{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 20.601336302895323,
  "eval_steps": 250,
  "global_step": 9250,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0022271714922048997,
      "grad_norm": 48.25035095214844,
      "learning_rate": 1e-06,
      "loss": 4.3999,
      "num_input_tokens_seen": 55440,
      "step": 1
    },
    {
      "epoch": 0.0022271714922048997,
      "loss": 4.068600654602051,
      "loss_ce": 1.3654756546020508,
      "loss_iou": 1.1171875,
      "loss_num": 0.09423828125,
      "loss_xval": 2.703125,
      "num_input_tokens_seen": 55440,
      "step": 1
    },
    {
      "epoch": 0.004454342984409799,
      "grad_norm": 130.20863342285156,
      "learning_rate": 1e-06,
      "loss": 3.4252,
      "num_input_tokens_seen": 111516,
      "step": 2
    },
    {
      "epoch": 0.004454342984409799,
      "loss": 3.1070704460144043,
      "loss_ce": 0.6607813835144043,
      "loss_iou": 1.0390625,
      "loss_num": 0.072265625,
      "loss_xval": 2.453125,
      "num_input_tokens_seen": 111516,
      "step": 2
    },
    {
      "epoch": 0.0066815144766146995,
      "grad_norm": 219.82708740234375,
      "learning_rate": 1e-06,
      "loss": 4.3295,
      "num_input_tokens_seen": 167164,
      "step": 3
    },
    {
      "epoch": 0.0066815144766146995,
      "loss": 4.420684814453125,
      "loss_ce": 1.4587706327438354,
      "loss_iou": 1.2109375,
      "loss_num": 0.10888671875,
      "loss_xval": 2.96875,
      "num_input_tokens_seen": 167164,
      "step": 3
    },
    {
      "epoch": 0.008908685968819599,
      "grad_norm": 42.13670349121094,
      "learning_rate": 1e-06,
      "loss": 4.1998,
      "num_input_tokens_seen": 221656,
      "step": 4
    },
    {
      "epoch": 0.008908685968819599,
      "loss": 4.317120552062988,
      "loss_ce": 1.9479798078536987,
      "loss_iou": 0.91015625,
      "loss_num": 0.10986328125,
      "loss_xval": 2.375,
      "num_input_tokens_seen": 221656,
      "step": 4
    },
    {
      "epoch": 0.011135857461024499,
      "grad_norm": 45.899261474609375,
      "learning_rate": 1e-06,
      "loss": 3.8544,
      "num_input_tokens_seen": 279772,
      "step": 5
    },
    {
      "epoch": 0.011135857461024499,
      "eval_seeclick_web_CIoU": 0.627178281545639,
      "eval_seeclick_web_GIoU": 0.6259045600891113,
      "eval_seeclick_web_IoU": 0.6463199555873871,
      "eval_seeclick_web_MAE_all": 0.015169001650065184,
      "eval_seeclick_web_MAE_h": 0.008642073255032301,
      "eval_seeclick_web_MAE_w": 0.0126046072691679,
      "eval_seeclick_web_MAE_x_boxes": 0.00952292513102293,
      "eval_seeclick_web_MAE_y_boxes": 0.022719496861100197,
      "eval_seeclick_web_inside_bbox": 0.8732638955116272,
      "eval_seeclick_web_loss": 0.8245152831077576,
      "eval_seeclick_web_loss_ce": 0.00025572007871232927,
      "eval_seeclick_web_loss_iou": 0.3670654296875,
      "eval_seeclick_web_loss_num": 0.011660575866699219,
      "eval_seeclick_web_loss_xval": 0.7923583984375,
      "eval_seeclick_web_runtime": 17.4436,
      "eval_seeclick_web_samples_per_second": 2.866,
      "eval_seeclick_web_steps_per_second": 0.115,
      "num_input_tokens_seen": 279772,
      "step": 5
    },
    {
      "epoch": 0.011135857461024499,
      "eval_icons_CIoU": 0.4010816812515259,
      "eval_icons_GIoU": 0.4324014186859131,
      "eval_icons_IoU": 0.49571336805820465,
      "eval_icons_MAE_all": 0.05109180323779583,
      "eval_icons_MAE_h": 0.020274315029382706,
      "eval_icons_MAE_w": 0.07078993320465088,
      "eval_icons_MAE_x_boxes": 0.05979071371257305,
      "eval_icons_MAE_y_boxes": 0.02338168118149042,
      "eval_icons_inside_bbox": 0.7361111044883728,
      "eval_icons_loss": 1.3772772550582886,
      "eval_icons_loss_ce": 9.112022598856129e-05,
      "eval_icons_loss_iou": 0.539794921875,
      "eval_icons_loss_num": 0.04870033264160156,
      "eval_icons_loss_xval": 1.323486328125,
      "eval_icons_runtime": 16.5509,
      "eval_icons_samples_per_second": 3.021,
      "eval_icons_steps_per_second": 0.121,
      "num_input_tokens_seen": 279772,
      "step": 5
    },
    {
      "epoch": 0.011135857461024499,
      "eval_screenspot_CIoU": 0.10782323777675629,
      "eval_screenspot_GIoU": 0.12048953274885814,
      "eval_screenspot_IoU": 0.26688433190186817,
      "eval_screenspot_MAE_all": 0.1258778969446818,
      "eval_screenspot_MAE_h": 0.07123650113741557,
      "eval_screenspot_MAE_w": 0.16419320305188498,
      "eval_screenspot_MAE_x_boxes": 0.1336704045534134,
      "eval_screenspot_MAE_y_boxes": 0.13144449392954508,
      "eval_screenspot_inside_bbox": 0.451666663090388,
      "eval_screenspot_loss": 2.3829233646392822,
      "eval_screenspot_loss_ce": 0.0014819869732794662,
      "eval_screenspot_loss_iou": 0.888671875,
      "eval_screenspot_loss_num": 0.1301116943359375,
      "eval_screenspot_loss_xval": 2.427734375,
      "eval_screenspot_runtime": 26.6936,
      "eval_screenspot_samples_per_second": 3.334,
      "eval_screenspot_steps_per_second": 0.112,
      "num_input_tokens_seen": 279772,
      "step": 5
    },
    {
      "epoch": 0.011135857461024499,
      "eval_compot_CIoU": 0.3970271050930023,
      "eval_compot_GIoU": 0.4374036639928818,
      "eval_compot_IoU": 0.45621901750564575,
      "eval_compot_MAE_all": 0.01914477813988924,
      "eval_compot_MAE_h": 0.005190690280869603,
      "eval_compot_MAE_w": 0.027808972634375095,
      "eval_compot_MAE_x_boxes": 0.03283216618001461,
      "eval_compot_MAE_y_boxes": 0.0045464420691132545,
      "eval_compot_inside_bbox": 0.6458333432674408,
      "eval_compot_loss": 1.301793098449707,
      "eval_compot_loss_ce": 0.0011670971289277077,
      "eval_compot_loss_iou": 0.60400390625,
      "eval_compot_loss_num": 0.018798828125,
      "eval_compot_loss_xval": 1.302001953125,
      "eval_compot_runtime": 16.8489,
      "eval_compot_samples_per_second": 2.968,
      "eval_compot_steps_per_second": 0.119,
      "num_input_tokens_seen": 279772,
      "step": 5
    },
    {
      "epoch": 0.011135857461024499,
      "eval_custom_ui_val_CIoU": -0.2649508896801207,
      "eval_custom_ui_val_GIoU": -0.28727365616295075,
      "eval_custom_ui_val_IoU": 0.02925413821099533,
      "eval_custom_ui_val_MAE_all": 0.09884808709224065,
      "eval_custom_ui_val_MAE_h": 0.0829097247785992,
      "eval_custom_ui_val_MAE_w": 0.06122648136483298,
      "eval_custom_ui_val_MAE_x_boxes": 0.05147905213137468,
      "eval_custom_ui_val_MAE_y_boxes": 0.1948982576529185,
      "eval_custom_ui_val_inside_bbox": 0.034722222222222224,
      "eval_custom_ui_val_loss": 3.0896787643432617,
      "eval_custom_ui_val_loss_ce": 0.0023468418318467834,
      "eval_custom_ui_val_loss_iou": 1.2951931423611112,
      "eval_custom_ui_val_loss_num": 0.09839375813802083,
      "eval_custom_ui_val_loss_xval": 3.082248263888889,
      "eval_custom_ui_val_runtime": 54.828,
      "eval_custom_ui_val_samples_per_second": 4.833,
      "eval_custom_ui_val_steps_per_second": 0.164,
      "num_input_tokens_seen": 279772,
      "step": 5
    },
    {
      "epoch": 0.011135857461024499,
      "loss": 2.992961883544922,
      "loss_ce": 0.004680817015469074,
      "loss_iou": 1.3046875,
      "loss_num": 0.07470703125,
      "loss_xval": 2.984375,
      "num_input_tokens_seen": 279772,
      "step": 5
    },
    {
      "epoch": 0.013363028953229399,
      "grad_norm": 51.13984298706055,
      "learning_rate": 1e-06,
      "loss": 3.8227,
      "num_input_tokens_seen": 336424,
      "step": 6
    },
    {
      "epoch": 0.013363028953229399,
      "loss": 3.419996738433838,
      "loss_ce": 1.4341567754745483,
      "loss_iou": 0.7578125,
      "loss_num": 0.09375,
      "loss_xval": 1.984375,
      "num_input_tokens_seen": 336424,
      "step": 6
    },
    {
      "epoch": 0.015590200445434299,
      "grad_norm": 43.559844970703125,
      "learning_rate": 1e-06,
      "loss": 3.5541,
      "num_input_tokens_seen": 391856,
      "step": 7
    },
    {
      "epoch": 0.015590200445434299,
      "loss": 3.7645998001098633,
      "loss_ce": 0.9286624193191528,
      "loss_iou": 1.140625,
      "loss_num": 0.1123046875,
      "loss_xval": 2.84375,
      "num_input_tokens_seen": 391856,
      "step": 7
    },
    {
      "epoch": 0.017817371937639197,
      "grad_norm": 58.651023864746094,
      "learning_rate": 1e-06,
      "loss": 4.0592,
      "num_input_tokens_seen": 446076,
      "step": 8
    },
    {
      "epoch": 0.017817371937639197,
      "loss": 3.763951301574707,
      "loss_ce": 1.845982551574707,
      "loss_iou": 0.69140625,
      "loss_num": 0.107421875,
      "loss_xval": 1.921875,
      "num_input_tokens_seen": 446076,
      "step": 8
    },
    {
      "epoch": 0.0200445434298441,
      "grad_norm": 46.13543701171875,
      "learning_rate": 1e-06,
      "loss": 3.4206,
      "num_input_tokens_seen": 503600,
      "step": 9
    },
    {
      "epoch": 0.0200445434298441,
      "loss": 3.4812936782836914,
      "loss_ce": 1.1902780532836914,
      "loss_iou": 0.90234375,
      "loss_num": 0.0966796875,
      "loss_xval": 2.296875,
      "num_input_tokens_seen": 503600,
      "step": 9
    },
    {
      "epoch": 0.022271714922048998,
      "grad_norm": 42.953372955322266,
      "learning_rate": 1e-06,
      "loss": 4.2745,
      "num_input_tokens_seen": 557604,
      "step": 10
    },
    {
      "epoch": 0.022271714922048998,
      "loss": 4.187029838562012,
      "loss_ce": 1.5722839832305908,
      "loss_iou": 1.0390625,
      "loss_num": 0.107421875,
      "loss_xval": 2.609375,
      "num_input_tokens_seen": 557604,
      "step": 10
    },
    {
      "epoch": 0.024498886414253896,
      "grad_norm": 53.018638610839844,
      "learning_rate": 1e-06,
      "loss": 4.4355,
      "num_input_tokens_seen": 614780,
      "step": 11
    },
    {
      "epoch": 0.024498886414253896,
      "loss": 3.7926254272460938,
      "loss_ce": 1.1353989839553833,
      "loss_iou": 1.0703125,
      "loss_num": 0.103515625,
      "loss_xval": 2.65625,
      "num_input_tokens_seen": 614780,
      "step": 11
    },
    {
      "epoch": 0.026726057906458798,
      "grad_norm": 52.364288330078125,
      "learning_rate": 1e-06,
      "loss": 3.4929,
      "num_input_tokens_seen": 672264,
      "step": 12
    },
    {
      "epoch": 0.026726057906458798,
      "loss": 3.3297152519226074,
      "loss_ce": 0.778934121131897,
      "loss_iou": 1.0859375,
      "loss_num": 0.0751953125,
      "loss_xval": 2.546875,
      "num_input_tokens_seen": 672264,
      "step": 12
    },
    {
      "epoch": 0.028953229398663696,
      "grad_norm": 160.2591552734375,
      "learning_rate": 1e-06,
      "loss": 4.0254,
      "num_input_tokens_seen": 728504,
      "step": 13
    },
    {
      "epoch": 0.028953229398663696,
      "loss": 4.575099945068359,
      "loss_ce": 1.8211936950683594,
      "loss_iou": 1.125,
      "loss_num": 0.10107421875,
      "loss_xval": 2.75,
      "num_input_tokens_seen": 728504,
      "step": 13
    },
    {
      "epoch": 0.031180400890868598,
      "grad_norm": 48.962005615234375,
      "learning_rate": 1e-06,
      "loss": 3.9505,
      "num_input_tokens_seen": 783692,
      "step": 14
    },
    {
      "epoch": 0.031180400890868598,
      "loss": 4.143004417419434,
      "loss_ce": 1.8163442611694336,
      "loss_iou": 0.94921875,
      "loss_num": 0.08642578125,
      "loss_xval": 2.328125,
      "num_input_tokens_seen": 783692,
      "step": 14
    },
    {
      "epoch": 0.0334075723830735,
      "grad_norm": 39.02128982543945,
      "learning_rate": 1e-06,
      "loss": 4.0185,
      "num_input_tokens_seen": 838168,
      "step": 15
    },
    {
      "epoch": 0.0334075723830735,
      "loss": 4.348506927490234,
      "loss_ce": 2.2537801265716553,
      "loss_iou": 0.78125,
      "loss_num": 0.1064453125,
      "loss_xval": 2.09375,
      "num_input_tokens_seen": 838168,
      "step": 15
    },
    {
      "epoch": 0.035634743875278395,
      "grad_norm": 35.96154022216797,
      "learning_rate": 1e-06,
      "loss": 3.6832,
      "num_input_tokens_seen": 895276,
      "step": 16
    },
    {
      "epoch": 0.035634743875278395,
      "loss": 3.568532943725586,
      "loss_ce": 1.519704818725586,
      "loss_iou": 0.83203125,
      "loss_num": 0.0771484375,
      "loss_xval": 2.046875,
      "num_input_tokens_seen": 895276,
      "step": 16
    },
    {
      "epoch": 0.0378619153674833,
      "grad_norm": 49.04829788208008,
      "learning_rate": 1e-06,
      "loss": 3.519,
      "num_input_tokens_seen": 951180,
      "step": 17
    },
    {
      "epoch": 0.0378619153674833,
      "loss": 3.1685400009155273,
      "loss_ce": 0.9458837509155273,
      "loss_iou": 0.88671875,
      "loss_num": 0.0908203125,
      "loss_xval": 2.21875,
      "num_input_tokens_seen": 951180,
      "step": 17
    },
    {
      "epoch": 0.0400890868596882,
      "grad_norm": 105.61235809326172,
      "learning_rate": 1e-06,
      "loss": 3.7186,
      "num_input_tokens_seen": 1008684,
      "step": 18
    },
    {
      "epoch": 0.0400890868596882,
      "loss": 3.8322296142578125,
      "loss_ce": 1.353714108467102,
      "loss_iou": 0.96484375,
      "loss_num": 0.10986328125,
      "loss_xval": 2.484375,
      "num_input_tokens_seen": 1008684,
      "step": 18
    },
    {
      "epoch": 0.042316258351893093,
      "grad_norm": 114.26968383789062,
      "learning_rate": 1e-06,
      "loss": 4.0211,
      "num_input_tokens_seen": 1062212,
      "step": 19
    },
    {
      "epoch": 0.042316258351893093,
      "loss": 3.5515711307525635,
      "loss_ce": 1.0281336307525635,
      "loss_iou": 1.0234375,
      "loss_num": 0.09521484375,
      "loss_xval": 2.53125,
      "num_input_tokens_seen": 1062212,
      "step": 19
    },
    {
      "epoch": 0.044543429844097995,
      "grad_norm": 53.97627258300781,
      "learning_rate": 1e-06,
      "loss": 4.2972,
      "num_input_tokens_seen": 1114576,
      "step": 20
    },
    {
      "epoch": 0.044543429844097995,
      "loss": 4.559757709503174,
      "loss_ce": 1.2492105960845947,
      "loss_iou": 1.3046875,
      "loss_num": 0.138671875,
      "loss_xval": 3.3125,
      "num_input_tokens_seen": 1114576,
      "step": 20
    },
    {
      "epoch": 0.0467706013363029,
      "grad_norm": 38.33933639526367,
      "learning_rate": 1e-06,
      "loss": 3.5725,
      "num_input_tokens_seen": 1169836,
      "step": 21
    },
    {
      "epoch": 0.0467706013363029,
      "loss": 3.7388784885406494,
      "loss_ce": 1.3677849769592285,
      "loss_iou": 0.98828125,
      "loss_num": 0.0791015625,
      "loss_xval": 2.375,
      "num_input_tokens_seen": 1169836,
      "step": 21
    },
    {
      "epoch": 0.04899777282850779,
      "grad_norm": 59.448753356933594,
      "learning_rate": 1e-06,
      "loss": 3.8666,
      "num_input_tokens_seen": 1228308,
      "step": 22
    },
    {
      "epoch": 0.04899777282850779,
      "loss": 3.630049705505371,
      "loss_ce": 1.1505573987960815,
      "loss_iou": 0.9921875,
      "loss_num": 0.0986328125,
      "loss_xval": 2.484375,
      "num_input_tokens_seen": 1228308,
      "step": 22
    },
    {
      "epoch": 0.051224944320712694,
      "grad_norm": 52.553958892822266,
      "learning_rate": 1e-06,
      "loss": 3.362,
      "num_input_tokens_seen": 1287564,
      "step": 23
    },
    {
      "epoch": 0.051224944320712694,
      "loss": 2.8519859313964844,
      "loss_ce": 1.0302085876464844,
      "loss_iou": 0.75390625,
      "loss_num": 0.0625,
      "loss_xval": 1.8203125,
      "num_input_tokens_seen": 1287564,
      "step": 23
    },
    {
      "epoch": 0.053452115812917596,
      "grad_norm": 49.04064178466797,
      "learning_rate": 1e-06,
      "loss": 4.1017,
      "num_input_tokens_seen": 1343120,
      "step": 24
    },
    {
      "epoch": 0.053452115812917596,
      "loss": 4.231411933898926,
      "loss_ce": 1.8066072463989258,
      "loss_iou": 0.96875,
      "loss_num": 0.09716796875,
      "loss_xval": 2.421875,
      "num_input_tokens_seen": 1343120,
      "step": 24
    },
    {
      "epoch": 0.0556792873051225,
      "grad_norm": 56.048038482666016,
      "learning_rate": 1e-06,
      "loss": 4.3803,
      "num_input_tokens_seen": 1396700,
      "step": 25
    },
    {
      "epoch": 0.0556792873051225,
      "loss": 4.276466369628906,
      "loss_ce": 1.6143567562103271,
      "loss_iou": 1.0234375,
      "loss_num": 0.1240234375,
      "loss_xval": 2.65625,
      "num_input_tokens_seen": 1396700,
      "step": 25
    },
    {
      "epoch": 0.05790645879732739,
      "grad_norm": 39.70867919921875,
      "learning_rate": 1e-06,
      "loss": 3.3416,
      "num_input_tokens_seen": 1453876,
      "step": 26
    },
    {
      "epoch": 0.05790645879732739,
      "loss": 3.4961788654327393,
      "loss_ce": 1.0899291038513184,
      "loss_iou": 1.0234375,
      "loss_num": 0.0732421875,
      "loss_xval": 2.40625,
      "num_input_tokens_seen": 1453876,
      "step": 26
    },
    {
      "epoch": 0.060133630289532294,
      "grad_norm": 60.43865966796875,
      "learning_rate": 1e-06,
      "loss": 3.5507,
      "num_input_tokens_seen": 1509556,
      "step": 27
    },
    {
      "epoch": 0.060133630289532294,
      "loss": 3.679036855697632,
      "loss_ce": 1.376302719116211,
      "loss_iou": 0.890625,
      "loss_num": 0.103515625,
      "loss_xval": 2.296875,
      "num_input_tokens_seen": 1509556,
      "step": 27
    },
    {
      "epoch": 0.062360801781737196,
      "grad_norm": 33.97861099243164,
      "learning_rate": 1e-06,
      "loss": 3.5827,
      "num_input_tokens_seen": 1566848,
      "step": 28
    },
    {
      "epoch": 0.062360801781737196,
      "loss": 3.9071435928344727,
      "loss_ce": 1.2176905870437622,
      "loss_iou": 1.0703125,
      "loss_num": 0.1103515625,
      "loss_xval": 2.6875,
      "num_input_tokens_seen": 1566848,
      "step": 28
    },
    {
      "epoch": 0.0645879732739421,
      "grad_norm": 71.39791107177734,
      "learning_rate": 1e-06,
      "loss": 3.9493,
      "num_input_tokens_seen": 1622132,
      "step": 29
    },
    {
      "epoch": 0.0645879732739421,
      "loss": 4.430097579956055,
      "loss_ce": 1.3754103183746338,
      "loss_iou": 1.203125,
      "loss_num": 0.1298828125,
      "loss_xval": 3.0625,
      "num_input_tokens_seen": 1622132,
      "step": 29
    },
    {
      "epoch": 0.066815144766147,
      "grad_norm": 111.8135757446289,
      "learning_rate": 1e-06,
      "loss": 3.5729,
      "num_input_tokens_seen": 1680504,
      "step": 30
    },
    {
      "epoch": 0.066815144766147,
      "loss": 3.6753127574920654,
      "loss_ce": 1.620625376701355,
      "loss_iou": 0.76953125,
      "loss_num": 0.10302734375,
      "loss_xval": 2.0625,
      "num_input_tokens_seen": 1680504,
      "step": 30
    },
    {
      "epoch": 0.06904231625835189,
      "grad_norm": 48.25176239013672,
      "learning_rate": 1e-06,
      "loss": 3.9768,
      "num_input_tokens_seen": 1735984,
      "step": 31
    },
    {
      "epoch": 0.06904231625835189,
      "loss": 4.000043869018555,
      "loss_ce": 1.2251414060592651,
      "loss_iou": 1.1171875,
      "loss_num": 0.107421875,
      "loss_xval": 2.78125,
      "num_input_tokens_seen": 1735984,
      "step": 31
    },
    {
      "epoch": 0.07126948775055679,
      "grad_norm": 105.31439208984375,
      "learning_rate": 1e-06,
      "loss": 4.4586,
      "num_input_tokens_seen": 1791036,
      "step": 32
    },
    {
      "epoch": 0.07126948775055679,
      "loss": 4.315090656280518,
      "loss_ce": 1.6012234687805176,
      "loss_iou": 1.078125,
      "loss_num": 0.1103515625,
      "loss_xval": 2.71875,
      "num_input_tokens_seen": 1791036,
      "step": 32
    },
    {
      "epoch": 0.07349665924276169,
      "grad_norm": 61.652164459228516,
      "learning_rate": 1e-06,
      "loss": 4.2882,
      "num_input_tokens_seen": 1846348,
      "step": 33
    },
    {
      "epoch": 0.07349665924276169,
      "loss": 3.7140510082244873,
      "loss_ce": 1.3263558149337769,
      "loss_iou": 0.9375,
      "loss_num": 0.1025390625,
      "loss_xval": 2.390625,
      "num_input_tokens_seen": 1846348,
      "step": 33
    },
    {
      "epoch": 0.0757238307349666,
      "grad_norm": 40.629302978515625,
      "learning_rate": 1e-06,
      "loss": 3.8681,
      "num_input_tokens_seen": 1902912,
      "step": 34
    },
    {
      "epoch": 0.0757238307349666,
      "loss": 3.7667856216430664,
      "loss_ce": 1.078309178352356,
      "loss_iou": 1.109375,
      "loss_num": 0.09375,
      "loss_xval": 2.6875,
      "num_input_tokens_seen": 1902912,
      "step": 34
    },
    {
      "epoch": 0.0779510022271715,
      "grad_norm": 68.22655487060547,
      "learning_rate": 1e-06,
      "loss": 4.0071,
      "num_input_tokens_seen": 1958520,
      "step": 35
    },
    {
      "epoch": 0.0779510022271715,
      "loss": 3.5187363624572754,
      "loss_ce": 1.068541169166565,
      "loss_iou": 0.9296875,
      "loss_num": 0.11865234375,
      "loss_xval": 2.453125,
      "num_input_tokens_seen": 1958520,
      "step": 35
    },
    {
      "epoch": 0.0801781737193764,
      "grad_norm": 76.44446563720703,
      "learning_rate": 1e-06,
      "loss": 4.2565,
      "num_input_tokens_seen": 2011292,
      "step": 36
    },
    {
      "epoch": 0.0801781737193764,
      "loss": 3.761260509490967,
      "loss_ce": 1.2253718376159668,
      "loss_iou": 1.015625,
      "loss_num": 0.10205078125,
      "loss_xval": 2.53125,
      "num_input_tokens_seen": 2011292,
      "step": 36
    },
    {
      "epoch": 0.08240534521158129,
      "grad_norm": 40.924400329589844,
      "learning_rate": 1e-06,
      "loss": 3.7774,
      "num_input_tokens_seen": 2068200,
      "step": 37
    },
    {
      "epoch": 0.08240534521158129,
      "loss": 3.715888500213623,
      "loss_ce": 1.778876781463623,
      "loss_iou": 0.75390625,
      "loss_num": 0.0859375,
      "loss_xval": 1.9375,
      "num_input_tokens_seen": 2068200,
      "step": 37
    },
    {
      "epoch": 0.08463251670378619,
      "grad_norm": 63.25859069824219,
      "learning_rate": 1e-06,
      "loss": 3.4187,
      "num_input_tokens_seen": 2124620,
      "step": 38
    },
    {
      "epoch": 0.08463251670378619,
      "loss": 3.6475610733032227,
      "loss_ce": 1.252053141593933,
      "loss_iou": 0.953125,
      "loss_num": 0.09765625,
      "loss_xval": 2.390625,
      "num_input_tokens_seen": 2124620,
      "step": 38
    },
    {
      "epoch": 0.08685968819599109,
      "grad_norm": 53.20210266113281,
      "learning_rate": 1e-06,
      "loss": 3.69,
      "num_input_tokens_seen": 2180648,
      "step": 39
    },
    {
      "epoch": 0.08685968819599109,
      "loss": 3.29498291015625,
      "loss_ce": 1.070373773574829,
      "loss_iou": 0.921875,
      "loss_num": 0.076171875,
      "loss_xval": 2.21875,
      "num_input_tokens_seen": 2180648,
      "step": 39
    },
    {
      "epoch": 0.08908685968819599,
      "grad_norm": 56.693077087402344,
      "learning_rate": 1e-06,
      "loss": 3.8869,
      "num_input_tokens_seen": 2239320,
      "step": 40
    },
    {
      "epoch": 0.08908685968819599,
      "loss": 4.501180648803711,
      "loss_ce": 1.5011805295944214,
      "loss_iou": 1.1796875,
      "loss_num": 0.1298828125,
      "loss_xval": 3.0,
      "num_input_tokens_seen": 2239320,
      "step": 40
    },
    {
      "epoch": 0.09131403118040089,
      "grad_norm": 40.52269744873047,
      "learning_rate": 1e-06,
      "loss": 3.3113,
      "num_input_tokens_seen": 2294136,
      "step": 41
    },
    {
      "epoch": 0.09131403118040089,
      "loss": 3.6476950645446777,
      "loss_ce": 1.1711325645446777,
      "loss_iou": 0.9609375,
      "loss_num": 0.1103515625,
      "loss_xval": 2.46875,
      "num_input_tokens_seen": 2294136,
      "step": 41
    },
    {
      "epoch": 0.0935412026726058,
      "grad_norm": 45.07380676269531,
      "learning_rate": 1e-06,
      "loss": 4.2714,
      "num_input_tokens_seen": 2349572,
      "step": 42
    },
    {
      "epoch": 0.0935412026726058,
      "loss": 4.451573371887207,
      "loss_ce": 1.497471570968628,
      "loss_iou": 1.1953125,
      "loss_num": 0.11279296875,
      "loss_xval": 2.953125,
      "num_input_tokens_seen": 2349572,
      "step": 42
    },
    {
      "epoch": 0.0957683741648107,
      "grad_norm": 43.15333557128906,
      "learning_rate": 1e-06,
      "loss": 4.3193,
      "num_input_tokens_seen": 2405120,
      "step": 43
    },
    {
      "epoch": 0.0957683741648107,
      "loss": 4.689324378967285,
      "loss_ce": 1.3651058673858643,
      "loss_iou": 1.34375,
      "loss_num": 0.125,
      "loss_xval": 3.328125,
      "num_input_tokens_seen": 2405120,
      "step": 43
    },
    {
      "epoch": 0.09799554565701558,
      "grad_norm": 43.444942474365234,
      "learning_rate": 1e-06,
      "loss": 3.2982,
      "num_input_tokens_seen": 2459840,
      "step": 44
    },
    {
      "epoch": 0.09799554565701558,
      "loss": 3.244344472885132,
      "loss_ce": 0.8068445920944214,
      "loss_iou": 1.03125,
      "loss_num": 0.07470703125,
      "loss_xval": 2.4375,
      "num_input_tokens_seen": 2459840,
      "step": 44
    },
    {
      "epoch": 0.10022271714922049,
      "grad_norm": 43.407318115234375,
      "learning_rate": 1e-06,
      "loss": 3.8361,
      "num_input_tokens_seen": 2513464,
      "step": 45
    },
    {
      "epoch": 0.10022271714922049,
      "loss": 3.987287998199463,
      "loss_ce": 1.3007644414901733,
      "loss_iou": 1.0625,
      "loss_num": 0.111328125,
      "loss_xval": 2.6875,
      "num_input_tokens_seen": 2513464,
      "step": 45
    },
    {
      "epoch": 0.10244988864142539,
      "grad_norm": 47.13986587524414,
      "learning_rate": 1e-06,
      "loss": 3.9219,
      "num_input_tokens_seen": 2572040,
      "step": 46
    },
    {
      "epoch": 0.10244988864142539,
      "loss": 4.833956718444824,
      "loss_ce": 1.9491908550262451,
      "loss_iou": 1.0703125,
      "loss_num": 0.1484375,
      "loss_xval": 2.890625,
      "num_input_tokens_seen": 2572040,
      "step": 46
    },
    {
      "epoch": 0.10467706013363029,
      "grad_norm": 48.692970275878906,
      "learning_rate": 1e-06,
      "loss": 3.7242,
      "num_input_tokens_seen": 2627832,
      "step": 47
    },
    {
      "epoch": 0.10467706013363029,
      "loss": 3.7583022117614746,
      "loss_ce": 1.0444347858428955,
      "loss_iou": 1.078125,
      "loss_num": 0.11181640625,
      "loss_xval": 2.71875,
      "num_input_tokens_seen": 2627832,
      "step": 47
    },
    {
      "epoch": 0.10690423162583519,
      "grad_norm": 39.18439483642578,
      "learning_rate": 1e-06,
      "loss": 4.2013,
      "num_input_tokens_seen": 2682316,
      "step": 48
    },
    {
      "epoch": 0.10690423162583519,
      "loss": 4.251529216766357,
      "loss_ce": 1.5435214042663574,
      "loss_iou": 1.0625,
      "loss_num": 0.1162109375,
      "loss_xval": 2.703125,
      "num_input_tokens_seen": 2682316,
      "step": 48
    },
    {
      "epoch": 0.1091314031180401,
      "grad_norm": 40.1721305847168,
      "learning_rate": 1e-06,
      "loss": 3.7247,
      "num_input_tokens_seen": 2738948,
      "step": 49
    },
    {
      "epoch": 0.1091314031180401,
      "loss": 3.429368734359741,
      "loss_ce": 1.0699937343597412,
      "loss_iou": 0.96875,
      "loss_num": 0.0849609375,
      "loss_xval": 2.359375,
      "num_input_tokens_seen": 2738948,
      "step": 49
    },
    {
      "epoch": 0.111358574610245,
      "grad_norm": 53.2962646484375,
      "learning_rate": 1e-06,
      "loss": 3.3685,
      "num_input_tokens_seen": 2795512,
      "step": 50
    },
    {
      "epoch": 0.111358574610245,
      "loss": 3.372053861618042,
      "loss_ce": 1.0507646799087524,
      "loss_iou": 0.9375,
      "loss_num": 0.08935546875,
      "loss_xval": 2.328125,
      "num_input_tokens_seen": 2795512,
      "step": 50
    },
    {
      "epoch": 0.11358574610244988,
      "grad_norm": 39.22996139526367,
      "learning_rate": 1e-06,
      "loss": 3.5006,
      "num_input_tokens_seen": 2853264,
      "step": 51
    },
    {
      "epoch": 0.11358574610244988,
      "loss": 3.352569580078125,
      "loss_ce": 0.9648742079734802,
      "loss_iou": 0.9609375,
      "loss_num": 0.09423828125,
      "loss_xval": 2.390625,
      "num_input_tokens_seen": 2853264,
      "step": 51
    },
    {
      "epoch": 0.11581291759465479,
      "grad_norm": 51.9208869934082,
      "learning_rate": 1e-06,
      "loss": 4.002,
      "num_input_tokens_seen": 2909744,
      "step": 52
    },
    {
      "epoch": 0.11581291759465479,
      "loss": 3.849562168121338,
      "loss_ce": 1.003859043121338,
      "loss_iou": 1.0546875,
      "loss_num": 0.1455078125,
      "loss_xval": 2.84375,
      "num_input_tokens_seen": 2909744,
      "step": 52
    },
    {
      "epoch": 0.11804008908685969,
      "grad_norm": 36.589988708496094,
      "learning_rate": 1e-06,
      "loss": 3.8423,
      "num_input_tokens_seen": 2963684,
      "step": 53
    },
    {
      "epoch": 0.11804008908685969,
      "loss": 3.315175771713257,
      "loss_ce": 1.3342187404632568,
      "loss_iou": 0.77734375,
      "loss_num": 0.0859375,
      "loss_xval": 1.984375,
      "num_input_tokens_seen": 2963684,
      "step": 53
    },
    {
      "epoch": 0.12026726057906459,
      "grad_norm": 93.62321472167969,
      "learning_rate": 1e-06,
      "loss": 3.6344,
      "num_input_tokens_seen": 3019120,
      "step": 54
    },
    {
      "epoch": 0.12026726057906459,
      "loss": 3.678225040435791,
      "loss_ce": 1.3032249212265015,
      "loss_iou": 0.95703125,
      "loss_num": 0.09228515625,
      "loss_xval": 2.375,
      "num_input_tokens_seen": 3019120,
      "step": 54
    },
    {
      "epoch": 0.12249443207126949,
      "grad_norm": 73.76739501953125,
      "learning_rate": 1e-06,
      "loss": 4.5231,
      "num_input_tokens_seen": 3074692,
      "step": 55
    },
    {
      "epoch": 0.12249443207126949,
      "loss": 4.42036247253418,
      "loss_ce": 1.8959481716156006,
      "loss_iou": 0.91796875,
      "loss_num": 0.1376953125,
      "loss_xval": 2.53125,
      "num_input_tokens_seen": 3074692,
      "step": 55
    },
    {
      "epoch": 0.12472160356347439,
      "grad_norm": 58.37528991699219,
      "learning_rate": 1e-06,
      "loss": 4.4363,
      "num_input_tokens_seen": 3128300,
      "step": 56
    },
    {
      "epoch": 0.12472160356347439,
      "loss": 4.376946926116943,
      "loss_ce": 1.451165795326233,
      "loss_iou": 1.140625,
      "loss_num": 0.12890625,
      "loss_xval": 2.921875,
      "num_input_tokens_seen": 3128300,
      "step": 56
    },
    {
      "epoch": 0.12694877505567928,
      "grad_norm": 50.1368293762207,
      "learning_rate": 1e-06,
      "loss": 3.0486,
      "num_input_tokens_seen": 3186576,
      "step": 57
    },
    {
      "epoch": 0.12694877505567928,
      "loss": 3.0530099868774414,
      "loss_ce": 0.8830881118774414,
      "loss_iou": 0.828125,
      "loss_num": 0.1025390625,
      "loss_xval": 2.171875,
      "num_input_tokens_seen": 3186576,
      "step": 57
    },
    {
      "epoch": 0.1291759465478842,
      "grad_norm": 88.61648559570312,
      "learning_rate": 1e-06,
      "loss": 3.6031,
      "num_input_tokens_seen": 3243908,
      "step": 58
    },
    {
      "epoch": 0.1291759465478842,
      "loss": 3.6260769367218018,
      "loss_ce": 1.0352567434310913,
      "loss_iou": 1.0546875,
      "loss_num": 0.095703125,
      "loss_xval": 2.59375,
      "num_input_tokens_seen": 3243908,
      "step": 58
    },
    {
      "epoch": 0.13140311804008908,
      "grad_norm": 366.4051818847656,
      "learning_rate": 1e-06,
      "loss": 2.9703,
      "num_input_tokens_seen": 3301940,
      "step": 59
    },
    {
      "epoch": 0.13140311804008908,
      "loss": 2.9112725257873535,
      "loss_ce": 0.6681084632873535,
      "loss_iou": 0.890625,
      "loss_num": 0.09326171875,
      "loss_xval": 2.25,
      "num_input_tokens_seen": 3301940,
      "step": 59
    },
    {
      "epoch": 0.133630289532294,
      "grad_norm": 60.00054931640625,
      "learning_rate": 1e-06,
      "loss": 3.4228,
      "num_input_tokens_seen": 3357440,
      "step": 60
    },
    {
      "epoch": 0.133630289532294,
      "loss": 4.004701137542725,
      "loss_ce": 1.8259902000427246,
      "loss_iou": 0.8125,
      "loss_num": 0.10986328125,
      "loss_xval": 2.171875,
      "num_input_tokens_seen": 3357440,
      "step": 60
    },
    {
      "epoch": 0.1358574610244989,
      "grad_norm": 75.00879669189453,
      "learning_rate": 1e-06,
      "loss": 3.1458,
      "num_input_tokens_seen": 3416064,
      "step": 61
    },
    {
      "epoch": 0.1358574610244989,
      "loss": 3.1031904220581055,
      "loss_ce": 1.0651044845581055,
      "loss_iou": 0.828125,
      "loss_num": 0.076171875,
      "loss_xval": 2.03125,
      "num_input_tokens_seen": 3416064,
      "step": 61
    },
    {
      "epoch": 0.13808463251670378,
      "grad_norm": 44.26047134399414,
      "learning_rate": 1e-06,
      "loss": 3.4534,
      "num_input_tokens_seen": 3474300,
      "step": 62
    },
    {
      "epoch": 0.13808463251670378,
      "loss": 3.3606114387512207,
      "loss_ce": 0.8108068704605103,
      "loss_iou": 0.98828125,
      "loss_num": 0.1142578125,
      "loss_xval": 2.546875,
      "num_input_tokens_seen": 3474300,
      "step": 62
    },
    {
      "epoch": 0.1403118040089087,
      "grad_norm": 43.13284683227539,
      "learning_rate": 1e-06,
      "loss": 4.0539,
      "num_input_tokens_seen": 3526756,
      "step": 63
    },
    {
      "epoch": 0.1403118040089087,
      "loss": 3.9055140018463135,
      "loss_ce": 1.0422327518463135,
      "loss_iou": 1.1015625,
      "loss_num": 0.1328125,
      "loss_xval": 2.859375,
      "num_input_tokens_seen": 3526756,
      "step": 63
    },
    {
      "epoch": 0.14253897550111358,
      "grad_norm": 48.860225677490234,
      "learning_rate": 1e-06,
      "loss": 3.7684,
      "num_input_tokens_seen": 3582620,
      "step": 64
    },
    {
      "epoch": 0.14253897550111358,
      "loss": 3.5178382396698,
      "loss_ce": 0.9514319896697998,
      "loss_iou": 1.0234375,
      "loss_num": 0.10400390625,
      "loss_xval": 2.5625,
      "num_input_tokens_seen": 3582620,
      "step": 64
    },
    {
      "epoch": 0.1447661469933185,
      "grad_norm": 51.4886474609375,
      "learning_rate": 1e-06,
      "loss": 4.0096,
      "num_input_tokens_seen": 3636136,
      "step": 65
    },
    {
      "epoch": 0.1447661469933185,
      "loss": 4.009946823120117,
      "loss_ce": 1.5128767490386963,
      "loss_iou": 0.9765625,
      "loss_num": 0.10888671875,
      "loss_xval": 2.5,
      "num_input_tokens_seen": 3636136,
      "step": 65
    },
    {
      "epoch": 0.14699331848552338,
      "grad_norm": 75.6802978515625,
      "learning_rate": 1e-06,
      "loss": 4.0291,
      "num_input_tokens_seen": 3692172,
      "step": 66
    },
    {
      "epoch": 0.14699331848552338,
      "loss": 3.9964442253112793,
      "loss_ce": 1.5921471118927002,
      "loss_iou": 0.84765625,
      "loss_num": 0.1416015625,
      "loss_xval": 2.40625,
      "num_input_tokens_seen": 3692172,
      "step": 66
    },
    {
      "epoch": 0.1492204899777283,
      "grad_norm": 44.05568313598633,
      "learning_rate": 1e-06,
      "loss": 2.7517,
      "num_input_tokens_seen": 3750032,
      "step": 67
    },
    {
      "epoch": 0.1492204899777283,
      "loss": 2.7267603874206543,
      "loss_ce": 0.6720730662345886,
      "loss_iou": 0.8671875,
      "loss_num": 0.06396484375,
      "loss_xval": 2.0625,
      "num_input_tokens_seen": 3750032,
      "step": 67
    },
    {
      "epoch": 0.1514476614699332,
      "grad_norm": 33.830841064453125,
      "learning_rate": 1e-06,
      "loss": 3.3022,
      "num_input_tokens_seen": 3804556,
      "step": 68
    },
    {
      "epoch": 0.1514476614699332,
      "loss": 3.4905872344970703,
      "loss_ce": 0.7855091094970703,
      "loss_iou": 1.046875,
      "loss_num": 0.12451171875,
      "loss_xval": 2.703125,
      "num_input_tokens_seen": 3804556,
      "step": 68
    },
    {
      "epoch": 0.15367483296213807,
      "grad_norm": 47.166080474853516,
      "learning_rate": 1e-06,
      "loss": 3.6395,
      "num_input_tokens_seen": 3860224,
      "step": 69
    },
    {
      "epoch": 0.15367483296213807,
      "loss": 3.921863555908203,
      "loss_ce": 1.5263557434082031,
      "loss_iou": 0.8828125,
      "loss_num": 0.1259765625,
      "loss_xval": 2.390625,
      "num_input_tokens_seen": 3860224,
      "step": 69
    },
    {
      "epoch": 0.155902004454343,
      "grad_norm": 60.735130310058594,
      "learning_rate": 1e-06,
      "loss": 4.3284,
      "num_input_tokens_seen": 3911976,
      "step": 70
    },
    {
      "epoch": 0.155902004454343,
      "loss": 4.273301124572754,
      "loss_ce": 1.4632422924041748,
      "loss_iou": 1.1328125,
      "loss_num": 0.10791015625,
      "loss_xval": 2.8125,
      "num_input_tokens_seen": 3911976,
      "step": 70
    },
    {
      "epoch": 0.15812917594654788,
      "grad_norm": 40.85783004760742,
      "learning_rate": 1e-06,
      "loss": 3.3411,
      "num_input_tokens_seen": 3969416,
      "step": 71
    },
    {
      "epoch": 0.15812917594654788,
      "loss": 3.9168148040771484,
      "loss_ce": 1.2000181674957275,
      "loss_iou": 0.99609375,
      "loss_num": 0.1455078125,
      "loss_xval": 2.71875,
      "num_input_tokens_seen": 3969416,
      "step": 71
    },
    {
      "epoch": 0.1603563474387528,
      "grad_norm": 37.3762092590332,
      "learning_rate": 1e-06,
      "loss": 4.0877,
      "num_input_tokens_seen": 4024764,
      "step": 72
    },
    {
      "epoch": 0.1603563474387528,
      "loss": 3.9297492504119873,
      "loss_ce": 1.3613898754119873,
      "loss_iou": 1.0390625,
      "loss_num": 0.09912109375,
      "loss_xval": 2.5625,
      "num_input_tokens_seen": 4024764,
      "step": 72
    },
    {
      "epoch": 0.16258351893095768,
      "grad_norm": 37.51104736328125,
      "learning_rate": 1e-06,
      "loss": 3.6427,
      "num_input_tokens_seen": 4081604,
      "step": 73
    },
    {
      "epoch": 0.16258351893095768,
      "loss": 3.7234535217285156,
      "loss_ce": 1.1570473909378052,
      "loss_iou": 1.0,
      "loss_num": 0.11328125,
      "loss_xval": 2.5625,
      "num_input_tokens_seen": 4081604,
      "step": 73
    },
    {
      "epoch": 0.16481069042316257,
      "grad_norm": 58.83037567138672,
      "learning_rate": 1e-06,
      "loss": 3.6275,
      "num_input_tokens_seen": 4138792,
      "step": 74
    },
    {
      "epoch": 0.16481069042316257,
      "loss": 3.8506052494049072,
      "loss_ce": 1.0537302494049072,
      "loss_iou": 1.078125,
      "loss_num": 0.12890625,
      "loss_xval": 2.796875,
      "num_input_tokens_seen": 4138792,
      "step": 74
    },
    {
      "epoch": 0.16703786191536749,
      "grad_norm": 74.164794921875,
      "learning_rate": 1e-06,
      "loss": 3.8793,
      "num_input_tokens_seen": 4195448,
      "step": 75
    },
    {
      "epoch": 0.16703786191536749,
      "loss": 4.057165145874023,
      "loss_ce": 0.9848995208740234,
      "loss_iou": 1.1796875,
      "loss_num": 0.1416015625,
      "loss_xval": 3.078125,
      "num_input_tokens_seen": 4195448,
      "step": 75
    },
    {
      "epoch": 0.16926503340757237,
      "grad_norm": 103.62609100341797,
      "learning_rate": 1e-06,
      "loss": 3.7588,
      "num_input_tokens_seen": 4247988,
      "step": 76
    },
    {
      "epoch": 0.16926503340757237,
      "loss": 3.4790072441101074,
      "loss_ce": 1.3315460681915283,
      "loss_iou": 0.83203125,
      "loss_num": 0.0966796875,
      "loss_xval": 2.140625,
      "num_input_tokens_seen": 4247988,
      "step": 76
    },
    {
      "epoch": 0.1714922048997773,
      "grad_norm": 36.90887451171875,
      "learning_rate": 1e-06,
      "loss": 3.8448,
      "num_input_tokens_seen": 4301000,
      "step": 77
    },
    {
      "epoch": 0.1714922048997773,
      "loss": 4.019002437591553,
      "loss_ce": 1.0971274375915527,
      "loss_iou": 1.1328125,
      "loss_num": 0.1328125,
      "loss_xval": 2.921875,
      "num_input_tokens_seen": 4301000,
      "step": 77
    },
    {
      "epoch": 0.17371937639198218,
      "grad_norm": 60.749755859375,
      "learning_rate": 1e-06,
      "loss": 3.6311,
      "num_input_tokens_seen": 4358284,
      "step": 78
    },
    {
      "epoch": 0.17371937639198218,
      "loss": 3.9924979209899902,
      "loss_ce": 1.3831228017807007,
      "loss_iou": 1.0390625,
      "loss_num": 0.10693359375,
      "loss_xval": 2.609375,
      "num_input_tokens_seen": 4358284,
      "step": 78
    },
    {
      "epoch": 0.1759465478841871,
      "grad_norm": 43.58686065673828,
      "learning_rate": 1e-06,
      "loss": 3.7598,
      "num_input_tokens_seen": 4412984,
      "step": 79
    },
    {
      "epoch": 0.1759465478841871,
      "loss": 4.099183559417725,
      "loss_ce": 1.1939101219177246,
      "loss_iou": 1.0859375,
      "loss_num": 0.1474609375,
      "loss_xval": 2.90625,
      "num_input_tokens_seen": 4412984,
      "step": 79
    },
    {
      "epoch": 0.17817371937639198,
      "grad_norm": 46.925628662109375,
      "learning_rate": 1e-06,
      "loss": 3.6665,
      "num_input_tokens_seen": 4469440,
      "step": 80
    },
    {
      "epoch": 0.17817371937639198,
      "loss": 3.7074530124664307,
      "loss_ce": 1.1742497682571411,
      "loss_iou": 0.9609375,
      "loss_num": 0.12109375,
      "loss_xval": 2.53125,
      "num_input_tokens_seen": 4469440,
      "step": 80
    },
    {
      "epoch": 0.18040089086859687,
      "grad_norm": 37.318233489990234,
      "learning_rate": 1e-06,
      "loss": 3.108,
      "num_input_tokens_seen": 4528044,
      "step": 81
    },
    {
      "epoch": 0.18040089086859687,
      "loss": 3.302272319793701,
      "loss_ce": 0.8784441947937012,
      "loss_iou": 0.91796875,
      "loss_num": 0.11767578125,
      "loss_xval": 2.421875,
      "num_input_tokens_seen": 4528044,
      "step": 81
    },
    {
      "epoch": 0.18262806236080179,
      "grad_norm": 89.16364288330078,
      "learning_rate": 1e-06,
      "loss": 2.8823,
      "num_input_tokens_seen": 4585856,
      "step": 82
    },
    {
      "epoch": 0.18262806236080179,
      "loss": 2.5211141109466553,
      "loss_ce": 0.4742392301559448,
      "loss_iou": 0.84765625,
      "loss_num": 0.06982421875,
      "loss_xval": 2.046875,
      "num_input_tokens_seen": 4585856,
      "step": 82
    },
    {
      "epoch": 0.18485523385300667,
      "grad_norm": 75.21393585205078,
      "learning_rate": 1e-06,
      "loss": 3.7455,
      "num_input_tokens_seen": 4640456,
      "step": 83
    },
    {
      "epoch": 0.18485523385300667,
      "loss": 4.00370979309082,
      "loss_ce": 1.2458975315093994,
      "loss_iou": 1.09375,
      "loss_num": 0.11328125,
      "loss_xval": 2.75,
      "num_input_tokens_seen": 4640456,
      "step": 83
    },
    {
      "epoch": 0.1870824053452116,
      "grad_norm": 47.31288146972656,
      "learning_rate": 1e-06,
      "loss": 3.7102,
      "num_input_tokens_seen": 4695604,
      "step": 84
    },
    {
      "epoch": 0.1870824053452116,
      "loss": 3.4363784790039062,
      "loss_ce": 1.1434097290039062,
      "loss_iou": 0.8671875,
      "loss_num": 0.1123046875,
      "loss_xval": 2.296875,
      "num_input_tokens_seen": 4695604,
      "step": 84
    },
    {
      "epoch": 0.18930957683741648,
      "grad_norm": 103.16661071777344,
      "learning_rate": 1e-06,
      "loss": 3.9615,
      "num_input_tokens_seen": 4750568,
      "step": 85
    },
    {
      "epoch": 0.18930957683741648,
      "loss": 3.598466396331787,
      "loss_ce": 0.9617477655410767,
      "loss_iou": 0.98046875,
      "loss_num": 0.134765625,
      "loss_xval": 2.640625,
      "num_input_tokens_seen": 4750568,
      "step": 85
    },
    {
      "epoch": 0.1915367483296214,
      "grad_norm": 30.83519172668457,
      "learning_rate": 1e-06,
      "loss": 3.4149,
      "num_input_tokens_seen": 4802048,
      "step": 86
    },
    {
      "epoch": 0.1915367483296214,
      "loss": 3.000054359436035,
      "loss_ce": 0.6123592257499695,
      "loss_iou": 0.890625,
      "loss_num": 0.12158203125,
      "loss_xval": 2.390625,
      "num_input_tokens_seen": 4802048,
      "step": 86
    },
    {
      "epoch": 0.19376391982182628,
      "grad_norm": 44.40241241455078,
      "learning_rate": 1e-06,
      "loss": 3.8462,
      "num_input_tokens_seen": 4857292,
      "step": 87
    },
    {
      "epoch": 0.19376391982182628,
      "loss": 3.8453128337860107,
      "loss_ce": 1.0943361520767212,
      "loss_iou": 1.0546875,
      "loss_num": 0.1279296875,
      "loss_xval": 2.75,
      "num_input_tokens_seen": 4857292,
      "step": 87
    },
    {
      "epoch": 0.19599109131403117,
      "grad_norm": 36.97007751464844,
      "learning_rate": 1e-06,
      "loss": 3.4149,
      "num_input_tokens_seen": 4914020,
      "step": 88
    },
    {
      "epoch": 0.19599109131403117,
      "loss": 3.3789005279541016,
      "loss_ce": 1.074212908744812,
      "loss_iou": 0.84375,
      "loss_num": 0.1220703125,
      "loss_xval": 2.3125,
      "num_input_tokens_seen": 4914020,
      "step": 88
    },
    {
      "epoch": 0.19821826280623608,
      "grad_norm": 100.41455841064453,
      "learning_rate": 1e-06,
      "loss": 3.7294,
      "num_input_tokens_seen": 4970596,
      "step": 89
    },
    {
      "epoch": 0.19821826280623608,
      "loss": 3.8161191940307617,
      "loss_ce": 1.1676816940307617,
      "loss_iou": 1.046875,
      "loss_num": 0.1123046875,
      "loss_xval": 2.65625,
      "num_input_tokens_seen": 4970596,
      "step": 89
    },
    {
      "epoch": 0.20044543429844097,
      "grad_norm": 56.474788665771484,
      "learning_rate": 1e-06,
      "loss": 3.628,
      "num_input_tokens_seen": 5022084,
      "step": 90
    },
    {
      "epoch": 0.20044543429844097,
      "loss": 3.246203899383545,
      "loss_ce": 0.770129919052124,
      "loss_iou": 0.89453125,
      "loss_num": 0.13671875,
      "loss_xval": 2.46875,
      "num_input_tokens_seen": 5022084,
      "step": 90
    },
    {
      "epoch": 0.2026726057906459,
      "grad_norm": 45.0289192199707,
      "learning_rate": 1e-06,
      "loss": 2.9406,
      "num_input_tokens_seen": 5076540,
      "step": 91
    },
    {
      "epoch": 0.2026726057906459,
      "loss": 2.8975300788879395,
      "loss_ce": 0.491280198097229,
      "loss_iou": 0.9296875,
      "loss_num": 0.109375,
      "loss_xval": 2.40625,
      "num_input_tokens_seen": 5076540,
      "step": 91
    },
    {
      "epoch": 0.20489977728285078,
      "grad_norm": 33.83285140991211,
      "learning_rate": 1e-06,
      "loss": 2.8553,
      "num_input_tokens_seen": 5132312,
      "step": 92
    },
    {
      "epoch": 0.20489977728285078,
      "loss": 2.73876953125,
      "loss_ce": 0.7338864803314209,
      "loss_iou": 0.75390625,
      "loss_num": 0.09912109375,
      "loss_xval": 2.0,
      "num_input_tokens_seen": 5132312,
      "step": 92
    },
    {
      "epoch": 0.2071269487750557,
      "grad_norm": 55.93543243408203,
      "learning_rate": 1e-06,
      "loss": 3.3996,
      "num_input_tokens_seen": 5188188,
      "step": 93
    },
    {
      "epoch": 0.2071269487750557,
      "loss": 3.369142770767212,
      "loss_ce": 0.7919942140579224,
      "loss_iou": 0.95703125,
      "loss_num": 0.1318359375,
      "loss_xval": 2.578125,
      "num_input_tokens_seen": 5188188,
      "step": 93
    },
    {
      "epoch": 0.20935412026726058,
      "grad_norm": 42.92362976074219,
      "learning_rate": 1e-06,
      "loss": 3.3957,
      "num_input_tokens_seen": 5245580,
      "step": 94
    },
    {
      "epoch": 0.20935412026726058,
      "loss": 3.3484046459198,
      "loss_ce": 0.7302405834197998,
      "loss_iou": 0.9609375,
      "loss_num": 0.1396484375,
      "loss_xval": 2.625,
      "num_input_tokens_seen": 5245580,
      "step": 94
    },
    {
      "epoch": 0.21158129175946547,
      "grad_norm": 38.73051452636719,
      "learning_rate": 1e-06,
      "loss": 2.8372,
      "num_input_tokens_seen": 5302720,
      "step": 95
    },
    {
      "epoch": 0.21158129175946547,
      "loss": 3.1018481254577637,
      "loss_ce": 0.6829028129577637,
      "loss_iou": 0.90625,
      "loss_num": 0.12109375,
      "loss_xval": 2.421875,
      "num_input_tokens_seen": 5302720,
      "step": 95
    },
    {
      "epoch": 0.21380846325167038,
      "grad_norm": 51.393985748291016,
      "learning_rate": 1e-06,
      "loss": 3.4311,
      "num_input_tokens_seen": 5359268,
      "step": 96
    },
    {
      "epoch": 0.21380846325167038,
      "loss": 3.3091301918029785,
      "loss_ce": 1.068895697593689,
      "loss_iou": 0.796875,
      "loss_num": 0.1298828125,
      "loss_xval": 2.234375,
      "num_input_tokens_seen": 5359268,
      "step": 96
    },
    {
      "epoch": 0.21603563474387527,
      "grad_norm": 34.125511169433594,
      "learning_rate": 1e-06,
      "loss": 3.3903,
      "num_input_tokens_seen": 5415716,
      "step": 97
    },
    {
      "epoch": 0.21603563474387527,
      "loss": 3.3011648654937744,
      "loss_ce": 0.8431569933891296,
      "loss_iou": 0.8828125,
      "loss_num": 0.138671875,
      "loss_xval": 2.453125,
      "num_input_tokens_seen": 5415716,
      "step": 97
    },
    {
      "epoch": 0.2182628062360802,
      "grad_norm": 47.519710540771484,
      "learning_rate": 1e-06,
      "loss": 2.9414,
      "num_input_tokens_seen": 5473288,
      "step": 98
    },
    {
      "epoch": 0.2182628062360802,
      "loss": 3.0692644119262695,
      "loss_ce": 0.71477210521698,
      "loss_iou": 0.88671875,
      "loss_num": 0.11669921875,
      "loss_xval": 2.359375,
      "num_input_tokens_seen": 5473288,
      "step": 98
    },
    {
      "epoch": 0.22048997772828507,
      "grad_norm": 26.452730178833008,
      "learning_rate": 1e-06,
      "loss": 3.3211,
      "num_input_tokens_seen": 5528484,
      "step": 99
    },
    {
      "epoch": 0.22048997772828507,
      "loss": 3.3662614822387695,
      "loss_ce": 0.9463395476341248,
      "loss_iou": 0.875,
      "loss_num": 0.1328125,
      "loss_xval": 2.421875,
      "num_input_tokens_seen": 5528484,
      "step": 99
    },
    {
      "epoch": 0.22271714922049,
      "grad_norm": 60.19447708129883,
      "learning_rate": 1e-06,
      "loss": 3.2708,
      "num_input_tokens_seen": 5584888,
      "step": 100
    },
    {
      "epoch": 0.22271714922049,
      "loss": 3.1879868507385254,
      "loss_ce": 0.9902329444885254,
      "loss_iou": 0.79296875,
      "loss_num": 0.1220703125,
      "loss_xval": 2.203125,
      "num_input_tokens_seen": 5584888,
      "step": 100
    },
    {
      "epoch": 0.22494432071269488,
      "grad_norm": 70.54967498779297,
      "learning_rate": 1e-06,
      "loss": 2.7226,
      "num_input_tokens_seen": 5642204,
      "step": 101
    },
    {
      "epoch": 0.22494432071269488,
      "loss": 2.864590644836426,
      "loss_ce": 0.7434969544410706,
      "loss_iou": 0.765625,
      "loss_num": 0.11865234375,
      "loss_xval": 2.125,
      "num_input_tokens_seen": 5642204,
      "step": 101
    },
    {
      "epoch": 0.22717149220489977,
      "grad_norm": 30.121597290039062,
      "learning_rate": 1e-06,
      "loss": 3.6091,
      "num_input_tokens_seen": 5696676,
      "step": 102
    },
    {
      "epoch": 0.22717149220489977,
      "loss": 3.1128854751586914,
      "loss_ce": 0.7256782054901123,
      "loss_iou": 0.8828125,
      "loss_num": 0.12353515625,
      "loss_xval": 2.390625,
      "num_input_tokens_seen": 5696676,
      "step": 102
    },
    {
      "epoch": 0.22939866369710468,
      "grad_norm": 38.39320373535156,
      "learning_rate": 1e-06,
      "loss": 2.9509,
      "num_input_tokens_seen": 5751512,
      "step": 103
    },
    {
      "epoch": 0.22939866369710468,
      "loss": 3.0732927322387695,
      "loss_ce": 0.5625505447387695,
      "loss_iou": 0.90234375,
      "loss_num": 0.140625,
      "loss_xval": 2.515625,
      "num_input_tokens_seen": 5751512,
      "step": 103
    },
    {
      "epoch": 0.23162583518930957,
      "grad_norm": 38.619789123535156,
      "learning_rate": 1e-06,
      "loss": 3.2933,
      "num_input_tokens_seen": 5806704,
      "step": 104
    },
    {
      "epoch": 0.23162583518930957,
      "loss": 3.1526646614074707,
      "loss_ce": 0.8069615960121155,
      "loss_iou": 0.89453125,
      "loss_num": 0.11083984375,
      "loss_xval": 2.34375,
      "num_input_tokens_seen": 5806704,
      "step": 104
    },
    {
      "epoch": 0.23385300668151449,
      "grad_norm": 39.458309173583984,
      "learning_rate": 1e-06,
      "loss": 2.705,
      "num_input_tokens_seen": 5864592,
      "step": 105
    },
    {
      "epoch": 0.23385300668151449,
      "loss": 2.9264371395111084,
      "loss_ce": 0.6334683895111084,
      "loss_iou": 0.859375,
      "loss_num": 0.11474609375,
      "loss_xval": 2.296875,
      "num_input_tokens_seen": 5864592,
      "step": 105
    },
    {
      "epoch": 0.23608017817371937,
      "grad_norm": 91.0936050415039,
      "learning_rate": 1e-06,
      "loss": 2.6971,
      "num_input_tokens_seen": 5920932,
      "step": 106
    },
    {
      "epoch": 0.23608017817371937,
      "loss": 2.743122100830078,
      "loss_ce": 0.6098213195800781,
      "loss_iou": 0.75,
      "loss_num": 0.126953125,
      "loss_xval": 2.140625,
      "num_input_tokens_seen": 5920932,
      "step": 106
    },
    {
      "epoch": 0.2383073496659243,
      "grad_norm": 58.84850311279297,
      "learning_rate": 1e-06,
      "loss": 3.047,
      "num_input_tokens_seen": 5978660,
      "step": 107
    },
    {
      "epoch": 0.2383073496659243,
      "loss": 2.7701728343963623,
      "loss_ce": 0.4518134593963623,
      "loss_iou": 0.8125,
      "loss_num": 0.138671875,
      "loss_xval": 2.3125,
      "num_input_tokens_seen": 5978660,
      "step": 107
    },
    {
      "epoch": 0.24053452115812918,
      "grad_norm": 37.72629928588867,
      "learning_rate": 1e-06,
      "loss": 3.1347,
      "num_input_tokens_seen": 6032308,
      "step": 108
    },
    {
      "epoch": 0.24053452115812918,
      "loss": 3.364459991455078,
      "loss_ce": 0.720416784286499,
      "loss_iou": 0.98828125,
      "loss_num": 0.1337890625,
      "loss_xval": 2.640625,
      "num_input_tokens_seen": 6032308,
      "step": 108
    },
    {
      "epoch": 0.24276169265033407,
      "grad_norm": 35.40589141845703,
      "learning_rate": 1e-06,
      "loss": 2.8689,
      "num_input_tokens_seen": 6087012,
      "step": 109
    },
    {
      "epoch": 0.24276169265033407,
      "loss": 2.9507226943969727,
      "loss_ce": 0.40775376558303833,
      "loss_iou": 0.9453125,
      "loss_num": 0.1298828125,
      "loss_xval": 2.546875,
      "num_input_tokens_seen": 6087012,
      "step": 109
    },
    {
      "epoch": 0.24498886414253898,
      "grad_norm": 42.018192291259766,
      "learning_rate": 1e-06,
      "loss": 2.9651,
      "num_input_tokens_seen": 6144992,
      "step": 110
    },
    {
      "epoch": 0.24498886414253898,
      "loss": 2.959064483642578,
      "loss_ce": 0.5215646624565125,
      "loss_iou": 0.88671875,
      "loss_num": 0.1318359375,
      "loss_xval": 2.4375,
      "num_input_tokens_seen": 6144992,
      "step": 110
    },
    {
      "epoch": 0.24721603563474387,
      "grad_norm": 80.21202850341797,
      "learning_rate": 1e-06,
      "loss": 2.9912,
      "num_input_tokens_seen": 6201808,
      "step": 111
    },
    {
      "epoch": 0.24721603563474387,
      "loss": 3.193042755126953,
      "loss_ce": 0.9244881868362427,
      "loss_iou": 0.875,
      "loss_num": 0.10400390625,
      "loss_xval": 2.265625,
      "num_input_tokens_seen": 6201808,
      "step": 111
    },
    {
      "epoch": 0.24944320712694878,
      "grad_norm": 54.696163177490234,
      "learning_rate": 1e-06,
      "loss": 3.2172,
      "num_input_tokens_seen": 6253116,
      "step": 112
    },
    {
      "epoch": 0.24944320712694878,
      "loss": 3.254600763320923,
      "loss_ce": 0.6891711354255676,
      "loss_iou": 0.94140625,
      "loss_num": 0.13671875,
      "loss_xval": 2.5625,
      "num_input_tokens_seen": 6253116,
      "step": 112
    },
    {
      "epoch": 0.2516703786191537,
      "grad_norm": 42.921669006347656,
      "learning_rate": 1e-06,
      "loss": 3.186,
      "num_input_tokens_seen": 6308684,
      "step": 113
    },
    {
      "epoch": 0.2516703786191537,
      "loss": 3.4225220680236816,
      "loss_ce": 0.48892807960510254,
      "loss_iou": 1.046875,
      "loss_num": 0.16796875,
      "loss_xval": 2.9375,
      "num_input_tokens_seen": 6308684,
      "step": 113
    },
    {
      "epoch": 0.25389755011135856,
      "grad_norm": 78.24114227294922,
      "learning_rate": 1e-06,
      "loss": 3.0514,
      "num_input_tokens_seen": 6364580,
      "step": 114
    },
    {
      "epoch": 0.25389755011135856,
      "loss": 3.6505331993103027,
      "loss_ce": 0.7813925743103027,
      "loss_iou": 1.0546875,
      "loss_num": 0.1533203125,
      "loss_xval": 2.875,
      "num_input_tokens_seen": 6364580,
      "step": 114
    },
    {
      "epoch": 0.2561247216035635,
      "grad_norm": 43.09880447387695,
      "learning_rate": 1e-06,
      "loss": 3.0621,
      "num_input_tokens_seen": 6418640,
      "step": 115
    },
    {
      "epoch": 0.2561247216035635,
      "loss": 2.922844648361206,
      "loss_ce": 0.6381767392158508,
      "loss_iou": 0.8125,
      "loss_num": 0.1328125,
      "loss_xval": 2.28125,
      "num_input_tokens_seen": 6418640,
      "step": 115
    },
    {
      "epoch": 0.2583518930957684,
      "grad_norm": 78.85059356689453,
      "learning_rate": 1e-06,
      "loss": 2.7225,
      "num_input_tokens_seen": 6476332,
      "step": 116
    },
    {
      "epoch": 0.2583518930957684,
      "loss": 2.682021141052246,
      "loss_ce": 0.4828024208545685,
      "loss_iou": 0.8046875,
      "loss_num": 0.119140625,
      "loss_xval": 2.203125,
      "num_input_tokens_seen": 6476332,
      "step": 116
    },
    {
      "epoch": 0.26057906458797325,
      "grad_norm": 39.069583892822266,
      "learning_rate": 1e-06,
      "loss": 2.7889,
      "num_input_tokens_seen": 6531328,
      "step": 117
    },
    {
      "epoch": 0.26057906458797325,
      "loss": 2.9253664016723633,
      "loss_ce": 0.6001709699630737,
      "loss_iou": 0.859375,
      "loss_num": 0.12060546875,
      "loss_xval": 2.328125,
      "num_input_tokens_seen": 6531328,
      "step": 117
    },
    {
      "epoch": 0.26280623608017817,
      "grad_norm": 40.309391021728516,
      "learning_rate": 1e-06,
      "loss": 2.8465,
      "num_input_tokens_seen": 6589548,
      "step": 118
    },
    {
      "epoch": 0.26280623608017817,
      "loss": 2.8932766914367676,
      "loss_ce": 0.43429216742515564,
      "loss_iou": 0.87890625,
      "loss_num": 0.1396484375,
      "loss_xval": 2.453125,
      "num_input_tokens_seen": 6589548,
      "step": 118
    },
    {
      "epoch": 0.2650334075723831,
      "grad_norm": 31.454940795898438,
      "learning_rate": 1e-06,
      "loss": 2.6428,
      "num_input_tokens_seen": 6647160,
      "step": 119
    },
    {
      "epoch": 0.2650334075723831,
      "loss": 2.7675888538360596,
      "loss_ce": 0.48755955696105957,
      "loss_iou": 0.78515625,
      "loss_num": 0.142578125,
      "loss_xval": 2.28125,
      "num_input_tokens_seen": 6647160,
      "step": 119
    },
    {
      "epoch": 0.267260579064588,
      "grad_norm": 108.88129425048828,
      "learning_rate": 1e-06,
      "loss": 2.7364,
      "num_input_tokens_seen": 6705976,
      "step": 120
    },
    {
      "epoch": 0.267260579064588,
      "loss": 2.662890911102295,
      "loss_ce": 0.4177738428115845,
      "loss_iou": 0.796875,
      "loss_num": 0.1298828125,
      "loss_xval": 2.25,
      "num_input_tokens_seen": 6705976,
      "step": 120
    },
    {
      "epoch": 0.26948775055679286,
      "grad_norm": 63.05554962158203,
      "learning_rate": 1e-06,
      "loss": 3.2316,
      "num_input_tokens_seen": 6760876,
      "step": 121
    },
    {
      "epoch": 0.26948775055679286,
      "loss": 3.1148786544799805,
      "loss_ce": 0.5191754102706909,
      "loss_iou": 0.96875,
      "loss_num": 0.130859375,
      "loss_xval": 2.59375,
      "num_input_tokens_seen": 6760876,
      "step": 121
    },
    {
      "epoch": 0.2717149220489978,
      "grad_norm": 39.52189254760742,
      "learning_rate": 1e-06,
      "loss": 3.0243,
      "num_input_tokens_seen": 6815624,
      "step": 122
    },
    {
      "epoch": 0.2717149220489978,
      "loss": 2.8611388206481934,
      "loss_ce": 0.37090444564819336,
      "loss_iou": 0.91796875,
      "loss_num": 0.1298828125,
      "loss_xval": 2.484375,
      "num_input_tokens_seen": 6815624,
      "step": 122
    },
    {
      "epoch": 0.2739420935412027,
      "grad_norm": 44.9962158203125,
      "learning_rate": 1e-06,
      "loss": 2.9217,
      "num_input_tokens_seen": 6870556,
      "step": 123
    },
    {
      "epoch": 0.2739420935412027,
      "loss": 3.207249164581299,
      "loss_ce": 0.7170149087905884,
      "loss_iou": 0.94921875,
      "loss_num": 0.11865234375,
      "loss_xval": 2.484375,
      "num_input_tokens_seen": 6870556,
      "step": 123
    },
    {
      "epoch": 0.27616926503340755,
      "grad_norm": 25.661319732666016,
      "learning_rate": 1e-06,
      "loss": 2.9669,
      "num_input_tokens_seen": 6925992,
      "step": 124
    },
    {
      "epoch": 0.27616926503340755,
      "loss": 2.76767897605896,
      "loss_ce": 0.3028353154659271,
      "loss_iou": 0.91015625,
      "loss_num": 0.1298828125,
      "loss_xval": 2.46875,
      "num_input_tokens_seen": 6925992,
      "step": 124
    },
    {
      "epoch": 0.27839643652561247,
      "grad_norm": 41.912784576416016,
      "learning_rate": 1e-06,
      "loss": 2.6341,
      "num_input_tokens_seen": 6983364,
      "step": 125
    },
    {
      "epoch": 0.27839643652561247,
      "loss": 2.6220178604125977,
      "loss_ce": 0.4081503748893738,
      "loss_iou": 0.86328125,
      "loss_num": 0.09765625,
      "loss_xval": 2.21875,
      "num_input_tokens_seen": 6983364,
      "step": 125
    },
    {
      "epoch": 0.2806236080178174,
      "grad_norm": 60.407127380371094,
      "learning_rate": 1e-06,
      "loss": 2.6668,
      "num_input_tokens_seen": 7037256,
      "step": 126
    },
    {
      "epoch": 0.2806236080178174,
      "loss": 2.8176770210266113,
      "loss_ce": 0.3831067681312561,
      "loss_iou": 0.86328125,
      "loss_num": 0.140625,
      "loss_xval": 2.4375,
      "num_input_tokens_seen": 7037256,
      "step": 126
    },
    {
      "epoch": 0.2828507795100223,
      "grad_norm": 102.43517303466797,
      "learning_rate": 1e-06,
      "loss": 2.8805,
      "num_input_tokens_seen": 7091296,
      "step": 127
    },
    {
      "epoch": 0.2828507795100223,
      "loss": 2.8854475021362305,
      "loss_ce": 0.7106426954269409,
      "loss_iou": 0.79296875,
      "loss_num": 0.1171875,
      "loss_xval": 2.171875,
      "num_input_tokens_seen": 7091296,
      "step": 127
    },
    {
      "epoch": 0.28507795100222716,
      "grad_norm": 44.01148223876953,
      "learning_rate": 1e-06,
      "loss": 2.8418,
      "num_input_tokens_seen": 7146940,
      "step": 128
    },
    {
      "epoch": 0.28507795100222716,
      "loss": 2.973205089569092,
      "loss_ce": 0.391173779964447,
      "loss_iou": 0.90625,
      "loss_num": 0.154296875,
      "loss_xval": 2.578125,
      "num_input_tokens_seen": 7146940,
      "step": 128
    },
    {
      "epoch": 0.2873051224944321,
      "grad_norm": 116.4285659790039,
      "learning_rate": 1e-06,
      "loss": 2.5866,
      "num_input_tokens_seen": 7203656,
      "step": 129
    },
    {
      "epoch": 0.2873051224944321,
      "loss": 3.1285247802734375,
      "loss_ce": 0.44102469086647034,
      "loss_iou": 0.9609375,
      "loss_num": 0.15234375,
      "loss_xval": 2.6875,
      "num_input_tokens_seen": 7203656,
      "step": 129
    },
    {
      "epoch": 0.289532293986637,
      "grad_norm": 24.682891845703125,
      "learning_rate": 1e-06,
      "loss": 2.6812,
      "num_input_tokens_seen": 7258748,
      "step": 130
    },
    {
      "epoch": 0.289532293986637,
      "loss": 2.730166435241699,
      "loss_ce": 0.5348541140556335,
      "loss_iou": 0.8046875,
      "loss_num": 0.11767578125,
      "loss_xval": 2.1875,
      "num_input_tokens_seen": 7258748,
      "step": 130
    },
    {
      "epoch": 0.29175946547884185,
      "grad_norm": 34.84796142578125,
      "learning_rate": 1e-06,
      "loss": 2.5544,
      "num_input_tokens_seen": 7315644,
      "step": 131
    },
    {
      "epoch": 0.29175946547884185,
      "loss": 2.3853559494018555,
      "loss_ce": 0.3248090147972107,
      "loss_iou": 0.78515625,
      "loss_num": 0.09716796875,
      "loss_xval": 2.0625,
      "num_input_tokens_seen": 7315644,
      "step": 131
    },
    {
      "epoch": 0.29398663697104677,
      "grad_norm": 33.59828186035156,
      "learning_rate": 1e-06,
      "loss": 2.6018,
      "num_input_tokens_seen": 7371296,
      "step": 132
    },
    {
      "epoch": 0.29398663697104677,
      "loss": 2.7992615699768066,
      "loss_ce": 0.5941836833953857,
      "loss_iou": 0.78125,
      "loss_num": 0.12890625,
      "loss_xval": 2.203125,
      "num_input_tokens_seen": 7371296,
      "step": 132
    },
    {
      "epoch": 0.2962138084632517,
      "grad_norm": 26.216712951660156,
      "learning_rate": 1e-06,
      "loss": 2.4007,
      "num_input_tokens_seen": 7427892,
      "step": 133
    },
    {
      "epoch": 0.2962138084632517,
      "loss": 2.259786605834961,
      "loss_ce": 0.35256001353263855,
      "loss_iou": 0.73828125,
      "loss_num": 0.0859375,
      "loss_xval": 1.90625,
      "num_input_tokens_seen": 7427892,
      "step": 133
    },
    {
      "epoch": 0.2984409799554566,
      "grad_norm": 148.89346313476562,
      "learning_rate": 1e-06,
      "loss": 2.645,
      "num_input_tokens_seen": 7484960,
      "step": 134
    },
    {
      "epoch": 0.2984409799554566,
      "loss": 2.5535106658935547,
      "loss_ce": 0.44901835918426514,
      "loss_iou": 0.80078125,
      "loss_num": 0.099609375,
      "loss_xval": 2.109375,
      "num_input_tokens_seen": 7484960,
      "step": 134
    },
    {
      "epoch": 0.30066815144766146,
      "grad_norm": 29.48415184020996,
      "learning_rate": 1e-06,
      "loss": 2.4559,
      "num_input_tokens_seen": 7543972,
      "step": 135
    },
    {
      "epoch": 0.30066815144766146,
      "loss": 2.225175380706787,
      "loss_ce": 0.2398238629102707,
      "loss_iou": 0.7421875,
      "loss_num": 0.10009765625,
      "loss_xval": 1.984375,
      "num_input_tokens_seen": 7543972,
      "step": 135
    },
    {
      "epoch": 0.3028953229398664,
      "grad_norm": 50.64659881591797,
      "learning_rate": 1e-06,
      "loss": 2.7228,
      "num_input_tokens_seen": 7600460,
      "step": 136
    },
    {
      "epoch": 0.3028953229398664,
      "loss": 2.643749475479126,
      "loss_ce": 0.2501947581768036,
      "loss_iou": 0.8515625,
      "loss_num": 0.1376953125,
      "loss_xval": 2.390625,
      "num_input_tokens_seen": 7600460,
      "step": 136
    },
    {
      "epoch": 0.3051224944320713,
      "grad_norm": 27.465696334838867,
      "learning_rate": 1e-06,
      "loss": 2.5444,
      "num_input_tokens_seen": 7657896,
      "step": 137
    },
    {
      "epoch": 0.3051224944320713,
      "loss": 2.5869100093841553,
      "loss_ce": 0.3886679410934448,
      "loss_iou": 0.796875,
      "loss_num": 0.12158203125,
      "loss_xval": 2.203125,
      "num_input_tokens_seen": 7657896,
      "step": 137
    },
    {
      "epoch": 0.30734966592427615,
      "grad_norm": 30.95069122314453,
      "learning_rate": 1e-06,
      "loss": 2.6368,
      "num_input_tokens_seen": 7710764,
      "step": 138
    },
    {
      "epoch": 0.30734966592427615,
      "loss": 2.530655860900879,
      "loss_ce": 0.41444480419158936,
      "loss_iou": 0.70703125,
      "loss_num": 0.140625,
      "loss_xval": 2.109375,
      "num_input_tokens_seen": 7710764,
      "step": 138
    },
    {
      "epoch": 0.30957683741648107,
      "grad_norm": 27.392942428588867,
      "learning_rate": 1e-06,
      "loss": 2.3889,
      "num_input_tokens_seen": 7767452,
      "step": 139
    },
    {
      "epoch": 0.30957683741648107,
      "loss": 2.26643705368042,
      "loss_ce": 0.3201477825641632,
      "loss_iou": 0.7421875,
      "loss_num": 0.0927734375,
      "loss_xval": 1.9453125,
      "num_input_tokens_seen": 7767452,
      "step": 139
    },
    {
      "epoch": 0.311804008908686,
      "grad_norm": 52.42769241333008,
      "learning_rate": 1e-06,
      "loss": 2.5987,
      "num_input_tokens_seen": 7821504,
      "step": 140
    },
    {
      "epoch": 0.311804008908686,
      "loss": 2.6357715129852295,
      "loss_ce": 0.3486621379852295,
      "loss_iou": 0.8671875,
      "loss_num": 0.10986328125,
      "loss_xval": 2.28125,
      "num_input_tokens_seen": 7821504,
      "step": 140
    },
    {
      "epoch": 0.31403118040089084,
      "grad_norm": 23.9682674407959,
      "learning_rate": 1e-06,
      "loss": 2.6711,
      "num_input_tokens_seen": 7875900,
      "step": 141
    },
    {
      "epoch": 0.31403118040089084,
      "loss": 2.9198169708251953,
      "loss_ce": 0.44423115253448486,
      "loss_iou": 0.9140625,
      "loss_num": 0.12890625,
      "loss_xval": 2.46875,
      "num_input_tokens_seen": 7875900,
      "step": 141
    },
    {
      "epoch": 0.31625835189309576,
      "grad_norm": 103.93788146972656,
      "learning_rate": 1e-06,
      "loss": 2.5893,
      "num_input_tokens_seen": 7931732,
      "step": 142
    },
    {
      "epoch": 0.31625835189309576,
      "loss": 2.553285837173462,
      "loss_ce": 0.41266071796417236,
      "loss_iou": 0.82421875,
      "loss_num": 0.09814453125,
      "loss_xval": 2.140625,
      "num_input_tokens_seen": 7931732,
      "step": 142
    },
    {
      "epoch": 0.3184855233853007,
      "grad_norm": 44.492820739746094,
      "learning_rate": 1e-06,
      "loss": 2.4701,
      "num_input_tokens_seen": 7987608,
      "step": 143
    },
    {
      "epoch": 0.3184855233853007,
      "loss": 2.3122029304504395,
      "loss_ce": 0.3502885699272156,
      "loss_iou": 0.7265625,
      "loss_num": 0.1025390625,
      "loss_xval": 1.9609375,
      "num_input_tokens_seen": 7987608,
      "step": 143
    },
    {
      "epoch": 0.3207126948775056,
      "grad_norm": 23.186241149902344,
      "learning_rate": 1e-06,
      "loss": 2.4726,
      "num_input_tokens_seen": 8043360,
      "step": 144
    },
    {
      "epoch": 0.3207126948775056,
      "loss": 2.7117819786071777,
      "loss_ce": 0.3328755497932434,
      "loss_iou": 0.88671875,
      "loss_num": 0.12060546875,
      "loss_xval": 2.375,
      "num_input_tokens_seen": 8043360,
      "step": 144
    },
    {
      "epoch": 0.32293986636971045,
      "grad_norm": 60.27910614013672,
      "learning_rate": 1e-06,
      "loss": 2.6172,
      "num_input_tokens_seen": 8098064,
      "step": 145
    },
    {
      "epoch": 0.32293986636971045,
      "loss": 2.32407283782959,
      "loss_ce": 0.28696349263191223,
      "loss_iou": 0.796875,
      "loss_num": 0.08837890625,
      "loss_xval": 2.03125,
      "num_input_tokens_seen": 8098064,
      "step": 145
    },
    {
      "epoch": 0.32516703786191536,
      "grad_norm": 28.506025314331055,
      "learning_rate": 1e-06,
      "loss": 2.534,
      "num_input_tokens_seen": 8153264,
      "step": 146
    },
    {
      "epoch": 0.32516703786191536,
      "loss": 2.474531650543213,
      "loss_ce": 0.3632035255432129,
      "loss_iou": 0.7421875,
      "loss_num": 0.1259765625,
      "loss_xval": 2.109375,
      "num_input_tokens_seen": 8153264,
      "step": 146
    },
    {
      "epoch": 0.3273942093541203,
      "grad_norm": 55.345096588134766,
      "learning_rate": 1e-06,
      "loss": 2.5464,
      "num_input_tokens_seen": 8209784,
      "step": 147
    },
    {
      "epoch": 0.3273942093541203,
      "loss": 2.5160865783691406,
      "loss_ce": 0.48581331968307495,
      "loss_iou": 0.7578125,
      "loss_num": 0.10205078125,
      "loss_xval": 2.03125,
      "num_input_tokens_seen": 8209784,
      "step": 147
    },
    {
      "epoch": 0.32962138084632514,
      "grad_norm": 44.24598693847656,
      "learning_rate": 1e-06,
      "loss": 2.7815,
      "num_input_tokens_seen": 8262396,
      "step": 148
    },
    {
      "epoch": 0.32962138084632514,
      "loss": 2.644756317138672,
      "loss_ce": 0.3185845911502838,
      "loss_iou": 0.88671875,
      "loss_num": 0.11083984375,
      "loss_xval": 2.328125,
      "num_input_tokens_seen": 8262396,
      "step": 148
    },
    {
      "epoch": 0.33184855233853006,
      "grad_norm": 39.58821105957031,
      "learning_rate": 1e-06,
      "loss": 2.3255,
      "num_input_tokens_seen": 8316924,
      "step": 149
    },
    {
      "epoch": 0.33184855233853006,
      "loss": 2.2807397842407227,
      "loss_ce": 0.3002711534500122,
      "loss_iou": 0.73046875,
      "loss_num": 0.103515625,
      "loss_xval": 1.984375,
      "num_input_tokens_seen": 8316924,
      "step": 149
    },
    {
      "epoch": 0.33407572383073497,
      "grad_norm": 30.283483505249023,
      "learning_rate": 1e-06,
      "loss": 2.3857,
      "num_input_tokens_seen": 8372780,
      "step": 150
    },
    {
      "epoch": 0.33407572383073497,
      "loss": 2.6044678688049316,
      "loss_ce": 0.25729984045028687,
      "loss_iou": 0.84375,
      "loss_num": 0.1318359375,
      "loss_xval": 2.34375,
      "num_input_tokens_seen": 8372780,
      "step": 150
    },
    {
      "epoch": 0.3363028953229399,
      "grad_norm": 30.648853302001953,
      "learning_rate": 1e-06,
      "loss": 2.5253,
      "num_input_tokens_seen": 8426544,
      "step": 151
    },
    {
      "epoch": 0.3363028953229399,
      "loss": 2.3786497116088867,
      "loss_ce": 0.2819698452949524,
      "loss_iou": 0.7734375,
      "loss_num": 0.10986328125,
      "loss_xval": 2.09375,
      "num_input_tokens_seen": 8426544,
      "step": 151
    },
    {
      "epoch": 0.33853006681514475,
      "grad_norm": 41.66792297363281,
      "learning_rate": 1e-06,
      "loss": 2.3784,
      "num_input_tokens_seen": 8483968,
      "step": 152
    },
    {
      "epoch": 0.33853006681514475,
      "loss": 2.490370273590088,
      "loss_ce": 0.3272841274738312,
      "loss_iou": 0.79296875,
      "loss_num": 0.115234375,
      "loss_xval": 2.15625,
      "num_input_tokens_seen": 8483968,
      "step": 152
    },
    {
      "epoch": 0.34075723830734966,
      "grad_norm": 39.65934753417969,
      "learning_rate": 1e-06,
      "loss": 2.9073,
      "num_input_tokens_seen": 8537416,
      "step": 153
    },
    {
      "epoch": 0.34075723830734966,
      "loss": 3.037890911102295,
      "loss_ce": 0.40898463129997253,
      "loss_iou": 1.03125,
      "loss_num": 0.11328125,
      "loss_xval": 2.625,
      "num_input_tokens_seen": 8537416,
      "step": 153
    },
    {
      "epoch": 0.3429844097995546,
      "grad_norm": 43.49128341674805,
      "learning_rate": 1e-06,
      "loss": 2.6995,
      "num_input_tokens_seen": 8592392,
      "step": 154
    },
    {
      "epoch": 0.3429844097995546,
      "loss": 2.778602123260498,
      "loss_ce": 0.24051626026630402,
      "loss_iou": 0.8984375,
      "loss_num": 0.146484375,
      "loss_xval": 2.53125,
      "num_input_tokens_seen": 8592392,
      "step": 154
    },
    {
      "epoch": 0.34521158129175944,
      "grad_norm": 158.33724975585938,
      "learning_rate": 1e-06,
      "loss": 2.5526,
      "num_input_tokens_seen": 8647228,
      "step": 155
    },
    {
      "epoch": 0.34521158129175944,
      "loss": 2.642277717590332,
      "loss_ce": 0.29755133390426636,
      "loss_iou": 0.89453125,
      "loss_num": 0.11083984375,
      "loss_xval": 2.34375,
      "num_input_tokens_seen": 8647228,
      "step": 155
    },
    {
      "epoch": 0.34743875278396436,
      "grad_norm": 173.70913696289062,
      "learning_rate": 1e-06,
      "loss": 2.8128,
      "num_input_tokens_seen": 8701864,
      "step": 156
    },
    {
      "epoch": 0.34743875278396436,
      "loss": 2.8622255325317383,
      "loss_ce": 0.30753791332244873,
      "loss_iou": 0.921875,
      "loss_num": 0.1416015625,
      "loss_xval": 2.5625,
      "num_input_tokens_seen": 8701864,
      "step": 156
    },
    {
      "epoch": 0.34966592427616927,
      "grad_norm": 410.6438903808594,
      "learning_rate": 1e-06,
      "loss": 2.4924,
      "num_input_tokens_seen": 8759524,
      "step": 157
    },
    {
      "epoch": 0.34966592427616927,
      "loss": 2.3690528869628906,
      "loss_ce": 0.3465919494628906,
      "loss_iou": 0.7578125,
      "loss_num": 0.1025390625,
      "loss_xval": 2.015625,
      "num_input_tokens_seen": 8759524,
      "step": 157
    },
    {
      "epoch": 0.3518930957683742,
      "grad_norm": 40.04865646362305,
      "learning_rate": 1e-06,
      "loss": 2.43,
      "num_input_tokens_seen": 8814804,
      "step": 158
    },
    {
      "epoch": 0.3518930957683742,
      "loss": 2.7295665740966797,
      "loss_ce": 0.2725353240966797,
      "loss_iou": 0.94921875,
      "loss_num": 0.111328125,
      "loss_xval": 2.453125,
      "num_input_tokens_seen": 8814804,
      "step": 158
    },
    {
      "epoch": 0.35412026726057905,
      "grad_norm": 62.50952911376953,
      "learning_rate": 1e-06,
      "loss": 2.4288,
      "num_input_tokens_seen": 8870392,
      "step": 159
    },
    {
      "epoch": 0.35412026726057905,
      "loss": 2.16571307182312,
      "loss_ce": 0.24676772952079773,
      "loss_iou": 0.7734375,
      "loss_num": 0.07568359375,
      "loss_xval": 1.921875,
      "num_input_tokens_seen": 8870392,
      "step": 159
    },
    {
      "epoch": 0.35634743875278396,
      "grad_norm": 28.229555130004883,
      "learning_rate": 1e-06,
      "loss": 2.3698,
      "num_input_tokens_seen": 8927972,
      "step": 160
    },
    {
      "epoch": 0.35634743875278396,
      "loss": 2.608736753463745,
      "loss_ce": 0.24057263135910034,
      "loss_iou": 0.8828125,
      "loss_num": 0.12060546875,
      "loss_xval": 2.375,
      "num_input_tokens_seen": 8927972,
      "step": 160
    },
    {
      "epoch": 0.3585746102449889,
      "grad_norm": 21.421159744262695,
      "learning_rate": 1e-06,
      "loss": 2.3587,
      "num_input_tokens_seen": 8983156,
      "step": 161
    },
    {
      "epoch": 0.3585746102449889,
      "loss": 1.9863007068634033,
      "loss_ce": 0.13766781985759735,
      "loss_iou": 0.75390625,
      "loss_num": 0.0673828125,
      "loss_xval": 1.8515625,
      "num_input_tokens_seen": 8983156,
      "step": 161
    },
    {
      "epoch": 0.36080178173719374,
      "grad_norm": 23.95926284790039,
      "learning_rate": 1e-06,
      "loss": 2.4225,
      "num_input_tokens_seen": 9038652,
      "step": 162
    },
    {
      "epoch": 0.36080178173719374,
      "loss": 2.301337480545044,
      "loss_ce": 0.2510446012020111,
      "loss_iou": 0.7734375,
      "loss_num": 0.10009765625,
      "loss_xval": 2.046875,
      "num_input_tokens_seen": 9038652,
      "step": 162
    },
    {
      "epoch": 0.36302895322939865,
      "grad_norm": 47.72632598876953,
      "learning_rate": 1e-06,
      "loss": 2.5188,
      "num_input_tokens_seen": 9096844,
      "step": 163
    },
    {
      "epoch": 0.36302895322939865,
      "loss": 2.3073041439056396,
      "loss_ce": 0.20134715735912323,
      "loss_iou": 0.75,
      "loss_num": 0.12060546875,
      "loss_xval": 2.109375,
      "num_input_tokens_seen": 9096844,
      "step": 163
    },
    {
      "epoch": 0.36525612472160357,
      "grad_norm": 75.89527893066406,
      "learning_rate": 1e-06,
      "loss": 2.5799,
      "num_input_tokens_seen": 9151260,
      "step": 164
    },
    {
      "epoch": 0.36525612472160357,
      "loss": 2.6282100677490234,
      "loss_ce": 0.3332880735397339,
      "loss_iou": 0.87890625,
      "loss_num": 0.10693359375,
      "loss_xval": 2.296875,
      "num_input_tokens_seen": 9151260,
      "step": 164
    },
    {
      "epoch": 0.3674832962138085,
      "grad_norm": 44.563873291015625,
      "learning_rate": 1e-06,
      "loss": 2.5729,
      "num_input_tokens_seen": 9206156,
      "step": 165
    },
    {
      "epoch": 0.3674832962138085,
      "loss": 2.5874228477478027,
      "loss_ce": 0.30128994584083557,
      "loss_iou": 0.90625,
      "loss_num": 0.0947265625,
      "loss_xval": 2.28125,
      "num_input_tokens_seen": 9206156,
      "step": 165
    },
    {
      "epoch": 0.36971046770601335,
      "grad_norm": 57.38001251220703,
      "learning_rate": 1e-06,
      "loss": 2.588,
      "num_input_tokens_seen": 9259864,
      "step": 166
    },
    {
      "epoch": 0.36971046770601335,
      "loss": 2.448063373565674,
      "loss_ce": 0.2586103677749634,
      "loss_iou": 0.875,
      "loss_num": 0.08740234375,
      "loss_xval": 2.1875,
      "num_input_tokens_seen": 9259864,
      "step": 166
    },
    {
      "epoch": 0.37193763919821826,
      "grad_norm": 53.09568405151367,
      "learning_rate": 1e-06,
      "loss": 2.5351,
      "num_input_tokens_seen": 9317272,
      "step": 167
    },
    {
      "epoch": 0.37193763919821826,
      "loss": 2.441850185394287,
      "loss_ce": 0.3842327892780304,
      "loss_iou": 0.7890625,
      "loss_num": 0.0966796875,
      "loss_xval": 2.0625,
      "num_input_tokens_seen": 9317272,
      "step": 167
    },
    {
      "epoch": 0.3741648106904232,
      "grad_norm": 36.94442367553711,
      "learning_rate": 1e-06,
      "loss": 2.1125,
      "num_input_tokens_seen": 9372720,
      "step": 168
    },
    {
      "epoch": 0.3741648106904232,
      "loss": 2.203955888748169,
      "loss_ce": 0.25278398394584656,
      "loss_iou": 0.74609375,
      "loss_num": 0.091796875,
      "loss_xval": 1.953125,
      "num_input_tokens_seen": 9372720,
      "step": 168
    },
    {
      "epoch": 0.37639198218262804,
      "grad_norm": 61.23322296142578,
      "learning_rate": 1e-06,
      "loss": 2.5849,
      "num_input_tokens_seen": 9429244,
      "step": 169
    },
    {
      "epoch": 0.37639198218262804,
      "loss": 2.3757333755493164,
      "loss_ce": 0.22045986354351044,
      "loss_iou": 0.81640625,
      "loss_num": 0.1044921875,
      "loss_xval": 2.15625,
      "num_input_tokens_seen": 9429244,
      "step": 169
    },
    {
      "epoch": 0.37861915367483295,
      "grad_norm": 44.59577178955078,
      "learning_rate": 1e-06,
      "loss": 2.3184,
      "num_input_tokens_seen": 9485716,
      "step": 170
    },
    {
      "epoch": 0.37861915367483295,
      "loss": 2.2477307319641113,
      "loss_ce": 0.21745747327804565,
      "loss_iou": 0.83203125,
      "loss_num": 0.0732421875,
      "loss_xval": 2.03125,
      "num_input_tokens_seen": 9485716,
      "step": 170
    },
    {
      "epoch": 0.38084632516703787,
      "grad_norm": 26.56584358215332,
      "learning_rate": 1e-06,
      "loss": 2.3085,
      "num_input_tokens_seen": 9542880,
      "step": 171
    },
    {
      "epoch": 0.38084632516703787,
      "loss": 2.289217948913574,
      "loss_ce": 0.17203053832054138,
      "loss_iou": 0.859375,
      "loss_num": 0.080078125,
      "loss_xval": 2.125,
      "num_input_tokens_seen": 9542880,
      "step": 171
    },
    {
      "epoch": 0.3830734966592428,
      "grad_norm": 40.40160369873047,
      "learning_rate": 1e-06,
      "loss": 2.4104,
      "num_input_tokens_seen": 9599024,
      "step": 172
    },
    {
      "epoch": 0.3830734966592428,
      "loss": 2.4211387634277344,
      "loss_ce": 0.155513733625412,
      "loss_iou": 0.84765625,
      "loss_num": 0.11376953125,
      "loss_xval": 2.265625,
      "num_input_tokens_seen": 9599024,
      "step": 172
    },
    {
      "epoch": 0.38530066815144765,
      "grad_norm": 23.74785614013672,
      "learning_rate": 1e-06,
      "loss": 2.5009,
      "num_input_tokens_seen": 9651488,
      "step": 173
    },
    {
      "epoch": 0.38530066815144765,
      "loss": 2.5731005668640137,
      "loss_ce": 0.2664598822593689,
      "loss_iou": 0.890625,
      "loss_num": 0.1044921875,
      "loss_xval": 2.3125,
      "num_input_tokens_seen": 9651488,
      "step": 173
    },
    {
      "epoch": 0.38752783964365256,
      "grad_norm": 51.20659255981445,
      "learning_rate": 1e-06,
      "loss": 2.7359,
      "num_input_tokens_seen": 9708124,
      "step": 174
    },
    {
      "epoch": 0.38752783964365256,
      "loss": 2.863722324371338,
      "loss_ce": 0.31489402055740356,
      "loss_iou": 0.95703125,
      "loss_num": 0.126953125,
      "loss_xval": 2.546875,
      "num_input_tokens_seen": 9708124,
      "step": 174
    },
    {
      "epoch": 0.3897550111358575,
      "grad_norm": 18.966318130493164,
      "learning_rate": 1e-06,
      "loss": 2.3377,
      "num_input_tokens_seen": 9766496,
      "step": 175
    },
    {
      "epoch": 0.3897550111358575,
      "loss": 2.530661106109619,
      "loss_ce": 0.2992156147956848,
      "loss_iou": 0.8515625,
      "loss_num": 0.10546875,
      "loss_xval": 2.234375,
      "num_input_tokens_seen": 9766496,
      "step": 175
    },
    {
      "epoch": 0.39198218262806234,
      "grad_norm": 43.686466217041016,
      "learning_rate": 1e-06,
      "loss": 2.5387,
      "num_input_tokens_seen": 9823256,
      "step": 176
    },
    {
      "epoch": 0.39198218262806234,
      "loss": 2.393760919570923,
      "loss_ce": 0.18672963976860046,
      "loss_iou": 0.8203125,
      "loss_num": 0.11279296875,
      "loss_xval": 2.203125,
      "num_input_tokens_seen": 9823256,
      "step": 176
    },
    {
      "epoch": 0.39420935412026725,
      "grad_norm": 117.024658203125,
      "learning_rate": 1e-06,
      "loss": 2.5409,
      "num_input_tokens_seen": 9881900,
      "step": 177
    },
    {
      "epoch": 0.39420935412026725,
      "loss": 2.7967357635498047,
      "loss_ce": 0.2508372664451599,
      "loss_iou": 0.90234375,
      "loss_num": 0.1474609375,
      "loss_xval": 2.546875,
      "num_input_tokens_seen": 9881900,
      "step": 177
    },
    {
      "epoch": 0.39643652561247217,
      "grad_norm": 77.61784362792969,
      "learning_rate": 1e-06,
      "loss": 2.5997,
      "num_input_tokens_seen": 9937420,
      "step": 178
    },
    {
      "epoch": 0.39643652561247217,
      "loss": 2.7027268409729004,
      "loss_ce": 0.18905505537986755,
      "loss_iou": 0.9453125,
      "loss_num": 0.125,
      "loss_xval": 2.515625,
      "num_input_tokens_seen": 9937420,
      "step": 178
    },
    {
      "epoch": 0.3986636971046771,
      "grad_norm": 22.932493209838867,
      "learning_rate": 1e-06,
      "loss": 2.0248,
      "num_input_tokens_seen": 9994316,
      "step": 179
    },
    {
      "epoch": 0.3986636971046771,
      "loss": 2.1930716037750244,
      "loss_ce": 0.20479029417037964,
      "loss_iou": 0.76953125,
      "loss_num": 0.08984375,
      "loss_xval": 1.984375,
      "num_input_tokens_seen": 9994316,
      "step": 179
    },
    {
      "epoch": 0.40089086859688194,
      "grad_norm": 61.33550262451172,
      "learning_rate": 1e-06,
      "loss": 2.55,
      "num_input_tokens_seen": 10046516,
      "step": 180
    },
    {
      "epoch": 0.40089086859688194,
      "loss": 2.5461134910583496,
      "loss_ce": 0.25509777665138245,
      "loss_iou": 0.91796875,
      "loss_num": 0.0908203125,
      "loss_xval": 2.296875,
      "num_input_tokens_seen": 10046516,
      "step": 180
    },
    {
      "epoch": 0.40311804008908686,
      "grad_norm": 29.507160186767578,
      "learning_rate": 1e-06,
      "loss": 2.4409,
      "num_input_tokens_seen": 10103372,
      "step": 181
    },
    {
      "epoch": 0.40311804008908686,
      "loss": 2.6029810905456543,
      "loss_ce": 0.23091065883636475,
      "loss_iou": 0.8984375,
      "loss_num": 0.11572265625,
      "loss_xval": 2.375,
      "num_input_tokens_seen": 10103372,
      "step": 181
    },
    {
      "epoch": 0.4053452115812918,
      "grad_norm": 90.82682037353516,
      "learning_rate": 1e-06,
      "loss": 2.5212,
      "num_input_tokens_seen": 10159204,
      "step": 182
    },
    {
      "epoch": 0.4053452115812918,
      "loss": 2.695434331893921,
      "loss_ce": 0.3145750164985657,
      "loss_iou": 0.8828125,
      "loss_num": 0.123046875,
      "loss_xval": 2.375,
      "num_input_tokens_seen": 10159204,
      "step": 182
    },
    {
      "epoch": 0.40757238307349664,
      "grad_norm": 27.87862205505371,
      "learning_rate": 1e-06,
      "loss": 2.2165,
      "num_input_tokens_seen": 10214768,
      "step": 183
    },
    {
      "epoch": 0.40757238307349664,
      "loss": 2.1518983840942383,
      "loss_ce": 0.1807069331407547,
      "loss_iou": 0.73828125,
      "loss_num": 0.09912109375,
      "loss_xval": 1.96875,
      "num_input_tokens_seen": 10214768,
      "step": 183
    },
    {
      "epoch": 0.40979955456570155,
      "grad_norm": 72.37561798095703,
      "learning_rate": 1e-06,
      "loss": 2.2339,
      "num_input_tokens_seen": 10270868,
      "step": 184
    },
    {
      "epoch": 0.40979955456570155,
      "loss": 2.1870460510253906,
      "loss_ce": 0.20511233806610107,
      "loss_iou": 0.7734375,
      "loss_num": 0.08642578125,
      "loss_xval": 1.984375,
      "num_input_tokens_seen": 10270868,
      "step": 184
    },
    {
      "epoch": 0.41202672605790647,
      "grad_norm": 22.659595489501953,
      "learning_rate": 1e-06,
      "loss": 2.4928,
      "num_input_tokens_seen": 10322388,
      "step": 185
    },
    {
      "epoch": 0.41202672605790647,
      "loss": 2.4523706436157227,
      "loss_ce": 0.2468043565750122,
      "loss_iou": 0.83984375,
      "loss_num": 0.1044921875,
      "loss_xval": 2.203125,
      "num_input_tokens_seen": 10322388,
      "step": 185
    },
    {
      "epoch": 0.4142538975501114,
      "grad_norm": 27.231279373168945,
      "learning_rate": 1e-06,
      "loss": 2.2191,
      "num_input_tokens_seen": 10379116,
      "step": 186
    },
    {
      "epoch": 0.4142538975501114,
      "loss": 2.0813474655151367,
      "loss_ce": 0.14189457893371582,
      "loss_iou": 0.765625,
      "loss_num": 0.08203125,
      "loss_xval": 1.9375,
      "num_input_tokens_seen": 10379116,
      "step": 186
    },
    {
      "epoch": 0.41648106904231624,
      "grad_norm": 20.43091583251953,
      "learning_rate": 1e-06,
      "loss": 2.2453,
      "num_input_tokens_seen": 10434552,
      "step": 187
    },
    {
      "epoch": 0.41648106904231624,
      "loss": 2.5419094562530518,
      "loss_ce": 0.25821810960769653,
      "loss_iou": 0.796875,
      "loss_num": 0.138671875,
      "loss_xval": 2.28125,
      "num_input_tokens_seen": 10434552,
      "step": 187
    },
    {
      "epoch": 0.41870824053452116,
      "grad_norm": 25.056270599365234,
      "learning_rate": 1e-06,
      "loss": 2.2543,
      "num_input_tokens_seen": 10493740,
      "step": 188
    },
    {
      "epoch": 0.41870824053452116,
      "loss": 2.3587117195129395,
      "loss_ce": 0.16730527579784393,
      "loss_iou": 0.8203125,
      "loss_num": 0.109375,
      "loss_xval": 2.1875,
      "num_input_tokens_seen": 10493740,
      "step": 188
    },
    {
      "epoch": 0.4209354120267261,
      "grad_norm": 26.688255310058594,
      "learning_rate": 1e-06,
      "loss": 2.2537,
      "num_input_tokens_seen": 10548228,
      "step": 189
    },
    {
      "epoch": 0.4209354120267261,
      "loss": 2.1556687355041504,
      "loss_ce": 0.173246830701828,
      "loss_iou": 0.7890625,
      "loss_num": 0.08203125,
      "loss_xval": 1.984375,
      "num_input_tokens_seen": 10548228,
      "step": 189
    },
    {
      "epoch": 0.42316258351893093,
      "grad_norm": 56.197242736816406,
      "learning_rate": 1e-06,
      "loss": 2.0274,
      "num_input_tokens_seen": 10604476,
      "step": 190
    },
    {
      "epoch": 0.42316258351893093,
      "loss": 2.1501314640045166,
      "loss_ce": 0.188217431306839,
      "loss_iou": 0.734375,
      "loss_num": 0.09716796875,
      "loss_xval": 1.9609375,
      "num_input_tokens_seen": 10604476,
      "step": 190
    },
    {
      "epoch": 0.42538975501113585,
      "grad_norm": 32.006675720214844,
      "learning_rate": 1e-06,
      "loss": 2.4196,
      "num_input_tokens_seen": 10660088,
      "step": 191
    },
    {
      "epoch": 0.42538975501113585,
      "loss": 2.4907290935516357,
      "loss_ce": 0.21826806664466858,
      "loss_iou": 0.859375,
      "loss_num": 0.1103515625,
      "loss_xval": 2.265625,
      "num_input_tokens_seen": 10660088,
      "step": 191
    },
    {
      "epoch": 0.42761692650334077,
      "grad_norm": 23.622976303100586,
      "learning_rate": 1e-06,
      "loss": 1.8407,
      "num_input_tokens_seen": 10719304,
      "step": 192
    },
    {
      "epoch": 0.42761692650334077,
      "loss": 1.853776454925537,
      "loss_ce": 0.0735030248761177,
      "loss_iou": 0.76171875,
      "loss_num": 0.05078125,
      "loss_xval": 1.78125,
      "num_input_tokens_seen": 10719304,
      "step": 192
    },
    {
      "epoch": 0.4298440979955457,
      "grad_norm": 132.57778930664062,
      "learning_rate": 1e-06,
      "loss": 1.9273,
      "num_input_tokens_seen": 10776824,
      "step": 193
    },
    {
      "epoch": 0.4298440979955457,
      "loss": 2.2028584480285645,
      "loss_ce": 0.19748730957508087,
      "loss_iou": 0.74609375,
      "loss_num": 0.10302734375,
      "loss_xval": 2.0,
      "num_input_tokens_seen": 10776824,
      "step": 193
    },
    {
      "epoch": 0.43207126948775054,
      "grad_norm": 78.81751251220703,
      "learning_rate": 1e-06,
      "loss": 2.282,
      "num_input_tokens_seen": 10832888,
      "step": 194
    },
    {
      "epoch": 0.43207126948775054,
      "loss": 2.0263118743896484,
      "loss_ce": 0.08978863805532455,
      "loss_iou": 0.78515625,
      "loss_num": 0.0732421875,
      "loss_xval": 1.9375,
      "num_input_tokens_seen": 10832888,
      "step": 194
    },
    {
      "epoch": 0.43429844097995546,
      "grad_norm": 33.7293586730957,
      "learning_rate": 1e-06,
      "loss": 1.9715,
      "num_input_tokens_seen": 10889816,
      "step": 195
    },
    {
      "epoch": 0.43429844097995546,
      "loss": 1.9463962316513062,
      "loss_ce": 0.15049774944782257,
      "loss_iou": 0.71484375,
      "loss_num": 0.07373046875,
      "loss_xval": 1.796875,
      "num_input_tokens_seen": 10889816,
      "step": 195
    },
    {
      "epoch": 0.4365256124721604,
      "grad_norm": 42.51150131225586,
      "learning_rate": 1e-06,
      "loss": 2.2224,
      "num_input_tokens_seen": 10947644,
      "step": 196
    },
    {
      "epoch": 0.4365256124721604,
      "loss": 2.052577495574951,
      "loss_ce": 0.1531633585691452,
      "loss_iou": 0.7734375,
      "loss_num": 0.0703125,
      "loss_xval": 1.8984375,
      "num_input_tokens_seen": 10947644,
      "step": 196
    },
    {
      "epoch": 0.43875278396436523,
      "grad_norm": 21.027219772338867,
      "learning_rate": 1e-06,
      "loss": 2.0492,
      "num_input_tokens_seen": 11004624,
      "step": 197
    },
    {
      "epoch": 0.43875278396436523,
      "loss": 2.039621353149414,
      "loss_ce": 0.1343478262424469,
      "loss_iou": 0.77734375,
      "loss_num": 0.0703125,
      "loss_xval": 1.90625,
      "num_input_tokens_seen": 11004624,
      "step": 197
    },
    {
      "epoch": 0.44097995545657015,
      "grad_norm": 30.52896499633789,
      "learning_rate": 1e-06,
      "loss": 2.0472,
      "num_input_tokens_seen": 11060372,
      "step": 198
    },
    {
      "epoch": 0.44097995545657015,
      "loss": 2.0091304779052734,
      "loss_ce": 0.12045860290527344,
      "loss_iou": 0.76953125,
      "loss_num": 0.0703125,
      "loss_xval": 1.890625,
      "num_input_tokens_seen": 11060372,
      "step": 198
    },
    {
      "epoch": 0.44320712694877507,
      "grad_norm": 40.47418975830078,
      "learning_rate": 1e-06,
      "loss": 2.1817,
      "num_input_tokens_seen": 11117700,
      "step": 199
    },
    {
      "epoch": 0.44320712694877507,
      "loss": 2.1524715423583984,
      "loss_ce": 0.1846981942653656,
      "loss_iou": 0.75,
      "loss_num": 0.09375,
      "loss_xval": 1.96875,
      "num_input_tokens_seen": 11117700,
      "step": 199
    },
    {
      "epoch": 0.44543429844098,
      "grad_norm": 42.97262191772461,
      "learning_rate": 1e-06,
      "loss": 2.1472,
      "num_input_tokens_seen": 11176308,
      "step": 200
    },
    {
      "epoch": 0.44543429844098,
      "loss": 2.278196096420288,
      "loss_ce": 0.12389924377202988,
      "loss_iou": 0.8359375,
      "loss_num": 0.09716796875,
      "loss_xval": 2.15625,
      "num_input_tokens_seen": 11176308,
      "step": 200
    },
    {
      "epoch": 0.44766146993318484,
      "grad_norm": 46.24988555908203,
      "learning_rate": 1e-06,
      "loss": 2.2836,
      "num_input_tokens_seen": 11232984,
      "step": 201
    },
    {
      "epoch": 0.44766146993318484,
      "loss": 2.1692562103271484,
      "loss_ce": 0.16632673144340515,
      "loss_iou": 0.76953125,
      "loss_num": 0.09375,
      "loss_xval": 2.0,
      "num_input_tokens_seen": 11232984,
      "step": 201
    },
    {
      "epoch": 0.44988864142538976,
      "grad_norm": 19.978219985961914,
      "learning_rate": 1e-06,
      "loss": 1.9362,
      "num_input_tokens_seen": 11289736,
      "step": 202
    },
    {
      "epoch": 0.44988864142538976,
      "loss": 2.002622127532959,
      "loss_ce": 0.14520032703876495,
      "loss_iou": 0.75,
      "loss_num": 0.07080078125,
      "loss_xval": 1.859375,
      "num_input_tokens_seen": 11289736,
      "step": 202
    },
    {
      "epoch": 0.4521158129175947,
      "grad_norm": 40.11283493041992,
      "learning_rate": 1e-06,
      "loss": 2.0214,
      "num_input_tokens_seen": 11347688,
      "step": 203
    },
    {
      "epoch": 0.4521158129175947,
      "loss": 2.089003801345825,
      "loss_ce": 0.12367182970046997,
      "loss_iou": 0.76171875,
      "loss_num": 0.087890625,
      "loss_xval": 1.96875,
      "num_input_tokens_seen": 11347688,
      "step": 203
    },
    {
      "epoch": 0.45434298440979953,
      "grad_norm": 45.47671890258789,
      "learning_rate": 1e-06,
      "loss": 1.7514,
      "num_input_tokens_seen": 11405500,
      "step": 204
    },
    {
      "epoch": 0.45434298440979953,
      "loss": 1.679763913154602,
      "loss_ce": 0.08992011845111847,
      "loss_iou": 0.6328125,
      "loss_num": 0.06494140625,
      "loss_xval": 1.59375,
      "num_input_tokens_seen": 11405500,
      "step": 204
    },
    {
      "epoch": 0.45657015590200445,
      "grad_norm": 37.67365646362305,
      "learning_rate": 1e-06,
      "loss": 2.1156,
      "num_input_tokens_seen": 11463076,
      "step": 205
    },
    {
      "epoch": 0.45657015590200445,
      "loss": 1.912656545639038,
      "loss_ce": 0.1499611884355545,
      "loss_iou": 0.63671875,
      "loss_num": 0.09814453125,
      "loss_xval": 1.765625,
      "num_input_tokens_seen": 11463076,
      "step": 205
    },
    {
      "epoch": 0.45879732739420936,
      "grad_norm": 22.140178680419922,
      "learning_rate": 1e-06,
      "loss": 2.2171,
      "num_input_tokens_seen": 11517904,
      "step": 206
    },
    {
      "epoch": 0.45879732739420936,
      "loss": 2.3471362590789795,
      "loss_ce": 0.13815191388130188,
      "loss_iou": 0.875,
      "loss_num": 0.0908203125,
      "loss_xval": 2.203125,
      "num_input_tokens_seen": 11517904,
      "step": 206
    },
    {
      "epoch": 0.4610244988864143,
      "grad_norm": 42.47764205932617,
      "learning_rate": 1e-06,
      "loss": 1.9804,
      "num_input_tokens_seen": 11572968,
      "step": 207
    },
    {
      "epoch": 0.4610244988864143,
      "loss": 2.3084888458251953,
      "loss_ce": 0.2538011372089386,
      "loss_iou": 0.796875,
      "loss_num": 0.09228515625,
      "loss_xval": 2.0625,
      "num_input_tokens_seen": 11572968,
      "step": 207
    },
    {
      "epoch": 0.46325167037861914,
      "grad_norm": 32.766109466552734,
      "learning_rate": 1e-06,
      "loss": 1.8334,
      "num_input_tokens_seen": 11629228,
      "step": 208
    },
    {
      "epoch": 0.46325167037861914,
      "loss": 1.8067898750305176,
      "loss_ce": 0.10049097239971161,
      "loss_iou": 0.67578125,
      "loss_num": 0.07177734375,
      "loss_xval": 1.703125,
      "num_input_tokens_seen": 11629228,
      "step": 208
    },
    {
      "epoch": 0.46547884187082406,
      "grad_norm": 21.969463348388672,
      "learning_rate": 1e-06,
      "loss": 1.7864,
      "num_input_tokens_seen": 11687208,
      "step": 209
    },
    {
      "epoch": 0.46547884187082406,
      "loss": 1.507708191871643,
      "loss_ce": 0.09462223947048187,
      "loss_iou": 0.56640625,
      "loss_num": 0.055419921875,
      "loss_xval": 1.4140625,
      "num_input_tokens_seen": 11687208,
      "step": 209
    },
    {
      "epoch": 0.46770601336302897,
      "grad_norm": 66.28128814697266,
      "learning_rate": 1e-06,
      "loss": 2.1212,
      "num_input_tokens_seen": 11744372,
      "step": 210
    },
    {
      "epoch": 0.46770601336302897,
      "loss": 1.94266939163208,
      "loss_ce": 0.18949554860591888,
      "loss_iou": 0.66015625,
      "loss_num": 0.0869140625,
      "loss_xval": 1.75,
      "num_input_tokens_seen": 11744372,
      "step": 210
    },
    {
      "epoch": 0.46993318485523383,
      "grad_norm": 27.227609634399414,
      "learning_rate": 1e-06,
      "loss": 2.1373,
      "num_input_tokens_seen": 11800196,
      "step": 211
    },
    {
      "epoch": 0.46993318485523383,
      "loss": 2.1795947551727295,
      "loss_ce": 0.12490727007389069,
      "loss_iou": 0.79296875,
      "loss_num": 0.09375,
      "loss_xval": 2.0625,
      "num_input_tokens_seen": 11800196,
      "step": 211
    },
    {
      "epoch": 0.47216035634743875,
      "grad_norm": 27.855268478393555,
      "learning_rate": 1e-06,
      "loss": 2.2184,
      "num_input_tokens_seen": 11857088,
      "step": 212
    },
    {
      "epoch": 0.47216035634743875,
      "loss": 2.045684337615967,
      "loss_ce": 0.13650476932525635,
      "loss_iou": 0.765625,
      "loss_num": 0.076171875,
      "loss_xval": 1.90625,
      "num_input_tokens_seen": 11857088,
      "step": 212
    },
    {
      "epoch": 0.47438752783964366,
      "grad_norm": 18.745811462402344,
      "learning_rate": 1e-06,
      "loss": 1.8398,
      "num_input_tokens_seen": 11908736,
      "step": 213
    },
    {
      "epoch": 0.47438752783964366,
      "loss": 1.898270845413208,
      "loss_ce": 0.1253216713666916,
      "loss_iou": 0.71875,
      "loss_num": 0.06689453125,
      "loss_xval": 1.7734375,
      "num_input_tokens_seen": 11908736,
      "step": 213
    },
    {
      "epoch": 0.4766146993318486,
      "grad_norm": 50.83546447753906,
      "learning_rate": 1e-06,
      "loss": 1.8401,
      "num_input_tokens_seen": 11963016,
      "step": 214
    },
    {
      "epoch": 0.4766146993318486,
      "loss": 1.990050196647644,
      "loss_ce": 0.1697378158569336,
      "loss_iou": 0.671875,
      "loss_num": 0.0947265625,
      "loss_xval": 1.8203125,
      "num_input_tokens_seen": 11963016,
      "step": 214
    },
    {
      "epoch": 0.47884187082405344,
      "grad_norm": 15.607227325439453,
      "learning_rate": 1e-06,
      "loss": 2.0483,
      "num_input_tokens_seen": 12019844,
      "step": 215
    },
    {
      "epoch": 0.47884187082405344,
      "loss": 1.621078610420227,
      "loss_ce": 0.10240183770656586,
      "loss_iou": 0.6328125,
      "loss_num": 0.05078125,
      "loss_xval": 1.515625,
      "num_input_tokens_seen": 12019844,
      "step": 215
    },
    {
      "epoch": 0.48106904231625836,
      "grad_norm": 561.2518920898438,
      "learning_rate": 1e-06,
      "loss": 2.0352,
      "num_input_tokens_seen": 12079540,
      "step": 216
    },
    {
      "epoch": 0.48106904231625836,
      "loss": 2.1394166946411133,
      "loss_ce": 0.18922138214111328,
      "loss_iou": 0.71484375,
      "loss_num": 0.1044921875,
      "loss_xval": 1.953125,
      "num_input_tokens_seen": 12079540,
      "step": 216
    },
    {
      "epoch": 0.48329621380846327,
      "grad_norm": 75.96478271484375,
      "learning_rate": 1e-06,
      "loss": 1.9004,
      "num_input_tokens_seen": 12136556,
      "step": 217
    },
    {
      "epoch": 0.48329621380846327,
      "loss": 2.064810276031494,
      "loss_ce": 0.13121652603149414,
      "loss_iou": 0.72265625,
      "loss_num": 0.09765625,
      "loss_xval": 1.9375,
      "num_input_tokens_seen": 12136556,
      "step": 217
    },
    {
      "epoch": 0.48552338530066813,
      "grad_norm": 21.790206909179688,
      "learning_rate": 1e-06,
      "loss": 1.9754,
      "num_input_tokens_seen": 12190060,
      "step": 218
    },
    {
      "epoch": 0.48552338530066813,
      "loss": 2.044926404953003,
      "loss_ce": 0.13574674725532532,
      "loss_iou": 0.70703125,
      "loss_num": 0.09912109375,
      "loss_xval": 1.90625,
      "num_input_tokens_seen": 12190060,
      "step": 218
    },
    {
      "epoch": 0.48775055679287305,
      "grad_norm": 127.96965026855469,
      "learning_rate": 1e-06,
      "loss": 2.0309,
      "num_input_tokens_seen": 12247900,
      "step": 219
    },
    {
      "epoch": 0.48775055679287305,
      "loss": 2.0088348388671875,
      "loss_ce": 0.12211604416370392,
      "loss_iou": 0.74609375,
      "loss_num": 0.0791015625,
      "loss_xval": 1.890625,
      "num_input_tokens_seen": 12247900,
      "step": 219
    },
    {
      "epoch": 0.48997772828507796,
      "grad_norm": 37.08074951171875,
      "learning_rate": 1e-06,
      "loss": 1.8653,
      "num_input_tokens_seen": 12303552,
      "step": 220
    },
    {
      "epoch": 0.48997772828507796,
      "loss": 1.8836941719055176,
      "loss_ce": 0.10244403779506683,
      "loss_iou": 0.72265625,
      "loss_num": 0.0673828125,
      "loss_xval": 1.78125,
      "num_input_tokens_seen": 12303552,
      "step": 220
    },
    {
      "epoch": 0.4922048997772829,
      "grad_norm": 32.377262115478516,
      "learning_rate": 1e-06,
      "loss": 1.8778,
      "num_input_tokens_seen": 12359432,
      "step": 221
    },
    {
      "epoch": 0.4922048997772829,
      "loss": 1.9408328533172607,
      "loss_ce": 0.10782508552074432,
      "loss_iou": 0.7421875,
      "loss_num": 0.06884765625,
      "loss_xval": 1.8359375,
      "num_input_tokens_seen": 12359432,
      "step": 221
    },
    {
      "epoch": 0.49443207126948774,
      "grad_norm": 45.29218673706055,
      "learning_rate": 1e-06,
      "loss": 2.1275,
      "num_input_tokens_seen": 12415808,
      "step": 222
    },
    {
      "epoch": 0.49443207126948774,
      "loss": 2.1390140056610107,
      "loss_ce": 0.1058109700679779,
      "loss_iou": 0.77734375,
      "loss_num": 0.09619140625,
      "loss_xval": 2.03125,
      "num_input_tokens_seen": 12415808,
      "step": 222
    },
    {
      "epoch": 0.49665924276169265,
      "grad_norm": 23.249635696411133,
      "learning_rate": 1e-06,
      "loss": 2.2482,
      "num_input_tokens_seen": 12468328,
      "step": 223
    },
    {
      "epoch": 0.49665924276169265,
      "loss": 2.2763051986694336,
      "loss_ce": 0.13763317465782166,
      "loss_iou": 0.8125,
      "loss_num": 0.10302734375,
      "loss_xval": 2.140625,
      "num_input_tokens_seen": 12468328,
      "step": 223
    },
    {
      "epoch": 0.49888641425389757,
      "grad_norm": 26.10118293762207,
      "learning_rate": 1e-06,
      "loss": 1.745,
      "num_input_tokens_seen": 12526672,
      "step": 224
    },
    {
      "epoch": 0.49888641425389757,
      "loss": 1.4877617359161377,
      "loss_ce": 0.06539853662252426,
      "loss_iou": 0.57421875,
      "loss_num": 0.055419921875,
      "loss_xval": 1.421875,
      "num_input_tokens_seen": 12526672,
      "step": 224
    },
    {
      "epoch": 0.5011135857461024,
      "grad_norm": 28.74011993408203,
      "learning_rate": 1e-06,
      "loss": 1.8918,
      "num_input_tokens_seen": 12582496,
      "step": 225
    },
    {
      "epoch": 0.5011135857461024,
      "loss": 2.0152716636657715,
      "loss_ce": 0.0636114627122879,
      "loss_iou": 0.77734375,
      "loss_num": 0.0791015625,
      "loss_xval": 1.953125,
      "num_input_tokens_seen": 12582496,
      "step": 225
    },
    {
      "epoch": 0.5033407572383074,
      "grad_norm": 39.25506591796875,
      "learning_rate": 1e-06,
      "loss": 2.072,
      "num_input_tokens_seen": 12637180,
      "step": 226
    },
    {
      "epoch": 0.5033407572383074,
      "loss": 2.0042176246643066,
      "loss_ce": 0.0706239864230156,
      "loss_iou": 0.76171875,
      "loss_num": 0.08154296875,
      "loss_xval": 1.9375,
      "num_input_tokens_seen": 12637180,
      "step": 226
    },
    {
      "epoch": 0.5055679287305123,
      "grad_norm": 58.19468307495117,
      "learning_rate": 1e-06,
      "loss": 1.9587,
      "num_input_tokens_seen": 12696256,
      "step": 227
    },
    {
      "epoch": 0.5055679287305123,
      "loss": 2.141087532043457,
      "loss_ce": 0.10886099934577942,
      "loss_iou": 0.80078125,
      "loss_num": 0.08642578125,
      "loss_xval": 2.03125,
      "num_input_tokens_seen": 12696256,
      "step": 227
    },
    {
      "epoch": 0.5077951002227171,
      "grad_norm": 26.209997177124023,
      "learning_rate": 1e-06,
      "loss": 2.0101,
      "num_input_tokens_seen": 12751124,
      "step": 228
    },
    {
      "epoch": 0.5077951002227171,
      "loss": 1.975238561630249,
      "loss_ce": 0.10316816717386246,
      "loss_iou": 0.76953125,
      "loss_num": 0.06689453125,
      "loss_xval": 1.875,
      "num_input_tokens_seen": 12751124,
      "step": 228
    },
    {
      "epoch": 0.5100222717149221,
      "grad_norm": 29.183605194091797,
      "learning_rate": 1e-06,
      "loss": 1.8975,
      "num_input_tokens_seen": 12807368,
      "step": 229
    },
    {
      "epoch": 0.5100222717149221,
      "loss": 2.0033857822418213,
      "loss_ce": 0.1225263923406601,
      "loss_iou": 0.71875,
      "loss_num": 0.08837890625,
      "loss_xval": 1.8828125,
      "num_input_tokens_seen": 12807368,
      "step": 229
    },
    {
      "epoch": 0.512249443207127,
      "grad_norm": 32.19590377807617,
      "learning_rate": 1e-06,
      "loss": 1.9127,
      "num_input_tokens_seen": 12866044,
      "step": 230
    },
    {
      "epoch": 0.512249443207127,
      "loss": 1.722662329673767,
      "loss_ce": 0.10645144432783127,
      "loss_iou": 0.63671875,
      "loss_num": 0.068359375,
      "loss_xval": 1.6171875,
      "num_input_tokens_seen": 12866044,
      "step": 230
    },
    {
      "epoch": 0.5144766146993318,
      "grad_norm": 88.9583740234375,
      "learning_rate": 1e-06,
      "loss": 1.8966,
      "num_input_tokens_seen": 12919828,
      "step": 231
    },
    {
      "epoch": 0.5144766146993318,
      "loss": 1.7427191734313965,
      "loss_ce": 0.08256293088197708,
      "loss_iou": 0.66796875,
      "loss_num": 0.06494140625,
      "loss_xval": 1.65625,
      "num_input_tokens_seen": 12919828,
      "step": 231
    },
    {
      "epoch": 0.5167037861915368,
      "grad_norm": 38.6852912902832,
      "learning_rate": 1e-06,
      "loss": 1.8535,
      "num_input_tokens_seen": 12976452,
      "step": 232
    },
    {
      "epoch": 0.5167037861915368,
      "loss": 1.5481868982315063,
      "loss_ce": 0.09603846073150635,
      "loss_iou": 0.60546875,
      "loss_num": 0.04833984375,
      "loss_xval": 1.453125,
      "num_input_tokens_seen": 12976452,
      "step": 232
    },
    {
      "epoch": 0.5189309576837416,
      "grad_norm": 40.1108512878418,
      "learning_rate": 1e-06,
      "loss": 1.7714,
      "num_input_tokens_seen": 13032624,
      "step": 233
    },
    {
      "epoch": 0.5189309576837416,
      "loss": 1.8302245140075684,
      "loss_ce": 0.043115146458148956,
      "loss_iou": 0.765625,
      "loss_num": 0.05126953125,
      "loss_xval": 1.7890625,
      "num_input_tokens_seen": 13032624,
      "step": 233
    },
    {
      "epoch": 0.5211581291759465,
      "grad_norm": 35.25422286987305,
      "learning_rate": 1e-06,
      "loss": 1.9139,
      "num_input_tokens_seen": 13089524,
      "step": 234
    },
    {
      "epoch": 0.5211581291759465,
      "loss": 2.1705551147460938,
      "loss_ce": 0.11000814288854599,
      "loss_iou": 0.8203125,
      "loss_num": 0.083984375,
      "loss_xval": 2.0625,
      "num_input_tokens_seen": 13089524,
      "step": 234
    },
    {
      "epoch": 0.5233853006681515,
      "grad_norm": 92.00837707519531,
      "learning_rate": 1e-06,
      "loss": 2.0319,
      "num_input_tokens_seen": 13146916,
      "step": 235
    },
    {
      "epoch": 0.5233853006681515,
      "loss": 1.8411858081817627,
      "loss_ce": 0.04724044352769852,
      "loss_iou": 0.7578125,
      "loss_num": 0.05517578125,
      "loss_xval": 1.796875,
      "num_input_tokens_seen": 13146916,
      "step": 235
    },
    {
      "epoch": 0.5256124721603563,
      "grad_norm": 61.55561065673828,
      "learning_rate": 1e-06,
      "loss": 1.938,
      "num_input_tokens_seen": 13206136,
      "step": 236
    },
    {
      "epoch": 0.5256124721603563,
      "loss": 1.859898328781128,
      "loss_ce": 0.0937848836183548,
      "loss_iou": 0.6875,
      "loss_num": 0.078125,
      "loss_xval": 1.765625,
      "num_input_tokens_seen": 13206136,
      "step": 236
    },
    {
      "epoch": 0.5278396436525612,
      "grad_norm": 24.729263305664062,
      "learning_rate": 1e-06,
      "loss": 1.6522,
      "num_input_tokens_seen": 13262240,
      "step": 237
    },
    {
      "epoch": 0.5278396436525612,
      "loss": 1.7375231981277466,
      "loss_ce": 0.05246463418006897,
      "loss_iou": 0.671875,
      "loss_num": 0.06884765625,
      "loss_xval": 1.6875,
      "num_input_tokens_seen": 13262240,
      "step": 237
    },
    {
      "epoch": 0.5300668151447662,
      "grad_norm": 46.568321228027344,
      "learning_rate": 1e-06,
      "loss": 1.8631,
      "num_input_tokens_seen": 13317024,
      "step": 238
    },
    {
      "epoch": 0.5300668151447662,
      "loss": 1.6511387825012207,
      "loss_ce": 0.043716952204704285,
      "loss_iou": 0.66796875,
      "loss_num": 0.05517578125,
      "loss_xval": 1.609375,
      "num_input_tokens_seen": 13317024,
      "step": 238
    },
    {
      "epoch": 0.532293986636971,
      "grad_norm": 433.0378112792969,
      "learning_rate": 1e-06,
      "loss": 1.9798,
      "num_input_tokens_seen": 13370596,
      "step": 239
    },
    {
      "epoch": 0.532293986636971,
      "loss": 1.9186971187591553,
      "loss_ce": 0.052486199885606766,
      "loss_iou": 0.7421875,
      "loss_num": 0.076171875,
      "loss_xval": 1.8671875,
      "num_input_tokens_seen": 13370596,
      "step": 239
    },
    {
      "epoch": 0.534521158129176,
      "grad_norm": 25.267301559448242,
      "learning_rate": 1e-06,
      "loss": 1.6908,
      "num_input_tokens_seen": 13427652,
      "step": 240
    },
    {
      "epoch": 0.534521158129176,
      "loss": 1.6441099643707275,
      "loss_ce": 0.04157081991434097,
      "loss_iou": 0.625,
      "loss_num": 0.06982421875,
      "loss_xval": 1.6015625,
      "num_input_tokens_seen": 13427652,
      "step": 240
    },
    {
      "epoch": 0.5367483296213809,
      "grad_norm": 23.9971981048584,
      "learning_rate": 1e-06,
      "loss": 1.944,
      "num_input_tokens_seen": 13484652,
      "step": 241
    },
    {
      "epoch": 0.5367483296213809,
      "loss": 1.9427827596664429,
      "loss_ce": 0.0628998875617981,
      "loss_iou": 0.78515625,
      "loss_num": 0.061279296875,
      "loss_xval": 1.8828125,
      "num_input_tokens_seen": 13484652,
      "step": 241
    },
    {
      "epoch": 0.5389755011135857,
      "grad_norm": 28.519479751586914,
      "learning_rate": 1e-06,
      "loss": 1.7093,
      "num_input_tokens_seen": 13539076,
      "step": 242
    },
    {
      "epoch": 0.5389755011135857,
      "loss": 1.5857552289962769,
      "loss_ce": 0.030091160908341408,
      "loss_iou": 0.61328125,
      "loss_num": 0.06494140625,
      "loss_xval": 1.5546875,
      "num_input_tokens_seen": 13539076,
      "step": 242
    },
    {
      "epoch": 0.5412026726057907,
      "grad_norm": 25.626075744628906,
      "learning_rate": 1e-06,
      "loss": 1.8502,
      "num_input_tokens_seen": 13596724,
      "step": 243
    },
    {
      "epoch": 0.5412026726057907,
      "loss": 1.6758081912994385,
      "loss_ce": 0.09328873455524445,
      "loss_iou": 0.6640625,
      "loss_num": 0.05029296875,
      "loss_xval": 1.5859375,
      "num_input_tokens_seen": 13596724,
      "step": 243
    },
    {
      "epoch": 0.5434298440979956,
      "grad_norm": 41.99237823486328,
      "learning_rate": 1e-06,
      "loss": 1.6974,
      "num_input_tokens_seen": 13652844,
      "step": 244
    },
    {
      "epoch": 0.5434298440979956,
      "loss": 1.7878342866897583,
      "loss_ce": 0.05052957311272621,
      "loss_iou": 0.7421875,
      "loss_num": 0.05029296875,
      "loss_xval": 1.734375,
      "num_input_tokens_seen": 13652844,
      "step": 244
    },
    {
      "epoch": 0.5456570155902004,
      "grad_norm": 36.3553352355957,
      "learning_rate": 1e-06,
      "loss": 1.9218,
      "num_input_tokens_seen": 13707612,
      "step": 245
    },
    {
      "epoch": 0.5456570155902004,
      "loss": 1.8159754276275635,
      "loss_ce": 0.04546765610575676,
      "loss_iou": 0.76171875,
      "loss_num": 0.04931640625,
      "loss_xval": 1.7734375,
      "num_input_tokens_seen": 13707612,
      "step": 245
    },
    {
      "epoch": 0.5478841870824054,
      "grad_norm": 25.488935470581055,
      "learning_rate": 1e-06,
      "loss": 1.7327,
      "num_input_tokens_seen": 13764768,
      "step": 246
    },
    {
      "epoch": 0.5478841870824054,
      "loss": 1.66471266746521,
      "loss_ce": 0.040689267218112946,
      "loss_iou": 0.65625,
      "loss_num": 0.06298828125,
      "loss_xval": 1.625,
      "num_input_tokens_seen": 13764768,
      "step": 246
    },
    {
      "epoch": 0.5501113585746102,
      "grad_norm": 43.198577880859375,
      "learning_rate": 1e-06,
      "loss": 2.2297,
      "num_input_tokens_seen": 13817324,
      "step": 247
    },
    {
      "epoch": 0.5501113585746102,
      "loss": 2.3575921058654785,
      "loss_ce": 0.06559999287128448,
      "loss_iou": 0.859375,
      "loss_num": 0.115234375,
      "loss_xval": 2.296875,
      "num_input_tokens_seen": 13817324,
      "step": 247
    },
    {
      "epoch": 0.5523385300668151,
      "grad_norm": 40.075462341308594,
      "learning_rate": 1e-06,
      "loss": 1.6199,
      "num_input_tokens_seen": 13873804,
      "step": 248
    },
    {
      "epoch": 0.5523385300668151,
      "loss": 1.5571107864379883,
      "loss_ce": 0.028302079066634178,
      "loss_iou": 0.61328125,
      "loss_num": 0.060546875,
      "loss_xval": 1.53125,
      "num_input_tokens_seen": 13873804,
      "step": 248
    },
    {
      "epoch": 0.5545657015590201,
      "grad_norm": 36.94553756713867,
      "learning_rate": 1e-06,
      "loss": 1.4597,
      "num_input_tokens_seen": 13932404,
      "step": 249
    },
    {
      "epoch": 0.5545657015590201,
      "loss": 1.564407229423523,
      "loss_ce": 0.05122363194823265,
      "loss_iou": 0.64453125,
      "loss_num": 0.044677734375,
      "loss_xval": 1.515625,
      "num_input_tokens_seen": 13932404,
      "step": 249
    },
    {
      "epoch": 0.5567928730512249,
      "grad_norm": 48.372737884521484,
      "learning_rate": 1e-06,
      "loss": 1.6582,
      "num_input_tokens_seen": 13988996,
      "step": 250
    },
    {
      "epoch": 0.5567928730512249,
      "eval_seeclick_web_CIoU": 0.45204322040081024,
      "eval_seeclick_web_GIoU": 0.446009561419487,
      "eval_seeclick_web_IoU": 0.48508621752262115,
      "eval_seeclick_web_MAE_all": 0.015304200816899538,
      "eval_seeclick_web_MAE_h": 0.01132917869836092,
      "eval_seeclick_web_MAE_w": 0.01562582701444626,
      "eval_seeclick_web_MAE_x_boxes": 0.010999062564224005,
      "eval_seeclick_web_MAE_y_boxes": 0.02148408070206642,
      "eval_seeclick_web_inside_bbox": 0.8263888955116272,
      "eval_seeclick_web_loss": 1.2039486169815063,
      "eval_seeclick_web_loss_ce": 0.0032457184279337525,
      "eval_seeclick_web_loss_iou": 0.554443359375,
      "eval_seeclick_web_loss_num": 0.013319015502929688,
      "eval_seeclick_web_loss_xval": 1.176025390625,
      "eval_seeclick_web_runtime": 17.9468,
      "eval_seeclick_web_samples_per_second": 2.786,
      "eval_seeclick_web_steps_per_second": 0.111,
      "num_input_tokens_seen": 13988996,
      "step": 250
    },
    {
      "epoch": 0.5567928730512249,
      "eval_icons_CIoU": 0.16195975244045258,
      "eval_icons_GIoU": 0.18568327277898788,
      "eval_icons_IoU": 0.2861369401216507,
      "eval_icons_MAE_all": 0.05986557714641094,
      "eval_icons_MAE_h": 0.0345042385160923,
      "eval_icons_MAE_w": 0.07578632980585098,
      "eval_icons_MAE_x_boxes": 0.06385871395468712,
      "eval_icons_MAE_y_boxes": 0.02914611343294382,
      "eval_icons_inside_bbox": 0.4322916716337204,
      "eval_icons_loss": 1.9311987161636353,
      "eval_icons_loss_ce": 0.01602194458246231,
      "eval_icons_loss_iou": 0.7734375,
      "eval_icons_loss_num": 0.05643463134765625,
      "eval_icons_loss_xval": 1.828125,
      "eval_icons_runtime": 16.6678,
      "eval_icons_samples_per_second": 3.0,
      "eval_icons_steps_per_second": 0.12,
      "num_input_tokens_seen": 13988996,
      "step": 250
    },
    {
      "epoch": 0.5567928730512249,
      "eval_screenspot_CIoU": 0.13644553472598395,
      "eval_screenspot_GIoU": 0.13833926369746527,
      "eval_screenspot_IoU": 0.27226150035858154,
      "eval_screenspot_MAE_all": 0.11441413809855779,
      "eval_screenspot_MAE_h": 0.06734236205617587,
      "eval_screenspot_MAE_w": 0.15528815736373267,
      "eval_screenspot_MAE_x_boxes": 0.13489964107672373,
      "eval_screenspot_MAE_y_boxes": 0.08852330843607585,
      "eval_screenspot_inside_bbox": 0.4887500007947286,
      "eval_screenspot_loss": 2.3626174926757812,
      "eval_screenspot_loss_ce": 0.06506530692179997,
      "eval_screenspot_loss_iou": 0.8843587239583334,
      "eval_screenspot_loss_num": 0.11939748128255208,
      "eval_screenspot_loss_xval": 2.365234375,
      "eval_screenspot_runtime": 26.8445,
      "eval_screenspot_samples_per_second": 3.315,
      "eval_screenspot_steps_per_second": 0.112,
      "num_input_tokens_seen": 13988996,
      "step": 250
    },
    {
      "epoch": 0.5567928730512249,
      "eval_compot_CIoU": 0.2119811549782753,
      "eval_compot_GIoU": 0.22996322065591812,
      "eval_compot_IoU": 0.2962482124567032,
      "eval_compot_MAE_all": 0.029227093793451786,
      "eval_compot_MAE_h": 0.021410066168755293,
      "eval_compot_MAE_w": 0.032643974758684635,
      "eval_compot_MAE_x_boxes": 0.03914828971028328,
      "eval_compot_MAE_y_boxes": 0.012671195901930332,
      "eval_compot_inside_bbox": 0.46875,
      "eval_compot_loss": 1.7058837413787842,
      "eval_compot_loss_ce": 0.0029119880637153983,
      "eval_compot_loss_iou": 0.745849609375,
      "eval_compot_loss_num": 0.02825927734375,
      "eval_compot_loss_xval": 1.6337890625,
      "eval_compot_runtime": 18.0854,
      "eval_compot_samples_per_second": 2.765,
      "eval_compot_steps_per_second": 0.111,
      "num_input_tokens_seen": 13988996,
      "step": 250
    },
    {
      "epoch": 0.5567928730512249,
      "eval_custom_ui_val_CIoU": 0.18997877753443188,
      "eval_custom_ui_val_GIoU": 0.199455168719093,
      "eval_custom_ui_val_IoU": 0.26526735723018646,
      "eval_custom_ui_val_MAE_all": 0.06052247662511137,
      "eval_custom_ui_val_MAE_h": 0.03889055632882648,
      "eval_custom_ui_val_MAE_w": 0.07028790439168613,
      "eval_custom_ui_val_MAE_x_boxes": 0.058271253067586154,
      "eval_custom_ui_val_MAE_y_boxes": 0.050626704883244306,
      "eval_custom_ui_val_inside_bbox": 0.5111882719728682,
      "eval_custom_ui_val_loss": 1.941453218460083,
      "eval_custom_ui_val_loss_ce": 0.020650964023338422,
      "eval_custom_ui_val_loss_iou": 0.7940809461805556,
      "eval_custom_ui_val_loss_num": 0.05839665730794271,
      "eval_custom_ui_val_loss_xval": 1.8801540798611112,
      "eval_custom_ui_val_runtime": 55.8224,
      "eval_custom_ui_val_samples_per_second": 4.747,
      "eval_custom_ui_val_steps_per_second": 0.161,
      "num_input_tokens_seen": 13988996,
      "step": 250
    },
    {
      "epoch": 0.5567928730512249,
      "loss": 1.535244107246399,
      "loss_ce": 0.023525364696979523,
      "loss_iou": 0.66015625,
      "loss_num": 0.038330078125,
      "loss_xval": 1.515625,
      "num_input_tokens_seen": 13988996,
      "step": 250
    },
    {
      "epoch": 0.5590200445434298,
      "grad_norm": 25.04877281188965,
      "learning_rate": 1e-06,
      "loss": 1.6859,
      "num_input_tokens_seen": 14044712,
      "step": 251
    },
    {
      "epoch": 0.5590200445434298,
      "loss": 1.5795611143112183,
      "loss_ce": 0.04782275855541229,
      "loss_iou": 0.609375,
      "loss_num": 0.06298828125,
      "loss_xval": 1.53125,
      "num_input_tokens_seen": 14044712,
      "step": 251
    },
    {
      "epoch": 0.5612472160356348,
      "grad_norm": 88.63025665283203,
      "learning_rate": 1e-06,
      "loss": 1.3939,
      "num_input_tokens_seen": 14101200,
      "step": 252
    },
    {
      "epoch": 0.5612472160356348,
      "loss": 1.3761667013168335,
      "loss_ce": 0.0680612325668335,
      "loss_iou": 0.54296875,
      "loss_num": 0.044921875,
      "loss_xval": 1.3046875,
      "num_input_tokens_seen": 14101200,
      "step": 252
    },
    {
      "epoch": 0.5634743875278396,
      "grad_norm": 28.615373611450195,
      "learning_rate": 1e-06,
      "loss": 1.5713,
      "num_input_tokens_seen": 14158672,
      "step": 253
    },
    {
      "epoch": 0.5634743875278396,
      "loss": 1.540935754776001,
      "loss_ce": 0.01456859614700079,
      "loss_iou": 0.64453125,
      "loss_num": 0.046875,
      "loss_xval": 1.5234375,
      "num_input_tokens_seen": 14158672,
      "step": 253
    },
    {
      "epoch": 0.5657015590200446,
      "grad_norm": 39.433143615722656,
      "learning_rate": 1e-06,
      "loss": 1.6708,
      "num_input_tokens_seen": 14216228,
      "step": 254
    },
    {
      "epoch": 0.5657015590200446,
      "loss": 1.7911148071289062,
      "loss_ce": 0.05381014943122864,
      "loss_iou": 0.71484375,
      "loss_num": 0.060546875,
      "loss_xval": 1.734375,
      "num_input_tokens_seen": 14216228,
      "step": 254
    },
    {
      "epoch": 0.5679287305122495,
      "grad_norm": 30.27321434020996,
      "learning_rate": 1e-06,
      "loss": 1.9151,
      "num_input_tokens_seen": 14270952,
      "step": 255
    },
    {
      "epoch": 0.5679287305122495,
      "loss": 2.0560014247894287,
      "loss_ce": 0.04672405868768692,
      "loss_iou": 0.79296875,
      "loss_num": 0.08447265625,
      "loss_xval": 2.015625,
      "num_input_tokens_seen": 14270952,
      "step": 255
    },
    {
      "epoch": 0.5701559020044543,
      "grad_norm": 25.623844146728516,
      "learning_rate": 1e-06,
      "loss": 1.5495,
      "num_input_tokens_seen": 14327860,
      "step": 256
    },
    {
      "epoch": 0.5701559020044543,
      "loss": 1.5679357051849365,
      "loss_ce": 0.032779376953840256,
      "loss_iou": 0.6015625,
      "loss_num": 0.0654296875,
      "loss_xval": 1.53125,
      "num_input_tokens_seen": 14327860,
      "step": 256
    },
    {
      "epoch": 0.5723830734966593,
      "grad_norm": 34.12678909301758,
      "learning_rate": 1e-06,
      "loss": 1.5758,
      "num_input_tokens_seen": 14383192,
      "step": 257
    },
    {
      "epoch": 0.5723830734966593,
      "loss": 1.398708701133728,
      "loss_ce": 0.030056362971663475,
      "loss_iou": 0.54296875,
      "loss_num": 0.056396484375,
      "loss_xval": 1.3671875,
      "num_input_tokens_seen": 14383192,
      "step": 257
    },
    {
      "epoch": 0.5746102449888641,
      "grad_norm": 22.465286254882812,
      "learning_rate": 1e-06,
      "loss": 1.618,
      "num_input_tokens_seen": 14438224,
      "step": 258
    },
    {
      "epoch": 0.5746102449888641,
      "loss": 1.6812279224395752,
      "loss_ce": 0.019118648022413254,
      "loss_iou": 0.72265625,
      "loss_num": 0.04345703125,
      "loss_xval": 1.6640625,
      "num_input_tokens_seen": 14438224,
      "step": 258
    },
    {
      "epoch": 0.576837416481069,
      "grad_norm": 45.041263580322266,
      "learning_rate": 1e-06,
      "loss": 1.7445,
      "num_input_tokens_seen": 14490804,
      "step": 259
    },
    {
      "epoch": 0.576837416481069,
      "loss": 1.58518648147583,
      "loss_ce": 0.05833101272583008,
      "loss_iou": 0.62890625,
      "loss_num": 0.053955078125,
      "loss_xval": 1.5234375,
      "num_input_tokens_seen": 14490804,
      "step": 259
    },
    {
      "epoch": 0.579064587973274,
      "grad_norm": 32.25990295410156,
      "learning_rate": 1e-06,
      "loss": 1.7383,
      "num_input_tokens_seen": 14548204,
      "step": 260
    },
    {
      "epoch": 0.579064587973274,
      "loss": 1.8021881580352783,
      "loss_ce": 0.019961677491664886,
      "loss_iou": 0.73046875,
      "loss_num": 0.06396484375,
      "loss_xval": 1.78125,
      "num_input_tokens_seen": 14548204,
      "step": 260
    },
    {
      "epoch": 0.5812917594654788,
      "grad_norm": 24.800201416015625,
      "learning_rate": 1e-06,
      "loss": 1.9518,
      "num_input_tokens_seen": 14602976,
      "step": 261
    },
    {
      "epoch": 0.5812917594654788,
      "loss": 1.7187559604644775,
      "loss_ce": 0.05420520156621933,
      "loss_iou": 0.70703125,
      "loss_num": 0.04931640625,
      "loss_xval": 1.6640625,
      "num_input_tokens_seen": 14602976,
      "step": 261
    },
    {
      "epoch": 0.5835189309576837,
      "grad_norm": 38.99089431762695,
      "learning_rate": 1e-06,
      "loss": 1.3557,
      "num_input_tokens_seen": 14659812,
      "step": 262
    },
    {
      "epoch": 0.5835189309576837,
      "loss": 1.3884127140045166,
      "loss_ce": 0.003158772364258766,
      "loss_iou": 0.609375,
      "loss_num": 0.032470703125,
      "loss_xval": 1.3828125,
      "num_input_tokens_seen": 14659812,
      "step": 262
    },
    {
      "epoch": 0.5857461024498887,
      "grad_norm": 36.124290466308594,
      "learning_rate": 1e-06,
      "loss": 1.9584,
      "num_input_tokens_seen": 14716812,
      "step": 263
    },
    {
      "epoch": 0.5857461024498887,
      "loss": 2.1227617263793945,
      "loss_ce": 0.07393358647823334,
      "loss_iou": 0.83203125,
      "loss_num": 0.07763671875,
      "loss_xval": 2.046875,
      "num_input_tokens_seen": 14716812,
      "step": 263
    },
    {
      "epoch": 0.5879732739420935,
      "grad_norm": 27.93045997619629,
      "learning_rate": 1e-06,
      "loss": 1.6202,
      "num_input_tokens_seen": 14770856,
      "step": 264
    },
    {
      "epoch": 0.5879732739420935,
      "loss": 1.2804536819458008,
      "loss_ce": 0.0348481610417366,
      "loss_iou": 0.51953125,
      "loss_num": 0.041748046875,
      "loss_xval": 1.2421875,
      "num_input_tokens_seen": 14770856,
      "step": 264
    },
    {
      "epoch": 0.5902004454342984,
      "grad_norm": 25.890186309814453,
      "learning_rate": 1e-06,
      "loss": 1.7844,
      "num_input_tokens_seen": 14828476,
      "step": 265
    },
    {
      "epoch": 0.5902004454342984,
      "loss": 1.8109703063964844,
      "loss_ce": 0.03265002369880676,
      "loss_iou": 0.7421875,
      "loss_num": 0.059326171875,
      "loss_xval": 1.78125,
      "num_input_tokens_seen": 14828476,
      "step": 265
    },
    {
      "epoch": 0.5924276169265034,
      "grad_norm": 44.33429718017578,
      "learning_rate": 1e-06,
      "loss": 1.8674,
      "num_input_tokens_seen": 14880352,
      "step": 266
    },
    {
      "epoch": 0.5924276169265034,
      "loss": 1.7880139350891113,
      "loss_ce": 0.041920170187950134,
      "loss_iou": 0.70703125,
      "loss_num": 0.06640625,
      "loss_xval": 1.75,
      "num_input_tokens_seen": 14880352,
      "step": 266
    },
    {
      "epoch": 0.5946547884187082,
      "grad_norm": 46.44974136352539,
      "learning_rate": 1e-06,
      "loss": 1.6601,
      "num_input_tokens_seen": 14934660,
      "step": 267
    },
    {
      "epoch": 0.5946547884187082,
      "loss": 1.5294612646102905,
      "loss_ce": 0.031414370983839035,
      "loss_iou": 0.6171875,
      "loss_num": 0.052978515625,
      "loss_xval": 1.5,
      "num_input_tokens_seen": 14934660,
      "step": 267
    },
    {
      "epoch": 0.5968819599109132,
      "grad_norm": 50.88404846191406,
      "learning_rate": 1e-06,
      "loss": 1.4847,
      "num_input_tokens_seen": 14990032,
      "step": 268
    },
    {
      "epoch": 0.5968819599109132,
      "loss": 1.1937447786331177,
      "loss_ce": 0.016986995935440063,
      "loss_iou": 0.5,
      "loss_num": 0.03564453125,
      "loss_xval": 1.1796875,
      "num_input_tokens_seen": 14990032,
      "step": 268
    },
    {
      "epoch": 0.5991091314031181,
      "grad_norm": 27.827131271362305,
      "learning_rate": 1e-06,
      "loss": 1.5246,
      "num_input_tokens_seen": 15048824,
      "step": 269
    },
    {
      "epoch": 0.5991091314031181,
      "loss": 1.5418243408203125,
      "loss_ce": 0.0242463368922472,
      "loss_iou": 0.63671875,
      "loss_num": 0.049072265625,
      "loss_xval": 1.515625,
      "num_input_tokens_seen": 15048824,
      "step": 269
    },
    {
      "epoch": 0.6013363028953229,
      "grad_norm": 19.647504806518555,
      "learning_rate": 1e-06,
      "loss": 1.637,
      "num_input_tokens_seen": 15107692,
      "step": 270
    },
    {
      "epoch": 0.6013363028953229,
      "loss": 1.218379259109497,
      "loss_ce": 0.010371430777013302,
      "loss_iou": 0.53125,
      "loss_num": 0.0284423828125,
      "loss_xval": 1.2109375,
      "num_input_tokens_seen": 15107692,
      "step": 270
    },
    {
      "epoch": 0.6035634743875279,
      "grad_norm": 24.345657348632812,
      "learning_rate": 1e-06,
      "loss": 1.8571,
      "num_input_tokens_seen": 15161936,
      "step": 271
    },
    {
      "epoch": 0.6035634743875279,
      "loss": 1.8928604125976562,
      "loss_ce": 0.01151271816343069,
      "loss_iou": 0.80078125,
      "loss_num": 0.055908203125,
      "loss_xval": 1.8828125,
      "num_input_tokens_seen": 15161936,
      "step": 271
    },
    {
      "epoch": 0.6057906458797327,
      "grad_norm": 28.753704071044922,
      "learning_rate": 1e-06,
      "loss": 1.4391,
      "num_input_tokens_seen": 15218400,
      "step": 272
    },
    {
      "epoch": 0.6057906458797327,
      "loss": 1.046234130859375,
      "loss_ce": 0.030609235167503357,
      "loss_iou": 0.431640625,
      "loss_num": 0.0303955078125,
      "loss_xval": 1.015625,
      "num_input_tokens_seen": 15218400,
      "step": 272
    },
    {
      "epoch": 0.6080178173719376,
      "grad_norm": 28.32134437561035,
      "learning_rate": 1e-06,
      "loss": 1.6383,
      "num_input_tokens_seen": 15271292,
      "step": 273
    },
    {
      "epoch": 0.6080178173719376,
      "loss": 1.7744327783584595,
      "loss_ce": 0.0346866175532341,
      "loss_iou": 0.7578125,
      "loss_num": 0.044921875,
      "loss_xval": 1.7421875,
      "num_input_tokens_seen": 15271292,
      "step": 273
    },
    {
      "epoch": 0.6102449888641426,
      "grad_norm": 26.63081932067871,
      "learning_rate": 1e-06,
      "loss": 1.6534,
      "num_input_tokens_seen": 15324252,
      "step": 274
    },
    {
      "epoch": 0.6102449888641426,
      "loss": 1.8621397018432617,
      "loss_ce": 0.015460037626326084,
      "loss_iou": 0.77734375,
      "loss_num": 0.05859375,
      "loss_xval": 1.84375,
      "num_input_tokens_seen": 15324252,
      "step": 274
    },
    {
      "epoch": 0.6124721603563474,
      "grad_norm": 159.9815673828125,
      "learning_rate": 1e-06,
      "loss": 1.6268,
      "num_input_tokens_seen": 15380232,
      "step": 275
    },
    {
      "epoch": 0.6124721603563474,
      "loss": 1.5270476341247559,
      "loss_ce": 0.024606265127658844,
      "loss_iou": 0.6015625,
      "loss_num": 0.060546875,
      "loss_xval": 1.5,
      "num_input_tokens_seen": 15380232,
      "step": 275
    },
    {
      "epoch": 0.6146993318485523,
      "grad_norm": 16.36348533630371,
      "learning_rate": 1e-06,
      "loss": 1.8404,
      "num_input_tokens_seen": 15434716,
      "step": 276
    },
    {
      "epoch": 0.6146993318485523,
      "loss": 2.1927452087402344,
      "loss_ce": 0.04479604959487915,
      "loss_iou": 0.81640625,
      "loss_num": 0.10400390625,
      "loss_xval": 2.140625,
      "num_input_tokens_seen": 15434716,
      "step": 276
    },
    {
      "epoch": 0.6169265033407573,
      "grad_norm": 35.74172592163086,
      "learning_rate": 1e-06,
      "loss": 1.4253,
      "num_input_tokens_seen": 15491324,
      "step": 277
    },
    {
      "epoch": 0.6169265033407573,
      "loss": 1.4241523742675781,
      "loss_ce": 0.007160228211432695,
      "loss_iou": 0.59765625,
      "loss_num": 0.045166015625,
      "loss_xval": 1.4140625,
      "num_input_tokens_seen": 15491324,
      "step": 277
    },
    {
      "epoch": 0.6191536748329621,
      "grad_norm": 30.04996681213379,
      "learning_rate": 1e-06,
      "loss": 1.7018,
      "num_input_tokens_seen": 15545848,
      "step": 278
    },
    {
      "epoch": 0.6191536748329621,
      "loss": 1.6422159671783447,
      "loss_ce": 0.028934601694345474,
      "loss_iou": 0.67578125,
      "loss_num": 0.051513671875,
      "loss_xval": 1.609375,
      "num_input_tokens_seen": 15545848,
      "step": 278
    },
    {
      "epoch": 0.621380846325167,
      "grad_norm": 32.703819274902344,
      "learning_rate": 1e-06,
      "loss": 1.6324,
      "num_input_tokens_seen": 15603716,
      "step": 279
    },
    {
      "epoch": 0.621380846325167,
      "loss": 1.5165987014770508,
      "loss_ce": 0.0112276840955019,
      "loss_iou": 0.6484375,
      "loss_num": 0.041015625,
      "loss_xval": 1.5078125,
      "num_input_tokens_seen": 15603716,
      "step": 279
    },
    {
      "epoch": 0.623608017817372,
      "grad_norm": 35.6143798828125,
      "learning_rate": 1e-06,
      "loss": 1.6699,
      "num_input_tokens_seen": 15658112,
      "step": 280
    },
    {
      "epoch": 0.623608017817372,
      "loss": 1.777186393737793,
      "loss_ce": 0.02865123562514782,
      "loss_iou": 0.6796875,
      "loss_num": 0.07763671875,
      "loss_xval": 1.75,
      "num_input_tokens_seen": 15658112,
      "step": 280
    },
    {
      "epoch": 0.6258351893095768,
      "grad_norm": 28.780593872070312,
      "learning_rate": 1e-06,
      "loss": 1.6865,
      "num_input_tokens_seen": 15716760,
      "step": 281
    },
    {
      "epoch": 0.6258351893095768,
      "loss": 1.422225832939148,
      "loss_ce": 0.011092978529632092,
      "loss_iou": 0.6328125,
      "loss_num": 0.0284423828125,
      "loss_xval": 1.4140625,
      "num_input_tokens_seen": 15716760,
      "step": 281
    },
    {
      "epoch": 0.6280623608017817,
      "grad_norm": 39.134151458740234,
      "learning_rate": 1e-06,
      "loss": 2.0141,
      "num_input_tokens_seen": 15773952,
      "step": 282
    },
    {
      "epoch": 0.6280623608017817,
      "loss": 2.0597357749938965,
      "loss_ce": 0.03629831597208977,
      "loss_iou": 0.79296875,
      "loss_num": 0.087890625,
      "loss_xval": 2.03125,
      "num_input_tokens_seen": 15773952,
      "step": 282
    },
    {
      "epoch": 0.6302895322939867,
      "grad_norm": 21.123931884765625,
      "learning_rate": 1e-06,
      "loss": 1.4487,
      "num_input_tokens_seen": 15832456,
      "step": 283
    },
    {
      "epoch": 0.6302895322939867,
      "loss": 1.4959959983825684,
      "loss_ce": 0.018945157527923584,
      "loss_iou": 0.6328125,
      "loss_num": 0.0419921875,
      "loss_xval": 1.4765625,
      "num_input_tokens_seen": 15832456,
      "step": 283
    },
    {
      "epoch": 0.6325167037861915,
      "grad_norm": 137.96141052246094,
      "learning_rate": 1e-06,
      "loss": 1.5979,
      "num_input_tokens_seen": 15890888,
      "step": 284
    },
    {
      "epoch": 0.6325167037861915,
      "loss": 1.8737279176712036,
      "loss_ce": 0.05341540277004242,
      "loss_iou": 0.71875,
      "loss_num": 0.07763671875,
      "loss_xval": 1.8203125,
      "num_input_tokens_seen": 15890888,
      "step": 284
    },
    {
      "epoch": 0.6347438752783965,
      "grad_norm": 26.675518035888672,
      "learning_rate": 1e-06,
      "loss": 1.7712,
      "num_input_tokens_seen": 15945736,
      "step": 285
    },
    {
      "epoch": 0.6347438752783965,
      "loss": 1.6105928421020508,
      "loss_ce": 0.03197958692908287,
      "loss_iou": 0.65234375,
      "loss_num": 0.054931640625,
      "loss_xval": 1.578125,
      "num_input_tokens_seen": 15945736,
      "step": 285
    },
    {
      "epoch": 0.6369710467706013,
      "grad_norm": 24.4492130279541,
      "learning_rate": 1e-06,
      "loss": 1.5191,
      "num_input_tokens_seen": 16002312,
      "step": 286
    },
    {
      "epoch": 0.6369710467706013,
      "loss": 1.766524314880371,
      "loss_ce": 0.0673055648803711,
      "loss_iou": 0.6796875,
      "loss_num": 0.0673828125,
      "loss_xval": 1.703125,
      "num_input_tokens_seen": 16002312,
      "step": 286
    },
    {
      "epoch": 0.6391982182628062,
      "grad_norm": 30.305042266845703,
      "learning_rate": 1e-06,
      "loss": 1.4741,
      "num_input_tokens_seen": 16059496,
      "step": 287
    },
    {
      "epoch": 0.6391982182628062,
      "loss": 1.4600698947906494,
      "loss_ce": 0.04307776317000389,
      "loss_iou": 0.53125,
      "loss_num": 0.06982421875,
      "loss_xval": 1.4140625,
      "num_input_tokens_seen": 16059496,
      "step": 287
    },
    {
      "epoch": 0.6414253897550112,
      "grad_norm": 43.9190788269043,
      "learning_rate": 1e-06,
      "loss": 2.0036,
      "num_input_tokens_seen": 16114212,
      "step": 288
    },
    {
      "epoch": 0.6414253897550112,
      "loss": 2.116396188735962,
      "loss_ce": 0.04706018790602684,
      "loss_iou": 0.7890625,
      "loss_num": 0.09912109375,
      "loss_xval": 2.0625,
      "num_input_tokens_seen": 16114212,
      "step": 288
    },
    {
      "epoch": 0.643652561247216,
      "grad_norm": 32.69955825805664,
      "learning_rate": 1e-06,
      "loss": 1.4369,
      "num_input_tokens_seen": 16171136,
      "step": 289
    },
    {
      "epoch": 0.643652561247216,
      "loss": 1.4707211256027222,
      "loss_ce": 0.019060947000980377,
      "loss_iou": 0.6015625,
      "loss_num": 0.049560546875,
      "loss_xval": 1.453125,
      "num_input_tokens_seen": 16171136,
      "step": 289
    },
    {
      "epoch": 0.6458797327394209,
      "grad_norm": 23.189529418945312,
      "learning_rate": 1e-06,
      "loss": 1.6011,
      "num_input_tokens_seen": 16227256,
      "step": 290
    },
    {
      "epoch": 0.6458797327394209,
      "loss": 1.716025948524475,
      "loss_ce": 0.02217838540673256,
      "loss_iou": 0.7109375,
      "loss_num": 0.0537109375,
      "loss_xval": 1.6953125,
      "num_input_tokens_seen": 16227256,
      "step": 290
    },
    {
      "epoch": 0.6481069042316259,
      "grad_norm": 83.02802276611328,
      "learning_rate": 1e-06,
      "loss": 1.5053,
      "num_input_tokens_seen": 16283164,
      "step": 291
    },
    {
      "epoch": 0.6481069042316259,
      "loss": 1.8076732158660889,
      "loss_ce": 0.03325919434428215,
      "loss_iou": 0.7265625,
      "loss_num": 0.06494140625,
      "loss_xval": 1.7734375,
      "num_input_tokens_seen": 16283164,
      "step": 291
    },
    {
      "epoch": 0.6503340757238307,
      "grad_norm": 26.41913604736328,
      "learning_rate": 1e-06,
      "loss": 1.482,
      "num_input_tokens_seen": 16341224,
      "step": 292
    },
    {
      "epoch": 0.6503340757238307,
      "loss": 1.52240788936615,
      "loss_ce": 0.02973209135234356,
      "loss_iou": 0.60546875,
      "loss_num": 0.055908203125,
      "loss_xval": 1.4921875,
      "num_input_tokens_seen": 16341224,
      "step": 292
    },
    {
      "epoch": 0.6525612472160356,
      "grad_norm": 24.008352279663086,
      "learning_rate": 1e-06,
      "loss": 1.2855,
      "num_input_tokens_seen": 16399488,
      "step": 293
    },
    {
      "epoch": 0.6525612472160356,
      "loss": 1.323553442955017,
      "loss_ce": 0.02570188418030739,
      "loss_iou": 0.515625,
      "loss_num": 0.052490234375,
      "loss_xval": 1.296875,
      "num_input_tokens_seen": 16399488,
      "step": 293
    },
    {
      "epoch": 0.6547884187082406,
      "grad_norm": 40.153358459472656,
      "learning_rate": 1e-06,
      "loss": 2.0784,
      "num_input_tokens_seen": 16452992,
      "step": 294
    },
    {
      "epoch": 0.6547884187082406,
      "loss": 2.15143084526062,
      "loss_ce": 0.020571384578943253,
      "loss_iou": 0.828125,
      "loss_num": 0.0947265625,
      "loss_xval": 2.125,
      "num_input_tokens_seen": 16452992,
      "step": 294
    },
    {
      "epoch": 0.6570155902004454,
      "grad_norm": 19.569934844970703,
      "learning_rate": 1e-06,
      "loss": 1.6381,
      "num_input_tokens_seen": 16510300,
      "step": 295
    },
    {
      "epoch": 0.6570155902004454,
      "loss": 1.743638515472412,
      "loss_ce": 0.018052466213703156,
      "loss_iou": 0.7265625,
      "loss_num": 0.053466796875,
      "loss_xval": 1.7265625,
      "num_input_tokens_seen": 16510300,
      "step": 295
    },
    {
      "epoch": 0.6592427616926503,
      "grad_norm": 24.097932815551758,
      "learning_rate": 1e-06,
      "loss": 1.4483,
      "num_input_tokens_seen": 16565272,
      "step": 296
    },
    {
      "epoch": 0.6592427616926503,
      "loss": 1.252844214439392,
      "loss_ce": 0.0033324414398521185,
      "loss_iou": 0.49609375,
      "loss_num": 0.051513671875,
      "loss_xval": 1.25,
      "num_input_tokens_seen": 16565272,
      "step": 296
    },
    {
      "epoch": 0.6614699331848553,
      "grad_norm": 20.85422706604004,
      "learning_rate": 1e-06,
      "loss": 1.4426,
      "num_input_tokens_seen": 16620292,
      "step": 297
    },
    {
      "epoch": 0.6614699331848553,
      "loss": 1.5032652616500854,
      "loss_ce": 0.02230823040008545,
      "loss_iou": 0.6015625,
      "loss_num": 0.055419921875,
      "loss_xval": 1.484375,
      "num_input_tokens_seen": 16620292,
      "step": 297
    },
    {
      "epoch": 0.6636971046770601,
      "grad_norm": 31.075130462646484,
      "learning_rate": 1e-06,
      "loss": 1.7549,
      "num_input_tokens_seen": 16677532,
      "step": 298
    },
    {
      "epoch": 0.6636971046770601,
      "loss": 1.6268385648727417,
      "loss_ce": 0.030647173523902893,
      "loss_iou": 0.64453125,
      "loss_num": 0.0615234375,
      "loss_xval": 1.59375,
      "num_input_tokens_seen": 16677532,
      "step": 298
    },
    {
      "epoch": 0.6659242761692651,
      "grad_norm": 179.63589477539062,
      "learning_rate": 1e-06,
      "loss": 1.5358,
      "num_input_tokens_seen": 16731944,
      "step": 299
    },
    {
      "epoch": 0.6659242761692651,
      "loss": 1.3143196105957031,
      "loss_ce": 0.01890948787331581,
      "loss_iou": 0.52734375,
      "loss_num": 0.04736328125,
      "loss_xval": 1.296875,
      "num_input_tokens_seen": 16731944,
      "step": 299
    },
    {
      "epoch": 0.6681514476614699,
      "grad_norm": 78.36969757080078,
      "learning_rate": 1e-06,
      "loss": 1.6417,
      "num_input_tokens_seen": 16785832,
      "step": 300
    },
    {
      "epoch": 0.6681514476614699,
      "loss": 1.405898928642273,
      "loss_ce": 0.023086415603756905,
      "loss_iou": 0.5546875,
      "loss_num": 0.054931640625,
      "loss_xval": 1.3828125,
      "num_input_tokens_seen": 16785832,
      "step": 300
    },
    {
      "epoch": 0.6703786191536748,
      "grad_norm": 20.544160842895508,
      "learning_rate": 1e-06,
      "loss": 1.3546,
      "num_input_tokens_seen": 16843352,
      "step": 301
    },
    {
      "epoch": 0.6703786191536748,
      "loss": 1.2849873304367065,
      "loss_ce": 0.00910840556025505,
      "loss_iou": 0.55859375,
      "loss_num": 0.03125,
      "loss_xval": 1.2734375,
      "num_input_tokens_seen": 16843352,
      "step": 301
    },
    {
      "epoch": 0.6726057906458798,
      "grad_norm": 33.55524826049805,
      "learning_rate": 1e-06,
      "loss": 1.5479,
      "num_input_tokens_seen": 16899880,
      "step": 302
    },
    {
      "epoch": 0.6726057906458798,
      "loss": 1.4881852865219116,
      "loss_ce": 0.01504078321158886,
      "loss_iou": 0.625,
      "loss_num": 0.04443359375,
      "loss_xval": 1.4765625,
      "num_input_tokens_seen": 16899880,
      "step": 302
    },
    {
      "epoch": 0.6748329621380846,
      "grad_norm": 25.37139320373535,
      "learning_rate": 1e-06,
      "loss": 1.3608,
      "num_input_tokens_seen": 16955144,
      "step": 303
    },
    {
      "epoch": 0.6748329621380846,
      "loss": 1.1685500144958496,
      "loss_ce": 0.01718279719352722,
      "loss_iou": 0.484375,
      "loss_num": 0.03662109375,
      "loss_xval": 1.1484375,
      "num_input_tokens_seen": 16955144,
      "step": 303
    },
    {
      "epoch": 0.6770601336302895,
      "grad_norm": 26.81342887878418,
      "learning_rate": 1e-06,
      "loss": 1.3982,
      "num_input_tokens_seen": 17011748,
      "step": 304
    },
    {
      "epoch": 0.6770601336302895,
      "loss": 1.2066997289657593,
      "loss_ce": 0.01138727180659771,
      "loss_iou": 0.53125,
      "loss_num": 0.02587890625,
      "loss_xval": 1.1953125,
      "num_input_tokens_seen": 17011748,
      "step": 304
    },
    {
      "epoch": 0.6792873051224945,
      "grad_norm": 27.548656463623047,
      "learning_rate": 1e-06,
      "loss": 1.5679,
      "num_input_tokens_seen": 17068112,
      "step": 305
    },
    {
      "epoch": 0.6792873051224945,
      "loss": 1.7167648077011108,
      "loss_ce": 0.04781951755285263,
      "loss_iou": 0.671875,
      "loss_num": 0.06396484375,
      "loss_xval": 1.671875,
      "num_input_tokens_seen": 17068112,
      "step": 305
    },
    {
      "epoch": 0.6815144766146993,
      "grad_norm": 40.20294952392578,
      "learning_rate": 1e-06,
      "loss": 1.4371,
      "num_input_tokens_seen": 17122104,
      "step": 306
    },
    {
      "epoch": 0.6815144766146993,
      "loss": 1.477067470550537,
      "loss_ce": 0.01710660383105278,
      "loss_iou": 0.58984375,
      "loss_num": 0.055419921875,
      "loss_xval": 1.4609375,
      "num_input_tokens_seen": 17122104,
      "step": 306
    },
    {
      "epoch": 0.6837416481069042,
      "grad_norm": 31.41501808166504,
      "learning_rate": 1e-06,
      "loss": 1.4743,
      "num_input_tokens_seen": 17179616,
      "step": 307
    },
    {
      "epoch": 0.6837416481069042,
      "loss": 1.279785394668579,
      "loss_ce": 0.05468768998980522,
      "loss_iou": 0.515625,
      "loss_num": 0.03857421875,
      "loss_xval": 1.2265625,
      "num_input_tokens_seen": 17179616,
      "step": 307
    },
    {
      "epoch": 0.6859688195991092,
      "grad_norm": 31.942686080932617,
      "learning_rate": 1e-06,
      "loss": 1.2797,
      "num_input_tokens_seen": 17236176,
      "step": 308
    },
    {
      "epoch": 0.6859688195991092,
      "loss": 1.2551771402359009,
      "loss_ce": 0.027149761095643044,
      "loss_iou": 0.5390625,
      "loss_num": 0.029541015625,
      "loss_xval": 1.2265625,
      "num_input_tokens_seen": 17236176,
      "step": 308
    },
    {
      "epoch": 0.688195991091314,
      "grad_norm": 30.1185302734375,
      "learning_rate": 1e-06,
      "loss": 1.3682,
      "num_input_tokens_seen": 17293040,
      "step": 309
    },
    {
      "epoch": 0.688195991091314,
      "loss": 1.479234218597412,
      "loss_ce": 0.037827931344509125,
      "loss_iou": 0.60546875,
      "loss_num": 0.0458984375,
      "loss_xval": 1.4375,
      "num_input_tokens_seen": 17293040,
      "step": 309
    },
    {
      "epoch": 0.6904231625835189,
      "grad_norm": 17.059524536132812,
      "learning_rate": 1e-06,
      "loss": 1.5945,
      "num_input_tokens_seen": 17349008,
      "step": 310
    },
    {
      "epoch": 0.6904231625835189,
      "loss": 1.7185901403427124,
      "loss_ce": 0.01155892200767994,
      "loss_iou": 0.71484375,
      "loss_num": 0.054443359375,
      "loss_xval": 1.703125,
      "num_input_tokens_seen": 17349008,
      "step": 310
    },
    {
      "epoch": 0.6926503340757239,
      "grad_norm": 20.258180618286133,
      "learning_rate": 1e-06,
      "loss": 1.4631,
      "num_input_tokens_seen": 17404740,
      "step": 311
    },
    {
      "epoch": 0.6926503340757239,
      "loss": 1.595271348953247,
      "loss_ce": 0.05718539282679558,
      "loss_iou": 0.625,
      "loss_num": 0.058349609375,
      "loss_xval": 1.5390625,
      "num_input_tokens_seen": 17404740,
      "step": 311
    },
    {
      "epoch": 0.6948775055679287,
      "grad_norm": 51.31781005859375,
      "learning_rate": 1e-06,
      "loss": 1.9344,
      "num_input_tokens_seen": 17460352,
      "step": 312
    },
    {
      "epoch": 0.6948775055679287,
      "loss": 1.8748557567596436,
      "loss_ce": 0.0301292035728693,
      "loss_iou": 0.7734375,
      "loss_num": 0.0595703125,
      "loss_xval": 1.84375,
      "num_input_tokens_seen": 17460352,
      "step": 312
    },
    {
      "epoch": 0.6971046770601337,
      "grad_norm": 42.53439712524414,
      "learning_rate": 1e-06,
      "loss": 1.2032,
      "num_input_tokens_seen": 17518148,
      "step": 313
    },
    {
      "epoch": 0.6971046770601337,
      "loss": 1.2251746654510498,
      "loss_ce": 0.024491025134921074,
      "loss_iou": 0.51171875,
      "loss_num": 0.03515625,
      "loss_xval": 1.203125,
      "num_input_tokens_seen": 17518148,
      "step": 313
    },
    {
      "epoch": 0.6993318485523385,
      "grad_norm": 190.15823364257812,
      "learning_rate": 1e-06,
      "loss": 1.5881,
      "num_input_tokens_seen": 17574740,
      "step": 314
    },
    {
      "epoch": 0.6993318485523385,
      "loss": 1.6861159801483154,
      "loss_ce": 0.023518286645412445,
      "loss_iou": 0.6875,
      "loss_num": 0.056640625,
      "loss_xval": 1.6640625,
      "num_input_tokens_seen": 17574740,
      "step": 314
    },
    {
      "epoch": 0.7015590200445434,
      "grad_norm": 23.019386291503906,
      "learning_rate": 1e-06,
      "loss": 1.336,
      "num_input_tokens_seen": 17630988,
      "step": 315
    },
    {
      "epoch": 0.7015590200445434,
      "loss": 1.2523000240325928,
      "loss_ce": 0.0032766717486083508,
      "loss_iou": 0.5078125,
      "loss_num": 0.047119140625,
      "loss_xval": 1.25,
      "num_input_tokens_seen": 17630988,
      "step": 315
    },
    {
      "epoch": 0.7037861915367484,
      "grad_norm": 33.44746017456055,
      "learning_rate": 1e-06,
      "loss": 1.4686,
      "num_input_tokens_seen": 17688132,
      "step": 316
    },
    {
      "epoch": 0.7037861915367484,
      "loss": 1.3316755294799805,
      "loss_ce": 0.024058308452367783,
      "loss_iou": 0.53125,
      "loss_num": 0.049560546875,
      "loss_xval": 1.3046875,
      "num_input_tokens_seen": 17688132,
      "step": 316
    },
    {
      "epoch": 0.7060133630289532,
      "grad_norm": 26.666362762451172,
      "learning_rate": 1e-06,
      "loss": 1.4387,
      "num_input_tokens_seen": 17744136,
      "step": 317
    },
    {
      "epoch": 0.7060133630289532,
      "loss": 1.4408788681030273,
      "loss_ce": 0.038046881556510925,
      "loss_iou": 0.58984375,
      "loss_num": 0.044677734375,
      "loss_xval": 1.40625,
      "num_input_tokens_seen": 17744136,
      "step": 317
    },
    {
      "epoch": 0.7082405345211581,
      "grad_norm": 29.52345848083496,
      "learning_rate": 1e-06,
      "loss": 1.6314,
      "num_input_tokens_seen": 17802324,
      "step": 318
    },
    {
      "epoch": 0.7082405345211581,
      "loss": 1.7159836292266846,
      "loss_ce": 0.018229741603136063,
      "loss_iou": 0.6875,
      "loss_num": 0.06396484375,
      "loss_xval": 1.6953125,
      "num_input_tokens_seen": 17802324,
      "step": 318
    },
    {
      "epoch": 0.7104677060133631,
      "grad_norm": 31.622154235839844,
      "learning_rate": 1e-06,
      "loss": 1.765,
      "num_input_tokens_seen": 17860648,
      "step": 319
    },
    {
      "epoch": 0.7104677060133631,
      "loss": 1.6791552305221558,
      "loss_ce": 0.010209913365542889,
      "loss_iou": 0.671875,
      "loss_num": 0.06494140625,
      "loss_xval": 1.671875,
      "num_input_tokens_seen": 17860648,
      "step": 319
    },
    {
      "epoch": 0.7126948775055679,
      "grad_norm": 38.03776931762695,
      "learning_rate": 1e-06,
      "loss": 1.3441,
      "num_input_tokens_seen": 17915988,
      "step": 320
    },
    {
      "epoch": 0.7126948775055679,
      "loss": 1.6516163349151611,
      "loss_ce": 0.017338957637548447,
      "loss_iou": 0.65234375,
      "loss_num": 0.06640625,
      "loss_xval": 1.6328125,
      "num_input_tokens_seen": 17915988,
      "step": 320
    },
    {
      "epoch": 0.7149220489977728,
      "grad_norm": 30.399452209472656,
      "learning_rate": 1e-06,
      "loss": 1.4676,
      "num_input_tokens_seen": 17974328,
      "step": 321
    },
    {
      "epoch": 0.7149220489977728,
      "loss": 1.5550212860107422,
      "loss_ce": 0.01888836920261383,
      "loss_iou": 0.62109375,
      "loss_num": 0.05908203125,
      "loss_xval": 1.5390625,
      "num_input_tokens_seen": 17974328,
      "step": 321
    },
    {
      "epoch": 0.7171492204899778,
      "grad_norm": 34.999053955078125,
      "learning_rate": 1e-06,
      "loss": 1.5055,
      "num_input_tokens_seen": 18027768,
      "step": 322
    },
    {
      "epoch": 0.7171492204899778,
      "loss": 1.5647876262664795,
      "loss_ce": 0.014494719915091991,
      "loss_iou": 0.65625,
      "loss_num": 0.048095703125,
      "loss_xval": 1.546875,
      "num_input_tokens_seen": 18027768,
      "step": 322
    },
    {
      "epoch": 0.7193763919821826,
      "grad_norm": 48.161949157714844,
      "learning_rate": 1e-06,
      "loss": 1.5941,
      "num_input_tokens_seen": 18086484,
      "step": 323
    },
    {
      "epoch": 0.7193763919821826,
      "loss": 1.713207483291626,
      "loss_ce": 0.02326606959104538,
      "loss_iou": 0.70703125,
      "loss_num": 0.05517578125,
      "loss_xval": 1.6875,
      "num_input_tokens_seen": 18086484,
      "step": 323
    },
    {
      "epoch": 0.7216035634743875,
      "grad_norm": 68.18085479736328,
      "learning_rate": 1e-06,
      "loss": 1.604,
      "num_input_tokens_seen": 18142296,
      "step": 324
    },
    {
      "epoch": 0.7216035634743875,
      "loss": 1.4922294616699219,
      "loss_ce": 0.012249022722244263,
      "loss_iou": 0.59375,
      "loss_num": 0.05810546875,
      "loss_xval": 1.4765625,
      "num_input_tokens_seen": 18142296,
      "step": 324
    },
    {
      "epoch": 0.7238307349665924,
      "grad_norm": 23.24517250061035,
      "learning_rate": 1e-06,
      "loss": 1.6705,
      "num_input_tokens_seen": 18201016,
      "step": 325
    },
    {
      "epoch": 0.7238307349665924,
      "loss": 1.8428983688354492,
      "loss_ce": 0.0069608502089977264,
      "loss_iou": 0.77734375,
      "loss_num": 0.056396484375,
      "loss_xval": 1.8359375,
      "num_input_tokens_seen": 18201016,
      "step": 325
    },
    {
      "epoch": 0.7260579064587973,
      "grad_norm": 95.58949279785156,
      "learning_rate": 1e-06,
      "loss": 1.7858,
      "num_input_tokens_seen": 18256504,
      "step": 326
    },
    {
      "epoch": 0.7260579064587973,
      "loss": 1.7894692420959473,
      "loss_ce": 0.013102035038173199,
      "loss_iou": 0.69921875,
      "loss_num": 0.07568359375,
      "loss_xval": 1.7734375,
      "num_input_tokens_seen": 18256504,
      "step": 326
    },
    {
      "epoch": 0.7282850779510023,
      "grad_norm": 23.247690200805664,
      "learning_rate": 1e-06,
      "loss": 1.5796,
      "num_input_tokens_seen": 18310452,
      "step": 327
    },
    {
      "epoch": 0.7282850779510023,
      "loss": 1.59755539894104,
      "loss_ce": 0.010641279630362988,
      "loss_iou": 0.65234375,
      "loss_num": 0.05615234375,
      "loss_xval": 1.5859375,
      "num_input_tokens_seen": 18310452,
      "step": 327
    },
    {
      "epoch": 0.7305122494432071,
      "grad_norm": 67.12216186523438,
      "learning_rate": 1e-06,
      "loss": 1.4647,
      "num_input_tokens_seen": 18367084,
      "step": 328
    },
    {
      "epoch": 0.7305122494432071,
      "loss": 1.48179030418396,
      "loss_ce": 0.014016897417604923,
      "loss_iou": 0.5859375,
      "loss_num": 0.05908203125,
      "loss_xval": 1.46875,
      "num_input_tokens_seen": 18367084,
      "step": 328
    },
    {
      "epoch": 0.732739420935412,
      "grad_norm": 31.022777557373047,
      "learning_rate": 1e-06,
      "loss": 1.4833,
      "num_input_tokens_seen": 18424920,
      "step": 329
    },
    {
      "epoch": 0.732739420935412,
      "loss": 1.486365795135498,
      "loss_ce": 0.014686101116240025,
      "loss_iou": 0.62109375,
      "loss_num": 0.0458984375,
      "loss_xval": 1.46875,
      "num_input_tokens_seen": 18424920,
      "step": 329
    },
    {
      "epoch": 0.734966592427617,
      "grad_norm": 29.17088508605957,
      "learning_rate": 1e-06,
      "loss": 1.1757,
      "num_input_tokens_seen": 18478976,
      "step": 330
    },
    {
      "epoch": 0.734966592427617,
      "loss": 1.1004878282546997,
      "loss_ce": 0.005761317443102598,
      "loss_iou": 0.453125,
      "loss_num": 0.037841796875,
      "loss_xval": 1.09375,
      "num_input_tokens_seen": 18478976,
      "step": 330
    },
    {
      "epoch": 0.7371937639198218,
      "grad_norm": 20.834753036499023,
      "learning_rate": 1e-06,
      "loss": 1.1889,
      "num_input_tokens_seen": 18536144,
      "step": 331
    },
    {
      "epoch": 0.7371937639198218,
      "loss": 1.3901407718658447,
      "loss_ce": 0.0014689104864373803,
      "loss_iou": 0.578125,
      "loss_num": 0.04638671875,
      "loss_xval": 1.390625,
      "num_input_tokens_seen": 18536144,
      "step": 331
    },
    {
      "epoch": 0.7394209354120267,
      "grad_norm": 68.39385986328125,
      "learning_rate": 1e-06,
      "loss": 1.5586,
      "num_input_tokens_seen": 18593860,
      "step": 332
    },
    {
      "epoch": 0.7394209354120267,
      "loss": 1.5588513612747192,
      "loss_ce": 0.005140438210219145,
      "loss_iou": 0.6484375,
      "loss_num": 0.052001953125,
      "loss_xval": 1.5546875,
      "num_input_tokens_seen": 18593860,
      "step": 332
    },
    {
      "epoch": 0.7416481069042317,
      "grad_norm": 19.545167922973633,
      "learning_rate": 1e-06,
      "loss": 1.5776,
      "num_input_tokens_seen": 18647892,
      "step": 333
    },
    {
      "epoch": 0.7416481069042317,
      "loss": 1.6709346771240234,
      "loss_ce": 0.017126010730862617,
      "loss_iou": 0.640625,
      "loss_num": 0.07421875,
      "loss_xval": 1.65625,
      "num_input_tokens_seen": 18647892,
      "step": 333
    },
    {
      "epoch": 0.7438752783964365,
      "grad_norm": 47.12783432006836,
      "learning_rate": 1e-06,
      "loss": 1.503,
      "num_input_tokens_seen": 18704784,
      "step": 334
    },
    {
      "epoch": 0.7438752783964365,
      "loss": 1.865045428276062,
      "loss_ce": 0.012994609773159027,
      "loss_iou": 0.796875,
      "loss_num": 0.052001953125,
      "loss_xval": 1.8515625,
      "num_input_tokens_seen": 18704784,
      "step": 334
    },
    {
      "epoch": 0.7461024498886414,
      "grad_norm": 25.380210876464844,
      "learning_rate": 1e-06,
      "loss": 1.1623,
      "num_input_tokens_seen": 18761996,
      "step": 335
    },
    {
      "epoch": 0.7461024498886414,
      "loss": 1.0759522914886475,
      "loss_ce": 0.015649594366550446,
      "loss_iou": 0.423828125,
      "loss_num": 0.042724609375,
      "loss_xval": 1.0625,
      "num_input_tokens_seen": 18761996,
      "step": 335
    },
    {
      "epoch": 0.7483296213808464,
      "grad_norm": 19.682802200317383,
      "learning_rate": 1e-06,
      "loss": 1.4553,
      "num_input_tokens_seen": 18819652,
      "step": 336
    },
    {
      "epoch": 0.7483296213808464,
      "loss": 1.3878556489944458,
      "loss_ce": 0.011879058554768562,
      "loss_iou": 0.5546875,
      "loss_num": 0.053466796875,
      "loss_xval": 1.375,
      "num_input_tokens_seen": 18819652,
      "step": 336
    },
    {
      "epoch": 0.7505567928730512,
      "grad_norm": 91.77301025390625,
      "learning_rate": 1e-06,
      "loss": 1.6139,
      "num_input_tokens_seen": 18877576,
      "step": 337
    },
    {
      "epoch": 0.7505567928730512,
      "loss": 1.334415316581726,
      "loss_ce": 0.008731753565371037,
      "loss_iou": 0.546875,
      "loss_num": 0.0458984375,
      "loss_xval": 1.328125,
      "num_input_tokens_seen": 18877576,
      "step": 337
    },
    {
      "epoch": 0.7527839643652561,
      "grad_norm": 28.926156997680664,
      "learning_rate": 1e-06,
      "loss": 1.5567,
      "num_input_tokens_seen": 18935300,
      "step": 338
    },
    {
      "epoch": 0.7527839643652561,
      "loss": 1.7415505647659302,
      "loss_ce": 0.01205837819725275,
      "loss_iou": 0.671875,
      "loss_num": 0.076171875,
      "loss_xval": 1.7265625,
      "num_input_tokens_seen": 18935300,
      "step": 338
    },
    {
      "epoch": 0.755011135857461,
      "grad_norm": 36.80242919921875,
      "learning_rate": 1e-06,
      "loss": 1.783,
      "num_input_tokens_seen": 18989472,
      "step": 339
    },
    {
      "epoch": 0.755011135857461,
      "loss": 1.8453316688537598,
      "loss_ce": 0.012812146916985512,
      "loss_iou": 0.77734375,
      "loss_num": 0.0546875,
      "loss_xval": 1.8359375,
      "num_input_tokens_seen": 18989472,
      "step": 339
    },
    {
      "epoch": 0.7572383073496659,
      "grad_norm": 19.47812271118164,
      "learning_rate": 1e-06,
      "loss": 1.5968,
      "num_input_tokens_seen": 19041164,
      "step": 340
    },
    {
      "epoch": 0.7572383073496659,
      "loss": 1.3463078737258911,
      "loss_ce": 0.017938785254955292,
      "loss_iou": 0.56640625,
      "loss_num": 0.039306640625,
      "loss_xval": 1.328125,
      "num_input_tokens_seen": 19041164,
      "step": 340
    },
    {
      "epoch": 0.7594654788418709,
      "grad_norm": 21.5255126953125,
      "learning_rate": 1e-06,
      "loss": 1.2247,
      "num_input_tokens_seen": 19098404,
      "step": 341
    },
    {
      "epoch": 0.7594654788418709,
      "loss": 1.1793166399002075,
      "loss_ce": 0.015254099853336811,
      "loss_iou": 0.470703125,
      "loss_num": 0.04443359375,
      "loss_xval": 1.1640625,
      "num_input_tokens_seen": 19098404,
      "step": 341
    },
    {
      "epoch": 0.7616926503340757,
      "grad_norm": 21.701622009277344,
      "learning_rate": 1e-06,
      "loss": 1.4873,
      "num_input_tokens_seen": 19152172,
      "step": 342
    },
    {
      "epoch": 0.7616926503340757,
      "loss": 1.5115392208099365,
      "loss_ce": 0.016422055661678314,
      "loss_iou": 0.5859375,
      "loss_num": 0.0634765625,
      "loss_xval": 1.4921875,
      "num_input_tokens_seen": 19152172,
      "step": 342
    },
    {
      "epoch": 0.7639198218262806,
      "grad_norm": 24.85000228881836,
      "learning_rate": 1e-06,
      "loss": 1.3042,
      "num_input_tokens_seen": 19208312,
      "step": 343
    },
    {
      "epoch": 0.7639198218262806,
      "loss": 1.3061310052871704,
      "loss_ce": 0.011697422713041306,
      "loss_iou": 0.54296875,
      "loss_num": 0.041259765625,
      "loss_xval": 1.296875,
      "num_input_tokens_seen": 19208312,
      "step": 343
    },
    {
      "epoch": 0.7661469933184856,
      "grad_norm": 28.263137817382812,
      "learning_rate": 1e-06,
      "loss": 1.4686,
      "num_input_tokens_seen": 19263288,
      "step": 344
    },
    {
      "epoch": 0.7661469933184856,
      "loss": 1.6164791584014893,
      "loss_ce": 0.008080787025392056,
      "loss_iou": 0.64453125,
      "loss_num": 0.064453125,
      "loss_xval": 1.609375,
      "num_input_tokens_seen": 19263288,
      "step": 344
    },
    {
      "epoch": 0.7683741648106904,
      "grad_norm": 26.359323501586914,
      "learning_rate": 1e-06,
      "loss": 1.5675,
      "num_input_tokens_seen": 19322276,
      "step": 345
    },
    {
      "epoch": 0.7683741648106904,
      "loss": 1.6035141944885254,
      "loss_ce": 0.007078767288476229,
      "loss_iou": 0.6484375,
      "loss_num": 0.0595703125,
      "loss_xval": 1.59375,
      "num_input_tokens_seen": 19322276,
      "step": 345
    },
    {
      "epoch": 0.7706013363028953,
      "grad_norm": 28.58954429626465,
      "learning_rate": 1e-06,
      "loss": 1.4959,
      "num_input_tokens_seen": 19377088,
      "step": 346
    },
    {
      "epoch": 0.7706013363028953,
      "loss": 1.7983763217926025,
      "loss_ce": 0.02982161194086075,
      "loss_iou": 0.65234375,
      "loss_num": 0.09326171875,
      "loss_xval": 1.765625,
      "num_input_tokens_seen": 19377088,
      "step": 346
    },
    {
      "epoch": 0.7728285077951003,
      "grad_norm": 32.53110885620117,
      "learning_rate": 1e-06,
      "loss": 1.468,
      "num_input_tokens_seen": 19432868,
      "step": 347
    },
    {
      "epoch": 0.7728285077951003,
      "loss": 1.644303560256958,
      "loss_ce": 0.006120047532021999,
      "loss_iou": 0.703125,
      "loss_num": 0.04638671875,
      "loss_xval": 1.640625,
      "num_input_tokens_seen": 19432868,
      "step": 347
    },
    {
      "epoch": 0.7750556792873051,
      "grad_norm": 19.344207763671875,
      "learning_rate": 1e-06,
      "loss": 1.4323,
      "num_input_tokens_seen": 19490488,
      "step": 348
    },
    {
      "epoch": 0.7750556792873051,
      "loss": 1.4876625537872314,
      "loss_ce": 0.005729038268327713,
      "loss_iou": 0.5546875,
      "loss_num": 0.07373046875,
      "loss_xval": 1.484375,
      "num_input_tokens_seen": 19490488,
      "step": 348
    },
    {
      "epoch": 0.77728285077951,
      "grad_norm": 142.5430145263672,
      "learning_rate": 1e-06,
      "loss": 1.4212,
      "num_input_tokens_seen": 19546108,
      "step": 349
    },
    {
      "epoch": 0.77728285077951,
      "loss": 1.081338882446289,
      "loss_ce": 0.0024814759381115437,
      "loss_iou": 0.4453125,
      "loss_num": 0.037841796875,
      "loss_xval": 1.078125,
      "num_input_tokens_seen": 19546108,
      "step": 349
    },
    {
      "epoch": 0.779510022271715,
      "grad_norm": 27.834339141845703,
      "learning_rate": 1e-06,
      "loss": 1.4574,
      "num_input_tokens_seen": 19602608,
      "step": 350
    },
    {
      "epoch": 0.779510022271715,
      "loss": 1.220694899559021,
      "loss_ce": 0.0029214350506663322,
      "loss_iou": 0.494140625,
      "loss_num": 0.0458984375,
      "loss_xval": 1.21875,
      "num_input_tokens_seen": 19602608,
      "step": 350
    },
    {
      "epoch": 0.7817371937639198,
      "grad_norm": 51.78019332885742,
      "learning_rate": 1e-06,
      "loss": 1.3015,
      "num_input_tokens_seen": 19659844,
      "step": 351
    },
    {
      "epoch": 0.7817371937639198,
      "loss": 0.9219905138015747,
      "loss_ce": 0.007439759094268084,
      "loss_iou": 0.388671875,
      "loss_num": 0.0277099609375,
      "loss_xval": 0.9140625,
      "num_input_tokens_seen": 19659844,
      "step": 351
    },
    {
      "epoch": 0.7839643652561247,
      "grad_norm": 19.256717681884766,
      "learning_rate": 1e-06,
      "loss": 1.552,
      "num_input_tokens_seen": 19716640,
      "step": 352
    },
    {
      "epoch": 0.7839643652561247,
      "loss": 1.5087199211120605,
      "loss_ce": 0.00481368275359273,
      "loss_iou": 0.6171875,
      "loss_num": 0.0537109375,
      "loss_xval": 1.5,
      "num_input_tokens_seen": 19716640,
      "step": 352
    },
    {
      "epoch": 0.7861915367483296,
      "grad_norm": 25.72173500061035,
      "learning_rate": 1e-06,
      "loss": 1.4851,
      "num_input_tokens_seen": 19771512,
      "step": 353
    },
    {
      "epoch": 0.7861915367483296,
      "loss": 1.3254516124725342,
      "loss_ce": 0.028576675802469254,
      "loss_iou": 0.51953125,
      "loss_num": 0.051025390625,
      "loss_xval": 1.296875,
      "num_input_tokens_seen": 19771512,
      "step": 353
    },
    {
      "epoch": 0.7884187082405345,
      "grad_norm": 27.796306610107422,
      "learning_rate": 1e-06,
      "loss": 1.1745,
      "num_input_tokens_seen": 19827200,
      "step": 354
    },
    {
      "epoch": 0.7884187082405345,
      "loss": 1.0663065910339355,
      "loss_ce": 0.0013651238987222314,
      "loss_iou": 0.458984375,
      "loss_num": 0.029296875,
      "loss_xval": 1.0625,
      "num_input_tokens_seen": 19827200,
      "step": 354
    },
    {
      "epoch": 0.7906458797327395,
      "grad_norm": 29.534481048583984,
      "learning_rate": 1e-06,
      "loss": 1.2979,
      "num_input_tokens_seen": 19883360,
      "step": 355
    },
    {
      "epoch": 0.7906458797327395,
      "loss": 1.285064697265625,
      "loss_ce": 0.014556895941495895,
      "loss_iou": 0.484375,
      "loss_num": 0.060791015625,
      "loss_xval": 1.2734375,
      "num_input_tokens_seen": 19883360,
      "step": 355
    },
    {
      "epoch": 0.7928730512249443,
      "grad_norm": 20.099090576171875,
      "learning_rate": 1e-06,
      "loss": 1.4715,
      "num_input_tokens_seen": 19940216,
      "step": 356
    },
    {
      "epoch": 0.7928730512249443,
      "loss": 1.6332509517669678,
      "loss_ce": 0.00483296625316143,
      "loss_iou": 0.6640625,
      "loss_num": 0.060546875,
      "loss_xval": 1.625,
      "num_input_tokens_seen": 19940216,
      "step": 356
    },
    {
      "epoch": 0.7951002227171492,
      "grad_norm": 23.28925323486328,
      "learning_rate": 1e-06,
      "loss": 1.4499,
      "num_input_tokens_seen": 19996716,
      "step": 357
    },
    {
      "epoch": 0.7951002227171492,
      "loss": 1.103065848350525,
      "loss_ce": 0.007850958965718746,
      "loss_iou": 0.453125,
      "loss_num": 0.038330078125,
      "loss_xval": 1.09375,
      "num_input_tokens_seen": 19996716,
      "step": 357
    },
    {
      "epoch": 0.7973273942093542,
      "grad_norm": 26.855409622192383,
      "learning_rate": 1e-06,
      "loss": 1.1852,
      "num_input_tokens_seen": 20052192,
      "step": 358
    },
    {
      "epoch": 0.7973273942093542,
      "loss": 1.3929839134216309,
      "loss_ce": 0.003823714330792427,
      "loss_iou": 0.5703125,
      "loss_num": 0.050537109375,
      "loss_xval": 1.390625,
      "num_input_tokens_seen": 20052192,
      "step": 358
    },
    {
      "epoch": 0.799554565701559,
      "grad_norm": 32.07172393798828,
      "learning_rate": 1e-06,
      "loss": 1.5933,
      "num_input_tokens_seen": 20110068,
      "step": 359
    },
    {
      "epoch": 0.799554565701559,
      "loss": 1.588189721107483,
      "loss_ce": 0.005181873217225075,
      "loss_iou": 0.65234375,
      "loss_num": 0.05517578125,
      "loss_xval": 1.5859375,
      "num_input_tokens_seen": 20110068,
      "step": 359
    },
    {
      "epoch": 0.8017817371937639,
      "grad_norm": 32.26838684082031,
      "learning_rate": 1e-06,
      "loss": 1.451,
      "num_input_tokens_seen": 20166760,
      "step": 360
    },
    {
      "epoch": 0.8017817371937639,
      "loss": 1.4686858654022217,
      "loss_ce": 0.017513982951641083,
      "loss_iou": 0.609375,
      "loss_num": 0.04638671875,
      "loss_xval": 1.453125,
      "num_input_tokens_seen": 20166760,
      "step": 360
    },
    {
      "epoch": 0.8040089086859689,
      "grad_norm": 31.04205894470215,
      "learning_rate": 1e-06,
      "loss": 1.3697,
      "num_input_tokens_seen": 20223556,
      "step": 361
    },
    {
      "epoch": 0.8040089086859689,
      "loss": 1.2945581674575806,
      "loss_ce": 0.006228114478290081,
      "loss_iou": 0.515625,
      "loss_num": 0.05078125,
      "loss_xval": 1.2890625,
      "num_input_tokens_seen": 20223556,
      "step": 361
    },
    {
      "epoch": 0.8062360801781737,
      "grad_norm": 28.388748168945312,
      "learning_rate": 1e-06,
      "loss": 1.4442,
      "num_input_tokens_seen": 20283264,
      "step": 362
    },
    {
      "epoch": 0.8062360801781737,
      "loss": 1.3582143783569336,
      "loss_ce": 0.002745626959949732,
      "loss_iou": 0.5625,
      "loss_num": 0.04541015625,
      "loss_xval": 1.359375,
      "num_input_tokens_seen": 20283264,
      "step": 362
    },
    {
      "epoch": 0.8084632516703786,
      "grad_norm": 49.78532028198242,
      "learning_rate": 1e-06,
      "loss": 1.3255,
      "num_input_tokens_seen": 20337404,
      "step": 363
    },
    {
      "epoch": 0.8084632516703786,
      "loss": 1.1450395584106445,
      "loss_ce": 0.015156792476773262,
      "loss_iou": 0.412109375,
      "loss_num": 0.061767578125,
      "loss_xval": 1.1328125,
      "num_input_tokens_seen": 20337404,
      "step": 363
    },
    {
      "epoch": 0.8106904231625836,
      "grad_norm": 48.754234313964844,
      "learning_rate": 1e-06,
      "loss": 1.2584,
      "num_input_tokens_seen": 20393236,
      "step": 364
    },
    {
      "epoch": 0.8106904231625836,
      "loss": 1.1171329021453857,
      "loss_ce": 0.007513846270740032,
      "loss_iou": 0.4453125,
      "loss_num": 0.04345703125,
      "loss_xval": 1.109375,
      "num_input_tokens_seen": 20393236,
      "step": 364
    },
    {
      "epoch": 0.8129175946547884,
      "grad_norm": 49.22020721435547,
      "learning_rate": 1e-06,
      "loss": 1.4507,
      "num_input_tokens_seen": 20451716,
      "step": 365
    },
    {
      "epoch": 0.8129175946547884,
      "loss": 1.6032345294952393,
      "loss_ce": 0.046593837440013885,
      "loss_iou": 0.6171875,
      "loss_num": 0.064453125,
      "loss_xval": 1.5546875,
      "num_input_tokens_seen": 20451716,
      "step": 365
    },
    {
      "epoch": 0.8151447661469933,
      "grad_norm": 81.98351287841797,
      "learning_rate": 1e-06,
      "loss": 1.5286,
      "num_input_tokens_seen": 20507752,
      "step": 366
    },
    {
      "epoch": 0.8151447661469933,
      "loss": 1.5305746793746948,
      "loss_ce": 0.0032309277448803186,
      "loss_iou": 0.62890625,
      "loss_num": 0.054443359375,
      "loss_xval": 1.53125,
      "num_input_tokens_seen": 20507752,
      "step": 366
    },
    {
      "epoch": 0.8173719376391982,
      "grad_norm": 20.340085983276367,
      "learning_rate": 1e-06,
      "loss": 1.245,
      "num_input_tokens_seen": 20561536,
      "step": 367
    },
    {
      "epoch": 0.8173719376391982,
      "loss": 1.0740153789520264,
      "loss_ce": 0.01566578447818756,
      "loss_iou": 0.400390625,
      "loss_num": 0.05126953125,
      "loss_xval": 1.0546875,
      "num_input_tokens_seen": 20561536,
      "step": 367
    },
    {
      "epoch": 0.8195991091314031,
      "grad_norm": 28.017810821533203,
      "learning_rate": 1e-06,
      "loss": 1.4366,
      "num_input_tokens_seen": 20614508,
      "step": 368
    },
    {
      "epoch": 0.8195991091314031,
      "loss": 1.5188279151916504,
      "loss_ce": 0.033476315438747406,
      "loss_iou": 0.6015625,
      "loss_num": 0.05615234375,
      "loss_xval": 1.484375,
      "num_input_tokens_seen": 20614508,
      "step": 368
    },
    {
      "epoch": 0.821826280623608,
      "grad_norm": 26.68250846862793,
      "learning_rate": 1e-06,
      "loss": 1.1561,
      "num_input_tokens_seen": 20669580,
      "step": 369
    },
    {
      "epoch": 0.821826280623608,
      "loss": 1.2030680179595947,
      "loss_ce": 0.0014078648528084159,
      "loss_iou": 0.5,
      "loss_num": 0.040283203125,
      "loss_xval": 1.203125,
      "num_input_tokens_seen": 20669580,
      "step": 369
    },
    {
      "epoch": 0.8240534521158129,
      "grad_norm": 19.57592010498047,
      "learning_rate": 1e-06,
      "loss": 1.2601,
      "num_input_tokens_seen": 20723132,
      "step": 370
    },
    {
      "epoch": 0.8240534521158129,
      "loss": 1.2386668920516968,
      "loss_ce": 0.013080945238471031,
      "loss_iou": 0.5390625,
      "loss_num": 0.030029296875,
      "loss_xval": 1.2265625,
      "num_input_tokens_seen": 20723132,
      "step": 370
    },
    {
      "epoch": 0.8262806236080178,
      "grad_norm": 38.58708190917969,
      "learning_rate": 1e-06,
      "loss": 1.5873,
      "num_input_tokens_seen": 20779144,
      "step": 371
    },
    {
      "epoch": 0.8262806236080178,
      "loss": 1.2857810258865356,
      "loss_ce": 0.013320127502083778,
      "loss_iou": 0.49609375,
      "loss_num": 0.055908203125,
      "loss_xval": 1.2734375,
      "num_input_tokens_seen": 20779144,
      "step": 371
    },
    {
      "epoch": 0.8285077951002228,
      "grad_norm": 27.13821792602539,
      "learning_rate": 1e-06,
      "loss": 1.7652,
      "num_input_tokens_seen": 20835076,
      "step": 372
    },
    {
      "epoch": 0.8285077951002228,
      "loss": 1.7804100513458252,
      "loss_ce": 0.006972476840019226,
      "loss_iou": 0.69921875,
      "loss_num": 0.0751953125,
      "loss_xval": 1.7734375,
      "num_input_tokens_seen": 20835076,
      "step": 372
    },
    {
      "epoch": 0.8307349665924276,
      "grad_norm": 18.151918411254883,
      "learning_rate": 1e-06,
      "loss": 1.4218,
      "num_input_tokens_seen": 20891112,
      "step": 373
    },
    {
      "epoch": 0.8307349665924276,
      "loss": 1.4063541889190674,
      "loss_ce": 0.005475334823131561,
      "loss_iou": 0.56640625,
      "loss_num": 0.052978515625,
      "loss_xval": 1.3984375,
      "num_input_tokens_seen": 20891112,
      "step": 373
    },
    {
      "epoch": 0.8329621380846325,
      "grad_norm": 20.646316528320312,
      "learning_rate": 1e-06,
      "loss": 1.2937,
      "num_input_tokens_seen": 20947924,
      "step": 374
    },
    {
      "epoch": 0.8329621380846325,
      "loss": 1.1752688884735107,
      "loss_ce": 0.02146025560796261,
      "loss_iou": 0.466796875,
      "loss_num": 0.043701171875,
      "loss_xval": 1.15625,
      "num_input_tokens_seen": 20947924,
      "step": 374
    },
    {
      "epoch": 0.8351893095768375,
      "grad_norm": 65.90545654296875,
      "learning_rate": 1e-06,
      "loss": 1.1793,
      "num_input_tokens_seen": 21003736,
      "step": 375
    },
    {
      "epoch": 0.8351893095768375,
      "loss": 1.0895462036132812,
      "loss_ce": 0.006538336630910635,
      "loss_iou": 0.4609375,
      "loss_num": 0.0322265625,
      "loss_xval": 1.0859375,
      "num_input_tokens_seen": 21003736,
      "step": 375
    },
    {
      "epoch": 0.8374164810690423,
      "grad_norm": 30.747459411621094,
      "learning_rate": 1e-06,
      "loss": 1.4332,
      "num_input_tokens_seen": 21058432,
      "step": 376
    },
    {
      "epoch": 0.8374164810690423,
      "loss": 1.4749113321304321,
      "loss_ce": 0.005673029460012913,
      "loss_iou": 0.54296875,
      "loss_num": 0.076171875,
      "loss_xval": 1.46875,
      "num_input_tokens_seen": 21058432,
      "step": 376
    },
    {
      "epoch": 0.8396436525612472,
      "grad_norm": 48.859676361083984,
      "learning_rate": 1e-06,
      "loss": 1.288,
      "num_input_tokens_seen": 21114724,
      "step": 377
    },
    {
      "epoch": 0.8396436525612472,
      "loss": 1.3059370517730713,
      "loss_ce": 0.02029254473745823,
      "loss_iou": 0.53515625,
      "loss_num": 0.0439453125,
      "loss_xval": 1.2890625,
      "num_input_tokens_seen": 21114724,
      "step": 377
    },
    {
      "epoch": 0.8418708240534521,
      "grad_norm": 34.50962448120117,
      "learning_rate": 1e-06,
      "loss": 1.237,
      "num_input_tokens_seen": 21169136,
      "step": 378
    },
    {
      "epoch": 0.8418708240534521,
      "loss": 1.1840803623199463,
      "loss_ce": 0.0024397093802690506,
      "loss_iou": 0.44921875,
      "loss_num": 0.05712890625,
      "loss_xval": 1.1796875,
      "num_input_tokens_seen": 21169136,
      "step": 378
    },
    {
      "epoch": 0.844097995545657,
      "grad_norm": 50.90530776977539,
      "learning_rate": 1e-06,
      "loss": 1.3962,
      "num_input_tokens_seen": 21225176,
      "step": 379
    },
    {
      "epoch": 0.844097995545657,
      "loss": 1.6650288105010986,
      "loss_ce": 0.02660096064209938,
      "loss_iou": 0.65625,
      "loss_num": 0.0654296875,
      "loss_xval": 1.640625,
      "num_input_tokens_seen": 21225176,
      "step": 379
    },
    {
      "epoch": 0.8463251670378619,
      "grad_norm": 30.996505737304688,
      "learning_rate": 1e-06,
      "loss": 1.2789,
      "num_input_tokens_seen": 21284272,
      "step": 380
    },
    {
      "epoch": 0.8463251670378619,
      "loss": 1.3432170152664185,
      "loss_ce": 0.0019084562081843615,
      "loss_iou": 0.52734375,
      "loss_num": 0.056884765625,
      "loss_xval": 1.34375,
      "num_input_tokens_seen": 21284272,
      "step": 380
    },
    {
      "epoch": 0.8485523385300668,
      "grad_norm": 20.33319854736328,
      "learning_rate": 1e-06,
      "loss": 1.1925,
      "num_input_tokens_seen": 21339980,
      "step": 381
    },
    {
      "epoch": 0.8485523385300668,
      "loss": 1.164018154144287,
      "loss_ce": 0.0019087546970695257,
      "loss_iou": 0.47265625,
      "loss_num": 0.043212890625,
      "loss_xval": 1.1640625,
      "num_input_tokens_seen": 21339980,
      "step": 381
    },
    {
      "epoch": 0.8507795100222717,
      "grad_norm": 47.0328369140625,
      "learning_rate": 1e-06,
      "loss": 1.4929,
      "num_input_tokens_seen": 21397864,
      "step": 382
    },
    {
      "epoch": 0.8507795100222717,
      "loss": 1.355049729347229,
      "loss_ce": 0.005440343637019396,
      "loss_iou": 0.55859375,
      "loss_num": 0.046142578125,
      "loss_xval": 1.3515625,
      "num_input_tokens_seen": 21397864,
      "step": 382
    },
    {
      "epoch": 0.8530066815144766,
      "grad_norm": 84.89624786376953,
      "learning_rate": 1e-06,
      "loss": 1.4523,
      "num_input_tokens_seen": 21453536,
      "step": 383
    },
    {
      "epoch": 0.8530066815144766,
      "loss": 1.6871811151504517,
      "loss_ce": 0.00944666936993599,
      "loss_iou": 0.65625,
      "loss_num": 0.0732421875,
      "loss_xval": 1.6796875,
      "num_input_tokens_seen": 21453536,
      "step": 383
    },
    {
      "epoch": 0.8552338530066815,
      "grad_norm": 17.764833450317383,
      "learning_rate": 1e-06,
      "loss": 1.4082,
      "num_input_tokens_seen": 21508716,
      "step": 384
    },
    {
      "epoch": 0.8552338530066815,
      "loss": 1.6076984405517578,
      "loss_ce": 0.015901662409305573,
      "loss_iou": 0.66015625,
      "loss_num": 0.0546875,
      "loss_xval": 1.59375,
      "num_input_tokens_seen": 21508716,
      "step": 384
    },
    {
      "epoch": 0.8574610244988864,
      "grad_norm": 21.195423126220703,
      "learning_rate": 1e-06,
      "loss": 1.1628,
      "num_input_tokens_seen": 21563776,
      "step": 385
    },
    {
      "epoch": 0.8574610244988864,
      "loss": 1.30448317527771,
      "loss_ce": 0.010049499571323395,
      "loss_iou": 0.546875,
      "loss_num": 0.039794921875,
      "loss_xval": 1.296875,
      "num_input_tokens_seen": 21563776,
      "step": 385
    },
    {
      "epoch": 0.8596881959910914,
      "grad_norm": 31.222137451171875,
      "learning_rate": 1e-06,
      "loss": 1.2988,
      "num_input_tokens_seen": 21617956,
      "step": 386
    },
    {
      "epoch": 0.8596881959910914,
      "loss": 1.2785615921020508,
      "loss_ce": 0.0092744380235672,
      "loss_iou": 0.49609375,
      "loss_num": 0.05517578125,
      "loss_xval": 1.265625,
      "num_input_tokens_seen": 21617956,
      "step": 386
    },
    {
      "epoch": 0.8619153674832962,
      "grad_norm": 43.02681350708008,
      "learning_rate": 1e-06,
      "loss": 1.5262,
      "num_input_tokens_seen": 21673748,
      "step": 387
    },
    {
      "epoch": 0.8619153674832962,
      "loss": 1.531335711479187,
      "loss_ce": 0.003991919104009867,
      "loss_iou": 0.58203125,
      "loss_num": 0.07177734375,
      "loss_xval": 1.53125,
      "num_input_tokens_seen": 21673748,
      "step": 387
    },
    {
      "epoch": 0.8641425389755011,
      "grad_norm": 41.39469909667969,
      "learning_rate": 1e-06,
      "loss": 1.2424,
      "num_input_tokens_seen": 21729848,
      "step": 388
    },
    {
      "epoch": 0.8641425389755011,
      "loss": 1.398338794708252,
      "loss_ce": 0.0013661817647516727,
      "loss_iou": 0.55859375,
      "loss_num": 0.055419921875,
      "loss_xval": 1.3984375,
      "num_input_tokens_seen": 21729848,
      "step": 388
    },
    {
      "epoch": 0.8663697104677061,
      "grad_norm": 24.085895538330078,
      "learning_rate": 1e-06,
      "loss": 1.3652,
      "num_input_tokens_seen": 21789184,
      "step": 389
    },
    {
      "epoch": 0.8663697104677061,
      "loss": 1.4368267059326172,
      "loss_ce": 0.012022039853036404,
      "loss_iou": 0.57421875,
      "loss_num": 0.0546875,
      "loss_xval": 1.421875,
      "num_input_tokens_seen": 21789184,
      "step": 389
    },
    {
      "epoch": 0.8685968819599109,
      "grad_norm": 21.240936279296875,
      "learning_rate": 1e-06,
      "loss": 1.1302,
      "num_input_tokens_seen": 21845372,
      "step": 390
    },
    {
      "epoch": 0.8685968819599109,
      "loss": 1.243898868560791,
      "loss_ce": 0.01001210231333971,
      "loss_iou": 0.515625,
      "loss_num": 0.040283203125,
      "loss_xval": 1.234375,
      "num_input_tokens_seen": 21845372,
      "step": 390
    },
    {
      "epoch": 0.8708240534521158,
      "grad_norm": 26.22933578491211,
      "learning_rate": 1e-06,
      "loss": 1.2685,
      "num_input_tokens_seen": 21898080,
      "step": 391
    },
    {
      "epoch": 0.8708240534521158,
      "loss": 1.6092109680175781,
      "loss_ce": 0.0179024338722229,
      "loss_iou": 0.65234375,
      "loss_num": 0.05810546875,
      "loss_xval": 1.59375,
      "num_input_tokens_seen": 21898080,
      "step": 391
    },
    {
      "epoch": 0.8730512249443207,
      "grad_norm": 68.2789077758789,
      "learning_rate": 1e-06,
      "loss": 1.4469,
      "num_input_tokens_seen": 21954884,
      "step": 392
    },
    {
      "epoch": 0.8730512249443207,
      "loss": 1.568161964416504,
      "loss_ce": 0.002732297871261835,
      "loss_iou": 0.64453125,
      "loss_num": 0.0556640625,
      "loss_xval": 1.5625,
      "num_input_tokens_seen": 21954884,
      "step": 392
    },
    {
      "epoch": 0.8752783964365256,
      "grad_norm": 27.22370719909668,
      "learning_rate": 1e-06,
      "loss": 1.2137,
      "num_input_tokens_seen": 22008924,
      "step": 393
    },
    {
      "epoch": 0.8752783964365256,
      "loss": 1.0929490327835083,
      "loss_ce": 0.004081860650330782,
      "loss_iou": 0.44921875,
      "loss_num": 0.038330078125,
      "loss_xval": 1.0859375,
      "num_input_tokens_seen": 22008924,
      "step": 393
    },
    {
      "epoch": 0.8775055679287305,
      "grad_norm": 82.91708374023438,
      "learning_rate": 1e-06,
      "loss": 1.6931,
      "num_input_tokens_seen": 22064736,
      "step": 394
    },
    {
      "epoch": 0.8775055679287305,
      "loss": 1.915067434310913,
      "loss_ce": 0.007840840145945549,
      "loss_iou": 0.765625,
      "loss_num": 0.07568359375,
      "loss_xval": 1.90625,
      "num_input_tokens_seen": 22064736,
      "step": 394
    },
    {
      "epoch": 0.8797327394209354,
      "grad_norm": 25.599044799804688,
      "learning_rate": 1e-06,
      "loss": 1.3419,
      "num_input_tokens_seen": 22119196,
      "step": 395
    },
    {
      "epoch": 0.8797327394209354,
      "loss": 1.3939309120178223,
      "loss_ce": 0.003794201649725437,
      "loss_iou": 0.55078125,
      "loss_num": 0.05712890625,
      "loss_xval": 1.390625,
      "num_input_tokens_seen": 22119196,
      "step": 395
    },
    {
      "epoch": 0.8819599109131403,
      "grad_norm": 38.43741226196289,
      "learning_rate": 1e-06,
      "loss": 1.3195,
      "num_input_tokens_seen": 22175900,
      "step": 396
    },
    {
      "epoch": 0.8819599109131403,
      "loss": 1.2486144304275513,
      "loss_ce": 0.013262815773487091,
      "loss_iou": 0.4921875,
      "loss_num": 0.050048828125,
      "loss_xval": 1.234375,
      "num_input_tokens_seen": 22175900,
      "step": 396
    },
    {
      "epoch": 0.8841870824053452,
      "grad_norm": 51.51115417480469,
      "learning_rate": 1e-06,
      "loss": 1.3812,
      "num_input_tokens_seen": 22232828,
      "step": 397
    },
    {
      "epoch": 0.8841870824053452,
      "loss": 1.6601974964141846,
      "loss_ce": 0.029338089749217033,
      "loss_iou": 0.6171875,
      "loss_num": 0.080078125,
      "loss_xval": 1.6328125,
      "num_input_tokens_seen": 22232828,
      "step": 397
    },
    {
      "epoch": 0.8864142538975501,
      "grad_norm": 29.84437370300293,
      "learning_rate": 1e-06,
      "loss": 1.145,
      "num_input_tokens_seen": 22290996,
      "step": 398
    },
    {
      "epoch": 0.8864142538975501,
      "loss": 1.0916481018066406,
      "loss_ce": 0.0008277894230559468,
      "loss_iou": 0.412109375,
      "loss_num": 0.053466796875,
      "loss_xval": 1.09375,
      "num_input_tokens_seen": 22290996,
      "step": 398
    },
    {
      "epoch": 0.888641425389755,
      "grad_norm": 34.59022903442383,
      "learning_rate": 1e-06,
      "loss": 1.1759,
      "num_input_tokens_seen": 22345820,
      "step": 399
    },
    {
      "epoch": 0.888641425389755,
      "loss": 1.1103270053863525,
      "loss_ce": 0.0009520421735942364,
      "loss_iou": 0.46875,
      "loss_num": 0.0341796875,
      "loss_xval": 1.109375,
      "num_input_tokens_seen": 22345820,
      "step": 399
    },
    {
      "epoch": 0.89086859688196,
      "grad_norm": 36.79861068725586,
      "learning_rate": 1e-06,
      "loss": 1.3186,
      "num_input_tokens_seen": 22403684,
      "step": 400
    },
    {
      "epoch": 0.89086859688196,
      "loss": 1.1463134288787842,
      "loss_ce": 0.005688320379704237,
      "loss_iou": 0.47265625,
      "loss_num": 0.0390625,
      "loss_xval": 1.140625,
      "num_input_tokens_seen": 22403684,
      "step": 400
    },
    {
      "epoch": 0.8930957683741648,
      "grad_norm": 21.335620880126953,
      "learning_rate": 1e-06,
      "loss": 1.3037,
      "num_input_tokens_seen": 22461644,
      "step": 401
    },
    {
      "epoch": 0.8930957683741648,
      "loss": 1.3235667943954468,
      "loss_ce": 0.007160472683608532,
      "loss_iou": 0.5390625,
      "loss_num": 0.048095703125,
      "loss_xval": 1.3125,
      "num_input_tokens_seen": 22461644,
      "step": 401
    },
    {
      "epoch": 0.8953229398663697,
      "grad_norm": 50.855709075927734,
      "learning_rate": 1e-06,
      "loss": 1.6207,
      "num_input_tokens_seen": 22518540,
      "step": 402
    },
    {
      "epoch": 0.8953229398663697,
      "loss": 1.8505743741989136,
      "loss_ce": 0.002918071812018752,
      "loss_iou": 0.71875,
      "loss_num": 0.08203125,
      "loss_xval": 1.84375,
      "num_input_tokens_seen": 22518540,
      "step": 402
    },
    {
      "epoch": 0.8975501113585747,
      "grad_norm": 25.960264205932617,
      "learning_rate": 1e-06,
      "loss": 1.412,
      "num_input_tokens_seen": 22575984,
      "step": 403
    },
    {
      "epoch": 0.8975501113585747,
      "loss": 1.4437758922576904,
      "loss_ce": 0.0028578725177794695,
      "loss_iou": 0.59375,
      "loss_num": 0.05126953125,
      "loss_xval": 1.4375,
      "num_input_tokens_seen": 22575984,
      "step": 403
    },
    {
      "epoch": 0.8997772828507795,
      "grad_norm": 61.44157791137695,
      "learning_rate": 1e-06,
      "loss": 1.1071,
      "num_input_tokens_seen": 22632724,
      "step": 404
    },
    {
      "epoch": 0.8997772828507795,
      "loss": 1.1888924837112427,
      "loss_ce": 0.002613153774291277,
      "loss_iou": 0.498046875,
      "loss_num": 0.0380859375,
      "loss_xval": 1.1875,
      "num_input_tokens_seen": 22632724,
      "step": 404
    },
    {
      "epoch": 0.9020044543429844,
      "grad_norm": 32.13711929321289,
      "learning_rate": 1e-06,
      "loss": 1.3325,
      "num_input_tokens_seen": 22689348,
      "step": 405
    },
    {
      "epoch": 0.9020044543429844,
      "loss": 1.1436662673950195,
      "loss_ce": 0.002064717700704932,
      "loss_iou": 0.494140625,
      "loss_num": 0.030517578125,
      "loss_xval": 1.140625,
      "num_input_tokens_seen": 22689348,
      "step": 405
    },
    {
      "epoch": 0.9042316258351893,
      "grad_norm": 22.586368560791016,
      "learning_rate": 1e-06,
      "loss": 1.3851,
      "num_input_tokens_seen": 22744560,
      "step": 406
    },
    {
      "epoch": 0.9042316258351893,
      "loss": 1.374314546585083,
      "loss_ce": 0.010056735947728157,
      "loss_iou": 0.5390625,
      "loss_num": 0.056884765625,
      "loss_xval": 1.3671875,
      "num_input_tokens_seen": 22744560,
      "step": 406
    },
    {
      "epoch": 0.9064587973273942,
      "grad_norm": 19.449066162109375,
      "learning_rate": 1e-06,
      "loss": 1.4187,
      "num_input_tokens_seen": 22796036,
      "step": 407
    },
    {
      "epoch": 0.9064587973273942,
      "loss": 1.272933006286621,
      "loss_ce": 0.0038900894578546286,
      "loss_iou": 0.5390625,
      "loss_num": 0.03857421875,
      "loss_xval": 1.265625,
      "num_input_tokens_seen": 22796036,
      "step": 407
    },
    {
      "epoch": 0.9086859688195991,
      "grad_norm": 17.967327117919922,
      "learning_rate": 1e-06,
      "loss": 1.101,
      "num_input_tokens_seen": 22851848,
      "step": 408
    },
    {
      "epoch": 0.9086859688195991,
      "loss": 0.8791401386260986,
      "loss_ce": 0.009511251002550125,
      "loss_iou": 0.35546875,
      "loss_num": 0.0311279296875,
      "loss_xval": 0.87109375,
      "num_input_tokens_seen": 22851848,
      "step": 408
    },
    {
      "epoch": 0.910913140311804,
      "grad_norm": 34.27125549316406,
      "learning_rate": 1e-06,
      "loss": 1.279,
      "num_input_tokens_seen": 22906068,
      "step": 409
    },
    {
      "epoch": 0.910913140311804,
      "loss": 1.3037530183792114,
      "loss_ce": 0.008098645135760307,
      "loss_iou": 0.546875,
      "loss_num": 0.041015625,
      "loss_xval": 1.296875,
      "num_input_tokens_seen": 22906068,
      "step": 409
    },
    {
      "epoch": 0.9131403118040089,
      "grad_norm": 21.108455657958984,
      "learning_rate": 1e-06,
      "loss": 1.3891,
      "num_input_tokens_seen": 22963800,
      "step": 410
    },
    {
      "epoch": 0.9131403118040089,
      "loss": 1.512423038482666,
      "loss_ce": 0.009981658309698105,
      "loss_iou": 0.6640625,
      "loss_num": 0.034912109375,
      "loss_xval": 1.5,
      "num_input_tokens_seen": 22963800,
      "step": 410
    },
    {
      "epoch": 0.9153674832962138,
      "grad_norm": 26.992612838745117,
      "learning_rate": 1e-06,
      "loss": 1.39,
      "num_input_tokens_seen": 23019888,
      "step": 411
    },
    {
      "epoch": 0.9153674832962138,
      "loss": 1.4112193584442139,
      "loss_ce": 0.0020396006293594837,
      "loss_iou": 0.5703125,
      "loss_num": 0.052490234375,
      "loss_xval": 1.40625,
      "num_input_tokens_seen": 23019888,
      "step": 411
    },
    {
      "epoch": 0.9175946547884187,
      "grad_norm": 25.17546272277832,
      "learning_rate": 1e-06,
      "loss": 1.2354,
      "num_input_tokens_seen": 23077784,
      "step": 412
    },
    {
      "epoch": 0.9175946547884187,
      "loss": 1.2591149806976318,
      "loss_ce": 0.02327517233788967,
      "loss_iou": 0.515625,
      "loss_num": 0.04052734375,
      "loss_xval": 1.234375,
      "num_input_tokens_seen": 23077784,
      "step": 412
    },
    {
      "epoch": 0.9198218262806236,
      "grad_norm": 50.736148834228516,
      "learning_rate": 1e-06,
      "loss": 1.1223,
      "num_input_tokens_seen": 23133656,
      "step": 413
    },
    {
      "epoch": 0.9198218262806236,
      "loss": 1.0429635047912598,
      "loss_ce": 0.0029244269244372845,
      "loss_iou": 0.404296875,
      "loss_num": 0.046630859375,
      "loss_xval": 1.0390625,
      "num_input_tokens_seen": 23133656,
      "step": 413
    },
    {
      "epoch": 0.9220489977728286,
      "grad_norm": 49.43931579589844,
      "learning_rate": 1e-06,
      "loss": 1.1577,
      "num_input_tokens_seen": 23190832,
      "step": 414
    },
    {
      "epoch": 0.9220489977728286,
      "loss": 1.1161892414093018,
      "loss_ce": 0.009988099336624146,
      "loss_iou": 0.431640625,
      "loss_num": 0.04833984375,
      "loss_xval": 1.109375,
      "num_input_tokens_seen": 23190832,
      "step": 414
    },
    {
      "epoch": 0.9242761692650334,
      "grad_norm": 31.02313232421875,
      "learning_rate": 1e-06,
      "loss": 1.4122,
      "num_input_tokens_seen": 23247304,
      "step": 415
    },
    {
      "epoch": 0.9242761692650334,
      "loss": 1.5239001512527466,
      "loss_ce": 0.003880674485117197,
      "loss_iou": 0.640625,
      "loss_num": 0.04833984375,
      "loss_xval": 1.5234375,
      "num_input_tokens_seen": 23247304,
      "step": 415
    },
    {
      "epoch": 0.9265033407572383,
      "grad_norm": 24.66490936279297,
      "learning_rate": 1e-06,
      "loss": 1.3191,
      "num_input_tokens_seen": 23302872,
      "step": 416
    },
    {
      "epoch": 0.9265033407572383,
      "loss": 1.2839527130126953,
      "loss_ce": 0.00563238188624382,
      "loss_iou": 0.51953125,
      "loss_num": 0.04833984375,
      "loss_xval": 1.28125,
      "num_input_tokens_seen": 23302872,
      "step": 416
    },
    {
      "epoch": 0.9287305122494433,
      "grad_norm": 29.012544631958008,
      "learning_rate": 1e-06,
      "loss": 1.4372,
      "num_input_tokens_seen": 23361228,
      "step": 417
    },
    {
      "epoch": 0.9287305122494433,
      "loss": 1.4470521211624146,
      "loss_ce": 0.008087254129350185,
      "loss_iou": 0.578125,
      "loss_num": 0.056640625,
      "loss_xval": 1.4375,
      "num_input_tokens_seen": 23361228,
      "step": 417
    },
    {
      "epoch": 0.9309576837416481,
      "grad_norm": 29.133703231811523,
      "learning_rate": 1e-06,
      "loss": 1.3035,
      "num_input_tokens_seen": 23417480,
      "step": 418
    },
    {
      "epoch": 0.9309576837416481,
      "loss": 1.2170838117599487,
      "loss_ce": 0.001263482728973031,
      "loss_iou": 0.482421875,
      "loss_num": 0.050537109375,
      "loss_xval": 1.21875,
      "num_input_tokens_seen": 23417480,
      "step": 418
    },
    {
      "epoch": 0.933184855233853,
      "grad_norm": 24.859573364257812,
      "learning_rate": 1e-06,
      "loss": 1.6284,
      "num_input_tokens_seen": 23474768,
      "step": 419
    },
    {
      "epoch": 0.933184855233853,
      "loss": 1.9370301961898804,
      "loss_ce": 0.03078020177781582,
      "loss_iou": 0.76171875,
      "loss_num": 0.0771484375,
      "loss_xval": 1.90625,
      "num_input_tokens_seen": 23474768,
      "step": 419
    },
    {
      "epoch": 0.9354120267260579,
      "grad_norm": 25.39378547668457,
      "learning_rate": 1e-06,
      "loss": 1.1607,
      "num_input_tokens_seen": 23530896,
      "step": 420
    },
    {
      "epoch": 0.9354120267260579,
      "loss": 1.2405550479888916,
      "loss_ce": 0.00471511110663414,
      "loss_iou": 0.53515625,
      "loss_num": 0.033935546875,
      "loss_xval": 1.234375,
      "num_input_tokens_seen": 23530896,
      "step": 420
    },
    {
      "epoch": 0.9376391982182628,
      "grad_norm": 39.46179962158203,
      "learning_rate": 1e-06,
      "loss": 1.4029,
      "num_input_tokens_seen": 23587564,
      "step": 421
    },
    {
      "epoch": 0.9376391982182628,
      "loss": 1.2013354301452637,
      "loss_ce": 0.0026050377637147903,
      "loss_iou": 0.466796875,
      "loss_num": 0.053466796875,
      "loss_xval": 1.1953125,
      "num_input_tokens_seen": 23587564,
      "step": 421
    },
    {
      "epoch": 0.9398663697104677,
      "grad_norm": 20.98834991455078,
      "learning_rate": 1e-06,
      "loss": 1.4864,
      "num_input_tokens_seen": 23641996,
      "step": 422
    },
    {
      "epoch": 0.9398663697104677,
      "loss": 1.4933122396469116,
      "loss_ce": 0.005030961707234383,
      "loss_iou": 0.6015625,
      "loss_num": 0.057373046875,
      "loss_xval": 1.484375,
      "num_input_tokens_seen": 23641996,
      "step": 422
    },
    {
      "epoch": 0.9420935412026726,
      "grad_norm": 28.44550895690918,
      "learning_rate": 1e-06,
      "loss": 1.3014,
      "num_input_tokens_seen": 23695464,
      "step": 423
    },
    {
      "epoch": 0.9420935412026726,
      "loss": 1.5061261653900146,
      "loss_ce": 0.021751180291175842,
      "loss_iou": 0.61328125,
      "loss_num": 0.05224609375,
      "loss_xval": 1.484375,
      "num_input_tokens_seen": 23695464,
      "step": 423
    },
    {
      "epoch": 0.9443207126948775,
      "grad_norm": 24.33575439453125,
      "learning_rate": 1e-06,
      "loss": 1.104,
      "num_input_tokens_seen": 23747676,
      "step": 424
    },
    {
      "epoch": 0.9443207126948775,
      "loss": 1.0683302879333496,
      "loss_ce": 0.0016799941658973694,
      "loss_iou": 0.439453125,
      "loss_num": 0.03759765625,
      "loss_xval": 1.0703125,
      "num_input_tokens_seen": 23747676,
      "step": 424
    },
    {
      "epoch": 0.9465478841870824,
      "grad_norm": 29.6976375579834,
      "learning_rate": 1e-06,
      "loss": 1.1804,
      "num_input_tokens_seen": 23805080,
      "step": 425
    },
    {
      "epoch": 0.9465478841870824,
      "loss": 1.1131433248519897,
      "loss_ce": 0.00132695899810642,
      "loss_iou": 0.453125,
      "loss_num": 0.041748046875,
      "loss_xval": 1.109375,
      "num_input_tokens_seen": 23805080,
      "step": 425
    },
    {
      "epoch": 0.9487750556792873,
      "grad_norm": 24.570606231689453,
      "learning_rate": 1e-06,
      "loss": 1.0864,
      "num_input_tokens_seen": 23862348,
      "step": 426
    },
    {
      "epoch": 0.9487750556792873,
      "loss": 0.9711905717849731,
      "loss_ce": 0.000975710921920836,
      "loss_iou": 0.416015625,
      "loss_num": 0.0277099609375,
      "loss_xval": 0.96875,
      "num_input_tokens_seen": 23862348,
      "step": 426
    },
    {
      "epoch": 0.9510022271714922,
      "grad_norm": 34.240631103515625,
      "learning_rate": 1e-06,
      "loss": 1.1878,
      "num_input_tokens_seen": 23919428,
      "step": 427
    },
    {
      "epoch": 0.9510022271714922,
      "loss": 1.3001813888549805,
      "loss_ce": 0.0015973602421581745,
      "loss_iou": 0.53125,
      "loss_num": 0.04736328125,
      "loss_xval": 1.296875,
      "num_input_tokens_seen": 23919428,
      "step": 427
    },
    {
      "epoch": 0.9532293986636972,
      "grad_norm": 22.28962516784668,
      "learning_rate": 1e-06,
      "loss": 1.0995,
      "num_input_tokens_seen": 23975564,
      "step": 428
    },
    {
      "epoch": 0.9532293986636972,
      "loss": 1.0272424221038818,
      "loss_ce": 0.01918586902320385,
      "loss_iou": 0.41796875,
      "loss_num": 0.0341796875,
      "loss_xval": 1.0078125,
      "num_input_tokens_seen": 23975564,
      "step": 428
    },
    {
      "epoch": 0.955456570155902,
      "grad_norm": 50.77242660522461,
      "learning_rate": 1e-06,
      "loss": 1.4387,
      "num_input_tokens_seen": 24032664,
      "step": 429
    },
    {
      "epoch": 0.955456570155902,
      "loss": 1.2324780225753784,
      "loss_ce": 0.019099093973636627,
      "loss_iou": 0.451171875,
      "loss_num": 0.0625,
      "loss_xval": 1.2109375,
      "num_input_tokens_seen": 24032664,
      "step": 429
    },
    {
      "epoch": 0.9576837416481069,
      "grad_norm": 17.031267166137695,
      "learning_rate": 1e-06,
      "loss": 1.0689,
      "num_input_tokens_seen": 24087656,
      "step": 430
    },
    {
      "epoch": 0.9576837416481069,
      "loss": 1.104911208152771,
      "loss_ce": 0.008475645445287228,
      "loss_iou": 0.4296875,
      "loss_num": 0.0478515625,
      "loss_xval": 1.09375,
      "num_input_tokens_seen": 24087656,
      "step": 430
    },
    {
      "epoch": 0.9599109131403119,
      "grad_norm": 29.331024169921875,
      "learning_rate": 1e-06,
      "loss": 1.404,
      "num_input_tokens_seen": 24141584,
      "step": 431
    },
    {
      "epoch": 0.9599109131403119,
      "loss": 1.457468032836914,
      "loss_ce": 0.027536382898688316,
      "loss_iou": 0.55859375,
      "loss_num": 0.06298828125,
      "loss_xval": 1.4296875,
      "num_input_tokens_seen": 24141584,
      "step": 431
    },
    {
      "epoch": 0.9621380846325167,
      "grad_norm": 23.636056900024414,
      "learning_rate": 1e-06,
      "loss": 1.379,
      "num_input_tokens_seen": 24197980,
      "step": 432
    },
    {
      "epoch": 0.9621380846325167,
      "loss": 1.2647449970245361,
      "loss_ce": 0.002049737609922886,
      "loss_iou": 0.51171875,
      "loss_num": 0.04833984375,
      "loss_xval": 1.265625,
      "num_input_tokens_seen": 24197980,
      "step": 432
    },
    {
      "epoch": 0.9643652561247216,
      "grad_norm": 94.4756851196289,
      "learning_rate": 1e-06,
      "loss": 1.3875,
      "num_input_tokens_seen": 24252980,
      "step": 433
    },
    {
      "epoch": 0.9643652561247216,
      "loss": 1.2504823207855225,
      "loss_ce": 0.0009705987758934498,
      "loss_iou": 0.474609375,
      "loss_num": 0.059814453125,
      "loss_xval": 1.25,
      "num_input_tokens_seen": 24252980,
      "step": 433
    },
    {
      "epoch": 0.9665924276169265,
      "grad_norm": 29.35396957397461,
      "learning_rate": 1e-06,
      "loss": 1.313,
      "num_input_tokens_seen": 24310388,
      "step": 434
    },
    {
      "epoch": 0.9665924276169265,
      "loss": 1.3304047584533691,
      "loss_ce": 0.015951739624142647,
      "loss_iou": 0.484375,
      "loss_num": 0.06884765625,
      "loss_xval": 1.3125,
      "num_input_tokens_seen": 24310388,
      "step": 434
    },
    {
      "epoch": 0.9688195991091314,
      "grad_norm": 19.95311737060547,
      "learning_rate": 1e-06,
      "loss": 1.0571,
      "num_input_tokens_seen": 24367988,
      "step": 435
    },
    {
      "epoch": 0.9688195991091314,
      "loss": 0.8543475270271301,
      "loss_ce": 0.0008319243206642568,
      "loss_iou": 0.345703125,
      "loss_num": 0.03271484375,
      "loss_xval": 0.8515625,
      "num_input_tokens_seen": 24367988,
      "step": 435
    },
    {
      "epoch": 0.9710467706013363,
      "grad_norm": 17.18592071533203,
      "learning_rate": 1e-06,
      "loss": 1.4076,
      "num_input_tokens_seen": 24421072,
      "step": 436
    },
    {
      "epoch": 0.9710467706013363,
      "loss": 1.2863093614578247,
      "loss_ce": 0.0038386958185583353,
      "loss_iou": 0.5078125,
      "loss_num": 0.053466796875,
      "loss_xval": 1.28125,
      "num_input_tokens_seen": 24421072,
      "step": 436
    },
    {
      "epoch": 0.9732739420935412,
      "grad_norm": 45.679954528808594,
      "learning_rate": 1e-06,
      "loss": 1.4846,
      "num_input_tokens_seen": 24476364,
      "step": 437
    },
    {
      "epoch": 0.9732739420935412,
      "loss": 1.2888422012329102,
      "loss_ce": 0.0017329129623249173,
      "loss_iou": 0.5078125,
      "loss_num": 0.0537109375,
      "loss_xval": 1.2890625,
      "num_input_tokens_seen": 24476364,
      "step": 437
    },
    {
      "epoch": 0.9755011135857461,
      "grad_norm": 26.56547737121582,
      "learning_rate": 1e-06,
      "loss": 1.4019,
      "num_input_tokens_seen": 24532516,
      "step": 438
    },
    {
      "epoch": 0.9755011135857461,
      "loss": 1.554681420326233,
      "loss_ce": 0.003900158451870084,
      "loss_iou": 0.609375,
      "loss_num": 0.06640625,
      "loss_xval": 1.546875,
      "num_input_tokens_seen": 24532516,
      "step": 438
    },
    {
      "epoch": 0.977728285077951,
      "grad_norm": 37.293983459472656,
      "learning_rate": 1e-06,
      "loss": 1.2445,
      "num_input_tokens_seen": 24586972,
      "step": 439
    },
    {
      "epoch": 0.977728285077951,
      "loss": 1.148227572441101,
      "loss_ce": 0.0100439777597785,
      "loss_iou": 0.4140625,
      "loss_num": 0.06201171875,
      "loss_xval": 1.140625,
      "num_input_tokens_seen": 24586972,
      "step": 439
    },
    {
      "epoch": 0.9799554565701559,
      "grad_norm": 29.18305778503418,
      "learning_rate": 1e-06,
      "loss": 1.1585,
      "num_input_tokens_seen": 24643012,
      "step": 440
    },
    {
      "epoch": 0.9799554565701559,
      "loss": 1.3668557405471802,
      "loss_ce": 0.004306901711970568,
      "loss_iou": 0.5625,
      "loss_num": 0.0478515625,
      "loss_xval": 1.359375,
      "num_input_tokens_seen": 24643012,
      "step": 440
    },
    {
      "epoch": 0.9821826280623608,
      "grad_norm": 47.63218688964844,
      "learning_rate": 1e-06,
      "loss": 1.4442,
      "num_input_tokens_seen": 24697816,
      "step": 441
    },
    {
      "epoch": 0.9821826280623608,
      "loss": 1.5267517566680908,
      "loss_ce": 0.0018494933610782027,
      "loss_iou": 0.60546875,
      "loss_num": 0.0625,
      "loss_xval": 1.5234375,
      "num_input_tokens_seen": 24697816,
      "step": 441
    },
    {
      "epoch": 0.9844097995545658,
      "grad_norm": 21.678260803222656,
      "learning_rate": 1e-06,
      "loss": 1.3717,
      "num_input_tokens_seen": 24752152,
      "step": 442
    },
    {
      "epoch": 0.9844097995545658,
      "loss": 1.6792147159576416,
      "loss_ce": 0.005386614240705967,
      "loss_iou": 0.60546875,
      "loss_num": 0.0927734375,
      "loss_xval": 1.671875,
      "num_input_tokens_seen": 24752152,
      "step": 442
    },
    {
      "epoch": 0.9866369710467706,
      "grad_norm": 52.10433578491211,
      "learning_rate": 1e-06,
      "loss": 1.3047,
      "num_input_tokens_seen": 24810384,
      "step": 443
    },
    {
      "epoch": 0.9866369710467706,
      "loss": 1.4768739938735962,
      "loss_ce": 0.005682523362338543,
      "loss_iou": 0.58203125,
      "loss_num": 0.06201171875,
      "loss_xval": 1.46875,
      "num_input_tokens_seen": 24810384,
      "step": 443
    },
    {
      "epoch": 0.9888641425389755,
      "grad_norm": 35.90444564819336,
      "learning_rate": 1e-06,
      "loss": 1.4123,
      "num_input_tokens_seen": 24864948,
      "step": 444
    },
    {
      "epoch": 0.9888641425389755,
      "loss": 1.7221654653549194,
      "loss_ce": 0.0039037105161696672,
      "loss_iou": 0.69921875,
      "loss_num": 0.06494140625,
      "loss_xval": 1.71875,
      "num_input_tokens_seen": 24864948,
      "step": 444
    },
    {
      "epoch": 0.9910913140311804,
      "grad_norm": 20.43412208557129,
      "learning_rate": 1e-06,
      "loss": 1.3585,
      "num_input_tokens_seen": 24919520,
      "step": 445
    },
    {
      "epoch": 0.9910913140311804,
      "loss": 1.4697803258895874,
      "loss_ce": 0.01421389076858759,
      "loss_iou": 0.58984375,
      "loss_num": 0.05517578125,
      "loss_xval": 1.453125,
      "num_input_tokens_seen": 24919520,
      "step": 445
    },
    {
      "epoch": 0.9933184855233853,
      "grad_norm": 48.228004455566406,
      "learning_rate": 1e-06,
      "loss": 1.6479,
      "num_input_tokens_seen": 24977308,
      "step": 446
    },
    {
      "epoch": 0.9933184855233853,
      "loss": 1.6608734130859375,
      "loss_ce": 0.0016936406027525663,
      "loss_iou": 0.67578125,
      "loss_num": 0.0615234375,
      "loss_xval": 1.65625,
      "num_input_tokens_seen": 24977308,
      "step": 446
    },
    {
      "epoch": 0.9955456570155902,
      "grad_norm": 43.12616729736328,
      "learning_rate": 1e-06,
      "loss": 1.6302,
      "num_input_tokens_seen": 25033208,
      "step": 447
    },
    {
      "epoch": 0.9955456570155902,
      "loss": 1.7152773141860962,
      "loss_ce": 0.004828128032386303,
      "loss_iou": 0.61328125,
      "loss_num": 0.0966796875,
      "loss_xval": 1.7109375,
      "num_input_tokens_seen": 25033208,
      "step": 447
    },
    {
      "epoch": 0.9977728285077951,
      "grad_norm": 32.99639892578125,
      "learning_rate": 1e-06,
      "loss": 1.43,
      "num_input_tokens_seen": 25089796,
      "step": 448
    },
    {
      "epoch": 0.9977728285077951,
      "loss": 1.0105788707733154,
      "loss_ce": 0.005207820795476437,
      "loss_iou": 0.423828125,
      "loss_num": 0.031494140625,
      "loss_xval": 1.0078125,
      "num_input_tokens_seen": 25089796,
      "step": 448
    },
    {
      "epoch": 1.0,
      "grad_norm": 22.11530303955078,
      "learning_rate": 1e-06,
      "loss": 1.0883,
      "num_input_tokens_seen": 25146032,
      "step": 449
    },
    {
      "epoch": 1.0,
      "loss": 1.1784168481826782,
      "loss_ce": 0.006053549237549305,
      "loss_iou": 0.4609375,
      "loss_num": 0.05029296875,
      "loss_xval": 1.171875,
      "num_input_tokens_seen": 25146032,
      "step": 449
    },
    {
      "epoch": 1.0022271714922049,
      "grad_norm": 18.13511848449707,
      "learning_rate": 1e-06,
      "loss": 1.2001,
      "num_input_tokens_seen": 25204016,
      "step": 450
    },
    {
      "epoch": 1.0022271714922049,
      "loss": 1.3148114681243896,
      "loss_ce": 0.0042645614594221115,
      "loss_iou": 0.55078125,
      "loss_num": 0.042236328125,
      "loss_xval": 1.3125,
      "num_input_tokens_seen": 25204016,
      "step": 450
    },
    {
      "epoch": 1.0044543429844097,
      "grad_norm": 72.61564636230469,
      "learning_rate": 1e-06,
      "loss": 1.3781,
      "num_input_tokens_seen": 25258404,
      "step": 451
    },
    {
      "epoch": 1.0044543429844097,
      "loss": 1.4480741024017334,
      "loss_ce": 0.015945199877023697,
      "loss_iou": 0.55078125,
      "loss_num": 0.06689453125,
      "loss_xval": 1.4296875,
      "num_input_tokens_seen": 25258404,
      "step": 451
    },
    {
      "epoch": 1.0066815144766148,
      "grad_norm": 40.180419921875,
      "learning_rate": 1e-06,
      "loss": 1.4411,
      "num_input_tokens_seen": 25315080,
      "step": 452
    },
    {
      "epoch": 1.0066815144766148,
      "loss": 1.4863982200622559,
      "loss_ce": 0.005441202782094479,
      "loss_iou": 0.5703125,
      "loss_num": 0.068359375,
      "loss_xval": 1.484375,
      "num_input_tokens_seen": 25315080,
      "step": 452
    },
    {
      "epoch": 1.0089086859688197,
      "grad_norm": 28.12888526916504,
      "learning_rate": 1e-06,
      "loss": 1.1451,
      "num_input_tokens_seen": 25370644,
      "step": 453
    },
    {
      "epoch": 1.0089086859688197,
      "loss": 1.1297539472579956,
      "loss_ce": 0.01940234750509262,
      "loss_iou": 0.4453125,
      "loss_num": 0.043701171875,
      "loss_xval": 1.109375,
      "num_input_tokens_seen": 25370644,
      "step": 453
    },
    {
      "epoch": 1.0111358574610245,
      "grad_norm": 26.153356552124023,
      "learning_rate": 1e-06,
      "loss": 1.2351,
      "num_input_tokens_seen": 25425148,
      "step": 454
    },
    {
      "epoch": 1.0111358574610245,
      "loss": 1.6115047931671143,
      "loss_ce": 0.002618103986606002,
      "loss_iou": 0.5859375,
      "loss_num": 0.08837890625,
      "loss_xval": 1.609375,
      "num_input_tokens_seen": 25425148,
      "step": 454
    },
    {
      "epoch": 1.0133630289532294,
      "grad_norm": 41.62250900268555,
      "learning_rate": 1e-06,
      "loss": 1.4871,
      "num_input_tokens_seen": 25478780,
      "step": 455
    },
    {
      "epoch": 1.0133630289532294,
      "loss": 1.8347097635269165,
      "loss_ce": 0.004143323749303818,
      "loss_iou": 0.703125,
      "loss_num": 0.08544921875,
      "loss_xval": 1.828125,
      "num_input_tokens_seen": 25478780,
      "step": 455
    },
    {
      "epoch": 1.0155902004454342,
      "grad_norm": 54.628746032714844,
      "learning_rate": 1e-06,
      "loss": 1.2763,
      "num_input_tokens_seen": 25533068,
      "step": 456
    },
    {
      "epoch": 1.0155902004454342,
      "loss": 1.017959713935852,
      "loss_ce": 0.007217581383883953,
      "loss_iou": 0.384765625,
      "loss_num": 0.048583984375,
      "loss_xval": 1.0078125,
      "num_input_tokens_seen": 25533068,
      "step": 456
    },
    {
      "epoch": 1.017817371937639,
      "grad_norm": 19.17976951599121,
      "learning_rate": 1e-06,
      "loss": 1.1245,
      "num_input_tokens_seen": 25588116,
      "step": 457
    },
    {
      "epoch": 1.017817371937639,
      "loss": 1.1571969985961914,
      "loss_ce": 0.0024117890279740095,
      "loss_iou": 0.5078125,
      "loss_num": 0.02734375,
      "loss_xval": 1.15625,
      "num_input_tokens_seen": 25588116,
      "step": 457
    },
    {
      "epoch": 1.0200445434298442,
      "grad_norm": 59.96177291870117,
      "learning_rate": 1e-06,
      "loss": 1.213,
      "num_input_tokens_seen": 25645984,
      "step": 458
    },
    {
      "epoch": 1.0200445434298442,
      "loss": 1.2157493829727173,
      "loss_ce": 0.006276742089539766,
      "loss_iou": 0.455078125,
      "loss_num": 0.06005859375,
      "loss_xval": 1.2109375,
      "num_input_tokens_seen": 25645984,
      "step": 458
    },
    {
      "epoch": 1.022271714922049,
      "grad_norm": 22.60318946838379,
      "learning_rate": 1e-06,
      "loss": 1.5715,
      "num_input_tokens_seen": 25701456,
      "step": 459
    },
    {
      "epoch": 1.022271714922049,
      "loss": 1.8298416137695312,
      "loss_ce": 0.006599403452128172,
      "loss_iou": 0.75390625,
      "loss_num": 0.06298828125,
      "loss_xval": 1.8203125,
      "num_input_tokens_seen": 25701456,
      "step": 459
    },
    {
      "epoch": 1.024498886414254,
      "grad_norm": 30.009733200073242,
      "learning_rate": 1e-06,
      "loss": 1.2751,
      "num_input_tokens_seen": 25760304,
      "step": 460
    },
    {
      "epoch": 1.024498886414254,
      "loss": 0.9823180437088013,
      "loss_ce": 0.008441124111413956,
      "loss_iou": 0.388671875,
      "loss_num": 0.0390625,
      "loss_xval": 0.97265625,
      "num_input_tokens_seen": 25760304,
      "step": 460
    },
    {
      "epoch": 1.0267260579064588,
      "grad_norm": 29.346010208129883,
      "learning_rate": 1e-06,
      "loss": 1.0664,
      "num_input_tokens_seen": 25812724,
      "step": 461
    },
    {
      "epoch": 1.0267260579064588,
      "loss": 0.9884135723114014,
      "loss_ce": 0.006968258880078793,
      "loss_iou": 0.396484375,
      "loss_num": 0.03759765625,
      "loss_xval": 0.98046875,
      "num_input_tokens_seen": 25812724,
      "step": 461
    },
    {
      "epoch": 1.0289532293986636,
      "grad_norm": 75.70279693603516,
      "learning_rate": 1e-06,
      "loss": 1.2436,
      "num_input_tokens_seen": 25866680,
      "step": 462
    },
    {
      "epoch": 1.0289532293986636,
      "loss": 1.3562824726104736,
      "loss_ce": 0.010579358786344528,
      "loss_iou": 0.5625,
      "loss_num": 0.04443359375,
      "loss_xval": 1.34375,
      "num_input_tokens_seen": 25866680,
      "step": 462
    },
    {
      "epoch": 1.0311804008908685,
      "grad_norm": 54.88785171508789,
      "learning_rate": 1e-06,
      "loss": 1.2655,
      "num_input_tokens_seen": 25920128,
      "step": 463
    },
    {
      "epoch": 1.0311804008908685,
      "loss": 1.265822172164917,
      "loss_ce": 0.002150336978957057,
      "loss_iou": 0.486328125,
      "loss_num": 0.05810546875,
      "loss_xval": 1.265625,
      "num_input_tokens_seen": 25920128,
      "step": 463
    },
    {
      "epoch": 1.0334075723830736,
      "grad_norm": 24.761333465576172,
      "learning_rate": 1e-06,
      "loss": 1.2714,
      "num_input_tokens_seen": 25973036,
      "step": 464
    },
    {
      "epoch": 1.0334075723830736,
      "loss": 1.3659954071044922,
      "loss_ce": 0.0012494358234107494,
      "loss_iou": 0.515625,
      "loss_num": 0.0673828125,
      "loss_xval": 1.3671875,
      "num_input_tokens_seen": 25973036,
      "step": 464
    },
    {
      "epoch": 1.0356347438752784,
      "grad_norm": 23.98271369934082,
      "learning_rate": 1e-06,
      "loss": 1.3715,
      "num_input_tokens_seen": 26025288,
      "step": 465
    },
    {
      "epoch": 1.0356347438752784,
      "loss": 1.3842101097106934,
      "loss_ce": 0.007867315784096718,
      "loss_iou": 0.546875,
      "loss_num": 0.056640625,
      "loss_xval": 1.375,
      "num_input_tokens_seen": 26025288,
      "step": 465
    },
    {
      "epoch": 1.0378619153674833,
      "grad_norm": 21.38801383972168,
      "learning_rate": 1e-06,
      "loss": 1.2095,
      "num_input_tokens_seen": 26082480,
      "step": 466
    },
    {
      "epoch": 1.0378619153674833,
      "loss": 1.2851924896240234,
      "loss_ce": 0.0010128666181117296,
      "loss_iou": 0.50390625,
      "loss_num": 0.054931640625,
      "loss_xval": 1.28125,
      "num_input_tokens_seen": 26082480,
      "step": 466
    },
    {
      "epoch": 1.0400890868596881,
      "grad_norm": 25.396892547607422,
      "learning_rate": 1e-06,
      "loss": 1.0663,
      "num_input_tokens_seen": 26139504,
      "step": 467
    },
    {
      "epoch": 1.0400890868596881,
      "loss": 1.0588932037353516,
      "loss_ce": 0.0007877358002588153,
      "loss_iou": 0.44140625,
      "loss_num": 0.03466796875,
      "loss_xval": 1.0546875,
      "num_input_tokens_seen": 26139504,
      "step": 467
    },
    {
      "epoch": 1.042316258351893,
      "grad_norm": 18.479541778564453,
      "learning_rate": 1e-06,
      "loss": 1.2228,
      "num_input_tokens_seen": 26196448,
      "step": 468
    },
    {
      "epoch": 1.042316258351893,
      "loss": 1.2266954183578491,
      "loss_ce": 0.0011094606015831232,
      "loss_iou": 0.482421875,
      "loss_num": 0.052001953125,
      "loss_xval": 1.2265625,
      "num_input_tokens_seen": 26196448,
      "step": 468
    },
    {
      "epoch": 1.044543429844098,
      "grad_norm": 19.95421028137207,
      "learning_rate": 1e-06,
      "loss": 1.1337,
      "num_input_tokens_seen": 26250848,
      "step": 469
    },
    {
      "epoch": 1.044543429844098,
      "loss": 1.1349093914031982,
      "loss_ce": 0.0011203193571418524,
      "loss_iou": 0.421875,
      "loss_num": 0.05859375,
      "loss_xval": 1.1328125,
      "num_input_tokens_seen": 26250848,
      "step": 469
    },
    {
      "epoch": 1.046770601336303,
      "grad_norm": 29.218976974487305,
      "learning_rate": 1e-06,
      "loss": 1.0506,
      "num_input_tokens_seen": 26304848,
      "step": 470
    },
    {
      "epoch": 1.046770601336303,
      "loss": 0.7857218980789185,
      "loss_ce": 0.0015422652941197157,
      "loss_iou": 0.267578125,
      "loss_num": 0.049560546875,
      "loss_xval": 0.78515625,
      "num_input_tokens_seen": 26304848,
      "step": 470
    },
    {
      "epoch": 1.0489977728285078,
      "grad_norm": 26.657873153686523,
      "learning_rate": 1e-06,
      "loss": 1.2585,
      "num_input_tokens_seen": 26362040,
      "step": 471
    },
    {
      "epoch": 1.0489977728285078,
      "loss": 1.1070148944854736,
      "loss_ce": 0.0015461579896509647,
      "loss_iou": 0.4453125,
      "loss_num": 0.04345703125,
      "loss_xval": 1.109375,
      "num_input_tokens_seen": 26362040,
      "step": 471
    },
    {
      "epoch": 1.0512249443207127,
      "grad_norm": 24.098112106323242,
      "learning_rate": 1e-06,
      "loss": 1.2629,
      "num_input_tokens_seen": 26418744,
      "step": 472
    },
    {
      "epoch": 1.0512249443207127,
      "loss": 1.3726317882537842,
      "loss_ce": 0.008373986929655075,
      "loss_iou": 0.53125,
      "loss_num": 0.0595703125,
      "loss_xval": 1.3671875,
      "num_input_tokens_seen": 26418744,
      "step": 472
    },
    {
      "epoch": 1.0534521158129175,
      "grad_norm": 18.4561767578125,
      "learning_rate": 1e-06,
      "loss": 1.0389,
      "num_input_tokens_seen": 26476200,
      "step": 473
    },
    {
      "epoch": 1.0534521158129175,
      "loss": 1.1621673107147217,
      "loss_ce": 0.0010345308110117912,
      "loss_iou": 0.453125,
      "loss_num": 0.051025390625,
      "loss_xval": 1.1640625,
      "num_input_tokens_seen": 26476200,
      "step": 473
    },
    {
      "epoch": 1.0556792873051224,
      "grad_norm": 22.120088577270508,
      "learning_rate": 1e-06,
      "loss": 1.2004,
      "num_input_tokens_seen": 26532420,
      "step": 474
    },
    {
      "epoch": 1.0556792873051224,
      "loss": 1.0202289819717407,
      "loss_ce": 0.0014302136842161417,
      "loss_iou": 0.384765625,
      "loss_num": 0.050048828125,
      "loss_xval": 1.015625,
      "num_input_tokens_seen": 26532420,
      "step": 474
    },
    {
      "epoch": 1.0579064587973275,
      "grad_norm": 28.964649200439453,
      "learning_rate": 1e-06,
      "loss": 1.2759,
      "num_input_tokens_seen": 26585900,
      "step": 475
    },
    {
      "epoch": 1.0579064587973275,
      "loss": 1.5081238746643066,
      "loss_ce": 0.0017761469352990389,
      "loss_iou": 0.57421875,
      "loss_num": 0.072265625,
      "loss_xval": 1.5078125,
      "num_input_tokens_seen": 26585900,
      "step": 475
    },
    {
      "epoch": 1.0601336302895323,
      "grad_norm": 18.955963134765625,
      "learning_rate": 1e-06,
      "loss": 1.2483,
      "num_input_tokens_seen": 26640324,
      "step": 476
    },
    {
      "epoch": 1.0601336302895323,
      "loss": 1.3863422870635986,
      "loss_ce": 0.0040180860087275505,
      "loss_iou": 0.6015625,
      "loss_num": 0.03564453125,
      "loss_xval": 1.3828125,
      "num_input_tokens_seen": 26640324,
      "step": 476
    },
    {
      "epoch": 1.0623608017817372,
      "grad_norm": 29.491914749145508,
      "learning_rate": 1e-06,
      "loss": 1.2475,
      "num_input_tokens_seen": 26694532,
      "step": 477
    },
    {
      "epoch": 1.0623608017817372,
      "loss": 0.9849135875701904,
      "loss_ce": 0.0007827761583030224,
      "loss_iou": 0.369140625,
      "loss_num": 0.048828125,
      "loss_xval": 0.984375,
      "num_input_tokens_seen": 26694532,
      "step": 477
    },
    {
      "epoch": 1.064587973273942,
      "grad_norm": 65.8954849243164,
      "learning_rate": 1e-06,
      "loss": 1.4668,
      "num_input_tokens_seen": 26749932,
      "step": 478
    },
    {
      "epoch": 1.064587973273942,
      "loss": 1.4816560745239258,
      "loss_ce": 0.001187233254313469,
      "loss_iou": 0.5703125,
      "loss_num": 0.06787109375,
      "loss_xval": 1.484375,
      "num_input_tokens_seen": 26749932,
      "step": 478
    },
    {
      "epoch": 1.066815144766147,
      "grad_norm": 19.884841918945312,
      "learning_rate": 1e-06,
      "loss": 1.1807,
      "num_input_tokens_seen": 26804456,
      "step": 479
    },
    {
      "epoch": 1.066815144766147,
      "loss": 1.2711074352264404,
      "loss_ce": 0.002064512576907873,
      "loss_iou": 0.5234375,
      "loss_num": 0.044189453125,
      "loss_xval": 1.265625,
      "num_input_tokens_seen": 26804456,
      "step": 479
    },
    {
      "epoch": 1.069042316258352,
      "grad_norm": 22.818063735961914,
      "learning_rate": 1e-06,
      "loss": 1.3075,
      "num_input_tokens_seen": 26860280,
      "step": 480
    },
    {
      "epoch": 1.069042316258352,
      "loss": 1.333120584487915,
      "loss_ce": 0.0010893936268985271,
      "loss_iou": 0.5546875,
      "loss_num": 0.045166015625,
      "loss_xval": 1.328125,
      "num_input_tokens_seen": 26860280,
      "step": 480
    },
    {
      "epoch": 1.0712694877505569,
      "grad_norm": 21.60323715209961,
      "learning_rate": 1e-06,
      "loss": 1.1227,
      "num_input_tokens_seen": 26915848,
      "step": 481
    },
    {
      "epoch": 1.0712694877505569,
      "loss": 1.1749460697174072,
      "loss_ce": 0.002582841319963336,
      "loss_iou": 0.498046875,
      "loss_num": 0.03564453125,
      "loss_xval": 1.171875,
      "num_input_tokens_seen": 26915848,
      "step": 481
    },
    {
      "epoch": 1.0734966592427617,
      "grad_norm": 17.057666778564453,
      "learning_rate": 1e-06,
      "loss": 0.963,
      "num_input_tokens_seen": 26973348,
      "step": 482
    },
    {
      "epoch": 1.0734966592427617,
      "loss": 1.024942398071289,
      "loss_ce": 0.001016615773551166,
      "loss_iou": 0.4296875,
      "loss_num": 0.033203125,
      "loss_xval": 1.0234375,
      "num_input_tokens_seen": 26973348,
      "step": 482
    },
    {
      "epoch": 1.0757238307349666,
      "grad_norm": 26.440898895263672,
      "learning_rate": 1e-06,
      "loss": 1.2304,
      "num_input_tokens_seen": 27030864,
      "step": 483
    },
    {
      "epoch": 1.0757238307349666,
      "loss": 1.3861130475997925,
      "loss_ce": 0.00134742702357471,
      "loss_iou": 0.52734375,
      "loss_num": 0.06640625,
      "loss_xval": 1.3828125,
      "num_input_tokens_seen": 27030864,
      "step": 483
    },
    {
      "epoch": 1.0779510022271714,
      "grad_norm": 19.473052978515625,
      "learning_rate": 1e-06,
      "loss": 1.494,
      "num_input_tokens_seen": 27084476,
      "step": 484
    },
    {
      "epoch": 1.0779510022271714,
      "loss": 1.7679762840270996,
      "loss_ce": 0.0008863758994266391,
      "loss_iou": 0.70703125,
      "loss_num": 0.07080078125,
      "loss_xval": 1.765625,
      "num_input_tokens_seen": 27084476,
      "step": 484
    },
    {
      "epoch": 1.0801781737193763,
      "grad_norm": 22.302900314331055,
      "learning_rate": 1e-06,
      "loss": 1.3245,
      "num_input_tokens_seen": 27138164,
      "step": 485
    },
    {
      "epoch": 1.0801781737193763,
      "loss": 1.4490606784820557,
      "loss_ce": 0.0008185390033759177,
      "loss_iou": 0.6171875,
      "loss_num": 0.042724609375,
      "loss_xval": 1.4453125,
      "num_input_tokens_seen": 27138164,
      "step": 485
    },
    {
      "epoch": 1.0824053452115814,
      "grad_norm": 27.68915557861328,
      "learning_rate": 1e-06,
      "loss": 0.9771,
      "num_input_tokens_seen": 27195028,
      "step": 486
    },
    {
      "epoch": 1.0824053452115814,
      "loss": 0.880718469619751,
      "loss_ce": 0.0018122631590813398,
      "loss_iou": 0.369140625,
      "loss_num": 0.0281982421875,
      "loss_xval": 0.87890625,
      "num_input_tokens_seen": 27195028,
      "step": 486
    },
    {
      "epoch": 1.0846325167037862,
      "grad_norm": 27.754865646362305,
      "learning_rate": 1e-06,
      "loss": 1.3244,
      "num_input_tokens_seen": 27250504,
      "step": 487
    },
    {
      "epoch": 1.0846325167037862,
      "loss": 1.085121750831604,
      "loss_ce": 0.0006490740925073624,
      "loss_iou": 0.4296875,
      "loss_num": 0.044677734375,
      "loss_xval": 1.0859375,
      "num_input_tokens_seen": 27250504,
      "step": 487
    },
    {
      "epoch": 1.086859688195991,
      "grad_norm": 29.184783935546875,
      "learning_rate": 1e-06,
      "loss": 1.131,
      "num_input_tokens_seen": 27306092,
      "step": 488
    },
    {
      "epoch": 1.086859688195991,
      "loss": 1.2078437805175781,
      "loss_ce": 0.007404354866594076,
      "loss_iou": 0.453125,
      "loss_num": 0.058837890625,
      "loss_xval": 1.203125,
      "num_input_tokens_seen": 27306092,
      "step": 488
    },
    {
      "epoch": 1.089086859688196,
      "grad_norm": 16.471162796020508,
      "learning_rate": 1e-06,
      "loss": 1.1771,
      "num_input_tokens_seen": 27362708,
      "step": 489
    },
    {
      "epoch": 1.089086859688196,
      "loss": 1.2551052570343018,
      "loss_ce": 0.0007107860874384642,
      "loss_iou": 0.52734375,
      "loss_num": 0.040283203125,
      "loss_xval": 1.2578125,
      "num_input_tokens_seen": 27362708,
      "step": 489
    },
    {
      "epoch": 1.0913140311804008,
      "grad_norm": 36.433353424072266,
      "learning_rate": 1e-06,
      "loss": 1.1646,
      "num_input_tokens_seen": 27421108,
      "step": 490
    },
    {
      "epoch": 1.0913140311804008,
      "loss": 1.1176813840866089,
      "loss_ce": 0.0014704548520967364,
      "loss_iou": 0.46484375,
      "loss_num": 0.037841796875,
      "loss_xval": 1.1171875,
      "num_input_tokens_seen": 27421108,
      "step": 490
    },
    {
      "epoch": 1.093541202672606,
      "grad_norm": 21.38726806640625,
      "learning_rate": 1e-06,
      "loss": 1.1923,
      "num_input_tokens_seen": 27475860,
      "step": 491
    },
    {
      "epoch": 1.093541202672606,
      "loss": 1.053163766860962,
      "loss_ce": 0.0058004953898489475,
      "loss_iou": 0.41796875,
      "loss_num": 0.042236328125,
      "loss_xval": 1.046875,
      "num_input_tokens_seen": 27475860,
      "step": 491
    },
    {
      "epoch": 1.0957683741648108,
      "grad_norm": 32.688072204589844,
      "learning_rate": 1e-06,
      "loss": 1.0674,
      "num_input_tokens_seen": 27533372,
      "step": 492
    },
    {
      "epoch": 1.0957683741648108,
      "loss": 0.7940797805786133,
      "loss_ce": 0.01527119055390358,
      "loss_iou": 0.3359375,
      "loss_num": 0.021240234375,
      "loss_xval": 0.77734375,
      "num_input_tokens_seen": 27533372,
      "step": 492
    },
    {
      "epoch": 1.0979955456570156,
      "grad_norm": 20.838420867919922,
      "learning_rate": 1e-06,
      "loss": 1.4187,
      "num_input_tokens_seen": 27588776,
      "step": 493
    },
    {
      "epoch": 1.0979955456570156,
      "loss": 1.4859492778778076,
      "loss_ce": 0.0015742178075015545,
      "loss_iou": 0.5625,
      "loss_num": 0.0712890625,
      "loss_xval": 1.484375,
      "num_input_tokens_seen": 27588776,
      "step": 493
    },
    {
      "epoch": 1.1002227171492205,
      "grad_norm": 48.34200668334961,
      "learning_rate": 1e-06,
      "loss": 1.3155,
      "num_input_tokens_seen": 27646412,
      "step": 494
    },
    {
      "epoch": 1.1002227171492205,
      "loss": 1.6553877592086792,
      "loss_ce": 0.004997197538614273,
      "loss_iou": 0.65234375,
      "loss_num": 0.06884765625,
      "loss_xval": 1.6484375,
      "num_input_tokens_seen": 27646412,
      "step": 494
    },
    {
      "epoch": 1.1024498886414253,
      "grad_norm": 23.207468032836914,
      "learning_rate": 1e-06,
      "loss": 1.1153,
      "num_input_tokens_seen": 27702240,
      "step": 495
    },
    {
      "epoch": 1.1024498886414253,
      "loss": 1.4165308475494385,
      "loss_ce": 0.003933266270905733,
      "loss_iou": 0.57421875,
      "loss_num": 0.05322265625,
      "loss_xval": 1.4140625,
      "num_input_tokens_seen": 27702240,
      "step": 495
    },
    {
      "epoch": 1.1046770601336302,
      "grad_norm": 15.092357635498047,
      "learning_rate": 1e-06,
      "loss": 0.9602,
      "num_input_tokens_seen": 27759140,
      "step": 496
    },
    {
      "epoch": 1.1046770601336302,
      "loss": 1.09522545337677,
      "loss_ce": 0.002940312959253788,
      "loss_iou": 0.439453125,
      "loss_num": 0.04248046875,
      "loss_xval": 1.09375,
      "num_input_tokens_seen": 27759140,
      "step": 496
    },
    {
      "epoch": 1.1069042316258353,
      "grad_norm": 23.996767044067383,
      "learning_rate": 1e-06,
      "loss": 1.2124,
      "num_input_tokens_seen": 27816900,
      "step": 497
    },
    {
      "epoch": 1.1069042316258353,
      "loss": 1.2346551418304443,
      "loss_ce": 0.004186346661299467,
      "loss_iou": 0.53125,
      "loss_num": 0.033447265625,
      "loss_xval": 1.234375,
      "num_input_tokens_seen": 27816900,
      "step": 497
    },
    {
      "epoch": 1.1091314031180401,
      "grad_norm": 154.3430633544922,
      "learning_rate": 1e-06,
      "loss": 1.2454,
      "num_input_tokens_seen": 27875428,
      "step": 498
    },
    {
      "epoch": 1.1091314031180401,
      "loss": 1.4324363470077515,
      "loss_ce": 0.0007956642657518387,
      "loss_iou": 0.59765625,
      "loss_num": 0.047607421875,
      "loss_xval": 1.4296875,
      "num_input_tokens_seen": 27875428,
      "step": 498
    },
    {
      "epoch": 1.111358574610245,
      "grad_norm": 30.189279556274414,
      "learning_rate": 1e-06,
      "loss": 1.4237,
      "num_input_tokens_seen": 27930704,
      "step": 499
    },
    {
      "epoch": 1.111358574610245,
      "loss": 1.6740535497665405,
      "loss_ce": 0.0021784906275570393,
      "loss_iou": 0.62109375,
      "loss_num": 0.08642578125,
      "loss_xval": 1.671875,
      "num_input_tokens_seen": 27930704,
      "step": 499
    },
    {
      "epoch": 1.1135857461024499,
      "grad_norm": 49.67326354980469,
      "learning_rate": 1e-06,
      "loss": 1.2495,
      "num_input_tokens_seen": 27984260,
      "step": 500
    },
    {
      "epoch": 1.1135857461024499,
      "eval_seeclick_web_CIoU": 0.48640576004981995,
      "eval_seeclick_web_GIoU": 0.47375138103961945,
      "eval_seeclick_web_IoU": 0.5116834044456482,
      "eval_seeclick_web_MAE_all": 0.01661441382020712,
      "eval_seeclick_web_MAE_h": 0.01039309287443757,
      "eval_seeclick_web_MAE_w": 0.017452615778893232,
      "eval_seeclick_web_MAE_x_boxes": 0.010556747671216726,
      "eval_seeclick_web_MAE_y_boxes": 0.020513739669695497,
      "eval_seeclick_web_inside_bbox": 0.8576388955116272,
      "eval_seeclick_web_loss": 1.1193333864212036,
      "eval_seeclick_web_loss_ce": 0.0005516507662832737,
      "eval_seeclick_web_loss_iou": 0.519775390625,
      "eval_seeclick_web_loss_num": 0.014141082763671875,
      "eval_seeclick_web_loss_xval": 1.110107421875,
      "eval_seeclick_web_runtime": 19.1191,
      "eval_seeclick_web_samples_per_second": 2.615,
      "eval_seeclick_web_steps_per_second": 0.105,
      "num_input_tokens_seen": 27984260,
      "step": 500
    },
    {
      "epoch": 1.1135857461024499,
      "eval_icons_CIoU": 0.3478478938341141,
      "eval_icons_GIoU": 0.39389291405677795,
      "eval_icons_IoU": 0.4208338111639023,
      "eval_icons_MAE_all": 0.06633740290999413,
      "eval_icons_MAE_h": 0.03294991096481681,
      "eval_icons_MAE_w": 0.06817016191780567,
      "eval_icons_MAE_x_boxes": 0.06793619319796562,
      "eval_icons_MAE_y_boxes": 0.033266451209783554,
      "eval_icons_inside_bbox": 0.7048611044883728,
      "eval_icons_loss": 1.5932660102844238,
      "eval_icons_loss_ce": 0.0014974797377362847,
      "eval_icons_loss_iou": 0.614501953125,
      "eval_icons_loss_num": 0.06303977966308594,
      "eval_icons_loss_xval": 1.544677734375,
      "eval_icons_runtime": 16.8196,
      "eval_icons_samples_per_second": 2.973,
      "eval_icons_steps_per_second": 0.119,
      "num_input_tokens_seen": 27984260,
      "step": 500
    },
    {
      "epoch": 1.1135857461024499,
      "eval_screenspot_CIoU": 0.2469616780678431,
      "eval_screenspot_GIoU": 0.2680433491865794,
      "eval_screenspot_IoU": 0.33675894141197205,
      "eval_screenspot_MAE_all": 0.09882631152868271,
      "eval_screenspot_MAE_h": 0.06025169417262077,
      "eval_screenspot_MAE_w": 0.10315311948458354,
      "eval_screenspot_MAE_x_boxes": 0.11736861368020375,
      "eval_screenspot_MAE_y_boxes": 0.0693823571006457,
      "eval_screenspot_inside_bbox": 0.5570833285649618,
      "eval_screenspot_loss": 1.9871058464050293,
      "eval_screenspot_loss_ce": 0.006458223797380924,
      "eval_screenspot_loss_iou": 0.7589518229166666,
      "eval_screenspot_loss_num": 0.10675303141276042,
      "eval_screenspot_loss_xval": 2.0504557291666665,
      "eval_screenspot_runtime": 27.6396,
      "eval_screenspot_samples_per_second": 3.22,
      "eval_screenspot_steps_per_second": 0.109,
      "num_input_tokens_seen": 27984260,
      "step": 500
    },
    {
      "epoch": 1.1135857461024499,
      "eval_compot_CIoU": 0.2672244608402252,
      "eval_compot_GIoU": 0.2929573655128479,
      "eval_compot_IoU": 0.333434134721756,
      "eval_compot_MAE_all": 0.03275496046990156,
      "eval_compot_MAE_h": 0.015514453873038292,
      "eval_compot_MAE_w": 0.043023936450481415,
      "eval_compot_MAE_x_boxes": 0.04298969078809023,
      "eval_compot_MAE_y_boxes": 0.009457055712118745,
      "eval_compot_inside_bbox": 0.5590277910232544,
      "eval_compot_loss": 1.6115740537643433,
      "eval_compot_loss_ce": 0.0006396450917236507,
      "eval_compot_loss_iou": 0.701904296875,
      "eval_compot_loss_num": 0.02944183349609375,
      "eval_compot_loss_xval": 1.55078125,
      "eval_compot_runtime": 18.007,
      "eval_compot_samples_per_second": 2.777,
      "eval_compot_steps_per_second": 0.111,
      "num_input_tokens_seen": 27984260,
      "step": 500
    },
    {
      "epoch": 1.1135857461024499,
      "eval_custom_ui_val_CIoU": 0.3800675223271052,
      "eval_custom_ui_val_GIoU": 0.4050305353270637,
      "eval_custom_ui_val_IoU": 0.43855932023790145,
      "eval_custom_ui_val_MAE_all": 0.046337926760315895,
      "eval_custom_ui_val_MAE_h": 0.026893117154637974,
      "eval_custom_ui_val_MAE_w": 0.05207906601329645,
      "eval_custom_ui_val_MAE_x_boxes": 0.046116245082683034,
      "eval_custom_ui_val_MAE_y_boxes": 0.030045698396861553,
      "eval_custom_ui_val_inside_bbox": 0.6608796318372091,
      "eval_custom_ui_val_loss": 1.4429850578308105,
      "eval_custom_ui_val_loss_ce": 0.0016363520019998152,
      "eval_custom_ui_val_loss_iou": 0.5924343532986112,
      "eval_custom_ui_val_loss_num": 0.04485532972547743,
      "eval_custom_ui_val_loss_xval": 1.4086642795138888,
      "eval_custom_ui_val_runtime": 55.8893,
      "eval_custom_ui_val_samples_per_second": 4.742,
      "eval_custom_ui_val_steps_per_second": 0.161,
      "num_input_tokens_seen": 27984260,
      "step": 500
    },
    {
      "epoch": 1.1135857461024499,
      "loss": 1.1450822353363037,
      "loss_ce": 0.0012833788059651852,
      "loss_iou": 0.482421875,
      "loss_num": 0.0361328125,
      "loss_xval": 1.140625,
      "num_input_tokens_seen": 27984260,
      "step": 500
    },
    {
      "epoch": 1.1158129175946547,
      "grad_norm": 604.319580078125,
      "learning_rate": 1e-06,
      "loss": 1.4076,
      "num_input_tokens_seen": 28041640,
      "step": 501
    },
    {
      "epoch": 1.1158129175946547,
      "loss": 1.3554702997207642,
      "loss_ce": 0.0037856735289096832,
      "loss_iou": 0.55078125,
      "loss_num": 0.0498046875,
      "loss_xval": 1.3515625,
      "num_input_tokens_seen": 28041640,
      "step": 501
    },
    {
      "epoch": 1.1180400890868596,
      "grad_norm": 25.75654411315918,
      "learning_rate": 1e-06,
      "loss": 1.1217,
      "num_input_tokens_seen": 28095584,
      "step": 502
    },
    {
      "epoch": 1.1180400890868596,
      "loss": 1.170928955078125,
      "loss_ce": 0.01809690147638321,
      "loss_iou": 0.45703125,
      "loss_num": 0.047607421875,
      "loss_xval": 1.15625,
      "num_input_tokens_seen": 28095584,
      "step": 502
    },
    {
      "epoch": 1.1202672605790647,
      "grad_norm": 25.978004455566406,
      "learning_rate": 1e-06,
      "loss": 1.3928,
      "num_input_tokens_seen": 28152476,
      "step": 503
    },
    {
      "epoch": 1.1202672605790647,
      "loss": 1.4479464292526245,
      "loss_ce": 0.001657373970374465,
      "loss_iou": 0.5625,
      "loss_num": 0.06396484375,
      "loss_xval": 1.4453125,
      "num_input_tokens_seen": 28152476,
      "step": 503
    },
    {
      "epoch": 1.1224944320712695,
      "grad_norm": 26.65591049194336,
      "learning_rate": 1e-06,
      "loss": 1.2271,
      "num_input_tokens_seen": 28207068,
      "step": 504
    },
    {
      "epoch": 1.1224944320712695,
      "loss": 1.1432322263717651,
      "loss_ce": 0.007001784630119801,
      "loss_iou": 0.455078125,
      "loss_num": 0.045654296875,
      "loss_xval": 1.1328125,
      "num_input_tokens_seen": 28207068,
      "step": 504
    },
    {
      "epoch": 1.1247216035634744,
      "grad_norm": 54.66550064086914,
      "learning_rate": 1e-06,
      "loss": 1.1682,
      "num_input_tokens_seen": 28259960,
      "step": 505
    },
    {
      "epoch": 1.1247216035634744,
      "loss": 1.2690539360046387,
      "loss_ce": 0.0009875732939690351,
      "loss_iou": 0.51953125,
      "loss_num": 0.0458984375,
      "loss_xval": 1.265625,
      "num_input_tokens_seen": 28259960,
      "step": 505
    },
    {
      "epoch": 1.1269487750556793,
      "grad_norm": 47.31111145019531,
      "learning_rate": 1e-06,
      "loss": 1.1786,
      "num_input_tokens_seen": 28320020,
      "step": 506
    },
    {
      "epoch": 1.1269487750556793,
      "loss": 1.4288506507873535,
      "loss_ce": 0.0045341607183218,
      "loss_iou": 0.58984375,
      "loss_num": 0.04931640625,
      "loss_xval": 1.421875,
      "num_input_tokens_seen": 28320020,
      "step": 506
    },
    {
      "epoch": 1.1291759465478841,
      "grad_norm": 19.817325592041016,
      "learning_rate": 1e-06,
      "loss": 1.2798,
      "num_input_tokens_seen": 28376252,
      "step": 507
    },
    {
      "epoch": 1.1291759465478841,
      "loss": 0.7173053622245789,
      "loss_ce": 0.0009967784862965345,
      "loss_iou": 0.294921875,
      "loss_num": 0.025634765625,
      "loss_xval": 0.71484375,
      "num_input_tokens_seen": 28376252,
      "step": 507
    },
    {
      "epoch": 1.131403118040089,
      "grad_norm": 29.8699951171875,
      "learning_rate": 1e-06,
      "loss": 1.0068,
      "num_input_tokens_seen": 28428956,
      "step": 508
    },
    {
      "epoch": 1.131403118040089,
      "loss": 1.0889060497283936,
      "loss_ce": 0.0010153307812288404,
      "loss_iou": 0.46484375,
      "loss_num": 0.031982421875,
      "loss_xval": 1.0859375,
      "num_input_tokens_seen": 28428956,
      "step": 508
    },
    {
      "epoch": 1.133630289532294,
      "grad_norm": 33.13240051269531,
      "learning_rate": 1e-06,
      "loss": 1.3036,
      "num_input_tokens_seen": 28484500,
      "step": 509
    },
    {
      "epoch": 1.133630289532294,
      "loss": 1.068528652191162,
      "loss_ce": 0.003343157237395644,
      "loss_iou": 0.4375,
      "loss_num": 0.037841796875,
      "loss_xval": 1.0625,
      "num_input_tokens_seen": 28484500,
      "step": 509
    },
    {
      "epoch": 1.135857461024499,
      "grad_norm": 28.9699764251709,
      "learning_rate": 1e-06,
      "loss": 0.9363,
      "num_input_tokens_seen": 28537936,
      "step": 510
    },
    {
      "epoch": 1.135857461024499,
      "loss": 0.9706677198410034,
      "loss_ce": 0.005823939107358456,
      "loss_iou": 0.419921875,
      "loss_num": 0.0252685546875,
      "loss_xval": 0.96484375,
      "num_input_tokens_seen": 28537936,
      "step": 510
    },
    {
      "epoch": 1.1380846325167038,
      "grad_norm": 24.593473434448242,
      "learning_rate": 1e-06,
      "loss": 1.2115,
      "num_input_tokens_seen": 28595228,
      "step": 511
    },
    {
      "epoch": 1.1380846325167038,
      "loss": 1.0602378845214844,
      "loss_ce": 0.0006675816257484257,
      "loss_iou": 0.455078125,
      "loss_num": 0.0294189453125,
      "loss_xval": 1.0625,
      "num_input_tokens_seen": 28595228,
      "step": 511
    },
    {
      "epoch": 1.1403118040089086,
      "grad_norm": 24.511798858642578,
      "learning_rate": 1e-06,
      "loss": 1.2554,
      "num_input_tokens_seen": 28650712,
      "step": 512
    },
    {
      "epoch": 1.1403118040089086,
      "loss": 1.273465871810913,
      "loss_ce": 0.0049112411215901375,
      "loss_iou": 0.55859375,
      "loss_num": 0.0306396484375,
      "loss_xval": 1.265625,
      "num_input_tokens_seen": 28650712,
      "step": 512
    },
    {
      "epoch": 1.1425389755011135,
      "grad_norm": 23.49129295349121,
      "learning_rate": 1e-06,
      "loss": 1.1208,
      "num_input_tokens_seen": 28708008,
      "step": 513
    },
    {
      "epoch": 1.1425389755011135,
      "loss": 1.1443027257919312,
      "loss_ce": 0.005630870349705219,
      "loss_iou": 0.46875,
      "loss_num": 0.040283203125,
      "loss_xval": 1.140625,
      "num_input_tokens_seen": 28708008,
      "step": 513
    },
    {
      "epoch": 1.1447661469933186,
      "grad_norm": 22.09214973449707,
      "learning_rate": 1e-06,
      "loss": 1.2605,
      "num_input_tokens_seen": 28762704,
      "step": 514
    },
    {
      "epoch": 1.1447661469933186,
      "loss": 1.2734076976776123,
      "loss_ce": 0.018280737102031708,
      "loss_iou": 0.484375,
      "loss_num": 0.05712890625,
      "loss_xval": 1.2578125,
      "num_input_tokens_seen": 28762704,
      "step": 514
    },
    {
      "epoch": 1.1469933184855234,
      "grad_norm": 36.97422790527344,
      "learning_rate": 1e-06,
      "loss": 1.2453,
      "num_input_tokens_seen": 28820084,
      "step": 515
    },
    {
      "epoch": 1.1469933184855234,
      "loss": 1.241039514541626,
      "loss_ce": 0.0012934368569403887,
      "loss_iou": 0.52734375,
      "loss_num": 0.037109375,
      "loss_xval": 1.2421875,
      "num_input_tokens_seen": 28820084,
      "step": 515
    },
    {
      "epoch": 1.1492204899777283,
      "grad_norm": 19.85295867919922,
      "learning_rate": 1e-06,
      "loss": 1.346,
      "num_input_tokens_seen": 28874500,
      "step": 516
    },
    {
      "epoch": 1.1492204899777283,
      "loss": 1.231957197189331,
      "loss_ce": 0.0005119675770401955,
      "loss_iou": 0.5,
      "loss_num": 0.046630859375,
      "loss_xval": 1.234375,
      "num_input_tokens_seen": 28874500,
      "step": 516
    },
    {
      "epoch": 1.1514476614699332,
      "grad_norm": 34.39645767211914,
      "learning_rate": 1e-06,
      "loss": 1.3985,
      "num_input_tokens_seen": 28931164,
      "step": 517
    },
    {
      "epoch": 1.1514476614699332,
      "loss": 1.2988076210021973,
      "loss_ce": 0.0019326311303302646,
      "loss_iou": 0.50390625,
      "loss_num": 0.05810546875,
      "loss_xval": 1.296875,
      "num_input_tokens_seen": 28931164,
      "step": 517
    },
    {
      "epoch": 1.153674832962138,
      "grad_norm": 20.85076141357422,
      "learning_rate": 1e-06,
      "loss": 1.3304,
      "num_input_tokens_seen": 28987576,
      "step": 518
    },
    {
      "epoch": 1.153674832962138,
      "loss": 1.326310157775879,
      "loss_ce": 0.005509458482265472,
      "loss_iou": 0.5234375,
      "loss_num": 0.0546875,
      "loss_xval": 1.3203125,
      "num_input_tokens_seen": 28987576,
      "step": 518
    },
    {
      "epoch": 1.1559020044543429,
      "grad_norm": 25.91316795349121,
      "learning_rate": 1e-06,
      "loss": 1.0788,
      "num_input_tokens_seen": 29045416,
      "step": 519
    },
    {
      "epoch": 1.1559020044543429,
      "loss": 0.9822598695755005,
      "loss_ce": 0.0008145694737322628,
      "loss_iou": 0.4296875,
      "loss_num": 0.0242919921875,
      "loss_xval": 0.98046875,
      "num_input_tokens_seen": 29045416,
      "step": 519
    },
    {
      "epoch": 1.158129175946548,
      "grad_norm": 19.741249084472656,
      "learning_rate": 1e-06,
      "loss": 1.179,
      "num_input_tokens_seen": 29100408,
      "step": 520
    },
    {
      "epoch": 1.158129175946548,
      "loss": 1.2618613243103027,
      "loss_ce": 0.0020956983789801598,
      "loss_iou": 0.486328125,
      "loss_num": 0.05712890625,
      "loss_xval": 1.2578125,
      "num_input_tokens_seen": 29100408,
      "step": 520
    },
    {
      "epoch": 1.1603563474387528,
      "grad_norm": 18.815519332885742,
      "learning_rate": 1e-06,
      "loss": 1.0117,
      "num_input_tokens_seen": 29158568,
      "step": 521
    },
    {
      "epoch": 1.1603563474387528,
      "loss": 1.15644371509552,
      "loss_ce": 0.00605308311060071,
      "loss_iou": 0.482421875,
      "loss_num": 0.037109375,
      "loss_xval": 1.1484375,
      "num_input_tokens_seen": 29158568,
      "step": 521
    },
    {
      "epoch": 1.1625835189309577,
      "grad_norm": 37.82133865356445,
      "learning_rate": 1e-06,
      "loss": 1.617,
      "num_input_tokens_seen": 29216180,
      "step": 522
    },
    {
      "epoch": 1.1625835189309577,
      "loss": 1.670715570449829,
      "loss_ce": 0.003723357105627656,
      "loss_iou": 0.63671875,
      "loss_num": 0.0791015625,
      "loss_xval": 1.6640625,
      "num_input_tokens_seen": 29216180,
      "step": 522
    },
    {
      "epoch": 1.1648106904231625,
      "grad_norm": 31.2655029296875,
      "learning_rate": 1e-06,
      "loss": 1.1777,
      "num_input_tokens_seen": 29270996,
      "step": 523
    },
    {
      "epoch": 1.1648106904231625,
      "loss": 1.0179691314697266,
      "loss_ce": 0.0013676062226295471,
      "loss_iou": 0.419921875,
      "loss_num": 0.035400390625,
      "loss_xval": 1.015625,
      "num_input_tokens_seen": 29270996,
      "step": 523
    },
    {
      "epoch": 1.1670378619153674,
      "grad_norm": 32.841373443603516,
      "learning_rate": 1e-06,
      "loss": 1.4131,
      "num_input_tokens_seen": 29327600,
      "step": 524
    },
    {
      "epoch": 1.1670378619153674,
      "loss": 1.0806894302368164,
      "loss_ce": 0.0006113001727499068,
      "loss_iou": 0.439453125,
      "loss_num": 0.039794921875,
      "loss_xval": 1.078125,
      "num_input_tokens_seen": 29327600,
      "step": 524
    },
    {
      "epoch": 1.1692650334075725,
      "grad_norm": 14.620657920837402,
      "learning_rate": 1e-06,
      "loss": 1.228,
      "num_input_tokens_seen": 29383468,
      "step": 525
    },
    {
      "epoch": 1.1692650334075725,
      "loss": 1.1438016891479492,
      "loss_ce": 0.0007352291722781956,
      "loss_iou": 0.49609375,
      "loss_num": 0.030517578125,
      "loss_xval": 1.140625,
      "num_input_tokens_seen": 29383468,
      "step": 525
    },
    {
      "epoch": 1.1714922048997773,
      "grad_norm": 38.983917236328125,
      "learning_rate": 1e-06,
      "loss": 1.0997,
      "num_input_tokens_seen": 29438716,
      "step": 526
    },
    {
      "epoch": 1.1714922048997773,
      "loss": 1.1212483644485474,
      "loss_ce": 0.005525712855160236,
      "loss_iou": 0.447265625,
      "loss_num": 0.04443359375,
      "loss_xval": 1.1171875,
      "num_input_tokens_seen": 29438716,
      "step": 526
    },
    {
      "epoch": 1.1737193763919822,
      "grad_norm": 38.69413375854492,
      "learning_rate": 1e-06,
      "loss": 1.2774,
      "num_input_tokens_seen": 29494100,
      "step": 527
    },
    {
      "epoch": 1.1737193763919822,
      "loss": 0.9909718036651611,
      "loss_ce": 0.0007374268025159836,
      "loss_iou": 0.375,
      "loss_num": 0.0478515625,
      "loss_xval": 0.9921875,
      "num_input_tokens_seen": 29494100,
      "step": 527
    },
    {
      "epoch": 1.175946547884187,
      "grad_norm": 20.045331954956055,
      "learning_rate": 1e-06,
      "loss": 1.0815,
      "num_input_tokens_seen": 29549828,
      "step": 528
    },
    {
      "epoch": 1.175946547884187,
      "loss": 0.9373371005058289,
      "loss_ce": 0.004231642000377178,
      "loss_iou": 0.376953125,
      "loss_num": 0.03564453125,
      "loss_xval": 0.93359375,
      "num_input_tokens_seen": 29549828,
      "step": 528
    },
    {
      "epoch": 1.178173719376392,
      "grad_norm": 29.08791732788086,
      "learning_rate": 1e-06,
      "loss": 1.0875,
      "num_input_tokens_seen": 29605928,
      "step": 529
    },
    {
      "epoch": 1.178173719376392,
      "loss": 1.0599498748779297,
      "loss_ce": 0.0018443877343088388,
      "loss_iou": 0.404296875,
      "loss_num": 0.050537109375,
      "loss_xval": 1.0546875,
      "num_input_tokens_seen": 29605928,
      "step": 529
    },
    {
      "epoch": 1.1804008908685968,
      "grad_norm": 19.786157608032227,
      "learning_rate": 1e-06,
      "loss": 1.2501,
      "num_input_tokens_seen": 29661192,
      "step": 530
    },
    {
      "epoch": 1.1804008908685968,
      "loss": 1.2494618892669678,
      "loss_ce": 0.0014150183415040374,
      "loss_iou": 0.46484375,
      "loss_num": 0.0634765625,
      "loss_xval": 1.25,
      "num_input_tokens_seen": 29661192,
      "step": 530
    },
    {
      "epoch": 1.1826280623608019,
      "grad_norm": 56.539695739746094,
      "learning_rate": 1e-06,
      "loss": 1.2396,
      "num_input_tokens_seen": 29716704,
      "step": 531
    },
    {
      "epoch": 1.1826280623608019,
      "loss": 1.2581884860992432,
      "loss_ce": 0.0008642220636829734,
      "loss_iou": 0.546875,
      "loss_num": 0.033447265625,
      "loss_xval": 1.2578125,
      "num_input_tokens_seen": 29716704,
      "step": 531
    },
    {
      "epoch": 1.1848552338530067,
      "grad_norm": 15.872360229492188,
      "learning_rate": 1e-06,
      "loss": 1.0118,
      "num_input_tokens_seen": 29775024,
      "step": 532
    },
    {
      "epoch": 1.1848552338530067,
      "loss": 0.7084531784057617,
      "loss_ce": 0.0009336533839814365,
      "loss_iou": 0.296875,
      "loss_num": 0.0228271484375,
      "loss_xval": 0.70703125,
      "num_input_tokens_seen": 29775024,
      "step": 532
    },
    {
      "epoch": 1.1870824053452116,
      "grad_norm": 20.187984466552734,
      "learning_rate": 1e-06,
      "loss": 1.2427,
      "num_input_tokens_seen": 29830880,
      "step": 533
    },
    {
      "epoch": 1.1870824053452116,
      "loss": 1.1666477918624878,
      "loss_ce": 0.001608791295439005,
      "loss_iou": 0.50390625,
      "loss_num": 0.03125,
      "loss_xval": 1.1640625,
      "num_input_tokens_seen": 29830880,
      "step": 533
    },
    {
      "epoch": 1.1893095768374164,
      "grad_norm": 30.441001892089844,
      "learning_rate": 1e-06,
      "loss": 0.9994,
      "num_input_tokens_seen": 29888668,
      "step": 534
    },
    {
      "epoch": 1.1893095768374164,
      "loss": 1.0735273361206055,
      "loss_ce": 0.00077339808922261,
      "loss_iou": 0.453125,
      "loss_num": 0.033203125,
      "loss_xval": 1.0703125,
      "num_input_tokens_seen": 29888668,
      "step": 534
    },
    {
      "epoch": 1.1915367483296213,
      "grad_norm": 24.198322296142578,
      "learning_rate": 1e-06,
      "loss": 0.9697,
      "num_input_tokens_seen": 29946592,
      "step": 535
    },
    {
      "epoch": 1.1915367483296213,
      "loss": 0.9628783464431763,
      "loss_ce": 0.0009642738732509315,
      "loss_iou": 0.376953125,
      "loss_num": 0.04150390625,
      "loss_xval": 0.9609375,
      "num_input_tokens_seen": 29946592,
      "step": 535
    },
    {
      "epoch": 1.1937639198218264,
      "grad_norm": 30.70039939880371,
      "learning_rate": 1e-06,
      "loss": 1.2909,
      "num_input_tokens_seen": 30004576,
      "step": 536
    },
    {
      "epoch": 1.1937639198218264,
      "loss": 1.1397030353546143,
      "loss_ce": 0.0010311845690011978,
      "loss_iou": 0.45703125,
      "loss_num": 0.04541015625,
      "loss_xval": 1.140625,
      "num_input_tokens_seen": 30004576,
      "step": 536
    },
    {
      "epoch": 1.1959910913140313,
      "grad_norm": 40.41281509399414,
      "learning_rate": 1e-06,
      "loss": 1.3186,
      "num_input_tokens_seen": 30056972,
      "step": 537
    },
    {
      "epoch": 1.1959910913140313,
      "loss": 1.1098320484161377,
      "loss_ce": 0.0004569512093439698,
      "loss_iou": 0.490234375,
      "loss_num": 0.0260009765625,
      "loss_xval": 1.109375,
      "num_input_tokens_seen": 30056972,
      "step": 537
    },
    {
      "epoch": 1.1982182628062361,
      "grad_norm": 25.64579963684082,
      "learning_rate": 1e-06,
      "loss": 1.2267,
      "num_input_tokens_seen": 30111776,
      "step": 538
    },
    {
      "epoch": 1.1982182628062361,
      "loss": 1.4003875255584717,
      "loss_ce": 0.002682518446817994,
      "loss_iou": 0.5625,
      "loss_num": 0.05517578125,
      "loss_xval": 1.3984375,
      "num_input_tokens_seen": 30111776,
      "step": 538
    },
    {
      "epoch": 1.200445434298441,
      "grad_norm": 62.563392639160156,
      "learning_rate": 1e-06,
      "loss": 1.1343,
      "num_input_tokens_seen": 30168204,
      "step": 539
    },
    {
      "epoch": 1.200445434298441,
      "loss": 1.2498741149902344,
      "loss_ce": 0.00353623297996819,
      "loss_iou": 0.48828125,
      "loss_num": 0.053955078125,
      "loss_xval": 1.25,
      "num_input_tokens_seen": 30168204,
      "step": 539
    },
    {
      "epoch": 1.2026726057906458,
      "grad_norm": 24.099092483520508,
      "learning_rate": 1e-06,
      "loss": 1.5677,
      "num_input_tokens_seen": 30222172,
      "step": 540
    },
    {
      "epoch": 1.2026726057906458,
      "loss": 1.189444899559021,
      "loss_ce": 0.001456591533496976,
      "loss_iou": 0.470703125,
      "loss_num": 0.049560546875,
      "loss_xval": 1.1875,
      "num_input_tokens_seen": 30222172,
      "step": 540
    },
    {
      "epoch": 1.2048997772828507,
      "grad_norm": 14.803204536437988,
      "learning_rate": 1e-06,
      "loss": 1.2895,
      "num_input_tokens_seen": 30277260,
      "step": 541
    },
    {
      "epoch": 1.2048997772828507,
      "loss": 1.3928616046905518,
      "loss_ce": 0.0027249492704868317,
      "loss_iou": 0.5625,
      "loss_num": 0.052490234375,
      "loss_xval": 1.390625,
      "num_input_tokens_seen": 30277260,
      "step": 541
    },
    {
      "epoch": 1.2071269487750558,
      "grad_norm": 56.08818054199219,
      "learning_rate": 1e-06,
      "loss": 1.4108,
      "num_input_tokens_seen": 30333120,
      "step": 542
    },
    {
      "epoch": 1.2071269487750558,
      "loss": 1.4470124244689941,
      "loss_ce": 0.001211727038025856,
      "loss_iou": 0.59765625,
      "loss_num": 0.049560546875,
      "loss_xval": 1.4453125,
      "num_input_tokens_seen": 30333120,
      "step": 542
    },
    {
      "epoch": 1.2093541202672606,
      "grad_norm": 33.44521713256836,
      "learning_rate": 1e-06,
      "loss": 1.0997,
      "num_input_tokens_seen": 30385752,
      "step": 543
    },
    {
      "epoch": 1.2093541202672606,
      "loss": 0.8862332105636597,
      "loss_ce": 0.0009793277131393552,
      "loss_iou": 0.375,
      "loss_num": 0.027099609375,
      "loss_xval": 0.88671875,
      "num_input_tokens_seen": 30385752,
      "step": 543
    },
    {
      "epoch": 1.2115812917594655,
      "grad_norm": 20.260774612426758,
      "learning_rate": 1e-06,
      "loss": 1.3352,
      "num_input_tokens_seen": 30441532,
      "step": 544
    },
    {
      "epoch": 1.2115812917594655,
      "loss": 1.403501033782959,
      "loss_ce": 0.0006689808797091246,
      "loss_iou": 0.5234375,
      "loss_num": 0.0712890625,
      "loss_xval": 1.40625,
      "num_input_tokens_seen": 30441532,
      "step": 544
    },
    {
      "epoch": 1.2138084632516704,
      "grad_norm": 20.101301193237305,
      "learning_rate": 1e-06,
      "loss": 1.0824,
      "num_input_tokens_seen": 30495056,
      "step": 545
    },
    {
      "epoch": 1.2138084632516704,
      "loss": 1.0261245965957642,
      "loss_ce": 0.0004898015176877379,
      "loss_iou": 0.431640625,
      "loss_num": 0.0322265625,
      "loss_xval": 1.0234375,
      "num_input_tokens_seen": 30495056,
      "step": 545
    },
    {
      "epoch": 1.2160356347438752,
      "grad_norm": 27.02719497680664,
      "learning_rate": 1e-06,
      "loss": 1.1653,
      "num_input_tokens_seen": 30550680,
      "step": 546
    },
    {
      "epoch": 1.2160356347438752,
      "loss": 1.1834897994995117,
      "loss_ce": 0.0008725329535081983,
      "loss_iou": 0.5078125,
      "loss_num": 0.033447265625,
      "loss_xval": 1.1796875,
      "num_input_tokens_seen": 30550680,
      "step": 546
    },
    {
      "epoch": 1.2182628062360803,
      "grad_norm": 17.25623893737793,
      "learning_rate": 1e-06,
      "loss": 1.4252,
      "num_input_tokens_seen": 30606964,
      "step": 547
    },
    {
      "epoch": 1.2182628062360803,
      "loss": 1.1636087894439697,
      "loss_ce": 0.0036967378109693527,
      "loss_iou": 0.45703125,
      "loss_num": 0.049072265625,
      "loss_xval": 1.15625,
      "num_input_tokens_seen": 30606964,
      "step": 547
    },
    {
      "epoch": 1.2204899777282852,
      "grad_norm": 19.519060134887695,
      "learning_rate": 1e-06,
      "loss": 1.1066,
      "num_input_tokens_seen": 30663460,
      "step": 548
    },
    {
      "epoch": 1.2204899777282852,
      "loss": 1.1835308074951172,
      "loss_ce": 0.0018901234725490212,
      "loss_iou": 0.50390625,
      "loss_num": 0.034912109375,
      "loss_xval": 1.1796875,
      "num_input_tokens_seen": 30663460,
      "step": 548
    },
    {
      "epoch": 1.22271714922049,
      "grad_norm": 25.55568504333496,
      "learning_rate": 1e-06,
      "loss": 1.2514,
      "num_input_tokens_seen": 30721628,
      "step": 549
    },
    {
      "epoch": 1.22271714922049,
      "loss": 1.3971433639526367,
      "loss_ce": 0.0011472116457298398,
      "loss_iou": 0.5625,
      "loss_num": 0.0546875,
      "loss_xval": 1.3984375,
      "num_input_tokens_seen": 30721628,
      "step": 549
    },
    {
      "epoch": 1.2249443207126949,
      "grad_norm": 31.964738845825195,
      "learning_rate": 1e-06,
      "loss": 1.3476,
      "num_input_tokens_seen": 30777708,
      "step": 550
    },
    {
      "epoch": 1.2249443207126949,
      "loss": 1.316117286682129,
      "loss_ce": 0.0006876069819554687,
      "loss_iou": 0.5546875,
      "loss_num": 0.041259765625,
      "loss_xval": 1.3125,
      "num_input_tokens_seen": 30777708,
      "step": 550
    },
    {
      "epoch": 1.2271714922048997,
      "grad_norm": 19.710111618041992,
      "learning_rate": 1e-06,
      "loss": 1.3899,
      "num_input_tokens_seen": 30829420,
      "step": 551
    },
    {
      "epoch": 1.2271714922048997,
      "loss": 1.4299864768981934,
      "loss_ce": 0.01006464846432209,
      "loss_iou": 0.57421875,
      "loss_num": 0.053955078125,
      "loss_xval": 1.421875,
      "num_input_tokens_seen": 30829420,
      "step": 551
    },
    {
      "epoch": 1.2293986636971046,
      "grad_norm": 49.47341537475586,
      "learning_rate": 1e-06,
      "loss": 1.1639,
      "num_input_tokens_seen": 30887668,
      "step": 552
    },
    {
      "epoch": 1.2293986636971046,
      "loss": 1.0400408506393433,
      "loss_ce": 0.0014665467897430062,
      "loss_iou": 0.396484375,
      "loss_num": 0.04931640625,
      "loss_xval": 1.0390625,
      "num_input_tokens_seen": 30887668,
      "step": 552
    },
    {
      "epoch": 1.2316258351893095,
      "grad_norm": 21.37660789489746,
      "learning_rate": 1e-06,
      "loss": 1.2537,
      "num_input_tokens_seen": 30942632,
      "step": 553
    },
    {
      "epoch": 1.2316258351893095,
      "loss": 1.3482673168182373,
      "loss_ce": 0.0020758796017616987,
      "loss_iou": 0.53515625,
      "loss_num": 0.0556640625,
      "loss_xval": 1.34375,
      "num_input_tokens_seen": 30942632,
      "step": 553
    },
    {
      "epoch": 1.2338530066815145,
      "grad_norm": 15.992125511169434,
      "learning_rate": 1e-06,
      "loss": 1.1998,
      "num_input_tokens_seen": 31000952,
      "step": 554
    },
    {
      "epoch": 1.2338530066815145,
      "loss": 1.3668153285980225,
      "loss_ce": 0.0006044998299330473,
      "loss_iou": 0.5859375,
      "loss_num": 0.039306640625,
      "loss_xval": 1.3671875,
      "num_input_tokens_seen": 31000952,
      "step": 554
    },
    {
      "epoch": 1.2360801781737194,
      "grad_norm": 22.379846572875977,
      "learning_rate": 1e-06,
      "loss": 1.3172,
      "num_input_tokens_seen": 31058164,
      "step": 555
    },
    {
      "epoch": 1.2360801781737194,
      "loss": 1.7307417392730713,
      "loss_ce": 0.004179192706942558,
      "loss_iou": 0.71484375,
      "loss_num": 0.0595703125,
      "loss_xval": 1.7265625,
      "num_input_tokens_seen": 31058164,
      "step": 555
    },
    {
      "epoch": 1.2383073496659243,
      "grad_norm": 17.48279571533203,
      "learning_rate": 1e-06,
      "loss": 1.0391,
      "num_input_tokens_seen": 31113116,
      "step": 556
    },
    {
      "epoch": 1.2383073496659243,
      "loss": 1.0837606191635132,
      "loss_ce": 0.0009969680104404688,
      "loss_iou": 0.380859375,
      "loss_num": 0.06396484375,
      "loss_xval": 1.0859375,
      "num_input_tokens_seen": 31113116,
      "step": 556
    },
    {
      "epoch": 1.2405345211581291,
      "grad_norm": 26.064071655273438,
      "learning_rate": 1e-06,
      "loss": 1.2376,
      "num_input_tokens_seen": 31172128,
      "step": 557
    },
    {
      "epoch": 1.2405345211581291,
      "loss": 1.2718275785446167,
      "loss_ce": 0.0022963983938097954,
      "loss_iou": 0.52734375,
      "loss_num": 0.04248046875,
      "loss_xval": 1.265625,
      "num_input_tokens_seen": 31172128,
      "step": 557
    },
    {
      "epoch": 1.242761692650334,
      "grad_norm": 27.61939811706543,
      "learning_rate": 1e-06,
      "loss": 0.9737,
      "num_input_tokens_seen": 31227572,
      "step": 558
    },
    {
      "epoch": 1.242761692650334,
      "loss": 1.1241004467010498,
      "loss_ce": 0.0010536747286096215,
      "loss_iou": 0.47265625,
      "loss_num": 0.03564453125,
      "loss_xval": 1.125,
      "num_input_tokens_seen": 31227572,
      "step": 558
    },
    {
      "epoch": 1.244988864142539,
      "grad_norm": 23.97410774230957,
      "learning_rate": 1e-06,
      "loss": 1.2226,
      "num_input_tokens_seen": 31282092,
      "step": 559
    },
    {
      "epoch": 1.244988864142539,
      "loss": 1.2522716522216797,
      "loss_ce": 0.000806744210422039,
      "loss_iou": 0.484375,
      "loss_num": 0.056396484375,
      "loss_xval": 1.25,
      "num_input_tokens_seen": 31282092,
      "step": 559
    },
    {
      "epoch": 1.247216035634744,
      "grad_norm": 18.71125602722168,
      "learning_rate": 1e-06,
      "loss": 1.0672,
      "num_input_tokens_seen": 31340076,
      "step": 560
    },
    {
      "epoch": 1.247216035634744,
      "loss": 1.1468454599380493,
      "loss_ce": 0.0008493656641803682,
      "loss_iou": 0.482421875,
      "loss_num": 0.036376953125,
      "loss_xval": 1.1484375,
      "num_input_tokens_seen": 31340076,
      "step": 560
    },
    {
      "epoch": 1.2494432071269488,
      "grad_norm": 70.7640151977539,
      "learning_rate": 1e-06,
      "loss": 1.2216,
      "num_input_tokens_seen": 31396448,
      "step": 561
    },
    {
      "epoch": 1.2494432071269488,
      "loss": 1.2371618747711182,
      "loss_ce": 0.013528996147215366,
      "loss_iou": 0.515625,
      "loss_num": 0.038330078125,
      "loss_xval": 1.2265625,
      "num_input_tokens_seen": 31396448,
      "step": 561
    },
    {
      "epoch": 1.2516703786191536,
      "grad_norm": 19.904014587402344,
      "learning_rate": 1e-06,
      "loss": 1.354,
      "num_input_tokens_seen": 31453460,
      "step": 562
    },
    {
      "epoch": 1.2516703786191536,
      "loss": 1.3166056871414185,
      "loss_ce": 0.00117595330812037,
      "loss_iou": 0.5234375,
      "loss_num": 0.05322265625,
      "loss_xval": 1.3125,
      "num_input_tokens_seen": 31453460,
      "step": 562
    },
    {
      "epoch": 1.2538975501113585,
      "grad_norm": 24.498027801513672,
      "learning_rate": 1e-06,
      "loss": 1.0603,
      "num_input_tokens_seen": 31509260,
      "step": 563
    },
    {
      "epoch": 1.2538975501113585,
      "loss": 1.2601665258407593,
      "loss_ce": 0.000645035644993186,
      "loss_iou": 0.53515625,
      "loss_num": 0.037841796875,
      "loss_xval": 1.2578125,
      "num_input_tokens_seen": 31509260,
      "step": 563
    },
    {
      "epoch": 1.2561247216035634,
      "grad_norm": 17.997339248657227,
      "learning_rate": 1e-06,
      "loss": 1.2419,
      "num_input_tokens_seen": 31561832,
      "step": 564
    },
    {
      "epoch": 1.2561247216035634,
      "loss": 1.050682783126831,
      "loss_ce": 0.002831157995387912,
      "loss_iou": 0.4453125,
      "loss_num": 0.031494140625,
      "loss_xval": 1.046875,
      "num_input_tokens_seen": 31561832,
      "step": 564
    },
    {
      "epoch": 1.2583518930957684,
      "grad_norm": 25.669340133666992,
      "learning_rate": 1e-06,
      "loss": 1.0692,
      "num_input_tokens_seen": 31619832,
      "step": 565
    },
    {
      "epoch": 1.2583518930957684,
      "loss": 1.0688444375991821,
      "loss_ce": 0.0009733220795169473,
      "loss_iou": 0.4453125,
      "loss_num": 0.035888671875,
      "loss_xval": 1.0703125,
      "num_input_tokens_seen": 31619832,
      "step": 565
    },
    {
      "epoch": 1.2605790645879733,
      "grad_norm": 16.837244033813477,
      "learning_rate": 1e-06,
      "loss": 1.2095,
      "num_input_tokens_seen": 31676760,
      "step": 566
    },
    {
      "epoch": 1.2605790645879733,
      "loss": 1.536379337310791,
      "loss_ce": 0.0017113613430410624,
      "loss_iou": 0.61328125,
      "loss_num": 0.0615234375,
      "loss_xval": 1.53125,
      "num_input_tokens_seen": 31676760,
      "step": 566
    },
    {
      "epoch": 1.2628062360801782,
      "grad_norm": 16.443872451782227,
      "learning_rate": 1e-06,
      "loss": 1.4209,
      "num_input_tokens_seen": 31730956,
      "step": 567
    },
    {
      "epoch": 1.2628062360801782,
      "loss": 1.459883213043213,
      "loss_ce": 0.0011430047452449799,
      "loss_iou": 0.56640625,
      "loss_num": 0.06591796875,
      "loss_xval": 1.4609375,
      "num_input_tokens_seen": 31730956,
      "step": 567
    },
    {
      "epoch": 1.265033407572383,
      "grad_norm": 24.123031616210938,
      "learning_rate": 1e-06,
      "loss": 1.0963,
      "num_input_tokens_seen": 31786932,
      "step": 568
    },
    {
      "epoch": 1.265033407572383,
      "loss": 1.188434362411499,
      "loss_ce": 0.0009342934936285019,
      "loss_iou": 0.47265625,
      "loss_num": 0.048828125,
      "loss_xval": 1.1875,
      "num_input_tokens_seen": 31786932,
      "step": 568
    },
    {
      "epoch": 1.267260579064588,
      "grad_norm": 20.135202407836914,
      "learning_rate": 1e-06,
      "loss": 1.1357,
      "num_input_tokens_seen": 31843536,
      "step": 569
    },
    {
      "epoch": 1.267260579064588,
      "loss": 1.4373984336853027,
      "loss_ce": 0.001607370562851429,
      "loss_iou": 0.53125,
      "loss_num": 0.07470703125,
      "loss_xval": 1.4375,
      "num_input_tokens_seen": 31843536,
      "step": 569
    },
    {
      "epoch": 1.269487750556793,
      "grad_norm": 34.270626068115234,
      "learning_rate": 1e-06,
      "loss": 1.2551,
      "num_input_tokens_seen": 31898080,
      "step": 570
    },
    {
      "epoch": 1.269487750556793,
      "loss": 1.2200250625610352,
      "loss_ce": 0.0007868170505389571,
      "loss_iou": 0.478515625,
      "loss_num": 0.05224609375,
      "loss_xval": 1.21875,
      "num_input_tokens_seen": 31898080,
      "step": 570
    },
    {
      "epoch": 1.2717149220489978,
      "grad_norm": 19.55361557006836,
      "learning_rate": 1e-06,
      "loss": 1.1386,
      "num_input_tokens_seen": 31951824,
      "step": 571
    },
    {
      "epoch": 1.2717149220489978,
      "loss": 1.242485761642456,
      "loss_ce": 0.0007865370716899633,
      "loss_iou": 0.5078125,
      "loss_num": 0.045654296875,
      "loss_xval": 1.2421875,
      "num_input_tokens_seen": 31951824,
      "step": 571
    },
    {
      "epoch": 1.2739420935412027,
      "grad_norm": 64.2752456665039,
      "learning_rate": 1e-06,
      "loss": 1.3344,
      "num_input_tokens_seen": 32006760,
      "step": 572
    },
    {
      "epoch": 1.2739420935412027,
      "loss": 0.9924131035804749,
      "loss_ce": 0.002422844059765339,
      "loss_iou": 0.41796875,
      "loss_num": 0.03076171875,
      "loss_xval": 0.98828125,
      "num_input_tokens_seen": 32006760,
      "step": 572
    },
    {
      "epoch": 1.2761692650334076,
      "grad_norm": 18.450927734375,
      "learning_rate": 1e-06,
      "loss": 1.1749,
      "num_input_tokens_seen": 32065980,
      "step": 573
    },
    {
      "epoch": 1.2761692650334076,
      "loss": 1.0965838432312012,
      "loss_ce": 0.0023455810733139515,
      "loss_iou": 0.42578125,
      "loss_num": 0.04833984375,
      "loss_xval": 1.09375,
      "num_input_tokens_seen": 32065980,
      "step": 573
    },
    {
      "epoch": 1.2783964365256124,
      "grad_norm": 20.623170852661133,
      "learning_rate": 1e-06,
      "loss": 1.0463,
      "num_input_tokens_seen": 32120756,
      "step": 574
    },
    {
      "epoch": 1.2783964365256124,
      "loss": 1.1162598133087158,
      "loss_ce": 0.0005371532752178609,
      "loss_iou": 0.48046875,
      "loss_num": 0.0306396484375,
      "loss_xval": 1.1171875,
      "num_input_tokens_seen": 32120756,
      "step": 574
    },
    {
      "epoch": 1.2806236080178173,
      "grad_norm": 30.17593002319336,
      "learning_rate": 1e-06,
      "loss": 1.0251,
      "num_input_tokens_seen": 32175548,
      "step": 575
    },
    {
      "epoch": 1.2806236080178173,
      "loss": 1.20136296749115,
      "loss_ce": 0.000679333577863872,
      "loss_iou": 0.5078125,
      "loss_num": 0.037353515625,
      "loss_xval": 1.203125,
      "num_input_tokens_seen": 32175548,
      "step": 575
    },
    {
      "epoch": 1.2828507795100224,
      "grad_norm": 24.09958839416504,
      "learning_rate": 1e-06,
      "loss": 1.2353,
      "num_input_tokens_seen": 32230428,
      "step": 576
    },
    {
      "epoch": 1.2828507795100224,
      "loss": 1.3486398458480835,
      "loss_ce": 0.002448498737066984,
      "loss_iou": 0.5625,
      "loss_num": 0.0439453125,
      "loss_xval": 1.34375,
      "num_input_tokens_seen": 32230428,
      "step": 576
    },
    {
      "epoch": 1.2850779510022272,
      "grad_norm": 25.068777084350586,
      "learning_rate": 1e-06,
      "loss": 1.1451,
      "num_input_tokens_seen": 32282324,
      "step": 577
    },
    {
      "epoch": 1.2850779510022272,
      "loss": 1.3363144397735596,
      "loss_ce": 0.0011094561778008938,
      "loss_iou": 0.498046875,
      "loss_num": 0.06787109375,
      "loss_xval": 1.3359375,
      "num_input_tokens_seen": 32282324,
      "step": 577
    },
    {
      "epoch": 1.287305122494432,
      "grad_norm": 21.740449905395508,
      "learning_rate": 1e-06,
      "loss": 0.9655,
      "num_input_tokens_seen": 32339156,
      "step": 578
    },
    {
      "epoch": 1.287305122494432,
      "loss": 0.8180431127548218,
      "loss_ce": 0.0009044220205396414,
      "loss_iou": 0.33984375,
      "loss_num": 0.02783203125,
      "loss_xval": 0.81640625,
      "num_input_tokens_seen": 32339156,
      "step": 578
    },
    {
      "epoch": 1.289532293986637,
      "grad_norm": 70.63128662109375,
      "learning_rate": 1e-06,
      "loss": 1.2932,
      "num_input_tokens_seen": 32394596,
      "step": 579
    },
    {
      "epoch": 1.289532293986637,
      "loss": 1.4371894598007202,
      "loss_ce": 0.0006660318467766047,
      "loss_iou": 0.5546875,
      "loss_num": 0.06640625,
      "loss_xval": 1.4375,
      "num_input_tokens_seen": 32394596,
      "step": 579
    },
    {
      "epoch": 1.2917594654788418,
      "grad_norm": 23.278512954711914,
      "learning_rate": 1e-06,
      "loss": 1.1461,
      "num_input_tokens_seen": 32452512,
      "step": 580
    },
    {
      "epoch": 1.2917594654788418,
      "loss": 1.3186371326446533,
      "loss_ce": 0.003695748746395111,
      "loss_iou": 0.494140625,
      "loss_num": 0.0654296875,
      "loss_xval": 1.3125,
      "num_input_tokens_seen": 32452512,
      "step": 580
    },
    {
      "epoch": 1.2939866369710469,
      "grad_norm": 21.15460205078125,
      "learning_rate": 1e-06,
      "loss": 1.1878,
      "num_input_tokens_seen": 32507568,
      "step": 581
    },
    {
      "epoch": 1.2939866369710469,
      "loss": 1.161703109741211,
      "loss_ce": 0.0027676241006702185,
      "loss_iou": 0.474609375,
      "loss_num": 0.0419921875,
      "loss_xval": 1.15625,
      "num_input_tokens_seen": 32507568,
      "step": 581
    },
    {
      "epoch": 1.2962138084632517,
      "grad_norm": 20.2482852935791,
      "learning_rate": 1e-06,
      "loss": 1.0799,
      "num_input_tokens_seen": 32564432,
      "step": 582
    },
    {
      "epoch": 1.2962138084632517,
      "loss": 1.0239415168762207,
      "loss_ce": 0.0012363542336970568,
      "loss_iou": 0.419921875,
      "loss_num": 0.036376953125,
      "loss_xval": 1.0234375,
      "num_input_tokens_seen": 32564432,
      "step": 582
    },
    {
      "epoch": 1.2984409799554566,
      "grad_norm": 39.36937713623047,
      "learning_rate": 1e-06,
      "loss": 0.9524,
      "num_input_tokens_seen": 32618096,
      "step": 583
    },
    {
      "epoch": 1.2984409799554566,
      "loss": 0.9080791473388672,
      "loss_ce": 0.0008525372250005603,
      "loss_iou": 0.34375,
      "loss_num": 0.044189453125,
      "loss_xval": 0.90625,
      "num_input_tokens_seen": 32618096,
      "step": 583
    },
    {
      "epoch": 1.3006681514476615,
      "grad_norm": 21.902511596679688,
      "learning_rate": 1e-06,
      "loss": 1.3062,
      "num_input_tokens_seen": 32674700,
      "step": 584
    },
    {
      "epoch": 1.3006681514476615,
      "loss": 1.2775499820709229,
      "loss_ce": 0.004112505819648504,
      "loss_iou": 0.5078125,
      "loss_num": 0.051025390625,
      "loss_xval": 1.2734375,
      "num_input_tokens_seen": 32674700,
      "step": 584
    },
    {
      "epoch": 1.3028953229398663,
      "grad_norm": 18.999221801757812,
      "learning_rate": 1e-06,
      "loss": 1.1638,
      "num_input_tokens_seen": 32732620,
      "step": 585
    },
    {
      "epoch": 1.3028953229398663,
      "loss": 1.1583878993988037,
      "loss_ce": 0.0018937456188723445,
      "loss_iou": 0.431640625,
      "loss_num": 0.05859375,
      "loss_xval": 1.15625,
      "num_input_tokens_seen": 32732620,
      "step": 585
    },
    {
      "epoch": 1.3051224944320712,
      "grad_norm": 29.080703735351562,
      "learning_rate": 1e-06,
      "loss": 1.1476,
      "num_input_tokens_seen": 32790356,
      "step": 586
    },
    {
      "epoch": 1.3051224944320712,
      "loss": 1.168489933013916,
      "loss_ce": 0.0005212133983150125,
      "loss_iou": 0.48828125,
      "loss_num": 0.0380859375,
      "loss_xval": 1.171875,
      "num_input_tokens_seen": 32790356,
      "step": 586
    },
    {
      "epoch": 1.307349665924276,
      "grad_norm": 26.149927139282227,
      "learning_rate": 1e-06,
      "loss": 1.1816,
      "num_input_tokens_seen": 32846304,
      "step": 587
    },
    {
      "epoch": 1.307349665924276,
      "loss": 1.3775864839553833,
      "loss_ce": 0.0006334002828225493,
      "loss_iou": 0.56640625,
      "loss_num": 0.04931640625,
      "loss_xval": 1.375,
      "num_input_tokens_seen": 32846304,
      "step": 587
    },
    {
      "epoch": 1.3095768374164811,
      "grad_norm": 15.33914566040039,
      "learning_rate": 1e-06,
      "loss": 0.9549,
      "num_input_tokens_seen": 32901992,
      "step": 588
    },
    {
      "epoch": 1.3095768374164811,
      "loss": 0.846169114112854,
      "loss_ce": 0.001198451267555356,
      "loss_iou": 0.345703125,
      "loss_num": 0.03125,
      "loss_xval": 0.84375,
      "num_input_tokens_seen": 32901992,
      "step": 588
    },
    {
      "epoch": 1.311804008908686,
      "grad_norm": 45.818695068359375,
      "learning_rate": 1e-06,
      "loss": 1.2682,
      "num_input_tokens_seen": 32961136,
      "step": 589
    },
    {
      "epoch": 1.311804008908686,
      "loss": 1.3090804815292358,
      "loss_ce": 0.00048678729217499495,
      "loss_iou": 0.5703125,
      "loss_num": 0.033203125,
      "loss_xval": 1.3125,
      "num_input_tokens_seen": 32961136,
      "step": 589
    },
    {
      "epoch": 1.3140311804008908,
      "grad_norm": 31.00553321838379,
      "learning_rate": 1e-06,
      "loss": 1.1881,
      "num_input_tokens_seen": 33015672,
      "step": 590
    },
    {
      "epoch": 1.3140311804008908,
      "loss": 1.2077548503875732,
      "loss_ce": 0.0007235349621623755,
      "loss_iou": 0.48828125,
      "loss_num": 0.04541015625,
      "loss_xval": 1.203125,
      "num_input_tokens_seen": 33015672,
      "step": 590
    },
    {
      "epoch": 1.3162583518930957,
      "grad_norm": 32.82173156738281,
      "learning_rate": 1e-06,
      "loss": 1.4302,
      "num_input_tokens_seen": 33071504,
      "step": 591
    },
    {
      "epoch": 1.3162583518930957,
      "loss": 1.373945713043213,
      "loss_ce": 0.00041063432581722736,
      "loss_iou": 0.546875,
      "loss_num": 0.056640625,
      "loss_xval": 1.375,
      "num_input_tokens_seen": 33071504,
      "step": 591
    },
    {
      "epoch": 1.3184855233853008,
      "grad_norm": 19.010250091552734,
      "learning_rate": 1e-06,
      "loss": 0.9531,
      "num_input_tokens_seen": 33126464,
      "step": 592
    },
    {
      "epoch": 1.3184855233853008,
      "loss": 0.9081621766090393,
      "loss_ce": 0.002156274626031518,
      "loss_iou": 0.37109375,
      "loss_num": 0.032958984375,
      "loss_xval": 0.90625,
      "num_input_tokens_seen": 33126464,
      "step": 592
    },
    {
      "epoch": 1.3207126948775056,
      "grad_norm": 19.802019119262695,
      "learning_rate": 1e-06,
      "loss": 1.3826,
      "num_input_tokens_seen": 33182964,
      "step": 593
    },
    {
      "epoch": 1.3207126948775056,
      "loss": 1.3006547689437866,
      "loss_ce": 0.0008500526309944689,
      "loss_iou": 0.52734375,
      "loss_num": 0.049560546875,
      "loss_xval": 1.296875,
      "num_input_tokens_seen": 33182964,
      "step": 593
    },
    {
      "epoch": 1.3229398663697105,
      "grad_norm": 19.390378952026367,
      "learning_rate": 1e-06,
      "loss": 1.3825,
      "num_input_tokens_seen": 33237572,
      "step": 594
    },
    {
      "epoch": 1.3229398663697105,
      "loss": 1.126906156539917,
      "loss_ce": 0.0004412978305481374,
      "loss_iou": 0.486328125,
      "loss_num": 0.031005859375,
      "loss_xval": 1.125,
      "num_input_tokens_seen": 33237572,
      "step": 594
    },
    {
      "epoch": 1.3251670378619154,
      "grad_norm": 17.320398330688477,
      "learning_rate": 1e-06,
      "loss": 0.8407,
      "num_input_tokens_seen": 33292472,
      "step": 595
    },
    {
      "epoch": 1.3251670378619154,
      "loss": 0.7980577945709229,
      "loss_ce": 0.00679800333455205,
      "loss_iou": 0.33984375,
      "loss_num": 0.0223388671875,
      "loss_xval": 0.79296875,
      "num_input_tokens_seen": 33292472,
      "step": 595
    },
    {
      "epoch": 1.3273942093541202,
      "grad_norm": 22.934152603149414,
      "learning_rate": 1e-06,
      "loss": 1.2177,
      "num_input_tokens_seen": 33349680,
      "step": 596
    },
    {
      "epoch": 1.3273942093541202,
      "loss": 1.237870216369629,
      "loss_ce": 0.000809596327599138,
      "loss_iou": 0.50390625,
      "loss_num": 0.046875,
      "loss_xval": 1.234375,
      "num_input_tokens_seen": 33349680,
      "step": 596
    },
    {
      "epoch": 1.329621380846325,
      "grad_norm": 19.58114242553711,
      "learning_rate": 1e-06,
      "loss": 0.9529,
      "num_input_tokens_seen": 33407408,
      "step": 597
    },
    {
      "epoch": 1.329621380846325,
      "loss": 1.1202585697174072,
      "loss_ce": 0.0006297538056969643,
      "loss_iou": 0.443359375,
      "loss_num": 0.046630859375,
      "loss_xval": 1.1171875,
      "num_input_tokens_seen": 33407408,
      "step": 597
    },
    {
      "epoch": 1.33184855233853,
      "grad_norm": 34.02892303466797,
      "learning_rate": 1e-06,
      "loss": 1.4897,
      "num_input_tokens_seen": 33461988,
      "step": 598
    },
    {
      "epoch": 1.33184855233853,
      "loss": 1.4317845106124878,
      "loss_ce": 0.0016087474068626761,
      "loss_iou": 0.59765625,
      "loss_num": 0.04736328125,
      "loss_xval": 1.4296875,
      "num_input_tokens_seen": 33461988,
      "step": 598
    },
    {
      "epoch": 1.334075723830735,
      "grad_norm": 51.48478698730469,
      "learning_rate": 1e-06,
      "loss": 1.156,
      "num_input_tokens_seen": 33517444,
      "step": 599
    },
    {
      "epoch": 1.334075723830735,
      "loss": 0.9608005881309509,
      "loss_ce": 0.0010837747249752283,
      "loss_iou": 0.400390625,
      "loss_num": 0.03173828125,
      "loss_xval": 0.9609375,
      "num_input_tokens_seen": 33517444,
      "step": 599
    },
    {
      "epoch": 1.3363028953229399,
      "grad_norm": 24.541122436523438,
      "learning_rate": 1e-06,
      "loss": 1.1167,
      "num_input_tokens_seen": 33573432,
      "step": 600
    },
    {
      "epoch": 1.3363028953229399,
      "loss": 1.3346822261810303,
      "loss_ce": 0.0014302851632237434,
      "loss_iou": 0.578125,
      "loss_num": 0.035888671875,
      "loss_xval": 1.3359375,
      "num_input_tokens_seen": 33573432,
      "step": 600
    },
    {
      "epoch": 1.3385300668151447,
      "grad_norm": 15.516777038574219,
      "learning_rate": 1e-06,
      "loss": 1.2706,
      "num_input_tokens_seen": 33627888,
      "step": 601
    },
    {
      "epoch": 1.3385300668151447,
      "loss": 1.197658896446228,
      "loss_ce": 0.0008815636392682791,
      "loss_iou": 0.484375,
      "loss_num": 0.04541015625,
      "loss_xval": 1.1953125,
      "num_input_tokens_seen": 33627888,
      "step": 601
    },
    {
      "epoch": 1.3407572383073496,
      "grad_norm": 22.7963924407959,
      "learning_rate": 1e-06,
      "loss": 1.2063,
      "num_input_tokens_seen": 33685960,
      "step": 602
    },
    {
      "epoch": 1.3407572383073496,
      "loss": 1.1104426383972168,
      "loss_ce": 0.0008235453860834241,
      "loss_iou": 0.46484375,
      "loss_num": 0.0361328125,
      "loss_xval": 1.109375,
      "num_input_tokens_seen": 33685960,
      "step": 602
    },
    {
      "epoch": 1.3429844097995547,
      "grad_norm": 32.18126678466797,
      "learning_rate": 1e-06,
      "loss": 1.2209,
      "num_input_tokens_seen": 33742816,
      "step": 603
    },
    {
      "epoch": 1.3429844097995547,
      "loss": 1.1211140155792236,
      "loss_ce": 0.0007527406560257077,
      "loss_iou": 0.453125,
      "loss_num": 0.042724609375,
      "loss_xval": 1.1171875,
      "num_input_tokens_seen": 33742816,
      "step": 603
    },
    {
      "epoch": 1.3452115812917596,
      "grad_norm": 24.347137451171875,
      "learning_rate": 1e-06,
      "loss": 1.1429,
      "num_input_tokens_seen": 33800708,
      "step": 604
    },
    {
      "epoch": 1.3452115812917596,
      "loss": 1.146965742111206,
      "loss_ce": 0.0012137566227465868,
      "loss_iou": 0.453125,
      "loss_num": 0.047607421875,
      "loss_xval": 1.1484375,
      "num_input_tokens_seen": 33800708,
      "step": 604
    },
    {
      "epoch": 1.3474387527839644,
      "grad_norm": 27.303401947021484,
      "learning_rate": 1e-06,
      "loss": 1.1817,
      "num_input_tokens_seen": 33853544,
      "step": 605
    },
    {
      "epoch": 1.3474387527839644,
      "loss": 1.0051369667053223,
      "loss_ce": 0.006113563664257526,
      "loss_iou": 0.390625,
      "loss_num": 0.04345703125,
      "loss_xval": 1.0,
      "num_input_tokens_seen": 33853544,
      "step": 605
    },
    {
      "epoch": 1.3496659242761693,
      "grad_norm": 24.224193572998047,
      "learning_rate": 1e-06,
      "loss": 1.0283,
      "num_input_tokens_seen": 33908172,
      "step": 606
    },
    {
      "epoch": 1.3496659242761693,
      "loss": 1.0841628313064575,
      "loss_ce": 0.0006667570560239255,
      "loss_iou": 0.41015625,
      "loss_num": 0.05224609375,
      "loss_xval": 1.0859375,
      "num_input_tokens_seen": 33908172,
      "step": 606
    },
    {
      "epoch": 1.3518930957683741,
      "grad_norm": 23.933223724365234,
      "learning_rate": 1e-06,
      "loss": 1.3531,
      "num_input_tokens_seen": 33964128,
      "step": 607
    },
    {
      "epoch": 1.3518930957683741,
      "loss": 1.0083250999450684,
      "loss_ce": 0.0005124981980770826,
      "loss_iou": 0.42578125,
      "loss_num": 0.03125,
      "loss_xval": 1.0078125,
      "num_input_tokens_seen": 33964128,
      "step": 607
    },
    {
      "epoch": 1.354120267260579,
      "grad_norm": 49.8543815612793,
      "learning_rate": 1e-06,
      "loss": 0.9476,
      "num_input_tokens_seen": 34018964,
      "step": 608
    },
    {
      "epoch": 1.354120267260579,
      "loss": 1.1164300441741943,
      "loss_ce": 0.0011956640519201756,
      "loss_iou": 0.447265625,
      "loss_num": 0.043701171875,
      "loss_xval": 1.1171875,
      "num_input_tokens_seen": 34018964,
      "step": 608
    },
    {
      "epoch": 1.3563474387527839,
      "grad_norm": 20.391830444335938,
      "learning_rate": 1e-06,
      "loss": 1.0838,
      "num_input_tokens_seen": 34074960,
      "step": 609
    },
    {
      "epoch": 1.3563474387527839,
      "loss": 1.0016207695007324,
      "loss_ce": 0.0008883203845471144,
      "loss_iou": 0.390625,
      "loss_num": 0.044189453125,
      "loss_xval": 1.0,
      "num_input_tokens_seen": 34074960,
      "step": 609
    },
    {
      "epoch": 1.358574610244989,
      "grad_norm": 21.543336868286133,
      "learning_rate": 1e-06,
      "loss": 1.0178,
      "num_input_tokens_seen": 34131548,
      "step": 610
    },
    {
      "epoch": 1.358574610244989,
      "loss": 0.7443816661834717,
      "loss_ce": 0.0036590369418263435,
      "loss_iou": 0.2890625,
      "loss_num": 0.03271484375,
      "loss_xval": 0.7421875,
      "num_input_tokens_seen": 34131548,
      "step": 610
    },
    {
      "epoch": 1.3608017817371938,
      "grad_norm": 205.24061584472656,
      "learning_rate": 1e-06,
      "loss": 1.1072,
      "num_input_tokens_seen": 34187164,
      "step": 611
    },
    {
      "epoch": 1.3608017817371938,
      "loss": 1.3798058032989502,
      "loss_ce": 0.0013879577163606882,
      "loss_iou": 0.5625,
      "loss_num": 0.05078125,
      "loss_xval": 1.375,
      "num_input_tokens_seen": 34187164,
      "step": 611
    },
    {
      "epoch": 1.3630289532293987,
      "grad_norm": 50.390380859375,
      "learning_rate": 1e-06,
      "loss": 1.4146,
      "num_input_tokens_seen": 34240892,
      "step": 612
    },
    {
      "epoch": 1.3630289532293987,
      "loss": 1.688087821006775,
      "loss_ce": 0.00644719647243619,
      "loss_iou": 0.6328125,
      "loss_num": 0.08251953125,
      "loss_xval": 1.6796875,
      "num_input_tokens_seen": 34240892,
      "step": 612
    },
    {
      "epoch": 1.3652561247216035,
      "grad_norm": 22.16488265991211,
      "learning_rate": 1e-06,
      "loss": 1.0745,
      "num_input_tokens_seen": 34296472,
      "step": 613
    },
    {
      "epoch": 1.3652561247216035,
      "loss": 1.2694756984710693,
      "loss_ce": 0.0006769584724679589,
      "loss_iou": 0.5078125,
      "loss_num": 0.050537109375,
      "loss_xval": 1.265625,
      "num_input_tokens_seen": 34296472,
      "step": 613
    },
    {
      "epoch": 1.3674832962138086,
      "grad_norm": 44.119998931884766,
      "learning_rate": 1e-06,
      "loss": 1.1704,
      "num_input_tokens_seen": 34352164,
      "step": 614
    },
    {
      "epoch": 1.3674832962138086,
      "loss": 1.0769069194793701,
      "loss_ce": 0.0012233321322128177,
      "loss_iou": 0.431640625,
      "loss_num": 0.04248046875,
      "loss_xval": 1.078125,
      "num_input_tokens_seen": 34352164,
      "step": 614
    },
    {
      "epoch": 1.3697104677060135,
      "grad_norm": 38.12681198120117,
      "learning_rate": 1e-06,
      "loss": 1.2304,
      "num_input_tokens_seen": 34408956,
      "step": 615
    },
    {
      "epoch": 1.3697104677060135,
      "loss": 1.0864746570587158,
      "loss_ce": 0.0005371640436351299,
      "loss_iou": 0.447265625,
      "loss_num": 0.038330078125,
      "loss_xval": 1.0859375,
      "num_input_tokens_seen": 34408956,
      "step": 615
    },
    {
      "epoch": 1.3719376391982183,
      "grad_norm": 22.228532791137695,
      "learning_rate": 1e-06,
      "loss": 1.0929,
      "num_input_tokens_seen": 34462868,
      "step": 616
    },
    {
      "epoch": 1.3719376391982183,
      "loss": 1.1881909370422363,
      "loss_ce": 0.0011792225996032357,
      "loss_iou": 0.458984375,
      "loss_num": 0.053955078125,
      "loss_xval": 1.1875,
      "num_input_tokens_seen": 34462868,
      "step": 616
    },
    {
      "epoch": 1.3741648106904232,
      "grad_norm": 16.62993812561035,
      "learning_rate": 1e-06,
      "loss": 0.8739,
      "num_input_tokens_seen": 34519148,
      "step": 617
    },
    {
      "epoch": 1.3741648106904232,
      "loss": 0.939282238483429,
      "loss_ce": 0.0008056251681409776,
      "loss_iou": 0.3828125,
      "loss_num": 0.0341796875,
      "loss_xval": 0.9375,
      "num_input_tokens_seen": 34519148,
      "step": 617
    },
    {
      "epoch": 1.376391982182628,
      "grad_norm": 26.92157745361328,
      "learning_rate": 1e-06,
      "loss": 1.16,
      "num_input_tokens_seen": 34576480,
      "step": 618
    },
    {
      "epoch": 1.376391982182628,
      "loss": 1.1556055545806885,
      "loss_ce": 0.002041085623204708,
      "loss_iou": 0.494140625,
      "loss_num": 0.033447265625,
      "loss_xval": 1.15625,
      "num_input_tokens_seen": 34576480,
      "step": 618
    },
    {
      "epoch": 1.378619153674833,
      "grad_norm": 78.8863525390625,
      "learning_rate": 1e-06,
      "loss": 0.8745,
      "num_input_tokens_seen": 34631844,
      "step": 619
    },
    {
      "epoch": 1.378619153674833,
      "loss": 0.6508793234825134,
      "loss_ce": 0.0004887055838480592,
      "loss_iou": 0.263671875,
      "loss_num": 0.0242919921875,
      "loss_xval": 0.6484375,
      "num_input_tokens_seen": 34631844,
      "step": 619
    },
    {
      "epoch": 1.3808463251670378,
      "grad_norm": 18.949325561523438,
      "learning_rate": 1e-06,
      "loss": 0.8934,
      "num_input_tokens_seen": 34687740,
      "step": 620
    },
    {
      "epoch": 1.3808463251670378,
      "loss": 0.8919047117233276,
      "loss_ce": 0.000791468657553196,
      "loss_iou": 0.373046875,
      "loss_num": 0.029052734375,
      "loss_xval": 0.890625,
      "num_input_tokens_seen": 34687740,
      "step": 620
    },
    {
      "epoch": 1.3830734966592428,
      "grad_norm": 21.888832092285156,
      "learning_rate": 1e-06,
      "loss": 1.0868,
      "num_input_tokens_seen": 34744896,
      "step": 621
    },
    {
      "epoch": 1.3830734966592428,
      "loss": 1.120755910873413,
      "loss_ce": 0.0006387863541021943,
      "loss_iou": 0.486328125,
      "loss_num": 0.0291748046875,
      "loss_xval": 1.1171875,
      "num_input_tokens_seen": 34744896,
      "step": 621
    },
    {
      "epoch": 1.3853006681514477,
      "grad_norm": 62.403011322021484,
      "learning_rate": 1e-06,
      "loss": 0.8124,
      "num_input_tokens_seen": 34800684,
      "step": 622
    },
    {
      "epoch": 1.3853006681514477,
      "loss": 0.9201839566230774,
      "loss_ce": 0.0012385983718559146,
      "loss_iou": 0.3828125,
      "loss_num": 0.031005859375,
      "loss_xval": 0.91796875,
      "num_input_tokens_seen": 34800684,
      "step": 622
    },
    {
      "epoch": 1.3875278396436526,
      "grad_norm": 28.83528709411621,
      "learning_rate": 1e-06,
      "loss": 1.2444,
      "num_input_tokens_seen": 34856832,
      "step": 623
    },
    {
      "epoch": 1.3875278396436526,
      "loss": 1.0172494649887085,
      "loss_ce": 0.00113617442548275,
      "loss_iou": 0.41796875,
      "loss_num": 0.0361328125,
      "loss_xval": 1.015625,
      "num_input_tokens_seen": 34856832,
      "step": 623
    },
    {
      "epoch": 1.3897550111358574,
      "grad_norm": 62.715023040771484,
      "learning_rate": 1e-06,
      "loss": 1.3,
      "num_input_tokens_seen": 34914236,
      "step": 624
    },
    {
      "epoch": 1.3897550111358574,
      "loss": 1.4728524684906006,
      "loss_ce": 0.002149291103705764,
      "loss_iou": 0.5859375,
      "loss_num": 0.060302734375,
      "loss_xval": 1.46875,
      "num_input_tokens_seen": 34914236,
      "step": 624
    },
    {
      "epoch": 1.3919821826280623,
      "grad_norm": 22.02700424194336,
      "learning_rate": 1e-06,
      "loss": 1.218,
      "num_input_tokens_seen": 34970880,
      "step": 625
    },
    {
      "epoch": 1.3919821826280623,
      "loss": 1.4092342853546143,
      "loss_ce": 0.0010311320656910539,
      "loss_iou": 0.59375,
      "loss_num": 0.044189453125,
      "loss_xval": 1.40625,
      "num_input_tokens_seen": 34970880,
      "step": 625
    },
    {
      "epoch": 1.3942093541202674,
      "grad_norm": 82.06532287597656,
      "learning_rate": 1e-06,
      "loss": 1.2257,
      "num_input_tokens_seen": 35026436,
      "step": 626
    },
    {
      "epoch": 1.3942093541202674,
      "loss": 1.3661468029022217,
      "loss_ce": 0.0018889284692704678,
      "loss_iou": 0.4921875,
      "loss_num": 0.0751953125,
      "loss_xval": 1.3671875,
      "num_input_tokens_seen": 35026436,
      "step": 626
    },
    {
      "epoch": 1.3964365256124722,
      "grad_norm": 21.959980010986328,
      "learning_rate": 1e-06,
      "loss": 1.0485,
      "num_input_tokens_seen": 35083640,
      "step": 627
    },
    {
      "epoch": 1.3964365256124722,
      "loss": 1.2543073892593384,
      "loss_ce": 0.0008894825004972517,
      "loss_iou": 0.47265625,
      "loss_num": 0.061767578125,
      "loss_xval": 1.25,
      "num_input_tokens_seen": 35083640,
      "step": 627
    },
    {
      "epoch": 1.398663697104677,
      "grad_norm": 69.06570434570312,
      "learning_rate": 1e-06,
      "loss": 1.1878,
      "num_input_tokens_seen": 35139632,
      "step": 628
    },
    {
      "epoch": 1.398663697104677,
      "loss": 1.040403127670288,
      "loss_ce": 0.003171587362885475,
      "loss_iou": 0.4375,
      "loss_num": 0.032958984375,
      "loss_xval": 1.0390625,
      "num_input_tokens_seen": 35139632,
      "step": 628
    },
    {
      "epoch": 1.400890868596882,
      "grad_norm": 32.16525650024414,
      "learning_rate": 1e-06,
      "loss": 1.4466,
      "num_input_tokens_seen": 35194568,
      "step": 629
    },
    {
      "epoch": 1.400890868596882,
      "loss": 1.3845007419586182,
      "loss_ce": 0.006571032106876373,
      "loss_iou": 0.474609375,
      "loss_num": 0.0849609375,
      "loss_xval": 1.375,
      "num_input_tokens_seen": 35194568,
      "step": 629
    },
    {
      "epoch": 1.4031180400890868,
      "grad_norm": 19.882808685302734,
      "learning_rate": 1e-06,
      "loss": 0.7151,
      "num_input_tokens_seen": 35252680,
      "step": 630
    },
    {
      "epoch": 1.4031180400890868,
      "loss": 0.6794017553329468,
      "loss_ce": 0.00044663704466074705,
      "loss_iou": 0.251953125,
      "loss_num": 0.034912109375,
      "loss_xval": 0.6796875,
      "num_input_tokens_seen": 35252680,
      "step": 630
    },
    {
      "epoch": 1.4053452115812917,
      "grad_norm": 23.41751480102539,
      "learning_rate": 1e-06,
      "loss": 0.8731,
      "num_input_tokens_seen": 35308480,
      "step": 631
    },
    {
      "epoch": 1.4053452115812917,
      "loss": 0.9022737741470337,
      "loss_ce": 0.0009065663907676935,
      "loss_iou": 0.376953125,
      "loss_num": 0.02978515625,
      "loss_xval": 0.90234375,
      "num_input_tokens_seen": 35308480,
      "step": 631
    },
    {
      "epoch": 1.4075723830734965,
      "grad_norm": 23.227956771850586,
      "learning_rate": 1e-06,
      "loss": 1.2639,
      "num_input_tokens_seen": 35364808,
      "step": 632
    },
    {
      "epoch": 1.4075723830734965,
      "loss": 1.2337548732757568,
      "loss_ce": 0.0023096189834177494,
      "loss_iou": 0.5,
      "loss_num": 0.045654296875,
      "loss_xval": 1.234375,
      "num_input_tokens_seen": 35364808,
      "step": 632
    },
    {
      "epoch": 1.4097995545657016,
      "grad_norm": 55.23591232299805,
      "learning_rate": 1e-06,
      "loss": 1.1625,
      "num_input_tokens_seen": 35420704,
      "step": 633
    },
    {
      "epoch": 1.4097995545657016,
      "loss": 0.9497365951538086,
      "loss_ce": 0.0005178386345505714,
      "loss_iou": 0.365234375,
      "loss_num": 0.043701171875,
      "loss_xval": 0.94921875,
      "num_input_tokens_seen": 35420704,
      "step": 633
    },
    {
      "epoch": 1.4120267260579065,
      "grad_norm": 26.891746520996094,
      "learning_rate": 1e-06,
      "loss": 1.1738,
      "num_input_tokens_seen": 35473184,
      "step": 634
    },
    {
      "epoch": 1.4120267260579065,
      "loss": 1.1008825302124023,
      "loss_ce": 0.0012731605675071478,
      "loss_iou": 0.453125,
      "loss_num": 0.03857421875,
      "loss_xval": 1.1015625,
      "num_input_tokens_seen": 35473184,
      "step": 634
    },
    {
      "epoch": 1.4142538975501113,
      "grad_norm": 17.678611755371094,
      "learning_rate": 1e-06,
      "loss": 1.1334,
      "num_input_tokens_seen": 35530148,
      "step": 635
    },
    {
      "epoch": 1.4142538975501113,
      "loss": 1.0327593088150024,
      "loss_ce": 0.0005327487015165389,
      "loss_iou": 0.443359375,
      "loss_num": 0.0294189453125,
      "loss_xval": 1.03125,
      "num_input_tokens_seen": 35530148,
      "step": 635
    },
    {
      "epoch": 1.4164810690423162,
      "grad_norm": 313.0267028808594,
      "learning_rate": 1e-06,
      "loss": 1.3829,
      "num_input_tokens_seen": 35583856,
      "step": 636
    },
    {
      "epoch": 1.4164810690423162,
      "loss": 1.357082486152649,
      "loss_ce": 0.0023461126256734133,
      "loss_iou": 0.5625,
      "loss_num": 0.046142578125,
      "loss_xval": 1.3515625,
      "num_input_tokens_seen": 35583856,
      "step": 636
    },
    {
      "epoch": 1.4187082405345213,
      "grad_norm": 20.430328369140625,
      "learning_rate": 1e-06,
      "loss": 1.2449,
      "num_input_tokens_seen": 35640836,
      "step": 637
    },
    {
      "epoch": 1.4187082405345213,
      "loss": 1.4005608558654785,
      "loss_ce": 0.0006584060029126704,
      "loss_iou": 0.56640625,
      "loss_num": 0.052978515625,
      "loss_xval": 1.3984375,
      "num_input_tokens_seen": 35640836,
      "step": 637
    },
    {
      "epoch": 1.4209354120267261,
      "grad_norm": 23.457950592041016,
      "learning_rate": 1e-06,
      "loss": 0.8546,
      "num_input_tokens_seen": 35695288,
      "step": 638
    },
    {
      "epoch": 1.4209354120267261,
      "loss": 0.8239110112190247,
      "loss_ce": 0.0017674551345407963,
      "loss_iou": 0.34375,
      "loss_num": 0.02685546875,
      "loss_xval": 0.8203125,
      "num_input_tokens_seen": 35695288,
      "step": 638
    },
    {
      "epoch": 1.423162583518931,
      "grad_norm": 22.952999114990234,
      "learning_rate": 1e-06,
      "loss": 0.8464,
      "num_input_tokens_seen": 35751468,
      "step": 639
    },
    {
      "epoch": 1.423162583518931,
      "loss": 1.025329351425171,
      "loss_ce": 0.0009152949205599725,
      "loss_iou": 0.423828125,
      "loss_num": 0.03515625,
      "loss_xval": 1.0234375,
      "num_input_tokens_seen": 35751468,
      "step": 639
    },
    {
      "epoch": 1.4253897550111359,
      "grad_norm": 17.500404357910156,
      "learning_rate": 1e-06,
      "loss": 0.8751,
      "num_input_tokens_seen": 35809384,
      "step": 640
    },
    {
      "epoch": 1.4253897550111359,
      "loss": 0.8379029035568237,
      "loss_ce": 0.0005006167921237648,
      "loss_iou": 0.34765625,
      "loss_num": 0.0286865234375,
      "loss_xval": 0.8359375,
      "num_input_tokens_seen": 35809384,
      "step": 640
    },
    {
      "epoch": 1.4276169265033407,
      "grad_norm": 30.681381225585938,
      "learning_rate": 1e-06,
      "loss": 1.1528,
      "num_input_tokens_seen": 35865544,
      "step": 641
    },
    {
      "epoch": 1.4276169265033407,
      "loss": 1.2849559783935547,
      "loss_ce": 0.0005320889176800847,
      "loss_iou": 0.55078125,
      "loss_num": 0.037353515625,
      "loss_xval": 1.28125,
      "num_input_tokens_seen": 35865544,
      "step": 641
    },
    {
      "epoch": 1.4298440979955456,
      "grad_norm": 19.769329071044922,
      "learning_rate": 1e-06,
      "loss": 1.2354,
      "num_input_tokens_seen": 35921472,
      "step": 642
    },
    {
      "epoch": 1.4298440979955456,
      "loss": 1.344813346862793,
      "loss_ce": 0.0008191849919967353,
      "loss_iou": 0.5703125,
      "loss_num": 0.041259765625,
      "loss_xval": 1.34375,
      "num_input_tokens_seen": 35921472,
      "step": 642
    },
    {
      "epoch": 1.4320712694877504,
      "grad_norm": 166.1317901611328,
      "learning_rate": 1e-06,
      "loss": 1.3808,
      "num_input_tokens_seen": 35978028,
      "step": 643
    },
    {
      "epoch": 1.4320712694877504,
      "loss": 1.7419133186340332,
      "loss_ce": 0.002655471907928586,
      "loss_iou": 0.68359375,
      "loss_num": 0.0751953125,
      "loss_xval": 1.7421875,
      "num_input_tokens_seen": 35978028,
      "step": 643
    },
    {
      "epoch": 1.4342984409799555,
      "grad_norm": 21.949459075927734,
      "learning_rate": 1e-06,
      "loss": 0.9716,
      "num_input_tokens_seen": 36033152,
      "step": 644
    },
    {
      "epoch": 1.4342984409799555,
      "loss": 0.8689587116241455,
      "loss_ce": 0.0007946894620545208,
      "loss_iou": 0.376953125,
      "loss_num": 0.022705078125,
      "loss_xval": 0.8671875,
      "num_input_tokens_seen": 36033152,
      "step": 644
    },
    {
      "epoch": 1.4365256124721604,
      "grad_norm": 53.461631774902344,
      "learning_rate": 1e-06,
      "loss": 1.3388,
      "num_input_tokens_seen": 36088644,
      "step": 645
    },
    {
      "epoch": 1.4365256124721604,
      "loss": 1.3191097974777222,
      "loss_ce": 0.0017270214157178998,
      "loss_iou": 0.490234375,
      "loss_num": 0.0673828125,
      "loss_xval": 1.3203125,
      "num_input_tokens_seen": 36088644,
      "step": 645
    },
    {
      "epoch": 1.4387527839643652,
      "grad_norm": 17.125713348388672,
      "learning_rate": 1e-06,
      "loss": 1.1615,
      "num_input_tokens_seen": 36147156,
      "step": 646
    },
    {
      "epoch": 1.4387527839643652,
      "loss": 1.204886555671692,
      "loss_ce": 0.0012732010800391436,
      "loss_iou": 0.46875,
      "loss_num": 0.052978515625,
      "loss_xval": 1.203125,
      "num_input_tokens_seen": 36147156,
      "step": 646
    },
    {
      "epoch": 1.44097995545657,
      "grad_norm": 20.655792236328125,
      "learning_rate": 1e-06,
      "loss": 0.8804,
      "num_input_tokens_seen": 36205528,
      "step": 647
    },
    {
      "epoch": 1.44097995545657,
      "loss": 0.9184082746505737,
      "loss_ce": 0.0004395167634356767,
      "loss_iou": 0.36328125,
      "loss_num": 0.037841796875,
      "loss_xval": 0.91796875,
      "num_input_tokens_seen": 36205528,
      "step": 647
    },
    {
      "epoch": 1.4432071269487752,
      "grad_norm": 27.488887786865234,
      "learning_rate": 1e-06,
      "loss": 1.0496,
      "num_input_tokens_seen": 36263608,
      "step": 648
    },
    {
      "epoch": 1.4432071269487752,
      "loss": 1.052678108215332,
      "loss_ce": 0.000431923137512058,
      "loss_iou": 0.423828125,
      "loss_num": 0.041259765625,
      "loss_xval": 1.0546875,
      "num_input_tokens_seen": 36263608,
      "step": 648
    },
    {
      "epoch": 1.44543429844098,
      "grad_norm": 19.03914451599121,
      "learning_rate": 1e-06,
      "loss": 1.0058,
      "num_input_tokens_seen": 36320740,
      "step": 649
    },
    {
      "epoch": 1.44543429844098,
      "loss": 1.1877658367156982,
      "loss_ce": 0.0014864858239889145,
      "loss_iou": 0.46484375,
      "loss_num": 0.05126953125,
      "loss_xval": 1.1875,
      "num_input_tokens_seen": 36320740,
      "step": 649
    },
    {
      "epoch": 1.447661469933185,
      "grad_norm": 30.350322723388672,
      "learning_rate": 1e-06,
      "loss": 1.1674,
      "num_input_tokens_seen": 36375948,
      "step": 650
    },
    {
      "epoch": 1.447661469933185,
      "loss": 1.6142632961273193,
      "loss_ce": 0.0014703237684443593,
      "loss_iou": 0.6484375,
      "loss_num": 0.0625,
      "loss_xval": 1.609375,
      "num_input_tokens_seen": 36375948,
      "step": 650
    },
    {
      "epoch": 1.4498886414253898,
      "grad_norm": 31.322311401367188,
      "learning_rate": 1e-06,
      "loss": 1.0709,
      "num_input_tokens_seen": 36433352,
      "step": 651
    },
    {
      "epoch": 1.4498886414253898,
      "loss": 1.1504631042480469,
      "loss_ce": 0.0005606971681118011,
      "loss_iou": 0.482421875,
      "loss_num": 0.037109375,
      "loss_xval": 1.1484375,
      "num_input_tokens_seen": 36433352,
      "step": 651
    },
    {
      "epoch": 1.4521158129175946,
      "grad_norm": 58.9162712097168,
      "learning_rate": 1e-06,
      "loss": 1.1301,
      "num_input_tokens_seen": 36489816,
      "step": 652
    },
    {
      "epoch": 1.4521158129175946,
      "loss": 1.2538143396377563,
      "loss_ce": 0.001372913713566959,
      "loss_iou": 0.494140625,
      "loss_num": 0.052490234375,
      "loss_xval": 1.25,
      "num_input_tokens_seen": 36489816,
      "step": 652
    },
    {
      "epoch": 1.4543429844097995,
      "grad_norm": 17.665956497192383,
      "learning_rate": 1e-06,
      "loss": 1.1987,
      "num_input_tokens_seen": 36542532,
      "step": 653
    },
    {
      "epoch": 1.4543429844097995,
      "loss": 1.2465870380401611,
      "loss_ce": 0.0004933135933242738,
      "loss_iou": 0.50390625,
      "loss_num": 0.047607421875,
      "loss_xval": 1.25,
      "num_input_tokens_seen": 36542532,
      "step": 653
    },
    {
      "epoch": 1.4565701559020043,
      "grad_norm": 63.455204010009766,
      "learning_rate": 1e-06,
      "loss": 1.0736,
      "num_input_tokens_seen": 36599520,
      "step": 654
    },
    {
      "epoch": 1.4565701559020043,
      "loss": 1.2124230861663818,
      "loss_ce": 0.0005090509075671434,
      "loss_iou": 0.4765625,
      "loss_num": 0.0517578125,
      "loss_xval": 1.2109375,
      "num_input_tokens_seen": 36599520,
      "step": 654
    },
    {
      "epoch": 1.4587973273942094,
      "grad_norm": 26.223678588867188,
      "learning_rate": 1e-06,
      "loss": 1.0227,
      "num_input_tokens_seen": 36656080,
      "step": 655
    },
    {
      "epoch": 1.4587973273942094,
      "loss": 0.9443738460540771,
      "loss_ce": 0.000526183401234448,
      "loss_iou": 0.353515625,
      "loss_num": 0.047119140625,
      "loss_xval": 0.9453125,
      "num_input_tokens_seen": 36656080,
      "step": 655
    },
    {
      "epoch": 1.4610244988864143,
      "grad_norm": 41.6849365234375,
      "learning_rate": 1e-06,
      "loss": 1.0155,
      "num_input_tokens_seen": 36712280,
      "step": 656
    },
    {
      "epoch": 1.4610244988864143,
      "loss": 1.218764066696167,
      "loss_ce": 0.008314890787005424,
      "loss_iou": 0.47265625,
      "loss_num": 0.052978515625,
      "loss_xval": 1.2109375,
      "num_input_tokens_seen": 36712280,
      "step": 656
    },
    {
      "epoch": 1.4632516703786191,
      "grad_norm": 26.14752197265625,
      "learning_rate": 1e-06,
      "loss": 1.06,
      "num_input_tokens_seen": 36769340,
      "step": 657
    },
    {
      "epoch": 1.4632516703786191,
      "loss": 1.0224517583847046,
      "loss_ce": 0.0007232209318317473,
      "loss_iou": 0.39453125,
      "loss_num": 0.04638671875,
      "loss_xval": 1.0234375,
      "num_input_tokens_seen": 36769340,
      "step": 657
    },
    {
      "epoch": 1.465478841870824,
      "grad_norm": 25.274215698242188,
      "learning_rate": 1e-06,
      "loss": 1.2905,
      "num_input_tokens_seen": 36824936,
      "step": 658
    },
    {
      "epoch": 1.465478841870824,
      "loss": 1.3570466041564941,
      "loss_ce": 0.0010896207531914115,
      "loss_iou": 0.57421875,
      "loss_num": 0.041748046875,
      "loss_xval": 1.359375,
      "num_input_tokens_seen": 36824936,
      "step": 658
    },
    {
      "epoch": 1.467706013363029,
      "grad_norm": 18.70472526550293,
      "learning_rate": 1e-06,
      "loss": 0.9063,
      "num_input_tokens_seen": 36883996,
      "step": 659
    },
    {
      "epoch": 1.467706013363029,
      "loss": 0.9597538709640503,
      "loss_ce": 0.005408107303082943,
      "loss_iou": 0.376953125,
      "loss_num": 0.0400390625,
      "loss_xval": 0.953125,
      "num_input_tokens_seen": 36883996,
      "step": 659
    },
    {
      "epoch": 1.469933184855234,
      "grad_norm": 24.98056411743164,
      "learning_rate": 1e-06,
      "loss": 0.9572,
      "num_input_tokens_seen": 36936316,
      "step": 660
    },
    {
      "epoch": 1.469933184855234,
      "loss": 0.7073796391487122,
      "loss_ce": 0.006451886147260666,
      "loss_iou": 0.283203125,
      "loss_num": 0.02685546875,
      "loss_xval": 0.69921875,
      "num_input_tokens_seen": 36936316,
      "step": 660
    },
    {
      "epoch": 1.4721603563474388,
      "grad_norm": 22.775863647460938,
      "learning_rate": 1e-06,
      "loss": 1.0446,
      "num_input_tokens_seen": 36991328,
      "step": 661
    },
    {
      "epoch": 1.4721603563474388,
      "loss": 1.084336757659912,
      "loss_ce": 0.0008406001143157482,
      "loss_iou": 0.4296875,
      "loss_num": 0.044921875,
      "loss_xval": 1.0859375,
      "num_input_tokens_seen": 36991328,
      "step": 661
    },
    {
      "epoch": 1.4743875278396437,
      "grad_norm": 25.190065383911133,
      "learning_rate": 1e-06,
      "loss": 1.2837,
      "num_input_tokens_seen": 37044136,
      "step": 662
    },
    {
      "epoch": 1.4743875278396437,
      "loss": 1.5641846656799316,
      "loss_ce": 0.0026612321380525827,
      "loss_iou": 0.609375,
      "loss_num": 0.06787109375,
      "loss_xval": 1.5625,
      "num_input_tokens_seen": 37044136,
      "step": 662
    },
    {
      "epoch": 1.4766146993318485,
      "grad_norm": 28.881362915039062,
      "learning_rate": 1e-06,
      "loss": 0.8291,
      "num_input_tokens_seen": 37100336,
      "step": 663
    },
    {
      "epoch": 1.4766146993318485,
      "loss": 0.8968852758407593,
      "loss_ce": 0.003330609295517206,
      "loss_iou": 0.35546875,
      "loss_num": 0.036376953125,
      "loss_xval": 0.89453125,
      "num_input_tokens_seen": 37100336,
      "step": 663
    },
    {
      "epoch": 1.4788418708240534,
      "grad_norm": 24.36873435974121,
      "learning_rate": 1e-06,
      "loss": 1.2756,
      "num_input_tokens_seen": 37154992,
      "step": 664
    },
    {
      "epoch": 1.4788418708240534,
      "loss": 1.3726325035095215,
      "loss_ce": 0.0030035879462957382,
      "loss_iou": 0.5546875,
      "loss_num": 0.052001953125,
      "loss_xval": 1.3671875,
      "num_input_tokens_seen": 37154992,
      "step": 664
    },
    {
      "epoch": 1.4810690423162582,
      "grad_norm": 40.89757537841797,
      "learning_rate": 1e-06,
      "loss": 1.1327,
      "num_input_tokens_seen": 37208516,
      "step": 665
    },
    {
      "epoch": 1.4810690423162582,
      "loss": 0.8330258131027222,
      "loss_ce": 0.0005062957643531263,
      "loss_iou": 0.35546875,
      "loss_num": 0.024169921875,
      "loss_xval": 0.83203125,
      "num_input_tokens_seen": 37208516,
      "step": 665
    },
    {
      "epoch": 1.4832962138084633,
      "grad_norm": 33.48430252075195,
      "learning_rate": 1e-06,
      "loss": 1.0491,
      "num_input_tokens_seen": 37263476,
      "step": 666
    },
    {
      "epoch": 1.4832962138084633,
      "loss": 1.249821424484253,
      "loss_ce": 0.001530370325781405,
      "loss_iou": 0.55078125,
      "loss_num": 0.02978515625,
      "loss_xval": 1.25,
      "num_input_tokens_seen": 37263476,
      "step": 666
    },
    {
      "epoch": 1.4855233853006682,
      "grad_norm": 31.53962516784668,
      "learning_rate": 1e-06,
      "loss": 0.8733,
      "num_input_tokens_seen": 37319864,
      "step": 667
    },
    {
      "epoch": 1.4855233853006682,
      "loss": 0.6637530326843262,
      "loss_ce": 0.00213192543014884,
      "loss_iou": 0.27734375,
      "loss_num": 0.0218505859375,
      "loss_xval": 0.66015625,
      "num_input_tokens_seen": 37319864,
      "step": 667
    },
    {
      "epoch": 1.487750556792873,
      "grad_norm": 19.535804748535156,
      "learning_rate": 1e-06,
      "loss": 1.0933,
      "num_input_tokens_seen": 37378472,
      "step": 668
    },
    {
      "epoch": 1.487750556792873,
      "loss": 1.066902756690979,
      "loss_ce": 0.000496495165862143,
      "loss_iou": 0.451171875,
      "loss_num": 0.03271484375,
      "loss_xval": 1.0625,
      "num_input_tokens_seen": 37378472,
      "step": 668
    },
    {
      "epoch": 1.489977728285078,
      "grad_norm": 35.252281188964844,
      "learning_rate": 1e-06,
      "loss": 1.4544,
      "num_input_tokens_seen": 37433128,
      "step": 669
    },
    {
      "epoch": 1.489977728285078,
      "loss": 1.5701709985733032,
      "loss_ce": 0.0008350461139343679,
      "loss_iou": 0.640625,
      "loss_num": 0.05712890625,
      "loss_xval": 1.5703125,
      "num_input_tokens_seen": 37433128,
      "step": 669
    },
    {
      "epoch": 1.492204899777283,
      "grad_norm": 22.478748321533203,
      "learning_rate": 1e-06,
      "loss": 1.0887,
      "num_input_tokens_seen": 37487148,
      "step": 670
    },
    {
      "epoch": 1.492204899777283,
      "loss": 0.9821290373802185,
      "loss_ce": 0.0026368550024926662,
      "loss_iou": 0.333984375,
      "loss_num": 0.06201171875,
      "loss_xval": 0.98046875,
      "num_input_tokens_seen": 37487148,
      "step": 670
    },
    {
      "epoch": 1.4944320712694878,
      "grad_norm": 26.4521484375,
      "learning_rate": 1e-06,
      "loss": 1.0069,
      "num_input_tokens_seen": 37545644,
      "step": 671
    },
    {
      "epoch": 1.4944320712694878,
      "loss": 1.1284525394439697,
      "loss_ce": 0.0176127590239048,
      "loss_iou": 0.423828125,
      "loss_num": 0.052734375,
      "loss_xval": 1.109375,
      "num_input_tokens_seen": 37545644,
      "step": 671
    },
    {
      "epoch": 1.4966592427616927,
      "grad_norm": 20.00921058654785,
      "learning_rate": 1e-06,
      "loss": 1.1729,
      "num_input_tokens_seen": 37602744,
      "step": 672
    },
    {
      "epoch": 1.4966592427616927,
      "loss": 1.4309265613555908,
      "loss_ce": 0.0017273698467761278,
      "loss_iou": 0.57421875,
      "loss_num": 0.05615234375,
      "loss_xval": 1.4296875,
      "num_input_tokens_seen": 37602744,
      "step": 672
    },
    {
      "epoch": 1.4988864142538976,
      "grad_norm": 21.240440368652344,
      "learning_rate": 1e-06,
      "loss": 0.9773,
      "num_input_tokens_seen": 37658080,
      "step": 673
    },
    {
      "epoch": 1.4988864142538976,
      "loss": 1.0557523965835571,
      "loss_ce": 0.0005766096874140203,
      "loss_iou": 0.42578125,
      "loss_num": 0.040771484375,
      "loss_xval": 1.0546875,
      "num_input_tokens_seen": 37658080,
      "step": 673
    },
    {
      "epoch": 1.5011135857461024,
      "grad_norm": 20.119047164916992,
      "learning_rate": 1e-06,
      "loss": 0.8459,
      "num_input_tokens_seen": 37713888,
      "step": 674
    },
    {
      "epoch": 1.5011135857461024,
      "loss": 0.8226086497306824,
      "loss_ce": 0.0013196287909522653,
      "loss_iou": 0.34375,
      "loss_num": 0.0263671875,
      "loss_xval": 0.8203125,
      "num_input_tokens_seen": 37713888,
      "step": 674
    },
    {
      "epoch": 1.5033407572383073,
      "grad_norm": 29.80697250366211,
      "learning_rate": 1e-06,
      "loss": 0.9022,
      "num_input_tokens_seen": 37768756,
      "step": 675
    },
    {
      "epoch": 1.5033407572383073,
      "loss": 0.9731925129890442,
      "loss_ce": 0.001024498138576746,
      "loss_iou": 0.423828125,
      "loss_num": 0.0247802734375,
      "loss_xval": 0.97265625,
      "num_input_tokens_seen": 37768756,
      "step": 675
    },
    {
      "epoch": 1.5055679287305122,
      "grad_norm": 17.169862747192383,
      "learning_rate": 1e-06,
      "loss": 1.2668,
      "num_input_tokens_seen": 37821224,
      "step": 676
    },
    {
      "epoch": 1.5055679287305122,
      "loss": 1.339667558670044,
      "loss_ce": 0.0008004190749488771,
      "loss_iou": 0.5390625,
      "loss_num": 0.052490234375,
      "loss_xval": 1.3359375,
      "num_input_tokens_seen": 37821224,
      "step": 676
    },
    {
      "epoch": 1.507795100222717,
      "grad_norm": 48.648136138916016,
      "learning_rate": 1e-06,
      "loss": 1.1283,
      "num_input_tokens_seen": 37876864,
      "step": 677
    },
    {
      "epoch": 1.507795100222717,
      "loss": 1.0091381072998047,
      "loss_ce": 0.00047117803478613496,
      "loss_iou": 0.412109375,
      "loss_num": 0.03662109375,
      "loss_xval": 1.0078125,
      "num_input_tokens_seen": 37876864,
      "step": 677
    },
    {
      "epoch": 1.510022271714922,
      "grad_norm": 31.37755012512207,
      "learning_rate": 1e-06,
      "loss": 1.0842,
      "num_input_tokens_seen": 37934036,
      "step": 678
    },
    {
      "epoch": 1.510022271714922,
      "loss": 0.8647042512893677,
      "loss_ce": 0.0004463824152480811,
      "loss_iou": 0.37109375,
      "loss_num": 0.0245361328125,
      "loss_xval": 0.86328125,
      "num_input_tokens_seen": 37934036,
      "step": 678
    },
    {
      "epoch": 1.512249443207127,
      "grad_norm": 75.93875885009766,
      "learning_rate": 1e-06,
      "loss": 1.3985,
      "num_input_tokens_seen": 37987492,
      "step": 679
    },
    {
      "epoch": 1.512249443207127,
      "loss": 1.21080482006073,
      "loss_ce": 0.0013321400620043278,
      "loss_iou": 0.5,
      "loss_num": 0.041015625,
      "loss_xval": 1.2109375,
      "num_input_tokens_seen": 37987492,
      "step": 679
    },
    {
      "epoch": 1.5144766146993318,
      "grad_norm": 28.709468841552734,
      "learning_rate": 1e-06,
      "loss": 1.0242,
      "num_input_tokens_seen": 38046268,
      "step": 680
    },
    {
      "epoch": 1.5144766146993318,
      "loss": 0.7978775501251221,
      "loss_ce": 0.0007584316190332174,
      "loss_iou": 0.3046875,
      "loss_num": 0.0380859375,
      "loss_xval": 0.796875,
      "num_input_tokens_seen": 38046268,
      "step": 680
    },
    {
      "epoch": 1.516703786191537,
      "grad_norm": 15.11223030090332,
      "learning_rate": 1e-06,
      "loss": 1.0458,
      "num_input_tokens_seen": 38096816,
      "step": 681
    },
    {
      "epoch": 1.516703786191537,
      "loss": 0.8654755353927612,
      "loss_ce": 0.0004853087302763015,
      "loss_iou": 0.3515625,
      "loss_num": 0.032470703125,
      "loss_xval": 0.86328125,
      "num_input_tokens_seen": 38096816,
      "step": 681
    },
    {
      "epoch": 1.5189309576837418,
      "grad_norm": 28.030420303344727,
      "learning_rate": 1e-06,
      "loss": 1.1786,
      "num_input_tokens_seen": 38152668,
      "step": 682
    },
    {
      "epoch": 1.5189309576837418,
      "loss": 0.9609849452972412,
      "loss_ce": 0.0010240338742733002,
      "loss_iou": 0.392578125,
      "loss_num": 0.034912109375,
      "loss_xval": 0.9609375,
      "num_input_tokens_seen": 38152668,
      "step": 682
    },
    {
      "epoch": 1.5211581291759466,
      "grad_norm": 18.825443267822266,
      "learning_rate": 1e-06,
      "loss": 1.1633,
      "num_input_tokens_seen": 38208028,
      "step": 683
    },
    {
      "epoch": 1.5211581291759466,
      "loss": 1.3014280796051025,
      "loss_ce": 0.0006468580104410648,
      "loss_iou": 0.5390625,
      "loss_num": 0.04541015625,
      "loss_xval": 1.296875,
      "num_input_tokens_seen": 38208028,
      "step": 683
    },
    {
      "epoch": 1.5233853006681515,
      "grad_norm": 33.09317398071289,
      "learning_rate": 1e-06,
      "loss": 1.0005,
      "num_input_tokens_seen": 38266628,
      "step": 684
    },
    {
      "epoch": 1.5233853006681515,
      "loss": 0.8423627614974976,
      "loss_ce": 0.0005658682784996927,
      "loss_iou": 0.326171875,
      "loss_num": 0.037841796875,
      "loss_xval": 0.84375,
      "num_input_tokens_seen": 38266628,
      "step": 684
    },
    {
      "epoch": 1.5256124721603563,
      "grad_norm": 21.19887924194336,
      "learning_rate": 1e-06,
      "loss": 1.2159,
      "num_input_tokens_seen": 38321396,
      "step": 685
    },
    {
      "epoch": 1.5256124721603563,
      "loss": 1.512146234512329,
      "loss_ce": 0.001403960632160306,
      "loss_iou": 0.6015625,
      "loss_num": 0.06201171875,
      "loss_xval": 1.5078125,
      "num_input_tokens_seen": 38321396,
      "step": 685
    },
    {
      "epoch": 1.5278396436525612,
      "grad_norm": 108.64661407470703,
      "learning_rate": 1e-06,
      "loss": 1.1361,
      "num_input_tokens_seen": 38376092,
      "step": 686
    },
    {
      "epoch": 1.5278396436525612,
      "loss": 1.0967187881469727,
      "loss_ce": 0.0010156568605452776,
      "loss_iou": 0.453125,
      "loss_num": 0.038330078125,
      "loss_xval": 1.09375,
      "num_input_tokens_seen": 38376092,
      "step": 686
    },
    {
      "epoch": 1.530066815144766,
      "grad_norm": 18.550582885742188,
      "learning_rate": 1e-06,
      "loss": 1.2406,
      "num_input_tokens_seen": 38432308,
      "step": 687
    },
    {
      "epoch": 1.530066815144766,
      "loss": 1.231999158859253,
      "loss_ce": 0.0005538459517993033,
      "loss_iou": 0.515625,
      "loss_num": 0.040283203125,
      "loss_xval": 1.234375,
      "num_input_tokens_seen": 38432308,
      "step": 687
    },
    {
      "epoch": 1.532293986636971,
      "grad_norm": 54.29788589477539,
      "learning_rate": 1e-06,
      "loss": 0.9656,
      "num_input_tokens_seen": 38488264,
      "step": 688
    },
    {
      "epoch": 1.532293986636971,
      "loss": 0.8684132099151611,
      "loss_ce": 0.0007374430424533784,
      "loss_iou": 0.318359375,
      "loss_num": 0.046142578125,
      "loss_xval": 0.8671875,
      "num_input_tokens_seen": 38488264,
      "step": 688
    },
    {
      "epoch": 1.534521158129176,
      "grad_norm": 26.847867965698242,
      "learning_rate": 1e-06,
      "loss": 1.318,
      "num_input_tokens_seen": 38544928,
      "step": 689
    },
    {
      "epoch": 1.534521158129176,
      "loss": 1.185499906539917,
      "loss_ce": 0.0019061192870140076,
      "loss_iou": 0.49609375,
      "loss_num": 0.03857421875,
      "loss_xval": 1.1875,
      "num_input_tokens_seen": 38544928,
      "step": 689
    },
    {
      "epoch": 1.5367483296213809,
      "grad_norm": 17.0640926361084,
      "learning_rate": 1e-06,
      "loss": 1.1527,
      "num_input_tokens_seen": 38600752,
      "step": 690
    },
    {
      "epoch": 1.5367483296213809,
      "loss": 1.185793399810791,
      "loss_ce": 0.0007348479703068733,
      "loss_iou": 0.484375,
      "loss_num": 0.04345703125,
      "loss_xval": 1.1875,
      "num_input_tokens_seen": 38600752,
      "step": 690
    },
    {
      "epoch": 1.5389755011135857,
      "grad_norm": 125.89933013916016,
      "learning_rate": 1e-06,
      "loss": 1.053,
      "num_input_tokens_seen": 38656252,
      "step": 691
    },
    {
      "epoch": 1.5389755011135857,
      "loss": 1.032954454421997,
      "loss_ce": 0.0009720420930534601,
      "loss_iou": 0.39453125,
      "loss_num": 0.048583984375,
      "loss_xval": 1.03125,
      "num_input_tokens_seen": 38656252,
      "step": 691
    },
    {
      "epoch": 1.5412026726057908,
      "grad_norm": 50.515708923339844,
      "learning_rate": 1e-06,
      "loss": 1.1837,
      "num_input_tokens_seen": 38712256,
      "step": 692
    },
    {
      "epoch": 1.5412026726057908,
      "loss": 1.369874358177185,
      "loss_ce": 0.001466137240640819,
      "loss_iou": 0.53515625,
      "loss_num": 0.05908203125,
      "loss_xval": 1.3671875,
      "num_input_tokens_seen": 38712256,
      "step": 692
    },
    {
      "epoch": 1.5434298440979957,
      "grad_norm": 25.205232620239258,
      "learning_rate": 1e-06,
      "loss": 1.059,
      "num_input_tokens_seen": 38769176,
      "step": 693
    },
    {
      "epoch": 1.5434298440979957,
      "loss": 1.1137714385986328,
      "loss_ce": 0.0004902197397314012,
      "loss_iou": 0.462890625,
      "loss_num": 0.037353515625,
      "loss_xval": 1.109375,
      "num_input_tokens_seen": 38769176,
      "step": 693
    },
    {
      "epoch": 1.5456570155902005,
      "grad_norm": 17.223934173583984,
      "learning_rate": 1e-06,
      "loss": 1.1755,
      "num_input_tokens_seen": 38825140,
      "step": 694
    },
    {
      "epoch": 1.5456570155902005,
      "loss": 1.2859344482421875,
      "loss_ce": 0.0017547393217682838,
      "loss_iou": 0.48828125,
      "loss_num": 0.0615234375,
      "loss_xval": 1.28125,
      "num_input_tokens_seen": 38825140,
      "step": 694
    },
    {
      "epoch": 1.5478841870824054,
      "grad_norm": 36.42241287231445,
      "learning_rate": 1e-06,
      "loss": 1.0896,
      "num_input_tokens_seen": 38881164,
      "step": 695
    },
    {
      "epoch": 1.5478841870824054,
      "loss": 0.9976003170013428,
      "loss_ce": 0.0010183160193264484,
      "loss_iou": 0.416015625,
      "loss_num": 0.033203125,
      "loss_xval": 0.99609375,
      "num_input_tokens_seen": 38881164,
      "step": 695
    },
    {
      "epoch": 1.5501113585746102,
      "grad_norm": 53.83267593383789,
      "learning_rate": 1e-06,
      "loss": 1.0186,
      "num_input_tokens_seen": 38937620,
      "step": 696
    },
    {
      "epoch": 1.5501113585746102,
      "loss": 1.0366376638412476,
      "loss_ce": 0.0014814026653766632,
      "loss_iou": 0.435546875,
      "loss_num": 0.032958984375,
      "loss_xval": 1.03125,
      "num_input_tokens_seen": 38937620,
      "step": 696
    },
    {
      "epoch": 1.552338530066815,
      "grad_norm": 26.581382751464844,
      "learning_rate": 1e-06,
      "loss": 0.9886,
      "num_input_tokens_seen": 38990964,
      "step": 697
    },
    {
      "epoch": 1.552338530066815,
      "loss": 1.0419948101043701,
      "loss_ce": 0.0004909674171358347,
      "loss_iou": 0.42578125,
      "loss_num": 0.0380859375,
      "loss_xval": 1.0390625,
      "num_input_tokens_seen": 38990964,
      "step": 697
    },
    {
      "epoch": 1.55456570155902,
      "grad_norm": 20.961688995361328,
      "learning_rate": 1e-06,
      "loss": 1.2405,
      "num_input_tokens_seen": 39046004,
      "step": 698
    },
    {
      "epoch": 1.55456570155902,
      "loss": 1.3255722522735596,
      "loss_ce": 0.0023301024921238422,
      "loss_iou": 0.57421875,
      "loss_num": 0.035400390625,
      "loss_xval": 1.3203125,
      "num_input_tokens_seen": 39046004,
      "step": 698
    },
    {
      "epoch": 1.5567928730512248,
      "grad_norm": 18.719867706298828,
      "learning_rate": 1e-06,
      "loss": 1.0962,
      "num_input_tokens_seen": 39104440,
      "step": 699
    },
    {
      "epoch": 1.5567928730512248,
      "loss": 0.9274605512619019,
      "loss_ce": 0.0011909835739061236,
      "loss_iou": 0.38671875,
      "loss_num": 0.0311279296875,
      "loss_xval": 0.92578125,
      "num_input_tokens_seen": 39104440,
      "step": 699
    },
    {
      "epoch": 1.5590200445434297,
      "grad_norm": 27.276357650756836,
      "learning_rate": 1e-06,
      "loss": 1.0064,
      "num_input_tokens_seen": 39161992,
      "step": 700
    },
    {
      "epoch": 1.5590200445434297,
      "loss": 0.9984862208366394,
      "loss_ce": 0.0011718107853084803,
      "loss_iou": 0.408203125,
      "loss_num": 0.036376953125,
      "loss_xval": 0.99609375,
      "num_input_tokens_seen": 39161992,
      "step": 700
    },
    {
      "epoch": 1.5612472160356348,
      "grad_norm": 17.313262939453125,
      "learning_rate": 1e-06,
      "loss": 1.0717,
      "num_input_tokens_seen": 39216700,
      "step": 701
    },
    {
      "epoch": 1.5612472160356348,
      "loss": 0.8593438863754272,
      "loss_ce": 0.0034632175229489803,
      "loss_iou": 0.35546875,
      "loss_num": 0.0289306640625,
      "loss_xval": 0.85546875,
      "num_input_tokens_seen": 39216700,
      "step": 701
    },
    {
      "epoch": 1.5634743875278396,
      "grad_norm": 26.77741241455078,
      "learning_rate": 1e-06,
      "loss": 0.9991,
      "num_input_tokens_seen": 39273188,
      "step": 702
    },
    {
      "epoch": 1.5634743875278396,
      "loss": 1.1402983665466309,
      "loss_ce": 0.0006498623406514525,
      "loss_iou": 0.48828125,
      "loss_num": 0.032958984375,
      "loss_xval": 1.140625,
      "num_input_tokens_seen": 39273188,
      "step": 702
    },
    {
      "epoch": 1.5657015590200447,
      "grad_norm": 25.94588851928711,
      "learning_rate": 1e-06,
      "loss": 1.2285,
      "num_input_tokens_seen": 39326596,
      "step": 703
    },
    {
      "epoch": 1.5657015590200447,
      "loss": 1.107351541519165,
      "loss_ce": 0.0016386474017053843,
      "loss_iou": 0.392578125,
      "loss_num": 0.064453125,
      "loss_xval": 1.109375,
      "num_input_tokens_seen": 39326596,
      "step": 703
    },
    {
      "epoch": 1.5679287305122496,
      "grad_norm": 108.70765686035156,
      "learning_rate": 1e-06,
      "loss": 1.2865,
      "num_input_tokens_seen": 39381292,
      "step": 704
    },
    {
      "epoch": 1.5679287305122496,
      "loss": 1.2144091129302979,
      "loss_ce": 0.0024950681254267693,
      "loss_iou": 0.4765625,
      "loss_num": 0.051513671875,
      "loss_xval": 1.2109375,
      "num_input_tokens_seen": 39381292,
      "step": 704
    },
    {
      "epoch": 1.5701559020044544,
      "grad_norm": 27.831453323364258,
      "learning_rate": 1e-06,
      "loss": 1.1255,
      "num_input_tokens_seen": 39438056,
      "step": 705
    },
    {
      "epoch": 1.5701559020044544,
      "loss": 1.3021111488342285,
      "loss_ce": 0.0015739843947812915,
      "loss_iou": 0.5078125,
      "loss_num": 0.056884765625,
      "loss_xval": 1.296875,
      "num_input_tokens_seen": 39438056,
      "step": 705
    },
    {
      "epoch": 1.5723830734966593,
      "grad_norm": 35.32048797607422,
      "learning_rate": 1e-06,
      "loss": 0.9482,
      "num_input_tokens_seen": 39492676,
      "step": 706
    },
    {
      "epoch": 1.5723830734966593,
      "loss": 0.9542930722236633,
      "loss_ce": 0.0006797942915000021,
      "loss_iou": 0.390625,
      "loss_num": 0.03515625,
      "loss_xval": 0.953125,
      "num_input_tokens_seen": 39492676,
      "step": 706
    },
    {
      "epoch": 1.5746102449888641,
      "grad_norm": 65.93453216552734,
      "learning_rate": 1e-06,
      "loss": 1.0207,
      "num_input_tokens_seen": 39551388,
      "step": 707
    },
    {
      "epoch": 1.5746102449888641,
      "loss": 0.9525820016860962,
      "loss_ce": 0.0006776798400096595,
      "loss_iou": 0.421875,
      "loss_num": 0.021484375,
      "loss_xval": 0.953125,
      "num_input_tokens_seen": 39551388,
      "step": 707
    },
    {
      "epoch": 1.576837416481069,
      "grad_norm": 14.823631286621094,
      "learning_rate": 1e-06,
      "loss": 1.1039,
      "num_input_tokens_seen": 39607732,
      "step": 708
    },
    {
      "epoch": 1.576837416481069,
      "loss": 0.9811808466911316,
      "loss_ce": 0.00046793223009444773,
      "loss_iou": 0.404296875,
      "loss_num": 0.03466796875,
      "loss_xval": 0.98046875,
      "num_input_tokens_seen": 39607732,
      "step": 708
    },
    {
      "epoch": 1.5790645879732739,
      "grad_norm": 33.79949188232422,
      "learning_rate": 1e-06,
      "loss": 1.0272,
      "num_input_tokens_seen": 39664892,
      "step": 709
    },
    {
      "epoch": 1.5790645879732739,
      "loss": 0.8985086679458618,
      "loss_ce": 0.0005594115937128663,
      "loss_iou": 0.392578125,
      "loss_num": 0.0228271484375,
      "loss_xval": 0.8984375,
      "num_input_tokens_seen": 39664892,
      "step": 709
    },
    {
      "epoch": 1.5812917594654787,
      "grad_norm": 44.01935958862305,
      "learning_rate": 1e-06,
      "loss": 1.2837,
      "num_input_tokens_seen": 39720468,
      "step": 710
    },
    {
      "epoch": 1.5812917594654787,
      "loss": 1.2381861209869385,
      "loss_ce": 0.0008813057793304324,
      "loss_iou": 0.435546875,
      "loss_num": 0.0732421875,
      "loss_xval": 1.234375,
      "num_input_tokens_seen": 39720468,
      "step": 710
    },
    {
      "epoch": 1.5835189309576836,
      "grad_norm": 16.639955520629883,
      "learning_rate": 1e-06,
      "loss": 1.2093,
      "num_input_tokens_seen": 39776680,
      "step": 711
    },
    {
      "epoch": 1.5835189309576836,
      "loss": 1.147930383682251,
      "loss_ce": 0.001445973408408463,
      "loss_iou": 0.453125,
      "loss_num": 0.04833984375,
      "loss_xval": 1.1484375,
      "num_input_tokens_seen": 39776680,
      "step": 711
    },
    {
      "epoch": 1.5857461024498887,
      "grad_norm": 21.6805477142334,
      "learning_rate": 1e-06,
      "loss": 1.0205,
      "num_input_tokens_seen": 39830636,
      "step": 712
    },
    {
      "epoch": 1.5857461024498887,
      "loss": 1.0399500131607056,
      "loss_ce": 0.0016199484234675765,
      "loss_iou": 0.427734375,
      "loss_num": 0.036376953125,
      "loss_xval": 1.0390625,
      "num_input_tokens_seen": 39830636,
      "step": 712
    },
    {
      "epoch": 1.5879732739420935,
      "grad_norm": 50.981754302978516,
      "learning_rate": 1e-06,
      "loss": 1.0619,
      "num_input_tokens_seen": 39887336,
      "step": 713
    },
    {
      "epoch": 1.5879732739420935,
      "loss": 1.1285226345062256,
      "loss_ce": 0.0005930241313762963,
      "loss_iou": 0.43359375,
      "loss_num": 0.052001953125,
      "loss_xval": 1.125,
      "num_input_tokens_seen": 39887336,
      "step": 713
    },
    {
      "epoch": 1.5902004454342984,
      "grad_norm": 1745.5362548828125,
      "learning_rate": 1e-06,
      "loss": 1.0216,
      "num_input_tokens_seen": 39945364,
      "step": 714
    },
    {
      "epoch": 1.5902004454342984,
      "loss": 0.8426003456115723,
      "loss_ce": 0.0008034344646148384,
      "loss_iou": 0.34375,
      "loss_num": 0.031005859375,
      "loss_xval": 0.84375,
      "num_input_tokens_seen": 39945364,
      "step": 714
    },
    {
      "epoch": 1.5924276169265035,
      "grad_norm": 141.83999633789062,
      "learning_rate": 1e-06,
      "loss": 0.9186,
      "num_input_tokens_seen": 39998372,
      "step": 715
    },
    {
      "epoch": 1.5924276169265035,
      "loss": 0.9666658639907837,
      "loss_ce": 0.00035729241790249944,
      "loss_iou": 0.376953125,
      "loss_num": 0.04296875,
      "loss_xval": 0.96484375,
      "num_input_tokens_seen": 39998372,
      "step": 715
    },
    {
      "epoch": 1.5946547884187083,
      "grad_norm": 16.73200225830078,
      "learning_rate": 1e-06,
      "loss": 1.0519,
      "num_input_tokens_seen": 40053772,
      "step": 716
    },
    {
      "epoch": 1.5946547884187083,
      "loss": 1.3080520629882812,
      "loss_ce": 0.005317714065313339,
      "loss_iou": 0.55078125,
      "loss_num": 0.040771484375,
      "loss_xval": 1.3046875,
      "num_input_tokens_seen": 40053772,
      "step": 716
    },
    {
      "epoch": 1.5968819599109132,
      "grad_norm": 19.21235466003418,
      "learning_rate": 1e-06,
      "loss": 0.903,
      "num_input_tokens_seen": 40111152,
      "step": 717
    },
    {
      "epoch": 1.5968819599109132,
      "loss": 0.8331512808799744,
      "loss_ce": 0.00649112556129694,
      "loss_iou": 0.341796875,
      "loss_num": 0.0283203125,
      "loss_xval": 0.828125,
      "num_input_tokens_seen": 40111152,
      "step": 717
    },
    {
      "epoch": 1.599109131403118,
      "grad_norm": 26.08073616027832,
      "learning_rate": 1e-06,
      "loss": 1.0622,
      "num_input_tokens_seen": 40170092,
      "step": 718
    },
    {
      "epoch": 1.599109131403118,
      "loss": 1.0961337089538574,
      "loss_ce": 0.0006747127044945955,
      "loss_iou": 0.43359375,
      "loss_num": 0.045654296875,
      "loss_xval": 1.09375,
      "num_input_tokens_seen": 40170092,
      "step": 718
    },
    {
      "epoch": 1.601336302895323,
      "grad_norm": 38.96022033691406,
      "learning_rate": 1e-06,
      "loss": 1.0349,
      "num_input_tokens_seen": 40222380,
      "step": 719
    },
    {
      "epoch": 1.601336302895323,
      "loss": 1.0462639331817627,
      "loss_ce": 0.00048747030086815357,
      "loss_iou": 0.43359375,
      "loss_num": 0.03564453125,
      "loss_xval": 1.046875,
      "num_input_tokens_seen": 40222380,
      "step": 719
    },
    {
      "epoch": 1.6035634743875278,
      "grad_norm": 15.093145370483398,
      "learning_rate": 1e-06,
      "loss": 1.0315,
      "num_input_tokens_seen": 40279864,
      "step": 720
    },
    {
      "epoch": 1.6035634743875278,
      "loss": 1.010891079902649,
      "loss_ce": 0.0040551056154072285,
      "loss_iou": 0.416015625,
      "loss_num": 0.034912109375,
      "loss_xval": 1.0078125,
      "num_input_tokens_seen": 40279864,
      "step": 720
    },
    {
      "epoch": 1.6057906458797326,
      "grad_norm": 24.701709747314453,
      "learning_rate": 1e-06,
      "loss": 1.314,
      "num_input_tokens_seen": 40334520,
      "step": 721
    },
    {
      "epoch": 1.6057906458797326,
      "loss": 1.5099828243255615,
      "loss_ce": 0.0011937202652916312,
      "loss_iou": 0.5625,
      "loss_num": 0.07763671875,
      "loss_xval": 1.5078125,
      "num_input_tokens_seen": 40334520,
      "step": 721
    },
    {
      "epoch": 1.6080178173719375,
      "grad_norm": 18.33966064453125,
      "learning_rate": 1e-06,
      "loss": 1.0704,
      "num_input_tokens_seen": 40389100,
      "step": 722
    },
    {
      "epoch": 1.6080178173719375,
      "loss": 0.9851089715957642,
      "loss_ce": 0.0004898360930383205,
      "loss_iou": 0.43359375,
      "loss_num": 0.0233154296875,
      "loss_xval": 0.984375,
      "num_input_tokens_seen": 40389100,
      "step": 722
    },
    {
      "epoch": 1.6102449888641426,
      "grad_norm": 51.41325759887695,
      "learning_rate": 1e-06,
      "loss": 0.8882,
      "num_input_tokens_seen": 40446016,
      "step": 723
    },
    {
      "epoch": 1.6102449888641426,
      "loss": 1.013216257095337,
      "loss_ce": 0.0005209506489336491,
      "loss_iou": 0.431640625,
      "loss_num": 0.0296630859375,
      "loss_xval": 1.015625,
      "num_input_tokens_seen": 40446016,
      "step": 723
    },
    {
      "epoch": 1.6124721603563474,
      "grad_norm": 22.967365264892578,
      "learning_rate": 1e-06,
      "loss": 1.1492,
      "num_input_tokens_seen": 40500448,
      "step": 724
    },
    {
      "epoch": 1.6124721603563474,
      "loss": 1.168558120727539,
      "loss_ce": 0.0005894272471778095,
      "loss_iou": 0.4453125,
      "loss_num": 0.0556640625,
      "loss_xval": 1.171875,
      "num_input_tokens_seen": 40500448,
      "step": 724
    },
    {
      "epoch": 1.6146993318485523,
      "grad_norm": 54.2728157043457,
      "learning_rate": 1e-06,
      "loss": 1.0244,
      "num_input_tokens_seen": 40557300,
      "step": 725
    },
    {
      "epoch": 1.6146993318485523,
      "loss": 0.8771815896034241,
      "loss_ce": 0.0009608692489564419,
      "loss_iou": 0.353515625,
      "loss_num": 0.033935546875,
      "loss_xval": 0.875,
      "num_input_tokens_seen": 40557300,
      "step": 725
    },
    {
      "epoch": 1.6169265033407574,
      "grad_norm": 55.31407928466797,
      "learning_rate": 1e-06,
      "loss": 1.0433,
      "num_input_tokens_seen": 40613104,
      "step": 726
    },
    {
      "epoch": 1.6169265033407574,
      "loss": 1.0284351110458374,
      "loss_ce": 0.003044519107788801,
      "loss_iou": 0.4140625,
      "loss_num": 0.0390625,
      "loss_xval": 1.0234375,
      "num_input_tokens_seen": 40613104,
      "step": 726
    },
    {
      "epoch": 1.6191536748329622,
      "grad_norm": 24.382633209228516,
      "learning_rate": 1e-06,
      "loss": 0.9911,
      "num_input_tokens_seen": 40667412,
      "step": 727
    },
    {
      "epoch": 1.6191536748329622,
      "loss": 0.7859052419662476,
      "loss_ce": 0.0005047998856753111,
      "loss_iou": 0.33984375,
      "loss_num": 0.0216064453125,
      "loss_xval": 0.78515625,
      "num_input_tokens_seen": 40667412,
      "step": 727
    },
    {
      "epoch": 1.621380846325167,
      "grad_norm": 22.375062942504883,
      "learning_rate": 1e-06,
      "loss": 0.8341,
      "num_input_tokens_seen": 40724616,
      "step": 728
    },
    {
      "epoch": 1.621380846325167,
      "loss": 0.7646250128746033,
      "loss_ce": 0.0004648742906283587,
      "loss_iou": 0.328125,
      "loss_num": 0.021728515625,
      "loss_xval": 0.765625,
      "num_input_tokens_seen": 40724616,
      "step": 728
    },
    {
      "epoch": 1.623608017817372,
      "grad_norm": 15.496011734008789,
      "learning_rate": 1e-06,
      "loss": 1.1697,
      "num_input_tokens_seen": 40780212,
      "step": 729
    },
    {
      "epoch": 1.623608017817372,
      "loss": 1.1705524921417236,
      "loss_ce": 0.0006306255236268044,
      "loss_iou": 0.484375,
      "loss_num": 0.04052734375,
      "loss_xval": 1.171875,
      "num_input_tokens_seen": 40780212,
      "step": 729
    },
    {
      "epoch": 1.6258351893095768,
      "grad_norm": 24.535715103149414,
      "learning_rate": 1e-06,
      "loss": 1.0096,
      "num_input_tokens_seen": 40837012,
      "step": 730
    },
    {
      "epoch": 1.6258351893095768,
      "loss": 0.9559507369995117,
      "loss_ce": 0.0008726270170882344,
      "loss_iou": 0.39453125,
      "loss_num": 0.033203125,
      "loss_xval": 0.953125,
      "num_input_tokens_seen": 40837012,
      "step": 730
    },
    {
      "epoch": 1.6280623608017817,
      "grad_norm": 16.106016159057617,
      "learning_rate": 1e-06,
      "loss": 1.144,
      "num_input_tokens_seen": 40891652,
      "step": 731
    },
    {
      "epoch": 1.6280623608017817,
      "loss": 1.1018345355987549,
      "loss_ce": 0.000516209052875638,
      "loss_iou": 0.482421875,
      "loss_num": 0.0274658203125,
      "loss_xval": 1.1015625,
      "num_input_tokens_seen": 40891652,
      "step": 731
    },
    {
      "epoch": 1.6302895322939865,
      "grad_norm": 34.02077102661133,
      "learning_rate": 1e-06,
      "loss": 1.173,
      "num_input_tokens_seen": 40947216,
      "step": 732
    },
    {
      "epoch": 1.6302895322939865,
      "loss": 1.2337870597839355,
      "loss_ce": 0.003562505356967449,
      "loss_iou": 0.45703125,
      "loss_num": 0.06298828125,
      "loss_xval": 1.2265625,
      "num_input_tokens_seen": 40947216,
      "step": 732
    },
    {
      "epoch": 1.6325167037861914,
      "grad_norm": 23.39971351623535,
      "learning_rate": 1e-06,
      "loss": 0.7367,
      "num_input_tokens_seen": 41005556,
      "step": 733
    },
    {
      "epoch": 1.6325167037861914,
      "loss": 0.6531933546066284,
      "loss_ce": 0.000361293728929013,
      "loss_iou": 0.2578125,
      "loss_num": 0.02783203125,
      "loss_xval": 0.65234375,
      "num_input_tokens_seen": 41005556,
      "step": 733
    },
    {
      "epoch": 1.6347438752783965,
      "grad_norm": 15.164018630981445,
      "learning_rate": 1e-06,
      "loss": 1.2361,
      "num_input_tokens_seen": 41060208,
      "step": 734
    },
    {
      "epoch": 1.6347438752783965,
      "loss": 1.2857666015625,
      "loss_ce": 0.001342757255770266,
      "loss_iou": 0.484375,
      "loss_num": 0.06298828125,
      "loss_xval": 1.28125,
      "num_input_tokens_seen": 41060208,
      "step": 734
    },
    {
      "epoch": 1.6369710467706013,
      "grad_norm": 24.00090980529785,
      "learning_rate": 1e-06,
      "loss": 1.0606,
      "num_input_tokens_seen": 41116160,
      "step": 735
    },
    {
      "epoch": 1.6369710467706013,
      "loss": 0.6998146772384644,
      "loss_ce": 0.001450410927645862,
      "loss_iou": 0.283203125,
      "loss_num": 0.0262451171875,
      "loss_xval": 0.69921875,
      "num_input_tokens_seen": 41116160,
      "step": 735
    },
    {
      "epoch": 1.6391982182628062,
      "grad_norm": 42.26320266723633,
      "learning_rate": 1e-06,
      "loss": 0.9948,
      "num_input_tokens_seen": 41173264,
      "step": 736
    },
    {
      "epoch": 1.6391982182628062,
      "loss": 0.8883851766586304,
      "loss_ce": 0.0006898957653902471,
      "loss_iou": 0.380859375,
      "loss_num": 0.0252685546875,
      "loss_xval": 0.88671875,
      "num_input_tokens_seen": 41173264,
      "step": 736
    },
    {
      "epoch": 1.6414253897550113,
      "grad_norm": 17.52292823791504,
      "learning_rate": 1e-06,
      "loss": 1.1069,
      "num_input_tokens_seen": 41228928,
      "step": 737
    },
    {
      "epoch": 1.6414253897550113,
      "loss": 1.1075830459594727,
      "loss_ce": 0.001137720886617899,
      "loss_iou": 0.427734375,
      "loss_num": 0.050048828125,
      "loss_xval": 1.109375,
      "num_input_tokens_seen": 41228928,
      "step": 737
    },
    {
      "epoch": 1.6436525612472161,
      "grad_norm": 17.622894287109375,
      "learning_rate": 1e-06,
      "loss": 1.1064,
      "num_input_tokens_seen": 41285968,
      "step": 738
    },
    {
      "epoch": 1.6436525612472161,
      "loss": 0.900823712348938,
      "loss_ce": 0.0004331024829298258,
      "loss_iou": 0.37890625,
      "loss_num": 0.0281982421875,
      "loss_xval": 0.8984375,
      "num_input_tokens_seen": 41285968,
      "step": 738
    },
    {
      "epoch": 1.645879732739421,
      "grad_norm": 40.00740051269531,
      "learning_rate": 1e-06,
      "loss": 0.9503,
      "num_input_tokens_seen": 41341372,
      "step": 739
    },
    {
      "epoch": 1.645879732739421,
      "loss": 0.6320021152496338,
      "loss_ce": 0.0004103146493434906,
      "loss_iou": 0.2734375,
      "loss_num": 0.0167236328125,
      "loss_xval": 0.6328125,
      "num_input_tokens_seen": 41341372,
      "step": 739
    },
    {
      "epoch": 1.6481069042316259,
      "grad_norm": 23.410932540893555,
      "learning_rate": 1e-06,
      "loss": 1.1302,
      "num_input_tokens_seen": 41394872,
      "step": 740
    },
    {
      "epoch": 1.6481069042316259,
      "loss": 1.1568620204925537,
      "loss_ce": 0.001100304420106113,
      "loss_iou": 0.43359375,
      "loss_num": 0.05810546875,
      "loss_xval": 1.15625,
      "num_input_tokens_seen": 41394872,
      "step": 740
    },
    {
      "epoch": 1.6503340757238307,
      "grad_norm": 15.308507919311523,
      "learning_rate": 1e-06,
      "loss": 1.0673,
      "num_input_tokens_seen": 41451676,
      "step": 741
    },
    {
      "epoch": 1.6503340757238307,
      "loss": 1.1007647514343262,
      "loss_ce": 0.0033525261096656322,
      "loss_iou": 0.416015625,
      "loss_num": 0.052734375,
      "loss_xval": 1.09375,
      "num_input_tokens_seen": 41451676,
      "step": 741
    },
    {
      "epoch": 1.6525612472160356,
      "grad_norm": 18.964244842529297,
      "learning_rate": 1e-06,
      "loss": 1.0727,
      "num_input_tokens_seen": 41509472,
      "step": 742
    },
    {
      "epoch": 1.6525612472160356,
      "loss": 0.9488592743873596,
      "loss_ce": 0.001105397124774754,
      "loss_iou": 0.404296875,
      "loss_num": 0.028076171875,
      "loss_xval": 0.94921875,
      "num_input_tokens_seen": 41509472,
      "step": 742
    },
    {
      "epoch": 1.6547884187082404,
      "grad_norm": 25.588586807250977,
      "learning_rate": 1e-06,
      "loss": 0.9794,
      "num_input_tokens_seen": 41565372,
      "step": 743
    },
    {
      "epoch": 1.6547884187082404,
      "loss": 0.8954076766967773,
      "loss_ce": 0.00038817772292532027,
      "loss_iou": 0.3125,
      "loss_num": 0.054443359375,
      "loss_xval": 0.89453125,
      "num_input_tokens_seen": 41565372,
      "step": 743
    },
    {
      "epoch": 1.6570155902004453,
      "grad_norm": 58.01374435424805,
      "learning_rate": 1e-06,
      "loss": 1.0396,
      "num_input_tokens_seen": 41622096,
      "step": 744
    },
    {
      "epoch": 1.6570155902004453,
      "loss": 1.068095088005066,
      "loss_ce": 0.00046813933295197785,
      "loss_iou": 0.44140625,
      "loss_num": 0.036865234375,
      "loss_xval": 1.0703125,
      "num_input_tokens_seen": 41622096,
      "step": 744
    },
    {
      "epoch": 1.6592427616926502,
      "grad_norm": 19.471540451049805,
      "learning_rate": 1e-06,
      "loss": 1.2403,
      "num_input_tokens_seen": 41679236,
      "step": 745
    },
    {
      "epoch": 1.6592427616926502,
      "loss": 1.3742785453796387,
      "loss_ce": 0.00123158423230052,
      "loss_iou": 0.5546875,
      "loss_num": 0.05322265625,
      "loss_xval": 1.375,
      "num_input_tokens_seen": 41679236,
      "step": 745
    },
    {
      "epoch": 1.6614699331848553,
      "grad_norm": 28.21709442138672,
      "learning_rate": 1e-06,
      "loss": 1.0587,
      "num_input_tokens_seen": 41736156,
      "step": 746
    },
    {
      "epoch": 1.6614699331848553,
      "loss": 0.78886479139328,
      "loss_ce": 0.0005347341066226363,
      "loss_iou": 0.33203125,
      "loss_num": 0.02490234375,
      "loss_xval": 0.7890625,
      "num_input_tokens_seen": 41736156,
      "step": 746
    },
    {
      "epoch": 1.6636971046770601,
      "grad_norm": 18.029945373535156,
      "learning_rate": 1e-06,
      "loss": 1.2296,
      "num_input_tokens_seen": 41791268,
      "step": 747
    },
    {
      "epoch": 1.6636971046770601,
      "loss": 1.125550389289856,
      "loss_ce": 0.0007945015095174313,
      "loss_iou": 0.4609375,
      "loss_num": 0.041015625,
      "loss_xval": 1.125,
      "num_input_tokens_seen": 41791268,
      "step": 747
    },
    {
      "epoch": 1.6659242761692652,
      "grad_norm": 19.177194595336914,
      "learning_rate": 1e-06,
      "loss": 1.143,
      "num_input_tokens_seen": 41845948,
      "step": 748
    },
    {
      "epoch": 1.6659242761692652,
      "loss": 1.0622222423553467,
      "loss_ce": 0.0006987220258451998,
      "loss_iou": 0.451171875,
      "loss_num": 0.0322265625,
      "loss_xval": 1.0625,
      "num_input_tokens_seen": 41845948,
      "step": 748
    },
    {
      "epoch": 1.66815144766147,
      "grad_norm": 18.510656356811523,
      "learning_rate": 1e-06,
      "loss": 0.9679,
      "num_input_tokens_seen": 41901836,
      "step": 749
    },
    {
      "epoch": 1.66815144766147,
      "loss": 1.2074395418167114,
      "loss_ce": 0.0008966219611465931,
      "loss_iou": 0.490234375,
      "loss_num": 0.04541015625,
      "loss_xval": 1.203125,
      "num_input_tokens_seen": 41901836,
      "step": 749
    },
    {
      "epoch": 1.670378619153675,
      "grad_norm": 18.269397735595703,
      "learning_rate": 1e-06,
      "loss": 0.8143,
      "num_input_tokens_seen": 41960764,
      "step": 750
    },
    {
      "epoch": 1.670378619153675,
      "eval_seeclick_web_CIoU": 0.5411946177482605,
      "eval_seeclick_web_GIoU": 0.5299433916807175,
      "eval_seeclick_web_IoU": 0.560671478509903,
      "eval_seeclick_web_MAE_all": 0.01778533821925521,
      "eval_seeclick_web_MAE_h": 0.011779951397329569,
      "eval_seeclick_web_MAE_w": 0.017373694106936455,
      "eval_seeclick_web_MAE_x_boxes": 0.01111887488514185,
      "eval_seeclick_web_MAE_y_boxes": 0.02299651806242764,
      "eval_seeclick_web_inside_bbox": 0.9010416567325592,
      "eval_seeclick_web_loss": 1.0158196687698364,
      "eval_seeclick_web_loss_ce": 0.00048441681428812444,
      "eval_seeclick_web_loss_iou": 0.462646484375,
      "eval_seeclick_web_loss_num": 0.014404296875,
      "eval_seeclick_web_loss_xval": 0.9970703125,
      "eval_seeclick_web_runtime": 17.3473,
      "eval_seeclick_web_samples_per_second": 2.882,
      "eval_seeclick_web_steps_per_second": 0.115,
      "num_input_tokens_seen": 41960764,
      "step": 750
    },
    {
      "epoch": 1.670378619153675,
      "eval_icons_CIoU": 0.3312116116285324,
      "eval_icons_GIoU": 0.36676979064941406,
      "eval_icons_IoU": 0.40202146768569946,
      "eval_icons_MAE_all": 0.06723485328257084,
      "eval_icons_MAE_h": 0.03537856135517359,
      "eval_icons_MAE_w": 0.07818298228085041,
      "eval_icons_MAE_x_boxes": 0.058233313262462616,
      "eval_icons_MAE_y_boxes": 0.03735906444489956,
      "eval_icons_inside_bbox": 0.6336805522441864,
      "eval_icons_loss": 1.6458972692489624,
      "eval_icons_loss_ce": 0.0018369617173448205,
      "eval_icons_loss_iou": 0.6297607421875,
      "eval_icons_loss_num": 0.06496238708496094,
      "eval_icons_loss_xval": 1.583984375,
      "eval_icons_runtime": 17.556,
      "eval_icons_samples_per_second": 2.848,
      "eval_icons_steps_per_second": 0.114,
      "num_input_tokens_seen": 41960764,
      "step": 750
    },
    {
      "epoch": 1.670378619153675,
      "eval_screenspot_CIoU": 0.2800278961658478,
      "eval_screenspot_GIoU": 0.3007667362689972,
      "eval_screenspot_IoU": 0.36394914984703064,
      "eval_screenspot_MAE_all": 0.08516304691632588,
      "eval_screenspot_MAE_h": 0.04900899901986122,
      "eval_screenspot_MAE_w": 0.08881760636965434,
      "eval_screenspot_MAE_x_boxes": 0.11530703057845433,
      "eval_screenspot_MAE_y_boxes": 0.05522619063655535,
      "eval_screenspot_inside_bbox": 0.5808333357175192,
      "eval_screenspot_loss": 1.8781194686889648,
      "eval_screenspot_loss_ce": 0.00343730168727537,
      "eval_screenspot_loss_iou": 0.73583984375,
      "eval_screenspot_loss_num": 0.09470876057942708,
      "eval_screenspot_loss_xval": 1.9451497395833333,
      "eval_screenspot_runtime": 27.4038,
      "eval_screenspot_samples_per_second": 3.248,
      "eval_screenspot_steps_per_second": 0.109,
      "num_input_tokens_seen": 41960764,
      "step": 750
    },
    {
      "epoch": 1.670378619153675,
      "eval_compot_CIoU": 0.3159261643886566,
      "eval_compot_GIoU": 0.3490176349878311,
      "eval_compot_IoU": 0.380667582154274,
      "eval_compot_MAE_all": 0.03044109046459198,
      "eval_compot_MAE_h": 0.012972671538591385,
      "eval_compot_MAE_w": 0.040612708777189255,
      "eval_compot_MAE_x_boxes": 0.03857684042304754,
      "eval_compot_MAE_y_boxes": 0.007390682585537434,
      "eval_compot_inside_bbox": 0.5868055522441864,
      "eval_compot_loss": 1.4768078327178955,
      "eval_compot_loss_ce": 0.0005472496850416064,
      "eval_compot_loss_iou": 0.6441650390625,
      "eval_compot_loss_num": 0.027063369750976562,
      "eval_compot_loss_xval": 1.423095703125,
      "eval_compot_runtime": 17.7125,
      "eval_compot_samples_per_second": 2.823,
      "eval_compot_steps_per_second": 0.113,
      "num_input_tokens_seen": 41960764,
      "step": 750
    },
    {
      "epoch": 1.670378619153675,
      "eval_custom_ui_val_CIoU": 0.3929840202132861,
      "eval_custom_ui_val_GIoU": 0.4221853729751375,
      "eval_custom_ui_val_IoU": 0.45630496740341187,
      "eval_custom_ui_val_MAE_all": 0.043078110449843936,
      "eval_custom_ui_val_MAE_h": 0.025929110849069223,
      "eval_custom_ui_val_MAE_w": 0.048152227161659136,
      "eval_custom_ui_val_MAE_x_boxes": 0.04743420394758383,
      "eval_custom_ui_val_MAE_y_boxes": 0.02334741482304202,
      "eval_custom_ui_val_inside_bbox": 0.6608796318372091,
      "eval_custom_ui_val_loss": 1.3928289413452148,
      "eval_custom_ui_val_loss_ce": 0.0016487750755105582,
      "eval_custom_ui_val_loss_iou": 0.5744357638888888,
      "eval_custom_ui_val_loss_num": 0.0423272450764974,
      "eval_custom_ui_val_loss_xval": 1.3607584635416667,
      "eval_custom_ui_val_runtime": 56.9195,
      "eval_custom_ui_val_samples_per_second": 4.656,
      "eval_custom_ui_val_steps_per_second": 0.158,
      "num_input_tokens_seen": 41960764,
      "step": 750
    },
    {
      "epoch": 1.670378619153675,
      "loss": 1.1135873794555664,
      "loss_ce": 0.0010384945198893547,
      "loss_iou": 0.466796875,
      "loss_num": 0.03515625,
      "loss_xval": 1.109375,
      "num_input_tokens_seen": 41960764,
      "step": 750
    },
    {
      "epoch": 1.6726057906458798,
      "grad_norm": 22.51813507080078,
      "learning_rate": 1e-06,
      "loss": 1.0466,
      "num_input_tokens_seen": 42015972,
      "step": 751
    },
    {
      "epoch": 1.6726057906458798,
      "loss": 1.0174639225006104,
      "loss_ce": 0.0008623974863439798,
      "loss_iou": 0.419921875,
      "loss_num": 0.034912109375,
      "loss_xval": 1.015625,
      "num_input_tokens_seen": 42015972,
      "step": 751
    },
    {
      "epoch": 1.6748329621380846,
      "grad_norm": 22.246463775634766,
      "learning_rate": 1e-06,
      "loss": 1.4559,
      "num_input_tokens_seen": 42071584,
      "step": 752
    },
    {
      "epoch": 1.6748329621380846,
      "loss": 1.3008112907409668,
      "loss_ce": 0.001006676466204226,
      "loss_iou": 0.4921875,
      "loss_num": 0.06298828125,
      "loss_xval": 1.296875,
      "num_input_tokens_seen": 42071584,
      "step": 752
    },
    {
      "epoch": 1.6770601336302895,
      "grad_norm": 25.3890380859375,
      "learning_rate": 1e-06,
      "loss": 1.0625,
      "num_input_tokens_seen": 42129356,
      "step": 753
    },
    {
      "epoch": 1.6770601336302895,
      "loss": 0.8968948721885681,
      "loss_ce": 0.0004105077823624015,
      "loss_iou": 0.37109375,
      "loss_num": 0.0311279296875,
      "loss_xval": 0.8984375,
      "num_input_tokens_seen": 42129356,
      "step": 753
    },
    {
      "epoch": 1.6792873051224944,
      "grad_norm": 19.59492301940918,
      "learning_rate": 1e-06,
      "loss": 0.9136,
      "num_input_tokens_seen": 42186100,
      "step": 754
    },
    {
      "epoch": 1.6792873051224944,
      "loss": 0.7982625961303711,
      "loss_ce": 0.0004110607551410794,
      "loss_iou": 0.33203125,
      "loss_num": 0.0267333984375,
      "loss_xval": 0.796875,
      "num_input_tokens_seen": 42186100,
      "step": 754
    },
    {
      "epoch": 1.6815144766146992,
      "grad_norm": 24.47226905822754,
      "learning_rate": 1e-06,
      "loss": 1.1686,
      "num_input_tokens_seen": 42240244,
      "step": 755
    },
    {
      "epoch": 1.6815144766146992,
      "loss": 1.1603378057479858,
      "loss_ce": 0.0004257457912899554,
      "loss_iou": 0.470703125,
      "loss_num": 0.043701171875,
      "loss_xval": 1.15625,
      "num_input_tokens_seen": 42240244,
      "step": 755
    },
    {
      "epoch": 1.683741648106904,
      "grad_norm": 23.5756893157959,
      "learning_rate": 1e-06,
      "loss": 1.1523,
      "num_input_tokens_seen": 42299200,
      "step": 756
    },
    {
      "epoch": 1.683741648106904,
      "loss": 0.9354584217071533,
      "loss_ce": 0.00039978878339752555,
      "loss_iou": 0.388671875,
      "loss_num": 0.03125,
      "loss_xval": 0.93359375,
      "num_input_tokens_seen": 42299200,
      "step": 756
    },
    {
      "epoch": 1.6859688195991092,
      "grad_norm": 29.683124542236328,
      "learning_rate": 1e-06,
      "loss": 1.2437,
      "num_input_tokens_seen": 42355712,
      "step": 757
    },
    {
      "epoch": 1.6859688195991092,
      "loss": 1.3413095474243164,
      "loss_ce": 0.0007333762478083372,
      "loss_iou": 0.55078125,
      "loss_num": 0.046875,
      "loss_xval": 1.34375,
      "num_input_tokens_seen": 42355712,
      "step": 757
    },
    {
      "epoch": 1.688195991091314,
      "grad_norm": 15.45508861541748,
      "learning_rate": 1e-06,
      "loss": 0.7795,
      "num_input_tokens_seen": 42413232,
      "step": 758
    },
    {
      "epoch": 1.688195991091314,
      "loss": 0.9429343938827515,
      "loss_ce": 0.0005515510565601289,
      "loss_iou": 0.396484375,
      "loss_num": 0.0299072265625,
      "loss_xval": 0.94140625,
      "num_input_tokens_seen": 42413232,
      "step": 758
    },
    {
      "epoch": 1.6904231625835189,
      "grad_norm": 24.212417602539062,
      "learning_rate": 1e-06,
      "loss": 0.9237,
      "num_input_tokens_seen": 42470224,
      "step": 759
    },
    {
      "epoch": 1.6904231625835189,
      "loss": 1.025328516960144,
      "loss_ce": 0.005308972671627998,
      "loss_iou": 0.419921875,
      "loss_num": 0.035888671875,
      "loss_xval": 1.0234375,
      "num_input_tokens_seen": 42470224,
      "step": 759
    },
    {
      "epoch": 1.692650334075724,
      "grad_norm": 18.754596710205078,
      "learning_rate": 1e-06,
      "loss": 1.0195,
      "num_input_tokens_seen": 42525456,
      "step": 760
    },
    {
      "epoch": 1.692650334075724,
      "loss": 1.1706150770187378,
      "loss_ce": 0.0036228555254638195,
      "loss_iou": 0.4921875,
      "loss_num": 0.036376953125,
      "loss_xval": 1.1640625,
      "num_input_tokens_seen": 42525456,
      "step": 760
    },
    {
      "epoch": 1.6948775055679288,
      "grad_norm": 19.588882446289062,
      "learning_rate": 1e-06,
      "loss": 0.695,
      "num_input_tokens_seen": 42582136,
      "step": 761
    },
    {
      "epoch": 1.6948775055679288,
      "loss": 0.852080225944519,
      "loss_ce": 0.0005177696002647281,
      "loss_iou": 0.376953125,
      "loss_num": 0.019287109375,
      "loss_xval": 0.8515625,
      "num_input_tokens_seen": 42582136,
      "step": 761
    },
    {
      "epoch": 1.6971046770601337,
      "grad_norm": 32.498355865478516,
      "learning_rate": 1e-06,
      "loss": 1.1963,
      "num_input_tokens_seen": 42637868,
      "step": 762
    },
    {
      "epoch": 1.6971046770601337,
      "loss": 1.1873914003372192,
      "loss_ce": 0.0008679982274770737,
      "loss_iou": 0.484375,
      "loss_num": 0.04345703125,
      "loss_xval": 1.1875,
      "num_input_tokens_seen": 42637868,
      "step": 762
    },
    {
      "epoch": 1.6993318485523385,
      "grad_norm": 24.11650848388672,
      "learning_rate": 1e-06,
      "loss": 1.1282,
      "num_input_tokens_seen": 42689848,
      "step": 763
    },
    {
      "epoch": 1.6993318485523385,
      "loss": 1.3757023811340332,
      "loss_ce": 0.0007023118087090552,
      "loss_iou": 0.578125,
      "loss_num": 0.04443359375,
      "loss_xval": 1.375,
      "num_input_tokens_seen": 42689848,
      "step": 763
    },
    {
      "epoch": 1.7015590200445434,
      "grad_norm": 17.393308639526367,
      "learning_rate": 1e-06,
      "loss": 1.0435,
      "num_input_tokens_seen": 42745660,
      "step": 764
    },
    {
      "epoch": 1.7015590200445434,
      "loss": 0.8240088224411011,
      "loss_ce": 0.0014990322524681687,
      "loss_iou": 0.34375,
      "loss_num": 0.02685546875,
      "loss_xval": 0.82421875,
      "num_input_tokens_seen": 42745660,
      "step": 764
    },
    {
      "epoch": 1.7037861915367483,
      "grad_norm": 26.520469665527344,
      "learning_rate": 1e-06,
      "loss": 1.0743,
      "num_input_tokens_seen": 42802272,
      "step": 765
    },
    {
      "epoch": 1.7037861915367483,
      "loss": 1.2456159591674805,
      "loss_ce": 0.0019635630305856466,
      "loss_iou": 0.5,
      "loss_num": 0.048095703125,
      "loss_xval": 1.2421875,
      "num_input_tokens_seen": 42802272,
      "step": 765
    },
    {
      "epoch": 1.7060133630289531,
      "grad_norm": 34.5315055847168,
      "learning_rate": 1e-06,
      "loss": 1.1088,
      "num_input_tokens_seen": 42859164,
      "step": 766
    },
    {
      "epoch": 1.7060133630289531,
      "loss": 1.120169758796692,
      "loss_ce": 0.0015174560248851776,
      "loss_iou": 0.47265625,
      "loss_num": 0.03466796875,
      "loss_xval": 1.1171875,
      "num_input_tokens_seen": 42859164,
      "step": 766
    },
    {
      "epoch": 1.708240534521158,
      "grad_norm": 28.206703186035156,
      "learning_rate": 1e-06,
      "loss": 1.1139,
      "num_input_tokens_seen": 42911928,
      "step": 767
    },
    {
      "epoch": 1.708240534521158,
      "loss": 1.1372349262237549,
      "loss_ce": 0.0010045571252703667,
      "loss_iou": 0.4765625,
      "loss_num": 0.03662109375,
      "loss_xval": 1.1328125,
      "num_input_tokens_seen": 42911928,
      "step": 767
    },
    {
      "epoch": 1.710467706013363,
      "grad_norm": 44.07477951049805,
      "learning_rate": 1e-06,
      "loss": 0.6435,
      "num_input_tokens_seen": 42970112,
      "step": 768
    },
    {
      "epoch": 1.710467706013363,
      "loss": 0.7950579524040222,
      "loss_ce": 0.0003801731509156525,
      "loss_iou": 0.318359375,
      "loss_num": 0.031494140625,
      "loss_xval": 0.79296875,
      "num_input_tokens_seen": 42970112,
      "step": 768
    },
    {
      "epoch": 1.712694877505568,
      "grad_norm": 39.26905059814453,
      "learning_rate": 1e-06,
      "loss": 1.1469,
      "num_input_tokens_seen": 43025648,
      "step": 769
    },
    {
      "epoch": 1.712694877505568,
      "loss": 1.2926554679870605,
      "loss_ce": 0.0006633971352130175,
      "loss_iou": 0.50390625,
      "loss_num": 0.05712890625,
      "loss_xval": 1.2890625,
      "num_input_tokens_seen": 43025648,
      "step": 769
    },
    {
      "epoch": 1.7149220489977728,
      "grad_norm": 39.54010009765625,
      "learning_rate": 1e-06,
      "loss": 1.1173,
      "num_input_tokens_seen": 43083808,
      "step": 770
    },
    {
      "epoch": 1.7149220489977728,
      "loss": 1.2310657501220703,
      "loss_ce": 0.0010852674022316933,
      "loss_iou": 0.482421875,
      "loss_num": 0.052978515625,
      "loss_xval": 1.2265625,
      "num_input_tokens_seen": 43083808,
      "step": 770
    },
    {
      "epoch": 1.7171492204899779,
      "grad_norm": 19.132938385009766,
      "learning_rate": 1e-06,
      "loss": 1.06,
      "num_input_tokens_seen": 43142508,
      "step": 771
    },
    {
      "epoch": 1.7171492204899779,
      "loss": 0.9760845303535461,
      "loss_ce": 0.0004985497798770666,
      "loss_iou": 0.412109375,
      "loss_num": 0.0308837890625,
      "loss_xval": 0.9765625,
      "num_input_tokens_seen": 43142508,
      "step": 771
    },
    {
      "epoch": 1.7193763919821827,
      "grad_norm": 23.097333908081055,
      "learning_rate": 1e-06,
      "loss": 1.0792,
      "num_input_tokens_seen": 43196132,
      "step": 772
    },
    {
      "epoch": 1.7193763919821827,
      "loss": 0.7085492014884949,
      "loss_ce": 0.0005413593607954681,
      "loss_iou": 0.28515625,
      "loss_num": 0.027099609375,
      "loss_xval": 0.70703125,
      "num_input_tokens_seen": 43196132,
      "step": 772
    },
    {
      "epoch": 1.7216035634743876,
      "grad_norm": 30.94361686706543,
      "learning_rate": 1e-06,
      "loss": 1.0867,
      "num_input_tokens_seen": 43253588,
      "step": 773
    },
    {
      "epoch": 1.7216035634743876,
      "loss": 0.9410488605499268,
      "loss_ce": 0.0011074627982452512,
      "loss_iou": 0.3828125,
      "loss_num": 0.034912109375,
      "loss_xval": 0.94140625,
      "num_input_tokens_seen": 43253588,
      "step": 773
    },
    {
      "epoch": 1.7238307349665924,
      "grad_norm": 23.98365592956543,
      "learning_rate": 1e-06,
      "loss": 1.3351,
      "num_input_tokens_seen": 43308080,
      "step": 774
    },
    {
      "epoch": 1.7238307349665924,
      "loss": 1.3561934232711792,
      "loss_ce": 0.0007246616296470165,
      "loss_iou": 0.515625,
      "loss_num": 0.0654296875,
      "loss_xval": 1.359375,
      "num_input_tokens_seen": 43308080,
      "step": 774
    },
    {
      "epoch": 1.7260579064587973,
      "grad_norm": 21.451553344726562,
      "learning_rate": 1e-06,
      "loss": 0.835,
      "num_input_tokens_seen": 43365784,
      "step": 775
    },
    {
      "epoch": 1.7260579064587973,
      "loss": 0.7267694473266602,
      "loss_ce": 0.0019159411313012242,
      "loss_iou": 0.306640625,
      "loss_num": 0.0224609375,
      "loss_xval": 0.7265625,
      "num_input_tokens_seen": 43365784,
      "step": 775
    },
    {
      "epoch": 1.7282850779510022,
      "grad_norm": 32.42961502075195,
      "learning_rate": 1e-06,
      "loss": 0.7633,
      "num_input_tokens_seen": 43420724,
      "step": 776
    },
    {
      "epoch": 1.7282850779510022,
      "loss": 0.8495993614196777,
      "loss_ce": 0.0009665663237683475,
      "loss_iou": 0.345703125,
      "loss_num": 0.031494140625,
      "loss_xval": 0.84765625,
      "num_input_tokens_seen": 43420724,
      "step": 776
    },
    {
      "epoch": 1.730512249443207,
      "grad_norm": 12.944496154785156,
      "learning_rate": 1e-06,
      "loss": 1.1678,
      "num_input_tokens_seen": 43478128,
      "step": 777
    },
    {
      "epoch": 1.730512249443207,
      "loss": 1.2062785625457764,
      "loss_ce": 0.0004680473357439041,
      "loss_iou": 0.5,
      "loss_num": 0.041259765625,
      "loss_xval": 1.203125,
      "num_input_tokens_seen": 43478128,
      "step": 777
    },
    {
      "epoch": 1.732739420935412,
      "grad_norm": 24.311969757080078,
      "learning_rate": 1e-06,
      "loss": 1.3848,
      "num_input_tokens_seen": 43535104,
      "step": 778
    },
    {
      "epoch": 1.732739420935412,
      "loss": 1.19753098487854,
      "loss_ce": 0.0007536107441410422,
      "loss_iou": 0.486328125,
      "loss_num": 0.04443359375,
      "loss_xval": 1.1953125,
      "num_input_tokens_seen": 43535104,
      "step": 778
    },
    {
      "epoch": 1.734966592427617,
      "grad_norm": 24.397178649902344,
      "learning_rate": 1e-06,
      "loss": 1.0169,
      "num_input_tokens_seen": 43592232,
      "step": 779
    },
    {
      "epoch": 1.734966592427617,
      "loss": 1.0153961181640625,
      "loss_ce": 0.003433296922594309,
      "loss_iou": 0.40625,
      "loss_num": 0.040283203125,
      "loss_xval": 1.015625,
      "num_input_tokens_seen": 43592232,
      "step": 779
    },
    {
      "epoch": 1.7371937639198218,
      "grad_norm": 18.340551376342773,
      "learning_rate": 1e-06,
      "loss": 1.0615,
      "num_input_tokens_seen": 43645688,
      "step": 780
    },
    {
      "epoch": 1.7371937639198218,
      "loss": 1.023805856704712,
      "loss_ce": 0.0007346307393163443,
      "loss_iou": 0.404296875,
      "loss_num": 0.04248046875,
      "loss_xval": 1.0234375,
      "num_input_tokens_seen": 43645688,
      "step": 780
    },
    {
      "epoch": 1.7394209354120267,
      "grad_norm": 16.935611724853516,
      "learning_rate": 1e-06,
      "loss": 1.0056,
      "num_input_tokens_seen": 43702564,
      "step": 781
    },
    {
      "epoch": 1.7394209354120267,
      "loss": 0.7179310321807861,
      "loss_ce": 0.002354835858568549,
      "loss_iou": 0.291015625,
      "loss_num": 0.026611328125,
      "loss_xval": 0.71484375,
      "num_input_tokens_seen": 43702564,
      "step": 781
    },
    {
      "epoch": 1.7416481069042318,
      "grad_norm": 22.93832015991211,
      "learning_rate": 1e-06,
      "loss": 0.9551,
      "num_input_tokens_seen": 43760636,
      "step": 782
    },
    {
      "epoch": 1.7416481069042318,
      "loss": 1.01814603805542,
      "loss_ce": 0.0010561385424807668,
      "loss_iou": 0.423828125,
      "loss_num": 0.033935546875,
      "loss_xval": 1.015625,
      "num_input_tokens_seen": 43760636,
      "step": 782
    },
    {
      "epoch": 1.7438752783964366,
      "grad_norm": 26.112279891967773,
      "learning_rate": 1e-06,
      "loss": 1.104,
      "num_input_tokens_seen": 43818316,
      "step": 783
    },
    {
      "epoch": 1.7438752783964366,
      "loss": 0.8660245537757874,
      "loss_ce": 0.0020108623430132866,
      "loss_iou": 0.37109375,
      "loss_num": 0.0244140625,
      "loss_xval": 0.86328125,
      "num_input_tokens_seen": 43818316,
      "step": 783
    },
    {
      "epoch": 1.7461024498886415,
      "grad_norm": 20.74223518371582,
      "learning_rate": 1e-06,
      "loss": 0.9333,
      "num_input_tokens_seen": 43874416,
      "step": 784
    },
    {
      "epoch": 1.7461024498886415,
      "loss": 1.0379929542541504,
      "loss_ce": 0.0003951968683395535,
      "loss_iou": 0.42578125,
      "loss_num": 0.037109375,
      "loss_xval": 1.0390625,
      "num_input_tokens_seen": 43874416,
      "step": 784
    },
    {
      "epoch": 1.7483296213808464,
      "grad_norm": 16.81831169128418,
      "learning_rate": 1e-06,
      "loss": 0.8192,
      "num_input_tokens_seen": 43929028,
      "step": 785
    },
    {
      "epoch": 1.7483296213808464,
      "loss": 0.7447642087936401,
      "loss_ce": 0.0003794525982812047,
      "loss_iou": 0.322265625,
      "loss_num": 0.020263671875,
      "loss_xval": 0.74609375,
      "num_input_tokens_seen": 43929028,
      "step": 785
    },
    {
      "epoch": 1.7505567928730512,
      "grad_norm": 15.588628768920898,
      "learning_rate": 1e-06,
      "loss": 1.0575,
      "num_input_tokens_seen": 43985228,
      "step": 786
    },
    {
      "epoch": 1.7505567928730512,
      "loss": 0.9384199380874634,
      "loss_ce": 0.0004316343110986054,
      "loss_iou": 0.376953125,
      "loss_num": 0.03662109375,
      "loss_xval": 0.9375,
      "num_input_tokens_seen": 43985228,
      "step": 786
    },
    {
      "epoch": 1.752783964365256,
      "grad_norm": 47.489871978759766,
      "learning_rate": 1e-06,
      "loss": 0.8388,
      "num_input_tokens_seen": 44042064,
      "step": 787
    },
    {
      "epoch": 1.752783964365256,
      "loss": 0.9427950382232666,
      "loss_ce": 0.0006563607603311539,
      "loss_iou": 0.3828125,
      "loss_num": 0.034912109375,
      "loss_xval": 0.94140625,
      "num_input_tokens_seen": 44042064,
      "step": 787
    },
    {
      "epoch": 1.755011135857461,
      "grad_norm": 23.84717559814453,
      "learning_rate": 1e-06,
      "loss": 1.1627,
      "num_input_tokens_seen": 44099536,
      "step": 788
    },
    {
      "epoch": 1.755011135857461,
      "loss": 1.0481215715408325,
      "loss_ce": 0.0007583041442558169,
      "loss_iou": 0.42578125,
      "loss_num": 0.038818359375,
      "loss_xval": 1.046875,
      "num_input_tokens_seen": 44099536,
      "step": 788
    },
    {
      "epoch": 1.7572383073496658,
      "grad_norm": 15.595168113708496,
      "learning_rate": 1e-06,
      "loss": 0.9104,
      "num_input_tokens_seen": 44156140,
      "step": 789
    },
    {
      "epoch": 1.7572383073496658,
      "loss": 0.9101336002349854,
      "loss_ce": 0.0004656048258766532,
      "loss_iou": 0.369140625,
      "loss_num": 0.034912109375,
      "loss_xval": 0.91015625,
      "num_input_tokens_seen": 44156140,
      "step": 789
    },
    {
      "epoch": 1.7594654788418709,
      "grad_norm": 19.3514404296875,
      "learning_rate": 1e-06,
      "loss": 1.3284,
      "num_input_tokens_seen": 44212504,
      "step": 790
    },
    {
      "epoch": 1.7594654788418709,
      "loss": 1.7789078950881958,
      "loss_ce": 0.0010758922435343266,
      "loss_iou": 0.6953125,
      "loss_num": 0.07861328125,
      "loss_xval": 1.78125,
      "num_input_tokens_seen": 44212504,
      "step": 790
    },
    {
      "epoch": 1.7616926503340757,
      "grad_norm": 20.171802520751953,
      "learning_rate": 1e-06,
      "loss": 0.8222,
      "num_input_tokens_seen": 44268412,
      "step": 791
    },
    {
      "epoch": 1.7616926503340757,
      "loss": 0.7054498195648193,
      "loss_ce": 0.00037169185816310346,
      "loss_iou": 0.283203125,
      "loss_num": 0.02783203125,
      "loss_xval": 0.703125,
      "num_input_tokens_seen": 44268412,
      "step": 791
    },
    {
      "epoch": 1.7639198218262806,
      "grad_norm": 26.683834075927734,
      "learning_rate": 1e-06,
      "loss": 0.9773,
      "num_input_tokens_seen": 44326872,
      "step": 792
    },
    {
      "epoch": 1.7639198218262806,
      "loss": 1.1897196769714355,
      "loss_ce": 0.0007548188441433012,
      "loss_iou": 0.47265625,
      "loss_num": 0.048583984375,
      "loss_xval": 1.1875,
      "num_input_tokens_seen": 44326872,
      "step": 792
    },
    {
      "epoch": 1.7661469933184857,
      "grad_norm": 32.8150749206543,
      "learning_rate": 1e-06,
      "loss": 0.9811,
      "num_input_tokens_seen": 44380344,
      "step": 793
    },
    {
      "epoch": 1.7661469933184857,
      "loss": 1.1400426626205444,
      "loss_ce": 0.0006384018342941999,
      "loss_iou": 0.462890625,
      "loss_num": 0.042724609375,
      "loss_xval": 1.140625,
      "num_input_tokens_seen": 44380344,
      "step": 793
    },
    {
      "epoch": 1.7683741648106905,
      "grad_norm": 117.90059661865234,
      "learning_rate": 1e-06,
      "loss": 1.0223,
      "num_input_tokens_seen": 44435364,
      "step": 794
    },
    {
      "epoch": 1.7683741648106905,
      "loss": 0.6915375590324402,
      "loss_ce": 0.0008637503487989306,
      "loss_iou": 0.279296875,
      "loss_num": 0.02685546875,
      "loss_xval": 0.69140625,
      "num_input_tokens_seen": 44435364,
      "step": 794
    },
    {
      "epoch": 1.7706013363028954,
      "grad_norm": 20.04206085205078,
      "learning_rate": 1e-06,
      "loss": 0.6872,
      "num_input_tokens_seen": 44492192,
      "step": 795
    },
    {
      "epoch": 1.7706013363028954,
      "loss": 0.6449465751647949,
      "loss_ce": 0.00041528072324581444,
      "loss_iou": 0.267578125,
      "loss_num": 0.022216796875,
      "loss_xval": 0.64453125,
      "num_input_tokens_seen": 44492192,
      "step": 795
    },
    {
      "epoch": 1.7728285077951003,
      "grad_norm": 17.540992736816406,
      "learning_rate": 1e-06,
      "loss": 0.8579,
      "num_input_tokens_seen": 44548236,
      "step": 796
    },
    {
      "epoch": 1.7728285077951003,
      "loss": 0.8379471302032471,
      "loss_ce": 0.0005447610164992511,
      "loss_iou": 0.33984375,
      "loss_num": 0.031494140625,
      "loss_xval": 0.8359375,
      "num_input_tokens_seen": 44548236,
      "step": 796
    },
    {
      "epoch": 1.7750556792873051,
      "grad_norm": 21.848678588867188,
      "learning_rate": 1e-06,
      "loss": 0.9379,
      "num_input_tokens_seen": 44603416,
      "step": 797
    },
    {
      "epoch": 1.7750556792873051,
      "loss": 1.0255460739135742,
      "loss_ce": 0.0005215964047238231,
      "loss_iou": 0.416015625,
      "loss_num": 0.038818359375,
      "loss_xval": 1.0234375,
      "num_input_tokens_seen": 44603416,
      "step": 797
    },
    {
      "epoch": 1.77728285077951,
      "grad_norm": 17.641189575195312,
      "learning_rate": 1e-06,
      "loss": 1.009,
      "num_input_tokens_seen": 44659364,
      "step": 798
    },
    {
      "epoch": 1.77728285077951,
      "loss": 1.0162487030029297,
      "loss_ce": 0.0006236857152543962,
      "loss_iou": 0.431640625,
      "loss_num": 0.0301513671875,
      "loss_xval": 1.015625,
      "num_input_tokens_seen": 44659364,
      "step": 798
    },
    {
      "epoch": 1.7795100222717148,
      "grad_norm": 32.69601058959961,
      "learning_rate": 1e-06,
      "loss": 1.0617,
      "num_input_tokens_seen": 44716108,
      "step": 799
    },
    {
      "epoch": 1.7795100222717148,
      "loss": 1.2807564735412598,
      "loss_ce": 0.0009712378960102797,
      "loss_iou": 0.484375,
      "loss_num": 0.062255859375,
      "loss_xval": 1.28125,
      "num_input_tokens_seen": 44716108,
      "step": 799
    },
    {
      "epoch": 1.7817371937639197,
      "grad_norm": 18.248733520507812,
      "learning_rate": 1e-06,
      "loss": 0.9217,
      "num_input_tokens_seen": 44775240,
      "step": 800
    },
    {
      "epoch": 1.7817371937639197,
      "loss": 0.9542760252952576,
      "loss_ce": 0.0026158532127738,
      "loss_iou": 0.392578125,
      "loss_num": 0.03369140625,
      "loss_xval": 0.953125,
      "num_input_tokens_seen": 44775240,
      "step": 800
    },
    {
      "epoch": 1.7839643652561246,
      "grad_norm": 31.044193267822266,
      "learning_rate": 1e-06,
      "loss": 1.0257,
      "num_input_tokens_seen": 44830904,
      "step": 801
    },
    {
      "epoch": 1.7839643652561246,
      "loss": 0.9924131035804749,
      "loss_ce": 0.000713902700226754,
      "loss_iou": 0.365234375,
      "loss_num": 0.0517578125,
      "loss_xval": 0.9921875,
      "num_input_tokens_seen": 44830904,
      "step": 801
    },
    {
      "epoch": 1.7861915367483296,
      "grad_norm": 16.06730842590332,
      "learning_rate": 1e-06,
      "loss": 1.0,
      "num_input_tokens_seen": 44885496,
      "step": 802
    },
    {
      "epoch": 1.7861915367483296,
      "loss": 1.0914322137832642,
      "loss_ce": 0.0006118253222666681,
      "loss_iou": 0.4453125,
      "loss_num": 0.0400390625,
      "loss_xval": 1.09375,
      "num_input_tokens_seen": 44885496,
      "step": 802
    },
    {
      "epoch": 1.7884187082405345,
      "grad_norm": 18.782161712646484,
      "learning_rate": 1e-06,
      "loss": 1.0227,
      "num_input_tokens_seen": 44939520,
      "step": 803
    },
    {
      "epoch": 1.7884187082405345,
      "loss": 1.2903541326522827,
      "loss_ce": 0.0008032987243495882,
      "loss_iou": 0.53125,
      "loss_num": 0.045166015625,
      "loss_xval": 1.2890625,
      "num_input_tokens_seen": 44939520,
      "step": 803
    },
    {
      "epoch": 1.7906458797327396,
      "grad_norm": 14.930334091186523,
      "learning_rate": 1e-06,
      "loss": 0.9842,
      "num_input_tokens_seen": 44994976,
      "step": 804
    },
    {
      "epoch": 1.7906458797327396,
      "loss": 0.6712737679481506,
      "loss_ce": 0.0028167327400296926,
      "loss_iou": 0.27734375,
      "loss_num": 0.0223388671875,
      "loss_xval": 0.66796875,
      "num_input_tokens_seen": 44994976,
      "step": 804
    },
    {
      "epoch": 1.7928730512249444,
      "grad_norm": 21.465778350830078,
      "learning_rate": 1e-06,
      "loss": 1.161,
      "num_input_tokens_seen": 45053556,
      "step": 805
    },
    {
      "epoch": 1.7928730512249444,
      "loss": 1.2845568656921387,
      "loss_ce": 0.0006212838925421238,
      "loss_iou": 0.486328125,
      "loss_num": 0.062255859375,
      "loss_xval": 1.28125,
      "num_input_tokens_seen": 45053556,
      "step": 805
    },
    {
      "epoch": 1.7951002227171493,
      "grad_norm": 20.920679092407227,
      "learning_rate": 1e-06,
      "loss": 1.1158,
      "num_input_tokens_seen": 45110760,
      "step": 806
    },
    {
      "epoch": 1.7951002227171493,
      "loss": 1.0931113958358765,
      "loss_ce": 0.000582136504817754,
      "loss_iou": 0.44140625,
      "loss_num": 0.04150390625,
      "loss_xval": 1.09375,
      "num_input_tokens_seen": 45110760,
      "step": 806
    },
    {
      "epoch": 1.7973273942093542,
      "grad_norm": 29.755828857421875,
      "learning_rate": 1e-06,
      "loss": 1.2162,
      "num_input_tokens_seen": 45166492,
      "step": 807
    },
    {
      "epoch": 1.7973273942093542,
      "loss": 1.4112908840179443,
      "loss_ce": 0.0006463178433477879,
      "loss_iou": 0.5703125,
      "loss_num": 0.0537109375,
      "loss_xval": 1.4140625,
      "num_input_tokens_seen": 45166492,
      "step": 807
    },
    {
      "epoch": 1.799554565701559,
      "grad_norm": 25.334617614746094,
      "learning_rate": 1e-06,
      "loss": 1.0345,
      "num_input_tokens_seen": 45222840,
      "step": 808
    },
    {
      "epoch": 1.799554565701559,
      "loss": 0.8354628086090088,
      "loss_ce": 0.0005018864176236093,
      "loss_iou": 0.32421875,
      "loss_num": 0.03759765625,
      "loss_xval": 0.8359375,
      "num_input_tokens_seen": 45222840,
      "step": 808
    },
    {
      "epoch": 1.8017817371937639,
      "grad_norm": 12.594086647033691,
      "learning_rate": 1e-06,
      "loss": 1.0409,
      "num_input_tokens_seen": 45278916,
      "step": 809
    },
    {
      "epoch": 1.8017817371937639,
      "loss": 0.8014848232269287,
      "loss_ce": 0.0004594190395437181,
      "loss_iou": 0.326171875,
      "loss_num": 0.029541015625,
      "loss_xval": 0.80078125,
      "num_input_tokens_seen": 45278916,
      "step": 809
    },
    {
      "epoch": 1.8040089086859687,
      "grad_norm": 19.810697555541992,
      "learning_rate": 1e-06,
      "loss": 0.8261,
      "num_input_tokens_seen": 45331148,
      "step": 810
    },
    {
      "epoch": 1.8040089086859687,
      "loss": 0.7200347185134888,
      "loss_ce": 0.0005523251602426171,
      "loss_iou": 0.279296875,
      "loss_num": 0.031982421875,
      "loss_xval": 0.71875,
      "num_input_tokens_seen": 45331148,
      "step": 810
    },
    {
      "epoch": 1.8062360801781736,
      "grad_norm": 15.187623023986816,
      "learning_rate": 1e-06,
      "loss": 0.9049,
      "num_input_tokens_seen": 45385580,
      "step": 811
    },
    {
      "epoch": 1.8062360801781736,
      "loss": 0.8386489748954773,
      "loss_ce": 0.001979056978598237,
      "loss_iou": 0.322265625,
      "loss_num": 0.03857421875,
      "loss_xval": 0.8359375,
      "num_input_tokens_seen": 45385580,
      "step": 811
    },
    {
      "epoch": 1.8084632516703785,
      "grad_norm": 30.983192443847656,
      "learning_rate": 1e-06,
      "loss": 1.2259,
      "num_input_tokens_seen": 45440584,
      "step": 812
    },
    {
      "epoch": 1.8084632516703785,
      "loss": 0.9607589244842529,
      "loss_ce": 0.001286255195736885,
      "loss_iou": 0.41015625,
      "loss_num": 0.0281982421875,
      "loss_xval": 0.9609375,
      "num_input_tokens_seen": 45440584,
      "step": 812
    },
    {
      "epoch": 1.8106904231625836,
      "grad_norm": 16.581043243408203,
      "learning_rate": 1e-06,
      "loss": 0.9445,
      "num_input_tokens_seen": 45493384,
      "step": 813
    },
    {
      "epoch": 1.8106904231625836,
      "loss": 0.8210749626159668,
      "loss_ce": 0.0007623857818543911,
      "loss_iou": 0.34765625,
      "loss_num": 0.0252685546875,
      "loss_xval": 0.8203125,
      "num_input_tokens_seen": 45493384,
      "step": 813
    },
    {
      "epoch": 1.8129175946547884,
      "grad_norm": 20.215229034423828,
      "learning_rate": 1e-06,
      "loss": 1.0194,
      "num_input_tokens_seen": 45548060,
      "step": 814
    },
    {
      "epoch": 1.8129175946547884,
      "loss": 1.2104692459106445,
      "loss_ce": 0.0029496951028704643,
      "loss_iou": 0.46484375,
      "loss_num": 0.055908203125,
      "loss_xval": 1.2109375,
      "num_input_tokens_seen": 45548060,
      "step": 814
    },
    {
      "epoch": 1.8151447661469933,
      "grad_norm": 17.977659225463867,
      "learning_rate": 1e-06,
      "loss": 0.9403,
      "num_input_tokens_seen": 45605948,
      "step": 815
    },
    {
      "epoch": 1.8151447661469933,
      "loss": 0.7051939964294434,
      "loss_ce": 0.0003600172349251807,
      "loss_iou": 0.302734375,
      "loss_num": 0.019775390625,
      "loss_xval": 0.703125,
      "num_input_tokens_seen": 45605948,
      "step": 815
    },
    {
      "epoch": 1.8173719376391984,
      "grad_norm": 21.293136596679688,
      "learning_rate": 1e-06,
      "loss": 0.9601,
      "num_input_tokens_seen": 45660788,
      "step": 816
    },
    {
      "epoch": 1.8173719376391984,
      "loss": 0.7573980093002319,
      "loss_ce": 0.0005621028249152005,
      "loss_iou": 0.326171875,
      "loss_num": 0.020751953125,
      "loss_xval": 0.7578125,
      "num_input_tokens_seen": 45660788,
      "step": 816
    },
    {
      "epoch": 1.8195991091314032,
      "grad_norm": 31.912277221679688,
      "learning_rate": 1e-06,
      "loss": 1.2548,
      "num_input_tokens_seen": 45713720,
      "step": 817
    },
    {
      "epoch": 1.8195991091314032,
      "loss": 1.1719120740890503,
      "loss_ce": 0.012366149574518204,
      "loss_iou": 0.490234375,
      "loss_num": 0.03564453125,
      "loss_xval": 1.15625,
      "num_input_tokens_seen": 45713720,
      "step": 817
    },
    {
      "epoch": 1.821826280623608,
      "grad_norm": 43.602935791015625,
      "learning_rate": 1e-06,
      "loss": 1.0844,
      "num_input_tokens_seen": 45769628,
      "step": 818
    },
    {
      "epoch": 1.821826280623608,
      "loss": 1.0855807065963745,
      "loss_ce": 0.0006197973270900548,
      "loss_iou": 0.451171875,
      "loss_num": 0.036865234375,
      "loss_xval": 1.0859375,
      "num_input_tokens_seen": 45769628,
      "step": 818
    },
    {
      "epoch": 1.824053452115813,
      "grad_norm": 23.616613388061523,
      "learning_rate": 1e-06,
      "loss": 1.1116,
      "num_input_tokens_seen": 45827072,
      "step": 819
    },
    {
      "epoch": 1.824053452115813,
      "loss": 1.052821159362793,
      "loss_ce": 0.0005750858690589666,
      "loss_iou": 0.4375,
      "loss_num": 0.03515625,
      "loss_xval": 1.0546875,
      "num_input_tokens_seen": 45827072,
      "step": 819
    },
    {
      "epoch": 1.8262806236080178,
      "grad_norm": 24.936237335205078,
      "learning_rate": 1e-06,
      "loss": 1.1643,
      "num_input_tokens_seen": 45882452,
      "step": 820
    },
    {
      "epoch": 1.8262806236080178,
      "loss": 1.1679127216339111,
      "loss_ce": 0.00043235800694674253,
      "loss_iou": 0.466796875,
      "loss_num": 0.046630859375,
      "loss_xval": 1.1640625,
      "num_input_tokens_seen": 45882452,
      "step": 820
    },
    {
      "epoch": 1.8285077951002227,
      "grad_norm": 27.94110107421875,
      "learning_rate": 1e-06,
      "loss": 1.0027,
      "num_input_tokens_seen": 45937344,
      "step": 821
    },
    {
      "epoch": 1.8285077951002227,
      "loss": 1.220947265625,
      "loss_ce": 0.0007323599420487881,
      "loss_iou": 0.46875,
      "loss_num": 0.056396484375,
      "loss_xval": 1.21875,
      "num_input_tokens_seen": 45937344,
      "step": 821
    },
    {
      "epoch": 1.8307349665924275,
      "grad_norm": 19.486064910888672,
      "learning_rate": 1e-06,
      "loss": 0.9696,
      "num_input_tokens_seen": 45994040,
      "step": 822
    },
    {
      "epoch": 1.8307349665924275,
      "loss": 0.7867767214775085,
      "loss_ce": 0.0003997594176325947,
      "loss_iou": 0.32421875,
      "loss_num": 0.028076171875,
      "loss_xval": 0.78515625,
      "num_input_tokens_seen": 45994040,
      "step": 822
    },
    {
      "epoch": 1.8329621380846324,
      "grad_norm": 33.13214111328125,
      "learning_rate": 1e-06,
      "loss": 1.2236,
      "num_input_tokens_seen": 46049792,
      "step": 823
    },
    {
      "epoch": 1.8329621380846324,
      "loss": 1.2174769639968872,
      "loss_ce": 0.0011683637276291847,
      "loss_iou": 0.486328125,
      "loss_num": 0.049072265625,
      "loss_xval": 1.21875,
      "num_input_tokens_seen": 46049792,
      "step": 823
    },
    {
      "epoch": 1.8351893095768375,
      "grad_norm": 22.1630802154541,
      "learning_rate": 1e-06,
      "loss": 1.0282,
      "num_input_tokens_seen": 46106908,
      "step": 824
    },
    {
      "epoch": 1.8351893095768375,
      "loss": 1.025941252708435,
      "loss_ce": 0.0005506377201527357,
      "loss_iou": 0.4453125,
      "loss_num": 0.02685546875,
      "loss_xval": 1.0234375,
      "num_input_tokens_seen": 46106908,
      "step": 824
    },
    {
      "epoch": 1.8374164810690423,
      "grad_norm": 18.732593536376953,
      "learning_rate": 1e-06,
      "loss": 1.1442,
      "num_input_tokens_seen": 46161732,
      "step": 825
    },
    {
      "epoch": 1.8374164810690423,
      "loss": 1.1100322008132935,
      "loss_ce": 0.0004130484303459525,
      "loss_iou": 0.4453125,
      "loss_num": 0.044189453125,
      "loss_xval": 1.109375,
      "num_input_tokens_seen": 46161732,
      "step": 825
    },
    {
      "epoch": 1.8396436525612472,
      "grad_norm": 16.942766189575195,
      "learning_rate": 1e-06,
      "loss": 1.1521,
      "num_input_tokens_seen": 46217944,
      "step": 826
    },
    {
      "epoch": 1.8396436525612472,
      "loss": 0.9271703362464905,
      "loss_ce": 0.0018773877527564764,
      "loss_iou": 0.3828125,
      "loss_num": 0.031982421875,
      "loss_xval": 0.92578125,
      "num_input_tokens_seen": 46217944,
      "step": 826
    },
    {
      "epoch": 1.8418708240534523,
      "grad_norm": 19.92278289794922,
      "learning_rate": 1e-06,
      "loss": 1.1781,
      "num_input_tokens_seen": 46274360,
      "step": 827
    },
    {
      "epoch": 1.8418708240534523,
      "loss": 1.132466197013855,
      "loss_ce": 0.0008743547950871289,
      "loss_iou": 0.443359375,
      "loss_num": 0.04931640625,
      "loss_xval": 1.1328125,
      "num_input_tokens_seen": 46274360,
      "step": 827
    },
    {
      "epoch": 1.8440979955456571,
      "grad_norm": 27.525070190429688,
      "learning_rate": 1e-06,
      "loss": 0.9065,
      "num_input_tokens_seen": 46331236,
      "step": 828
    },
    {
      "epoch": 1.8440979955456571,
      "loss": 0.9962807893753052,
      "loss_ce": 0.000431208114605397,
      "loss_iou": 0.412109375,
      "loss_num": 0.034423828125,
      "loss_xval": 0.99609375,
      "num_input_tokens_seen": 46331236,
      "step": 828
    },
    {
      "epoch": 1.846325167037862,
      "grad_norm": 19.961835861206055,
      "learning_rate": 1e-06,
      "loss": 1.0033,
      "num_input_tokens_seen": 46387876,
      "step": 829
    },
    {
      "epoch": 1.846325167037862,
      "loss": 0.8641173243522644,
      "loss_ce": 0.0008360765059478581,
      "loss_iou": 0.376953125,
      "loss_num": 0.021728515625,
      "loss_xval": 0.86328125,
      "num_input_tokens_seen": 46387876,
      "step": 829
    },
    {
      "epoch": 1.8485523385300668,
      "grad_norm": 23.06169891357422,
      "learning_rate": 1e-06,
      "loss": 1.2062,
      "num_input_tokens_seen": 46441284,
      "step": 830
    },
    {
      "epoch": 1.8485523385300668,
      "loss": 1.086785078048706,
      "loss_ce": 0.0003593094297684729,
      "loss_iou": 0.447265625,
      "loss_num": 0.037841796875,
      "loss_xval": 1.0859375,
      "num_input_tokens_seen": 46441284,
      "step": 830
    },
    {
      "epoch": 1.8507795100222717,
      "grad_norm": 23.462730407714844,
      "learning_rate": 1e-06,
      "loss": 0.932,
      "num_input_tokens_seen": 46498148,
      "step": 831
    },
    {
      "epoch": 1.8507795100222717,
      "loss": 1.0123460292816162,
      "loss_ce": 0.005265878979116678,
      "loss_iou": 0.41015625,
      "loss_num": 0.037841796875,
      "loss_xval": 1.0078125,
      "num_input_tokens_seen": 46498148,
      "step": 831
    },
    {
      "epoch": 1.8530066815144766,
      "grad_norm": 18.19162368774414,
      "learning_rate": 1e-06,
      "loss": 1.1176,
      "num_input_tokens_seen": 46552656,
      "step": 832
    },
    {
      "epoch": 1.8530066815144766,
      "loss": 1.0369131565093994,
      "loss_ce": 0.0007803026819601655,
      "loss_iou": 0.4296875,
      "loss_num": 0.03466796875,
      "loss_xval": 1.0390625,
      "num_input_tokens_seen": 46552656,
      "step": 832
    },
    {
      "epoch": 1.8552338530066814,
      "grad_norm": 27.384538650512695,
      "learning_rate": 1e-06,
      "loss": 1.327,
      "num_input_tokens_seen": 46609180,
      "step": 833
    },
    {
      "epoch": 1.8552338530066814,
      "loss": 1.4005694389343262,
      "loss_ce": 0.002132008085027337,
      "loss_iou": 0.55078125,
      "loss_num": 0.05859375,
      "loss_xval": 1.3984375,
      "num_input_tokens_seen": 46609180,
      "step": 833
    },
    {
      "epoch": 1.8574610244988863,
      "grad_norm": 17.754493713378906,
      "learning_rate": 1e-06,
      "loss": 0.8976,
      "num_input_tokens_seen": 46667088,
      "step": 834
    },
    {
      "epoch": 1.8574610244988863,
      "loss": 0.7892530560493469,
      "loss_ce": 0.00043467164505273104,
      "loss_iou": 0.326171875,
      "loss_num": 0.0274658203125,
      "loss_xval": 0.7890625,
      "num_input_tokens_seen": 46667088,
      "step": 834
    },
    {
      "epoch": 1.8596881959910914,
      "grad_norm": 22.643587112426758,
      "learning_rate": 1e-06,
      "loss": 1.2626,
      "num_input_tokens_seen": 46722560,
      "step": 835
    },
    {
      "epoch": 1.8596881959910914,
      "loss": 1.1584439277648926,
      "loss_ce": 0.0014614604879170656,
      "loss_iou": 0.4453125,
      "loss_num": 0.05322265625,
      "loss_xval": 1.15625,
      "num_input_tokens_seen": 46722560,
      "step": 835
    },
    {
      "epoch": 1.8619153674832962,
      "grad_norm": 16.647260665893555,
      "learning_rate": 1e-06,
      "loss": 0.9988,
      "num_input_tokens_seen": 46780772,
      "step": 836
    },
    {
      "epoch": 1.8619153674832962,
      "loss": 0.8128501176834106,
      "loss_ce": 0.0010825353674590588,
      "loss_iou": 0.30859375,
      "loss_num": 0.03857421875,
      "loss_xval": 0.8125,
      "num_input_tokens_seen": 46780772,
      "step": 836
    },
    {
      "epoch": 1.864142538975501,
      "grad_norm": 17.087894439697266,
      "learning_rate": 1e-06,
      "loss": 1.3527,
      "num_input_tokens_seen": 46837680,
      "step": 837
    },
    {
      "epoch": 1.864142538975501,
      "loss": 1.4123845100402832,
      "loss_ce": 0.0007634205976501107,
      "loss_iou": 0.56640625,
      "loss_num": 0.05517578125,
      "loss_xval": 1.4140625,
      "num_input_tokens_seen": 46837680,
      "step": 837
    },
    {
      "epoch": 1.8663697104677062,
      "grad_norm": 17.794803619384766,
      "learning_rate": 1e-06,
      "loss": 1.1507,
      "num_input_tokens_seen": 46894204,
      "step": 838
    },
    {
      "epoch": 1.8663697104677062,
      "loss": 1.2315852642059326,
      "loss_ce": 0.015764975920319557,
      "loss_iou": 0.474609375,
      "loss_num": 0.053466796875,
      "loss_xval": 1.21875,
      "num_input_tokens_seen": 46894204,
      "step": 838
    },
    {
      "epoch": 1.868596881959911,
      "grad_norm": 20.756214141845703,
      "learning_rate": 1e-06,
      "loss": 1.0918,
      "num_input_tokens_seen": 46953264,
      "step": 839
    },
    {
      "epoch": 1.868596881959911,
      "loss": 1.1783467531204224,
      "loss_ce": 0.00036822014953941107,
      "loss_iou": 0.478515625,
      "loss_num": 0.044189453125,
      "loss_xval": 1.1796875,
      "num_input_tokens_seen": 46953264,
      "step": 839
    },
    {
      "epoch": 1.8708240534521159,
      "grad_norm": 17.135751724243164,
      "learning_rate": 1e-06,
      "loss": 0.8869,
      "num_input_tokens_seen": 47008696,
      "step": 840
    },
    {
      "epoch": 1.8708240534521159,
      "loss": 0.8081262111663818,
      "loss_ce": 0.000631108705420047,
      "loss_iou": 0.302734375,
      "loss_num": 0.040283203125,
      "loss_xval": 0.80859375,
      "num_input_tokens_seen": 47008696,
      "step": 840
    },
    {
      "epoch": 1.8730512249443207,
      "grad_norm": 24.36521339416504,
      "learning_rate": 1e-06,
      "loss": 1.1049,
      "num_input_tokens_seen": 47066420,
      "step": 841
    },
    {
      "epoch": 1.8730512249443207,
      "loss": 0.8385207653045654,
      "loss_ce": 0.00038598667015321553,
      "loss_iou": 0.373046875,
      "loss_num": 0.01806640625,
      "loss_xval": 0.83984375,
      "num_input_tokens_seen": 47066420,
      "step": 841
    },
    {
      "epoch": 1.8752783964365256,
      "grad_norm": 29.640390396118164,
      "learning_rate": 1e-06,
      "loss": 1.0064,
      "num_input_tokens_seen": 47123592,
      "step": 842
    },
    {
      "epoch": 1.8752783964365256,
      "loss": 0.8564976453781128,
      "loss_ce": 0.0005406375275924802,
      "loss_iou": 0.34765625,
      "loss_num": 0.0322265625,
      "loss_xval": 0.85546875,
      "num_input_tokens_seen": 47123592,
      "step": 842
    },
    {
      "epoch": 1.8775055679287305,
      "grad_norm": 23.356630325317383,
      "learning_rate": 1e-06,
      "loss": 1.1621,
      "num_input_tokens_seen": 47178084,
      "step": 843
    },
    {
      "epoch": 1.8775055679287305,
      "loss": 1.280181646347046,
      "loss_ce": 0.0003964488860219717,
      "loss_iou": 0.55859375,
      "loss_num": 0.032470703125,
      "loss_xval": 1.28125,
      "num_input_tokens_seen": 47178084,
      "step": 843
    },
    {
      "epoch": 1.8797327394209353,
      "grad_norm": 22.576688766479492,
      "learning_rate": 1e-06,
      "loss": 0.9159,
      "num_input_tokens_seen": 47235124,
      "step": 844
    },
    {
      "epoch": 1.8797327394209353,
      "loss": 0.9658552408218384,
      "loss_ce": 0.0005232463008724153,
      "loss_iou": 0.3984375,
      "loss_num": 0.033935546875,
      "loss_xval": 0.96484375,
      "num_input_tokens_seen": 47235124,
      "step": 844
    },
    {
      "epoch": 1.8819599109131402,
      "grad_norm": 16.825740814208984,
      "learning_rate": 1e-06,
      "loss": 1.0477,
      "num_input_tokens_seen": 47291416,
      "step": 845
    },
    {
      "epoch": 1.8819599109131402,
      "loss": 1.1294829845428467,
      "loss_ce": 0.0013091352302581072,
      "loss_iou": 0.453125,
      "loss_num": 0.044189453125,
      "loss_xval": 1.125,
      "num_input_tokens_seen": 47291416,
      "step": 845
    },
    {
      "epoch": 1.884187082405345,
      "grad_norm": 26.097557067871094,
      "learning_rate": 1e-06,
      "loss": 0.9203,
      "num_input_tokens_seen": 47348868,
      "step": 846
    },
    {
      "epoch": 1.884187082405345,
      "loss": 1.0167232751846313,
      "loss_ce": 0.0013424212811514735,
      "loss_iou": 0.443359375,
      "loss_num": 0.025390625,
      "loss_xval": 1.015625,
      "num_input_tokens_seen": 47348868,
      "step": 846
    },
    {
      "epoch": 1.8864142538975501,
      "grad_norm": 25.306549072265625,
      "learning_rate": 1e-06,
      "loss": 1.1072,
      "num_input_tokens_seen": 47405380,
      "step": 847
    },
    {
      "epoch": 1.8864142538975501,
      "loss": 0.9173081517219543,
      "loss_ce": 0.0008042958797886968,
      "loss_iou": 0.359375,
      "loss_num": 0.039794921875,
      "loss_xval": 0.91796875,
      "num_input_tokens_seen": 47405380,
      "step": 847
    },
    {
      "epoch": 1.888641425389755,
      "grad_norm": 14.741415023803711,
      "learning_rate": 1e-06,
      "loss": 0.7495,
      "num_input_tokens_seen": 47460792,
      "step": 848
    },
    {
      "epoch": 1.888641425389755,
      "loss": 0.6206876635551453,
      "loss_ce": 0.0003263273974880576,
      "loss_iou": 0.265625,
      "loss_num": 0.0177001953125,
      "loss_xval": 0.62109375,
      "num_input_tokens_seen": 47460792,
      "step": 848
    },
    {
      "epoch": 1.89086859688196,
      "grad_norm": 16.20488739013672,
      "learning_rate": 1e-06,
      "loss": 0.9648,
      "num_input_tokens_seen": 47518560,
      "step": 849
    },
    {
      "epoch": 1.89086859688196,
      "loss": 0.8242032527923584,
      "loss_ce": 0.010482522659003735,
      "loss_iou": 0.287109375,
      "loss_num": 0.0478515625,
      "loss_xval": 0.8125,
      "num_input_tokens_seen": 47518560,
      "step": 849
    },
    {
      "epoch": 1.893095768374165,
      "grad_norm": 19.45951271057129,
      "learning_rate": 1e-06,
      "loss": 1.2282,
      "num_input_tokens_seen": 47574244,
      "step": 850
    },
    {
      "epoch": 1.893095768374165,
      "loss": 1.3258323669433594,
      "loss_ce": 0.0013694913359358907,
      "loss_iou": 0.515625,
      "loss_num": 0.059326171875,
      "loss_xval": 1.328125,
      "num_input_tokens_seen": 47574244,
      "step": 850
    },
    {
      "epoch": 1.8953229398663698,
      "grad_norm": 18.448118209838867,
      "learning_rate": 1e-06,
      "loss": 1.1335,
      "num_input_tokens_seen": 47627784,
      "step": 851
    },
    {
      "epoch": 1.8953229398663698,
      "loss": 1.3810503482818604,
      "loss_ce": 0.001167478272691369,
      "loss_iou": 0.5625,
      "loss_num": 0.051025390625,
      "loss_xval": 1.3828125,
      "num_input_tokens_seen": 47627784,
      "step": 851
    },
    {
      "epoch": 1.8975501113585747,
      "grad_norm": 28.44631004333496,
      "learning_rate": 1e-06,
      "loss": 0.9616,
      "num_input_tokens_seen": 47685004,
      "step": 852
    },
    {
      "epoch": 1.8975501113585747,
      "loss": 0.8197891116142273,
      "loss_ce": 0.00045313817099668086,
      "loss_iou": 0.33984375,
      "loss_num": 0.02783203125,
      "loss_xval": 0.8203125,
      "num_input_tokens_seen": 47685004,
      "step": 852
    },
    {
      "epoch": 1.8997772828507795,
      "grad_norm": 15.586737632751465,
      "learning_rate": 1e-06,
      "loss": 1.1863,
      "num_input_tokens_seen": 47740520,
      "step": 853
    },
    {
      "epoch": 1.8997772828507795,
      "loss": 1.022131323814392,
      "loss_ce": 0.00040277119842357934,
      "loss_iou": 0.431640625,
      "loss_num": 0.031982421875,
      "loss_xval": 1.0234375,
      "num_input_tokens_seen": 47740520,
      "step": 853
    },
    {
      "epoch": 1.9020044543429844,
      "grad_norm": 35.565025329589844,
      "learning_rate": 1e-06,
      "loss": 1.2452,
      "num_input_tokens_seen": 47793144,
      "step": 854
    },
    {
      "epoch": 1.9020044543429844,
      "loss": 1.1260898113250732,
      "loss_ce": 0.000601442065089941,
      "loss_iou": 0.453125,
      "loss_num": 0.0439453125,
      "loss_xval": 1.125,
      "num_input_tokens_seen": 47793144,
      "step": 854
    },
    {
      "epoch": 1.9042316258351892,
      "grad_norm": 21.033187866210938,
      "learning_rate": 1e-06,
      "loss": 1.1192,
      "num_input_tokens_seen": 47848176,
      "step": 855
    },
    {
      "epoch": 1.9042316258351892,
      "loss": 1.057677984237671,
      "loss_ce": 0.00042704385123215616,
      "loss_iou": 0.43359375,
      "loss_num": 0.037841796875,
      "loss_xval": 1.0546875,
      "num_input_tokens_seen": 47848176,
      "step": 855
    },
    {
      "epoch": 1.906458797327394,
      "grad_norm": 16.94105339050293,
      "learning_rate": 1e-06,
      "loss": 1.0397,
      "num_input_tokens_seen": 47901820,
      "step": 856
    },
    {
      "epoch": 1.906458797327394,
      "loss": 0.8765525817871094,
      "loss_ce": 0.0005760163185186684,
      "loss_iou": 0.37890625,
      "loss_num": 0.02392578125,
      "loss_xval": 0.875,
      "num_input_tokens_seen": 47901820,
      "step": 856
    },
    {
      "epoch": 1.908685968819599,
      "grad_norm": 22.052213668823242,
      "learning_rate": 1e-06,
      "loss": 1.058,
      "num_input_tokens_seen": 47958540,
      "step": 857
    },
    {
      "epoch": 1.908685968819599,
      "loss": 0.9184067845344543,
      "loss_ce": 0.0004380404716357589,
      "loss_iou": 0.376953125,
      "loss_num": 0.033203125,
      "loss_xval": 0.91796875,
      "num_input_tokens_seen": 47958540,
      "step": 857
    },
    {
      "epoch": 1.910913140311804,
      "grad_norm": 18.550006866455078,
      "learning_rate": 1e-06,
      "loss": 0.8463,
      "num_input_tokens_seen": 48015240,
      "step": 858
    },
    {
      "epoch": 1.910913140311804,
      "loss": 0.8902982473373413,
      "loss_ce": 0.0004057343176100403,
      "loss_iou": 0.330078125,
      "loss_num": 0.045654296875,
      "loss_xval": 0.890625,
      "num_input_tokens_seen": 48015240,
      "step": 858
    },
    {
      "epoch": 1.913140311804009,
      "grad_norm": 19.243886947631836,
      "learning_rate": 1e-06,
      "loss": 0.8471,
      "num_input_tokens_seen": 48071956,
      "step": 859
    },
    {
      "epoch": 1.913140311804009,
      "loss": 0.8775541186332703,
      "loss_ce": 0.0010892475256696343,
      "loss_iou": 0.36328125,
      "loss_num": 0.02978515625,
      "loss_xval": 0.875,
      "num_input_tokens_seen": 48071956,
      "step": 859
    },
    {
      "epoch": 1.9153674832962138,
      "grad_norm": 43.45414733886719,
      "learning_rate": 1e-06,
      "loss": 0.8823,
      "num_input_tokens_seen": 48125932,
      "step": 860
    },
    {
      "epoch": 1.9153674832962138,
      "loss": 0.9675887823104858,
      "loss_ce": 0.0005477914237417281,
      "loss_iou": 0.421875,
      "loss_num": 0.0245361328125,
      "loss_xval": 0.96875,
      "num_input_tokens_seen": 48125932,
      "step": 860
    },
    {
      "epoch": 1.9175946547884188,
      "grad_norm": 17.62973976135254,
      "learning_rate": 1e-06,
      "loss": 0.8766,
      "num_input_tokens_seen": 48183412,
      "step": 861
    },
    {
      "epoch": 1.9175946547884188,
      "loss": 1.0546504259109497,
      "loss_ce": 0.0006953147239983082,
      "loss_iou": 0.439453125,
      "loss_num": 0.034912109375,
      "loss_xval": 1.0546875,
      "num_input_tokens_seen": 48183412,
      "step": 861
    },
    {
      "epoch": 1.9198218262806237,
      "grad_norm": 15.39229965209961,
      "learning_rate": 1e-06,
      "loss": 1.0492,
      "num_input_tokens_seen": 48237972,
      "step": 862
    },
    {
      "epoch": 1.9198218262806237,
      "loss": 1.0228043794631958,
      "loss_ce": 0.0005876163486391306,
      "loss_iou": 0.396484375,
      "loss_num": 0.046142578125,
      "loss_xval": 1.0234375,
      "num_input_tokens_seen": 48237972,
      "step": 862
    },
    {
      "epoch": 1.9220489977728286,
      "grad_norm": 35.672637939453125,
      "learning_rate": 1e-06,
      "loss": 1.1325,
      "num_input_tokens_seen": 48294312,
      "step": 863
    },
    {
      "epoch": 1.9220489977728286,
      "loss": 1.1503536701202393,
      "loss_ce": 0.00045127171324566007,
      "loss_iou": 0.474609375,
      "loss_num": 0.039794921875,
      "loss_xval": 1.1484375,
      "num_input_tokens_seen": 48294312,
      "step": 863
    },
    {
      "epoch": 1.9242761692650334,
      "grad_norm": 18.644657135009766,
      "learning_rate": 1e-06,
      "loss": 1.1068,
      "num_input_tokens_seen": 48352424,
      "step": 864
    },
    {
      "epoch": 1.9242761692650334,
      "loss": 0.9790323972702026,
      "loss_ce": 0.001493359450250864,
      "loss_iou": 0.3984375,
      "loss_num": 0.0361328125,
      "loss_xval": 0.9765625,
      "num_input_tokens_seen": 48352424,
      "step": 864
    },
    {
      "epoch": 1.9265033407572383,
      "grad_norm": 19.904115676879883,
      "learning_rate": 1e-06,
      "loss": 1.2304,
      "num_input_tokens_seen": 48407952,
      "step": 865
    },
    {
      "epoch": 1.9265033407572383,
      "loss": 1.5032316446304321,
      "loss_ce": 0.0037199431098997593,
      "loss_iou": 0.5703125,
      "loss_num": 0.072265625,
      "loss_xval": 1.5,
      "num_input_tokens_seen": 48407952,
      "step": 865
    },
    {
      "epoch": 1.9287305122494431,
      "grad_norm": 16.542858123779297,
      "learning_rate": 1e-06,
      "loss": 0.8461,
      "num_input_tokens_seen": 48463412,
      "step": 866
    },
    {
      "epoch": 1.9287305122494431,
      "loss": 0.9278074502944946,
      "loss_ce": 0.0025144873652607203,
      "loss_iou": 0.36328125,
      "loss_num": 0.039306640625,
      "loss_xval": 0.92578125,
      "num_input_tokens_seen": 48463412,
      "step": 866
    },
    {
      "epoch": 1.930957683741648,
      "grad_norm": 28.734394073486328,
      "learning_rate": 1e-06,
      "loss": 1.206,
      "num_input_tokens_seen": 48521584,
      "step": 867
    },
    {
      "epoch": 1.930957683741648,
      "loss": 1.0615025758743286,
      "loss_ce": 0.00046744622522965074,
      "loss_iou": 0.45703125,
      "loss_num": 0.0291748046875,
      "loss_xval": 1.0625,
      "num_input_tokens_seen": 48521584,
      "step": 867
    },
    {
      "epoch": 1.9331848552338529,
      "grad_norm": 16.990293502807617,
      "learning_rate": 1e-06,
      "loss": 1.0472,
      "num_input_tokens_seen": 48575076,
      "step": 868
    },
    {
      "epoch": 1.9331848552338529,
      "loss": 1.2131456136703491,
      "loss_ce": 0.0012315567582845688,
      "loss_iou": 0.482421875,
      "loss_num": 0.049560546875,
      "loss_xval": 1.2109375,
      "num_input_tokens_seen": 48575076,
      "step": 868
    },
    {
      "epoch": 1.935412026726058,
      "grad_norm": 20.22574234008789,
      "learning_rate": 1e-06,
      "loss": 1.0024,
      "num_input_tokens_seen": 48632592,
      "step": 869
    },
    {
      "epoch": 1.935412026726058,
      "loss": 1.0861210823059082,
      "loss_ce": 0.0006718788645230234,
      "loss_iou": 0.474609375,
      "loss_num": 0.0272216796875,
      "loss_xval": 1.0859375,
      "num_input_tokens_seen": 48632592,
      "step": 869
    },
    {
      "epoch": 1.9376391982182628,
      "grad_norm": 16.520029067993164,
      "learning_rate": 1e-06,
      "loss": 0.9542,
      "num_input_tokens_seen": 48687552,
      "step": 870
    },
    {
      "epoch": 1.9376391982182628,
      "loss": 1.0474047660827637,
      "loss_ce": 0.0005297240568324924,
      "loss_iou": 0.46484375,
      "loss_num": 0.023193359375,
      "loss_xval": 1.046875,
      "num_input_tokens_seen": 48687552,
      "step": 870
    },
    {
      "epoch": 1.9398663697104677,
      "grad_norm": 20.480531692504883,
      "learning_rate": 1e-06,
      "loss": 0.7512,
      "num_input_tokens_seen": 48742604,
      "step": 871
    },
    {
      "epoch": 1.9398663697104677,
      "loss": 0.876989483833313,
      "loss_ce": 0.0005246583605185151,
      "loss_iou": 0.37890625,
      "loss_num": 0.024169921875,
      "loss_xval": 0.875,
      "num_input_tokens_seen": 48742604,
      "step": 871
    },
    {
      "epoch": 1.9420935412026727,
      "grad_norm": 24.84157371520996,
      "learning_rate": 1e-06,
      "loss": 0.8841,
      "num_input_tokens_seen": 48800660,
      "step": 872
    },
    {
      "epoch": 1.9420935412026727,
      "loss": 0.9676753282546997,
      "loss_ce": 0.0008784872479736805,
      "loss_iou": 0.40234375,
      "loss_num": 0.03271484375,
      "loss_xval": 0.96875,
      "num_input_tokens_seen": 48800660,
      "step": 872
    },
    {
      "epoch": 1.9443207126948776,
      "grad_norm": 22.539447784423828,
      "learning_rate": 1e-06,
      "loss": 1.1714,
      "num_input_tokens_seen": 48855408,
      "step": 873
    },
    {
      "epoch": 1.9443207126948776,
      "loss": 1.1532479524612427,
      "loss_ce": 0.0004159705131314695,
      "loss_iou": 0.515625,
      "loss_num": 0.0247802734375,
      "loss_xval": 1.15625,
      "num_input_tokens_seen": 48855408,
      "step": 873
    },
    {
      "epoch": 1.9465478841870825,
      "grad_norm": 41.367530822753906,
      "learning_rate": 1e-06,
      "loss": 1.2651,
      "num_input_tokens_seen": 48910668,
      "step": 874
    },
    {
      "epoch": 1.9465478841870825,
      "loss": 1.3369231224060059,
      "loss_ce": 0.0014739749021828175,
      "loss_iou": 0.5625,
      "loss_num": 0.041259765625,
      "loss_xval": 1.3359375,
      "num_input_tokens_seen": 48910668,
      "step": 874
    },
    {
      "epoch": 1.9487750556792873,
      "grad_norm": 15.942052841186523,
      "learning_rate": 1e-06,
      "loss": 0.9964,
      "num_input_tokens_seen": 48967356,
      "step": 875
    },
    {
      "epoch": 1.9487750556792873,
      "loss": 1.1262295246124268,
      "loss_ce": 0.0007413048297166824,
      "loss_iou": 0.427734375,
      "loss_num": 0.053955078125,
      "loss_xval": 1.125,
      "num_input_tokens_seen": 48967356,
      "step": 875
    },
    {
      "epoch": 1.9510022271714922,
      "grad_norm": 45.24485778808594,
      "learning_rate": 1e-06,
      "loss": 1.0602,
      "num_input_tokens_seen": 49023520,
      "step": 876
    },
    {
      "epoch": 1.9510022271714922,
      "loss": 1.150200366973877,
      "loss_ce": 0.0007863342761993408,
      "loss_iou": 0.5,
      "loss_num": 0.029296875,
      "loss_xval": 1.1484375,
      "num_input_tokens_seen": 49023520,
      "step": 876
    },
    {
      "epoch": 1.953229398663697,
      "grad_norm": 24.16380500793457,
      "learning_rate": 1e-06,
      "loss": 0.9545,
      "num_input_tokens_seen": 49079940,
      "step": 877
    },
    {
      "epoch": 1.953229398663697,
      "loss": 0.8877462148666382,
      "loss_ce": 0.0005392197053879499,
      "loss_iou": 0.37109375,
      "loss_num": 0.0289306640625,
      "loss_xval": 0.88671875,
      "num_input_tokens_seen": 49079940,
      "step": 877
    },
    {
      "epoch": 1.955456570155902,
      "grad_norm": 22.12382698059082,
      "learning_rate": 1e-06,
      "loss": 1.1521,
      "num_input_tokens_seen": 49138012,
      "step": 878
    },
    {
      "epoch": 1.955456570155902,
      "loss": 1.5235247611999512,
      "loss_ce": 0.0017962402198463678,
      "loss_iou": 0.59765625,
      "loss_num": 0.06494140625,
      "loss_xval": 1.5234375,
      "num_input_tokens_seen": 49138012,
      "step": 878
    },
    {
      "epoch": 1.9576837416481068,
      "grad_norm": 18.208831787109375,
      "learning_rate": 1e-06,
      "loss": 0.9084,
      "num_input_tokens_seen": 49194864,
      "step": 879
    },
    {
      "epoch": 1.9576837416481068,
      "loss": 0.9498248100280762,
      "loss_ce": 0.0003619292110670358,
      "loss_iou": 0.3671875,
      "loss_num": 0.042724609375,
      "loss_xval": 0.94921875,
      "num_input_tokens_seen": 49194864,
      "step": 879
    },
    {
      "epoch": 1.9599109131403119,
      "grad_norm": 16.236581802368164,
      "learning_rate": 1e-06,
      "loss": 0.9203,
      "num_input_tokens_seen": 49251060,
      "step": 880
    },
    {
      "epoch": 1.9599109131403119,
      "loss": 0.9726754426956177,
      "loss_ce": 0.0007516111945733428,
      "loss_iou": 0.396484375,
      "loss_num": 0.03564453125,
      "loss_xval": 0.97265625,
      "num_input_tokens_seen": 49251060,
      "step": 880
    },
    {
      "epoch": 1.9621380846325167,
      "grad_norm": 23.136600494384766,
      "learning_rate": 1e-06,
      "loss": 1.1897,
      "num_input_tokens_seen": 49309208,
      "step": 881
    },
    {
      "epoch": 1.9621380846325167,
      "loss": 1.0319995880126953,
      "loss_ce": 0.002946855966001749,
      "loss_iou": 0.43359375,
      "loss_num": 0.032470703125,
      "loss_xval": 1.03125,
      "num_input_tokens_seen": 49309208,
      "step": 881
    },
    {
      "epoch": 1.9643652561247216,
      "grad_norm": 54.83725357055664,
      "learning_rate": 1e-06,
      "loss": 1.0684,
      "num_input_tokens_seen": 49365352,
      "step": 882
    },
    {
      "epoch": 1.9643652561247216,
      "loss": 1.0005229711532593,
      "loss_ce": 0.0005229845410212874,
      "loss_iou": 0.39453125,
      "loss_num": 0.041748046875,
      "loss_xval": 1.0,
      "num_input_tokens_seen": 49365352,
      "step": 882
    },
    {
      "epoch": 1.9665924276169267,
      "grad_norm": 20.81184959411621,
      "learning_rate": 1e-06,
      "loss": 1.3279,
      "num_input_tokens_seen": 49420316,
      "step": 883
    },
    {
      "epoch": 1.9665924276169267,
      "loss": 1.0947761535644531,
      "loss_ce": 0.0005379515350796282,
      "loss_iou": 0.451171875,
      "loss_num": 0.03857421875,
      "loss_xval": 1.09375,
      "num_input_tokens_seen": 49420316,
      "step": 883
    },
    {
      "epoch": 1.9688195991091315,
      "grad_norm": 24.571720123291016,
      "learning_rate": 1e-06,
      "loss": 0.9046,
      "num_input_tokens_seen": 49478252,
      "step": 884
    },
    {
      "epoch": 1.9688195991091315,
      "loss": 1.0495492219924927,
      "loss_ce": 0.0007211226620711386,
      "loss_iou": 0.41796875,
      "loss_num": 0.042236328125,
      "loss_xval": 1.046875,
      "num_input_tokens_seen": 49478252,
      "step": 884
    },
    {
      "epoch": 1.9710467706013364,
      "grad_norm": 17.04405403137207,
      "learning_rate": 1e-06,
      "loss": 0.9426,
      "num_input_tokens_seen": 49533784,
      "step": 885
    },
    {
      "epoch": 1.9710467706013364,
      "loss": 0.8944860696792603,
      "loss_ce": 0.0009313831687904894,
      "loss_iou": 0.361328125,
      "loss_num": 0.033935546875,
      "loss_xval": 0.89453125,
      "num_input_tokens_seen": 49533784,
      "step": 885
    },
    {
      "epoch": 1.9732739420935412,
      "grad_norm": 21.639575958251953,
      "learning_rate": 1e-06,
      "loss": 1.0049,
      "num_input_tokens_seen": 49589404,
      "step": 886
    },
    {
      "epoch": 1.9732739420935412,
      "loss": 0.9363880157470703,
      "loss_ce": 0.0008411717135459185,
      "loss_iou": 0.369140625,
      "loss_num": 0.038818359375,
      "loss_xval": 0.9375,
      "num_input_tokens_seen": 49589404,
      "step": 886
    },
    {
      "epoch": 1.975501113585746,
      "grad_norm": 15.142632484436035,
      "learning_rate": 1e-06,
      "loss": 1.0436,
      "num_input_tokens_seen": 49646820,
      "step": 887
    },
    {
      "epoch": 1.975501113585746,
      "loss": 0.9690762758255005,
      "loss_ce": 0.010580191388726234,
      "loss_iou": 0.37890625,
      "loss_num": 0.039794921875,
      "loss_xval": 0.95703125,
      "num_input_tokens_seen": 49646820,
      "step": 887
    },
    {
      "epoch": 1.977728285077951,
      "grad_norm": 132.77593994140625,
      "learning_rate": 1e-06,
      "loss": 1.0238,
      "num_input_tokens_seen": 49702832,
      "step": 888
    },
    {
      "epoch": 1.977728285077951,
      "loss": 1.0730714797973633,
      "loss_ce": 0.0008057774393819273,
      "loss_iou": 0.44140625,
      "loss_num": 0.037109375,
      "loss_xval": 1.0703125,
      "num_input_tokens_seen": 49702832,
      "step": 888
    },
    {
      "epoch": 1.9799554565701558,
      "grad_norm": 46.88450622558594,
      "learning_rate": 1e-06,
      "loss": 1.0213,
      "num_input_tokens_seen": 49759608,
      "step": 889
    },
    {
      "epoch": 1.9799554565701558,
      "loss": 0.5785812139511108,
      "loss_ce": 0.0004562578978948295,
      "loss_iou": 0.23828125,
      "loss_num": 0.020263671875,
      "loss_xval": 0.578125,
      "num_input_tokens_seen": 49759608,
      "step": 889
    },
    {
      "epoch": 1.9821826280623607,
      "grad_norm": 13.415266990661621,
      "learning_rate": 1e-06,
      "loss": 0.6749,
      "num_input_tokens_seen": 49816076,
      "step": 890
    },
    {
      "epoch": 1.9821826280623607,
      "loss": 0.6280744075775146,
      "loss_ce": 0.00038887892151251435,
      "loss_iou": 0.24609375,
      "loss_num": 0.0269775390625,
      "loss_xval": 0.62890625,
      "num_input_tokens_seen": 49816076,
      "step": 890
    },
    {
      "epoch": 1.9844097995545658,
      "grad_norm": 15.064915657043457,
      "learning_rate": 1e-06,
      "loss": 0.9421,
      "num_input_tokens_seen": 49873896,
      "step": 891
    },
    {
      "epoch": 1.9844097995545658,
      "loss": 0.9301207065582275,
      "loss_ce": 0.00043319491669535637,
      "loss_iou": 0.388671875,
      "loss_num": 0.0302734375,
      "loss_xval": 0.9296875,
      "num_input_tokens_seen": 49873896,
      "step": 891
    },
    {
      "epoch": 1.9866369710467706,
      "grad_norm": 19.410207748413086,
      "learning_rate": 1e-06,
      "loss": 0.8304,
      "num_input_tokens_seen": 49929944,
      "step": 892
    },
    {
      "epoch": 1.9866369710467706,
      "loss": 0.8065379858016968,
      "loss_ce": 0.0003856243856716901,
      "loss_iou": 0.345703125,
      "loss_num": 0.0230712890625,
      "loss_xval": 0.8046875,
      "num_input_tokens_seen": 49929944,
      "step": 892
    },
    {
      "epoch": 1.9888641425389755,
      "grad_norm": 18.616121292114258,
      "learning_rate": 1e-06,
      "loss": 0.9924,
      "num_input_tokens_seen": 49987336,
      "step": 893
    },
    {
      "epoch": 1.9888641425389755,
      "loss": 1.0283457040786743,
      "loss_ce": 0.0005136135150678456,
      "loss_iou": 0.392578125,
      "loss_num": 0.048583984375,
      "loss_xval": 1.03125,
      "num_input_tokens_seen": 49987336,
      "step": 893
    },
    {
      "epoch": 1.9910913140311806,
      "grad_norm": 20.091445922851562,
      "learning_rate": 1e-06,
      "loss": 1.0854,
      "num_input_tokens_seen": 50045032,
      "step": 894
    },
    {
      "epoch": 1.9910913140311806,
      "loss": 0.9809556007385254,
      "loss_ce": 0.000730981701053679,
      "loss_iou": 0.396484375,
      "loss_num": 0.037353515625,
      "loss_xval": 0.98046875,
      "num_input_tokens_seen": 50045032,
      "step": 894
    },
    {
      "epoch": 1.9933184855233854,
      "grad_norm": 19.74848175048828,
      "learning_rate": 1e-06,
      "loss": 1.0705,
      "num_input_tokens_seen": 50100472,
      "step": 895
    },
    {
      "epoch": 1.9933184855233854,
      "loss": 0.8635627031326294,
      "loss_ce": 0.0012580410111695528,
      "loss_iou": 0.326171875,
      "loss_num": 0.04248046875,
      "loss_xval": 0.86328125,
      "num_input_tokens_seen": 50100472,
      "step": 895
    },
    {
      "epoch": 1.9955456570155903,
      "grad_norm": 17.26140022277832,
      "learning_rate": 1e-06,
      "loss": 0.9529,
      "num_input_tokens_seen": 50154436,
      "step": 896
    },
    {
      "epoch": 1.9955456570155903,
      "loss": 0.9841878414154053,
      "loss_ce": 0.0006672587478533387,
      "loss_iou": 0.38671875,
      "loss_num": 0.041748046875,
      "loss_xval": 0.984375,
      "num_input_tokens_seen": 50154436,
      "step": 896
    },
    {
      "epoch": 1.9977728285077951,
      "grad_norm": 20.126981735229492,
      "learning_rate": 1e-06,
      "loss": 1.0437,
      "num_input_tokens_seen": 50211096,
      "step": 897
    },
    {
      "epoch": 1.9977728285077951,
      "loss": 1.0043690204620361,
      "loss_ce": 0.0004628162132576108,
      "loss_iou": 0.408203125,
      "loss_num": 0.037841796875,
      "loss_xval": 1.0,
      "num_input_tokens_seen": 50211096,
      "step": 897
    },
    {
      "epoch": 2.0,
      "grad_norm": 23.061552047729492,
      "learning_rate": 1e-06,
      "loss": 1.0086,
      "num_input_tokens_seen": 50270028,
      "step": 898
    },
    {
      "epoch": 2.0,
      "loss": 0.6891332864761353,
      "loss_ce": 0.0004125749983359128,
      "loss_iou": 0.294921875,
      "loss_num": 0.0198974609375,
      "loss_xval": 0.6875,
      "num_input_tokens_seen": 50270028,
      "step": 898
    },
    {
      "epoch": 2.002227171492205,
      "grad_norm": 18.92450523376465,
      "learning_rate": 1e-06,
      "loss": 1.1896,
      "num_input_tokens_seen": 50325468,
      "step": 899
    },
    {
      "epoch": 2.002227171492205,
      "loss": 1.5556280612945557,
      "loss_ce": 0.0019170554587617517,
      "loss_iou": 0.578125,
      "loss_num": 0.080078125,
      "loss_xval": 1.5546875,
      "num_input_tokens_seen": 50325468,
      "step": 899
    },
    {
      "epoch": 2.0044543429844097,
      "grad_norm": 20.99635887145996,
      "learning_rate": 1e-06,
      "loss": 0.8291,
      "num_input_tokens_seen": 50384784,
      "step": 900
    },
    {
      "epoch": 2.0044543429844097,
      "loss": 0.8486037254333496,
      "loss_ce": 0.0004591494216583669,
      "loss_iou": 0.341796875,
      "loss_num": 0.033447265625,
      "loss_xval": 0.84765625,
      "num_input_tokens_seen": 50384784,
      "step": 900
    },
    {
      "epoch": 2.0066815144766146,
      "grad_norm": 31.64121437072754,
      "learning_rate": 1e-06,
      "loss": 0.8485,
      "num_input_tokens_seen": 50440300,
      "step": 901
    },
    {
      "epoch": 2.0066815144766146,
      "loss": 0.8760842084884644,
      "loss_ce": 0.0008400966180488467,
      "loss_iou": 0.35546875,
      "loss_num": 0.033447265625,
      "loss_xval": 0.875,
      "num_input_tokens_seen": 50440300,
      "step": 901
    },
    {
      "epoch": 2.0089086859688194,
      "grad_norm": 17.200613021850586,
      "learning_rate": 1e-06,
      "loss": 1.2305,
      "num_input_tokens_seen": 50497772,
      "step": 902
    },
    {
      "epoch": 2.0089086859688194,
      "loss": 1.1643800735473633,
      "loss_ce": 0.0008058917010203004,
      "loss_iou": 0.46875,
      "loss_num": 0.044677734375,
      "loss_xval": 1.1640625,
      "num_input_tokens_seen": 50497772,
      "step": 902
    },
    {
      "epoch": 2.0111358574610243,
      "grad_norm": 16.72941780090332,
      "learning_rate": 1e-06,
      "loss": 0.7154,
      "num_input_tokens_seen": 50555944,
      "step": 903
    },
    {
      "epoch": 2.0111358574610243,
      "loss": 0.8082038164138794,
      "loss_ce": 0.0008307406678795815,
      "loss_iou": 0.302734375,
      "loss_num": 0.040283203125,
      "loss_xval": 0.80859375,
      "num_input_tokens_seen": 50555944,
      "step": 903
    },
    {
      "epoch": 2.0133630289532296,
      "grad_norm": 20.396434783935547,
      "learning_rate": 1e-06,
      "loss": 1.1836,
      "num_input_tokens_seen": 50613104,
      "step": 904
    },
    {
      "epoch": 2.0133630289532296,
      "loss": 1.1902202367782593,
      "loss_ce": 0.0012554043205454946,
      "loss_iou": 0.466796875,
      "loss_num": 0.051025390625,
      "loss_xval": 1.1875,
      "num_input_tokens_seen": 50613104,
      "step": 904
    },
    {
      "epoch": 2.0155902004454345,
      "grad_norm": 21.122411727905273,
      "learning_rate": 1e-06,
      "loss": 0.9534,
      "num_input_tokens_seen": 50668660,
      "step": 905
    },
    {
      "epoch": 2.0155902004454345,
      "loss": 0.9574775695800781,
      "loss_ce": 0.00044635249651037157,
      "loss_iou": 0.390625,
      "loss_num": 0.035400390625,
      "loss_xval": 0.95703125,
      "num_input_tokens_seen": 50668660,
      "step": 905
    },
    {
      "epoch": 2.0178173719376393,
      "grad_norm": 12.18447494506836,
      "learning_rate": 1e-06,
      "loss": 1.2904,
      "num_input_tokens_seen": 50721156,
      "step": 906
    },
    {
      "epoch": 2.0178173719376393,
      "loss": 0.806514322757721,
      "loss_ce": 0.0004840257461182773,
      "loss_iou": 0.333984375,
      "loss_num": 0.0277099609375,
      "loss_xval": 0.8046875,
      "num_input_tokens_seen": 50721156,
      "step": 906
    },
    {
      "epoch": 2.020044543429844,
      "grad_norm": 20.636756896972656,
      "learning_rate": 1e-06,
      "loss": 1.0138,
      "num_input_tokens_seen": 50777508,
      "step": 907
    },
    {
      "epoch": 2.020044543429844,
      "loss": 0.9786741733551025,
      "loss_ce": 0.0006468579522334039,
      "loss_iou": 0.41796875,
      "loss_num": 0.02880859375,
      "loss_xval": 0.9765625,
      "num_input_tokens_seen": 50777508,
      "step": 907
    },
    {
      "epoch": 2.022271714922049,
      "grad_norm": 41.294925689697266,
      "learning_rate": 1e-06,
      "loss": 1.1918,
      "num_input_tokens_seen": 50833956,
      "step": 908
    },
    {
      "epoch": 2.022271714922049,
      "loss": 0.883208692073822,
      "loss_ce": 0.000396185350837186,
      "loss_iou": 0.322265625,
      "loss_num": 0.047607421875,
      "loss_xval": 0.8828125,
      "num_input_tokens_seen": 50833956,
      "step": 908
    },
    {
      "epoch": 2.024498886414254,
      "grad_norm": 18.029842376708984,
      "learning_rate": 1e-06,
      "loss": 1.185,
      "num_input_tokens_seen": 50892188,
      "step": 909
    },
    {
      "epoch": 2.024498886414254,
      "loss": 1.0920612812042236,
      "loss_ce": 0.0005085701122879982,
      "loss_iou": 0.451171875,
      "loss_num": 0.038330078125,
      "loss_xval": 1.09375,
      "num_input_tokens_seen": 50892188,
      "step": 909
    },
    {
      "epoch": 2.0267260579064588,
      "grad_norm": 17.75556755065918,
      "learning_rate": 1e-06,
      "loss": 0.9455,
      "num_input_tokens_seen": 50948976,
      "step": 910
    },
    {
      "epoch": 2.0267260579064588,
      "loss": 0.8729127645492554,
      "loss_ce": 0.0005983302253298461,
      "loss_iou": 0.37109375,
      "loss_num": 0.0264892578125,
      "loss_xval": 0.87109375,
      "num_input_tokens_seen": 50948976,
      "step": 910
    },
    {
      "epoch": 2.0289532293986636,
      "grad_norm": 22.336694717407227,
      "learning_rate": 1e-06,
      "loss": 0.811,
      "num_input_tokens_seen": 51005372,
      "step": 911
    },
    {
      "epoch": 2.0289532293986636,
      "loss": 0.8184252977371216,
      "loss_ce": 0.0010424721986055374,
      "loss_iou": 0.33203125,
      "loss_num": 0.0306396484375,
      "loss_xval": 0.81640625,
      "num_input_tokens_seen": 51005372,
      "step": 911
    },
    {
      "epoch": 2.0311804008908685,
      "grad_norm": 24.3977108001709,
      "learning_rate": 1e-06,
      "loss": 0.7044,
      "num_input_tokens_seen": 51060608,
      "step": 912
    },
    {
      "epoch": 2.0311804008908685,
      "loss": 0.8316826820373535,
      "loss_ce": 0.00038386776577681303,
      "loss_iou": 0.349609375,
      "loss_num": 0.0267333984375,
      "loss_xval": 0.83203125,
      "num_input_tokens_seen": 51060608,
      "step": 912
    },
    {
      "epoch": 2.0334075723830733,
      "grad_norm": 21.548065185546875,
      "learning_rate": 1e-06,
      "loss": 0.9748,
      "num_input_tokens_seen": 51119092,
      "step": 913
    },
    {
      "epoch": 2.0334075723830733,
      "loss": 0.853911280632019,
      "loss_ce": 0.0003957064473070204,
      "loss_iou": 0.34765625,
      "loss_num": 0.031494140625,
      "loss_xval": 0.8515625,
      "num_input_tokens_seen": 51119092,
      "step": 913
    },
    {
      "epoch": 2.035634743875278,
      "grad_norm": 17.78352928161621,
      "learning_rate": 1e-06,
      "loss": 0.8966,
      "num_input_tokens_seen": 51174088,
      "step": 914
    },
    {
      "epoch": 2.035634743875278,
      "loss": 0.6430214047431946,
      "loss_ce": 0.0004432816640473902,
      "loss_iou": 0.275390625,
      "loss_num": 0.0184326171875,
      "loss_xval": 0.640625,
      "num_input_tokens_seen": 51174088,
      "step": 914
    },
    {
      "epoch": 2.0378619153674835,
      "grad_norm": 18.09092903137207,
      "learning_rate": 1e-06,
      "loss": 1.0737,
      "num_input_tokens_seen": 51231032,
      "step": 915
    },
    {
      "epoch": 2.0378619153674835,
      "loss": 1.0431830883026123,
      "loss_ce": 0.0007026625098660588,
      "loss_iou": 0.412109375,
      "loss_num": 0.0439453125,
      "loss_xval": 1.0390625,
      "num_input_tokens_seen": 51231032,
      "step": 915
    },
    {
      "epoch": 2.0400890868596884,
      "grad_norm": 21.492124557495117,
      "learning_rate": 1e-06,
      "loss": 0.849,
      "num_input_tokens_seen": 51288688,
      "step": 916
    },
    {
      "epoch": 2.0400890868596884,
      "loss": 0.8285926580429077,
      "loss_ce": 0.0004676440730690956,
      "loss_iou": 0.3515625,
      "loss_num": 0.02490234375,
      "loss_xval": 0.828125,
      "num_input_tokens_seen": 51288688,
      "step": 916
    },
    {
      "epoch": 2.0423162583518932,
      "grad_norm": 24.50774574279785,
      "learning_rate": 1e-06,
      "loss": 1.1289,
      "num_input_tokens_seen": 51343280,
      "step": 917
    },
    {
      "epoch": 2.0423162583518932,
      "loss": 1.1621774435043335,
      "loss_ce": 0.0005563499871641397,
      "loss_iou": 0.482421875,
      "loss_num": 0.0390625,
      "loss_xval": 1.1640625,
      "num_input_tokens_seen": 51343280,
      "step": 917
    },
    {
      "epoch": 2.044543429844098,
      "grad_norm": 23.351022720336914,
      "learning_rate": 1e-06,
      "loss": 0.8928,
      "num_input_tokens_seen": 51399164,
      "step": 918
    },
    {
      "epoch": 2.044543429844098,
      "loss": 0.76546311378479,
      "loss_ce": 0.0005705538205802441,
      "loss_iou": 0.31640625,
      "loss_num": 0.0263671875,
      "loss_xval": 0.765625,
      "num_input_tokens_seen": 51399164,
      "step": 918
    },
    {
      "epoch": 2.046770601336303,
      "grad_norm": 23.73122215270996,
      "learning_rate": 1e-06,
      "loss": 1.2118,
      "num_input_tokens_seen": 51454772,
      "step": 919
    },
    {
      "epoch": 2.046770601336303,
      "loss": 1.1581388711929321,
      "loss_ce": 0.0006681361701339483,
      "loss_iou": 0.5078125,
      "loss_num": 0.0281982421875,
      "loss_xval": 1.15625,
      "num_input_tokens_seen": 51454772,
      "step": 919
    },
    {
      "epoch": 2.048997772828508,
      "grad_norm": 18.12936782836914,
      "learning_rate": 1e-06,
      "loss": 1.1355,
      "num_input_tokens_seen": 51508836,
      "step": 920
    },
    {
      "epoch": 2.048997772828508,
      "loss": 0.9356842041015625,
      "loss_ce": 0.0016021885676309466,
      "loss_iou": 0.3359375,
      "loss_num": 0.05224609375,
      "loss_xval": 0.93359375,
      "num_input_tokens_seen": 51508836,
      "step": 920
    },
    {
      "epoch": 2.0512249443207127,
      "grad_norm": 18.033832550048828,
      "learning_rate": 1e-06,
      "loss": 0.897,
      "num_input_tokens_seen": 51562932,
      "step": 921
    },
    {
      "epoch": 2.0512249443207127,
      "loss": 0.9213240742683411,
      "loss_ce": 0.001158121507614851,
      "loss_iou": 0.4140625,
      "loss_num": 0.0185546875,
      "loss_xval": 0.921875,
      "num_input_tokens_seen": 51562932,
      "step": 921
    },
    {
      "epoch": 2.0534521158129175,
      "grad_norm": 22.288450241088867,
      "learning_rate": 1e-06,
      "loss": 1.1296,
      "num_input_tokens_seen": 51617208,
      "step": 922
    },
    {
      "epoch": 2.0534521158129175,
      "loss": 1.1455841064453125,
      "loss_ce": 0.017166122794151306,
      "loss_iou": 0.447265625,
      "loss_num": 0.046630859375,
      "loss_xval": 1.125,
      "num_input_tokens_seen": 51617208,
      "step": 922
    },
    {
      "epoch": 2.0556792873051224,
      "grad_norm": 13.78810977935791,
      "learning_rate": 1e-06,
      "loss": 1.0396,
      "num_input_tokens_seen": 51669524,
      "step": 923
    },
    {
      "epoch": 2.0556792873051224,
      "loss": 1.045444130897522,
      "loss_ce": 0.0005222847685217857,
      "loss_iou": 0.421875,
      "loss_num": 0.04052734375,
      "loss_xval": 1.046875,
      "num_input_tokens_seen": 51669524,
      "step": 923
    },
    {
      "epoch": 2.0579064587973273,
      "grad_norm": 15.79564380645752,
      "learning_rate": 1e-06,
      "loss": 0.9122,
      "num_input_tokens_seen": 51727060,
      "step": 924
    },
    {
      "epoch": 2.0579064587973273,
      "loss": 0.7765226364135742,
      "loss_ce": 0.000399627722799778,
      "loss_iou": 0.326171875,
      "loss_num": 0.0245361328125,
      "loss_xval": 0.77734375,
      "num_input_tokens_seen": 51727060,
      "step": 924
    },
    {
      "epoch": 2.060133630289532,
      "grad_norm": 27.499584197998047,
      "learning_rate": 1e-06,
      "loss": 1.1524,
      "num_input_tokens_seen": 51782548,
      "step": 925
    },
    {
      "epoch": 2.060133630289532,
      "loss": 1.210993766784668,
      "loss_ce": 0.00042247679084539413,
      "loss_iou": 0.51171875,
      "loss_num": 0.037353515625,
      "loss_xval": 1.2109375,
      "num_input_tokens_seen": 51782548,
      "step": 925
    },
    {
      "epoch": 2.062360801781737,
      "grad_norm": 21.81817626953125,
      "learning_rate": 1e-06,
      "loss": 0.7765,
      "num_input_tokens_seen": 51839240,
      "step": 926
    },
    {
      "epoch": 2.062360801781737,
      "loss": 0.8635718822479248,
      "loss_ce": 0.0005347240949049592,
      "loss_iou": 0.365234375,
      "loss_num": 0.02685546875,
      "loss_xval": 0.86328125,
      "num_input_tokens_seen": 51839240,
      "step": 926
    },
    {
      "epoch": 2.0645879732739423,
      "grad_norm": 27.51004409790039,
      "learning_rate": 1e-06,
      "loss": 0.885,
      "num_input_tokens_seen": 51896564,
      "step": 927
    },
    {
      "epoch": 2.0645879732739423,
      "loss": 0.7961362600326538,
      "loss_ce": 0.0007261328864842653,
      "loss_iou": 0.357421875,
      "loss_num": 0.016357421875,
      "loss_xval": 0.796875,
      "num_input_tokens_seen": 51896564,
      "step": 927
    },
    {
      "epoch": 2.066815144766147,
      "grad_norm": 19.509262084960938,
      "learning_rate": 1e-06,
      "loss": 0.9487,
      "num_input_tokens_seen": 51954088,
      "step": 928
    },
    {
      "epoch": 2.066815144766147,
      "loss": 1.0089809894561768,
      "loss_ce": 0.0006801420240662992,
      "loss_iou": 0.396484375,
      "loss_num": 0.04296875,
      "loss_xval": 1.0078125,
      "num_input_tokens_seen": 51954088,
      "step": 928
    },
    {
      "epoch": 2.069042316258352,
      "grad_norm": 21.640626907348633,
      "learning_rate": 1e-06,
      "loss": 0.9389,
      "num_input_tokens_seen": 52010852,
      "step": 929
    },
    {
      "epoch": 2.069042316258352,
      "loss": 1.0190068483352661,
      "loss_ce": 0.0014287333469837904,
      "loss_iou": 0.40625,
      "loss_num": 0.04150390625,
      "loss_xval": 1.015625,
      "num_input_tokens_seen": 52010852,
      "step": 929
    },
    {
      "epoch": 2.071269487750557,
      "grad_norm": 18.838546752929688,
      "learning_rate": 1e-06,
      "loss": 1.213,
      "num_input_tokens_seen": 52069620,
      "step": 930
    },
    {
      "epoch": 2.071269487750557,
      "loss": 1.1467416286468506,
      "loss_ce": 0.0009896388510242105,
      "loss_iou": 0.498046875,
      "loss_num": 0.030029296875,
      "loss_xval": 1.1484375,
      "num_input_tokens_seen": 52069620,
      "step": 930
    },
    {
      "epoch": 2.0734966592427617,
      "grad_norm": 22.098485946655273,
      "learning_rate": 1e-06,
      "loss": 0.9992,
      "num_input_tokens_seen": 52125176,
      "step": 931
    },
    {
      "epoch": 2.0734966592427617,
      "loss": 1.115182638168335,
      "loss_ce": 0.0004365970380604267,
      "loss_iou": 0.42578125,
      "loss_num": 0.05224609375,
      "loss_xval": 1.1171875,
      "num_input_tokens_seen": 52125176,
      "step": 931
    },
    {
      "epoch": 2.0757238307349666,
      "grad_norm": 15.070525169372559,
      "learning_rate": 1e-06,
      "loss": 0.9082,
      "num_input_tokens_seen": 52182404,
      "step": 932
    },
    {
      "epoch": 2.0757238307349666,
      "loss": 0.8831936717033386,
      "loss_ce": 0.0005032622721046209,
      "loss_iou": 0.3828125,
      "loss_num": 0.023681640625,
      "loss_xval": 0.8828125,
      "num_input_tokens_seen": 52182404,
      "step": 932
    },
    {
      "epoch": 2.0779510022271714,
      "grad_norm": 22.24066925048828,
      "learning_rate": 1e-06,
      "loss": 0.9648,
      "num_input_tokens_seen": 52238920,
      "step": 933
    },
    {
      "epoch": 2.0779510022271714,
      "loss": 1.1474614143371582,
      "loss_ce": 0.0004888359108008444,
      "loss_iou": 0.462890625,
      "loss_num": 0.044189453125,
      "loss_xval": 1.1484375,
      "num_input_tokens_seen": 52238920,
      "step": 933
    },
    {
      "epoch": 2.0801781737193763,
      "grad_norm": 17.68231964111328,
      "learning_rate": 1e-06,
      "loss": 0.9119,
      "num_input_tokens_seen": 52293392,
      "step": 934
    },
    {
      "epoch": 2.0801781737193763,
      "loss": 0.9804922342300415,
      "loss_ce": 0.0010000698966905475,
      "loss_iou": 0.38671875,
      "loss_num": 0.041015625,
      "loss_xval": 0.98046875,
      "num_input_tokens_seen": 52293392,
      "step": 934
    },
    {
      "epoch": 2.082405345211581,
      "grad_norm": 16.659244537353516,
      "learning_rate": 1e-06,
      "loss": 0.9513,
      "num_input_tokens_seen": 52345928,
      "step": 935
    },
    {
      "epoch": 2.082405345211581,
      "loss": 0.7912268042564392,
      "loss_ce": 0.00045530046918429434,
      "loss_iou": 0.337890625,
      "loss_num": 0.0228271484375,
      "loss_xval": 0.7890625,
      "num_input_tokens_seen": 52345928,
      "step": 935
    },
    {
      "epoch": 2.084632516703786,
      "grad_norm": 17.690183639526367,
      "learning_rate": 1e-06,
      "loss": 0.8534,
      "num_input_tokens_seen": 52401488,
      "step": 936
    },
    {
      "epoch": 2.084632516703786,
      "loss": 0.9090708494186401,
      "loss_ce": 0.00037944965879432857,
      "loss_iou": 0.33984375,
      "loss_num": 0.04541015625,
      "loss_xval": 0.91015625,
      "num_input_tokens_seen": 52401488,
      "step": 936
    },
    {
      "epoch": 2.086859688195991,
      "grad_norm": 18.25249481201172,
      "learning_rate": 1e-06,
      "loss": 0.9348,
      "num_input_tokens_seen": 52454132,
      "step": 937
    },
    {
      "epoch": 2.086859688195991,
      "loss": 0.9369181990623474,
      "loss_ce": 0.00039476132951676846,
      "loss_iou": 0.3828125,
      "loss_num": 0.0341796875,
      "loss_xval": 0.9375,
      "num_input_tokens_seen": 52454132,
      "step": 937
    },
    {
      "epoch": 2.089086859688196,
      "grad_norm": 23.943071365356445,
      "learning_rate": 1e-06,
      "loss": 0.9978,
      "num_input_tokens_seen": 52508724,
      "step": 938
    },
    {
      "epoch": 2.089086859688196,
      "loss": 0.9771347045898438,
      "loss_ce": 0.0025253635831177235,
      "loss_iou": 0.40234375,
      "loss_num": 0.033935546875,
      "loss_xval": 0.9765625,
      "num_input_tokens_seen": 52508724,
      "step": 938
    },
    {
      "epoch": 2.091314031180401,
      "grad_norm": 19.138086318969727,
      "learning_rate": 1e-06,
      "loss": 0.9259,
      "num_input_tokens_seen": 52567416,
      "step": 939
    },
    {
      "epoch": 2.091314031180401,
      "loss": 0.8586698174476624,
      "loss_ce": 0.0019803589675575495,
      "loss_iou": 0.3515625,
      "loss_num": 0.03125,
      "loss_xval": 0.85546875,
      "num_input_tokens_seen": 52567416,
      "step": 939
    },
    {
      "epoch": 2.093541202672606,
      "grad_norm": 29.820850372314453,
      "learning_rate": 1e-06,
      "loss": 1.0665,
      "num_input_tokens_seen": 52625816,
      "step": 940
    },
    {
      "epoch": 2.093541202672606,
      "loss": 1.4001514911651611,
      "loss_ce": 0.0012256972258910537,
      "loss_iou": 0.5703125,
      "loss_num": 0.051513671875,
      "loss_xval": 1.3984375,
      "num_input_tokens_seen": 52625816,
      "step": 940
    },
    {
      "epoch": 2.0957683741648108,
      "grad_norm": 14.36317253112793,
      "learning_rate": 1e-06,
      "loss": 0.5621,
      "num_input_tokens_seen": 52682968,
      "step": 941
    },
    {
      "epoch": 2.0957683741648108,
      "loss": 0.5812158584594727,
      "loss_ce": 0.00040527121745981276,
      "loss_iou": 0.2197265625,
      "loss_num": 0.0284423828125,
      "loss_xval": 0.58203125,
      "num_input_tokens_seen": 52682968,
      "step": 941
    },
    {
      "epoch": 2.0979955456570156,
      "grad_norm": 19.567962646484375,
      "learning_rate": 1e-06,
      "loss": 0.9139,
      "num_input_tokens_seen": 52737516,
      "step": 942
    },
    {
      "epoch": 2.0979955456570156,
      "loss": 1.065826654434204,
      "loss_ce": 0.00039695383748039603,
      "loss_iou": 0.453125,
      "loss_num": 0.031982421875,
      "loss_xval": 1.0625,
      "num_input_tokens_seen": 52737516,
      "step": 942
    },
    {
      "epoch": 2.1002227171492205,
      "grad_norm": 28.01074981689453,
      "learning_rate": 1e-06,
      "loss": 1.1929,
      "num_input_tokens_seen": 52793536,
      "step": 943
    },
    {
      "epoch": 2.1002227171492205,
      "loss": 1.1947014331817627,
      "loss_ce": 0.00036561937304213643,
      "loss_iou": 0.48046875,
      "loss_num": 0.046630859375,
      "loss_xval": 1.1953125,
      "num_input_tokens_seen": 52793536,
      "step": 943
    },
    {
      "epoch": 2.1024498886414253,
      "grad_norm": 20.295827865600586,
      "learning_rate": 1e-06,
      "loss": 1.2702,
      "num_input_tokens_seen": 52847904,
      "step": 944
    },
    {
      "epoch": 2.1024498886414253,
      "loss": 1.352550983428955,
      "loss_ce": 0.0005001933313906193,
      "loss_iou": 0.578125,
      "loss_num": 0.0380859375,
      "loss_xval": 1.3515625,
      "num_input_tokens_seen": 52847904,
      "step": 944
    },
    {
      "epoch": 2.10467706013363,
      "grad_norm": 16.32712745666504,
      "learning_rate": 1e-06,
      "loss": 1.0087,
      "num_input_tokens_seen": 52904284,
      "step": 945
    },
    {
      "epoch": 2.10467706013363,
      "loss": 1.0332987308502197,
      "loss_ce": 0.0005839248769916594,
      "loss_iou": 0.404296875,
      "loss_num": 0.044921875,
      "loss_xval": 1.03125,
      "num_input_tokens_seen": 52904284,
      "step": 945
    },
    {
      "epoch": 2.106904231625835,
      "grad_norm": 29.043960571289062,
      "learning_rate": 1e-06,
      "loss": 1.0885,
      "num_input_tokens_seen": 52960784,
      "step": 946
    },
    {
      "epoch": 2.106904231625835,
      "loss": 0.9473793506622314,
      "loss_ce": 0.0006019732682034373,
      "loss_iou": 0.3671875,
      "loss_num": 0.04248046875,
      "loss_xval": 0.9453125,
      "num_input_tokens_seen": 52960784,
      "step": 946
    },
    {
      "epoch": 2.10913140311804,
      "grad_norm": 14.747257232666016,
      "learning_rate": 1e-06,
      "loss": 0.9877,
      "num_input_tokens_seen": 53015812,
      "step": 947
    },
    {
      "epoch": 2.10913140311804,
      "loss": 1.0732710361480713,
      "loss_ce": 0.0007612318731844425,
      "loss_iou": 0.42578125,
      "loss_num": 0.044189453125,
      "loss_xval": 1.0703125,
      "num_input_tokens_seen": 53015812,
      "step": 947
    },
    {
      "epoch": 2.111358574610245,
      "grad_norm": 103.11000061035156,
      "learning_rate": 1e-06,
      "loss": 0.9914,
      "num_input_tokens_seen": 53072308,
      "step": 948
    },
    {
      "epoch": 2.111358574610245,
      "loss": 0.9887726902961731,
      "loss_ce": 0.000491408514790237,
      "loss_iou": 0.447265625,
      "loss_num": 0.0185546875,
      "loss_xval": 0.98828125,
      "num_input_tokens_seen": 53072308,
      "step": 948
    },
    {
      "epoch": 2.11358574610245,
      "grad_norm": 23.691804885864258,
      "learning_rate": 1e-06,
      "loss": 1.2725,
      "num_input_tokens_seen": 53127836,
      "step": 949
    },
    {
      "epoch": 2.11358574610245,
      "loss": 1.3882193565368652,
      "loss_ce": 0.001500481041148305,
      "loss_iou": 0.51171875,
      "loss_num": 0.07373046875,
      "loss_xval": 1.390625,
      "num_input_tokens_seen": 53127836,
      "step": 949
    },
    {
      "epoch": 2.115812917594655,
      "grad_norm": 13.513452529907227,
      "learning_rate": 1e-06,
      "loss": 0.9153,
      "num_input_tokens_seen": 53184644,
      "step": 950
    },
    {
      "epoch": 2.115812917594655,
      "loss": 1.0870821475982666,
      "loss_ce": 0.0004122781683690846,
      "loss_iou": 0.435546875,
      "loss_num": 0.042724609375,
      "loss_xval": 1.0859375,
      "num_input_tokens_seen": 53184644,
      "step": 950
    },
    {
      "epoch": 2.11804008908686,
      "grad_norm": 14.758941650390625,
      "learning_rate": 1e-06,
      "loss": 0.8424,
      "num_input_tokens_seen": 53242056,
      "step": 951
    },
    {
      "epoch": 2.11804008908686,
      "loss": 0.9238724708557129,
      "loss_ce": 0.00041059922659769654,
      "loss_iou": 0.384765625,
      "loss_num": 0.0311279296875,
      "loss_xval": 0.921875,
      "num_input_tokens_seen": 53242056,
      "step": 951
    },
    {
      "epoch": 2.1202672605790647,
      "grad_norm": 50.94358825683594,
      "learning_rate": 1e-06,
      "loss": 0.9876,
      "num_input_tokens_seen": 53296836,
      "step": 952
    },
    {
      "epoch": 2.1202672605790647,
      "loss": 0.8739633560180664,
      "loss_ce": 0.00042815617052838206,
      "loss_iou": 0.35546875,
      "loss_num": 0.0322265625,
      "loss_xval": 0.875,
      "num_input_tokens_seen": 53296836,
      "step": 952
    },
    {
      "epoch": 2.1224944320712695,
      "grad_norm": 29.346717834472656,
      "learning_rate": 1e-06,
      "loss": 0.8476,
      "num_input_tokens_seen": 53356080,
      "step": 953
    },
    {
      "epoch": 2.1224944320712695,
      "loss": 0.7091416120529175,
      "loss_ce": 0.00040138079202733934,
      "loss_iou": 0.298828125,
      "loss_num": 0.022705078125,
      "loss_xval": 0.70703125,
      "num_input_tokens_seen": 53356080,
      "step": 953
    },
    {
      "epoch": 2.1247216035634744,
      "grad_norm": 30.826066970825195,
      "learning_rate": 1e-06,
      "loss": 0.8384,
      "num_input_tokens_seen": 53411060,
      "step": 954
    },
    {
      "epoch": 2.1247216035634744,
      "loss": 0.8927929401397705,
      "loss_ce": 0.00045894747017882764,
      "loss_iou": 0.330078125,
      "loss_num": 0.04638671875,
      "loss_xval": 0.890625,
      "num_input_tokens_seen": 53411060,
      "step": 954
    },
    {
      "epoch": 2.1269487750556793,
      "grad_norm": 26.217329025268555,
      "learning_rate": 1e-06,
      "loss": 1.2109,
      "num_input_tokens_seen": 53467452,
      "step": 955
    },
    {
      "epoch": 2.1269487750556793,
      "loss": 1.171452283859253,
      "loss_ce": 0.0005538875702768564,
      "loss_iou": 0.50390625,
      "loss_num": 0.031982421875,
      "loss_xval": 1.171875,
      "num_input_tokens_seen": 53467452,
      "step": 955
    },
    {
      "epoch": 2.129175946547884,
      "grad_norm": 15.275303840637207,
      "learning_rate": 1e-06,
      "loss": 1.0093,
      "num_input_tokens_seen": 53523000,
      "step": 956
    },
    {
      "epoch": 2.129175946547884,
      "loss": 0.9647861123085022,
      "loss_ce": 0.0004306259215809405,
      "loss_iou": 0.3984375,
      "loss_num": 0.033447265625,
      "loss_xval": 0.96484375,
      "num_input_tokens_seen": 53523000,
      "step": 956
    },
    {
      "epoch": 2.131403118040089,
      "grad_norm": 20.987869262695312,
      "learning_rate": 1e-06,
      "loss": 0.8899,
      "num_input_tokens_seen": 53577268,
      "step": 957
    },
    {
      "epoch": 2.131403118040089,
      "loss": 0.8194369077682495,
      "loss_ce": 0.00034510315163061023,
      "loss_iou": 0.3359375,
      "loss_num": 0.0296630859375,
      "loss_xval": 0.8203125,
      "num_input_tokens_seen": 53577268,
      "step": 957
    },
    {
      "epoch": 2.133630289532294,
      "grad_norm": 17.244304656982422,
      "learning_rate": 1e-06,
      "loss": 0.9678,
      "num_input_tokens_seen": 53632968,
      "step": 958
    },
    {
      "epoch": 2.133630289532294,
      "loss": 0.7954371571540833,
      "loss_ce": 0.0010035325540229678,
      "loss_iou": 0.32421875,
      "loss_num": 0.02880859375,
      "loss_xval": 0.79296875,
      "num_input_tokens_seen": 53632968,
      "step": 958
    },
    {
      "epoch": 2.1358574610244987,
      "grad_norm": 27.137948989868164,
      "learning_rate": 1e-06,
      "loss": 0.8922,
      "num_input_tokens_seen": 53686276,
      "step": 959
    },
    {
      "epoch": 2.1358574610244987,
      "loss": 0.8764593005180359,
      "loss_ce": 0.007318664342164993,
      "loss_iou": 0.345703125,
      "loss_num": 0.035888671875,
      "loss_xval": 0.8671875,
      "num_input_tokens_seen": 53686276,
      "step": 959
    },
    {
      "epoch": 2.138084632516704,
      "grad_norm": 13.352965354919434,
      "learning_rate": 1e-06,
      "loss": 1.2179,
      "num_input_tokens_seen": 53738980,
      "step": 960
    },
    {
      "epoch": 2.138084632516704,
      "loss": 1.0099034309387207,
      "loss_ce": 0.0003819413250312209,
      "loss_iou": 0.4375,
      "loss_num": 0.0272216796875,
      "loss_xval": 1.0078125,
      "num_input_tokens_seen": 53738980,
      "step": 960
    },
    {
      "epoch": 2.140311804008909,
      "grad_norm": 26.12520980834961,
      "learning_rate": 1e-06,
      "loss": 1.0853,
      "num_input_tokens_seen": 53795408,
      "step": 961
    },
    {
      "epoch": 2.140311804008909,
      "loss": 1.0113427639007568,
      "loss_ce": 0.00035642064176499844,
      "loss_iou": 0.42578125,
      "loss_num": 0.0322265625,
      "loss_xval": 1.0078125,
      "num_input_tokens_seen": 53795408,
      "step": 961
    },
    {
      "epoch": 2.1425389755011137,
      "grad_norm": 25.219985961914062,
      "learning_rate": 1e-06,
      "loss": 1.0355,
      "num_input_tokens_seen": 53849140,
      "step": 962
    },
    {
      "epoch": 2.1425389755011137,
      "loss": 0.8350330591201782,
      "loss_ce": 0.0008045152062550187,
      "loss_iou": 0.33984375,
      "loss_num": 0.030517578125,
      "loss_xval": 0.8359375,
      "num_input_tokens_seen": 53849140,
      "step": 962
    },
    {
      "epoch": 2.1447661469933186,
      "grad_norm": 22.237886428833008,
      "learning_rate": 1e-06,
      "loss": 1.0859,
      "num_input_tokens_seen": 53904680,
      "step": 963
    },
    {
      "epoch": 2.1447661469933186,
      "loss": 0.9965952634811401,
      "loss_ce": 0.000501530768815428,
      "loss_iou": 0.41015625,
      "loss_num": 0.034912109375,
      "loss_xval": 0.99609375,
      "num_input_tokens_seen": 53904680,
      "step": 963
    },
    {
      "epoch": 2.1469933184855234,
      "grad_norm": 27.480417251586914,
      "learning_rate": 1e-06,
      "loss": 1.0019,
      "num_input_tokens_seen": 53962276,
      "step": 964
    },
    {
      "epoch": 2.1469933184855234,
      "loss": 0.9119249582290649,
      "loss_ce": 0.0007921320502646267,
      "loss_iou": 0.373046875,
      "loss_num": 0.033203125,
      "loss_xval": 0.91015625,
      "num_input_tokens_seen": 53962276,
      "step": 964
    },
    {
      "epoch": 2.1492204899777283,
      "grad_norm": 16.98265266418457,
      "learning_rate": 1e-06,
      "loss": 0.9955,
      "num_input_tokens_seen": 54015980,
      "step": 965
    },
    {
      "epoch": 2.1492204899777283,
      "loss": 0.8996407985687256,
      "loss_ce": 0.0004708755586761981,
      "loss_iou": 0.376953125,
      "loss_num": 0.029052734375,
      "loss_xval": 0.8984375,
      "num_input_tokens_seen": 54015980,
      "step": 965
    },
    {
      "epoch": 2.151447661469933,
      "grad_norm": 17.36399269104004,
      "learning_rate": 1e-06,
      "loss": 0.9589,
      "num_input_tokens_seen": 54072540,
      "step": 966
    },
    {
      "epoch": 2.151447661469933,
      "loss": 0.8351460099220276,
      "loss_ce": 0.00042922317516058683,
      "loss_iou": 0.341796875,
      "loss_num": 0.0301513671875,
      "loss_xval": 0.8359375,
      "num_input_tokens_seen": 54072540,
      "step": 966
    },
    {
      "epoch": 2.153674832962138,
      "grad_norm": 25.51936149597168,
      "learning_rate": 1e-06,
      "loss": 1.0361,
      "num_input_tokens_seen": 54128332,
      "step": 967
    },
    {
      "epoch": 2.153674832962138,
      "loss": 0.8526254892349243,
      "loss_ce": 0.0005747509421780705,
      "loss_iou": 0.365234375,
      "loss_num": 0.02392578125,
      "loss_xval": 0.8515625,
      "num_input_tokens_seen": 54128332,
      "step": 967
    },
    {
      "epoch": 2.155902004454343,
      "grad_norm": 25.46952247619629,
      "learning_rate": 1e-06,
      "loss": 1.2554,
      "num_input_tokens_seen": 54184776,
      "step": 968
    },
    {
      "epoch": 2.155902004454343,
      "loss": 1.3332043886184692,
      "loss_ce": 0.0011730894912034273,
      "loss_iou": 0.52734375,
      "loss_num": 0.05615234375,
      "loss_xval": 1.328125,
      "num_input_tokens_seen": 54184776,
      "step": 968
    },
    {
      "epoch": 2.1581291759465477,
      "grad_norm": 16.16930389404297,
      "learning_rate": 1e-06,
      "loss": 0.8004,
      "num_input_tokens_seen": 54241804,
      "step": 969
    },
    {
      "epoch": 2.1581291759465477,
      "loss": 0.612720787525177,
      "loss_ce": 0.006275475956499577,
      "loss_iou": 0.2412109375,
      "loss_num": 0.0245361328125,
      "loss_xval": 0.60546875,
      "num_input_tokens_seen": 54241804,
      "step": 969
    },
    {
      "epoch": 2.1603563474387526,
      "grad_norm": 12.317280769348145,
      "learning_rate": 1e-06,
      "loss": 0.7401,
      "num_input_tokens_seen": 54297804,
      "step": 970
    },
    {
      "epoch": 2.1603563474387526,
      "loss": 0.8193256855010986,
      "loss_ce": 0.00047803280176594853,
      "loss_iou": 0.349609375,
      "loss_num": 0.0240478515625,
      "loss_xval": 0.8203125,
      "num_input_tokens_seen": 54297804,
      "step": 970
    },
    {
      "epoch": 2.1625835189309575,
      "grad_norm": 25.277143478393555,
      "learning_rate": 1e-06,
      "loss": 0.9579,
      "num_input_tokens_seen": 54352416,
      "step": 971
    },
    {
      "epoch": 2.1625835189309575,
      "loss": 1.0242624282836914,
      "loss_ce": 0.0003366165910847485,
      "loss_iou": 0.435546875,
      "loss_num": 0.030517578125,
      "loss_xval": 1.0234375,
      "num_input_tokens_seen": 54352416,
      "step": 971
    },
    {
      "epoch": 2.1648106904231628,
      "grad_norm": 23.600387573242188,
      "learning_rate": 1e-06,
      "loss": 1.0226,
      "num_input_tokens_seen": 54408836,
      "step": 972
    },
    {
      "epoch": 2.1648106904231628,
      "loss": 0.9383513927459717,
      "loss_ce": 0.0003631227882578969,
      "loss_iou": 0.392578125,
      "loss_num": 0.0303955078125,
      "loss_xval": 0.9375,
      "num_input_tokens_seen": 54408836,
      "step": 972
    },
    {
      "epoch": 2.1670378619153676,
      "grad_norm": 28.22263526916504,
      "learning_rate": 1e-06,
      "loss": 1.102,
      "num_input_tokens_seen": 54461188,
      "step": 973
    },
    {
      "epoch": 2.1670378619153676,
      "loss": 1.2241106033325195,
      "loss_ce": 0.0004167944425716996,
      "loss_iou": 0.498046875,
      "loss_num": 0.04541015625,
      "loss_xval": 1.2265625,
      "num_input_tokens_seen": 54461188,
      "step": 973
    },
    {
      "epoch": 2.1692650334075725,
      "grad_norm": 23.0220947265625,
      "learning_rate": 1e-06,
      "loss": 1.0596,
      "num_input_tokens_seen": 54516224,
      "step": 974
    },
    {
      "epoch": 2.1692650334075725,
      "loss": 1.1534892320632935,
      "loss_ce": 0.0006571857957169414,
      "loss_iou": 0.47265625,
      "loss_num": 0.04150390625,
      "loss_xval": 1.15625,
      "num_input_tokens_seen": 54516224,
      "step": 974
    },
    {
      "epoch": 2.1714922048997773,
      "grad_norm": 50.8327751159668,
      "learning_rate": 1e-06,
      "loss": 0.962,
      "num_input_tokens_seen": 54571536,
      "step": 975
    },
    {
      "epoch": 2.1714922048997773,
      "loss": 0.8975195288658142,
      "loss_ce": 0.0004247867036610842,
      "loss_iou": 0.3828125,
      "loss_num": 0.0264892578125,
      "loss_xval": 0.8984375,
      "num_input_tokens_seen": 54571536,
      "step": 975
    },
    {
      "epoch": 2.173719376391982,
      "grad_norm": 18.573394775390625,
      "learning_rate": 1e-06,
      "loss": 1.0714,
      "num_input_tokens_seen": 54624220,
      "step": 976
    },
    {
      "epoch": 2.173719376391982,
      "loss": 0.7970374822616577,
      "loss_ce": 0.0011390313738957047,
      "loss_iou": 0.3046875,
      "loss_num": 0.037109375,
      "loss_xval": 0.796875,
      "num_input_tokens_seen": 54624220,
      "step": 976
    },
    {
      "epoch": 2.175946547884187,
      "grad_norm": 26.248809814453125,
      "learning_rate": 1e-06,
      "loss": 0.9597,
      "num_input_tokens_seen": 54681684,
      "step": 977
    },
    {
      "epoch": 2.175946547884187,
      "loss": 0.7498317956924438,
      "loss_ce": 0.0003200596256647259,
      "loss_iou": 0.314453125,
      "loss_num": 0.02392578125,
      "loss_xval": 0.75,
      "num_input_tokens_seen": 54681684,
      "step": 977
    },
    {
      "epoch": 2.178173719376392,
      "grad_norm": 17.984722137451172,
      "learning_rate": 1e-06,
      "loss": 0.8119,
      "num_input_tokens_seen": 54739128,
      "step": 978
    },
    {
      "epoch": 2.178173719376392,
      "loss": 0.8240878582000732,
      "loss_ce": 0.00035743031185120344,
      "loss_iou": 0.345703125,
      "loss_num": 0.0263671875,
      "loss_xval": 0.82421875,
      "num_input_tokens_seen": 54739128,
      "step": 978
    },
    {
      "epoch": 2.180400890868597,
      "grad_norm": 16.49453353881836,
      "learning_rate": 1e-06,
      "loss": 1.1017,
      "num_input_tokens_seen": 54795068,
      "step": 979
    },
    {
      "epoch": 2.180400890868597,
      "loss": 1.3098089694976807,
      "loss_ce": 0.0004827585944440216,
      "loss_iou": 0.546875,
      "loss_num": 0.042724609375,
      "loss_xval": 1.3125,
      "num_input_tokens_seen": 54795068,
      "step": 979
    },
    {
      "epoch": 2.1826280623608016,
      "grad_norm": 25.356178283691406,
      "learning_rate": 1e-06,
      "loss": 0.9565,
      "num_input_tokens_seen": 54851036,
      "step": 980
    },
    {
      "epoch": 2.1826280623608016,
      "loss": 0.794826328754425,
      "loss_ce": 0.0003927270008716732,
      "loss_iou": 0.3515625,
      "loss_num": 0.018310546875,
      "loss_xval": 0.79296875,
      "num_input_tokens_seen": 54851036,
      "step": 980
    },
    {
      "epoch": 2.1848552338530065,
      "grad_norm": 20.565073013305664,
      "learning_rate": 1e-06,
      "loss": 0.8048,
      "num_input_tokens_seen": 54903052,
      "step": 981
    },
    {
      "epoch": 2.1848552338530065,
      "loss": 0.7406924366950989,
      "loss_ce": 0.00045808005961589515,
      "loss_iou": 0.318359375,
      "loss_num": 0.020751953125,
      "loss_xval": 0.7421875,
      "num_input_tokens_seen": 54903052,
      "step": 981
    },
    {
      "epoch": 2.187082405345212,
      "grad_norm": 25.87790870666504,
      "learning_rate": 1e-06,
      "loss": 0.8178,
      "num_input_tokens_seen": 54960984,
      "step": 982
    },
    {
      "epoch": 2.187082405345212,
      "loss": 0.7066792249679565,
      "loss_ce": 0.0006245420081540942,
      "loss_iou": 0.294921875,
      "loss_num": 0.0235595703125,
      "loss_xval": 0.70703125,
      "num_input_tokens_seen": 54960984,
      "step": 982
    },
    {
      "epoch": 2.1893095768374167,
      "grad_norm": 19.50031089782715,
      "learning_rate": 1e-06,
      "loss": 1.1593,
      "num_input_tokens_seen": 55017748,
      "step": 983
    },
    {
      "epoch": 2.1893095768374167,
      "loss": 0.988908588886261,
      "loss_ce": 0.0003832111251540482,
      "loss_iou": 0.423828125,
      "loss_num": 0.028076171875,
      "loss_xval": 0.98828125,
      "num_input_tokens_seen": 55017748,
      "step": 983
    },
    {
      "epoch": 2.1915367483296215,
      "grad_norm": 16.748844146728516,
      "learning_rate": 1e-06,
      "loss": 0.8003,
      "num_input_tokens_seen": 55074808,
      "step": 984
    },
    {
      "epoch": 2.1915367483296215,
      "loss": 0.9688905477523804,
      "loss_ce": 0.00038473017048090696,
      "loss_iou": 0.419921875,
      "loss_num": 0.0262451171875,
      "loss_xval": 0.96875,
      "num_input_tokens_seen": 55074808,
      "step": 984
    },
    {
      "epoch": 2.1937639198218264,
      "grad_norm": 68.782958984375,
      "learning_rate": 1e-06,
      "loss": 1.0795,
      "num_input_tokens_seen": 55131672,
      "step": 985
    },
    {
      "epoch": 2.1937639198218264,
      "loss": 0.8397893309593201,
      "loss_ce": 0.0006780020194128156,
      "loss_iou": 0.353515625,
      "loss_num": 0.02685546875,
      "loss_xval": 0.83984375,
      "num_input_tokens_seen": 55131672,
      "step": 985
    },
    {
      "epoch": 2.1959910913140313,
      "grad_norm": 21.74385643005371,
      "learning_rate": 1e-06,
      "loss": 0.966,
      "num_input_tokens_seen": 55186240,
      "step": 986
    },
    {
      "epoch": 2.1959910913140313,
      "loss": 1.022130012512207,
      "loss_ce": 0.005528404843062162,
      "loss_iou": 0.404296875,
      "loss_num": 0.04150390625,
      "loss_xval": 1.015625,
      "num_input_tokens_seen": 55186240,
      "step": 986
    },
    {
      "epoch": 2.198218262806236,
      "grad_norm": 27.58509063720703,
      "learning_rate": 1e-06,
      "loss": 1.0463,
      "num_input_tokens_seen": 55243732,
      "step": 987
    },
    {
      "epoch": 2.198218262806236,
      "loss": 1.0700922012329102,
      "loss_ce": 0.0005121089052408934,
      "loss_iou": 0.4375,
      "loss_num": 0.038330078125,
      "loss_xval": 1.0703125,
      "num_input_tokens_seen": 55243732,
      "step": 987
    },
    {
      "epoch": 2.200445434298441,
      "grad_norm": 19.08161735534668,
      "learning_rate": 1e-06,
      "loss": 1.0416,
      "num_input_tokens_seen": 55300384,
      "step": 988
    },
    {
      "epoch": 2.200445434298441,
      "loss": 1.1714062690734863,
      "loss_ce": 0.0007519207429140806,
      "loss_iou": 0.4375,
      "loss_num": 0.0595703125,
      "loss_xval": 1.171875,
      "num_input_tokens_seen": 55300384,
      "step": 988
    },
    {
      "epoch": 2.202672605790646,
      "grad_norm": 21.197906494140625,
      "learning_rate": 1e-06,
      "loss": 0.9063,
      "num_input_tokens_seen": 55355972,
      "step": 989
    },
    {
      "epoch": 2.202672605790646,
      "loss": 0.8070752024650574,
      "loss_ce": 0.0011670144740492105,
      "loss_iou": 0.330078125,
      "loss_num": 0.02880859375,
      "loss_xval": 0.8046875,
      "num_input_tokens_seen": 55355972,
      "step": 989
    },
    {
      "epoch": 2.2048997772828507,
      "grad_norm": 57.792137145996094,
      "learning_rate": 1e-06,
      "loss": 0.7107,
      "num_input_tokens_seen": 55411904,
      "step": 990
    },
    {
      "epoch": 2.2048997772828507,
      "loss": 0.8724934458732605,
      "loss_ce": 0.00042314609163440764,
      "loss_iou": 0.357421875,
      "loss_num": 0.03125,
      "loss_xval": 0.87109375,
      "num_input_tokens_seen": 55411904,
      "step": 990
    },
    {
      "epoch": 2.2071269487750556,
      "grad_norm": 18.81410026550293,
      "learning_rate": 1e-06,
      "loss": 0.921,
      "num_input_tokens_seen": 55469032,
      "step": 991
    },
    {
      "epoch": 2.2071269487750556,
      "loss": 0.5198028087615967,
      "loss_ce": 0.0003936740104109049,
      "loss_iou": 0.2275390625,
      "loss_num": 0.01287841796875,
      "loss_xval": 0.51953125,
      "num_input_tokens_seen": 55469032,
      "step": 991
    },
    {
      "epoch": 2.2093541202672604,
      "grad_norm": 19.535879135131836,
      "learning_rate": 1e-06,
      "loss": 0.8962,
      "num_input_tokens_seen": 55528000,
      "step": 992
    },
    {
      "epoch": 2.2093541202672604,
      "loss": 1.013750433921814,
      "loss_ce": 0.0005668357480317354,
      "loss_iou": 0.423828125,
      "loss_num": 0.033203125,
      "loss_xval": 1.015625,
      "num_input_tokens_seen": 55528000,
      "step": 992
    },
    {
      "epoch": 2.2115812917594653,
      "grad_norm": 19.903423309326172,
      "learning_rate": 1e-06,
      "loss": 1.1321,
      "num_input_tokens_seen": 55585248,
      "step": 993
    },
    {
      "epoch": 2.2115812917594653,
      "loss": 1.1713190078735352,
      "loss_ce": 0.0009089382365345955,
      "loss_iou": 0.486328125,
      "loss_num": 0.03955078125,
      "loss_xval": 1.171875,
      "num_input_tokens_seen": 55585248,
      "step": 993
    },
    {
      "epoch": 2.2138084632516706,
      "grad_norm": 20.668163299560547,
      "learning_rate": 1e-06,
      "loss": 1.093,
      "num_input_tokens_seen": 55643640,
      "step": 994
    },
    {
      "epoch": 2.2138084632516706,
      "loss": 1.0199007987976074,
      "loss_ce": 0.0006136804004199803,
      "loss_iou": 0.400390625,
      "loss_num": 0.04345703125,
      "loss_xval": 1.015625,
      "num_input_tokens_seen": 55643640,
      "step": 994
    },
    {
      "epoch": 2.2160356347438754,
      "grad_norm": 14.16650676727295,
      "learning_rate": 1e-06,
      "loss": 0.8074,
      "num_input_tokens_seen": 55699880,
      "step": 995
    },
    {
      "epoch": 2.2160356347438754,
      "loss": 0.7972851991653442,
      "loss_ce": 0.0004102342645637691,
      "loss_iou": 0.330078125,
      "loss_num": 0.0274658203125,
      "loss_xval": 0.796875,
      "num_input_tokens_seen": 55699880,
      "step": 995
    },
    {
      "epoch": 2.2182628062360803,
      "grad_norm": 21.781885147094727,
      "learning_rate": 1e-06,
      "loss": 1.2838,
      "num_input_tokens_seen": 55757468,
      "step": 996
    },
    {
      "epoch": 2.2182628062360803,
      "loss": 1.5526351928710938,
      "loss_ce": 0.0008773244917392731,
      "loss_iou": 0.625,
      "loss_num": 0.060546875,
      "loss_xval": 1.5546875,
      "num_input_tokens_seen": 55757468,
      "step": 996
    },
    {
      "epoch": 2.220489977728285,
      "grad_norm": 13.922673225402832,
      "learning_rate": 1e-06,
      "loss": 0.8959,
      "num_input_tokens_seen": 55811116,
      "step": 997
    },
    {
      "epoch": 2.220489977728285,
      "loss": 0.8420233130455017,
      "loss_ce": 0.00047059194184839725,
      "loss_iou": 0.3515625,
      "loss_num": 0.02783203125,
      "loss_xval": 0.83984375,
      "num_input_tokens_seen": 55811116,
      "step": 997
    },
    {
      "epoch": 2.22271714922049,
      "grad_norm": 31.61424446105957,
      "learning_rate": 1e-06,
      "loss": 1.0489,
      "num_input_tokens_seen": 55867056,
      "step": 998
    },
    {
      "epoch": 2.22271714922049,
      "loss": 0.9875812530517578,
      "loss_ce": 0.0016193470219150186,
      "loss_iou": 0.4140625,
      "loss_num": 0.031982421875,
      "loss_xval": 0.984375,
      "num_input_tokens_seen": 55867056,
      "step": 998
    },
    {
      "epoch": 2.224944320712695,
      "grad_norm": 18.81981658935547,
      "learning_rate": 1e-06,
      "loss": 1.1018,
      "num_input_tokens_seen": 55922508,
      "step": 999
    },
    {
      "epoch": 2.224944320712695,
      "loss": 1.031872272491455,
      "loss_ce": 0.0011104578152298927,
      "loss_iou": 0.41796875,
      "loss_num": 0.0390625,
      "loss_xval": 1.03125,
      "num_input_tokens_seen": 55922508,
      "step": 999
    },
    {
      "epoch": 2.2271714922048997,
      "grad_norm": 30.99945640563965,
      "learning_rate": 1e-06,
      "loss": 0.8712,
      "num_input_tokens_seen": 55979796,
      "step": 1000
    },
    {
      "epoch": 2.2271714922048997,
      "eval_seeclick_web_CIoU": 0.5659106969833374,
      "eval_seeclick_web_GIoU": 0.5565789341926575,
      "eval_seeclick_web_IoU": 0.5834561288356781,
      "eval_seeclick_web_MAE_all": 0.016565547324717045,
      "eval_seeclick_web_MAE_h": 0.010058594401925802,
      "eval_seeclick_web_MAE_w": 0.016721216030418873,
      "eval_seeclick_web_MAE_x_boxes": 0.009843998588621616,
      "eval_seeclick_web_MAE_y_boxes": 0.02248636749573052,
      "eval_seeclick_web_inside_bbox": 0.9166666567325592,
      "eval_seeclick_web_loss": 0.9639145135879517,
      "eval_seeclick_web_loss_ce": 0.0004584374401019886,
      "eval_seeclick_web_loss_iou": 0.4400634765625,
      "eval_seeclick_web_loss_num": 0.013164520263671875,
      "eval_seeclick_web_loss_xval": 0.946044921875,
      "eval_seeclick_web_runtime": 18.6045,
      "eval_seeclick_web_samples_per_second": 2.688,
      "eval_seeclick_web_steps_per_second": 0.108,
      "num_input_tokens_seen": 55979796,
      "step": 1000
    },
    {
      "epoch": 2.2271714922048997,
      "eval_icons_CIoU": 0.3126373589038849,
      "eval_icons_GIoU": 0.3512195348739624,
      "eval_icons_IoU": 0.38992173969745636,
      "eval_icons_MAE_all": 0.06972917914390564,
      "eval_icons_MAE_h": 0.035466980654746294,
      "eval_icons_MAE_w": 0.07942605763673782,
      "eval_icons_MAE_x_boxes": 0.06474643759429455,
      "eval_icons_MAE_y_boxes": 0.03778674267232418,
      "eval_icons_inside_bbox": 0.6336805522441864,
      "eval_icons_loss": 1.6989706754684448,
      "eval_icons_loss_ce": 0.0012570468825288117,
      "eval_icons_loss_iou": 0.6531982421875,
      "eval_icons_loss_num": 0.06581878662109375,
      "eval_icons_loss_xval": 1.634765625,
      "eval_icons_runtime": 18.1838,
      "eval_icons_samples_per_second": 2.75,
      "eval_icons_steps_per_second": 0.11,
      "num_input_tokens_seen": 55979796,
      "step": 1000
    },
    {
      "epoch": 2.2271714922048997,
      "eval_screenspot_CIoU": 0.30459535121917725,
      "eval_screenspot_GIoU": 0.326509823401769,
      "eval_screenspot_IoU": 0.38954149683316547,
      "eval_screenspot_MAE_all": 0.0834270715713501,
      "eval_screenspot_MAE_h": 0.04727879042426745,
      "eval_screenspot_MAE_w": 0.07983010758956273,
      "eval_screenspot_MAE_x_boxes": 0.12237335244814555,
      "eval_screenspot_MAE_y_boxes": 0.0545121505856514,
      "eval_screenspot_inside_bbox": 0.6045833428700765,
      "eval_screenspot_loss": 1.8311774730682373,
      "eval_screenspot_loss_ce": 0.002042266229788462,
      "eval_screenspot_loss_iou": 0.721923828125,
      "eval_screenspot_loss_num": 0.09384409586588542,
      "eval_screenspot_loss_xval": 1.9124348958333333,
      "eval_screenspot_runtime": 27.92,
      "eval_screenspot_samples_per_second": 3.188,
      "eval_screenspot_steps_per_second": 0.107,
      "num_input_tokens_seen": 55979796,
      "step": 1000
    },
    {
      "epoch": 2.2271714922048997,
      "eval_compot_CIoU": 0.3376414477825165,
      "eval_compot_GIoU": 0.3708181381225586,
      "eval_compot_IoU": 0.40139785408973694,
      "eval_compot_MAE_all": 0.02684677764773369,
      "eval_compot_MAE_h": 0.011378975585103035,
      "eval_compot_MAE_w": 0.037089540623128414,
      "eval_compot_MAE_x_boxes": 0.034720015712082386,
      "eval_compot_MAE_y_boxes": 0.007318113464862108,
      "eval_compot_inside_bbox": 0.5868055522441864,
      "eval_compot_loss": 1.426712989807129,
      "eval_compot_loss_ce": 0.0004611587501130998,
      "eval_compot_loss_iou": 0.6314697265625,
      "eval_compot_loss_num": 0.024566650390625,
      "eval_compot_loss_xval": 1.38525390625,
      "eval_compot_runtime": 18.3333,
      "eval_compot_samples_per_second": 2.727,
      "eval_compot_steps_per_second": 0.109,
      "num_input_tokens_seen": 55979796,
      "step": 1000
    },
    {
      "epoch": 2.2271714922048997,
      "eval_custom_ui_val_CIoU": 0.411270619266563,
      "eval_custom_ui_val_GIoU": 0.4448644651307,
      "eval_custom_ui_val_IoU": 0.4744107557667626,
      "eval_custom_ui_val_MAE_all": 0.040074211441808276,
      "eval_custom_ui_val_MAE_h": 0.023370659496221278,
      "eval_custom_ui_val_MAE_w": 0.04196054426332315,
      "eval_custom_ui_val_MAE_x_boxes": 0.04563241203625997,
      "eval_custom_ui_val_MAE_y_boxes": 0.02151624123669333,
      "eval_custom_ui_val_inside_bbox": 0.6712962985038757,
      "eval_custom_ui_val_loss": 1.345447063446045,
      "eval_custom_ui_val_loss_ce": 0.0011988434901771445,
      "eval_custom_ui_val_loss_iou": 0.5582411024305556,
      "eval_custom_ui_val_loss_num": 0.03997251722547743,
      "eval_custom_ui_val_loss_xval": 1.3163519965277777,
      "eval_custom_ui_val_runtime": 55.6004,
      "eval_custom_ui_val_samples_per_second": 4.766,
      "eval_custom_ui_val_steps_per_second": 0.162,
      "num_input_tokens_seen": 55979796,
      "step": 1000
    },
    {
      "epoch": 2.2271714922048997,
      "loss": 1.0719767808914185,
      "loss_ce": 0.000687746680341661,
      "loss_iou": 0.45703125,
      "loss_num": 0.03125,
      "loss_xval": 1.0703125,
      "num_input_tokens_seen": 55979796,
      "step": 1000
    },
    {
      "epoch": 2.2293986636971046,
      "grad_norm": 23.139463424682617,
      "learning_rate": 1e-06,
      "loss": 1.2981,
      "num_input_tokens_seen": 56034828,
      "step": 1001
    },
    {
      "epoch": 2.2293986636971046,
      "loss": 1.173054575920105,
      "loss_ce": 0.0004471880674827844,
      "loss_iou": 0.484375,
      "loss_num": 0.04052734375,
      "loss_xval": 1.171875,
      "num_input_tokens_seen": 56034828,
      "step": 1001
    },
    {
      "epoch": 2.2316258351893095,
      "grad_norm": 14.833909034729004,
      "learning_rate": 1e-06,
      "loss": 0.7852,
      "num_input_tokens_seen": 56090360,
      "step": 1002
    },
    {
      "epoch": 2.2316258351893095,
      "loss": 0.7579194903373718,
      "loss_ce": 0.00035115116043016315,
      "loss_iou": 0.32421875,
      "loss_num": 0.0216064453125,
      "loss_xval": 0.7578125,
      "num_input_tokens_seen": 56090360,
      "step": 1002
    },
    {
      "epoch": 2.2338530066815143,
      "grad_norm": 15.422577857971191,
      "learning_rate": 1e-06,
      "loss": 0.9412,
      "num_input_tokens_seen": 56146060,
      "step": 1003
    },
    {
      "epoch": 2.2338530066815143,
      "loss": 1.1035232543945312,
      "loss_ce": 0.0006180062773637474,
      "loss_iou": 0.474609375,
      "loss_num": 0.030517578125,
      "loss_xval": 1.1015625,
      "num_input_tokens_seen": 56146060,
      "step": 1003
    },
    {
      "epoch": 2.236080178173719,
      "grad_norm": 18.446163177490234,
      "learning_rate": 1e-06,
      "loss": 0.7755,
      "num_input_tokens_seen": 56202244,
      "step": 1004
    },
    {
      "epoch": 2.236080178173719,
      "loss": 0.8598674535751343,
      "loss_ce": 0.000492453167680651,
      "loss_iou": 0.349609375,
      "loss_num": 0.031982421875,
      "loss_xval": 0.859375,
      "num_input_tokens_seen": 56202244,
      "step": 1004
    },
    {
      "epoch": 2.2383073496659245,
      "grad_norm": 15.064460754394531,
      "learning_rate": 1e-06,
      "loss": 0.9762,
      "num_input_tokens_seen": 56257360,
      "step": 1005
    },
    {
      "epoch": 2.2383073496659245,
      "loss": 1.0436865091323853,
      "loss_ce": 0.0007178318337537348,
      "loss_iou": 0.404296875,
      "loss_num": 0.047119140625,
      "loss_xval": 1.046875,
      "num_input_tokens_seen": 56257360,
      "step": 1005
    },
    {
      "epoch": 2.2405345211581293,
      "grad_norm": 37.368812561035156,
      "learning_rate": 1e-06,
      "loss": 1.0937,
      "num_input_tokens_seen": 56309892,
      "step": 1006
    },
    {
      "epoch": 2.2405345211581293,
      "loss": 0.8646106123924255,
      "loss_ce": 0.00035280632437206805,
      "loss_iou": 0.3671875,
      "loss_num": 0.0257568359375,
      "loss_xval": 0.86328125,
      "num_input_tokens_seen": 56309892,
      "step": 1006
    },
    {
      "epoch": 2.242761692650334,
      "grad_norm": 20.58150863647461,
      "learning_rate": 1e-06,
      "loss": 1.0268,
      "num_input_tokens_seen": 56365936,
      "step": 1007
    },
    {
      "epoch": 2.242761692650334,
      "loss": 1.112687587738037,
      "loss_ce": 0.00038295358535833657,
      "loss_iou": 0.443359375,
      "loss_num": 0.044921875,
      "loss_xval": 1.109375,
      "num_input_tokens_seen": 56365936,
      "step": 1007
    },
    {
      "epoch": 2.244988864142539,
      "grad_norm": 19.283090591430664,
      "learning_rate": 1e-06,
      "loss": 0.9734,
      "num_input_tokens_seen": 56422576,
      "step": 1008
    },
    {
      "epoch": 2.244988864142539,
      "loss": 1.0260212421417236,
      "loss_ce": 0.00038648463669233024,
      "loss_iou": 0.4453125,
      "loss_num": 0.02685546875,
      "loss_xval": 1.0234375,
      "num_input_tokens_seen": 56422576,
      "step": 1008
    },
    {
      "epoch": 2.247216035634744,
      "grad_norm": 15.9085111618042,
      "learning_rate": 1e-06,
      "loss": 1.1175,
      "num_input_tokens_seen": 56481320,
      "step": 1009
    },
    {
      "epoch": 2.247216035634744,
      "loss": 0.8886697888374329,
      "loss_ce": 0.0004862197383772582,
      "loss_iou": 0.33203125,
      "loss_num": 0.04443359375,
      "loss_xval": 0.88671875,
      "num_input_tokens_seen": 56481320,
      "step": 1009
    },
    {
      "epoch": 2.249443207126949,
      "grad_norm": 20.693801879882812,
      "learning_rate": 1e-06,
      "loss": 0.8815,
      "num_input_tokens_seen": 56537736,
      "step": 1010
    },
    {
      "epoch": 2.249443207126949,
      "loss": 0.7114344835281372,
      "loss_ce": 0.000619084108620882,
      "loss_iou": 0.294921875,
      "loss_num": 0.024658203125,
      "loss_xval": 0.7109375,
      "num_input_tokens_seen": 56537736,
      "step": 1010
    },
    {
      "epoch": 2.2516703786191536,
      "grad_norm": 16.531198501586914,
      "learning_rate": 1e-06,
      "loss": 1.0092,
      "num_input_tokens_seen": 56594736,
      "step": 1011
    },
    {
      "epoch": 2.2516703786191536,
      "loss": 0.9720375537872314,
      "loss_ce": 0.0003578157047741115,
      "loss_iou": 0.419921875,
      "loss_num": 0.0264892578125,
      "loss_xval": 0.97265625,
      "num_input_tokens_seen": 56594736,
      "step": 1011
    },
    {
      "epoch": 2.2538975501113585,
      "grad_norm": 36.781532287597656,
      "learning_rate": 1e-06,
      "loss": 1.0224,
      "num_input_tokens_seen": 56651788,
      "step": 1012
    },
    {
      "epoch": 2.2538975501113585,
      "loss": 1.0697126388549805,
      "loss_ce": 0.00037672443431802094,
      "loss_iou": 0.462890625,
      "loss_num": 0.0281982421875,
      "loss_xval": 1.0703125,
      "num_input_tokens_seen": 56651788,
      "step": 1012
    },
    {
      "epoch": 2.2561247216035634,
      "grad_norm": 18.51349449157715,
      "learning_rate": 1e-06,
      "loss": 0.7345,
      "num_input_tokens_seen": 56707000,
      "step": 1013
    },
    {
      "epoch": 2.2561247216035634,
      "loss": 0.8490347862243652,
      "loss_ce": 0.000401980709284544,
      "loss_iou": 0.3828125,
      "loss_num": 0.0166015625,
      "loss_xval": 0.84765625,
      "num_input_tokens_seen": 56707000,
      "step": 1013
    },
    {
      "epoch": 2.2583518930957682,
      "grad_norm": 23.15570068359375,
      "learning_rate": 1e-06,
      "loss": 1.0084,
      "num_input_tokens_seen": 56761980,
      "step": 1014
    },
    {
      "epoch": 2.2583518930957682,
      "loss": 0.8266038298606873,
      "loss_ce": 0.0005540476413443685,
      "loss_iou": 0.34375,
      "loss_num": 0.02734375,
      "loss_xval": 0.82421875,
      "num_input_tokens_seen": 56761980,
      "step": 1014
    },
    {
      "epoch": 2.260579064587973,
      "grad_norm": 34.847137451171875,
      "learning_rate": 1e-06,
      "loss": 1.0461,
      "num_input_tokens_seen": 56820908,
      "step": 1015
    },
    {
      "epoch": 2.260579064587973,
      "loss": 0.9271408319473267,
      "loss_ce": 0.0006271736929193139,
      "loss_iou": 0.36328125,
      "loss_num": 0.0400390625,
      "loss_xval": 0.92578125,
      "num_input_tokens_seen": 56820908,
      "step": 1015
    },
    {
      "epoch": 2.262806236080178,
      "grad_norm": 34.261474609375,
      "learning_rate": 1e-06,
      "loss": 1.0021,
      "num_input_tokens_seen": 56877908,
      "step": 1016
    },
    {
      "epoch": 2.262806236080178,
      "loss": 0.912013590335846,
      "loss_ce": 0.00039250371628440917,
      "loss_iou": 0.384765625,
      "loss_num": 0.028076171875,
      "loss_xval": 0.91015625,
      "num_input_tokens_seen": 56877908,
      "step": 1016
    },
    {
      "epoch": 2.2650334075723833,
      "grad_norm": 30.62000846862793,
      "learning_rate": 1e-06,
      "loss": 1.1779,
      "num_input_tokens_seen": 56933112,
      "step": 1017
    },
    {
      "epoch": 2.2650334075723833,
      "loss": 1.2571220397949219,
      "loss_ce": 0.0012625895906239748,
      "loss_iou": 0.5078125,
      "loss_num": 0.04736328125,
      "loss_xval": 1.2578125,
      "num_input_tokens_seen": 56933112,
      "step": 1017
    },
    {
      "epoch": 2.267260579064588,
      "grad_norm": 21.855445861816406,
      "learning_rate": 1e-06,
      "loss": 0.8362,
      "num_input_tokens_seen": 56988708,
      "step": 1018
    },
    {
      "epoch": 2.267260579064588,
      "loss": 0.8795033097267151,
      "loss_ce": 0.00035289014340378344,
      "loss_iou": 0.380859375,
      "loss_num": 0.023193359375,
      "loss_xval": 0.87890625,
      "num_input_tokens_seen": 56988708,
      "step": 1018
    },
    {
      "epoch": 2.269487750556793,
      "grad_norm": 26.169057846069336,
      "learning_rate": 1e-06,
      "loss": 0.6786,
      "num_input_tokens_seen": 57046476,
      "step": 1019
    },
    {
      "epoch": 2.269487750556793,
      "loss": 0.7506150007247925,
      "loss_ce": 0.00037088984390720725,
      "loss_iou": 0.328125,
      "loss_num": 0.0191650390625,
      "loss_xval": 0.75,
      "num_input_tokens_seen": 57046476,
      "step": 1019
    },
    {
      "epoch": 2.271714922048998,
      "grad_norm": 13.190043449401855,
      "learning_rate": 1e-06,
      "loss": 0.6864,
      "num_input_tokens_seen": 57104372,
      "step": 1020
    },
    {
      "epoch": 2.271714922048998,
      "loss": 0.7020950317382812,
      "loss_ce": 0.0004348924267105758,
      "loss_iou": 0.298828125,
      "loss_num": 0.0211181640625,
      "loss_xval": 0.703125,
      "num_input_tokens_seen": 57104372,
      "step": 1020
    },
    {
      "epoch": 2.2739420935412027,
      "grad_norm": 16.335966110229492,
      "learning_rate": 1e-06,
      "loss": 0.7953,
      "num_input_tokens_seen": 57159292,
      "step": 1021
    },
    {
      "epoch": 2.2739420935412027,
      "loss": 0.8143704533576965,
      "loss_ce": 0.0004056024190504104,
      "loss_iou": 0.359375,
      "loss_num": 0.01904296875,
      "loss_xval": 0.8125,
      "num_input_tokens_seen": 57159292,
      "step": 1021
    },
    {
      "epoch": 2.2761692650334076,
      "grad_norm": 27.681427001953125,
      "learning_rate": 1e-06,
      "loss": 0.9381,
      "num_input_tokens_seen": 57216432,
      "step": 1022
    },
    {
      "epoch": 2.2761692650334076,
      "loss": 0.8490380048751831,
      "loss_ce": 0.0004051811702083796,
      "loss_iou": 0.369140625,
      "loss_num": 0.0223388671875,
      "loss_xval": 0.84765625,
      "num_input_tokens_seen": 57216432,
      "step": 1022
    },
    {
      "epoch": 2.2783964365256124,
      "grad_norm": 19.292463302612305,
      "learning_rate": 1e-06,
      "loss": 0.9677,
      "num_input_tokens_seen": 57273288,
      "step": 1023
    },
    {
      "epoch": 2.2783964365256124,
      "loss": 0.9046714305877686,
      "loss_ce": 0.000374538212781772,
      "loss_iou": 0.3515625,
      "loss_num": 0.04052734375,
      "loss_xval": 0.90625,
      "num_input_tokens_seen": 57273288,
      "step": 1023
    },
    {
      "epoch": 2.2806236080178173,
      "grad_norm": 22.14390754699707,
      "learning_rate": 1e-06,
      "loss": 0.8592,
      "num_input_tokens_seen": 57329664,
      "step": 1024
    },
    {
      "epoch": 2.2806236080178173,
      "loss": 1.0152591466903687,
      "loss_ce": 0.00036650500260293484,
      "loss_iou": 0.427734375,
      "loss_num": 0.031982421875,
      "loss_xval": 1.015625,
      "num_input_tokens_seen": 57329664,
      "step": 1024
    },
    {
      "epoch": 2.282850779510022,
      "grad_norm": 27.401857376098633,
      "learning_rate": 1e-06,
      "loss": 0.936,
      "num_input_tokens_seen": 57384724,
      "step": 1025
    },
    {
      "epoch": 2.282850779510022,
      "loss": 0.8641769886016846,
      "loss_ce": 0.00040744812577031553,
      "loss_iou": 0.357421875,
      "loss_num": 0.0294189453125,
      "loss_xval": 0.86328125,
      "num_input_tokens_seen": 57384724,
      "step": 1025
    },
    {
      "epoch": 2.285077951002227,
      "grad_norm": 13.289790153503418,
      "learning_rate": 1e-06,
      "loss": 0.8172,
      "num_input_tokens_seen": 57441660,
      "step": 1026
    },
    {
      "epoch": 2.285077951002227,
      "loss": 0.7961374521255493,
      "loss_ce": 0.0017038530204445124,
      "loss_iou": 0.30859375,
      "loss_num": 0.03515625,
      "loss_xval": 0.79296875,
      "num_input_tokens_seen": 57441660,
      "step": 1026
    },
    {
      "epoch": 2.2873051224944323,
      "grad_norm": 14.736815452575684,
      "learning_rate": 1e-06,
      "loss": 0.8734,
      "num_input_tokens_seen": 57497360,
      "step": 1027
    },
    {
      "epoch": 2.2873051224944323,
      "loss": 0.8420823216438293,
      "loss_ce": 0.0005296375602483749,
      "loss_iou": 0.345703125,
      "loss_num": 0.0299072265625,
      "loss_xval": 0.83984375,
      "num_input_tokens_seen": 57497360,
      "step": 1027
    },
    {
      "epoch": 2.289532293986637,
      "grad_norm": 18.238637924194336,
      "learning_rate": 1e-06,
      "loss": 1.3031,
      "num_input_tokens_seen": 57556032,
      "step": 1028
    },
    {
      "epoch": 2.289532293986637,
      "loss": 1.1549606323242188,
      "loss_ce": 0.0009079101146198809,
      "loss_iou": 0.44140625,
      "loss_num": 0.0546875,
      "loss_xval": 1.15625,
      "num_input_tokens_seen": 57556032,
      "step": 1028
    },
    {
      "epoch": 2.291759465478842,
      "grad_norm": 20.993566513061523,
      "learning_rate": 1e-06,
      "loss": 0.9124,
      "num_input_tokens_seen": 57610720,
      "step": 1029
    },
    {
      "epoch": 2.291759465478842,
      "loss": 0.8001816272735596,
      "loss_ce": 0.00031593156745657325,
      "loss_iou": 0.345703125,
      "loss_num": 0.0216064453125,
      "loss_xval": 0.80078125,
      "num_input_tokens_seen": 57610720,
      "step": 1029
    },
    {
      "epoch": 2.293986636971047,
      "grad_norm": 31.266921997070312,
      "learning_rate": 1e-06,
      "loss": 0.9937,
      "num_input_tokens_seen": 57669460,
      "step": 1030
    },
    {
      "epoch": 2.293986636971047,
      "loss": 1.138893485069275,
      "loss_ce": 0.0021747485734522343,
      "loss_iou": 0.45703125,
      "loss_num": 0.044921875,
      "loss_xval": 1.140625,
      "num_input_tokens_seen": 57669460,
      "step": 1030
    },
    {
      "epoch": 2.2962138084632517,
      "grad_norm": 29.804710388183594,
      "learning_rate": 1e-06,
      "loss": 0.9726,
      "num_input_tokens_seen": 57726024,
      "step": 1031
    },
    {
      "epoch": 2.2962138084632517,
      "loss": 1.0021300315856934,
      "loss_ce": 0.00042098466656170785,
      "loss_iou": 0.388671875,
      "loss_num": 0.045654296875,
      "loss_xval": 1.0,
      "num_input_tokens_seen": 57726024,
      "step": 1031
    },
    {
      "epoch": 2.2984409799554566,
      "grad_norm": 16.084012985229492,
      "learning_rate": 1e-06,
      "loss": 1.006,
      "num_input_tokens_seen": 57783076,
      "step": 1032
    },
    {
      "epoch": 2.2984409799554566,
      "loss": 1.1425867080688477,
      "loss_ce": 0.0004969405708834529,
      "loss_iou": 0.4609375,
      "loss_num": 0.04443359375,
      "loss_xval": 1.140625,
      "num_input_tokens_seen": 57783076,
      "step": 1032
    },
    {
      "epoch": 2.3006681514476615,
      "grad_norm": 27.19474983215332,
      "learning_rate": 1e-06,
      "loss": 0.992,
      "num_input_tokens_seen": 57837908,
      "step": 1033
    },
    {
      "epoch": 2.3006681514476615,
      "loss": 0.9429985284805298,
      "loss_ce": 0.0006157412426546216,
      "loss_iou": 0.3828125,
      "loss_num": 0.03564453125,
      "loss_xval": 0.94140625,
      "num_input_tokens_seen": 57837908,
      "step": 1033
    },
    {
      "epoch": 2.3028953229398663,
      "grad_norm": 30.541584014892578,
      "learning_rate": 1e-06,
      "loss": 1.1811,
      "num_input_tokens_seen": 57893948,
      "step": 1034
    },
    {
      "epoch": 2.3028953229398663,
      "loss": 1.2314056158065796,
      "loss_ce": 0.00044855731539428234,
      "loss_iou": 0.490234375,
      "loss_num": 0.049560546875,
      "loss_xval": 1.234375,
      "num_input_tokens_seen": 57893948,
      "step": 1034
    },
    {
      "epoch": 2.305122494432071,
      "grad_norm": 16.411218643188477,
      "learning_rate": 1e-06,
      "loss": 0.8391,
      "num_input_tokens_seen": 57950452,
      "step": 1035
    },
    {
      "epoch": 2.305122494432071,
      "loss": 0.8310631513595581,
      "loss_ce": 0.0004967194981873035,
      "loss_iou": 0.328125,
      "loss_num": 0.034423828125,
      "loss_xval": 0.83203125,
      "num_input_tokens_seen": 57950452,
      "step": 1035
    },
    {
      "epoch": 2.307349665924276,
      "grad_norm": 20.609291076660156,
      "learning_rate": 1e-06,
      "loss": 0.9038,
      "num_input_tokens_seen": 58006408,
      "step": 1036
    },
    {
      "epoch": 2.307349665924276,
      "loss": 0.7673088312149048,
      "loss_ce": 0.0004631901392713189,
      "loss_iou": 0.326171875,
      "loss_num": 0.0230712890625,
      "loss_xval": 0.765625,
      "num_input_tokens_seen": 58006408,
      "step": 1036
    },
    {
      "epoch": 2.309576837416481,
      "grad_norm": 31.944068908691406,
      "learning_rate": 1e-06,
      "loss": 1.2189,
      "num_input_tokens_seen": 58062244,
      "step": 1037
    },
    {
      "epoch": 2.309576837416481,
      "loss": 0.9620157480239868,
      "loss_ce": 0.0003457932034507394,
      "loss_iou": 0.396484375,
      "loss_num": 0.033447265625,
      "loss_xval": 0.9609375,
      "num_input_tokens_seen": 58062244,
      "step": 1037
    },
    {
      "epoch": 2.3118040089086858,
      "grad_norm": 28.491863250732422,
      "learning_rate": 1e-06,
      "loss": 1.0784,
      "num_input_tokens_seen": 58116528,
      "step": 1038
    },
    {
      "epoch": 2.3118040089086858,
      "loss": 0.8677387237548828,
      "loss_ce": 0.0003071002720389515,
      "loss_iou": 0.359375,
      "loss_num": 0.0299072265625,
      "loss_xval": 0.8671875,
      "num_input_tokens_seen": 58116528,
      "step": 1038
    },
    {
      "epoch": 2.3140311804008906,
      "grad_norm": 16.422298431396484,
      "learning_rate": 1e-06,
      "loss": 1.1027,
      "num_input_tokens_seen": 58172968,
      "step": 1039
    },
    {
      "epoch": 2.3140311804008906,
      "loss": 1.250427484512329,
      "loss_ce": 0.0006715654162690043,
      "loss_iou": 0.482421875,
      "loss_num": 0.05712890625,
      "loss_xval": 1.25,
      "num_input_tokens_seen": 58172968,
      "step": 1039
    },
    {
      "epoch": 2.316258351893096,
      "grad_norm": 33.38667297363281,
      "learning_rate": 1e-06,
      "loss": 1.3044,
      "num_input_tokens_seen": 58228600,
      "step": 1040
    },
    {
      "epoch": 2.316258351893096,
      "loss": 1.3572330474853516,
      "loss_ce": 0.00029941959655843675,
      "loss_iou": 0.5703125,
      "loss_num": 0.042724609375,
      "loss_xval": 1.359375,
      "num_input_tokens_seen": 58228600,
      "step": 1040
    },
    {
      "epoch": 2.318485523385301,
      "grad_norm": 25.701223373413086,
      "learning_rate": 1e-06,
      "loss": 0.8746,
      "num_input_tokens_seen": 58284388,
      "step": 1041
    },
    {
      "epoch": 2.318485523385301,
      "loss": 0.9398232102394104,
      "loss_ce": 0.0006142100319266319,
      "loss_iou": 0.412109375,
      "loss_num": 0.0233154296875,
      "loss_xval": 0.9375,
      "num_input_tokens_seen": 58284388,
      "step": 1041
    },
    {
      "epoch": 2.3207126948775056,
      "grad_norm": 30.38446807861328,
      "learning_rate": 1e-06,
      "loss": 0.9229,
      "num_input_tokens_seen": 58342404,
      "step": 1042
    },
    {
      "epoch": 2.3207126948775056,
      "loss": 1.0379189252853394,
      "loss_ce": 0.0005654151318594813,
      "loss_iou": 0.43359375,
      "loss_num": 0.033447265625,
      "loss_xval": 1.0390625,
      "num_input_tokens_seen": 58342404,
      "step": 1042
    },
    {
      "epoch": 2.3229398663697105,
      "grad_norm": 26.981040954589844,
      "learning_rate": 1e-06,
      "loss": 1.1894,
      "num_input_tokens_seen": 58398820,
      "step": 1043
    },
    {
      "epoch": 2.3229398663697105,
      "loss": 1.1476168632507324,
      "loss_ce": 0.0004000905028078705,
      "loss_iou": 0.494140625,
      "loss_num": 0.031494140625,
      "loss_xval": 1.1484375,
      "num_input_tokens_seen": 58398820,
      "step": 1043
    },
    {
      "epoch": 2.3251670378619154,
      "grad_norm": 17.203502655029297,
      "learning_rate": 1e-06,
      "loss": 1.1498,
      "num_input_tokens_seen": 58455324,
      "step": 1044
    },
    {
      "epoch": 2.3251670378619154,
      "loss": 1.3662068843841553,
      "loss_ce": 0.00048417344805784523,
      "loss_iou": 0.578125,
      "loss_num": 0.04248046875,
      "loss_xval": 1.3671875,
      "num_input_tokens_seen": 58455324,
      "step": 1044
    },
    {
      "epoch": 2.3273942093541202,
      "grad_norm": 26.704832077026367,
      "learning_rate": 1e-06,
      "loss": 1.1157,
      "num_input_tokens_seen": 58512188,
      "step": 1045
    },
    {
      "epoch": 2.3273942093541202,
      "loss": 1.0862228870391846,
      "loss_ce": 0.000529451877810061,
      "loss_iou": 0.451171875,
      "loss_num": 0.03662109375,
      "loss_xval": 1.0859375,
      "num_input_tokens_seen": 58512188,
      "step": 1045
    },
    {
      "epoch": 2.329621380846325,
      "grad_norm": 18.440515518188477,
      "learning_rate": 1e-06,
      "loss": 0.9097,
      "num_input_tokens_seen": 58570316,
      "step": 1046
    },
    {
      "epoch": 2.329621380846325,
      "loss": 0.8449282050132751,
      "loss_ce": 0.0004457966424524784,
      "loss_iou": 0.35546875,
      "loss_num": 0.0262451171875,
      "loss_xval": 0.84375,
      "num_input_tokens_seen": 58570316,
      "step": 1046
    },
    {
      "epoch": 2.33184855233853,
      "grad_norm": 15.037751197814941,
      "learning_rate": 1e-06,
      "loss": 0.7533,
      "num_input_tokens_seen": 58628892,
      "step": 1047
    },
    {
      "epoch": 2.33184855233853,
      "loss": 0.7818921804428101,
      "loss_ce": 0.00039804144762456417,
      "loss_iou": 0.353515625,
      "loss_num": 0.01507568359375,
      "loss_xval": 0.78125,
      "num_input_tokens_seen": 58628892,
      "step": 1047
    },
    {
      "epoch": 2.334075723830735,
      "grad_norm": 19.363588333129883,
      "learning_rate": 1e-06,
      "loss": 0.8876,
      "num_input_tokens_seen": 58685104,
      "step": 1048
    },
    {
      "epoch": 2.334075723830735,
      "loss": 1.0165672302246094,
      "loss_ce": 0.000453831598861143,
      "loss_iou": 0.408203125,
      "loss_num": 0.039794921875,
      "loss_xval": 1.015625,
      "num_input_tokens_seen": 58685104,
      "step": 1048
    },
    {
      "epoch": 2.33630289532294,
      "grad_norm": 28.217439651489258,
      "learning_rate": 1e-06,
      "loss": 1.1681,
      "num_input_tokens_seen": 58738936,
      "step": 1049
    },
    {
      "epoch": 2.33630289532294,
      "loss": 1.4371428489685059,
      "loss_ce": 0.0006193204899318516,
      "loss_iou": 0.55078125,
      "loss_num": 0.06787109375,
      "loss_xval": 1.4375,
      "num_input_tokens_seen": 58738936,
      "step": 1049
    },
    {
      "epoch": 2.338530066815145,
      "grad_norm": 17.534543991088867,
      "learning_rate": 1e-06,
      "loss": 0.6902,
      "num_input_tokens_seen": 58798988,
      "step": 1050
    },
    {
      "epoch": 2.338530066815145,
      "loss": 0.47515368461608887,
      "loss_ce": 0.00030017929384484887,
      "loss_iou": 0.212890625,
      "loss_num": 0.0098876953125,
      "loss_xval": 0.474609375,
      "num_input_tokens_seen": 58798988,
      "step": 1050
    },
    {
      "epoch": 2.34075723830735,
      "grad_norm": 18.249549865722656,
      "learning_rate": 1e-06,
      "loss": 0.9676,
      "num_input_tokens_seen": 58854660,
      "step": 1051
    },
    {
      "epoch": 2.34075723830735,
      "loss": 0.977979302406311,
      "loss_ce": 0.0004402369959279895,
      "loss_iou": 0.40234375,
      "loss_num": 0.0341796875,
      "loss_xval": 0.9765625,
      "num_input_tokens_seen": 58854660,
      "step": 1051
    },
    {
      "epoch": 2.3429844097995547,
      "grad_norm": 21.75814437866211,
      "learning_rate": 1e-06,
      "loss": 0.9833,
      "num_input_tokens_seen": 58911232,
      "step": 1052
    },
    {
      "epoch": 2.3429844097995547,
      "loss": 1.1573677062988281,
      "loss_ce": 0.0005073855281807482,
      "loss_iou": 0.484375,
      "loss_num": 0.0380859375,
      "loss_xval": 1.15625,
      "num_input_tokens_seen": 58911232,
      "step": 1052
    },
    {
      "epoch": 2.3452115812917596,
      "grad_norm": 18.00012969970703,
      "learning_rate": 1e-06,
      "loss": 0.7443,
      "num_input_tokens_seen": 58968392,
      "step": 1053
    },
    {
      "epoch": 2.3452115812917596,
      "loss": 0.7719112634658813,
      "loss_ce": 0.0004268469929229468,
      "loss_iou": 0.314453125,
      "loss_num": 0.0286865234375,
      "loss_xval": 0.7734375,
      "num_input_tokens_seen": 58968392,
      "step": 1053
    },
    {
      "epoch": 2.3474387527839644,
      "grad_norm": 17.328489303588867,
      "learning_rate": 1e-06,
      "loss": 0.9562,
      "num_input_tokens_seen": 59026080,
      "step": 1054
    },
    {
      "epoch": 2.3474387527839644,
      "loss": 0.7599896192550659,
      "loss_ce": 0.00046816159738227725,
      "loss_iou": 0.330078125,
      "loss_num": 0.019775390625,
      "loss_xval": 0.7578125,
      "num_input_tokens_seen": 59026080,
      "step": 1054
    },
    {
      "epoch": 2.3496659242761693,
      "grad_norm": 24.498754501342773,
      "learning_rate": 1e-06,
      "loss": 1.0251,
      "num_input_tokens_seen": 59082868,
      "step": 1055
    },
    {
      "epoch": 2.3496659242761693,
      "loss": 1.0223811864852905,
      "loss_ce": 0.0024837690871208906,
      "loss_iou": 0.443359375,
      "loss_num": 0.0267333984375,
      "loss_xval": 1.0234375,
      "num_input_tokens_seen": 59082868,
      "step": 1055
    },
    {
      "epoch": 2.351893095768374,
      "grad_norm": 258.32928466796875,
      "learning_rate": 1e-06,
      "loss": 1.0218,
      "num_input_tokens_seen": 59139812,
      "step": 1056
    },
    {
      "epoch": 2.351893095768374,
      "loss": 0.9473181962966919,
      "loss_ce": 0.0005408285651355982,
      "loss_iou": 0.3828125,
      "loss_num": 0.0361328125,
      "loss_xval": 0.9453125,
      "num_input_tokens_seen": 59139812,
      "step": 1056
    },
    {
      "epoch": 2.354120267260579,
      "grad_norm": 41.9471435546875,
      "learning_rate": 1e-06,
      "loss": 0.9818,
      "num_input_tokens_seen": 59195536,
      "step": 1057
    },
    {
      "epoch": 2.354120267260579,
      "loss": 1.0145529508590698,
      "loss_ce": 0.0003928400401491672,
      "loss_iou": 0.44140625,
      "loss_num": 0.02587890625,
      "loss_xval": 1.015625,
      "num_input_tokens_seen": 59195536,
      "step": 1057
    },
    {
      "epoch": 2.356347438752784,
      "grad_norm": 20.948766708374023,
      "learning_rate": 1e-06,
      "loss": 1.0111,
      "num_input_tokens_seen": 59252604,
      "step": 1058
    },
    {
      "epoch": 2.356347438752784,
      "loss": 1.2552516460418701,
      "loss_ce": 0.0003688871511258185,
      "loss_iou": 0.4921875,
      "loss_num": 0.053955078125,
      "loss_xval": 1.2578125,
      "num_input_tokens_seen": 59252604,
      "step": 1058
    },
    {
      "epoch": 2.3585746102449887,
      "grad_norm": 33.876991271972656,
      "learning_rate": 1e-06,
      "loss": 0.8549,
      "num_input_tokens_seen": 59308408,
      "step": 1059
    },
    {
      "epoch": 2.3585746102449887,
      "loss": 0.6373202204704285,
      "loss_ce": 0.0003573211724869907,
      "loss_iou": 0.251953125,
      "loss_num": 0.0269775390625,
      "loss_xval": 0.63671875,
      "num_input_tokens_seen": 59308408,
      "step": 1059
    },
    {
      "epoch": 2.3608017817371936,
      "grad_norm": 18.77747917175293,
      "learning_rate": 1e-06,
      "loss": 0.7226,
      "num_input_tokens_seen": 59365048,
      "step": 1060
    },
    {
      "epoch": 2.3608017817371936,
      "loss": 0.7437995076179504,
      "loss_ce": 0.0008795711910352111,
      "loss_iou": 0.31640625,
      "loss_num": 0.0220947265625,
      "loss_xval": 0.7421875,
      "num_input_tokens_seen": 59365048,
      "step": 1060
    },
    {
      "epoch": 2.3630289532293984,
      "grad_norm": 28.6199893951416,
      "learning_rate": 1e-06,
      "loss": 1.1339,
      "num_input_tokens_seen": 59421260,
      "step": 1061
    },
    {
      "epoch": 2.3630289532293984,
      "loss": 0.9527925848960876,
      "loss_ce": 0.000644176616333425,
      "loss_iou": 0.396484375,
      "loss_num": 0.03173828125,
      "loss_xval": 0.953125,
      "num_input_tokens_seen": 59421260,
      "step": 1061
    },
    {
      "epoch": 2.3652561247216037,
      "grad_norm": 60.87717056274414,
      "learning_rate": 1e-06,
      "loss": 0.9096,
      "num_input_tokens_seen": 59479676,
      "step": 1062
    },
    {
      "epoch": 2.3652561247216037,
      "loss": 0.817761242389679,
      "loss_ce": 0.0006225479301065207,
      "loss_iou": 0.34765625,
      "loss_num": 0.024658203125,
      "loss_xval": 0.81640625,
      "num_input_tokens_seen": 59479676,
      "step": 1062
    },
    {
      "epoch": 2.3674832962138086,
      "grad_norm": 21.296977996826172,
      "learning_rate": 1e-06,
      "loss": 0.9039,
      "num_input_tokens_seen": 59534276,
      "step": 1063
    },
    {
      "epoch": 2.3674832962138086,
      "loss": 0.9175959825515747,
      "loss_ce": 0.0003596206079237163,
      "loss_iou": 0.35546875,
      "loss_num": 0.041015625,
      "loss_xval": 0.91796875,
      "num_input_tokens_seen": 59534276,
      "step": 1063
    },
    {
      "epoch": 2.3697104677060135,
      "grad_norm": 28.44460105895996,
      "learning_rate": 1e-06,
      "loss": 1.0105,
      "num_input_tokens_seen": 59589900,
      "step": 1064
    },
    {
      "epoch": 2.3697104677060135,
      "loss": 1.1641950607299805,
      "loss_ce": 0.0006208861595951021,
      "loss_iou": 0.482421875,
      "loss_num": 0.03955078125,
      "loss_xval": 1.1640625,
      "num_input_tokens_seen": 59589900,
      "step": 1064
    },
    {
      "epoch": 2.3719376391982183,
      "grad_norm": 35.73344421386719,
      "learning_rate": 1e-06,
      "loss": 0.8737,
      "num_input_tokens_seen": 59649136,
      "step": 1065
    },
    {
      "epoch": 2.3719376391982183,
      "loss": 0.921160101890564,
      "loss_ce": 0.000505772652104497,
      "loss_iou": 0.37109375,
      "loss_num": 0.03564453125,
      "loss_xval": 0.921875,
      "num_input_tokens_seen": 59649136,
      "step": 1065
    },
    {
      "epoch": 2.374164810690423,
      "grad_norm": 22.40777015686035,
      "learning_rate": 1e-06,
      "loss": 0.9714,
      "num_input_tokens_seen": 59701696,
      "step": 1066
    },
    {
      "epoch": 2.374164810690423,
      "loss": 0.939410388469696,
      "loss_ce": 0.00044557781075127423,
      "loss_iou": 0.390625,
      "loss_num": 0.031982421875,
      "loss_xval": 0.9375,
      "num_input_tokens_seen": 59701696,
      "step": 1066
    },
    {
      "epoch": 2.376391982182628,
      "grad_norm": 19.74895668029785,
      "learning_rate": 1e-06,
      "loss": 0.7167,
      "num_input_tokens_seen": 59759464,
      "step": 1067
    },
    {
      "epoch": 2.376391982182628,
      "loss": 0.7840393781661987,
      "loss_ce": 0.00034799351124092937,
      "loss_iou": 0.333984375,
      "loss_num": 0.0230712890625,
      "loss_xval": 0.78515625,
      "num_input_tokens_seen": 59759464,
      "step": 1067
    },
    {
      "epoch": 2.378619153674833,
      "grad_norm": 45.2192268371582,
      "learning_rate": 1e-06,
      "loss": 0.9642,
      "num_input_tokens_seen": 59813332,
      "step": 1068
    },
    {
      "epoch": 2.378619153674833,
      "loss": 0.961778998374939,
      "loss_ce": 0.0003532259142957628,
      "loss_iou": 0.404296875,
      "loss_num": 0.030517578125,
      "loss_xval": 0.9609375,
      "num_input_tokens_seen": 59813332,
      "step": 1068
    },
    {
      "epoch": 2.3808463251670378,
      "grad_norm": 21.226028442382812,
      "learning_rate": 1e-06,
      "loss": 0.8399,
      "num_input_tokens_seen": 59867084,
      "step": 1069
    },
    {
      "epoch": 2.3808463251670378,
      "loss": 0.6262215375900269,
      "loss_ce": 0.00036704522790387273,
      "loss_iou": 0.2294921875,
      "loss_num": 0.033447265625,
      "loss_xval": 0.625,
      "num_input_tokens_seen": 59867084,
      "step": 1069
    },
    {
      "epoch": 2.3830734966592426,
      "grad_norm": 20.79296112060547,
      "learning_rate": 1e-06,
      "loss": 0.949,
      "num_input_tokens_seen": 59923596,
      "step": 1070
    },
    {
      "epoch": 2.3830734966592426,
      "loss": 0.9344473481178284,
      "loss_ce": 0.0003653277817647904,
      "loss_iou": 0.39453125,
      "loss_num": 0.029052734375,
      "loss_xval": 0.93359375,
      "num_input_tokens_seen": 59923596,
      "step": 1070
    },
    {
      "epoch": 2.3853006681514475,
      "grad_norm": 23.782529830932617,
      "learning_rate": 1e-06,
      "loss": 0.9666,
      "num_input_tokens_seen": 59978588,
      "step": 1071
    },
    {
      "epoch": 2.3853006681514475,
      "loss": 0.9463720917701721,
      "loss_ce": 0.0005713239079341292,
      "loss_iou": 0.3828125,
      "loss_num": 0.03564453125,
      "loss_xval": 0.9453125,
      "num_input_tokens_seen": 59978588,
      "step": 1071
    },
    {
      "epoch": 2.387527839643653,
      "grad_norm": 25.347000122070312,
      "learning_rate": 1e-06,
      "loss": 1.0797,
      "num_input_tokens_seen": 60035216,
      "step": 1072
    },
    {
      "epoch": 2.387527839643653,
      "loss": 1.156226396560669,
      "loss_ce": 0.0004647444002330303,
      "loss_iou": 0.482421875,
      "loss_num": 0.037841796875,
      "loss_xval": 1.15625,
      "num_input_tokens_seen": 60035216,
      "step": 1072
    },
    {
      "epoch": 2.3897550111358576,
      "grad_norm": 104.3314208984375,
      "learning_rate": 1e-06,
      "loss": 0.8764,
      "num_input_tokens_seen": 60089212,
      "step": 1073
    },
    {
      "epoch": 2.3897550111358576,
      "loss": 0.868368923664093,
      "loss_ce": 0.00044899751082994044,
      "loss_iou": 0.3671875,
      "loss_num": 0.02685546875,
      "loss_xval": 0.8671875,
      "num_input_tokens_seen": 60089212,
      "step": 1073
    },
    {
      "epoch": 2.3919821826280625,
      "grad_norm": 17.936288833618164,
      "learning_rate": 1e-06,
      "loss": 1.1517,
      "num_input_tokens_seen": 60145076,
      "step": 1074
    },
    {
      "epoch": 2.3919821826280625,
      "loss": 1.1004548072814941,
      "loss_ce": 0.0006012015510350466,
      "loss_iou": 0.431640625,
      "loss_num": 0.0478515625,
      "loss_xval": 1.1015625,
      "num_input_tokens_seen": 60145076,
      "step": 1074
    },
    {
      "epoch": 2.3942093541202674,
      "grad_norm": 71.37808227539062,
      "learning_rate": 1e-06,
      "loss": 0.9387,
      "num_input_tokens_seen": 60201880,
      "step": 1075
    },
    {
      "epoch": 2.3942093541202674,
      "loss": 0.8587861061096191,
      "loss_ce": 0.00038765568751841784,
      "loss_iou": 0.35546875,
      "loss_num": 0.0294189453125,
      "loss_xval": 0.859375,
      "num_input_tokens_seen": 60201880,
      "step": 1075
    },
    {
      "epoch": 2.3964365256124722,
      "grad_norm": 15.348673820495605,
      "learning_rate": 1e-06,
      "loss": 0.7725,
      "num_input_tokens_seen": 60257620,
      "step": 1076
    },
    {
      "epoch": 2.3964365256124722,
      "loss": 0.8453316688537598,
      "loss_ce": 0.00036094876122660935,
      "loss_iou": 0.373046875,
      "loss_num": 0.0196533203125,
      "loss_xval": 0.84375,
      "num_input_tokens_seen": 60257620,
      "step": 1076
    },
    {
      "epoch": 2.398663697104677,
      "grad_norm": 13.168649673461914,
      "learning_rate": 1e-06,
      "loss": 0.7955,
      "num_input_tokens_seen": 60315036,
      "step": 1077
    },
    {
      "epoch": 2.398663697104677,
      "loss": 1.0594782829284668,
      "loss_ce": 0.000884645152837038,
      "loss_iou": 0.435546875,
      "loss_num": 0.037353515625,
      "loss_xval": 1.0625,
      "num_input_tokens_seen": 60315036,
      "step": 1077
    },
    {
      "epoch": 2.400890868596882,
      "grad_norm": 22.153902053833008,
      "learning_rate": 1e-06,
      "loss": 0.9847,
      "num_input_tokens_seen": 60371132,
      "step": 1078
    },
    {
      "epoch": 2.400890868596882,
      "loss": 0.6142248511314392,
      "loss_ce": 0.0004552791069727391,
      "loss_iou": 0.2470703125,
      "loss_num": 0.0240478515625,
      "loss_xval": 0.61328125,
      "num_input_tokens_seen": 60371132,
      "step": 1078
    },
    {
      "epoch": 2.403118040089087,
      "grad_norm": 16.075576782226562,
      "learning_rate": 1e-06,
      "loss": 0.9945,
      "num_input_tokens_seen": 60425052,
      "step": 1079
    },
    {
      "epoch": 2.403118040089087,
      "loss": 0.8446251153945923,
      "loss_ce": 0.0003868020430672914,
      "loss_iou": 0.34765625,
      "loss_num": 0.030029296875,
      "loss_xval": 0.84375,
      "num_input_tokens_seen": 60425052,
      "step": 1079
    },
    {
      "epoch": 2.4053452115812917,
      "grad_norm": 30.00712013244629,
      "learning_rate": 1e-06,
      "loss": 0.9475,
      "num_input_tokens_seen": 60481796,
      "step": 1080
    },
    {
      "epoch": 2.4053452115812917,
      "loss": 1.1728054285049438,
      "loss_ce": 0.0004421064513735473,
      "loss_iou": 0.46875,
      "loss_num": 0.047119140625,
      "loss_xval": 1.171875,
      "num_input_tokens_seen": 60481796,
      "step": 1080
    },
    {
      "epoch": 2.4075723830734965,
      "grad_norm": 37.473915100097656,
      "learning_rate": 1e-06,
      "loss": 1.0343,
      "num_input_tokens_seen": 60535540,
      "step": 1081
    },
    {
      "epoch": 2.4075723830734965,
      "loss": 0.8917477130889893,
      "loss_ce": 0.00039027928141877055,
      "loss_iou": 0.369140625,
      "loss_num": 0.0308837890625,
      "loss_xval": 0.890625,
      "num_input_tokens_seen": 60535540,
      "step": 1081
    },
    {
      "epoch": 2.4097995545657014,
      "grad_norm": 26.055566787719727,
      "learning_rate": 1e-06,
      "loss": 0.9813,
      "num_input_tokens_seen": 60591724,
      "step": 1082
    },
    {
      "epoch": 2.4097995545657014,
      "loss": 0.8454374670982361,
      "loss_ce": 0.00046675774501636624,
      "loss_iou": 0.341796875,
      "loss_num": 0.0322265625,
      "loss_xval": 0.84375,
      "num_input_tokens_seen": 60591724,
      "step": 1082
    },
    {
      "epoch": 2.4120267260579062,
      "grad_norm": 19.031705856323242,
      "learning_rate": 1e-06,
      "loss": 0.9273,
      "num_input_tokens_seen": 60646416,
      "step": 1083
    },
    {
      "epoch": 2.4120267260579062,
      "loss": 1.0545891523361206,
      "loss_ce": 0.000389896216802299,
      "loss_iou": 0.458984375,
      "loss_num": 0.0277099609375,
      "loss_xval": 1.0546875,
      "num_input_tokens_seen": 60646416,
      "step": 1083
    },
    {
      "epoch": 2.4142538975501115,
      "grad_norm": 13.565470695495605,
      "learning_rate": 1e-06,
      "loss": 0.8106,
      "num_input_tokens_seen": 60703608,
      "step": 1084
    },
    {
      "epoch": 2.4142538975501115,
      "loss": 0.821467399597168,
      "loss_ce": 0.0004224562435410917,
      "loss_iou": 0.3203125,
      "loss_num": 0.036376953125,
      "loss_xval": 0.8203125,
      "num_input_tokens_seen": 60703608,
      "step": 1084
    },
    {
      "epoch": 2.4164810690423164,
      "grad_norm": 12.959884643554688,
      "learning_rate": 1e-06,
      "loss": 0.8867,
      "num_input_tokens_seen": 60760392,
      "step": 1085
    },
    {
      "epoch": 2.4164810690423164,
      "loss": 0.7961212396621704,
      "loss_ce": 0.00046693626791238785,
      "loss_iou": 0.310546875,
      "loss_num": 0.034912109375,
      "loss_xval": 0.796875,
      "num_input_tokens_seen": 60760392,
      "step": 1085
    },
    {
      "epoch": 2.4187082405345213,
      "grad_norm": 15.361568450927734,
      "learning_rate": 1e-06,
      "loss": 0.8654,
      "num_input_tokens_seen": 60817288,
      "step": 1086
    },
    {
      "epoch": 2.4187082405345213,
      "loss": 0.6870740056037903,
      "loss_ce": 0.00030639575561508536,
      "loss_iou": 0.294921875,
      "loss_num": 0.01953125,
      "loss_xval": 0.6875,
      "num_input_tokens_seen": 60817288,
      "step": 1086
    },
    {
      "epoch": 2.420935412026726,
      "grad_norm": 22.026248931884766,
      "learning_rate": 1e-06,
      "loss": 1.1047,
      "num_input_tokens_seen": 60874444,
      "step": 1087
    },
    {
      "epoch": 2.420935412026726,
      "loss": 0.8688787221908569,
      "loss_ce": 0.0007146652205847204,
      "loss_iou": 0.3828125,
      "loss_num": 0.0208740234375,
      "loss_xval": 0.8671875,
      "num_input_tokens_seen": 60874444,
      "step": 1087
    },
    {
      "epoch": 2.423162583518931,
      "grad_norm": 20.72212028503418,
      "learning_rate": 1e-06,
      "loss": 1.1223,
      "num_input_tokens_seen": 60932720,
      "step": 1088
    },
    {
      "epoch": 2.423162583518931,
      "loss": 1.190258264541626,
      "loss_ce": 0.00080519710900262,
      "loss_iou": 0.4609375,
      "loss_num": 0.0537109375,
      "loss_xval": 1.1875,
      "num_input_tokens_seen": 60932720,
      "step": 1088
    },
    {
      "epoch": 2.425389755011136,
      "grad_norm": 22.80421257019043,
      "learning_rate": 1e-06,
      "loss": 1.0687,
      "num_input_tokens_seen": 60990336,
      "step": 1089
    },
    {
      "epoch": 2.425389755011136,
      "loss": 1.2014228105545044,
      "loss_ce": 0.0007392432307824492,
      "loss_iou": 0.4921875,
      "loss_num": 0.04345703125,
      "loss_xval": 1.203125,
      "num_input_tokens_seen": 60990336,
      "step": 1089
    },
    {
      "epoch": 2.4276169265033407,
      "grad_norm": 23.451295852661133,
      "learning_rate": 1e-06,
      "loss": 1.0575,
      "num_input_tokens_seen": 61045524,
      "step": 1090
    },
    {
      "epoch": 2.4276169265033407,
      "loss": 1.208905816078186,
      "loss_ce": 0.000409751373808831,
      "loss_iou": 0.49609375,
      "loss_num": 0.043212890625,
      "loss_xval": 1.2109375,
      "num_input_tokens_seen": 61045524,
      "step": 1090
    },
    {
      "epoch": 2.4298440979955456,
      "grad_norm": 17.383296966552734,
      "learning_rate": 1e-06,
      "loss": 0.7853,
      "num_input_tokens_seen": 61104956,
      "step": 1091
    },
    {
      "epoch": 2.4298440979955456,
      "loss": 0.7447309494018555,
      "loss_ce": 0.000834420439787209,
      "loss_iou": 0.32421875,
      "loss_num": 0.0189208984375,
      "loss_xval": 0.7421875,
      "num_input_tokens_seen": 61104956,
      "step": 1091
    },
    {
      "epoch": 2.4320712694877504,
      "grad_norm": 18.50430679321289,
      "learning_rate": 1e-06,
      "loss": 0.9737,
      "num_input_tokens_seen": 61159236,
      "step": 1092
    },
    {
      "epoch": 2.4320712694877504,
      "loss": 0.9158762693405151,
      "loss_ce": 0.00034890550887212157,
      "loss_iou": 0.38671875,
      "loss_num": 0.028076171875,
      "loss_xval": 0.9140625,
      "num_input_tokens_seen": 61159236,
      "step": 1092
    },
    {
      "epoch": 2.4342984409799553,
      "grad_norm": 18.052396774291992,
      "learning_rate": 1e-06,
      "loss": 1.0082,
      "num_input_tokens_seen": 61217816,
      "step": 1093
    },
    {
      "epoch": 2.4342984409799553,
      "loss": 0.8641993999481201,
      "loss_ce": 0.00042982713785022497,
      "loss_iou": 0.36328125,
      "loss_num": 0.027587890625,
      "loss_xval": 0.86328125,
      "num_input_tokens_seen": 61217816,
      "step": 1093
    },
    {
      "epoch": 2.4365256124721606,
      "grad_norm": 20.244873046875,
      "learning_rate": 1e-06,
      "loss": 0.5503,
      "num_input_tokens_seen": 61274332,
      "step": 1094
    },
    {
      "epoch": 2.4365256124721606,
      "loss": 0.40073591470718384,
      "loss_ce": 0.0003452802775427699,
      "loss_iou": 0.1748046875,
      "loss_num": 0.010009765625,
      "loss_xval": 0.400390625,
      "num_input_tokens_seen": 61274332,
      "step": 1094
    },
    {
      "epoch": 2.4387527839643655,
      "grad_norm": 18.268579483032227,
      "learning_rate": 1e-06,
      "loss": 0.9795,
      "num_input_tokens_seen": 61327972,
      "step": 1095
    },
    {
      "epoch": 2.4387527839643655,
      "loss": 0.696860671043396,
      "loss_ce": 0.00044950933079235256,
      "loss_iou": 0.291015625,
      "loss_num": 0.02294921875,
      "loss_xval": 0.6953125,
      "num_input_tokens_seen": 61327972,
      "step": 1095
    },
    {
      "epoch": 2.4409799554565703,
      "grad_norm": 13.835185050964355,
      "learning_rate": 1e-06,
      "loss": 0.8244,
      "num_input_tokens_seen": 61384388,
      "step": 1096
    },
    {
      "epoch": 2.4409799554565703,
      "loss": 0.6497031450271606,
      "loss_ce": 0.00028911407571285963,
      "loss_iou": 0.2421875,
      "loss_num": 0.03271484375,
      "loss_xval": 0.6484375,
      "num_input_tokens_seen": 61384388,
      "step": 1096
    },
    {
      "epoch": 2.443207126948775,
      "grad_norm": 17.01195526123047,
      "learning_rate": 1e-06,
      "loss": 0.9698,
      "num_input_tokens_seen": 61439968,
      "step": 1097
    },
    {
      "epoch": 2.443207126948775,
      "loss": 1.043675184249878,
      "loss_ce": 0.0007063635857775807,
      "loss_iou": 0.45703125,
      "loss_num": 0.0260009765625,
      "loss_xval": 1.046875,
      "num_input_tokens_seen": 61439968,
      "step": 1097
    },
    {
      "epoch": 2.44543429844098,
      "grad_norm": 12.948113441467285,
      "learning_rate": 1e-06,
      "loss": 0.8937,
      "num_input_tokens_seen": 61496432,
      "step": 1098
    },
    {
      "epoch": 2.44543429844098,
      "loss": 0.6754014492034912,
      "loss_ce": 0.00035263324389234185,
      "loss_iou": 0.291015625,
      "loss_num": 0.0184326171875,
      "loss_xval": 0.67578125,
      "num_input_tokens_seen": 61496432,
      "step": 1098
    },
    {
      "epoch": 2.447661469933185,
      "grad_norm": 15.429597854614258,
      "learning_rate": 1e-06,
      "loss": 1.3256,
      "num_input_tokens_seen": 61555176,
      "step": 1099
    },
    {
      "epoch": 2.447661469933185,
      "loss": 1.1492905616760254,
      "loss_ce": 0.0003646724799182266,
      "loss_iou": 0.4609375,
      "loss_num": 0.045654296875,
      "loss_xval": 1.1484375,
      "num_input_tokens_seen": 61555176,
      "step": 1099
    },
    {
      "epoch": 2.4498886414253898,
      "grad_norm": 13.81130313873291,
      "learning_rate": 1e-06,
      "loss": 0.9434,
      "num_input_tokens_seen": 61614452,
      "step": 1100
    },
    {
      "epoch": 2.4498886414253898,
      "loss": 1.213216781616211,
      "loss_ce": 0.00032615740201435983,
      "loss_iou": 0.5234375,
      "loss_num": 0.03271484375,
      "loss_xval": 1.2109375,
      "num_input_tokens_seen": 61614452,
      "step": 1100
    },
    {
      "epoch": 2.4521158129175946,
      "grad_norm": 36.95311737060547,
      "learning_rate": 1e-06,
      "loss": 0.6854,
      "num_input_tokens_seen": 61671728,
      "step": 1101
    },
    {
      "epoch": 2.4521158129175946,
      "loss": 0.6432477235794067,
      "loss_ce": 0.0006696175551041961,
      "loss_iou": 0.24609375,
      "loss_num": 0.0302734375,
      "loss_xval": 0.640625,
      "num_input_tokens_seen": 61671728,
      "step": 1101
    },
    {
      "epoch": 2.4543429844097995,
      "grad_norm": 18.458709716796875,
      "learning_rate": 1e-06,
      "loss": 0.8445,
      "num_input_tokens_seen": 61723072,
      "step": 1102
    },
    {
      "epoch": 2.4543429844097995,
      "loss": 1.0072041749954224,
      "loss_ce": 0.00036824517883360386,
      "loss_iou": 0.412109375,
      "loss_num": 0.036865234375,
      "loss_xval": 1.0078125,
      "num_input_tokens_seen": 61723072,
      "step": 1102
    },
    {
      "epoch": 2.4565701559020043,
      "grad_norm": 15.849419593811035,
      "learning_rate": 1e-06,
      "loss": 1.1075,
      "num_input_tokens_seen": 61780896,
      "step": 1103
    },
    {
      "epoch": 2.4565701559020043,
      "loss": 0.9099684953689575,
      "loss_ce": 0.006648160517215729,
      "loss_iou": 0.3984375,
      "loss_num": 0.021240234375,
      "loss_xval": 0.90234375,
      "num_input_tokens_seen": 61780896,
      "step": 1103
    },
    {
      "epoch": 2.458797327394209,
      "grad_norm": 19.382238388061523,
      "learning_rate": 1e-06,
      "loss": 1.0016,
      "num_input_tokens_seen": 61836688,
      "step": 1104
    },
    {
      "epoch": 2.458797327394209,
      "loss": 0.6892969608306885,
      "loss_ce": 0.0003321617841720581,
      "loss_iou": 0.28125,
      "loss_num": 0.025146484375,
      "loss_xval": 0.6875,
      "num_input_tokens_seen": 61836688,
      "step": 1104
    },
    {
      "epoch": 2.461024498886414,
      "grad_norm": 19.609500885009766,
      "learning_rate": 1e-06,
      "loss": 1.1058,
      "num_input_tokens_seen": 61892164,
      "step": 1105
    },
    {
      "epoch": 2.461024498886414,
      "loss": 1.3998944759368896,
      "loss_ce": 0.00048047915333881974,
      "loss_iou": 0.546875,
      "loss_num": 0.06103515625,
      "loss_xval": 1.3984375,
      "num_input_tokens_seen": 61892164,
      "step": 1105
    },
    {
      "epoch": 2.463251670378619,
      "grad_norm": 19.078765869140625,
      "learning_rate": 1e-06,
      "loss": 0.8841,
      "num_input_tokens_seen": 61948728,
      "step": 1106
    },
    {
      "epoch": 2.463251670378619,
      "loss": 0.673430323600769,
      "loss_ce": 0.0003346248995512724,
      "loss_iou": 0.251953125,
      "loss_num": 0.033935546875,
      "loss_xval": 0.671875,
      "num_input_tokens_seen": 61948728,
      "step": 1106
    },
    {
      "epoch": 2.4654788418708242,
      "grad_norm": 28.457544326782227,
      "learning_rate": 1e-06,
      "loss": 0.7968,
      "num_input_tokens_seen": 62002888,
      "step": 1107
    },
    {
      "epoch": 2.4654788418708242,
      "loss": 0.6821703314781189,
      "loss_ce": 0.00028555351309478283,
      "loss_iou": 0.251953125,
      "loss_num": 0.035888671875,
      "loss_xval": 0.68359375,
      "num_input_tokens_seen": 62002888,
      "step": 1107
    },
    {
      "epoch": 2.467706013363029,
      "grad_norm": 15.16339111328125,
      "learning_rate": 1e-06,
      "loss": 0.8623,
      "num_input_tokens_seen": 62058460,
      "step": 1108
    },
    {
      "epoch": 2.467706013363029,
      "loss": 0.8493123650550842,
      "loss_ce": 0.00043543853098526597,
      "loss_iou": 0.3515625,
      "loss_num": 0.029296875,
      "loss_xval": 0.84765625,
      "num_input_tokens_seen": 62058460,
      "step": 1108
    },
    {
      "epoch": 2.469933184855234,
      "grad_norm": 21.689632415771484,
      "learning_rate": 1e-06,
      "loss": 0.8395,
      "num_input_tokens_seen": 62116612,
      "step": 1109
    },
    {
      "epoch": 2.469933184855234,
      "loss": 0.6302697658538818,
      "loss_ce": 0.00038696054252795875,
      "loss_iou": 0.267578125,
      "loss_num": 0.0194091796875,
      "loss_xval": 0.62890625,
      "num_input_tokens_seen": 62116612,
      "step": 1109
    },
    {
      "epoch": 2.472160356347439,
      "grad_norm": 32.711299896240234,
      "learning_rate": 1e-06,
      "loss": 0.9001,
      "num_input_tokens_seen": 62173120,
      "step": 1110
    },
    {
      "epoch": 2.472160356347439,
      "loss": 1.1314702033996582,
      "loss_ce": 0.0003667679848149419,
      "loss_iou": 0.5078125,
      "loss_num": 0.0223388671875,
      "loss_xval": 1.1328125,
      "num_input_tokens_seen": 62173120,
      "step": 1110
    },
    {
      "epoch": 2.4743875278396437,
      "grad_norm": 63.607330322265625,
      "learning_rate": 1e-06,
      "loss": 0.928,
      "num_input_tokens_seen": 62231148,
      "step": 1111
    },
    {
      "epoch": 2.4743875278396437,
      "loss": 0.8059262037277222,
      "loss_ce": 0.00038422548095695674,
      "loss_iou": 0.3359375,
      "loss_num": 0.026611328125,
      "loss_xval": 0.8046875,
      "num_input_tokens_seen": 62231148,
      "step": 1111
    },
    {
      "epoch": 2.4766146993318485,
      "grad_norm": 17.097545623779297,
      "learning_rate": 1e-06,
      "loss": 0.9165,
      "num_input_tokens_seen": 62288036,
      "step": 1112
    },
    {
      "epoch": 2.4766146993318485,
      "loss": 0.9176112413406372,
      "loss_ce": 0.0003749003808479756,
      "loss_iou": 0.400390625,
      "loss_num": 0.02294921875,
      "loss_xval": 0.91796875,
      "num_input_tokens_seen": 62288036,
      "step": 1112
    },
    {
      "epoch": 2.4788418708240534,
      "grad_norm": 19.068836212158203,
      "learning_rate": 1e-06,
      "loss": 1.0021,
      "num_input_tokens_seen": 62343840,
      "step": 1113
    },
    {
      "epoch": 2.4788418708240534,
      "loss": 0.8066169619560242,
      "loss_ce": 0.0007087617414072156,
      "loss_iou": 0.326171875,
      "loss_num": 0.0311279296875,
      "loss_xval": 0.8046875,
      "num_input_tokens_seen": 62343840,
      "step": 1113
    },
    {
      "epoch": 2.4810690423162582,
      "grad_norm": 18.9963321685791,
      "learning_rate": 1e-06,
      "loss": 0.8238,
      "num_input_tokens_seen": 62399636,
      "step": 1114
    },
    {
      "epoch": 2.4810690423162582,
      "loss": 0.6373310089111328,
      "loss_ce": 0.00036811293102800846,
      "loss_iou": 0.27734375,
      "loss_num": 0.01611328125,
      "loss_xval": 0.63671875,
      "num_input_tokens_seen": 62399636,
      "step": 1114
    },
    {
      "epoch": 2.483296213808463,
      "grad_norm": 18.128734588623047,
      "learning_rate": 1e-06,
      "loss": 1.0068,
      "num_input_tokens_seen": 62457020,
      "step": 1115
    },
    {
      "epoch": 2.483296213808463,
      "loss": 0.7235212326049805,
      "loss_ce": 0.0003766651498153806,
      "loss_iou": 0.30859375,
      "loss_num": 0.021240234375,
      "loss_xval": 0.72265625,
      "num_input_tokens_seen": 62457020,
      "step": 1115
    },
    {
      "epoch": 2.485523385300668,
      "grad_norm": 53.74859619140625,
      "learning_rate": 1e-06,
      "loss": 0.8546,
      "num_input_tokens_seen": 62515924,
      "step": 1116
    },
    {
      "epoch": 2.485523385300668,
      "loss": 1.067058801651001,
      "loss_ce": 0.0006525892531499267,
      "loss_iou": 0.4375,
      "loss_num": 0.0380859375,
      "loss_xval": 1.0625,
      "num_input_tokens_seen": 62515924,
      "step": 1116
    },
    {
      "epoch": 2.4877505567928733,
      "grad_norm": 16.628570556640625,
      "learning_rate": 1e-06,
      "loss": 0.9406,
      "num_input_tokens_seen": 62575028,
      "step": 1117
    },
    {
      "epoch": 2.4877505567928733,
      "loss": 1.1619257926940918,
      "loss_ce": 0.0005488308379426599,
      "loss_iou": 0.4765625,
      "loss_num": 0.041748046875,
      "loss_xval": 1.1640625,
      "num_input_tokens_seen": 62575028,
      "step": 1117
    },
    {
      "epoch": 2.489977728285078,
      "grad_norm": 18.023460388183594,
      "learning_rate": 1e-06,
      "loss": 0.9142,
      "num_input_tokens_seen": 62630184,
      "step": 1118
    },
    {
      "epoch": 2.489977728285078,
      "loss": 1.1649341583251953,
      "loss_ce": 0.00038348851376213133,
      "loss_iou": 0.458984375,
      "loss_num": 0.049560546875,
      "loss_xval": 1.1640625,
      "num_input_tokens_seen": 62630184,
      "step": 1118
    },
    {
      "epoch": 2.492204899777283,
      "grad_norm": 28.666088104248047,
      "learning_rate": 1e-06,
      "loss": 1.1036,
      "num_input_tokens_seen": 62686600,
      "step": 1119
    },
    {
      "epoch": 2.492204899777283,
      "loss": 1.091188907623291,
      "loss_ce": 0.0008568049524910748,
      "loss_iou": 0.443359375,
      "loss_num": 0.040771484375,
      "loss_xval": 1.09375,
      "num_input_tokens_seen": 62686600,
      "step": 1119
    },
    {
      "epoch": 2.494432071269488,
      "grad_norm": 34.05298614501953,
      "learning_rate": 1e-06,
      "loss": 1.3067,
      "num_input_tokens_seen": 62740120,
      "step": 1120
    },
    {
      "epoch": 2.494432071269488,
      "loss": 1.4040180444717407,
      "loss_ce": 0.00094178831204772,
      "loss_iou": 0.53125,
      "loss_num": 0.0673828125,
      "loss_xval": 1.40625,
      "num_input_tokens_seen": 62740120,
      "step": 1120
    },
    {
      "epoch": 2.4966592427616927,
      "grad_norm": 23.138212203979492,
      "learning_rate": 1e-06,
      "loss": 0.9225,
      "num_input_tokens_seen": 62796964,
      "step": 1121
    },
    {
      "epoch": 2.4966592427616927,
      "loss": 1.013495922088623,
      "loss_ce": 0.0003123595961369574,
      "loss_iou": 0.404296875,
      "loss_num": 0.04052734375,
      "loss_xval": 1.015625,
      "num_input_tokens_seen": 62796964,
      "step": 1121
    },
    {
      "epoch": 2.4988864142538976,
      "grad_norm": 17.127120971679688,
      "learning_rate": 1e-06,
      "loss": 1.2497,
      "num_input_tokens_seen": 62853264,
      "step": 1122
    },
    {
      "epoch": 2.4988864142538976,
      "loss": 1.4592686891555786,
      "loss_ce": 0.0007725717732682824,
      "loss_iou": 0.546875,
      "loss_num": 0.07373046875,
      "loss_xval": 1.4609375,
      "num_input_tokens_seen": 62853264,
      "step": 1122
    },
    {
      "epoch": 2.5011135857461024,
      "grad_norm": 14.743764877319336,
      "learning_rate": 1e-06,
      "loss": 0.8212,
      "num_input_tokens_seen": 62909164,
      "step": 1123
    },
    {
      "epoch": 2.5011135857461024,
      "loss": 0.598504900932312,
      "loss_ce": 0.0003604079829528928,
      "loss_iou": 0.26171875,
      "loss_num": 0.01519775390625,
      "loss_xval": 0.59765625,
      "num_input_tokens_seen": 62909164,
      "step": 1123
    },
    {
      "epoch": 2.5033407572383073,
      "grad_norm": 13.89709758758545,
      "learning_rate": 1e-06,
      "loss": 0.9177,
      "num_input_tokens_seen": 62965280,
      "step": 1124
    },
    {
      "epoch": 2.5033407572383073,
      "loss": 0.7613284587860107,
      "loss_ce": 0.0003421393339522183,
      "loss_iou": 0.3046875,
      "loss_num": 0.0301513671875,
      "loss_xval": 0.76171875,
      "num_input_tokens_seen": 62965280,
      "step": 1124
    },
    {
      "epoch": 2.505567928730512,
      "grad_norm": 27.25965690612793,
      "learning_rate": 1e-06,
      "loss": 0.8435,
      "num_input_tokens_seen": 63019164,
      "step": 1125
    },
    {
      "epoch": 2.505567928730512,
      "loss": 0.9681985974311829,
      "loss_ce": 0.0004251442151144147,
      "loss_iou": 0.390625,
      "loss_num": 0.037109375,
      "loss_xval": 0.96875,
      "num_input_tokens_seen": 63019164,
      "step": 1125
    },
    {
      "epoch": 2.507795100222717,
      "grad_norm": 19.890287399291992,
      "learning_rate": 1e-06,
      "loss": 0.8424,
      "num_input_tokens_seen": 63073888,
      "step": 1126
    },
    {
      "epoch": 2.507795100222717,
      "loss": 0.7713284492492676,
      "loss_ce": 0.00033233320573344827,
      "loss_iou": 0.32421875,
      "loss_num": 0.0242919921875,
      "loss_xval": 0.76953125,
      "num_input_tokens_seen": 63073888,
      "step": 1126
    },
    {
      "epoch": 2.510022271714922,
      "grad_norm": 19.229721069335938,
      "learning_rate": 1e-06,
      "loss": 0.799,
      "num_input_tokens_seen": 63131120,
      "step": 1127
    },
    {
      "epoch": 2.510022271714922,
      "loss": 0.7538760900497437,
      "loss_ce": 0.0004581384710036218,
      "loss_iou": 0.306640625,
      "loss_num": 0.0279541015625,
      "loss_xval": 0.75390625,
      "num_input_tokens_seen": 63131120,
      "step": 1127
    },
    {
      "epoch": 2.5122494432071267,
      "grad_norm": 18.243764877319336,
      "learning_rate": 1e-06,
      "loss": 1.0801,
      "num_input_tokens_seen": 63184712,
      "step": 1128
    },
    {
      "epoch": 2.5122494432071267,
      "loss": 1.014033555984497,
      "loss_ce": 0.00036173040280118585,
      "loss_iou": 0.431640625,
      "loss_num": 0.0302734375,
      "loss_xval": 1.015625,
      "num_input_tokens_seen": 63184712,
      "step": 1128
    },
    {
      "epoch": 2.5144766146993316,
      "grad_norm": 27.520559310913086,
      "learning_rate": 1e-06,
      "loss": 1.0515,
      "num_input_tokens_seen": 63239260,
      "step": 1129
    },
    {
      "epoch": 2.5144766146993316,
      "loss": 0.9654926061630249,
      "loss_ce": 0.0005268162931315601,
      "loss_iou": 0.369140625,
      "loss_num": 0.0458984375,
      "loss_xval": 0.96484375,
      "num_input_tokens_seen": 63239260,
      "step": 1129
    },
    {
      "epoch": 2.516703786191537,
      "grad_norm": 23.411195755004883,
      "learning_rate": 1e-06,
      "loss": 0.5888,
      "num_input_tokens_seen": 63294036,
      "step": 1130
    },
    {
      "epoch": 2.516703786191537,
      "loss": 0.5716757774353027,
      "loss_ce": 0.00038673574454151094,
      "loss_iou": 0.248046875,
      "loss_num": 0.0150146484375,
      "loss_xval": 0.5703125,
      "num_input_tokens_seen": 63294036,
      "step": 1130
    },
    {
      "epoch": 2.5189309576837418,
      "grad_norm": 27.774524688720703,
      "learning_rate": 1e-06,
      "loss": 0.9061,
      "num_input_tokens_seen": 63350920,
      "step": 1131
    },
    {
      "epoch": 2.5189309576837418,
      "loss": 0.7308274507522583,
      "loss_ce": 0.00035869883140549064,
      "loss_iou": 0.298828125,
      "loss_num": 0.026123046875,
      "loss_xval": 0.73046875,
      "num_input_tokens_seen": 63350920,
      "step": 1131
    },
    {
      "epoch": 2.5211581291759466,
      "grad_norm": 42.71458053588867,
      "learning_rate": 1e-06,
      "loss": 1.0052,
      "num_input_tokens_seen": 63408328,
      "step": 1132
    },
    {
      "epoch": 2.5211581291759466,
      "loss": 0.9227426052093506,
      "loss_ce": 0.00037933725980110466,
      "loss_iou": 0.41015625,
      "loss_num": 0.0203857421875,
      "loss_xval": 0.921875,
      "num_input_tokens_seen": 63408328,
      "step": 1132
    },
    {
      "epoch": 2.5233853006681515,
      "grad_norm": 21.282625198364258,
      "learning_rate": 1e-06,
      "loss": 1.0933,
      "num_input_tokens_seen": 63463376,
      "step": 1133
    },
    {
      "epoch": 2.5233853006681515,
      "loss": 0.8489203453063965,
      "loss_ce": 0.0005317054456099868,
      "loss_iou": 0.380859375,
      "loss_num": 0.017333984375,
      "loss_xval": 0.84765625,
      "num_input_tokens_seen": 63463376,
      "step": 1133
    },
    {
      "epoch": 2.5256124721603563,
      "grad_norm": 13.566144943237305,
      "learning_rate": 1e-06,
      "loss": 0.9126,
      "num_input_tokens_seen": 63521748,
      "step": 1134
    },
    {
      "epoch": 2.5256124721603563,
      "loss": 1.027003288269043,
      "loss_ce": 0.0006361395353451371,
      "loss_iou": 0.42578125,
      "loss_num": 0.03466796875,
      "loss_xval": 1.0234375,
      "num_input_tokens_seen": 63521748,
      "step": 1134
    },
    {
      "epoch": 2.527839643652561,
      "grad_norm": 22.058757781982422,
      "learning_rate": 1e-06,
      "loss": 0.8642,
      "num_input_tokens_seen": 63575808,
      "step": 1135
    },
    {
      "epoch": 2.527839643652561,
      "loss": 0.5723444223403931,
      "loss_ce": 0.00032291823299601674,
      "loss_iou": 0.224609375,
      "loss_num": 0.024658203125,
      "loss_xval": 0.5703125,
      "num_input_tokens_seen": 63575808,
      "step": 1135
    },
    {
      "epoch": 2.530066815144766,
      "grad_norm": 17.351253509521484,
      "learning_rate": 1e-06,
      "loss": 0.9297,
      "num_input_tokens_seen": 63629380,
      "step": 1136
    },
    {
      "epoch": 2.530066815144766,
      "loss": 1.1235007047653198,
      "loss_ce": 0.0006980298785492778,
      "loss_iou": 0.466796875,
      "loss_num": 0.037841796875,
      "loss_xval": 1.125,
      "num_input_tokens_seen": 63629380,
      "step": 1136
    },
    {
      "epoch": 2.532293986636971,
      "grad_norm": 25.26835060119629,
      "learning_rate": 1e-06,
      "loss": 0.8625,
      "num_input_tokens_seen": 63685404,
      "step": 1137
    },
    {
      "epoch": 2.532293986636971,
      "loss": 0.7635213136672974,
      "loss_ce": 0.0003376836539246142,
      "loss_iou": 0.314453125,
      "loss_num": 0.026611328125,
      "loss_xval": 0.76171875,
      "num_input_tokens_seen": 63685404,
      "step": 1137
    },
    {
      "epoch": 2.534521158129176,
      "grad_norm": 23.412189483642578,
      "learning_rate": 1e-06,
      "loss": 1.0618,
      "num_input_tokens_seen": 63742344,
      "step": 1138
    },
    {
      "epoch": 2.534521158129176,
      "loss": 1.1275867223739624,
      "loss_ce": 0.0006335656507872045,
      "loss_iou": 0.470703125,
      "loss_num": 0.03759765625,
      "loss_xval": 1.125,
      "num_input_tokens_seen": 63742344,
      "step": 1138
    },
    {
      "epoch": 2.536748329621381,
      "grad_norm": 59.32328414916992,
      "learning_rate": 1e-06,
      "loss": 0.9894,
      "num_input_tokens_seen": 63798548,
      "step": 1139
    },
    {
      "epoch": 2.536748329621381,
      "loss": 0.9086170196533203,
      "loss_ce": 0.0006579948822036386,
      "loss_iou": 0.37890625,
      "loss_num": 0.030029296875,
      "loss_xval": 0.90625,
      "num_input_tokens_seen": 63798548,
      "step": 1139
    },
    {
      "epoch": 2.538975501113586,
      "grad_norm": 34.01222610473633,
      "learning_rate": 1e-06,
      "loss": 1.2089,
      "num_input_tokens_seen": 63851244,
      "step": 1140
    },
    {
      "epoch": 2.538975501113586,
      "loss": 1.4944491386413574,
      "loss_ce": 0.0003085851203650236,
      "loss_iou": 0.66796875,
      "loss_num": 0.03173828125,
      "loss_xval": 1.4921875,
      "num_input_tokens_seen": 63851244,
      "step": 1140
    },
    {
      "epoch": 2.541202672605791,
      "grad_norm": 23.09809112548828,
      "learning_rate": 1e-06,
      "loss": 1.0927,
      "num_input_tokens_seen": 63908384,
      "step": 1141
    },
    {
      "epoch": 2.541202672605791,
      "loss": 0.905631959438324,
      "loss_ce": 0.00035852432483807206,
      "loss_iou": 0.375,
      "loss_num": 0.0311279296875,
      "loss_xval": 0.90625,
      "num_input_tokens_seen": 63908384,
      "step": 1141
    },
    {
      "epoch": 2.5434298440979957,
      "grad_norm": 12.117959022521973,
      "learning_rate": 1e-06,
      "loss": 1.2241,
      "num_input_tokens_seen": 63965248,
      "step": 1142
    },
    {
      "epoch": 2.5434298440979957,
      "loss": 1.4422664642333984,
      "loss_ce": 0.0006161456694826484,
      "loss_iou": 0.515625,
      "loss_num": 0.0830078125,
      "loss_xval": 1.4453125,
      "num_input_tokens_seen": 63965248,
      "step": 1142
    },
    {
      "epoch": 2.5456570155902005,
      "grad_norm": 19.716156005859375,
      "learning_rate": 1e-06,
      "loss": 0.9176,
      "num_input_tokens_seen": 64018792,
      "step": 1143
    },
    {
      "epoch": 2.5456570155902005,
      "loss": 0.611274778842926,
      "loss_ce": 0.000434944755397737,
      "loss_iou": 0.248046875,
      "loss_num": 0.023193359375,
      "loss_xval": 0.609375,
      "num_input_tokens_seen": 64018792,
      "step": 1143
    },
    {
      "epoch": 2.5478841870824054,
      "grad_norm": 16.11542320251465,
      "learning_rate": 1e-06,
      "loss": 0.9638,
      "num_input_tokens_seen": 64073780,
      "step": 1144
    },
    {
      "epoch": 2.5478841870824054,
      "loss": 1.0830767154693604,
      "loss_ce": 0.0004351097741164267,
      "loss_iou": 0.3984375,
      "loss_num": 0.05712890625,
      "loss_xval": 1.0859375,
      "num_input_tokens_seen": 64073780,
      "step": 1144
    },
    {
      "epoch": 2.5501113585746102,
      "grad_norm": 27.197479248046875,
      "learning_rate": 1e-06,
      "loss": 0.8059,
      "num_input_tokens_seen": 64130436,
      "step": 1145
    },
    {
      "epoch": 2.5501113585746102,
      "loss": 1.001615285873413,
      "loss_ce": 0.0008829243597574532,
      "loss_iou": 0.39453125,
      "loss_num": 0.042236328125,
      "loss_xval": 1.0,
      "num_input_tokens_seen": 64130436,
      "step": 1145
    },
    {
      "epoch": 2.552338530066815,
      "grad_norm": 17.01374053955078,
      "learning_rate": 1e-06,
      "loss": 0.9329,
      "num_input_tokens_seen": 64183692,
      "step": 1146
    },
    {
      "epoch": 2.552338530066815,
      "loss": 0.9256563186645508,
      "loss_ce": 0.00036334185278974473,
      "loss_iou": 0.375,
      "loss_num": 0.034912109375,
      "loss_xval": 0.92578125,
      "num_input_tokens_seen": 64183692,
      "step": 1146
    },
    {
      "epoch": 2.55456570155902,
      "grad_norm": 18.95051383972168,
      "learning_rate": 1e-06,
      "loss": 0.9019,
      "num_input_tokens_seen": 64239220,
      "step": 1147
    },
    {
      "epoch": 2.55456570155902,
      "loss": 1.1268051862716675,
      "loss_ce": 0.0003403578884899616,
      "loss_iou": 0.462890625,
      "loss_num": 0.040283203125,
      "loss_xval": 1.125,
      "num_input_tokens_seen": 64239220,
      "step": 1147
    },
    {
      "epoch": 2.556792873051225,
      "grad_norm": 26.1760196685791,
      "learning_rate": 1e-06,
      "loss": 0.8004,
      "num_input_tokens_seen": 64292876,
      "step": 1148
    },
    {
      "epoch": 2.556792873051225,
      "loss": 0.802330493927002,
      "loss_ce": 0.0004505745891947299,
      "loss_iou": 0.3125,
      "loss_num": 0.03564453125,
      "loss_xval": 0.80078125,
      "num_input_tokens_seen": 64292876,
      "step": 1148
    },
    {
      "epoch": 2.5590200445434297,
      "grad_norm": 18.058563232421875,
      "learning_rate": 1e-06,
      "loss": 0.9304,
      "num_input_tokens_seen": 64350980,
      "step": 1149
    },
    {
      "epoch": 2.5590200445434297,
      "loss": 1.0028624534606934,
      "loss_ce": 0.00066518341191113,
      "loss_iou": 0.41796875,
      "loss_num": 0.032958984375,
      "loss_xval": 1.0,
      "num_input_tokens_seen": 64350980,
      "step": 1149
    },
    {
      "epoch": 2.5612472160356345,
      "grad_norm": 15.238943099975586,
      "learning_rate": 1e-06,
      "loss": 0.8567,
      "num_input_tokens_seen": 64407372,
      "step": 1150
    },
    {
      "epoch": 2.5612472160356345,
      "loss": 0.6080912351608276,
      "loss_ce": 0.00042517349356785417,
      "loss_iou": 0.255859375,
      "loss_num": 0.0194091796875,
      "loss_xval": 0.609375,
      "num_input_tokens_seen": 64407372,
      "step": 1150
    },
    {
      "epoch": 2.5634743875278394,
      "grad_norm": 24.09518814086914,
      "learning_rate": 1e-06,
      "loss": 0.9458,
      "num_input_tokens_seen": 64459936,
      "step": 1151
    },
    {
      "epoch": 2.5634743875278394,
      "loss": 0.8809654712677002,
      "loss_ce": 0.0003502329345792532,
      "loss_iou": 0.40234375,
      "loss_num": 0.01507568359375,
      "loss_xval": 0.87890625,
      "num_input_tokens_seen": 64459936,
      "step": 1151
    },
    {
      "epoch": 2.5657015590200447,
      "grad_norm": 18.098283767700195,
      "learning_rate": 1e-06,
      "loss": 0.9909,
      "num_input_tokens_seen": 64516664,
      "step": 1152
    },
    {
      "epoch": 2.5657015590200447,
      "loss": 0.8294593095779419,
      "loss_ce": 0.00035776515142060816,
      "loss_iou": 0.337890625,
      "loss_num": 0.03076171875,
      "loss_xval": 0.828125,
      "num_input_tokens_seen": 64516664,
      "step": 1152
    },
    {
      "epoch": 2.5679287305122496,
      "grad_norm": 23.110363006591797,
      "learning_rate": 1e-06,
      "loss": 0.8105,
      "num_input_tokens_seen": 64573704,
      "step": 1153
    },
    {
      "epoch": 2.5679287305122496,
      "loss": 0.7701290845870972,
      "loss_ce": 0.0003536652075126767,
      "loss_iou": 0.326171875,
      "loss_num": 0.0234375,
      "loss_xval": 0.76953125,
      "num_input_tokens_seen": 64573704,
      "step": 1153
    },
    {
      "epoch": 2.5701559020044544,
      "grad_norm": 18.98008155822754,
      "learning_rate": 1e-06,
      "loss": 0.8578,
      "num_input_tokens_seen": 64632268,
      "step": 1154
    },
    {
      "epoch": 2.5701559020044544,
      "loss": 0.7266561985015869,
      "loss_ce": 0.00033786421408876777,
      "loss_iou": 0.32421875,
      "loss_num": 0.015625,
      "loss_xval": 0.7265625,
      "num_input_tokens_seen": 64632268,
      "step": 1154
    },
    {
      "epoch": 2.5723830734966593,
      "grad_norm": 17.440505981445312,
      "learning_rate": 1e-06,
      "loss": 1.0514,
      "num_input_tokens_seen": 64687228,
      "step": 1155
    },
    {
      "epoch": 2.5723830734966593,
      "loss": 1.1310064792633057,
      "loss_ce": 0.0003913280088454485,
      "loss_iou": 0.470703125,
      "loss_num": 0.03759765625,
      "loss_xval": 1.1328125,
      "num_input_tokens_seen": 64687228,
      "step": 1155
    },
    {
      "epoch": 2.574610244988864,
      "grad_norm": 17.068138122558594,
      "learning_rate": 1e-06,
      "loss": 0.7717,
      "num_input_tokens_seen": 64744716,
      "step": 1156
    },
    {
      "epoch": 2.574610244988864,
      "loss": 0.6678210496902466,
      "loss_ce": 0.00034053760464303195,
      "loss_iou": 0.291015625,
      "loss_num": 0.01708984375,
      "loss_xval": 0.66796875,
      "num_input_tokens_seen": 64744716,
      "step": 1156
    },
    {
      "epoch": 2.576837416481069,
      "grad_norm": 16.1224308013916,
      "learning_rate": 1e-06,
      "loss": 0.894,
      "num_input_tokens_seen": 64803444,
      "step": 1157
    },
    {
      "epoch": 2.576837416481069,
      "loss": 0.783063530921936,
      "loss_ce": 0.0003486335917841643,
      "loss_iou": 0.34375,
      "loss_num": 0.0194091796875,
      "loss_xval": 0.78125,
      "num_input_tokens_seen": 64803444,
      "step": 1157
    },
    {
      "epoch": 2.579064587973274,
      "grad_norm": 31.78223991394043,
      "learning_rate": 1e-06,
      "loss": 0.8672,
      "num_input_tokens_seen": 64859080,
      "step": 1158
    },
    {
      "epoch": 2.579064587973274,
      "loss": 0.8566364645957947,
      "loss_ce": 0.0004353098920546472,
      "loss_iou": 0.33984375,
      "loss_num": 0.03515625,
      "loss_xval": 0.85546875,
      "num_input_tokens_seen": 64859080,
      "step": 1158
    },
    {
      "epoch": 2.5812917594654787,
      "grad_norm": 22.2586612701416,
      "learning_rate": 1e-06,
      "loss": 1.2238,
      "num_input_tokens_seen": 64916436,
      "step": 1159
    },
    {
      "epoch": 2.5812917594654787,
      "loss": 1.3102490901947021,
      "loss_ce": 0.0009229998104274273,
      "loss_iou": 0.5078125,
      "loss_num": 0.059326171875,
      "loss_xval": 1.3125,
      "num_input_tokens_seen": 64916436,
      "step": 1159
    },
    {
      "epoch": 2.5835189309576836,
      "grad_norm": 47.00844955444336,
      "learning_rate": 1e-06,
      "loss": 1.2237,
      "num_input_tokens_seen": 64972836,
      "step": 1160
    },
    {
      "epoch": 2.5835189309576836,
      "loss": 1.0932590961456299,
      "loss_ce": 0.0004857148160226643,
      "loss_iou": 0.4375,
      "loss_num": 0.04345703125,
      "loss_xval": 1.09375,
      "num_input_tokens_seen": 64972836,
      "step": 1160
    },
    {
      "epoch": 2.585746102449889,
      "grad_norm": 20.3150634765625,
      "learning_rate": 1e-06,
      "loss": 1.236,
      "num_input_tokens_seen": 65027704,
      "step": 1161
    },
    {
      "epoch": 2.585746102449889,
      "loss": 1.0992555618286133,
      "loss_ce": 0.001111081801354885,
      "loss_iou": 0.474609375,
      "loss_num": 0.0296630859375,
      "loss_xval": 1.1015625,
      "num_input_tokens_seen": 65027704,
      "step": 1161
    },
    {
      "epoch": 2.5879732739420938,
      "grad_norm": 18.011165618896484,
      "learning_rate": 1e-06,
      "loss": 0.932,
      "num_input_tokens_seen": 65083788,
      "step": 1162
    },
    {
      "epoch": 2.5879732739420938,
      "loss": 1.0766229629516602,
      "loss_ce": 0.00045114755630493164,
      "loss_iou": 0.470703125,
      "loss_num": 0.0267333984375,
      "loss_xval": 1.078125,
      "num_input_tokens_seen": 65083788,
      "step": 1162
    },
    {
      "epoch": 2.5902004454342986,
      "grad_norm": 82.84613037109375,
      "learning_rate": 1e-06,
      "loss": 1.2092,
      "num_input_tokens_seen": 65139956,
      "step": 1163
    },
    {
      "epoch": 2.5902004454342986,
      "loss": 1.3553651571273804,
      "loss_ce": 0.00038471657899208367,
      "loss_iou": 0.546875,
      "loss_num": 0.051513671875,
      "loss_xval": 1.3515625,
      "num_input_tokens_seen": 65139956,
      "step": 1163
    },
    {
      "epoch": 2.5924276169265035,
      "grad_norm": 17.478727340698242,
      "learning_rate": 1e-06,
      "loss": 0.9212,
      "num_input_tokens_seen": 65192748,
      "step": 1164
    },
    {
      "epoch": 2.5924276169265035,
      "loss": 0.8342557549476624,
      "loss_ce": 0.0005154828540980816,
      "loss_iou": 0.337890625,
      "loss_num": 0.03173828125,
      "loss_xval": 0.83203125,
      "num_input_tokens_seen": 65192748,
      "step": 1164
    },
    {
      "epoch": 2.5946547884187083,
      "grad_norm": 16.476207733154297,
      "learning_rate": 1e-06,
      "loss": 0.9718,
      "num_input_tokens_seen": 65247340,
      "step": 1165
    },
    {
      "epoch": 2.5946547884187083,
      "loss": 0.9732043743133545,
      "loss_ce": 0.002012967597693205,
      "loss_iou": 0.4140625,
      "loss_num": 0.029052734375,
      "loss_xval": 0.97265625,
      "num_input_tokens_seen": 65247340,
      "step": 1165
    },
    {
      "epoch": 2.596881959910913,
      "grad_norm": 22.207536697387695,
      "learning_rate": 1e-06,
      "loss": 1.1978,
      "num_input_tokens_seen": 65304100,
      "step": 1166
    },
    {
      "epoch": 2.596881959910913,
      "loss": 1.1994106769561768,
      "loss_ce": 0.00043600943172350526,
      "loss_iou": 0.49609375,
      "loss_num": 0.04150390625,
      "loss_xval": 1.1953125,
      "num_input_tokens_seen": 65304100,
      "step": 1166
    },
    {
      "epoch": 2.599109131403118,
      "grad_norm": 28.934432983398438,
      "learning_rate": 1e-06,
      "loss": 0.7938,
      "num_input_tokens_seen": 65362144,
      "step": 1167
    },
    {
      "epoch": 2.599109131403118,
      "loss": 0.9534372091293335,
      "loss_ce": 0.0051950025372207165,
      "loss_iou": 0.400390625,
      "loss_num": 0.0294189453125,
      "loss_xval": 0.94921875,
      "num_input_tokens_seen": 65362144,
      "step": 1167
    },
    {
      "epoch": 2.601336302895323,
      "grad_norm": 18.83047103881836,
      "learning_rate": 1e-06,
      "loss": 0.8324,
      "num_input_tokens_seen": 65417596,
      "step": 1168
    },
    {
      "epoch": 2.601336302895323,
      "loss": 1.0699725151062012,
      "loss_ce": 0.0008807817357592285,
      "loss_iou": 0.421875,
      "loss_num": 0.044677734375,
      "loss_xval": 1.0703125,
      "num_input_tokens_seen": 65417596,
      "step": 1168
    },
    {
      "epoch": 2.6035634743875278,
      "grad_norm": 24.358144760131836,
      "learning_rate": 1e-06,
      "loss": 1.0814,
      "num_input_tokens_seen": 65470364,
      "step": 1169
    },
    {
      "epoch": 2.6035634743875278,
      "loss": 1.0574841499328613,
      "loss_ce": 0.0003552237758412957,
      "loss_iou": 0.421875,
      "loss_num": 0.04296875,
      "loss_xval": 1.0546875,
      "num_input_tokens_seen": 65470364,
      "step": 1169
    },
    {
      "epoch": 2.6057906458797326,
      "grad_norm": 93.62918090820312,
      "learning_rate": 1e-06,
      "loss": 1.0645,
      "num_input_tokens_seen": 65526196,
      "step": 1170
    },
    {
      "epoch": 2.6057906458797326,
      "loss": 0.9251736998558044,
      "loss_ce": 0.0003689858131110668,
      "loss_iou": 0.40625,
      "loss_num": 0.0223388671875,
      "loss_xval": 0.92578125,
      "num_input_tokens_seen": 65526196,
      "step": 1170
    },
    {
      "epoch": 2.6080178173719375,
      "grad_norm": 18.0775146484375,
      "learning_rate": 1e-06,
      "loss": 0.9244,
      "num_input_tokens_seen": 65580796,
      "step": 1171
    },
    {
      "epoch": 2.6080178173719375,
      "loss": 0.8135091662406921,
      "loss_ce": 0.0005208852817304432,
      "loss_iou": 0.330078125,
      "loss_num": 0.03076171875,
      "loss_xval": 0.8125,
      "num_input_tokens_seen": 65580796,
      "step": 1171
    },
    {
      "epoch": 2.6102449888641424,
      "grad_norm": 20.29970932006836,
      "learning_rate": 1e-06,
      "loss": 0.9885,
      "num_input_tokens_seen": 65633880,
      "step": 1172
    },
    {
      "epoch": 2.6102449888641424,
      "loss": 0.7842642068862915,
      "loss_ce": 0.00032866618130356073,
      "loss_iou": 0.318359375,
      "loss_num": 0.0294189453125,
      "loss_xval": 0.78515625,
      "num_input_tokens_seen": 65633880,
      "step": 1172
    },
    {
      "epoch": 2.612472160356347,
      "grad_norm": 16.221195220947266,
      "learning_rate": 1e-06,
      "loss": 0.7178,
      "num_input_tokens_seen": 65689656,
      "step": 1173
    },
    {
      "epoch": 2.612472160356347,
      "loss": 0.6070936322212219,
      "loss_ce": 0.00040419274591840804,
      "loss_iou": 0.2578125,
      "loss_num": 0.0185546875,
      "loss_xval": 0.60546875,
      "num_input_tokens_seen": 65689656,
      "step": 1173
    },
    {
      "epoch": 2.614699331848552,
      "grad_norm": 21.689422607421875,
      "learning_rate": 1e-06,
      "loss": 0.9887,
      "num_input_tokens_seen": 65748144,
      "step": 1174
    },
    {
      "epoch": 2.614699331848552,
      "loss": 0.829987108707428,
      "loss_ce": 0.00039724778616800904,
      "loss_iou": 0.369140625,
      "loss_num": 0.0179443359375,
      "loss_xval": 0.828125,
      "num_input_tokens_seen": 65748144,
      "step": 1174
    },
    {
      "epoch": 2.6169265033407574,
      "grad_norm": 26.423198699951172,
      "learning_rate": 1e-06,
      "loss": 1.0744,
      "num_input_tokens_seen": 65804144,
      "step": 1175
    },
    {
      "epoch": 2.6169265033407574,
      "loss": 1.1194689273834229,
      "loss_ce": 0.00032840511994436383,
      "loss_iou": 0.4609375,
      "loss_num": 0.03955078125,
      "loss_xval": 1.1171875,
      "num_input_tokens_seen": 65804144,
      "step": 1175
    },
    {
      "epoch": 2.6191536748329622,
      "grad_norm": 24.602550506591797,
      "learning_rate": 1e-06,
      "loss": 1.1684,
      "num_input_tokens_seen": 65858964,
      "step": 1176
    },
    {
      "epoch": 2.6191536748329622,
      "loss": 0.9794027805328369,
      "loss_ce": 0.00039887180901132524,
      "loss_iou": 0.40625,
      "loss_num": 0.033203125,
      "loss_xval": 0.98046875,
      "num_input_tokens_seen": 65858964,
      "step": 1176
    },
    {
      "epoch": 2.621380846325167,
      "grad_norm": 20.516910552978516,
      "learning_rate": 1e-06,
      "loss": 0.8049,
      "num_input_tokens_seen": 65914988,
      "step": 1177
    },
    {
      "epoch": 2.621380846325167,
      "loss": 0.8118576407432556,
      "loss_ce": 0.00033423834247514606,
      "loss_iou": 0.3515625,
      "loss_num": 0.0216064453125,
      "loss_xval": 0.8125,
      "num_input_tokens_seen": 65914988,
      "step": 1177
    },
    {
      "epoch": 2.623608017817372,
      "grad_norm": 20.05683135986328,
      "learning_rate": 1e-06,
      "loss": 1.0382,
      "num_input_tokens_seen": 65971964,
      "step": 1178
    },
    {
      "epoch": 2.623608017817372,
      "loss": 1.1247992515563965,
      "loss_ce": 0.0005316782626323402,
      "loss_iou": 0.451171875,
      "loss_num": 0.044189453125,
      "loss_xval": 1.125,
      "num_input_tokens_seen": 65971964,
      "step": 1178
    },
    {
      "epoch": 2.625835189309577,
      "grad_norm": 33.51689910888672,
      "learning_rate": 1e-06,
      "loss": 0.919,
      "num_input_tokens_seen": 66030480,
      "step": 1179
    },
    {
      "epoch": 2.625835189309577,
      "loss": 1.0413250923156738,
      "loss_ce": 0.00030942526063881814,
      "loss_iou": 0.435546875,
      "loss_num": 0.03369140625,
      "loss_xval": 1.0390625,
      "num_input_tokens_seen": 66030480,
      "step": 1179
    },
    {
      "epoch": 2.6280623608017817,
      "grad_norm": 21.715831756591797,
      "learning_rate": 1e-06,
      "loss": 1.1688,
      "num_input_tokens_seen": 66084552,
      "step": 1180
    },
    {
      "epoch": 2.6280623608017817,
      "loss": 1.3229314088821411,
      "loss_ce": 0.0006658075144514441,
      "loss_iou": 0.52734375,
      "loss_num": 0.052978515625,
      "loss_xval": 1.3203125,
      "num_input_tokens_seen": 66084552,
      "step": 1180
    },
    {
      "epoch": 2.6302895322939865,
      "grad_norm": 16.5938777923584,
      "learning_rate": 1e-06,
      "loss": 0.8147,
      "num_input_tokens_seen": 66140764,
      "step": 1181
    },
    {
      "epoch": 2.6302895322939865,
      "loss": 0.7720546722412109,
      "loss_ce": 0.0005702448543161154,
      "loss_iou": 0.306640625,
      "loss_num": 0.031494140625,
      "loss_xval": 0.7734375,
      "num_input_tokens_seen": 66140764,
      "step": 1181
    },
    {
      "epoch": 2.6325167037861914,
      "grad_norm": 20.419330596923828,
      "learning_rate": 1e-06,
      "loss": 0.9905,
      "num_input_tokens_seen": 66196076,
      "step": 1182
    },
    {
      "epoch": 2.6325167037861914,
      "loss": 1.1305346488952637,
      "loss_ce": 0.0006517736474052072,
      "loss_iou": 0.50390625,
      "loss_num": 0.0245361328125,
      "loss_xval": 1.1328125,
      "num_input_tokens_seen": 66196076,
      "step": 1182
    },
    {
      "epoch": 2.6347438752783967,
      "grad_norm": 27.849132537841797,
      "learning_rate": 1e-06,
      "loss": 0.7988,
      "num_input_tokens_seen": 66247728,
      "step": 1183
    },
    {
      "epoch": 2.6347438752783967,
      "loss": 0.901054859161377,
      "loss_ce": 0.000420065043726936,
      "loss_iou": 0.375,
      "loss_num": 0.0296630859375,
      "loss_xval": 0.90234375,
      "num_input_tokens_seen": 66247728,
      "step": 1183
    },
    {
      "epoch": 2.6369710467706016,
      "grad_norm": 20.54474449157715,
      "learning_rate": 1e-06,
      "loss": 1.0772,
      "num_input_tokens_seen": 66301456,
      "step": 1184
    },
    {
      "epoch": 2.6369710467706016,
      "loss": 1.1666040420532227,
      "loss_ce": 0.000588378170505166,
      "loss_iou": 0.43359375,
      "loss_num": 0.060302734375,
      "loss_xval": 1.1640625,
      "num_input_tokens_seen": 66301456,
      "step": 1184
    },
    {
      "epoch": 2.6391982182628064,
      "grad_norm": 21.712364196777344,
      "learning_rate": 1e-06,
      "loss": 0.8525,
      "num_input_tokens_seen": 66358172,
      "step": 1185
    },
    {
      "epoch": 2.6391982182628064,
      "loss": 0.7973392009735107,
      "loss_ce": 0.0003421990550123155,
      "loss_iou": 0.35546875,
      "loss_num": 0.017333984375,
      "loss_xval": 0.796875,
      "num_input_tokens_seen": 66358172,
      "step": 1185
    },
    {
      "epoch": 2.6414253897550113,
      "grad_norm": 46.89886474609375,
      "learning_rate": 1e-06,
      "loss": 1.2667,
      "num_input_tokens_seen": 66415012,
      "step": 1186
    },
    {
      "epoch": 2.6414253897550113,
      "loss": 1.1698817014694214,
      "loss_ce": 0.0004481581272557378,
      "loss_iou": 0.47265625,
      "loss_num": 0.045166015625,
      "loss_xval": 1.171875,
      "num_input_tokens_seen": 66415012,
      "step": 1186
    },
    {
      "epoch": 2.643652561247216,
      "grad_norm": 18.3448429107666,
      "learning_rate": 1e-06,
      "loss": 1.0143,
      "num_input_tokens_seen": 66468628,
      "step": 1187
    },
    {
      "epoch": 2.643652561247216,
      "loss": 1.03020179271698,
      "loss_ce": 0.0004165967693552375,
      "loss_iou": 0.439453125,
      "loss_num": 0.030029296875,
      "loss_xval": 1.03125,
      "num_input_tokens_seen": 66468628,
      "step": 1187
    },
    {
      "epoch": 2.645879732739421,
      "grad_norm": 36.016719818115234,
      "learning_rate": 1e-06,
      "loss": 0.8041,
      "num_input_tokens_seen": 66524504,
      "step": 1188
    },
    {
      "epoch": 2.645879732739421,
      "loss": 0.7011793851852417,
      "loss_ce": 0.0004957778146490455,
      "loss_iou": 0.275390625,
      "loss_num": 0.0301513671875,
      "loss_xval": 0.69921875,
      "num_input_tokens_seen": 66524504,
      "step": 1188
    },
    {
      "epoch": 2.648106904231626,
      "grad_norm": 31.499238967895508,
      "learning_rate": 1e-06,
      "loss": 0.923,
      "num_input_tokens_seen": 66580192,
      "step": 1189
    },
    {
      "epoch": 2.648106904231626,
      "loss": 0.7815631031990051,
      "loss_ce": 0.00031307380413636565,
      "loss_iou": 0.287109375,
      "loss_num": 0.0419921875,
      "loss_xval": 0.78125,
      "num_input_tokens_seen": 66580192,
      "step": 1189
    },
    {
      "epoch": 2.6503340757238307,
      "grad_norm": 17.818838119506836,
      "learning_rate": 1e-06,
      "loss": 1.0473,
      "num_input_tokens_seen": 66637228,
      "step": 1190
    },
    {
      "epoch": 2.6503340757238307,
      "loss": 0.8321281671524048,
      "loss_ce": 0.0003410530334804207,
      "loss_iou": 0.3515625,
      "loss_num": 0.0260009765625,
      "loss_xval": 0.83203125,
      "num_input_tokens_seen": 66637228,
      "step": 1190
    },
    {
      "epoch": 2.6525612472160356,
      "grad_norm": 21.78716278076172,
      "learning_rate": 1e-06,
      "loss": 0.8932,
      "num_input_tokens_seen": 66695384,
      "step": 1191
    },
    {
      "epoch": 2.6525612472160356,
      "loss": 0.9930351376533508,
      "loss_ce": 0.0003593713045120239,
      "loss_iou": 0.41796875,
      "loss_num": 0.03125,
      "loss_xval": 0.9921875,
      "num_input_tokens_seen": 66695384,
      "step": 1191
    },
    {
      "epoch": 2.6547884187082404,
      "grad_norm": 19.491619110107422,
      "learning_rate": 1e-06,
      "loss": 0.8489,
      "num_input_tokens_seen": 66749192,
      "step": 1192
    },
    {
      "epoch": 2.6547884187082404,
      "loss": 0.7878109216690063,
      "loss_ce": 0.0011898394441232085,
      "loss_iou": 0.3359375,
      "loss_num": 0.023193359375,
      "loss_xval": 0.78515625,
      "num_input_tokens_seen": 66749192,
      "step": 1192
    },
    {
      "epoch": 2.6570155902004453,
      "grad_norm": 26.39838981628418,
      "learning_rate": 1e-06,
      "loss": 1.1329,
      "num_input_tokens_seen": 66803576,
      "step": 1193
    },
    {
      "epoch": 2.6570155902004453,
      "loss": 1.0078489780426025,
      "loss_ce": 0.0005247670924291015,
      "loss_iou": 0.421875,
      "loss_num": 0.03271484375,
      "loss_xval": 1.0078125,
      "num_input_tokens_seen": 66803576,
      "step": 1193
    },
    {
      "epoch": 2.65924276169265,
      "grad_norm": 23.46156883239746,
      "learning_rate": 1e-06,
      "loss": 0.8324,
      "num_input_tokens_seen": 66859052,
      "step": 1194
    },
    {
      "epoch": 2.65924276169265,
      "loss": 0.7302192449569702,
      "loss_ce": 0.000482962466776371,
      "loss_iou": 0.296875,
      "loss_num": 0.0274658203125,
      "loss_xval": 0.73046875,
      "num_input_tokens_seen": 66859052,
      "step": 1194
    },
    {
      "epoch": 2.661469933184855,
      "grad_norm": 16.806400299072266,
      "learning_rate": 1e-06,
      "loss": 1.0672,
      "num_input_tokens_seen": 66914164,
      "step": 1195
    },
    {
      "epoch": 2.661469933184855,
      "loss": 1.1170461177825928,
      "loss_ce": 0.000591110554523766,
      "loss_iou": 0.427734375,
      "loss_num": 0.052001953125,
      "loss_xval": 1.1171875,
      "num_input_tokens_seen": 66914164,
      "step": 1195
    },
    {
      "epoch": 2.66369710467706,
      "grad_norm": 18.408832550048828,
      "learning_rate": 1e-06,
      "loss": 1.0065,
      "num_input_tokens_seen": 66969884,
      "step": 1196
    },
    {
      "epoch": 2.66369710467706,
      "loss": 0.8114081621170044,
      "loss_ce": 0.00037299515679478645,
      "loss_iou": 0.341796875,
      "loss_num": 0.02587890625,
      "loss_xval": 0.8125,
      "num_input_tokens_seen": 66969884,
      "step": 1196
    },
    {
      "epoch": 2.665924276169265,
      "grad_norm": 23.257755279541016,
      "learning_rate": 1e-06,
      "loss": 1.0099,
      "num_input_tokens_seen": 67028592,
      "step": 1197
    },
    {
      "epoch": 2.665924276169265,
      "loss": 1.1483575105667114,
      "loss_ce": 0.0004082891682628542,
      "loss_iou": 0.451171875,
      "loss_num": 0.04931640625,
      "loss_xval": 1.1484375,
      "num_input_tokens_seen": 67028592,
      "step": 1197
    },
    {
      "epoch": 2.66815144766147,
      "grad_norm": 14.403682708740234,
      "learning_rate": 1e-06,
      "loss": 0.917,
      "num_input_tokens_seen": 67085904,
      "step": 1198
    },
    {
      "epoch": 2.66815144766147,
      "loss": 0.916401207447052,
      "loss_ce": 0.000385561550501734,
      "loss_iou": 0.3984375,
      "loss_num": 0.0234375,
      "loss_xval": 0.9140625,
      "num_input_tokens_seen": 67085904,
      "step": 1198
    },
    {
      "epoch": 2.670378619153675,
      "grad_norm": 16.39507484436035,
      "learning_rate": 1e-06,
      "loss": 0.7198,
      "num_input_tokens_seen": 67143124,
      "step": 1199
    },
    {
      "epoch": 2.670378619153675,
      "loss": 0.6678426861763,
      "loss_ce": 0.0003622480435296893,
      "loss_iou": 0.27734375,
      "loss_num": 0.0224609375,
      "loss_xval": 0.66796875,
      "num_input_tokens_seen": 67143124,
      "step": 1199
    },
    {
      "epoch": 2.6726057906458798,
      "grad_norm": 12.098701477050781,
      "learning_rate": 1e-06,
      "loss": 0.867,
      "num_input_tokens_seen": 67196720,
      "step": 1200
    },
    {
      "epoch": 2.6726057906458798,
      "loss": 1.0529296398162842,
      "loss_ce": 0.0005615358240902424,
      "loss_iou": 0.40625,
      "loss_num": 0.0478515625,
      "loss_xval": 1.0546875,
      "num_input_tokens_seen": 67196720,
      "step": 1200
    },
    {
      "epoch": 2.6748329621380846,
      "grad_norm": 23.761484146118164,
      "learning_rate": 1e-06,
      "loss": 0.9273,
      "num_input_tokens_seen": 67251244,
      "step": 1201
    },
    {
      "epoch": 2.6748329621380846,
      "loss": 0.9254390001296997,
      "loss_ce": 0.0003901528543792665,
      "loss_iou": 0.3359375,
      "loss_num": 0.051025390625,
      "loss_xval": 0.92578125,
      "num_input_tokens_seen": 67251244,
      "step": 1201
    },
    {
      "epoch": 2.6770601336302895,
      "grad_norm": 13.946866989135742,
      "learning_rate": 1e-06,
      "loss": 0.9489,
      "num_input_tokens_seen": 67305792,
      "step": 1202
    },
    {
      "epoch": 2.6770601336302895,
      "loss": 1.0536152124404907,
      "loss_ce": 0.0003925645723938942,
      "loss_iou": 0.439453125,
      "loss_num": 0.03466796875,
      "loss_xval": 1.0546875,
      "num_input_tokens_seen": 67305792,
      "step": 1202
    },
    {
      "epoch": 2.6792873051224944,
      "grad_norm": 14.783041954040527,
      "learning_rate": 1e-06,
      "loss": 0.7948,
      "num_input_tokens_seen": 67363808,
      "step": 1203
    },
    {
      "epoch": 2.6792873051224944,
      "loss": 0.7142338156700134,
      "loss_ce": 0.0010990574955940247,
      "loss_iou": 0.30859375,
      "loss_num": 0.0191650390625,
      "loss_xval": 0.71484375,
      "num_input_tokens_seen": 67363808,
      "step": 1203
    },
    {
      "epoch": 2.681514476614699,
      "grad_norm": 85.64546966552734,
      "learning_rate": 1e-06,
      "loss": 0.7912,
      "num_input_tokens_seen": 67419668,
      "step": 1204
    },
    {
      "epoch": 2.681514476614699,
      "loss": 0.7728584408760071,
      "loss_ce": 0.0003975207218900323,
      "loss_iou": 0.296875,
      "loss_num": 0.036376953125,
      "loss_xval": 0.7734375,
      "num_input_tokens_seen": 67419668,
      "step": 1204
    },
    {
      "epoch": 2.683741648106904,
      "grad_norm": 26.301057815551758,
      "learning_rate": 1e-06,
      "loss": 0.7419,
      "num_input_tokens_seen": 67477960,
      "step": 1205
    },
    {
      "epoch": 2.683741648106904,
      "loss": 0.5447392463684082,
      "loss_ce": 0.0003056719433516264,
      "loss_iou": 0.2138671875,
      "loss_num": 0.0235595703125,
      "loss_xval": 0.54296875,
      "num_input_tokens_seen": 67477960,
      "step": 1205
    },
    {
      "epoch": 2.6859688195991094,
      "grad_norm": 20.807857513427734,
      "learning_rate": 1e-06,
      "loss": 1.0459,
      "num_input_tokens_seen": 67530876,
      "step": 1206
    },
    {
      "epoch": 2.6859688195991094,
      "loss": 1.1529654264450073,
      "loss_ce": 0.0003775313380174339,
      "loss_iou": 0.50390625,
      "loss_num": 0.029052734375,
      "loss_xval": 1.15625,
      "num_input_tokens_seen": 67530876,
      "step": 1206
    },
    {
      "epoch": 2.6881959910913142,
      "grad_norm": 23.6107234954834,
      "learning_rate": 1e-06,
      "loss": 0.826,
      "num_input_tokens_seen": 67586056,
      "step": 1207
    },
    {
      "epoch": 2.6881959910913142,
      "loss": 0.9727563858032227,
      "loss_ce": 0.0005884337006136775,
      "loss_iou": 0.392578125,
      "loss_num": 0.037353515625,
      "loss_xval": 0.97265625,
      "num_input_tokens_seen": 67586056,
      "step": 1207
    },
    {
      "epoch": 2.690423162583519,
      "grad_norm": 17.839683532714844,
      "learning_rate": 1e-06,
      "loss": 0.9884,
      "num_input_tokens_seen": 67639676,
      "step": 1208
    },
    {
      "epoch": 2.690423162583519,
      "loss": 0.9732263088226318,
      "loss_ce": 0.0003259408404119313,
      "loss_iou": 0.40625,
      "loss_num": 0.0322265625,
      "loss_xval": 0.97265625,
      "num_input_tokens_seen": 67639676,
      "step": 1208
    },
    {
      "epoch": 2.692650334075724,
      "grad_norm": 71.87188720703125,
      "learning_rate": 1e-06,
      "loss": 1.0553,
      "num_input_tokens_seen": 67695312,
      "step": 1209
    },
    {
      "epoch": 2.692650334075724,
      "loss": 1.0907628536224365,
      "loss_ce": 0.0004308174247853458,
      "loss_iou": 0.4296875,
      "loss_num": 0.046142578125,
      "loss_xval": 1.09375,
      "num_input_tokens_seen": 67695312,
      "step": 1209
    },
    {
      "epoch": 2.694877505567929,
      "grad_norm": 21.322093963623047,
      "learning_rate": 1e-06,
      "loss": 0.7315,
      "num_input_tokens_seen": 67754484,
      "step": 1210
    },
    {
      "epoch": 2.694877505567929,
      "loss": 0.6453206539154053,
      "loss_ce": 0.00030111317755654454,
      "loss_iou": 0.279296875,
      "loss_num": 0.0169677734375,
      "loss_xval": 0.64453125,
      "num_input_tokens_seen": 67754484,
      "step": 1210
    },
    {
      "epoch": 2.6971046770601337,
      "grad_norm": 17.93378448486328,
      "learning_rate": 1e-06,
      "loss": 0.834,
      "num_input_tokens_seen": 67806972,
      "step": 1211
    },
    {
      "epoch": 2.6971046770601337,
      "loss": 0.879258930683136,
      "loss_ce": 0.00035265146289020777,
      "loss_iou": 0.375,
      "loss_num": 0.0257568359375,
      "loss_xval": 0.87890625,
      "num_input_tokens_seen": 67806972,
      "step": 1211
    },
    {
      "epoch": 2.6993318485523385,
      "grad_norm": 17.423023223876953,
      "learning_rate": 1e-06,
      "loss": 0.9924,
      "num_input_tokens_seen": 67865540,
      "step": 1212
    },
    {
      "epoch": 2.6993318485523385,
      "loss": 1.1357271671295166,
      "loss_ce": 0.0004733309615403414,
      "loss_iou": 0.46875,
      "loss_num": 0.0390625,
      "loss_xval": 1.1328125,
      "num_input_tokens_seen": 67865540,
      "step": 1212
    },
    {
      "epoch": 2.7015590200445434,
      "grad_norm": 16.1553955078125,
      "learning_rate": 1e-06,
      "loss": 0.8851,
      "num_input_tokens_seen": 67923660,
      "step": 1213
    },
    {
      "epoch": 2.7015590200445434,
      "loss": 0.8159765005111694,
      "loss_ce": 0.0004247480828780681,
      "loss_iou": 0.357421875,
      "loss_num": 0.0201416015625,
      "loss_xval": 0.81640625,
      "num_input_tokens_seen": 67923660,
      "step": 1213
    },
    {
      "epoch": 2.7037861915367483,
      "grad_norm": 27.426773071289062,
      "learning_rate": 1e-06,
      "loss": 1.0762,
      "num_input_tokens_seen": 67979792,
      "step": 1214
    },
    {
      "epoch": 2.7037861915367483,
      "loss": 0.9264326095581055,
      "loss_ce": 0.00040718415402807295,
      "loss_iou": 0.361328125,
      "loss_num": 0.04052734375,
      "loss_xval": 0.92578125,
      "num_input_tokens_seen": 67979792,
      "step": 1214
    },
    {
      "epoch": 2.706013363028953,
      "grad_norm": 143.9352569580078,
      "learning_rate": 1e-06,
      "loss": 1.2101,
      "num_input_tokens_seen": 68036028,
      "step": 1215
    },
    {
      "epoch": 2.706013363028953,
      "loss": 1.3111499547958374,
      "loss_ce": 0.0006030529621057212,
      "loss_iou": 0.53125,
      "loss_num": 0.0498046875,
      "loss_xval": 1.3125,
      "num_input_tokens_seen": 68036028,
      "step": 1215
    },
    {
      "epoch": 2.708240534521158,
      "grad_norm": 15.904016494750977,
      "learning_rate": 1e-06,
      "loss": 0.8784,
      "num_input_tokens_seen": 68090580,
      "step": 1216
    },
    {
      "epoch": 2.708240534521158,
      "loss": 0.8531776070594788,
      "loss_ce": 0.00039442608249373734,
      "loss_iou": 0.384765625,
      "loss_num": 0.016845703125,
      "loss_xval": 0.8515625,
      "num_input_tokens_seen": 68090580,
      "step": 1216
    },
    {
      "epoch": 2.710467706013363,
      "grad_norm": 16.765676498413086,
      "learning_rate": 1e-06,
      "loss": 0.7818,
      "num_input_tokens_seen": 68146632,
      "step": 1217
    },
    {
      "epoch": 2.710467706013363,
      "loss": 0.8600921630859375,
      "loss_ce": 0.00047299021389335394,
      "loss_iou": 0.3671875,
      "loss_num": 0.02490234375,
      "loss_xval": 0.859375,
      "num_input_tokens_seen": 68146632,
      "step": 1217
    },
    {
      "epoch": 2.7126948775055677,
      "grad_norm": 16.13719367980957,
      "learning_rate": 1e-06,
      "loss": 0.8074,
      "num_input_tokens_seen": 68202376,
      "step": 1218
    },
    {
      "epoch": 2.7126948775055677,
      "loss": 0.9842748045921326,
      "loss_ce": 0.00038809922989457846,
      "loss_iou": 0.40625,
      "loss_num": 0.0341796875,
      "loss_xval": 0.984375,
      "num_input_tokens_seen": 68202376,
      "step": 1218
    },
    {
      "epoch": 2.7149220489977726,
      "grad_norm": 16.392423629760742,
      "learning_rate": 1e-06,
      "loss": 0.832,
      "num_input_tokens_seen": 68257756,
      "step": 1219
    },
    {
      "epoch": 2.7149220489977726,
      "loss": 0.6619054079055786,
      "loss_ce": 0.00028435979038476944,
      "loss_iou": 0.275390625,
      "loss_num": 0.0224609375,
      "loss_xval": 0.66015625,
      "num_input_tokens_seen": 68257756,
      "step": 1219
    },
    {
      "epoch": 2.717149220489978,
      "grad_norm": 39.93339920043945,
      "learning_rate": 1e-06,
      "loss": 0.8111,
      "num_input_tokens_seen": 68317524,
      "step": 1220
    },
    {
      "epoch": 2.717149220489978,
      "loss": 0.8016022443771362,
      "loss_ce": 0.0003327628946863115,
      "loss_iou": 0.3203125,
      "loss_num": 0.0322265625,
      "loss_xval": 0.80078125,
      "num_input_tokens_seen": 68317524,
      "step": 1220
    },
    {
      "epoch": 2.7193763919821827,
      "grad_norm": 32.62158203125,
      "learning_rate": 1e-06,
      "loss": 0.8964,
      "num_input_tokens_seen": 68370704,
      "step": 1221
    },
    {
      "epoch": 2.7193763919821827,
      "loss": 0.9595242738723755,
      "loss_ce": 0.0005399275105446577,
      "loss_iou": 0.404296875,
      "loss_num": 0.02978515625,
      "loss_xval": 0.9609375,
      "num_input_tokens_seen": 68370704,
      "step": 1221
    },
    {
      "epoch": 2.7216035634743876,
      "grad_norm": 26.76487159729004,
      "learning_rate": 1e-06,
      "loss": 1.0057,
      "num_input_tokens_seen": 68426772,
      "step": 1222
    },
    {
      "epoch": 2.7216035634743876,
      "loss": 0.91068434715271,
      "loss_ce": 0.00028396639390848577,
      "loss_iou": 0.37890625,
      "loss_num": 0.0303955078125,
      "loss_xval": 0.91015625,
      "num_input_tokens_seen": 68426772,
      "step": 1222
    },
    {
      "epoch": 2.7238307349665924,
      "grad_norm": 18.720125198364258,
      "learning_rate": 1e-06,
      "loss": 0.8459,
      "num_input_tokens_seen": 68483180,
      "step": 1223
    },
    {
      "epoch": 2.7238307349665924,
      "loss": 0.9395471215248108,
      "loss_ce": 0.0003381132846698165,
      "loss_iou": 0.380859375,
      "loss_num": 0.03564453125,
      "loss_xval": 0.9375,
      "num_input_tokens_seen": 68483180,
      "step": 1223
    },
    {
      "epoch": 2.7260579064587973,
      "grad_norm": 15.632884979248047,
      "learning_rate": 1e-06,
      "loss": 1.093,
      "num_input_tokens_seen": 68538720,
      "step": 1224
    },
    {
      "epoch": 2.7260579064587973,
      "loss": 1.2165559530258179,
      "loss_ce": 0.0004914908786304295,
      "loss_iou": 0.484375,
      "loss_num": 0.0498046875,
      "loss_xval": 1.21875,
      "num_input_tokens_seen": 68538720,
      "step": 1224
    },
    {
      "epoch": 2.728285077951002,
      "grad_norm": 16.02979850769043,
      "learning_rate": 1e-06,
      "loss": 0.8688,
      "num_input_tokens_seen": 68596884,
      "step": 1225
    },
    {
      "epoch": 2.728285077951002,
      "loss": 1.0046154260635376,
      "loss_ce": 0.00046502824989147484,
      "loss_iou": 0.427734375,
      "loss_num": 0.0299072265625,
      "loss_xval": 1.0078125,
      "num_input_tokens_seen": 68596884,
      "step": 1225
    },
    {
      "epoch": 2.730512249443207,
      "grad_norm": 19.00936508178711,
      "learning_rate": 1e-06,
      "loss": 0.8638,
      "num_input_tokens_seen": 68654332,
      "step": 1226
    },
    {
      "epoch": 2.730512249443207,
      "loss": 0.9798787832260132,
      "loss_ce": 0.00038662960287183523,
      "loss_iou": 0.373046875,
      "loss_num": 0.046875,
      "loss_xval": 0.98046875,
      "num_input_tokens_seen": 68654332,
      "step": 1226
    },
    {
      "epoch": 2.732739420935412,
      "grad_norm": 18.884357452392578,
      "learning_rate": 1e-06,
      "loss": 0.7695,
      "num_input_tokens_seen": 68713232,
      "step": 1227
    },
    {
      "epoch": 2.732739420935412,
      "loss": 0.8073458671569824,
      "loss_ce": 0.00046107626985758543,
      "loss_iou": 0.349609375,
      "loss_num": 0.0213623046875,
      "loss_xval": 0.80859375,
      "num_input_tokens_seen": 68713232,
      "step": 1227
    },
    {
      "epoch": 2.734966592427617,
      "grad_norm": 14.613882064819336,
      "learning_rate": 1e-06,
      "loss": 0.8354,
      "num_input_tokens_seen": 68768872,
      "step": 1228
    },
    {
      "epoch": 2.734966592427617,
      "loss": 0.870397686958313,
      "loss_ce": 0.0005245659267529845,
      "loss_iou": 0.353515625,
      "loss_num": 0.032470703125,
      "loss_xval": 0.87109375,
      "num_input_tokens_seen": 68768872,
      "step": 1228
    },
    {
      "epoch": 2.737193763919822,
      "grad_norm": 16.955501556396484,
      "learning_rate": 1e-06,
      "loss": 1.1796,
      "num_input_tokens_seen": 68822780,
      "step": 1229
    },
    {
      "epoch": 2.737193763919822,
      "loss": 1.0951062440872192,
      "loss_ce": 0.0003797074896283448,
      "loss_iou": 0.439453125,
      "loss_num": 0.043701171875,
      "loss_xval": 1.09375,
      "num_input_tokens_seen": 68822780,
      "step": 1229
    },
    {
      "epoch": 2.739420935412027,
      "grad_norm": 20.716533660888672,
      "learning_rate": 1e-06,
      "loss": 1.0516,
      "num_input_tokens_seen": 68877892,
      "step": 1230
    },
    {
      "epoch": 2.739420935412027,
      "loss": 1.0252530574798584,
      "loss_ce": 0.0008389821741729975,
      "loss_iou": 0.412109375,
      "loss_num": 0.0400390625,
      "loss_xval": 1.0234375,
      "num_input_tokens_seen": 68877892,
      "step": 1230
    },
    {
      "epoch": 2.7416481069042318,
      "grad_norm": 40.377017974853516,
      "learning_rate": 1e-06,
      "loss": 0.7578,
      "num_input_tokens_seen": 68935384,
      "step": 1231
    },
    {
      "epoch": 2.7416481069042318,
      "loss": 0.8238745331764221,
      "loss_ce": 0.0003882101736962795,
      "loss_iou": 0.34375,
      "loss_num": 0.0274658203125,
      "loss_xval": 0.82421875,
      "num_input_tokens_seen": 68935384,
      "step": 1231
    },
    {
      "epoch": 2.7438752783964366,
      "grad_norm": 20.555801391601562,
      "learning_rate": 1e-06,
      "loss": 0.9902,
      "num_input_tokens_seen": 68991268,
      "step": 1232
    },
    {
      "epoch": 2.7438752783964366,
      "loss": 1.0188990831375122,
      "loss_ce": 0.0003444222966209054,
      "loss_iou": 0.453125,
      "loss_num": 0.0225830078125,
      "loss_xval": 1.015625,
      "num_input_tokens_seen": 68991268,
      "step": 1232
    },
    {
      "epoch": 2.7461024498886415,
      "grad_norm": 17.75653648376465,
      "learning_rate": 1e-06,
      "loss": 0.9883,
      "num_input_tokens_seen": 69048292,
      "step": 1233
    },
    {
      "epoch": 2.7461024498886415,
      "loss": 1.4018105268478394,
      "loss_ce": 0.0009315803181380033,
      "loss_iou": 0.5390625,
      "loss_num": 0.06494140625,
      "loss_xval": 1.3984375,
      "num_input_tokens_seen": 69048292,
      "step": 1233
    },
    {
      "epoch": 2.7483296213808464,
      "grad_norm": 25.2138614654541,
      "learning_rate": 1e-06,
      "loss": 0.9916,
      "num_input_tokens_seen": 69102780,
      "step": 1234
    },
    {
      "epoch": 2.7483296213808464,
      "loss": 1.0661447048187256,
      "loss_ce": 0.00047087084385566413,
      "loss_iou": 0.421875,
      "loss_num": 0.044677734375,
      "loss_xval": 1.0625,
      "num_input_tokens_seen": 69102780,
      "step": 1234
    },
    {
      "epoch": 2.750556792873051,
      "grad_norm": 41.74817657470703,
      "learning_rate": 1e-06,
      "loss": 0.7839,
      "num_input_tokens_seen": 69157904,
      "step": 1235
    },
    {
      "epoch": 2.750556792873051,
      "loss": 0.9228337407112122,
      "loss_ce": 0.00047044423990882933,
      "loss_iou": 0.3828125,
      "loss_num": 0.03173828125,
      "loss_xval": 0.921875,
      "num_input_tokens_seen": 69157904,
      "step": 1235
    },
    {
      "epoch": 2.752783964365256,
      "grad_norm": 17.140514373779297,
      "learning_rate": 1e-06,
      "loss": 0.7179,
      "num_input_tokens_seen": 69215832,
      "step": 1236
    },
    {
      "epoch": 2.752783964365256,
      "loss": 0.8632500171661377,
      "loss_ce": 0.00045710656559094787,
      "loss_iou": 0.330078125,
      "loss_num": 0.040283203125,
      "loss_xval": 0.86328125,
      "num_input_tokens_seen": 69215832,
      "step": 1236
    },
    {
      "epoch": 2.755011135857461,
      "grad_norm": 43.60929870605469,
      "learning_rate": 1e-06,
      "loss": 0.9365,
      "num_input_tokens_seen": 69271512,
      "step": 1237
    },
    {
      "epoch": 2.755011135857461,
      "loss": 1.1083848476409912,
      "loss_ce": 0.0004747234925162047,
      "loss_iou": 0.486328125,
      "loss_num": 0.0272216796875,
      "loss_xval": 1.109375,
      "num_input_tokens_seen": 69271512,
      "step": 1237
    },
    {
      "epoch": 2.757238307349666,
      "grad_norm": 82.30256652832031,
      "learning_rate": 1e-06,
      "loss": 0.9312,
      "num_input_tokens_seen": 69328124,
      "step": 1238
    },
    {
      "epoch": 2.757238307349666,
      "loss": 1.1663775444030762,
      "loss_ce": 0.000606063287705183,
      "loss_iou": 0.466796875,
      "loss_num": 0.046142578125,
      "loss_xval": 1.1640625,
      "num_input_tokens_seen": 69328124,
      "step": 1238
    },
    {
      "epoch": 2.7594654788418707,
      "grad_norm": 18.209518432617188,
      "learning_rate": 1e-06,
      "loss": 0.9739,
      "num_input_tokens_seen": 69385368,
      "step": 1239
    },
    {
      "epoch": 2.7594654788418707,
      "loss": 1.1721243858337402,
      "loss_ce": 0.0022025578655302525,
      "loss_iou": 0.466796875,
      "loss_num": 0.047119140625,
      "loss_xval": 1.171875,
      "num_input_tokens_seen": 69385368,
      "step": 1239
    },
    {
      "epoch": 2.7616926503340755,
      "grad_norm": 27.468372344970703,
      "learning_rate": 1e-06,
      "loss": 0.7438,
      "num_input_tokens_seen": 69440864,
      "step": 1240
    },
    {
      "epoch": 2.7616926503340755,
      "loss": 0.7869477868080139,
      "loss_ce": 0.00032670435030013323,
      "loss_iou": 0.333984375,
      "loss_num": 0.0233154296875,
      "loss_xval": 0.78515625,
      "num_input_tokens_seen": 69440864,
      "step": 1240
    },
    {
      "epoch": 2.7639198218262804,
      "grad_norm": 29.165250778198242,
      "learning_rate": 1e-06,
      "loss": 0.9046,
      "num_input_tokens_seen": 69497304,
      "step": 1241
    },
    {
      "epoch": 2.7639198218262804,
      "loss": 0.9364851117134094,
      "loss_ce": 0.0004500235663726926,
      "loss_iou": 0.34375,
      "loss_num": 0.0498046875,
      "loss_xval": 0.9375,
      "num_input_tokens_seen": 69497304,
      "step": 1241
    },
    {
      "epoch": 2.7661469933184857,
      "grad_norm": 17.164466857910156,
      "learning_rate": 1e-06,
      "loss": 0.8518,
      "num_input_tokens_seen": 69554524,
      "step": 1242
    },
    {
      "epoch": 2.7661469933184857,
      "loss": 0.9595236778259277,
      "loss_ce": 0.0005393511964939535,
      "loss_iou": 0.404296875,
      "loss_num": 0.0301513671875,
      "loss_xval": 0.9609375,
      "num_input_tokens_seen": 69554524,
      "step": 1242
    },
    {
      "epoch": 2.7683741648106905,
      "grad_norm": 19.568635940551758,
      "learning_rate": 1e-06,
      "loss": 0.8746,
      "num_input_tokens_seen": 69611204,
      "step": 1243
    },
    {
      "epoch": 2.7683741648106905,
      "loss": 1.1134837865829468,
      "loss_ce": 0.004352906718850136,
      "loss_iou": 0.47265625,
      "loss_num": 0.033203125,
      "loss_xval": 1.109375,
      "num_input_tokens_seen": 69611204,
      "step": 1243
    },
    {
      "epoch": 2.7706013363028954,
      "grad_norm": 16.17546844482422,
      "learning_rate": 1e-06,
      "loss": 0.7277,
      "num_input_tokens_seen": 69668100,
      "step": 1244
    },
    {
      "epoch": 2.7706013363028954,
      "loss": 0.7887986898422241,
      "loss_ce": 0.00046862097224220634,
      "loss_iou": 0.318359375,
      "loss_num": 0.0301513671875,
      "loss_xval": 0.7890625,
      "num_input_tokens_seen": 69668100,
      "step": 1244
    },
    {
      "epoch": 2.7728285077951003,
      "grad_norm": 22.0662784576416,
      "learning_rate": 1e-06,
      "loss": 0.9972,
      "num_input_tokens_seen": 69723688,
      "step": 1245
    },
    {
      "epoch": 2.7728285077951003,
      "loss": 1.2928619384765625,
      "loss_ce": 0.00038146309088915586,
      "loss_iou": 0.5234375,
      "loss_num": 0.04931640625,
      "loss_xval": 1.2890625,
      "num_input_tokens_seen": 69723688,
      "step": 1245
    },
    {
      "epoch": 2.775055679287305,
      "grad_norm": 19.460460662841797,
      "learning_rate": 1e-06,
      "loss": 0.601,
      "num_input_tokens_seen": 69778896,
      "step": 1246
    },
    {
      "epoch": 2.775055679287305,
      "loss": 0.5189080238342285,
      "loss_ce": 0.000353323295712471,
      "loss_iou": 0.2197265625,
      "loss_num": 0.015869140625,
      "loss_xval": 0.51953125,
      "num_input_tokens_seen": 69778896,
      "step": 1246
    },
    {
      "epoch": 2.77728285077951,
      "grad_norm": 20.321802139282227,
      "learning_rate": 1e-06,
      "loss": 1.1584,
      "num_input_tokens_seen": 69836796,
      "step": 1247
    },
    {
      "epoch": 2.77728285077951,
      "loss": 1.031623125076294,
      "loss_ce": 0.00037302449345588684,
      "loss_iou": 0.4296875,
      "loss_num": 0.034912109375,
      "loss_xval": 1.03125,
      "num_input_tokens_seen": 69836796,
      "step": 1247
    },
    {
      "epoch": 2.779510022271715,
      "grad_norm": 15.81376838684082,
      "learning_rate": 1e-06,
      "loss": 0.8184,
      "num_input_tokens_seen": 69893548,
      "step": 1248
    },
    {
      "epoch": 2.779510022271715,
      "loss": 0.9024643898010254,
      "loss_ce": 0.00036481593269854784,
      "loss_iou": 0.365234375,
      "loss_num": 0.03466796875,
      "loss_xval": 0.90234375,
      "num_input_tokens_seen": 69893548,
      "step": 1248
    },
    {
      "epoch": 2.7817371937639197,
      "grad_norm": 131.67137145996094,
      "learning_rate": 1e-06,
      "loss": 1.0096,
      "num_input_tokens_seen": 69944348,
      "step": 1249
    },
    {
      "epoch": 2.7817371937639197,
      "loss": 0.9452086687088013,
      "loss_ce": 0.00038447632687166333,
      "loss_iou": 0.40625,
      "loss_num": 0.0264892578125,
      "loss_xval": 0.9453125,
      "num_input_tokens_seen": 69944348,
      "step": 1249
    },
    {
      "epoch": 2.7839643652561246,
      "grad_norm": 13.716216087341309,
      "learning_rate": 1e-06,
      "loss": 1.1417,
      "num_input_tokens_seen": 70001912,
      "step": 1250
    },
    {
      "epoch": 2.7839643652561246,
      "eval_seeclick_web_CIoU": 0.5651521682739258,
      "eval_seeclick_web_GIoU": 0.5592525601387024,
      "eval_seeclick_web_IoU": 0.5811098515987396,
      "eval_seeclick_web_MAE_all": 0.017407238017767668,
      "eval_seeclick_web_MAE_h": 0.01108331186696887,
      "eval_seeclick_web_MAE_w": 0.01834576530382037,
      "eval_seeclick_web_MAE_x_boxes": 0.008968821726739407,
      "eval_seeclick_web_MAE_y_boxes": 0.02286715735681355,
      "eval_seeclick_web_inside_bbox": 0.9166666567325592,
      "eval_seeclick_web_loss": 0.9634284973144531,
      "eval_seeclick_web_loss_ce": 0.00043979176552966237,
      "eval_seeclick_web_loss_iou": 0.439697265625,
      "eval_seeclick_web_loss_num": 0.013711929321289062,
      "eval_seeclick_web_loss_xval": 0.9478759765625,
      "eval_seeclick_web_runtime": 32.3264,
      "eval_seeclick_web_samples_per_second": 1.547,
      "eval_seeclick_web_steps_per_second": 0.062,
      "num_input_tokens_seen": 70001912,
      "step": 1250
    },
    {
      "epoch": 2.7839643652561246,
      "eval_icons_CIoU": 0.3112401217222214,
      "eval_icons_GIoU": 0.33725370466709137,
      "eval_icons_IoU": 0.38187770545482635,
      "eval_icons_MAE_all": 0.06834794208407402,
      "eval_icons_MAE_h": 0.03893335722386837,
      "eval_icons_MAE_w": 0.08004930429160595,
      "eval_icons_MAE_x_boxes": 0.05875684879720211,
      "eval_icons_MAE_y_boxes": 0.039277092553675175,
      "eval_icons_inside_bbox": 0.6336805522441864,
      "eval_icons_loss": 1.721555471420288,
      "eval_icons_loss_ce": 0.0008201654127333313,
      "eval_icons_loss_iou": 0.66845703125,
      "eval_icons_loss_num": 0.06448173522949219,
      "eval_icons_loss_xval": 1.65869140625,
      "eval_icons_runtime": 30.6205,
      "eval_icons_samples_per_second": 1.633,
      "eval_icons_steps_per_second": 0.065,
      "num_input_tokens_seen": 70001912,
      "step": 1250
    },
    {
      "epoch": 2.7839643652561246,
      "eval_screenspot_CIoU": 0.31326837340990704,
      "eval_screenspot_GIoU": 0.3382245600223541,
      "eval_screenspot_IoU": 0.3957456350326538,
      "eval_screenspot_MAE_all": 0.08052034179369609,
      "eval_screenspot_MAE_h": 0.04156584106385708,
      "eval_screenspot_MAE_w": 0.0848269909620285,
      "eval_screenspot_MAE_x_boxes": 0.11255322148402531,
      "eval_screenspot_MAE_y_boxes": 0.04932925725976626,
      "eval_screenspot_inside_bbox": 0.6045833428700765,
      "eval_screenspot_loss": 1.7774311304092407,
      "eval_screenspot_loss_ce": 0.0008034493657760322,
      "eval_screenspot_loss_iou": 0.7025553385416666,
      "eval_screenspot_loss_num": 0.09097544352213542,
      "eval_screenspot_loss_xval": 1.8603515625,
      "eval_screenspot_runtime": 54.3229,
      "eval_screenspot_samples_per_second": 1.638,
      "eval_screenspot_steps_per_second": 0.055,
      "num_input_tokens_seen": 70001912,
      "step": 1250
    },
    {
      "epoch": 2.7839643652561246,
      "eval_compot_CIoU": 0.3437846302986145,
      "eval_compot_GIoU": 0.3697910010814667,
      "eval_compot_IoU": 0.4004169702529907,
      "eval_compot_MAE_all": 0.021469497121870518,
      "eval_compot_MAE_h": 0.011028026696294546,
      "eval_compot_MAE_w": 0.02703993208706379,
      "eval_compot_MAE_x_boxes": 0.030306325759738684,
      "eval_compot_MAE_y_boxes": 0.007093302207067609,
      "eval_compot_inside_bbox": 0.6145833432674408,
      "eval_compot_loss": 1.3977539539337158,
      "eval_compot_loss_ce": 0.0004101828089915216,
      "eval_compot_loss_iou": 0.63037109375,
      "eval_compot_loss_num": 0.020694732666015625,
      "eval_compot_loss_xval": 1.36376953125,
      "eval_compot_runtime": 32.0716,
      "eval_compot_samples_per_second": 1.559,
      "eval_compot_steps_per_second": 0.062,
      "num_input_tokens_seen": 70001912,
      "step": 1250
    },
    {
      "epoch": 2.7839643652561246,
      "eval_custom_ui_val_CIoU": 0.41552355140447617,
      "eval_custom_ui_val_GIoU": 0.44775263799561393,
      "eval_custom_ui_val_IoU": 0.47426238159338635,
      "eval_custom_ui_val_MAE_all": 0.03807516168389055,
      "eval_custom_ui_val_MAE_h": 0.020889455763002236,
      "eval_custom_ui_val_MAE_w": 0.04122109152376652,
      "eval_custom_ui_val_MAE_x_boxes": 0.04385989842315515,
      "eval_custom_ui_val_MAE_y_boxes": 0.020979333875907794,
      "eval_custom_ui_val_inside_bbox": 0.6678240762816535,
      "eval_custom_ui_val_loss": 1.312791347503662,
      "eval_custom_ui_val_loss_ce": 0.0007190946586585293,
      "eval_custom_ui_val_loss_iou": 0.5456407335069444,
      "eval_custom_ui_val_loss_num": 0.03783988952636719,
      "eval_custom_ui_val_loss_xval": 1.2806260850694444,
      "eval_custom_ui_val_runtime": 95.8302,
      "eval_custom_ui_val_samples_per_second": 2.765,
      "eval_custom_ui_val_steps_per_second": 0.094,
      "num_input_tokens_seen": 70001912,
      "step": 1250
    },
    {
      "epoch": 2.7839643652561246,
      "loss": 1.004927635192871,
      "loss_ce": 0.0005332073196768761,
      "loss_iou": 0.4296875,
      "loss_num": 0.0286865234375,
      "loss_xval": 1.0078125,
      "num_input_tokens_seen": 70001912,
      "step": 1250
    },
    {
      "epoch": 2.78619153674833,
      "grad_norm": 19.672489166259766,
      "learning_rate": 1e-06,
      "loss": 0.8286,
      "num_input_tokens_seen": 70058016,
      "step": 1251
    },
    {
      "epoch": 2.78619153674833,
      "loss": 0.6981078386306763,
      "loss_ce": 0.00035395551822148263,
      "loss_iou": 0.294921875,
      "loss_num": 0.0213623046875,
      "loss_xval": 0.69921875,
      "num_input_tokens_seen": 70058016,
      "step": 1251
    },
    {
      "epoch": 2.7884187082405347,
      "grad_norm": 17.608375549316406,
      "learning_rate": 1e-06,
      "loss": 0.9146,
      "num_input_tokens_seen": 70112088,
      "step": 1252
    },
    {
      "epoch": 2.7884187082405347,
      "loss": 0.9438179135322571,
      "loss_ce": 0.0003364117001183331,
      "loss_iou": 0.38671875,
      "loss_num": 0.033935546875,
      "loss_xval": 0.9453125,
      "num_input_tokens_seen": 70112088,
      "step": 1252
    },
    {
      "epoch": 2.7906458797327396,
      "grad_norm": 14.264594078063965,
      "learning_rate": 1e-06,
      "loss": 0.8837,
      "num_input_tokens_seen": 70171184,
      "step": 1253
    },
    {
      "epoch": 2.7906458797327396,
      "loss": 0.9383166432380676,
      "loss_ce": 0.00032835284946486354,
      "loss_iou": 0.400390625,
      "loss_num": 0.0272216796875,
      "loss_xval": 0.9375,
      "num_input_tokens_seen": 70171184,
      "step": 1253
    },
    {
      "epoch": 2.7928730512249444,
      "grad_norm": 21.34050750732422,
      "learning_rate": 1e-06,
      "loss": 1.0205,
      "num_input_tokens_seen": 70227628,
      "step": 1254
    },
    {
      "epoch": 2.7928730512249444,
      "loss": 1.0269427299499512,
      "loss_ce": 0.0004534609033726156,
      "loss_iou": 0.4296875,
      "loss_num": 0.03369140625,
      "loss_xval": 1.0234375,
      "num_input_tokens_seen": 70227628,
      "step": 1254
    },
    {
      "epoch": 2.7951002227171493,
      "grad_norm": 14.754793167114258,
      "learning_rate": 1e-06,
      "loss": 0.9247,
      "num_input_tokens_seen": 70283180,
      "step": 1255
    },
    {
      "epoch": 2.7951002227171493,
      "loss": 0.9754384160041809,
      "loss_ce": 0.00034071545815095305,
      "loss_iou": 0.392578125,
      "loss_num": 0.037841796875,
      "loss_xval": 0.9765625,
      "num_input_tokens_seen": 70283180,
      "step": 1255
    },
    {
      "epoch": 2.797327394209354,
      "grad_norm": 21.97988510131836,
      "learning_rate": 1e-06,
      "loss": 1.0107,
      "num_input_tokens_seen": 70335640,
      "step": 1256
    },
    {
      "epoch": 2.797327394209354,
      "loss": 1.1570931673049927,
      "loss_ce": 0.0003548713284544647,
      "loss_iou": 0.50390625,
      "loss_num": 0.029541015625,
      "loss_xval": 1.15625,
      "num_input_tokens_seen": 70335640,
      "step": 1256
    },
    {
      "epoch": 2.799554565701559,
      "grad_norm": 38.17330551147461,
      "learning_rate": 1e-06,
      "loss": 0.7453,
      "num_input_tokens_seen": 70390748,
      "step": 1257
    },
    {
      "epoch": 2.799554565701559,
      "loss": 0.8263010382652283,
      "loss_ce": 0.00037330458872020245,
      "loss_iou": 0.37109375,
      "loss_num": 0.01611328125,
      "loss_xval": 0.82421875,
      "num_input_tokens_seen": 70390748,
      "step": 1257
    },
    {
      "epoch": 2.801781737193764,
      "grad_norm": 22.141918182373047,
      "learning_rate": 1e-06,
      "loss": 1.0046,
      "num_input_tokens_seen": 70446116,
      "step": 1258
    },
    {
      "epoch": 2.801781737193764,
      "loss": 1.124394178390503,
      "loss_ce": 0.0003707860014401376,
      "loss_iou": 0.4609375,
      "loss_num": 0.040283203125,
      "loss_xval": 1.125,
      "num_input_tokens_seen": 70446116,
      "step": 1258
    },
    {
      "epoch": 2.8040089086859687,
      "grad_norm": 115.46956634521484,
      "learning_rate": 1e-06,
      "loss": 0.9891,
      "num_input_tokens_seen": 70503256,
      "step": 1259
    },
    {
      "epoch": 2.8040089086859687,
      "loss": 0.8390759229660034,
      "loss_ce": 0.0004528118879534304,
      "loss_iou": 0.3828125,
      "loss_num": 0.0147705078125,
      "loss_xval": 0.83984375,
      "num_input_tokens_seen": 70503256,
      "step": 1259
    },
    {
      "epoch": 2.8062360801781736,
      "grad_norm": 18.191608428955078,
      "learning_rate": 1e-06,
      "loss": 0.6494,
      "num_input_tokens_seen": 70559428,
      "step": 1260
    },
    {
      "epoch": 2.8062360801781736,
      "loss": 0.4827492833137512,
      "loss_ce": 0.0003274133778177202,
      "loss_iou": 0.2041015625,
      "loss_num": 0.014892578125,
      "loss_xval": 0.482421875,
      "num_input_tokens_seen": 70559428,
      "step": 1260
    },
    {
      "epoch": 2.8084632516703785,
      "grad_norm": 47.72921371459961,
      "learning_rate": 1e-06,
      "loss": 0.937,
      "num_input_tokens_seen": 70614552,
      "step": 1261
    },
    {
      "epoch": 2.8084632516703785,
      "loss": 0.9824411869049072,
      "loss_ce": 0.0036814198829233646,
      "loss_iou": 0.41015625,
      "loss_num": 0.031494140625,
      "loss_xval": 0.98046875,
      "num_input_tokens_seen": 70614552,
      "step": 1261
    },
    {
      "epoch": 2.8106904231625833,
      "grad_norm": 15.726812362670898,
      "learning_rate": 1e-06,
      "loss": 0.8683,
      "num_input_tokens_seen": 70669544,
      "step": 1262
    },
    {
      "epoch": 2.8106904231625833,
      "loss": 1.0436663627624512,
      "loss_ce": 0.0004534028994385153,
      "loss_iou": 0.42578125,
      "loss_num": 0.0380859375,
      "loss_xval": 1.046875,
      "num_input_tokens_seen": 70669544,
      "step": 1262
    },
    {
      "epoch": 2.812917594654788,
      "grad_norm": 42.43077087402344,
      "learning_rate": 1e-06,
      "loss": 0.8308,
      "num_input_tokens_seen": 70725468,
      "step": 1263
    },
    {
      "epoch": 2.812917594654788,
      "loss": 1.070866346359253,
      "loss_ce": 0.0005538529367186129,
      "loss_iou": 0.400390625,
      "loss_num": 0.0537109375,
      "loss_xval": 1.0703125,
      "num_input_tokens_seen": 70725468,
      "step": 1263
    },
    {
      "epoch": 2.815144766146993,
      "grad_norm": 15.935296058654785,
      "learning_rate": 1e-06,
      "loss": 0.8946,
      "num_input_tokens_seen": 70780344,
      "step": 1264
    },
    {
      "epoch": 2.815144766146993,
      "loss": 0.5159022212028503,
      "loss_ce": 0.00027721430524252355,
      "loss_iou": 0.2197265625,
      "loss_num": 0.015380859375,
      "loss_xval": 0.515625,
      "num_input_tokens_seen": 70780344,
      "step": 1264
    },
    {
      "epoch": 2.8173719376391984,
      "grad_norm": 14.334254264831543,
      "learning_rate": 1e-06,
      "loss": 0.7528,
      "num_input_tokens_seen": 70836060,
      "step": 1265
    },
    {
      "epoch": 2.8173719376391984,
      "loss": 0.7817806601524353,
      "loss_ce": 0.0002865034039132297,
      "loss_iou": 0.33203125,
      "loss_num": 0.0233154296875,
      "loss_xval": 0.78125,
      "num_input_tokens_seen": 70836060,
      "step": 1265
    },
    {
      "epoch": 2.819599109131403,
      "grad_norm": 19.307138442993164,
      "learning_rate": 1e-06,
      "loss": 0.8431,
      "num_input_tokens_seen": 70893088,
      "step": 1266
    },
    {
      "epoch": 2.819599109131403,
      "loss": 0.9788370132446289,
      "loss_ce": 0.00032134755747392774,
      "loss_iou": 0.412109375,
      "loss_num": 0.030517578125,
      "loss_xval": 0.9765625,
      "num_input_tokens_seen": 70893088,
      "step": 1266
    },
    {
      "epoch": 2.821826280623608,
      "grad_norm": 25.13027000427246,
      "learning_rate": 1e-06,
      "loss": 0.9889,
      "num_input_tokens_seen": 70949580,
      "step": 1267
    },
    {
      "epoch": 2.821826280623608,
      "loss": 1.048335313796997,
      "loss_ce": 0.00048371872981078923,
      "loss_iou": 0.421875,
      "loss_num": 0.040771484375,
      "loss_xval": 1.046875,
      "num_input_tokens_seen": 70949580,
      "step": 1267
    },
    {
      "epoch": 2.824053452115813,
      "grad_norm": 23.69377899169922,
      "learning_rate": 1e-06,
      "loss": 0.7746,
      "num_input_tokens_seen": 71005544,
      "step": 1268
    },
    {
      "epoch": 2.824053452115813,
      "loss": 0.871030330657959,
      "loss_ce": 0.00042489959741942585,
      "loss_iou": 0.3203125,
      "loss_num": 0.046142578125,
      "loss_xval": 0.87109375,
      "num_input_tokens_seen": 71005544,
      "step": 1268
    },
    {
      "epoch": 2.826280623608018,
      "grad_norm": 17.32742691040039,
      "learning_rate": 1e-06,
      "loss": 0.6273,
      "num_input_tokens_seen": 71061960,
      "step": 1269
    },
    {
      "epoch": 2.826280623608018,
      "loss": 0.5913735628128052,
      "loss_ce": 0.0003091061080340296,
      "loss_iou": 0.25,
      "loss_num": 0.0184326171875,
      "loss_xval": 0.58984375,
      "num_input_tokens_seen": 71061960,
      "step": 1269
    },
    {
      "epoch": 2.8285077951002227,
      "grad_norm": 14.68806266784668,
      "learning_rate": 1e-06,
      "loss": 0.9952,
      "num_input_tokens_seen": 71118024,
      "step": 1270
    },
    {
      "epoch": 2.8285077951002227,
      "loss": 0.9375334978103638,
      "loss_ce": 0.00027759268414229155,
      "loss_iou": 0.400390625,
      "loss_num": 0.0274658203125,
      "loss_xval": 0.9375,
      "num_input_tokens_seen": 71118024,
      "step": 1270
    },
    {
      "epoch": 2.8307349665924275,
      "grad_norm": 11.978171348571777,
      "learning_rate": 1e-06,
      "loss": 0.933,
      "num_input_tokens_seen": 71170604,
      "step": 1271
    },
    {
      "epoch": 2.8307349665924275,
      "loss": 1.082184076309204,
      "loss_ce": 0.0006410967325791717,
      "loss_iou": 0.455078125,
      "loss_num": 0.033935546875,
      "loss_xval": 1.078125,
      "num_input_tokens_seen": 71170604,
      "step": 1271
    },
    {
      "epoch": 2.8329621380846324,
      "grad_norm": 16.340866088867188,
      "learning_rate": 1e-06,
      "loss": 0.7549,
      "num_input_tokens_seen": 71227240,
      "step": 1272
    },
    {
      "epoch": 2.8329621380846324,
      "loss": 0.6341564059257507,
      "loss_ce": 0.000367343716789037,
      "loss_iou": 0.27734375,
      "loss_num": 0.01556396484375,
      "loss_xval": 0.6328125,
      "num_input_tokens_seen": 71227240,
      "step": 1272
    },
    {
      "epoch": 2.8351893095768377,
      "grad_norm": 21.42453384399414,
      "learning_rate": 1e-06,
      "loss": 1.3223,
      "num_input_tokens_seen": 71281800,
      "step": 1273
    },
    {
      "epoch": 2.8351893095768377,
      "loss": 1.39503812789917,
      "loss_ce": 0.0005069249891676009,
      "loss_iou": 0.609375,
      "loss_num": 0.034912109375,
      "loss_xval": 1.390625,
      "num_input_tokens_seen": 71281800,
      "step": 1273
    },
    {
      "epoch": 2.8374164810690425,
      "grad_norm": 14.59418773651123,
      "learning_rate": 1e-06,
      "loss": 0.858,
      "num_input_tokens_seen": 71338428,
      "step": 1274
    },
    {
      "epoch": 2.8374164810690425,
      "loss": 0.6660067439079285,
      "loss_ce": 0.00047936852206476033,
      "loss_iou": 0.302734375,
      "loss_num": 0.011962890625,
      "loss_xval": 0.6640625,
      "num_input_tokens_seen": 71338428,
      "step": 1274
    },
    {
      "epoch": 2.8396436525612474,
      "grad_norm": 23.267770767211914,
      "learning_rate": 1e-06,
      "loss": 1.0233,
      "num_input_tokens_seen": 71393332,
      "step": 1275
    },
    {
      "epoch": 2.8396436525612474,
      "loss": 1.2549712657928467,
      "loss_ce": 0.0005767274415120482,
      "loss_iou": 0.4765625,
      "loss_num": 0.06103515625,
      "loss_xval": 1.2578125,
      "num_input_tokens_seen": 71393332,
      "step": 1275
    },
    {
      "epoch": 2.8418708240534523,
      "grad_norm": 19.297468185424805,
      "learning_rate": 1e-06,
      "loss": 0.863,
      "num_input_tokens_seen": 71450628,
      "step": 1276
    },
    {
      "epoch": 2.8418708240534523,
      "loss": 0.8579794764518738,
      "loss_ce": 0.00031341775320470333,
      "loss_iou": 0.349609375,
      "loss_num": 0.03125,
      "loss_xval": 0.859375,
      "num_input_tokens_seen": 71450628,
      "step": 1276
    },
    {
      "epoch": 2.844097995545657,
      "grad_norm": 20.925310134887695,
      "learning_rate": 1e-06,
      "loss": 0.785,
      "num_input_tokens_seen": 71509096,
      "step": 1277
    },
    {
      "epoch": 2.844097995545657,
      "loss": 0.7992924451828003,
      "loss_ce": 0.0003422028967179358,
      "loss_iou": 0.310546875,
      "loss_num": 0.03564453125,
      "loss_xval": 0.80078125,
      "num_input_tokens_seen": 71509096,
      "step": 1277
    },
    {
      "epoch": 2.846325167037862,
      "grad_norm": 21.453073501586914,
      "learning_rate": 1e-06,
      "loss": 0.7625,
      "num_input_tokens_seen": 71567296,
      "step": 1278
    },
    {
      "epoch": 2.846325167037862,
      "loss": 0.7048360705375671,
      "loss_ce": 0.0004903356893919408,
      "loss_iou": 0.3046875,
      "loss_num": 0.01904296875,
      "loss_xval": 0.703125,
      "num_input_tokens_seen": 71567296,
      "step": 1278
    },
    {
      "epoch": 2.848552338530067,
      "grad_norm": 33.181209564208984,
      "learning_rate": 1e-06,
      "loss": 0.891,
      "num_input_tokens_seen": 71623012,
      "step": 1279
    },
    {
      "epoch": 2.848552338530067,
      "loss": 0.5747618675231934,
      "loss_ce": 0.00029898268985562027,
      "loss_iou": 0.23828125,
      "loss_num": 0.0194091796875,
      "loss_xval": 0.57421875,
      "num_input_tokens_seen": 71623012,
      "step": 1279
    },
    {
      "epoch": 2.8507795100222717,
      "grad_norm": 17.72443199157715,
      "learning_rate": 1e-06,
      "loss": 0.8168,
      "num_input_tokens_seen": 71680860,
      "step": 1280
    },
    {
      "epoch": 2.8507795100222717,
      "loss": 0.709847629070282,
      "loss_ce": 0.00037496426375582814,
      "loss_iou": 0.3046875,
      "loss_num": 0.0205078125,
      "loss_xval": 0.7109375,
      "num_input_tokens_seen": 71680860,
      "step": 1280
    },
    {
      "epoch": 2.8530066815144766,
      "grad_norm": 24.76962661743164,
      "learning_rate": 1e-06,
      "loss": 0.9463,
      "num_input_tokens_seen": 71734508,
      "step": 1281
    },
    {
      "epoch": 2.8530066815144766,
      "loss": 0.9339801669120789,
      "loss_ce": 0.0003863995661959052,
      "loss_iou": 0.3828125,
      "loss_num": 0.03369140625,
      "loss_xval": 0.93359375,
      "num_input_tokens_seen": 71734508,
      "step": 1281
    },
    {
      "epoch": 2.8552338530066814,
      "grad_norm": 22.486467361450195,
      "learning_rate": 1e-06,
      "loss": 0.7574,
      "num_input_tokens_seen": 71790040,
      "step": 1282
    },
    {
      "epoch": 2.8552338530066814,
      "loss": 0.8373603820800781,
      "loss_ce": 0.000446313846623525,
      "loss_iou": 0.3359375,
      "loss_num": 0.032470703125,
      "loss_xval": 0.8359375,
      "num_input_tokens_seen": 71790040,
      "step": 1282
    },
    {
      "epoch": 2.8574610244988863,
      "grad_norm": 22.6174373626709,
      "learning_rate": 1e-06,
      "loss": 0.9784,
      "num_input_tokens_seen": 71843828,
      "step": 1283
    },
    {
      "epoch": 2.8574610244988863,
      "loss": 1.0318081378936768,
      "loss_ce": 0.0005582119338214397,
      "loss_iou": 0.40625,
      "loss_num": 0.043212890625,
      "loss_xval": 1.03125,
      "num_input_tokens_seen": 71843828,
      "step": 1283
    },
    {
      "epoch": 2.859688195991091,
      "grad_norm": 18.821727752685547,
      "learning_rate": 1e-06,
      "loss": 0.939,
      "num_input_tokens_seen": 71900448,
      "step": 1284
    },
    {
      "epoch": 2.859688195991091,
      "loss": 0.8391966819763184,
      "loss_ce": 0.0006956788711249828,
      "loss_iou": 0.32421875,
      "loss_num": 0.037841796875,
      "loss_xval": 0.83984375,
      "num_input_tokens_seen": 71900448,
      "step": 1284
    },
    {
      "epoch": 2.861915367483296,
      "grad_norm": 13.810822486877441,
      "learning_rate": 1e-06,
      "loss": 0.9658,
      "num_input_tokens_seen": 71958344,
      "step": 1285
    },
    {
      "epoch": 2.861915367483296,
      "loss": 0.921851396560669,
      "loss_ce": 0.0004646632296498865,
      "loss_iou": 0.380859375,
      "loss_num": 0.031494140625,
      "loss_xval": 0.921875,
      "num_input_tokens_seen": 71958344,
      "step": 1285
    },
    {
      "epoch": 2.864142538975501,
      "grad_norm": 19.463180541992188,
      "learning_rate": 1e-06,
      "loss": 0.9619,
      "num_input_tokens_seen": 72014732,
      "step": 1286
    },
    {
      "epoch": 2.864142538975501,
      "loss": 1.0936338901519775,
      "loss_ce": 0.0003722285036928952,
      "loss_iou": 0.44140625,
      "loss_num": 0.041748046875,
      "loss_xval": 1.09375,
      "num_input_tokens_seen": 72014732,
      "step": 1286
    },
    {
      "epoch": 2.866369710467706,
      "grad_norm": 20.0224666595459,
      "learning_rate": 1e-06,
      "loss": 0.8855,
      "num_input_tokens_seen": 72071312,
      "step": 1287
    },
    {
      "epoch": 2.866369710467706,
      "loss": 0.9935523271560669,
      "loss_ce": 0.0011207055067643523,
      "loss_iou": 0.36328125,
      "loss_num": 0.053466796875,
      "loss_xval": 0.9921875,
      "num_input_tokens_seen": 72071312,
      "step": 1287
    },
    {
      "epoch": 2.868596881959911,
      "grad_norm": 18.678556442260742,
      "learning_rate": 1e-06,
      "loss": 0.8911,
      "num_input_tokens_seen": 72126324,
      "step": 1288
    },
    {
      "epoch": 2.868596881959911,
      "loss": 0.7727195024490356,
      "loss_ce": 0.00025854576961137354,
      "loss_iou": 0.3203125,
      "loss_num": 0.0260009765625,
      "loss_xval": 0.7734375,
      "num_input_tokens_seen": 72126324,
      "step": 1288
    },
    {
      "epoch": 2.870824053452116,
      "grad_norm": 13.80396556854248,
      "learning_rate": 1e-06,
      "loss": 0.9222,
      "num_input_tokens_seen": 72181044,
      "step": 1289
    },
    {
      "epoch": 2.870824053452116,
      "loss": 1.003150224685669,
      "loss_ce": 0.0004646561574190855,
      "loss_iou": 0.3984375,
      "loss_num": 0.041259765625,
      "loss_xval": 1.0,
      "num_input_tokens_seen": 72181044,
      "step": 1289
    },
    {
      "epoch": 2.8730512249443207,
      "grad_norm": 30.7203369140625,
      "learning_rate": 1e-06,
      "loss": 1.0782,
      "num_input_tokens_seen": 72238480,
      "step": 1290
    },
    {
      "epoch": 2.8730512249443207,
      "loss": 1.1529991626739502,
      "loss_ce": 0.0004112512688152492,
      "loss_iou": 0.51171875,
      "loss_num": 0.0262451171875,
      "loss_xval": 1.15625,
      "num_input_tokens_seen": 72238480,
      "step": 1290
    },
    {
      "epoch": 2.8752783964365256,
      "grad_norm": 25.847972869873047,
      "learning_rate": 1e-06,
      "loss": 0.8204,
      "num_input_tokens_seen": 72293396,
      "step": 1291
    },
    {
      "epoch": 2.8752783964365256,
      "loss": 1.011704444885254,
      "loss_ce": 0.0009622069192118943,
      "loss_iou": 0.380859375,
      "loss_num": 0.0498046875,
      "loss_xval": 1.0078125,
      "num_input_tokens_seen": 72293396,
      "step": 1291
    },
    {
      "epoch": 2.8775055679287305,
      "grad_norm": 100.3687744140625,
      "learning_rate": 1e-06,
      "loss": 0.8099,
      "num_input_tokens_seen": 72349376,
      "step": 1292
    },
    {
      "epoch": 2.8775055679287305,
      "loss": 0.7852460741996765,
      "loss_ce": 0.0003339627292007208,
      "loss_iou": 0.287109375,
      "loss_num": 0.0419921875,
      "loss_xval": 0.78515625,
      "num_input_tokens_seen": 72349376,
      "step": 1292
    },
    {
      "epoch": 2.8797327394209353,
      "grad_norm": 31.29021644592285,
      "learning_rate": 1e-06,
      "loss": 0.8552,
      "num_input_tokens_seen": 72405340,
      "step": 1293
    },
    {
      "epoch": 2.8797327394209353,
      "loss": 0.8164255619049072,
      "loss_ce": 0.0005075936205685139,
      "loss_iou": 0.333984375,
      "loss_num": 0.0296630859375,
      "loss_xval": 0.81640625,
      "num_input_tokens_seen": 72405340,
      "step": 1293
    },
    {
      "epoch": 2.88195991091314,
      "grad_norm": 18.81956672668457,
      "learning_rate": 1e-06,
      "loss": 1.1343,
      "num_input_tokens_seen": 72462008,
      "step": 1294
    },
    {
      "epoch": 2.88195991091314,
      "loss": 0.831671416759491,
      "loss_ce": 0.0003726041177287698,
      "loss_iou": 0.3359375,
      "loss_num": 0.031982421875,
      "loss_xval": 0.83203125,
      "num_input_tokens_seen": 72462008,
      "step": 1294
    },
    {
      "epoch": 2.884187082405345,
      "grad_norm": 28.06816291809082,
      "learning_rate": 1e-06,
      "loss": 0.7352,
      "num_input_tokens_seen": 72520860,
      "step": 1295
    },
    {
      "epoch": 2.884187082405345,
      "loss": 0.8308737277984619,
      "loss_ce": 0.0005514706717804074,
      "loss_iou": 0.3671875,
      "loss_num": 0.0191650390625,
      "loss_xval": 0.83203125,
      "num_input_tokens_seen": 72520860,
      "step": 1295
    },
    {
      "epoch": 2.8864142538975504,
      "grad_norm": 119.70653533935547,
      "learning_rate": 1e-06,
      "loss": 0.9987,
      "num_input_tokens_seen": 72578972,
      "step": 1296
    },
    {
      "epoch": 2.8864142538975504,
      "loss": 1.2035748958587646,
      "loss_ce": 0.00044986687134951353,
      "loss_iou": 0.482421875,
      "loss_num": 0.04736328125,
      "loss_xval": 1.203125,
      "num_input_tokens_seen": 72578972,
      "step": 1296
    },
    {
      "epoch": 2.888641425389755,
      "grad_norm": 23.493093490600586,
      "learning_rate": 1e-06,
      "loss": 0.9778,
      "num_input_tokens_seen": 72634868,
      "step": 1297
    },
    {
      "epoch": 2.888641425389755,
      "loss": 0.8210999965667725,
      "loss_ce": 0.0002992242225445807,
      "loss_iou": 0.353515625,
      "loss_num": 0.0224609375,
      "loss_xval": 0.8203125,
      "num_input_tokens_seen": 72634868,
      "step": 1297
    },
    {
      "epoch": 2.89086859688196,
      "grad_norm": 112.7999038696289,
      "learning_rate": 1e-06,
      "loss": 1.1468,
      "num_input_tokens_seen": 72686616,
      "step": 1298
    },
    {
      "epoch": 2.89086859688196,
      "loss": 1.0459107160568237,
      "loss_ce": 0.0007446431554853916,
      "loss_iou": 0.4140625,
      "loss_num": 0.04345703125,
      "loss_xval": 1.046875,
      "num_input_tokens_seen": 72686616,
      "step": 1298
    },
    {
      "epoch": 2.893095768374165,
      "grad_norm": 17.558181762695312,
      "learning_rate": 1e-06,
      "loss": 0.9396,
      "num_input_tokens_seen": 72740388,
      "step": 1299
    },
    {
      "epoch": 2.893095768374165,
      "loss": 0.843862771987915,
      "loss_ce": 0.0008451773319393396,
      "loss_iou": 0.359375,
      "loss_num": 0.0250244140625,
      "loss_xval": 0.84375,
      "num_input_tokens_seen": 72740388,
      "step": 1299
    },
    {
      "epoch": 2.89532293986637,
      "grad_norm": 20.879600524902344,
      "learning_rate": 1e-06,
      "loss": 0.9484,
      "num_input_tokens_seen": 72796572,
      "step": 1300
    },
    {
      "epoch": 2.89532293986637,
      "loss": 1.0628635883331299,
      "loss_ce": 0.00036358111537992954,
      "loss_iou": 0.4296875,
      "loss_num": 0.041015625,
      "loss_xval": 1.0625,
      "num_input_tokens_seen": 72796572,
      "step": 1300
    },
    {
      "epoch": 2.8975501113585747,
      "grad_norm": 24.84024429321289,
      "learning_rate": 1e-06,
      "loss": 0.8657,
      "num_input_tokens_seen": 72854948,
      "step": 1301
    },
    {
      "epoch": 2.8975501113585747,
      "loss": 0.7214281558990479,
      "loss_ce": 0.00035886449040845037,
      "loss_iou": 0.306640625,
      "loss_num": 0.021728515625,
      "loss_xval": 0.72265625,
      "num_input_tokens_seen": 72854948,
      "step": 1301
    },
    {
      "epoch": 2.8997772828507795,
      "grad_norm": 22.32291030883789,
      "learning_rate": 1e-06,
      "loss": 1.0913,
      "num_input_tokens_seen": 72907880,
      "step": 1302
    },
    {
      "epoch": 2.8997772828507795,
      "loss": 1.042357325553894,
      "loss_ce": 0.00036513677332550287,
      "loss_iou": 0.458984375,
      "loss_num": 0.0245361328125,
      "loss_xval": 1.0390625,
      "num_input_tokens_seen": 72907880,
      "step": 1302
    },
    {
      "epoch": 2.9020044543429844,
      "grad_norm": 94.72842407226562,
      "learning_rate": 1e-06,
      "loss": 0.8327,
      "num_input_tokens_seen": 72965036,
      "step": 1303
    },
    {
      "epoch": 2.9020044543429844,
      "loss": 0.7842406034469604,
      "loss_ce": 0.00030507519841194153,
      "loss_iou": 0.3203125,
      "loss_num": 0.028564453125,
      "loss_xval": 0.78515625,
      "num_input_tokens_seen": 72965036,
      "step": 1303
    },
    {
      "epoch": 2.9042316258351892,
      "grad_norm": 64.44385528564453,
      "learning_rate": 1e-06,
      "loss": 1.0017,
      "num_input_tokens_seen": 73019812,
      "step": 1304
    },
    {
      "epoch": 2.9042316258351892,
      "loss": 1.2140803337097168,
      "loss_ce": 0.0009456113912165165,
      "loss_iou": 0.46875,
      "loss_num": 0.055419921875,
      "loss_xval": 1.2109375,
      "num_input_tokens_seen": 73019812,
      "step": 1304
    },
    {
      "epoch": 2.906458797327394,
      "grad_norm": 63.43607711791992,
      "learning_rate": 1e-06,
      "loss": 0.6413,
      "num_input_tokens_seen": 73077144,
      "step": 1305
    },
    {
      "epoch": 2.906458797327394,
      "loss": 0.6670837998390198,
      "loss_ce": 0.0003357888199388981,
      "loss_iou": 0.302734375,
      "loss_num": 0.01202392578125,
      "loss_xval": 0.66796875,
      "num_input_tokens_seen": 73077144,
      "step": 1305
    },
    {
      "epoch": 2.908685968819599,
      "grad_norm": 19.118846893310547,
      "learning_rate": 1e-06,
      "loss": 0.8712,
      "num_input_tokens_seen": 73134624,
      "step": 1306
    },
    {
      "epoch": 2.908685968819599,
      "loss": 0.8654381036758423,
      "loss_ce": 0.00044786007492803037,
      "loss_iou": 0.37890625,
      "loss_num": 0.0208740234375,
      "loss_xval": 0.86328125,
      "num_input_tokens_seen": 73134624,
      "step": 1306
    },
    {
      "epoch": 2.910913140311804,
      "grad_norm": 15.382525444030762,
      "learning_rate": 1e-06,
      "loss": 0.9095,
      "num_input_tokens_seen": 73189224,
      "step": 1307
    },
    {
      "epoch": 2.910913140311804,
      "loss": 0.9012154340744019,
      "loss_ce": 0.0003364897274877876,
      "loss_iou": 0.384765625,
      "loss_num": 0.0260009765625,
      "loss_xval": 0.90234375,
      "num_input_tokens_seen": 73189224,
      "step": 1307
    },
    {
      "epoch": 2.9131403118040087,
      "grad_norm": 19.829851150512695,
      "learning_rate": 1e-06,
      "loss": 0.6906,
      "num_input_tokens_seen": 73245992,
      "step": 1308
    },
    {
      "epoch": 2.9131403118040087,
      "loss": 0.6446969509124756,
      "loss_ce": 0.00040984569932334125,
      "loss_iou": 0.267578125,
      "loss_num": 0.02197265625,
      "loss_xval": 0.64453125,
      "num_input_tokens_seen": 73245992,
      "step": 1308
    },
    {
      "epoch": 2.9153674832962135,
      "grad_norm": 29.750280380249023,
      "learning_rate": 1e-06,
      "loss": 0.9183,
      "num_input_tokens_seen": 73299640,
      "step": 1309
    },
    {
      "epoch": 2.9153674832962135,
      "loss": 0.7756353616714478,
      "loss_ce": 0.0004888884141109884,
      "loss_iou": 0.33203125,
      "loss_num": 0.0224609375,
      "loss_xval": 0.7734375,
      "num_input_tokens_seen": 73299640,
      "step": 1309
    },
    {
      "epoch": 2.917594654788419,
      "grad_norm": 14.211082458496094,
      "learning_rate": 1e-06,
      "loss": 0.8288,
      "num_input_tokens_seen": 73354816,
      "step": 1310
    },
    {
      "epoch": 2.917594654788419,
      "loss": 0.7674408555030823,
      "loss_ce": 0.0003510417591314763,
      "loss_iou": 0.328125,
      "loss_num": 0.0220947265625,
      "loss_xval": 0.765625,
      "num_input_tokens_seen": 73354816,
      "step": 1310
    },
    {
      "epoch": 2.9198218262806237,
      "grad_norm": 23.373992919921875,
      "learning_rate": 1e-06,
      "loss": 0.789,
      "num_input_tokens_seen": 73412668,
      "step": 1311
    },
    {
      "epoch": 2.9198218262806237,
      "loss": 0.7085351943969727,
      "loss_ce": 0.0005274016875773668,
      "loss_iou": 0.3125,
      "loss_num": 0.016845703125,
      "loss_xval": 0.70703125,
      "num_input_tokens_seen": 73412668,
      "step": 1311
    },
    {
      "epoch": 2.9220489977728286,
      "grad_norm": 16.870405197143555,
      "learning_rate": 1e-06,
      "loss": 1.0438,
      "num_input_tokens_seen": 73468408,
      "step": 1312
    },
    {
      "epoch": 2.9220489977728286,
      "loss": 0.9538618922233582,
      "loss_ce": 0.002201691037043929,
      "loss_iou": 0.421875,
      "loss_num": 0.0218505859375,
      "loss_xval": 0.953125,
      "num_input_tokens_seen": 73468408,
      "step": 1312
    },
    {
      "epoch": 2.9242761692650334,
      "grad_norm": 17.644731521606445,
      "learning_rate": 1e-06,
      "loss": 0.9705,
      "num_input_tokens_seen": 73525120,
      "step": 1313
    },
    {
      "epoch": 2.9242761692650334,
      "loss": 1.1021685600280762,
      "loss_ce": 0.0006060994928702712,
      "loss_iou": 0.4296875,
      "loss_num": 0.048828125,
      "loss_xval": 1.1015625,
      "num_input_tokens_seen": 73525120,
      "step": 1313
    },
    {
      "epoch": 2.9265033407572383,
      "grad_norm": 12.963172912597656,
      "learning_rate": 1e-06,
      "loss": 0.6575,
      "num_input_tokens_seen": 73581328,
      "step": 1314
    },
    {
      "epoch": 2.9265033407572383,
      "loss": 0.7298256754875183,
      "loss_ce": 0.00033350015291944146,
      "loss_iou": 0.30859375,
      "loss_num": 0.022216796875,
      "loss_xval": 0.73046875,
      "num_input_tokens_seen": 73581328,
      "step": 1314
    },
    {
      "epoch": 2.928730512249443,
      "grad_norm": 15.870047569274902,
      "learning_rate": 1e-06,
      "loss": 0.9587,
      "num_input_tokens_seen": 73636896,
      "step": 1315
    },
    {
      "epoch": 2.928730512249443,
      "loss": 0.8428421020507812,
      "loss_ce": 0.0003127763920929283,
      "loss_iou": 0.376953125,
      "loss_num": 0.0179443359375,
      "loss_xval": 0.84375,
      "num_input_tokens_seen": 73636896,
      "step": 1315
    },
    {
      "epoch": 2.930957683741648,
      "grad_norm": 14.364209175109863,
      "learning_rate": 1e-06,
      "loss": 0.8652,
      "num_input_tokens_seen": 73693568,
      "step": 1316
    },
    {
      "epoch": 2.930957683741648,
      "loss": 0.7832823395729065,
      "loss_ce": 0.0003234214964322746,
      "loss_iou": 0.337890625,
      "loss_num": 0.021240234375,
      "loss_xval": 0.78125,
      "num_input_tokens_seen": 73693568,
      "step": 1316
    },
    {
      "epoch": 2.933184855233853,
      "grad_norm": 32.526336669921875,
      "learning_rate": 1e-06,
      "loss": 0.9016,
      "num_input_tokens_seen": 73752056,
      "step": 1317
    },
    {
      "epoch": 2.933184855233853,
      "loss": 0.7434287071228027,
      "loss_ce": 0.0003867132472805679,
      "loss_iou": 0.33203125,
      "loss_num": 0.015869140625,
      "loss_xval": 0.7421875,
      "num_input_tokens_seen": 73752056,
      "step": 1317
    },
    {
      "epoch": 2.935412026726058,
      "grad_norm": 16.808130264282227,
      "learning_rate": 1e-06,
      "loss": 0.8978,
      "num_input_tokens_seen": 73809184,
      "step": 1318
    },
    {
      "epoch": 2.935412026726058,
      "loss": 0.7791517972946167,
      "loss_ce": 0.0003431940567679703,
      "loss_iou": 0.326171875,
      "loss_num": 0.0255126953125,
      "loss_xval": 0.77734375,
      "num_input_tokens_seen": 73809184,
      "step": 1318
    },
    {
      "epoch": 2.937639198218263,
      "grad_norm": 25.0362548828125,
      "learning_rate": 1e-06,
      "loss": 0.9062,
      "num_input_tokens_seen": 73863224,
      "step": 1319
    },
    {
      "epoch": 2.937639198218263,
      "loss": 0.8815177083015442,
      "loss_ce": 0.00041416779276914895,
      "loss_iou": 0.375,
      "loss_num": 0.026123046875,
      "loss_xval": 0.8828125,
      "num_input_tokens_seen": 73863224,
      "step": 1319
    },
    {
      "epoch": 2.939866369710468,
      "grad_norm": 23.442148208618164,
      "learning_rate": 1e-06,
      "loss": 1.0457,
      "num_input_tokens_seen": 73919812,
      "step": 1320
    },
    {
      "epoch": 2.939866369710468,
      "loss": 0.9350742101669312,
      "loss_ce": 0.0002597759012132883,
      "loss_iou": 0.388671875,
      "loss_num": 0.03125,
      "loss_xval": 0.93359375,
      "num_input_tokens_seen": 73919812,
      "step": 1320
    },
    {
      "epoch": 2.9420935412026727,
      "grad_norm": 39.34425735473633,
      "learning_rate": 1e-06,
      "loss": 0.7751,
      "num_input_tokens_seen": 73974388,
      "step": 1321
    },
    {
      "epoch": 2.9420935412026727,
      "loss": 0.6952590346336365,
      "loss_ce": 0.00031273809145204723,
      "loss_iou": 0.302734375,
      "loss_num": 0.0179443359375,
      "loss_xval": 0.6953125,
      "num_input_tokens_seen": 73974388,
      "step": 1321
    },
    {
      "epoch": 2.9443207126948776,
      "grad_norm": 15.462008476257324,
      "learning_rate": 1e-06,
      "loss": 0.9421,
      "num_input_tokens_seen": 74032576,
      "step": 1322
    },
    {
      "epoch": 2.9443207126948776,
      "loss": 0.9447818994522095,
      "loss_ce": 0.0009342257399111986,
      "loss_iou": 0.33984375,
      "loss_num": 0.052978515625,
      "loss_xval": 0.9453125,
      "num_input_tokens_seen": 74032576,
      "step": 1322
    },
    {
      "epoch": 2.9465478841870825,
      "grad_norm": 14.060518264770508,
      "learning_rate": 1e-06,
      "loss": 0.7786,
      "num_input_tokens_seen": 74087264,
      "step": 1323
    },
    {
      "epoch": 2.9465478841870825,
      "loss": 0.6352978944778442,
      "loss_ce": 0.0005322670331224799,
      "loss_iou": 0.28125,
      "loss_num": 0.01483154296875,
      "loss_xval": 0.6328125,
      "num_input_tokens_seen": 74087264,
      "step": 1323
    },
    {
      "epoch": 2.9487750556792873,
      "grad_norm": 22.72332763671875,
      "learning_rate": 1e-06,
      "loss": 1.0088,
      "num_input_tokens_seen": 74142868,
      "step": 1324
    },
    {
      "epoch": 2.9487750556792873,
      "loss": 1.0528690814971924,
      "loss_ce": 0.0006230201106518507,
      "loss_iou": 0.455078125,
      "loss_num": 0.028564453125,
      "loss_xval": 1.0546875,
      "num_input_tokens_seen": 74142868,
      "step": 1324
    },
    {
      "epoch": 2.951002227171492,
      "grad_norm": 12.463337898254395,
      "learning_rate": 1e-06,
      "loss": 1.0772,
      "num_input_tokens_seen": 74195720,
      "step": 1325
    },
    {
      "epoch": 2.951002227171492,
      "loss": 1.2419848442077637,
      "loss_ce": 0.0002855784259736538,
      "loss_iou": 0.53125,
      "loss_num": 0.036376953125,
      "loss_xval": 1.2421875,
      "num_input_tokens_seen": 74195720,
      "step": 1325
    },
    {
      "epoch": 2.953229398663697,
      "grad_norm": 21.965835571289062,
      "learning_rate": 1e-06,
      "loss": 1.1917,
      "num_input_tokens_seen": 74249536,
      "step": 1326
    },
    {
      "epoch": 2.953229398663697,
      "loss": 0.998505711555481,
      "loss_ce": 0.00045889458851888776,
      "loss_iou": 0.4140625,
      "loss_num": 0.033935546875,
      "loss_xval": 1.0,
      "num_input_tokens_seen": 74249536,
      "step": 1326
    },
    {
      "epoch": 2.955456570155902,
      "grad_norm": 19.335357666015625,
      "learning_rate": 1e-06,
      "loss": 0.79,
      "num_input_tokens_seen": 74306892,
      "step": 1327
    },
    {
      "epoch": 2.955456570155902,
      "loss": 0.7617478370666504,
      "loss_ce": 0.0003953152918256819,
      "loss_iou": 0.314453125,
      "loss_num": 0.026611328125,
      "loss_xval": 0.76171875,
      "num_input_tokens_seen": 74306892,
      "step": 1327
    },
    {
      "epoch": 2.9576837416481068,
      "grad_norm": 17.01285743713379,
      "learning_rate": 1e-06,
      "loss": 1.0855,
      "num_input_tokens_seen": 74363500,
      "step": 1328
    },
    {
      "epoch": 2.9576837416481068,
      "loss": 1.1959009170532227,
      "loss_ce": 0.0015648790867999196,
      "loss_iou": 0.46484375,
      "loss_num": 0.05224609375,
      "loss_xval": 1.1953125,
      "num_input_tokens_seen": 74363500,
      "step": 1328
    },
    {
      "epoch": 2.9599109131403116,
      "grad_norm": 15.02244758605957,
      "learning_rate": 1e-06,
      "loss": 0.8086,
      "num_input_tokens_seen": 74420444,
      "step": 1329
    },
    {
      "epoch": 2.9599109131403116,
      "loss": 0.6169714331626892,
      "loss_ce": 0.00027223769575357437,
      "loss_iou": 0.265625,
      "loss_num": 0.0172119140625,
      "loss_xval": 0.6171875,
      "num_input_tokens_seen": 74420444,
      "step": 1329
    },
    {
      "epoch": 2.9621380846325165,
      "grad_norm": 28.502256393432617,
      "learning_rate": 1e-06,
      "loss": 0.9974,
      "num_input_tokens_seen": 74475096,
      "step": 1330
    },
    {
      "epoch": 2.9621380846325165,
      "loss": 1.2459745407104492,
      "loss_ce": 0.00036911843926645815,
      "loss_iou": 0.53515625,
      "loss_num": 0.03564453125,
      "loss_xval": 1.2421875,
      "num_input_tokens_seen": 74475096,
      "step": 1330
    },
    {
      "epoch": 2.9643652561247213,
      "grad_norm": 78.01403045654297,
      "learning_rate": 1e-06,
      "loss": 0.695,
      "num_input_tokens_seen": 74532852,
      "step": 1331
    },
    {
      "epoch": 2.9643652561247213,
      "loss": 0.6845568418502808,
      "loss_ce": 0.000352742470568046,
      "loss_iou": 0.306640625,
      "loss_num": 0.014404296875,
      "loss_xval": 0.68359375,
      "num_input_tokens_seen": 74532852,
      "step": 1331
    },
    {
      "epoch": 2.9665924276169267,
      "grad_norm": 15.447741508483887,
      "learning_rate": 1e-06,
      "loss": 0.778,
      "num_input_tokens_seen": 74588216,
      "step": 1332
    },
    {
      "epoch": 2.9665924276169267,
      "loss": 0.9468116760253906,
      "loss_ce": 0.00027843567659147084,
      "loss_iou": 0.388671875,
      "loss_num": 0.0341796875,
      "loss_xval": 0.9453125,
      "num_input_tokens_seen": 74588216,
      "step": 1332
    },
    {
      "epoch": 2.9688195991091315,
      "grad_norm": 20.878141403198242,
      "learning_rate": 1e-06,
      "loss": 0.8829,
      "num_input_tokens_seen": 74645576,
      "step": 1333
    },
    {
      "epoch": 2.9688195991091315,
      "loss": 1.0510952472686768,
      "loss_ce": 0.0003140345506835729,
      "loss_iou": 0.443359375,
      "loss_num": 0.033447265625,
      "loss_xval": 1.046875,
      "num_input_tokens_seen": 74645576,
      "step": 1333
    },
    {
      "epoch": 2.9710467706013364,
      "grad_norm": 15.299370765686035,
      "learning_rate": 1e-06,
      "loss": 1.2055,
      "num_input_tokens_seen": 74704232,
      "step": 1334
    },
    {
      "epoch": 2.9710467706013364,
      "loss": 1.2119064331054688,
      "loss_ce": 0.0012130287941545248,
      "loss_iou": 0.51171875,
      "loss_num": 0.037353515625,
      "loss_xval": 1.2109375,
      "num_input_tokens_seen": 74704232,
      "step": 1334
    },
    {
      "epoch": 2.9732739420935412,
      "grad_norm": 19.733963012695312,
      "learning_rate": 1e-06,
      "loss": 0.8997,
      "num_input_tokens_seen": 74757716,
      "step": 1335
    },
    {
      "epoch": 2.9732739420935412,
      "loss": 0.8220304250717163,
      "loss_ce": 0.0004972000606358051,
      "loss_iou": 0.345703125,
      "loss_num": 0.0257568359375,
      "loss_xval": 0.8203125,
      "num_input_tokens_seen": 74757716,
      "step": 1335
    },
    {
      "epoch": 2.975501113585746,
      "grad_norm": 23.857521057128906,
      "learning_rate": 1e-06,
      "loss": 0.9514,
      "num_input_tokens_seen": 74811252,
      "step": 1336
    },
    {
      "epoch": 2.975501113585746,
      "loss": 1.0858557224273682,
      "loss_ce": 0.0004064875829499215,
      "loss_iou": 0.455078125,
      "loss_num": 0.035400390625,
      "loss_xval": 1.0859375,
      "num_input_tokens_seen": 74811252,
      "step": 1336
    },
    {
      "epoch": 2.977728285077951,
      "grad_norm": 17.06792449951172,
      "learning_rate": 1e-06,
      "loss": 0.8545,
      "num_input_tokens_seen": 74867424,
      "step": 1337
    },
    {
      "epoch": 2.977728285077951,
      "loss": 0.698322057723999,
      "loss_ce": 0.0003239895449951291,
      "loss_iou": 0.32421875,
      "loss_num": 0.009765625,
      "loss_xval": 0.69921875,
      "num_input_tokens_seen": 74867424,
      "step": 1337
    },
    {
      "epoch": 2.979955456570156,
      "grad_norm": 21.56948471069336,
      "learning_rate": 1e-06,
      "loss": 0.8931,
      "num_input_tokens_seen": 74923808,
      "step": 1338
    },
    {
      "epoch": 2.979955456570156,
      "loss": 1.2361738681793213,
      "loss_ce": 0.0003339699178468436,
      "loss_iou": 0.494140625,
      "loss_num": 0.0498046875,
      "loss_xval": 1.234375,
      "num_input_tokens_seen": 74923808,
      "step": 1338
    },
    {
      "epoch": 2.9821826280623607,
      "grad_norm": 18.15468978881836,
      "learning_rate": 1e-06,
      "loss": 0.7672,
      "num_input_tokens_seen": 74978584,
      "step": 1339
    },
    {
      "epoch": 2.9821826280623607,
      "loss": 0.5524653792381287,
      "loss_ce": 0.0003413406084291637,
      "loss_iou": 0.232421875,
      "loss_num": 0.0174560546875,
      "loss_xval": 0.55078125,
      "num_input_tokens_seen": 74978584,
      "step": 1339
    },
    {
      "epoch": 2.984409799554566,
      "grad_norm": 19.710803985595703,
      "learning_rate": 1e-06,
      "loss": 0.9056,
      "num_input_tokens_seen": 75033200,
      "step": 1340
    },
    {
      "epoch": 2.984409799554566,
      "loss": 0.7972690463066101,
      "loss_ce": 0.00039405166171491146,
      "loss_iou": 0.328125,
      "loss_num": 0.02783203125,
      "loss_xval": 0.796875,
      "num_input_tokens_seen": 75033200,
      "step": 1340
    },
    {
      "epoch": 2.986636971046771,
      "grad_norm": 17.216930389404297,
      "learning_rate": 1e-06,
      "loss": 0.7153,
      "num_input_tokens_seen": 75088396,
      "step": 1341
    },
    {
      "epoch": 2.986636971046771,
      "loss": 0.7058947682380676,
      "loss_ce": 0.00032835971796885133,
      "loss_iou": 0.291015625,
      "loss_num": 0.0247802734375,
      "loss_xval": 0.70703125,
      "num_input_tokens_seen": 75088396,
      "step": 1341
    },
    {
      "epoch": 2.9888641425389757,
      "grad_norm": 40.40977478027344,
      "learning_rate": 1e-06,
      "loss": 1.0308,
      "num_input_tokens_seen": 75146988,
      "step": 1342
    },
    {
      "epoch": 2.9888641425389757,
      "loss": 0.981564462184906,
      "loss_ce": 0.000363294588169083,
      "loss_iou": 0.42578125,
      "loss_num": 0.0255126953125,
      "loss_xval": 0.98046875,
      "num_input_tokens_seen": 75146988,
      "step": 1342
    },
    {
      "epoch": 2.9910913140311806,
      "grad_norm": 409.5078430175781,
      "learning_rate": 1e-06,
      "loss": 1.0702,
      "num_input_tokens_seen": 75204000,
      "step": 1343
    },
    {
      "epoch": 2.9910913140311806,
      "loss": 0.9847346544265747,
      "loss_ce": 0.0003596529713831842,
      "loss_iou": 0.3984375,
      "loss_num": 0.037841796875,
      "loss_xval": 0.984375,
      "num_input_tokens_seen": 75204000,
      "step": 1343
    },
    {
      "epoch": 2.9933184855233854,
      "grad_norm": 14.489014625549316,
      "learning_rate": 1e-06,
      "loss": 0.9728,
      "num_input_tokens_seen": 75260556,
      "step": 1344
    },
    {
      "epoch": 2.9933184855233854,
      "loss": 1.0310882329940796,
      "loss_ce": 0.0003265210543759167,
      "loss_iou": 0.390625,
      "loss_num": 0.050048828125,
      "loss_xval": 1.03125,
      "num_input_tokens_seen": 75260556,
      "step": 1344
    },
    {
      "epoch": 2.9955456570155903,
      "grad_norm": 21.714834213256836,
      "learning_rate": 1e-06,
      "loss": 0.8475,
      "num_input_tokens_seen": 75318464,
      "step": 1345
    },
    {
      "epoch": 2.9955456570155903,
      "loss": 0.8694126605987549,
      "loss_ce": 0.0005162069573998451,
      "loss_iou": 0.384765625,
      "loss_num": 0.0198974609375,
      "loss_xval": 0.8671875,
      "num_input_tokens_seen": 75318464,
      "step": 1345
    },
    {
      "epoch": 2.997772828507795,
      "grad_norm": 18.179824829101562,
      "learning_rate": 1e-06,
      "loss": 0.8263,
      "num_input_tokens_seen": 75375036,
      "step": 1346
    },
    {
      "epoch": 2.997772828507795,
      "loss": 0.9615136384963989,
      "loss_ce": 0.0003319892566651106,
      "loss_iou": 0.376953125,
      "loss_num": 0.04150390625,
      "loss_xval": 0.9609375,
      "num_input_tokens_seen": 75375036,
      "step": 1346
    },
    {
      "epoch": 3.0,
      "grad_norm": 17.222766876220703,
      "learning_rate": 1e-06,
      "loss": 0.969,
      "num_input_tokens_seen": 75431008,
      "step": 1347
    },
    {
      "epoch": 3.0,
      "loss": 0.4970862567424774,
      "loss_ce": 0.00038217363180592656,
      "loss_iou": 0.2158203125,
      "loss_num": 0.01287841796875,
      "loss_xval": 0.49609375,
      "num_input_tokens_seen": 75431008,
      "step": 1347
    },
    {
      "epoch": 3.002227171492205,
      "grad_norm": 15.791449546813965,
      "learning_rate": 1e-06,
      "loss": 1.1072,
      "num_input_tokens_seen": 75486424,
      "step": 1348
    },
    {
      "epoch": 3.002227171492205,
      "loss": 1.2740671634674072,
      "loss_ce": 0.0006296620704233646,
      "loss_iou": 0.515625,
      "loss_num": 0.04833984375,
      "loss_xval": 1.2734375,
      "num_input_tokens_seen": 75486424,
      "step": 1348
    },
    {
      "epoch": 3.0044543429844097,
      "grad_norm": 18.678091049194336,
      "learning_rate": 1e-06,
      "loss": 0.8753,
      "num_input_tokens_seen": 75542876,
      "step": 1349
    },
    {
      "epoch": 3.0044543429844097,
      "loss": 0.7326757907867432,
      "loss_ce": 0.00025389608344994485,
      "loss_iou": 0.328125,
      "loss_num": 0.01513671875,
      "loss_xval": 0.734375,
      "num_input_tokens_seen": 75542876,
      "step": 1349
    },
    {
      "epoch": 3.0066815144766146,
      "grad_norm": 33.28176498413086,
      "learning_rate": 1e-06,
      "loss": 0.8996,
      "num_input_tokens_seen": 75599980,
      "step": 1350
    },
    {
      "epoch": 3.0066815144766146,
      "loss": 0.8423166275024414,
      "loss_ce": 0.00051976612303406,
      "loss_iou": 0.34375,
      "loss_num": 0.0308837890625,
      "loss_xval": 0.84375,
      "num_input_tokens_seen": 75599980,
      "step": 1350
    },
    {
      "epoch": 3.0089086859688194,
      "grad_norm": 16.385852813720703,
      "learning_rate": 1e-06,
      "loss": 0.9101,
      "num_input_tokens_seen": 75656456,
      "step": 1351
    },
    {
      "epoch": 3.0089086859688194,
      "loss": 1.0261783599853516,
      "loss_ce": 0.0005435821949504316,
      "loss_iou": 0.44140625,
      "loss_num": 0.0283203125,
      "loss_xval": 1.0234375,
      "num_input_tokens_seen": 75656456,
      "step": 1351
    },
    {
      "epoch": 3.0111358574610243,
      "grad_norm": 23.76874351501465,
      "learning_rate": 1e-06,
      "loss": 0.9519,
      "num_input_tokens_seen": 75715892,
      "step": 1352
    },
    {
      "epoch": 3.0111358574610243,
      "loss": 0.8311777114868164,
      "loss_ce": 0.0003672138263937086,
      "loss_iou": 0.353515625,
      "loss_num": 0.0244140625,
      "loss_xval": 0.83203125,
      "num_input_tokens_seen": 75715892,
      "step": 1352
    },
    {
      "epoch": 3.0133630289532296,
      "grad_norm": 20.76018714904785,
      "learning_rate": 1e-06,
      "loss": 0.8318,
      "num_input_tokens_seen": 75774108,
      "step": 1353
    },
    {
      "epoch": 3.0133630289532296,
      "loss": 0.8268052935600281,
      "loss_ce": 0.0006334498757496476,
      "loss_iou": 0.328125,
      "loss_num": 0.034423828125,
      "loss_xval": 0.828125,
      "num_input_tokens_seen": 75774108,
      "step": 1353
    },
    {
      "epoch": 3.0155902004454345,
      "grad_norm": 13.831488609313965,
      "learning_rate": 1e-06,
      "loss": 0.8734,
      "num_input_tokens_seen": 75829580,
      "step": 1354
    },
    {
      "epoch": 3.0155902004454345,
      "loss": 0.7394604682922363,
      "loss_ce": 0.00032472447492182255,
      "loss_iou": 0.30859375,
      "loss_num": 0.0242919921875,
      "loss_xval": 0.73828125,
      "num_input_tokens_seen": 75829580,
      "step": 1354
    },
    {
      "epoch": 3.0178173719376393,
      "grad_norm": 32.255157470703125,
      "learning_rate": 1e-06,
      "loss": 0.7957,
      "num_input_tokens_seen": 75888104,
      "step": 1355
    },
    {
      "epoch": 3.0178173719376393,
      "loss": 0.7028149366378784,
      "loss_ce": 0.0003003134625032544,
      "loss_iou": 0.29296875,
      "loss_num": 0.023193359375,
      "loss_xval": 0.703125,
      "num_input_tokens_seen": 75888104,
      "step": 1355
    },
    {
      "epoch": 3.020044543429844,
      "grad_norm": 38.09650802612305,
      "learning_rate": 1e-06,
      "loss": 0.8811,
      "num_input_tokens_seen": 75946020,
      "step": 1356
    },
    {
      "epoch": 3.020044543429844,
      "loss": 0.7325311899185181,
      "loss_ce": 0.00035350650432519615,
      "loss_iou": 0.306640625,
      "loss_num": 0.0240478515625,
      "loss_xval": 0.73046875,
      "num_input_tokens_seen": 75946020,
      "step": 1356
    },
    {
      "epoch": 3.022271714922049,
      "grad_norm": 16.011274337768555,
      "learning_rate": 1e-06,
      "loss": 0.7803,
      "num_input_tokens_seen": 76003528,
      "step": 1357
    },
    {
      "epoch": 3.022271714922049,
      "loss": 0.8663822412490845,
      "loss_ce": 0.0004154411144554615,
      "loss_iou": 0.380859375,
      "loss_num": 0.0211181640625,
      "loss_xval": 0.8671875,
      "num_input_tokens_seen": 76003528,
      "step": 1357
    },
    {
      "epoch": 3.024498886414254,
      "grad_norm": 33.57292556762695,
      "learning_rate": 1e-06,
      "loss": 0.9545,
      "num_input_tokens_seen": 76059112,
      "step": 1358
    },
    {
      "epoch": 3.024498886414254,
      "loss": 0.9820139408111572,
      "loss_ce": 0.00032445252873003483,
      "loss_iou": 0.37890625,
      "loss_num": 0.044921875,
      "loss_xval": 0.98046875,
      "num_input_tokens_seen": 76059112,
      "step": 1358
    },
    {
      "epoch": 3.0267260579064588,
      "grad_norm": 18.756921768188477,
      "learning_rate": 1e-06,
      "loss": 0.8035,
      "num_input_tokens_seen": 76115888,
      "step": 1359
    },
    {
      "epoch": 3.0267260579064588,
      "loss": 0.6882917284965515,
      "loss_ce": 0.00030343266553245485,
      "loss_iou": 0.3046875,
      "loss_num": 0.0155029296875,
      "loss_xval": 0.6875,
      "num_input_tokens_seen": 76115888,
      "step": 1359
    },
    {
      "epoch": 3.0289532293986636,
      "grad_norm": 46.1540641784668,
      "learning_rate": 1e-06,
      "loss": 0.9699,
      "num_input_tokens_seen": 76175268,
      "step": 1360
    },
    {
      "epoch": 3.0289532293986636,
      "loss": 0.7430423498153687,
      "loss_ce": 0.00036659964825958014,
      "loss_iou": 0.291015625,
      "loss_num": 0.032470703125,
      "loss_xval": 0.7421875,
      "num_input_tokens_seen": 76175268,
      "step": 1360
    },
    {
      "epoch": 3.0311804008908685,
      "grad_norm": 18.095458984375,
      "learning_rate": 1e-06,
      "loss": 1.0666,
      "num_input_tokens_seen": 76232400,
      "step": 1361
    },
    {
      "epoch": 3.0311804008908685,
      "loss": 0.8669412732124329,
      "loss_ce": 0.00048618431901559234,
      "loss_iou": 0.34765625,
      "loss_num": 0.0341796875,
      "loss_xval": 0.8671875,
      "num_input_tokens_seen": 76232400,
      "step": 1361
    },
    {
      "epoch": 3.0334075723830733,
      "grad_norm": 16.167116165161133,
      "learning_rate": 1e-06,
      "loss": 0.7831,
      "num_input_tokens_seen": 76287924,
      "step": 1362
    },
    {
      "epoch": 3.0334075723830733,
      "loss": 0.9040330648422241,
      "loss_ce": 0.00046859911526553333,
      "loss_iou": 0.3515625,
      "loss_num": 0.0400390625,
      "loss_xval": 0.90234375,
      "num_input_tokens_seen": 76287924,
      "step": 1362
    },
    {
      "epoch": 3.035634743875278,
      "grad_norm": 25.74004364013672,
      "learning_rate": 1e-06,
      "loss": 1.0122,
      "num_input_tokens_seen": 76344832,
      "step": 1363
    },
    {
      "epoch": 3.035634743875278,
      "loss": 1.1795121431350708,
      "loss_ce": 0.0005570236244238913,
      "loss_iou": 0.447265625,
      "loss_num": 0.05712890625,
      "loss_xval": 1.1796875,
      "num_input_tokens_seen": 76344832,
      "step": 1363
    },
    {
      "epoch": 3.0378619153674835,
      "grad_norm": 25.54564094543457,
      "learning_rate": 1e-06,
      "loss": 0.9171,
      "num_input_tokens_seen": 76399332,
      "step": 1364
    },
    {
      "epoch": 3.0378619153674835,
      "loss": 0.7992980480194092,
      "loss_ce": 0.0004698993288911879,
      "loss_iou": 0.359375,
      "loss_num": 0.0157470703125,
      "loss_xval": 0.796875,
      "num_input_tokens_seen": 76399332,
      "step": 1364
    },
    {
      "epoch": 3.0400890868596884,
      "grad_norm": 14.032575607299805,
      "learning_rate": 1e-06,
      "loss": 1.0658,
      "num_input_tokens_seen": 76452824,
      "step": 1365
    },
    {
      "epoch": 3.0400890868596884,
      "loss": 1.1477856636047363,
      "loss_ce": 0.000324697612086311,
      "loss_iou": 0.486328125,
      "loss_num": 0.034912109375,
      "loss_xval": 1.1484375,
      "num_input_tokens_seen": 76452824,
      "step": 1365
    },
    {
      "epoch": 3.0423162583518932,
      "grad_norm": 27.30165672302246,
      "learning_rate": 1e-06,
      "loss": 0.838,
      "num_input_tokens_seen": 76507928,
      "step": 1366
    },
    {
      "epoch": 3.0423162583518932,
      "loss": 1.014754056930542,
      "loss_ce": 0.00034971325658261776,
      "loss_iou": 0.41796875,
      "loss_num": 0.03515625,
      "loss_xval": 1.015625,
      "num_input_tokens_seen": 76507928,
      "step": 1366
    },
    {
      "epoch": 3.044543429844098,
      "grad_norm": 30.125919342041016,
      "learning_rate": 1e-06,
      "loss": 0.9891,
      "num_input_tokens_seen": 76561868,
      "step": 1367
    },
    {
      "epoch": 3.044543429844098,
      "loss": 1.0369231700897217,
      "loss_ce": 0.0007904000231064856,
      "loss_iou": 0.421875,
      "loss_num": 0.038330078125,
      "loss_xval": 1.0390625,
      "num_input_tokens_seen": 76561868,
      "step": 1367
    },
    {
      "epoch": 3.046770601336303,
      "grad_norm": 15.93181324005127,
      "learning_rate": 1e-06,
      "loss": 1.4316,
      "num_input_tokens_seen": 76615632,
      "step": 1368
    },
    {
      "epoch": 3.046770601336303,
      "loss": 1.6312259435653687,
      "loss_ce": 0.00036655933945439756,
      "loss_iou": 0.68359375,
      "loss_num": 0.052734375,
      "loss_xval": 1.6328125,
      "num_input_tokens_seen": 76615632,
      "step": 1368
    },
    {
      "epoch": 3.048997772828508,
      "grad_norm": 12.806370735168457,
      "learning_rate": 1e-06,
      "loss": 0.6985,
      "num_input_tokens_seen": 76672624,
      "step": 1369
    },
    {
      "epoch": 3.048997772828508,
      "loss": 0.7415366172790527,
      "loss_ce": 0.0003256290510762483,
      "loss_iou": 0.32421875,
      "loss_num": 0.018310546875,
      "loss_xval": 0.7421875,
      "num_input_tokens_seen": 76672624,
      "step": 1369
    },
    {
      "epoch": 3.0512249443207127,
      "grad_norm": 19.221446990966797,
      "learning_rate": 1e-06,
      "loss": 0.8891,
      "num_input_tokens_seen": 76729096,
      "step": 1370
    },
    {
      "epoch": 3.0512249443207127,
      "loss": 0.729907751083374,
      "loss_ce": 0.0004155657079536468,
      "loss_iou": 0.3125,
      "loss_num": 0.0208740234375,
      "loss_xval": 0.73046875,
      "num_input_tokens_seen": 76729096,
      "step": 1370
    },
    {
      "epoch": 3.0534521158129175,
      "grad_norm": 41.225067138671875,
      "learning_rate": 1e-06,
      "loss": 1.0878,
      "num_input_tokens_seen": 76783684,
      "step": 1371
    },
    {
      "epoch": 3.0534521158129175,
      "loss": 1.0934221744537354,
      "loss_ce": 0.0004046167596243322,
      "loss_iou": 0.44921875,
      "loss_num": 0.03955078125,
      "loss_xval": 1.09375,
      "num_input_tokens_seen": 76783684,
      "step": 1371
    },
    {
      "epoch": 3.0556792873051224,
      "grad_norm": 75.78993225097656,
      "learning_rate": 1e-06,
      "loss": 0.736,
      "num_input_tokens_seen": 76841952,
      "step": 1372
    },
    {
      "epoch": 3.0556792873051224,
      "loss": 0.587233304977417,
      "loss_ce": 0.00031919381581246853,
      "loss_iou": 0.251953125,
      "loss_num": 0.0162353515625,
      "loss_xval": 0.5859375,
      "num_input_tokens_seen": 76841952,
      "step": 1372
    },
    {
      "epoch": 3.0579064587973273,
      "grad_norm": 21.043071746826172,
      "learning_rate": 1e-06,
      "loss": 0.796,
      "num_input_tokens_seen": 76897836,
      "step": 1373
    },
    {
      "epoch": 3.0579064587973273,
      "loss": 0.8417444229125977,
      "loss_ce": 0.0004358206642791629,
      "loss_iou": 0.31640625,
      "loss_num": 0.0419921875,
      "loss_xval": 0.83984375,
      "num_input_tokens_seen": 76897836,
      "step": 1373
    },
    {
      "epoch": 3.060133630289532,
      "grad_norm": 15.805767059326172,
      "learning_rate": 1e-06,
      "loss": 0.9074,
      "num_input_tokens_seen": 76955312,
      "step": 1374
    },
    {
      "epoch": 3.060133630289532,
      "loss": 0.8958030343055725,
      "loss_ce": 0.0002952259674202651,
      "loss_iou": 0.369140625,
      "loss_num": 0.03173828125,
      "loss_xval": 0.89453125,
      "num_input_tokens_seen": 76955312,
      "step": 1374
    },
    {
      "epoch": 3.062360801781737,
      "grad_norm": 17.199810028076172,
      "learning_rate": 1e-06,
      "loss": 0.9986,
      "num_input_tokens_seen": 77010728,
      "step": 1375
    },
    {
      "epoch": 3.062360801781737,
      "loss": 0.8748060464859009,
      "loss_ce": 0.00029435683973133564,
      "loss_iou": 0.390625,
      "loss_num": 0.018310546875,
      "loss_xval": 0.875,
      "num_input_tokens_seen": 77010728,
      "step": 1375
    },
    {
      "epoch": 3.0645879732739423,
      "grad_norm": 25.19390869140625,
      "learning_rate": 1e-06,
      "loss": 0.8589,
      "num_input_tokens_seen": 77067240,
      "step": 1376
    },
    {
      "epoch": 3.0645879732739423,
      "loss": 0.8482306003570557,
      "loss_ce": 0.00033024855656549335,
      "loss_iou": 0.349609375,
      "loss_num": 0.0301513671875,
      "loss_xval": 0.84765625,
      "num_input_tokens_seen": 77067240,
      "step": 1376
    },
    {
      "epoch": 3.066815144766147,
      "grad_norm": 32.438499450683594,
      "learning_rate": 1e-06,
      "loss": 0.7083,
      "num_input_tokens_seen": 77124068,
      "step": 1377
    },
    {
      "epoch": 3.066815144766147,
      "loss": 0.721910297870636,
      "loss_ce": 0.0035265071783214808,
      "loss_iou": 0.310546875,
      "loss_num": 0.019287109375,
      "loss_xval": 0.71875,
      "num_input_tokens_seen": 77124068,
      "step": 1377
    },
    {
      "epoch": 3.069042316258352,
      "grad_norm": 17.08912467956543,
      "learning_rate": 1e-06,
      "loss": 0.9961,
      "num_input_tokens_seen": 77183416,
      "step": 1378
    },
    {
      "epoch": 3.069042316258352,
      "loss": 1.1389042139053345,
      "loss_ce": 0.0004764240875374526,
      "loss_iou": 0.48046875,
      "loss_num": 0.035400390625,
      "loss_xval": 1.140625,
      "num_input_tokens_seen": 77183416,
      "step": 1378
    },
    {
      "epoch": 3.071269487750557,
      "grad_norm": 23.56962776184082,
      "learning_rate": 1e-06,
      "loss": 0.7918,
      "num_input_tokens_seen": 77241704,
      "step": 1379
    },
    {
      "epoch": 3.071269487750557,
      "loss": 0.9982140064239502,
      "loss_ce": 0.000411315995734185,
      "loss_iou": 0.375,
      "loss_num": 0.050048828125,
      "loss_xval": 0.99609375,
      "num_input_tokens_seen": 77241704,
      "step": 1379
    },
    {
      "epoch": 3.0734966592427617,
      "grad_norm": 55.8472785949707,
      "learning_rate": 1e-06,
      "loss": 0.9511,
      "num_input_tokens_seen": 77297284,
      "step": 1380
    },
    {
      "epoch": 3.0734966592427617,
      "loss": 0.7722369432449341,
      "loss_ce": 0.0005083876312710345,
      "loss_iou": 0.33984375,
      "loss_num": 0.01806640625,
      "loss_xval": 0.7734375,
      "num_input_tokens_seen": 77297284,
      "step": 1380
    },
    {
      "epoch": 3.0757238307349666,
      "grad_norm": 19.836666107177734,
      "learning_rate": 1e-06,
      "loss": 0.9226,
      "num_input_tokens_seen": 77351728,
      "step": 1381
    },
    {
      "epoch": 3.0757238307349666,
      "loss": 0.816044807434082,
      "loss_ce": 0.0003710235469043255,
      "loss_iou": 0.333984375,
      "loss_num": 0.029296875,
      "loss_xval": 0.81640625,
      "num_input_tokens_seen": 77351728,
      "step": 1381
    },
    {
      "epoch": 3.0779510022271714,
      "grad_norm": 32.599266052246094,
      "learning_rate": 1e-06,
      "loss": 1.1038,
      "num_input_tokens_seen": 77408188,
      "step": 1382
    },
    {
      "epoch": 3.0779510022271714,
      "loss": 1.1422669887542725,
      "loss_ce": 0.0006654143799096346,
      "loss_iou": 0.474609375,
      "loss_num": 0.038818359375,
      "loss_xval": 1.140625,
      "num_input_tokens_seen": 77408188,
      "step": 1382
    },
    {
      "epoch": 3.0801781737193763,
      "grad_norm": 24.89826774597168,
      "learning_rate": 1e-06,
      "loss": 0.9169,
      "num_input_tokens_seen": 77463240,
      "step": 1383
    },
    {
      "epoch": 3.0801781737193763,
      "loss": 0.7601173520088196,
      "loss_ce": 0.000351747585227713,
      "loss_iou": 0.283203125,
      "loss_num": 0.0390625,
      "loss_xval": 0.7578125,
      "num_input_tokens_seen": 77463240,
      "step": 1383
    },
    {
      "epoch": 3.082405345211581,
      "grad_norm": 11.264007568359375,
      "learning_rate": 1e-06,
      "loss": 0.5929,
      "num_input_tokens_seen": 77520856,
      "step": 1384
    },
    {
      "epoch": 3.082405345211581,
      "loss": 0.5832231044769287,
      "loss_ce": 0.0009477226412855089,
      "loss_iou": 0.2177734375,
      "loss_num": 0.029052734375,
      "loss_xval": 0.58203125,
      "num_input_tokens_seen": 77520856,
      "step": 1384
    },
    {
      "epoch": 3.084632516703786,
      "grad_norm": 20.889442443847656,
      "learning_rate": 1e-06,
      "loss": 0.9322,
      "num_input_tokens_seen": 77575336,
      "step": 1385
    },
    {
      "epoch": 3.084632516703786,
      "loss": 0.9303987622261047,
      "loss_ce": 0.000467123172711581,
      "loss_iou": 0.39453125,
      "loss_num": 0.0283203125,
      "loss_xval": 0.9296875,
      "num_input_tokens_seen": 77575336,
      "step": 1385
    },
    {
      "epoch": 3.086859688195991,
      "grad_norm": 21.084487915039062,
      "learning_rate": 1e-06,
      "loss": 0.7207,
      "num_input_tokens_seen": 77631024,
      "step": 1386
    },
    {
      "epoch": 3.086859688195991,
      "loss": 0.8573043346405029,
      "loss_ce": 0.0003706898423843086,
      "loss_iou": 0.359375,
      "loss_num": 0.0274658203125,
      "loss_xval": 0.85546875,
      "num_input_tokens_seen": 77631024,
      "step": 1386
    },
    {
      "epoch": 3.089086859688196,
      "grad_norm": 27.087024688720703,
      "learning_rate": 1e-06,
      "loss": 0.8915,
      "num_input_tokens_seen": 77689520,
      "step": 1387
    },
    {
      "epoch": 3.089086859688196,
      "loss": 1.0073137283325195,
      "loss_ce": 0.00047768885269761086,
      "loss_iou": 0.416015625,
      "loss_num": 0.03515625,
      "loss_xval": 1.0078125,
      "num_input_tokens_seen": 77689520,
      "step": 1387
    },
    {
      "epoch": 3.091314031180401,
      "grad_norm": 13.960137367248535,
      "learning_rate": 1e-06,
      "loss": 0.8432,
      "num_input_tokens_seen": 77744956,
      "step": 1388
    },
    {
      "epoch": 3.091314031180401,
      "loss": 0.9717408418655396,
      "loss_ce": 0.0003052502288483083,
      "loss_iou": 0.4140625,
      "loss_num": 0.029052734375,
      "loss_xval": 0.97265625,
      "num_input_tokens_seen": 77744956,
      "step": 1388
    },
    {
      "epoch": 3.093541202672606,
      "grad_norm": 59.64553451538086,
      "learning_rate": 1e-06,
      "loss": 0.8163,
      "num_input_tokens_seen": 77800484,
      "step": 1389
    },
    {
      "epoch": 3.093541202672606,
      "loss": 0.7509998679161072,
      "loss_ce": 0.00026743774651549757,
      "loss_iou": 0.30859375,
      "loss_num": 0.02685546875,
      "loss_xval": 0.75,
      "num_input_tokens_seen": 77800484,
      "step": 1389
    },
    {
      "epoch": 3.0957683741648108,
      "grad_norm": 29.264339447021484,
      "learning_rate": 1e-06,
      "loss": 1.0133,
      "num_input_tokens_seen": 77857748,
      "step": 1390
    },
    {
      "epoch": 3.0957683741648108,
      "loss": 1.0564740896224976,
      "loss_ce": 0.0003217510529793799,
      "loss_iou": 0.43359375,
      "loss_num": 0.0380859375,
      "loss_xval": 1.0546875,
      "num_input_tokens_seen": 77857748,
      "step": 1390
    },
    {
      "epoch": 3.0979955456570156,
      "grad_norm": 16.573060989379883,
      "learning_rate": 1e-06,
      "loss": 0.8231,
      "num_input_tokens_seen": 77912436,
      "step": 1391
    },
    {
      "epoch": 3.0979955456570156,
      "loss": 0.7225488424301147,
      "loss_ce": 0.0006250177975744009,
      "loss_iou": 0.265625,
      "loss_num": 0.03857421875,
      "loss_xval": 0.72265625,
      "num_input_tokens_seen": 77912436,
      "step": 1391
    },
    {
      "epoch": 3.1002227171492205,
      "grad_norm": 23.289533615112305,
      "learning_rate": 1e-06,
      "loss": 0.6688,
      "num_input_tokens_seen": 77971256,
      "step": 1392
    },
    {
      "epoch": 3.1002227171492205,
      "loss": 0.6361066102981567,
      "loss_ce": 0.0015851398929953575,
      "loss_iou": 0.26171875,
      "loss_num": 0.02197265625,
      "loss_xval": 0.6328125,
      "num_input_tokens_seen": 77971256,
      "step": 1392
    },
    {
      "epoch": 3.1024498886414253,
      "grad_norm": 37.72678756713867,
      "learning_rate": 1e-06,
      "loss": 1.0144,
      "num_input_tokens_seen": 78026280,
      "step": 1393
    },
    {
      "epoch": 3.1024498886414253,
      "loss": 0.9221377968788147,
      "loss_ce": 0.0007510941359214485,
      "loss_iou": 0.392578125,
      "loss_num": 0.027587890625,
      "loss_xval": 0.921875,
      "num_input_tokens_seen": 78026280,
      "step": 1393
    },
    {
      "epoch": 3.10467706013363,
      "grad_norm": 62.46157455444336,
      "learning_rate": 1e-06,
      "loss": 0.8603,
      "num_input_tokens_seen": 78083452,
      "step": 1394
    },
    {
      "epoch": 3.10467706013363,
      "loss": 0.6961342096328735,
      "loss_ce": 0.000333470175974071,
      "loss_iou": 0.3125,
      "loss_num": 0.01434326171875,
      "loss_xval": 0.6953125,
      "num_input_tokens_seen": 78083452,
      "step": 1394
    },
    {
      "epoch": 3.106904231625835,
      "grad_norm": 29.57560920715332,
      "learning_rate": 1e-06,
      "loss": 0.8214,
      "num_input_tokens_seen": 78138744,
      "step": 1395
    },
    {
      "epoch": 3.106904231625835,
      "loss": 0.8320956230163574,
      "loss_ce": 0.00030853800126351416,
      "loss_iou": 0.36328125,
      "loss_num": 0.0211181640625,
      "loss_xval": 0.83203125,
      "num_input_tokens_seen": 78138744,
      "step": 1395
    },
    {
      "epoch": 3.10913140311804,
      "grad_norm": 23.34572982788086,
      "learning_rate": 1e-06,
      "loss": 0.9635,
      "num_input_tokens_seen": 78194916,
      "step": 1396
    },
    {
      "epoch": 3.10913140311804,
      "loss": 1.1087517738342285,
      "loss_ce": 0.0003533945418894291,
      "loss_iou": 0.458984375,
      "loss_num": 0.037841796875,
      "loss_xval": 1.109375,
      "num_input_tokens_seen": 78194916,
      "step": 1396
    },
    {
      "epoch": 3.111358574610245,
      "grad_norm": 19.57944679260254,
      "learning_rate": 1e-06,
      "loss": 0.7877,
      "num_input_tokens_seen": 78251972,
      "step": 1397
    },
    {
      "epoch": 3.111358574610245,
      "loss": 0.6399654150009155,
      "loss_ce": 0.0003169975243508816,
      "loss_iou": 0.28515625,
      "loss_num": 0.01422119140625,
      "loss_xval": 0.640625,
      "num_input_tokens_seen": 78251972,
      "step": 1397
    },
    {
      "epoch": 3.11358574610245,
      "grad_norm": 15.57333755493164,
      "learning_rate": 1e-06,
      "loss": 0.7701,
      "num_input_tokens_seen": 78304736,
      "step": 1398
    },
    {
      "epoch": 3.11358574610245,
      "loss": 0.7310658693313599,
      "loss_ce": 0.00035301962634548545,
      "loss_iou": 0.28125,
      "loss_num": 0.03369140625,
      "loss_xval": 0.73046875,
      "num_input_tokens_seen": 78304736,
      "step": 1398
    },
    {
      "epoch": 3.115812917594655,
      "grad_norm": 20.79068946838379,
      "learning_rate": 1e-06,
      "loss": 0.9479,
      "num_input_tokens_seen": 78360580,
      "step": 1399
    },
    {
      "epoch": 3.115812917594655,
      "loss": 0.7796562910079956,
      "loss_ce": 0.0006035708356648684,
      "loss_iou": 0.3125,
      "loss_num": 0.03076171875,
      "loss_xval": 0.77734375,
      "num_input_tokens_seen": 78360580,
      "step": 1399
    },
    {
      "epoch": 3.11804008908686,
      "grad_norm": 21.23663330078125,
      "learning_rate": 1e-06,
      "loss": 0.8893,
      "num_input_tokens_seen": 78417556,
      "step": 1400
    },
    {
      "epoch": 3.11804008908686,
      "loss": 0.6944925785064697,
      "loss_ce": 0.0004008029936812818,
      "loss_iou": 0.30078125,
      "loss_num": 0.018310546875,
      "loss_xval": 0.6953125,
      "num_input_tokens_seen": 78417556,
      "step": 1400
    },
    {
      "epoch": 3.1202672605790647,
      "grad_norm": 20.82147789001465,
      "learning_rate": 1e-06,
      "loss": 0.7757,
      "num_input_tokens_seen": 78472356,
      "step": 1401
    },
    {
      "epoch": 3.1202672605790647,
      "loss": 0.6926823258399963,
      "loss_ce": 0.00029954389901831746,
      "loss_iou": 0.294921875,
      "loss_num": 0.0203857421875,
      "loss_xval": 0.69140625,
      "num_input_tokens_seen": 78472356,
      "step": 1401
    },
    {
      "epoch": 3.1224944320712695,
      "grad_norm": 24.593259811401367,
      "learning_rate": 1e-06,
      "loss": 1.1074,
      "num_input_tokens_seen": 78529324,
      "step": 1402
    },
    {
      "epoch": 3.1224944320712695,
      "loss": 1.332648515701294,
      "loss_ce": 0.0006172613939270377,
      "loss_iou": 0.52734375,
      "loss_num": 0.0556640625,
      "loss_xval": 1.328125,
      "num_input_tokens_seen": 78529324,
      "step": 1402
    },
    {
      "epoch": 3.1247216035634744,
      "grad_norm": 16.219890594482422,
      "learning_rate": 1e-06,
      "loss": 0.8194,
      "num_input_tokens_seen": 78582848,
      "step": 1403
    },
    {
      "epoch": 3.1247216035634744,
      "loss": 0.8170123100280762,
      "loss_ce": 0.00036187097430229187,
      "loss_iou": 0.369140625,
      "loss_num": 0.01556396484375,
      "loss_xval": 0.81640625,
      "num_input_tokens_seen": 78582848,
      "step": 1403
    },
    {
      "epoch": 3.1269487750556793,
      "grad_norm": 24.382253646850586,
      "learning_rate": 1e-06,
      "loss": 0.9872,
      "num_input_tokens_seen": 78637908,
      "step": 1404
    },
    {
      "epoch": 3.1269487750556793,
      "loss": 0.9994413256645203,
      "loss_ce": 0.0004178856033831835,
      "loss_iou": 0.412109375,
      "loss_num": 0.03515625,
      "loss_xval": 1.0,
      "num_input_tokens_seen": 78637908,
      "step": 1404
    },
    {
      "epoch": 3.129175946547884,
      "grad_norm": 16.891210556030273,
      "learning_rate": 1e-06,
      "loss": 0.7293,
      "num_input_tokens_seen": 78692304,
      "step": 1405
    },
    {
      "epoch": 3.129175946547884,
      "loss": 0.6750854253768921,
      "loss_ce": 0.0002806965203490108,
      "loss_iou": 0.27734375,
      "loss_num": 0.0240478515625,
      "loss_xval": 0.67578125,
      "num_input_tokens_seen": 78692304,
      "step": 1405
    },
    {
      "epoch": 3.131403118040089,
      "grad_norm": 22.945178985595703,
      "learning_rate": 1e-06,
      "loss": 0.944,
      "num_input_tokens_seen": 78746900,
      "step": 1406
    },
    {
      "epoch": 3.131403118040089,
      "loss": 1.02711820602417,
      "loss_ce": 0.000506936979945749,
      "loss_iou": 0.431640625,
      "loss_num": 0.03271484375,
      "loss_xval": 1.0234375,
      "num_input_tokens_seen": 78746900,
      "step": 1406
    },
    {
      "epoch": 3.133630289532294,
      "grad_norm": 17.015195846557617,
      "learning_rate": 1e-06,
      "loss": 0.8693,
      "num_input_tokens_seen": 78804192,
      "step": 1407
    },
    {
      "epoch": 3.133630289532294,
      "loss": 0.7983866930007935,
      "loss_ce": 0.0002909940667450428,
      "loss_iou": 0.36328125,
      "loss_num": 0.01446533203125,
      "loss_xval": 0.796875,
      "num_input_tokens_seen": 78804192,
      "step": 1407
    },
    {
      "epoch": 3.1358574610244987,
      "grad_norm": 38.74119567871094,
      "learning_rate": 1e-06,
      "loss": 0.9071,
      "num_input_tokens_seen": 78858704,
      "step": 1408
    },
    {
      "epoch": 3.1358574610244987,
      "loss": 0.9215487241744995,
      "loss_ce": 0.00028408068465068936,
      "loss_iou": 0.40625,
      "loss_num": 0.021484375,
      "loss_xval": 0.921875,
      "num_input_tokens_seen": 78858704,
      "step": 1408
    },
    {
      "epoch": 3.138084632516704,
      "grad_norm": 16.403173446655273,
      "learning_rate": 1e-06,
      "loss": 0.8264,
      "num_input_tokens_seen": 78913596,
      "step": 1409
    },
    {
      "epoch": 3.138084632516704,
      "loss": 0.7442107796669006,
      "loss_ce": 0.0003143020730931312,
      "loss_iou": 0.28515625,
      "loss_num": 0.03466796875,
      "loss_xval": 0.7421875,
      "num_input_tokens_seen": 78913596,
      "step": 1409
    },
    {
      "epoch": 3.140311804008909,
      "grad_norm": 125.99543762207031,
      "learning_rate": 1e-06,
      "loss": 0.7983,
      "num_input_tokens_seen": 78971496,
      "step": 1410
    },
    {
      "epoch": 3.140311804008909,
      "loss": 0.7936596870422363,
      "loss_ce": 0.0006909238873049617,
      "loss_iou": 0.34765625,
      "loss_num": 0.01953125,
      "loss_xval": 0.79296875,
      "num_input_tokens_seen": 78971496,
      "step": 1410
    },
    {
      "epoch": 3.1425389755011137,
      "grad_norm": 29.1822509765625,
      "learning_rate": 1e-06,
      "loss": 0.7606,
      "num_input_tokens_seen": 79027964,
      "step": 1411
    },
    {
      "epoch": 3.1425389755011137,
      "loss": 0.5414670705795288,
      "loss_ce": 0.00032938801450654864,
      "loss_iou": 0.22265625,
      "loss_num": 0.0191650390625,
      "loss_xval": 0.54296875,
      "num_input_tokens_seen": 79027964,
      "step": 1411
    },
    {
      "epoch": 3.1447661469933186,
      "grad_norm": 62.62162399291992,
      "learning_rate": 1e-06,
      "loss": 1.0511,
      "num_input_tokens_seen": 79085588,
      "step": 1412
    },
    {
      "epoch": 3.1447661469933186,
      "loss": 0.763961911201477,
      "loss_ce": 0.00029001818620599806,
      "loss_iou": 0.3359375,
      "loss_num": 0.0184326171875,
      "loss_xval": 0.765625,
      "num_input_tokens_seen": 79085588,
      "step": 1412
    },
    {
      "epoch": 3.1469933184855234,
      "grad_norm": 20.36224937438965,
      "learning_rate": 1e-06,
      "loss": 0.9567,
      "num_input_tokens_seen": 79143720,
      "step": 1413
    },
    {
      "epoch": 3.1469933184855234,
      "loss": 0.7871879935264587,
      "loss_ce": 0.00032272058888338506,
      "loss_iou": 0.333984375,
      "loss_num": 0.0234375,
      "loss_xval": 0.78515625,
      "num_input_tokens_seen": 79143720,
      "step": 1413
    },
    {
      "epoch": 3.1492204899777283,
      "grad_norm": 16.99781608581543,
      "learning_rate": 1e-06,
      "loss": 0.9043,
      "num_input_tokens_seen": 79200260,
      "step": 1414
    },
    {
      "epoch": 3.1492204899777283,
      "loss": 1.0250790119171143,
      "loss_ce": 0.00042086507892236114,
      "loss_iou": 0.421875,
      "loss_num": 0.035888671875,
      "loss_xval": 1.0234375,
      "num_input_tokens_seen": 79200260,
      "step": 1414
    },
    {
      "epoch": 3.151447661469933,
      "grad_norm": 16.20554542541504,
      "learning_rate": 1e-06,
      "loss": 1.0504,
      "num_input_tokens_seen": 79255792,
      "step": 1415
    },
    {
      "epoch": 3.151447661469933,
      "loss": 1.2998144626617432,
      "loss_ce": 0.0004980739904567599,
      "loss_iou": 0.53515625,
      "loss_num": 0.0458984375,
      "loss_xval": 1.296875,
      "num_input_tokens_seen": 79255792,
      "step": 1415
    },
    {
      "epoch": 3.153674832962138,
      "grad_norm": 25.340084075927734,
      "learning_rate": 1e-06,
      "loss": 1.1359,
      "num_input_tokens_seen": 79312352,
      "step": 1416
    },
    {
      "epoch": 3.153674832962138,
      "loss": 0.8494139313697815,
      "loss_ce": 0.0007811552495695651,
      "loss_iou": 0.353515625,
      "loss_num": 0.0279541015625,
      "loss_xval": 0.84765625,
      "num_input_tokens_seen": 79312352,
      "step": 1416
    },
    {
      "epoch": 3.155902004454343,
      "grad_norm": 11.992765426635742,
      "learning_rate": 1e-06,
      "loss": 0.8142,
      "num_input_tokens_seen": 79369780,
      "step": 1417
    },
    {
      "epoch": 3.155902004454343,
      "loss": 0.6079621911048889,
      "loss_ce": 0.00029617524705827236,
      "loss_iou": 0.2734375,
      "loss_num": 0.01220703125,
      "loss_xval": 0.609375,
      "num_input_tokens_seen": 79369780,
      "step": 1417
    },
    {
      "epoch": 3.1581291759465477,
      "grad_norm": 20.040773391723633,
      "learning_rate": 1e-06,
      "loss": 0.9713,
      "num_input_tokens_seen": 79424212,
      "step": 1418
    },
    {
      "epoch": 3.1581291759465477,
      "loss": 0.855168342590332,
      "loss_ce": 0.00030997302383184433,
      "loss_iou": 0.37109375,
      "loss_num": 0.022705078125,
      "loss_xval": 0.85546875,
      "num_input_tokens_seen": 79424212,
      "step": 1418
    },
    {
      "epoch": 3.1603563474387526,
      "grad_norm": 24.867382049560547,
      "learning_rate": 1e-06,
      "loss": 1.0417,
      "num_input_tokens_seen": 79481792,
      "step": 1419
    },
    {
      "epoch": 3.1603563474387526,
      "loss": 1.1068477630615234,
      "loss_ce": 0.00040248059667646885,
      "loss_iou": 0.4296875,
      "loss_num": 0.04931640625,
      "loss_xval": 1.109375,
      "num_input_tokens_seen": 79481792,
      "step": 1419
    },
    {
      "epoch": 3.1625835189309575,
      "grad_norm": 26.066076278686523,
      "learning_rate": 1e-06,
      "loss": 1.0346,
      "num_input_tokens_seen": 79538200,
      "step": 1420
    },
    {
      "epoch": 3.1625835189309575,
      "loss": 0.9904955625534058,
      "loss_ce": 0.0005052955239079893,
      "loss_iou": 0.392578125,
      "loss_num": 0.041015625,
      "loss_xval": 0.98828125,
      "num_input_tokens_seen": 79538200,
      "step": 1420
    },
    {
      "epoch": 3.1648106904231628,
      "grad_norm": 32.84996032714844,
      "learning_rate": 1e-06,
      "loss": 0.7832,
      "num_input_tokens_seen": 79594752,
      "step": 1421
    },
    {
      "epoch": 3.1648106904231628,
      "loss": 0.6035779714584351,
      "loss_ce": 0.0003064596385229379,
      "loss_iou": 0.244140625,
      "loss_num": 0.0230712890625,
      "loss_xval": 0.6015625,
      "num_input_tokens_seen": 79594752,
      "step": 1421
    },
    {
      "epoch": 3.1670378619153676,
      "grad_norm": 12.997925758361816,
      "learning_rate": 1e-06,
      "loss": 0.7086,
      "num_input_tokens_seen": 79649304,
      "step": 1422
    },
    {
      "epoch": 3.1670378619153676,
      "loss": 0.6079559326171875,
      "loss_ce": 0.0002899472019635141,
      "loss_iou": 0.2578125,
      "loss_num": 0.0186767578125,
      "loss_xval": 0.609375,
      "num_input_tokens_seen": 79649304,
      "step": 1422
    },
    {
      "epoch": 3.1692650334075725,
      "grad_norm": 14.003304481506348,
      "learning_rate": 1e-06,
      "loss": 0.9371,
      "num_input_tokens_seen": 79705016,
      "step": 1423
    },
    {
      "epoch": 3.1692650334075725,
      "loss": 0.9424291253089905,
      "loss_ce": 0.00029046309646219015,
      "loss_iou": 0.408203125,
      "loss_num": 0.0255126953125,
      "loss_xval": 0.94140625,
      "num_input_tokens_seen": 79705016,
      "step": 1423
    },
    {
      "epoch": 3.1714922048997773,
      "grad_norm": 15.485200881958008,
      "learning_rate": 1e-06,
      "loss": 0.7606,
      "num_input_tokens_seen": 79760764,
      "step": 1424
    },
    {
      "epoch": 3.1714922048997773,
      "loss": 0.867828369140625,
      "loss_ce": 0.0003967147204093635,
      "loss_iou": 0.375,
      "loss_num": 0.02294921875,
      "loss_xval": 0.8671875,
      "num_input_tokens_seen": 79760764,
      "step": 1424
    },
    {
      "epoch": 3.173719376391982,
      "grad_norm": 19.317201614379883,
      "learning_rate": 1e-06,
      "loss": 0.7716,
      "num_input_tokens_seen": 79816396,
      "step": 1425
    },
    {
      "epoch": 3.173719376391982,
      "loss": 0.697672963142395,
      "loss_ce": 0.0004073456802871078,
      "loss_iou": 0.291015625,
      "loss_num": 0.0234375,
      "loss_xval": 0.6953125,
      "num_input_tokens_seen": 79816396,
      "step": 1425
    },
    {
      "epoch": 3.175946547884187,
      "grad_norm": 16.760774612426758,
      "learning_rate": 1e-06,
      "loss": 0.8792,
      "num_input_tokens_seen": 79871868,
      "step": 1426
    },
    {
      "epoch": 3.175946547884187,
      "loss": 0.954727292060852,
      "loss_ce": 0.00038158154347911477,
      "loss_iou": 0.3828125,
      "loss_num": 0.0380859375,
      "loss_xval": 0.953125,
      "num_input_tokens_seen": 79871868,
      "step": 1426
    },
    {
      "epoch": 3.178173719376392,
      "grad_norm": 18.781694412231445,
      "learning_rate": 1e-06,
      "loss": 0.6773,
      "num_input_tokens_seen": 79928048,
      "step": 1427
    },
    {
      "epoch": 3.178173719376392,
      "loss": 0.6321383714675903,
      "loss_ce": 0.0003024227044079453,
      "loss_iou": 0.2734375,
      "loss_num": 0.0167236328125,
      "loss_xval": 0.6328125,
      "num_input_tokens_seen": 79928048,
      "step": 1427
    },
    {
      "epoch": 3.180400890868597,
      "grad_norm": 14.42310905456543,
      "learning_rate": 1e-06,
      "loss": 0.7926,
      "num_input_tokens_seen": 79983508,
      "step": 1428
    },
    {
      "epoch": 3.180400890868597,
      "loss": 0.8372380137443542,
      "loss_ce": 0.0003239199868403375,
      "loss_iou": 0.353515625,
      "loss_num": 0.02587890625,
      "loss_xval": 0.8359375,
      "num_input_tokens_seen": 79983508,
      "step": 1428
    },
    {
      "epoch": 3.1826280623608016,
      "grad_norm": 14.923147201538086,
      "learning_rate": 1e-06,
      "loss": 0.9376,
      "num_input_tokens_seen": 80042264,
      "step": 1429
    },
    {
      "epoch": 3.1826280623608016,
      "loss": 0.7183116674423218,
      "loss_ce": 0.00029408183763734996,
      "loss_iou": 0.294921875,
      "loss_num": 0.025146484375,
      "loss_xval": 0.71875,
      "num_input_tokens_seen": 80042264,
      "step": 1429
    },
    {
      "epoch": 3.1848552338530065,
      "grad_norm": 14.917871475219727,
      "learning_rate": 1e-06,
      "loss": 0.9401,
      "num_input_tokens_seen": 80098272,
      "step": 1430
    },
    {
      "epoch": 3.1848552338530065,
      "loss": 1.095562219619751,
      "loss_ce": 0.00034750672057271004,
      "loss_iou": 0.451171875,
      "loss_num": 0.038330078125,
      "loss_xval": 1.09375,
      "num_input_tokens_seen": 80098272,
      "step": 1430
    },
    {
      "epoch": 3.187082405345212,
      "grad_norm": 21.82032585144043,
      "learning_rate": 1e-06,
      "loss": 0.813,
      "num_input_tokens_seen": 80153352,
      "step": 1431
    },
    {
      "epoch": 3.187082405345212,
      "loss": 0.6784963607788086,
      "loss_ce": 0.00027368031442165375,
      "loss_iou": 0.2890625,
      "loss_num": 0.0196533203125,
      "loss_xval": 0.6796875,
      "num_input_tokens_seen": 80153352,
      "step": 1431
    },
    {
      "epoch": 3.1893095768374167,
      "grad_norm": 20.35378074645996,
      "learning_rate": 1e-06,
      "loss": 0.8377,
      "num_input_tokens_seen": 80209672,
      "step": 1432
    },
    {
      "epoch": 3.1893095768374167,
      "loss": 0.9991272687911987,
      "loss_ce": 0.0003479632141534239,
      "loss_iou": 0.419921875,
      "loss_num": 0.03173828125,
      "loss_xval": 1.0,
      "num_input_tokens_seen": 80209672,
      "step": 1432
    },
    {
      "epoch": 3.1915367483296215,
      "grad_norm": 18.134105682373047,
      "learning_rate": 1e-06,
      "loss": 0.7804,
      "num_input_tokens_seen": 80264404,
      "step": 1433
    },
    {
      "epoch": 3.1915367483296215,
      "loss": 0.8298586010932922,
      "loss_ce": 0.000268744770437479,
      "loss_iou": 0.349609375,
      "loss_num": 0.0257568359375,
      "loss_xval": 0.828125,
      "num_input_tokens_seen": 80264404,
      "step": 1433
    },
    {
      "epoch": 3.1937639198218264,
      "grad_norm": 25.565568923950195,
      "learning_rate": 1e-06,
      "loss": 0.7445,
      "num_input_tokens_seen": 80322156,
      "step": 1434
    },
    {
      "epoch": 3.1937639198218264,
      "loss": 0.8270922303199768,
      "loss_ce": 0.00043206167174503207,
      "loss_iou": 0.345703125,
      "loss_num": 0.0269775390625,
      "loss_xval": 0.828125,
      "num_input_tokens_seen": 80322156,
      "step": 1434
    },
    {
      "epoch": 3.1959910913140313,
      "grad_norm": 19.158845901489258,
      "learning_rate": 1e-06,
      "loss": 0.7464,
      "num_input_tokens_seen": 80378616,
      "step": 1435
    },
    {
      "epoch": 3.1959910913140313,
      "loss": 0.6876095533370972,
      "loss_ce": 0.0005977747496217489,
      "loss_iou": 0.26171875,
      "loss_num": 0.032470703125,
      "loss_xval": 0.6875,
      "num_input_tokens_seen": 80378616,
      "step": 1435
    },
    {
      "epoch": 3.198218262806236,
      "grad_norm": 12.419325828552246,
      "learning_rate": 1e-06,
      "loss": 0.5989,
      "num_input_tokens_seen": 80437060,
      "step": 1436
    },
    {
      "epoch": 3.198218262806236,
      "loss": 0.6812313199043274,
      "loss_ce": 0.00032312102848663926,
      "loss_iou": 0.287109375,
      "loss_num": 0.0213623046875,
      "loss_xval": 0.6796875,
      "num_input_tokens_seen": 80437060,
      "step": 1436
    },
    {
      "epoch": 3.200445434298441,
      "grad_norm": 27.797161102294922,
      "learning_rate": 1e-06,
      "loss": 0.8508,
      "num_input_tokens_seen": 80492916,
      "step": 1437
    },
    {
      "epoch": 3.200445434298441,
      "loss": 0.8674960732460022,
      "loss_ce": 0.00030855537625029683,
      "loss_iou": 0.34765625,
      "loss_num": 0.0341796875,
      "loss_xval": 0.8671875,
      "num_input_tokens_seen": 80492916,
      "step": 1437
    },
    {
      "epoch": 3.202672605790646,
      "grad_norm": 26.252103805541992,
      "learning_rate": 1e-06,
      "loss": 0.8936,
      "num_input_tokens_seen": 80550292,
      "step": 1438
    },
    {
      "epoch": 3.202672605790646,
      "loss": 0.885981559753418,
      "loss_ce": 0.00048353534657508135,
      "loss_iou": 0.384765625,
      "loss_num": 0.022705078125,
      "loss_xval": 0.88671875,
      "num_input_tokens_seen": 80550292,
      "step": 1438
    },
    {
      "epoch": 3.2048997772828507,
      "grad_norm": 18.954618453979492,
      "learning_rate": 1e-06,
      "loss": 1.1562,
      "num_input_tokens_seen": 80606748,
      "step": 1439
    },
    {
      "epoch": 3.2048997772828507,
      "loss": 0.9013663530349731,
      "loss_ce": 0.00048747207620181143,
      "loss_iou": 0.35546875,
      "loss_num": 0.038330078125,
      "loss_xval": 0.90234375,
      "num_input_tokens_seen": 80606748,
      "step": 1439
    },
    {
      "epoch": 3.2071269487750556,
      "grad_norm": 17.839231491088867,
      "learning_rate": 1e-06,
      "loss": 0.7546,
      "num_input_tokens_seen": 80665732,
      "step": 1440
    },
    {
      "epoch": 3.2071269487750556,
      "loss": 0.7411551475524902,
      "loss_ce": 0.0004324775072745979,
      "loss_iou": 0.3203125,
      "loss_num": 0.0198974609375,
      "loss_xval": 0.7421875,
      "num_input_tokens_seen": 80665732,
      "step": 1440
    },
    {
      "epoch": 3.2093541202672604,
      "grad_norm": 16.883333206176758,
      "learning_rate": 1e-06,
      "loss": 0.8151,
      "num_input_tokens_seen": 80720260,
      "step": 1441
    },
    {
      "epoch": 3.2093541202672604,
      "loss": 1.001312255859375,
      "loss_ce": 0.00033568451181054115,
      "loss_iou": 0.40234375,
      "loss_num": 0.039794921875,
      "loss_xval": 1.0,
      "num_input_tokens_seen": 80720260,
      "step": 1441
    },
    {
      "epoch": 3.2115812917594653,
      "grad_norm": 19.242998123168945,
      "learning_rate": 1e-06,
      "loss": 0.8271,
      "num_input_tokens_seen": 80771032,
      "step": 1442
    },
    {
      "epoch": 3.2115812917594653,
      "loss": 0.8663139343261719,
      "loss_ce": 0.0005913155619055033,
      "loss_iou": 0.345703125,
      "loss_num": 0.034912109375,
      "loss_xval": 0.8671875,
      "num_input_tokens_seen": 80771032,
      "step": 1442
    },
    {
      "epoch": 3.2138084632516706,
      "grad_norm": 18.97730827331543,
      "learning_rate": 1e-06,
      "loss": 0.7771,
      "num_input_tokens_seen": 80823748,
      "step": 1443
    },
    {
      "epoch": 3.2138084632516706,
      "loss": 0.7416298389434814,
      "loss_ce": 0.0002968141343444586,
      "loss_iou": 0.314453125,
      "loss_num": 0.0224609375,
      "loss_xval": 0.7421875,
      "num_input_tokens_seen": 80823748,
      "step": 1443
    },
    {
      "epoch": 3.2160356347438754,
      "grad_norm": 71.22569274902344,
      "learning_rate": 1e-06,
      "loss": 0.966,
      "num_input_tokens_seen": 80879232,
      "step": 1444
    },
    {
      "epoch": 3.2160356347438754,
      "loss": 0.8303946852684021,
      "loss_ce": 0.000316560675855726,
      "loss_iou": 0.345703125,
      "loss_num": 0.02783203125,
      "loss_xval": 0.828125,
      "num_input_tokens_seen": 80879232,
      "step": 1444
    },
    {
      "epoch": 3.2182628062360803,
      "grad_norm": 16.554031372070312,
      "learning_rate": 1e-06,
      "loss": 0.772,
      "num_input_tokens_seen": 80936056,
      "step": 1445
    },
    {
      "epoch": 3.2182628062360803,
      "loss": 0.7341041564941406,
      "loss_ce": 0.0004616554651875049,
      "loss_iou": 0.294921875,
      "loss_num": 0.0286865234375,
      "loss_xval": 0.734375,
      "num_input_tokens_seen": 80936056,
      "step": 1445
    },
    {
      "epoch": 3.220489977728285,
      "grad_norm": 12.350296020507812,
      "learning_rate": 1e-06,
      "loss": 0.983,
      "num_input_tokens_seen": 80992044,
      "step": 1446
    },
    {
      "epoch": 3.220489977728285,
      "loss": 1.119089126586914,
      "loss_ce": 0.00043676019413396716,
      "loss_iou": 0.4375,
      "loss_num": 0.048583984375,
      "loss_xval": 1.1171875,
      "num_input_tokens_seen": 80992044,
      "step": 1446
    },
    {
      "epoch": 3.22271714922049,
      "grad_norm": 35.66438293457031,
      "learning_rate": 1e-06,
      "loss": 0.9671,
      "num_input_tokens_seen": 81047568,
      "step": 1447
    },
    {
      "epoch": 3.22271714922049,
      "loss": 1.1087019443511963,
      "loss_ce": 0.000303465174511075,
      "loss_iou": 0.46875,
      "loss_num": 0.0341796875,
      "loss_xval": 1.109375,
      "num_input_tokens_seen": 81047568,
      "step": 1447
    },
    {
      "epoch": 3.224944320712695,
      "grad_norm": 23.309444427490234,
      "learning_rate": 1e-06,
      "loss": 1.0038,
      "num_input_tokens_seen": 81105220,
      "step": 1448
    },
    {
      "epoch": 3.224944320712695,
      "loss": 0.9000189304351807,
      "loss_ce": 0.0008490003529004753,
      "loss_iou": 0.37109375,
      "loss_num": 0.031494140625,
      "loss_xval": 0.8984375,
      "num_input_tokens_seen": 81105220,
      "step": 1448
    },
    {
      "epoch": 3.2271714922048997,
      "grad_norm": 20.23189926147461,
      "learning_rate": 1e-06,
      "loss": 0.7571,
      "num_input_tokens_seen": 81157732,
      "step": 1449
    },
    {
      "epoch": 3.2271714922048997,
      "loss": 0.5965100526809692,
      "loss_ce": 0.00031864526681602,
      "loss_iou": 0.2412109375,
      "loss_num": 0.022705078125,
      "loss_xval": 0.59765625,
      "num_input_tokens_seen": 81157732,
      "step": 1449
    },
    {
      "epoch": 3.2293986636971046,
      "grad_norm": 20.063142776489258,
      "learning_rate": 1e-06,
      "loss": 0.7535,
      "num_input_tokens_seen": 81215292,
      "step": 1450
    },
    {
      "epoch": 3.2293986636971046,
      "loss": 0.9407631158828735,
      "loss_ce": 0.00033346362761221826,
      "loss_iou": 0.40625,
      "loss_num": 0.0255126953125,
      "loss_xval": 0.94140625,
      "num_input_tokens_seen": 81215292,
      "step": 1450
    },
    {
      "epoch": 3.2316258351893095,
      "grad_norm": 19.31182289123535,
      "learning_rate": 1e-06,
      "loss": 0.817,
      "num_input_tokens_seen": 81272552,
      "step": 1451
    },
    {
      "epoch": 3.2316258351893095,
      "loss": 1.0058592557907104,
      "loss_ce": 0.0004881612549070269,
      "loss_iou": 0.376953125,
      "loss_num": 0.050048828125,
      "loss_xval": 1.0078125,
      "num_input_tokens_seen": 81272552,
      "step": 1451
    },
    {
      "epoch": 3.2338530066815143,
      "grad_norm": 34.98418426513672,
      "learning_rate": 1e-06,
      "loss": 0.9567,
      "num_input_tokens_seen": 81322452,
      "step": 1452
    },
    {
      "epoch": 3.2338530066815143,
      "loss": 0.8940277099609375,
      "loss_ce": 0.000717109243851155,
      "loss_iou": 0.33203125,
      "loss_num": 0.04541015625,
      "loss_xval": 0.89453125,
      "num_input_tokens_seen": 81322452,
      "step": 1452
    },
    {
      "epoch": 3.236080178173719,
      "grad_norm": 22.681028366088867,
      "learning_rate": 1e-06,
      "loss": 1.0263,
      "num_input_tokens_seen": 81378452,
      "step": 1453
    },
    {
      "epoch": 3.236080178173719,
      "loss": 1.1674110889434814,
      "loss_ce": 0.0006629737326875329,
      "loss_iou": 0.46484375,
      "loss_num": 0.047607421875,
      "loss_xval": 1.1640625,
      "num_input_tokens_seen": 81378452,
      "step": 1453
    },
    {
      "epoch": 3.2383073496659245,
      "grad_norm": 15.543540954589844,
      "learning_rate": 1e-06,
      "loss": 0.7653,
      "num_input_tokens_seen": 81432916,
      "step": 1454
    },
    {
      "epoch": 3.2383073496659245,
      "loss": 0.5583458542823792,
      "loss_ce": 0.0002403960534138605,
      "loss_iou": 0.2392578125,
      "loss_num": 0.015869140625,
      "loss_xval": 0.55859375,
      "num_input_tokens_seen": 81432916,
      "step": 1454
    },
    {
      "epoch": 3.2405345211581293,
      "grad_norm": 35.686893463134766,
      "learning_rate": 1e-06,
      "loss": 1.0985,
      "num_input_tokens_seen": 81488412,
      "step": 1455
    },
    {
      "epoch": 3.2405345211581293,
      "loss": 1.1553540229797363,
      "loss_ce": 0.0003246957203373313,
      "loss_iou": 0.490234375,
      "loss_num": 0.034423828125,
      "loss_xval": 1.15625,
      "num_input_tokens_seen": 81488412,
      "step": 1455
    },
    {
      "epoch": 3.242761692650334,
      "grad_norm": 16.909992218017578,
      "learning_rate": 1e-06,
      "loss": 0.962,
      "num_input_tokens_seen": 81543048,
      "step": 1456
    },
    {
      "epoch": 3.242761692650334,
      "loss": 1.0606622695922852,
      "loss_ce": 0.00035947078140452504,
      "loss_iou": 0.41796875,
      "loss_num": 0.045654296875,
      "loss_xval": 1.0625,
      "num_input_tokens_seen": 81543048,
      "step": 1456
    },
    {
      "epoch": 3.244988864142539,
      "grad_norm": 17.404476165771484,
      "learning_rate": 1e-06,
      "loss": 0.5719,
      "num_input_tokens_seen": 81599572,
      "step": 1457
    },
    {
      "epoch": 3.244988864142539,
      "loss": 0.43446803092956543,
      "loss_ce": 0.00026391935534775257,
      "loss_iou": 0.1845703125,
      "loss_num": 0.01312255859375,
      "loss_xval": 0.43359375,
      "num_input_tokens_seen": 81599572,
      "step": 1457
    },
    {
      "epoch": 3.247216035634744,
      "grad_norm": 15.075811386108398,
      "learning_rate": 1e-06,
      "loss": 0.8541,
      "num_input_tokens_seen": 81658816,
      "step": 1458
    },
    {
      "epoch": 3.247216035634744,
      "loss": 0.969578206539154,
      "loss_ce": 0.0003399454872123897,
      "loss_iou": 0.400390625,
      "loss_num": 0.0341796875,
      "loss_xval": 0.96875,
      "num_input_tokens_seen": 81658816,
      "step": 1458
    },
    {
      "epoch": 3.249443207126949,
      "grad_norm": 189.14511108398438,
      "learning_rate": 1e-06,
      "loss": 0.767,
      "num_input_tokens_seen": 81714620,
      "step": 1459
    },
    {
      "epoch": 3.249443207126949,
      "loss": 0.8307961225509644,
      "loss_ce": 0.00022976743639446795,
      "loss_iou": 0.361328125,
      "loss_num": 0.0213623046875,
      "loss_xval": 0.83203125,
      "num_input_tokens_seen": 81714620,
      "step": 1459
    },
    {
      "epoch": 3.2516703786191536,
      "grad_norm": 16.0401611328125,
      "learning_rate": 1e-06,
      "loss": 0.6961,
      "num_input_tokens_seen": 81768192,
      "step": 1460
    },
    {
      "epoch": 3.2516703786191536,
      "loss": 0.4611513614654541,
      "loss_ce": 0.00027491431683301926,
      "loss_iou": 0.1875,
      "loss_num": 0.0172119140625,
      "loss_xval": 0.4609375,
      "num_input_tokens_seen": 81768192,
      "step": 1460
    },
    {
      "epoch": 3.2538975501113585,
      "grad_norm": 24.038867950439453,
      "learning_rate": 1e-06,
      "loss": 0.945,
      "num_input_tokens_seen": 81823076,
      "step": 1461
    },
    {
      "epoch": 3.2538975501113585,
      "loss": 1.0662281513214111,
      "loss_ce": 0.00043232861207798123,
      "loss_iou": 0.453125,
      "loss_num": 0.031494140625,
      "loss_xval": 1.0625,
      "num_input_tokens_seen": 81823076,
      "step": 1461
    },
    {
      "epoch": 3.2561247216035634,
      "grad_norm": 27.399744033813477,
      "learning_rate": 1e-06,
      "loss": 0.8288,
      "num_input_tokens_seen": 81878532,
      "step": 1462
    },
    {
      "epoch": 3.2561247216035634,
      "loss": 1.0180044174194336,
      "loss_ce": 0.00042639480670914054,
      "loss_iou": 0.4453125,
      "loss_num": 0.0252685546875,
      "loss_xval": 1.015625,
      "num_input_tokens_seen": 81878532,
      "step": 1462
    },
    {
      "epoch": 3.2583518930957682,
      "grad_norm": 30.450820922851562,
      "learning_rate": 1e-06,
      "loss": 0.9573,
      "num_input_tokens_seen": 81934472,
      "step": 1463
    },
    {
      "epoch": 3.2583518930957682,
      "loss": 0.7863764762878418,
      "loss_ce": 0.00024365229182876647,
      "loss_iou": 0.310546875,
      "loss_num": 0.03271484375,
      "loss_xval": 0.78515625,
      "num_input_tokens_seen": 81934472,
      "step": 1463
    },
    {
      "epoch": 3.260579064587973,
      "grad_norm": 73.2469482421875,
      "learning_rate": 1e-06,
      "loss": 0.8654,
      "num_input_tokens_seen": 81992352,
      "step": 1464
    },
    {
      "epoch": 3.260579064587973,
      "loss": 0.8775367736816406,
      "loss_ce": 0.0003394762461539358,
      "loss_iou": 0.373046875,
      "loss_num": 0.0264892578125,
      "loss_xval": 0.87890625,
      "num_input_tokens_seen": 81992352,
      "step": 1464
    },
    {
      "epoch": 3.262806236080178,
      "grad_norm": 17.928956985473633,
      "learning_rate": 1e-06,
      "loss": 0.9042,
      "num_input_tokens_seen": 82047880,
      "step": 1465
    },
    {
      "epoch": 3.262806236080178,
      "loss": 0.8798074722290039,
      "loss_ce": 0.009446126408874989,
      "loss_iou": 0.37109375,
      "loss_num": 0.025390625,
      "loss_xval": 0.87109375,
      "num_input_tokens_seen": 82047880,
      "step": 1465
    },
    {
      "epoch": 3.2650334075723833,
      "grad_norm": 13.953981399536133,
      "learning_rate": 1e-06,
      "loss": 0.7979,
      "num_input_tokens_seen": 82105036,
      "step": 1466
    },
    {
      "epoch": 3.2650334075723833,
      "loss": 0.9124147891998291,
      "loss_ce": 0.0003053955442737788,
      "loss_iou": 0.408203125,
      "loss_num": 0.01953125,
      "loss_xval": 0.9140625,
      "num_input_tokens_seen": 82105036,
      "step": 1466
    },
    {
      "epoch": 3.267260579064588,
      "grad_norm": 19.903533935546875,
      "learning_rate": 1e-06,
      "loss": 0.7328,
      "num_input_tokens_seen": 82161524,
      "step": 1467
    },
    {
      "epoch": 3.267260579064588,
      "loss": 0.7695513963699341,
      "loss_ce": 0.0002642880426719785,
      "loss_iou": 0.328125,
      "loss_num": 0.0225830078125,
      "loss_xval": 0.76953125,
      "num_input_tokens_seen": 82161524,
      "step": 1467
    },
    {
      "epoch": 3.269487750556793,
      "grad_norm": 16.431427001953125,
      "learning_rate": 1e-06,
      "loss": 0.7701,
      "num_input_tokens_seen": 82216912,
      "step": 1468
    },
    {
      "epoch": 3.269487750556793,
      "loss": 0.7241719365119934,
      "loss_ce": 0.00029497878858819604,
      "loss_iou": 0.3125,
      "loss_num": 0.0194091796875,
      "loss_xval": 0.72265625,
      "num_input_tokens_seen": 82216912,
      "step": 1468
    },
    {
      "epoch": 3.271714922048998,
      "grad_norm": 21.81812286376953,
      "learning_rate": 1e-06,
      "loss": 0.7118,
      "num_input_tokens_seen": 82273656,
      "step": 1469
    },
    {
      "epoch": 3.271714922048998,
      "loss": 0.6574879884719849,
      "loss_ce": 0.0002613977121654898,
      "loss_iou": 0.28125,
      "loss_num": 0.0191650390625,
      "loss_xval": 0.65625,
      "num_input_tokens_seen": 82273656,
      "step": 1469
    },
    {
      "epoch": 3.2739420935412027,
      "grad_norm": 16.43152618408203,
      "learning_rate": 1e-06,
      "loss": 0.8491,
      "num_input_tokens_seen": 82329848,
      "step": 1470
    },
    {
      "epoch": 3.2739420935412027,
      "loss": 0.6564966440200806,
      "loss_ce": 0.0002466381702106446,
      "loss_iou": 0.26171875,
      "loss_num": 0.0263671875,
      "loss_xval": 0.65625,
      "num_input_tokens_seen": 82329848,
      "step": 1470
    },
    {
      "epoch": 3.2761692650334076,
      "grad_norm": 26.318092346191406,
      "learning_rate": 1e-06,
      "loss": 0.7104,
      "num_input_tokens_seen": 82388348,
      "step": 1471
    },
    {
      "epoch": 3.2761692650334076,
      "loss": 0.7334426045417786,
      "loss_ce": 0.00028831767849624157,
      "loss_iou": 0.283203125,
      "loss_num": 0.033447265625,
      "loss_xval": 0.734375,
      "num_input_tokens_seen": 82388348,
      "step": 1471
    },
    {
      "epoch": 3.2783964365256124,
      "grad_norm": 15.957210540771484,
      "learning_rate": 1e-06,
      "loss": 0.5911,
      "num_input_tokens_seen": 82447224,
      "step": 1472
    },
    {
      "epoch": 3.2783964365256124,
      "loss": 0.5700989961624146,
      "loss_ce": 0.0002747434191405773,
      "loss_iou": 0.2451171875,
      "loss_num": 0.0159912109375,
      "loss_xval": 0.5703125,
      "num_input_tokens_seen": 82447224,
      "step": 1472
    },
    {
      "epoch": 3.2806236080178173,
      "grad_norm": 21.27912712097168,
      "learning_rate": 1e-06,
      "loss": 0.9417,
      "num_input_tokens_seen": 82503100,
      "step": 1473
    },
    {
      "epoch": 3.2806236080178173,
      "loss": 1.111638069152832,
      "loss_ce": 0.0003100019530393183,
      "loss_iou": 0.490234375,
      "loss_num": 0.0257568359375,
      "loss_xval": 1.109375,
      "num_input_tokens_seen": 82503100,
      "step": 1473
    },
    {
      "epoch": 3.282850779510022,
      "grad_norm": 36.66477584838867,
      "learning_rate": 1e-06,
      "loss": 0.8626,
      "num_input_tokens_seen": 82560744,
      "step": 1474
    },
    {
      "epoch": 3.282850779510022,
      "loss": 0.839104413986206,
      "loss_ce": 0.00048139755381271243,
      "loss_iou": 0.359375,
      "loss_num": 0.0244140625,
      "loss_xval": 0.83984375,
      "num_input_tokens_seen": 82560744,
      "step": 1474
    },
    {
      "epoch": 3.285077951002227,
      "grad_norm": 20.039661407470703,
      "learning_rate": 1e-06,
      "loss": 0.8975,
      "num_input_tokens_seen": 82617300,
      "step": 1475
    },
    {
      "epoch": 3.285077951002227,
      "loss": 1.0146886110305786,
      "loss_ce": 0.0002843354013748467,
      "loss_iou": 0.439453125,
      "loss_num": 0.0274658203125,
      "loss_xval": 1.015625,
      "num_input_tokens_seen": 82617300,
      "step": 1475
    },
    {
      "epoch": 3.2873051224944323,
      "grad_norm": 15.592321395874023,
      "learning_rate": 1e-06,
      "loss": 0.93,
      "num_input_tokens_seen": 82670780,
      "step": 1476
    },
    {
      "epoch": 3.2873051224944323,
      "loss": 0.9146976470947266,
      "loss_ce": 0.0008792462758719921,
      "loss_iou": 0.373046875,
      "loss_num": 0.03369140625,
      "loss_xval": 0.9140625,
      "num_input_tokens_seen": 82670780,
      "step": 1476
    },
    {
      "epoch": 3.289532293986637,
      "grad_norm": 15.493447303771973,
      "learning_rate": 1e-06,
      "loss": 0.7573,
      "num_input_tokens_seen": 82729732,
      "step": 1477
    },
    {
      "epoch": 3.289532293986637,
      "loss": 0.6953732371330261,
      "loss_ce": 0.0003048654180020094,
      "loss_iou": 0.291015625,
      "loss_num": 0.022705078125,
      "loss_xval": 0.6953125,
      "num_input_tokens_seen": 82729732,
      "step": 1477
    },
    {
      "epoch": 3.291759465478842,
      "grad_norm": 26.018573760986328,
      "learning_rate": 1e-06,
      "loss": 0.8349,
      "num_input_tokens_seen": 82787736,
      "step": 1478
    },
    {
      "epoch": 3.291759465478842,
      "loss": 0.6441195607185364,
      "loss_ce": 0.0003207349800504744,
      "loss_iou": 0.28515625,
      "loss_num": 0.01446533203125,
      "loss_xval": 0.64453125,
      "num_input_tokens_seen": 82787736,
      "step": 1478
    },
    {
      "epoch": 3.293986636971047,
      "grad_norm": 18.60807991027832,
      "learning_rate": 1e-06,
      "loss": 0.9953,
      "num_input_tokens_seen": 82843952,
      "step": 1479
    },
    {
      "epoch": 3.293986636971047,
      "loss": 0.9405885934829712,
      "loss_ce": 0.0004030146519653499,
      "loss_iou": 0.375,
      "loss_num": 0.038330078125,
      "loss_xval": 0.94140625,
      "num_input_tokens_seen": 82843952,
      "step": 1479
    },
    {
      "epoch": 3.2962138084632517,
      "grad_norm": 71.17398834228516,
      "learning_rate": 1e-06,
      "loss": 0.9742,
      "num_input_tokens_seen": 82901868,
      "step": 1480
    },
    {
      "epoch": 3.2962138084632517,
      "loss": 1.149244785308838,
      "loss_ce": 0.00031894395942799747,
      "loss_iou": 0.50390625,
      "loss_num": 0.0283203125,
      "loss_xval": 1.1484375,
      "num_input_tokens_seen": 82901868,
      "step": 1480
    },
    {
      "epoch": 3.2984409799554566,
      "grad_norm": 53.66946792602539,
      "learning_rate": 1e-06,
      "loss": 0.941,
      "num_input_tokens_seen": 82955420,
      "step": 1481
    },
    {
      "epoch": 3.2984409799554566,
      "loss": 1.0035812854766846,
      "loss_ce": 0.004924064036458731,
      "loss_iou": 0.39453125,
      "loss_num": 0.042236328125,
      "loss_xval": 1.0,
      "num_input_tokens_seen": 82955420,
      "step": 1481
    },
    {
      "epoch": 3.3006681514476615,
      "grad_norm": 26.24509048461914,
      "learning_rate": 1e-06,
      "loss": 0.9412,
      "num_input_tokens_seen": 83009936,
      "step": 1482
    },
    {
      "epoch": 3.3006681514476615,
      "loss": 1.0405406951904297,
      "loss_ce": 0.0006236102781258523,
      "loss_iou": 0.4296875,
      "loss_num": 0.036376953125,
      "loss_xval": 1.0390625,
      "num_input_tokens_seen": 83009936,
      "step": 1482
    },
    {
      "epoch": 3.3028953229398663,
      "grad_norm": 14.86413860321045,
      "learning_rate": 1e-06,
      "loss": 0.8586,
      "num_input_tokens_seen": 83067024,
      "step": 1483
    },
    {
      "epoch": 3.3028953229398663,
      "loss": 0.9363436102867126,
      "loss_ce": 0.00030847761081531644,
      "loss_iou": 0.404296875,
      "loss_num": 0.0255126953125,
      "loss_xval": 0.9375,
      "num_input_tokens_seen": 83067024,
      "step": 1483
    },
    {
      "epoch": 3.305122494432071,
      "grad_norm": 16.618406295776367,
      "learning_rate": 1e-06,
      "loss": 0.8834,
      "num_input_tokens_seen": 83122648,
      "step": 1484
    },
    {
      "epoch": 3.305122494432071,
      "loss": 1.071757197380066,
      "loss_ce": 0.0004681579302996397,
      "loss_iou": 0.435546875,
      "loss_num": 0.039794921875,
      "loss_xval": 1.0703125,
      "num_input_tokens_seen": 83122648,
      "step": 1484
    },
    {
      "epoch": 3.307349665924276,
      "grad_norm": 23.63321876525879,
      "learning_rate": 1e-06,
      "loss": 1.0184,
      "num_input_tokens_seen": 83176484,
      "step": 1485
    },
    {
      "epoch": 3.307349665924276,
      "loss": 0.7948580980300903,
      "loss_ce": 0.000424501282395795,
      "loss_iou": 0.328125,
      "loss_num": 0.0277099609375,
      "loss_xval": 0.79296875,
      "num_input_tokens_seen": 83176484,
      "step": 1485
    },
    {
      "epoch": 3.309576837416481,
      "grad_norm": 18.189393997192383,
      "learning_rate": 1e-06,
      "loss": 0.8297,
      "num_input_tokens_seen": 83233140,
      "step": 1486
    },
    {
      "epoch": 3.309576837416481,
      "loss": 0.7806228399276733,
      "loss_ce": 0.0003493690746836364,
      "loss_iou": 0.31640625,
      "loss_num": 0.02978515625,
      "loss_xval": 0.78125,
      "num_input_tokens_seen": 83233140,
      "step": 1486
    },
    {
      "epoch": 3.3118040089086858,
      "grad_norm": 21.735517501831055,
      "learning_rate": 1e-06,
      "loss": 0.8333,
      "num_input_tokens_seen": 83288588,
      "step": 1487
    },
    {
      "epoch": 3.3118040089086858,
      "loss": 0.6785666942596436,
      "loss_ce": 0.0005881287506781518,
      "loss_iou": 0.287109375,
      "loss_num": 0.0203857421875,
      "loss_xval": 0.6796875,
      "num_input_tokens_seen": 83288588,
      "step": 1487
    },
    {
      "epoch": 3.3140311804008906,
      "grad_norm": 239.3049774169922,
      "learning_rate": 1e-06,
      "loss": 0.896,
      "num_input_tokens_seen": 83343812,
      "step": 1488
    },
    {
      "epoch": 3.3140311804008906,
      "loss": 0.8566187024116516,
      "loss_ce": 0.00041753414552658796,
      "loss_iou": 0.349609375,
      "loss_num": 0.031494140625,
      "loss_xval": 0.85546875,
      "num_input_tokens_seen": 83343812,
      "step": 1488
    },
    {
      "epoch": 3.316258351893096,
      "grad_norm": 29.564838409423828,
      "learning_rate": 1e-06,
      "loss": 0.9014,
      "num_input_tokens_seen": 83399608,
      "step": 1489
    },
    {
      "epoch": 3.316258351893096,
      "loss": 0.8447275161743164,
      "loss_ce": 0.0004892901633866131,
      "loss_iou": 0.294921875,
      "loss_num": 0.05126953125,
      "loss_xval": 0.84375,
      "num_input_tokens_seen": 83399608,
      "step": 1489
    },
    {
      "epoch": 3.318485523385301,
      "grad_norm": 27.595678329467773,
      "learning_rate": 1e-06,
      "loss": 0.918,
      "num_input_tokens_seen": 83455260,
      "step": 1490
    },
    {
      "epoch": 3.318485523385301,
      "loss": 0.9408430457115173,
      "loss_ce": 0.00041330509702675045,
      "loss_iou": 0.37890625,
      "loss_num": 0.036865234375,
      "loss_xval": 0.94140625,
      "num_input_tokens_seen": 83455260,
      "step": 1490
    },
    {
      "epoch": 3.3207126948775056,
      "grad_norm": 12.939767837524414,
      "learning_rate": 1e-06,
      "loss": 0.7262,
      "num_input_tokens_seen": 83512012,
      "step": 1491
    },
    {
      "epoch": 3.3207126948775056,
      "loss": 0.773056149482727,
      "loss_ce": 0.00035112208570353687,
      "loss_iou": 0.298828125,
      "loss_num": 0.034912109375,
      "loss_xval": 0.7734375,
      "num_input_tokens_seen": 83512012,
      "step": 1491
    },
    {
      "epoch": 3.3229398663697105,
      "grad_norm": 17.990427017211914,
      "learning_rate": 1e-06,
      "loss": 0.9087,
      "num_input_tokens_seen": 83566724,
      "step": 1492
    },
    {
      "epoch": 3.3229398663697105,
      "loss": 0.7485735416412354,
      "loss_ce": 0.00028249542810954154,
      "loss_iou": 0.28515625,
      "loss_num": 0.035400390625,
      "loss_xval": 0.75,
      "num_input_tokens_seen": 83566724,
      "step": 1492
    },
    {
      "epoch": 3.3251670378619154,
      "grad_norm": 22.296096801757812,
      "learning_rate": 1e-06,
      "loss": 0.8386,
      "num_input_tokens_seen": 83623084,
      "step": 1493
    },
    {
      "epoch": 3.3251670378619154,
      "loss": 0.6770757436752319,
      "loss_ce": 0.0003179589402861893,
      "loss_iou": 0.283203125,
      "loss_num": 0.0223388671875,
      "loss_xval": 0.67578125,
      "num_input_tokens_seen": 83623084,
      "step": 1493
    },
    {
      "epoch": 3.3273942093541202,
      "grad_norm": 23.940916061401367,
      "learning_rate": 1e-06,
      "loss": 0.8714,
      "num_input_tokens_seen": 83679600,
      "step": 1494
    },
    {
      "epoch": 3.3273942093541202,
      "loss": 0.8882294297218323,
      "loss_ce": 0.0002899998507928103,
      "loss_iou": 0.396484375,
      "loss_num": 0.018798828125,
      "loss_xval": 0.88671875,
      "num_input_tokens_seen": 83679600,
      "step": 1494
    },
    {
      "epoch": 3.329621380846325,
      "grad_norm": 27.720958709716797,
      "learning_rate": 1e-06,
      "loss": 0.7311,
      "num_input_tokens_seen": 83736204,
      "step": 1495
    },
    {
      "epoch": 3.329621380846325,
      "loss": 0.7086691856384277,
      "loss_ce": 0.00041724531911313534,
      "loss_iou": 0.302734375,
      "loss_num": 0.0206298828125,
      "loss_xval": 0.70703125,
      "num_input_tokens_seen": 83736204,
      "step": 1495
    },
    {
      "epoch": 3.33184855233853,
      "grad_norm": 23.420194625854492,
      "learning_rate": 1e-06,
      "loss": 0.8413,
      "num_input_tokens_seen": 83792900,
      "step": 1496
    },
    {
      "epoch": 3.33184855233853,
      "loss": 0.7663967609405518,
      "loss_ce": 0.00028348196065053344,
      "loss_iou": 0.328125,
      "loss_num": 0.021728515625,
      "loss_xval": 0.765625,
      "num_input_tokens_seen": 83792900,
      "step": 1496
    },
    {
      "epoch": 3.334075723830735,
      "grad_norm": 35.49727249145508,
      "learning_rate": 1e-06,
      "loss": 0.8009,
      "num_input_tokens_seen": 83847828,
      "step": 1497
    },
    {
      "epoch": 3.334075723830735,
      "loss": 0.8171854615211487,
      "loss_ce": 0.0002908838796429336,
      "loss_iou": 0.34765625,
      "loss_num": 0.024169921875,
      "loss_xval": 0.81640625,
      "num_input_tokens_seen": 83847828,
      "step": 1497
    },
    {
      "epoch": 3.33630289532294,
      "grad_norm": 20.75282859802246,
      "learning_rate": 1e-06,
      "loss": 0.9144,
      "num_input_tokens_seen": 83903316,
      "step": 1498
    },
    {
      "epoch": 3.33630289532294,
      "loss": 1.1358669996261597,
      "loss_ce": 0.00036891031777486205,
      "loss_iou": 0.431640625,
      "loss_num": 0.05517578125,
      "loss_xval": 1.1328125,
      "num_input_tokens_seen": 83903316,
      "step": 1498
    },
    {
      "epoch": 3.338530066815145,
      "grad_norm": 28.5906925201416,
      "learning_rate": 1e-06,
      "loss": 0.8748,
      "num_input_tokens_seen": 83958736,
      "step": 1499
    },
    {
      "epoch": 3.338530066815145,
      "loss": 1.1274135112762451,
      "loss_ce": 0.00046035420382395387,
      "loss_iou": 0.4921875,
      "loss_num": 0.028564453125,
      "loss_xval": 1.125,
      "num_input_tokens_seen": 83958736,
      "step": 1499
    },
    {
      "epoch": 3.34075723830735,
      "grad_norm": 23.704599380493164,
      "learning_rate": 1e-06,
      "loss": 0.8677,
      "num_input_tokens_seen": 84012288,
      "step": 1500
    },
    {
      "epoch": 3.34075723830735,
      "eval_seeclick_web_CIoU": 0.5594667792320251,
      "eval_seeclick_web_GIoU": 0.5541926324367523,
      "eval_seeclick_web_IoU": 0.5765488147735596,
      "eval_seeclick_web_MAE_all": 0.017576972022652626,
      "eval_seeclick_web_MAE_h": 0.010464820079505444,
      "eval_seeclick_web_MAE_w": 0.018719857558608055,
      "eval_seeclick_web_MAE_x_boxes": 0.00861422996968031,
      "eval_seeclick_web_MAE_y_boxes": 0.022560626734048128,
      "eval_seeclick_web_inside_bbox": 0.9166666567325592,
      "eval_seeclick_web_loss": 0.9647981524467468,
      "eval_seeclick_web_loss_ce": 0.0004042425425723195,
      "eval_seeclick_web_loss_iou": 0.4376220703125,
      "eval_seeclick_web_loss_num": 0.013916015625,
      "eval_seeclick_web_loss_xval": 0.94482421875,
      "eval_seeclick_web_runtime": 33.8762,
      "eval_seeclick_web_samples_per_second": 1.476,
      "eval_seeclick_web_steps_per_second": 0.059,
      "num_input_tokens_seen": 84012288,
      "step": 1500
    },
    {
      "epoch": 3.34075723830735,
      "eval_icons_CIoU": 0.3017665892839432,
      "eval_icons_GIoU": 0.3255244195461273,
      "eval_icons_IoU": 0.37552230060100555,
      "eval_icons_MAE_all": 0.06774608045816422,
      "eval_icons_MAE_h": 0.03980293497443199,
      "eval_icons_MAE_w": 0.08278198912739754,
      "eval_icons_MAE_x_boxes": 0.05543145537376404,
      "eval_icons_MAE_y_boxes": 0.038841537199914455,
      "eval_icons_inside_bbox": 0.6336805522441864,
      "eval_icons_loss": 1.737123966217041,
      "eval_icons_loss_ce": 0.0005422734539024532,
      "eval_icons_loss_iou": 0.670654296875,
      "eval_icons_loss_num": 0.05875968933105469,
      "eval_icons_loss_xval": 1.63427734375,
      "eval_icons_runtime": 35.2512,
      "eval_icons_samples_per_second": 1.418,
      "eval_icons_steps_per_second": 0.057,
      "num_input_tokens_seen": 84012288,
      "step": 1500
    },
    {
      "epoch": 3.34075723830735,
      "eval_screenspot_CIoU": 0.31236544251441956,
      "eval_screenspot_GIoU": 0.3288701077302297,
      "eval_screenspot_IoU": 0.3995972275733948,
      "eval_screenspot_MAE_all": 0.0782718521853288,
      "eval_screenspot_MAE_h": 0.04242397534350554,
      "eval_screenspot_MAE_w": 0.088284266491731,
      "eval_screenspot_MAE_x_boxes": 0.10724017272392909,
      "eval_screenspot_MAE_y_boxes": 0.047414361188809075,
      "eval_screenspot_inside_bbox": 0.6150000095367432,
      "eval_screenspot_loss": 1.7993109226226807,
      "eval_screenspot_loss_ce": 0.0005223362823016942,
      "eval_screenspot_loss_iou": 0.7171223958333334,
      "eval_screenspot_loss_num": 0.08854929606119792,
      "eval_screenspot_loss_xval": 1.8779296875,
      "eval_screenspot_runtime": 49.5236,
      "eval_screenspot_samples_per_second": 1.797,
      "eval_screenspot_steps_per_second": 0.061,
      "num_input_tokens_seen": 84012288,
      "step": 1500
    },
    {
      "epoch": 3.34075723830735,
      "eval_compot_CIoU": 0.34500832855701447,
      "eval_compot_GIoU": 0.37410005927085876,
      "eval_compot_IoU": 0.40422166883945465,
      "eval_compot_MAE_all": 0.020579061470925808,
      "eval_compot_MAE_h": 0.009703563060611486,
      "eval_compot_MAE_w": 0.026576916687190533,
      "eval_compot_MAE_x_boxes": 0.03108334168791771,
      "eval_compot_MAE_y_boxes": 0.006412426475435495,
      "eval_compot_inside_bbox": 0.6145833432674408,
      "eval_compot_loss": 1.364414930343628,
      "eval_compot_loss_ce": 0.0003521185863064602,
      "eval_compot_loss_iou": 0.6151123046875,
      "eval_compot_loss_num": 0.019611358642578125,
      "eval_compot_loss_xval": 1.329833984375,
      "eval_compot_runtime": 23.3555,
      "eval_compot_samples_per_second": 2.141,
      "eval_compot_steps_per_second": 0.086,
      "num_input_tokens_seen": 84012288,
      "step": 1500
    },
    {
      "epoch": 3.34075723830735,
      "eval_custom_ui_val_CIoU": 0.4252634909417894,
      "eval_custom_ui_val_GIoU": 0.4529141320122613,
      "eval_custom_ui_val_IoU": 0.48057351344161564,
      "eval_custom_ui_val_MAE_all": 0.03534992608345217,
      "eval_custom_ui_val_MAE_h": 0.020278693590727117,
      "eval_custom_ui_val_MAE_w": 0.039640864771273404,
      "eval_custom_ui_val_MAE_x_boxes": 0.03883321676403284,
      "eval_custom_ui_val_MAE_y_boxes": 0.020278344189541206,
      "eval_custom_ui_val_inside_bbox": 0.6658950646718343,
      "eval_custom_ui_val_loss": 1.297371745109558,
      "eval_custom_ui_val_loss_ce": 0.0004658611885841108,
      "eval_custom_ui_val_loss_iou": 0.5472547743055556,
      "eval_custom_ui_val_loss_num": 0.03508186340332031,
      "eval_custom_ui_val_loss_xval": 1.2700737847222223,
      "eval_custom_ui_val_runtime": 65.9243,
      "eval_custom_ui_val_samples_per_second": 4.02,
      "eval_custom_ui_val_steps_per_second": 0.137,
      "num_input_tokens_seen": 84012288,
      "step": 1500
    },
    {
      "epoch": 3.34075723830735,
      "loss": 1.047553539276123,
      "loss_ce": 0.00043441198067739606,
      "loss_iou": 0.453125,
      "loss_num": 0.0279541015625,
      "loss_xval": 1.046875,
      "num_input_tokens_seen": 84012288,
      "step": 1500
    },
    {
      "epoch": 3.3429844097995547,
      "grad_norm": 17.67667007446289,
      "learning_rate": 1e-06,
      "loss": 0.8425,
      "num_input_tokens_seen": 84069404,
      "step": 1501
    },
    {
      "epoch": 3.3429844097995547,
      "loss": 0.8640942573547363,
      "loss_ce": 0.00032469953293912113,
      "loss_iou": 0.33203125,
      "loss_num": 0.039794921875,
      "loss_xval": 0.86328125,
      "num_input_tokens_seen": 84069404,
      "step": 1501
    },
    {
      "epoch": 3.3452115812917596,
      "grad_norm": 17.004056930541992,
      "learning_rate": 1e-06,
      "loss": 0.9548,
      "num_input_tokens_seen": 84124248,
      "step": 1502
    },
    {
      "epoch": 3.3452115812917596,
      "loss": 1.012810468673706,
      "loss_ce": 0.0003593353903852403,
      "loss_iou": 0.4296875,
      "loss_num": 0.0306396484375,
      "loss_xval": 1.015625,
      "num_input_tokens_seen": 84124248,
      "step": 1502
    },
    {
      "epoch": 3.3474387527839644,
      "grad_norm": 17.8640193939209,
      "learning_rate": 1e-06,
      "loss": 0.8929,
      "num_input_tokens_seen": 84182520,
      "step": 1503
    },
    {
      "epoch": 3.3474387527839644,
      "loss": 1.041400671005249,
      "loss_ce": 0.0003850944631267339,
      "loss_iou": 0.453125,
      "loss_num": 0.02685546875,
      "loss_xval": 1.0390625,
      "num_input_tokens_seen": 84182520,
      "step": 1503
    },
    {
      "epoch": 3.3496659242761693,
      "grad_norm": 31.79290199279785,
      "learning_rate": 1e-06,
      "loss": 0.6244,
      "num_input_tokens_seen": 84241328,
      "step": 1504
    },
    {
      "epoch": 3.3496659242761693,
      "loss": 0.537376880645752,
      "loss_ce": 0.0002675021532922983,
      "loss_iou": 0.2177734375,
      "loss_num": 0.0201416015625,
      "loss_xval": 0.5390625,
      "num_input_tokens_seen": 84241328,
      "step": 1504
    },
    {
      "epoch": 3.351893095768374,
      "grad_norm": 19.212589263916016,
      "learning_rate": 1e-06,
      "loss": 0.8643,
      "num_input_tokens_seen": 84296620,
      "step": 1505
    },
    {
      "epoch": 3.351893095768374,
      "loss": 0.6515594124794006,
      "loss_ce": 0.0004363722400739789,
      "loss_iou": 0.236328125,
      "loss_num": 0.035400390625,
      "loss_xval": 0.65234375,
      "num_input_tokens_seen": 84296620,
      "step": 1505
    },
    {
      "epoch": 3.354120267260579,
      "grad_norm": 20.98340606689453,
      "learning_rate": 1e-06,
      "loss": 0.6714,
      "num_input_tokens_seen": 84350084,
      "step": 1506
    },
    {
      "epoch": 3.354120267260579,
      "loss": 0.7122939825057983,
      "loss_ce": 0.0003799225087277591,
      "loss_iou": 0.296875,
      "loss_num": 0.023681640625,
      "loss_xval": 0.7109375,
      "num_input_tokens_seen": 84350084,
      "step": 1506
    },
    {
      "epoch": 3.356347438752784,
      "grad_norm": 16.530118942260742,
      "learning_rate": 1e-06,
      "loss": 0.8061,
      "num_input_tokens_seen": 84402640,
      "step": 1507
    },
    {
      "epoch": 3.356347438752784,
      "loss": 0.8870806694030762,
      "loss_ce": 0.0003618546761572361,
      "loss_iou": 0.375,
      "loss_num": 0.027587890625,
      "loss_xval": 0.88671875,
      "num_input_tokens_seen": 84402640,
      "step": 1507
    },
    {
      "epoch": 3.3585746102449887,
      "grad_norm": 22.20184326171875,
      "learning_rate": 1e-06,
      "loss": 0.818,
      "num_input_tokens_seen": 84458176,
      "step": 1508
    },
    {
      "epoch": 3.3585746102449887,
      "loss": 0.7754631638526917,
      "loss_ce": 0.0003166621900163591,
      "loss_iou": 0.341796875,
      "loss_num": 0.0181884765625,
      "loss_xval": 0.7734375,
      "num_input_tokens_seen": 84458176,
      "step": 1508
    },
    {
      "epoch": 3.3608017817371936,
      "grad_norm": 17.622278213500977,
      "learning_rate": 1e-06,
      "loss": 0.6034,
      "num_input_tokens_seen": 84514432,
      "step": 1509
    },
    {
      "epoch": 3.3608017817371936,
      "loss": 0.4961128532886505,
      "loss_ce": 0.0002632543910294771,
      "loss_iou": 0.2109375,
      "loss_num": 0.01507568359375,
      "loss_xval": 0.49609375,
      "num_input_tokens_seen": 84514432,
      "step": 1509
    },
    {
      "epoch": 3.3630289532293984,
      "grad_norm": 19.41647720336914,
      "learning_rate": 1e-06,
      "loss": 0.9464,
      "num_input_tokens_seen": 84572280,
      "step": 1510
    },
    {
      "epoch": 3.3630289532293984,
      "loss": 0.9257339239120483,
      "loss_ce": 0.0004409342654980719,
      "loss_iou": 0.3984375,
      "loss_num": 0.0257568359375,
      "loss_xval": 0.92578125,
      "num_input_tokens_seen": 84572280,
      "step": 1510
    },
    {
      "epoch": 3.3652561247216037,
      "grad_norm": 30.21278190612793,
      "learning_rate": 1e-06,
      "loss": 0.8469,
      "num_input_tokens_seen": 84629116,
      "step": 1511
    },
    {
      "epoch": 3.3652561247216037,
      "loss": 0.769070029258728,
      "loss_ce": 0.0002712323039304465,
      "loss_iou": 0.333984375,
      "loss_num": 0.0203857421875,
      "loss_xval": 0.76953125,
      "num_input_tokens_seen": 84629116,
      "step": 1511
    },
    {
      "epoch": 3.3674832962138086,
      "grad_norm": 27.09176254272461,
      "learning_rate": 1e-06,
      "loss": 1.0643,
      "num_input_tokens_seen": 84685628,
      "step": 1512
    },
    {
      "epoch": 3.3674832962138086,
      "loss": 1.0953569412231445,
      "loss_ce": 0.0003862777375616133,
      "loss_iou": 0.421875,
      "loss_num": 0.050048828125,
      "loss_xval": 1.09375,
      "num_input_tokens_seen": 84685628,
      "step": 1512
    },
    {
      "epoch": 3.3697104677060135,
      "grad_norm": 29.004039764404297,
      "learning_rate": 1e-06,
      "loss": 0.8344,
      "num_input_tokens_seen": 84743456,
      "step": 1513
    },
    {
      "epoch": 3.3697104677060135,
      "loss": 1.0513304471969604,
      "loss_ce": 0.0003050968807656318,
      "loss_iou": 0.40625,
      "loss_num": 0.048095703125,
      "loss_xval": 1.0546875,
      "num_input_tokens_seen": 84743456,
      "step": 1513
    },
    {
      "epoch": 3.3719376391982183,
      "grad_norm": 14.021499633789062,
      "learning_rate": 1e-06,
      "loss": 0.7724,
      "num_input_tokens_seen": 84800152,
      "step": 1514
    },
    {
      "epoch": 3.3719376391982183,
      "loss": 0.7707792520523071,
      "loss_ce": 0.004665941931307316,
      "loss_iou": 0.296875,
      "loss_num": 0.0341796875,
      "loss_xval": 0.765625,
      "num_input_tokens_seen": 84800152,
      "step": 1514
    },
    {
      "epoch": 3.374164810690423,
      "grad_norm": 21.85736656188965,
      "learning_rate": 1e-06,
      "loss": 0.9805,
      "num_input_tokens_seen": 84856756,
      "step": 1515
    },
    {
      "epoch": 3.374164810690423,
      "loss": 1.0318880081176758,
      "loss_ce": 0.0016146359266713262,
      "loss_iou": 0.404296875,
      "loss_num": 0.044189453125,
      "loss_xval": 1.03125,
      "num_input_tokens_seen": 84856756,
      "step": 1515
    },
    {
      "epoch": 3.376391982182628,
      "grad_norm": 17.330366134643555,
      "learning_rate": 1e-06,
      "loss": 0.8845,
      "num_input_tokens_seen": 84913964,
      "step": 1516
    },
    {
      "epoch": 3.376391982182628,
      "loss": 0.9630565643310547,
      "loss_ce": 0.00041007917025126517,
      "loss_iou": 0.4140625,
      "loss_num": 0.0264892578125,
      "loss_xval": 0.9609375,
      "num_input_tokens_seen": 84913964,
      "step": 1516
    },
    {
      "epoch": 3.378619153674833,
      "grad_norm": 28.434722900390625,
      "learning_rate": 1e-06,
      "loss": 0.7971,
      "num_input_tokens_seen": 84973372,
      "step": 1517
    },
    {
      "epoch": 3.378619153674833,
      "loss": 0.6269816160202026,
      "loss_ce": 0.0002726696548052132,
      "loss_iou": 0.255859375,
      "loss_num": 0.023193359375,
      "loss_xval": 0.625,
      "num_input_tokens_seen": 84973372,
      "step": 1517
    },
    {
      "epoch": 3.3808463251670378,
      "grad_norm": 22.015655517578125,
      "learning_rate": 1e-06,
      "loss": 0.7052,
      "num_input_tokens_seen": 85029976,
      "step": 1518
    },
    {
      "epoch": 3.3808463251670378,
      "loss": 0.7049768567085266,
      "loss_ce": 0.00038701502489857376,
      "loss_iou": 0.314453125,
      "loss_num": 0.01513671875,
      "loss_xval": 0.703125,
      "num_input_tokens_seen": 85029976,
      "step": 1518
    },
    {
      "epoch": 3.3830734966592426,
      "grad_norm": 18.153146743774414,
      "learning_rate": 1e-06,
      "loss": 0.7883,
      "num_input_tokens_seen": 85085816,
      "step": 1519
    },
    {
      "epoch": 3.3830734966592426,
      "loss": 0.9505753517150879,
      "loss_ce": 0.00025792684755288064,
      "loss_iou": 0.390625,
      "loss_num": 0.0341796875,
      "loss_xval": 0.94921875,
      "num_input_tokens_seen": 85085816,
      "step": 1519
    },
    {
      "epoch": 3.3853006681514475,
      "grad_norm": 21.065345764160156,
      "learning_rate": 1e-06,
      "loss": 0.9315,
      "num_input_tokens_seen": 85143668,
      "step": 1520
    },
    {
      "epoch": 3.3853006681514475,
      "loss": 0.7659118175506592,
      "loss_ce": 0.00028681475669145584,
      "loss_iou": 0.3359375,
      "loss_num": 0.0185546875,
      "loss_xval": 0.765625,
      "num_input_tokens_seen": 85143668,
      "step": 1520
    },
    {
      "epoch": 3.387527839643653,
      "grad_norm": 25.462947845458984,
      "learning_rate": 1e-06,
      "loss": 0.9036,
      "num_input_tokens_seen": 85199320,
      "step": 1521
    },
    {
      "epoch": 3.387527839643653,
      "loss": 0.8738066554069519,
      "loss_ce": 0.00027149327797815204,
      "loss_iou": 0.380859375,
      "loss_num": 0.0224609375,
      "loss_xval": 0.875,
      "num_input_tokens_seen": 85199320,
      "step": 1521
    },
    {
      "epoch": 3.3897550111358576,
      "grad_norm": 22.21038055419922,
      "learning_rate": 1e-06,
      "loss": 0.8248,
      "num_input_tokens_seen": 85256340,
      "step": 1522
    },
    {
      "epoch": 3.3897550111358576,
      "loss": 0.8254960179328918,
      "loss_ce": 0.0003007104678545147,
      "loss_iou": 0.359375,
      "loss_num": 0.0208740234375,
      "loss_xval": 0.82421875,
      "num_input_tokens_seen": 85256340,
      "step": 1522
    },
    {
      "epoch": 3.3919821826280625,
      "grad_norm": 17.12187385559082,
      "learning_rate": 1e-06,
      "loss": 0.826,
      "num_input_tokens_seen": 85312876,
      "step": 1523
    },
    {
      "epoch": 3.3919821826280625,
      "loss": 0.9279178380966187,
      "loss_ce": 0.0004276382096577436,
      "loss_iou": 0.37890625,
      "loss_num": 0.033935546875,
      "loss_xval": 0.92578125,
      "num_input_tokens_seen": 85312876,
      "step": 1523
    },
    {
      "epoch": 3.3942093541202674,
      "grad_norm": 20.59052085876465,
      "learning_rate": 1e-06,
      "loss": 0.8034,
      "num_input_tokens_seen": 85369068,
      "step": 1524
    },
    {
      "epoch": 3.3942093541202674,
      "loss": 0.9275302886962891,
      "loss_ce": 0.00028415530687198043,
      "loss_iou": 0.36328125,
      "loss_num": 0.0400390625,
      "loss_xval": 0.92578125,
      "num_input_tokens_seen": 85369068,
      "step": 1524
    },
    {
      "epoch": 3.3964365256124722,
      "grad_norm": 23.074607849121094,
      "learning_rate": 1e-06,
      "loss": 1.0088,
      "num_input_tokens_seen": 85424964,
      "step": 1525
    },
    {
      "epoch": 3.3964365256124722,
      "loss": 1.4848394393920898,
      "loss_ce": 0.00046440563164651394,
      "loss_iou": 0.5859375,
      "loss_num": 0.0634765625,
      "loss_xval": 1.484375,
      "num_input_tokens_seen": 85424964,
      "step": 1525
    },
    {
      "epoch": 3.398663697104677,
      "grad_norm": 103.47501373291016,
      "learning_rate": 1e-06,
      "loss": 0.7614,
      "num_input_tokens_seen": 85481460,
      "step": 1526
    },
    {
      "epoch": 3.398663697104677,
      "loss": 0.9770767688751221,
      "loss_ce": 0.00027014350052922964,
      "loss_iou": 0.400390625,
      "loss_num": 0.03564453125,
      "loss_xval": 0.9765625,
      "num_input_tokens_seen": 85481460,
      "step": 1526
    },
    {
      "epoch": 3.400890868596882,
      "grad_norm": 16.694969177246094,
      "learning_rate": 1e-06,
      "loss": 0.7637,
      "num_input_tokens_seen": 85538076,
      "step": 1527
    },
    {
      "epoch": 3.400890868596882,
      "loss": 0.7700480818748474,
      "loss_ce": 0.00027271179715171456,
      "loss_iou": 0.341796875,
      "loss_num": 0.01708984375,
      "loss_xval": 0.76953125,
      "num_input_tokens_seen": 85538076,
      "step": 1527
    },
    {
      "epoch": 3.403118040089087,
      "grad_norm": 15.084866523742676,
      "learning_rate": 1e-06,
      "loss": 0.8101,
      "num_input_tokens_seen": 85593752,
      "step": 1528
    },
    {
      "epoch": 3.403118040089087,
      "loss": 0.7312592267990112,
      "loss_ce": 0.0003021882730536163,
      "loss_iou": 0.30859375,
      "loss_num": 0.0225830078125,
      "loss_xval": 0.73046875,
      "num_input_tokens_seen": 85593752,
      "step": 1528
    },
    {
      "epoch": 3.4053452115812917,
      "grad_norm": 137.56297302246094,
      "learning_rate": 1e-06,
      "loss": 1.0569,
      "num_input_tokens_seen": 85649224,
      "step": 1529
    },
    {
      "epoch": 3.4053452115812917,
      "loss": 0.6406855583190918,
      "loss_ce": 0.00030470843194052577,
      "loss_iou": 0.2890625,
      "loss_num": 0.0125732421875,
      "loss_xval": 0.640625,
      "num_input_tokens_seen": 85649224,
      "step": 1529
    },
    {
      "epoch": 3.4075723830734965,
      "grad_norm": 17.245445251464844,
      "learning_rate": 1e-06,
      "loss": 0.8775,
      "num_input_tokens_seen": 85704284,
      "step": 1530
    },
    {
      "epoch": 3.4075723830734965,
      "loss": 0.7688406109809875,
      "loss_ce": 0.000285913614789024,
      "loss_iou": 0.326171875,
      "loss_num": 0.0233154296875,
      "loss_xval": 0.76953125,
      "num_input_tokens_seen": 85704284,
      "step": 1530
    },
    {
      "epoch": 3.4097995545657014,
      "grad_norm": 28.163143157958984,
      "learning_rate": 1e-06,
      "loss": 0.6158,
      "num_input_tokens_seen": 85758312,
      "step": 1531
    },
    {
      "epoch": 3.4097995545657014,
      "loss": 0.6164954900741577,
      "loss_ce": 0.000772814848460257,
      "loss_iou": 0.28125,
      "loss_num": 0.01080322265625,
      "loss_xval": 0.6171875,
      "num_input_tokens_seen": 85758312,
      "step": 1531
    },
    {
      "epoch": 3.4120267260579062,
      "grad_norm": 20.04214096069336,
      "learning_rate": 1e-06,
      "loss": 0.8842,
      "num_input_tokens_seen": 85814628,
      "step": 1532
    },
    {
      "epoch": 3.4120267260579062,
      "loss": 1.005671739578247,
      "loss_ce": 0.00030057106050662696,
      "loss_iou": 0.4296875,
      "loss_num": 0.029052734375,
      "loss_xval": 1.0078125,
      "num_input_tokens_seen": 85814628,
      "step": 1532
    },
    {
      "epoch": 3.4142538975501115,
      "grad_norm": 31.450151443481445,
      "learning_rate": 1e-06,
      "loss": 0.8459,
      "num_input_tokens_seen": 85870152,
      "step": 1533
    },
    {
      "epoch": 3.4142538975501115,
      "loss": 1.0091707706451416,
      "loss_ce": 0.0003816866665147245,
      "loss_iou": 0.41796875,
      "loss_num": 0.0341796875,
      "loss_xval": 1.0078125,
      "num_input_tokens_seen": 85870152,
      "step": 1533
    },
    {
      "epoch": 3.4164810690423164,
      "grad_norm": 16.301393508911133,
      "learning_rate": 1e-06,
      "loss": 1.1544,
      "num_input_tokens_seen": 85923552,
      "step": 1534
    },
    {
      "epoch": 3.4164810690423164,
      "loss": 1.0733964443206787,
      "loss_ce": 0.0003984392969869077,
      "loss_iou": 0.412109375,
      "loss_num": 0.04931640625,
      "loss_xval": 1.0703125,
      "num_input_tokens_seen": 85923552,
      "step": 1534
    },
    {
      "epoch": 3.4187082405345213,
      "grad_norm": 22.28922462463379,
      "learning_rate": 1e-06,
      "loss": 1.2395,
      "num_input_tokens_seen": 85978024,
      "step": 1535
    },
    {
      "epoch": 3.4187082405345213,
      "loss": 1.1963934898376465,
      "loss_ce": 0.0005926065496169031,
      "loss_iou": 0.4609375,
      "loss_num": 0.0546875,
      "loss_xval": 1.1953125,
      "num_input_tokens_seen": 85978024,
      "step": 1535
    },
    {
      "epoch": 3.420935412026726,
      "grad_norm": 17.531831741333008,
      "learning_rate": 1e-06,
      "loss": 0.857,
      "num_input_tokens_seen": 86032808,
      "step": 1536
    },
    {
      "epoch": 3.420935412026726,
      "loss": 0.6863161325454712,
      "loss_ce": 0.00028098217444494367,
      "loss_iou": 0.267578125,
      "loss_num": 0.0299072265625,
      "loss_xval": 0.6875,
      "num_input_tokens_seen": 86032808,
      "step": 1536
    },
    {
      "epoch": 3.423162583518931,
      "grad_norm": 40.43157958984375,
      "learning_rate": 1e-06,
      "loss": 0.9083,
      "num_input_tokens_seen": 86084988,
      "step": 1537
    },
    {
      "epoch": 3.423162583518931,
      "loss": 1.0269807577133179,
      "loss_ce": 0.00036941259168088436,
      "loss_iou": 0.416015625,
      "loss_num": 0.038330078125,
      "loss_xval": 1.0234375,
      "num_input_tokens_seen": 86084988,
      "step": 1537
    },
    {
      "epoch": 3.425389755011136,
      "grad_norm": 13.929858207702637,
      "learning_rate": 1e-06,
      "loss": 0.5897,
      "num_input_tokens_seen": 86141804,
      "step": 1538
    },
    {
      "epoch": 3.425389755011136,
      "loss": 0.5110272765159607,
      "loss_ce": 0.00028507091337814927,
      "loss_iou": 0.228515625,
      "loss_num": 0.0106201171875,
      "loss_xval": 0.51171875,
      "num_input_tokens_seen": 86141804,
      "step": 1538
    },
    {
      "epoch": 3.4276169265033407,
      "grad_norm": 22.480680465698242,
      "learning_rate": 1e-06,
      "loss": 1.0234,
      "num_input_tokens_seen": 86196392,
      "step": 1539
    },
    {
      "epoch": 3.4276169265033407,
      "loss": 1.0525072813034058,
      "loss_ce": 0.0002611761447042227,
      "loss_iou": 0.3828125,
      "loss_num": 0.05712890625,
      "loss_xval": 1.0546875,
      "num_input_tokens_seen": 86196392,
      "step": 1539
    },
    {
      "epoch": 3.4298440979955456,
      "grad_norm": 20.415674209594727,
      "learning_rate": 1e-06,
      "loss": 0.7179,
      "num_input_tokens_seen": 86254652,
      "step": 1540
    },
    {
      "epoch": 3.4298440979955456,
      "loss": 0.7557092905044556,
      "loss_ce": 0.0003381772548891604,
      "loss_iou": 0.3125,
      "loss_num": 0.025390625,
      "loss_xval": 0.75390625,
      "num_input_tokens_seen": 86254652,
      "step": 1540
    },
    {
      "epoch": 3.4320712694877504,
      "grad_norm": 14.296008110046387,
      "learning_rate": 1e-06,
      "loss": 0.9989,
      "num_input_tokens_seen": 86310292,
      "step": 1541
    },
    {
      "epoch": 3.4320712694877504,
      "loss": 1.1000525951385498,
      "loss_ce": 0.00044316527782939374,
      "loss_iou": 0.44921875,
      "loss_num": 0.0400390625,
      "loss_xval": 1.1015625,
      "num_input_tokens_seen": 86310292,
      "step": 1541
    },
    {
      "epoch": 3.4342984409799553,
      "grad_norm": 15.6137056350708,
      "learning_rate": 1e-06,
      "loss": 0.825,
      "num_input_tokens_seen": 86365832,
      "step": 1542
    },
    {
      "epoch": 3.4342984409799553,
      "loss": 0.7097923755645752,
      "loss_ce": 0.00031970589770935476,
      "loss_iou": 0.32421875,
      "loss_num": 0.01190185546875,
      "loss_xval": 0.7109375,
      "num_input_tokens_seen": 86365832,
      "step": 1542
    },
    {
      "epoch": 3.4365256124721606,
      "grad_norm": 405.90081787109375,
      "learning_rate": 1e-06,
      "loss": 0.8,
      "num_input_tokens_seen": 86423384,
      "step": 1543
    },
    {
      "epoch": 3.4365256124721606,
      "loss": 1.0093533992767334,
      "loss_ce": 0.0008084863657131791,
      "loss_iou": 0.40625,
      "loss_num": 0.0390625,
      "loss_xval": 1.0078125,
      "num_input_tokens_seen": 86423384,
      "step": 1543
    },
    {
      "epoch": 3.4387527839643655,
      "grad_norm": 23.693382263183594,
      "learning_rate": 1e-06,
      "loss": 0.9363,
      "num_input_tokens_seen": 86479144,
      "step": 1544
    },
    {
      "epoch": 3.4387527839643655,
      "loss": 0.854164719581604,
      "loss_ce": 0.0004049596609547734,
      "loss_iou": 0.37109375,
      "loss_num": 0.022216796875,
      "loss_xval": 0.85546875,
      "num_input_tokens_seen": 86479144,
      "step": 1544
    },
    {
      "epoch": 3.4409799554565703,
      "grad_norm": 16.630285263061523,
      "learning_rate": 1e-06,
      "loss": 1.0941,
      "num_input_tokens_seen": 86535676,
      "step": 1545
    },
    {
      "epoch": 3.4409799554565703,
      "loss": 1.3825310468673706,
      "loss_ce": 0.00032891728915274143,
      "loss_iou": 0.5546875,
      "loss_num": 0.053955078125,
      "loss_xval": 1.3828125,
      "num_input_tokens_seen": 86535676,
      "step": 1545
    },
    {
      "epoch": 3.443207126948775,
      "grad_norm": 15.714624404907227,
      "learning_rate": 1e-06,
      "loss": 0.9159,
      "num_input_tokens_seen": 86593604,
      "step": 1546
    },
    {
      "epoch": 3.443207126948775,
      "loss": 0.6130160093307495,
      "loss_ce": 0.00028408144135028124,
      "loss_iou": 0.259765625,
      "loss_num": 0.018310546875,
      "loss_xval": 0.61328125,
      "num_input_tokens_seen": 86593604,
      "step": 1546
    },
    {
      "epoch": 3.44543429844098,
      "grad_norm": 13.893007278442383,
      "learning_rate": 1e-06,
      "loss": 0.7666,
      "num_input_tokens_seen": 86649364,
      "step": 1547
    },
    {
      "epoch": 3.44543429844098,
      "loss": 0.6706589460372925,
      "loss_ce": 0.0002487250021658838,
      "loss_iou": 0.271484375,
      "loss_num": 0.025390625,
      "loss_xval": 0.671875,
      "num_input_tokens_seen": 86649364,
      "step": 1547
    },
    {
      "epoch": 3.447661469933185,
      "grad_norm": 19.675296783447266,
      "learning_rate": 1e-06,
      "loss": 0.8681,
      "num_input_tokens_seen": 86703940,
      "step": 1548
    },
    {
      "epoch": 3.447661469933185,
      "loss": 0.9216609001159668,
      "loss_ce": 0.0002741321222856641,
      "loss_iou": 0.412109375,
      "loss_num": 0.0198974609375,
      "loss_xval": 0.921875,
      "num_input_tokens_seen": 86703940,
      "step": 1548
    },
    {
      "epoch": 3.4498886414253898,
      "grad_norm": 19.515134811401367,
      "learning_rate": 1e-06,
      "loss": 1.096,
      "num_input_tokens_seen": 86760584,
      "step": 1549
    },
    {
      "epoch": 3.4498886414253898,
      "loss": 1.3511126041412354,
      "loss_ce": 0.00028249574825167656,
      "loss_iou": 0.51953125,
      "loss_num": 0.06201171875,
      "loss_xval": 1.3515625,
      "num_input_tokens_seen": 86760584,
      "step": 1549
    },
    {
      "epoch": 3.4521158129175946,
      "grad_norm": 17.653553009033203,
      "learning_rate": 1e-06,
      "loss": 0.7243,
      "num_input_tokens_seen": 86815208,
      "step": 1550
    },
    {
      "epoch": 3.4521158129175946,
      "loss": 0.7034174203872681,
      "loss_ce": 0.0002923937572631985,
      "loss_iou": 0.318359375,
      "loss_num": 0.01348876953125,
      "loss_xval": 0.703125,
      "num_input_tokens_seen": 86815208,
      "step": 1550
    },
    {
      "epoch": 3.4543429844097995,
      "grad_norm": 13.833768844604492,
      "learning_rate": 1e-06,
      "loss": 0.9415,
      "num_input_tokens_seen": 86869804,
      "step": 1551
    },
    {
      "epoch": 3.4543429844097995,
      "loss": 0.8110692501068115,
      "loss_ce": 0.00027830369072034955,
      "loss_iou": 0.3515625,
      "loss_num": 0.0218505859375,
      "loss_xval": 0.8125,
      "num_input_tokens_seen": 86869804,
      "step": 1551
    },
    {
      "epoch": 3.4565701559020043,
      "grad_norm": 21.14474868774414,
      "learning_rate": 1e-06,
      "loss": 0.8695,
      "num_input_tokens_seen": 86925632,
      "step": 1552
    },
    {
      "epoch": 3.4565701559020043,
      "loss": 0.7101556062698364,
      "loss_ce": 0.0004388463275972754,
      "loss_iou": 0.26953125,
      "loss_num": 0.0341796875,
      "loss_xval": 0.7109375,
      "num_input_tokens_seen": 86925632,
      "step": 1552
    },
    {
      "epoch": 3.458797327394209,
      "grad_norm": 23.79280662536621,
      "learning_rate": 1e-06,
      "loss": 1.0303,
      "num_input_tokens_seen": 86983484,
      "step": 1553
    },
    {
      "epoch": 3.458797327394209,
      "loss": 1.0085933208465576,
      "loss_ce": 0.0002924925647675991,
      "loss_iou": 0.443359375,
      "loss_num": 0.0244140625,
      "loss_xval": 1.0078125,
      "num_input_tokens_seen": 86983484,
      "step": 1553
    },
    {
      "epoch": 3.461024498886414,
      "grad_norm": 21.37234115600586,
      "learning_rate": 1e-06,
      "loss": 0.6182,
      "num_input_tokens_seen": 87040560,
      "step": 1554
    },
    {
      "epoch": 3.461024498886414,
      "loss": 0.8163036108016968,
      "loss_ce": 0.0003856316034216434,
      "loss_iou": 0.33203125,
      "loss_num": 0.0306396484375,
      "loss_xval": 0.81640625,
      "num_input_tokens_seen": 87040560,
      "step": 1554
    },
    {
      "epoch": 3.463251670378619,
      "grad_norm": 22.6509952545166,
      "learning_rate": 1e-06,
      "loss": 0.7905,
      "num_input_tokens_seen": 87096772,
      "step": 1555
    },
    {
      "epoch": 3.463251670378619,
      "loss": 0.7041604518890381,
      "loss_ce": 0.00030306505504995584,
      "loss_iou": 0.28515625,
      "loss_num": 0.0267333984375,
      "loss_xval": 0.703125,
      "num_input_tokens_seen": 87096772,
      "step": 1555
    },
    {
      "epoch": 3.4654788418708242,
      "grad_norm": 16.446748733520508,
      "learning_rate": 1e-06,
      "loss": 0.9045,
      "num_input_tokens_seen": 87153824,
      "step": 1556
    },
    {
      "epoch": 3.4654788418708242,
      "loss": 0.9084920287132263,
      "loss_ce": 0.0002888740273192525,
      "loss_iou": 0.392578125,
      "loss_num": 0.0244140625,
      "loss_xval": 0.90625,
      "num_input_tokens_seen": 87153824,
      "step": 1556
    },
    {
      "epoch": 3.467706013363029,
      "grad_norm": 13.511149406433105,
      "learning_rate": 1e-06,
      "loss": 1.0573,
      "num_input_tokens_seen": 87209524,
      "step": 1557
    },
    {
      "epoch": 3.467706013363029,
      "loss": 1.2329661846160889,
      "loss_ce": 0.0003001574077643454,
      "loss_iou": 0.5,
      "loss_num": 0.045654296875,
      "loss_xval": 1.234375,
      "num_input_tokens_seen": 87209524,
      "step": 1557
    },
    {
      "epoch": 3.469933184855234,
      "grad_norm": 57.4884147644043,
      "learning_rate": 1e-06,
      "loss": 0.8035,
      "num_input_tokens_seen": 87264100,
      "step": 1558
    },
    {
      "epoch": 3.469933184855234,
      "loss": 0.8297716379165649,
      "loss_ce": 0.00030388019513338804,
      "loss_iou": 0.359375,
      "loss_num": 0.0224609375,
      "loss_xval": 0.828125,
      "num_input_tokens_seen": 87264100,
      "step": 1558
    },
    {
      "epoch": 3.472160356347439,
      "grad_norm": 15.150367736816406,
      "learning_rate": 1e-06,
      "loss": 0.9682,
      "num_input_tokens_seen": 87321728,
      "step": 1559
    },
    {
      "epoch": 3.472160356347439,
      "loss": 1.013202428817749,
      "loss_ce": 0.0002630281960591674,
      "loss_iou": 0.435546875,
      "loss_num": 0.0286865234375,
      "loss_xval": 1.015625,
      "num_input_tokens_seen": 87321728,
      "step": 1559
    },
    {
      "epoch": 3.4743875278396437,
      "grad_norm": 26.730871200561523,
      "learning_rate": 1e-06,
      "loss": 0.8033,
      "num_input_tokens_seen": 87378128,
      "step": 1560
    },
    {
      "epoch": 3.4743875278396437,
      "loss": 0.7272953391075134,
      "loss_ce": 0.0004887018585577607,
      "loss_iou": 0.326171875,
      "loss_num": 0.014892578125,
      "loss_xval": 0.7265625,
      "num_input_tokens_seen": 87378128,
      "step": 1560
    },
    {
      "epoch": 3.4766146993318485,
      "grad_norm": 63.709983825683594,
      "learning_rate": 1e-06,
      "loss": 0.9815,
      "num_input_tokens_seen": 87432972,
      "step": 1561
    },
    {
      "epoch": 3.4766146993318485,
      "loss": 0.9485345482826233,
      "loss_ce": 0.0002923659631051123,
      "loss_iou": 0.3828125,
      "loss_num": 0.03662109375,
      "loss_xval": 0.94921875,
      "num_input_tokens_seen": 87432972,
      "step": 1561
    },
    {
      "epoch": 3.4788418708240534,
      "grad_norm": 19.27259063720703,
      "learning_rate": 1e-06,
      "loss": 0.7692,
      "num_input_tokens_seen": 87486680,
      "step": 1562
    },
    {
      "epoch": 3.4788418708240534,
      "loss": 0.7349532842636108,
      "loss_ce": 0.00033419817918911576,
      "loss_iou": 0.33203125,
      "loss_num": 0.01409912109375,
      "loss_xval": 0.734375,
      "num_input_tokens_seen": 87486680,
      "step": 1562
    },
    {
      "epoch": 3.4810690423162582,
      "grad_norm": 17.00948143005371,
      "learning_rate": 1e-06,
      "loss": 0.8309,
      "num_input_tokens_seen": 87543352,
      "step": 1563
    },
    {
      "epoch": 3.4810690423162582,
      "loss": 0.9362764954566956,
      "loss_ce": 0.00024133155238814652,
      "loss_iou": 0.412109375,
      "loss_num": 0.022216796875,
      "loss_xval": 0.9375,
      "num_input_tokens_seen": 87543352,
      "step": 1563
    },
    {
      "epoch": 3.483296213808463,
      "grad_norm": 24.209060668945312,
      "learning_rate": 1e-06,
      "loss": 0.9046,
      "num_input_tokens_seen": 87600896,
      "step": 1564
    },
    {
      "epoch": 3.483296213808463,
      "loss": 1.042458176612854,
      "loss_ce": 0.00046601821668446064,
      "loss_iou": 0.43359375,
      "loss_num": 0.03515625,
      "loss_xval": 1.0390625,
      "num_input_tokens_seen": 87600896,
      "step": 1564
    },
    {
      "epoch": 3.485523385300668,
      "grad_norm": 14.757556915283203,
      "learning_rate": 1e-06,
      "loss": 0.9321,
      "num_input_tokens_seen": 87658544,
      "step": 1565
    },
    {
      "epoch": 3.485523385300668,
      "loss": 0.7475967407226562,
      "loss_ce": 0.000282272812910378,
      "loss_iou": 0.3046875,
      "loss_num": 0.02783203125,
      "loss_xval": 0.74609375,
      "num_input_tokens_seen": 87658544,
      "step": 1565
    },
    {
      "epoch": 3.4877505567928733,
      "grad_norm": 29.213010787963867,
      "learning_rate": 1e-06,
      "loss": 0.9172,
      "num_input_tokens_seen": 87715960,
      "step": 1566
    },
    {
      "epoch": 3.4877505567928733,
      "loss": 0.8977096676826477,
      "loss_ce": 0.00049288832815364,
      "loss_iou": 0.3515625,
      "loss_num": 0.038818359375,
      "loss_xval": 0.8984375,
      "num_input_tokens_seen": 87715960,
      "step": 1566
    },
    {
      "epoch": 3.489977728285078,
      "grad_norm": 13.978538513183594,
      "learning_rate": 1e-06,
      "loss": 0.8842,
      "num_input_tokens_seen": 87769772,
      "step": 1567
    },
    {
      "epoch": 3.489977728285078,
      "loss": 0.8801678419113159,
      "loss_ce": 0.0002850402379408479,
      "loss_iou": 0.37109375,
      "loss_num": 0.0272216796875,
      "loss_xval": 0.87890625,
      "num_input_tokens_seen": 87769772,
      "step": 1567
    },
    {
      "epoch": 3.492204899777283,
      "grad_norm": 28.59236717224121,
      "learning_rate": 1e-06,
      "loss": 0.646,
      "num_input_tokens_seen": 87825804,
      "step": 1568
    },
    {
      "epoch": 3.492204899777283,
      "loss": 0.6284915208816528,
      "loss_ce": 0.00031768600456416607,
      "loss_iou": 0.28125,
      "loss_num": 0.0130615234375,
      "loss_xval": 0.62890625,
      "num_input_tokens_seen": 87825804,
      "step": 1568
    },
    {
      "epoch": 3.494432071269488,
      "grad_norm": 27.629867553710938,
      "learning_rate": 1e-06,
      "loss": 0.8455,
      "num_input_tokens_seen": 87881064,
      "step": 1569
    },
    {
      "epoch": 3.494432071269488,
      "loss": 0.7841451168060303,
      "loss_ce": 0.00027057109400629997,
      "loss_iou": 0.34375,
      "loss_num": 0.0191650390625,
      "loss_xval": 0.78515625,
      "num_input_tokens_seen": 87881064,
      "step": 1569
    },
    {
      "epoch": 3.4966592427616927,
      "grad_norm": 28.423208236694336,
      "learning_rate": 1e-06,
      "loss": 1.0515,
      "num_input_tokens_seen": 87938884,
      "step": 1570
    },
    {
      "epoch": 3.4966592427616927,
      "loss": 1.0150330066680908,
      "loss_ce": 0.00038457888877019286,
      "loss_iou": 0.4375,
      "loss_num": 0.0279541015625,
      "loss_xval": 1.015625,
      "num_input_tokens_seen": 87938884,
      "step": 1570
    },
    {
      "epoch": 3.4988864142538976,
      "grad_norm": 80.1189956665039,
      "learning_rate": 1e-06,
      "loss": 0.8306,
      "num_input_tokens_seen": 87991052,
      "step": 1571
    },
    {
      "epoch": 3.4988864142538976,
      "loss": 0.9712834358215332,
      "loss_ce": 0.0005803282838314772,
      "loss_iou": 0.412109375,
      "loss_num": 0.0294189453125,
      "loss_xval": 0.96875,
      "num_input_tokens_seen": 87991052,
      "step": 1571
    },
    {
      "epoch": 3.5011135857461024,
      "grad_norm": 33.21288299560547,
      "learning_rate": 1e-06,
      "loss": 0.999,
      "num_input_tokens_seen": 88047996,
      "step": 1572
    },
    {
      "epoch": 3.5011135857461024,
      "loss": 0.9822986721992493,
      "loss_ce": 0.00036505749449133873,
      "loss_iou": 0.412109375,
      "loss_num": 0.031982421875,
      "loss_xval": 0.98046875,
      "num_input_tokens_seen": 88047996,
      "step": 1572
    },
    {
      "epoch": 3.5033407572383073,
      "grad_norm": 27.994529724121094,
      "learning_rate": 1e-06,
      "loss": 0.8276,
      "num_input_tokens_seen": 88102512,
      "step": 1573
    },
    {
      "epoch": 3.5033407572383073,
      "loss": 0.9002130031585693,
      "loss_ce": 0.0003106549265794456,
      "loss_iou": 0.392578125,
      "loss_num": 0.0233154296875,
      "loss_xval": 0.8984375,
      "num_input_tokens_seen": 88102512,
      "step": 1573
    },
    {
      "epoch": 3.505567928730512,
      "grad_norm": 29.116310119628906,
      "learning_rate": 1e-06,
      "loss": 1.0666,
      "num_input_tokens_seen": 88156400,
      "step": 1574
    },
    {
      "epoch": 3.505567928730512,
      "loss": 1.0999196767807007,
      "loss_ce": 0.0003103648195974529,
      "loss_iou": 0.47265625,
      "loss_num": 0.031005859375,
      "loss_xval": 1.1015625,
      "num_input_tokens_seen": 88156400,
      "step": 1574
    },
    {
      "epoch": 3.507795100222717,
      "grad_norm": 18.990516662597656,
      "learning_rate": 1e-06,
      "loss": 0.7131,
      "num_input_tokens_seen": 88211188,
      "step": 1575
    },
    {
      "epoch": 3.507795100222717,
      "loss": 0.8792534470558167,
      "loss_ce": 0.000347210094332695,
      "loss_iou": 0.388671875,
      "loss_num": 0.02001953125,
      "loss_xval": 0.87890625,
      "num_input_tokens_seen": 88211188,
      "step": 1575
    },
    {
      "epoch": 3.510022271714922,
      "grad_norm": 21.322250366210938,
      "learning_rate": 1e-06,
      "loss": 0.8679,
      "num_input_tokens_seen": 88267384,
      "step": 1576
    },
    {
      "epoch": 3.510022271714922,
      "loss": 1.0999305248260498,
      "loss_ce": 0.0003211447037756443,
      "loss_iou": 0.482421875,
      "loss_num": 0.0272216796875,
      "loss_xval": 1.1015625,
      "num_input_tokens_seen": 88267384,
      "step": 1576
    },
    {
      "epoch": 3.5122494432071267,
      "grad_norm": 26.06310272216797,
      "learning_rate": 1e-06,
      "loss": 0.9818,
      "num_input_tokens_seen": 88323956,
      "step": 1577
    },
    {
      "epoch": 3.5122494432071267,
      "loss": 1.0118680000305176,
      "loss_ce": 0.0003933944390155375,
      "loss_iou": 0.40625,
      "loss_num": 0.03955078125,
      "loss_xval": 1.0078125,
      "num_input_tokens_seen": 88323956,
      "step": 1577
    },
    {
      "epoch": 3.5144766146993316,
      "grad_norm": 17.4585018157959,
      "learning_rate": 1e-06,
      "loss": 0.9207,
      "num_input_tokens_seen": 88377432,
      "step": 1578
    },
    {
      "epoch": 3.5144766146993316,
      "loss": 0.7224873304367065,
      "loss_ce": 0.00031937778112478554,
      "loss_iou": 0.314453125,
      "loss_num": 0.01904296875,
      "loss_xval": 0.72265625,
      "num_input_tokens_seen": 88377432,
      "step": 1578
    },
    {
      "epoch": 3.516703786191537,
      "grad_norm": 24.705142974853516,
      "learning_rate": 1e-06,
      "loss": 0.9077,
      "num_input_tokens_seen": 88433012,
      "step": 1579
    },
    {
      "epoch": 3.516703786191537,
      "loss": 1.0539125204086304,
      "loss_ce": 0.00044574099592864513,
      "loss_iou": 0.400390625,
      "loss_num": 0.05078125,
      "loss_xval": 1.0546875,
      "num_input_tokens_seen": 88433012,
      "step": 1579
    },
    {
      "epoch": 3.5189309576837418,
      "grad_norm": 30.141464233398438,
      "learning_rate": 1e-06,
      "loss": 0.7654,
      "num_input_tokens_seen": 88490756,
      "step": 1580
    },
    {
      "epoch": 3.5189309576837418,
      "loss": 0.8266535997390747,
      "loss_ce": 0.00048178687575273216,
      "loss_iou": 0.306640625,
      "loss_num": 0.042236328125,
      "loss_xval": 0.828125,
      "num_input_tokens_seen": 88490756,
      "step": 1580
    },
    {
      "epoch": 3.5211581291759466,
      "grad_norm": 21.32522964477539,
      "learning_rate": 1e-06,
      "loss": 0.8354,
      "num_input_tokens_seen": 88548592,
      "step": 1581
    },
    {
      "epoch": 3.5211581291759466,
      "loss": 0.9959642291069031,
      "loss_ce": 0.0006028971401974559,
      "loss_iou": 0.443359375,
      "loss_num": 0.021484375,
      "loss_xval": 0.99609375,
      "num_input_tokens_seen": 88548592,
      "step": 1581
    },
    {
      "epoch": 3.5233853006681515,
      "grad_norm": 26.187313079833984,
      "learning_rate": 1e-06,
      "loss": 0.7145,
      "num_input_tokens_seen": 88605516,
      "step": 1582
    },
    {
      "epoch": 3.5233853006681515,
      "loss": 0.4769740104675293,
      "loss_ce": 0.0002894522622227669,
      "loss_iou": 0.212890625,
      "loss_num": 0.01025390625,
      "loss_xval": 0.4765625,
      "num_input_tokens_seen": 88605516,
      "step": 1582
    },
    {
      "epoch": 3.5256124721603563,
      "grad_norm": 27.296201705932617,
      "learning_rate": 1e-06,
      "loss": 0.8584,
      "num_input_tokens_seen": 88660980,
      "step": 1583
    },
    {
      "epoch": 3.5256124721603563,
      "loss": 0.7954574823379517,
      "loss_ce": 0.0002915282384492457,
      "loss_iou": 0.33203125,
      "loss_num": 0.026123046875,
      "loss_xval": 0.796875,
      "num_input_tokens_seen": 88660980,
      "step": 1583
    },
    {
      "epoch": 3.527839643652561,
      "grad_norm": 18.88905143737793,
      "learning_rate": 1e-06,
      "loss": 0.898,
      "num_input_tokens_seen": 88715208,
      "step": 1584
    },
    {
      "epoch": 3.527839643652561,
      "loss": 0.7566571235656738,
      "loss_ce": 0.0003094491839874536,
      "loss_iou": 0.328125,
      "loss_num": 0.02001953125,
      "loss_xval": 0.7578125,
      "num_input_tokens_seen": 88715208,
      "step": 1584
    },
    {
      "epoch": 3.530066815144766,
      "grad_norm": 17.91330337524414,
      "learning_rate": 1e-06,
      "loss": 0.7627,
      "num_input_tokens_seen": 88773392,
      "step": 1585
    },
    {
      "epoch": 3.530066815144766,
      "loss": 0.8976699113845825,
      "loss_ce": 0.0004530604637693614,
      "loss_iou": 0.380859375,
      "loss_num": 0.02685546875,
      "loss_xval": 0.8984375,
      "num_input_tokens_seen": 88773392,
      "step": 1585
    },
    {
      "epoch": 3.532293986636971,
      "grad_norm": 15.070040702819824,
      "learning_rate": 1e-06,
      "loss": 1.0047,
      "num_input_tokens_seen": 88830772,
      "step": 1586
    },
    {
      "epoch": 3.532293986636971,
      "loss": 0.675557017326355,
      "loss_ce": 0.0002640365855768323,
      "loss_iou": 0.30078125,
      "loss_num": 0.0145263671875,
      "loss_xval": 0.67578125,
      "num_input_tokens_seen": 88830772,
      "step": 1586
    },
    {
      "epoch": 3.534521158129176,
      "grad_norm": 15.70621109008789,
      "learning_rate": 1e-06,
      "loss": 0.9061,
      "num_input_tokens_seen": 88887288,
      "step": 1587
    },
    {
      "epoch": 3.534521158129176,
      "loss": 1.0725452899932861,
      "loss_ce": 0.000279579107882455,
      "loss_iou": 0.431640625,
      "loss_num": 0.041259765625,
      "loss_xval": 1.0703125,
      "num_input_tokens_seen": 88887288,
      "step": 1587
    },
    {
      "epoch": 3.536748329621381,
      "grad_norm": 16.636899948120117,
      "learning_rate": 1e-06,
      "loss": 0.7762,
      "num_input_tokens_seen": 88943304,
      "step": 1588
    },
    {
      "epoch": 3.536748329621381,
      "loss": 0.7620242834091187,
      "loss_ce": 0.00030552022508345544,
      "loss_iou": 0.333984375,
      "loss_num": 0.0189208984375,
      "loss_xval": 0.76171875,
      "num_input_tokens_seen": 88943304,
      "step": 1588
    },
    {
      "epoch": 3.538975501113586,
      "grad_norm": 19.53682518005371,
      "learning_rate": 1e-06,
      "loss": 0.7223,
      "num_input_tokens_seen": 89000188,
      "step": 1589
    },
    {
      "epoch": 3.538975501113586,
      "loss": 0.6042957305908203,
      "loss_ce": 0.00029182256548665464,
      "loss_iou": 0.24609375,
      "loss_num": 0.0224609375,
      "loss_xval": 0.60546875,
      "num_input_tokens_seen": 89000188,
      "step": 1589
    },
    {
      "epoch": 3.541202672605791,
      "grad_norm": 16.928892135620117,
      "learning_rate": 1e-06,
      "loss": 1.0069,
      "num_input_tokens_seen": 89055864,
      "step": 1590
    },
    {
      "epoch": 3.541202672605791,
      "loss": 1.0560284852981567,
      "loss_ce": 0.0003644293174147606,
      "loss_iou": 0.439453125,
      "loss_num": 0.035400390625,
      "loss_xval": 1.0546875,
      "num_input_tokens_seen": 89055864,
      "step": 1590
    },
    {
      "epoch": 3.5434298440979957,
      "grad_norm": 21.396242141723633,
      "learning_rate": 1e-06,
      "loss": 0.7472,
      "num_input_tokens_seen": 89114124,
      "step": 1591
    },
    {
      "epoch": 3.5434298440979957,
      "loss": 0.6990212202072144,
      "loss_ce": 0.00029076545615680516,
      "loss_iou": 0.314453125,
      "loss_num": 0.013916015625,
      "loss_xval": 0.69921875,
      "num_input_tokens_seen": 89114124,
      "step": 1591
    },
    {
      "epoch": 3.5456570155902005,
      "grad_norm": 14.654166221618652,
      "learning_rate": 1e-06,
      "loss": 1.0152,
      "num_input_tokens_seen": 89170248,
      "step": 1592
    },
    {
      "epoch": 3.5456570155902005,
      "loss": 0.6828731894493103,
      "loss_ce": 0.0002559710410423577,
      "loss_iou": 0.263671875,
      "loss_num": 0.031494140625,
      "loss_xval": 0.68359375,
      "num_input_tokens_seen": 89170248,
      "step": 1592
    },
    {
      "epoch": 3.5478841870824054,
      "grad_norm": 15.999966621398926,
      "learning_rate": 1e-06,
      "loss": 0.7778,
      "num_input_tokens_seen": 89226156,
      "step": 1593
    },
    {
      "epoch": 3.5478841870824054,
      "loss": 0.9013996124267578,
      "loss_ce": 0.0002766078105196357,
      "loss_iou": 0.3984375,
      "loss_num": 0.0203857421875,
      "loss_xval": 0.90234375,
      "num_input_tokens_seen": 89226156,
      "step": 1593
    },
    {
      "epoch": 3.5501113585746102,
      "grad_norm": 39.26863479614258,
      "learning_rate": 1e-06,
      "loss": 1.2987,
      "num_input_tokens_seen": 89281032,
      "step": 1594
    },
    {
      "epoch": 3.5501113585746102,
      "loss": 1.2463330030441284,
      "loss_ce": 0.00048335548490285873,
      "loss_iou": 0.466796875,
      "loss_num": 0.06298828125,
      "loss_xval": 1.2421875,
      "num_input_tokens_seen": 89281032,
      "step": 1594
    },
    {
      "epoch": 3.552338530066815,
      "grad_norm": 18.40961456298828,
      "learning_rate": 1e-06,
      "loss": 1.1405,
      "num_input_tokens_seen": 89338460,
      "step": 1595
    },
    {
      "epoch": 3.552338530066815,
      "loss": 1.0916733741760254,
      "loss_ce": 0.000364768726285547,
      "loss_iou": 0.466796875,
      "loss_num": 0.031494140625,
      "loss_xval": 1.09375,
      "num_input_tokens_seen": 89338460,
      "step": 1595
    },
    {
      "epoch": 3.55456570155902,
      "grad_norm": 21.472373962402344,
      "learning_rate": 1e-06,
      "loss": 0.9278,
      "num_input_tokens_seen": 89396420,
      "step": 1596
    },
    {
      "epoch": 3.55456570155902,
      "loss": 1.1413955688476562,
      "loss_ce": 0.0002824011608026922,
      "loss_iou": 0.474609375,
      "loss_num": 0.03857421875,
      "loss_xval": 1.140625,
      "num_input_tokens_seen": 89396420,
      "step": 1596
    },
    {
      "epoch": 3.556792873051225,
      "grad_norm": 14.41667366027832,
      "learning_rate": 1e-06,
      "loss": 0.8159,
      "num_input_tokens_seen": 89450448,
      "step": 1597
    },
    {
      "epoch": 3.556792873051225,
      "loss": 1.1159937381744385,
      "loss_ce": 0.00027105180197395384,
      "loss_iou": 0.447265625,
      "loss_num": 0.044677734375,
      "loss_xval": 1.1171875,
      "num_input_tokens_seen": 89450448,
      "step": 1597
    },
    {
      "epoch": 3.5590200445434297,
      "grad_norm": 20.48485565185547,
      "learning_rate": 1e-06,
      "loss": 1.0086,
      "num_input_tokens_seen": 89508572,
      "step": 1598
    },
    {
      "epoch": 3.5590200445434297,
      "loss": 0.9943912029266357,
      "loss_ce": 0.00025058950996026397,
      "loss_iou": 0.41796875,
      "loss_num": 0.031982421875,
      "loss_xval": 0.9921875,
      "num_input_tokens_seen": 89508572,
      "step": 1598
    },
    {
      "epoch": 3.5612472160356345,
      "grad_norm": 33.4490852355957,
      "learning_rate": 1e-06,
      "loss": 1.206,
      "num_input_tokens_seen": 89561964,
      "step": 1599
    },
    {
      "epoch": 3.5612472160356345,
      "loss": 1.290135383605957,
      "loss_ce": 0.0003403525915928185,
      "loss_iou": 0.50390625,
      "loss_num": 0.055908203125,
      "loss_xval": 1.2890625,
      "num_input_tokens_seen": 89561964,
      "step": 1599
    },
    {
      "epoch": 3.5634743875278394,
      "grad_norm": 14.543143272399902,
      "learning_rate": 1e-06,
      "loss": 0.748,
      "num_input_tokens_seen": 89616600,
      "step": 1600
    },
    {
      "epoch": 3.5634743875278394,
      "loss": 0.8427092432975769,
      "loss_ce": 0.0009123453055508435,
      "loss_iou": 0.328125,
      "loss_num": 0.036865234375,
      "loss_xval": 0.84375,
      "num_input_tokens_seen": 89616600,
      "step": 1600
    },
    {
      "epoch": 3.5657015590200447,
      "grad_norm": 16.943578720092773,
      "learning_rate": 1e-06,
      "loss": 0.8685,
      "num_input_tokens_seen": 89672620,
      "step": 1601
    },
    {
      "epoch": 3.5657015590200447,
      "loss": 0.8163316249847412,
      "loss_ce": 0.00041370143298991024,
      "loss_iou": 0.337890625,
      "loss_num": 0.028076171875,
      "loss_xval": 0.81640625,
      "num_input_tokens_seen": 89672620,
      "step": 1601
    },
    {
      "epoch": 3.5679287305122496,
      "grad_norm": 16.56560707092285,
      "learning_rate": 1e-06,
      "loss": 0.8267,
      "num_input_tokens_seen": 89727488,
      "step": 1602
    },
    {
      "epoch": 3.5679287305122496,
      "loss": 0.9919605255126953,
      "loss_ce": 0.000261296343524009,
      "loss_iou": 0.423828125,
      "loss_num": 0.0291748046875,
      "loss_xval": 0.9921875,
      "num_input_tokens_seen": 89727488,
      "step": 1602
    },
    {
      "epoch": 3.5701559020044544,
      "grad_norm": 17.976058959960938,
      "learning_rate": 1e-06,
      "loss": 0.8624,
      "num_input_tokens_seen": 89781992,
      "step": 1603
    },
    {
      "epoch": 3.5701559020044544,
      "loss": 0.9566484093666077,
      "loss_ce": 0.0003496097633615136,
      "loss_iou": 0.40234375,
      "loss_num": 0.030029296875,
      "loss_xval": 0.95703125,
      "num_input_tokens_seen": 89781992,
      "step": 1603
    },
    {
      "epoch": 3.5723830734966593,
      "grad_norm": 21.834300994873047,
      "learning_rate": 1e-06,
      "loss": 1.0587,
      "num_input_tokens_seen": 89837948,
      "step": 1604
    },
    {
      "epoch": 3.5723830734966593,
      "loss": 1.1624748706817627,
      "loss_ce": 0.00036550615914165974,
      "loss_iou": 0.50390625,
      "loss_num": 0.0302734375,
      "loss_xval": 1.1640625,
      "num_input_tokens_seen": 89837948,
      "step": 1604
    },
    {
      "epoch": 3.574610244988864,
      "grad_norm": 33.87417221069336,
      "learning_rate": 1e-06,
      "loss": 0.9384,
      "num_input_tokens_seen": 89892432,
      "step": 1605
    },
    {
      "epoch": 3.574610244988864,
      "loss": 1.0320757627487183,
      "loss_ce": 0.0003374941006768495,
      "loss_iou": 0.4453125,
      "loss_num": 0.0284423828125,
      "loss_xval": 1.03125,
      "num_input_tokens_seen": 89892432,
      "step": 1605
    },
    {
      "epoch": 3.576837416481069,
      "grad_norm": 31.87815284729004,
      "learning_rate": 1e-06,
      "loss": 1.0271,
      "num_input_tokens_seen": 89949104,
      "step": 1606
    },
    {
      "epoch": 3.576837416481069,
      "loss": 0.6741030216217041,
      "loss_ce": 0.0002749357954598963,
      "loss_iou": 0.302734375,
      "loss_num": 0.01397705078125,
      "loss_xval": 0.671875,
      "num_input_tokens_seen": 89949104,
      "step": 1606
    },
    {
      "epoch": 3.579064587973274,
      "grad_norm": 19.64597511291504,
      "learning_rate": 1e-06,
      "loss": 0.688,
      "num_input_tokens_seen": 90007552,
      "step": 1607
    },
    {
      "epoch": 3.579064587973274,
      "loss": 0.7588503956794739,
      "loss_ce": 0.0003054735716432333,
      "loss_iou": 0.345703125,
      "loss_num": 0.0135498046875,
      "loss_xval": 0.7578125,
      "num_input_tokens_seen": 90007552,
      "step": 1607
    },
    {
      "epoch": 3.5812917594654787,
      "grad_norm": 16.6953067779541,
      "learning_rate": 1e-06,
      "loss": 0.841,
      "num_input_tokens_seen": 90065300,
      "step": 1608
    },
    {
      "epoch": 3.5812917594654787,
      "loss": 1.0172404050827026,
      "loss_ce": 0.00039471167838200927,
      "loss_iou": 0.42578125,
      "loss_num": 0.03271484375,
      "loss_xval": 1.015625,
      "num_input_tokens_seen": 90065300,
      "step": 1608
    },
    {
      "epoch": 3.5835189309576836,
      "grad_norm": 18.6095027923584,
      "learning_rate": 1e-06,
      "loss": 0.8472,
      "num_input_tokens_seen": 90117120,
      "step": 1609
    },
    {
      "epoch": 3.5835189309576836,
      "loss": 0.7410043478012085,
      "loss_ce": 0.00028169897268526256,
      "loss_iou": 0.310546875,
      "loss_num": 0.023681640625,
      "loss_xval": 0.7421875,
      "num_input_tokens_seen": 90117120,
      "step": 1609
    },
    {
      "epoch": 3.585746102449889,
      "grad_norm": 19.11724090576172,
      "learning_rate": 1e-06,
      "loss": 0.9566,
      "num_input_tokens_seen": 90171580,
      "step": 1610
    },
    {
      "epoch": 3.585746102449889,
      "loss": 0.7324100136756897,
      "loss_ce": 0.0004764144541695714,
      "loss_iou": 0.30859375,
      "loss_num": 0.0233154296875,
      "loss_xval": 0.73046875,
      "num_input_tokens_seen": 90171580,
      "step": 1610
    },
    {
      "epoch": 3.5879732739420938,
      "grad_norm": 47.91438674926758,
      "learning_rate": 1e-06,
      "loss": 0.9755,
      "num_input_tokens_seen": 90227144,
      "step": 1611
    },
    {
      "epoch": 3.5879732739420938,
      "loss": 1.2943049669265747,
      "loss_ce": 0.000359668250894174,
      "loss_iou": 0.54296875,
      "loss_num": 0.04248046875,
      "loss_xval": 1.296875,
      "num_input_tokens_seen": 90227144,
      "step": 1611
    },
    {
      "epoch": 3.5902004454342986,
      "grad_norm": 14.350851058959961,
      "learning_rate": 1e-06,
      "loss": 0.7686,
      "num_input_tokens_seen": 90285220,
      "step": 1612
    },
    {
      "epoch": 3.5902004454342986,
      "loss": 0.6935353875160217,
      "loss_ce": 0.0004201638512313366,
      "loss_iou": 0.3046875,
      "loss_num": 0.0166015625,
      "loss_xval": 0.69140625,
      "num_input_tokens_seen": 90285220,
      "step": 1612
    },
    {
      "epoch": 3.5924276169265035,
      "grad_norm": 87.72432708740234,
      "learning_rate": 1e-06,
      "loss": 0.9391,
      "num_input_tokens_seen": 90341444,
      "step": 1613
    },
    {
      "epoch": 3.5924276169265035,
      "loss": 0.8298938274383545,
      "loss_ce": 0.00030400152900256217,
      "loss_iou": 0.345703125,
      "loss_num": 0.0274658203125,
      "loss_xval": 0.828125,
      "num_input_tokens_seen": 90341444,
      "step": 1613
    },
    {
      "epoch": 3.5946547884187083,
      "grad_norm": 18.14371681213379,
      "learning_rate": 1e-06,
      "loss": 0.8685,
      "num_input_tokens_seen": 90398880,
      "step": 1614
    },
    {
      "epoch": 3.5946547884187083,
      "loss": 0.9817934632301331,
      "loss_ce": 0.0003481835883576423,
      "loss_iou": 0.400390625,
      "loss_num": 0.0361328125,
      "loss_xval": 0.98046875,
      "num_input_tokens_seen": 90398880,
      "step": 1614
    },
    {
      "epoch": 3.596881959910913,
      "grad_norm": 18.2519588470459,
      "learning_rate": 1e-06,
      "loss": 0.8095,
      "num_input_tokens_seen": 90453864,
      "step": 1615
    },
    {
      "epoch": 3.596881959910913,
      "loss": 0.7134523391723633,
      "loss_ce": 0.00031755882082507014,
      "loss_iou": 0.31640625,
      "loss_num": 0.016357421875,
      "loss_xval": 0.71484375,
      "num_input_tokens_seen": 90453864,
      "step": 1615
    },
    {
      "epoch": 3.599109131403118,
      "grad_norm": 11.235553741455078,
      "learning_rate": 1e-06,
      "loss": 0.9908,
      "num_input_tokens_seen": 90508336,
      "step": 1616
    },
    {
      "epoch": 3.599109131403118,
      "loss": 1.2815724611282349,
      "loss_ce": 0.0003224927349947393,
      "loss_iou": 0.5390625,
      "loss_num": 0.041015625,
      "loss_xval": 1.28125,
      "num_input_tokens_seen": 90508336,
      "step": 1616
    },
    {
      "epoch": 3.601336302895323,
      "grad_norm": 19.534269332885742,
      "learning_rate": 1e-06,
      "loss": 0.9323,
      "num_input_tokens_seen": 90565836,
      "step": 1617
    },
    {
      "epoch": 3.601336302895323,
      "loss": 1.1978850364685059,
      "loss_ce": 0.0006193737499415874,
      "loss_iou": 0.435546875,
      "loss_num": 0.06494140625,
      "loss_xval": 1.1953125,
      "num_input_tokens_seen": 90565836,
      "step": 1617
    },
    {
      "epoch": 3.6035634743875278,
      "grad_norm": 16.99414825439453,
      "learning_rate": 1e-06,
      "loss": 0.791,
      "num_input_tokens_seen": 90621756,
      "step": 1618
    },
    {
      "epoch": 3.6035634743875278,
      "loss": 0.6094855070114136,
      "loss_ce": 0.0003546725492924452,
      "loss_iou": 0.265625,
      "loss_num": 0.015869140625,
      "loss_xval": 0.609375,
      "num_input_tokens_seen": 90621756,
      "step": 1618
    },
    {
      "epoch": 3.6057906458797326,
      "grad_norm": 12.391351699829102,
      "learning_rate": 1e-06,
      "loss": 0.736,
      "num_input_tokens_seen": 90679300,
      "step": 1619
    },
    {
      "epoch": 3.6057906458797326,
      "loss": 0.6363743543624878,
      "loss_ce": 0.0002659702149685472,
      "loss_iou": 0.26953125,
      "loss_num": 0.01904296875,
      "loss_xval": 0.63671875,
      "num_input_tokens_seen": 90679300,
      "step": 1619
    },
    {
      "epoch": 3.6080178173719375,
      "grad_norm": 19.21337127685547,
      "learning_rate": 1e-06,
      "loss": 0.9495,
      "num_input_tokens_seen": 90738552,
      "step": 1620
    },
    {
      "epoch": 3.6080178173719375,
      "loss": 0.9752398729324341,
      "loss_ce": 0.00038639939157292247,
      "loss_iou": 0.41796875,
      "loss_num": 0.0279541015625,
      "loss_xval": 0.9765625,
      "num_input_tokens_seen": 90738552,
      "step": 1620
    },
    {
      "epoch": 3.6102449888641424,
      "grad_norm": 14.483777046203613,
      "learning_rate": 1e-06,
      "loss": 0.7827,
      "num_input_tokens_seen": 90796600,
      "step": 1621
    },
    {
      "epoch": 3.6102449888641424,
      "loss": 0.7163407802581787,
      "loss_ce": 0.00027631394914351404,
      "loss_iou": 0.298828125,
      "loss_num": 0.0235595703125,
      "loss_xval": 0.71484375,
      "num_input_tokens_seen": 90796600,
      "step": 1621
    },
    {
      "epoch": 3.612472160356347,
      "grad_norm": 18.561866760253906,
      "learning_rate": 1e-06,
      "loss": 0.9083,
      "num_input_tokens_seen": 90852108,
      "step": 1622
    },
    {
      "epoch": 3.612472160356347,
      "loss": 0.7986100912094116,
      "loss_ce": 0.00027019885601475835,
      "loss_iou": 0.345703125,
      "loss_num": 0.021240234375,
      "loss_xval": 0.796875,
      "num_input_tokens_seen": 90852108,
      "step": 1622
    },
    {
      "epoch": 3.614699331848552,
      "grad_norm": 13.72581672668457,
      "learning_rate": 1e-06,
      "loss": 0.9149,
      "num_input_tokens_seen": 90909780,
      "step": 1623
    },
    {
      "epoch": 3.614699331848552,
      "loss": 0.8790796995162964,
      "loss_ce": 0.0002955020754598081,
      "loss_iou": 0.3515625,
      "loss_num": 0.035400390625,
      "loss_xval": 0.87890625,
      "num_input_tokens_seen": 90909780,
      "step": 1623
    },
    {
      "epoch": 3.6169265033407574,
      "grad_norm": 19.92896842956543,
      "learning_rate": 1e-06,
      "loss": 0.8137,
      "num_input_tokens_seen": 90964636,
      "step": 1624
    },
    {
      "epoch": 3.6169265033407574,
      "loss": 0.4973217248916626,
      "loss_ce": 0.00025139018543995917,
      "loss_iou": 0.2001953125,
      "loss_num": 0.0194091796875,
      "loss_xval": 0.49609375,
      "num_input_tokens_seen": 90964636,
      "step": 1624
    },
    {
      "epoch": 3.6191536748329622,
      "grad_norm": 24.191030502319336,
      "learning_rate": 1e-06,
      "loss": 0.8298,
      "num_input_tokens_seen": 91020328,
      "step": 1625
    },
    {
      "epoch": 3.6191536748329622,
      "loss": 0.8796758651733398,
      "loss_ce": 0.0002813548780977726,
      "loss_iou": 0.34765625,
      "loss_num": 0.036865234375,
      "loss_xval": 0.87890625,
      "num_input_tokens_seen": 91020328,
      "step": 1625
    },
    {
      "epoch": 3.621380846325167,
      "grad_norm": 18.71809959411621,
      "learning_rate": 1e-06,
      "loss": 0.7477,
      "num_input_tokens_seen": 91076856,
      "step": 1626
    },
    {
      "epoch": 3.621380846325167,
      "loss": 0.6939845085144043,
      "loss_ce": 0.0003809723712038249,
      "loss_iou": 0.27734375,
      "loss_num": 0.027587890625,
      "loss_xval": 0.6953125,
      "num_input_tokens_seen": 91076856,
      "step": 1626
    },
    {
      "epoch": 3.623608017817372,
      "grad_norm": 15.862205505371094,
      "learning_rate": 1e-06,
      "loss": 0.8313,
      "num_input_tokens_seen": 91136100,
      "step": 1627
    },
    {
      "epoch": 3.623608017817372,
      "loss": 0.8305182456970215,
      "loss_ce": 0.0007453373400494456,
      "loss_iou": 0.3359375,
      "loss_num": 0.0311279296875,
      "loss_xval": 0.828125,
      "num_input_tokens_seen": 91136100,
      "step": 1627
    },
    {
      "epoch": 3.625835189309577,
      "grad_norm": 23.61324691772461,
      "learning_rate": 1e-06,
      "loss": 0.8207,
      "num_input_tokens_seen": 91191620,
      "step": 1628
    },
    {
      "epoch": 3.625835189309577,
      "loss": 0.6336687207221985,
      "loss_ce": 0.0002458438102621585,
      "loss_iou": 0.27734375,
      "loss_num": 0.0157470703125,
      "loss_xval": 0.6328125,
      "num_input_tokens_seen": 91191620,
      "step": 1628
    },
    {
      "epoch": 3.6280623608017817,
      "grad_norm": 18.18329620361328,
      "learning_rate": 1e-06,
      "loss": 0.9281,
      "num_input_tokens_seen": 91250588,
      "step": 1629
    },
    {
      "epoch": 3.6280623608017817,
      "loss": 1.0516316890716553,
      "loss_ce": 0.0003620931529439986,
      "loss_iou": 0.421875,
      "loss_num": 0.04150390625,
      "loss_xval": 1.0546875,
      "num_input_tokens_seen": 91250588,
      "step": 1629
    },
    {
      "epoch": 3.6302895322939865,
      "grad_norm": 18.133012771606445,
      "learning_rate": 1e-06,
      "loss": 0.5779,
      "num_input_tokens_seen": 91308444,
      "step": 1630
    },
    {
      "epoch": 3.6302895322939865,
      "loss": 0.7188634872436523,
      "loss_ce": 0.0002355621982133016,
      "loss_iou": 0.29296875,
      "loss_num": 0.026611328125,
      "loss_xval": 0.71875,
      "num_input_tokens_seen": 91308444,
      "step": 1630
    },
    {
      "epoch": 3.6325167037861914,
      "grad_norm": 21.32465934753418,
      "learning_rate": 1e-06,
      "loss": 0.7386,
      "num_input_tokens_seen": 91363304,
      "step": 1631
    },
    {
      "epoch": 3.6325167037861914,
      "loss": 0.7637060284614563,
      "loss_ce": 0.0002782873052638024,
      "loss_iou": 0.337890625,
      "loss_num": 0.0172119140625,
      "loss_xval": 0.76171875,
      "num_input_tokens_seen": 91363304,
      "step": 1631
    },
    {
      "epoch": 3.6347438752783967,
      "grad_norm": 15.17698860168457,
      "learning_rate": 1e-06,
      "loss": 1.0197,
      "num_input_tokens_seen": 91417932,
      "step": 1632
    },
    {
      "epoch": 3.6347438752783967,
      "loss": 0.6608976125717163,
      "loss_ce": 0.0002531085046939552,
      "loss_iou": 0.244140625,
      "loss_num": 0.034423828125,
      "loss_xval": 0.66015625,
      "num_input_tokens_seen": 91417932,
      "step": 1632
    },
    {
      "epoch": 3.6369710467706016,
      "grad_norm": 19.845325469970703,
      "learning_rate": 1e-06,
      "loss": 0.737,
      "num_input_tokens_seen": 91474376,
      "step": 1633
    },
    {
      "epoch": 3.6369710467706016,
      "loss": 0.8098554611206055,
      "loss_ce": 0.0002851108438335359,
      "loss_iou": 0.3359375,
      "loss_num": 0.02783203125,
      "loss_xval": 0.80859375,
      "num_input_tokens_seen": 91474376,
      "step": 1633
    },
    {
      "epoch": 3.6391982182628064,
      "grad_norm": 18.74712371826172,
      "learning_rate": 1e-06,
      "loss": 0.8073,
      "num_input_tokens_seen": 91529408,
      "step": 1634
    },
    {
      "epoch": 3.6391982182628064,
      "loss": 0.799332857131958,
      "loss_ce": 0.0002606004709377885,
      "loss_iou": 0.33984375,
      "loss_num": 0.024169921875,
      "loss_xval": 0.80078125,
      "num_input_tokens_seen": 91529408,
      "step": 1634
    },
    {
      "epoch": 3.6414253897550113,
      "grad_norm": 16.073997497558594,
      "learning_rate": 1e-06,
      "loss": 0.6983,
      "num_input_tokens_seen": 91584964,
      "step": 1635
    },
    {
      "epoch": 3.6414253897550113,
      "loss": 0.5875762104988098,
      "loss_ce": 0.000418026524130255,
      "loss_iou": 0.271484375,
      "loss_num": 0.0091552734375,
      "loss_xval": 0.5859375,
      "num_input_tokens_seen": 91584964,
      "step": 1635
    },
    {
      "epoch": 3.643652561247216,
      "grad_norm": 26.88153076171875,
      "learning_rate": 1e-06,
      "loss": 0.737,
      "num_input_tokens_seen": 91642164,
      "step": 1636
    },
    {
      "epoch": 3.643652561247216,
      "loss": 0.8176705837249756,
      "loss_ce": 0.00028783181915059686,
      "loss_iou": 0.33203125,
      "loss_num": 0.03076171875,
      "loss_xval": 0.81640625,
      "num_input_tokens_seen": 91642164,
      "step": 1636
    },
    {
      "epoch": 3.645879732739421,
      "grad_norm": 29.69915771484375,
      "learning_rate": 1e-06,
      "loss": 1.0198,
      "num_input_tokens_seen": 91697608,
      "step": 1637
    },
    {
      "epoch": 3.645879732739421,
      "loss": 0.971588134765625,
      "loss_ce": 0.0003966961521655321,
      "loss_iou": 0.39453125,
      "loss_num": 0.0361328125,
      "loss_xval": 0.97265625,
      "num_input_tokens_seen": 91697608,
      "step": 1637
    },
    {
      "epoch": 3.648106904231626,
      "grad_norm": 14.602875709533691,
      "learning_rate": 1e-06,
      "loss": 0.7628,
      "num_input_tokens_seen": 91755904,
      "step": 1638
    },
    {
      "epoch": 3.648106904231626,
      "loss": 0.7990860342979431,
      "loss_ce": 0.0002579537685960531,
      "loss_iou": 0.34765625,
      "loss_num": 0.0208740234375,
      "loss_xval": 0.796875,
      "num_input_tokens_seen": 91755904,
      "step": 1638
    },
    {
      "epoch": 3.6503340757238307,
      "grad_norm": 19.639419555664062,
      "learning_rate": 1e-06,
      "loss": 0.9729,
      "num_input_tokens_seen": 91813308,
      "step": 1639
    },
    {
      "epoch": 3.6503340757238307,
      "loss": 0.7160975933074951,
      "loss_ce": 0.0002772239677142352,
      "loss_iou": 0.30078125,
      "loss_num": 0.0225830078125,
      "loss_xval": 0.71484375,
      "num_input_tokens_seen": 91813308,
      "step": 1639
    },
    {
      "epoch": 3.6525612472160356,
      "grad_norm": 26.63214683532715,
      "learning_rate": 1e-06,
      "loss": 0.7171,
      "num_input_tokens_seen": 91870888,
      "step": 1640
    },
    {
      "epoch": 3.6525612472160356,
      "loss": 0.6993926763534546,
      "loss_ce": 0.0004180770483799279,
      "loss_iou": 0.30859375,
      "loss_num": 0.0166015625,
      "loss_xval": 0.69921875,
      "num_input_tokens_seen": 91870888,
      "step": 1640
    },
    {
      "epoch": 3.6547884187082404,
      "grad_norm": 24.945594787597656,
      "learning_rate": 1e-06,
      "loss": 0.7824,
      "num_input_tokens_seen": 91928644,
      "step": 1641
    },
    {
      "epoch": 3.6547884187082404,
      "loss": 0.8066414594650269,
      "loss_ce": 0.00036706856917589903,
      "loss_iou": 0.322265625,
      "loss_num": 0.0322265625,
      "loss_xval": 0.8046875,
      "num_input_tokens_seen": 91928644,
      "step": 1641
    },
    {
      "epoch": 3.6570155902004453,
      "grad_norm": 28.210344314575195,
      "learning_rate": 1e-06,
      "loss": 0.7365,
      "num_input_tokens_seen": 91982148,
      "step": 1642
    },
    {
      "epoch": 3.6570155902004453,
      "loss": 0.6948254704475403,
      "loss_ce": 0.0014660632004961371,
      "loss_iou": 0.3046875,
      "loss_num": 0.016357421875,
      "loss_xval": 0.6953125,
      "num_input_tokens_seen": 91982148,
      "step": 1642
    },
    {
      "epoch": 3.65924276169265,
      "grad_norm": 83.1476821899414,
      "learning_rate": 1e-06,
      "loss": 0.8233,
      "num_input_tokens_seen": 92038604,
      "step": 1643
    },
    {
      "epoch": 3.65924276169265,
      "loss": 1.0671532154083252,
      "loss_ce": 0.000502894283272326,
      "loss_iou": 0.453125,
      "loss_num": 0.031982421875,
      "loss_xval": 1.0703125,
      "num_input_tokens_seen": 92038604,
      "step": 1643
    },
    {
      "epoch": 3.661469933184855,
      "grad_norm": 19.53736114501953,
      "learning_rate": 1e-06,
      "loss": 0.6483,
      "num_input_tokens_seen": 92094664,
      "step": 1644
    },
    {
      "epoch": 3.661469933184855,
      "loss": 0.4442579448223114,
      "loss_ce": 0.0002882296103052795,
      "loss_iou": 0.1650390625,
      "loss_num": 0.022705078125,
      "loss_xval": 0.443359375,
      "num_input_tokens_seen": 92094664,
      "step": 1644
    },
    {
      "epoch": 3.66369710467706,
      "grad_norm": 30.144088745117188,
      "learning_rate": 1e-06,
      "loss": 0.8099,
      "num_input_tokens_seen": 92149264,
      "step": 1645
    },
    {
      "epoch": 3.66369710467706,
      "loss": 0.6731590032577515,
      "loss_ce": 0.0003074193373322487,
      "loss_iou": 0.275390625,
      "loss_num": 0.0242919921875,
      "loss_xval": 0.671875,
      "num_input_tokens_seen": 92149264,
      "step": 1645
    },
    {
      "epoch": 3.665924276169265,
      "grad_norm": 24.664552688598633,
      "learning_rate": 1e-06,
      "loss": 0.9853,
      "num_input_tokens_seen": 92207820,
      "step": 1646
    },
    {
      "epoch": 3.665924276169265,
      "loss": 0.8198744654655457,
      "loss_ce": 0.000294416124233976,
      "loss_iou": 0.3515625,
      "loss_num": 0.0235595703125,
      "loss_xval": 0.8203125,
      "num_input_tokens_seen": 92207820,
      "step": 1646
    },
    {
      "epoch": 3.66815144766147,
      "grad_norm": 17.04414939880371,
      "learning_rate": 1e-06,
      "loss": 0.6132,
      "num_input_tokens_seen": 92263900,
      "step": 1647
    },
    {
      "epoch": 3.66815144766147,
      "loss": 0.5955994725227356,
      "loss_ce": 0.0003235829062759876,
      "loss_iou": 0.2451171875,
      "loss_num": 0.0211181640625,
      "loss_xval": 0.59375,
      "num_input_tokens_seen": 92263900,
      "step": 1647
    },
    {
      "epoch": 3.670378619153675,
      "grad_norm": 26.547033309936523,
      "learning_rate": 1e-06,
      "loss": 0.7967,
      "num_input_tokens_seen": 92319188,
      "step": 1648
    },
    {
      "epoch": 3.670378619153675,
      "loss": 0.6533565521240234,
      "loss_ce": 0.00028039264725521207,
      "loss_iou": 0.291015625,
      "loss_num": 0.01446533203125,
      "loss_xval": 0.65234375,
      "num_input_tokens_seen": 92319188,
      "step": 1648
    },
    {
      "epoch": 3.6726057906458798,
      "grad_norm": 20.542510986328125,
      "learning_rate": 1e-06,
      "loss": 0.804,
      "num_input_tokens_seen": 92375980,
      "step": 1649
    },
    {
      "epoch": 3.6726057906458798,
      "loss": 0.8124038577079773,
      "loss_ce": 0.00039214000571519136,
      "loss_iou": 0.30859375,
      "loss_num": 0.0390625,
      "loss_xval": 0.8125,
      "num_input_tokens_seen": 92375980,
      "step": 1649
    },
    {
      "epoch": 3.6748329621380846,
      "grad_norm": 77.64947509765625,
      "learning_rate": 1e-06,
      "loss": 1.049,
      "num_input_tokens_seen": 92427792,
      "step": 1650
    },
    {
      "epoch": 3.6748329621380846,
      "loss": 0.7283252477645874,
      "loss_ce": 0.00029791187262162566,
      "loss_iou": 0.31640625,
      "loss_num": 0.019287109375,
      "loss_xval": 0.7265625,
      "num_input_tokens_seen": 92427792,
      "step": 1650
    },
    {
      "epoch": 3.6770601336302895,
      "grad_norm": 20.271570205688477,
      "learning_rate": 1e-06,
      "loss": 0.7777,
      "num_input_tokens_seen": 92484212,
      "step": 1651
    },
    {
      "epoch": 3.6770601336302895,
      "loss": 0.6192773580551147,
      "loss_ce": 0.00025881448527798057,
      "loss_iou": 0.267578125,
      "loss_num": 0.0172119140625,
      "loss_xval": 0.6171875,
      "num_input_tokens_seen": 92484212,
      "step": 1651
    },
    {
      "epoch": 3.6792873051224944,
      "grad_norm": 15.280570030212402,
      "learning_rate": 1e-06,
      "loss": 0.7805,
      "num_input_tokens_seen": 92538032,
      "step": 1652
    },
    {
      "epoch": 3.6792873051224944,
      "loss": 0.6725317239761353,
      "loss_ce": 0.00029047008138149977,
      "loss_iou": 0.283203125,
      "loss_num": 0.0211181640625,
      "loss_xval": 0.671875,
      "num_input_tokens_seen": 92538032,
      "step": 1652
    },
    {
      "epoch": 3.681514476614699,
      "grad_norm": 17.816753387451172,
      "learning_rate": 1e-06,
      "loss": 1.012,
      "num_input_tokens_seen": 92592792,
      "step": 1653
    },
    {
      "epoch": 3.681514476614699,
      "loss": 0.8765089511871338,
      "loss_ce": 0.0002882035914808512,
      "loss_iou": 0.384765625,
      "loss_num": 0.0213623046875,
      "loss_xval": 0.875,
      "num_input_tokens_seen": 92592792,
      "step": 1653
    },
    {
      "epoch": 3.683741648106904,
      "grad_norm": 34.85762405395508,
      "learning_rate": 1e-06,
      "loss": 0.9599,
      "num_input_tokens_seen": 92649552,
      "step": 1654
    },
    {
      "epoch": 3.683741648106904,
      "loss": 1.0684025287628174,
      "loss_ce": 0.00028726100572384894,
      "loss_iou": 0.453125,
      "loss_num": 0.0322265625,
      "loss_xval": 1.0703125,
      "num_input_tokens_seen": 92649552,
      "step": 1654
    },
    {
      "epoch": 3.6859688195991094,
      "grad_norm": 16.367671966552734,
      "learning_rate": 1e-06,
      "loss": 0.7476,
      "num_input_tokens_seen": 92705912,
      "step": 1655
    },
    {
      "epoch": 3.6859688195991094,
      "loss": 0.467061311006546,
      "loss_ce": 0.0002644392370712012,
      "loss_iou": 0.2080078125,
      "loss_num": 0.01025390625,
      "loss_xval": 0.466796875,
      "num_input_tokens_seen": 92705912,
      "step": 1655
    },
    {
      "epoch": 3.6881959910913142,
      "grad_norm": 28.718902587890625,
      "learning_rate": 1e-06,
      "loss": 0.8724,
      "num_input_tokens_seen": 92764576,
      "step": 1656
    },
    {
      "epoch": 3.6881959910913142,
      "loss": 0.7795916795730591,
      "loss_ce": 0.0002947830653283745,
      "loss_iou": 0.328125,
      "loss_num": 0.0244140625,
      "loss_xval": 0.78125,
      "num_input_tokens_seen": 92764576,
      "step": 1656
    },
    {
      "epoch": 3.690423162583519,
      "grad_norm": 25.93492317199707,
      "learning_rate": 1e-06,
      "loss": 0.9498,
      "num_input_tokens_seen": 92818528,
      "step": 1657
    },
    {
      "epoch": 3.690423162583519,
      "loss": 1.08326256275177,
      "loss_ce": 0.00025476596783846617,
      "loss_iou": 0.474609375,
      "loss_num": 0.026611328125,
      "loss_xval": 1.0859375,
      "num_input_tokens_seen": 92818528,
      "step": 1657
    },
    {
      "epoch": 3.692650334075724,
      "grad_norm": 16.00531768798828,
      "learning_rate": 1e-06,
      "loss": 1.0849,
      "num_input_tokens_seen": 92874168,
      "step": 1658
    },
    {
      "epoch": 3.692650334075724,
      "loss": 0.8632932901382446,
      "loss_ce": 0.0002562026202213019,
      "loss_iou": 0.384765625,
      "loss_num": 0.018798828125,
      "loss_xval": 0.86328125,
      "num_input_tokens_seen": 92874168,
      "step": 1658
    },
    {
      "epoch": 3.694877505567929,
      "grad_norm": 48.529300689697266,
      "learning_rate": 1e-06,
      "loss": 0.9435,
      "num_input_tokens_seen": 92926172,
      "step": 1659
    },
    {
      "epoch": 3.694877505567929,
      "loss": 0.9439380168914795,
      "loss_ce": 0.000822762493044138,
      "loss_iou": 0.40234375,
      "loss_num": 0.02783203125,
      "loss_xval": 0.94140625,
      "num_input_tokens_seen": 92926172,
      "step": 1659
    },
    {
      "epoch": 3.6971046770601337,
      "grad_norm": 16.016910552978516,
      "learning_rate": 1e-06,
      "loss": 0.7698,
      "num_input_tokens_seen": 92981500,
      "step": 1660
    },
    {
      "epoch": 3.6971046770601337,
      "loss": 0.7871308326721191,
      "loss_ce": 0.0002656061842571944,
      "loss_iou": 0.341796875,
      "loss_num": 0.0201416015625,
      "loss_xval": 0.78515625,
      "num_input_tokens_seen": 92981500,
      "step": 1660
    },
    {
      "epoch": 3.6993318485523385,
      "grad_norm": 18.289608001708984,
      "learning_rate": 1e-06,
      "loss": 1.0147,
      "num_input_tokens_seen": 93036804,
      "step": 1661
    },
    {
      "epoch": 3.6993318485523385,
      "loss": 1.0226683616638184,
      "loss_ce": 0.00032947887666523457,
      "loss_iou": 0.4296875,
      "loss_num": 0.031982421875,
      "loss_xval": 1.0234375,
      "num_input_tokens_seen": 93036804,
      "step": 1661
    },
    {
      "epoch": 3.7015590200445434,
      "grad_norm": 15.377629280090332,
      "learning_rate": 1e-06,
      "loss": 0.7615,
      "num_input_tokens_seen": 93093888,
      "step": 1662
    },
    {
      "epoch": 3.7015590200445434,
      "loss": 0.8022950887680054,
      "loss_ce": 0.00029318686574697495,
      "loss_iou": 0.328125,
      "loss_num": 0.0291748046875,
      "loss_xval": 0.80078125,
      "num_input_tokens_seen": 93093888,
      "step": 1662
    },
    {
      "epoch": 3.7037861915367483,
      "grad_norm": 16.651514053344727,
      "learning_rate": 1e-06,
      "loss": 0.8527,
      "num_input_tokens_seen": 93151972,
      "step": 1663
    },
    {
      "epoch": 3.7037861915367483,
      "loss": 0.8089163303375244,
      "loss_ce": 0.0003225764958187938,
      "loss_iou": 0.353515625,
      "loss_num": 0.0203857421875,
      "loss_xval": 0.80859375,
      "num_input_tokens_seen": 93151972,
      "step": 1663
    },
    {
      "epoch": 3.706013363028953,
      "grad_norm": 87.34024047851562,
      "learning_rate": 1e-06,
      "loss": 0.9063,
      "num_input_tokens_seen": 93208668,
      "step": 1664
    },
    {
      "epoch": 3.706013363028953,
      "loss": 0.6086684465408325,
      "loss_ce": 0.00026995883672498167,
      "loss_iou": 0.25390625,
      "loss_num": 0.019775390625,
      "loss_xval": 0.609375,
      "num_input_tokens_seen": 93208668,
      "step": 1664
    },
    {
      "epoch": 3.708240534521158,
      "grad_norm": 22.259021759033203,
      "learning_rate": 1e-06,
      "loss": 0.9715,
      "num_input_tokens_seen": 93266120,
      "step": 1665
    },
    {
      "epoch": 3.708240534521158,
      "loss": 1.0723727941513062,
      "loss_ce": 0.0003513463889248669,
      "loss_iou": 0.4453125,
      "loss_num": 0.03662109375,
      "loss_xval": 1.0703125,
      "num_input_tokens_seen": 93266120,
      "step": 1665
    },
    {
      "epoch": 3.710467706013363,
      "grad_norm": 13.237698554992676,
      "learning_rate": 1e-06,
      "loss": 0.7824,
      "num_input_tokens_seen": 93322796,
      "step": 1666
    },
    {
      "epoch": 3.710467706013363,
      "loss": 0.803238034248352,
      "loss_ce": 0.0002595084370113909,
      "loss_iou": 0.33984375,
      "loss_num": 0.0245361328125,
      "loss_xval": 0.8046875,
      "num_input_tokens_seen": 93322796,
      "step": 1666
    },
    {
      "epoch": 3.7126948775055677,
      "grad_norm": 30.929845809936523,
      "learning_rate": 1e-06,
      "loss": 0.9966,
      "num_input_tokens_seen": 93378408,
      "step": 1667
    },
    {
      "epoch": 3.7126948775055677,
      "loss": 0.9339191913604736,
      "loss_ce": 0.0003254116454627365,
      "loss_iou": 0.423828125,
      "loss_num": 0.0172119140625,
      "loss_xval": 0.93359375,
      "num_input_tokens_seen": 93378408,
      "step": 1667
    },
    {
      "epoch": 3.7149220489977726,
      "grad_norm": 17.648576736450195,
      "learning_rate": 1e-06,
      "loss": 0.9418,
      "num_input_tokens_seen": 93435448,
      "step": 1668
    },
    {
      "epoch": 3.7149220489977726,
      "loss": 0.8993349075317383,
      "loss_ce": 0.0013856550212949514,
      "loss_iou": 0.373046875,
      "loss_num": 0.0306396484375,
      "loss_xval": 0.8984375,
      "num_input_tokens_seen": 93435448,
      "step": 1668
    },
    {
      "epoch": 3.717149220489978,
      "grad_norm": 18.18744659423828,
      "learning_rate": 1e-06,
      "loss": 0.5543,
      "num_input_tokens_seen": 93490744,
      "step": 1669
    },
    {
      "epoch": 3.717149220489978,
      "loss": 0.5734395980834961,
      "loss_ce": 0.00044156977673992515,
      "loss_iou": 0.2470703125,
      "loss_num": 0.0157470703125,
      "loss_xval": 0.57421875,
      "num_input_tokens_seen": 93490744,
      "step": 1669
    },
    {
      "epoch": 3.7193763919821827,
      "grad_norm": 26.080154418945312,
      "learning_rate": 1e-06,
      "loss": 0.8468,
      "num_input_tokens_seen": 93546192,
      "step": 1670
    },
    {
      "epoch": 3.7193763919821827,
      "loss": 1.0555299520492554,
      "loss_ce": 0.00035410295822657645,
      "loss_iou": 0.455078125,
      "loss_num": 0.0286865234375,
      "loss_xval": 1.0546875,
      "num_input_tokens_seen": 93546192,
      "step": 1670
    },
    {
      "epoch": 3.7216035634743876,
      "grad_norm": 19.251693725585938,
      "learning_rate": 1e-06,
      "loss": 0.9076,
      "num_input_tokens_seen": 93604348,
      "step": 1671
    },
    {
      "epoch": 3.7216035634743876,
      "loss": 0.8526378273963928,
      "loss_ce": 0.00034295275690965354,
      "loss_iou": 0.361328125,
      "loss_num": 0.0263671875,
      "loss_xval": 0.8515625,
      "num_input_tokens_seen": 93604348,
      "step": 1671
    },
    {
      "epoch": 3.7238307349665924,
      "grad_norm": 15.625646591186523,
      "learning_rate": 1e-06,
      "loss": 0.881,
      "num_input_tokens_seen": 93659012,
      "step": 1672
    },
    {
      "epoch": 3.7238307349665924,
      "loss": 0.9302690625190735,
      "loss_ce": 0.00033744628308340907,
      "loss_iou": 0.3984375,
      "loss_num": 0.0262451171875,
      "loss_xval": 0.9296875,
      "num_input_tokens_seen": 93659012,
      "step": 1672
    },
    {
      "epoch": 3.7260579064587973,
      "grad_norm": 47.977176666259766,
      "learning_rate": 1e-06,
      "loss": 1.0519,
      "num_input_tokens_seen": 93710680,
      "step": 1673
    },
    {
      "epoch": 3.7260579064587973,
      "loss": 0.8930314779281616,
      "loss_ce": 0.0006975086871534586,
      "loss_iou": 0.375,
      "loss_num": 0.028564453125,
      "loss_xval": 0.890625,
      "num_input_tokens_seen": 93710680,
      "step": 1673
    },
    {
      "epoch": 3.728285077951002,
      "grad_norm": 27.666378021240234,
      "learning_rate": 1e-06,
      "loss": 0.8055,
      "num_input_tokens_seen": 93766956,
      "step": 1674
    },
    {
      "epoch": 3.728285077951002,
      "loss": 0.6055097579956055,
      "loss_ce": 0.000285172660369426,
      "loss_iou": 0.2490234375,
      "loss_num": 0.0213623046875,
      "loss_xval": 0.60546875,
      "num_input_tokens_seen": 93766956,
      "step": 1674
    },
    {
      "epoch": 3.730512249443207,
      "grad_norm": 28.261253356933594,
      "learning_rate": 1e-06,
      "loss": 0.6353,
      "num_input_tokens_seen": 93824432,
      "step": 1675
    },
    {
      "epoch": 3.730512249443207,
      "loss": 0.7097867727279663,
      "loss_ce": 0.00031414441764354706,
      "loss_iou": 0.30078125,
      "loss_num": 0.021240234375,
      "loss_xval": 0.7109375,
      "num_input_tokens_seen": 93824432,
      "step": 1675
    },
    {
      "epoch": 3.732739420935412,
      "grad_norm": 17.860763549804688,
      "learning_rate": 1e-06,
      "loss": 0.7996,
      "num_input_tokens_seen": 93882600,
      "step": 1676
    },
    {
      "epoch": 3.732739420935412,
      "loss": 0.6038820743560791,
      "loss_ce": 0.00021385436411947012,
      "loss_iou": 0.24609375,
      "loss_num": 0.022216796875,
      "loss_xval": 0.60546875,
      "num_input_tokens_seen": 93882600,
      "step": 1676
    },
    {
      "epoch": 3.734966592427617,
      "grad_norm": 22.68053436279297,
      "learning_rate": 1e-06,
      "loss": 1.0956,
      "num_input_tokens_seen": 93938548,
      "step": 1677
    },
    {
      "epoch": 3.734966592427617,
      "loss": 0.991020917892456,
      "loss_ce": 0.00029826798709109426,
      "loss_iou": 0.44140625,
      "loss_num": 0.0216064453125,
      "loss_xval": 0.9921875,
      "num_input_tokens_seen": 93938548,
      "step": 1677
    },
    {
      "epoch": 3.737193763919822,
      "grad_norm": 22.026281356811523,
      "learning_rate": 1e-06,
      "loss": 0.7681,
      "num_input_tokens_seen": 93993212,
      "step": 1678
    },
    {
      "epoch": 3.737193763919822,
      "loss": 0.945850670337677,
      "loss_ce": 0.000294047174975276,
      "loss_iou": 0.365234375,
      "loss_num": 0.04296875,
      "loss_xval": 0.9453125,
      "num_input_tokens_seen": 93993212,
      "step": 1678
    },
    {
      "epoch": 3.739420935412027,
      "grad_norm": 25.18975067138672,
      "learning_rate": 1e-06,
      "loss": 0.8155,
      "num_input_tokens_seen": 94051056,
      "step": 1679
    },
    {
      "epoch": 3.739420935412027,
      "loss": 0.9602963328361511,
      "loss_ce": 0.00106780044734478,
      "loss_iou": 0.384765625,
      "loss_num": 0.038330078125,
      "loss_xval": 0.9609375,
      "num_input_tokens_seen": 94051056,
      "step": 1679
    },
    {
      "epoch": 3.7416481069042318,
      "grad_norm": 22.7869873046875,
      "learning_rate": 1e-06,
      "loss": 0.7987,
      "num_input_tokens_seen": 94110436,
      "step": 1680
    },
    {
      "epoch": 3.7416481069042318,
      "loss": 0.8000843524932861,
      "loss_ce": 0.0002796592016238719,
      "loss_iou": 0.337890625,
      "loss_num": 0.024658203125,
      "loss_xval": 0.80078125,
      "num_input_tokens_seen": 94110436,
      "step": 1680
    },
    {
      "epoch": 3.7438752783964366,
      "grad_norm": 24.386417388916016,
      "learning_rate": 1e-06,
      "loss": 1.0109,
      "num_input_tokens_seen": 94167244,
      "step": 1681
    },
    {
      "epoch": 3.7438752783964366,
      "loss": 1.1265201568603516,
      "loss_ce": 0.0005436294013634324,
      "loss_iou": 0.4453125,
      "loss_num": 0.047119140625,
      "loss_xval": 1.125,
      "num_input_tokens_seen": 94167244,
      "step": 1681
    },
    {
      "epoch": 3.7461024498886415,
      "grad_norm": 19.943164825439453,
      "learning_rate": 1e-06,
      "loss": 1.1193,
      "num_input_tokens_seen": 94223920,
      "step": 1682
    },
    {
      "epoch": 3.7461024498886415,
      "loss": 0.9446117877960205,
      "loss_ce": 0.00027577788569033146,
      "loss_iou": 0.41796875,
      "loss_num": 0.021728515625,
      "loss_xval": 0.9453125,
      "num_input_tokens_seen": 94223920,
      "step": 1682
    },
    {
      "epoch": 3.7483296213808464,
      "grad_norm": 20.862585067749023,
      "learning_rate": 1e-06,
      "loss": 0.6461,
      "num_input_tokens_seen": 94279276,
      "step": 1683
    },
    {
      "epoch": 3.7483296213808464,
      "loss": 0.621061384677887,
      "loss_ce": 0.0002728351391851902,
      "loss_iou": 0.267578125,
      "loss_num": 0.01708984375,
      "loss_xval": 0.62109375,
      "num_input_tokens_seen": 94279276,
      "step": 1683
    },
    {
      "epoch": 3.750556792873051,
      "grad_norm": 19.76258087158203,
      "learning_rate": 1e-06,
      "loss": 0.8812,
      "num_input_tokens_seen": 94333312,
      "step": 1684
    },
    {
      "epoch": 3.750556792873051,
      "loss": 1.0115703344345093,
      "loss_ce": 0.00033991390955634415,
      "loss_iou": 0.4375,
      "loss_num": 0.0277099609375,
      "loss_xval": 1.0078125,
      "num_input_tokens_seen": 94333312,
      "step": 1684
    },
    {
      "epoch": 3.752783964365256,
      "grad_norm": 13.520242691040039,
      "learning_rate": 1e-06,
      "loss": 0.9022,
      "num_input_tokens_seen": 94389944,
      "step": 1685
    },
    {
      "epoch": 3.752783964365256,
      "loss": 1.0272419452667236,
      "loss_ce": 0.00038654671516269445,
      "loss_iou": 0.439453125,
      "loss_num": 0.0296630859375,
      "loss_xval": 1.0234375,
      "num_input_tokens_seen": 94389944,
      "step": 1685
    },
    {
      "epoch": 3.755011135857461,
      "grad_norm": 13.537060737609863,
      "learning_rate": 1e-06,
      "loss": 0.7416,
      "num_input_tokens_seen": 94445592,
      "step": 1686
    },
    {
      "epoch": 3.755011135857461,
      "loss": 0.7466691136360168,
      "loss_ce": 0.00033120866282843053,
      "loss_iou": 0.33203125,
      "loss_num": 0.016357421875,
      "loss_xval": 0.74609375,
      "num_input_tokens_seen": 94445592,
      "step": 1686
    },
    {
      "epoch": 3.757238307349666,
      "grad_norm": 17.105846405029297,
      "learning_rate": 1e-06,
      "loss": 0.8223,
      "num_input_tokens_seen": 94504800,
      "step": 1687
    },
    {
      "epoch": 3.757238307349666,
      "loss": 0.8430700302124023,
      "loss_ce": 0.0002966249012388289,
      "loss_iou": 0.337890625,
      "loss_num": 0.033447265625,
      "loss_xval": 0.84375,
      "num_input_tokens_seen": 94504800,
      "step": 1687
    },
    {
      "epoch": 3.7594654788418707,
      "grad_norm": 18.66309356689453,
      "learning_rate": 1e-06,
      "loss": 1.0448,
      "num_input_tokens_seen": 94560456,
      "step": 1688
    },
    {
      "epoch": 3.7594654788418707,
      "loss": 0.9472213983535767,
      "loss_ce": 0.00044402258936315775,
      "loss_iou": 0.412109375,
      "loss_num": 0.0245361328125,
      "loss_xval": 0.9453125,
      "num_input_tokens_seen": 94560456,
      "step": 1688
    },
    {
      "epoch": 3.7616926503340755,
      "grad_norm": 16.696016311645508,
      "learning_rate": 1e-06,
      "loss": 0.7974,
      "num_input_tokens_seen": 94617952,
      "step": 1689
    },
    {
      "epoch": 3.7616926503340755,
      "loss": 0.7895834445953369,
      "loss_ce": 0.0003989200631622225,
      "loss_iou": 0.322265625,
      "loss_num": 0.029052734375,
      "loss_xval": 0.7890625,
      "num_input_tokens_seen": 94617952,
      "step": 1689
    },
    {
      "epoch": 3.7639198218262804,
      "grad_norm": 27.074752807617188,
      "learning_rate": 1e-06,
      "loss": 0.7939,
      "num_input_tokens_seen": 94673300,
      "step": 1690
    },
    {
      "epoch": 3.7639198218262804,
      "loss": 0.47075653076171875,
      "loss_ce": 0.00029751902911812067,
      "loss_iou": 0.2041015625,
      "loss_num": 0.012451171875,
      "loss_xval": 0.470703125,
      "num_input_tokens_seen": 94673300,
      "step": 1690
    },
    {
      "epoch": 3.7661469933184857,
      "grad_norm": 23.333097457885742,
      "learning_rate": 1e-06,
      "loss": 0.901,
      "num_input_tokens_seen": 94727112,
      "step": 1691
    },
    {
      "epoch": 3.7661469933184857,
      "loss": 0.8041844964027405,
      "loss_ce": 0.0004735640832222998,
      "loss_iou": 0.349609375,
      "loss_num": 0.0213623046875,
      "loss_xval": 0.8046875,
      "num_input_tokens_seen": 94727112,
      "step": 1691
    },
    {
      "epoch": 3.7683741648106905,
      "grad_norm": 14.201781272888184,
      "learning_rate": 1e-06,
      "loss": 0.8346,
      "num_input_tokens_seen": 94784200,
      "step": 1692
    },
    {
      "epoch": 3.7683741648106905,
      "loss": 0.82025545835495,
      "loss_ce": 0.0003091579128522426,
      "loss_iou": 0.322265625,
      "loss_num": 0.034912109375,
      "loss_xval": 0.8203125,
      "num_input_tokens_seen": 94784200,
      "step": 1692
    },
    {
      "epoch": 3.7706013363028954,
      "grad_norm": 17.050172805786133,
      "learning_rate": 1e-06,
      "loss": 0.9037,
      "num_input_tokens_seen": 94842204,
      "step": 1693
    },
    {
      "epoch": 3.7706013363028954,
      "loss": 0.7915303707122803,
      "loss_ce": 0.0002705698716454208,
      "loss_iou": 0.3359375,
      "loss_num": 0.02392578125,
      "loss_xval": 0.79296875,
      "num_input_tokens_seen": 94842204,
      "step": 1693
    },
    {
      "epoch": 3.7728285077951003,
      "grad_norm": 20.24613380432129,
      "learning_rate": 1e-06,
      "loss": 0.6938,
      "num_input_tokens_seen": 94897320,
      "step": 1694
    },
    {
      "epoch": 3.7728285077951003,
      "loss": 0.6334647536277771,
      "loss_ce": 0.000285994668956846,
      "loss_iou": 0.275390625,
      "loss_num": 0.0164794921875,
      "loss_xval": 0.6328125,
      "num_input_tokens_seen": 94897320,
      "step": 1694
    },
    {
      "epoch": 3.775055679287305,
      "grad_norm": 23.602615356445312,
      "learning_rate": 1e-06,
      "loss": 0.8466,
      "num_input_tokens_seen": 94956064,
      "step": 1695
    },
    {
      "epoch": 3.775055679287305,
      "loss": 0.8750972747802734,
      "loss_ce": 0.00034144739038310945,
      "loss_iou": 0.380859375,
      "loss_num": 0.0230712890625,
      "loss_xval": 0.875,
      "num_input_tokens_seen": 94956064,
      "step": 1695
    },
    {
      "epoch": 3.77728285077951,
      "grad_norm": 19.40206527709961,
      "learning_rate": 1e-06,
      "loss": 0.7123,
      "num_input_tokens_seen": 95011940,
      "step": 1696
    },
    {
      "epoch": 3.77728285077951,
      "loss": 0.9245999455451965,
      "loss_ce": 0.0002835210179910064,
      "loss_iou": 0.375,
      "loss_num": 0.03466796875,
      "loss_xval": 0.92578125,
      "num_input_tokens_seen": 95011940,
      "step": 1696
    },
    {
      "epoch": 3.779510022271715,
      "grad_norm": 22.941545486450195,
      "learning_rate": 1e-06,
      "loss": 0.7155,
      "num_input_tokens_seen": 95065572,
      "step": 1697
    },
    {
      "epoch": 3.779510022271715,
      "loss": 0.5341686606407166,
      "loss_ce": 0.00023310747928917408,
      "loss_iou": 0.21484375,
      "loss_num": 0.0208740234375,
      "loss_xval": 0.53515625,
      "num_input_tokens_seen": 95065572,
      "step": 1697
    },
    {
      "epoch": 3.7817371937639197,
      "grad_norm": 19.74363899230957,
      "learning_rate": 1e-06,
      "loss": 0.7572,
      "num_input_tokens_seen": 95120584,
      "step": 1698
    },
    {
      "epoch": 3.7817371937639197,
      "loss": 0.7215688228607178,
      "loss_ce": 0.00037740974221378565,
      "loss_iou": 0.294921875,
      "loss_num": 0.0263671875,
      "loss_xval": 0.72265625,
      "num_input_tokens_seen": 95120584,
      "step": 1698
    },
    {
      "epoch": 3.7839643652561246,
      "grad_norm": 34.252037048339844,
      "learning_rate": 1e-06,
      "loss": 0.8692,
      "num_input_tokens_seen": 95176160,
      "step": 1699
    },
    {
      "epoch": 3.7839643652561246,
      "loss": 0.9719327688217163,
      "loss_ce": 0.0002530916826799512,
      "loss_iou": 0.421875,
      "loss_num": 0.025634765625,
      "loss_xval": 0.97265625,
      "num_input_tokens_seen": 95176160,
      "step": 1699
    },
    {
      "epoch": 3.78619153674833,
      "grad_norm": 17.124095916748047,
      "learning_rate": 1e-06,
      "loss": 0.7826,
      "num_input_tokens_seen": 95230472,
      "step": 1700
    },
    {
      "epoch": 3.78619153674833,
      "loss": 0.7893081903457642,
      "loss_ce": 0.0007340057054534554,
      "loss_iou": 0.330078125,
      "loss_num": 0.0255126953125,
      "loss_xval": 0.7890625,
      "num_input_tokens_seen": 95230472,
      "step": 1700
    },
    {
      "epoch": 3.7884187082405347,
      "grad_norm": 18.04917335510254,
      "learning_rate": 1e-06,
      "loss": 0.7732,
      "num_input_tokens_seen": 95286580,
      "step": 1701
    },
    {
      "epoch": 3.7884187082405347,
      "loss": 0.8537644743919373,
      "loss_ce": 0.0004929610877297819,
      "loss_iou": 0.361328125,
      "loss_num": 0.0257568359375,
      "loss_xval": 0.8515625,
      "num_input_tokens_seen": 95286580,
      "step": 1701
    },
    {
      "epoch": 3.7906458797327396,
      "grad_norm": 19.060016632080078,
      "learning_rate": 1e-06,
      "loss": 0.7842,
      "num_input_tokens_seen": 95341332,
      "step": 1702
    },
    {
      "epoch": 3.7906458797327396,
      "loss": 0.5456702709197998,
      "loss_ce": 0.00026007683482021093,
      "loss_iou": 0.2236328125,
      "loss_num": 0.01953125,
      "loss_xval": 0.546875,
      "num_input_tokens_seen": 95341332,
      "step": 1702
    },
    {
      "epoch": 3.7928730512249444,
      "grad_norm": 14.60232925415039,
      "learning_rate": 1e-06,
      "loss": 0.6544,
      "num_input_tokens_seen": 95397800,
      "step": 1703
    },
    {
      "epoch": 3.7928730512249444,
      "loss": 0.5783922076225281,
      "loss_ce": 0.00026723096380010247,
      "loss_iou": 0.248046875,
      "loss_num": 0.016357421875,
      "loss_xval": 0.578125,
      "num_input_tokens_seen": 95397800,
      "step": 1703
    },
    {
      "epoch": 3.7951002227171493,
      "grad_norm": 237.25967407226562,
      "learning_rate": 1e-06,
      "loss": 0.8609,
      "num_input_tokens_seen": 95453292,
      "step": 1704
    },
    {
      "epoch": 3.7951002227171493,
      "loss": 0.7244041562080383,
      "loss_ce": 0.0002830714511219412,
      "loss_iou": 0.314453125,
      "loss_num": 0.018798828125,
      "loss_xval": 0.72265625,
      "num_input_tokens_seen": 95453292,
      "step": 1704
    },
    {
      "epoch": 3.797327394209354,
      "grad_norm": 23.924535751342773,
      "learning_rate": 1e-06,
      "loss": 0.8929,
      "num_input_tokens_seen": 95509636,
      "step": 1705
    },
    {
      "epoch": 3.797327394209354,
      "loss": 0.9049976468086243,
      "loss_ce": 0.00045664224307984114,
      "loss_iou": 0.376953125,
      "loss_num": 0.0299072265625,
      "loss_xval": 0.90625,
      "num_input_tokens_seen": 95509636,
      "step": 1705
    },
    {
      "epoch": 3.799554565701559,
      "grad_norm": 13.076998710632324,
      "learning_rate": 1e-06,
      "loss": 0.7502,
      "num_input_tokens_seen": 95567220,
      "step": 1706
    },
    {
      "epoch": 3.799554565701559,
      "loss": 0.8397418260574341,
      "loss_ce": 0.0006305105634965003,
      "loss_iou": 0.3359375,
      "loss_num": 0.03369140625,
      "loss_xval": 0.83984375,
      "num_input_tokens_seen": 95567220,
      "step": 1706
    },
    {
      "epoch": 3.801781737193764,
      "grad_norm": 16.36588478088379,
      "learning_rate": 1e-06,
      "loss": 0.8995,
      "num_input_tokens_seen": 95621892,
      "step": 1707
    },
    {
      "epoch": 3.801781737193764,
      "loss": 0.8961795568466187,
      "loss_ce": 0.00042754405876621604,
      "loss_iou": 0.36328125,
      "loss_num": 0.034423828125,
      "loss_xval": 0.89453125,
      "num_input_tokens_seen": 95621892,
      "step": 1707
    },
    {
      "epoch": 3.8040089086859687,
      "grad_norm": 31.586313247680664,
      "learning_rate": 1e-06,
      "loss": 0.8272,
      "num_input_tokens_seen": 95678952,
      "step": 1708
    },
    {
      "epoch": 3.8040089086859687,
      "loss": 0.7548704147338867,
      "loss_ce": 0.0002317242615390569,
      "loss_iou": 0.333984375,
      "loss_num": 0.0169677734375,
      "loss_xval": 0.75390625,
      "num_input_tokens_seen": 95678952,
      "step": 1708
    },
    {
      "epoch": 3.8062360801781736,
      "grad_norm": 21.018543243408203,
      "learning_rate": 1e-06,
      "loss": 0.8788,
      "num_input_tokens_seen": 95734860,
      "step": 1709
    },
    {
      "epoch": 3.8062360801781736,
      "loss": 0.8822214603424072,
      "loss_ce": 0.0011179400607943535,
      "loss_iou": 0.37109375,
      "loss_num": 0.0279541015625,
      "loss_xval": 0.8828125,
      "num_input_tokens_seen": 95734860,
      "step": 1709
    },
    {
      "epoch": 3.8084632516703785,
      "grad_norm": 46.00482177734375,
      "learning_rate": 1e-06,
      "loss": 0.6713,
      "num_input_tokens_seen": 95791268,
      "step": 1710
    },
    {
      "epoch": 3.8084632516703785,
      "loss": 0.6450284123420715,
      "loss_ce": 0.00025300466222688556,
      "loss_iou": 0.296875,
      "loss_num": 0.01031494140625,
      "loss_xval": 0.64453125,
      "num_input_tokens_seen": 95791268,
      "step": 1710
    },
    {
      "epoch": 3.8106904231625833,
      "grad_norm": 19.820987701416016,
      "learning_rate": 1e-06,
      "loss": 0.9956,
      "num_input_tokens_seen": 95847960,
      "step": 1711
    },
    {
      "epoch": 3.8106904231625833,
      "loss": 0.8101856708526611,
      "loss_ce": 0.00037119118496775627,
      "loss_iou": 0.34765625,
      "loss_num": 0.0230712890625,
      "loss_xval": 0.80859375,
      "num_input_tokens_seen": 95847960,
      "step": 1711
    },
    {
      "epoch": 3.812917594654788,
      "grad_norm": 16.08196449279785,
      "learning_rate": 1e-06,
      "loss": 1.0883,
      "num_input_tokens_seen": 95904500,
      "step": 1712
    },
    {
      "epoch": 3.812917594654788,
      "loss": 0.8286195993423462,
      "loss_ce": 0.0004946578992530704,
      "loss_iou": 0.33984375,
      "loss_num": 0.029541015625,
      "loss_xval": 0.828125,
      "num_input_tokens_seen": 95904500,
      "step": 1712
    },
    {
      "epoch": 3.815144766146993,
      "grad_norm": 25.859769821166992,
      "learning_rate": 1e-06,
      "loss": 0.747,
      "num_input_tokens_seen": 95959000,
      "step": 1713
    },
    {
      "epoch": 3.815144766146993,
      "loss": 0.749795138835907,
      "loss_ce": 0.00028341758297756314,
      "loss_iou": 0.31640625,
      "loss_num": 0.0235595703125,
      "loss_xval": 0.75,
      "num_input_tokens_seen": 95959000,
      "step": 1713
    },
    {
      "epoch": 3.8173719376391984,
      "grad_norm": 16.63954734802246,
      "learning_rate": 1e-06,
      "loss": 0.7589,
      "num_input_tokens_seen": 96015556,
      "step": 1714
    },
    {
      "epoch": 3.8173719376391984,
      "loss": 0.7214926481246948,
      "loss_ce": 0.00030121111194603145,
      "loss_iou": 0.32421875,
      "loss_num": 0.01458740234375,
      "loss_xval": 0.72265625,
      "num_input_tokens_seen": 96015556,
      "step": 1714
    },
    {
      "epoch": 3.819599109131403,
      "grad_norm": 21.721033096313477,
      "learning_rate": 1e-06,
      "loss": 0.7904,
      "num_input_tokens_seen": 96072212,
      "step": 1715
    },
    {
      "epoch": 3.819599109131403,
      "loss": 0.7724882364273071,
      "loss_ce": 0.00027147267246618867,
      "loss_iou": 0.333984375,
      "loss_num": 0.0206298828125,
      "loss_xval": 0.7734375,
      "num_input_tokens_seen": 96072212,
      "step": 1715
    },
    {
      "epoch": 3.821826280623608,
      "grad_norm": 13.081016540527344,
      "learning_rate": 1e-06,
      "loss": 0.7134,
      "num_input_tokens_seen": 96129256,
      "step": 1716
    },
    {
      "epoch": 3.821826280623608,
      "loss": 0.8351633548736572,
      "loss_ce": 0.00026344467187300324,
      "loss_iou": 0.375,
      "loss_num": 0.0167236328125,
      "loss_xval": 0.8359375,
      "num_input_tokens_seen": 96129256,
      "step": 1716
    },
    {
      "epoch": 3.824053452115813,
      "grad_norm": 23.20547866821289,
      "learning_rate": 1e-06,
      "loss": 0.9529,
      "num_input_tokens_seen": 96186100,
      "step": 1717
    },
    {
      "epoch": 3.824053452115813,
      "loss": 1.0769813060760498,
      "loss_ce": 0.00032106111757457256,
      "loss_iou": 0.42578125,
      "loss_num": 0.044677734375,
      "loss_xval": 1.078125,
      "num_input_tokens_seen": 96186100,
      "step": 1717
    },
    {
      "epoch": 3.826280623608018,
      "grad_norm": 19.76165008544922,
      "learning_rate": 1e-06,
      "loss": 0.9206,
      "num_input_tokens_seen": 96242056,
      "step": 1718
    },
    {
      "epoch": 3.826280623608018,
      "loss": 0.7002924680709839,
      "loss_ce": 0.0003413489321246743,
      "loss_iou": 0.310546875,
      "loss_num": 0.0155029296875,
      "loss_xval": 0.69921875,
      "num_input_tokens_seen": 96242056,
      "step": 1718
    },
    {
      "epoch": 3.8285077951002227,
      "grad_norm": 19.560970306396484,
      "learning_rate": 1e-06,
      "loss": 0.9795,
      "num_input_tokens_seen": 96296820,
      "step": 1719
    },
    {
      "epoch": 3.8285077951002227,
      "loss": 1.1053590774536133,
      "loss_ce": 0.00037851842353120446,
      "loss_iou": 0.453125,
      "loss_num": 0.039306640625,
      "loss_xval": 1.1015625,
      "num_input_tokens_seen": 96296820,
      "step": 1719
    },
    {
      "epoch": 3.8307349665924275,
      "grad_norm": 13.943714141845703,
      "learning_rate": 1e-06,
      "loss": 0.7494,
      "num_input_tokens_seen": 96352772,
      "step": 1720
    },
    {
      "epoch": 3.8307349665924275,
      "loss": 0.9070804715156555,
      "loss_ce": 0.00034218025393784046,
      "loss_iou": 0.369140625,
      "loss_num": 0.033935546875,
      "loss_xval": 0.90625,
      "num_input_tokens_seen": 96352772,
      "step": 1720
    },
    {
      "epoch": 3.8329621380846324,
      "grad_norm": 24.83785629272461,
      "learning_rate": 1e-06,
      "loss": 0.8379,
      "num_input_tokens_seen": 96409184,
      "step": 1721
    },
    {
      "epoch": 3.8329621380846324,
      "loss": 0.9676505327224731,
      "loss_ce": 0.0003653773164842278,
      "loss_iou": 0.4140625,
      "loss_num": 0.027587890625,
      "loss_xval": 0.96875,
      "num_input_tokens_seen": 96409184,
      "step": 1721
    },
    {
      "epoch": 3.8351893095768377,
      "grad_norm": 18.294523239135742,
      "learning_rate": 1e-06,
      "loss": 0.7214,
      "num_input_tokens_seen": 96465040,
      "step": 1722
    },
    {
      "epoch": 3.8351893095768377,
      "loss": 0.5652261972427368,
      "loss_ce": 0.00028477917658165097,
      "loss_iou": 0.240234375,
      "loss_num": 0.0169677734375,
      "loss_xval": 0.56640625,
      "num_input_tokens_seen": 96465040,
      "step": 1722
    },
    {
      "epoch": 3.8374164810690425,
      "grad_norm": 14.921290397644043,
      "learning_rate": 1e-06,
      "loss": 0.7268,
      "num_input_tokens_seen": 96523672,
      "step": 1723
    },
    {
      "epoch": 3.8374164810690425,
      "loss": 0.7978357076644897,
      "loss_ce": 0.00022831102251075208,
      "loss_iou": 0.33203125,
      "loss_num": 0.0269775390625,
      "loss_xval": 0.796875,
      "num_input_tokens_seen": 96523672,
      "step": 1723
    },
    {
      "epoch": 3.8396436525612474,
      "grad_norm": 25.57061767578125,
      "learning_rate": 1e-06,
      "loss": 0.7983,
      "num_input_tokens_seen": 96579016,
      "step": 1724
    },
    {
      "epoch": 3.8396436525612474,
      "loss": 0.9255920648574829,
      "loss_ce": 0.0002990873181261122,
      "loss_iou": 0.404296875,
      "loss_num": 0.0235595703125,
      "loss_xval": 0.92578125,
      "num_input_tokens_seen": 96579016,
      "step": 1724
    },
    {
      "epoch": 3.8418708240534523,
      "grad_norm": 25.671611785888672,
      "learning_rate": 1e-06,
      "loss": 0.7852,
      "num_input_tokens_seen": 96633780,
      "step": 1725
    },
    {
      "epoch": 3.8418708240534523,
      "loss": 0.7190070152282715,
      "loss_ce": 0.0002570503856986761,
      "loss_iou": 0.3203125,
      "loss_num": 0.0157470703125,
      "loss_xval": 0.71875,
      "num_input_tokens_seen": 96633780,
      "step": 1725
    },
    {
      "epoch": 3.844097995545657,
      "grad_norm": 21.268661499023438,
      "learning_rate": 1e-06,
      "loss": 0.5661,
      "num_input_tokens_seen": 96689492,
      "step": 1726
    },
    {
      "epoch": 3.844097995545657,
      "loss": 0.563983678817749,
      "loss_ce": 0.0002629789523780346,
      "loss_iou": 0.2294921875,
      "loss_num": 0.02099609375,
      "loss_xval": 0.5625,
      "num_input_tokens_seen": 96689492,
      "step": 1726
    },
    {
      "epoch": 3.846325167037862,
      "grad_norm": 26.356407165527344,
      "learning_rate": 1e-06,
      "loss": 0.7125,
      "num_input_tokens_seen": 96746656,
      "step": 1727
    },
    {
      "epoch": 3.846325167037862,
      "loss": 0.8865300416946411,
      "loss_ce": 0.0004216782108414918,
      "loss_iou": 0.375,
      "loss_num": 0.027587890625,
      "loss_xval": 0.88671875,
      "num_input_tokens_seen": 96746656,
      "step": 1727
    },
    {
      "epoch": 3.848552338530067,
      "grad_norm": 16.691015243530273,
      "learning_rate": 1e-06,
      "loss": 0.9871,
      "num_input_tokens_seen": 96803292,
      "step": 1728
    },
    {
      "epoch": 3.848552338530067,
      "loss": 0.9029386043548584,
      "loss_ce": 0.0003506758948788047,
      "loss_iou": 0.380859375,
      "loss_num": 0.0279541015625,
      "loss_xval": 0.90234375,
      "num_input_tokens_seen": 96803292,
      "step": 1728
    },
    {
      "epoch": 3.8507795100222717,
      "grad_norm": 25.19491195678711,
      "learning_rate": 1e-06,
      "loss": 0.723,
      "num_input_tokens_seen": 96859896,
      "step": 1729
    },
    {
      "epoch": 3.8507795100222717,
      "loss": 0.7351480722427368,
      "loss_ce": 0.00028477725572884083,
      "loss_iou": 0.2890625,
      "loss_num": 0.0311279296875,
      "loss_xval": 0.734375,
      "num_input_tokens_seen": 96859896,
      "step": 1729
    },
    {
      "epoch": 3.8530066815144766,
      "grad_norm": 16.253820419311523,
      "learning_rate": 1e-06,
      "loss": 0.6062,
      "num_input_tokens_seen": 96917348,
      "step": 1730
    },
    {
      "epoch": 3.8530066815144766,
      "loss": 0.6677459478378296,
      "loss_ce": 0.0002654629643075168,
      "loss_iou": 0.28515625,
      "loss_num": 0.0189208984375,
      "loss_xval": 0.66796875,
      "num_input_tokens_seen": 96917348,
      "step": 1730
    },
    {
      "epoch": 3.8552338530066814,
      "grad_norm": 12.308496475219727,
      "learning_rate": 1e-06,
      "loss": 0.7134,
      "num_input_tokens_seen": 96975804,
      "step": 1731
    },
    {
      "epoch": 3.8552338530066814,
      "loss": 0.6045119166374207,
      "loss_ce": 0.0003859363787341863,
      "loss_iou": 0.248046875,
      "loss_num": 0.021484375,
      "loss_xval": 0.60546875,
      "num_input_tokens_seen": 96975804,
      "step": 1731
    },
    {
      "epoch": 3.8574610244988863,
      "grad_norm": 26.0831298828125,
      "learning_rate": 1e-06,
      "loss": 1.1291,
      "num_input_tokens_seen": 97033224,
      "step": 1732
    },
    {
      "epoch": 3.8574610244988863,
      "loss": 1.0747888088226318,
      "loss_ce": 0.00032590571208857,
      "loss_iou": 0.435546875,
      "loss_num": 0.04052734375,
      "loss_xval": 1.078125,
      "num_input_tokens_seen": 97033224,
      "step": 1732
    },
    {
      "epoch": 3.859688195991091,
      "grad_norm": 22.14068031311035,
      "learning_rate": 1e-06,
      "loss": 0.9682,
      "num_input_tokens_seen": 97092012,
      "step": 1733
    },
    {
      "epoch": 3.859688195991091,
      "loss": 1.2439754009246826,
      "loss_ce": 0.00032304698834195733,
      "loss_iou": 0.515625,
      "loss_num": 0.042724609375,
      "loss_xval": 1.2421875,
      "num_input_tokens_seen": 97092012,
      "step": 1733
    },
    {
      "epoch": 3.861915367483296,
      "grad_norm": 17.057979583740234,
      "learning_rate": 1e-06,
      "loss": 0.893,
      "num_input_tokens_seen": 97145840,
      "step": 1734
    },
    {
      "epoch": 3.861915367483296,
      "loss": 1.0669821500778198,
      "loss_ce": 0.0003318190574645996,
      "loss_iou": 0.39453125,
      "loss_num": 0.05517578125,
      "loss_xval": 1.0703125,
      "num_input_tokens_seen": 97145840,
      "step": 1734
    },
    {
      "epoch": 3.864142538975501,
      "grad_norm": 14.770302772521973,
      "learning_rate": 1e-06,
      "loss": 0.8305,
      "num_input_tokens_seen": 97201400,
      "step": 1735
    },
    {
      "epoch": 3.864142538975501,
      "loss": 0.9496909379959106,
      "loss_ce": 0.00047216590610332787,
      "loss_iou": 0.39453125,
      "loss_num": 0.031982421875,
      "loss_xval": 0.94921875,
      "num_input_tokens_seen": 97201400,
      "step": 1735
    },
    {
      "epoch": 3.866369710467706,
      "grad_norm": 19.580652236938477,
      "learning_rate": 1e-06,
      "loss": 0.8005,
      "num_input_tokens_seen": 97258904,
      "step": 1736
    },
    {
      "epoch": 3.866369710467706,
      "loss": 0.6937247514724731,
      "loss_ce": 0.0003653843014035374,
      "loss_iou": 0.298828125,
      "loss_num": 0.0194091796875,
      "loss_xval": 0.6953125,
      "num_input_tokens_seen": 97258904,
      "step": 1736
    },
    {
      "epoch": 3.868596881959911,
      "grad_norm": 29.482715606689453,
      "learning_rate": 1e-06,
      "loss": 0.8335,
      "num_input_tokens_seen": 97317044,
      "step": 1737
    },
    {
      "epoch": 3.868596881959911,
      "loss": 0.874514102935791,
      "loss_ce": 0.00024650723207741976,
      "loss_iou": 0.39453125,
      "loss_num": 0.017333984375,
      "loss_xval": 0.875,
      "num_input_tokens_seen": 97317044,
      "step": 1737
    },
    {
      "epoch": 3.870824053452116,
      "grad_norm": 17.633045196533203,
      "learning_rate": 1e-06,
      "loss": 0.965,
      "num_input_tokens_seen": 97372048,
      "step": 1738
    },
    {
      "epoch": 3.870824053452116,
      "loss": 1.020100474357605,
      "loss_ce": 0.0005692073609679937,
      "loss_iou": 0.40234375,
      "loss_num": 0.04248046875,
      "loss_xval": 1.015625,
      "num_input_tokens_seen": 97372048,
      "step": 1738
    },
    {
      "epoch": 3.8730512249443207,
      "grad_norm": 18.022729873657227,
      "learning_rate": 1e-06,
      "loss": 0.7467,
      "num_input_tokens_seen": 97427968,
      "step": 1739
    },
    {
      "epoch": 3.8730512249443207,
      "loss": 0.7238982915878296,
      "loss_ce": 0.0002655313292052597,
      "loss_iou": 0.294921875,
      "loss_num": 0.0267333984375,
      "loss_xval": 0.72265625,
      "num_input_tokens_seen": 97427968,
      "step": 1739
    },
    {
      "epoch": 3.8752783964365256,
      "grad_norm": 20.802419662475586,
      "learning_rate": 1e-06,
      "loss": 0.8692,
      "num_input_tokens_seen": 97483504,
      "step": 1740
    },
    {
      "epoch": 3.8752783964365256,
      "loss": 0.9595529437065125,
      "loss_ce": 0.00032444443786516786,
      "loss_iou": 0.419921875,
      "loss_num": 0.0240478515625,
      "loss_xval": 0.9609375,
      "num_input_tokens_seen": 97483504,
      "step": 1740
    },
    {
      "epoch": 3.8775055679287305,
      "grad_norm": 25.31464385986328,
      "learning_rate": 1e-06,
      "loss": 0.9623,
      "num_input_tokens_seen": 97539960,
      "step": 1741
    },
    {
      "epoch": 3.8775055679287305,
      "loss": 0.9504677057266235,
      "loss_ce": 0.0002723511715885252,
      "loss_iou": 0.3828125,
      "loss_num": 0.037109375,
      "loss_xval": 0.94921875,
      "num_input_tokens_seen": 97539960,
      "step": 1741
    },
    {
      "epoch": 3.8797327394209353,
      "grad_norm": 18.857027053833008,
      "learning_rate": 1e-06,
      "loss": 0.6928,
      "num_input_tokens_seen": 97597248,
      "step": 1742
    },
    {
      "epoch": 3.8797327394209353,
      "loss": 0.7760539650917053,
      "loss_ce": 0.00023611923097632825,
      "loss_iou": 0.33984375,
      "loss_num": 0.0196533203125,
      "loss_xval": 0.77734375,
      "num_input_tokens_seen": 97597248,
      "step": 1742
    },
    {
      "epoch": 3.88195991091314,
      "grad_norm": 24.135141372680664,
      "learning_rate": 1e-06,
      "loss": 0.7342,
      "num_input_tokens_seen": 97652256,
      "step": 1743
    },
    {
      "epoch": 3.88195991091314,
      "loss": 0.636237621307373,
      "loss_ce": 0.00025132749578915536,
      "loss_iou": 0.26171875,
      "loss_num": 0.02197265625,
      "loss_xval": 0.63671875,
      "num_input_tokens_seen": 97652256,
      "step": 1743
    },
    {
      "epoch": 3.884187082405345,
      "grad_norm": 27.896411895751953,
      "learning_rate": 1e-06,
      "loss": 0.7845,
      "num_input_tokens_seen": 97706708,
      "step": 1744
    },
    {
      "epoch": 3.884187082405345,
      "loss": 0.7319508790969849,
      "loss_ce": 0.00026141630951315165,
      "loss_iou": 0.328125,
      "loss_num": 0.01544189453125,
      "loss_xval": 0.73046875,
      "num_input_tokens_seen": 97706708,
      "step": 1744
    },
    {
      "epoch": 3.8864142538975504,
      "grad_norm": 16.71023178100586,
      "learning_rate": 1e-06,
      "loss": 0.7643,
      "num_input_tokens_seen": 97765116,
      "step": 1745
    },
    {
      "epoch": 3.8864142538975504,
      "loss": 0.8230664730072021,
      "loss_ce": 0.00031261841650120914,
      "loss_iou": 0.328125,
      "loss_num": 0.033447265625,
      "loss_xval": 0.82421875,
      "num_input_tokens_seen": 97765116,
      "step": 1745
    },
    {
      "epoch": 3.888641425389755,
      "grad_norm": 17.422271728515625,
      "learning_rate": 1e-06,
      "loss": 0.7545,
      "num_input_tokens_seen": 97822628,
      "step": 1746
    },
    {
      "epoch": 3.888641425389755,
      "loss": 0.6438745260238647,
      "loss_ce": 0.0003198395133949816,
      "loss_iou": 0.2734375,
      "loss_num": 0.0191650390625,
      "loss_xval": 0.64453125,
      "num_input_tokens_seen": 97822628,
      "step": 1746
    },
    {
      "epoch": 3.89086859688196,
      "grad_norm": 14.700139999389648,
      "learning_rate": 1e-06,
      "loss": 0.8227,
      "num_input_tokens_seen": 97878372,
      "step": 1747
    },
    {
      "epoch": 3.89086859688196,
      "loss": 0.8928521275520325,
      "loss_ce": 0.00027394542121328413,
      "loss_iou": 0.380859375,
      "loss_num": 0.0260009765625,
      "loss_xval": 0.890625,
      "num_input_tokens_seen": 97878372,
      "step": 1747
    },
    {
      "epoch": 3.893095768374165,
      "grad_norm": 25.434099197387695,
      "learning_rate": 1e-06,
      "loss": 0.6795,
      "num_input_tokens_seen": 97932476,
      "step": 1748
    },
    {
      "epoch": 3.893095768374165,
      "loss": 0.6719607710838318,
      "loss_ce": 0.0013064806116744876,
      "loss_iou": 0.2578125,
      "loss_num": 0.03076171875,
      "loss_xval": 0.671875,
      "num_input_tokens_seen": 97932476,
      "step": 1748
    },
    {
      "epoch": 3.89532293986637,
      "grad_norm": 12.71288776397705,
      "learning_rate": 1e-06,
      "loss": 0.9539,
      "num_input_tokens_seen": 97990880,
      "step": 1749
    },
    {
      "epoch": 3.89532293986637,
      "loss": 0.8964526653289795,
      "loss_ce": 0.000578734208829701,
      "loss_iou": 0.3828125,
      "loss_num": 0.02587890625,
      "loss_xval": 0.89453125,
      "num_input_tokens_seen": 97990880,
      "step": 1749
    },
    {
      "epoch": 3.8975501113585747,
      "grad_norm": 20.1357421875,
      "learning_rate": 1e-06,
      "loss": 0.8062,
      "num_input_tokens_seen": 98046088,
      "step": 1750
    },
    {
      "epoch": 3.8975501113585747,
      "eval_seeclick_web_CIoU": 0.5617943108081818,
      "eval_seeclick_web_GIoU": 0.558159202337265,
      "eval_seeclick_web_IoU": 0.5777357220649719,
      "eval_seeclick_web_MAE_all": 0.017627435736358166,
      "eval_seeclick_web_MAE_h": 0.010515023721382022,
      "eval_seeclick_web_MAE_w": 0.018624153919517994,
      "eval_seeclick_web_MAE_x_boxes": 0.00829980755224824,
      "eval_seeclick_web_MAE_y_boxes": 0.02257556258700788,
      "eval_seeclick_web_inside_bbox": 0.9166666567325592,
      "eval_seeclick_web_loss": 0.9613375663757324,
      "eval_seeclick_web_loss_ce": 0.0003832996590062976,
      "eval_seeclick_web_loss_iou": 0.43701171875,
      "eval_seeclick_web_loss_num": 0.013935089111328125,
      "eval_seeclick_web_loss_xval": 0.9443359375,
      "eval_seeclick_web_runtime": 27.2225,
      "eval_seeclick_web_samples_per_second": 1.837,
      "eval_seeclick_web_steps_per_second": 0.073,
      "num_input_tokens_seen": 98046088,
      "step": 1750
    },
    {
      "epoch": 3.8975501113585747,
      "eval_icons_CIoU": 0.3064361959695816,
      "eval_icons_GIoU": 0.32649458944797516,
      "eval_icons_IoU": 0.38111720979213715,
      "eval_icons_MAE_all": 0.06658957526087761,
      "eval_icons_MAE_h": 0.040302242152392864,
      "eval_icons_MAE_w": 0.07293413020670414,
      "eval_icons_MAE_x_boxes": 0.06224925257265568,
      "eval_icons_MAE_y_boxes": 0.03964635543525219,
      "eval_icons_inside_bbox": 0.6336805522441864,
      "eval_icons_loss": 1.7142175436019897,
      "eval_icons_loss_ce": 0.0004412867419887334,
      "eval_icons_loss_iou": 0.6680908203125,
      "eval_icons_loss_num": 0.06289100646972656,
      "eval_icons_loss_xval": 1.64990234375,
      "eval_icons_runtime": 24.1165,
      "eval_icons_samples_per_second": 2.073,
      "eval_icons_steps_per_second": 0.083,
      "num_input_tokens_seen": 98046088,
      "step": 1750
    },
    {
      "epoch": 3.8975501113585747,
      "eval_screenspot_CIoU": 0.31633803248405457,
      "eval_screenspot_GIoU": 0.3353947401046753,
      "eval_screenspot_IoU": 0.3984935482343038,
      "eval_screenspot_MAE_all": 0.06795181334018707,
      "eval_screenspot_MAE_h": 0.03754324403901895,
      "eval_screenspot_MAE_w": 0.07950195794304211,
      "eval_screenspot_MAE_x_boxes": 0.08205063392718633,
      "eval_screenspot_MAE_y_boxes": 0.044876331463456154,
      "eval_screenspot_inside_bbox": 0.6358333428700765,
      "eval_screenspot_loss": 1.7439180612564087,
      "eval_screenspot_loss_ce": 0.0004279320516313116,
      "eval_screenspot_loss_iou": 0.712890625,
      "eval_screenspot_loss_num": 0.07890574137369792,
      "eval_screenspot_loss_xval": 1.8204752604166667,
      "eval_screenspot_runtime": 34.5027,
      "eval_screenspot_samples_per_second": 2.58,
      "eval_screenspot_steps_per_second": 0.087,
      "num_input_tokens_seen": 98046088,
      "step": 1750
    },
    {
      "epoch": 3.8975501113585747,
      "eval_compot_CIoU": 0.3428885042667389,
      "eval_compot_GIoU": 0.3697793632745743,
      "eval_compot_IoU": 0.40237635374069214,
      "eval_compot_MAE_all": 0.019974621944129467,
      "eval_compot_MAE_h": 0.009356681257486343,
      "eval_compot_MAE_w": 0.02669452875852585,
      "eval_compot_MAE_x_boxes": 0.0306751299649477,
      "eval_compot_MAE_y_boxes": 0.00634155492298305,
      "eval_compot_inside_bbox": 0.6302083432674408,
      "eval_compot_loss": 1.3990836143493652,
      "eval_compot_loss_ce": 0.000333975360263139,
      "eval_compot_loss_iou": 0.637939453125,
      "eval_compot_loss_num": 0.018901824951171875,
      "eval_compot_loss_xval": 1.37109375,
      "eval_compot_runtime": 19.3662,
      "eval_compot_samples_per_second": 2.582,
      "eval_compot_steps_per_second": 0.103,
      "num_input_tokens_seen": 98046088,
      "step": 1750
    },
    {
      "epoch": 3.8975501113585747,
      "eval_custom_ui_val_CIoU": 0.4409982098473443,
      "eval_custom_ui_val_GIoU": 0.46577325297726524,
      "eval_custom_ui_val_IoU": 0.49479154580169254,
      "eval_custom_ui_val_MAE_all": 0.03411407251324919,
      "eval_custom_ui_val_MAE_h": 0.0205486583419972,
      "eval_custom_ui_val_MAE_w": 0.03924656597276529,
      "eval_custom_ui_val_MAE_x_boxes": 0.03744201848490371,
      "eval_custom_ui_val_MAE_y_boxes": 0.019695010320800874,
      "eval_custom_ui_val_inside_bbox": 0.6867283980051676,
      "eval_custom_ui_val_loss": 1.2667858600616455,
      "eval_custom_ui_val_loss_ce": 0.0004089692018977884,
      "eval_custom_ui_val_loss_iou": 0.5344373914930556,
      "eval_custom_ui_val_loss_num": 0.03371832105848524,
      "eval_custom_ui_val_loss_xval": 1.2374403211805556,
      "eval_custom_ui_val_runtime": 56.4523,
      "eval_custom_ui_val_samples_per_second": 4.694,
      "eval_custom_ui_val_steps_per_second": 0.159,
      "num_input_tokens_seen": 98046088,
      "step": 1750
    },
    {
      "epoch": 3.8975501113585747,
      "loss": 0.9943009614944458,
      "loss_ce": 0.00040448884828947484,
      "loss_iou": 0.42578125,
      "loss_num": 0.0283203125,
      "loss_xval": 0.9921875,
      "num_input_tokens_seen": 98046088,
      "step": 1750
    },
    {
      "epoch": 3.8997772828507795,
      "grad_norm": 17.32667350769043,
      "learning_rate": 1e-06,
      "loss": 0.9158,
      "num_input_tokens_seen": 98101832,
      "step": 1751
    },
    {
      "epoch": 3.8997772828507795,
      "loss": 0.8142361640930176,
      "loss_ce": 0.00027136015705764294,
      "loss_iou": 0.365234375,
      "loss_num": 0.0167236328125,
      "loss_xval": 0.8125,
      "num_input_tokens_seen": 98101832,
      "step": 1751
    },
    {
      "epoch": 3.9020044543429844,
      "grad_norm": 14.840397834777832,
      "learning_rate": 1e-06,
      "loss": 0.7732,
      "num_input_tokens_seen": 98158656,
      "step": 1752
    },
    {
      "epoch": 3.9020044543429844,
      "loss": 0.9086735248565674,
      "loss_ce": 0.0004703607992269099,
      "loss_iou": 0.341796875,
      "loss_num": 0.044677734375,
      "loss_xval": 0.90625,
      "num_input_tokens_seen": 98158656,
      "step": 1752
    },
    {
      "epoch": 3.9042316258351892,
      "grad_norm": 16.115385055541992,
      "learning_rate": 1e-06,
      "loss": 0.7102,
      "num_input_tokens_seen": 98216120,
      "step": 1753
    },
    {
      "epoch": 3.9042316258351892,
      "loss": 0.6780398488044739,
      "loss_ce": 0.0003054844564758241,
      "loss_iou": 0.2890625,
      "loss_num": 0.020263671875,
      "loss_xval": 0.6796875,
      "num_input_tokens_seen": 98216120,
      "step": 1753
    },
    {
      "epoch": 3.906458797327394,
      "grad_norm": 22.08925437927246,
      "learning_rate": 1e-06,
      "loss": 0.8575,
      "num_input_tokens_seen": 98273000,
      "step": 1754
    },
    {
      "epoch": 3.906458797327394,
      "loss": 0.6279443502426147,
      "loss_ce": 0.0002587907947599888,
      "loss_iou": 0.291015625,
      "loss_num": 0.00958251953125,
      "loss_xval": 0.62890625,
      "num_input_tokens_seen": 98273000,
      "step": 1754
    },
    {
      "epoch": 3.908685968819599,
      "grad_norm": 15.868555068969727,
      "learning_rate": 1e-06,
      "loss": 0.8731,
      "num_input_tokens_seen": 98330372,
      "step": 1755
    },
    {
      "epoch": 3.908685968819599,
      "loss": 1.0122852325439453,
      "loss_ce": 0.00032239314168691635,
      "loss_iou": 0.443359375,
      "loss_num": 0.0250244140625,
      "loss_xval": 1.015625,
      "num_input_tokens_seen": 98330372,
      "step": 1755
    },
    {
      "epoch": 3.910913140311804,
      "grad_norm": 20.903287887573242,
      "learning_rate": 1e-06,
      "loss": 1.0159,
      "num_input_tokens_seen": 98387768,
      "step": 1756
    },
    {
      "epoch": 3.910913140311804,
      "loss": 0.9633511900901794,
      "loss_ce": 0.00046056750579737127,
      "loss_iou": 0.384765625,
      "loss_num": 0.038818359375,
      "loss_xval": 0.9609375,
      "num_input_tokens_seen": 98387768,
      "step": 1756
    },
    {
      "epoch": 3.9131403118040087,
      "grad_norm": 14.697265625,
      "learning_rate": 1e-06,
      "loss": 1.0258,
      "num_input_tokens_seen": 98443212,
      "step": 1757
    },
    {
      "epoch": 3.9131403118040087,
      "loss": 1.067274808883667,
      "loss_ce": 0.00038028520066291094,
      "loss_iou": 0.435546875,
      "loss_num": 0.03857421875,
      "loss_xval": 1.0703125,
      "num_input_tokens_seen": 98443212,
      "step": 1757
    },
    {
      "epoch": 3.9153674832962135,
      "grad_norm": 18.442344665527344,
      "learning_rate": 1e-06,
      "loss": 0.8647,
      "num_input_tokens_seen": 98498060,
      "step": 1758
    },
    {
      "epoch": 3.9153674832962135,
      "loss": 0.6922543048858643,
      "loss_ce": 0.0002377206110395491,
      "loss_iou": 0.27734375,
      "loss_num": 0.027587890625,
      "loss_xval": 0.69140625,
      "num_input_tokens_seen": 98498060,
      "step": 1758
    },
    {
      "epoch": 3.917594654788419,
      "grad_norm": 22.285242080688477,
      "learning_rate": 1e-06,
      "loss": 0.716,
      "num_input_tokens_seen": 98553040,
      "step": 1759
    },
    {
      "epoch": 3.917594654788419,
      "loss": 0.8192723393440247,
      "loss_ce": 0.0006687932182103395,
      "loss_iou": 0.36328125,
      "loss_num": 0.0181884765625,
      "loss_xval": 0.8203125,
      "num_input_tokens_seen": 98553040,
      "step": 1759
    },
    {
      "epoch": 3.9198218262806237,
      "grad_norm": 19.2437801361084,
      "learning_rate": 1e-06,
      "loss": 0.8274,
      "num_input_tokens_seen": 98607992,
      "step": 1760
    },
    {
      "epoch": 3.9198218262806237,
      "loss": 1.0185580253601074,
      "loss_ce": 0.00024743605172261596,
      "loss_iou": 0.451171875,
      "loss_num": 0.0234375,
      "loss_xval": 1.015625,
      "num_input_tokens_seen": 98607992,
      "step": 1760
    },
    {
      "epoch": 3.9220489977728286,
      "grad_norm": 28.81540298461914,
      "learning_rate": 1e-06,
      "loss": 0.8937,
      "num_input_tokens_seen": 98663688,
      "step": 1761
    },
    {
      "epoch": 3.9220489977728286,
      "loss": 0.7943629026412964,
      "loss_ce": 0.0004176311194896698,
      "loss_iou": 0.32421875,
      "loss_num": 0.0291748046875,
      "loss_xval": 0.79296875,
      "num_input_tokens_seen": 98663688,
      "step": 1761
    },
    {
      "epoch": 3.9242761692650334,
      "grad_norm": 13.910033226013184,
      "learning_rate": 1e-06,
      "loss": 0.8205,
      "num_input_tokens_seen": 98720668,
      "step": 1762
    },
    {
      "epoch": 3.9242761692650334,
      "loss": 0.7007368803024292,
      "loss_ce": 0.0023726352956146,
      "loss_iou": 0.298828125,
      "loss_num": 0.0205078125,
      "loss_xval": 0.69921875,
      "num_input_tokens_seen": 98720668,
      "step": 1762
    },
    {
      "epoch": 3.9265033407572383,
      "grad_norm": 19.399320602416992,
      "learning_rate": 1e-06,
      "loss": 0.6872,
      "num_input_tokens_seen": 98777756,
      "step": 1763
    },
    {
      "epoch": 3.9265033407572383,
      "loss": 0.4689710736274719,
      "loss_ce": 0.00022106988762971014,
      "loss_iou": 0.203125,
      "loss_num": 0.0123291015625,
      "loss_xval": 0.46875,
      "num_input_tokens_seen": 98777756,
      "step": 1763
    },
    {
      "epoch": 3.928730512249443,
      "grad_norm": 18.18318748474121,
      "learning_rate": 1e-06,
      "loss": 0.7284,
      "num_input_tokens_seen": 98835320,
      "step": 1764
    },
    {
      "epoch": 3.928730512249443,
      "loss": 0.7946980595588684,
      "loss_ce": 0.0002644615597091615,
      "loss_iou": 0.359375,
      "loss_num": 0.01495361328125,
      "loss_xval": 0.79296875,
      "num_input_tokens_seen": 98835320,
      "step": 1764
    },
    {
      "epoch": 3.930957683741648,
      "grad_norm": 17.789981842041016,
      "learning_rate": 1e-06,
      "loss": 0.846,
      "num_input_tokens_seen": 98891900,
      "step": 1765
    },
    {
      "epoch": 3.930957683741648,
      "loss": 0.5826168060302734,
      "loss_ce": 0.0005855775089003146,
      "loss_iou": 0.255859375,
      "loss_num": 0.01446533203125,
      "loss_xval": 0.58203125,
      "num_input_tokens_seen": 98891900,
      "step": 1765
    },
    {
      "epoch": 3.933184855233853,
      "grad_norm": 14.104592323303223,
      "learning_rate": 1e-06,
      "loss": 0.8345,
      "num_input_tokens_seen": 98948704,
      "step": 1766
    },
    {
      "epoch": 3.933184855233853,
      "loss": 1.0315115451812744,
      "loss_ce": 0.0002615266712382436,
      "loss_iou": 0.43359375,
      "loss_num": 0.033203125,
      "loss_xval": 1.03125,
      "num_input_tokens_seen": 98948704,
      "step": 1766
    },
    {
      "epoch": 3.935412026726058,
      "grad_norm": 17.459924697875977,
      "learning_rate": 1e-06,
      "loss": 0.8304,
      "num_input_tokens_seen": 99005660,
      "step": 1767
    },
    {
      "epoch": 3.935412026726058,
      "loss": 0.9141278266906738,
      "loss_ce": 0.00030949406209401786,
      "loss_iou": 0.404296875,
      "loss_num": 0.0208740234375,
      "loss_xval": 0.9140625,
      "num_input_tokens_seen": 99005660,
      "step": 1767
    },
    {
      "epoch": 3.937639198218263,
      "grad_norm": 18.547863006591797,
      "learning_rate": 1e-06,
      "loss": 1.1333,
      "num_input_tokens_seen": 99061216,
      "step": 1768
    },
    {
      "epoch": 3.937639198218263,
      "loss": 0.909751296043396,
      "loss_ce": 0.00032748148078098893,
      "loss_iou": 0.380859375,
      "loss_num": 0.029541015625,
      "loss_xval": 0.91015625,
      "num_input_tokens_seen": 99061216,
      "step": 1768
    },
    {
      "epoch": 3.939866369710468,
      "grad_norm": 17.022930145263672,
      "learning_rate": 1e-06,
      "loss": 0.7887,
      "num_input_tokens_seen": 99120308,
      "step": 1769
    },
    {
      "epoch": 3.939866369710468,
      "loss": 0.8347407579421997,
      "loss_ce": 0.0002681155747268349,
      "loss_iou": 0.36328125,
      "loss_num": 0.0216064453125,
      "loss_xval": 0.8359375,
      "num_input_tokens_seen": 99120308,
      "step": 1769
    },
    {
      "epoch": 3.9420935412026727,
      "grad_norm": 18.062122344970703,
      "learning_rate": 1e-06,
      "loss": 0.846,
      "num_input_tokens_seen": 99176104,
      "step": 1770
    },
    {
      "epoch": 3.9420935412026727,
      "loss": 0.7344276905059814,
      "loss_ce": 0.00029679658473469317,
      "loss_iou": 0.29296875,
      "loss_num": 0.030029296875,
      "loss_xval": 0.734375,
      "num_input_tokens_seen": 99176104,
      "step": 1770
    },
    {
      "epoch": 3.9443207126948776,
      "grad_norm": 24.9909610748291,
      "learning_rate": 1e-06,
      "loss": 0.9439,
      "num_input_tokens_seen": 99230876,
      "step": 1771
    },
    {
      "epoch": 3.9443207126948776,
      "loss": 0.9261212944984436,
      "loss_ce": 0.0003400646965019405,
      "loss_iou": 0.41015625,
      "loss_num": 0.0211181640625,
      "loss_xval": 0.92578125,
      "num_input_tokens_seen": 99230876,
      "step": 1771
    },
    {
      "epoch": 3.9465478841870825,
      "grad_norm": 14.309945106506348,
      "learning_rate": 1e-06,
      "loss": 0.7448,
      "num_input_tokens_seen": 99286860,
      "step": 1772
    },
    {
      "epoch": 3.9465478841870825,
      "loss": 0.7100058794021606,
      "loss_ce": 0.000289075484033674,
      "loss_iou": 0.310546875,
      "loss_num": 0.017578125,
      "loss_xval": 0.7109375,
      "num_input_tokens_seen": 99286860,
      "step": 1772
    },
    {
      "epoch": 3.9487750556792873,
      "grad_norm": 12.438628196716309,
      "learning_rate": 1e-06,
      "loss": 1.0524,
      "num_input_tokens_seen": 99344372,
      "step": 1773
    },
    {
      "epoch": 3.9487750556792873,
      "loss": 1.3829472064971924,
      "loss_ce": 0.00037877040449529886,
      "loss_iou": 0.59375,
      "loss_num": 0.038330078125,
      "loss_xval": 1.3828125,
      "num_input_tokens_seen": 99344372,
      "step": 1773
    },
    {
      "epoch": 3.951002227171492,
      "grad_norm": 19.387874603271484,
      "learning_rate": 1e-06,
      "loss": 0.8822,
      "num_input_tokens_seen": 99401420,
      "step": 1774
    },
    {
      "epoch": 3.951002227171492,
      "loss": 1.1286416053771973,
      "loss_ce": 0.0004677603137679398,
      "loss_iou": 0.455078125,
      "loss_num": 0.043701171875,
      "loss_xval": 1.125,
      "num_input_tokens_seen": 99401420,
      "step": 1774
    },
    {
      "epoch": 3.953229398663697,
      "grad_norm": 16.427181243896484,
      "learning_rate": 1e-06,
      "loss": 0.9915,
      "num_input_tokens_seen": 99456744,
      "step": 1775
    },
    {
      "epoch": 3.953229398663697,
      "loss": 0.9648089408874512,
      "loss_ce": 0.00045345089165493846,
      "loss_iou": 0.40234375,
      "loss_num": 0.03173828125,
      "loss_xval": 0.96484375,
      "num_input_tokens_seen": 99456744,
      "step": 1775
    },
    {
      "epoch": 3.955456570155902,
      "grad_norm": 16.38910675048828,
      "learning_rate": 1e-06,
      "loss": 0.8111,
      "num_input_tokens_seen": 99511464,
      "step": 1776
    },
    {
      "epoch": 3.955456570155902,
      "loss": 1.0158613920211792,
      "loss_ce": 0.0002364249958191067,
      "loss_iou": 0.42578125,
      "loss_num": 0.03271484375,
      "loss_xval": 1.015625,
      "num_input_tokens_seen": 99511464,
      "step": 1776
    },
    {
      "epoch": 3.9576837416481068,
      "grad_norm": 28.928443908691406,
      "learning_rate": 1e-06,
      "loss": 0.8633,
      "num_input_tokens_seen": 99567452,
      "step": 1777
    },
    {
      "epoch": 3.9576837416481068,
      "loss": 1.1553375720977783,
      "loss_ce": 0.0003082446346525103,
      "loss_iou": 0.49609375,
      "loss_num": 0.03271484375,
      "loss_xval": 1.15625,
      "num_input_tokens_seen": 99567452,
      "step": 1777
    },
    {
      "epoch": 3.9599109131403116,
      "grad_norm": 21.195846557617188,
      "learning_rate": 1e-06,
      "loss": 0.846,
      "num_input_tokens_seen": 99624932,
      "step": 1778
    },
    {
      "epoch": 3.9599109131403116,
      "loss": 0.6980462074279785,
      "loss_ce": 0.0002923067077063024,
      "loss_iou": 0.2734375,
      "loss_num": 0.0299072265625,
      "loss_xval": 0.69921875,
      "num_input_tokens_seen": 99624932,
      "step": 1778
    },
    {
      "epoch": 3.9621380846325165,
      "grad_norm": 16.269763946533203,
      "learning_rate": 1e-06,
      "loss": 0.6815,
      "num_input_tokens_seen": 99681052,
      "step": 1779
    },
    {
      "epoch": 3.9621380846325165,
      "loss": 0.7171025276184082,
      "loss_ce": 0.00030569382943212986,
      "loss_iou": 0.291015625,
      "loss_num": 0.0267333984375,
      "loss_xval": 0.71875,
      "num_input_tokens_seen": 99681052,
      "step": 1779
    },
    {
      "epoch": 3.9643652561247213,
      "grad_norm": 17.27662467956543,
      "learning_rate": 1e-06,
      "loss": 0.7603,
      "num_input_tokens_seen": 99738304,
      "step": 1780
    },
    {
      "epoch": 3.9643652561247213,
      "loss": 0.9583814740180969,
      "loss_ce": 0.0002515443484298885,
      "loss_iou": 0.431640625,
      "loss_num": 0.01904296875,
      "loss_xval": 0.95703125,
      "num_input_tokens_seen": 99738304,
      "step": 1780
    },
    {
      "epoch": 3.9665924276169267,
      "grad_norm": 16.905935287475586,
      "learning_rate": 1e-06,
      "loss": 0.701,
      "num_input_tokens_seen": 99793864,
      "step": 1781
    },
    {
      "epoch": 3.9665924276169267,
      "loss": 0.8442895412445068,
      "loss_ce": 0.0002953645307570696,
      "loss_iou": 0.359375,
      "loss_num": 0.025146484375,
      "loss_xval": 0.84375,
      "num_input_tokens_seen": 99793864,
      "step": 1781
    },
    {
      "epoch": 3.9688195991091315,
      "grad_norm": 17.269075393676758,
      "learning_rate": 1e-06,
      "loss": 0.6253,
      "num_input_tokens_seen": 99849596,
      "step": 1782
    },
    {
      "epoch": 3.9688195991091315,
      "loss": 0.6899313926696777,
      "loss_ce": 0.0004782522446475923,
      "loss_iou": 0.279296875,
      "loss_num": 0.0262451171875,
      "loss_xval": 0.6875,
      "num_input_tokens_seen": 99849596,
      "step": 1782
    },
    {
      "epoch": 3.9710467706013364,
      "grad_norm": 12.376049995422363,
      "learning_rate": 1e-06,
      "loss": 0.7759,
      "num_input_tokens_seen": 99905032,
      "step": 1783
    },
    {
      "epoch": 3.9710467706013364,
      "loss": 0.8750180602073669,
      "loss_ce": 0.0002622200991027057,
      "loss_iou": 0.361328125,
      "loss_num": 0.0302734375,
      "loss_xval": 0.875,
      "num_input_tokens_seen": 99905032,
      "step": 1783
    },
    {
      "epoch": 3.9732739420935412,
      "grad_norm": 15.985518455505371,
      "learning_rate": 1e-06,
      "loss": 0.7939,
      "num_input_tokens_seen": 99961684,
      "step": 1784
    },
    {
      "epoch": 3.9732739420935412,
      "loss": 0.7649087905883789,
      "loss_ce": 0.00026036915369331837,
      "loss_iou": 0.349609375,
      "loss_num": 0.0128173828125,
      "loss_xval": 0.765625,
      "num_input_tokens_seen": 99961684,
      "step": 1784
    },
    {
      "epoch": 3.975501113585746,
      "grad_norm": 17.415231704711914,
      "learning_rate": 1e-06,
      "loss": 0.9524,
      "num_input_tokens_seen": 100018708,
      "step": 1785
    },
    {
      "epoch": 3.975501113585746,
      "loss": 1.1517889499664307,
      "loss_ce": 0.00042183659388683736,
      "loss_iou": 0.48828125,
      "loss_num": 0.034912109375,
      "loss_xval": 1.1484375,
      "num_input_tokens_seen": 100018708,
      "step": 1785
    },
    {
      "epoch": 3.977728285077951,
      "grad_norm": 21.656478881835938,
      "learning_rate": 1e-06,
      "loss": 0.7554,
      "num_input_tokens_seen": 100073244,
      "step": 1786
    },
    {
      "epoch": 3.977728285077951,
      "loss": 0.634032130241394,
      "loss_ce": 0.0002430875028949231,
      "loss_iou": 0.287109375,
      "loss_num": 0.01214599609375,
      "loss_xval": 0.6328125,
      "num_input_tokens_seen": 100073244,
      "step": 1786
    },
    {
      "epoch": 3.979955456570156,
      "grad_norm": 55.99577713012695,
      "learning_rate": 1e-06,
      "loss": 1.0249,
      "num_input_tokens_seen": 100127112,
      "step": 1787
    },
    {
      "epoch": 3.979955456570156,
      "loss": 0.8672703504562378,
      "loss_ce": 0.00032695726258680224,
      "loss_iou": 0.328125,
      "loss_num": 0.04248046875,
      "loss_xval": 0.8671875,
      "num_input_tokens_seen": 100127112,
      "step": 1787
    },
    {
      "epoch": 3.9821826280623607,
      "grad_norm": 20.532896041870117,
      "learning_rate": 1e-06,
      "loss": 0.8107,
      "num_input_tokens_seen": 100181880,
      "step": 1788
    },
    {
      "epoch": 3.9821826280623607,
      "loss": 0.7985996603965759,
      "loss_ce": 0.000259819149505347,
      "loss_iou": 0.349609375,
      "loss_num": 0.02001953125,
      "loss_xval": 0.796875,
      "num_input_tokens_seen": 100181880,
      "step": 1788
    },
    {
      "epoch": 3.984409799554566,
      "grad_norm": 22.649169921875,
      "learning_rate": 1e-06,
      "loss": 0.7598,
      "num_input_tokens_seen": 100238956,
      "step": 1789
    },
    {
      "epoch": 3.984409799554566,
      "loss": 0.7128958702087402,
      "loss_ce": 0.00024937212583608925,
      "loss_iou": 0.30859375,
      "loss_num": 0.019287109375,
      "loss_xval": 0.7109375,
      "num_input_tokens_seen": 100238956,
      "step": 1789
    },
    {
      "epoch": 3.986636971046771,
      "grad_norm": 37.473751068115234,
      "learning_rate": 1e-06,
      "loss": 0.9002,
      "num_input_tokens_seen": 100293580,
      "step": 1790
    },
    {
      "epoch": 3.986636971046771,
      "loss": 0.7505241632461548,
      "loss_ce": 0.0005241355975158513,
      "loss_iou": 0.330078125,
      "loss_num": 0.017822265625,
      "loss_xval": 0.75,
      "num_input_tokens_seen": 100293580,
      "step": 1790
    },
    {
      "epoch": 3.9888641425389757,
      "grad_norm": 17.84164810180664,
      "learning_rate": 1e-06,
      "loss": 0.8214,
      "num_input_tokens_seen": 100347096,
      "step": 1791
    },
    {
      "epoch": 3.9888641425389757,
      "loss": 0.83009272813797,
      "loss_ce": 0.000258743908489123,
      "loss_iou": 0.32421875,
      "loss_num": 0.035888671875,
      "loss_xval": 0.828125,
      "num_input_tokens_seen": 100347096,
      "step": 1791
    },
    {
      "epoch": 3.9910913140311806,
      "grad_norm": 15.902947425842285,
      "learning_rate": 1e-06,
      "loss": 0.8695,
      "num_input_tokens_seen": 100403084,
      "step": 1792
    },
    {
      "epoch": 3.9910913140311806,
      "loss": 0.8725968599319458,
      "loss_ce": 0.0002824169350787997,
      "loss_iou": 0.376953125,
      "loss_num": 0.0238037109375,
      "loss_xval": 0.87109375,
      "num_input_tokens_seen": 100403084,
      "step": 1792
    },
    {
      "epoch": 3.9933184855233854,
      "grad_norm": 32.91238784790039,
      "learning_rate": 1e-06,
      "loss": 0.6368,
      "num_input_tokens_seen": 100460528,
      "step": 1793
    },
    {
      "epoch": 3.9933184855233854,
      "loss": 0.6155407428741455,
      "loss_ce": 0.00030632468406111,
      "loss_iou": 0.2578125,
      "loss_num": 0.020263671875,
      "loss_xval": 0.6171875,
      "num_input_tokens_seen": 100460528,
      "step": 1793
    },
    {
      "epoch": 3.9955456570155903,
      "grad_norm": 17.373069763183594,
      "learning_rate": 1e-06,
      "loss": 0.6237,
      "num_input_tokens_seen": 100518164,
      "step": 1794
    },
    {
      "epoch": 3.9955456570155903,
      "loss": 0.748991847038269,
      "loss_ce": 0.00045668811071664095,
      "loss_iou": 0.3125,
      "loss_num": 0.024658203125,
      "loss_xval": 0.75,
      "num_input_tokens_seen": 100518164,
      "step": 1794
    },
    {
      "epoch": 3.997772828507795,
      "grad_norm": 18.933780670166016,
      "learning_rate": 1e-06,
      "loss": 0.7479,
      "num_input_tokens_seen": 100576824,
      "step": 1795
    },
    {
      "epoch": 3.997772828507795,
      "loss": 0.8931920528411865,
      "loss_ce": 0.0003698251966852695,
      "loss_iou": 0.390625,
      "loss_num": 0.0220947265625,
      "loss_xval": 0.89453125,
      "num_input_tokens_seen": 100576824,
      "step": 1795
    },
    {
      "epoch": 4.0,
      "grad_norm": 18.814178466796875,
      "learning_rate": 1e-06,
      "loss": 0.8102,
      "num_input_tokens_seen": 100631512,
      "step": 1796
    },
    {
      "epoch": 4.0,
      "loss": 0.976672887802124,
      "loss_ce": 0.0003544443752616644,
      "loss_iou": 0.419921875,
      "loss_num": 0.0274658203125,
      "loss_xval": 0.9765625,
      "num_input_tokens_seen": 100631512,
      "step": 1796
    },
    {
      "epoch": 4.002227171492205,
      "grad_norm": 18.784257888793945,
      "learning_rate": 1e-06,
      "loss": 1.0043,
      "num_input_tokens_seen": 100688068,
      "step": 1797
    },
    {
      "epoch": 4.002227171492205,
      "loss": 1.069665551185608,
      "loss_ce": 0.00032957567600533366,
      "loss_iou": 0.431640625,
      "loss_num": 0.041015625,
      "loss_xval": 1.0703125,
      "num_input_tokens_seen": 100688068,
      "step": 1797
    },
    {
      "epoch": 4.00445434298441,
      "grad_norm": 15.86024284362793,
      "learning_rate": 1e-06,
      "loss": 0.9139,
      "num_input_tokens_seen": 100740940,
      "step": 1798
    },
    {
      "epoch": 4.00445434298441,
      "loss": 0.6887216567993164,
      "loss_ce": 0.00024506475892849267,
      "loss_iou": 0.302734375,
      "loss_num": 0.0166015625,
      "loss_xval": 0.6875,
      "num_input_tokens_seen": 100740940,
      "step": 1798
    },
    {
      "epoch": 4.006681514476615,
      "grad_norm": 41.829795837402344,
      "learning_rate": 1e-06,
      "loss": 0.7138,
      "num_input_tokens_seen": 100797540,
      "step": 1799
    },
    {
      "epoch": 4.006681514476615,
      "loss": 0.5916675329208374,
      "loss_ce": 0.00035895261680707335,
      "loss_iou": 0.248046875,
      "loss_num": 0.0191650390625,
      "loss_xval": 0.58984375,
      "num_input_tokens_seen": 100797540,
      "step": 1799
    },
    {
      "epoch": 4.008908685968819,
      "grad_norm": 15.94815731048584,
      "learning_rate": 1e-06,
      "loss": 0.8192,
      "num_input_tokens_seen": 100852428,
      "step": 1800
    },
    {
      "epoch": 4.008908685968819,
      "loss": 0.628261148929596,
      "loss_ce": 0.000331463961629197,
      "loss_iou": 0.259765625,
      "loss_num": 0.0218505859375,
      "loss_xval": 0.62890625,
      "num_input_tokens_seen": 100852428,
      "step": 1800
    },
    {
      "epoch": 4.011135857461024,
      "grad_norm": 17.08213233947754,
      "learning_rate": 1e-06,
      "loss": 0.6745,
      "num_input_tokens_seen": 100909988,
      "step": 1801
    },
    {
      "epoch": 4.011135857461024,
      "loss": 0.49927961826324463,
      "loss_ce": 0.00025619484949856997,
      "loss_iou": 0.2138671875,
      "loss_num": 0.0142822265625,
      "loss_xval": 0.5,
      "num_input_tokens_seen": 100909988,
      "step": 1801
    },
    {
      "epoch": 4.013363028953229,
      "grad_norm": 15.652112007141113,
      "learning_rate": 1e-06,
      "loss": 0.7803,
      "num_input_tokens_seen": 100967540,
      "step": 1802
    },
    {
      "epoch": 4.013363028953229,
      "loss": 0.6135426759719849,
      "loss_ce": 0.0002614644472487271,
      "loss_iou": 0.26171875,
      "loss_num": 0.0179443359375,
      "loss_xval": 0.61328125,
      "num_input_tokens_seen": 100967540,
      "step": 1802
    },
    {
      "epoch": 4.015590200445434,
      "grad_norm": 18.24171257019043,
      "learning_rate": 1e-06,
      "loss": 0.9651,
      "num_input_tokens_seen": 101023900,
      "step": 1803
    },
    {
      "epoch": 4.015590200445434,
      "loss": 0.9055318832397461,
      "loss_ce": 0.00025842676404863596,
      "loss_iou": 0.376953125,
      "loss_num": 0.030029296875,
      "loss_xval": 0.90625,
      "num_input_tokens_seen": 101023900,
      "step": 1803
    },
    {
      "epoch": 4.017817371937639,
      "grad_norm": 26.98381996154785,
      "learning_rate": 1e-06,
      "loss": 0.9335,
      "num_input_tokens_seen": 101081336,
      "step": 1804
    },
    {
      "epoch": 4.017817371937639,
      "loss": 0.942715048789978,
      "loss_ce": 0.00033223340869881213,
      "loss_iou": 0.3828125,
      "loss_num": 0.03564453125,
      "loss_xval": 0.94140625,
      "num_input_tokens_seen": 101081336,
      "step": 1804
    },
    {
      "epoch": 4.020044543429844,
      "grad_norm": 14.921833038330078,
      "learning_rate": 1e-06,
      "loss": 0.8849,
      "num_input_tokens_seen": 101139376,
      "step": 1805
    },
    {
      "epoch": 4.020044543429844,
      "loss": 1.0176324844360352,
      "loss_ce": 0.0007868300890550017,
      "loss_iou": 0.44921875,
      "loss_num": 0.024169921875,
      "loss_xval": 1.015625,
      "num_input_tokens_seen": 101139376,
      "step": 1805
    },
    {
      "epoch": 4.022271714922049,
      "grad_norm": 53.253265380859375,
      "learning_rate": 1e-06,
      "loss": 0.6965,
      "num_input_tokens_seen": 101197336,
      "step": 1806
    },
    {
      "epoch": 4.022271714922049,
      "loss": 0.5695779323577881,
      "loss_ce": 0.00024200681946240366,
      "loss_iou": 0.24609375,
      "loss_num": 0.0155029296875,
      "loss_xval": 0.5703125,
      "num_input_tokens_seen": 101197336,
      "step": 1806
    },
    {
      "epoch": 4.0244988864142535,
      "grad_norm": 12.943678855895996,
      "learning_rate": 1e-06,
      "loss": 0.6534,
      "num_input_tokens_seen": 101253816,
      "step": 1807
    },
    {
      "epoch": 4.0244988864142535,
      "loss": 0.6627930402755737,
      "loss_ce": 0.00043953658314421773,
      "loss_iou": 0.26953125,
      "loss_num": 0.0250244140625,
      "loss_xval": 0.6640625,
      "num_input_tokens_seen": 101253816,
      "step": 1807
    },
    {
      "epoch": 4.026726057906459,
      "grad_norm": 45.154788970947266,
      "learning_rate": 1e-06,
      "loss": 0.9206,
      "num_input_tokens_seen": 101307840,
      "step": 1808
    },
    {
      "epoch": 4.026726057906459,
      "loss": 0.8648847341537476,
      "loss_ce": 0.0002606944181025028,
      "loss_iou": 0.34375,
      "loss_num": 0.03564453125,
      "loss_xval": 0.86328125,
      "num_input_tokens_seen": 101307840,
      "step": 1808
    },
    {
      "epoch": 4.028953229398664,
      "grad_norm": 18.014551162719727,
      "learning_rate": 1e-06,
      "loss": 0.7334,
      "num_input_tokens_seen": 101363116,
      "step": 1809
    },
    {
      "epoch": 4.028953229398664,
      "loss": 0.5737777352333069,
      "loss_ce": 0.0002914007636718452,
      "loss_iou": 0.255859375,
      "loss_num": 0.0126953125,
      "loss_xval": 0.57421875,
      "num_input_tokens_seen": 101363116,
      "step": 1809
    },
    {
      "epoch": 4.031180400890869,
      "grad_norm": 18.99349021911621,
      "learning_rate": 1e-06,
      "loss": 0.8291,
      "num_input_tokens_seen": 101419912,
      "step": 1810
    },
    {
      "epoch": 4.031180400890869,
      "loss": 0.7891653180122375,
      "loss_ce": 0.0003469590446911752,
      "loss_iou": 0.30859375,
      "loss_num": 0.034423828125,
      "loss_xval": 0.7890625,
      "num_input_tokens_seen": 101419912,
      "step": 1810
    },
    {
      "epoch": 4.033407572383074,
      "grad_norm": 19.479049682617188,
      "learning_rate": 1e-06,
      "loss": 0.6363,
      "num_input_tokens_seen": 101478732,
      "step": 1811
    },
    {
      "epoch": 4.033407572383074,
      "loss": 0.5214917659759521,
      "loss_ce": 0.0002515409141778946,
      "loss_iou": 0.2236328125,
      "loss_num": 0.0146484375,
      "loss_xval": 0.51953125,
      "num_input_tokens_seen": 101478732,
      "step": 1811
    },
    {
      "epoch": 4.035634743875279,
      "grad_norm": 21.70351791381836,
      "learning_rate": 1e-06,
      "loss": 0.8533,
      "num_input_tokens_seen": 101536500,
      "step": 1812
    },
    {
      "epoch": 4.035634743875279,
      "loss": 0.8468989133834839,
      "loss_ce": 0.0003413098747842014,
      "loss_iou": 0.365234375,
      "loss_num": 0.0234375,
      "loss_xval": 0.84765625,
      "num_input_tokens_seen": 101536500,
      "step": 1812
    },
    {
      "epoch": 4.0378619153674835,
      "grad_norm": 15.182201385498047,
      "learning_rate": 1e-06,
      "loss": 0.9404,
      "num_input_tokens_seen": 101592900,
      "step": 1813
    },
    {
      "epoch": 4.0378619153674835,
      "loss": 1.334275484085083,
      "loss_ce": 0.00029111921321600676,
      "loss_iou": 0.59375,
      "loss_num": 0.030029296875,
      "loss_xval": 1.3359375,
      "num_input_tokens_seen": 101592900,
      "step": 1813
    },
    {
      "epoch": 4.040089086859688,
      "grad_norm": 18.792926788330078,
      "learning_rate": 1e-06,
      "loss": 0.7653,
      "num_input_tokens_seen": 101648332,
      "step": 1814
    },
    {
      "epoch": 4.040089086859688,
      "loss": 0.7761302590370178,
      "loss_ce": 0.0002513654180802405,
      "loss_iou": 0.328125,
      "loss_num": 0.0245361328125,
      "loss_xval": 0.77734375,
      "num_input_tokens_seen": 101648332,
      "step": 1814
    },
    {
      "epoch": 4.042316258351893,
      "grad_norm": 22.391029357910156,
      "learning_rate": 1e-06,
      "loss": 0.7972,
      "num_input_tokens_seen": 101702948,
      "step": 1815
    },
    {
      "epoch": 4.042316258351893,
      "loss": 0.9708037376403809,
      "loss_ce": 0.00034468824742361903,
      "loss_iou": 0.427734375,
      "loss_num": 0.0230712890625,
      "loss_xval": 0.96875,
      "num_input_tokens_seen": 101702948,
      "step": 1815
    },
    {
      "epoch": 4.044543429844098,
      "grad_norm": 19.350894927978516,
      "learning_rate": 1e-06,
      "loss": 0.8455,
      "num_input_tokens_seen": 101756712,
      "step": 1816
    },
    {
      "epoch": 4.044543429844098,
      "loss": 0.6643034219741821,
      "loss_ce": 0.00024090104852803051,
      "loss_iou": 0.275390625,
      "loss_num": 0.0228271484375,
      "loss_xval": 0.6640625,
      "num_input_tokens_seen": 101756712,
      "step": 1816
    },
    {
      "epoch": 4.046770601336303,
      "grad_norm": 18.292842864990234,
      "learning_rate": 1e-06,
      "loss": 0.8264,
      "num_input_tokens_seen": 101812980,
      "step": 1817
    },
    {
      "epoch": 4.046770601336303,
      "loss": 0.9234464168548584,
      "loss_ce": 0.00035073645994998515,
      "loss_iou": 0.392578125,
      "loss_num": 0.027587890625,
      "loss_xval": 0.921875,
      "num_input_tokens_seen": 101812980,
      "step": 1817
    },
    {
      "epoch": 4.048997772828508,
      "grad_norm": 19.12327003479004,
      "learning_rate": 1e-06,
      "loss": 0.7468,
      "num_input_tokens_seen": 101868528,
      "step": 1818
    },
    {
      "epoch": 4.048997772828508,
      "loss": 0.8360873460769653,
      "loss_ce": 0.0003940344322472811,
      "loss_iou": 0.349609375,
      "loss_num": 0.02734375,
      "loss_xval": 0.8359375,
      "num_input_tokens_seen": 101868528,
      "step": 1818
    },
    {
      "epoch": 4.051224944320713,
      "grad_norm": 15.702198028564453,
      "learning_rate": 1e-06,
      "loss": 0.6871,
      "num_input_tokens_seen": 101923832,
      "step": 1819
    },
    {
      "epoch": 4.051224944320713,
      "loss": 0.7370009422302246,
      "loss_ce": 0.0014051980106160045,
      "loss_iou": 0.30859375,
      "loss_num": 0.0234375,
      "loss_xval": 0.734375,
      "num_input_tokens_seen": 101923832,
      "step": 1819
    },
    {
      "epoch": 4.0534521158129175,
      "grad_norm": 21.061260223388672,
      "learning_rate": 1e-06,
      "loss": 0.7905,
      "num_input_tokens_seen": 101978472,
      "step": 1820
    },
    {
      "epoch": 4.0534521158129175,
      "loss": 0.8799552917480469,
      "loss_ce": 0.0006828849436715245,
      "loss_iou": 0.353515625,
      "loss_num": 0.034423828125,
      "loss_xval": 0.87890625,
      "num_input_tokens_seen": 101978472,
      "step": 1820
    },
    {
      "epoch": 4.055679287305122,
      "grad_norm": 23.500896453857422,
      "learning_rate": 1e-06,
      "loss": 0.9895,
      "num_input_tokens_seen": 102035200,
      "step": 1821
    },
    {
      "epoch": 4.055679287305122,
      "loss": 0.9768585562705994,
      "loss_ce": 0.0002960490819532424,
      "loss_iou": 0.40234375,
      "loss_num": 0.034423828125,
      "loss_xval": 0.9765625,
      "num_input_tokens_seen": 102035200,
      "step": 1821
    },
    {
      "epoch": 4.057906458797327,
      "grad_norm": 16.685222625732422,
      "learning_rate": 1e-06,
      "loss": 0.7959,
      "num_input_tokens_seen": 102088752,
      "step": 1822
    },
    {
      "epoch": 4.057906458797327,
      "loss": 0.8308522701263428,
      "loss_ce": 0.0002858432417269796,
      "loss_iou": 0.34375,
      "loss_num": 0.0283203125,
      "loss_xval": 0.83203125,
      "num_input_tokens_seen": 102088752,
      "step": 1822
    },
    {
      "epoch": 4.060133630289532,
      "grad_norm": 36.30082702636719,
      "learning_rate": 1e-06,
      "loss": 0.7506,
      "num_input_tokens_seen": 102139624,
      "step": 1823
    },
    {
      "epoch": 4.060133630289532,
      "loss": 0.8599013090133667,
      "loss_ce": 0.0002821852976921946,
      "loss_iou": 0.375,
      "loss_num": 0.02197265625,
      "loss_xval": 0.859375,
      "num_input_tokens_seen": 102139624,
      "step": 1823
    },
    {
      "epoch": 4.062360801781737,
      "grad_norm": 17.49234962463379,
      "learning_rate": 1e-06,
      "loss": 0.7182,
      "num_input_tokens_seen": 102196280,
      "step": 1824
    },
    {
      "epoch": 4.062360801781737,
      "loss": 0.7073689103126526,
      "loss_ce": 0.0003376654349267483,
      "loss_iou": 0.3125,
      "loss_num": 0.016845703125,
      "loss_xval": 0.70703125,
      "num_input_tokens_seen": 102196280,
      "step": 1824
    },
    {
      "epoch": 4.064587973273942,
      "grad_norm": 23.85994529724121,
      "learning_rate": 1e-06,
      "loss": 0.9767,
      "num_input_tokens_seen": 102252520,
      "step": 1825
    },
    {
      "epoch": 4.064587973273942,
      "loss": 1.1830354928970337,
      "loss_ce": 0.0002961636637337506,
      "loss_iou": 0.482421875,
      "loss_num": 0.043212890625,
      "loss_xval": 1.1796875,
      "num_input_tokens_seen": 102252520,
      "step": 1825
    },
    {
      "epoch": 4.066815144766147,
      "grad_norm": 15.231406211853027,
      "learning_rate": 1e-06,
      "loss": 0.7989,
      "num_input_tokens_seen": 102308852,
      "step": 1826
    },
    {
      "epoch": 4.066815144766147,
      "loss": 0.7088180780410767,
      "loss_ce": 0.0003219395875930786,
      "loss_iou": 0.291015625,
      "loss_num": 0.02490234375,
      "loss_xval": 0.70703125,
      "num_input_tokens_seen": 102308852,
      "step": 1826
    },
    {
      "epoch": 4.0690423162583516,
      "grad_norm": 20.627899169921875,
      "learning_rate": 1e-06,
      "loss": 0.8451,
      "num_input_tokens_seen": 102363256,
      "step": 1827
    },
    {
      "epoch": 4.0690423162583516,
      "loss": 0.43590307235717773,
      "loss_ce": 0.0003562027995940298,
      "loss_iou": 0.177734375,
      "loss_num": 0.0159912109375,
      "loss_xval": 0.435546875,
      "num_input_tokens_seen": 102363256,
      "step": 1827
    },
    {
      "epoch": 4.071269487750556,
      "grad_norm": 15.15736198425293,
      "learning_rate": 1e-06,
      "loss": 0.8284,
      "num_input_tokens_seen": 102417712,
      "step": 1828
    },
    {
      "epoch": 4.071269487750556,
      "loss": 0.876508355140686,
      "loss_ce": 0.0002876708167605102,
      "loss_iou": 0.390625,
      "loss_num": 0.01904296875,
      "loss_xval": 0.875,
      "num_input_tokens_seen": 102417712,
      "step": 1828
    },
    {
      "epoch": 4.073496659242761,
      "grad_norm": 16.91136360168457,
      "learning_rate": 1e-06,
      "loss": 0.7477,
      "num_input_tokens_seen": 102476200,
      "step": 1829
    },
    {
      "epoch": 4.073496659242761,
      "loss": 0.862647294998169,
      "loss_ce": 0.0003425973991397768,
      "loss_iou": 0.3828125,
      "loss_num": 0.0196533203125,
      "loss_xval": 0.86328125,
      "num_input_tokens_seen": 102476200,
      "step": 1829
    },
    {
      "epoch": 4.075723830734967,
      "grad_norm": 21.360063552856445,
      "learning_rate": 1e-06,
      "loss": 0.6998,
      "num_input_tokens_seen": 102533876,
      "step": 1830
    },
    {
      "epoch": 4.075723830734967,
      "loss": 0.8524512052536011,
      "loss_ce": 0.0004004453949164599,
      "loss_iou": 0.35546875,
      "loss_num": 0.0284423828125,
      "loss_xval": 0.8515625,
      "num_input_tokens_seen": 102533876,
      "step": 1830
    },
    {
      "epoch": 4.077951002227172,
      "grad_norm": 18.112483978271484,
      "learning_rate": 1e-06,
      "loss": 0.6848,
      "num_input_tokens_seen": 102590972,
      "step": 1831
    },
    {
      "epoch": 4.077951002227172,
      "loss": 0.6477828621864319,
      "loss_ce": 0.0003219213103875518,
      "loss_iou": 0.279296875,
      "loss_num": 0.0177001953125,
      "loss_xval": 0.6484375,
      "num_input_tokens_seen": 102590972,
      "step": 1831
    },
    {
      "epoch": 4.080178173719377,
      "grad_norm": 20.459156036376953,
      "learning_rate": 1e-06,
      "loss": 0.8724,
      "num_input_tokens_seen": 102648508,
      "step": 1832
    },
    {
      "epoch": 4.080178173719377,
      "loss": 0.8328180909156799,
      "loss_ce": 0.00029852997977286577,
      "loss_iou": 0.345703125,
      "loss_num": 0.0281982421875,
      "loss_xval": 0.83203125,
      "num_input_tokens_seen": 102648508,
      "step": 1832
    },
    {
      "epoch": 4.082405345211582,
      "grad_norm": 21.345972061157227,
      "learning_rate": 1e-06,
      "loss": 0.7649,
      "num_input_tokens_seen": 102705228,
      "step": 1833
    },
    {
      "epoch": 4.082405345211582,
      "loss": 0.7076017260551453,
      "loss_ce": 0.0003263298131059855,
      "loss_iou": 0.29296875,
      "loss_num": 0.0238037109375,
      "loss_xval": 0.70703125,
      "num_input_tokens_seen": 102705228,
      "step": 1833
    },
    {
      "epoch": 4.0846325167037865,
      "grad_norm": 17.683818817138672,
      "learning_rate": 1e-06,
      "loss": 0.8536,
      "num_input_tokens_seen": 102761536,
      "step": 1834
    },
    {
      "epoch": 4.0846325167037865,
      "loss": 0.8223095536231995,
      "loss_ce": 0.00028804835164919496,
      "loss_iou": 0.341796875,
      "loss_num": 0.027099609375,
      "loss_xval": 0.8203125,
      "num_input_tokens_seen": 102761536,
      "step": 1834
    },
    {
      "epoch": 4.086859688195991,
      "grad_norm": 20.6676082611084,
      "learning_rate": 1e-06,
      "loss": 0.9164,
      "num_input_tokens_seen": 102818308,
      "step": 1835
    },
    {
      "epoch": 4.086859688195991,
      "loss": 0.900152325630188,
      "loss_ce": 0.0002499477122910321,
      "loss_iou": 0.384765625,
      "loss_num": 0.0263671875,
      "loss_xval": 0.8984375,
      "num_input_tokens_seen": 102818308,
      "step": 1835
    },
    {
      "epoch": 4.089086859688196,
      "grad_norm": 16.989944458007812,
      "learning_rate": 1e-06,
      "loss": 0.7689,
      "num_input_tokens_seen": 102870992,
      "step": 1836
    },
    {
      "epoch": 4.089086859688196,
      "loss": 0.6965099573135376,
      "loss_ce": 0.00022090264246799052,
      "loss_iou": 0.2890625,
      "loss_num": 0.023681640625,
      "loss_xval": 0.6953125,
      "num_input_tokens_seen": 102870992,
      "step": 1836
    },
    {
      "epoch": 4.091314031180401,
      "grad_norm": 18.3848934173584,
      "learning_rate": 1e-06,
      "loss": 0.9445,
      "num_input_tokens_seen": 102925516,
      "step": 1837
    },
    {
      "epoch": 4.091314031180401,
      "loss": 0.8509833812713623,
      "loss_ce": 0.00027541263261809945,
      "loss_iou": 0.345703125,
      "loss_num": 0.03173828125,
      "loss_xval": 0.8515625,
      "num_input_tokens_seen": 102925516,
      "step": 1837
    },
    {
      "epoch": 4.093541202672606,
      "grad_norm": 19.348962783813477,
      "learning_rate": 1e-06,
      "loss": 1.0582,
      "num_input_tokens_seen": 102980780,
      "step": 1838
    },
    {
      "epoch": 4.093541202672606,
      "loss": 0.901675820350647,
      "loss_ce": 0.00030863762367516756,
      "loss_iou": 0.39453125,
      "loss_num": 0.022705078125,
      "loss_xval": 0.90234375,
      "num_input_tokens_seen": 102980780,
      "step": 1838
    },
    {
      "epoch": 4.095768374164811,
      "grad_norm": 18.351146697998047,
      "learning_rate": 1e-06,
      "loss": 0.6857,
      "num_input_tokens_seen": 103038208,
      "step": 1839
    },
    {
      "epoch": 4.095768374164811,
      "loss": 0.7659989595413208,
      "loss_ce": 0.00025187875144183636,
      "loss_iou": 0.310546875,
      "loss_num": 0.02880859375,
      "loss_xval": 0.765625,
      "num_input_tokens_seen": 103038208,
      "step": 1839
    },
    {
      "epoch": 4.097995545657016,
      "grad_norm": 15.127579689025879,
      "learning_rate": 1e-06,
      "loss": 0.8049,
      "num_input_tokens_seen": 103090740,
      "step": 1840
    },
    {
      "epoch": 4.097995545657016,
      "loss": 0.9890508055686951,
      "loss_ce": 0.0002812229795381427,
      "loss_iou": 0.439453125,
      "loss_num": 0.022216796875,
      "loss_xval": 0.98828125,
      "num_input_tokens_seen": 103090740,
      "step": 1840
    },
    {
      "epoch": 4.1002227171492205,
      "grad_norm": 25.431516647338867,
      "learning_rate": 1e-06,
      "loss": 0.7564,
      "num_input_tokens_seen": 103147724,
      "step": 1841
    },
    {
      "epoch": 4.1002227171492205,
      "loss": 0.5847120881080627,
      "loss_ce": 0.00023942730331327766,
      "loss_iou": 0.25390625,
      "loss_num": 0.0155029296875,
      "loss_xval": 0.5859375,
      "num_input_tokens_seen": 103147724,
      "step": 1841
    },
    {
      "epoch": 4.102449888641425,
      "grad_norm": 14.01602554321289,
      "learning_rate": 1e-06,
      "loss": 0.6801,
      "num_input_tokens_seen": 103206520,
      "step": 1842
    },
    {
      "epoch": 4.102449888641425,
      "loss": 0.661648690700531,
      "loss_ce": 0.004055918660014868,
      "loss_iou": 0.27734375,
      "loss_num": 0.0203857421875,
      "loss_xval": 0.65625,
      "num_input_tokens_seen": 103206520,
      "step": 1842
    },
    {
      "epoch": 4.10467706013363,
      "grad_norm": 30.822895050048828,
      "learning_rate": 1e-06,
      "loss": 0.8477,
      "num_input_tokens_seen": 103260592,
      "step": 1843
    },
    {
      "epoch": 4.10467706013363,
      "loss": 1.0794177055358887,
      "loss_ce": 0.0003160441410727799,
      "loss_iou": 0.48046875,
      "loss_num": 0.023681640625,
      "loss_xval": 1.078125,
      "num_input_tokens_seen": 103260592,
      "step": 1843
    },
    {
      "epoch": 4.106904231625835,
      "grad_norm": 19.740049362182617,
      "learning_rate": 1e-06,
      "loss": 0.8343,
      "num_input_tokens_seen": 103318044,
      "step": 1844
    },
    {
      "epoch": 4.106904231625835,
      "loss": 0.7112168669700623,
      "loss_ce": 0.00027938373386859894,
      "loss_iou": 0.3125,
      "loss_num": 0.01708984375,
      "loss_xval": 0.7109375,
      "num_input_tokens_seen": 103318044,
      "step": 1844
    },
    {
      "epoch": 4.10913140311804,
      "grad_norm": 23.76945686340332,
      "learning_rate": 1e-06,
      "loss": 0.5955,
      "num_input_tokens_seen": 103371804,
      "step": 1845
    },
    {
      "epoch": 4.10913140311804,
      "loss": 0.5928046107292175,
      "loss_ce": 0.00027534179389476776,
      "loss_iou": 0.240234375,
      "loss_num": 0.0224609375,
      "loss_xval": 0.59375,
      "num_input_tokens_seen": 103371804,
      "step": 1845
    },
    {
      "epoch": 4.111358574610245,
      "grad_norm": 19.958959579467773,
      "learning_rate": 1e-06,
      "loss": 0.7486,
      "num_input_tokens_seen": 103428372,
      "step": 1846
    },
    {
      "epoch": 4.111358574610245,
      "loss": 0.8037309646606445,
      "loss_ce": 0.00026413900195620954,
      "loss_iou": 0.375,
      "loss_num": 0.0108642578125,
      "loss_xval": 0.8046875,
      "num_input_tokens_seen": 103428372,
      "step": 1846
    },
    {
      "epoch": 4.11358574610245,
      "grad_norm": 15.336278915405273,
      "learning_rate": 1e-06,
      "loss": 0.6876,
      "num_input_tokens_seen": 103484976,
      "step": 1847
    },
    {
      "epoch": 4.11358574610245,
      "loss": 0.7304455637931824,
      "loss_ce": 0.0002209786616731435,
      "loss_iou": 0.302734375,
      "loss_num": 0.02490234375,
      "loss_xval": 0.73046875,
      "num_input_tokens_seen": 103484976,
      "step": 1847
    },
    {
      "epoch": 4.1158129175946545,
      "grad_norm": 13.736832618713379,
      "learning_rate": 1e-06,
      "loss": 0.8494,
      "num_input_tokens_seen": 103542500,
      "step": 1848
    },
    {
      "epoch": 4.1158129175946545,
      "loss": 0.8857908248901367,
      "loss_ce": 0.00041487094131298363,
      "loss_iou": 0.35546875,
      "loss_num": 0.034912109375,
      "loss_xval": 0.88671875,
      "num_input_tokens_seen": 103542500,
      "step": 1848
    },
    {
      "epoch": 4.118040089086859,
      "grad_norm": 18.131500244140625,
      "learning_rate": 1e-06,
      "loss": 0.9464,
      "num_input_tokens_seen": 103597080,
      "step": 1849
    },
    {
      "epoch": 4.118040089086859,
      "loss": 0.9518570899963379,
      "loss_ce": 0.0003189902927260846,
      "loss_iou": 0.38671875,
      "loss_num": 0.035888671875,
      "loss_xval": 0.953125,
      "num_input_tokens_seen": 103597080,
      "step": 1849
    },
    {
      "epoch": 4.120267260579064,
      "grad_norm": 18.16317367553711,
      "learning_rate": 1e-06,
      "loss": 1.0965,
      "num_input_tokens_seen": 103651008,
      "step": 1850
    },
    {
      "epoch": 4.120267260579064,
      "loss": 1.1296725273132324,
      "loss_ce": 0.00027801876422017813,
      "loss_iou": 0.458984375,
      "loss_num": 0.042236328125,
      "loss_xval": 1.1328125,
      "num_input_tokens_seen": 103651008,
      "step": 1850
    },
    {
      "epoch": 4.122494432071269,
      "grad_norm": 29.750585556030273,
      "learning_rate": 1e-06,
      "loss": 0.8961,
      "num_input_tokens_seen": 103709900,
      "step": 1851
    },
    {
      "epoch": 4.122494432071269,
      "loss": 0.8951970934867859,
      "loss_ce": 0.00042168618529103696,
      "loss_iou": 0.373046875,
      "loss_num": 0.0296630859375,
      "loss_xval": 0.89453125,
      "num_input_tokens_seen": 103709900,
      "step": 1851
    },
    {
      "epoch": 4.124721603563474,
      "grad_norm": 27.113773345947266,
      "learning_rate": 1e-06,
      "loss": 0.6256,
      "num_input_tokens_seen": 103766964,
      "step": 1852
    },
    {
      "epoch": 4.124721603563474,
      "loss": 0.6550110578536987,
      "loss_ce": 0.00022586948762182146,
      "loss_iou": 0.298828125,
      "loss_num": 0.01153564453125,
      "loss_xval": 0.65625,
      "num_input_tokens_seen": 103766964,
      "step": 1852
    },
    {
      "epoch": 4.12694877505568,
      "grad_norm": 15.007980346679688,
      "learning_rate": 1e-06,
      "loss": 1.0075,
      "num_input_tokens_seen": 103821128,
      "step": 1853
    },
    {
      "epoch": 4.12694877505568,
      "loss": 0.9304366707801819,
      "loss_ce": 0.00026090393657796085,
      "loss_iou": 0.3984375,
      "loss_num": 0.0267333984375,
      "loss_xval": 0.9296875,
      "num_input_tokens_seen": 103821128,
      "step": 1853
    },
    {
      "epoch": 4.129175946547885,
      "grad_norm": 16.54915428161621,
      "learning_rate": 1e-06,
      "loss": 0.8283,
      "num_input_tokens_seen": 103874556,
      "step": 1854
    },
    {
      "epoch": 4.129175946547885,
      "loss": 0.6098688840866089,
      "loss_ce": 0.00024977908469736576,
      "loss_iou": 0.25390625,
      "loss_num": 0.0201416015625,
      "loss_xval": 0.609375,
      "num_input_tokens_seen": 103874556,
      "step": 1854
    },
    {
      "epoch": 4.131403118040089,
      "grad_norm": 18.67920684814453,
      "learning_rate": 1e-06,
      "loss": 0.8078,
      "num_input_tokens_seen": 103929176,
      "step": 1855
    },
    {
      "epoch": 4.131403118040089,
      "loss": 0.6376897692680359,
      "loss_ce": 0.00023860471264924854,
      "loss_iou": 0.28515625,
      "loss_num": 0.01318359375,
      "loss_xval": 0.63671875,
      "num_input_tokens_seen": 103929176,
      "step": 1855
    },
    {
      "epoch": 4.133630289532294,
      "grad_norm": 14.864916801452637,
      "learning_rate": 1e-06,
      "loss": 0.7531,
      "num_input_tokens_seen": 103985216,
      "step": 1856
    },
    {
      "epoch": 4.133630289532294,
      "loss": 0.606116533279419,
      "loss_ce": 0.0002816014748532325,
      "loss_iou": 0.244140625,
      "loss_num": 0.023681640625,
      "loss_xval": 0.60546875,
      "num_input_tokens_seen": 103985216,
      "step": 1856
    },
    {
      "epoch": 4.135857461024499,
      "grad_norm": 15.250370025634766,
      "learning_rate": 1e-06,
      "loss": 0.7684,
      "num_input_tokens_seen": 104043664,
      "step": 1857
    },
    {
      "epoch": 4.135857461024499,
      "loss": 0.810298502445221,
      "loss_ce": 0.00023990226327441633,
      "loss_iou": 0.337890625,
      "loss_num": 0.02734375,
      "loss_xval": 0.80859375,
      "num_input_tokens_seen": 104043664,
      "step": 1857
    },
    {
      "epoch": 4.138084632516704,
      "grad_norm": 33.826820373535156,
      "learning_rate": 1e-06,
      "loss": 0.6527,
      "num_input_tokens_seen": 104101224,
      "step": 1858
    },
    {
      "epoch": 4.138084632516704,
      "loss": 0.5385033488273621,
      "loss_ce": 0.00029532582266256213,
      "loss_iou": 0.23046875,
      "loss_num": 0.01519775390625,
      "loss_xval": 0.5390625,
      "num_input_tokens_seen": 104101224,
      "step": 1858
    },
    {
      "epoch": 4.140311804008909,
      "grad_norm": 26.74302101135254,
      "learning_rate": 1e-06,
      "loss": 0.7609,
      "num_input_tokens_seen": 104159992,
      "step": 1859
    },
    {
      "epoch": 4.140311804008909,
      "loss": 0.7529861927032471,
      "loss_ce": 0.00030061625875532627,
      "loss_iou": 0.32421875,
      "loss_num": 0.0208740234375,
      "loss_xval": 0.75390625,
      "num_input_tokens_seen": 104159992,
      "step": 1859
    },
    {
      "epoch": 4.142538975501114,
      "grad_norm": 16.13300895690918,
      "learning_rate": 1e-06,
      "loss": 0.8149,
      "num_input_tokens_seen": 104217472,
      "step": 1860
    },
    {
      "epoch": 4.142538975501114,
      "loss": 0.7612708806991577,
      "loss_ce": 0.00028455047868192196,
      "loss_iou": 0.328125,
      "loss_num": 0.02099609375,
      "loss_xval": 0.76171875,
      "num_input_tokens_seen": 104217472,
      "step": 1860
    },
    {
      "epoch": 4.144766146993319,
      "grad_norm": 14.439784049987793,
      "learning_rate": 1e-06,
      "loss": 0.7813,
      "num_input_tokens_seen": 104274900,
      "step": 1861
    },
    {
      "epoch": 4.144766146993319,
      "loss": 0.7901536226272583,
      "loss_ce": 0.0003587682731449604,
      "loss_iou": 0.302734375,
      "loss_num": 0.037109375,
      "loss_xval": 0.7890625,
      "num_input_tokens_seen": 104274900,
      "step": 1861
    },
    {
      "epoch": 4.146993318485523,
      "grad_norm": 15.760991096496582,
      "learning_rate": 1e-06,
      "loss": 0.9137,
      "num_input_tokens_seen": 104330380,
      "step": 1862
    },
    {
      "epoch": 4.146993318485523,
      "loss": 0.831478476524353,
      "loss_ce": 0.0004237755201756954,
      "loss_iou": 0.33203125,
      "loss_num": 0.033203125,
      "loss_xval": 0.83203125,
      "num_input_tokens_seen": 104330380,
      "step": 1862
    },
    {
      "epoch": 4.149220489977728,
      "grad_norm": 16.417325973510742,
      "learning_rate": 1e-06,
      "loss": 0.796,
      "num_input_tokens_seen": 104386476,
      "step": 1863
    },
    {
      "epoch": 4.149220489977728,
      "loss": 0.8149293065071106,
      "loss_ce": 0.0002320479543413967,
      "loss_iou": 0.365234375,
      "loss_num": 0.0169677734375,
      "loss_xval": 0.81640625,
      "num_input_tokens_seen": 104386476,
      "step": 1863
    },
    {
      "epoch": 4.151447661469933,
      "grad_norm": 26.965940475463867,
      "learning_rate": 1e-06,
      "loss": 0.6169,
      "num_input_tokens_seen": 104442536,
      "step": 1864
    },
    {
      "epoch": 4.151447661469933,
      "loss": 0.6312210559844971,
      "loss_ce": 0.00023957040684763342,
      "loss_iou": 0.275390625,
      "loss_num": 0.0159912109375,
      "loss_xval": 0.6328125,
      "num_input_tokens_seen": 104442536,
      "step": 1864
    },
    {
      "epoch": 4.153674832962138,
      "grad_norm": 26.452974319458008,
      "learning_rate": 1e-06,
      "loss": 0.8482,
      "num_input_tokens_seen": 104497928,
      "step": 1865
    },
    {
      "epoch": 4.153674832962138,
      "loss": 0.9214756488800049,
      "loss_ce": 0.00033305271062999964,
      "loss_iou": 0.388671875,
      "loss_num": 0.0289306640625,
      "loss_xval": 0.921875,
      "num_input_tokens_seen": 104497928,
      "step": 1865
    },
    {
      "epoch": 4.155902004454343,
      "grad_norm": 22.56464385986328,
      "learning_rate": 1e-06,
      "loss": 0.807,
      "num_input_tokens_seen": 104553604,
      "step": 1866
    },
    {
      "epoch": 4.155902004454343,
      "loss": 0.9364668130874634,
      "loss_ce": 0.00030959240393713117,
      "loss_iou": 0.36328125,
      "loss_num": 0.042236328125,
      "loss_xval": 0.9375,
      "num_input_tokens_seen": 104553604,
      "step": 1866
    },
    {
      "epoch": 4.158129175946548,
      "grad_norm": 16.754383087158203,
      "learning_rate": 1e-06,
      "loss": 0.7331,
      "num_input_tokens_seen": 104609096,
      "step": 1867
    },
    {
      "epoch": 4.158129175946548,
      "loss": 0.7379820942878723,
      "loss_ce": 0.0003112004487775266,
      "loss_iou": 0.30859375,
      "loss_num": 0.0244140625,
      "loss_xval": 0.73828125,
      "num_input_tokens_seen": 104609096,
      "step": 1867
    },
    {
      "epoch": 4.160356347438753,
      "grad_norm": 21.52208709716797,
      "learning_rate": 1e-06,
      "loss": 0.9237,
      "num_input_tokens_seen": 104662120,
      "step": 1868
    },
    {
      "epoch": 4.160356347438753,
      "loss": 0.9280723333358765,
      "loss_ce": 0.00033793269540183246,
      "loss_iou": 0.37109375,
      "loss_num": 0.036865234375,
      "loss_xval": 0.9296875,
      "num_input_tokens_seen": 104662120,
      "step": 1868
    },
    {
      "epoch": 4.1625835189309575,
      "grad_norm": 21.401424407958984,
      "learning_rate": 1e-06,
      "loss": 0.6609,
      "num_input_tokens_seen": 104720356,
      "step": 1869
    },
    {
      "epoch": 4.1625835189309575,
      "loss": 0.6265453100204468,
      "loss_ce": 0.00032457264023832977,
      "loss_iou": 0.27734375,
      "loss_num": 0.01416015625,
      "loss_xval": 0.625,
      "num_input_tokens_seen": 104720356,
      "step": 1869
    },
    {
      "epoch": 4.164810690423162,
      "grad_norm": 20.714725494384766,
      "learning_rate": 1e-06,
      "loss": 0.8871,
      "num_input_tokens_seen": 104777020,
      "step": 1870
    },
    {
      "epoch": 4.164810690423162,
      "loss": 0.9072417616844177,
      "loss_ce": 0.00038138843956403434,
      "loss_iou": 0.3984375,
      "loss_num": 0.0220947265625,
      "loss_xval": 0.90625,
      "num_input_tokens_seen": 104777020,
      "step": 1870
    },
    {
      "epoch": 4.167037861915367,
      "grad_norm": 19.34101104736328,
      "learning_rate": 1e-06,
      "loss": 0.9844,
      "num_input_tokens_seen": 104832044,
      "step": 1871
    },
    {
      "epoch": 4.167037861915367,
      "loss": 1.2446290254592896,
      "loss_ce": 0.0004884034278802574,
      "loss_iou": 0.5234375,
      "loss_num": 0.038818359375,
      "loss_xval": 1.2421875,
      "num_input_tokens_seen": 104832044,
      "step": 1871
    },
    {
      "epoch": 4.169265033407572,
      "grad_norm": 22.785999298095703,
      "learning_rate": 1e-06,
      "loss": 0.7377,
      "num_input_tokens_seen": 104889672,
      "step": 1872
    },
    {
      "epoch": 4.169265033407572,
      "loss": 0.5561378002166748,
      "loss_ce": 0.0002295885351486504,
      "loss_iou": 0.224609375,
      "loss_num": 0.021484375,
      "loss_xval": 0.5546875,
      "num_input_tokens_seen": 104889672,
      "step": 1872
    },
    {
      "epoch": 4.171492204899777,
      "grad_norm": 24.524354934692383,
      "learning_rate": 1e-06,
      "loss": 0.7842,
      "num_input_tokens_seen": 104946972,
      "step": 1873
    },
    {
      "epoch": 4.171492204899777,
      "loss": 0.7845892906188965,
      "loss_ce": 0.0010199534008279443,
      "loss_iou": 0.314453125,
      "loss_num": 0.0311279296875,
      "loss_xval": 0.78515625,
      "num_input_tokens_seen": 104946972,
      "step": 1873
    },
    {
      "epoch": 4.173719376391982,
      "grad_norm": 21.119625091552734,
      "learning_rate": 1e-06,
      "loss": 0.7332,
      "num_input_tokens_seen": 105001336,
      "step": 1874
    },
    {
      "epoch": 4.173719376391982,
      "loss": 0.803102970123291,
      "loss_ce": 0.00036857047234661877,
      "loss_iou": 0.33984375,
      "loss_num": 0.024658203125,
      "loss_xval": 0.8046875,
      "num_input_tokens_seen": 105001336,
      "step": 1874
    },
    {
      "epoch": 4.1759465478841875,
      "grad_norm": 21.471126556396484,
      "learning_rate": 1e-06,
      "loss": 0.9933,
      "num_input_tokens_seen": 105056692,
      "step": 1875
    },
    {
      "epoch": 4.1759465478841875,
      "loss": 0.8526608943939209,
      "loss_ce": 0.0003660406800918281,
      "loss_iou": 0.369140625,
      "loss_num": 0.0228271484375,
      "loss_xval": 0.8515625,
      "num_input_tokens_seen": 105056692,
      "step": 1875
    },
    {
      "epoch": 4.178173719376392,
      "grad_norm": 16.687490463256836,
      "learning_rate": 1e-06,
      "loss": 0.8024,
      "num_input_tokens_seen": 105113712,
      "step": 1876
    },
    {
      "epoch": 4.178173719376392,
      "loss": 0.668755054473877,
      "loss_ce": 0.0002979860291816294,
      "loss_iou": 0.28515625,
      "loss_num": 0.01953125,
      "loss_xval": 0.66796875,
      "num_input_tokens_seen": 105113712,
      "step": 1876
    },
    {
      "epoch": 4.180400890868597,
      "grad_norm": 13.53281021118164,
      "learning_rate": 1e-06,
      "loss": 0.9623,
      "num_input_tokens_seen": 105169732,
      "step": 1877
    },
    {
      "epoch": 4.180400890868597,
      "loss": 0.9789065718650818,
      "loss_ce": 0.00026888877619057894,
      "loss_iou": 0.3828125,
      "loss_num": 0.04248046875,
      "loss_xval": 0.98046875,
      "num_input_tokens_seen": 105169732,
      "step": 1877
    },
    {
      "epoch": 4.182628062360802,
      "grad_norm": 21.907808303833008,
      "learning_rate": 1e-06,
      "loss": 0.8223,
      "num_input_tokens_seen": 105221624,
      "step": 1878
    },
    {
      "epoch": 4.182628062360802,
      "loss": 0.7880837917327881,
      "loss_ce": 0.00024200028565246612,
      "loss_iou": 0.33984375,
      "loss_num": 0.0216064453125,
      "loss_xval": 0.7890625,
      "num_input_tokens_seen": 105221624,
      "step": 1878
    },
    {
      "epoch": 4.184855233853007,
      "grad_norm": 28.09380340576172,
      "learning_rate": 1e-06,
      "loss": 1.0485,
      "num_input_tokens_seen": 105278504,
      "step": 1879
    },
    {
      "epoch": 4.184855233853007,
      "loss": 1.0873863697052002,
      "loss_ce": 0.00035019166534766555,
      "loss_iou": 0.458984375,
      "loss_num": 0.03369140625,
      "loss_xval": 1.0859375,
      "num_input_tokens_seen": 105278504,
      "step": 1879
    },
    {
      "epoch": 4.187082405345212,
      "grad_norm": 28.955883026123047,
      "learning_rate": 1e-06,
      "loss": 0.7196,
      "num_input_tokens_seen": 105335440,
      "step": 1880
    },
    {
      "epoch": 4.187082405345212,
      "loss": 0.7075688242912292,
      "loss_ce": 0.0002934132644440979,
      "loss_iou": 0.31640625,
      "loss_num": 0.01556396484375,
      "loss_xval": 0.70703125,
      "num_input_tokens_seen": 105335440,
      "step": 1880
    },
    {
      "epoch": 4.189309576837417,
      "grad_norm": 16.112701416015625,
      "learning_rate": 1e-06,
      "loss": 0.9407,
      "num_input_tokens_seen": 105391556,
      "step": 1881
    },
    {
      "epoch": 4.189309576837417,
      "loss": 1.010768175125122,
      "loss_ce": 0.00027016678359359503,
      "loss_iou": 0.365234375,
      "loss_num": 0.056396484375,
      "loss_xval": 1.0078125,
      "num_input_tokens_seen": 105391556,
      "step": 1881
    },
    {
      "epoch": 4.1915367483296215,
      "grad_norm": 12.533539772033691,
      "learning_rate": 1e-06,
      "loss": 0.8191,
      "num_input_tokens_seen": 105444616,
      "step": 1882
    },
    {
      "epoch": 4.1915367483296215,
      "loss": 0.994451642036438,
      "loss_ce": 0.00031105236848816276,
      "loss_iou": 0.431640625,
      "loss_num": 0.026123046875,
      "loss_xval": 0.9921875,
      "num_input_tokens_seen": 105444616,
      "step": 1882
    },
    {
      "epoch": 4.193763919821826,
      "grad_norm": 18.672292709350586,
      "learning_rate": 1e-06,
      "loss": 0.7595,
      "num_input_tokens_seen": 105502508,
      "step": 1883
    },
    {
      "epoch": 4.193763919821826,
      "loss": 0.787238359451294,
      "loss_ce": 0.00025101928622461855,
      "loss_iou": 0.326171875,
      "loss_num": 0.026611328125,
      "loss_xval": 0.78515625,
      "num_input_tokens_seen": 105502508,
      "step": 1883
    },
    {
      "epoch": 4.195991091314031,
      "grad_norm": 18.28311538696289,
      "learning_rate": 1e-06,
      "loss": 0.7337,
      "num_input_tokens_seen": 105559380,
      "step": 1884
    },
    {
      "epoch": 4.195991091314031,
      "loss": 0.6529708504676819,
      "loss_ce": 0.000993293710052967,
      "loss_iou": 0.283203125,
      "loss_num": 0.0169677734375,
      "loss_xval": 0.65234375,
      "num_input_tokens_seen": 105559380,
      "step": 1884
    },
    {
      "epoch": 4.198218262806236,
      "grad_norm": 23.031770706176758,
      "learning_rate": 1e-06,
      "loss": 0.7621,
      "num_input_tokens_seen": 105614328,
      "step": 1885
    },
    {
      "epoch": 4.198218262806236,
      "loss": 0.542103111743927,
      "loss_ce": 0.00023299809254240245,
      "loss_iou": 0.220703125,
      "loss_num": 0.0198974609375,
      "loss_xval": 0.54296875,
      "num_input_tokens_seen": 105614328,
      "step": 1885
    },
    {
      "epoch": 4.200445434298441,
      "grad_norm": 23.01361656188965,
      "learning_rate": 1e-06,
      "loss": 0.7555,
      "num_input_tokens_seen": 105670768,
      "step": 1886
    },
    {
      "epoch": 4.200445434298441,
      "loss": 0.7036082744598389,
      "loss_ce": 0.00023906549904495478,
      "loss_iou": 0.28125,
      "loss_num": 0.0281982421875,
      "loss_xval": 0.703125,
      "num_input_tokens_seen": 105670768,
      "step": 1886
    },
    {
      "epoch": 4.202672605790646,
      "grad_norm": 12.780369758605957,
      "learning_rate": 1e-06,
      "loss": 0.7383,
      "num_input_tokens_seen": 105727652,
      "step": 1887
    },
    {
      "epoch": 4.202672605790646,
      "loss": 0.7529648542404175,
      "loss_ce": 0.00027933655655942857,
      "loss_iou": 0.330078125,
      "loss_num": 0.018310546875,
      "loss_xval": 0.75390625,
      "num_input_tokens_seen": 105727652,
      "step": 1887
    },
    {
      "epoch": 4.204899777282851,
      "grad_norm": 17.18299674987793,
      "learning_rate": 1e-06,
      "loss": 0.775,
      "num_input_tokens_seen": 105784604,
      "step": 1888
    },
    {
      "epoch": 4.204899777282851,
      "loss": 0.6293338537216187,
      "loss_ce": 0.0003055296838283539,
      "loss_iou": 0.2353515625,
      "loss_num": 0.031494140625,
      "loss_xval": 0.62890625,
      "num_input_tokens_seen": 105784604,
      "step": 1888
    },
    {
      "epoch": 4.2071269487750556,
      "grad_norm": 48.9861946105957,
      "learning_rate": 1e-06,
      "loss": 0.9114,
      "num_input_tokens_seen": 105840236,
      "step": 1889
    },
    {
      "epoch": 4.2071269487750556,
      "loss": 1.198516607284546,
      "loss_ce": 0.0002743962104432285,
      "loss_iou": 0.48828125,
      "loss_num": 0.043701171875,
      "loss_xval": 1.1953125,
      "num_input_tokens_seen": 105840236,
      "step": 1889
    },
    {
      "epoch": 4.20935412026726,
      "grad_norm": 18.402477264404297,
      "learning_rate": 1e-06,
      "loss": 0.6669,
      "num_input_tokens_seen": 105895768,
      "step": 1890
    },
    {
      "epoch": 4.20935412026726,
      "loss": 0.6518987417221069,
      "loss_ce": 0.0002874084748327732,
      "loss_iou": 0.283203125,
      "loss_num": 0.017333984375,
      "loss_xval": 0.65234375,
      "num_input_tokens_seen": 105895768,
      "step": 1890
    },
    {
      "epoch": 4.211581291759465,
      "grad_norm": 19.16950035095215,
      "learning_rate": 1e-06,
      "loss": 0.9874,
      "num_input_tokens_seen": 105950604,
      "step": 1891
    },
    {
      "epoch": 4.211581291759465,
      "loss": 1.1375155448913574,
      "loss_ce": 0.00030853564385324717,
      "loss_iou": 0.462890625,
      "loss_num": 0.042236328125,
      "loss_xval": 1.140625,
      "num_input_tokens_seen": 105950604,
      "step": 1891
    },
    {
      "epoch": 4.21380846325167,
      "grad_norm": 12.357964515686035,
      "learning_rate": 1e-06,
      "loss": 0.6145,
      "num_input_tokens_seen": 106007040,
      "step": 1892
    },
    {
      "epoch": 4.21380846325167,
      "loss": 0.6721287965774536,
      "loss_ce": 0.000253824342507869,
      "loss_iou": 0.291015625,
      "loss_num": 0.0177001953125,
      "loss_xval": 0.671875,
      "num_input_tokens_seen": 106007040,
      "step": 1892
    },
    {
      "epoch": 4.216035634743875,
      "grad_norm": 16.27143669128418,
      "learning_rate": 1e-06,
      "loss": 0.9689,
      "num_input_tokens_seen": 106063332,
      "step": 1893
    },
    {
      "epoch": 4.216035634743875,
      "loss": 0.9934124946594238,
      "loss_ce": 0.00024840355035848916,
      "loss_iou": 0.43359375,
      "loss_num": 0.0255126953125,
      "loss_xval": 0.9921875,
      "num_input_tokens_seen": 106063332,
      "step": 1893
    },
    {
      "epoch": 4.21826280623608,
      "grad_norm": 14.465191841125488,
      "learning_rate": 1e-06,
      "loss": 0.8634,
      "num_input_tokens_seen": 106119888,
      "step": 1894
    },
    {
      "epoch": 4.21826280623608,
      "loss": 0.752957820892334,
      "loss_ce": 0.0002722431090660393,
      "loss_iou": 0.3046875,
      "loss_num": 0.02880859375,
      "loss_xval": 0.75390625,
      "num_input_tokens_seen": 106119888,
      "step": 1894
    },
    {
      "epoch": 4.220489977728285,
      "grad_norm": 13.513664245605469,
      "learning_rate": 1e-06,
      "loss": 0.7989,
      "num_input_tokens_seen": 106177424,
      "step": 1895
    },
    {
      "epoch": 4.220489977728285,
      "loss": 0.869196891784668,
      "loss_ce": 0.00030032347422093153,
      "loss_iou": 0.376953125,
      "loss_num": 0.022705078125,
      "loss_xval": 0.8671875,
      "num_input_tokens_seen": 106177424,
      "step": 1895
    },
    {
      "epoch": 4.22271714922049,
      "grad_norm": 18.78656005859375,
      "learning_rate": 1e-06,
      "loss": 0.7666,
      "num_input_tokens_seen": 106233892,
      "step": 1896
    },
    {
      "epoch": 4.22271714922049,
      "loss": 0.7184998989105225,
      "loss_ce": 0.00023819129273761064,
      "loss_iou": 0.32421875,
      "loss_num": 0.0142822265625,
      "loss_xval": 0.71875,
      "num_input_tokens_seen": 106233892,
      "step": 1896
    },
    {
      "epoch": 4.224944320712694,
      "grad_norm": 41.630496978759766,
      "learning_rate": 1e-06,
      "loss": 0.8828,
      "num_input_tokens_seen": 106289772,
      "step": 1897
    },
    {
      "epoch": 4.224944320712694,
      "loss": 0.8243512511253357,
      "loss_ce": 0.00037670054007321596,
      "loss_iou": 0.35546875,
      "loss_num": 0.0230712890625,
      "loss_xval": 0.82421875,
      "num_input_tokens_seen": 106289772,
      "step": 1897
    },
    {
      "epoch": 4.2271714922049,
      "grad_norm": 14.743143081665039,
      "learning_rate": 1e-06,
      "loss": 0.9489,
      "num_input_tokens_seen": 106344316,
      "step": 1898
    },
    {
      "epoch": 4.2271714922049,
      "loss": 0.888018012046814,
      "loss_ce": 0.0003226427361369133,
      "loss_iou": 0.41015625,
      "loss_num": 0.0130615234375,
      "loss_xval": 0.88671875,
      "num_input_tokens_seen": 106344316,
      "step": 1898
    },
    {
      "epoch": 4.229398663697105,
      "grad_norm": 16.215017318725586,
      "learning_rate": 1e-06,
      "loss": 0.8319,
      "num_input_tokens_seen": 106399904,
      "step": 1899
    },
    {
      "epoch": 4.229398663697105,
      "loss": 0.7622650265693665,
      "loss_ce": 0.00030216382583603263,
      "loss_iou": 0.349609375,
      "loss_num": 0.0125732421875,
      "loss_xval": 0.76171875,
      "num_input_tokens_seen": 106399904,
      "step": 1899
    },
    {
      "epoch": 4.23162583518931,
      "grad_norm": 21.640851974487305,
      "learning_rate": 1e-06,
      "loss": 0.7937,
      "num_input_tokens_seen": 106456716,
      "step": 1900
    },
    {
      "epoch": 4.23162583518931,
      "loss": 0.8867244124412537,
      "loss_ce": 0.0002498445101082325,
      "loss_iou": 0.365234375,
      "loss_num": 0.03173828125,
      "loss_xval": 0.88671875,
      "num_input_tokens_seen": 106456716,
      "step": 1900
    },
    {
      "epoch": 4.233853006681515,
      "grad_norm": 45.25251770019531,
      "learning_rate": 1e-06,
      "loss": 0.9745,
      "num_input_tokens_seen": 106514668,
      "step": 1901
    },
    {
      "epoch": 4.233853006681515,
      "loss": 0.8965962529182434,
      "loss_ce": 0.0003560082404874265,
      "loss_iou": 0.38671875,
      "loss_num": 0.0242919921875,
      "loss_xval": 0.89453125,
      "num_input_tokens_seen": 106514668,
      "step": 1901
    },
    {
      "epoch": 4.23608017817372,
      "grad_norm": 23.69890594482422,
      "learning_rate": 1e-06,
      "loss": 0.6128,
      "num_input_tokens_seen": 106570596,
      "step": 1902
    },
    {
      "epoch": 4.23608017817372,
      "loss": 0.7368512153625488,
      "loss_ce": 0.0002789643476717174,
      "loss_iou": 0.318359375,
      "loss_num": 0.0201416015625,
      "loss_xval": 0.73828125,
      "num_input_tokens_seen": 106570596,
      "step": 1902
    },
    {
      "epoch": 4.2383073496659245,
      "grad_norm": 14.473897933959961,
      "learning_rate": 1e-06,
      "loss": 0.8265,
      "num_input_tokens_seen": 106625068,
      "step": 1903
    },
    {
      "epoch": 4.2383073496659245,
      "loss": 0.7507649660110474,
      "loss_ce": 0.00027672675787471235,
      "loss_iou": 0.310546875,
      "loss_num": 0.025634765625,
      "loss_xval": 0.75,
      "num_input_tokens_seen": 106625068,
      "step": 1903
    },
    {
      "epoch": 4.240534521158129,
      "grad_norm": 15.77379322052002,
      "learning_rate": 1e-06,
      "loss": 0.651,
      "num_input_tokens_seen": 106683920,
      "step": 1904
    },
    {
      "epoch": 4.240534521158129,
      "loss": 0.5201445817947388,
      "loss_ce": 0.0002471142797730863,
      "loss_iou": 0.2265625,
      "loss_num": 0.0133056640625,
      "loss_xval": 0.51953125,
      "num_input_tokens_seen": 106683920,
      "step": 1904
    },
    {
      "epoch": 4.242761692650334,
      "grad_norm": 16.723033905029297,
      "learning_rate": 1e-06,
      "loss": 0.6274,
      "num_input_tokens_seen": 106740932,
      "step": 1905
    },
    {
      "epoch": 4.242761692650334,
      "loss": 0.5735186338424683,
      "loss_ce": 0.00027646025409922004,
      "loss_iou": 0.2578125,
      "loss_num": 0.0115966796875,
      "loss_xval": 0.57421875,
      "num_input_tokens_seen": 106740932,
      "step": 1905
    },
    {
      "epoch": 4.244988864142539,
      "grad_norm": 23.549467086791992,
      "learning_rate": 1e-06,
      "loss": 0.9234,
      "num_input_tokens_seen": 106798536,
      "step": 1906
    },
    {
      "epoch": 4.244988864142539,
      "loss": 0.8834173679351807,
      "loss_ce": 0.0006048293435014784,
      "loss_iou": 0.33984375,
      "loss_num": 0.040771484375,
      "loss_xval": 0.8828125,
      "num_input_tokens_seen": 106798536,
      "step": 1906
    },
    {
      "epoch": 4.247216035634744,
      "grad_norm": 20.045562744140625,
      "learning_rate": 1e-06,
      "loss": 0.9631,
      "num_input_tokens_seen": 106855984,
      "step": 1907
    },
    {
      "epoch": 4.247216035634744,
      "loss": 1.116750955581665,
      "loss_ce": 0.0002958837430924177,
      "loss_iou": 0.423828125,
      "loss_num": 0.053466796875,
      "loss_xval": 1.1171875,
      "num_input_tokens_seen": 106855984,
      "step": 1907
    },
    {
      "epoch": 4.249443207126949,
      "grad_norm": 20.793867111206055,
      "learning_rate": 1e-06,
      "loss": 0.7098,
      "num_input_tokens_seen": 106910552,
      "step": 1908
    },
    {
      "epoch": 4.249443207126949,
      "loss": 0.691659152507782,
      "loss_ce": 0.0002528924378566444,
      "loss_iou": 0.318359375,
      "loss_num": 0.01092529296875,
      "loss_xval": 0.69140625,
      "num_input_tokens_seen": 106910552,
      "step": 1908
    },
    {
      "epoch": 4.251670378619154,
      "grad_norm": 31.35016632080078,
      "learning_rate": 1e-06,
      "loss": 0.8538,
      "num_input_tokens_seen": 106966684,
      "step": 1909
    },
    {
      "epoch": 4.251670378619154,
      "loss": 0.7839844226837158,
      "loss_ce": 0.000293065735604614,
      "loss_iou": 0.34765625,
      "loss_num": 0.0174560546875,
      "loss_xval": 0.78515625,
      "num_input_tokens_seen": 106966684,
      "step": 1909
    },
    {
      "epoch": 4.2538975501113585,
      "grad_norm": 34.71610641479492,
      "learning_rate": 1e-06,
      "loss": 0.6133,
      "num_input_tokens_seen": 107025092,
      "step": 1910
    },
    {
      "epoch": 4.2538975501113585,
      "loss": 0.6420915722846985,
      "loss_ce": 0.0002459048992022872,
      "loss_iou": 0.27734375,
      "loss_num": 0.0174560546875,
      "loss_xval": 0.640625,
      "num_input_tokens_seen": 107025092,
      "step": 1910
    },
    {
      "epoch": 4.256124721603563,
      "grad_norm": 14.96351146697998,
      "learning_rate": 1e-06,
      "loss": 0.6374,
      "num_input_tokens_seen": 107081992,
      "step": 1911
    },
    {
      "epoch": 4.256124721603563,
      "loss": 0.4527207911014557,
      "loss_ce": 0.00032822368666529655,
      "loss_iou": 0.1943359375,
      "loss_num": 0.0125732421875,
      "loss_xval": 0.453125,
      "num_input_tokens_seen": 107081992,
      "step": 1911
    },
    {
      "epoch": 4.258351893095768,
      "grad_norm": 14.8378267288208,
      "learning_rate": 1e-06,
      "loss": 0.7082,
      "num_input_tokens_seen": 107135848,
      "step": 1912
    },
    {
      "epoch": 4.258351893095768,
      "loss": 0.5522253513336182,
      "loss_ce": 0.00022335897665470839,
      "loss_iou": 0.21875,
      "loss_num": 0.0230712890625,
      "loss_xval": 0.55078125,
      "num_input_tokens_seen": 107135848,
      "step": 1912
    },
    {
      "epoch": 4.260579064587973,
      "grad_norm": 21.59320068359375,
      "learning_rate": 1e-06,
      "loss": 0.8173,
      "num_input_tokens_seen": 107191440,
      "step": 1913
    },
    {
      "epoch": 4.260579064587973,
      "loss": 1.090116024017334,
      "loss_ce": 0.0002723358920775354,
      "loss_iou": 0.486328125,
      "loss_num": 0.0235595703125,
      "loss_xval": 1.09375,
      "num_input_tokens_seen": 107191440,
      "step": 1913
    },
    {
      "epoch": 4.262806236080178,
      "grad_norm": 16.01348876953125,
      "learning_rate": 1e-06,
      "loss": 0.8553,
      "num_input_tokens_seen": 107247028,
      "step": 1914
    },
    {
      "epoch": 4.262806236080178,
      "loss": 0.773563027381897,
      "loss_ce": 0.0002476052031852305,
      "loss_iou": 0.296875,
      "loss_num": 0.0361328125,
      "loss_xval": 0.7734375,
      "num_input_tokens_seen": 107247028,
      "step": 1914
    },
    {
      "epoch": 4.265033407572383,
      "grad_norm": 16.764503479003906,
      "learning_rate": 1e-06,
      "loss": 1.0386,
      "num_input_tokens_seen": 107301488,
      "step": 1915
    },
    {
      "epoch": 4.265033407572383,
      "loss": 1.2570042610168457,
      "loss_ce": 0.0004123870749026537,
      "loss_iou": 0.515625,
      "loss_num": 0.04443359375,
      "loss_xval": 1.2578125,
      "num_input_tokens_seen": 107301488,
      "step": 1915
    },
    {
      "epoch": 4.267260579064588,
      "grad_norm": 17.2139949798584,
      "learning_rate": 1e-06,
      "loss": 0.7674,
      "num_input_tokens_seen": 107357008,
      "step": 1916
    },
    {
      "epoch": 4.267260579064588,
      "loss": 0.8225438594818115,
      "loss_ce": 0.0002781503717415035,
      "loss_iou": 0.34765625,
      "loss_num": 0.0252685546875,
      "loss_xval": 0.8203125,
      "num_input_tokens_seen": 107357008,
      "step": 1916
    },
    {
      "epoch": 4.2694877505567925,
      "grad_norm": 18.4807186126709,
      "learning_rate": 1e-06,
      "loss": 0.4431,
      "num_input_tokens_seen": 107411784,
      "step": 1917
    },
    {
      "epoch": 4.2694877505567925,
      "loss": 0.4435253143310547,
      "loss_ce": 0.00041007628897204995,
      "loss_iou": 0.1630859375,
      "loss_num": 0.0234375,
      "loss_xval": 0.443359375,
      "num_input_tokens_seen": 107411784,
      "step": 1917
    },
    {
      "epoch": 4.271714922048997,
      "grad_norm": 16.725303649902344,
      "learning_rate": 1e-06,
      "loss": 0.7385,
      "num_input_tokens_seen": 107463544,
      "step": 1918
    },
    {
      "epoch": 4.271714922048997,
      "loss": 0.7992796897888184,
      "loss_ce": 0.00032942870166152716,
      "loss_iou": 0.3515625,
      "loss_num": 0.01904296875,
      "loss_xval": 0.80078125,
      "num_input_tokens_seen": 107463544,
      "step": 1918
    },
    {
      "epoch": 4.273942093541203,
      "grad_norm": 17.902894973754883,
      "learning_rate": 1e-06,
      "loss": 0.6089,
      "num_input_tokens_seen": 107518944,
      "step": 1919
    },
    {
      "epoch": 4.273942093541203,
      "loss": 0.7945454120635986,
      "loss_ce": 0.0002338896010769531,
      "loss_iou": 0.328125,
      "loss_num": 0.027587890625,
      "loss_xval": 0.79296875,
      "num_input_tokens_seen": 107518944,
      "step": 1919
    },
    {
      "epoch": 4.276169265033408,
      "grad_norm": 21.469985961914062,
      "learning_rate": 1e-06,
      "loss": 0.7481,
      "num_input_tokens_seen": 107574840,
      "step": 1920
    },
    {
      "epoch": 4.276169265033408,
      "loss": 0.7355300188064575,
      "loss_ce": 0.0003004681202583015,
      "loss_iou": 0.310546875,
      "loss_num": 0.0228271484375,
      "loss_xval": 0.734375,
      "num_input_tokens_seen": 107574840,
      "step": 1920
    },
    {
      "epoch": 4.278396436525613,
      "grad_norm": 18.499727249145508,
      "learning_rate": 1e-06,
      "loss": 0.8113,
      "num_input_tokens_seen": 107629476,
      "step": 1921
    },
    {
      "epoch": 4.278396436525613,
      "loss": 0.7343944311141968,
      "loss_ce": 0.00026357907336205244,
      "loss_iou": 0.30859375,
      "loss_num": 0.02294921875,
      "loss_xval": 0.734375,
      "num_input_tokens_seen": 107629476,
      "step": 1921
    },
    {
      "epoch": 4.280623608017818,
      "grad_norm": 25.940811157226562,
      "learning_rate": 1e-06,
      "loss": 0.7132,
      "num_input_tokens_seen": 107686892,
      "step": 1922
    },
    {
      "epoch": 4.280623608017818,
      "loss": 0.6143178343772888,
      "loss_ce": 0.0003041609888896346,
      "loss_iou": 0.2578125,
      "loss_num": 0.0194091796875,
      "loss_xval": 0.61328125,
      "num_input_tokens_seen": 107686892,
      "step": 1922
    },
    {
      "epoch": 4.282850779510023,
      "grad_norm": 21.650733947753906,
      "learning_rate": 1e-06,
      "loss": 0.726,
      "num_input_tokens_seen": 107743916,
      "step": 1923
    },
    {
      "epoch": 4.282850779510023,
      "loss": 0.6625887751579285,
      "loss_ce": 0.00023527286248281598,
      "loss_iou": 0.27734375,
      "loss_num": 0.02197265625,
      "loss_xval": 0.6640625,
      "num_input_tokens_seen": 107743916,
      "step": 1923
    },
    {
      "epoch": 4.285077951002227,
      "grad_norm": 15.383871078491211,
      "learning_rate": 1e-06,
      "loss": 0.7414,
      "num_input_tokens_seen": 107800000,
      "step": 1924
    },
    {
      "epoch": 4.285077951002227,
      "loss": 0.7688758373260498,
      "loss_ce": 0.0003211286966688931,
      "loss_iou": 0.33203125,
      "loss_num": 0.02099609375,
      "loss_xval": 0.76953125,
      "num_input_tokens_seen": 107800000,
      "step": 1924
    },
    {
      "epoch": 4.287305122494432,
      "grad_norm": 18.70506477355957,
      "learning_rate": 1e-06,
      "loss": 0.6572,
      "num_input_tokens_seen": 107856952,
      "step": 1925
    },
    {
      "epoch": 4.287305122494432,
      "loss": 0.5217354893684387,
      "loss_ce": 0.0002510941121727228,
      "loss_iou": 0.2314453125,
      "loss_num": 0.01177978515625,
      "loss_xval": 0.5234375,
      "num_input_tokens_seen": 107856952,
      "step": 1925
    },
    {
      "epoch": 4.289532293986637,
      "grad_norm": 15.325700759887695,
      "learning_rate": 1e-06,
      "loss": 0.8531,
      "num_input_tokens_seen": 107914656,
      "step": 1926
    },
    {
      "epoch": 4.289532293986637,
      "loss": 0.7412204146385193,
      "loss_ce": 0.0002536483807489276,
      "loss_iou": 0.328125,
      "loss_num": 0.0169677734375,
      "loss_xval": 0.7421875,
      "num_input_tokens_seen": 107914656,
      "step": 1926
    },
    {
      "epoch": 4.291759465478842,
      "grad_norm": 36.90719985961914,
      "learning_rate": 1e-06,
      "loss": 1.0012,
      "num_input_tokens_seen": 107971084,
      "step": 1927
    },
    {
      "epoch": 4.291759465478842,
      "loss": 0.8567878007888794,
      "loss_ce": 0.00034247490111738443,
      "loss_iou": 0.376953125,
      "loss_num": 0.0206298828125,
      "loss_xval": 0.85546875,
      "num_input_tokens_seen": 107971084,
      "step": 1927
    },
    {
      "epoch": 4.293986636971047,
      "grad_norm": 21.110570907592773,
      "learning_rate": 1e-06,
      "loss": 0.5608,
      "num_input_tokens_seen": 108027536,
      "step": 1928
    },
    {
      "epoch": 4.293986636971047,
      "loss": 0.49611085653305054,
      "loss_ce": 0.0002612844982650131,
      "loss_iou": 0.2041015625,
      "loss_num": 0.017578125,
      "loss_xval": 0.49609375,
      "num_input_tokens_seen": 108027536,
      "step": 1928
    },
    {
      "epoch": 4.296213808463252,
      "grad_norm": 19.765779495239258,
      "learning_rate": 1e-06,
      "loss": 0.6742,
      "num_input_tokens_seen": 108086688,
      "step": 1929
    },
    {
      "epoch": 4.296213808463252,
      "loss": 0.7515699863433838,
      "loss_ce": 0.0002272275451105088,
      "loss_iou": 0.32421875,
      "loss_num": 0.0203857421875,
      "loss_xval": 0.75,
      "num_input_tokens_seen": 108086688,
      "step": 1929
    },
    {
      "epoch": 4.298440979955457,
      "grad_norm": 27.658580780029297,
      "learning_rate": 1e-06,
      "loss": 0.8779,
      "num_input_tokens_seen": 108138612,
      "step": 1930
    },
    {
      "epoch": 4.298440979955457,
      "loss": 0.9459295272827148,
      "loss_ce": 0.0002508389297872782,
      "loss_iou": 0.40625,
      "loss_num": 0.0267333984375,
      "loss_xval": 0.9453125,
      "num_input_tokens_seen": 108138612,
      "step": 1930
    },
    {
      "epoch": 4.3006681514476615,
      "grad_norm": 18.092632293701172,
      "learning_rate": 1e-06,
      "loss": 0.9169,
      "num_input_tokens_seen": 108197300,
      "step": 1931
    },
    {
      "epoch": 4.3006681514476615,
      "loss": 1.1985375881195068,
      "loss_ce": 0.00029541528783738613,
      "loss_iou": 0.46875,
      "loss_num": 0.0517578125,
      "loss_xval": 1.1953125,
      "num_input_tokens_seen": 108197300,
      "step": 1931
    },
    {
      "epoch": 4.302895322939866,
      "grad_norm": 22.87906265258789,
      "learning_rate": 1e-06,
      "loss": 0.7375,
      "num_input_tokens_seen": 108253652,
      "step": 1932
    },
    {
      "epoch": 4.302895322939866,
      "loss": 0.6818982362747192,
      "loss_ce": 0.0002576185797806829,
      "loss_iou": 0.30078125,
      "loss_num": 0.015625,
      "loss_xval": 0.6796875,
      "num_input_tokens_seen": 108253652,
      "step": 1932
    },
    {
      "epoch": 4.305122494432071,
      "grad_norm": 19.443403244018555,
      "learning_rate": 1e-06,
      "loss": 0.7365,
      "num_input_tokens_seen": 108306668,
      "step": 1933
    },
    {
      "epoch": 4.305122494432071,
      "loss": 0.8623777627944946,
      "loss_ce": 0.0003171888238284737,
      "loss_iou": 0.376953125,
      "loss_num": 0.0216064453125,
      "loss_xval": 0.86328125,
      "num_input_tokens_seen": 108306668,
      "step": 1933
    },
    {
      "epoch": 4.307349665924276,
      "grad_norm": 19.513713836669922,
      "learning_rate": 1e-06,
      "loss": 0.8245,
      "num_input_tokens_seen": 108364872,
      "step": 1934
    },
    {
      "epoch": 4.307349665924276,
      "loss": 0.8435297608375549,
      "loss_ce": 0.00026804511435329914,
      "loss_iou": 0.3671875,
      "loss_num": 0.0220947265625,
      "loss_xval": 0.84375,
      "num_input_tokens_seen": 108364872,
      "step": 1934
    },
    {
      "epoch": 4.309576837416481,
      "grad_norm": 17.910724639892578,
      "learning_rate": 1e-06,
      "loss": 0.8137,
      "num_input_tokens_seen": 108421172,
      "step": 1935
    },
    {
      "epoch": 4.309576837416481,
      "loss": 1.0195529460906982,
      "loss_ce": 0.000510023906826973,
      "loss_iou": 0.41796875,
      "loss_num": 0.03662109375,
      "loss_xval": 1.015625,
      "num_input_tokens_seen": 108421172,
      "step": 1935
    },
    {
      "epoch": 4.311804008908686,
      "grad_norm": 20.821613311767578,
      "learning_rate": 1e-06,
      "loss": 0.7728,
      "num_input_tokens_seen": 108476148,
      "step": 1936
    },
    {
      "epoch": 4.311804008908686,
      "loss": 0.7290493249893188,
      "loss_ce": 0.00028955648303963244,
      "loss_iou": 0.2890625,
      "loss_num": 0.0301513671875,
      "loss_xval": 0.73046875,
      "num_input_tokens_seen": 108476148,
      "step": 1936
    },
    {
      "epoch": 4.314031180400891,
      "grad_norm": 14.239187240600586,
      "learning_rate": 1e-06,
      "loss": 0.7459,
      "num_input_tokens_seen": 108530832,
      "step": 1937
    },
    {
      "epoch": 4.314031180400891,
      "loss": 0.7683401107788086,
      "loss_ce": 0.0002737450413405895,
      "loss_iou": 0.34375,
      "loss_num": 0.0164794921875,
      "loss_xval": 0.76953125,
      "num_input_tokens_seen": 108530832,
      "step": 1937
    },
    {
      "epoch": 4.3162583518930955,
      "grad_norm": 22.064966201782227,
      "learning_rate": 1e-06,
      "loss": 0.827,
      "num_input_tokens_seen": 108586484,
      "step": 1938
    },
    {
      "epoch": 4.3162583518930955,
      "loss": 0.7754453420639038,
      "loss_ce": 0.00029886612901464105,
      "loss_iou": 0.349609375,
      "loss_num": 0.0150146484375,
      "loss_xval": 0.7734375,
      "num_input_tokens_seen": 108586484,
      "step": 1938
    },
    {
      "epoch": 4.3184855233853,
      "grad_norm": 22.323854446411133,
      "learning_rate": 1e-06,
      "loss": 0.6395,
      "num_input_tokens_seen": 108643224,
      "step": 1939
    },
    {
      "epoch": 4.3184855233853,
      "loss": 0.7868392467498779,
      "loss_ce": 0.000950582732912153,
      "loss_iou": 0.34765625,
      "loss_num": 0.0184326171875,
      "loss_xval": 0.78515625,
      "num_input_tokens_seen": 108643224,
      "step": 1939
    },
    {
      "epoch": 4.320712694877505,
      "grad_norm": 18.545169830322266,
      "learning_rate": 1e-06,
      "loss": 0.6774,
      "num_input_tokens_seen": 108699852,
      "step": 1940
    },
    {
      "epoch": 4.320712694877505,
      "loss": 0.8520459532737732,
      "loss_ce": 0.0003613817389123142,
      "loss_iou": 0.337890625,
      "loss_num": 0.035400390625,
      "loss_xval": 0.8515625,
      "num_input_tokens_seen": 108699852,
      "step": 1940
    },
    {
      "epoch": 4.32293986636971,
      "grad_norm": 20.88629150390625,
      "learning_rate": 1e-06,
      "loss": 0.8882,
      "num_input_tokens_seen": 108755364,
      "step": 1941
    },
    {
      "epoch": 4.32293986636971,
      "loss": 0.8283944129943848,
      "loss_ce": 0.000269408046733588,
      "loss_iou": 0.322265625,
      "loss_num": 0.037109375,
      "loss_xval": 0.828125,
      "num_input_tokens_seen": 108755364,
      "step": 1941
    },
    {
      "epoch": 4.325167037861915,
      "grad_norm": 26.91864585876465,
      "learning_rate": 1e-06,
      "loss": 0.573,
      "num_input_tokens_seen": 108814096,
      "step": 1942
    },
    {
      "epoch": 4.325167037861915,
      "loss": 0.6568059921264648,
      "loss_ce": 0.00031184524414129555,
      "loss_iou": 0.298828125,
      "loss_num": 0.0120849609375,
      "loss_xval": 0.65625,
      "num_input_tokens_seen": 108814096,
      "step": 1942
    },
    {
      "epoch": 4.327394209354121,
      "grad_norm": 14.774848937988281,
      "learning_rate": 1e-06,
      "loss": 0.9021,
      "num_input_tokens_seen": 108870164,
      "step": 1943
    },
    {
      "epoch": 4.327394209354121,
      "loss": 0.8290045261383057,
      "loss_ce": 0.001856090733781457,
      "loss_iou": 0.33984375,
      "loss_num": 0.0294189453125,
      "loss_xval": 0.828125,
      "num_input_tokens_seen": 108870164,
      "step": 1943
    },
    {
      "epoch": 4.3296213808463255,
      "grad_norm": 18.11646270751953,
      "learning_rate": 1e-06,
      "loss": 0.8127,
      "num_input_tokens_seen": 108927584,
      "step": 1944
    },
    {
      "epoch": 4.3296213808463255,
      "loss": 0.6893248558044434,
      "loss_ce": 0.0002379524812567979,
      "loss_iou": 0.29296875,
      "loss_num": 0.0201416015625,
      "loss_xval": 0.6875,
      "num_input_tokens_seen": 108927584,
      "step": 1944
    },
    {
      "epoch": 4.33184855233853,
      "grad_norm": 14.257638931274414,
      "learning_rate": 1e-06,
      "loss": 0.8333,
      "num_input_tokens_seen": 108983712,
      "step": 1945
    },
    {
      "epoch": 4.33184855233853,
      "loss": 0.9684985876083374,
      "loss_ce": 0.0003589991247281432,
      "loss_iou": 0.400390625,
      "loss_num": 0.03369140625,
      "loss_xval": 0.96875,
      "num_input_tokens_seen": 108983712,
      "step": 1945
    },
    {
      "epoch": 4.334075723830735,
      "grad_norm": 24.499502182006836,
      "learning_rate": 1e-06,
      "loss": 0.5697,
      "num_input_tokens_seen": 109040176,
      "step": 1946
    },
    {
      "epoch": 4.334075723830735,
      "loss": 0.4902326464653015,
      "loss_ce": 0.00024243266670964658,
      "loss_iou": 0.2158203125,
      "loss_num": 0.01153564453125,
      "loss_xval": 0.490234375,
      "num_input_tokens_seen": 109040176,
      "step": 1946
    },
    {
      "epoch": 4.33630289532294,
      "grad_norm": 28.52815055847168,
      "learning_rate": 1e-06,
      "loss": 0.8584,
      "num_input_tokens_seen": 109096776,
      "step": 1947
    },
    {
      "epoch": 4.33630289532294,
      "loss": 0.8685885667800903,
      "loss_ce": 0.0003024664765689522,
      "loss_iou": 0.345703125,
      "loss_num": 0.035400390625,
      "loss_xval": 0.8671875,
      "num_input_tokens_seen": 109096776,
      "step": 1947
    },
    {
      "epoch": 4.338530066815145,
      "grad_norm": 24.575611114501953,
      "learning_rate": 1e-06,
      "loss": 1.0167,
      "num_input_tokens_seen": 109153828,
      "step": 1948
    },
    {
      "epoch": 4.338530066815145,
      "loss": 0.993535041809082,
      "loss_ce": 0.0002489146136213094,
      "loss_iou": 0.43359375,
      "loss_num": 0.025634765625,
      "loss_xval": 0.9921875,
      "num_input_tokens_seen": 109153828,
      "step": 1948
    },
    {
      "epoch": 4.34075723830735,
      "grad_norm": 15.715718269348145,
      "learning_rate": 1e-06,
      "loss": 0.8722,
      "num_input_tokens_seen": 109212620,
      "step": 1949
    },
    {
      "epoch": 4.34075723830735,
      "loss": 0.8171886205673218,
      "loss_ce": 0.00029411769355647266,
      "loss_iou": 0.34765625,
      "loss_num": 0.0240478515625,
      "loss_xval": 0.81640625,
      "num_input_tokens_seen": 109212620,
      "step": 1949
    },
    {
      "epoch": 4.342984409799555,
      "grad_norm": 23.913692474365234,
      "learning_rate": 1e-06,
      "loss": 0.6369,
      "num_input_tokens_seen": 109270456,
      "step": 1950
    },
    {
      "epoch": 4.342984409799555,
      "loss": 0.641880989074707,
      "loss_ce": 0.0002794649626594037,
      "loss_iou": 0.28125,
      "loss_num": 0.015625,
      "loss_xval": 0.640625,
      "num_input_tokens_seen": 109270456,
      "step": 1950
    },
    {
      "epoch": 4.3452115812917596,
      "grad_norm": 18.232446670532227,
      "learning_rate": 1e-06,
      "loss": 0.768,
      "num_input_tokens_seen": 109325660,
      "step": 1951
    },
    {
      "epoch": 4.3452115812917596,
      "loss": 0.9707180857658386,
      "loss_ce": 0.0002590723452158272,
      "loss_iou": 0.416015625,
      "loss_num": 0.0277099609375,
      "loss_xval": 0.96875,
      "num_input_tokens_seen": 109325660,
      "step": 1951
    },
    {
      "epoch": 4.347438752783964,
      "grad_norm": 16.996084213256836,
      "learning_rate": 1e-06,
      "loss": 0.9126,
      "num_input_tokens_seen": 109381844,
      "step": 1952
    },
    {
      "epoch": 4.347438752783964,
      "loss": 0.7782900333404541,
      "loss_ce": 0.00027485546888783574,
      "loss_iou": 0.3515625,
      "loss_num": 0.01483154296875,
      "loss_xval": 0.77734375,
      "num_input_tokens_seen": 109381844,
      "step": 1952
    },
    {
      "epoch": 4.349665924276169,
      "grad_norm": 17.12710189819336,
      "learning_rate": 1e-06,
      "loss": 0.826,
      "num_input_tokens_seen": 109440512,
      "step": 1953
    },
    {
      "epoch": 4.349665924276169,
      "loss": 1.0132062435150146,
      "loss_ce": 0.00026673171669244766,
      "loss_iou": 0.427734375,
      "loss_num": 0.03173828125,
      "loss_xval": 1.015625,
      "num_input_tokens_seen": 109440512,
      "step": 1953
    },
    {
      "epoch": 4.351893095768374,
      "grad_norm": 14.912611961364746,
      "learning_rate": 1e-06,
      "loss": 0.8326,
      "num_input_tokens_seen": 109497420,
      "step": 1954
    },
    {
      "epoch": 4.351893095768374,
      "loss": 0.42840301990509033,
      "loss_ce": 0.00030242273351177573,
      "loss_iou": 0.17578125,
      "loss_num": 0.01519775390625,
      "loss_xval": 0.427734375,
      "num_input_tokens_seen": 109497420,
      "step": 1954
    },
    {
      "epoch": 4.354120267260579,
      "grad_norm": 17.690128326416016,
      "learning_rate": 1e-06,
      "loss": 1.0669,
      "num_input_tokens_seen": 109554340,
      "step": 1955
    },
    {
      "epoch": 4.354120267260579,
      "loss": 1.157071590423584,
      "loss_ce": 0.00033330710721202195,
      "loss_iou": 0.466796875,
      "loss_num": 0.044677734375,
      "loss_xval": 1.15625,
      "num_input_tokens_seen": 109554340,
      "step": 1955
    },
    {
      "epoch": 4.356347438752784,
      "grad_norm": 279.8905029296875,
      "learning_rate": 1e-06,
      "loss": 0.7712,
      "num_input_tokens_seen": 109611648,
      "step": 1956
    },
    {
      "epoch": 4.356347438752784,
      "loss": 0.8818769454956055,
      "loss_ce": 0.0010175781790167093,
      "loss_iou": 0.384765625,
      "loss_num": 0.0224609375,
      "loss_xval": 0.8828125,
      "num_input_tokens_seen": 109611648,
      "step": 1956
    },
    {
      "epoch": 4.358574610244989,
      "grad_norm": 21.142581939697266,
      "learning_rate": 1e-06,
      "loss": 0.8436,
      "num_input_tokens_seen": 109670164,
      "step": 1957
    },
    {
      "epoch": 4.358574610244989,
      "loss": 0.8566855192184448,
      "loss_ce": 0.00024018189287744462,
      "loss_iou": 0.353515625,
      "loss_num": 0.02978515625,
      "loss_xval": 0.85546875,
      "num_input_tokens_seen": 109670164,
      "step": 1957
    },
    {
      "epoch": 4.360801781737194,
      "grad_norm": 23.72075080871582,
      "learning_rate": 1e-06,
      "loss": 0.5878,
      "num_input_tokens_seen": 109727780,
      "step": 1958
    },
    {
      "epoch": 4.360801781737194,
      "loss": 0.7790469527244568,
      "loss_ce": 0.00023835319734644145,
      "loss_iou": 0.318359375,
      "loss_num": 0.0289306640625,
      "loss_xval": 0.77734375,
      "num_input_tokens_seen": 109727780,
      "step": 1958
    },
    {
      "epoch": 4.363028953229398,
      "grad_norm": 22.97592544555664,
      "learning_rate": 1e-06,
      "loss": 0.9028,
      "num_input_tokens_seen": 109786720,
      "step": 1959
    },
    {
      "epoch": 4.363028953229398,
      "loss": 1.0544261932373047,
      "loss_ce": 0.00022698812244925648,
      "loss_iou": 0.4453125,
      "loss_num": 0.032470703125,
      "loss_xval": 1.0546875,
      "num_input_tokens_seen": 109786720,
      "step": 1959
    },
    {
      "epoch": 4.365256124721603,
      "grad_norm": 35.19451904296875,
      "learning_rate": 1e-06,
      "loss": 1.0087,
      "num_input_tokens_seen": 109840344,
      "step": 1960
    },
    {
      "epoch": 4.365256124721603,
      "loss": 0.9346234202384949,
      "loss_ce": 0.00029728777008131146,
      "loss_iou": 0.412109375,
      "loss_num": 0.0218505859375,
      "loss_xval": 0.93359375,
      "num_input_tokens_seen": 109840344,
      "step": 1960
    },
    {
      "epoch": 4.367483296213808,
      "grad_norm": 21.005878448486328,
      "learning_rate": 1e-06,
      "loss": 0.8335,
      "num_input_tokens_seen": 109898760,
      "step": 1961
    },
    {
      "epoch": 4.367483296213808,
      "loss": 0.7780290842056274,
      "loss_ce": 0.0004412271664477885,
      "loss_iou": 0.306640625,
      "loss_num": 0.033203125,
      "loss_xval": 0.77734375,
      "num_input_tokens_seen": 109898760,
      "step": 1961
    },
    {
      "epoch": 4.369710467706013,
      "grad_norm": 20.942707061767578,
      "learning_rate": 1e-06,
      "loss": 0.7858,
      "num_input_tokens_seen": 109956856,
      "step": 1962
    },
    {
      "epoch": 4.369710467706013,
      "loss": 0.6604050397872925,
      "loss_ce": 0.0002488172031007707,
      "loss_iou": 0.296875,
      "loss_num": 0.0133056640625,
      "loss_xval": 0.66015625,
      "num_input_tokens_seen": 109956856,
      "step": 1962
    },
    {
      "epoch": 4.371937639198218,
      "grad_norm": 14.403862953186035,
      "learning_rate": 1e-06,
      "loss": 0.5435,
      "num_input_tokens_seen": 110010668,
      "step": 1963
    },
    {
      "epoch": 4.371937639198218,
      "loss": 0.32872194051742554,
      "loss_ce": 0.00023072000476531684,
      "loss_iou": 0.1455078125,
      "loss_num": 0.007476806640625,
      "loss_xval": 0.328125,
      "num_input_tokens_seen": 110010668,
      "step": 1963
    },
    {
      "epoch": 4.374164810690424,
      "grad_norm": 27.356098175048828,
      "learning_rate": 1e-06,
      "loss": 0.5899,
      "num_input_tokens_seen": 110067848,
      "step": 1964
    },
    {
      "epoch": 4.374164810690424,
      "loss": 0.6242412328720093,
      "loss_ce": 0.00021780356473755091,
      "loss_iou": 0.25390625,
      "loss_num": 0.0233154296875,
      "loss_xval": 0.625,
      "num_input_tokens_seen": 110067848,
      "step": 1964
    },
    {
      "epoch": 4.3763919821826285,
      "grad_norm": 24.027332305908203,
      "learning_rate": 1e-06,
      "loss": 0.6839,
      "num_input_tokens_seen": 110122296,
      "step": 1965
    },
    {
      "epoch": 4.3763919821826285,
      "loss": 0.6491550207138062,
      "loss_ce": 0.00022922824427951127,
      "loss_iou": 0.28125,
      "loss_num": 0.0169677734375,
      "loss_xval": 0.6484375,
      "num_input_tokens_seen": 110122296,
      "step": 1965
    },
    {
      "epoch": 4.378619153674833,
      "grad_norm": 18.06757354736328,
      "learning_rate": 1e-06,
      "loss": 0.8689,
      "num_input_tokens_seen": 110177296,
      "step": 1966
    },
    {
      "epoch": 4.378619153674833,
      "loss": 0.7419674396514893,
      "loss_ce": 0.0002682540216483176,
      "loss_iou": 0.31640625,
      "loss_num": 0.021484375,
      "loss_xval": 0.7421875,
      "num_input_tokens_seen": 110177296,
      "step": 1966
    },
    {
      "epoch": 4.380846325167038,
      "grad_norm": 22.51265525817871,
      "learning_rate": 1e-06,
      "loss": 0.8991,
      "num_input_tokens_seen": 110233976,
      "step": 1967
    },
    {
      "epoch": 4.380846325167038,
      "loss": 0.7884268760681152,
      "loss_ce": 0.000340936123393476,
      "loss_iou": 0.318359375,
      "loss_num": 0.030029296875,
      "loss_xval": 0.7890625,
      "num_input_tokens_seen": 110233976,
      "step": 1967
    },
    {
      "epoch": 4.383073496659243,
      "grad_norm": 34.43399429321289,
      "learning_rate": 1e-06,
      "loss": 1.0184,
      "num_input_tokens_seen": 110289432,
      "step": 1968
    },
    {
      "epoch": 4.383073496659243,
      "loss": 1.0707261562347412,
      "loss_ce": 0.00041369907557964325,
      "loss_iou": 0.4609375,
      "loss_num": 0.0296630859375,
      "loss_xval": 1.0703125,
      "num_input_tokens_seen": 110289432,
      "step": 1968
    },
    {
      "epoch": 4.385300668151448,
      "grad_norm": 21.057336807250977,
      "learning_rate": 1e-06,
      "loss": 0.8572,
      "num_input_tokens_seen": 110345920,
      "step": 1969
    },
    {
      "epoch": 4.385300668151448,
      "loss": 0.8917399644851685,
      "loss_ce": 0.0003824798041023314,
      "loss_iou": 0.369140625,
      "loss_num": 0.0303955078125,
      "loss_xval": 0.890625,
      "num_input_tokens_seen": 110345920,
      "step": 1969
    },
    {
      "epoch": 4.387527839643653,
      "grad_norm": 17.0823917388916,
      "learning_rate": 1e-06,
      "loss": 0.8178,
      "num_input_tokens_seen": 110398524,
      "step": 1970
    },
    {
      "epoch": 4.387527839643653,
      "loss": 0.9053047895431519,
      "loss_ce": 0.0002754859742708504,
      "loss_iou": 0.349609375,
      "loss_num": 0.041015625,
      "loss_xval": 0.90625,
      "num_input_tokens_seen": 110398524,
      "step": 1970
    },
    {
      "epoch": 4.389755011135858,
      "grad_norm": 19.87203598022461,
      "learning_rate": 1e-06,
      "loss": 0.89,
      "num_input_tokens_seen": 110454132,
      "step": 1971
    },
    {
      "epoch": 4.389755011135858,
      "loss": 0.851262092590332,
      "loss_ce": 0.00030996621353551745,
      "loss_iou": 0.3515625,
      "loss_num": 0.02978515625,
      "loss_xval": 0.8515625,
      "num_input_tokens_seen": 110454132,
      "step": 1971
    },
    {
      "epoch": 4.3919821826280625,
      "grad_norm": 25.85304069519043,
      "learning_rate": 1e-06,
      "loss": 0.668,
      "num_input_tokens_seen": 110510644,
      "step": 1972
    },
    {
      "epoch": 4.3919821826280625,
      "loss": 0.6331183314323425,
      "loss_ce": 0.0003058624570257962,
      "loss_iou": 0.265625,
      "loss_num": 0.0201416015625,
      "loss_xval": 0.6328125,
      "num_input_tokens_seen": 110510644,
      "step": 1972
    },
    {
      "epoch": 4.394209354120267,
      "grad_norm": 20.608179092407227,
      "learning_rate": 1e-06,
      "loss": 0.4726,
      "num_input_tokens_seen": 110568336,
      "step": 1973
    },
    {
      "epoch": 4.394209354120267,
      "loss": 0.5503981113433838,
      "loss_ce": 0.00022721837740391493,
      "loss_iou": 0.25,
      "loss_num": 0.00994873046875,
      "loss_xval": 0.55078125,
      "num_input_tokens_seen": 110568336,
      "step": 1973
    },
    {
      "epoch": 4.396436525612472,
      "grad_norm": 27.215190887451172,
      "learning_rate": 1e-06,
      "loss": 0.7294,
      "num_input_tokens_seen": 110624496,
      "step": 1974
    },
    {
      "epoch": 4.396436525612472,
      "loss": 0.8059403896331787,
      "loss_ce": 0.00027631851844489574,
      "loss_iou": 0.341796875,
      "loss_num": 0.02392578125,
      "loss_xval": 0.8046875,
      "num_input_tokens_seen": 110624496,
      "step": 1974
    },
    {
      "epoch": 4.398663697104677,
      "grad_norm": 17.576311111450195,
      "learning_rate": 1e-06,
      "loss": 0.5477,
      "num_input_tokens_seen": 110679224,
      "step": 1975
    },
    {
      "epoch": 4.398663697104677,
      "loss": 0.3737618923187256,
      "loss_ce": 0.0002267223026137799,
      "loss_iou": 0.1455078125,
      "loss_num": 0.0167236328125,
      "loss_xval": 0.373046875,
      "num_input_tokens_seen": 110679224,
      "step": 1975
    },
    {
      "epoch": 4.400890868596882,
      "grad_norm": 22.326269149780273,
      "learning_rate": 1e-06,
      "loss": 0.7698,
      "num_input_tokens_seen": 110735720,
      "step": 1976
    },
    {
      "epoch": 4.400890868596882,
      "loss": 0.7260935306549072,
      "loss_ce": 0.0002634518896229565,
      "loss_iou": 0.33203125,
      "loss_num": 0.0120849609375,
      "loss_xval": 0.7265625,
      "num_input_tokens_seen": 110735720,
      "step": 1976
    },
    {
      "epoch": 4.403118040089087,
      "grad_norm": 31.376184463500977,
      "learning_rate": 1e-06,
      "loss": 0.7239,
      "num_input_tokens_seen": 110790824,
      "step": 1977
    },
    {
      "epoch": 4.403118040089087,
      "loss": 0.8379002213478088,
      "loss_ce": 0.0002537188120186329,
      "loss_iou": 0.359375,
      "loss_num": 0.024169921875,
      "loss_xval": 0.8359375,
      "num_input_tokens_seen": 110790824,
      "step": 1977
    },
    {
      "epoch": 4.405345211581292,
      "grad_norm": 22.093713760375977,
      "learning_rate": 1e-06,
      "loss": 1.2021,
      "num_input_tokens_seen": 110845432,
      "step": 1978
    },
    {
      "epoch": 4.405345211581292,
      "loss": 1.1360301971435547,
      "loss_ce": 0.00028802291490137577,
      "loss_iou": 0.482421875,
      "loss_num": 0.033935546875,
      "loss_xval": 1.1328125,
      "num_input_tokens_seen": 110845432,
      "step": 1978
    },
    {
      "epoch": 4.4075723830734965,
      "grad_norm": 21.3291072845459,
      "learning_rate": 1e-06,
      "loss": 0.7553,
      "num_input_tokens_seen": 110901496,
      "step": 1979
    },
    {
      "epoch": 4.4075723830734965,
      "loss": 0.6370877623558044,
      "loss_ce": 0.0003689899167511612,
      "loss_iou": 0.279296875,
      "loss_num": 0.015625,
      "loss_xval": 0.63671875,
      "num_input_tokens_seen": 110901496,
      "step": 1979
    },
    {
      "epoch": 4.409799554565701,
      "grad_norm": 26.429105758666992,
      "learning_rate": 1e-06,
      "loss": 0.8782,
      "num_input_tokens_seen": 110960112,
      "step": 1980
    },
    {
      "epoch": 4.409799554565701,
      "loss": 0.8666300773620605,
      "loss_ce": 0.00029704906046390533,
      "loss_iou": 0.365234375,
      "loss_num": 0.02734375,
      "loss_xval": 0.8671875,
      "num_input_tokens_seen": 110960112,
      "step": 1980
    },
    {
      "epoch": 4.412026726057906,
      "grad_norm": 22.83293342590332,
      "learning_rate": 1e-06,
      "loss": 0.6349,
      "num_input_tokens_seen": 111016200,
      "step": 1981
    },
    {
      "epoch": 4.412026726057906,
      "loss": 0.6940905451774597,
      "loss_ce": 0.0002428823063382879,
      "loss_iou": 0.30078125,
      "loss_num": 0.0184326171875,
      "loss_xval": 0.6953125,
      "num_input_tokens_seen": 111016200,
      "step": 1981
    },
    {
      "epoch": 4.414253897550111,
      "grad_norm": 31.386951446533203,
      "learning_rate": 1e-06,
      "loss": 0.9319,
      "num_input_tokens_seen": 111070980,
      "step": 1982
    },
    {
      "epoch": 4.414253897550111,
      "loss": 0.7642084956169128,
      "loss_ce": 0.00029245836776681244,
      "loss_iou": 0.330078125,
      "loss_num": 0.0206298828125,
      "loss_xval": 0.765625,
      "num_input_tokens_seen": 111070980,
      "step": 1982
    },
    {
      "epoch": 4.416481069042316,
      "grad_norm": 14.445569038391113,
      "learning_rate": 1e-06,
      "loss": 0.7298,
      "num_input_tokens_seen": 111127432,
      "step": 1983
    },
    {
      "epoch": 4.416481069042316,
      "loss": 0.7684084177017212,
      "loss_ce": 0.00028093665605410933,
      "loss_iou": 0.31640625,
      "loss_num": 0.02734375,
      "loss_xval": 0.76953125,
      "num_input_tokens_seen": 111127432,
      "step": 1983
    },
    {
      "epoch": 4.418708240534521,
      "grad_norm": 27.805484771728516,
      "learning_rate": 1e-06,
      "loss": 0.8484,
      "num_input_tokens_seen": 111182920,
      "step": 1984
    },
    {
      "epoch": 4.418708240534521,
      "loss": 1.1093249320983887,
      "loss_ce": 0.0006822688737884164,
      "loss_iou": 0.498046875,
      "loss_num": 0.02294921875,
      "loss_xval": 1.109375,
      "num_input_tokens_seen": 111182920,
      "step": 1984
    },
    {
      "epoch": 4.420935412026726,
      "grad_norm": 18.803524017333984,
      "learning_rate": 1e-06,
      "loss": 0.8397,
      "num_input_tokens_seen": 111236748,
      "step": 1985
    },
    {
      "epoch": 4.420935412026726,
      "loss": 0.7825398445129395,
      "loss_ce": 0.0003133030259050429,
      "loss_iou": 0.3125,
      "loss_num": 0.03125,
      "loss_xval": 0.78125,
      "num_input_tokens_seen": 111236748,
      "step": 1985
    },
    {
      "epoch": 4.4231625835189305,
      "grad_norm": 15.333048820495605,
      "learning_rate": 1e-06,
      "loss": 0.7332,
      "num_input_tokens_seen": 111292696,
      "step": 1986
    },
    {
      "epoch": 4.4231625835189305,
      "loss": 0.8349952697753906,
      "loss_ce": 0.00027839711401611567,
      "loss_iou": 0.3359375,
      "loss_num": 0.03271484375,
      "loss_xval": 0.8359375,
      "num_input_tokens_seen": 111292696,
      "step": 1986
    },
    {
      "epoch": 4.425389755011135,
      "grad_norm": 19.713153839111328,
      "learning_rate": 1e-06,
      "loss": 0.8684,
      "num_input_tokens_seen": 111348792,
      "step": 1987
    },
    {
      "epoch": 4.425389755011135,
      "loss": 0.7836449146270752,
      "loss_ce": 0.0009300732635892928,
      "loss_iou": 0.310546875,
      "loss_num": 0.0322265625,
      "loss_xval": 0.78125,
      "num_input_tokens_seen": 111348792,
      "step": 1987
    },
    {
      "epoch": 4.427616926503341,
      "grad_norm": 16.41890525817871,
      "learning_rate": 1e-06,
      "loss": 0.9639,
      "num_input_tokens_seen": 111405636,
      "step": 1988
    },
    {
      "epoch": 4.427616926503341,
      "loss": 0.8102682828903198,
      "loss_ce": 0.0003318004310131073,
      "loss_iou": 0.345703125,
      "loss_num": 0.023193359375,
      "loss_xval": 0.80859375,
      "num_input_tokens_seen": 111405636,
      "step": 1988
    },
    {
      "epoch": 4.429844097995546,
      "grad_norm": 17.086889266967773,
      "learning_rate": 1e-06,
      "loss": 0.9369,
      "num_input_tokens_seen": 111463272,
      "step": 1989
    },
    {
      "epoch": 4.429844097995546,
      "loss": 0.8751001358032227,
      "loss_ce": 0.0003442912711761892,
      "loss_iou": 0.36328125,
      "loss_num": 0.0299072265625,
      "loss_xval": 0.875,
      "num_input_tokens_seen": 111463272,
      "step": 1989
    },
    {
      "epoch": 4.432071269487751,
      "grad_norm": 25.840059280395508,
      "learning_rate": 1e-06,
      "loss": 0.8088,
      "num_input_tokens_seen": 111518036,
      "step": 1990
    },
    {
      "epoch": 4.432071269487751,
      "loss": 0.8438689708709717,
      "loss_ce": 0.0003630804130807519,
      "loss_iou": 0.37109375,
      "loss_num": 0.020751953125,
      "loss_xval": 0.84375,
      "num_input_tokens_seen": 111518036,
      "step": 1990
    },
    {
      "epoch": 4.434298440979956,
      "grad_norm": 59.44672775268555,
      "learning_rate": 1e-06,
      "loss": 0.739,
      "num_input_tokens_seen": 111575688,
      "step": 1991
    },
    {
      "epoch": 4.434298440979956,
      "loss": 0.8342432975769043,
      "loss_ce": 0.00025894519058056176,
      "loss_iou": 0.3671875,
      "loss_num": 0.019775390625,
      "loss_xval": 0.8359375,
      "num_input_tokens_seen": 111575688,
      "step": 1991
    },
    {
      "epoch": 4.436525612472161,
      "grad_norm": 18.12961769104004,
      "learning_rate": 1e-06,
      "loss": 0.7432,
      "num_input_tokens_seen": 111628236,
      "step": 1992
    },
    {
      "epoch": 4.436525612472161,
      "loss": 0.42511090636253357,
      "loss_ce": 0.00030621426412835717,
      "loss_iou": 0.1845703125,
      "loss_num": 0.01092529296875,
      "loss_xval": 0.42578125,
      "num_input_tokens_seen": 111628236,
      "step": 1992
    },
    {
      "epoch": 4.4387527839643655,
      "grad_norm": 34.39860916137695,
      "learning_rate": 1e-06,
      "loss": 0.6518,
      "num_input_tokens_seen": 111685388,
      "step": 1993
    },
    {
      "epoch": 4.4387527839643655,
      "loss": 0.8032548427581787,
      "loss_ce": 0.0002763355150818825,
      "loss_iou": 0.326171875,
      "loss_num": 0.0299072265625,
      "loss_xval": 0.8046875,
      "num_input_tokens_seen": 111685388,
      "step": 1993
    },
    {
      "epoch": 4.44097995545657,
      "grad_norm": 18.156936645507812,
      "learning_rate": 1e-06,
      "loss": 0.6957,
      "num_input_tokens_seen": 111740404,
      "step": 1994
    },
    {
      "epoch": 4.44097995545657,
      "loss": 0.7895779013633728,
      "loss_ce": 0.0002712547720875591,
      "loss_iou": 0.318359375,
      "loss_num": 0.0303955078125,
      "loss_xval": 0.7890625,
      "num_input_tokens_seen": 111740404,
      "step": 1994
    },
    {
      "epoch": 4.443207126948775,
      "grad_norm": 16.068462371826172,
      "learning_rate": 1e-06,
      "loss": 0.8716,
      "num_input_tokens_seen": 111795960,
      "step": 1995
    },
    {
      "epoch": 4.443207126948775,
      "loss": 1.020226001739502,
      "loss_ce": 0.0004505877732299268,
      "loss_iou": 0.40625,
      "loss_num": 0.041259765625,
      "loss_xval": 1.0234375,
      "num_input_tokens_seen": 111795960,
      "step": 1995
    },
    {
      "epoch": 4.44543429844098,
      "grad_norm": 17.97376251220703,
      "learning_rate": 1e-06,
      "loss": 0.5944,
      "num_input_tokens_seen": 111854164,
      "step": 1996
    },
    {
      "epoch": 4.44543429844098,
      "loss": 0.6652653813362122,
      "loss_ce": 0.00034838789724744856,
      "loss_iou": 0.279296875,
      "loss_num": 0.021484375,
      "loss_xval": 0.6640625,
      "num_input_tokens_seen": 111854164,
      "step": 1996
    },
    {
      "epoch": 4.447661469933185,
      "grad_norm": 17.494909286499023,
      "learning_rate": 1e-06,
      "loss": 0.7858,
      "num_input_tokens_seen": 111908960,
      "step": 1997
    },
    {
      "epoch": 4.447661469933185,
      "loss": 0.6890178918838501,
      "loss_ce": 0.0002971593930851668,
      "loss_iou": 0.30078125,
      "loss_num": 0.0172119140625,
      "loss_xval": 0.6875,
      "num_input_tokens_seen": 111908960,
      "step": 1997
    },
    {
      "epoch": 4.44988864142539,
      "grad_norm": 41.67517852783203,
      "learning_rate": 1e-06,
      "loss": 0.8325,
      "num_input_tokens_seen": 111961680,
      "step": 1998
    },
    {
      "epoch": 4.44988864142539,
      "loss": 0.6864081621170044,
      "loss_ce": 0.0011054326314479113,
      "loss_iou": 0.2890625,
      "loss_num": 0.021728515625,
      "loss_xval": 0.68359375,
      "num_input_tokens_seen": 111961680,
      "step": 1998
    },
    {
      "epoch": 4.452115812917595,
      "grad_norm": 31.639928817749023,
      "learning_rate": 1e-06,
      "loss": 0.6754,
      "num_input_tokens_seen": 112020664,
      "step": 1999
    },
    {
      "epoch": 4.452115812917595,
      "loss": 0.7533044219017029,
      "loss_ce": 0.00025264680152758956,
      "loss_iou": 0.326171875,
      "loss_num": 0.02001953125,
      "loss_xval": 0.75390625,
      "num_input_tokens_seen": 112020664,
      "step": 1999
    },
    {
      "epoch": 4.4543429844097995,
      "grad_norm": 17.85369300842285,
      "learning_rate": 1e-06,
      "loss": 0.6786,
      "num_input_tokens_seen": 112077028,
      "step": 2000
    },
    {
      "epoch": 4.4543429844097995,
      "eval_seeclick_web_CIoU": 0.565800279378891,
      "eval_seeclick_web_GIoU": 0.5634768307209015,
      "eval_seeclick_web_IoU": 0.581417441368103,
      "eval_seeclick_web_MAE_all": 0.017455607652664185,
      "eval_seeclick_web_MAE_h": 0.010133389849215746,
      "eval_seeclick_web_MAE_w": 0.01889999955892563,
      "eval_seeclick_web_MAE_x_boxes": 0.00820656050927937,
      "eval_seeclick_web_MAE_y_boxes": 0.02206642786040902,
      "eval_seeclick_web_inside_bbox": 0.9166666567325592,
      "eval_seeclick_web_loss": 0.9660096764564514,
      "eval_seeclick_web_loss_ce": 0.0003684775874717161,
      "eval_seeclick_web_loss_iou": 0.4342041015625,
      "eval_seeclick_web_loss_num": 0.013675689697265625,
      "eval_seeclick_web_loss_xval": 0.9365234375,
      "eval_seeclick_web_runtime": 30.3044,
      "eval_seeclick_web_samples_per_second": 1.65,
      "eval_seeclick_web_steps_per_second": 0.066,
      "num_input_tokens_seen": 112077028,
      "step": 2000
    },
    {
      "epoch": 4.4543429844097995,
      "eval_icons_CIoU": 0.30845747888088226,
      "eval_icons_GIoU": 0.33300746977329254,
      "eval_icons_IoU": 0.38466861844062805,
      "eval_icons_MAE_all": 0.06535855308175087,
      "eval_icons_MAE_h": 0.03866162151098251,
      "eval_icons_MAE_w": 0.0722741037607193,
      "eval_icons_MAE_x_boxes": 0.06343554332852364,
      "eval_icons_MAE_y_boxes": 0.03758078906685114,
      "eval_icons_inside_bbox": 0.6336805522441864,
      "eval_icons_loss": 1.709810495376587,
      "eval_icons_loss_ce": 0.00041042393422685564,
      "eval_icons_loss_iou": 0.6688232421875,
      "eval_icons_loss_num": 0.06202888488769531,
      "eval_icons_loss_xval": 1.6484375,
      "eval_icons_runtime": 23.8114,
      "eval_icons_samples_per_second": 2.1,
      "eval_icons_steps_per_second": 0.084,
      "num_input_tokens_seen": 112077028,
      "step": 2000
    },
    {
      "epoch": 4.4543429844097995,
      "eval_screenspot_CIoU": 0.3293018539746602,
      "eval_screenspot_GIoU": 0.34788986047108966,
      "eval_screenspot_IoU": 0.4084410071372986,
      "eval_screenspot_MAE_all": 0.06550942113002141,
      "eval_screenspot_MAE_h": 0.036151558781663574,
      "eval_screenspot_MAE_w": 0.07488848641514778,
      "eval_screenspot_MAE_x_boxes": 0.07839654758572578,
      "eval_screenspot_MAE_y_boxes": 0.04824645258486271,
      "eval_screenspot_inside_bbox": 0.6462500095367432,
      "eval_screenspot_loss": 1.6914161443710327,
      "eval_screenspot_loss_ce": 0.00041740476929893094,
      "eval_screenspot_loss_iou": 0.695556640625,
      "eval_screenspot_loss_num": 0.07661692301432292,
      "eval_screenspot_loss_xval": 1.7740885416666667,
      "eval_screenspot_runtime": 39.3803,
      "eval_screenspot_samples_per_second": 2.26,
      "eval_screenspot_steps_per_second": 0.076,
      "num_input_tokens_seen": 112077028,
      "step": 2000
    },
    {
      "epoch": 4.4543429844097995,
      "eval_compot_CIoU": 0.35170771181583405,
      "eval_compot_GIoU": 0.37089845538139343,
      "eval_compot_IoU": 0.4089939594268799,
      "eval_compot_MAE_all": 0.018402607180178165,
      "eval_compot_MAE_h": 0.008481500204652548,
      "eval_compot_MAE_w": 0.02456105500459671,
      "eval_compot_MAE_x_boxes": 0.029113260563462973,
      "eval_compot_MAE_y_boxes": 0.0059411004185676575,
      "eval_compot_inside_bbox": 0.6302083432674408,
      "eval_compot_loss": 1.3837639093399048,
      "eval_compot_loss_ce": 0.0003268021682742983,
      "eval_compot_loss_iou": 0.6363525390625,
      "eval_compot_loss_num": 0.017194747924804688,
      "eval_compot_loss_xval": 1.357666015625,
      "eval_compot_runtime": 23.2732,
      "eval_compot_samples_per_second": 2.148,
      "eval_compot_steps_per_second": 0.086,
      "num_input_tokens_seen": 112077028,
      "step": 2000
    },
    {
      "epoch": 4.4543429844097995,
      "eval_custom_ui_val_CIoU": 0.4485127362940047,
      "eval_custom_ui_val_GIoU": 0.4697931508223216,
      "eval_custom_ui_val_IoU": 0.5003485398160087,
      "eval_custom_ui_val_MAE_all": 0.03264946728530857,
      "eval_custom_ui_val_MAE_h": 0.01890565103126897,
      "eval_custom_ui_val_MAE_w": 0.03845942123896546,
      "eval_custom_ui_val_MAE_x_boxes": 0.03562343731108639,
      "eval_custom_ui_val_MAE_y_boxes": 0.017556848521861766,
      "eval_custom_ui_val_inside_bbox": 0.7040895091162788,
      "eval_custom_ui_val_loss": 1.2602308988571167,
      "eval_custom_ui_val_loss_ce": 0.000424561229819018,
      "eval_custom_ui_val_loss_iou": 0.533935546875,
      "eval_custom_ui_val_loss_num": 0.03228992886013455,
      "eval_custom_ui_val_loss_xval": 1.2295193142361112,
      "eval_custom_ui_val_runtime": 61.5934,
      "eval_custom_ui_val_samples_per_second": 4.302,
      "eval_custom_ui_val_steps_per_second": 0.146,
      "num_input_tokens_seen": 112077028,
      "step": 2000
    },
    {
      "epoch": 4.4543429844097995,
      "loss": 0.9637588858604431,
      "loss_ce": 0.0003799735859502107,
      "loss_iou": 0.41796875,
      "loss_num": 0.025146484375,
      "loss_xval": 0.96484375,
      "num_input_tokens_seen": 112077028,
      "step": 2000
    },
    {
      "epoch": 4.456570155902004,
      "grad_norm": 21.966798782348633,
      "learning_rate": 1e-06,
      "loss": 0.8968,
      "num_input_tokens_seen": 112130504,
      "step": 2001
    },
    {
      "epoch": 4.456570155902004,
      "loss": 0.7630659341812134,
      "loss_ce": 0.0003706031129695475,
      "loss_iou": 0.31640625,
      "loss_num": 0.025634765625,
      "loss_xval": 0.76171875,
      "num_input_tokens_seen": 112130504,
      "step": 2001
    },
    {
      "epoch": 4.458797327394209,
      "grad_norm": 16.523204803466797,
      "learning_rate": 1e-06,
      "loss": 0.7768,
      "num_input_tokens_seen": 112188064,
      "step": 2002
    },
    {
      "epoch": 4.458797327394209,
      "loss": 0.6959242820739746,
      "loss_ce": 0.00024555198615416884,
      "loss_iou": 0.28515625,
      "loss_num": 0.025390625,
      "loss_xval": 0.6953125,
      "num_input_tokens_seen": 112188064,
      "step": 2002
    },
    {
      "epoch": 4.461024498886414,
      "grad_norm": 14.921442985534668,
      "learning_rate": 1e-06,
      "loss": 0.8676,
      "num_input_tokens_seen": 112245504,
      "step": 2003
    },
    {
      "epoch": 4.461024498886414,
      "loss": 0.7521045207977295,
      "loss_ce": 0.00027342155226506293,
      "loss_iou": 0.322265625,
      "loss_num": 0.021728515625,
      "loss_xval": 0.75,
      "num_input_tokens_seen": 112245504,
      "step": 2003
    },
    {
      "epoch": 4.463251670378619,
      "grad_norm": 28.89473533630371,
      "learning_rate": 1e-06,
      "loss": 0.8408,
      "num_input_tokens_seen": 112300972,
      "step": 2004
    },
    {
      "epoch": 4.463251670378619,
      "loss": 0.6384103298187256,
      "loss_ce": 0.00022677732340525836,
      "loss_iou": 0.265625,
      "loss_num": 0.021484375,
      "loss_xval": 0.63671875,
      "num_input_tokens_seen": 112300972,
      "step": 2004
    },
    {
      "epoch": 4.465478841870824,
      "grad_norm": 21.826824188232422,
      "learning_rate": 1e-06,
      "loss": 0.9587,
      "num_input_tokens_seen": 112353748,
      "step": 2005
    },
    {
      "epoch": 4.465478841870824,
      "loss": 0.8835301399230957,
      "loss_ce": 0.00022938736947253346,
      "loss_iou": 0.380859375,
      "loss_num": 0.0244140625,
      "loss_xval": 0.8828125,
      "num_input_tokens_seen": 112353748,
      "step": 2005
    },
    {
      "epoch": 4.467706013363029,
      "grad_norm": 18.621387481689453,
      "learning_rate": 1e-06,
      "loss": 0.857,
      "num_input_tokens_seen": 112408688,
      "step": 2006
    },
    {
      "epoch": 4.467706013363029,
      "loss": 0.7712559700012207,
      "loss_ce": 0.0002599011640995741,
      "loss_iou": 0.35546875,
      "loss_num": 0.0123291015625,
      "loss_xval": 0.76953125,
      "num_input_tokens_seen": 112408688,
      "step": 2006
    },
    {
      "epoch": 4.4699331848552335,
      "grad_norm": 20.207508087158203,
      "learning_rate": 1e-06,
      "loss": 0.7325,
      "num_input_tokens_seen": 112465128,
      "step": 2007
    },
    {
      "epoch": 4.4699331848552335,
      "loss": 0.6955528855323792,
      "loss_ce": 0.00024040245625656098,
      "loss_iou": 0.294921875,
      "loss_num": 0.0211181640625,
      "loss_xval": 0.6953125,
      "num_input_tokens_seen": 112465128,
      "step": 2007
    },
    {
      "epoch": 4.472160356347438,
      "grad_norm": 15.423189163208008,
      "learning_rate": 1e-06,
      "loss": 0.7108,
      "num_input_tokens_seen": 112520084,
      "step": 2008
    },
    {
      "epoch": 4.472160356347438,
      "loss": 0.7231366038322449,
      "loss_ce": 0.0002362322702538222,
      "loss_iou": 0.28125,
      "loss_num": 0.03173828125,
      "loss_xval": 0.72265625,
      "num_input_tokens_seen": 112520084,
      "step": 2008
    },
    {
      "epoch": 4.474387527839644,
      "grad_norm": 23.1674861907959,
      "learning_rate": 1e-06,
      "loss": 0.7111,
      "num_input_tokens_seen": 112577288,
      "step": 2009
    },
    {
      "epoch": 4.474387527839644,
      "loss": 0.6543446779251099,
      "loss_ce": 0.0002919051912613213,
      "loss_iou": 0.279296875,
      "loss_num": 0.0194091796875,
      "loss_xval": 0.65234375,
      "num_input_tokens_seen": 112577288,
      "step": 2009
    },
    {
      "epoch": 4.476614699331849,
      "grad_norm": 21.063098907470703,
      "learning_rate": 1e-06,
      "loss": 0.8211,
      "num_input_tokens_seen": 112632456,
      "step": 2010
    },
    {
      "epoch": 4.476614699331849,
      "loss": 0.9507274627685547,
      "loss_ce": 0.000532110221683979,
      "loss_iou": 0.421875,
      "loss_num": 0.0208740234375,
      "loss_xval": 0.94921875,
      "num_input_tokens_seen": 112632456,
      "step": 2010
    },
    {
      "epoch": 4.478841870824054,
      "grad_norm": 14.694646835327148,
      "learning_rate": 1e-06,
      "loss": 0.7868,
      "num_input_tokens_seen": 112685992,
      "step": 2011
    },
    {
      "epoch": 4.478841870824054,
      "loss": 0.8718289732933044,
      "loss_ce": 0.0002469439641572535,
      "loss_iou": 0.3359375,
      "loss_num": 0.03955078125,
      "loss_xval": 0.87109375,
      "num_input_tokens_seen": 112685992,
      "step": 2011
    },
    {
      "epoch": 4.481069042316259,
      "grad_norm": 15.614250183105469,
      "learning_rate": 1e-06,
      "loss": 0.6904,
      "num_input_tokens_seen": 112736456,
      "step": 2012
    },
    {
      "epoch": 4.481069042316259,
      "loss": 0.7561224699020386,
      "loss_ce": 0.0002630477538332343,
      "loss_iou": 0.30078125,
      "loss_num": 0.031005859375,
      "loss_xval": 0.7578125,
      "num_input_tokens_seen": 112736456,
      "step": 2012
    },
    {
      "epoch": 4.4832962138084635,
      "grad_norm": 19.059675216674805,
      "learning_rate": 1e-06,
      "loss": 0.7422,
      "num_input_tokens_seen": 112791472,
      "step": 2013
    },
    {
      "epoch": 4.4832962138084635,
      "loss": 0.5962854623794556,
      "loss_ce": 0.00021605889196507633,
      "loss_iou": 0.267578125,
      "loss_num": 0.0120849609375,
      "loss_xval": 0.59765625,
      "num_input_tokens_seen": 112791472,
      "step": 2013
    },
    {
      "epoch": 4.485523385300668,
      "grad_norm": 16.322175979614258,
      "learning_rate": 1e-06,
      "loss": 0.6247,
      "num_input_tokens_seen": 112847224,
      "step": 2014
    },
    {
      "epoch": 4.485523385300668,
      "loss": 0.6646714806556702,
      "loss_ce": 0.00024277158081531525,
      "loss_iou": 0.267578125,
      "loss_num": 0.025390625,
      "loss_xval": 0.6640625,
      "num_input_tokens_seen": 112847224,
      "step": 2014
    },
    {
      "epoch": 4.487750556792873,
      "grad_norm": 35.126644134521484,
      "learning_rate": 1e-06,
      "loss": 0.6962,
      "num_input_tokens_seen": 112905124,
      "step": 2015
    },
    {
      "epoch": 4.487750556792873,
      "loss": 0.7029978632926941,
      "loss_ce": 0.00023909028095658869,
      "loss_iou": 0.291015625,
      "loss_num": 0.0242919921875,
      "loss_xval": 0.703125,
      "num_input_tokens_seen": 112905124,
      "step": 2015
    },
    {
      "epoch": 4.489977728285078,
      "grad_norm": 17.807477951049805,
      "learning_rate": 1e-06,
      "loss": 0.7735,
      "num_input_tokens_seen": 112963088,
      "step": 2016
    },
    {
      "epoch": 4.489977728285078,
      "loss": 0.7800430655479431,
      "loss_ce": 0.0002578938438091427,
      "loss_iou": 0.3515625,
      "loss_num": 0.01495361328125,
      "loss_xval": 0.78125,
      "num_input_tokens_seen": 112963088,
      "step": 2016
    },
    {
      "epoch": 4.492204899777283,
      "grad_norm": 25.92805290222168,
      "learning_rate": 1e-06,
      "loss": 0.8668,
      "num_input_tokens_seen": 113012908,
      "step": 2017
    },
    {
      "epoch": 4.492204899777283,
      "loss": 0.7654862403869629,
      "loss_ce": 0.0003495719865895808,
      "loss_iou": 0.3359375,
      "loss_num": 0.0181884765625,
      "loss_xval": 0.765625,
      "num_input_tokens_seen": 113012908,
      "step": 2017
    },
    {
      "epoch": 4.494432071269488,
      "grad_norm": 26.185386657714844,
      "learning_rate": 1e-06,
      "loss": 0.666,
      "num_input_tokens_seen": 113069828,
      "step": 2018
    },
    {
      "epoch": 4.494432071269488,
      "loss": 0.5187711715698242,
      "loss_ce": 0.0003385008021723479,
      "loss_iou": 0.2294921875,
      "loss_num": 0.0118408203125,
      "loss_xval": 0.51953125,
      "num_input_tokens_seen": 113069828,
      "step": 2018
    },
    {
      "epoch": 4.496659242761693,
      "grad_norm": 16.547481536865234,
      "learning_rate": 1e-06,
      "loss": 0.7828,
      "num_input_tokens_seen": 113126292,
      "step": 2019
    },
    {
      "epoch": 4.496659242761693,
      "loss": 0.7228785753250122,
      "loss_ce": 0.0002223363844677806,
      "loss_iou": 0.302734375,
      "loss_num": 0.02294921875,
      "loss_xval": 0.72265625,
      "num_input_tokens_seen": 113126292,
      "step": 2019
    },
    {
      "epoch": 4.498886414253898,
      "grad_norm": 16.149658203125,
      "learning_rate": 1e-06,
      "loss": 0.7314,
      "num_input_tokens_seen": 113180940,
      "step": 2020
    },
    {
      "epoch": 4.498886414253898,
      "loss": 0.8957573771476746,
      "loss_ce": 0.0002495555963832885,
      "loss_iou": 0.392578125,
      "loss_num": 0.021728515625,
      "loss_xval": 0.89453125,
      "num_input_tokens_seen": 113180940,
      "step": 2020
    },
    {
      "epoch": 4.501113585746102,
      "grad_norm": 19.080398559570312,
      "learning_rate": 1e-06,
      "loss": 0.6275,
      "num_input_tokens_seen": 113237920,
      "step": 2021
    },
    {
      "epoch": 4.501113585746102,
      "loss": 0.8777415752410889,
      "loss_ce": 0.0004222409042995423,
      "loss_iou": 0.396484375,
      "loss_num": 0.0166015625,
      "loss_xval": 0.87890625,
      "num_input_tokens_seen": 113237920,
      "step": 2021
    },
    {
      "epoch": 4.503340757238307,
      "grad_norm": 18.565778732299805,
      "learning_rate": 1e-06,
      "loss": 0.8399,
      "num_input_tokens_seen": 113292456,
      "step": 2022
    },
    {
      "epoch": 4.503340757238307,
      "loss": 0.8745211362838745,
      "loss_ce": 0.0002535720122978091,
      "loss_iou": 0.373046875,
      "loss_num": 0.025390625,
      "loss_xval": 0.875,
      "num_input_tokens_seen": 113292456,
      "step": 2022
    },
    {
      "epoch": 4.505567928730512,
      "grad_norm": 16.41952133178711,
      "learning_rate": 1e-06,
      "loss": 0.6277,
      "num_input_tokens_seen": 113348476,
      "step": 2023
    },
    {
      "epoch": 4.505567928730512,
      "loss": 0.6538951992988586,
      "loss_ce": 0.0005749051342718303,
      "loss_iou": 0.302734375,
      "loss_num": 0.00946044921875,
      "loss_xval": 0.65234375,
      "num_input_tokens_seen": 113348476,
      "step": 2023
    },
    {
      "epoch": 4.507795100222717,
      "grad_norm": 21.629709243774414,
      "learning_rate": 1e-06,
      "loss": 0.6398,
      "num_input_tokens_seen": 113406428,
      "step": 2024
    },
    {
      "epoch": 4.507795100222717,
      "loss": 0.7275168299674988,
      "loss_ce": 0.00022192415781319141,
      "loss_iou": 0.298828125,
      "loss_num": 0.0260009765625,
      "loss_xval": 0.7265625,
      "num_input_tokens_seen": 113406428,
      "step": 2024
    },
    {
      "epoch": 4.510022271714922,
      "grad_norm": 20.57599449157715,
      "learning_rate": 1e-06,
      "loss": 0.7676,
      "num_input_tokens_seen": 113458468,
      "step": 2025
    },
    {
      "epoch": 4.510022271714922,
      "loss": 0.8076622486114502,
      "loss_ce": 0.0002891695185098797,
      "loss_iou": 0.349609375,
      "loss_num": 0.0218505859375,
      "loss_xval": 0.80859375,
      "num_input_tokens_seen": 113458468,
      "step": 2025
    },
    {
      "epoch": 4.512249443207127,
      "grad_norm": 17.691875457763672,
      "learning_rate": 1e-06,
      "loss": 0.7612,
      "num_input_tokens_seen": 113514212,
      "step": 2026
    },
    {
      "epoch": 4.512249443207127,
      "loss": 0.46555206179618835,
      "loss_ce": 0.00046417216071859,
      "loss_iou": 0.2060546875,
      "loss_num": 0.01043701171875,
      "loss_xval": 0.46484375,
      "num_input_tokens_seen": 113514212,
      "step": 2026
    },
    {
      "epoch": 4.514476614699332,
      "grad_norm": 21.262399673461914,
      "learning_rate": 1e-06,
      "loss": 0.7058,
      "num_input_tokens_seen": 113568052,
      "step": 2027
    },
    {
      "epoch": 4.514476614699332,
      "loss": 0.7981184720993042,
      "loss_ce": 0.00026689909282140434,
      "loss_iou": 0.328125,
      "loss_num": 0.0281982421875,
      "loss_xval": 0.796875,
      "num_input_tokens_seen": 113568052,
      "step": 2027
    },
    {
      "epoch": 4.5167037861915365,
      "grad_norm": 19.464096069335938,
      "learning_rate": 1e-06,
      "loss": 0.7808,
      "num_input_tokens_seen": 113625848,
      "step": 2028
    },
    {
      "epoch": 4.5167037861915365,
      "loss": 0.9021308422088623,
      "loss_ce": 0.0016181376995518804,
      "loss_iou": 0.400390625,
      "loss_num": 0.019775390625,
      "loss_xval": 0.90234375,
      "num_input_tokens_seen": 113625848,
      "step": 2028
    },
    {
      "epoch": 4.518930957683741,
      "grad_norm": 18.429672241210938,
      "learning_rate": 1e-06,
      "loss": 0.723,
      "num_input_tokens_seen": 113680916,
      "step": 2029
    },
    {
      "epoch": 4.518930957683741,
      "loss": 0.9203833341598511,
      "loss_ce": 0.00021731224842369556,
      "loss_iou": 0.390625,
      "loss_num": 0.027587890625,
      "loss_xval": 0.921875,
      "num_input_tokens_seen": 113680916,
      "step": 2029
    },
    {
      "epoch": 4.521158129175946,
      "grad_norm": 15.999838829040527,
      "learning_rate": 1e-06,
      "loss": 0.8875,
      "num_input_tokens_seen": 113736420,
      "step": 2030
    },
    {
      "epoch": 4.521158129175946,
      "loss": 0.799384593963623,
      "loss_ce": 0.0003123145434074104,
      "loss_iou": 0.33203125,
      "loss_num": 0.0267333984375,
      "loss_xval": 0.80078125,
      "num_input_tokens_seen": 113736420,
      "step": 2030
    },
    {
      "epoch": 4.523385300668151,
      "grad_norm": 15.953963279724121,
      "learning_rate": 1e-06,
      "loss": 0.7948,
      "num_input_tokens_seen": 113792556,
      "step": 2031
    },
    {
      "epoch": 4.523385300668151,
      "loss": 0.6049848794937134,
      "loss_ce": 0.0002485427539795637,
      "loss_iou": 0.26171875,
      "loss_num": 0.0162353515625,
      "loss_xval": 0.60546875,
      "num_input_tokens_seen": 113792556,
      "step": 2031
    },
    {
      "epoch": 4.525612472160356,
      "grad_norm": 24.632400512695312,
      "learning_rate": 1e-06,
      "loss": 0.707,
      "num_input_tokens_seen": 113845716,
      "step": 2032
    },
    {
      "epoch": 4.525612472160356,
      "loss": 0.9207833409309387,
      "loss_ce": 0.00025110485148616135,
      "loss_iou": 0.3984375,
      "loss_num": 0.0244140625,
      "loss_xval": 0.921875,
      "num_input_tokens_seen": 113845716,
      "step": 2032
    },
    {
      "epoch": 4.527839643652561,
      "grad_norm": 17.756380081176758,
      "learning_rate": 1e-06,
      "loss": 1.0081,
      "num_input_tokens_seen": 113902444,
      "step": 2033
    },
    {
      "epoch": 4.527839643652561,
      "loss": 1.0257360935211182,
      "loss_ce": 0.00034554791636765003,
      "loss_iou": 0.384765625,
      "loss_num": 0.051025390625,
      "loss_xval": 1.0234375,
      "num_input_tokens_seen": 113902444,
      "step": 2033
    },
    {
      "epoch": 4.5300668151447665,
      "grad_norm": 18.746212005615234,
      "learning_rate": 1e-06,
      "loss": 0.731,
      "num_input_tokens_seen": 113957736,
      "step": 2034
    },
    {
      "epoch": 4.5300668151447665,
      "loss": 0.6118181943893433,
      "loss_ce": 0.00024596002185717225,
      "loss_iou": 0.265625,
      "loss_num": 0.015869140625,
      "loss_xval": 0.61328125,
      "num_input_tokens_seen": 113957736,
      "step": 2034
    },
    {
      "epoch": 4.532293986636971,
      "grad_norm": 18.45545196533203,
      "learning_rate": 1e-06,
      "loss": 0.7164,
      "num_input_tokens_seen": 114014944,
      "step": 2035
    },
    {
      "epoch": 4.532293986636971,
      "loss": 0.799105167388916,
      "loss_ce": 0.00027704003150574863,
      "loss_iou": 0.353515625,
      "loss_num": 0.0185546875,
      "loss_xval": 0.796875,
      "num_input_tokens_seen": 114014944,
      "step": 2035
    },
    {
      "epoch": 4.534521158129176,
      "grad_norm": 16.944326400756836,
      "learning_rate": 1e-06,
      "loss": 0.7828,
      "num_input_tokens_seen": 114073000,
      "step": 2036
    },
    {
      "epoch": 4.534521158129176,
      "loss": 0.7347084283828735,
      "loss_ce": 0.00033347506541758776,
      "loss_iou": 0.3203125,
      "loss_num": 0.018798828125,
      "loss_xval": 0.734375,
      "num_input_tokens_seen": 114073000,
      "step": 2036
    },
    {
      "epoch": 4.536748329621381,
      "grad_norm": 42.41679000854492,
      "learning_rate": 1e-06,
      "loss": 0.6489,
      "num_input_tokens_seen": 114129412,
      "step": 2037
    },
    {
      "epoch": 4.536748329621381,
      "loss": 0.591286838054657,
      "loss_ce": 0.0004665802407544106,
      "loss_iou": 0.2392578125,
      "loss_num": 0.0223388671875,
      "loss_xval": 0.58984375,
      "num_input_tokens_seen": 114129412,
      "step": 2037
    },
    {
      "epoch": 4.538975501113586,
      "grad_norm": 20.462520599365234,
      "learning_rate": 1e-06,
      "loss": 0.8208,
      "num_input_tokens_seen": 114182588,
      "step": 2038
    },
    {
      "epoch": 4.538975501113586,
      "loss": 0.692272424697876,
      "loss_ce": 0.00037790805799886584,
      "loss_iou": 0.28125,
      "loss_num": 0.0257568359375,
      "loss_xval": 0.69140625,
      "num_input_tokens_seen": 114182588,
      "step": 2038
    },
    {
      "epoch": 4.541202672605791,
      "grad_norm": 23.208797454833984,
      "learning_rate": 1e-06,
      "loss": 1.2438,
      "num_input_tokens_seen": 114236432,
      "step": 2039
    },
    {
      "epoch": 4.541202672605791,
      "loss": 1.3942797183990479,
      "loss_ce": 0.000725053483620286,
      "loss_iou": 0.55078125,
      "loss_num": 0.0576171875,
      "loss_xval": 1.390625,
      "num_input_tokens_seen": 114236432,
      "step": 2039
    },
    {
      "epoch": 4.543429844097996,
      "grad_norm": 20.509309768676758,
      "learning_rate": 1e-06,
      "loss": 0.894,
      "num_input_tokens_seen": 114291080,
      "step": 2040
    },
    {
      "epoch": 4.543429844097996,
      "loss": 0.9274067878723145,
      "loss_ce": 0.0002827763673849404,
      "loss_iou": 0.3828125,
      "loss_num": 0.03271484375,
      "loss_xval": 0.92578125,
      "num_input_tokens_seen": 114291080,
      "step": 2040
    },
    {
      "epoch": 4.5456570155902005,
      "grad_norm": 36.92770767211914,
      "learning_rate": 1e-06,
      "loss": 0.7234,
      "num_input_tokens_seen": 114345108,
      "step": 2041
    },
    {
      "epoch": 4.5456570155902005,
      "loss": 0.7667504549026489,
      "loss_ce": 0.00027098384452983737,
      "loss_iou": 0.3046875,
      "loss_num": 0.03125,
      "loss_xval": 0.765625,
      "num_input_tokens_seen": 114345108,
      "step": 2041
    },
    {
      "epoch": 4.547884187082405,
      "grad_norm": 21.100217819213867,
      "learning_rate": 1e-06,
      "loss": 0.7731,
      "num_input_tokens_seen": 114399096,
      "step": 2042
    },
    {
      "epoch": 4.547884187082405,
      "loss": 0.7943999767303467,
      "loss_ce": 0.00045465261791832745,
      "loss_iou": 0.349609375,
      "loss_num": 0.018798828125,
      "loss_xval": 0.79296875,
      "num_input_tokens_seen": 114399096,
      "step": 2042
    },
    {
      "epoch": 4.55011135857461,
      "grad_norm": 20.20854949951172,
      "learning_rate": 1e-06,
      "loss": 0.6397,
      "num_input_tokens_seen": 114454576,
      "step": 2043
    },
    {
      "epoch": 4.55011135857461,
      "loss": 0.620108962059021,
      "loss_ce": 0.00023591173521708697,
      "loss_iou": 0.2431640625,
      "loss_num": 0.026611328125,
      "loss_xval": 0.62109375,
      "num_input_tokens_seen": 114454576,
      "step": 2043
    },
    {
      "epoch": 4.552338530066815,
      "grad_norm": 14.868428230285645,
      "learning_rate": 1e-06,
      "loss": 0.7342,
      "num_input_tokens_seen": 114508528,
      "step": 2044
    },
    {
      "epoch": 4.552338530066815,
      "loss": 0.7006996870040894,
      "loss_ce": 0.0002602189779281616,
      "loss_iou": 0.30078125,
      "loss_num": 0.01953125,
      "loss_xval": 0.69921875,
      "num_input_tokens_seen": 114508528,
      "step": 2044
    },
    {
      "epoch": 4.55456570155902,
      "grad_norm": 14.888712882995605,
      "learning_rate": 1e-06,
      "loss": 0.8115,
      "num_input_tokens_seen": 114565380,
      "step": 2045
    },
    {
      "epoch": 4.55456570155902,
      "loss": 0.7298033237457275,
      "loss_ce": 0.00031118281185626984,
      "loss_iou": 0.3203125,
      "loss_num": 0.017578125,
      "loss_xval": 0.73046875,
      "num_input_tokens_seen": 114565380,
      "step": 2045
    },
    {
      "epoch": 4.556792873051225,
      "grad_norm": 18.843162536621094,
      "learning_rate": 1e-06,
      "loss": 0.8365,
      "num_input_tokens_seen": 114621044,
      "step": 2046
    },
    {
      "epoch": 4.556792873051225,
      "loss": 0.9204500913619995,
      "loss_ce": 0.0002840836241375655,
      "loss_iou": 0.41796875,
      "loss_num": 0.0167236328125,
      "loss_xval": 0.921875,
      "num_input_tokens_seen": 114621044,
      "step": 2046
    },
    {
      "epoch": 4.55902004454343,
      "grad_norm": 22.14583969116211,
      "learning_rate": 1e-06,
      "loss": 0.7857,
      "num_input_tokens_seen": 114678684,
      "step": 2047
    },
    {
      "epoch": 4.55902004454343,
      "loss": 0.9415132403373718,
      "loss_ce": 0.0003511565155349672,
      "loss_iou": 0.37890625,
      "loss_num": 0.036865234375,
      "loss_xval": 0.94140625,
      "num_input_tokens_seen": 114678684,
      "step": 2047
    },
    {
      "epoch": 4.5612472160356345,
      "grad_norm": 17.67383575439453,
      "learning_rate": 1e-06,
      "loss": 0.8186,
      "num_input_tokens_seen": 114736248,
      "step": 2048
    },
    {
      "epoch": 4.5612472160356345,
      "loss": 0.8276271820068359,
      "loss_ce": 0.0004787354846484959,
      "loss_iou": 0.35546875,
      "loss_num": 0.023193359375,
      "loss_xval": 0.828125,
      "num_input_tokens_seen": 114736248,
      "step": 2048
    },
    {
      "epoch": 4.563474387527839,
      "grad_norm": 20.229520797729492,
      "learning_rate": 1e-06,
      "loss": 0.7565,
      "num_input_tokens_seen": 114793940,
      "step": 2049
    },
    {
      "epoch": 4.563474387527839,
      "loss": 0.8892278671264648,
      "loss_ce": 0.00031182251404970884,
      "loss_iou": 0.369140625,
      "loss_num": 0.02978515625,
      "loss_xval": 0.890625,
      "num_input_tokens_seen": 114793940,
      "step": 2049
    },
    {
      "epoch": 4.565701559020044,
      "grad_norm": 18.42917823791504,
      "learning_rate": 1e-06,
      "loss": 0.611,
      "num_input_tokens_seen": 114851312,
      "step": 2050
    },
    {
      "epoch": 4.565701559020044,
      "loss": 0.5913413763046265,
      "loss_ce": 0.0002769285929389298,
      "loss_iou": 0.267578125,
      "loss_num": 0.01153564453125,
      "loss_xval": 0.58984375,
      "num_input_tokens_seen": 114851312,
      "step": 2050
    },
    {
      "epoch": 4.567928730512249,
      "grad_norm": 15.126654624938965,
      "learning_rate": 1e-06,
      "loss": 0.7946,
      "num_input_tokens_seen": 114904084,
      "step": 2051
    },
    {
      "epoch": 4.567928730512249,
      "loss": 0.9705009460449219,
      "loss_ce": 0.0002861037792172283,
      "loss_iou": 0.404296875,
      "loss_num": 0.031982421875,
      "loss_xval": 0.96875,
      "num_input_tokens_seen": 114904084,
      "step": 2051
    },
    {
      "epoch": 4.570155902004454,
      "grad_norm": 101.9856948852539,
      "learning_rate": 1e-06,
      "loss": 0.7333,
      "num_input_tokens_seen": 114958232,
      "step": 2052
    },
    {
      "epoch": 4.570155902004454,
      "loss": 0.759264349937439,
      "loss_ce": 0.00023114567738957703,
      "loss_iou": 0.337890625,
      "loss_num": 0.016845703125,
      "loss_xval": 0.7578125,
      "num_input_tokens_seen": 114958232,
      "step": 2052
    },
    {
      "epoch": 4.57238307349666,
      "grad_norm": 16.39866065979004,
      "learning_rate": 1e-06,
      "loss": 0.6814,
      "num_input_tokens_seen": 115015688,
      "step": 2053
    },
    {
      "epoch": 4.57238307349666,
      "loss": 0.8523510694503784,
      "loss_ce": 0.00030033523216843605,
      "loss_iou": 0.361328125,
      "loss_num": 0.025634765625,
      "loss_xval": 0.8515625,
      "num_input_tokens_seen": 115015688,
      "step": 2053
    },
    {
      "epoch": 4.574610244988865,
      "grad_norm": 24.29990005493164,
      "learning_rate": 1e-06,
      "loss": 0.8799,
      "num_input_tokens_seen": 115072520,
      "step": 2054
    },
    {
      "epoch": 4.574610244988865,
      "loss": 0.7028564214706421,
      "loss_ce": 0.00046386715257540345,
      "loss_iou": 0.30859375,
      "loss_num": 0.0172119140625,
      "loss_xval": 0.703125,
      "num_input_tokens_seen": 115072520,
      "step": 2054
    },
    {
      "epoch": 4.5768374164810695,
      "grad_norm": 42.747581481933594,
      "learning_rate": 1e-06,
      "loss": 0.7572,
      "num_input_tokens_seen": 115128176,
      "step": 2055
    },
    {
      "epoch": 4.5768374164810695,
      "loss": 0.7119477987289429,
      "loss_ce": 0.00027786268037743866,
      "loss_iou": 0.291015625,
      "loss_num": 0.02587890625,
      "loss_xval": 0.7109375,
      "num_input_tokens_seen": 115128176,
      "step": 2055
    },
    {
      "epoch": 4.579064587973274,
      "grad_norm": 28.751867294311523,
      "learning_rate": 1e-06,
      "loss": 0.7886,
      "num_input_tokens_seen": 115185668,
      "step": 2056
    },
    {
      "epoch": 4.579064587973274,
      "loss": 0.7783513069152832,
      "loss_ce": 0.0002751080028247088,
      "loss_iou": 0.330078125,
      "loss_num": 0.0238037109375,
      "loss_xval": 0.77734375,
      "num_input_tokens_seen": 115185668,
      "step": 2056
    },
    {
      "epoch": 4.581291759465479,
      "grad_norm": 20.60097885131836,
      "learning_rate": 1e-06,
      "loss": 0.8628,
      "num_input_tokens_seen": 115242368,
      "step": 2057
    },
    {
      "epoch": 4.581291759465479,
      "loss": 0.954852283000946,
      "loss_ce": 0.00026240124134346843,
      "loss_iou": 0.41796875,
      "loss_num": 0.023681640625,
      "loss_xval": 0.953125,
      "num_input_tokens_seen": 115242368,
      "step": 2057
    },
    {
      "epoch": 4.583518930957684,
      "grad_norm": 31.440357208251953,
      "learning_rate": 1e-06,
      "loss": 1.0689,
      "num_input_tokens_seen": 115297348,
      "step": 2058
    },
    {
      "epoch": 4.583518930957684,
      "loss": 1.095000982284546,
      "loss_ce": 0.000274458434432745,
      "loss_iou": 0.46484375,
      "loss_num": 0.033203125,
      "loss_xval": 1.09375,
      "num_input_tokens_seen": 115297348,
      "step": 2058
    },
    {
      "epoch": 4.585746102449889,
      "grad_norm": 22.026607513427734,
      "learning_rate": 1e-06,
      "loss": 0.9932,
      "num_input_tokens_seen": 115351760,
      "step": 2059
    },
    {
      "epoch": 4.585746102449889,
      "loss": 0.8425930738449097,
      "loss_ce": 0.0003078907902818173,
      "loss_iou": 0.341796875,
      "loss_num": 0.03173828125,
      "loss_xval": 0.84375,
      "num_input_tokens_seen": 115351760,
      "step": 2059
    },
    {
      "epoch": 4.587973273942094,
      "grad_norm": 21.124006271362305,
      "learning_rate": 1e-06,
      "loss": 0.7942,
      "num_input_tokens_seen": 115407436,
      "step": 2060
    },
    {
      "epoch": 4.587973273942094,
      "loss": 0.8991901874542236,
      "loss_ce": 0.0002643706393428147,
      "loss_iou": 0.369140625,
      "loss_num": 0.032470703125,
      "loss_xval": 0.8984375,
      "num_input_tokens_seen": 115407436,
      "step": 2060
    },
    {
      "epoch": 4.590200445434299,
      "grad_norm": 13.804505348205566,
      "learning_rate": 1e-06,
      "loss": 0.8642,
      "num_input_tokens_seen": 115464140,
      "step": 2061
    },
    {
      "epoch": 4.590200445434299,
      "loss": 0.6700685620307922,
      "loss_ce": 0.0002687171217985451,
      "loss_iou": 0.2578125,
      "loss_num": 0.0308837890625,
      "loss_xval": 0.66796875,
      "num_input_tokens_seen": 115464140,
      "step": 2061
    },
    {
      "epoch": 4.5924276169265035,
      "grad_norm": 30.673608779907227,
      "learning_rate": 1e-06,
      "loss": 0.8756,
      "num_input_tokens_seen": 115517956,
      "step": 2062
    },
    {
      "epoch": 4.5924276169265035,
      "loss": 0.6949182748794556,
      "loss_ce": 0.00033817399526014924,
      "loss_iou": 0.287109375,
      "loss_num": 0.024169921875,
      "loss_xval": 0.6953125,
      "num_input_tokens_seen": 115517956,
      "step": 2062
    },
    {
      "epoch": 4.594654788418708,
      "grad_norm": 16.69051170349121,
      "learning_rate": 1e-06,
      "loss": 0.7354,
      "num_input_tokens_seen": 115575676,
      "step": 2063
    },
    {
      "epoch": 4.594654788418708,
      "loss": 0.8584408760070801,
      "loss_ce": 0.00028658530209213495,
      "loss_iou": 0.330078125,
      "loss_num": 0.039306640625,
      "loss_xval": 0.859375,
      "num_input_tokens_seen": 115575676,
      "step": 2063
    },
    {
      "epoch": 4.596881959910913,
      "grad_norm": 16.270042419433594,
      "learning_rate": 1e-06,
      "loss": 0.7017,
      "num_input_tokens_seen": 115631496,
      "step": 2064
    },
    {
      "epoch": 4.596881959910913,
      "loss": 0.5482097268104553,
      "loss_ce": 0.0006023210007697344,
      "loss_iou": 0.212890625,
      "loss_num": 0.024169921875,
      "loss_xval": 0.546875,
      "num_input_tokens_seen": 115631496,
      "step": 2064
    },
    {
      "epoch": 4.599109131403118,
      "grad_norm": 18.328039169311523,
      "learning_rate": 1e-06,
      "loss": 0.9602,
      "num_input_tokens_seen": 115685200,
      "step": 2065
    },
    {
      "epoch": 4.599109131403118,
      "loss": 0.9518949389457703,
      "loss_ce": 0.00023480196250602603,
      "loss_iou": 0.41796875,
      "loss_num": 0.022705078125,
      "loss_xval": 0.953125,
      "num_input_tokens_seen": 115685200,
      "step": 2065
    },
    {
      "epoch": 4.601336302895323,
      "grad_norm": 18.395797729492188,
      "learning_rate": 1e-06,
      "loss": 0.9051,
      "num_input_tokens_seen": 115740640,
      "step": 2066
    },
    {
      "epoch": 4.601336302895323,
      "loss": 0.9198845624923706,
      "loss_ce": 0.00032886205008253455,
      "loss_iou": 0.369140625,
      "loss_num": 0.035888671875,
      "loss_xval": 0.91796875,
      "num_input_tokens_seen": 115740640,
      "step": 2066
    },
    {
      "epoch": 4.603563474387528,
      "grad_norm": 78.51664733886719,
      "learning_rate": 1e-06,
      "loss": 0.7135,
      "num_input_tokens_seen": 115794096,
      "step": 2067
    },
    {
      "epoch": 4.603563474387528,
      "loss": 0.912857174873352,
      "loss_ce": 0.00025952988653443754,
      "loss_iou": 0.392578125,
      "loss_num": 0.0260009765625,
      "loss_xval": 0.9140625,
      "num_input_tokens_seen": 115794096,
      "step": 2067
    },
    {
      "epoch": 4.605790645879733,
      "grad_norm": 13.404295921325684,
      "learning_rate": 1e-06,
      "loss": 0.6822,
      "num_input_tokens_seen": 115852012,
      "step": 2068
    },
    {
      "epoch": 4.605790645879733,
      "loss": 0.6375604867935181,
      "loss_ce": 0.00023134646471589804,
      "loss_iou": 0.271484375,
      "loss_num": 0.018798828125,
      "loss_xval": 0.63671875,
      "num_input_tokens_seen": 115852012,
      "step": 2068
    },
    {
      "epoch": 4.6080178173719375,
      "grad_norm": 35.90913772583008,
      "learning_rate": 1e-06,
      "loss": 0.9463,
      "num_input_tokens_seen": 115910272,
      "step": 2069
    },
    {
      "epoch": 4.6080178173719375,
      "loss": 0.7615780830383301,
      "loss_ce": 0.0003476125421002507,
      "loss_iou": 0.33984375,
      "loss_num": 0.0166015625,
      "loss_xval": 0.76171875,
      "num_input_tokens_seen": 115910272,
      "step": 2069
    },
    {
      "epoch": 4.610244988864142,
      "grad_norm": 48.58989715576172,
      "learning_rate": 1e-06,
      "loss": 0.8322,
      "num_input_tokens_seen": 115966060,
      "step": 2070
    },
    {
      "epoch": 4.610244988864142,
      "loss": 0.9044246077537537,
      "loss_ce": 0.00024984654737636447,
      "loss_iou": 0.3828125,
      "loss_num": 0.027587890625,
      "loss_xval": 0.90234375,
      "num_input_tokens_seen": 115966060,
      "step": 2070
    },
    {
      "epoch": 4.612472160356347,
      "grad_norm": 24.12740135192871,
      "learning_rate": 1e-06,
      "loss": 0.8331,
      "num_input_tokens_seen": 116023692,
      "step": 2071
    },
    {
      "epoch": 4.612472160356347,
      "loss": 0.7075637578964233,
      "loss_ce": 0.0002883510896936059,
      "loss_iou": 0.296875,
      "loss_num": 0.0228271484375,
      "loss_xval": 0.70703125,
      "num_input_tokens_seen": 116023692,
      "step": 2071
    },
    {
      "epoch": 4.614699331848552,
      "grad_norm": 28.676658630371094,
      "learning_rate": 1e-06,
      "loss": 0.8544,
      "num_input_tokens_seen": 116078540,
      "step": 2072
    },
    {
      "epoch": 4.614699331848552,
      "loss": 0.9685863852500916,
      "loss_ce": 0.00032463445677421987,
      "loss_iou": 0.408203125,
      "loss_num": 0.0303955078125,
      "loss_xval": 0.96875,
      "num_input_tokens_seen": 116078540,
      "step": 2072
    },
    {
      "epoch": 4.616926503340757,
      "grad_norm": 19.603900909423828,
      "learning_rate": 1e-06,
      "loss": 0.7623,
      "num_input_tokens_seen": 116133732,
      "step": 2073
    },
    {
      "epoch": 4.616926503340757,
      "loss": 0.7463170886039734,
      "loss_ce": 0.0002233087579952553,
      "loss_iou": 0.27734375,
      "loss_num": 0.0380859375,
      "loss_xval": 0.74609375,
      "num_input_tokens_seen": 116133732,
      "step": 2073
    },
    {
      "epoch": 4.619153674832962,
      "grad_norm": 15.20378589630127,
      "learning_rate": 1e-06,
      "loss": 0.7644,
      "num_input_tokens_seen": 116188860,
      "step": 2074
    },
    {
      "epoch": 4.619153674832962,
      "loss": 0.748953104019165,
      "loss_ce": 0.0004179720126558095,
      "loss_iou": 0.287109375,
      "loss_num": 0.034423828125,
      "loss_xval": 0.75,
      "num_input_tokens_seen": 116188860,
      "step": 2074
    },
    {
      "epoch": 4.621380846325167,
      "grad_norm": 15.175333976745605,
      "learning_rate": 1e-06,
      "loss": 0.8667,
      "num_input_tokens_seen": 116243408,
      "step": 2075
    },
    {
      "epoch": 4.621380846325167,
      "loss": 0.890459418296814,
      "loss_ce": 0.00032265347545035183,
      "loss_iou": 0.384765625,
      "loss_num": 0.024169921875,
      "loss_xval": 0.890625,
      "num_input_tokens_seen": 116243408,
      "step": 2075
    },
    {
      "epoch": 4.6236080178173715,
      "grad_norm": 13.573973655700684,
      "learning_rate": 1e-06,
      "loss": 1.0168,
      "num_input_tokens_seen": 116300164,
      "step": 2076
    },
    {
      "epoch": 4.6236080178173715,
      "loss": 1.1477470397949219,
      "loss_ce": 0.0002860655076801777,
      "loss_iou": 0.486328125,
      "loss_num": 0.035400390625,
      "loss_xval": 1.1484375,
      "num_input_tokens_seen": 116300164,
      "step": 2076
    },
    {
      "epoch": 4.625835189309576,
      "grad_norm": 19.694929122924805,
      "learning_rate": 1e-06,
      "loss": 0.8291,
      "num_input_tokens_seen": 116357772,
      "step": 2077
    },
    {
      "epoch": 4.625835189309576,
      "loss": 0.9049163460731506,
      "loss_ce": 0.00037533161230385303,
      "loss_iou": 0.35546875,
      "loss_num": 0.038330078125,
      "loss_xval": 0.90625,
      "num_input_tokens_seen": 116357772,
      "step": 2077
    },
    {
      "epoch": 4.628062360801781,
      "grad_norm": 14.48517894744873,
      "learning_rate": 1e-06,
      "loss": 0.8624,
      "num_input_tokens_seen": 116413892,
      "step": 2078
    },
    {
      "epoch": 4.628062360801781,
      "loss": 0.7932718992233276,
      "loss_ce": 0.0005472734337672591,
      "loss_iou": 0.330078125,
      "loss_num": 0.02685546875,
      "loss_xval": 0.79296875,
      "num_input_tokens_seen": 116413892,
      "step": 2078
    },
    {
      "epoch": 4.630289532293987,
      "grad_norm": 26.142356872558594,
      "learning_rate": 1e-06,
      "loss": 0.9727,
      "num_input_tokens_seen": 116471888,
      "step": 2079
    },
    {
      "epoch": 4.630289532293987,
      "loss": 1.051328420639038,
      "loss_ce": 0.0003031002124771476,
      "loss_iou": 0.458984375,
      "loss_num": 0.026611328125,
      "loss_xval": 1.0546875,
      "num_input_tokens_seen": 116471888,
      "step": 2079
    },
    {
      "epoch": 4.632516703786192,
      "grad_norm": 38.62444305419922,
      "learning_rate": 1e-06,
      "loss": 0.7796,
      "num_input_tokens_seen": 116527800,
      "step": 2080
    },
    {
      "epoch": 4.632516703786192,
      "loss": 0.8833059668540955,
      "loss_ce": 0.00024934165412560105,
      "loss_iou": 0.404296875,
      "loss_num": 0.01519775390625,
      "loss_xval": 0.8828125,
      "num_input_tokens_seen": 116527800,
      "step": 2080
    },
    {
      "epoch": 4.634743875278397,
      "grad_norm": 26.539600372314453,
      "learning_rate": 1e-06,
      "loss": 0.9048,
      "num_input_tokens_seen": 116583864,
      "step": 2081
    },
    {
      "epoch": 4.634743875278397,
      "loss": 0.9949434995651245,
      "loss_ce": 0.0005587629275396466,
      "loss_iou": 0.4453125,
      "loss_num": 0.0211181640625,
      "loss_xval": 0.99609375,
      "num_input_tokens_seen": 116583864,
      "step": 2081
    },
    {
      "epoch": 4.636971046770602,
      "grad_norm": 19.73139190673828,
      "learning_rate": 1e-06,
      "loss": 0.6691,
      "num_input_tokens_seen": 116635540,
      "step": 2082
    },
    {
      "epoch": 4.636971046770602,
      "loss": 0.6694454550743103,
      "loss_ce": 0.0009884194005280733,
      "loss_iou": 0.2578125,
      "loss_num": 0.0302734375,
      "loss_xval": 0.66796875,
      "num_input_tokens_seen": 116635540,
      "step": 2082
    },
    {
      "epoch": 4.639198218262806,
      "grad_norm": 22.97101593017578,
      "learning_rate": 1e-06,
      "loss": 0.6584,
      "num_input_tokens_seen": 116691464,
      "step": 2083
    },
    {
      "epoch": 4.639198218262806,
      "loss": 0.5923573970794678,
      "loss_ce": 0.0003163928631693125,
      "loss_iou": 0.265625,
      "loss_num": 0.012451171875,
      "loss_xval": 0.59375,
      "num_input_tokens_seen": 116691464,
      "step": 2083
    },
    {
      "epoch": 4.641425389755011,
      "grad_norm": 21.242584228515625,
      "learning_rate": 1e-06,
      "loss": 0.7134,
      "num_input_tokens_seen": 116743160,
      "step": 2084
    },
    {
      "epoch": 4.641425389755011,
      "loss": 0.5542248487472534,
      "loss_ce": 0.0002697905874811113,
      "loss_iou": 0.2177734375,
      "loss_num": 0.02392578125,
      "loss_xval": 0.5546875,
      "num_input_tokens_seen": 116743160,
      "step": 2084
    },
    {
      "epoch": 4.643652561247216,
      "grad_norm": 25.813098907470703,
      "learning_rate": 1e-06,
      "loss": 0.7164,
      "num_input_tokens_seen": 116797988,
      "step": 2085
    },
    {
      "epoch": 4.643652561247216,
      "loss": 0.5944602489471436,
      "loss_ce": 0.00022195720521267503,
      "loss_iou": 0.2578125,
      "loss_num": 0.0159912109375,
      "loss_xval": 0.59375,
      "num_input_tokens_seen": 116797988,
      "step": 2085
    },
    {
      "epoch": 4.645879732739421,
      "grad_norm": 35.45105743408203,
      "learning_rate": 1e-06,
      "loss": 0.9267,
      "num_input_tokens_seen": 116853140,
      "step": 2086
    },
    {
      "epoch": 4.645879732739421,
      "loss": 0.6704750657081604,
      "loss_ce": 0.0005531828501261771,
      "loss_iou": 0.296875,
      "loss_num": 0.0155029296875,
      "loss_xval": 0.671875,
      "num_input_tokens_seen": 116853140,
      "step": 2086
    },
    {
      "epoch": 4.648106904231626,
      "grad_norm": 19.315475463867188,
      "learning_rate": 1e-06,
      "loss": 0.7122,
      "num_input_tokens_seen": 116909096,
      "step": 2087
    },
    {
      "epoch": 4.648106904231626,
      "loss": 0.5913118720054626,
      "loss_ce": 0.0002473921631462872,
      "loss_iou": 0.251953125,
      "loss_num": 0.0177001953125,
      "loss_xval": 0.58984375,
      "num_input_tokens_seen": 116909096,
      "step": 2087
    },
    {
      "epoch": 4.650334075723831,
      "grad_norm": 27.988327026367188,
      "learning_rate": 1e-06,
      "loss": 0.8224,
      "num_input_tokens_seen": 116966212,
      "step": 2088
    },
    {
      "epoch": 4.650334075723831,
      "loss": 0.9297181367874146,
      "loss_ce": 0.00027486798353493214,
      "loss_iou": 0.376953125,
      "loss_num": 0.03515625,
      "loss_xval": 0.9296875,
      "num_input_tokens_seen": 116966212,
      "step": 2088
    },
    {
      "epoch": 4.652561247216036,
      "grad_norm": 16.152359008789062,
      "learning_rate": 1e-06,
      "loss": 0.5982,
      "num_input_tokens_seen": 117022116,
      "step": 2089
    },
    {
      "epoch": 4.652561247216036,
      "loss": 0.6331831216812134,
      "loss_ce": 0.0002486219455022365,
      "loss_iou": 0.265625,
      "loss_num": 0.0201416015625,
      "loss_xval": 0.6328125,
      "num_input_tokens_seen": 117022116,
      "step": 2089
    },
    {
      "epoch": 4.6547884187082404,
      "grad_norm": 14.966622352600098,
      "learning_rate": 1e-06,
      "loss": 0.8524,
      "num_input_tokens_seen": 117077144,
      "step": 2090
    },
    {
      "epoch": 4.6547884187082404,
      "loss": 0.8334739208221436,
      "loss_ce": 0.0007102236268110573,
      "loss_iou": 0.341796875,
      "loss_num": 0.0294189453125,
      "loss_xval": 0.83203125,
      "num_input_tokens_seen": 117077144,
      "step": 2090
    },
    {
      "epoch": 4.657015590200445,
      "grad_norm": 18.069534301757812,
      "learning_rate": 1e-06,
      "loss": 0.6994,
      "num_input_tokens_seen": 117132324,
      "step": 2091
    },
    {
      "epoch": 4.657015590200445,
      "loss": 0.6397680044174194,
      "loss_ce": 0.00024168557138182223,
      "loss_iou": 0.283203125,
      "loss_num": 0.014404296875,
      "loss_xval": 0.640625,
      "num_input_tokens_seen": 117132324,
      "step": 2091
    },
    {
      "epoch": 4.65924276169265,
      "grad_norm": 16.776460647583008,
      "learning_rate": 1e-06,
      "loss": 0.6939,
      "num_input_tokens_seen": 117192008,
      "step": 2092
    },
    {
      "epoch": 4.65924276169265,
      "loss": 0.8577221632003784,
      "loss_ce": 0.0003002659068442881,
      "loss_iou": 0.3671875,
      "loss_num": 0.0250244140625,
      "loss_xval": 0.859375,
      "num_input_tokens_seen": 117192008,
      "step": 2092
    },
    {
      "epoch": 4.661469933184855,
      "grad_norm": 12.129549980163574,
      "learning_rate": 1e-06,
      "loss": 0.7361,
      "num_input_tokens_seen": 117248744,
      "step": 2093
    },
    {
      "epoch": 4.661469933184855,
      "loss": 0.8417858481407166,
      "loss_ce": 0.00035519120865501463,
      "loss_iou": 0.3671875,
      "loss_num": 0.0216064453125,
      "loss_xval": 0.83984375,
      "num_input_tokens_seen": 117248744,
      "step": 2093
    },
    {
      "epoch": 4.66369710467706,
      "grad_norm": 47.98014450073242,
      "learning_rate": 1e-06,
      "loss": 1.0221,
      "num_input_tokens_seen": 117303972,
      "step": 2094
    },
    {
      "epoch": 4.66369710467706,
      "loss": 0.8582909107208252,
      "loss_ce": 0.0002586581977084279,
      "loss_iou": 0.365234375,
      "loss_num": 0.0255126953125,
      "loss_xval": 0.859375,
      "num_input_tokens_seen": 117303972,
      "step": 2094
    },
    {
      "epoch": 4.665924276169265,
      "grad_norm": 17.894067764282227,
      "learning_rate": 1e-06,
      "loss": 0.7274,
      "num_input_tokens_seen": 117360916,
      "step": 2095
    },
    {
      "epoch": 4.665924276169265,
      "loss": 0.9711999297142029,
      "loss_ce": 0.0002527014003135264,
      "loss_iou": 0.39453125,
      "loss_num": 0.036376953125,
      "loss_xval": 0.97265625,
      "num_input_tokens_seen": 117360916,
      "step": 2095
    },
    {
      "epoch": 4.66815144766147,
      "grad_norm": 22.916641235351562,
      "learning_rate": 1e-06,
      "loss": 0.7064,
      "num_input_tokens_seen": 117416304,
      "step": 2096
    },
    {
      "epoch": 4.66815144766147,
      "loss": 0.7692503929138184,
      "loss_ce": 0.00045159138971939683,
      "loss_iou": 0.3125,
      "loss_num": 0.0289306640625,
      "loss_xval": 0.76953125,
      "num_input_tokens_seen": 117416304,
      "step": 2096
    },
    {
      "epoch": 4.6703786191536745,
      "grad_norm": 17.155508041381836,
      "learning_rate": 1e-06,
      "loss": 0.9104,
      "num_input_tokens_seen": 117473004,
      "step": 2097
    },
    {
      "epoch": 4.6703786191536745,
      "loss": 1.1037812232971191,
      "loss_ce": 0.0002655387215781957,
      "loss_iou": 0.453125,
      "loss_num": 0.03955078125,
      "loss_xval": 1.1015625,
      "num_input_tokens_seen": 117473004,
      "step": 2097
    },
    {
      "epoch": 4.67260579064588,
      "grad_norm": 17.130844116210938,
      "learning_rate": 1e-06,
      "loss": 0.7475,
      "num_input_tokens_seen": 117528424,
      "step": 2098
    },
    {
      "epoch": 4.67260579064588,
      "loss": 0.8190168738365173,
      "loss_ce": 0.0004133854527026415,
      "loss_iou": 0.33984375,
      "loss_num": 0.02783203125,
      "loss_xval": 0.8203125,
      "num_input_tokens_seen": 117528424,
      "step": 2098
    },
    {
      "epoch": 4.674832962138085,
      "grad_norm": 16.85631561279297,
      "learning_rate": 1e-06,
      "loss": 0.5101,
      "num_input_tokens_seen": 117581876,
      "step": 2099
    },
    {
      "epoch": 4.674832962138085,
      "loss": 0.469228595495224,
      "loss_ce": 0.00023444523685611784,
      "loss_iou": 0.19921875,
      "loss_num": 0.01416015625,
      "loss_xval": 0.46875,
      "num_input_tokens_seen": 117581876,
      "step": 2099
    },
    {
      "epoch": 4.67706013363029,
      "grad_norm": 17.609556198120117,
      "learning_rate": 1e-06,
      "loss": 1.0525,
      "num_input_tokens_seen": 117638468,
      "step": 2100
    },
    {
      "epoch": 4.67706013363029,
      "loss": 0.9739843606948853,
      "loss_ce": 0.0003515969729050994,
      "loss_iou": 0.43359375,
      "loss_num": 0.0213623046875,
      "loss_xval": 0.97265625,
      "num_input_tokens_seen": 117638468,
      "step": 2100
    },
    {
      "epoch": 4.679287305122495,
      "grad_norm": 20.728130340576172,
      "learning_rate": 1e-06,
      "loss": 0.7308,
      "num_input_tokens_seen": 117694144,
      "step": 2101
    },
    {
      "epoch": 4.679287305122495,
      "loss": 0.7154548168182373,
      "loss_ce": 0.0003668900462798774,
      "loss_iou": 0.298828125,
      "loss_num": 0.0238037109375,
      "loss_xval": 0.71484375,
      "num_input_tokens_seen": 117694144,
      "step": 2101
    },
    {
      "epoch": 4.6815144766147,
      "grad_norm": 18.357759475708008,
      "learning_rate": 1e-06,
      "loss": 0.9824,
      "num_input_tokens_seen": 117747644,
      "step": 2102
    },
    {
      "epoch": 4.6815144766147,
      "loss": 0.903360903263092,
      "loss_ce": 0.0002847156720235944,
      "loss_iou": 0.3515625,
      "loss_num": 0.039794921875,
      "loss_xval": 0.90234375,
      "num_input_tokens_seen": 117747644,
      "step": 2102
    },
    {
      "epoch": 4.6837416481069045,
      "grad_norm": 18.632221221923828,
      "learning_rate": 1e-06,
      "loss": 0.7674,
      "num_input_tokens_seen": 117802256,
      "step": 2103
    },
    {
      "epoch": 4.6837416481069045,
      "loss": 0.8225224018096924,
      "loss_ce": 0.00025684869615361094,
      "loss_iou": 0.337890625,
      "loss_num": 0.029541015625,
      "loss_xval": 0.8203125,
      "num_input_tokens_seen": 117802256,
      "step": 2103
    },
    {
      "epoch": 4.685968819599109,
      "grad_norm": 22.77741813659668,
      "learning_rate": 1e-06,
      "loss": 0.6407,
      "num_input_tokens_seen": 117857308,
      "step": 2104
    },
    {
      "epoch": 4.685968819599109,
      "loss": 0.6071910858154297,
      "loss_ce": 0.0002575043763499707,
      "loss_iou": 0.2216796875,
      "loss_num": 0.03271484375,
      "loss_xval": 0.60546875,
      "num_input_tokens_seen": 117857308,
      "step": 2104
    },
    {
      "epoch": 4.688195991091314,
      "grad_norm": 12.895187377929688,
      "learning_rate": 1e-06,
      "loss": 0.5916,
      "num_input_tokens_seen": 117913116,
      "step": 2105
    },
    {
      "epoch": 4.688195991091314,
      "loss": 0.7664899230003357,
      "loss_ce": 0.0002545661700423807,
      "loss_iou": 0.330078125,
      "loss_num": 0.020751953125,
      "loss_xval": 0.765625,
      "num_input_tokens_seen": 117913116,
      "step": 2105
    },
    {
      "epoch": 4.690423162583519,
      "grad_norm": 36.03581237792969,
      "learning_rate": 1e-06,
      "loss": 0.7424,
      "num_input_tokens_seen": 117970296,
      "step": 2106
    },
    {
      "epoch": 4.690423162583519,
      "loss": 0.785170316696167,
      "loss_ce": 0.00025823366013355553,
      "loss_iou": 0.341796875,
      "loss_num": 0.0205078125,
      "loss_xval": 0.78515625,
      "num_input_tokens_seen": 117970296,
      "step": 2106
    },
    {
      "epoch": 4.692650334075724,
      "grad_norm": 24.407381057739258,
      "learning_rate": 1e-06,
      "loss": 0.7918,
      "num_input_tokens_seen": 118024948,
      "step": 2107
    },
    {
      "epoch": 4.692650334075724,
      "loss": 0.6611515283584595,
      "loss_ce": 0.00026292851543985307,
      "loss_iou": 0.29296875,
      "loss_num": 0.01519775390625,
      "loss_xval": 0.66015625,
      "num_input_tokens_seen": 118024948,
      "step": 2107
    },
    {
      "epoch": 4.694877505567929,
      "grad_norm": 30.365440368652344,
      "learning_rate": 1e-06,
      "loss": 0.8189,
      "num_input_tokens_seen": 118081340,
      "step": 2108
    },
    {
      "epoch": 4.694877505567929,
      "loss": 0.9890696406364441,
      "loss_ce": 0.0003001365694217384,
      "loss_iou": 0.384765625,
      "loss_num": 0.043701171875,
      "loss_xval": 0.98828125,
      "num_input_tokens_seen": 118081340,
      "step": 2108
    },
    {
      "epoch": 4.697104677060134,
      "grad_norm": 14.271038055419922,
      "learning_rate": 1e-06,
      "loss": 0.8856,
      "num_input_tokens_seen": 118139104,
      "step": 2109
    },
    {
      "epoch": 4.697104677060134,
      "loss": 1.0308204889297485,
      "loss_ce": 0.0003028618230018765,
      "loss_iou": 0.4296875,
      "loss_num": 0.03466796875,
      "loss_xval": 1.03125,
      "num_input_tokens_seen": 118139104,
      "step": 2109
    },
    {
      "epoch": 4.6993318485523385,
      "grad_norm": 22.121456146240234,
      "learning_rate": 1e-06,
      "loss": 0.7422,
      "num_input_tokens_seen": 118194740,
      "step": 2110
    },
    {
      "epoch": 4.6993318485523385,
      "loss": 0.7307307124137878,
      "loss_ce": 0.00026196378166787326,
      "loss_iou": 0.32421875,
      "loss_num": 0.0167236328125,
      "loss_xval": 0.73046875,
      "num_input_tokens_seen": 118194740,
      "step": 2110
    },
    {
      "epoch": 4.701559020044543,
      "grad_norm": 21.364805221557617,
      "learning_rate": 1e-06,
      "loss": 0.6672,
      "num_input_tokens_seen": 118248584,
      "step": 2111
    },
    {
      "epoch": 4.701559020044543,
      "loss": 0.8600109815597534,
      "loss_ce": 0.000391850684536621,
      "loss_iou": 0.37890625,
      "loss_num": 0.02001953125,
      "loss_xval": 0.859375,
      "num_input_tokens_seen": 118248584,
      "step": 2111
    },
    {
      "epoch": 4.703786191536748,
      "grad_norm": 21.18739128112793,
      "learning_rate": 1e-06,
      "loss": 0.7989,
      "num_input_tokens_seen": 118306024,
      "step": 2112
    },
    {
      "epoch": 4.703786191536748,
      "loss": 0.8781961798667908,
      "loss_ce": 0.00026655703550204635,
      "loss_iou": 0.39453125,
      "loss_num": 0.0177001953125,
      "loss_xval": 0.87890625,
      "num_input_tokens_seen": 118306024,
      "step": 2112
    },
    {
      "epoch": 4.706013363028953,
      "grad_norm": 16.3918399810791,
      "learning_rate": 1e-06,
      "loss": 0.7998,
      "num_input_tokens_seen": 118361808,
      "step": 2113
    },
    {
      "epoch": 4.706013363028953,
      "loss": 0.5792876482009888,
      "loss_ce": 0.0003081717586610466,
      "loss_iou": 0.25390625,
      "loss_num": 0.01397705078125,
      "loss_xval": 0.578125,
      "num_input_tokens_seen": 118361808,
      "step": 2113
    },
    {
      "epoch": 4.708240534521158,
      "grad_norm": 21.2349910736084,
      "learning_rate": 1e-06,
      "loss": 0.6949,
      "num_input_tokens_seen": 118419444,
      "step": 2114
    },
    {
      "epoch": 4.708240534521158,
      "loss": 0.6933687925338745,
      "loss_ce": 0.00025352693046443164,
      "loss_iou": 0.296875,
      "loss_num": 0.019775390625,
      "loss_xval": 0.69140625,
      "num_input_tokens_seen": 118419444,
      "step": 2114
    },
    {
      "epoch": 4.710467706013363,
      "grad_norm": 20.82132911682129,
      "learning_rate": 1e-06,
      "loss": 0.6637,
      "num_input_tokens_seen": 118476632,
      "step": 2115
    },
    {
      "epoch": 4.710467706013363,
      "loss": 0.6345095038414001,
      "loss_ce": 0.0004763126198668033,
      "loss_iou": 0.28125,
      "loss_num": 0.01422119140625,
      "loss_xval": 0.6328125,
      "num_input_tokens_seen": 118476632,
      "step": 2115
    },
    {
      "epoch": 4.712694877505568,
      "grad_norm": 17.699792861938477,
      "learning_rate": 1e-06,
      "loss": 0.6965,
      "num_input_tokens_seen": 118532612,
      "step": 2116
    },
    {
      "epoch": 4.712694877505568,
      "loss": 0.6084548234939575,
      "loss_ce": 0.0002699895412661135,
      "loss_iou": 0.265625,
      "loss_num": 0.015625,
      "loss_xval": 0.609375,
      "num_input_tokens_seen": 118532612,
      "step": 2116
    },
    {
      "epoch": 4.714922048997773,
      "grad_norm": 15.86410903930664,
      "learning_rate": 1e-06,
      "loss": 0.9501,
      "num_input_tokens_seen": 118589372,
      "step": 2117
    },
    {
      "epoch": 4.714922048997773,
      "loss": 1.0604746341705322,
      "loss_ce": 0.000416038790717721,
      "loss_iou": 0.42578125,
      "loss_num": 0.041748046875,
      "loss_xval": 1.0625,
      "num_input_tokens_seen": 118589372,
      "step": 2117
    },
    {
      "epoch": 4.717149220489977,
      "grad_norm": 20.29158592224121,
      "learning_rate": 1e-06,
      "loss": 0.6542,
      "num_input_tokens_seen": 118644080,
      "step": 2118
    },
    {
      "epoch": 4.717149220489977,
      "loss": 0.6954843401908875,
      "loss_ce": 0.00041598049574531615,
      "loss_iou": 0.29296875,
      "loss_num": 0.02197265625,
      "loss_xval": 0.6953125,
      "num_input_tokens_seen": 118644080,
      "step": 2118
    },
    {
      "epoch": 4.719376391982182,
      "grad_norm": 16.680042266845703,
      "learning_rate": 1e-06,
      "loss": 0.7097,
      "num_input_tokens_seen": 118698048,
      "step": 2119
    },
    {
      "epoch": 4.719376391982182,
      "loss": 0.8289152383804321,
      "loss_ce": 0.00030192872509360313,
      "loss_iou": 0.365234375,
      "loss_num": 0.0196533203125,
      "loss_xval": 0.828125,
      "num_input_tokens_seen": 118698048,
      "step": 2119
    },
    {
      "epoch": 4.721603563474387,
      "grad_norm": 20.128711700439453,
      "learning_rate": 1e-06,
      "loss": 0.7996,
      "num_input_tokens_seen": 118754912,
      "step": 2120
    },
    {
      "epoch": 4.721603563474387,
      "loss": 0.7679637670516968,
      "loss_ce": 0.0003856367547996342,
      "loss_iou": 0.33984375,
      "loss_num": 0.017578125,
      "loss_xval": 0.765625,
      "num_input_tokens_seen": 118754912,
      "step": 2120
    },
    {
      "epoch": 4.723830734966592,
      "grad_norm": 26.825408935546875,
      "learning_rate": 1e-06,
      "loss": 1.0429,
      "num_input_tokens_seen": 118809528,
      "step": 2121
    },
    {
      "epoch": 4.723830734966592,
      "loss": 0.6894991993904114,
      "loss_ce": 0.0002902133564930409,
      "loss_iou": 0.275390625,
      "loss_num": 0.02783203125,
      "loss_xval": 0.6875,
      "num_input_tokens_seen": 118809528,
      "step": 2121
    },
    {
      "epoch": 4.726057906458797,
      "grad_norm": 35.720130920410156,
      "learning_rate": 1e-06,
      "loss": 0.7786,
      "num_input_tokens_seen": 118866560,
      "step": 2122
    },
    {
      "epoch": 4.726057906458797,
      "loss": 0.766879677772522,
      "loss_ce": 0.0002781344810500741,
      "loss_iou": 0.298828125,
      "loss_num": 0.033935546875,
      "loss_xval": 0.765625,
      "num_input_tokens_seen": 118866560,
      "step": 2122
    },
    {
      "epoch": 4.728285077951003,
      "grad_norm": 39.63307189941406,
      "learning_rate": 1e-06,
      "loss": 0.7441,
      "num_input_tokens_seen": 118922208,
      "step": 2123
    },
    {
      "epoch": 4.728285077951003,
      "loss": 0.6776679158210754,
      "loss_ce": 0.0002997248957399279,
      "loss_iou": 0.271484375,
      "loss_num": 0.02685546875,
      "loss_xval": 0.67578125,
      "num_input_tokens_seen": 118922208,
      "step": 2123
    },
    {
      "epoch": 4.7305122494432075,
      "grad_norm": 29.788860321044922,
      "learning_rate": 1e-06,
      "loss": 0.7002,
      "num_input_tokens_seen": 118976912,
      "step": 2124
    },
    {
      "epoch": 4.7305122494432075,
      "loss": 0.8606002926826477,
      "loss_ce": 0.00024869441404007375,
      "loss_iou": 0.3359375,
      "loss_num": 0.0380859375,
      "loss_xval": 0.859375,
      "num_input_tokens_seen": 118976912,
      "step": 2124
    },
    {
      "epoch": 4.732739420935412,
      "grad_norm": 17.970624923706055,
      "learning_rate": 1e-06,
      "loss": 0.6788,
      "num_input_tokens_seen": 119032532,
      "step": 2125
    },
    {
      "epoch": 4.732739420935412,
      "loss": 0.9126179814338684,
      "loss_ce": 0.0005085880402475595,
      "loss_iou": 0.38671875,
      "loss_num": 0.0279541015625,
      "loss_xval": 0.9140625,
      "num_input_tokens_seen": 119032532,
      "step": 2125
    },
    {
      "epoch": 4.734966592427617,
      "grad_norm": 40.0213508605957,
      "learning_rate": 1e-06,
      "loss": 0.6293,
      "num_input_tokens_seen": 119089356,
      "step": 2126
    },
    {
      "epoch": 4.734966592427617,
      "loss": 0.6347508430480957,
      "loss_ce": 0.00022931784042157233,
      "loss_iou": 0.28515625,
      "loss_num": 0.0130615234375,
      "loss_xval": 0.6328125,
      "num_input_tokens_seen": 119089356,
      "step": 2126
    },
    {
      "epoch": 4.737193763919822,
      "grad_norm": 19.045475006103516,
      "learning_rate": 1e-06,
      "loss": 0.7774,
      "num_input_tokens_seen": 119144504,
      "step": 2127
    },
    {
      "epoch": 4.737193763919822,
      "loss": 0.9141049385070801,
      "loss_ce": 0.0002866187132894993,
      "loss_iou": 0.400390625,
      "loss_num": 0.0223388671875,
      "loss_xval": 0.9140625,
      "num_input_tokens_seen": 119144504,
      "step": 2127
    },
    {
      "epoch": 4.739420935412027,
      "grad_norm": 17.02509880065918,
      "learning_rate": 1e-06,
      "loss": 0.7687,
      "num_input_tokens_seen": 119201984,
      "step": 2128
    },
    {
      "epoch": 4.739420935412027,
      "loss": 0.7351812124252319,
      "loss_ce": 0.0003179654595442116,
      "loss_iou": 0.322265625,
      "loss_num": 0.018310546875,
      "loss_xval": 0.734375,
      "num_input_tokens_seen": 119201984,
      "step": 2128
    },
    {
      "epoch": 4.741648106904232,
      "grad_norm": 42.980289459228516,
      "learning_rate": 1e-06,
      "loss": 0.6333,
      "num_input_tokens_seen": 119259860,
      "step": 2129
    },
    {
      "epoch": 4.741648106904232,
      "loss": 0.538473904132843,
      "loss_ce": 0.0002658716693986207,
      "loss_iou": 0.24609375,
      "loss_num": 0.0091552734375,
      "loss_xval": 0.5390625,
      "num_input_tokens_seen": 119259860,
      "step": 2129
    },
    {
      "epoch": 4.743875278396437,
      "grad_norm": 15.055540084838867,
      "learning_rate": 1e-06,
      "loss": 0.8069,
      "num_input_tokens_seen": 119317616,
      "step": 2130
    },
    {
      "epoch": 4.743875278396437,
      "loss": 0.8891835808753967,
      "loss_ce": 0.0002675617579370737,
      "loss_iou": 0.3671875,
      "loss_num": 0.0306396484375,
      "loss_xval": 0.890625,
      "num_input_tokens_seen": 119317616,
      "step": 2130
    },
    {
      "epoch": 4.7461024498886415,
      "grad_norm": 20.52086067199707,
      "learning_rate": 1e-06,
      "loss": 0.8202,
      "num_input_tokens_seen": 119373908,
      "step": 2131
    },
    {
      "epoch": 4.7461024498886415,
      "loss": 0.8626622557640076,
      "loss_ce": 0.00035758258309215307,
      "loss_iou": 0.359375,
      "loss_num": 0.0286865234375,
      "loss_xval": 0.86328125,
      "num_input_tokens_seen": 119373908,
      "step": 2131
    },
    {
      "epoch": 4.748329621380846,
      "grad_norm": 24.670711517333984,
      "learning_rate": 1e-06,
      "loss": 0.8469,
      "num_input_tokens_seen": 119429720,
      "step": 2132
    },
    {
      "epoch": 4.748329621380846,
      "loss": 0.7893211841583252,
      "loss_ce": 0.00025862548500299454,
      "loss_iou": 0.34375,
      "loss_num": 0.0201416015625,
      "loss_xval": 0.7890625,
      "num_input_tokens_seen": 119429720,
      "step": 2132
    },
    {
      "epoch": 4.750556792873051,
      "grad_norm": 21.29181671142578,
      "learning_rate": 1e-06,
      "loss": 0.6766,
      "num_input_tokens_seen": 119484036,
      "step": 2133
    },
    {
      "epoch": 4.750556792873051,
      "loss": 0.609359860420227,
      "loss_ce": 0.00022898372844792902,
      "loss_iou": 0.240234375,
      "loss_num": 0.02587890625,
      "loss_xval": 0.609375,
      "num_input_tokens_seen": 119484036,
      "step": 2133
    },
    {
      "epoch": 4.752783964365256,
      "grad_norm": 54.08604431152344,
      "learning_rate": 1e-06,
      "loss": 0.6534,
      "num_input_tokens_seen": 119537184,
      "step": 2134
    },
    {
      "epoch": 4.752783964365256,
      "loss": 0.6004340648651123,
      "loss_ce": 0.0003363891737535596,
      "loss_iou": 0.234375,
      "loss_num": 0.026123046875,
      "loss_xval": 0.6015625,
      "num_input_tokens_seen": 119537184,
      "step": 2134
    },
    {
      "epoch": 4.755011135857461,
      "grad_norm": 24.42340660095215,
      "learning_rate": 1e-06,
      "loss": 0.66,
      "num_input_tokens_seen": 119591936,
      "step": 2135
    },
    {
      "epoch": 4.755011135857461,
      "loss": 0.762969970703125,
      "loss_ce": 0.0002747006365098059,
      "loss_iou": 0.322265625,
      "loss_num": 0.023681640625,
      "loss_xval": 0.76171875,
      "num_input_tokens_seen": 119591936,
      "step": 2135
    },
    {
      "epoch": 4.757238307349666,
      "grad_norm": 18.103748321533203,
      "learning_rate": 1e-06,
      "loss": 0.7105,
      "num_input_tokens_seen": 119650452,
      "step": 2136
    },
    {
      "epoch": 4.757238307349666,
      "loss": 0.8064308166503906,
      "loss_ce": 0.0002785040996968746,
      "loss_iou": 0.35546875,
      "loss_num": 0.0191650390625,
      "loss_xval": 0.8046875,
      "num_input_tokens_seen": 119650452,
      "step": 2136
    },
    {
      "epoch": 4.759465478841871,
      "grad_norm": 25.047883987426758,
      "learning_rate": 1e-06,
      "loss": 0.9671,
      "num_input_tokens_seen": 119705364,
      "step": 2137
    },
    {
      "epoch": 4.759465478841871,
      "loss": 0.9405218958854675,
      "loss_ce": 0.0003363378345966339,
      "loss_iou": 0.390625,
      "loss_num": 0.03173828125,
      "loss_xval": 0.94140625,
      "num_input_tokens_seen": 119705364,
      "step": 2137
    },
    {
      "epoch": 4.7616926503340755,
      "grad_norm": 11.171337127685547,
      "learning_rate": 1e-06,
      "loss": 0.4534,
      "num_input_tokens_seen": 119761084,
      "step": 2138
    },
    {
      "epoch": 4.7616926503340755,
      "loss": 0.43443742394447327,
      "loss_ce": 0.00023330794647336006,
      "loss_iou": 0.1904296875,
      "loss_num": 0.01055908203125,
      "loss_xval": 0.43359375,
      "num_input_tokens_seen": 119761084,
      "step": 2138
    },
    {
      "epoch": 4.76391982182628,
      "grad_norm": 14.953879356384277,
      "learning_rate": 1e-06,
      "loss": 0.8728,
      "num_input_tokens_seen": 119816688,
      "step": 2139
    },
    {
      "epoch": 4.76391982182628,
      "loss": 0.6874940991401672,
      "loss_ce": 0.00029927384457550943,
      "loss_iou": 0.3046875,
      "loss_num": 0.0157470703125,
      "loss_xval": 0.6875,
      "num_input_tokens_seen": 119816688,
      "step": 2139
    },
    {
      "epoch": 4.766146993318485,
      "grad_norm": 25.45313262939453,
      "learning_rate": 1e-06,
      "loss": 0.8034,
      "num_input_tokens_seen": 119874648,
      "step": 2140
    },
    {
      "epoch": 4.766146993318485,
      "loss": 0.8113042116165161,
      "loss_ce": 0.00026898583746515214,
      "loss_iou": 0.328125,
      "loss_num": 0.03125,
      "loss_xval": 0.8125,
      "num_input_tokens_seen": 119874648,
      "step": 2140
    },
    {
      "epoch": 4.76837416481069,
      "grad_norm": 22.319425582885742,
      "learning_rate": 1e-06,
      "loss": 0.7391,
      "num_input_tokens_seen": 119930376,
      "step": 2141
    },
    {
      "epoch": 4.76837416481069,
      "loss": 0.7610592246055603,
      "loss_ce": 0.0005611696396954358,
      "loss_iou": 0.314453125,
      "loss_num": 0.0267333984375,
      "loss_xval": 0.76171875,
      "num_input_tokens_seen": 119930376,
      "step": 2141
    },
    {
      "epoch": 4.770601336302895,
      "grad_norm": 27.872591018676758,
      "learning_rate": 1e-06,
      "loss": 0.7955,
      "num_input_tokens_seen": 119985268,
      "step": 2142
    },
    {
      "epoch": 4.770601336302895,
      "loss": 0.5949513912200928,
      "loss_ce": 0.00022480121697299182,
      "loss_iou": 0.2490234375,
      "loss_num": 0.019287109375,
      "loss_xval": 0.59375,
      "num_input_tokens_seen": 119985268,
      "step": 2142
    },
    {
      "epoch": 4.772828507795101,
      "grad_norm": 21.563512802124023,
      "learning_rate": 1e-06,
      "loss": 0.6652,
      "num_input_tokens_seen": 120041572,
      "step": 2143
    },
    {
      "epoch": 4.772828507795101,
      "loss": 0.6074680089950562,
      "loss_ce": 0.0002902495616581291,
      "loss_iou": 0.26171875,
      "loss_num": 0.0166015625,
      "loss_xval": 0.60546875,
      "num_input_tokens_seen": 120041572,
      "step": 2143
    },
    {
      "epoch": 4.775055679287306,
      "grad_norm": 21.816646575927734,
      "learning_rate": 1e-06,
      "loss": 0.7993,
      "num_input_tokens_seen": 120097280,
      "step": 2144
    },
    {
      "epoch": 4.775055679287306,
      "loss": 0.8655096292495728,
      "loss_ce": 0.00027528638020157814,
      "loss_iou": 0.3828125,
      "loss_num": 0.01953125,
      "loss_xval": 0.8671875,
      "num_input_tokens_seen": 120097280,
      "step": 2144
    },
    {
      "epoch": 4.77728285077951,
      "grad_norm": 17.00129508972168,
      "learning_rate": 1e-06,
      "loss": 0.8654,
      "num_input_tokens_seen": 120152928,
      "step": 2145
    },
    {
      "epoch": 4.77728285077951,
      "loss": 1.0312559604644775,
      "loss_ce": 0.00025006328360177577,
      "loss_iou": 0.427734375,
      "loss_num": 0.035400390625,
      "loss_xval": 1.03125,
      "num_input_tokens_seen": 120152928,
      "step": 2145
    },
    {
      "epoch": 4.779510022271715,
      "grad_norm": 23.580554962158203,
      "learning_rate": 1e-06,
      "loss": 0.912,
      "num_input_tokens_seen": 120211296,
      "step": 2146
    },
    {
      "epoch": 4.779510022271715,
      "loss": 0.8288871645927429,
      "loss_ce": 0.0005180308944545686,
      "loss_iou": 0.34375,
      "loss_num": 0.02783203125,
      "loss_xval": 0.828125,
      "num_input_tokens_seen": 120211296,
      "step": 2146
    },
    {
      "epoch": 4.78173719376392,
      "grad_norm": 18.791494369506836,
      "learning_rate": 1e-06,
      "loss": 0.7229,
      "num_input_tokens_seen": 120268204,
      "step": 2147
    },
    {
      "epoch": 4.78173719376392,
      "loss": 0.7734935283660889,
      "loss_ce": 0.000300163053907454,
      "loss_iou": 0.310546875,
      "loss_num": 0.0308837890625,
      "loss_xval": 0.7734375,
      "num_input_tokens_seen": 120268204,
      "step": 2147
    },
    {
      "epoch": 4.783964365256125,
      "grad_norm": 17.211793899536133,
      "learning_rate": 1e-06,
      "loss": 0.6943,
      "num_input_tokens_seen": 120320000,
      "step": 2148
    },
    {
      "epoch": 4.783964365256125,
      "loss": 0.8804046511650085,
      "loss_ce": 0.00027765982667915523,
      "loss_iou": 0.357421875,
      "loss_num": 0.032958984375,
      "loss_xval": 0.87890625,
      "num_input_tokens_seen": 120320000,
      "step": 2148
    },
    {
      "epoch": 4.78619153674833,
      "grad_norm": 16.56837272644043,
      "learning_rate": 1e-06,
      "loss": 0.555,
      "num_input_tokens_seen": 120375736,
      "step": 2149
    },
    {
      "epoch": 4.78619153674833,
      "loss": 0.4850603938102722,
      "loss_ce": 0.0003192013828083873,
      "loss_iou": 0.2138671875,
      "loss_num": 0.01116943359375,
      "loss_xval": 0.484375,
      "num_input_tokens_seen": 120375736,
      "step": 2149
    },
    {
      "epoch": 4.788418708240535,
      "grad_norm": 60.28794860839844,
      "learning_rate": 1e-06,
      "loss": 0.8792,
      "num_input_tokens_seen": 120429608,
      "step": 2150
    },
    {
      "epoch": 4.788418708240535,
      "loss": 0.890438437461853,
      "loss_ce": 0.00030169825186021626,
      "loss_iou": 0.38671875,
      "loss_num": 0.023193359375,
      "loss_xval": 0.890625,
      "num_input_tokens_seen": 120429608,
      "step": 2150
    },
    {
      "epoch": 4.79064587973274,
      "grad_norm": 25.588119506835938,
      "learning_rate": 1e-06,
      "loss": 0.705,
      "num_input_tokens_seen": 120484200,
      "step": 2151
    },
    {
      "epoch": 4.79064587973274,
      "loss": 0.8378915786743164,
      "loss_ce": 0.00036717430339194834,
      "loss_iou": 0.345703125,
      "loss_num": 0.0289306640625,
      "loss_xval": 0.8359375,
      "num_input_tokens_seen": 120484200,
      "step": 2151
    },
    {
      "epoch": 4.7928730512249444,
      "grad_norm": 20.40205955505371,
      "learning_rate": 1e-06,
      "loss": 0.7429,
      "num_input_tokens_seen": 120539644,
      "step": 2152
    },
    {
      "epoch": 4.7928730512249444,
      "loss": 0.6924837827682495,
      "loss_ce": 0.00034514523576945066,
      "loss_iou": 0.302734375,
      "loss_num": 0.017578125,
      "loss_xval": 0.69140625,
      "num_input_tokens_seen": 120539644,
      "step": 2152
    },
    {
      "epoch": 4.795100222717149,
      "grad_norm": 24.60614013671875,
      "learning_rate": 1e-06,
      "loss": 0.7615,
      "num_input_tokens_seen": 120594500,
      "step": 2153
    },
    {
      "epoch": 4.795100222717149,
      "loss": 0.7266230583190918,
      "loss_ce": 0.000487854442326352,
      "loss_iou": 0.302734375,
      "loss_num": 0.023681640625,
      "loss_xval": 0.7265625,
      "num_input_tokens_seen": 120594500,
      "step": 2153
    },
    {
      "epoch": 4.797327394209354,
      "grad_norm": 16.96647834777832,
      "learning_rate": 1e-06,
      "loss": 0.7819,
      "num_input_tokens_seen": 120649288,
      "step": 2154
    },
    {
      "epoch": 4.797327394209354,
      "loss": 0.6350046396255493,
      "loss_ce": 0.00023901589156594127,
      "loss_iou": 0.28125,
      "loss_num": 0.0145263671875,
      "loss_xval": 0.6328125,
      "num_input_tokens_seen": 120649288,
      "step": 2154
    },
    {
      "epoch": 4.799554565701559,
      "grad_norm": 18.894594192504883,
      "learning_rate": 1e-06,
      "loss": 0.9572,
      "num_input_tokens_seen": 120704828,
      "step": 2155
    },
    {
      "epoch": 4.799554565701559,
      "loss": 0.9965546131134033,
      "loss_ce": 0.000460872077383101,
      "loss_iou": 0.41796875,
      "loss_num": 0.031982421875,
      "loss_xval": 0.99609375,
      "num_input_tokens_seen": 120704828,
      "step": 2155
    },
    {
      "epoch": 4.801781737193764,
      "grad_norm": 19.677793502807617,
      "learning_rate": 1e-06,
      "loss": 0.8639,
      "num_input_tokens_seen": 120760452,
      "step": 2156
    },
    {
      "epoch": 4.801781737193764,
      "loss": 0.9670339822769165,
      "loss_ce": 0.00023707067884970456,
      "loss_iou": 0.40625,
      "loss_num": 0.0306396484375,
      "loss_xval": 0.96875,
      "num_input_tokens_seen": 120760452,
      "step": 2156
    },
    {
      "epoch": 4.804008908685969,
      "grad_norm": 28.458290100097656,
      "learning_rate": 1e-06,
      "loss": 0.8885,
      "num_input_tokens_seen": 120813932,
      "step": 2157
    },
    {
      "epoch": 4.804008908685969,
      "loss": 0.9386167526245117,
      "loss_ce": 0.0004453619185369462,
      "loss_iou": 0.37109375,
      "loss_num": 0.039794921875,
      "loss_xval": 0.9375,
      "num_input_tokens_seen": 120813932,
      "step": 2157
    },
    {
      "epoch": 4.806236080178174,
      "grad_norm": 19.169998168945312,
      "learning_rate": 1e-06,
      "loss": 0.7322,
      "num_input_tokens_seen": 120867148,
      "step": 2158
    },
    {
      "epoch": 4.806236080178174,
      "loss": 0.6552798748016357,
      "loss_ce": 0.0008609433425590396,
      "loss_iou": 0.275390625,
      "loss_num": 0.02099609375,
      "loss_xval": 0.65625,
      "num_input_tokens_seen": 120867148,
      "step": 2158
    },
    {
      "epoch": 4.8084632516703785,
      "grad_norm": 100.8769302368164,
      "learning_rate": 1e-06,
      "loss": 0.8376,
      "num_input_tokens_seen": 120921256,
      "step": 2159
    },
    {
      "epoch": 4.8084632516703785,
      "loss": 0.7510162591934204,
      "loss_ce": 0.0002837868523783982,
      "loss_iou": 0.330078125,
      "loss_num": 0.0184326171875,
      "loss_xval": 0.75,
      "num_input_tokens_seen": 120921256,
      "step": 2159
    },
    {
      "epoch": 4.810690423162583,
      "grad_norm": 18.4330997467041,
      "learning_rate": 1e-06,
      "loss": 0.9006,
      "num_input_tokens_seen": 120976736,
      "step": 2160
    },
    {
      "epoch": 4.810690423162583,
      "loss": 1.1066399812698364,
      "loss_ce": 0.00043876029667444527,
      "loss_iou": 0.447265625,
      "loss_num": 0.0419921875,
      "loss_xval": 1.109375,
      "num_input_tokens_seen": 120976736,
      "step": 2160
    },
    {
      "epoch": 4.812917594654788,
      "grad_norm": 19.734941482543945,
      "learning_rate": 1e-06,
      "loss": 0.814,
      "num_input_tokens_seen": 121034524,
      "step": 2161
    },
    {
      "epoch": 4.812917594654788,
      "loss": 0.8756116628646851,
      "loss_ce": 0.0006117259617894888,
      "loss_iou": 0.3671875,
      "loss_num": 0.0281982421875,
      "loss_xval": 0.875,
      "num_input_tokens_seen": 121034524,
      "step": 2161
    },
    {
      "epoch": 4.815144766146993,
      "grad_norm": 18.92926788330078,
      "learning_rate": 1e-06,
      "loss": 0.6339,
      "num_input_tokens_seen": 121090056,
      "step": 2162
    },
    {
      "epoch": 4.815144766146993,
      "loss": 0.6168254017829895,
      "loss_ce": 0.0003703091642819345,
      "loss_iou": 0.27734375,
      "loss_num": 0.01251220703125,
      "loss_xval": 0.6171875,
      "num_input_tokens_seen": 121090056,
      "step": 2162
    },
    {
      "epoch": 4.817371937639198,
      "grad_norm": 30.59477996826172,
      "learning_rate": 1e-06,
      "loss": 0.8952,
      "num_input_tokens_seen": 121148352,
      "step": 2163
    },
    {
      "epoch": 4.817371937639198,
      "loss": 0.9438307881355286,
      "loss_ce": 0.0004713647358585149,
      "loss_iou": 0.388671875,
      "loss_num": 0.03271484375,
      "loss_xval": 0.9453125,
      "num_input_tokens_seen": 121148352,
      "step": 2163
    },
    {
      "epoch": 4.819599109131403,
      "grad_norm": 21.30016326904297,
      "learning_rate": 1e-06,
      "loss": 0.9444,
      "num_input_tokens_seen": 121206584,
      "step": 2164
    },
    {
      "epoch": 4.819599109131403,
      "loss": 0.7458754777908325,
      "loss_ce": 0.000392090150853619,
      "loss_iou": 0.298828125,
      "loss_num": 0.0296630859375,
      "loss_xval": 0.74609375,
      "num_input_tokens_seen": 121206584,
      "step": 2164
    },
    {
      "epoch": 4.821826280623608,
      "grad_norm": 25.39438247680664,
      "learning_rate": 1e-06,
      "loss": 0.7381,
      "num_input_tokens_seen": 121265252,
      "step": 2165
    },
    {
      "epoch": 4.821826280623608,
      "loss": 0.7435488700866699,
      "loss_ce": 0.00026276521384716034,
      "loss_iou": 0.326171875,
      "loss_num": 0.01806640625,
      "loss_xval": 0.7421875,
      "num_input_tokens_seen": 121265252,
      "step": 2165
    },
    {
      "epoch": 4.8240534521158125,
      "grad_norm": 36.96885681152344,
      "learning_rate": 1e-06,
      "loss": 0.7863,
      "num_input_tokens_seen": 121321320,
      "step": 2166
    },
    {
      "epoch": 4.8240534521158125,
      "loss": 0.82174152135849,
      "loss_ce": 0.0002082870778394863,
      "loss_iou": 0.34375,
      "loss_num": 0.0267333984375,
      "loss_xval": 0.8203125,
      "num_input_tokens_seen": 121321320,
      "step": 2166
    },
    {
      "epoch": 4.826280623608017,
      "grad_norm": 18.965267181396484,
      "learning_rate": 1e-06,
      "loss": 0.7267,
      "num_input_tokens_seen": 121375740,
      "step": 2167
    },
    {
      "epoch": 4.826280623608017,
      "loss": 0.8186416625976562,
      "loss_ce": 0.0005264327628538013,
      "loss_iou": 0.3671875,
      "loss_num": 0.016845703125,
      "loss_xval": 0.81640625,
      "num_input_tokens_seen": 121375740,
      "step": 2167
    },
    {
      "epoch": 4.828507795100223,
      "grad_norm": 14.640192031860352,
      "learning_rate": 1e-06,
      "loss": 0.6366,
      "num_input_tokens_seen": 121433144,
      "step": 2168
    },
    {
      "epoch": 4.828507795100223,
      "loss": 0.5576098561286926,
      "loss_ce": 0.00029783969512209296,
      "loss_iou": 0.2470703125,
      "loss_num": 0.012451171875,
      "loss_xval": 0.55859375,
      "num_input_tokens_seen": 121433144,
      "step": 2168
    },
    {
      "epoch": 4.830734966592428,
      "grad_norm": 16.91661834716797,
      "learning_rate": 1e-06,
      "loss": 0.7211,
      "num_input_tokens_seen": 121491312,
      "step": 2169
    },
    {
      "epoch": 4.830734966592428,
      "loss": 0.4785498082637787,
      "loss_ce": 0.00027831370243802667,
      "loss_iou": 0.18359375,
      "loss_num": 0.0220947265625,
      "loss_xval": 0.478515625,
      "num_input_tokens_seen": 121491312,
      "step": 2169
    },
    {
      "epoch": 4.832962138084633,
      "grad_norm": 31.92616844177246,
      "learning_rate": 1e-06,
      "loss": 0.9147,
      "num_input_tokens_seen": 121546584,
      "step": 2170
    },
    {
      "epoch": 4.832962138084633,
      "loss": 0.9893652200698853,
      "loss_ce": 0.00035156396916136146,
      "loss_iou": 0.384765625,
      "loss_num": 0.044189453125,
      "loss_xval": 0.98828125,
      "num_input_tokens_seen": 121546584,
      "step": 2170
    },
    {
      "epoch": 4.835189309576838,
      "grad_norm": 18.34603500366211,
      "learning_rate": 1e-06,
      "loss": 0.7332,
      "num_input_tokens_seen": 121601848,
      "step": 2171
    },
    {
      "epoch": 4.835189309576838,
      "loss": 0.6528722047805786,
      "loss_ce": 0.00022326521866489202,
      "loss_iou": 0.2578125,
      "loss_num": 0.0277099609375,
      "loss_xval": 0.65234375,
      "num_input_tokens_seen": 121601848,
      "step": 2171
    },
    {
      "epoch": 4.8374164810690425,
      "grad_norm": 23.596956253051758,
      "learning_rate": 1e-06,
      "loss": 0.6937,
      "num_input_tokens_seen": 121654980,
      "step": 2172
    },
    {
      "epoch": 4.8374164810690425,
      "loss": 0.7717235088348389,
      "loss_ce": 0.00023919279919937253,
      "loss_iou": 0.326171875,
      "loss_num": 0.0240478515625,
      "loss_xval": 0.7734375,
      "num_input_tokens_seen": 121654980,
      "step": 2172
    },
    {
      "epoch": 4.839643652561247,
      "grad_norm": 17.30280876159668,
      "learning_rate": 1e-06,
      "loss": 0.8213,
      "num_input_tokens_seen": 121710832,
      "step": 2173
    },
    {
      "epoch": 4.839643652561247,
      "loss": 0.890198290348053,
      "loss_ce": 0.0003057056455872953,
      "loss_iou": 0.388671875,
      "loss_num": 0.022705078125,
      "loss_xval": 0.890625,
      "num_input_tokens_seen": 121710832,
      "step": 2173
    },
    {
      "epoch": 4.841870824053452,
      "grad_norm": 21.682880401611328,
      "learning_rate": 1e-06,
      "loss": 0.7604,
      "num_input_tokens_seen": 121769308,
      "step": 2174
    },
    {
      "epoch": 4.841870824053452,
      "loss": 0.8933024406433105,
      "loss_ce": 0.00023604354646522552,
      "loss_iou": 0.392578125,
      "loss_num": 0.0216064453125,
      "loss_xval": 0.89453125,
      "num_input_tokens_seen": 121769308,
      "step": 2174
    },
    {
      "epoch": 4.844097995545657,
      "grad_norm": 23.18276596069336,
      "learning_rate": 1e-06,
      "loss": 0.852,
      "num_input_tokens_seen": 121825936,
      "step": 2175
    },
    {
      "epoch": 4.844097995545657,
      "loss": 0.8222508430480957,
      "loss_ce": 0.0004735908005386591,
      "loss_iou": 0.34765625,
      "loss_num": 0.0255126953125,
      "loss_xval": 0.8203125,
      "num_input_tokens_seen": 121825936,
      "step": 2175
    },
    {
      "epoch": 4.846325167037862,
      "grad_norm": 16.410219192504883,
      "learning_rate": 1e-06,
      "loss": 0.5924,
      "num_input_tokens_seen": 121880128,
      "step": 2176
    },
    {
      "epoch": 4.846325167037862,
      "loss": 0.5815201997756958,
      "loss_ce": 0.00022139312932267785,
      "loss_iou": 0.2451171875,
      "loss_num": 0.0181884765625,
      "loss_xval": 0.58203125,
      "num_input_tokens_seen": 121880128,
      "step": 2176
    },
    {
      "epoch": 4.848552338530067,
      "grad_norm": 16.55893898010254,
      "learning_rate": 1e-06,
      "loss": 0.5514,
      "num_input_tokens_seen": 121937432,
      "step": 2177
    },
    {
      "epoch": 4.848552338530067,
      "loss": 0.5693067312240601,
      "loss_ce": 0.00021495725377462804,
      "loss_iou": 0.25,
      "loss_num": 0.013671875,
      "loss_xval": 0.5703125,
      "num_input_tokens_seen": 121937432,
      "step": 2177
    },
    {
      "epoch": 4.850779510022272,
      "grad_norm": 22.6440372467041,
      "learning_rate": 1e-06,
      "loss": 0.694,
      "num_input_tokens_seen": 121993900,
      "step": 2178
    },
    {
      "epoch": 4.850779510022272,
      "loss": 0.8454983234405518,
      "loss_ce": 0.0002834856859408319,
      "loss_iou": 0.359375,
      "loss_num": 0.0252685546875,
      "loss_xval": 0.84375,
      "num_input_tokens_seen": 121993900,
      "step": 2178
    },
    {
      "epoch": 4.853006681514477,
      "grad_norm": 47.70549392700195,
      "learning_rate": 1e-06,
      "loss": 0.8097,
      "num_input_tokens_seen": 122050492,
      "step": 2179
    },
    {
      "epoch": 4.853006681514477,
      "loss": 0.8886595964431763,
      "loss_ce": 0.00035399917396716774,
      "loss_iou": 0.38671875,
      "loss_num": 0.0228271484375,
      "loss_xval": 0.88671875,
      "num_input_tokens_seen": 122050492,
      "step": 2179
    },
    {
      "epoch": 4.855233853006681,
      "grad_norm": 15.52509593963623,
      "learning_rate": 1e-06,
      "loss": 0.6431,
      "num_input_tokens_seen": 122105120,
      "step": 2180
    },
    {
      "epoch": 4.855233853006681,
      "loss": 0.5089912414550781,
      "loss_ce": 0.00020220015721861273,
      "loss_iou": 0.2001953125,
      "loss_num": 0.0218505859375,
      "loss_xval": 0.5078125,
      "num_input_tokens_seen": 122105120,
      "step": 2180
    },
    {
      "epoch": 4.857461024498886,
      "grad_norm": 21.397817611694336,
      "learning_rate": 1e-06,
      "loss": 0.5721,
      "num_input_tokens_seen": 122162916,
      "step": 2181
    },
    {
      "epoch": 4.857461024498886,
      "loss": 0.5020558834075928,
      "loss_ce": 0.0002248543023597449,
      "loss_iou": 0.2236328125,
      "loss_num": 0.01080322265625,
      "loss_xval": 0.5,
      "num_input_tokens_seen": 122162916,
      "step": 2181
    },
    {
      "epoch": 4.859688195991091,
      "grad_norm": 18.195024490356445,
      "learning_rate": 1e-06,
      "loss": 0.6435,
      "num_input_tokens_seen": 122219996,
      "step": 2182
    },
    {
      "epoch": 4.859688195991091,
      "loss": 0.6611765027046204,
      "loss_ce": 0.0002878474479075521,
      "loss_iou": 0.283203125,
      "loss_num": 0.0191650390625,
      "loss_xval": 0.66015625,
      "num_input_tokens_seen": 122219996,
      "step": 2182
    },
    {
      "epoch": 4.861915367483296,
      "grad_norm": 72.6725082397461,
      "learning_rate": 1e-06,
      "loss": 0.7279,
      "num_input_tokens_seen": 122274148,
      "step": 2183
    },
    {
      "epoch": 4.861915367483296,
      "loss": 0.6504148244857788,
      "loss_ce": 0.0002683540806174278,
      "loss_iou": 0.30078125,
      "loss_num": 0.010009765625,
      "loss_xval": 0.6484375,
      "num_input_tokens_seen": 122274148,
      "step": 2183
    },
    {
      "epoch": 4.864142538975501,
      "grad_norm": 31.536542892456055,
      "learning_rate": 1e-06,
      "loss": 0.8554,
      "num_input_tokens_seen": 122329776,
      "step": 2184
    },
    {
      "epoch": 4.864142538975501,
      "loss": 0.7600322365760803,
      "loss_ce": 0.00026659879949875176,
      "loss_iou": 0.3125,
      "loss_num": 0.027099609375,
      "loss_xval": 0.7578125,
      "num_input_tokens_seen": 122329776,
      "step": 2184
    },
    {
      "epoch": 4.866369710467706,
      "grad_norm": 25.865697860717773,
      "learning_rate": 1e-06,
      "loss": 0.8837,
      "num_input_tokens_seen": 122387312,
      "step": 2185
    },
    {
      "epoch": 4.866369710467706,
      "loss": 0.8069164752960205,
      "loss_ce": 0.00027582579059526324,
      "loss_iou": 0.33203125,
      "loss_num": 0.02880859375,
      "loss_xval": 0.8046875,
      "num_input_tokens_seen": 122387312,
      "step": 2185
    },
    {
      "epoch": 4.868596881959911,
      "grad_norm": 20.460817337036133,
      "learning_rate": 1e-06,
      "loss": 0.7866,
      "num_input_tokens_seen": 122444908,
      "step": 2186
    },
    {
      "epoch": 4.868596881959911,
      "loss": 0.9665539264678955,
      "loss_ce": 0.00036741438088938594,
      "loss_iou": 0.396484375,
      "loss_num": 0.03466796875,
      "loss_xval": 0.96484375,
      "num_input_tokens_seen": 122444908,
      "step": 2186
    },
    {
      "epoch": 4.870824053452115,
      "grad_norm": 29.2179012298584,
      "learning_rate": 1e-06,
      "loss": 0.6808,
      "num_input_tokens_seen": 122499836,
      "step": 2187
    },
    {
      "epoch": 4.870824053452115,
      "loss": 0.8551727533340454,
      "loss_ce": 0.00043644450488500297,
      "loss_iou": 0.337890625,
      "loss_num": 0.035400390625,
      "loss_xval": 0.85546875,
      "num_input_tokens_seen": 122499836,
      "step": 2187
    },
    {
      "epoch": 4.873051224944321,
      "grad_norm": 26.670900344848633,
      "learning_rate": 1e-06,
      "loss": 0.7637,
      "num_input_tokens_seen": 122556144,
      "step": 2188
    },
    {
      "epoch": 4.873051224944321,
      "loss": 0.7377973794937134,
      "loss_ce": 0.00024859551922418177,
      "loss_iou": 0.33203125,
      "loss_num": 0.01507568359375,
      "loss_xval": 0.73828125,
      "num_input_tokens_seen": 122556144,
      "step": 2188
    },
    {
      "epoch": 4.875278396436526,
      "grad_norm": 24.723434448242188,
      "learning_rate": 1e-06,
      "loss": 0.72,
      "num_input_tokens_seen": 122612748,
      "step": 2189
    },
    {
      "epoch": 4.875278396436526,
      "loss": 0.6694687008857727,
      "loss_ce": 0.0002792471495922655,
      "loss_iou": 0.298828125,
      "loss_num": 0.01422119140625,
      "loss_xval": 0.66796875,
      "num_input_tokens_seen": 122612748,
      "step": 2189
    },
    {
      "epoch": 4.877505567928731,
      "grad_norm": 15.66889476776123,
      "learning_rate": 1e-06,
      "loss": 0.6759,
      "num_input_tokens_seen": 122668312,
      "step": 2190
    },
    {
      "epoch": 4.877505567928731,
      "loss": 0.7348698973655701,
      "loss_ce": 0.00025075351004488766,
      "loss_iou": 0.326171875,
      "loss_num": 0.0166015625,
      "loss_xval": 0.734375,
      "num_input_tokens_seen": 122668312,
      "step": 2190
    },
    {
      "epoch": 4.879732739420936,
      "grad_norm": 25.57622718811035,
      "learning_rate": 1e-06,
      "loss": 0.8345,
      "num_input_tokens_seen": 122721432,
      "step": 2191
    },
    {
      "epoch": 4.879732739420936,
      "loss": 0.7397469282150269,
      "loss_ce": 0.00024497421691194177,
      "loss_iou": 0.31640625,
      "loss_num": 0.021240234375,
      "loss_xval": 0.73828125,
      "num_input_tokens_seen": 122721432,
      "step": 2191
    },
    {
      "epoch": 4.881959910913141,
      "grad_norm": 24.592416763305664,
      "learning_rate": 1e-06,
      "loss": 0.6403,
      "num_input_tokens_seen": 122777480,
      "step": 2192
    },
    {
      "epoch": 4.881959910913141,
      "loss": 0.6197606325149536,
      "loss_ce": 0.00037585641257464886,
      "loss_iou": 0.265625,
      "loss_num": 0.0177001953125,
      "loss_xval": 0.62109375,
      "num_input_tokens_seen": 122777480,
      "step": 2192
    },
    {
      "epoch": 4.8841870824053455,
      "grad_norm": 17.536779403686523,
      "learning_rate": 1e-06,
      "loss": 0.6577,
      "num_input_tokens_seen": 122835032,
      "step": 2193
    },
    {
      "epoch": 4.8841870824053455,
      "loss": 0.7718772888183594,
      "loss_ce": 0.0002708572428673506,
      "loss_iou": 0.337890625,
      "loss_num": 0.0185546875,
      "loss_xval": 0.7734375,
      "num_input_tokens_seen": 122835032,
      "step": 2193
    },
    {
      "epoch": 4.88641425389755,
      "grad_norm": 32.589019775390625,
      "learning_rate": 1e-06,
      "loss": 0.9171,
      "num_input_tokens_seen": 122888828,
      "step": 2194
    },
    {
      "epoch": 4.88641425389755,
      "loss": 1.2385810613632202,
      "loss_ce": 0.00029984532739035785,
      "loss_iou": 0.5234375,
      "loss_num": 0.03759765625,
      "loss_xval": 1.234375,
      "num_input_tokens_seen": 122888828,
      "step": 2194
    },
    {
      "epoch": 4.888641425389755,
      "grad_norm": 22.544490814208984,
      "learning_rate": 1e-06,
      "loss": 0.7052,
      "num_input_tokens_seen": 122946980,
      "step": 2195
    },
    {
      "epoch": 4.888641425389755,
      "loss": 0.7754490971565247,
      "loss_ce": 0.00030256161699071527,
      "loss_iou": 0.314453125,
      "loss_num": 0.0296630859375,
      "loss_xval": 0.7734375,
      "num_input_tokens_seen": 122946980,
      "step": 2195
    },
    {
      "epoch": 4.89086859688196,
      "grad_norm": 16.414432525634766,
      "learning_rate": 1e-06,
      "loss": 0.6703,
      "num_input_tokens_seen": 123006736,
      "step": 2196
    },
    {
      "epoch": 4.89086859688196,
      "loss": 0.6196362972259521,
      "loss_ce": 0.00025152770103886724,
      "loss_iou": 0.263671875,
      "loss_num": 0.0189208984375,
      "loss_xval": 0.62109375,
      "num_input_tokens_seen": 123006736,
      "step": 2196
    },
    {
      "epoch": 4.893095768374165,
      "grad_norm": 21.864694595336914,
      "learning_rate": 1e-06,
      "loss": 0.8546,
      "num_input_tokens_seen": 123065160,
      "step": 2197
    },
    {
      "epoch": 4.893095768374165,
      "loss": 0.8675484657287598,
      "loss_ce": 0.00036098493728786707,
      "loss_iou": 0.365234375,
      "loss_num": 0.0274658203125,
      "loss_xval": 0.8671875,
      "num_input_tokens_seen": 123065160,
      "step": 2197
    },
    {
      "epoch": 4.89532293986637,
      "grad_norm": 31.04578971862793,
      "learning_rate": 1e-06,
      "loss": 0.735,
      "num_input_tokens_seen": 123117600,
      "step": 2198
    },
    {
      "epoch": 4.89532293986637,
      "loss": 0.6225305795669556,
      "loss_ce": 0.00021616063895635307,
      "loss_iou": 0.275390625,
      "loss_num": 0.01409912109375,
      "loss_xval": 0.62109375,
      "num_input_tokens_seen": 123117600,
      "step": 2198
    },
    {
      "epoch": 4.897550111358575,
      "grad_norm": 13.26624870300293,
      "learning_rate": 1e-06,
      "loss": 0.6433,
      "num_input_tokens_seen": 123171304,
      "step": 2199
    },
    {
      "epoch": 4.897550111358575,
      "loss": 0.6789191961288452,
      "loss_ce": 0.00023874480393715203,
      "loss_iou": 0.259765625,
      "loss_num": 0.03173828125,
      "loss_xval": 0.6796875,
      "num_input_tokens_seen": 123171304,
      "step": 2199
    },
    {
      "epoch": 4.8997772828507795,
      "grad_norm": 19.680660247802734,
      "learning_rate": 1e-06,
      "loss": 0.8179,
      "num_input_tokens_seen": 123227968,
      "step": 2200
    },
    {
      "epoch": 4.8997772828507795,
      "loss": 1.005420446395874,
      "loss_ce": 0.0002934019430540502,
      "loss_iou": 0.421875,
      "loss_num": 0.0322265625,
      "loss_xval": 1.0078125,
      "num_input_tokens_seen": 123227968,
      "step": 2200
    },
    {
      "epoch": 4.902004454342984,
      "grad_norm": 14.69334888458252,
      "learning_rate": 1e-06,
      "loss": 0.6017,
      "num_input_tokens_seen": 123286024,
      "step": 2201
    },
    {
      "epoch": 4.902004454342984,
      "loss": 0.7753942012786865,
      "loss_ce": 0.0002477114903740585,
      "loss_iou": 0.345703125,
      "loss_num": 0.0164794921875,
      "loss_xval": 0.7734375,
      "num_input_tokens_seen": 123286024,
      "step": 2201
    },
    {
      "epoch": 4.904231625835189,
      "grad_norm": 23.2365665435791,
      "learning_rate": 1e-06,
      "loss": 0.6886,
      "num_input_tokens_seen": 123337296,
      "step": 2202
    },
    {
      "epoch": 4.904231625835189,
      "loss": 0.6639537811279297,
      "loss_ce": 0.00037955871084704995,
      "loss_iou": 0.287109375,
      "loss_num": 0.017578125,
      "loss_xval": 0.6640625,
      "num_input_tokens_seen": 123337296,
      "step": 2202
    },
    {
      "epoch": 4.906458797327394,
      "grad_norm": 21.47087287902832,
      "learning_rate": 1e-06,
      "loss": 0.7853,
      "num_input_tokens_seen": 123391896,
      "step": 2203
    },
    {
      "epoch": 4.906458797327394,
      "loss": 0.8630619049072266,
      "loss_ce": 0.0002689045504666865,
      "loss_iou": 0.34375,
      "loss_num": 0.03466796875,
      "loss_xval": 0.86328125,
      "num_input_tokens_seen": 123391896,
      "step": 2203
    },
    {
      "epoch": 4.908685968819599,
      "grad_norm": 15.741186141967773,
      "learning_rate": 1e-06,
      "loss": 0.644,
      "num_input_tokens_seen": 123448656,
      "step": 2204
    },
    {
      "epoch": 4.908685968819599,
      "loss": 0.6564854383468628,
      "loss_ce": 0.00023539297399111092,
      "loss_iou": 0.2734375,
      "loss_num": 0.021728515625,
      "loss_xval": 0.65625,
      "num_input_tokens_seen": 123448656,
      "step": 2204
    },
    {
      "epoch": 4.910913140311804,
      "grad_norm": 21.5997257232666,
      "learning_rate": 1e-06,
      "loss": 0.823,
      "num_input_tokens_seen": 123503840,
      "step": 2205
    },
    {
      "epoch": 4.910913140311804,
      "loss": 0.8479044437408447,
      "loss_ce": 0.00024817389203235507,
      "loss_iou": 0.35546875,
      "loss_num": 0.0269775390625,
      "loss_xval": 0.84765625,
      "num_input_tokens_seen": 123503840,
      "step": 2205
    },
    {
      "epoch": 4.913140311804009,
      "grad_norm": 16.790218353271484,
      "learning_rate": 1e-06,
      "loss": 0.5964,
      "num_input_tokens_seen": 123558308,
      "step": 2206
    },
    {
      "epoch": 4.913140311804009,
      "loss": 0.6172899007797241,
      "loss_ce": 0.00022445424110628664,
      "loss_iou": 0.275390625,
      "loss_num": 0.01287841796875,
      "loss_xval": 0.6171875,
      "num_input_tokens_seen": 123558308,
      "step": 2206
    },
    {
      "epoch": 4.9153674832962135,
      "grad_norm": 66.33501434326172,
      "learning_rate": 1e-06,
      "loss": 0.8602,
      "num_input_tokens_seen": 123615124,
      "step": 2207
    },
    {
      "epoch": 4.9153674832962135,
      "loss": 0.8113021850585938,
      "loss_ce": 0.00026704196352511644,
      "loss_iou": 0.359375,
      "loss_num": 0.01806640625,
      "loss_xval": 0.8125,
      "num_input_tokens_seen": 123615124,
      "step": 2207
    },
    {
      "epoch": 4.917594654788418,
      "grad_norm": 16.30253028869629,
      "learning_rate": 1e-06,
      "loss": 0.4585,
      "num_input_tokens_seen": 123671784,
      "step": 2208
    },
    {
      "epoch": 4.917594654788418,
      "loss": 0.4267365336418152,
      "loss_ce": 0.00022286101011559367,
      "loss_iou": 0.181640625,
      "loss_num": 0.01263427734375,
      "loss_xval": 0.42578125,
      "num_input_tokens_seen": 123671784,
      "step": 2208
    },
    {
      "epoch": 4.919821826280623,
      "grad_norm": 16.062482833862305,
      "learning_rate": 1e-06,
      "loss": 0.7222,
      "num_input_tokens_seen": 123729044,
      "step": 2209
    },
    {
      "epoch": 4.919821826280623,
      "loss": 0.7194496393203735,
      "loss_ce": 0.0004554773331619799,
      "loss_iou": 0.298828125,
      "loss_num": 0.02392578125,
      "loss_xval": 0.71875,
      "num_input_tokens_seen": 123729044,
      "step": 2209
    },
    {
      "epoch": 4.922048997772828,
      "grad_norm": 19.926166534423828,
      "learning_rate": 1e-06,
      "loss": 0.9957,
      "num_input_tokens_seen": 123784456,
      "step": 2210
    },
    {
      "epoch": 4.922048997772828,
      "loss": 0.9393208026885986,
      "loss_ce": 0.0003559546312317252,
      "loss_iou": 0.376953125,
      "loss_num": 0.037109375,
      "loss_xval": 0.9375,
      "num_input_tokens_seen": 123784456,
      "step": 2210
    },
    {
      "epoch": 4.924276169265033,
      "grad_norm": 26.751087188720703,
      "learning_rate": 1e-06,
      "loss": 0.8433,
      "num_input_tokens_seen": 123839024,
      "step": 2211
    },
    {
      "epoch": 4.924276169265033,
      "loss": 0.7800534963607788,
      "loss_ce": 0.0002683330385480076,
      "loss_iou": 0.330078125,
      "loss_num": 0.023681640625,
      "loss_xval": 0.78125,
      "num_input_tokens_seen": 123839024,
      "step": 2211
    },
    {
      "epoch": 4.926503340757238,
      "grad_norm": 20.585493087768555,
      "learning_rate": 1e-06,
      "loss": 0.6179,
      "num_input_tokens_seen": 123895700,
      "step": 2212
    },
    {
      "epoch": 4.926503340757238,
      "loss": 0.6982933282852173,
      "loss_ce": 0.000295252597425133,
      "loss_iou": 0.31640625,
      "loss_num": 0.01312255859375,
      "loss_xval": 0.69921875,
      "num_input_tokens_seen": 123895700,
      "step": 2212
    },
    {
      "epoch": 4.928730512249444,
      "grad_norm": 16.665607452392578,
      "learning_rate": 1e-06,
      "loss": 0.7618,
      "num_input_tokens_seen": 123950376,
      "step": 2213
    },
    {
      "epoch": 4.928730512249444,
      "loss": 0.5816483497619629,
      "loss_ce": 0.0002274463913636282,
      "loss_iou": 0.2490234375,
      "loss_num": 0.0167236328125,
      "loss_xval": 0.58203125,
      "num_input_tokens_seen": 123950376,
      "step": 2213
    },
    {
      "epoch": 4.9309576837416484,
      "grad_norm": 21.802547454833984,
      "learning_rate": 1e-06,
      "loss": 0.6218,
      "num_input_tokens_seen": 124006408,
      "step": 2214
    },
    {
      "epoch": 4.9309576837416484,
      "loss": 0.6234120726585388,
      "loss_ce": 0.00024313261383213103,
      "loss_iou": 0.259765625,
      "loss_num": 0.0206298828125,
      "loss_xval": 0.625,
      "num_input_tokens_seen": 124006408,
      "step": 2214
    },
    {
      "epoch": 4.933184855233853,
      "grad_norm": 50.704349517822266,
      "learning_rate": 1e-06,
      "loss": 0.7325,
      "num_input_tokens_seen": 124064376,
      "step": 2215
    },
    {
      "epoch": 4.933184855233853,
      "loss": 0.8308193683624268,
      "loss_ce": 0.0004971576854586601,
      "loss_iou": 0.337890625,
      "loss_num": 0.0311279296875,
      "loss_xval": 0.83203125,
      "num_input_tokens_seen": 124064376,
      "step": 2215
    },
    {
      "epoch": 4.935412026726058,
      "grad_norm": 18.29207992553711,
      "learning_rate": 1e-06,
      "loss": 0.8372,
      "num_input_tokens_seen": 124120104,
      "step": 2216
    },
    {
      "epoch": 4.935412026726058,
      "loss": 0.6730678081512451,
      "loss_ce": 0.0002162476594094187,
      "loss_iou": 0.283203125,
      "loss_num": 0.0211181640625,
      "loss_xval": 0.671875,
      "num_input_tokens_seen": 124120104,
      "step": 2216
    },
    {
      "epoch": 4.937639198218263,
      "grad_norm": 18.550386428833008,
      "learning_rate": 1e-06,
      "loss": 0.7834,
      "num_input_tokens_seen": 124175912,
      "step": 2217
    },
    {
      "epoch": 4.937639198218263,
      "loss": 0.635769784450531,
      "loss_ce": 0.00027175506693311036,
      "loss_iou": 0.27734375,
      "loss_num": 0.0164794921875,
      "loss_xval": 0.63671875,
      "num_input_tokens_seen": 124175912,
      "step": 2217
    },
    {
      "epoch": 4.939866369710468,
      "grad_norm": 20.139272689819336,
      "learning_rate": 1e-06,
      "loss": 0.9253,
      "num_input_tokens_seen": 124231184,
      "step": 2218
    },
    {
      "epoch": 4.939866369710468,
      "loss": 0.9445996284484863,
      "loss_ce": 0.00038581539411097765,
      "loss_iou": 0.365234375,
      "loss_num": 0.04248046875,
      "loss_xval": 0.9453125,
      "num_input_tokens_seen": 124231184,
      "step": 2218
    },
    {
      "epoch": 4.942093541202673,
      "grad_norm": 23.283292770385742,
      "learning_rate": 1e-06,
      "loss": 0.8611,
      "num_input_tokens_seen": 124288768,
      "step": 2219
    },
    {
      "epoch": 4.942093541202673,
      "loss": 0.8784899711608887,
      "loss_ce": 0.0003162057837471366,
      "loss_iou": 0.36328125,
      "loss_num": 0.03076171875,
      "loss_xval": 0.87890625,
      "num_input_tokens_seen": 124288768,
      "step": 2219
    },
    {
      "epoch": 4.944320712694878,
      "grad_norm": 15.636147499084473,
      "learning_rate": 1e-06,
      "loss": 0.5499,
      "num_input_tokens_seen": 124344436,
      "step": 2220
    },
    {
      "epoch": 4.944320712694878,
      "loss": 0.592854380607605,
      "loss_ce": 0.0003251142334192991,
      "loss_iou": 0.267578125,
      "loss_num": 0.01129150390625,
      "loss_xval": 0.59375,
      "num_input_tokens_seen": 124344436,
      "step": 2220
    },
    {
      "epoch": 4.9465478841870825,
      "grad_norm": 14.876840591430664,
      "learning_rate": 1e-06,
      "loss": 0.6573,
      "num_input_tokens_seen": 124400952,
      "step": 2221
    },
    {
      "epoch": 4.9465478841870825,
      "loss": 0.7517122626304626,
      "loss_ce": 0.0002474116045050323,
      "loss_iou": 0.32421875,
      "loss_num": 0.0205078125,
      "loss_xval": 0.75,
      "num_input_tokens_seen": 124400952,
      "step": 2221
    },
    {
      "epoch": 4.948775055679287,
      "grad_norm": 28.595050811767578,
      "learning_rate": 1e-06,
      "loss": 0.901,
      "num_input_tokens_seen": 124457384,
      "step": 2222
    },
    {
      "epoch": 4.948775055679287,
      "loss": 0.9143351316452026,
      "loss_ce": 0.00027260559727437794,
      "loss_iou": 0.404296875,
      "loss_num": 0.0213623046875,
      "loss_xval": 0.9140625,
      "num_input_tokens_seen": 124457384,
      "step": 2222
    },
    {
      "epoch": 4.951002227171492,
      "grad_norm": 24.853782653808594,
      "learning_rate": 1e-06,
      "loss": 0.7986,
      "num_input_tokens_seen": 124514252,
      "step": 2223
    },
    {
      "epoch": 4.951002227171492,
      "loss": 0.6873869895935059,
      "loss_ce": 0.00043627433478832245,
      "loss_iou": 0.302734375,
      "loss_num": 0.0164794921875,
      "loss_xval": 0.6875,
      "num_input_tokens_seen": 124514252,
      "step": 2223
    },
    {
      "epoch": 4.953229398663697,
      "grad_norm": 22.14551544189453,
      "learning_rate": 1e-06,
      "loss": 0.691,
      "num_input_tokens_seen": 124569752,
      "step": 2224
    },
    {
      "epoch": 4.953229398663697,
      "loss": 0.6962725520133972,
      "loss_ce": 0.00022760960564482957,
      "loss_iou": 0.294921875,
      "loss_num": 0.021240234375,
      "loss_xval": 0.6953125,
      "num_input_tokens_seen": 124569752,
      "step": 2224
    },
    {
      "epoch": 4.955456570155902,
      "grad_norm": 20.33577537536621,
      "learning_rate": 1e-06,
      "loss": 0.6513,
      "num_input_tokens_seen": 124626644,
      "step": 2225
    },
    {
      "epoch": 4.955456570155902,
      "loss": 0.7576655149459839,
      "loss_ce": 0.0003413379890844226,
      "loss_iou": 0.302734375,
      "loss_num": 0.0306396484375,
      "loss_xval": 0.7578125,
      "num_input_tokens_seen": 124626644,
      "step": 2225
    },
    {
      "epoch": 4.957683741648107,
      "grad_norm": 19.96550178527832,
      "learning_rate": 1e-06,
      "loss": 0.4482,
      "num_input_tokens_seen": 124684520,
      "step": 2226
    },
    {
      "epoch": 4.957683741648107,
      "loss": 0.46445581316947937,
      "loss_ce": 0.0002224127674708143,
      "loss_iou": 0.201171875,
      "loss_num": 0.01226806640625,
      "loss_xval": 0.46484375,
      "num_input_tokens_seen": 124684520,
      "step": 2226
    },
    {
      "epoch": 4.959910913140312,
      "grad_norm": 16.052396774291992,
      "learning_rate": 1e-06,
      "loss": 0.6533,
      "num_input_tokens_seen": 124739472,
      "step": 2227
    },
    {
      "epoch": 4.959910913140312,
      "loss": 0.8050886988639832,
      "loss_ce": 0.0004011690034531057,
      "loss_iou": 0.33203125,
      "loss_num": 0.0277099609375,
      "loss_xval": 0.8046875,
      "num_input_tokens_seen": 124739472,
      "step": 2227
    },
    {
      "epoch": 4.9621380846325165,
      "grad_norm": 19.418689727783203,
      "learning_rate": 1e-06,
      "loss": 0.7515,
      "num_input_tokens_seen": 124795168,
      "step": 2228
    },
    {
      "epoch": 4.9621380846325165,
      "loss": 0.6984738707542419,
      "loss_ce": 0.00023169181076809764,
      "loss_iou": 0.287109375,
      "loss_num": 0.02490234375,
      "loss_xval": 0.69921875,
      "num_input_tokens_seen": 124795168,
      "step": 2228
    },
    {
      "epoch": 4.964365256124721,
      "grad_norm": 19.798582077026367,
      "learning_rate": 1e-06,
      "loss": 0.8594,
      "num_input_tokens_seen": 124850384,
      "step": 2229
    },
    {
      "epoch": 4.964365256124721,
      "loss": 0.8433953523635864,
      "loss_ce": 0.0002557524712756276,
      "loss_iou": 0.349609375,
      "loss_num": 0.0291748046875,
      "loss_xval": 0.84375,
      "num_input_tokens_seen": 124850384,
      "step": 2229
    },
    {
      "epoch": 4.966592427616926,
      "grad_norm": 28.708772659301758,
      "learning_rate": 1e-06,
      "loss": 0.8574,
      "num_input_tokens_seen": 124905988,
      "step": 2230
    },
    {
      "epoch": 4.966592427616926,
      "loss": 0.7451651692390442,
      "loss_ce": 0.00029212021036073565,
      "loss_iou": 0.306640625,
      "loss_num": 0.0267333984375,
      "loss_xval": 0.74609375,
      "num_input_tokens_seen": 124905988,
      "step": 2230
    },
    {
      "epoch": 4.968819599109131,
      "grad_norm": 24.822715759277344,
      "learning_rate": 1e-06,
      "loss": 0.7711,
      "num_input_tokens_seen": 124961660,
      "step": 2231
    },
    {
      "epoch": 4.968819599109131,
      "loss": 0.6456351280212402,
      "loss_ce": 0.00018830940825864673,
      "loss_iou": 0.275390625,
      "loss_num": 0.018798828125,
      "loss_xval": 0.64453125,
      "num_input_tokens_seen": 124961660,
      "step": 2231
    },
    {
      "epoch": 4.971046770601336,
      "grad_norm": 18.94268035888672,
      "learning_rate": 1e-06,
      "loss": 0.9299,
      "num_input_tokens_seen": 125015652,
      "step": 2232
    },
    {
      "epoch": 4.971046770601336,
      "loss": 0.9597747325897217,
      "loss_ce": 0.00030205969233065844,
      "loss_iou": 0.400390625,
      "loss_num": 0.031494140625,
      "loss_xval": 0.9609375,
      "num_input_tokens_seen": 125015652,
      "step": 2232
    },
    {
      "epoch": 4.973273942093542,
      "grad_norm": 17.5726318359375,
      "learning_rate": 1e-06,
      "loss": 0.7039,
      "num_input_tokens_seen": 125073992,
      "step": 2233
    },
    {
      "epoch": 4.973273942093542,
      "loss": 0.5961167216300964,
      "loss_ce": 0.00023052276810631156,
      "loss_iou": 0.26953125,
      "loss_num": 0.0115966796875,
      "loss_xval": 0.59765625,
      "num_input_tokens_seen": 125073992,
      "step": 2233
    },
    {
      "epoch": 4.9755011135857465,
      "grad_norm": 33.96272277832031,
      "learning_rate": 1e-06,
      "loss": 0.8093,
      "num_input_tokens_seen": 125130660,
      "step": 2234
    },
    {
      "epoch": 4.9755011135857465,
      "loss": 0.8671445846557617,
      "loss_ce": 0.00020118044631090015,
      "loss_iou": 0.345703125,
      "loss_num": 0.034423828125,
      "loss_xval": 0.8671875,
      "num_input_tokens_seen": 125130660,
      "step": 2234
    },
    {
      "epoch": 4.977728285077951,
      "grad_norm": 25.479049682617188,
      "learning_rate": 1e-06,
      "loss": 0.7355,
      "num_input_tokens_seen": 125188292,
      "step": 2235
    },
    {
      "epoch": 4.977728285077951,
      "loss": 0.6173162460327148,
      "loss_ce": 0.0002507962053641677,
      "loss_iou": 0.26953125,
      "loss_num": 0.0157470703125,
      "loss_xval": 0.6171875,
      "num_input_tokens_seen": 125188292,
      "step": 2235
    },
    {
      "epoch": 4.979955456570156,
      "grad_norm": 17.59937286376953,
      "learning_rate": 1e-06,
      "loss": 0.5309,
      "num_input_tokens_seen": 125245472,
      "step": 2236
    },
    {
      "epoch": 4.979955456570156,
      "loss": 0.4158274531364441,
      "loss_ce": 0.0002390348818153143,
      "loss_iou": 0.17578125,
      "loss_num": 0.012939453125,
      "loss_xval": 0.416015625,
      "num_input_tokens_seen": 125245472,
      "step": 2236
    },
    {
      "epoch": 4.982182628062361,
      "grad_norm": 24.452638626098633,
      "learning_rate": 1e-06,
      "loss": 0.7179,
      "num_input_tokens_seen": 125302944,
      "step": 2237
    },
    {
      "epoch": 4.982182628062361,
      "loss": 0.5437171459197998,
      "loss_ce": 0.0002601124288048595,
      "loss_iou": 0.251953125,
      "loss_num": 0.007659912109375,
      "loss_xval": 0.54296875,
      "num_input_tokens_seen": 125302944,
      "step": 2237
    },
    {
      "epoch": 4.984409799554566,
      "grad_norm": 15.169961929321289,
      "learning_rate": 1e-06,
      "loss": 0.8227,
      "num_input_tokens_seen": 125362732,
      "step": 2238
    },
    {
      "epoch": 4.984409799554566,
      "loss": 0.7834569811820984,
      "loss_ce": 0.00025388289941474795,
      "loss_iou": 0.34375,
      "loss_num": 0.01953125,
      "loss_xval": 0.78125,
      "num_input_tokens_seen": 125362732,
      "step": 2238
    },
    {
      "epoch": 4.986636971046771,
      "grad_norm": 15.975359916687012,
      "learning_rate": 1e-06,
      "loss": 0.7685,
      "num_input_tokens_seen": 125419796,
      "step": 2239
    },
    {
      "epoch": 4.986636971046771,
      "loss": 0.6306621432304382,
      "loss_ce": 0.0002910442417487502,
      "loss_iou": 0.265625,
      "loss_num": 0.019287109375,
      "loss_xval": 0.62890625,
      "num_input_tokens_seen": 125419796,
      "step": 2239
    },
    {
      "epoch": 4.988864142538976,
      "grad_norm": 41.811767578125,
      "learning_rate": 1e-06,
      "loss": 0.5101,
      "num_input_tokens_seen": 125478088,
      "step": 2240
    },
    {
      "epoch": 4.988864142538976,
      "loss": 0.5742073059082031,
      "loss_ce": 0.0002326710382476449,
      "loss_iou": 0.24609375,
      "loss_num": 0.016357421875,
      "loss_xval": 0.57421875,
      "num_input_tokens_seen": 125478088,
      "step": 2240
    },
    {
      "epoch": 4.991091314031181,
      "grad_norm": 21.81138801574707,
      "learning_rate": 1e-06,
      "loss": 0.8395,
      "num_input_tokens_seen": 125533344,
      "step": 2241
    },
    {
      "epoch": 4.991091314031181,
      "loss": 0.7543541789054871,
      "loss_ce": 0.001912805950269103,
      "loss_iou": 0.287109375,
      "loss_num": 0.03564453125,
      "loss_xval": 0.75390625,
      "num_input_tokens_seen": 125533344,
      "step": 2241
    },
    {
      "epoch": 4.993318485523385,
      "grad_norm": 30.80375099182129,
      "learning_rate": 1e-06,
      "loss": 0.9426,
      "num_input_tokens_seen": 125588860,
      "step": 2242
    },
    {
      "epoch": 4.993318485523385,
      "loss": 0.8981271386146545,
      "loss_ce": 0.0005441233515739441,
      "loss_iou": 0.384765625,
      "loss_num": 0.0260009765625,
      "loss_xval": 0.8984375,
      "num_input_tokens_seen": 125588860,
      "step": 2242
    },
    {
      "epoch": 4.99554565701559,
      "grad_norm": 18.436973571777344,
      "learning_rate": 1e-06,
      "loss": 0.6176,
      "num_input_tokens_seen": 125646052,
      "step": 2243
    },
    {
      "epoch": 4.99554565701559,
      "loss": 0.5616306066513062,
      "loss_ce": 0.0002292500576004386,
      "loss_iou": 0.236328125,
      "loss_num": 0.017822265625,
      "loss_xval": 0.5625,
      "num_input_tokens_seen": 125646052,
      "step": 2243
    },
    {
      "epoch": 4.997772828507795,
      "grad_norm": 20.010271072387695,
      "learning_rate": 1e-06,
      "loss": 0.9739,
      "num_input_tokens_seen": 125701944,
      "step": 2244
    },
    {
      "epoch": 4.997772828507795,
      "loss": 0.7831923961639404,
      "loss_ce": 0.0002333829615963623,
      "loss_iou": 0.34765625,
      "loss_num": 0.017333984375,
      "loss_xval": 0.78125,
      "num_input_tokens_seen": 125701944,
      "step": 2244
    },
    {
      "epoch": 5.0,
      "grad_norm": 22.875764846801758,
      "learning_rate": 1e-06,
      "loss": 0.8608,
      "num_input_tokens_seen": 125761180,
      "step": 2245
    },
    {
      "epoch": 5.0,
      "loss": 0.8516440391540527,
      "loss_ce": 0.0003256534691900015,
      "loss_iou": 0.349609375,
      "loss_num": 0.03076171875,
      "loss_xval": 0.8515625,
      "num_input_tokens_seen": 125761180,
      "step": 2245
    },
    {
      "epoch": 5.002227171492205,
      "grad_norm": 32.292686462402344,
      "learning_rate": 1e-06,
      "loss": 0.7312,
      "num_input_tokens_seen": 125816488,
      "step": 2246
    },
    {
      "epoch": 5.002227171492205,
      "loss": 0.7869899868965149,
      "loss_ce": 0.00036892099888063967,
      "loss_iou": 0.341796875,
      "loss_num": 0.0205078125,
      "loss_xval": 0.78515625,
      "num_input_tokens_seen": 125816488,
      "step": 2246
    },
    {
      "epoch": 5.00445434298441,
      "grad_norm": 22.563552856445312,
      "learning_rate": 1e-06,
      "loss": 0.6971,
      "num_input_tokens_seen": 125871976,
      "step": 2247
    },
    {
      "epoch": 5.00445434298441,
      "loss": 0.7504916191101074,
      "loss_ce": 0.0002475018845871091,
      "loss_iou": 0.333984375,
      "loss_num": 0.016357421875,
      "loss_xval": 0.75,
      "num_input_tokens_seen": 125871976,
      "step": 2247
    },
    {
      "epoch": 5.006681514476615,
      "grad_norm": 14.274714469909668,
      "learning_rate": 1e-06,
      "loss": 0.7045,
      "num_input_tokens_seen": 125930388,
      "step": 2248
    },
    {
      "epoch": 5.006681514476615,
      "loss": 0.598191499710083,
      "loss_ce": 0.000291119038593024,
      "loss_iou": 0.26953125,
      "loss_num": 0.01202392578125,
      "loss_xval": 0.59765625,
      "num_input_tokens_seen": 125930388,
      "step": 2248
    },
    {
      "epoch": 5.008908685968819,
      "grad_norm": 16.532169342041016,
      "learning_rate": 1e-06,
      "loss": 0.7869,
      "num_input_tokens_seen": 125986868,
      "step": 2249
    },
    {
      "epoch": 5.008908685968819,
      "loss": 0.7695136666297913,
      "loss_ce": 0.0002265319344587624,
      "loss_iou": 0.337890625,
      "loss_num": 0.01904296875,
      "loss_xval": 0.76953125,
      "num_input_tokens_seen": 125986868,
      "step": 2249
    },
    {
      "epoch": 5.011135857461024,
      "grad_norm": 14.793770790100098,
      "learning_rate": 1e-06,
      "loss": 0.901,
      "num_input_tokens_seen": 126043356,
      "step": 2250
    },
    {
      "epoch": 5.011135857461024,
      "eval_seeclick_web_CIoU": 0.5627106428146362,
      "eval_seeclick_web_GIoU": 0.5593923330307007,
      "eval_seeclick_web_IoU": 0.5781354308128357,
      "eval_seeclick_web_MAE_all": 0.01731129875406623,
      "eval_seeclick_web_MAE_h": 0.009559806901961565,
      "eval_seeclick_web_MAE_w": 0.01916863350197673,
      "eval_seeclick_web_MAE_x_boxes": 0.00826259353198111,
      "eval_seeclick_web_MAE_y_boxes": 0.021791240433230996,
      "eval_seeclick_web_inside_bbox": 0.9166666567325592,
      "eval_seeclick_web_loss": 0.9538028240203857,
      "eval_seeclick_web_loss_ce": 0.00035019248025491834,
      "eval_seeclick_web_loss_iou": 0.4365234375,
      "eval_seeclick_web_loss_num": 0.013622283935546875,
      "eval_seeclick_web_loss_xval": 0.940673828125,
      "eval_seeclick_web_runtime": 29.6362,
      "eval_seeclick_web_samples_per_second": 1.687,
      "eval_seeclick_web_steps_per_second": 0.067,
      "num_input_tokens_seen": 126043356,
      "step": 2250
    },
    {
      "epoch": 5.011135857461024,
      "eval_icons_CIoU": 0.3190227448940277,
      "eval_icons_GIoU": 0.3375135064125061,
      "eval_icons_IoU": 0.38641299307346344,
      "eval_icons_MAE_all": 0.06205383501946926,
      "eval_icons_MAE_h": 0.03881475329399109,
      "eval_icons_MAE_w": 0.06065435893833637,
      "eval_icons_MAE_x_boxes": 0.06400249153375626,
      "eval_icons_MAE_y_boxes": 0.037100257351994514,
      "eval_icons_inside_bbox": 0.6493055522441864,
      "eval_icons_loss": 1.6723029613494873,
      "eval_icons_loss_ce": 0.0004043810913572088,
      "eval_icons_loss_iou": 0.663818359375,
      "eval_icons_loss_num": 0.05780982971191406,
      "eval_icons_loss_xval": 1.615234375,
      "eval_icons_runtime": 27.5554,
      "eval_icons_samples_per_second": 1.815,
      "eval_icons_steps_per_second": 0.073,
      "num_input_tokens_seen": 126043356,
      "step": 2250
    },
    {
      "epoch": 5.011135857461024,
      "eval_screenspot_CIoU": 0.33845322330792743,
      "eval_screenspot_GIoU": 0.3534944951534271,
      "eval_screenspot_IoU": 0.41957201560338336,
      "eval_screenspot_MAE_all": 0.06564747542142868,
      "eval_screenspot_MAE_h": 0.03897890945275625,
      "eval_screenspot_MAE_w": 0.07650155077377956,
      "eval_screenspot_MAE_x_boxes": 0.07791633903980255,
      "eval_screenspot_MAE_y_boxes": 0.04749378779282173,
      "eval_screenspot_inside_bbox": 0.6566666762034098,
      "eval_screenspot_loss": 1.6820822954177856,
      "eval_screenspot_loss_ce": 0.00038540839644459385,
      "eval_screenspot_loss_iou": 0.6896158854166666,
      "eval_screenspot_loss_num": 0.07676951090494792,
      "eval_screenspot_loss_xval": 1.7643229166666667,
      "eval_screenspot_runtime": 47.089,
      "eval_screenspot_samples_per_second": 1.89,
      "eval_screenspot_steps_per_second": 0.064,
      "num_input_tokens_seen": 126043356,
      "step": 2250
    },
    {
      "epoch": 5.011135857461024,
      "eval_compot_CIoU": 0.35792700946331024,
      "eval_compot_GIoU": 0.37573473155498505,
      "eval_compot_IoU": 0.4099069982767105,
      "eval_compot_MAE_all": 0.0185648575425148,
      "eval_compot_MAE_h": 0.008833811618387699,
      "eval_compot_MAE_w": 0.02340342104434967,
      "eval_compot_MAE_x_boxes": 0.02984704216942191,
      "eval_compot_MAE_y_boxes": 0.006203887518495321,
      "eval_compot_inside_bbox": 0.6458333432674408,
      "eval_compot_loss": 1.367824673652649,
      "eval_compot_loss_ce": 0.00032587358145974576,
      "eval_compot_loss_iou": 0.623779296875,
      "eval_compot_loss_num": 0.017635345458984375,
      "eval_compot_loss_xval": 1.33447265625,
      "eval_compot_runtime": 24.6368,
      "eval_compot_samples_per_second": 2.029,
      "eval_compot_steps_per_second": 0.081,
      "num_input_tokens_seen": 126043356,
      "step": 2250
    },
    {
      "epoch": 5.011135857461024,
      "eval_custom_ui_val_CIoU": 0.45421523518032497,
      "eval_custom_ui_val_GIoU": 0.47301916943656075,
      "eval_custom_ui_val_IoU": 0.5073628789848752,
      "eval_custom_ui_val_MAE_all": 0.03390951289070977,
      "eval_custom_ui_val_MAE_h": 0.020180532294842932,
      "eval_custom_ui_val_MAE_w": 0.03951867276595698,
      "eval_custom_ui_val_MAE_x_boxes": 0.039389809800518885,
      "eval_custom_ui_val_MAE_y_boxes": 0.016320706371011004,
      "eval_custom_ui_val_inside_bbox": 0.7353395091162788,
      "eval_custom_ui_val_loss": 1.2387813329696655,
      "eval_custom_ui_val_loss_ce": 0.0004057693327518387,
      "eval_custom_ui_val_loss_iou": 0.521484375,
      "eval_custom_ui_val_loss_num": 0.033151202731662326,
      "eval_custom_ui_val_loss_xval": 1.2088758680555556,
      "eval_custom_ui_val_runtime": 73.0296,
      "eval_custom_ui_val_samples_per_second": 3.629,
      "eval_custom_ui_val_steps_per_second": 0.123,
      "num_input_tokens_seen": 126043356,
      "step": 2250
    },
    {
      "epoch": 5.011135857461024,
      "loss": 0.9642201066017151,
      "loss_ce": 0.00035298120928928256,
      "loss_iou": 0.419921875,
      "loss_num": 0.024169921875,
      "loss_xval": 0.96484375,
      "num_input_tokens_seen": 126043356,
      "step": 2250
    },
    {
      "epoch": 5.013363028953229,
      "grad_norm": 21.67910385131836,
      "learning_rate": 1e-06,
      "loss": 0.6784,
      "num_input_tokens_seen": 126099280,
      "step": 2251
    },
    {
      "epoch": 5.013363028953229,
      "loss": 0.6512374877929688,
      "loss_ce": 0.0003585785743780434,
      "loss_iou": 0.28125,
      "loss_num": 0.017822265625,
      "loss_xval": 0.65234375,
      "num_input_tokens_seen": 126099280,
      "step": 2251
    },
    {
      "epoch": 5.015590200445434,
      "grad_norm": 29.783891677856445,
      "learning_rate": 1e-06,
      "loss": 1.0225,
      "num_input_tokens_seen": 126154972,
      "step": 2252
    },
    {
      "epoch": 5.015590200445434,
      "loss": 0.9770213961601257,
      "loss_ce": 0.0007030742126516998,
      "loss_iou": 0.380859375,
      "loss_num": 0.04296875,
      "loss_xval": 0.9765625,
      "num_input_tokens_seen": 126154972,
      "step": 2252
    },
    {
      "epoch": 5.017817371937639,
      "grad_norm": 28.99271011352539,
      "learning_rate": 1e-06,
      "loss": 0.713,
      "num_input_tokens_seen": 126208548,
      "step": 2253
    },
    {
      "epoch": 5.017817371937639,
      "loss": 0.7910330295562744,
      "loss_ce": 0.0005056494846940041,
      "loss_iou": 0.3046875,
      "loss_num": 0.0361328125,
      "loss_xval": 0.7890625,
      "num_input_tokens_seen": 126208548,
      "step": 2253
    },
    {
      "epoch": 5.020044543429844,
      "grad_norm": 13.615692138671875,
      "learning_rate": 1e-06,
      "loss": 0.522,
      "num_input_tokens_seen": 126262364,
      "step": 2254
    },
    {
      "epoch": 5.020044543429844,
      "loss": 0.4722355008125305,
      "loss_ce": 0.0002506262972019613,
      "loss_iou": 0.1796875,
      "loss_num": 0.022705078125,
      "loss_xval": 0.47265625,
      "num_input_tokens_seen": 126262364,
      "step": 2254
    },
    {
      "epoch": 5.022271714922049,
      "grad_norm": 13.118451118469238,
      "learning_rate": 1e-06,
      "loss": 0.7156,
      "num_input_tokens_seen": 126319680,
      "step": 2255
    },
    {
      "epoch": 5.022271714922049,
      "loss": 0.7510231733322144,
      "loss_ce": 0.00029075576458126307,
      "loss_iou": 0.31640625,
      "loss_num": 0.0233154296875,
      "loss_xval": 0.75,
      "num_input_tokens_seen": 126319680,
      "step": 2255
    },
    {
      "epoch": 5.0244988864142535,
      "grad_norm": 51.616050720214844,
      "learning_rate": 1e-06,
      "loss": 0.9654,
      "num_input_tokens_seen": 126375108,
      "step": 2256
    },
    {
      "epoch": 5.0244988864142535,
      "loss": 0.7489110231399536,
      "loss_ce": 0.00025383190950378776,
      "loss_iou": 0.3125,
      "loss_num": 0.025146484375,
      "loss_xval": 0.75,
      "num_input_tokens_seen": 126375108,
      "step": 2256
    },
    {
      "epoch": 5.026726057906459,
      "grad_norm": 15.720102310180664,
      "learning_rate": 1e-06,
      "loss": 0.8735,
      "num_input_tokens_seen": 126432556,
      "step": 2257
    },
    {
      "epoch": 5.026726057906459,
      "loss": 0.6794352531433105,
      "loss_ce": 0.00023599226551596075,
      "loss_iou": 0.30078125,
      "loss_num": 0.015380859375,
      "loss_xval": 0.6796875,
      "num_input_tokens_seen": 126432556,
      "step": 2257
    },
    {
      "epoch": 5.028953229398664,
      "grad_norm": 35.91153335571289,
      "learning_rate": 1e-06,
      "loss": 0.6807,
      "num_input_tokens_seen": 126489040,
      "step": 2258
    },
    {
      "epoch": 5.028953229398664,
      "loss": 0.6047276258468628,
      "loss_ce": 0.00023541940026916564,
      "loss_iou": 0.2470703125,
      "loss_num": 0.0220947265625,
      "loss_xval": 0.60546875,
      "num_input_tokens_seen": 126489040,
      "step": 2258
    },
    {
      "epoch": 5.031180400890869,
      "grad_norm": 19.358016967773438,
      "learning_rate": 1e-06,
      "loss": 0.7165,
      "num_input_tokens_seen": 126546532,
      "step": 2259
    },
    {
      "epoch": 5.031180400890869,
      "loss": 0.5831573605537415,
      "loss_ce": 0.0006378005491569638,
      "loss_iou": 0.25390625,
      "loss_num": 0.01519775390625,
      "loss_xval": 0.58203125,
      "num_input_tokens_seen": 126546532,
      "step": 2259
    },
    {
      "epoch": 5.033407572383074,
      "grad_norm": 29.59214973449707,
      "learning_rate": 1e-06,
      "loss": 0.8328,
      "num_input_tokens_seen": 126600020,
      "step": 2260
    },
    {
      "epoch": 5.033407572383074,
      "loss": 0.8578157424926758,
      "loss_ce": 0.0002718048053793609,
      "loss_iou": 0.365234375,
      "loss_num": 0.025634765625,
      "loss_xval": 0.859375,
      "num_input_tokens_seen": 126600020,
      "step": 2260
    },
    {
      "epoch": 5.035634743875279,
      "grad_norm": 24.364622116088867,
      "learning_rate": 1e-06,
      "loss": 0.6877,
      "num_input_tokens_seen": 126658112,
      "step": 2261
    },
    {
      "epoch": 5.035634743875279,
      "loss": 0.7512416839599609,
      "loss_ce": 0.00026507957954891026,
      "loss_iou": 0.333984375,
      "loss_num": 0.0166015625,
      "loss_xval": 0.75,
      "num_input_tokens_seen": 126658112,
      "step": 2261
    },
    {
      "epoch": 5.0378619153674835,
      "grad_norm": 20.649215698242188,
      "learning_rate": 1e-06,
      "loss": 0.8893,
      "num_input_tokens_seen": 126715528,
      "step": 2262
    },
    {
      "epoch": 5.0378619153674835,
      "loss": 0.9017385244369507,
      "loss_ce": 0.0002493016654625535,
      "loss_iou": 0.3359375,
      "loss_num": 0.04638671875,
      "loss_xval": 0.90234375,
      "num_input_tokens_seen": 126715528,
      "step": 2262
    },
    {
      "epoch": 5.040089086859688,
      "grad_norm": 27.44195556640625,
      "learning_rate": 1e-06,
      "loss": 0.531,
      "num_input_tokens_seen": 126769300,
      "step": 2263
    },
    {
      "epoch": 5.040089086859688,
      "loss": 0.6221789121627808,
      "loss_ce": 0.00023064325796440244,
      "loss_iou": 0.283203125,
      "loss_num": 0.01092529296875,
      "loss_xval": 0.62109375,
      "num_input_tokens_seen": 126769300,
      "step": 2263
    },
    {
      "epoch": 5.042316258351893,
      "grad_norm": 16.656320571899414,
      "learning_rate": 1e-06,
      "loss": 0.5794,
      "num_input_tokens_seen": 126825972,
      "step": 2264
    },
    {
      "epoch": 5.042316258351893,
      "loss": 0.6281655430793762,
      "loss_ce": 0.0002358591154916212,
      "loss_iou": 0.271484375,
      "loss_num": 0.0166015625,
      "loss_xval": 0.62890625,
      "num_input_tokens_seen": 126825972,
      "step": 2264
    },
    {
      "epoch": 5.044543429844098,
      "grad_norm": 29.34036636352539,
      "learning_rate": 1e-06,
      "loss": 0.7923,
      "num_input_tokens_seen": 126883048,
      "step": 2265
    },
    {
      "epoch": 5.044543429844098,
      "loss": 0.6961706876754761,
      "loss_ce": 0.000247849035076797,
      "loss_iou": 0.3046875,
      "loss_num": 0.0169677734375,
      "loss_xval": 0.6953125,
      "num_input_tokens_seen": 126883048,
      "step": 2265
    },
    {
      "epoch": 5.046770601336303,
      "grad_norm": 28.162872314453125,
      "learning_rate": 1e-06,
      "loss": 1.0247,
      "num_input_tokens_seen": 126937628,
      "step": 2266
    },
    {
      "epoch": 5.046770601336303,
      "loss": 0.8716781139373779,
      "loss_ce": 0.0003402004367671907,
      "loss_iou": 0.392578125,
      "loss_num": 0.0172119140625,
      "loss_xval": 0.87109375,
      "num_input_tokens_seen": 126937628,
      "step": 2266
    },
    {
      "epoch": 5.048997772828508,
      "grad_norm": 20.128053665161133,
      "learning_rate": 1e-06,
      "loss": 1.1153,
      "num_input_tokens_seen": 126989292,
      "step": 2267
    },
    {
      "epoch": 5.048997772828508,
      "loss": 1.152024507522583,
      "loss_ce": 0.0004131349269300699,
      "loss_iou": 0.51171875,
      "loss_num": 0.02587890625,
      "loss_xval": 1.1484375,
      "num_input_tokens_seen": 126989292,
      "step": 2267
    },
    {
      "epoch": 5.051224944320713,
      "grad_norm": 17.286222457885742,
      "learning_rate": 1e-06,
      "loss": 0.5722,
      "num_input_tokens_seen": 127046740,
      "step": 2268
    },
    {
      "epoch": 5.051224944320713,
      "loss": 0.39940914511680603,
      "loss_ce": 0.00023922644322738051,
      "loss_iou": 0.16015625,
      "loss_num": 0.015869140625,
      "loss_xval": 0.3984375,
      "num_input_tokens_seen": 127046740,
      "step": 2268
    },
    {
      "epoch": 5.0534521158129175,
      "grad_norm": 17.838388442993164,
      "learning_rate": 1e-06,
      "loss": 0.6537,
      "num_input_tokens_seen": 127103684,
      "step": 2269
    },
    {
      "epoch": 5.0534521158129175,
      "loss": 0.643367350101471,
      "loss_ce": 0.0003009595675393939,
      "loss_iou": 0.291015625,
      "loss_num": 0.011962890625,
      "loss_xval": 0.64453125,
      "num_input_tokens_seen": 127103684,
      "step": 2269
    },
    {
      "epoch": 5.055679287305122,
      "grad_norm": 22.380460739135742,
      "learning_rate": 1e-06,
      "loss": 0.8209,
      "num_input_tokens_seen": 127160244,
      "step": 2270
    },
    {
      "epoch": 5.055679287305122,
      "loss": 0.7935837507247925,
      "loss_ce": 0.000370887661119923,
      "loss_iou": 0.32421875,
      "loss_num": 0.028564453125,
      "loss_xval": 0.79296875,
      "num_input_tokens_seen": 127160244,
      "step": 2270
    },
    {
      "epoch": 5.057906458797327,
      "grad_norm": 19.87263298034668,
      "learning_rate": 1e-06,
      "loss": 0.7521,
      "num_input_tokens_seen": 127215716,
      "step": 2271
    },
    {
      "epoch": 5.057906458797327,
      "loss": 0.8534216284751892,
      "loss_ce": 0.0002721815253607929,
      "loss_iou": 0.357421875,
      "loss_num": 0.027587890625,
      "loss_xval": 0.8515625,
      "num_input_tokens_seen": 127215716,
      "step": 2271
    },
    {
      "epoch": 5.060133630289532,
      "grad_norm": 24.000696182250977,
      "learning_rate": 1e-06,
      "loss": 0.5931,
      "num_input_tokens_seen": 127274152,
      "step": 2272
    },
    {
      "epoch": 5.060133630289532,
      "loss": 0.5939878225326538,
      "loss_ce": 0.00023785245139151812,
      "loss_iou": 0.2451171875,
      "loss_num": 0.0208740234375,
      "loss_xval": 0.59375,
      "num_input_tokens_seen": 127274152,
      "step": 2272
    },
    {
      "epoch": 5.062360801781737,
      "grad_norm": 17.99005889892578,
      "learning_rate": 1e-06,
      "loss": 0.6256,
      "num_input_tokens_seen": 127330676,
      "step": 2273
    },
    {
      "epoch": 5.062360801781737,
      "loss": 0.5732445120811462,
      "loss_ce": 0.00024645167286507785,
      "loss_iou": 0.244140625,
      "loss_num": 0.016845703125,
      "loss_xval": 0.57421875,
      "num_input_tokens_seen": 127330676,
      "step": 2273
    },
    {
      "epoch": 5.064587973273942,
      "grad_norm": 17.27025032043457,
      "learning_rate": 1e-06,
      "loss": 0.6703,
      "num_input_tokens_seen": 127385060,
      "step": 2274
    },
    {
      "epoch": 5.064587973273942,
      "loss": 0.6232882738113403,
      "loss_ce": 0.000241378482314758,
      "loss_iou": 0.287109375,
      "loss_num": 0.00958251953125,
      "loss_xval": 0.625,
      "num_input_tokens_seen": 127385060,
      "step": 2274
    },
    {
      "epoch": 5.066815144766147,
      "grad_norm": 18.195852279663086,
      "learning_rate": 1e-06,
      "loss": 0.7011,
      "num_input_tokens_seen": 127441056,
      "step": 2275
    },
    {
      "epoch": 5.066815144766147,
      "loss": 0.6917237043380737,
      "loss_ce": 0.00031748326728120446,
      "loss_iou": 0.2734375,
      "loss_num": 0.0289306640625,
      "loss_xval": 0.69140625,
      "num_input_tokens_seen": 127441056,
      "step": 2275
    },
    {
      "epoch": 5.0690423162583516,
      "grad_norm": 17.40542221069336,
      "learning_rate": 1e-06,
      "loss": 0.6412,
      "num_input_tokens_seen": 127495888,
      "step": 2276
    },
    {
      "epoch": 5.0690423162583516,
      "loss": 0.7902520895004272,
      "loss_ce": 0.00021307067072484642,
      "loss_iou": 0.322265625,
      "loss_num": 0.029052734375,
      "loss_xval": 0.7890625,
      "num_input_tokens_seen": 127495888,
      "step": 2276
    },
    {
      "epoch": 5.071269487750556,
      "grad_norm": 15.051342010498047,
      "learning_rate": 1e-06,
      "loss": 0.9653,
      "num_input_tokens_seen": 127552560,
      "step": 2277
    },
    {
      "epoch": 5.071269487750556,
      "loss": 1.024023175239563,
      "loss_ce": 0.0005856686620973051,
      "loss_iou": 0.43359375,
      "loss_num": 0.03173828125,
      "loss_xval": 1.0234375,
      "num_input_tokens_seen": 127552560,
      "step": 2277
    },
    {
      "epoch": 5.073496659242761,
      "grad_norm": 15.868587493896484,
      "learning_rate": 1e-06,
      "loss": 0.6495,
      "num_input_tokens_seen": 127609712,
      "step": 2278
    },
    {
      "epoch": 5.073496659242761,
      "loss": 0.6852548718452454,
      "loss_ce": 0.00019630559836514294,
      "loss_iou": 0.29296875,
      "loss_num": 0.0198974609375,
      "loss_xval": 0.68359375,
      "num_input_tokens_seen": 127609712,
      "step": 2278
    },
    {
      "epoch": 5.075723830734967,
      "grad_norm": 23.950061798095703,
      "learning_rate": 1e-06,
      "loss": 0.6507,
      "num_input_tokens_seen": 127667116,
      "step": 2279
    },
    {
      "epoch": 5.075723830734967,
      "loss": 0.6977980136871338,
      "loss_ce": 0.00028824395849369466,
      "loss_iou": 0.31640625,
      "loss_num": 0.01287841796875,
      "loss_xval": 0.69921875,
      "num_input_tokens_seen": 127667116,
      "step": 2279
    },
    {
      "epoch": 5.077951002227172,
      "grad_norm": 83.21546936035156,
      "learning_rate": 1e-06,
      "loss": 0.6931,
      "num_input_tokens_seen": 127724152,
      "step": 2280
    },
    {
      "epoch": 5.077951002227172,
      "loss": 0.575419545173645,
      "loss_ce": 0.00022418698063120246,
      "loss_iou": 0.2578125,
      "loss_num": 0.01214599609375,
      "loss_xval": 0.57421875,
      "num_input_tokens_seen": 127724152,
      "step": 2280
    },
    {
      "epoch": 5.080178173719377,
      "grad_norm": 16.499300003051758,
      "learning_rate": 1e-06,
      "loss": 0.7207,
      "num_input_tokens_seen": 127779712,
      "step": 2281
    },
    {
      "epoch": 5.080178173719377,
      "loss": 0.7757115364074707,
      "loss_ce": 0.00025984214153140783,
      "loss_iou": 0.30078125,
      "loss_num": 0.034912109375,
      "loss_xval": 0.77734375,
      "num_input_tokens_seen": 127779712,
      "step": 2281
    },
    {
      "epoch": 5.082405345211582,
      "grad_norm": 21.090255737304688,
      "learning_rate": 1e-06,
      "loss": 0.8105,
      "num_input_tokens_seen": 127835904,
      "step": 2282
    },
    {
      "epoch": 5.082405345211582,
      "loss": 0.7296441793441772,
      "loss_ce": 0.00027409923495724797,
      "loss_iou": 0.328125,
      "loss_num": 0.0142822265625,
      "loss_xval": 0.73046875,
      "num_input_tokens_seen": 127835904,
      "step": 2282
    },
    {
      "epoch": 5.0846325167037865,
      "grad_norm": 21.14578628540039,
      "learning_rate": 1e-06,
      "loss": 0.8618,
      "num_input_tokens_seen": 127892840,
      "step": 2283
    },
    {
      "epoch": 5.0846325167037865,
      "loss": 0.9588485360145569,
      "loss_ce": 0.00023040127416606992,
      "loss_iou": 0.3984375,
      "loss_num": 0.0322265625,
      "loss_xval": 0.95703125,
      "num_input_tokens_seen": 127892840,
      "step": 2283
    },
    {
      "epoch": 5.086859688195991,
      "grad_norm": 48.02599334716797,
      "learning_rate": 1e-06,
      "loss": 0.5714,
      "num_input_tokens_seen": 127951528,
      "step": 2284
    },
    {
      "epoch": 5.086859688195991,
      "loss": 0.5024325847625732,
      "loss_ce": 0.00023529936152044684,
      "loss_iou": 0.2265625,
      "loss_num": 0.00994873046875,
      "loss_xval": 0.50390625,
      "num_input_tokens_seen": 127951528,
      "step": 2284
    },
    {
      "epoch": 5.089086859688196,
      "grad_norm": 15.565524101257324,
      "learning_rate": 1e-06,
      "loss": 0.8572,
      "num_input_tokens_seen": 128008636,
      "step": 2285
    },
    {
      "epoch": 5.089086859688196,
      "loss": 0.9711748361587524,
      "loss_ce": 0.0002275363076478243,
      "loss_iou": 0.40625,
      "loss_num": 0.03125,
      "loss_xval": 0.97265625,
      "num_input_tokens_seen": 128008636,
      "step": 2285
    },
    {
      "epoch": 5.091314031180401,
      "grad_norm": 23.433475494384766,
      "learning_rate": 1e-06,
      "loss": 0.8038,
      "num_input_tokens_seen": 128066100,
      "step": 2286
    },
    {
      "epoch": 5.091314031180401,
      "loss": 0.787333607673645,
      "loss_ce": 0.00022421692847274244,
      "loss_iou": 0.345703125,
      "loss_num": 0.0194091796875,
      "loss_xval": 0.7890625,
      "num_input_tokens_seen": 128066100,
      "step": 2286
    },
    {
      "epoch": 5.093541202672606,
      "grad_norm": 24.027969360351562,
      "learning_rate": 1e-06,
      "loss": 0.7235,
      "num_input_tokens_seen": 128121516,
      "step": 2287
    },
    {
      "epoch": 5.093541202672606,
      "loss": 0.7629618644714355,
      "loss_ce": 0.0002665651263669133,
      "loss_iou": 0.349609375,
      "loss_num": 0.012451171875,
      "loss_xval": 0.76171875,
      "num_input_tokens_seen": 128121516,
      "step": 2287
    },
    {
      "epoch": 5.095768374164811,
      "grad_norm": 19.88593292236328,
      "learning_rate": 1e-06,
      "loss": 0.9881,
      "num_input_tokens_seen": 128174484,
      "step": 2288
    },
    {
      "epoch": 5.095768374164811,
      "loss": 1.1999242305755615,
      "loss_ce": 0.00021719752112403512,
      "loss_iou": 0.4921875,
      "loss_num": 0.042724609375,
      "loss_xval": 1.203125,
      "num_input_tokens_seen": 128174484,
      "step": 2288
    },
    {
      "epoch": 5.097995545657016,
      "grad_norm": 45.70637130737305,
      "learning_rate": 1e-06,
      "loss": 0.5952,
      "num_input_tokens_seen": 128232784,
      "step": 2289
    },
    {
      "epoch": 5.097995545657016,
      "loss": 0.6135152578353882,
      "loss_ce": 0.00023402433726005256,
      "loss_iou": 0.267578125,
      "loss_num": 0.015869140625,
      "loss_xval": 0.61328125,
      "num_input_tokens_seen": 128232784,
      "step": 2289
    },
    {
      "epoch": 5.1002227171492205,
      "grad_norm": 16.309289932250977,
      "learning_rate": 1e-06,
      "loss": 0.7085,
      "num_input_tokens_seen": 128287280,
      "step": 2290
    },
    {
      "epoch": 5.1002227171492205,
      "loss": 0.734100878238678,
      "loss_ce": 0.0002141737350029871,
      "loss_iou": 0.30859375,
      "loss_num": 0.02294921875,
      "loss_xval": 0.734375,
      "num_input_tokens_seen": 128287280,
      "step": 2290
    },
    {
      "epoch": 5.102449888641425,
      "grad_norm": 15.12460994720459,
      "learning_rate": 1e-06,
      "loss": 0.5811,
      "num_input_tokens_seen": 128342260,
      "step": 2291
    },
    {
      "epoch": 5.102449888641425,
      "loss": 0.5832387208938599,
      "loss_ce": 0.000230930614634417,
      "loss_iou": 0.2578125,
      "loss_num": 0.013916015625,
      "loss_xval": 0.58203125,
      "num_input_tokens_seen": 128342260,
      "step": 2291
    },
    {
      "epoch": 5.10467706013363,
      "grad_norm": 16.05999183654785,
      "learning_rate": 1e-06,
      "loss": 0.6387,
      "num_input_tokens_seen": 128394812,
      "step": 2292
    },
    {
      "epoch": 5.10467706013363,
      "loss": 0.6987403035163879,
      "loss_ce": 0.00025396081036888063,
      "loss_iou": 0.298828125,
      "loss_num": 0.02001953125,
      "loss_xval": 0.69921875,
      "num_input_tokens_seen": 128394812,
      "step": 2292
    },
    {
      "epoch": 5.106904231625835,
      "grad_norm": 17.23577117919922,
      "learning_rate": 1e-06,
      "loss": 0.5956,
      "num_input_tokens_seen": 128451808,
      "step": 2293
    },
    {
      "epoch": 5.106904231625835,
      "loss": 0.5036342740058899,
      "loss_ce": 0.0002162978344131261,
      "loss_iou": 0.212890625,
      "loss_num": 0.0157470703125,
      "loss_xval": 0.50390625,
      "num_input_tokens_seen": 128451808,
      "step": 2293
    },
    {
      "epoch": 5.10913140311804,
      "grad_norm": 28.054798126220703,
      "learning_rate": 1e-06,
      "loss": 1.03,
      "num_input_tokens_seen": 128505588,
      "step": 2294
    },
    {
      "epoch": 5.10913140311804,
      "loss": 0.8542524576187134,
      "loss_ce": 0.0002485929289832711,
      "loss_iou": 0.365234375,
      "loss_num": 0.02490234375,
      "loss_xval": 0.85546875,
      "num_input_tokens_seen": 128505588,
      "step": 2294
    },
    {
      "epoch": 5.111358574610245,
      "grad_norm": 15.79360580444336,
      "learning_rate": 1e-06,
      "loss": 0.7536,
      "num_input_tokens_seen": 128564004,
      "step": 2295
    },
    {
      "epoch": 5.111358574610245,
      "loss": 0.846659779548645,
      "loss_ce": 0.00022423264454118907,
      "loss_iou": 0.349609375,
      "loss_num": 0.029541015625,
      "loss_xval": 0.84765625,
      "num_input_tokens_seen": 128564004,
      "step": 2295
    },
    {
      "epoch": 5.11358574610245,
      "grad_norm": 35.329349517822266,
      "learning_rate": 1e-06,
      "loss": 0.8545,
      "num_input_tokens_seen": 128622824,
      "step": 2296
    },
    {
      "epoch": 5.11358574610245,
      "loss": 0.8776304721832275,
      "loss_ce": 0.00043314468348398805,
      "loss_iou": 0.369140625,
      "loss_num": 0.02783203125,
      "loss_xval": 0.87890625,
      "num_input_tokens_seen": 128622824,
      "step": 2296
    },
    {
      "epoch": 5.1158129175946545,
      "grad_norm": 17.822633743286133,
      "learning_rate": 1e-06,
      "loss": 0.7707,
      "num_input_tokens_seen": 128679316,
      "step": 2297
    },
    {
      "epoch": 5.1158129175946545,
      "loss": 0.49904176592826843,
      "loss_ce": 0.0002624738262966275,
      "loss_iou": 0.201171875,
      "loss_num": 0.0194091796875,
      "loss_xval": 0.498046875,
      "num_input_tokens_seen": 128679316,
      "step": 2297
    },
    {
      "epoch": 5.118040089086859,
      "grad_norm": 24.6488094329834,
      "learning_rate": 1e-06,
      "loss": 0.7886,
      "num_input_tokens_seen": 128734780,
      "step": 2298
    },
    {
      "epoch": 5.118040089086859,
      "loss": 0.8342887163162231,
      "loss_ce": 0.00030440062982961535,
      "loss_iou": 0.384765625,
      "loss_num": 0.0133056640625,
      "loss_xval": 0.8359375,
      "num_input_tokens_seen": 128734780,
      "step": 2298
    },
    {
      "epoch": 5.120267260579064,
      "grad_norm": 21.603078842163086,
      "learning_rate": 1e-06,
      "loss": 0.8968,
      "num_input_tokens_seen": 128790756,
      "step": 2299
    },
    {
      "epoch": 5.120267260579064,
      "loss": 1.144487738609314,
      "loss_ce": 0.0004448448889888823,
      "loss_iou": 0.451171875,
      "loss_num": 0.048095703125,
      "loss_xval": 1.140625,
      "num_input_tokens_seen": 128790756,
      "step": 2299
    },
    {
      "epoch": 5.122494432071269,
      "grad_norm": 22.150482177734375,
      "learning_rate": 1e-06,
      "loss": 0.8125,
      "num_input_tokens_seen": 128847224,
      "step": 2300
    },
    {
      "epoch": 5.122494432071269,
      "loss": 0.8337835669517517,
      "loss_ce": 0.0002874800411518663,
      "loss_iou": 0.330078125,
      "loss_num": 0.034423828125,
      "loss_xval": 0.83203125,
      "num_input_tokens_seen": 128847224,
      "step": 2300
    },
    {
      "epoch": 5.124721603563474,
      "grad_norm": 17.031408309936523,
      "learning_rate": 1e-06,
      "loss": 0.5427,
      "num_input_tokens_seen": 128904848,
      "step": 2301
    },
    {
      "epoch": 5.124721603563474,
      "loss": 0.5903265476226807,
      "loss_ce": 0.00023866641276981682,
      "loss_iou": 0.2333984375,
      "loss_num": 0.02490234375,
      "loss_xval": 0.58984375,
      "num_input_tokens_seen": 128904848,
      "step": 2301
    },
    {
      "epoch": 5.12694877505568,
      "grad_norm": 17.657089233398438,
      "learning_rate": 1e-06,
      "loss": 0.5752,
      "num_input_tokens_seen": 128960196,
      "step": 2302
    },
    {
      "epoch": 5.12694877505568,
      "loss": 0.4570813775062561,
      "loss_ce": 0.00023321554181165993,
      "loss_iou": 0.193359375,
      "loss_num": 0.01397705078125,
      "loss_xval": 0.45703125,
      "num_input_tokens_seen": 128960196,
      "step": 2302
    },
    {
      "epoch": 5.129175946547885,
      "grad_norm": 16.3359317779541,
      "learning_rate": 1e-06,
      "loss": 0.6331,
      "num_input_tokens_seen": 129017636,
      "step": 2303
    },
    {
      "epoch": 5.129175946547885,
      "loss": 0.6099591255187988,
      "loss_ce": 0.0005841399542987347,
      "loss_iou": 0.263671875,
      "loss_num": 0.0167236328125,
      "loss_xval": 0.609375,
      "num_input_tokens_seen": 129017636,
      "step": 2303
    },
    {
      "epoch": 5.131403118040089,
      "grad_norm": 21.530765533447266,
      "learning_rate": 1e-06,
      "loss": 0.9466,
      "num_input_tokens_seen": 129073348,
      "step": 2304
    },
    {
      "epoch": 5.131403118040089,
      "loss": 1.1084952354431152,
      "loss_ce": 0.00034102535573765635,
      "loss_iou": 0.474609375,
      "loss_num": 0.03173828125,
      "loss_xval": 1.109375,
      "num_input_tokens_seen": 129073348,
      "step": 2304
    },
    {
      "epoch": 5.133630289532294,
      "grad_norm": 37.55251693725586,
      "learning_rate": 1e-06,
      "loss": 0.9198,
      "num_input_tokens_seen": 129130900,
      "step": 2305
    },
    {
      "epoch": 5.133630289532294,
      "loss": 0.638685941696167,
      "loss_ce": 0.0002581804874353111,
      "loss_iou": 0.267578125,
      "loss_num": 0.020751953125,
      "loss_xval": 0.63671875,
      "num_input_tokens_seen": 129130900,
      "step": 2305
    },
    {
      "epoch": 5.135857461024499,
      "grad_norm": 28.42352867126465,
      "learning_rate": 1e-06,
      "loss": 0.737,
      "num_input_tokens_seen": 129183076,
      "step": 2306
    },
    {
      "epoch": 5.135857461024499,
      "loss": 0.6569973826408386,
      "loss_ce": 0.0002591205993667245,
      "loss_iou": 0.29296875,
      "loss_num": 0.0145263671875,
      "loss_xval": 0.65625,
      "num_input_tokens_seen": 129183076,
      "step": 2306
    },
    {
      "epoch": 5.138084632516704,
      "grad_norm": 15.96066951751709,
      "learning_rate": 1e-06,
      "loss": 0.8596,
      "num_input_tokens_seen": 129239480,
      "step": 2307
    },
    {
      "epoch": 5.138084632516704,
      "loss": 0.7568236589431763,
      "loss_ce": 0.00023182888980954885,
      "loss_iou": 0.330078125,
      "loss_num": 0.01953125,
      "loss_xval": 0.7578125,
      "num_input_tokens_seen": 129239480,
      "step": 2307
    },
    {
      "epoch": 5.140311804008909,
      "grad_norm": 21.160701751708984,
      "learning_rate": 1e-06,
      "loss": 0.7788,
      "num_input_tokens_seen": 129295984,
      "step": 2308
    },
    {
      "epoch": 5.140311804008909,
      "loss": 0.9611057043075562,
      "loss_ce": 0.00041232837247662246,
      "loss_iou": 0.36328125,
      "loss_num": 0.04638671875,
      "loss_xval": 0.9609375,
      "num_input_tokens_seen": 129295984,
      "step": 2308
    },
    {
      "epoch": 5.142538975501114,
      "grad_norm": 23.850879669189453,
      "learning_rate": 1e-06,
      "loss": 0.8738,
      "num_input_tokens_seen": 129350436,
      "step": 2309
    },
    {
      "epoch": 5.142538975501114,
      "loss": 0.835231602191925,
      "loss_ce": 0.0002706579980440438,
      "loss_iou": 0.337890625,
      "loss_num": 0.031982421875,
      "loss_xval": 0.8359375,
      "num_input_tokens_seen": 129350436,
      "step": 2309
    },
    {
      "epoch": 5.144766146993319,
      "grad_norm": 40.127952575683594,
      "learning_rate": 1e-06,
      "loss": 0.7161,
      "num_input_tokens_seen": 129406620,
      "step": 2310
    },
    {
      "epoch": 5.144766146993319,
      "loss": 0.691180944442749,
      "loss_ce": 0.00026295060524716973,
      "loss_iou": 0.3125,
      "loss_num": 0.01348876953125,
      "loss_xval": 0.69140625,
      "num_input_tokens_seen": 129406620,
      "step": 2310
    },
    {
      "epoch": 5.146993318485523,
      "grad_norm": 21.454145431518555,
      "learning_rate": 1e-06,
      "loss": 0.5811,
      "num_input_tokens_seen": 129462148,
      "step": 2311
    },
    {
      "epoch": 5.146993318485523,
      "loss": 0.5917257070541382,
      "loss_ce": 0.0002950755297206342,
      "loss_iou": 0.234375,
      "loss_num": 0.024658203125,
      "loss_xval": 0.58984375,
      "num_input_tokens_seen": 129462148,
      "step": 2311
    },
    {
      "epoch": 5.149220489977728,
      "grad_norm": 16.070457458496094,
      "learning_rate": 1e-06,
      "loss": 0.458,
      "num_input_tokens_seen": 129518980,
      "step": 2312
    },
    {
      "epoch": 5.149220489977728,
      "loss": 0.5424692630767822,
      "loss_ce": 0.0002329152193851769,
      "loss_iou": 0.2314453125,
      "loss_num": 0.015869140625,
      "loss_xval": 0.54296875,
      "num_input_tokens_seen": 129518980,
      "step": 2312
    },
    {
      "epoch": 5.151447661469933,
      "grad_norm": 27.548030853271484,
      "learning_rate": 1e-06,
      "loss": 0.7495,
      "num_input_tokens_seen": 129574880,
      "step": 2313
    },
    {
      "epoch": 5.151447661469933,
      "loss": 0.7275496125221252,
      "loss_ce": 0.00025467833620496094,
      "loss_iou": 0.3203125,
      "loss_num": 0.0169677734375,
      "loss_xval": 0.7265625,
      "num_input_tokens_seen": 129574880,
      "step": 2313
    },
    {
      "epoch": 5.153674832962138,
      "grad_norm": 17.572242736816406,
      "learning_rate": 1e-06,
      "loss": 0.6046,
      "num_input_tokens_seen": 129631804,
      "step": 2314
    },
    {
      "epoch": 5.153674832962138,
      "loss": 0.7023811340332031,
      "loss_ce": 0.0007209961186163127,
      "loss_iou": 0.28125,
      "loss_num": 0.028076171875,
      "loss_xval": 0.703125,
      "num_input_tokens_seen": 129631804,
      "step": 2314
    },
    {
      "epoch": 5.155902004454343,
      "grad_norm": 17.386499404907227,
      "learning_rate": 1e-06,
      "loss": 0.8311,
      "num_input_tokens_seen": 129684280,
      "step": 2315
    },
    {
      "epoch": 5.155902004454343,
      "loss": 0.8039573431015015,
      "loss_ce": 0.00024638883769512177,
      "loss_iou": 0.357421875,
      "loss_num": 0.017578125,
      "loss_xval": 0.8046875,
      "num_input_tokens_seen": 129684280,
      "step": 2315
    },
    {
      "epoch": 5.158129175946548,
      "grad_norm": 18.181276321411133,
      "learning_rate": 1e-06,
      "loss": 0.8983,
      "num_input_tokens_seen": 129742428,
      "step": 2316
    },
    {
      "epoch": 5.158129175946548,
      "loss": 0.8774632811546326,
      "loss_ce": 0.000266014103544876,
      "loss_iou": 0.34375,
      "loss_num": 0.037841796875,
      "loss_xval": 0.87890625,
      "num_input_tokens_seen": 129742428,
      "step": 2316
    },
    {
      "epoch": 5.160356347438753,
      "grad_norm": 21.5535888671875,
      "learning_rate": 1e-06,
      "loss": 0.8057,
      "num_input_tokens_seen": 129796972,
      "step": 2317
    },
    {
      "epoch": 5.160356347438753,
      "loss": 0.9526360630989075,
      "loss_ce": 0.00024349387967959046,
      "loss_iou": 0.38671875,
      "loss_num": 0.0361328125,
      "loss_xval": 0.953125,
      "num_input_tokens_seen": 129796972,
      "step": 2317
    },
    {
      "epoch": 5.1625835189309575,
      "grad_norm": 21.653419494628906,
      "learning_rate": 1e-06,
      "loss": 0.7653,
      "num_input_tokens_seen": 129851752,
      "step": 2318
    },
    {
      "epoch": 5.1625835189309575,
      "loss": 0.8452595472335815,
      "loss_ce": 0.0002888377639465034,
      "loss_iou": 0.345703125,
      "loss_num": 0.030517578125,
      "loss_xval": 0.84375,
      "num_input_tokens_seen": 129851752,
      "step": 2318
    },
    {
      "epoch": 5.164810690423162,
      "grad_norm": 23.30936622619629,
      "learning_rate": 1e-06,
      "loss": 0.7064,
      "num_input_tokens_seen": 129907124,
      "step": 2319
    },
    {
      "epoch": 5.164810690423162,
      "loss": 0.541006326675415,
      "loss_ce": 0.0002348569978494197,
      "loss_iou": 0.228515625,
      "loss_num": 0.0164794921875,
      "loss_xval": 0.5390625,
      "num_input_tokens_seen": 129907124,
      "step": 2319
    },
    {
      "epoch": 5.167037861915367,
      "grad_norm": 16.756088256835938,
      "learning_rate": 1e-06,
      "loss": 0.5693,
      "num_input_tokens_seen": 129963388,
      "step": 2320
    },
    {
      "epoch": 5.167037861915367,
      "loss": 0.481838583946228,
      "loss_ce": 0.00027121190214529634,
      "loss_iou": 0.203125,
      "loss_num": 0.0150146484375,
      "loss_xval": 0.482421875,
      "num_input_tokens_seen": 129963388,
      "step": 2320
    },
    {
      "epoch": 5.169265033407572,
      "grad_norm": 18.584272384643555,
      "learning_rate": 1e-06,
      "loss": 0.896,
      "num_input_tokens_seen": 130019724,
      "step": 2321
    },
    {
      "epoch": 5.169265033407572,
      "loss": 0.9863563776016235,
      "loss_ce": 0.00027232500724494457,
      "loss_iou": 0.41796875,
      "loss_num": 0.0299072265625,
      "loss_xval": 0.984375,
      "num_input_tokens_seen": 130019724,
      "step": 2321
    },
    {
      "epoch": 5.171492204899777,
      "grad_norm": 23.582571029663086,
      "learning_rate": 1e-06,
      "loss": 0.6948,
      "num_input_tokens_seen": 130077412,
      "step": 2322
    },
    {
      "epoch": 5.171492204899777,
      "loss": 0.7178100347518921,
      "loss_ce": 0.0002807402634061873,
      "loss_iou": 0.3125,
      "loss_num": 0.018798828125,
      "loss_xval": 0.71875,
      "num_input_tokens_seen": 130077412,
      "step": 2322
    },
    {
      "epoch": 5.173719376391982,
      "grad_norm": 15.536462783813477,
      "learning_rate": 1e-06,
      "loss": 0.8434,
      "num_input_tokens_seen": 130133952,
      "step": 2323
    },
    {
      "epoch": 5.173719376391982,
      "loss": 0.7571786642074585,
      "loss_ce": 0.00076994055416435,
      "loss_iou": 0.32421875,
      "loss_num": 0.0216064453125,
      "loss_xval": 0.7578125,
      "num_input_tokens_seen": 130133952,
      "step": 2323
    },
    {
      "epoch": 5.1759465478841875,
      "grad_norm": 21.30929946899414,
      "learning_rate": 1e-06,
      "loss": 0.832,
      "num_input_tokens_seen": 130190140,
      "step": 2324
    },
    {
      "epoch": 5.1759465478841875,
      "loss": 0.9763731956481934,
      "loss_ce": 0.0002989959320984781,
      "loss_iou": 0.380859375,
      "loss_num": 0.042724609375,
      "loss_xval": 0.9765625,
      "num_input_tokens_seen": 130190140,
      "step": 2324
    },
    {
      "epoch": 5.178173719376392,
      "grad_norm": 55.067665100097656,
      "learning_rate": 1e-06,
      "loss": 0.8182,
      "num_input_tokens_seen": 130244552,
      "step": 2325
    },
    {
      "epoch": 5.178173719376392,
      "loss": 1.0063856840133667,
      "loss_ce": 0.0002821519155986607,
      "loss_iou": 0.41796875,
      "loss_num": 0.033935546875,
      "loss_xval": 1.0078125,
      "num_input_tokens_seen": 130244552,
      "step": 2325
    },
    {
      "epoch": 5.180400890868597,
      "grad_norm": 19.690275192260742,
      "learning_rate": 1e-06,
      "loss": 0.9497,
      "num_input_tokens_seen": 130301252,
      "step": 2326
    },
    {
      "epoch": 5.180400890868597,
      "loss": 0.765385627746582,
      "loss_ce": 0.00024891988141462207,
      "loss_iou": 0.3359375,
      "loss_num": 0.018310546875,
      "loss_xval": 0.765625,
      "num_input_tokens_seen": 130301252,
      "step": 2326
    },
    {
      "epoch": 5.182628062360802,
      "grad_norm": 14.636940002441406,
      "learning_rate": 1e-06,
      "loss": 0.5755,
      "num_input_tokens_seen": 130357596,
      "step": 2327
    },
    {
      "epoch": 5.182628062360802,
      "loss": 0.5990309715270996,
      "loss_ce": 0.00027604633942246437,
      "loss_iou": 0.26953125,
      "loss_num": 0.011474609375,
      "loss_xval": 0.59765625,
      "num_input_tokens_seen": 130357596,
      "step": 2327
    },
    {
      "epoch": 5.184855233853007,
      "grad_norm": 19.635517120361328,
      "learning_rate": 1e-06,
      "loss": 0.6214,
      "num_input_tokens_seen": 130413372,
      "step": 2328
    },
    {
      "epoch": 5.184855233853007,
      "loss": 0.7523684501647949,
      "loss_ce": 0.00029323421767912805,
      "loss_iou": 0.31640625,
      "loss_num": 0.023681640625,
      "loss_xval": 0.75390625,
      "num_input_tokens_seen": 130413372,
      "step": 2328
    },
    {
      "epoch": 5.187082405345212,
      "grad_norm": 15.182498931884766,
      "learning_rate": 1e-06,
      "loss": 0.628,
      "num_input_tokens_seen": 130468056,
      "step": 2329
    },
    {
      "epoch": 5.187082405345212,
      "loss": 0.513440728187561,
      "loss_ce": 0.00025709014153108,
      "loss_iou": 0.220703125,
      "loss_num": 0.01434326171875,
      "loss_xval": 0.51171875,
      "num_input_tokens_seen": 130468056,
      "step": 2329
    },
    {
      "epoch": 5.189309576837417,
      "grad_norm": 23.259883880615234,
      "learning_rate": 1e-06,
      "loss": 0.5599,
      "num_input_tokens_seen": 130522728,
      "step": 2330
    },
    {
      "epoch": 5.189309576837417,
      "loss": 0.483401358127594,
      "loss_ce": 0.00024706803378649056,
      "loss_iou": 0.1953125,
      "loss_num": 0.018798828125,
      "loss_xval": 0.482421875,
      "num_input_tokens_seen": 130522728,
      "step": 2330
    },
    {
      "epoch": 5.1915367483296215,
      "grad_norm": 27.320947647094727,
      "learning_rate": 1e-06,
      "loss": 0.7893,
      "num_input_tokens_seen": 130577656,
      "step": 2331
    },
    {
      "epoch": 5.1915367483296215,
      "loss": 0.6123491525650024,
      "loss_ce": 0.00028857935103587806,
      "loss_iou": 0.271484375,
      "loss_num": 0.0135498046875,
      "loss_xval": 0.61328125,
      "num_input_tokens_seen": 130577656,
      "step": 2331
    },
    {
      "epoch": 5.193763919821826,
      "grad_norm": 74.38851928710938,
      "learning_rate": 1e-06,
      "loss": 0.6629,
      "num_input_tokens_seen": 130635684,
      "step": 2332
    },
    {
      "epoch": 5.193763919821826,
      "loss": 0.7268493175506592,
      "loss_ce": 0.0002868149313144386,
      "loss_iou": 0.326171875,
      "loss_num": 0.0146484375,
      "loss_xval": 0.7265625,
      "num_input_tokens_seen": 130635684,
      "step": 2332
    },
    {
      "epoch": 5.195991091314031,
      "grad_norm": 21.843446731567383,
      "learning_rate": 1e-06,
      "loss": 0.6575,
      "num_input_tokens_seen": 130693564,
      "step": 2333
    },
    {
      "epoch": 5.195991091314031,
      "loss": 0.5599232912063599,
      "loss_ce": 0.00023093904019333422,
      "loss_iou": 0.251953125,
      "loss_num": 0.011474609375,
      "loss_xval": 0.55859375,
      "num_input_tokens_seen": 130693564,
      "step": 2333
    },
    {
      "epoch": 5.198218262806236,
      "grad_norm": 27.58201789855957,
      "learning_rate": 1e-06,
      "loss": 0.6391,
      "num_input_tokens_seen": 130751884,
      "step": 2334
    },
    {
      "epoch": 5.198218262806236,
      "loss": 0.6432796716690063,
      "loss_ce": 0.00021332071628421545,
      "loss_iou": 0.28515625,
      "loss_num": 0.01409912109375,
      "loss_xval": 0.64453125,
      "num_input_tokens_seen": 130751884,
      "step": 2334
    },
    {
      "epoch": 5.200445434298441,
      "grad_norm": 68.57735443115234,
      "learning_rate": 1e-06,
      "loss": 0.6312,
      "num_input_tokens_seen": 130808940,
      "step": 2335
    },
    {
      "epoch": 5.200445434298441,
      "loss": 0.45080137252807617,
      "loss_ce": 0.00023984728613868356,
      "loss_iou": 0.203125,
      "loss_num": 0.0087890625,
      "loss_xval": 0.451171875,
      "num_input_tokens_seen": 130808940,
      "step": 2335
    },
    {
      "epoch": 5.202672605790646,
      "grad_norm": 23.095355987548828,
      "learning_rate": 1e-06,
      "loss": 0.7388,
      "num_input_tokens_seen": 130864876,
      "step": 2336
    },
    {
      "epoch": 5.202672605790646,
      "loss": 0.8970399498939514,
      "loss_ce": 0.0003114146529696882,
      "loss_iou": 0.380859375,
      "loss_num": 0.0269775390625,
      "loss_xval": 0.8984375,
      "num_input_tokens_seen": 130864876,
      "step": 2336
    },
    {
      "epoch": 5.204899777282851,
      "grad_norm": 16.609811782836914,
      "learning_rate": 1e-06,
      "loss": 0.8964,
      "num_input_tokens_seen": 130919176,
      "step": 2337
    },
    {
      "epoch": 5.204899777282851,
      "loss": 1.08896803855896,
      "loss_ce": 0.0003450649091973901,
      "loss_iou": 0.484375,
      "loss_num": 0.024169921875,
      "loss_xval": 1.0859375,
      "num_input_tokens_seen": 130919176,
      "step": 2337
    },
    {
      "epoch": 5.2071269487750556,
      "grad_norm": 17.34803009033203,
      "learning_rate": 1e-06,
      "loss": 0.6675,
      "num_input_tokens_seen": 130977396,
      "step": 2338
    },
    {
      "epoch": 5.2071269487750556,
      "loss": 0.661956250667572,
      "loss_ce": 0.00021310552256181836,
      "loss_iou": 0.259765625,
      "loss_num": 0.0281982421875,
      "loss_xval": 0.66015625,
      "num_input_tokens_seen": 130977396,
      "step": 2338
    },
    {
      "epoch": 5.20935412026726,
      "grad_norm": 19.648202896118164,
      "learning_rate": 1e-06,
      "loss": 0.8492,
      "num_input_tokens_seen": 131032596,
      "step": 2339
    },
    {
      "epoch": 5.20935412026726,
      "loss": 0.8113721013069153,
      "loss_ce": 0.0003369731712155044,
      "loss_iou": 0.34765625,
      "loss_num": 0.022705078125,
      "loss_xval": 0.8125,
      "num_input_tokens_seen": 131032596,
      "step": 2339
    },
    {
      "epoch": 5.211581291759465,
      "grad_norm": 28.624279022216797,
      "learning_rate": 1e-06,
      "loss": 0.5556,
      "num_input_tokens_seen": 131089720,
      "step": 2340
    },
    {
      "epoch": 5.211581291759465,
      "loss": 0.615825891494751,
      "loss_ce": 0.0003474131808616221,
      "loss_iou": 0.25390625,
      "loss_num": 0.021728515625,
      "loss_xval": 0.6171875,
      "num_input_tokens_seen": 131089720,
      "step": 2340
    },
    {
      "epoch": 5.21380846325167,
      "grad_norm": 32.367496490478516,
      "learning_rate": 1e-06,
      "loss": 0.8943,
      "num_input_tokens_seen": 131147500,
      "step": 2341
    },
    {
      "epoch": 5.21380846325167,
      "loss": 1.0326426029205322,
      "loss_ce": 0.000355017080437392,
      "loss_iou": 0.40234375,
      "loss_num": 0.045166015625,
      "loss_xval": 1.03125,
      "num_input_tokens_seen": 131147500,
      "step": 2341
    },
    {
      "epoch": 5.216035634743875,
      "grad_norm": 32.042381286621094,
      "learning_rate": 1e-06,
      "loss": 0.7757,
      "num_input_tokens_seen": 131202024,
      "step": 2342
    },
    {
      "epoch": 5.216035634743875,
      "loss": 0.7637057900428772,
      "loss_ce": 0.0002780459471978247,
      "loss_iou": 0.31640625,
      "loss_num": 0.02587890625,
      "loss_xval": 0.76171875,
      "num_input_tokens_seen": 131202024,
      "step": 2342
    },
    {
      "epoch": 5.21826280623608,
      "grad_norm": 24.583831787109375,
      "learning_rate": 1e-06,
      "loss": 0.859,
      "num_input_tokens_seen": 131258048,
      "step": 2343
    },
    {
      "epoch": 5.21826280623608,
      "loss": 0.815448522567749,
      "loss_ce": 0.0002629937371239066,
      "loss_iou": 0.31640625,
      "loss_num": 0.036865234375,
      "loss_xval": 0.81640625,
      "num_input_tokens_seen": 131258048,
      "step": 2343
    },
    {
      "epoch": 5.220489977728285,
      "grad_norm": 19.044597625732422,
      "learning_rate": 1e-06,
      "loss": 0.7225,
      "num_input_tokens_seen": 131316644,
      "step": 2344
    },
    {
      "epoch": 5.220489977728285,
      "loss": 0.6471991539001465,
      "loss_ce": 0.00022645638091489673,
      "loss_iou": 0.2451171875,
      "loss_num": 0.03125,
      "loss_xval": 0.6484375,
      "num_input_tokens_seen": 131316644,
      "step": 2344
    },
    {
      "epoch": 5.22271714922049,
      "grad_norm": 16.071741104125977,
      "learning_rate": 1e-06,
      "loss": 0.5489,
      "num_input_tokens_seen": 131372184,
      "step": 2345
    },
    {
      "epoch": 5.22271714922049,
      "loss": 0.6769649386405945,
      "loss_ce": 0.00020714016864076257,
      "loss_iou": 0.296875,
      "loss_num": 0.0162353515625,
      "loss_xval": 0.67578125,
      "num_input_tokens_seen": 131372184,
      "step": 2345
    },
    {
      "epoch": 5.224944320712694,
      "grad_norm": 18.349136352539062,
      "learning_rate": 1e-06,
      "loss": 0.6446,
      "num_input_tokens_seen": 131430784,
      "step": 2346
    },
    {
      "epoch": 5.224944320712694,
      "loss": 0.6362409591674805,
      "loss_ce": 0.00025463022757321596,
      "loss_iou": 0.28515625,
      "loss_num": 0.0133056640625,
      "loss_xval": 0.63671875,
      "num_input_tokens_seen": 131430784,
      "step": 2346
    },
    {
      "epoch": 5.2271714922049,
      "grad_norm": 18.53408432006836,
      "learning_rate": 1e-06,
      "loss": 0.5966,
      "num_input_tokens_seen": 131488492,
      "step": 2347
    },
    {
      "epoch": 5.2271714922049,
      "loss": 0.5330299139022827,
      "loss_ce": 0.0003150389820802957,
      "loss_iou": 0.220703125,
      "loss_num": 0.0184326171875,
      "loss_xval": 0.53125,
      "num_input_tokens_seen": 131488492,
      "step": 2347
    },
    {
      "epoch": 5.229398663697105,
      "grad_norm": 21.853288650512695,
      "learning_rate": 1e-06,
      "loss": 0.5483,
      "num_input_tokens_seen": 131545288,
      "step": 2348
    },
    {
      "epoch": 5.229398663697105,
      "loss": 0.47664040327072144,
      "loss_ce": 0.00019999593496322632,
      "loss_iou": 0.2080078125,
      "loss_num": 0.01220703125,
      "loss_xval": 0.4765625,
      "num_input_tokens_seen": 131545288,
      "step": 2348
    },
    {
      "epoch": 5.23162583518931,
      "grad_norm": 31.041902542114258,
      "learning_rate": 1e-06,
      "loss": 0.8796,
      "num_input_tokens_seen": 131601784,
      "step": 2349
    },
    {
      "epoch": 5.23162583518931,
      "loss": 0.9917435646057129,
      "loss_ce": 0.0002884728601202369,
      "loss_iou": 0.390625,
      "loss_num": 0.042236328125,
      "loss_xval": 0.9921875,
      "num_input_tokens_seen": 131601784,
      "step": 2349
    },
    {
      "epoch": 5.233853006681515,
      "grad_norm": 56.980098724365234,
      "learning_rate": 1e-06,
      "loss": 0.7844,
      "num_input_tokens_seen": 131658228,
      "step": 2350
    },
    {
      "epoch": 5.233853006681515,
      "loss": 0.7722232341766357,
      "loss_ce": 0.0002506303135305643,
      "loss_iou": 0.33203125,
      "loss_num": 0.021728515625,
      "loss_xval": 0.7734375,
      "num_input_tokens_seen": 131658228,
      "step": 2350
    },
    {
      "epoch": 5.23608017817372,
      "grad_norm": 10.815868377685547,
      "learning_rate": 1e-06,
      "loss": 0.6814,
      "num_input_tokens_seen": 131713244,
      "step": 2351
    },
    {
      "epoch": 5.23608017817372,
      "loss": 0.8227336406707764,
      "loss_ce": 0.00034591203439049423,
      "loss_iou": 0.333984375,
      "loss_num": 0.03125,
      "loss_xval": 0.82421875,
      "num_input_tokens_seen": 131713244,
      "step": 2351
    },
    {
      "epoch": 5.2383073496659245,
      "grad_norm": 34.09395217895508,
      "learning_rate": 1e-06,
      "loss": 0.5397,
      "num_input_tokens_seen": 131771288,
      "step": 2352
    },
    {
      "epoch": 5.2383073496659245,
      "loss": 0.5100328922271729,
      "loss_ce": 0.0002672626869753003,
      "loss_iou": 0.2109375,
      "loss_num": 0.0177001953125,
      "loss_xval": 0.5078125,
      "num_input_tokens_seen": 131771288,
      "step": 2352
    },
    {
      "epoch": 5.240534521158129,
      "grad_norm": 18.458494186401367,
      "learning_rate": 1e-06,
      "loss": 0.7453,
      "num_input_tokens_seen": 131829120,
      "step": 2353
    },
    {
      "epoch": 5.240534521158129,
      "loss": 0.4893495738506317,
      "loss_ce": 0.00021382731210906059,
      "loss_iou": 0.197265625,
      "loss_num": 0.0191650390625,
      "loss_xval": 0.48828125,
      "num_input_tokens_seen": 131829120,
      "step": 2353
    },
    {
      "epoch": 5.242761692650334,
      "grad_norm": 21.484622955322266,
      "learning_rate": 1e-06,
      "loss": 0.8861,
      "num_input_tokens_seen": 131884604,
      "step": 2354
    },
    {
      "epoch": 5.242761692650334,
      "loss": 0.9834548830986023,
      "loss_ce": 0.0003006171027664095,
      "loss_iou": 0.416015625,
      "loss_num": 0.030029296875,
      "loss_xval": 0.984375,
      "num_input_tokens_seen": 131884604,
      "step": 2354
    },
    {
      "epoch": 5.244988864142539,
      "grad_norm": 24.594648361206055,
      "learning_rate": 1e-06,
      "loss": 0.7533,
      "num_input_tokens_seen": 131942424,
      "step": 2355
    },
    {
      "epoch": 5.244988864142539,
      "loss": 0.7038974761962891,
      "loss_ce": 0.00028417640714906156,
      "loss_iou": 0.310546875,
      "loss_num": 0.0162353515625,
      "loss_xval": 0.703125,
      "num_input_tokens_seen": 131942424,
      "step": 2355
    },
    {
      "epoch": 5.247216035634744,
      "grad_norm": 46.79767990112305,
      "learning_rate": 1e-06,
      "loss": 0.6838,
      "num_input_tokens_seen": 131998944,
      "step": 2356
    },
    {
      "epoch": 5.247216035634744,
      "loss": 0.5929880738258362,
      "loss_ce": 0.00021467695478349924,
      "loss_iou": 0.259765625,
      "loss_num": 0.01446533203125,
      "loss_xval": 0.59375,
      "num_input_tokens_seen": 131998944,
      "step": 2356
    },
    {
      "epoch": 5.249443207126949,
      "grad_norm": 18.105363845825195,
      "learning_rate": 1e-06,
      "loss": 0.6012,
      "num_input_tokens_seen": 132055476,
      "step": 2357
    },
    {
      "epoch": 5.249443207126949,
      "loss": 0.45707041025161743,
      "loss_ce": 0.00028330169152468443,
      "loss_iou": 0.189453125,
      "loss_num": 0.015380859375,
      "loss_xval": 0.45703125,
      "num_input_tokens_seen": 132055476,
      "step": 2357
    },
    {
      "epoch": 5.251670378619154,
      "grad_norm": 18.88360595703125,
      "learning_rate": 1e-06,
      "loss": 0.8412,
      "num_input_tokens_seen": 132112904,
      "step": 2358
    },
    {
      "epoch": 5.251670378619154,
      "loss": 0.8789228200912476,
      "loss_ce": 0.0002607441274449229,
      "loss_iou": 0.37109375,
      "loss_num": 0.02685546875,
      "loss_xval": 0.87890625,
      "num_input_tokens_seen": 132112904,
      "step": 2358
    },
    {
      "epoch": 5.2538975501113585,
      "grad_norm": 26.32551383972168,
      "learning_rate": 1e-06,
      "loss": 0.8518,
      "num_input_tokens_seen": 132168780,
      "step": 2359
    },
    {
      "epoch": 5.2538975501113585,
      "loss": 0.9214595556259155,
      "loss_ce": 0.0003170058480463922,
      "loss_iou": 0.42578125,
      "loss_num": 0.01373291015625,
      "loss_xval": 0.921875,
      "num_input_tokens_seen": 132168780,
      "step": 2359
    },
    {
      "epoch": 5.256124721603563,
      "grad_norm": 39.1221809387207,
      "learning_rate": 1e-06,
      "loss": 0.5728,
      "num_input_tokens_seen": 132223368,
      "step": 2360
    },
    {
      "epoch": 5.256124721603563,
      "loss": 0.574551522731781,
      "loss_ce": 0.00021067832130938768,
      "loss_iou": 0.2265625,
      "loss_num": 0.024169921875,
      "loss_xval": 0.57421875,
      "num_input_tokens_seen": 132223368,
      "step": 2360
    },
    {
      "epoch": 5.258351893095768,
      "grad_norm": 24.009830474853516,
      "learning_rate": 1e-06,
      "loss": 0.8335,
      "num_input_tokens_seen": 132279872,
      "step": 2361
    },
    {
      "epoch": 5.258351893095768,
      "loss": 0.8674226999282837,
      "loss_ce": 0.00023520024842582643,
      "loss_iou": 0.388671875,
      "loss_num": 0.017822265625,
      "loss_xval": 0.8671875,
      "num_input_tokens_seen": 132279872,
      "step": 2361
    },
    {
      "epoch": 5.260579064587973,
      "grad_norm": 16.209796905517578,
      "learning_rate": 1e-06,
      "loss": 0.8201,
      "num_input_tokens_seen": 132337356,
      "step": 2362
    },
    {
      "epoch": 5.260579064587973,
      "loss": 1.021734356880188,
      "loss_ce": 0.0004940725630149245,
      "loss_iou": 0.435546875,
      "loss_num": 0.0302734375,
      "loss_xval": 1.0234375,
      "num_input_tokens_seen": 132337356,
      "step": 2362
    },
    {
      "epoch": 5.262806236080178,
      "grad_norm": 21.380468368530273,
      "learning_rate": 1e-06,
      "loss": 0.6083,
      "num_input_tokens_seen": 132389212,
      "step": 2363
    },
    {
      "epoch": 5.262806236080178,
      "loss": 0.5677247643470764,
      "loss_ce": 0.00034196508931927383,
      "loss_iou": 0.203125,
      "loss_num": 0.0322265625,
      "loss_xval": 0.56640625,
      "num_input_tokens_seen": 132389212,
      "step": 2363
    },
    {
      "epoch": 5.265033407572383,
      "grad_norm": 18.931297302246094,
      "learning_rate": 1e-06,
      "loss": 1.0995,
      "num_input_tokens_seen": 132445236,
      "step": 2364
    },
    {
      "epoch": 5.265033407572383,
      "loss": 0.9282673597335815,
      "loss_ce": 0.0002889031311497092,
      "loss_iou": 0.3671875,
      "loss_num": 0.038818359375,
      "loss_xval": 0.9296875,
      "num_input_tokens_seen": 132445236,
      "step": 2364
    },
    {
      "epoch": 5.267260579064588,
      "grad_norm": 51.42475128173828,
      "learning_rate": 1e-06,
      "loss": 0.7182,
      "num_input_tokens_seen": 132499704,
      "step": 2365
    },
    {
      "epoch": 5.267260579064588,
      "loss": 0.6144786477088928,
      "loss_ce": 0.00022083369549363852,
      "loss_iou": 0.2578125,
      "loss_num": 0.0196533203125,
      "loss_xval": 0.61328125,
      "num_input_tokens_seen": 132499704,
      "step": 2365
    },
    {
      "epoch": 5.2694877505567925,
      "grad_norm": 15.084627151489258,
      "learning_rate": 1e-06,
      "loss": 0.5506,
      "num_input_tokens_seen": 132557716,
      "step": 2366
    },
    {
      "epoch": 5.2694877505567925,
      "loss": 0.5173279047012329,
      "loss_ce": 0.00023811672872398049,
      "loss_iou": 0.2255859375,
      "loss_num": 0.01312255859375,
      "loss_xval": 0.515625,
      "num_input_tokens_seen": 132557716,
      "step": 2366
    },
    {
      "epoch": 5.271714922048997,
      "grad_norm": 23.721725463867188,
      "learning_rate": 1e-06,
      "loss": 0.8121,
      "num_input_tokens_seen": 132611180,
      "step": 2367
    },
    {
      "epoch": 5.271714922048997,
      "loss": 0.6462873220443726,
      "loss_ce": 0.00029126679874025285,
      "loss_iou": 0.25390625,
      "loss_num": 0.02734375,
      "loss_xval": 0.64453125,
      "num_input_tokens_seen": 132611180,
      "step": 2367
    },
    {
      "epoch": 5.273942093541203,
      "grad_norm": 15.91121768951416,
      "learning_rate": 1e-06,
      "loss": 0.6395,
      "num_input_tokens_seen": 132667040,
      "step": 2368
    },
    {
      "epoch": 5.273942093541203,
      "loss": 0.6589342951774597,
      "loss_ce": 0.00024289448629133403,
      "loss_iou": 0.279296875,
      "loss_num": 0.02001953125,
      "loss_xval": 0.66015625,
      "num_input_tokens_seen": 132667040,
      "step": 2368
    },
    {
      "epoch": 5.276169265033408,
      "grad_norm": 17.777984619140625,
      "learning_rate": 1e-06,
      "loss": 0.6871,
      "num_input_tokens_seen": 132722572,
      "step": 2369
    },
    {
      "epoch": 5.276169265033408,
      "loss": 0.376442015171051,
      "loss_ce": 0.00022133463062345982,
      "loss_iou": 0.16015625,
      "loss_num": 0.0111083984375,
      "loss_xval": 0.376953125,
      "num_input_tokens_seen": 132722572,
      "step": 2369
    },
    {
      "epoch": 5.278396436525613,
      "grad_norm": 25.825546264648438,
      "learning_rate": 1e-06,
      "loss": 0.9053,
      "num_input_tokens_seen": 132780576,
      "step": 2370
    },
    {
      "epoch": 5.278396436525613,
      "loss": 0.6252577304840088,
      "loss_ce": 0.0002577258856035769,
      "loss_iou": 0.283203125,
      "loss_num": 0.01141357421875,
      "loss_xval": 0.625,
      "num_input_tokens_seen": 132780576,
      "step": 2370
    },
    {
      "epoch": 5.280623608017818,
      "grad_norm": 20.95437240600586,
      "learning_rate": 1e-06,
      "loss": 0.7428,
      "num_input_tokens_seen": 132836940,
      "step": 2371
    },
    {
      "epoch": 5.280623608017818,
      "loss": 0.8691422343254089,
      "loss_ce": 0.00024575780844315886,
      "loss_iou": 0.36328125,
      "loss_num": 0.0284423828125,
      "loss_xval": 0.8671875,
      "num_input_tokens_seen": 132836940,
      "step": 2371
    },
    {
      "epoch": 5.282850779510023,
      "grad_norm": 16.127458572387695,
      "learning_rate": 1e-06,
      "loss": 0.7024,
      "num_input_tokens_seen": 132891444,
      "step": 2372
    },
    {
      "epoch": 5.282850779510023,
      "loss": 0.7618474364280701,
      "loss_ce": 0.0003727960283868015,
      "loss_iou": 0.31640625,
      "loss_num": 0.0255126953125,
      "loss_xval": 0.76171875,
      "num_input_tokens_seen": 132891444,
      "step": 2372
    },
    {
      "epoch": 5.285077951002227,
      "grad_norm": 19.69451141357422,
      "learning_rate": 1e-06,
      "loss": 0.9777,
      "num_input_tokens_seen": 132948012,
      "step": 2373
    },
    {
      "epoch": 5.285077951002227,
      "loss": 0.8320986032485962,
      "loss_ce": 0.00031142972875386477,
      "loss_iou": 0.333984375,
      "loss_num": 0.032470703125,
      "loss_xval": 0.83203125,
      "num_input_tokens_seen": 132948012,
      "step": 2373
    },
    {
      "epoch": 5.287305122494432,
      "grad_norm": 24.581279754638672,
      "learning_rate": 1e-06,
      "loss": 1.0444,
      "num_input_tokens_seen": 133005612,
      "step": 2374
    },
    {
      "epoch": 5.287305122494432,
      "loss": 0.9788609743118286,
      "loss_ce": 0.00034534052247181535,
      "loss_iou": 0.40625,
      "loss_num": 0.032958984375,
      "loss_xval": 0.9765625,
      "num_input_tokens_seen": 133005612,
      "step": 2374
    },
    {
      "epoch": 5.289532293986637,
      "grad_norm": 20.55710220336914,
      "learning_rate": 1e-06,
      "loss": 0.835,
      "num_input_tokens_seen": 133062216,
      "step": 2375
    },
    {
      "epoch": 5.289532293986637,
      "loss": 0.9570013880729675,
      "loss_ce": 0.00033630896359682083,
      "loss_iou": 0.390625,
      "loss_num": 0.03466796875,
      "loss_xval": 0.95703125,
      "num_input_tokens_seen": 133062216,
      "step": 2375
    },
    {
      "epoch": 5.291759465478842,
      "grad_norm": 25.395736694335938,
      "learning_rate": 1e-06,
      "loss": 0.7496,
      "num_input_tokens_seen": 133117644,
      "step": 2376
    },
    {
      "epoch": 5.291759465478842,
      "loss": 0.5810257196426392,
      "loss_ce": 0.00021513670799322426,
      "loss_iou": 0.25,
      "loss_num": 0.0164794921875,
      "loss_xval": 0.58203125,
      "num_input_tokens_seen": 133117644,
      "step": 2376
    },
    {
      "epoch": 5.293986636971047,
      "grad_norm": 58.38376235961914,
      "learning_rate": 1e-06,
      "loss": 0.8065,
      "num_input_tokens_seen": 133176104,
      "step": 2377
    },
    {
      "epoch": 5.293986636971047,
      "loss": 0.6448827981948853,
      "loss_ce": 0.0002294883888680488,
      "loss_iou": 0.28515625,
      "loss_num": 0.014404296875,
      "loss_xval": 0.64453125,
      "num_input_tokens_seen": 133176104,
      "step": 2377
    },
    {
      "epoch": 5.296213808463252,
      "grad_norm": 47.06428909301758,
      "learning_rate": 1e-06,
      "loss": 0.6075,
      "num_input_tokens_seen": 133231892,
      "step": 2378
    },
    {
      "epoch": 5.296213808463252,
      "loss": 0.6965197920799255,
      "loss_ce": 0.00023071446048561484,
      "loss_iou": 0.287109375,
      "loss_num": 0.0245361328125,
      "loss_xval": 0.6953125,
      "num_input_tokens_seen": 133231892,
      "step": 2378
    },
    {
      "epoch": 5.298440979955457,
      "grad_norm": 17.276321411132812,
      "learning_rate": 1e-06,
      "loss": 0.7652,
      "num_input_tokens_seen": 133286408,
      "step": 2379
    },
    {
      "epoch": 5.298440979955457,
      "loss": 0.8292036652565002,
      "loss_ce": 0.00034621963277459145,
      "loss_iou": 0.37109375,
      "loss_num": 0.01708984375,
      "loss_xval": 0.828125,
      "num_input_tokens_seen": 133286408,
      "step": 2379
    },
    {
      "epoch": 5.3006681514476615,
      "grad_norm": 252.65249633789062,
      "learning_rate": 1e-06,
      "loss": 0.7102,
      "num_input_tokens_seen": 133341916,
      "step": 2380
    },
    {
      "epoch": 5.3006681514476615,
      "loss": 0.6878687143325806,
      "loss_ce": 0.00024663194199092686,
      "loss_iou": 0.30859375,
      "loss_num": 0.01397705078125,
      "loss_xval": 0.6875,
      "num_input_tokens_seen": 133341916,
      "step": 2380
    },
    {
      "epoch": 5.302895322939866,
      "grad_norm": 22.686996459960938,
      "learning_rate": 1e-06,
      "loss": 0.7836,
      "num_input_tokens_seen": 133397788,
      "step": 2381
    },
    {
      "epoch": 5.302895322939866,
      "loss": 0.9040675759315491,
      "loss_ce": 0.0002589549985714257,
      "loss_iou": 0.41015625,
      "loss_num": 0.0167236328125,
      "loss_xval": 0.90234375,
      "num_input_tokens_seen": 133397788,
      "step": 2381
    },
    {
      "epoch": 5.305122494432071,
      "grad_norm": 18.216825485229492,
      "learning_rate": 1e-06,
      "loss": 0.7332,
      "num_input_tokens_seen": 133455328,
      "step": 2382
    },
    {
      "epoch": 5.305122494432071,
      "loss": 0.8532497882843018,
      "loss_ce": 0.000222432630835101,
      "loss_iou": 0.36328125,
      "loss_num": 0.02490234375,
      "loss_xval": 0.8515625,
      "num_input_tokens_seen": 133455328,
      "step": 2382
    },
    {
      "epoch": 5.307349665924276,
      "grad_norm": 66.18022918701172,
      "learning_rate": 1e-06,
      "loss": 0.7825,
      "num_input_tokens_seen": 133509936,
      "step": 2383
    },
    {
      "epoch": 5.307349665924276,
      "loss": 0.7538729906082153,
      "loss_ce": 0.0002108832122758031,
      "loss_iou": 0.31640625,
      "loss_num": 0.0242919921875,
      "loss_xval": 0.75390625,
      "num_input_tokens_seen": 133509936,
      "step": 2383
    },
    {
      "epoch": 5.309576837416481,
      "grad_norm": 19.18753433227539,
      "learning_rate": 1e-06,
      "loss": 0.5276,
      "num_input_tokens_seen": 133566504,
      "step": 2384
    },
    {
      "epoch": 5.309576837416481,
      "loss": 0.6280398964881897,
      "loss_ce": 0.00023228148347698152,
      "loss_iou": 0.26953125,
      "loss_num": 0.017822265625,
      "loss_xval": 0.62890625,
      "num_input_tokens_seen": 133566504,
      "step": 2384
    },
    {
      "epoch": 5.311804008908686,
      "grad_norm": 21.568161010742188,
      "learning_rate": 1e-06,
      "loss": 0.6863,
      "num_input_tokens_seen": 133624932,
      "step": 2385
    },
    {
      "epoch": 5.311804008908686,
      "loss": 0.569631814956665,
      "loss_ce": 0.00029586919117718935,
      "loss_iou": 0.251953125,
      "loss_num": 0.01275634765625,
      "loss_xval": 0.5703125,
      "num_input_tokens_seen": 133624932,
      "step": 2385
    },
    {
      "epoch": 5.314031180400891,
      "grad_norm": 19.216602325439453,
      "learning_rate": 1e-06,
      "loss": 0.5267,
      "num_input_tokens_seen": 133679920,
      "step": 2386
    },
    {
      "epoch": 5.314031180400891,
      "loss": 0.5746643543243408,
      "loss_ce": 0.00020149891497567296,
      "loss_iou": 0.2412109375,
      "loss_num": 0.0185546875,
      "loss_xval": 0.57421875,
      "num_input_tokens_seen": 133679920,
      "step": 2386
    },
    {
      "epoch": 5.3162583518930955,
      "grad_norm": 25.528783798217773,
      "learning_rate": 1e-06,
      "loss": 0.8468,
      "num_input_tokens_seen": 133735508,
      "step": 2387
    },
    {
      "epoch": 5.3162583518930955,
      "loss": 0.7472808361053467,
      "loss_ce": 0.00021054709213785827,
      "loss_iou": 0.318359375,
      "loss_num": 0.0216064453125,
      "loss_xval": 0.74609375,
      "num_input_tokens_seen": 133735508,
      "step": 2387
    },
    {
      "epoch": 5.3184855233853,
      "grad_norm": 19.144241333007812,
      "learning_rate": 1e-06,
      "loss": 0.7418,
      "num_input_tokens_seen": 133788248,
      "step": 2388
    },
    {
      "epoch": 5.3184855233853,
      "loss": 0.7324228286743164,
      "loss_ce": 0.0002451083273626864,
      "loss_iou": 0.3203125,
      "loss_num": 0.0186767578125,
      "loss_xval": 0.73046875,
      "num_input_tokens_seen": 133788248,
      "step": 2388
    },
    {
      "epoch": 5.320712694877505,
      "grad_norm": 26.32198143005371,
      "learning_rate": 1e-06,
      "loss": 0.6614,
      "num_input_tokens_seen": 133843512,
      "step": 2389
    },
    {
      "epoch": 5.320712694877505,
      "loss": 0.6170359253883362,
      "loss_ce": 0.00021462509175762534,
      "loss_iou": 0.26171875,
      "loss_num": 0.01904296875,
      "loss_xval": 0.6171875,
      "num_input_tokens_seen": 133843512,
      "step": 2389
    },
    {
      "epoch": 5.32293986636971,
      "grad_norm": 17.643627166748047,
      "learning_rate": 1e-06,
      "loss": 0.6772,
      "num_input_tokens_seen": 133898856,
      "step": 2390
    },
    {
      "epoch": 5.32293986636971,
      "loss": 0.6501243710517883,
      "loss_ce": 0.00022201667889021337,
      "loss_iou": 0.27734375,
      "loss_num": 0.01904296875,
      "loss_xval": 0.6484375,
      "num_input_tokens_seen": 133898856,
      "step": 2390
    },
    {
      "epoch": 5.325167037861915,
      "grad_norm": 31.93773651123047,
      "learning_rate": 1e-06,
      "loss": 0.7977,
      "num_input_tokens_seen": 133956748,
      "step": 2391
    },
    {
      "epoch": 5.325167037861915,
      "loss": 0.6381719708442688,
      "loss_ce": 0.00023252921528182924,
      "loss_iou": 0.28125,
      "loss_num": 0.01531982421875,
      "loss_xval": 0.63671875,
      "num_input_tokens_seen": 133956748,
      "step": 2391
    },
    {
      "epoch": 5.327394209354121,
      "grad_norm": 19.902048110961914,
      "learning_rate": 1e-06,
      "loss": 0.6616,
      "num_input_tokens_seen": 134011496,
      "step": 2392
    },
    {
      "epoch": 5.327394209354121,
      "loss": 0.6387029886245728,
      "loss_ce": 0.0007635352667421103,
      "loss_iou": 0.2890625,
      "loss_num": 0.01226806640625,
      "loss_xval": 0.63671875,
      "num_input_tokens_seen": 134011496,
      "step": 2392
    },
    {
      "epoch": 5.3296213808463255,
      "grad_norm": 16.311479568481445,
      "learning_rate": 1e-06,
      "loss": 0.4794,
      "num_input_tokens_seen": 134069204,
      "step": 2393
    },
    {
      "epoch": 5.3296213808463255,
      "loss": 0.4752798080444336,
      "loss_ce": 0.00018213533621747047,
      "loss_iou": 0.19921875,
      "loss_num": 0.01544189453125,
      "loss_xval": 0.474609375,
      "num_input_tokens_seen": 134069204,
      "step": 2393
    },
    {
      "epoch": 5.33184855233853,
      "grad_norm": 39.13882827758789,
      "learning_rate": 1e-06,
      "loss": 0.6946,
      "num_input_tokens_seen": 134126304,
      "step": 2394
    },
    {
      "epoch": 5.33184855233853,
      "loss": 0.837151288986206,
      "loss_ce": 0.00023718301963526756,
      "loss_iou": 0.34375,
      "loss_num": 0.02978515625,
      "loss_xval": 0.8359375,
      "num_input_tokens_seen": 134126304,
      "step": 2394
    },
    {
      "epoch": 5.334075723830735,
      "grad_norm": 20.73467445373535,
      "learning_rate": 1e-06,
      "loss": 0.7741,
      "num_input_tokens_seen": 134183476,
      "step": 2395
    },
    {
      "epoch": 5.334075723830735,
      "loss": 0.7032053470611572,
      "loss_ce": 0.00032445762190036476,
      "loss_iou": 0.29296875,
      "loss_num": 0.023681640625,
      "loss_xval": 0.703125,
      "num_input_tokens_seen": 134183476,
      "step": 2395
    },
    {
      "epoch": 5.33630289532294,
      "grad_norm": 25.634654998779297,
      "learning_rate": 1e-06,
      "loss": 0.564,
      "num_input_tokens_seen": 134240096,
      "step": 2396
    },
    {
      "epoch": 5.33630289532294,
      "loss": 0.699744701385498,
      "loss_ce": 0.0002818358479999006,
      "loss_iou": 0.31640625,
      "loss_num": 0.01373291015625,
      "loss_xval": 0.69921875,
      "num_input_tokens_seen": 134240096,
      "step": 2396
    },
    {
      "epoch": 5.338530066815145,
      "grad_norm": 19.363285064697266,
      "learning_rate": 1e-06,
      "loss": 0.4604,
      "num_input_tokens_seen": 134296184,
      "step": 2397
    },
    {
      "epoch": 5.338530066815145,
      "loss": 0.5098617076873779,
      "loss_ce": 0.00034024479100480676,
      "loss_iou": 0.224609375,
      "loss_num": 0.01226806640625,
      "loss_xval": 0.5078125,
      "num_input_tokens_seen": 134296184,
      "step": 2397
    },
    {
      "epoch": 5.34075723830735,
      "grad_norm": 31.034711837768555,
      "learning_rate": 1e-06,
      "loss": 0.8775,
      "num_input_tokens_seen": 134347520,
      "step": 2398
    },
    {
      "epoch": 5.34075723830735,
      "loss": 0.730765163898468,
      "loss_ce": 0.00029640039429068565,
      "loss_iou": 0.322265625,
      "loss_num": 0.01708984375,
      "loss_xval": 0.73046875,
      "num_input_tokens_seen": 134347520,
      "step": 2398
    },
    {
      "epoch": 5.342984409799555,
      "grad_norm": 21.829448699951172,
      "learning_rate": 1e-06,
      "loss": 1.003,
      "num_input_tokens_seen": 134402984,
      "step": 2399
    },
    {
      "epoch": 5.342984409799555,
      "loss": 1.1158628463745117,
      "loss_ce": 0.0006285187555477023,
      "loss_iou": 0.462890625,
      "loss_num": 0.0380859375,
      "loss_xval": 1.1171875,
      "num_input_tokens_seen": 134402984,
      "step": 2399
    },
    {
      "epoch": 5.3452115812917596,
      "grad_norm": 21.801376342773438,
      "learning_rate": 1e-06,
      "loss": 0.8284,
      "num_input_tokens_seen": 134457812,
      "step": 2400
    },
    {
      "epoch": 5.3452115812917596,
      "loss": 0.8780108094215393,
      "loss_ce": 0.00032530241878703237,
      "loss_iou": 0.380859375,
      "loss_num": 0.02294921875,
      "loss_xval": 0.87890625,
      "num_input_tokens_seen": 134457812,
      "step": 2400
    },
    {
      "epoch": 5.347438752783964,
      "grad_norm": 33.11177444458008,
      "learning_rate": 1e-06,
      "loss": 0.8474,
      "num_input_tokens_seen": 134512096,
      "step": 2401
    },
    {
      "epoch": 5.347438752783964,
      "loss": 0.9323829412460327,
      "loss_ce": 0.0002540295827202499,
      "loss_iou": 0.41015625,
      "loss_num": 0.0223388671875,
      "loss_xval": 0.93359375,
      "num_input_tokens_seen": 134512096,
      "step": 2401
    },
    {
      "epoch": 5.349665924276169,
      "grad_norm": 19.453571319580078,
      "learning_rate": 1e-06,
      "loss": 0.7402,
      "num_input_tokens_seen": 134565996,
      "step": 2402
    },
    {
      "epoch": 5.349665924276169,
      "loss": 0.7348358631134033,
      "loss_ce": 0.00021668968838639557,
      "loss_iou": 0.2890625,
      "loss_num": 0.031494140625,
      "loss_xval": 0.734375,
      "num_input_tokens_seen": 134565996,
      "step": 2402
    },
    {
      "epoch": 5.351893095768374,
      "grad_norm": 26.344636917114258,
      "learning_rate": 1e-06,
      "loss": 1.0379,
      "num_input_tokens_seen": 134623536,
      "step": 2403
    },
    {
      "epoch": 5.351893095768374,
      "loss": 0.9678047895431519,
      "loss_ce": 0.0002755335299298167,
      "loss_iou": 0.416015625,
      "loss_num": 0.026611328125,
      "loss_xval": 0.96875,
      "num_input_tokens_seen": 134623536,
      "step": 2403
    },
    {
      "epoch": 5.354120267260579,
      "grad_norm": 24.26504898071289,
      "learning_rate": 1e-06,
      "loss": 0.7085,
      "num_input_tokens_seen": 134676664,
      "step": 2404
    },
    {
      "epoch": 5.354120267260579,
      "loss": 0.7473124265670776,
      "loss_ce": 0.00024206144735217094,
      "loss_iou": 0.283203125,
      "loss_num": 0.0361328125,
      "loss_xval": 0.74609375,
      "num_input_tokens_seen": 134676664,
      "step": 2404
    },
    {
      "epoch": 5.356347438752784,
      "grad_norm": 37.98905563354492,
      "learning_rate": 1e-06,
      "loss": 0.6818,
      "num_input_tokens_seen": 134734292,
      "step": 2405
    },
    {
      "epoch": 5.356347438752784,
      "loss": 0.8488801717758179,
      "loss_ce": 0.00024738311185501516,
      "loss_iou": 0.34765625,
      "loss_num": 0.0302734375,
      "loss_xval": 0.84765625,
      "num_input_tokens_seen": 134734292,
      "step": 2405
    },
    {
      "epoch": 5.358574610244989,
      "grad_norm": 110.85530090332031,
      "learning_rate": 1e-06,
      "loss": 0.5897,
      "num_input_tokens_seen": 134791600,
      "step": 2406
    },
    {
      "epoch": 5.358574610244989,
      "loss": 0.5577570796012878,
      "loss_ce": 0.0002619822043925524,
      "loss_iou": 0.25390625,
      "loss_num": 0.00970458984375,
      "loss_xval": 0.55859375,
      "num_input_tokens_seen": 134791600,
      "step": 2406
    },
    {
      "epoch": 5.360801781737194,
      "grad_norm": 25.807218551635742,
      "learning_rate": 1e-06,
      "loss": 0.8237,
      "num_input_tokens_seen": 134847120,
      "step": 2407
    },
    {
      "epoch": 5.360801781737194,
      "loss": 0.7333089709281921,
      "loss_ce": 0.0002767038531601429,
      "loss_iou": 0.29296875,
      "loss_num": 0.0294189453125,
      "loss_xval": 0.734375,
      "num_input_tokens_seen": 134847120,
      "step": 2407
    },
    {
      "epoch": 5.363028953229398,
      "grad_norm": 23.73634147644043,
      "learning_rate": 1e-06,
      "loss": 0.7855,
      "num_input_tokens_seen": 134901488,
      "step": 2408
    },
    {
      "epoch": 5.363028953229398,
      "loss": 0.6962625980377197,
      "loss_ce": 0.000339773076120764,
      "loss_iou": 0.2890625,
      "loss_num": 0.0234375,
      "loss_xval": 0.6953125,
      "num_input_tokens_seen": 134901488,
      "step": 2408
    },
    {
      "epoch": 5.365256124721603,
      "grad_norm": 23.408971786499023,
      "learning_rate": 1e-06,
      "loss": 0.7951,
      "num_input_tokens_seen": 134956808,
      "step": 2409
    },
    {
      "epoch": 5.365256124721603,
      "loss": 0.8180850744247437,
      "loss_ce": 0.00021401085541583598,
      "loss_iou": 0.349609375,
      "loss_num": 0.0240478515625,
      "loss_xval": 0.81640625,
      "num_input_tokens_seen": 134956808,
      "step": 2409
    },
    {
      "epoch": 5.367483296213808,
      "grad_norm": 24.673824310302734,
      "learning_rate": 1e-06,
      "loss": 0.6417,
      "num_input_tokens_seen": 135013228,
      "step": 2410
    },
    {
      "epoch": 5.367483296213808,
      "loss": 0.3933447301387787,
      "loss_ce": 0.0002172911772504449,
      "loss_iou": 0.162109375,
      "loss_num": 0.01373291015625,
      "loss_xval": 0.392578125,
      "num_input_tokens_seen": 135013228,
      "step": 2410
    },
    {
      "epoch": 5.369710467706013,
      "grad_norm": 19.17323875427246,
      "learning_rate": 1e-06,
      "loss": 0.6747,
      "num_input_tokens_seen": 135070756,
      "step": 2411
    },
    {
      "epoch": 5.369710467706013,
      "loss": 0.7497762441635132,
      "loss_ce": 0.00026456580962985754,
      "loss_iou": 0.326171875,
      "loss_num": 0.0189208984375,
      "loss_xval": 0.75,
      "num_input_tokens_seen": 135070756,
      "step": 2411
    },
    {
      "epoch": 5.371937639198218,
      "grad_norm": 21.6716365814209,
      "learning_rate": 1e-06,
      "loss": 0.7877,
      "num_input_tokens_seen": 135123704,
      "step": 2412
    },
    {
      "epoch": 5.371937639198218,
      "loss": 0.9468029737472534,
      "loss_ce": 0.0002697420713957399,
      "loss_iou": 0.40234375,
      "loss_num": 0.028076171875,
      "loss_xval": 0.9453125,
      "num_input_tokens_seen": 135123704,
      "step": 2412
    },
    {
      "epoch": 5.374164810690424,
      "grad_norm": 28.3837833404541,
      "learning_rate": 1e-06,
      "loss": 0.8981,
      "num_input_tokens_seen": 135177764,
      "step": 2413
    },
    {
      "epoch": 5.374164810690424,
      "loss": 1.0344269275665283,
      "loss_ce": 0.00024724419927224517,
      "loss_iou": 0.421875,
      "loss_num": 0.0380859375,
      "loss_xval": 1.03125,
      "num_input_tokens_seen": 135177764,
      "step": 2413
    },
    {
      "epoch": 5.3763919821826285,
      "grad_norm": 38.46763229370117,
      "learning_rate": 1e-06,
      "loss": 0.7782,
      "num_input_tokens_seen": 135233172,
      "step": 2414
    },
    {
      "epoch": 5.3763919821826285,
      "loss": 0.8738818764686584,
      "loss_ce": 0.00034671538742259145,
      "loss_iou": 0.37890625,
      "loss_num": 0.0228271484375,
      "loss_xval": 0.875,
      "num_input_tokens_seen": 135233172,
      "step": 2414
    },
    {
      "epoch": 5.378619153674833,
      "grad_norm": 20.032617568969727,
      "learning_rate": 1e-06,
      "loss": 0.661,
      "num_input_tokens_seen": 135287744,
      "step": 2415
    },
    {
      "epoch": 5.378619153674833,
      "loss": 0.6660585403442383,
      "loss_ce": 0.00040915622957982123,
      "loss_iou": 0.2890625,
      "loss_num": 0.017822265625,
      "loss_xval": 0.6640625,
      "num_input_tokens_seen": 135287744,
      "step": 2415
    },
    {
      "epoch": 5.380846325167038,
      "grad_norm": 33.80507278442383,
      "learning_rate": 1e-06,
      "loss": 0.8263,
      "num_input_tokens_seen": 135344300,
      "step": 2416
    },
    {
      "epoch": 5.380846325167038,
      "loss": 0.8908694386482239,
      "loss_ce": 0.0002443883859086782,
      "loss_iou": 0.39453125,
      "loss_num": 0.0206298828125,
      "loss_xval": 0.890625,
      "num_input_tokens_seen": 135344300,
      "step": 2416
    },
    {
      "epoch": 5.383073496659243,
      "grad_norm": 18.646997451782227,
      "learning_rate": 1e-06,
      "loss": 0.6062,
      "num_input_tokens_seen": 135402736,
      "step": 2417
    },
    {
      "epoch": 5.383073496659243,
      "loss": 0.6154400110244751,
      "loss_ce": 0.0002056759112747386,
      "loss_iou": 0.279296875,
      "loss_num": 0.011474609375,
      "loss_xval": 0.6171875,
      "num_input_tokens_seen": 135402736,
      "step": 2417
    },
    {
      "epoch": 5.385300668151448,
      "grad_norm": 27.358457565307617,
      "learning_rate": 1e-06,
      "loss": 0.7446,
      "num_input_tokens_seen": 135457252,
      "step": 2418
    },
    {
      "epoch": 5.385300668151448,
      "loss": 0.8736233711242676,
      "loss_ce": 0.0003323230193927884,
      "loss_iou": 0.38671875,
      "loss_num": 0.0201416015625,
      "loss_xval": 0.875,
      "num_input_tokens_seen": 135457252,
      "step": 2418
    },
    {
      "epoch": 5.387527839643653,
      "grad_norm": 37.07065200805664,
      "learning_rate": 1e-06,
      "loss": 0.9104,
      "num_input_tokens_seen": 135514764,
      "step": 2419
    },
    {
      "epoch": 5.387527839643653,
      "loss": 0.8482790589332581,
      "loss_ce": 0.000378611555788666,
      "loss_iou": 0.33984375,
      "loss_num": 0.033447265625,
      "loss_xval": 0.84765625,
      "num_input_tokens_seen": 135514764,
      "step": 2419
    },
    {
      "epoch": 5.389755011135858,
      "grad_norm": 15.80694580078125,
      "learning_rate": 1e-06,
      "loss": 0.7015,
      "num_input_tokens_seen": 135571288,
      "step": 2420
    },
    {
      "epoch": 5.389755011135858,
      "loss": 0.6910269856452942,
      "loss_ce": 0.00023109573521651328,
      "loss_iou": 0.306640625,
      "loss_num": 0.015625,
      "loss_xval": 0.69140625,
      "num_input_tokens_seen": 135571288,
      "step": 2420
    },
    {
      "epoch": 5.3919821826280625,
      "grad_norm": 16.644014358520508,
      "learning_rate": 1e-06,
      "loss": 0.5917,
      "num_input_tokens_seen": 135627180,
      "step": 2421
    },
    {
      "epoch": 5.3919821826280625,
      "loss": 0.6370264291763306,
      "loss_ce": 0.00030768115539103746,
      "loss_iou": 0.244140625,
      "loss_num": 0.0299072265625,
      "loss_xval": 0.63671875,
      "num_input_tokens_seen": 135627180,
      "step": 2421
    },
    {
      "epoch": 5.394209354120267,
      "grad_norm": 23.295394897460938,
      "learning_rate": 1e-06,
      "loss": 0.851,
      "num_input_tokens_seen": 135681732,
      "step": 2422
    },
    {
      "epoch": 5.394209354120267,
      "loss": 0.890271782875061,
      "loss_ce": 0.0003792433417402208,
      "loss_iou": 0.37109375,
      "loss_num": 0.0296630859375,
      "loss_xval": 0.890625,
      "num_input_tokens_seen": 135681732,
      "step": 2422
    },
    {
      "epoch": 5.396436525612472,
      "grad_norm": 17.34937286376953,
      "learning_rate": 1e-06,
      "loss": 0.7305,
      "num_input_tokens_seen": 135739488,
      "step": 2423
    },
    {
      "epoch": 5.396436525612472,
      "loss": 0.7664504051208496,
      "loss_ce": 0.00033708903356455266,
      "loss_iou": 0.314453125,
      "loss_num": 0.0272216796875,
      "loss_xval": 0.765625,
      "num_input_tokens_seen": 135739488,
      "step": 2423
    },
    {
      "epoch": 5.398663697104677,
      "grad_norm": 17.406497955322266,
      "learning_rate": 1e-06,
      "loss": 0.5762,
      "num_input_tokens_seen": 135793620,
      "step": 2424
    },
    {
      "epoch": 5.398663697104677,
      "loss": 0.5314889550209045,
      "loss_ce": 0.00023894087644293904,
      "loss_iou": 0.2158203125,
      "loss_num": 0.019775390625,
      "loss_xval": 0.53125,
      "num_input_tokens_seen": 135793620,
      "step": 2424
    },
    {
      "epoch": 5.400890868596882,
      "grad_norm": 15.964725494384766,
      "learning_rate": 1e-06,
      "loss": 0.6212,
      "num_input_tokens_seen": 135850120,
      "step": 2425
    },
    {
      "epoch": 5.400890868596882,
      "loss": 0.5310900211334229,
      "loss_ce": 0.0002062540443148464,
      "loss_iou": 0.2275390625,
      "loss_num": 0.01507568359375,
      "loss_xval": 0.53125,
      "num_input_tokens_seen": 135850120,
      "step": 2425
    },
    {
      "epoch": 5.403118040089087,
      "grad_norm": 18.38140296936035,
      "learning_rate": 1e-06,
      "loss": 0.5365,
      "num_input_tokens_seen": 135901984,
      "step": 2426
    },
    {
      "epoch": 5.403118040089087,
      "loss": 0.48569074273109436,
      "loss_ce": 0.0002171028172597289,
      "loss_iou": 0.21484375,
      "loss_num": 0.01123046875,
      "loss_xval": 0.486328125,
      "num_input_tokens_seen": 135901984,
      "step": 2426
    },
    {
      "epoch": 5.405345211581292,
      "grad_norm": 17.759723663330078,
      "learning_rate": 1e-06,
      "loss": 0.6546,
      "num_input_tokens_seen": 135956660,
      "step": 2427
    },
    {
      "epoch": 5.405345211581292,
      "loss": 0.6573410034179688,
      "loss_ce": 0.00035854033194482327,
      "loss_iou": 0.28125,
      "loss_num": 0.0186767578125,
      "loss_xval": 0.65625,
      "num_input_tokens_seen": 135956660,
      "step": 2427
    },
    {
      "epoch": 5.4075723830734965,
      "grad_norm": 18.03878402709961,
      "learning_rate": 1e-06,
      "loss": 0.5775,
      "num_input_tokens_seen": 136013048,
      "step": 2428
    },
    {
      "epoch": 5.4075723830734965,
      "loss": 0.592266857624054,
      "loss_ce": 0.00022585978149436414,
      "loss_iou": 0.2451171875,
      "loss_num": 0.0201416015625,
      "loss_xval": 0.59375,
      "num_input_tokens_seen": 136013048,
      "step": 2428
    },
    {
      "epoch": 5.409799554565701,
      "grad_norm": 30.479446411132812,
      "learning_rate": 1e-06,
      "loss": 0.7551,
      "num_input_tokens_seen": 136068760,
      "step": 2429
    },
    {
      "epoch": 5.409799554565701,
      "loss": 1.0719916820526123,
      "loss_ce": 0.00033644368522800505,
      "loss_iou": 0.46875,
      "loss_num": 0.0263671875,
      "loss_xval": 1.0703125,
      "num_input_tokens_seen": 136068760,
      "step": 2429
    },
    {
      "epoch": 5.412026726057906,
      "grad_norm": 22.04293441772461,
      "learning_rate": 1e-06,
      "loss": 0.7282,
      "num_input_tokens_seen": 136122268,
      "step": 2430
    },
    {
      "epoch": 5.412026726057906,
      "loss": 0.7684208750724792,
      "loss_ce": 0.00023237511049956083,
      "loss_iou": 0.306640625,
      "loss_num": 0.0306396484375,
      "loss_xval": 0.76953125,
      "num_input_tokens_seen": 136122268,
      "step": 2430
    },
    {
      "epoch": 5.414253897550111,
      "grad_norm": 49.4771614074707,
      "learning_rate": 1e-06,
      "loss": 0.5288,
      "num_input_tokens_seen": 136178720,
      "step": 2431
    },
    {
      "epoch": 5.414253897550111,
      "loss": 0.4244188964366913,
      "loss_ce": 0.00022456918668467551,
      "loss_iou": 0.1728515625,
      "loss_num": 0.01556396484375,
      "loss_xval": 0.423828125,
      "num_input_tokens_seen": 136178720,
      "step": 2431
    },
    {
      "epoch": 5.416481069042316,
      "grad_norm": 31.579322814941406,
      "learning_rate": 1e-06,
      "loss": 0.5512,
      "num_input_tokens_seen": 136237672,
      "step": 2432
    },
    {
      "epoch": 5.416481069042316,
      "loss": 0.5803066492080688,
      "loss_ce": 0.00022852106485515833,
      "loss_iou": 0.2431640625,
      "loss_num": 0.018798828125,
      "loss_xval": 0.578125,
      "num_input_tokens_seen": 136237672,
      "step": 2432
    },
    {
      "epoch": 5.418708240534521,
      "grad_norm": 30.84029197692871,
      "learning_rate": 1e-06,
      "loss": 0.574,
      "num_input_tokens_seen": 136293760,
      "step": 2433
    },
    {
      "epoch": 5.418708240534521,
      "loss": 0.6513895988464355,
      "loss_ce": 0.00026654996327124536,
      "loss_iou": 0.265625,
      "loss_num": 0.02392578125,
      "loss_xval": 0.65234375,
      "num_input_tokens_seen": 136293760,
      "step": 2433
    },
    {
      "epoch": 5.420935412026726,
      "grad_norm": 16.020021438598633,
      "learning_rate": 1e-06,
      "loss": 0.5865,
      "num_input_tokens_seen": 136350844,
      "step": 2434
    },
    {
      "epoch": 5.420935412026726,
      "loss": 0.5578700304031372,
      "loss_ce": 0.0002528943296056241,
      "loss_iou": 0.2353515625,
      "loss_num": 0.017333984375,
      "loss_xval": 0.55859375,
      "num_input_tokens_seen": 136350844,
      "step": 2434
    },
    {
      "epoch": 5.4231625835189305,
      "grad_norm": 22.808626174926758,
      "learning_rate": 1e-06,
      "loss": 0.9401,
      "num_input_tokens_seen": 136404744,
      "step": 2435
    },
    {
      "epoch": 5.4231625835189305,
      "loss": 0.8737362027168274,
      "loss_ce": 0.00020110802142880857,
      "loss_iou": 0.359375,
      "loss_num": 0.03076171875,
      "loss_xval": 0.875,
      "num_input_tokens_seen": 136404744,
      "step": 2435
    },
    {
      "epoch": 5.425389755011135,
      "grad_norm": 25.185213088989258,
      "learning_rate": 1e-06,
      "loss": 0.5613,
      "num_input_tokens_seen": 136462516,
      "step": 2436
    },
    {
      "epoch": 5.425389755011135,
      "loss": 0.5026748776435852,
      "loss_ce": 0.00023346173111349344,
      "loss_iou": 0.22265625,
      "loss_num": 0.0113525390625,
      "loss_xval": 0.50390625,
      "num_input_tokens_seen": 136462516,
      "step": 2436
    },
    {
      "epoch": 5.427616926503341,
      "grad_norm": 16.586254119873047,
      "learning_rate": 1e-06,
      "loss": 0.7677,
      "num_input_tokens_seen": 136520164,
      "step": 2437
    },
    {
      "epoch": 5.427616926503341,
      "loss": 0.6672698259353638,
      "loss_ce": 0.0002775907050818205,
      "loss_iou": 0.275390625,
      "loss_num": 0.0230712890625,
      "loss_xval": 0.66796875,
      "num_input_tokens_seen": 136520164,
      "step": 2437
    },
    {
      "epoch": 5.429844097995546,
      "grad_norm": 15.319040298461914,
      "learning_rate": 1e-06,
      "loss": 0.5602,
      "num_input_tokens_seen": 136579040,
      "step": 2438
    },
    {
      "epoch": 5.429844097995546,
      "loss": 0.39814403653144836,
      "loss_ce": 0.00019482253992464393,
      "loss_iou": 0.16015625,
      "loss_num": 0.0155029296875,
      "loss_xval": 0.3984375,
      "num_input_tokens_seen": 136579040,
      "step": 2438
    },
    {
      "epoch": 5.432071269487751,
      "grad_norm": 21.42112159729004,
      "learning_rate": 1e-06,
      "loss": 0.9151,
      "num_input_tokens_seen": 136635020,
      "step": 2439
    },
    {
      "epoch": 5.432071269487751,
      "loss": 0.9438084363937378,
      "loss_ce": 0.0003270409069955349,
      "loss_iou": 0.392578125,
      "loss_num": 0.0322265625,
      "loss_xval": 0.9453125,
      "num_input_tokens_seen": 136635020,
      "step": 2439
    },
    {
      "epoch": 5.434298440979956,
      "grad_norm": 19.967939376831055,
      "learning_rate": 1e-06,
      "loss": 0.9586,
      "num_input_tokens_seen": 136693516,
      "step": 2440
    },
    {
      "epoch": 5.434298440979956,
      "loss": 0.9183446168899536,
      "loss_ce": 0.0003758435195777565,
      "loss_iou": 0.390625,
      "loss_num": 0.027587890625,
      "loss_xval": 0.91796875,
      "num_input_tokens_seen": 136693516,
      "step": 2440
    },
    {
      "epoch": 5.436525612472161,
      "grad_norm": 21.792882919311523,
      "learning_rate": 1e-06,
      "loss": 0.7106,
      "num_input_tokens_seen": 136750260,
      "step": 2441
    },
    {
      "epoch": 5.436525612472161,
      "loss": 0.6200574636459351,
      "loss_ce": 0.00018441499560140073,
      "loss_iou": 0.2578125,
      "loss_num": 0.0206298828125,
      "loss_xval": 0.62109375,
      "num_input_tokens_seen": 136750260,
      "step": 2441
    },
    {
      "epoch": 5.4387527839643655,
      "grad_norm": 18.052919387817383,
      "learning_rate": 1e-06,
      "loss": 0.5861,
      "num_input_tokens_seen": 136804772,
      "step": 2442
    },
    {
      "epoch": 5.4387527839643655,
      "loss": 0.48152902722358704,
      "loss_ce": 0.000449933490017429,
      "loss_iou": 0.2021484375,
      "loss_num": 0.015625,
      "loss_xval": 0.48046875,
      "num_input_tokens_seen": 136804772,
      "step": 2442
    },
    {
      "epoch": 5.44097995545657,
      "grad_norm": 20.881345748901367,
      "learning_rate": 1e-06,
      "loss": 0.8096,
      "num_input_tokens_seen": 136861740,
      "step": 2443
    },
    {
      "epoch": 5.44097995545657,
      "loss": 0.6257365942001343,
      "loss_ce": 0.0002483331772964448,
      "loss_iou": 0.248046875,
      "loss_num": 0.02587890625,
      "loss_xval": 0.625,
      "num_input_tokens_seen": 136861740,
      "step": 2443
    },
    {
      "epoch": 5.443207126948775,
      "grad_norm": 17.746784210205078,
      "learning_rate": 1e-06,
      "loss": 0.7268,
      "num_input_tokens_seen": 136920452,
      "step": 2444
    },
    {
      "epoch": 5.443207126948775,
      "loss": 0.8379260897636414,
      "loss_ce": 0.00027960725128650665,
      "loss_iou": 0.349609375,
      "loss_num": 0.02783203125,
      "loss_xval": 0.8359375,
      "num_input_tokens_seen": 136920452,
      "step": 2444
    },
    {
      "epoch": 5.44543429844098,
      "grad_norm": 17.058738708496094,
      "learning_rate": 1e-06,
      "loss": 0.8293,
      "num_input_tokens_seen": 136980184,
      "step": 2445
    },
    {
      "epoch": 5.44543429844098,
      "loss": 0.8342658281326294,
      "loss_ce": 0.00028141395887359977,
      "loss_iou": 0.357421875,
      "loss_num": 0.024169921875,
      "loss_xval": 0.8359375,
      "num_input_tokens_seen": 136980184,
      "step": 2445
    },
    {
      "epoch": 5.447661469933185,
      "grad_norm": 14.320577621459961,
      "learning_rate": 1e-06,
      "loss": 0.6234,
      "num_input_tokens_seen": 137037052,
      "step": 2446
    },
    {
      "epoch": 5.447661469933185,
      "loss": 0.6837434768676758,
      "loss_ce": 0.00027177410083822906,
      "loss_iou": 0.2734375,
      "loss_num": 0.027587890625,
      "loss_xval": 0.68359375,
      "num_input_tokens_seen": 137037052,
      "step": 2446
    },
    {
      "epoch": 5.44988864142539,
      "grad_norm": 19.701732635498047,
      "learning_rate": 1e-06,
      "loss": 0.6273,
      "num_input_tokens_seen": 137094316,
      "step": 2447
    },
    {
      "epoch": 5.44988864142539,
      "loss": 0.5671048164367676,
      "loss_ce": 0.00021026638569310308,
      "loss_iou": 0.2578125,
      "loss_num": 0.01031494140625,
      "loss_xval": 0.56640625,
      "num_input_tokens_seen": 137094316,
      "step": 2447
    },
    {
      "epoch": 5.452115812917595,
      "grad_norm": 47.633155822753906,
      "learning_rate": 1e-06,
      "loss": 0.7027,
      "num_input_tokens_seen": 137152472,
      "step": 2448
    },
    {
      "epoch": 5.452115812917595,
      "loss": 0.7645977735519409,
      "loss_ce": 0.00019351180526427925,
      "loss_iou": 0.33984375,
      "loss_num": 0.0172119140625,
      "loss_xval": 0.765625,
      "num_input_tokens_seen": 137152472,
      "step": 2448
    },
    {
      "epoch": 5.4543429844097995,
      "grad_norm": 18.31052017211914,
      "learning_rate": 1e-06,
      "loss": 0.8306,
      "num_input_tokens_seen": 137208136,
      "step": 2449
    },
    {
      "epoch": 5.4543429844097995,
      "loss": 0.5993661284446716,
      "loss_ce": 0.0002450407773721963,
      "loss_iou": 0.263671875,
      "loss_num": 0.01422119140625,
      "loss_xval": 0.59765625,
      "num_input_tokens_seen": 137208136,
      "step": 2449
    },
    {
      "epoch": 5.456570155902004,
      "grad_norm": 19.636415481567383,
      "learning_rate": 1e-06,
      "loss": 0.9006,
      "num_input_tokens_seen": 137262276,
      "step": 2450
    },
    {
      "epoch": 5.456570155902004,
      "loss": 0.8564736843109131,
      "loss_ce": 0.0002724617370404303,
      "loss_iou": 0.3671875,
      "loss_num": 0.0242919921875,
      "loss_xval": 0.85546875,
      "num_input_tokens_seen": 137262276,
      "step": 2450
    },
    {
      "epoch": 5.458797327394209,
      "grad_norm": 21.784557342529297,
      "learning_rate": 1e-06,
      "loss": 0.6682,
      "num_input_tokens_seen": 137320112,
      "step": 2451
    },
    {
      "epoch": 5.458797327394209,
      "loss": 0.6689052581787109,
      "loss_ce": 0.0002041200641542673,
      "loss_iou": 0.287109375,
      "loss_num": 0.0191650390625,
      "loss_xval": 0.66796875,
      "num_input_tokens_seen": 137320112,
      "step": 2451
    },
    {
      "epoch": 5.461024498886414,
      "grad_norm": 19.112836837768555,
      "learning_rate": 1e-06,
      "loss": 0.7381,
      "num_input_tokens_seen": 137375784,
      "step": 2452
    },
    {
      "epoch": 5.461024498886414,
      "loss": 0.6511021852493286,
      "loss_ce": 0.00022325036115944386,
      "loss_iou": 0.259765625,
      "loss_num": 0.0267333984375,
      "loss_xval": 0.65234375,
      "num_input_tokens_seen": 137375784,
      "step": 2452
    },
    {
      "epoch": 5.463251670378619,
      "grad_norm": 25.689016342163086,
      "learning_rate": 1e-06,
      "loss": 0.7655,
      "num_input_tokens_seen": 137431796,
      "step": 2453
    },
    {
      "epoch": 5.463251670378619,
      "loss": 0.9367631673812866,
      "loss_ce": 0.0006059646257199347,
      "loss_iou": 0.376953125,
      "loss_num": 0.03662109375,
      "loss_xval": 0.9375,
      "num_input_tokens_seen": 137431796,
      "step": 2453
    },
    {
      "epoch": 5.465478841870824,
      "grad_norm": 24.619190216064453,
      "learning_rate": 1e-06,
      "loss": 0.8185,
      "num_input_tokens_seen": 137487700,
      "step": 2454
    },
    {
      "epoch": 5.465478841870824,
      "loss": 0.683560311794281,
      "loss_ce": 0.00021070845832582563,
      "loss_iou": 0.27734375,
      "loss_num": 0.02587890625,
      "loss_xval": 0.68359375,
      "num_input_tokens_seen": 137487700,
      "step": 2454
    },
    {
      "epoch": 5.467706013363029,
      "grad_norm": 19.307693481445312,
      "learning_rate": 1e-06,
      "loss": 0.6936,
      "num_input_tokens_seen": 137544564,
      "step": 2455
    },
    {
      "epoch": 5.467706013363029,
      "loss": 0.6617897748947144,
      "loss_ce": 0.0002907552698161453,
      "loss_iou": 0.27734375,
      "loss_num": 0.021484375,
      "loss_xval": 0.66015625,
      "num_input_tokens_seen": 137544564,
      "step": 2455
    },
    {
      "epoch": 5.4699331848552335,
      "grad_norm": 20.65030860900879,
      "learning_rate": 1e-06,
      "loss": 0.8517,
      "num_input_tokens_seen": 137602024,
      "step": 2456
    },
    {
      "epoch": 5.4699331848552335,
      "loss": 0.7545545101165771,
      "loss_ce": 0.0004040825879201293,
      "loss_iou": 0.29296875,
      "loss_num": 0.033447265625,
      "loss_xval": 0.75390625,
      "num_input_tokens_seen": 137602024,
      "step": 2456
    },
    {
      "epoch": 5.472160356347438,
      "grad_norm": 19.01409339904785,
      "learning_rate": 1e-06,
      "loss": 0.593,
      "num_input_tokens_seen": 137657656,
      "step": 2457
    },
    {
      "epoch": 5.472160356347438,
      "loss": 0.577867329120636,
      "loss_ce": 0.00023062352556735277,
      "loss_iou": 0.2431640625,
      "loss_num": 0.018310546875,
      "loss_xval": 0.578125,
      "num_input_tokens_seen": 137657656,
      "step": 2457
    },
    {
      "epoch": 5.474387527839644,
      "grad_norm": 25.600095748901367,
      "learning_rate": 1e-06,
      "loss": 0.9298,
      "num_input_tokens_seen": 137713816,
      "step": 2458
    },
    {
      "epoch": 5.474387527839644,
      "loss": 0.7569461464881897,
      "loss_ce": 0.00023227237397804856,
      "loss_iou": 0.34375,
      "loss_num": 0.01361083984375,
      "loss_xval": 0.7578125,
      "num_input_tokens_seen": 137713816,
      "step": 2458
    },
    {
      "epoch": 5.476614699331849,
      "grad_norm": 16.993986129760742,
      "learning_rate": 1e-06,
      "loss": 0.695,
      "num_input_tokens_seen": 137771028,
      "step": 2459
    },
    {
      "epoch": 5.476614699331849,
      "loss": 0.7477880120277405,
      "loss_ce": 0.00022941670613363385,
      "loss_iou": 0.302734375,
      "loss_num": 0.0279541015625,
      "loss_xval": 0.74609375,
      "num_input_tokens_seen": 137771028,
      "step": 2459
    },
    {
      "epoch": 5.478841870824054,
      "grad_norm": 25.62329864501953,
      "learning_rate": 1e-06,
      "loss": 0.755,
      "num_input_tokens_seen": 137826816,
      "step": 2460
    },
    {
      "epoch": 5.478841870824054,
      "loss": 0.8492061495780945,
      "loss_ce": 0.0003292199980933219,
      "loss_iou": 0.35546875,
      "loss_num": 0.02734375,
      "loss_xval": 0.84765625,
      "num_input_tokens_seen": 137826816,
      "step": 2460
    },
    {
      "epoch": 5.481069042316259,
      "grad_norm": 18.305076599121094,
      "learning_rate": 1e-06,
      "loss": 0.6416,
      "num_input_tokens_seen": 137879896,
      "step": 2461
    },
    {
      "epoch": 5.481069042316259,
      "loss": 0.6674618721008301,
      "loss_ce": 0.00022550678113475442,
      "loss_iou": 0.30078125,
      "loss_num": 0.01275634765625,
      "loss_xval": 0.66796875,
      "num_input_tokens_seen": 137879896,
      "step": 2461
    },
    {
      "epoch": 5.4832962138084635,
      "grad_norm": 14.69758129119873,
      "learning_rate": 1e-06,
      "loss": 0.5432,
      "num_input_tokens_seen": 137937088,
      "step": 2462
    },
    {
      "epoch": 5.4832962138084635,
      "loss": 0.4212685227394104,
      "loss_ce": 0.0002480166731402278,
      "loss_iou": 0.1875,
      "loss_num": 0.009033203125,
      "loss_xval": 0.421875,
      "num_input_tokens_seen": 137937088,
      "step": 2462
    },
    {
      "epoch": 5.485523385300668,
      "grad_norm": 15.696910858154297,
      "learning_rate": 1e-06,
      "loss": 0.8152,
      "num_input_tokens_seen": 137994708,
      "step": 2463
    },
    {
      "epoch": 5.485523385300668,
      "loss": 0.7479099035263062,
      "loss_ce": 0.00022926190285943449,
      "loss_iou": 0.328125,
      "loss_num": 0.0185546875,
      "loss_xval": 0.74609375,
      "num_input_tokens_seen": 137994708,
      "step": 2463
    },
    {
      "epoch": 5.487750556792873,
      "grad_norm": 14.623006820678711,
      "learning_rate": 1e-06,
      "loss": 0.5012,
      "num_input_tokens_seen": 138052660,
      "step": 2464
    },
    {
      "epoch": 5.487750556792873,
      "loss": 0.48292332887649536,
      "loss_ce": 0.00025733254733495414,
      "loss_iou": 0.193359375,
      "loss_num": 0.019287109375,
      "loss_xval": 0.482421875,
      "num_input_tokens_seen": 138052660,
      "step": 2464
    },
    {
      "epoch": 5.489977728285078,
      "grad_norm": 60.842769622802734,
      "learning_rate": 1e-06,
      "loss": 0.7927,
      "num_input_tokens_seen": 138108180,
      "step": 2465
    },
    {
      "epoch": 5.489977728285078,
      "loss": 0.8183541297912598,
      "loss_ce": 0.00023886302369646728,
      "loss_iou": 0.3203125,
      "loss_num": 0.035888671875,
      "loss_xval": 0.81640625,
      "num_input_tokens_seen": 138108180,
      "step": 2465
    },
    {
      "epoch": 5.492204899777283,
      "grad_norm": 13.957056045532227,
      "learning_rate": 1e-06,
      "loss": 0.5878,
      "num_input_tokens_seen": 138166092,
      "step": 2466
    },
    {
      "epoch": 5.492204899777283,
      "loss": 0.6444277763366699,
      "loss_ce": 0.0003848428023047745,
      "loss_iou": 0.2734375,
      "loss_num": 0.01953125,
      "loss_xval": 0.64453125,
      "num_input_tokens_seen": 138166092,
      "step": 2466
    },
    {
      "epoch": 5.494432071269488,
      "grad_norm": 21.401012420654297,
      "learning_rate": 1e-06,
      "loss": 0.7396,
      "num_input_tokens_seen": 138223400,
      "step": 2467
    },
    {
      "epoch": 5.494432071269488,
      "loss": 0.7348067760467529,
      "loss_ce": 0.0003097508451901376,
      "loss_iou": 0.3046875,
      "loss_num": 0.0247802734375,
      "loss_xval": 0.734375,
      "num_input_tokens_seen": 138223400,
      "step": 2467
    },
    {
      "epoch": 5.496659242761693,
      "grad_norm": 31.345884323120117,
      "learning_rate": 1e-06,
      "loss": 0.6474,
      "num_input_tokens_seen": 138281468,
      "step": 2468
    },
    {
      "epoch": 5.496659242761693,
      "loss": 0.5778509378433228,
      "loss_ce": 0.00021416423260234296,
      "loss_iou": 0.255859375,
      "loss_num": 0.0133056640625,
      "loss_xval": 0.578125,
      "num_input_tokens_seen": 138281468,
      "step": 2468
    },
    {
      "epoch": 5.498886414253898,
      "grad_norm": 14.698542594909668,
      "learning_rate": 1e-06,
      "loss": 0.5857,
      "num_input_tokens_seen": 138337216,
      "step": 2469
    },
    {
      "epoch": 5.498886414253898,
      "loss": 0.5800629258155823,
      "loss_ce": 0.00022892668494023383,
      "loss_iou": 0.2412109375,
      "loss_num": 0.01953125,
      "loss_xval": 0.578125,
      "num_input_tokens_seen": 138337216,
      "step": 2469
    },
    {
      "epoch": 5.501113585746102,
      "grad_norm": 18.81879425048828,
      "learning_rate": 1e-06,
      "loss": 0.5896,
      "num_input_tokens_seen": 138395452,
      "step": 2470
    },
    {
      "epoch": 5.501113585746102,
      "loss": 0.6011573672294617,
      "loss_ce": 0.00020525051513686776,
      "loss_iou": 0.26171875,
      "loss_num": 0.015625,
      "loss_xval": 0.6015625,
      "num_input_tokens_seen": 138395452,
      "step": 2470
    },
    {
      "epoch": 5.503340757238307,
      "grad_norm": 14.421637535095215,
      "learning_rate": 1e-06,
      "loss": 0.5715,
      "num_input_tokens_seen": 138453528,
      "step": 2471
    },
    {
      "epoch": 5.503340757238307,
      "loss": 0.5524652600288391,
      "loss_ce": 0.00021915776596870273,
      "loss_iou": 0.2216796875,
      "loss_num": 0.02197265625,
      "loss_xval": 0.55078125,
      "num_input_tokens_seen": 138453528,
      "step": 2471
    },
    {
      "epoch": 5.505567928730512,
      "grad_norm": 32.5775146484375,
      "learning_rate": 1e-06,
      "loss": 0.5319,
      "num_input_tokens_seen": 138508352,
      "step": 2472
    },
    {
      "epoch": 5.505567928730512,
      "loss": 0.5072603225708008,
      "loss_ce": 0.00018024971359409392,
      "loss_iou": 0.224609375,
      "loss_num": 0.01141357421875,
      "loss_xval": 0.5078125,
      "num_input_tokens_seen": 138508352,
      "step": 2472
    },
    {
      "epoch": 5.507795100222717,
      "grad_norm": 14.467394828796387,
      "learning_rate": 1e-06,
      "loss": 0.9051,
      "num_input_tokens_seen": 138562664,
      "step": 2473
    },
    {
      "epoch": 5.507795100222717,
      "loss": 1.1971768140792847,
      "loss_ce": 0.0003995007718913257,
      "loss_iou": 0.474609375,
      "loss_num": 0.04931640625,
      "loss_xval": 1.1953125,
      "num_input_tokens_seen": 138562664,
      "step": 2473
    },
    {
      "epoch": 5.510022271714922,
      "grad_norm": 15.590438842773438,
      "learning_rate": 1e-06,
      "loss": 0.7975,
      "num_input_tokens_seen": 138614632,
      "step": 2474
    },
    {
      "epoch": 5.510022271714922,
      "loss": 0.7288182973861694,
      "loss_ce": 0.00030269764829427004,
      "loss_iou": 0.32421875,
      "loss_num": 0.0162353515625,
      "loss_xval": 0.7265625,
      "num_input_tokens_seen": 138614632,
      "step": 2474
    },
    {
      "epoch": 5.512249443207127,
      "grad_norm": 16.527565002441406,
      "learning_rate": 1e-06,
      "loss": 0.6321,
      "num_input_tokens_seen": 138672144,
      "step": 2475
    },
    {
      "epoch": 5.512249443207127,
      "loss": 0.5714784860610962,
      "loss_ce": 0.00018944795010611415,
      "loss_iou": 0.2255859375,
      "loss_num": 0.02392578125,
      "loss_xval": 0.5703125,
      "num_input_tokens_seen": 138672144,
      "step": 2475
    },
    {
      "epoch": 5.514476614699332,
      "grad_norm": 16.16336441040039,
      "learning_rate": 1e-06,
      "loss": 0.6376,
      "num_input_tokens_seen": 138727908,
      "step": 2476
    },
    {
      "epoch": 5.514476614699332,
      "loss": 0.5442032814025879,
      "loss_ce": 0.0002579318534117192,
      "loss_iou": 0.2421875,
      "loss_num": 0.01177978515625,
      "loss_xval": 0.54296875,
      "num_input_tokens_seen": 138727908,
      "step": 2476
    },
    {
      "epoch": 5.5167037861915365,
      "grad_norm": 19.183643341064453,
      "learning_rate": 1e-06,
      "loss": 0.6079,
      "num_input_tokens_seen": 138783588,
      "step": 2477
    },
    {
      "epoch": 5.5167037861915365,
      "loss": 0.5135550498962402,
      "loss_ce": 0.00037142602377571166,
      "loss_iou": 0.1826171875,
      "loss_num": 0.029541015625,
      "loss_xval": 0.51171875,
      "num_input_tokens_seen": 138783588,
      "step": 2477
    },
    {
      "epoch": 5.518930957683741,
      "grad_norm": 34.50285339355469,
      "learning_rate": 1e-06,
      "loss": 0.588,
      "num_input_tokens_seen": 138836984,
      "step": 2478
    },
    {
      "epoch": 5.518930957683741,
      "loss": 0.5181179046630859,
      "loss_ce": 0.00029565999284386635,
      "loss_iou": 0.2001953125,
      "loss_num": 0.0233154296875,
      "loss_xval": 0.51953125,
      "num_input_tokens_seen": 138836984,
      "step": 2478
    },
    {
      "epoch": 5.521158129175946,
      "grad_norm": 16.072481155395508,
      "learning_rate": 1e-06,
      "loss": 0.5809,
      "num_input_tokens_seen": 138895444,
      "step": 2479
    },
    {
      "epoch": 5.521158129175946,
      "loss": 0.4480636417865753,
      "loss_ce": 0.00018766321591101587,
      "loss_iou": 0.1806640625,
      "loss_num": 0.0174560546875,
      "loss_xval": 0.447265625,
      "num_input_tokens_seen": 138895444,
      "step": 2479
    },
    {
      "epoch": 5.523385300668151,
      "grad_norm": 27.131000518798828,
      "learning_rate": 1e-06,
      "loss": 0.8815,
      "num_input_tokens_seen": 138948860,
      "step": 2480
    },
    {
      "epoch": 5.523385300668151,
      "loss": 0.9381879568099976,
      "loss_ce": 0.00019970518769696355,
      "loss_iou": 0.390625,
      "loss_num": 0.03173828125,
      "loss_xval": 0.9375,
      "num_input_tokens_seen": 138948860,
      "step": 2480
    },
    {
      "epoch": 5.525612472160356,
      "grad_norm": 21.947586059570312,
      "learning_rate": 1e-06,
      "loss": 0.6556,
      "num_input_tokens_seen": 139003808,
      "step": 2481
    },
    {
      "epoch": 5.525612472160356,
      "loss": 0.6769753694534302,
      "loss_ce": 0.0004616759833879769,
      "loss_iou": 0.287109375,
      "loss_num": 0.020263671875,
      "loss_xval": 0.67578125,
      "num_input_tokens_seen": 139003808,
      "step": 2481
    },
    {
      "epoch": 5.527839643652561,
      "grad_norm": 18.34388542175293,
      "learning_rate": 1e-06,
      "loss": 1.0344,
      "num_input_tokens_seen": 139060604,
      "step": 2482
    },
    {
      "epoch": 5.527839643652561,
      "loss": 1.1095937490463257,
      "loss_ce": 0.00021878261759411544,
      "loss_iou": 0.4296875,
      "loss_num": 0.0498046875,
      "loss_xval": 1.109375,
      "num_input_tokens_seen": 139060604,
      "step": 2482
    },
    {
      "epoch": 5.5300668151447665,
      "grad_norm": 28.046768188476562,
      "learning_rate": 1e-06,
      "loss": 0.9061,
      "num_input_tokens_seen": 139115064,
      "step": 2483
    },
    {
      "epoch": 5.5300668151447665,
      "loss": 0.8196831941604614,
      "loss_ce": 0.00022515948512591422,
      "loss_iou": 0.353515625,
      "loss_num": 0.0225830078125,
      "loss_xval": 0.8203125,
      "num_input_tokens_seen": 139115064,
      "step": 2483
    },
    {
      "epoch": 5.532293986636971,
      "grad_norm": 38.94261169433594,
      "learning_rate": 1e-06,
      "loss": 0.5469,
      "num_input_tokens_seen": 139171664,
      "step": 2484
    },
    {
      "epoch": 5.532293986636971,
      "loss": 0.418997585773468,
      "loss_ce": 0.0002963862498290837,
      "loss_iou": 0.1787109375,
      "loss_num": 0.01220703125,
      "loss_xval": 0.41796875,
      "num_input_tokens_seen": 139171664,
      "step": 2484
    },
    {
      "epoch": 5.534521158129176,
      "grad_norm": 21.334020614624023,
      "learning_rate": 1e-06,
      "loss": 0.7203,
      "num_input_tokens_seen": 139228464,
      "step": 2485
    },
    {
      "epoch": 5.534521158129176,
      "loss": 0.8227799534797668,
      "loss_ce": 0.0002701581106521189,
      "loss_iou": 0.349609375,
      "loss_num": 0.0244140625,
      "loss_xval": 0.82421875,
      "num_input_tokens_seen": 139228464,
      "step": 2485
    },
    {
      "epoch": 5.536748329621381,
      "grad_norm": 30.529455184936523,
      "learning_rate": 1e-06,
      "loss": 0.6742,
      "num_input_tokens_seen": 139281716,
      "step": 2486
    },
    {
      "epoch": 5.536748329621381,
      "loss": 0.769308865070343,
      "loss_ce": 0.00026593299116939306,
      "loss_iou": 0.353515625,
      "loss_num": 0.01190185546875,
      "loss_xval": 0.76953125,
      "num_input_tokens_seen": 139281716,
      "step": 2486
    },
    {
      "epoch": 5.538975501113586,
      "grad_norm": 15.466273307800293,
      "learning_rate": 1e-06,
      "loss": 0.5857,
      "num_input_tokens_seen": 139339380,
      "step": 2487
    },
    {
      "epoch": 5.538975501113586,
      "loss": 0.6729258298873901,
      "loss_ce": 0.00031836878042668104,
      "loss_iou": 0.28515625,
      "loss_num": 0.0206298828125,
      "loss_xval": 0.671875,
      "num_input_tokens_seen": 139339380,
      "step": 2487
    },
    {
      "epoch": 5.541202672605791,
      "grad_norm": 21.989582061767578,
      "learning_rate": 1e-06,
      "loss": 0.6376,
      "num_input_tokens_seen": 139398944,
      "step": 2488
    },
    {
      "epoch": 5.541202672605791,
      "loss": 0.7792110443115234,
      "loss_ce": 0.0004024332156404853,
      "loss_iou": 0.31640625,
      "loss_num": 0.029541015625,
      "loss_xval": 0.77734375,
      "num_input_tokens_seen": 139398944,
      "step": 2488
    },
    {
      "epoch": 5.543429844097996,
      "grad_norm": 35.42259979248047,
      "learning_rate": 1e-06,
      "loss": 0.8917,
      "num_input_tokens_seen": 139454540,
      "step": 2489
    },
    {
      "epoch": 5.543429844097996,
      "loss": 1.0833284854888916,
      "loss_ce": 0.00032068698783405125,
      "loss_iou": 0.44140625,
      "loss_num": 0.0400390625,
      "loss_xval": 1.0859375,
      "num_input_tokens_seen": 139454540,
      "step": 2489
    },
    {
      "epoch": 5.5456570155902005,
      "grad_norm": 20.069564819335938,
      "learning_rate": 1e-06,
      "loss": 0.7142,
      "num_input_tokens_seen": 139510416,
      "step": 2490
    },
    {
      "epoch": 5.5456570155902005,
      "loss": 0.6588754653930664,
      "loss_ce": 0.0005502753192558885,
      "loss_iou": 0.28515625,
      "loss_num": 0.017822265625,
      "loss_xval": 0.66015625,
      "num_input_tokens_seen": 139510416,
      "step": 2490
    },
    {
      "epoch": 5.547884187082405,
      "grad_norm": 28.32633399963379,
      "learning_rate": 1e-06,
      "loss": 0.5564,
      "num_input_tokens_seen": 139567320,
      "step": 2491
    },
    {
      "epoch": 5.547884187082405,
      "loss": 0.4439275860786438,
      "loss_ce": 0.00026305546634830534,
      "loss_iou": 0.2021484375,
      "loss_num": 0.00799560546875,
      "loss_xval": 0.443359375,
      "num_input_tokens_seen": 139567320,
      "step": 2491
    },
    {
      "epoch": 5.55011135857461,
      "grad_norm": 29.56279182434082,
      "learning_rate": 1e-06,
      "loss": 0.6047,
      "num_input_tokens_seen": 139625440,
      "step": 2492
    },
    {
      "epoch": 5.55011135857461,
      "loss": 0.5217064619064331,
      "loss_ce": 0.00022206196445040405,
      "loss_iou": 0.2431640625,
      "loss_num": 0.006988525390625,
      "loss_xval": 0.5234375,
      "num_input_tokens_seen": 139625440,
      "step": 2492
    },
    {
      "epoch": 5.552338530066815,
      "grad_norm": 46.10884475708008,
      "learning_rate": 1e-06,
      "loss": 0.5618,
      "num_input_tokens_seen": 139682200,
      "step": 2493
    },
    {
      "epoch": 5.552338530066815,
      "loss": 0.41375732421875,
      "loss_ce": 0.00018312688916921616,
      "loss_iou": 0.1875,
      "loss_num": 0.00775146484375,
      "loss_xval": 0.4140625,
      "num_input_tokens_seen": 139682200,
      "step": 2493
    },
    {
      "epoch": 5.55456570155902,
      "grad_norm": 18.429887771606445,
      "learning_rate": 1e-06,
      "loss": 0.5575,
      "num_input_tokens_seen": 139738800,
      "step": 2494
    },
    {
      "epoch": 5.55456570155902,
      "loss": 0.6128218770027161,
      "loss_ce": 0.00027302149101160467,
      "loss_iou": 0.28515625,
      "loss_num": 0.008544921875,
      "loss_xval": 0.61328125,
      "num_input_tokens_seen": 139738800,
      "step": 2494
    },
    {
      "epoch": 5.556792873051225,
      "grad_norm": 24.69832420349121,
      "learning_rate": 1e-06,
      "loss": 0.7079,
      "num_input_tokens_seen": 139796508,
      "step": 2495
    },
    {
      "epoch": 5.556792873051225,
      "loss": 0.8797301054000854,
      "loss_ce": 0.00027458026306703687,
      "loss_iou": 0.3515625,
      "loss_num": 0.03466796875,
      "loss_xval": 0.87890625,
      "num_input_tokens_seen": 139796508,
      "step": 2495
    },
    {
      "epoch": 5.55902004454343,
      "grad_norm": 20.031938552856445,
      "learning_rate": 1e-06,
      "loss": 0.8567,
      "num_input_tokens_seen": 139852360,
      "step": 2496
    },
    {
      "epoch": 5.55902004454343,
      "loss": 0.9052836894989014,
      "loss_ce": 0.000254325830610469,
      "loss_iou": 0.412109375,
      "loss_num": 0.0166015625,
      "loss_xval": 0.90625,
      "num_input_tokens_seen": 139852360,
      "step": 2496
    },
    {
      "epoch": 5.5612472160356345,
      "grad_norm": 24.965023040771484,
      "learning_rate": 1e-06,
      "loss": 0.7616,
      "num_input_tokens_seen": 139907408,
      "step": 2497
    },
    {
      "epoch": 5.5612472160356345,
      "loss": 0.6985030770301819,
      "loss_ce": 0.0002608746290206909,
      "loss_iou": 0.287109375,
      "loss_num": 0.0250244140625,
      "loss_xval": 0.69921875,
      "num_input_tokens_seen": 139907408,
      "step": 2497
    },
    {
      "epoch": 5.563474387527839,
      "grad_norm": 22.246639251708984,
      "learning_rate": 1e-06,
      "loss": 0.9545,
      "num_input_tokens_seen": 139963936,
      "step": 2498
    },
    {
      "epoch": 5.563474387527839,
      "loss": 0.8662214875221252,
      "loss_ce": 0.0002546662581153214,
      "loss_iou": 0.35546875,
      "loss_num": 0.031005859375,
      "loss_xval": 0.8671875,
      "num_input_tokens_seen": 139963936,
      "step": 2498
    },
    {
      "epoch": 5.565701559020044,
      "grad_norm": 16.71168327331543,
      "learning_rate": 1e-06,
      "loss": 0.5694,
      "num_input_tokens_seen": 140018732,
      "step": 2499
    },
    {
      "epoch": 5.565701559020044,
      "loss": 0.6181437373161316,
      "loss_ce": 0.00022381696908269078,
      "loss_iou": 0.263671875,
      "loss_num": 0.0177001953125,
      "loss_xval": 0.6171875,
      "num_input_tokens_seen": 140018732,
      "step": 2499
    },
    {
      "epoch": 5.567928730512249,
      "grad_norm": 17.26616096496582,
      "learning_rate": 1e-06,
      "loss": 0.6123,
      "num_input_tokens_seen": 140075980,
      "step": 2500
    },
    {
      "epoch": 5.567928730512249,
      "eval_seeclick_web_CIoU": 0.5732596218585968,
      "eval_seeclick_web_GIoU": 0.5691904425621033,
      "eval_seeclick_web_IoU": 0.5895318686962128,
      "eval_seeclick_web_MAE_all": 0.016760945785790682,
      "eval_seeclick_web_MAE_h": 0.009123492753133178,
      "eval_seeclick_web_MAE_w": 0.01760054472833872,
      "eval_seeclick_web_MAE_x_boxes": 0.008502837270498276,
      "eval_seeclick_web_MAE_y_boxes": 0.02224468719214201,
      "eval_seeclick_web_inside_bbox": 0.9166666567325592,
      "eval_seeclick_web_loss": 0.9365757703781128,
      "eval_seeclick_web_loss_ce": 0.00031520209449809045,
      "eval_seeclick_web_loss_iou": 0.4267578125,
      "eval_seeclick_web_loss_num": 0.013166427612304688,
      "eval_seeclick_web_loss_xval": 0.918701171875,
      "eval_seeclick_web_runtime": 28.1194,
      "eval_seeclick_web_samples_per_second": 1.778,
      "eval_seeclick_web_steps_per_second": 0.071,
      "num_input_tokens_seen": 140075980,
      "step": 2500
    },
    {
      "epoch": 5.567928730512249,
      "eval_icons_CIoU": 0.32149502635002136,
      "eval_icons_GIoU": 0.33890318870544434,
      "eval_icons_IoU": 0.3914993554353714,
      "eval_icons_MAE_all": 0.05996252968907356,
      "eval_icons_MAE_h": 0.03914587013423443,
      "eval_icons_MAE_w": 0.05691290646791458,
      "eval_icons_MAE_x_boxes": 0.05956794694066048,
      "eval_icons_MAE_y_boxes": 0.037835730239748955,
      "eval_icons_inside_bbox": 0.6493055522441864,
      "eval_icons_loss": 1.6447449922561646,
      "eval_icons_loss_ce": 0.0003619373310357332,
      "eval_icons_loss_iou": 0.6527099609375,
      "eval_icons_loss_num": 0.05636787414550781,
      "eval_icons_loss_xval": 1.58837890625,
      "eval_icons_runtime": 25.1446,
      "eval_icons_samples_per_second": 1.988,
      "eval_icons_steps_per_second": 0.08,
      "num_input_tokens_seen": 140075980,
      "step": 2500
    },
    {
      "epoch": 5.567928730512249,
      "eval_screenspot_CIoU": 0.33252301812171936,
      "eval_screenspot_GIoU": 0.3458707630634308,
      "eval_screenspot_IoU": 0.4156401753425598,
      "eval_screenspot_MAE_all": 0.066184946646293,
      "eval_screenspot_MAE_h": 0.03784913197159767,
      "eval_screenspot_MAE_w": 0.08022025724252065,
      "eval_screenspot_MAE_x_boxes": 0.08052412172158559,
      "eval_screenspot_MAE_y_boxes": 0.0472174392392238,
      "eval_screenspot_inside_bbox": 0.6462500095367432,
      "eval_screenspot_loss": 1.7031261920928955,
      "eval_screenspot_loss_ce": 0.0003634931442017357,
      "eval_screenspot_loss_iou": 0.6932779947916666,
      "eval_screenspot_loss_num": 0.07747904459635417,
      "eval_screenspot_loss_xval": 1.7745768229166667,
      "eval_screenspot_runtime": 39.8835,
      "eval_screenspot_samples_per_second": 2.231,
      "eval_screenspot_steps_per_second": 0.075,
      "num_input_tokens_seen": 140075980,
      "step": 2500
    },
    {
      "epoch": 5.567928730512249,
      "eval_compot_CIoU": 0.36012406647205353,
      "eval_compot_GIoU": 0.3780812919139862,
      "eval_compot_IoU": 0.41496771574020386,
      "eval_compot_MAE_all": 0.017852995079010725,
      "eval_compot_MAE_h": 0.007951512467116117,
      "eval_compot_MAE_w": 0.02203182876110077,
      "eval_compot_MAE_x_boxes": 0.02926653064787388,
      "eval_compot_MAE_y_boxes": 0.006283238530158997,
      "eval_compot_inside_bbox": 0.6458333432674408,
      "eval_compot_loss": 1.3696753978729248,
      "eval_compot_loss_ce": 0.00030196372244972736,
      "eval_compot_loss_iou": 0.6297607421875,
      "eval_compot_loss_num": 0.016778945922851562,
      "eval_compot_loss_xval": 1.343505859375,
      "eval_compot_runtime": 24.2244,
      "eval_compot_samples_per_second": 2.064,
      "eval_compot_steps_per_second": 0.083,
      "num_input_tokens_seen": 140075980,
      "step": 2500
    },
    {
      "epoch": 5.567928730512249,
      "eval_custom_ui_val_CIoU": 0.4640722307893965,
      "eval_custom_ui_val_GIoU": 0.48160041703118217,
      "eval_custom_ui_val_IoU": 0.518558962477578,
      "eval_custom_ui_val_MAE_all": 0.03118372191157606,
      "eval_custom_ui_val_MAE_h": 0.018310750601813197,
      "eval_custom_ui_val_MAE_w": 0.036142679759197764,
      "eval_custom_ui_val_MAE_x_boxes": 0.03573724896543556,
      "eval_custom_ui_val_MAE_y_boxes": 0.01598744459139804,
      "eval_custom_ui_val_inside_bbox": 0.738811731338501,
      "eval_custom_ui_val_loss": 1.2015693187713623,
      "eval_custom_ui_val_loss_ce": 0.0003550093032471422,
      "eval_custom_ui_val_loss_iou": 0.5086941189236112,
      "eval_custom_ui_val_loss_num": 0.029562632242838543,
      "eval_custom_ui_val_loss_xval": 1.1652560763888888,
      "eval_custom_ui_val_runtime": 74.2518,
      "eval_custom_ui_val_samples_per_second": 3.569,
      "eval_custom_ui_val_steps_per_second": 0.121,
      "num_input_tokens_seen": 140075980,
      "step": 2500
    },
    {
      "epoch": 5.567928730512249,
      "loss": 0.907296895980835,
      "loss_ce": 0.0003145075461361557,
      "loss_iou": 0.3984375,
      "loss_num": 0.0216064453125,
      "loss_xval": 0.90625,
      "num_input_tokens_seen": 140075980,
      "step": 2500
    },
    {
      "epoch": 5.570155902004454,
      "grad_norm": 96.08206939697266,
      "learning_rate": 1e-06,
      "loss": 0.6085,
      "num_input_tokens_seen": 140130244,
      "step": 2501
    },
    {
      "epoch": 5.570155902004454,
      "loss": 0.6808421611785889,
      "loss_ce": 0.00023916579084470868,
      "loss_iou": 0.28125,
      "loss_num": 0.0234375,
      "loss_xval": 0.6796875,
      "num_input_tokens_seen": 140130244,
      "step": 2501
    },
    {
      "epoch": 5.57238307349666,
      "grad_norm": 23.38459587097168,
      "learning_rate": 1e-06,
      "loss": 0.7066,
      "num_input_tokens_seen": 140187016,
      "step": 2502
    },
    {
      "epoch": 5.57238307349666,
      "loss": 0.6098702549934387,
      "loss_ce": 0.0002510999620426446,
      "loss_iou": 0.267578125,
      "loss_num": 0.0147705078125,
      "loss_xval": 0.609375,
      "num_input_tokens_seen": 140187016,
      "step": 2502
    },
    {
      "epoch": 5.574610244988865,
      "grad_norm": 18.638948440551758,
      "learning_rate": 1e-06,
      "loss": 0.6391,
      "num_input_tokens_seen": 140245660,
      "step": 2503
    },
    {
      "epoch": 5.574610244988865,
      "loss": 0.7389647364616394,
      "loss_ce": 0.0001952238380908966,
      "loss_iou": 0.287109375,
      "loss_num": 0.03271484375,
      "loss_xval": 0.73828125,
      "num_input_tokens_seen": 140245660,
      "step": 2503
    },
    {
      "epoch": 5.5768374164810695,
      "grad_norm": 15.6986722946167,
      "learning_rate": 1e-06,
      "loss": 0.6478,
      "num_input_tokens_seen": 140300240,
      "step": 2504
    },
    {
      "epoch": 5.5768374164810695,
      "loss": 0.4837605357170105,
      "loss_ce": 0.00024005114391911775,
      "loss_iou": 0.2177734375,
      "loss_num": 0.00958251953125,
      "loss_xval": 0.484375,
      "num_input_tokens_seen": 140300240,
      "step": 2504
    },
    {
      "epoch": 5.579064587973274,
      "grad_norm": 15.869756698608398,
      "learning_rate": 1e-06,
      "loss": 0.6703,
      "num_input_tokens_seen": 140357396,
      "step": 2505
    },
    {
      "epoch": 5.579064587973274,
      "loss": 0.8984197378158569,
      "loss_ce": 0.00022636953508481383,
      "loss_iou": 0.365234375,
      "loss_num": 0.03369140625,
      "loss_xval": 0.8984375,
      "num_input_tokens_seen": 140357396,
      "step": 2505
    },
    {
      "epoch": 5.581291759465479,
      "grad_norm": 15.25345516204834,
      "learning_rate": 1e-06,
      "loss": 0.8536,
      "num_input_tokens_seen": 140413868,
      "step": 2506
    },
    {
      "epoch": 5.581291759465479,
      "loss": 0.951309084892273,
      "loss_ce": 0.0006254723994061351,
      "loss_iou": 0.41015625,
      "loss_num": 0.0257568359375,
      "loss_xval": 0.94921875,
      "num_input_tokens_seen": 140413868,
      "step": 2506
    },
    {
      "epoch": 5.583518930957684,
      "grad_norm": 35.47723388671875,
      "learning_rate": 1e-06,
      "loss": 0.6912,
      "num_input_tokens_seen": 140471300,
      "step": 2507
    },
    {
      "epoch": 5.583518930957684,
      "loss": 0.6750940084457397,
      "loss_ce": 0.0002893685596063733,
      "loss_iou": 0.306640625,
      "loss_num": 0.011962890625,
      "loss_xval": 0.67578125,
      "num_input_tokens_seen": 140471300,
      "step": 2507
    },
    {
      "epoch": 5.585746102449889,
      "grad_norm": 45.554256439208984,
      "learning_rate": 1e-06,
      "loss": 0.6581,
      "num_input_tokens_seen": 140528484,
      "step": 2508
    },
    {
      "epoch": 5.585746102449889,
      "loss": 0.7441803216934204,
      "loss_ce": 0.00028382547316141427,
      "loss_iou": 0.326171875,
      "loss_num": 0.0181884765625,
      "loss_xval": 0.7421875,
      "num_input_tokens_seen": 140528484,
      "step": 2508
    },
    {
      "epoch": 5.587973273942094,
      "grad_norm": 13.767509460449219,
      "learning_rate": 1e-06,
      "loss": 0.7014,
      "num_input_tokens_seen": 140585164,
      "step": 2509
    },
    {
      "epoch": 5.587973273942094,
      "loss": 0.6096572875976562,
      "loss_ce": 0.00028234010096639395,
      "loss_iou": 0.25390625,
      "loss_num": 0.0201416015625,
      "loss_xval": 0.609375,
      "num_input_tokens_seen": 140585164,
      "step": 2509
    },
    {
      "epoch": 5.590200445434299,
      "grad_norm": 42.34712219238281,
      "learning_rate": 1e-06,
      "loss": 0.7912,
      "num_input_tokens_seen": 140640568,
      "step": 2510
    },
    {
      "epoch": 5.590200445434299,
      "loss": 0.9159772396087646,
      "loss_ce": 0.00020570913329720497,
      "loss_iou": 0.3984375,
      "loss_num": 0.0234375,
      "loss_xval": 0.9140625,
      "num_input_tokens_seen": 140640568,
      "step": 2510
    },
    {
      "epoch": 5.5924276169265035,
      "grad_norm": 12.886595726013184,
      "learning_rate": 1e-06,
      "loss": 0.8791,
      "num_input_tokens_seen": 140697100,
      "step": 2511
    },
    {
      "epoch": 5.5924276169265035,
      "loss": 0.7073045372962952,
      "loss_ce": 0.0002733045257627964,
      "loss_iou": 0.306640625,
      "loss_num": 0.0189208984375,
      "loss_xval": 0.70703125,
      "num_input_tokens_seen": 140697100,
      "step": 2511
    },
    {
      "epoch": 5.594654788418708,
      "grad_norm": 20.680938720703125,
      "learning_rate": 1e-06,
      "loss": 0.8405,
      "num_input_tokens_seen": 140754100,
      "step": 2512
    },
    {
      "epoch": 5.594654788418708,
      "loss": 0.854705274105072,
      "loss_ce": 0.00021311425371095538,
      "loss_iou": 0.341796875,
      "loss_num": 0.034423828125,
      "loss_xval": 0.85546875,
      "num_input_tokens_seen": 140754100,
      "step": 2512
    },
    {
      "epoch": 5.596881959910913,
      "grad_norm": 23.940448760986328,
      "learning_rate": 1e-06,
      "loss": 0.8587,
      "num_input_tokens_seen": 140809880,
      "step": 2513
    },
    {
      "epoch": 5.596881959910913,
      "loss": 0.8762653470039368,
      "loss_ce": 0.0002887820010073483,
      "loss_iou": 0.40625,
      "loss_num": 0.012451171875,
      "loss_xval": 0.875,
      "num_input_tokens_seen": 140809880,
      "step": 2513
    },
    {
      "epoch": 5.599109131403118,
      "grad_norm": 22.817508697509766,
      "learning_rate": 1e-06,
      "loss": 0.6931,
      "num_input_tokens_seen": 140865016,
      "step": 2514
    },
    {
      "epoch": 5.599109131403118,
      "loss": 0.8307744860649109,
      "loss_ce": 0.00020807303371839225,
      "loss_iou": 0.353515625,
      "loss_num": 0.0250244140625,
      "loss_xval": 0.83203125,
      "num_input_tokens_seen": 140865016,
      "step": 2514
    },
    {
      "epoch": 5.601336302895323,
      "grad_norm": 27.795007705688477,
      "learning_rate": 1e-06,
      "loss": 0.6248,
      "num_input_tokens_seen": 140921912,
      "step": 2515
    },
    {
      "epoch": 5.601336302895323,
      "loss": 0.6642851829528809,
      "loss_ce": 0.00022267791791819036,
      "loss_iou": 0.294921875,
      "loss_num": 0.014892578125,
      "loss_xval": 0.6640625,
      "num_input_tokens_seen": 140921912,
      "step": 2515
    },
    {
      "epoch": 5.603563474387528,
      "grad_norm": 17.664352416992188,
      "learning_rate": 1e-06,
      "loss": 0.5839,
      "num_input_tokens_seen": 140976712,
      "step": 2516
    },
    {
      "epoch": 5.603563474387528,
      "loss": 0.4930035173892975,
      "loss_ce": 0.00020567109459079802,
      "loss_iou": 0.2109375,
      "loss_num": 0.013916015625,
      "loss_xval": 0.4921875,
      "num_input_tokens_seen": 140976712,
      "step": 2516
    },
    {
      "epoch": 5.605790645879733,
      "grad_norm": 16.923248291015625,
      "learning_rate": 1e-06,
      "loss": 0.6882,
      "num_input_tokens_seen": 141030776,
      "step": 2517
    },
    {
      "epoch": 5.605790645879733,
      "loss": 0.6687592267990112,
      "loss_ce": 0.0001800994505174458,
      "loss_iou": 0.275390625,
      "loss_num": 0.0233154296875,
      "loss_xval": 0.66796875,
      "num_input_tokens_seen": 141030776,
      "step": 2517
    },
    {
      "epoch": 5.6080178173719375,
      "grad_norm": 21.347900390625,
      "learning_rate": 1e-06,
      "loss": 0.6674,
      "num_input_tokens_seen": 141087176,
      "step": 2518
    },
    {
      "epoch": 5.6080178173719375,
      "loss": 0.7416476011276245,
      "loss_ce": 0.00019255219376645982,
      "loss_iou": 0.333984375,
      "loss_num": 0.01458740234375,
      "loss_xval": 0.7421875,
      "num_input_tokens_seen": 141087176,
      "step": 2518
    },
    {
      "epoch": 5.610244988864142,
      "grad_norm": 29.194229125976562,
      "learning_rate": 1e-06,
      "loss": 0.7402,
      "num_input_tokens_seen": 141143912,
      "step": 2519
    },
    {
      "epoch": 5.610244988864142,
      "loss": 0.937484860420227,
      "loss_ce": 0.00022902997443452477,
      "loss_iou": 0.421875,
      "loss_num": 0.0186767578125,
      "loss_xval": 0.9375,
      "num_input_tokens_seen": 141143912,
      "step": 2519
    },
    {
      "epoch": 5.612472160356347,
      "grad_norm": 25.27153778076172,
      "learning_rate": 1e-06,
      "loss": 0.6045,
      "num_input_tokens_seen": 141199920,
      "step": 2520
    },
    {
      "epoch": 5.612472160356347,
      "loss": 0.7792827486991882,
      "loss_ce": 0.00023000439978204668,
      "loss_iou": 0.318359375,
      "loss_num": 0.028564453125,
      "loss_xval": 0.77734375,
      "num_input_tokens_seen": 141199920,
      "step": 2520
    },
    {
      "epoch": 5.614699331848552,
      "grad_norm": 20.523181915283203,
      "learning_rate": 1e-06,
      "loss": 0.5798,
      "num_input_tokens_seen": 141257784,
      "step": 2521
    },
    {
      "epoch": 5.614699331848552,
      "loss": 0.6049889326095581,
      "loss_ce": 0.0002525739837437868,
      "loss_iou": 0.271484375,
      "loss_num": 0.01226806640625,
      "loss_xval": 0.60546875,
      "num_input_tokens_seen": 141257784,
      "step": 2521
    },
    {
      "epoch": 5.616926503340757,
      "grad_norm": 26.476478576660156,
      "learning_rate": 1e-06,
      "loss": 0.7458,
      "num_input_tokens_seen": 141311380,
      "step": 2522
    },
    {
      "epoch": 5.616926503340757,
      "loss": 0.8652767539024353,
      "loss_ce": 0.000286536494968459,
      "loss_iou": 0.3984375,
      "loss_num": 0.01385498046875,
      "loss_xval": 0.86328125,
      "num_input_tokens_seen": 141311380,
      "step": 2522
    },
    {
      "epoch": 5.619153674832962,
      "grad_norm": 25.442533493041992,
      "learning_rate": 1e-06,
      "loss": 0.8765,
      "num_input_tokens_seen": 141367500,
      "step": 2523
    },
    {
      "epoch": 5.619153674832962,
      "loss": 1.0669339895248413,
      "loss_ce": 0.00028354296227917075,
      "loss_iou": 0.44140625,
      "loss_num": 0.037109375,
      "loss_xval": 1.0703125,
      "num_input_tokens_seen": 141367500,
      "step": 2523
    },
    {
      "epoch": 5.621380846325167,
      "grad_norm": 24.357839584350586,
      "learning_rate": 1e-06,
      "loss": 0.6908,
      "num_input_tokens_seen": 141425116,
      "step": 2524
    },
    {
      "epoch": 5.621380846325167,
      "loss": 0.751206636428833,
      "loss_ce": 0.0002300894120708108,
      "loss_iou": 0.3359375,
      "loss_num": 0.0162353515625,
      "loss_xval": 0.75,
      "num_input_tokens_seen": 141425116,
      "step": 2524
    },
    {
      "epoch": 5.6236080178173715,
      "grad_norm": 17.895536422729492,
      "learning_rate": 1e-06,
      "loss": 0.7793,
      "num_input_tokens_seen": 141481940,
      "step": 2525
    },
    {
      "epoch": 5.6236080178173715,
      "loss": 0.9199173450469971,
      "loss_ce": 0.0002396509371465072,
      "loss_iou": 0.390625,
      "loss_num": 0.027587890625,
      "loss_xval": 0.91796875,
      "num_input_tokens_seen": 141481940,
      "step": 2525
    },
    {
      "epoch": 5.625835189309576,
      "grad_norm": 13.240893363952637,
      "learning_rate": 1e-06,
      "loss": 0.686,
      "num_input_tokens_seen": 141538040,
      "step": 2526
    },
    {
      "epoch": 5.625835189309576,
      "loss": 0.52378910779953,
      "loss_ce": 0.00022953613370191306,
      "loss_iou": 0.234375,
      "loss_num": 0.0107421875,
      "loss_xval": 0.5234375,
      "num_input_tokens_seen": 141538040,
      "step": 2526
    },
    {
      "epoch": 5.628062360801781,
      "grad_norm": 16.763574600219727,
      "learning_rate": 1e-06,
      "loss": 0.6977,
      "num_input_tokens_seen": 141595016,
      "step": 2527
    },
    {
      "epoch": 5.628062360801781,
      "loss": 0.8276025056838989,
      "loss_ce": 0.00020992739882785827,
      "loss_iou": 0.32421875,
      "loss_num": 0.03515625,
      "loss_xval": 0.828125,
      "num_input_tokens_seen": 141595016,
      "step": 2527
    },
    {
      "epoch": 5.630289532293987,
      "grad_norm": 23.006132125854492,
      "learning_rate": 1e-06,
      "loss": 0.788,
      "num_input_tokens_seen": 141653304,
      "step": 2528
    },
    {
      "epoch": 5.630289532293987,
      "loss": 0.8520069122314453,
      "loss_ce": 0.0002002965338760987,
      "loss_iou": 0.35546875,
      "loss_num": 0.0281982421875,
      "loss_xval": 0.8515625,
      "num_input_tokens_seen": 141653304,
      "step": 2528
    },
    {
      "epoch": 5.632516703786192,
      "grad_norm": 15.624357223510742,
      "learning_rate": 1e-06,
      "loss": 0.651,
      "num_input_tokens_seen": 141710920,
      "step": 2529
    },
    {
      "epoch": 5.632516703786192,
      "loss": 0.6601657867431641,
      "loss_ce": 0.0002536678221076727,
      "loss_iou": 0.296875,
      "loss_num": 0.01312255859375,
      "loss_xval": 0.66015625,
      "num_input_tokens_seen": 141710920,
      "step": 2529
    },
    {
      "epoch": 5.634743875278397,
      "grad_norm": 18.147720336914062,
      "learning_rate": 1e-06,
      "loss": 0.7309,
      "num_input_tokens_seen": 141765836,
      "step": 2530
    },
    {
      "epoch": 5.634743875278397,
      "loss": 0.8750065565109253,
      "loss_ce": 0.000250728742685169,
      "loss_iou": 0.36328125,
      "loss_num": 0.029541015625,
      "loss_xval": 0.875,
      "num_input_tokens_seen": 141765836,
      "step": 2530
    },
    {
      "epoch": 5.636971046770602,
      "grad_norm": 18.959402084350586,
      "learning_rate": 1e-06,
      "loss": 0.6962,
      "num_input_tokens_seen": 141821112,
      "step": 2531
    },
    {
      "epoch": 5.636971046770602,
      "loss": 0.8564548492431641,
      "loss_ce": 0.00025371278752572834,
      "loss_iou": 0.357421875,
      "loss_num": 0.0281982421875,
      "loss_xval": 0.85546875,
      "num_input_tokens_seen": 141821112,
      "step": 2531
    },
    {
      "epoch": 5.639198218262806,
      "grad_norm": 28.660118103027344,
      "learning_rate": 1e-06,
      "loss": 0.6739,
      "num_input_tokens_seen": 141877664,
      "step": 2532
    },
    {
      "epoch": 5.639198218262806,
      "loss": 0.8321478366851807,
      "loss_ce": 0.0004828467790503055,
      "loss_iou": 0.30078125,
      "loss_num": 0.04541015625,
      "loss_xval": 0.83203125,
      "num_input_tokens_seen": 141877664,
      "step": 2532
    },
    {
      "epoch": 5.641425389755011,
      "grad_norm": 33.3441162109375,
      "learning_rate": 1e-06,
      "loss": 0.6838,
      "num_input_tokens_seen": 141934636,
      "step": 2533
    },
    {
      "epoch": 5.641425389755011,
      "loss": 0.7307108640670776,
      "loss_ce": 0.0002421102544758469,
      "loss_iou": 0.318359375,
      "loss_num": 0.0185546875,
      "loss_xval": 0.73046875,
      "num_input_tokens_seen": 141934636,
      "step": 2533
    },
    {
      "epoch": 5.643652561247216,
      "grad_norm": 22.631315231323242,
      "learning_rate": 1e-06,
      "loss": 0.8627,
      "num_input_tokens_seen": 141991128,
      "step": 2534
    },
    {
      "epoch": 5.643652561247216,
      "loss": 0.9368560910224915,
      "loss_ce": 0.0003326181322336197,
      "loss_iou": 0.3984375,
      "loss_num": 0.0283203125,
      "loss_xval": 0.9375,
      "num_input_tokens_seen": 141991128,
      "step": 2534
    },
    {
      "epoch": 5.645879732739421,
      "grad_norm": 13.450477600097656,
      "learning_rate": 1e-06,
      "loss": 0.61,
      "num_input_tokens_seen": 142045436,
      "step": 2535
    },
    {
      "epoch": 5.645879732739421,
      "loss": 0.5746713280677795,
      "loss_ce": 0.00020842923549935222,
      "loss_iou": 0.2470703125,
      "loss_num": 0.015869140625,
      "loss_xval": 0.57421875,
      "num_input_tokens_seen": 142045436,
      "step": 2535
    },
    {
      "epoch": 5.648106904231626,
      "grad_norm": 14.741578102111816,
      "learning_rate": 1e-06,
      "loss": 0.5904,
      "num_input_tokens_seen": 142102496,
      "step": 2536
    },
    {
      "epoch": 5.648106904231626,
      "loss": 0.4626786708831787,
      "loss_ce": 0.0005204671761021018,
      "loss_iou": 0.197265625,
      "loss_num": 0.01361083984375,
      "loss_xval": 0.462890625,
      "num_input_tokens_seen": 142102496,
      "step": 2536
    },
    {
      "epoch": 5.650334075723831,
      "grad_norm": 17.967416763305664,
      "learning_rate": 1e-06,
      "loss": 0.6889,
      "num_input_tokens_seen": 142160328,
      "step": 2537
    },
    {
      "epoch": 5.650334075723831,
      "loss": 0.5751903057098389,
      "loss_ce": 0.00017807658878155053,
      "loss_iou": 0.2265625,
      "loss_num": 0.0242919921875,
      "loss_xval": 0.57421875,
      "num_input_tokens_seen": 142160328,
      "step": 2537
    },
    {
      "epoch": 5.652561247216036,
      "grad_norm": 19.637807846069336,
      "learning_rate": 1e-06,
      "loss": 0.5937,
      "num_input_tokens_seen": 142215832,
      "step": 2538
    },
    {
      "epoch": 5.652561247216036,
      "loss": 0.5415201187133789,
      "loss_ce": 0.0002603091998025775,
      "loss_iou": 0.25,
      "loss_num": 0.0081787109375,
      "loss_xval": 0.54296875,
      "num_input_tokens_seen": 142215832,
      "step": 2538
    },
    {
      "epoch": 5.6547884187082404,
      "grad_norm": 27.54079246520996,
      "learning_rate": 1e-06,
      "loss": 0.8718,
      "num_input_tokens_seen": 142272748,
      "step": 2539
    },
    {
      "epoch": 5.6547884187082404,
      "loss": 0.9330868721008301,
      "loss_ce": 0.00022556885960511863,
      "loss_iou": 0.380859375,
      "loss_num": 0.0341796875,
      "loss_xval": 0.93359375,
      "num_input_tokens_seen": 142272748,
      "step": 2539
    },
    {
      "epoch": 5.657015590200445,
      "grad_norm": 23.226085662841797,
      "learning_rate": 1e-06,
      "loss": 0.6973,
      "num_input_tokens_seen": 142329648,
      "step": 2540
    },
    {
      "epoch": 5.657015590200445,
      "loss": 0.5698352456092834,
      "loss_ce": 0.00025519938208162785,
      "loss_iou": 0.232421875,
      "loss_num": 0.0208740234375,
      "loss_xval": 0.5703125,
      "num_input_tokens_seen": 142329648,
      "step": 2540
    },
    {
      "epoch": 5.65924276169265,
      "grad_norm": 28.16909408569336,
      "learning_rate": 1e-06,
      "loss": 0.6034,
      "num_input_tokens_seen": 142385652,
      "step": 2541
    },
    {
      "epoch": 5.65924276169265,
      "loss": 0.7484890222549438,
      "loss_ce": 0.00019804939802270383,
      "loss_iou": 0.3125,
      "loss_num": 0.0247802734375,
      "loss_xval": 0.75,
      "num_input_tokens_seen": 142385652,
      "step": 2541
    },
    {
      "epoch": 5.661469933184855,
      "grad_norm": 20.69179916381836,
      "learning_rate": 1e-06,
      "loss": 0.6642,
      "num_input_tokens_seen": 142439416,
      "step": 2542
    },
    {
      "epoch": 5.661469933184855,
      "loss": 0.7014844417572021,
      "loss_ce": 0.0001904680102597922,
      "loss_iou": 0.3046875,
      "loss_num": 0.018798828125,
      "loss_xval": 0.703125,
      "num_input_tokens_seen": 142439416,
      "step": 2542
    },
    {
      "epoch": 5.66369710467706,
      "grad_norm": 22.986743927001953,
      "learning_rate": 1e-06,
      "loss": 0.6988,
      "num_input_tokens_seen": 142494088,
      "step": 2543
    },
    {
      "epoch": 5.66369710467706,
      "loss": 0.6153843402862549,
      "loss_ce": 0.0003941247705370188,
      "loss_iou": 0.27734375,
      "loss_num": 0.01214599609375,
      "loss_xval": 0.61328125,
      "num_input_tokens_seen": 142494088,
      "step": 2543
    },
    {
      "epoch": 5.665924276169265,
      "grad_norm": 15.41435432434082,
      "learning_rate": 1e-06,
      "loss": 0.584,
      "num_input_tokens_seen": 142548856,
      "step": 2544
    },
    {
      "epoch": 5.665924276169265,
      "loss": 0.4866850674152374,
      "loss_ce": 0.0002348774141864851,
      "loss_iou": 0.21875,
      "loss_num": 0.00994873046875,
      "loss_xval": 0.486328125,
      "num_input_tokens_seen": 142548856,
      "step": 2544
    },
    {
      "epoch": 5.66815144766147,
      "grad_norm": 19.401214599609375,
      "learning_rate": 1e-06,
      "loss": 0.803,
      "num_input_tokens_seen": 142605360,
      "step": 2545
    },
    {
      "epoch": 5.66815144766147,
      "loss": 0.8005963563919067,
      "loss_ce": 0.0003033963148482144,
      "loss_iou": 0.345703125,
      "loss_num": 0.021728515625,
      "loss_xval": 0.80078125,
      "num_input_tokens_seen": 142605360,
      "step": 2545
    },
    {
      "epoch": 5.6703786191536745,
      "grad_norm": 15.095193862915039,
      "learning_rate": 1e-06,
      "loss": 0.7953,
      "num_input_tokens_seen": 142660940,
      "step": 2546
    },
    {
      "epoch": 5.6703786191536745,
      "loss": 0.6447733640670776,
      "loss_ce": 0.00024208366812672466,
      "loss_iou": 0.265625,
      "loss_num": 0.022705078125,
      "loss_xval": 0.64453125,
      "num_input_tokens_seen": 142660940,
      "step": 2546
    },
    {
      "epoch": 5.67260579064588,
      "grad_norm": 17.20547103881836,
      "learning_rate": 1e-06,
      "loss": 0.6511,
      "num_input_tokens_seen": 142717892,
      "step": 2547
    },
    {
      "epoch": 5.67260579064588,
      "loss": 0.6638010740280151,
      "loss_ce": 0.0002268501848448068,
      "loss_iou": 0.267578125,
      "loss_num": 0.0257568359375,
      "loss_xval": 0.6640625,
      "num_input_tokens_seen": 142717892,
      "step": 2547
    },
    {
      "epoch": 5.674832962138085,
      "grad_norm": 26.371519088745117,
      "learning_rate": 1e-06,
      "loss": 0.7449,
      "num_input_tokens_seen": 142769660,
      "step": 2548
    },
    {
      "epoch": 5.674832962138085,
      "loss": 0.6620455980300903,
      "loss_ce": 0.0003024160396307707,
      "loss_iou": 0.267578125,
      "loss_num": 0.025390625,
      "loss_xval": 0.66015625,
      "num_input_tokens_seen": 142769660,
      "step": 2548
    },
    {
      "epoch": 5.67706013363029,
      "grad_norm": 27.11166763305664,
      "learning_rate": 1e-06,
      "loss": 0.9627,
      "num_input_tokens_seen": 142825764,
      "step": 2549
    },
    {
      "epoch": 5.67706013363029,
      "loss": 1.015166997909546,
      "loss_ce": 0.00027431544731371105,
      "loss_iou": 0.3984375,
      "loss_num": 0.043701171875,
      "loss_xval": 1.015625,
      "num_input_tokens_seen": 142825764,
      "step": 2549
    },
    {
      "epoch": 5.679287305122495,
      "grad_norm": 14.629166603088379,
      "learning_rate": 1e-06,
      "loss": 0.6238,
      "num_input_tokens_seen": 142880500,
      "step": 2550
    },
    {
      "epoch": 5.679287305122495,
      "loss": 0.4876967668533325,
      "loss_ce": 0.00039207623922266066,
      "loss_iou": 0.2109375,
      "loss_num": 0.01318359375,
      "loss_xval": 0.48828125,
      "num_input_tokens_seen": 142880500,
      "step": 2550
    },
    {
      "epoch": 5.6815144766147,
      "grad_norm": 41.370121002197266,
      "learning_rate": 1e-06,
      "loss": 0.7918,
      "num_input_tokens_seen": 142932608,
      "step": 2551
    },
    {
      "epoch": 5.6815144766147,
      "loss": 0.9631957411766052,
      "loss_ce": 0.0004271506331861019,
      "loss_iou": 0.3984375,
      "loss_num": 0.033203125,
      "loss_xval": 0.9609375,
      "num_input_tokens_seen": 142932608,
      "step": 2551
    },
    {
      "epoch": 5.6837416481069045,
      "grad_norm": 17.59296417236328,
      "learning_rate": 1e-06,
      "loss": 0.6625,
      "num_input_tokens_seen": 142988896,
      "step": 2552
    },
    {
      "epoch": 5.6837416481069045,
      "loss": 0.51399827003479,
      "loss_ce": 0.0003264106926508248,
      "loss_iou": 0.2333984375,
      "loss_num": 0.00933837890625,
      "loss_xval": 0.515625,
      "num_input_tokens_seen": 142988896,
      "step": 2552
    },
    {
      "epoch": 5.685968819599109,
      "grad_norm": 28.058643341064453,
      "learning_rate": 1e-06,
      "loss": 0.7143,
      "num_input_tokens_seen": 143043040,
      "step": 2553
    },
    {
      "epoch": 5.685968819599109,
      "loss": 0.6291226744651794,
      "loss_ce": 0.00021646139794029295,
      "loss_iou": 0.2890625,
      "loss_num": 0.0098876953125,
      "loss_xval": 0.62890625,
      "num_input_tokens_seen": 143043040,
      "step": 2553
    },
    {
      "epoch": 5.688195991091314,
      "grad_norm": 25.84493064880371,
      "learning_rate": 1e-06,
      "loss": 0.6757,
      "num_input_tokens_seen": 143096936,
      "step": 2554
    },
    {
      "epoch": 5.688195991091314,
      "loss": 0.8105074167251587,
      "loss_ce": 0.0002046898298431188,
      "loss_iou": 0.33203125,
      "loss_num": 0.0291748046875,
      "loss_xval": 0.80859375,
      "num_input_tokens_seen": 143096936,
      "step": 2554
    },
    {
      "epoch": 5.690423162583519,
      "grad_norm": 21.50450897216797,
      "learning_rate": 1e-06,
      "loss": 0.5824,
      "num_input_tokens_seen": 143155644,
      "step": 2555
    },
    {
      "epoch": 5.690423162583519,
      "loss": 0.5691958665847778,
      "loss_ce": 0.00022610818268731236,
      "loss_iou": 0.2578125,
      "loss_num": 0.01043701171875,
      "loss_xval": 0.5703125,
      "num_input_tokens_seen": 143155644,
      "step": 2555
    },
    {
      "epoch": 5.692650334075724,
      "grad_norm": 18.279918670654297,
      "learning_rate": 1e-06,
      "loss": 0.6493,
      "num_input_tokens_seen": 143213008,
      "step": 2556
    },
    {
      "epoch": 5.692650334075724,
      "loss": 0.828661322593689,
      "loss_ce": 0.0002921561535913497,
      "loss_iou": 0.373046875,
      "loss_num": 0.0162353515625,
      "loss_xval": 0.828125,
      "num_input_tokens_seen": 143213008,
      "step": 2556
    },
    {
      "epoch": 5.694877505567929,
      "grad_norm": 16.526641845703125,
      "learning_rate": 1e-06,
      "loss": 0.6586,
      "num_input_tokens_seen": 143269428,
      "step": 2557
    },
    {
      "epoch": 5.694877505567929,
      "loss": 0.6889902353286743,
      "loss_ce": 0.00026959230308420956,
      "loss_iou": 0.2890625,
      "loss_num": 0.02197265625,
      "loss_xval": 0.6875,
      "num_input_tokens_seen": 143269428,
      "step": 2557
    },
    {
      "epoch": 5.697104677060134,
      "grad_norm": 16.533336639404297,
      "learning_rate": 1e-06,
      "loss": 0.5711,
      "num_input_tokens_seen": 143326208,
      "step": 2558
    },
    {
      "epoch": 5.697104677060134,
      "loss": 0.5453148484230042,
      "loss_ce": 0.0002708985994104296,
      "loss_iou": 0.23828125,
      "loss_num": 0.013671875,
      "loss_xval": 0.546875,
      "num_input_tokens_seen": 143326208,
      "step": 2558
    },
    {
      "epoch": 5.6993318485523385,
      "grad_norm": 85.07674407958984,
      "learning_rate": 1e-06,
      "loss": 0.7295,
      "num_input_tokens_seen": 143383116,
      "step": 2559
    },
    {
      "epoch": 5.6993318485523385,
      "loss": 0.5694315433502197,
      "loss_ce": 0.00021770322928205132,
      "loss_iou": 0.2236328125,
      "loss_num": 0.024169921875,
      "loss_xval": 0.5703125,
      "num_input_tokens_seen": 143383116,
      "step": 2559
    },
    {
      "epoch": 5.701559020044543,
      "grad_norm": 23.484725952148438,
      "learning_rate": 1e-06,
      "loss": 0.6961,
      "num_input_tokens_seen": 143437420,
      "step": 2560
    },
    {
      "epoch": 5.701559020044543,
      "loss": 0.736332893371582,
      "loss_ce": 0.0002489334437996149,
      "loss_iou": 0.28515625,
      "loss_num": 0.033203125,
      "loss_xval": 0.734375,
      "num_input_tokens_seen": 143437420,
      "step": 2560
    },
    {
      "epoch": 5.703786191536748,
      "grad_norm": 18.436580657958984,
      "learning_rate": 1e-06,
      "loss": 0.7144,
      "num_input_tokens_seen": 143493204,
      "step": 2561
    },
    {
      "epoch": 5.703786191536748,
      "loss": 0.8088289499282837,
      "loss_ce": 0.0002351756556890905,
      "loss_iou": 0.3671875,
      "loss_num": 0.01513671875,
      "loss_xval": 0.80859375,
      "num_input_tokens_seen": 143493204,
      "step": 2561
    },
    {
      "epoch": 5.706013363028953,
      "grad_norm": 406.9714660644531,
      "learning_rate": 1e-06,
      "loss": 0.8129,
      "num_input_tokens_seen": 143548960,
      "step": 2562
    },
    {
      "epoch": 5.706013363028953,
      "loss": 0.7016435861587524,
      "loss_ce": 0.00022760604042559862,
      "loss_iou": 0.318359375,
      "loss_num": 0.01287841796875,
      "loss_xval": 0.703125,
      "num_input_tokens_seen": 143548960,
      "step": 2562
    },
    {
      "epoch": 5.708240534521158,
      "grad_norm": 20.201797485351562,
      "learning_rate": 1e-06,
      "loss": 0.7235,
      "num_input_tokens_seen": 143606860,
      "step": 2563
    },
    {
      "epoch": 5.708240534521158,
      "loss": 0.6854332685470581,
      "loss_ce": 0.0003746763104572892,
      "loss_iou": 0.3125,
      "loss_num": 0.01190185546875,
      "loss_xval": 0.68359375,
      "num_input_tokens_seen": 143606860,
      "step": 2563
    },
    {
      "epoch": 5.710467706013363,
      "grad_norm": 17.661972045898438,
      "learning_rate": 1e-06,
      "loss": 0.6369,
      "num_input_tokens_seen": 143661512,
      "step": 2564
    },
    {
      "epoch": 5.710467706013363,
      "loss": 0.589299201965332,
      "loss_ce": 0.00018784166604746133,
      "loss_iou": 0.2578125,
      "loss_num": 0.01434326171875,
      "loss_xval": 0.58984375,
      "num_input_tokens_seen": 143661512,
      "step": 2564
    },
    {
      "epoch": 5.712694877505568,
      "grad_norm": 30.762678146362305,
      "learning_rate": 1e-06,
      "loss": 0.8426,
      "num_input_tokens_seen": 143719200,
      "step": 2565
    },
    {
      "epoch": 5.712694877505568,
      "loss": 0.8480957746505737,
      "loss_ce": 0.00019540796347428113,
      "loss_iou": 0.345703125,
      "loss_num": 0.03125,
      "loss_xval": 0.84765625,
      "num_input_tokens_seen": 143719200,
      "step": 2565
    },
    {
      "epoch": 5.714922048997773,
      "grad_norm": 18.01441764831543,
      "learning_rate": 1e-06,
      "loss": 0.7997,
      "num_input_tokens_seen": 143772696,
      "step": 2566
    },
    {
      "epoch": 5.714922048997773,
      "loss": 0.9778006076812744,
      "loss_ce": 0.00020047195721417665,
      "loss_iou": 0.400390625,
      "loss_num": 0.03515625,
      "loss_xval": 0.9765625,
      "num_input_tokens_seen": 143772696,
      "step": 2566
    },
    {
      "epoch": 5.717149220489977,
      "grad_norm": 19.236597061157227,
      "learning_rate": 1e-06,
      "loss": 0.6094,
      "num_input_tokens_seen": 143828632,
      "step": 2567
    },
    {
      "epoch": 5.717149220489977,
      "loss": 0.7114354372024536,
      "loss_ce": 0.0002537810942158103,
      "loss_iou": 0.30859375,
      "loss_num": 0.018798828125,
      "loss_xval": 0.7109375,
      "num_input_tokens_seen": 143828632,
      "step": 2567
    },
    {
      "epoch": 5.719376391982182,
      "grad_norm": 16.775983810424805,
      "learning_rate": 1e-06,
      "loss": 0.5711,
      "num_input_tokens_seen": 143882672,
      "step": 2568
    },
    {
      "epoch": 5.719376391982182,
      "loss": 0.6285956501960754,
      "loss_ce": 0.00017768185352906585,
      "loss_iou": 0.2734375,
      "loss_num": 0.0159912109375,
      "loss_xval": 0.62890625,
      "num_input_tokens_seen": 143882672,
      "step": 2568
    },
    {
      "epoch": 5.721603563474387,
      "grad_norm": 15.795881271362305,
      "learning_rate": 1e-06,
      "loss": 0.7436,
      "num_input_tokens_seen": 143938316,
      "step": 2569
    },
    {
      "epoch": 5.721603563474387,
      "loss": 0.8578826189041138,
      "loss_ce": 0.0002165931509807706,
      "loss_iou": 0.392578125,
      "loss_num": 0.0147705078125,
      "loss_xval": 0.859375,
      "num_input_tokens_seen": 143938316,
      "step": 2569
    },
    {
      "epoch": 5.723830734966592,
      "grad_norm": 21.437118530273438,
      "learning_rate": 1e-06,
      "loss": 0.7572,
      "num_input_tokens_seen": 143995348,
      "step": 2570
    },
    {
      "epoch": 5.723830734966592,
      "loss": 0.7881063222885132,
      "loss_ce": 0.00026450445875525475,
      "loss_iou": 0.345703125,
      "loss_num": 0.019287109375,
      "loss_xval": 0.7890625,
      "num_input_tokens_seen": 143995348,
      "step": 2570
    },
    {
      "epoch": 5.726057906458797,
      "grad_norm": 16.0108585357666,
      "learning_rate": 1e-06,
      "loss": 0.7053,
      "num_input_tokens_seen": 144052660,
      "step": 2571
    },
    {
      "epoch": 5.726057906458797,
      "loss": 0.58664470911026,
      "loss_ce": 0.00021891661162953824,
      "loss_iou": 0.251953125,
      "loss_num": 0.0164794921875,
      "loss_xval": 0.5859375,
      "num_input_tokens_seen": 144052660,
      "step": 2571
    },
    {
      "epoch": 5.728285077951003,
      "grad_norm": 26.33615493774414,
      "learning_rate": 1e-06,
      "loss": 0.6633,
      "num_input_tokens_seen": 144105148,
      "step": 2572
    },
    {
      "epoch": 5.728285077951003,
      "loss": 0.4833102822303772,
      "loss_ce": 0.00021702511003240943,
      "loss_iou": 0.205078125,
      "loss_num": 0.0146484375,
      "loss_xval": 0.482421875,
      "num_input_tokens_seen": 144105148,
      "step": 2572
    },
    {
      "epoch": 5.7305122494432075,
      "grad_norm": 21.177268981933594,
      "learning_rate": 1e-06,
      "loss": 0.9853,
      "num_input_tokens_seen": 144161172,
      "step": 2573
    },
    {
      "epoch": 5.7305122494432075,
      "loss": 1.158959150314331,
      "loss_ce": 0.0007559259538538754,
      "loss_iou": 0.48046875,
      "loss_num": 0.039306640625,
      "loss_xval": 1.15625,
      "num_input_tokens_seen": 144161172,
      "step": 2573
    },
    {
      "epoch": 5.732739420935412,
      "grad_norm": 25.209182739257812,
      "learning_rate": 1e-06,
      "loss": 0.6585,
      "num_input_tokens_seen": 144215804,
      "step": 2574
    },
    {
      "epoch": 5.732739420935412,
      "loss": 0.6748742461204529,
      "loss_ce": 0.00031366333132609725,
      "loss_iou": 0.291015625,
      "loss_num": 0.0185546875,
      "loss_xval": 0.67578125,
      "num_input_tokens_seen": 144215804,
      "step": 2574
    },
    {
      "epoch": 5.734966592427617,
      "grad_norm": 16.51605796813965,
      "learning_rate": 1e-06,
      "loss": 0.6987,
      "num_input_tokens_seen": 144273436,
      "step": 2575
    },
    {
      "epoch": 5.734966592427617,
      "loss": 0.564425528049469,
      "loss_ce": 0.00021653338626492769,
      "loss_iou": 0.248046875,
      "loss_num": 0.0135498046875,
      "loss_xval": 0.5625,
      "num_input_tokens_seen": 144273436,
      "step": 2575
    },
    {
      "epoch": 5.737193763919822,
      "grad_norm": 21.38966941833496,
      "learning_rate": 1e-06,
      "loss": 0.7136,
      "num_input_tokens_seen": 144330508,
      "step": 2576
    },
    {
      "epoch": 5.737193763919822,
      "loss": 0.7851696610450745,
      "loss_ce": 0.00025755877140909433,
      "loss_iou": 0.3125,
      "loss_num": 0.0322265625,
      "loss_xval": 0.78515625,
      "num_input_tokens_seen": 144330508,
      "step": 2576
    },
    {
      "epoch": 5.739420935412027,
      "grad_norm": 21.6599178314209,
      "learning_rate": 1e-06,
      "loss": 0.7422,
      "num_input_tokens_seen": 144387136,
      "step": 2577
    },
    {
      "epoch": 5.739420935412027,
      "loss": 0.8027091026306152,
      "loss_ce": 0.00021889799972996116,
      "loss_iou": 0.326171875,
      "loss_num": 0.0303955078125,
      "loss_xval": 0.80078125,
      "num_input_tokens_seen": 144387136,
      "step": 2577
    },
    {
      "epoch": 5.741648106904232,
      "grad_norm": 21.33940887451172,
      "learning_rate": 1e-06,
      "loss": 0.7533,
      "num_input_tokens_seen": 144441500,
      "step": 2578
    },
    {
      "epoch": 5.741648106904232,
      "loss": 0.6317015290260315,
      "loss_ce": 0.0002318147598998621,
      "loss_iou": 0.28515625,
      "loss_num": 0.012451171875,
      "loss_xval": 0.6328125,
      "num_input_tokens_seen": 144441500,
      "step": 2578
    },
    {
      "epoch": 5.743875278396437,
      "grad_norm": 18.354366302490234,
      "learning_rate": 1e-06,
      "loss": 0.6682,
      "num_input_tokens_seen": 144499348,
      "step": 2579
    },
    {
      "epoch": 5.743875278396437,
      "loss": 0.7804338335990906,
      "loss_ce": 0.0002824625698849559,
      "loss_iou": 0.28125,
      "loss_num": 0.043701171875,
      "loss_xval": 0.78125,
      "num_input_tokens_seen": 144499348,
      "step": 2579
    },
    {
      "epoch": 5.7461024498886415,
      "grad_norm": 20.35249900817871,
      "learning_rate": 1e-06,
      "loss": 0.8315,
      "num_input_tokens_seen": 144553584,
      "step": 2580
    },
    {
      "epoch": 5.7461024498886415,
      "loss": 0.8381065130233765,
      "loss_ce": 0.00021590120741166174,
      "loss_iou": 0.365234375,
      "loss_num": 0.0211181640625,
      "loss_xval": 0.8359375,
      "num_input_tokens_seen": 144553584,
      "step": 2580
    },
    {
      "epoch": 5.748329621380846,
      "grad_norm": 16.173368453979492,
      "learning_rate": 1e-06,
      "loss": 0.4855,
      "num_input_tokens_seen": 144610324,
      "step": 2581
    },
    {
      "epoch": 5.748329621380846,
      "loss": 0.5163002610206604,
      "loss_ce": 0.00018697154882829636,
      "loss_iou": 0.2353515625,
      "loss_num": 0.0091552734375,
      "loss_xval": 0.515625,
      "num_input_tokens_seen": 144610324,
      "step": 2581
    },
    {
      "epoch": 5.750556792873051,
      "grad_norm": 31.1335506439209,
      "learning_rate": 1e-06,
      "loss": 0.7519,
      "num_input_tokens_seen": 144665984,
      "step": 2582
    },
    {
      "epoch": 5.750556792873051,
      "loss": 0.5313022136688232,
      "loss_ce": 0.0002963669830933213,
      "loss_iou": 0.2353515625,
      "loss_num": 0.0118408203125,
      "loss_xval": 0.53125,
      "num_input_tokens_seen": 144665984,
      "step": 2582
    },
    {
      "epoch": 5.752783964365256,
      "grad_norm": 24.663740158081055,
      "learning_rate": 1e-06,
      "loss": 0.5619,
      "num_input_tokens_seen": 144722060,
      "step": 2583
    },
    {
      "epoch": 5.752783964365256,
      "loss": 0.6767467260360718,
      "loss_ce": 0.00023307063383981586,
      "loss_iou": 0.30078125,
      "loss_num": 0.015380859375,
      "loss_xval": 0.67578125,
      "num_input_tokens_seen": 144722060,
      "step": 2583
    },
    {
      "epoch": 5.755011135857461,
      "grad_norm": 13.527139663696289,
      "learning_rate": 1e-06,
      "loss": 0.5107,
      "num_input_tokens_seen": 144777528,
      "step": 2584
    },
    {
      "epoch": 5.755011135857461,
      "loss": 0.5506859421730042,
      "loss_ce": 0.0002709058462642133,
      "loss_iou": 0.2373046875,
      "loss_num": 0.01519775390625,
      "loss_xval": 0.55078125,
      "num_input_tokens_seen": 144777528,
      "step": 2584
    },
    {
      "epoch": 5.757238307349666,
      "grad_norm": 19.172088623046875,
      "learning_rate": 1e-06,
      "loss": 0.7847,
      "num_input_tokens_seen": 144833756,
      "step": 2585
    },
    {
      "epoch": 5.757238307349666,
      "loss": 0.7893670797348022,
      "loss_ce": 0.0003045589546673,
      "loss_iou": 0.330078125,
      "loss_num": 0.025634765625,
      "loss_xval": 0.7890625,
      "num_input_tokens_seen": 144833756,
      "step": 2585
    },
    {
      "epoch": 5.759465478841871,
      "grad_norm": 17.034088134765625,
      "learning_rate": 1e-06,
      "loss": 0.6934,
      "num_input_tokens_seen": 144892408,
      "step": 2586
    },
    {
      "epoch": 5.759465478841871,
      "loss": 0.6261861324310303,
      "loss_ce": 0.00020951575424987823,
      "loss_iou": 0.26953125,
      "loss_num": 0.0174560546875,
      "loss_xval": 0.625,
      "num_input_tokens_seen": 144892408,
      "step": 2586
    },
    {
      "epoch": 5.7616926503340755,
      "grad_norm": 22.121641159057617,
      "learning_rate": 1e-06,
      "loss": 0.7483,
      "num_input_tokens_seen": 144946344,
      "step": 2587
    },
    {
      "epoch": 5.7616926503340755,
      "loss": 0.8284502029418945,
      "loss_ce": 0.00032524295966140926,
      "loss_iou": 0.34765625,
      "loss_num": 0.026611328125,
      "loss_xval": 0.828125,
      "num_input_tokens_seen": 144946344,
      "step": 2587
    },
    {
      "epoch": 5.76391982182628,
      "grad_norm": 18.808324813842773,
      "learning_rate": 1e-06,
      "loss": 0.706,
      "num_input_tokens_seen": 145002944,
      "step": 2588
    },
    {
      "epoch": 5.76391982182628,
      "loss": 0.5951967239379883,
      "loss_ce": 0.00022597931092604995,
      "loss_iou": 0.2578125,
      "loss_num": 0.0157470703125,
      "loss_xval": 0.59375,
      "num_input_tokens_seen": 145002944,
      "step": 2588
    },
    {
      "epoch": 5.766146993318485,
      "grad_norm": 17.422155380249023,
      "learning_rate": 1e-06,
      "loss": 0.689,
      "num_input_tokens_seen": 145056836,
      "step": 2589
    },
    {
      "epoch": 5.766146993318485,
      "loss": 0.7689008712768555,
      "loss_ce": 0.00022416308638639748,
      "loss_iou": 0.310546875,
      "loss_num": 0.029541015625,
      "loss_xval": 0.76953125,
      "num_input_tokens_seen": 145056836,
      "step": 2589
    },
    {
      "epoch": 5.76837416481069,
      "grad_norm": 19.195316314697266,
      "learning_rate": 1e-06,
      "loss": 0.7372,
      "num_input_tokens_seen": 145113532,
      "step": 2590
    },
    {
      "epoch": 5.76837416481069,
      "loss": 0.8400857448577881,
      "loss_ce": 0.0002420624950900674,
      "loss_iou": 0.34375,
      "loss_num": 0.0303955078125,
      "loss_xval": 0.83984375,
      "num_input_tokens_seen": 145113532,
      "step": 2590
    },
    {
      "epoch": 5.770601336302895,
      "grad_norm": 18.72977638244629,
      "learning_rate": 1e-06,
      "loss": 0.7385,
      "num_input_tokens_seen": 145169580,
      "step": 2591
    },
    {
      "epoch": 5.770601336302895,
      "loss": 0.9092113375663757,
      "loss_ce": 0.0002757691836450249,
      "loss_iou": 0.361328125,
      "loss_num": 0.037841796875,
      "loss_xval": 0.91015625,
      "num_input_tokens_seen": 145169580,
      "step": 2591
    },
    {
      "epoch": 5.772828507795101,
      "grad_norm": 26.984228134155273,
      "learning_rate": 1e-06,
      "loss": 0.7739,
      "num_input_tokens_seen": 145225448,
      "step": 2592
    },
    {
      "epoch": 5.772828507795101,
      "loss": 0.6938083171844482,
      "loss_ce": 0.0004489576967898756,
      "loss_iou": 0.2890625,
      "loss_num": 0.023193359375,
      "loss_xval": 0.6953125,
      "num_input_tokens_seen": 145225448,
      "step": 2592
    },
    {
      "epoch": 5.775055679287306,
      "grad_norm": 26.926746368408203,
      "learning_rate": 1e-06,
      "loss": 0.8225,
      "num_input_tokens_seen": 145282560,
      "step": 2593
    },
    {
      "epoch": 5.775055679287306,
      "loss": 0.8718444108963013,
      "loss_ce": 0.00026242341846227646,
      "loss_iou": 0.37109375,
      "loss_num": 0.0262451171875,
      "loss_xval": 0.87109375,
      "num_input_tokens_seen": 145282560,
      "step": 2593
    },
    {
      "epoch": 5.77728285077951,
      "grad_norm": 17.646190643310547,
      "learning_rate": 1e-06,
      "loss": 0.7529,
      "num_input_tokens_seen": 145339244,
      "step": 2594
    },
    {
      "epoch": 5.77728285077951,
      "loss": 0.878664493560791,
      "loss_ce": 0.00024651282001286745,
      "loss_iou": 0.34375,
      "loss_num": 0.037841796875,
      "loss_xval": 0.87890625,
      "num_input_tokens_seen": 145339244,
      "step": 2594
    },
    {
      "epoch": 5.779510022271715,
      "grad_norm": 18.98468780517578,
      "learning_rate": 1e-06,
      "loss": 0.6249,
      "num_input_tokens_seen": 145394856,
      "step": 2595
    },
    {
      "epoch": 5.779510022271715,
      "loss": 0.6960242390632629,
      "loss_ce": 0.00022345452453009784,
      "loss_iou": 0.318359375,
      "loss_num": 0.01214599609375,
      "loss_xval": 0.6953125,
      "num_input_tokens_seen": 145394856,
      "step": 2595
    },
    {
      "epoch": 5.78173719376392,
      "grad_norm": 31.605703353881836,
      "learning_rate": 1e-06,
      "loss": 0.6959,
      "num_input_tokens_seen": 145450920,
      "step": 2596
    },
    {
      "epoch": 5.78173719376392,
      "loss": 0.6217830181121826,
      "loss_ce": 0.00020095528452657163,
      "loss_iou": 0.263671875,
      "loss_num": 0.0191650390625,
      "loss_xval": 0.62109375,
      "num_input_tokens_seen": 145450920,
      "step": 2596
    },
    {
      "epoch": 5.783964365256125,
      "grad_norm": 17.69574546813965,
      "learning_rate": 1e-06,
      "loss": 0.6234,
      "num_input_tokens_seen": 145504552,
      "step": 2597
    },
    {
      "epoch": 5.783964365256125,
      "loss": 0.8066386580467224,
      "loss_ce": 0.0002421344688627869,
      "loss_iou": 0.328125,
      "loss_num": 0.02978515625,
      "loss_xval": 0.8046875,
      "num_input_tokens_seen": 145504552,
      "step": 2597
    },
    {
      "epoch": 5.78619153674833,
      "grad_norm": 31.138301849365234,
      "learning_rate": 1e-06,
      "loss": 0.7631,
      "num_input_tokens_seen": 145559244,
      "step": 2598
    },
    {
      "epoch": 5.78619153674833,
      "loss": 0.7720120549201965,
      "loss_ce": 0.00028353132074698806,
      "loss_iou": 0.3515625,
      "loss_num": 0.01336669921875,
      "loss_xval": 0.7734375,
      "num_input_tokens_seen": 145559244,
      "step": 2598
    },
    {
      "epoch": 5.788418708240535,
      "grad_norm": 22.075687408447266,
      "learning_rate": 1e-06,
      "loss": 0.7088,
      "num_input_tokens_seen": 145617544,
      "step": 2599
    },
    {
      "epoch": 5.788418708240535,
      "loss": 0.8822649717330933,
      "loss_ce": 0.00042899814434349537,
      "loss_iou": 0.392578125,
      "loss_num": 0.0196533203125,
      "loss_xval": 0.8828125,
      "num_input_tokens_seen": 145617544,
      "step": 2599
    },
    {
      "epoch": 5.79064587973274,
      "grad_norm": 22.63301658630371,
      "learning_rate": 1e-06,
      "loss": 0.7705,
      "num_input_tokens_seen": 145675480,
      "step": 2600
    },
    {
      "epoch": 5.79064587973274,
      "loss": 0.7863603234291077,
      "loss_ce": 0.0002275097358506173,
      "loss_iou": 0.33203125,
      "loss_num": 0.024169921875,
      "loss_xval": 0.78515625,
      "num_input_tokens_seen": 145675480,
      "step": 2600
    },
    {
      "epoch": 5.7928730512249444,
      "grad_norm": 23.069360733032227,
      "learning_rate": 1e-06,
      "loss": 0.7148,
      "num_input_tokens_seen": 145726312,
      "step": 2601
    },
    {
      "epoch": 5.7928730512249444,
      "loss": 0.8112291097640991,
      "loss_ce": 0.00019390250963624567,
      "loss_iou": 0.349609375,
      "loss_num": 0.0224609375,
      "loss_xval": 0.8125,
      "num_input_tokens_seen": 145726312,
      "step": 2601
    },
    {
      "epoch": 5.795100222717149,
      "grad_norm": 27.435791015625,
      "learning_rate": 1e-06,
      "loss": 0.7378,
      "num_input_tokens_seen": 145783560,
      "step": 2602
    },
    {
      "epoch": 5.795100222717149,
      "loss": 0.7958614826202393,
      "loss_ce": 0.00020716458675451577,
      "loss_iou": 0.35546875,
      "loss_num": 0.01708984375,
      "loss_xval": 0.796875,
      "num_input_tokens_seen": 145783560,
      "step": 2602
    },
    {
      "epoch": 5.797327394209354,
      "grad_norm": 24.120868682861328,
      "learning_rate": 1e-06,
      "loss": 0.7342,
      "num_input_tokens_seen": 145836256,
      "step": 2603
    },
    {
      "epoch": 5.797327394209354,
      "loss": 0.8412899971008301,
      "loss_ce": 0.00028652820037677884,
      "loss_iou": 0.33203125,
      "loss_num": 0.035400390625,
      "loss_xval": 0.83984375,
      "num_input_tokens_seen": 145836256,
      "step": 2603
    },
    {
      "epoch": 5.799554565701559,
      "grad_norm": 28.57222557067871,
      "learning_rate": 1e-06,
      "loss": 0.7984,
      "num_input_tokens_seen": 145892380,
      "step": 2604
    },
    {
      "epoch": 5.799554565701559,
      "loss": 0.7485337853431702,
      "loss_ce": 0.00024279108038172126,
      "loss_iou": 0.318359375,
      "loss_num": 0.0228271484375,
      "loss_xval": 0.75,
      "num_input_tokens_seen": 145892380,
      "step": 2604
    },
    {
      "epoch": 5.801781737193764,
      "grad_norm": 30.485336303710938,
      "learning_rate": 1e-06,
      "loss": 0.7688,
      "num_input_tokens_seen": 145948016,
      "step": 2605
    },
    {
      "epoch": 5.801781737193764,
      "loss": 0.7458430528640747,
      "loss_ce": 0.0002376111369812861,
      "loss_iou": 0.3125,
      "loss_num": 0.02392578125,
      "loss_xval": 0.74609375,
      "num_input_tokens_seen": 145948016,
      "step": 2605
    },
    {
      "epoch": 5.804008908685969,
      "grad_norm": 18.45177459716797,
      "learning_rate": 1e-06,
      "loss": 0.8566,
      "num_input_tokens_seen": 146005492,
      "step": 2606
    },
    {
      "epoch": 5.804008908685969,
      "loss": 0.9877591133117676,
      "loss_ce": 0.00021025189198553562,
      "loss_iou": 0.40625,
      "loss_num": 0.034912109375,
      "loss_xval": 0.98828125,
      "num_input_tokens_seen": 146005492,
      "step": 2606
    },
    {
      "epoch": 5.806236080178174,
      "grad_norm": 23.423572540283203,
      "learning_rate": 1e-06,
      "loss": 0.5956,
      "num_input_tokens_seen": 146062328,
      "step": 2607
    },
    {
      "epoch": 5.806236080178174,
      "loss": 0.6337399482727051,
      "loss_ce": 0.00019499435438774526,
      "loss_iou": 0.28125,
      "loss_num": 0.013671875,
      "loss_xval": 0.6328125,
      "num_input_tokens_seen": 146062328,
      "step": 2607
    },
    {
      "epoch": 5.8084632516703785,
      "grad_norm": 23.9626407623291,
      "learning_rate": 1e-06,
      "loss": 0.8475,
      "num_input_tokens_seen": 146116388,
      "step": 2608
    },
    {
      "epoch": 5.8084632516703785,
      "loss": 0.6824861764907837,
      "loss_ce": 0.00023519776004832238,
      "loss_iou": 0.2470703125,
      "loss_num": 0.037841796875,
      "loss_xval": 0.68359375,
      "num_input_tokens_seen": 146116388,
      "step": 2608
    },
    {
      "epoch": 5.810690423162583,
      "grad_norm": 19.422828674316406,
      "learning_rate": 1e-06,
      "loss": 0.7589,
      "num_input_tokens_seen": 146172864,
      "step": 2609
    },
    {
      "epoch": 5.810690423162583,
      "loss": 0.7359454035758972,
      "loss_ce": 0.0002276489103678614,
      "loss_iou": 0.330078125,
      "loss_num": 0.0152587890625,
      "loss_xval": 0.734375,
      "num_input_tokens_seen": 146172864,
      "step": 2609
    },
    {
      "epoch": 5.812917594654788,
      "grad_norm": 26.152000427246094,
      "learning_rate": 1e-06,
      "loss": 0.6478,
      "num_input_tokens_seen": 146230640,
      "step": 2610
    },
    {
      "epoch": 5.812917594654788,
      "loss": 0.7595158815383911,
      "loss_ce": 0.00023854889150243253,
      "loss_iou": 0.310546875,
      "loss_num": 0.0279541015625,
      "loss_xval": 0.7578125,
      "num_input_tokens_seen": 146230640,
      "step": 2610
    },
    {
      "epoch": 5.815144766146993,
      "grad_norm": 40.59634780883789,
      "learning_rate": 1e-06,
      "loss": 0.5484,
      "num_input_tokens_seen": 146287508,
      "step": 2611
    },
    {
      "epoch": 5.815144766146993,
      "loss": 0.6367042660713196,
      "loss_ce": 0.00022962281946092844,
      "loss_iou": 0.259765625,
      "loss_num": 0.023193359375,
      "loss_xval": 0.63671875,
      "num_input_tokens_seen": 146287508,
      "step": 2611
    },
    {
      "epoch": 5.817371937639198,
      "grad_norm": 16.996164321899414,
      "learning_rate": 1e-06,
      "loss": 0.8022,
      "num_input_tokens_seen": 146343108,
      "step": 2612
    },
    {
      "epoch": 5.817371937639198,
      "loss": 0.7761521339416504,
      "loss_ce": 0.00027319122455082834,
      "loss_iou": 0.310546875,
      "loss_num": 0.03125,
      "loss_xval": 0.77734375,
      "num_input_tokens_seen": 146343108,
      "step": 2612
    },
    {
      "epoch": 5.819599109131403,
      "grad_norm": 14.195480346679688,
      "learning_rate": 1e-06,
      "loss": 0.5662,
      "num_input_tokens_seen": 146401944,
      "step": 2613
    },
    {
      "epoch": 5.819599109131403,
      "loss": 0.6162877082824707,
      "loss_ce": 0.00032092921901494265,
      "loss_iou": 0.2421875,
      "loss_num": 0.0262451171875,
      "loss_xval": 0.6171875,
      "num_input_tokens_seen": 146401944,
      "step": 2613
    },
    {
      "epoch": 5.821826280623608,
      "grad_norm": 13.408287048339844,
      "learning_rate": 1e-06,
      "loss": 0.676,
      "num_input_tokens_seen": 146459796,
      "step": 2614
    },
    {
      "epoch": 5.821826280623608,
      "loss": 0.6257205009460449,
      "loss_ce": 0.00023219409922603518,
      "loss_iou": 0.26953125,
      "loss_num": 0.0174560546875,
      "loss_xval": 0.625,
      "num_input_tokens_seen": 146459796,
      "step": 2614
    },
    {
      "epoch": 5.8240534521158125,
      "grad_norm": 37.40284729003906,
      "learning_rate": 1e-06,
      "loss": 0.6556,
      "num_input_tokens_seen": 146515484,
      "step": 2615
    },
    {
      "epoch": 5.8240534521158125,
      "loss": 0.6754716634750366,
      "loss_ce": 0.00017868283612187952,
      "loss_iou": 0.27734375,
      "loss_num": 0.023681640625,
      "loss_xval": 0.67578125,
      "num_input_tokens_seen": 146515484,
      "step": 2615
    },
    {
      "epoch": 5.826280623608017,
      "grad_norm": 22.057910919189453,
      "learning_rate": 1e-06,
      "loss": 0.7252,
      "num_input_tokens_seen": 146570828,
      "step": 2616
    },
    {
      "epoch": 5.826280623608017,
      "loss": 0.7309498190879822,
      "loss_ce": 0.00023692671675235033,
      "loss_iou": 0.296875,
      "loss_num": 0.02734375,
      "loss_xval": 0.73046875,
      "num_input_tokens_seen": 146570828,
      "step": 2616
    },
    {
      "epoch": 5.828507795100223,
      "grad_norm": 63.05039596557617,
      "learning_rate": 1e-06,
      "loss": 0.577,
      "num_input_tokens_seen": 146627412,
      "step": 2617
    },
    {
      "epoch": 5.828507795100223,
      "loss": 0.5437142848968506,
      "loss_ce": 0.0002571970981080085,
      "loss_iou": 0.23828125,
      "loss_num": 0.01336669921875,
      "loss_xval": 0.54296875,
      "num_input_tokens_seen": 146627412,
      "step": 2617
    },
    {
      "epoch": 5.830734966592428,
      "grad_norm": 29.350282669067383,
      "learning_rate": 1e-06,
      "loss": 0.6785,
      "num_input_tokens_seen": 146684332,
      "step": 2618
    },
    {
      "epoch": 5.830734966592428,
      "loss": 0.5683233141899109,
      "loss_ce": 0.00020807163673453033,
      "loss_iou": 0.236328125,
      "loss_num": 0.01904296875,
      "loss_xval": 0.56640625,
      "num_input_tokens_seen": 146684332,
      "step": 2618
    },
    {
      "epoch": 5.832962138084633,
      "grad_norm": 23.34769058227539,
      "learning_rate": 1e-06,
      "loss": 0.6697,
      "num_input_tokens_seen": 146737280,
      "step": 2619
    },
    {
      "epoch": 5.832962138084633,
      "loss": 0.6157218217849731,
      "loss_ce": 0.0004874598525930196,
      "loss_iou": 0.271484375,
      "loss_num": 0.01446533203125,
      "loss_xval": 0.6171875,
      "num_input_tokens_seen": 146737280,
      "step": 2619
    },
    {
      "epoch": 5.835189309576838,
      "grad_norm": 22.184803009033203,
      "learning_rate": 1e-06,
      "loss": 0.6025,
      "num_input_tokens_seen": 146794400,
      "step": 2620
    },
    {
      "epoch": 5.835189309576838,
      "loss": 0.6838958263397217,
      "loss_ce": 0.0003020889707840979,
      "loss_iou": 0.3125,
      "loss_num": 0.0115966796875,
      "loss_xval": 0.68359375,
      "num_input_tokens_seen": 146794400,
      "step": 2620
    },
    {
      "epoch": 5.8374164810690425,
      "grad_norm": 16.25792694091797,
      "learning_rate": 1e-06,
      "loss": 0.71,
      "num_input_tokens_seen": 146849192,
      "step": 2621
    },
    {
      "epoch": 5.8374164810690425,
      "loss": 0.6701442003250122,
      "loss_ce": 0.00022233014169614762,
      "loss_iou": 0.25390625,
      "loss_num": 0.032470703125,
      "loss_xval": 0.671875,
      "num_input_tokens_seen": 146849192,
      "step": 2621
    },
    {
      "epoch": 5.839643652561247,
      "grad_norm": 17.911888122558594,
      "learning_rate": 1e-06,
      "loss": 0.6373,
      "num_input_tokens_seen": 146906416,
      "step": 2622
    },
    {
      "epoch": 5.839643652561247,
      "loss": 0.5712316632270813,
      "loss_ce": 0.00018675101455301046,
      "loss_iou": 0.2421875,
      "loss_num": 0.0172119140625,
      "loss_xval": 0.5703125,
      "num_input_tokens_seen": 146906416,
      "step": 2622
    },
    {
      "epoch": 5.841870824053452,
      "grad_norm": 15.35204029083252,
      "learning_rate": 1e-06,
      "loss": 0.6543,
      "num_input_tokens_seen": 146963888,
      "step": 2623
    },
    {
      "epoch": 5.841870824053452,
      "loss": 0.6473047733306885,
      "loss_ce": 0.00021000676497351378,
      "loss_iou": 0.265625,
      "loss_num": 0.0233154296875,
      "loss_xval": 0.6484375,
      "num_input_tokens_seen": 146963888,
      "step": 2623
    },
    {
      "epoch": 5.844097995545657,
      "grad_norm": 19.907550811767578,
      "learning_rate": 1e-06,
      "loss": 0.8551,
      "num_input_tokens_seen": 147019984,
      "step": 2624
    },
    {
      "epoch": 5.844097995545657,
      "loss": 0.6151929497718811,
      "loss_ce": 0.0002027209848165512,
      "loss_iou": 0.26953125,
      "loss_num": 0.01483154296875,
      "loss_xval": 0.61328125,
      "num_input_tokens_seen": 147019984,
      "step": 2624
    },
    {
      "epoch": 5.846325167037862,
      "grad_norm": 35.5843391418457,
      "learning_rate": 1e-06,
      "loss": 0.8085,
      "num_input_tokens_seen": 147074124,
      "step": 2625
    },
    {
      "epoch": 5.846325167037862,
      "loss": 0.6340126991271973,
      "loss_ce": 0.00022365737822838128,
      "loss_iou": 0.283203125,
      "loss_num": 0.0135498046875,
      "loss_xval": 0.6328125,
      "num_input_tokens_seen": 147074124,
      "step": 2625
    },
    {
      "epoch": 5.848552338530067,
      "grad_norm": 22.104284286499023,
      "learning_rate": 1e-06,
      "loss": 0.7518,
      "num_input_tokens_seen": 147129964,
      "step": 2626
    },
    {
      "epoch": 5.848552338530067,
      "loss": 0.7092376947402954,
      "loss_ce": 0.0002533411025069654,
      "loss_iou": 0.310546875,
      "loss_num": 0.017578125,
      "loss_xval": 0.7109375,
      "num_input_tokens_seen": 147129964,
      "step": 2626
    },
    {
      "epoch": 5.850779510022272,
      "grad_norm": 18.39105987548828,
      "learning_rate": 1e-06,
      "loss": 0.6966,
      "num_input_tokens_seen": 147184592,
      "step": 2627
    },
    {
      "epoch": 5.850779510022272,
      "loss": 0.661689043045044,
      "loss_ce": 0.0003120756009593606,
      "loss_iou": 0.27734375,
      "loss_num": 0.021728515625,
      "loss_xval": 0.66015625,
      "num_input_tokens_seen": 147184592,
      "step": 2627
    },
    {
      "epoch": 5.853006681514477,
      "grad_norm": 22.63771629333496,
      "learning_rate": 1e-06,
      "loss": 0.9064,
      "num_input_tokens_seen": 147241272,
      "step": 2628
    },
    {
      "epoch": 5.853006681514477,
      "loss": 0.5712363719940186,
      "loss_ce": 0.00019146442355122417,
      "loss_iou": 0.251953125,
      "loss_num": 0.0135498046875,
      "loss_xval": 0.5703125,
      "num_input_tokens_seen": 147241272,
      "step": 2628
    },
    {
      "epoch": 5.855233853006681,
      "grad_norm": 21.009822845458984,
      "learning_rate": 1e-06,
      "loss": 0.8111,
      "num_input_tokens_seen": 147294076,
      "step": 2629
    },
    {
      "epoch": 5.855233853006681,
      "loss": 0.9777696132659912,
      "loss_ce": 0.00023052690085023642,
      "loss_iou": 0.443359375,
      "loss_num": 0.0179443359375,
      "loss_xval": 0.9765625,
      "num_input_tokens_seen": 147294076,
      "step": 2629
    },
    {
      "epoch": 5.857461024498886,
      "grad_norm": 19.79799461364746,
      "learning_rate": 1e-06,
      "loss": 0.7446,
      "num_input_tokens_seen": 147349408,
      "step": 2630
    },
    {
      "epoch": 5.857461024498886,
      "loss": 0.6237329244613647,
      "loss_ce": 0.00019774649990722537,
      "loss_iou": 0.263671875,
      "loss_num": 0.01904296875,
      "loss_xval": 0.625,
      "num_input_tokens_seen": 147349408,
      "step": 2630
    },
    {
      "epoch": 5.859688195991091,
      "grad_norm": 22.230148315429688,
      "learning_rate": 1e-06,
      "loss": 0.7763,
      "num_input_tokens_seen": 147405948,
      "step": 2631
    },
    {
      "epoch": 5.859688195991091,
      "loss": 0.8681101202964783,
      "loss_ce": 0.0003122506313957274,
      "loss_iou": 0.37890625,
      "loss_num": 0.021728515625,
      "loss_xval": 0.8671875,
      "num_input_tokens_seen": 147405948,
      "step": 2631
    },
    {
      "epoch": 5.861915367483296,
      "grad_norm": 16.93956756591797,
      "learning_rate": 1e-06,
      "loss": 0.7856,
      "num_input_tokens_seen": 147460328,
      "step": 2632
    },
    {
      "epoch": 5.861915367483296,
      "loss": 0.8456612825393677,
      "loss_ce": 0.00020235308329574764,
      "loss_iou": 0.34375,
      "loss_num": 0.031494140625,
      "loss_xval": 0.84375,
      "num_input_tokens_seen": 147460328,
      "step": 2632
    },
    {
      "epoch": 5.864142538975501,
      "grad_norm": 28.430946350097656,
      "learning_rate": 1e-06,
      "loss": 0.7788,
      "num_input_tokens_seen": 147515700,
      "step": 2633
    },
    {
      "epoch": 5.864142538975501,
      "loss": 0.8373937606811523,
      "loss_ce": 0.00023549118486698717,
      "loss_iou": 0.369140625,
      "loss_num": 0.0198974609375,
      "loss_xval": 0.8359375,
      "num_input_tokens_seen": 147515700,
      "step": 2633
    },
    {
      "epoch": 5.866369710467706,
      "grad_norm": 111.71736145019531,
      "learning_rate": 1e-06,
      "loss": 0.9358,
      "num_input_tokens_seen": 147570320,
      "step": 2634
    },
    {
      "epoch": 5.866369710467706,
      "loss": 0.9798734188079834,
      "loss_ce": 0.0003812336944974959,
      "loss_iou": 0.400390625,
      "loss_num": 0.035888671875,
      "loss_xval": 0.98046875,
      "num_input_tokens_seen": 147570320,
      "step": 2634
    },
    {
      "epoch": 5.868596881959911,
      "grad_norm": 20.809370040893555,
      "learning_rate": 1e-06,
      "loss": 0.7524,
      "num_input_tokens_seen": 147625424,
      "step": 2635
    },
    {
      "epoch": 5.868596881959911,
      "loss": 0.6215413808822632,
      "loss_ce": 0.00020347216923255473,
      "loss_iou": 0.255859375,
      "loss_num": 0.021728515625,
      "loss_xval": 0.62109375,
      "num_input_tokens_seen": 147625424,
      "step": 2635
    },
    {
      "epoch": 5.870824053452115,
      "grad_norm": 23.472732543945312,
      "learning_rate": 1e-06,
      "loss": 0.896,
      "num_input_tokens_seen": 147682576,
      "step": 2636
    },
    {
      "epoch": 5.870824053452115,
      "loss": 0.7243614196777344,
      "loss_ce": 0.00024032902729231864,
      "loss_iou": 0.322265625,
      "loss_num": 0.0159912109375,
      "loss_xval": 0.72265625,
      "num_input_tokens_seen": 147682576,
      "step": 2636
    },
    {
      "epoch": 5.873051224944321,
      "grad_norm": 19.540515899658203,
      "learning_rate": 1e-06,
      "loss": 0.8665,
      "num_input_tokens_seen": 147738480,
      "step": 2637
    },
    {
      "epoch": 5.873051224944321,
      "loss": 0.94027179479599,
      "loss_ce": 0.0003304004785604775,
      "loss_iou": 0.33984375,
      "loss_num": 0.0517578125,
      "loss_xval": 0.94140625,
      "num_input_tokens_seen": 147738480,
      "step": 2637
    },
    {
      "epoch": 5.875278396436526,
      "grad_norm": 18.206157684326172,
      "learning_rate": 1e-06,
      "loss": 0.9815,
      "num_input_tokens_seen": 147795268,
      "step": 2638
    },
    {
      "epoch": 5.875278396436526,
      "loss": 0.8137099146842957,
      "loss_ce": 0.00023330387193709612,
      "loss_iou": 0.328125,
      "loss_num": 0.03125,
      "loss_xval": 0.8125,
      "num_input_tokens_seen": 147795268,
      "step": 2638
    },
    {
      "epoch": 5.877505567928731,
      "grad_norm": 36.962310791015625,
      "learning_rate": 1e-06,
      "loss": 0.8831,
      "num_input_tokens_seen": 147851048,
      "step": 2639
    },
    {
      "epoch": 5.877505567928731,
      "loss": 1.2098288536071777,
      "loss_ce": 0.0003562095225788653,
      "loss_iou": 0.5234375,
      "loss_num": 0.03271484375,
      "loss_xval": 1.2109375,
      "num_input_tokens_seen": 147851048,
      "step": 2639
    },
    {
      "epoch": 5.879732739420936,
      "grad_norm": 20.242557525634766,
      "learning_rate": 1e-06,
      "loss": 0.6815,
      "num_input_tokens_seen": 147908012,
      "step": 2640
    },
    {
      "epoch": 5.879732739420936,
      "loss": 0.6303223371505737,
      "loss_ce": 0.0001953527971636504,
      "loss_iou": 0.27734375,
      "loss_num": 0.01495361328125,
      "loss_xval": 0.62890625,
      "num_input_tokens_seen": 147908012,
      "step": 2640
    },
    {
      "epoch": 5.881959910913141,
      "grad_norm": 34.177433013916016,
      "learning_rate": 1e-06,
      "loss": 0.7303,
      "num_input_tokens_seen": 147960696,
      "step": 2641
    },
    {
      "epoch": 5.881959910913141,
      "loss": 0.9157909154891968,
      "loss_ce": 0.000263590132817626,
      "loss_iou": 0.39453125,
      "loss_num": 0.0255126953125,
      "loss_xval": 0.9140625,
      "num_input_tokens_seen": 147960696,
      "step": 2641
    },
    {
      "epoch": 5.8841870824053455,
      "grad_norm": 19.002622604370117,
      "learning_rate": 1e-06,
      "loss": 0.6177,
      "num_input_tokens_seen": 148016204,
      "step": 2642
    },
    {
      "epoch": 5.8841870824053455,
      "loss": 0.528780460357666,
      "loss_ce": 0.0002160129661206156,
      "loss_iou": 0.2265625,
      "loss_num": 0.0150146484375,
      "loss_xval": 0.52734375,
      "num_input_tokens_seen": 148016204,
      "step": 2642
    },
    {
      "epoch": 5.88641425389755,
      "grad_norm": 20.485605239868164,
      "learning_rate": 1e-06,
      "loss": 0.7832,
      "num_input_tokens_seen": 148072492,
      "step": 2643
    },
    {
      "epoch": 5.88641425389755,
      "loss": 1.044980764389038,
      "loss_ce": 0.0015236493200063705,
      "loss_iou": 0.455078125,
      "loss_num": 0.0269775390625,
      "loss_xval": 1.046875,
      "num_input_tokens_seen": 148072492,
      "step": 2643
    },
    {
      "epoch": 5.888641425389755,
      "grad_norm": 25.896760940551758,
      "learning_rate": 1e-06,
      "loss": 0.6733,
      "num_input_tokens_seen": 148124560,
      "step": 2644
    },
    {
      "epoch": 5.888641425389755,
      "loss": 0.5408051013946533,
      "loss_ce": 0.00021672301227226853,
      "loss_iou": 0.248046875,
      "loss_num": 0.00909423828125,
      "loss_xval": 0.5390625,
      "num_input_tokens_seen": 148124560,
      "step": 2644
    },
    {
      "epoch": 5.89086859688196,
      "grad_norm": 15.1829252243042,
      "learning_rate": 1e-06,
      "loss": 0.7851,
      "num_input_tokens_seen": 148182164,
      "step": 2645
    },
    {
      "epoch": 5.89086859688196,
      "loss": 0.6976854801177979,
      "loss_ce": 0.000297736085485667,
      "loss_iou": 0.287109375,
      "loss_num": 0.0242919921875,
      "loss_xval": 0.69921875,
      "num_input_tokens_seen": 148182164,
      "step": 2645
    },
    {
      "epoch": 5.893095768374165,
      "grad_norm": 20.109758377075195,
      "learning_rate": 1e-06,
      "loss": 0.4663,
      "num_input_tokens_seen": 148237212,
      "step": 2646
    },
    {
      "epoch": 5.893095768374165,
      "loss": 0.403061181306839,
      "loss_ce": 0.00016811591922305524,
      "loss_iou": 0.1708984375,
      "loss_num": 0.0120849609375,
      "loss_xval": 0.40234375,
      "num_input_tokens_seen": 148237212,
      "step": 2646
    },
    {
      "epoch": 5.89532293986637,
      "grad_norm": 22.928607940673828,
      "learning_rate": 1e-06,
      "loss": 0.7238,
      "num_input_tokens_seen": 148294320,
      "step": 2647
    },
    {
      "epoch": 5.89532293986637,
      "loss": 0.7370385527610779,
      "loss_ce": 0.00022214508499018848,
      "loss_iou": 0.3203125,
      "loss_num": 0.01953125,
      "loss_xval": 0.73828125,
      "num_input_tokens_seen": 148294320,
      "step": 2647
    },
    {
      "epoch": 5.897550111358575,
      "grad_norm": 18.266637802124023,
      "learning_rate": 1e-06,
      "loss": 0.7101,
      "num_input_tokens_seen": 148350012,
      "step": 2648
    },
    {
      "epoch": 5.897550111358575,
      "loss": 0.841547429561615,
      "loss_ce": 0.00023882737150415778,
      "loss_iou": 0.357421875,
      "loss_num": 0.0250244140625,
      "loss_xval": 0.83984375,
      "num_input_tokens_seen": 148350012,
      "step": 2648
    },
    {
      "epoch": 5.8997772828507795,
      "grad_norm": 16.207386016845703,
      "learning_rate": 1e-06,
      "loss": 0.7109,
      "num_input_tokens_seen": 148405388,
      "step": 2649
    },
    {
      "epoch": 5.8997772828507795,
      "loss": 0.6293390989303589,
      "loss_ce": 0.00018869154155254364,
      "loss_iou": 0.275390625,
      "loss_num": 0.0159912109375,
      "loss_xval": 0.62890625,
      "num_input_tokens_seen": 148405388,
      "step": 2649
    },
    {
      "epoch": 5.902004454342984,
      "grad_norm": 16.40260124206543,
      "learning_rate": 1e-06,
      "loss": 0.4936,
      "num_input_tokens_seen": 148462940,
      "step": 2650
    },
    {
      "epoch": 5.902004454342984,
      "loss": 0.45450687408447266,
      "loss_ce": 0.00016120026702992618,
      "loss_iou": 0.1953125,
      "loss_num": 0.01287841796875,
      "loss_xval": 0.455078125,
      "num_input_tokens_seen": 148462940,
      "step": 2650
    },
    {
      "epoch": 5.904231625835189,
      "grad_norm": 22.90458869934082,
      "learning_rate": 1e-06,
      "loss": 0.7208,
      "num_input_tokens_seen": 148518860,
      "step": 2651
    },
    {
      "epoch": 5.904231625835189,
      "loss": 0.7937043309211731,
      "loss_ce": 0.00024726998526602983,
      "loss_iou": 0.34375,
      "loss_num": 0.020751953125,
      "loss_xval": 0.79296875,
      "num_input_tokens_seen": 148518860,
      "step": 2651
    },
    {
      "epoch": 5.906458797327394,
      "grad_norm": 15.67951488494873,
      "learning_rate": 1e-06,
      "loss": 0.478,
      "num_input_tokens_seen": 148577372,
      "step": 2652
    },
    {
      "epoch": 5.906458797327394,
      "loss": 0.41467005014419556,
      "loss_ce": 0.0002413251786492765,
      "loss_iou": 0.1962890625,
      "loss_num": 0.00433349609375,
      "loss_xval": 0.4140625,
      "num_input_tokens_seen": 148577372,
      "step": 2652
    },
    {
      "epoch": 5.908685968819599,
      "grad_norm": 23.35599708557129,
      "learning_rate": 1e-06,
      "loss": 0.698,
      "num_input_tokens_seen": 148633988,
      "step": 2653
    },
    {
      "epoch": 5.908685968819599,
      "loss": 0.7189466953277588,
      "loss_ce": 0.0001966924173757434,
      "loss_iou": 0.306640625,
      "loss_num": 0.0211181640625,
      "loss_xval": 0.71875,
      "num_input_tokens_seen": 148633988,
      "step": 2653
    },
    {
      "epoch": 5.910913140311804,
      "grad_norm": 22.419492721557617,
      "learning_rate": 1e-06,
      "loss": 0.8223,
      "num_input_tokens_seen": 148686904,
      "step": 2654
    },
    {
      "epoch": 5.910913140311804,
      "loss": 0.5142409205436707,
      "loss_ce": 0.0003248959837947041,
      "loss_iou": 0.2138671875,
      "loss_num": 0.01708984375,
      "loss_xval": 0.515625,
      "num_input_tokens_seen": 148686904,
      "step": 2654
    },
    {
      "epoch": 5.913140311804009,
      "grad_norm": 22.227079391479492,
      "learning_rate": 1e-06,
      "loss": 0.5734,
      "num_input_tokens_seen": 148741860,
      "step": 2655
    },
    {
      "epoch": 5.913140311804009,
      "loss": 0.6701195240020752,
      "loss_ce": 0.00019767896446865052,
      "loss_iou": 0.2578125,
      "loss_num": 0.0308837890625,
      "loss_xval": 0.671875,
      "num_input_tokens_seen": 148741860,
      "step": 2655
    },
    {
      "epoch": 5.9153674832962135,
      "grad_norm": 24.240558624267578,
      "learning_rate": 1e-06,
      "loss": 0.7875,
      "num_input_tokens_seen": 148798228,
      "step": 2656
    },
    {
      "epoch": 5.9153674832962135,
      "loss": 0.7565721869468689,
      "loss_ce": 0.0002245493233203888,
      "loss_iou": 0.3203125,
      "loss_num": 0.0228271484375,
      "loss_xval": 0.7578125,
      "num_input_tokens_seen": 148798228,
      "step": 2656
    },
    {
      "epoch": 5.917594654788418,
      "grad_norm": 21.239978790283203,
      "learning_rate": 1e-06,
      "loss": 0.8062,
      "num_input_tokens_seen": 148852728,
      "step": 2657
    },
    {
      "epoch": 5.917594654788418,
      "loss": 0.9770278930664062,
      "loss_ce": 0.00022124522365629673,
      "loss_iou": 0.40234375,
      "loss_num": 0.034423828125,
      "loss_xval": 0.9765625,
      "num_input_tokens_seen": 148852728,
      "step": 2657
    },
    {
      "epoch": 5.919821826280623,
      "grad_norm": 21.198122024536133,
      "learning_rate": 1e-06,
      "loss": 0.5775,
      "num_input_tokens_seen": 148910144,
      "step": 2658
    },
    {
      "epoch": 5.919821826280623,
      "loss": 0.5937119126319885,
      "loss_ce": 0.00020606812904588878,
      "loss_iou": 0.25,
      "loss_num": 0.0186767578125,
      "loss_xval": 0.59375,
      "num_input_tokens_seen": 148910144,
      "step": 2658
    },
    {
      "epoch": 5.922048997772828,
      "grad_norm": 21.191734313964844,
      "learning_rate": 1e-06,
      "loss": 0.6275,
      "num_input_tokens_seen": 148965064,
      "step": 2659
    },
    {
      "epoch": 5.922048997772828,
      "loss": 0.5399864912033081,
      "loss_ce": 0.0001916133624035865,
      "loss_iou": 0.244140625,
      "loss_num": 0.01031494140625,
      "loss_xval": 0.5390625,
      "num_input_tokens_seen": 148965064,
      "step": 2659
    },
    {
      "epoch": 5.924276169265033,
      "grad_norm": 22.334716796875,
      "learning_rate": 1e-06,
      "loss": 0.881,
      "num_input_tokens_seen": 149019216,
      "step": 2660
    },
    {
      "epoch": 5.924276169265033,
      "loss": 0.7107348442077637,
      "loss_ce": 0.00022451579570770264,
      "loss_iou": 0.298828125,
      "loss_num": 0.0228271484375,
      "loss_xval": 0.7109375,
      "num_input_tokens_seen": 149019216,
      "step": 2660
    },
    {
      "epoch": 5.926503340757238,
      "grad_norm": 26.160076141357422,
      "learning_rate": 1e-06,
      "loss": 0.6856,
      "num_input_tokens_seen": 149076832,
      "step": 2661
    },
    {
      "epoch": 5.926503340757238,
      "loss": 0.7112481594085693,
      "loss_ce": 0.0001885854871943593,
      "loss_iou": 0.294921875,
      "loss_num": 0.0245361328125,
      "loss_xval": 0.7109375,
      "num_input_tokens_seen": 149076832,
      "step": 2661
    },
    {
      "epoch": 5.928730512249444,
      "grad_norm": 35.00899887084961,
      "learning_rate": 1e-06,
      "loss": 0.7615,
      "num_input_tokens_seen": 149133992,
      "step": 2662
    },
    {
      "epoch": 5.928730512249444,
      "loss": 0.8671593070030212,
      "loss_ce": 0.00021591814584098756,
      "loss_iou": 0.396484375,
      "loss_num": 0.01483154296875,
      "loss_xval": 0.8671875,
      "num_input_tokens_seen": 149133992,
      "step": 2662
    },
    {
      "epoch": 5.9309576837416484,
      "grad_norm": 23.221233367919922,
      "learning_rate": 1e-06,
      "loss": 0.6566,
      "num_input_tokens_seen": 149190436,
      "step": 2663
    },
    {
      "epoch": 5.9309576837416484,
      "loss": 0.43981656432151794,
      "loss_ce": 0.0002413657057331875,
      "loss_iou": 0.1943359375,
      "loss_num": 0.010009765625,
      "loss_xval": 0.439453125,
      "num_input_tokens_seen": 149190436,
      "step": 2663
    },
    {
      "epoch": 5.933184855233853,
      "grad_norm": 19.708267211914062,
      "learning_rate": 1e-06,
      "loss": 0.7044,
      "num_input_tokens_seen": 149247672,
      "step": 2664
    },
    {
      "epoch": 5.933184855233853,
      "loss": 0.8448212742805481,
      "loss_ce": 0.00021680007921531796,
      "loss_iou": 0.375,
      "loss_num": 0.018798828125,
      "loss_xval": 0.84375,
      "num_input_tokens_seen": 149247672,
      "step": 2664
    },
    {
      "epoch": 5.935412026726058,
      "grad_norm": 15.226134300231934,
      "learning_rate": 1e-06,
      "loss": 0.5869,
      "num_input_tokens_seen": 149305572,
      "step": 2665
    },
    {
      "epoch": 5.935412026726058,
      "loss": 0.5407141447067261,
      "loss_ce": 0.00018680887296795845,
      "loss_iou": 0.2333984375,
      "loss_num": 0.0146484375,
      "loss_xval": 0.5390625,
      "num_input_tokens_seen": 149305572,
      "step": 2665
    },
    {
      "epoch": 5.937639198218263,
      "grad_norm": 18.039369583129883,
      "learning_rate": 1e-06,
      "loss": 0.6593,
      "num_input_tokens_seen": 149363100,
      "step": 2666
    },
    {
      "epoch": 5.937639198218263,
      "loss": 0.4833727180957794,
      "loss_ce": 0.00021840460249222815,
      "loss_iou": 0.21875,
      "loss_num": 0.0091552734375,
      "loss_xval": 0.482421875,
      "num_input_tokens_seen": 149363100,
      "step": 2666
    },
    {
      "epoch": 5.939866369710468,
      "grad_norm": 14.021678924560547,
      "learning_rate": 1e-06,
      "loss": 0.9587,
      "num_input_tokens_seen": 149418140,
      "step": 2667
    },
    {
      "epoch": 5.939866369710468,
      "loss": 0.6920410990715027,
      "loss_ce": 0.0002686180523596704,
      "loss_iou": 0.306640625,
      "loss_num": 0.0159912109375,
      "loss_xval": 0.69140625,
      "num_input_tokens_seen": 149418140,
      "step": 2667
    },
    {
      "epoch": 5.942093541202673,
      "grad_norm": 17.604162216186523,
      "learning_rate": 1e-06,
      "loss": 0.5922,
      "num_input_tokens_seen": 149474312,
      "step": 2668
    },
    {
      "epoch": 5.942093541202673,
      "loss": 0.5115793347358704,
      "loss_ce": 0.0002268127864226699,
      "loss_iou": 0.2353515625,
      "loss_num": 0.00811767578125,
      "loss_xval": 0.51171875,
      "num_input_tokens_seen": 149474312,
      "step": 2668
    },
    {
      "epoch": 5.944320712694878,
      "grad_norm": 23.896390914916992,
      "learning_rate": 1e-06,
      "loss": 0.7063,
      "num_input_tokens_seen": 149530356,
      "step": 2669
    },
    {
      "epoch": 5.944320712694878,
      "loss": 0.5715218186378479,
      "loss_ce": 0.00023274502018466592,
      "loss_iou": 0.2490234375,
      "loss_num": 0.0145263671875,
      "loss_xval": 0.5703125,
      "num_input_tokens_seen": 149530356,
      "step": 2669
    },
    {
      "epoch": 5.9465478841870825,
      "grad_norm": 20.610628128051758,
      "learning_rate": 1e-06,
      "loss": 0.721,
      "num_input_tokens_seen": 149587276,
      "step": 2670
    },
    {
      "epoch": 5.9465478841870825,
      "loss": 0.7370386123657227,
      "loss_ce": 0.00022224214626476169,
      "loss_iou": 0.328125,
      "loss_num": 0.0162353515625,
      "loss_xval": 0.73828125,
      "num_input_tokens_seen": 149587276,
      "step": 2670
    },
    {
      "epoch": 5.948775055679287,
      "grad_norm": 18.218326568603516,
      "learning_rate": 1e-06,
      "loss": 0.8538,
      "num_input_tokens_seen": 149643160,
      "step": 2671
    },
    {
      "epoch": 5.948775055679287,
      "loss": 1.0529261827468872,
      "loss_ce": 0.00043590826680883765,
      "loss_iou": 0.40625,
      "loss_num": 0.0478515625,
      "loss_xval": 1.0546875,
      "num_input_tokens_seen": 149643160,
      "step": 2671
    },
    {
      "epoch": 5.951002227171492,
      "grad_norm": 20.644004821777344,
      "learning_rate": 1e-06,
      "loss": 0.7755,
      "num_input_tokens_seen": 149697316,
      "step": 2672
    },
    {
      "epoch": 5.951002227171492,
      "loss": 0.9467422962188721,
      "loss_ce": 0.0003311632899567485,
      "loss_iou": 0.357421875,
      "loss_num": 0.04638671875,
      "loss_xval": 0.9453125,
      "num_input_tokens_seen": 149697316,
      "step": 2672
    },
    {
      "epoch": 5.953229398663697,
      "grad_norm": 15.27214527130127,
      "learning_rate": 1e-06,
      "loss": 0.5603,
      "num_input_tokens_seen": 149754696,
      "step": 2673
    },
    {
      "epoch": 5.953229398663697,
      "loss": 0.5694996118545532,
      "loss_ce": 0.00040784955490380526,
      "loss_iou": 0.23828125,
      "loss_num": 0.0186767578125,
      "loss_xval": 0.5703125,
      "num_input_tokens_seen": 149754696,
      "step": 2673
    },
    {
      "epoch": 5.955456570155902,
      "grad_norm": 18.75067138671875,
      "learning_rate": 1e-06,
      "loss": 0.7863,
      "num_input_tokens_seen": 149809432,
      "step": 2674
    },
    {
      "epoch": 5.955456570155902,
      "loss": 0.8103808760643005,
      "loss_ce": 0.00032227032352238894,
      "loss_iou": 0.365234375,
      "loss_num": 0.01556396484375,
      "loss_xval": 0.80859375,
      "num_input_tokens_seen": 149809432,
      "step": 2674
    },
    {
      "epoch": 5.957683741648107,
      "grad_norm": 25.840362548828125,
      "learning_rate": 1e-06,
      "loss": 0.8198,
      "num_input_tokens_seen": 149865132,
      "step": 2675
    },
    {
      "epoch": 5.957683741648107,
      "loss": 0.85858553647995,
      "loss_ce": 0.00018709682626649737,
      "loss_iou": 0.384765625,
      "loss_num": 0.0177001953125,
      "loss_xval": 0.859375,
      "num_input_tokens_seen": 149865132,
      "step": 2675
    },
    {
      "epoch": 5.959910913140312,
      "grad_norm": 15.168061256408691,
      "learning_rate": 1e-06,
      "loss": 0.5927,
      "num_input_tokens_seen": 149921252,
      "step": 2676
    },
    {
      "epoch": 5.959910913140312,
      "loss": 0.5525012016296387,
      "loss_ce": 0.00025514926528558135,
      "loss_iou": 0.2412109375,
      "loss_num": 0.01385498046875,
      "loss_xval": 0.55078125,
      "num_input_tokens_seen": 149921252,
      "step": 2676
    },
    {
      "epoch": 5.9621380846325165,
      "grad_norm": 359.7155456542969,
      "learning_rate": 1e-06,
      "loss": 0.8264,
      "num_input_tokens_seen": 149975912,
      "step": 2677
    },
    {
      "epoch": 5.9621380846325165,
      "loss": 0.7250722646713257,
      "loss_ce": 0.00021878087136428803,
      "loss_iou": 0.30859375,
      "loss_num": 0.0218505859375,
      "loss_xval": 0.7265625,
      "num_input_tokens_seen": 149975912,
      "step": 2677
    },
    {
      "epoch": 5.964365256124721,
      "grad_norm": 20.11498260498047,
      "learning_rate": 1e-06,
      "loss": 0.6048,
      "num_input_tokens_seen": 150032272,
      "step": 2678
    },
    {
      "epoch": 5.964365256124721,
      "loss": 0.6044542789459229,
      "loss_ce": 0.00020618733833543956,
      "loss_iou": 0.255859375,
      "loss_num": 0.018310546875,
      "loss_xval": 0.60546875,
      "num_input_tokens_seen": 150032272,
      "step": 2678
    },
    {
      "epoch": 5.966592427616926,
      "grad_norm": 40.01982116699219,
      "learning_rate": 1e-06,
      "loss": 0.6705,
      "num_input_tokens_seen": 150089664,
      "step": 2679
    },
    {
      "epoch": 5.966592427616926,
      "loss": 0.48896247148513794,
      "loss_ce": 0.00019291002536192536,
      "loss_iou": 0.1884765625,
      "loss_num": 0.0223388671875,
      "loss_xval": 0.48828125,
      "num_input_tokens_seen": 150089664,
      "step": 2679
    },
    {
      "epoch": 5.968819599109131,
      "grad_norm": 13.833683013916016,
      "learning_rate": 1e-06,
      "loss": 0.5939,
      "num_input_tokens_seen": 150145520,
      "step": 2680
    },
    {
      "epoch": 5.968819599109131,
      "loss": 0.5350244641304016,
      "loss_ce": 0.000356461969204247,
      "loss_iou": 0.236328125,
      "loss_num": 0.0125732421875,
      "loss_xval": 0.53515625,
      "num_input_tokens_seen": 150145520,
      "step": 2680
    },
    {
      "epoch": 5.971046770601336,
      "grad_norm": 28.182466506958008,
      "learning_rate": 1e-06,
      "loss": 0.9304,
      "num_input_tokens_seen": 150198876,
      "step": 2681
    },
    {
      "epoch": 5.971046770601336,
      "loss": 0.5284985899925232,
      "loss_ce": 0.00017829591524787247,
      "loss_iou": 0.220703125,
      "loss_num": 0.017333984375,
      "loss_xval": 0.52734375,
      "num_input_tokens_seen": 150198876,
      "step": 2681
    },
    {
      "epoch": 5.973273942093542,
      "grad_norm": 15.853994369506836,
      "learning_rate": 1e-06,
      "loss": 0.6288,
      "num_input_tokens_seen": 150255024,
      "step": 2682
    },
    {
      "epoch": 5.973273942093542,
      "loss": 0.5141396522521973,
      "loss_ce": 0.0004677815013565123,
      "loss_iou": 0.1982421875,
      "loss_num": 0.023681640625,
      "loss_xval": 0.515625,
      "num_input_tokens_seen": 150255024,
      "step": 2682
    },
    {
      "epoch": 5.9755011135857465,
      "grad_norm": 19.383615493774414,
      "learning_rate": 1e-06,
      "loss": 0.9422,
      "num_input_tokens_seen": 150309836,
      "step": 2683
    },
    {
      "epoch": 5.9755011135857465,
      "loss": 0.8364056944847107,
      "loss_ce": 0.00022408382210414857,
      "loss_iou": 0.33203125,
      "loss_num": 0.03466796875,
      "loss_xval": 0.8359375,
      "num_input_tokens_seen": 150309836,
      "step": 2683
    },
    {
      "epoch": 5.977728285077951,
      "grad_norm": 23.231369018554688,
      "learning_rate": 1e-06,
      "loss": 0.7372,
      "num_input_tokens_seen": 150365848,
      "step": 2684
    },
    {
      "epoch": 5.977728285077951,
      "loss": 0.8447408676147461,
      "loss_ce": 0.0002584208268672228,
      "loss_iou": 0.34765625,
      "loss_num": 0.02978515625,
      "loss_xval": 0.84375,
      "num_input_tokens_seen": 150365848,
      "step": 2684
    },
    {
      "epoch": 5.979955456570156,
      "grad_norm": 22.7137393951416,
      "learning_rate": 1e-06,
      "loss": 0.6089,
      "num_input_tokens_seen": 150423628,
      "step": 2685
    },
    {
      "epoch": 5.979955456570156,
      "loss": 0.5462290048599243,
      "loss_ce": 0.00020853537716902792,
      "loss_iou": 0.23828125,
      "loss_num": 0.0137939453125,
      "loss_xval": 0.546875,
      "num_input_tokens_seen": 150423628,
      "step": 2685
    },
    {
      "epoch": 5.982182628062361,
      "grad_norm": 24.549579620361328,
      "learning_rate": 1e-06,
      "loss": 0.7922,
      "num_input_tokens_seen": 150478840,
      "step": 2686
    },
    {
      "epoch": 5.982182628062361,
      "loss": 0.8291885256767273,
      "loss_ce": 0.00020902017422486097,
      "loss_iou": 0.373046875,
      "loss_num": 0.0166015625,
      "loss_xval": 0.828125,
      "num_input_tokens_seen": 150478840,
      "step": 2686
    },
    {
      "epoch": 5.984409799554566,
      "grad_norm": 26.245058059692383,
      "learning_rate": 1e-06,
      "loss": 0.6099,
      "num_input_tokens_seen": 150534692,
      "step": 2687
    },
    {
      "epoch": 5.984409799554566,
      "loss": 0.49533137679100037,
      "loss_ce": 0.00021419850236270577,
      "loss_iou": 0.224609375,
      "loss_num": 0.00909423828125,
      "loss_xval": 0.49609375,
      "num_input_tokens_seen": 150534692,
      "step": 2687
    },
    {
      "epoch": 5.986636971046771,
      "grad_norm": 24.997560501098633,
      "learning_rate": 1e-06,
      "loss": 0.5429,
      "num_input_tokens_seen": 150591508,
      "step": 2688
    },
    {
      "epoch": 5.986636971046771,
      "loss": 0.5927173495292664,
      "loss_ce": 0.00031011985265649855,
      "loss_iou": 0.267578125,
      "loss_num": 0.01177978515625,
      "loss_xval": 0.59375,
      "num_input_tokens_seen": 150591508,
      "step": 2688
    },
    {
      "epoch": 5.988864142538976,
      "grad_norm": 16.52888298034668,
      "learning_rate": 1e-06,
      "loss": 0.7054,
      "num_input_tokens_seen": 150648216,
      "step": 2689
    },
    {
      "epoch": 5.988864142538976,
      "loss": 0.44299080967903137,
      "loss_ce": 0.00022651677136309445,
      "loss_iou": 0.171875,
      "loss_num": 0.0198974609375,
      "loss_xval": 0.443359375,
      "num_input_tokens_seen": 150648216,
      "step": 2689
    },
    {
      "epoch": 5.991091314031181,
      "grad_norm": 19.582372665405273,
      "learning_rate": 1e-06,
      "loss": 0.901,
      "num_input_tokens_seen": 150705220,
      "step": 2690
    },
    {
      "epoch": 5.991091314031181,
      "loss": 0.9914488196372986,
      "loss_ce": 0.00023790652630850673,
      "loss_iou": 0.43359375,
      "loss_num": 0.0244140625,
      "loss_xval": 0.9921875,
      "num_input_tokens_seen": 150705220,
      "step": 2690
    },
    {
      "epoch": 5.993318485523385,
      "grad_norm": 19.67958641052246,
      "learning_rate": 1e-06,
      "loss": 0.7656,
      "num_input_tokens_seen": 150759608,
      "step": 2691
    },
    {
      "epoch": 5.993318485523385,
      "loss": 0.586168110370636,
      "loss_ce": 0.00023060785315465182,
      "loss_iou": 0.2431640625,
      "loss_num": 0.0198974609375,
      "loss_xval": 0.5859375,
      "num_input_tokens_seen": 150759608,
      "step": 2691
    },
    {
      "epoch": 5.99554565701559,
      "grad_norm": 15.20195198059082,
      "learning_rate": 1e-06,
      "loss": 0.6715,
      "num_input_tokens_seen": 150816908,
      "step": 2692
    },
    {
      "epoch": 5.99554565701559,
      "loss": 0.6716079115867615,
      "loss_ce": 0.00022117490880191326,
      "loss_iou": 0.28515625,
      "loss_num": 0.020263671875,
      "loss_xval": 0.671875,
      "num_input_tokens_seen": 150816908,
      "step": 2692
    },
    {
      "epoch": 5.997772828507795,
      "grad_norm": 25.80984878540039,
      "learning_rate": 1e-06,
      "loss": 0.8853,
      "num_input_tokens_seen": 150872492,
      "step": 2693
    },
    {
      "epoch": 5.997772828507795,
      "loss": 1.047957181930542,
      "loss_ce": 0.00034974192385561764,
      "loss_iou": 0.43359375,
      "loss_num": 0.0361328125,
      "loss_xval": 1.046875,
      "num_input_tokens_seen": 150872492,
      "step": 2693
    },
    {
      "epoch": 6.0,
      "grad_norm": 18.474491119384766,
      "learning_rate": 1e-06,
      "loss": 0.8475,
      "num_input_tokens_seen": 150929208,
      "step": 2694
    },
    {
      "epoch": 6.0,
      "loss": 1.0649967193603516,
      "loss_ce": 0.0002994451788254082,
      "loss_iou": 0.4453125,
      "loss_num": 0.034912109375,
      "loss_xval": 1.0625,
      "num_input_tokens_seen": 150929208,
      "step": 2694
    },
    {
      "epoch": 6.002227171492205,
      "grad_norm": 25.486400604248047,
      "learning_rate": 1e-06,
      "loss": 0.7665,
      "num_input_tokens_seen": 150983692,
      "step": 2695
    },
    {
      "epoch": 6.002227171492205,
      "loss": 0.7017991542816162,
      "loss_ce": 0.0002610695082694292,
      "loss_iou": 0.283203125,
      "loss_num": 0.0269775390625,
      "loss_xval": 0.703125,
      "num_input_tokens_seen": 150983692,
      "step": 2695
    },
    {
      "epoch": 6.00445434298441,
      "grad_norm": 15.804261207580566,
      "learning_rate": 1e-06,
      "loss": 0.621,
      "num_input_tokens_seen": 151040928,
      "step": 2696
    },
    {
      "epoch": 6.00445434298441,
      "loss": 0.6565641164779663,
      "loss_ce": 0.000192045554285869,
      "loss_iou": 0.28125,
      "loss_num": 0.0185546875,
      "loss_xval": 0.65625,
      "num_input_tokens_seen": 151040928,
      "step": 2696
    },
    {
      "epoch": 6.006681514476615,
      "grad_norm": 18.53908348083496,
      "learning_rate": 1e-06,
      "loss": 0.7812,
      "num_input_tokens_seen": 151095388,
      "step": 2697
    },
    {
      "epoch": 6.006681514476615,
      "loss": 0.8815810084342957,
      "loss_ce": 0.00023334722209256142,
      "loss_iou": 0.369140625,
      "loss_num": 0.02880859375,
      "loss_xval": 0.8828125,
      "num_input_tokens_seen": 151095388,
      "step": 2697
    },
    {
      "epoch": 6.008908685968819,
      "grad_norm": 23.245758056640625,
      "learning_rate": 1e-06,
      "loss": 0.7773,
      "num_input_tokens_seen": 151148968,
      "step": 2698
    },
    {
      "epoch": 6.008908685968819,
      "loss": 0.8959696292877197,
      "loss_ce": 0.00021769374143332243,
      "loss_iou": 0.35546875,
      "loss_num": 0.03759765625,
      "loss_xval": 0.89453125,
      "num_input_tokens_seen": 151148968,
      "step": 2698
    },
    {
      "epoch": 6.011135857461024,
      "grad_norm": 13.859575271606445,
      "learning_rate": 1e-06,
      "loss": 0.54,
      "num_input_tokens_seen": 151205420,
      "step": 2699
    },
    {
      "epoch": 6.011135857461024,
      "loss": 0.5023019313812256,
      "loss_ce": 0.00022672602790407836,
      "loss_iou": 0.21484375,
      "loss_num": 0.01434326171875,
      "loss_xval": 0.50390625,
      "num_input_tokens_seen": 151205420,
      "step": 2699
    },
    {
      "epoch": 6.013363028953229,
      "grad_norm": 17.814104080200195,
      "learning_rate": 1e-06,
      "loss": 0.5899,
      "num_input_tokens_seen": 151262892,
      "step": 2700
    },
    {
      "epoch": 6.013363028953229,
      "loss": 0.6286102533340454,
      "loss_ce": 0.0001923007657751441,
      "loss_iou": 0.255859375,
      "loss_num": 0.02294921875,
      "loss_xval": 0.62890625,
      "num_input_tokens_seen": 151262892,
      "step": 2700
    },
    {
      "epoch": 6.015590200445434,
      "grad_norm": 20.40790557861328,
      "learning_rate": 1e-06,
      "loss": 0.7099,
      "num_input_tokens_seen": 151316164,
      "step": 2701
    },
    {
      "epoch": 6.015590200445434,
      "loss": 0.8200287818908691,
      "loss_ce": 0.00020452812896110117,
      "loss_iou": 0.34765625,
      "loss_num": 0.025146484375,
      "loss_xval": 0.8203125,
      "num_input_tokens_seen": 151316164,
      "step": 2701
    },
    {
      "epoch": 6.017817371937639,
      "grad_norm": 19.481016159057617,
      "learning_rate": 1e-06,
      "loss": 0.6958,
      "num_input_tokens_seen": 151371160,
      "step": 2702
    },
    {
      "epoch": 6.017817371937639,
      "loss": 0.6471884250640869,
      "loss_ce": 0.00021577253937721252,
      "loss_iou": 0.25,
      "loss_num": 0.0296630859375,
      "loss_xval": 0.6484375,
      "num_input_tokens_seen": 151371160,
      "step": 2702
    },
    {
      "epoch": 6.020044543429844,
      "grad_norm": 32.87615203857422,
      "learning_rate": 1e-06,
      "loss": 0.7168,
      "num_input_tokens_seen": 151428440,
      "step": 2703
    },
    {
      "epoch": 6.020044543429844,
      "loss": 0.737989604473114,
      "loss_ce": 0.00019663787679746747,
      "loss_iou": 0.330078125,
      "loss_num": 0.015625,
      "loss_xval": 0.73828125,
      "num_input_tokens_seen": 151428440,
      "step": 2703
    },
    {
      "epoch": 6.022271714922049,
      "grad_norm": 11.81971263885498,
      "learning_rate": 1e-06,
      "loss": 0.6217,
      "num_input_tokens_seen": 151483012,
      "step": 2704
    },
    {
      "epoch": 6.022271714922049,
      "loss": 0.6774269342422485,
      "loss_ce": 0.00018085945339407772,
      "loss_iou": 0.291015625,
      "loss_num": 0.0189208984375,
      "loss_xval": 0.67578125,
      "num_input_tokens_seen": 151483012,
      "step": 2704
    },
    {
      "epoch": 6.0244988864142535,
      "grad_norm": 30.31116485595703,
      "learning_rate": 1e-06,
      "loss": 0.6544,
      "num_input_tokens_seen": 151538432,
      "step": 2705
    },
    {
      "epoch": 6.0244988864142535,
      "loss": 0.8250452280044556,
      "loss_ce": 0.00033821084070950747,
      "loss_iou": 0.341796875,
      "loss_num": 0.0281982421875,
      "loss_xval": 0.82421875,
      "num_input_tokens_seen": 151538432,
      "step": 2705
    },
    {
      "epoch": 6.026726057906459,
      "grad_norm": 18.400726318359375,
      "learning_rate": 1e-06,
      "loss": 0.5954,
      "num_input_tokens_seen": 151594204,
      "step": 2706
    },
    {
      "epoch": 6.026726057906459,
      "loss": 0.5231574177742004,
      "loss_ce": 0.00020822283113375306,
      "loss_iou": 0.21875,
      "loss_num": 0.016845703125,
      "loss_xval": 0.5234375,
      "num_input_tokens_seen": 151594204,
      "step": 2706
    },
    {
      "epoch": 6.028953229398664,
      "grad_norm": 15.375958442687988,
      "learning_rate": 1e-06,
      "loss": 0.7347,
      "num_input_tokens_seen": 151651444,
      "step": 2707
    },
    {
      "epoch": 6.028953229398664,
      "loss": 0.5390880107879639,
      "loss_ce": 0.00026969151804223657,
      "loss_iou": 0.2333984375,
      "loss_num": 0.0146484375,
      "loss_xval": 0.5390625,
      "num_input_tokens_seen": 151651444,
      "step": 2707
    },
    {
      "epoch": 6.031180400890869,
      "grad_norm": 20.725507736206055,
      "learning_rate": 1e-06,
      "loss": 0.546,
      "num_input_tokens_seen": 151708908,
      "step": 2708
    },
    {
      "epoch": 6.031180400890869,
      "loss": 0.49738240242004395,
      "loss_ce": 0.0009224280365742743,
      "loss_iou": 0.203125,
      "loss_num": 0.01806640625,
      "loss_xval": 0.49609375,
      "num_input_tokens_seen": 151708908,
      "step": 2708
    },
    {
      "epoch": 6.033407572383074,
      "grad_norm": 21.394176483154297,
      "learning_rate": 1e-06,
      "loss": 0.4111,
      "num_input_tokens_seen": 151765488,
      "step": 2709
    },
    {
      "epoch": 6.033407572383074,
      "loss": 0.4496401846408844,
      "loss_ce": 0.0001773049880284816,
      "loss_iou": 0.1826171875,
      "loss_num": 0.0167236328125,
      "loss_xval": 0.44921875,
      "num_input_tokens_seen": 151765488,
      "step": 2709
    },
    {
      "epoch": 6.035634743875279,
      "grad_norm": 26.451292037963867,
      "learning_rate": 1e-06,
      "loss": 0.5001,
      "num_input_tokens_seen": 151820864,
      "step": 2710
    },
    {
      "epoch": 6.035634743875279,
      "loss": 0.41302353143692017,
      "loss_ce": 0.0001817169541027397,
      "loss_iou": 0.177734375,
      "loss_num": 0.0115966796875,
      "loss_xval": 0.412109375,
      "num_input_tokens_seen": 151820864,
      "step": 2710
    },
    {
      "epoch": 6.0378619153674835,
      "grad_norm": 18.717567443847656,
      "learning_rate": 1e-06,
      "loss": 0.6072,
      "num_input_tokens_seen": 151879976,
      "step": 2711
    },
    {
      "epoch": 6.0378619153674835,
      "loss": 0.7275397181510925,
      "loss_ce": 0.00024480142747052014,
      "loss_iou": 0.31640625,
      "loss_num": 0.0191650390625,
      "loss_xval": 0.7265625,
      "num_input_tokens_seen": 151879976,
      "step": 2711
    },
    {
      "epoch": 6.040089086859688,
      "grad_norm": 15.812241554260254,
      "learning_rate": 1e-06,
      "loss": 0.6669,
      "num_input_tokens_seen": 151938396,
      "step": 2712
    },
    {
      "epoch": 6.040089086859688,
      "loss": 0.5586850047111511,
      "loss_ce": 0.00021330438903532922,
      "loss_iou": 0.2275390625,
      "loss_num": 0.020751953125,
      "loss_xval": 0.55859375,
      "num_input_tokens_seen": 151938396,
      "step": 2712
    },
    {
      "epoch": 6.042316258351893,
      "grad_norm": 28.703325271606445,
      "learning_rate": 1e-06,
      "loss": 1.0115,
      "num_input_tokens_seen": 151992472,
      "step": 2713
    },
    {
      "epoch": 6.042316258351893,
      "loss": 0.8724164366722107,
      "loss_ce": 0.0003461412852630019,
      "loss_iou": 0.384765625,
      "loss_num": 0.020263671875,
      "loss_xval": 0.87109375,
      "num_input_tokens_seen": 151992472,
      "step": 2713
    },
    {
      "epoch": 6.044543429844098,
      "grad_norm": 21.311323165893555,
      "learning_rate": 1e-06,
      "loss": 0.9992,
      "num_input_tokens_seen": 152046528,
      "step": 2714
    },
    {
      "epoch": 6.044543429844098,
      "loss": 0.9513915777206421,
      "loss_ce": 0.00021972534887026995,
      "loss_iou": 0.408203125,
      "loss_num": 0.0269775390625,
      "loss_xval": 0.953125,
      "num_input_tokens_seen": 152046528,
      "step": 2714
    },
    {
      "epoch": 6.046770601336303,
      "grad_norm": 16.139442443847656,
      "learning_rate": 1e-06,
      "loss": 0.5864,
      "num_input_tokens_seen": 152103856,
      "step": 2715
    },
    {
      "epoch": 6.046770601336303,
      "loss": 0.7112746238708496,
      "loss_ce": 0.00021506489429157227,
      "loss_iou": 0.306640625,
      "loss_num": 0.01953125,
      "loss_xval": 0.7109375,
      "num_input_tokens_seen": 152103856,
      "step": 2715
    },
    {
      "epoch": 6.048997772828508,
      "grad_norm": 18.766185760498047,
      "learning_rate": 1e-06,
      "loss": 0.6731,
      "num_input_tokens_seen": 152161252,
      "step": 2716
    },
    {
      "epoch": 6.048997772828508,
      "loss": 0.6426092386245728,
      "loss_ce": 0.0002752981963567436,
      "loss_iou": 0.283203125,
      "loss_num": 0.01483154296875,
      "loss_xval": 0.640625,
      "num_input_tokens_seen": 152161252,
      "step": 2716
    },
    {
      "epoch": 6.051224944320713,
      "grad_norm": 20.689594268798828,
      "learning_rate": 1e-06,
      "loss": 0.6403,
      "num_input_tokens_seen": 152219268,
      "step": 2717
    },
    {
      "epoch": 6.051224944320713,
      "loss": 0.6734442114830017,
      "loss_ce": 0.00022642976546194404,
      "loss_iou": 0.291015625,
      "loss_num": 0.0181884765625,
      "loss_xval": 0.671875,
      "num_input_tokens_seen": 152219268,
      "step": 2717
    },
    {
      "epoch": 6.0534521158129175,
      "grad_norm": 20.202369689941406,
      "learning_rate": 1e-06,
      "loss": 0.9098,
      "num_input_tokens_seen": 152275040,
      "step": 2718
    },
    {
      "epoch": 6.0534521158129175,
      "loss": 1.088090181350708,
      "loss_ce": 0.00019952133879996836,
      "loss_iou": 0.462890625,
      "loss_num": 0.032470703125,
      "loss_xval": 1.0859375,
      "num_input_tokens_seen": 152275040,
      "step": 2718
    },
    {
      "epoch": 6.055679287305122,
      "grad_norm": 24.603801727294922,
      "learning_rate": 1e-06,
      "loss": 0.5698,
      "num_input_tokens_seen": 152329728,
      "step": 2719
    },
    {
      "epoch": 6.055679287305122,
      "loss": 0.7060960531234741,
      "loss_ce": 0.0002855417551472783,
      "loss_iou": 0.3125,
      "loss_num": 0.0164794921875,
      "loss_xval": 0.70703125,
      "num_input_tokens_seen": 152329728,
      "step": 2719
    },
    {
      "epoch": 6.057906458797327,
      "grad_norm": 17.9193172454834,
      "learning_rate": 1e-06,
      "loss": 0.528,
      "num_input_tokens_seen": 152388868,
      "step": 2720
    },
    {
      "epoch": 6.057906458797327,
      "loss": 0.4527170658111572,
      "loss_ce": 0.00020240643061697483,
      "loss_iou": 0.17578125,
      "loss_num": 0.02001953125,
      "loss_xval": 0.453125,
      "num_input_tokens_seen": 152388868,
      "step": 2720
    },
    {
      "epoch": 6.060133630289532,
      "grad_norm": 16.661762237548828,
      "learning_rate": 1e-06,
      "loss": 0.6847,
      "num_input_tokens_seen": 152442380,
      "step": 2721
    },
    {
      "epoch": 6.060133630289532,
      "loss": 0.6767435073852539,
      "loss_ce": 0.0007180861430242658,
      "loss_iou": 0.263671875,
      "loss_num": 0.030029296875,
      "loss_xval": 0.67578125,
      "num_input_tokens_seen": 152442380,
      "step": 2721
    },
    {
      "epoch": 6.062360801781737,
      "grad_norm": 16.014999389648438,
      "learning_rate": 1e-06,
      "loss": 0.5119,
      "num_input_tokens_seen": 152499432,
      "step": 2722
    },
    {
      "epoch": 6.062360801781737,
      "loss": 0.574461817741394,
      "loss_ce": 0.000243054106249474,
      "loss_iou": 0.2431640625,
      "loss_num": 0.017822265625,
      "loss_xval": 0.57421875,
      "num_input_tokens_seen": 152499432,
      "step": 2722
    },
    {
      "epoch": 6.064587973273942,
      "grad_norm": 39.888118743896484,
      "learning_rate": 1e-06,
      "loss": 0.8712,
      "num_input_tokens_seen": 152551644,
      "step": 2723
    },
    {
      "epoch": 6.064587973273942,
      "loss": 0.9099514484405518,
      "loss_ce": 0.0002834274200722575,
      "loss_iou": 0.38671875,
      "loss_num": 0.0277099609375,
      "loss_xval": 0.91015625,
      "num_input_tokens_seen": 152551644,
      "step": 2723
    },
    {
      "epoch": 6.066815144766147,
      "grad_norm": 16.595260620117188,
      "learning_rate": 1e-06,
      "loss": 0.6684,
      "num_input_tokens_seen": 152608960,
      "step": 2724
    },
    {
      "epoch": 6.066815144766147,
      "loss": 0.7639758586883545,
      "loss_ce": 0.0001819009194150567,
      "loss_iou": 0.283203125,
      "loss_num": 0.03955078125,
      "loss_xval": 0.765625,
      "num_input_tokens_seen": 152608960,
      "step": 2724
    },
    {
      "epoch": 6.0690423162583516,
      "grad_norm": 15.64448070526123,
      "learning_rate": 1e-06,
      "loss": 0.6383,
      "num_input_tokens_seen": 152661612,
      "step": 2725
    },
    {
      "epoch": 6.0690423162583516,
      "loss": 0.5945534706115723,
      "loss_ce": 0.0003151525743305683,
      "loss_iou": 0.234375,
      "loss_num": 0.02490234375,
      "loss_xval": 0.59375,
      "num_input_tokens_seen": 152661612,
      "step": 2725
    },
    {
      "epoch": 6.071269487750556,
      "grad_norm": 14.511089324951172,
      "learning_rate": 1e-06,
      "loss": 0.7527,
      "num_input_tokens_seen": 152717140,
      "step": 2726
    },
    {
      "epoch": 6.071269487750556,
      "loss": 0.7612234354019165,
      "loss_ce": 0.00023707791115157306,
      "loss_iou": 0.337890625,
      "loss_num": 0.0166015625,
      "loss_xval": 0.76171875,
      "num_input_tokens_seen": 152717140,
      "step": 2726
    },
    {
      "epoch": 6.073496659242761,
      "grad_norm": 23.56868553161621,
      "learning_rate": 1e-06,
      "loss": 0.69,
      "num_input_tokens_seen": 152772512,
      "step": 2727
    },
    {
      "epoch": 6.073496659242761,
      "loss": 0.8910683393478394,
      "loss_ce": 0.00019922200590372086,
      "loss_iou": 0.373046875,
      "loss_num": 0.029296875,
      "loss_xval": 0.890625,
      "num_input_tokens_seen": 152772512,
      "step": 2727
    },
    {
      "epoch": 6.075723830734967,
      "grad_norm": 15.447635650634766,
      "learning_rate": 1e-06,
      "loss": 0.5218,
      "num_input_tokens_seen": 152828880,
      "step": 2728
    },
    {
      "epoch": 6.075723830734967,
      "loss": 0.5673261284828186,
      "loss_ce": 0.00018746175919659436,
      "loss_iou": 0.21875,
      "loss_num": 0.0257568359375,
      "loss_xval": 0.56640625,
      "num_input_tokens_seen": 152828880,
      "step": 2728
    },
    {
      "epoch": 6.077951002227172,
      "grad_norm": 24.025293350219727,
      "learning_rate": 1e-06,
      "loss": 0.862,
      "num_input_tokens_seen": 152884388,
      "step": 2729
    },
    {
      "epoch": 6.077951002227172,
      "loss": 0.7360300421714783,
      "loss_ce": 0.00019018063903786242,
      "loss_iou": 0.32421875,
      "loss_num": 0.01708984375,
      "loss_xval": 0.734375,
      "num_input_tokens_seen": 152884388,
      "step": 2729
    },
    {
      "epoch": 6.080178173719377,
      "grad_norm": 16.803165435791016,
      "learning_rate": 1e-06,
      "loss": 0.8949,
      "num_input_tokens_seen": 152942756,
      "step": 2730
    },
    {
      "epoch": 6.080178173719377,
      "loss": 0.9147862195968628,
      "loss_ce": 0.00023543770657852292,
      "loss_iou": 0.349609375,
      "loss_num": 0.042724609375,
      "loss_xval": 0.9140625,
      "num_input_tokens_seen": 152942756,
      "step": 2730
    },
    {
      "epoch": 6.082405345211582,
      "grad_norm": 18.61233139038086,
      "learning_rate": 1e-06,
      "loss": 0.6134,
      "num_input_tokens_seen": 152998352,
      "step": 2731
    },
    {
      "epoch": 6.082405345211582,
      "loss": 0.46365272998809814,
      "loss_ce": 0.0004569324664771557,
      "loss_iou": 0.2041015625,
      "loss_num": 0.0108642578125,
      "loss_xval": 0.462890625,
      "num_input_tokens_seen": 152998352,
      "step": 2731
    },
    {
      "epoch": 6.0846325167037865,
      "grad_norm": 22.512348175048828,
      "learning_rate": 1e-06,
      "loss": 0.8523,
      "num_input_tokens_seen": 153052908,
      "step": 2732
    },
    {
      "epoch": 6.0846325167037865,
      "loss": 0.7741538882255554,
      "loss_ce": 0.00022813121904619038,
      "loss_iou": 0.310546875,
      "loss_num": 0.03076171875,
      "loss_xval": 0.7734375,
      "num_input_tokens_seen": 153052908,
      "step": 2732
    },
    {
      "epoch": 6.086859688195991,
      "grad_norm": 36.26032257080078,
      "learning_rate": 1e-06,
      "loss": 0.8099,
      "num_input_tokens_seen": 153107940,
      "step": 2733
    },
    {
      "epoch": 6.086859688195991,
      "loss": 0.8505562543869019,
      "loss_ce": 0.00021446403115987778,
      "loss_iou": 0.39453125,
      "loss_num": 0.0123291015625,
      "loss_xval": 0.8515625,
      "num_input_tokens_seen": 153107940,
      "step": 2733
    },
    {
      "epoch": 6.089086859688196,
      "grad_norm": 23.493366241455078,
      "learning_rate": 1e-06,
      "loss": 0.7301,
      "num_input_tokens_seen": 153166016,
      "step": 2734
    },
    {
      "epoch": 6.089086859688196,
      "loss": 0.6998932361602783,
      "loss_ce": 0.00018625493976287544,
      "loss_iou": 0.298828125,
      "loss_num": 0.020263671875,
      "loss_xval": 0.69921875,
      "num_input_tokens_seen": 153166016,
      "step": 2734
    },
    {
      "epoch": 6.091314031180401,
      "grad_norm": 14.597694396972656,
      "learning_rate": 1e-06,
      "loss": 0.6806,
      "num_input_tokens_seen": 153222732,
      "step": 2735
    },
    {
      "epoch": 6.091314031180401,
      "loss": 0.44916197657585144,
      "loss_ce": 0.00018737564096227288,
      "loss_iou": 0.1962890625,
      "loss_num": 0.01123046875,
      "loss_xval": 0.44921875,
      "num_input_tokens_seen": 153222732,
      "step": 2735
    },
    {
      "epoch": 6.093541202672606,
      "grad_norm": 22.396705627441406,
      "learning_rate": 1e-06,
      "loss": 0.7905,
      "num_input_tokens_seen": 153279676,
      "step": 2736
    },
    {
      "epoch": 6.093541202672606,
      "loss": 0.8422399163246155,
      "loss_ce": 0.00019887213420588523,
      "loss_iou": 0.361328125,
      "loss_num": 0.02392578125,
      "loss_xval": 0.84375,
      "num_input_tokens_seen": 153279676,
      "step": 2736
    },
    {
      "epoch": 6.095768374164811,
      "grad_norm": 125.70928192138672,
      "learning_rate": 1e-06,
      "loss": 0.6704,
      "num_input_tokens_seen": 153335272,
      "step": 2737
    },
    {
      "epoch": 6.095768374164811,
      "loss": 0.6769283413887024,
      "loss_ce": 0.0001705446484265849,
      "loss_iou": 0.298828125,
      "loss_num": 0.01611328125,
      "loss_xval": 0.67578125,
      "num_input_tokens_seen": 153335272,
      "step": 2737
    },
    {
      "epoch": 6.097995545657016,
      "grad_norm": 16.32554817199707,
      "learning_rate": 1e-06,
      "loss": 0.6954,
      "num_input_tokens_seen": 153390464,
      "step": 2738
    },
    {
      "epoch": 6.097995545657016,
      "loss": 0.9421560764312744,
      "loss_ce": 0.0002615359262563288,
      "loss_iou": 0.419921875,
      "loss_num": 0.0206298828125,
      "loss_xval": 0.94140625,
      "num_input_tokens_seen": 153390464,
      "step": 2738
    },
    {
      "epoch": 6.1002227171492205,
      "grad_norm": 28.368751525878906,
      "learning_rate": 1e-06,
      "loss": 0.806,
      "num_input_tokens_seen": 153448492,
      "step": 2739
    },
    {
      "epoch": 6.1002227171492205,
      "loss": 0.846657395362854,
      "loss_ce": 0.00022182743123266846,
      "loss_iou": 0.357421875,
      "loss_num": 0.0264892578125,
      "loss_xval": 0.84765625,
      "num_input_tokens_seen": 153448492,
      "step": 2739
    },
    {
      "epoch": 6.102449888641425,
      "grad_norm": 16.504480361938477,
      "learning_rate": 1e-06,
      "loss": 0.651,
      "num_input_tokens_seen": 153502140,
      "step": 2740
    },
    {
      "epoch": 6.102449888641425,
      "loss": 0.6750026345252991,
      "loss_ce": 0.0001979665830731392,
      "loss_iou": 0.298828125,
      "loss_num": 0.01531982421875,
      "loss_xval": 0.67578125,
      "num_input_tokens_seen": 153502140,
      "step": 2740
    },
    {
      "epoch": 6.10467706013363,
      "grad_norm": 17.08942413330078,
      "learning_rate": 1e-06,
      "loss": 0.5972,
      "num_input_tokens_seen": 153558592,
      "step": 2741
    },
    {
      "epoch": 6.10467706013363,
      "loss": 0.7375233769416809,
      "loss_ce": 0.00021869146439712495,
      "loss_iou": 0.29296875,
      "loss_num": 0.030029296875,
      "loss_xval": 0.73828125,
      "num_input_tokens_seen": 153558592,
      "step": 2741
    },
    {
      "epoch": 6.106904231625835,
      "grad_norm": 39.28758239746094,
      "learning_rate": 1e-06,
      "loss": 0.6328,
      "num_input_tokens_seen": 153615344,
      "step": 2742
    },
    {
      "epoch": 6.106904231625835,
      "loss": 0.6711956858634949,
      "loss_ce": 0.00029724877094849944,
      "loss_iou": 0.287109375,
      "loss_num": 0.0194091796875,
      "loss_xval": 0.671875,
      "num_input_tokens_seen": 153615344,
      "step": 2742
    },
    {
      "epoch": 6.10913140311804,
      "grad_norm": 25.463825225830078,
      "learning_rate": 1e-06,
      "loss": 0.6359,
      "num_input_tokens_seen": 153673836,
      "step": 2743
    },
    {
      "epoch": 6.10913140311804,
      "loss": 0.6788397431373596,
      "loss_ce": 0.0002203606127295643,
      "loss_iou": 0.3046875,
      "loss_num": 0.01385498046875,
      "loss_xval": 0.6796875,
      "num_input_tokens_seen": 153673836,
      "step": 2743
    },
    {
      "epoch": 6.111358574610245,
      "grad_norm": 20.883134841918945,
      "learning_rate": 1e-06,
      "loss": 0.7075,
      "num_input_tokens_seen": 153728812,
      "step": 2744
    },
    {
      "epoch": 6.111358574610245,
      "loss": 0.9753950834274292,
      "loss_ce": 0.00029741463367827237,
      "loss_iou": 0.43359375,
      "loss_num": 0.021728515625,
      "loss_xval": 0.9765625,
      "num_input_tokens_seen": 153728812,
      "step": 2744
    },
    {
      "epoch": 6.11358574610245,
      "grad_norm": 24.50876808166504,
      "learning_rate": 1e-06,
      "loss": 0.8218,
      "num_input_tokens_seen": 153784252,
      "step": 2745
    },
    {
      "epoch": 6.11358574610245,
      "loss": 0.9287269115447998,
      "loss_ce": 0.0009925166377797723,
      "loss_iou": 0.3671875,
      "loss_num": 0.038818359375,
      "loss_xval": 0.9296875,
      "num_input_tokens_seen": 153784252,
      "step": 2745
    },
    {
      "epoch": 6.1158129175946545,
      "grad_norm": 22.58624839782715,
      "learning_rate": 1e-06,
      "loss": 0.6866,
      "num_input_tokens_seen": 153839176,
      "step": 2746
    },
    {
      "epoch": 6.1158129175946545,
      "loss": 0.7910230755805969,
      "loss_ce": 0.0002516076201573014,
      "loss_iou": 0.33203125,
      "loss_num": 0.0247802734375,
      "loss_xval": 0.7890625,
      "num_input_tokens_seen": 153839176,
      "step": 2746
    },
    {
      "epoch": 6.118040089086859,
      "grad_norm": 12.76305103302002,
      "learning_rate": 1e-06,
      "loss": 0.6154,
      "num_input_tokens_seen": 153897360,
      "step": 2747
    },
    {
      "epoch": 6.118040089086859,
      "loss": 0.6281359791755676,
      "loss_ce": 0.0002062909334199503,
      "loss_iou": 0.291015625,
      "loss_num": 0.0089111328125,
      "loss_xval": 0.62890625,
      "num_input_tokens_seen": 153897360,
      "step": 2747
    },
    {
      "epoch": 6.120267260579064,
      "grad_norm": 17.98402214050293,
      "learning_rate": 1e-06,
      "loss": 0.6908,
      "num_input_tokens_seen": 153951844,
      "step": 2748
    },
    {
      "epoch": 6.120267260579064,
      "loss": 0.8605327606201172,
      "loss_ce": 0.0001811749825719744,
      "loss_iou": 0.34765625,
      "loss_num": 0.033203125,
      "loss_xval": 0.859375,
      "num_input_tokens_seen": 153951844,
      "step": 2748
    },
    {
      "epoch": 6.122494432071269,
      "grad_norm": 19.222530364990234,
      "learning_rate": 1e-06,
      "loss": 0.7594,
      "num_input_tokens_seen": 154006004,
      "step": 2749
    },
    {
      "epoch": 6.122494432071269,
      "loss": 0.7193889617919922,
      "loss_ce": 0.0003948350786231458,
      "loss_iou": 0.3046875,
      "loss_num": 0.02197265625,
      "loss_xval": 0.71875,
      "num_input_tokens_seen": 154006004,
      "step": 2749
    },
    {
      "epoch": 6.124721603563474,
      "grad_norm": 16.589632034301758,
      "learning_rate": 1e-06,
      "loss": 0.7099,
      "num_input_tokens_seen": 154061564,
      "step": 2750
    },
    {
      "epoch": 6.124721603563474,
      "eval_seeclick_web_CIoU": 0.5717557966709137,
      "eval_seeclick_web_GIoU": 0.5666035413742065,
      "eval_seeclick_web_IoU": 0.5889425873756409,
      "eval_seeclick_web_MAE_all": 0.016641407273709774,
      "eval_seeclick_web_MAE_h": 0.009463720256462693,
      "eval_seeclick_web_MAE_w": 0.0170047702267766,
      "eval_seeclick_web_MAE_x_boxes": 0.009009606204926968,
      "eval_seeclick_web_MAE_y_boxes": 0.022332632914185524,
      "eval_seeclick_web_inside_bbox": 0.9166666567325592,
      "eval_seeclick_web_loss": 0.9315496683120728,
      "eval_seeclick_web_loss_ce": 0.0002911543124355376,
      "eval_seeclick_web_loss_iou": 0.423828125,
      "eval_seeclick_web_loss_num": 0.0130767822265625,
      "eval_seeclick_web_loss_xval": 0.912841796875,
      "eval_seeclick_web_runtime": 26.4518,
      "eval_seeclick_web_samples_per_second": 1.89,
      "eval_seeclick_web_steps_per_second": 0.076,
      "num_input_tokens_seen": 154061564,
      "step": 2750
    },
    {
      "epoch": 6.124721603563474,
      "eval_icons_CIoU": 0.2957966476678848,
      "eval_icons_GIoU": 0.3207407593727112,
      "eval_icons_IoU": 0.3739467114210129,
      "eval_icons_MAE_all": 0.06642757169902325,
      "eval_icons_MAE_h": 0.03895352780818939,
      "eval_icons_MAE_w": 0.07084468938410282,
      "eval_icons_MAE_x_boxes": 0.06239369884133339,
      "eval_icons_MAE_y_boxes": 0.03859401401132345,
      "eval_icons_inside_bbox": 0.6336805522441864,
      "eval_icons_loss": 1.7069156169891357,
      "eval_icons_loss_ce": 0.00033613041159696877,
      "eval_icons_loss_iou": 0.6629638671875,
      "eval_icons_loss_num": 0.06306648254394531,
      "eval_icons_loss_xval": 1.64111328125,
      "eval_icons_runtime": 25.4698,
      "eval_icons_samples_per_second": 1.963,
      "eval_icons_steps_per_second": 0.079,
      "num_input_tokens_seen": 154061564,
      "step": 2750
    },
    {
      "epoch": 6.124721603563474,
      "eval_screenspot_CIoU": 0.3402452568213145,
      "eval_screenspot_GIoU": 0.3564106822013855,
      "eval_screenspot_IoU": 0.4236031075318654,
      "eval_screenspot_MAE_all": 0.06354892750581105,
      "eval_screenspot_MAE_h": 0.03823430463671684,
      "eval_screenspot_MAE_w": 0.07551725829641025,
      "eval_screenspot_MAE_x_boxes": 0.07483576859037082,
      "eval_screenspot_MAE_y_boxes": 0.046273874429365,
      "eval_screenspot_inside_bbox": 0.659583330154419,
      "eval_screenspot_loss": 1.6718511581420898,
      "eval_screenspot_loss_ce": 0.00034005365644892055,
      "eval_screenspot_loss_iou": 0.6853841145833334,
      "eval_screenspot_loss_num": 0.07482655843098958,
      "eval_screenspot_loss_xval": 1.7449544270833333,
      "eval_screenspot_runtime": 44.3938,
      "eval_screenspot_samples_per_second": 2.005,
      "eval_screenspot_steps_per_second": 0.068,
      "num_input_tokens_seen": 154061564,
      "step": 2750
    },
    {
      "epoch": 6.124721603563474,
      "eval_compot_CIoU": 0.3488970696926117,
      "eval_compot_GIoU": 0.363851934671402,
      "eval_compot_IoU": 0.40621405839920044,
      "eval_compot_MAE_all": 0.017889784649014473,
      "eval_compot_MAE_h": 0.008526601362973452,
      "eval_compot_MAE_w": 0.021996816620230675,
      "eval_compot_MAE_x_boxes": 0.02964367438107729,
      "eval_compot_MAE_y_boxes": 0.006707766558974981,
      "eval_compot_inside_bbox": 0.6458333432674408,
      "eval_compot_loss": 1.3785548210144043,
      "eval_compot_loss_ce": 0.0002755048044491559,
      "eval_compot_loss_iou": 0.6339111328125,
      "eval_compot_loss_num": 0.016744613647460938,
      "eval_compot_loss_xval": 1.35205078125,
      "eval_compot_runtime": 24.1215,
      "eval_compot_samples_per_second": 2.073,
      "eval_compot_steps_per_second": 0.083,
      "num_input_tokens_seen": 154061564,
      "step": 2750
    },
    {
      "epoch": 6.124721603563474,
      "eval_custom_ui_val_CIoU": 0.4712728477186627,
      "eval_custom_ui_val_GIoU": 0.48705925544102985,
      "eval_custom_ui_val_IoU": 0.5247346328364478,
      "eval_custom_ui_val_MAE_all": 0.03072897769096825,
      "eval_custom_ui_val_MAE_h": 0.017728664524232347,
      "eval_custom_ui_val_MAE_w": 0.03841559050811662,
      "eval_custom_ui_val_MAE_x_boxes": 0.03411081122855345,
      "eval_custom_ui_val_MAE_y_boxes": 0.015722092292788956,
      "eval_custom_ui_val_inside_bbox": 0.7353395091162788,
      "eval_custom_ui_val_loss": 1.1931304931640625,
      "eval_custom_ui_val_loss_ce": 0.0003192785694005175,
      "eval_custom_ui_val_loss_iou": 0.5072292751736112,
      "eval_custom_ui_val_loss_num": 0.02840868631998698,
      "eval_custom_ui_val_loss_xval": 1.1561957465277777,
      "eval_custom_ui_val_runtime": 72.27,
      "eval_custom_ui_val_samples_per_second": 3.667,
      "eval_custom_ui_val_steps_per_second": 0.125,
      "num_input_tokens_seen": 154061564,
      "step": 2750
    },
    {
      "epoch": 6.124721603563474,
      "loss": 0.908734917640686,
      "loss_ce": 0.0002876613289117813,
      "loss_iou": 0.39453125,
      "loss_num": 0.024169921875,
      "loss_xval": 0.91015625,
      "num_input_tokens_seen": 154061564,
      "step": 2750
    },
    {
      "epoch": 6.12694877505568,
      "grad_norm": 21.767990112304688,
      "learning_rate": 1e-06,
      "loss": 0.7365,
      "num_input_tokens_seen": 154117696,
      "step": 2751
    },
    {
      "epoch": 6.12694877505568,
      "loss": 0.7305013537406921,
      "loss_ce": 0.0002767470432445407,
      "loss_iou": 0.283203125,
      "loss_num": 0.032958984375,
      "loss_xval": 0.73046875,
      "num_input_tokens_seen": 154117696,
      "step": 2751
    },
    {
      "epoch": 6.129175946547885,
      "grad_norm": 26.993640899658203,
      "learning_rate": 1e-06,
      "loss": 0.8128,
      "num_input_tokens_seen": 154173524,
      "step": 2752
    },
    {
      "epoch": 6.129175946547885,
      "loss": 0.9162663221359253,
      "loss_ce": 0.0002507681492716074,
      "loss_iou": 0.3828125,
      "loss_num": 0.0296630859375,
      "loss_xval": 0.9140625,
      "num_input_tokens_seen": 154173524,
      "step": 2752
    },
    {
      "epoch": 6.131403118040089,
      "grad_norm": 18.245227813720703,
      "learning_rate": 1e-06,
      "loss": 0.5699,
      "num_input_tokens_seen": 154231616,
      "step": 2753
    },
    {
      "epoch": 6.131403118040089,
      "loss": 0.5690479278564453,
      "loss_ce": 0.00020031025633215904,
      "loss_iou": 0.251953125,
      "loss_num": 0.0126953125,
      "loss_xval": 0.5703125,
      "num_input_tokens_seen": 154231616,
      "step": 2753
    },
    {
      "epoch": 6.133630289532294,
      "grad_norm": 15.791337013244629,
      "learning_rate": 1e-06,
      "loss": 0.7701,
      "num_input_tokens_seen": 154288064,
      "step": 2754
    },
    {
      "epoch": 6.133630289532294,
      "loss": 0.6845183968544006,
      "loss_ce": 0.00019223052368033677,
      "loss_iou": 0.291015625,
      "loss_num": 0.0206298828125,
      "loss_xval": 0.68359375,
      "num_input_tokens_seen": 154288064,
      "step": 2754
    },
    {
      "epoch": 6.135857461024499,
      "grad_norm": 18.88831901550293,
      "learning_rate": 1e-06,
      "loss": 0.5883,
      "num_input_tokens_seen": 154340732,
      "step": 2755
    },
    {
      "epoch": 6.135857461024499,
      "loss": 0.5457676649093628,
      "loss_ce": 0.00023538943787571043,
      "loss_iou": 0.25390625,
      "loss_num": 0.00787353515625,
      "loss_xval": 0.546875,
      "num_input_tokens_seen": 154340732,
      "step": 2755
    },
    {
      "epoch": 6.138084632516704,
      "grad_norm": 15.97834300994873,
      "learning_rate": 1e-06,
      "loss": 0.5491,
      "num_input_tokens_seen": 154392456,
      "step": 2756
    },
    {
      "epoch": 6.138084632516704,
      "loss": 0.5473060011863708,
      "loss_ce": 0.00018683119560591877,
      "loss_iou": 0.24609375,
      "loss_num": 0.0107421875,
      "loss_xval": 0.546875,
      "num_input_tokens_seen": 154392456,
      "step": 2756
    },
    {
      "epoch": 6.140311804008909,
      "grad_norm": 13.842836380004883,
      "learning_rate": 1e-06,
      "loss": 0.6017,
      "num_input_tokens_seen": 154448276,
      "step": 2757
    },
    {
      "epoch": 6.140311804008909,
      "loss": 0.6372648477554321,
      "loss_ce": 0.00017988680338021368,
      "loss_iou": 0.271484375,
      "loss_num": 0.018798828125,
      "loss_xval": 0.63671875,
      "num_input_tokens_seen": 154448276,
      "step": 2757
    },
    {
      "epoch": 6.142538975501114,
      "grad_norm": 20.629444122314453,
      "learning_rate": 1e-06,
      "loss": 0.6205,
      "num_input_tokens_seen": 154507936,
      "step": 2758
    },
    {
      "epoch": 6.142538975501114,
      "loss": 0.643352210521698,
      "loss_ce": 0.00028582755476236343,
      "loss_iou": 0.25390625,
      "loss_num": 0.0274658203125,
      "loss_xval": 0.64453125,
      "num_input_tokens_seen": 154507936,
      "step": 2758
    },
    {
      "epoch": 6.144766146993319,
      "grad_norm": 26.82574462890625,
      "learning_rate": 1e-06,
      "loss": 0.7818,
      "num_input_tokens_seen": 154564448,
      "step": 2759
    },
    {
      "epoch": 6.144766146993319,
      "loss": 0.9775708913803101,
      "loss_ce": 0.0002760253846645355,
      "loss_iou": 0.427734375,
      "loss_num": 0.025146484375,
      "loss_xval": 0.9765625,
      "num_input_tokens_seen": 154564448,
      "step": 2759
    },
    {
      "epoch": 6.146993318485523,
      "grad_norm": 21.474916458129883,
      "learning_rate": 1e-06,
      "loss": 0.8228,
      "num_input_tokens_seen": 154621152,
      "step": 2760
    },
    {
      "epoch": 6.146993318485523,
      "loss": 0.8825445175170898,
      "loss_ce": 0.00022031497792340815,
      "loss_iou": 0.33203125,
      "loss_num": 0.044189453125,
      "loss_xval": 0.8828125,
      "num_input_tokens_seen": 154621152,
      "step": 2760
    },
    {
      "epoch": 6.149220489977728,
      "grad_norm": 21.773693084716797,
      "learning_rate": 1e-06,
      "loss": 0.9334,
      "num_input_tokens_seen": 154676312,
      "step": 2761
    },
    {
      "epoch": 6.149220489977728,
      "loss": 0.9210202693939209,
      "loss_ce": 0.00024393736384809017,
      "loss_iou": 0.384765625,
      "loss_num": 0.0299072265625,
      "loss_xval": 0.921875,
      "num_input_tokens_seen": 154676312,
      "step": 2761
    },
    {
      "epoch": 6.151447661469933,
      "grad_norm": 18.154090881347656,
      "learning_rate": 1e-06,
      "loss": 0.563,
      "num_input_tokens_seen": 154732704,
      "step": 2762
    },
    {
      "epoch": 6.151447661469933,
      "loss": 0.3975493311882019,
      "loss_ce": 0.00033253489527851343,
      "loss_iou": 0.158203125,
      "loss_num": 0.0159912109375,
      "loss_xval": 0.396484375,
      "num_input_tokens_seen": 154732704,
      "step": 2762
    },
    {
      "epoch": 6.153674832962138,
      "grad_norm": 20.072853088378906,
      "learning_rate": 1e-06,
      "loss": 0.5174,
      "num_input_tokens_seen": 154790408,
      "step": 2763
    },
    {
      "epoch": 6.153674832962138,
      "loss": 0.5634435415267944,
      "loss_ce": 0.00021113727416377515,
      "loss_iou": 0.2470703125,
      "loss_num": 0.01397705078125,
      "loss_xval": 0.5625,
      "num_input_tokens_seen": 154790408,
      "step": 2763
    },
    {
      "epoch": 6.155902004454343,
      "grad_norm": 16.74744987487793,
      "learning_rate": 1e-06,
      "loss": 0.729,
      "num_input_tokens_seen": 154847920,
      "step": 2764
    },
    {
      "epoch": 6.155902004454343,
      "loss": 0.8036819696426392,
      "loss_ce": 0.00021515009575523436,
      "loss_iou": 0.34375,
      "loss_num": 0.023193359375,
      "loss_xval": 0.8046875,
      "num_input_tokens_seen": 154847920,
      "step": 2764
    },
    {
      "epoch": 6.158129175946548,
      "grad_norm": 31.824892044067383,
      "learning_rate": 1e-06,
      "loss": 0.8517,
      "num_input_tokens_seen": 154904104,
      "step": 2765
    },
    {
      "epoch": 6.158129175946548,
      "loss": 0.874335527420044,
      "loss_ce": 0.00018998724408447742,
      "loss_iou": 0.38671875,
      "loss_num": 0.02001953125,
      "loss_xval": 0.875,
      "num_input_tokens_seen": 154904104,
      "step": 2765
    },
    {
      "epoch": 6.160356347438753,
      "grad_norm": 18.871536254882812,
      "learning_rate": 1e-06,
      "loss": 0.6084,
      "num_input_tokens_seen": 154960796,
      "step": 2766
    },
    {
      "epoch": 6.160356347438753,
      "loss": 0.6629579067230225,
      "loss_ce": 0.000360234291292727,
      "loss_iou": 0.265625,
      "loss_num": 0.0264892578125,
      "loss_xval": 0.6640625,
      "num_input_tokens_seen": 154960796,
      "step": 2766
    },
    {
      "epoch": 6.1625835189309575,
      "grad_norm": 18.299470901489258,
      "learning_rate": 1e-06,
      "loss": 0.5538,
      "num_input_tokens_seen": 155013768,
      "step": 2767
    },
    {
      "epoch": 6.1625835189309575,
      "loss": 0.6405810117721558,
      "loss_ce": 0.00020016740018036216,
      "loss_iou": 0.28515625,
      "loss_num": 0.01409912109375,
      "loss_xval": 0.640625,
      "num_input_tokens_seen": 155013768,
      "step": 2767
    },
    {
      "epoch": 6.164810690423162,
      "grad_norm": 13.490962982177734,
      "learning_rate": 1e-06,
      "loss": 0.8451,
      "num_input_tokens_seen": 155069988,
      "step": 2768
    },
    {
      "epoch": 6.164810690423162,
      "loss": 0.865939199924469,
      "loss_ce": 0.00021657557226717472,
      "loss_iou": 0.36328125,
      "loss_num": 0.027587890625,
      "loss_xval": 0.8671875,
      "num_input_tokens_seen": 155069988,
      "step": 2768
    },
    {
      "epoch": 6.167037861915367,
      "grad_norm": 15.8577241897583,
      "learning_rate": 1e-06,
      "loss": 0.6934,
      "num_input_tokens_seen": 155125976,
      "step": 2769
    },
    {
      "epoch": 6.167037861915367,
      "loss": 0.7875275611877441,
      "loss_ce": 0.00017407389532309026,
      "loss_iou": 0.3359375,
      "loss_num": 0.023193359375,
      "loss_xval": 0.7890625,
      "num_input_tokens_seen": 155125976,
      "step": 2769
    },
    {
      "epoch": 6.169265033407572,
      "grad_norm": 15.64920711517334,
      "learning_rate": 1e-06,
      "loss": 0.584,
      "num_input_tokens_seen": 155181288,
      "step": 2770
    },
    {
      "epoch": 6.169265033407572,
      "loss": 0.6932950615882874,
      "loss_ce": 0.0004239326517563313,
      "loss_iou": 0.291015625,
      "loss_num": 0.0224609375,
      "loss_xval": 0.69140625,
      "num_input_tokens_seen": 155181288,
      "step": 2770
    },
    {
      "epoch": 6.171492204899777,
      "grad_norm": 25.973970413208008,
      "learning_rate": 1e-06,
      "loss": 0.5113,
      "num_input_tokens_seen": 155238888,
      "step": 2771
    },
    {
      "epoch": 6.171492204899777,
      "loss": 0.5262500047683716,
      "loss_ce": 0.0002490488113835454,
      "loss_iou": 0.228515625,
      "loss_num": 0.01409912109375,
      "loss_xval": 0.52734375,
      "num_input_tokens_seen": 155238888,
      "step": 2771
    },
    {
      "epoch": 6.173719376391982,
      "grad_norm": 23.871335983276367,
      "learning_rate": 1e-06,
      "loss": 0.8139,
      "num_input_tokens_seen": 155294792,
      "step": 2772
    },
    {
      "epoch": 6.173719376391982,
      "loss": 0.7459384202957153,
      "loss_ce": 0.0004549958393909037,
      "loss_iou": 0.3125,
      "loss_num": 0.0238037109375,
      "loss_xval": 0.74609375,
      "num_input_tokens_seen": 155294792,
      "step": 2772
    },
    {
      "epoch": 6.1759465478841875,
      "grad_norm": 24.223966598510742,
      "learning_rate": 1e-06,
      "loss": 0.8061,
      "num_input_tokens_seen": 155350512,
      "step": 2773
    },
    {
      "epoch": 6.1759465478841875,
      "loss": 1.045411229133606,
      "loss_ce": 0.00024516499252058566,
      "loss_iou": 0.462890625,
      "loss_num": 0.0235595703125,
      "loss_xval": 1.046875,
      "num_input_tokens_seen": 155350512,
      "step": 2773
    },
    {
      "epoch": 6.178173719376392,
      "grad_norm": 15.486029624938965,
      "learning_rate": 1e-06,
      "loss": 0.6705,
      "num_input_tokens_seen": 155404252,
      "step": 2774
    },
    {
      "epoch": 6.178173719376392,
      "loss": 0.8749101758003235,
      "loss_ce": 0.0001542975369375199,
      "loss_iou": 0.333984375,
      "loss_num": 0.041748046875,
      "loss_xval": 0.875,
      "num_input_tokens_seen": 155404252,
      "step": 2774
    },
    {
      "epoch": 6.180400890868597,
      "grad_norm": 20.75381851196289,
      "learning_rate": 1e-06,
      "loss": 0.6957,
      "num_input_tokens_seen": 155459700,
      "step": 2775
    },
    {
      "epoch": 6.180400890868597,
      "loss": 0.5211961269378662,
      "loss_ce": 0.00020005203259643167,
      "loss_iou": 0.2373046875,
      "loss_num": 0.00921630859375,
      "loss_xval": 0.51953125,
      "num_input_tokens_seen": 155459700,
      "step": 2775
    },
    {
      "epoch": 6.182628062360802,
      "grad_norm": 19.334497451782227,
      "learning_rate": 1e-06,
      "loss": 0.6821,
      "num_input_tokens_seen": 155517152,
      "step": 2776
    },
    {
      "epoch": 6.182628062360802,
      "loss": 0.7436808347702026,
      "loss_ce": 0.0002726712264120579,
      "loss_iou": 0.314453125,
      "loss_num": 0.02294921875,
      "loss_xval": 0.7421875,
      "num_input_tokens_seen": 155517152,
      "step": 2776
    },
    {
      "epoch": 6.184855233853007,
      "grad_norm": 13.525506019592285,
      "learning_rate": 1e-06,
      "loss": 0.5351,
      "num_input_tokens_seen": 155574720,
      "step": 2777
    },
    {
      "epoch": 6.184855233853007,
      "loss": 0.42132747173309326,
      "loss_ce": 0.0001848653773777187,
      "loss_iou": 0.1845703125,
      "loss_num": 0.01055908203125,
      "loss_xval": 0.421875,
      "num_input_tokens_seen": 155574720,
      "step": 2777
    },
    {
      "epoch": 6.187082405345212,
      "grad_norm": 13.232222557067871,
      "learning_rate": 1e-06,
      "loss": 0.5728,
      "num_input_tokens_seen": 155631268,
      "step": 2778
    },
    {
      "epoch": 6.187082405345212,
      "loss": 0.5074321627616882,
      "loss_ce": 0.00023002157104201615,
      "loss_iou": 0.2021484375,
      "loss_num": 0.020751953125,
      "loss_xval": 0.5078125,
      "num_input_tokens_seen": 155631268,
      "step": 2778
    },
    {
      "epoch": 6.189309576837417,
      "grad_norm": 31.53081703186035,
      "learning_rate": 1e-06,
      "loss": 0.6913,
      "num_input_tokens_seen": 155688784,
      "step": 2779
    },
    {
      "epoch": 6.189309576837417,
      "loss": 0.9023219347000122,
      "loss_ce": 0.0002223325427621603,
      "loss_iou": 0.345703125,
      "loss_num": 0.0419921875,
      "loss_xval": 0.90234375,
      "num_input_tokens_seen": 155688784,
      "step": 2779
    },
    {
      "epoch": 6.1915367483296215,
      "grad_norm": 20.483375549316406,
      "learning_rate": 1e-06,
      "loss": 0.6798,
      "num_input_tokens_seen": 155744636,
      "step": 2780
    },
    {
      "epoch": 6.1915367483296215,
      "loss": 0.7736812233924866,
      "loss_ce": 0.00024372681218665093,
      "loss_iou": 0.34765625,
      "loss_num": 0.0159912109375,
      "loss_xval": 0.7734375,
      "num_input_tokens_seen": 155744636,
      "step": 2780
    },
    {
      "epoch": 6.193763919821826,
      "grad_norm": 13.428009986877441,
      "learning_rate": 1e-06,
      "loss": 0.5512,
      "num_input_tokens_seen": 155803476,
      "step": 2781
    },
    {
      "epoch": 6.193763919821826,
      "loss": 0.5040408372879028,
      "loss_ce": 0.0002566420880611986,
      "loss_iou": 0.2216796875,
      "loss_num": 0.01202392578125,
      "loss_xval": 0.50390625,
      "num_input_tokens_seen": 155803476,
      "step": 2781
    },
    {
      "epoch": 6.195991091314031,
      "grad_norm": 23.016326904296875,
      "learning_rate": 1e-06,
      "loss": 0.6668,
      "num_input_tokens_seen": 155858004,
      "step": 2782
    },
    {
      "epoch": 6.195991091314031,
      "loss": 0.6006724834442139,
      "loss_ce": 0.0002086429885821417,
      "loss_iou": 0.26953125,
      "loss_num": 0.01202392578125,
      "loss_xval": 0.6015625,
      "num_input_tokens_seen": 155858004,
      "step": 2782
    },
    {
      "epoch": 6.198218262806236,
      "grad_norm": 18.91625213623047,
      "learning_rate": 1e-06,
      "loss": 0.5554,
      "num_input_tokens_seen": 155908832,
      "step": 2783
    },
    {
      "epoch": 6.198218262806236,
      "loss": 0.4532051086425781,
      "loss_ce": 0.00020219493308104575,
      "loss_iou": 0.1884765625,
      "loss_num": 0.0152587890625,
      "loss_xval": 0.453125,
      "num_input_tokens_seen": 155908832,
      "step": 2783
    },
    {
      "epoch": 6.200445434298441,
      "grad_norm": 21.05170440673828,
      "learning_rate": 1e-06,
      "loss": 0.5745,
      "num_input_tokens_seen": 155965312,
      "step": 2784
    },
    {
      "epoch": 6.200445434298441,
      "loss": 0.6314213275909424,
      "loss_ce": 0.00019577116472646594,
      "loss_iou": 0.279296875,
      "loss_num": 0.01416015625,
      "loss_xval": 0.6328125,
      "num_input_tokens_seen": 155965312,
      "step": 2784
    },
    {
      "epoch": 6.202672605790646,
      "grad_norm": 20.375423431396484,
      "learning_rate": 1e-06,
      "loss": 0.6475,
      "num_input_tokens_seen": 156018284,
      "step": 2785
    },
    {
      "epoch": 6.202672605790646,
      "loss": 0.7216565608978271,
      "loss_ce": 0.0002210296515841037,
      "loss_iou": 0.2890625,
      "loss_num": 0.0284423828125,
      "loss_xval": 0.72265625,
      "num_input_tokens_seen": 156018284,
      "step": 2785
    },
    {
      "epoch": 6.204899777282851,
      "grad_norm": 16.35026741027832,
      "learning_rate": 1e-06,
      "loss": 0.7937,
      "num_input_tokens_seen": 156075156,
      "step": 2786
    },
    {
      "epoch": 6.204899777282851,
      "loss": 0.8708457946777344,
      "loss_ce": 0.00024033612862695009,
      "loss_iou": 0.37890625,
      "loss_num": 0.0228271484375,
      "loss_xval": 0.87109375,
      "num_input_tokens_seen": 156075156,
      "step": 2786
    },
    {
      "epoch": 6.2071269487750556,
      "grad_norm": 23.919958114624023,
      "learning_rate": 1e-06,
      "loss": 0.5475,
      "num_input_tokens_seen": 156130308,
      "step": 2787
    },
    {
      "epoch": 6.2071269487750556,
      "loss": 0.5468391180038452,
      "loss_ce": 0.00020827000844292343,
      "loss_iou": 0.2021484375,
      "loss_num": 0.0284423828125,
      "loss_xval": 0.546875,
      "num_input_tokens_seen": 156130308,
      "step": 2787
    },
    {
      "epoch": 6.20935412026726,
      "grad_norm": 14.023747444152832,
      "learning_rate": 1e-06,
      "loss": 0.5289,
      "num_input_tokens_seen": 156185484,
      "step": 2788
    },
    {
      "epoch": 6.20935412026726,
      "loss": 0.6045721769332886,
      "loss_ce": 0.00020206648332532495,
      "loss_iou": 0.25390625,
      "loss_num": 0.0191650390625,
      "loss_xval": 0.60546875,
      "num_input_tokens_seen": 156185484,
      "step": 2788
    },
    {
      "epoch": 6.211581291759465,
      "grad_norm": 33.71453094482422,
      "learning_rate": 1e-06,
      "loss": 0.7773,
      "num_input_tokens_seen": 156240220,
      "step": 2789
    },
    {
      "epoch": 6.211581291759465,
      "loss": 0.7553761601448059,
      "loss_ce": 0.0002492480562068522,
      "loss_iou": 0.318359375,
      "loss_num": 0.0240478515625,
      "loss_xval": 0.75390625,
      "num_input_tokens_seen": 156240220,
      "step": 2789
    },
    {
      "epoch": 6.21380846325167,
      "grad_norm": 15.885749816894531,
      "learning_rate": 1e-06,
      "loss": 0.7896,
      "num_input_tokens_seen": 156294752,
      "step": 2790
    },
    {
      "epoch": 6.21380846325167,
      "loss": 0.7173956632614136,
      "loss_ce": 0.00035462420783005655,
      "loss_iou": 0.326171875,
      "loss_num": 0.01300048828125,
      "loss_xval": 0.71875,
      "num_input_tokens_seen": 156294752,
      "step": 2790
    },
    {
      "epoch": 6.216035634743875,
      "grad_norm": 23.221410751342773,
      "learning_rate": 1e-06,
      "loss": 0.5949,
      "num_input_tokens_seen": 156350172,
      "step": 2791
    },
    {
      "epoch": 6.216035634743875,
      "loss": 0.4679461717605591,
      "loss_ce": 0.00017272785771638155,
      "loss_iou": 0.1923828125,
      "loss_num": 0.0166015625,
      "loss_xval": 0.46875,
      "num_input_tokens_seen": 156350172,
      "step": 2791
    },
    {
      "epoch": 6.21826280623608,
      "grad_norm": 18.390562057495117,
      "learning_rate": 1e-06,
      "loss": 0.7924,
      "num_input_tokens_seen": 156405748,
      "step": 2792
    },
    {
      "epoch": 6.21826280623608,
      "loss": 0.5683201551437378,
      "loss_ce": 0.0002049248432740569,
      "loss_iou": 0.251953125,
      "loss_num": 0.012451171875,
      "loss_xval": 0.56640625,
      "num_input_tokens_seen": 156405748,
      "step": 2792
    },
    {
      "epoch": 6.220489977728285,
      "grad_norm": 21.410785675048828,
      "learning_rate": 1e-06,
      "loss": 0.6411,
      "num_input_tokens_seen": 156458592,
      "step": 2793
    },
    {
      "epoch": 6.220489977728285,
      "loss": 0.49654620885849,
      "loss_ce": 0.0002083319704979658,
      "loss_iou": 0.21484375,
      "loss_num": 0.01348876953125,
      "loss_xval": 0.49609375,
      "num_input_tokens_seen": 156458592,
      "step": 2793
    },
    {
      "epoch": 6.22271714922049,
      "grad_norm": 16.62869644165039,
      "learning_rate": 1e-06,
      "loss": 0.6773,
      "num_input_tokens_seen": 156515956,
      "step": 2794
    },
    {
      "epoch": 6.22271714922049,
      "loss": 0.6638964414596558,
      "loss_ce": 0.00020011054584756494,
      "loss_iou": 0.2734375,
      "loss_num": 0.0233154296875,
      "loss_xval": 0.6640625,
      "num_input_tokens_seen": 156515956,
      "step": 2794
    },
    {
      "epoch": 6.224944320712694,
      "grad_norm": 17.955623626708984,
      "learning_rate": 1e-06,
      "loss": 0.5419,
      "num_input_tokens_seen": 156571336,
      "step": 2795
    },
    {
      "epoch": 6.224944320712694,
      "loss": 0.5947801470756531,
      "loss_ce": 0.00017566801398061216,
      "loss_iou": 0.25,
      "loss_num": 0.01904296875,
      "loss_xval": 0.59375,
      "num_input_tokens_seen": 156571336,
      "step": 2795
    },
    {
      "epoch": 6.2271714922049,
      "grad_norm": 16.663782119750977,
      "learning_rate": 1e-06,
      "loss": 0.5893,
      "num_input_tokens_seen": 156627776,
      "step": 2796
    },
    {
      "epoch": 6.2271714922049,
      "loss": 0.7995222806930542,
      "loss_ce": 0.00020588196639437228,
      "loss_iou": 0.30859375,
      "loss_num": 0.03662109375,
      "loss_xval": 0.80078125,
      "num_input_tokens_seen": 156627776,
      "step": 2796
    },
    {
      "epoch": 6.229398663697105,
      "grad_norm": 16.558795928955078,
      "learning_rate": 1e-06,
      "loss": 0.7245,
      "num_input_tokens_seen": 156682772,
      "step": 2797
    },
    {
      "epoch": 6.229398663697105,
      "loss": 0.7288368344306946,
      "loss_ce": 0.0003212342271581292,
      "loss_iou": 0.306640625,
      "loss_num": 0.022705078125,
      "loss_xval": 0.7265625,
      "num_input_tokens_seen": 156682772,
      "step": 2797
    },
    {
      "epoch": 6.23162583518931,
      "grad_norm": 56.9061164855957,
      "learning_rate": 1e-06,
      "loss": 0.6461,
      "num_input_tokens_seen": 156739268,
      "step": 2798
    },
    {
      "epoch": 6.23162583518931,
      "loss": 0.5878466367721558,
      "loss_ce": 0.00020016153575852513,
      "loss_iou": 0.265625,
      "loss_num": 0.0108642578125,
      "loss_xval": 0.5859375,
      "num_input_tokens_seen": 156739268,
      "step": 2798
    },
    {
      "epoch": 6.233853006681515,
      "grad_norm": 17.540464401245117,
      "learning_rate": 1e-06,
      "loss": 0.8046,
      "num_input_tokens_seen": 156797304,
      "step": 2799
    },
    {
      "epoch": 6.233853006681515,
      "loss": 0.8160011768341064,
      "loss_ce": 0.0002052470954367891,
      "loss_iou": 0.357421875,
      "loss_num": 0.0201416015625,
      "loss_xval": 0.81640625,
      "num_input_tokens_seen": 156797304,
      "step": 2799
    },
    {
      "epoch": 6.23608017817372,
      "grad_norm": 23.366252899169922,
      "learning_rate": 1e-06,
      "loss": 0.7489,
      "num_input_tokens_seen": 156852524,
      "step": 2800
    },
    {
      "epoch": 6.23608017817372,
      "loss": 0.7832313776016235,
      "loss_ce": 0.0002724075166042894,
      "loss_iou": 0.298828125,
      "loss_num": 0.037109375,
      "loss_xval": 0.78125,
      "num_input_tokens_seen": 156852524,
      "step": 2800
    },
    {
      "epoch": 6.2383073496659245,
      "grad_norm": 18.745590209960938,
      "learning_rate": 1e-06,
      "loss": 0.602,
      "num_input_tokens_seen": 156909388,
      "step": 2801
    },
    {
      "epoch": 6.2383073496659245,
      "loss": 0.377541720867157,
      "loss_ce": 0.0002223837363999337,
      "loss_iou": 0.16015625,
      "loss_num": 0.01141357421875,
      "loss_xval": 0.376953125,
      "num_input_tokens_seen": 156909388,
      "step": 2801
    },
    {
      "epoch": 6.240534521158129,
      "grad_norm": 16.821367263793945,
      "learning_rate": 1e-06,
      "loss": 0.6813,
      "num_input_tokens_seen": 156965248,
      "step": 2802
    },
    {
      "epoch": 6.240534521158129,
      "loss": 0.6255514621734619,
      "loss_ce": 0.00018525280756875873,
      "loss_iou": 0.2490234375,
      "loss_num": 0.025634765625,
      "loss_xval": 0.625,
      "num_input_tokens_seen": 156965248,
      "step": 2802
    },
    {
      "epoch": 6.242761692650334,
      "grad_norm": 17.60284996032715,
      "learning_rate": 1e-06,
      "loss": 0.6629,
      "num_input_tokens_seen": 157021544,
      "step": 2803
    },
    {
      "epoch": 6.242761692650334,
      "loss": 0.7033724784851074,
      "loss_ce": 0.00024746524286456406,
      "loss_iou": 0.310546875,
      "loss_num": 0.0166015625,
      "loss_xval": 0.703125,
      "num_input_tokens_seen": 157021544,
      "step": 2803
    },
    {
      "epoch": 6.244988864142539,
      "grad_norm": 18.265342712402344,
      "learning_rate": 1e-06,
      "loss": 0.5529,
      "num_input_tokens_seen": 157077008,
      "step": 2804
    },
    {
      "epoch": 6.244988864142539,
      "loss": 0.495932936668396,
      "loss_ce": 0.0002053846837952733,
      "loss_iou": 0.1982421875,
      "loss_num": 0.019775390625,
      "loss_xval": 0.49609375,
      "num_input_tokens_seen": 157077008,
      "step": 2804
    },
    {
      "epoch": 6.247216035634744,
      "grad_norm": 36.90097427368164,
      "learning_rate": 1e-06,
      "loss": 0.6764,
      "num_input_tokens_seen": 157131556,
      "step": 2805
    },
    {
      "epoch": 6.247216035634744,
      "loss": 0.6967709064483643,
      "loss_ce": 0.00023769214749336243,
      "loss_iou": 0.306640625,
      "loss_num": 0.0164794921875,
      "loss_xval": 0.6953125,
      "num_input_tokens_seen": 157131556,
      "step": 2805
    },
    {
      "epoch": 6.249443207126949,
      "grad_norm": 20.952041625976562,
      "learning_rate": 1e-06,
      "loss": 0.6065,
      "num_input_tokens_seen": 157187128,
      "step": 2806
    },
    {
      "epoch": 6.249443207126949,
      "loss": 0.496565043926239,
      "loss_ce": 0.00022713570797350258,
      "loss_iou": 0.208984375,
      "loss_num": 0.0157470703125,
      "loss_xval": 0.49609375,
      "num_input_tokens_seen": 157187128,
      "step": 2806
    },
    {
      "epoch": 6.251670378619154,
      "grad_norm": 27.254335403442383,
      "learning_rate": 1e-06,
      "loss": 0.7668,
      "num_input_tokens_seen": 157242080,
      "step": 2807
    },
    {
      "epoch": 6.251670378619154,
      "loss": 0.721343457698822,
      "loss_ce": 0.0002741218195296824,
      "loss_iou": 0.28125,
      "loss_num": 0.03173828125,
      "loss_xval": 0.72265625,
      "num_input_tokens_seen": 157242080,
      "step": 2807
    },
    {
      "epoch": 6.2538975501113585,
      "grad_norm": 19.683401107788086,
      "learning_rate": 1e-06,
      "loss": 0.4727,
      "num_input_tokens_seen": 157299408,
      "step": 2808
    },
    {
      "epoch": 6.2538975501113585,
      "loss": 0.5055980682373047,
      "loss_ce": 0.00022694582003168762,
      "loss_iou": 0.22265625,
      "loss_num": 0.01202392578125,
      "loss_xval": 0.50390625,
      "num_input_tokens_seen": 157299408,
      "step": 2808
    },
    {
      "epoch": 6.256124721603563,
      "grad_norm": 16.38343620300293,
      "learning_rate": 1e-06,
      "loss": 0.6176,
      "num_input_tokens_seen": 157354524,
      "step": 2809
    },
    {
      "epoch": 6.256124721603563,
      "loss": 0.5820066332817078,
      "loss_ce": 0.0002195206907344982,
      "loss_iou": 0.232421875,
      "loss_num": 0.0234375,
      "loss_xval": 0.58203125,
      "num_input_tokens_seen": 157354524,
      "step": 2809
    },
    {
      "epoch": 6.258351893095768,
      "grad_norm": 23.633302688598633,
      "learning_rate": 1e-06,
      "loss": 0.7708,
      "num_input_tokens_seen": 157409136,
      "step": 2810
    },
    {
      "epoch": 6.258351893095768,
      "loss": 1.0537309646606445,
      "loss_ce": 0.00026410428108647466,
      "loss_iou": 0.46875,
      "loss_num": 0.0225830078125,
      "loss_xval": 1.0546875,
      "num_input_tokens_seen": 157409136,
      "step": 2810
    },
    {
      "epoch": 6.260579064587973,
      "grad_norm": 17.914230346679688,
      "learning_rate": 1e-06,
      "loss": 0.8414,
      "num_input_tokens_seen": 157463620,
      "step": 2811
    },
    {
      "epoch": 6.260579064587973,
      "loss": 1.0271397829055786,
      "loss_ce": 0.00028427952202036977,
      "loss_iou": 0.45703125,
      "loss_num": 0.022216796875,
      "loss_xval": 1.0234375,
      "num_input_tokens_seen": 157463620,
      "step": 2811
    },
    {
      "epoch": 6.262806236080178,
      "grad_norm": 20.162099838256836,
      "learning_rate": 1e-06,
      "loss": 0.6692,
      "num_input_tokens_seen": 157518172,
      "step": 2812
    },
    {
      "epoch": 6.262806236080178,
      "loss": 0.709923267364502,
      "loss_ce": 0.0004505990073084831,
      "loss_iou": 0.30078125,
      "loss_num": 0.0213623046875,
      "loss_xval": 0.7109375,
      "num_input_tokens_seen": 157518172,
      "step": 2812
    },
    {
      "epoch": 6.265033407572383,
      "grad_norm": 20.278295516967773,
      "learning_rate": 1e-06,
      "loss": 0.6405,
      "num_input_tokens_seen": 157573484,
      "step": 2813
    },
    {
      "epoch": 6.265033407572383,
      "loss": 0.7189103364944458,
      "loss_ce": 0.0001603504060767591,
      "loss_iou": 0.2890625,
      "loss_num": 0.028076171875,
      "loss_xval": 0.71875,
      "num_input_tokens_seen": 157573484,
      "step": 2813
    },
    {
      "epoch": 6.267260579064588,
      "grad_norm": 19.324562072753906,
      "learning_rate": 1e-06,
      "loss": 0.6074,
      "num_input_tokens_seen": 157627908,
      "step": 2814
    },
    {
      "epoch": 6.267260579064588,
      "loss": 0.7765587568283081,
      "loss_ce": 0.0002525679301470518,
      "loss_iou": 0.330078125,
      "loss_num": 0.0234375,
      "loss_xval": 0.77734375,
      "num_input_tokens_seen": 157627908,
      "step": 2814
    },
    {
      "epoch": 6.2694877505567925,
      "grad_norm": 15.711827278137207,
      "learning_rate": 1e-06,
      "loss": 0.676,
      "num_input_tokens_seen": 157685372,
      "step": 2815
    },
    {
      "epoch": 6.2694877505567925,
      "loss": 0.6160228252410889,
      "loss_ce": 0.0003001945442520082,
      "loss_iou": 0.259765625,
      "loss_num": 0.0189208984375,
      "loss_xval": 0.6171875,
      "num_input_tokens_seen": 157685372,
      "step": 2815
    },
    {
      "epoch": 6.271714922048997,
      "grad_norm": 24.523929595947266,
      "learning_rate": 1e-06,
      "loss": 0.7507,
      "num_input_tokens_seen": 157740784,
      "step": 2816
    },
    {
      "epoch": 6.271714922048997,
      "loss": 0.6751787662506104,
      "loss_ce": 0.0003741044201888144,
      "loss_iou": 0.28125,
      "loss_num": 0.0225830078125,
      "loss_xval": 0.67578125,
      "num_input_tokens_seen": 157740784,
      "step": 2816
    },
    {
      "epoch": 6.273942093541203,
      "grad_norm": 17.52358627319336,
      "learning_rate": 1e-06,
      "loss": 0.627,
      "num_input_tokens_seen": 157797720,
      "step": 2817
    },
    {
      "epoch": 6.273942093541203,
      "loss": 0.6270310878753662,
      "loss_ce": 0.00032214989187195897,
      "loss_iou": 0.263671875,
      "loss_num": 0.02001953125,
      "loss_xval": 0.625,
      "num_input_tokens_seen": 157797720,
      "step": 2817
    },
    {
      "epoch": 6.276169265033408,
      "grad_norm": 15.597443580627441,
      "learning_rate": 1e-06,
      "loss": 0.511,
      "num_input_tokens_seen": 157855748,
      "step": 2818
    },
    {
      "epoch": 6.276169265033408,
      "loss": 0.45937642455101013,
      "loss_ce": 0.0005141303990967572,
      "loss_iou": 0.1962890625,
      "loss_num": 0.01336669921875,
      "loss_xval": 0.458984375,
      "num_input_tokens_seen": 157855748,
      "step": 2818
    },
    {
      "epoch": 6.278396436525613,
      "grad_norm": 20.574974060058594,
      "learning_rate": 1e-06,
      "loss": 0.8684,
      "num_input_tokens_seen": 157906192,
      "step": 2819
    },
    {
      "epoch": 6.278396436525613,
      "loss": 0.9832384586334229,
      "loss_ce": 0.0003282871039118618,
      "loss_iou": 0.4140625,
      "loss_num": 0.0311279296875,
      "loss_xval": 0.984375,
      "num_input_tokens_seen": 157906192,
      "step": 2819
    },
    {
      "epoch": 6.280623608017818,
      "grad_norm": 16.417438507080078,
      "learning_rate": 1e-06,
      "loss": 0.878,
      "num_input_tokens_seen": 157960736,
      "step": 2820
    },
    {
      "epoch": 6.280623608017818,
      "loss": 1.0158679485321045,
      "loss_ce": 0.0002430274907965213,
      "loss_iou": 0.416015625,
      "loss_num": 0.036865234375,
      "loss_xval": 1.015625,
      "num_input_tokens_seen": 157960736,
      "step": 2820
    },
    {
      "epoch": 6.282850779510023,
      "grad_norm": 21.210569381713867,
      "learning_rate": 1e-06,
      "loss": 0.724,
      "num_input_tokens_seen": 158016856,
      "step": 2821
    },
    {
      "epoch": 6.282850779510023,
      "loss": 0.6245869398117065,
      "loss_ce": 0.0001973453618120402,
      "loss_iou": 0.267578125,
      "loss_num": 0.018310546875,
      "loss_xval": 0.625,
      "num_input_tokens_seen": 158016856,
      "step": 2821
    },
    {
      "epoch": 6.285077951002227,
      "grad_norm": 18.306804656982422,
      "learning_rate": 1e-06,
      "loss": 0.7915,
      "num_input_tokens_seen": 158071804,
      "step": 2822
    },
    {
      "epoch": 6.285077951002227,
      "loss": 0.865458071231842,
      "loss_ce": 0.0002236901200376451,
      "loss_iou": 0.361328125,
      "loss_num": 0.0284423828125,
      "loss_xval": 0.8671875,
      "num_input_tokens_seen": 158071804,
      "step": 2822
    },
    {
      "epoch": 6.287305122494432,
      "grad_norm": 18.746957778930664,
      "learning_rate": 1e-06,
      "loss": 0.8188,
      "num_input_tokens_seen": 158126148,
      "step": 2823
    },
    {
      "epoch": 6.287305122494432,
      "loss": 0.8055477142333984,
      "loss_ce": 0.0003718675870914012,
      "loss_iou": 0.359375,
      "loss_num": 0.016845703125,
      "loss_xval": 0.8046875,
      "num_input_tokens_seen": 158126148,
      "step": 2823
    },
    {
      "epoch": 6.289532293986637,
      "grad_norm": 20.189741134643555,
      "learning_rate": 1e-06,
      "loss": 0.7432,
      "num_input_tokens_seen": 158181076,
      "step": 2824
    },
    {
      "epoch": 6.289532293986637,
      "loss": 0.8805779218673706,
      "loss_ce": 0.00020679559384007007,
      "loss_iou": 0.375,
      "loss_num": 0.026123046875,
      "loss_xval": 0.87890625,
      "num_input_tokens_seen": 158181076,
      "step": 2824
    },
    {
      "epoch": 6.291759465478842,
      "grad_norm": 29.9582576751709,
      "learning_rate": 1e-06,
      "loss": 0.8211,
      "num_input_tokens_seen": 158236796,
      "step": 2825
    },
    {
      "epoch": 6.291759465478842,
      "loss": 0.7246193289756775,
      "loss_ce": 0.0002540839195717126,
      "loss_iou": 0.310546875,
      "loss_num": 0.0206298828125,
      "loss_xval": 0.72265625,
      "num_input_tokens_seen": 158236796,
      "step": 2825
    },
    {
      "epoch": 6.293986636971047,
      "grad_norm": 24.43955421447754,
      "learning_rate": 1e-06,
      "loss": 0.6298,
      "num_input_tokens_seen": 158292708,
      "step": 2826
    },
    {
      "epoch": 6.293986636971047,
      "loss": 0.560718297958374,
      "loss_ce": 0.00017140517593361437,
      "loss_iou": 0.251953125,
      "loss_num": 0.01123046875,
      "loss_xval": 0.5625,
      "num_input_tokens_seen": 158292708,
      "step": 2826
    },
    {
      "epoch": 6.296213808463252,
      "grad_norm": 15.927586555480957,
      "learning_rate": 1e-06,
      "loss": 0.7068,
      "num_input_tokens_seen": 158347268,
      "step": 2827
    },
    {
      "epoch": 6.296213808463252,
      "loss": 0.6577511429786682,
      "loss_ce": 0.0002804349351208657,
      "loss_iou": 0.291015625,
      "loss_num": 0.01519775390625,
      "loss_xval": 0.65625,
      "num_input_tokens_seen": 158347268,
      "step": 2827
    },
    {
      "epoch": 6.298440979955457,
      "grad_norm": 17.409822463989258,
      "learning_rate": 1e-06,
      "loss": 0.6776,
      "num_input_tokens_seen": 158402892,
      "step": 2828
    },
    {
      "epoch": 6.298440979955457,
      "loss": 0.6074726581573486,
      "loss_ce": 0.00029491446912288666,
      "loss_iou": 0.26171875,
      "loss_num": 0.0167236328125,
      "loss_xval": 0.60546875,
      "num_input_tokens_seen": 158402892,
      "step": 2828
    },
    {
      "epoch": 6.3006681514476615,
      "grad_norm": 27.617082595825195,
      "learning_rate": 1e-06,
      "loss": 0.604,
      "num_input_tokens_seen": 158457584,
      "step": 2829
    },
    {
      "epoch": 6.3006681514476615,
      "loss": 0.6342703104019165,
      "loss_ce": 0.00023712051915936172,
      "loss_iou": 0.267578125,
      "loss_num": 0.0194091796875,
      "loss_xval": 0.6328125,
      "num_input_tokens_seen": 158457584,
      "step": 2829
    },
    {
      "epoch": 6.302895322939866,
      "grad_norm": 15.794069290161133,
      "learning_rate": 1e-06,
      "loss": 0.5568,
      "num_input_tokens_seen": 158514100,
      "step": 2830
    },
    {
      "epoch": 6.302895322939866,
      "loss": 0.591174840927124,
      "loss_ce": 0.0003545153886079788,
      "loss_iou": 0.255859375,
      "loss_num": 0.015625,
      "loss_xval": 0.58984375,
      "num_input_tokens_seen": 158514100,
      "step": 2830
    },
    {
      "epoch": 6.305122494432071,
      "grad_norm": 19.552928924560547,
      "learning_rate": 1e-06,
      "loss": 0.7125,
      "num_input_tokens_seen": 158571860,
      "step": 2831
    },
    {
      "epoch": 6.305122494432071,
      "loss": 0.5463184118270874,
      "loss_ce": 0.00017578111146576703,
      "loss_iou": 0.2392578125,
      "loss_num": 0.013671875,
      "loss_xval": 0.546875,
      "num_input_tokens_seen": 158571860,
      "step": 2831
    },
    {
      "epoch": 6.307349665924276,
      "grad_norm": 31.347753524780273,
      "learning_rate": 1e-06,
      "loss": 0.5719,
      "num_input_tokens_seen": 158627292,
      "step": 2832
    },
    {
      "epoch": 6.307349665924276,
      "loss": 0.46991389989852905,
      "loss_ce": 0.00018733731121756136,
      "loss_iou": 0.189453125,
      "loss_num": 0.0181884765625,
      "loss_xval": 0.46875,
      "num_input_tokens_seen": 158627292,
      "step": 2832
    },
    {
      "epoch": 6.309576837416481,
      "grad_norm": 19.185462951660156,
      "learning_rate": 1e-06,
      "loss": 0.6727,
      "num_input_tokens_seen": 158682192,
      "step": 2833
    },
    {
      "epoch": 6.309576837416481,
      "loss": 0.6861748695373535,
      "loss_ce": 0.0002617766731418669,
      "loss_iou": 0.275390625,
      "loss_num": 0.02685546875,
      "loss_xval": 0.6875,
      "num_input_tokens_seen": 158682192,
      "step": 2833
    },
    {
      "epoch": 6.311804008908686,
      "grad_norm": 23.759424209594727,
      "learning_rate": 1e-06,
      "loss": 0.5722,
      "num_input_tokens_seen": 158739748,
      "step": 2834
    },
    {
      "epoch": 6.311804008908686,
      "loss": 0.6177768707275391,
      "loss_ce": 0.0002231545513495803,
      "loss_iou": 0.263671875,
      "loss_num": 0.01806640625,
      "loss_xval": 0.6171875,
      "num_input_tokens_seen": 158739748,
      "step": 2834
    },
    {
      "epoch": 6.314031180400891,
      "grad_norm": 26.93903350830078,
      "learning_rate": 1e-06,
      "loss": 0.7831,
      "num_input_tokens_seen": 158792348,
      "step": 2835
    },
    {
      "epoch": 6.314031180400891,
      "loss": 1.0019464492797852,
      "loss_ce": 0.0004816856817342341,
      "loss_iou": 0.42578125,
      "loss_num": 0.02978515625,
      "loss_xval": 1.0,
      "num_input_tokens_seen": 158792348,
      "step": 2835
    },
    {
      "epoch": 6.3162583518930955,
      "grad_norm": 55.32765197753906,
      "learning_rate": 1e-06,
      "loss": 0.7881,
      "num_input_tokens_seen": 158850492,
      "step": 2836
    },
    {
      "epoch": 6.3162583518930955,
      "loss": 0.7746649384498596,
      "loss_ce": 0.00025085339439101517,
      "loss_iou": 0.302734375,
      "loss_num": 0.033935546875,
      "loss_xval": 0.7734375,
      "num_input_tokens_seen": 158850492,
      "step": 2836
    },
    {
      "epoch": 6.3184855233853,
      "grad_norm": 18.002099990844727,
      "learning_rate": 1e-06,
      "loss": 0.5655,
      "num_input_tokens_seen": 158907516,
      "step": 2837
    },
    {
      "epoch": 6.3184855233853,
      "loss": 0.5236201882362366,
      "loss_ce": 0.00018271096632815897,
      "loss_iou": 0.19140625,
      "loss_num": 0.0283203125,
      "loss_xval": 0.5234375,
      "num_input_tokens_seen": 158907516,
      "step": 2837
    },
    {
      "epoch": 6.320712694877505,
      "grad_norm": 20.92586326599121,
      "learning_rate": 1e-06,
      "loss": 0.6947,
      "num_input_tokens_seen": 158962112,
      "step": 2838
    },
    {
      "epoch": 6.320712694877505,
      "loss": 0.7641561031341553,
      "loss_ce": 0.0002400341909378767,
      "loss_iou": 0.328125,
      "loss_num": 0.0213623046875,
      "loss_xval": 0.765625,
      "num_input_tokens_seen": 158962112,
      "step": 2838
    },
    {
      "epoch": 6.32293986636971,
      "grad_norm": 24.1262264251709,
      "learning_rate": 1e-06,
      "loss": 0.8267,
      "num_input_tokens_seen": 159017948,
      "step": 2839
    },
    {
      "epoch": 6.32293986636971,
      "loss": 0.6984615325927734,
      "loss_ce": 0.0002193648397224024,
      "loss_iou": 0.283203125,
      "loss_num": 0.026611328125,
      "loss_xval": 0.69921875,
      "num_input_tokens_seen": 159017948,
      "step": 2839
    },
    {
      "epoch": 6.325167037861915,
      "grad_norm": 14.128987312316895,
      "learning_rate": 1e-06,
      "loss": 0.7434,
      "num_input_tokens_seen": 159072920,
      "step": 2840
    },
    {
      "epoch": 6.325167037861915,
      "loss": 0.782150387763977,
      "loss_ce": 0.00029003899544477463,
      "loss_iou": 0.326171875,
      "loss_num": 0.0257568359375,
      "loss_xval": 0.78125,
      "num_input_tokens_seen": 159072920,
      "step": 2840
    },
    {
      "epoch": 6.327394209354121,
      "grad_norm": 15.3905029296875,
      "learning_rate": 1e-06,
      "loss": 0.5419,
      "num_input_tokens_seen": 159130120,
      "step": 2841
    },
    {
      "epoch": 6.327394209354121,
      "loss": 0.3986847996711731,
      "loss_ce": 0.00024727691197767854,
      "loss_iou": 0.169921875,
      "loss_num": 0.01165771484375,
      "loss_xval": 0.3984375,
      "num_input_tokens_seen": 159130120,
      "step": 2841
    },
    {
      "epoch": 6.3296213808463255,
      "grad_norm": 22.618736267089844,
      "learning_rate": 1e-06,
      "loss": 0.7229,
      "num_input_tokens_seen": 159185880,
      "step": 2842
    },
    {
      "epoch": 6.3296213808463255,
      "loss": 0.5595074892044067,
      "loss_ce": 0.00018136684957426041,
      "loss_iou": 0.240234375,
      "loss_num": 0.015869140625,
      "loss_xval": 0.55859375,
      "num_input_tokens_seen": 159185880,
      "step": 2842
    },
    {
      "epoch": 6.33184855233853,
      "grad_norm": 26.56134605407715,
      "learning_rate": 1e-06,
      "loss": 0.5805,
      "num_input_tokens_seen": 159243236,
      "step": 2843
    },
    {
      "epoch": 6.33184855233853,
      "loss": 0.5183169841766357,
      "loss_ce": 0.0002505767624825239,
      "loss_iou": 0.2255859375,
      "loss_num": 0.01348876953125,
      "loss_xval": 0.51953125,
      "num_input_tokens_seen": 159243236,
      "step": 2843
    },
    {
      "epoch": 6.334075723830735,
      "grad_norm": 18.414081573486328,
      "learning_rate": 1e-06,
      "loss": 0.833,
      "num_input_tokens_seen": 159300220,
      "step": 2844
    },
    {
      "epoch": 6.334075723830735,
      "loss": 0.8649712800979614,
      "loss_ce": 0.00022521875507663935,
      "loss_iou": 0.37109375,
      "loss_num": 0.024658203125,
      "loss_xval": 0.86328125,
      "num_input_tokens_seen": 159300220,
      "step": 2844
    },
    {
      "epoch": 6.33630289532294,
      "grad_norm": 55.4049072265625,
      "learning_rate": 1e-06,
      "loss": 0.5908,
      "num_input_tokens_seen": 159358836,
      "step": 2845
    },
    {
      "epoch": 6.33630289532294,
      "loss": 0.5839510560035706,
      "loss_ce": 0.0002108067856170237,
      "loss_iou": 0.24609375,
      "loss_num": 0.0181884765625,
      "loss_xval": 0.58203125,
      "num_input_tokens_seen": 159358836,
      "step": 2845
    },
    {
      "epoch": 6.338530066815145,
      "grad_norm": 15.77271556854248,
      "learning_rate": 1e-06,
      "loss": 0.6155,
      "num_input_tokens_seen": 159416428,
      "step": 2846
    },
    {
      "epoch": 6.338530066815145,
      "loss": 0.5473378896713257,
      "loss_ce": 0.00021875005040783435,
      "loss_iou": 0.2451171875,
      "loss_num": 0.0115966796875,
      "loss_xval": 0.546875,
      "num_input_tokens_seen": 159416428,
      "step": 2846
    },
    {
      "epoch": 6.34075723830735,
      "grad_norm": 17.767343521118164,
      "learning_rate": 1e-06,
      "loss": 0.4326,
      "num_input_tokens_seen": 159474868,
      "step": 2847
    },
    {
      "epoch": 6.34075723830735,
      "loss": 0.420124888420105,
      "loss_ce": 0.00020299111201893538,
      "loss_iou": 0.1865234375,
      "loss_num": 0.00921630859375,
      "loss_xval": 0.419921875,
      "num_input_tokens_seen": 159474868,
      "step": 2847
    },
    {
      "epoch": 6.342984409799555,
      "grad_norm": 17.833757400512695,
      "learning_rate": 1e-06,
      "loss": 0.6052,
      "num_input_tokens_seen": 159530900,
      "step": 2848
    },
    {
      "epoch": 6.342984409799555,
      "loss": 0.6891921758651733,
      "loss_ce": 0.0002273364079883322,
      "loss_iou": 0.306640625,
      "loss_num": 0.01519775390625,
      "loss_xval": 0.6875,
      "num_input_tokens_seen": 159530900,
      "step": 2848
    },
    {
      "epoch": 6.3452115812917596,
      "grad_norm": 36.907676696777344,
      "learning_rate": 1e-06,
      "loss": 0.9243,
      "num_input_tokens_seen": 159587904,
      "step": 2849
    },
    {
      "epoch": 6.3452115812917596,
      "loss": 0.7456858158111572,
      "loss_ce": 0.00020244505139999092,
      "loss_iou": 0.31640625,
      "loss_num": 0.022705078125,
      "loss_xval": 0.74609375,
      "num_input_tokens_seen": 159587904,
      "step": 2849
    },
    {
      "epoch": 6.347438752783964,
      "grad_norm": 25.969263076782227,
      "learning_rate": 1e-06,
      "loss": 0.671,
      "num_input_tokens_seen": 159643948,
      "step": 2850
    },
    {
      "epoch": 6.347438752783964,
      "loss": 0.7656386494636536,
      "loss_ce": 0.000257776933722198,
      "loss_iou": 0.330078125,
      "loss_num": 0.021240234375,
      "loss_xval": 0.765625,
      "num_input_tokens_seen": 159643948,
      "step": 2850
    },
    {
      "epoch": 6.349665924276169,
      "grad_norm": 16.245384216308594,
      "learning_rate": 1e-06,
      "loss": 0.6789,
      "num_input_tokens_seen": 159701440,
      "step": 2851
    },
    {
      "epoch": 6.349665924276169,
      "loss": 0.6860156655311584,
      "loss_ce": 0.00034670589957386255,
      "loss_iou": 0.27734375,
      "loss_num": 0.026611328125,
      "loss_xval": 0.6875,
      "num_input_tokens_seen": 159701440,
      "step": 2851
    },
    {
      "epoch": 6.351893095768374,
      "grad_norm": 23.11394691467285,
      "learning_rate": 1e-06,
      "loss": 0.6914,
      "num_input_tokens_seen": 159759676,
      "step": 2852
    },
    {
      "epoch": 6.351893095768374,
      "loss": 0.4833386540412903,
      "loss_ce": 0.00018433517834637314,
      "loss_iou": 0.2197265625,
      "loss_num": 0.0089111328125,
      "loss_xval": 0.482421875,
      "num_input_tokens_seen": 159759676,
      "step": 2852
    },
    {
      "epoch": 6.354120267260579,
      "grad_norm": 46.09978103637695,
      "learning_rate": 1e-06,
      "loss": 0.8392,
      "num_input_tokens_seen": 159817828,
      "step": 2853
    },
    {
      "epoch": 6.354120267260579,
      "loss": 0.5963437557220459,
      "loss_ce": 0.00039652473060414195,
      "loss_iou": 0.2333984375,
      "loss_num": 0.02587890625,
      "loss_xval": 0.59765625,
      "num_input_tokens_seen": 159817828,
      "step": 2853
    },
    {
      "epoch": 6.356347438752784,
      "grad_norm": 25.764183044433594,
      "learning_rate": 1e-06,
      "loss": 0.8404,
      "num_input_tokens_seen": 159871560,
      "step": 2854
    },
    {
      "epoch": 6.356347438752784,
      "loss": 0.6220239400863647,
      "loss_ce": 0.00019775178225245327,
      "loss_iou": 0.25,
      "loss_num": 0.02392578125,
      "loss_xval": 0.62109375,
      "num_input_tokens_seen": 159871560,
      "step": 2854
    },
    {
      "epoch": 6.358574610244989,
      "grad_norm": 19.76352882385254,
      "learning_rate": 1e-06,
      "loss": 0.575,
      "num_input_tokens_seen": 159923780,
      "step": 2855
    },
    {
      "epoch": 6.358574610244989,
      "loss": 0.5392453074455261,
      "loss_ce": 0.00018279827781952918,
      "loss_iou": 0.2265625,
      "loss_num": 0.01708984375,
      "loss_xval": 0.5390625,
      "num_input_tokens_seen": 159923780,
      "step": 2855
    },
    {
      "epoch": 6.360801781737194,
      "grad_norm": 19.6981201171875,
      "learning_rate": 1e-06,
      "loss": 0.7724,
      "num_input_tokens_seen": 159977408,
      "step": 2856
    },
    {
      "epoch": 6.360801781737194,
      "loss": 0.7918369770050049,
      "loss_ce": 0.00021093878604006022,
      "loss_iou": 0.33203125,
      "loss_num": 0.025390625,
      "loss_xval": 0.79296875,
      "num_input_tokens_seen": 159977408,
      "step": 2856
    },
    {
      "epoch": 6.363028953229398,
      "grad_norm": 19.517127990722656,
      "learning_rate": 1e-06,
      "loss": 0.4262,
      "num_input_tokens_seen": 160035044,
      "step": 2857
    },
    {
      "epoch": 6.363028953229398,
      "loss": 0.45344364643096924,
      "loss_ce": 0.00019658930250443518,
      "loss_iou": 0.2001953125,
      "loss_num": 0.0107421875,
      "loss_xval": 0.453125,
      "num_input_tokens_seen": 160035044,
      "step": 2857
    },
    {
      "epoch": 6.365256124721603,
      "grad_norm": 27.655757904052734,
      "learning_rate": 1e-06,
      "loss": 0.6266,
      "num_input_tokens_seen": 160091256,
      "step": 2858
    },
    {
      "epoch": 6.365256124721603,
      "loss": 0.45751261711120605,
      "loss_ce": 0.00023723256890662014,
      "loss_iou": 0.1787109375,
      "loss_num": 0.02001953125,
      "loss_xval": 0.45703125,
      "num_input_tokens_seen": 160091256,
      "step": 2858
    },
    {
      "epoch": 6.367483296213808,
      "grad_norm": 13.697670936584473,
      "learning_rate": 1e-06,
      "loss": 0.6436,
      "num_input_tokens_seen": 160146776,
      "step": 2859
    },
    {
      "epoch": 6.367483296213808,
      "loss": 0.7326347827911377,
      "loss_ce": 0.00021289548021741211,
      "loss_iou": 0.306640625,
      "loss_num": 0.02392578125,
      "loss_xval": 0.734375,
      "num_input_tokens_seen": 160146776,
      "step": 2859
    },
    {
      "epoch": 6.369710467706013,
      "grad_norm": 20.36834144592285,
      "learning_rate": 1e-06,
      "loss": 0.7276,
      "num_input_tokens_seen": 160202160,
      "step": 2860
    },
    {
      "epoch": 6.369710467706013,
      "loss": 0.8042212128639221,
      "loss_ce": 0.0002660886093508452,
      "loss_iou": 0.357421875,
      "loss_num": 0.0179443359375,
      "loss_xval": 0.8046875,
      "num_input_tokens_seen": 160202160,
      "step": 2860
    },
    {
      "epoch": 6.371937639198218,
      "grad_norm": 28.106962203979492,
      "learning_rate": 1e-06,
      "loss": 0.6299,
      "num_input_tokens_seen": 160255840,
      "step": 2861
    },
    {
      "epoch": 6.371937639198218,
      "loss": 0.5891226530075073,
      "loss_ce": 0.00025547394761815667,
      "loss_iou": 0.2734375,
      "loss_num": 0.00823974609375,
      "loss_xval": 0.58984375,
      "num_input_tokens_seen": 160255840,
      "step": 2861
    },
    {
      "epoch": 6.374164810690424,
      "grad_norm": 17.660959243774414,
      "learning_rate": 1e-06,
      "loss": 0.6536,
      "num_input_tokens_seen": 160312148,
      "step": 2862
    },
    {
      "epoch": 6.374164810690424,
      "loss": 0.6094207763671875,
      "loss_ce": 0.000167851394508034,
      "loss_iou": 0.263671875,
      "loss_num": 0.0166015625,
      "loss_xval": 0.609375,
      "num_input_tokens_seen": 160312148,
      "step": 2862
    },
    {
      "epoch": 6.3763919821826285,
      "grad_norm": 17.073429107666016,
      "learning_rate": 1e-06,
      "loss": 0.7522,
      "num_input_tokens_seen": 160368156,
      "step": 2863
    },
    {
      "epoch": 6.3763919821826285,
      "loss": 0.7373796105384827,
      "loss_ce": 0.00019699129916261882,
      "loss_iou": 0.33203125,
      "loss_num": 0.01458740234375,
      "loss_xval": 0.73828125,
      "num_input_tokens_seen": 160368156,
      "step": 2863
    },
    {
      "epoch": 6.378619153674833,
      "grad_norm": 92.50753021240234,
      "learning_rate": 1e-06,
      "loss": 0.7973,
      "num_input_tokens_seen": 160426052,
      "step": 2864
    },
    {
      "epoch": 6.378619153674833,
      "loss": 0.7172311544418335,
      "loss_ce": 0.00019013139535672963,
      "loss_iou": 0.318359375,
      "loss_num": 0.01611328125,
      "loss_xval": 0.71875,
      "num_input_tokens_seen": 160426052,
      "step": 2864
    },
    {
      "epoch": 6.380846325167038,
      "grad_norm": 18.557222366333008,
      "learning_rate": 1e-06,
      "loss": 0.6129,
      "num_input_tokens_seen": 160482740,
      "step": 2865
    },
    {
      "epoch": 6.380846325167038,
      "loss": 0.49754250049591064,
      "loss_ce": 0.00022808092762716115,
      "loss_iou": 0.2373046875,
      "loss_num": 0.004638671875,
      "loss_xval": 0.498046875,
      "num_input_tokens_seen": 160482740,
      "step": 2865
    },
    {
      "epoch": 6.383073496659243,
      "grad_norm": 19.247840881347656,
      "learning_rate": 1e-06,
      "loss": 0.8082,
      "num_input_tokens_seen": 160538876,
      "step": 2866
    },
    {
      "epoch": 6.383073496659243,
      "loss": 0.753126859664917,
      "loss_ce": 0.00019718983094207942,
      "loss_iou": 0.33984375,
      "loss_num": 0.01483154296875,
      "loss_xval": 0.75390625,
      "num_input_tokens_seen": 160538876,
      "step": 2866
    },
    {
      "epoch": 6.385300668151448,
      "grad_norm": 19.878429412841797,
      "learning_rate": 1e-06,
      "loss": 0.5645,
      "num_input_tokens_seen": 160592400,
      "step": 2867
    },
    {
      "epoch": 6.385300668151448,
      "loss": 0.44196611642837524,
      "loss_ce": 0.0001936689077410847,
      "loss_iou": 0.189453125,
      "loss_num": 0.0128173828125,
      "loss_xval": 0.44140625,
      "num_input_tokens_seen": 160592400,
      "step": 2867
    },
    {
      "epoch": 6.387527839643653,
      "grad_norm": 23.526199340820312,
      "learning_rate": 1e-06,
      "loss": 0.7953,
      "num_input_tokens_seen": 160648336,
      "step": 2868
    },
    {
      "epoch": 6.387527839643653,
      "loss": 0.9042606353759766,
      "loss_ce": 0.00032996918889693916,
      "loss_iou": 0.39453125,
      "loss_num": 0.02294921875,
      "loss_xval": 0.90234375,
      "num_input_tokens_seen": 160648336,
      "step": 2868
    },
    {
      "epoch": 6.389755011135858,
      "grad_norm": 24.269256591796875,
      "learning_rate": 1e-06,
      "loss": 0.6684,
      "num_input_tokens_seen": 160706796,
      "step": 2869
    },
    {
      "epoch": 6.389755011135858,
      "loss": 0.7116117477416992,
      "loss_ce": 0.00030804408015683293,
      "loss_iou": 0.283203125,
      "loss_num": 0.0286865234375,
      "loss_xval": 0.7109375,
      "num_input_tokens_seen": 160706796,
      "step": 2869
    },
    {
      "epoch": 6.3919821826280625,
      "grad_norm": 31.342178344726562,
      "learning_rate": 1e-06,
      "loss": 0.647,
      "num_input_tokens_seen": 160762660,
      "step": 2870
    },
    {
      "epoch": 6.3919821826280625,
      "loss": 0.4261830449104309,
      "loss_ce": 0.00021868752082809806,
      "loss_iou": 0.189453125,
      "loss_num": 0.00946044921875,
      "loss_xval": 0.42578125,
      "num_input_tokens_seen": 160762660,
      "step": 2870
    },
    {
      "epoch": 6.394209354120267,
      "grad_norm": 16.151552200317383,
      "learning_rate": 1e-06,
      "loss": 0.5474,
      "num_input_tokens_seen": 160818852,
      "step": 2871
    },
    {
      "epoch": 6.394209354120267,
      "loss": 0.5044623017311096,
      "loss_ce": 0.0001898359478218481,
      "loss_iou": 0.2265625,
      "loss_num": 0.01007080078125,
      "loss_xval": 0.50390625,
      "num_input_tokens_seen": 160818852,
      "step": 2871
    },
    {
      "epoch": 6.396436525612472,
      "grad_norm": 15.111863136291504,
      "learning_rate": 1e-06,
      "loss": 0.6786,
      "num_input_tokens_seen": 160875212,
      "step": 2872
    },
    {
      "epoch": 6.396436525612472,
      "loss": 0.6591340899467468,
      "loss_ce": 0.00019854362471960485,
      "loss_iou": 0.28125,
      "loss_num": 0.0189208984375,
      "loss_xval": 0.66015625,
      "num_input_tokens_seen": 160875212,
      "step": 2872
    },
    {
      "epoch": 6.398663697104677,
      "grad_norm": 20.276626586914062,
      "learning_rate": 1e-06,
      "loss": 0.5893,
      "num_input_tokens_seen": 160933880,
      "step": 2873
    },
    {
      "epoch": 6.398663697104677,
      "loss": 0.4171835482120514,
      "loss_ce": 0.00019134554895572364,
      "loss_iou": 0.1865234375,
      "loss_num": 0.00872802734375,
      "loss_xval": 0.41796875,
      "num_input_tokens_seen": 160933880,
      "step": 2873
    },
    {
      "epoch": 6.400890868596882,
      "grad_norm": 33.857666015625,
      "learning_rate": 1e-06,
      "loss": 0.6731,
      "num_input_tokens_seen": 160988996,
      "step": 2874
    },
    {
      "epoch": 6.400890868596882,
      "loss": 0.8410516977310181,
      "loss_ce": 0.00023136789968702942,
      "loss_iou": 0.357421875,
      "loss_num": 0.0252685546875,
      "loss_xval": 0.83984375,
      "num_input_tokens_seen": 160988996,
      "step": 2874
    },
    {
      "epoch": 6.403118040089087,
      "grad_norm": 12.741671562194824,
      "learning_rate": 1e-06,
      "loss": 0.7232,
      "num_input_tokens_seen": 161043496,
      "step": 2875
    },
    {
      "epoch": 6.403118040089087,
      "loss": 0.6495035886764526,
      "loss_ce": 0.00021157064475119114,
      "loss_iou": 0.28125,
      "loss_num": 0.0172119140625,
      "loss_xval": 0.6484375,
      "num_input_tokens_seen": 161043496,
      "step": 2875
    },
    {
      "epoch": 6.405345211581292,
      "grad_norm": 18.550369262695312,
      "learning_rate": 1e-06,
      "loss": 0.6694,
      "num_input_tokens_seen": 161099512,
      "step": 2876
    },
    {
      "epoch": 6.405345211581292,
      "loss": 0.7444691061973572,
      "loss_ce": 0.00032850331626832485,
      "loss_iou": 0.296875,
      "loss_num": 0.030029296875,
      "loss_xval": 0.7421875,
      "num_input_tokens_seen": 161099512,
      "step": 2876
    },
    {
      "epoch": 6.4075723830734965,
      "grad_norm": 23.05417823791504,
      "learning_rate": 1e-06,
      "loss": 0.4981,
      "num_input_tokens_seen": 161155008,
      "step": 2877
    },
    {
      "epoch": 6.4075723830734965,
      "loss": 0.6640802621841431,
      "loss_ce": 0.00020084393327124417,
      "loss_iou": 0.3046875,
      "loss_num": 0.0107421875,
      "loss_xval": 0.6640625,
      "num_input_tokens_seen": 161155008,
      "step": 2877
    },
    {
      "epoch": 6.409799554565701,
      "grad_norm": 16.260465621948242,
      "learning_rate": 1e-06,
      "loss": 0.8791,
      "num_input_tokens_seen": 161211428,
      "step": 2878
    },
    {
      "epoch": 6.409799554565701,
      "loss": 0.9614474773406982,
      "loss_ce": 0.0005100243142805994,
      "loss_iou": 0.40234375,
      "loss_num": 0.031005859375,
      "loss_xval": 0.9609375,
      "num_input_tokens_seen": 161211428,
      "step": 2878
    },
    {
      "epoch": 6.412026726057906,
      "grad_norm": 25.339269638061523,
      "learning_rate": 1e-06,
      "loss": 0.6356,
      "num_input_tokens_seen": 161268840,
      "step": 2879
    },
    {
      "epoch": 6.412026726057906,
      "loss": 0.5295264720916748,
      "loss_ce": 0.00022961897775530815,
      "loss_iou": 0.2353515625,
      "loss_num": 0.01165771484375,
      "loss_xval": 0.53125,
      "num_input_tokens_seen": 161268840,
      "step": 2879
    },
    {
      "epoch": 6.414253897550111,
      "grad_norm": 17.69098663330078,
      "learning_rate": 1e-06,
      "loss": 0.7186,
      "num_input_tokens_seen": 161326592,
      "step": 2880
    },
    {
      "epoch": 6.414253897550111,
      "loss": 0.6872239708900452,
      "loss_ce": 0.00021221938368398696,
      "loss_iou": 0.3046875,
      "loss_num": 0.015380859375,
      "loss_xval": 0.6875,
      "num_input_tokens_seen": 161326592,
      "step": 2880
    },
    {
      "epoch": 6.416481069042316,
      "grad_norm": 20.64066505432129,
      "learning_rate": 1e-06,
      "loss": 0.6329,
      "num_input_tokens_seen": 161379560,
      "step": 2881
    },
    {
      "epoch": 6.416481069042316,
      "loss": 0.8198229074478149,
      "loss_ce": 0.00024278149066958576,
      "loss_iou": 0.36328125,
      "loss_num": 0.0184326171875,
      "loss_xval": 0.8203125,
      "num_input_tokens_seen": 161379560,
      "step": 2881
    },
    {
      "epoch": 6.418708240534521,
      "grad_norm": 16.438508987426758,
      "learning_rate": 1e-06,
      "loss": 0.6852,
      "num_input_tokens_seen": 161436220,
      "step": 2882
    },
    {
      "epoch": 6.418708240534521,
      "loss": 0.7768357992172241,
      "loss_ce": 0.0002244812058052048,
      "loss_iou": 0.298828125,
      "loss_num": 0.0361328125,
      "loss_xval": 0.77734375,
      "num_input_tokens_seen": 161436220,
      "step": 2882
    },
    {
      "epoch": 6.420935412026726,
      "grad_norm": 20.693113327026367,
      "learning_rate": 1e-06,
      "loss": 0.6301,
      "num_input_tokens_seen": 161489760,
      "step": 2883
    },
    {
      "epoch": 6.420935412026726,
      "loss": 0.6213924884796143,
      "loss_ce": 0.00029870617436245084,
      "loss_iou": 0.271484375,
      "loss_num": 0.01531982421875,
      "loss_xval": 0.62109375,
      "num_input_tokens_seen": 161489760,
      "step": 2883
    },
    {
      "epoch": 6.4231625835189305,
      "grad_norm": 89.39657592773438,
      "learning_rate": 1e-06,
      "loss": 0.8208,
      "num_input_tokens_seen": 161546596,
      "step": 2884
    },
    {
      "epoch": 6.4231625835189305,
      "loss": 0.8657987117767334,
      "loss_ce": 0.00032019149512052536,
      "loss_iou": 0.361328125,
      "loss_num": 0.028564453125,
      "loss_xval": 0.8671875,
      "num_input_tokens_seen": 161546596,
      "step": 2884
    },
    {
      "epoch": 6.425389755011135,
      "grad_norm": 19.063907623291016,
      "learning_rate": 1e-06,
      "loss": 0.5801,
      "num_input_tokens_seen": 161600504,
      "step": 2885
    },
    {
      "epoch": 6.425389755011135,
      "loss": 0.5580852627754211,
      "loss_ce": 0.00019339279970154166,
      "loss_iou": 0.2578125,
      "loss_num": 0.00860595703125,
      "loss_xval": 0.55859375,
      "num_input_tokens_seen": 161600504,
      "step": 2885
    },
    {
      "epoch": 6.427616926503341,
      "grad_norm": 21.565977096557617,
      "learning_rate": 1e-06,
      "loss": 0.6976,
      "num_input_tokens_seen": 161659560,
      "step": 2886
    },
    {
      "epoch": 6.427616926503341,
      "loss": 0.8639631271362305,
      "loss_ce": 0.0004377923032734543,
      "loss_iou": 0.400390625,
      "loss_num": 0.01287841796875,
      "loss_xval": 0.86328125,
      "num_input_tokens_seen": 161659560,
      "step": 2886
    },
    {
      "epoch": 6.429844097995546,
      "grad_norm": 17.81540870666504,
      "learning_rate": 1e-06,
      "loss": 0.4801,
      "num_input_tokens_seen": 161717364,
      "step": 2887
    },
    {
      "epoch": 6.429844097995546,
      "loss": 0.49559223651885986,
      "loss_ce": 0.00023093473282642663,
      "loss_iou": 0.2216796875,
      "loss_num": 0.01043701171875,
      "loss_xval": 0.49609375,
      "num_input_tokens_seen": 161717364,
      "step": 2887
    },
    {
      "epoch": 6.432071269487751,
      "grad_norm": 46.8631591796875,
      "learning_rate": 1e-06,
      "loss": 0.7548,
      "num_input_tokens_seen": 161772008,
      "step": 2888
    },
    {
      "epoch": 6.432071269487751,
      "loss": 0.8915398716926575,
      "loss_ce": 0.0004266298783477396,
      "loss_iou": 0.353515625,
      "loss_num": 0.03662109375,
      "loss_xval": 0.890625,
      "num_input_tokens_seen": 161772008,
      "step": 2888
    },
    {
      "epoch": 6.434298440979956,
      "grad_norm": 17.266054153442383,
      "learning_rate": 1e-06,
      "loss": 0.5134,
      "num_input_tokens_seen": 161831844,
      "step": 2889
    },
    {
      "epoch": 6.434298440979956,
      "loss": 0.5959360003471375,
      "loss_ce": 0.00023290744866244495,
      "loss_iou": 0.25,
      "loss_num": 0.0189208984375,
      "loss_xval": 0.59375,
      "num_input_tokens_seen": 161831844,
      "step": 2889
    },
    {
      "epoch": 6.436525612472161,
      "grad_norm": 14.699010848999023,
      "learning_rate": 1e-06,
      "loss": 0.7692,
      "num_input_tokens_seen": 161889280,
      "step": 2890
    },
    {
      "epoch": 6.436525612472161,
      "loss": 0.817835807800293,
      "loss_ce": 0.0004529977450147271,
      "loss_iou": 0.330078125,
      "loss_num": 0.031494140625,
      "loss_xval": 0.81640625,
      "num_input_tokens_seen": 161889280,
      "step": 2890
    },
    {
      "epoch": 6.4387527839643655,
      "grad_norm": 15.680800437927246,
      "learning_rate": 1e-06,
      "loss": 0.6009,
      "num_input_tokens_seen": 161944376,
      "step": 2891
    },
    {
      "epoch": 6.4387527839643655,
      "loss": 0.6331249475479126,
      "loss_ce": 0.0003124309587292373,
      "loss_iou": 0.27734375,
      "loss_num": 0.0159912109375,
      "loss_xval": 0.6328125,
      "num_input_tokens_seen": 161944376,
      "step": 2891
    },
    {
      "epoch": 6.44097995545657,
      "grad_norm": 15.222108840942383,
      "learning_rate": 1e-06,
      "loss": 0.7805,
      "num_input_tokens_seen": 162001860,
      "step": 2892
    },
    {
      "epoch": 6.44097995545657,
      "loss": 0.8359357714653015,
      "loss_ce": 0.00024238668265752494,
      "loss_iou": 0.345703125,
      "loss_num": 0.028564453125,
      "loss_xval": 0.8359375,
      "num_input_tokens_seen": 162001860,
      "step": 2892
    },
    {
      "epoch": 6.443207126948775,
      "grad_norm": 44.775211334228516,
      "learning_rate": 1e-06,
      "loss": 0.7518,
      "num_input_tokens_seen": 162057500,
      "step": 2893
    },
    {
      "epoch": 6.443207126948775,
      "loss": 0.6410810351371765,
      "loss_ce": 0.0007001558551564813,
      "loss_iou": 0.287109375,
      "loss_num": 0.01312255859375,
      "loss_xval": 0.640625,
      "num_input_tokens_seen": 162057500,
      "step": 2893
    },
    {
      "epoch": 6.44543429844098,
      "grad_norm": 16.15131378173828,
      "learning_rate": 1e-06,
      "loss": 0.7321,
      "num_input_tokens_seen": 162112232,
      "step": 2894
    },
    {
      "epoch": 6.44543429844098,
      "loss": 0.7305101156234741,
      "loss_ce": 0.0002855030761566013,
      "loss_iou": 0.32421875,
      "loss_num": 0.0164794921875,
      "loss_xval": 0.73046875,
      "num_input_tokens_seen": 162112232,
      "step": 2894
    },
    {
      "epoch": 6.447661469933185,
      "grad_norm": 62.584171295166016,
      "learning_rate": 1e-06,
      "loss": 0.6187,
      "num_input_tokens_seen": 162166996,
      "step": 2895
    },
    {
      "epoch": 6.447661469933185,
      "loss": 0.5436509847640991,
      "loss_ce": 0.00019392551621422172,
      "loss_iou": 0.2431640625,
      "loss_num": 0.011474609375,
      "loss_xval": 0.54296875,
      "num_input_tokens_seen": 162166996,
      "step": 2895
    },
    {
      "epoch": 6.44988864142539,
      "grad_norm": 25.94312858581543,
      "learning_rate": 1e-06,
      "loss": 0.8496,
      "num_input_tokens_seen": 162221552,
      "step": 2896
    },
    {
      "epoch": 6.44988864142539,
      "loss": 0.8812910914421082,
      "loss_ce": 0.00018758632359094918,
      "loss_iou": 0.37890625,
      "loss_num": 0.0247802734375,
      "loss_xval": 0.8828125,
      "num_input_tokens_seen": 162221552,
      "step": 2896
    },
    {
      "epoch": 6.452115812917595,
      "grad_norm": 17.239500045776367,
      "learning_rate": 1e-06,
      "loss": 0.6339,
      "num_input_tokens_seen": 162278460,
      "step": 2897
    },
    {
      "epoch": 6.452115812917595,
      "loss": 0.5580621957778931,
      "loss_ce": 0.00020086884615011513,
      "loss_iou": 0.220703125,
      "loss_num": 0.0233154296875,
      "loss_xval": 0.55859375,
      "num_input_tokens_seen": 162278460,
      "step": 2897
    },
    {
      "epoch": 6.4543429844097995,
      "grad_norm": 30.1453800201416,
      "learning_rate": 1e-06,
      "loss": 0.6447,
      "num_input_tokens_seen": 162335104,
      "step": 2898
    },
    {
      "epoch": 6.4543429844097995,
      "loss": 0.529179036617279,
      "loss_ce": 0.00037046882789582014,
      "loss_iou": 0.244140625,
      "loss_num": 0.00799560546875,
      "loss_xval": 0.52734375,
      "num_input_tokens_seen": 162335104,
      "step": 2898
    },
    {
      "epoch": 6.456570155902004,
      "grad_norm": 14.278246879577637,
      "learning_rate": 1e-06,
      "loss": 0.8011,
      "num_input_tokens_seen": 162390036,
      "step": 2899
    },
    {
      "epoch": 6.456570155902004,
      "loss": 0.6543991565704346,
      "loss_ce": 0.00022433955746237189,
      "loss_iou": 0.255859375,
      "loss_num": 0.0283203125,
      "loss_xval": 0.65234375,
      "num_input_tokens_seen": 162390036,
      "step": 2899
    },
    {
      "epoch": 6.458797327394209,
      "grad_norm": 27.82345962524414,
      "learning_rate": 1e-06,
      "loss": 0.5877,
      "num_input_tokens_seen": 162446176,
      "step": 2900
    },
    {
      "epoch": 6.458797327394209,
      "loss": 0.5277654528617859,
      "loss_ce": 0.0001775633281795308,
      "loss_iou": 0.2353515625,
      "loss_num": 0.01129150390625,
      "loss_xval": 0.52734375,
      "num_input_tokens_seen": 162446176,
      "step": 2900
    },
    {
      "epoch": 6.461024498886414,
      "grad_norm": 16.007659912109375,
      "learning_rate": 1e-06,
      "loss": 0.7475,
      "num_input_tokens_seen": 162502660,
      "step": 2901
    },
    {
      "epoch": 6.461024498886414,
      "loss": 0.9330655932426453,
      "loss_ce": 0.00020428383140824735,
      "loss_iou": 0.349609375,
      "loss_num": 0.046630859375,
      "loss_xval": 0.93359375,
      "num_input_tokens_seen": 162502660,
      "step": 2901
    },
    {
      "epoch": 6.463251670378619,
      "grad_norm": 21.829668045043945,
      "learning_rate": 1e-06,
      "loss": 0.6332,
      "num_input_tokens_seen": 162559356,
      "step": 2902
    },
    {
      "epoch": 6.463251670378619,
      "loss": 0.623560905456543,
      "loss_ce": 0.0002699050819501281,
      "loss_iou": 0.23828125,
      "loss_num": 0.0294189453125,
      "loss_xval": 0.625,
      "num_input_tokens_seen": 162559356,
      "step": 2902
    },
    {
      "epoch": 6.465478841870824,
      "grad_norm": 16.68356704711914,
      "learning_rate": 1e-06,
      "loss": 0.6161,
      "num_input_tokens_seen": 162617696,
      "step": 2903
    },
    {
      "epoch": 6.465478841870824,
      "loss": 0.5705079436302185,
      "loss_ce": 0.00019540925859473646,
      "loss_iou": 0.240234375,
      "loss_num": 0.01806640625,
      "loss_xval": 0.5703125,
      "num_input_tokens_seen": 162617696,
      "step": 2903
    },
    {
      "epoch": 6.467706013363029,
      "grad_norm": 19.659059524536133,
      "learning_rate": 1e-06,
      "loss": 0.6401,
      "num_input_tokens_seen": 162674516,
      "step": 2904
    },
    {
      "epoch": 6.467706013363029,
      "loss": 0.7844167351722717,
      "loss_ce": 0.00023700644669588655,
      "loss_iou": 0.33203125,
      "loss_num": 0.0238037109375,
      "loss_xval": 0.78515625,
      "num_input_tokens_seen": 162674516,
      "step": 2904
    },
    {
      "epoch": 6.4699331848552335,
      "grad_norm": 21.870929718017578,
      "learning_rate": 1e-06,
      "loss": 0.6373,
      "num_input_tokens_seen": 162731232,
      "step": 2905
    },
    {
      "epoch": 6.4699331848552335,
      "loss": 0.6285732984542847,
      "loss_ce": 0.00039952859515324235,
      "loss_iou": 0.28515625,
      "loss_num": 0.01171875,
      "loss_xval": 0.62890625,
      "num_input_tokens_seen": 162731232,
      "step": 2905
    },
    {
      "epoch": 6.472160356347438,
      "grad_norm": 22.746726989746094,
      "learning_rate": 1e-06,
      "loss": 0.5828,
      "num_input_tokens_seen": 162786624,
      "step": 2906
    },
    {
      "epoch": 6.472160356347438,
      "loss": 0.6897099018096924,
      "loss_ce": 0.0005009524757042527,
      "loss_iou": 0.30078125,
      "loss_num": 0.017822265625,
      "loss_xval": 0.6875,
      "num_input_tokens_seen": 162786624,
      "step": 2906
    },
    {
      "epoch": 6.474387527839644,
      "grad_norm": 22.810585021972656,
      "learning_rate": 1e-06,
      "loss": 0.858,
      "num_input_tokens_seen": 162844052,
      "step": 2907
    },
    {
      "epoch": 6.474387527839644,
      "loss": 0.7649703025817871,
      "loss_ce": 0.00032186286989599466,
      "loss_iou": 0.34375,
      "loss_num": 0.01513671875,
      "loss_xval": 0.765625,
      "num_input_tokens_seen": 162844052,
      "step": 2907
    },
    {
      "epoch": 6.476614699331849,
      "grad_norm": 14.657524108886719,
      "learning_rate": 1e-06,
      "loss": 0.7491,
      "num_input_tokens_seen": 162903604,
      "step": 2908
    },
    {
      "epoch": 6.476614699331849,
      "loss": 0.6694676280021667,
      "loss_ce": 0.00027815584326162934,
      "loss_iou": 0.255859375,
      "loss_num": 0.03173828125,
      "loss_xval": 0.66796875,
      "num_input_tokens_seen": 162903604,
      "step": 2908
    },
    {
      "epoch": 6.478841870824054,
      "grad_norm": 24.23907470703125,
      "learning_rate": 1e-06,
      "loss": 0.6508,
      "num_input_tokens_seen": 162959780,
      "step": 2909
    },
    {
      "epoch": 6.478841870824054,
      "loss": 0.7415847778320312,
      "loss_ce": 0.0001907538971863687,
      "loss_iou": 0.326171875,
      "loss_num": 0.01806640625,
      "loss_xval": 0.7421875,
      "num_input_tokens_seen": 162959780,
      "step": 2909
    },
    {
      "epoch": 6.481069042316259,
      "grad_norm": 20.807043075561523,
      "learning_rate": 1e-06,
      "loss": 0.5773,
      "num_input_tokens_seen": 163016212,
      "step": 2910
    },
    {
      "epoch": 6.481069042316259,
      "loss": 0.7409608364105225,
      "loss_ce": 0.00048227497609332204,
      "loss_iou": 0.3203125,
      "loss_num": 0.0196533203125,
      "loss_xval": 0.7421875,
      "num_input_tokens_seen": 163016212,
      "step": 2910
    },
    {
      "epoch": 6.4832962138084635,
      "grad_norm": 22.38340187072754,
      "learning_rate": 1e-06,
      "loss": 0.69,
      "num_input_tokens_seen": 163071812,
      "step": 2911
    },
    {
      "epoch": 6.4832962138084635,
      "loss": 0.6240810751914978,
      "loss_ce": 0.00030178556335158646,
      "loss_iou": 0.26953125,
      "loss_num": 0.017333984375,
      "loss_xval": 0.625,
      "num_input_tokens_seen": 163071812,
      "step": 2911
    },
    {
      "epoch": 6.485523385300668,
      "grad_norm": 17.558237075805664,
      "learning_rate": 1e-06,
      "loss": 0.942,
      "num_input_tokens_seen": 163127468,
      "step": 2912
    },
    {
      "epoch": 6.485523385300668,
      "loss": 1.0024060010910034,
      "loss_ce": 0.00020868651336058974,
      "loss_iou": 0.443359375,
      "loss_num": 0.02294921875,
      "loss_xval": 1.0,
      "num_input_tokens_seen": 163127468,
      "step": 2912
    },
    {
      "epoch": 6.487750556792873,
      "grad_norm": 20.10137367248535,
      "learning_rate": 1e-06,
      "loss": 0.6397,
      "num_input_tokens_seen": 163182160,
      "step": 2913
    },
    {
      "epoch": 6.487750556792873,
      "loss": 0.37796449661254883,
      "loss_ce": 0.0002789545978885144,
      "loss_iou": 0.173828125,
      "loss_num": 0.00579833984375,
      "loss_xval": 0.376953125,
      "num_input_tokens_seen": 163182160,
      "step": 2913
    },
    {
      "epoch": 6.489977728285078,
      "grad_norm": 59.727386474609375,
      "learning_rate": 1e-06,
      "loss": 0.7155,
      "num_input_tokens_seen": 163238980,
      "step": 2914
    },
    {
      "epoch": 6.489977728285078,
      "loss": 0.6282615661621094,
      "loss_ce": 0.00020979228429496288,
      "loss_iou": 0.271484375,
      "loss_num": 0.01708984375,
      "loss_xval": 0.62890625,
      "num_input_tokens_seen": 163238980,
      "step": 2914
    },
    {
      "epoch": 6.492204899777283,
      "grad_norm": 25.564571380615234,
      "learning_rate": 1e-06,
      "loss": 0.737,
      "num_input_tokens_seen": 163293456,
      "step": 2915
    },
    {
      "epoch": 6.492204899777283,
      "loss": 0.5602612495422363,
      "loss_ce": 0.00020264273916836828,
      "loss_iou": 0.263671875,
      "loss_num": 0.006744384765625,
      "loss_xval": 0.55859375,
      "num_input_tokens_seen": 163293456,
      "step": 2915
    },
    {
      "epoch": 6.494432071269488,
      "grad_norm": 19.56485939025879,
      "learning_rate": 1e-06,
      "loss": 0.6112,
      "num_input_tokens_seen": 163347456,
      "step": 2916
    },
    {
      "epoch": 6.494432071269488,
      "loss": 0.6698689460754395,
      "loss_ce": 0.00028277470846660435,
      "loss_iou": 0.251953125,
      "loss_num": 0.032958984375,
      "loss_xval": 0.66796875,
      "num_input_tokens_seen": 163347456,
      "step": 2916
    },
    {
      "epoch": 6.496659242761693,
      "grad_norm": 16.429033279418945,
      "learning_rate": 1e-06,
      "loss": 0.6937,
      "num_input_tokens_seen": 163405948,
      "step": 2917
    },
    {
      "epoch": 6.496659242761693,
      "loss": 0.42375442385673523,
      "loss_ce": 0.00017044274136424065,
      "loss_iou": 0.17578125,
      "loss_num": 0.014404296875,
      "loss_xval": 0.423828125,
      "num_input_tokens_seen": 163405948,
      "step": 2917
    },
    {
      "epoch": 6.498886414253898,
      "grad_norm": 18.85441780090332,
      "learning_rate": 1e-06,
      "loss": 0.7792,
      "num_input_tokens_seen": 163461656,
      "step": 2918
    },
    {
      "epoch": 6.498886414253898,
      "loss": 0.5861859321594238,
      "loss_ce": 0.00024842872517183423,
      "loss_iou": 0.26171875,
      "loss_num": 0.01263427734375,
      "loss_xval": 0.5859375,
      "num_input_tokens_seen": 163461656,
      "step": 2918
    },
    {
      "epoch": 6.501113585746102,
      "grad_norm": 23.43665885925293,
      "learning_rate": 1e-06,
      "loss": 0.5503,
      "num_input_tokens_seen": 163519632,
      "step": 2919
    },
    {
      "epoch": 6.501113585746102,
      "loss": 0.4104507565498352,
      "loss_ce": 0.00017243438924197108,
      "loss_iou": 0.1845703125,
      "loss_num": 0.0081787109375,
      "loss_xval": 0.41015625,
      "num_input_tokens_seen": 163519632,
      "step": 2919
    },
    {
      "epoch": 6.503340757238307,
      "grad_norm": 28.950546264648438,
      "learning_rate": 1e-06,
      "loss": 0.741,
      "num_input_tokens_seen": 163577312,
      "step": 2920
    },
    {
      "epoch": 6.503340757238307,
      "loss": 0.7677791714668274,
      "loss_ce": 0.0002010307798627764,
      "loss_iou": 0.330078125,
      "loss_num": 0.0216064453125,
      "loss_xval": 0.765625,
      "num_input_tokens_seen": 163577312,
      "step": 2920
    },
    {
      "epoch": 6.505567928730512,
      "grad_norm": 24.427560806274414,
      "learning_rate": 1e-06,
      "loss": 0.5989,
      "num_input_tokens_seen": 163632820,
      "step": 2921
    },
    {
      "epoch": 6.505567928730512,
      "loss": 0.7374780178070068,
      "loss_ce": 0.00017326742818113416,
      "loss_iou": 0.296875,
      "loss_num": 0.0281982421875,
      "loss_xval": 0.73828125,
      "num_input_tokens_seen": 163632820,
      "step": 2921
    },
    {
      "epoch": 6.507795100222717,
      "grad_norm": 18.299259185791016,
      "learning_rate": 1e-06,
      "loss": 0.676,
      "num_input_tokens_seen": 163689372,
      "step": 2922
    },
    {
      "epoch": 6.507795100222717,
      "loss": 0.6376951932907104,
      "loss_ce": 0.00024403775751125067,
      "loss_iou": 0.298828125,
      "loss_num": 0.00830078125,
      "loss_xval": 0.63671875,
      "num_input_tokens_seen": 163689372,
      "step": 2922
    },
    {
      "epoch": 6.510022271714922,
      "grad_norm": 31.762420654296875,
      "learning_rate": 1e-06,
      "loss": 0.8069,
      "num_input_tokens_seen": 163745852,
      "step": 2923
    },
    {
      "epoch": 6.510022271714922,
      "loss": 1.0176315307617188,
      "loss_ce": 0.00029755313880741596,
      "loss_iou": 0.4140625,
      "loss_num": 0.037353515625,
      "loss_xval": 1.015625,
      "num_input_tokens_seen": 163745852,
      "step": 2923
    },
    {
      "epoch": 6.512249443207127,
      "grad_norm": 25.43143653869629,
      "learning_rate": 1e-06,
      "loss": 1.0766,
      "num_input_tokens_seen": 163801732,
      "step": 2924
    },
    {
      "epoch": 6.512249443207127,
      "loss": 1.3304567337036133,
      "loss_ce": 0.00025651464238762856,
      "loss_iou": 0.53125,
      "loss_num": 0.0537109375,
      "loss_xval": 1.328125,
      "num_input_tokens_seen": 163801732,
      "step": 2924
    },
    {
      "epoch": 6.514476614699332,
      "grad_norm": 15.741415977478027,
      "learning_rate": 1e-06,
      "loss": 0.6474,
      "num_input_tokens_seen": 163859180,
      "step": 2925
    },
    {
      "epoch": 6.514476614699332,
      "loss": 0.6122488975524902,
      "loss_ce": 0.00018836073286365718,
      "loss_iou": 0.279296875,
      "loss_num": 0.0106201171875,
      "loss_xval": 0.61328125,
      "num_input_tokens_seen": 163859180,
      "step": 2925
    },
    {
      "epoch": 6.5167037861915365,
      "grad_norm": 20.761812210083008,
      "learning_rate": 1e-06,
      "loss": 0.8235,
      "num_input_tokens_seen": 163916244,
      "step": 2926
    },
    {
      "epoch": 6.5167037861915365,
      "loss": 0.5054827332496643,
      "loss_ce": 0.00023371227143798023,
      "loss_iou": 0.2216796875,
      "loss_num": 0.01226806640625,
      "loss_xval": 0.50390625,
      "num_input_tokens_seen": 163916244,
      "step": 2926
    },
    {
      "epoch": 6.518930957683741,
      "grad_norm": 61.213077545166016,
      "learning_rate": 1e-06,
      "loss": 0.9006,
      "num_input_tokens_seen": 163970816,
      "step": 2927
    },
    {
      "epoch": 6.518930957683741,
      "loss": 0.9136002659797668,
      "loss_ce": 0.0003922681207768619,
      "loss_iou": 0.3984375,
      "loss_num": 0.023681640625,
      "loss_xval": 0.9140625,
      "num_input_tokens_seen": 163970816,
      "step": 2927
    },
    {
      "epoch": 6.521158129175946,
      "grad_norm": 21.650136947631836,
      "learning_rate": 1e-06,
      "loss": 0.7797,
      "num_input_tokens_seen": 164028208,
      "step": 2928
    },
    {
      "epoch": 6.521158129175946,
      "loss": 0.8101509213447571,
      "loss_ce": 0.00021437015675473958,
      "loss_iou": 0.337890625,
      "loss_num": 0.0269775390625,
      "loss_xval": 0.80859375,
      "num_input_tokens_seen": 164028208,
      "step": 2928
    },
    {
      "epoch": 6.523385300668151,
      "grad_norm": 20.427627563476562,
      "learning_rate": 1e-06,
      "loss": 0.8737,
      "num_input_tokens_seen": 164083608,
      "step": 2929
    },
    {
      "epoch": 6.523385300668151,
      "loss": 0.9638548493385315,
      "loss_ce": 0.00023182647419162095,
      "loss_iou": 0.43359375,
      "loss_num": 0.01953125,
      "loss_xval": 0.96484375,
      "num_input_tokens_seen": 164083608,
      "step": 2929
    },
    {
      "epoch": 6.525612472160356,
      "grad_norm": 17.497631072998047,
      "learning_rate": 1e-06,
      "loss": 0.7885,
      "num_input_tokens_seen": 164138104,
      "step": 2930
    },
    {
      "epoch": 6.525612472160356,
      "loss": 0.9440581798553467,
      "loss_ce": 0.00021053646923974156,
      "loss_iou": 0.3515625,
      "loss_num": 0.04833984375,
      "loss_xval": 0.9453125,
      "num_input_tokens_seen": 164138104,
      "step": 2930
    },
    {
      "epoch": 6.527839643652561,
      "grad_norm": 26.011911392211914,
      "learning_rate": 1e-06,
      "loss": 0.7072,
      "num_input_tokens_seen": 164194992,
      "step": 2931
    },
    {
      "epoch": 6.527839643652561,
      "loss": 0.5633350610733032,
      "loss_ce": 0.00022474557044915855,
      "loss_iou": 0.251953125,
      "loss_num": 0.0118408203125,
      "loss_xval": 0.5625,
      "num_input_tokens_seen": 164194992,
      "step": 2931
    },
    {
      "epoch": 6.5300668151447665,
      "grad_norm": 14.881753921508789,
      "learning_rate": 1e-06,
      "loss": 0.6276,
      "num_input_tokens_seen": 164252436,
      "step": 2932
    },
    {
      "epoch": 6.5300668151447665,
      "loss": 0.7083165645599365,
      "loss_ce": 0.0005529068876057863,
      "loss_iou": 0.2890625,
      "loss_num": 0.02587890625,
      "loss_xval": 0.70703125,
      "num_input_tokens_seen": 164252436,
      "step": 2932
    },
    {
      "epoch": 6.532293986636971,
      "grad_norm": 22.067697525024414,
      "learning_rate": 1e-06,
      "loss": 0.6472,
      "num_input_tokens_seen": 164309220,
      "step": 2933
    },
    {
      "epoch": 6.532293986636971,
      "loss": 0.6984990835189819,
      "loss_ce": 0.0002568979107309133,
      "loss_iou": 0.3125,
      "loss_num": 0.01416015625,
      "loss_xval": 0.69921875,
      "num_input_tokens_seen": 164309220,
      "step": 2933
    },
    {
      "epoch": 6.534521158129176,
      "grad_norm": 18.131603240966797,
      "learning_rate": 1e-06,
      "loss": 0.6463,
      "num_input_tokens_seen": 164366052,
      "step": 2934
    },
    {
      "epoch": 6.534521158129176,
      "loss": 0.6132932901382446,
      "loss_ce": 0.00025620122323744,
      "loss_iou": 0.228515625,
      "loss_num": 0.031005859375,
      "loss_xval": 0.61328125,
      "num_input_tokens_seen": 164366052,
      "step": 2934
    },
    {
      "epoch": 6.536748329621381,
      "grad_norm": 28.302825927734375,
      "learning_rate": 1e-06,
      "loss": 0.5853,
      "num_input_tokens_seen": 164417760,
      "step": 2935
    },
    {
      "epoch": 6.536748329621381,
      "loss": 0.7578645944595337,
      "loss_ce": 0.0002961784484796226,
      "loss_iou": 0.318359375,
      "loss_num": 0.024169921875,
      "loss_xval": 0.7578125,
      "num_input_tokens_seen": 164417760,
      "step": 2935
    },
    {
      "epoch": 6.538975501113586,
      "grad_norm": 18.224735260009766,
      "learning_rate": 1e-06,
      "loss": 0.6522,
      "num_input_tokens_seen": 164472700,
      "step": 2936
    },
    {
      "epoch": 6.538975501113586,
      "loss": 0.6816818118095398,
      "loss_ce": 0.000285336805973202,
      "loss_iou": 0.29296875,
      "loss_num": 0.018798828125,
      "loss_xval": 0.6796875,
      "num_input_tokens_seen": 164472700,
      "step": 2936
    },
    {
      "epoch": 6.541202672605791,
      "grad_norm": 21.562236785888672,
      "learning_rate": 1e-06,
      "loss": 0.6994,
      "num_input_tokens_seen": 164528444,
      "step": 2937
    },
    {
      "epoch": 6.541202672605791,
      "loss": 0.641094446182251,
      "loss_ce": 0.00022529340640176088,
      "loss_iou": 0.296875,
      "loss_num": 0.00982666015625,
      "loss_xval": 0.640625,
      "num_input_tokens_seen": 164528444,
      "step": 2937
    },
    {
      "epoch": 6.543429844097996,
      "grad_norm": 44.151615142822266,
      "learning_rate": 1e-06,
      "loss": 0.7169,
      "num_input_tokens_seen": 164583964,
      "step": 2938
    },
    {
      "epoch": 6.543429844097996,
      "loss": 0.6977666616439819,
      "loss_ce": 0.00025689357426017523,
      "loss_iou": 0.31640625,
      "loss_num": 0.01287841796875,
      "loss_xval": 0.69921875,
      "num_input_tokens_seen": 164583964,
      "step": 2938
    },
    {
      "epoch": 6.5456570155902005,
      "grad_norm": 27.166345596313477,
      "learning_rate": 1e-06,
      "loss": 0.6231,
      "num_input_tokens_seen": 164638820,
      "step": 2939
    },
    {
      "epoch": 6.5456570155902005,
      "loss": 0.6275501251220703,
      "loss_ce": 0.0002307354734512046,
      "loss_iou": 0.251953125,
      "loss_num": 0.0242919921875,
      "loss_xval": 0.62890625,
      "num_input_tokens_seen": 164638820,
      "step": 2939
    },
    {
      "epoch": 6.547884187082405,
      "grad_norm": 15.208316802978516,
      "learning_rate": 1e-06,
      "loss": 0.4751,
      "num_input_tokens_seen": 164693132,
      "step": 2940
    },
    {
      "epoch": 6.547884187082405,
      "loss": 0.3183910846710205,
      "loss_ce": 0.00018428399926051497,
      "loss_iou": 0.1455078125,
      "loss_num": 0.005584716796875,
      "loss_xval": 0.318359375,
      "num_input_tokens_seen": 164693132,
      "step": 2940
    },
    {
      "epoch": 6.55011135857461,
      "grad_norm": 32.17951965332031,
      "learning_rate": 1e-06,
      "loss": 0.7,
      "num_input_tokens_seen": 164749436,
      "step": 2941
    },
    {
      "epoch": 6.55011135857461,
      "loss": 0.7634174823760986,
      "loss_ce": 0.0004780220042448491,
      "loss_iou": 0.33203125,
      "loss_num": 0.02001953125,
      "loss_xval": 0.76171875,
      "num_input_tokens_seen": 164749436,
      "step": 2941
    },
    {
      "epoch": 6.552338530066815,
      "grad_norm": 21.82436752319336,
      "learning_rate": 1e-06,
      "loss": 0.6542,
      "num_input_tokens_seen": 164805000,
      "step": 2942
    },
    {
      "epoch": 6.552338530066815,
      "loss": 0.6028134822845459,
      "loss_ce": 0.0002744359662756324,
      "loss_iou": 0.25390625,
      "loss_num": 0.018798828125,
      "loss_xval": 0.6015625,
      "num_input_tokens_seen": 164805000,
      "step": 2942
    },
    {
      "epoch": 6.55456570155902,
      "grad_norm": 20.991344451904297,
      "learning_rate": 1e-06,
      "loss": 0.7364,
      "num_input_tokens_seen": 164861344,
      "step": 2943
    },
    {
      "epoch": 6.55456570155902,
      "loss": 0.7187168002128601,
      "loss_ce": 0.00021093602117616683,
      "loss_iou": 0.26953125,
      "loss_num": 0.03564453125,
      "loss_xval": 0.71875,
      "num_input_tokens_seen": 164861344,
      "step": 2943
    },
    {
      "epoch": 6.556792873051225,
      "grad_norm": 21.933055877685547,
      "learning_rate": 1e-06,
      "loss": 0.4837,
      "num_input_tokens_seen": 164919076,
      "step": 2944
    },
    {
      "epoch": 6.556792873051225,
      "loss": 0.5052962899208069,
      "loss_ce": 0.00016931179561652243,
      "loss_iou": 0.2099609375,
      "loss_num": 0.0167236328125,
      "loss_xval": 0.50390625,
      "num_input_tokens_seen": 164919076,
      "step": 2944
    },
    {
      "epoch": 6.55902004454343,
      "grad_norm": 19.509490966796875,
      "learning_rate": 1e-06,
      "loss": 0.9163,
      "num_input_tokens_seen": 164972212,
      "step": 2945
    },
    {
      "epoch": 6.55902004454343,
      "loss": 0.810613751411438,
      "loss_ce": 0.00025000711320899427,
      "loss_iou": 0.33984375,
      "loss_num": 0.026123046875,
      "loss_xval": 0.80859375,
      "num_input_tokens_seen": 164972212,
      "step": 2945
    },
    {
      "epoch": 6.5612472160356345,
      "grad_norm": 33.0357780456543,
      "learning_rate": 1e-06,
      "loss": 0.6847,
      "num_input_tokens_seen": 165028004,
      "step": 2946
    },
    {
      "epoch": 6.5612472160356345,
      "loss": 0.7473153471946716,
      "loss_ce": 0.0002450351894367486,
      "loss_iou": 0.31640625,
      "loss_num": 0.0223388671875,
      "loss_xval": 0.74609375,
      "num_input_tokens_seen": 165028004,
      "step": 2946
    },
    {
      "epoch": 6.563474387527839,
      "grad_norm": 19.9017391204834,
      "learning_rate": 1e-06,
      "loss": 0.6142,
      "num_input_tokens_seen": 165083812,
      "step": 2947
    },
    {
      "epoch": 6.563474387527839,
      "loss": 0.6369014382362366,
      "loss_ce": 0.0001826911757234484,
      "loss_iou": 0.28515625,
      "loss_num": 0.0130615234375,
      "loss_xval": 0.63671875,
      "num_input_tokens_seen": 165083812,
      "step": 2947
    },
    {
      "epoch": 6.565701559020044,
      "grad_norm": 30.455806732177734,
      "learning_rate": 1e-06,
      "loss": 0.601,
      "num_input_tokens_seen": 165139816,
      "step": 2948
    },
    {
      "epoch": 6.565701559020044,
      "loss": 0.6039369106292725,
      "loss_ce": 0.00023821514332666993,
      "loss_iou": 0.2333984375,
      "loss_num": 0.027587890625,
      "loss_xval": 0.60546875,
      "num_input_tokens_seen": 165139816,
      "step": 2948
    },
    {
      "epoch": 6.567928730512249,
      "grad_norm": 20.041481018066406,
      "learning_rate": 1e-06,
      "loss": 0.62,
      "num_input_tokens_seen": 165195512,
      "step": 2949
    },
    {
      "epoch": 6.567928730512249,
      "loss": 0.6124828457832336,
      "loss_ce": 0.00017815582395996898,
      "loss_iou": 0.2490234375,
      "loss_num": 0.0228271484375,
      "loss_xval": 0.61328125,
      "num_input_tokens_seen": 165195512,
      "step": 2949
    },
    {
      "epoch": 6.570155902004454,
      "grad_norm": 17.582809448242188,
      "learning_rate": 1e-06,
      "loss": 0.5694,
      "num_input_tokens_seen": 165251664,
      "step": 2950
    },
    {
      "epoch": 6.570155902004454,
      "loss": 0.6081294417381287,
      "loss_ce": 0.00021927471971139312,
      "loss_iou": 0.263671875,
      "loss_num": 0.015869140625,
      "loss_xval": 0.609375,
      "num_input_tokens_seen": 165251664,
      "step": 2950
    },
    {
      "epoch": 6.57238307349666,
      "grad_norm": 18.174318313598633,
      "learning_rate": 1e-06,
      "loss": 0.4721,
      "num_input_tokens_seen": 165307188,
      "step": 2951
    },
    {
      "epoch": 6.57238307349666,
      "loss": 0.49299192428588867,
      "loss_ce": 0.00016358881839551032,
      "loss_iou": 0.2099609375,
      "loss_num": 0.01434326171875,
      "loss_xval": 0.4921875,
      "num_input_tokens_seen": 165307188,
      "step": 2951
    },
    {
      "epoch": 6.574610244988865,
      "grad_norm": 21.09537696838379,
      "learning_rate": 1e-06,
      "loss": 0.6628,
      "num_input_tokens_seen": 165362820,
      "step": 2952
    },
    {
      "epoch": 6.574610244988865,
      "loss": 0.7518377304077148,
      "loss_ce": 0.0002508063626009971,
      "loss_iou": 0.326171875,
      "loss_num": 0.02001953125,
      "loss_xval": 0.75,
      "num_input_tokens_seen": 165362820,
      "step": 2952
    },
    {
      "epoch": 6.5768374164810695,
      "grad_norm": 30.387908935546875,
      "learning_rate": 1e-06,
      "loss": 0.5611,
      "num_input_tokens_seen": 165419408,
      "step": 2953
    },
    {
      "epoch": 6.5768374164810695,
      "loss": 0.5768758058547974,
      "loss_ce": 0.00021561238099820912,
      "loss_iou": 0.2578125,
      "loss_num": 0.01251220703125,
      "loss_xval": 0.578125,
      "num_input_tokens_seen": 165419408,
      "step": 2953
    },
    {
      "epoch": 6.579064587973274,
      "grad_norm": 20.413816452026367,
      "learning_rate": 1e-06,
      "loss": 0.7898,
      "num_input_tokens_seen": 165472128,
      "step": 2954
    },
    {
      "epoch": 6.579064587973274,
      "loss": 0.7371137142181396,
      "loss_ce": 0.0001751929521560669,
      "loss_iou": 0.287109375,
      "loss_num": 0.03271484375,
      "loss_xval": 0.73828125,
      "num_input_tokens_seen": 165472128,
      "step": 2954
    },
    {
      "epoch": 6.581291759465479,
      "grad_norm": 20.204029083251953,
      "learning_rate": 1e-06,
      "loss": 0.5606,
      "num_input_tokens_seen": 165529144,
      "step": 2955
    },
    {
      "epoch": 6.581291759465479,
      "loss": 0.6010243892669678,
      "loss_ce": 0.00019429507665336132,
      "loss_iou": 0.28125,
      "loss_num": 0.00787353515625,
      "loss_xval": 0.6015625,
      "num_input_tokens_seen": 165529144,
      "step": 2955
    },
    {
      "epoch": 6.583518930957684,
      "grad_norm": 17.47038459777832,
      "learning_rate": 1e-06,
      "loss": 0.6566,
      "num_input_tokens_seen": 165588264,
      "step": 2956
    },
    {
      "epoch": 6.583518930957684,
      "loss": 0.7859352827072144,
      "loss_ce": 0.0002907089365180582,
      "loss_iou": 0.333984375,
      "loss_num": 0.023681640625,
      "loss_xval": 0.78515625,
      "num_input_tokens_seen": 165588264,
      "step": 2956
    },
    {
      "epoch": 6.585746102449889,
      "grad_norm": 16.517253875732422,
      "learning_rate": 1e-06,
      "loss": 0.547,
      "num_input_tokens_seen": 165644784,
      "step": 2957
    },
    {
      "epoch": 6.585746102449889,
      "loss": 0.6574386358261108,
      "loss_ce": 0.00021205886150710285,
      "loss_iou": 0.287109375,
      "loss_num": 0.016357421875,
      "loss_xval": 0.65625,
      "num_input_tokens_seen": 165644784,
      "step": 2957
    },
    {
      "epoch": 6.587973273942094,
      "grad_norm": 18.46898078918457,
      "learning_rate": 1e-06,
      "loss": 0.6594,
      "num_input_tokens_seen": 165701372,
      "step": 2958
    },
    {
      "epoch": 6.587973273942094,
      "loss": 0.38792580366134644,
      "loss_ce": 0.0003525797219481319,
      "loss_iou": 0.162109375,
      "loss_num": 0.01251220703125,
      "loss_xval": 0.38671875,
      "num_input_tokens_seen": 165701372,
      "step": 2958
    },
    {
      "epoch": 6.590200445434299,
      "grad_norm": 28.243539810180664,
      "learning_rate": 1e-06,
      "loss": 0.7292,
      "num_input_tokens_seen": 165756960,
      "step": 2959
    },
    {
      "epoch": 6.590200445434299,
      "loss": 0.9738781452178955,
      "loss_ce": 0.0002453392662573606,
      "loss_iou": 0.42578125,
      "loss_num": 0.0244140625,
      "loss_xval": 0.97265625,
      "num_input_tokens_seen": 165756960,
      "step": 2959
    },
    {
      "epoch": 6.5924276169265035,
      "grad_norm": 15.872271537780762,
      "learning_rate": 1e-06,
      "loss": 0.5439,
      "num_input_tokens_seen": 165813860,
      "step": 2960
    },
    {
      "epoch": 6.5924276169265035,
      "loss": 0.624319314956665,
      "loss_ce": 0.00029592958162538707,
      "loss_iou": 0.26171875,
      "loss_num": 0.0201416015625,
      "loss_xval": 0.625,
      "num_input_tokens_seen": 165813860,
      "step": 2960
    },
    {
      "epoch": 6.594654788418708,
      "grad_norm": 109.15684509277344,
      "learning_rate": 1e-06,
      "loss": 0.7413,
      "num_input_tokens_seen": 165871416,
      "step": 2961
    },
    {
      "epoch": 6.594654788418708,
      "loss": 0.9980896711349487,
      "loss_ce": 0.00016483690706081688,
      "loss_iou": 0.44140625,
      "loss_num": 0.023193359375,
      "loss_xval": 0.99609375,
      "num_input_tokens_seen": 165871416,
      "step": 2961
    },
    {
      "epoch": 6.596881959910913,
      "grad_norm": 16.21949577331543,
      "learning_rate": 1e-06,
      "loss": 0.9368,
      "num_input_tokens_seen": 165925904,
      "step": 2962
    },
    {
      "epoch": 6.596881959910913,
      "loss": 0.8550323247909546,
      "loss_ce": 0.00017394759925082326,
      "loss_iou": 0.33984375,
      "loss_num": 0.034912109375,
      "loss_xval": 0.85546875,
      "num_input_tokens_seen": 165925904,
      "step": 2962
    },
    {
      "epoch": 6.599109131403118,
      "grad_norm": 31.012943267822266,
      "learning_rate": 1e-06,
      "loss": 0.6417,
      "num_input_tokens_seen": 165983760,
      "step": 2963
    },
    {
      "epoch": 6.599109131403118,
      "loss": 0.6024689674377441,
      "loss_ce": 0.000662269361782819,
      "loss_iou": 0.25390625,
      "loss_num": 0.018798828125,
      "loss_xval": 0.6015625,
      "num_input_tokens_seen": 165983760,
      "step": 2963
    },
    {
      "epoch": 6.601336302895323,
      "grad_norm": 20.181808471679688,
      "learning_rate": 1e-06,
      "loss": 0.6337,
      "num_input_tokens_seen": 166040296,
      "step": 2964
    },
    {
      "epoch": 6.601336302895323,
      "loss": 0.724341094493866,
      "loss_ce": 0.0003420562425162643,
      "loss_iou": 0.298828125,
      "loss_num": 0.0252685546875,
      "loss_xval": 0.72265625,
      "num_input_tokens_seen": 166040296,
      "step": 2964
    },
    {
      "epoch": 6.603563474387528,
      "grad_norm": 28.568126678466797,
      "learning_rate": 1e-06,
      "loss": 0.7788,
      "num_input_tokens_seen": 166092104,
      "step": 2965
    },
    {
      "epoch": 6.603563474387528,
      "loss": 0.7236001491546631,
      "loss_ce": 0.00021149440726730973,
      "loss_iou": 0.275390625,
      "loss_num": 0.034912109375,
      "loss_xval": 0.72265625,
      "num_input_tokens_seen": 166092104,
      "step": 2965
    },
    {
      "epoch": 6.605790645879733,
      "grad_norm": 14.282336235046387,
      "learning_rate": 1e-06,
      "loss": 0.7572,
      "num_input_tokens_seen": 166147648,
      "step": 2966
    },
    {
      "epoch": 6.605790645879733,
      "loss": 0.9260650873184204,
      "loss_ce": 0.0002839115622919053,
      "loss_iou": 0.365234375,
      "loss_num": 0.0390625,
      "loss_xval": 0.92578125,
      "num_input_tokens_seen": 166147648,
      "step": 2966
    },
    {
      "epoch": 6.6080178173719375,
      "grad_norm": 18.43797492980957,
      "learning_rate": 1e-06,
      "loss": 0.8469,
      "num_input_tokens_seen": 166205440,
      "step": 2967
    },
    {
      "epoch": 6.6080178173719375,
      "loss": 1.090362548828125,
      "loss_ce": 0.0002746962709352374,
      "loss_iou": 0.4453125,
      "loss_num": 0.03955078125,
      "loss_xval": 1.09375,
      "num_input_tokens_seen": 166205440,
      "step": 2967
    },
    {
      "epoch": 6.610244988864142,
      "grad_norm": 20.853721618652344,
      "learning_rate": 1e-06,
      "loss": 0.7179,
      "num_input_tokens_seen": 166262912,
      "step": 2968
    },
    {
      "epoch": 6.610244988864142,
      "loss": 0.7870877981185913,
      "loss_ce": 0.00022257049567997456,
      "loss_iou": 0.34375,
      "loss_num": 0.020263671875,
      "loss_xval": 0.78515625,
      "num_input_tokens_seen": 166262912,
      "step": 2968
    },
    {
      "epoch": 6.612472160356347,
      "grad_norm": 119.40055847167969,
      "learning_rate": 1e-06,
      "loss": 0.6213,
      "num_input_tokens_seen": 166320972,
      "step": 2969
    },
    {
      "epoch": 6.612472160356347,
      "loss": 0.6227630376815796,
      "loss_ce": 0.00020442584354896098,
      "loss_iou": 0.275390625,
      "loss_num": 0.01434326171875,
      "loss_xval": 0.62109375,
      "num_input_tokens_seen": 166320972,
      "step": 2969
    },
    {
      "epoch": 6.614699331848552,
      "grad_norm": 21.16520881652832,
      "learning_rate": 1e-06,
      "loss": 0.7879,
      "num_input_tokens_seen": 166377852,
      "step": 2970
    },
    {
      "epoch": 6.614699331848552,
      "loss": 0.6298392415046692,
      "loss_ce": 0.0003226206754334271,
      "loss_iou": 0.27734375,
      "loss_num": 0.01470947265625,
      "loss_xval": 0.62890625,
      "num_input_tokens_seen": 166377852,
      "step": 2970
    },
    {
      "epoch": 6.616926503340757,
      "grad_norm": 24.545520782470703,
      "learning_rate": 1e-06,
      "loss": 0.761,
      "num_input_tokens_seen": 166434064,
      "step": 2971
    },
    {
      "epoch": 6.616926503340757,
      "loss": 0.798297643661499,
      "loss_ce": 0.00020199231221340597,
      "loss_iou": 0.345703125,
      "loss_num": 0.021240234375,
      "loss_xval": 0.796875,
      "num_input_tokens_seen": 166434064,
      "step": 2971
    },
    {
      "epoch": 6.619153674832962,
      "grad_norm": 14.242857933044434,
      "learning_rate": 1e-06,
      "loss": 0.7568,
      "num_input_tokens_seen": 166486632,
      "step": 2972
    },
    {
      "epoch": 6.619153674832962,
      "loss": 0.8938003778457642,
      "loss_ce": 0.00024563493207097054,
      "loss_iou": 0.35546875,
      "loss_num": 0.036376953125,
      "loss_xval": 0.89453125,
      "num_input_tokens_seen": 166486632,
      "step": 2972
    },
    {
      "epoch": 6.621380846325167,
      "grad_norm": 13.78774642944336,
      "learning_rate": 1e-06,
      "loss": 0.5507,
      "num_input_tokens_seen": 166543024,
      "step": 2973
    },
    {
      "epoch": 6.621380846325167,
      "loss": 0.5280297994613647,
      "loss_ce": 0.00019778512069024146,
      "loss_iou": 0.2314453125,
      "loss_num": 0.01312255859375,
      "loss_xval": 0.52734375,
      "num_input_tokens_seen": 166543024,
      "step": 2973
    },
    {
      "epoch": 6.6236080178173715,
      "grad_norm": 16.257577896118164,
      "learning_rate": 1e-06,
      "loss": 0.5946,
      "num_input_tokens_seen": 166601376,
      "step": 2974
    },
    {
      "epoch": 6.6236080178173715,
      "loss": 0.705248236656189,
      "loss_ce": 0.00017008130089379847,
      "loss_iou": 0.30859375,
      "loss_num": 0.0179443359375,
      "loss_xval": 0.703125,
      "num_input_tokens_seen": 166601376,
      "step": 2974
    },
    {
      "epoch": 6.625835189309576,
      "grad_norm": 19.095983505249023,
      "learning_rate": 1e-06,
      "loss": 0.7031,
      "num_input_tokens_seen": 166654948,
      "step": 2975
    },
    {
      "epoch": 6.625835189309576,
      "loss": 0.48734956979751587,
      "loss_ce": 0.00016694515943527222,
      "loss_iou": 0.201171875,
      "loss_num": 0.0169677734375,
      "loss_xval": 0.486328125,
      "num_input_tokens_seen": 166654948,
      "step": 2975
    },
    {
      "epoch": 6.628062360801781,
      "grad_norm": 21.924911499023438,
      "learning_rate": 1e-06,
      "loss": 0.6498,
      "num_input_tokens_seen": 166711656,
      "step": 2976
    },
    {
      "epoch": 6.628062360801781,
      "loss": 0.5872310996055603,
      "loss_ce": 0.0001949925208464265,
      "loss_iou": 0.25,
      "loss_num": 0.0174560546875,
      "loss_xval": 0.5859375,
      "num_input_tokens_seen": 166711656,
      "step": 2976
    },
    {
      "epoch": 6.630289532293987,
      "grad_norm": 20.422719955444336,
      "learning_rate": 1e-06,
      "loss": 0.736,
      "num_input_tokens_seen": 166770256,
      "step": 2977
    },
    {
      "epoch": 6.630289532293987,
      "loss": 0.9298211932182312,
      "loss_ce": 0.0002557823609095067,
      "loss_iou": 0.3515625,
      "loss_num": 0.045654296875,
      "loss_xval": 0.9296875,
      "num_input_tokens_seen": 166770256,
      "step": 2977
    },
    {
      "epoch": 6.632516703786192,
      "grad_norm": 18.542539596557617,
      "learning_rate": 1e-06,
      "loss": 0.6461,
      "num_input_tokens_seen": 166826808,
      "step": 2978
    },
    {
      "epoch": 6.632516703786192,
      "loss": 0.5421923398971558,
      "loss_ce": 0.00020013400353491306,
      "loss_iou": 0.22265625,
      "loss_num": 0.0191650390625,
      "loss_xval": 0.54296875,
      "num_input_tokens_seen": 166826808,
      "step": 2978
    },
    {
      "epoch": 6.634743875278397,
      "grad_norm": 21.711483001708984,
      "learning_rate": 1e-06,
      "loss": 0.833,
      "num_input_tokens_seen": 166881468,
      "step": 2979
    },
    {
      "epoch": 6.634743875278397,
      "loss": 0.8562052249908447,
      "loss_ce": 0.000248211930738762,
      "loss_iou": 0.33984375,
      "loss_num": 0.034912109375,
      "loss_xval": 0.85546875,
      "num_input_tokens_seen": 166881468,
      "step": 2979
    },
    {
      "epoch": 6.636971046770602,
      "grad_norm": 19.191953659057617,
      "learning_rate": 1e-06,
      "loss": 0.5379,
      "num_input_tokens_seen": 166938348,
      "step": 2980
    },
    {
      "epoch": 6.636971046770602,
      "loss": 0.47580990195274353,
      "loss_ce": 0.0002239710884168744,
      "loss_iou": 0.216796875,
      "loss_num": 0.00836181640625,
      "loss_xval": 0.4765625,
      "num_input_tokens_seen": 166938348,
      "step": 2980
    },
    {
      "epoch": 6.639198218262806,
      "grad_norm": 19.030431747436523,
      "learning_rate": 1e-06,
      "loss": 0.5368,
      "num_input_tokens_seen": 166994888,
      "step": 2981
    },
    {
      "epoch": 6.639198218262806,
      "loss": 0.5439237952232361,
      "loss_ce": 0.00022264779545366764,
      "loss_iou": 0.220703125,
      "loss_num": 0.0205078125,
      "loss_xval": 0.54296875,
      "num_input_tokens_seen": 166994888,
      "step": 2981
    },
    {
      "epoch": 6.641425389755011,
      "grad_norm": 17.75498390197754,
      "learning_rate": 1e-06,
      "loss": 0.5462,
      "num_input_tokens_seen": 167050384,
      "step": 2982
    },
    {
      "epoch": 6.641425389755011,
      "loss": 0.4909123182296753,
      "loss_ce": 0.00018965097842738032,
      "loss_iou": 0.2158203125,
      "loss_num": 0.0118408203125,
      "loss_xval": 0.490234375,
      "num_input_tokens_seen": 167050384,
      "step": 2982
    },
    {
      "epoch": 6.643652561247216,
      "grad_norm": 15.332694053649902,
      "learning_rate": 1e-06,
      "loss": 0.4491,
      "num_input_tokens_seen": 167108932,
      "step": 2983
    },
    {
      "epoch": 6.643652561247216,
      "loss": 0.5531620979309082,
      "loss_ce": 0.00018356410146225244,
      "loss_iou": 0.2490234375,
      "loss_num": 0.01116943359375,
      "loss_xval": 0.5546875,
      "num_input_tokens_seen": 167108932,
      "step": 2983
    },
    {
      "epoch": 6.645879732739421,
      "grad_norm": 26.91922378540039,
      "learning_rate": 1e-06,
      "loss": 0.6703,
      "num_input_tokens_seen": 167167308,
      "step": 2984
    },
    {
      "epoch": 6.645879732739421,
      "loss": 0.5900508165359497,
      "loss_ce": 0.0002071046328637749,
      "loss_iou": 0.2490234375,
      "loss_num": 0.0181884765625,
      "loss_xval": 0.58984375,
      "num_input_tokens_seen": 167167308,
      "step": 2984
    },
    {
      "epoch": 6.648106904231626,
      "grad_norm": 15.314505577087402,
      "learning_rate": 1e-06,
      "loss": 0.6925,
      "num_input_tokens_seen": 167223976,
      "step": 2985
    },
    {
      "epoch": 6.648106904231626,
      "loss": 0.6426599621772766,
      "loss_ce": 0.00032597355311736465,
      "loss_iou": 0.28515625,
      "loss_num": 0.0147705078125,
      "loss_xval": 0.640625,
      "num_input_tokens_seen": 167223976,
      "step": 2985
    },
    {
      "epoch": 6.650334075723831,
      "grad_norm": 20.561725616455078,
      "learning_rate": 1e-06,
      "loss": 0.7403,
      "num_input_tokens_seen": 167279468,
      "step": 2986
    },
    {
      "epoch": 6.650334075723831,
      "loss": 0.6292366981506348,
      "loss_ce": 0.00020838108321186155,
      "loss_iou": 0.26953125,
      "loss_num": 0.0177001953125,
      "loss_xval": 0.62890625,
      "num_input_tokens_seen": 167279468,
      "step": 2986
    },
    {
      "epoch": 6.652561247216036,
      "grad_norm": 22.052919387817383,
      "learning_rate": 1e-06,
      "loss": 0.6358,
      "num_input_tokens_seen": 167338024,
      "step": 2987
    },
    {
      "epoch": 6.652561247216036,
      "loss": 0.6393663287162781,
      "loss_ce": 0.00032825471134856343,
      "loss_iou": 0.27734375,
      "loss_num": 0.0172119140625,
      "loss_xval": 0.640625,
      "num_input_tokens_seen": 167338024,
      "step": 2987
    },
    {
      "epoch": 6.6547884187082404,
      "grad_norm": 25.92134666442871,
      "learning_rate": 1e-06,
      "loss": 0.9487,
      "num_input_tokens_seen": 167394980,
      "step": 2988
    },
    {
      "epoch": 6.6547884187082404,
      "loss": 0.8820292949676514,
      "loss_ce": 0.00019336529658176005,
      "loss_iou": 0.376953125,
      "loss_num": 0.0255126953125,
      "loss_xval": 0.8828125,
      "num_input_tokens_seen": 167394980,
      "step": 2988
    },
    {
      "epoch": 6.657015590200445,
      "grad_norm": 18.905071258544922,
      "learning_rate": 1e-06,
      "loss": 0.5956,
      "num_input_tokens_seen": 167451864,
      "step": 2989
    },
    {
      "epoch": 6.657015590200445,
      "loss": 0.5608692169189453,
      "loss_ce": 0.00032231814111582935,
      "loss_iou": 0.251953125,
      "loss_num": 0.01153564453125,
      "loss_xval": 0.5625,
      "num_input_tokens_seen": 167451864,
      "step": 2989
    },
    {
      "epoch": 6.65924276169265,
      "grad_norm": 18.516260147094727,
      "learning_rate": 1e-06,
      "loss": 0.6818,
      "num_input_tokens_seen": 167507760,
      "step": 2990
    },
    {
      "epoch": 6.65924276169265,
      "loss": 0.5940439701080322,
      "loss_ce": 0.0002939357655122876,
      "loss_iou": 0.259765625,
      "loss_num": 0.01519775390625,
      "loss_xval": 0.59375,
      "num_input_tokens_seen": 167507760,
      "step": 2990
    },
    {
      "epoch": 6.661469933184855,
      "grad_norm": 16.960506439208984,
      "learning_rate": 1e-06,
      "loss": 0.6908,
      "num_input_tokens_seen": 167560344,
      "step": 2991
    },
    {
      "epoch": 6.661469933184855,
      "loss": 0.620684802532196,
      "loss_ce": 0.00020142148423474282,
      "loss_iou": 0.263671875,
      "loss_num": 0.018798828125,
      "loss_xval": 0.62109375,
      "num_input_tokens_seen": 167560344,
      "step": 2991
    },
    {
      "epoch": 6.66369710467706,
      "grad_norm": 24.14487648010254,
      "learning_rate": 1e-06,
      "loss": 0.5732,
      "num_input_tokens_seen": 167618800,
      "step": 2992
    },
    {
      "epoch": 6.66369710467706,
      "loss": 0.6208098530769348,
      "loss_ce": 0.0003264534752815962,
      "loss_iou": 0.279296875,
      "loss_num": 0.01226806640625,
      "loss_xval": 0.62109375,
      "num_input_tokens_seen": 167618800,
      "step": 2992
    },
    {
      "epoch": 6.665924276169265,
      "grad_norm": 16.674419403076172,
      "learning_rate": 1e-06,
      "loss": 0.4944,
      "num_input_tokens_seen": 167674316,
      "step": 2993
    },
    {
      "epoch": 6.665924276169265,
      "loss": 0.5599268674850464,
      "loss_ce": 0.0002344850217923522,
      "loss_iou": 0.2255859375,
      "loss_num": 0.0216064453125,
      "loss_xval": 0.55859375,
      "num_input_tokens_seen": 167674316,
      "step": 2993
    },
    {
      "epoch": 6.66815144766147,
      "grad_norm": 14.498659133911133,
      "learning_rate": 1e-06,
      "loss": 0.8444,
      "num_input_tokens_seen": 167730912,
      "step": 2994
    },
    {
      "epoch": 6.66815144766147,
      "loss": 0.7760642766952515,
      "loss_ce": 0.0009178139735013247,
      "loss_iou": 0.31640625,
      "loss_num": 0.0283203125,
      "loss_xval": 0.7734375,
      "num_input_tokens_seen": 167730912,
      "step": 2994
    },
    {
      "epoch": 6.6703786191536745,
      "grad_norm": 25.62771987915039,
      "learning_rate": 1e-06,
      "loss": 0.7389,
      "num_input_tokens_seen": 167786696,
      "step": 2995
    },
    {
      "epoch": 6.6703786191536745,
      "loss": 0.6130115985870361,
      "loss_ce": 0.00021857366664335132,
      "loss_iou": 0.27734375,
      "loss_num": 0.01190185546875,
      "loss_xval": 0.61328125,
      "num_input_tokens_seen": 167786696,
      "step": 2995
    },
    {
      "epoch": 6.67260579064588,
      "grad_norm": 84.09690856933594,
      "learning_rate": 1e-06,
      "loss": 0.7173,
      "num_input_tokens_seen": 167843364,
      "step": 2996
    },
    {
      "epoch": 6.67260579064588,
      "loss": 0.5746430158615112,
      "loss_ce": 0.00018016056856140494,
      "loss_iou": 0.2578125,
      "loss_num": 0.011962890625,
      "loss_xval": 0.57421875,
      "num_input_tokens_seen": 167843364,
      "step": 2996
    },
    {
      "epoch": 6.674832962138085,
      "grad_norm": 25.761995315551758,
      "learning_rate": 1e-06,
      "loss": 0.7212,
      "num_input_tokens_seen": 167900084,
      "step": 2997
    },
    {
      "epoch": 6.674832962138085,
      "loss": 0.848706066608429,
      "loss_ce": 0.0003174202283844352,
      "loss_iou": 0.3671875,
      "loss_num": 0.0230712890625,
      "loss_xval": 0.84765625,
      "num_input_tokens_seen": 167900084,
      "step": 2997
    },
    {
      "epoch": 6.67706013363029,
      "grad_norm": 20.495059967041016,
      "learning_rate": 1e-06,
      "loss": 0.6009,
      "num_input_tokens_seen": 167955952,
      "step": 2998
    },
    {
      "epoch": 6.67706013363029,
      "loss": 0.6451590061187744,
      "loss_ce": 0.0002615168341435492,
      "loss_iou": 0.2890625,
      "loss_num": 0.0135498046875,
      "loss_xval": 0.64453125,
      "num_input_tokens_seen": 167955952,
      "step": 2998
    },
    {
      "epoch": 6.679287305122495,
      "grad_norm": 18.190731048583984,
      "learning_rate": 1e-06,
      "loss": 0.5682,
      "num_input_tokens_seen": 168011540,
      "step": 2999
    },
    {
      "epoch": 6.679287305122495,
      "loss": 0.5880811810493469,
      "loss_ce": 0.0001905930694192648,
      "loss_iou": 0.25390625,
      "loss_num": 0.0157470703125,
      "loss_xval": 0.5859375,
      "num_input_tokens_seen": 168011540,
      "step": 2999
    },
    {
      "epoch": 6.6815144766147,
      "grad_norm": 15.931896209716797,
      "learning_rate": 1e-06,
      "loss": 0.556,
      "num_input_tokens_seen": 168068960,
      "step": 3000
    },
    {
      "epoch": 6.6815144766147,
      "eval_seeclick_web_CIoU": 0.5745645761489868,
      "eval_seeclick_web_GIoU": 0.5703821778297424,
      "eval_seeclick_web_IoU": 0.5911202728748322,
      "eval_seeclick_web_MAE_all": 0.016741125378757715,
      "eval_seeclick_web_MAE_h": 0.009390837512910366,
      "eval_seeclick_web_MAE_w": 0.01750754565000534,
      "eval_seeclick_web_MAE_x_boxes": 0.009469148702919483,
      "eval_seeclick_web_MAE_y_boxes": 0.022113264771178365,
      "eval_seeclick_web_inside_bbox": 0.9166666567325592,
      "eval_seeclick_web_loss": 0.9277820587158203,
      "eval_seeclick_web_loss_ce": 0.00027503210003487766,
      "eval_seeclick_web_loss_iou": 0.421875,
      "eval_seeclick_web_loss_num": 0.013193130493164062,
      "eval_seeclick_web_loss_xval": 0.909423828125,
      "eval_seeclick_web_runtime": 22.6295,
      "eval_seeclick_web_samples_per_second": 2.21,
      "eval_seeclick_web_steps_per_second": 0.088,
      "num_input_tokens_seen": 168068960,
      "step": 3000
    },
    {
      "epoch": 6.6815144766147,
      "eval_icons_CIoU": 0.28626811504364014,
      "eval_icons_GIoU": 0.31125396490097046,
      "eval_icons_IoU": 0.36675940454006195,
      "eval_icons_MAE_all": 0.06613602861762047,
      "eval_icons_MAE_h": 0.04001020174473524,
      "eval_icons_MAE_w": 0.07005784474313259,
      "eval_icons_MAE_x_boxes": 0.06103345938026905,
      "eval_icons_MAE_y_boxes": 0.039336864836513996,
      "eval_icons_inside_bbox": 0.6336805522441864,
      "eval_icons_loss": 1.7287678718566895,
      "eval_icons_loss_ce": 0.00033969079959206283,
      "eval_icons_loss_iou": 0.669921875,
      "eval_icons_loss_num": 0.06263542175292969,
      "eval_icons_loss_xval": 1.65380859375,
      "eval_icons_runtime": 20.8368,
      "eval_icons_samples_per_second": 2.4,
      "eval_icons_steps_per_second": 0.096,
      "num_input_tokens_seen": 168068960,
      "step": 3000
    },
    {
      "epoch": 6.6815144766147,
      "eval_screenspot_CIoU": 0.34714128573735553,
      "eval_screenspot_GIoU": 0.36292509237925213,
      "eval_screenspot_IoU": 0.42989224195480347,
      "eval_screenspot_MAE_all": 0.06254567454258601,
      "eval_screenspot_MAE_h": 0.03752323302129904,
      "eval_screenspot_MAE_w": 0.07452885061502457,
      "eval_screenspot_MAE_x_boxes": 0.07285770525534947,
      "eval_screenspot_MAE_y_boxes": 0.04644376132637262,
      "eval_screenspot_inside_bbox": 0.659583330154419,
      "eval_screenspot_loss": 1.6588377952575684,
      "eval_screenspot_loss_ce": 0.00031722194398753345,
      "eval_screenspot_loss_iou": 0.6805826822916666,
      "eval_screenspot_loss_num": 0.07418060302734375,
      "eval_screenspot_loss_xval": 1.7322591145833333,
      "eval_screenspot_runtime": 32.6151,
      "eval_screenspot_samples_per_second": 2.729,
      "eval_screenspot_steps_per_second": 0.092,
      "num_input_tokens_seen": 168068960,
      "step": 3000
    },
    {
      "epoch": 6.6815144766147,
      "eval_compot_CIoU": 0.3531677573919296,
      "eval_compot_GIoU": 0.3627708703279495,
      "eval_compot_IoU": 0.4095195233821869,
      "eval_compot_MAE_all": 0.01802563015371561,
      "eval_compot_MAE_h": 0.008967408444732428,
      "eval_compot_MAE_w": 0.022056237794458866,
      "eval_compot_MAE_x_boxes": 0.02982906997203827,
      "eval_compot_MAE_y_boxes": 0.006724046776071191,
      "eval_compot_inside_bbox": 0.6458333432674408,
      "eval_compot_loss": 1.3883819580078125,
      "eval_compot_loss_ce": 0.0002577164559625089,
      "eval_compot_loss_iou": 0.633544921875,
      "eval_compot_loss_num": 0.016773223876953125,
      "eval_compot_loss_xval": 1.350830078125,
      "eval_compot_runtime": 20.0639,
      "eval_compot_samples_per_second": 2.492,
      "eval_compot_steps_per_second": 0.1,
      "num_input_tokens_seen": 168068960,
      "step": 3000
    },
    {
      "epoch": 6.6815144766147,
      "eval_custom_ui_val_CIoU": 0.46863051421112484,
      "eval_custom_ui_val_GIoU": 0.4837728473875258,
      "eval_custom_ui_val_IoU": 0.5239554312494066,
      "eval_custom_ui_val_MAE_all": 0.030405950939489737,
      "eval_custom_ui_val_MAE_h": 0.01713582917323543,
      "eval_custom_ui_val_MAE_w": 0.038816668921046786,
      "eval_custom_ui_val_MAE_x_boxes": 0.034813721767730184,
      "eval_custom_ui_val_MAE_y_boxes": 0.015315383543363877,
      "eval_custom_ui_val_inside_bbox": 0.7422839535607232,
      "eval_custom_ui_val_loss": 1.1846333742141724,
      "eval_custom_ui_val_loss_ce": 0.00029205658291983936,
      "eval_custom_ui_val_loss_iou": 0.5049913194444444,
      "eval_custom_ui_val_loss_num": 0.02808909946017795,
      "eval_custom_ui_val_loss_xval": 1.1504720052083333,
      "eval_custom_ui_val_runtime": 56.2868,
      "eval_custom_ui_val_samples_per_second": 4.708,
      "eval_custom_ui_val_steps_per_second": 0.16,
      "num_input_tokens_seen": 168068960,
      "step": 3000
    },
    {
      "epoch": 6.6815144766147,
      "loss": 0.9355728030204773,
      "loss_ce": 0.00027009687619283795,
      "loss_iou": 0.40625,
      "loss_num": 0.0244140625,
      "loss_xval": 0.93359375,
      "num_input_tokens_seen": 168068960,
      "step": 3000
    },
    {
      "epoch": 6.6837416481069045,
      "grad_norm": 13.867417335510254,
      "learning_rate": 1e-06,
      "loss": 0.778,
      "num_input_tokens_seen": 168123220,
      "step": 3001
    },
    {
      "epoch": 6.6837416481069045,
      "loss": 0.8318126201629639,
      "loss_ce": 0.00026967888697981834,
      "loss_iou": 0.345703125,
      "loss_num": 0.0283203125,
      "loss_xval": 0.83203125,
      "num_input_tokens_seen": 168123220,
      "step": 3001
    },
    {
      "epoch": 6.685968819599109,
      "grad_norm": 23.010339736938477,
      "learning_rate": 1e-06,
      "loss": 0.8531,
      "num_input_tokens_seen": 168177788,
      "step": 3002
    },
    {
      "epoch": 6.685968819599109,
      "loss": 0.9224424958229065,
      "loss_ce": 0.00032336413278244436,
      "loss_iou": 0.390625,
      "loss_num": 0.028076171875,
      "loss_xval": 0.921875,
      "num_input_tokens_seen": 168177788,
      "step": 3002
    },
    {
      "epoch": 6.688195991091314,
      "grad_norm": 17.75686264038086,
      "learning_rate": 1e-06,
      "loss": 0.6219,
      "num_input_tokens_seen": 168232444,
      "step": 3003
    },
    {
      "epoch": 6.688195991091314,
      "loss": 0.38138991594314575,
      "loss_ce": 0.00016430205141659826,
      "loss_iou": 0.166015625,
      "loss_num": 0.00994873046875,
      "loss_xval": 0.380859375,
      "num_input_tokens_seen": 168232444,
      "step": 3003
    },
    {
      "epoch": 6.690423162583519,
      "grad_norm": 25.14574432373047,
      "learning_rate": 1e-06,
      "loss": 0.6382,
      "num_input_tokens_seen": 168285212,
      "step": 3004
    },
    {
      "epoch": 6.690423162583519,
      "loss": 0.7175732851028442,
      "loss_ce": 0.000288100796751678,
      "loss_iou": 0.2890625,
      "loss_num": 0.028076171875,
      "loss_xval": 0.71875,
      "num_input_tokens_seen": 168285212,
      "step": 3004
    },
    {
      "epoch": 6.692650334075724,
      "grad_norm": 15.96047592163086,
      "learning_rate": 1e-06,
      "loss": 0.8216,
      "num_input_tokens_seen": 168341780,
      "step": 3005
    },
    {
      "epoch": 6.692650334075724,
      "loss": 1.042626142501831,
      "loss_ce": 0.0002676715375855565,
      "loss_iou": 0.408203125,
      "loss_num": 0.045166015625,
      "loss_xval": 1.0390625,
      "num_input_tokens_seen": 168341780,
      "step": 3005
    },
    {
      "epoch": 6.694877505567929,
      "grad_norm": 14.52697467803955,
      "learning_rate": 1e-06,
      "loss": 0.4355,
      "num_input_tokens_seen": 168397592,
      "step": 3006
    },
    {
      "epoch": 6.694877505567929,
      "loss": 0.4221663177013397,
      "loss_ce": 0.00023030643933452666,
      "loss_iou": 0.17578125,
      "loss_num": 0.0140380859375,
      "loss_xval": 0.421875,
      "num_input_tokens_seen": 168397592,
      "step": 3006
    },
    {
      "epoch": 6.697104677060134,
      "grad_norm": 17.11386489868164,
      "learning_rate": 1e-06,
      "loss": 0.7575,
      "num_input_tokens_seen": 168453048,
      "step": 3007
    },
    {
      "epoch": 6.697104677060134,
      "loss": 0.9401570558547974,
      "loss_ce": 0.0002156144182663411,
      "loss_iou": 0.408203125,
      "loss_num": 0.0247802734375,
      "loss_xval": 0.94140625,
      "num_input_tokens_seen": 168453048,
      "step": 3007
    },
    {
      "epoch": 6.6993318485523385,
      "grad_norm": 23.60494041442871,
      "learning_rate": 1e-06,
      "loss": 0.6955,
      "num_input_tokens_seen": 168508760,
      "step": 3008
    },
    {
      "epoch": 6.6993318485523385,
      "loss": 0.49571555852890015,
      "loss_ce": 0.00023218031856231391,
      "loss_iou": 0.2314453125,
      "loss_num": 0.006805419921875,
      "loss_xval": 0.49609375,
      "num_input_tokens_seen": 168508760,
      "step": 3008
    },
    {
      "epoch": 6.701559020044543,
      "grad_norm": 14.749323844909668,
      "learning_rate": 1e-06,
      "loss": 0.6877,
      "num_input_tokens_seen": 168565604,
      "step": 3009
    },
    {
      "epoch": 6.701559020044543,
      "loss": 0.6759235858917236,
      "loss_ce": 0.0003864543105009943,
      "loss_iou": 0.271484375,
      "loss_num": 0.0267333984375,
      "loss_xval": 0.67578125,
      "num_input_tokens_seen": 168565604,
      "step": 3009
    },
    {
      "epoch": 6.703786191536748,
      "grad_norm": 24.45879554748535,
      "learning_rate": 1e-06,
      "loss": 0.6726,
      "num_input_tokens_seen": 168623400,
      "step": 3010
    },
    {
      "epoch": 6.703786191536748,
      "loss": 0.5331798791885376,
      "loss_ce": 0.00022087327670305967,
      "loss_iou": 0.23828125,
      "loss_num": 0.01153564453125,
      "loss_xval": 0.53125,
      "num_input_tokens_seen": 168623400,
      "step": 3010
    },
    {
      "epoch": 6.706013363028953,
      "grad_norm": 22.486263275146484,
      "learning_rate": 1e-06,
      "loss": 0.5951,
      "num_input_tokens_seen": 168678412,
      "step": 3011
    },
    {
      "epoch": 6.706013363028953,
      "loss": 0.6491339206695557,
      "loss_ce": 0.0002081873535644263,
      "loss_iou": 0.259765625,
      "loss_num": 0.0263671875,
      "loss_xval": 0.6484375,
      "num_input_tokens_seen": 168678412,
      "step": 3011
    },
    {
      "epoch": 6.708240534521158,
      "grad_norm": 12.50680923461914,
      "learning_rate": 1e-06,
      "loss": 0.721,
      "num_input_tokens_seen": 168734048,
      "step": 3012
    },
    {
      "epoch": 6.708240534521158,
      "loss": 0.7475176453590393,
      "loss_ce": 0.00020322672207839787,
      "loss_iou": 0.296875,
      "loss_num": 0.0306396484375,
      "loss_xval": 0.74609375,
      "num_input_tokens_seen": 168734048,
      "step": 3012
    },
    {
      "epoch": 6.710467706013363,
      "grad_norm": 28.51459503173828,
      "learning_rate": 1e-06,
      "loss": 0.5842,
      "num_input_tokens_seen": 168790568,
      "step": 3013
    },
    {
      "epoch": 6.710467706013363,
      "loss": 0.6308811902999878,
      "loss_ce": 0.000265932030742988,
      "loss_iou": 0.275390625,
      "loss_num": 0.0159912109375,
      "loss_xval": 0.62890625,
      "num_input_tokens_seen": 168790568,
      "step": 3013
    },
    {
      "epoch": 6.712694877505568,
      "grad_norm": 18.395198822021484,
      "learning_rate": 1e-06,
      "loss": 0.8135,
      "num_input_tokens_seen": 168844228,
      "step": 3014
    },
    {
      "epoch": 6.712694877505568,
      "loss": 0.7819979190826416,
      "loss_ce": 0.00025961300707422197,
      "loss_iou": 0.333984375,
      "loss_num": 0.0230712890625,
      "loss_xval": 0.78125,
      "num_input_tokens_seen": 168844228,
      "step": 3014
    },
    {
      "epoch": 6.714922048997773,
      "grad_norm": 23.87611198425293,
      "learning_rate": 1e-06,
      "loss": 0.7382,
      "num_input_tokens_seen": 168899728,
      "step": 3015
    },
    {
      "epoch": 6.714922048997773,
      "loss": 0.7516616582870483,
      "loss_ce": 0.00019676925148814917,
      "loss_iou": 0.326171875,
      "loss_num": 0.0198974609375,
      "loss_xval": 0.75,
      "num_input_tokens_seen": 168899728,
      "step": 3015
    },
    {
      "epoch": 6.717149220489977,
      "grad_norm": 20.738792419433594,
      "learning_rate": 1e-06,
      "loss": 0.786,
      "num_input_tokens_seen": 168956592,
      "step": 3016
    },
    {
      "epoch": 6.717149220489977,
      "loss": 0.8078007102012634,
      "loss_ce": 0.00018351592007093132,
      "loss_iou": 0.306640625,
      "loss_num": 0.0390625,
      "loss_xval": 0.80859375,
      "num_input_tokens_seen": 168956592,
      "step": 3016
    },
    {
      "epoch": 6.719376391982182,
      "grad_norm": 19.863662719726562,
      "learning_rate": 1e-06,
      "loss": 0.7038,
      "num_input_tokens_seen": 169012588,
      "step": 3017
    },
    {
      "epoch": 6.719376391982182,
      "loss": 0.5791813135147095,
      "loss_ce": 0.00020178730483166873,
      "loss_iou": 0.24609375,
      "loss_num": 0.017333984375,
      "loss_xval": 0.578125,
      "num_input_tokens_seen": 169012588,
      "step": 3017
    },
    {
      "epoch": 6.721603563474387,
      "grad_norm": 18.61119842529297,
      "learning_rate": 1e-06,
      "loss": 0.6261,
      "num_input_tokens_seen": 169067608,
      "step": 3018
    },
    {
      "epoch": 6.721603563474387,
      "loss": 0.6960831880569458,
      "loss_ce": 0.0002824185648933053,
      "loss_iou": 0.31640625,
      "loss_num": 0.01263427734375,
      "loss_xval": 0.6953125,
      "num_input_tokens_seen": 169067608,
      "step": 3018
    },
    {
      "epoch": 6.723830734966592,
      "grad_norm": 19.781635284423828,
      "learning_rate": 1e-06,
      "loss": 0.6468,
      "num_input_tokens_seen": 169122988,
      "step": 3019
    },
    {
      "epoch": 6.723830734966592,
      "loss": 0.7083573341369629,
      "loss_ce": 0.0002274075523018837,
      "loss_iou": 0.26953125,
      "loss_num": 0.033447265625,
      "loss_xval": 0.70703125,
      "num_input_tokens_seen": 169122988,
      "step": 3019
    },
    {
      "epoch": 6.726057906458797,
      "grad_norm": 17.748201370239258,
      "learning_rate": 1e-06,
      "loss": 0.5458,
      "num_input_tokens_seen": 169178700,
      "step": 3020
    },
    {
      "epoch": 6.726057906458797,
      "loss": 0.5851503610610962,
      "loss_ce": 0.00018942491442430764,
      "loss_iou": 0.25390625,
      "loss_num": 0.01513671875,
      "loss_xval": 0.5859375,
      "num_input_tokens_seen": 169178700,
      "step": 3020
    },
    {
      "epoch": 6.728285077951003,
      "grad_norm": 14.647541999816895,
      "learning_rate": 1e-06,
      "loss": 0.4199,
      "num_input_tokens_seen": 169234748,
      "step": 3021
    },
    {
      "epoch": 6.728285077951003,
      "loss": 0.4449549615383148,
      "loss_ce": 0.00019176513887941837,
      "loss_iou": 0.17578125,
      "loss_num": 0.0185546875,
      "loss_xval": 0.4453125,
      "num_input_tokens_seen": 169234748,
      "step": 3021
    },
    {
      "epoch": 6.7305122494432075,
      "grad_norm": 22.735586166381836,
      "learning_rate": 1e-06,
      "loss": 0.8144,
      "num_input_tokens_seen": 169291708,
      "step": 3022
    },
    {
      "epoch": 6.7305122494432075,
      "loss": 0.8748047351837158,
      "loss_ce": 0.0002930166956502944,
      "loss_iou": 0.328125,
      "loss_num": 0.0439453125,
      "loss_xval": 0.875,
      "num_input_tokens_seen": 169291708,
      "step": 3022
    },
    {
      "epoch": 6.732739420935412,
      "grad_norm": 10.555468559265137,
      "learning_rate": 1e-06,
      "loss": 0.8329,
      "num_input_tokens_seen": 169346188,
      "step": 3023
    },
    {
      "epoch": 6.732739420935412,
      "loss": 0.5163635611534119,
      "loss_ce": 0.00018923793686553836,
      "loss_iou": 0.1982421875,
      "loss_num": 0.0240478515625,
      "loss_xval": 0.515625,
      "num_input_tokens_seen": 169346188,
      "step": 3023
    },
    {
      "epoch": 6.734966592427617,
      "grad_norm": 25.756729125976562,
      "learning_rate": 1e-06,
      "loss": 0.7476,
      "num_input_tokens_seen": 169401748,
      "step": 3024
    },
    {
      "epoch": 6.734966592427617,
      "loss": 0.6750069856643677,
      "loss_ce": 0.00020230711379554123,
      "loss_iou": 0.291015625,
      "loss_num": 0.0186767578125,
      "loss_xval": 0.67578125,
      "num_input_tokens_seen": 169401748,
      "step": 3024
    },
    {
      "epoch": 6.737193763919822,
      "grad_norm": 16.90264129638672,
      "learning_rate": 1e-06,
      "loss": 0.5714,
      "num_input_tokens_seen": 169456820,
      "step": 3025
    },
    {
      "epoch": 6.737193763919822,
      "loss": 0.4832545220851898,
      "loss_ce": 0.0004664373118430376,
      "loss_iou": 0.2041015625,
      "loss_num": 0.01483154296875,
      "loss_xval": 0.482421875,
      "num_input_tokens_seen": 169456820,
      "step": 3025
    },
    {
      "epoch": 6.739420935412027,
      "grad_norm": 19.030052185058594,
      "learning_rate": 1e-06,
      "loss": 0.8233,
      "num_input_tokens_seen": 169512288,
      "step": 3026
    },
    {
      "epoch": 6.739420935412027,
      "loss": 0.8015903234481812,
      "loss_ce": 0.00032082191319204867,
      "loss_iou": 0.3515625,
      "loss_num": 0.0194091796875,
      "loss_xval": 0.80078125,
      "num_input_tokens_seen": 169512288,
      "step": 3026
    },
    {
      "epoch": 6.741648106904232,
      "grad_norm": 22.391340255737305,
      "learning_rate": 1e-06,
      "loss": 0.5371,
      "num_input_tokens_seen": 169568768,
      "step": 3027
    },
    {
      "epoch": 6.741648106904232,
      "loss": 0.39635348320007324,
      "loss_ce": 0.00017430493608117104,
      "loss_iou": 0.154296875,
      "loss_num": 0.017578125,
      "loss_xval": 0.396484375,
      "num_input_tokens_seen": 169568768,
      "step": 3027
    },
    {
      "epoch": 6.743875278396437,
      "grad_norm": 23.041179656982422,
      "learning_rate": 1e-06,
      "loss": 0.6764,
      "num_input_tokens_seen": 169623344,
      "step": 3028
    },
    {
      "epoch": 6.743875278396437,
      "loss": 0.6840122938156128,
      "loss_ce": 0.00017437028873246163,
      "loss_iou": 0.296875,
      "loss_num": 0.0185546875,
      "loss_xval": 0.68359375,
      "num_input_tokens_seen": 169623344,
      "step": 3028
    },
    {
      "epoch": 6.7461024498886415,
      "grad_norm": 22.090497970581055,
      "learning_rate": 1e-06,
      "loss": 0.6771,
      "num_input_tokens_seen": 169677776,
      "step": 3029
    },
    {
      "epoch": 6.7461024498886415,
      "loss": 0.7656627297401428,
      "loss_ce": 0.00015980687749106437,
      "loss_iou": 0.33203125,
      "loss_num": 0.020263671875,
      "loss_xval": 0.765625,
      "num_input_tokens_seen": 169677776,
      "step": 3029
    },
    {
      "epoch": 6.748329621380846,
      "grad_norm": 16.185291290283203,
      "learning_rate": 1e-06,
      "loss": 0.6104,
      "num_input_tokens_seen": 169734036,
      "step": 3030
    },
    {
      "epoch": 6.748329621380846,
      "loss": 0.643264651298523,
      "loss_ce": 0.00019819998124148697,
      "loss_iou": 0.279296875,
      "loss_num": 0.01708984375,
      "loss_xval": 0.64453125,
      "num_input_tokens_seen": 169734036,
      "step": 3030
    },
    {
      "epoch": 6.750556792873051,
      "grad_norm": 19.104719161987305,
      "learning_rate": 1e-06,
      "loss": 0.5948,
      "num_input_tokens_seen": 169788020,
      "step": 3031
    },
    {
      "epoch": 6.750556792873051,
      "loss": 0.5878357887268066,
      "loss_ce": 0.00018932024249807,
      "loss_iou": 0.251953125,
      "loss_num": 0.0164794921875,
      "loss_xval": 0.5859375,
      "num_input_tokens_seen": 169788020,
      "step": 3031
    },
    {
      "epoch": 6.752783964365256,
      "grad_norm": 23.02263641357422,
      "learning_rate": 1e-06,
      "loss": 0.629,
      "num_input_tokens_seen": 169844864,
      "step": 3032
    },
    {
      "epoch": 6.752783964365256,
      "loss": 0.5922929048538208,
      "loss_ce": 0.0002518455730751157,
      "loss_iou": 0.2373046875,
      "loss_num": 0.0235595703125,
      "loss_xval": 0.59375,
      "num_input_tokens_seen": 169844864,
      "step": 3032
    },
    {
      "epoch": 6.755011135857461,
      "grad_norm": 21.84469985961914,
      "learning_rate": 1e-06,
      "loss": 0.6563,
      "num_input_tokens_seen": 169901280,
      "step": 3033
    },
    {
      "epoch": 6.755011135857461,
      "loss": 0.6547205448150635,
      "loss_ce": 0.0001795107964426279,
      "loss_iou": 0.263671875,
      "loss_num": 0.025146484375,
      "loss_xval": 0.65625,
      "num_input_tokens_seen": 169901280,
      "step": 3033
    },
    {
      "epoch": 6.757238307349666,
      "grad_norm": 24.893165588378906,
      "learning_rate": 1e-06,
      "loss": 0.8051,
      "num_input_tokens_seen": 169955720,
      "step": 3034
    },
    {
      "epoch": 6.757238307349666,
      "loss": 0.9663440585136414,
      "loss_ce": 0.00027958687860518694,
      "loss_iou": 0.4140625,
      "loss_num": 0.0274658203125,
      "loss_xval": 0.96484375,
      "num_input_tokens_seen": 169955720,
      "step": 3034
    },
    {
      "epoch": 6.759465478841871,
      "grad_norm": 29.10569953918457,
      "learning_rate": 1e-06,
      "loss": 0.6784,
      "num_input_tokens_seen": 170011236,
      "step": 3035
    },
    {
      "epoch": 6.759465478841871,
      "loss": 0.6698566675186157,
      "loss_ce": 0.00017896070494316518,
      "loss_iou": 0.287109375,
      "loss_num": 0.0191650390625,
      "loss_xval": 0.66796875,
      "num_input_tokens_seen": 170011236,
      "step": 3035
    },
    {
      "epoch": 6.7616926503340755,
      "grad_norm": 22.953475952148438,
      "learning_rate": 1e-06,
      "loss": 1.0044,
      "num_input_tokens_seen": 170068108,
      "step": 3036
    },
    {
      "epoch": 6.7616926503340755,
      "loss": 1.014899492263794,
      "loss_ce": 0.0002510766498744488,
      "loss_iou": 0.416015625,
      "loss_num": 0.0361328125,
      "loss_xval": 1.015625,
      "num_input_tokens_seen": 170068108,
      "step": 3036
    },
    {
      "epoch": 6.76391982182628,
      "grad_norm": 18.038188934326172,
      "learning_rate": 1e-06,
      "loss": 0.6545,
      "num_input_tokens_seen": 170127512,
      "step": 3037
    },
    {
      "epoch": 6.76391982182628,
      "loss": 0.6023421883583069,
      "loss_ce": 0.00029138405807316303,
      "loss_iou": 0.271484375,
      "loss_num": 0.0118408203125,
      "loss_xval": 0.6015625,
      "num_input_tokens_seen": 170127512,
      "step": 3037
    },
    {
      "epoch": 6.766146993318485,
      "grad_norm": 20.037540435791016,
      "learning_rate": 1e-06,
      "loss": 0.5158,
      "num_input_tokens_seen": 170181356,
      "step": 3038
    },
    {
      "epoch": 6.766146993318485,
      "loss": 0.5402380228042603,
      "loss_ce": 0.00019893058924935758,
      "loss_iou": 0.2294921875,
      "loss_num": 0.0162353515625,
      "loss_xval": 0.5390625,
      "num_input_tokens_seen": 170181356,
      "step": 3038
    },
    {
      "epoch": 6.76837416481069,
      "grad_norm": 16.7264347076416,
      "learning_rate": 1e-06,
      "loss": 0.9252,
      "num_input_tokens_seen": 170237844,
      "step": 3039
    },
    {
      "epoch": 6.76837416481069,
      "loss": 1.0988414287567139,
      "loss_ce": 0.00020856756600551307,
      "loss_iou": 0.46875,
      "loss_num": 0.032470703125,
      "loss_xval": 1.1015625,
      "num_input_tokens_seen": 170237844,
      "step": 3039
    },
    {
      "epoch": 6.770601336302895,
      "grad_norm": 14.285926818847656,
      "learning_rate": 1e-06,
      "loss": 0.5733,
      "num_input_tokens_seen": 170291264,
      "step": 3040
    },
    {
      "epoch": 6.770601336302895,
      "loss": 0.44891494512557983,
      "loss_ce": 0.00018448734772391617,
      "loss_iou": 0.1826171875,
      "loss_num": 0.0167236328125,
      "loss_xval": 0.44921875,
      "num_input_tokens_seen": 170291264,
      "step": 3040
    },
    {
      "epoch": 6.772828507795101,
      "grad_norm": 29.201885223388672,
      "learning_rate": 1e-06,
      "loss": 0.5023,
      "num_input_tokens_seen": 170348048,
      "step": 3041
    },
    {
      "epoch": 6.772828507795101,
      "loss": 0.4634665846824646,
      "loss_ce": 0.000209752848604694,
      "loss_iou": 0.2021484375,
      "loss_num": 0.01190185546875,
      "loss_xval": 0.462890625,
      "num_input_tokens_seen": 170348048,
      "step": 3041
    },
    {
      "epoch": 6.775055679287306,
      "grad_norm": 41.29672622680664,
      "learning_rate": 1e-06,
      "loss": 0.7947,
      "num_input_tokens_seen": 170402500,
      "step": 3042
    },
    {
      "epoch": 6.775055679287306,
      "loss": 0.8640490770339966,
      "loss_ce": 0.00027952369418926537,
      "loss_iou": 0.37109375,
      "loss_num": 0.02392578125,
      "loss_xval": 0.86328125,
      "num_input_tokens_seen": 170402500,
      "step": 3042
    },
    {
      "epoch": 6.77728285077951,
      "grad_norm": 31.665985107421875,
      "learning_rate": 1e-06,
      "loss": 0.5856,
      "num_input_tokens_seen": 170458260,
      "step": 3043
    },
    {
      "epoch": 6.77728285077951,
      "loss": 0.45639604330062866,
      "loss_ce": 0.00021927341003902256,
      "loss_iou": 0.18359375,
      "loss_num": 0.0177001953125,
      "loss_xval": 0.45703125,
      "num_input_tokens_seen": 170458260,
      "step": 3043
    },
    {
      "epoch": 6.779510022271715,
      "grad_norm": 43.43659591674805,
      "learning_rate": 1e-06,
      "loss": 0.7794,
      "num_input_tokens_seen": 170514684,
      "step": 3044
    },
    {
      "epoch": 6.779510022271715,
      "loss": 0.8676573038101196,
      "loss_ce": 0.0009580720216035843,
      "loss_iou": 0.365234375,
      "loss_num": 0.02734375,
      "loss_xval": 0.8671875,
      "num_input_tokens_seen": 170514684,
      "step": 3044
    },
    {
      "epoch": 6.78173719376392,
      "grad_norm": 24.82050323486328,
      "learning_rate": 1e-06,
      "loss": 0.531,
      "num_input_tokens_seen": 170569476,
      "step": 3045
    },
    {
      "epoch": 6.78173719376392,
      "loss": 0.5165289640426636,
      "loss_ce": 0.00017151227802969515,
      "loss_iou": 0.2255859375,
      "loss_num": 0.01318359375,
      "loss_xval": 0.515625,
      "num_input_tokens_seen": 170569476,
      "step": 3045
    },
    {
      "epoch": 6.783964365256125,
      "grad_norm": 19.520343780517578,
      "learning_rate": 1e-06,
      "loss": 0.7418,
      "num_input_tokens_seen": 170622728,
      "step": 3046
    },
    {
      "epoch": 6.783964365256125,
      "loss": 0.9240363836288452,
      "loss_ce": 0.00026935621281154454,
      "loss_iou": 0.380859375,
      "loss_num": 0.032470703125,
      "loss_xval": 0.921875,
      "num_input_tokens_seen": 170622728,
      "step": 3046
    },
    {
      "epoch": 6.78619153674833,
      "grad_norm": 18.211183547973633,
      "learning_rate": 1e-06,
      "loss": 0.6673,
      "num_input_tokens_seen": 170679272,
      "step": 3047
    },
    {
      "epoch": 6.78619153674833,
      "loss": 0.7375867962837219,
      "loss_ce": 0.000282141612842679,
      "loss_iou": 0.328125,
      "loss_num": 0.0162353515625,
      "loss_xval": 0.73828125,
      "num_input_tokens_seen": 170679272,
      "step": 3047
    },
    {
      "epoch": 6.788418708240535,
      "grad_norm": 21.0445499420166,
      "learning_rate": 1e-06,
      "loss": 0.5899,
      "num_input_tokens_seen": 170733732,
      "step": 3048
    },
    {
      "epoch": 6.788418708240535,
      "loss": 0.5784125924110413,
      "loss_ce": 0.00016554353351239115,
      "loss_iou": 0.248046875,
      "loss_num": 0.016357421875,
      "loss_xval": 0.578125,
      "num_input_tokens_seen": 170733732,
      "step": 3048
    },
    {
      "epoch": 6.79064587973274,
      "grad_norm": 25.12030601501465,
      "learning_rate": 1e-06,
      "loss": 0.7525,
      "num_input_tokens_seen": 170790864,
      "step": 3049
    },
    {
      "epoch": 6.79064587973274,
      "loss": 0.8492100238800049,
      "loss_ce": 0.0011875506024807692,
      "loss_iou": 0.349609375,
      "loss_num": 0.0299072265625,
      "loss_xval": 0.84765625,
      "num_input_tokens_seen": 170790864,
      "step": 3049
    },
    {
      "epoch": 6.7928730512249444,
      "grad_norm": 14.534224510192871,
      "learning_rate": 1e-06,
      "loss": 0.5735,
      "num_input_tokens_seen": 170846928,
      "step": 3050
    },
    {
      "epoch": 6.7928730512249444,
      "loss": 0.6872592568397522,
      "loss_ce": 0.00024754402693361044,
      "loss_iou": 0.30078125,
      "loss_num": 0.01708984375,
      "loss_xval": 0.6875,
      "num_input_tokens_seen": 170846928,
      "step": 3050
    },
    {
      "epoch": 6.795100222717149,
      "grad_norm": 32.43819808959961,
      "learning_rate": 1e-06,
      "loss": 0.7584,
      "num_input_tokens_seen": 170901412,
      "step": 3051
    },
    {
      "epoch": 6.795100222717149,
      "loss": 0.47796866297721863,
      "loss_ce": 0.00018544365593697876,
      "loss_iou": 0.2138671875,
      "loss_num": 0.00994873046875,
      "loss_xval": 0.478515625,
      "num_input_tokens_seen": 170901412,
      "step": 3051
    },
    {
      "epoch": 6.797327394209354,
      "grad_norm": 26.212574005126953,
      "learning_rate": 1e-06,
      "loss": 0.7598,
      "num_input_tokens_seen": 170955608,
      "step": 3052
    },
    {
      "epoch": 6.797327394209354,
      "loss": 0.8884332776069641,
      "loss_ce": 0.0002496936358511448,
      "loss_iou": 0.39453125,
      "loss_num": 0.019775390625,
      "loss_xval": 0.88671875,
      "num_input_tokens_seen": 170955608,
      "step": 3052
    },
    {
      "epoch": 6.799554565701559,
      "grad_norm": 18.317108154296875,
      "learning_rate": 1e-06,
      "loss": 0.5841,
      "num_input_tokens_seen": 171010808,
      "step": 3053
    },
    {
      "epoch": 6.799554565701559,
      "loss": 0.6397278308868408,
      "loss_ce": 0.00020147013128735125,
      "loss_iou": 0.29296875,
      "loss_num": 0.0111083984375,
      "loss_xval": 0.640625,
      "num_input_tokens_seen": 171010808,
      "step": 3053
    },
    {
      "epoch": 6.801781737193764,
      "grad_norm": 17.737529754638672,
      "learning_rate": 1e-06,
      "loss": 0.8411,
      "num_input_tokens_seen": 171066148,
      "step": 3054
    },
    {
      "epoch": 6.801781737193764,
      "loss": 0.8929228186607361,
      "loss_ce": 0.0003446534974500537,
      "loss_iou": 0.375,
      "loss_num": 0.0284423828125,
      "loss_xval": 0.890625,
      "num_input_tokens_seen": 171066148,
      "step": 3054
    },
    {
      "epoch": 6.804008908685969,
      "grad_norm": 18.264446258544922,
      "learning_rate": 1e-06,
      "loss": 0.8505,
      "num_input_tokens_seen": 171121628,
      "step": 3055
    },
    {
      "epoch": 6.804008908685969,
      "loss": 0.6899810433387756,
      "loss_ce": 0.00028378370916470885,
      "loss_iou": 0.296875,
      "loss_num": 0.019287109375,
      "loss_xval": 0.69140625,
      "num_input_tokens_seen": 171121628,
      "step": 3055
    },
    {
      "epoch": 6.806236080178174,
      "grad_norm": 16.88157844543457,
      "learning_rate": 1e-06,
      "loss": 0.5521,
      "num_input_tokens_seen": 171177668,
      "step": 3056
    },
    {
      "epoch": 6.806236080178174,
      "loss": 0.5917320251464844,
      "loss_ce": 0.00017928854504134506,
      "loss_iou": 0.263671875,
      "loss_num": 0.0130615234375,
      "loss_xval": 0.58984375,
      "num_input_tokens_seen": 171177668,
      "step": 3056
    },
    {
      "epoch": 6.8084632516703785,
      "grad_norm": 19.890636444091797,
      "learning_rate": 1e-06,
      "loss": 0.7256,
      "num_input_tokens_seen": 171234924,
      "step": 3057
    },
    {
      "epoch": 6.8084632516703785,
      "loss": 0.9203733205795288,
      "loss_ce": 0.0002072805364150554,
      "loss_iou": 0.35546875,
      "loss_num": 0.0419921875,
      "loss_xval": 0.921875,
      "num_input_tokens_seen": 171234924,
      "step": 3057
    },
    {
      "epoch": 6.810690423162583,
      "grad_norm": 24.549684524536133,
      "learning_rate": 1e-06,
      "loss": 0.6066,
      "num_input_tokens_seen": 171289624,
      "step": 3058
    },
    {
      "epoch": 6.810690423162583,
      "loss": 0.6476849317550659,
      "loss_ce": 0.00022400161833502352,
      "loss_iou": 0.28125,
      "loss_num": 0.0167236328125,
      "loss_xval": 0.6484375,
      "num_input_tokens_seen": 171289624,
      "step": 3058
    },
    {
      "epoch": 6.812917594654788,
      "grad_norm": 29.96609878540039,
      "learning_rate": 1e-06,
      "loss": 0.544,
      "num_input_tokens_seen": 171345904,
      "step": 3059
    },
    {
      "epoch": 6.812917594654788,
      "loss": 0.37130942940711975,
      "loss_ce": 0.00021568889496847987,
      "loss_iou": 0.150390625,
      "loss_num": 0.0137939453125,
      "loss_xval": 0.37109375,
      "num_input_tokens_seen": 171345904,
      "step": 3059
    },
    {
      "epoch": 6.815144766146993,
      "grad_norm": 17.452735900878906,
      "learning_rate": 1e-06,
      "loss": 0.6359,
      "num_input_tokens_seen": 171403776,
      "step": 3060
    },
    {
      "epoch": 6.815144766146993,
      "loss": 0.5905120372772217,
      "loss_ce": 0.00018000410636886954,
      "loss_iou": 0.24609375,
      "loss_num": 0.019775390625,
      "loss_xval": 0.58984375,
      "num_input_tokens_seen": 171403776,
      "step": 3060
    },
    {
      "epoch": 6.817371937639198,
      "grad_norm": 24.2515926361084,
      "learning_rate": 1e-06,
      "loss": 0.6321,
      "num_input_tokens_seen": 171459812,
      "step": 3061
    },
    {
      "epoch": 6.817371937639198,
      "loss": 0.6198359131813049,
      "loss_ce": 0.00020700221648439765,
      "loss_iou": 0.27734375,
      "loss_num": 0.01318359375,
      "loss_xval": 0.62109375,
      "num_input_tokens_seen": 171459812,
      "step": 3061
    },
    {
      "epoch": 6.819599109131403,
      "grad_norm": 26.85552215576172,
      "learning_rate": 1e-06,
      "loss": 0.583,
      "num_input_tokens_seen": 171517492,
      "step": 3062
    },
    {
      "epoch": 6.819599109131403,
      "loss": 0.5180150270462036,
      "loss_ce": 0.0001927339908434078,
      "loss_iou": 0.1943359375,
      "loss_num": 0.02587890625,
      "loss_xval": 0.51953125,
      "num_input_tokens_seen": 171517492,
      "step": 3062
    },
    {
      "epoch": 6.821826280623608,
      "grad_norm": 17.88697624206543,
      "learning_rate": 1e-06,
      "loss": 0.587,
      "num_input_tokens_seen": 171572508,
      "step": 3063
    },
    {
      "epoch": 6.821826280623608,
      "loss": 0.6976819634437561,
      "loss_ce": 0.00017221369489561766,
      "loss_iou": 0.294921875,
      "loss_num": 0.021728515625,
      "loss_xval": 0.69921875,
      "num_input_tokens_seen": 171572508,
      "step": 3063
    },
    {
      "epoch": 6.8240534521158125,
      "grad_norm": 18.11661148071289,
      "learning_rate": 1e-06,
      "loss": 0.5901,
      "num_input_tokens_seen": 171630664,
      "step": 3064
    },
    {
      "epoch": 6.8240534521158125,
      "loss": 0.6340024471282959,
      "loss_ce": 0.00021340540843084455,
      "loss_iou": 0.2734375,
      "loss_num": 0.017578125,
      "loss_xval": 0.6328125,
      "num_input_tokens_seen": 171630664,
      "step": 3064
    },
    {
      "epoch": 6.826280623608017,
      "grad_norm": 15.912714004516602,
      "learning_rate": 1e-06,
      "loss": 0.7199,
      "num_input_tokens_seen": 171688464,
      "step": 3065
    },
    {
      "epoch": 6.826280623608017,
      "loss": 0.7723828554153442,
      "loss_ce": 0.0002881219261325896,
      "loss_iou": 0.34765625,
      "loss_num": 0.015625,
      "loss_xval": 0.7734375,
      "num_input_tokens_seen": 171688464,
      "step": 3065
    },
    {
      "epoch": 6.828507795100223,
      "grad_norm": 104.20018768310547,
      "learning_rate": 1e-06,
      "loss": 0.829,
      "num_input_tokens_seen": 171743324,
      "step": 3066
    },
    {
      "epoch": 6.828507795100223,
      "loss": 1.210741400718689,
      "loss_ce": 0.0007805360364727676,
      "loss_iou": 0.51953125,
      "loss_num": 0.033935546875,
      "loss_xval": 1.2109375,
      "num_input_tokens_seen": 171743324,
      "step": 3066
    },
    {
      "epoch": 6.830734966592428,
      "grad_norm": 16.49806785583496,
      "learning_rate": 1e-06,
      "loss": 0.6798,
      "num_input_tokens_seen": 171799732,
      "step": 3067
    },
    {
      "epoch": 6.830734966592428,
      "loss": 0.5802951455116272,
      "loss_ce": 0.00021701655350625515,
      "loss_iou": 0.2294921875,
      "loss_num": 0.024169921875,
      "loss_xval": 0.578125,
      "num_input_tokens_seen": 171799732,
      "step": 3067
    },
    {
      "epoch": 6.832962138084633,
      "grad_norm": 20.636322021484375,
      "learning_rate": 1e-06,
      "loss": 0.6688,
      "num_input_tokens_seen": 171855072,
      "step": 3068
    },
    {
      "epoch": 6.832962138084633,
      "loss": 0.7048916816711426,
      "loss_ce": 0.00017975937225855887,
      "loss_iou": 0.310546875,
      "loss_num": 0.016357421875,
      "loss_xval": 0.703125,
      "num_input_tokens_seen": 171855072,
      "step": 3068
    },
    {
      "epoch": 6.835189309576838,
      "grad_norm": 14.782151222229004,
      "learning_rate": 1e-06,
      "loss": 0.7034,
      "num_input_tokens_seen": 171912412,
      "step": 3069
    },
    {
      "epoch": 6.835189309576838,
      "loss": 0.6367747187614441,
      "loss_ce": 0.0015207845717668533,
      "loss_iou": 0.2734375,
      "loss_num": 0.0177001953125,
      "loss_xval": 0.63671875,
      "num_input_tokens_seen": 171912412,
      "step": 3069
    },
    {
      "epoch": 6.8374164810690425,
      "grad_norm": 22.69580078125,
      "learning_rate": 1e-06,
      "loss": 0.6691,
      "num_input_tokens_seen": 171967604,
      "step": 3070
    },
    {
      "epoch": 6.8374164810690425,
      "loss": 0.6948800086975098,
      "loss_ce": 0.00017786939861252904,
      "loss_iou": 0.318359375,
      "loss_num": 0.01171875,
      "loss_xval": 0.6953125,
      "num_input_tokens_seen": 171967604,
      "step": 3070
    },
    {
      "epoch": 6.839643652561247,
      "grad_norm": 16.151691436767578,
      "learning_rate": 1e-06,
      "loss": 0.7264,
      "num_input_tokens_seen": 172024480,
      "step": 3071
    },
    {
      "epoch": 6.839643652561247,
      "loss": 0.7126206159591675,
      "loss_ce": 0.0002182956231990829,
      "loss_iou": 0.31640625,
      "loss_num": 0.0157470703125,
      "loss_xval": 0.7109375,
      "num_input_tokens_seen": 172024480,
      "step": 3071
    },
    {
      "epoch": 6.841870824053452,
      "grad_norm": 23.6954402923584,
      "learning_rate": 1e-06,
      "loss": 0.5101,
      "num_input_tokens_seen": 172081368,
      "step": 3072
    },
    {
      "epoch": 6.841870824053452,
      "loss": 0.41563284397125244,
      "loss_ce": 0.00034961808705702424,
      "loss_iou": 0.171875,
      "loss_num": 0.014404296875,
      "loss_xval": 0.416015625,
      "num_input_tokens_seen": 172081368,
      "step": 3072
    },
    {
      "epoch": 6.844097995545657,
      "grad_norm": 14.069080352783203,
      "learning_rate": 1e-06,
      "loss": 0.4567,
      "num_input_tokens_seen": 172134880,
      "step": 3073
    },
    {
      "epoch": 6.844097995545657,
      "loss": 0.4463517367839813,
      "loss_ce": 0.00018475353135727346,
      "loss_iou": 0.18359375,
      "loss_num": 0.0159912109375,
      "loss_xval": 0.4453125,
      "num_input_tokens_seen": 172134880,
      "step": 3073
    },
    {
      "epoch": 6.846325167037862,
      "grad_norm": 16.129291534423828,
      "learning_rate": 1e-06,
      "loss": 0.7115,
      "num_input_tokens_seen": 172191948,
      "step": 3074
    },
    {
      "epoch": 6.846325167037862,
      "loss": 1.0690743923187256,
      "loss_ce": 0.00022675658692605793,
      "loss_iou": 0.486328125,
      "loss_num": 0.0194091796875,
      "loss_xval": 1.0703125,
      "num_input_tokens_seen": 172191948,
      "step": 3074
    },
    {
      "epoch": 6.848552338530067,
      "grad_norm": 15.745563507080078,
      "learning_rate": 1e-06,
      "loss": 0.4251,
      "num_input_tokens_seen": 172250304,
      "step": 3075
    },
    {
      "epoch": 6.848552338530067,
      "loss": 0.40329307317733765,
      "loss_ce": 0.0002169150102417916,
      "loss_iou": 0.1806640625,
      "loss_num": 0.00848388671875,
      "loss_xval": 0.40234375,
      "num_input_tokens_seen": 172250304,
      "step": 3075
    },
    {
      "epoch": 6.850779510022272,
      "grad_norm": 23.94293975830078,
      "learning_rate": 1e-06,
      "loss": 0.7518,
      "num_input_tokens_seen": 172304700,
      "step": 3076
    },
    {
      "epoch": 6.850779510022272,
      "loss": 0.6821050643920898,
      "loss_ce": 0.00022032001288607717,
      "loss_iou": 0.2890625,
      "loss_num": 0.0205078125,
      "loss_xval": 0.68359375,
      "num_input_tokens_seen": 172304700,
      "step": 3076
    },
    {
      "epoch": 6.853006681514477,
      "grad_norm": 22.777400970458984,
      "learning_rate": 1e-06,
      "loss": 0.6088,
      "num_input_tokens_seen": 172360340,
      "step": 3077
    },
    {
      "epoch": 6.853006681514477,
      "loss": 0.36834925413131714,
      "loss_ce": 0.00030725146643817425,
      "loss_iou": 0.1416015625,
      "loss_num": 0.016845703125,
      "loss_xval": 0.3671875,
      "num_input_tokens_seen": 172360340,
      "step": 3077
    },
    {
      "epoch": 6.855233853006681,
      "grad_norm": 23.746376037597656,
      "learning_rate": 1e-06,
      "loss": 0.592,
      "num_input_tokens_seen": 172418912,
      "step": 3078
    },
    {
      "epoch": 6.855233853006681,
      "loss": 0.6163959503173828,
      "loss_ce": 0.0001850596017902717,
      "loss_iou": 0.265625,
      "loss_num": 0.0166015625,
      "loss_xval": 0.6171875,
      "num_input_tokens_seen": 172418912,
      "step": 3078
    },
    {
      "epoch": 6.857461024498886,
      "grad_norm": 35.40282440185547,
      "learning_rate": 1e-06,
      "loss": 0.8302,
      "num_input_tokens_seen": 172477360,
      "step": 3079
    },
    {
      "epoch": 6.857461024498886,
      "loss": 0.8708090782165527,
      "loss_ce": 0.00020358533947728574,
      "loss_iou": 0.3828125,
      "loss_num": 0.020751953125,
      "loss_xval": 0.87109375,
      "num_input_tokens_seen": 172477360,
      "step": 3079
    },
    {
      "epoch": 6.859688195991091,
      "grad_norm": 15.791609764099121,
      "learning_rate": 1e-06,
      "loss": 0.5824,
      "num_input_tokens_seen": 172533360,
      "step": 3080
    },
    {
      "epoch": 6.859688195991091,
      "loss": 0.8971776962280273,
      "loss_ce": 0.00020497874356806278,
      "loss_iou": 0.388671875,
      "loss_num": 0.0244140625,
      "loss_xval": 0.8984375,
      "num_input_tokens_seen": 172533360,
      "step": 3080
    },
    {
      "epoch": 6.861915367483296,
      "grad_norm": 16.064542770385742,
      "learning_rate": 1e-06,
      "loss": 0.6309,
      "num_input_tokens_seen": 172586820,
      "step": 3081
    },
    {
      "epoch": 6.861915367483296,
      "loss": 0.5737239122390747,
      "loss_ce": 0.0002375697804382071,
      "loss_iou": 0.267578125,
      "loss_num": 0.007598876953125,
      "loss_xval": 0.57421875,
      "num_input_tokens_seen": 172586820,
      "step": 3081
    },
    {
      "epoch": 6.864142538975501,
      "grad_norm": 23.688007354736328,
      "learning_rate": 1e-06,
      "loss": 0.7019,
      "num_input_tokens_seen": 172644792,
      "step": 3082
    },
    {
      "epoch": 6.864142538975501,
      "loss": 0.7206803560256958,
      "loss_ce": 0.00022136216284707189,
      "loss_iou": 0.322265625,
      "loss_num": 0.01519775390625,
      "loss_xval": 0.71875,
      "num_input_tokens_seen": 172644792,
      "step": 3082
    },
    {
      "epoch": 6.866369710467706,
      "grad_norm": 16.832897186279297,
      "learning_rate": 1e-06,
      "loss": 0.4801,
      "num_input_tokens_seen": 172703448,
      "step": 3083
    },
    {
      "epoch": 6.866369710467706,
      "loss": 0.4925019443035126,
      "loss_ce": 0.00019236501248087734,
      "loss_iou": 0.2158203125,
      "loss_num": 0.01202392578125,
      "loss_xval": 0.4921875,
      "num_input_tokens_seen": 172703448,
      "step": 3083
    },
    {
      "epoch": 6.868596881959911,
      "grad_norm": 19.060598373413086,
      "learning_rate": 1e-06,
      "loss": 0.6689,
      "num_input_tokens_seen": 172759888,
      "step": 3084
    },
    {
      "epoch": 6.868596881959911,
      "loss": 0.6370642185211182,
      "loss_ce": 0.0003454496618360281,
      "loss_iou": 0.275390625,
      "loss_num": 0.01708984375,
      "loss_xval": 0.63671875,
      "num_input_tokens_seen": 172759888,
      "step": 3084
    },
    {
      "epoch": 6.870824053452115,
      "grad_norm": 16.259754180908203,
      "learning_rate": 1e-06,
      "loss": 0.4995,
      "num_input_tokens_seen": 172815816,
      "step": 3085
    },
    {
      "epoch": 6.870824053452115,
      "loss": 0.529482364654541,
      "loss_ce": 0.00018545490456745028,
      "loss_iou": 0.228515625,
      "loss_num": 0.01434326171875,
      "loss_xval": 0.53125,
      "num_input_tokens_seen": 172815816,
      "step": 3085
    },
    {
      "epoch": 6.873051224944321,
      "grad_norm": 21.378643035888672,
      "learning_rate": 1e-06,
      "loss": 0.6019,
      "num_input_tokens_seen": 172871560,
      "step": 3086
    },
    {
      "epoch": 6.873051224944321,
      "loss": 0.6235019564628601,
      "loss_ce": 0.00021095495321787894,
      "loss_iou": 0.255859375,
      "loss_num": 0.02294921875,
      "loss_xval": 0.625,
      "num_input_tokens_seen": 172871560,
      "step": 3086
    },
    {
      "epoch": 6.875278396436526,
      "grad_norm": 36.62589645385742,
      "learning_rate": 1e-06,
      "loss": 0.7479,
      "num_input_tokens_seen": 172926412,
      "step": 3087
    },
    {
      "epoch": 6.875278396436526,
      "loss": 0.9482687711715698,
      "loss_ce": 0.00027069164207205176,
      "loss_iou": 0.412109375,
      "loss_num": 0.024658203125,
      "loss_xval": 0.94921875,
      "num_input_tokens_seen": 172926412,
      "step": 3087
    },
    {
      "epoch": 6.877505567928731,
      "grad_norm": 21.01188087463379,
      "learning_rate": 1e-06,
      "loss": 0.548,
      "num_input_tokens_seen": 172984452,
      "step": 3088
    },
    {
      "epoch": 6.877505567928731,
      "loss": 0.5002177357673645,
      "loss_ce": 0.00021776201901957393,
      "loss_iou": 0.2216796875,
      "loss_num": 0.01129150390625,
      "loss_xval": 0.5,
      "num_input_tokens_seen": 172984452,
      "step": 3088
    },
    {
      "epoch": 6.879732739420936,
      "grad_norm": 16.86750030517578,
      "learning_rate": 1e-06,
      "loss": 0.6602,
      "num_input_tokens_seen": 173040592,
      "step": 3089
    },
    {
      "epoch": 6.879732739420936,
      "loss": 0.757439911365509,
      "loss_ce": 0.0003598659241106361,
      "loss_iou": 0.314453125,
      "loss_num": 0.02587890625,
      "loss_xval": 0.7578125,
      "num_input_tokens_seen": 173040592,
      "step": 3089
    },
    {
      "epoch": 6.881959910913141,
      "grad_norm": 21.569618225097656,
      "learning_rate": 1e-06,
      "loss": 0.7411,
      "num_input_tokens_seen": 173097352,
      "step": 3090
    },
    {
      "epoch": 6.881959910913141,
      "loss": 0.8854343891143799,
      "loss_ce": 0.0001805470819817856,
      "loss_iou": 0.376953125,
      "loss_num": 0.0263671875,
      "loss_xval": 0.88671875,
      "num_input_tokens_seen": 173097352,
      "step": 3090
    },
    {
      "epoch": 6.8841870824053455,
      "grad_norm": 49.518394470214844,
      "learning_rate": 1e-06,
      "loss": 0.5467,
      "num_input_tokens_seen": 173156100,
      "step": 3091
    },
    {
      "epoch": 6.8841870824053455,
      "loss": 0.465660035610199,
      "loss_ce": 0.00020591478096321225,
      "loss_iou": 0.2041015625,
      "loss_num": 0.01153564453125,
      "loss_xval": 0.46484375,
      "num_input_tokens_seen": 173156100,
      "step": 3091
    },
    {
      "epoch": 6.88641425389755,
      "grad_norm": 75.19503784179688,
      "learning_rate": 1e-06,
      "loss": 0.7018,
      "num_input_tokens_seen": 173212776,
      "step": 3092
    },
    {
      "epoch": 6.88641425389755,
      "loss": 0.8085219860076904,
      "loss_ce": 0.00029449607245624065,
      "loss_iou": 0.34375,
      "loss_num": 0.024169921875,
      "loss_xval": 0.80859375,
      "num_input_tokens_seen": 173212776,
      "step": 3092
    },
    {
      "epoch": 6.888641425389755,
      "grad_norm": 18.865760803222656,
      "learning_rate": 1e-06,
      "loss": 0.7596,
      "num_input_tokens_seen": 173267624,
      "step": 3093
    },
    {
      "epoch": 6.888641425389755,
      "loss": 0.8561751246452332,
      "loss_ce": 0.00021812312479596585,
      "loss_iou": 0.3671875,
      "loss_num": 0.024658203125,
      "loss_xval": 0.85546875,
      "num_input_tokens_seen": 173267624,
      "step": 3093
    },
    {
      "epoch": 6.89086859688196,
      "grad_norm": 26.241153717041016,
      "learning_rate": 1e-06,
      "loss": 0.6993,
      "num_input_tokens_seen": 173324276,
      "step": 3094
    },
    {
      "epoch": 6.89086859688196,
      "loss": 0.7165099382400513,
      "loss_ce": 0.00020136788953095675,
      "loss_iou": 0.294921875,
      "loss_num": 0.025634765625,
      "loss_xval": 0.71484375,
      "num_input_tokens_seen": 173324276,
      "step": 3094
    },
    {
      "epoch": 6.893095768374165,
      "grad_norm": 17.569229125976562,
      "learning_rate": 1e-06,
      "loss": 0.5308,
      "num_input_tokens_seen": 173382708,
      "step": 3095
    },
    {
      "epoch": 6.893095768374165,
      "loss": 0.4684421718120575,
      "loss_ce": 0.0001804671046556905,
      "loss_iou": 0.1953125,
      "loss_num": 0.015625,
      "loss_xval": 0.46875,
      "num_input_tokens_seen": 173382708,
      "step": 3095
    },
    {
      "epoch": 6.89532293986637,
      "grad_norm": 13.874095916748047,
      "learning_rate": 1e-06,
      "loss": 0.7751,
      "num_input_tokens_seen": 173438884,
      "step": 3096
    },
    {
      "epoch": 6.89532293986637,
      "loss": 0.9164137840270996,
      "loss_ce": 0.00021500332513824105,
      "loss_iou": 0.345703125,
      "loss_num": 0.045166015625,
      "loss_xval": 0.91796875,
      "num_input_tokens_seen": 173438884,
      "step": 3096
    },
    {
      "epoch": 6.897550111358575,
      "grad_norm": 19.509395599365234,
      "learning_rate": 1e-06,
      "loss": 0.523,
      "num_input_tokens_seen": 173493924,
      "step": 3097
    },
    {
      "epoch": 6.897550111358575,
      "loss": 0.6444209814071655,
      "loss_ce": 0.000255966791883111,
      "loss_iou": 0.28125,
      "loss_num": 0.0167236328125,
      "loss_xval": 0.64453125,
      "num_input_tokens_seen": 173493924,
      "step": 3097
    },
    {
      "epoch": 6.8997772828507795,
      "grad_norm": 19.37684440612793,
      "learning_rate": 1e-06,
      "loss": 0.6321,
      "num_input_tokens_seen": 173549456,
      "step": 3098
    },
    {
      "epoch": 6.8997772828507795,
      "loss": 0.6488611698150635,
      "loss_ce": 0.00017949687025975436,
      "loss_iou": 0.271484375,
      "loss_num": 0.021484375,
      "loss_xval": 0.6484375,
      "num_input_tokens_seen": 173549456,
      "step": 3098
    },
    {
      "epoch": 6.902004454342984,
      "grad_norm": 17.55590057373047,
      "learning_rate": 1e-06,
      "loss": 0.5922,
      "num_input_tokens_seen": 173608184,
      "step": 3099
    },
    {
      "epoch": 6.902004454342984,
      "loss": 0.43987372517585754,
      "loss_ce": 0.00017645125626586378,
      "loss_iou": 0.19140625,
      "loss_num": 0.01141357421875,
      "loss_xval": 0.439453125,
      "num_input_tokens_seen": 173608184,
      "step": 3099
    },
    {
      "epoch": 6.904231625835189,
      "grad_norm": 17.88399887084961,
      "learning_rate": 1e-06,
      "loss": 0.5861,
      "num_input_tokens_seen": 173663616,
      "step": 3100
    },
    {
      "epoch": 6.904231625835189,
      "loss": 0.5784924030303955,
      "loss_ce": 0.00018430282943882048,
      "loss_iou": 0.2470703125,
      "loss_num": 0.016845703125,
      "loss_xval": 0.578125,
      "num_input_tokens_seen": 173663616,
      "step": 3100
    },
    {
      "epoch": 6.906458797327394,
      "grad_norm": 16.59260368347168,
      "learning_rate": 1e-06,
      "loss": 0.5021,
      "num_input_tokens_seen": 173716408,
      "step": 3101
    },
    {
      "epoch": 6.906458797327394,
      "loss": 0.5853925943374634,
      "loss_ce": 0.0001875399611890316,
      "loss_iou": 0.2392578125,
      "loss_num": 0.0213623046875,
      "loss_xval": 0.5859375,
      "num_input_tokens_seen": 173716408,
      "step": 3101
    },
    {
      "epoch": 6.908685968819599,
      "grad_norm": 17.521724700927734,
      "learning_rate": 1e-06,
      "loss": 0.6656,
      "num_input_tokens_seen": 173772800,
      "step": 3102
    },
    {
      "epoch": 6.908685968819599,
      "loss": 0.6075997948646545,
      "loss_ce": 0.0001779411395546049,
      "loss_iou": 0.2392578125,
      "loss_num": 0.025634765625,
      "loss_xval": 0.609375,
      "num_input_tokens_seen": 173772800,
      "step": 3102
    },
    {
      "epoch": 6.910913140311804,
      "grad_norm": 28.117921829223633,
      "learning_rate": 1e-06,
      "loss": 0.6282,
      "num_input_tokens_seen": 173829652,
      "step": 3103
    },
    {
      "epoch": 6.910913140311804,
      "loss": 0.5315419435501099,
      "loss_ce": 0.00016988006245810539,
      "loss_iou": 0.2294921875,
      "loss_num": 0.01446533203125,
      "loss_xval": 0.53125,
      "num_input_tokens_seen": 173829652,
      "step": 3103
    },
    {
      "epoch": 6.913140311804009,
      "grad_norm": 22.19889259338379,
      "learning_rate": 1e-06,
      "loss": 0.7273,
      "num_input_tokens_seen": 173884456,
      "step": 3104
    },
    {
      "epoch": 6.913140311804009,
      "loss": 0.7211758494377136,
      "loss_ce": 0.0002286143571836874,
      "loss_iou": 0.30859375,
      "loss_num": 0.0208740234375,
      "loss_xval": 0.72265625,
      "num_input_tokens_seen": 173884456,
      "step": 3104
    },
    {
      "epoch": 6.9153674832962135,
      "grad_norm": 11.665279388427734,
      "learning_rate": 1e-06,
      "loss": 0.62,
      "num_input_tokens_seen": 173939820,
      "step": 3105
    },
    {
      "epoch": 6.9153674832962135,
      "loss": 0.4306148886680603,
      "loss_ce": 0.0001950013975147158,
      "loss_iou": 0.1943359375,
      "loss_num": 0.00836181640625,
      "loss_xval": 0.4296875,
      "num_input_tokens_seen": 173939820,
      "step": 3105
    },
    {
      "epoch": 6.917594654788418,
      "grad_norm": 20.158309936523438,
      "learning_rate": 1e-06,
      "loss": 0.5542,
      "num_input_tokens_seen": 173995772,
      "step": 3106
    },
    {
      "epoch": 6.917594654788418,
      "loss": 0.44402259588241577,
      "loss_ce": 0.00017491859034635127,
      "loss_iou": 0.1796875,
      "loss_num": 0.0169677734375,
      "loss_xval": 0.443359375,
      "num_input_tokens_seen": 173995772,
      "step": 3106
    },
    {
      "epoch": 6.919821826280623,
      "grad_norm": 18.162250518798828,
      "learning_rate": 1e-06,
      "loss": 0.482,
      "num_input_tokens_seen": 174053676,
      "step": 3107
    },
    {
      "epoch": 6.919821826280623,
      "loss": 0.42890018224716187,
      "loss_ce": 0.0001892539585242048,
      "loss_iou": 0.17578125,
      "loss_num": 0.015380859375,
      "loss_xval": 0.4296875,
      "num_input_tokens_seen": 174053676,
      "step": 3107
    },
    {
      "epoch": 6.922048997772828,
      "grad_norm": 48.48846435546875,
      "learning_rate": 1e-06,
      "loss": 0.7151,
      "num_input_tokens_seen": 174112208,
      "step": 3108
    },
    {
      "epoch": 6.922048997772828,
      "loss": 0.827580451965332,
      "loss_ce": 0.00018786173313856125,
      "loss_iou": 0.373046875,
      "loss_num": 0.0164794921875,
      "loss_xval": 0.828125,
      "num_input_tokens_seen": 174112208,
      "step": 3108
    },
    {
      "epoch": 6.924276169265033,
      "grad_norm": 25.14827537536621,
      "learning_rate": 1e-06,
      "loss": 0.7384,
      "num_input_tokens_seen": 174168272,
      "step": 3109
    },
    {
      "epoch": 6.924276169265033,
      "loss": 0.7882997393608093,
      "loss_ce": 0.00021379378449637443,
      "loss_iou": 0.35546875,
      "loss_num": 0.0157470703125,
      "loss_xval": 0.7890625,
      "num_input_tokens_seen": 174168272,
      "step": 3109
    },
    {
      "epoch": 6.926503340757238,
      "grad_norm": 18.243032455444336,
      "learning_rate": 1e-06,
      "loss": 0.7986,
      "num_input_tokens_seen": 174226660,
      "step": 3110
    },
    {
      "epoch": 6.926503340757238,
      "loss": 0.8874292969703674,
      "loss_ce": 0.00022228219313547015,
      "loss_iou": 0.359375,
      "loss_num": 0.033447265625,
      "loss_xval": 0.88671875,
      "num_input_tokens_seen": 174226660,
      "step": 3110
    },
    {
      "epoch": 6.928730512249444,
      "grad_norm": 14.311541557312012,
      "learning_rate": 1e-06,
      "loss": 0.6963,
      "num_input_tokens_seen": 174279336,
      "step": 3111
    },
    {
      "epoch": 6.928730512249444,
      "loss": 0.7939878702163696,
      "loss_ce": 0.00028668707818724215,
      "loss_iou": 0.318359375,
      "loss_num": 0.031494140625,
      "loss_xval": 0.79296875,
      "num_input_tokens_seen": 174279336,
      "step": 3111
    },
    {
      "epoch": 6.9309576837416484,
      "grad_norm": 20.134523391723633,
      "learning_rate": 1e-06,
      "loss": 0.7007,
      "num_input_tokens_seen": 174335124,
      "step": 3112
    },
    {
      "epoch": 6.9309576837416484,
      "loss": 0.7140600681304932,
      "loss_ce": 0.00019287687609903514,
      "loss_iou": 0.302734375,
      "loss_num": 0.021728515625,
      "loss_xval": 0.71484375,
      "num_input_tokens_seen": 174335124,
      "step": 3112
    },
    {
      "epoch": 6.933184855233853,
      "grad_norm": 18.099767684936523,
      "learning_rate": 1e-06,
      "loss": 0.8233,
      "num_input_tokens_seen": 174391556,
      "step": 3113
    },
    {
      "epoch": 6.933184855233853,
      "loss": 0.7465390563011169,
      "loss_ce": 0.00020114146173000336,
      "loss_iou": 0.32421875,
      "loss_num": 0.02001953125,
      "loss_xval": 0.74609375,
      "num_input_tokens_seen": 174391556,
      "step": 3113
    },
    {
      "epoch": 6.935412026726058,
      "grad_norm": 22.79920768737793,
      "learning_rate": 1e-06,
      "loss": 0.6987,
      "num_input_tokens_seen": 174446136,
      "step": 3114
    },
    {
      "epoch": 6.935412026726058,
      "loss": 0.6254494190216064,
      "loss_ce": 0.00020533311180770397,
      "loss_iou": 0.2490234375,
      "loss_num": 0.0252685546875,
      "loss_xval": 0.625,
      "num_input_tokens_seen": 174446136,
      "step": 3114
    },
    {
      "epoch": 6.937639198218263,
      "grad_norm": 25.27695655822754,
      "learning_rate": 1e-06,
      "loss": 0.6646,
      "num_input_tokens_seen": 174500684,
      "step": 3115
    },
    {
      "epoch": 6.937639198218263,
      "loss": 0.7533596754074097,
      "loss_ce": 0.000185904442332685,
      "loss_iou": 0.328125,
      "loss_num": 0.0191650390625,
      "loss_xval": 0.75390625,
      "num_input_tokens_seen": 174500684,
      "step": 3115
    },
    {
      "epoch": 6.939866369710468,
      "grad_norm": 15.809171676635742,
      "learning_rate": 1e-06,
      "loss": 0.4947,
      "num_input_tokens_seen": 174554360,
      "step": 3116
    },
    {
      "epoch": 6.939866369710468,
      "loss": 0.6411548852920532,
      "loss_ce": 0.0002857603249140084,
      "loss_iou": 0.2734375,
      "loss_num": 0.0184326171875,
      "loss_xval": 0.640625,
      "num_input_tokens_seen": 174554360,
      "step": 3116
    },
    {
      "epoch": 6.942093541202673,
      "grad_norm": 21.57350730895996,
      "learning_rate": 1e-06,
      "loss": 0.5634,
      "num_input_tokens_seen": 174608692,
      "step": 3117
    },
    {
      "epoch": 6.942093541202673,
      "loss": 0.550343930721283,
      "loss_ce": 0.00017302096239291131,
      "loss_iou": 0.2421875,
      "loss_num": 0.0130615234375,
      "loss_xval": 0.55078125,
      "num_input_tokens_seen": 174608692,
      "step": 3117
    },
    {
      "epoch": 6.944320712694878,
      "grad_norm": 16.204471588134766,
      "learning_rate": 1e-06,
      "loss": 0.7834,
      "num_input_tokens_seen": 174665252,
      "step": 3118
    },
    {
      "epoch": 6.944320712694878,
      "loss": 0.9167299270629883,
      "loss_ce": 0.0008363361121155322,
      "loss_iou": 0.41015625,
      "loss_num": 0.0189208984375,
      "loss_xval": 0.9140625,
      "num_input_tokens_seen": 174665252,
      "step": 3118
    },
    {
      "epoch": 6.9465478841870825,
      "grad_norm": 17.026905059814453,
      "learning_rate": 1e-06,
      "loss": 0.5704,
      "num_input_tokens_seen": 174719864,
      "step": 3119
    },
    {
      "epoch": 6.9465478841870825,
      "loss": 0.657658576965332,
      "loss_ce": 0.0001878339098766446,
      "loss_iou": 0.28125,
      "loss_num": 0.0191650390625,
      "loss_xval": 0.65625,
      "num_input_tokens_seen": 174719864,
      "step": 3119
    },
    {
      "epoch": 6.948775055679287,
      "grad_norm": 22.539409637451172,
      "learning_rate": 1e-06,
      "loss": 0.75,
      "num_input_tokens_seen": 174773804,
      "step": 3120
    },
    {
      "epoch": 6.948775055679287,
      "loss": 0.5996890068054199,
      "loss_ce": 0.00020167973707430065,
      "loss_iou": 0.26171875,
      "loss_num": 0.014892578125,
      "loss_xval": 0.59765625,
      "num_input_tokens_seen": 174773804,
      "step": 3120
    },
    {
      "epoch": 6.951002227171492,
      "grad_norm": 22.28173828125,
      "learning_rate": 1e-06,
      "loss": 0.7117,
      "num_input_tokens_seen": 174828396,
      "step": 3121
    },
    {
      "epoch": 6.951002227171492,
      "loss": 0.5873724818229675,
      "loss_ce": 0.00021425656450446695,
      "loss_iou": 0.25,
      "loss_num": 0.017333984375,
      "loss_xval": 0.5859375,
      "num_input_tokens_seen": 174828396,
      "step": 3121
    },
    {
      "epoch": 6.953229398663697,
      "grad_norm": 54.56577682495117,
      "learning_rate": 1e-06,
      "loss": 0.4691,
      "num_input_tokens_seen": 174884540,
      "step": 3122
    },
    {
      "epoch": 6.953229398663697,
      "loss": 0.5313279628753662,
      "loss_ce": 0.0003220998914912343,
      "loss_iou": 0.2333984375,
      "loss_num": 0.0128173828125,
      "loss_xval": 0.53125,
      "num_input_tokens_seen": 174884540,
      "step": 3122
    },
    {
      "epoch": 6.955456570155902,
      "grad_norm": 33.31269454956055,
      "learning_rate": 1e-06,
      "loss": 0.8895,
      "num_input_tokens_seen": 174939284,
      "step": 3123
    },
    {
      "epoch": 6.955456570155902,
      "loss": 0.7173866033554077,
      "loss_ce": 0.0003456372069194913,
      "loss_iou": 0.2734375,
      "loss_num": 0.03369140625,
      "loss_xval": 0.71875,
      "num_input_tokens_seen": 174939284,
      "step": 3123
    },
    {
      "epoch": 6.957683741648107,
      "grad_norm": 25.145395278930664,
      "learning_rate": 1e-06,
      "loss": 0.5721,
      "num_input_tokens_seen": 174994868,
      "step": 3124
    },
    {
      "epoch": 6.957683741648107,
      "loss": 0.6920799612998962,
      "loss_ce": 0.00018543089390732348,
      "loss_iou": 0.27734375,
      "loss_num": 0.0277099609375,
      "loss_xval": 0.69140625,
      "num_input_tokens_seen": 174994868,
      "step": 3124
    },
    {
      "epoch": 6.959910913140312,
      "grad_norm": 31.5352840423584,
      "learning_rate": 1e-06,
      "loss": 0.6128,
      "num_input_tokens_seen": 175050648,
      "step": 3125
    },
    {
      "epoch": 6.959910913140312,
      "loss": 0.5849337577819824,
      "loss_ce": 0.00021696032490581274,
      "loss_iou": 0.26171875,
      "loss_num": 0.0123291015625,
      "loss_xval": 0.5859375,
      "num_input_tokens_seen": 175050648,
      "step": 3125
    },
    {
      "epoch": 6.9621380846325165,
      "grad_norm": 28.711498260498047,
      "learning_rate": 1e-06,
      "loss": 0.7421,
      "num_input_tokens_seen": 175109124,
      "step": 3126
    },
    {
      "epoch": 6.9621380846325165,
      "loss": 0.5786285400390625,
      "loss_ce": 0.0002593994140625,
      "loss_iou": 0.25390625,
      "loss_num": 0.0142822265625,
      "loss_xval": 0.578125,
      "num_input_tokens_seen": 175109124,
      "step": 3126
    },
    {
      "epoch": 6.964365256124721,
      "grad_norm": 81.01925659179688,
      "learning_rate": 1e-06,
      "loss": 0.6552,
      "num_input_tokens_seen": 175165652,
      "step": 3127
    },
    {
      "epoch": 6.964365256124721,
      "loss": 0.4998403489589691,
      "loss_ce": 0.00020655704429373145,
      "loss_iou": 0.212890625,
      "loss_num": 0.01483154296875,
      "loss_xval": 0.5,
      "num_input_tokens_seen": 175165652,
      "step": 3127
    },
    {
      "epoch": 6.966592427616926,
      "grad_norm": 20.51897621154785,
      "learning_rate": 1e-06,
      "loss": 0.5369,
      "num_input_tokens_seen": 175223100,
      "step": 3128
    },
    {
      "epoch": 6.966592427616926,
      "loss": 0.5199558734893799,
      "loss_ce": 0.00018052573432214558,
      "loss_iou": 0.240234375,
      "loss_num": 0.0081787109375,
      "loss_xval": 0.51953125,
      "num_input_tokens_seen": 175223100,
      "step": 3128
    },
    {
      "epoch": 6.968819599109131,
      "grad_norm": 13.959333419799805,
      "learning_rate": 1e-06,
      "loss": 0.5609,
      "num_input_tokens_seen": 175280568,
      "step": 3129
    },
    {
      "epoch": 6.968819599109131,
      "loss": 0.7189445495605469,
      "loss_ce": 0.0001945712574524805,
      "loss_iou": 0.333984375,
      "loss_num": 0.010009765625,
      "loss_xval": 0.71875,
      "num_input_tokens_seen": 175280568,
      "step": 3129
    },
    {
      "epoch": 6.971046770601336,
      "grad_norm": 22.97583770751953,
      "learning_rate": 1e-06,
      "loss": 1.0313,
      "num_input_tokens_seen": 175337324,
      "step": 3130
    },
    {
      "epoch": 6.971046770601336,
      "loss": 0.8275806903839111,
      "loss_ce": 0.00018814984650816768,
      "loss_iou": 0.359375,
      "loss_num": 0.0213623046875,
      "loss_xval": 0.828125,
      "num_input_tokens_seen": 175337324,
      "step": 3130
    },
    {
      "epoch": 6.973273942093542,
      "grad_norm": 34.06120681762695,
      "learning_rate": 1e-06,
      "loss": 0.8319,
      "num_input_tokens_seen": 175392008,
      "step": 3131
    },
    {
      "epoch": 6.973273942093542,
      "loss": 0.9467490315437317,
      "loss_ce": 0.00021585801732726395,
      "loss_iou": 0.412109375,
      "loss_num": 0.024658203125,
      "loss_xval": 0.9453125,
      "num_input_tokens_seen": 175392008,
      "step": 3131
    },
    {
      "epoch": 6.9755011135857465,
      "grad_norm": 16.659931182861328,
      "learning_rate": 1e-06,
      "loss": 0.5561,
      "num_input_tokens_seen": 175450048,
      "step": 3132
    },
    {
      "epoch": 6.9755011135857465,
      "loss": 0.5903693437576294,
      "loss_ce": 0.0002204178017564118,
      "loss_iou": 0.255859375,
      "loss_num": 0.0159912109375,
      "loss_xval": 0.58984375,
      "num_input_tokens_seen": 175450048,
      "step": 3132
    },
    {
      "epoch": 6.977728285077951,
      "grad_norm": 31.275850296020508,
      "learning_rate": 1e-06,
      "loss": 0.7978,
      "num_input_tokens_seen": 175504996,
      "step": 3133
    },
    {
      "epoch": 6.977728285077951,
      "loss": 1.0408833026885986,
      "loss_ce": 0.0002338237245567143,
      "loss_iou": 0.419921875,
      "loss_num": 0.039794921875,
      "loss_xval": 1.0390625,
      "num_input_tokens_seen": 175504996,
      "step": 3133
    },
    {
      "epoch": 6.979955456570156,
      "grad_norm": 19.884502410888672,
      "learning_rate": 1e-06,
      "loss": 0.5947,
      "num_input_tokens_seen": 175562712,
      "step": 3134
    },
    {
      "epoch": 6.979955456570156,
      "loss": 0.4893288016319275,
      "loss_ce": 0.00019303697627037764,
      "loss_iou": 0.2314453125,
      "loss_num": 0.00555419921875,
      "loss_xval": 0.48828125,
      "num_input_tokens_seen": 175562712,
      "step": 3134
    },
    {
      "epoch": 6.982182628062361,
      "grad_norm": 13.180671691894531,
      "learning_rate": 1e-06,
      "loss": 0.7,
      "num_input_tokens_seen": 175620280,
      "step": 3135
    },
    {
      "epoch": 6.982182628062361,
      "loss": 0.8063207864761353,
      "loss_ce": 0.0004125875420868397,
      "loss_iou": 0.35546875,
      "loss_num": 0.01904296875,
      "loss_xval": 0.8046875,
      "num_input_tokens_seen": 175620280,
      "step": 3135
    },
    {
      "epoch": 6.984409799554566,
      "grad_norm": 21.996858596801758,
      "learning_rate": 1e-06,
      "loss": 0.5783,
      "num_input_tokens_seen": 175678996,
      "step": 3136
    },
    {
      "epoch": 6.984409799554566,
      "loss": 0.7316303253173828,
      "loss_ce": 0.00018497416749596596,
      "loss_iou": 0.302734375,
      "loss_num": 0.025146484375,
      "loss_xval": 0.73046875,
      "num_input_tokens_seen": 175678996,
      "step": 3136
    },
    {
      "epoch": 6.986636971046771,
      "grad_norm": 19.231781005859375,
      "learning_rate": 1e-06,
      "loss": 0.8218,
      "num_input_tokens_seen": 175733860,
      "step": 3137
    },
    {
      "epoch": 6.986636971046771,
      "loss": 0.6332485675811768,
      "loss_ce": 0.00019193578918930143,
      "loss_iou": 0.25,
      "loss_num": 0.0262451171875,
      "loss_xval": 0.6328125,
      "num_input_tokens_seen": 175733860,
      "step": 3137
    },
    {
      "epoch": 6.988864142538976,
      "grad_norm": 11.28582763671875,
      "learning_rate": 1e-06,
      "loss": 0.7356,
      "num_input_tokens_seen": 175792092,
      "step": 3138
    },
    {
      "epoch": 6.988864142538976,
      "loss": 0.7703654766082764,
      "loss_ce": 0.00022392123355530202,
      "loss_iou": 0.3046875,
      "loss_num": 0.032470703125,
      "loss_xval": 0.76953125,
      "num_input_tokens_seen": 175792092,
      "step": 3138
    },
    {
      "epoch": 6.991091314031181,
      "grad_norm": 18.971759796142578,
      "learning_rate": 1e-06,
      "loss": 0.6312,
      "num_input_tokens_seen": 175848452,
      "step": 3139
    },
    {
      "epoch": 6.991091314031181,
      "loss": 0.7722006440162659,
      "loss_ce": 0.00016694323858246207,
      "loss_iou": 0.3046875,
      "loss_num": 0.0322265625,
      "loss_xval": 0.7734375,
      "num_input_tokens_seen": 175848452,
      "step": 3139
    },
    {
      "epoch": 6.993318485523385,
      "grad_norm": 16.247879028320312,
      "learning_rate": 1e-06,
      "loss": 0.5357,
      "num_input_tokens_seen": 175906920,
      "step": 3140
    },
    {
      "epoch": 6.993318485523385,
      "loss": 0.5915122628211975,
      "loss_ce": 0.00020364229567348957,
      "loss_iou": 0.25390625,
      "loss_num": 0.0162353515625,
      "loss_xval": 0.58984375,
      "num_input_tokens_seen": 175906920,
      "step": 3140
    },
    {
      "epoch": 6.99554565701559,
      "grad_norm": 14.442233085632324,
      "learning_rate": 1e-06,
      "loss": 0.4958,
      "num_input_tokens_seen": 175965616,
      "step": 3141
    },
    {
      "epoch": 6.99554565701559,
      "loss": 0.5104349851608276,
      "loss_ce": 0.00018110190285369754,
      "loss_iou": 0.2373046875,
      "loss_num": 0.0069580078125,
      "loss_xval": 0.51171875,
      "num_input_tokens_seen": 175965616,
      "step": 3141
    },
    {
      "epoch": 6.997772828507795,
      "grad_norm": 13.822257995605469,
      "learning_rate": 1e-06,
      "loss": 0.9366,
      "num_input_tokens_seen": 176025056,
      "step": 3142
    },
    {
      "epoch": 6.997772828507795,
      "loss": 0.7665646076202393,
      "loss_ce": 0.00045135943219065666,
      "loss_iou": 0.3359375,
      "loss_num": 0.0185546875,
      "loss_xval": 0.765625,
      "num_input_tokens_seen": 176025056,
      "step": 3142
    },
    {
      "epoch": 7.0,
      "grad_norm": 20.180612564086914,
      "learning_rate": 1e-06,
      "loss": 0.7645,
      "num_input_tokens_seen": 176083108,
      "step": 3143
    },
    {
      "epoch": 7.0,
      "loss": 0.6660090088844299,
      "loss_ce": 0.00023752517881803215,
      "loss_iou": 0.291015625,
      "loss_num": 0.0167236328125,
      "loss_xval": 0.6640625,
      "num_input_tokens_seen": 176083108,
      "step": 3143
    },
    {
      "epoch": 7.002227171492205,
      "grad_norm": 15.68157958984375,
      "learning_rate": 1e-06,
      "loss": 0.6653,
      "num_input_tokens_seen": 176139524,
      "step": 3144
    },
    {
      "epoch": 7.002227171492205,
      "loss": 0.6556283831596375,
      "loss_ce": 0.00023284553026314825,
      "loss_iou": 0.287109375,
      "loss_num": 0.0164794921875,
      "loss_xval": 0.65625,
      "num_input_tokens_seen": 176139524,
      "step": 3144
    },
    {
      "epoch": 7.00445434298441,
      "grad_norm": 31.179643630981445,
      "learning_rate": 1e-06,
      "loss": 0.5867,
      "num_input_tokens_seen": 176194724,
      "step": 3145
    },
    {
      "epoch": 7.00445434298441,
      "loss": 0.49978122115135193,
      "loss_ce": 0.0002695125003810972,
      "loss_iou": 0.23046875,
      "loss_num": 0.00750732421875,
      "loss_xval": 0.5,
      "num_input_tokens_seen": 176194724,
      "step": 3145
    },
    {
      "epoch": 7.006681514476615,
      "grad_norm": 32.13369369506836,
      "learning_rate": 1e-06,
      "loss": 0.7949,
      "num_input_tokens_seen": 176250820,
      "step": 3146
    },
    {
      "epoch": 7.006681514476615,
      "loss": 0.7021375298500061,
      "loss_ce": 0.00023322663037106395,
      "loss_iou": 0.302734375,
      "loss_num": 0.019287109375,
      "loss_xval": 0.703125,
      "num_input_tokens_seen": 176250820,
      "step": 3146
    },
    {
      "epoch": 7.008908685968819,
      "grad_norm": 18.130083084106445,
      "learning_rate": 1e-06,
      "loss": 0.755,
      "num_input_tokens_seen": 176306224,
      "step": 3147
    },
    {
      "epoch": 7.008908685968819,
      "loss": 0.9072054624557495,
      "loss_ce": 0.00022300847922451794,
      "loss_iou": 0.365234375,
      "loss_num": 0.03564453125,
      "loss_xval": 0.90625,
      "num_input_tokens_seen": 176306224,
      "step": 3147
    },
    {
      "epoch": 7.011135857461024,
      "grad_norm": 24.085601806640625,
      "learning_rate": 1e-06,
      "loss": 0.5465,
      "num_input_tokens_seen": 176362740,
      "step": 3148
    },
    {
      "epoch": 7.011135857461024,
      "loss": 0.5722109079360962,
      "loss_ce": 0.00018940077279694378,
      "loss_iou": 0.25390625,
      "loss_num": 0.01275634765625,
      "loss_xval": 0.5703125,
      "num_input_tokens_seen": 176362740,
      "step": 3148
    },
    {
      "epoch": 7.013363028953229,
      "grad_norm": 16.56767463684082,
      "learning_rate": 1e-06,
      "loss": 0.6452,
      "num_input_tokens_seen": 176420184,
      "step": 3149
    },
    {
      "epoch": 7.013363028953229,
      "loss": 0.7920135259628296,
      "loss_ce": 0.00026545586297288537,
      "loss_iou": 0.322265625,
      "loss_num": 0.029052734375,
      "loss_xval": 0.79296875,
      "num_input_tokens_seen": 176420184,
      "step": 3149
    },
    {
      "epoch": 7.015590200445434,
      "grad_norm": 21.7204532623291,
      "learning_rate": 1e-06,
      "loss": 0.7989,
      "num_input_tokens_seen": 176475432,
      "step": 3150
    },
    {
      "epoch": 7.015590200445434,
      "loss": 0.5816566944122314,
      "loss_ce": 0.0003578995238058269,
      "loss_iou": 0.251953125,
      "loss_num": 0.01507568359375,
      "loss_xval": 0.58203125,
      "num_input_tokens_seen": 176475432,
      "step": 3150
    },
    {
      "epoch": 7.017817371937639,
      "grad_norm": 18.18883514404297,
      "learning_rate": 1e-06,
      "loss": 0.8199,
      "num_input_tokens_seen": 176532060,
      "step": 3151
    },
    {
      "epoch": 7.017817371937639,
      "loss": 0.8285827040672302,
      "loss_ce": 0.0002135752292815596,
      "loss_iou": 0.349609375,
      "loss_num": 0.02587890625,
      "loss_xval": 0.828125,
      "num_input_tokens_seen": 176532060,
      "step": 3151
    },
    {
      "epoch": 7.020044543429844,
      "grad_norm": 15.747713088989258,
      "learning_rate": 1e-06,
      "loss": 0.8188,
      "num_input_tokens_seen": 176589504,
      "step": 3152
    },
    {
      "epoch": 7.020044543429844,
      "loss": 0.8652061223983765,
      "loss_ce": 0.00021594867575913668,
      "loss_iou": 0.375,
      "loss_num": 0.0228271484375,
      "loss_xval": 0.86328125,
      "num_input_tokens_seen": 176589504,
      "step": 3152
    },
    {
      "epoch": 7.022271714922049,
      "grad_norm": 22.65471649169922,
      "learning_rate": 1e-06,
      "loss": 0.6007,
      "num_input_tokens_seen": 176647648,
      "step": 3153
    },
    {
      "epoch": 7.022271714922049,
      "loss": 0.7232741117477417,
      "loss_ce": 0.0003737136139534414,
      "loss_iou": 0.3203125,
      "loss_num": 0.0167236328125,
      "loss_xval": 0.72265625,
      "num_input_tokens_seen": 176647648,
      "step": 3153
    },
    {
      "epoch": 7.0244988864142535,
      "grad_norm": 16.48569107055664,
      "learning_rate": 1e-06,
      "loss": 0.7161,
      "num_input_tokens_seen": 176704740,
      "step": 3154
    },
    {
      "epoch": 7.0244988864142535,
      "loss": 0.6486464738845825,
      "loss_ce": 0.00045317187323234975,
      "loss_iou": 0.265625,
      "loss_num": 0.0235595703125,
      "loss_xval": 0.6484375,
      "num_input_tokens_seen": 176704740,
      "step": 3154
    },
    {
      "epoch": 7.026726057906459,
      "grad_norm": 26.504276275634766,
      "learning_rate": 1e-06,
      "loss": 0.6456,
      "num_input_tokens_seen": 176761396,
      "step": 3155
    },
    {
      "epoch": 7.026726057906459,
      "loss": 0.7807806730270386,
      "loss_ce": 0.00026307476218789816,
      "loss_iou": 0.296875,
      "loss_num": 0.037353515625,
      "loss_xval": 0.78125,
      "num_input_tokens_seen": 176761396,
      "step": 3155
    },
    {
      "epoch": 7.028953229398664,
      "grad_norm": 18.16960334777832,
      "learning_rate": 1e-06,
      "loss": 0.4767,
      "num_input_tokens_seen": 176816940,
      "step": 3156
    },
    {
      "epoch": 7.028953229398664,
      "loss": 0.40581727027893066,
      "loss_ce": 0.00017763671348802745,
      "loss_iou": 0.177734375,
      "loss_num": 0.01019287109375,
      "loss_xval": 0.40625,
      "num_input_tokens_seen": 176816940,
      "step": 3156
    },
    {
      "epoch": 7.031180400890869,
      "grad_norm": 16.1363468170166,
      "learning_rate": 1e-06,
      "loss": 0.5441,
      "num_input_tokens_seen": 176873404,
      "step": 3157
    },
    {
      "epoch": 7.031180400890869,
      "loss": 0.65606689453125,
      "loss_ce": 0.00018308302969671786,
      "loss_iou": 0.25390625,
      "loss_num": 0.0299072265625,
      "loss_xval": 0.65625,
      "num_input_tokens_seen": 176873404,
      "step": 3157
    },
    {
      "epoch": 7.033407572383074,
      "grad_norm": 21.781497955322266,
      "learning_rate": 1e-06,
      "loss": 0.4695,
      "num_input_tokens_seen": 176931848,
      "step": 3158
    },
    {
      "epoch": 7.033407572383074,
      "loss": 0.3050188720226288,
      "loss_ce": 0.00020930425671394914,
      "loss_iou": 0.1376953125,
      "loss_num": 0.00604248046875,
      "loss_xval": 0.3046875,
      "num_input_tokens_seen": 176931848,
      "step": 3158
    },
    {
      "epoch": 7.035634743875279,
      "grad_norm": 23.22254753112793,
      "learning_rate": 1e-06,
      "loss": 0.7001,
      "num_input_tokens_seen": 176988340,
      "step": 3159
    },
    {
      "epoch": 7.035634743875279,
      "loss": 0.7503657341003418,
      "loss_ce": 0.0020747713278979063,
      "loss_iou": 0.33203125,
      "loss_num": 0.01708984375,
      "loss_xval": 0.75,
      "num_input_tokens_seen": 176988340,
      "step": 3159
    },
    {
      "epoch": 7.0378619153674835,
      "grad_norm": 15.618306159973145,
      "learning_rate": 1e-06,
      "loss": 0.5591,
      "num_input_tokens_seen": 177043904,
      "step": 3160
    },
    {
      "epoch": 7.0378619153674835,
      "loss": 0.41194963455200195,
      "loss_ce": 0.0009388765902258456,
      "loss_iou": 0.189453125,
      "loss_num": 0.006500244140625,
      "loss_xval": 0.41015625,
      "num_input_tokens_seen": 177043904,
      "step": 3160
    },
    {
      "epoch": 7.040089086859688,
      "grad_norm": 17.742382049560547,
      "learning_rate": 1e-06,
      "loss": 0.5555,
      "num_input_tokens_seen": 177098532,
      "step": 3161
    },
    {
      "epoch": 7.040089086859688,
      "loss": 0.557831883430481,
      "loss_ce": 0.00021466660837177187,
      "loss_iou": 0.265625,
      "loss_num": 0.0050048828125,
      "loss_xval": 0.55859375,
      "num_input_tokens_seen": 177098532,
      "step": 3161
    },
    {
      "epoch": 7.042316258351893,
      "grad_norm": 22.0341854095459,
      "learning_rate": 1e-06,
      "loss": 0.7129,
      "num_input_tokens_seen": 177154748,
      "step": 3162
    },
    {
      "epoch": 7.042316258351893,
      "loss": 0.8048324584960938,
      "loss_ce": 0.0002669950481504202,
      "loss_iou": 0.353515625,
      "loss_num": 0.01904296875,
      "loss_xval": 0.8046875,
      "num_input_tokens_seen": 177154748,
      "step": 3162
    },
    {
      "epoch": 7.044543429844098,
      "grad_norm": 19.422056198120117,
      "learning_rate": 1e-06,
      "loss": 0.5373,
      "num_input_tokens_seen": 177208936,
      "step": 3163
    },
    {
      "epoch": 7.044543429844098,
      "loss": 0.5412275791168213,
      "loss_ce": 0.0031416614074259996,
      "loss_iou": 0.2490234375,
      "loss_num": 0.00799560546875,
      "loss_xval": 0.5390625,
      "num_input_tokens_seen": 177208936,
      "step": 3163
    },
    {
      "epoch": 7.046770601336303,
      "grad_norm": 16.3929443359375,
      "learning_rate": 1e-06,
      "loss": 0.819,
      "num_input_tokens_seen": 177262612,
      "step": 3164
    },
    {
      "epoch": 7.046770601336303,
      "loss": 0.8707385063171387,
      "loss_ce": 0.00025512618594802916,
      "loss_iou": 0.369140625,
      "loss_num": 0.0262451171875,
      "loss_xval": 0.87109375,
      "num_input_tokens_seen": 177262612,
      "step": 3164
    },
    {
      "epoch": 7.048997772828508,
      "grad_norm": 28.08599281311035,
      "learning_rate": 1e-06,
      "loss": 0.7073,
      "num_input_tokens_seen": 177319504,
      "step": 3165
    },
    {
      "epoch": 7.048997772828508,
      "loss": 0.7279913425445557,
      "loss_ce": 0.00020817822951357812,
      "loss_iou": 0.3125,
      "loss_num": 0.020751953125,
      "loss_xval": 0.7265625,
      "num_input_tokens_seen": 177319504,
      "step": 3165
    },
    {
      "epoch": 7.051224944320713,
      "grad_norm": 18.15179443359375,
      "learning_rate": 1e-06,
      "loss": 0.6731,
      "num_input_tokens_seen": 177375208,
      "step": 3166
    },
    {
      "epoch": 7.051224944320713,
      "loss": 0.7211670279502869,
      "loss_ce": 0.00021975839626975358,
      "loss_iou": 0.32421875,
      "loss_num": 0.01434326171875,
      "loss_xval": 0.72265625,
      "num_input_tokens_seen": 177375208,
      "step": 3166
    },
    {
      "epoch": 7.0534521158129175,
      "grad_norm": 24.333948135375977,
      "learning_rate": 1e-06,
      "loss": 0.7331,
      "num_input_tokens_seen": 177432468,
      "step": 3167
    },
    {
      "epoch": 7.0534521158129175,
      "loss": 0.6586510539054871,
      "loss_ce": 0.00020380858040880412,
      "loss_iou": 0.26953125,
      "loss_num": 0.02392578125,
      "loss_xval": 0.66015625,
      "num_input_tokens_seen": 177432468,
      "step": 3167
    },
    {
      "epoch": 7.055679287305122,
      "grad_norm": 24.297325134277344,
      "learning_rate": 1e-06,
      "loss": 0.7255,
      "num_input_tokens_seen": 177487920,
      "step": 3168
    },
    {
      "epoch": 7.055679287305122,
      "loss": 0.5983210206031799,
      "loss_ce": 0.0001764908665791154,
      "loss_iou": 0.2578125,
      "loss_num": 0.0166015625,
      "loss_xval": 0.59765625,
      "num_input_tokens_seen": 177487920,
      "step": 3168
    },
    {
      "epoch": 7.057906458797327,
      "grad_norm": 24.517797470092773,
      "learning_rate": 1e-06,
      "loss": 0.7016,
      "num_input_tokens_seen": 177543680,
      "step": 3169
    },
    {
      "epoch": 7.057906458797327,
      "loss": 0.697583019733429,
      "loss_ce": 0.00019531394354999065,
      "loss_iou": 0.296875,
      "loss_num": 0.0206298828125,
      "loss_xval": 0.69921875,
      "num_input_tokens_seen": 177543680,
      "step": 3169
    },
    {
      "epoch": 7.060133630289532,
      "grad_norm": 15.318553924560547,
      "learning_rate": 1e-06,
      "loss": 0.6435,
      "num_input_tokens_seen": 177600092,
      "step": 3170
    },
    {
      "epoch": 7.060133630289532,
      "loss": 0.6004433035850525,
      "loss_ce": 0.0002235766005469486,
      "loss_iou": 0.26953125,
      "loss_num": 0.012451171875,
      "loss_xval": 0.6015625,
      "num_input_tokens_seen": 177600092,
      "step": 3170
    },
    {
      "epoch": 7.062360801781737,
      "grad_norm": 26.361791610717773,
      "learning_rate": 1e-06,
      "loss": 0.6517,
      "num_input_tokens_seen": 177652892,
      "step": 3171
    },
    {
      "epoch": 7.062360801781737,
      "loss": 0.7172492146492004,
      "loss_ce": 0.00020820970530621707,
      "loss_iou": 0.30859375,
      "loss_num": 0.0203857421875,
      "loss_xval": 0.71875,
      "num_input_tokens_seen": 177652892,
      "step": 3171
    },
    {
      "epoch": 7.064587973273942,
      "grad_norm": 18.68074607849121,
      "learning_rate": 1e-06,
      "loss": 0.6691,
      "num_input_tokens_seen": 177705568,
      "step": 3172
    },
    {
      "epoch": 7.064587973273942,
      "loss": 0.696946382522583,
      "loss_ce": 0.0001690261415205896,
      "loss_iou": 0.30078125,
      "loss_num": 0.018798828125,
      "loss_xval": 0.6953125,
      "num_input_tokens_seen": 177705568,
      "step": 3172
    },
    {
      "epoch": 7.066815144766147,
      "grad_norm": 28.12506866455078,
      "learning_rate": 1e-06,
      "loss": 0.7585,
      "num_input_tokens_seen": 177759380,
      "step": 3173
    },
    {
      "epoch": 7.066815144766147,
      "loss": 0.8229764699935913,
      "loss_ce": 0.0002226082724519074,
      "loss_iou": 0.341796875,
      "loss_num": 0.0281982421875,
      "loss_xval": 0.82421875,
      "num_input_tokens_seen": 177759380,
      "step": 3173
    },
    {
      "epoch": 7.0690423162583516,
      "grad_norm": 25.993946075439453,
      "learning_rate": 1e-06,
      "loss": 0.6376,
      "num_input_tokens_seen": 177814872,
      "step": 3174
    },
    {
      "epoch": 7.0690423162583516,
      "loss": 0.8440980911254883,
      "loss_ce": 0.0015688535058870912,
      "loss_iou": 0.341796875,
      "loss_num": 0.031494140625,
      "loss_xval": 0.84375,
      "num_input_tokens_seen": 177814872,
      "step": 3174
    },
    {
      "epoch": 7.071269487750556,
      "grad_norm": 30.0826416015625,
      "learning_rate": 1e-06,
      "loss": 0.656,
      "num_input_tokens_seen": 177871144,
      "step": 3175
    },
    {
      "epoch": 7.071269487750556,
      "loss": 0.46419858932495117,
      "loss_ce": 0.00020933072664774954,
      "loss_iou": 0.197265625,
      "loss_num": 0.0140380859375,
      "loss_xval": 0.46484375,
      "num_input_tokens_seen": 177871144,
      "step": 3175
    },
    {
      "epoch": 7.073496659242761,
      "grad_norm": 22.121746063232422,
      "learning_rate": 1e-06,
      "loss": 0.6732,
      "num_input_tokens_seen": 177924664,
      "step": 3176
    },
    {
      "epoch": 7.073496659242761,
      "loss": 0.5211669206619263,
      "loss_ce": 0.00017085002036765218,
      "loss_iou": 0.232421875,
      "loss_num": 0.0111083984375,
      "loss_xval": 0.51953125,
      "num_input_tokens_seen": 177924664,
      "step": 3176
    },
    {
      "epoch": 7.075723830734967,
      "grad_norm": 52.88250732421875,
      "learning_rate": 1e-06,
      "loss": 0.5979,
      "num_input_tokens_seen": 177977040,
      "step": 3177
    },
    {
      "epoch": 7.075723830734967,
      "loss": 0.6259975433349609,
      "loss_ce": 0.00026510769384913146,
      "loss_iou": 0.25,
      "loss_num": 0.02490234375,
      "loss_xval": 0.625,
      "num_input_tokens_seen": 177977040,
      "step": 3177
    },
    {
      "epoch": 7.077951002227172,
      "grad_norm": 23.79888916015625,
      "learning_rate": 1e-06,
      "loss": 0.8311,
      "num_input_tokens_seen": 178030552,
      "step": 3178
    },
    {
      "epoch": 7.077951002227172,
      "loss": 0.6989488005638123,
      "loss_ce": 0.0002183011529268697,
      "loss_iou": 0.298828125,
      "loss_num": 0.0201416015625,
      "loss_xval": 0.69921875,
      "num_input_tokens_seen": 178030552,
      "step": 3178
    },
    {
      "epoch": 7.080178173719377,
      "grad_norm": 18.478776931762695,
      "learning_rate": 1e-06,
      "loss": 0.6773,
      "num_input_tokens_seen": 178089092,
      "step": 3179
    },
    {
      "epoch": 7.080178173719377,
      "loss": 0.6097142696380615,
      "loss_ce": 0.0004613480414263904,
      "loss_iou": 0.275390625,
      "loss_num": 0.01141357421875,
      "loss_xval": 0.609375,
      "num_input_tokens_seen": 178089092,
      "step": 3179
    },
    {
      "epoch": 7.082405345211582,
      "grad_norm": 37.656944274902344,
      "learning_rate": 1e-06,
      "loss": 0.7103,
      "num_input_tokens_seen": 178139908,
      "step": 3180
    },
    {
      "epoch": 7.082405345211582,
      "loss": 0.7475607395172119,
      "loss_ce": 0.001344920601695776,
      "loss_iou": 0.333984375,
      "loss_num": 0.015380859375,
      "loss_xval": 0.74609375,
      "num_input_tokens_seen": 178139908,
      "step": 3180
    },
    {
      "epoch": 7.0846325167037865,
      "grad_norm": 21.057231903076172,
      "learning_rate": 1e-06,
      "loss": 0.7319,
      "num_input_tokens_seen": 178196864,
      "step": 3181
    },
    {
      "epoch": 7.0846325167037865,
      "loss": 0.6234728097915649,
      "loss_ce": 0.00018174458818975836,
      "loss_iou": 0.25390625,
      "loss_num": 0.0228271484375,
      "loss_xval": 0.625,
      "num_input_tokens_seen": 178196864,
      "step": 3181
    },
    {
      "epoch": 7.086859688195991,
      "grad_norm": 13.602341651916504,
      "learning_rate": 1e-06,
      "loss": 0.4496,
      "num_input_tokens_seen": 178254848,
      "step": 3182
    },
    {
      "epoch": 7.086859688195991,
      "loss": 0.40568333864212036,
      "loss_ce": 0.00016574125038459897,
      "loss_iou": 0.17578125,
      "loss_num": 0.01080322265625,
      "loss_xval": 0.40625,
      "num_input_tokens_seen": 178254848,
      "step": 3182
    },
    {
      "epoch": 7.089086859688196,
      "grad_norm": 19.736835479736328,
      "learning_rate": 1e-06,
      "loss": 0.6679,
      "num_input_tokens_seen": 178312356,
      "step": 3183
    },
    {
      "epoch": 7.089086859688196,
      "loss": 0.8520088195800781,
      "loss_ce": 0.00020223407773301005,
      "loss_iou": 0.373046875,
      "loss_num": 0.0213623046875,
      "loss_xval": 0.8515625,
      "num_input_tokens_seen": 178312356,
      "step": 3183
    },
    {
      "epoch": 7.091314031180401,
      "grad_norm": 18.998903274536133,
      "learning_rate": 1e-06,
      "loss": 0.577,
      "num_input_tokens_seen": 178369760,
      "step": 3184
    },
    {
      "epoch": 7.091314031180401,
      "loss": 0.6121712327003479,
      "loss_ce": 0.0004158708034083247,
      "loss_iou": 0.232421875,
      "loss_num": 0.0294189453125,
      "loss_xval": 0.61328125,
      "num_input_tokens_seen": 178369760,
      "step": 3184
    },
    {
      "epoch": 7.093541202672606,
      "grad_norm": 17.410633087158203,
      "learning_rate": 1e-06,
      "loss": 0.5794,
      "num_input_tokens_seen": 178424408,
      "step": 3185
    },
    {
      "epoch": 7.093541202672606,
      "loss": 0.6892697215080261,
      "loss_ce": 0.00030487452750094235,
      "loss_iou": 0.294921875,
      "loss_num": 0.0196533203125,
      "loss_xval": 0.6875,
      "num_input_tokens_seen": 178424408,
      "step": 3185
    },
    {
      "epoch": 7.095768374164811,
      "grad_norm": 25.50342559814453,
      "learning_rate": 1e-06,
      "loss": 0.752,
      "num_input_tokens_seen": 178478552,
      "step": 3186
    },
    {
      "epoch": 7.095768374164811,
      "loss": 0.6170066595077515,
      "loss_ce": 0.00018539902521297336,
      "loss_iou": 0.267578125,
      "loss_num": 0.0164794921875,
      "loss_xval": 0.6171875,
      "num_input_tokens_seen": 178478552,
      "step": 3186
    },
    {
      "epoch": 7.097995545657016,
      "grad_norm": 23.802183151245117,
      "learning_rate": 1e-06,
      "loss": 0.6589,
      "num_input_tokens_seen": 178533624,
      "step": 3187
    },
    {
      "epoch": 7.097995545657016,
      "loss": 0.632292628288269,
      "loss_ce": 0.0002125354076270014,
      "loss_iou": 0.271484375,
      "loss_num": 0.017578125,
      "loss_xval": 0.6328125,
      "num_input_tokens_seen": 178533624,
      "step": 3187
    },
    {
      "epoch": 7.1002227171492205,
      "grad_norm": 22.64967155456543,
      "learning_rate": 1e-06,
      "loss": 0.4275,
      "num_input_tokens_seen": 178590396,
      "step": 3188
    },
    {
      "epoch": 7.1002227171492205,
      "loss": 0.49148237705230713,
      "loss_ce": 0.00014936855586711317,
      "loss_iou": 0.22265625,
      "loss_num": 0.00933837890625,
      "loss_xval": 0.4921875,
      "num_input_tokens_seen": 178590396,
      "step": 3188
    },
    {
      "epoch": 7.102449888641425,
      "grad_norm": 17.186386108398438,
      "learning_rate": 1e-06,
      "loss": 0.6466,
      "num_input_tokens_seen": 178648316,
      "step": 3189
    },
    {
      "epoch": 7.102449888641425,
      "loss": 0.7294230461120605,
      "loss_ce": 0.00017502682749181986,
      "loss_iou": 0.333984375,
      "loss_num": 0.0125732421875,
      "loss_xval": 0.73046875,
      "num_input_tokens_seen": 178648316,
      "step": 3189
    },
    {
      "epoch": 7.10467706013363,
      "grad_norm": 14.459593772888184,
      "learning_rate": 1e-06,
      "loss": 0.7198,
      "num_input_tokens_seen": 178705096,
      "step": 3190
    },
    {
      "epoch": 7.10467706013363,
      "loss": 0.62459796667099,
      "loss_ce": 0.00020834297174587846,
      "loss_iou": 0.265625,
      "loss_num": 0.018798828125,
      "loss_xval": 0.625,
      "num_input_tokens_seen": 178705096,
      "step": 3190
    },
    {
      "epoch": 7.106904231625835,
      "grad_norm": 19.029233932495117,
      "learning_rate": 1e-06,
      "loss": 0.7772,
      "num_input_tokens_seen": 178761896,
      "step": 3191
    },
    {
      "epoch": 7.106904231625835,
      "loss": 0.5232353210449219,
      "loss_ce": 0.00016402616165578365,
      "loss_iou": 0.23046875,
      "loss_num": 0.0125732421875,
      "loss_xval": 0.5234375,
      "num_input_tokens_seen": 178761896,
      "step": 3191
    },
    {
      "epoch": 7.10913140311804,
      "grad_norm": 27.418760299682617,
      "learning_rate": 1e-06,
      "loss": 0.7062,
      "num_input_tokens_seen": 178816456,
      "step": 3192
    },
    {
      "epoch": 7.10913140311804,
      "loss": 0.9633630514144897,
      "loss_ce": 0.0004724356404040009,
      "loss_iou": 0.390625,
      "loss_num": 0.036376953125,
      "loss_xval": 0.9609375,
      "num_input_tokens_seen": 178816456,
      "step": 3192
    },
    {
      "epoch": 7.111358574610245,
      "grad_norm": 14.343728065490723,
      "learning_rate": 1e-06,
      "loss": 0.4866,
      "num_input_tokens_seen": 178872960,
      "step": 3193
    },
    {
      "epoch": 7.111358574610245,
      "loss": 0.5601356029510498,
      "loss_ce": 0.0001990435121115297,
      "loss_iou": 0.2080078125,
      "loss_num": 0.0286865234375,
      "loss_xval": 0.55859375,
      "num_input_tokens_seen": 178872960,
      "step": 3193
    },
    {
      "epoch": 7.11358574610245,
      "grad_norm": 32.39339828491211,
      "learning_rate": 1e-06,
      "loss": 0.6818,
      "num_input_tokens_seen": 178927728,
      "step": 3194
    },
    {
      "epoch": 7.11358574610245,
      "loss": 0.7690376043319702,
      "loss_ce": 0.00023876590421423316,
      "loss_iou": 0.296875,
      "loss_num": 0.034912109375,
      "loss_xval": 0.76953125,
      "num_input_tokens_seen": 178927728,
      "step": 3194
    },
    {
      "epoch": 7.1158129175946545,
      "grad_norm": 21.278362274169922,
      "learning_rate": 1e-06,
      "loss": 0.6444,
      "num_input_tokens_seen": 178984676,
      "step": 3195
    },
    {
      "epoch": 7.1158129175946545,
      "loss": 0.5341231822967529,
      "loss_ce": 0.0001876165042631328,
      "loss_iou": 0.2216796875,
      "loss_num": 0.0181884765625,
      "loss_xval": 0.53515625,
      "num_input_tokens_seen": 178984676,
      "step": 3195
    },
    {
      "epoch": 7.118040089086859,
      "grad_norm": 30.706249237060547,
      "learning_rate": 1e-06,
      "loss": 0.7029,
      "num_input_tokens_seen": 179040412,
      "step": 3196
    },
    {
      "epoch": 7.118040089086859,
      "loss": 0.627882719039917,
      "loss_ce": 0.00019718779367394745,
      "loss_iou": 0.291015625,
      "loss_num": 0.00909423828125,
      "loss_xval": 0.62890625,
      "num_input_tokens_seen": 179040412,
      "step": 3196
    },
    {
      "epoch": 7.120267260579064,
      "grad_norm": 14.424131393432617,
      "learning_rate": 1e-06,
      "loss": 0.4814,
      "num_input_tokens_seen": 179096912,
      "step": 3197
    },
    {
      "epoch": 7.120267260579064,
      "loss": 0.46601158380508423,
      "loss_ce": 0.0001913084415718913,
      "loss_iou": 0.2099609375,
      "loss_num": 0.0093994140625,
      "loss_xval": 0.46484375,
      "num_input_tokens_seen": 179096912,
      "step": 3197
    },
    {
      "epoch": 7.122494432071269,
      "grad_norm": 17.124177932739258,
      "learning_rate": 1e-06,
      "loss": 0.6929,
      "num_input_tokens_seen": 179153884,
      "step": 3198
    },
    {
      "epoch": 7.122494432071269,
      "loss": 0.7391812801361084,
      "loss_ce": 0.00016765895998105407,
      "loss_iou": 0.3203125,
      "loss_num": 0.01953125,
      "loss_xval": 0.73828125,
      "num_input_tokens_seen": 179153884,
      "step": 3198
    },
    {
      "epoch": 7.124721603563474,
      "grad_norm": 71.17576599121094,
      "learning_rate": 1e-06,
      "loss": 0.7984,
      "num_input_tokens_seen": 179208364,
      "step": 3199
    },
    {
      "epoch": 7.124721603563474,
      "loss": 0.8881849646568298,
      "loss_ce": 0.0002455237554386258,
      "loss_iou": 0.357421875,
      "loss_num": 0.03466796875,
      "loss_xval": 0.88671875,
      "num_input_tokens_seen": 179208364,
      "step": 3199
    },
    {
      "epoch": 7.12694877505568,
      "grad_norm": 16.409549713134766,
      "learning_rate": 1e-06,
      "loss": 0.5439,
      "num_input_tokens_seen": 179262820,
      "step": 3200
    },
    {
      "epoch": 7.12694877505568,
      "loss": 0.5343835353851318,
      "loss_ce": 0.00020382186630740762,
      "loss_iou": 0.232421875,
      "loss_num": 0.01385498046875,
      "loss_xval": 0.53515625,
      "num_input_tokens_seen": 179262820,
      "step": 3200
    },
    {
      "epoch": 7.129175946547885,
      "grad_norm": 19.769577026367188,
      "learning_rate": 1e-06,
      "loss": 0.8334,
      "num_input_tokens_seen": 179320280,
      "step": 3201
    },
    {
      "epoch": 7.129175946547885,
      "loss": 0.5122010111808777,
      "loss_ce": 0.00017708051018416882,
      "loss_iou": 0.224609375,
      "loss_num": 0.0128173828125,
      "loss_xval": 0.51171875,
      "num_input_tokens_seen": 179320280,
      "step": 3201
    },
    {
      "epoch": 7.131403118040089,
      "grad_norm": 15.5679292678833,
      "learning_rate": 1e-06,
      "loss": 0.5519,
      "num_input_tokens_seen": 179377732,
      "step": 3202
    },
    {
      "epoch": 7.131403118040089,
      "loss": 0.5272064805030823,
      "loss_ce": 0.00022892479319125414,
      "loss_iou": 0.240234375,
      "loss_num": 0.00921630859375,
      "loss_xval": 0.52734375,
      "num_input_tokens_seen": 179377732,
      "step": 3202
    },
    {
      "epoch": 7.133630289532294,
      "grad_norm": 20.086763381958008,
      "learning_rate": 1e-06,
      "loss": 0.7019,
      "num_input_tokens_seen": 179433624,
      "step": 3203
    },
    {
      "epoch": 7.133630289532294,
      "loss": 0.6670153141021729,
      "loss_ce": 0.0002672579721547663,
      "loss_iou": 0.2890625,
      "loss_num": 0.01806640625,
      "loss_xval": 0.66796875,
      "num_input_tokens_seen": 179433624,
      "step": 3203
    },
    {
      "epoch": 7.135857461024499,
      "grad_norm": 23.608802795410156,
      "learning_rate": 1e-06,
      "loss": 0.511,
      "num_input_tokens_seen": 179491448,
      "step": 3204
    },
    {
      "epoch": 7.135857461024499,
      "loss": 0.4480966031551361,
      "loss_ce": 0.00022062801872380078,
      "loss_iou": 0.1953125,
      "loss_num": 0.01141357421875,
      "loss_xval": 0.447265625,
      "num_input_tokens_seen": 179491448,
      "step": 3204
    },
    {
      "epoch": 7.138084632516704,
      "grad_norm": 15.496456146240234,
      "learning_rate": 1e-06,
      "loss": 0.4442,
      "num_input_tokens_seen": 179550108,
      "step": 3205
    },
    {
      "epoch": 7.138084632516704,
      "loss": 0.369751513004303,
      "loss_ce": 0.00036674662260338664,
      "loss_iou": 0.1572265625,
      "loss_num": 0.01104736328125,
      "loss_xval": 0.369140625,
      "num_input_tokens_seen": 179550108,
      "step": 3205
    },
    {
      "epoch": 7.140311804008909,
      "grad_norm": 16.712627410888672,
      "learning_rate": 1e-06,
      "loss": 0.6253,
      "num_input_tokens_seen": 179609256,
      "step": 3206
    },
    {
      "epoch": 7.140311804008909,
      "loss": 0.561935305595398,
      "loss_ce": 0.00016772476374171674,
      "loss_iou": 0.2353515625,
      "loss_num": 0.0181884765625,
      "loss_xval": 0.5625,
      "num_input_tokens_seen": 179609256,
      "step": 3206
    },
    {
      "epoch": 7.142538975501114,
      "grad_norm": 18.983253479003906,
      "learning_rate": 1e-06,
      "loss": 0.8887,
      "num_input_tokens_seen": 179664724,
      "step": 3207
    },
    {
      "epoch": 7.142538975501114,
      "loss": 1.041954755783081,
      "loss_ce": 0.00020668681827373803,
      "loss_iou": 0.421875,
      "loss_num": 0.039306640625,
      "loss_xval": 1.0390625,
      "num_input_tokens_seen": 179664724,
      "step": 3207
    },
    {
      "epoch": 7.144766146993319,
      "grad_norm": 31.973432540893555,
      "learning_rate": 1e-06,
      "loss": 0.6015,
      "num_input_tokens_seen": 179723472,
      "step": 3208
    },
    {
      "epoch": 7.144766146993319,
      "loss": 0.5833202600479126,
      "loss_ce": 0.00019040628103539348,
      "loss_iou": 0.267578125,
      "loss_num": 0.0098876953125,
      "loss_xval": 0.58203125,
      "num_input_tokens_seen": 179723472,
      "step": 3208
    },
    {
      "epoch": 7.146993318485523,
      "grad_norm": 17.996723175048828,
      "learning_rate": 1e-06,
      "loss": 0.5506,
      "num_input_tokens_seen": 179781296,
      "step": 3209
    },
    {
      "epoch": 7.146993318485523,
      "loss": 0.613503098487854,
      "loss_ce": 0.00022186974820215255,
      "loss_iou": 0.259765625,
      "loss_num": 0.0185546875,
      "loss_xval": 0.61328125,
      "num_input_tokens_seen": 179781296,
      "step": 3209
    },
    {
      "epoch": 7.149220489977728,
      "grad_norm": 12.357635498046875,
      "learning_rate": 1e-06,
      "loss": 0.5798,
      "num_input_tokens_seen": 179836852,
      "step": 3210
    },
    {
      "epoch": 7.149220489977728,
      "loss": 0.6682331562042236,
      "loss_ce": 0.0002034001227002591,
      "loss_iou": 0.2373046875,
      "loss_num": 0.03857421875,
      "loss_xval": 0.66796875,
      "num_input_tokens_seen": 179836852,
      "step": 3210
    },
    {
      "epoch": 7.151447661469933,
      "grad_norm": 16.912948608398438,
      "learning_rate": 1e-06,
      "loss": 0.6662,
      "num_input_tokens_seen": 179892884,
      "step": 3211
    },
    {
      "epoch": 7.151447661469933,
      "loss": 0.8070770502090454,
      "loss_ce": 0.00019233419152442366,
      "loss_iou": 0.337890625,
      "loss_num": 0.0263671875,
      "loss_xval": 0.80859375,
      "num_input_tokens_seen": 179892884,
      "step": 3211
    },
    {
      "epoch": 7.153674832962138,
      "grad_norm": 17.714216232299805,
      "learning_rate": 1e-06,
      "loss": 0.4681,
      "num_input_tokens_seen": 179949768,
      "step": 3212
    },
    {
      "epoch": 7.153674832962138,
      "loss": 0.4969853460788727,
      "loss_ce": 0.00015915744006633759,
      "loss_iou": 0.205078125,
      "loss_num": 0.01708984375,
      "loss_xval": 0.49609375,
      "num_input_tokens_seen": 179949768,
      "step": 3212
    },
    {
      "epoch": 7.155902004454343,
      "grad_norm": 26.61876106262207,
      "learning_rate": 1e-06,
      "loss": 0.8391,
      "num_input_tokens_seen": 180005168,
      "step": 3213
    },
    {
      "epoch": 7.155902004454343,
      "loss": 0.7213802337646484,
      "loss_ce": 0.00018884365272242576,
      "loss_iou": 0.31640625,
      "loss_num": 0.0177001953125,
      "loss_xval": 0.72265625,
      "num_input_tokens_seen": 180005168,
      "step": 3213
    },
    {
      "epoch": 7.158129175946548,
      "grad_norm": 15.515600204467773,
      "learning_rate": 1e-06,
      "loss": 0.6922,
      "num_input_tokens_seen": 180061648,
      "step": 3214
    },
    {
      "epoch": 7.158129175946548,
      "loss": 0.6254113912582397,
      "loss_ce": 0.0001672459620749578,
      "loss_iou": 0.26171875,
      "loss_num": 0.02001953125,
      "loss_xval": 0.625,
      "num_input_tokens_seen": 180061648,
      "step": 3214
    },
    {
      "epoch": 7.160356347438753,
      "grad_norm": 25.33653450012207,
      "learning_rate": 1e-06,
      "loss": 0.6162,
      "num_input_tokens_seen": 180118080,
      "step": 3215
    },
    {
      "epoch": 7.160356347438753,
      "loss": 0.5609911680221558,
      "loss_ce": 0.00020018592476844788,
      "loss_iou": 0.2431640625,
      "loss_num": 0.0147705078125,
      "loss_xval": 0.5625,
      "num_input_tokens_seen": 180118080,
      "step": 3215
    },
    {
      "epoch": 7.1625835189309575,
      "grad_norm": 16.110671997070312,
      "learning_rate": 1e-06,
      "loss": 0.6583,
      "num_input_tokens_seen": 180173604,
      "step": 3216
    },
    {
      "epoch": 7.1625835189309575,
      "loss": 0.7682472467422485,
      "loss_ce": 0.00018078596622217447,
      "loss_iou": 0.328125,
      "loss_num": 0.0223388671875,
      "loss_xval": 0.76953125,
      "num_input_tokens_seen": 180173604,
      "step": 3216
    },
    {
      "epoch": 7.164810690423162,
      "grad_norm": 20.151470184326172,
      "learning_rate": 1e-06,
      "loss": 0.5956,
      "num_input_tokens_seen": 180230020,
      "step": 3217
    },
    {
      "epoch": 7.164810690423162,
      "loss": 0.5044952630996704,
      "loss_ce": 0.0003448851057328284,
      "loss_iou": 0.2265625,
      "loss_num": 0.01007080078125,
      "loss_xval": 0.50390625,
      "num_input_tokens_seen": 180230020,
      "step": 3217
    },
    {
      "epoch": 7.167037861915367,
      "grad_norm": 13.990897178649902,
      "learning_rate": 1e-06,
      "loss": 0.8846,
      "num_input_tokens_seen": 180283952,
      "step": 3218
    },
    {
      "epoch": 7.167037861915367,
      "loss": 0.638586699962616,
      "loss_ce": 0.0002810558071359992,
      "loss_iou": 0.283203125,
      "loss_num": 0.0142822265625,
      "loss_xval": 0.63671875,
      "num_input_tokens_seen": 180283952,
      "step": 3218
    },
    {
      "epoch": 7.169265033407572,
      "grad_norm": 17.823301315307617,
      "learning_rate": 1e-06,
      "loss": 0.8448,
      "num_input_tokens_seen": 180340276,
      "step": 3219
    },
    {
      "epoch": 7.169265033407572,
      "loss": 0.7374616861343384,
      "loss_ce": 0.00015702965902164578,
      "loss_iou": 0.310546875,
      "loss_num": 0.0235595703125,
      "loss_xval": 0.73828125,
      "num_input_tokens_seen": 180340276,
      "step": 3219
    },
    {
      "epoch": 7.171492204899777,
      "grad_norm": 21.112295150756836,
      "learning_rate": 1e-06,
      "loss": 0.6465,
      "num_input_tokens_seen": 180397884,
      "step": 3220
    },
    {
      "epoch": 7.171492204899777,
      "loss": 0.4930199384689331,
      "loss_ce": 0.00022211679606698453,
      "loss_iou": 0.2197265625,
      "loss_num": 0.01043701171875,
      "loss_xval": 0.4921875,
      "num_input_tokens_seen": 180397884,
      "step": 3220
    },
    {
      "epoch": 7.173719376391982,
      "grad_norm": 14.755711555480957,
      "learning_rate": 1e-06,
      "loss": 0.4308,
      "num_input_tokens_seen": 180452508,
      "step": 3221
    },
    {
      "epoch": 7.173719376391982,
      "loss": 0.5091297030448914,
      "loss_ce": 0.00021859334083274007,
      "loss_iou": 0.2138671875,
      "loss_num": 0.0164794921875,
      "loss_xval": 0.5078125,
      "num_input_tokens_seen": 180452508,
      "step": 3221
    },
    {
      "epoch": 7.1759465478841875,
      "grad_norm": 17.953371047973633,
      "learning_rate": 1e-06,
      "loss": 0.5174,
      "num_input_tokens_seen": 180510496,
      "step": 3222
    },
    {
      "epoch": 7.1759465478841875,
      "loss": 0.6082879304885864,
      "loss_ce": 0.0004388437664601952,
      "loss_iou": 0.25390625,
      "loss_num": 0.0203857421875,
      "loss_xval": 0.609375,
      "num_input_tokens_seen": 180510496,
      "step": 3222
    },
    {
      "epoch": 7.178173719376392,
      "grad_norm": 27.752344131469727,
      "learning_rate": 1e-06,
      "loss": 0.6884,
      "num_input_tokens_seen": 180567600,
      "step": 3223
    },
    {
      "epoch": 7.178173719376392,
      "loss": 0.8676443696022034,
      "loss_ce": 0.00021275154722388834,
      "loss_iou": 0.35546875,
      "loss_num": 0.03173828125,
      "loss_xval": 0.8671875,
      "num_input_tokens_seen": 180567600,
      "step": 3223
    },
    {
      "epoch": 7.180400890868597,
      "grad_norm": 20.237045288085938,
      "learning_rate": 1e-06,
      "loss": 0.7489,
      "num_input_tokens_seen": 180622160,
      "step": 3224
    },
    {
      "epoch": 7.180400890868597,
      "loss": 0.7350777387619019,
      "loss_ce": 0.0002144520840374753,
      "loss_iou": 0.318359375,
      "loss_num": 0.01953125,
      "loss_xval": 0.734375,
      "num_input_tokens_seen": 180622160,
      "step": 3224
    },
    {
      "epoch": 7.182628062360802,
      "grad_norm": 17.729013442993164,
      "learning_rate": 1e-06,
      "loss": 0.6288,
      "num_input_tokens_seen": 180678492,
      "step": 3225
    },
    {
      "epoch": 7.182628062360802,
      "loss": 0.7737811803817749,
      "loss_ce": 0.00034367607440799475,
      "loss_iou": 0.3515625,
      "loss_num": 0.01458740234375,
      "loss_xval": 0.7734375,
      "num_input_tokens_seen": 180678492,
      "step": 3225
    },
    {
      "epoch": 7.184855233853007,
      "grad_norm": 17.447938919067383,
      "learning_rate": 1e-06,
      "loss": 0.8197,
      "num_input_tokens_seen": 180734308,
      "step": 3226
    },
    {
      "epoch": 7.184855233853007,
      "loss": 0.6977049708366394,
      "loss_ce": 0.00019519682973623276,
      "loss_iou": 0.298828125,
      "loss_num": 0.02001953125,
      "loss_xval": 0.69921875,
      "num_input_tokens_seen": 180734308,
      "step": 3226
    },
    {
      "epoch": 7.187082405345212,
      "grad_norm": 17.61049461364746,
      "learning_rate": 1e-06,
      "loss": 0.7649,
      "num_input_tokens_seen": 180789952,
      "step": 3227
    },
    {
      "epoch": 7.187082405345212,
      "loss": 0.6152649521827698,
      "loss_ce": 0.00027470148052088916,
      "loss_iou": 0.255859375,
      "loss_num": 0.0206298828125,
      "loss_xval": 0.61328125,
      "num_input_tokens_seen": 180789952,
      "step": 3227
    },
    {
      "epoch": 7.189309576837417,
      "grad_norm": 17.236371994018555,
      "learning_rate": 1e-06,
      "loss": 0.6585,
      "num_input_tokens_seen": 180844252,
      "step": 3228
    },
    {
      "epoch": 7.189309576837417,
      "loss": 0.5438805818557739,
      "loss_ce": 0.00017939825193025172,
      "loss_iou": 0.25,
      "loss_num": 0.00823974609375,
      "loss_xval": 0.54296875,
      "num_input_tokens_seen": 180844252,
      "step": 3228
    },
    {
      "epoch": 7.1915367483296215,
      "grad_norm": 15.305548667907715,
      "learning_rate": 1e-06,
      "loss": 0.543,
      "num_input_tokens_seen": 180898788,
      "step": 3229
    },
    {
      "epoch": 7.1915367483296215,
      "loss": 0.5575687885284424,
      "loss_ce": 0.00019571834127418697,
      "loss_iou": 0.2421875,
      "loss_num": 0.0146484375,
      "loss_xval": 0.55859375,
      "num_input_tokens_seen": 180898788,
      "step": 3229
    },
    {
      "epoch": 7.193763919821826,
      "grad_norm": 36.117584228515625,
      "learning_rate": 1e-06,
      "loss": 0.8034,
      "num_input_tokens_seen": 180955360,
      "step": 3230
    },
    {
      "epoch": 7.193763919821826,
      "loss": 0.7606761455535889,
      "loss_ce": 0.00017807254334911704,
      "loss_iou": 0.31640625,
      "loss_num": 0.0257568359375,
      "loss_xval": 0.76171875,
      "num_input_tokens_seen": 180955360,
      "step": 3230
    },
    {
      "epoch": 7.195991091314031,
      "grad_norm": 26.037805557250977,
      "learning_rate": 1e-06,
      "loss": 0.5336,
      "num_input_tokens_seen": 181010364,
      "step": 3231
    },
    {
      "epoch": 7.195991091314031,
      "loss": 0.4867013990879059,
      "loss_ce": 0.0001901786308735609,
      "loss_iou": 0.2119140625,
      "loss_num": 0.012451171875,
      "loss_xval": 0.486328125,
      "num_input_tokens_seen": 181010364,
      "step": 3231
    },
    {
      "epoch": 7.198218262806236,
      "grad_norm": 20.942747116088867,
      "learning_rate": 1e-06,
      "loss": 0.6008,
      "num_input_tokens_seen": 181065840,
      "step": 3232
    },
    {
      "epoch": 7.198218262806236,
      "loss": 0.5934904217720032,
      "loss_ce": 0.00022868883388582617,
      "loss_iou": 0.25390625,
      "loss_num": 0.01708984375,
      "loss_xval": 0.59375,
      "num_input_tokens_seen": 181065840,
      "step": 3232
    },
    {
      "epoch": 7.200445434298441,
      "grad_norm": 14.775635719299316,
      "learning_rate": 1e-06,
      "loss": 0.6616,
      "num_input_tokens_seen": 181122380,
      "step": 3233
    },
    {
      "epoch": 7.200445434298441,
      "loss": 0.5939116477966309,
      "loss_ce": 0.0001616643276065588,
      "loss_iou": 0.267578125,
      "loss_num": 0.0115966796875,
      "loss_xval": 0.59375,
      "num_input_tokens_seen": 181122380,
      "step": 3233
    },
    {
      "epoch": 7.202672605790646,
      "grad_norm": 16.27145004272461,
      "learning_rate": 1e-06,
      "loss": 0.84,
      "num_input_tokens_seen": 181179864,
      "step": 3234
    },
    {
      "epoch": 7.202672605790646,
      "loss": 0.9337013959884644,
      "loss_ce": 0.0008400785736739635,
      "loss_iou": 0.38671875,
      "loss_num": 0.03173828125,
      "loss_xval": 0.93359375,
      "num_input_tokens_seen": 181179864,
      "step": 3234
    },
    {
      "epoch": 7.204899777282851,
      "grad_norm": 29.473827362060547,
      "learning_rate": 1e-06,
      "loss": 0.7444,
      "num_input_tokens_seen": 181233212,
      "step": 3235
    },
    {
      "epoch": 7.204899777282851,
      "loss": 0.8354470729827881,
      "loss_ce": 0.0004861447378061712,
      "loss_iou": 0.3671875,
      "loss_num": 0.020263671875,
      "loss_xval": 0.8359375,
      "num_input_tokens_seen": 181233212,
      "step": 3235
    },
    {
      "epoch": 7.2071269487750556,
      "grad_norm": 22.71228790283203,
      "learning_rate": 1e-06,
      "loss": 0.6487,
      "num_input_tokens_seen": 181288260,
      "step": 3236
    },
    {
      "epoch": 7.2071269487750556,
      "loss": 0.7758283615112305,
      "loss_ce": 0.00019357707060407847,
      "loss_iou": 0.32421875,
      "loss_num": 0.025146484375,
      "loss_xval": 0.77734375,
      "num_input_tokens_seen": 181288260,
      "step": 3236
    },
    {
      "epoch": 7.20935412026726,
      "grad_norm": 16.81915283203125,
      "learning_rate": 1e-06,
      "loss": 0.7454,
      "num_input_tokens_seen": 181343576,
      "step": 3237
    },
    {
      "epoch": 7.20935412026726,
      "loss": 0.8085384368896484,
      "loss_ce": 0.00031090632546693087,
      "loss_iou": 0.357421875,
      "loss_num": 0.018310546875,
      "loss_xval": 0.80859375,
      "num_input_tokens_seen": 181343576,
      "step": 3237
    },
    {
      "epoch": 7.211581291759465,
      "grad_norm": 15.39242172241211,
      "learning_rate": 1e-06,
      "loss": 0.5735,
      "num_input_tokens_seen": 181400228,
      "step": 3238
    },
    {
      "epoch": 7.211581291759465,
      "loss": 0.6538634300231934,
      "loss_ce": 0.0002989704080391675,
      "loss_iou": 0.279296875,
      "loss_num": 0.018798828125,
      "loss_xval": 0.65234375,
      "num_input_tokens_seen": 181400228,
      "step": 3238
    },
    {
      "epoch": 7.21380846325167,
      "grad_norm": 59.97997283935547,
      "learning_rate": 1e-06,
      "loss": 0.6083,
      "num_input_tokens_seen": 181458112,
      "step": 3239
    },
    {
      "epoch": 7.21380846325167,
      "loss": 0.7236316204071045,
      "loss_ce": 0.0002429374580970034,
      "loss_iou": 0.318359375,
      "loss_num": 0.0172119140625,
      "loss_xval": 0.72265625,
      "num_input_tokens_seen": 181458112,
      "step": 3239
    },
    {
      "epoch": 7.216035634743875,
      "grad_norm": 13.519067764282227,
      "learning_rate": 1e-06,
      "loss": 0.5369,
      "num_input_tokens_seen": 181514996,
      "step": 3240
    },
    {
      "epoch": 7.216035634743875,
      "loss": 0.36852580308914185,
      "loss_ce": 0.00017863643006421626,
      "loss_iou": 0.1572265625,
      "loss_num": 0.01080322265625,
      "loss_xval": 0.369140625,
      "num_input_tokens_seen": 181514996,
      "step": 3240
    },
    {
      "epoch": 7.21826280623608,
      "grad_norm": 26.277721405029297,
      "learning_rate": 1e-06,
      "loss": 0.4887,
      "num_input_tokens_seen": 181571304,
      "step": 3241
    },
    {
      "epoch": 7.21826280623608,
      "loss": 0.5072898268699646,
      "loss_ce": 0.0006980198086239398,
      "loss_iou": 0.2265625,
      "loss_num": 0.01055908203125,
      "loss_xval": 0.5078125,
      "num_input_tokens_seen": 181571304,
      "step": 3241
    },
    {
      "epoch": 7.220489977728285,
      "grad_norm": 146.0947265625,
      "learning_rate": 1e-06,
      "loss": 0.62,
      "num_input_tokens_seen": 181626812,
      "step": 3242
    },
    {
      "epoch": 7.220489977728285,
      "loss": 0.640555202960968,
      "loss_ce": 0.000174342465470545,
      "loss_iou": 0.294921875,
      "loss_num": 0.010009765625,
      "loss_xval": 0.640625,
      "num_input_tokens_seen": 181626812,
      "step": 3242
    },
    {
      "epoch": 7.22271714922049,
      "grad_norm": 26.887720108032227,
      "learning_rate": 1e-06,
      "loss": 0.8177,
      "num_input_tokens_seen": 181681800,
      "step": 3243
    },
    {
      "epoch": 7.22271714922049,
      "loss": 0.6879905462265015,
      "loss_ce": 0.0004905810346826911,
      "loss_iou": 0.296875,
      "loss_num": 0.018798828125,
      "loss_xval": 0.6875,
      "num_input_tokens_seen": 181681800,
      "step": 3243
    },
    {
      "epoch": 7.224944320712694,
      "grad_norm": 17.908056259155273,
      "learning_rate": 1e-06,
      "loss": 0.6145,
      "num_input_tokens_seen": 181736656,
      "step": 3244
    },
    {
      "epoch": 7.224944320712694,
      "loss": 0.5160667896270752,
      "loss_ce": 0.00019765175238717347,
      "loss_iou": 0.2314453125,
      "loss_num": 0.0107421875,
      "loss_xval": 0.515625,
      "num_input_tokens_seen": 181736656,
      "step": 3244
    },
    {
      "epoch": 7.2271714922049,
      "grad_norm": 13.982433319091797,
      "learning_rate": 1e-06,
      "loss": 0.5749,
      "num_input_tokens_seen": 181794728,
      "step": 3245
    },
    {
      "epoch": 7.2271714922049,
      "loss": 0.4950827658176422,
      "loss_ce": 0.0002097318647429347,
      "loss_iou": 0.1953125,
      "loss_num": 0.02099609375,
      "loss_xval": 0.494140625,
      "num_input_tokens_seen": 181794728,
      "step": 3245
    },
    {
      "epoch": 7.229398663697105,
      "grad_norm": 19.076622009277344,
      "learning_rate": 1e-06,
      "loss": 0.5997,
      "num_input_tokens_seen": 181850576,
      "step": 3246
    },
    {
      "epoch": 7.229398663697105,
      "loss": 0.5747532844543457,
      "loss_ce": 0.00029037147760391235,
      "loss_iou": 0.23046875,
      "loss_num": 0.0225830078125,
      "loss_xval": 0.57421875,
      "num_input_tokens_seen": 181850576,
      "step": 3246
    },
    {
      "epoch": 7.23162583518931,
      "grad_norm": 36.7861442565918,
      "learning_rate": 1e-06,
      "loss": 0.8128,
      "num_input_tokens_seen": 181905156,
      "step": 3247
    },
    {
      "epoch": 7.23162583518931,
      "loss": 0.9604984521865845,
      "loss_ce": 0.0010258200345560908,
      "loss_iou": 0.40234375,
      "loss_num": 0.03076171875,
      "loss_xval": 0.9609375,
      "num_input_tokens_seen": 181905156,
      "step": 3247
    },
    {
      "epoch": 7.233853006681515,
      "grad_norm": 28.286663055419922,
      "learning_rate": 1e-06,
      "loss": 0.7149,
      "num_input_tokens_seen": 181961164,
      "step": 3248
    },
    {
      "epoch": 7.233853006681515,
      "loss": 0.7203893065452576,
      "loss_ce": 0.0001744742039591074,
      "loss_iou": 0.33203125,
      "loss_num": 0.01141357421875,
      "loss_xval": 0.71875,
      "num_input_tokens_seen": 181961164,
      "step": 3248
    },
    {
      "epoch": 7.23608017817372,
      "grad_norm": 18.691064834594727,
      "learning_rate": 1e-06,
      "loss": 0.6303,
      "num_input_tokens_seen": 182020592,
      "step": 3249
    },
    {
      "epoch": 7.23608017817372,
      "loss": 0.6998783946037292,
      "loss_ce": 0.00017135526286438107,
      "loss_iou": 0.296875,
      "loss_num": 0.0211181640625,
      "loss_xval": 0.69921875,
      "num_input_tokens_seen": 182020592,
      "step": 3249
    },
    {
      "epoch": 7.2383073496659245,
      "grad_norm": 15.494166374206543,
      "learning_rate": 1e-06,
      "loss": 0.6059,
      "num_input_tokens_seen": 182075240,
      "step": 3250
    },
    {
      "epoch": 7.2383073496659245,
      "eval_seeclick_web_CIoU": 0.5736072361469269,
      "eval_seeclick_web_GIoU": 0.5713272094726562,
      "eval_seeclick_web_IoU": 0.5900914669036865,
      "eval_seeclick_web_MAE_all": 0.01693468587473035,
      "eval_seeclick_web_MAE_h": 0.009493312099948525,
      "eval_seeclick_web_MAE_w": 0.017465373501181602,
      "eval_seeclick_web_MAE_x_boxes": 0.009315244387835264,
      "eval_seeclick_web_MAE_y_boxes": 0.022717589512467384,
      "eval_seeclick_web_inside_bbox": 0.9166666567325592,
      "eval_seeclick_web_loss": 0.9355779886245728,
      "eval_seeclick_web_loss_ce": 0.00025881038163788617,
      "eval_seeclick_web_loss_iou": 0.4263916015625,
      "eval_seeclick_web_loss_num": 0.013437271118164062,
      "eval_seeclick_web_loss_xval": 0.920166015625,
      "eval_seeclick_web_runtime": 22.493,
      "eval_seeclick_web_samples_per_second": 2.223,
      "eval_seeclick_web_steps_per_second": 0.089,
      "num_input_tokens_seen": 182075240,
      "step": 3250
    },
    {
      "epoch": 7.2383073496659245,
      "eval_icons_CIoU": 0.2808741182088852,
      "eval_icons_GIoU": 0.3052249103784561,
      "eval_icons_IoU": 0.3607981503009796,
      "eval_icons_MAE_all": 0.06516656465828419,
      "eval_icons_MAE_h": 0.03764822334051132,
      "eval_icons_MAE_w": 0.07068785838782787,
      "eval_icons_MAE_x_boxes": 0.06141933798789978,
      "eval_icons_MAE_y_boxes": 0.03828867059201002,
      "eval_icons_inside_bbox": 0.6336805522441864,
      "eval_icons_loss": 1.7471874952316284,
      "eval_icons_loss_ce": 0.0003105820360360667,
      "eval_icons_loss_iou": 0.67919921875,
      "eval_icons_loss_num": 0.06205558776855469,
      "eval_icons_loss_xval": 1.66943359375,
      "eval_icons_runtime": 21.7009,
      "eval_icons_samples_per_second": 2.304,
      "eval_icons_steps_per_second": 0.092,
      "num_input_tokens_seen": 182075240,
      "step": 3250
    },
    {
      "epoch": 7.2383073496659245,
      "eval_screenspot_CIoU": 0.342861811319987,
      "eval_screenspot_GIoU": 0.3614034950733185,
      "eval_screenspot_IoU": 0.42649880051612854,
      "eval_screenspot_MAE_all": 0.06335503856341045,
      "eval_screenspot_MAE_h": 0.03755492903292179,
      "eval_screenspot_MAE_w": 0.07218488802512486,
      "eval_screenspot_MAE_x_boxes": 0.07481345720589161,
      "eval_screenspot_MAE_y_boxes": 0.04668992726753155,
      "eval_screenspot_inside_bbox": 0.6729166706403097,
      "eval_screenspot_loss": 1.6553105115890503,
      "eval_screenspot_loss_ce": 0.00030076557110684615,
      "eval_screenspot_loss_iou": 0.6756998697916666,
      "eval_screenspot_loss_num": 0.07468668619791667,
      "eval_screenspot_loss_xval": 1.7252604166666667,
      "eval_screenspot_runtime": 36.4126,
      "eval_screenspot_samples_per_second": 2.444,
      "eval_screenspot_steps_per_second": 0.082,
      "num_input_tokens_seen": 182075240,
      "step": 3250
    },
    {
      "epoch": 7.2383073496659245,
      "eval_compot_CIoU": 0.35138705372810364,
      "eval_compot_GIoU": 0.3623329848051071,
      "eval_compot_IoU": 0.4089464396238327,
      "eval_compot_MAE_all": 0.01784850051626563,
      "eval_compot_MAE_h": 0.009071170818060637,
      "eval_compot_MAE_w": 0.02207251265645027,
      "eval_compot_MAE_x_boxes": 0.029478789307177067,
      "eval_compot_MAE_y_boxes": 0.0062866308726370335,
      "eval_compot_inside_bbox": 0.6458333432674408,
      "eval_compot_loss": 1.3766494989395142,
      "eval_compot_loss_ce": 0.00024336049682460725,
      "eval_compot_loss_iou": 0.6270751953125,
      "eval_compot_loss_num": 0.0166168212890625,
      "eval_compot_loss_xval": 1.3369140625,
      "eval_compot_runtime": 22.3641,
      "eval_compot_samples_per_second": 2.236,
      "eval_compot_steps_per_second": 0.089,
      "num_input_tokens_seen": 182075240,
      "step": 3250
    },
    {
      "epoch": 7.2383073496659245,
      "eval_custom_ui_val_CIoU": 0.46513410409291583,
      "eval_custom_ui_val_GIoU": 0.47927956614229417,
      "eval_custom_ui_val_IoU": 0.5227026873164706,
      "eval_custom_ui_val_MAE_all": 0.03009833147128423,
      "eval_custom_ui_val_MAE_h": 0.017143823982526857,
      "eval_custom_ui_val_MAE_w": 0.038110896220637694,
      "eval_custom_ui_val_MAE_x_boxes": 0.03446255738122596,
      "eval_custom_ui_val_MAE_y_boxes": 0.015643620294415288,
      "eval_custom_ui_val_inside_bbox": 0.7527006202273898,
      "eval_custom_ui_val_loss": 1.198036789894104,
      "eval_custom_ui_val_loss_ce": 0.000274412335582181,
      "eval_custom_ui_val_loss_iou": 0.5103217230902778,
      "eval_custom_ui_val_loss_num": 0.028072569105360243,
      "eval_custom_ui_val_loss_xval": 1.1607801649305556,
      "eval_custom_ui_val_runtime": 60.3528,
      "eval_custom_ui_val_samples_per_second": 4.391,
      "eval_custom_ui_val_steps_per_second": 0.149,
      "num_input_tokens_seen": 182075240,
      "step": 3250
    },
    {
      "epoch": 7.2383073496659245,
      "loss": 0.9084522724151611,
      "loss_ce": 0.00024910393403843045,
      "loss_iou": 0.396484375,
      "loss_num": 0.023193359375,
      "loss_xval": 0.90625,
      "num_input_tokens_seen": 182075240,
      "step": 3250
    },
    {
      "epoch": 7.240534521158129,
      "grad_norm": 24.0264892578125,
      "learning_rate": 1e-06,
      "loss": 0.7723,
      "num_input_tokens_seen": 182134612,
      "step": 3251
    },
    {
      "epoch": 7.240534521158129,
      "loss": 0.7384548187255859,
      "loss_ce": 0.00017352268332615495,
      "loss_iou": 0.291015625,
      "loss_num": 0.03125,
      "loss_xval": 0.73828125,
      "num_input_tokens_seen": 182134612,
      "step": 3251
    },
    {
      "epoch": 7.242761692650334,
      "grad_norm": 17.387958526611328,
      "learning_rate": 1e-06,
      "loss": 0.4481,
      "num_input_tokens_seen": 182192124,
      "step": 3252
    },
    {
      "epoch": 7.242761692650334,
      "loss": 0.40355396270751953,
      "loss_ce": 0.004872324876487255,
      "loss_iou": 0.171875,
      "loss_num": 0.01104736328125,
      "loss_xval": 0.3984375,
      "num_input_tokens_seen": 182192124,
      "step": 3252
    },
    {
      "epoch": 7.244988864142539,
      "grad_norm": 26.076507568359375,
      "learning_rate": 1e-06,
      "loss": 0.6101,
      "num_input_tokens_seen": 182248140,
      "step": 3253
    },
    {
      "epoch": 7.244988864142539,
      "loss": 0.5168513059616089,
      "loss_ce": 0.0002497411041986197,
      "loss_iou": 0.220703125,
      "loss_num": 0.01513671875,
      "loss_xval": 0.515625,
      "num_input_tokens_seen": 182248140,
      "step": 3253
    },
    {
      "epoch": 7.247216035634744,
      "grad_norm": 15.936118125915527,
      "learning_rate": 1e-06,
      "loss": 0.6972,
      "num_input_tokens_seen": 182302560,
      "step": 3254
    },
    {
      "epoch": 7.247216035634744,
      "loss": 0.6615881323814392,
      "loss_ce": 0.00027220824267715216,
      "loss_iou": 0.251953125,
      "loss_num": 0.031982421875,
      "loss_xval": 0.66015625,
      "num_input_tokens_seen": 182302560,
      "step": 3254
    },
    {
      "epoch": 7.249443207126949,
      "grad_norm": 21.23505210876465,
      "learning_rate": 1e-06,
      "loss": 0.6678,
      "num_input_tokens_seen": 182359212,
      "step": 3255
    },
    {
      "epoch": 7.249443207126949,
      "loss": 0.6819499135017395,
      "loss_ce": 0.00024824903812259436,
      "loss_iou": 0.263671875,
      "loss_num": 0.030517578125,
      "loss_xval": 0.68359375,
      "num_input_tokens_seen": 182359212,
      "step": 3255
    },
    {
      "epoch": 7.251670378619154,
      "grad_norm": 14.325440406799316,
      "learning_rate": 1e-06,
      "loss": 0.6858,
      "num_input_tokens_seen": 182412908,
      "step": 3256
    },
    {
      "epoch": 7.251670378619154,
      "loss": 0.6170239448547363,
      "loss_ce": 0.0008129666093736887,
      "loss_iou": 0.251953125,
      "loss_num": 0.022705078125,
      "loss_xval": 0.6171875,
      "num_input_tokens_seen": 182412908,
      "step": 3256
    },
    {
      "epoch": 7.2538975501113585,
      "grad_norm": 16.26470184326172,
      "learning_rate": 1e-06,
      "loss": 0.9041,
      "num_input_tokens_seen": 182471924,
      "step": 3257
    },
    {
      "epoch": 7.2538975501113585,
      "loss": 0.7431246042251587,
      "loss_ce": 0.00044888071715831757,
      "loss_iou": 0.310546875,
      "loss_num": 0.0240478515625,
      "loss_xval": 0.7421875,
      "num_input_tokens_seen": 182471924,
      "step": 3257
    },
    {
      "epoch": 7.256124721603563,
      "grad_norm": 18.07449722290039,
      "learning_rate": 1e-06,
      "loss": 0.5345,
      "num_input_tokens_seen": 182526908,
      "step": 3258
    },
    {
      "epoch": 7.256124721603563,
      "loss": 0.549538791179657,
      "loss_ce": 0.00022235384676605463,
      "loss_iou": 0.2421875,
      "loss_num": 0.012939453125,
      "loss_xval": 0.55078125,
      "num_input_tokens_seen": 182526908,
      "step": 3258
    },
    {
      "epoch": 7.258351893095768,
      "grad_norm": 15.7253999710083,
      "learning_rate": 1e-06,
      "loss": 0.6923,
      "num_input_tokens_seen": 182583592,
      "step": 3259
    },
    {
      "epoch": 7.258351893095768,
      "loss": 0.9013206958770752,
      "loss_ce": 0.001052145496942103,
      "loss_iou": 0.4140625,
      "loss_num": 0.01446533203125,
      "loss_xval": 0.8984375,
      "num_input_tokens_seen": 182583592,
      "step": 3259
    },
    {
      "epoch": 7.260579064587973,
      "grad_norm": 20.088714599609375,
      "learning_rate": 1e-06,
      "loss": 0.535,
      "num_input_tokens_seen": 182641252,
      "step": 3260
    },
    {
      "epoch": 7.260579064587973,
      "loss": 0.5077439546585083,
      "loss_ce": 0.00017562185530550778,
      "loss_iou": 0.228515625,
      "loss_num": 0.01019287109375,
      "loss_xval": 0.5078125,
      "num_input_tokens_seen": 182641252,
      "step": 3260
    },
    {
      "epoch": 7.262806236080178,
      "grad_norm": 15.09283447265625,
      "learning_rate": 1e-06,
      "loss": 0.5227,
      "num_input_tokens_seen": 182697120,
      "step": 3261
    },
    {
      "epoch": 7.262806236080178,
      "loss": 0.43932655453681946,
      "loss_ce": 0.00017860504158306867,
      "loss_iou": 0.1826171875,
      "loss_num": 0.01470947265625,
      "loss_xval": 0.439453125,
      "num_input_tokens_seen": 182697120,
      "step": 3261
    },
    {
      "epoch": 7.265033407572383,
      "grad_norm": 19.676998138427734,
      "learning_rate": 1e-06,
      "loss": 0.6918,
      "num_input_tokens_seen": 182752892,
      "step": 3262
    },
    {
      "epoch": 7.265033407572383,
      "loss": 0.630585789680481,
      "loss_ce": 0.00021469607600010931,
      "loss_iou": 0.275390625,
      "loss_num": 0.016357421875,
      "loss_xval": 0.62890625,
      "num_input_tokens_seen": 182752892,
      "step": 3262
    },
    {
      "epoch": 7.267260579064588,
      "grad_norm": 110.68122863769531,
      "learning_rate": 1e-06,
      "loss": 0.7325,
      "num_input_tokens_seen": 182810764,
      "step": 3263
    },
    {
      "epoch": 7.267260579064588,
      "loss": 0.7132278084754944,
      "loss_ce": 0.0002150905056623742,
      "loss_iou": 0.310546875,
      "loss_num": 0.01806640625,
      "loss_xval": 0.71484375,
      "num_input_tokens_seen": 182810764,
      "step": 3263
    },
    {
      "epoch": 7.2694877505567925,
      "grad_norm": 23.634044647216797,
      "learning_rate": 1e-06,
      "loss": 0.7038,
      "num_input_tokens_seen": 182868244,
      "step": 3264
    },
    {
      "epoch": 7.2694877505567925,
      "loss": 0.6935637593269348,
      "loss_ce": 0.002157476032152772,
      "loss_iou": 0.28515625,
      "loss_num": 0.0244140625,
      "loss_xval": 0.69140625,
      "num_input_tokens_seen": 182868244,
      "step": 3264
    },
    {
      "epoch": 7.271714922048997,
      "grad_norm": 23.025278091430664,
      "learning_rate": 1e-06,
      "loss": 0.5485,
      "num_input_tokens_seen": 182924824,
      "step": 3265
    },
    {
      "epoch": 7.271714922048997,
      "loss": 0.5591219067573547,
      "loss_ce": 0.00028401942108757794,
      "loss_iou": 0.2578125,
      "loss_num": 0.00860595703125,
      "loss_xval": 0.55859375,
      "num_input_tokens_seen": 182924824,
      "step": 3265
    },
    {
      "epoch": 7.273942093541203,
      "grad_norm": 15.47992992401123,
      "learning_rate": 1e-06,
      "loss": 0.6323,
      "num_input_tokens_seen": 182979672,
      "step": 3266
    },
    {
      "epoch": 7.273942093541203,
      "loss": 0.7570323944091797,
      "loss_ce": 0.00019647592853289098,
      "loss_iou": 0.31640625,
      "loss_num": 0.0252685546875,
      "loss_xval": 0.7578125,
      "num_input_tokens_seen": 182979672,
      "step": 3266
    },
    {
      "epoch": 7.276169265033408,
      "grad_norm": 19.16633415222168,
      "learning_rate": 1e-06,
      "loss": 0.6991,
      "num_input_tokens_seen": 183035308,
      "step": 3267
    },
    {
      "epoch": 7.276169265033408,
      "loss": 0.8601139783859253,
      "loss_ce": 0.00037270993925631046,
      "loss_iou": 0.34765625,
      "loss_num": 0.03271484375,
      "loss_xval": 0.859375,
      "num_input_tokens_seen": 183035308,
      "step": 3267
    },
    {
      "epoch": 7.278396436525613,
      "grad_norm": 17.217559814453125,
      "learning_rate": 1e-06,
      "loss": 0.8387,
      "num_input_tokens_seen": 183093172,
      "step": 3268
    },
    {
      "epoch": 7.278396436525613,
      "loss": 0.6842477321624756,
      "loss_ce": 0.00016567618877161294,
      "loss_iou": 0.2734375,
      "loss_num": 0.0274658203125,
      "loss_xval": 0.68359375,
      "num_input_tokens_seen": 183093172,
      "step": 3268
    },
    {
      "epoch": 7.280623608017818,
      "grad_norm": 16.485475540161133,
      "learning_rate": 1e-06,
      "loss": 0.6404,
      "num_input_tokens_seen": 183148780,
      "step": 3269
    },
    {
      "epoch": 7.280623608017818,
      "loss": 0.7775435447692871,
      "loss_ce": 0.00019982852973043919,
      "loss_iou": 0.328125,
      "loss_num": 0.0240478515625,
      "loss_xval": 0.77734375,
      "num_input_tokens_seen": 183148780,
      "step": 3269
    },
    {
      "epoch": 7.282850779510023,
      "grad_norm": 14.587860107421875,
      "learning_rate": 1e-06,
      "loss": 0.6758,
      "num_input_tokens_seen": 183205300,
      "step": 3270
    },
    {
      "epoch": 7.282850779510023,
      "loss": 0.5569312572479248,
      "loss_ce": 0.0001685761963017285,
      "loss_iou": 0.2197265625,
      "loss_num": 0.0233154296875,
      "loss_xval": 0.55859375,
      "num_input_tokens_seen": 183205300,
      "step": 3270
    },
    {
      "epoch": 7.285077951002227,
      "grad_norm": 20.71013832092285,
      "learning_rate": 1e-06,
      "loss": 0.4886,
      "num_input_tokens_seen": 183261900,
      "step": 3271
    },
    {
      "epoch": 7.285077951002227,
      "loss": 0.45970138907432556,
      "loss_ce": 0.00022872036788612604,
      "loss_iou": 0.1806640625,
      "loss_num": 0.01953125,
      "loss_xval": 0.458984375,
      "num_input_tokens_seen": 183261900,
      "step": 3271
    },
    {
      "epoch": 7.287305122494432,
      "grad_norm": 15.918169975280762,
      "learning_rate": 1e-06,
      "loss": 0.8508,
      "num_input_tokens_seen": 183317060,
      "step": 3272
    },
    {
      "epoch": 7.287305122494432,
      "loss": 0.6296719312667847,
      "loss_ce": 0.00015531133976764977,
      "loss_iou": 0.26171875,
      "loss_num": 0.0213623046875,
      "loss_xval": 0.62890625,
      "num_input_tokens_seen": 183317060,
      "step": 3272
    },
    {
      "epoch": 7.289532293986637,
      "grad_norm": 13.940874099731445,
      "learning_rate": 1e-06,
      "loss": 0.6049,
      "num_input_tokens_seen": 183372896,
      "step": 3273
    },
    {
      "epoch": 7.289532293986637,
      "loss": 0.6454752087593079,
      "loss_ce": 0.00021151437249500304,
      "loss_iou": 0.263671875,
      "loss_num": 0.0233154296875,
      "loss_xval": 0.64453125,
      "num_input_tokens_seen": 183372896,
      "step": 3273
    },
    {
      "epoch": 7.291759465478842,
      "grad_norm": 13.392332077026367,
      "learning_rate": 1e-06,
      "loss": 0.5429,
      "num_input_tokens_seen": 183431176,
      "step": 3274
    },
    {
      "epoch": 7.291759465478842,
      "loss": 0.5099714398384094,
      "loss_ce": 0.00020580792624969035,
      "loss_iou": 0.2314453125,
      "loss_num": 0.0093994140625,
      "loss_xval": 0.5078125,
      "num_input_tokens_seen": 183431176,
      "step": 3274
    },
    {
      "epoch": 7.293986636971047,
      "grad_norm": 31.767141342163086,
      "learning_rate": 1e-06,
      "loss": 0.6066,
      "num_input_tokens_seen": 183485436,
      "step": 3275
    },
    {
      "epoch": 7.293986636971047,
      "loss": 0.7388736605644226,
      "loss_ce": 0.00022620504023507237,
      "loss_iou": 0.287109375,
      "loss_num": 0.03271484375,
      "loss_xval": 0.73828125,
      "num_input_tokens_seen": 183485436,
      "step": 3275
    },
    {
      "epoch": 7.296213808463252,
      "grad_norm": 27.66456413269043,
      "learning_rate": 1e-06,
      "loss": 0.6369,
      "num_input_tokens_seen": 183542084,
      "step": 3276
    },
    {
      "epoch": 7.296213808463252,
      "loss": 0.4608932137489319,
      "loss_ce": 0.00019985750259365886,
      "loss_iou": 0.203125,
      "loss_num": 0.010986328125,
      "loss_xval": 0.4609375,
      "num_input_tokens_seen": 183542084,
      "step": 3276
    },
    {
      "epoch": 7.298440979955457,
      "grad_norm": 23.609827041625977,
      "learning_rate": 1e-06,
      "loss": 0.7595,
      "num_input_tokens_seen": 183597960,
      "step": 3277
    },
    {
      "epoch": 7.298440979955457,
      "loss": 0.683929443359375,
      "loss_ce": 0.0003357165842317045,
      "loss_iou": 0.3046875,
      "loss_num": 0.01446533203125,
      "loss_xval": 0.68359375,
      "num_input_tokens_seen": 183597960,
      "step": 3277
    },
    {
      "epoch": 7.3006681514476615,
      "grad_norm": 19.84480094909668,
      "learning_rate": 1e-06,
      "loss": 0.5777,
      "num_input_tokens_seen": 183655488,
      "step": 3278
    },
    {
      "epoch": 7.3006681514476615,
      "loss": 0.7001070976257324,
      "loss_ce": 0.00015596086450386792,
      "loss_iou": 0.3046875,
      "loss_num": 0.017822265625,
      "loss_xval": 0.69921875,
      "num_input_tokens_seen": 183655488,
      "step": 3278
    },
    {
      "epoch": 7.302895322939866,
      "grad_norm": 20.375669479370117,
      "learning_rate": 1e-06,
      "loss": 0.6575,
      "num_input_tokens_seen": 183712908,
      "step": 3279
    },
    {
      "epoch": 7.302895322939866,
      "loss": 0.7356512546539307,
      "loss_ce": 0.00029969203751534224,
      "loss_iou": 0.30859375,
      "loss_num": 0.0235595703125,
      "loss_xval": 0.734375,
      "num_input_tokens_seen": 183712908,
      "step": 3279
    },
    {
      "epoch": 7.305122494432071,
      "grad_norm": 24.645275115966797,
      "learning_rate": 1e-06,
      "loss": 0.6105,
      "num_input_tokens_seen": 183770608,
      "step": 3280
    },
    {
      "epoch": 7.305122494432071,
      "loss": 0.6635428667068481,
      "loss_ce": 0.0002127783081959933,
      "loss_iou": 0.30078125,
      "loss_num": 0.01239013671875,
      "loss_xval": 0.6640625,
      "num_input_tokens_seen": 183770608,
      "step": 3280
    },
    {
      "epoch": 7.307349665924276,
      "grad_norm": 14.512449264526367,
      "learning_rate": 1e-06,
      "loss": 0.7044,
      "num_input_tokens_seen": 183825576,
      "step": 3281
    },
    {
      "epoch": 7.307349665924276,
      "loss": 0.9417954087257385,
      "loss_ce": 0.00038921748637221754,
      "loss_iou": 0.384765625,
      "loss_num": 0.0341796875,
      "loss_xval": 0.94140625,
      "num_input_tokens_seen": 183825576,
      "step": 3281
    },
    {
      "epoch": 7.309576837416481,
      "grad_norm": 19.870107650756836,
      "learning_rate": 1e-06,
      "loss": 0.6247,
      "num_input_tokens_seen": 183881588,
      "step": 3282
    },
    {
      "epoch": 7.309576837416481,
      "loss": 0.8251048922538757,
      "loss_ce": 0.00015370306209661067,
      "loss_iou": 0.345703125,
      "loss_num": 0.026611328125,
      "loss_xval": 0.82421875,
      "num_input_tokens_seen": 183881588,
      "step": 3282
    },
    {
      "epoch": 7.311804008908686,
      "grad_norm": 55.68107604980469,
      "learning_rate": 1e-06,
      "loss": 0.5833,
      "num_input_tokens_seen": 183940448,
      "step": 3283
    },
    {
      "epoch": 7.311804008908686,
      "loss": 0.564154326915741,
      "loss_ce": 0.00018950334924738854,
      "loss_iou": 0.244140625,
      "loss_num": 0.0150146484375,
      "loss_xval": 0.5625,
      "num_input_tokens_seen": 183940448,
      "step": 3283
    },
    {
      "epoch": 7.314031180400891,
      "grad_norm": 22.890865325927734,
      "learning_rate": 1e-06,
      "loss": 0.5107,
      "num_input_tokens_seen": 183997440,
      "step": 3284
    },
    {
      "epoch": 7.314031180400891,
      "loss": 0.40723344683647156,
      "loss_ce": 0.000251030403887853,
      "loss_iou": 0.185546875,
      "loss_num": 0.00732421875,
      "loss_xval": 0.40625,
      "num_input_tokens_seen": 183997440,
      "step": 3284
    },
    {
      "epoch": 7.3162583518930955,
      "grad_norm": 38.493743896484375,
      "learning_rate": 1e-06,
      "loss": 0.7763,
      "num_input_tokens_seen": 184054300,
      "step": 3285
    },
    {
      "epoch": 7.3162583518930955,
      "loss": 0.7445439100265503,
      "loss_ce": 0.00015913081006146967,
      "loss_iou": 0.330078125,
      "loss_num": 0.01708984375,
      "loss_xval": 0.74609375,
      "num_input_tokens_seen": 184054300,
      "step": 3285
    },
    {
      "epoch": 7.3184855233853,
      "grad_norm": 12.925439834594727,
      "learning_rate": 1e-06,
      "loss": 0.4989,
      "num_input_tokens_seen": 184109764,
      "step": 3286
    },
    {
      "epoch": 7.3184855233853,
      "loss": 0.5402973294258118,
      "loss_ce": 0.00025825732154771686,
      "loss_iou": 0.240234375,
      "loss_num": 0.01171875,
      "loss_xval": 0.5390625,
      "num_input_tokens_seen": 184109764,
      "step": 3286
    },
    {
      "epoch": 7.320712694877505,
      "grad_norm": 14.311683654785156,
      "learning_rate": 1e-06,
      "loss": 0.7607,
      "num_input_tokens_seen": 184166476,
      "step": 3287
    },
    {
      "epoch": 7.320712694877505,
      "loss": 0.5723039507865906,
      "loss_ce": 0.0004045310488436371,
      "loss_iou": 0.255859375,
      "loss_num": 0.01214599609375,
      "loss_xval": 0.5703125,
      "num_input_tokens_seen": 184166476,
      "step": 3287
    },
    {
      "epoch": 7.32293986636971,
      "grad_norm": 21.740203857421875,
      "learning_rate": 1e-06,
      "loss": 0.7618,
      "num_input_tokens_seen": 184220904,
      "step": 3288
    },
    {
      "epoch": 7.32293986636971,
      "loss": 0.7254713773727417,
      "loss_ce": 0.00031269137980416417,
      "loss_iou": 0.283203125,
      "loss_num": 0.031982421875,
      "loss_xval": 0.7265625,
      "num_input_tokens_seen": 184220904,
      "step": 3288
    },
    {
      "epoch": 7.325167037861915,
      "grad_norm": 20.65580177307129,
      "learning_rate": 1e-06,
      "loss": 0.7546,
      "num_input_tokens_seen": 184277784,
      "step": 3289
    },
    {
      "epoch": 7.325167037861915,
      "loss": 0.6745328903198242,
      "loss_ce": 0.0002165070327464491,
      "loss_iou": 0.271484375,
      "loss_num": 0.0262451171875,
      "loss_xval": 0.67578125,
      "num_input_tokens_seen": 184277784,
      "step": 3289
    },
    {
      "epoch": 7.327394209354121,
      "grad_norm": 22.88839340209961,
      "learning_rate": 1e-06,
      "loss": 0.6154,
      "num_input_tokens_seen": 184334484,
      "step": 3290
    },
    {
      "epoch": 7.327394209354121,
      "loss": 0.6290842294692993,
      "loss_ce": 0.0004221061826683581,
      "loss_iou": 0.287109375,
      "loss_num": 0.01129150390625,
      "loss_xval": 0.62890625,
      "num_input_tokens_seen": 184334484,
      "step": 3290
    },
    {
      "epoch": 7.3296213808463255,
      "grad_norm": 19.395837783813477,
      "learning_rate": 1e-06,
      "loss": 0.8562,
      "num_input_tokens_seen": 184389964,
      "step": 3291
    },
    {
      "epoch": 7.3296213808463255,
      "loss": 0.7204995155334473,
      "loss_ce": 0.00028470673714764416,
      "loss_iou": 0.30078125,
      "loss_num": 0.023681640625,
      "loss_xval": 0.71875,
      "num_input_tokens_seen": 184389964,
      "step": 3291
    },
    {
      "epoch": 7.33184855233853,
      "grad_norm": 18.469728469848633,
      "learning_rate": 1e-06,
      "loss": 0.6447,
      "num_input_tokens_seen": 184441476,
      "step": 3292
    },
    {
      "epoch": 7.33184855233853,
      "loss": 0.6571934819221497,
      "loss_ce": 0.0002110537316184491,
      "loss_iou": 0.2890625,
      "loss_num": 0.0162353515625,
      "loss_xval": 0.65625,
      "num_input_tokens_seen": 184441476,
      "step": 3292
    },
    {
      "epoch": 7.334075723830735,
      "grad_norm": 21.28434944152832,
      "learning_rate": 1e-06,
      "loss": 0.637,
      "num_input_tokens_seen": 184499576,
      "step": 3293
    },
    {
      "epoch": 7.334075723830735,
      "loss": 0.7633267045021057,
      "loss_ce": 0.00026520711253397167,
      "loss_iou": 0.294921875,
      "loss_num": 0.034912109375,
      "loss_xval": 0.76171875,
      "num_input_tokens_seen": 184499576,
      "step": 3293
    },
    {
      "epoch": 7.33630289532294,
      "grad_norm": 14.56684398651123,
      "learning_rate": 1e-06,
      "loss": 0.7462,
      "num_input_tokens_seen": 184557016,
      "step": 3294
    },
    {
      "epoch": 7.33630289532294,
      "loss": 0.6941665410995483,
      "loss_ce": 0.0001968404685612768,
      "loss_iou": 0.30859375,
      "loss_num": 0.01519775390625,
      "loss_xval": 0.6953125,
      "num_input_tokens_seen": 184557016,
      "step": 3294
    },
    {
      "epoch": 7.338530066815145,
      "grad_norm": 33.30550003051758,
      "learning_rate": 1e-06,
      "loss": 0.7178,
      "num_input_tokens_seen": 184612596,
      "step": 3295
    },
    {
      "epoch": 7.338530066815145,
      "loss": 0.8899809122085571,
      "loss_ce": 0.00033249915577471256,
      "loss_iou": 0.349609375,
      "loss_num": 0.037841796875,
      "loss_xval": 0.890625,
      "num_input_tokens_seen": 184612596,
      "step": 3295
    },
    {
      "epoch": 7.34075723830735,
      "grad_norm": 77.08248138427734,
      "learning_rate": 1e-06,
      "loss": 0.5869,
      "num_input_tokens_seen": 184669968,
      "step": 3296
    },
    {
      "epoch": 7.34075723830735,
      "loss": 0.4362117350101471,
      "loss_ce": 0.0004207201418466866,
      "loss_iou": 0.189453125,
      "loss_num": 0.01123046875,
      "loss_xval": 0.435546875,
      "num_input_tokens_seen": 184669968,
      "step": 3296
    },
    {
      "epoch": 7.342984409799555,
      "grad_norm": 17.855411529541016,
      "learning_rate": 1e-06,
      "loss": 0.606,
      "num_input_tokens_seen": 184727100,
      "step": 3297
    },
    {
      "epoch": 7.342984409799555,
      "loss": 0.5758390426635742,
      "loss_ce": 0.00015543993504252285,
      "loss_iou": 0.2353515625,
      "loss_num": 0.0208740234375,
      "loss_xval": 0.57421875,
      "num_input_tokens_seen": 184727100,
      "step": 3297
    },
    {
      "epoch": 7.3452115812917596,
      "grad_norm": 18.587966918945312,
      "learning_rate": 1e-06,
      "loss": 0.5917,
      "num_input_tokens_seen": 184781776,
      "step": 3298
    },
    {
      "epoch": 7.3452115812917596,
      "loss": 0.4892728626728058,
      "loss_ce": 0.0003202102379873395,
      "loss_iou": 0.2109375,
      "loss_num": 0.01318359375,
      "loss_xval": 0.48828125,
      "num_input_tokens_seen": 184781776,
      "step": 3298
    },
    {
      "epoch": 7.347438752783964,
      "grad_norm": 45.11115646362305,
      "learning_rate": 1e-06,
      "loss": 0.5559,
      "num_input_tokens_seen": 184833932,
      "step": 3299
    },
    {
      "epoch": 7.347438752783964,
      "loss": 0.6328990459442139,
      "loss_ce": 0.00020860567747149616,
      "loss_iou": 0.271484375,
      "loss_num": 0.017822265625,
      "loss_xval": 0.6328125,
      "num_input_tokens_seen": 184833932,
      "step": 3299
    },
    {
      "epoch": 7.349665924276169,
      "grad_norm": 22.717544555664062,
      "learning_rate": 1e-06,
      "loss": 0.5208,
      "num_input_tokens_seen": 184888688,
      "step": 3300
    },
    {
      "epoch": 7.349665924276169,
      "loss": 0.46978557109832764,
      "loss_ce": 0.00018108604126609862,
      "loss_iou": 0.2099609375,
      "loss_num": 0.00994873046875,
      "loss_xval": 0.46875,
      "num_input_tokens_seen": 184888688,
      "step": 3300
    },
    {
      "epoch": 7.351893095768374,
      "grad_norm": 20.01310920715332,
      "learning_rate": 1e-06,
      "loss": 0.5374,
      "num_input_tokens_seen": 184943680,
      "step": 3301
    },
    {
      "epoch": 7.351893095768374,
      "loss": 0.451114296913147,
      "loss_ce": 0.00018655930762179196,
      "loss_iou": 0.1767578125,
      "loss_num": 0.01953125,
      "loss_xval": 0.451171875,
      "num_input_tokens_seen": 184943680,
      "step": 3301
    },
    {
      "epoch": 7.354120267260579,
      "grad_norm": 36.787132263183594,
      "learning_rate": 1e-06,
      "loss": 0.8746,
      "num_input_tokens_seen": 184996220,
      "step": 3302
    },
    {
      "epoch": 7.354120267260579,
      "loss": 0.8494809865951538,
      "loss_ce": 0.0003598902840167284,
      "loss_iou": 0.33984375,
      "loss_num": 0.033447265625,
      "loss_xval": 0.84765625,
      "num_input_tokens_seen": 184996220,
      "step": 3302
    },
    {
      "epoch": 7.356347438752784,
      "grad_norm": 22.735063552856445,
      "learning_rate": 1e-06,
      "loss": 0.7577,
      "num_input_tokens_seen": 185047952,
      "step": 3303
    },
    {
      "epoch": 7.356347438752784,
      "loss": 0.5736503601074219,
      "loss_ce": 0.00016400867025367916,
      "loss_iou": 0.255859375,
      "loss_num": 0.01214599609375,
      "loss_xval": 0.57421875,
      "num_input_tokens_seen": 185047952,
      "step": 3303
    },
    {
      "epoch": 7.358574610244989,
      "grad_norm": 15.140356063842773,
      "learning_rate": 1e-06,
      "loss": 0.6265,
      "num_input_tokens_seen": 185104504,
      "step": 3304
    },
    {
      "epoch": 7.358574610244989,
      "loss": 0.8113583922386169,
      "loss_ce": 0.00020119798136875033,
      "loss_iou": 0.34375,
      "loss_num": 0.0247802734375,
      "loss_xval": 0.8125,
      "num_input_tokens_seen": 185104504,
      "step": 3304
    },
    {
      "epoch": 7.360801781737194,
      "grad_norm": 25.078630447387695,
      "learning_rate": 1e-06,
      "loss": 0.7108,
      "num_input_tokens_seen": 185162056,
      "step": 3305
    },
    {
      "epoch": 7.360801781737194,
      "loss": 0.7138417959213257,
      "loss_ce": 0.00021878939878661186,
      "loss_iou": 0.3125,
      "loss_num": 0.017822265625,
      "loss_xval": 0.71484375,
      "num_input_tokens_seen": 185162056,
      "step": 3305
    },
    {
      "epoch": 7.363028953229398,
      "grad_norm": 21.57823371887207,
      "learning_rate": 1e-06,
      "loss": 0.7628,
      "num_input_tokens_seen": 185218544,
      "step": 3306
    },
    {
      "epoch": 7.363028953229398,
      "loss": 1.0595169067382812,
      "loss_ce": 0.00019062710634898394,
      "loss_iou": 0.466796875,
      "loss_num": 0.0252685546875,
      "loss_xval": 1.0625,
      "num_input_tokens_seen": 185218544,
      "step": 3306
    },
    {
      "epoch": 7.365256124721603,
      "grad_norm": 17.32356071472168,
      "learning_rate": 1e-06,
      "loss": 0.574,
      "num_input_tokens_seen": 185276144,
      "step": 3307
    },
    {
      "epoch": 7.365256124721603,
      "loss": 0.49967584013938904,
      "loss_ce": 0.00016412066179327667,
      "loss_iou": 0.21484375,
      "loss_num": 0.01397705078125,
      "loss_xval": 0.5,
      "num_input_tokens_seen": 185276144,
      "step": 3307
    },
    {
      "epoch": 7.367483296213808,
      "grad_norm": 17.607749938964844,
      "learning_rate": 1e-06,
      "loss": 0.6742,
      "num_input_tokens_seen": 185333240,
      "step": 3308
    },
    {
      "epoch": 7.367483296213808,
      "loss": 0.6454328298568726,
      "loss_ce": 0.00016912329010665417,
      "loss_iou": 0.27734375,
      "loss_num": 0.01806640625,
      "loss_xval": 0.64453125,
      "num_input_tokens_seen": 185333240,
      "step": 3308
    },
    {
      "epoch": 7.369710467706013,
      "grad_norm": 16.388914108276367,
      "learning_rate": 1e-06,
      "loss": 0.5042,
      "num_input_tokens_seen": 185390156,
      "step": 3309
    },
    {
      "epoch": 7.369710467706013,
      "loss": 0.4991820454597473,
      "loss_ce": 0.00021965075575280935,
      "loss_iou": 0.2275390625,
      "loss_num": 0.009033203125,
      "loss_xval": 0.498046875,
      "num_input_tokens_seen": 185390156,
      "step": 3309
    },
    {
      "epoch": 7.371937639198218,
      "grad_norm": 18.82445526123047,
      "learning_rate": 1e-06,
      "loss": 0.6255,
      "num_input_tokens_seen": 185443792,
      "step": 3310
    },
    {
      "epoch": 7.371937639198218,
      "loss": 0.8694823980331421,
      "loss_ce": 0.00034184064134024084,
      "loss_iou": 0.3671875,
      "loss_num": 0.02685546875,
      "loss_xval": 0.8671875,
      "num_input_tokens_seen": 185443792,
      "step": 3310
    },
    {
      "epoch": 7.374164810690424,
      "grad_norm": 23.051231384277344,
      "learning_rate": 1e-06,
      "loss": 0.5239,
      "num_input_tokens_seen": 185500112,
      "step": 3311
    },
    {
      "epoch": 7.374164810690424,
      "loss": 0.6354948282241821,
      "loss_ce": 0.00024095734988804907,
      "loss_iou": 0.2734375,
      "loss_num": 0.01806640625,
      "loss_xval": 0.63671875,
      "num_input_tokens_seen": 185500112,
      "step": 3311
    },
    {
      "epoch": 7.3763919821826285,
      "grad_norm": 12.503715515136719,
      "learning_rate": 1e-06,
      "loss": 0.6206,
      "num_input_tokens_seen": 185556836,
      "step": 3312
    },
    {
      "epoch": 7.3763919821826285,
      "loss": 0.6298484206199646,
      "loss_ce": 0.00020972295897081494,
      "loss_iou": 0.279296875,
      "loss_num": 0.01434326171875,
      "loss_xval": 0.62890625,
      "num_input_tokens_seen": 185556836,
      "step": 3312
    },
    {
      "epoch": 7.378619153674833,
      "grad_norm": 39.04258728027344,
      "learning_rate": 1e-06,
      "loss": 0.7489,
      "num_input_tokens_seen": 185611472,
      "step": 3313
    },
    {
      "epoch": 7.378619153674833,
      "loss": 0.6686286330223083,
      "loss_ce": 0.000171576626598835,
      "loss_iou": 0.279296875,
      "loss_num": 0.0216064453125,
      "loss_xval": 0.66796875,
      "num_input_tokens_seen": 185611472,
      "step": 3313
    },
    {
      "epoch": 7.380846325167038,
      "grad_norm": 20.53398323059082,
      "learning_rate": 1e-06,
      "loss": 0.5664,
      "num_input_tokens_seen": 185664932,
      "step": 3314
    },
    {
      "epoch": 7.380846325167038,
      "loss": 0.5182798504829407,
      "loss_ce": 0.00021345698041841388,
      "loss_iou": 0.2265625,
      "loss_num": 0.01275634765625,
      "loss_xval": 0.51953125,
      "num_input_tokens_seen": 185664932,
      "step": 3314
    },
    {
      "epoch": 7.383073496659243,
      "grad_norm": 17.228731155395508,
      "learning_rate": 1e-06,
      "loss": 0.7039,
      "num_input_tokens_seen": 185719144,
      "step": 3315
    },
    {
      "epoch": 7.383073496659243,
      "loss": 0.6190841197967529,
      "loss_ce": 0.00018761484534479678,
      "loss_iou": 0.2890625,
      "loss_num": 0.00836181640625,
      "loss_xval": 0.6171875,
      "num_input_tokens_seen": 185719144,
      "step": 3315
    },
    {
      "epoch": 7.385300668151448,
      "grad_norm": 25.301191329956055,
      "learning_rate": 1e-06,
      "loss": 0.6395,
      "num_input_tokens_seen": 185773492,
      "step": 3316
    },
    {
      "epoch": 7.385300668151448,
      "loss": 0.483335018157959,
      "loss_ce": 0.00018073133833240718,
      "loss_iou": 0.1982421875,
      "loss_num": 0.017578125,
      "loss_xval": 0.482421875,
      "num_input_tokens_seen": 185773492,
      "step": 3316
    },
    {
      "epoch": 7.387527839643653,
      "grad_norm": 21.02423095703125,
      "learning_rate": 1e-06,
      "loss": 0.4614,
      "num_input_tokens_seen": 185830376,
      "step": 3317
    },
    {
      "epoch": 7.387527839643653,
      "loss": 0.5402176380157471,
      "loss_ce": 0.00017858328646980226,
      "loss_iou": 0.2421875,
      "loss_num": 0.010986328125,
      "loss_xval": 0.5390625,
      "num_input_tokens_seen": 185830376,
      "step": 3317
    },
    {
      "epoch": 7.389755011135858,
      "grad_norm": 17.576339721679688,
      "learning_rate": 1e-06,
      "loss": 0.607,
      "num_input_tokens_seen": 185884160,
      "step": 3318
    },
    {
      "epoch": 7.389755011135858,
      "loss": 0.7509297728538513,
      "loss_ce": 0.00019736467220354825,
      "loss_iou": 0.33203125,
      "loss_num": 0.0172119140625,
      "loss_xval": 0.75,
      "num_input_tokens_seen": 185884160,
      "step": 3318
    },
    {
      "epoch": 7.3919821826280625,
      "grad_norm": 26.350475311279297,
      "learning_rate": 1e-06,
      "loss": 0.5901,
      "num_input_tokens_seen": 185939524,
      "step": 3319
    },
    {
      "epoch": 7.3919821826280625,
      "loss": 0.5638967156410217,
      "loss_ce": 0.0001760525774443522,
      "loss_iou": 0.2373046875,
      "loss_num": 0.017822265625,
      "loss_xval": 0.5625,
      "num_input_tokens_seen": 185939524,
      "step": 3319
    },
    {
      "epoch": 7.394209354120267,
      "grad_norm": 16.871248245239258,
      "learning_rate": 1e-06,
      "loss": 0.6936,
      "num_input_tokens_seen": 185996464,
      "step": 3320
    },
    {
      "epoch": 7.394209354120267,
      "loss": 0.6505089998245239,
      "loss_ce": 0.0007286987965926528,
      "loss_iou": 0.28125,
      "loss_num": 0.01708984375,
      "loss_xval": 0.6484375,
      "num_input_tokens_seen": 185996464,
      "step": 3320
    },
    {
      "epoch": 7.396436525612472,
      "grad_norm": 28.474199295043945,
      "learning_rate": 1e-06,
      "loss": 0.5359,
      "num_input_tokens_seen": 186051964,
      "step": 3321
    },
    {
      "epoch": 7.396436525612472,
      "loss": 0.5825638771057129,
      "loss_ce": 0.00016640947433188558,
      "loss_iou": 0.25390625,
      "loss_num": 0.0146484375,
      "loss_xval": 0.58203125,
      "num_input_tokens_seen": 186051964,
      "step": 3321
    },
    {
      "epoch": 7.398663697104677,
      "grad_norm": 15.175790786743164,
      "learning_rate": 1e-06,
      "loss": 0.5039,
      "num_input_tokens_seen": 186106580,
      "step": 3322
    },
    {
      "epoch": 7.398663697104677,
      "loss": 0.4596654176712036,
      "loss_ce": 0.00019277595856692642,
      "loss_iou": 0.203125,
      "loss_num": 0.0106201171875,
      "loss_xval": 0.458984375,
      "num_input_tokens_seen": 186106580,
      "step": 3322
    },
    {
      "epoch": 7.400890868596882,
      "grad_norm": 20.897014617919922,
      "learning_rate": 1e-06,
      "loss": 0.6625,
      "num_input_tokens_seen": 186163020,
      "step": 3323
    },
    {
      "epoch": 7.400890868596882,
      "loss": 0.5817232131958008,
      "loss_ce": 0.00018024235032498837,
      "loss_iou": 0.265625,
      "loss_num": 0.009765625,
      "loss_xval": 0.58203125,
      "num_input_tokens_seen": 186163020,
      "step": 3323
    },
    {
      "epoch": 7.403118040089087,
      "grad_norm": 17.83477210998535,
      "learning_rate": 1e-06,
      "loss": 0.5627,
      "num_input_tokens_seen": 186218364,
      "step": 3324
    },
    {
      "epoch": 7.403118040089087,
      "loss": 0.580369234085083,
      "loss_ce": 0.0002911181654781103,
      "loss_iou": 0.2294921875,
      "loss_num": 0.0242919921875,
      "loss_xval": 0.578125,
      "num_input_tokens_seen": 186218364,
      "step": 3324
    },
    {
      "epoch": 7.405345211581292,
      "grad_norm": 18.244247436523438,
      "learning_rate": 1e-06,
      "loss": 0.7525,
      "num_input_tokens_seen": 186271824,
      "step": 3325
    },
    {
      "epoch": 7.405345211581292,
      "loss": 0.5565927028656006,
      "loss_ce": 0.00019619996601250023,
      "loss_iou": 0.25,
      "loss_num": 0.01092529296875,
      "loss_xval": 0.5546875,
      "num_input_tokens_seen": 186271824,
      "step": 3325
    },
    {
      "epoch": 7.4075723830734965,
      "grad_norm": 25.529314041137695,
      "learning_rate": 1e-06,
      "loss": 0.7204,
      "num_input_tokens_seen": 186326604,
      "step": 3326
    },
    {
      "epoch": 7.4075723830734965,
      "loss": 0.7397571802139282,
      "loss_ce": 0.0002552796504460275,
      "loss_iou": 0.32421875,
      "loss_num": 0.0181884765625,
      "loss_xval": 0.73828125,
      "num_input_tokens_seen": 186326604,
      "step": 3326
    },
    {
      "epoch": 7.409799554565701,
      "grad_norm": 15.39941692352295,
      "learning_rate": 1e-06,
      "loss": 0.562,
      "num_input_tokens_seen": 186383040,
      "step": 3327
    },
    {
      "epoch": 7.409799554565701,
      "loss": 0.6352710723876953,
      "loss_ce": 0.00026131048798561096,
      "loss_iou": 0.27734375,
      "loss_num": 0.016357421875,
      "loss_xval": 0.63671875,
      "num_input_tokens_seen": 186383040,
      "step": 3327
    },
    {
      "epoch": 7.412026726057906,
      "grad_norm": 14.803224563598633,
      "learning_rate": 1e-06,
      "loss": 0.641,
      "num_input_tokens_seen": 186435624,
      "step": 3328
    },
    {
      "epoch": 7.412026726057906,
      "loss": 0.512509286403656,
      "loss_ce": 0.0005464061396196485,
      "loss_iou": 0.2080078125,
      "loss_num": 0.0191650390625,
      "loss_xval": 0.51171875,
      "num_input_tokens_seen": 186435624,
      "step": 3328
    },
    {
      "epoch": 7.414253897550111,
      "grad_norm": 19.84122085571289,
      "learning_rate": 1e-06,
      "loss": 0.6622,
      "num_input_tokens_seen": 186492116,
      "step": 3329
    },
    {
      "epoch": 7.414253897550111,
      "loss": 0.579806923866272,
      "loss_ce": 0.00021710841974709183,
      "loss_iou": 0.2490234375,
      "loss_num": 0.0162353515625,
      "loss_xval": 0.578125,
      "num_input_tokens_seen": 186492116,
      "step": 3329
    },
    {
      "epoch": 7.416481069042316,
      "grad_norm": 18.60784149169922,
      "learning_rate": 1e-06,
      "loss": 0.7456,
      "num_input_tokens_seen": 186549676,
      "step": 3330
    },
    {
      "epoch": 7.416481069042316,
      "loss": 0.7443963885307312,
      "loss_ce": 0.0002557601546868682,
      "loss_iou": 0.302734375,
      "loss_num": 0.0274658203125,
      "loss_xval": 0.7421875,
      "num_input_tokens_seen": 186549676,
      "step": 3330
    },
    {
      "epoch": 7.418708240534521,
      "grad_norm": 21.594797134399414,
      "learning_rate": 1e-06,
      "loss": 0.7149,
      "num_input_tokens_seen": 186604212,
      "step": 3331
    },
    {
      "epoch": 7.418708240534521,
      "loss": 0.7223727107048035,
      "loss_ce": 0.00020475327619351447,
      "loss_iou": 0.314453125,
      "loss_num": 0.0189208984375,
      "loss_xval": 0.72265625,
      "num_input_tokens_seen": 186604212,
      "step": 3331
    },
    {
      "epoch": 7.420935412026726,
      "grad_norm": 20.443817138671875,
      "learning_rate": 1e-06,
      "loss": 0.6048,
      "num_input_tokens_seen": 186658836,
      "step": 3332
    },
    {
      "epoch": 7.420935412026726,
      "loss": 0.5924441814422607,
      "loss_ce": 0.000281118496786803,
      "loss_iou": 0.24609375,
      "loss_num": 0.0198974609375,
      "loss_xval": 0.59375,
      "num_input_tokens_seen": 186658836,
      "step": 3332
    },
    {
      "epoch": 7.4231625835189305,
      "grad_norm": 22.519840240478516,
      "learning_rate": 1e-06,
      "loss": 0.4982,
      "num_input_tokens_seen": 186716272,
      "step": 3333
    },
    {
      "epoch": 7.4231625835189305,
      "loss": 0.5731725096702576,
      "loss_ce": 0.0002965159364975989,
      "loss_iou": 0.263671875,
      "loss_num": 0.00927734375,
      "loss_xval": 0.57421875,
      "num_input_tokens_seen": 186716272,
      "step": 3333
    },
    {
      "epoch": 7.425389755011135,
      "grad_norm": 16.776288986206055,
      "learning_rate": 1e-06,
      "loss": 0.5875,
      "num_input_tokens_seen": 186769964,
      "step": 3334
    },
    {
      "epoch": 7.425389755011135,
      "loss": 0.5091996192932129,
      "loss_ce": 0.0001664264709688723,
      "loss_iou": 0.205078125,
      "loss_num": 0.019775390625,
      "loss_xval": 0.5078125,
      "num_input_tokens_seen": 186769964,
      "step": 3334
    },
    {
      "epoch": 7.427616926503341,
      "grad_norm": 19.484222412109375,
      "learning_rate": 1e-06,
      "loss": 0.5844,
      "num_input_tokens_seen": 186826344,
      "step": 3335
    },
    {
      "epoch": 7.427616926503341,
      "loss": 0.6198055744171143,
      "loss_ce": 0.0001766337372828275,
      "loss_iou": 0.259765625,
      "loss_num": 0.0205078125,
      "loss_xval": 0.62109375,
      "num_input_tokens_seen": 186826344,
      "step": 3335
    },
    {
      "epoch": 7.429844097995546,
      "grad_norm": 17.908937454223633,
      "learning_rate": 1e-06,
      "loss": 0.624,
      "num_input_tokens_seen": 186881588,
      "step": 3336
    },
    {
      "epoch": 7.429844097995546,
      "loss": 0.6411352753639221,
      "loss_ce": 0.0002661352336872369,
      "loss_iou": 0.279296875,
      "loss_num": 0.0164794921875,
      "loss_xval": 0.640625,
      "num_input_tokens_seen": 186881588,
      "step": 3336
    },
    {
      "epoch": 7.432071269487751,
      "grad_norm": 24.69992446899414,
      "learning_rate": 1e-06,
      "loss": 0.5701,
      "num_input_tokens_seen": 186939068,
      "step": 3337
    },
    {
      "epoch": 7.432071269487751,
      "loss": 0.5966298580169678,
      "loss_ce": 0.00019431284454185516,
      "loss_iou": 0.2392578125,
      "loss_num": 0.0235595703125,
      "loss_xval": 0.59765625,
      "num_input_tokens_seen": 186939068,
      "step": 3337
    },
    {
      "epoch": 7.434298440979956,
      "grad_norm": 14.345844268798828,
      "learning_rate": 1e-06,
      "loss": 0.4817,
      "num_input_tokens_seen": 186996924,
      "step": 3338
    },
    {
      "epoch": 7.434298440979956,
      "loss": 0.6793899536132812,
      "loss_ce": 0.00019069564586970955,
      "loss_iou": 0.296875,
      "loss_num": 0.0169677734375,
      "loss_xval": 0.6796875,
      "num_input_tokens_seen": 186996924,
      "step": 3338
    },
    {
      "epoch": 7.436525612472161,
      "grad_norm": 14.285273551940918,
      "learning_rate": 1e-06,
      "loss": 0.5295,
      "num_input_tokens_seen": 187053760,
      "step": 3339
    },
    {
      "epoch": 7.436525612472161,
      "loss": 0.7244080305099487,
      "loss_ce": 0.0001648097822908312,
      "loss_iou": 0.3125,
      "loss_num": 0.0201416015625,
      "loss_xval": 0.72265625,
      "num_input_tokens_seen": 187053760,
      "step": 3339
    },
    {
      "epoch": 7.4387527839643655,
      "grad_norm": 16.611595153808594,
      "learning_rate": 1e-06,
      "loss": 0.8745,
      "num_input_tokens_seen": 187108896,
      "step": 3340
    },
    {
      "epoch": 7.4387527839643655,
      "loss": 1.1862220764160156,
      "loss_ce": 0.0001869781408458948,
      "loss_iou": 0.498046875,
      "loss_num": 0.037353515625,
      "loss_xval": 1.1875,
      "num_input_tokens_seen": 187108896,
      "step": 3340
    },
    {
      "epoch": 7.44097995545657,
      "grad_norm": 22.56696891784668,
      "learning_rate": 1e-06,
      "loss": 0.6369,
      "num_input_tokens_seen": 187166280,
      "step": 3341
    },
    {
      "epoch": 7.44097995545657,
      "loss": 0.7133273482322693,
      "loss_ce": 0.00019260836415924132,
      "loss_iou": 0.298828125,
      "loss_num": 0.022705078125,
      "loss_xval": 0.71484375,
      "num_input_tokens_seen": 187166280,
      "step": 3341
    },
    {
      "epoch": 7.443207126948775,
      "grad_norm": 31.73601531982422,
      "learning_rate": 1e-06,
      "loss": 0.7095,
      "num_input_tokens_seen": 187223724,
      "step": 3342
    },
    {
      "epoch": 7.443207126948775,
      "loss": 0.5013738870620728,
      "loss_ce": 0.00027527124620974064,
      "loss_iou": 0.2109375,
      "loss_num": 0.0157470703125,
      "loss_xval": 0.5,
      "num_input_tokens_seen": 187223724,
      "step": 3342
    },
    {
      "epoch": 7.44543429844098,
      "grad_norm": 15.37706184387207,
      "learning_rate": 1e-06,
      "loss": 0.457,
      "num_input_tokens_seen": 187281536,
      "step": 3343
    },
    {
      "epoch": 7.44543429844098,
      "loss": 0.43623799085617065,
      "loss_ce": 0.00020283354388084263,
      "loss_iou": 0.185546875,
      "loss_num": 0.01287841796875,
      "loss_xval": 0.435546875,
      "num_input_tokens_seen": 187281536,
      "step": 3343
    },
    {
      "epoch": 7.447661469933185,
      "grad_norm": 14.091643333435059,
      "learning_rate": 1e-06,
      "loss": 0.7446,
      "num_input_tokens_seen": 187339076,
      "step": 3344
    },
    {
      "epoch": 7.447661469933185,
      "loss": 0.8033976554870605,
      "loss_ce": 0.00026650671497918665,
      "loss_iou": 0.318359375,
      "loss_num": 0.033203125,
      "loss_xval": 0.8046875,
      "num_input_tokens_seen": 187339076,
      "step": 3344
    },
    {
      "epoch": 7.44988864142539,
      "grad_norm": 17.009387969970703,
      "learning_rate": 1e-06,
      "loss": 0.6225,
      "num_input_tokens_seen": 187395496,
      "step": 3345
    },
    {
      "epoch": 7.44988864142539,
      "loss": 0.7989903688430786,
      "loss_ce": 0.001260860008187592,
      "loss_iou": 0.28125,
      "loss_num": 0.046875,
      "loss_xval": 0.796875,
      "num_input_tokens_seen": 187395496,
      "step": 3345
    },
    {
      "epoch": 7.452115812917595,
      "grad_norm": 16.3013916015625,
      "learning_rate": 1e-06,
      "loss": 0.7418,
      "num_input_tokens_seen": 187452228,
      "step": 3346
    },
    {
      "epoch": 7.452115812917595,
      "loss": 0.7919469475746155,
      "loss_ce": 0.0001989098673220724,
      "loss_iou": 0.3359375,
      "loss_num": 0.024169921875,
      "loss_xval": 0.79296875,
      "num_input_tokens_seen": 187452228,
      "step": 3346
    },
    {
      "epoch": 7.4543429844097995,
      "grad_norm": 25.37590217590332,
      "learning_rate": 1e-06,
      "loss": 0.6981,
      "num_input_tokens_seen": 187507640,
      "step": 3347
    },
    {
      "epoch": 7.4543429844097995,
      "loss": 0.5329313278198242,
      "loss_ce": 0.0007047686958685517,
      "loss_iou": 0.2314453125,
      "loss_num": 0.013916015625,
      "loss_xval": 0.53125,
      "num_input_tokens_seen": 187507640,
      "step": 3347
    },
    {
      "epoch": 7.456570155902004,
      "grad_norm": 15.719435691833496,
      "learning_rate": 1e-06,
      "loss": 0.6663,
      "num_input_tokens_seen": 187564908,
      "step": 3348
    },
    {
      "epoch": 7.456570155902004,
      "loss": 0.7233518362045288,
      "loss_ce": 0.00020732844131998718,
      "loss_iou": 0.30078125,
      "loss_num": 0.0240478515625,
      "loss_xval": 0.72265625,
      "num_input_tokens_seen": 187564908,
      "step": 3348
    },
    {
      "epoch": 7.458797327394209,
      "grad_norm": 14.450872421264648,
      "learning_rate": 1e-06,
      "loss": 0.6857,
      "num_input_tokens_seen": 187616496,
      "step": 3349
    },
    {
      "epoch": 7.458797327394209,
      "loss": 0.5809314250946045,
      "loss_ce": 0.0006091895047575235,
      "loss_iou": 0.25390625,
      "loss_num": 0.0147705078125,
      "loss_xval": 0.58203125,
      "num_input_tokens_seen": 187616496,
      "step": 3349
    },
    {
      "epoch": 7.461024498886414,
      "grad_norm": 16.598936080932617,
      "learning_rate": 1e-06,
      "loss": 0.6052,
      "num_input_tokens_seen": 187670232,
      "step": 3350
    },
    {
      "epoch": 7.461024498886414,
      "loss": 0.479515016078949,
      "loss_ce": 0.00017542547720950097,
      "loss_iou": 0.20703125,
      "loss_num": 0.0130615234375,
      "loss_xval": 0.478515625,
      "num_input_tokens_seen": 187670232,
      "step": 3350
    },
    {
      "epoch": 7.463251670378619,
      "grad_norm": 29.393518447875977,
      "learning_rate": 1e-06,
      "loss": 0.6938,
      "num_input_tokens_seen": 187724992,
      "step": 3351
    },
    {
      "epoch": 7.463251670378619,
      "loss": 0.8341479897499084,
      "loss_ce": 0.00040772499050945044,
      "loss_iou": 0.375,
      "loss_num": 0.0167236328125,
      "loss_xval": 0.83203125,
      "num_input_tokens_seen": 187724992,
      "step": 3351
    },
    {
      "epoch": 7.465478841870824,
      "grad_norm": 17.579273223876953,
      "learning_rate": 1e-06,
      "loss": 0.6167,
      "num_input_tokens_seen": 187782612,
      "step": 3352
    },
    {
      "epoch": 7.465478841870824,
      "loss": 0.7424448132514954,
      "loss_ce": 0.0002573099918663502,
      "loss_iou": 0.31640625,
      "loss_num": 0.021728515625,
      "loss_xval": 0.7421875,
      "num_input_tokens_seen": 187782612,
      "step": 3352
    },
    {
      "epoch": 7.467706013363029,
      "grad_norm": 20.376893997192383,
      "learning_rate": 1e-06,
      "loss": 0.6322,
      "num_input_tokens_seen": 187838060,
      "step": 3353
    },
    {
      "epoch": 7.467706013363029,
      "loss": 0.6124618649482727,
      "loss_ce": 0.00040132386493496597,
      "loss_iou": 0.2578125,
      "loss_num": 0.019287109375,
      "loss_xval": 0.61328125,
      "num_input_tokens_seen": 187838060,
      "step": 3353
    },
    {
      "epoch": 7.4699331848552335,
      "grad_norm": 27.1313533782959,
      "learning_rate": 1e-06,
      "loss": 0.8042,
      "num_input_tokens_seen": 187891800,
      "step": 3354
    },
    {
      "epoch": 7.4699331848552335,
      "loss": 0.8244494795799255,
      "loss_ce": 0.00023073022020980716,
      "loss_iou": 0.3828125,
      "loss_num": 0.01165771484375,
      "loss_xval": 0.82421875,
      "num_input_tokens_seen": 187891800,
      "step": 3354
    },
    {
      "epoch": 7.472160356347438,
      "grad_norm": 17.113248825073242,
      "learning_rate": 1e-06,
      "loss": 0.5967,
      "num_input_tokens_seen": 187949628,
      "step": 3355
    },
    {
      "epoch": 7.472160356347438,
      "loss": 0.414481520652771,
      "loss_ce": 0.00017491632024757564,
      "loss_iou": 0.1767578125,
      "loss_num": 0.0123291015625,
      "loss_xval": 0.4140625,
      "num_input_tokens_seen": 187949628,
      "step": 3355
    },
    {
      "epoch": 7.474387527839644,
      "grad_norm": 34.95979309082031,
      "learning_rate": 1e-06,
      "loss": 0.7369,
      "num_input_tokens_seen": 188006312,
      "step": 3356
    },
    {
      "epoch": 7.474387527839644,
      "loss": 0.7557699680328369,
      "loss_ce": 0.00030729081481695175,
      "loss_iou": 0.298828125,
      "loss_num": 0.031982421875,
      "loss_xval": 0.75390625,
      "num_input_tokens_seen": 188006312,
      "step": 3356
    },
    {
      "epoch": 7.476614699331849,
      "grad_norm": 18.19429588317871,
      "learning_rate": 1e-06,
      "loss": 0.7208,
      "num_input_tokens_seen": 188062748,
      "step": 3357
    },
    {
      "epoch": 7.476614699331849,
      "loss": 0.5873679518699646,
      "loss_ce": 0.00027081643929705024,
      "loss_iou": 0.25,
      "loss_num": 0.01708984375,
      "loss_xval": 0.5859375,
      "num_input_tokens_seen": 188062748,
      "step": 3357
    },
    {
      "epoch": 7.478841870824054,
      "grad_norm": 25.23208236694336,
      "learning_rate": 1e-06,
      "loss": 0.462,
      "num_input_tokens_seen": 188118952,
      "step": 3358
    },
    {
      "epoch": 7.478841870824054,
      "loss": 0.6015306711196899,
      "loss_ce": 0.00021232501603662968,
      "loss_iou": 0.26171875,
      "loss_num": 0.0159912109375,
      "loss_xval": 0.6015625,
      "num_input_tokens_seen": 188118952,
      "step": 3358
    },
    {
      "epoch": 7.481069042316259,
      "grad_norm": 13.118025779724121,
      "learning_rate": 1e-06,
      "loss": 0.7661,
      "num_input_tokens_seen": 188172408,
      "step": 3359
    },
    {
      "epoch": 7.481069042316259,
      "loss": 0.7937259674072266,
      "loss_ce": 0.00026893243193626404,
      "loss_iou": 0.33984375,
      "loss_num": 0.0224609375,
      "loss_xval": 0.79296875,
      "num_input_tokens_seen": 188172408,
      "step": 3359
    },
    {
      "epoch": 7.4832962138084635,
      "grad_norm": 21.09063148498535,
      "learning_rate": 1e-06,
      "loss": 0.655,
      "num_input_tokens_seen": 188228848,
      "step": 3360
    },
    {
      "epoch": 7.4832962138084635,
      "loss": 0.7106525897979736,
      "loss_ce": 0.000203342831810005,
      "loss_iou": 0.271484375,
      "loss_num": 0.03369140625,
      "loss_xval": 0.7109375,
      "num_input_tokens_seen": 188228848,
      "step": 3360
    },
    {
      "epoch": 7.485523385300668,
      "grad_norm": 19.18509292602539,
      "learning_rate": 1e-06,
      "loss": 0.737,
      "num_input_tokens_seen": 188284528,
      "step": 3361
    },
    {
      "epoch": 7.485523385300668,
      "loss": 0.8408026695251465,
      "loss_ce": 0.00022645562421530485,
      "loss_iou": 0.359375,
      "loss_num": 0.024658203125,
      "loss_xval": 0.83984375,
      "num_input_tokens_seen": 188284528,
      "step": 3361
    },
    {
      "epoch": 7.487750556792873,
      "grad_norm": 21.143802642822266,
      "learning_rate": 1e-06,
      "loss": 0.6618,
      "num_input_tokens_seen": 188340712,
      "step": 3362
    },
    {
      "epoch": 7.487750556792873,
      "loss": 0.7464203238487244,
      "loss_ce": 0.00020453488104976714,
      "loss_iou": 0.33203125,
      "loss_num": 0.016357421875,
      "loss_xval": 0.74609375,
      "num_input_tokens_seen": 188340712,
      "step": 3362
    },
    {
      "epoch": 7.489977728285078,
      "grad_norm": 25.710002899169922,
      "learning_rate": 1e-06,
      "loss": 0.7553,
      "num_input_tokens_seen": 188395496,
      "step": 3363
    },
    {
      "epoch": 7.489977728285078,
      "loss": 0.898524820804596,
      "loss_ce": 0.00033146512578241527,
      "loss_iou": 0.373046875,
      "loss_num": 0.030517578125,
      "loss_xval": 0.8984375,
      "num_input_tokens_seen": 188395496,
      "step": 3363
    },
    {
      "epoch": 7.492204899777283,
      "grad_norm": 18.37621307373047,
      "learning_rate": 1e-06,
      "loss": 0.6617,
      "num_input_tokens_seen": 188451292,
      "step": 3364
    },
    {
      "epoch": 7.492204899777283,
      "loss": 0.6165107488632202,
      "loss_ce": 0.00029981633997522295,
      "loss_iou": 0.25,
      "loss_num": 0.0234375,
      "loss_xval": 0.6171875,
      "num_input_tokens_seen": 188451292,
      "step": 3364
    },
    {
      "epoch": 7.494432071269488,
      "grad_norm": 23.23649787902832,
      "learning_rate": 1e-06,
      "loss": 0.723,
      "num_input_tokens_seen": 188510116,
      "step": 3365
    },
    {
      "epoch": 7.494432071269488,
      "loss": 0.5230201482772827,
      "loss_ce": 0.00019301672000437975,
      "loss_iou": 0.232421875,
      "loss_num": 0.0115966796875,
      "loss_xval": 0.5234375,
      "num_input_tokens_seen": 188510116,
      "step": 3365
    },
    {
      "epoch": 7.496659242761693,
      "grad_norm": 19.028898239135742,
      "learning_rate": 1e-06,
      "loss": 0.6373,
      "num_input_tokens_seen": 188566168,
      "step": 3366
    },
    {
      "epoch": 7.496659242761693,
      "loss": 0.5705129504203796,
      "loss_ce": 0.00020043338008690625,
      "loss_iou": 0.255859375,
      "loss_num": 0.01165771484375,
      "loss_xval": 0.5703125,
      "num_input_tokens_seen": 188566168,
      "step": 3366
    },
    {
      "epoch": 7.498886414253898,
      "grad_norm": 60.06413650512695,
      "learning_rate": 1e-06,
      "loss": 0.6584,
      "num_input_tokens_seen": 188622556,
      "step": 3367
    },
    {
      "epoch": 7.498886414253898,
      "loss": 0.5773051977157593,
      "loss_ce": 0.00021782593103125691,
      "loss_iou": 0.248046875,
      "loss_num": 0.01611328125,
      "loss_xval": 0.578125,
      "num_input_tokens_seen": 188622556,
      "step": 3367
    },
    {
      "epoch": 7.501113585746102,
      "grad_norm": 29.536270141601562,
      "learning_rate": 1e-06,
      "loss": 0.67,
      "num_input_tokens_seen": 188675804,
      "step": 3368
    },
    {
      "epoch": 7.501113585746102,
      "loss": 0.7318795919418335,
      "loss_ce": 0.0001901389186969027,
      "loss_iou": 0.326171875,
      "loss_num": 0.015869140625,
      "loss_xval": 0.73046875,
      "num_input_tokens_seen": 188675804,
      "step": 3368
    },
    {
      "epoch": 7.503340757238307,
      "grad_norm": 19.667911529541016,
      "learning_rate": 1e-06,
      "loss": 0.7901,
      "num_input_tokens_seen": 188729532,
      "step": 3369
    },
    {
      "epoch": 7.503340757238307,
      "loss": 0.6141993999481201,
      "loss_ce": 0.00018573581473901868,
      "loss_iou": 0.275390625,
      "loss_num": 0.0123291015625,
      "loss_xval": 0.61328125,
      "num_input_tokens_seen": 188729532,
      "step": 3369
    },
    {
      "epoch": 7.505567928730512,
      "grad_norm": 19.440563201904297,
      "learning_rate": 1e-06,
      "loss": 0.6208,
      "num_input_tokens_seen": 188784192,
      "step": 3370
    },
    {
      "epoch": 7.505567928730512,
      "loss": 0.608165979385376,
      "loss_ce": 0.0002558160631451756,
      "loss_iou": 0.259765625,
      "loss_num": 0.0177001953125,
      "loss_xval": 0.609375,
      "num_input_tokens_seen": 188784192,
      "step": 3370
    },
    {
      "epoch": 7.507795100222717,
      "grad_norm": 20.731998443603516,
      "learning_rate": 1e-06,
      "loss": 0.6593,
      "num_input_tokens_seen": 188838624,
      "step": 3371
    },
    {
      "epoch": 7.507795100222717,
      "loss": 0.6856630444526672,
      "loss_ce": 0.00023825542302802205,
      "loss_iou": 0.322265625,
      "loss_num": 0.008056640625,
      "loss_xval": 0.68359375,
      "num_input_tokens_seen": 188838624,
      "step": 3371
    },
    {
      "epoch": 7.510022271714922,
      "grad_norm": 21.571271896362305,
      "learning_rate": 1e-06,
      "loss": 0.6617,
      "num_input_tokens_seen": 188896352,
      "step": 3372
    },
    {
      "epoch": 7.510022271714922,
      "loss": 0.7389297485351562,
      "loss_ce": 0.0001602266274858266,
      "loss_iou": 0.333984375,
      "loss_num": 0.01397705078125,
      "loss_xval": 0.73828125,
      "num_input_tokens_seen": 188896352,
      "step": 3372
    },
    {
      "epoch": 7.512249443207127,
      "grad_norm": 26.354827880859375,
      "learning_rate": 1e-06,
      "loss": 0.5974,
      "num_input_tokens_seen": 188953156,
      "step": 3373
    },
    {
      "epoch": 7.512249443207127,
      "loss": 0.5054886341094971,
      "loss_ce": 0.00023960010730661452,
      "loss_iou": 0.228515625,
      "loss_num": 0.00946044921875,
      "loss_xval": 0.50390625,
      "num_input_tokens_seen": 188953156,
      "step": 3373
    },
    {
      "epoch": 7.514476614699332,
      "grad_norm": 33.86724853515625,
      "learning_rate": 1e-06,
      "loss": 0.529,
      "num_input_tokens_seen": 189007540,
      "step": 3374
    },
    {
      "epoch": 7.514476614699332,
      "loss": 0.34347233176231384,
      "loss_ce": 0.0002106213360093534,
      "loss_iou": 0.1474609375,
      "loss_num": 0.0096435546875,
      "loss_xval": 0.34375,
      "num_input_tokens_seen": 189007540,
      "step": 3374
    },
    {
      "epoch": 7.5167037861915365,
      "grad_norm": 23.846332550048828,
      "learning_rate": 1e-06,
      "loss": 0.8868,
      "num_input_tokens_seen": 189063288,
      "step": 3375
    },
    {
      "epoch": 7.5167037861915365,
      "loss": 1.2485597133636475,
      "loss_ce": 0.00026864674873650074,
      "loss_iou": 0.51953125,
      "loss_num": 0.041748046875,
      "loss_xval": 1.25,
      "num_input_tokens_seen": 189063288,
      "step": 3375
    },
    {
      "epoch": 7.518930957683741,
      "grad_norm": 34.920448303222656,
      "learning_rate": 1e-06,
      "loss": 0.619,
      "num_input_tokens_seen": 189121948,
      "step": 3376
    },
    {
      "epoch": 7.518930957683741,
      "loss": 0.4434422254562378,
      "loss_ce": 0.0002049226895906031,
      "loss_iou": 0.1640625,
      "loss_num": 0.0233154296875,
      "loss_xval": 0.443359375,
      "num_input_tokens_seen": 189121948,
      "step": 3376
    },
    {
      "epoch": 7.521158129175946,
      "grad_norm": 45.45869445800781,
      "learning_rate": 1e-06,
      "loss": 0.6966,
      "num_input_tokens_seen": 189178620,
      "step": 3377
    },
    {
      "epoch": 7.521158129175946,
      "loss": 0.9557803869247437,
      "loss_ce": 0.00021394254872575402,
      "loss_iou": 0.376953125,
      "loss_num": 0.04052734375,
      "loss_xval": 0.95703125,
      "num_input_tokens_seen": 189178620,
      "step": 3377
    },
    {
      "epoch": 7.523385300668151,
      "grad_norm": 21.550317764282227,
      "learning_rate": 1e-06,
      "loss": 0.5571,
      "num_input_tokens_seen": 189234636,
      "step": 3378
    },
    {
      "epoch": 7.523385300668151,
      "loss": 0.6044266819953918,
      "loss_ce": 0.00017864728579297662,
      "loss_iou": 0.279296875,
      "loss_num": 0.00946044921875,
      "loss_xval": 0.60546875,
      "num_input_tokens_seen": 189234636,
      "step": 3378
    },
    {
      "epoch": 7.525612472160356,
      "grad_norm": 17.31205177307129,
      "learning_rate": 1e-06,
      "loss": 0.5324,
      "num_input_tokens_seen": 189293248,
      "step": 3379
    },
    {
      "epoch": 7.525612472160356,
      "loss": 0.5417789220809937,
      "loss_ce": 0.00027502982993610203,
      "loss_iou": 0.2275390625,
      "loss_num": 0.017333984375,
      "loss_xval": 0.54296875,
      "num_input_tokens_seen": 189293248,
      "step": 3379
    },
    {
      "epoch": 7.527839643652561,
      "grad_norm": 18.366722106933594,
      "learning_rate": 1e-06,
      "loss": 0.6955,
      "num_input_tokens_seen": 189351236,
      "step": 3380
    },
    {
      "epoch": 7.527839643652561,
      "loss": 0.6718355417251587,
      "loss_ce": 0.00020462839165702462,
      "loss_iou": 0.30859375,
      "loss_num": 0.01055908203125,
      "loss_xval": 0.671875,
      "num_input_tokens_seen": 189351236,
      "step": 3380
    },
    {
      "epoch": 7.5300668151447665,
      "grad_norm": 28.49734878540039,
      "learning_rate": 1e-06,
      "loss": 0.6942,
      "num_input_tokens_seen": 189408204,
      "step": 3381
    },
    {
      "epoch": 7.5300668151447665,
      "loss": 0.6963520050048828,
      "loss_ce": 0.00018498743884265423,
      "loss_iou": 0.306640625,
      "loss_num": 0.0162353515625,
      "loss_xval": 0.6953125,
      "num_input_tokens_seen": 189408204,
      "step": 3381
    },
    {
      "epoch": 7.532293986636971,
      "grad_norm": 21.700590133666992,
      "learning_rate": 1e-06,
      "loss": 0.4877,
      "num_input_tokens_seen": 189466496,
      "step": 3382
    },
    {
      "epoch": 7.532293986636971,
      "loss": 0.6716306209564209,
      "loss_ce": 0.00024389507598243654,
      "loss_iou": 0.28125,
      "loss_num": 0.02197265625,
      "loss_xval": 0.671875,
      "num_input_tokens_seen": 189466496,
      "step": 3382
    },
    {
      "epoch": 7.534521158129176,
      "grad_norm": 18.741180419921875,
      "learning_rate": 1e-06,
      "loss": 0.6188,
      "num_input_tokens_seen": 189524108,
      "step": 3383
    },
    {
      "epoch": 7.534521158129176,
      "loss": 0.4249844551086426,
      "loss_ce": 0.0003018215938936919,
      "loss_iou": 0.185546875,
      "loss_num": 0.0108642578125,
      "loss_xval": 0.423828125,
      "num_input_tokens_seen": 189524108,
      "step": 3383
    },
    {
      "epoch": 7.536748329621381,
      "grad_norm": 18.38628578186035,
      "learning_rate": 1e-06,
      "loss": 0.5341,
      "num_input_tokens_seen": 189580564,
      "step": 3384
    },
    {
      "epoch": 7.536748329621381,
      "loss": 0.6349685192108154,
      "loss_ce": 0.00020285190839786083,
      "loss_iou": 0.263671875,
      "loss_num": 0.0211181640625,
      "loss_xval": 0.6328125,
      "num_input_tokens_seen": 189580564,
      "step": 3384
    },
    {
      "epoch": 7.538975501113586,
      "grad_norm": 17.587261199951172,
      "learning_rate": 1e-06,
      "loss": 0.5419,
      "num_input_tokens_seen": 189633448,
      "step": 3385
    },
    {
      "epoch": 7.538975501113586,
      "loss": 0.590072512626648,
      "loss_ce": 0.00022877063020132482,
      "loss_iou": 0.2353515625,
      "loss_num": 0.023681640625,
      "loss_xval": 0.58984375,
      "num_input_tokens_seen": 189633448,
      "step": 3385
    },
    {
      "epoch": 7.541202672605791,
      "grad_norm": 23.872756958007812,
      "learning_rate": 1e-06,
      "loss": 0.5987,
      "num_input_tokens_seen": 189690868,
      "step": 3386
    },
    {
      "epoch": 7.541202672605791,
      "loss": 0.568671464920044,
      "loss_ce": 0.00019003619672730565,
      "loss_iou": 0.24609375,
      "loss_num": 0.01544189453125,
      "loss_xval": 0.5703125,
      "num_input_tokens_seen": 189690868,
      "step": 3386
    },
    {
      "epoch": 7.543429844097996,
      "grad_norm": 20.22979164123535,
      "learning_rate": 1e-06,
      "loss": 0.585,
      "num_input_tokens_seen": 189746736,
      "step": 3387
    },
    {
      "epoch": 7.543429844097996,
      "loss": 0.5246177911758423,
      "loss_ce": 0.00020370190031826496,
      "loss_iou": 0.2177734375,
      "loss_num": 0.0174560546875,
      "loss_xval": 0.5234375,
      "num_input_tokens_seen": 189746736,
      "step": 3387
    },
    {
      "epoch": 7.5456570155902005,
      "grad_norm": 18.41916275024414,
      "learning_rate": 1e-06,
      "loss": 0.6426,
      "num_input_tokens_seen": 189803012,
      "step": 3388
    },
    {
      "epoch": 7.5456570155902005,
      "loss": 0.621086597442627,
      "loss_ce": 0.0002370403817621991,
      "loss_iou": 0.28125,
      "loss_num": 0.01202392578125,
      "loss_xval": 0.62109375,
      "num_input_tokens_seen": 189803012,
      "step": 3388
    },
    {
      "epoch": 7.547884187082405,
      "grad_norm": 189.67298889160156,
      "learning_rate": 1e-06,
      "loss": 0.5411,
      "num_input_tokens_seen": 189859212,
      "step": 3389
    },
    {
      "epoch": 7.547884187082405,
      "loss": 0.5852431654930115,
      "loss_ce": 0.0002822045935317874,
      "loss_iou": 0.251953125,
      "loss_num": 0.01611328125,
      "loss_xval": 0.5859375,
      "num_input_tokens_seen": 189859212,
      "step": 3389
    },
    {
      "epoch": 7.55011135857461,
      "grad_norm": 15.547959327697754,
      "learning_rate": 1e-06,
      "loss": 0.8417,
      "num_input_tokens_seen": 189916504,
      "step": 3390
    },
    {
      "epoch": 7.55011135857461,
      "loss": 0.5157856941223145,
      "loss_ce": 0.00022171747696120292,
      "loss_iou": 0.2138671875,
      "loss_num": 0.0174560546875,
      "loss_xval": 0.515625,
      "num_input_tokens_seen": 189916504,
      "step": 3390
    },
    {
      "epoch": 7.552338530066815,
      "grad_norm": 26.860252380371094,
      "learning_rate": 1e-06,
      "loss": 0.7319,
      "num_input_tokens_seen": 189973312,
      "step": 3391
    },
    {
      "epoch": 7.552338530066815,
      "loss": 0.7284436225891113,
      "loss_ce": 0.00017210739315487444,
      "loss_iou": 0.28515625,
      "loss_num": 0.031494140625,
      "loss_xval": 0.7265625,
      "num_input_tokens_seen": 189973312,
      "step": 3391
    },
    {
      "epoch": 7.55456570155902,
      "grad_norm": 19.536853790283203,
      "learning_rate": 1e-06,
      "loss": 0.4387,
      "num_input_tokens_seen": 190030136,
      "step": 3392
    },
    {
      "epoch": 7.55456570155902,
      "loss": 0.4148935079574585,
      "loss_ce": 0.00022065457596909255,
      "loss_iou": 0.173828125,
      "loss_num": 0.013427734375,
      "loss_xval": 0.4140625,
      "num_input_tokens_seen": 190030136,
      "step": 3392
    },
    {
      "epoch": 7.556792873051225,
      "grad_norm": 20.840858459472656,
      "learning_rate": 1e-06,
      "loss": 0.6976,
      "num_input_tokens_seen": 190086148,
      "step": 3393
    },
    {
      "epoch": 7.556792873051225,
      "loss": 0.8776832222938538,
      "loss_ce": 0.00024186470545828342,
      "loss_iou": 0.369140625,
      "loss_num": 0.02783203125,
      "loss_xval": 0.87890625,
      "num_input_tokens_seen": 190086148,
      "step": 3393
    },
    {
      "epoch": 7.55902004454343,
      "grad_norm": 16.58432960510254,
      "learning_rate": 1e-06,
      "loss": 0.736,
      "num_input_tokens_seen": 190141764,
      "step": 3394
    },
    {
      "epoch": 7.55902004454343,
      "loss": 0.7458630204200745,
      "loss_ce": 0.0002575473044998944,
      "loss_iou": 0.291015625,
      "loss_num": 0.032470703125,
      "loss_xval": 0.74609375,
      "num_input_tokens_seen": 190141764,
      "step": 3394
    },
    {
      "epoch": 7.5612472160356345,
      "grad_norm": 15.23847770690918,
      "learning_rate": 1e-06,
      "loss": 0.7859,
      "num_input_tokens_seen": 190197356,
      "step": 3395
    },
    {
      "epoch": 7.5612472160356345,
      "loss": 0.8410194516181946,
      "loss_ce": 0.00019913521828129888,
      "loss_iou": 0.33984375,
      "loss_num": 0.032470703125,
      "loss_xval": 0.83984375,
      "num_input_tokens_seen": 190197356,
      "step": 3395
    },
    {
      "epoch": 7.563474387527839,
      "grad_norm": 17.79561424255371,
      "learning_rate": 1e-06,
      "loss": 0.5632,
      "num_input_tokens_seen": 190256636,
      "step": 3396
    },
    {
      "epoch": 7.563474387527839,
      "loss": 0.458886057138443,
      "loss_ce": 0.00026789220282807946,
      "loss_iou": 0.2158203125,
      "loss_num": 0.0054931640625,
      "loss_xval": 0.458984375,
      "num_input_tokens_seen": 190256636,
      "step": 3396
    },
    {
      "epoch": 7.565701559020044,
      "grad_norm": 15.726015090942383,
      "learning_rate": 1e-06,
      "loss": 0.4185,
      "num_input_tokens_seen": 190311620,
      "step": 3397
    },
    {
      "epoch": 7.565701559020044,
      "loss": 0.4230450987815857,
      "loss_ce": 0.0010480263736099005,
      "loss_iou": 0.1787109375,
      "loss_num": 0.0128173828125,
      "loss_xval": 0.421875,
      "num_input_tokens_seen": 190311620,
      "step": 3397
    },
    {
      "epoch": 7.567928730512249,
      "grad_norm": 25.931352615356445,
      "learning_rate": 1e-06,
      "loss": 0.7059,
      "num_input_tokens_seen": 190366792,
      "step": 3398
    },
    {
      "epoch": 7.567928730512249,
      "loss": 0.7654964923858643,
      "loss_ce": 0.0003598046605475247,
      "loss_iou": 0.337890625,
      "loss_num": 0.017822265625,
      "loss_xval": 0.765625,
      "num_input_tokens_seen": 190366792,
      "step": 3398
    },
    {
      "epoch": 7.570155902004454,
      "grad_norm": 21.16999053955078,
      "learning_rate": 1e-06,
      "loss": 0.7098,
      "num_input_tokens_seen": 190422324,
      "step": 3399
    },
    {
      "epoch": 7.570155902004454,
      "loss": 0.48839089274406433,
      "loss_ce": 0.0001706905895844102,
      "loss_iou": 0.212890625,
      "loss_num": 0.01214599609375,
      "loss_xval": 0.48828125,
      "num_input_tokens_seen": 190422324,
      "step": 3399
    },
    {
      "epoch": 7.57238307349666,
      "grad_norm": 23.648658752441406,
      "learning_rate": 1e-06,
      "loss": 0.4927,
      "num_input_tokens_seen": 190480076,
      "step": 3400
    },
    {
      "epoch": 7.57238307349666,
      "loss": 0.5361147522926331,
      "loss_ce": 0.00016503711231052876,
      "loss_iou": 0.2236328125,
      "loss_num": 0.0179443359375,
      "loss_xval": 0.53515625,
      "num_input_tokens_seen": 190480076,
      "step": 3400
    },
    {
      "epoch": 7.574610244988865,
      "grad_norm": 15.166297912597656,
      "learning_rate": 1e-06,
      "loss": 0.5387,
      "num_input_tokens_seen": 190535360,
      "step": 3401
    },
    {
      "epoch": 7.574610244988865,
      "loss": 0.6972156763076782,
      "loss_ce": 0.0001941586670000106,
      "loss_iou": 0.294921875,
      "loss_num": 0.021240234375,
      "loss_xval": 0.6953125,
      "num_input_tokens_seen": 190535360,
      "step": 3401
    },
    {
      "epoch": 7.5768374164810695,
      "grad_norm": 21.948759078979492,
      "learning_rate": 1e-06,
      "loss": 0.6897,
      "num_input_tokens_seen": 190589160,
      "step": 3402
    },
    {
      "epoch": 7.5768374164810695,
      "loss": 0.5853911638259888,
      "loss_ce": 0.0001860598858911544,
      "loss_iou": 0.26171875,
      "loss_num": 0.0126953125,
      "loss_xval": 0.5859375,
      "num_input_tokens_seen": 190589160,
      "step": 3402
    },
    {
      "epoch": 7.579064587973274,
      "grad_norm": 36.27766799926758,
      "learning_rate": 1e-06,
      "loss": 0.9359,
      "num_input_tokens_seen": 190644804,
      "step": 3403
    },
    {
      "epoch": 7.579064587973274,
      "loss": 0.9691004753112793,
      "loss_ce": 0.00022840322344563901,
      "loss_iou": 0.41015625,
      "loss_num": 0.030029296875,
      "loss_xval": 0.96875,
      "num_input_tokens_seen": 190644804,
      "step": 3403
    },
    {
      "epoch": 7.581291759465479,
      "grad_norm": 27.834016799926758,
      "learning_rate": 1e-06,
      "loss": 0.6663,
      "num_input_tokens_seen": 190702760,
      "step": 3404
    },
    {
      "epoch": 7.581291759465479,
      "loss": 0.7742569446563721,
      "loss_ce": 0.000209053119760938,
      "loss_iou": 0.345703125,
      "loss_num": 0.0167236328125,
      "loss_xval": 0.7734375,
      "num_input_tokens_seen": 190702760,
      "step": 3404
    },
    {
      "epoch": 7.583518930957684,
      "grad_norm": 18.967069625854492,
      "learning_rate": 1e-06,
      "loss": 0.5449,
      "num_input_tokens_seen": 190758396,
      "step": 3405
    },
    {
      "epoch": 7.583518930957684,
      "loss": 0.6806260943412781,
      "loss_ce": 0.0002061867417069152,
      "loss_iou": 0.296875,
      "loss_num": 0.0174560546875,
      "loss_xval": 0.6796875,
      "num_input_tokens_seen": 190758396,
      "step": 3405
    },
    {
      "epoch": 7.585746102449889,
      "grad_norm": 22.21535873413086,
      "learning_rate": 1e-06,
      "loss": 0.668,
      "num_input_tokens_seen": 190810616,
      "step": 3406
    },
    {
      "epoch": 7.585746102449889,
      "loss": 0.7401872873306274,
      "loss_ce": 0.0001970837765838951,
      "loss_iou": 0.32421875,
      "loss_num": 0.01806640625,
      "loss_xval": 0.73828125,
      "num_input_tokens_seen": 190810616,
      "step": 3406
    },
    {
      "epoch": 7.587973273942094,
      "grad_norm": 16.877315521240234,
      "learning_rate": 1e-06,
      "loss": 0.5691,
      "num_input_tokens_seen": 190866152,
      "step": 3407
    },
    {
      "epoch": 7.587973273942094,
      "loss": 0.545766294002533,
      "loss_ce": 0.00023407851404044777,
      "loss_iou": 0.232421875,
      "loss_num": 0.01611328125,
      "loss_xval": 0.546875,
      "num_input_tokens_seen": 190866152,
      "step": 3407
    },
    {
      "epoch": 7.590200445434299,
      "grad_norm": 29.44357681274414,
      "learning_rate": 1e-06,
      "loss": 0.5684,
      "num_input_tokens_seen": 190924720,
      "step": 3408
    },
    {
      "epoch": 7.590200445434299,
      "loss": 0.6174308061599731,
      "loss_ce": 0.0002433276386000216,
      "loss_iou": 0.25390625,
      "loss_num": 0.022216796875,
      "loss_xval": 0.6171875,
      "num_input_tokens_seen": 190924720,
      "step": 3408
    },
    {
      "epoch": 7.5924276169265035,
      "grad_norm": 22.013757705688477,
      "learning_rate": 1e-06,
      "loss": 0.6058,
      "num_input_tokens_seen": 190982220,
      "step": 3409
    },
    {
      "epoch": 7.5924276169265035,
      "loss": 0.6500968337059021,
      "loss_ce": 0.0001944488030858338,
      "loss_iou": 0.28515625,
      "loss_num": 0.015869140625,
      "loss_xval": 0.6484375,
      "num_input_tokens_seen": 190982220,
      "step": 3409
    },
    {
      "epoch": 7.594654788418708,
      "grad_norm": 19.142648696899414,
      "learning_rate": 1e-06,
      "loss": 0.6046,
      "num_input_tokens_seen": 191039340,
      "step": 3410
    },
    {
      "epoch": 7.594654788418708,
      "loss": 0.35581400990486145,
      "loss_ce": 0.0001621594128664583,
      "loss_iou": 0.1640625,
      "loss_num": 0.00531005859375,
      "loss_xval": 0.35546875,
      "num_input_tokens_seen": 191039340,
      "step": 3410
    },
    {
      "epoch": 7.596881959910913,
      "grad_norm": 19.273056030273438,
      "learning_rate": 1e-06,
      "loss": 0.6247,
      "num_input_tokens_seen": 191097160,
      "step": 3411
    },
    {
      "epoch": 7.596881959910913,
      "loss": 0.5918649435043335,
      "loss_ce": 0.00019010651158168912,
      "loss_iou": 0.2490234375,
      "loss_num": 0.0186767578125,
      "loss_xval": 0.58984375,
      "num_input_tokens_seen": 191097160,
      "step": 3411
    },
    {
      "epoch": 7.599109131403118,
      "grad_norm": 12.594167709350586,
      "learning_rate": 1e-06,
      "loss": 0.5376,
      "num_input_tokens_seen": 191153944,
      "step": 3412
    },
    {
      "epoch": 7.599109131403118,
      "loss": 0.5286160707473755,
      "loss_ce": 0.00017369385750498623,
      "loss_iou": 0.2314453125,
      "loss_num": 0.01287841796875,
      "loss_xval": 0.52734375,
      "num_input_tokens_seen": 191153944,
      "step": 3412
    },
    {
      "epoch": 7.601336302895323,
      "grad_norm": 21.295249938964844,
      "learning_rate": 1e-06,
      "loss": 0.7192,
      "num_input_tokens_seen": 191210128,
      "step": 3413
    },
    {
      "epoch": 7.601336302895323,
      "loss": 0.8668521642684937,
      "loss_ce": 0.00015292070747818798,
      "loss_iou": 0.3359375,
      "loss_num": 0.038330078125,
      "loss_xval": 0.8671875,
      "num_input_tokens_seen": 191210128,
      "step": 3413
    },
    {
      "epoch": 7.603563474387528,
      "grad_norm": 20.785724639892578,
      "learning_rate": 1e-06,
      "loss": 0.7145,
      "num_input_tokens_seen": 191267092,
      "step": 3414
    },
    {
      "epoch": 7.603563474387528,
      "loss": 0.6779365539550781,
      "loss_ce": 0.0002021369436988607,
      "loss_iou": 0.27734375,
      "loss_num": 0.0247802734375,
      "loss_xval": 0.6796875,
      "num_input_tokens_seen": 191267092,
      "step": 3414
    },
    {
      "epoch": 7.605790645879733,
      "grad_norm": 15.27004337310791,
      "learning_rate": 1e-06,
      "loss": 0.6446,
      "num_input_tokens_seen": 191323100,
      "step": 3415
    },
    {
      "epoch": 7.605790645879733,
      "loss": 0.7496963739395142,
      "loss_ce": 0.00018462681327946484,
      "loss_iou": 0.328125,
      "loss_num": 0.018310546875,
      "loss_xval": 0.75,
      "num_input_tokens_seen": 191323100,
      "step": 3415
    },
    {
      "epoch": 7.6080178173719375,
      "grad_norm": 34.315914154052734,
      "learning_rate": 1e-06,
      "loss": 0.5219,
      "num_input_tokens_seen": 191379484,
      "step": 3416
    },
    {
      "epoch": 7.6080178173719375,
      "loss": 0.558527410030365,
      "loss_ce": 0.00017779026529751718,
      "loss_iou": 0.25,
      "loss_num": 0.01190185546875,
      "loss_xval": 0.55859375,
      "num_input_tokens_seen": 191379484,
      "step": 3416
    },
    {
      "epoch": 7.610244988864142,
      "grad_norm": 16.876907348632812,
      "learning_rate": 1e-06,
      "loss": 0.875,
      "num_input_tokens_seen": 191434760,
      "step": 3417
    },
    {
      "epoch": 7.610244988864142,
      "loss": 1.0747464895248413,
      "loss_ce": 0.00028365751495584846,
      "loss_iou": 0.47265625,
      "loss_num": 0.02587890625,
      "loss_xval": 1.078125,
      "num_input_tokens_seen": 191434760,
      "step": 3417
    },
    {
      "epoch": 7.612472160356347,
      "grad_norm": 13.883415222167969,
      "learning_rate": 1e-06,
      "loss": 0.7571,
      "num_input_tokens_seen": 191492628,
      "step": 3418
    },
    {
      "epoch": 7.612472160356347,
      "loss": 0.7818785309791565,
      "loss_ce": 0.00026232993695884943,
      "loss_iou": 0.28515625,
      "loss_num": 0.04248046875,
      "loss_xval": 0.78125,
      "num_input_tokens_seen": 191492628,
      "step": 3418
    },
    {
      "epoch": 7.614699331848552,
      "grad_norm": 12.954174995422363,
      "learning_rate": 1e-06,
      "loss": 0.5369,
      "num_input_tokens_seen": 191549316,
      "step": 3419
    },
    {
      "epoch": 7.614699331848552,
      "loss": 0.5483337640762329,
      "loss_ce": 0.00023806520039215684,
      "loss_iou": 0.2236328125,
      "loss_num": 0.02001953125,
      "loss_xval": 0.546875,
      "num_input_tokens_seen": 191549316,
      "step": 3419
    },
    {
      "epoch": 7.616926503340757,
      "grad_norm": 14.275961875915527,
      "learning_rate": 1e-06,
      "loss": 0.7559,
      "num_input_tokens_seen": 191605056,
      "step": 3420
    },
    {
      "epoch": 7.616926503340757,
      "loss": 0.5231179594993591,
      "loss_ce": 0.0001687241019681096,
      "loss_iou": 0.23046875,
      "loss_num": 0.01239013671875,
      "loss_xval": 0.5234375,
      "num_input_tokens_seen": 191605056,
      "step": 3420
    },
    {
      "epoch": 7.619153674832962,
      "grad_norm": 18.700197219848633,
      "learning_rate": 1e-06,
      "loss": 0.6053,
      "num_input_tokens_seen": 191662236,
      "step": 3421
    },
    {
      "epoch": 7.619153674832962,
      "loss": 0.7925351858139038,
      "loss_ce": 0.00029885018011555076,
      "loss_iou": 0.341796875,
      "loss_num": 0.0218505859375,
      "loss_xval": 0.79296875,
      "num_input_tokens_seen": 191662236,
      "step": 3421
    },
    {
      "epoch": 7.621380846325167,
      "grad_norm": 24.45431900024414,
      "learning_rate": 1e-06,
      "loss": 0.6113,
      "num_input_tokens_seen": 191720748,
      "step": 3422
    },
    {
      "epoch": 7.621380846325167,
      "loss": 0.6171329617500305,
      "loss_ce": 0.00018957394058816135,
      "loss_iou": 0.267578125,
      "loss_num": 0.0162353515625,
      "loss_xval": 0.6171875,
      "num_input_tokens_seen": 191720748,
      "step": 3422
    },
    {
      "epoch": 7.6236080178173715,
      "grad_norm": 20.672542572021484,
      "learning_rate": 1e-06,
      "loss": 0.8138,
      "num_input_tokens_seen": 191774632,
      "step": 3423
    },
    {
      "epoch": 7.6236080178173715,
      "loss": 0.8238697052001953,
      "loss_ce": 0.00013917218893766403,
      "loss_iou": 0.361328125,
      "loss_num": 0.0205078125,
      "loss_xval": 0.82421875,
      "num_input_tokens_seen": 191774632,
      "step": 3423
    },
    {
      "epoch": 7.625835189309576,
      "grad_norm": 23.883867263793945,
      "learning_rate": 1e-06,
      "loss": 0.7723,
      "num_input_tokens_seen": 191831068,
      "step": 3424
    },
    {
      "epoch": 7.625835189309576,
      "loss": 0.6435463428497314,
      "loss_ce": 0.0002357645716983825,
      "loss_iou": 0.283203125,
      "loss_num": 0.01495361328125,
      "loss_xval": 0.64453125,
      "num_input_tokens_seen": 191831068,
      "step": 3424
    },
    {
      "epoch": 7.628062360801781,
      "grad_norm": 17.740406036376953,
      "learning_rate": 1e-06,
      "loss": 0.4917,
      "num_input_tokens_seen": 191888888,
      "step": 3425
    },
    {
      "epoch": 7.628062360801781,
      "loss": 0.5690467953681946,
      "loss_ce": 0.00019914188305847347,
      "loss_iou": 0.2412109375,
      "loss_num": 0.0174560546875,
      "loss_xval": 0.5703125,
      "num_input_tokens_seen": 191888888,
      "step": 3425
    },
    {
      "epoch": 7.630289532293987,
      "grad_norm": 34.307125091552734,
      "learning_rate": 1e-06,
      "loss": 0.518,
      "num_input_tokens_seen": 191944488,
      "step": 3426
    },
    {
      "epoch": 7.630289532293987,
      "loss": 0.5533403158187866,
      "loss_ce": 0.00036179396556690335,
      "loss_iou": 0.20703125,
      "loss_num": 0.0277099609375,
      "loss_xval": 0.5546875,
      "num_input_tokens_seen": 191944488,
      "step": 3426
    },
    {
      "epoch": 7.632516703786192,
      "grad_norm": 15.701261520385742,
      "learning_rate": 1e-06,
      "loss": 0.5397,
      "num_input_tokens_seen": 191999632,
      "step": 3427
    },
    {
      "epoch": 7.632516703786192,
      "loss": 0.6593526601791382,
      "loss_ce": 0.0004170782049186528,
      "loss_iou": 0.28125,
      "loss_num": 0.019287109375,
      "loss_xval": 0.66015625,
      "num_input_tokens_seen": 191999632,
      "step": 3427
    },
    {
      "epoch": 7.634743875278397,
      "grad_norm": 20.72135353088379,
      "learning_rate": 1e-06,
      "loss": 0.6254,
      "num_input_tokens_seen": 192055488,
      "step": 3428
    },
    {
      "epoch": 7.634743875278397,
      "loss": 0.6992897987365723,
      "loss_ce": 0.00019309617346152663,
      "loss_iou": 0.302734375,
      "loss_num": 0.0185546875,
      "loss_xval": 0.69921875,
      "num_input_tokens_seen": 192055488,
      "step": 3428
    },
    {
      "epoch": 7.636971046770602,
      "grad_norm": 30.77797508239746,
      "learning_rate": 1e-06,
      "loss": 0.5652,
      "num_input_tokens_seen": 192112044,
      "step": 3429
    },
    {
      "epoch": 7.636971046770602,
      "loss": 0.5464463233947754,
      "loss_ce": 0.00018164291395805776,
      "loss_iou": 0.2470703125,
      "loss_num": 0.0106201171875,
      "loss_xval": 0.546875,
      "num_input_tokens_seen": 192112044,
      "step": 3429
    },
    {
      "epoch": 7.639198218262806,
      "grad_norm": 17.127363204956055,
      "learning_rate": 1e-06,
      "loss": 0.942,
      "num_input_tokens_seen": 192165832,
      "step": 3430
    },
    {
      "epoch": 7.639198218262806,
      "loss": 0.7089158892631531,
      "loss_ce": 0.00017566155293025076,
      "loss_iou": 0.294921875,
      "loss_num": 0.0240478515625,
      "loss_xval": 0.70703125,
      "num_input_tokens_seen": 192165832,
      "step": 3430
    },
    {
      "epoch": 7.641425389755011,
      "grad_norm": 19.867774963378906,
      "learning_rate": 1e-06,
      "loss": 0.5599,
      "num_input_tokens_seen": 192219304,
      "step": 3431
    },
    {
      "epoch": 7.641425389755011,
      "loss": 0.4574645757675171,
      "loss_ce": 0.0001891769061330706,
      "loss_iou": 0.2119140625,
      "loss_num": 0.0067138671875,
      "loss_xval": 0.45703125,
      "num_input_tokens_seen": 192219304,
      "step": 3431
    },
    {
      "epoch": 7.643652561247216,
      "grad_norm": 13.734804153442383,
      "learning_rate": 1e-06,
      "loss": 0.6357,
      "num_input_tokens_seen": 192276704,
      "step": 3432
    },
    {
      "epoch": 7.643652561247216,
      "loss": 0.6115978956222534,
      "loss_ce": 0.0002697639574762434,
      "loss_iou": 0.2451171875,
      "loss_num": 0.0245361328125,
      "loss_xval": 0.609375,
      "num_input_tokens_seen": 192276704,
      "step": 3432
    },
    {
      "epoch": 7.645879732739421,
      "grad_norm": 18.749895095825195,
      "learning_rate": 1e-06,
      "loss": 0.7049,
      "num_input_tokens_seen": 192335292,
      "step": 3433
    },
    {
      "epoch": 7.645879732739421,
      "loss": 0.6146724820137024,
      "loss_ce": 0.00017052568728104234,
      "loss_iou": 0.26171875,
      "loss_num": 0.0185546875,
      "loss_xval": 0.61328125,
      "num_input_tokens_seen": 192335292,
      "step": 3433
    },
    {
      "epoch": 7.648106904231626,
      "grad_norm": 17.647464752197266,
      "learning_rate": 1e-06,
      "loss": 0.5601,
      "num_input_tokens_seen": 192391092,
      "step": 3434
    },
    {
      "epoch": 7.648106904231626,
      "loss": 0.519446611404419,
      "loss_ce": 0.00015951888053677976,
      "loss_iou": 0.228515625,
      "loss_num": 0.0123291015625,
      "loss_xval": 0.51953125,
      "num_input_tokens_seen": 192391092,
      "step": 3434
    },
    {
      "epoch": 7.650334075723831,
      "grad_norm": 17.00111961364746,
      "learning_rate": 1e-06,
      "loss": 0.5881,
      "num_input_tokens_seen": 192447236,
      "step": 3435
    },
    {
      "epoch": 7.650334075723831,
      "loss": 0.4986279606819153,
      "loss_ce": 0.00033695262391120195,
      "loss_iou": 0.1923828125,
      "loss_num": 0.0225830078125,
      "loss_xval": 0.498046875,
      "num_input_tokens_seen": 192447236,
      "step": 3435
    },
    {
      "epoch": 7.652561247216036,
      "grad_norm": 16.545228958129883,
      "learning_rate": 1e-06,
      "loss": 0.7619,
      "num_input_tokens_seen": 192502120,
      "step": 3436
    },
    {
      "epoch": 7.652561247216036,
      "loss": 0.8621032238006592,
      "loss_ce": 0.0002867898438125849,
      "loss_iou": 0.353515625,
      "loss_num": 0.0306396484375,
      "loss_xval": 0.86328125,
      "num_input_tokens_seen": 192502120,
      "step": 3436
    },
    {
      "epoch": 7.6547884187082404,
      "grad_norm": 26.11361312866211,
      "learning_rate": 1e-06,
      "loss": 0.7085,
      "num_input_tokens_seen": 192556024,
      "step": 3437
    },
    {
      "epoch": 7.6547884187082404,
      "loss": 0.9188729524612427,
      "loss_ce": 0.00017173260857816786,
      "loss_iou": 0.375,
      "loss_num": 0.0341796875,
      "loss_xval": 0.91796875,
      "num_input_tokens_seen": 192556024,
      "step": 3437
    },
    {
      "epoch": 7.657015590200445,
      "grad_norm": 14.627326011657715,
      "learning_rate": 1e-06,
      "loss": 0.4922,
      "num_input_tokens_seen": 192614920,
      "step": 3438
    },
    {
      "epoch": 7.657015590200445,
      "loss": 0.47406378388404846,
      "loss_ce": 0.0007971928571350873,
      "loss_iou": 0.212890625,
      "loss_num": 0.0093994140625,
      "loss_xval": 0.47265625,
      "num_input_tokens_seen": 192614920,
      "step": 3438
    },
    {
      "epoch": 7.65924276169265,
      "grad_norm": 22.492679595947266,
      "learning_rate": 1e-06,
      "loss": 0.502,
      "num_input_tokens_seen": 192670472,
      "step": 3439
    },
    {
      "epoch": 7.65924276169265,
      "loss": 0.49001675844192505,
      "loss_ce": 0.000270665914285928,
      "loss_iou": 0.220703125,
      "loss_num": 0.00982666015625,
      "loss_xval": 0.490234375,
      "num_input_tokens_seen": 192670472,
      "step": 3439
    },
    {
      "epoch": 7.661469933184855,
      "grad_norm": 25.153066635131836,
      "learning_rate": 1e-06,
      "loss": 0.6003,
      "num_input_tokens_seen": 192724972,
      "step": 3440
    },
    {
      "epoch": 7.661469933184855,
      "loss": 0.6613759994506836,
      "loss_ce": 0.0002431726170470938,
      "loss_iou": 0.2890625,
      "loss_num": 0.0169677734375,
      "loss_xval": 0.66015625,
      "num_input_tokens_seen": 192724972,
      "step": 3440
    },
    {
      "epoch": 7.66369710467706,
      "grad_norm": 23.217100143432617,
      "learning_rate": 1e-06,
      "loss": 0.4673,
      "num_input_tokens_seen": 192780272,
      "step": 3441
    },
    {
      "epoch": 7.66369710467706,
      "loss": 0.49449318647384644,
      "loss_ce": 0.00023047745344229043,
      "loss_iou": 0.21484375,
      "loss_num": 0.0128173828125,
      "loss_xval": 0.494140625,
      "num_input_tokens_seen": 192780272,
      "step": 3441
    },
    {
      "epoch": 7.665924276169265,
      "grad_norm": 19.738222122192383,
      "learning_rate": 1e-06,
      "loss": 0.5292,
      "num_input_tokens_seen": 192834056,
      "step": 3442
    },
    {
      "epoch": 7.665924276169265,
      "loss": 0.4528142213821411,
      "loss_ce": 0.0001774845877662301,
      "loss_iou": 0.193359375,
      "loss_num": 0.01318359375,
      "loss_xval": 0.453125,
      "num_input_tokens_seen": 192834056,
      "step": 3442
    },
    {
      "epoch": 7.66815144766147,
      "grad_norm": 17.538352966308594,
      "learning_rate": 1e-06,
      "loss": 0.5892,
      "num_input_tokens_seen": 192887184,
      "step": 3443
    },
    {
      "epoch": 7.66815144766147,
      "loss": 0.5604883432388306,
      "loss_ce": 0.00018561651813797653,
      "loss_iou": 0.2451171875,
      "loss_num": 0.01416015625,
      "loss_xval": 0.55859375,
      "num_input_tokens_seen": 192887184,
      "step": 3443
    },
    {
      "epoch": 7.6703786191536745,
      "grad_norm": 20.0687198638916,
      "learning_rate": 1e-06,
      "loss": 0.6592,
      "num_input_tokens_seen": 192944640,
      "step": 3444
    },
    {
      "epoch": 7.6703786191536745,
      "loss": 0.5089391469955444,
      "loss_ce": 0.00027216560556553304,
      "loss_iou": 0.228515625,
      "loss_num": 0.01043701171875,
      "loss_xval": 0.5078125,
      "num_input_tokens_seen": 192944640,
      "step": 3444
    },
    {
      "epoch": 7.67260579064588,
      "grad_norm": 15.960465431213379,
      "learning_rate": 1e-06,
      "loss": 0.5891,
      "num_input_tokens_seen": 193000164,
      "step": 3445
    },
    {
      "epoch": 7.67260579064588,
      "loss": 0.6576277017593384,
      "loss_ce": 0.00040116519085131586,
      "loss_iou": 0.267578125,
      "loss_num": 0.024169921875,
      "loss_xval": 0.65625,
      "num_input_tokens_seen": 193000164,
      "step": 3445
    },
    {
      "epoch": 7.674832962138085,
      "grad_norm": 18.63033103942871,
      "learning_rate": 1e-06,
      "loss": 0.623,
      "num_input_tokens_seen": 193053128,
      "step": 3446
    },
    {
      "epoch": 7.674832962138085,
      "loss": 0.5889174342155457,
      "loss_ce": 0.00017231784295290709,
      "loss_iou": 0.25,
      "loss_num": 0.017822265625,
      "loss_xval": 0.58984375,
      "num_input_tokens_seen": 193053128,
      "step": 3446
    },
    {
      "epoch": 7.67706013363029,
      "grad_norm": 14.499951362609863,
      "learning_rate": 1e-06,
      "loss": 0.5724,
      "num_input_tokens_seen": 193108676,
      "step": 3447
    },
    {
      "epoch": 7.67706013363029,
      "loss": 0.657649576663971,
      "loss_ce": 0.000178877409780398,
      "loss_iou": 0.3046875,
      "loss_num": 0.009765625,
      "loss_xval": 0.65625,
      "num_input_tokens_seen": 193108676,
      "step": 3447
    },
    {
      "epoch": 7.679287305122495,
      "grad_norm": 18.459136962890625,
      "learning_rate": 1e-06,
      "loss": 0.5688,
      "num_input_tokens_seen": 193166308,
      "step": 3448
    },
    {
      "epoch": 7.679287305122495,
      "loss": 0.7646270990371704,
      "loss_ce": 0.00022284439182840288,
      "loss_iou": 0.34375,
      "loss_num": 0.015625,
      "loss_xval": 0.765625,
      "num_input_tokens_seen": 193166308,
      "step": 3448
    },
    {
      "epoch": 7.6815144766147,
      "grad_norm": 35.68458938598633,
      "learning_rate": 1e-06,
      "loss": 0.5712,
      "num_input_tokens_seen": 193222788,
      "step": 3449
    },
    {
      "epoch": 7.6815144766147,
      "loss": 0.48069441318511963,
      "loss_ce": 0.0001646471064304933,
      "loss_iou": 0.212890625,
      "loss_num": 0.01104736328125,
      "loss_xval": 0.48046875,
      "num_input_tokens_seen": 193222788,
      "step": 3449
    },
    {
      "epoch": 7.6837416481069045,
      "grad_norm": 23.960975646972656,
      "learning_rate": 1e-06,
      "loss": 0.7771,
      "num_input_tokens_seen": 193273416,
      "step": 3450
    },
    {
      "epoch": 7.6837416481069045,
      "loss": 0.5310087203979492,
      "loss_ce": 0.0002469586324878037,
      "loss_iou": 0.2255859375,
      "loss_num": 0.015869140625,
      "loss_xval": 0.53125,
      "num_input_tokens_seen": 193273416,
      "step": 3450
    },
    {
      "epoch": 7.685968819599109,
      "grad_norm": 21.065771102905273,
      "learning_rate": 1e-06,
      "loss": 0.6134,
      "num_input_tokens_seen": 193327328,
      "step": 3451
    },
    {
      "epoch": 7.685968819599109,
      "loss": 0.8937476873397827,
      "loss_ce": 0.0001929743157234043,
      "loss_iou": 0.365234375,
      "loss_num": 0.032470703125,
      "loss_xval": 0.89453125,
      "num_input_tokens_seen": 193327328,
      "step": 3451
    },
    {
      "epoch": 7.688195991091314,
      "grad_norm": 16.559165954589844,
      "learning_rate": 1e-06,
      "loss": 0.5826,
      "num_input_tokens_seen": 193384244,
      "step": 3452
    },
    {
      "epoch": 7.688195991091314,
      "loss": 0.6634224653244019,
      "loss_ce": 0.0002144278259947896,
      "loss_iou": 0.302734375,
      "loss_num": 0.01177978515625,
      "loss_xval": 0.6640625,
      "num_input_tokens_seen": 193384244,
      "step": 3452
    },
    {
      "epoch": 7.690423162583519,
      "grad_norm": 16.13030433654785,
      "learning_rate": 1e-06,
      "loss": 0.6468,
      "num_input_tokens_seen": 193439784,
      "step": 3453
    },
    {
      "epoch": 7.690423162583519,
      "loss": 0.6554673910140991,
      "loss_ce": 0.0001939490030054003,
      "loss_iou": 0.283203125,
      "loss_num": 0.0179443359375,
      "loss_xval": 0.65625,
      "num_input_tokens_seen": 193439784,
      "step": 3453
    },
    {
      "epoch": 7.692650334075724,
      "grad_norm": 17.711097717285156,
      "learning_rate": 1e-06,
      "loss": 0.6227,
      "num_input_tokens_seen": 193493800,
      "step": 3454
    },
    {
      "epoch": 7.692650334075724,
      "loss": 0.6291646957397461,
      "loss_ce": 0.0002584658795967698,
      "loss_iou": 0.25390625,
      "loss_num": 0.0244140625,
      "loss_xval": 0.62890625,
      "num_input_tokens_seen": 193493800,
      "step": 3454
    },
    {
      "epoch": 7.694877505567929,
      "grad_norm": 16.89303207397461,
      "learning_rate": 1e-06,
      "loss": 0.5846,
      "num_input_tokens_seen": 193546256,
      "step": 3455
    },
    {
      "epoch": 7.694877505567929,
      "loss": 0.550576388835907,
      "loss_ce": 0.00016140022489707917,
      "loss_iou": 0.2060546875,
      "loss_num": 0.027587890625,
      "loss_xval": 0.55078125,
      "num_input_tokens_seen": 193546256,
      "step": 3455
    },
    {
      "epoch": 7.697104677060134,
      "grad_norm": 34.59210968017578,
      "learning_rate": 1e-06,
      "loss": 0.6684,
      "num_input_tokens_seen": 193602852,
      "step": 3456
    },
    {
      "epoch": 7.697104677060134,
      "loss": 0.5592398643493652,
      "loss_ce": 0.000157813512487337,
      "loss_iou": 0.263671875,
      "loss_num": 0.00653076171875,
      "loss_xval": 0.55859375,
      "num_input_tokens_seen": 193602852,
      "step": 3456
    },
    {
      "epoch": 7.6993318485523385,
      "grad_norm": 20.46678352355957,
      "learning_rate": 1e-06,
      "loss": 0.5734,
      "num_input_tokens_seen": 193661456,
      "step": 3457
    },
    {
      "epoch": 7.6993318485523385,
      "loss": 0.5596364140510559,
      "loss_ce": 0.00018814706709235907,
      "loss_iou": 0.2421875,
      "loss_num": 0.01513671875,
      "loss_xval": 0.55859375,
      "num_input_tokens_seen": 193661456,
      "step": 3457
    },
    {
      "epoch": 7.701559020044543,
      "grad_norm": 14.73276424407959,
      "learning_rate": 1e-06,
      "loss": 0.6287,
      "num_input_tokens_seen": 193720104,
      "step": 3458
    },
    {
      "epoch": 7.701559020044543,
      "loss": 0.5451168417930603,
      "loss_ce": 0.00019496420281939209,
      "loss_iou": 0.25,
      "loss_num": 0.00860595703125,
      "loss_xval": 0.546875,
      "num_input_tokens_seen": 193720104,
      "step": 3458
    },
    {
      "epoch": 7.703786191536748,
      "grad_norm": 14.44981575012207,
      "learning_rate": 1e-06,
      "loss": 0.3805,
      "num_input_tokens_seen": 193776696,
      "step": 3459
    },
    {
      "epoch": 7.703786191536748,
      "loss": 0.3983895778656006,
      "loss_ce": 0.00019621921819634736,
      "loss_iou": 0.177734375,
      "loss_num": 0.00830078125,
      "loss_xval": 0.3984375,
      "num_input_tokens_seen": 193776696,
      "step": 3459
    },
    {
      "epoch": 7.706013363028953,
      "grad_norm": 19.433345794677734,
      "learning_rate": 1e-06,
      "loss": 0.4632,
      "num_input_tokens_seen": 193834220,
      "step": 3460
    },
    {
      "epoch": 7.706013363028953,
      "loss": 0.4960872530937195,
      "loss_ce": 0.00023767323000356555,
      "loss_iou": 0.185546875,
      "loss_num": 0.0247802734375,
      "loss_xval": 0.49609375,
      "num_input_tokens_seen": 193834220,
      "step": 3460
    },
    {
      "epoch": 7.708240534521158,
      "grad_norm": 55.87489318847656,
      "learning_rate": 1e-06,
      "loss": 0.5689,
      "num_input_tokens_seen": 193888720,
      "step": 3461
    },
    {
      "epoch": 7.708240534521158,
      "loss": 0.6605603098869324,
      "loss_ce": 0.0004040383209940046,
      "loss_iou": 0.2890625,
      "loss_num": 0.0162353515625,
      "loss_xval": 0.66015625,
      "num_input_tokens_seen": 193888720,
      "step": 3461
    },
    {
      "epoch": 7.710467706013363,
      "grad_norm": 28.30753517150879,
      "learning_rate": 1e-06,
      "loss": 0.7928,
      "num_input_tokens_seen": 193943548,
      "step": 3462
    },
    {
      "epoch": 7.710467706013363,
      "loss": 0.6849916577339172,
      "loss_ce": 0.00017721363110467792,
      "loss_iou": 0.287109375,
      "loss_num": 0.02197265625,
      "loss_xval": 0.68359375,
      "num_input_tokens_seen": 193943548,
      "step": 3462
    },
    {
      "epoch": 7.712694877505568,
      "grad_norm": 23.885019302368164,
      "learning_rate": 1e-06,
      "loss": 0.7603,
      "num_input_tokens_seen": 194000004,
      "step": 3463
    },
    {
      "epoch": 7.712694877505568,
      "loss": 0.6882323026657104,
      "loss_ce": 0.00024399746325798333,
      "loss_iou": 0.27734375,
      "loss_num": 0.0262451171875,
      "loss_xval": 0.6875,
      "num_input_tokens_seen": 194000004,
      "step": 3463
    },
    {
      "epoch": 7.714922048997773,
      "grad_norm": 14.701726913452148,
      "learning_rate": 1e-06,
      "loss": 0.5217,
      "num_input_tokens_seen": 194056588,
      "step": 3464
    },
    {
      "epoch": 7.714922048997773,
      "loss": 0.5858080387115479,
      "loss_ce": 0.00020622329611796886,
      "loss_iou": 0.240234375,
      "loss_num": 0.0208740234375,
      "loss_xval": 0.5859375,
      "num_input_tokens_seen": 194056588,
      "step": 3464
    },
    {
      "epoch": 7.717149220489977,
      "grad_norm": 16.439701080322266,
      "learning_rate": 1e-06,
      "loss": 0.5724,
      "num_input_tokens_seen": 194111836,
      "step": 3465
    },
    {
      "epoch": 7.717149220489977,
      "loss": 0.5037325620651245,
      "loss_ce": 0.0001925197138916701,
      "loss_iou": 0.21484375,
      "loss_num": 0.0146484375,
      "loss_xval": 0.50390625,
      "num_input_tokens_seen": 194111836,
      "step": 3465
    },
    {
      "epoch": 7.719376391982182,
      "grad_norm": 21.691720962524414,
      "learning_rate": 1e-06,
      "loss": 0.628,
      "num_input_tokens_seen": 194168228,
      "step": 3466
    },
    {
      "epoch": 7.719376391982182,
      "loss": 0.7247686982154846,
      "loss_ce": 0.00015930971130728722,
      "loss_iou": 0.306640625,
      "loss_num": 0.0218505859375,
      "loss_xval": 0.7265625,
      "num_input_tokens_seen": 194168228,
      "step": 3466
    },
    {
      "epoch": 7.721603563474387,
      "grad_norm": 13.828973770141602,
      "learning_rate": 1e-06,
      "loss": 0.5437,
      "num_input_tokens_seen": 194223784,
      "step": 3467
    },
    {
      "epoch": 7.721603563474387,
      "loss": 0.5051822662353516,
      "loss_ce": 0.00017740536713972688,
      "loss_iou": 0.2333984375,
      "loss_num": 0.007659912109375,
      "loss_xval": 0.50390625,
      "num_input_tokens_seen": 194223784,
      "step": 3467
    },
    {
      "epoch": 7.723830734966592,
      "grad_norm": 12.506599426269531,
      "learning_rate": 1e-06,
      "loss": 0.5575,
      "num_input_tokens_seen": 194280580,
      "step": 3468
    },
    {
      "epoch": 7.723830734966592,
      "loss": 0.5337764620780945,
      "loss_ce": 0.00032918865326792,
      "loss_iou": 0.2119140625,
      "loss_num": 0.0218505859375,
      "loss_xval": 0.53515625,
      "num_input_tokens_seen": 194280580,
      "step": 3468
    },
    {
      "epoch": 7.726057906458797,
      "grad_norm": 52.32142639160156,
      "learning_rate": 1e-06,
      "loss": 0.6348,
      "num_input_tokens_seen": 194336084,
      "step": 3469
    },
    {
      "epoch": 7.726057906458797,
      "loss": 0.6026872992515564,
      "loss_ce": 0.0001482292718719691,
      "loss_iou": 0.27734375,
      "loss_num": 0.0098876953125,
      "loss_xval": 0.6015625,
      "num_input_tokens_seen": 194336084,
      "step": 3469
    },
    {
      "epoch": 7.728285077951003,
      "grad_norm": 26.636455535888672,
      "learning_rate": 1e-06,
      "loss": 0.5941,
      "num_input_tokens_seen": 194389144,
      "step": 3470
    },
    {
      "epoch": 7.728285077951003,
      "loss": 0.5668909549713135,
      "loss_ce": 0.00024056396796368062,
      "loss_iou": 0.234375,
      "loss_num": 0.0196533203125,
      "loss_xval": 0.56640625,
      "num_input_tokens_seen": 194389144,
      "step": 3470
    },
    {
      "epoch": 7.7305122494432075,
      "grad_norm": 25.504764556884766,
      "learning_rate": 1e-06,
      "loss": 0.6929,
      "num_input_tokens_seen": 194446564,
      "step": 3471
    },
    {
      "epoch": 7.7305122494432075,
      "loss": 0.7231018543243408,
      "loss_ce": 0.00020141596905887127,
      "loss_iou": 0.31640625,
      "loss_num": 0.0181884765625,
      "loss_xval": 0.72265625,
      "num_input_tokens_seen": 194446564,
      "step": 3471
    },
    {
      "epoch": 7.732739420935412,
      "grad_norm": 18.57636070251465,
      "learning_rate": 1e-06,
      "loss": 0.7205,
      "num_input_tokens_seen": 194501216,
      "step": 3472
    },
    {
      "epoch": 7.732739420935412,
      "loss": 0.757434606552124,
      "loss_ce": 0.00017141405260190368,
      "loss_iou": 0.326171875,
      "loss_num": 0.021240234375,
      "loss_xval": 0.7578125,
      "num_input_tokens_seen": 194501216,
      "step": 3472
    },
    {
      "epoch": 7.734966592427617,
      "grad_norm": 23.02729034423828,
      "learning_rate": 1e-06,
      "loss": 0.6108,
      "num_input_tokens_seen": 194558088,
      "step": 3473
    },
    {
      "epoch": 7.734966592427617,
      "loss": 0.591259241104126,
      "loss_ce": 0.0001947856944752857,
      "loss_iou": 0.248046875,
      "loss_num": 0.0189208984375,
      "loss_xval": 0.58984375,
      "num_input_tokens_seen": 194558088,
      "step": 3473
    },
    {
      "epoch": 7.737193763919822,
      "grad_norm": 16.5152587890625,
      "learning_rate": 1e-06,
      "loss": 0.591,
      "num_input_tokens_seen": 194616640,
      "step": 3474
    },
    {
      "epoch": 7.737193763919822,
      "loss": 0.42105597257614136,
      "loss_ce": 0.00015753699699416757,
      "loss_iou": 0.189453125,
      "loss_num": 0.00830078125,
      "loss_xval": 0.421875,
      "num_input_tokens_seen": 194616640,
      "step": 3474
    },
    {
      "epoch": 7.739420935412027,
      "grad_norm": 20.325769424438477,
      "learning_rate": 1e-06,
      "loss": 0.8947,
      "num_input_tokens_seen": 194674240,
      "step": 3475
    },
    {
      "epoch": 7.739420935412027,
      "loss": 1.23799729347229,
      "loss_ce": 0.0003264172119088471,
      "loss_iou": 0.478515625,
      "loss_num": 0.055908203125,
      "loss_xval": 1.234375,
      "num_input_tokens_seen": 194674240,
      "step": 3475
    },
    {
      "epoch": 7.741648106904232,
      "grad_norm": 21.625873565673828,
      "learning_rate": 1e-06,
      "loss": 0.5786,
      "num_input_tokens_seen": 194729012,
      "step": 3476
    },
    {
      "epoch": 7.741648106904232,
      "loss": 0.391615629196167,
      "loss_ce": 0.000136150760226883,
      "loss_iou": 0.16796875,
      "loss_num": 0.01104736328125,
      "loss_xval": 0.390625,
      "num_input_tokens_seen": 194729012,
      "step": 3476
    },
    {
      "epoch": 7.743875278396437,
      "grad_norm": 14.393267631530762,
      "learning_rate": 1e-06,
      "loss": 0.6558,
      "num_input_tokens_seen": 194786480,
      "step": 3477
    },
    {
      "epoch": 7.743875278396437,
      "loss": 0.5332167148590088,
      "loss_ce": 0.0001356468565063551,
      "loss_iou": 0.220703125,
      "loss_num": 0.0181884765625,
      "loss_xval": 0.53125,
      "num_input_tokens_seen": 194786480,
      "step": 3477
    },
    {
      "epoch": 7.7461024498886415,
      "grad_norm": 33.26962661743164,
      "learning_rate": 1e-06,
      "loss": 0.9039,
      "num_input_tokens_seen": 194842640,
      "step": 3478
    },
    {
      "epoch": 7.7461024498886415,
      "loss": 0.8585567474365234,
      "loss_ce": 0.00015828575124032795,
      "loss_iou": 0.37109375,
      "loss_num": 0.0234375,
      "loss_xval": 0.859375,
      "num_input_tokens_seen": 194842640,
      "step": 3478
    },
    {
      "epoch": 7.748329621380846,
      "grad_norm": 20.943241119384766,
      "learning_rate": 1e-06,
      "loss": 0.5986,
      "num_input_tokens_seen": 194900196,
      "step": 3479
    },
    {
      "epoch": 7.748329621380846,
      "loss": 0.5145847797393799,
      "loss_ce": 0.00018048740457743406,
      "loss_iou": 0.232421875,
      "loss_num": 0.01007080078125,
      "loss_xval": 0.515625,
      "num_input_tokens_seen": 194900196,
      "step": 3479
    },
    {
      "epoch": 7.750556792873051,
      "grad_norm": 27.85867691040039,
      "learning_rate": 1e-06,
      "loss": 0.4849,
      "num_input_tokens_seen": 194956328,
      "step": 3480
    },
    {
      "epoch": 7.750556792873051,
      "loss": 0.5256341695785522,
      "loss_ce": 0.0007318383431993425,
      "loss_iou": 0.2421875,
      "loss_num": 0.00799560546875,
      "loss_xval": 0.5234375,
      "num_input_tokens_seen": 194956328,
      "step": 3480
    },
    {
      "epoch": 7.752783964365256,
      "grad_norm": 23.426788330078125,
      "learning_rate": 1e-06,
      "loss": 0.8095,
      "num_input_tokens_seen": 195011508,
      "step": 3481
    },
    {
      "epoch": 7.752783964365256,
      "loss": 0.9079303741455078,
      "loss_ce": 0.00021555817511398345,
      "loss_iou": 0.3828125,
      "loss_num": 0.02880859375,
      "loss_xval": 0.90625,
      "num_input_tokens_seen": 195011508,
      "step": 3481
    },
    {
      "epoch": 7.755011135857461,
      "grad_norm": 26.646705627441406,
      "learning_rate": 1e-06,
      "loss": 0.6848,
      "num_input_tokens_seen": 195066276,
      "step": 3482
    },
    {
      "epoch": 7.755011135857461,
      "loss": 0.5551425218582153,
      "loss_ce": 0.00021085041225887835,
      "loss_iou": 0.2314453125,
      "loss_num": 0.0185546875,
      "loss_xval": 0.5546875,
      "num_input_tokens_seen": 195066276,
      "step": 3482
    },
    {
      "epoch": 7.757238307349666,
      "grad_norm": 21.877819061279297,
      "learning_rate": 1e-06,
      "loss": 0.75,
      "num_input_tokens_seen": 195121820,
      "step": 3483
    },
    {
      "epoch": 7.757238307349666,
      "loss": 0.87198805809021,
      "loss_ce": 0.00016191550821531564,
      "loss_iou": 0.37890625,
      "loss_num": 0.0224609375,
      "loss_xval": 0.87109375,
      "num_input_tokens_seen": 195121820,
      "step": 3483
    },
    {
      "epoch": 7.759465478841871,
      "grad_norm": 17.727195739746094,
      "learning_rate": 1e-06,
      "loss": 0.7847,
      "num_input_tokens_seen": 195179468,
      "step": 3484
    },
    {
      "epoch": 7.759465478841871,
      "loss": 0.8190126419067383,
      "loss_ce": 0.00016502838116139174,
      "loss_iou": 0.353515625,
      "loss_num": 0.0224609375,
      "loss_xval": 0.8203125,
      "num_input_tokens_seen": 195179468,
      "step": 3484
    },
    {
      "epoch": 7.7616926503340755,
      "grad_norm": 18.596248626708984,
      "learning_rate": 1e-06,
      "loss": 0.4576,
      "num_input_tokens_seen": 195233616,
      "step": 3485
    },
    {
      "epoch": 7.7616926503340755,
      "loss": 0.4784582555294037,
      "loss_ce": 0.0001867622195277363,
      "loss_iou": 0.203125,
      "loss_num": 0.01446533203125,
      "loss_xval": 0.478515625,
      "num_input_tokens_seen": 195233616,
      "step": 3485
    },
    {
      "epoch": 7.76391982182628,
      "grad_norm": 21.53468894958496,
      "learning_rate": 1e-06,
      "loss": 0.5638,
      "num_input_tokens_seen": 195289464,
      "step": 3486
    },
    {
      "epoch": 7.76391982182628,
      "loss": 0.6328760981559753,
      "loss_ce": 0.00018564131460152566,
      "loss_iou": 0.265625,
      "loss_num": 0.0198974609375,
      "loss_xval": 0.6328125,
      "num_input_tokens_seen": 195289464,
      "step": 3486
    },
    {
      "epoch": 7.766146993318485,
      "grad_norm": 17.920970916748047,
      "learning_rate": 1e-06,
      "loss": 0.8169,
      "num_input_tokens_seen": 195344536,
      "step": 3487
    },
    {
      "epoch": 7.766146993318485,
      "loss": 0.45805519819259644,
      "loss_ce": 0.00016945481183938682,
      "loss_iou": 0.2080078125,
      "loss_num": 0.0084228515625,
      "loss_xval": 0.45703125,
      "num_input_tokens_seen": 195344536,
      "step": 3487
    },
    {
      "epoch": 7.76837416481069,
      "grad_norm": 17.782272338867188,
      "learning_rate": 1e-06,
      "loss": 0.5641,
      "num_input_tokens_seen": 195399244,
      "step": 3488
    },
    {
      "epoch": 7.76837416481069,
      "loss": 0.5128250122070312,
      "loss_ce": 0.00025179749354720116,
      "loss_iou": 0.2255859375,
      "loss_num": 0.01214599609375,
      "loss_xval": 0.51171875,
      "num_input_tokens_seen": 195399244,
      "step": 3488
    },
    {
      "epoch": 7.770601336302895,
      "grad_norm": 28.824478149414062,
      "learning_rate": 1e-06,
      "loss": 0.4858,
      "num_input_tokens_seen": 195455096,
      "step": 3489
    },
    {
      "epoch": 7.770601336302895,
      "loss": 0.5841382741928101,
      "loss_ce": 0.0001538892393000424,
      "loss_iou": 0.259765625,
      "loss_num": 0.0125732421875,
      "loss_xval": 0.5859375,
      "num_input_tokens_seen": 195455096,
      "step": 3489
    },
    {
      "epoch": 7.772828507795101,
      "grad_norm": 16.21584129333496,
      "learning_rate": 1e-06,
      "loss": 0.6751,
      "num_input_tokens_seen": 195507752,
      "step": 3490
    },
    {
      "epoch": 7.772828507795101,
      "loss": 0.7475310564041138,
      "loss_ce": 0.0002166137855965644,
      "loss_iou": 0.322265625,
      "loss_num": 0.0206298828125,
      "loss_xval": 0.74609375,
      "num_input_tokens_seen": 195507752,
      "step": 3490
    },
    {
      "epoch": 7.775055679287306,
      "grad_norm": 16.735774993896484,
      "learning_rate": 1e-06,
      "loss": 0.5796,
      "num_input_tokens_seen": 195565384,
      "step": 3491
    },
    {
      "epoch": 7.775055679287306,
      "loss": 0.6273143291473389,
      "loss_ce": 0.00023911299649626017,
      "loss_iou": 0.25390625,
      "loss_num": 0.02392578125,
      "loss_xval": 0.62890625,
      "num_input_tokens_seen": 195565384,
      "step": 3491
    },
    {
      "epoch": 7.77728285077951,
      "grad_norm": 17.522539138793945,
      "learning_rate": 1e-06,
      "loss": 0.7027,
      "num_input_tokens_seen": 195619800,
      "step": 3492
    },
    {
      "epoch": 7.77728285077951,
      "loss": 0.7621063590049744,
      "loss_ce": 0.00014349556295201182,
      "loss_iou": 0.322265625,
      "loss_num": 0.0235595703125,
      "loss_xval": 0.76171875,
      "num_input_tokens_seen": 195619800,
      "step": 3492
    },
    {
      "epoch": 7.779510022271715,
      "grad_norm": 26.973304748535156,
      "learning_rate": 1e-06,
      "loss": 0.9673,
      "num_input_tokens_seen": 195673328,
      "step": 3493
    },
    {
      "epoch": 7.779510022271715,
      "loss": 1.0272767543792725,
      "loss_ce": 0.0006654445314779878,
      "loss_iou": 0.435546875,
      "loss_num": 0.03125,
      "loss_xval": 1.0234375,
      "num_input_tokens_seen": 195673328,
      "step": 3493
    },
    {
      "epoch": 7.78173719376392,
      "grad_norm": 19.321470260620117,
      "learning_rate": 1e-06,
      "loss": 0.6188,
      "num_input_tokens_seen": 195728884,
      "step": 3494
    },
    {
      "epoch": 7.78173719376392,
      "loss": 0.5602940320968628,
      "loss_ce": 0.0002354723692405969,
      "loss_iou": 0.2265625,
      "loss_num": 0.0213623046875,
      "loss_xval": 0.55859375,
      "num_input_tokens_seen": 195728884,
      "step": 3494
    },
    {
      "epoch": 7.783964365256125,
      "grad_norm": 19.93840789794922,
      "learning_rate": 1e-06,
      "loss": 0.5698,
      "num_input_tokens_seen": 195783412,
      "step": 3495
    },
    {
      "epoch": 7.783964365256125,
      "loss": 0.5731680393218994,
      "loss_ce": 0.00016997376224026084,
      "loss_iou": 0.224609375,
      "loss_num": 0.02490234375,
      "loss_xval": 0.57421875,
      "num_input_tokens_seen": 195783412,
      "step": 3495
    },
    {
      "epoch": 7.78619153674833,
      "grad_norm": 23.796375274658203,
      "learning_rate": 1e-06,
      "loss": 0.6692,
      "num_input_tokens_seen": 195838768,
      "step": 3496
    },
    {
      "epoch": 7.78619153674833,
      "loss": 0.717685341835022,
      "loss_ce": 0.00015606911620125175,
      "loss_iou": 0.3125,
      "loss_num": 0.01806640625,
      "loss_xval": 0.71875,
      "num_input_tokens_seen": 195838768,
      "step": 3496
    },
    {
      "epoch": 7.788418708240535,
      "grad_norm": 15.419715881347656,
      "learning_rate": 1e-06,
      "loss": 0.6021,
      "num_input_tokens_seen": 195894208,
      "step": 3497
    },
    {
      "epoch": 7.788418708240535,
      "loss": 0.7013325691223145,
      "loss_ce": 0.00016067746037151664,
      "loss_iou": 0.310546875,
      "loss_num": 0.0159912109375,
      "loss_xval": 0.703125,
      "num_input_tokens_seen": 195894208,
      "step": 3497
    },
    {
      "epoch": 7.79064587973274,
      "grad_norm": 21.454118728637695,
      "learning_rate": 1e-06,
      "loss": 0.6689,
      "num_input_tokens_seen": 195949980,
      "step": 3498
    },
    {
      "epoch": 7.79064587973274,
      "loss": 0.6081476211547852,
      "loss_ce": 0.00023741269251331687,
      "loss_iou": 0.2470703125,
      "loss_num": 0.02294921875,
      "loss_xval": 0.609375,
      "num_input_tokens_seen": 195949980,
      "step": 3498
    },
    {
      "epoch": 7.7928730512249444,
      "grad_norm": 21.61153793334961,
      "learning_rate": 1e-06,
      "loss": 0.7103,
      "num_input_tokens_seen": 196005100,
      "step": 3499
    },
    {
      "epoch": 7.7928730512249444,
      "loss": 0.5294746160507202,
      "loss_ce": 0.0012153343996033072,
      "loss_iou": 0.2353515625,
      "loss_num": 0.0113525390625,
      "loss_xval": 0.52734375,
      "num_input_tokens_seen": 196005100,
      "step": 3499
    },
    {
      "epoch": 7.795100222717149,
      "grad_norm": 15.124077796936035,
      "learning_rate": 1e-06,
      "loss": 0.7598,
      "num_input_tokens_seen": 196060760,
      "step": 3500
    },
    {
      "epoch": 7.795100222717149,
      "eval_seeclick_web_CIoU": 0.5788059234619141,
      "eval_seeclick_web_GIoU": 0.5771978497505188,
      "eval_seeclick_web_IoU": 0.5963517725467682,
      "eval_seeclick_web_MAE_all": 0.01642331574112177,
      "eval_seeclick_web_MAE_h": 0.008296339772641659,
      "eval_seeclick_web_MAE_w": 0.01685692649334669,
      "eval_seeclick_web_MAE_x_boxes": 0.009794581681489944,
      "eval_seeclick_web_MAE_y_boxes": 0.02204720745794475,
      "eval_seeclick_web_inside_bbox": 0.9166666567325592,
      "eval_seeclick_web_loss": 0.9286947846412659,
      "eval_seeclick_web_loss_ce": 0.0002503915602574125,
      "eval_seeclick_web_loss_iou": 0.4237060546875,
      "eval_seeclick_web_loss_num": 0.013225555419921875,
      "eval_seeclick_web_loss_xval": 0.913330078125,
      "eval_seeclick_web_runtime": 20.0349,
      "eval_seeclick_web_samples_per_second": 2.496,
      "eval_seeclick_web_steps_per_second": 0.1,
      "num_input_tokens_seen": 196060760,
      "step": 3500
    },
    {
      "epoch": 7.795100222717149,
      "eval_icons_CIoU": 0.3098849505186081,
      "eval_icons_GIoU": 0.3192085027694702,
      "eval_icons_IoU": 0.3817008137702942,
      "eval_icons_MAE_all": 0.0535985603928566,
      "eval_icons_MAE_h": 0.038005659356713295,
      "eval_icons_MAE_w": 0.04143812507390976,
      "eval_icons_MAE_x_boxes": 0.05823635682463646,
      "eval_icons_MAE_y_boxes": 0.037908594124019146,
      "eval_icons_inside_bbox": 0.6493055522441864,
      "eval_icons_loss": 1.6437243223190308,
      "eval_icons_loss_ce": 0.00029484537662938237,
      "eval_icons_loss_iou": 0.649658203125,
      "eval_icons_loss_num": 0.04685401916503906,
      "eval_icons_loss_xval": 1.533203125,
      "eval_icons_runtime": 21.1399,
      "eval_icons_samples_per_second": 2.365,
      "eval_icons_steps_per_second": 0.095,
      "num_input_tokens_seen": 196060760,
      "step": 3500
    },
    {
      "epoch": 7.795100222717149,
      "eval_screenspot_CIoU": 0.34179479877154034,
      "eval_screenspot_GIoU": 0.35942623019218445,
      "eval_screenspot_IoU": 0.4252944588661194,
      "eval_screenspot_MAE_all": 0.06343474984169006,
      "eval_screenspot_MAE_h": 0.037477921694517136,
      "eval_screenspot_MAE_w": 0.07201713944474857,
      "eval_screenspot_MAE_x_boxes": 0.07516818679869175,
      "eval_screenspot_MAE_y_boxes": 0.046938162917892136,
      "eval_screenspot_inside_bbox": 0.6966666579246521,
      "eval_screenspot_loss": 1.6630264520645142,
      "eval_screenspot_loss_ce": 0.0002933433085369567,
      "eval_screenspot_loss_iou": 0.6790364583333334,
      "eval_screenspot_loss_num": 0.07456461588541667,
      "eval_screenspot_loss_xval": 1.7312825520833333,
      "eval_screenspot_runtime": 33.7448,
      "eval_screenspot_samples_per_second": 2.637,
      "eval_screenspot_steps_per_second": 0.089,
      "num_input_tokens_seen": 196060760,
      "step": 3500
    },
    {
      "epoch": 7.795100222717149,
      "eval_compot_CIoU": 0.3524170517921448,
      "eval_compot_GIoU": 0.3639347553253174,
      "eval_compot_IoU": 0.40965285897254944,
      "eval_compot_MAE_all": 0.017891014460474253,
      "eval_compot_MAE_h": 0.00863239774480462,
      "eval_compot_MAE_w": 0.022040129639208317,
      "eval_compot_MAE_x_boxes": 0.029601704329252243,
      "eval_compot_MAE_y_boxes": 0.00674438988789916,
      "eval_compot_inside_bbox": 0.6302083432674408,
      "eval_compot_loss": 1.3697645664215088,
      "eval_compot_loss_ce": 0.00023254087136592716,
      "eval_compot_loss_iou": 0.62646484375,
      "eval_compot_loss_num": 0.016546249389648438,
      "eval_compot_loss_xval": 1.335205078125,
      "eval_compot_runtime": 20.9161,
      "eval_compot_samples_per_second": 2.391,
      "eval_compot_steps_per_second": 0.096,
      "num_input_tokens_seen": 196060760,
      "step": 3500
    },
    {
      "epoch": 7.795100222717149,
      "eval_custom_ui_val_CIoU": 0.4692361321714189,
      "eval_custom_ui_val_GIoU": 0.48255549867947894,
      "eval_custom_ui_val_IoU": 0.5274171696768867,
      "eval_custom_ui_val_MAE_all": 0.029507537372410297,
      "eval_custom_ui_val_MAE_h": 0.016804457098866504,
      "eval_custom_ui_val_MAE_w": 0.03681204499055942,
      "eval_custom_ui_val_MAE_x_boxes": 0.03406558599736956,
      "eval_custom_ui_val_MAE_y_boxes": 0.015829176952441532,
      "eval_custom_ui_val_inside_bbox": 0.7457561757829454,
      "eval_custom_ui_val_loss": 1.1831005811691284,
      "eval_custom_ui_val_loss_ce": 0.0002701325170669912,
      "eval_custom_ui_val_loss_iou": 0.5026584201388888,
      "eval_custom_ui_val_loss_num": 0.027642779880099826,
      "eval_custom_ui_val_loss_xval": 1.1439887152777777,
      "eval_custom_ui_val_runtime": 59.4212,
      "eval_custom_ui_val_samples_per_second": 4.46,
      "eval_custom_ui_val_steps_per_second": 0.151,
      "num_input_tokens_seen": 196060760,
      "step": 3500
    },
    {
      "epoch": 7.795100222717149,
      "loss": 0.8771896958351135,
      "loss_ce": 0.0002365570981055498,
      "loss_iou": 0.3828125,
      "loss_num": 0.0220947265625,
      "loss_xval": 0.875,
      "num_input_tokens_seen": 196060760,
      "step": 3500
    },
    {
      "epoch": 7.797327394209354,
      "grad_norm": 20.44601058959961,
      "learning_rate": 1e-06,
      "loss": 0.7573,
      "num_input_tokens_seen": 196116308,
      "step": 3501
    },
    {
      "epoch": 7.797327394209354,
      "loss": 0.790895402431488,
      "loss_ce": 0.0003680819063447416,
      "loss_iou": 0.306640625,
      "loss_num": 0.035400390625,
      "loss_xval": 0.7890625,
      "num_input_tokens_seen": 196116308,
      "step": 3501
    },
    {
      "epoch": 7.799554565701559,
      "grad_norm": 19.4777774810791,
      "learning_rate": 1e-06,
      "loss": 0.6974,
      "num_input_tokens_seen": 196172864,
      "step": 3502
    },
    {
      "epoch": 7.799554565701559,
      "loss": 0.7237052917480469,
      "loss_ce": 0.00019457802409306169,
      "loss_iou": 0.3125,
      "loss_num": 0.0196533203125,
      "loss_xval": 0.72265625,
      "num_input_tokens_seen": 196172864,
      "step": 3502
    },
    {
      "epoch": 7.801781737193764,
      "grad_norm": 18.440637588500977,
      "learning_rate": 1e-06,
      "loss": 0.5529,
      "num_input_tokens_seen": 196229132,
      "step": 3503
    },
    {
      "epoch": 7.801781737193764,
      "loss": 0.5999140739440918,
      "loss_ce": 0.0005488179158419371,
      "loss_iou": 0.2734375,
      "loss_num": 0.0107421875,
      "loss_xval": 0.59765625,
      "num_input_tokens_seen": 196229132,
      "step": 3503
    },
    {
      "epoch": 7.804008908685969,
      "grad_norm": 24.30689811706543,
      "learning_rate": 1e-06,
      "loss": 0.4758,
      "num_input_tokens_seen": 196287452,
      "step": 3504
    },
    {
      "epoch": 7.804008908685969,
      "loss": 0.4462117552757263,
      "loss_ce": 0.00016680760018061846,
      "loss_iou": 0.1962890625,
      "loss_num": 0.0106201171875,
      "loss_xval": 0.4453125,
      "num_input_tokens_seen": 196287452,
      "step": 3504
    },
    {
      "epoch": 7.806236080178174,
      "grad_norm": 21.83576011657715,
      "learning_rate": 1e-06,
      "loss": 0.7566,
      "num_input_tokens_seen": 196340952,
      "step": 3505
    },
    {
      "epoch": 7.806236080178174,
      "loss": 0.8626251220703125,
      "loss_ce": 0.0003204582317266613,
      "loss_iou": 0.38671875,
      "loss_num": 0.018310546875,
      "loss_xval": 0.86328125,
      "num_input_tokens_seen": 196340952,
      "step": 3505
    },
    {
      "epoch": 7.8084632516703785,
      "grad_norm": 41.28773880004883,
      "learning_rate": 1e-06,
      "loss": 0.5113,
      "num_input_tokens_seen": 196397792,
      "step": 3506
    },
    {
      "epoch": 7.8084632516703785,
      "loss": 0.5087121725082397,
      "loss_ce": 0.0001672878279350698,
      "loss_iou": 0.216796875,
      "loss_num": 0.01519775390625,
      "loss_xval": 0.5078125,
      "num_input_tokens_seen": 196397792,
      "step": 3506
    },
    {
      "epoch": 7.810690423162583,
      "grad_norm": 14.272880554199219,
      "learning_rate": 1e-06,
      "loss": 0.505,
      "num_input_tokens_seen": 196455836,
      "step": 3507
    },
    {
      "epoch": 7.810690423162583,
      "loss": 0.4430527091026306,
      "loss_ce": 0.0001816436561057344,
      "loss_iou": 0.19140625,
      "loss_num": 0.0120849609375,
      "loss_xval": 0.443359375,
      "num_input_tokens_seen": 196455836,
      "step": 3507
    },
    {
      "epoch": 7.812917594654788,
      "grad_norm": 16.481369018554688,
      "learning_rate": 1e-06,
      "loss": 0.6081,
      "num_input_tokens_seen": 196513684,
      "step": 3508
    },
    {
      "epoch": 7.812917594654788,
      "loss": 0.6896613836288452,
      "loss_ce": 0.00020824806415475905,
      "loss_iou": 0.3046875,
      "loss_num": 0.01611328125,
      "loss_xval": 0.6875,
      "num_input_tokens_seen": 196513684,
      "step": 3508
    },
    {
      "epoch": 7.815144766146993,
      "grad_norm": 16.66682243347168,
      "learning_rate": 1e-06,
      "loss": 0.561,
      "num_input_tokens_seen": 196566484,
      "step": 3509
    },
    {
      "epoch": 7.815144766146993,
      "loss": 0.47554415464401245,
      "loss_ce": 0.00020236926502548158,
      "loss_iou": 0.1865234375,
      "loss_num": 0.0206298828125,
      "loss_xval": 0.474609375,
      "num_input_tokens_seen": 196566484,
      "step": 3509
    },
    {
      "epoch": 7.817371937639198,
      "grad_norm": 31.580455780029297,
      "learning_rate": 1e-06,
      "loss": 0.584,
      "num_input_tokens_seen": 196621400,
      "step": 3510
    },
    {
      "epoch": 7.817371937639198,
      "loss": 0.6484927535057068,
      "loss_ce": 0.00017732605920173228,
      "loss_iou": 0.275390625,
      "loss_num": 0.019775390625,
      "loss_xval": 0.6484375,
      "num_input_tokens_seen": 196621400,
      "step": 3510
    },
    {
      "epoch": 7.819599109131403,
      "grad_norm": 14.395856857299805,
      "learning_rate": 1e-06,
      "loss": 0.5939,
      "num_input_tokens_seen": 196678068,
      "step": 3511
    },
    {
      "epoch": 7.819599109131403,
      "loss": 0.5976732969284058,
      "loss_ce": 0.0001391283149132505,
      "loss_iou": 0.248046875,
      "loss_num": 0.020263671875,
      "loss_xval": 0.59765625,
      "num_input_tokens_seen": 196678068,
      "step": 3511
    },
    {
      "epoch": 7.821826280623608,
      "grad_norm": 22.546255111694336,
      "learning_rate": 1e-06,
      "loss": 0.5801,
      "num_input_tokens_seen": 196733972,
      "step": 3512
    },
    {
      "epoch": 7.821826280623608,
      "loss": 0.467131644487381,
      "loss_ce": 0.00015164985961746424,
      "loss_iou": 0.212890625,
      "loss_num": 0.00811767578125,
      "loss_xval": 0.466796875,
      "num_input_tokens_seen": 196733972,
      "step": 3512
    },
    {
      "epoch": 7.8240534521158125,
      "grad_norm": 16.682117462158203,
      "learning_rate": 1e-06,
      "loss": 0.5618,
      "num_input_tokens_seen": 196790868,
      "step": 3513
    },
    {
      "epoch": 7.8240534521158125,
      "loss": 0.6464253664016724,
      "loss_ce": 0.00018512400856707245,
      "loss_iou": 0.267578125,
      "loss_num": 0.02197265625,
      "loss_xval": 0.64453125,
      "num_input_tokens_seen": 196790868,
      "step": 3513
    },
    {
      "epoch": 7.826280623608017,
      "grad_norm": 18.157453536987305,
      "learning_rate": 1e-06,
      "loss": 0.6781,
      "num_input_tokens_seen": 196847364,
      "step": 3514
    },
    {
      "epoch": 7.826280623608017,
      "loss": 0.6460493803024292,
      "loss_ce": 0.0001753381366143003,
      "loss_iou": 0.279296875,
      "loss_num": 0.017578125,
      "loss_xval": 0.64453125,
      "num_input_tokens_seen": 196847364,
      "step": 3514
    },
    {
      "epoch": 7.828507795100223,
      "grad_norm": 20.77778434753418,
      "learning_rate": 1e-06,
      "loss": 0.743,
      "num_input_tokens_seen": 196904288,
      "step": 3515
    },
    {
      "epoch": 7.828507795100223,
      "loss": 0.7536814212799072,
      "loss_ce": 0.00026343436911702156,
      "loss_iou": 0.318359375,
      "loss_num": 0.0235595703125,
      "loss_xval": 0.75390625,
      "num_input_tokens_seen": 196904288,
      "step": 3515
    },
    {
      "epoch": 7.830734966592428,
      "grad_norm": 15.515265464782715,
      "learning_rate": 1e-06,
      "loss": 0.4926,
      "num_input_tokens_seen": 196961928,
      "step": 3516
    },
    {
      "epoch": 7.830734966592428,
      "loss": 0.4176163673400879,
      "loss_ce": 0.00013586709974333644,
      "loss_iou": 0.1689453125,
      "loss_num": 0.01611328125,
      "loss_xval": 0.41796875,
      "num_input_tokens_seen": 196961928,
      "step": 3516
    },
    {
      "epoch": 7.832962138084633,
      "grad_norm": 23.900028228759766,
      "learning_rate": 1e-06,
      "loss": 0.4868,
      "num_input_tokens_seen": 197017236,
      "step": 3517
    },
    {
      "epoch": 7.832962138084633,
      "loss": 0.49435505270957947,
      "loss_ce": 0.00015339165111072361,
      "loss_iou": 0.2275390625,
      "loss_num": 0.0079345703125,
      "loss_xval": 0.494140625,
      "num_input_tokens_seen": 197017236,
      "step": 3517
    },
    {
      "epoch": 7.835189309576838,
      "grad_norm": 15.74838638305664,
      "learning_rate": 1e-06,
      "loss": 0.7655,
      "num_input_tokens_seen": 197073028,
      "step": 3518
    },
    {
      "epoch": 7.835189309576838,
      "loss": 1.0873568058013916,
      "loss_ce": 0.00019860133761540055,
      "loss_iou": 0.443359375,
      "loss_num": 0.04052734375,
      "loss_xval": 1.0859375,
      "num_input_tokens_seen": 197073028,
      "step": 3518
    },
    {
      "epoch": 7.8374164810690425,
      "grad_norm": 25.064544677734375,
      "learning_rate": 1e-06,
      "loss": 0.552,
      "num_input_tokens_seen": 197130308,
      "step": 3519
    },
    {
      "epoch": 7.8374164810690425,
      "loss": 0.512490451335907,
      "loss_ce": 0.0001613302156329155,
      "loss_iou": 0.2158203125,
      "loss_num": 0.01611328125,
      "loss_xval": 0.51171875,
      "num_input_tokens_seen": 197130308,
      "step": 3519
    },
    {
      "epoch": 7.839643652561247,
      "grad_norm": 20.62727928161621,
      "learning_rate": 1e-06,
      "loss": 0.4757,
      "num_input_tokens_seen": 197186840,
      "step": 3520
    },
    {
      "epoch": 7.839643652561247,
      "loss": 0.6267750263214111,
      "loss_ce": 0.000188069258001633,
      "loss_iou": 0.2734375,
      "loss_num": 0.015625,
      "loss_xval": 0.625,
      "num_input_tokens_seen": 197186840,
      "step": 3520
    },
    {
      "epoch": 7.841870824053452,
      "grad_norm": 15.101804733276367,
      "learning_rate": 1e-06,
      "loss": 0.6523,
      "num_input_tokens_seen": 197244324,
      "step": 3521
    },
    {
      "epoch": 7.841870824053452,
      "loss": 0.8312093019485474,
      "loss_ce": 0.0002156283298972994,
      "loss_iou": 0.357421875,
      "loss_num": 0.0233154296875,
      "loss_xval": 0.83203125,
      "num_input_tokens_seen": 197244324,
      "step": 3521
    },
    {
      "epoch": 7.844097995545657,
      "grad_norm": 27.03868865966797,
      "learning_rate": 1e-06,
      "loss": 0.52,
      "num_input_tokens_seen": 197299300,
      "step": 3522
    },
    {
      "epoch": 7.844097995545657,
      "loss": 0.5687766075134277,
      "loss_ce": 0.0002951917704194784,
      "loss_iou": 0.244140625,
      "loss_num": 0.015869140625,
      "loss_xval": 0.5703125,
      "num_input_tokens_seen": 197299300,
      "step": 3522
    },
    {
      "epoch": 7.846325167037862,
      "grad_norm": 51.59347915649414,
      "learning_rate": 1e-06,
      "loss": 0.6587,
      "num_input_tokens_seen": 197355036,
      "step": 3523
    },
    {
      "epoch": 7.846325167037862,
      "loss": 0.6813713908195496,
      "loss_ce": 0.0002190661762142554,
      "loss_iou": 0.298828125,
      "loss_num": 0.0166015625,
      "loss_xval": 0.6796875,
      "num_input_tokens_seen": 197355036,
      "step": 3523
    },
    {
      "epoch": 7.848552338530067,
      "grad_norm": 14.566356658935547,
      "learning_rate": 1e-06,
      "loss": 0.5917,
      "num_input_tokens_seen": 197410556,
      "step": 3524
    },
    {
      "epoch": 7.848552338530067,
      "loss": 0.5617358684539795,
      "loss_ce": 0.0002124165475834161,
      "loss_iou": 0.2255859375,
      "loss_num": 0.0220947265625,
      "loss_xval": 0.5625,
      "num_input_tokens_seen": 197410556,
      "step": 3524
    },
    {
      "epoch": 7.850779510022272,
      "grad_norm": 17.55548858642578,
      "learning_rate": 1e-06,
      "loss": 0.5818,
      "num_input_tokens_seen": 197468284,
      "step": 3525
    },
    {
      "epoch": 7.850779510022272,
      "loss": 0.6032076478004456,
      "loss_ce": 0.0001802719198167324,
      "loss_iou": 0.26953125,
      "loss_num": 0.01300048828125,
      "loss_xval": 0.6015625,
      "num_input_tokens_seen": 197468284,
      "step": 3525
    },
    {
      "epoch": 7.853006681514477,
      "grad_norm": 48.917110443115234,
      "learning_rate": 1e-06,
      "loss": 0.7346,
      "num_input_tokens_seen": 197525216,
      "step": 3526
    },
    {
      "epoch": 7.853006681514477,
      "loss": 0.9218271970748901,
      "loss_ce": 0.0001963673421414569,
      "loss_iou": 0.375,
      "loss_num": 0.033935546875,
      "loss_xval": 0.921875,
      "num_input_tokens_seen": 197525216,
      "step": 3526
    },
    {
      "epoch": 7.855233853006681,
      "grad_norm": 32.04833984375,
      "learning_rate": 1e-06,
      "loss": 0.5597,
      "num_input_tokens_seen": 197580472,
      "step": 3527
    },
    {
      "epoch": 7.855233853006681,
      "loss": 0.5170796513557434,
      "loss_ce": 0.00023395352764055133,
      "loss_iou": 0.2265625,
      "loss_num": 0.012451171875,
      "loss_xval": 0.515625,
      "num_input_tokens_seen": 197580472,
      "step": 3527
    },
    {
      "epoch": 7.857461024498886,
      "grad_norm": 20.859832763671875,
      "learning_rate": 1e-06,
      "loss": 0.6578,
      "num_input_tokens_seen": 197636904,
      "step": 3528
    },
    {
      "epoch": 7.857461024498886,
      "loss": 0.8837323188781738,
      "loss_ce": 0.00018741836538538337,
      "loss_iou": 0.34765625,
      "loss_num": 0.03759765625,
      "loss_xval": 0.8828125,
      "num_input_tokens_seen": 197636904,
      "step": 3528
    },
    {
      "epoch": 7.859688195991091,
      "grad_norm": 20.60695457458496,
      "learning_rate": 1e-06,
      "loss": 0.7797,
      "num_input_tokens_seen": 197689184,
      "step": 3529
    },
    {
      "epoch": 7.859688195991091,
      "loss": 0.9426758289337158,
      "loss_ce": 0.0003540791803970933,
      "loss_iou": 0.39453125,
      "loss_num": 0.0303955078125,
      "loss_xval": 0.94140625,
      "num_input_tokens_seen": 197689184,
      "step": 3529
    },
    {
      "epoch": 7.861915367483296,
      "grad_norm": 23.411840438842773,
      "learning_rate": 1e-06,
      "loss": 0.7374,
      "num_input_tokens_seen": 197745056,
      "step": 3530
    },
    {
      "epoch": 7.861915367483296,
      "loss": 0.8417353630065918,
      "loss_ce": 0.00018263611127622426,
      "loss_iou": 0.337890625,
      "loss_num": 0.032958984375,
      "loss_xval": 0.83984375,
      "num_input_tokens_seen": 197745056,
      "step": 3530
    },
    {
      "epoch": 7.864142538975501,
      "grad_norm": 14.93423843383789,
      "learning_rate": 1e-06,
      "loss": 0.4174,
      "num_input_tokens_seen": 197801836,
      "step": 3531
    },
    {
      "epoch": 7.864142538975501,
      "loss": 0.4481639862060547,
      "loss_ce": 0.00016593134205322713,
      "loss_iou": 0.171875,
      "loss_num": 0.0208740234375,
      "loss_xval": 0.447265625,
      "num_input_tokens_seen": 197801836,
      "step": 3531
    },
    {
      "epoch": 7.866369710467706,
      "grad_norm": 19.369924545288086,
      "learning_rate": 1e-06,
      "loss": 0.6988,
      "num_input_tokens_seen": 197858596,
      "step": 3532
    },
    {
      "epoch": 7.866369710467706,
      "loss": 0.987027108669281,
      "loss_ce": 0.0002106587344314903,
      "loss_iou": 0.412109375,
      "loss_num": 0.032470703125,
      "loss_xval": 0.98828125,
      "num_input_tokens_seen": 197858596,
      "step": 3532
    },
    {
      "epoch": 7.868596881959911,
      "grad_norm": 37.832054138183594,
      "learning_rate": 1e-06,
      "loss": 0.7192,
      "num_input_tokens_seen": 197915224,
      "step": 3533
    },
    {
      "epoch": 7.868596881959911,
      "loss": 0.7180832028388977,
      "loss_ce": 0.000309744878904894,
      "loss_iou": 0.30078125,
      "loss_num": 0.023681640625,
      "loss_xval": 0.71875,
      "num_input_tokens_seen": 197915224,
      "step": 3533
    },
    {
      "epoch": 7.870824053452115,
      "grad_norm": 13.525866508483887,
      "learning_rate": 1e-06,
      "loss": 0.4544,
      "num_input_tokens_seen": 197971236,
      "step": 3534
    },
    {
      "epoch": 7.870824053452115,
      "loss": 0.5098298788070679,
      "loss_ce": 0.0001863364304881543,
      "loss_iou": 0.2158203125,
      "loss_num": 0.015380859375,
      "loss_xval": 0.5078125,
      "num_input_tokens_seen": 197971236,
      "step": 3534
    },
    {
      "epoch": 7.873051224944321,
      "grad_norm": 20.633840560913086,
      "learning_rate": 1e-06,
      "loss": 0.6103,
      "num_input_tokens_seen": 198029292,
      "step": 3535
    },
    {
      "epoch": 7.873051224944321,
      "loss": 0.7734463810920715,
      "loss_ce": 0.0002530375204514712,
      "loss_iou": 0.361328125,
      "loss_num": 0.0103759765625,
      "loss_xval": 0.7734375,
      "num_input_tokens_seen": 198029292,
      "step": 3535
    },
    {
      "epoch": 7.875278396436526,
      "grad_norm": 169.28097534179688,
      "learning_rate": 1e-06,
      "loss": 0.6063,
      "num_input_tokens_seen": 198083980,
      "step": 3536
    },
    {
      "epoch": 7.875278396436526,
      "loss": 0.52978515625,
      "loss_ce": 0.0004882930079475045,
      "loss_iou": 0.2294921875,
      "loss_num": 0.0140380859375,
      "loss_xval": 0.53125,
      "num_input_tokens_seen": 198083980,
      "step": 3536
    },
    {
      "epoch": 7.877505567928731,
      "grad_norm": 20.492658615112305,
      "learning_rate": 1e-06,
      "loss": 0.5758,
      "num_input_tokens_seen": 198139948,
      "step": 3537
    },
    {
      "epoch": 7.877505567928731,
      "loss": 0.6573839783668518,
      "loss_ce": 0.00015742500545457006,
      "loss_iou": 0.271484375,
      "loss_num": 0.0230712890625,
      "loss_xval": 0.65625,
      "num_input_tokens_seen": 198139948,
      "step": 3537
    },
    {
      "epoch": 7.879732739420936,
      "grad_norm": 27.89311981201172,
      "learning_rate": 1e-06,
      "loss": 0.8226,
      "num_input_tokens_seen": 198198180,
      "step": 3538
    },
    {
      "epoch": 7.879732739420936,
      "loss": 0.9203274250030518,
      "loss_ce": 0.00016145262634381652,
      "loss_iou": 0.40625,
      "loss_num": 0.021728515625,
      "loss_xval": 0.921875,
      "num_input_tokens_seen": 198198180,
      "step": 3538
    },
    {
      "epoch": 7.881959910913141,
      "grad_norm": 21.136817932128906,
      "learning_rate": 1e-06,
      "loss": 0.5017,
      "num_input_tokens_seen": 198253928,
      "step": 3539
    },
    {
      "epoch": 7.881959910913141,
      "loss": 0.5626224279403687,
      "loss_ce": 0.0003665737749543041,
      "loss_iou": 0.248046875,
      "loss_num": 0.0133056640625,
      "loss_xval": 0.5625,
      "num_input_tokens_seen": 198253928,
      "step": 3539
    },
    {
      "epoch": 7.8841870824053455,
      "grad_norm": 28.12201499938965,
      "learning_rate": 1e-06,
      "loss": 0.4507,
      "num_input_tokens_seen": 198309852,
      "step": 3540
    },
    {
      "epoch": 7.8841870824053455,
      "loss": 0.5494855642318726,
      "loss_ce": 0.00016917182074394077,
      "loss_iou": 0.22265625,
      "loss_num": 0.0206298828125,
      "loss_xval": 0.55078125,
      "num_input_tokens_seen": 198309852,
      "step": 3540
    },
    {
      "epoch": 7.88641425389755,
      "grad_norm": 47.74600601196289,
      "learning_rate": 1e-06,
      "loss": 0.5923,
      "num_input_tokens_seen": 198365504,
      "step": 3541
    },
    {
      "epoch": 7.88641425389755,
      "loss": 0.6581393480300903,
      "loss_ce": 0.00042452660272829235,
      "loss_iou": 0.267578125,
      "loss_num": 0.0240478515625,
      "loss_xval": 0.65625,
      "num_input_tokens_seen": 198365504,
      "step": 3541
    },
    {
      "epoch": 7.888641425389755,
      "grad_norm": 11.964970588684082,
      "learning_rate": 1e-06,
      "loss": 0.4837,
      "num_input_tokens_seen": 198420992,
      "step": 3542
    },
    {
      "epoch": 7.888641425389755,
      "loss": 0.46798640489578247,
      "loss_ce": 0.00021299449144862592,
      "loss_iou": 0.2021484375,
      "loss_num": 0.01263427734375,
      "loss_xval": 0.46875,
      "num_input_tokens_seen": 198420992,
      "step": 3542
    },
    {
      "epoch": 7.89086859688196,
      "grad_norm": 14.015028953552246,
      "learning_rate": 1e-06,
      "loss": 0.5561,
      "num_input_tokens_seen": 198478588,
      "step": 3543
    },
    {
      "epoch": 7.89086859688196,
      "loss": 0.5441027879714966,
      "loss_ce": 0.0001574681227793917,
      "loss_iou": 0.234375,
      "loss_num": 0.01495361328125,
      "loss_xval": 0.54296875,
      "num_input_tokens_seen": 198478588,
      "step": 3543
    },
    {
      "epoch": 7.893095768374165,
      "grad_norm": 13.727971076965332,
      "learning_rate": 1e-06,
      "loss": 0.7564,
      "num_input_tokens_seen": 198532012,
      "step": 3544
    },
    {
      "epoch": 7.893095768374165,
      "loss": 0.9394014477729797,
      "loss_ce": 0.0001924369134940207,
      "loss_iou": 0.3671875,
      "loss_num": 0.040771484375,
      "loss_xval": 0.9375,
      "num_input_tokens_seen": 198532012,
      "step": 3544
    },
    {
      "epoch": 7.89532293986637,
      "grad_norm": 24.72553253173828,
      "learning_rate": 1e-06,
      "loss": 0.7666,
      "num_input_tokens_seen": 198587972,
      "step": 3545
    },
    {
      "epoch": 7.89532293986637,
      "loss": 0.6909983158111572,
      "loss_ce": 0.0003245328553020954,
      "loss_iou": 0.2470703125,
      "loss_num": 0.0390625,
      "loss_xval": 0.69140625,
      "num_input_tokens_seen": 198587972,
      "step": 3545
    },
    {
      "epoch": 7.897550111358575,
      "grad_norm": 16.695913314819336,
      "learning_rate": 1e-06,
      "loss": 0.6233,
      "num_input_tokens_seen": 198644912,
      "step": 3546
    },
    {
      "epoch": 7.897550111358575,
      "loss": 0.5989300608634949,
      "loss_ce": 0.00017520345863886178,
      "loss_iou": 0.2412109375,
      "loss_num": 0.023193359375,
      "loss_xval": 0.59765625,
      "num_input_tokens_seen": 198644912,
      "step": 3546
    },
    {
      "epoch": 7.8997772828507795,
      "grad_norm": 23.642194747924805,
      "learning_rate": 1e-06,
      "loss": 0.8229,
      "num_input_tokens_seen": 198701832,
      "step": 3547
    },
    {
      "epoch": 7.8997772828507795,
      "loss": 0.7591318488121033,
      "loss_ce": 0.0008310843259096146,
      "loss_iou": 0.310546875,
      "loss_num": 0.0274658203125,
      "loss_xval": 0.7578125,
      "num_input_tokens_seen": 198701832,
      "step": 3547
    },
    {
      "epoch": 7.902004454342984,
      "grad_norm": 17.38911247253418,
      "learning_rate": 1e-06,
      "loss": 0.4997,
      "num_input_tokens_seen": 198759072,
      "step": 3548
    },
    {
      "epoch": 7.902004454342984,
      "loss": 0.4457376301288605,
      "loss_ce": 0.004087251145392656,
      "loss_iou": 0.1796875,
      "loss_num": 0.016357421875,
      "loss_xval": 0.44140625,
      "num_input_tokens_seen": 198759072,
      "step": 3548
    },
    {
      "epoch": 7.904231625835189,
      "grad_norm": 15.332737922668457,
      "learning_rate": 1e-06,
      "loss": 0.9554,
      "num_input_tokens_seen": 198815520,
      "step": 3549
    },
    {
      "epoch": 7.904231625835189,
      "loss": 0.7602176666259766,
      "loss_ce": 0.00020788329129572958,
      "loss_iou": 0.322265625,
      "loss_num": 0.0235595703125,
      "loss_xval": 0.76171875,
      "num_input_tokens_seen": 198815520,
      "step": 3549
    },
    {
      "epoch": 7.906458797327394,
      "grad_norm": 17.26343536376953,
      "learning_rate": 1e-06,
      "loss": 0.6921,
      "num_input_tokens_seen": 198871008,
      "step": 3550
    },
    {
      "epoch": 7.906458797327394,
      "loss": 0.7331159710884094,
      "loss_ce": 0.00020580444834195077,
      "loss_iou": 0.30078125,
      "loss_num": 0.026123046875,
      "loss_xval": 0.734375,
      "num_input_tokens_seen": 198871008,
      "step": 3550
    },
    {
      "epoch": 7.908685968819599,
      "grad_norm": 18.0777587890625,
      "learning_rate": 1e-06,
      "loss": 0.9367,
      "num_input_tokens_seen": 198927764,
      "step": 3551
    },
    {
      "epoch": 7.908685968819599,
      "loss": 0.9931344985961914,
      "loss_ce": 0.0002145655162166804,
      "loss_iou": 0.4140625,
      "loss_num": 0.032958984375,
      "loss_xval": 0.9921875,
      "num_input_tokens_seen": 198927764,
      "step": 3551
    },
    {
      "epoch": 7.910913140311804,
      "grad_norm": 15.072328567504883,
      "learning_rate": 1e-06,
      "loss": 0.4566,
      "num_input_tokens_seen": 198983208,
      "step": 3552
    },
    {
      "epoch": 7.910913140311804,
      "loss": 0.3726545572280884,
      "loss_ce": 0.00021803012350574136,
      "loss_iou": 0.1689453125,
      "loss_num": 0.00689697265625,
      "loss_xval": 0.373046875,
      "num_input_tokens_seen": 198983208,
      "step": 3552
    },
    {
      "epoch": 7.913140311804009,
      "grad_norm": 23.05716896057129,
      "learning_rate": 1e-06,
      "loss": 0.541,
      "num_input_tokens_seen": 199038544,
      "step": 3553
    },
    {
      "epoch": 7.913140311804009,
      "loss": 0.4792807102203369,
      "loss_ce": 0.000154730340000242,
      "loss_iou": 0.2138671875,
      "loss_num": 0.0101318359375,
      "loss_xval": 0.478515625,
      "num_input_tokens_seen": 199038544,
      "step": 3553
    },
    {
      "epoch": 7.9153674832962135,
      "grad_norm": 14.205634117126465,
      "learning_rate": 1e-06,
      "loss": 0.511,
      "num_input_tokens_seen": 199093300,
      "step": 3554
    },
    {
      "epoch": 7.9153674832962135,
      "loss": 0.5586905479431152,
      "loss_ce": 0.00021891451615374535,
      "loss_iou": 0.21875,
      "loss_num": 0.0242919921875,
      "loss_xval": 0.55859375,
      "num_input_tokens_seen": 199093300,
      "step": 3554
    },
    {
      "epoch": 7.917594654788418,
      "grad_norm": 20.449451446533203,
      "learning_rate": 1e-06,
      "loss": 0.529,
      "num_input_tokens_seen": 199151908,
      "step": 3555
    },
    {
      "epoch": 7.917594654788418,
      "loss": 0.3755338788032532,
      "loss_ce": 0.0001676726678851992,
      "loss_iou": 0.1689453125,
      "loss_num": 0.00750732421875,
      "loss_xval": 0.375,
      "num_input_tokens_seen": 199151908,
      "step": 3555
    },
    {
      "epoch": 7.919821826280623,
      "grad_norm": 24.3996639251709,
      "learning_rate": 1e-06,
      "loss": 0.775,
      "num_input_tokens_seen": 199204824,
      "step": 3556
    },
    {
      "epoch": 7.919821826280623,
      "loss": 0.6280686855316162,
      "loss_ce": 0.00016949971904978156,
      "loss_iou": 0.265625,
      "loss_num": 0.01953125,
      "loss_xval": 0.62890625,
      "num_input_tokens_seen": 199204824,
      "step": 3556
    },
    {
      "epoch": 7.922048997772828,
      "grad_norm": 16.553987503051758,
      "learning_rate": 1e-06,
      "loss": 0.7373,
      "num_input_tokens_seen": 199259004,
      "step": 3557
    },
    {
      "epoch": 7.922048997772828,
      "loss": 0.8635649681091309,
      "loss_ce": 0.00016163403051905334,
      "loss_iou": 0.3671875,
      "loss_num": 0.0255126953125,
      "loss_xval": 0.86328125,
      "num_input_tokens_seen": 199259004,
      "step": 3557
    },
    {
      "epoch": 7.924276169265033,
      "grad_norm": 20.715532302856445,
      "learning_rate": 1e-06,
      "loss": 0.5043,
      "num_input_tokens_seen": 199314500,
      "step": 3558
    },
    {
      "epoch": 7.924276169265033,
      "loss": 0.5213112235069275,
      "loss_ce": 0.00019308787886984646,
      "loss_iou": 0.234375,
      "loss_num": 0.01043701171875,
      "loss_xval": 0.51953125,
      "num_input_tokens_seen": 199314500,
      "step": 3558
    },
    {
      "epoch": 7.926503340757238,
      "grad_norm": 33.767356872558594,
      "learning_rate": 1e-06,
      "loss": 0.4129,
      "num_input_tokens_seen": 199372168,
      "step": 3559
    },
    {
      "epoch": 7.926503340757238,
      "loss": 0.31914597749710083,
      "loss_ce": 0.0001762679312378168,
      "loss_iou": 0.1484375,
      "loss_num": 0.0042724609375,
      "loss_xval": 0.318359375,
      "num_input_tokens_seen": 199372168,
      "step": 3559
    },
    {
      "epoch": 7.928730512249444,
      "grad_norm": 23.694744110107422,
      "learning_rate": 1e-06,
      "loss": 0.713,
      "num_input_tokens_seen": 199425588,
      "step": 3560
    },
    {
      "epoch": 7.928730512249444,
      "loss": 0.7644591331481934,
      "loss_ce": 0.00042102765291929245,
      "loss_iou": 0.302734375,
      "loss_num": 0.031494140625,
      "loss_xval": 0.765625,
      "num_input_tokens_seen": 199425588,
      "step": 3560
    },
    {
      "epoch": 7.9309576837416484,
      "grad_norm": 22.84273338317871,
      "learning_rate": 1e-06,
      "loss": 0.6157,
      "num_input_tokens_seen": 199483636,
      "step": 3561
    },
    {
      "epoch": 7.9309576837416484,
      "loss": 0.646425724029541,
      "loss_ce": 0.00018549279775470495,
      "loss_iou": 0.275390625,
      "loss_num": 0.01904296875,
      "loss_xval": 0.64453125,
      "num_input_tokens_seen": 199483636,
      "step": 3561
    },
    {
      "epoch": 7.933184855233853,
      "grad_norm": 16.44258689880371,
      "learning_rate": 1e-06,
      "loss": 0.507,
      "num_input_tokens_seen": 199538272,
      "step": 3562
    },
    {
      "epoch": 7.933184855233853,
      "loss": 0.5787392854690552,
      "loss_ce": 0.00024805517750792205,
      "loss_iou": 0.2578125,
      "loss_num": 0.0125732421875,
      "loss_xval": 0.578125,
      "num_input_tokens_seen": 199538272,
      "step": 3562
    },
    {
      "epoch": 7.935412026726058,
      "grad_norm": 20.36518669128418,
      "learning_rate": 1e-06,
      "loss": 0.6295,
      "num_input_tokens_seen": 199596328,
      "step": 3563
    },
    {
      "epoch": 7.935412026726058,
      "loss": 0.641238272190094,
      "loss_ce": 0.00036911331699229777,
      "loss_iou": 0.296875,
      "loss_num": 0.009521484375,
      "loss_xval": 0.640625,
      "num_input_tokens_seen": 199596328,
      "step": 3563
    },
    {
      "epoch": 7.937639198218263,
      "grad_norm": 49.14147186279297,
      "learning_rate": 1e-06,
      "loss": 0.6839,
      "num_input_tokens_seen": 199653252,
      "step": 3564
    },
    {
      "epoch": 7.937639198218263,
      "loss": 0.6464834213256836,
      "loss_ce": 0.00024316800408996642,
      "loss_iou": 0.30078125,
      "loss_num": 0.00927734375,
      "loss_xval": 0.64453125,
      "num_input_tokens_seen": 199653252,
      "step": 3564
    },
    {
      "epoch": 7.939866369710468,
      "grad_norm": 16.839582443237305,
      "learning_rate": 1e-06,
      "loss": 0.6352,
      "num_input_tokens_seen": 199708264,
      "step": 3565
    },
    {
      "epoch": 7.939866369710468,
      "loss": 0.49529188871383667,
      "loss_ce": 0.0001746913476381451,
      "loss_iou": 0.212890625,
      "loss_num": 0.01397705078125,
      "loss_xval": 0.49609375,
      "num_input_tokens_seen": 199708264,
      "step": 3565
    },
    {
      "epoch": 7.942093541202673,
      "grad_norm": 16.844385147094727,
      "learning_rate": 1e-06,
      "loss": 0.6126,
      "num_input_tokens_seen": 199762920,
      "step": 3566
    },
    {
      "epoch": 7.942093541202673,
      "loss": 0.6229934096336365,
      "loss_ce": 0.00019066702225245535,
      "loss_iou": 0.263671875,
      "loss_num": 0.0194091796875,
      "loss_xval": 0.62109375,
      "num_input_tokens_seen": 199762920,
      "step": 3566
    },
    {
      "epoch": 7.944320712694878,
      "grad_norm": 23.8248233795166,
      "learning_rate": 1e-06,
      "loss": 0.6024,
      "num_input_tokens_seen": 199820520,
      "step": 3567
    },
    {
      "epoch": 7.944320712694878,
      "loss": 0.6216274499893188,
      "loss_ce": 0.00016751314979046583,
      "loss_iou": 0.26953125,
      "loss_num": 0.0164794921875,
      "loss_xval": 0.62109375,
      "num_input_tokens_seen": 199820520,
      "step": 3567
    },
    {
      "epoch": 7.9465478841870825,
      "grad_norm": 13.384178161621094,
      "learning_rate": 1e-06,
      "loss": 0.6889,
      "num_input_tokens_seen": 199874968,
      "step": 3568
    },
    {
      "epoch": 7.9465478841870825,
      "loss": 0.6415170431137085,
      "loss_ce": 0.00022065843222662807,
      "loss_iou": 0.259765625,
      "loss_num": 0.0242919921875,
      "loss_xval": 0.640625,
      "num_input_tokens_seen": 199874968,
      "step": 3568
    },
    {
      "epoch": 7.948775055679287,
      "grad_norm": 17.003244400024414,
      "learning_rate": 1e-06,
      "loss": 0.7046,
      "num_input_tokens_seen": 199931032,
      "step": 3569
    },
    {
      "epoch": 7.948775055679287,
      "loss": 0.7286696434020996,
      "loss_ce": 0.00015403382712975144,
      "loss_iou": 0.326171875,
      "loss_num": 0.0150146484375,
      "loss_xval": 0.7265625,
      "num_input_tokens_seen": 199931032,
      "step": 3569
    },
    {
      "epoch": 7.951002227171492,
      "grad_norm": 18.784358978271484,
      "learning_rate": 1e-06,
      "loss": 0.6467,
      "num_input_tokens_seen": 199986240,
      "step": 3570
    },
    {
      "epoch": 7.951002227171492,
      "loss": 0.711881160736084,
      "loss_ce": 0.00021116853167768568,
      "loss_iou": 0.3203125,
      "loss_num": 0.01458740234375,
      "loss_xval": 0.7109375,
      "num_input_tokens_seen": 199986240,
      "step": 3570
    },
    {
      "epoch": 7.953229398663697,
      "grad_norm": 25.438982009887695,
      "learning_rate": 1e-06,
      "loss": 0.6809,
      "num_input_tokens_seen": 200041048,
      "step": 3571
    },
    {
      "epoch": 7.953229398663697,
      "loss": 0.8664193153381348,
      "loss_ce": 0.00020840237266384065,
      "loss_iou": 0.34375,
      "loss_num": 0.03564453125,
      "loss_xval": 0.8671875,
      "num_input_tokens_seen": 200041048,
      "step": 3571
    },
    {
      "epoch": 7.955456570155902,
      "grad_norm": 18.567123413085938,
      "learning_rate": 1e-06,
      "loss": 0.6805,
      "num_input_tokens_seen": 200098136,
      "step": 3572
    },
    {
      "epoch": 7.955456570155902,
      "loss": 0.8434309363365173,
      "loss_ce": 0.0001692144141998142,
      "loss_iou": 0.35546875,
      "loss_num": 0.02685546875,
      "loss_xval": 0.84375,
      "num_input_tokens_seen": 200098136,
      "step": 3572
    },
    {
      "epoch": 7.957683741648107,
      "grad_norm": 271.4332275390625,
      "learning_rate": 1e-06,
      "loss": 0.6131,
      "num_input_tokens_seen": 200152360,
      "step": 3573
    },
    {
      "epoch": 7.957683741648107,
      "loss": 0.7838395833969116,
      "loss_ce": 0.00014815322356298566,
      "loss_iou": 0.3359375,
      "loss_num": 0.0225830078125,
      "loss_xval": 0.78515625,
      "num_input_tokens_seen": 200152360,
      "step": 3573
    },
    {
      "epoch": 7.959910913140312,
      "grad_norm": 18.194717407226562,
      "learning_rate": 1e-06,
      "loss": 0.5837,
      "num_input_tokens_seen": 200208800,
      "step": 3574
    },
    {
      "epoch": 7.959910913140312,
      "loss": 0.3856315314769745,
      "loss_ce": 0.00013348979700822383,
      "loss_iou": 0.15625,
      "loss_num": 0.0146484375,
      "loss_xval": 0.384765625,
      "num_input_tokens_seen": 200208800,
      "step": 3574
    },
    {
      "epoch": 7.9621380846325165,
      "grad_norm": 23.228425979614258,
      "learning_rate": 1e-06,
      "loss": 0.8644,
      "num_input_tokens_seen": 200265476,
      "step": 3575
    },
    {
      "epoch": 7.9621380846325165,
      "loss": 0.6910591125488281,
      "loss_ce": 0.0001411863195244223,
      "loss_iou": 0.263671875,
      "loss_num": 0.032958984375,
      "loss_xval": 0.69140625,
      "num_input_tokens_seen": 200265476,
      "step": 3575
    },
    {
      "epoch": 7.964365256124721,
      "grad_norm": 28.046964645385742,
      "learning_rate": 1e-06,
      "loss": 0.6866,
      "num_input_tokens_seen": 200322152,
      "step": 3576
    },
    {
      "epoch": 7.964365256124721,
      "loss": 0.7025761604309082,
      "loss_ce": 0.00018362130504101515,
      "loss_iou": 0.294921875,
      "loss_num": 0.0224609375,
      "loss_xval": 0.703125,
      "num_input_tokens_seen": 200322152,
      "step": 3576
    },
    {
      "epoch": 7.966592427616926,
      "grad_norm": 23.809234619140625,
      "learning_rate": 1e-06,
      "loss": 0.5926,
      "num_input_tokens_seen": 200375664,
      "step": 3577
    },
    {
      "epoch": 7.966592427616926,
      "loss": 0.6842349767684937,
      "loss_ce": 0.00015298397920560092,
      "loss_iou": 0.31640625,
      "loss_num": 0.010498046875,
      "loss_xval": 0.68359375,
      "num_input_tokens_seen": 200375664,
      "step": 3577
    },
    {
      "epoch": 7.968819599109131,
      "grad_norm": 35.44462966918945,
      "learning_rate": 1e-06,
      "loss": 0.6293,
      "num_input_tokens_seen": 200430944,
      "step": 3578
    },
    {
      "epoch": 7.968819599109131,
      "loss": 0.6284739375114441,
      "loss_ce": 0.0001780486200004816,
      "loss_iou": 0.28515625,
      "loss_num": 0.01129150390625,
      "loss_xval": 0.62890625,
      "num_input_tokens_seen": 200430944,
      "step": 3578
    },
    {
      "epoch": 7.971046770601336,
      "grad_norm": 22.342145919799805,
      "learning_rate": 1e-06,
      "loss": 0.5531,
      "num_input_tokens_seen": 200486764,
      "step": 3579
    },
    {
      "epoch": 7.971046770601336,
      "loss": 0.5724341869354248,
      "loss_ce": 0.0001685347524471581,
      "loss_iou": 0.259765625,
      "loss_num": 0.01055908203125,
      "loss_xval": 0.5703125,
      "num_input_tokens_seen": 200486764,
      "step": 3579
    },
    {
      "epoch": 7.973273942093542,
      "grad_norm": 18.398874282836914,
      "learning_rate": 1e-06,
      "loss": 0.641,
      "num_input_tokens_seen": 200544420,
      "step": 3580
    },
    {
      "epoch": 7.973273942093542,
      "loss": 0.6851996779441833,
      "loss_ce": 0.0001410768600180745,
      "loss_iou": 0.28515625,
      "loss_num": 0.0228271484375,
      "loss_xval": 0.68359375,
      "num_input_tokens_seen": 200544420,
      "step": 3580
    },
    {
      "epoch": 7.9755011135857465,
      "grad_norm": 12.79372787475586,
      "learning_rate": 1e-06,
      "loss": 0.3946,
      "num_input_tokens_seen": 200601156,
      "step": 3581
    },
    {
      "epoch": 7.9755011135857465,
      "loss": 0.40876662731170654,
      "loss_ce": 0.0010518098715692759,
      "loss_iou": 0.177734375,
      "loss_num": 0.01043701171875,
      "loss_xval": 0.408203125,
      "num_input_tokens_seen": 200601156,
      "step": 3581
    },
    {
      "epoch": 7.977728285077951,
      "grad_norm": 15.637197494506836,
      "learning_rate": 1e-06,
      "loss": 0.8963,
      "num_input_tokens_seen": 200657656,
      "step": 3582
    },
    {
      "epoch": 7.977728285077951,
      "loss": 1.1127707958221436,
      "loss_ce": 0.0002219329762738198,
      "loss_iou": 0.4453125,
      "loss_num": 0.043701171875,
      "loss_xval": 1.109375,
      "num_input_tokens_seen": 200657656,
      "step": 3582
    },
    {
      "epoch": 7.979955456570156,
      "grad_norm": 19.720239639282227,
      "learning_rate": 1e-06,
      "loss": 0.7492,
      "num_input_tokens_seen": 200713320,
      "step": 3583
    },
    {
      "epoch": 7.979955456570156,
      "loss": 0.5831568241119385,
      "loss_ce": 0.000149025785503909,
      "loss_iou": 0.251953125,
      "loss_num": 0.0155029296875,
      "loss_xval": 0.58203125,
      "num_input_tokens_seen": 200713320,
      "step": 3583
    },
    {
      "epoch": 7.982182628062361,
      "grad_norm": 13.927251815795898,
      "learning_rate": 1e-06,
      "loss": 0.371,
      "num_input_tokens_seen": 200768308,
      "step": 3584
    },
    {
      "epoch": 7.982182628062361,
      "loss": 0.3358955383300781,
      "loss_ce": 0.00044634263031184673,
      "loss_iou": 0.1416015625,
      "loss_num": 0.01055908203125,
      "loss_xval": 0.3359375,
      "num_input_tokens_seen": 200768308,
      "step": 3584
    },
    {
      "epoch": 7.984409799554566,
      "grad_norm": 19.705570220947266,
      "learning_rate": 1e-06,
      "loss": 0.5878,
      "num_input_tokens_seen": 200821804,
      "step": 3585
    },
    {
      "epoch": 7.984409799554566,
      "loss": 0.4889172911643982,
      "loss_ce": 0.00020879982912447304,
      "loss_iou": 0.212890625,
      "loss_num": 0.01275634765625,
      "loss_xval": 0.48828125,
      "num_input_tokens_seen": 200821804,
      "step": 3585
    },
    {
      "epoch": 7.986636971046771,
      "grad_norm": 125.53588104248047,
      "learning_rate": 1e-06,
      "loss": 0.778,
      "num_input_tokens_seen": 200880704,
      "step": 3586
    },
    {
      "epoch": 7.986636971046771,
      "loss": 0.7231355905532837,
      "loss_ce": 0.00023522632545791566,
      "loss_iou": 0.296875,
      "loss_num": 0.0257568359375,
      "loss_xval": 0.72265625,
      "num_input_tokens_seen": 200880704,
      "step": 3586
    },
    {
      "epoch": 7.988864142538976,
      "grad_norm": 15.870854377746582,
      "learning_rate": 1e-06,
      "loss": 0.7217,
      "num_input_tokens_seen": 200937412,
      "step": 3587
    },
    {
      "epoch": 7.988864142538976,
      "loss": 0.3839578628540039,
      "loss_ce": 0.00022984019597060978,
      "loss_iou": 0.1318359375,
      "loss_num": 0.02392578125,
      "loss_xval": 0.3828125,
      "num_input_tokens_seen": 200937412,
      "step": 3587
    },
    {
      "epoch": 7.991091314031181,
      "grad_norm": 17.978586196899414,
      "learning_rate": 1e-06,
      "loss": 0.495,
      "num_input_tokens_seen": 200992028,
      "step": 3588
    },
    {
      "epoch": 7.991091314031181,
      "loss": 0.49199825525283813,
      "loss_ce": 0.0001769806258380413,
      "loss_iou": 0.2138671875,
      "loss_num": 0.0126953125,
      "loss_xval": 0.4921875,
      "num_input_tokens_seen": 200992028,
      "step": 3588
    },
    {
      "epoch": 7.993318485523385,
      "grad_norm": 18.851573944091797,
      "learning_rate": 1e-06,
      "loss": 0.6107,
      "num_input_tokens_seen": 201049456,
      "step": 3589
    },
    {
      "epoch": 7.993318485523385,
      "loss": 0.6161330938339233,
      "loss_ce": 0.00016627684817649424,
      "loss_iou": 0.26171875,
      "loss_num": 0.01904296875,
      "loss_xval": 0.6171875,
      "num_input_tokens_seen": 201049456,
      "step": 3589
    },
    {
      "epoch": 7.99554565701559,
      "grad_norm": 24.63906478881836,
      "learning_rate": 1e-06,
      "loss": 0.6317,
      "num_input_tokens_seen": 201104932,
      "step": 3590
    },
    {
      "epoch": 7.99554565701559,
      "loss": 0.6456716060638428,
      "loss_ce": 0.00016378730651922524,
      "loss_iou": 0.298828125,
      "loss_num": 0.00946044921875,
      "loss_xval": 0.64453125,
      "num_input_tokens_seen": 201104932,
      "step": 3590
    },
    {
      "epoch": 7.997772828507795,
      "grad_norm": 19.594945907592773,
      "learning_rate": 1e-06,
      "loss": 0.6995,
      "num_input_tokens_seen": 201161408,
      "step": 3591
    },
    {
      "epoch": 7.997772828507795,
      "loss": 0.6261618137359619,
      "loss_ce": 0.00018528125656303018,
      "loss_iou": 0.2734375,
      "loss_num": 0.015869140625,
      "loss_xval": 0.625,
      "num_input_tokens_seen": 201161408,
      "step": 3591
    },
    {
      "epoch": 8.0,
      "grad_norm": 13.603879928588867,
      "learning_rate": 1e-06,
      "loss": 0.6194,
      "num_input_tokens_seen": 201216936,
      "step": 3592
    },
    {
      "epoch": 8.0,
      "loss": 0.7476929426193237,
      "loss_ce": 0.00013434665743261576,
      "loss_iou": 0.33203125,
      "loss_num": 0.0172119140625,
      "loss_xval": 0.74609375,
      "num_input_tokens_seen": 201216936,
      "step": 3592
    },
    {
      "epoch": 8.002227171492205,
      "grad_norm": 18.30463409423828,
      "learning_rate": 1e-06,
      "loss": 0.7265,
      "num_input_tokens_seen": 201274140,
      "step": 3593
    },
    {
      "epoch": 8.002227171492205,
      "loss": 0.8002690076828003,
      "loss_ce": 0.00022022916527930647,
      "loss_iou": 0.353515625,
      "loss_num": 0.0185546875,
      "loss_xval": 0.80078125,
      "num_input_tokens_seen": 201274140,
      "step": 3593
    },
    {
      "epoch": 8.00445434298441,
      "grad_norm": 21.524507522583008,
      "learning_rate": 1e-06,
      "loss": 0.7233,
      "num_input_tokens_seen": 201330520,
      "step": 3594
    },
    {
      "epoch": 8.00445434298441,
      "loss": 0.7201758027076721,
      "loss_ce": 0.00020510726608335972,
      "loss_iou": 0.333984375,
      "loss_num": 0.01055908203125,
      "loss_xval": 0.71875,
      "num_input_tokens_seen": 201330520,
      "step": 3594
    },
    {
      "epoch": 8.006681514476615,
      "grad_norm": 23.734619140625,
      "learning_rate": 1e-06,
      "loss": 0.8149,
      "num_input_tokens_seen": 201386452,
      "step": 3595
    },
    {
      "epoch": 8.006681514476615,
      "loss": 0.6124993562698364,
      "loss_ce": 0.00019466172670945525,
      "loss_iou": 0.2734375,
      "loss_num": 0.01312255859375,
      "loss_xval": 0.61328125,
      "num_input_tokens_seen": 201386452,
      "step": 3595
    },
    {
      "epoch": 8.00890868596882,
      "grad_norm": 20.260971069335938,
      "learning_rate": 1e-06,
      "loss": 0.8065,
      "num_input_tokens_seen": 201444920,
      "step": 3596
    },
    {
      "epoch": 8.00890868596882,
      "loss": 0.48429691791534424,
      "loss_ce": 0.00016606590361334383,
      "loss_iou": 0.20703125,
      "loss_num": 0.01397705078125,
      "loss_xval": 0.484375,
      "num_input_tokens_seen": 201444920,
      "step": 3596
    },
    {
      "epoch": 8.011135857461024,
      "grad_norm": 13.375497817993164,
      "learning_rate": 1e-06,
      "loss": 0.4723,
      "num_input_tokens_seen": 201504192,
      "step": 3597
    },
    {
      "epoch": 8.011135857461024,
      "loss": 0.42642343044281006,
      "loss_ce": 0.00015389773761853576,
      "loss_iou": 0.1943359375,
      "loss_num": 0.0074462890625,
      "loss_xval": 0.42578125,
      "num_input_tokens_seen": 201504192,
      "step": 3597
    },
    {
      "epoch": 8.01336302895323,
      "grad_norm": 13.759163856506348,
      "learning_rate": 1e-06,
      "loss": 0.4227,
      "num_input_tokens_seen": 201560760,
      "step": 3598
    },
    {
      "epoch": 8.01336302895323,
      "loss": 0.5462967157363892,
      "loss_ce": 0.0002762216317933053,
      "loss_iou": 0.2431640625,
      "loss_num": 0.01190185546875,
      "loss_xval": 0.546875,
      "num_input_tokens_seen": 201560760,
      "step": 3598
    },
    {
      "epoch": 8.015590200445434,
      "grad_norm": 13.41096305847168,
      "learning_rate": 1e-06,
      "loss": 0.6107,
      "num_input_tokens_seen": 201614240,
      "step": 3599
    },
    {
      "epoch": 8.015590200445434,
      "loss": 0.5510572791099548,
      "loss_ce": 0.0001539345394121483,
      "loss_iou": 0.2470703125,
      "loss_num": 0.01123046875,
      "loss_xval": 0.55078125,
      "num_input_tokens_seen": 201614240,
      "step": 3599
    },
    {
      "epoch": 8.017817371937639,
      "grad_norm": 23.751489639282227,
      "learning_rate": 1e-06,
      "loss": 0.6214,
      "num_input_tokens_seen": 201672716,
      "step": 3600
    },
    {
      "epoch": 8.017817371937639,
      "loss": 0.7004799842834473,
      "loss_ce": 0.00040674611227586865,
      "loss_iou": 0.314453125,
      "loss_num": 0.0142822265625,
      "loss_xval": 0.69921875,
      "num_input_tokens_seen": 201672716,
      "step": 3600
    },
    {
      "epoch": 8.020044543429844,
      "grad_norm": 14.586628913879395,
      "learning_rate": 1e-06,
      "loss": 0.7327,
      "num_input_tokens_seen": 201729168,
      "step": 3601
    },
    {
      "epoch": 8.020044543429844,
      "loss": 0.636683464050293,
      "loss_ce": 0.00020884581317659467,
      "loss_iou": 0.263671875,
      "loss_num": 0.021728515625,
      "loss_xval": 0.63671875,
      "num_input_tokens_seen": 201729168,
      "step": 3601
    },
    {
      "epoch": 8.022271714922049,
      "grad_norm": 19.21128273010254,
      "learning_rate": 1e-06,
      "loss": 0.5544,
      "num_input_tokens_seen": 201784708,
      "step": 3602
    },
    {
      "epoch": 8.022271714922049,
      "loss": 0.4761182367801666,
      "loss_ce": 0.00016609346494078636,
      "loss_iou": 0.2001953125,
      "loss_num": 0.0152587890625,
      "loss_xval": 0.4765625,
      "num_input_tokens_seen": 201784708,
      "step": 3602
    },
    {
      "epoch": 8.024498886414253,
      "grad_norm": 16.912395477294922,
      "learning_rate": 1e-06,
      "loss": 0.671,
      "num_input_tokens_seen": 201843236,
      "step": 3603
    },
    {
      "epoch": 8.024498886414253,
      "loss": 0.7302785515785217,
      "loss_ce": 0.00017601058061700314,
      "loss_iou": 0.3046875,
      "loss_num": 0.023681640625,
      "loss_xval": 0.73046875,
      "num_input_tokens_seen": 201843236,
      "step": 3603
    },
    {
      "epoch": 8.026726057906458,
      "grad_norm": 18.214143753051758,
      "learning_rate": 1e-06,
      "loss": 0.6397,
      "num_input_tokens_seen": 201897720,
      "step": 3604
    },
    {
      "epoch": 8.026726057906458,
      "loss": 0.41655105352401733,
      "loss_ce": 0.00016921485075727105,
      "loss_iou": 0.19140625,
      "loss_num": 0.006988525390625,
      "loss_xval": 0.416015625,
      "num_input_tokens_seen": 201897720,
      "step": 3604
    },
    {
      "epoch": 8.028953229398663,
      "grad_norm": 18.966310501098633,
      "learning_rate": 1e-06,
      "loss": 0.6243,
      "num_input_tokens_seen": 201953360,
      "step": 3605
    },
    {
      "epoch": 8.028953229398663,
      "loss": 0.5837928056716919,
      "loss_ce": 0.00017461951938457787,
      "loss_iou": 0.232421875,
      "loss_num": 0.0238037109375,
      "loss_xval": 0.58203125,
      "num_input_tokens_seen": 201953360,
      "step": 3605
    },
    {
      "epoch": 8.031180400890868,
      "grad_norm": 18.62992286682129,
      "learning_rate": 1e-06,
      "loss": 0.6118,
      "num_input_tokens_seen": 202008704,
      "step": 3606
    },
    {
      "epoch": 8.031180400890868,
      "loss": 0.5211237668991089,
      "loss_ce": 0.0003718439256772399,
      "loss_iou": 0.216796875,
      "loss_num": 0.0177001953125,
      "loss_xval": 0.51953125,
      "num_input_tokens_seen": 202008704,
      "step": 3606
    },
    {
      "epoch": 8.033407572383073,
      "grad_norm": 24.204988479614258,
      "learning_rate": 1e-06,
      "loss": 0.5671,
      "num_input_tokens_seen": 202068704,
      "step": 3607
    },
    {
      "epoch": 8.033407572383073,
      "loss": 0.4836837947368622,
      "loss_ce": 0.0005294845905154943,
      "loss_iou": 0.1923828125,
      "loss_num": 0.01953125,
      "loss_xval": 0.482421875,
      "num_input_tokens_seen": 202068704,
      "step": 3607
    },
    {
      "epoch": 8.035634743875278,
      "grad_norm": 13.87259292602539,
      "learning_rate": 1e-06,
      "loss": 0.6244,
      "num_input_tokens_seen": 202123772,
      "step": 3608
    },
    {
      "epoch": 8.035634743875278,
      "loss": 0.5231030583381653,
      "loss_ce": 0.0001537989010103047,
      "loss_iou": 0.220703125,
      "loss_num": 0.01611328125,
      "loss_xval": 0.5234375,
      "num_input_tokens_seen": 202123772,
      "step": 3608
    },
    {
      "epoch": 8.037861915367483,
      "grad_norm": 23.374591827392578,
      "learning_rate": 1e-06,
      "loss": 0.7716,
      "num_input_tokens_seen": 202179184,
      "step": 3609
    },
    {
      "epoch": 8.037861915367483,
      "loss": 0.7146378755569458,
      "loss_ce": 0.00016029010294005275,
      "loss_iou": 0.310546875,
      "loss_num": 0.0185546875,
      "loss_xval": 0.71484375,
      "num_input_tokens_seen": 202179184,
      "step": 3609
    },
    {
      "epoch": 8.040089086859687,
      "grad_norm": 18.180028915405273,
      "learning_rate": 1e-06,
      "loss": 0.6477,
      "num_input_tokens_seen": 202234064,
      "step": 3610
    },
    {
      "epoch": 8.040089086859687,
      "loss": 0.8393857479095459,
      "loss_ce": 0.0001523745886515826,
      "loss_iou": 0.359375,
      "loss_num": 0.024169921875,
      "loss_xval": 0.83984375,
      "num_input_tokens_seen": 202234064,
      "step": 3610
    },
    {
      "epoch": 8.042316258351892,
      "grad_norm": 22.855134963989258,
      "learning_rate": 1e-06,
      "loss": 0.565,
      "num_input_tokens_seen": 202290848,
      "step": 3611
    },
    {
      "epoch": 8.042316258351892,
      "loss": 0.4523416757583618,
      "loss_ce": 0.0001932748273247853,
      "loss_iou": 0.193359375,
      "loss_num": 0.01300048828125,
      "loss_xval": 0.453125,
      "num_input_tokens_seen": 202290848,
      "step": 3611
    },
    {
      "epoch": 8.044543429844097,
      "grad_norm": 22.91286849975586,
      "learning_rate": 1e-06,
      "loss": 0.7699,
      "num_input_tokens_seen": 202349456,
      "step": 3612
    },
    {
      "epoch": 8.044543429844097,
      "loss": 0.7767907381057739,
      "loss_ce": 0.0001794241979951039,
      "loss_iou": 0.322265625,
      "loss_num": 0.0267333984375,
      "loss_xval": 0.77734375,
      "num_input_tokens_seen": 202349456,
      "step": 3612
    },
    {
      "epoch": 8.046770601336302,
      "grad_norm": 22.00181770324707,
      "learning_rate": 1e-06,
      "loss": 0.5741,
      "num_input_tokens_seen": 202406208,
      "step": 3613
    },
    {
      "epoch": 8.046770601336302,
      "loss": 0.6100356578826904,
      "loss_ce": 0.00017238240980077535,
      "loss_iou": 0.2451171875,
      "loss_num": 0.0238037109375,
      "loss_xval": 0.609375,
      "num_input_tokens_seen": 202406208,
      "step": 3613
    },
    {
      "epoch": 8.048997772828507,
      "grad_norm": 23.352352142333984,
      "learning_rate": 1e-06,
      "loss": 0.4346,
      "num_input_tokens_seen": 202461324,
      "step": 3614
    },
    {
      "epoch": 8.048997772828507,
      "loss": 0.4557119607925415,
      "loss_ce": 0.00014553280198015273,
      "loss_iou": 0.1845703125,
      "loss_num": 0.0172119140625,
      "loss_xval": 0.455078125,
      "num_input_tokens_seen": 202461324,
      "step": 3614
    },
    {
      "epoch": 8.051224944320714,
      "grad_norm": 24.288408279418945,
      "learning_rate": 1e-06,
      "loss": 0.5669,
      "num_input_tokens_seen": 202520016,
      "step": 3615
    },
    {
      "epoch": 8.051224944320714,
      "loss": 0.8215472102165222,
      "loss_ce": 0.0006243445095606148,
      "loss_iou": 0.33203125,
      "loss_num": 0.0311279296875,
      "loss_xval": 0.8203125,
      "num_input_tokens_seen": 202520016,
      "step": 3615
    },
    {
      "epoch": 8.053452115812918,
      "grad_norm": 18.481889724731445,
      "learning_rate": 1e-06,
      "loss": 0.44,
      "num_input_tokens_seen": 202576916,
      "step": 3616
    },
    {
      "epoch": 8.053452115812918,
      "loss": 0.38955333828926086,
      "loss_ce": 0.0001490543072577566,
      "loss_iou": 0.1708984375,
      "loss_num": 0.0096435546875,
      "loss_xval": 0.388671875,
      "num_input_tokens_seen": 202576916,
      "step": 3616
    },
    {
      "epoch": 8.055679287305123,
      "grad_norm": 23.070777893066406,
      "learning_rate": 1e-06,
      "loss": 0.6434,
      "num_input_tokens_seen": 202633128,
      "step": 3617
    },
    {
      "epoch": 8.055679287305123,
      "loss": 0.6288352608680725,
      "loss_ce": 0.00017316042794845998,
      "loss_iou": 0.267578125,
      "loss_num": 0.0186767578125,
      "loss_xval": 0.62890625,
      "num_input_tokens_seen": 202633128,
      "step": 3617
    },
    {
      "epoch": 8.057906458797328,
      "grad_norm": 21.540128707885742,
      "learning_rate": 1e-06,
      "loss": 0.5371,
      "num_input_tokens_seen": 202686364,
      "step": 3618
    },
    {
      "epoch": 8.057906458797328,
      "loss": 0.5090727210044861,
      "loss_ce": 0.0001615592191228643,
      "loss_iou": 0.224609375,
      "loss_num": 0.01177978515625,
      "loss_xval": 0.5078125,
      "num_input_tokens_seen": 202686364,
      "step": 3618
    },
    {
      "epoch": 8.060133630289533,
      "grad_norm": 29.778921127319336,
      "learning_rate": 1e-06,
      "loss": 0.8185,
      "num_input_tokens_seen": 202740904,
      "step": 3619
    },
    {
      "epoch": 8.060133630289533,
      "loss": 0.6977142691612244,
      "loss_ce": 0.000204495110665448,
      "loss_iou": 0.30078125,
      "loss_num": 0.0194091796875,
      "loss_xval": 0.69921875,
      "num_input_tokens_seen": 202740904,
      "step": 3619
    },
    {
      "epoch": 8.062360801781738,
      "grad_norm": 15.424278259277344,
      "learning_rate": 1e-06,
      "loss": 0.5548,
      "num_input_tokens_seen": 202796944,
      "step": 3620
    },
    {
      "epoch": 8.062360801781738,
      "loss": 0.6742639541625977,
      "loss_ce": 0.00019166519632562995,
      "loss_iou": 0.279296875,
      "loss_num": 0.023193359375,
      "loss_xval": 0.67578125,
      "num_input_tokens_seen": 202796944,
      "step": 3620
    },
    {
      "epoch": 8.064587973273943,
      "grad_norm": 15.881271362304688,
      "learning_rate": 1e-06,
      "loss": 0.6079,
      "num_input_tokens_seen": 202852488,
      "step": 3621
    },
    {
      "epoch": 8.064587973273943,
      "loss": 0.6368354558944702,
      "loss_ce": 0.00017772393766790628,
      "loss_iou": 0.29296875,
      "loss_num": 0.01031494140625,
      "loss_xval": 0.63671875,
      "num_input_tokens_seen": 202852488,
      "step": 3621
    },
    {
      "epoch": 8.066815144766148,
      "grad_norm": 20.29325294494629,
      "learning_rate": 1e-06,
      "loss": 0.7275,
      "num_input_tokens_seen": 202905252,
      "step": 3622
    },
    {
      "epoch": 8.066815144766148,
      "loss": 0.5565529465675354,
      "loss_ce": 0.00015647773398086429,
      "loss_iou": 0.234375,
      "loss_num": 0.0174560546875,
      "loss_xval": 0.5546875,
      "num_input_tokens_seen": 202905252,
      "step": 3622
    },
    {
      "epoch": 8.069042316258352,
      "grad_norm": 18.05845069885254,
      "learning_rate": 1e-06,
      "loss": 0.5679,
      "num_input_tokens_seen": 202962516,
      "step": 3623
    },
    {
      "epoch": 8.069042316258352,
      "loss": 0.5240879058837891,
      "loss_ce": 0.0001620947732590139,
      "loss_iou": 0.2314453125,
      "loss_num": 0.01239013671875,
      "loss_xval": 0.5234375,
      "num_input_tokens_seen": 202962516,
      "step": 3623
    },
    {
      "epoch": 8.071269487750557,
      "grad_norm": 16.5575008392334,
      "learning_rate": 1e-06,
      "loss": 0.3642,
      "num_input_tokens_seen": 203018252,
      "step": 3624
    },
    {
      "epoch": 8.071269487750557,
      "loss": 0.23819105327129364,
      "loss_ce": 0.00015395878290291876,
      "loss_iou": 0.10693359375,
      "loss_num": 0.00482177734375,
      "loss_xval": 0.23828125,
      "num_input_tokens_seen": 203018252,
      "step": 3624
    },
    {
      "epoch": 8.073496659242762,
      "grad_norm": 14.785902976989746,
      "learning_rate": 1e-06,
      "loss": 0.7054,
      "num_input_tokens_seen": 203075692,
      "step": 3625
    },
    {
      "epoch": 8.073496659242762,
      "loss": 0.8658159971237183,
      "loss_ce": 0.0003374355146661401,
      "loss_iou": 0.34765625,
      "loss_num": 0.0341796875,
      "loss_xval": 0.8671875,
      "num_input_tokens_seen": 203075692,
      "step": 3625
    },
    {
      "epoch": 8.075723830734967,
      "grad_norm": 30.066858291625977,
      "learning_rate": 1e-06,
      "loss": 0.6224,
      "num_input_tokens_seen": 203132688,
      "step": 3626
    },
    {
      "epoch": 8.075723830734967,
      "loss": 0.5541576743125916,
      "loss_ce": 0.00020257396681699902,
      "loss_iou": 0.21875,
      "loss_num": 0.0233154296875,
      "loss_xval": 0.5546875,
      "num_input_tokens_seen": 203132688,
      "step": 3626
    },
    {
      "epoch": 8.077951002227172,
      "grad_norm": 13.862289428710938,
      "learning_rate": 1e-06,
      "loss": 0.6046,
      "num_input_tokens_seen": 203189732,
      "step": 3627
    },
    {
      "epoch": 8.077951002227172,
      "loss": 0.6278433203697205,
      "loss_ce": 0.00015777646331116557,
      "loss_iou": 0.2578125,
      "loss_num": 0.0220947265625,
      "loss_xval": 0.62890625,
      "num_input_tokens_seen": 203189732,
      "step": 3627
    },
    {
      "epoch": 8.080178173719377,
      "grad_norm": 14.706859588623047,
      "learning_rate": 1e-06,
      "loss": 0.5169,
      "num_input_tokens_seen": 203247688,
      "step": 3628
    },
    {
      "epoch": 8.080178173719377,
      "loss": 0.48699110746383667,
      "loss_ce": 0.00017472410399932414,
      "loss_iou": 0.220703125,
      "loss_num": 0.00909423828125,
      "loss_xval": 0.486328125,
      "num_input_tokens_seen": 203247688,
      "step": 3628
    },
    {
      "epoch": 8.082405345211582,
      "grad_norm": 15.710479736328125,
      "learning_rate": 1e-06,
      "loss": 0.661,
      "num_input_tokens_seen": 203304544,
      "step": 3629
    },
    {
      "epoch": 8.082405345211582,
      "loss": 0.6681832075119019,
      "loss_ce": 0.00021445300080813468,
      "loss_iou": 0.271484375,
      "loss_num": 0.0247802734375,
      "loss_xval": 0.66796875,
      "num_input_tokens_seen": 203304544,
      "step": 3629
    },
    {
      "epoch": 8.084632516703786,
      "grad_norm": 64.82049560546875,
      "learning_rate": 1e-06,
      "loss": 0.6353,
      "num_input_tokens_seen": 203358916,
      "step": 3630
    },
    {
      "epoch": 8.084632516703786,
      "loss": 0.7306484580039978,
      "loss_ce": 0.00024074350949376822,
      "loss_iou": 0.314453125,
      "loss_num": 0.0203857421875,
      "loss_xval": 0.73046875,
      "num_input_tokens_seen": 203358916,
      "step": 3630
    },
    {
      "epoch": 8.086859688195991,
      "grad_norm": 17.783905029296875,
      "learning_rate": 1e-06,
      "loss": 0.558,
      "num_input_tokens_seen": 203414864,
      "step": 3631
    },
    {
      "epoch": 8.086859688195991,
      "loss": 0.6720456480979919,
      "loss_ce": 0.00017064949497580528,
      "loss_iou": 0.298828125,
      "loss_num": 0.01458740234375,
      "loss_xval": 0.671875,
      "num_input_tokens_seen": 203414864,
      "step": 3631
    },
    {
      "epoch": 8.089086859688196,
      "grad_norm": 13.846236228942871,
      "learning_rate": 1e-06,
      "loss": 0.5735,
      "num_input_tokens_seen": 203469972,
      "step": 3632
    },
    {
      "epoch": 8.089086859688196,
      "loss": 0.61847984790802,
      "loss_ce": 0.00019372851238586009,
      "loss_iou": 0.27734375,
      "loss_num": 0.01312255859375,
      "loss_xval": 0.6171875,
      "num_input_tokens_seen": 203469972,
      "step": 3632
    },
    {
      "epoch": 8.091314031180401,
      "grad_norm": 49.66973876953125,
      "learning_rate": 1e-06,
      "loss": 0.7761,
      "num_input_tokens_seen": 203528200,
      "step": 3633
    },
    {
      "epoch": 8.091314031180401,
      "loss": 0.869318962097168,
      "loss_ce": 0.00017833446327131242,
      "loss_iou": 0.365234375,
      "loss_num": 0.0274658203125,
      "loss_xval": 0.8671875,
      "num_input_tokens_seen": 203528200,
      "step": 3633
    },
    {
      "epoch": 8.093541202672606,
      "grad_norm": 23.811777114868164,
      "learning_rate": 1e-06,
      "loss": 0.4735,
      "num_input_tokens_seen": 203584436,
      "step": 3634
    },
    {
      "epoch": 8.093541202672606,
      "loss": 0.4863722026348114,
      "loss_ce": 0.00016612093895673752,
      "loss_iou": 0.21875,
      "loss_num": 0.00982666015625,
      "loss_xval": 0.486328125,
      "num_input_tokens_seen": 203584436,
      "step": 3634
    },
    {
      "epoch": 8.09576837416481,
      "grad_norm": 15.80212116241455,
      "learning_rate": 1e-06,
      "loss": 0.5608,
      "num_input_tokens_seen": 203641428,
      "step": 3635
    },
    {
      "epoch": 8.09576837416481,
      "loss": 0.48964816331863403,
      "loss_ce": 0.00014622273738496006,
      "loss_iou": 0.2099609375,
      "loss_num": 0.0140380859375,
      "loss_xval": 0.490234375,
      "num_input_tokens_seen": 203641428,
      "step": 3635
    },
    {
      "epoch": 8.097995545657016,
      "grad_norm": 24.546077728271484,
      "learning_rate": 1e-06,
      "loss": 0.6726,
      "num_input_tokens_seen": 203693980,
      "step": 3636
    },
    {
      "epoch": 8.097995545657016,
      "loss": 0.569033682346344,
      "loss_ce": 0.0001860179763752967,
      "loss_iou": 0.2578125,
      "loss_num": 0.01104736328125,
      "loss_xval": 0.5703125,
      "num_input_tokens_seen": 203693980,
      "step": 3636
    },
    {
      "epoch": 8.10022271714922,
      "grad_norm": 45.66901397705078,
      "learning_rate": 1e-06,
      "loss": 0.8868,
      "num_input_tokens_seen": 203747768,
      "step": 3637
    },
    {
      "epoch": 8.10022271714922,
      "loss": 0.6677889227867126,
      "loss_ce": 0.00018637420726008713,
      "loss_iou": 0.291015625,
      "loss_num": 0.016845703125,
      "loss_xval": 0.66796875,
      "num_input_tokens_seen": 203747768,
      "step": 3637
    },
    {
      "epoch": 8.102449888641425,
      "grad_norm": 25.512130737304688,
      "learning_rate": 1e-06,
      "loss": 0.5878,
      "num_input_tokens_seen": 203805264,
      "step": 3638
    },
    {
      "epoch": 8.102449888641425,
      "loss": 0.47341108322143555,
      "loss_ce": 0.00014449466834776103,
      "loss_iou": 0.201171875,
      "loss_num": 0.01422119140625,
      "loss_xval": 0.47265625,
      "num_input_tokens_seen": 203805264,
      "step": 3638
    },
    {
      "epoch": 8.10467706013363,
      "grad_norm": 16.746042251586914,
      "learning_rate": 1e-06,
      "loss": 0.5147,
      "num_input_tokens_seen": 203862852,
      "step": 3639
    },
    {
      "epoch": 8.10467706013363,
      "loss": 0.5026716589927673,
      "loss_ce": 0.0003523434279486537,
      "loss_iou": 0.2060546875,
      "loss_num": 0.0179443359375,
      "loss_xval": 0.50390625,
      "num_input_tokens_seen": 203862852,
      "step": 3639
    },
    {
      "epoch": 8.106904231625835,
      "grad_norm": 14.684697151184082,
      "learning_rate": 1e-06,
      "loss": 0.6076,
      "num_input_tokens_seen": 203920432,
      "step": 3640
    },
    {
      "epoch": 8.106904231625835,
      "loss": 0.7979079484939575,
      "loss_ce": 0.00017849741561803967,
      "loss_iou": 0.30859375,
      "loss_num": 0.035888671875,
      "loss_xval": 0.796875,
      "num_input_tokens_seen": 203920432,
      "step": 3640
    },
    {
      "epoch": 8.10913140311804,
      "grad_norm": 29.783206939697266,
      "learning_rate": 1e-06,
      "loss": 0.6487,
      "num_input_tokens_seen": 203977000,
      "step": 3641
    },
    {
      "epoch": 8.10913140311804,
      "loss": 0.7514079809188843,
      "loss_ce": 0.00043139857007190585,
      "loss_iou": 0.34375,
      "loss_num": 0.012451171875,
      "loss_xval": 0.75,
      "num_input_tokens_seen": 203977000,
      "step": 3641
    },
    {
      "epoch": 8.111358574610245,
      "grad_norm": 17.605884552001953,
      "learning_rate": 1e-06,
      "loss": 0.5144,
      "num_input_tokens_seen": 204034360,
      "step": 3642
    },
    {
      "epoch": 8.111358574610245,
      "loss": 0.6749836206436157,
      "loss_ce": 0.00017892984033096582,
      "loss_iou": 0.30859375,
      "loss_num": 0.01153564453125,
      "loss_xval": 0.67578125,
      "num_input_tokens_seen": 204034360,
      "step": 3642
    },
    {
      "epoch": 8.11358574610245,
      "grad_norm": 17.724246978759766,
      "learning_rate": 1e-06,
      "loss": 0.6901,
      "num_input_tokens_seen": 204093164,
      "step": 3643
    },
    {
      "epoch": 8.11358574610245,
      "loss": 0.8570829033851624,
      "loss_ce": 0.00033243943471461535,
      "loss_iou": 0.365234375,
      "loss_num": 0.0250244140625,
      "loss_xval": 0.85546875,
      "num_input_tokens_seen": 204093164,
      "step": 3643
    },
    {
      "epoch": 8.115812917594655,
      "grad_norm": 23.620380401611328,
      "learning_rate": 1e-06,
      "loss": 0.6023,
      "num_input_tokens_seen": 204147504,
      "step": 3644
    },
    {
      "epoch": 8.115812917594655,
      "loss": 0.5590322017669678,
      "loss_ce": 0.0001943043462233618,
      "loss_iou": 0.25,
      "loss_num": 0.0113525390625,
      "loss_xval": 0.55859375,
      "num_input_tokens_seen": 204147504,
      "step": 3644
    },
    {
      "epoch": 8.11804008908686,
      "grad_norm": 20.3369140625,
      "learning_rate": 1e-06,
      "loss": 0.5411,
      "num_input_tokens_seen": 204204044,
      "step": 3645
    },
    {
      "epoch": 8.11804008908686,
      "loss": 0.5622153878211975,
      "loss_ce": 0.00020367707475088537,
      "loss_iou": 0.22265625,
      "loss_num": 0.0233154296875,
      "loss_xval": 0.5625,
      "num_input_tokens_seen": 204204044,
      "step": 3645
    },
    {
      "epoch": 8.120267260579064,
      "grad_norm": 20.101959228515625,
      "learning_rate": 1e-06,
      "loss": 0.5644,
      "num_input_tokens_seen": 204262612,
      "step": 3646
    },
    {
      "epoch": 8.120267260579064,
      "loss": 0.5232473611831665,
      "loss_ce": 0.0001760848390404135,
      "loss_iou": 0.228515625,
      "loss_num": 0.01318359375,
      "loss_xval": 0.5234375,
      "num_input_tokens_seen": 204262612,
      "step": 3646
    },
    {
      "epoch": 8.122494432071269,
      "grad_norm": 40.90657043457031,
      "learning_rate": 1e-06,
      "loss": 0.6132,
      "num_input_tokens_seen": 204318668,
      "step": 3647
    },
    {
      "epoch": 8.122494432071269,
      "loss": 0.5699893832206726,
      "loss_ce": 0.00016517913900315762,
      "loss_iou": 0.2294921875,
      "loss_num": 0.0220947265625,
      "loss_xval": 0.5703125,
      "num_input_tokens_seen": 204318668,
      "step": 3647
    },
    {
      "epoch": 8.124721603563474,
      "grad_norm": 22.63620948791504,
      "learning_rate": 1e-06,
      "loss": 0.5056,
      "num_input_tokens_seen": 204376132,
      "step": 3648
    },
    {
      "epoch": 8.124721603563474,
      "loss": 0.478777140378952,
      "loss_ce": 0.00013943444355390966,
      "loss_iou": 0.2158203125,
      "loss_num": 0.00921630859375,
      "loss_xval": 0.478515625,
      "num_input_tokens_seen": 204376132,
      "step": 3648
    },
    {
      "epoch": 8.126948775055679,
      "grad_norm": 13.594147682189941,
      "learning_rate": 1e-06,
      "loss": 0.4997,
      "num_input_tokens_seen": 204435292,
      "step": 3649
    },
    {
      "epoch": 8.126948775055679,
      "loss": 0.5463213920593262,
      "loss_ce": 0.00017879356164485216,
      "loss_iou": 0.2578125,
      "loss_num": 0.006378173828125,
      "loss_xval": 0.546875,
      "num_input_tokens_seen": 204435292,
      "step": 3649
    },
    {
      "epoch": 8.129175946547884,
      "grad_norm": 22.09722137451172,
      "learning_rate": 1e-06,
      "loss": 0.6183,
      "num_input_tokens_seen": 204491100,
      "step": 3650
    },
    {
      "epoch": 8.129175946547884,
      "loss": 0.5094276666641235,
      "loss_ce": 0.00015031076327431947,
      "loss_iou": 0.208984375,
      "loss_num": 0.0181884765625,
      "loss_xval": 0.5078125,
      "num_input_tokens_seen": 204491100,
      "step": 3650
    },
    {
      "epoch": 8.131403118040089,
      "grad_norm": 19.736661911010742,
      "learning_rate": 1e-06,
      "loss": 0.6618,
      "num_input_tokens_seen": 204546644,
      "step": 3651
    },
    {
      "epoch": 8.131403118040089,
      "loss": 0.7262549996376038,
      "loss_ce": 0.00018080734298564494,
      "loss_iou": 0.298828125,
      "loss_num": 0.0260009765625,
      "loss_xval": 0.7265625,
      "num_input_tokens_seen": 204546644,
      "step": 3651
    },
    {
      "epoch": 8.133630289532293,
      "grad_norm": 46.30784606933594,
      "learning_rate": 1e-06,
      "loss": 0.78,
      "num_input_tokens_seen": 204601144,
      "step": 3652
    },
    {
      "epoch": 8.133630289532293,
      "loss": 0.8595629930496216,
      "loss_ce": 0.00018801141413860023,
      "loss_iou": 0.333984375,
      "loss_num": 0.0380859375,
      "loss_xval": 0.859375,
      "num_input_tokens_seen": 204601144,
      "step": 3652
    },
    {
      "epoch": 8.135857461024498,
      "grad_norm": 19.52750587463379,
      "learning_rate": 1e-06,
      "loss": 0.7199,
      "num_input_tokens_seen": 204657164,
      "step": 3653
    },
    {
      "epoch": 8.135857461024498,
      "loss": 0.9219685196876526,
      "loss_ce": 0.0002156377595383674,
      "loss_iou": 0.38671875,
      "loss_num": 0.029296875,
      "loss_xval": 0.921875,
      "num_input_tokens_seen": 204657164,
      "step": 3653
    },
    {
      "epoch": 8.138084632516703,
      "grad_norm": 16.912254333496094,
      "learning_rate": 1e-06,
      "loss": 0.7982,
      "num_input_tokens_seen": 204714952,
      "step": 3654
    },
    {
      "epoch": 8.138084632516703,
      "loss": 0.9020793437957764,
      "loss_ce": 0.00022387836361303926,
      "loss_iou": 0.3828125,
      "loss_num": 0.0272216796875,
      "loss_xval": 0.90234375,
      "num_input_tokens_seen": 204714952,
      "step": 3654
    },
    {
      "epoch": 8.140311804008908,
      "grad_norm": 16.71441650390625,
      "learning_rate": 1e-06,
      "loss": 0.4875,
      "num_input_tokens_seen": 204768916,
      "step": 3655
    },
    {
      "epoch": 8.140311804008908,
      "loss": 0.5125894546508789,
      "loss_ce": 0.00013829523231834173,
      "loss_iou": 0.19921875,
      "loss_num": 0.0228271484375,
      "loss_xval": 0.51171875,
      "num_input_tokens_seen": 204768916,
      "step": 3655
    },
    {
      "epoch": 8.142538975501113,
      "grad_norm": 20.34796714782715,
      "learning_rate": 1e-06,
      "loss": 0.6757,
      "num_input_tokens_seen": 204825592,
      "step": 3656
    },
    {
      "epoch": 8.142538975501113,
      "loss": 0.7248254418373108,
      "loss_ce": 0.00021602565539069474,
      "loss_iou": 0.28515625,
      "loss_num": 0.031005859375,
      "loss_xval": 0.7265625,
      "num_input_tokens_seen": 204825592,
      "step": 3656
    },
    {
      "epoch": 8.144766146993318,
      "grad_norm": 29.74313735961914,
      "learning_rate": 1e-06,
      "loss": 0.7219,
      "num_input_tokens_seen": 204878420,
      "step": 3657
    },
    {
      "epoch": 8.144766146993318,
      "loss": 0.5412226319313049,
      "loss_ce": 0.0002070144983008504,
      "loss_iou": 0.2392578125,
      "loss_num": 0.01251220703125,
      "loss_xval": 0.5390625,
      "num_input_tokens_seen": 204878420,
      "step": 3657
    },
    {
      "epoch": 8.146993318485523,
      "grad_norm": 17.01011085510254,
      "learning_rate": 1e-06,
      "loss": 0.8195,
      "num_input_tokens_seen": 204933092,
      "step": 3658
    },
    {
      "epoch": 8.146993318485523,
      "loss": 0.9253840446472168,
      "loss_ce": 0.0003352175117470324,
      "loss_iou": 0.38671875,
      "loss_num": 0.030029296875,
      "loss_xval": 0.92578125,
      "num_input_tokens_seen": 204933092,
      "step": 3658
    },
    {
      "epoch": 8.14922048997773,
      "grad_norm": 44.65693664550781,
      "learning_rate": 1e-06,
      "loss": 0.7511,
      "num_input_tokens_seen": 204988252,
      "step": 3659
    },
    {
      "epoch": 8.14922048997773,
      "loss": 0.5185383558273315,
      "loss_ce": 0.00016674870857968926,
      "loss_iou": 0.234375,
      "loss_num": 0.0098876953125,
      "loss_xval": 0.51953125,
      "num_input_tokens_seen": 204988252,
      "step": 3659
    },
    {
      "epoch": 8.151447661469934,
      "grad_norm": 12.71816635131836,
      "learning_rate": 1e-06,
      "loss": 0.4983,
      "num_input_tokens_seen": 205044848,
      "step": 3660
    },
    {
      "epoch": 8.151447661469934,
      "loss": 0.5829252600669861,
      "loss_ce": 0.0003446881892159581,
      "loss_iou": 0.2578125,
      "loss_num": 0.0130615234375,
      "loss_xval": 0.58203125,
      "num_input_tokens_seen": 205044848,
      "step": 3660
    },
    {
      "epoch": 8.153674832962139,
      "grad_norm": 16.42379379272461,
      "learning_rate": 1e-06,
      "loss": 0.6397,
      "num_input_tokens_seen": 205101712,
      "step": 3661
    },
    {
      "epoch": 8.153674832962139,
      "loss": 0.8078476190567017,
      "loss_ce": 0.00023041786334943026,
      "loss_iou": 0.34375,
      "loss_num": 0.0240478515625,
      "loss_xval": 0.80859375,
      "num_input_tokens_seen": 205101712,
      "step": 3661
    },
    {
      "epoch": 8.155902004454344,
      "grad_norm": 17.691226959228516,
      "learning_rate": 1e-06,
      "loss": 0.5595,
      "num_input_tokens_seen": 205158256,
      "step": 3662
    },
    {
      "epoch": 8.155902004454344,
      "loss": 0.5641229152679443,
      "loss_ce": 0.00015803641872480512,
      "loss_iou": 0.2578125,
      "loss_num": 0.0093994140625,
      "loss_xval": 0.5625,
      "num_input_tokens_seen": 205158256,
      "step": 3662
    },
    {
      "epoch": 8.158129175946549,
      "grad_norm": 27.215301513671875,
      "learning_rate": 1e-06,
      "loss": 0.738,
      "num_input_tokens_seen": 205215004,
      "step": 3663
    },
    {
      "epoch": 8.158129175946549,
      "loss": 0.6344413757324219,
      "loss_ce": 0.0002250707766506821,
      "loss_iou": 0.24609375,
      "loss_num": 0.028564453125,
      "loss_xval": 0.6328125,
      "num_input_tokens_seen": 205215004,
      "step": 3663
    },
    {
      "epoch": 8.160356347438753,
      "grad_norm": 27.16131591796875,
      "learning_rate": 1e-06,
      "loss": 0.5939,
      "num_input_tokens_seen": 205269476,
      "step": 3664
    },
    {
      "epoch": 8.160356347438753,
      "loss": 0.729198694229126,
      "loss_ce": 0.00019480480113998055,
      "loss_iou": 0.306640625,
      "loss_num": 0.022705078125,
      "loss_xval": 0.73046875,
      "num_input_tokens_seen": 205269476,
      "step": 3664
    },
    {
      "epoch": 8.162583518930958,
      "grad_norm": 14.612595558166504,
      "learning_rate": 1e-06,
      "loss": 0.7695,
      "num_input_tokens_seen": 205321936,
      "step": 3665
    },
    {
      "epoch": 8.162583518930958,
      "loss": 0.7426295280456543,
      "loss_ce": 0.00019791701924987137,
      "loss_iou": 0.3125,
      "loss_num": 0.02392578125,
      "loss_xval": 0.7421875,
      "num_input_tokens_seen": 205321936,
      "step": 3665
    },
    {
      "epoch": 8.164810690423163,
      "grad_norm": 23.188566207885742,
      "learning_rate": 1e-06,
      "loss": 0.9123,
      "num_input_tokens_seen": 205373464,
      "step": 3666
    },
    {
      "epoch": 8.164810690423163,
      "loss": 1.0600320100784302,
      "loss_ce": 0.0003396637039259076,
      "loss_iou": 0.455078125,
      "loss_num": 0.0301513671875,
      "loss_xval": 1.0625,
      "num_input_tokens_seen": 205373464,
      "step": 3666
    },
    {
      "epoch": 8.167037861915368,
      "grad_norm": 17.917818069458008,
      "learning_rate": 1e-06,
      "loss": 0.6325,
      "num_input_tokens_seen": 205424324,
      "step": 3667
    },
    {
      "epoch": 8.167037861915368,
      "loss": 0.43619126081466675,
      "loss_ce": 0.00015609552792739123,
      "loss_iou": 0.189453125,
      "loss_num": 0.011474609375,
      "loss_xval": 0.435546875,
      "num_input_tokens_seen": 205424324,
      "step": 3667
    },
    {
      "epoch": 8.169265033407573,
      "grad_norm": 22.07512092590332,
      "learning_rate": 1e-06,
      "loss": 1.011,
      "num_input_tokens_seen": 205479044,
      "step": 3668
    },
    {
      "epoch": 8.169265033407573,
      "loss": 1.0271341800689697,
      "loss_ce": 0.00027870899066329,
      "loss_iou": 0.3984375,
      "loss_num": 0.0458984375,
      "loss_xval": 1.0234375,
      "num_input_tokens_seen": 205479044,
      "step": 3668
    },
    {
      "epoch": 8.171492204899778,
      "grad_norm": 47.829078674316406,
      "learning_rate": 1e-06,
      "loss": 0.4878,
      "num_input_tokens_seen": 205533932,
      "step": 3669
    },
    {
      "epoch": 8.171492204899778,
      "loss": 0.40675288438796997,
      "loss_ce": 0.0001671954378252849,
      "loss_iou": 0.18359375,
      "loss_num": 0.007720947265625,
      "loss_xval": 0.40625,
      "num_input_tokens_seen": 205533932,
      "step": 3669
    },
    {
      "epoch": 8.173719376391983,
      "grad_norm": 18.006383895874023,
      "learning_rate": 1e-06,
      "loss": 0.4668,
      "num_input_tokens_seen": 205589416,
      "step": 3670
    },
    {
      "epoch": 8.173719376391983,
      "loss": 0.3985961675643921,
      "loss_ce": 0.00015864020679146051,
      "loss_iou": 0.17578125,
      "loss_num": 0.00921630859375,
      "loss_xval": 0.3984375,
      "num_input_tokens_seen": 205589416,
      "step": 3670
    },
    {
      "epoch": 8.175946547884188,
      "grad_norm": 27.27997589111328,
      "learning_rate": 1e-06,
      "loss": 0.5638,
      "num_input_tokens_seen": 205643196,
      "step": 3671
    },
    {
      "epoch": 8.175946547884188,
      "loss": 0.5613173246383667,
      "loss_ce": 0.00016013637650758028,
      "loss_iou": 0.2216796875,
      "loss_num": 0.0234375,
      "loss_xval": 0.5625,
      "num_input_tokens_seen": 205643196,
      "step": 3671
    },
    {
      "epoch": 8.178173719376392,
      "grad_norm": 19.659027099609375,
      "learning_rate": 1e-06,
      "loss": 0.7405,
      "num_input_tokens_seen": 205698748,
      "step": 3672
    },
    {
      "epoch": 8.178173719376392,
      "loss": 0.9643779397010803,
      "loss_ce": 0.0002666152431629598,
      "loss_iou": 0.369140625,
      "loss_num": 0.044921875,
      "loss_xval": 0.96484375,
      "num_input_tokens_seen": 205698748,
      "step": 3672
    },
    {
      "epoch": 8.180400890868597,
      "grad_norm": 25.99626922607422,
      "learning_rate": 1e-06,
      "loss": 0.6958,
      "num_input_tokens_seen": 205753176,
      "step": 3673
    },
    {
      "epoch": 8.180400890868597,
      "loss": 0.4478331208229065,
      "loss_ce": 0.00020127877360209823,
      "loss_iou": 0.203125,
      "loss_num": 0.0084228515625,
      "loss_xval": 0.447265625,
      "num_input_tokens_seen": 205753176,
      "step": 3673
    },
    {
      "epoch": 8.182628062360802,
      "grad_norm": 15.863686561584473,
      "learning_rate": 1e-06,
      "loss": 0.5438,
      "num_input_tokens_seen": 205807732,
      "step": 3674
    },
    {
      "epoch": 8.182628062360802,
      "loss": 0.5248715877532959,
      "loss_ce": 0.00015238260675687343,
      "loss_iou": 0.2353515625,
      "loss_num": 0.0107421875,
      "loss_xval": 0.5234375,
      "num_input_tokens_seen": 205807732,
      "step": 3674
    },
    {
      "epoch": 8.184855233853007,
      "grad_norm": 21.04659652709961,
      "learning_rate": 1e-06,
      "loss": 0.6401,
      "num_input_tokens_seen": 205860244,
      "step": 3675
    },
    {
      "epoch": 8.184855233853007,
      "loss": 0.5119701623916626,
      "loss_ce": 0.0001293848908971995,
      "loss_iou": 0.2177734375,
      "loss_num": 0.01519775390625,
      "loss_xval": 0.51171875,
      "num_input_tokens_seen": 205860244,
      "step": 3675
    },
    {
      "epoch": 8.187082405345212,
      "grad_norm": 15.499350547790527,
      "learning_rate": 1e-06,
      "loss": 0.6712,
      "num_input_tokens_seen": 205918972,
      "step": 3676
    },
    {
      "epoch": 8.187082405345212,
      "loss": 0.7695894241333008,
      "loss_ce": 0.00018028570048045367,
      "loss_iou": 0.302734375,
      "loss_num": 0.033203125,
      "loss_xval": 0.76953125,
      "num_input_tokens_seen": 205918972,
      "step": 3676
    },
    {
      "epoch": 8.189309576837417,
      "grad_norm": 19.32471466064453,
      "learning_rate": 1e-06,
      "loss": 0.592,
      "num_input_tokens_seen": 205975060,
      "step": 3677
    },
    {
      "epoch": 8.189309576837417,
      "loss": 0.6343322992324829,
      "loss_ce": 0.00017700789612717927,
      "loss_iou": 0.283203125,
      "loss_num": 0.01336669921875,
      "loss_xval": 0.6328125,
      "num_input_tokens_seen": 205975060,
      "step": 3677
    },
    {
      "epoch": 8.191536748329622,
      "grad_norm": 18.84845542907715,
      "learning_rate": 1e-06,
      "loss": 0.69,
      "num_input_tokens_seen": 206030576,
      "step": 3678
    },
    {
      "epoch": 8.191536748329622,
      "loss": 0.8888704776763916,
      "loss_ce": 0.00019854953279718757,
      "loss_iou": 0.375,
      "loss_num": 0.02734375,
      "loss_xval": 0.890625,
      "num_input_tokens_seen": 206030576,
      "step": 3678
    },
    {
      "epoch": 8.193763919821826,
      "grad_norm": 16.970340728759766,
      "learning_rate": 1e-06,
      "loss": 0.6626,
      "num_input_tokens_seen": 206087296,
      "step": 3679
    },
    {
      "epoch": 8.193763919821826,
      "loss": 0.6373680830001831,
      "loss_ce": 0.0002831476158462465,
      "loss_iou": 0.28125,
      "loss_num": 0.01470947265625,
      "loss_xval": 0.63671875,
      "num_input_tokens_seen": 206087296,
      "step": 3679
    },
    {
      "epoch": 8.195991091314031,
      "grad_norm": 31.894102096557617,
      "learning_rate": 1e-06,
      "loss": 0.5481,
      "num_input_tokens_seen": 206141220,
      "step": 3680
    },
    {
      "epoch": 8.195991091314031,
      "loss": 0.3562151789665222,
      "loss_ce": 0.0001360624737571925,
      "loss_iou": 0.16015625,
      "loss_num": 0.007049560546875,
      "loss_xval": 0.35546875,
      "num_input_tokens_seen": 206141220,
      "step": 3680
    },
    {
      "epoch": 8.198218262806236,
      "grad_norm": 15.175886154174805,
      "learning_rate": 1e-06,
      "loss": 0.4565,
      "num_input_tokens_seen": 206199092,
      "step": 3681
    },
    {
      "epoch": 8.198218262806236,
      "loss": 0.38468533754348755,
      "loss_ce": 0.00022489193361252546,
      "loss_iou": 0.1669921875,
      "loss_num": 0.01019287109375,
      "loss_xval": 0.384765625,
      "num_input_tokens_seen": 206199092,
      "step": 3681
    },
    {
      "epoch": 8.200445434298441,
      "grad_norm": 14.856992721557617,
      "learning_rate": 1e-06,
      "loss": 0.5933,
      "num_input_tokens_seen": 206255596,
      "step": 3682
    },
    {
      "epoch": 8.200445434298441,
      "loss": 0.6677369475364685,
      "loss_ce": 0.0002565070753917098,
      "loss_iou": 0.28125,
      "loss_num": 0.0211181640625,
      "loss_xval": 0.66796875,
      "num_input_tokens_seen": 206255596,
      "step": 3682
    },
    {
      "epoch": 8.202672605790646,
      "grad_norm": 26.978513717651367,
      "learning_rate": 1e-06,
      "loss": 0.6554,
      "num_input_tokens_seen": 206310404,
      "step": 3683
    },
    {
      "epoch": 8.202672605790646,
      "loss": 0.6696029901504517,
      "loss_ce": 0.0004135652561672032,
      "loss_iou": 0.28125,
      "loss_num": 0.021240234375,
      "loss_xval": 0.66796875,
      "num_input_tokens_seen": 206310404,
      "step": 3683
    },
    {
      "epoch": 8.20489977728285,
      "grad_norm": 15.110715866088867,
      "learning_rate": 1e-06,
      "loss": 0.5443,
      "num_input_tokens_seen": 206364188,
      "step": 3684
    },
    {
      "epoch": 8.20489977728285,
      "loss": 0.4510454833507538,
      "loss_ce": 0.00017877297068480402,
      "loss_iou": 0.1943359375,
      "loss_num": 0.01251220703125,
      "loss_xval": 0.451171875,
      "num_input_tokens_seen": 206364188,
      "step": 3684
    },
    {
      "epoch": 8.207126948775056,
      "grad_norm": 18.641998291015625,
      "learning_rate": 1e-06,
      "loss": 0.563,
      "num_input_tokens_seen": 206419384,
      "step": 3685
    },
    {
      "epoch": 8.207126948775056,
      "loss": 0.5365896821022034,
      "loss_ce": 0.00021273433230817318,
      "loss_iou": 0.2197265625,
      "loss_num": 0.01953125,
      "loss_xval": 0.53515625,
      "num_input_tokens_seen": 206419384,
      "step": 3685
    },
    {
      "epoch": 8.20935412026726,
      "grad_norm": 16.16904640197754,
      "learning_rate": 1e-06,
      "loss": 0.6996,
      "num_input_tokens_seen": 206474860,
      "step": 3686
    },
    {
      "epoch": 8.20935412026726,
      "loss": 0.9597395062446594,
      "loss_ce": 0.00026684050681069493,
      "loss_iou": 0.369140625,
      "loss_num": 0.0439453125,
      "loss_xval": 0.9609375,
      "num_input_tokens_seen": 206474860,
      "step": 3686
    },
    {
      "epoch": 8.211581291759465,
      "grad_norm": 14.339240074157715,
      "learning_rate": 1e-06,
      "loss": 0.5036,
      "num_input_tokens_seen": 206532756,
      "step": 3687
    },
    {
      "epoch": 8.211581291759465,
      "loss": 0.5385290384292603,
      "loss_ce": 0.0001989899465115741,
      "loss_iou": 0.2392578125,
      "loss_num": 0.01202392578125,
      "loss_xval": 0.5390625,
      "num_input_tokens_seen": 206532756,
      "step": 3687
    },
    {
      "epoch": 8.21380846325167,
      "grad_norm": 16.02851104736328,
      "learning_rate": 1e-06,
      "loss": 0.7048,
      "num_input_tokens_seen": 206588744,
      "step": 3688
    },
    {
      "epoch": 8.21380846325167,
      "loss": 0.8307337164878845,
      "loss_ce": 0.000167329068062827,
      "loss_iou": 0.36328125,
      "loss_num": 0.0205078125,
      "loss_xval": 0.83203125,
      "num_input_tokens_seen": 206588744,
      "step": 3688
    },
    {
      "epoch": 8.216035634743875,
      "grad_norm": 20.067358016967773,
      "learning_rate": 1e-06,
      "loss": 0.5148,
      "num_input_tokens_seen": 206645832,
      "step": 3689
    },
    {
      "epoch": 8.216035634743875,
      "loss": 0.4788054823875427,
      "loss_ce": 0.0001677550608292222,
      "loss_iou": 0.2001953125,
      "loss_num": 0.01544189453125,
      "loss_xval": 0.478515625,
      "num_input_tokens_seen": 206645832,
      "step": 3689
    },
    {
      "epoch": 8.21826280623608,
      "grad_norm": 13.61365795135498,
      "learning_rate": 1e-06,
      "loss": 0.4008,
      "num_input_tokens_seen": 206701388,
      "step": 3690
    },
    {
      "epoch": 8.21826280623608,
      "loss": 0.34245729446411133,
      "loss_ce": 0.00017215096158906817,
      "loss_iou": 0.1513671875,
      "loss_num": 0.007781982421875,
      "loss_xval": 0.341796875,
      "num_input_tokens_seen": 206701388,
      "step": 3690
    },
    {
      "epoch": 8.220489977728285,
      "grad_norm": 16.062326431274414,
      "learning_rate": 1e-06,
      "loss": 0.6755,
      "num_input_tokens_seen": 206756396,
      "step": 3691
    },
    {
      "epoch": 8.220489977728285,
      "loss": 0.7819403409957886,
      "loss_ce": 0.00020207473426125944,
      "loss_iou": 0.337890625,
      "loss_num": 0.02099609375,
      "loss_xval": 0.78125,
      "num_input_tokens_seen": 206756396,
      "step": 3691
    },
    {
      "epoch": 8.22271714922049,
      "grad_norm": 16.20405387878418,
      "learning_rate": 1e-06,
      "loss": 0.8364,
      "num_input_tokens_seen": 206813452,
      "step": 3692
    },
    {
      "epoch": 8.22271714922049,
      "loss": 0.7428736686706543,
      "loss_ce": 0.00019789818907156587,
      "loss_iou": 0.34765625,
      "loss_num": 0.00885009765625,
      "loss_xval": 0.7421875,
      "num_input_tokens_seen": 206813452,
      "step": 3692
    },
    {
      "epoch": 8.224944320712694,
      "grad_norm": 15.899574279785156,
      "learning_rate": 1e-06,
      "loss": 0.5448,
      "num_input_tokens_seen": 206871936,
      "step": 3693
    },
    {
      "epoch": 8.224944320712694,
      "loss": 0.4710107445716858,
      "loss_ce": 0.00018556615395937115,
      "loss_iou": 0.2080078125,
      "loss_num": 0.010986328125,
      "loss_xval": 0.470703125,
      "num_input_tokens_seen": 206871936,
      "step": 3693
    },
    {
      "epoch": 8.2271714922049,
      "grad_norm": 21.231388092041016,
      "learning_rate": 1e-06,
      "loss": 0.6426,
      "num_input_tokens_seen": 206927444,
      "step": 3694
    },
    {
      "epoch": 8.2271714922049,
      "loss": 0.717461347579956,
      "loss_ce": 0.00017620844300836325,
      "loss_iou": 0.291015625,
      "loss_num": 0.02685546875,
      "loss_xval": 0.71875,
      "num_input_tokens_seen": 206927444,
      "step": 3694
    },
    {
      "epoch": 8.229398663697104,
      "grad_norm": 14.745061874389648,
      "learning_rate": 1e-06,
      "loss": 0.4019,
      "num_input_tokens_seen": 206982152,
      "step": 3695
    },
    {
      "epoch": 8.229398663697104,
      "loss": 0.32421159744262695,
      "loss_ce": 0.00014541992277372628,
      "loss_iou": 0.1279296875,
      "loss_num": 0.01361083984375,
      "loss_xval": 0.32421875,
      "num_input_tokens_seen": 206982152,
      "step": 3695
    },
    {
      "epoch": 8.231625835189309,
      "grad_norm": 16.86565589904785,
      "learning_rate": 1e-06,
      "loss": 0.577,
      "num_input_tokens_seen": 207036748,
      "step": 3696
    },
    {
      "epoch": 8.231625835189309,
      "loss": 0.5805333852767944,
      "loss_ce": 0.00015002592408563942,
      "loss_iou": 0.23828125,
      "loss_num": 0.020751953125,
      "loss_xval": 0.58203125,
      "num_input_tokens_seen": 207036748,
      "step": 3696
    },
    {
      "epoch": 8.233853006681514,
      "grad_norm": 16.574954986572266,
      "learning_rate": 1e-06,
      "loss": 0.751,
      "num_input_tokens_seen": 207090424,
      "step": 3697
    },
    {
      "epoch": 8.233853006681514,
      "loss": 0.7723691463470459,
      "loss_ce": 0.0001523814134998247,
      "loss_iou": 0.33984375,
      "loss_num": 0.018798828125,
      "loss_xval": 0.7734375,
      "num_input_tokens_seen": 207090424,
      "step": 3697
    },
    {
      "epoch": 8.236080178173719,
      "grad_norm": 32.75465774536133,
      "learning_rate": 1e-06,
      "loss": 0.4883,
      "num_input_tokens_seen": 207146916,
      "step": 3698
    },
    {
      "epoch": 8.236080178173719,
      "loss": 0.5894994735717773,
      "loss_ce": 0.0001440244377590716,
      "loss_iou": 0.263671875,
      "loss_num": 0.0126953125,
      "loss_xval": 0.58984375,
      "num_input_tokens_seen": 207146916,
      "step": 3698
    },
    {
      "epoch": 8.238307349665924,
      "grad_norm": 19.782665252685547,
      "learning_rate": 1e-06,
      "loss": 1.013,
      "num_input_tokens_seen": 207202692,
      "step": 3699
    },
    {
      "epoch": 8.238307349665924,
      "loss": 1.0240821838378906,
      "loss_ce": 0.00015636572788935155,
      "loss_iou": 0.384765625,
      "loss_num": 0.050537109375,
      "loss_xval": 1.0234375,
      "num_input_tokens_seen": 207202692,
      "step": 3699
    },
    {
      "epoch": 8.240534521158128,
      "grad_norm": 25.966341018676758,
      "learning_rate": 1e-06,
      "loss": 0.6403,
      "num_input_tokens_seen": 207253188,
      "step": 3700
    },
    {
      "epoch": 8.240534521158128,
      "loss": 0.602780818939209,
      "loss_ce": 0.00024172097619157284,
      "loss_iou": 0.232421875,
      "loss_num": 0.027587890625,
      "loss_xval": 0.6015625,
      "num_input_tokens_seen": 207253188,
      "step": 3700
    },
    {
      "epoch": 8.242761692650333,
      "grad_norm": 19.4698543548584,
      "learning_rate": 1e-06,
      "loss": 0.565,
      "num_input_tokens_seen": 207307820,
      "step": 3701
    },
    {
      "epoch": 8.242761692650333,
      "loss": 0.6106410026550293,
      "loss_ce": 0.00016741504077799618,
      "loss_iou": 0.2734375,
      "loss_num": 0.0128173828125,
      "loss_xval": 0.609375,
      "num_input_tokens_seen": 207307820,
      "step": 3701
    },
    {
      "epoch": 8.244988864142538,
      "grad_norm": 15.045053482055664,
      "learning_rate": 1e-06,
      "loss": 0.4617,
      "num_input_tokens_seen": 207363340,
      "step": 3702
    },
    {
      "epoch": 8.244988864142538,
      "loss": 0.40211230516433716,
      "loss_ce": 0.00013477080210577697,
      "loss_iou": 0.1767578125,
      "loss_num": 0.0096435546875,
      "loss_xval": 0.40234375,
      "num_input_tokens_seen": 207363340,
      "step": 3702
    },
    {
      "epoch": 8.247216035634743,
      "grad_norm": 15.657466888427734,
      "learning_rate": 1e-06,
      "loss": 0.721,
      "num_input_tokens_seen": 207416744,
      "step": 3703
    },
    {
      "epoch": 8.247216035634743,
      "loss": 0.7376276254653931,
      "loss_ce": 0.00026186800096184015,
      "loss_iou": 0.298828125,
      "loss_num": 0.0281982421875,
      "loss_xval": 0.73828125,
      "num_input_tokens_seen": 207416744,
      "step": 3703
    },
    {
      "epoch": 8.249443207126948,
      "grad_norm": 16.417505264282227,
      "learning_rate": 1e-06,
      "loss": 0.4049,
      "num_input_tokens_seen": 207473136,
      "step": 3704
    },
    {
      "epoch": 8.249443207126948,
      "loss": 0.3186110258102417,
      "loss_ce": 0.00012959179002791643,
      "loss_iou": 0.1259765625,
      "loss_num": 0.01324462890625,
      "loss_xval": 0.318359375,
      "num_input_tokens_seen": 207473136,
      "step": 3704
    },
    {
      "epoch": 8.251670378619155,
      "grad_norm": 21.20426368713379,
      "learning_rate": 1e-06,
      "loss": 0.5388,
      "num_input_tokens_seen": 207528736,
      "step": 3705
    },
    {
      "epoch": 8.251670378619155,
      "loss": 0.6479160189628601,
      "loss_ce": 0.00014987270697019994,
      "loss_iou": 0.275390625,
      "loss_num": 0.01953125,
      "loss_xval": 0.6484375,
      "num_input_tokens_seen": 207528736,
      "step": 3705
    },
    {
      "epoch": 8.25389755011136,
      "grad_norm": 27.81067657470703,
      "learning_rate": 1e-06,
      "loss": 0.6099,
      "num_input_tokens_seen": 207579912,
      "step": 3706
    },
    {
      "epoch": 8.25389755011136,
      "loss": 0.6510103344917297,
      "loss_ce": 0.00019250249897595495,
      "loss_iou": 0.267578125,
      "loss_num": 0.023193359375,
      "loss_xval": 0.65234375,
      "num_input_tokens_seen": 207579912,
      "step": 3706
    },
    {
      "epoch": 8.256124721603564,
      "grad_norm": 24.32451629638672,
      "learning_rate": 1e-06,
      "loss": 0.6658,
      "num_input_tokens_seen": 207635324,
      "step": 3707
    },
    {
      "epoch": 8.256124721603564,
      "loss": 0.6035555601119995,
      "loss_ce": 0.00016204667917918414,
      "loss_iou": 0.25390625,
      "loss_num": 0.018798828125,
      "loss_xval": 0.6015625,
      "num_input_tokens_seen": 207635324,
      "step": 3707
    },
    {
      "epoch": 8.25835189309577,
      "grad_norm": 19.096254348754883,
      "learning_rate": 1e-06,
      "loss": 0.6809,
      "num_input_tokens_seen": 207691984,
      "step": 3708
    },
    {
      "epoch": 8.25835189309577,
      "loss": 0.6975338459014893,
      "loss_ce": 0.0002682044287212193,
      "loss_iou": 0.298828125,
      "loss_num": 0.0201416015625,
      "loss_xval": 0.6953125,
      "num_input_tokens_seen": 207691984,
      "step": 3708
    },
    {
      "epoch": 8.260579064587974,
      "grad_norm": 19.96952247619629,
      "learning_rate": 1e-06,
      "loss": 0.9264,
      "num_input_tokens_seen": 207742476,
      "step": 3709
    },
    {
      "epoch": 8.260579064587974,
      "loss": 0.9751963019371033,
      "loss_ce": 0.00022070904378779233,
      "loss_iou": 0.416015625,
      "loss_num": 0.0283203125,
      "loss_xval": 0.9765625,
      "num_input_tokens_seen": 207742476,
      "step": 3709
    },
    {
      "epoch": 8.262806236080179,
      "grad_norm": 15.585442543029785,
      "learning_rate": 1e-06,
      "loss": 0.4606,
      "num_input_tokens_seen": 207800484,
      "step": 3710
    },
    {
      "epoch": 8.262806236080179,
      "loss": 0.5031048655509949,
      "loss_ce": 0.00017520022811368108,
      "loss_iou": 0.2265625,
      "loss_num": 0.0098876953125,
      "loss_xval": 0.50390625,
      "num_input_tokens_seen": 207800484,
      "step": 3710
    },
    {
      "epoch": 8.265033407572384,
      "grad_norm": 36.97969055175781,
      "learning_rate": 1e-06,
      "loss": 0.9413,
      "num_input_tokens_seen": 207855020,
      "step": 3711
    },
    {
      "epoch": 8.265033407572384,
      "loss": 0.9603962898254395,
      "loss_ce": 0.00019123686070088297,
      "loss_iou": 0.373046875,
      "loss_num": 0.043212890625,
      "loss_xval": 0.9609375,
      "num_input_tokens_seen": 207855020,
      "step": 3711
    },
    {
      "epoch": 8.267260579064589,
      "grad_norm": 15.044690132141113,
      "learning_rate": 1e-06,
      "loss": 0.7137,
      "num_input_tokens_seen": 207911256,
      "step": 3712
    },
    {
      "epoch": 8.267260579064589,
      "loss": 0.6639913320541382,
      "loss_ce": 0.00017301499610766768,
      "loss_iou": 0.279296875,
      "loss_num": 0.0211181640625,
      "loss_xval": 0.6640625,
      "num_input_tokens_seen": 207911256,
      "step": 3712
    },
    {
      "epoch": 8.269487750556793,
      "grad_norm": 19.051246643066406,
      "learning_rate": 1e-06,
      "loss": 0.6152,
      "num_input_tokens_seen": 207968744,
      "step": 3713
    },
    {
      "epoch": 8.269487750556793,
      "loss": 0.5926859378814697,
      "loss_ce": 0.00015662802616134286,
      "loss_iou": 0.259765625,
      "loss_num": 0.0145263671875,
      "loss_xval": 0.59375,
      "num_input_tokens_seen": 207968744,
      "step": 3713
    },
    {
      "epoch": 8.271714922048998,
      "grad_norm": 17.641752243041992,
      "learning_rate": 1e-06,
      "loss": 0.6043,
      "num_input_tokens_seen": 208024108,
      "step": 3714
    },
    {
      "epoch": 8.271714922048998,
      "loss": 0.5748984813690186,
      "loss_ce": 0.0001914296008180827,
      "loss_iou": 0.251953125,
      "loss_num": 0.01397705078125,
      "loss_xval": 0.57421875,
      "num_input_tokens_seen": 208024108,
      "step": 3714
    },
    {
      "epoch": 8.273942093541203,
      "grad_norm": 17.993663787841797,
      "learning_rate": 1e-06,
      "loss": 0.7099,
      "num_input_tokens_seen": 208075840,
      "step": 3715
    },
    {
      "epoch": 8.273942093541203,
      "loss": 0.6750204563140869,
      "loss_ce": 0.0004599187523126602,
      "loss_iou": 0.30078125,
      "loss_num": 0.014892578125,
      "loss_xval": 0.67578125,
      "num_input_tokens_seen": 208075840,
      "step": 3715
    },
    {
      "epoch": 8.276169265033408,
      "grad_norm": 23.302898406982422,
      "learning_rate": 1e-06,
      "loss": 0.4742,
      "num_input_tokens_seen": 208132584,
      "step": 3716
    },
    {
      "epoch": 8.276169265033408,
      "loss": 0.5063949823379517,
      "loss_ce": 0.0001694063248578459,
      "loss_iou": 0.2275390625,
      "loss_num": 0.01031494140625,
      "loss_xval": 0.5078125,
      "num_input_tokens_seen": 208132584,
      "step": 3716
    },
    {
      "epoch": 8.278396436525613,
      "grad_norm": 19.531972885131836,
      "learning_rate": 1e-06,
      "loss": 0.5614,
      "num_input_tokens_seen": 208189256,
      "step": 3717
    },
    {
      "epoch": 8.278396436525613,
      "loss": 0.46560224890708923,
      "loss_ce": 0.0001481281651649624,
      "loss_iou": 0.1953125,
      "loss_num": 0.014892578125,
      "loss_xval": 0.46484375,
      "num_input_tokens_seen": 208189256,
      "step": 3717
    },
    {
      "epoch": 8.280623608017818,
      "grad_norm": 23.255966186523438,
      "learning_rate": 1e-06,
      "loss": 0.6233,
      "num_input_tokens_seen": 208247268,
      "step": 3718
    },
    {
      "epoch": 8.280623608017818,
      "loss": 0.630403995513916,
      "loss_ce": 0.0002770504215732217,
      "loss_iou": 0.263671875,
      "loss_num": 0.020263671875,
      "loss_xval": 0.62890625,
      "num_input_tokens_seen": 208247268,
      "step": 3718
    },
    {
      "epoch": 8.282850779510023,
      "grad_norm": 31.8346004486084,
      "learning_rate": 1e-06,
      "loss": 0.4895,
      "num_input_tokens_seen": 208303824,
      "step": 3719
    },
    {
      "epoch": 8.282850779510023,
      "loss": 0.49452388286590576,
      "loss_ce": 0.00020015303744003177,
      "loss_iou": 0.212890625,
      "loss_num": 0.0140380859375,
      "loss_xval": 0.494140625,
      "num_input_tokens_seen": 208303824,
      "step": 3719
    },
    {
      "epoch": 8.285077951002227,
      "grad_norm": 16.450027465820312,
      "learning_rate": 1e-06,
      "loss": 0.8151,
      "num_input_tokens_seen": 208361756,
      "step": 3720
    },
    {
      "epoch": 8.285077951002227,
      "loss": 0.8614362478256226,
      "loss_ce": 0.00023015934857539833,
      "loss_iou": 0.361328125,
      "loss_num": 0.0274658203125,
      "loss_xval": 0.859375,
      "num_input_tokens_seen": 208361756,
      "step": 3720
    },
    {
      "epoch": 8.287305122494432,
      "grad_norm": 73.26863098144531,
      "learning_rate": 1e-06,
      "loss": 0.8532,
      "num_input_tokens_seen": 208418208,
      "step": 3721
    },
    {
      "epoch": 8.287305122494432,
      "loss": 1.0102782249450684,
      "loss_ce": 0.0002685172366909683,
      "loss_iou": 0.3515625,
      "loss_num": 0.0615234375,
      "loss_xval": 1.0078125,
      "num_input_tokens_seen": 208418208,
      "step": 3721
    },
    {
      "epoch": 8.289532293986637,
      "grad_norm": 14.410170555114746,
      "learning_rate": 1e-06,
      "loss": 0.5911,
      "num_input_tokens_seen": 208475988,
      "step": 3722
    },
    {
      "epoch": 8.289532293986637,
      "loss": 0.736660897731781,
      "loss_ce": 0.00033277933835051954,
      "loss_iou": 0.322265625,
      "loss_num": 0.018310546875,
      "loss_xval": 0.734375,
      "num_input_tokens_seen": 208475988,
      "step": 3722
    },
    {
      "epoch": 8.291759465478842,
      "grad_norm": 17.486080169677734,
      "learning_rate": 1e-06,
      "loss": 0.6389,
      "num_input_tokens_seen": 208532072,
      "step": 3723
    },
    {
      "epoch": 8.291759465478842,
      "loss": 0.7401440143585205,
      "loss_ce": 0.00015374486974906176,
      "loss_iou": 0.333984375,
      "loss_num": 0.01446533203125,
      "loss_xval": 0.73828125,
      "num_input_tokens_seen": 208532072,
      "step": 3723
    },
    {
      "epoch": 8.293986636971047,
      "grad_norm": 27.663610458374023,
      "learning_rate": 1e-06,
      "loss": 0.5406,
      "num_input_tokens_seen": 208587856,
      "step": 3724
    },
    {
      "epoch": 8.293986636971047,
      "loss": 0.46732282638549805,
      "loss_ce": 0.0001597225054865703,
      "loss_iou": 0.2021484375,
      "loss_num": 0.0125732421875,
      "loss_xval": 0.466796875,
      "num_input_tokens_seen": 208587856,
      "step": 3724
    },
    {
      "epoch": 8.296213808463252,
      "grad_norm": 23.040565490722656,
      "learning_rate": 1e-06,
      "loss": 0.7325,
      "num_input_tokens_seen": 208640652,
      "step": 3725
    },
    {
      "epoch": 8.296213808463252,
      "loss": 0.9635574817657471,
      "loss_ce": 0.00017856716294772923,
      "loss_iou": 0.42578125,
      "loss_num": 0.0220947265625,
      "loss_xval": 0.96484375,
      "num_input_tokens_seen": 208640652,
      "step": 3725
    },
    {
      "epoch": 8.298440979955457,
      "grad_norm": 17.724363327026367,
      "learning_rate": 1e-06,
      "loss": 0.4894,
      "num_input_tokens_seen": 208694956,
      "step": 3726
    },
    {
      "epoch": 8.298440979955457,
      "loss": 0.42843109369277954,
      "loss_ce": 0.0002084198349621147,
      "loss_iou": 0.189453125,
      "loss_num": 0.00970458984375,
      "loss_xval": 0.427734375,
      "num_input_tokens_seen": 208694956,
      "step": 3726
    },
    {
      "epoch": 8.300668151447661,
      "grad_norm": 21.658283233642578,
      "learning_rate": 1e-06,
      "loss": 0.5328,
      "num_input_tokens_seen": 208749444,
      "step": 3727
    },
    {
      "epoch": 8.300668151447661,
      "loss": 0.5249356627464294,
      "loss_ce": 0.0001553678303025663,
      "loss_iou": 0.2138671875,
      "loss_num": 0.019287109375,
      "loss_xval": 0.5234375,
      "num_input_tokens_seen": 208749444,
      "step": 3727
    },
    {
      "epoch": 8.302895322939866,
      "grad_norm": 25.13924789428711,
      "learning_rate": 1e-06,
      "loss": 0.5197,
      "num_input_tokens_seen": 208807968,
      "step": 3728
    },
    {
      "epoch": 8.302895322939866,
      "loss": 0.7015814781188965,
      "loss_ce": 0.00016545310791116208,
      "loss_iou": 0.3046875,
      "loss_num": 0.01806640625,
      "loss_xval": 0.703125,
      "num_input_tokens_seen": 208807968,
      "step": 3728
    },
    {
      "epoch": 8.305122494432071,
      "grad_norm": 19.753189086914062,
      "learning_rate": 1e-06,
      "loss": 0.6141,
      "num_input_tokens_seen": 208865056,
      "step": 3729
    },
    {
      "epoch": 8.305122494432071,
      "loss": 0.7641258835792542,
      "loss_ce": 0.0002098674012813717,
      "loss_iou": 0.330078125,
      "loss_num": 0.0211181640625,
      "loss_xval": 0.765625,
      "num_input_tokens_seen": 208865056,
      "step": 3729
    },
    {
      "epoch": 8.307349665924276,
      "grad_norm": 18.014732360839844,
      "learning_rate": 1e-06,
      "loss": 0.5394,
      "num_input_tokens_seen": 208923552,
      "step": 3730
    },
    {
      "epoch": 8.307349665924276,
      "loss": 0.6470487713813782,
      "loss_ce": 0.0001981953828362748,
      "loss_iou": 0.28515625,
      "loss_num": 0.01507568359375,
      "loss_xval": 0.6484375,
      "num_input_tokens_seen": 208923552,
      "step": 3730
    },
    {
      "epoch": 8.309576837416481,
      "grad_norm": 35.129364013671875,
      "learning_rate": 1e-06,
      "loss": 0.6561,
      "num_input_tokens_seen": 208979572,
      "step": 3731
    },
    {
      "epoch": 8.309576837416481,
      "loss": 0.6719157695770264,
      "loss_ce": 0.0002849046722985804,
      "loss_iou": 0.30078125,
      "loss_num": 0.01373291015625,
      "loss_xval": 0.671875,
      "num_input_tokens_seen": 208979572,
      "step": 3731
    },
    {
      "epoch": 8.311804008908686,
      "grad_norm": 22.689146041870117,
      "learning_rate": 1e-06,
      "loss": 0.7743,
      "num_input_tokens_seen": 209034148,
      "step": 3732
    },
    {
      "epoch": 8.311804008908686,
      "loss": 0.7121527194976807,
      "loss_ce": 0.00023866846458986402,
      "loss_iou": 0.28125,
      "loss_num": 0.0302734375,
      "loss_xval": 0.7109375,
      "num_input_tokens_seen": 209034148,
      "step": 3732
    },
    {
      "epoch": 8.31403118040089,
      "grad_norm": 91.16738891601562,
      "learning_rate": 1e-06,
      "loss": 0.5123,
      "num_input_tokens_seen": 209090968,
      "step": 3733
    },
    {
      "epoch": 8.31403118040089,
      "loss": 0.5099412202835083,
      "loss_ce": 0.00017555063823238015,
      "loss_iou": 0.2294921875,
      "loss_num": 0.010009765625,
      "loss_xval": 0.5078125,
      "num_input_tokens_seen": 209090968,
      "step": 3733
    },
    {
      "epoch": 8.316258351893095,
      "grad_norm": 19.895280838012695,
      "learning_rate": 1e-06,
      "loss": 0.6358,
      "num_input_tokens_seen": 209145588,
      "step": 3734
    },
    {
      "epoch": 8.316258351893095,
      "loss": 0.8830050826072693,
      "loss_ce": 0.00019260949920862913,
      "loss_iou": 0.369140625,
      "loss_num": 0.0294189453125,
      "loss_xval": 0.8828125,
      "num_input_tokens_seen": 209145588,
      "step": 3734
    },
    {
      "epoch": 8.3184855233853,
      "grad_norm": 28.7868709564209,
      "learning_rate": 1e-06,
      "loss": 0.6875,
      "num_input_tokens_seen": 209201476,
      "step": 3735
    },
    {
      "epoch": 8.3184855233853,
      "loss": 0.7821846008300781,
      "loss_ce": 0.00020213823881931603,
      "loss_iou": 0.3515625,
      "loss_num": 0.015380859375,
      "loss_xval": 0.78125,
      "num_input_tokens_seen": 209201476,
      "step": 3735
    },
    {
      "epoch": 8.320712694877505,
      "grad_norm": 21.692697525024414,
      "learning_rate": 1e-06,
      "loss": 0.5106,
      "num_input_tokens_seen": 209257012,
      "step": 3736
    },
    {
      "epoch": 8.320712694877505,
      "loss": 0.5338290929794312,
      "loss_ce": 0.0001377178414259106,
      "loss_iou": 0.2392578125,
      "loss_num": 0.01104736328125,
      "loss_xval": 0.53515625,
      "num_input_tokens_seen": 209257012,
      "step": 3736
    },
    {
      "epoch": 8.32293986636971,
      "grad_norm": 24.94318199157715,
      "learning_rate": 1e-06,
      "loss": 0.8265,
      "num_input_tokens_seen": 209312432,
      "step": 3737
    },
    {
      "epoch": 8.32293986636971,
      "loss": 0.6608332395553589,
      "loss_ce": 0.0001887211692519486,
      "loss_iou": 0.2734375,
      "loss_num": 0.022705078125,
      "loss_xval": 0.66015625,
      "num_input_tokens_seen": 209312432,
      "step": 3737
    },
    {
      "epoch": 8.325167037861915,
      "grad_norm": 17.53243637084961,
      "learning_rate": 1e-06,
      "loss": 0.5902,
      "num_input_tokens_seen": 209368084,
      "step": 3738
    },
    {
      "epoch": 8.325167037861915,
      "loss": 0.7750497460365295,
      "loss_ce": 0.00014740778715349734,
      "loss_iou": 0.306640625,
      "loss_num": 0.032470703125,
      "loss_xval": 0.7734375,
      "num_input_tokens_seen": 209368084,
      "step": 3738
    },
    {
      "epoch": 8.32739420935412,
      "grad_norm": 14.224679946899414,
      "learning_rate": 1e-06,
      "loss": 0.5397,
      "num_input_tokens_seen": 209421340,
      "step": 3739
    },
    {
      "epoch": 8.32739420935412,
      "loss": 0.43520334362983704,
      "loss_ce": 0.00014473804912995547,
      "loss_iou": 0.1962890625,
      "loss_num": 0.0086669921875,
      "loss_xval": 0.435546875,
      "num_input_tokens_seen": 209421340,
      "step": 3739
    },
    {
      "epoch": 8.329621380846325,
      "grad_norm": 17.133176803588867,
      "learning_rate": 1e-06,
      "loss": 0.5827,
      "num_input_tokens_seen": 209478168,
      "step": 3740
    },
    {
      "epoch": 8.329621380846325,
      "loss": 0.5902957916259766,
      "loss_ce": 0.0002079373225569725,
      "loss_iou": 0.2431640625,
      "loss_num": 0.0206298828125,
      "loss_xval": 0.58984375,
      "num_input_tokens_seen": 209478168,
      "step": 3740
    },
    {
      "epoch": 8.33184855233853,
      "grad_norm": 23.81924057006836,
      "learning_rate": 1e-06,
      "loss": 0.5458,
      "num_input_tokens_seen": 209533220,
      "step": 3741
    },
    {
      "epoch": 8.33184855233853,
      "loss": 0.7213869094848633,
      "loss_ce": 0.00019548808631952852,
      "loss_iou": 0.322265625,
      "loss_num": 0.0155029296875,
      "loss_xval": 0.72265625,
      "num_input_tokens_seen": 209533220,
      "step": 3741
    },
    {
      "epoch": 8.334075723830734,
      "grad_norm": 16.98560333251953,
      "learning_rate": 1e-06,
      "loss": 0.4944,
      "num_input_tokens_seen": 209589144,
      "step": 3742
    },
    {
      "epoch": 8.334075723830734,
      "loss": 0.3810575604438782,
      "loss_ce": 0.00019817678548861295,
      "loss_iou": 0.1640625,
      "loss_num": 0.01055908203125,
      "loss_xval": 0.380859375,
      "num_input_tokens_seen": 209589144,
      "step": 3742
    },
    {
      "epoch": 8.33630289532294,
      "grad_norm": 18.857044219970703,
      "learning_rate": 1e-06,
      "loss": 0.4678,
      "num_input_tokens_seen": 209640328,
      "step": 3743
    },
    {
      "epoch": 8.33630289532294,
      "loss": 0.413556307554245,
      "loss_ce": 0.00016517053882125765,
      "loss_iou": 0.173828125,
      "loss_num": 0.01312255859375,
      "loss_xval": 0.4140625,
      "num_input_tokens_seen": 209640328,
      "step": 3743
    },
    {
      "epoch": 8.338530066815144,
      "grad_norm": 16.573524475097656,
      "learning_rate": 1e-06,
      "loss": 0.6259,
      "num_input_tokens_seen": 209696820,
      "step": 3744
    },
    {
      "epoch": 8.338530066815144,
      "loss": 0.6830825805664062,
      "loss_ce": 0.00022125753457657993,
      "loss_iou": 0.296875,
      "loss_num": 0.01806640625,
      "loss_xval": 0.68359375,
      "num_input_tokens_seen": 209696820,
      "step": 3744
    },
    {
      "epoch": 8.340757238307349,
      "grad_norm": 29.868751525878906,
      "learning_rate": 1e-06,
      "loss": 0.8029,
      "num_input_tokens_seen": 209750812,
      "step": 3745
    },
    {
      "epoch": 8.340757238307349,
      "loss": 0.7501974701881409,
      "loss_ce": 0.0001974825281649828,
      "loss_iou": 0.31640625,
      "loss_num": 0.0238037109375,
      "loss_xval": 0.75,
      "num_input_tokens_seen": 209750812,
      "step": 3745
    },
    {
      "epoch": 8.342984409799554,
      "grad_norm": 13.687642097473145,
      "learning_rate": 1e-06,
      "loss": 0.6552,
      "num_input_tokens_seen": 209805656,
      "step": 3746
    },
    {
      "epoch": 8.342984409799554,
      "loss": 0.8280588984489441,
      "loss_ce": 0.00017803689115680754,
      "loss_iou": 0.337890625,
      "loss_num": 0.0308837890625,
      "loss_xval": 0.828125,
      "num_input_tokens_seen": 209805656,
      "step": 3746
    },
    {
      "epoch": 8.345211581291759,
      "grad_norm": 40.525962829589844,
      "learning_rate": 1e-06,
      "loss": 0.7054,
      "num_input_tokens_seen": 209859280,
      "step": 3747
    },
    {
      "epoch": 8.345211581291759,
      "loss": 0.7600662708282471,
      "loss_ce": 0.00017863856919575483,
      "loss_iou": 0.3203125,
      "loss_num": 0.02392578125,
      "loss_xval": 0.76171875,
      "num_input_tokens_seen": 209859280,
      "step": 3747
    },
    {
      "epoch": 8.347438752783964,
      "grad_norm": 81.74958801269531,
      "learning_rate": 1e-06,
      "loss": 0.5614,
      "num_input_tokens_seen": 209917268,
      "step": 3748
    },
    {
      "epoch": 8.347438752783964,
      "loss": 0.37694764137268066,
      "loss_ce": 0.0002996893017552793,
      "loss_iou": 0.16015625,
      "loss_num": 0.01129150390625,
      "loss_xval": 0.376953125,
      "num_input_tokens_seen": 209917268,
      "step": 3748
    },
    {
      "epoch": 8.34966592427617,
      "grad_norm": 58.23908996582031,
      "learning_rate": 1e-06,
      "loss": 0.6929,
      "num_input_tokens_seen": 209973232,
      "step": 3749
    },
    {
      "epoch": 8.34966592427617,
      "loss": 0.5084662437438965,
      "loss_ce": 0.0001654803636483848,
      "loss_iou": 0.2294921875,
      "loss_num": 0.00994873046875,
      "loss_xval": 0.5078125,
      "num_input_tokens_seen": 209973232,
      "step": 3749
    },
    {
      "epoch": 8.351893095768375,
      "grad_norm": 14.04624080657959,
      "learning_rate": 1e-06,
      "loss": 0.4782,
      "num_input_tokens_seen": 210030984,
      "step": 3750
    },
    {
      "epoch": 8.351893095768375,
      "eval_seeclick_web_CIoU": 0.5785337090492249,
      "eval_seeclick_web_GIoU": 0.5769274234771729,
      "eval_seeclick_web_IoU": 0.5960685312747955,
      "eval_seeclick_web_MAE_all": 0.01612033136188984,
      "eval_seeclick_web_MAE_h": 0.008123957552015781,
      "eval_seeclick_web_MAE_w": 0.016420952044427395,
      "eval_seeclick_web_MAE_x_boxes": 0.009630883112549782,
      "eval_seeclick_web_MAE_y_boxes": 0.021957224002107978,
      "eval_seeclick_web_inside_bbox": 0.9166666567325592,
      "eval_seeclick_web_loss": 0.9174290299415588,
      "eval_seeclick_web_loss_ce": 0.00023504487762693316,
      "eval_seeclick_web_loss_iou": 0.41943359375,
      "eval_seeclick_web_loss_num": 0.012887954711914062,
      "eval_seeclick_web_loss_xval": 0.90283203125,
      "eval_seeclick_web_runtime": 22.7475,
      "eval_seeclick_web_samples_per_second": 2.198,
      "eval_seeclick_web_steps_per_second": 0.088,
      "num_input_tokens_seen": 210030984,
      "step": 3750
    },
    {
      "epoch": 8.351893095768375,
      "eval_icons_CIoU": 0.2745140343904495,
      "eval_icons_GIoU": 0.2994941622018814,
      "eval_icons_IoU": 0.35809725522994995,
      "eval_icons_MAE_all": 0.06353667378425598,
      "eval_icons_MAE_h": 0.03699115989729762,
      "eval_icons_MAE_w": 0.06941121257841587,
      "eval_icons_MAE_x_boxes": 0.05955472029745579,
      "eval_icons_MAE_y_boxes": 0.038531024008989334,
      "eval_icons_inside_bbox": 0.6059027910232544,
      "eval_icons_loss": 1.762790560722351,
      "eval_icons_loss_ce": 0.0002956779644591734,
      "eval_icons_loss_iou": 0.687744140625,
      "eval_icons_loss_num": 0.06077384948730469,
      "eval_icons_loss_xval": 1.6787109375,
      "eval_icons_runtime": 22.6059,
      "eval_icons_samples_per_second": 2.212,
      "eval_icons_steps_per_second": 0.088,
      "num_input_tokens_seen": 210030984,
      "step": 3750
    },
    {
      "epoch": 8.351893095768375,
      "eval_screenspot_CIoU": 0.34976951281229657,
      "eval_screenspot_GIoU": 0.36631672581036884,
      "eval_screenspot_IoU": 0.4275648792584737,
      "eval_screenspot_MAE_all": 0.05975629389286041,
      "eval_screenspot_MAE_h": 0.03780995992322763,
      "eval_screenspot_MAE_w": 0.06876554464300473,
      "eval_screenspot_MAE_x_boxes": 0.06643692528208096,
      "eval_screenspot_MAE_y_boxes": 0.04602641022453705,
      "eval_screenspot_inside_bbox": 0.6862499912579855,
      "eval_screenspot_loss": 1.6282535791397095,
      "eval_screenspot_loss_ce": 0.0002775423345156014,
      "eval_screenspot_loss_iou": 0.6735026041666666,
      "eval_screenspot_loss_num": 0.07061513264973958,
      "eval_screenspot_loss_xval": 1.7000325520833333,
      "eval_screenspot_runtime": 35.522,
      "eval_screenspot_samples_per_second": 2.505,
      "eval_screenspot_steps_per_second": 0.084,
      "num_input_tokens_seen": 210030984,
      "step": 3750
    },
    {
      "epoch": 8.351893095768375,
      "eval_compot_CIoU": 0.3505849689245224,
      "eval_compot_GIoU": 0.3601333200931549,
      "eval_compot_IoU": 0.40839655697345734,
      "eval_compot_MAE_all": 0.017904515843838453,
      "eval_compot_MAE_h": 0.009080663323402405,
      "eval_compot_MAE_w": 0.022023603320121765,
      "eval_compot_MAE_x_boxes": 0.0294346297159791,
      "eval_compot_MAE_y_boxes": 0.0067735526245087385,
      "eval_compot_inside_bbox": 0.6302083432674408,
      "eval_compot_loss": 1.3908497095108032,
      "eval_compot_loss_ce": 0.00022490947594633326,
      "eval_compot_loss_iou": 0.63720703125,
      "eval_compot_loss_num": 0.01686859130859375,
      "eval_compot_loss_xval": 1.358642578125,
      "eval_compot_runtime": 19.7321,
      "eval_compot_samples_per_second": 2.534,
      "eval_compot_steps_per_second": 0.101,
      "num_input_tokens_seen": 210030984,
      "step": 3750
    },
    {
      "epoch": 8.351893095768375,
      "eval_custom_ui_val_CIoU": 0.470332317882114,
      "eval_custom_ui_val_GIoU": 0.48397915727562374,
      "eval_custom_ui_val_IoU": 0.530869291888343,
      "eval_custom_ui_val_MAE_all": 0.0304783143930965,
      "eval_custom_ui_val_MAE_h": 0.015579992827648917,
      "eval_custom_ui_val_MAE_w": 0.04120294677300586,
      "eval_custom_ui_val_MAE_x_boxes": 0.039027334190905094,
      "eval_custom_ui_val_MAE_y_boxes": 0.015175239571059743,
      "eval_custom_ui_val_inside_bbox": 0.738811731338501,
      "eval_custom_ui_val_loss": 1.183510184288025,
      "eval_custom_ui_val_loss_ce": 0.00025910464238323685,
      "eval_custom_ui_val_loss_iou": 0.5034722222222222,
      "eval_custom_ui_val_loss_num": 0.02816009521484375,
      "eval_custom_ui_val_loss_xval": 1.1477593315972223,
      "eval_custom_ui_val_runtime": 64.4781,
      "eval_custom_ui_val_samples_per_second": 4.11,
      "eval_custom_ui_val_steps_per_second": 0.14,
      "num_input_tokens_seen": 210030984,
      "step": 3750
    },
    {
      "epoch": 8.351893095768375,
      "loss": 0.9277232885360718,
      "loss_ce": 0.00023301383771467954,
      "loss_iou": 0.40234375,
      "loss_num": 0.024169921875,
      "loss_xval": 0.92578125,
      "num_input_tokens_seen": 210030984,
      "step": 3750
    },
    {
      "epoch": 8.35412026726058,
      "grad_norm": 20.638172149658203,
      "learning_rate": 1e-06,
      "loss": 0.6147,
      "num_input_tokens_seen": 210089604,
      "step": 3751
    },
    {
      "epoch": 8.35412026726058,
      "loss": 0.7052416205406189,
      "loss_ce": 0.00016353550017811358,
      "loss_iou": 0.291015625,
      "loss_num": 0.025146484375,
      "loss_xval": 0.703125,
      "num_input_tokens_seen": 210089604,
      "step": 3751
    },
    {
      "epoch": 8.356347438752785,
      "grad_norm": 17.781017303466797,
      "learning_rate": 1e-06,
      "loss": 0.5201,
      "num_input_tokens_seen": 210145624,
      "step": 3752
    },
    {
      "epoch": 8.356347438752785,
      "loss": 0.6434928178787231,
      "loss_ce": 0.00018225307576358318,
      "loss_iou": 0.265625,
      "loss_num": 0.02197265625,
      "loss_xval": 0.64453125,
      "num_input_tokens_seen": 210145624,
      "step": 3752
    },
    {
      "epoch": 8.35857461024499,
      "grad_norm": 88.13903045654297,
      "learning_rate": 1e-06,
      "loss": 0.6379,
      "num_input_tokens_seen": 210203036,
      "step": 3753
    },
    {
      "epoch": 8.35857461024499,
      "loss": 0.6185895204544067,
      "loss_ce": 0.00018133444245904684,
      "loss_iou": 0.28515625,
      "loss_num": 0.00958251953125,
      "loss_xval": 0.6171875,
      "num_input_tokens_seen": 210203036,
      "step": 3753
    },
    {
      "epoch": 8.360801781737194,
      "grad_norm": 15.603206634521484,
      "learning_rate": 1e-06,
      "loss": 0.738,
      "num_input_tokens_seen": 210259024,
      "step": 3754
    },
    {
      "epoch": 8.360801781737194,
      "loss": 0.7939836978912354,
      "loss_ce": 0.0006487197242677212,
      "loss_iou": 0.33984375,
      "loss_num": 0.02294921875,
      "loss_xval": 0.79296875,
      "num_input_tokens_seen": 210259024,
      "step": 3754
    },
    {
      "epoch": 8.3630289532294,
      "grad_norm": 21.302343368530273,
      "learning_rate": 1e-06,
      "loss": 0.7286,
      "num_input_tokens_seen": 210316388,
      "step": 3755
    },
    {
      "epoch": 8.3630289532294,
      "loss": 0.8514755964279175,
      "loss_ce": 0.00015728248399682343,
      "loss_iou": 0.34375,
      "loss_num": 0.03271484375,
      "loss_xval": 0.8515625,
      "num_input_tokens_seen": 210316388,
      "step": 3755
    },
    {
      "epoch": 8.365256124721604,
      "grad_norm": 15.152788162231445,
      "learning_rate": 1e-06,
      "loss": 0.5393,
      "num_input_tokens_seen": 210373268,
      "step": 3756
    },
    {
      "epoch": 8.365256124721604,
      "loss": 0.6328523755073547,
      "loss_ce": 0.00016193960618693382,
      "loss_iou": 0.275390625,
      "loss_num": 0.0166015625,
      "loss_xval": 0.6328125,
      "num_input_tokens_seen": 210373268,
      "step": 3756
    },
    {
      "epoch": 8.367483296213809,
      "grad_norm": 19.08024787902832,
      "learning_rate": 1e-06,
      "loss": 0.6706,
      "num_input_tokens_seen": 210426048,
      "step": 3757
    },
    {
      "epoch": 8.367483296213809,
      "loss": 0.7494222521781921,
      "loss_ce": 0.0001546498097013682,
      "loss_iou": 0.33984375,
      "loss_num": 0.013916015625,
      "loss_xval": 0.75,
      "num_input_tokens_seen": 210426048,
      "step": 3757
    },
    {
      "epoch": 8.369710467706014,
      "grad_norm": 19.812162399291992,
      "learning_rate": 1e-06,
      "loss": 0.6775,
      "num_input_tokens_seen": 210483032,
      "step": 3758
    },
    {
      "epoch": 8.369710467706014,
      "loss": 0.6561557650566101,
      "loss_ce": 0.0001498895580880344,
      "loss_iou": 0.287109375,
      "loss_num": 0.016357421875,
      "loss_xval": 0.65625,
      "num_input_tokens_seen": 210483032,
      "step": 3758
    },
    {
      "epoch": 8.371937639198219,
      "grad_norm": 22.67157554626465,
      "learning_rate": 1e-06,
      "loss": 0.7241,
      "num_input_tokens_seen": 210540188,
      "step": 3759
    },
    {
      "epoch": 8.371937639198219,
      "loss": 0.745628833770752,
      "loss_ce": 0.0002675401628948748,
      "loss_iou": 0.322265625,
      "loss_num": 0.0201416015625,
      "loss_xval": 0.74609375,
      "num_input_tokens_seen": 210540188,
      "step": 3759
    },
    {
      "epoch": 8.374164810690424,
      "grad_norm": 24.26649284362793,
      "learning_rate": 1e-06,
      "loss": 0.6399,
      "num_input_tokens_seen": 210597540,
      "step": 3760
    },
    {
      "epoch": 8.374164810690424,
      "loss": 0.5113849639892578,
      "loss_ce": 0.0001544796396046877,
      "loss_iou": 0.224609375,
      "loss_num": 0.0125732421875,
      "loss_xval": 0.51171875,
      "num_input_tokens_seen": 210597540,
      "step": 3760
    },
    {
      "epoch": 8.376391982182628,
      "grad_norm": 18.92394256591797,
      "learning_rate": 1e-06,
      "loss": 0.6188,
      "num_input_tokens_seen": 210651116,
      "step": 3761
    },
    {
      "epoch": 8.376391982182628,
      "loss": 0.6872795224189758,
      "loss_ce": 0.0002677679876796901,
      "loss_iou": 0.279296875,
      "loss_num": 0.025634765625,
      "loss_xval": 0.6875,
      "num_input_tokens_seen": 210651116,
      "step": 3761
    },
    {
      "epoch": 8.378619153674833,
      "grad_norm": 20.54817771911621,
      "learning_rate": 1e-06,
      "loss": 0.5593,
      "num_input_tokens_seen": 210705180,
      "step": 3762
    },
    {
      "epoch": 8.378619153674833,
      "loss": 0.4156179428100586,
      "loss_ce": 0.00018215348245576024,
      "loss_iou": 0.1796875,
      "loss_num": 0.01104736328125,
      "loss_xval": 0.416015625,
      "num_input_tokens_seen": 210705180,
      "step": 3762
    },
    {
      "epoch": 8.380846325167038,
      "grad_norm": 17.212587356567383,
      "learning_rate": 1e-06,
      "loss": 0.4299,
      "num_input_tokens_seen": 210763276,
      "step": 3763
    },
    {
      "epoch": 8.380846325167038,
      "loss": 0.373112291097641,
      "loss_ce": 0.00018747567082755268,
      "loss_iou": 0.171875,
      "loss_num": 0.0059814453125,
      "loss_xval": 0.373046875,
      "num_input_tokens_seen": 210763276,
      "step": 3763
    },
    {
      "epoch": 8.383073496659243,
      "grad_norm": 19.85719871520996,
      "learning_rate": 1e-06,
      "loss": 0.5694,
      "num_input_tokens_seen": 210818156,
      "step": 3764
    },
    {
      "epoch": 8.383073496659243,
      "loss": 0.5267565250396729,
      "loss_ce": 0.00014518463285639882,
      "loss_iou": 0.2392578125,
      "loss_num": 0.00946044921875,
      "loss_xval": 0.52734375,
      "num_input_tokens_seen": 210818156,
      "step": 3764
    },
    {
      "epoch": 8.385300668151448,
      "grad_norm": 22.43527603149414,
      "learning_rate": 1e-06,
      "loss": 0.7907,
      "num_input_tokens_seen": 210873660,
      "step": 3765
    },
    {
      "epoch": 8.385300668151448,
      "loss": 0.8294920921325684,
      "loss_ce": 0.0001463915396016091,
      "loss_iou": 0.330078125,
      "loss_num": 0.033935546875,
      "loss_xval": 0.828125,
      "num_input_tokens_seen": 210873660,
      "step": 3765
    },
    {
      "epoch": 8.387527839643653,
      "grad_norm": 15.67119312286377,
      "learning_rate": 1e-06,
      "loss": 0.4889,
      "num_input_tokens_seen": 210928384,
      "step": 3766
    },
    {
      "epoch": 8.387527839643653,
      "loss": 0.5981723070144653,
      "loss_ce": 0.0002108315529767424,
      "loss_iou": 0.251953125,
      "loss_num": 0.018798828125,
      "loss_xval": 0.59765625,
      "num_input_tokens_seen": 210928384,
      "step": 3766
    },
    {
      "epoch": 8.389755011135858,
      "grad_norm": 84.53565216064453,
      "learning_rate": 1e-06,
      "loss": 0.6099,
      "num_input_tokens_seen": 210983112,
      "step": 3767
    },
    {
      "epoch": 8.389755011135858,
      "loss": 0.5748885869979858,
      "loss_ce": 0.00018153036944568157,
      "loss_iou": 0.251953125,
      "loss_num": 0.01446533203125,
      "loss_xval": 0.57421875,
      "num_input_tokens_seen": 210983112,
      "step": 3767
    },
    {
      "epoch": 8.391982182628063,
      "grad_norm": 21.295007705688477,
      "learning_rate": 1e-06,
      "loss": 0.5424,
      "num_input_tokens_seen": 211039244,
      "step": 3768
    },
    {
      "epoch": 8.391982182628063,
      "loss": 0.5405082702636719,
      "loss_ce": 0.0002250939724035561,
      "loss_iou": 0.23828125,
      "loss_num": 0.01275634765625,
      "loss_xval": 0.5390625,
      "num_input_tokens_seen": 211039244,
      "step": 3768
    },
    {
      "epoch": 8.394209354120267,
      "grad_norm": 28.580411911010742,
      "learning_rate": 1e-06,
      "loss": 0.4886,
      "num_input_tokens_seen": 211093216,
      "step": 3769
    },
    {
      "epoch": 8.394209354120267,
      "loss": 0.4398452043533325,
      "loss_ce": 0.0001479371276218444,
      "loss_iou": 0.1953125,
      "loss_num": 0.009765625,
      "loss_xval": 0.439453125,
      "num_input_tokens_seen": 211093216,
      "step": 3769
    },
    {
      "epoch": 8.396436525612472,
      "grad_norm": 23.4449462890625,
      "learning_rate": 1e-06,
      "loss": 0.6098,
      "num_input_tokens_seen": 211145896,
      "step": 3770
    },
    {
      "epoch": 8.396436525612472,
      "loss": 0.4624381363391876,
      "loss_ce": 0.0002799497451633215,
      "loss_iou": 0.201171875,
      "loss_num": 0.0120849609375,
      "loss_xval": 0.462890625,
      "num_input_tokens_seen": 211145896,
      "step": 3770
    },
    {
      "epoch": 8.398663697104677,
      "grad_norm": 21.414167404174805,
      "learning_rate": 1e-06,
      "loss": 0.5727,
      "num_input_tokens_seen": 211200436,
      "step": 3771
    },
    {
      "epoch": 8.398663697104677,
      "loss": 0.6622140407562256,
      "loss_ce": 0.00022672140039503574,
      "loss_iou": 0.287109375,
      "loss_num": 0.017333984375,
      "loss_xval": 0.66015625,
      "num_input_tokens_seen": 211200436,
      "step": 3771
    },
    {
      "epoch": 8.400890868596882,
      "grad_norm": 24.094797134399414,
      "learning_rate": 1e-06,
      "loss": 0.4669,
      "num_input_tokens_seen": 211253492,
      "step": 3772
    },
    {
      "epoch": 8.400890868596882,
      "loss": 0.3054664731025696,
      "loss_ce": 0.00016864070494193584,
      "loss_iou": 0.1337890625,
      "loss_num": 0.0076904296875,
      "loss_xval": 0.3046875,
      "num_input_tokens_seen": 211253492,
      "step": 3772
    },
    {
      "epoch": 8.403118040089087,
      "grad_norm": 16.602676391601562,
      "learning_rate": 1e-06,
      "loss": 0.6977,
      "num_input_tokens_seen": 211309996,
      "step": 3773
    },
    {
      "epoch": 8.403118040089087,
      "loss": 0.8234143853187561,
      "loss_ce": 0.00017219953588210046,
      "loss_iou": 0.3046875,
      "loss_num": 0.042724609375,
      "loss_xval": 0.82421875,
      "num_input_tokens_seen": 211309996,
      "step": 3773
    },
    {
      "epoch": 8.405345211581292,
      "grad_norm": 15.6812744140625,
      "learning_rate": 1e-06,
      "loss": 0.5358,
      "num_input_tokens_seen": 211365484,
      "step": 3774
    },
    {
      "epoch": 8.405345211581292,
      "loss": 0.5414432287216187,
      "loss_ce": 0.0001529736036900431,
      "loss_iou": 0.2333984375,
      "loss_num": 0.01495361328125,
      "loss_xval": 0.54296875,
      "num_input_tokens_seen": 211365484,
      "step": 3774
    },
    {
      "epoch": 8.407572383073497,
      "grad_norm": 15.80637264251709,
      "learning_rate": 1e-06,
      "loss": 0.5815,
      "num_input_tokens_seen": 211421912,
      "step": 3775
    },
    {
      "epoch": 8.407572383073497,
      "loss": 0.4960188865661621,
      "loss_ce": 0.000169286533491686,
      "loss_iou": 0.212890625,
      "loss_num": 0.01409912109375,
      "loss_xval": 0.49609375,
      "num_input_tokens_seen": 211421912,
      "step": 3775
    },
    {
      "epoch": 8.409799554565701,
      "grad_norm": 22.33266830444336,
      "learning_rate": 1e-06,
      "loss": 0.7466,
      "num_input_tokens_seen": 211478792,
      "step": 3776
    },
    {
      "epoch": 8.409799554565701,
      "loss": 0.7512620091438293,
      "loss_ce": 0.00016335461987182498,
      "loss_iou": 0.302734375,
      "loss_num": 0.0286865234375,
      "loss_xval": 0.75,
      "num_input_tokens_seen": 211478792,
      "step": 3776
    },
    {
      "epoch": 8.412026726057906,
      "grad_norm": 18.270977020263672,
      "learning_rate": 1e-06,
      "loss": 0.6434,
      "num_input_tokens_seen": 211535448,
      "step": 3777
    },
    {
      "epoch": 8.412026726057906,
      "loss": 0.44083136320114136,
      "loss_ce": 0.00015753594925627112,
      "loss_iou": 0.1962890625,
      "loss_num": 0.00946044921875,
      "loss_xval": 0.44140625,
      "num_input_tokens_seen": 211535448,
      "step": 3777
    },
    {
      "epoch": 8.414253897550111,
      "grad_norm": 26.987186431884766,
      "learning_rate": 1e-06,
      "loss": 0.6894,
      "num_input_tokens_seen": 211590956,
      "step": 3778
    },
    {
      "epoch": 8.414253897550111,
      "loss": 0.9005333781242371,
      "loss_ce": 0.00014275358989834785,
      "loss_iou": 0.3671875,
      "loss_num": 0.032958984375,
      "loss_xval": 0.8984375,
      "num_input_tokens_seen": 211590956,
      "step": 3778
    },
    {
      "epoch": 8.416481069042316,
      "grad_norm": 36.464210510253906,
      "learning_rate": 1e-06,
      "loss": 0.5899,
      "num_input_tokens_seen": 211645452,
      "step": 3779
    },
    {
      "epoch": 8.416481069042316,
      "loss": 0.5824276804924011,
      "loss_ce": 0.00015230441931635141,
      "loss_iou": 0.251953125,
      "loss_num": 0.0157470703125,
      "loss_xval": 0.58203125,
      "num_input_tokens_seen": 211645452,
      "step": 3779
    },
    {
      "epoch": 8.41870824053452,
      "grad_norm": 18.14861297607422,
      "learning_rate": 1e-06,
      "loss": 0.6356,
      "num_input_tokens_seen": 211701036,
      "step": 3780
    },
    {
      "epoch": 8.41870824053452,
      "loss": 0.8778401613235474,
      "loss_ce": 0.0001545833656564355,
      "loss_iou": 0.353515625,
      "loss_num": 0.03369140625,
      "loss_xval": 0.87890625,
      "num_input_tokens_seen": 211701036,
      "step": 3780
    },
    {
      "epoch": 8.420935412026726,
      "grad_norm": 13.912912368774414,
      "learning_rate": 1e-06,
      "loss": 0.6765,
      "num_input_tokens_seen": 211757800,
      "step": 3781
    },
    {
      "epoch": 8.420935412026726,
      "loss": 0.7191988229751587,
      "loss_ce": 0.0002046736772172153,
      "loss_iou": 0.314453125,
      "loss_num": 0.0181884765625,
      "loss_xval": 0.71875,
      "num_input_tokens_seen": 211757800,
      "step": 3781
    },
    {
      "epoch": 8.42316258351893,
      "grad_norm": 28.029550552368164,
      "learning_rate": 1e-06,
      "loss": 0.5842,
      "num_input_tokens_seen": 211815608,
      "step": 3782
    },
    {
      "epoch": 8.42316258351893,
      "loss": 0.5234848260879517,
      "loss_ce": 0.00016937771579250693,
      "loss_iou": 0.220703125,
      "loss_num": 0.01611328125,
      "loss_xval": 0.5234375,
      "num_input_tokens_seen": 211815608,
      "step": 3782
    },
    {
      "epoch": 8.425389755011135,
      "grad_norm": 14.431218147277832,
      "learning_rate": 1e-06,
      "loss": 0.7867,
      "num_input_tokens_seen": 211872108,
      "step": 3783
    },
    {
      "epoch": 8.425389755011135,
      "loss": 0.6559756994247437,
      "loss_ce": 0.0002139731077477336,
      "loss_iou": 0.26171875,
      "loss_num": 0.0262451171875,
      "loss_xval": 0.65625,
      "num_input_tokens_seen": 211872108,
      "step": 3783
    },
    {
      "epoch": 8.42761692650334,
      "grad_norm": 35.63063430786133,
      "learning_rate": 1e-06,
      "loss": 0.7438,
      "num_input_tokens_seen": 211927844,
      "step": 3784
    },
    {
      "epoch": 8.42761692650334,
      "loss": 0.7180646657943726,
      "loss_ce": 0.0001691640354692936,
      "loss_iou": 0.330078125,
      "loss_num": 0.01165771484375,
      "loss_xval": 0.71875,
      "num_input_tokens_seen": 211927844,
      "step": 3784
    },
    {
      "epoch": 8.429844097995545,
      "grad_norm": 29.35708999633789,
      "learning_rate": 1e-06,
      "loss": 0.574,
      "num_input_tokens_seen": 211979040,
      "step": 3785
    },
    {
      "epoch": 8.429844097995545,
      "loss": 0.4887080192565918,
      "loss_ce": 0.0001826368534239009,
      "loss_iou": 0.2197265625,
      "loss_num": 0.00982666015625,
      "loss_xval": 0.48828125,
      "num_input_tokens_seen": 211979040,
      "step": 3785
    },
    {
      "epoch": 8.43207126948775,
      "grad_norm": 21.319759368896484,
      "learning_rate": 1e-06,
      "loss": 0.5513,
      "num_input_tokens_seen": 212035452,
      "step": 3786
    },
    {
      "epoch": 8.43207126948775,
      "loss": 0.363552451133728,
      "loss_ce": 0.00014914579514879733,
      "loss_iou": 0.15625,
      "loss_num": 0.0101318359375,
      "loss_xval": 0.36328125,
      "num_input_tokens_seen": 212035452,
      "step": 3786
    },
    {
      "epoch": 8.434298440979955,
      "grad_norm": 23.589679718017578,
      "learning_rate": 1e-06,
      "loss": 0.7261,
      "num_input_tokens_seen": 212093292,
      "step": 3787
    },
    {
      "epoch": 8.434298440979955,
      "loss": 0.6076536178588867,
      "loss_ce": 0.00023178444826044142,
      "loss_iou": 0.240234375,
      "loss_num": 0.025634765625,
      "loss_xval": 0.609375,
      "num_input_tokens_seen": 212093292,
      "step": 3787
    },
    {
      "epoch": 8.43652561247216,
      "grad_norm": 14.83303451538086,
      "learning_rate": 1e-06,
      "loss": 0.4061,
      "num_input_tokens_seen": 212150376,
      "step": 3788
    },
    {
      "epoch": 8.43652561247216,
      "loss": 0.35096991062164307,
      "loss_ce": 0.0001398474269080907,
      "loss_iou": 0.1474609375,
      "loss_num": 0.0111083984375,
      "loss_xval": 0.3515625,
      "num_input_tokens_seen": 212150376,
      "step": 3788
    },
    {
      "epoch": 8.438752783964365,
      "grad_norm": 20.81564712524414,
      "learning_rate": 1e-06,
      "loss": 0.6,
      "num_input_tokens_seen": 212202624,
      "step": 3789
    },
    {
      "epoch": 8.438752783964365,
      "loss": 0.5509481430053711,
      "loss_ce": 0.00016686462913639843,
      "loss_iou": 0.2451171875,
      "loss_num": 0.0120849609375,
      "loss_xval": 0.55078125,
      "num_input_tokens_seen": 212202624,
      "step": 3789
    },
    {
      "epoch": 8.44097995545657,
      "grad_norm": 15.132460594177246,
      "learning_rate": 1e-06,
      "loss": 0.7095,
      "num_input_tokens_seen": 212258148,
      "step": 3790
    },
    {
      "epoch": 8.44097995545657,
      "loss": 0.7777798175811768,
      "loss_ce": 0.00019189229351468384,
      "loss_iou": 0.322265625,
      "loss_num": 0.0269775390625,
      "loss_xval": 0.77734375,
      "num_input_tokens_seen": 212258148,
      "step": 3790
    },
    {
      "epoch": 8.443207126948774,
      "grad_norm": 16.23349952697754,
      "learning_rate": 1e-06,
      "loss": 0.4949,
      "num_input_tokens_seen": 212315600,
      "step": 3791
    },
    {
      "epoch": 8.443207126948774,
      "loss": 0.6320212483406067,
      "loss_ce": 0.00018532070680521429,
      "loss_iou": 0.25390625,
      "loss_num": 0.02490234375,
      "loss_xval": 0.6328125,
      "num_input_tokens_seen": 212315600,
      "step": 3791
    },
    {
      "epoch": 8.44543429844098,
      "grad_norm": 29.736553192138672,
      "learning_rate": 1e-06,
      "loss": 0.6874,
      "num_input_tokens_seen": 212371436,
      "step": 3792
    },
    {
      "epoch": 8.44543429844098,
      "loss": 0.7760206460952759,
      "loss_ce": 0.00014173590170685202,
      "loss_iou": 0.34765625,
      "loss_num": 0.0164794921875,
      "loss_xval": 0.77734375,
      "num_input_tokens_seen": 212371436,
      "step": 3792
    },
    {
      "epoch": 8.447661469933184,
      "grad_norm": 17.218870162963867,
      "learning_rate": 1e-06,
      "loss": 0.5741,
      "num_input_tokens_seen": 212427368,
      "step": 3793
    },
    {
      "epoch": 8.447661469933184,
      "loss": 0.7538229823112488,
      "loss_ce": 0.00016089789278339595,
      "loss_iou": 0.30859375,
      "loss_num": 0.0272216796875,
      "loss_xval": 0.75390625,
      "num_input_tokens_seen": 212427368,
      "step": 3793
    },
    {
      "epoch": 8.449888641425389,
      "grad_norm": 17.727853775024414,
      "learning_rate": 1e-06,
      "loss": 0.4507,
      "num_input_tokens_seen": 212482836,
      "step": 3794
    },
    {
      "epoch": 8.449888641425389,
      "loss": 0.5649484395980835,
      "loss_ce": 0.0004953413736075163,
      "loss_iou": 0.2353515625,
      "loss_num": 0.018798828125,
      "loss_xval": 0.5625,
      "num_input_tokens_seen": 212482836,
      "step": 3794
    },
    {
      "epoch": 8.452115812917596,
      "grad_norm": 21.763526916503906,
      "learning_rate": 1e-06,
      "loss": 0.518,
      "num_input_tokens_seen": 212537884,
      "step": 3795
    },
    {
      "epoch": 8.452115812917596,
      "loss": 0.43735483288764954,
      "loss_ce": 0.00022105529205873609,
      "loss_iou": 0.189453125,
      "loss_num": 0.01190185546875,
      "loss_xval": 0.4375,
      "num_input_tokens_seen": 212537884,
      "step": 3795
    },
    {
      "epoch": 8.4543429844098,
      "grad_norm": 29.629959106445312,
      "learning_rate": 1e-06,
      "loss": 0.5047,
      "num_input_tokens_seen": 212592316,
      "step": 3796
    },
    {
      "epoch": 8.4543429844098,
      "loss": 0.4535437524318695,
      "loss_ce": 0.00017461413517594337,
      "loss_iou": 0.1845703125,
      "loss_num": 0.0167236328125,
      "loss_xval": 0.453125,
      "num_input_tokens_seen": 212592316,
      "step": 3796
    },
    {
      "epoch": 8.456570155902005,
      "grad_norm": 33.515464782714844,
      "learning_rate": 1e-06,
      "loss": 0.7209,
      "num_input_tokens_seen": 212649160,
      "step": 3797
    },
    {
      "epoch": 8.456570155902005,
      "loss": 0.8188025951385498,
      "loss_ce": 0.0001990534656215459,
      "loss_iou": 0.357421875,
      "loss_num": 0.02099609375,
      "loss_xval": 0.8203125,
      "num_input_tokens_seen": 212649160,
      "step": 3797
    },
    {
      "epoch": 8.45879732739421,
      "grad_norm": 25.849811553955078,
      "learning_rate": 1e-06,
      "loss": 0.6433,
      "num_input_tokens_seen": 212705208,
      "step": 3798
    },
    {
      "epoch": 8.45879732739421,
      "loss": 0.7545800805091858,
      "loss_ce": 0.0001855643349699676,
      "loss_iou": 0.34375,
      "loss_num": 0.01324462890625,
      "loss_xval": 0.75390625,
      "num_input_tokens_seen": 212705208,
      "step": 3798
    },
    {
      "epoch": 8.461024498886415,
      "grad_norm": 27.304214477539062,
      "learning_rate": 1e-06,
      "loss": 0.6453,
      "num_input_tokens_seen": 212760512,
      "step": 3799
    },
    {
      "epoch": 8.461024498886415,
      "loss": 0.5861043334007263,
      "loss_ce": 0.00016680179396644235,
      "loss_iou": 0.240234375,
      "loss_num": 0.02099609375,
      "loss_xval": 0.5859375,
      "num_input_tokens_seen": 212760512,
      "step": 3799
    },
    {
      "epoch": 8.46325167037862,
      "grad_norm": 23.8879451751709,
      "learning_rate": 1e-06,
      "loss": 0.6665,
      "num_input_tokens_seen": 212817608,
      "step": 3800
    },
    {
      "epoch": 8.46325167037862,
      "loss": 0.6634845733642578,
      "loss_ce": 0.00015447995974682271,
      "loss_iou": 0.310546875,
      "loss_num": 0.00811767578125,
      "loss_xval": 0.6640625,
      "num_input_tokens_seen": 212817608,
      "step": 3800
    },
    {
      "epoch": 8.465478841870825,
      "grad_norm": 42.28864288330078,
      "learning_rate": 1e-06,
      "loss": 0.7315,
      "num_input_tokens_seen": 212873236,
      "step": 3801
    },
    {
      "epoch": 8.465478841870825,
      "loss": 0.7518316507339478,
      "loss_ce": 0.0003667787532322109,
      "loss_iou": 0.3203125,
      "loss_num": 0.0223388671875,
      "loss_xval": 0.75,
      "num_input_tokens_seen": 212873236,
      "step": 3801
    },
    {
      "epoch": 8.46770601336303,
      "grad_norm": 17.901504516601562,
      "learning_rate": 1e-06,
      "loss": 0.6369,
      "num_input_tokens_seen": 212929028,
      "step": 3802
    },
    {
      "epoch": 8.46770601336303,
      "loss": 0.6398534178733826,
      "loss_ce": 0.000204975571250543,
      "loss_iou": 0.26953125,
      "loss_num": 0.020263671875,
      "loss_xval": 0.640625,
      "num_input_tokens_seen": 212929028,
      "step": 3802
    },
    {
      "epoch": 8.469933184855234,
      "grad_norm": 12.724729537963867,
      "learning_rate": 1e-06,
      "loss": 0.726,
      "num_input_tokens_seen": 212985068,
      "step": 3803
    },
    {
      "epoch": 8.469933184855234,
      "loss": 0.5726535320281982,
      "loss_ce": 0.00014379943604581058,
      "loss_iou": 0.23046875,
      "loss_num": 0.0224609375,
      "loss_xval": 0.57421875,
      "num_input_tokens_seen": 212985068,
      "step": 3803
    },
    {
      "epoch": 8.47216035634744,
      "grad_norm": 17.61178207397461,
      "learning_rate": 1e-06,
      "loss": 0.4873,
      "num_input_tokens_seen": 213042936,
      "step": 3804
    },
    {
      "epoch": 8.47216035634744,
      "loss": 0.37965720891952515,
      "loss_ce": 0.00014062756963539869,
      "loss_iou": 0.1708984375,
      "loss_num": 0.00750732421875,
      "loss_xval": 0.37890625,
      "num_input_tokens_seen": 213042936,
      "step": 3804
    },
    {
      "epoch": 8.474387527839644,
      "grad_norm": 17.672943115234375,
      "learning_rate": 1e-06,
      "loss": 0.5562,
      "num_input_tokens_seen": 213095496,
      "step": 3805
    },
    {
      "epoch": 8.474387527839644,
      "loss": 0.5798885822296143,
      "loss_ce": 0.00017669444787316024,
      "loss_iou": 0.251953125,
      "loss_num": 0.015380859375,
      "loss_xval": 0.578125,
      "num_input_tokens_seen": 213095496,
      "step": 3805
    },
    {
      "epoch": 8.476614699331849,
      "grad_norm": 15.858698844909668,
      "learning_rate": 1e-06,
      "loss": 0.5357,
      "num_input_tokens_seen": 213151276,
      "step": 3806
    },
    {
      "epoch": 8.476614699331849,
      "loss": 0.3039921820163727,
      "loss_ce": 0.00015917010023258626,
      "loss_iou": 0.126953125,
      "loss_num": 0.01007080078125,
      "loss_xval": 0.3046875,
      "num_input_tokens_seen": 213151276,
      "step": 3806
    },
    {
      "epoch": 8.478841870824054,
      "grad_norm": 19.456628799438477,
      "learning_rate": 1e-06,
      "loss": 0.5587,
      "num_input_tokens_seen": 213208404,
      "step": 3807
    },
    {
      "epoch": 8.478841870824054,
      "loss": 0.4867333769798279,
      "loss_ce": 0.0001611171173863113,
      "loss_iou": 0.208984375,
      "loss_num": 0.01385498046875,
      "loss_xval": 0.486328125,
      "num_input_tokens_seen": 213208404,
      "step": 3807
    },
    {
      "epoch": 8.481069042316259,
      "grad_norm": 16.330257415771484,
      "learning_rate": 1e-06,
      "loss": 0.592,
      "num_input_tokens_seen": 213266032,
      "step": 3808
    },
    {
      "epoch": 8.481069042316259,
      "loss": 0.8233369588851929,
      "loss_ce": 0.00015580856415908784,
      "loss_iou": 0.330078125,
      "loss_num": 0.033203125,
      "loss_xval": 0.82421875,
      "num_input_tokens_seen": 213266032,
      "step": 3808
    },
    {
      "epoch": 8.483296213808464,
      "grad_norm": 18.079784393310547,
      "learning_rate": 1e-06,
      "loss": 0.5542,
      "num_input_tokens_seen": 213321096,
      "step": 3809
    },
    {
      "epoch": 8.483296213808464,
      "loss": 0.5775792598724365,
      "loss_ce": 0.00018672105215955526,
      "loss_iou": 0.267578125,
      "loss_num": 0.00811767578125,
      "loss_xval": 0.578125,
      "num_input_tokens_seen": 213321096,
      "step": 3809
    },
    {
      "epoch": 8.485523385300668,
      "grad_norm": 14.717164039611816,
      "learning_rate": 1e-06,
      "loss": 0.5484,
      "num_input_tokens_seen": 213379256,
      "step": 3810
    },
    {
      "epoch": 8.485523385300668,
      "loss": 0.6559404134750366,
      "loss_ce": 0.00017873873002827168,
      "loss_iou": 0.2890625,
      "loss_num": 0.01531982421875,
      "loss_xval": 0.65625,
      "num_input_tokens_seen": 213379256,
      "step": 3810
    },
    {
      "epoch": 8.487750556792873,
      "grad_norm": 15.925959587097168,
      "learning_rate": 1e-06,
      "loss": 0.5738,
      "num_input_tokens_seen": 213436860,
      "step": 3811
    },
    {
      "epoch": 8.487750556792873,
      "loss": 0.6409112811088562,
      "loss_ce": 0.00016421903274022043,
      "loss_iou": 0.275390625,
      "loss_num": 0.01806640625,
      "loss_xval": 0.640625,
      "num_input_tokens_seen": 213436860,
      "step": 3811
    },
    {
      "epoch": 8.489977728285078,
      "grad_norm": 16.596834182739258,
      "learning_rate": 1e-06,
      "loss": 0.5186,
      "num_input_tokens_seen": 213494208,
      "step": 3812
    },
    {
      "epoch": 8.489977728285078,
      "loss": 0.4774528443813324,
      "loss_ce": 0.00015789938333909959,
      "loss_iou": 0.2109375,
      "loss_num": 0.0113525390625,
      "loss_xval": 0.4765625,
      "num_input_tokens_seen": 213494208,
      "step": 3812
    },
    {
      "epoch": 8.492204899777283,
      "grad_norm": 20.76615333557129,
      "learning_rate": 1e-06,
      "loss": 0.5608,
      "num_input_tokens_seen": 213550704,
      "step": 3813
    },
    {
      "epoch": 8.492204899777283,
      "loss": 0.6620367765426636,
      "loss_ce": 0.00017153860244434327,
      "loss_iou": 0.296875,
      "loss_num": 0.01361083984375,
      "loss_xval": 0.66015625,
      "num_input_tokens_seen": 213550704,
      "step": 3813
    },
    {
      "epoch": 8.494432071269488,
      "grad_norm": 35.46400451660156,
      "learning_rate": 1e-06,
      "loss": 0.4233,
      "num_input_tokens_seen": 213605856,
      "step": 3814
    },
    {
      "epoch": 8.494432071269488,
      "loss": 0.4266274869441986,
      "loss_ce": 0.00011381316289771348,
      "loss_iou": 0.1982421875,
      "loss_num": 0.00592041015625,
      "loss_xval": 0.42578125,
      "num_input_tokens_seen": 213605856,
      "step": 3814
    },
    {
      "epoch": 8.496659242761693,
      "grad_norm": 17.748003005981445,
      "learning_rate": 1e-06,
      "loss": 0.4427,
      "num_input_tokens_seen": 213661372,
      "step": 3815
    },
    {
      "epoch": 8.496659242761693,
      "loss": 0.5624146461486816,
      "loss_ce": 0.00015880668070167303,
      "loss_iou": 0.2216796875,
      "loss_num": 0.023681640625,
      "loss_xval": 0.5625,
      "num_input_tokens_seen": 213661372,
      "step": 3815
    },
    {
      "epoch": 8.498886414253898,
      "grad_norm": 17.674402236938477,
      "learning_rate": 1e-06,
      "loss": 0.5981,
      "num_input_tokens_seen": 213716564,
      "step": 3816
    },
    {
      "epoch": 8.498886414253898,
      "loss": 0.5442743897438049,
      "loss_ce": 0.00020702678011730313,
      "loss_iou": 0.2412109375,
      "loss_num": 0.0125732421875,
      "loss_xval": 0.54296875,
      "num_input_tokens_seen": 213716564,
      "step": 3816
    },
    {
      "epoch": 8.501113585746102,
      "grad_norm": 15.914905548095703,
      "learning_rate": 1e-06,
      "loss": 0.6234,
      "num_input_tokens_seen": 213775080,
      "step": 3817
    },
    {
      "epoch": 8.501113585746102,
      "loss": 0.6505662202835083,
      "loss_ce": 0.00017557844694238156,
      "loss_iou": 0.296875,
      "loss_num": 0.0111083984375,
      "loss_xval": 0.6484375,
      "num_input_tokens_seen": 213775080,
      "step": 3817
    },
    {
      "epoch": 8.503340757238307,
      "grad_norm": 13.035726547241211,
      "learning_rate": 1e-06,
      "loss": 0.4907,
      "num_input_tokens_seen": 213832808,
      "step": 3818
    },
    {
      "epoch": 8.503340757238307,
      "loss": 0.5575791001319885,
      "loss_ce": 0.00020606406906154007,
      "loss_iou": 0.2578125,
      "loss_num": 0.00848388671875,
      "loss_xval": 0.55859375,
      "num_input_tokens_seen": 213832808,
      "step": 3818
    },
    {
      "epoch": 8.505567928730512,
      "grad_norm": 21.998821258544922,
      "learning_rate": 1e-06,
      "loss": 0.6187,
      "num_input_tokens_seen": 213892096,
      "step": 3819
    },
    {
      "epoch": 8.505567928730512,
      "loss": 0.5462846755981445,
      "loss_ce": 0.0001420707703800872,
      "loss_iou": 0.2353515625,
      "loss_num": 0.01495361328125,
      "loss_xval": 0.546875,
      "num_input_tokens_seen": 213892096,
      "step": 3819
    },
    {
      "epoch": 8.507795100222717,
      "grad_norm": 38.21397399902344,
      "learning_rate": 1e-06,
      "loss": 0.6985,
      "num_input_tokens_seen": 213946800,
      "step": 3820
    },
    {
      "epoch": 8.507795100222717,
      "loss": 0.7257634401321411,
      "loss_ce": 0.00017748677055351436,
      "loss_iou": 0.326171875,
      "loss_num": 0.01470947265625,
      "loss_xval": 0.7265625,
      "num_input_tokens_seen": 213946800,
      "step": 3820
    },
    {
      "epoch": 8.510022271714922,
      "grad_norm": 24.83524513244629,
      "learning_rate": 1e-06,
      "loss": 0.5032,
      "num_input_tokens_seen": 214003200,
      "step": 3821
    },
    {
      "epoch": 8.510022271714922,
      "loss": 0.464314341545105,
      "loss_ce": 0.0004471480497159064,
      "loss_iou": 0.1953125,
      "loss_num": 0.014892578125,
      "loss_xval": 0.46484375,
      "num_input_tokens_seen": 214003200,
      "step": 3821
    },
    {
      "epoch": 8.512249443207127,
      "grad_norm": 18.419172286987305,
      "learning_rate": 1e-06,
      "loss": 0.6254,
      "num_input_tokens_seen": 214059832,
      "step": 3822
    },
    {
      "epoch": 8.512249443207127,
      "loss": 0.5757325887680054,
      "loss_ce": 0.00017112254863604903,
      "loss_iou": 0.255859375,
      "loss_num": 0.0125732421875,
      "loss_xval": 0.57421875,
      "num_input_tokens_seen": 214059832,
      "step": 3822
    },
    {
      "epoch": 8.514476614699332,
      "grad_norm": 23.88714599609375,
      "learning_rate": 1e-06,
      "loss": 0.7178,
      "num_input_tokens_seen": 214115268,
      "step": 3823
    },
    {
      "epoch": 8.514476614699332,
      "loss": 0.9213709235191345,
      "loss_ce": 0.00022837005963083357,
      "loss_iou": 0.365234375,
      "loss_num": 0.037841796875,
      "loss_xval": 0.921875,
      "num_input_tokens_seen": 214115268,
      "step": 3823
    },
    {
      "epoch": 8.516703786191536,
      "grad_norm": 16.07703399658203,
      "learning_rate": 1e-06,
      "loss": 0.774,
      "num_input_tokens_seen": 214173736,
      "step": 3824
    },
    {
      "epoch": 8.516703786191536,
      "loss": 0.5605735778808594,
      "loss_ce": 0.00014879286754876375,
      "loss_iou": 0.23828125,
      "loss_num": 0.0167236328125,
      "loss_xval": 0.55859375,
      "num_input_tokens_seen": 214173736,
      "step": 3824
    },
    {
      "epoch": 8.518930957683741,
      "grad_norm": 17.80256462097168,
      "learning_rate": 1e-06,
      "loss": 0.5559,
      "num_input_tokens_seen": 214231872,
      "step": 3825
    },
    {
      "epoch": 8.518930957683741,
      "loss": 0.5602927207946777,
      "loss_ce": 0.00017310409748461097,
      "loss_iou": 0.251953125,
      "loss_num": 0.01141357421875,
      "loss_xval": 0.55859375,
      "num_input_tokens_seen": 214231872,
      "step": 3825
    },
    {
      "epoch": 8.521158129175946,
      "grad_norm": 15.230652809143066,
      "learning_rate": 1e-06,
      "loss": 0.5637,
      "num_input_tokens_seen": 214289616,
      "step": 3826
    },
    {
      "epoch": 8.521158129175946,
      "loss": 0.6788380146026611,
      "loss_ce": 0.0002491388004273176,
      "loss_iou": 0.28125,
      "loss_num": 0.0234375,
      "loss_xval": 0.6796875,
      "num_input_tokens_seen": 214289616,
      "step": 3826
    },
    {
      "epoch": 8.523385300668151,
      "grad_norm": 19.55664825439453,
      "learning_rate": 1e-06,
      "loss": 0.4918,
      "num_input_tokens_seen": 214347108,
      "step": 3827
    },
    {
      "epoch": 8.523385300668151,
      "loss": 0.4482664167881012,
      "loss_ce": 0.00014631151861976832,
      "loss_iou": 0.185546875,
      "loss_num": 0.01519775390625,
      "loss_xval": 0.447265625,
      "num_input_tokens_seen": 214347108,
      "step": 3827
    },
    {
      "epoch": 8.525612472160356,
      "grad_norm": 19.749784469604492,
      "learning_rate": 1e-06,
      "loss": 0.5827,
      "num_input_tokens_seen": 214399800,
      "step": 3828
    },
    {
      "epoch": 8.525612472160356,
      "loss": 0.7083353996276855,
      "loss_ce": 0.00020547436724882573,
      "loss_iou": 0.3125,
      "loss_num": 0.0166015625,
      "loss_xval": 0.70703125,
      "num_input_tokens_seen": 214399800,
      "step": 3828
    },
    {
      "epoch": 8.52783964365256,
      "grad_norm": 16.604900360107422,
      "learning_rate": 1e-06,
      "loss": 0.7405,
      "num_input_tokens_seen": 214454464,
      "step": 3829
    },
    {
      "epoch": 8.52783964365256,
      "loss": 0.6651185750961304,
      "loss_ce": 0.00020155691890977323,
      "loss_iou": 0.298828125,
      "loss_num": 0.01373291015625,
      "loss_xval": 0.6640625,
      "num_input_tokens_seen": 214454464,
      "step": 3829
    },
    {
      "epoch": 8.530066815144766,
      "grad_norm": 39.781227111816406,
      "learning_rate": 1e-06,
      "loss": 0.5991,
      "num_input_tokens_seen": 214511448,
      "step": 3830
    },
    {
      "epoch": 8.530066815144766,
      "loss": 0.589692234992981,
      "loss_ce": 0.0002146851911675185,
      "loss_iou": 0.24609375,
      "loss_num": 0.0191650390625,
      "loss_xval": 0.58984375,
      "num_input_tokens_seen": 214511448,
      "step": 3830
    },
    {
      "epoch": 8.53229398663697,
      "grad_norm": 26.230976104736328,
      "learning_rate": 1e-06,
      "loss": 0.517,
      "num_input_tokens_seen": 214565856,
      "step": 3831
    },
    {
      "epoch": 8.53229398663697,
      "loss": 0.389570951461792,
      "loss_ce": 0.00016667514864820987,
      "loss_iou": 0.1767578125,
      "loss_num": 0.007354736328125,
      "loss_xval": 0.388671875,
      "num_input_tokens_seen": 214565856,
      "step": 3831
    },
    {
      "epoch": 8.534521158129175,
      "grad_norm": 10.636022567749023,
      "learning_rate": 1e-06,
      "loss": 0.4324,
      "num_input_tokens_seen": 214623292,
      "step": 3832
    },
    {
      "epoch": 8.534521158129175,
      "loss": 0.4710007309913635,
      "loss_ce": 0.00017552266945131123,
      "loss_iou": 0.2138671875,
      "loss_num": 0.00836181640625,
      "loss_xval": 0.470703125,
      "num_input_tokens_seen": 214623292,
      "step": 3832
    },
    {
      "epoch": 8.53674832962138,
      "grad_norm": 20.143753051757812,
      "learning_rate": 1e-06,
      "loss": 0.5645,
      "num_input_tokens_seen": 214679760,
      "step": 3833
    },
    {
      "epoch": 8.53674832962138,
      "loss": 0.5450610518455505,
      "loss_ce": 0.00013918953482061625,
      "loss_iou": 0.2294921875,
      "loss_num": 0.01708984375,
      "loss_xval": 0.546875,
      "num_input_tokens_seen": 214679760,
      "step": 3833
    },
    {
      "epoch": 8.538975501113585,
      "grad_norm": 29.548900604248047,
      "learning_rate": 1e-06,
      "loss": 0.6879,
      "num_input_tokens_seen": 214734852,
      "step": 3834
    },
    {
      "epoch": 8.538975501113585,
      "loss": 0.43624764680862427,
      "loss_ce": 0.00015146093210205436,
      "loss_iou": 0.193359375,
      "loss_num": 0.00994873046875,
      "loss_xval": 0.435546875,
      "num_input_tokens_seen": 214734852,
      "step": 3834
    },
    {
      "epoch": 8.54120267260579,
      "grad_norm": 21.732810974121094,
      "learning_rate": 1e-06,
      "loss": 0.639,
      "num_input_tokens_seen": 214787916,
      "step": 3835
    },
    {
      "epoch": 8.54120267260579,
      "loss": 0.7973939180374146,
      "loss_ce": 0.0001527119311504066,
      "loss_iou": 0.3515625,
      "loss_num": 0.01904296875,
      "loss_xval": 0.796875,
      "num_input_tokens_seen": 214787916,
      "step": 3835
    },
    {
      "epoch": 8.543429844097995,
      "grad_norm": 10.797616958618164,
      "learning_rate": 1e-06,
      "loss": 0.4759,
      "num_input_tokens_seen": 214842424,
      "step": 3836
    },
    {
      "epoch": 8.543429844097995,
      "loss": 0.5132704377174377,
      "loss_ce": 0.00020891126769129187,
      "loss_iou": 0.197265625,
      "loss_num": 0.0235595703125,
      "loss_xval": 0.51171875,
      "num_input_tokens_seen": 214842424,
      "step": 3836
    },
    {
      "epoch": 8.5456570155902,
      "grad_norm": 21.894956588745117,
      "learning_rate": 1e-06,
      "loss": 0.5809,
      "num_input_tokens_seen": 214897440,
      "step": 3837
    },
    {
      "epoch": 8.5456570155902,
      "loss": 0.5928073525428772,
      "loss_ce": 0.0001559797237860039,
      "loss_iou": 0.255859375,
      "loss_num": 0.0162353515625,
      "loss_xval": 0.59375,
      "num_input_tokens_seen": 214897440,
      "step": 3837
    },
    {
      "epoch": 8.547884187082406,
      "grad_norm": 16.91735076904297,
      "learning_rate": 1e-06,
      "loss": 0.6461,
      "num_input_tokens_seen": 214951216,
      "step": 3838
    },
    {
      "epoch": 8.547884187082406,
      "loss": 0.7435756325721741,
      "loss_ce": 0.00016743276501074433,
      "loss_iou": 0.328125,
      "loss_num": 0.0174560546875,
      "loss_xval": 0.7421875,
      "num_input_tokens_seen": 214951216,
      "step": 3838
    },
    {
      "epoch": 8.550111358574611,
      "grad_norm": 16.6099796295166,
      "learning_rate": 1e-06,
      "loss": 0.7447,
      "num_input_tokens_seen": 215004404,
      "step": 3839
    },
    {
      "epoch": 8.550111358574611,
      "loss": 0.4547840356826782,
      "loss_ce": 0.00019417837029322982,
      "loss_iou": 0.2041015625,
      "loss_num": 0.0093994140625,
      "loss_xval": 0.455078125,
      "num_input_tokens_seen": 215004404,
      "step": 3839
    },
    {
      "epoch": 8.552338530066816,
      "grad_norm": 18.755542755126953,
      "learning_rate": 1e-06,
      "loss": 0.7003,
      "num_input_tokens_seen": 215061452,
      "step": 3840
    },
    {
      "epoch": 8.552338530066816,
      "loss": 0.7906391620635986,
      "loss_ce": 0.00023385511303786188,
      "loss_iou": 0.34375,
      "loss_num": 0.02099609375,
      "loss_xval": 0.7890625,
      "num_input_tokens_seen": 215061452,
      "step": 3840
    },
    {
      "epoch": 8.55456570155902,
      "grad_norm": 15.474801063537598,
      "learning_rate": 1e-06,
      "loss": 0.5902,
      "num_input_tokens_seen": 215119176,
      "step": 3841
    },
    {
      "epoch": 8.55456570155902,
      "loss": 0.6494476199150085,
      "loss_ce": 0.00015564775094389915,
      "loss_iou": 0.271484375,
      "loss_num": 0.0211181640625,
      "loss_xval": 0.6484375,
      "num_input_tokens_seen": 215119176,
      "step": 3841
    },
    {
      "epoch": 8.556792873051226,
      "grad_norm": 22.1204891204834,
      "learning_rate": 1e-06,
      "loss": 0.5713,
      "num_input_tokens_seen": 215174944,
      "step": 3842
    },
    {
      "epoch": 8.556792873051226,
      "loss": 0.5475232601165771,
      "loss_ce": 0.00015993315901141614,
      "loss_iou": 0.2392578125,
      "loss_num": 0.01397705078125,
      "loss_xval": 0.546875,
      "num_input_tokens_seen": 215174944,
      "step": 3842
    },
    {
      "epoch": 8.55902004454343,
      "grad_norm": 13.736021995544434,
      "learning_rate": 1e-06,
      "loss": 0.5063,
      "num_input_tokens_seen": 215230864,
      "step": 3843
    },
    {
      "epoch": 8.55902004454343,
      "loss": 0.4863549768924713,
      "loss_ce": 0.00014891059254296124,
      "loss_iou": 0.2138671875,
      "loss_num": 0.01171875,
      "loss_xval": 0.486328125,
      "num_input_tokens_seen": 215230864,
      "step": 3843
    },
    {
      "epoch": 8.561247216035635,
      "grad_norm": 22.418909072875977,
      "learning_rate": 1e-06,
      "loss": 0.7348,
      "num_input_tokens_seen": 215289312,
      "step": 3844
    },
    {
      "epoch": 8.561247216035635,
      "loss": 0.7660500407218933,
      "loss_ce": 0.00018092009122483432,
      "loss_iou": 0.361328125,
      "loss_num": 0.00830078125,
      "loss_xval": 0.765625,
      "num_input_tokens_seen": 215289312,
      "step": 3844
    },
    {
      "epoch": 8.56347438752784,
      "grad_norm": 16.7957820892334,
      "learning_rate": 1e-06,
      "loss": 0.6157,
      "num_input_tokens_seen": 215346324,
      "step": 3845
    },
    {
      "epoch": 8.56347438752784,
      "loss": 0.4588056802749634,
      "loss_ce": 0.0001875289308372885,
      "loss_iou": 0.201171875,
      "loss_num": 0.01129150390625,
      "loss_xval": 0.458984375,
      "num_input_tokens_seen": 215346324,
      "step": 3845
    },
    {
      "epoch": 8.565701559020045,
      "grad_norm": 18.19455909729004,
      "learning_rate": 1e-06,
      "loss": 0.4958,
      "num_input_tokens_seen": 215402928,
      "step": 3846
    },
    {
      "epoch": 8.565701559020045,
      "loss": 0.4269152879714966,
      "loss_ce": 0.00015747133875265718,
      "loss_iou": 0.1884765625,
      "loss_num": 0.00994873046875,
      "loss_xval": 0.42578125,
      "num_input_tokens_seen": 215402928,
      "step": 3846
    },
    {
      "epoch": 8.56792873051225,
      "grad_norm": 21.939851760864258,
      "learning_rate": 1e-06,
      "loss": 0.649,
      "num_input_tokens_seen": 215458620,
      "step": 3847
    },
    {
      "epoch": 8.56792873051225,
      "loss": 0.5985796451568604,
      "loss_ce": 0.00019095309835392982,
      "loss_iou": 0.251953125,
      "loss_num": 0.018798828125,
      "loss_xval": 0.59765625,
      "num_input_tokens_seen": 215458620,
      "step": 3847
    },
    {
      "epoch": 8.570155902004455,
      "grad_norm": 17.403221130371094,
      "learning_rate": 1e-06,
      "loss": 0.5387,
      "num_input_tokens_seen": 215515404,
      "step": 3848
    },
    {
      "epoch": 8.570155902004455,
      "loss": 0.6373581886291504,
      "loss_ce": 0.0001511847076471895,
      "loss_iou": 0.283203125,
      "loss_num": 0.0146484375,
      "loss_xval": 0.63671875,
      "num_input_tokens_seen": 215515404,
      "step": 3848
    },
    {
      "epoch": 8.57238307349666,
      "grad_norm": 17.169862747192383,
      "learning_rate": 1e-06,
      "loss": 0.5251,
      "num_input_tokens_seen": 215571552,
      "step": 3849
    },
    {
      "epoch": 8.57238307349666,
      "loss": 0.4762542247772217,
      "loss_ce": 0.0001799672027118504,
      "loss_iou": 0.2119140625,
      "loss_num": 0.01043701171875,
      "loss_xval": 0.4765625,
      "num_input_tokens_seen": 215571552,
      "step": 3849
    },
    {
      "epoch": 8.574610244988865,
      "grad_norm": 15.021171569824219,
      "learning_rate": 1e-06,
      "loss": 0.737,
      "num_input_tokens_seen": 215627964,
      "step": 3850
    },
    {
      "epoch": 8.574610244988865,
      "loss": 0.8246517777442932,
      "loss_ce": 0.00018888208433054388,
      "loss_iou": 0.328125,
      "loss_num": 0.033203125,
      "loss_xval": 0.82421875,
      "num_input_tokens_seen": 215627964,
      "step": 3850
    },
    {
      "epoch": 8.57683741648107,
      "grad_norm": 25.07132339477539,
      "learning_rate": 1e-06,
      "loss": 0.6764,
      "num_input_tokens_seen": 215683576,
      "step": 3851
    },
    {
      "epoch": 8.57683741648107,
      "loss": 0.5619097352027893,
      "loss_ce": 0.00014213821850717068,
      "loss_iou": 0.25,
      "loss_num": 0.01202392578125,
      "loss_xval": 0.5625,
      "num_input_tokens_seen": 215683576,
      "step": 3851
    },
    {
      "epoch": 8.579064587973274,
      "grad_norm": 15.985095024108887,
      "learning_rate": 1e-06,
      "loss": 0.5649,
      "num_input_tokens_seen": 215740424,
      "step": 3852
    },
    {
      "epoch": 8.579064587973274,
      "loss": 0.5519498586654663,
      "loss_ce": 0.000802424328867346,
      "loss_iou": 0.2275390625,
      "loss_num": 0.0189208984375,
      "loss_xval": 0.55078125,
      "num_input_tokens_seen": 215740424,
      "step": 3852
    },
    {
      "epoch": 8.58129175946548,
      "grad_norm": 17.084421157836914,
      "learning_rate": 1e-06,
      "loss": 0.4632,
      "num_input_tokens_seen": 215797432,
      "step": 3853
    },
    {
      "epoch": 8.58129175946548,
      "loss": 0.45524704456329346,
      "loss_ce": 0.00016892084386199713,
      "loss_iou": 0.2060546875,
      "loss_num": 0.0087890625,
      "loss_xval": 0.455078125,
      "num_input_tokens_seen": 215797432,
      "step": 3853
    },
    {
      "epoch": 8.583518930957684,
      "grad_norm": 25.7215633392334,
      "learning_rate": 1e-06,
      "loss": 0.827,
      "num_input_tokens_seen": 215855860,
      "step": 3854
    },
    {
      "epoch": 8.583518930957684,
      "loss": 0.8243607878684998,
      "loss_ce": 0.0001420074113411829,
      "loss_iou": 0.322265625,
      "loss_num": 0.035888671875,
      "loss_xval": 0.82421875,
      "num_input_tokens_seen": 215855860,
      "step": 3854
    },
    {
      "epoch": 8.585746102449889,
      "grad_norm": 20.765113830566406,
      "learning_rate": 1e-06,
      "loss": 0.5694,
      "num_input_tokens_seen": 215913020,
      "step": 3855
    },
    {
      "epoch": 8.585746102449889,
      "loss": 0.4681831896305084,
      "loss_ce": 0.00016561683150939643,
      "loss_iou": 0.201171875,
      "loss_num": 0.01318359375,
      "loss_xval": 0.46875,
      "num_input_tokens_seen": 215913020,
      "step": 3855
    },
    {
      "epoch": 8.587973273942094,
      "grad_norm": 22.44451141357422,
      "learning_rate": 1e-06,
      "loss": 0.5467,
      "num_input_tokens_seen": 215968572,
      "step": 3856
    },
    {
      "epoch": 8.587973273942094,
      "loss": 0.5845038890838623,
      "loss_ce": 0.00015330014866776764,
      "loss_iou": 0.26171875,
      "loss_num": 0.01251220703125,
      "loss_xval": 0.5859375,
      "num_input_tokens_seen": 215968572,
      "step": 3856
    },
    {
      "epoch": 8.590200445434299,
      "grad_norm": 20.8133487701416,
      "learning_rate": 1e-06,
      "loss": 0.5643,
      "num_input_tokens_seen": 216022120,
      "step": 3857
    },
    {
      "epoch": 8.590200445434299,
      "loss": 0.5692482590675354,
      "loss_ce": 0.00015648285625502467,
      "loss_iou": 0.248046875,
      "loss_num": 0.0145263671875,
      "loss_xval": 0.5703125,
      "num_input_tokens_seen": 216022120,
      "step": 3857
    },
    {
      "epoch": 8.592427616926503,
      "grad_norm": 15.802310943603516,
      "learning_rate": 1e-06,
      "loss": 0.7726,
      "num_input_tokens_seen": 216077540,
      "step": 3858
    },
    {
      "epoch": 8.592427616926503,
      "loss": 0.732232391834259,
      "loss_ce": 0.00017669174121692777,
      "loss_iou": 0.30859375,
      "loss_num": 0.0224609375,
      "loss_xval": 0.73046875,
      "num_input_tokens_seen": 216077540,
      "step": 3858
    },
    {
      "epoch": 8.594654788418708,
      "grad_norm": 21.7590389251709,
      "learning_rate": 1e-06,
      "loss": 0.4892,
      "num_input_tokens_seen": 216132424,
      "step": 3859
    },
    {
      "epoch": 8.594654788418708,
      "loss": 0.3683167099952698,
      "loss_ce": 0.0001526352425571531,
      "loss_iou": 0.1318359375,
      "loss_num": 0.0206298828125,
      "loss_xval": 0.3671875,
      "num_input_tokens_seen": 216132424,
      "step": 3859
    },
    {
      "epoch": 8.596881959910913,
      "grad_norm": 29.03900718688965,
      "learning_rate": 1e-06,
      "loss": 0.7531,
      "num_input_tokens_seen": 216185120,
      "step": 3860
    },
    {
      "epoch": 8.596881959910913,
      "loss": 0.7123351097106934,
      "loss_ce": 0.0001769287046045065,
      "loss_iou": 0.3203125,
      "loss_num": 0.01458740234375,
      "loss_xval": 0.7109375,
      "num_input_tokens_seen": 216185120,
      "step": 3860
    },
    {
      "epoch": 8.599109131403118,
      "grad_norm": 14.911413192749023,
      "learning_rate": 1e-06,
      "loss": 0.4934,
      "num_input_tokens_seen": 216239524,
      "step": 3861
    },
    {
      "epoch": 8.599109131403118,
      "loss": 0.5576772689819336,
      "loss_ce": 0.00018213198927696794,
      "loss_iou": 0.25390625,
      "loss_num": 0.00982666015625,
      "loss_xval": 0.55859375,
      "num_input_tokens_seen": 216239524,
      "step": 3861
    },
    {
      "epoch": 8.601336302895323,
      "grad_norm": 23.764856338500977,
      "learning_rate": 1e-06,
      "loss": 0.535,
      "num_input_tokens_seen": 216295068,
      "step": 3862
    },
    {
      "epoch": 8.601336302895323,
      "loss": 0.6164814233779907,
      "loss_ce": 0.0001483721862314269,
      "loss_iou": 0.27734375,
      "loss_num": 0.01263427734375,
      "loss_xval": 0.6171875,
      "num_input_tokens_seen": 216295068,
      "step": 3862
    },
    {
      "epoch": 8.603563474387528,
      "grad_norm": 43.42782211303711,
      "learning_rate": 1e-06,
      "loss": 0.8414,
      "num_input_tokens_seen": 216349744,
      "step": 3863
    },
    {
      "epoch": 8.603563474387528,
      "loss": 0.4933130145072937,
      "loss_ce": 0.00014895344793330878,
      "loss_iou": 0.224609375,
      "loss_num": 0.00872802734375,
      "loss_xval": 0.4921875,
      "num_input_tokens_seen": 216349744,
      "step": 3863
    },
    {
      "epoch": 8.605790645879733,
      "grad_norm": 20.517236709594727,
      "learning_rate": 1e-06,
      "loss": 0.6078,
      "num_input_tokens_seen": 216407720,
      "step": 3864
    },
    {
      "epoch": 8.605790645879733,
      "loss": 0.5921906232833862,
      "loss_ce": 0.00014958585961721838,
      "loss_iou": 0.26171875,
      "loss_num": 0.01373291015625,
      "loss_xval": 0.59375,
      "num_input_tokens_seen": 216407720,
      "step": 3864
    },
    {
      "epoch": 8.608017817371937,
      "grad_norm": 11.897843360900879,
      "learning_rate": 1e-06,
      "loss": 0.3836,
      "num_input_tokens_seen": 216466340,
      "step": 3865
    },
    {
      "epoch": 8.608017817371937,
      "loss": 0.3420777916908264,
      "loss_ce": 0.00015885230095591396,
      "loss_iou": 0.1552734375,
      "loss_num": 0.006317138671875,
      "loss_xval": 0.341796875,
      "num_input_tokens_seen": 216466340,
      "step": 3865
    },
    {
      "epoch": 8.610244988864142,
      "grad_norm": 20.564950942993164,
      "learning_rate": 1e-06,
      "loss": 0.6276,
      "num_input_tokens_seen": 216521392,
      "step": 3866
    },
    {
      "epoch": 8.610244988864142,
      "loss": 0.4767248034477234,
      "loss_ce": 0.0001622969430172816,
      "loss_iou": 0.2080078125,
      "loss_num": 0.01226806640625,
      "loss_xval": 0.4765625,
      "num_input_tokens_seen": 216521392,
      "step": 3866
    },
    {
      "epoch": 8.612472160356347,
      "grad_norm": 22.76060676574707,
      "learning_rate": 1e-06,
      "loss": 0.9585,
      "num_input_tokens_seen": 216572244,
      "step": 3867
    },
    {
      "epoch": 8.612472160356347,
      "loss": 1.402675747871399,
      "loss_ce": 0.00033200866892002523,
      "loss_iou": 0.5546875,
      "loss_num": 0.05810546875,
      "loss_xval": 1.40625,
      "num_input_tokens_seen": 216572244,
      "step": 3867
    },
    {
      "epoch": 8.614699331848552,
      "grad_norm": 17.371736526489258,
      "learning_rate": 1e-06,
      "loss": 0.5416,
      "num_input_tokens_seen": 216630272,
      "step": 3868
    },
    {
      "epoch": 8.614699331848552,
      "loss": 0.49856650829315186,
      "loss_ce": 0.000153428060002625,
      "loss_iou": 0.2216796875,
      "loss_num": 0.01092529296875,
      "loss_xval": 0.498046875,
      "num_input_tokens_seen": 216630272,
      "step": 3868
    },
    {
      "epoch": 8.616926503340757,
      "grad_norm": 18.579317092895508,
      "learning_rate": 1e-06,
      "loss": 0.5887,
      "num_input_tokens_seen": 216686540,
      "step": 3869
    },
    {
      "epoch": 8.616926503340757,
      "loss": 0.5541001558303833,
      "loss_ce": 0.00014507281593978405,
      "loss_iou": 0.2138671875,
      "loss_num": 0.0252685546875,
      "loss_xval": 0.5546875,
      "num_input_tokens_seen": 216686540,
      "step": 3869
    },
    {
      "epoch": 8.619153674832962,
      "grad_norm": 17.833322525024414,
      "learning_rate": 1e-06,
      "loss": 0.5787,
      "num_input_tokens_seen": 216742544,
      "step": 3870
    },
    {
      "epoch": 8.619153674832962,
      "loss": 0.5232726335525513,
      "loss_ce": 0.00020131460041739047,
      "loss_iou": 0.23046875,
      "loss_num": 0.01263427734375,
      "loss_xval": 0.5234375,
      "num_input_tokens_seen": 216742544,
      "step": 3870
    },
    {
      "epoch": 8.621380846325167,
      "grad_norm": 16.937902450561523,
      "learning_rate": 1e-06,
      "loss": 0.6145,
      "num_input_tokens_seen": 216798684,
      "step": 3871
    },
    {
      "epoch": 8.621380846325167,
      "loss": 0.4764711260795593,
      "loss_ce": 0.000152770007844083,
      "loss_iou": 0.203125,
      "loss_num": 0.01397705078125,
      "loss_xval": 0.4765625,
      "num_input_tokens_seen": 216798684,
      "step": 3871
    },
    {
      "epoch": 8.623608017817372,
      "grad_norm": 15.312004089355469,
      "learning_rate": 1e-06,
      "loss": 0.5381,
      "num_input_tokens_seen": 216852448,
      "step": 3872
    },
    {
      "epoch": 8.623608017817372,
      "loss": 0.47357112169265747,
      "loss_ce": 0.00018245888350065798,
      "loss_iou": 0.2119140625,
      "loss_num": 0.00982666015625,
      "loss_xval": 0.47265625,
      "num_input_tokens_seen": 216852448,
      "step": 3872
    },
    {
      "epoch": 8.625835189309576,
      "grad_norm": 19.796167373657227,
      "learning_rate": 1e-06,
      "loss": 0.5406,
      "num_input_tokens_seen": 216906968,
      "step": 3873
    },
    {
      "epoch": 8.625835189309576,
      "loss": 0.42670130729675293,
      "loss_ce": 0.00018764834385365248,
      "loss_iou": 0.19921875,
      "loss_num": 0.0057373046875,
      "loss_xval": 0.42578125,
      "num_input_tokens_seen": 216906968,
      "step": 3873
    },
    {
      "epoch": 8.628062360801781,
      "grad_norm": 17.857717514038086,
      "learning_rate": 1e-06,
      "loss": 0.8762,
      "num_input_tokens_seen": 216959064,
      "step": 3874
    },
    {
      "epoch": 8.628062360801781,
      "loss": 0.8671058416366577,
      "loss_ce": 0.00016246383893303573,
      "loss_iou": 0.40234375,
      "loss_num": 0.01300048828125,
      "loss_xval": 0.8671875,
      "num_input_tokens_seen": 216959064,
      "step": 3874
    },
    {
      "epoch": 8.630289532293986,
      "grad_norm": 26.22123908996582,
      "learning_rate": 1e-06,
      "loss": 0.571,
      "num_input_tokens_seen": 217016020,
      "step": 3875
    },
    {
      "epoch": 8.630289532293986,
      "loss": 0.6762186288833618,
      "loss_ce": 0.0001932468730956316,
      "loss_iou": 0.310546875,
      "loss_num": 0.0108642578125,
      "loss_xval": 0.67578125,
      "num_input_tokens_seen": 217016020,
      "step": 3875
    },
    {
      "epoch": 8.632516703786191,
      "grad_norm": 16.814098358154297,
      "learning_rate": 1e-06,
      "loss": 0.5477,
      "num_input_tokens_seen": 217069532,
      "step": 3876
    },
    {
      "epoch": 8.632516703786191,
      "loss": 0.3917490839958191,
      "loss_ce": 0.00014753625146113336,
      "loss_iou": 0.1708984375,
      "loss_num": 0.00994873046875,
      "loss_xval": 0.390625,
      "num_input_tokens_seen": 217069532,
      "step": 3876
    },
    {
      "epoch": 8.634743875278396,
      "grad_norm": 14.337661743164062,
      "learning_rate": 1e-06,
      "loss": 0.4624,
      "num_input_tokens_seen": 217128076,
      "step": 3877
    },
    {
      "epoch": 8.634743875278396,
      "loss": 0.49962735176086426,
      "loss_ce": 0.00023766841331962496,
      "loss_iou": 0.2236328125,
      "loss_num": 0.0107421875,
      "loss_xval": 0.5,
      "num_input_tokens_seen": 217128076,
      "step": 3877
    },
    {
      "epoch": 8.6369710467706,
      "grad_norm": 19.950851440429688,
      "learning_rate": 1e-06,
      "loss": 0.4853,
      "num_input_tokens_seen": 217187092,
      "step": 3878
    },
    {
      "epoch": 8.6369710467706,
      "loss": 0.4780520796775818,
      "loss_ce": 0.0001620454277144745,
      "loss_iou": 0.197265625,
      "loss_num": 0.0167236328125,
      "loss_xval": 0.478515625,
      "num_input_tokens_seen": 217187092,
      "step": 3878
    },
    {
      "epoch": 8.639198218262806,
      "grad_norm": 16.780813217163086,
      "learning_rate": 1e-06,
      "loss": 0.6026,
      "num_input_tokens_seen": 217244100,
      "step": 3879
    },
    {
      "epoch": 8.639198218262806,
      "loss": 0.7703104019165039,
      "loss_ce": 0.00029087584698572755,
      "loss_iou": 0.314453125,
      "loss_num": 0.02783203125,
      "loss_xval": 0.76953125,
      "num_input_tokens_seen": 217244100,
      "step": 3879
    },
    {
      "epoch": 8.64142538975501,
      "grad_norm": 20.776565551757812,
      "learning_rate": 1e-06,
      "loss": 0.5547,
      "num_input_tokens_seen": 217300800,
      "step": 3880
    },
    {
      "epoch": 8.64142538975501,
      "loss": 0.5758423805236816,
      "loss_ce": 0.0001587883016327396,
      "loss_iou": 0.244140625,
      "loss_num": 0.017578125,
      "loss_xval": 0.57421875,
      "num_input_tokens_seen": 217300800,
      "step": 3880
    },
    {
      "epoch": 8.643652561247215,
      "grad_norm": 16.412261962890625,
      "learning_rate": 1e-06,
      "loss": 0.5354,
      "num_input_tokens_seen": 217357076,
      "step": 3881
    },
    {
      "epoch": 8.643652561247215,
      "loss": 0.48521143198013306,
      "loss_ce": 0.00047024147352203727,
      "loss_iou": 0.2080078125,
      "loss_num": 0.013671875,
      "loss_xval": 0.484375,
      "num_input_tokens_seen": 217357076,
      "step": 3881
    },
    {
      "epoch": 8.64587973273942,
      "grad_norm": 40.7680778503418,
      "learning_rate": 1e-06,
      "loss": 0.7236,
      "num_input_tokens_seen": 217413124,
      "step": 3882
    },
    {
      "epoch": 8.64587973273942,
      "loss": 0.7792381644248962,
      "loss_ce": 0.00018543524492997676,
      "loss_iou": 0.341796875,
      "loss_num": 0.019287109375,
      "loss_xval": 0.77734375,
      "num_input_tokens_seen": 217413124,
      "step": 3882
    },
    {
      "epoch": 8.648106904231625,
      "grad_norm": 29.648727416992188,
      "learning_rate": 1e-06,
      "loss": 0.6271,
      "num_input_tokens_seen": 217468736,
      "step": 3883
    },
    {
      "epoch": 8.648106904231625,
      "loss": 0.6764873266220093,
      "loss_ce": 0.00027885870076715946,
      "loss_iou": 0.27734375,
      "loss_num": 0.024169921875,
      "loss_xval": 0.67578125,
      "num_input_tokens_seen": 217468736,
      "step": 3883
    },
    {
      "epoch": 8.65033407572383,
      "grad_norm": 16.55933380126953,
      "learning_rate": 1e-06,
      "loss": 0.6713,
      "num_input_tokens_seen": 217524312,
      "step": 3884
    },
    {
      "epoch": 8.65033407572383,
      "loss": 0.46768736839294434,
      "loss_ce": 0.00015806582814548165,
      "loss_iou": 0.201171875,
      "loss_num": 0.01300048828125,
      "loss_xval": 0.466796875,
      "num_input_tokens_seen": 217524312,
      "step": 3884
    },
    {
      "epoch": 8.652561247216035,
      "grad_norm": 24.027820587158203,
      "learning_rate": 1e-06,
      "loss": 0.6617,
      "num_input_tokens_seen": 217582960,
      "step": 3885
    },
    {
      "epoch": 8.652561247216035,
      "loss": 0.5760955810546875,
      "loss_ce": 0.00016780085570644587,
      "loss_iou": 0.251953125,
      "loss_num": 0.01446533203125,
      "loss_xval": 0.57421875,
      "num_input_tokens_seen": 217582960,
      "step": 3885
    },
    {
      "epoch": 8.654788418708241,
      "grad_norm": 12.898533821105957,
      "learning_rate": 1e-06,
      "loss": 0.4816,
      "num_input_tokens_seen": 217640700,
      "step": 3886
    },
    {
      "epoch": 8.654788418708241,
      "loss": 0.39870375394821167,
      "loss_ce": 0.00014417324564419687,
      "loss_iou": 0.169921875,
      "loss_num": 0.01171875,
      "loss_xval": 0.3984375,
      "num_input_tokens_seen": 217640700,
      "step": 3886
    },
    {
      "epoch": 8.657015590200446,
      "grad_norm": 24.518152236938477,
      "learning_rate": 1e-06,
      "loss": 0.7044,
      "num_input_tokens_seen": 217698148,
      "step": 3887
    },
    {
      "epoch": 8.657015590200446,
      "loss": 0.8448777198791504,
      "loss_ce": 0.0006394129595719278,
      "loss_iou": 0.357421875,
      "loss_num": 0.0257568359375,
      "loss_xval": 0.84375,
      "num_input_tokens_seen": 217698148,
      "step": 3887
    },
    {
      "epoch": 8.659242761692651,
      "grad_norm": 20.064247131347656,
      "learning_rate": 1e-06,
      "loss": 0.5771,
      "num_input_tokens_seen": 217755788,
      "step": 3888
    },
    {
      "epoch": 8.659242761692651,
      "loss": 0.602916955947876,
      "loss_ce": 0.00013376196147873998,
      "loss_iou": 0.26171875,
      "loss_num": 0.01611328125,
      "loss_xval": 0.6015625,
      "num_input_tokens_seen": 217755788,
      "step": 3888
    },
    {
      "epoch": 8.661469933184856,
      "grad_norm": 25.421236038208008,
      "learning_rate": 1e-06,
      "loss": 0.4772,
      "num_input_tokens_seen": 217813576,
      "step": 3889
    },
    {
      "epoch": 8.661469933184856,
      "loss": 0.5371390581130981,
      "loss_ce": 0.00015175854787230492,
      "loss_iou": 0.2470703125,
      "loss_num": 0.0086669921875,
      "loss_xval": 0.53515625,
      "num_input_tokens_seen": 217813576,
      "step": 3889
    },
    {
      "epoch": 8.66369710467706,
      "grad_norm": 23.681997299194336,
      "learning_rate": 1e-06,
      "loss": 0.6563,
      "num_input_tokens_seen": 217868180,
      "step": 3890
    },
    {
      "epoch": 8.66369710467706,
      "loss": 0.8451223373413086,
      "loss_ce": 0.00015169865218922496,
      "loss_iou": 0.36328125,
      "loss_num": 0.0238037109375,
      "loss_xval": 0.84375,
      "num_input_tokens_seen": 217868180,
      "step": 3890
    },
    {
      "epoch": 8.665924276169266,
      "grad_norm": 19.213150024414062,
      "learning_rate": 1e-06,
      "loss": 0.7518,
      "num_input_tokens_seen": 217926136,
      "step": 3891
    },
    {
      "epoch": 8.665924276169266,
      "loss": 0.5428782105445862,
      "loss_ce": 0.00015362550038844347,
      "loss_iou": 0.23046875,
      "loss_num": 0.0166015625,
      "loss_xval": 0.54296875,
      "num_input_tokens_seen": 217926136,
      "step": 3891
    },
    {
      "epoch": 8.66815144766147,
      "grad_norm": 22.411832809448242,
      "learning_rate": 1e-06,
      "loss": 0.6165,
      "num_input_tokens_seen": 217983252,
      "step": 3892
    },
    {
      "epoch": 8.66815144766147,
      "loss": 0.7387982606887817,
      "loss_ce": 0.0001508084242232144,
      "loss_iou": 0.3203125,
      "loss_num": 0.01953125,
      "loss_xval": 0.73828125,
      "num_input_tokens_seen": 217983252,
      "step": 3892
    },
    {
      "epoch": 8.670378619153675,
      "grad_norm": 22.501161575317383,
      "learning_rate": 1e-06,
      "loss": 0.5534,
      "num_input_tokens_seen": 218039972,
      "step": 3893
    },
    {
      "epoch": 8.670378619153675,
      "loss": 0.3819955587387085,
      "loss_ce": 0.0002206652716267854,
      "loss_iou": 0.1591796875,
      "loss_num": 0.01275634765625,
      "loss_xval": 0.380859375,
      "num_input_tokens_seen": 218039972,
      "step": 3893
    },
    {
      "epoch": 8.67260579064588,
      "grad_norm": 27.5329532623291,
      "learning_rate": 1e-06,
      "loss": 0.6429,
      "num_input_tokens_seen": 218097220,
      "step": 3894
    },
    {
      "epoch": 8.67260579064588,
      "loss": 0.3702659010887146,
      "loss_ce": 0.0001486857217969373,
      "loss_iou": 0.1640625,
      "loss_num": 0.00860595703125,
      "loss_xval": 0.37109375,
      "num_input_tokens_seen": 218097220,
      "step": 3894
    },
    {
      "epoch": 8.674832962138085,
      "grad_norm": 13.485652923583984,
      "learning_rate": 1e-06,
      "loss": 0.432,
      "num_input_tokens_seen": 218154216,
      "step": 3895
    },
    {
      "epoch": 8.674832962138085,
      "loss": 0.40601831674575806,
      "loss_ce": 0.00013454348663799465,
      "loss_iou": 0.1640625,
      "loss_num": 0.0157470703125,
      "loss_xval": 0.40625,
      "num_input_tokens_seen": 218154216,
      "step": 3895
    },
    {
      "epoch": 8.67706013363029,
      "grad_norm": 26.522533416748047,
      "learning_rate": 1e-06,
      "loss": 0.5957,
      "num_input_tokens_seen": 218206464,
      "step": 3896
    },
    {
      "epoch": 8.67706013363029,
      "loss": 0.751849889755249,
      "loss_ce": 0.0001408641110174358,
      "loss_iou": 0.32421875,
      "loss_num": 0.0206298828125,
      "loss_xval": 0.75,
      "num_input_tokens_seen": 218206464,
      "step": 3896
    },
    {
      "epoch": 8.679287305122495,
      "grad_norm": 24.25346565246582,
      "learning_rate": 1e-06,
      "loss": 0.8235,
      "num_input_tokens_seen": 218263664,
      "step": 3897
    },
    {
      "epoch": 8.679287305122495,
      "loss": 0.8781342506408691,
      "loss_ce": 0.0002045307046500966,
      "loss_iou": 0.333984375,
      "loss_num": 0.041748046875,
      "loss_xval": 0.87890625,
      "num_input_tokens_seen": 218263664,
      "step": 3897
    },
    {
      "epoch": 8.6815144766147,
      "grad_norm": 17.56336212158203,
      "learning_rate": 1e-06,
      "loss": 0.7218,
      "num_input_tokens_seen": 218319228,
      "step": 3898
    },
    {
      "epoch": 8.6815144766147,
      "loss": 0.8000770807266235,
      "loss_ce": 0.0001502884115325287,
      "loss_iou": 0.361328125,
      "loss_num": 0.01507568359375,
      "loss_xval": 0.80078125,
      "num_input_tokens_seen": 218319228,
      "step": 3898
    },
    {
      "epoch": 8.683741648106905,
      "grad_norm": 21.46102523803711,
      "learning_rate": 1e-06,
      "loss": 0.5721,
      "num_input_tokens_seen": 218374208,
      "step": 3899
    },
    {
      "epoch": 8.683741648106905,
      "loss": 0.6461607813835144,
      "loss_ce": 0.00016468125977553427,
      "loss_iou": 0.287109375,
      "loss_num": 0.0146484375,
      "loss_xval": 0.64453125,
      "num_input_tokens_seen": 218374208,
      "step": 3899
    },
    {
      "epoch": 8.68596881959911,
      "grad_norm": 15.952383995056152,
      "learning_rate": 1e-06,
      "loss": 0.6203,
      "num_input_tokens_seen": 218432844,
      "step": 3900
    },
    {
      "epoch": 8.68596881959911,
      "loss": 0.5890233516693115,
      "loss_ce": 0.0001561893877806142,
      "loss_iou": 0.265625,
      "loss_num": 0.01165771484375,
      "loss_xval": 0.58984375,
      "num_input_tokens_seen": 218432844,
      "step": 3900
    },
    {
      "epoch": 8.688195991091314,
      "grad_norm": 20.294042587280273,
      "learning_rate": 1e-06,
      "loss": 0.5964,
      "num_input_tokens_seen": 218489864,
      "step": 3901
    },
    {
      "epoch": 8.688195991091314,
      "loss": 0.6337062120437622,
      "loss_ce": 0.00040536391315981746,
      "loss_iou": 0.28125,
      "loss_num": 0.013916015625,
      "loss_xval": 0.6328125,
      "num_input_tokens_seen": 218489864,
      "step": 3901
    },
    {
      "epoch": 8.690423162583519,
      "grad_norm": 14.53677749633789,
      "learning_rate": 1e-06,
      "loss": 0.6919,
      "num_input_tokens_seen": 218545668,
      "step": 3902
    },
    {
      "epoch": 8.690423162583519,
      "loss": 0.4630255401134491,
      "loss_ce": 0.00013489379489328712,
      "loss_iou": 0.2080078125,
      "loss_num": 0.0093994140625,
      "loss_xval": 0.462890625,
      "num_input_tokens_seen": 218545668,
      "step": 3902
    },
    {
      "epoch": 8.692650334075724,
      "grad_norm": 21.7016544342041,
      "learning_rate": 1e-06,
      "loss": 0.614,
      "num_input_tokens_seen": 218602996,
      "step": 3903
    },
    {
      "epoch": 8.692650334075724,
      "loss": 0.573477566242218,
      "loss_ce": 0.000479543989058584,
      "loss_iou": 0.2392578125,
      "loss_num": 0.0189208984375,
      "loss_xval": 0.57421875,
      "num_input_tokens_seen": 218602996,
      "step": 3903
    },
    {
      "epoch": 8.694877505567929,
      "grad_norm": 17.698266983032227,
      "learning_rate": 1e-06,
      "loss": 0.4877,
      "num_input_tokens_seen": 218659176,
      "step": 3904
    },
    {
      "epoch": 8.694877505567929,
      "loss": 0.47621312737464905,
      "loss_ce": 0.00013888333342038095,
      "loss_iou": 0.203125,
      "loss_num": 0.0137939453125,
      "loss_xval": 0.4765625,
      "num_input_tokens_seen": 218659176,
      "step": 3904
    },
    {
      "epoch": 8.697104677060134,
      "grad_norm": 21.86427116394043,
      "learning_rate": 1e-06,
      "loss": 0.5896,
      "num_input_tokens_seen": 218713336,
      "step": 3905
    },
    {
      "epoch": 8.697104677060134,
      "loss": 0.5707519054412842,
      "loss_ce": 0.0001952478487510234,
      "loss_iou": 0.259765625,
      "loss_num": 0.01007080078125,
      "loss_xval": 0.5703125,
      "num_input_tokens_seen": 218713336,
      "step": 3905
    },
    {
      "epoch": 8.699331848552339,
      "grad_norm": 21.313528060913086,
      "learning_rate": 1e-06,
      "loss": 0.5622,
      "num_input_tokens_seen": 218771932,
      "step": 3906
    },
    {
      "epoch": 8.699331848552339,
      "loss": 0.5350602865219116,
      "loss_ce": 0.00014820430078543723,
      "loss_iou": 0.23828125,
      "loss_num": 0.01177978515625,
      "loss_xval": 0.53515625,
      "num_input_tokens_seen": 218771932,
      "step": 3906
    },
    {
      "epoch": 8.701559020044543,
      "grad_norm": 25.534914016723633,
      "learning_rate": 1e-06,
      "loss": 0.4969,
      "num_input_tokens_seen": 218829644,
      "step": 3907
    },
    {
      "epoch": 8.701559020044543,
      "loss": 0.4723511338233948,
      "loss_ce": 0.00018313938926439732,
      "loss_iou": 0.21875,
      "loss_num": 0.00689697265625,
      "loss_xval": 0.47265625,
      "num_input_tokens_seen": 218829644,
      "step": 3907
    },
    {
      "epoch": 8.703786191536748,
      "grad_norm": 31.614242553710938,
      "learning_rate": 1e-06,
      "loss": 0.7103,
      "num_input_tokens_seen": 218884672,
      "step": 3908
    },
    {
      "epoch": 8.703786191536748,
      "loss": 0.6449835300445557,
      "loss_ce": 0.0002081613929476589,
      "loss_iou": 0.27734375,
      "loss_num": 0.0177001953125,
      "loss_xval": 0.64453125,
      "num_input_tokens_seen": 218884672,
      "step": 3908
    },
    {
      "epoch": 8.706013363028953,
      "grad_norm": 15.506339073181152,
      "learning_rate": 1e-06,
      "loss": 0.5959,
      "num_input_tokens_seen": 218938064,
      "step": 3909
    },
    {
      "epoch": 8.706013363028953,
      "loss": 0.729173481464386,
      "loss_ce": 0.00016962323570623994,
      "loss_iou": 0.283203125,
      "loss_num": 0.031982421875,
      "loss_xval": 0.73046875,
      "num_input_tokens_seen": 218938064,
      "step": 3909
    },
    {
      "epoch": 8.708240534521158,
      "grad_norm": 23.147897720336914,
      "learning_rate": 1e-06,
      "loss": 0.4788,
      "num_input_tokens_seen": 218995120,
      "step": 3910
    },
    {
      "epoch": 8.708240534521158,
      "loss": 0.3938792645931244,
      "loss_ce": 0.0002025177382165566,
      "loss_iou": 0.1787109375,
      "loss_num": 0.007080078125,
      "loss_xval": 0.39453125,
      "num_input_tokens_seen": 218995120,
      "step": 3910
    },
    {
      "epoch": 8.710467706013363,
      "grad_norm": 23.020849227905273,
      "learning_rate": 1e-06,
      "loss": 0.6817,
      "num_input_tokens_seen": 219050528,
      "step": 3911
    },
    {
      "epoch": 8.710467706013363,
      "loss": 0.6399117708206177,
      "loss_ce": 0.00026331457775086164,
      "loss_iou": 0.255859375,
      "loss_num": 0.02587890625,
      "loss_xval": 0.640625,
      "num_input_tokens_seen": 219050528,
      "step": 3911
    },
    {
      "epoch": 8.712694877505568,
      "grad_norm": 16.904033660888672,
      "learning_rate": 1e-06,
      "loss": 0.5594,
      "num_input_tokens_seen": 219105692,
      "step": 3912
    },
    {
      "epoch": 8.712694877505568,
      "loss": 0.7058773636817932,
      "loss_ce": 0.00018892009393312037,
      "loss_iou": 0.3046875,
      "loss_num": 0.01904296875,
      "loss_xval": 0.70703125,
      "num_input_tokens_seen": 219105692,
      "step": 3912
    },
    {
      "epoch": 8.714922048997773,
      "grad_norm": 15.886432647705078,
      "learning_rate": 1e-06,
      "loss": 0.5245,
      "num_input_tokens_seen": 219160744,
      "step": 3913
    },
    {
      "epoch": 8.714922048997773,
      "loss": 0.6087524890899658,
      "loss_ce": 0.00023196196707431227,
      "loss_iou": 0.2578125,
      "loss_num": 0.01904296875,
      "loss_xval": 0.609375,
      "num_input_tokens_seen": 219160744,
      "step": 3913
    },
    {
      "epoch": 8.717149220489977,
      "grad_norm": 14.26981258392334,
      "learning_rate": 1e-06,
      "loss": 0.5525,
      "num_input_tokens_seen": 219217212,
      "step": 3914
    },
    {
      "epoch": 8.717149220489977,
      "loss": 0.7599831819534302,
      "loss_ce": 0.00021756268688477576,
      "loss_iou": 0.279296875,
      "loss_num": 0.040283203125,
      "loss_xval": 0.7578125,
      "num_input_tokens_seen": 219217212,
      "step": 3914
    },
    {
      "epoch": 8.719376391982182,
      "grad_norm": 15.31400203704834,
      "learning_rate": 1e-06,
      "loss": 0.6814,
      "num_input_tokens_seen": 219275172,
      "step": 3915
    },
    {
      "epoch": 8.719376391982182,
      "loss": 0.5874415040016174,
      "loss_ce": 0.00016121604130603373,
      "loss_iou": 0.236328125,
      "loss_num": 0.022705078125,
      "loss_xval": 0.5859375,
      "num_input_tokens_seen": 219275172,
      "step": 3915
    },
    {
      "epoch": 8.721603563474387,
      "grad_norm": 13.000571250915527,
      "learning_rate": 1e-06,
      "loss": 0.43,
      "num_input_tokens_seen": 219329864,
      "step": 3916
    },
    {
      "epoch": 8.721603563474387,
      "loss": 0.35176119208335876,
      "loss_ce": 0.00019871644326485693,
      "loss_iou": 0.158203125,
      "loss_num": 0.007232666015625,
      "loss_xval": 0.3515625,
      "num_input_tokens_seen": 219329864,
      "step": 3916
    },
    {
      "epoch": 8.723830734966592,
      "grad_norm": 24.805957794189453,
      "learning_rate": 1e-06,
      "loss": 0.6412,
      "num_input_tokens_seen": 219387476,
      "step": 3917
    },
    {
      "epoch": 8.723830734966592,
      "loss": 0.6872168183326721,
      "loss_ce": 0.00020506742293946445,
      "loss_iou": 0.26171875,
      "loss_num": 0.032958984375,
      "loss_xval": 0.6875,
      "num_input_tokens_seen": 219387476,
      "step": 3917
    },
    {
      "epoch": 8.726057906458797,
      "grad_norm": 17.83942413330078,
      "learning_rate": 1e-06,
      "loss": 0.6982,
      "num_input_tokens_seen": 219445780,
      "step": 3918
    },
    {
      "epoch": 8.726057906458797,
      "loss": 0.7072159051895142,
      "loss_ce": 0.00018466576875653118,
      "loss_iou": 0.29296875,
      "loss_num": 0.02392578125,
      "loss_xval": 0.70703125,
      "num_input_tokens_seen": 219445780,
      "step": 3918
    },
    {
      "epoch": 8.728285077951002,
      "grad_norm": 18.335899353027344,
      "learning_rate": 1e-06,
      "loss": 0.554,
      "num_input_tokens_seen": 219501992,
      "step": 3919
    },
    {
      "epoch": 8.728285077951002,
      "loss": 0.6365430355072021,
      "loss_ce": 0.0001904442033264786,
      "loss_iou": 0.279296875,
      "loss_num": 0.015380859375,
      "loss_xval": 0.63671875,
      "num_input_tokens_seen": 219501992,
      "step": 3919
    },
    {
      "epoch": 8.730512249443207,
      "grad_norm": 54.56409454345703,
      "learning_rate": 1e-06,
      "loss": 0.7071,
      "num_input_tokens_seen": 219557492,
      "step": 3920
    },
    {
      "epoch": 8.730512249443207,
      "loss": 0.7284532785415649,
      "loss_ce": 0.00018174288561567664,
      "loss_iou": 0.29296875,
      "loss_num": 0.0289306640625,
      "loss_xval": 0.7265625,
      "num_input_tokens_seen": 219557492,
      "step": 3920
    },
    {
      "epoch": 8.732739420935411,
      "grad_norm": 19.81359100341797,
      "learning_rate": 1e-06,
      "loss": 0.4295,
      "num_input_tokens_seen": 219612520,
      "step": 3921
    },
    {
      "epoch": 8.732739420935411,
      "loss": 0.35280388593673706,
      "loss_ce": 0.00014273943088483065,
      "loss_iou": 0.15234375,
      "loss_num": 0.009521484375,
      "loss_xval": 0.353515625,
      "num_input_tokens_seen": 219612520,
      "step": 3921
    },
    {
      "epoch": 8.734966592427616,
      "grad_norm": 20.431453704833984,
      "learning_rate": 1e-06,
      "loss": 0.6836,
      "num_input_tokens_seen": 219666936,
      "step": 3922
    },
    {
      "epoch": 8.734966592427616,
      "loss": 0.7043083906173706,
      "loss_ce": 0.00020681662135757506,
      "loss_iou": 0.29296875,
      "loss_num": 0.02392578125,
      "loss_xval": 0.703125,
      "num_input_tokens_seen": 219666936,
      "step": 3922
    },
    {
      "epoch": 8.737193763919821,
      "grad_norm": 74.53073120117188,
      "learning_rate": 1e-06,
      "loss": 0.6355,
      "num_input_tokens_seen": 219721856,
      "step": 3923
    },
    {
      "epoch": 8.737193763919821,
      "loss": 0.5503079891204834,
      "loss_ce": 0.00013712375948671252,
      "loss_iou": 0.2216796875,
      "loss_num": 0.0216064453125,
      "loss_xval": 0.55078125,
      "num_input_tokens_seen": 219721856,
      "step": 3923
    },
    {
      "epoch": 8.739420935412026,
      "grad_norm": 18.951021194458008,
      "learning_rate": 1e-06,
      "loss": 0.4582,
      "num_input_tokens_seen": 219777316,
      "step": 3924
    },
    {
      "epoch": 8.739420935412026,
      "loss": 0.4863456189632416,
      "loss_ce": 0.00026161997811868787,
      "loss_iou": 0.22265625,
      "loss_num": 0.00836181640625,
      "loss_xval": 0.486328125,
      "num_input_tokens_seen": 219777316,
      "step": 3924
    },
    {
      "epoch": 8.74164810690423,
      "grad_norm": 17.503719329833984,
      "learning_rate": 1e-06,
      "loss": 0.6413,
      "num_input_tokens_seen": 219834932,
      "step": 3925
    },
    {
      "epoch": 8.74164810690423,
      "loss": 0.6336848735809326,
      "loss_ce": 0.0001399153989041224,
      "loss_iou": 0.244140625,
      "loss_num": 0.0291748046875,
      "loss_xval": 0.6328125,
      "num_input_tokens_seen": 219834932,
      "step": 3925
    },
    {
      "epoch": 8.743875278396436,
      "grad_norm": 15.789238929748535,
      "learning_rate": 1e-06,
      "loss": 0.5391,
      "num_input_tokens_seen": 219891608,
      "step": 3926
    },
    {
      "epoch": 8.743875278396436,
      "loss": 0.559868574142456,
      "loss_ce": 0.00029824947705492377,
      "loss_iou": 0.236328125,
      "loss_num": 0.0172119140625,
      "loss_xval": 0.55859375,
      "num_input_tokens_seen": 219891608,
      "step": 3926
    },
    {
      "epoch": 8.74610244988864,
      "grad_norm": 22.618701934814453,
      "learning_rate": 1e-06,
      "loss": 0.6993,
      "num_input_tokens_seen": 219948044,
      "step": 3927
    },
    {
      "epoch": 8.74610244988864,
      "loss": 0.7226395606994629,
      "loss_ce": 0.00022747760522179306,
      "loss_iou": 0.3046875,
      "loss_num": 0.022216796875,
      "loss_xval": 0.72265625,
      "num_input_tokens_seen": 219948044,
      "step": 3927
    },
    {
      "epoch": 8.748329621380847,
      "grad_norm": 16.85779571533203,
      "learning_rate": 1e-06,
      "loss": 0.5143,
      "num_input_tokens_seen": 220002632,
      "step": 3928
    },
    {
      "epoch": 8.748329621380847,
      "loss": 0.6813238859176636,
      "loss_ce": 0.00017151121573988348,
      "loss_iou": 0.291015625,
      "loss_num": 0.0201416015625,
      "loss_xval": 0.6796875,
      "num_input_tokens_seen": 220002632,
      "step": 3928
    },
    {
      "epoch": 8.750556792873052,
      "grad_norm": 22.382169723510742,
      "learning_rate": 1e-06,
      "loss": 0.6557,
      "num_input_tokens_seen": 220058412,
      "step": 3929
    },
    {
      "epoch": 8.750556792873052,
      "loss": 0.7504002451896667,
      "loss_ce": 0.00015609902038704604,
      "loss_iou": 0.3203125,
      "loss_num": 0.02197265625,
      "loss_xval": 0.75,
      "num_input_tokens_seen": 220058412,
      "step": 3929
    },
    {
      "epoch": 8.752783964365257,
      "grad_norm": 15.235098838806152,
      "learning_rate": 1e-06,
      "loss": 0.6484,
      "num_input_tokens_seen": 220114060,
      "step": 3930
    },
    {
      "epoch": 8.752783964365257,
      "loss": 0.7623621225357056,
      "loss_ce": 0.00015506052295677364,
      "loss_iou": 0.337890625,
      "loss_num": 0.017333984375,
      "loss_xval": 0.76171875,
      "num_input_tokens_seen": 220114060,
      "step": 3930
    },
    {
      "epoch": 8.755011135857462,
      "grad_norm": 19.23377227783203,
      "learning_rate": 1e-06,
      "loss": 0.7264,
      "num_input_tokens_seen": 220169768,
      "step": 3931
    },
    {
      "epoch": 8.755011135857462,
      "loss": 0.7468116283416748,
      "loss_ce": 0.0007179292151704431,
      "loss_iou": 0.3203125,
      "loss_num": 0.0206298828125,
      "loss_xval": 0.74609375,
      "num_input_tokens_seen": 220169768,
      "step": 3931
    },
    {
      "epoch": 8.757238307349667,
      "grad_norm": 24.401901245117188,
      "learning_rate": 1e-06,
      "loss": 0.603,
      "num_input_tokens_seen": 220227336,
      "step": 3932
    },
    {
      "epoch": 8.757238307349667,
      "loss": 0.422521710395813,
      "loss_ce": 0.00015843103756196797,
      "loss_iou": 0.1865234375,
      "loss_num": 0.010009765625,
      "loss_xval": 0.421875,
      "num_input_tokens_seen": 220227336,
      "step": 3932
    },
    {
      "epoch": 8.759465478841872,
      "grad_norm": 24.561513900756836,
      "learning_rate": 1e-06,
      "loss": 0.5883,
      "num_input_tokens_seen": 220281280,
      "step": 3933
    },
    {
      "epoch": 8.759465478841872,
      "loss": 0.6444734930992126,
      "loss_ce": 0.00018640572670847178,
      "loss_iou": 0.291015625,
      "loss_num": 0.0126953125,
      "loss_xval": 0.64453125,
      "num_input_tokens_seen": 220281280,
      "step": 3933
    },
    {
      "epoch": 8.761692650334076,
      "grad_norm": 22.442155838012695,
      "learning_rate": 1e-06,
      "loss": 0.5317,
      "num_input_tokens_seen": 220337648,
      "step": 3934
    },
    {
      "epoch": 8.761692650334076,
      "loss": 0.662487268447876,
      "loss_ce": 0.00013376369315665215,
      "loss_iou": 0.287109375,
      "loss_num": 0.017333984375,
      "loss_xval": 0.6640625,
      "num_input_tokens_seen": 220337648,
      "step": 3934
    },
    {
      "epoch": 8.763919821826281,
      "grad_norm": 17.563997268676758,
      "learning_rate": 1e-06,
      "loss": 0.578,
      "num_input_tokens_seen": 220395500,
      "step": 3935
    },
    {
      "epoch": 8.763919821826281,
      "loss": 0.7258895635604858,
      "loss_ce": 0.0007919379277154803,
      "loss_iou": 0.30078125,
      "loss_num": 0.0244140625,
      "loss_xval": 0.7265625,
      "num_input_tokens_seen": 220395500,
      "step": 3935
    },
    {
      "epoch": 8.766146993318486,
      "grad_norm": 21.255687713623047,
      "learning_rate": 1e-06,
      "loss": 0.6404,
      "num_input_tokens_seen": 220450940,
      "step": 3936
    },
    {
      "epoch": 8.766146993318486,
      "loss": 0.6532711386680603,
      "loss_ce": 0.00019493838772177696,
      "loss_iou": 0.26171875,
      "loss_num": 0.0257568359375,
      "loss_xval": 0.65234375,
      "num_input_tokens_seen": 220450940,
      "step": 3936
    },
    {
      "epoch": 8.768374164810691,
      "grad_norm": 14.73167610168457,
      "learning_rate": 1e-06,
      "loss": 0.605,
      "num_input_tokens_seen": 220505768,
      "step": 3937
    },
    {
      "epoch": 8.768374164810691,
      "loss": 0.4992978870868683,
      "loss_ce": 0.00015238078776746988,
      "loss_iou": 0.2197265625,
      "loss_num": 0.01214599609375,
      "loss_xval": 0.5,
      "num_input_tokens_seen": 220505768,
      "step": 3937
    },
    {
      "epoch": 8.770601336302896,
      "grad_norm": 20.76053237915039,
      "learning_rate": 1e-06,
      "loss": 0.5365,
      "num_input_tokens_seen": 220562928,
      "step": 3938
    },
    {
      "epoch": 8.770601336302896,
      "loss": 0.5216740369796753,
      "loss_ce": 0.00018964534683618695,
      "loss_iou": 0.2373046875,
      "loss_num": 0.0093994140625,
      "loss_xval": 0.5234375,
      "num_input_tokens_seen": 220562928,
      "step": 3938
    },
    {
      "epoch": 8.7728285077951,
      "grad_norm": 16.428634643554688,
      "learning_rate": 1e-06,
      "loss": 0.6054,
      "num_input_tokens_seen": 220619840,
      "step": 3939
    },
    {
      "epoch": 8.7728285077951,
      "loss": 0.6325225830078125,
      "loss_ce": 0.00019834056729450822,
      "loss_iou": 0.2578125,
      "loss_num": 0.023681640625,
      "loss_xval": 0.6328125,
      "num_input_tokens_seen": 220619840,
      "step": 3939
    },
    {
      "epoch": 8.775055679287306,
      "grad_norm": 14.64647388458252,
      "learning_rate": 1e-06,
      "loss": 0.5719,
      "num_input_tokens_seen": 220674660,
      "step": 3940
    },
    {
      "epoch": 8.775055679287306,
      "loss": 0.6691886782646179,
      "loss_ce": 0.00024336397473234683,
      "loss_iou": 0.275390625,
      "loss_num": 0.023681640625,
      "loss_xval": 0.66796875,
      "num_input_tokens_seen": 220674660,
      "step": 3940
    },
    {
      "epoch": 8.77728285077951,
      "grad_norm": 12.952407836914062,
      "learning_rate": 1e-06,
      "loss": 0.4744,
      "num_input_tokens_seen": 220733140,
      "step": 3941
    },
    {
      "epoch": 8.77728285077951,
      "loss": 0.6572137475013733,
      "loss_ce": 0.00023132127535063773,
      "loss_iou": 0.28125,
      "loss_num": 0.019287109375,
      "loss_xval": 0.65625,
      "num_input_tokens_seen": 220733140,
      "step": 3941
    },
    {
      "epoch": 8.779510022271715,
      "grad_norm": 18.830053329467773,
      "learning_rate": 1e-06,
      "loss": 0.6262,
      "num_input_tokens_seen": 220789288,
      "step": 3942
    },
    {
      "epoch": 8.779510022271715,
      "loss": 0.6996396780014038,
      "loss_ce": 0.0001767998473951593,
      "loss_iou": 0.279296875,
      "loss_num": 0.0283203125,
      "loss_xval": 0.69921875,
      "num_input_tokens_seen": 220789288,
      "step": 3942
    },
    {
      "epoch": 8.78173719376392,
      "grad_norm": 24.526212692260742,
      "learning_rate": 1e-06,
      "loss": 0.7996,
      "num_input_tokens_seen": 220844188,
      "step": 3943
    },
    {
      "epoch": 8.78173719376392,
      "loss": 0.6259015798568726,
      "loss_ce": 0.0001691749203018844,
      "loss_iou": 0.283203125,
      "loss_num": 0.01190185546875,
      "loss_xval": 0.625,
      "num_input_tokens_seen": 220844188,
      "step": 3943
    },
    {
      "epoch": 8.783964365256125,
      "grad_norm": 17.90644073486328,
      "learning_rate": 1e-06,
      "loss": 0.5722,
      "num_input_tokens_seen": 220901768,
      "step": 3944
    },
    {
      "epoch": 8.783964365256125,
      "loss": 0.5738731622695923,
      "loss_ce": 0.00014269730309024453,
      "loss_iou": 0.26171875,
      "loss_num": 0.0098876953125,
      "loss_xval": 0.57421875,
      "num_input_tokens_seen": 220901768,
      "step": 3944
    },
    {
      "epoch": 8.78619153674833,
      "grad_norm": 40.18233871459961,
      "learning_rate": 1e-06,
      "loss": 0.7066,
      "num_input_tokens_seen": 220957076,
      "step": 3945
    },
    {
      "epoch": 8.78619153674833,
      "loss": 0.8366864919662476,
      "loss_ce": 0.0001997254294110462,
      "loss_iou": 0.359375,
      "loss_num": 0.0233154296875,
      "loss_xval": 0.8359375,
      "num_input_tokens_seen": 220957076,
      "step": 3945
    },
    {
      "epoch": 8.788418708240535,
      "grad_norm": 35.03602600097656,
      "learning_rate": 1e-06,
      "loss": 0.683,
      "num_input_tokens_seen": 221011744,
      "step": 3946
    },
    {
      "epoch": 8.788418708240535,
      "loss": 0.5396614074707031,
      "loss_ce": 0.00023272512771654874,
      "loss_iou": 0.224609375,
      "loss_num": 0.0179443359375,
      "loss_xval": 0.5390625,
      "num_input_tokens_seen": 221011744,
      "step": 3946
    },
    {
      "epoch": 8.79064587973274,
      "grad_norm": 19.676462173461914,
      "learning_rate": 1e-06,
      "loss": 0.6669,
      "num_input_tokens_seen": 221068528,
      "step": 3947
    },
    {
      "epoch": 8.79064587973274,
      "loss": 0.7179461717605591,
      "loss_ce": 0.00017279275925830007,
      "loss_iou": 0.291015625,
      "loss_num": 0.02685546875,
      "loss_xval": 0.71875,
      "num_input_tokens_seen": 221068528,
      "step": 3947
    },
    {
      "epoch": 8.792873051224944,
      "grad_norm": 26.63570785522461,
      "learning_rate": 1e-06,
      "loss": 0.5588,
      "num_input_tokens_seen": 221125112,
      "step": 3948
    },
    {
      "epoch": 8.792873051224944,
      "loss": 0.6211436986923218,
      "loss_ce": 0.0005382976960390806,
      "loss_iou": 0.259765625,
      "loss_num": 0.0201416015625,
      "loss_xval": 0.62109375,
      "num_input_tokens_seen": 221125112,
      "step": 3948
    },
    {
      "epoch": 8.79510022271715,
      "grad_norm": 17.29355239868164,
      "learning_rate": 1e-06,
      "loss": 0.5911,
      "num_input_tokens_seen": 221181924,
      "step": 3949
    },
    {
      "epoch": 8.79510022271715,
      "loss": 0.6676521897315979,
      "loss_ce": 0.0001717579725664109,
      "loss_iou": 0.298828125,
      "loss_num": 0.01373291015625,
      "loss_xval": 0.66796875,
      "num_input_tokens_seen": 221181924,
      "step": 3949
    },
    {
      "epoch": 8.797327394209354,
      "grad_norm": 20.77406120300293,
      "learning_rate": 1e-06,
      "loss": 0.6492,
      "num_input_tokens_seen": 221239948,
      "step": 3950
    },
    {
      "epoch": 8.797327394209354,
      "loss": 0.7762324213981628,
      "loss_ce": 0.00035347635275684297,
      "loss_iou": 0.310546875,
      "loss_num": 0.0311279296875,
      "loss_xval": 0.77734375,
      "num_input_tokens_seen": 221239948,
      "step": 3950
    },
    {
      "epoch": 8.799554565701559,
      "grad_norm": 25.013608932495117,
      "learning_rate": 1e-06,
      "loss": 0.7604,
      "num_input_tokens_seen": 221296624,
      "step": 3951
    },
    {
      "epoch": 8.799554565701559,
      "loss": 0.6159054040908813,
      "loss_ce": 0.0001827623782446608,
      "loss_iou": 0.263671875,
      "loss_num": 0.017578125,
      "loss_xval": 0.6171875,
      "num_input_tokens_seen": 221296624,
      "step": 3951
    },
    {
      "epoch": 8.801781737193764,
      "grad_norm": 22.501018524169922,
      "learning_rate": 1e-06,
      "loss": 0.6662,
      "num_input_tokens_seen": 221352544,
      "step": 3952
    },
    {
      "epoch": 8.801781737193764,
      "loss": 0.8238818645477295,
      "loss_ce": 0.0001514081668574363,
      "loss_iou": 0.341796875,
      "loss_num": 0.0277099609375,
      "loss_xval": 0.82421875,
      "num_input_tokens_seen": 221352544,
      "step": 3952
    },
    {
      "epoch": 8.804008908685969,
      "grad_norm": 16.182146072387695,
      "learning_rate": 1e-06,
      "loss": 0.7461,
      "num_input_tokens_seen": 221409016,
      "step": 3953
    },
    {
      "epoch": 8.804008908685969,
      "loss": 1.03929603099823,
      "loss_ce": 0.00023350719129666686,
      "loss_iou": 0.41015625,
      "loss_num": 0.04345703125,
      "loss_xval": 1.0390625,
      "num_input_tokens_seen": 221409016,
      "step": 3953
    },
    {
      "epoch": 8.806236080178174,
      "grad_norm": 20.251689910888672,
      "learning_rate": 1e-06,
      "loss": 0.625,
      "num_input_tokens_seen": 221464600,
      "step": 3954
    },
    {
      "epoch": 8.806236080178174,
      "loss": 0.6593621969223022,
      "loss_ce": 0.00018251534493174404,
      "loss_iou": 0.296875,
      "loss_num": 0.01324462890625,
      "loss_xval": 0.66015625,
      "num_input_tokens_seen": 221464600,
      "step": 3954
    },
    {
      "epoch": 8.808463251670378,
      "grad_norm": 22.65058708190918,
      "learning_rate": 1e-06,
      "loss": 0.4196,
      "num_input_tokens_seen": 221519316,
      "step": 3955
    },
    {
      "epoch": 8.808463251670378,
      "loss": 0.5017726421356201,
      "loss_ce": 0.0001857746101450175,
      "loss_iou": 0.205078125,
      "loss_num": 0.01806640625,
      "loss_xval": 0.5,
      "num_input_tokens_seen": 221519316,
      "step": 3955
    },
    {
      "epoch": 8.810690423162583,
      "grad_norm": 16.896793365478516,
      "learning_rate": 1e-06,
      "loss": 0.6443,
      "num_input_tokens_seen": 221574936,
      "step": 3956
    },
    {
      "epoch": 8.810690423162583,
      "loss": 0.4596463441848755,
      "loss_ce": 0.00017369160195812583,
      "loss_iou": 0.2001953125,
      "loss_num": 0.0120849609375,
      "loss_xval": 0.458984375,
      "num_input_tokens_seen": 221574936,
      "step": 3956
    },
    {
      "epoch": 8.812917594654788,
      "grad_norm": 19.71438217163086,
      "learning_rate": 1e-06,
      "loss": 0.589,
      "num_input_tokens_seen": 221629572,
      "step": 3957
    },
    {
      "epoch": 8.812917594654788,
      "loss": 0.5652260780334473,
      "loss_ce": 0.00016256351955235004,
      "loss_iou": 0.2421875,
      "loss_num": 0.01611328125,
      "loss_xval": 0.56640625,
      "num_input_tokens_seen": 221629572,
      "step": 3957
    },
    {
      "epoch": 8.815144766146993,
      "grad_norm": 14.786736488342285,
      "learning_rate": 1e-06,
      "loss": 0.4319,
      "num_input_tokens_seen": 221687656,
      "step": 3958
    },
    {
      "epoch": 8.815144766146993,
      "loss": 0.45927944779396057,
      "loss_ce": 0.00017301308980677277,
      "loss_iou": 0.20703125,
      "loss_num": 0.00921630859375,
      "loss_xval": 0.458984375,
      "num_input_tokens_seen": 221687656,
      "step": 3958
    },
    {
      "epoch": 8.817371937639198,
      "grad_norm": 17.631956100463867,
      "learning_rate": 1e-06,
      "loss": 0.6813,
      "num_input_tokens_seen": 221745792,
      "step": 3959
    },
    {
      "epoch": 8.817371937639198,
      "loss": 0.5624639987945557,
      "loss_ce": 0.0002081410784740001,
      "loss_iou": 0.2431640625,
      "loss_num": 0.01544189453125,
      "loss_xval": 0.5625,
      "num_input_tokens_seen": 221745792,
      "step": 3959
    },
    {
      "epoch": 8.819599109131403,
      "grad_norm": 22.315784454345703,
      "learning_rate": 1e-06,
      "loss": 0.7343,
      "num_input_tokens_seen": 221800120,
      "step": 3960
    },
    {
      "epoch": 8.819599109131403,
      "loss": 0.8773585557937622,
      "loss_ce": 0.00016129494179040194,
      "loss_iou": 0.37890625,
      "loss_num": 0.0238037109375,
      "loss_xval": 0.87890625,
      "num_input_tokens_seen": 221800120,
      "step": 3960
    },
    {
      "epoch": 8.821826280623608,
      "grad_norm": 14.803606986999512,
      "learning_rate": 1e-06,
      "loss": 0.6183,
      "num_input_tokens_seen": 221856540,
      "step": 3961
    },
    {
      "epoch": 8.821826280623608,
      "loss": 0.5055595636367798,
      "loss_ce": 0.00018847928731702268,
      "loss_iou": 0.2236328125,
      "loss_num": 0.0115966796875,
      "loss_xval": 0.50390625,
      "num_input_tokens_seen": 221856540,
      "step": 3961
    },
    {
      "epoch": 8.824053452115812,
      "grad_norm": 22.032180786132812,
      "learning_rate": 1e-06,
      "loss": 0.5228,
      "num_input_tokens_seen": 221908880,
      "step": 3962
    },
    {
      "epoch": 8.824053452115812,
      "loss": 0.551054835319519,
      "loss_ce": 0.00027354987105354667,
      "loss_iou": 0.24609375,
      "loss_num": 0.0115966796875,
      "loss_xval": 0.55078125,
      "num_input_tokens_seen": 221908880,
      "step": 3962
    },
    {
      "epoch": 8.826280623608017,
      "grad_norm": 64.0085678100586,
      "learning_rate": 1e-06,
      "loss": 0.8304,
      "num_input_tokens_seen": 221964260,
      "step": 3963
    },
    {
      "epoch": 8.826280623608017,
      "loss": 0.7420934438705444,
      "loss_ce": 0.00015010150673333555,
      "loss_iou": 0.314453125,
      "loss_num": 0.0225830078125,
      "loss_xval": 0.7421875,
      "num_input_tokens_seen": 221964260,
      "step": 3963
    },
    {
      "epoch": 8.828507795100222,
      "grad_norm": 16.832103729248047,
      "learning_rate": 1e-06,
      "loss": 0.8408,
      "num_input_tokens_seen": 222022236,
      "step": 3964
    },
    {
      "epoch": 8.828507795100222,
      "loss": 1.2050738334655762,
      "loss_ce": 0.00023982246057130396,
      "loss_iou": 0.466796875,
      "loss_num": 0.053955078125,
      "loss_xval": 1.203125,
      "num_input_tokens_seen": 222022236,
      "step": 3964
    },
    {
      "epoch": 8.830734966592427,
      "grad_norm": 20.819835662841797,
      "learning_rate": 1e-06,
      "loss": 0.7145,
      "num_input_tokens_seen": 222077780,
      "step": 3965
    },
    {
      "epoch": 8.830734966592427,
      "loss": 0.9205887317657471,
      "loss_ce": 0.00017864895926322788,
      "loss_iou": 0.37109375,
      "loss_num": 0.035400390625,
      "loss_xval": 0.921875,
      "num_input_tokens_seen": 222077780,
      "step": 3965
    },
    {
      "epoch": 8.832962138084632,
      "grad_norm": 18.16783332824707,
      "learning_rate": 1e-06,
      "loss": 0.4841,
      "num_input_tokens_seen": 222136840,
      "step": 3966
    },
    {
      "epoch": 8.832962138084632,
      "loss": 0.5124315023422241,
      "loss_ce": 0.00019395150593481958,
      "loss_iou": 0.2080078125,
      "loss_num": 0.01953125,
      "loss_xval": 0.51171875,
      "num_input_tokens_seen": 222136840,
      "step": 3966
    },
    {
      "epoch": 8.835189309576837,
      "grad_norm": 25.93811798095703,
      "learning_rate": 1e-06,
      "loss": 0.4811,
      "num_input_tokens_seen": 222191848,
      "step": 3967
    },
    {
      "epoch": 8.835189309576837,
      "loss": 0.5905265212059021,
      "loss_ce": 0.00019449519459158182,
      "loss_iou": 0.248046875,
      "loss_num": 0.018798828125,
      "loss_xval": 0.58984375,
      "num_input_tokens_seen": 222191848,
      "step": 3967
    },
    {
      "epoch": 8.837416481069042,
      "grad_norm": 26.04351043701172,
      "learning_rate": 1e-06,
      "loss": 0.4815,
      "num_input_tokens_seen": 222246416,
      "step": 3968
    },
    {
      "epoch": 8.837416481069042,
      "loss": 0.5717356204986572,
      "loss_ce": 0.00020242987375240773,
      "loss_iou": 0.251953125,
      "loss_num": 0.013671875,
      "loss_xval": 0.5703125,
      "num_input_tokens_seen": 222246416,
      "step": 3968
    },
    {
      "epoch": 8.839643652561247,
      "grad_norm": 32.86327362060547,
      "learning_rate": 1e-06,
      "loss": 0.6211,
      "num_input_tokens_seen": 222300948,
      "step": 3969
    },
    {
      "epoch": 8.839643652561247,
      "loss": 0.7585898637771606,
      "loss_ce": 0.000167002814123407,
      "loss_iou": 0.3203125,
      "loss_num": 0.023681640625,
      "loss_xval": 0.7578125,
      "num_input_tokens_seen": 222300948,
      "step": 3969
    },
    {
      "epoch": 8.841870824053451,
      "grad_norm": 29.39252471923828,
      "learning_rate": 1e-06,
      "loss": 0.7207,
      "num_input_tokens_seen": 222356800,
      "step": 3970
    },
    {
      "epoch": 8.841870824053451,
      "loss": 0.7579694390296936,
      "loss_ce": 0.00015694380272179842,
      "loss_iou": 0.310546875,
      "loss_num": 0.0277099609375,
      "loss_xval": 0.7578125,
      "num_input_tokens_seen": 222356800,
      "step": 3970
    },
    {
      "epoch": 8.844097995545656,
      "grad_norm": 47.64936828613281,
      "learning_rate": 1e-06,
      "loss": 0.7561,
      "num_input_tokens_seen": 222414448,
      "step": 3971
    },
    {
      "epoch": 8.844097995545656,
      "loss": 0.6730362772941589,
      "loss_ce": 0.00018474232638254762,
      "loss_iou": 0.2890625,
      "loss_num": 0.01904296875,
      "loss_xval": 0.671875,
      "num_input_tokens_seen": 222414448,
      "step": 3971
    },
    {
      "epoch": 8.846325167037861,
      "grad_norm": 19.512876510620117,
      "learning_rate": 1e-06,
      "loss": 0.7139,
      "num_input_tokens_seen": 222468120,
      "step": 3972
    },
    {
      "epoch": 8.846325167037861,
      "loss": 0.6079658269882202,
      "loss_ce": 0.0001777580037014559,
      "loss_iou": 0.263671875,
      "loss_num": 0.015869140625,
      "loss_xval": 0.609375,
      "num_input_tokens_seen": 222468120,
      "step": 3972
    },
    {
      "epoch": 8.848552338530066,
      "grad_norm": 17.036836624145508,
      "learning_rate": 1e-06,
      "loss": 0.6141,
      "num_input_tokens_seen": 222523584,
      "step": 3973
    },
    {
      "epoch": 8.848552338530066,
      "loss": 0.5712729692459106,
      "loss_ce": 0.00016703552682884037,
      "loss_iou": 0.23828125,
      "loss_num": 0.0189208984375,
      "loss_xval": 0.5703125,
      "num_input_tokens_seen": 222523584,
      "step": 3973
    },
    {
      "epoch": 8.85077951002227,
      "grad_norm": 17.4189395904541,
      "learning_rate": 1e-06,
      "loss": 0.4694,
      "num_input_tokens_seen": 222581532,
      "step": 3974
    },
    {
      "epoch": 8.85077951002227,
      "loss": 0.42311620712280273,
      "loss_ce": 0.00014257154543884099,
      "loss_iou": 0.1904296875,
      "loss_num": 0.00860595703125,
      "loss_xval": 0.423828125,
      "num_input_tokens_seen": 222581532,
      "step": 3974
    },
    {
      "epoch": 8.853006681514476,
      "grad_norm": 40.079444885253906,
      "learning_rate": 1e-06,
      "loss": 0.5725,
      "num_input_tokens_seen": 222638036,
      "step": 3975
    },
    {
      "epoch": 8.853006681514476,
      "loss": 0.61489337682724,
      "loss_ce": 0.00014727030065841973,
      "loss_iou": 0.2734375,
      "loss_num": 0.0135498046875,
      "loss_xval": 0.61328125,
      "num_input_tokens_seen": 222638036,
      "step": 3975
    },
    {
      "epoch": 8.855233853006682,
      "grad_norm": 29.330018997192383,
      "learning_rate": 1e-06,
      "loss": 0.468,
      "num_input_tokens_seen": 222693340,
      "step": 3976
    },
    {
      "epoch": 8.855233853006682,
      "loss": 0.531175971031189,
      "loss_ce": 0.00017008304712362587,
      "loss_iou": 0.23828125,
      "loss_num": 0.01104736328125,
      "loss_xval": 0.53125,
      "num_input_tokens_seen": 222693340,
      "step": 3976
    },
    {
      "epoch": 8.857461024498887,
      "grad_norm": 11.950536727905273,
      "learning_rate": 1e-06,
      "loss": 0.4903,
      "num_input_tokens_seen": 222750828,
      "step": 3977
    },
    {
      "epoch": 8.857461024498887,
      "loss": 0.36172372102737427,
      "loss_ce": 0.00015144153439905494,
      "loss_iou": 0.1708984375,
      "loss_num": 0.00408935546875,
      "loss_xval": 0.361328125,
      "num_input_tokens_seen": 222750828,
      "step": 3977
    },
    {
      "epoch": 8.859688195991092,
      "grad_norm": 17.319873809814453,
      "learning_rate": 1e-06,
      "loss": 0.569,
      "num_input_tokens_seen": 222807080,
      "step": 3978
    },
    {
      "epoch": 8.859688195991092,
      "loss": 0.5596229434013367,
      "loss_ce": 0.00017472056788392365,
      "loss_iou": 0.212890625,
      "loss_num": 0.026611328125,
      "loss_xval": 0.55859375,
      "num_input_tokens_seen": 222807080,
      "step": 3978
    },
    {
      "epoch": 8.861915367483297,
      "grad_norm": 25.0523624420166,
      "learning_rate": 1e-06,
      "loss": 0.6712,
      "num_input_tokens_seen": 222864000,
      "step": 3979
    },
    {
      "epoch": 8.861915367483297,
      "loss": 0.670589804649353,
      "loss_ce": 0.00017965443839784712,
      "loss_iou": 0.3046875,
      "loss_num": 0.01214599609375,
      "loss_xval": 0.671875,
      "num_input_tokens_seen": 222864000,
      "step": 3979
    },
    {
      "epoch": 8.864142538975502,
      "grad_norm": 18.790050506591797,
      "learning_rate": 1e-06,
      "loss": 0.4437,
      "num_input_tokens_seen": 222918468,
      "step": 3980
    },
    {
      "epoch": 8.864142538975502,
      "loss": 0.5265804529190063,
      "loss_ce": 0.00021330692106857896,
      "loss_iou": 0.2333984375,
      "loss_num": 0.01177978515625,
      "loss_xval": 0.52734375,
      "num_input_tokens_seen": 222918468,
      "step": 3980
    },
    {
      "epoch": 8.866369710467707,
      "grad_norm": 39.062259674072266,
      "learning_rate": 1e-06,
      "loss": 0.5382,
      "num_input_tokens_seen": 222971988,
      "step": 3981
    },
    {
      "epoch": 8.866369710467707,
      "loss": 0.4984826147556305,
      "loss_ce": 0.00019161765521857888,
      "loss_iou": 0.2177734375,
      "loss_num": 0.0126953125,
      "loss_xval": 0.498046875,
      "num_input_tokens_seen": 222971988,
      "step": 3981
    },
    {
      "epoch": 8.868596881959911,
      "grad_norm": 17.326452255249023,
      "learning_rate": 1e-06,
      "loss": 0.6576,
      "num_input_tokens_seen": 223027000,
      "step": 3982
    },
    {
      "epoch": 8.868596881959911,
      "loss": 0.6256543397903442,
      "loss_ce": 0.00041018627234734595,
      "loss_iou": 0.275390625,
      "loss_num": 0.01513671875,
      "loss_xval": 0.625,
      "num_input_tokens_seen": 223027000,
      "step": 3982
    },
    {
      "epoch": 8.870824053452116,
      "grad_norm": 16.484027862548828,
      "learning_rate": 1e-06,
      "loss": 0.7406,
      "num_input_tokens_seen": 223081684,
      "step": 3983
    },
    {
      "epoch": 8.870824053452116,
      "loss": 0.6387948989868164,
      "loss_ce": 0.0001230332418344915,
      "loss_iou": 0.275390625,
      "loss_num": 0.017578125,
      "loss_xval": 0.640625,
      "num_input_tokens_seen": 223081684,
      "step": 3983
    },
    {
      "epoch": 8.873051224944321,
      "grad_norm": 26.997020721435547,
      "learning_rate": 1e-06,
      "loss": 0.6737,
      "num_input_tokens_seen": 223135544,
      "step": 3984
    },
    {
      "epoch": 8.873051224944321,
      "loss": 0.745303213596344,
      "loss_ce": 0.00018600517068989575,
      "loss_iou": 0.310546875,
      "loss_num": 0.024658203125,
      "loss_xval": 0.74609375,
      "num_input_tokens_seen": 223135544,
      "step": 3984
    },
    {
      "epoch": 8.875278396436526,
      "grad_norm": 16.463109970092773,
      "learning_rate": 1e-06,
      "loss": 0.6391,
      "num_input_tokens_seen": 223192264,
      "step": 3985
    },
    {
      "epoch": 8.875278396436526,
      "loss": 0.6673752069473267,
      "loss_ce": 0.00038302806206047535,
      "loss_iou": 0.283203125,
      "loss_num": 0.0205078125,
      "loss_xval": 0.66796875,
      "num_input_tokens_seen": 223192264,
      "step": 3985
    },
    {
      "epoch": 8.877505567928731,
      "grad_norm": 26.982946395874023,
      "learning_rate": 1e-06,
      "loss": 0.5375,
      "num_input_tokens_seen": 223247844,
      "step": 3986
    },
    {
      "epoch": 8.877505567928731,
      "loss": 0.5868437886238098,
      "loss_ce": 0.00041803409112617373,
      "loss_iou": 0.255859375,
      "loss_num": 0.0147705078125,
      "loss_xval": 0.5859375,
      "num_input_tokens_seen": 223247844,
      "step": 3986
    },
    {
      "epoch": 8.879732739420936,
      "grad_norm": 17.86455726623535,
      "learning_rate": 1e-06,
      "loss": 0.5355,
      "num_input_tokens_seen": 223304620,
      "step": 3987
    },
    {
      "epoch": 8.879732739420936,
      "loss": 0.46156710386276245,
      "loss_ce": 0.0001413182180840522,
      "loss_iou": 0.203125,
      "loss_num": 0.01080322265625,
      "loss_xval": 0.4609375,
      "num_input_tokens_seen": 223304620,
      "step": 3987
    },
    {
      "epoch": 8.88195991091314,
      "grad_norm": 19.752405166625977,
      "learning_rate": 1e-06,
      "loss": 0.3951,
      "num_input_tokens_seen": 223361520,
      "step": 3988
    },
    {
      "epoch": 8.88195991091314,
      "loss": 0.3074072003364563,
      "loss_ce": 0.0001561975514050573,
      "loss_iou": 0.140625,
      "loss_num": 0.0052490234375,
      "loss_xval": 0.306640625,
      "num_input_tokens_seen": 223361520,
      "step": 3988
    },
    {
      "epoch": 8.884187082405345,
      "grad_norm": 12.55148983001709,
      "learning_rate": 1e-06,
      "loss": 0.7263,
      "num_input_tokens_seen": 223419580,
      "step": 3989
    },
    {
      "epoch": 8.884187082405345,
      "loss": 1.0404853820800781,
      "loss_ce": 0.00020214373944327235,
      "loss_iou": 0.3984375,
      "loss_num": 0.048583984375,
      "loss_xval": 1.0390625,
      "num_input_tokens_seen": 223419580,
      "step": 3989
    },
    {
      "epoch": 8.88641425389755,
      "grad_norm": 19.007347106933594,
      "learning_rate": 1e-06,
      "loss": 0.5547,
      "num_input_tokens_seen": 223476520,
      "step": 3990
    },
    {
      "epoch": 8.88641425389755,
      "loss": 0.45726656913757324,
      "loss_ce": 0.00017428494174964726,
      "loss_iou": 0.1953125,
      "loss_num": 0.01312255859375,
      "loss_xval": 0.45703125,
      "num_input_tokens_seen": 223476520,
      "step": 3990
    },
    {
      "epoch": 8.888641425389755,
      "grad_norm": 21.701292037963867,
      "learning_rate": 1e-06,
      "loss": 0.6151,
      "num_input_tokens_seen": 223532396,
      "step": 3991
    },
    {
      "epoch": 8.888641425389755,
      "loss": 0.6684781908988953,
      "loss_ce": 0.00014323292998597026,
      "loss_iou": 0.291015625,
      "loss_num": 0.0174560546875,
      "loss_xval": 0.66796875,
      "num_input_tokens_seen": 223532396,
      "step": 3991
    },
    {
      "epoch": 8.89086859688196,
      "grad_norm": 20.25054359436035,
      "learning_rate": 1e-06,
      "loss": 0.7418,
      "num_input_tokens_seen": 223588248,
      "step": 3992
    },
    {
      "epoch": 8.89086859688196,
      "loss": 0.6505246162414551,
      "loss_ce": 0.00013402706827037036,
      "loss_iou": 0.294921875,
      "loss_num": 0.011962890625,
      "loss_xval": 0.6484375,
      "num_input_tokens_seen": 223588248,
      "step": 3992
    },
    {
      "epoch": 8.893095768374165,
      "grad_norm": 39.547183990478516,
      "learning_rate": 1e-06,
      "loss": 0.6393,
      "num_input_tokens_seen": 223643372,
      "step": 3993
    },
    {
      "epoch": 8.893095768374165,
      "loss": 0.47359946370124817,
      "loss_ce": 0.0009432291844859719,
      "loss_iou": 0.2080078125,
      "loss_num": 0.0111083984375,
      "loss_xval": 0.47265625,
      "num_input_tokens_seen": 223643372,
      "step": 3993
    },
    {
      "epoch": 8.89532293986637,
      "grad_norm": 15.449313163757324,
      "learning_rate": 1e-06,
      "loss": 0.4557,
      "num_input_tokens_seen": 223701152,
      "step": 3994
    },
    {
      "epoch": 8.89532293986637,
      "loss": 0.4065554440021515,
      "loss_ce": 0.00018337275832891464,
      "loss_iou": 0.17578125,
      "loss_num": 0.01092529296875,
      "loss_xval": 0.40625,
      "num_input_tokens_seen": 223701152,
      "step": 3994
    },
    {
      "epoch": 8.897550111358575,
      "grad_norm": 20.84811782836914,
      "learning_rate": 1e-06,
      "loss": 0.5546,
      "num_input_tokens_seen": 223754800,
      "step": 3995
    },
    {
      "epoch": 8.897550111358575,
      "loss": 0.5366966724395752,
      "loss_ce": 0.00013655968359671533,
      "loss_iou": 0.2275390625,
      "loss_num": 0.0162353515625,
      "loss_xval": 0.53515625,
      "num_input_tokens_seen": 223754800,
      "step": 3995
    },
    {
      "epoch": 8.89977728285078,
      "grad_norm": 22.598325729370117,
      "learning_rate": 1e-06,
      "loss": 0.4089,
      "num_input_tokens_seen": 223811716,
      "step": 3996
    },
    {
      "epoch": 8.89977728285078,
      "loss": 0.5239717364311218,
      "loss_ce": 0.00016802808386273682,
      "loss_iou": 0.2158203125,
      "loss_num": 0.0185546875,
      "loss_xval": 0.5234375,
      "num_input_tokens_seen": 223811716,
      "step": 3996
    },
    {
      "epoch": 8.902004454342984,
      "grad_norm": 16.426435470581055,
      "learning_rate": 1e-06,
      "loss": 0.5465,
      "num_input_tokens_seen": 223867872,
      "step": 3997
    },
    {
      "epoch": 8.902004454342984,
      "loss": 0.5754693746566772,
      "loss_ce": 0.00015196282765828073,
      "loss_iou": 0.24609375,
      "loss_num": 0.0164794921875,
      "loss_xval": 0.57421875,
      "num_input_tokens_seen": 223867872,
      "step": 3997
    },
    {
      "epoch": 8.90423162583519,
      "grad_norm": 35.01568603515625,
      "learning_rate": 1e-06,
      "loss": 0.8874,
      "num_input_tokens_seen": 223922524,
      "step": 3998
    },
    {
      "epoch": 8.90423162583519,
      "loss": 0.7457464337348938,
      "loss_ce": 0.00014096120139583945,
      "loss_iou": 0.326171875,
      "loss_num": 0.0185546875,
      "loss_xval": 0.74609375,
      "num_input_tokens_seen": 223922524,
      "step": 3998
    },
    {
      "epoch": 8.906458797327394,
      "grad_norm": 52.03001403808594,
      "learning_rate": 1e-06,
      "loss": 0.7269,
      "num_input_tokens_seen": 223977692,
      "step": 3999
    },
    {
      "epoch": 8.906458797327394,
      "loss": 0.9718111753463745,
      "loss_ce": 0.00013149988080840558,
      "loss_iou": 0.392578125,
      "loss_num": 0.036865234375,
      "loss_xval": 0.97265625,
      "num_input_tokens_seen": 223977692,
      "step": 3999
    },
    {
      "epoch": 8.908685968819599,
      "grad_norm": 23.651729583740234,
      "learning_rate": 1e-06,
      "loss": 0.6773,
      "num_input_tokens_seen": 224031804,
      "step": 4000
    },
    {
      "epoch": 8.908685968819599,
      "eval_seeclick_web_CIoU": 0.584255576133728,
      "eval_seeclick_web_GIoU": 0.5831271409988403,
      "eval_seeclick_web_IoU": 0.6019724309444427,
      "eval_seeclick_web_MAE_all": 0.0160904498770833,
      "eval_seeclick_web_MAE_h": 0.00791524676606059,
      "eval_seeclick_web_MAE_w": 0.016444522887468338,
      "eval_seeclick_web_MAE_x_boxes": 0.009520682971924543,
      "eval_seeclick_web_MAE_y_boxes": 0.022019447991624475,
      "eval_seeclick_web_inside_bbox": 0.9166666567325592,
      "eval_seeclick_web_loss": 0.9091385006904602,
      "eval_seeclick_web_loss_ce": 0.00022534980962518603,
      "eval_seeclick_web_loss_iou": 0.41326904296875,
      "eval_seeclick_web_loss_num": 0.012880325317382812,
      "eval_seeclick_web_loss_xval": 0.8909912109375,
      "eval_seeclick_web_runtime": 23.4695,
      "eval_seeclick_web_samples_per_second": 2.13,
      "eval_seeclick_web_steps_per_second": 0.085,
      "num_input_tokens_seen": 224031804,
      "step": 4000
    },
    {
      "epoch": 8.908685968819599,
      "eval_icons_CIoU": 0.280209019780159,
      "eval_icons_GIoU": 0.3042156249284744,
      "eval_icons_IoU": 0.3636113852262497,
      "eval_icons_MAE_all": 0.06306547299027443,
      "eval_icons_MAE_h": 0.03832197841256857,
      "eval_icons_MAE_w": 0.06806911528110504,
      "eval_icons_MAE_x_boxes": 0.05768297426402569,
      "eval_icons_MAE_y_boxes": 0.03917842544615269,
      "eval_icons_inside_bbox": 0.6336805522441864,
      "eval_icons_loss": 1.7296451330184937,
      "eval_icons_loss_ce": 0.00027250249695498496,
      "eval_icons_loss_iou": 0.6729736328125,
      "eval_icons_loss_num": 0.06075096130371094,
      "eval_icons_loss_xval": 1.648193359375,
      "eval_icons_runtime": 20.6404,
      "eval_icons_samples_per_second": 2.422,
      "eval_icons_steps_per_second": 0.097,
      "num_input_tokens_seen": 224031804,
      "step": 4000
    },
    {
      "epoch": 8.908685968819599,
      "eval_screenspot_CIoU": 0.344086229801178,
      "eval_screenspot_GIoU": 0.36166812976201373,
      "eval_screenspot_IoU": 0.42560062805811566,
      "eval_screenspot_MAE_all": 0.06235171233614286,
      "eval_screenspot_MAE_h": 0.038458424930771194,
      "eval_screenspot_MAE_w": 0.07053530837098758,
      "eval_screenspot_MAE_x_boxes": 0.07286067555348079,
      "eval_screenspot_MAE_y_boxes": 0.04639405757188797,
      "eval_screenspot_inside_bbox": 0.6862499912579855,
      "eval_screenspot_loss": 1.6449319124221802,
      "eval_screenspot_loss_ce": 0.0002812407910823822,
      "eval_screenspot_loss_iou": 0.6747233072916666,
      "eval_screenspot_loss_num": 0.07325236002604167,
      "eval_screenspot_loss_xval": 1.7154947916666667,
      "eval_screenspot_runtime": 37.0539,
      "eval_screenspot_samples_per_second": 2.402,
      "eval_screenspot_steps_per_second": 0.081,
      "num_input_tokens_seen": 224031804,
      "step": 4000
    },
    {
      "epoch": 8.908685968819599,
      "eval_compot_CIoU": 0.3489241451025009,
      "eval_compot_GIoU": 0.3593568354845047,
      "eval_compot_IoU": 0.4082530289888382,
      "eval_compot_MAE_all": 0.01762966625392437,
      "eval_compot_MAE_h": 0.009061001241207123,
      "eval_compot_MAE_w": 0.022062174044549465,
      "eval_compot_MAE_x_boxes": 0.028892694041132927,
      "eval_compot_MAE_y_boxes": 0.006829841528087854,
      "eval_compot_inside_bbox": 0.6302083432674408,
      "eval_compot_loss": 1.407405138015747,
      "eval_compot_loss_ce": 0.00021665637177648023,
      "eval_compot_loss_iou": 0.6473388671875,
      "eval_compot_loss_num": 0.016569137573242188,
      "eval_compot_loss_xval": 1.377197265625,
      "eval_compot_runtime": 21.8205,
      "eval_compot_samples_per_second": 2.291,
      "eval_compot_steps_per_second": 0.092,
      "num_input_tokens_seen": 224031804,
      "step": 4000
    },
    {
      "epoch": 8.908685968819599,
      "eval_custom_ui_val_CIoU": 0.4698154992527432,
      "eval_custom_ui_val_GIoU": 0.483290907409456,
      "eval_custom_ui_val_IoU": 0.5295593506760068,
      "eval_custom_ui_val_MAE_all": 0.030179924021164577,
      "eval_custom_ui_val_MAE_h": 0.016545448245273695,
      "eval_custom_ui_val_MAE_w": 0.03918004294650422,
      "eval_custom_ui_val_MAE_x_boxes": 0.03695020234833161,
      "eval_custom_ui_val_MAE_y_boxes": 0.01497822223852078,
      "eval_custom_ui_val_inside_bbox": 0.7457561757829454,
      "eval_custom_ui_val_loss": 1.1855908632278442,
      "eval_custom_ui_val_loss_ce": 0.00024014294386789616,
      "eval_custom_ui_val_loss_iou": 0.5057237413194444,
      "eval_custom_ui_val_loss_num": 0.027684529622395832,
      "eval_custom_ui_val_loss_xval": 1.1498480902777777,
      "eval_custom_ui_val_runtime": 70.3473,
      "eval_custom_ui_val_samples_per_second": 3.767,
      "eval_custom_ui_val_steps_per_second": 0.128,
      "num_input_tokens_seen": 224031804,
      "step": 4000
    },
    {
      "epoch": 8.908685968819599,
      "loss": 0.9238047003746033,
      "loss_ce": 0.00022073142463341355,
      "loss_iou": 0.400390625,
      "loss_num": 0.0245361328125,
      "loss_xval": 0.921875,
      "num_input_tokens_seen": 224031804,
      "step": 4000
    },
    {
      "epoch": 8.910913140311804,
      "grad_norm": 17.125661849975586,
      "learning_rate": 1e-06,
      "loss": 0.6028,
      "num_input_tokens_seen": 224088664,
      "step": 4001
    },
    {
      "epoch": 8.910913140311804,
      "loss": 0.6688809394836426,
      "loss_ce": 0.00017975796072278172,
      "loss_iou": 0.279296875,
      "loss_num": 0.0224609375,
      "loss_xval": 0.66796875,
      "num_input_tokens_seen": 224088664,
      "step": 4001
    },
    {
      "epoch": 8.913140311804009,
      "grad_norm": 11.797565460205078,
      "learning_rate": 1e-06,
      "loss": 0.4654,
      "num_input_tokens_seen": 224146092,
      "step": 4002
    },
    {
      "epoch": 8.913140311804009,
      "loss": 0.43066659569740295,
      "loss_ce": 0.00012459905701689422,
      "loss_iou": 0.181640625,
      "loss_num": 0.013427734375,
      "loss_xval": 0.4296875,
      "num_input_tokens_seen": 224146092,
      "step": 4002
    },
    {
      "epoch": 8.915367483296214,
      "grad_norm": 18.511592864990234,
      "learning_rate": 1e-06,
      "loss": 0.4735,
      "num_input_tokens_seen": 224201576,
      "step": 4003
    },
    {
      "epoch": 8.915367483296214,
      "loss": 0.44720643758773804,
      "loss_ce": 0.00018495078256819397,
      "loss_iou": 0.1953125,
      "loss_num": 0.0113525390625,
      "loss_xval": 0.447265625,
      "num_input_tokens_seen": 224201576,
      "step": 4003
    },
    {
      "epoch": 8.917594654788418,
      "grad_norm": 19.493896484375,
      "learning_rate": 1e-06,
      "loss": 0.658,
      "num_input_tokens_seen": 224258016,
      "step": 4004
    },
    {
      "epoch": 8.917594654788418,
      "loss": 0.9191592931747437,
      "loss_ce": 0.00021400017431005836,
      "loss_iou": 0.388671875,
      "loss_num": 0.0281982421875,
      "loss_xval": 0.91796875,
      "num_input_tokens_seen": 224258016,
      "step": 4004
    },
    {
      "epoch": 8.919821826280623,
      "grad_norm": 19.006305694580078,
      "learning_rate": 1e-06,
      "loss": 0.6289,
      "num_input_tokens_seen": 224311936,
      "step": 4005
    },
    {
      "epoch": 8.919821826280623,
      "loss": 0.7552786469459534,
      "loss_ce": 0.0001517195487394929,
      "loss_iou": 0.322265625,
      "loss_num": 0.021728515625,
      "loss_xval": 0.75390625,
      "num_input_tokens_seen": 224311936,
      "step": 4005
    },
    {
      "epoch": 8.922048997772828,
      "grad_norm": 58.83499526977539,
      "learning_rate": 1e-06,
      "loss": 0.6666,
      "num_input_tokens_seen": 224368388,
      "step": 4006
    },
    {
      "epoch": 8.922048997772828,
      "loss": 0.746427595615387,
      "loss_ce": 0.00021177891176193953,
      "loss_iou": 0.33984375,
      "loss_num": 0.01336669921875,
      "loss_xval": 0.74609375,
      "num_input_tokens_seen": 224368388,
      "step": 4006
    },
    {
      "epoch": 8.924276169265033,
      "grad_norm": 17.04109001159668,
      "learning_rate": 1e-06,
      "loss": 0.6045,
      "num_input_tokens_seen": 224423224,
      "step": 4007
    },
    {
      "epoch": 8.924276169265033,
      "loss": 0.6434751749038696,
      "loss_ce": 0.00016464036889374256,
      "loss_iou": 0.298828125,
      "loss_num": 0.00958251953125,
      "loss_xval": 0.64453125,
      "num_input_tokens_seen": 224423224,
      "step": 4007
    },
    {
      "epoch": 8.926503340757238,
      "grad_norm": 16.078716278076172,
      "learning_rate": 1e-06,
      "loss": 0.4969,
      "num_input_tokens_seen": 224479088,
      "step": 4008
    },
    {
      "epoch": 8.926503340757238,
      "loss": 0.47645998001098633,
      "loss_ce": 0.00014161772560328245,
      "loss_iou": 0.2080078125,
      "loss_num": 0.011962890625,
      "loss_xval": 0.4765625,
      "num_input_tokens_seen": 224479088,
      "step": 4008
    },
    {
      "epoch": 8.928730512249443,
      "grad_norm": 18.10431671142578,
      "learning_rate": 1e-06,
      "loss": 0.5381,
      "num_input_tokens_seen": 224536168,
      "step": 4009
    },
    {
      "epoch": 8.928730512249443,
      "loss": 0.5730590224266052,
      "loss_ce": 0.00018303040997125208,
      "loss_iou": 0.220703125,
      "loss_num": 0.0264892578125,
      "loss_xval": 0.57421875,
      "num_input_tokens_seen": 224536168,
      "step": 4009
    },
    {
      "epoch": 8.930957683741648,
      "grad_norm": 20.611072540283203,
      "learning_rate": 1e-06,
      "loss": 0.5397,
      "num_input_tokens_seen": 224594284,
      "step": 4010
    },
    {
      "epoch": 8.930957683741648,
      "loss": 0.5585442781448364,
      "loss_ce": 0.00019466917729005218,
      "loss_iou": 0.2578125,
      "loss_num": 0.0087890625,
      "loss_xval": 0.55859375,
      "num_input_tokens_seen": 224594284,
      "step": 4010
    },
    {
      "epoch": 8.933184855233852,
      "grad_norm": 20.54092788696289,
      "learning_rate": 1e-06,
      "loss": 0.5272,
      "num_input_tokens_seen": 224651224,
      "step": 4011
    },
    {
      "epoch": 8.933184855233852,
      "loss": 0.577430248260498,
      "loss_ce": 0.00015977630391716957,
      "loss_iou": 0.259765625,
      "loss_num": 0.01165771484375,
      "loss_xval": 0.578125,
      "num_input_tokens_seen": 224651224,
      "step": 4011
    },
    {
      "epoch": 8.935412026726057,
      "grad_norm": 15.826013565063477,
      "learning_rate": 1e-06,
      "loss": 0.6706,
      "num_input_tokens_seen": 224707328,
      "step": 4012
    },
    {
      "epoch": 8.935412026726057,
      "loss": 0.4253334701061249,
      "loss_ce": 0.0001625774457352236,
      "loss_iou": 0.193359375,
      "loss_num": 0.007568359375,
      "loss_xval": 0.42578125,
      "num_input_tokens_seen": 224707328,
      "step": 4012
    },
    {
      "epoch": 8.937639198218262,
      "grad_norm": 25.152671813964844,
      "learning_rate": 1e-06,
      "loss": 0.5631,
      "num_input_tokens_seen": 224766028,
      "step": 4013
    },
    {
      "epoch": 8.937639198218262,
      "loss": 0.7184352874755859,
      "loss_ce": 0.0001735725672915578,
      "loss_iou": 0.31640625,
      "loss_num": 0.01708984375,
      "loss_xval": 0.71875,
      "num_input_tokens_seen": 224766028,
      "step": 4013
    },
    {
      "epoch": 8.939866369710467,
      "grad_norm": 30.11623191833496,
      "learning_rate": 1e-06,
      "loss": 0.605,
      "num_input_tokens_seen": 224822620,
      "step": 4014
    },
    {
      "epoch": 8.939866369710467,
      "loss": 0.7723772525787354,
      "loss_ce": 0.00016042383504100144,
      "loss_iou": 0.326171875,
      "loss_num": 0.0244140625,
      "loss_xval": 0.7734375,
      "num_input_tokens_seen": 224822620,
      "step": 4014
    },
    {
      "epoch": 8.942093541202672,
      "grad_norm": 45.63913345336914,
      "learning_rate": 1e-06,
      "loss": 0.5652,
      "num_input_tokens_seen": 224878404,
      "step": 4015
    },
    {
      "epoch": 8.942093541202672,
      "loss": 0.6024213433265686,
      "loss_ce": 0.0001264153397642076,
      "loss_iou": 0.2314453125,
      "loss_num": 0.0281982421875,
      "loss_xval": 0.6015625,
      "num_input_tokens_seen": 224878404,
      "step": 4015
    },
    {
      "epoch": 8.944320712694877,
      "grad_norm": 17.508010864257812,
      "learning_rate": 1e-06,
      "loss": 0.4491,
      "num_input_tokens_seen": 224936428,
      "step": 4016
    },
    {
      "epoch": 8.944320712694877,
      "loss": 0.5184711217880249,
      "loss_ce": 0.0001606106961844489,
      "loss_iou": 0.220703125,
      "loss_num": 0.01531982421875,
      "loss_xval": 0.51953125,
      "num_input_tokens_seen": 224936428,
      "step": 4016
    },
    {
      "epoch": 8.946547884187082,
      "grad_norm": 15.416549682617188,
      "learning_rate": 1e-06,
      "loss": 0.6399,
      "num_input_tokens_seen": 224994188,
      "step": 4017
    },
    {
      "epoch": 8.946547884187082,
      "loss": 0.7011144161224365,
      "loss_ce": 0.00018664849631022662,
      "loss_iou": 0.287109375,
      "loss_num": 0.0252685546875,
      "loss_xval": 0.69921875,
      "num_input_tokens_seen": 224994188,
      "step": 4017
    },
    {
      "epoch": 8.948775055679288,
      "grad_norm": 21.26276969909668,
      "learning_rate": 1e-06,
      "loss": 0.4372,
      "num_input_tokens_seen": 225051360,
      "step": 4018
    },
    {
      "epoch": 8.948775055679288,
      "loss": 0.3914967179298401,
      "loss_ce": 0.00013928582484368235,
      "loss_iou": 0.171875,
      "loss_num": 0.00958251953125,
      "loss_xval": 0.390625,
      "num_input_tokens_seen": 225051360,
      "step": 4018
    },
    {
      "epoch": 8.951002227171493,
      "grad_norm": 13.906608581542969,
      "learning_rate": 1e-06,
      "loss": 0.4571,
      "num_input_tokens_seen": 225108952,
      "step": 4019
    },
    {
      "epoch": 8.951002227171493,
      "loss": 0.4027478098869324,
      "loss_ce": 0.00015992176486179233,
      "loss_iou": 0.1767578125,
      "loss_num": 0.00970458984375,
      "loss_xval": 0.40234375,
      "num_input_tokens_seen": 225108952,
      "step": 4019
    },
    {
      "epoch": 8.953229398663698,
      "grad_norm": 13.286585807800293,
      "learning_rate": 1e-06,
      "loss": 0.6289,
      "num_input_tokens_seen": 225165484,
      "step": 4020
    },
    {
      "epoch": 8.953229398663698,
      "loss": 0.37464502453804016,
      "loss_ce": 0.00013330676301848143,
      "loss_iou": 0.158203125,
      "loss_num": 0.011474609375,
      "loss_xval": 0.375,
      "num_input_tokens_seen": 225165484,
      "step": 4020
    },
    {
      "epoch": 8.955456570155903,
      "grad_norm": 27.786767959594727,
      "learning_rate": 1e-06,
      "loss": 0.843,
      "num_input_tokens_seen": 225219788,
      "step": 4021
    },
    {
      "epoch": 8.955456570155903,
      "loss": 0.6932044625282288,
      "loss_ce": 0.00015029762289486825,
      "loss_iou": 0.30078125,
      "loss_num": 0.0179443359375,
      "loss_xval": 0.69140625,
      "num_input_tokens_seen": 225219788,
      "step": 4021
    },
    {
      "epoch": 8.957683741648108,
      "grad_norm": 15.158284187316895,
      "learning_rate": 1e-06,
      "loss": 0.5056,
      "num_input_tokens_seen": 225277592,
      "step": 4022
    },
    {
      "epoch": 8.957683741648108,
      "loss": 0.5200361013412476,
      "loss_ce": 0.00013863734784536064,
      "loss_iou": 0.212890625,
      "loss_num": 0.018798828125,
      "loss_xval": 0.51953125,
      "num_input_tokens_seen": 225277592,
      "step": 4022
    },
    {
      "epoch": 8.959910913140313,
      "grad_norm": 45.792728424072266,
      "learning_rate": 1e-06,
      "loss": 0.4745,
      "num_input_tokens_seen": 225334828,
      "step": 4023
    },
    {
      "epoch": 8.959910913140313,
      "loss": 0.40143540501594543,
      "loss_ce": 0.0001902944641187787,
      "loss_iou": 0.1826171875,
      "loss_num": 0.00714111328125,
      "loss_xval": 0.400390625,
      "num_input_tokens_seen": 225334828,
      "step": 4023
    },
    {
      "epoch": 8.962138084632517,
      "grad_norm": 19.045509338378906,
      "learning_rate": 1e-06,
      "loss": 0.5239,
      "num_input_tokens_seen": 225389700,
      "step": 4024
    },
    {
      "epoch": 8.962138084632517,
      "loss": 0.4376333951950073,
      "loss_ce": 0.00013340538134798408,
      "loss_iou": 0.1962890625,
      "loss_num": 0.0091552734375,
      "loss_xval": 0.4375,
      "num_input_tokens_seen": 225389700,
      "step": 4024
    },
    {
      "epoch": 8.964365256124722,
      "grad_norm": 15.347426414489746,
      "learning_rate": 1e-06,
      "loss": 0.6214,
      "num_input_tokens_seen": 225447624,
      "step": 4025
    },
    {
      "epoch": 8.964365256124722,
      "loss": 0.5260591506958008,
      "loss_ce": 0.000180262781213969,
      "loss_iou": 0.2216796875,
      "loss_num": 0.016357421875,
      "loss_xval": 0.52734375,
      "num_input_tokens_seen": 225447624,
      "step": 4025
    },
    {
      "epoch": 8.966592427616927,
      "grad_norm": 14.499673843383789,
      "learning_rate": 1e-06,
      "loss": 0.5167,
      "num_input_tokens_seen": 225506444,
      "step": 4026
    },
    {
      "epoch": 8.966592427616927,
      "loss": 0.3981502652168274,
      "loss_ce": 0.00014002776879351586,
      "loss_iou": 0.171875,
      "loss_num": 0.010986328125,
      "loss_xval": 0.3984375,
      "num_input_tokens_seen": 225506444,
      "step": 4026
    },
    {
      "epoch": 8.968819599109132,
      "grad_norm": 19.23043441772461,
      "learning_rate": 1e-06,
      "loss": 0.5612,
      "num_input_tokens_seen": 225560052,
      "step": 4027
    },
    {
      "epoch": 8.968819599109132,
      "loss": 0.32940202951431274,
      "loss_ce": 0.00017840655345935374,
      "loss_iou": 0.140625,
      "loss_num": 0.00958251953125,
      "loss_xval": 0.330078125,
      "num_input_tokens_seen": 225560052,
      "step": 4027
    },
    {
      "epoch": 8.971046770601337,
      "grad_norm": 34.80836486816406,
      "learning_rate": 1e-06,
      "loss": 0.5679,
      "num_input_tokens_seen": 225618236,
      "step": 4028
    },
    {
      "epoch": 8.971046770601337,
      "loss": 0.6726021766662598,
      "loss_ce": 0.00023891603632364422,
      "loss_iou": 0.30078125,
      "loss_num": 0.0142822265625,
      "loss_xval": 0.671875,
      "num_input_tokens_seen": 225618236,
      "step": 4028
    },
    {
      "epoch": 8.973273942093542,
      "grad_norm": 19.638723373413086,
      "learning_rate": 1e-06,
      "loss": 0.5453,
      "num_input_tokens_seen": 225674116,
      "step": 4029
    },
    {
      "epoch": 8.973273942093542,
      "loss": 0.520553708076477,
      "loss_ce": 0.0012665874091908336,
      "loss_iou": 0.1845703125,
      "loss_num": 0.030029296875,
      "loss_xval": 0.51953125,
      "num_input_tokens_seen": 225674116,
      "step": 4029
    },
    {
      "epoch": 8.975501113585747,
      "grad_norm": 18.89423942565918,
      "learning_rate": 1e-06,
      "loss": 0.6397,
      "num_input_tokens_seen": 225730220,
      "step": 4030
    },
    {
      "epoch": 8.975501113585747,
      "loss": 0.5465627908706665,
      "loss_ce": 0.00017607388144824654,
      "loss_iou": 0.244140625,
      "loss_num": 0.01165771484375,
      "loss_xval": 0.546875,
      "num_input_tokens_seen": 225730220,
      "step": 4030
    },
    {
      "epoch": 8.977728285077951,
      "grad_norm": 13.650494575500488,
      "learning_rate": 1e-06,
      "loss": 0.5068,
      "num_input_tokens_seen": 225786488,
      "step": 4031
    },
    {
      "epoch": 8.977728285077951,
      "loss": 0.5636086463928223,
      "loss_ce": 0.00013209109602030367,
      "loss_iou": 0.2275390625,
      "loss_num": 0.021728515625,
      "loss_xval": 0.5625,
      "num_input_tokens_seen": 225786488,
      "step": 4031
    },
    {
      "epoch": 8.979955456570156,
      "grad_norm": 16.062467575073242,
      "learning_rate": 1e-06,
      "loss": 0.4167,
      "num_input_tokens_seen": 225841372,
      "step": 4032
    },
    {
      "epoch": 8.979955456570156,
      "loss": 0.29786229133605957,
      "loss_ce": 0.00013277304242365062,
      "loss_iou": 0.1201171875,
      "loss_num": 0.011474609375,
      "loss_xval": 0.296875,
      "num_input_tokens_seen": 225841372,
      "step": 4032
    },
    {
      "epoch": 8.982182628062361,
      "grad_norm": 17.630416870117188,
      "learning_rate": 1e-06,
      "loss": 0.5697,
      "num_input_tokens_seen": 225894908,
      "step": 4033
    },
    {
      "epoch": 8.982182628062361,
      "loss": 0.676677942276001,
      "loss_ce": 0.00016425049398094416,
      "loss_iou": 0.2734375,
      "loss_num": 0.0257568359375,
      "loss_xval": 0.67578125,
      "num_input_tokens_seen": 225894908,
      "step": 4033
    },
    {
      "epoch": 8.984409799554566,
      "grad_norm": 20.209087371826172,
      "learning_rate": 1e-06,
      "loss": 0.4613,
      "num_input_tokens_seen": 225948016,
      "step": 4034
    },
    {
      "epoch": 8.984409799554566,
      "loss": 0.6483420729637146,
      "loss_ce": 0.00014869487495161593,
      "loss_iou": 0.298828125,
      "loss_num": 0.00982666015625,
      "loss_xval": 0.6484375,
      "num_input_tokens_seen": 225948016,
      "step": 4034
    },
    {
      "epoch": 8.98663697104677,
      "grad_norm": 31.209794998168945,
      "learning_rate": 1e-06,
      "loss": 0.62,
      "num_input_tokens_seen": 226005860,
      "step": 4035
    },
    {
      "epoch": 8.98663697104677,
      "loss": 0.6758050918579102,
      "loss_ce": 0.0002679667086340487,
      "loss_iou": 0.287109375,
      "loss_num": 0.0205078125,
      "loss_xval": 0.67578125,
      "num_input_tokens_seen": 226005860,
      "step": 4035
    },
    {
      "epoch": 8.988864142538976,
      "grad_norm": 13.588601112365723,
      "learning_rate": 1e-06,
      "loss": 0.4508,
      "num_input_tokens_seen": 226061580,
      "step": 4036
    },
    {
      "epoch": 8.988864142538976,
      "loss": 0.2981112003326416,
      "loss_ce": 0.00013759495050180703,
      "loss_iou": 0.134765625,
      "loss_num": 0.00555419921875,
      "loss_xval": 0.298828125,
      "num_input_tokens_seen": 226061580,
      "step": 4036
    },
    {
      "epoch": 8.99109131403118,
      "grad_norm": 12.43724250793457,
      "learning_rate": 1e-06,
      "loss": 0.6268,
      "num_input_tokens_seen": 226121088,
      "step": 4037
    },
    {
      "epoch": 8.99109131403118,
      "loss": 0.7063800096511841,
      "loss_ce": 0.00020320256589911878,
      "loss_iou": 0.298828125,
      "loss_num": 0.02197265625,
      "loss_xval": 0.70703125,
      "num_input_tokens_seen": 226121088,
      "step": 4037
    },
    {
      "epoch": 8.993318485523385,
      "grad_norm": 14.459733963012695,
      "learning_rate": 1e-06,
      "loss": 0.7679,
      "num_input_tokens_seen": 226178000,
      "step": 4038
    },
    {
      "epoch": 8.993318485523385,
      "loss": 0.6036978363990784,
      "loss_ce": 0.00018220869242213666,
      "loss_iou": 0.2294921875,
      "loss_num": 0.029052734375,
      "loss_xval": 0.6015625,
      "num_input_tokens_seen": 226178000,
      "step": 4038
    },
    {
      "epoch": 8.99554565701559,
      "grad_norm": 20.82482147216797,
      "learning_rate": 1e-06,
      "loss": 0.5971,
      "num_input_tokens_seen": 226234736,
      "step": 4039
    },
    {
      "epoch": 8.99554565701559,
      "loss": 0.47330033779144287,
      "loss_ce": 0.00015583749336656183,
      "loss_iou": 0.2158203125,
      "loss_num": 0.00830078125,
      "loss_xval": 0.47265625,
      "num_input_tokens_seen": 226234736,
      "step": 4039
    },
    {
      "epoch": 8.997772828507795,
      "grad_norm": 18.3863468170166,
      "learning_rate": 1e-06,
      "loss": 0.5675,
      "num_input_tokens_seen": 226287252,
      "step": 4040
    },
    {
      "epoch": 8.997772828507795,
      "loss": 0.5164123177528381,
      "loss_ce": 0.00017695256974548101,
      "loss_iou": 0.2314453125,
      "loss_num": 0.01068115234375,
      "loss_xval": 0.515625,
      "num_input_tokens_seen": 226287252,
      "step": 4040
    },
    {
      "epoch": 9.0,
      "grad_norm": 18.648279190063477,
      "learning_rate": 1e-06,
      "loss": 0.5718,
      "num_input_tokens_seen": 226345288,
      "step": 4041
    },
    {
      "epoch": 9.0,
      "loss": 0.6733693480491638,
      "loss_ce": 0.00015158146561589092,
      "loss_iou": 0.283203125,
      "loss_num": 0.0211181640625,
      "loss_xval": 0.671875,
      "num_input_tokens_seen": 226345288,
      "step": 4041
    },
    {
      "epoch": 9.002227171492205,
      "grad_norm": 56.01642608642578,
      "learning_rate": 1e-06,
      "loss": 0.5703,
      "num_input_tokens_seen": 226400596,
      "step": 4042
    },
    {
      "epoch": 9.002227171492205,
      "loss": 0.6565755605697632,
      "loss_ce": 0.00020350792328827083,
      "loss_iou": 0.25,
      "loss_num": 0.03125,
      "loss_xval": 0.65625,
      "num_input_tokens_seen": 226400596,
      "step": 4042
    },
    {
      "epoch": 9.00445434298441,
      "grad_norm": 21.13946533203125,
      "learning_rate": 1e-06,
      "loss": 0.5926,
      "num_input_tokens_seen": 226455584,
      "step": 4043
    },
    {
      "epoch": 9.00445434298441,
      "loss": 0.5227538347244263,
      "loss_ce": 0.00017086087609641254,
      "loss_iou": 0.240234375,
      "loss_num": 0.00830078125,
      "loss_xval": 0.5234375,
      "num_input_tokens_seen": 226455584,
      "step": 4043
    },
    {
      "epoch": 9.006681514476615,
      "grad_norm": 23.61444091796875,
      "learning_rate": 1e-06,
      "loss": 0.5942,
      "num_input_tokens_seen": 226508220,
      "step": 4044
    },
    {
      "epoch": 9.006681514476615,
      "loss": 0.588532030582428,
      "loss_ce": 0.00015310911112464964,
      "loss_iou": 0.24609375,
      "loss_num": 0.0194091796875,
      "loss_xval": 0.58984375,
      "num_input_tokens_seen": 226508220,
      "step": 4044
    },
    {
      "epoch": 9.00890868596882,
      "grad_norm": 26.408849716186523,
      "learning_rate": 1e-06,
      "loss": 0.4008,
      "num_input_tokens_seen": 226563516,
      "step": 4045
    },
    {
      "epoch": 9.00890868596882,
      "loss": 0.4769752323627472,
      "loss_ce": 0.00016859016614034772,
      "loss_iou": 0.20703125,
      "loss_num": 0.01251220703125,
      "loss_xval": 0.4765625,
      "num_input_tokens_seen": 226563516,
      "step": 4045
    },
    {
      "epoch": 9.011135857461024,
      "grad_norm": 16.130887985229492,
      "learning_rate": 1e-06,
      "loss": 0.4924,
      "num_input_tokens_seen": 226619892,
      "step": 4046
    },
    {
      "epoch": 9.011135857461024,
      "loss": 0.3370331823825836,
      "loss_ce": 0.0001191033807117492,
      "loss_iou": 0.11083984375,
      "loss_num": 0.0230712890625,
      "loss_xval": 0.3359375,
      "num_input_tokens_seen": 226619892,
      "step": 4046
    },
    {
      "epoch": 9.01336302895323,
      "grad_norm": 17.81732749938965,
      "learning_rate": 1e-06,
      "loss": 0.6139,
      "num_input_tokens_seen": 226678680,
      "step": 4047
    },
    {
      "epoch": 9.01336302895323,
      "loss": 0.826948881149292,
      "loss_ce": 0.00028875278076156974,
      "loss_iou": 0.345703125,
      "loss_num": 0.0274658203125,
      "loss_xval": 0.828125,
      "num_input_tokens_seen": 226678680,
      "step": 4047
    },
    {
      "epoch": 9.015590200445434,
      "grad_norm": 19.63979721069336,
      "learning_rate": 1e-06,
      "loss": 0.7731,
      "num_input_tokens_seen": 226736108,
      "step": 4048
    },
    {
      "epoch": 9.015590200445434,
      "loss": 0.8490115404129028,
      "loss_ce": 0.00025667509180493653,
      "loss_iou": 0.37890625,
      "loss_num": 0.0179443359375,
      "loss_xval": 0.84765625,
      "num_input_tokens_seen": 226736108,
      "step": 4048
    },
    {
      "epoch": 9.017817371937639,
      "grad_norm": 23.76055335998535,
      "learning_rate": 1e-06,
      "loss": 0.5958,
      "num_input_tokens_seen": 226791264,
      "step": 4049
    },
    {
      "epoch": 9.017817371937639,
      "loss": 0.4640924334526062,
      "loss_ce": 0.00022527157852891833,
      "loss_iou": 0.201171875,
      "loss_num": 0.012451171875,
      "loss_xval": 0.46484375,
      "num_input_tokens_seen": 226791264,
      "step": 4049
    },
    {
      "epoch": 9.020044543429844,
      "grad_norm": 12.326661109924316,
      "learning_rate": 1e-06,
      "loss": 0.49,
      "num_input_tokens_seen": 226849248,
      "step": 4050
    },
    {
      "epoch": 9.020044543429844,
      "loss": 0.4276803731918335,
      "loss_ce": 0.00019013047858607024,
      "loss_iou": 0.1953125,
      "loss_num": 0.007476806640625,
      "loss_xval": 0.427734375,
      "num_input_tokens_seen": 226849248,
      "step": 4050
    },
    {
      "epoch": 9.022271714922049,
      "grad_norm": 25.97770118713379,
      "learning_rate": 1e-06,
      "loss": 0.704,
      "num_input_tokens_seen": 226903776,
      "step": 4051
    },
    {
      "epoch": 9.022271714922049,
      "loss": 0.8058446049690247,
      "loss_ce": 0.0001805661740945652,
      "loss_iou": 0.361328125,
      "loss_num": 0.016357421875,
      "loss_xval": 0.8046875,
      "num_input_tokens_seen": 226903776,
      "step": 4051
    },
    {
      "epoch": 9.024498886414253,
      "grad_norm": 23.6370792388916,
      "learning_rate": 1e-06,
      "loss": 0.4962,
      "num_input_tokens_seen": 226960756,
      "step": 4052
    },
    {
      "epoch": 9.024498886414253,
      "loss": 0.5954891443252563,
      "loss_ce": 0.00015223291120491922,
      "loss_iou": 0.2734375,
      "loss_num": 0.010009765625,
      "loss_xval": 0.59375,
      "num_input_tokens_seen": 226960756,
      "step": 4052
    },
    {
      "epoch": 9.026726057906458,
      "grad_norm": 22.727136611938477,
      "learning_rate": 1e-06,
      "loss": 0.792,
      "num_input_tokens_seen": 227015196,
      "step": 4053
    },
    {
      "epoch": 9.026726057906458,
      "loss": 0.6129428148269653,
      "loss_ce": 0.000394008937291801,
      "loss_iou": 0.267578125,
      "loss_num": 0.01519775390625,
      "loss_xval": 0.61328125,
      "num_input_tokens_seen": 227015196,
      "step": 4053
    },
    {
      "epoch": 9.028953229398663,
      "grad_norm": 21.844358444213867,
      "learning_rate": 1e-06,
      "loss": 0.6825,
      "num_input_tokens_seen": 227069252,
      "step": 4054
    },
    {
      "epoch": 9.028953229398663,
      "loss": 0.7161485552787781,
      "loss_ce": 0.00020616035908460617,
      "loss_iou": 0.3125,
      "loss_num": 0.0185546875,
      "loss_xval": 0.71484375,
      "num_input_tokens_seen": 227069252,
      "step": 4054
    },
    {
      "epoch": 9.031180400890868,
      "grad_norm": 12.784194946289062,
      "learning_rate": 1e-06,
      "loss": 0.5135,
      "num_input_tokens_seen": 227124728,
      "step": 4055
    },
    {
      "epoch": 9.031180400890868,
      "loss": 0.46670782566070557,
      "loss_ce": 0.00015509186778217554,
      "loss_iou": 0.21484375,
      "loss_num": 0.00750732421875,
      "loss_xval": 0.466796875,
      "num_input_tokens_seen": 227124728,
      "step": 4055
    },
    {
      "epoch": 9.033407572383073,
      "grad_norm": 48.46086120605469,
      "learning_rate": 1e-06,
      "loss": 0.7413,
      "num_input_tokens_seen": 227181172,
      "step": 4056
    },
    {
      "epoch": 9.033407572383073,
      "loss": 1.0057790279388428,
      "loss_ce": 0.00016376111307181418,
      "loss_iou": 0.40625,
      "loss_num": 0.0390625,
      "loss_xval": 1.0078125,
      "num_input_tokens_seen": 227181172,
      "step": 4056
    },
    {
      "epoch": 9.035634743875278,
      "grad_norm": 20.736644744873047,
      "learning_rate": 1e-06,
      "loss": 0.5321,
      "num_input_tokens_seen": 227237508,
      "step": 4057
    },
    {
      "epoch": 9.035634743875278,
      "loss": 0.46913978457450867,
      "loss_ce": 0.0007559881778433919,
      "loss_iou": 0.2099609375,
      "loss_num": 0.00982666015625,
      "loss_xval": 0.46875,
      "num_input_tokens_seen": 227237508,
      "step": 4057
    },
    {
      "epoch": 9.037861915367483,
      "grad_norm": 69.1377944946289,
      "learning_rate": 1e-06,
      "loss": 0.7385,
      "num_input_tokens_seen": 227289364,
      "step": 4058
    },
    {
      "epoch": 9.037861915367483,
      "loss": 0.597074568271637,
      "loss_ce": 0.00015077766147442162,
      "loss_iou": 0.26171875,
      "loss_num": 0.01483154296875,
      "loss_xval": 0.59765625,
      "num_input_tokens_seen": 227289364,
      "step": 4058
    },
    {
      "epoch": 9.040089086859687,
      "grad_norm": 15.685859680175781,
      "learning_rate": 1e-06,
      "loss": 0.5525,
      "num_input_tokens_seen": 227345812,
      "step": 4059
    },
    {
      "epoch": 9.040089086859687,
      "loss": 0.5543001890182495,
      "loss_ce": 0.00016202838742174208,
      "loss_iou": 0.2373046875,
      "loss_num": 0.0159912109375,
      "loss_xval": 0.5546875,
      "num_input_tokens_seen": 227345812,
      "step": 4059
    },
    {
      "epoch": 9.042316258351892,
      "grad_norm": 15.678596496582031,
      "learning_rate": 1e-06,
      "loss": 0.6039,
      "num_input_tokens_seen": 227402332,
      "step": 4060
    },
    {
      "epoch": 9.042316258351892,
      "loss": 0.6879202127456665,
      "loss_ce": 0.00017606788605917245,
      "loss_iou": 0.28125,
      "loss_num": 0.02490234375,
      "loss_xval": 0.6875,
      "num_input_tokens_seen": 227402332,
      "step": 4060
    },
    {
      "epoch": 9.044543429844097,
      "grad_norm": 24.40289306640625,
      "learning_rate": 1e-06,
      "loss": 0.6604,
      "num_input_tokens_seen": 227456080,
      "step": 4061
    },
    {
      "epoch": 9.044543429844097,
      "loss": 0.7870203256607056,
      "loss_ce": 0.00015513686230406165,
      "loss_iou": 0.3046875,
      "loss_num": 0.035400390625,
      "loss_xval": 0.78515625,
      "num_input_tokens_seen": 227456080,
      "step": 4061
    },
    {
      "epoch": 9.046770601336302,
      "grad_norm": 20.38033103942871,
      "learning_rate": 1e-06,
      "loss": 0.5911,
      "num_input_tokens_seen": 227512720,
      "step": 4062
    },
    {
      "epoch": 9.046770601336302,
      "loss": 0.541382908821106,
      "loss_ce": 0.0001231818023370579,
      "loss_iou": 0.25,
      "loss_num": 0.00848388671875,
      "loss_xval": 0.54296875,
      "num_input_tokens_seen": 227512720,
      "step": 4062
    },
    {
      "epoch": 9.048997772828507,
      "grad_norm": 16.504854202270508,
      "learning_rate": 1e-06,
      "loss": 0.5148,
      "num_input_tokens_seen": 227570688,
      "step": 4063
    },
    {
      "epoch": 9.048997772828507,
      "loss": 0.4388590455055237,
      "loss_ce": 0.00013835716526955366,
      "loss_iou": 0.1884765625,
      "loss_num": 0.01214599609375,
      "loss_xval": 0.439453125,
      "num_input_tokens_seen": 227570688,
      "step": 4063
    },
    {
      "epoch": 9.051224944320714,
      "grad_norm": 23.736780166625977,
      "learning_rate": 1e-06,
      "loss": 0.5917,
      "num_input_tokens_seen": 227628128,
      "step": 4064
    },
    {
      "epoch": 9.051224944320714,
      "loss": 0.7519105672836304,
      "loss_ce": 0.0002015778300119564,
      "loss_iou": 0.31640625,
      "loss_num": 0.02392578125,
      "loss_xval": 0.75,
      "num_input_tokens_seen": 227628128,
      "step": 4064
    },
    {
      "epoch": 9.053452115812918,
      "grad_norm": 17.67232894897461,
      "learning_rate": 1e-06,
      "loss": 0.6101,
      "num_input_tokens_seen": 227683884,
      "step": 4065
    },
    {
      "epoch": 9.053452115812918,
      "loss": 0.618810772895813,
      "loss_ce": 0.00015848156181164086,
      "loss_iou": 0.265625,
      "loss_num": 0.017578125,
      "loss_xval": 0.6171875,
      "num_input_tokens_seen": 227683884,
      "step": 4065
    },
    {
      "epoch": 9.055679287305123,
      "grad_norm": 19.514118194580078,
      "learning_rate": 1e-06,
      "loss": 0.8178,
      "num_input_tokens_seen": 227735504,
      "step": 4066
    },
    {
      "epoch": 9.055679287305123,
      "loss": 0.6703531742095947,
      "loss_ce": 0.00018715695478022099,
      "loss_iou": 0.287109375,
      "loss_num": 0.01904296875,
      "loss_xval": 0.671875,
      "num_input_tokens_seen": 227735504,
      "step": 4066
    },
    {
      "epoch": 9.057906458797328,
      "grad_norm": 18.996246337890625,
      "learning_rate": 1e-06,
      "loss": 0.4703,
      "num_input_tokens_seen": 227793336,
      "step": 4067
    },
    {
      "epoch": 9.057906458797328,
      "loss": 0.37396594882011414,
      "loss_ce": 0.00018664645904209465,
      "loss_iou": 0.1630859375,
      "loss_num": 0.0096435546875,
      "loss_xval": 0.373046875,
      "num_input_tokens_seen": 227793336,
      "step": 4067
    },
    {
      "epoch": 9.060133630289533,
      "grad_norm": 13.91976547241211,
      "learning_rate": 1e-06,
      "loss": 0.4175,
      "num_input_tokens_seen": 227851892,
      "step": 4068
    },
    {
      "epoch": 9.060133630289533,
      "loss": 0.5668018460273743,
      "loss_ce": 0.00015145693032536656,
      "loss_iou": 0.255859375,
      "loss_num": 0.0108642578125,
      "loss_xval": 0.56640625,
      "num_input_tokens_seen": 227851892,
      "step": 4068
    },
    {
      "epoch": 9.062360801781738,
      "grad_norm": 15.925460815429688,
      "learning_rate": 1e-06,
      "loss": 0.4987,
      "num_input_tokens_seen": 227904312,
      "step": 4069
    },
    {
      "epoch": 9.062360801781738,
      "loss": 0.5191997289657593,
      "loss_ce": 0.00015678332420066,
      "loss_iou": 0.2431640625,
      "loss_num": 0.00665283203125,
      "loss_xval": 0.51953125,
      "num_input_tokens_seen": 227904312,
      "step": 4069
    },
    {
      "epoch": 9.064587973273943,
      "grad_norm": 19.193389892578125,
      "learning_rate": 1e-06,
      "loss": 0.6064,
      "num_input_tokens_seen": 227959496,
      "step": 4070
    },
    {
      "epoch": 9.064587973273943,
      "loss": 0.5843898057937622,
      "loss_ce": 0.00016130355652421713,
      "loss_iou": 0.25,
      "loss_num": 0.0169677734375,
      "loss_xval": 0.5859375,
      "num_input_tokens_seen": 227959496,
      "step": 4070
    },
    {
      "epoch": 9.066815144766148,
      "grad_norm": 23.820255279541016,
      "learning_rate": 1e-06,
      "loss": 0.4375,
      "num_input_tokens_seen": 228018492,
      "step": 4071
    },
    {
      "epoch": 9.066815144766148,
      "loss": 0.4991843104362488,
      "loss_ce": 0.00016086628602351993,
      "loss_iou": 0.228515625,
      "loss_num": 0.00836181640625,
      "loss_xval": 0.5,
      "num_input_tokens_seen": 228018492,
      "step": 4071
    },
    {
      "epoch": 9.069042316258352,
      "grad_norm": 17.966339111328125,
      "learning_rate": 1e-06,
      "loss": 0.4902,
      "num_input_tokens_seen": 228072824,
      "step": 4072
    },
    {
      "epoch": 9.069042316258352,
      "loss": 0.33019495010375977,
      "loss_ce": 0.00020839170610997826,
      "loss_iou": 0.1337890625,
      "loss_num": 0.01251220703125,
      "loss_xval": 0.330078125,
      "num_input_tokens_seen": 228072824,
      "step": 4072
    },
    {
      "epoch": 9.071269487750557,
      "grad_norm": 19.144359588623047,
      "learning_rate": 1e-06,
      "loss": 0.5818,
      "num_input_tokens_seen": 228126552,
      "step": 4073
    },
    {
      "epoch": 9.071269487750557,
      "loss": 0.5044497847557068,
      "loss_ce": 0.00017729168757796288,
      "loss_iou": 0.20703125,
      "loss_num": 0.01806640625,
      "loss_xval": 0.50390625,
      "num_input_tokens_seen": 228126552,
      "step": 4073
    },
    {
      "epoch": 9.073496659242762,
      "grad_norm": 27.93861961364746,
      "learning_rate": 1e-06,
      "loss": 0.542,
      "num_input_tokens_seen": 228181208,
      "step": 4074
    },
    {
      "epoch": 9.073496659242762,
      "loss": 0.4734833836555481,
      "loss_ce": 0.00015576003352180123,
      "loss_iou": 0.220703125,
      "loss_num": 0.006256103515625,
      "loss_xval": 0.47265625,
      "num_input_tokens_seen": 228181208,
      "step": 4074
    },
    {
      "epoch": 9.075723830734967,
      "grad_norm": 16.074905395507812,
      "learning_rate": 1e-06,
      "loss": 0.5862,
      "num_input_tokens_seen": 228237728,
      "step": 4075
    },
    {
      "epoch": 9.075723830734967,
      "loss": 0.6505739092826843,
      "loss_ce": 0.00018325743440072984,
      "loss_iou": 0.27734375,
      "loss_num": 0.019287109375,
      "loss_xval": 0.6484375,
      "num_input_tokens_seen": 228237728,
      "step": 4075
    },
    {
      "epoch": 9.077951002227172,
      "grad_norm": 26.28703498840332,
      "learning_rate": 1e-06,
      "loss": 0.5817,
      "num_input_tokens_seen": 228294284,
      "step": 4076
    },
    {
      "epoch": 9.077951002227172,
      "loss": 0.46036121249198914,
      "loss_ce": 0.00015614047879353166,
      "loss_iou": 0.205078125,
      "loss_num": 0.010009765625,
      "loss_xval": 0.4609375,
      "num_input_tokens_seen": 228294284,
      "step": 4076
    },
    {
      "epoch": 9.080178173719377,
      "grad_norm": 19.399211883544922,
      "learning_rate": 1e-06,
      "loss": 0.5473,
      "num_input_tokens_seen": 228349092,
      "step": 4077
    },
    {
      "epoch": 9.080178173719377,
      "loss": 0.5990505218505859,
      "loss_ce": 0.00017354718875139952,
      "loss_iou": 0.251953125,
      "loss_num": 0.0189208984375,
      "loss_xval": 0.59765625,
      "num_input_tokens_seen": 228349092,
      "step": 4077
    },
    {
      "epoch": 9.082405345211582,
      "grad_norm": 25.406312942504883,
      "learning_rate": 1e-06,
      "loss": 0.5724,
      "num_input_tokens_seen": 228404620,
      "step": 4078
    },
    {
      "epoch": 9.082405345211582,
      "loss": 0.5832569003105164,
      "loss_ce": 0.00012698877253569663,
      "loss_iou": 0.251953125,
      "loss_num": 0.0157470703125,
      "loss_xval": 0.58203125,
      "num_input_tokens_seen": 228404620,
      "step": 4078
    },
    {
      "epoch": 9.084632516703786,
      "grad_norm": 21.80425262451172,
      "learning_rate": 1e-06,
      "loss": 0.4839,
      "num_input_tokens_seen": 228460136,
      "step": 4079
    },
    {
      "epoch": 9.084632516703786,
      "loss": 0.3352894186973572,
      "loss_ce": 0.00014540574920829386,
      "loss_iou": 0.1376953125,
      "loss_num": 0.011962890625,
      "loss_xval": 0.3359375,
      "num_input_tokens_seen": 228460136,
      "step": 4079
    },
    {
      "epoch": 9.086859688195991,
      "grad_norm": 20.531923294067383,
      "learning_rate": 1e-06,
      "loss": 0.491,
      "num_input_tokens_seen": 228517568,
      "step": 4080
    },
    {
      "epoch": 9.086859688195991,
      "loss": 0.6657657623291016,
      "loss_ce": 0.00014688099327031523,
      "loss_iou": 0.2734375,
      "loss_num": 0.0234375,
      "loss_xval": 0.6640625,
      "num_input_tokens_seen": 228517568,
      "step": 4080
    },
    {
      "epoch": 9.089086859688196,
      "grad_norm": 21.13003921508789,
      "learning_rate": 1e-06,
      "loss": 0.6183,
      "num_input_tokens_seen": 228571092,
      "step": 4081
    },
    {
      "epoch": 9.089086859688196,
      "loss": 0.7186765670776367,
      "loss_ce": 0.0001707267656456679,
      "loss_iou": 0.314453125,
      "loss_num": 0.0181884765625,
      "loss_xval": 0.71875,
      "num_input_tokens_seen": 228571092,
      "step": 4081
    },
    {
      "epoch": 9.091314031180401,
      "grad_norm": 17.36870002746582,
      "learning_rate": 1e-06,
      "loss": 0.7185,
      "num_input_tokens_seen": 228626888,
      "step": 4082
    },
    {
      "epoch": 9.091314031180401,
      "loss": 0.9348142147064209,
      "loss_ce": 0.000365984917152673,
      "loss_iou": 0.384765625,
      "loss_num": 0.032470703125,
      "loss_xval": 0.93359375,
      "num_input_tokens_seen": 228626888,
      "step": 4082
    },
    {
      "epoch": 9.093541202672606,
      "grad_norm": 16.720142364501953,
      "learning_rate": 1e-06,
      "loss": 0.6332,
      "num_input_tokens_seen": 228684328,
      "step": 4083
    },
    {
      "epoch": 9.093541202672606,
      "loss": 0.5341998934745789,
      "loss_ce": 0.0001422719651600346,
      "loss_iou": 0.2265625,
      "loss_num": 0.01611328125,
      "loss_xval": 0.53515625,
      "num_input_tokens_seen": 228684328,
      "step": 4083
    },
    {
      "epoch": 9.09576837416481,
      "grad_norm": 21.377214431762695,
      "learning_rate": 1e-06,
      "loss": 0.6319,
      "num_input_tokens_seen": 228741784,
      "step": 4084
    },
    {
      "epoch": 9.09576837416481,
      "loss": 0.5814862251281738,
      "loss_ce": 0.0011639189906418324,
      "loss_iou": 0.2431640625,
      "loss_num": 0.0189208984375,
      "loss_xval": 0.58203125,
      "num_input_tokens_seen": 228741784,
      "step": 4084
    },
    {
      "epoch": 9.097995545657016,
      "grad_norm": 18.07195472717285,
      "learning_rate": 1e-06,
      "loss": 0.7059,
      "num_input_tokens_seen": 228797852,
      "step": 4085
    },
    {
      "epoch": 9.097995545657016,
      "loss": 0.8597797155380249,
      "loss_ce": 0.00016060096095316112,
      "loss_iou": 0.3828125,
      "loss_num": 0.0189208984375,
      "loss_xval": 0.859375,
      "num_input_tokens_seen": 228797852,
      "step": 4085
    },
    {
      "epoch": 9.10022271714922,
      "grad_norm": 15.08352279663086,
      "learning_rate": 1e-06,
      "loss": 0.5056,
      "num_input_tokens_seen": 228852356,
      "step": 4086
    },
    {
      "epoch": 9.10022271714922,
      "loss": 0.5499671697616577,
      "loss_ce": 0.00016245490405708551,
      "loss_iou": 0.25390625,
      "loss_num": 0.00830078125,
      "loss_xval": 0.55078125,
      "num_input_tokens_seen": 228852356,
      "step": 4086
    },
    {
      "epoch": 9.102449888641425,
      "grad_norm": 23.534421920776367,
      "learning_rate": 1e-06,
      "loss": 0.5026,
      "num_input_tokens_seen": 228904152,
      "step": 4087
    },
    {
      "epoch": 9.102449888641425,
      "loss": 0.4977412819862366,
      "loss_ce": 0.00018270721193403006,
      "loss_iou": 0.2080078125,
      "loss_num": 0.016357421875,
      "loss_xval": 0.498046875,
      "num_input_tokens_seen": 228904152,
      "step": 4087
    },
    {
      "epoch": 9.10467706013363,
      "grad_norm": 19.300045013427734,
      "learning_rate": 1e-06,
      "loss": 0.9714,
      "num_input_tokens_seen": 228957696,
      "step": 4088
    },
    {
      "epoch": 9.10467706013363,
      "loss": 0.8054395914077759,
      "loss_ce": 0.00014173405361361802,
      "loss_iou": 0.32421875,
      "loss_num": 0.03173828125,
      "loss_xval": 0.8046875,
      "num_input_tokens_seen": 228957696,
      "step": 4088
    },
    {
      "epoch": 9.106904231625835,
      "grad_norm": 20.3017578125,
      "learning_rate": 1e-06,
      "loss": 0.3923,
      "num_input_tokens_seen": 229012032,
      "step": 4089
    },
    {
      "epoch": 9.106904231625835,
      "loss": 0.40640291571617126,
      "loss_ce": 0.00015291740419343114,
      "loss_iou": 0.1865234375,
      "loss_num": 0.006866455078125,
      "loss_xval": 0.40625,
      "num_input_tokens_seen": 229012032,
      "step": 4089
    },
    {
      "epoch": 9.10913140311804,
      "grad_norm": 20.25347137451172,
      "learning_rate": 1e-06,
      "loss": 0.5511,
      "num_input_tokens_seen": 229067576,
      "step": 4090
    },
    {
      "epoch": 9.10913140311804,
      "loss": 0.5202056169509888,
      "loss_ce": 0.00012503171456046402,
      "loss_iou": 0.2177734375,
      "loss_num": 0.0169677734375,
      "loss_xval": 0.51953125,
      "num_input_tokens_seen": 229067576,
      "step": 4090
    },
    {
      "epoch": 9.111358574610245,
      "grad_norm": 19.562305450439453,
      "learning_rate": 1e-06,
      "loss": 0.6281,
      "num_input_tokens_seen": 229127092,
      "step": 4091
    },
    {
      "epoch": 9.111358574610245,
      "loss": 0.7418767213821411,
      "loss_ce": 0.00017745466902852058,
      "loss_iou": 0.298828125,
      "loss_num": 0.02880859375,
      "loss_xval": 0.7421875,
      "num_input_tokens_seen": 229127092,
      "step": 4091
    },
    {
      "epoch": 9.11358574610245,
      "grad_norm": 19.772968292236328,
      "learning_rate": 1e-06,
      "loss": 0.5598,
      "num_input_tokens_seen": 229181696,
      "step": 4092
    },
    {
      "epoch": 9.11358574610245,
      "loss": 0.6595751047134399,
      "loss_ce": 0.00015126551443245262,
      "loss_iou": 0.2890625,
      "loss_num": 0.0164794921875,
      "loss_xval": 0.66015625,
      "num_input_tokens_seen": 229181696,
      "step": 4092
    },
    {
      "epoch": 9.115812917594655,
      "grad_norm": 21.81011962890625,
      "learning_rate": 1e-06,
      "loss": 0.6088,
      "num_input_tokens_seen": 229235572,
      "step": 4093
    },
    {
      "epoch": 9.115812917594655,
      "loss": 0.6007593870162964,
      "loss_ce": 0.00017348321853205562,
      "loss_iou": 0.265625,
      "loss_num": 0.0135498046875,
      "loss_xval": 0.6015625,
      "num_input_tokens_seen": 229235572,
      "step": 4093
    },
    {
      "epoch": 9.11804008908686,
      "grad_norm": 19.07417106628418,
      "learning_rate": 1e-06,
      "loss": 0.4561,
      "num_input_tokens_seen": 229291708,
      "step": 4094
    },
    {
      "epoch": 9.11804008908686,
      "loss": 0.5785167217254639,
      "loss_ce": 0.00020860986842308193,
      "loss_iou": 0.236328125,
      "loss_num": 0.02099609375,
      "loss_xval": 0.578125,
      "num_input_tokens_seen": 229291708,
      "step": 4094
    },
    {
      "epoch": 9.120267260579064,
      "grad_norm": 27.573549270629883,
      "learning_rate": 1e-06,
      "loss": 0.6232,
      "num_input_tokens_seen": 229349472,
      "step": 4095
    },
    {
      "epoch": 9.120267260579064,
      "loss": 0.7934918403625488,
      "loss_ce": 0.0002789198188111186,
      "loss_iou": 0.322265625,
      "loss_num": 0.0296630859375,
      "loss_xval": 0.79296875,
      "num_input_tokens_seen": 229349472,
      "step": 4095
    },
    {
      "epoch": 9.122494432071269,
      "grad_norm": 13.100459098815918,
      "learning_rate": 1e-06,
      "loss": 0.7147,
      "num_input_tokens_seen": 229405976,
      "step": 4096
    },
    {
      "epoch": 9.122494432071269,
      "loss": 0.48514890670776367,
      "loss_ce": 0.00016353695536963642,
      "loss_iou": 0.22265625,
      "loss_num": 0.008056640625,
      "loss_xval": 0.484375,
      "num_input_tokens_seen": 229405976,
      "step": 4096
    },
    {
      "epoch": 9.124721603563474,
      "grad_norm": 25.375526428222656,
      "learning_rate": 1e-06,
      "loss": 0.6361,
      "num_input_tokens_seen": 229459716,
      "step": 4097
    },
    {
      "epoch": 9.124721603563474,
      "loss": 0.4884394705295563,
      "loss_ce": 0.00015823112335056067,
      "loss_iou": 0.216796875,
      "loss_num": 0.01080322265625,
      "loss_xval": 0.48828125,
      "num_input_tokens_seen": 229459716,
      "step": 4097
    },
    {
      "epoch": 9.126948775055679,
      "grad_norm": 18.099225997924805,
      "learning_rate": 1e-06,
      "loss": 0.6709,
      "num_input_tokens_seen": 229516212,
      "step": 4098
    },
    {
      "epoch": 9.126948775055679,
      "loss": 0.5384310483932495,
      "loss_ce": 0.0007113110623322427,
      "loss_iou": 0.2353515625,
      "loss_num": 0.01348876953125,
      "loss_xval": 0.5390625,
      "num_input_tokens_seen": 229516212,
      "step": 4098
    },
    {
      "epoch": 9.129175946547884,
      "grad_norm": 22.159957885742188,
      "learning_rate": 1e-06,
      "loss": 0.5206,
      "num_input_tokens_seen": 229570184,
      "step": 4099
    },
    {
      "epoch": 9.129175946547884,
      "loss": 0.638085126876831,
      "loss_ce": 0.00014572580403182656,
      "loss_iou": 0.26171875,
      "loss_num": 0.0234375,
      "loss_xval": 0.63671875,
      "num_input_tokens_seen": 229570184,
      "step": 4099
    },
    {
      "epoch": 9.131403118040089,
      "grad_norm": 19.935977935791016,
      "learning_rate": 1e-06,
      "loss": 0.5385,
      "num_input_tokens_seen": 229624652,
      "step": 4100
    },
    {
      "epoch": 9.131403118040089,
      "loss": 0.6349242925643921,
      "loss_ce": 0.00015866165631450713,
      "loss_iou": 0.28125,
      "loss_num": 0.01446533203125,
      "loss_xval": 0.6328125,
      "num_input_tokens_seen": 229624652,
      "step": 4100
    },
    {
      "epoch": 9.133630289532293,
      "grad_norm": 17.789283752441406,
      "learning_rate": 1e-06,
      "loss": 0.5755,
      "num_input_tokens_seen": 229681300,
      "step": 4101
    },
    {
      "epoch": 9.133630289532293,
      "loss": 0.5260428190231323,
      "loss_ce": 0.00016388525546062738,
      "loss_iou": 0.232421875,
      "loss_num": 0.01239013671875,
      "loss_xval": 0.52734375,
      "num_input_tokens_seen": 229681300,
      "step": 4101
    },
    {
      "epoch": 9.135857461024498,
      "grad_norm": 29.85516357421875,
      "learning_rate": 1e-06,
      "loss": 0.5757,
      "num_input_tokens_seen": 229735816,
      "step": 4102
    },
    {
      "epoch": 9.135857461024498,
      "loss": 0.6520152688026428,
      "loss_ce": 0.00015980206080712378,
      "loss_iou": 0.28125,
      "loss_num": 0.0179443359375,
      "loss_xval": 0.65234375,
      "num_input_tokens_seen": 229735816,
      "step": 4102
    },
    {
      "epoch": 9.138084632516703,
      "grad_norm": 17.78467559814453,
      "learning_rate": 1e-06,
      "loss": 0.4811,
      "num_input_tokens_seen": 229789440,
      "step": 4103
    },
    {
      "epoch": 9.138084632516703,
      "loss": 0.5187010169029236,
      "loss_ce": 0.00014634460967499763,
      "loss_iou": 0.2041015625,
      "loss_num": 0.02197265625,
      "loss_xval": 0.51953125,
      "num_input_tokens_seen": 229789440,
      "step": 4103
    },
    {
      "epoch": 9.140311804008908,
      "grad_norm": 13.813704490661621,
      "learning_rate": 1e-06,
      "loss": 0.5981,
      "num_input_tokens_seen": 229845256,
      "step": 4104
    },
    {
      "epoch": 9.140311804008908,
      "loss": 0.6361100673675537,
      "loss_ce": 0.000245812872890383,
      "loss_iou": 0.25390625,
      "loss_num": 0.0252685546875,
      "loss_xval": 0.63671875,
      "num_input_tokens_seen": 229845256,
      "step": 4104
    },
    {
      "epoch": 9.142538975501113,
      "grad_norm": 17.91861343383789,
      "learning_rate": 1e-06,
      "loss": 0.7475,
      "num_input_tokens_seen": 229902832,
      "step": 4105
    },
    {
      "epoch": 9.142538975501113,
      "loss": 0.8098193407058716,
      "loss_ce": 0.0002490263432264328,
      "loss_iou": 0.369140625,
      "loss_num": 0.0142822265625,
      "loss_xval": 0.80859375,
      "num_input_tokens_seen": 229902832,
      "step": 4105
    },
    {
      "epoch": 9.144766146993318,
      "grad_norm": 34.62513732910156,
      "learning_rate": 1e-06,
      "loss": 0.5171,
      "num_input_tokens_seen": 229959432,
      "step": 4106
    },
    {
      "epoch": 9.144766146993318,
      "loss": 0.6218581199645996,
      "loss_ce": 0.00015404779696837068,
      "loss_iou": 0.265625,
      "loss_num": 0.017822265625,
      "loss_xval": 0.62109375,
      "num_input_tokens_seen": 229959432,
      "step": 4106
    },
    {
      "epoch": 9.146993318485523,
      "grad_norm": 16.90599822998047,
      "learning_rate": 1e-06,
      "loss": 0.6897,
      "num_input_tokens_seen": 230015140,
      "step": 4107
    },
    {
      "epoch": 9.146993318485523,
      "loss": 0.9310408234596252,
      "loss_ce": 0.00013258628314360976,
      "loss_iou": 0.400390625,
      "loss_num": 0.0262451171875,
      "loss_xval": 0.9296875,
      "num_input_tokens_seen": 230015140,
      "step": 4107
    },
    {
      "epoch": 9.14922048997773,
      "grad_norm": 18.637083053588867,
      "learning_rate": 1e-06,
      "loss": 0.5436,
      "num_input_tokens_seen": 230074668,
      "step": 4108
    },
    {
      "epoch": 9.14922048997773,
      "loss": 0.5741239786148071,
      "loss_ce": 0.0001493622548878193,
      "loss_iou": 0.26171875,
      "loss_num": 0.0101318359375,
      "loss_xval": 0.57421875,
      "num_input_tokens_seen": 230074668,
      "step": 4108
    },
    {
      "epoch": 9.151447661469934,
      "grad_norm": 26.293113708496094,
      "learning_rate": 1e-06,
      "loss": 0.5738,
      "num_input_tokens_seen": 230131008,
      "step": 4109
    },
    {
      "epoch": 9.151447661469934,
      "loss": 0.5387140512466431,
      "loss_ce": 0.00013981794472783804,
      "loss_iou": 0.220703125,
      "loss_num": 0.0196533203125,
      "loss_xval": 0.5390625,
      "num_input_tokens_seen": 230131008,
      "step": 4109
    },
    {
      "epoch": 9.153674832962139,
      "grad_norm": 99.2131576538086,
      "learning_rate": 1e-06,
      "loss": 0.6848,
      "num_input_tokens_seen": 230186684,
      "step": 4110
    },
    {
      "epoch": 9.153674832962139,
      "loss": 0.8059197068214417,
      "loss_ce": 0.00025563541566953063,
      "loss_iou": 0.35546875,
      "loss_num": 0.0191650390625,
      "loss_xval": 0.8046875,
      "num_input_tokens_seen": 230186684,
      "step": 4110
    },
    {
      "epoch": 9.155902004454344,
      "grad_norm": 11.691954612731934,
      "learning_rate": 1e-06,
      "loss": 0.5424,
      "num_input_tokens_seen": 230244428,
      "step": 4111
    },
    {
      "epoch": 9.155902004454344,
      "loss": 0.5812520980834961,
      "loss_ce": 0.00019743737357202917,
      "loss_iou": 0.22265625,
      "loss_num": 0.0269775390625,
      "loss_xval": 0.58203125,
      "num_input_tokens_seen": 230244428,
      "step": 4111
    },
    {
      "epoch": 9.158129175946549,
      "grad_norm": 23.4112606048584,
      "learning_rate": 1e-06,
      "loss": 0.5201,
      "num_input_tokens_seen": 230298452,
      "step": 4112
    },
    {
      "epoch": 9.158129175946549,
      "loss": 0.6187844276428223,
      "loss_ce": 0.00013208799646236002,
      "loss_iou": 0.2490234375,
      "loss_num": 0.024169921875,
      "loss_xval": 0.6171875,
      "num_input_tokens_seen": 230298452,
      "step": 4112
    },
    {
      "epoch": 9.160356347438753,
      "grad_norm": 24.178136825561523,
      "learning_rate": 1e-06,
      "loss": 0.4805,
      "num_input_tokens_seen": 230352684,
      "step": 4113
    },
    {
      "epoch": 9.160356347438753,
      "loss": 0.5137137174606323,
      "loss_ce": 0.00016394033445976675,
      "loss_iou": 0.2451171875,
      "loss_num": 0.0045166015625,
      "loss_xval": 0.51171875,
      "num_input_tokens_seen": 230352684,
      "step": 4113
    },
    {
      "epoch": 9.162583518930958,
      "grad_norm": 17.317472457885742,
      "learning_rate": 1e-06,
      "loss": 0.5563,
      "num_input_tokens_seen": 230408152,
      "step": 4114
    },
    {
      "epoch": 9.162583518930958,
      "loss": 0.6397294998168945,
      "loss_ce": 0.0001421074557583779,
      "loss_iou": 0.26953125,
      "loss_num": 0.0198974609375,
      "loss_xval": 0.640625,
      "num_input_tokens_seen": 230408152,
      "step": 4114
    },
    {
      "epoch": 9.164810690423163,
      "grad_norm": 18.378246307373047,
      "learning_rate": 1e-06,
      "loss": 0.57,
      "num_input_tokens_seen": 230464552,
      "step": 4115
    },
    {
      "epoch": 9.164810690423163,
      "loss": 0.5939757823944092,
      "loss_ce": 0.00016474167932756245,
      "loss_iou": 0.212890625,
      "loss_num": 0.03369140625,
      "loss_xval": 0.59375,
      "num_input_tokens_seen": 230464552,
      "step": 4115
    },
    {
      "epoch": 9.167037861915368,
      "grad_norm": 13.769119262695312,
      "learning_rate": 1e-06,
      "loss": 0.6041,
      "num_input_tokens_seen": 230520760,
      "step": 4116
    },
    {
      "epoch": 9.167037861915368,
      "loss": 0.7275907397270203,
      "loss_ce": 0.0001737515558488667,
      "loss_iou": 0.306640625,
      "loss_num": 0.0224609375,
      "loss_xval": 0.7265625,
      "num_input_tokens_seen": 230520760,
      "step": 4116
    },
    {
      "epoch": 9.169265033407573,
      "grad_norm": 16.299108505249023,
      "learning_rate": 1e-06,
      "loss": 0.6447,
      "num_input_tokens_seen": 230577540,
      "step": 4117
    },
    {
      "epoch": 9.169265033407573,
      "loss": 0.7198707461357117,
      "loss_ce": 0.0002662655897438526,
      "loss_iou": 0.29296875,
      "loss_num": 0.0269775390625,
      "loss_xval": 0.71875,
      "num_input_tokens_seen": 230577540,
      "step": 4117
    },
    {
      "epoch": 9.171492204899778,
      "grad_norm": 20.457033157348633,
      "learning_rate": 1e-06,
      "loss": 0.7404,
      "num_input_tokens_seen": 230632328,
      "step": 4118
    },
    {
      "epoch": 9.171492204899778,
      "loss": 0.7303974032402039,
      "loss_ce": 0.0001728244824334979,
      "loss_iou": 0.3125,
      "loss_num": 0.021240234375,
      "loss_xval": 0.73046875,
      "num_input_tokens_seen": 230632328,
      "step": 4118
    },
    {
      "epoch": 9.173719376391983,
      "grad_norm": 17.638404846191406,
      "learning_rate": 1e-06,
      "loss": 0.5783,
      "num_input_tokens_seen": 230687228,
      "step": 4119
    },
    {
      "epoch": 9.173719376391983,
      "loss": 0.46454915404319763,
      "loss_ce": 0.00019369515939615667,
      "loss_iou": 0.177734375,
      "loss_num": 0.021728515625,
      "loss_xval": 0.46484375,
      "num_input_tokens_seen": 230687228,
      "step": 4119
    },
    {
      "epoch": 9.175946547884188,
      "grad_norm": 18.509496688842773,
      "learning_rate": 1e-06,
      "loss": 0.6066,
      "num_input_tokens_seen": 230743560,
      "step": 4120
    },
    {
      "epoch": 9.175946547884188,
      "loss": 0.7997314929962158,
      "loss_ce": 0.00017094583017751575,
      "loss_iou": 0.33984375,
      "loss_num": 0.0234375,
      "loss_xval": 0.80078125,
      "num_input_tokens_seen": 230743560,
      "step": 4120
    },
    {
      "epoch": 9.178173719376392,
      "grad_norm": 16.593032836914062,
      "learning_rate": 1e-06,
      "loss": 0.6728,
      "num_input_tokens_seen": 230800968,
      "step": 4121
    },
    {
      "epoch": 9.178173719376392,
      "loss": 0.949892520904541,
      "loss_ce": 0.00018545052444096655,
      "loss_iou": 0.37890625,
      "loss_num": 0.0380859375,
      "loss_xval": 0.94921875,
      "num_input_tokens_seen": 230800968,
      "step": 4121
    },
    {
      "epoch": 9.180400890868597,
      "grad_norm": 18.3552303314209,
      "learning_rate": 1e-06,
      "loss": 0.6935,
      "num_input_tokens_seen": 230857948,
      "step": 4122
    },
    {
      "epoch": 9.180400890868597,
      "loss": 0.7443253993988037,
      "loss_ce": 0.00018475553952157497,
      "loss_iou": 0.3125,
      "loss_num": 0.02392578125,
      "loss_xval": 0.7421875,
      "num_input_tokens_seen": 230857948,
      "step": 4122
    },
    {
      "epoch": 9.182628062360802,
      "grad_norm": 17.54131317138672,
      "learning_rate": 1e-06,
      "loss": 0.4531,
      "num_input_tokens_seen": 230912532,
      "step": 4123
    },
    {
      "epoch": 9.182628062360802,
      "loss": 0.43069642782211304,
      "loss_ce": 0.0001544352126074955,
      "loss_iou": 0.1796875,
      "loss_num": 0.01409912109375,
      "loss_xval": 0.4296875,
      "num_input_tokens_seen": 230912532,
      "step": 4123
    },
    {
      "epoch": 9.184855233853007,
      "grad_norm": 15.92164421081543,
      "learning_rate": 1e-06,
      "loss": 0.416,
      "num_input_tokens_seen": 230967364,
      "step": 4124
    },
    {
      "epoch": 9.184855233853007,
      "loss": 0.37866830825805664,
      "loss_ce": 0.00012826549937017262,
      "loss_iou": 0.158203125,
      "loss_num": 0.0123291015625,
      "loss_xval": 0.37890625,
      "num_input_tokens_seen": 230967364,
      "step": 4124
    },
    {
      "epoch": 9.187082405345212,
      "grad_norm": 14.523744583129883,
      "learning_rate": 1e-06,
      "loss": 0.5998,
      "num_input_tokens_seen": 231023780,
      "step": 4125
    },
    {
      "epoch": 9.187082405345212,
      "loss": 0.4754905700683594,
      "loss_ce": 0.00014873658074066043,
      "loss_iou": 0.2177734375,
      "loss_num": 0.00799560546875,
      "loss_xval": 0.474609375,
      "num_input_tokens_seen": 231023780,
      "step": 4125
    },
    {
      "epoch": 9.189309576837417,
      "grad_norm": 41.82172393798828,
      "learning_rate": 1e-06,
      "loss": 0.6878,
      "num_input_tokens_seen": 231081652,
      "step": 4126
    },
    {
      "epoch": 9.189309576837417,
      "loss": 0.8278828859329224,
      "loss_ce": 0.0002461716067045927,
      "loss_iou": 0.33984375,
      "loss_num": 0.029296875,
      "loss_xval": 0.828125,
      "num_input_tokens_seen": 231081652,
      "step": 4126
    },
    {
      "epoch": 9.191536748329622,
      "grad_norm": 13.996521949768066,
      "learning_rate": 1e-06,
      "loss": 0.627,
      "num_input_tokens_seen": 231137804,
      "step": 4127
    },
    {
      "epoch": 9.191536748329622,
      "loss": 0.6197777390480042,
      "loss_ce": 0.0001488196139689535,
      "loss_iou": 0.2470703125,
      "loss_num": 0.0250244140625,
      "loss_xval": 0.62109375,
      "num_input_tokens_seen": 231137804,
      "step": 4127
    },
    {
      "epoch": 9.193763919821826,
      "grad_norm": 23.2529239654541,
      "learning_rate": 1e-06,
      "loss": 0.5141,
      "num_input_tokens_seen": 231192268,
      "step": 4128
    },
    {
      "epoch": 9.193763919821826,
      "loss": 0.5340644717216492,
      "loss_ce": 0.00015944175538606942,
      "loss_iou": 0.236328125,
      "loss_num": 0.0120849609375,
      "loss_xval": 0.53515625,
      "num_input_tokens_seen": 231192268,
      "step": 4128
    },
    {
      "epoch": 9.195991091314031,
      "grad_norm": 16.674602508544922,
      "learning_rate": 1e-06,
      "loss": 0.5281,
      "num_input_tokens_seen": 231246748,
      "step": 4129
    },
    {
      "epoch": 9.195991091314031,
      "loss": 0.5064660310745239,
      "loss_ce": 0.00011834965698653832,
      "loss_iou": 0.2197265625,
      "loss_num": 0.0133056640625,
      "loss_xval": 0.5078125,
      "num_input_tokens_seen": 231246748,
      "step": 4129
    },
    {
      "epoch": 9.198218262806236,
      "grad_norm": 18.663047790527344,
      "learning_rate": 1e-06,
      "loss": 0.6424,
      "num_input_tokens_seen": 231302740,
      "step": 4130
    },
    {
      "epoch": 9.198218262806236,
      "loss": 0.5936636924743652,
      "loss_ce": 0.0001578406518092379,
      "loss_iou": 0.2392578125,
      "loss_num": 0.02294921875,
      "loss_xval": 0.59375,
      "num_input_tokens_seen": 231302740,
      "step": 4130
    },
    {
      "epoch": 9.200445434298441,
      "grad_norm": 39.11851119995117,
      "learning_rate": 1e-06,
      "loss": 0.4673,
      "num_input_tokens_seen": 231356712,
      "step": 4131
    },
    {
      "epoch": 9.200445434298441,
      "loss": 0.3150908052921295,
      "loss_ce": 0.0001494048337917775,
      "loss_iou": 0.126953125,
      "loss_num": 0.0123291015625,
      "loss_xval": 0.314453125,
      "num_input_tokens_seen": 231356712,
      "step": 4131
    },
    {
      "epoch": 9.202672605790646,
      "grad_norm": 16.948453903198242,
      "learning_rate": 1e-06,
      "loss": 0.537,
      "num_input_tokens_seen": 231411692,
      "step": 4132
    },
    {
      "epoch": 9.202672605790646,
      "loss": 0.3508445620536804,
      "loss_ce": 0.00013653644418809563,
      "loss_iou": 0.1572265625,
      "loss_num": 0.00726318359375,
      "loss_xval": 0.3515625,
      "num_input_tokens_seen": 231411692,
      "step": 4132
    },
    {
      "epoch": 9.20489977728285,
      "grad_norm": 18.197696685791016,
      "learning_rate": 1e-06,
      "loss": 0.5971,
      "num_input_tokens_seen": 231468140,
      "step": 4133
    },
    {
      "epoch": 9.20489977728285,
      "loss": 0.47916704416275024,
      "loss_ce": 0.0001631466147955507,
      "loss_iou": 0.203125,
      "loss_num": 0.01470947265625,
      "loss_xval": 0.478515625,
      "num_input_tokens_seen": 231468140,
      "step": 4133
    },
    {
      "epoch": 9.207126948775056,
      "grad_norm": 18.140928268432617,
      "learning_rate": 1e-06,
      "loss": 0.4389,
      "num_input_tokens_seen": 231524076,
      "step": 4134
    },
    {
      "epoch": 9.207126948775056,
      "loss": 0.484072208404541,
      "loss_ce": 0.0001854776928666979,
      "loss_iou": 0.2021484375,
      "loss_num": 0.0159912109375,
      "loss_xval": 0.484375,
      "num_input_tokens_seen": 231524076,
      "step": 4134
    },
    {
      "epoch": 9.20935412026726,
      "grad_norm": 31.31038475036621,
      "learning_rate": 1e-06,
      "loss": 0.5013,
      "num_input_tokens_seen": 231580384,
      "step": 4135
    },
    {
      "epoch": 9.20935412026726,
      "loss": 0.5428678393363953,
      "loss_ce": 0.00014321647176984698,
      "loss_iou": 0.23046875,
      "loss_num": 0.016357421875,
      "loss_xval": 0.54296875,
      "num_input_tokens_seen": 231580384,
      "step": 4135
    },
    {
      "epoch": 9.211581291759465,
      "grad_norm": 18.321279525756836,
      "learning_rate": 1e-06,
      "loss": 0.482,
      "num_input_tokens_seen": 231635920,
      "step": 4136
    },
    {
      "epoch": 9.211581291759465,
      "loss": 0.5265001058578491,
      "loss_ce": 0.0001329151273239404,
      "loss_iou": 0.2314453125,
      "loss_num": 0.01263427734375,
      "loss_xval": 0.52734375,
      "num_input_tokens_seen": 231635920,
      "step": 4136
    },
    {
      "epoch": 9.21380846325167,
      "grad_norm": 15.607020378112793,
      "learning_rate": 1e-06,
      "loss": 0.7133,
      "num_input_tokens_seen": 231692860,
      "step": 4137
    },
    {
      "epoch": 9.21380846325167,
      "loss": 0.6896048784255981,
      "loss_ce": 0.00015171918494161218,
      "loss_iou": 0.298828125,
      "loss_num": 0.0181884765625,
      "loss_xval": 0.6875,
      "num_input_tokens_seen": 231692860,
      "step": 4137
    },
    {
      "epoch": 9.216035634743875,
      "grad_norm": 20.895606994628906,
      "learning_rate": 1e-06,
      "loss": 0.5896,
      "num_input_tokens_seen": 231752072,
      "step": 4138
    },
    {
      "epoch": 9.216035634743875,
      "loss": 0.5861691236495972,
      "loss_ce": 0.00017053935152944177,
      "loss_iou": 0.2431640625,
      "loss_num": 0.0198974609375,
      "loss_xval": 0.5859375,
      "num_input_tokens_seen": 231752072,
      "step": 4138
    },
    {
      "epoch": 9.21826280623608,
      "grad_norm": 21.63951873779297,
      "learning_rate": 1e-06,
      "loss": 0.5425,
      "num_input_tokens_seen": 231806848,
      "step": 4139
    },
    {
      "epoch": 9.21826280623608,
      "loss": 0.6595204472541809,
      "loss_ce": 0.0004628373426385224,
      "loss_iou": 0.296875,
      "loss_num": 0.0133056640625,
      "loss_xval": 0.66015625,
      "num_input_tokens_seen": 231806848,
      "step": 4139
    },
    {
      "epoch": 9.220489977728285,
      "grad_norm": 17.077104568481445,
      "learning_rate": 1e-06,
      "loss": 0.6683,
      "num_input_tokens_seen": 231862472,
      "step": 4140
    },
    {
      "epoch": 9.220489977728285,
      "loss": 0.7157517075538635,
      "loss_ce": 0.00017556847888045013,
      "loss_iou": 0.30078125,
      "loss_num": 0.0228271484375,
      "loss_xval": 0.71484375,
      "num_input_tokens_seen": 231862472,
      "step": 4140
    },
    {
      "epoch": 9.22271714922049,
      "grad_norm": 15.992926597595215,
      "learning_rate": 1e-06,
      "loss": 0.5696,
      "num_input_tokens_seen": 231916984,
      "step": 4141
    },
    {
      "epoch": 9.22271714922049,
      "loss": 0.48255985975265503,
      "loss_ce": 0.00013797251449432224,
      "loss_iou": 0.2138671875,
      "loss_num": 0.01080322265625,
      "loss_xval": 0.482421875,
      "num_input_tokens_seen": 231916984,
      "step": 4141
    },
    {
      "epoch": 9.224944320712694,
      "grad_norm": 17.786970138549805,
      "learning_rate": 1e-06,
      "loss": 0.6369,
      "num_input_tokens_seen": 231971832,
      "step": 4142
    },
    {
      "epoch": 9.224944320712694,
      "loss": 0.581957221031189,
      "loss_ce": 0.00017014719196595252,
      "loss_iou": 0.267578125,
      "loss_num": 0.009521484375,
      "loss_xval": 0.58203125,
      "num_input_tokens_seen": 231971832,
      "step": 4142
    },
    {
      "epoch": 9.2271714922049,
      "grad_norm": 17.536426544189453,
      "learning_rate": 1e-06,
      "loss": 0.7411,
      "num_input_tokens_seen": 232029912,
      "step": 4143
    },
    {
      "epoch": 9.2271714922049,
      "loss": 0.6547312140464783,
      "loss_ce": 0.00019021191110368818,
      "loss_iou": 0.28515625,
      "loss_num": 0.01708984375,
      "loss_xval": 0.65625,
      "num_input_tokens_seen": 232029912,
      "step": 4143
    },
    {
      "epoch": 9.229398663697104,
      "grad_norm": 17.012691497802734,
      "learning_rate": 1e-06,
      "loss": 0.5203,
      "num_input_tokens_seen": 232089236,
      "step": 4144
    },
    {
      "epoch": 9.229398663697104,
      "loss": 0.5632002353668213,
      "loss_ce": 0.00021197632304392755,
      "loss_iou": 0.2451171875,
      "loss_num": 0.01446533203125,
      "loss_xval": 0.5625,
      "num_input_tokens_seen": 232089236,
      "step": 4144
    },
    {
      "epoch": 9.231625835189309,
      "grad_norm": 17.848848342895508,
      "learning_rate": 1e-06,
      "loss": 0.4004,
      "num_input_tokens_seen": 232144528,
      "step": 4145
    },
    {
      "epoch": 9.231625835189309,
      "loss": 0.3932184875011444,
      "loss_ce": 0.0002741398348007351,
      "loss_iou": 0.169921875,
      "loss_num": 0.0106201171875,
      "loss_xval": 0.392578125,
      "num_input_tokens_seen": 232144528,
      "step": 4145
    },
    {
      "epoch": 9.233853006681514,
      "grad_norm": 14.590682983398438,
      "learning_rate": 1e-06,
      "loss": 0.476,
      "num_input_tokens_seen": 232202272,
      "step": 4146
    },
    {
      "epoch": 9.233853006681514,
      "loss": 0.40485870838165283,
      "loss_ce": 0.00013458858302328736,
      "loss_iou": 0.1728515625,
      "loss_num": 0.0118408203125,
      "loss_xval": 0.404296875,
      "num_input_tokens_seen": 232202272,
      "step": 4146
    },
    {
      "epoch": 9.236080178173719,
      "grad_norm": 22.08498764038086,
      "learning_rate": 1e-06,
      "loss": 0.5566,
      "num_input_tokens_seen": 232257404,
      "step": 4147
    },
    {
      "epoch": 9.236080178173719,
      "loss": 0.687615156173706,
      "loss_ce": 0.0002372571761952713,
      "loss_iou": 0.27734375,
      "loss_num": 0.0263671875,
      "loss_xval": 0.6875,
      "num_input_tokens_seen": 232257404,
      "step": 4147
    },
    {
      "epoch": 9.238307349665924,
      "grad_norm": 33.60602569580078,
      "learning_rate": 1e-06,
      "loss": 0.5149,
      "num_input_tokens_seen": 232313104,
      "step": 4148
    },
    {
      "epoch": 9.238307349665924,
      "loss": 0.46257176995277405,
      "loss_ce": 0.00016943998343776911,
      "loss_iou": 0.208984375,
      "loss_num": 0.0087890625,
      "loss_xval": 0.462890625,
      "num_input_tokens_seen": 232313104,
      "step": 4148
    },
    {
      "epoch": 9.240534521158128,
      "grad_norm": 20.825382232666016,
      "learning_rate": 1e-06,
      "loss": 0.6308,
      "num_input_tokens_seen": 232368704,
      "step": 4149
    },
    {
      "epoch": 9.240534521158128,
      "loss": 0.7248323559761047,
      "loss_ce": 0.00022297201212495565,
      "loss_iou": 0.291015625,
      "loss_num": 0.0283203125,
      "loss_xval": 0.7265625,
      "num_input_tokens_seen": 232368704,
      "step": 4149
    },
    {
      "epoch": 9.242761692650333,
      "grad_norm": 16.274980545043945,
      "learning_rate": 1e-06,
      "loss": 0.611,
      "num_input_tokens_seen": 232423564,
      "step": 4150
    },
    {
      "epoch": 9.242761692650333,
      "loss": 0.8920093178749084,
      "loss_ce": 0.0001636209199205041,
      "loss_iou": 0.337890625,
      "loss_num": 0.04296875,
      "loss_xval": 0.890625,
      "num_input_tokens_seen": 232423564,
      "step": 4150
    },
    {
      "epoch": 9.244988864142538,
      "grad_norm": 17.573843002319336,
      "learning_rate": 1e-06,
      "loss": 0.4551,
      "num_input_tokens_seen": 232480304,
      "step": 4151
    },
    {
      "epoch": 9.244988864142538,
      "loss": 0.4830518662929535,
      "loss_ce": 0.0001416985469404608,
      "loss_iou": 0.2060546875,
      "loss_num": 0.01409912109375,
      "loss_xval": 0.482421875,
      "num_input_tokens_seen": 232480304,
      "step": 4151
    },
    {
      "epoch": 9.247216035634743,
      "grad_norm": 20.984481811523438,
      "learning_rate": 1e-06,
      "loss": 0.5364,
      "num_input_tokens_seen": 232536816,
      "step": 4152
    },
    {
      "epoch": 9.247216035634743,
      "loss": 0.46505963802337646,
      "loss_ce": 0.00015485798940062523,
      "loss_iou": 0.1982421875,
      "loss_num": 0.01373291015625,
      "loss_xval": 0.46484375,
      "num_input_tokens_seen": 232536816,
      "step": 4152
    },
    {
      "epoch": 9.249443207126948,
      "grad_norm": 16.171337127685547,
      "learning_rate": 1e-06,
      "loss": 0.4752,
      "num_input_tokens_seen": 232592992,
      "step": 4153
    },
    {
      "epoch": 9.249443207126948,
      "loss": 0.5165247917175293,
      "loss_ce": 0.0004115123301744461,
      "loss_iou": 0.220703125,
      "loss_num": 0.01519775390625,
      "loss_xval": 0.515625,
      "num_input_tokens_seen": 232592992,
      "step": 4153
    },
    {
      "epoch": 9.251670378619155,
      "grad_norm": 21.20688247680664,
      "learning_rate": 1e-06,
      "loss": 0.5654,
      "num_input_tokens_seen": 232649224,
      "step": 4154
    },
    {
      "epoch": 9.251670378619155,
      "loss": 0.4512087404727936,
      "loss_ce": 0.00015892238297965378,
      "loss_iou": 0.1865234375,
      "loss_num": 0.01544189453125,
      "loss_xval": 0.451171875,
      "num_input_tokens_seen": 232649224,
      "step": 4154
    },
    {
      "epoch": 9.25389755011136,
      "grad_norm": 12.871101379394531,
      "learning_rate": 1e-06,
      "loss": 0.4954,
      "num_input_tokens_seen": 232706068,
      "step": 4155
    },
    {
      "epoch": 9.25389755011136,
      "loss": 0.5504859685897827,
      "loss_ce": 0.00019304068700876087,
      "loss_iou": 0.248046875,
      "loss_num": 0.010986328125,
      "loss_xval": 0.55078125,
      "num_input_tokens_seen": 232706068,
      "step": 4155
    },
    {
      "epoch": 9.256124721603564,
      "grad_norm": 21.039005279541016,
      "learning_rate": 1e-06,
      "loss": 0.6367,
      "num_input_tokens_seen": 232759908,
      "step": 4156
    },
    {
      "epoch": 9.256124721603564,
      "loss": 0.6510411500930786,
      "loss_ce": 0.00016224203864112496,
      "loss_iou": 0.2890625,
      "loss_num": 0.014404296875,
      "loss_xval": 0.65234375,
      "num_input_tokens_seen": 232759908,
      "step": 4156
    },
    {
      "epoch": 9.25835189309577,
      "grad_norm": 17.5001163482666,
      "learning_rate": 1e-06,
      "loss": 0.5683,
      "num_input_tokens_seen": 232815104,
      "step": 4157
    },
    {
      "epoch": 9.25835189309577,
      "loss": 0.57974773645401,
      "loss_ce": 0.00015791512851137668,
      "loss_iou": 0.23046875,
      "loss_num": 0.0238037109375,
      "loss_xval": 0.578125,
      "num_input_tokens_seen": 232815104,
      "step": 4157
    },
    {
      "epoch": 9.260579064587974,
      "grad_norm": 13.760968208312988,
      "learning_rate": 1e-06,
      "loss": 0.49,
      "num_input_tokens_seen": 232870704,
      "step": 4158
    },
    {
      "epoch": 9.260579064587974,
      "loss": 0.553850531578064,
      "loss_ce": 0.00013953927555121481,
      "loss_iou": 0.2392578125,
      "loss_num": 0.01495361328125,
      "loss_xval": 0.5546875,
      "num_input_tokens_seen": 232870704,
      "step": 4158
    },
    {
      "epoch": 9.262806236080179,
      "grad_norm": 20.684478759765625,
      "learning_rate": 1e-06,
      "loss": 0.578,
      "num_input_tokens_seen": 232925348,
      "step": 4159
    },
    {
      "epoch": 9.262806236080179,
      "loss": 0.4780513644218445,
      "loss_ce": 0.000146102363942191,
      "loss_iou": 0.212890625,
      "loss_num": 0.010498046875,
      "loss_xval": 0.478515625,
      "num_input_tokens_seen": 232925348,
      "step": 4159
    },
    {
      "epoch": 9.265033407572384,
      "grad_norm": 13.17013931274414,
      "learning_rate": 1e-06,
      "loss": 0.6326,
      "num_input_tokens_seen": 232982220,
      "step": 4160
    },
    {
      "epoch": 9.265033407572384,
      "loss": 0.7166314125061035,
      "loss_ce": 0.00020072060578968376,
      "loss_iou": 0.302734375,
      "loss_num": 0.0224609375,
      "loss_xval": 0.71484375,
      "num_input_tokens_seen": 232982220,
      "step": 4160
    },
    {
      "epoch": 9.267260579064589,
      "grad_norm": 21.762144088745117,
      "learning_rate": 1e-06,
      "loss": 0.6647,
      "num_input_tokens_seen": 233039268,
      "step": 4161
    },
    {
      "epoch": 9.267260579064589,
      "loss": 0.6170930862426758,
      "loss_ce": 0.00014968152390792966,
      "loss_iou": 0.2451171875,
      "loss_num": 0.025390625,
      "loss_xval": 0.6171875,
      "num_input_tokens_seen": 233039268,
      "step": 4161
    },
    {
      "epoch": 9.269487750556793,
      "grad_norm": 19.671171188354492,
      "learning_rate": 1e-06,
      "loss": 0.4354,
      "num_input_tokens_seen": 233095824,
      "step": 4162
    },
    {
      "epoch": 9.269487750556793,
      "loss": 0.41189324855804443,
      "loss_ce": 0.00015009564231149852,
      "loss_iou": 0.1845703125,
      "loss_num": 0.0087890625,
      "loss_xval": 0.412109375,
      "num_input_tokens_seen": 233095824,
      "step": 4162
    },
    {
      "epoch": 9.271714922048998,
      "grad_norm": 19.537986755371094,
      "learning_rate": 1e-06,
      "loss": 0.5711,
      "num_input_tokens_seen": 233152836,
      "step": 4163
    },
    {
      "epoch": 9.271714922048998,
      "loss": 0.654711902141571,
      "loss_ce": 0.00017089179891627282,
      "loss_iou": 0.298828125,
      "loss_num": 0.01141357421875,
      "loss_xval": 0.65625,
      "num_input_tokens_seen": 233152836,
      "step": 4163
    },
    {
      "epoch": 9.273942093541203,
      "grad_norm": 15.77131462097168,
      "learning_rate": 1e-06,
      "loss": 0.566,
      "num_input_tokens_seen": 233207496,
      "step": 4164
    },
    {
      "epoch": 9.273942093541203,
      "loss": 0.5520115494728088,
      "loss_ce": 0.00019268158939667046,
      "loss_iou": 0.2353515625,
      "loss_num": 0.01611328125,
      "loss_xval": 0.55078125,
      "num_input_tokens_seen": 233207496,
      "step": 4164
    },
    {
      "epoch": 9.276169265033408,
      "grad_norm": 13.012890815734863,
      "learning_rate": 1e-06,
      "loss": 0.5237,
      "num_input_tokens_seen": 233264596,
      "step": 4165
    },
    {
      "epoch": 9.276169265033408,
      "loss": 0.46829238533973694,
      "loss_ce": 0.00015275325858965516,
      "loss_iou": 0.1875,
      "loss_num": 0.018798828125,
      "loss_xval": 0.46875,
      "num_input_tokens_seen": 233264596,
      "step": 4165
    },
    {
      "epoch": 9.278396436525613,
      "grad_norm": 24.705101013183594,
      "learning_rate": 1e-06,
      "loss": 0.6364,
      "num_input_tokens_seen": 233321472,
      "step": 4166
    },
    {
      "epoch": 9.278396436525613,
      "loss": 0.5150579214096069,
      "loss_ce": 0.00016535192844457924,
      "loss_iou": 0.228515625,
      "loss_num": 0.0115966796875,
      "loss_xval": 0.515625,
      "num_input_tokens_seen": 233321472,
      "step": 4166
    },
    {
      "epoch": 9.280623608017818,
      "grad_norm": 15.74944019317627,
      "learning_rate": 1e-06,
      "loss": 0.5495,
      "num_input_tokens_seen": 233377064,
      "step": 4167
    },
    {
      "epoch": 9.280623608017818,
      "loss": 0.5138007402420044,
      "loss_ce": 0.00022045343939680606,
      "loss_iou": 0.23046875,
      "loss_num": 0.01055908203125,
      "loss_xval": 0.51171875,
      "num_input_tokens_seen": 233377064,
      "step": 4167
    },
    {
      "epoch": 9.282850779510023,
      "grad_norm": 25.632081985473633,
      "learning_rate": 1e-06,
      "loss": 0.5602,
      "num_input_tokens_seen": 233432460,
      "step": 4168
    },
    {
      "epoch": 9.282850779510023,
      "loss": 0.6908320784568787,
      "loss_ce": 0.0001582444820087403,
      "loss_iou": 0.26953125,
      "loss_num": 0.0299072265625,
      "loss_xval": 0.69140625,
      "num_input_tokens_seen": 233432460,
      "step": 4168
    },
    {
      "epoch": 9.285077951002227,
      "grad_norm": 21.129497528076172,
      "learning_rate": 1e-06,
      "loss": 0.6149,
      "num_input_tokens_seen": 233489932,
      "step": 4169
    },
    {
      "epoch": 9.285077951002227,
      "loss": 0.8304713368415833,
      "loss_ce": 0.00014905552961863577,
      "loss_iou": 0.337890625,
      "loss_num": 0.0306396484375,
      "loss_xval": 0.83203125,
      "num_input_tokens_seen": 233489932,
      "step": 4169
    },
    {
      "epoch": 9.287305122494432,
      "grad_norm": 18.864761352539062,
      "learning_rate": 1e-06,
      "loss": 0.6289,
      "num_input_tokens_seen": 233546048,
      "step": 4170
    },
    {
      "epoch": 9.287305122494432,
      "loss": 0.4682886600494385,
      "loss_ce": 0.00014900718815624714,
      "loss_iou": 0.2109375,
      "loss_num": 0.00946044921875,
      "loss_xval": 0.46875,
      "num_input_tokens_seen": 233546048,
      "step": 4170
    },
    {
      "epoch": 9.289532293986637,
      "grad_norm": 20.65398597717285,
      "learning_rate": 1e-06,
      "loss": 0.5547,
      "num_input_tokens_seen": 233601528,
      "step": 4171
    },
    {
      "epoch": 9.289532293986637,
      "loss": 0.6244362592697144,
      "loss_ce": 0.00016863204655237496,
      "loss_iou": 0.2734375,
      "loss_num": 0.01556396484375,
      "loss_xval": 0.625,
      "num_input_tokens_seen": 233601528,
      "step": 4171
    },
    {
      "epoch": 9.291759465478842,
      "grad_norm": 23.496076583862305,
      "learning_rate": 1e-06,
      "loss": 0.5856,
      "num_input_tokens_seen": 233657160,
      "step": 4172
    },
    {
      "epoch": 9.291759465478842,
      "loss": 0.5482483506202698,
      "loss_ce": 0.00015264737885445356,
      "loss_iou": 0.1982421875,
      "loss_num": 0.0303955078125,
      "loss_xval": 0.546875,
      "num_input_tokens_seen": 233657160,
      "step": 4172
    },
    {
      "epoch": 9.293986636971047,
      "grad_norm": 19.034543991088867,
      "learning_rate": 1e-06,
      "loss": 0.7883,
      "num_input_tokens_seen": 233713276,
      "step": 4173
    },
    {
      "epoch": 9.293986636971047,
      "loss": 0.7952208518981934,
      "loss_ce": 0.00017690191452857107,
      "loss_iou": 0.3203125,
      "loss_num": 0.03076171875,
      "loss_xval": 0.796875,
      "num_input_tokens_seen": 233713276,
      "step": 4173
    },
    {
      "epoch": 9.296213808463252,
      "grad_norm": 17.01585578918457,
      "learning_rate": 1e-06,
      "loss": 0.3602,
      "num_input_tokens_seen": 233769732,
      "step": 4174
    },
    {
      "epoch": 9.296213808463252,
      "loss": 0.4717066287994385,
      "loss_ce": 0.0001490233844378963,
      "loss_iou": 0.20703125,
      "loss_num": 0.01141357421875,
      "loss_xval": 0.470703125,
      "num_input_tokens_seen": 233769732,
      "step": 4174
    },
    {
      "epoch": 9.298440979955457,
      "grad_norm": 18.97732925415039,
      "learning_rate": 1e-06,
      "loss": 0.4866,
      "num_input_tokens_seen": 233827660,
      "step": 4175
    },
    {
      "epoch": 9.298440979955457,
      "loss": 0.5509434938430786,
      "loss_ce": 0.0001622582640266046,
      "loss_iou": 0.25,
      "loss_num": 0.0098876953125,
      "loss_xval": 0.55078125,
      "num_input_tokens_seen": 233827660,
      "step": 4175
    },
    {
      "epoch": 9.300668151447661,
      "grad_norm": 23.303890228271484,
      "learning_rate": 1e-06,
      "loss": 0.67,
      "num_input_tokens_seen": 233884728,
      "step": 4176
    },
    {
      "epoch": 9.300668151447661,
      "loss": 0.754581093788147,
      "loss_ce": 0.00018660849309526384,
      "loss_iou": 0.298828125,
      "loss_num": 0.031494140625,
      "loss_xval": 0.75390625,
      "num_input_tokens_seen": 233884728,
      "step": 4176
    },
    {
      "epoch": 9.302895322939866,
      "grad_norm": 15.087565422058105,
      "learning_rate": 1e-06,
      "loss": 0.5322,
      "num_input_tokens_seen": 233942264,
      "step": 4177
    },
    {
      "epoch": 9.302895322939866,
      "loss": 0.692180871963501,
      "loss_ce": 0.00016421612235717475,
      "loss_iou": 0.296875,
      "loss_num": 0.019775390625,
      "loss_xval": 0.69140625,
      "num_input_tokens_seen": 233942264,
      "step": 4177
    },
    {
      "epoch": 9.305122494432071,
      "grad_norm": 21.872156143188477,
      "learning_rate": 1e-06,
      "loss": 0.4854,
      "num_input_tokens_seen": 233999788,
      "step": 4178
    },
    {
      "epoch": 9.305122494432071,
      "loss": 0.37598371505737305,
      "loss_ce": 0.00012923183385282755,
      "loss_iou": 0.166015625,
      "loss_num": 0.0089111328125,
      "loss_xval": 0.375,
      "num_input_tokens_seen": 233999788,
      "step": 4178
    },
    {
      "epoch": 9.307349665924276,
      "grad_norm": 15.20783519744873,
      "learning_rate": 1e-06,
      "loss": 0.7171,
      "num_input_tokens_seen": 234058404,
      "step": 4179
    },
    {
      "epoch": 9.307349665924276,
      "loss": 0.7457845211029053,
      "loss_ce": 0.00017908678273670375,
      "loss_iou": 0.3046875,
      "loss_num": 0.0274658203125,
      "loss_xval": 0.74609375,
      "num_input_tokens_seen": 234058404,
      "step": 4179
    },
    {
      "epoch": 9.309576837416481,
      "grad_norm": 18.751060485839844,
      "learning_rate": 1e-06,
      "loss": 0.5907,
      "num_input_tokens_seen": 234115236,
      "step": 4180
    },
    {
      "epoch": 9.309576837416481,
      "loss": 0.5246922969818115,
      "loss_ce": 0.00015614343283232301,
      "loss_iou": 0.2294921875,
      "loss_num": 0.013427734375,
      "loss_xval": 0.5234375,
      "num_input_tokens_seen": 234115236,
      "step": 4180
    },
    {
      "epoch": 9.311804008908686,
      "grad_norm": 21.91156005859375,
      "learning_rate": 1e-06,
      "loss": 0.4344,
      "num_input_tokens_seen": 234173444,
      "step": 4181
    },
    {
      "epoch": 9.311804008908686,
      "loss": 0.5160158276557922,
      "loss_ce": 0.0001467098481953144,
      "loss_iou": 0.244140625,
      "loss_num": 0.005767822265625,
      "loss_xval": 0.515625,
      "num_input_tokens_seen": 234173444,
      "step": 4181
    },
    {
      "epoch": 9.31403118040089,
      "grad_norm": 31.379457473754883,
      "learning_rate": 1e-06,
      "loss": 0.8267,
      "num_input_tokens_seen": 234228884,
      "step": 4182
    },
    {
      "epoch": 9.31403118040089,
      "loss": 0.8775807619094849,
      "loss_ce": 0.00013937248149886727,
      "loss_iou": 0.369140625,
      "loss_num": 0.027587890625,
      "loss_xval": 0.87890625,
      "num_input_tokens_seen": 234228884,
      "step": 4182
    },
    {
      "epoch": 9.316258351893095,
      "grad_norm": 18.9929256439209,
      "learning_rate": 1e-06,
      "loss": 0.5908,
      "num_input_tokens_seen": 234286752,
      "step": 4183
    },
    {
      "epoch": 9.316258351893095,
      "loss": 0.8116600513458252,
      "loss_ce": 0.000136615228257142,
      "loss_iou": 0.337890625,
      "loss_num": 0.02685546875,
      "loss_xval": 0.8125,
      "num_input_tokens_seen": 234286752,
      "step": 4183
    },
    {
      "epoch": 9.3184855233853,
      "grad_norm": 16.527360916137695,
      "learning_rate": 1e-06,
      "loss": 0.6648,
      "num_input_tokens_seen": 234342332,
      "step": 4184
    },
    {
      "epoch": 9.3184855233853,
      "loss": 0.7774823904037476,
      "loss_ce": 0.00013861866318620741,
      "loss_iou": 0.34765625,
      "loss_num": 0.0167236328125,
      "loss_xval": 0.77734375,
      "num_input_tokens_seen": 234342332,
      "step": 4184
    },
    {
      "epoch": 9.320712694877505,
      "grad_norm": 16.93600845336914,
      "learning_rate": 1e-06,
      "loss": 0.5782,
      "num_input_tokens_seen": 234398820,
      "step": 4185
    },
    {
      "epoch": 9.320712694877505,
      "loss": 0.7552863359451294,
      "loss_ce": 0.0001593705965206027,
      "loss_iou": 0.322265625,
      "loss_num": 0.022216796875,
      "loss_xval": 0.75390625,
      "num_input_tokens_seen": 234398820,
      "step": 4185
    },
    {
      "epoch": 9.32293986636971,
      "grad_norm": 22.869224548339844,
      "learning_rate": 1e-06,
      "loss": 0.7147,
      "num_input_tokens_seen": 234455760,
      "step": 4186
    },
    {
      "epoch": 9.32293986636971,
      "loss": 0.5804110169410706,
      "loss_ce": 0.00024132244288921356,
      "loss_iou": 0.24609375,
      "loss_num": 0.0177001953125,
      "loss_xval": 0.58203125,
      "num_input_tokens_seen": 234455760,
      "step": 4186
    },
    {
      "epoch": 9.325167037861915,
      "grad_norm": 15.121347427368164,
      "learning_rate": 1e-06,
      "loss": 0.4896,
      "num_input_tokens_seen": 234510508,
      "step": 4187
    },
    {
      "epoch": 9.325167037861915,
      "loss": 0.3788001537322998,
      "loss_ce": 0.00013804002082906663,
      "loss_iou": 0.166015625,
      "loss_num": 0.0093994140625,
      "loss_xval": 0.37890625,
      "num_input_tokens_seen": 234510508,
      "step": 4187
    },
    {
      "epoch": 9.32739420935412,
      "grad_norm": 18.335205078125,
      "learning_rate": 1e-06,
      "loss": 0.7204,
      "num_input_tokens_seen": 234567480,
      "step": 4188
    },
    {
      "epoch": 9.32739420935412,
      "loss": 0.6255242824554443,
      "loss_ce": 0.0001580739044584334,
      "loss_iou": 0.251953125,
      "loss_num": 0.024658203125,
      "loss_xval": 0.625,
      "num_input_tokens_seen": 234567480,
      "step": 4188
    },
    {
      "epoch": 9.329621380846325,
      "grad_norm": 12.408183097839355,
      "learning_rate": 1e-06,
      "loss": 0.58,
      "num_input_tokens_seen": 234625836,
      "step": 4189
    },
    {
      "epoch": 9.329621380846325,
      "loss": 0.36714544892311096,
      "loss_ce": 0.0002021001128014177,
      "loss_iou": 0.146484375,
      "loss_num": 0.0147705078125,
      "loss_xval": 0.3671875,
      "num_input_tokens_seen": 234625836,
      "step": 4189
    },
    {
      "epoch": 9.33184855233853,
      "grad_norm": 20.51274299621582,
      "learning_rate": 1e-06,
      "loss": 0.5372,
      "num_input_tokens_seen": 234683824,
      "step": 4190
    },
    {
      "epoch": 9.33184855233853,
      "loss": 0.4362949728965759,
      "loss_ce": 0.00013776315608993173,
      "loss_iou": 0.1865234375,
      "loss_num": 0.01263427734375,
      "loss_xval": 0.435546875,
      "num_input_tokens_seen": 234683824,
      "step": 4190
    },
    {
      "epoch": 9.334075723830734,
      "grad_norm": 67.99667358398438,
      "learning_rate": 1e-06,
      "loss": 0.5763,
      "num_input_tokens_seen": 234739016,
      "step": 4191
    },
    {
      "epoch": 9.334075723830734,
      "loss": 0.4585108757019043,
      "loss_ce": 0.00013684081204701215,
      "loss_iou": 0.1923828125,
      "loss_num": 0.0146484375,
      "loss_xval": 0.458984375,
      "num_input_tokens_seen": 234739016,
      "step": 4191
    },
    {
      "epoch": 9.33630289532294,
      "grad_norm": 17.7122859954834,
      "learning_rate": 1e-06,
      "loss": 0.6762,
      "num_input_tokens_seen": 234793360,
      "step": 4192
    },
    {
      "epoch": 9.33630289532294,
      "loss": 0.6960656642913818,
      "loss_ce": 0.00014278030721470714,
      "loss_iou": 0.32421875,
      "loss_num": 0.0093994140625,
      "loss_xval": 0.6953125,
      "num_input_tokens_seen": 234793360,
      "step": 4192
    },
    {
      "epoch": 9.338530066815144,
      "grad_norm": 14.239766120910645,
      "learning_rate": 1e-06,
      "loss": 0.4748,
      "num_input_tokens_seen": 234849088,
      "step": 4193
    },
    {
      "epoch": 9.338530066815144,
      "loss": 0.40485137701034546,
      "loss_ce": 0.0001883181685116142,
      "loss_iou": 0.18359375,
      "loss_num": 0.007598876953125,
      "loss_xval": 0.404296875,
      "num_input_tokens_seen": 234849088,
      "step": 4193
    },
    {
      "epoch": 9.340757238307349,
      "grad_norm": 22.02215576171875,
      "learning_rate": 1e-06,
      "loss": 0.4715,
      "num_input_tokens_seen": 234905732,
      "step": 4194
    },
    {
      "epoch": 9.340757238307349,
      "loss": 0.4357007145881653,
      "loss_ce": 0.0001538339420221746,
      "loss_iou": 0.193359375,
      "loss_num": 0.00982666015625,
      "loss_xval": 0.435546875,
      "num_input_tokens_seen": 234905732,
      "step": 4194
    },
    {
      "epoch": 9.342984409799554,
      "grad_norm": 24.65700340270996,
      "learning_rate": 1e-06,
      "loss": 0.6571,
      "num_input_tokens_seen": 234959588,
      "step": 4195
    },
    {
      "epoch": 9.342984409799554,
      "loss": 0.6442734003067017,
      "loss_ce": 0.00023038909421302378,
      "loss_iou": 0.2353515625,
      "loss_num": 0.03466796875,
      "loss_xval": 0.64453125,
      "num_input_tokens_seen": 234959588,
      "step": 4195
    },
    {
      "epoch": 9.345211581291759,
      "grad_norm": 18.189048767089844,
      "learning_rate": 1e-06,
      "loss": 0.5611,
      "num_input_tokens_seen": 235018980,
      "step": 4196
    },
    {
      "epoch": 9.345211581291759,
      "loss": 0.6549949645996094,
      "loss_ce": 0.00020979381224606186,
      "loss_iou": 0.29296875,
      "loss_num": 0.0140380859375,
      "loss_xval": 0.65625,
      "num_input_tokens_seen": 235018980,
      "step": 4196
    },
    {
      "epoch": 9.347438752783964,
      "grad_norm": 43.657958984375,
      "learning_rate": 1e-06,
      "loss": 0.7156,
      "num_input_tokens_seen": 235076428,
      "step": 4197
    },
    {
      "epoch": 9.347438752783964,
      "loss": 0.6480019092559814,
      "loss_ce": 0.0002968419576063752,
      "loss_iou": 0.28125,
      "loss_num": 0.0172119140625,
      "loss_xval": 0.6484375,
      "num_input_tokens_seen": 235076428,
      "step": 4197
    },
    {
      "epoch": 9.34966592427617,
      "grad_norm": 18.815326690673828,
      "learning_rate": 1e-06,
      "loss": 0.526,
      "num_input_tokens_seen": 235134848,
      "step": 4198
    },
    {
      "epoch": 9.34966592427617,
      "loss": 0.5375351905822754,
      "loss_ce": 0.00018168592941947281,
      "loss_iou": 0.2373046875,
      "loss_num": 0.01263427734375,
      "loss_xval": 0.5390625,
      "num_input_tokens_seen": 235134848,
      "step": 4198
    },
    {
      "epoch": 9.351893095768375,
      "grad_norm": 16.838205337524414,
      "learning_rate": 1e-06,
      "loss": 0.6095,
      "num_input_tokens_seen": 235189616,
      "step": 4199
    },
    {
      "epoch": 9.351893095768375,
      "loss": 0.5999011993408203,
      "loss_ce": 0.00016978610074147582,
      "loss_iou": 0.26171875,
      "loss_num": 0.0152587890625,
      "loss_xval": 0.6015625,
      "num_input_tokens_seen": 235189616,
      "step": 4199
    },
    {
      "epoch": 9.35412026726058,
      "grad_norm": 22.742488861083984,
      "learning_rate": 1e-06,
      "loss": 0.6383,
      "num_input_tokens_seen": 235243244,
      "step": 4200
    },
    {
      "epoch": 9.35412026726058,
      "loss": 0.49684494733810425,
      "loss_ce": 0.00014081134577281773,
      "loss_iou": 0.205078125,
      "loss_num": 0.0174560546875,
      "loss_xval": 0.49609375,
      "num_input_tokens_seen": 235243244,
      "step": 4200
    },
    {
      "epoch": 9.356347438752785,
      "grad_norm": 18.40520668029785,
      "learning_rate": 1e-06,
      "loss": 0.6173,
      "num_input_tokens_seen": 235300852,
      "step": 4201
    },
    {
      "epoch": 9.356347438752785,
      "loss": 0.7159720659255981,
      "loss_ce": 0.00015178298053797334,
      "loss_iou": 0.291015625,
      "loss_num": 0.02685546875,
      "loss_xval": 0.71484375,
      "num_input_tokens_seen": 235300852,
      "step": 4201
    },
    {
      "epoch": 9.35857461024499,
      "grad_norm": 18.26913833618164,
      "learning_rate": 1e-06,
      "loss": 0.6494,
      "num_input_tokens_seen": 235357404,
      "step": 4202
    },
    {
      "epoch": 9.35857461024499,
      "loss": 0.7315636873245239,
      "loss_ce": 0.00024043236044235528,
      "loss_iou": 0.328125,
      "loss_num": 0.0145263671875,
      "loss_xval": 0.73046875,
      "num_input_tokens_seen": 235357404,
      "step": 4202
    },
    {
      "epoch": 9.360801781737194,
      "grad_norm": 20.889400482177734,
      "learning_rate": 1e-06,
      "loss": 0.6694,
      "num_input_tokens_seen": 235412144,
      "step": 4203
    },
    {
      "epoch": 9.360801781737194,
      "loss": 0.6847151517868042,
      "loss_ce": 0.00014484771236311644,
      "loss_iou": 0.2890625,
      "loss_num": 0.0208740234375,
      "loss_xval": 0.68359375,
      "num_input_tokens_seen": 235412144,
      "step": 4203
    },
    {
      "epoch": 9.3630289532294,
      "grad_norm": 14.52232551574707,
      "learning_rate": 1e-06,
      "loss": 0.522,
      "num_input_tokens_seen": 235469972,
      "step": 4204
    },
    {
      "epoch": 9.3630289532294,
      "loss": 0.5159176588058472,
      "loss_ce": 0.00017058770754374564,
      "loss_iou": 0.2197265625,
      "loss_num": 0.01531982421875,
      "loss_xval": 0.515625,
      "num_input_tokens_seen": 235469972,
      "step": 4204
    },
    {
      "epoch": 9.365256124721604,
      "grad_norm": 18.92323875427246,
      "learning_rate": 1e-06,
      "loss": 0.5651,
      "num_input_tokens_seen": 235526988,
      "step": 4205
    },
    {
      "epoch": 9.365256124721604,
      "loss": 0.7476789951324463,
      "loss_ce": 0.00036454031942412257,
      "loss_iou": 0.330078125,
      "loss_num": 0.0172119140625,
      "loss_xval": 0.74609375,
      "num_input_tokens_seen": 235526988,
      "step": 4205
    },
    {
      "epoch": 9.367483296213809,
      "grad_norm": 25.92852210998535,
      "learning_rate": 1e-06,
      "loss": 0.4669,
      "num_input_tokens_seen": 235584824,
      "step": 4206
    },
    {
      "epoch": 9.367483296213809,
      "loss": 0.36491858959198,
      "loss_ce": 0.00017249592929147184,
      "loss_iou": 0.166015625,
      "loss_num": 0.006622314453125,
      "loss_xval": 0.365234375,
      "num_input_tokens_seen": 235584824,
      "step": 4206
    },
    {
      "epoch": 9.369710467706014,
      "grad_norm": 23.176660537719727,
      "learning_rate": 1e-06,
      "loss": 0.5446,
      "num_input_tokens_seen": 235642972,
      "step": 4207
    },
    {
      "epoch": 9.369710467706014,
      "loss": 0.425027072429657,
      "loss_ce": 0.00022239354439079762,
      "loss_iou": 0.171875,
      "loss_num": 0.0159912109375,
      "loss_xval": 0.42578125,
      "num_input_tokens_seen": 235642972,
      "step": 4207
    },
    {
      "epoch": 9.371937639198219,
      "grad_norm": 47.62214660644531,
      "learning_rate": 1e-06,
      "loss": 0.7125,
      "num_input_tokens_seen": 235698332,
      "step": 4208
    },
    {
      "epoch": 9.371937639198219,
      "loss": 0.8402742743492126,
      "loss_ce": 0.00018637791799847037,
      "loss_iou": 0.33203125,
      "loss_num": 0.035400390625,
      "loss_xval": 0.83984375,
      "num_input_tokens_seen": 235698332,
      "step": 4208
    },
    {
      "epoch": 9.374164810690424,
      "grad_norm": 15.558445930480957,
      "learning_rate": 1e-06,
      "loss": 0.4657,
      "num_input_tokens_seen": 235754988,
      "step": 4209
    },
    {
      "epoch": 9.374164810690424,
      "loss": 0.354988694190979,
      "loss_ce": 0.000130318061565049,
      "loss_iou": 0.1494140625,
      "loss_num": 0.0111083984375,
      "loss_xval": 0.35546875,
      "num_input_tokens_seen": 235754988,
      "step": 4209
    },
    {
      "epoch": 9.376391982182628,
      "grad_norm": 21.12288475036621,
      "learning_rate": 1e-06,
      "loss": 0.6511,
      "num_input_tokens_seen": 235812532,
      "step": 4210
    },
    {
      "epoch": 9.376391982182628,
      "loss": 0.7051099538803101,
      "loss_ce": 0.0001538842625450343,
      "loss_iou": 0.302734375,
      "loss_num": 0.01953125,
      "loss_xval": 0.703125,
      "num_input_tokens_seen": 235812532,
      "step": 4210
    },
    {
      "epoch": 9.378619153674833,
      "grad_norm": 16.031126022338867,
      "learning_rate": 1e-06,
      "loss": 0.4852,
      "num_input_tokens_seen": 235867920,
      "step": 4211
    },
    {
      "epoch": 9.378619153674833,
      "loss": 0.5213667750358582,
      "loss_ce": 0.0004927542759105563,
      "loss_iou": 0.2333984375,
      "loss_num": 0.0107421875,
      "loss_xval": 0.51953125,
      "num_input_tokens_seen": 235867920,
      "step": 4211
    },
    {
      "epoch": 9.380846325167038,
      "grad_norm": 16.82216453552246,
      "learning_rate": 1e-06,
      "loss": 0.468,
      "num_input_tokens_seen": 235923424,
      "step": 4212
    },
    {
      "epoch": 9.380846325167038,
      "loss": 0.5014048218727112,
      "loss_ce": 0.00018410818302072585,
      "loss_iou": 0.2109375,
      "loss_num": 0.015869140625,
      "loss_xval": 0.5,
      "num_input_tokens_seen": 235923424,
      "step": 4212
    },
    {
      "epoch": 9.383073496659243,
      "grad_norm": 15.369388580322266,
      "learning_rate": 1e-06,
      "loss": 0.4188,
      "num_input_tokens_seen": 235977656,
      "step": 4213
    },
    {
      "epoch": 9.383073496659243,
      "loss": 0.466952919960022,
      "loss_ce": 0.0001560571399750188,
      "loss_iou": 0.1943359375,
      "loss_num": 0.015625,
      "loss_xval": 0.466796875,
      "num_input_tokens_seen": 235977656,
      "step": 4213
    },
    {
      "epoch": 9.385300668151448,
      "grad_norm": 12.31820011138916,
      "learning_rate": 1e-06,
      "loss": 0.6936,
      "num_input_tokens_seen": 236034136,
      "step": 4214
    },
    {
      "epoch": 9.385300668151448,
      "loss": 0.6039036512374878,
      "loss_ce": 0.00014389277203008533,
      "loss_iou": 0.265625,
      "loss_num": 0.01409912109375,
      "loss_xval": 0.60546875,
      "num_input_tokens_seen": 236034136,
      "step": 4214
    },
    {
      "epoch": 9.387527839643653,
      "grad_norm": 20.034059524536133,
      "learning_rate": 1e-06,
      "loss": 0.5376,
      "num_input_tokens_seen": 236092456,
      "step": 4215
    },
    {
      "epoch": 9.387527839643653,
      "loss": 0.5938976407051086,
      "loss_ce": 0.0001476521574659273,
      "loss_iou": 0.267578125,
      "loss_num": 0.01141357421875,
      "loss_xval": 0.59375,
      "num_input_tokens_seen": 236092456,
      "step": 4215
    },
    {
      "epoch": 9.389755011135858,
      "grad_norm": 46.42024612426758,
      "learning_rate": 1e-06,
      "loss": 0.5741,
      "num_input_tokens_seen": 236147020,
      "step": 4216
    },
    {
      "epoch": 9.389755011135858,
      "loss": 0.5294545292854309,
      "loss_ce": 0.00015764265845064074,
      "loss_iou": 0.244140625,
      "loss_num": 0.0079345703125,
      "loss_xval": 0.53125,
      "num_input_tokens_seen": 236147020,
      "step": 4216
    },
    {
      "epoch": 9.391982182628063,
      "grad_norm": 26.644548416137695,
      "learning_rate": 1e-06,
      "loss": 0.6237,
      "num_input_tokens_seen": 236200772,
      "step": 4217
    },
    {
      "epoch": 9.391982182628063,
      "loss": 0.7529926300048828,
      "loss_ce": 0.00018498601275496185,
      "loss_iou": 0.33984375,
      "loss_num": 0.01483154296875,
      "loss_xval": 0.75390625,
      "num_input_tokens_seen": 236200772,
      "step": 4217
    },
    {
      "epoch": 9.394209354120267,
      "grad_norm": 18.652217864990234,
      "learning_rate": 1e-06,
      "loss": 0.5676,
      "num_input_tokens_seen": 236255604,
      "step": 4218
    },
    {
      "epoch": 9.394209354120267,
      "loss": 0.6091314554214478,
      "loss_ce": 0.00024474196834489703,
      "loss_iou": 0.28125,
      "loss_num": 0.0091552734375,
      "loss_xval": 0.609375,
      "num_input_tokens_seen": 236255604,
      "step": 4218
    },
    {
      "epoch": 9.396436525612472,
      "grad_norm": 14.043095588684082,
      "learning_rate": 1e-06,
      "loss": 0.7096,
      "num_input_tokens_seen": 236310220,
      "step": 4219
    },
    {
      "epoch": 9.396436525612472,
      "loss": 0.8942954540252686,
      "loss_ce": 0.0002524922601878643,
      "loss_iou": 0.390625,
      "loss_num": 0.022705078125,
      "loss_xval": 0.89453125,
      "num_input_tokens_seen": 236310220,
      "step": 4219
    },
    {
      "epoch": 9.398663697104677,
      "grad_norm": 17.981460571289062,
      "learning_rate": 1e-06,
      "loss": 0.5667,
      "num_input_tokens_seen": 236364604,
      "step": 4220
    },
    {
      "epoch": 9.398663697104677,
      "loss": 0.4448683559894562,
      "loss_ce": 0.0001662159920670092,
      "loss_iou": 0.2001953125,
      "loss_num": 0.00872802734375,
      "loss_xval": 0.4453125,
      "num_input_tokens_seen": 236364604,
      "step": 4220
    },
    {
      "epoch": 9.400890868596882,
      "grad_norm": 14.154129028320312,
      "learning_rate": 1e-06,
      "loss": 0.6947,
      "num_input_tokens_seen": 236421912,
      "step": 4221
    },
    {
      "epoch": 9.400890868596882,
      "loss": 0.7411311268806458,
      "loss_ce": 0.00016432552365586162,
      "loss_iou": 0.34765625,
      "loss_num": 0.00933837890625,
      "loss_xval": 0.7421875,
      "num_input_tokens_seen": 236421912,
      "step": 4221
    },
    {
      "epoch": 9.403118040089087,
      "grad_norm": 34.250003814697266,
      "learning_rate": 1e-06,
      "loss": 0.5965,
      "num_input_tokens_seen": 236478284,
      "step": 4222
    },
    {
      "epoch": 9.403118040089087,
      "loss": 0.5931603312492371,
      "loss_ce": 0.00014277252194005996,
      "loss_iou": 0.25,
      "loss_num": 0.0185546875,
      "loss_xval": 0.59375,
      "num_input_tokens_seen": 236478284,
      "step": 4222
    },
    {
      "epoch": 9.405345211581292,
      "grad_norm": 21.780054092407227,
      "learning_rate": 1e-06,
      "loss": 0.5467,
      "num_input_tokens_seen": 236534108,
      "step": 4223
    },
    {
      "epoch": 9.405345211581292,
      "loss": 0.5391287803649902,
      "loss_ce": 0.0001272944500669837,
      "loss_iou": 0.2158203125,
      "loss_num": 0.0213623046875,
      "loss_xval": 0.5390625,
      "num_input_tokens_seen": 236534108,
      "step": 4223
    },
    {
      "epoch": 9.407572383073497,
      "grad_norm": 22.980321884155273,
      "learning_rate": 1e-06,
      "loss": 0.7141,
      "num_input_tokens_seen": 236590772,
      "step": 4224
    },
    {
      "epoch": 9.407572383073497,
      "loss": 0.6423543691635132,
      "loss_ce": 0.00014242672477848828,
      "loss_iou": 0.27734375,
      "loss_num": 0.01708984375,
      "loss_xval": 0.640625,
      "num_input_tokens_seen": 236590772,
      "step": 4224
    },
    {
      "epoch": 9.409799554565701,
      "grad_norm": 21.113697052001953,
      "learning_rate": 1e-06,
      "loss": 0.4206,
      "num_input_tokens_seen": 236648468,
      "step": 4225
    },
    {
      "epoch": 9.409799554565701,
      "loss": 0.31469690799713135,
      "loss_ce": 0.0001217244571307674,
      "loss_iou": 0.12060546875,
      "loss_num": 0.0146484375,
      "loss_xval": 0.314453125,
      "num_input_tokens_seen": 236648468,
      "step": 4225
    },
    {
      "epoch": 9.412026726057906,
      "grad_norm": 9.035064697265625,
      "learning_rate": 1e-06,
      "loss": 0.3957,
      "num_input_tokens_seen": 236707008,
      "step": 4226
    },
    {
      "epoch": 9.412026726057906,
      "loss": 0.30737414956092834,
      "loss_ce": 0.00012316979700699449,
      "loss_iou": 0.125,
      "loss_num": 0.011474609375,
      "loss_xval": 0.306640625,
      "num_input_tokens_seen": 236707008,
      "step": 4226
    },
    {
      "epoch": 9.414253897550111,
      "grad_norm": 16.591636657714844,
      "learning_rate": 1e-06,
      "loss": 0.6342,
      "num_input_tokens_seen": 236761664,
      "step": 4227
    },
    {
      "epoch": 9.414253897550111,
      "loss": 0.7024285793304443,
      "loss_ce": 0.00015806331066414714,
      "loss_iou": 0.29296875,
      "loss_num": 0.0234375,
      "loss_xval": 0.703125,
      "num_input_tokens_seen": 236761664,
      "step": 4227
    },
    {
      "epoch": 9.416481069042316,
      "grad_norm": 21.84915542602539,
      "learning_rate": 1e-06,
      "loss": 0.6148,
      "num_input_tokens_seen": 236819440,
      "step": 4228
    },
    {
      "epoch": 9.416481069042316,
      "loss": 0.5935507416725159,
      "loss_ce": 0.0002889999595936388,
      "loss_iou": 0.2412109375,
      "loss_num": 0.02197265625,
      "loss_xval": 0.59375,
      "num_input_tokens_seen": 236819440,
      "step": 4228
    },
    {
      "epoch": 9.41870824053452,
      "grad_norm": 17.725332260131836,
      "learning_rate": 1e-06,
      "loss": 0.5218,
      "num_input_tokens_seen": 236876512,
      "step": 4229
    },
    {
      "epoch": 9.41870824053452,
      "loss": 0.546193540096283,
      "loss_ce": 0.00017301872139796615,
      "loss_iou": 0.24609375,
      "loss_num": 0.0106201171875,
      "loss_xval": 0.546875,
      "num_input_tokens_seen": 236876512,
      "step": 4229
    },
    {
      "epoch": 9.420935412026726,
      "grad_norm": 14.556755065917969,
      "learning_rate": 1e-06,
      "loss": 0.4876,
      "num_input_tokens_seen": 236933876,
      "step": 4230
    },
    {
      "epoch": 9.420935412026726,
      "loss": 0.5147203207015991,
      "loss_ce": 0.0008042675326578319,
      "loss_iou": 0.220703125,
      "loss_num": 0.0147705078125,
      "loss_xval": 0.515625,
      "num_input_tokens_seen": 236933876,
      "step": 4230
    },
    {
      "epoch": 9.42316258351893,
      "grad_norm": 30.021310806274414,
      "learning_rate": 1e-06,
      "loss": 0.5368,
      "num_input_tokens_seen": 236989724,
      "step": 4231
    },
    {
      "epoch": 9.42316258351893,
      "loss": 0.6030337810516357,
      "loss_ce": 0.000128521875012666,
      "loss_iou": 0.26171875,
      "loss_num": 0.01544189453125,
      "loss_xval": 0.6015625,
      "num_input_tokens_seen": 236989724,
      "step": 4231
    },
    {
      "epoch": 9.425389755011135,
      "grad_norm": 17.476322174072266,
      "learning_rate": 1e-06,
      "loss": 0.6046,
      "num_input_tokens_seen": 237044832,
      "step": 4232
    },
    {
      "epoch": 9.425389755011135,
      "loss": 0.6286413669586182,
      "loss_ce": 0.00022339157294481993,
      "loss_iou": 0.291015625,
      "loss_num": 0.0089111328125,
      "loss_xval": 0.62890625,
      "num_input_tokens_seen": 237044832,
      "step": 4232
    },
    {
      "epoch": 9.42761692650334,
      "grad_norm": 18.216716766357422,
      "learning_rate": 1e-06,
      "loss": 0.6325,
      "num_input_tokens_seen": 237102544,
      "step": 4233
    },
    {
      "epoch": 9.42761692650334,
      "loss": 0.7180312871932983,
      "loss_ce": 0.0002578936982899904,
      "loss_iou": 0.29296875,
      "loss_num": 0.026611328125,
      "loss_xval": 0.71875,
      "num_input_tokens_seen": 237102544,
      "step": 4233
    },
    {
      "epoch": 9.429844097995545,
      "grad_norm": 20.78913688659668,
      "learning_rate": 1e-06,
      "loss": 0.5074,
      "num_input_tokens_seen": 237158060,
      "step": 4234
    },
    {
      "epoch": 9.429844097995545,
      "loss": 0.5021035671234131,
      "loss_ce": 0.00015046056068968028,
      "loss_iou": 0.2275390625,
      "loss_num": 0.00933837890625,
      "loss_xval": 0.5,
      "num_input_tokens_seen": 237158060,
      "step": 4234
    },
    {
      "epoch": 9.43207126948775,
      "grad_norm": 138.009033203125,
      "learning_rate": 1e-06,
      "loss": 0.5155,
      "num_input_tokens_seen": 237213952,
      "step": 4235
    },
    {
      "epoch": 9.43207126948775,
      "loss": 0.5691501498222351,
      "loss_ce": 0.0005466500297188759,
      "loss_iou": 0.240234375,
      "loss_num": 0.017578125,
      "loss_xval": 0.5703125,
      "num_input_tokens_seen": 237213952,
      "step": 4235
    },
    {
      "epoch": 9.434298440979955,
      "grad_norm": 18.69287109375,
      "learning_rate": 1e-06,
      "loss": 0.5015,
      "num_input_tokens_seen": 237269936,
      "step": 4236
    },
    {
      "epoch": 9.434298440979955,
      "loss": 0.6309968829154968,
      "loss_ce": 0.00013751011283602566,
      "loss_iou": 0.251953125,
      "loss_num": 0.0255126953125,
      "loss_xval": 0.6328125,
      "num_input_tokens_seen": 237269936,
      "step": 4236
    },
    {
      "epoch": 9.43652561247216,
      "grad_norm": 11.285958290100098,
      "learning_rate": 1e-06,
      "loss": 0.5439,
      "num_input_tokens_seen": 237325496,
      "step": 4237
    },
    {
      "epoch": 9.43652561247216,
      "loss": 0.6378533840179443,
      "loss_ce": 0.0001580730895511806,
      "loss_iou": 0.27734375,
      "loss_num": 0.01708984375,
      "loss_xval": 0.63671875,
      "num_input_tokens_seen": 237325496,
      "step": 4237
    },
    {
      "epoch": 9.438752783964365,
      "grad_norm": 24.240333557128906,
      "learning_rate": 1e-06,
      "loss": 0.5876,
      "num_input_tokens_seen": 237383104,
      "step": 4238
    },
    {
      "epoch": 9.438752783964365,
      "loss": 0.6686908006668091,
      "loss_ce": 0.0002032603952102363,
      "loss_iou": 0.2734375,
      "loss_num": 0.02392578125,
      "loss_xval": 0.66796875,
      "num_input_tokens_seen": 237383104,
      "step": 4238
    },
    {
      "epoch": 9.44097995545657,
      "grad_norm": 16.92664337158203,
      "learning_rate": 1e-06,
      "loss": 0.4389,
      "num_input_tokens_seen": 237434992,
      "step": 4239
    },
    {
      "epoch": 9.44097995545657,
      "loss": 0.38501814007759094,
      "loss_ce": 0.00013044432853348553,
      "loss_iou": 0.1611328125,
      "loss_num": 0.012451171875,
      "loss_xval": 0.384765625,
      "num_input_tokens_seen": 237434992,
      "step": 4239
    },
    {
      "epoch": 9.443207126948774,
      "grad_norm": 20.529521942138672,
      "learning_rate": 1e-06,
      "loss": 0.6429,
      "num_input_tokens_seen": 237489668,
      "step": 4240
    },
    {
      "epoch": 9.443207126948774,
      "loss": 0.703424870967865,
      "loss_ce": 0.00017782000941224396,
      "loss_iou": 0.283203125,
      "loss_num": 0.0274658203125,
      "loss_xval": 0.703125,
      "num_input_tokens_seen": 237489668,
      "step": 4240
    },
    {
      "epoch": 9.44543429844098,
      "grad_norm": 18.514591217041016,
      "learning_rate": 1e-06,
      "loss": 0.5228,
      "num_input_tokens_seen": 237545776,
      "step": 4241
    },
    {
      "epoch": 9.44543429844098,
      "loss": 0.508231520652771,
      "loss_ce": 0.00041901745134964585,
      "loss_iou": 0.236328125,
      "loss_num": 0.007171630859375,
      "loss_xval": 0.5078125,
      "num_input_tokens_seen": 237545776,
      "step": 4241
    },
    {
      "epoch": 9.447661469933184,
      "grad_norm": 22.630306243896484,
      "learning_rate": 1e-06,
      "loss": 0.5125,
      "num_input_tokens_seen": 237603000,
      "step": 4242
    },
    {
      "epoch": 9.447661469933184,
      "loss": 0.5746460556983948,
      "loss_ce": 0.00018314782937522978,
      "loss_iou": 0.22265625,
      "loss_num": 0.02587890625,
      "loss_xval": 0.57421875,
      "num_input_tokens_seen": 237603000,
      "step": 4242
    },
    {
      "epoch": 9.449888641425389,
      "grad_norm": 21.890134811401367,
      "learning_rate": 1e-06,
      "loss": 0.5389,
      "num_input_tokens_seen": 237658508,
      "step": 4243
    },
    {
      "epoch": 9.449888641425389,
      "loss": 0.5602293610572815,
      "loss_ce": 0.00017077414668165147,
      "loss_iou": 0.248046875,
      "loss_num": 0.01263427734375,
      "loss_xval": 0.55859375,
      "num_input_tokens_seen": 237658508,
      "step": 4243
    },
    {
      "epoch": 9.452115812917596,
      "grad_norm": 17.784313201904297,
      "learning_rate": 1e-06,
      "loss": 0.5817,
      "num_input_tokens_seen": 237714064,
      "step": 4244
    },
    {
      "epoch": 9.452115812917596,
      "loss": 0.5421576499938965,
      "loss_ce": 0.0001654581428738311,
      "loss_iou": 0.2470703125,
      "loss_num": 0.00933837890625,
      "loss_xval": 0.54296875,
      "num_input_tokens_seen": 237714064,
      "step": 4244
    },
    {
      "epoch": 9.4543429844098,
      "grad_norm": 17.17974853515625,
      "learning_rate": 1e-06,
      "loss": 0.4908,
      "num_input_tokens_seen": 237768912,
      "step": 4245
    },
    {
      "epoch": 9.4543429844098,
      "loss": 0.34260523319244385,
      "loss_ce": 0.00013697342365048826,
      "loss_iou": 0.13671875,
      "loss_num": 0.01361083984375,
      "loss_xval": 0.341796875,
      "num_input_tokens_seen": 237768912,
      "step": 4245
    },
    {
      "epoch": 9.456570155902005,
      "grad_norm": 22.440263748168945,
      "learning_rate": 1e-06,
      "loss": 0.8688,
      "num_input_tokens_seen": 237822376,
      "step": 4246
    },
    {
      "epoch": 9.456570155902005,
      "loss": 0.7955905795097351,
      "loss_ce": 0.00018044964235741645,
      "loss_iou": 0.322265625,
      "loss_num": 0.030029296875,
      "loss_xval": 0.796875,
      "num_input_tokens_seen": 237822376,
      "step": 4246
    },
    {
      "epoch": 9.45879732739421,
      "grad_norm": 20.348644256591797,
      "learning_rate": 1e-06,
      "loss": 0.5448,
      "num_input_tokens_seen": 237879852,
      "step": 4247
    },
    {
      "epoch": 9.45879732739421,
      "loss": 0.7884389758110046,
      "loss_ce": 0.00023097131634131074,
      "loss_iou": 0.314453125,
      "loss_num": 0.031982421875,
      "loss_xval": 0.7890625,
      "num_input_tokens_seen": 237879852,
      "step": 4247
    },
    {
      "epoch": 9.461024498886415,
      "grad_norm": 98.13805389404297,
      "learning_rate": 1e-06,
      "loss": 0.7861,
      "num_input_tokens_seen": 237933556,
      "step": 4248
    },
    {
      "epoch": 9.461024498886415,
      "loss": 1.0614817142486572,
      "loss_ce": 0.0002023179258685559,
      "loss_iou": 0.453125,
      "loss_num": 0.03076171875,
      "loss_xval": 1.0625,
      "num_input_tokens_seen": 237933556,
      "step": 4248
    },
    {
      "epoch": 9.46325167037862,
      "grad_norm": 18.363706588745117,
      "learning_rate": 1e-06,
      "loss": 0.5792,
      "num_input_tokens_seen": 237990416,
      "step": 4249
    },
    {
      "epoch": 9.46325167037862,
      "loss": 0.6430014371871948,
      "loss_ce": 0.00017919728998094797,
      "loss_iou": 0.2734375,
      "loss_num": 0.01904296875,
      "loss_xval": 0.64453125,
      "num_input_tokens_seen": 237990416,
      "step": 4249
    },
    {
      "epoch": 9.465478841870825,
      "grad_norm": 22.31871223449707,
      "learning_rate": 1e-06,
      "loss": 0.5357,
      "num_input_tokens_seen": 238048204,
      "step": 4250
    },
    {
      "epoch": 9.465478841870825,
      "eval_seeclick_web_CIoU": 0.5783384740352631,
      "eval_seeclick_web_GIoU": 0.5768938064575195,
      "eval_seeclick_web_IoU": 0.596380203962326,
      "eval_seeclick_web_MAE_all": 0.016195162199437618,
      "eval_seeclick_web_MAE_h": 0.007964757736772299,
      "eval_seeclick_web_MAE_w": 0.01662321202456951,
      "eval_seeclick_web_MAE_x_boxes": 0.00983009533956647,
      "eval_seeclick_web_MAE_y_boxes": 0.021932302275672555,
      "eval_seeclick_web_inside_bbox": 0.9166666567325592,
      "eval_seeclick_web_loss": 0.9355359077453613,
      "eval_seeclick_web_loss_ce": 0.00021685881074517965,
      "eval_seeclick_web_loss_iou": 0.424072265625,
      "eval_seeclick_web_loss_num": 0.013062477111816406,
      "eval_seeclick_web_loss_xval": 0.9130859375,
      "eval_seeclick_web_runtime": 23.3851,
      "eval_seeclick_web_samples_per_second": 2.138,
      "eval_seeclick_web_steps_per_second": 0.086,
      "num_input_tokens_seen": 238048204,
      "step": 4250
    },
    {
      "epoch": 9.465478841870825,
      "eval_icons_CIoU": 0.2803712487220764,
      "eval_icons_GIoU": 0.30777665972709656,
      "eval_icons_IoU": 0.3619535565376282,
      "eval_icons_MAE_all": 0.06462299637496471,
      "eval_icons_MAE_h": 0.039737068116664886,
      "eval_icons_MAE_w": 0.06902601942420006,
      "eval_icons_MAE_x_boxes": 0.05829739384353161,
      "eval_icons_MAE_y_boxes": 0.039782424457371235,
      "eval_icons_inside_bbox": 0.6336805522441864,
      "eval_icons_loss": 1.738389253616333,
      "eval_icons_loss_ce": 0.00025170089793391526,
      "eval_icons_loss_iou": 0.6759033203125,
      "eval_icons_loss_num": 0.061977386474609375,
      "eval_icons_loss_xval": 1.659912109375,
      "eval_icons_runtime": 22.6548,
      "eval_icons_samples_per_second": 2.207,
      "eval_icons_steps_per_second": 0.088,
      "num_input_tokens_seen": 238048204,
      "step": 4250
    },
    {
      "epoch": 9.465478841870825,
      "eval_screenspot_CIoU": 0.34290045499801636,
      "eval_screenspot_GIoU": 0.36323591073354083,
      "eval_screenspot_IoU": 0.4251118103663127,
      "eval_screenspot_MAE_all": 0.06305227304498355,
      "eval_screenspot_MAE_h": 0.03816718918581804,
      "eval_screenspot_MAE_w": 0.07184332360823949,
      "eval_screenspot_MAE_x_boxes": 0.07595415661732356,
      "eval_screenspot_MAE_y_boxes": 0.04537342426677545,
      "eval_screenspot_inside_bbox": 0.6862499912579855,
      "eval_screenspot_loss": 1.6450954675674438,
      "eval_screenspot_loss_ce": 0.0002674317511264235,
      "eval_screenspot_loss_iou": 0.6758626302083334,
      "eval_screenspot_loss_num": 0.07347997029622395,
      "eval_screenspot_loss_xval": 1.7189127604166667,
      "eval_screenspot_runtime": 39.9574,
      "eval_screenspot_samples_per_second": 2.227,
      "eval_screenspot_steps_per_second": 0.075,
      "num_input_tokens_seen": 238048204,
      "step": 4250
    },
    {
      "epoch": 9.465478841870825,
      "eval_compot_CIoU": 0.3499128520488739,
      "eval_compot_GIoU": 0.3589301258325577,
      "eval_compot_IoU": 0.40847519040107727,
      "eval_compot_MAE_all": 0.018001767806708813,
      "eval_compot_MAE_h": 0.008998606353998184,
      "eval_compot_MAE_w": 0.021249551326036453,
      "eval_compot_MAE_x_boxes": 0.029959955252707005,
      "eval_compot_MAE_y_boxes": 0.007053635781630874,
      "eval_compot_inside_bbox": 0.6302083432674408,
      "eval_compot_loss": 1.3970842361450195,
      "eval_compot_loss_ce": 0.00020831655274378136,
      "eval_compot_loss_iou": 0.6414794921875,
      "eval_compot_loss_num": 0.016920089721679688,
      "eval_compot_loss_xval": 1.3681640625,
      "eval_compot_runtime": 21.8924,
      "eval_compot_samples_per_second": 2.284,
      "eval_compot_steps_per_second": 0.091,
      "num_input_tokens_seen": 238048204,
      "step": 4250
    },
    {
      "epoch": 9.465478841870825,
      "eval_custom_ui_val_CIoU": 0.4645145701037513,
      "eval_custom_ui_val_GIoU": 0.4788343757390976,
      "eval_custom_ui_val_IoU": 0.5250709156195322,
      "eval_custom_ui_val_MAE_all": 0.03104415287574132,
      "eval_custom_ui_val_MAE_h": 0.01666957584934102,
      "eval_custom_ui_val_MAE_w": 0.039596209612985454,
      "eval_custom_ui_val_MAE_x_boxes": 0.0380424867487616,
      "eval_custom_ui_val_MAE_y_boxes": 0.015655622765835788,
      "eval_custom_ui_val_inside_bbox": 0.7353395091162788,
      "eval_custom_ui_val_loss": 1.2056481838226318,
      "eval_custom_ui_val_loss_ce": 0.00023572022231140485,
      "eval_custom_ui_val_loss_iou": 0.5117323133680556,
      "eval_custom_ui_val_loss_num": 0.028479894002278645,
      "eval_custom_ui_val_loss_xval": 1.1659071180555556,
      "eval_custom_ui_val_runtime": 64.1276,
      "eval_custom_ui_val_samples_per_second": 4.132,
      "eval_custom_ui_val_steps_per_second": 0.14,
      "num_input_tokens_seen": 238048204,
      "step": 4250
    },
    {
      "epoch": 9.465478841870825,
      "loss": 0.9052478671073914,
      "loss_ce": 0.00021855966770090163,
      "loss_iou": 0.390625,
      "loss_num": 0.0250244140625,
      "loss_xval": 0.90625,
      "num_input_tokens_seen": 238048204,
      "step": 4250
    },
    {
      "epoch": 9.46770601336303,
      "grad_norm": 13.239431381225586,
      "learning_rate": 1e-06,
      "loss": 0.5228,
      "num_input_tokens_seen": 238102764,
      "step": 4251
    },
    {
      "epoch": 9.46770601336303,
      "loss": 0.4459550380706787,
      "loss_ce": 0.00015426415484398603,
      "loss_iou": 0.1982421875,
      "loss_num": 0.00994873046875,
      "loss_xval": 0.4453125,
      "num_input_tokens_seen": 238102764,
      "step": 4251
    },
    {
      "epoch": 9.469933184855234,
      "grad_norm": 16.412900924682617,
      "learning_rate": 1e-06,
      "loss": 0.5837,
      "num_input_tokens_seen": 238160024,
      "step": 4252
    },
    {
      "epoch": 9.469933184855234,
      "loss": 0.6651994585990906,
      "loss_ce": 0.00016038438479881734,
      "loss_iou": 0.265625,
      "loss_num": 0.0269775390625,
      "loss_xval": 0.6640625,
      "num_input_tokens_seen": 238160024,
      "step": 4252
    },
    {
      "epoch": 9.47216035634744,
      "grad_norm": 16.5249080657959,
      "learning_rate": 1e-06,
      "loss": 0.535,
      "num_input_tokens_seen": 238218512,
      "step": 4253
    },
    {
      "epoch": 9.47216035634744,
      "loss": 0.5461084246635437,
      "loss_ce": 0.00020996385137550533,
      "loss_iou": 0.248046875,
      "loss_num": 0.01007080078125,
      "loss_xval": 0.546875,
      "num_input_tokens_seen": 238218512,
      "step": 4253
    },
    {
      "epoch": 9.474387527839644,
      "grad_norm": 17.52018165588379,
      "learning_rate": 1e-06,
      "loss": 0.6211,
      "num_input_tokens_seen": 238277132,
      "step": 4254
    },
    {
      "epoch": 9.474387527839644,
      "loss": 0.5265886783599854,
      "loss_ce": 0.00022152790916152298,
      "loss_iou": 0.2216796875,
      "loss_num": 0.0166015625,
      "loss_xval": 0.52734375,
      "num_input_tokens_seen": 238277132,
      "step": 4254
    },
    {
      "epoch": 9.476614699331849,
      "grad_norm": 18.231037139892578,
      "learning_rate": 1e-06,
      "loss": 0.5999,
      "num_input_tokens_seen": 238331752,
      "step": 4255
    },
    {
      "epoch": 9.476614699331849,
      "loss": 0.7489705085754395,
      "loss_ce": 0.00019125922699458897,
      "loss_iou": 0.318359375,
      "loss_num": 0.022216796875,
      "loss_xval": 0.75,
      "num_input_tokens_seen": 238331752,
      "step": 4255
    },
    {
      "epoch": 9.478841870824054,
      "grad_norm": 27.039873123168945,
      "learning_rate": 1e-06,
      "loss": 0.6843,
      "num_input_tokens_seen": 238387640,
      "step": 4256
    },
    {
      "epoch": 9.478841870824054,
      "loss": 0.7631433010101318,
      "loss_ce": 0.00020384913659654558,
      "loss_iou": 0.3359375,
      "loss_num": 0.018310546875,
      "loss_xval": 0.76171875,
      "num_input_tokens_seen": 238387640,
      "step": 4256
    },
    {
      "epoch": 9.481069042316259,
      "grad_norm": 33.0074577331543,
      "learning_rate": 1e-06,
      "loss": 0.6398,
      "num_input_tokens_seen": 238440696,
      "step": 4257
    },
    {
      "epoch": 9.481069042316259,
      "loss": 0.8715030550956726,
      "loss_ce": 0.0001651808270253241,
      "loss_iou": 0.3828125,
      "loss_num": 0.0213623046875,
      "loss_xval": 0.87109375,
      "num_input_tokens_seen": 238440696,
      "step": 4257
    },
    {
      "epoch": 9.483296213808464,
      "grad_norm": 27.93197250366211,
      "learning_rate": 1e-06,
      "loss": 0.8218,
      "num_input_tokens_seen": 238494316,
      "step": 4258
    },
    {
      "epoch": 9.483296213808464,
      "loss": 0.602424681186676,
      "loss_ce": 0.0001297968119615689,
      "loss_iou": 0.255859375,
      "loss_num": 0.0184326171875,
      "loss_xval": 0.6015625,
      "num_input_tokens_seen": 238494316,
      "step": 4258
    },
    {
      "epoch": 9.485523385300668,
      "grad_norm": 23.008649826049805,
      "learning_rate": 1e-06,
      "loss": 0.7639,
      "num_input_tokens_seen": 238550660,
      "step": 4259
    },
    {
      "epoch": 9.485523385300668,
      "loss": 0.7036758065223694,
      "loss_ce": 0.0001846282830229029,
      "loss_iou": 0.296875,
      "loss_num": 0.0223388671875,
      "loss_xval": 0.703125,
      "num_input_tokens_seen": 238550660,
      "step": 4259
    },
    {
      "epoch": 9.487750556792873,
      "grad_norm": 18.077539443969727,
      "learning_rate": 1e-06,
      "loss": 0.5111,
      "num_input_tokens_seen": 238606896,
      "step": 4260
    },
    {
      "epoch": 9.487750556792873,
      "loss": 0.44276681542396545,
      "loss_ce": 0.00013988392311148345,
      "loss_iou": 0.1943359375,
      "loss_num": 0.010498046875,
      "loss_xval": 0.443359375,
      "num_input_tokens_seen": 238606896,
      "step": 4260
    },
    {
      "epoch": 9.489977728285078,
      "grad_norm": 12.999273300170898,
      "learning_rate": 1e-06,
      "loss": 0.6034,
      "num_input_tokens_seen": 238662636,
      "step": 4261
    },
    {
      "epoch": 9.489977728285078,
      "loss": 0.6766662001609802,
      "loss_ce": 0.00015251210425049067,
      "loss_iou": 0.283203125,
      "loss_num": 0.0216064453125,
      "loss_xval": 0.67578125,
      "num_input_tokens_seen": 238662636,
      "step": 4261
    },
    {
      "epoch": 9.492204899777283,
      "grad_norm": 22.135456085205078,
      "learning_rate": 1e-06,
      "loss": 0.5358,
      "num_input_tokens_seen": 238717560,
      "step": 4262
    },
    {
      "epoch": 9.492204899777283,
      "loss": 0.38586312532424927,
      "loss_ce": 0.000181969502591528,
      "loss_iou": 0.173828125,
      "loss_num": 0.00762939453125,
      "loss_xval": 0.384765625,
      "num_input_tokens_seen": 238717560,
      "step": 4262
    },
    {
      "epoch": 9.494432071269488,
      "grad_norm": 50.78253173828125,
      "learning_rate": 1e-06,
      "loss": 0.6531,
      "num_input_tokens_seen": 238775784,
      "step": 4263
    },
    {
      "epoch": 9.494432071269488,
      "loss": 0.7694522142410278,
      "loss_ce": 0.00016509005217812955,
      "loss_iou": 0.32421875,
      "loss_num": 0.0240478515625,
      "loss_xval": 0.76953125,
      "num_input_tokens_seen": 238775784,
      "step": 4263
    },
    {
      "epoch": 9.496659242761693,
      "grad_norm": 19.99468421936035,
      "learning_rate": 1e-06,
      "loss": 0.4683,
      "num_input_tokens_seen": 238834380,
      "step": 4264
    },
    {
      "epoch": 9.496659242761693,
      "loss": 0.5259066820144653,
      "loss_ce": 0.00014983075379859656,
      "loss_iou": 0.2197265625,
      "loss_num": 0.0169677734375,
      "loss_xval": 0.52734375,
      "num_input_tokens_seen": 238834380,
      "step": 4264
    },
    {
      "epoch": 9.498886414253898,
      "grad_norm": 20.870853424072266,
      "learning_rate": 1e-06,
      "loss": 0.7663,
      "num_input_tokens_seen": 238892804,
      "step": 4265
    },
    {
      "epoch": 9.498886414253898,
      "loss": 0.5954189300537109,
      "loss_ce": 0.00014309046673588455,
      "loss_iou": 0.2421875,
      "loss_num": 0.02197265625,
      "loss_xval": 0.59375,
      "num_input_tokens_seen": 238892804,
      "step": 4265
    },
    {
      "epoch": 9.501113585746102,
      "grad_norm": 16.311065673828125,
      "learning_rate": 1e-06,
      "loss": 0.5614,
      "num_input_tokens_seen": 238949684,
      "step": 4266
    },
    {
      "epoch": 9.501113585746102,
      "loss": 0.5311484336853027,
      "loss_ce": 0.00014256531721912324,
      "loss_iou": 0.2412109375,
      "loss_num": 0.010009765625,
      "loss_xval": 0.53125,
      "num_input_tokens_seen": 238949684,
      "step": 4266
    },
    {
      "epoch": 9.503340757238307,
      "grad_norm": 234.5572509765625,
      "learning_rate": 1e-06,
      "loss": 0.6156,
      "num_input_tokens_seen": 239004348,
      "step": 4267
    },
    {
      "epoch": 9.503340757238307,
      "loss": 0.7360097169876099,
      "loss_ce": 0.00041402934584766626,
      "loss_iou": 0.306640625,
      "loss_num": 0.0247802734375,
      "loss_xval": 0.734375,
      "num_input_tokens_seen": 239004348,
      "step": 4267
    },
    {
      "epoch": 9.505567928730512,
      "grad_norm": 22.837961196899414,
      "learning_rate": 1e-06,
      "loss": 0.5485,
      "num_input_tokens_seen": 239061392,
      "step": 4268
    },
    {
      "epoch": 9.505567928730512,
      "loss": 0.5970693826675415,
      "loss_ce": 0.0001455858291592449,
      "loss_iou": 0.2470703125,
      "loss_num": 0.0206298828125,
      "loss_xval": 0.59765625,
      "num_input_tokens_seen": 239061392,
      "step": 4268
    },
    {
      "epoch": 9.507795100222717,
      "grad_norm": 24.823139190673828,
      "learning_rate": 1e-06,
      "loss": 0.6112,
      "num_input_tokens_seen": 239115904,
      "step": 4269
    },
    {
      "epoch": 9.507795100222717,
      "loss": 0.7151393890380859,
      "loss_ce": 0.00017360490164719522,
      "loss_iou": 0.265625,
      "loss_num": 0.037109375,
      "loss_xval": 0.71484375,
      "num_input_tokens_seen": 239115904,
      "step": 4269
    },
    {
      "epoch": 9.510022271714922,
      "grad_norm": 31.860271453857422,
      "learning_rate": 1e-06,
      "loss": 0.5909,
      "num_input_tokens_seen": 239170464,
      "step": 4270
    },
    {
      "epoch": 9.510022271714922,
      "loss": 0.7071930170059204,
      "loss_ce": 0.00025335949612781405,
      "loss_iou": 0.3046875,
      "loss_num": 0.019287109375,
      "loss_xval": 0.70703125,
      "num_input_tokens_seen": 239170464,
      "step": 4270
    },
    {
      "epoch": 9.512249443207127,
      "grad_norm": 32.744850158691406,
      "learning_rate": 1e-06,
      "loss": 0.4666,
      "num_input_tokens_seen": 239227924,
      "step": 4271
    },
    {
      "epoch": 9.512249443207127,
      "loss": 0.31606927514076233,
      "loss_ce": 0.00027338817017152905,
      "loss_iou": 0.1435546875,
      "loss_num": 0.00567626953125,
      "loss_xval": 0.31640625,
      "num_input_tokens_seen": 239227924,
      "step": 4271
    },
    {
      "epoch": 9.514476614699332,
      "grad_norm": 24.312049865722656,
      "learning_rate": 1e-06,
      "loss": 0.6541,
      "num_input_tokens_seen": 239285676,
      "step": 4272
    },
    {
      "epoch": 9.514476614699332,
      "loss": 0.7336921691894531,
      "loss_ce": 0.0002937186509370804,
      "loss_iou": 0.291015625,
      "loss_num": 0.030517578125,
      "loss_xval": 0.734375,
      "num_input_tokens_seen": 239285676,
      "step": 4272
    },
    {
      "epoch": 9.516703786191536,
      "grad_norm": 16.174928665161133,
      "learning_rate": 1e-06,
      "loss": 0.6278,
      "num_input_tokens_seen": 239340228,
      "step": 4273
    },
    {
      "epoch": 9.516703786191536,
      "loss": 0.6358986496925354,
      "loss_ce": 0.0001564474805491045,
      "loss_iou": 0.263671875,
      "loss_num": 0.0216064453125,
      "loss_xval": 0.63671875,
      "num_input_tokens_seen": 239340228,
      "step": 4273
    },
    {
      "epoch": 9.518930957683741,
      "grad_norm": 46.78654861450195,
      "learning_rate": 1e-06,
      "loss": 0.5724,
      "num_input_tokens_seen": 239396412,
      "step": 4274
    },
    {
      "epoch": 9.518930957683741,
      "loss": 0.6387228965759277,
      "loss_ce": 0.00029518798692151904,
      "loss_iou": 0.26953125,
      "loss_num": 0.0198974609375,
      "loss_xval": 0.63671875,
      "num_input_tokens_seen": 239396412,
      "step": 4274
    },
    {
      "epoch": 9.521158129175946,
      "grad_norm": 14.414083480834961,
      "learning_rate": 1e-06,
      "loss": 0.374,
      "num_input_tokens_seen": 239450920,
      "step": 4275
    },
    {
      "epoch": 9.521158129175946,
      "loss": 0.27442389726638794,
      "loss_ce": 0.00013192339974921197,
      "loss_iou": 0.1044921875,
      "loss_num": 0.01318359375,
      "loss_xval": 0.2734375,
      "num_input_tokens_seen": 239450920,
      "step": 4275
    },
    {
      "epoch": 9.523385300668151,
      "grad_norm": 25.984390258789062,
      "learning_rate": 1e-06,
      "loss": 0.584,
      "num_input_tokens_seen": 239506768,
      "step": 4276
    },
    {
      "epoch": 9.523385300668151,
      "loss": 0.7418030500411987,
      "loss_ce": 0.0003479471488390118,
      "loss_iou": 0.314453125,
      "loss_num": 0.0220947265625,
      "loss_xval": 0.7421875,
      "num_input_tokens_seen": 239506768,
      "step": 4276
    },
    {
      "epoch": 9.525612472160356,
      "grad_norm": 25.130739212036133,
      "learning_rate": 1e-06,
      "loss": 0.5816,
      "num_input_tokens_seen": 239559000,
      "step": 4277
    },
    {
      "epoch": 9.525612472160356,
      "loss": 0.566669225692749,
      "loss_ce": 0.00014092770288698375,
      "loss_iou": 0.2333984375,
      "loss_num": 0.0201416015625,
      "loss_xval": 0.56640625,
      "num_input_tokens_seen": 239559000,
      "step": 4277
    },
    {
      "epoch": 9.52783964365256,
      "grad_norm": 18.2275447845459,
      "learning_rate": 1e-06,
      "loss": 0.4939,
      "num_input_tokens_seen": 239616448,
      "step": 4278
    },
    {
      "epoch": 9.52783964365256,
      "loss": 0.45471665263175964,
      "loss_ce": 0.0001268136693397537,
      "loss_iou": 0.2109375,
      "loss_num": 0.006591796875,
      "loss_xval": 0.455078125,
      "num_input_tokens_seen": 239616448,
      "step": 4278
    },
    {
      "epoch": 9.530066815144766,
      "grad_norm": 17.732545852661133,
      "learning_rate": 1e-06,
      "loss": 0.389,
      "num_input_tokens_seen": 239674876,
      "step": 4279
    },
    {
      "epoch": 9.530066815144766,
      "loss": 0.4159805476665497,
      "loss_ce": 0.00014803148224018514,
      "loss_iou": 0.17578125,
      "loss_num": 0.012939453125,
      "loss_xval": 0.416015625,
      "num_input_tokens_seen": 239674876,
      "step": 4279
    },
    {
      "epoch": 9.53229398663697,
      "grad_norm": 39.37736511230469,
      "learning_rate": 1e-06,
      "loss": 0.5871,
      "num_input_tokens_seen": 239732948,
      "step": 4280
    },
    {
      "epoch": 9.53229398663697,
      "loss": 0.49721166491508484,
      "loss_ce": 0.0001413416030118242,
      "loss_iou": 0.224609375,
      "loss_num": 0.00958251953125,
      "loss_xval": 0.49609375,
      "num_input_tokens_seen": 239732948,
      "step": 4280
    },
    {
      "epoch": 9.534521158129175,
      "grad_norm": 32.132293701171875,
      "learning_rate": 1e-06,
      "loss": 0.4756,
      "num_input_tokens_seen": 239789608,
      "step": 4281
    },
    {
      "epoch": 9.534521158129175,
      "loss": 0.44500744342803955,
      "loss_ce": 0.00018322419782634825,
      "loss_iou": 0.1904296875,
      "loss_num": 0.01263427734375,
      "loss_xval": 0.4453125,
      "num_input_tokens_seen": 239789608,
      "step": 4281
    },
    {
      "epoch": 9.53674832962138,
      "grad_norm": 14.220974922180176,
      "learning_rate": 1e-06,
      "loss": 0.4334,
      "num_input_tokens_seen": 239844376,
      "step": 4282
    },
    {
      "epoch": 9.53674832962138,
      "loss": 0.40624189376831055,
      "loss_ce": 0.00023602474539075047,
      "loss_iou": 0.1484375,
      "loss_num": 0.02197265625,
      "loss_xval": 0.40625,
      "num_input_tokens_seen": 239844376,
      "step": 4282
    },
    {
      "epoch": 9.538975501113585,
      "grad_norm": 23.4123592376709,
      "learning_rate": 1e-06,
      "loss": 0.5361,
      "num_input_tokens_seen": 239896980,
      "step": 4283
    },
    {
      "epoch": 9.538975501113585,
      "loss": 0.43205326795578003,
      "loss_ce": 0.00016851615509949625,
      "loss_iou": 0.1884765625,
      "loss_num": 0.01104736328125,
      "loss_xval": 0.431640625,
      "num_input_tokens_seen": 239896980,
      "step": 4283
    },
    {
      "epoch": 9.54120267260579,
      "grad_norm": 18.76468276977539,
      "learning_rate": 1e-06,
      "loss": 0.6264,
      "num_input_tokens_seen": 239953088,
      "step": 4284
    },
    {
      "epoch": 9.54120267260579,
      "loss": 0.6263715624809265,
      "loss_ce": 0.00015087236533872783,
      "loss_iou": 0.287109375,
      "loss_num": 0.01068115234375,
      "loss_xval": 0.625,
      "num_input_tokens_seen": 239953088,
      "step": 4284
    },
    {
      "epoch": 9.543429844097995,
      "grad_norm": 15.258724212646484,
      "learning_rate": 1e-06,
      "loss": 0.5531,
      "num_input_tokens_seen": 240011052,
      "step": 4285
    },
    {
      "epoch": 9.543429844097995,
      "loss": 0.5455443859100342,
      "loss_ce": 0.00013422755000647157,
      "loss_iou": 0.19921875,
      "loss_num": 0.029541015625,
      "loss_xval": 0.546875,
      "num_input_tokens_seen": 240011052,
      "step": 4285
    },
    {
      "epoch": 9.5456570155902,
      "grad_norm": 20.07108497619629,
      "learning_rate": 1e-06,
      "loss": 0.6354,
      "num_input_tokens_seen": 240061388,
      "step": 4286
    },
    {
      "epoch": 9.5456570155902,
      "loss": 0.670364260673523,
      "loss_ce": 0.00019821789464913309,
      "loss_iou": 0.287109375,
      "loss_num": 0.0194091796875,
      "loss_xval": 0.671875,
      "num_input_tokens_seen": 240061388,
      "step": 4286
    },
    {
      "epoch": 9.547884187082406,
      "grad_norm": 16.95429229736328,
      "learning_rate": 1e-06,
      "loss": 0.5388,
      "num_input_tokens_seen": 240120732,
      "step": 4287
    },
    {
      "epoch": 9.547884187082406,
      "loss": 0.6176313161849976,
      "loss_ce": 0.00013861627667210996,
      "loss_iou": 0.27734375,
      "loss_num": 0.012939453125,
      "loss_xval": 0.6171875,
      "num_input_tokens_seen": 240120732,
      "step": 4287
    },
    {
      "epoch": 9.550111358574611,
      "grad_norm": 30.3793888092041,
      "learning_rate": 1e-06,
      "loss": 0.7166,
      "num_input_tokens_seen": 240176776,
      "step": 4288
    },
    {
      "epoch": 9.550111358574611,
      "loss": 0.8734985589981079,
      "loss_ce": 0.000207598292035982,
      "loss_iou": 0.359375,
      "loss_num": 0.0308837890625,
      "loss_xval": 0.875,
      "num_input_tokens_seen": 240176776,
      "step": 4288
    },
    {
      "epoch": 9.552338530066816,
      "grad_norm": 23.587675094604492,
      "learning_rate": 1e-06,
      "loss": 0.6852,
      "num_input_tokens_seen": 240232852,
      "step": 4289
    },
    {
      "epoch": 9.552338530066816,
      "loss": 0.5840327739715576,
      "loss_ce": 0.00017050979658961296,
      "loss_iou": 0.263671875,
      "loss_num": 0.011474609375,
      "loss_xval": 0.58203125,
      "num_input_tokens_seen": 240232852,
      "step": 4289
    },
    {
      "epoch": 9.55456570155902,
      "grad_norm": 33.9605598449707,
      "learning_rate": 1e-06,
      "loss": 0.5779,
      "num_input_tokens_seen": 240288524,
      "step": 4290
    },
    {
      "epoch": 9.55456570155902,
      "loss": 0.6915979385375977,
      "loss_ce": 0.00019171604071743786,
      "loss_iou": 0.310546875,
      "loss_num": 0.01434326171875,
      "loss_xval": 0.69140625,
      "num_input_tokens_seen": 240288524,
      "step": 4290
    },
    {
      "epoch": 9.556792873051226,
      "grad_norm": 13.900967597961426,
      "learning_rate": 1e-06,
      "loss": 0.4135,
      "num_input_tokens_seen": 240344628,
      "step": 4291
    },
    {
      "epoch": 9.556792873051226,
      "loss": 0.5288236737251282,
      "loss_ce": 0.00013714321539737284,
      "loss_iou": 0.2255859375,
      "loss_num": 0.01544189453125,
      "loss_xval": 0.52734375,
      "num_input_tokens_seen": 240344628,
      "step": 4291
    },
    {
      "epoch": 9.55902004454343,
      "grad_norm": 40.7745246887207,
      "learning_rate": 1e-06,
      "loss": 0.5172,
      "num_input_tokens_seen": 240399768,
      "step": 4292
    },
    {
      "epoch": 9.55902004454343,
      "loss": 0.6885933876037598,
      "loss_ce": 0.0001778678852133453,
      "loss_iou": 0.30078125,
      "loss_num": 0.017333984375,
      "loss_xval": 0.6875,
      "num_input_tokens_seen": 240399768,
      "step": 4292
    },
    {
      "epoch": 9.561247216035635,
      "grad_norm": 24.22138786315918,
      "learning_rate": 1e-06,
      "loss": 0.6703,
      "num_input_tokens_seen": 240456252,
      "step": 4293
    },
    {
      "epoch": 9.561247216035635,
      "loss": 0.6952459812164307,
      "loss_ce": 0.00023865490220487118,
      "loss_iou": 0.265625,
      "loss_num": 0.03271484375,
      "loss_xval": 0.6953125,
      "num_input_tokens_seen": 240456252,
      "step": 4293
    },
    {
      "epoch": 9.56347438752784,
      "grad_norm": 19.527559280395508,
      "learning_rate": 1e-06,
      "loss": 0.6914,
      "num_input_tokens_seen": 240512636,
      "step": 4294
    },
    {
      "epoch": 9.56347438752784,
      "loss": 0.63869708776474,
      "loss_ce": 0.0001473345619160682,
      "loss_iou": 0.2451171875,
      "loss_num": 0.0301513671875,
      "loss_xval": 0.63671875,
      "num_input_tokens_seen": 240512636,
      "step": 4294
    },
    {
      "epoch": 9.565701559020045,
      "grad_norm": 17.53272819519043,
      "learning_rate": 1e-06,
      "loss": 0.6336,
      "num_input_tokens_seen": 240571148,
      "step": 4295
    },
    {
      "epoch": 9.565701559020045,
      "loss": 0.7293701767921448,
      "loss_ce": 0.00024417019449174404,
      "loss_iou": 0.310546875,
      "loss_num": 0.0218505859375,
      "loss_xval": 0.73046875,
      "num_input_tokens_seen": 240571148,
      "step": 4295
    },
    {
      "epoch": 9.56792873051225,
      "grad_norm": 20.05321502685547,
      "learning_rate": 1e-06,
      "loss": 0.6674,
      "num_input_tokens_seen": 240628292,
      "step": 4296
    },
    {
      "epoch": 9.56792873051225,
      "loss": 0.6266350746154785,
      "loss_ce": 0.0001702026347629726,
      "loss_iou": 0.28125,
      "loss_num": 0.0128173828125,
      "loss_xval": 0.625,
      "num_input_tokens_seen": 240628292,
      "step": 4296
    },
    {
      "epoch": 9.570155902004455,
      "grad_norm": 15.255680084228516,
      "learning_rate": 1e-06,
      "loss": 0.4882,
      "num_input_tokens_seen": 240686052,
      "step": 4297
    },
    {
      "epoch": 9.570155902004455,
      "loss": 0.5545688271522522,
      "loss_ce": 0.0001254748785868287,
      "loss_iou": 0.2265625,
      "loss_num": 0.020263671875,
      "loss_xval": 0.5546875,
      "num_input_tokens_seen": 240686052,
      "step": 4297
    },
    {
      "epoch": 9.57238307349666,
      "grad_norm": 18.931354522705078,
      "learning_rate": 1e-06,
      "loss": 0.5507,
      "num_input_tokens_seen": 240744448,
      "step": 4298
    },
    {
      "epoch": 9.57238307349666,
      "loss": 0.571701169013977,
      "loss_ce": 0.00016800707089714706,
      "loss_iou": 0.25390625,
      "loss_num": 0.0130615234375,
      "loss_xval": 0.5703125,
      "num_input_tokens_seen": 240744448,
      "step": 4298
    },
    {
      "epoch": 9.574610244988865,
      "grad_norm": 38.29421615600586,
      "learning_rate": 1e-06,
      "loss": 0.5789,
      "num_input_tokens_seen": 240800872,
      "step": 4299
    },
    {
      "epoch": 9.574610244988865,
      "loss": 0.6190475821495056,
      "loss_ce": 0.0001510926231276244,
      "loss_iou": 0.27734375,
      "loss_num": 0.01275634765625,
      "loss_xval": 0.6171875,
      "num_input_tokens_seen": 240800872,
      "step": 4299
    },
    {
      "epoch": 9.57683741648107,
      "grad_norm": 15.936655044555664,
      "learning_rate": 1e-06,
      "loss": 0.655,
      "num_input_tokens_seen": 240855308,
      "step": 4300
    },
    {
      "epoch": 9.57683741648107,
      "loss": 0.7067993879318237,
      "loss_ce": 0.00013432535342872143,
      "loss_iou": 0.29296875,
      "loss_num": 0.024658203125,
      "loss_xval": 0.70703125,
      "num_input_tokens_seen": 240855308,
      "step": 4300
    },
    {
      "epoch": 9.579064587973274,
      "grad_norm": 14.852972030639648,
      "learning_rate": 1e-06,
      "loss": 0.5593,
      "num_input_tokens_seen": 240910724,
      "step": 4301
    },
    {
      "epoch": 9.579064587973274,
      "loss": 0.6589754819869995,
      "loss_ce": 0.00016200373647734523,
      "loss_iou": 0.27734375,
      "loss_num": 0.0211181640625,
      "loss_xval": 0.66015625,
      "num_input_tokens_seen": 240910724,
      "step": 4301
    },
    {
      "epoch": 9.58129175946548,
      "grad_norm": 18.125749588012695,
      "learning_rate": 1e-06,
      "loss": 0.5183,
      "num_input_tokens_seen": 240967040,
      "step": 4302
    },
    {
      "epoch": 9.58129175946548,
      "loss": 0.404930055141449,
      "loss_ce": 0.00014489417662844062,
      "loss_iou": 0.166015625,
      "loss_num": 0.0142822265625,
      "loss_xval": 0.404296875,
      "num_input_tokens_seen": 240967040,
      "step": 4302
    },
    {
      "epoch": 9.583518930957684,
      "grad_norm": 16.510257720947266,
      "learning_rate": 1e-06,
      "loss": 0.6927,
      "num_input_tokens_seen": 241021748,
      "step": 4303
    },
    {
      "epoch": 9.583518930957684,
      "loss": 0.7239894866943359,
      "loss_ce": 0.0027980851009488106,
      "loss_iou": 0.3046875,
      "loss_num": 0.0220947265625,
      "loss_xval": 0.72265625,
      "num_input_tokens_seen": 241021748,
      "step": 4303
    },
    {
      "epoch": 9.585746102449889,
      "grad_norm": 17.652618408203125,
      "learning_rate": 1e-06,
      "loss": 0.5067,
      "num_input_tokens_seen": 241078380,
      "step": 4304
    },
    {
      "epoch": 9.585746102449889,
      "loss": 0.45593586564064026,
      "loss_ce": 0.00012532059918157756,
      "loss_iou": 0.18359375,
      "loss_num": 0.017578125,
      "loss_xval": 0.455078125,
      "num_input_tokens_seen": 241078380,
      "step": 4304
    },
    {
      "epoch": 9.587973273942094,
      "grad_norm": 21.635093688964844,
      "learning_rate": 1e-06,
      "loss": 0.6214,
      "num_input_tokens_seen": 241131336,
      "step": 4305
    },
    {
      "epoch": 9.587973273942094,
      "loss": 0.6010667085647583,
      "loss_ce": 0.00023660104488953948,
      "loss_iou": 0.263671875,
      "loss_num": 0.01507568359375,
      "loss_xval": 0.6015625,
      "num_input_tokens_seen": 241131336,
      "step": 4305
    },
    {
      "epoch": 9.590200445434299,
      "grad_norm": 21.13984489440918,
      "learning_rate": 1e-06,
      "loss": 0.5725,
      "num_input_tokens_seen": 241188112,
      "step": 4306
    },
    {
      "epoch": 9.590200445434299,
      "loss": 0.526428759098053,
      "loss_ce": 0.0003057234571315348,
      "loss_iou": 0.208984375,
      "loss_num": 0.0216064453125,
      "loss_xval": 0.52734375,
      "num_input_tokens_seen": 241188112,
      "step": 4306
    },
    {
      "epoch": 9.592427616926503,
      "grad_norm": 16.988536834716797,
      "learning_rate": 1e-06,
      "loss": 0.614,
      "num_input_tokens_seen": 241248100,
      "step": 4307
    },
    {
      "epoch": 9.592427616926503,
      "loss": 0.6281224489212036,
      "loss_ce": 0.00019275453814771026,
      "loss_iou": 0.2734375,
      "loss_num": 0.0162353515625,
      "loss_xval": 0.62890625,
      "num_input_tokens_seen": 241248100,
      "step": 4307
    },
    {
      "epoch": 9.594654788418708,
      "grad_norm": 21.287412643432617,
      "learning_rate": 1e-06,
      "loss": 0.7941,
      "num_input_tokens_seen": 241303028,
      "step": 4308
    },
    {
      "epoch": 9.594654788418708,
      "loss": 1.0175116062164307,
      "loss_ce": 0.00017758479225449264,
      "loss_iou": 0.416015625,
      "loss_num": 0.036865234375,
      "loss_xval": 1.015625,
      "num_input_tokens_seen": 241303028,
      "step": 4308
    },
    {
      "epoch": 9.596881959910913,
      "grad_norm": 17.50885009765625,
      "learning_rate": 1e-06,
      "loss": 0.5592,
      "num_input_tokens_seen": 241359508,
      "step": 4309
    },
    {
      "epoch": 9.596881959910913,
      "loss": 0.6614413261413574,
      "loss_ce": 0.00018642100621946156,
      "loss_iou": 0.283203125,
      "loss_num": 0.0194091796875,
      "loss_xval": 0.66015625,
      "num_input_tokens_seen": 241359508,
      "step": 4309
    },
    {
      "epoch": 9.599109131403118,
      "grad_norm": 23.472171783447266,
      "learning_rate": 1e-06,
      "loss": 0.7445,
      "num_input_tokens_seen": 241417160,
      "step": 4310
    },
    {
      "epoch": 9.599109131403118,
      "loss": 0.797520101070404,
      "loss_ce": 0.00015685184916947037,
      "loss_iou": 0.3359375,
      "loss_num": 0.0247802734375,
      "loss_xval": 0.796875,
      "num_input_tokens_seen": 241417160,
      "step": 4310
    },
    {
      "epoch": 9.601336302895323,
      "grad_norm": 29.62855339050293,
      "learning_rate": 1e-06,
      "loss": 0.7193,
      "num_input_tokens_seen": 241473892,
      "step": 4311
    },
    {
      "epoch": 9.601336302895323,
      "loss": 0.7771183848381042,
      "loss_ce": 0.00014080063556320965,
      "loss_iou": 0.33203125,
      "loss_num": 0.0230712890625,
      "loss_xval": 0.77734375,
      "num_input_tokens_seen": 241473892,
      "step": 4311
    },
    {
      "epoch": 9.603563474387528,
      "grad_norm": 19.706748962402344,
      "learning_rate": 1e-06,
      "loss": 0.6076,
      "num_input_tokens_seen": 241532188,
      "step": 4312
    },
    {
      "epoch": 9.603563474387528,
      "loss": 0.7907249927520752,
      "loss_ce": 0.00031976206810213625,
      "loss_iou": 0.333984375,
      "loss_num": 0.0250244140625,
      "loss_xval": 0.7890625,
      "num_input_tokens_seen": 241532188,
      "step": 4312
    },
    {
      "epoch": 9.605790645879733,
      "grad_norm": 19.344236373901367,
      "learning_rate": 1e-06,
      "loss": 0.565,
      "num_input_tokens_seen": 241587928,
      "step": 4313
    },
    {
      "epoch": 9.605790645879733,
      "loss": 0.6804521083831787,
      "loss_ce": 0.0001542992249596864,
      "loss_iou": 0.30078125,
      "loss_num": 0.01556396484375,
      "loss_xval": 0.6796875,
      "num_input_tokens_seen": 241587928,
      "step": 4313
    },
    {
      "epoch": 9.608017817371937,
      "grad_norm": 15.50113582611084,
      "learning_rate": 1e-06,
      "loss": 0.399,
      "num_input_tokens_seen": 241640704,
      "step": 4314
    },
    {
      "epoch": 9.608017817371937,
      "loss": 0.44986775517463684,
      "loss_ce": 0.00013020676851738244,
      "loss_iou": 0.2041015625,
      "loss_num": 0.00836181640625,
      "loss_xval": 0.44921875,
      "num_input_tokens_seen": 241640704,
      "step": 4314
    },
    {
      "epoch": 9.610244988864142,
      "grad_norm": 36.39356994628906,
      "learning_rate": 1e-06,
      "loss": 0.559,
      "num_input_tokens_seen": 241693484,
      "step": 4315
    },
    {
      "epoch": 9.610244988864142,
      "loss": 0.7055351138114929,
      "loss_ce": 0.000456988753285259,
      "loss_iou": 0.28125,
      "loss_num": 0.028076171875,
      "loss_xval": 0.703125,
      "num_input_tokens_seen": 241693484,
      "step": 4315
    },
    {
      "epoch": 9.612472160356347,
      "grad_norm": 16.385406494140625,
      "learning_rate": 1e-06,
      "loss": 0.6239,
      "num_input_tokens_seen": 241747860,
      "step": 4316
    },
    {
      "epoch": 9.612472160356347,
      "loss": 0.7735809087753296,
      "loss_ce": 0.0005096091772429645,
      "loss_iou": 0.345703125,
      "loss_num": 0.0166015625,
      "loss_xval": 0.7734375,
      "num_input_tokens_seen": 241747860,
      "step": 4316
    },
    {
      "epoch": 9.614699331848552,
      "grad_norm": 18.57633399963379,
      "learning_rate": 1e-06,
      "loss": 0.6124,
      "num_input_tokens_seen": 241802996,
      "step": 4317
    },
    {
      "epoch": 9.614699331848552,
      "loss": 0.6089174747467041,
      "loss_ce": 0.0001528042630525306,
      "loss_iou": 0.2578125,
      "loss_num": 0.0189208984375,
      "loss_xval": 0.609375,
      "num_input_tokens_seen": 241802996,
      "step": 4317
    },
    {
      "epoch": 9.616926503340757,
      "grad_norm": 17.626914978027344,
      "learning_rate": 1e-06,
      "loss": 0.4423,
      "num_input_tokens_seen": 241857684,
      "step": 4318
    },
    {
      "epoch": 9.616926503340757,
      "loss": 0.41496047377586365,
      "loss_ce": 0.0001655758242122829,
      "loss_iou": 0.1796875,
      "loss_num": 0.01092529296875,
      "loss_xval": 0.4140625,
      "num_input_tokens_seen": 241857684,
      "step": 4318
    },
    {
      "epoch": 9.619153674832962,
      "grad_norm": 23.88527488708496,
      "learning_rate": 1e-06,
      "loss": 0.6528,
      "num_input_tokens_seen": 241911896,
      "step": 4319
    },
    {
      "epoch": 9.619153674832962,
      "loss": 0.8219267129898071,
      "loss_ce": 0.00014938012463971972,
      "loss_iou": 0.34765625,
      "loss_num": 0.0250244140625,
      "loss_xval": 0.8203125,
      "num_input_tokens_seen": 241911896,
      "step": 4319
    },
    {
      "epoch": 9.621380846325167,
      "grad_norm": 14.68041706085205,
      "learning_rate": 1e-06,
      "loss": 0.3875,
      "num_input_tokens_seen": 241969944,
      "step": 4320
    },
    {
      "epoch": 9.621380846325167,
      "loss": 0.4286039471626282,
      "loss_ce": 0.00013716876856051385,
      "loss_iou": 0.1865234375,
      "loss_num": 0.01116943359375,
      "loss_xval": 0.427734375,
      "num_input_tokens_seen": 241969944,
      "step": 4320
    },
    {
      "epoch": 9.623608017817372,
      "grad_norm": 24.28359603881836,
      "learning_rate": 1e-06,
      "loss": 0.6069,
      "num_input_tokens_seen": 242029492,
      "step": 4321
    },
    {
      "epoch": 9.623608017817372,
      "loss": 0.6774266958236694,
      "loss_ce": 0.00018064204778056592,
      "loss_iou": 0.318359375,
      "loss_num": 0.00811767578125,
      "loss_xval": 0.67578125,
      "num_input_tokens_seen": 242029492,
      "step": 4321
    },
    {
      "epoch": 9.625835189309576,
      "grad_norm": 22.979413986206055,
      "learning_rate": 1e-06,
      "loss": 0.6953,
      "num_input_tokens_seen": 242084892,
      "step": 4322
    },
    {
      "epoch": 9.625835189309576,
      "loss": 0.7274185419082642,
      "loss_ce": 0.0001846232480602339,
      "loss_iou": 0.31640625,
      "loss_num": 0.0189208984375,
      "loss_xval": 0.7265625,
      "num_input_tokens_seen": 242084892,
      "step": 4322
    },
    {
      "epoch": 9.628062360801781,
      "grad_norm": 17.841794967651367,
      "learning_rate": 1e-06,
      "loss": 0.468,
      "num_input_tokens_seen": 242139172,
      "step": 4323
    },
    {
      "epoch": 9.628062360801781,
      "loss": 0.5101624131202698,
      "loss_ce": 0.00015266051923390478,
      "loss_iou": 0.2314453125,
      "loss_num": 0.00958251953125,
      "loss_xval": 0.51171875,
      "num_input_tokens_seen": 242139172,
      "step": 4323
    },
    {
      "epoch": 9.630289532293986,
      "grad_norm": 20.742664337158203,
      "learning_rate": 1e-06,
      "loss": 0.517,
      "num_input_tokens_seen": 242196196,
      "step": 4324
    },
    {
      "epoch": 9.630289532293986,
      "loss": 0.5766589045524597,
      "loss_ce": 0.0008532421197742224,
      "loss_iou": 0.2373046875,
      "loss_num": 0.020263671875,
      "loss_xval": 0.57421875,
      "num_input_tokens_seen": 242196196,
      "step": 4324
    },
    {
      "epoch": 9.632516703786191,
      "grad_norm": 60.28476333618164,
      "learning_rate": 1e-06,
      "loss": 0.6853,
      "num_input_tokens_seen": 242251812,
      "step": 4325
    },
    {
      "epoch": 9.632516703786191,
      "loss": 0.5879150629043579,
      "loss_ce": 0.00014649088552687317,
      "loss_iou": 0.2470703125,
      "loss_num": 0.0185546875,
      "loss_xval": 0.5859375,
      "num_input_tokens_seen": 242251812,
      "step": 4325
    },
    {
      "epoch": 9.634743875278396,
      "grad_norm": 66.29544067382812,
      "learning_rate": 1e-06,
      "loss": 0.6181,
      "num_input_tokens_seen": 242308236,
      "step": 4326
    },
    {
      "epoch": 9.634743875278396,
      "loss": 0.5924375057220459,
      "loss_ce": 0.00015232324949465692,
      "loss_iou": 0.232421875,
      "loss_num": 0.0255126953125,
      "loss_xval": 0.59375,
      "num_input_tokens_seen": 242308236,
      "step": 4326
    },
    {
      "epoch": 9.6369710467706,
      "grad_norm": 14.150559425354004,
      "learning_rate": 1e-06,
      "loss": 0.4069,
      "num_input_tokens_seen": 242364656,
      "step": 4327
    },
    {
      "epoch": 9.6369710467706,
      "loss": 0.4071110486984253,
      "loss_ce": 0.00012862987932749093,
      "loss_iou": 0.1826171875,
      "loss_num": 0.00848388671875,
      "loss_xval": 0.40625,
      "num_input_tokens_seen": 242364656,
      "step": 4327
    },
    {
      "epoch": 9.639198218262806,
      "grad_norm": 21.00891876220703,
      "learning_rate": 1e-06,
      "loss": 0.6717,
      "num_input_tokens_seen": 242422764,
      "step": 4328
    },
    {
      "epoch": 9.639198218262806,
      "loss": 0.8691731691360474,
      "loss_ce": 0.0029012321028858423,
      "loss_iou": 0.345703125,
      "loss_num": 0.03466796875,
      "loss_xval": 0.8671875,
      "num_input_tokens_seen": 242422764,
      "step": 4328
    },
    {
      "epoch": 9.64142538975501,
      "grad_norm": 21.65594482421875,
      "learning_rate": 1e-06,
      "loss": 0.5172,
      "num_input_tokens_seen": 242480756,
      "step": 4329
    },
    {
      "epoch": 9.64142538975501,
      "loss": 0.5980610251426697,
      "loss_ce": 0.00016064970986917615,
      "loss_iou": 0.2451171875,
      "loss_num": 0.0218505859375,
      "loss_xval": 0.59765625,
      "num_input_tokens_seen": 242480756,
      "step": 4329
    },
    {
      "epoch": 9.643652561247215,
      "grad_norm": 19.103317260742188,
      "learning_rate": 1e-06,
      "loss": 0.7951,
      "num_input_tokens_seen": 242536328,
      "step": 4330
    },
    {
      "epoch": 9.643652561247215,
      "loss": 0.6043500900268555,
      "loss_ce": 0.00022415511193685234,
      "loss_iou": 0.25,
      "loss_num": 0.0203857421875,
      "loss_xval": 0.60546875,
      "num_input_tokens_seen": 242536328,
      "step": 4330
    },
    {
      "epoch": 9.64587973273942,
      "grad_norm": 18.600261688232422,
      "learning_rate": 1e-06,
      "loss": 0.6719,
      "num_input_tokens_seen": 242594076,
      "step": 4331
    },
    {
      "epoch": 9.64587973273942,
      "loss": 0.7887901663780212,
      "loss_ce": 0.00021596970327664167,
      "loss_iou": 0.3203125,
      "loss_num": 0.0299072265625,
      "loss_xval": 0.7890625,
      "num_input_tokens_seen": 242594076,
      "step": 4331
    },
    {
      "epoch": 9.648106904231625,
      "grad_norm": 23.29888343811035,
      "learning_rate": 1e-06,
      "loss": 0.5155,
      "num_input_tokens_seen": 242647136,
      "step": 4332
    },
    {
      "epoch": 9.648106904231625,
      "loss": 0.44800740480422974,
      "loss_ce": 0.00013143845717422664,
      "loss_iou": 0.185546875,
      "loss_num": 0.015380859375,
      "loss_xval": 0.447265625,
      "num_input_tokens_seen": 242647136,
      "step": 4332
    },
    {
      "epoch": 9.65033407572383,
      "grad_norm": 24.5333194732666,
      "learning_rate": 1e-06,
      "loss": 0.5333,
      "num_input_tokens_seen": 242700824,
      "step": 4333
    },
    {
      "epoch": 9.65033407572383,
      "loss": 0.5556710958480835,
      "loss_ce": 0.00012911748490296304,
      "loss_iou": 0.25,
      "loss_num": 0.0108642578125,
      "loss_xval": 0.5546875,
      "num_input_tokens_seen": 242700824,
      "step": 4333
    },
    {
      "epoch": 9.652561247216035,
      "grad_norm": 17.157747268676758,
      "learning_rate": 1e-06,
      "loss": 0.7577,
      "num_input_tokens_seen": 242759000,
      "step": 4334
    },
    {
      "epoch": 9.652561247216035,
      "loss": 0.7171846628189087,
      "loss_ce": 0.00014364052913151681,
      "loss_iou": 0.31640625,
      "loss_num": 0.0172119140625,
      "loss_xval": 0.71875,
      "num_input_tokens_seen": 242759000,
      "step": 4334
    },
    {
      "epoch": 9.654788418708241,
      "grad_norm": 26.958030700683594,
      "learning_rate": 1e-06,
      "loss": 0.5635,
      "num_input_tokens_seen": 242814612,
      "step": 4335
    },
    {
      "epoch": 9.654788418708241,
      "loss": 0.517261266708374,
      "loss_ce": 0.00017140991985797882,
      "loss_iou": 0.234375,
      "loss_num": 0.0096435546875,
      "loss_xval": 0.515625,
      "num_input_tokens_seen": 242814612,
      "step": 4335
    },
    {
      "epoch": 9.657015590200446,
      "grad_norm": 13.9208345413208,
      "learning_rate": 1e-06,
      "loss": 0.5748,
      "num_input_tokens_seen": 242871532,
      "step": 4336
    },
    {
      "epoch": 9.657015590200446,
      "loss": 0.7538334131240845,
      "loss_ce": 0.00017131002095993608,
      "loss_iou": 0.283203125,
      "loss_num": 0.037109375,
      "loss_xval": 0.75390625,
      "num_input_tokens_seen": 242871532,
      "step": 4336
    },
    {
      "epoch": 9.659242761692651,
      "grad_norm": 19.94957733154297,
      "learning_rate": 1e-06,
      "loss": 0.5622,
      "num_input_tokens_seen": 242927132,
      "step": 4337
    },
    {
      "epoch": 9.659242761692651,
      "loss": 0.6652075052261353,
      "loss_ce": 0.00016842935292515904,
      "loss_iou": 0.2890625,
      "loss_num": 0.0169677734375,
      "loss_xval": 0.6640625,
      "num_input_tokens_seen": 242927132,
      "step": 4337
    },
    {
      "epoch": 9.661469933184856,
      "grad_norm": 18.527286529541016,
      "learning_rate": 1e-06,
      "loss": 0.7037,
      "num_input_tokens_seen": 242980644,
      "step": 4338
    },
    {
      "epoch": 9.661469933184856,
      "loss": 0.798882007598877,
      "loss_ce": 0.00017601504805497825,
      "loss_iou": 0.341796875,
      "loss_num": 0.0234375,
      "loss_xval": 0.796875,
      "num_input_tokens_seen": 242980644,
      "step": 4338
    },
    {
      "epoch": 9.66369710467706,
      "grad_norm": 24.58388328552246,
      "learning_rate": 1e-06,
      "loss": 0.623,
      "num_input_tokens_seen": 243034252,
      "step": 4339
    },
    {
      "epoch": 9.66369710467706,
      "loss": 0.56207674741745,
      "loss_ce": 0.0003091880353167653,
      "loss_iou": 0.2578125,
      "loss_num": 0.00958251953125,
      "loss_xval": 0.5625,
      "num_input_tokens_seen": 243034252,
      "step": 4339
    },
    {
      "epoch": 9.665924276169266,
      "grad_norm": 22.670303344726562,
      "learning_rate": 1e-06,
      "loss": 0.6743,
      "num_input_tokens_seen": 243091100,
      "step": 4340
    },
    {
      "epoch": 9.665924276169266,
      "loss": 0.5509505867958069,
      "loss_ce": 0.00016935347230173647,
      "loss_iou": 0.251953125,
      "loss_num": 0.0093994140625,
      "loss_xval": 0.55078125,
      "num_input_tokens_seen": 243091100,
      "step": 4340
    },
    {
      "epoch": 9.66815144766147,
      "grad_norm": 24.056676864624023,
      "learning_rate": 1e-06,
      "loss": 0.7283,
      "num_input_tokens_seen": 243144624,
      "step": 4341
    },
    {
      "epoch": 9.66815144766147,
      "loss": 0.876803994178772,
      "loss_ce": 0.0002171122032450512,
      "loss_iou": 0.375,
      "loss_num": 0.0255126953125,
      "loss_xval": 0.875,
      "num_input_tokens_seen": 243144624,
      "step": 4341
    },
    {
      "epoch": 9.670378619153675,
      "grad_norm": 24.195653915405273,
      "learning_rate": 1e-06,
      "loss": 0.5525,
      "num_input_tokens_seen": 243202736,
      "step": 4342
    },
    {
      "epoch": 9.670378619153675,
      "loss": 0.5127952098846436,
      "loss_ce": 0.00022197004000190645,
      "loss_iou": 0.2158203125,
      "loss_num": 0.0162353515625,
      "loss_xval": 0.51171875,
      "num_input_tokens_seen": 243202736,
      "step": 4342
    },
    {
      "epoch": 9.67260579064588,
      "grad_norm": 20.252214431762695,
      "learning_rate": 1e-06,
      "loss": 0.5979,
      "num_input_tokens_seen": 243260404,
      "step": 4343
    },
    {
      "epoch": 9.67260579064588,
      "loss": 0.569305956363678,
      "loss_ce": 0.00021413374633993953,
      "loss_iou": 0.2421875,
      "loss_num": 0.01708984375,
      "loss_xval": 0.5703125,
      "num_input_tokens_seen": 243260404,
      "step": 4343
    },
    {
      "epoch": 9.674832962138085,
      "grad_norm": 22.576656341552734,
      "learning_rate": 1e-06,
      "loss": 0.6199,
      "num_input_tokens_seen": 243315828,
      "step": 4344
    },
    {
      "epoch": 9.674832962138085,
      "loss": 0.7008746266365051,
      "loss_ce": 0.00019105095998384058,
      "loss_iou": 0.302734375,
      "loss_num": 0.0194091796875,
      "loss_xval": 0.69921875,
      "num_input_tokens_seen": 243315828,
      "step": 4344
    },
    {
      "epoch": 9.67706013363029,
      "grad_norm": 23.109046936035156,
      "learning_rate": 1e-06,
      "loss": 0.6222,
      "num_input_tokens_seen": 243371304,
      "step": 4345
    },
    {
      "epoch": 9.67706013363029,
      "loss": 0.6012382507324219,
      "loss_ce": 0.0001640522968955338,
      "loss_iou": 0.265625,
      "loss_num": 0.01409912109375,
      "loss_xval": 0.6015625,
      "num_input_tokens_seen": 243371304,
      "step": 4345
    },
    {
      "epoch": 9.679287305122495,
      "grad_norm": 17.9921875,
      "learning_rate": 1e-06,
      "loss": 0.6403,
      "num_input_tokens_seen": 243421812,
      "step": 4346
    },
    {
      "epoch": 9.679287305122495,
      "loss": 0.7237908840179443,
      "loss_ce": 0.00015808289754204452,
      "loss_iou": 0.30859375,
      "loss_num": 0.0216064453125,
      "loss_xval": 0.72265625,
      "num_input_tokens_seen": 243421812,
      "step": 4346
    },
    {
      "epoch": 9.6815144766147,
      "grad_norm": 22.943727493286133,
      "learning_rate": 1e-06,
      "loss": 0.5186,
      "num_input_tokens_seen": 243479024,
      "step": 4347
    },
    {
      "epoch": 9.6815144766147,
      "loss": 0.4597442150115967,
      "loss_ce": 0.0014922881964594126,
      "loss_iou": 0.2021484375,
      "loss_num": 0.0107421875,
      "loss_xval": 0.458984375,
      "num_input_tokens_seen": 243479024,
      "step": 4347
    },
    {
      "epoch": 9.683741648106905,
      "grad_norm": 24.374197006225586,
      "learning_rate": 1e-06,
      "loss": 0.6149,
      "num_input_tokens_seen": 243535892,
      "step": 4348
    },
    {
      "epoch": 9.683741648106905,
      "loss": 0.7361659407615662,
      "loss_ce": 0.00020402228983584791,
      "loss_iou": 0.3125,
      "loss_num": 0.0223388671875,
      "loss_xval": 0.734375,
      "num_input_tokens_seen": 243535892,
      "step": 4348
    },
    {
      "epoch": 9.68596881959911,
      "grad_norm": 22.339231491088867,
      "learning_rate": 1e-06,
      "loss": 0.735,
      "num_input_tokens_seen": 243591976,
      "step": 4349
    },
    {
      "epoch": 9.68596881959911,
      "loss": 0.8295117020606995,
      "loss_ce": 0.00016598097863607109,
      "loss_iou": 0.373046875,
      "loss_num": 0.016845703125,
      "loss_xval": 0.828125,
      "num_input_tokens_seen": 243591976,
      "step": 4349
    },
    {
      "epoch": 9.688195991091314,
      "grad_norm": 14.869282722473145,
      "learning_rate": 1e-06,
      "loss": 0.4473,
      "num_input_tokens_seen": 243648860,
      "step": 4350
    },
    {
      "epoch": 9.688195991091314,
      "loss": 0.5270088911056519,
      "loss_ce": 0.00015341459948103875,
      "loss_iou": 0.240234375,
      "loss_num": 0.00921630859375,
      "loss_xval": 0.52734375,
      "num_input_tokens_seen": 243648860,
      "step": 4350
    },
    {
      "epoch": 9.690423162583519,
      "grad_norm": 18.157405853271484,
      "learning_rate": 1e-06,
      "loss": 0.6345,
      "num_input_tokens_seen": 243705736,
      "step": 4351
    },
    {
      "epoch": 9.690423162583519,
      "loss": 0.6656748056411743,
      "loss_ce": 0.0001474730233894661,
      "loss_iou": 0.27734375,
      "loss_num": 0.0220947265625,
      "loss_xval": 0.6640625,
      "num_input_tokens_seen": 243705736,
      "step": 4351
    },
    {
      "epoch": 9.692650334075724,
      "grad_norm": 20.70793342590332,
      "learning_rate": 1e-06,
      "loss": 0.6055,
      "num_input_tokens_seen": 243763076,
      "step": 4352
    },
    {
      "epoch": 9.692650334075724,
      "loss": 0.6417899131774902,
      "loss_ce": 0.0001883807562990114,
      "loss_iou": 0.28125,
      "loss_num": 0.015625,
      "loss_xval": 0.640625,
      "num_input_tokens_seen": 243763076,
      "step": 4352
    },
    {
      "epoch": 9.694877505567929,
      "grad_norm": 16.69988441467285,
      "learning_rate": 1e-06,
      "loss": 0.4398,
      "num_input_tokens_seen": 243819884,
      "step": 4353
    },
    {
      "epoch": 9.694877505567929,
      "loss": 0.4072994291782379,
      "loss_ce": 0.00019493894069455564,
      "loss_iou": 0.1796875,
      "loss_num": 0.009521484375,
      "loss_xval": 0.40625,
      "num_input_tokens_seen": 243819884,
      "step": 4353
    },
    {
      "epoch": 9.697104677060134,
      "grad_norm": 18.394502639770508,
      "learning_rate": 1e-06,
      "loss": 0.709,
      "num_input_tokens_seen": 243874696,
      "step": 4354
    },
    {
      "epoch": 9.697104677060134,
      "loss": 0.7145628333091736,
      "loss_ce": 0.00020736586884595454,
      "loss_iou": 0.3125,
      "loss_num": 0.017822265625,
      "loss_xval": 0.71484375,
      "num_input_tokens_seen": 243874696,
      "step": 4354
    },
    {
      "epoch": 9.699331848552339,
      "grad_norm": 25.20227813720703,
      "learning_rate": 1e-06,
      "loss": 0.6424,
      "num_input_tokens_seen": 243930484,
      "step": 4355
    },
    {
      "epoch": 9.699331848552339,
      "loss": 0.6150010228157043,
      "loss_ce": 0.00013285188470035791,
      "loss_iou": 0.267578125,
      "loss_num": 0.0162353515625,
      "loss_xval": 0.61328125,
      "num_input_tokens_seen": 243930484,
      "step": 4355
    },
    {
      "epoch": 9.701559020044543,
      "grad_norm": 21.711627960205078,
      "learning_rate": 1e-06,
      "loss": 0.6181,
      "num_input_tokens_seen": 243988028,
      "step": 4356
    },
    {
      "epoch": 9.701559020044543,
      "loss": 0.7733531594276428,
      "loss_ce": 0.00015979193267412484,
      "loss_iou": 0.3359375,
      "loss_num": 0.0206298828125,
      "loss_xval": 0.7734375,
      "num_input_tokens_seen": 243988028,
      "step": 4356
    },
    {
      "epoch": 9.703786191536748,
      "grad_norm": 23.96913719177246,
      "learning_rate": 1e-06,
      "loss": 0.5364,
      "num_input_tokens_seen": 244042796,
      "step": 4357
    },
    {
      "epoch": 9.703786191536748,
      "loss": 0.43648862838745117,
      "loss_ce": 0.00020933088671881706,
      "loss_iou": 0.1904296875,
      "loss_num": 0.01123046875,
      "loss_xval": 0.435546875,
      "num_input_tokens_seen": 244042796,
      "step": 4357
    },
    {
      "epoch": 9.706013363028953,
      "grad_norm": 16.607881546020508,
      "learning_rate": 1e-06,
      "loss": 0.7013,
      "num_input_tokens_seen": 244100220,
      "step": 4358
    },
    {
      "epoch": 9.706013363028953,
      "loss": 0.651787281036377,
      "loss_ce": 0.00017597780970390886,
      "loss_iou": 0.302734375,
      "loss_num": 0.00897216796875,
      "loss_xval": 0.65234375,
      "num_input_tokens_seen": 244100220,
      "step": 4358
    },
    {
      "epoch": 9.708240534521158,
      "grad_norm": 17.134170532226562,
      "learning_rate": 1e-06,
      "loss": 0.596,
      "num_input_tokens_seen": 244158116,
      "step": 4359
    },
    {
      "epoch": 9.708240534521158,
      "loss": 0.6266348958015442,
      "loss_ce": 0.0001700021093711257,
      "loss_iou": 0.271484375,
      "loss_num": 0.0167236328125,
      "loss_xval": 0.625,
      "num_input_tokens_seen": 244158116,
      "step": 4359
    },
    {
      "epoch": 9.710467706013363,
      "grad_norm": 18.850582122802734,
      "learning_rate": 1e-06,
      "loss": 0.4285,
      "num_input_tokens_seen": 244214688,
      "step": 4360
    },
    {
      "epoch": 9.710467706013363,
      "loss": 0.4430268406867981,
      "loss_ce": 0.0001557298528496176,
      "loss_iou": 0.1943359375,
      "loss_num": 0.0106201171875,
      "loss_xval": 0.443359375,
      "num_input_tokens_seen": 244214688,
      "step": 4360
    },
    {
      "epoch": 9.712694877505568,
      "grad_norm": 15.182866096496582,
      "learning_rate": 1e-06,
      "loss": 0.461,
      "num_input_tokens_seen": 244272276,
      "step": 4361
    },
    {
      "epoch": 9.712694877505568,
      "loss": 0.4096102714538574,
      "loss_ce": 0.00018646058742888272,
      "loss_iou": 0.1826171875,
      "loss_num": 0.0087890625,
      "loss_xval": 0.41015625,
      "num_input_tokens_seen": 244272276,
      "step": 4361
    },
    {
      "epoch": 9.714922048997773,
      "grad_norm": 21.032060623168945,
      "learning_rate": 1e-06,
      "loss": 0.5245,
      "num_input_tokens_seen": 244326084,
      "step": 4362
    },
    {
      "epoch": 9.714922048997773,
      "loss": 0.5369995832443237,
      "loss_ce": 0.00013436358131002635,
      "loss_iou": 0.2353515625,
      "loss_num": 0.01318359375,
      "loss_xval": 0.53515625,
      "num_input_tokens_seen": 244326084,
      "step": 4362
    },
    {
      "epoch": 9.717149220489977,
      "grad_norm": 16.875015258789062,
      "learning_rate": 1e-06,
      "loss": 0.7523,
      "num_input_tokens_seen": 244383708,
      "step": 4363
    },
    {
      "epoch": 9.717149220489977,
      "loss": 0.8245912790298462,
      "loss_ce": 0.00025050187832675874,
      "loss_iou": 0.3203125,
      "loss_num": 0.036865234375,
      "loss_xval": 0.82421875,
      "num_input_tokens_seen": 244383708,
      "step": 4363
    },
    {
      "epoch": 9.719376391982182,
      "grad_norm": 15.091886520385742,
      "learning_rate": 1e-06,
      "loss": 0.4595,
      "num_input_tokens_seen": 244436700,
      "step": 4364
    },
    {
      "epoch": 9.719376391982182,
      "loss": 0.2986956834793091,
      "loss_ce": 0.00053892977302894,
      "loss_iou": 0.11376953125,
      "loss_num": 0.01416015625,
      "loss_xval": 0.298828125,
      "num_input_tokens_seen": 244436700,
      "step": 4364
    },
    {
      "epoch": 9.721603563474387,
      "grad_norm": 36.89336013793945,
      "learning_rate": 1e-06,
      "loss": 0.6042,
      "num_input_tokens_seen": 244492464,
      "step": 4365
    },
    {
      "epoch": 9.721603563474387,
      "loss": 0.7046360373497009,
      "loss_ce": 0.0010227648308500648,
      "loss_iou": 0.30078125,
      "loss_num": 0.0203857421875,
      "loss_xval": 0.703125,
      "num_input_tokens_seen": 244492464,
      "step": 4365
    },
    {
      "epoch": 9.723830734966592,
      "grad_norm": 21.095430374145508,
      "learning_rate": 1e-06,
      "loss": 0.7696,
      "num_input_tokens_seen": 244548864,
      "step": 4366
    },
    {
      "epoch": 9.723830734966592,
      "loss": 0.7162793874740601,
      "loss_ce": 0.0007032152498140931,
      "loss_iou": 0.318359375,
      "loss_num": 0.0152587890625,
      "loss_xval": 0.71484375,
      "num_input_tokens_seen": 244548864,
      "step": 4366
    },
    {
      "epoch": 9.726057906458797,
      "grad_norm": 17.61524772644043,
      "learning_rate": 1e-06,
      "loss": 0.4847,
      "num_input_tokens_seen": 244602692,
      "step": 4367
    },
    {
      "epoch": 9.726057906458797,
      "loss": 0.505713939666748,
      "loss_ce": 0.00022077930043451488,
      "loss_iou": 0.23046875,
      "loss_num": 0.00885009765625,
      "loss_xval": 0.50390625,
      "num_input_tokens_seen": 244602692,
      "step": 4367
    },
    {
      "epoch": 9.728285077951002,
      "grad_norm": 23.061426162719727,
      "learning_rate": 1e-06,
      "loss": 0.519,
      "num_input_tokens_seen": 244657716,
      "step": 4368
    },
    {
      "epoch": 9.728285077951002,
      "loss": 0.48245200514793396,
      "loss_ce": 0.00015220226487144828,
      "loss_iou": 0.208984375,
      "loss_num": 0.0128173828125,
      "loss_xval": 0.482421875,
      "num_input_tokens_seen": 244657716,
      "step": 4368
    },
    {
      "epoch": 9.730512249443207,
      "grad_norm": 19.16279411315918,
      "learning_rate": 1e-06,
      "loss": 0.5937,
      "num_input_tokens_seen": 244711224,
      "step": 4369
    },
    {
      "epoch": 9.730512249443207,
      "loss": 0.7714163661003113,
      "loss_ce": 0.0001761687162797898,
      "loss_iou": 0.333984375,
      "loss_num": 0.0203857421875,
      "loss_xval": 0.76953125,
      "num_input_tokens_seen": 244711224,
      "step": 4369
    },
    {
      "epoch": 9.732739420935411,
      "grad_norm": 17.79773712158203,
      "learning_rate": 1e-06,
      "loss": 0.5996,
      "num_input_tokens_seen": 244765972,
      "step": 4370
    },
    {
      "epoch": 9.732739420935411,
      "loss": 0.5670474171638489,
      "loss_ce": 0.00015290510782506317,
      "loss_iou": 0.244140625,
      "loss_num": 0.0157470703125,
      "loss_xval": 0.56640625,
      "num_input_tokens_seen": 244765972,
      "step": 4370
    },
    {
      "epoch": 9.734966592427616,
      "grad_norm": 16.366491317749023,
      "learning_rate": 1e-06,
      "loss": 0.6912,
      "num_input_tokens_seen": 244822468,
      "step": 4371
    },
    {
      "epoch": 9.734966592427616,
      "loss": 0.6842033863067627,
      "loss_ce": 0.0001213748037116602,
      "loss_iou": 0.3046875,
      "loss_num": 0.01470947265625,
      "loss_xval": 0.68359375,
      "num_input_tokens_seen": 244822468,
      "step": 4371
    },
    {
      "epoch": 9.737193763919821,
      "grad_norm": 21.454408645629883,
      "learning_rate": 1e-06,
      "loss": 0.4793,
      "num_input_tokens_seen": 244880364,
      "step": 4372
    },
    {
      "epoch": 9.737193763919821,
      "loss": 0.38819241523742676,
      "loss_ce": 0.00013089305139146745,
      "loss_iou": 0.177734375,
      "loss_num": 0.006561279296875,
      "loss_xval": 0.388671875,
      "num_input_tokens_seen": 244880364,
      "step": 4372
    },
    {
      "epoch": 9.739420935412026,
      "grad_norm": 52.516292572021484,
      "learning_rate": 1e-06,
      "loss": 0.4779,
      "num_input_tokens_seen": 244935344,
      "step": 4373
    },
    {
      "epoch": 9.739420935412026,
      "loss": 0.4206976294517517,
      "loss_ce": 0.0007757844286970794,
      "loss_iou": 0.162109375,
      "loss_num": 0.0194091796875,
      "loss_xval": 0.419921875,
      "num_input_tokens_seen": 244935344,
      "step": 4373
    },
    {
      "epoch": 9.74164810690423,
      "grad_norm": 12.701075553894043,
      "learning_rate": 1e-06,
      "loss": 0.5218,
      "num_input_tokens_seen": 244991972,
      "step": 4374
    },
    {
      "epoch": 9.74164810690423,
      "loss": 0.381612628698349,
      "loss_ce": 0.00014290204853750765,
      "loss_iou": 0.17578125,
      "loss_num": 0.006072998046875,
      "loss_xval": 0.380859375,
      "num_input_tokens_seen": 244991972,
      "step": 4374
    },
    {
      "epoch": 9.743875278396436,
      "grad_norm": 20.51826286315918,
      "learning_rate": 1e-06,
      "loss": 0.5475,
      "num_input_tokens_seen": 245048956,
      "step": 4375
    },
    {
      "epoch": 9.743875278396436,
      "loss": 0.5812922120094299,
      "loss_ce": 0.00029854726744815707,
      "loss_iou": 0.248046875,
      "loss_num": 0.01708984375,
      "loss_xval": 0.58203125,
      "num_input_tokens_seen": 245048956,
      "step": 4375
    },
    {
      "epoch": 9.74610244988864,
      "grad_norm": 19.916961669921875,
      "learning_rate": 1e-06,
      "loss": 0.5741,
      "num_input_tokens_seen": 245104028,
      "step": 4376
    },
    {
      "epoch": 9.74610244988864,
      "loss": 0.4663507342338562,
      "loss_ce": 0.0007135132909752429,
      "loss_iou": 0.1845703125,
      "loss_num": 0.01904296875,
      "loss_xval": 0.46484375,
      "num_input_tokens_seen": 245104028,
      "step": 4376
    },
    {
      "epoch": 9.748329621380847,
      "grad_norm": 18.798843383789062,
      "learning_rate": 1e-06,
      "loss": 0.7112,
      "num_input_tokens_seen": 245161592,
      "step": 4377
    },
    {
      "epoch": 9.748329621380847,
      "loss": 0.5074524283409119,
      "loss_ce": 0.00012819524272345006,
      "loss_iou": 0.2255859375,
      "loss_num": 0.01129150390625,
      "loss_xval": 0.5078125,
      "num_input_tokens_seen": 245161592,
      "step": 4377
    },
    {
      "epoch": 9.750556792873052,
      "grad_norm": 14.465723991394043,
      "learning_rate": 1e-06,
      "loss": 0.4476,
      "num_input_tokens_seen": 245218120,
      "step": 4378
    },
    {
      "epoch": 9.750556792873052,
      "loss": 0.5026047825813293,
      "loss_ce": 0.00016336789121851325,
      "loss_iou": 0.19921875,
      "loss_num": 0.020751953125,
      "loss_xval": 0.50390625,
      "num_input_tokens_seen": 245218120,
      "step": 4378
    },
    {
      "epoch": 9.752783964365257,
      "grad_norm": 148.81324768066406,
      "learning_rate": 1e-06,
      "loss": 0.4971,
      "num_input_tokens_seen": 245276528,
      "step": 4379
    },
    {
      "epoch": 9.752783964365257,
      "loss": 0.5717120170593262,
      "loss_ce": 0.00017881978419609368,
      "loss_iou": 0.255859375,
      "loss_num": 0.01214599609375,
      "loss_xval": 0.5703125,
      "num_input_tokens_seen": 245276528,
      "step": 4379
    },
    {
      "epoch": 9.755011135857462,
      "grad_norm": 22.322999954223633,
      "learning_rate": 1e-06,
      "loss": 0.8127,
      "num_input_tokens_seen": 245332196,
      "step": 4380
    },
    {
      "epoch": 9.755011135857462,
      "loss": 0.8225026726722717,
      "loss_ce": 0.0002370491565670818,
      "loss_iou": 0.359375,
      "loss_num": 0.0211181640625,
      "loss_xval": 0.8203125,
      "num_input_tokens_seen": 245332196,
      "step": 4380
    },
    {
      "epoch": 9.757238307349667,
      "grad_norm": 15.783463478088379,
      "learning_rate": 1e-06,
      "loss": 0.5557,
      "num_input_tokens_seen": 245386960,
      "step": 4381
    },
    {
      "epoch": 9.757238307349667,
      "loss": 0.670534610748291,
      "loss_ce": 0.0001244655722985044,
      "loss_iou": 0.2890625,
      "loss_num": 0.018798828125,
      "loss_xval": 0.671875,
      "num_input_tokens_seen": 245386960,
      "step": 4381
    },
    {
      "epoch": 9.759465478841872,
      "grad_norm": 16.41698455810547,
      "learning_rate": 1e-06,
      "loss": 0.3972,
      "num_input_tokens_seen": 245439152,
      "step": 4382
    },
    {
      "epoch": 9.759465478841872,
      "loss": 0.38251131772994995,
      "loss_ce": 0.00012604420771822333,
      "loss_iou": 0.166015625,
      "loss_num": 0.0098876953125,
      "loss_xval": 0.3828125,
      "num_input_tokens_seen": 245439152,
      "step": 4382
    },
    {
      "epoch": 9.761692650334076,
      "grad_norm": 21.44931983947754,
      "learning_rate": 1e-06,
      "loss": 0.6739,
      "num_input_tokens_seen": 245498164,
      "step": 4383
    },
    {
      "epoch": 9.761692650334076,
      "loss": 0.76970374584198,
      "loss_ce": 0.0001725103793432936,
      "loss_iou": 0.3203125,
      "loss_num": 0.0255126953125,
      "loss_xval": 0.76953125,
      "num_input_tokens_seen": 245498164,
      "step": 4383
    },
    {
      "epoch": 9.763919821826281,
      "grad_norm": 19.587390899658203,
      "learning_rate": 1e-06,
      "loss": 0.4272,
      "num_input_tokens_seen": 245552244,
      "step": 4384
    },
    {
      "epoch": 9.763919821826281,
      "loss": 0.4954010248184204,
      "loss_ce": 0.0001617551752133295,
      "loss_iou": 0.21484375,
      "loss_num": 0.0130615234375,
      "loss_xval": 0.49609375,
      "num_input_tokens_seen": 245552244,
      "step": 4384
    },
    {
      "epoch": 9.766146993318486,
      "grad_norm": 16.423864364624023,
      "learning_rate": 1e-06,
      "loss": 0.5937,
      "num_input_tokens_seen": 245607924,
      "step": 4385
    },
    {
      "epoch": 9.766146993318486,
      "loss": 0.659552276134491,
      "loss_ce": 0.0001284128229599446,
      "loss_iou": 0.2734375,
      "loss_num": 0.0224609375,
      "loss_xval": 0.66015625,
      "num_input_tokens_seen": 245607924,
      "step": 4385
    },
    {
      "epoch": 9.768374164810691,
      "grad_norm": 17.927560806274414,
      "learning_rate": 1e-06,
      "loss": 0.5699,
      "num_input_tokens_seen": 245662504,
      "step": 4386
    },
    {
      "epoch": 9.768374164810691,
      "loss": 0.5620256662368774,
      "loss_ce": 0.00013603617844637483,
      "loss_iou": 0.2314453125,
      "loss_num": 0.01953125,
      "loss_xval": 0.5625,
      "num_input_tokens_seen": 245662504,
      "step": 4386
    },
    {
      "epoch": 9.770601336302896,
      "grad_norm": 21.027870178222656,
      "learning_rate": 1e-06,
      "loss": 0.6231,
      "num_input_tokens_seen": 245719168,
      "step": 4387
    },
    {
      "epoch": 9.770601336302896,
      "loss": 0.7491781711578369,
      "loss_ce": 0.00015475715918000787,
      "loss_iou": 0.310546875,
      "loss_num": 0.025634765625,
      "loss_xval": 0.75,
      "num_input_tokens_seen": 245719168,
      "step": 4387
    },
    {
      "epoch": 9.7728285077951,
      "grad_norm": 209.35565185546875,
      "learning_rate": 1e-06,
      "loss": 0.7288,
      "num_input_tokens_seen": 245773520,
      "step": 4388
    },
    {
      "epoch": 9.7728285077951,
      "loss": 0.900350034236908,
      "loss_ce": 0.00020351973944343626,
      "loss_iou": 0.34375,
      "loss_num": 0.042236328125,
      "loss_xval": 0.8984375,
      "num_input_tokens_seen": 245773520,
      "step": 4388
    },
    {
      "epoch": 9.775055679287306,
      "grad_norm": 15.992069244384766,
      "learning_rate": 1e-06,
      "loss": 0.6249,
      "num_input_tokens_seen": 245829068,
      "step": 4389
    },
    {
      "epoch": 9.775055679287306,
      "loss": 0.4481305480003357,
      "loss_ce": 0.0002545609895605594,
      "loss_iou": 0.1943359375,
      "loss_num": 0.01177978515625,
      "loss_xval": 0.447265625,
      "num_input_tokens_seen": 245829068,
      "step": 4389
    },
    {
      "epoch": 9.77728285077951,
      "grad_norm": 65.46475219726562,
      "learning_rate": 1e-06,
      "loss": 0.5634,
      "num_input_tokens_seen": 245885128,
      "step": 4390
    },
    {
      "epoch": 9.77728285077951,
      "loss": 0.5616711378097534,
      "loss_ce": 0.00014769005065318197,
      "loss_iou": 0.2421875,
      "loss_num": 0.01519775390625,
      "loss_xval": 0.5625,
      "num_input_tokens_seen": 245885128,
      "step": 4390
    },
    {
      "epoch": 9.779510022271715,
      "grad_norm": 23.950044631958008,
      "learning_rate": 1e-06,
      "loss": 0.5287,
      "num_input_tokens_seen": 245939168,
      "step": 4391
    },
    {
      "epoch": 9.779510022271715,
      "loss": 0.6016286611557007,
      "loss_ce": 0.0003102673217654228,
      "loss_iou": 0.2470703125,
      "loss_num": 0.021484375,
      "loss_xval": 0.6015625,
      "num_input_tokens_seen": 245939168,
      "step": 4391
    },
    {
      "epoch": 9.78173719376392,
      "grad_norm": 13.293421745300293,
      "learning_rate": 1e-06,
      "loss": 0.4483,
      "num_input_tokens_seen": 245996116,
      "step": 4392
    },
    {
      "epoch": 9.78173719376392,
      "loss": 0.439850389957428,
      "loss_ce": 0.00015313076437450945,
      "loss_iou": 0.1806640625,
      "loss_num": 0.015869140625,
      "loss_xval": 0.439453125,
      "num_input_tokens_seen": 245996116,
      "step": 4392
    },
    {
      "epoch": 9.783964365256125,
      "grad_norm": 27.452177047729492,
      "learning_rate": 1e-06,
      "loss": 0.5466,
      "num_input_tokens_seen": 246051320,
      "step": 4393
    },
    {
      "epoch": 9.783964365256125,
      "loss": 0.5230746269226074,
      "loss_ce": 0.00012541524483822286,
      "loss_iou": 0.2373046875,
      "loss_num": 0.00958251953125,
      "loss_xval": 0.5234375,
      "num_input_tokens_seen": 246051320,
      "step": 4393
    },
    {
      "epoch": 9.78619153674833,
      "grad_norm": 18.695438385009766,
      "learning_rate": 1e-06,
      "loss": 0.6525,
      "num_input_tokens_seen": 246107564,
      "step": 4394
    },
    {
      "epoch": 9.78619153674833,
      "loss": 0.6207006573677063,
      "loss_ce": 0.0002172729728044942,
      "loss_iou": 0.263671875,
      "loss_num": 0.0186767578125,
      "loss_xval": 0.62109375,
      "num_input_tokens_seen": 246107564,
      "step": 4394
    },
    {
      "epoch": 9.788418708240535,
      "grad_norm": 16.555150985717773,
      "learning_rate": 1e-06,
      "loss": 0.6315,
      "num_input_tokens_seen": 246166380,
      "step": 4395
    },
    {
      "epoch": 9.788418708240535,
      "loss": 0.6279211044311523,
      "loss_ce": 0.00017448890139348805,
      "loss_iou": 0.2421875,
      "loss_num": 0.02880859375,
      "loss_xval": 0.62890625,
      "num_input_tokens_seen": 246166380,
      "step": 4395
    },
    {
      "epoch": 9.79064587973274,
      "grad_norm": 13.74927043914795,
      "learning_rate": 1e-06,
      "loss": 0.7028,
      "num_input_tokens_seen": 246222136,
      "step": 4396
    },
    {
      "epoch": 9.79064587973274,
      "loss": 0.8214117288589478,
      "loss_ce": 0.00036682139034383,
      "loss_iou": 0.3359375,
      "loss_num": 0.0301513671875,
      "loss_xval": 0.8203125,
      "num_input_tokens_seen": 246222136,
      "step": 4396
    },
    {
      "epoch": 9.792873051224944,
      "grad_norm": 17.71005630493164,
      "learning_rate": 1e-06,
      "loss": 0.5901,
      "num_input_tokens_seen": 246278428,
      "step": 4397
    },
    {
      "epoch": 9.792873051224944,
      "loss": 0.41447368264198303,
      "loss_ce": 0.00016704711015336215,
      "loss_iou": 0.1884765625,
      "loss_num": 0.007537841796875,
      "loss_xval": 0.4140625,
      "num_input_tokens_seen": 246278428,
      "step": 4397
    },
    {
      "epoch": 9.79510022271715,
      "grad_norm": 13.60590934753418,
      "learning_rate": 1e-06,
      "loss": 0.5328,
      "num_input_tokens_seen": 246335200,
      "step": 4398
    },
    {
      "epoch": 9.79510022271715,
      "loss": 0.4858427345752716,
      "loss_ce": 0.00012495677219703794,
      "loss_iou": 0.212890625,
      "loss_num": 0.0118408203125,
      "loss_xval": 0.486328125,
      "num_input_tokens_seen": 246335200,
      "step": 4398
    },
    {
      "epoch": 9.797327394209354,
      "grad_norm": 12.569108009338379,
      "learning_rate": 1e-06,
      "loss": 0.7097,
      "num_input_tokens_seen": 246391172,
      "step": 4399
    },
    {
      "epoch": 9.797327394209354,
      "loss": 0.8952298164367676,
      "loss_ce": 0.00021028138871770352,
      "loss_iou": 0.337890625,
      "loss_num": 0.043701171875,
      "loss_xval": 0.89453125,
      "num_input_tokens_seen": 246391172,
      "step": 4399
    },
    {
      "epoch": 9.799554565701559,
      "grad_norm": 16.045135498046875,
      "learning_rate": 1e-06,
      "loss": 0.5855,
      "num_input_tokens_seen": 246448824,
      "step": 4400
    },
    {
      "epoch": 9.799554565701559,
      "loss": 0.7127890586853027,
      "loss_ce": 0.00014254784036893398,
      "loss_iou": 0.306640625,
      "loss_num": 0.0201416015625,
      "loss_xval": 0.7109375,
      "num_input_tokens_seen": 246448824,
      "step": 4400
    },
    {
      "epoch": 9.801781737193764,
      "grad_norm": 18.482440948486328,
      "learning_rate": 1e-06,
      "loss": 0.5164,
      "num_input_tokens_seen": 246505864,
      "step": 4401
    },
    {
      "epoch": 9.801781737193764,
      "loss": 0.5470033884048462,
      "loss_ce": 0.00012841983698308468,
      "loss_iou": 0.2392578125,
      "loss_num": 0.01385498046875,
      "loss_xval": 0.546875,
      "num_input_tokens_seen": 246505864,
      "step": 4401
    },
    {
      "epoch": 9.804008908685969,
      "grad_norm": 17.286788940429688,
      "learning_rate": 1e-06,
      "loss": 0.6259,
      "num_input_tokens_seen": 246562828,
      "step": 4402
    },
    {
      "epoch": 9.804008908685969,
      "loss": 0.5885273218154907,
      "loss_ce": 0.00014838032075203955,
      "loss_iou": 0.2578125,
      "loss_num": 0.0146484375,
      "loss_xval": 0.58984375,
      "num_input_tokens_seen": 246562828,
      "step": 4402
    },
    {
      "epoch": 9.806236080178174,
      "grad_norm": 14.533248901367188,
      "learning_rate": 1e-06,
      "loss": 0.4501,
      "num_input_tokens_seen": 246619276,
      "step": 4403
    },
    {
      "epoch": 9.806236080178174,
      "loss": 0.4673030972480774,
      "loss_ce": 0.00014000001829117537,
      "loss_iou": 0.2080078125,
      "loss_num": 0.01025390625,
      "loss_xval": 0.466796875,
      "num_input_tokens_seen": 246619276,
      "step": 4403
    },
    {
      "epoch": 9.808463251670378,
      "grad_norm": 16.228418350219727,
      "learning_rate": 1e-06,
      "loss": 0.4324,
      "num_input_tokens_seen": 246678448,
      "step": 4404
    },
    {
      "epoch": 9.808463251670378,
      "loss": 0.47206440567970276,
      "loss_ce": 0.0001405745279043913,
      "loss_iou": 0.2109375,
      "loss_num": 0.0098876953125,
      "loss_xval": 0.47265625,
      "num_input_tokens_seen": 246678448,
      "step": 4404
    },
    {
      "epoch": 9.810690423162583,
      "grad_norm": 18.946840286254883,
      "learning_rate": 1e-06,
      "loss": 0.5187,
      "num_input_tokens_seen": 246732924,
      "step": 4405
    },
    {
      "epoch": 9.810690423162583,
      "loss": 0.5357993841171265,
      "loss_ce": 0.00015488323697354645,
      "loss_iou": 0.2294921875,
      "loss_num": 0.015380859375,
      "loss_xval": 0.53515625,
      "num_input_tokens_seen": 246732924,
      "step": 4405
    },
    {
      "epoch": 9.812917594654788,
      "grad_norm": 21.7833194732666,
      "learning_rate": 1e-06,
      "loss": 0.6752,
      "num_input_tokens_seen": 246790636,
      "step": 4406
    },
    {
      "epoch": 9.812917594654788,
      "loss": 0.5842469334602356,
      "loss_ce": 0.00014047868899069726,
      "loss_iou": 0.251953125,
      "loss_num": 0.01556396484375,
      "loss_xval": 0.5859375,
      "num_input_tokens_seen": 246790636,
      "step": 4406
    },
    {
      "epoch": 9.815144766146993,
      "grad_norm": 18.096515655517578,
      "learning_rate": 1e-06,
      "loss": 0.5097,
      "num_input_tokens_seen": 246847708,
      "step": 4407
    },
    {
      "epoch": 9.815144766146993,
      "loss": 0.634331226348877,
      "loss_ce": 0.0002980299177579582,
      "loss_iou": 0.263671875,
      "loss_num": 0.021484375,
      "loss_xval": 0.6328125,
      "num_input_tokens_seen": 246847708,
      "step": 4407
    },
    {
      "epoch": 9.817371937639198,
      "grad_norm": 21.39216423034668,
      "learning_rate": 1e-06,
      "loss": 0.6078,
      "num_input_tokens_seen": 246902376,
      "step": 4408
    },
    {
      "epoch": 9.817371937639198,
      "loss": 0.6875587701797485,
      "loss_ce": 0.00018085587362293154,
      "loss_iou": 0.283203125,
      "loss_num": 0.0244140625,
      "loss_xval": 0.6875,
      "num_input_tokens_seen": 246902376,
      "step": 4408
    },
    {
      "epoch": 9.819599109131403,
      "grad_norm": 29.385353088378906,
      "learning_rate": 1e-06,
      "loss": 0.6988,
      "num_input_tokens_seen": 246956228,
      "step": 4409
    },
    {
      "epoch": 9.819599109131403,
      "loss": 0.6361332535743713,
      "loss_ce": 0.00014688669762108475,
      "loss_iou": 0.28515625,
      "loss_num": 0.01318359375,
      "loss_xval": 0.63671875,
      "num_input_tokens_seen": 246956228,
      "step": 4409
    },
    {
      "epoch": 9.821826280623608,
      "grad_norm": 23.78947639465332,
      "learning_rate": 1e-06,
      "loss": 0.5749,
      "num_input_tokens_seen": 247012124,
      "step": 4410
    },
    {
      "epoch": 9.821826280623608,
      "loss": 0.4960480332374573,
      "loss_ce": 0.0001984003756660968,
      "loss_iou": 0.2197265625,
      "loss_num": 0.01129150390625,
      "loss_xval": 0.49609375,
      "num_input_tokens_seen": 247012124,
      "step": 4410
    },
    {
      "epoch": 9.824053452115812,
      "grad_norm": 24.32109260559082,
      "learning_rate": 1e-06,
      "loss": 0.6939,
      "num_input_tokens_seen": 247070408,
      "step": 4411
    },
    {
      "epoch": 9.824053452115812,
      "loss": 1.043853521347046,
      "loss_ce": 0.00015239304048009217,
      "loss_iou": 0.439453125,
      "loss_num": 0.032958984375,
      "loss_xval": 1.046875,
      "num_input_tokens_seen": 247070408,
      "step": 4411
    },
    {
      "epoch": 9.826280623608017,
      "grad_norm": 13.603236198425293,
      "learning_rate": 1e-06,
      "loss": 0.7308,
      "num_input_tokens_seen": 247130296,
      "step": 4412
    },
    {
      "epoch": 9.826280623608017,
      "loss": 0.8429387807846069,
      "loss_ce": 0.00016536489420104772,
      "loss_iou": 0.369140625,
      "loss_num": 0.0206298828125,
      "loss_xval": 0.84375,
      "num_input_tokens_seen": 247130296,
      "step": 4412
    },
    {
      "epoch": 9.828507795100222,
      "grad_norm": 23.041610717773438,
      "learning_rate": 1e-06,
      "loss": 0.4969,
      "num_input_tokens_seen": 247186440,
      "step": 4413
    },
    {
      "epoch": 9.828507795100222,
      "loss": 0.5199660062789917,
      "loss_ce": 0.0001905907120089978,
      "loss_iou": 0.2236328125,
      "loss_num": 0.0145263671875,
      "loss_xval": 0.51953125,
      "num_input_tokens_seen": 247186440,
      "step": 4413
    },
    {
      "epoch": 9.830734966592427,
      "grad_norm": 18.076107025146484,
      "learning_rate": 1e-06,
      "loss": 0.4157,
      "num_input_tokens_seen": 247239284,
      "step": 4414
    },
    {
      "epoch": 9.830734966592427,
      "loss": 0.3731992244720459,
      "loss_ce": 0.0001523523242212832,
      "loss_iou": 0.173828125,
      "loss_num": 0.005157470703125,
      "loss_xval": 0.373046875,
      "num_input_tokens_seen": 247239284,
      "step": 4414
    },
    {
      "epoch": 9.832962138084632,
      "grad_norm": 13.739655494689941,
      "learning_rate": 1e-06,
      "loss": 0.4094,
      "num_input_tokens_seen": 247295360,
      "step": 4415
    },
    {
      "epoch": 9.832962138084632,
      "loss": 0.3968695104122162,
      "loss_ce": 0.00014100654516369104,
      "loss_iou": 0.1796875,
      "loss_num": 0.007293701171875,
      "loss_xval": 0.396484375,
      "num_input_tokens_seen": 247295360,
      "step": 4415
    },
    {
      "epoch": 9.835189309576837,
      "grad_norm": 22.17353057861328,
      "learning_rate": 1e-06,
      "loss": 0.4095,
      "num_input_tokens_seen": 247353432,
      "step": 4416
    },
    {
      "epoch": 9.835189309576837,
      "loss": 0.38090550899505615,
      "loss_ce": 0.00016820061136968434,
      "loss_iou": 0.1748046875,
      "loss_num": 0.00616455078125,
      "loss_xval": 0.380859375,
      "num_input_tokens_seen": 247353432,
      "step": 4416
    },
    {
      "epoch": 9.837416481069042,
      "grad_norm": 28.909465789794922,
      "learning_rate": 1e-06,
      "loss": 0.5909,
      "num_input_tokens_seen": 247412760,
      "step": 4417
    },
    {
      "epoch": 9.837416481069042,
      "loss": 0.5692502856254578,
      "loss_ce": 0.00015849701594561338,
      "loss_iou": 0.251953125,
      "loss_num": 0.01300048828125,
      "loss_xval": 0.5703125,
      "num_input_tokens_seen": 247412760,
      "step": 4417
    },
    {
      "epoch": 9.839643652561247,
      "grad_norm": 18.436254501342773,
      "learning_rate": 1e-06,
      "loss": 0.7414,
      "num_input_tokens_seen": 247469036,
      "step": 4418
    },
    {
      "epoch": 9.839643652561247,
      "loss": 0.946731686592102,
      "loss_ce": 0.00019852191326208413,
      "loss_iou": 0.38671875,
      "loss_num": 0.0341796875,
      "loss_xval": 0.9453125,
      "num_input_tokens_seen": 247469036,
      "step": 4418
    },
    {
      "epoch": 9.841870824053451,
      "grad_norm": 17.95979881286621,
      "learning_rate": 1e-06,
      "loss": 0.5232,
      "num_input_tokens_seen": 247525680,
      "step": 4419
    },
    {
      "epoch": 9.841870824053451,
      "loss": 0.5357891917228699,
      "loss_ce": 0.00014465830463450402,
      "loss_iou": 0.228515625,
      "loss_num": 0.015869140625,
      "loss_xval": 0.53515625,
      "num_input_tokens_seen": 247525680,
      "step": 4419
    },
    {
      "epoch": 9.844097995545656,
      "grad_norm": 29.744184494018555,
      "learning_rate": 1e-06,
      "loss": 0.5656,
      "num_input_tokens_seen": 247580136,
      "step": 4420
    },
    {
      "epoch": 9.844097995545656,
      "loss": 0.6758188605308533,
      "loss_ce": 0.00015968517982400954,
      "loss_iou": 0.291015625,
      "loss_num": 0.0184326171875,
      "loss_xval": 0.67578125,
      "num_input_tokens_seen": 247580136,
      "step": 4420
    },
    {
      "epoch": 9.846325167037861,
      "grad_norm": 13.804282188415527,
      "learning_rate": 1e-06,
      "loss": 0.4793,
      "num_input_tokens_seen": 247634072,
      "step": 4421
    },
    {
      "epoch": 9.846325167037861,
      "loss": 0.34506338834762573,
      "loss_ce": 0.00033683725632727146,
      "loss_iou": 0.1484375,
      "loss_num": 0.0093994140625,
      "loss_xval": 0.34375,
      "num_input_tokens_seen": 247634072,
      "step": 4421
    },
    {
      "epoch": 9.848552338530066,
      "grad_norm": 53.691707611083984,
      "learning_rate": 1e-06,
      "loss": 0.5384,
      "num_input_tokens_seen": 247688788,
      "step": 4422
    },
    {
      "epoch": 9.848552338530066,
      "loss": 0.5472594499588013,
      "loss_ce": 0.0001402852067258209,
      "loss_iou": 0.24609375,
      "loss_num": 0.01116943359375,
      "loss_xval": 0.546875,
      "num_input_tokens_seen": 247688788,
      "step": 4422
    },
    {
      "epoch": 9.85077951002227,
      "grad_norm": 37.15926742553711,
      "learning_rate": 1e-06,
      "loss": 0.6395,
      "num_input_tokens_seen": 247746856,
      "step": 4423
    },
    {
      "epoch": 9.85077951002227,
      "loss": 0.6605561375617981,
      "loss_ce": 0.00015572294068988413,
      "loss_iou": 0.287109375,
      "loss_num": 0.016845703125,
      "loss_xval": 0.66015625,
      "num_input_tokens_seen": 247746856,
      "step": 4423
    },
    {
      "epoch": 9.853006681514476,
      "grad_norm": 16.28897476196289,
      "learning_rate": 1e-06,
      "loss": 0.7326,
      "num_input_tokens_seen": 247802588,
      "step": 4424
    },
    {
      "epoch": 9.853006681514476,
      "loss": 0.605131208896637,
      "loss_ce": 0.00015073509712237865,
      "loss_iou": 0.26953125,
      "loss_num": 0.012939453125,
      "loss_xval": 0.60546875,
      "num_input_tokens_seen": 247802588,
      "step": 4424
    },
    {
      "epoch": 9.855233853006682,
      "grad_norm": 24.916080474853516,
      "learning_rate": 1e-06,
      "loss": 0.5261,
      "num_input_tokens_seen": 247858736,
      "step": 4425
    },
    {
      "epoch": 9.855233853006682,
      "loss": 0.5880270004272461,
      "loss_ce": 0.0006245865952223539,
      "loss_iou": 0.267578125,
      "loss_num": 0.010498046875,
      "loss_xval": 0.5859375,
      "num_input_tokens_seen": 247858736,
      "step": 4425
    },
    {
      "epoch": 9.857461024498887,
      "grad_norm": 13.907861709594727,
      "learning_rate": 1e-06,
      "loss": 0.4909,
      "num_input_tokens_seen": 247916764,
      "step": 4426
    },
    {
      "epoch": 9.857461024498887,
      "loss": 0.5015993118286133,
      "loss_ce": 0.00013441478949971497,
      "loss_iou": 0.21875,
      "loss_num": 0.012939453125,
      "loss_xval": 0.5,
      "num_input_tokens_seen": 247916764,
      "step": 4426
    },
    {
      "epoch": 9.859688195991092,
      "grad_norm": 31.806222915649414,
      "learning_rate": 1e-06,
      "loss": 0.5157,
      "num_input_tokens_seen": 247971644,
      "step": 4427
    },
    {
      "epoch": 9.859688195991092,
      "loss": 0.5441058278083801,
      "loss_ce": 0.00016053869330789894,
      "loss_iou": 0.244140625,
      "loss_num": 0.01123046875,
      "loss_xval": 0.54296875,
      "num_input_tokens_seen": 247971644,
      "step": 4427
    },
    {
      "epoch": 9.861915367483297,
      "grad_norm": 13.702202796936035,
      "learning_rate": 1e-06,
      "loss": 0.4027,
      "num_input_tokens_seen": 248027780,
      "step": 4428
    },
    {
      "epoch": 9.861915367483297,
      "loss": 0.3753761351108551,
      "loss_ce": 0.00013199455861467868,
      "loss_iou": 0.1640625,
      "loss_num": 0.00946044921875,
      "loss_xval": 0.375,
      "num_input_tokens_seen": 248027780,
      "step": 4428
    },
    {
      "epoch": 9.864142538975502,
      "grad_norm": 22.021825790405273,
      "learning_rate": 1e-06,
      "loss": 0.369,
      "num_input_tokens_seen": 248085456,
      "step": 4429
    },
    {
      "epoch": 9.864142538975502,
      "loss": 0.34767791628837585,
      "loss_ce": 0.00014372303849086165,
      "loss_iou": 0.1572265625,
      "loss_num": 0.0064697265625,
      "loss_xval": 0.34765625,
      "num_input_tokens_seen": 248085456,
      "step": 4429
    },
    {
      "epoch": 9.866369710467707,
      "grad_norm": 15.745802879333496,
      "learning_rate": 1e-06,
      "loss": 0.4453,
      "num_input_tokens_seen": 248141132,
      "step": 4430
    },
    {
      "epoch": 9.866369710467707,
      "loss": 0.3866683840751648,
      "loss_ce": 0.0001937640190590173,
      "loss_iou": 0.15625,
      "loss_num": 0.01470947265625,
      "loss_xval": 0.38671875,
      "num_input_tokens_seen": 248141132,
      "step": 4430
    },
    {
      "epoch": 9.868596881959911,
      "grad_norm": 17.44371795654297,
      "learning_rate": 1e-06,
      "loss": 0.6266,
      "num_input_tokens_seen": 248199536,
      "step": 4431
    },
    {
      "epoch": 9.868596881959911,
      "loss": 0.4967302978038788,
      "loss_ce": 0.0001482525112805888,
      "loss_iou": 0.224609375,
      "loss_num": 0.0096435546875,
      "loss_xval": 0.49609375,
      "num_input_tokens_seen": 248199536,
      "step": 4431
    },
    {
      "epoch": 9.870824053452116,
      "grad_norm": 20.9477481842041,
      "learning_rate": 1e-06,
      "loss": 0.5961,
      "num_input_tokens_seen": 248254300,
      "step": 4432
    },
    {
      "epoch": 9.870824053452116,
      "loss": 0.5253418684005737,
      "loss_ce": 0.00019535439787432551,
      "loss_iou": 0.23828125,
      "loss_num": 0.009521484375,
      "loss_xval": 0.5234375,
      "num_input_tokens_seen": 248254300,
      "step": 4432
    },
    {
      "epoch": 9.873051224944321,
      "grad_norm": 32.99359893798828,
      "learning_rate": 1e-06,
      "loss": 0.6711,
      "num_input_tokens_seen": 248311172,
      "step": 4433
    },
    {
      "epoch": 9.873051224944321,
      "loss": 0.6815510988235474,
      "loss_ce": 0.00015456396795343608,
      "loss_iou": 0.294921875,
      "loss_num": 0.0186767578125,
      "loss_xval": 0.6796875,
      "num_input_tokens_seen": 248311172,
      "step": 4433
    },
    {
      "epoch": 9.875278396436526,
      "grad_norm": 31.431236267089844,
      "learning_rate": 1e-06,
      "loss": 0.4702,
      "num_input_tokens_seen": 248366780,
      "step": 4434
    },
    {
      "epoch": 9.875278396436526,
      "loss": 0.4480833113193512,
      "loss_ce": 0.0004514938627835363,
      "loss_iou": 0.19921875,
      "loss_num": 0.00982666015625,
      "loss_xval": 0.447265625,
      "num_input_tokens_seen": 248366780,
      "step": 4434
    },
    {
      "epoch": 9.877505567928731,
      "grad_norm": 14.6542329788208,
      "learning_rate": 1e-06,
      "loss": 0.6543,
      "num_input_tokens_seen": 248422596,
      "step": 4435
    },
    {
      "epoch": 9.877505567928731,
      "loss": 0.7738858461380005,
      "loss_ce": 0.00020423725072760135,
      "loss_iou": 0.3046875,
      "loss_num": 0.032958984375,
      "loss_xval": 0.7734375,
      "num_input_tokens_seen": 248422596,
      "step": 4435
    },
    {
      "epoch": 9.879732739420936,
      "grad_norm": 21.185422897338867,
      "learning_rate": 1e-06,
      "loss": 0.7841,
      "num_input_tokens_seen": 248477628,
      "step": 4436
    },
    {
      "epoch": 9.879732739420936,
      "loss": 0.7956980466842651,
      "loss_ce": 0.00016583640535827726,
      "loss_iou": 0.330078125,
      "loss_num": 0.026611328125,
      "loss_xval": 0.796875,
      "num_input_tokens_seen": 248477628,
      "step": 4436
    },
    {
      "epoch": 9.88195991091314,
      "grad_norm": 22.979459762573242,
      "learning_rate": 1e-06,
      "loss": 0.746,
      "num_input_tokens_seen": 248531252,
      "step": 4437
    },
    {
      "epoch": 9.88195991091314,
      "loss": 1.0274933576583862,
      "loss_ce": 0.00014961442502681166,
      "loss_iou": 0.4375,
      "loss_num": 0.030517578125,
      "loss_xval": 1.03125,
      "num_input_tokens_seen": 248531252,
      "step": 4437
    },
    {
      "epoch": 9.884187082405345,
      "grad_norm": 13.26347541809082,
      "learning_rate": 1e-06,
      "loss": 0.789,
      "num_input_tokens_seen": 248587712,
      "step": 4438
    },
    {
      "epoch": 9.884187082405345,
      "loss": 0.8224426507949829,
      "loss_ce": 0.0001770372618921101,
      "loss_iou": 0.34765625,
      "loss_num": 0.025390625,
      "loss_xval": 0.8203125,
      "num_input_tokens_seen": 248587712,
      "step": 4438
    },
    {
      "epoch": 9.88641425389755,
      "grad_norm": 19.138172149658203,
      "learning_rate": 1e-06,
      "loss": 0.6224,
      "num_input_tokens_seen": 248644168,
      "step": 4439
    },
    {
      "epoch": 9.88641425389755,
      "loss": 0.6742221117019653,
      "loss_ce": 0.00014987046597525477,
      "loss_iou": 0.2734375,
      "loss_num": 0.025146484375,
      "loss_xval": 0.67578125,
      "num_input_tokens_seen": 248644168,
      "step": 4439
    },
    {
      "epoch": 9.888641425389755,
      "grad_norm": 14.5908784866333,
      "learning_rate": 1e-06,
      "loss": 0.448,
      "num_input_tokens_seen": 248700288,
      "step": 4440
    },
    {
      "epoch": 9.888641425389755,
      "loss": 0.5669944286346436,
      "loss_ce": 0.00022195381461642683,
      "loss_iou": 0.2333984375,
      "loss_num": 0.020263671875,
      "loss_xval": 0.56640625,
      "num_input_tokens_seen": 248700288,
      "step": 4440
    },
    {
      "epoch": 9.89086859688196,
      "grad_norm": 15.528139114379883,
      "learning_rate": 1e-06,
      "loss": 0.5742,
      "num_input_tokens_seen": 248756128,
      "step": 4441
    },
    {
      "epoch": 9.89086859688196,
      "loss": 0.6929147243499756,
      "loss_ce": 0.00016564424731768668,
      "loss_iou": 0.283203125,
      "loss_num": 0.0255126953125,
      "loss_xval": 0.69140625,
      "num_input_tokens_seen": 248756128,
      "step": 4441
    },
    {
      "epoch": 9.893095768374165,
      "grad_norm": 25.997331619262695,
      "learning_rate": 1e-06,
      "loss": 0.4781,
      "num_input_tokens_seen": 248812652,
      "step": 4442
    },
    {
      "epoch": 9.893095768374165,
      "loss": 0.4952731132507324,
      "loss_ce": 0.0001559244265081361,
      "loss_iou": 0.23046875,
      "loss_num": 0.0069580078125,
      "loss_xval": 0.49609375,
      "num_input_tokens_seen": 248812652,
      "step": 4442
    },
    {
      "epoch": 9.89532293986637,
      "grad_norm": 17.757108688354492,
      "learning_rate": 1e-06,
      "loss": 0.457,
      "num_input_tokens_seen": 248870328,
      "step": 4443
    },
    {
      "epoch": 9.89532293986637,
      "loss": 0.5199582576751709,
      "loss_ce": 0.0003048997314181179,
      "loss_iou": 0.2294921875,
      "loss_num": 0.01220703125,
      "loss_xval": 0.51953125,
      "num_input_tokens_seen": 248870328,
      "step": 4443
    },
    {
      "epoch": 9.897550111358575,
      "grad_norm": 33.18953323364258,
      "learning_rate": 1e-06,
      "loss": 0.4442,
      "num_input_tokens_seen": 248927148,
      "step": 4444
    },
    {
      "epoch": 9.897550111358575,
      "loss": 0.4624471068382263,
      "loss_ce": 0.00016684165166225284,
      "loss_iou": 0.2060546875,
      "loss_num": 0.01007080078125,
      "loss_xval": 0.462890625,
      "num_input_tokens_seen": 248927148,
      "step": 4444
    },
    {
      "epoch": 9.89977728285078,
      "grad_norm": 16.396703720092773,
      "learning_rate": 1e-06,
      "loss": 0.695,
      "num_input_tokens_seen": 248982876,
      "step": 4445
    },
    {
      "epoch": 9.89977728285078,
      "loss": 0.759984016418457,
      "loss_ce": 0.0002184216573368758,
      "loss_iou": 0.32421875,
      "loss_num": 0.022705078125,
      "loss_xval": 0.7578125,
      "num_input_tokens_seen": 248982876,
      "step": 4445
    },
    {
      "epoch": 9.902004454342984,
      "grad_norm": 355.2792053222656,
      "learning_rate": 1e-06,
      "loss": 0.6844,
      "num_input_tokens_seen": 249036336,
      "step": 4446
    },
    {
      "epoch": 9.902004454342984,
      "loss": 0.8203428983688354,
      "loss_ce": 0.0007628132589161396,
      "loss_iou": 0.373046875,
      "loss_num": 0.01483154296875,
      "loss_xval": 0.8203125,
      "num_input_tokens_seen": 249036336,
      "step": 4446
    },
    {
      "epoch": 9.90423162583519,
      "grad_norm": 20.815412521362305,
      "learning_rate": 1e-06,
      "loss": 0.493,
      "num_input_tokens_seen": 249090512,
      "step": 4447
    },
    {
      "epoch": 9.90423162583519,
      "loss": 0.3990846276283264,
      "loss_ce": 0.00015884230379015207,
      "loss_iou": 0.173828125,
      "loss_num": 0.0103759765625,
      "loss_xval": 0.3984375,
      "num_input_tokens_seen": 249090512,
      "step": 4447
    },
    {
      "epoch": 9.906458797327394,
      "grad_norm": 22.289730072021484,
      "learning_rate": 1e-06,
      "loss": 0.5334,
      "num_input_tokens_seen": 249146280,
      "step": 4448
    },
    {
      "epoch": 9.906458797327394,
      "loss": 0.46716466546058655,
      "loss_ce": 0.00012365993461571634,
      "loss_iou": 0.205078125,
      "loss_num": 0.01153564453125,
      "loss_xval": 0.466796875,
      "num_input_tokens_seen": 249146280,
      "step": 4448
    },
    {
      "epoch": 9.908685968819599,
      "grad_norm": 18.984031677246094,
      "learning_rate": 1e-06,
      "loss": 0.482,
      "num_input_tokens_seen": 249199220,
      "step": 4449
    },
    {
      "epoch": 9.908685968819599,
      "loss": 0.3772040605545044,
      "loss_ce": 0.00012887499178759754,
      "loss_iou": 0.1669921875,
      "loss_num": 0.00848388671875,
      "loss_xval": 0.376953125,
      "num_input_tokens_seen": 249199220,
      "step": 4449
    },
    {
      "epoch": 9.910913140311804,
      "grad_norm": 22.167558670043945,
      "learning_rate": 1e-06,
      "loss": 0.3869,
      "num_input_tokens_seen": 249255288,
      "step": 4450
    },
    {
      "epoch": 9.910913140311804,
      "loss": 0.42713233828544617,
      "loss_ce": 0.0001304029137827456,
      "loss_iou": 0.1865234375,
      "loss_num": 0.01068115234375,
      "loss_xval": 0.427734375,
      "num_input_tokens_seen": 249255288,
      "step": 4450
    },
    {
      "epoch": 9.913140311804009,
      "grad_norm": 16.50093650817871,
      "learning_rate": 1e-06,
      "loss": 0.4574,
      "num_input_tokens_seen": 249309848,
      "step": 4451
    },
    {
      "epoch": 9.913140311804009,
      "loss": 0.3489049971103668,
      "loss_ce": 0.0001501307706348598,
      "loss_iou": 0.1484375,
      "loss_num": 0.0101318359375,
      "loss_xval": 0.349609375,
      "num_input_tokens_seen": 249309848,
      "step": 4451
    },
    {
      "epoch": 9.915367483296214,
      "grad_norm": 26.222118377685547,
      "learning_rate": 1e-06,
      "loss": 0.6307,
      "num_input_tokens_seen": 249365492,
      "step": 4452
    },
    {
      "epoch": 9.915367483296214,
      "loss": 0.5098949670791626,
      "loss_ce": 0.00012939177395310253,
      "loss_iou": 0.2177734375,
      "loss_num": 0.01513671875,
      "loss_xval": 0.5078125,
      "num_input_tokens_seen": 249365492,
      "step": 4452
    },
    {
      "epoch": 9.917594654788418,
      "grad_norm": 26.15787124633789,
      "learning_rate": 1e-06,
      "loss": 0.4824,
      "num_input_tokens_seen": 249420888,
      "step": 4453
    },
    {
      "epoch": 9.917594654788418,
      "loss": 0.34827083349227905,
      "loss_ce": 0.0001263211015611887,
      "loss_iou": 0.1533203125,
      "loss_num": 0.008056640625,
      "loss_xval": 0.34765625,
      "num_input_tokens_seen": 249420888,
      "step": 4453
    },
    {
      "epoch": 9.919821826280623,
      "grad_norm": 21.763919830322266,
      "learning_rate": 1e-06,
      "loss": 0.5638,
      "num_input_tokens_seen": 249474144,
      "step": 4454
    },
    {
      "epoch": 9.919821826280623,
      "loss": 0.48050248622894287,
      "loss_ce": 0.00015578939928673208,
      "loss_iou": 0.2177734375,
      "loss_num": 0.0089111328125,
      "loss_xval": 0.48046875,
      "num_input_tokens_seen": 249474144,
      "step": 4454
    },
    {
      "epoch": 9.922048997772828,
      "grad_norm": 22.63077163696289,
      "learning_rate": 1e-06,
      "loss": 0.5656,
      "num_input_tokens_seen": 249532508,
      "step": 4455
    },
    {
      "epoch": 9.922048997772828,
      "loss": 0.4286133050918579,
      "loss_ce": 0.00014652467507403344,
      "loss_iou": 0.1943359375,
      "loss_num": 0.008056640625,
      "loss_xval": 0.427734375,
      "num_input_tokens_seen": 249532508,
      "step": 4455
    },
    {
      "epoch": 9.924276169265033,
      "grad_norm": 19.111759185791016,
      "learning_rate": 1e-06,
      "loss": 0.4905,
      "num_input_tokens_seen": 249588256,
      "step": 4456
    },
    {
      "epoch": 9.924276169265033,
      "loss": 0.6744104623794556,
      "loss_ce": 0.0005213359836488962,
      "loss_iou": 0.279296875,
      "loss_num": 0.022705078125,
      "loss_xval": 0.67578125,
      "num_input_tokens_seen": 249588256,
      "step": 4456
    },
    {
      "epoch": 9.926503340757238,
      "grad_norm": 14.64742660522461,
      "learning_rate": 1e-06,
      "loss": 0.5547,
      "num_input_tokens_seen": 249645916,
      "step": 4457
    },
    {
      "epoch": 9.926503340757238,
      "loss": 0.6124250888824463,
      "loss_ce": 0.00036456523230299354,
      "loss_iou": 0.2578125,
      "loss_num": 0.018798828125,
      "loss_xval": 0.61328125,
      "num_input_tokens_seen": 249645916,
      "step": 4457
    },
    {
      "epoch": 9.928730512249443,
      "grad_norm": 30.41196060180664,
      "learning_rate": 1e-06,
      "loss": 0.6046,
      "num_input_tokens_seen": 249699704,
      "step": 4458
    },
    {
      "epoch": 9.928730512249443,
      "loss": 0.43876737356185913,
      "loss_ce": 0.00016875306027941406,
      "loss_iou": 0.2041015625,
      "loss_num": 0.0059814453125,
      "loss_xval": 0.439453125,
      "num_input_tokens_seen": 249699704,
      "step": 4458
    },
    {
      "epoch": 9.930957683741648,
      "grad_norm": 19.998111724853516,
      "learning_rate": 1e-06,
      "loss": 0.5264,
      "num_input_tokens_seen": 249754572,
      "step": 4459
    },
    {
      "epoch": 9.930957683741648,
      "loss": 0.41116297245025635,
      "loss_ce": 0.0001522265374660492,
      "loss_iou": 0.1806640625,
      "loss_num": 0.01007080078125,
      "loss_xval": 0.41015625,
      "num_input_tokens_seen": 249754572,
      "step": 4459
    },
    {
      "epoch": 9.933184855233852,
      "grad_norm": 33.828853607177734,
      "learning_rate": 1e-06,
      "loss": 0.7319,
      "num_input_tokens_seen": 249809712,
      "step": 4460
    },
    {
      "epoch": 9.933184855233852,
      "loss": 0.640634298324585,
      "loss_ce": 0.0002534303639549762,
      "loss_iou": 0.279296875,
      "loss_num": 0.01611328125,
      "loss_xval": 0.640625,
      "num_input_tokens_seen": 249809712,
      "step": 4460
    },
    {
      "epoch": 9.935412026726057,
      "grad_norm": 540.546875,
      "learning_rate": 1e-06,
      "loss": 0.5882,
      "num_input_tokens_seen": 249864220,
      "step": 4461
    },
    {
      "epoch": 9.935412026726057,
      "loss": 0.4697587788105011,
      "loss_ce": 0.00012375140795484185,
      "loss_iou": 0.19921875,
      "loss_num": 0.01422119140625,
      "loss_xval": 0.46875,
      "num_input_tokens_seen": 249864220,
      "step": 4461
    },
    {
      "epoch": 9.937639198218262,
      "grad_norm": 36.97883987426758,
      "learning_rate": 1e-06,
      "loss": 0.6369,
      "num_input_tokens_seen": 249917428,
      "step": 4462
    },
    {
      "epoch": 9.937639198218262,
      "loss": 0.5367482900619507,
      "loss_ce": 0.00012720399536192417,
      "loss_iou": 0.2216796875,
      "loss_num": 0.018798828125,
      "loss_xval": 0.53515625,
      "num_input_tokens_seen": 249917428,
      "step": 4462
    },
    {
      "epoch": 9.939866369710467,
      "grad_norm": 16.344486236572266,
      "learning_rate": 1e-06,
      "loss": 0.5017,
      "num_input_tokens_seen": 249975260,
      "step": 4463
    },
    {
      "epoch": 9.939866369710467,
      "loss": 0.4674408435821533,
      "loss_ce": 0.0001557128271088004,
      "loss_iou": 0.193359375,
      "loss_num": 0.0162353515625,
      "loss_xval": 0.466796875,
      "num_input_tokens_seen": 249975260,
      "step": 4463
    },
    {
      "epoch": 9.942093541202672,
      "grad_norm": 20.06435203552246,
      "learning_rate": 1e-06,
      "loss": 0.4822,
      "num_input_tokens_seen": 250028748,
      "step": 4464
    },
    {
      "epoch": 9.942093541202672,
      "loss": 0.562027096748352,
      "loss_ce": 0.00013746539480052888,
      "loss_iou": 0.2421875,
      "loss_num": 0.0152587890625,
      "loss_xval": 0.5625,
      "num_input_tokens_seen": 250028748,
      "step": 4464
    },
    {
      "epoch": 9.944320712694877,
      "grad_norm": 21.986183166503906,
      "learning_rate": 1e-06,
      "loss": 0.6695,
      "num_input_tokens_seen": 250086348,
      "step": 4465
    },
    {
      "epoch": 9.944320712694877,
      "loss": 0.6947346925735474,
      "loss_ce": 0.00015459363930858672,
      "loss_iou": 0.28515625,
      "loss_num": 0.0245361328125,
      "loss_xval": 0.6953125,
      "num_input_tokens_seen": 250086348,
      "step": 4465
    },
    {
      "epoch": 9.946547884187082,
      "grad_norm": 22.66680335998535,
      "learning_rate": 1e-06,
      "loss": 0.5404,
      "num_input_tokens_seen": 250140012,
      "step": 4466
    },
    {
      "epoch": 9.946547884187082,
      "loss": 0.5031921863555908,
      "loss_ce": 0.00014042215480003506,
      "loss_iou": 0.22265625,
      "loss_num": 0.01165771484375,
      "loss_xval": 0.50390625,
      "num_input_tokens_seen": 250140012,
      "step": 4466
    },
    {
      "epoch": 9.948775055679288,
      "grad_norm": 20.20932388305664,
      "learning_rate": 1e-06,
      "loss": 0.5721,
      "num_input_tokens_seen": 250197260,
      "step": 4467
    },
    {
      "epoch": 9.948775055679288,
      "loss": 0.5846377611160278,
      "loss_ce": 0.0001650652993703261,
      "loss_iou": 0.25,
      "loss_num": 0.016845703125,
      "loss_xval": 0.5859375,
      "num_input_tokens_seen": 250197260,
      "step": 4467
    },
    {
      "epoch": 9.951002227171493,
      "grad_norm": 13.735246658325195,
      "learning_rate": 1e-06,
      "loss": 0.5371,
      "num_input_tokens_seen": 250255264,
      "step": 4468
    },
    {
      "epoch": 9.951002227171493,
      "loss": 0.36416295170783997,
      "loss_ce": 0.0001492602750658989,
      "loss_iou": 0.16015625,
      "loss_num": 0.00872802734375,
      "loss_xval": 0.36328125,
      "num_input_tokens_seen": 250255264,
      "step": 4468
    },
    {
      "epoch": 9.953229398663698,
      "grad_norm": 31.190519332885742,
      "learning_rate": 1e-06,
      "loss": 0.6512,
      "num_input_tokens_seen": 250309688,
      "step": 4469
    },
    {
      "epoch": 9.953229398663698,
      "loss": 0.8156118392944336,
      "loss_ce": 0.00018213686416856945,
      "loss_iou": 0.35546875,
      "loss_num": 0.0208740234375,
      "loss_xval": 0.81640625,
      "num_input_tokens_seen": 250309688,
      "step": 4469
    },
    {
      "epoch": 9.955456570155903,
      "grad_norm": 20.936838150024414,
      "learning_rate": 1e-06,
      "loss": 0.6122,
      "num_input_tokens_seen": 250364732,
      "step": 4470
    },
    {
      "epoch": 9.955456570155903,
      "loss": 0.6811445951461792,
      "loss_ce": 0.0004805437056347728,
      "loss_iou": 0.28515625,
      "loss_num": 0.0220947265625,
      "loss_xval": 0.6796875,
      "num_input_tokens_seen": 250364732,
      "step": 4470
    },
    {
      "epoch": 9.957683741648108,
      "grad_norm": 20.30494499206543,
      "learning_rate": 1e-06,
      "loss": 0.5026,
      "num_input_tokens_seen": 250421248,
      "step": 4471
    },
    {
      "epoch": 9.957683741648108,
      "loss": 0.4135947823524475,
      "loss_ce": 0.00014261619071476161,
      "loss_iou": 0.1865234375,
      "loss_num": 0.00787353515625,
      "loss_xval": 0.4140625,
      "num_input_tokens_seen": 250421248,
      "step": 4471
    },
    {
      "epoch": 9.959910913140313,
      "grad_norm": 22.17571449279785,
      "learning_rate": 1e-06,
      "loss": 0.5305,
      "num_input_tokens_seen": 250476212,
      "step": 4472
    },
    {
      "epoch": 9.959910913140313,
      "loss": 0.5278682112693787,
      "loss_ce": 0.0001582720287842676,
      "loss_iou": 0.2275390625,
      "loss_num": 0.01446533203125,
      "loss_xval": 0.52734375,
      "num_input_tokens_seen": 250476212,
      "step": 4472
    },
    {
      "epoch": 9.962138084632517,
      "grad_norm": 21.48920249938965,
      "learning_rate": 1e-06,
      "loss": 0.4718,
      "num_input_tokens_seen": 250534156,
      "step": 4473
    },
    {
      "epoch": 9.962138084632517,
      "loss": 0.5655474662780762,
      "loss_ce": 0.0002398924989392981,
      "loss_iou": 0.265625,
      "loss_num": 0.00677490234375,
      "loss_xval": 0.56640625,
      "num_input_tokens_seen": 250534156,
      "step": 4473
    },
    {
      "epoch": 9.964365256124722,
      "grad_norm": 21.838937759399414,
      "learning_rate": 1e-06,
      "loss": 0.3754,
      "num_input_tokens_seen": 250590784,
      "step": 4474
    },
    {
      "epoch": 9.964365256124722,
      "loss": 0.44572579860687256,
      "loss_ce": 0.00016913507715798914,
      "loss_iou": 0.1875,
      "loss_num": 0.0142822265625,
      "loss_xval": 0.4453125,
      "num_input_tokens_seen": 250590784,
      "step": 4474
    },
    {
      "epoch": 9.966592427616927,
      "grad_norm": 15.84533977508545,
      "learning_rate": 1e-06,
      "loss": 0.5513,
      "num_input_tokens_seen": 250649576,
      "step": 4475
    },
    {
      "epoch": 9.966592427616927,
      "loss": 0.4788917899131775,
      "loss_ce": 0.0001320300216320902,
      "loss_iou": 0.203125,
      "loss_num": 0.0147705078125,
      "loss_xval": 0.478515625,
      "num_input_tokens_seen": 250649576,
      "step": 4475
    },
    {
      "epoch": 9.968819599109132,
      "grad_norm": 15.656432151794434,
      "learning_rate": 1e-06,
      "loss": 0.4874,
      "num_input_tokens_seen": 250705124,
      "step": 4476
    },
    {
      "epoch": 9.968819599109132,
      "loss": 0.49739590287208557,
      "loss_ce": 0.00014248676598072052,
      "loss_iou": 0.1904296875,
      "loss_num": 0.023193359375,
      "loss_xval": 0.498046875,
      "num_input_tokens_seen": 250705124,
      "step": 4476
    },
    {
      "epoch": 9.971046770601337,
      "grad_norm": 26.737167358398438,
      "learning_rate": 1e-06,
      "loss": 0.6391,
      "num_input_tokens_seen": 250761396,
      "step": 4477
    },
    {
      "epoch": 9.971046770601337,
      "loss": 0.6163727045059204,
      "loss_ce": 0.00016177864745259285,
      "loss_iou": 0.26171875,
      "loss_num": 0.0186767578125,
      "loss_xval": 0.6171875,
      "num_input_tokens_seen": 250761396,
      "step": 4477
    },
    {
      "epoch": 9.973273942093542,
      "grad_norm": 22.787172317504883,
      "learning_rate": 1e-06,
      "loss": 0.6781,
      "num_input_tokens_seen": 250815480,
      "step": 4478
    },
    {
      "epoch": 9.973273942093542,
      "loss": 0.8530862331390381,
      "loss_ce": 0.00018101731257047504,
      "loss_iou": 0.349609375,
      "loss_num": 0.0306396484375,
      "loss_xval": 0.8515625,
      "num_input_tokens_seen": 250815480,
      "step": 4478
    },
    {
      "epoch": 9.975501113585747,
      "grad_norm": 15.702953338623047,
      "learning_rate": 1e-06,
      "loss": 0.4059,
      "num_input_tokens_seen": 250872060,
      "step": 4479
    },
    {
      "epoch": 9.975501113585747,
      "loss": 0.3851672112941742,
      "loss_ce": 0.0001574572379468009,
      "loss_iou": 0.1640625,
      "loss_num": 0.01141357421875,
      "loss_xval": 0.384765625,
      "num_input_tokens_seen": 250872060,
      "step": 4479
    },
    {
      "epoch": 9.977728285077951,
      "grad_norm": 25.042709350585938,
      "learning_rate": 1e-06,
      "loss": 0.5131,
      "num_input_tokens_seen": 250929860,
      "step": 4480
    },
    {
      "epoch": 9.977728285077951,
      "loss": 0.5049949884414673,
      "loss_ce": 0.00011219277803320438,
      "loss_iou": 0.2275390625,
      "loss_num": 0.01019287109375,
      "loss_xval": 0.50390625,
      "num_input_tokens_seen": 250929860,
      "step": 4480
    },
    {
      "epoch": 9.979955456570156,
      "grad_norm": 15.756171226501465,
      "learning_rate": 1e-06,
      "loss": 0.5156,
      "num_input_tokens_seen": 250984516,
      "step": 4481
    },
    {
      "epoch": 9.979955456570156,
      "loss": 0.5473989248275757,
      "loss_ce": 0.00015769051969982684,
      "loss_iou": 0.251953125,
      "loss_num": 0.0084228515625,
      "loss_xval": 0.546875,
      "num_input_tokens_seen": 250984516,
      "step": 4481
    },
    {
      "epoch": 9.982182628062361,
      "grad_norm": 16.539581298828125,
      "learning_rate": 1e-06,
      "loss": 0.6737,
      "num_input_tokens_seen": 251041288,
      "step": 4482
    },
    {
      "epoch": 9.982182628062361,
      "loss": 0.6556758880615234,
      "loss_ce": 0.00015832131612114608,
      "loss_iou": 0.291015625,
      "loss_num": 0.01422119140625,
      "loss_xval": 0.65625,
      "num_input_tokens_seen": 251041288,
      "step": 4482
    },
    {
      "epoch": 9.984409799554566,
      "grad_norm": 19.558635711669922,
      "learning_rate": 1e-06,
      "loss": 0.5597,
      "num_input_tokens_seen": 251098176,
      "step": 4483
    },
    {
      "epoch": 9.984409799554566,
      "loss": 0.43492138385772705,
      "loss_ce": 0.00022901118791196495,
      "loss_iou": 0.197265625,
      "loss_num": 0.0079345703125,
      "loss_xval": 0.435546875,
      "num_input_tokens_seen": 251098176,
      "step": 4483
    },
    {
      "epoch": 9.98663697104677,
      "grad_norm": 14.561528205871582,
      "learning_rate": 1e-06,
      "loss": 0.5418,
      "num_input_tokens_seen": 251154472,
      "step": 4484
    },
    {
      "epoch": 9.98663697104677,
      "loss": 0.41586917638778687,
      "loss_ce": 0.0001434582518413663,
      "loss_iou": 0.181640625,
      "loss_num": 0.010498046875,
      "loss_xval": 0.416015625,
      "num_input_tokens_seen": 251154472,
      "step": 4484
    },
    {
      "epoch": 9.988864142538976,
      "grad_norm": 17.365453720092773,
      "learning_rate": 1e-06,
      "loss": 0.5246,
      "num_input_tokens_seen": 251211700,
      "step": 4485
    },
    {
      "epoch": 9.988864142538976,
      "loss": 0.41728872060775757,
      "loss_ce": 0.0001744481414789334,
      "loss_iou": 0.1884765625,
      "loss_num": 0.00811767578125,
      "loss_xval": 0.41796875,
      "num_input_tokens_seen": 251211700,
      "step": 4485
    },
    {
      "epoch": 9.99109131403118,
      "grad_norm": 21.09174346923828,
      "learning_rate": 1e-06,
      "loss": 0.586,
      "num_input_tokens_seen": 251266340,
      "step": 4486
    },
    {
      "epoch": 9.99109131403118,
      "loss": 0.6495558619499207,
      "loss_ce": 0.000141800323035568,
      "loss_iou": 0.27734375,
      "loss_num": 0.01904296875,
      "loss_xval": 0.6484375,
      "num_input_tokens_seen": 251266340,
      "step": 4486
    },
    {
      "epoch": 9.993318485523385,
      "grad_norm": 13.637633323669434,
      "learning_rate": 1e-06,
      "loss": 0.5915,
      "num_input_tokens_seen": 251323956,
      "step": 4487
    },
    {
      "epoch": 9.993318485523385,
      "loss": 0.6502304077148438,
      "loss_ce": 0.00020597720867954195,
      "loss_iou": 0.255859375,
      "loss_num": 0.0274658203125,
      "loss_xval": 0.6484375,
      "num_input_tokens_seen": 251323956,
      "step": 4487
    },
    {
      "epoch": 9.99554565701559,
      "grad_norm": 21.853805541992188,
      "learning_rate": 1e-06,
      "loss": 0.3252,
      "num_input_tokens_seen": 251383004,
      "step": 4488
    },
    {
      "epoch": 9.99554565701559,
      "loss": 0.4286557734012604,
      "loss_ce": 0.0001889891573227942,
      "loss_iou": 0.17578125,
      "loss_num": 0.0155029296875,
      "loss_xval": 0.427734375,
      "num_input_tokens_seen": 251383004,
      "step": 4488
    },
    {
      "epoch": 9.997772828507795,
      "grad_norm": 27.582048416137695,
      "learning_rate": 1e-06,
      "loss": 0.6094,
      "num_input_tokens_seen": 251442116,
      "step": 4489
    },
    {
      "epoch": 9.997772828507795,
      "loss": 0.6864469051361084,
      "loss_ce": 0.00016763756866566837,
      "loss_iou": 0.30078125,
      "loss_num": 0.016845703125,
      "loss_xval": 0.6875,
      "num_input_tokens_seen": 251442116,
      "step": 4489
    },
    {
      "epoch": 10.0,
      "grad_norm": 16.904457092285156,
      "learning_rate": 1e-06,
      "loss": 0.6229,
      "num_input_tokens_seen": 251498292,
      "step": 4490
    },
    {
      "epoch": 10.0,
      "loss": 0.6710529923439026,
      "loss_ce": 0.00015453548985533416,
      "loss_iou": 0.27734375,
      "loss_num": 0.0235595703125,
      "loss_xval": 0.671875,
      "num_input_tokens_seen": 251498292,
      "step": 4490
    },
    {
      "epoch": 10.002227171492205,
      "grad_norm": 24.753952026367188,
      "learning_rate": 1e-06,
      "loss": 0.5628,
      "num_input_tokens_seen": 251553736,
      "step": 4491
    },
    {
      "epoch": 10.002227171492205,
      "loss": 0.44396665692329407,
      "loss_ce": 0.00011899826495209709,
      "loss_iou": 0.1953125,
      "loss_num": 0.01055908203125,
      "loss_xval": 0.443359375,
      "num_input_tokens_seen": 251553736,
      "step": 4491
    },
    {
      "epoch": 10.00445434298441,
      "grad_norm": 23.54865074157715,
      "learning_rate": 1e-06,
      "loss": 0.6842,
      "num_input_tokens_seen": 251607144,
      "step": 4492
    },
    {
      "epoch": 10.00445434298441,
      "loss": 0.7414926290512085,
      "loss_ce": 0.00015957036521285772,
      "loss_iou": 0.298828125,
      "loss_num": 0.0284423828125,
      "loss_xval": 0.7421875,
      "num_input_tokens_seen": 251607144,
      "step": 4492
    },
    {
      "epoch": 10.006681514476615,
      "grad_norm": 17.976106643676758,
      "learning_rate": 1e-06,
      "loss": 0.5584,
      "num_input_tokens_seen": 251663048,
      "step": 4493
    },
    {
      "epoch": 10.006681514476615,
      "loss": 0.5196723937988281,
      "loss_ce": 0.0001411354896845296,
      "loss_iou": 0.201171875,
      "loss_num": 0.0234375,
      "loss_xval": 0.51953125,
      "num_input_tokens_seen": 251663048,
      "step": 4493
    },
    {
      "epoch": 10.00890868596882,
      "grad_norm": 14.608015060424805,
      "learning_rate": 1e-06,
      "loss": 0.4636,
      "num_input_tokens_seen": 251719488,
      "step": 4494
    },
    {
      "epoch": 10.00890868596882,
      "loss": 0.5168766975402832,
      "loss_ce": 0.00015306880231946707,
      "loss_iou": 0.2314453125,
      "loss_num": 0.01080322265625,
      "loss_xval": 0.515625,
      "num_input_tokens_seen": 251719488,
      "step": 4494
    },
    {
      "epoch": 10.011135857461024,
      "grad_norm": 18.117263793945312,
      "learning_rate": 1e-06,
      "loss": 0.6167,
      "num_input_tokens_seen": 251777652,
      "step": 4495
    },
    {
      "epoch": 10.011135857461024,
      "loss": 0.7012206315994263,
      "loss_ce": 0.0002929244365077466,
      "loss_iou": 0.283203125,
      "loss_num": 0.02685546875,
      "loss_xval": 0.69921875,
      "num_input_tokens_seen": 251777652,
      "step": 4495
    },
    {
      "epoch": 10.01336302895323,
      "grad_norm": 23.114547729492188,
      "learning_rate": 1e-06,
      "loss": 0.6466,
      "num_input_tokens_seen": 251834100,
      "step": 4496
    },
    {
      "epoch": 10.01336302895323,
      "loss": 0.4545919597148895,
      "loss_ce": 0.00012418595724739134,
      "loss_iou": 0.19921875,
      "loss_num": 0.010986328125,
      "loss_xval": 0.455078125,
      "num_input_tokens_seen": 251834100,
      "step": 4496
    },
    {
      "epoch": 10.015590200445434,
      "grad_norm": 35.5590705871582,
      "learning_rate": 1e-06,
      "loss": 0.5079,
      "num_input_tokens_seen": 251889480,
      "step": 4497
    },
    {
      "epoch": 10.015590200445434,
      "loss": 0.5074775218963623,
      "loss_ce": 0.00015331841132137924,
      "loss_iou": 0.220703125,
      "loss_num": 0.0133056640625,
      "loss_xval": 0.5078125,
      "num_input_tokens_seen": 251889480,
      "step": 4497
    },
    {
      "epoch": 10.017817371937639,
      "grad_norm": 35.230525970458984,
      "learning_rate": 1e-06,
      "loss": 0.4227,
      "num_input_tokens_seen": 251944744,
      "step": 4498
    },
    {
      "epoch": 10.017817371937639,
      "loss": 0.35149550437927246,
      "loss_ce": 0.00011612092202994972,
      "loss_iou": 0.1533203125,
      "loss_num": 0.0089111328125,
      "loss_xval": 0.3515625,
      "num_input_tokens_seen": 251944744,
      "step": 4498
    },
    {
      "epoch": 10.020044543429844,
      "grad_norm": 17.35042953491211,
      "learning_rate": 1e-06,
      "loss": 0.5238,
      "num_input_tokens_seen": 252001072,
      "step": 4499
    },
    {
      "epoch": 10.020044543429844,
      "loss": 0.5627619624137878,
      "loss_ce": 0.0002619452425278723,
      "loss_iou": 0.228515625,
      "loss_num": 0.02099609375,
      "loss_xval": 0.5625,
      "num_input_tokens_seen": 252001072,
      "step": 4499
    },
    {
      "epoch": 10.022271714922049,
      "grad_norm": 18.346031188964844,
      "learning_rate": 1e-06,
      "loss": 0.6645,
      "num_input_tokens_seen": 252054980,
      "step": 4500
    },
    {
      "epoch": 10.022271714922049,
      "eval_seeclick_web_CIoU": 0.5782820582389832,
      "eval_seeclick_web_GIoU": 0.5778241455554962,
      "eval_seeclick_web_IoU": 0.5961449146270752,
      "eval_seeclick_web_MAE_all": 0.016307780984789133,
      "eval_seeclick_web_MAE_h": 0.008026089053601027,
      "eval_seeclick_web_MAE_w": 0.01621978636831045,
      "eval_seeclick_web_MAE_x_boxes": 0.009718116372823715,
      "eval_seeclick_web_MAE_y_boxes": 0.022209799382835627,
      "eval_seeclick_web_inside_bbox": 0.9166666567325592,
      "eval_seeclick_web_loss": 0.9367778301239014,
      "eval_seeclick_web_loss_ce": 0.00020874126494163647,
      "eval_seeclick_web_loss_iou": 0.4254150390625,
      "eval_seeclick_web_loss_num": 0.013081550598144531,
      "eval_seeclick_web_loss_xval": 0.916259765625,
      "eval_seeclick_web_runtime": 18.3321,
      "eval_seeclick_web_samples_per_second": 2.727,
      "eval_seeclick_web_steps_per_second": 0.109,
      "num_input_tokens_seen": 252054980,
      "step": 4500
    },
    {
      "epoch": 10.022271714922049,
      "eval_icons_CIoU": 0.2818611115217209,
      "eval_icons_GIoU": 0.3074444532394409,
      "eval_icons_IoU": 0.3626803755760193,
      "eval_icons_MAE_all": 0.06376760825514793,
      "eval_icons_MAE_h": 0.038395908661186695,
      "eval_icons_MAE_w": 0.06800028681755066,
      "eval_icons_MAE_x_boxes": 0.059712398797273636,
      "eval_icons_MAE_y_boxes": 0.03969671577215195,
      "eval_icons_inside_bbox": 0.6336805522441864,
      "eval_icons_loss": 1.7186964750289917,
      "eval_icons_loss_ce": 0.00024977065913844854,
      "eval_icons_loss_iou": 0.6683349609375,
      "eval_icons_loss_num": 0.06123924255371094,
      "eval_icons_loss_xval": 1.6416015625,
      "eval_icons_runtime": 19.3941,
      "eval_icons_samples_per_second": 2.578,
      "eval_icons_steps_per_second": 0.103,
      "num_input_tokens_seen": 252054980,
      "step": 4500
    },
    {
      "epoch": 10.022271714922049,
      "eval_screenspot_CIoU": 0.34726441899935406,
      "eval_screenspot_GIoU": 0.36475805441538495,
      "eval_screenspot_IoU": 0.43036482731501263,
      "eval_screenspot_MAE_all": 0.06156049047907194,
      "eval_screenspot_MAE_h": 0.0375117938965559,
      "eval_screenspot_MAE_w": 0.07428983474771182,
      "eval_screenspot_MAE_x_boxes": 0.07907873081664245,
      "eval_screenspot_MAE_y_boxes": 0.04031235041717688,
      "eval_screenspot_inside_bbox": 0.6862499912579855,
      "eval_screenspot_loss": 1.6380661725997925,
      "eval_screenspot_loss_ce": 0.00026154937222599983,
      "eval_screenspot_loss_iou": 0.6749674479166666,
      "eval_screenspot_loss_num": 0.0705553690592448,
      "eval_screenspot_loss_xval": 1.7029622395833333,
      "eval_screenspot_runtime": 30.7953,
      "eval_screenspot_samples_per_second": 2.89,
      "eval_screenspot_steps_per_second": 0.097,
      "num_input_tokens_seen": 252054980,
      "step": 4500
    },
    {
      "epoch": 10.022271714922049,
      "eval_compot_CIoU": 0.35239382088184357,
      "eval_compot_GIoU": 0.36740949749946594,
      "eval_compot_IoU": 0.4099326878786087,
      "eval_compot_MAE_all": 0.01800244627520442,
      "eval_compot_MAE_h": 0.008867041673511267,
      "eval_compot_MAE_w": 0.02112545073032379,
      "eval_compot_MAE_x_boxes": 0.030241395346820354,
      "eval_compot_MAE_y_boxes": 0.0067769435700029135,
      "eval_compot_inside_bbox": 0.6302083432674408,
      "eval_compot_loss": 1.3512978553771973,
      "eval_compot_loss_ce": 0.0002027387818088755,
      "eval_compot_loss_iou": 0.6209716796875,
      "eval_compot_loss_num": 0.016811370849609375,
      "eval_compot_loss_xval": 1.325439453125,
      "eval_compot_runtime": 19.3355,
      "eval_compot_samples_per_second": 2.586,
      "eval_compot_steps_per_second": 0.103,
      "num_input_tokens_seen": 252054980,
      "step": 4500
    },
    {
      "epoch": 10.022271714922049,
      "eval_custom_ui_val_CIoU": 0.46736228962739307,
      "eval_custom_ui_val_GIoU": 0.48078184492058224,
      "eval_custom_ui_val_IoU": 0.5274362398518456,
      "eval_custom_ui_val_MAE_all": 0.03146956084916989,
      "eval_custom_ui_val_MAE_h": 0.016498709821866617,
      "eval_custom_ui_val_MAE_w": 0.0397563229004542,
      "eval_custom_ui_val_MAE_x_boxes": 0.039604716209901705,
      "eval_custom_ui_val_MAE_y_boxes": 0.015526494192373421,
      "eval_custom_ui_val_inside_bbox": 0.7527006202273898,
      "eval_custom_ui_val_loss": 1.2082252502441406,
      "eval_custom_ui_val_loss_ce": 0.00023262486178686636,
      "eval_custom_ui_val_loss_iou": 0.5123562282986112,
      "eval_custom_ui_val_loss_num": 0.02882491217719184,
      "eval_custom_ui_val_loss_xval": 1.1688096788194444,
      "eval_custom_ui_val_runtime": 55.4656,
      "eval_custom_ui_val_samples_per_second": 4.778,
      "eval_custom_ui_val_steps_per_second": 0.162,
      "num_input_tokens_seen": 252054980,
      "step": 4500
    },
    {
      "epoch": 10.022271714922049,
      "loss": 0.912562906742096,
      "loss_ce": 0.00020945594587828964,
      "loss_iou": 0.392578125,
      "loss_num": 0.0250244140625,
      "loss_xval": 0.9140625,
      "num_input_tokens_seen": 252054980,
      "step": 4500
    },
    {
      "epoch": 10.024498886414253,
      "grad_norm": 17.76951789855957,
      "learning_rate": 1e-06,
      "loss": 0.4816,
      "num_input_tokens_seen": 252110792,
      "step": 4501
    },
    {
      "epoch": 10.024498886414253,
      "loss": 0.40937352180480957,
      "loss_ce": 0.00019383057951927185,
      "loss_iou": 0.1826171875,
      "loss_num": 0.0087890625,
      "loss_xval": 0.41015625,
      "num_input_tokens_seen": 252110792,
      "step": 4501
    },
    {
      "epoch": 10.026726057906458,
      "grad_norm": 13.45481014251709,
      "learning_rate": 1e-06,
      "loss": 0.6639,
      "num_input_tokens_seen": 252168752,
      "step": 4502
    },
    {
      "epoch": 10.026726057906458,
      "loss": 0.7877389788627625,
      "loss_ce": 0.0001412917481502518,
      "loss_iou": 0.328125,
      "loss_num": 0.0264892578125,
      "loss_xval": 0.7890625,
      "num_input_tokens_seen": 252168752,
      "step": 4502
    },
    {
      "epoch": 10.028953229398663,
      "grad_norm": 14.474963188171387,
      "learning_rate": 1e-06,
      "loss": 0.5939,
      "num_input_tokens_seen": 252226176,
      "step": 4503
    },
    {
      "epoch": 10.028953229398663,
      "loss": 0.7263405323028564,
      "loss_ce": 0.00014424577238969505,
      "loss_iou": 0.279296875,
      "loss_num": 0.03369140625,
      "loss_xval": 0.7265625,
      "num_input_tokens_seen": 252226176,
      "step": 4503
    },
    {
      "epoch": 10.031180400890868,
      "grad_norm": 20.8179931640625,
      "learning_rate": 1e-06,
      "loss": 0.6298,
      "num_input_tokens_seen": 252285124,
      "step": 4504
    },
    {
      "epoch": 10.031180400890868,
      "loss": 0.5824363231658936,
      "loss_ce": 0.00016091388533823192,
      "loss_iou": 0.259765625,
      "loss_num": 0.01287841796875,
      "loss_xval": 0.58203125,
      "num_input_tokens_seen": 252285124,
      "step": 4504
    },
    {
      "epoch": 10.033407572383073,
      "grad_norm": 25.60984230041504,
      "learning_rate": 1e-06,
      "loss": 0.594,
      "num_input_tokens_seen": 252339596,
      "step": 4505
    },
    {
      "epoch": 10.033407572383073,
      "loss": 0.47377800941467285,
      "loss_ce": 0.00014518320676870644,
      "loss_iou": 0.203125,
      "loss_num": 0.01336669921875,
      "loss_xval": 0.47265625,
      "num_input_tokens_seen": 252339596,
      "step": 4505
    },
    {
      "epoch": 10.035634743875278,
      "grad_norm": 16.030750274658203,
      "learning_rate": 1e-06,
      "loss": 0.6331,
      "num_input_tokens_seen": 252395008,
      "step": 4506
    },
    {
      "epoch": 10.035634743875278,
      "loss": 0.6874623894691467,
      "loss_ce": 0.0001455222663935274,
      "loss_iou": 0.28515625,
      "loss_num": 0.0230712890625,
      "loss_xval": 0.6875,
      "num_input_tokens_seen": 252395008,
      "step": 4506
    },
    {
      "epoch": 10.037861915367483,
      "grad_norm": 24.892925262451172,
      "learning_rate": 1e-06,
      "loss": 0.6838,
      "num_input_tokens_seen": 252449524,
      "step": 4507
    },
    {
      "epoch": 10.037861915367483,
      "loss": 0.7407252788543701,
      "loss_ce": 0.00012467037595342845,
      "loss_iou": 0.33203125,
      "loss_num": 0.01507568359375,
      "loss_xval": 0.7421875,
      "num_input_tokens_seen": 252449524,
      "step": 4507
    },
    {
      "epoch": 10.040089086859687,
      "grad_norm": 20.39448356628418,
      "learning_rate": 1e-06,
      "loss": 0.4444,
      "num_input_tokens_seen": 252507044,
      "step": 4508
    },
    {
      "epoch": 10.040089086859687,
      "loss": 0.5464025139808655,
      "loss_ce": 0.0002599266008473933,
      "loss_iou": 0.2373046875,
      "loss_num": 0.01416015625,
      "loss_xval": 0.546875,
      "num_input_tokens_seen": 252507044,
      "step": 4508
    },
    {
      "epoch": 10.042316258351892,
      "grad_norm": 28.83409881591797,
      "learning_rate": 1e-06,
      "loss": 0.7085,
      "num_input_tokens_seen": 252561052,
      "step": 4509
    },
    {
      "epoch": 10.042316258351892,
      "loss": 0.6915697455406189,
      "loss_ce": 0.00016349003999494016,
      "loss_iou": 0.30859375,
      "loss_num": 0.0147705078125,
      "loss_xval": 0.69140625,
      "num_input_tokens_seen": 252561052,
      "step": 4509
    },
    {
      "epoch": 10.044543429844097,
      "grad_norm": 22.83576011657715,
      "learning_rate": 1e-06,
      "loss": 0.4975,
      "num_input_tokens_seen": 252613648,
      "step": 4510
    },
    {
      "epoch": 10.044543429844097,
      "loss": 0.3887307047843933,
      "loss_ce": 0.0005776156904175878,
      "loss_iou": 0.162109375,
      "loss_num": 0.012939453125,
      "loss_xval": 0.388671875,
      "num_input_tokens_seen": 252613648,
      "step": 4510
    },
    {
      "epoch": 10.046770601336302,
      "grad_norm": 20.531705856323242,
      "learning_rate": 1e-06,
      "loss": 0.6052,
      "num_input_tokens_seen": 252670492,
      "step": 4511
    },
    {
      "epoch": 10.046770601336302,
      "loss": 0.5842575430870056,
      "loss_ce": 0.00015114534471649677,
      "loss_iou": 0.25390625,
      "loss_num": 0.01507568359375,
      "loss_xval": 0.5859375,
      "num_input_tokens_seen": 252670492,
      "step": 4511
    },
    {
      "epoch": 10.048997772828507,
      "grad_norm": 19.739933013916016,
      "learning_rate": 1e-06,
      "loss": 0.9433,
      "num_input_tokens_seen": 252725212,
      "step": 4512
    },
    {
      "epoch": 10.048997772828507,
      "loss": 1.3185453414916992,
      "loss_ce": 0.0001860101765487343,
      "loss_iou": 0.53515625,
      "loss_num": 0.049072265625,
      "loss_xval": 1.3203125,
      "num_input_tokens_seen": 252725212,
      "step": 4512
    },
    {
      "epoch": 10.051224944320714,
      "grad_norm": 68.91185760498047,
      "learning_rate": 1e-06,
      "loss": 0.478,
      "num_input_tokens_seen": 252781356,
      "step": 4513
    },
    {
      "epoch": 10.051224944320714,
      "loss": 0.46619629859924316,
      "loss_ce": 0.00013182274415157735,
      "loss_iou": 0.1982421875,
      "loss_num": 0.01397705078125,
      "loss_xval": 0.466796875,
      "num_input_tokens_seen": 252781356,
      "step": 4513
    },
    {
      "epoch": 10.053452115812918,
      "grad_norm": 47.988834381103516,
      "learning_rate": 1e-06,
      "loss": 0.7151,
      "num_input_tokens_seen": 252835956,
      "step": 4514
    },
    {
      "epoch": 10.053452115812918,
      "loss": 0.7893850803375244,
      "loss_ce": 0.00032261922024190426,
      "loss_iou": 0.3359375,
      "loss_num": 0.023681640625,
      "loss_xval": 0.7890625,
      "num_input_tokens_seen": 252835956,
      "step": 4514
    },
    {
      "epoch": 10.055679287305123,
      "grad_norm": 13.075557708740234,
      "learning_rate": 1e-06,
      "loss": 0.5498,
      "num_input_tokens_seen": 252893772,
      "step": 4515
    },
    {
      "epoch": 10.055679287305123,
      "loss": 0.49821239709854126,
      "loss_ce": 0.00016552505258005112,
      "loss_iou": 0.2109375,
      "loss_num": 0.01507568359375,
      "loss_xval": 0.498046875,
      "num_input_tokens_seen": 252893772,
      "step": 4515
    },
    {
      "epoch": 10.057906458797328,
      "grad_norm": 31.43246078491211,
      "learning_rate": 1e-06,
      "loss": 0.5931,
      "num_input_tokens_seen": 252947940,
      "step": 4516
    },
    {
      "epoch": 10.057906458797328,
      "loss": 0.6125627756118774,
      "loss_ce": 0.0002580622094683349,
      "loss_iou": 0.279296875,
      "loss_num": 0.0106201171875,
      "loss_xval": 0.61328125,
      "num_input_tokens_seen": 252947940,
      "step": 4516
    },
    {
      "epoch": 10.060133630289533,
      "grad_norm": 27.630748748779297,
      "learning_rate": 1e-06,
      "loss": 0.6048,
      "num_input_tokens_seen": 253005340,
      "step": 4517
    },
    {
      "epoch": 10.060133630289533,
      "loss": 0.699859619140625,
      "loss_ce": 0.00015256297774612904,
      "loss_iou": 0.296875,
      "loss_num": 0.021484375,
      "loss_xval": 0.69921875,
      "num_input_tokens_seen": 253005340,
      "step": 4517
    },
    {
      "epoch": 10.062360801781738,
      "grad_norm": 23.77977180480957,
      "learning_rate": 1e-06,
      "loss": 0.5228,
      "num_input_tokens_seen": 253061712,
      "step": 4518
    },
    {
      "epoch": 10.062360801781738,
      "loss": 0.5870422124862671,
      "loss_ce": 0.000128203013446182,
      "loss_iou": 0.2451171875,
      "loss_num": 0.01904296875,
      "loss_xval": 0.5859375,
      "num_input_tokens_seen": 253061712,
      "step": 4518
    },
    {
      "epoch": 10.064587973273943,
      "grad_norm": 20.586177825927734,
      "learning_rate": 1e-06,
      "loss": 0.581,
      "num_input_tokens_seen": 253119552,
      "step": 4519
    },
    {
      "epoch": 10.064587973273943,
      "loss": 0.5409017205238342,
      "loss_ce": 0.000130253320094198,
      "loss_iou": 0.25390625,
      "loss_num": 0.00677490234375,
      "loss_xval": 0.5390625,
      "num_input_tokens_seen": 253119552,
      "step": 4519
    },
    {
      "epoch": 10.066815144766148,
      "grad_norm": 19.27606964111328,
      "learning_rate": 1e-06,
      "loss": 0.5846,
      "num_input_tokens_seen": 253173196,
      "step": 4520
    },
    {
      "epoch": 10.066815144766148,
      "loss": 0.5604423880577087,
      "loss_ce": 0.00013963712262921035,
      "loss_iou": 0.24609375,
      "loss_num": 0.01361083984375,
      "loss_xval": 0.55859375,
      "num_input_tokens_seen": 253173196,
      "step": 4520
    },
    {
      "epoch": 10.069042316258352,
      "grad_norm": 16.18434715270996,
      "learning_rate": 1e-06,
      "loss": 0.4942,
      "num_input_tokens_seen": 253230384,
      "step": 4521
    },
    {
      "epoch": 10.069042316258352,
      "loss": 0.5661949515342712,
      "loss_ce": 0.00015493093815166503,
      "loss_iou": 0.2421875,
      "loss_num": 0.0164794921875,
      "loss_xval": 0.56640625,
      "num_input_tokens_seen": 253230384,
      "step": 4521
    },
    {
      "epoch": 10.071269487750557,
      "grad_norm": 22.953781127929688,
      "learning_rate": 1e-06,
      "loss": 0.5444,
      "num_input_tokens_seen": 253286812,
      "step": 4522
    },
    {
      "epoch": 10.071269487750557,
      "loss": 0.5951232314109802,
      "loss_ce": 0.00015253589663188905,
      "loss_iou": 0.26953125,
      "loss_num": 0.011474609375,
      "loss_xval": 0.59375,
      "num_input_tokens_seen": 253286812,
      "step": 4522
    },
    {
      "epoch": 10.073496659242762,
      "grad_norm": 18.672510147094727,
      "learning_rate": 1e-06,
      "loss": 0.5558,
      "num_input_tokens_seen": 253343172,
      "step": 4523
    },
    {
      "epoch": 10.073496659242762,
      "loss": 0.5056633949279785,
      "loss_ce": 0.00017027268768288195,
      "loss_iou": 0.2177734375,
      "loss_num": 0.01409912109375,
      "loss_xval": 0.50390625,
      "num_input_tokens_seen": 253343172,
      "step": 4523
    },
    {
      "epoch": 10.075723830734967,
      "grad_norm": 17.472490310668945,
      "learning_rate": 1e-06,
      "loss": 0.6312,
      "num_input_tokens_seen": 253397752,
      "step": 4524
    },
    {
      "epoch": 10.075723830734967,
      "loss": 0.6256510019302368,
      "loss_ce": 0.0001627619785722345,
      "loss_iou": 0.287109375,
      "loss_num": 0.010009765625,
      "loss_xval": 0.625,
      "num_input_tokens_seen": 253397752,
      "step": 4524
    },
    {
      "epoch": 10.077951002227172,
      "grad_norm": 13.800442695617676,
      "learning_rate": 1e-06,
      "loss": 0.4006,
      "num_input_tokens_seen": 253454584,
      "step": 4525
    },
    {
      "epoch": 10.077951002227172,
      "loss": 0.4295790493488312,
      "loss_ce": 0.00013570842565968633,
      "loss_iou": 0.1826171875,
      "loss_num": 0.0130615234375,
      "loss_xval": 0.4296875,
      "num_input_tokens_seen": 253454584,
      "step": 4525
    },
    {
      "epoch": 10.080178173719377,
      "grad_norm": 83.25904846191406,
      "learning_rate": 1e-06,
      "loss": 0.5269,
      "num_input_tokens_seen": 253512564,
      "step": 4526
    },
    {
      "epoch": 10.080178173719377,
      "loss": 0.45021358132362366,
      "loss_ce": 0.00014032571925781667,
      "loss_iou": 0.2060546875,
      "loss_num": 0.007476806640625,
      "loss_xval": 0.44921875,
      "num_input_tokens_seen": 253512564,
      "step": 4526
    },
    {
      "epoch": 10.082405345211582,
      "grad_norm": 32.7124137878418,
      "learning_rate": 1e-06,
      "loss": 0.6008,
      "num_input_tokens_seen": 253569208,
      "step": 4527
    },
    {
      "epoch": 10.082405345211582,
      "loss": 0.5963279008865356,
      "loss_ce": 0.00013654123176820576,
      "loss_iou": 0.267578125,
      "loss_num": 0.01220703125,
      "loss_xval": 0.59765625,
      "num_input_tokens_seen": 253569208,
      "step": 4527
    },
    {
      "epoch": 10.084632516703786,
      "grad_norm": 18.75583839416504,
      "learning_rate": 1e-06,
      "loss": 0.436,
      "num_input_tokens_seen": 253625684,
      "step": 4528
    },
    {
      "epoch": 10.084632516703786,
      "loss": 0.42995163798332214,
      "loss_ce": 0.00014207106141839176,
      "loss_iou": 0.1767578125,
      "loss_num": 0.01519775390625,
      "loss_xval": 0.4296875,
      "num_input_tokens_seen": 253625684,
      "step": 4528
    },
    {
      "epoch": 10.086859688195991,
      "grad_norm": 55.7838020324707,
      "learning_rate": 1e-06,
      "loss": 0.7112,
      "num_input_tokens_seen": 253679644,
      "step": 4529
    },
    {
      "epoch": 10.086859688195991,
      "loss": 0.73432457447052,
      "loss_ce": 0.00019371899543330073,
      "loss_iou": 0.326171875,
      "loss_num": 0.0164794921875,
      "loss_xval": 0.734375,
      "num_input_tokens_seen": 253679644,
      "step": 4529
    },
    {
      "epoch": 10.089086859688196,
      "grad_norm": 15.54879379272461,
      "learning_rate": 1e-06,
      "loss": 0.5008,
      "num_input_tokens_seen": 253738164,
      "step": 4530
    },
    {
      "epoch": 10.089086859688196,
      "loss": 0.4535183012485504,
      "loss_ce": 0.00014916164218448102,
      "loss_iou": 0.1875,
      "loss_num": 0.015625,
      "loss_xval": 0.453125,
      "num_input_tokens_seen": 253738164,
      "step": 4530
    },
    {
      "epoch": 10.091314031180401,
      "grad_norm": 26.567584991455078,
      "learning_rate": 1e-06,
      "loss": 0.6987,
      "num_input_tokens_seen": 253793512,
      "step": 4531
    },
    {
      "epoch": 10.091314031180401,
      "loss": 0.8244662880897522,
      "loss_ce": 0.0002475357032380998,
      "loss_iou": 0.34375,
      "loss_num": 0.0274658203125,
      "loss_xval": 0.82421875,
      "num_input_tokens_seen": 253793512,
      "step": 4531
    },
    {
      "epoch": 10.093541202672606,
      "grad_norm": 17.407928466796875,
      "learning_rate": 1e-06,
      "loss": 0.6066,
      "num_input_tokens_seen": 253851088,
      "step": 4532
    },
    {
      "epoch": 10.093541202672606,
      "loss": 0.7939009666442871,
      "loss_ce": 0.0009321961551904678,
      "loss_iou": 0.32421875,
      "loss_num": 0.02880859375,
      "loss_xval": 0.79296875,
      "num_input_tokens_seen": 253851088,
      "step": 4532
    },
    {
      "epoch": 10.09576837416481,
      "grad_norm": 15.048906326293945,
      "learning_rate": 1e-06,
      "loss": 0.4981,
      "num_input_tokens_seen": 253906892,
      "step": 4533
    },
    {
      "epoch": 10.09576837416481,
      "loss": 0.5735203623771667,
      "loss_ce": 0.00015607741079293191,
      "loss_iou": 0.232421875,
      "loss_num": 0.0216064453125,
      "loss_xval": 0.57421875,
      "num_input_tokens_seen": 253906892,
      "step": 4533
    },
    {
      "epoch": 10.097995545657016,
      "grad_norm": 19.791410446166992,
      "learning_rate": 1e-06,
      "loss": 0.5475,
      "num_input_tokens_seen": 253964416,
      "step": 4534
    },
    {
      "epoch": 10.097995545657016,
      "loss": 0.5835007429122925,
      "loss_ce": 0.00014201825251802802,
      "loss_iou": 0.2490234375,
      "loss_num": 0.01708984375,
      "loss_xval": 0.58203125,
      "num_input_tokens_seen": 253964416,
      "step": 4534
    },
    {
      "epoch": 10.10022271714922,
      "grad_norm": 16.358074188232422,
      "learning_rate": 1e-06,
      "loss": 0.3828,
      "num_input_tokens_seen": 254021404,
      "step": 4535
    },
    {
      "epoch": 10.10022271714922,
      "loss": 0.48548585176467896,
      "loss_ce": 0.00013428418606054038,
      "loss_iou": 0.2099609375,
      "loss_num": 0.01300048828125,
      "loss_xval": 0.484375,
      "num_input_tokens_seen": 254021404,
      "step": 4535
    },
    {
      "epoch": 10.102449888641425,
      "grad_norm": 20.186296463012695,
      "learning_rate": 1e-06,
      "loss": 0.7477,
      "num_input_tokens_seen": 254076892,
      "step": 4536
    },
    {
      "epoch": 10.102449888641425,
      "loss": 0.8367500305175781,
      "loss_ce": 0.0001716263359412551,
      "loss_iou": 0.328125,
      "loss_num": 0.035888671875,
      "loss_xval": 0.8359375,
      "num_input_tokens_seen": 254076892,
      "step": 4536
    },
    {
      "epoch": 10.10467706013363,
      "grad_norm": 23.128829956054688,
      "learning_rate": 1e-06,
      "loss": 0.6005,
      "num_input_tokens_seen": 254132820,
      "step": 4537
    },
    {
      "epoch": 10.10467706013363,
      "loss": 0.7570396661758423,
      "loss_ce": 0.00020372896688058972,
      "loss_iou": 0.345703125,
      "loss_num": 0.01300048828125,
      "loss_xval": 0.7578125,
      "num_input_tokens_seen": 254132820,
      "step": 4537
    },
    {
      "epoch": 10.106904231625835,
      "grad_norm": 19.934675216674805,
      "learning_rate": 1e-06,
      "loss": 0.5398,
      "num_input_tokens_seen": 254190556,
      "step": 4538
    },
    {
      "epoch": 10.106904231625835,
      "loss": 0.46338510513305664,
      "loss_ce": 0.0004944695974700153,
      "loss_iou": 0.2080078125,
      "loss_num": 0.009521484375,
      "loss_xval": 0.462890625,
      "num_input_tokens_seen": 254190556,
      "step": 4538
    },
    {
      "epoch": 10.10913140311804,
      "grad_norm": 24.60442352294922,
      "learning_rate": 1e-06,
      "loss": 0.6456,
      "num_input_tokens_seen": 254248000,
      "step": 4539
    },
    {
      "epoch": 10.10913140311804,
      "loss": 0.7887189984321594,
      "loss_ce": 0.0001448030088795349,
      "loss_iou": 0.310546875,
      "loss_num": 0.033447265625,
      "loss_xval": 0.7890625,
      "num_input_tokens_seen": 254248000,
      "step": 4539
    },
    {
      "epoch": 10.111358574610245,
      "grad_norm": 16.316864013671875,
      "learning_rate": 1e-06,
      "loss": 0.5443,
      "num_input_tokens_seen": 254306248,
      "step": 4540
    },
    {
      "epoch": 10.111358574610245,
      "loss": 0.3741779625415802,
      "loss_ce": 0.00015453985542990267,
      "loss_iou": 0.15625,
      "loss_num": 0.01220703125,
      "loss_xval": 0.375,
      "num_input_tokens_seen": 254306248,
      "step": 4540
    },
    {
      "epoch": 10.11358574610245,
      "grad_norm": 19.49997901916504,
      "learning_rate": 1e-06,
      "loss": 0.6895,
      "num_input_tokens_seen": 254359852,
      "step": 4541
    },
    {
      "epoch": 10.11358574610245,
      "loss": 0.6783727407455444,
      "loss_ce": 0.00015005419845692813,
      "loss_iou": 0.27734375,
      "loss_num": 0.0250244140625,
      "loss_xval": 0.6796875,
      "num_input_tokens_seen": 254359852,
      "step": 4541
    },
    {
      "epoch": 10.115812917594655,
      "grad_norm": 16.376052856445312,
      "learning_rate": 1e-06,
      "loss": 0.4816,
      "num_input_tokens_seen": 254414044,
      "step": 4542
    },
    {
      "epoch": 10.115812917594655,
      "loss": 0.526771605014801,
      "loss_ce": 0.00016028305981308222,
      "loss_iou": 0.2294921875,
      "loss_num": 0.01385498046875,
      "loss_xval": 0.52734375,
      "num_input_tokens_seen": 254414044,
      "step": 4542
    },
    {
      "epoch": 10.11804008908686,
      "grad_norm": 20.01656723022461,
      "learning_rate": 1e-06,
      "loss": 0.5995,
      "num_input_tokens_seen": 254470696,
      "step": 4543
    },
    {
      "epoch": 10.11804008908686,
      "loss": 0.538500964641571,
      "loss_ce": 0.000170878556673415,
      "loss_iou": 0.2109375,
      "loss_num": 0.02294921875,
      "loss_xval": 0.5390625,
      "num_input_tokens_seen": 254470696,
      "step": 4543
    },
    {
      "epoch": 10.120267260579064,
      "grad_norm": 15.220352172851562,
      "learning_rate": 1e-06,
      "loss": 0.5189,
      "num_input_tokens_seen": 254526240,
      "step": 4544
    },
    {
      "epoch": 10.120267260579064,
      "loss": 0.5009964108467102,
      "loss_ce": 0.0002639779122546315,
      "loss_iou": 0.2236328125,
      "loss_num": 0.0106201171875,
      "loss_xval": 0.5,
      "num_input_tokens_seen": 254526240,
      "step": 4544
    },
    {
      "epoch": 10.122494432071269,
      "grad_norm": 15.65402889251709,
      "learning_rate": 1e-06,
      "loss": 0.5134,
      "num_input_tokens_seen": 254582940,
      "step": 4545
    },
    {
      "epoch": 10.122494432071269,
      "loss": 0.52699214220047,
      "loss_ce": 0.0001366714423056692,
      "loss_iou": 0.232421875,
      "loss_num": 0.012451171875,
      "loss_xval": 0.52734375,
      "num_input_tokens_seen": 254582940,
      "step": 4545
    },
    {
      "epoch": 10.124721603563474,
      "grad_norm": 21.156890869140625,
      "learning_rate": 1e-06,
      "loss": 0.5,
      "num_input_tokens_seen": 254638184,
      "step": 4546
    },
    {
      "epoch": 10.124721603563474,
      "loss": 0.5189378261566162,
      "loss_ce": 0.0001390243851346895,
      "loss_iou": 0.236328125,
      "loss_num": 0.009033203125,
      "loss_xval": 0.51953125,
      "num_input_tokens_seen": 254638184,
      "step": 4546
    },
    {
      "epoch": 10.126948775055679,
      "grad_norm": 20.83241844177246,
      "learning_rate": 1e-06,
      "loss": 0.4975,
      "num_input_tokens_seen": 254692884,
      "step": 4547
    },
    {
      "epoch": 10.126948775055679,
      "loss": 0.4995710849761963,
      "loss_ce": 0.00018145760986953974,
      "loss_iou": 0.21484375,
      "loss_num": 0.0137939453125,
      "loss_xval": 0.5,
      "num_input_tokens_seen": 254692884,
      "step": 4547
    },
    {
      "epoch": 10.129175946547884,
      "grad_norm": 146.1394805908203,
      "learning_rate": 1e-06,
      "loss": 0.5299,
      "num_input_tokens_seen": 254749104,
      "step": 4548
    },
    {
      "epoch": 10.129175946547884,
      "loss": 0.6289869546890259,
      "loss_ce": 0.001545518171042204,
      "loss_iou": 0.240234375,
      "loss_num": 0.0294189453125,
      "loss_xval": 0.62890625,
      "num_input_tokens_seen": 254749104,
      "step": 4548
    },
    {
      "epoch": 10.131403118040089,
      "grad_norm": 21.525724411010742,
      "learning_rate": 1e-06,
      "loss": 0.4352,
      "num_input_tokens_seen": 254806856,
      "step": 4549
    },
    {
      "epoch": 10.131403118040089,
      "loss": 0.44396907091140747,
      "loss_ce": 0.00012142823834437877,
      "loss_iou": 0.203125,
      "loss_num": 0.00738525390625,
      "loss_xval": 0.443359375,
      "num_input_tokens_seen": 254806856,
      "step": 4549
    },
    {
      "epoch": 10.133630289532293,
      "grad_norm": 15.656088829040527,
      "learning_rate": 1e-06,
      "loss": 0.486,
      "num_input_tokens_seen": 254862556,
      "step": 4550
    },
    {
      "epoch": 10.133630289532293,
      "loss": 0.3813803195953369,
      "loss_ce": 0.0001547573774587363,
      "loss_iou": 0.162109375,
      "loss_num": 0.011474609375,
      "loss_xval": 0.380859375,
      "num_input_tokens_seen": 254862556,
      "step": 4550
    },
    {
      "epoch": 10.135857461024498,
      "grad_norm": 18.48883628845215,
      "learning_rate": 1e-06,
      "loss": 0.4152,
      "num_input_tokens_seen": 254921356,
      "step": 4551
    },
    {
      "epoch": 10.135857461024498,
      "loss": 0.4063987135887146,
      "loss_ce": 0.00027079382562078536,
      "loss_iou": 0.1796875,
      "loss_num": 0.00958251953125,
      "loss_xval": 0.40625,
      "num_input_tokens_seen": 254921356,
      "step": 4551
    },
    {
      "epoch": 10.138084632516703,
      "grad_norm": 14.69189739227295,
      "learning_rate": 1e-06,
      "loss": 0.576,
      "num_input_tokens_seen": 254979752,
      "step": 4552
    },
    {
      "epoch": 10.138084632516703,
      "loss": 0.520897388458252,
      "loss_ce": 0.00014549962361343205,
      "loss_iou": 0.2216796875,
      "loss_num": 0.0152587890625,
      "loss_xval": 0.51953125,
      "num_input_tokens_seen": 254979752,
      "step": 4552
    },
    {
      "epoch": 10.140311804008908,
      "grad_norm": 16.03564453125,
      "learning_rate": 1e-06,
      "loss": 0.4583,
      "num_input_tokens_seen": 255037852,
      "step": 4553
    },
    {
      "epoch": 10.140311804008908,
      "loss": 0.500576376914978,
      "loss_ce": 0.0005153242964297533,
      "loss_iou": 0.2275390625,
      "loss_num": 0.0087890625,
      "loss_xval": 0.5,
      "num_input_tokens_seen": 255037852,
      "step": 4553
    },
    {
      "epoch": 10.142538975501113,
      "grad_norm": 14.397992134094238,
      "learning_rate": 1e-06,
      "loss": 0.5558,
      "num_input_tokens_seen": 255095480,
      "step": 4554
    },
    {
      "epoch": 10.142538975501113,
      "loss": 0.4591131806373596,
      "loss_ce": 0.0001288054045289755,
      "loss_iou": 0.2080078125,
      "loss_num": 0.0086669921875,
      "loss_xval": 0.458984375,
      "num_input_tokens_seen": 255095480,
      "step": 4554
    },
    {
      "epoch": 10.144766146993318,
      "grad_norm": 16.567859649658203,
      "learning_rate": 1e-06,
      "loss": 0.3819,
      "num_input_tokens_seen": 255154316,
      "step": 4555
    },
    {
      "epoch": 10.144766146993318,
      "loss": 0.3125069737434387,
      "loss_ce": 0.0001290409272769466,
      "loss_iou": 0.115234375,
      "loss_num": 0.0164794921875,
      "loss_xval": 0.3125,
      "num_input_tokens_seen": 255154316,
      "step": 4555
    },
    {
      "epoch": 10.146993318485523,
      "grad_norm": 13.150337219238281,
      "learning_rate": 1e-06,
      "loss": 0.4094,
      "num_input_tokens_seen": 255211692,
      "step": 4556
    },
    {
      "epoch": 10.146993318485523,
      "loss": 0.3635147213935852,
      "loss_ce": 0.00011139338312204927,
      "loss_iou": 0.15625,
      "loss_num": 0.0101318359375,
      "loss_xval": 0.36328125,
      "num_input_tokens_seen": 255211692,
      "step": 4556
    },
    {
      "epoch": 10.14922048997773,
      "grad_norm": 18.678987503051758,
      "learning_rate": 1e-06,
      "loss": 0.2899,
      "num_input_tokens_seen": 255265416,
      "step": 4557
    },
    {
      "epoch": 10.14922048997773,
      "loss": 0.3159905672073364,
      "loss_ce": 0.0001946770935319364,
      "loss_iou": 0.130859375,
      "loss_num": 0.01080322265625,
      "loss_xval": 0.31640625,
      "num_input_tokens_seen": 255265416,
      "step": 4557
    },
    {
      "epoch": 10.151447661469934,
      "grad_norm": 15.587052345275879,
      "learning_rate": 1e-06,
      "loss": 0.5265,
      "num_input_tokens_seen": 255319560,
      "step": 4558
    },
    {
      "epoch": 10.151447661469934,
      "loss": 0.3966302275657654,
      "loss_ce": 0.0001458691549487412,
      "loss_iou": 0.177734375,
      "loss_num": 0.00830078125,
      "loss_xval": 0.396484375,
      "num_input_tokens_seen": 255319560,
      "step": 4558
    },
    {
      "epoch": 10.153674832962139,
      "grad_norm": 19.554283142089844,
      "learning_rate": 1e-06,
      "loss": 0.5922,
      "num_input_tokens_seen": 255372816,
      "step": 4559
    },
    {
      "epoch": 10.153674832962139,
      "loss": 0.5624641180038452,
      "loss_ce": 0.0002082242863252759,
      "loss_iou": 0.2158203125,
      "loss_num": 0.0260009765625,
      "loss_xval": 0.5625,
      "num_input_tokens_seen": 255372816,
      "step": 4559
    },
    {
      "epoch": 10.155902004454344,
      "grad_norm": 27.854368209838867,
      "learning_rate": 1e-06,
      "loss": 0.5583,
      "num_input_tokens_seen": 255426296,
      "step": 4560
    },
    {
      "epoch": 10.155902004454344,
      "loss": 0.5693508386611938,
      "loss_ce": 0.00013697383110411465,
      "loss_iou": 0.2373046875,
      "loss_num": 0.018798828125,
      "loss_xval": 0.5703125,
      "num_input_tokens_seen": 255426296,
      "step": 4560
    },
    {
      "epoch": 10.158129175946549,
      "grad_norm": 22.34395408630371,
      "learning_rate": 1e-06,
      "loss": 0.4309,
      "num_input_tokens_seen": 255479824,
      "step": 4561
    },
    {
      "epoch": 10.158129175946549,
      "loss": 0.3726035952568054,
      "loss_ce": 0.0002891222247853875,
      "loss_iou": 0.1494140625,
      "loss_num": 0.0147705078125,
      "loss_xval": 0.373046875,
      "num_input_tokens_seen": 255479824,
      "step": 4561
    },
    {
      "epoch": 10.160356347438753,
      "grad_norm": 16.66666603088379,
      "learning_rate": 1e-06,
      "loss": 0.4812,
      "num_input_tokens_seen": 255533776,
      "step": 4562
    },
    {
      "epoch": 10.160356347438753,
      "loss": 0.46586695313453674,
      "loss_ce": 0.0002907839370891452,
      "loss_iou": 0.1748046875,
      "loss_num": 0.023193359375,
      "loss_xval": 0.46484375,
      "num_input_tokens_seen": 255533776,
      "step": 4562
    },
    {
      "epoch": 10.162583518930958,
      "grad_norm": 15.96545696258545,
      "learning_rate": 1e-06,
      "loss": 0.633,
      "num_input_tokens_seen": 255592708,
      "step": 4563
    },
    {
      "epoch": 10.162583518930958,
      "loss": 0.7283331751823425,
      "loss_ce": 0.00018375377112533897,
      "loss_iou": 0.3203125,
      "loss_num": 0.017578125,
      "loss_xval": 0.7265625,
      "num_input_tokens_seen": 255592708,
      "step": 4563
    },
    {
      "epoch": 10.164810690423163,
      "grad_norm": 17.769378662109375,
      "learning_rate": 1e-06,
      "loss": 0.5003,
      "num_input_tokens_seen": 255651044,
      "step": 4564
    },
    {
      "epoch": 10.164810690423163,
      "loss": 0.47229546308517456,
      "loss_ce": 0.00012747224536724389,
      "loss_iou": 0.1962890625,
      "loss_num": 0.0159912109375,
      "loss_xval": 0.47265625,
      "num_input_tokens_seen": 255651044,
      "step": 4564
    },
    {
      "epoch": 10.167037861915368,
      "grad_norm": 37.10343933105469,
      "learning_rate": 1e-06,
      "loss": 0.5678,
      "num_input_tokens_seen": 255706820,
      "step": 4565
    },
    {
      "epoch": 10.167037861915368,
      "loss": 0.5193136930465698,
      "loss_ce": 0.00027068209601566195,
      "loss_iou": 0.2255859375,
      "loss_num": 0.01373291015625,
      "loss_xval": 0.51953125,
      "num_input_tokens_seen": 255706820,
      "step": 4565
    },
    {
      "epoch": 10.169265033407573,
      "grad_norm": 16.142242431640625,
      "learning_rate": 1e-06,
      "loss": 0.5588,
      "num_input_tokens_seen": 255764732,
      "step": 4566
    },
    {
      "epoch": 10.169265033407573,
      "loss": 0.6505287885665894,
      "loss_ce": 0.00013815713464282453,
      "loss_iou": 0.29296875,
      "loss_num": 0.0133056640625,
      "loss_xval": 0.6484375,
      "num_input_tokens_seen": 255764732,
      "step": 4566
    },
    {
      "epoch": 10.171492204899778,
      "grad_norm": 26.361278533935547,
      "learning_rate": 1e-06,
      "loss": 0.5016,
      "num_input_tokens_seen": 255821704,
      "step": 4567
    },
    {
      "epoch": 10.171492204899778,
      "loss": 0.5541030168533325,
      "loss_ce": 0.0001479470229241997,
      "loss_iou": 0.240234375,
      "loss_num": 0.0150146484375,
      "loss_xval": 0.5546875,
      "num_input_tokens_seen": 255821704,
      "step": 4567
    },
    {
      "epoch": 10.173719376391983,
      "grad_norm": 14.757318496704102,
      "learning_rate": 1e-06,
      "loss": 0.5656,
      "num_input_tokens_seen": 255880672,
      "step": 4568
    },
    {
      "epoch": 10.173719376391983,
      "loss": 0.6377292275428772,
      "loss_ce": 0.00015599204925820231,
      "loss_iou": 0.263671875,
      "loss_num": 0.0223388671875,
      "loss_xval": 0.63671875,
      "num_input_tokens_seen": 255880672,
      "step": 4568
    },
    {
      "epoch": 10.175946547884188,
      "grad_norm": 20.087034225463867,
      "learning_rate": 1e-06,
      "loss": 0.5458,
      "num_input_tokens_seen": 255938112,
      "step": 4569
    },
    {
      "epoch": 10.175946547884188,
      "loss": 0.5018346905708313,
      "loss_ce": 0.002200859831646085,
      "loss_iou": 0.2265625,
      "loss_num": 0.0093994140625,
      "loss_xval": 0.5,
      "num_input_tokens_seen": 255938112,
      "step": 4569
    },
    {
      "epoch": 10.178173719376392,
      "grad_norm": 17.099891662597656,
      "learning_rate": 1e-06,
      "loss": 0.6128,
      "num_input_tokens_seen": 255994480,
      "step": 4570
    },
    {
      "epoch": 10.178173719376392,
      "loss": 0.6893445253372192,
      "loss_ce": 0.00013553063035942614,
      "loss_iou": 0.27734375,
      "loss_num": 0.0267333984375,
      "loss_xval": 0.6875,
      "num_input_tokens_seen": 255994480,
      "step": 4570
    },
    {
      "epoch": 10.180400890868597,
      "grad_norm": 24.685178756713867,
      "learning_rate": 1e-06,
      "loss": 0.5695,
      "num_input_tokens_seen": 256051472,
      "step": 4571
    },
    {
      "epoch": 10.180400890868597,
      "loss": 0.691739022731781,
      "loss_ce": 0.0013093581655994058,
      "loss_iou": 0.306640625,
      "loss_num": 0.015380859375,
      "loss_xval": 0.69140625,
      "num_input_tokens_seen": 256051472,
      "step": 4571
    },
    {
      "epoch": 10.182628062360802,
      "grad_norm": 22.65061378479004,
      "learning_rate": 1e-06,
      "loss": 0.5629,
      "num_input_tokens_seen": 256104576,
      "step": 4572
    },
    {
      "epoch": 10.182628062360802,
      "loss": 0.7198662757873535,
      "loss_ce": 0.00013974003377370536,
      "loss_iou": 0.306640625,
      "loss_num": 0.021240234375,
      "loss_xval": 0.71875,
      "num_input_tokens_seen": 256104576,
      "step": 4572
    },
    {
      "epoch": 10.184855233853007,
      "grad_norm": 21.435075759887695,
      "learning_rate": 1e-06,
      "loss": 0.4938,
      "num_input_tokens_seen": 256159868,
      "step": 4573
    },
    {
      "epoch": 10.184855233853007,
      "loss": 0.5475330352783203,
      "loss_ce": 0.00016974794561974704,
      "loss_iou": 0.240234375,
      "loss_num": 0.0133056640625,
      "loss_xval": 0.546875,
      "num_input_tokens_seen": 256159868,
      "step": 4573
    },
    {
      "epoch": 10.187082405345212,
      "grad_norm": 15.654433250427246,
      "learning_rate": 1e-06,
      "loss": 0.5428,
      "num_input_tokens_seen": 256216408,
      "step": 4574
    },
    {
      "epoch": 10.187082405345212,
      "loss": 0.6571834087371826,
      "loss_ce": 0.00020100505207665265,
      "loss_iou": 0.2734375,
      "loss_num": 0.0224609375,
      "loss_xval": 0.65625,
      "num_input_tokens_seen": 256216408,
      "step": 4574
    },
    {
      "epoch": 10.189309576837417,
      "grad_norm": 20.29172706604004,
      "learning_rate": 1e-06,
      "loss": 0.4126,
      "num_input_tokens_seen": 256272580,
      "step": 4575
    },
    {
      "epoch": 10.189309576837417,
      "loss": 0.3565613627433777,
      "loss_ce": 0.00011605105100898072,
      "loss_iou": 0.154296875,
      "loss_num": 0.00958251953125,
      "loss_xval": 0.35546875,
      "num_input_tokens_seen": 256272580,
      "step": 4575
    },
    {
      "epoch": 10.191536748329622,
      "grad_norm": 47.00114440917969,
      "learning_rate": 1e-06,
      "loss": 0.486,
      "num_input_tokens_seen": 256327452,
      "step": 4576
    },
    {
      "epoch": 10.191536748329622,
      "loss": 0.6286421418190002,
      "loss_ce": 0.00022419106971938163,
      "loss_iou": 0.263671875,
      "loss_num": 0.0198974609375,
      "loss_xval": 0.62890625,
      "num_input_tokens_seen": 256327452,
      "step": 4576
    },
    {
      "epoch": 10.193763919821826,
      "grad_norm": 15.635924339294434,
      "learning_rate": 1e-06,
      "loss": 0.547,
      "num_input_tokens_seen": 256383876,
      "step": 4577
    },
    {
      "epoch": 10.193763919821826,
      "loss": 0.6068173050880432,
      "loss_ce": 0.00012786718434654176,
      "loss_iou": 0.267578125,
      "loss_num": 0.01409912109375,
      "loss_xval": 0.60546875,
      "num_input_tokens_seen": 256383876,
      "step": 4577
    },
    {
      "epoch": 10.195991091314031,
      "grad_norm": 24.49083137512207,
      "learning_rate": 1e-06,
      "loss": 0.4668,
      "num_input_tokens_seen": 256441820,
      "step": 4578
    },
    {
      "epoch": 10.195991091314031,
      "loss": 0.49414756894111633,
      "loss_ce": 0.00012899917783215642,
      "loss_iou": 0.21875,
      "loss_num": 0.0113525390625,
      "loss_xval": 0.494140625,
      "num_input_tokens_seen": 256441820,
      "step": 4578
    },
    {
      "epoch": 10.198218262806236,
      "grad_norm": 18.868743896484375,
      "learning_rate": 1e-06,
      "loss": 0.7535,
      "num_input_tokens_seen": 256495464,
      "step": 4579
    },
    {
      "epoch": 10.198218262806236,
      "loss": 0.7945007681846619,
      "loss_ce": 0.0001892938744276762,
      "loss_iou": 0.30859375,
      "loss_num": 0.03564453125,
      "loss_xval": 0.79296875,
      "num_input_tokens_seen": 256495464,
      "step": 4579
    },
    {
      "epoch": 10.200445434298441,
      "grad_norm": 18.00078773498535,
      "learning_rate": 1e-06,
      "loss": 0.5301,
      "num_input_tokens_seen": 256551316,
      "step": 4580
    },
    {
      "epoch": 10.200445434298441,
      "loss": 0.6253817677497864,
      "loss_ce": 0.000137610943056643,
      "loss_iou": 0.2734375,
      "loss_num": 0.01531982421875,
      "loss_xval": 0.625,
      "num_input_tokens_seen": 256551316,
      "step": 4580
    },
    {
      "epoch": 10.202672605790646,
      "grad_norm": 14.672399520874023,
      "learning_rate": 1e-06,
      "loss": 0.5689,
      "num_input_tokens_seen": 256607460,
      "step": 4581
    },
    {
      "epoch": 10.202672605790646,
      "loss": 0.7001278400421143,
      "loss_ce": 0.0001766737550497055,
      "loss_iou": 0.283203125,
      "loss_num": 0.0267333984375,
      "loss_xval": 0.69921875,
      "num_input_tokens_seen": 256607460,
      "step": 4581
    },
    {
      "epoch": 10.20489977728285,
      "grad_norm": 27.121843338012695,
      "learning_rate": 1e-06,
      "loss": 0.3811,
      "num_input_tokens_seen": 256662608,
      "step": 4582
    },
    {
      "epoch": 10.20489977728285,
      "loss": 0.28326690196990967,
      "loss_ce": 0.00012479553697630763,
      "loss_iou": 0.107421875,
      "loss_num": 0.013671875,
      "loss_xval": 0.283203125,
      "num_input_tokens_seen": 256662608,
      "step": 4582
    },
    {
      "epoch": 10.207126948775056,
      "grad_norm": 22.3253231048584,
      "learning_rate": 1e-06,
      "loss": 0.49,
      "num_input_tokens_seen": 256719680,
      "step": 4583
    },
    {
      "epoch": 10.207126948775056,
      "loss": 0.5782710313796997,
      "loss_ce": 0.0001460201747249812,
      "loss_iou": 0.2451171875,
      "loss_num": 0.0174560546875,
      "loss_xval": 0.578125,
      "num_input_tokens_seen": 256719680,
      "step": 4583
    },
    {
      "epoch": 10.20935412026726,
      "grad_norm": 15.034200668334961,
      "learning_rate": 1e-06,
      "loss": 0.3748,
      "num_input_tokens_seen": 256775480,
      "step": 4584
    },
    {
      "epoch": 10.20935412026726,
      "loss": 0.3525688052177429,
      "loss_ce": 0.0001518014323664829,
      "loss_iou": 0.1484375,
      "loss_num": 0.01123046875,
      "loss_xval": 0.3515625,
      "num_input_tokens_seen": 256775480,
      "step": 4584
    },
    {
      "epoch": 10.211581291759465,
      "grad_norm": 20.57643699645996,
      "learning_rate": 1e-06,
      "loss": 0.4699,
      "num_input_tokens_seen": 256830920,
      "step": 4585
    },
    {
      "epoch": 10.211581291759465,
      "loss": 0.6310067772865295,
      "loss_ce": 0.00014741039194632322,
      "loss_iou": 0.28125,
      "loss_num": 0.01348876953125,
      "loss_xval": 0.6328125,
      "num_input_tokens_seen": 256830920,
      "step": 4585
    },
    {
      "epoch": 10.21380846325167,
      "grad_norm": 19.190690994262695,
      "learning_rate": 1e-06,
      "loss": 0.6572,
      "num_input_tokens_seen": 256885536,
      "step": 4586
    },
    {
      "epoch": 10.21380846325167,
      "loss": 0.8363385200500488,
      "loss_ce": 0.0001568598672747612,
      "loss_iou": 0.36328125,
      "loss_num": 0.022216796875,
      "loss_xval": 0.8359375,
      "num_input_tokens_seen": 256885536,
      "step": 4586
    },
    {
      "epoch": 10.216035634743875,
      "grad_norm": 19.48312759399414,
      "learning_rate": 1e-06,
      "loss": 0.7196,
      "num_input_tokens_seen": 256944132,
      "step": 4587
    },
    {
      "epoch": 10.216035634743875,
      "loss": 0.6926649808883667,
      "loss_ce": 0.00016006956866476685,
      "loss_iou": 0.3046875,
      "loss_num": 0.0164794921875,
      "loss_xval": 0.69140625,
      "num_input_tokens_seen": 256944132,
      "step": 4587
    },
    {
      "epoch": 10.21826280623608,
      "grad_norm": 28.042089462280273,
      "learning_rate": 1e-06,
      "loss": 0.6051,
      "num_input_tokens_seen": 256998580,
      "step": 4588
    },
    {
      "epoch": 10.21826280623608,
      "loss": 0.7248637080192566,
      "loss_ce": 0.0006205601966939867,
      "loss_iou": 0.30078125,
      "loss_num": 0.0247802734375,
      "loss_xval": 0.72265625,
      "num_input_tokens_seen": 256998580,
      "step": 4588
    },
    {
      "epoch": 10.220489977728285,
      "grad_norm": 17.185420989990234,
      "learning_rate": 1e-06,
      "loss": 0.427,
      "num_input_tokens_seen": 257053016,
      "step": 4589
    },
    {
      "epoch": 10.220489977728285,
      "loss": 0.4973496198654175,
      "loss_ce": 0.0002793156891129911,
      "loss_iou": 0.1982421875,
      "loss_num": 0.0203857421875,
      "loss_xval": 0.49609375,
      "num_input_tokens_seen": 257053016,
      "step": 4589
    },
    {
      "epoch": 10.22271714922049,
      "grad_norm": 22.403812408447266,
      "learning_rate": 1e-06,
      "loss": 0.4526,
      "num_input_tokens_seen": 257109612,
      "step": 4590
    },
    {
      "epoch": 10.22271714922049,
      "loss": 0.4690048098564148,
      "loss_ce": 0.00013278050755616277,
      "loss_iou": 0.2080078125,
      "loss_num": 0.010498046875,
      "loss_xval": 0.46875,
      "num_input_tokens_seen": 257109612,
      "step": 4590
    },
    {
      "epoch": 10.224944320712694,
      "grad_norm": 19.512828826904297,
      "learning_rate": 1e-06,
      "loss": 0.5368,
      "num_input_tokens_seen": 257166788,
      "step": 4591
    },
    {
      "epoch": 10.224944320712694,
      "loss": 0.5036743879318237,
      "loss_ce": 0.00013440893962979317,
      "loss_iou": 0.2294921875,
      "loss_num": 0.0086669921875,
      "loss_xval": 0.50390625,
      "num_input_tokens_seen": 257166788,
      "step": 4591
    },
    {
      "epoch": 10.2271714922049,
      "grad_norm": 29.91023826599121,
      "learning_rate": 1e-06,
      "loss": 0.6053,
      "num_input_tokens_seen": 257223768,
      "step": 4592
    },
    {
      "epoch": 10.2271714922049,
      "loss": 0.4671659767627716,
      "loss_ce": 0.00012497020361479372,
      "loss_iou": 0.2177734375,
      "loss_num": 0.00628662109375,
      "loss_xval": 0.466796875,
      "num_input_tokens_seen": 257223768,
      "step": 4592
    },
    {
      "epoch": 10.229398663697104,
      "grad_norm": 21.827224731445312,
      "learning_rate": 1e-06,
      "loss": 0.6082,
      "num_input_tokens_seen": 257281108,
      "step": 4593
    },
    {
      "epoch": 10.229398663697104,
      "loss": 0.4103159010410309,
      "loss_ce": 0.00015964708290994167,
      "loss_iou": 0.185546875,
      "loss_num": 0.007781982421875,
      "loss_xval": 0.41015625,
      "num_input_tokens_seen": 257281108,
      "step": 4593
    },
    {
      "epoch": 10.231625835189309,
      "grad_norm": 14.107495307922363,
      "learning_rate": 1e-06,
      "loss": 0.5821,
      "num_input_tokens_seen": 257334832,
      "step": 4594
    },
    {
      "epoch": 10.231625835189309,
      "loss": 0.5812970399856567,
      "loss_ce": 0.00012028503260808066,
      "loss_iou": 0.255859375,
      "loss_num": 0.01348876953125,
      "loss_xval": 0.58203125,
      "num_input_tokens_seen": 257334832,
      "step": 4594
    },
    {
      "epoch": 10.233853006681514,
      "grad_norm": 30.097774505615234,
      "learning_rate": 1e-06,
      "loss": 0.4502,
      "num_input_tokens_seen": 257388612,
      "step": 4595
    },
    {
      "epoch": 10.233853006681514,
      "loss": 0.47809213399887085,
      "loss_ce": 0.00018684033420868218,
      "loss_iou": 0.2158203125,
      "loss_num": 0.00927734375,
      "loss_xval": 0.478515625,
      "num_input_tokens_seen": 257388612,
      "step": 4595
    },
    {
      "epoch": 10.236080178173719,
      "grad_norm": 287.798095703125,
      "learning_rate": 1e-06,
      "loss": 0.613,
      "num_input_tokens_seen": 257445652,
      "step": 4596
    },
    {
      "epoch": 10.236080178173719,
      "loss": 0.5535175800323486,
      "loss_ce": 0.00029491656459867954,
      "loss_iou": 0.224609375,
      "loss_num": 0.0206298828125,
      "loss_xval": 0.5546875,
      "num_input_tokens_seen": 257445652,
      "step": 4596
    },
    {
      "epoch": 10.238307349665924,
      "grad_norm": 380.41595458984375,
      "learning_rate": 1e-06,
      "loss": 0.7191,
      "num_input_tokens_seen": 257499196,
      "step": 4597
    },
    {
      "epoch": 10.238307349665924,
      "loss": 0.7513624429702759,
      "loss_ce": 0.00014180070138536394,
      "loss_iou": 0.306640625,
      "loss_num": 0.027587890625,
      "loss_xval": 0.75,
      "num_input_tokens_seen": 257499196,
      "step": 4597
    },
    {
      "epoch": 10.240534521158128,
      "grad_norm": 48.11332321166992,
      "learning_rate": 1e-06,
      "loss": 0.6278,
      "num_input_tokens_seen": 257553432,
      "step": 4598
    },
    {
      "epoch": 10.240534521158128,
      "loss": 0.884353756904602,
      "loss_ce": 0.00019845434871967882,
      "loss_iou": 0.37890625,
      "loss_num": 0.0247802734375,
      "loss_xval": 0.8828125,
      "num_input_tokens_seen": 257553432,
      "step": 4598
    },
    {
      "epoch": 10.242761692650333,
      "grad_norm": 14.302517890930176,
      "learning_rate": 1e-06,
      "loss": 0.4885,
      "num_input_tokens_seen": 257611352,
      "step": 4599
    },
    {
      "epoch": 10.242761692650333,
      "loss": 0.5572800636291504,
      "loss_ce": 0.00015118328155949712,
      "loss_iou": 0.224609375,
      "loss_num": 0.021728515625,
      "loss_xval": 0.55859375,
      "num_input_tokens_seen": 257611352,
      "step": 4599
    },
    {
      "epoch": 10.244988864142538,
      "grad_norm": 15.14469051361084,
      "learning_rate": 1e-06,
      "loss": 0.5646,
      "num_input_tokens_seen": 257663240,
      "step": 4600
    },
    {
      "epoch": 10.244988864142538,
      "loss": 0.3984256088733673,
      "loss_ce": 0.000232246849918738,
      "loss_iou": 0.154296875,
      "loss_num": 0.01806640625,
      "loss_xval": 0.3984375,
      "num_input_tokens_seen": 257663240,
      "step": 4600
    },
    {
      "epoch": 10.247216035634743,
      "grad_norm": 17.56197738647461,
      "learning_rate": 1e-06,
      "loss": 0.537,
      "num_input_tokens_seen": 257719556,
      "step": 4601
    },
    {
      "epoch": 10.247216035634743,
      "loss": 0.42298391461372375,
      "loss_ce": 0.00013235666847322136,
      "loss_iou": 0.189453125,
      "loss_num": 0.008544921875,
      "loss_xval": 0.421875,
      "num_input_tokens_seen": 257719556,
      "step": 4601
    },
    {
      "epoch": 10.249443207126948,
      "grad_norm": 18.37726593017578,
      "learning_rate": 1e-06,
      "loss": 0.6735,
      "num_input_tokens_seen": 257774564,
      "step": 4602
    },
    {
      "epoch": 10.249443207126948,
      "loss": 0.4393607974052429,
      "loss_ce": 0.00015179984620772302,
      "loss_iou": 0.1923828125,
      "loss_num": 0.0108642578125,
      "loss_xval": 0.439453125,
      "num_input_tokens_seen": 257774564,
      "step": 4602
    },
    {
      "epoch": 10.251670378619155,
      "grad_norm": 13.660320281982422,
      "learning_rate": 1e-06,
      "loss": 0.4367,
      "num_input_tokens_seen": 257830240,
      "step": 4603
    },
    {
      "epoch": 10.251670378619155,
      "loss": 0.31909534335136414,
      "loss_ce": 0.0001256193791050464,
      "loss_iou": 0.1474609375,
      "loss_num": 0.00457763671875,
      "loss_xval": 0.318359375,
      "num_input_tokens_seen": 257830240,
      "step": 4603
    },
    {
      "epoch": 10.25389755011136,
      "grad_norm": 13.991618156433105,
      "learning_rate": 1e-06,
      "loss": 0.4515,
      "num_input_tokens_seen": 257883768,
      "step": 4604
    },
    {
      "epoch": 10.25389755011136,
      "loss": 0.39838650822639465,
      "loss_ce": 0.0001321271702181548,
      "loss_iou": 0.1806640625,
      "loss_num": 0.0074462890625,
      "loss_xval": 0.3984375,
      "num_input_tokens_seen": 257883768,
      "step": 4604
    },
    {
      "epoch": 10.256124721603564,
      "grad_norm": 15.985126495361328,
      "learning_rate": 1e-06,
      "loss": 0.6116,
      "num_input_tokens_seen": 257940856,
      "step": 4605
    },
    {
      "epoch": 10.256124721603564,
      "loss": 0.7079139947891235,
      "loss_ce": 0.0001502782106399536,
      "loss_iou": 0.296875,
      "loss_num": 0.022705078125,
      "loss_xval": 0.70703125,
      "num_input_tokens_seen": 257940856,
      "step": 4605
    },
    {
      "epoch": 10.25835189309577,
      "grad_norm": 17.861207962036133,
      "learning_rate": 1e-06,
      "loss": 0.6644,
      "num_input_tokens_seen": 257997188,
      "step": 4606
    },
    {
      "epoch": 10.25835189309577,
      "loss": 0.5625550150871277,
      "loss_ce": 0.00017705293430481106,
      "loss_iou": 0.232421875,
      "loss_num": 0.019287109375,
      "loss_xval": 0.5625,
      "num_input_tokens_seen": 257997188,
      "step": 4606
    },
    {
      "epoch": 10.260579064587974,
      "grad_norm": 16.24393653869629,
      "learning_rate": 1e-06,
      "loss": 0.3853,
      "num_input_tokens_seen": 258053824,
      "step": 4607
    },
    {
      "epoch": 10.260579064587974,
      "loss": 0.4628918766975403,
      "loss_ce": 0.00015384730068035424,
      "loss_iou": 0.201171875,
      "loss_num": 0.01214599609375,
      "loss_xval": 0.462890625,
      "num_input_tokens_seen": 258053824,
      "step": 4607
    },
    {
      "epoch": 10.262806236080179,
      "grad_norm": 20.257102966308594,
      "learning_rate": 1e-06,
      "loss": 0.4479,
      "num_input_tokens_seen": 258109152,
      "step": 4608
    },
    {
      "epoch": 10.262806236080179,
      "loss": 0.3674119710922241,
      "loss_ce": 0.00010239638504572213,
      "loss_iou": 0.1552734375,
      "loss_num": 0.0113525390625,
      "loss_xval": 0.3671875,
      "num_input_tokens_seen": 258109152,
      "step": 4608
    },
    {
      "epoch": 10.265033407572384,
      "grad_norm": 17.029348373413086,
      "learning_rate": 1e-06,
      "loss": 0.4576,
      "num_input_tokens_seen": 258166380,
      "step": 4609
    },
    {
      "epoch": 10.265033407572384,
      "loss": 0.4160352945327759,
      "loss_ce": 0.00014175890828482807,
      "loss_iou": 0.1904296875,
      "loss_num": 0.00714111328125,
      "loss_xval": 0.416015625,
      "num_input_tokens_seen": 258166380,
      "step": 4609
    },
    {
      "epoch": 10.267260579064589,
      "grad_norm": 21.380460739135742,
      "learning_rate": 1e-06,
      "loss": 0.5284,
      "num_input_tokens_seen": 258220732,
      "step": 4610
    },
    {
      "epoch": 10.267260579064589,
      "loss": 0.3585323393344879,
      "loss_ce": 0.00013390296953730285,
      "loss_iou": 0.1513671875,
      "loss_num": 0.0111083984375,
      "loss_xval": 0.359375,
      "num_input_tokens_seen": 258220732,
      "step": 4610
    },
    {
      "epoch": 10.269487750556793,
      "grad_norm": 14.722599983215332,
      "learning_rate": 1e-06,
      "loss": 0.58,
      "num_input_tokens_seen": 258277352,
      "step": 4611
    },
    {
      "epoch": 10.269487750556793,
      "loss": 0.4370768666267395,
      "loss_ce": 0.00018718844512477517,
      "loss_iou": 0.1953125,
      "loss_num": 0.00909423828125,
      "loss_xval": 0.4375,
      "num_input_tokens_seen": 258277352,
      "step": 4611
    },
    {
      "epoch": 10.271714922048998,
      "grad_norm": 21.076213836669922,
      "learning_rate": 1e-06,
      "loss": 0.5657,
      "num_input_tokens_seen": 258332080,
      "step": 4612
    },
    {
      "epoch": 10.271714922048998,
      "loss": 0.4650370478630066,
      "loss_ce": 0.0001932941668201238,
      "loss_iou": 0.2041015625,
      "loss_num": 0.01123046875,
      "loss_xval": 0.46484375,
      "num_input_tokens_seen": 258332080,
      "step": 4612
    },
    {
      "epoch": 10.273942093541203,
      "grad_norm": 85.08380126953125,
      "learning_rate": 1e-06,
      "loss": 0.6145,
      "num_input_tokens_seen": 258389684,
      "step": 4613
    },
    {
      "epoch": 10.273942093541203,
      "loss": 0.5986820459365845,
      "loss_ce": 0.0001713307574391365,
      "loss_iou": 0.251953125,
      "loss_num": 0.0186767578125,
      "loss_xval": 0.59765625,
      "num_input_tokens_seen": 258389684,
      "step": 4613
    },
    {
      "epoch": 10.276169265033408,
      "grad_norm": 32.875736236572266,
      "learning_rate": 1e-06,
      "loss": 0.6115,
      "num_input_tokens_seen": 258446596,
      "step": 4614
    },
    {
      "epoch": 10.276169265033408,
      "loss": 0.4154265522956848,
      "loss_ce": 0.00014332013961393386,
      "loss_iou": 0.1884765625,
      "loss_num": 0.007568359375,
      "loss_xval": 0.416015625,
      "num_input_tokens_seen": 258446596,
      "step": 4614
    },
    {
      "epoch": 10.278396436525613,
      "grad_norm": 25.337644577026367,
      "learning_rate": 1e-06,
      "loss": 0.6043,
      "num_input_tokens_seen": 258503504,
      "step": 4615
    },
    {
      "epoch": 10.278396436525613,
      "loss": 0.5890331268310547,
      "loss_ce": 0.00016596817295067012,
      "loss_iou": 0.2578125,
      "loss_num": 0.01434326171875,
      "loss_xval": 0.58984375,
      "num_input_tokens_seen": 258503504,
      "step": 4615
    },
    {
      "epoch": 10.280623608017818,
      "grad_norm": 22.491819381713867,
      "learning_rate": 1e-06,
      "loss": 0.4488,
      "num_input_tokens_seen": 258561264,
      "step": 4616
    },
    {
      "epoch": 10.280623608017818,
      "loss": 0.42713305354118347,
      "loss_ce": 0.00013111413863953203,
      "loss_iou": 0.1845703125,
      "loss_num": 0.011474609375,
      "loss_xval": 0.427734375,
      "num_input_tokens_seen": 258561264,
      "step": 4616
    },
    {
      "epoch": 10.282850779510023,
      "grad_norm": 16.053213119506836,
      "learning_rate": 1e-06,
      "loss": 0.6014,
      "num_input_tokens_seen": 258615720,
      "step": 4617
    },
    {
      "epoch": 10.282850779510023,
      "loss": 0.6224437952041626,
      "loss_ce": 0.00012936524581164122,
      "loss_iou": 0.2578125,
      "loss_num": 0.0213623046875,
      "loss_xval": 0.62109375,
      "num_input_tokens_seen": 258615720,
      "step": 4617
    },
    {
      "epoch": 10.285077951002227,
      "grad_norm": 21.939491271972656,
      "learning_rate": 1e-06,
      "loss": 0.447,
      "num_input_tokens_seen": 258671132,
      "step": 4618
    },
    {
      "epoch": 10.285077951002227,
      "loss": 0.33654463291168213,
      "loss_ce": 0.00011883860861416906,
      "loss_iou": 0.1474609375,
      "loss_num": 0.00830078125,
      "loss_xval": 0.3359375,
      "num_input_tokens_seen": 258671132,
      "step": 4618
    },
    {
      "epoch": 10.287305122494432,
      "grad_norm": 14.96814250946045,
      "learning_rate": 1e-06,
      "loss": 0.5002,
      "num_input_tokens_seen": 258728584,
      "step": 4619
    },
    {
      "epoch": 10.287305122494432,
      "loss": 0.5067825317382812,
      "loss_ce": 0.0002517920802347362,
      "loss_iou": 0.21875,
      "loss_num": 0.01385498046875,
      "loss_xval": 0.5078125,
      "num_input_tokens_seen": 258728584,
      "step": 4619
    },
    {
      "epoch": 10.289532293986637,
      "grad_norm": 17.478721618652344,
      "learning_rate": 1e-06,
      "loss": 0.3775,
      "num_input_tokens_seen": 258786092,
      "step": 4620
    },
    {
      "epoch": 10.289532293986637,
      "loss": 0.46181732416152954,
      "loss_ce": 0.00014738523168489337,
      "loss_iou": 0.203125,
      "loss_num": 0.01123046875,
      "loss_xval": 0.4609375,
      "num_input_tokens_seen": 258786092,
      "step": 4620
    },
    {
      "epoch": 10.291759465478842,
      "grad_norm": 22.687427520751953,
      "learning_rate": 1e-06,
      "loss": 0.5753,
      "num_input_tokens_seen": 258841324,
      "step": 4621
    },
    {
      "epoch": 10.291759465478842,
      "loss": 0.6752172708511353,
      "loss_ce": 0.00016845832578837872,
      "loss_iou": 0.3125,
      "loss_num": 0.00994873046875,
      "loss_xval": 0.67578125,
      "num_input_tokens_seen": 258841324,
      "step": 4621
    },
    {
      "epoch": 10.293986636971047,
      "grad_norm": 20.015989303588867,
      "learning_rate": 1e-06,
      "loss": 0.7111,
      "num_input_tokens_seen": 258895908,
      "step": 4622
    },
    {
      "epoch": 10.293986636971047,
      "loss": 0.5323803424835205,
      "loss_ce": 0.00015376918599940836,
      "loss_iou": 0.240234375,
      "loss_num": 0.0103759765625,
      "loss_xval": 0.53125,
      "num_input_tokens_seen": 258895908,
      "step": 4622
    },
    {
      "epoch": 10.296213808463252,
      "grad_norm": 19.902332305908203,
      "learning_rate": 1e-06,
      "loss": 0.3957,
      "num_input_tokens_seen": 258955124,
      "step": 4623
    },
    {
      "epoch": 10.296213808463252,
      "loss": 0.35737037658691406,
      "loss_ce": 0.0001926565309986472,
      "loss_iou": 0.1572265625,
      "loss_num": 0.0086669921875,
      "loss_xval": 0.357421875,
      "num_input_tokens_seen": 258955124,
      "step": 4623
    },
    {
      "epoch": 10.298440979955457,
      "grad_norm": 30.156911849975586,
      "learning_rate": 1e-06,
      "loss": 0.5562,
      "num_input_tokens_seen": 259010176,
      "step": 4624
    },
    {
      "epoch": 10.298440979955457,
      "loss": 0.5099859833717346,
      "loss_ce": 0.0002203439362347126,
      "loss_iou": 0.228515625,
      "loss_num": 0.01055908203125,
      "loss_xval": 0.5078125,
      "num_input_tokens_seen": 259010176,
      "step": 4624
    },
    {
      "epoch": 10.300668151447661,
      "grad_norm": 18.427066802978516,
      "learning_rate": 1e-06,
      "loss": 0.5903,
      "num_input_tokens_seen": 259065656,
      "step": 4625
    },
    {
      "epoch": 10.300668151447661,
      "loss": 0.5748733878135681,
      "loss_ce": 0.0001511350565124303,
      "loss_iou": 0.251953125,
      "loss_num": 0.014404296875,
      "loss_xval": 0.57421875,
      "num_input_tokens_seen": 259065656,
      "step": 4625
    },
    {
      "epoch": 10.302895322939866,
      "grad_norm": 16.115806579589844,
      "learning_rate": 1e-06,
      "loss": 0.4775,
      "num_input_tokens_seen": 259121088,
      "step": 4626
    },
    {
      "epoch": 10.302895322939866,
      "loss": 0.33922505378723145,
      "loss_ce": 0.0001137549479608424,
      "loss_iou": 0.1484375,
      "loss_num": 0.00860595703125,
      "loss_xval": 0.33984375,
      "num_input_tokens_seen": 259121088,
      "step": 4626
    },
    {
      "epoch": 10.305122494432071,
      "grad_norm": 21.94735336303711,
      "learning_rate": 1e-06,
      "loss": 0.6358,
      "num_input_tokens_seen": 259179752,
      "step": 4627
    },
    {
      "epoch": 10.305122494432071,
      "loss": 0.6556517481803894,
      "loss_ce": 0.0001341440947726369,
      "loss_iou": 0.279296875,
      "loss_num": 0.01904296875,
      "loss_xval": 0.65625,
      "num_input_tokens_seen": 259179752,
      "step": 4627
    },
    {
      "epoch": 10.307349665924276,
      "grad_norm": 18.129470825195312,
      "learning_rate": 1e-06,
      "loss": 0.5075,
      "num_input_tokens_seen": 259231148,
      "step": 4628
    },
    {
      "epoch": 10.307349665924276,
      "loss": 0.5028002858161926,
      "loss_ce": 0.00011475315841380507,
      "loss_iou": 0.2099609375,
      "loss_num": 0.0166015625,
      "loss_xval": 0.50390625,
      "num_input_tokens_seen": 259231148,
      "step": 4628
    },
    {
      "epoch": 10.309576837416481,
      "grad_norm": 16.93099021911621,
      "learning_rate": 1e-06,
      "loss": 0.5894,
      "num_input_tokens_seen": 259289500,
      "step": 4629
    },
    {
      "epoch": 10.309576837416481,
      "loss": 0.4424053430557251,
      "loss_ce": 0.00014459306839853525,
      "loss_iou": 0.203125,
      "loss_num": 0.007049560546875,
      "loss_xval": 0.44140625,
      "num_input_tokens_seen": 259289500,
      "step": 4629
    },
    {
      "epoch": 10.311804008908686,
      "grad_norm": 25.796844482421875,
      "learning_rate": 1e-06,
      "loss": 0.6209,
      "num_input_tokens_seen": 259346848,
      "step": 4630
    },
    {
      "epoch": 10.311804008908686,
      "loss": 0.6188187599182129,
      "loss_ce": 0.00016644690185785294,
      "loss_iou": 0.275390625,
      "loss_num": 0.01348876953125,
      "loss_xval": 0.6171875,
      "num_input_tokens_seen": 259346848,
      "step": 4630
    },
    {
      "epoch": 10.31403118040089,
      "grad_norm": 16.472164154052734,
      "learning_rate": 1e-06,
      "loss": 0.7976,
      "num_input_tokens_seen": 259404428,
      "step": 4631
    },
    {
      "epoch": 10.31403118040089,
      "loss": 0.7173308730125427,
      "loss_ce": 0.0002898685052059591,
      "loss_iou": 0.27734375,
      "loss_num": 0.032470703125,
      "loss_xval": 0.71875,
      "num_input_tokens_seen": 259404428,
      "step": 4631
    },
    {
      "epoch": 10.316258351893095,
      "grad_norm": 17.40803337097168,
      "learning_rate": 1e-06,
      "loss": 0.4205,
      "num_input_tokens_seen": 259460152,
      "step": 4632
    },
    {
      "epoch": 10.316258351893095,
      "loss": 0.3678438663482666,
      "loss_ce": 0.00016810771194286644,
      "loss_iou": 0.15625,
      "loss_num": 0.01104736328125,
      "loss_xval": 0.3671875,
      "num_input_tokens_seen": 259460152,
      "step": 4632
    },
    {
      "epoch": 10.3184855233853,
      "grad_norm": 17.019670486450195,
      "learning_rate": 1e-06,
      "loss": 0.6062,
      "num_input_tokens_seen": 259512592,
      "step": 4633
    },
    {
      "epoch": 10.3184855233853,
      "loss": 0.5542778968811035,
      "loss_ce": 0.00020081247203052044,
      "loss_iou": 0.2392578125,
      "loss_num": 0.01519775390625,
      "loss_xval": 0.5546875,
      "num_input_tokens_seen": 259512592,
      "step": 4633
    },
    {
      "epoch": 10.320712694877505,
      "grad_norm": 20.420970916748047,
      "learning_rate": 1e-06,
      "loss": 0.5013,
      "num_input_tokens_seen": 259569136,
      "step": 4634
    },
    {
      "epoch": 10.320712694877505,
      "loss": 0.5180292129516602,
      "loss_ce": 0.00020692471298389137,
      "loss_iou": 0.23046875,
      "loss_num": 0.01141357421875,
      "loss_xval": 0.51953125,
      "num_input_tokens_seen": 259569136,
      "step": 4634
    },
    {
      "epoch": 10.32293986636971,
      "grad_norm": 18.733917236328125,
      "learning_rate": 1e-06,
      "loss": 0.6095,
      "num_input_tokens_seen": 259626308,
      "step": 4635
    },
    {
      "epoch": 10.32293986636971,
      "loss": 0.5899757742881775,
      "loss_ce": 0.00013202980335336179,
      "loss_iou": 0.26171875,
      "loss_num": 0.0133056640625,
      "loss_xval": 0.58984375,
      "num_input_tokens_seen": 259626308,
      "step": 4635
    },
    {
      "epoch": 10.325167037861915,
      "grad_norm": 15.864693641662598,
      "learning_rate": 1e-06,
      "loss": 0.4543,
      "num_input_tokens_seen": 259683580,
      "step": 4636
    },
    {
      "epoch": 10.325167037861915,
      "loss": 0.36647483706474304,
      "loss_ce": 0.00014181638834998012,
      "loss_iou": 0.169921875,
      "loss_num": 0.0052490234375,
      "loss_xval": 0.3671875,
      "num_input_tokens_seen": 259683580,
      "step": 4636
    },
    {
      "epoch": 10.32739420935412,
      "grad_norm": 16.95244789123535,
      "learning_rate": 1e-06,
      "loss": 0.4863,
      "num_input_tokens_seen": 259740856,
      "step": 4637
    },
    {
      "epoch": 10.32739420935412,
      "loss": 0.4233607351779938,
      "loss_ce": 0.00014292271225713193,
      "loss_iou": 0.1826171875,
      "loss_num": 0.01165771484375,
      "loss_xval": 0.423828125,
      "num_input_tokens_seen": 259740856,
      "step": 4637
    },
    {
      "epoch": 10.329621380846325,
      "grad_norm": 14.428168296813965,
      "learning_rate": 1e-06,
      "loss": 0.5452,
      "num_input_tokens_seen": 259797892,
      "step": 4638
    },
    {
      "epoch": 10.329621380846325,
      "loss": 0.37976911664009094,
      "loss_ce": 0.00013044924708083272,
      "loss_iou": 0.173828125,
      "loss_num": 0.0064697265625,
      "loss_xval": 0.37890625,
      "num_input_tokens_seen": 259797892,
      "step": 4638
    },
    {
      "epoch": 10.33184855233853,
      "grad_norm": 20.9561767578125,
      "learning_rate": 1e-06,
      "loss": 0.5536,
      "num_input_tokens_seen": 259856484,
      "step": 4639
    },
    {
      "epoch": 10.33184855233853,
      "loss": 0.4375172257423401,
      "loss_ce": 0.00013931245484855026,
      "loss_iou": 0.19140625,
      "loss_num": 0.01068115234375,
      "loss_xval": 0.4375,
      "num_input_tokens_seen": 259856484,
      "step": 4639
    },
    {
      "epoch": 10.334075723830734,
      "grad_norm": 15.651261329650879,
      "learning_rate": 1e-06,
      "loss": 0.4453,
      "num_input_tokens_seen": 259916208,
      "step": 4640
    },
    {
      "epoch": 10.334075723830734,
      "loss": 0.548733115196228,
      "loss_ce": 0.00014910154277458787,
      "loss_iou": 0.2236328125,
      "loss_num": 0.0201416015625,
      "loss_xval": 0.546875,
      "num_input_tokens_seen": 259916208,
      "step": 4640
    },
    {
      "epoch": 10.33630289532294,
      "grad_norm": 13.660500526428223,
      "learning_rate": 1e-06,
      "loss": 0.4894,
      "num_input_tokens_seen": 259971708,
      "step": 4641
    },
    {
      "epoch": 10.33630289532294,
      "loss": 0.45058485865592957,
      "loss_ce": 0.000511608668603003,
      "loss_iou": 0.15234375,
      "loss_num": 0.02880859375,
      "loss_xval": 0.44921875,
      "num_input_tokens_seen": 259971708,
      "step": 4641
    },
    {
      "epoch": 10.338530066815144,
      "grad_norm": 25.716596603393555,
      "learning_rate": 1e-06,
      "loss": 0.5956,
      "num_input_tokens_seen": 260025852,
      "step": 4642
    },
    {
      "epoch": 10.338530066815144,
      "loss": 0.7315921783447266,
      "loss_ce": 0.00039098679553717375,
      "loss_iou": 0.298828125,
      "loss_num": 0.026611328125,
      "loss_xval": 0.73046875,
      "num_input_tokens_seen": 260025852,
      "step": 4642
    },
    {
      "epoch": 10.340757238307349,
      "grad_norm": 21.653146743774414,
      "learning_rate": 1e-06,
      "loss": 0.5072,
      "num_input_tokens_seen": 260079956,
      "step": 4643
    },
    {
      "epoch": 10.340757238307349,
      "loss": 0.434053510427475,
      "loss_ce": 0.0002766597317531705,
      "loss_iou": 0.1884765625,
      "loss_num": 0.0113525390625,
      "loss_xval": 0.43359375,
      "num_input_tokens_seen": 260079956,
      "step": 4643
    },
    {
      "epoch": 10.342984409799554,
      "grad_norm": 14.551214218139648,
      "learning_rate": 1e-06,
      "loss": 0.4545,
      "num_input_tokens_seen": 260136624,
      "step": 4644
    },
    {
      "epoch": 10.342984409799554,
      "loss": 0.5187681913375854,
      "loss_ce": 0.00021346815628930926,
      "loss_iou": 0.2294921875,
      "loss_num": 0.01177978515625,
      "loss_xval": 0.51953125,
      "num_input_tokens_seen": 260136624,
      "step": 4644
    },
    {
      "epoch": 10.345211581291759,
      "grad_norm": 23.34543228149414,
      "learning_rate": 1e-06,
      "loss": 0.5684,
      "num_input_tokens_seen": 260193936,
      "step": 4645
    },
    {
      "epoch": 10.345211581291759,
      "loss": 0.6667625904083252,
      "loss_ce": 0.00013660687545780092,
      "loss_iou": 0.30078125,
      "loss_num": 0.0128173828125,
      "loss_xval": 0.66796875,
      "num_input_tokens_seen": 260193936,
      "step": 4645
    },
    {
      "epoch": 10.347438752783964,
      "grad_norm": 15.129518508911133,
      "learning_rate": 1e-06,
      "loss": 0.4025,
      "num_input_tokens_seen": 260247704,
      "step": 4646
    },
    {
      "epoch": 10.347438752783964,
      "loss": 0.3118034601211548,
      "loss_ce": 0.00012743064144160599,
      "loss_iou": 0.130859375,
      "loss_num": 0.0101318359375,
      "loss_xval": 0.3125,
      "num_input_tokens_seen": 260247704,
      "step": 4646
    },
    {
      "epoch": 10.34966592427617,
      "grad_norm": 20.0349063873291,
      "learning_rate": 1e-06,
      "loss": 0.8892,
      "num_input_tokens_seen": 260305780,
      "step": 4647
    },
    {
      "epoch": 10.34966592427617,
      "loss": 0.9932963252067566,
      "loss_ce": 0.00013225735165178776,
      "loss_iou": 0.43359375,
      "loss_num": 0.02490234375,
      "loss_xval": 0.9921875,
      "num_input_tokens_seen": 260305780,
      "step": 4647
    },
    {
      "epoch": 10.351893095768375,
      "grad_norm": 24.84522247314453,
      "learning_rate": 1e-06,
      "loss": 0.6352,
      "num_input_tokens_seen": 260360704,
      "step": 4648
    },
    {
      "epoch": 10.351893095768375,
      "loss": 0.5557194352149963,
      "loss_ce": 0.0001774309203028679,
      "loss_iou": 0.25390625,
      "loss_num": 0.00994873046875,
      "loss_xval": 0.5546875,
      "num_input_tokens_seen": 260360704,
      "step": 4648
    },
    {
      "epoch": 10.35412026726058,
      "grad_norm": 16.85805892944336,
      "learning_rate": 1e-06,
      "loss": 0.5613,
      "num_input_tokens_seen": 260419212,
      "step": 4649
    },
    {
      "epoch": 10.35412026726058,
      "loss": 0.43383756279945374,
      "loss_ce": 0.00012176090240245685,
      "loss_iou": 0.173828125,
      "loss_num": 0.0169677734375,
      "loss_xval": 0.43359375,
      "num_input_tokens_seen": 260419212,
      "step": 4649
    },
    {
      "epoch": 10.356347438752785,
      "grad_norm": 19.93196678161621,
      "learning_rate": 1e-06,
      "loss": 0.5456,
      "num_input_tokens_seen": 260475968,
      "step": 4650
    },
    {
      "epoch": 10.356347438752785,
      "loss": 0.5350778698921204,
      "loss_ce": 0.00016572429740335792,
      "loss_iou": 0.2294921875,
      "loss_num": 0.01519775390625,
      "loss_xval": 0.53515625,
      "num_input_tokens_seen": 260475968,
      "step": 4650
    },
    {
      "epoch": 10.35857461024499,
      "grad_norm": 22.55901336669922,
      "learning_rate": 1e-06,
      "loss": 0.5437,
      "num_input_tokens_seen": 260530128,
      "step": 4651
    },
    {
      "epoch": 10.35857461024499,
      "loss": 0.6019728183746338,
      "loss_ce": 0.0001661243732087314,
      "loss_iou": 0.28125,
      "loss_num": 0.00799560546875,
      "loss_xval": 0.6015625,
      "num_input_tokens_seen": 260530128,
      "step": 4651
    },
    {
      "epoch": 10.360801781737194,
      "grad_norm": 24.937402725219727,
      "learning_rate": 1e-06,
      "loss": 0.5643,
      "num_input_tokens_seen": 260587396,
      "step": 4652
    },
    {
      "epoch": 10.360801781737194,
      "loss": 0.4734136760234833,
      "loss_ce": 0.00014707804075442255,
      "loss_iou": 0.2099609375,
      "loss_num": 0.01068115234375,
      "loss_xval": 0.47265625,
      "num_input_tokens_seen": 260587396,
      "step": 4652
    },
    {
      "epoch": 10.3630289532294,
      "grad_norm": 13.982348442077637,
      "learning_rate": 1e-06,
      "loss": 0.4673,
      "num_input_tokens_seen": 260644020,
      "step": 4653
    },
    {
      "epoch": 10.3630289532294,
      "loss": 0.3794419467449188,
      "loss_ce": 0.0004746583290398121,
      "loss_iou": 0.1689453125,
      "loss_num": 0.00823974609375,
      "loss_xval": 0.37890625,
      "num_input_tokens_seen": 260644020,
      "step": 4653
    },
    {
      "epoch": 10.365256124721604,
      "grad_norm": 16.844064712524414,
      "learning_rate": 1e-06,
      "loss": 0.6486,
      "num_input_tokens_seen": 260700716,
      "step": 4654
    },
    {
      "epoch": 10.365256124721604,
      "loss": 0.5358029007911682,
      "loss_ce": 0.00015838223043829203,
      "loss_iou": 0.244140625,
      "loss_num": 0.00946044921875,
      "loss_xval": 0.53515625,
      "num_input_tokens_seen": 260700716,
      "step": 4654
    },
    {
      "epoch": 10.367483296213809,
      "grad_norm": 16.441743850708008,
      "learning_rate": 1e-06,
      "loss": 0.6257,
      "num_input_tokens_seen": 260756304,
      "step": 4655
    },
    {
      "epoch": 10.367483296213809,
      "loss": 0.6969484090805054,
      "loss_ce": 0.00017108957399614155,
      "loss_iou": 0.29296875,
      "loss_num": 0.0228271484375,
      "loss_xval": 0.6953125,
      "num_input_tokens_seen": 260756304,
      "step": 4655
    },
    {
      "epoch": 10.369710467706014,
      "grad_norm": 16.78958511352539,
      "learning_rate": 1e-06,
      "loss": 0.6998,
      "num_input_tokens_seen": 260810784,
      "step": 4656
    },
    {
      "epoch": 10.369710467706014,
      "loss": 0.6009594798088074,
      "loss_ce": 0.00012937135761603713,
      "loss_iou": 0.251953125,
      "loss_num": 0.01904296875,
      "loss_xval": 0.6015625,
      "num_input_tokens_seen": 260810784,
      "step": 4656
    },
    {
      "epoch": 10.371937639198219,
      "grad_norm": 18.355632781982422,
      "learning_rate": 1e-06,
      "loss": 0.5985,
      "num_input_tokens_seen": 260867540,
      "step": 4657
    },
    {
      "epoch": 10.371937639198219,
      "loss": 0.7768846750259399,
      "loss_ce": 0.00015131058171391487,
      "loss_iou": 0.333984375,
      "loss_num": 0.021728515625,
      "loss_xval": 0.77734375,
      "num_input_tokens_seen": 260867540,
      "step": 4657
    },
    {
      "epoch": 10.374164810690424,
      "grad_norm": 17.681825637817383,
      "learning_rate": 1e-06,
      "loss": 0.5509,
      "num_input_tokens_seen": 260923584,
      "step": 4658
    },
    {
      "epoch": 10.374164810690424,
      "loss": 0.5139190554618835,
      "loss_ce": 0.00012508760846685618,
      "loss_iou": 0.21875,
      "loss_num": 0.01513671875,
      "loss_xval": 0.515625,
      "num_input_tokens_seen": 260923584,
      "step": 4658
    },
    {
      "epoch": 10.376391982182628,
      "grad_norm": 29.6978759765625,
      "learning_rate": 1e-06,
      "loss": 0.549,
      "num_input_tokens_seen": 260980476,
      "step": 4659
    },
    {
      "epoch": 10.376391982182628,
      "loss": 0.6463862657546997,
      "loss_ce": 0.00014600764552596956,
      "loss_iou": 0.2734375,
      "loss_num": 0.02001953125,
      "loss_xval": 0.64453125,
      "num_input_tokens_seen": 260980476,
      "step": 4659
    },
    {
      "epoch": 10.378619153674833,
      "grad_norm": 20.990150451660156,
      "learning_rate": 1e-06,
      "loss": 0.4872,
      "num_input_tokens_seen": 261038064,
      "step": 4660
    },
    {
      "epoch": 10.378619153674833,
      "loss": 0.5661933422088623,
      "loss_ce": 0.00015332447947002947,
      "loss_iou": 0.2578125,
      "loss_num": 0.00994873046875,
      "loss_xval": 0.56640625,
      "num_input_tokens_seen": 261038064,
      "step": 4660
    },
    {
      "epoch": 10.380846325167038,
      "grad_norm": 20.679819107055664,
      "learning_rate": 1e-06,
      "loss": 0.4146,
      "num_input_tokens_seen": 261093180,
      "step": 4661
    },
    {
      "epoch": 10.380846325167038,
      "loss": 0.479025661945343,
      "loss_ce": 0.00014384492533281446,
      "loss_iou": 0.2138671875,
      "loss_num": 0.010009765625,
      "loss_xval": 0.478515625,
      "num_input_tokens_seen": 261093180,
      "step": 4661
    },
    {
      "epoch": 10.383073496659243,
      "grad_norm": 12.441951751708984,
      "learning_rate": 1e-06,
      "loss": 0.5305,
      "num_input_tokens_seen": 261149604,
      "step": 4662
    },
    {
      "epoch": 10.383073496659243,
      "loss": 0.5029485821723938,
      "loss_ce": 0.00014094685320742428,
      "loss_iou": 0.2275390625,
      "loss_num": 0.0096435546875,
      "loss_xval": 0.50390625,
      "num_input_tokens_seen": 261149604,
      "step": 4662
    },
    {
      "epoch": 10.385300668151448,
      "grad_norm": 39.21706008911133,
      "learning_rate": 1e-06,
      "loss": 0.5123,
      "num_input_tokens_seen": 261207228,
      "step": 4663
    },
    {
      "epoch": 10.385300668151448,
      "loss": 0.44520843029022217,
      "loss_ce": 0.00014007699792273343,
      "loss_iou": 0.2021484375,
      "loss_num": 0.00836181640625,
      "loss_xval": 0.4453125,
      "num_input_tokens_seen": 261207228,
      "step": 4663
    },
    {
      "epoch": 10.387527839643653,
      "grad_norm": 29.102785110473633,
      "learning_rate": 1e-06,
      "loss": 0.5844,
      "num_input_tokens_seen": 261261652,
      "step": 4664
    },
    {
      "epoch": 10.387527839643653,
      "loss": 0.523324191570282,
      "loss_ce": 0.00013084019883535802,
      "loss_iou": 0.2294921875,
      "loss_num": 0.01300048828125,
      "loss_xval": 0.5234375,
      "num_input_tokens_seen": 261261652,
      "step": 4664
    },
    {
      "epoch": 10.389755011135858,
      "grad_norm": 17.95996856689453,
      "learning_rate": 1e-06,
      "loss": 0.5083,
      "num_input_tokens_seen": 261314420,
      "step": 4665
    },
    {
      "epoch": 10.389755011135858,
      "loss": 0.666547417640686,
      "loss_ce": 0.00016560900257900357,
      "loss_iou": 0.3046875,
      "loss_num": 0.01141357421875,
      "loss_xval": 0.66796875,
      "num_input_tokens_seen": 261314420,
      "step": 4665
    },
    {
      "epoch": 10.391982182628063,
      "grad_norm": 17.621519088745117,
      "learning_rate": 1e-06,
      "loss": 0.4285,
      "num_input_tokens_seen": 261369500,
      "step": 4666
    },
    {
      "epoch": 10.391982182628063,
      "loss": 0.3165377378463745,
      "loss_ce": 0.0001314996334258467,
      "loss_iou": 0.13671875,
      "loss_num": 0.00836181640625,
      "loss_xval": 0.31640625,
      "num_input_tokens_seen": 261369500,
      "step": 4666
    },
    {
      "epoch": 10.394209354120267,
      "grad_norm": 28.80778694152832,
      "learning_rate": 1e-06,
      "loss": 0.6926,
      "num_input_tokens_seen": 261424900,
      "step": 4667
    },
    {
      "epoch": 10.394209354120267,
      "loss": 0.6756924390792847,
      "loss_ce": 0.00015533142141066492,
      "loss_iou": 0.3046875,
      "loss_num": 0.01336669921875,
      "loss_xval": 0.67578125,
      "num_input_tokens_seen": 261424900,
      "step": 4667
    },
    {
      "epoch": 10.396436525612472,
      "grad_norm": 16.814796447753906,
      "learning_rate": 1e-06,
      "loss": 0.6278,
      "num_input_tokens_seen": 261481180,
      "step": 4668
    },
    {
      "epoch": 10.396436525612472,
      "loss": 0.6310012340545654,
      "loss_ce": 0.00020286736253183335,
      "loss_iou": 0.259765625,
      "loss_num": 0.0220947265625,
      "loss_xval": 0.62890625,
      "num_input_tokens_seen": 261481180,
      "step": 4668
    },
    {
      "epoch": 10.398663697104677,
      "grad_norm": 28.424175262451172,
      "learning_rate": 1e-06,
      "loss": 0.5187,
      "num_input_tokens_seen": 261537668,
      "step": 4669
    },
    {
      "epoch": 10.398663697104677,
      "loss": 0.7215948104858398,
      "loss_ce": 0.00015925764455460012,
      "loss_iou": 0.296875,
      "loss_num": 0.0252685546875,
      "loss_xval": 0.72265625,
      "num_input_tokens_seen": 261537668,
      "step": 4669
    },
    {
      "epoch": 10.400890868596882,
      "grad_norm": 42.25555419921875,
      "learning_rate": 1e-06,
      "loss": 0.5724,
      "num_input_tokens_seen": 261592388,
      "step": 4670
    },
    {
      "epoch": 10.400890868596882,
      "loss": 0.532261848449707,
      "loss_ce": 0.00015735380293335766,
      "loss_iou": 0.23046875,
      "loss_num": 0.01416015625,
      "loss_xval": 0.53125,
      "num_input_tokens_seen": 261592388,
      "step": 4670
    },
    {
      "epoch": 10.403118040089087,
      "grad_norm": 15.451391220092773,
      "learning_rate": 1e-06,
      "loss": 0.6405,
      "num_input_tokens_seen": 261649236,
      "step": 4671
    },
    {
      "epoch": 10.403118040089087,
      "loss": 0.8438804149627686,
      "loss_ce": 0.000130399945192039,
      "loss_iou": 0.337890625,
      "loss_num": 0.033203125,
      "loss_xval": 0.84375,
      "num_input_tokens_seen": 261649236,
      "step": 4671
    },
    {
      "epoch": 10.405345211581292,
      "grad_norm": 18.49277687072754,
      "learning_rate": 1e-06,
      "loss": 0.7358,
      "num_input_tokens_seen": 261705516,
      "step": 4672
    },
    {
      "epoch": 10.405345211581292,
      "loss": 0.5958718061447144,
      "loss_ce": 0.0001686769537627697,
      "loss_iou": 0.26171875,
      "loss_num": 0.01446533203125,
      "loss_xval": 0.59375,
      "num_input_tokens_seen": 261705516,
      "step": 4672
    },
    {
      "epoch": 10.407572383073497,
      "grad_norm": 18.64063262939453,
      "learning_rate": 1e-06,
      "loss": 0.5341,
      "num_input_tokens_seen": 261759996,
      "step": 4673
    },
    {
      "epoch": 10.407572383073497,
      "loss": 0.4139430522918701,
      "loss_ce": 0.00012470106594264507,
      "loss_iou": 0.169921875,
      "loss_num": 0.01483154296875,
      "loss_xval": 0.4140625,
      "num_input_tokens_seen": 261759996,
      "step": 4673
    },
    {
      "epoch": 10.409799554565701,
      "grad_norm": 15.402804374694824,
      "learning_rate": 1e-06,
      "loss": 0.5624,
      "num_input_tokens_seen": 261814640,
      "step": 4674
    },
    {
      "epoch": 10.409799554565701,
      "loss": 0.5486302375793457,
      "loss_ce": 0.00016830695676617324,
      "loss_iou": 0.22265625,
      "loss_num": 0.0205078125,
      "loss_xval": 0.546875,
      "num_input_tokens_seen": 261814640,
      "step": 4674
    },
    {
      "epoch": 10.412026726057906,
      "grad_norm": 16.70870590209961,
      "learning_rate": 1e-06,
      "loss": 0.4344,
      "num_input_tokens_seen": 261873280,
      "step": 4675
    },
    {
      "epoch": 10.412026726057906,
      "loss": 0.26441287994384766,
      "loss_ce": 0.00013062989455647767,
      "loss_iou": 0.1162109375,
      "loss_num": 0.00634765625,
      "loss_xval": 0.263671875,
      "num_input_tokens_seen": 261873280,
      "step": 4675
    },
    {
      "epoch": 10.414253897550111,
      "grad_norm": 16.716644287109375,
      "learning_rate": 1e-06,
      "loss": 0.7616,
      "num_input_tokens_seen": 261930832,
      "step": 4676
    },
    {
      "epoch": 10.414253897550111,
      "loss": 0.6276519298553467,
      "loss_ce": 0.00021053958334960043,
      "loss_iou": 0.263671875,
      "loss_num": 0.0196533203125,
      "loss_xval": 0.62890625,
      "num_input_tokens_seen": 261930832,
      "step": 4676
    },
    {
      "epoch": 10.416481069042316,
      "grad_norm": 24.789369583129883,
      "learning_rate": 1e-06,
      "loss": 0.5369,
      "num_input_tokens_seen": 261985440,
      "step": 4677
    },
    {
      "epoch": 10.416481069042316,
      "loss": 0.5724594593048096,
      "loss_ce": 0.00019378411525394768,
      "loss_iou": 0.255859375,
      "loss_num": 0.01220703125,
      "loss_xval": 0.5703125,
      "num_input_tokens_seen": 261985440,
      "step": 4677
    },
    {
      "epoch": 10.41870824053452,
      "grad_norm": 23.978904724121094,
      "learning_rate": 1e-06,
      "loss": 0.5775,
      "num_input_tokens_seen": 262040176,
      "step": 4678
    },
    {
      "epoch": 10.41870824053452,
      "loss": 0.7208090424537659,
      "loss_ce": 0.0001669653574936092,
      "loss_iou": 0.291015625,
      "loss_num": 0.02783203125,
      "loss_xval": 0.71875,
      "num_input_tokens_seen": 262040176,
      "step": 4678
    },
    {
      "epoch": 10.420935412026726,
      "grad_norm": 21.936405181884766,
      "learning_rate": 1e-06,
      "loss": 0.5377,
      "num_input_tokens_seen": 262096712,
      "step": 4679
    },
    {
      "epoch": 10.420935412026726,
      "loss": 0.6339254379272461,
      "loss_ce": 0.0001363815099466592,
      "loss_iou": 0.27734375,
      "loss_num": 0.0157470703125,
      "loss_xval": 0.6328125,
      "num_input_tokens_seen": 262096712,
      "step": 4679
    },
    {
      "epoch": 10.42316258351893,
      "grad_norm": 15.21117115020752,
      "learning_rate": 1e-06,
      "loss": 0.5304,
      "num_input_tokens_seen": 262151328,
      "step": 4680
    },
    {
      "epoch": 10.42316258351893,
      "loss": 0.6026941537857056,
      "loss_ce": 0.00015511347737628967,
      "loss_iou": 0.2470703125,
      "loss_num": 0.021728515625,
      "loss_xval": 0.6015625,
      "num_input_tokens_seen": 262151328,
      "step": 4680
    },
    {
      "epoch": 10.425389755011135,
      "grad_norm": 19.147802352905273,
      "learning_rate": 1e-06,
      "loss": 0.6639,
      "num_input_tokens_seen": 262207744,
      "step": 4681
    },
    {
      "epoch": 10.425389755011135,
      "loss": 0.6351706981658936,
      "loss_ce": 0.00016094453167170286,
      "loss_iou": 0.28125,
      "loss_num": 0.01416015625,
      "loss_xval": 0.63671875,
      "num_input_tokens_seen": 262207744,
      "step": 4681
    },
    {
      "epoch": 10.42761692650334,
      "grad_norm": 20.792253494262695,
      "learning_rate": 1e-06,
      "loss": 0.5855,
      "num_input_tokens_seen": 262266176,
      "step": 4682
    },
    {
      "epoch": 10.42761692650334,
      "loss": 0.7266117930412292,
      "loss_ce": 0.00017138014663942158,
      "loss_iou": 0.30859375,
      "loss_num": 0.0218505859375,
      "loss_xval": 0.7265625,
      "num_input_tokens_seen": 262266176,
      "step": 4682
    },
    {
      "epoch": 10.429844097995545,
      "grad_norm": 17.91155433654785,
      "learning_rate": 1e-06,
      "loss": 0.4326,
      "num_input_tokens_seen": 262322384,
      "step": 4683
    },
    {
      "epoch": 10.429844097995545,
      "loss": 0.4974979758262634,
      "loss_ce": 0.00018352872575633228,
      "loss_iou": 0.2158203125,
      "loss_num": 0.0130615234375,
      "loss_xval": 0.498046875,
      "num_input_tokens_seen": 262322384,
      "step": 4683
    },
    {
      "epoch": 10.43207126948775,
      "grad_norm": 15.88755989074707,
      "learning_rate": 1e-06,
      "loss": 0.4703,
      "num_input_tokens_seen": 262378432,
      "step": 4684
    },
    {
      "epoch": 10.43207126948775,
      "loss": 0.6226903200149536,
      "loss_ce": 0.0001317547430517152,
      "loss_iou": 0.27734375,
      "loss_num": 0.013427734375,
      "loss_xval": 0.62109375,
      "num_input_tokens_seen": 262378432,
      "step": 4684
    },
    {
      "epoch": 10.434298440979955,
      "grad_norm": 27.248088836669922,
      "learning_rate": 1e-06,
      "loss": 0.6203,
      "num_input_tokens_seen": 262433420,
      "step": 4685
    },
    {
      "epoch": 10.434298440979955,
      "loss": 0.8291926980018616,
      "loss_ce": 0.000579414947424084,
      "loss_iou": 0.3359375,
      "loss_num": 0.031494140625,
      "loss_xval": 0.828125,
      "num_input_tokens_seen": 262433420,
      "step": 4685
    },
    {
      "epoch": 10.43652561247216,
      "grad_norm": 23.795869827270508,
      "learning_rate": 1e-06,
      "loss": 0.4751,
      "num_input_tokens_seen": 262492420,
      "step": 4686
    },
    {
      "epoch": 10.43652561247216,
      "loss": 0.5592020750045776,
      "loss_ce": 0.0001200107071781531,
      "loss_iou": 0.255859375,
      "loss_num": 0.0096435546875,
      "loss_xval": 0.55859375,
      "num_input_tokens_seen": 262492420,
      "step": 4686
    },
    {
      "epoch": 10.438752783964365,
      "grad_norm": 119.42134857177734,
      "learning_rate": 1e-06,
      "loss": 0.4158,
      "num_input_tokens_seen": 262547884,
      "step": 4687
    },
    {
      "epoch": 10.438752783964365,
      "loss": 0.2656348645687103,
      "loss_ce": 0.0001319482980761677,
      "loss_iou": 0.111328125,
      "loss_num": 0.00860595703125,
      "loss_xval": 0.265625,
      "num_input_tokens_seen": 262547884,
      "step": 4687
    },
    {
      "epoch": 10.44097995545657,
      "grad_norm": 29.43088722229004,
      "learning_rate": 1e-06,
      "loss": 0.6806,
      "num_input_tokens_seen": 262605912,
      "step": 4688
    },
    {
      "epoch": 10.44097995545657,
      "loss": 0.7944400310516357,
      "loss_ce": 0.0001285005419049412,
      "loss_iou": 0.35546875,
      "loss_num": 0.0167236328125,
      "loss_xval": 0.79296875,
      "num_input_tokens_seen": 262605912,
      "step": 4688
    },
    {
      "epoch": 10.443207126948774,
      "grad_norm": 37.14841842651367,
      "learning_rate": 1e-06,
      "loss": 0.5946,
      "num_input_tokens_seen": 262658560,
      "step": 4689
    },
    {
      "epoch": 10.443207126948774,
      "loss": 0.5586202144622803,
      "loss_ce": 0.00014853276661597192,
      "loss_iou": 0.2216796875,
      "loss_num": 0.02294921875,
      "loss_xval": 0.55859375,
      "num_input_tokens_seen": 262658560,
      "step": 4689
    },
    {
      "epoch": 10.44543429844098,
      "grad_norm": 97.54398345947266,
      "learning_rate": 1e-06,
      "loss": 0.3942,
      "num_input_tokens_seen": 262714528,
      "step": 4690
    },
    {
      "epoch": 10.44543429844098,
      "loss": 0.3732898235321045,
      "loss_ce": 0.00012086871720384806,
      "loss_iou": 0.1591796875,
      "loss_num": 0.010986328125,
      "loss_xval": 0.373046875,
      "num_input_tokens_seen": 262714528,
      "step": 4690
    },
    {
      "epoch": 10.447661469933184,
      "grad_norm": 20.17695426940918,
      "learning_rate": 1e-06,
      "loss": 0.492,
      "num_input_tokens_seen": 262770576,
      "step": 4691
    },
    {
      "epoch": 10.447661469933184,
      "loss": 0.47415345907211304,
      "loss_ce": 0.00015445941244252026,
      "loss_iou": 0.20703125,
      "loss_num": 0.011962890625,
      "loss_xval": 0.474609375,
      "num_input_tokens_seen": 262770576,
      "step": 4691
    },
    {
      "epoch": 10.449888641425389,
      "grad_norm": 21.019207000732422,
      "learning_rate": 1e-06,
      "loss": 0.4976,
      "num_input_tokens_seen": 262827748,
      "step": 4692
    },
    {
      "epoch": 10.449888641425389,
      "loss": 0.44643697142601013,
      "loss_ce": 0.00014791959256399423,
      "loss_iou": 0.1806640625,
      "loss_num": 0.01708984375,
      "loss_xval": 0.4453125,
      "num_input_tokens_seen": 262827748,
      "step": 4692
    },
    {
      "epoch": 10.452115812917596,
      "grad_norm": 17.917194366455078,
      "learning_rate": 1e-06,
      "loss": 0.5386,
      "num_input_tokens_seen": 262884516,
      "step": 4693
    },
    {
      "epoch": 10.452115812917596,
      "loss": 0.40125101804733276,
      "loss_ce": 0.00012796126247849315,
      "loss_iou": 0.18359375,
      "loss_num": 0.00677490234375,
      "loss_xval": 0.400390625,
      "num_input_tokens_seen": 262884516,
      "step": 4693
    },
    {
      "epoch": 10.4543429844098,
      "grad_norm": 25.924360275268555,
      "learning_rate": 1e-06,
      "loss": 0.5668,
      "num_input_tokens_seen": 262939420,
      "step": 4694
    },
    {
      "epoch": 10.4543429844098,
      "loss": 0.43259525299072266,
      "loss_ce": 0.00022220781829673797,
      "loss_iou": 0.1845703125,
      "loss_num": 0.012451171875,
      "loss_xval": 0.431640625,
      "num_input_tokens_seen": 262939420,
      "step": 4694
    },
    {
      "epoch": 10.456570155902005,
      "grad_norm": 29.25006675720215,
      "learning_rate": 1e-06,
      "loss": 0.6523,
      "num_input_tokens_seen": 262998928,
      "step": 4695
    },
    {
      "epoch": 10.456570155902005,
      "loss": 0.7494416236877441,
      "loss_ce": 0.00017405254766345024,
      "loss_iou": 0.33203125,
      "loss_num": 0.0167236328125,
      "loss_xval": 0.75,
      "num_input_tokens_seen": 262998928,
      "step": 4695
    },
    {
      "epoch": 10.45879732739421,
      "grad_norm": 14.804512023925781,
      "learning_rate": 1e-06,
      "loss": 0.5316,
      "num_input_tokens_seen": 263055744,
      "step": 4696
    },
    {
      "epoch": 10.45879732739421,
      "loss": 0.5465176105499268,
      "loss_ce": 0.0001309234939981252,
      "loss_iou": 0.236328125,
      "loss_num": 0.014892578125,
      "loss_xval": 0.546875,
      "num_input_tokens_seen": 263055744,
      "step": 4696
    },
    {
      "epoch": 10.461024498886415,
      "grad_norm": 23.423185348510742,
      "learning_rate": 1e-06,
      "loss": 0.4285,
      "num_input_tokens_seen": 263111136,
      "step": 4697
    },
    {
      "epoch": 10.461024498886415,
      "loss": 0.35573017597198486,
      "loss_ce": 0.00013935507740825415,
      "loss_iou": 0.162109375,
      "loss_num": 0.00616455078125,
      "loss_xval": 0.35546875,
      "num_input_tokens_seen": 263111136,
      "step": 4697
    },
    {
      "epoch": 10.46325167037862,
      "grad_norm": 15.791742324829102,
      "learning_rate": 1e-06,
      "loss": 0.5727,
      "num_input_tokens_seen": 263168172,
      "step": 4698
    },
    {
      "epoch": 10.46325167037862,
      "loss": 0.7756419777870178,
      "loss_ce": 0.00012929517833981663,
      "loss_iou": 0.310546875,
      "loss_num": 0.0308837890625,
      "loss_xval": 0.77734375,
      "num_input_tokens_seen": 263168172,
      "step": 4698
    },
    {
      "epoch": 10.465478841870825,
      "grad_norm": 16.894933700561523,
      "learning_rate": 1e-06,
      "loss": 0.5117,
      "num_input_tokens_seen": 263224844,
      "step": 4699
    },
    {
      "epoch": 10.465478841870825,
      "loss": 0.4671807885169983,
      "loss_ce": 0.0001397810410708189,
      "loss_iou": 0.208984375,
      "loss_num": 0.0098876953125,
      "loss_xval": 0.466796875,
      "num_input_tokens_seen": 263224844,
      "step": 4699
    },
    {
      "epoch": 10.46770601336303,
      "grad_norm": 21.554258346557617,
      "learning_rate": 1e-06,
      "loss": 0.5666,
      "num_input_tokens_seen": 263278976,
      "step": 4700
    },
    {
      "epoch": 10.46770601336303,
      "loss": 0.5681118965148926,
      "loss_ce": 0.00011875165364472196,
      "loss_iou": 0.259765625,
      "loss_num": 0.00994873046875,
      "loss_xval": 0.56640625,
      "num_input_tokens_seen": 263278976,
      "step": 4700
    },
    {
      "epoch": 10.469933184855234,
      "grad_norm": 17.350711822509766,
      "learning_rate": 1e-06,
      "loss": 0.5042,
      "num_input_tokens_seen": 263336380,
      "step": 4701
    },
    {
      "epoch": 10.469933184855234,
      "loss": 0.4490933418273926,
      "loss_ce": 0.0002408274740446359,
      "loss_iou": 0.1962890625,
      "loss_num": 0.01123046875,
      "loss_xval": 0.44921875,
      "num_input_tokens_seen": 263336380,
      "step": 4701
    },
    {
      "epoch": 10.47216035634744,
      "grad_norm": 20.207218170166016,
      "learning_rate": 1e-06,
      "loss": 0.5642,
      "num_input_tokens_seen": 263390656,
      "step": 4702
    },
    {
      "epoch": 10.47216035634744,
      "loss": 0.5887584090232849,
      "loss_ce": 0.0001353670231765136,
      "loss_iou": 0.2099609375,
      "loss_num": 0.03369140625,
      "loss_xval": 0.58984375,
      "num_input_tokens_seen": 263390656,
      "step": 4702
    },
    {
      "epoch": 10.474387527839644,
      "grad_norm": 29.60268783569336,
      "learning_rate": 1e-06,
      "loss": 0.6127,
      "num_input_tokens_seen": 263448332,
      "step": 4703
    },
    {
      "epoch": 10.474387527839644,
      "loss": 0.6975966691970825,
      "loss_ce": 0.00014798392658121884,
      "loss_iou": 0.302734375,
      "loss_num": 0.0184326171875,
      "loss_xval": 0.69921875,
      "num_input_tokens_seen": 263448332,
      "step": 4703
    },
    {
      "epoch": 10.476614699331849,
      "grad_norm": 14.4660005569458,
      "learning_rate": 1e-06,
      "loss": 0.4241,
      "num_input_tokens_seen": 263503824,
      "step": 4704
    },
    {
      "epoch": 10.476614699331849,
      "loss": 0.3470431864261627,
      "loss_ce": 0.00011936050577787682,
      "loss_iou": 0.1484375,
      "loss_num": 0.010009765625,
      "loss_xval": 0.34765625,
      "num_input_tokens_seen": 263503824,
      "step": 4704
    },
    {
      "epoch": 10.478841870824054,
      "grad_norm": 17.75361442565918,
      "learning_rate": 1e-06,
      "loss": 0.5674,
      "num_input_tokens_seen": 263559724,
      "step": 4705
    },
    {
      "epoch": 10.478841870824054,
      "loss": 0.6192811131477356,
      "loss_ce": 0.0001404954819008708,
      "loss_iou": 0.267578125,
      "loss_num": 0.016845703125,
      "loss_xval": 0.6171875,
      "num_input_tokens_seen": 263559724,
      "step": 4705
    },
    {
      "epoch": 10.481069042316259,
      "grad_norm": 16.871896743774414,
      "learning_rate": 1e-06,
      "loss": 0.5026,
      "num_input_tokens_seen": 263615824,
      "step": 4706
    },
    {
      "epoch": 10.481069042316259,
      "loss": 0.42048490047454834,
      "loss_ce": 0.00019680103287100792,
      "loss_iou": 0.1728515625,
      "loss_num": 0.01483154296875,
      "loss_xval": 0.419921875,
      "num_input_tokens_seen": 263615824,
      "step": 4706
    },
    {
      "epoch": 10.483296213808464,
      "grad_norm": 18.86580467224121,
      "learning_rate": 1e-06,
      "loss": 0.4055,
      "num_input_tokens_seen": 263672988,
      "step": 4707
    },
    {
      "epoch": 10.483296213808464,
      "loss": 0.4523187577724457,
      "loss_ce": 0.0001703191373962909,
      "loss_iou": 0.1982421875,
      "loss_num": 0.01116943359375,
      "loss_xval": 0.453125,
      "num_input_tokens_seen": 263672988,
      "step": 4707
    },
    {
      "epoch": 10.485523385300668,
      "grad_norm": 18.27134132385254,
      "learning_rate": 1e-06,
      "loss": 0.4455,
      "num_input_tokens_seen": 263729920,
      "step": 4708
    },
    {
      "epoch": 10.485523385300668,
      "loss": 0.46423059701919556,
      "loss_ce": 0.00011923116107936949,
      "loss_iou": 0.2099609375,
      "loss_num": 0.00885009765625,
      "loss_xval": 0.46484375,
      "num_input_tokens_seen": 263729920,
      "step": 4708
    },
    {
      "epoch": 10.487750556792873,
      "grad_norm": 16.8643856048584,
      "learning_rate": 1e-06,
      "loss": 0.4758,
      "num_input_tokens_seen": 263786968,
      "step": 4709
    },
    {
      "epoch": 10.487750556792873,
      "loss": 0.41022104024887085,
      "loss_ce": 0.00018687370175030082,
      "loss_iou": 0.16796875,
      "loss_num": 0.01495361328125,
      "loss_xval": 0.41015625,
      "num_input_tokens_seen": 263786968,
      "step": 4709
    },
    {
      "epoch": 10.489977728285078,
      "grad_norm": 22.2263240814209,
      "learning_rate": 1e-06,
      "loss": 0.4912,
      "num_input_tokens_seen": 263842484,
      "step": 4710
    },
    {
      "epoch": 10.489977728285078,
      "loss": 0.5018296241760254,
      "loss_ce": 0.00012060045264661312,
      "loss_iou": 0.224609375,
      "loss_num": 0.0103759765625,
      "loss_xval": 0.5,
      "num_input_tokens_seen": 263842484,
      "step": 4710
    },
    {
      "epoch": 10.492204899777283,
      "grad_norm": 17.63686180114746,
      "learning_rate": 1e-06,
      "loss": 0.5927,
      "num_input_tokens_seen": 263899136,
      "step": 4711
    },
    {
      "epoch": 10.492204899777283,
      "loss": 0.675079345703125,
      "loss_ce": 0.00015263669774867594,
      "loss_iou": 0.26171875,
      "loss_num": 0.030517578125,
      "loss_xval": 0.67578125,
      "num_input_tokens_seen": 263899136,
      "step": 4711
    },
    {
      "epoch": 10.494432071269488,
      "grad_norm": 19.261802673339844,
      "learning_rate": 1e-06,
      "loss": 0.5165,
      "num_input_tokens_seen": 263956272,
      "step": 4712
    },
    {
      "epoch": 10.494432071269488,
      "loss": 0.4028354585170746,
      "loss_ce": 0.00012550255632959306,
      "loss_iou": 0.16796875,
      "loss_num": 0.013427734375,
      "loss_xval": 0.40234375,
      "num_input_tokens_seen": 263956272,
      "step": 4712
    },
    {
      "epoch": 10.496659242761693,
      "grad_norm": 22.613391876220703,
      "learning_rate": 1e-06,
      "loss": 0.5683,
      "num_input_tokens_seen": 264010704,
      "step": 4713
    },
    {
      "epoch": 10.496659242761693,
      "loss": 0.7420130968093872,
      "loss_ce": 0.0001918151247082278,
      "loss_iou": 0.306640625,
      "loss_num": 0.025390625,
      "loss_xval": 0.7421875,
      "num_input_tokens_seen": 264010704,
      "step": 4713
    },
    {
      "epoch": 10.498886414253898,
      "grad_norm": 18.75325584411621,
      "learning_rate": 1e-06,
      "loss": 0.805,
      "num_input_tokens_seen": 264065440,
      "step": 4714
    },
    {
      "epoch": 10.498886414253898,
      "loss": 0.7152308225631714,
      "loss_ce": 0.00014293221465777606,
      "loss_iou": 0.287109375,
      "loss_num": 0.0277099609375,
      "loss_xval": 0.71484375,
      "num_input_tokens_seen": 264065440,
      "step": 4714
    },
    {
      "epoch": 10.501113585746102,
      "grad_norm": 17.041898727416992,
      "learning_rate": 1e-06,
      "loss": 0.5754,
      "num_input_tokens_seen": 264122460,
      "step": 4715
    },
    {
      "epoch": 10.501113585746102,
      "loss": 0.5261731743812561,
      "loss_ce": 0.0002942573046311736,
      "loss_iou": 0.228515625,
      "loss_num": 0.01373291015625,
      "loss_xval": 0.52734375,
      "num_input_tokens_seen": 264122460,
      "step": 4715
    },
    {
      "epoch": 10.503340757238307,
      "grad_norm": 24.113428115844727,
      "learning_rate": 1e-06,
      "loss": 0.4715,
      "num_input_tokens_seen": 264176608,
      "step": 4716
    },
    {
      "epoch": 10.503340757238307,
      "loss": 0.6706770658493042,
      "loss_ce": 0.0001447929535061121,
      "loss_iou": 0.287109375,
      "loss_num": 0.0196533203125,
      "loss_xval": 0.671875,
      "num_input_tokens_seen": 264176608,
      "step": 4716
    },
    {
      "epoch": 10.505567928730512,
      "grad_norm": 25.971769332885742,
      "learning_rate": 1e-06,
      "loss": 0.6877,
      "num_input_tokens_seen": 264232344,
      "step": 4717
    },
    {
      "epoch": 10.505567928730512,
      "loss": 0.8002659678459167,
      "loss_ce": 0.00046127024688757956,
      "loss_iou": 0.3515625,
      "loss_num": 0.0196533203125,
      "loss_xval": 0.80078125,
      "num_input_tokens_seen": 264232344,
      "step": 4717
    },
    {
      "epoch": 10.507795100222717,
      "grad_norm": 18.61360740661621,
      "learning_rate": 1e-06,
      "loss": 0.6597,
      "num_input_tokens_seen": 264287720,
      "step": 4718
    },
    {
      "epoch": 10.507795100222717,
      "loss": 0.4703478217124939,
      "loss_ce": 0.0001329787337454036,
      "loss_iou": 0.21484375,
      "loss_num": 0.00799560546875,
      "loss_xval": 0.470703125,
      "num_input_tokens_seen": 264287720,
      "step": 4718
    },
    {
      "epoch": 10.510022271714922,
      "grad_norm": 22.263904571533203,
      "learning_rate": 1e-06,
      "loss": 0.3736,
      "num_input_tokens_seen": 264344596,
      "step": 4719
    },
    {
      "epoch": 10.510022271714922,
      "loss": 0.29570016264915466,
      "loss_ce": 0.00016795132250990719,
      "loss_iou": 0.1298828125,
      "loss_num": 0.007080078125,
      "loss_xval": 0.294921875,
      "num_input_tokens_seen": 264344596,
      "step": 4719
    },
    {
      "epoch": 10.512249443207127,
      "grad_norm": 16.474157333374023,
      "learning_rate": 1e-06,
      "loss": 0.4543,
      "num_input_tokens_seen": 264401628,
      "step": 4720
    },
    {
      "epoch": 10.512249443207127,
      "loss": 0.4919620752334595,
      "loss_ce": 0.00014076274237595499,
      "loss_iou": 0.21875,
      "loss_num": 0.0108642578125,
      "loss_xval": 0.4921875,
      "num_input_tokens_seen": 264401628,
      "step": 4720
    },
    {
      "epoch": 10.514476614699332,
      "grad_norm": 14.566910743713379,
      "learning_rate": 1e-06,
      "loss": 0.4783,
      "num_input_tokens_seen": 264457620,
      "step": 4721
    },
    {
      "epoch": 10.514476614699332,
      "loss": 0.4565121829509735,
      "loss_ce": 0.0007016496965661645,
      "loss_iou": 0.1962890625,
      "loss_num": 0.01263427734375,
      "loss_xval": 0.455078125,
      "num_input_tokens_seen": 264457620,
      "step": 4721
    },
    {
      "epoch": 10.516703786191536,
      "grad_norm": 11.910482406616211,
      "learning_rate": 1e-06,
      "loss": 0.584,
      "num_input_tokens_seen": 264514528,
      "step": 4722
    },
    {
      "epoch": 10.516703786191536,
      "loss": 0.4461956024169922,
      "loss_ce": 0.00015069925575517118,
      "loss_iou": 0.1962890625,
      "loss_num": 0.0107421875,
      "loss_xval": 0.4453125,
      "num_input_tokens_seen": 264514528,
      "step": 4722
    },
    {
      "epoch": 10.518930957683741,
      "grad_norm": 20.123058319091797,
      "learning_rate": 1e-06,
      "loss": 0.5795,
      "num_input_tokens_seen": 264571452,
      "step": 4723
    },
    {
      "epoch": 10.518930957683741,
      "loss": 0.7175799608230591,
      "loss_ce": 0.0001727462949929759,
      "loss_iou": 0.314453125,
      "loss_num": 0.017822265625,
      "loss_xval": 0.71875,
      "num_input_tokens_seen": 264571452,
      "step": 4723
    },
    {
      "epoch": 10.521158129175946,
      "grad_norm": 15.674975395202637,
      "learning_rate": 1e-06,
      "loss": 0.5361,
      "num_input_tokens_seen": 264627296,
      "step": 4724
    },
    {
      "epoch": 10.521158129175946,
      "loss": 0.5477256178855896,
      "loss_ce": 0.00011817881022579968,
      "loss_iou": 0.2265625,
      "loss_num": 0.018798828125,
      "loss_xval": 0.546875,
      "num_input_tokens_seen": 264627296,
      "step": 4724
    },
    {
      "epoch": 10.523385300668151,
      "grad_norm": 22.113855361938477,
      "learning_rate": 1e-06,
      "loss": 0.5957,
      "num_input_tokens_seen": 264681016,
      "step": 4725
    },
    {
      "epoch": 10.523385300668151,
      "loss": 0.6728705167770386,
      "loss_ce": 0.0001409871329087764,
      "loss_iou": 0.283203125,
      "loss_num": 0.02099609375,
      "loss_xval": 0.671875,
      "num_input_tokens_seen": 264681016,
      "step": 4725
    },
    {
      "epoch": 10.525612472160356,
      "grad_norm": 18.36030387878418,
      "learning_rate": 1e-06,
      "loss": 0.5032,
      "num_input_tokens_seen": 264738488,
      "step": 4726
    },
    {
      "epoch": 10.525612472160356,
      "loss": 0.3978542685508728,
      "loss_ce": 0.00014918479428160936,
      "loss_iou": 0.166015625,
      "loss_num": 0.01300048828125,
      "loss_xval": 0.3984375,
      "num_input_tokens_seen": 264738488,
      "step": 4726
    },
    {
      "epoch": 10.52783964365256,
      "grad_norm": 17.425609588623047,
      "learning_rate": 1e-06,
      "loss": 0.3512,
      "num_input_tokens_seen": 264795500,
      "step": 4727
    },
    {
      "epoch": 10.52783964365256,
      "loss": 0.3651301860809326,
      "loss_ce": 0.0001399611501256004,
      "loss_iou": 0.150390625,
      "loss_num": 0.01300048828125,
      "loss_xval": 0.365234375,
      "num_input_tokens_seen": 264795500,
      "step": 4727
    },
    {
      "epoch": 10.530066815144766,
      "grad_norm": 27.829883575439453,
      "learning_rate": 1e-06,
      "loss": 0.472,
      "num_input_tokens_seen": 264853260,
      "step": 4728
    },
    {
      "epoch": 10.530066815144766,
      "loss": 0.40601441264152527,
      "loss_ce": 0.00025269074831157923,
      "loss_iou": 0.1689453125,
      "loss_num": 0.0135498046875,
      "loss_xval": 0.40625,
      "num_input_tokens_seen": 264853260,
      "step": 4728
    },
    {
      "epoch": 10.53229398663697,
      "grad_norm": 10.47251033782959,
      "learning_rate": 1e-06,
      "loss": 0.5847,
      "num_input_tokens_seen": 264907728,
      "step": 4729
    },
    {
      "epoch": 10.53229398663697,
      "loss": 0.5061122179031372,
      "loss_ce": 0.00013078347546979785,
      "loss_iou": 0.2216796875,
      "loss_num": 0.0125732421875,
      "loss_xval": 0.5078125,
      "num_input_tokens_seen": 264907728,
      "step": 4729
    },
    {
      "epoch": 10.534521158129175,
      "grad_norm": 26.059276580810547,
      "learning_rate": 1e-06,
      "loss": 0.6593,
      "num_input_tokens_seen": 264961788,
      "step": 4730
    },
    {
      "epoch": 10.534521158129175,
      "loss": 0.6526414752006531,
      "loss_ce": 0.00017564307199791074,
      "loss_iou": 0.283203125,
      "loss_num": 0.0174560546875,
      "loss_xval": 0.65234375,
      "num_input_tokens_seen": 264961788,
      "step": 4730
    },
    {
      "epoch": 10.53674832962138,
      "grad_norm": 17.807939529418945,
      "learning_rate": 1e-06,
      "loss": 0.3761,
      "num_input_tokens_seen": 265015304,
      "step": 4731
    },
    {
      "epoch": 10.53674832962138,
      "loss": 0.3732970356941223,
      "loss_ce": 0.00015859381528571248,
      "loss_iou": 0.171875,
      "loss_num": 0.005950927734375,
      "loss_xval": 0.373046875,
      "num_input_tokens_seen": 265015304,
      "step": 4731
    },
    {
      "epoch": 10.538975501113585,
      "grad_norm": 92.26382446289062,
      "learning_rate": 1e-06,
      "loss": 0.5017,
      "num_input_tokens_seen": 265070648,
      "step": 4732
    },
    {
      "epoch": 10.538975501113585,
      "loss": 0.42812007665634155,
      "loss_ce": 0.0010571026941761374,
      "loss_iou": 0.171875,
      "loss_num": 0.0166015625,
      "loss_xval": 0.427734375,
      "num_input_tokens_seen": 265070648,
      "step": 4732
    },
    {
      "epoch": 10.54120267260579,
      "grad_norm": 18.604368209838867,
      "learning_rate": 1e-06,
      "loss": 0.5278,
      "num_input_tokens_seen": 265126172,
      "step": 4733
    },
    {
      "epoch": 10.54120267260579,
      "loss": 0.5804719924926758,
      "loss_ce": 0.00018023110169451684,
      "loss_iou": 0.234375,
      "loss_num": 0.022216796875,
      "loss_xval": 0.58203125,
      "num_input_tokens_seen": 265126172,
      "step": 4733
    },
    {
      "epoch": 10.543429844097995,
      "grad_norm": 17.7320556640625,
      "learning_rate": 1e-06,
      "loss": 0.6031,
      "num_input_tokens_seen": 265183528,
      "step": 4734
    },
    {
      "epoch": 10.543429844097995,
      "loss": 0.49740028381347656,
      "loss_ce": 0.00045207011862657964,
      "loss_iou": 0.1884765625,
      "loss_num": 0.02392578125,
      "loss_xval": 0.49609375,
      "num_input_tokens_seen": 265183528,
      "step": 4734
    },
    {
      "epoch": 10.5456570155902,
      "grad_norm": 29.795515060424805,
      "learning_rate": 1e-06,
      "loss": 0.3743,
      "num_input_tokens_seen": 265241408,
      "step": 4735
    },
    {
      "epoch": 10.5456570155902,
      "loss": 0.37365561723709106,
      "loss_ce": 0.00012047103518852964,
      "loss_iou": 0.1640625,
      "loss_num": 0.009033203125,
      "loss_xval": 0.373046875,
      "num_input_tokens_seen": 265241408,
      "step": 4735
    },
    {
      "epoch": 10.547884187082406,
      "grad_norm": 60.60297775268555,
      "learning_rate": 1e-06,
      "loss": 0.5276,
      "num_input_tokens_seen": 265296856,
      "step": 4736
    },
    {
      "epoch": 10.547884187082406,
      "loss": 0.3346078097820282,
      "loss_ce": 0.0001351345854345709,
      "loss_iou": 0.15625,
      "loss_num": 0.004180908203125,
      "loss_xval": 0.333984375,
      "num_input_tokens_seen": 265296856,
      "step": 4736
    },
    {
      "epoch": 10.550111358574611,
      "grad_norm": 28.60586929321289,
      "learning_rate": 1e-06,
      "loss": 0.5012,
      "num_input_tokens_seen": 265351804,
      "step": 4737
    },
    {
      "epoch": 10.550111358574611,
      "loss": 0.5089592933654785,
      "loss_ce": 0.00017022958490997553,
      "loss_iou": 0.1982421875,
      "loss_num": 0.0224609375,
      "loss_xval": 0.5078125,
      "num_input_tokens_seen": 265351804,
      "step": 4737
    },
    {
      "epoch": 10.552338530066816,
      "grad_norm": 16.353670120239258,
      "learning_rate": 1e-06,
      "loss": 0.6282,
      "num_input_tokens_seen": 265406476,
      "step": 4738
    },
    {
      "epoch": 10.552338530066816,
      "loss": 0.6063532829284668,
      "loss_ce": 0.00015206160605885088,
      "loss_iou": 0.255859375,
      "loss_num": 0.018798828125,
      "loss_xval": 0.60546875,
      "num_input_tokens_seen": 265406476,
      "step": 4738
    },
    {
      "epoch": 10.55456570155902,
      "grad_norm": 18.11619758605957,
      "learning_rate": 1e-06,
      "loss": 0.5011,
      "num_input_tokens_seen": 265463764,
      "step": 4739
    },
    {
      "epoch": 10.55456570155902,
      "loss": 0.6158651113510132,
      "loss_ce": 0.000386598490877077,
      "loss_iou": 0.271484375,
      "loss_num": 0.01458740234375,
      "loss_xval": 0.6171875,
      "num_input_tokens_seen": 265463764,
      "step": 4739
    },
    {
      "epoch": 10.556792873051226,
      "grad_norm": 21.920385360717773,
      "learning_rate": 1e-06,
      "loss": 0.7462,
      "num_input_tokens_seen": 265520116,
      "step": 4740
    },
    {
      "epoch": 10.556792873051226,
      "loss": 0.9230633974075317,
      "loss_ce": 0.00021183726494200528,
      "loss_iou": 0.37890625,
      "loss_num": 0.033203125,
      "loss_xval": 0.921875,
      "num_input_tokens_seen": 265520116,
      "step": 4740
    },
    {
      "epoch": 10.55902004454343,
      "grad_norm": 12.749909400939941,
      "learning_rate": 1e-06,
      "loss": 0.4596,
      "num_input_tokens_seen": 265578596,
      "step": 4741
    },
    {
      "epoch": 10.55902004454343,
      "loss": 0.5135598182678223,
      "loss_ce": 0.00013206491712480783,
      "loss_iou": 0.2255859375,
      "loss_num": 0.01275634765625,
      "loss_xval": 0.51171875,
      "num_input_tokens_seen": 265578596,
      "step": 4741
    },
    {
      "epoch": 10.561247216035635,
      "grad_norm": 23.8961124420166,
      "learning_rate": 1e-06,
      "loss": 0.669,
      "num_input_tokens_seen": 265634096,
      "step": 4742
    },
    {
      "epoch": 10.561247216035635,
      "loss": 0.8774167895317078,
      "loss_ce": 0.0029050512239336967,
      "loss_iou": 0.34765625,
      "loss_num": 0.03564453125,
      "loss_xval": 0.875,
      "num_input_tokens_seen": 265634096,
      "step": 4742
    },
    {
      "epoch": 10.56347438752784,
      "grad_norm": 27.8192081451416,
      "learning_rate": 1e-06,
      "loss": 0.6048,
      "num_input_tokens_seen": 265687780,
      "step": 4743
    },
    {
      "epoch": 10.56347438752784,
      "loss": 0.5951275825500488,
      "loss_ce": 0.00015690061263740063,
      "loss_iou": 0.2578125,
      "loss_num": 0.0162353515625,
      "loss_xval": 0.59375,
      "num_input_tokens_seen": 265687780,
      "step": 4743
    },
    {
      "epoch": 10.565701559020045,
      "grad_norm": 23.720129013061523,
      "learning_rate": 1e-06,
      "loss": 0.4395,
      "num_input_tokens_seen": 265746240,
      "step": 4744
    },
    {
      "epoch": 10.565701559020045,
      "loss": 0.43288612365722656,
      "loss_ce": 0.0001468704140279442,
      "loss_iou": 0.203125,
      "loss_num": 0.005126953125,
      "loss_xval": 0.43359375,
      "num_input_tokens_seen": 265746240,
      "step": 4744
    },
    {
      "epoch": 10.56792873051225,
      "grad_norm": 18.638071060180664,
      "learning_rate": 1e-06,
      "loss": 0.5585,
      "num_input_tokens_seen": 265800708,
      "step": 4745
    },
    {
      "epoch": 10.56792873051225,
      "loss": 0.690610408782959,
      "loss_ce": 0.00018075491243507713,
      "loss_iou": 0.291015625,
      "loss_num": 0.0216064453125,
      "loss_xval": 0.69140625,
      "num_input_tokens_seen": 265800708,
      "step": 4745
    },
    {
      "epoch": 10.570155902004455,
      "grad_norm": 27.448678970336914,
      "learning_rate": 1e-06,
      "loss": 0.5302,
      "num_input_tokens_seen": 265857676,
      "step": 4746
    },
    {
      "epoch": 10.570155902004455,
      "loss": 0.5328932404518127,
      "loss_ce": 0.0001783959160093218,
      "loss_iou": 0.2333984375,
      "loss_num": 0.01312255859375,
      "loss_xval": 0.53125,
      "num_input_tokens_seen": 265857676,
      "step": 4746
    },
    {
      "epoch": 10.57238307349666,
      "grad_norm": 18.024444580078125,
      "learning_rate": 1e-06,
      "loss": 0.5738,
      "num_input_tokens_seen": 265914688,
      "step": 4747
    },
    {
      "epoch": 10.57238307349666,
      "loss": 0.5035586357116699,
      "loss_ce": 0.0001406791852787137,
      "loss_iou": 0.2236328125,
      "loss_num": 0.0113525390625,
      "loss_xval": 0.50390625,
      "num_input_tokens_seen": 265914688,
      "step": 4747
    },
    {
      "epoch": 10.574610244988865,
      "grad_norm": 17.24481773376465,
      "learning_rate": 1e-06,
      "loss": 0.6003,
      "num_input_tokens_seen": 265970292,
      "step": 4748
    },
    {
      "epoch": 10.574610244988865,
      "loss": 0.5748478174209595,
      "loss_ce": 0.00014080088294576854,
      "loss_iou": 0.255859375,
      "loss_num": 0.0126953125,
      "loss_xval": 0.57421875,
      "num_input_tokens_seen": 265970292,
      "step": 4748
    },
    {
      "epoch": 10.57683741648107,
      "grad_norm": 21.86254119873047,
      "learning_rate": 1e-06,
      "loss": 0.5795,
      "num_input_tokens_seen": 266027040,
      "step": 4749
    },
    {
      "epoch": 10.57683741648107,
      "loss": 0.6045183539390564,
      "loss_ce": 0.00014822966477368027,
      "loss_iou": 0.2578125,
      "loss_num": 0.017822265625,
      "loss_xval": 0.60546875,
      "num_input_tokens_seen": 266027040,
      "step": 4749
    },
    {
      "epoch": 10.579064587973274,
      "grad_norm": 15.369706153869629,
      "learning_rate": 1e-06,
      "loss": 0.5245,
      "num_input_tokens_seen": 266085392,
      "step": 4750
    },
    {
      "epoch": 10.579064587973274,
      "eval_seeclick_web_CIoU": 0.5785282254219055,
      "eval_seeclick_web_GIoU": 0.5777834057807922,
      "eval_seeclick_web_IoU": 0.5963070392608643,
      "eval_seeclick_web_MAE_all": 0.016103142872452736,
      "eval_seeclick_web_MAE_h": 0.0080435904674232,
      "eval_seeclick_web_MAE_w": 0.015959544107317924,
      "eval_seeclick_web_MAE_x_boxes": 0.009686239995062351,
      "eval_seeclick_web_MAE_y_boxes": 0.021757841110229492,
      "eval_seeclick_web_inside_bbox": 0.9166666567325592,
      "eval_seeclick_web_loss": 0.9105224013328552,
      "eval_seeclick_web_loss_ce": 0.0002034704084508121,
      "eval_seeclick_web_loss_iou": 0.415283203125,
      "eval_seeclick_web_loss_num": 0.012907028198242188,
      "eval_seeclick_web_loss_xval": 0.89501953125,
      "eval_seeclick_web_runtime": 21.8476,
      "eval_seeclick_web_samples_per_second": 2.289,
      "eval_seeclick_web_steps_per_second": 0.092,
      "num_input_tokens_seen": 266085392,
      "step": 4750
    },
    {
      "epoch": 10.579064587973274,
      "eval_icons_CIoU": 0.2794487178325653,
      "eval_icons_GIoU": 0.3004266917705536,
      "eval_icons_IoU": 0.3560175597667694,
      "eval_icons_MAE_all": 0.060138389468193054,
      "eval_icons_MAE_h": 0.03905831277370453,
      "eval_icons_MAE_w": 0.05829060822725296,
      "eval_icons_MAE_x_boxes": 0.05876126326620579,
      "eval_icons_MAE_y_boxes": 0.0388388317078352,
      "eval_icons_inside_bbox": 0.6059027910232544,
      "eval_icons_loss": 1.7231128215789795,
      "eval_icons_loss_ce": 0.00027843897260027006,
      "eval_icons_loss_iou": 0.6767578125,
      "eval_icons_loss_num": 0.05956840515136719,
      "eval_icons_loss_xval": 1.6513671875,
      "eval_icons_runtime": 21.0355,
      "eval_icons_samples_per_second": 2.377,
      "eval_icons_steps_per_second": 0.095,
      "num_input_tokens_seen": 266085392,
      "step": 4750
    },
    {
      "epoch": 10.579064587973274,
      "eval_screenspot_CIoU": 0.3542960087458293,
      "eval_screenspot_GIoU": 0.373006671667099,
      "eval_screenspot_IoU": 0.4333365758260091,
      "eval_screenspot_MAE_all": 0.05960427472988764,
      "eval_screenspot_MAE_h": 0.03762041280666987,
      "eval_screenspot_MAE_w": 0.06836641455690066,
      "eval_screenspot_MAE_x_boxes": 0.07390400022268295,
      "eval_screenspot_MAE_y_boxes": 0.03899012443919977,
      "eval_screenspot_inside_bbox": 0.6862499912579855,
      "eval_screenspot_loss": 1.6113767623901367,
      "eval_screenspot_loss_ce": 0.0002595222613308579,
      "eval_screenspot_loss_iou": 0.6659342447916666,
      "eval_screenspot_loss_num": 0.0684064229329427,
      "eval_screenspot_loss_xval": 1.6736653645833333,
      "eval_screenspot_runtime": 35.1964,
      "eval_screenspot_samples_per_second": 2.529,
      "eval_screenspot_steps_per_second": 0.085,
      "num_input_tokens_seen": 266085392,
      "step": 4750
    },
    {
      "epoch": 10.579064587973274,
      "eval_compot_CIoU": 0.34722261130809784,
      "eval_compot_GIoU": 0.35881949961185455,
      "eval_compot_IoU": 0.4061947613954544,
      "eval_compot_MAE_all": 0.018287954851984978,
      "eval_compot_MAE_h": 0.009336336981505156,
      "eval_compot_MAE_w": 0.0211613979190588,
      "eval_compot_MAE_x_boxes": 0.030585231259465218,
      "eval_compot_MAE_y_boxes": 0.006953268777579069,
      "eval_compot_inside_bbox": 0.6302083432674408,
      "eval_compot_loss": 1.3919204473495483,
      "eval_compot_loss_ce": 0.00020033524197060615,
      "eval_compot_loss_iou": 0.6395263671875,
      "eval_compot_loss_num": 0.017185211181640625,
      "eval_compot_loss_xval": 1.364501953125,
      "eval_compot_runtime": 21.5206,
      "eval_compot_samples_per_second": 2.323,
      "eval_compot_steps_per_second": 0.093,
      "num_input_tokens_seen": 266085392,
      "step": 4750
    },
    {
      "epoch": 10.579064587973274,
      "eval_custom_ui_val_CIoU": 0.47197884652349686,
      "eval_custom_ui_val_GIoU": 0.48575951324568856,
      "eval_custom_ui_val_IoU": 0.5315978427728018,
      "eval_custom_ui_val_MAE_all": 0.03096564869499869,
      "eval_custom_ui_val_MAE_h": 0.016998413018882275,
      "eval_custom_ui_val_MAE_w": 0.039277435590823494,
      "eval_custom_ui_val_MAE_x_boxes": 0.038262664857837886,
      "eval_custom_ui_val_MAE_y_boxes": 0.01508552656095061,
      "eval_custom_ui_val_inside_bbox": 0.7527006202273898,
      "eval_custom_ui_val_loss": 1.1994266510009766,
      "eval_custom_ui_val_loss_ce": 0.00023388697849845307,
      "eval_custom_ui_val_loss_iou": 0.5101725260416666,
      "eval_custom_ui_val_loss_num": 0.028283013237847224,
      "eval_custom_ui_val_loss_xval": 1.1618109809027777,
      "eval_custom_ui_val_runtime": 65.3312,
      "eval_custom_ui_val_samples_per_second": 4.056,
      "eval_custom_ui_val_steps_per_second": 0.138,
      "num_input_tokens_seen": 266085392,
      "step": 4750
    },
    {
      "epoch": 10.579064587973274,
      "loss": 0.9167079925537109,
      "loss_ce": 0.00020406048861332238,
      "loss_iou": 0.39453125,
      "loss_num": 0.0252685546875,
      "loss_xval": 0.91796875,
      "num_input_tokens_seen": 266085392,
      "step": 4750
    },
    {
      "epoch": 10.58129175946548,
      "grad_norm": 22.031713485717773,
      "learning_rate": 1e-06,
      "loss": 0.5825,
      "num_input_tokens_seen": 266140888,
      "step": 4751
    },
    {
      "epoch": 10.58129175946548,
      "loss": 0.5439281463623047,
      "loss_ce": 0.00013536959886550903,
      "loss_iou": 0.24609375,
      "loss_num": 0.0101318359375,
      "loss_xval": 0.54296875,
      "num_input_tokens_seen": 266140888,
      "step": 4751
    },
    {
      "epoch": 10.583518930957684,
      "grad_norm": 28.358179092407227,
      "learning_rate": 1e-06,
      "loss": 0.5168,
      "num_input_tokens_seen": 266195452,
      "step": 4752
    },
    {
      "epoch": 10.583518930957684,
      "loss": 0.4804303050041199,
      "loss_ce": 0.0002057160745607689,
      "loss_iou": 0.2041015625,
      "loss_num": 0.014404296875,
      "loss_xval": 0.48046875,
      "num_input_tokens_seen": 266195452,
      "step": 4752
    },
    {
      "epoch": 10.585746102449889,
      "grad_norm": 20.244579315185547,
      "learning_rate": 1e-06,
      "loss": 0.6685,
      "num_input_tokens_seen": 266253420,
      "step": 4753
    },
    {
      "epoch": 10.585746102449889,
      "loss": 0.6307578086853027,
      "loss_ce": 0.00014260906027629972,
      "loss_iou": 0.283203125,
      "loss_num": 0.01300048828125,
      "loss_xval": 0.62890625,
      "num_input_tokens_seen": 266253420,
      "step": 4753
    },
    {
      "epoch": 10.587973273942094,
      "grad_norm": 23.68046760559082,
      "learning_rate": 1e-06,
      "loss": 0.5178,
      "num_input_tokens_seen": 266309880,
      "step": 4754
    },
    {
      "epoch": 10.587973273942094,
      "loss": 0.590294599533081,
      "loss_ce": 0.0002067518071271479,
      "loss_iou": 0.255859375,
      "loss_num": 0.015380859375,
      "loss_xval": 0.58984375,
      "num_input_tokens_seen": 266309880,
      "step": 4754
    },
    {
      "epoch": 10.590200445434299,
      "grad_norm": 14.198582649230957,
      "learning_rate": 1e-06,
      "loss": 0.6997,
      "num_input_tokens_seen": 266367096,
      "step": 4755
    },
    {
      "epoch": 10.590200445434299,
      "loss": 0.7059162855148315,
      "loss_ce": 0.00022783357417210937,
      "loss_iou": 0.3046875,
      "loss_num": 0.019287109375,
      "loss_xval": 0.70703125,
      "num_input_tokens_seen": 266367096,
      "step": 4755
    },
    {
      "epoch": 10.592427616926503,
      "grad_norm": 22.98945426940918,
      "learning_rate": 1e-06,
      "loss": 0.569,
      "num_input_tokens_seen": 266421524,
      "step": 4756
    },
    {
      "epoch": 10.592427616926503,
      "loss": 0.6033655405044556,
      "loss_ce": 0.0001550911256344989,
      "loss_iou": 0.267578125,
      "loss_num": 0.0137939453125,
      "loss_xval": 0.6015625,
      "num_input_tokens_seen": 266421524,
      "step": 4756
    },
    {
      "epoch": 10.594654788418708,
      "grad_norm": 22.537723541259766,
      "learning_rate": 1e-06,
      "loss": 0.3968,
      "num_input_tokens_seen": 266475392,
      "step": 4757
    },
    {
      "epoch": 10.594654788418708,
      "loss": 0.424811989068985,
      "loss_ce": 0.00012936009443365037,
      "loss_iou": 0.1875,
      "loss_num": 0.010009765625,
      "loss_xval": 0.423828125,
      "num_input_tokens_seen": 266475392,
      "step": 4757
    },
    {
      "epoch": 10.596881959910913,
      "grad_norm": 21.612504959106445,
      "learning_rate": 1e-06,
      "loss": 0.5241,
      "num_input_tokens_seen": 266531920,
      "step": 4758
    },
    {
      "epoch": 10.596881959910913,
      "loss": 0.5215491056442261,
      "loss_ce": 0.0003089057863689959,
      "loss_iou": 0.2275390625,
      "loss_num": 0.01348876953125,
      "loss_xval": 0.51953125,
      "num_input_tokens_seen": 266531920,
      "step": 4758
    },
    {
      "epoch": 10.599109131403118,
      "grad_norm": 14.450573921203613,
      "learning_rate": 1e-06,
      "loss": 0.5417,
      "num_input_tokens_seen": 266585680,
      "step": 4759
    },
    {
      "epoch": 10.599109131403118,
      "loss": 0.6613370180130005,
      "loss_ce": 0.00014318549074232578,
      "loss_iou": 0.26953125,
      "loss_num": 0.0242919921875,
      "loss_xval": 0.66015625,
      "num_input_tokens_seen": 266585680,
      "step": 4759
    },
    {
      "epoch": 10.601336302895323,
      "grad_norm": 24.842741012573242,
      "learning_rate": 1e-06,
      "loss": 0.7178,
      "num_input_tokens_seen": 266643272,
      "step": 4760
    },
    {
      "epoch": 10.601336302895323,
      "loss": 0.8287136554718018,
      "loss_ce": 0.00022248990717343986,
      "loss_iou": 0.365234375,
      "loss_num": 0.0196533203125,
      "loss_xval": 0.828125,
      "num_input_tokens_seen": 266643272,
      "step": 4760
    },
    {
      "epoch": 10.603563474387528,
      "grad_norm": 21.21478271484375,
      "learning_rate": 1e-06,
      "loss": 0.622,
      "num_input_tokens_seen": 266700996,
      "step": 4761
    },
    {
      "epoch": 10.603563474387528,
      "loss": 0.6402872800827026,
      "loss_ce": 0.00015057012205943465,
      "loss_iou": 0.28515625,
      "loss_num": 0.0140380859375,
      "loss_xval": 0.640625,
      "num_input_tokens_seen": 266700996,
      "step": 4761
    },
    {
      "epoch": 10.605790645879733,
      "grad_norm": 31.64474868774414,
      "learning_rate": 1e-06,
      "loss": 0.4276,
      "num_input_tokens_seen": 266758212,
      "step": 4762
    },
    {
      "epoch": 10.605790645879733,
      "loss": 0.43672651052474976,
      "loss_ce": 0.0002030893083428964,
      "loss_iou": 0.19140625,
      "loss_num": 0.0108642578125,
      "loss_xval": 0.4375,
      "num_input_tokens_seen": 266758212,
      "step": 4762
    },
    {
      "epoch": 10.608017817371937,
      "grad_norm": 32.779476165771484,
      "learning_rate": 1e-06,
      "loss": 0.6393,
      "num_input_tokens_seen": 266811900,
      "step": 4763
    },
    {
      "epoch": 10.608017817371937,
      "loss": 0.5560557842254639,
      "loss_ce": 0.00014755260781385005,
      "loss_iou": 0.2373046875,
      "loss_num": 0.016357421875,
      "loss_xval": 0.5546875,
      "num_input_tokens_seen": 266811900,
      "step": 4763
    },
    {
      "epoch": 10.610244988864142,
      "grad_norm": 17.99529266357422,
      "learning_rate": 1e-06,
      "loss": 0.4536,
      "num_input_tokens_seen": 266863520,
      "step": 4764
    },
    {
      "epoch": 10.610244988864142,
      "loss": 0.441336989402771,
      "loss_ce": 0.00011382217053323984,
      "loss_iou": 0.15625,
      "loss_num": 0.02587890625,
      "loss_xval": 0.44140625,
      "num_input_tokens_seen": 266863520,
      "step": 4764
    },
    {
      "epoch": 10.612472160356347,
      "grad_norm": 20.494014739990234,
      "learning_rate": 1e-06,
      "loss": 0.7394,
      "num_input_tokens_seen": 266918760,
      "step": 4765
    },
    {
      "epoch": 10.612472160356347,
      "loss": 0.49647536873817444,
      "loss_ce": 0.00013749409117735922,
      "loss_iou": 0.2216796875,
      "loss_num": 0.01055908203125,
      "loss_xval": 0.49609375,
      "num_input_tokens_seen": 266918760,
      "step": 4765
    },
    {
      "epoch": 10.614699331848552,
      "grad_norm": 19.456296920776367,
      "learning_rate": 1e-06,
      "loss": 0.4065,
      "num_input_tokens_seen": 266975336,
      "step": 4766
    },
    {
      "epoch": 10.614699331848552,
      "loss": 0.4348350167274475,
      "loss_ce": 0.00014263816410675645,
      "loss_iou": 0.1943359375,
      "loss_num": 0.009033203125,
      "loss_xval": 0.435546875,
      "num_input_tokens_seen": 266975336,
      "step": 4766
    },
    {
      "epoch": 10.616926503340757,
      "grad_norm": 23.569063186645508,
      "learning_rate": 1e-06,
      "loss": 0.4614,
      "num_input_tokens_seen": 267031440,
      "step": 4767
    },
    {
      "epoch": 10.616926503340757,
      "loss": 0.4862057566642761,
      "loss_ce": 0.00012178381439298391,
      "loss_iou": 0.2236328125,
      "loss_num": 0.0076904296875,
      "loss_xval": 0.486328125,
      "num_input_tokens_seen": 267031440,
      "step": 4767
    },
    {
      "epoch": 10.619153674832962,
      "grad_norm": 20.638471603393555,
      "learning_rate": 1e-06,
      "loss": 0.4967,
      "num_input_tokens_seen": 267086692,
      "step": 4768
    },
    {
      "epoch": 10.619153674832962,
      "loss": 0.5913643836975098,
      "loss_ce": 0.00017783006478566676,
      "loss_iou": 0.267578125,
      "loss_num": 0.0108642578125,
      "loss_xval": 0.58984375,
      "num_input_tokens_seen": 267086692,
      "step": 4768
    },
    {
      "epoch": 10.621380846325167,
      "grad_norm": 17.369722366333008,
      "learning_rate": 1e-06,
      "loss": 0.5522,
      "num_input_tokens_seen": 267143996,
      "step": 4769
    },
    {
      "epoch": 10.621380846325167,
      "loss": 0.6038705706596375,
      "loss_ce": 0.0001718288112897426,
      "loss_iou": 0.275390625,
      "loss_num": 0.01043701171875,
      "loss_xval": 0.60546875,
      "num_input_tokens_seen": 267143996,
      "step": 4769
    },
    {
      "epoch": 10.623608017817372,
      "grad_norm": 15.817121505737305,
      "learning_rate": 1e-06,
      "loss": 0.6093,
      "num_input_tokens_seen": 267202664,
      "step": 4770
    },
    {
      "epoch": 10.623608017817372,
      "loss": 0.608995795249939,
      "loss_ce": 0.00010910604032687843,
      "loss_iou": 0.26953125,
      "loss_num": 0.01434326171875,
      "loss_xval": 0.609375,
      "num_input_tokens_seen": 267202664,
      "step": 4770
    },
    {
      "epoch": 10.625835189309576,
      "grad_norm": 19.741697311401367,
      "learning_rate": 1e-06,
      "loss": 0.6764,
      "num_input_tokens_seen": 267259716,
      "step": 4771
    },
    {
      "epoch": 10.625835189309576,
      "loss": 0.5544874668121338,
      "loss_ce": 0.00016620635869912803,
      "loss_iou": 0.21875,
      "loss_num": 0.0233154296875,
      "loss_xval": 0.5546875,
      "num_input_tokens_seen": 267259716,
      "step": 4771
    },
    {
      "epoch": 10.628062360801781,
      "grad_norm": 14.572179794311523,
      "learning_rate": 1e-06,
      "loss": 0.4699,
      "num_input_tokens_seen": 267316276,
      "step": 4772
    },
    {
      "epoch": 10.628062360801781,
      "loss": 0.37722790241241455,
      "loss_ce": 0.0001526982377981767,
      "loss_iou": 0.169921875,
      "loss_num": 0.00738525390625,
      "loss_xval": 0.376953125,
      "num_input_tokens_seen": 267316276,
      "step": 4772
    },
    {
      "epoch": 10.630289532293986,
      "grad_norm": 35.8219108581543,
      "learning_rate": 1e-06,
      "loss": 0.7025,
      "num_input_tokens_seen": 267370620,
      "step": 4773
    },
    {
      "epoch": 10.630289532293986,
      "loss": 0.7943651080131531,
      "loss_ce": 0.00017563029541634023,
      "loss_iou": 0.359375,
      "loss_num": 0.0147705078125,
      "loss_xval": 0.79296875,
      "num_input_tokens_seen": 267370620,
      "step": 4773
    },
    {
      "epoch": 10.632516703786191,
      "grad_norm": 20.09000015258789,
      "learning_rate": 1e-06,
      "loss": 0.7104,
      "num_input_tokens_seen": 267427184,
      "step": 4774
    },
    {
      "epoch": 10.632516703786191,
      "loss": 0.812397837638855,
      "loss_ce": 0.00014197138079907745,
      "loss_iou": 0.318359375,
      "loss_num": 0.03515625,
      "loss_xval": 0.8125,
      "num_input_tokens_seen": 267427184,
      "step": 4774
    },
    {
      "epoch": 10.634743875278396,
      "grad_norm": 21.24492835998535,
      "learning_rate": 1e-06,
      "loss": 0.6861,
      "num_input_tokens_seen": 267481704,
      "step": 4775
    },
    {
      "epoch": 10.634743875278396,
      "loss": 0.7231541872024536,
      "loss_ce": 0.0001317608985118568,
      "loss_iou": 0.306640625,
      "loss_num": 0.0223388671875,
      "loss_xval": 0.72265625,
      "num_input_tokens_seen": 267481704,
      "step": 4775
    },
    {
      "epoch": 10.6369710467706,
      "grad_norm": 32.74296951293945,
      "learning_rate": 1e-06,
      "loss": 0.5468,
      "num_input_tokens_seen": 267539444,
      "step": 4776
    },
    {
      "epoch": 10.6369710467706,
      "loss": 0.5010995864868164,
      "loss_ce": 0.00030611734837293625,
      "loss_iou": 0.2109375,
      "loss_num": 0.015869140625,
      "loss_xval": 0.5,
      "num_input_tokens_seen": 267539444,
      "step": 4776
    },
    {
      "epoch": 10.639198218262806,
      "grad_norm": 21.132326126098633,
      "learning_rate": 1e-06,
      "loss": 0.6992,
      "num_input_tokens_seen": 267595728,
      "step": 4777
    },
    {
      "epoch": 10.639198218262806,
      "loss": 0.6195580959320068,
      "loss_ce": 0.00017328912508673966,
      "loss_iou": 0.271484375,
      "loss_num": 0.0152587890625,
      "loss_xval": 0.62109375,
      "num_input_tokens_seen": 267595728,
      "step": 4777
    },
    {
      "epoch": 10.64142538975501,
      "grad_norm": 13.380131721496582,
      "learning_rate": 1e-06,
      "loss": 0.3196,
      "num_input_tokens_seen": 267650392,
      "step": 4778
    },
    {
      "epoch": 10.64142538975501,
      "loss": 0.3022405207157135,
      "loss_ce": 0.00011649330554064363,
      "loss_iou": 0.1240234375,
      "loss_num": 0.0107421875,
      "loss_xval": 0.302734375,
      "num_input_tokens_seen": 267650392,
      "step": 4778
    },
    {
      "epoch": 10.643652561247215,
      "grad_norm": 16.144832611083984,
      "learning_rate": 1e-06,
      "loss": 0.5431,
      "num_input_tokens_seen": 267707784,
      "step": 4779
    },
    {
      "epoch": 10.643652561247215,
      "loss": 0.7454525828361511,
      "loss_ce": 0.00021334037592168897,
      "loss_iou": 0.302734375,
      "loss_num": 0.02783203125,
      "loss_xval": 0.74609375,
      "num_input_tokens_seen": 267707784,
      "step": 4779
    },
    {
      "epoch": 10.64587973273942,
      "grad_norm": 32.90868377685547,
      "learning_rate": 1e-06,
      "loss": 0.57,
      "num_input_tokens_seen": 267764500,
      "step": 4780
    },
    {
      "epoch": 10.64587973273942,
      "loss": 0.6209944486618042,
      "loss_ce": 0.0001448530238121748,
      "loss_iou": 0.263671875,
      "loss_num": 0.0189208984375,
      "loss_xval": 0.62109375,
      "num_input_tokens_seen": 267764500,
      "step": 4780
    },
    {
      "epoch": 10.648106904231625,
      "grad_norm": 90.17802429199219,
      "learning_rate": 1e-06,
      "loss": 0.5177,
      "num_input_tokens_seen": 267819128,
      "step": 4781
    },
    {
      "epoch": 10.648106904231625,
      "loss": 0.5430988073348999,
      "loss_ce": 0.00013003447384107858,
      "loss_iou": 0.21875,
      "loss_num": 0.021240234375,
      "loss_xval": 0.54296875,
      "num_input_tokens_seen": 267819128,
      "step": 4781
    },
    {
      "epoch": 10.65033407572383,
      "grad_norm": 20.155319213867188,
      "learning_rate": 1e-06,
      "loss": 0.6905,
      "num_input_tokens_seen": 267876160,
      "step": 4782
    },
    {
      "epoch": 10.65033407572383,
      "loss": 0.7085360288619995,
      "loss_ce": 0.00016200730169657618,
      "loss_iou": 0.287109375,
      "loss_num": 0.026611328125,
      "loss_xval": 0.70703125,
      "num_input_tokens_seen": 267876160,
      "step": 4782
    },
    {
      "epoch": 10.652561247216035,
      "grad_norm": 20.946746826171875,
      "learning_rate": 1e-06,
      "loss": 0.6077,
      "num_input_tokens_seen": 267933324,
      "step": 4783
    },
    {
      "epoch": 10.652561247216035,
      "loss": 0.5577636361122131,
      "loss_ce": 0.00014642757014371455,
      "loss_iou": 0.24609375,
      "loss_num": 0.01287841796875,
      "loss_xval": 0.55859375,
      "num_input_tokens_seen": 267933324,
      "step": 4783
    },
    {
      "epoch": 10.654788418708241,
      "grad_norm": 23.43316650390625,
      "learning_rate": 1e-06,
      "loss": 0.7215,
      "num_input_tokens_seen": 267990524,
      "step": 4784
    },
    {
      "epoch": 10.654788418708241,
      "loss": 0.5793881416320801,
      "loss_ce": 0.00016452360432595015,
      "loss_iou": 0.2578125,
      "loss_num": 0.01263427734375,
      "loss_xval": 0.578125,
      "num_input_tokens_seen": 267990524,
      "step": 4784
    },
    {
      "epoch": 10.657015590200446,
      "grad_norm": 29.320165634155273,
      "learning_rate": 1e-06,
      "loss": 0.6149,
      "num_input_tokens_seen": 268046904,
      "step": 4785
    },
    {
      "epoch": 10.657015590200446,
      "loss": 0.6889311075210571,
      "loss_ce": 0.00021046542678959668,
      "loss_iou": 0.30078125,
      "loss_num": 0.0177001953125,
      "loss_xval": 0.6875,
      "num_input_tokens_seen": 268046904,
      "step": 4785
    },
    {
      "epoch": 10.659242761692651,
      "grad_norm": 16.68779182434082,
      "learning_rate": 1e-06,
      "loss": 0.6267,
      "num_input_tokens_seen": 268105252,
      "step": 4786
    },
    {
      "epoch": 10.659242761692651,
      "loss": 0.6637170314788818,
      "loss_ce": 0.0001428277901140973,
      "loss_iou": 0.283203125,
      "loss_num": 0.01904296875,
      "loss_xval": 0.6640625,
      "num_input_tokens_seen": 268105252,
      "step": 4786
    },
    {
      "epoch": 10.661469933184856,
      "grad_norm": 41.306156158447266,
      "learning_rate": 1e-06,
      "loss": 0.7032,
      "num_input_tokens_seen": 268159052,
      "step": 4787
    },
    {
      "epoch": 10.661469933184856,
      "loss": 0.565805196762085,
      "loss_ce": 0.00013136808411218226,
      "loss_iou": 0.2490234375,
      "loss_num": 0.01361083984375,
      "loss_xval": 0.56640625,
      "num_input_tokens_seen": 268159052,
      "step": 4787
    },
    {
      "epoch": 10.66369710467706,
      "grad_norm": 12.712847709655762,
      "learning_rate": 1e-06,
      "loss": 0.3692,
      "num_input_tokens_seen": 268214984,
      "step": 4788
    },
    {
      "epoch": 10.66369710467706,
      "loss": 0.47919151186943054,
      "loss_ce": 0.0001876367605291307,
      "loss_iou": 0.21875,
      "loss_num": 0.00848388671875,
      "loss_xval": 0.478515625,
      "num_input_tokens_seen": 268214984,
      "step": 4788
    },
    {
      "epoch": 10.665924276169266,
      "grad_norm": 15.455141067504883,
      "learning_rate": 1e-06,
      "loss": 0.5079,
      "num_input_tokens_seen": 268269568,
      "step": 4789
    },
    {
      "epoch": 10.665924276169266,
      "loss": 0.4243258833885193,
      "loss_ce": 0.00013158305955585092,
      "loss_iou": 0.1650390625,
      "loss_num": 0.018798828125,
      "loss_xval": 0.423828125,
      "num_input_tokens_seen": 268269568,
      "step": 4789
    },
    {
      "epoch": 10.66815144766147,
      "grad_norm": 21.24195671081543,
      "learning_rate": 1e-06,
      "loss": 0.5663,
      "num_input_tokens_seen": 268325664,
      "step": 4790
    },
    {
      "epoch": 10.66815144766147,
      "loss": 0.4882197380065918,
      "loss_ce": 0.000182637624675408,
      "loss_iou": 0.2255859375,
      "loss_num": 0.007415771484375,
      "loss_xval": 0.48828125,
      "num_input_tokens_seen": 268325664,
      "step": 4790
    },
    {
      "epoch": 10.670378619153675,
      "grad_norm": 14.548911094665527,
      "learning_rate": 1e-06,
      "loss": 0.7085,
      "num_input_tokens_seen": 268378752,
      "step": 4791
    },
    {
      "epoch": 10.670378619153675,
      "loss": 0.6994498372077942,
      "loss_ce": 0.00023110417532734573,
      "loss_iou": 0.2734375,
      "loss_num": 0.030029296875,
      "loss_xval": 0.69921875,
      "num_input_tokens_seen": 268378752,
      "step": 4791
    },
    {
      "epoch": 10.67260579064588,
      "grad_norm": 14.135177612304688,
      "learning_rate": 1e-06,
      "loss": 0.6148,
      "num_input_tokens_seen": 268435256,
      "step": 4792
    },
    {
      "epoch": 10.67260579064588,
      "loss": 0.4595029354095459,
      "loss_ce": 0.00015233115118462592,
      "loss_iou": 0.2109375,
      "loss_num": 0.007232666015625,
      "loss_xval": 0.458984375,
      "num_input_tokens_seen": 268435256,
      "step": 4792
    },
    {
      "epoch": 10.674832962138085,
      "grad_norm": 20.141786575317383,
      "learning_rate": 1e-06,
      "loss": 0.7621,
      "num_input_tokens_seen": 268491724,
      "step": 4793
    },
    {
      "epoch": 10.674832962138085,
      "loss": 0.5482085943222046,
      "loss_ce": 0.00011288469249848276,
      "loss_iou": 0.236328125,
      "loss_num": 0.01507568359375,
      "loss_xval": 0.546875,
      "num_input_tokens_seen": 268491724,
      "step": 4793
    },
    {
      "epoch": 10.67706013363029,
      "grad_norm": 17.971179962158203,
      "learning_rate": 1e-06,
      "loss": 0.5321,
      "num_input_tokens_seen": 268548580,
      "step": 4794
    },
    {
      "epoch": 10.67706013363029,
      "loss": 0.5524182319641113,
      "loss_ce": 0.0001721275766612962,
      "loss_iou": 0.2353515625,
      "loss_num": 0.01611328125,
      "loss_xval": 0.55078125,
      "num_input_tokens_seen": 268548580,
      "step": 4794
    },
    {
      "epoch": 10.679287305122495,
      "grad_norm": 17.928714752197266,
      "learning_rate": 1e-06,
      "loss": 0.4135,
      "num_input_tokens_seen": 268604548,
      "step": 4795
    },
    {
      "epoch": 10.679287305122495,
      "loss": 0.3612457513809204,
      "loss_ce": 0.0001617687230464071,
      "loss_iou": 0.1669921875,
      "loss_num": 0.0054931640625,
      "loss_xval": 0.361328125,
      "num_input_tokens_seen": 268604548,
      "step": 4795
    },
    {
      "epoch": 10.6815144766147,
      "grad_norm": 17.0757999420166,
      "learning_rate": 1e-06,
      "loss": 0.5741,
      "num_input_tokens_seen": 268661340,
      "step": 4796
    },
    {
      "epoch": 10.6815144766147,
      "loss": 0.6327658891677856,
      "loss_ce": 0.0001975290069822222,
      "loss_iou": 0.255859375,
      "loss_num": 0.02392578125,
      "loss_xval": 0.6328125,
      "num_input_tokens_seen": 268661340,
      "step": 4796
    },
    {
      "epoch": 10.683741648106905,
      "grad_norm": 19.690837860107422,
      "learning_rate": 1e-06,
      "loss": 0.5216,
      "num_input_tokens_seen": 268715384,
      "step": 4797
    },
    {
      "epoch": 10.683741648106905,
      "loss": 0.3917364180088043,
      "loss_ce": 0.0001348454534308985,
      "loss_iou": 0.1806640625,
      "loss_num": 0.0059814453125,
      "loss_xval": 0.390625,
      "num_input_tokens_seen": 268715384,
      "step": 4797
    },
    {
      "epoch": 10.68596881959911,
      "grad_norm": 40.683292388916016,
      "learning_rate": 1e-06,
      "loss": 0.5789,
      "num_input_tokens_seen": 268772412,
      "step": 4798
    },
    {
      "epoch": 10.68596881959911,
      "loss": 0.6195793151855469,
      "loss_ce": 0.00019457080634310842,
      "loss_iou": 0.267578125,
      "loss_num": 0.0164794921875,
      "loss_xval": 0.62109375,
      "num_input_tokens_seen": 268772412,
      "step": 4798
    },
    {
      "epoch": 10.688195991091314,
      "grad_norm": 24.19279670715332,
      "learning_rate": 1e-06,
      "loss": 0.5381,
      "num_input_tokens_seen": 268830036,
      "step": 4799
    },
    {
      "epoch": 10.688195991091314,
      "loss": 0.528715193271637,
      "loss_ce": 0.0001507120905444026,
      "loss_iou": 0.244140625,
      "loss_num": 0.00811767578125,
      "loss_xval": 0.52734375,
      "num_input_tokens_seen": 268830036,
      "step": 4799
    },
    {
      "epoch": 10.690423162583519,
      "grad_norm": 15.297384262084961,
      "learning_rate": 1e-06,
      "loss": 0.5682,
      "num_input_tokens_seen": 268885000,
      "step": 4800
    },
    {
      "epoch": 10.690423162583519,
      "loss": 0.4110143184661865,
      "loss_ce": 0.00012564694043248892,
      "loss_iou": 0.1865234375,
      "loss_num": 0.00750732421875,
      "loss_xval": 0.41015625,
      "num_input_tokens_seen": 268885000,
      "step": 4800
    },
    {
      "epoch": 10.692650334075724,
      "grad_norm": 30.53861427307129,
      "learning_rate": 1e-06,
      "loss": 0.6299,
      "num_input_tokens_seen": 268941972,
      "step": 4801
    },
    {
      "epoch": 10.692650334075724,
      "loss": 0.45105427503585815,
      "loss_ce": 0.00012653737212531269,
      "loss_iou": 0.201171875,
      "loss_num": 0.009521484375,
      "loss_xval": 0.451171875,
      "num_input_tokens_seen": 268941972,
      "step": 4801
    },
    {
      "epoch": 10.694877505567929,
      "grad_norm": 22.213823318481445,
      "learning_rate": 1e-06,
      "loss": 0.4161,
      "num_input_tokens_seen": 268996852,
      "step": 4802
    },
    {
      "epoch": 10.694877505567929,
      "loss": 0.36823076009750366,
      "loss_ce": 0.00012772114132530987,
      "loss_iou": 0.1533203125,
      "loss_num": 0.0123291015625,
      "loss_xval": 0.3671875,
      "num_input_tokens_seen": 268996852,
      "step": 4802
    },
    {
      "epoch": 10.697104677060134,
      "grad_norm": 15.323545455932617,
      "learning_rate": 1e-06,
      "loss": 0.7173,
      "num_input_tokens_seen": 269052540,
      "step": 4803
    },
    {
      "epoch": 10.697104677060134,
      "loss": 0.6369800567626953,
      "loss_ce": 0.0002612703829072416,
      "loss_iou": 0.291015625,
      "loss_num": 0.0113525390625,
      "loss_xval": 0.63671875,
      "num_input_tokens_seen": 269052540,
      "step": 4803
    },
    {
      "epoch": 10.699331848552339,
      "grad_norm": 22.477542877197266,
      "learning_rate": 1e-06,
      "loss": 0.4287,
      "num_input_tokens_seen": 269107128,
      "step": 4804
    },
    {
      "epoch": 10.699331848552339,
      "loss": 0.31477898359298706,
      "loss_ce": 0.00014274932618718594,
      "loss_iou": 0.138671875,
      "loss_num": 0.007354736328125,
      "loss_xval": 0.314453125,
      "num_input_tokens_seen": 269107128,
      "step": 4804
    },
    {
      "epoch": 10.701559020044543,
      "grad_norm": 15.610451698303223,
      "learning_rate": 1e-06,
      "loss": 0.4464,
      "num_input_tokens_seen": 269162060,
      "step": 4805
    },
    {
      "epoch": 10.701559020044543,
      "loss": 0.3491445779800415,
      "loss_ce": 0.00020659035362768918,
      "loss_iou": 0.1552734375,
      "loss_num": 0.00775146484375,
      "loss_xval": 0.349609375,
      "num_input_tokens_seen": 269162060,
      "step": 4805
    },
    {
      "epoch": 10.703786191536748,
      "grad_norm": 28.197786331176758,
      "learning_rate": 1e-06,
      "loss": 0.6925,
      "num_input_tokens_seen": 269217312,
      "step": 4806
    },
    {
      "epoch": 10.703786191536748,
      "loss": 0.7023131847381592,
      "loss_ce": 0.0001647536118980497,
      "loss_iou": 0.30859375,
      "loss_num": 0.016845703125,
      "loss_xval": 0.703125,
      "num_input_tokens_seen": 269217312,
      "step": 4806
    },
    {
      "epoch": 10.706013363028953,
      "grad_norm": 15.568071365356445,
      "learning_rate": 1e-06,
      "loss": 0.3815,
      "num_input_tokens_seen": 269275728,
      "step": 4807
    },
    {
      "epoch": 10.706013363028953,
      "loss": 0.4796176552772522,
      "loss_ce": 0.00012549404345918447,
      "loss_iou": 0.220703125,
      "loss_num": 0.0074462890625,
      "loss_xval": 0.48046875,
      "num_input_tokens_seen": 269275728,
      "step": 4807
    },
    {
      "epoch": 10.708240534521158,
      "grad_norm": 17.49445915222168,
      "learning_rate": 1e-06,
      "loss": 0.4644,
      "num_input_tokens_seen": 269332812,
      "step": 4808
    },
    {
      "epoch": 10.708240534521158,
      "loss": 0.5414130687713623,
      "loss_ce": 0.0001532989408588037,
      "loss_iou": 0.2421875,
      "loss_num": 0.01116943359375,
      "loss_xval": 0.54296875,
      "num_input_tokens_seen": 269332812,
      "step": 4808
    },
    {
      "epoch": 10.710467706013363,
      "grad_norm": 28.943540573120117,
      "learning_rate": 1e-06,
      "loss": 0.5068,
      "num_input_tokens_seen": 269388828,
      "step": 4809
    },
    {
      "epoch": 10.710467706013363,
      "loss": 0.441672682762146,
      "loss_ce": 0.0001443543005734682,
      "loss_iou": 0.1923828125,
      "loss_num": 0.01123046875,
      "loss_xval": 0.44140625,
      "num_input_tokens_seen": 269388828,
      "step": 4809
    },
    {
      "epoch": 10.712694877505568,
      "grad_norm": 17.984169006347656,
      "learning_rate": 1e-06,
      "loss": 0.5561,
      "num_input_tokens_seen": 269446184,
      "step": 4810
    },
    {
      "epoch": 10.712694877505568,
      "loss": 0.5116062164306641,
      "loss_ce": 0.0001315802219323814,
      "loss_iou": 0.23828125,
      "loss_num": 0.007110595703125,
      "loss_xval": 0.51171875,
      "num_input_tokens_seen": 269446184,
      "step": 4810
    },
    {
      "epoch": 10.714922048997773,
      "grad_norm": 22.300188064575195,
      "learning_rate": 1e-06,
      "loss": 0.5555,
      "num_input_tokens_seen": 269502464,
      "step": 4811
    },
    {
      "epoch": 10.714922048997773,
      "loss": 0.6414833068847656,
      "loss_ce": 0.00012583400530274957,
      "loss_iou": 0.26171875,
      "loss_num": 0.0238037109375,
      "loss_xval": 0.640625,
      "num_input_tokens_seen": 269502464,
      "step": 4811
    },
    {
      "epoch": 10.717149220489977,
      "grad_norm": 22.90621566772461,
      "learning_rate": 1e-06,
      "loss": 0.5469,
      "num_input_tokens_seen": 269560400,
      "step": 4812
    },
    {
      "epoch": 10.717149220489977,
      "loss": 0.694524884223938,
      "loss_ce": 0.00018889480270445347,
      "loss_iou": 0.32421875,
      "loss_num": 0.009521484375,
      "loss_xval": 0.6953125,
      "num_input_tokens_seen": 269560400,
      "step": 4812
    },
    {
      "epoch": 10.719376391982182,
      "grad_norm": 20.077648162841797,
      "learning_rate": 1e-06,
      "loss": 0.665,
      "num_input_tokens_seen": 269617692,
      "step": 4813
    },
    {
      "epoch": 10.719376391982182,
      "loss": 0.9352070093154907,
      "loss_ce": 0.00014838551578577608,
      "loss_iou": 0.408203125,
      "loss_num": 0.023681640625,
      "loss_xval": 0.93359375,
      "num_input_tokens_seen": 269617692,
      "step": 4813
    },
    {
      "epoch": 10.721603563474387,
      "grad_norm": 45.49837875366211,
      "learning_rate": 1e-06,
      "loss": 0.6051,
      "num_input_tokens_seen": 269675808,
      "step": 4814
    },
    {
      "epoch": 10.721603563474387,
      "loss": 0.6009942889213562,
      "loss_ce": 0.00016420066822320223,
      "loss_iou": 0.263671875,
      "loss_num": 0.01446533203125,
      "loss_xval": 0.6015625,
      "num_input_tokens_seen": 269675808,
      "step": 4814
    },
    {
      "epoch": 10.723830734966592,
      "grad_norm": 18.847713470458984,
      "learning_rate": 1e-06,
      "loss": 0.6815,
      "num_input_tokens_seen": 269731972,
      "step": 4815
    },
    {
      "epoch": 10.723830734966592,
      "loss": 0.5592037439346313,
      "loss_ce": 0.00012171984417364001,
      "loss_iou": 0.2392578125,
      "loss_num": 0.01611328125,
      "loss_xval": 0.55859375,
      "num_input_tokens_seen": 269731972,
      "step": 4815
    },
    {
      "epoch": 10.726057906458797,
      "grad_norm": 21.027870178222656,
      "learning_rate": 1e-06,
      "loss": 0.6109,
      "num_input_tokens_seen": 269785520,
      "step": 4816
    },
    {
      "epoch": 10.726057906458797,
      "loss": 0.5723341107368469,
      "loss_ce": 0.0001295319088967517,
      "loss_iou": 0.251953125,
      "loss_num": 0.01373291015625,
      "loss_xval": 0.5703125,
      "num_input_tokens_seen": 269785520,
      "step": 4816
    },
    {
      "epoch": 10.728285077951002,
      "grad_norm": 22.105512619018555,
      "learning_rate": 1e-06,
      "loss": 0.4586,
      "num_input_tokens_seen": 269843136,
      "step": 4817
    },
    {
      "epoch": 10.728285077951002,
      "loss": 0.5098918676376343,
      "loss_ce": 0.0001262695004697889,
      "loss_iou": 0.220703125,
      "loss_num": 0.01373291015625,
      "loss_xval": 0.5078125,
      "num_input_tokens_seen": 269843136,
      "step": 4817
    },
    {
      "epoch": 10.730512249443207,
      "grad_norm": 20.30752944946289,
      "learning_rate": 1e-06,
      "loss": 0.6458,
      "num_input_tokens_seen": 269897972,
      "step": 4818
    },
    {
      "epoch": 10.730512249443207,
      "loss": 0.8840016722679138,
      "loss_ce": 0.00021258596098050475,
      "loss_iou": 0.3671875,
      "loss_num": 0.029541015625,
      "loss_xval": 0.8828125,
      "num_input_tokens_seen": 269897972,
      "step": 4818
    },
    {
      "epoch": 10.732739420935411,
      "grad_norm": 19.499021530151367,
      "learning_rate": 1e-06,
      "loss": 0.4868,
      "num_input_tokens_seen": 269955164,
      "step": 4819
    },
    {
      "epoch": 10.732739420935411,
      "loss": 0.4875670373439789,
      "loss_ce": 0.0001403049536747858,
      "loss_iou": 0.1865234375,
      "loss_num": 0.022705078125,
      "loss_xval": 0.48828125,
      "num_input_tokens_seen": 269955164,
      "step": 4819
    },
    {
      "epoch": 10.734966592427616,
      "grad_norm": 22.588951110839844,
      "learning_rate": 1e-06,
      "loss": 0.5172,
      "num_input_tokens_seen": 270014316,
      "step": 4820
    },
    {
      "epoch": 10.734966592427616,
      "loss": 0.3510775864124298,
      "loss_ce": 0.00012545072240754962,
      "loss_iou": 0.1533203125,
      "loss_num": 0.00885009765625,
      "loss_xval": 0.3515625,
      "num_input_tokens_seen": 270014316,
      "step": 4820
    },
    {
      "epoch": 10.737193763919821,
      "grad_norm": 20.028270721435547,
      "learning_rate": 1e-06,
      "loss": 0.6092,
      "num_input_tokens_seen": 270070980,
      "step": 4821
    },
    {
      "epoch": 10.737193763919821,
      "loss": 0.6935439109802246,
      "loss_ce": 0.00018454053497407585,
      "loss_iou": 0.29296875,
      "loss_num": 0.021240234375,
      "loss_xval": 0.6953125,
      "num_input_tokens_seen": 270070980,
      "step": 4821
    },
    {
      "epoch": 10.739420935412026,
      "grad_norm": 14.508070945739746,
      "learning_rate": 1e-06,
      "loss": 0.5061,
      "num_input_tokens_seen": 270128576,
      "step": 4822
    },
    {
      "epoch": 10.739420935412026,
      "loss": 0.6202830672264099,
      "loss_ce": 0.00016589835286140442,
      "loss_iou": 0.2578125,
      "loss_num": 0.0208740234375,
      "loss_xval": 0.62109375,
      "num_input_tokens_seen": 270128576,
      "step": 4822
    },
    {
      "epoch": 10.74164810690423,
      "grad_norm": 17.07376480102539,
      "learning_rate": 1e-06,
      "loss": 0.7271,
      "num_input_tokens_seen": 270184516,
      "step": 4823
    },
    {
      "epoch": 10.74164810690423,
      "loss": 0.64375239610672,
      "loss_ce": 0.00019769492791965604,
      "loss_iou": 0.298828125,
      "loss_num": 0.0089111328125,
      "loss_xval": 0.64453125,
      "num_input_tokens_seen": 270184516,
      "step": 4823
    },
    {
      "epoch": 10.743875278396436,
      "grad_norm": 20.712217330932617,
      "learning_rate": 1e-06,
      "loss": 0.5371,
      "num_input_tokens_seen": 270240064,
      "step": 4824
    },
    {
      "epoch": 10.743875278396436,
      "loss": 0.5007485747337341,
      "loss_ce": 0.0001382490445394069,
      "loss_iou": 0.216796875,
      "loss_num": 0.01312255859375,
      "loss_xval": 0.5,
      "num_input_tokens_seen": 270240064,
      "step": 4824
    },
    {
      "epoch": 10.74610244988864,
      "grad_norm": 20.345590591430664,
      "learning_rate": 1e-06,
      "loss": 0.5216,
      "num_input_tokens_seen": 270297036,
      "step": 4825
    },
    {
      "epoch": 10.74610244988864,
      "loss": 0.6117961406707764,
      "loss_ce": 0.00022386244381777942,
      "loss_iou": 0.267578125,
      "loss_num": 0.015625,
      "loss_xval": 0.61328125,
      "num_input_tokens_seen": 270297036,
      "step": 4825
    },
    {
      "epoch": 10.748329621380847,
      "grad_norm": 19.20088768005371,
      "learning_rate": 1e-06,
      "loss": 0.5529,
      "num_input_tokens_seen": 270352788,
      "step": 4826
    },
    {
      "epoch": 10.748329621380847,
      "loss": 0.6521322131156921,
      "loss_ce": 0.00015466928016394377,
      "loss_iou": 0.30078125,
      "loss_num": 0.01025390625,
      "loss_xval": 0.65234375,
      "num_input_tokens_seen": 270352788,
      "step": 4826
    },
    {
      "epoch": 10.750556792873052,
      "grad_norm": 16.478546142578125,
      "learning_rate": 1e-06,
      "loss": 0.458,
      "num_input_tokens_seen": 270407444,
      "step": 4827
    },
    {
      "epoch": 10.750556792873052,
      "loss": 0.5561587810516357,
      "loss_ce": 0.00012851686915382743,
      "loss_iou": 0.234375,
      "loss_num": 0.017578125,
      "loss_xval": 0.5546875,
      "num_input_tokens_seen": 270407444,
      "step": 4827
    },
    {
      "epoch": 10.752783964365257,
      "grad_norm": 30.999256134033203,
      "learning_rate": 1e-06,
      "loss": 0.4163,
      "num_input_tokens_seen": 270465300,
      "step": 4828
    },
    {
      "epoch": 10.752783964365257,
      "loss": 0.41544434428215027,
      "loss_ce": 0.00016115896869450808,
      "loss_iou": 0.1865234375,
      "loss_num": 0.00830078125,
      "loss_xval": 0.416015625,
      "num_input_tokens_seen": 270465300,
      "step": 4828
    },
    {
      "epoch": 10.755011135857462,
      "grad_norm": 21.072765350341797,
      "learning_rate": 1e-06,
      "loss": 0.507,
      "num_input_tokens_seen": 270524188,
      "step": 4829
    },
    {
      "epoch": 10.755011135857462,
      "loss": 0.682302713394165,
      "loss_ce": 0.0005400052759796381,
      "loss_iou": 0.294921875,
      "loss_num": 0.018310546875,
      "loss_xval": 0.68359375,
      "num_input_tokens_seen": 270524188,
      "step": 4829
    },
    {
      "epoch": 10.757238307349667,
      "grad_norm": 25.556285858154297,
      "learning_rate": 1e-06,
      "loss": 0.6146,
      "num_input_tokens_seen": 270580156,
      "step": 4830
    },
    {
      "epoch": 10.757238307349667,
      "loss": 0.43665558099746704,
      "loss_ce": 0.000132172426674515,
      "loss_iou": 0.189453125,
      "loss_num": 0.01165771484375,
      "loss_xval": 0.4375,
      "num_input_tokens_seen": 270580156,
      "step": 4830
    },
    {
      "epoch": 10.759465478841872,
      "grad_norm": 39.73129653930664,
      "learning_rate": 1e-06,
      "loss": 0.6867,
      "num_input_tokens_seen": 270637276,
      "step": 4831
    },
    {
      "epoch": 10.759465478841872,
      "loss": 0.7980098128318787,
      "loss_ce": 0.00015826240996830165,
      "loss_iou": 0.328125,
      "loss_num": 0.02783203125,
      "loss_xval": 0.796875,
      "num_input_tokens_seen": 270637276,
      "step": 4831
    },
    {
      "epoch": 10.761692650334076,
      "grad_norm": 31.106664657592773,
      "learning_rate": 1e-06,
      "loss": 0.5331,
      "num_input_tokens_seen": 270691744,
      "step": 4832
    },
    {
      "epoch": 10.761692650334076,
      "loss": 0.45221608877182007,
      "loss_ce": 0.00012868586054537445,
      "loss_iou": 0.1953125,
      "loss_num": 0.01226806640625,
      "loss_xval": 0.451171875,
      "num_input_tokens_seen": 270691744,
      "step": 4832
    },
    {
      "epoch": 10.763919821826281,
      "grad_norm": 19.635936737060547,
      "learning_rate": 1e-06,
      "loss": 0.4175,
      "num_input_tokens_seen": 270749204,
      "step": 4833
    },
    {
      "epoch": 10.763919821826281,
      "loss": 0.510043740272522,
      "loss_ce": 0.00015601412451360375,
      "loss_iou": 0.2197265625,
      "loss_num": 0.01397705078125,
      "loss_xval": 0.51171875,
      "num_input_tokens_seen": 270749204,
      "step": 4833
    },
    {
      "epoch": 10.766146993318486,
      "grad_norm": 30.472660064697266,
      "learning_rate": 1e-06,
      "loss": 0.6393,
      "num_input_tokens_seen": 270803020,
      "step": 4834
    },
    {
      "epoch": 10.766146993318486,
      "loss": 0.6473684906959534,
      "loss_ce": 0.0001517097553005442,
      "loss_iou": 0.28125,
      "loss_num": 0.0172119140625,
      "loss_xval": 0.6484375,
      "num_input_tokens_seen": 270803020,
      "step": 4834
    },
    {
      "epoch": 10.768374164810691,
      "grad_norm": 18.92053985595703,
      "learning_rate": 1e-06,
      "loss": 0.6669,
      "num_input_tokens_seen": 270859752,
      "step": 4835
    },
    {
      "epoch": 10.768374164810691,
      "loss": 0.4643715023994446,
      "loss_ce": 0.00013810490781906992,
      "loss_iou": 0.2138671875,
      "loss_num": 0.007476806640625,
      "loss_xval": 0.46484375,
      "num_input_tokens_seen": 270859752,
      "step": 4835
    },
    {
      "epoch": 10.770601336302896,
      "grad_norm": 16.783376693725586,
      "learning_rate": 1e-06,
      "loss": 0.4961,
      "num_input_tokens_seen": 270916024,
      "step": 4836
    },
    {
      "epoch": 10.770601336302896,
      "loss": 0.640272319316864,
      "loss_ce": 0.0001356131979264319,
      "loss_iou": 0.2734375,
      "loss_num": 0.018798828125,
      "loss_xval": 0.640625,
      "num_input_tokens_seen": 270916024,
      "step": 4836
    },
    {
      "epoch": 10.7728285077951,
      "grad_norm": 28.108156204223633,
      "learning_rate": 1e-06,
      "loss": 0.4965,
      "num_input_tokens_seen": 270972228,
      "step": 4837
    },
    {
      "epoch": 10.7728285077951,
      "loss": 0.6063258051872253,
      "loss_ce": 0.00012462104496080428,
      "loss_iou": 0.25,
      "loss_num": 0.021484375,
      "loss_xval": 0.60546875,
      "num_input_tokens_seen": 270972228,
      "step": 4837
    },
    {
      "epoch": 10.775055679287306,
      "grad_norm": 14.273365020751953,
      "learning_rate": 1e-06,
      "loss": 0.396,
      "num_input_tokens_seen": 271028976,
      "step": 4838
    },
    {
      "epoch": 10.775055679287306,
      "loss": 0.40957385301589966,
      "loss_ce": 0.00015002323198132217,
      "loss_iou": 0.19140625,
      "loss_num": 0.00531005859375,
      "loss_xval": 0.41015625,
      "num_input_tokens_seen": 271028976,
      "step": 4838
    },
    {
      "epoch": 10.77728285077951,
      "grad_norm": 16.653898239135742,
      "learning_rate": 1e-06,
      "loss": 0.4034,
      "num_input_tokens_seen": 271086780,
      "step": 4839
    },
    {
      "epoch": 10.77728285077951,
      "loss": 0.32855719327926636,
      "loss_ce": 0.0001270081556867808,
      "loss_iou": 0.1435546875,
      "loss_num": 0.0084228515625,
      "loss_xval": 0.328125,
      "num_input_tokens_seen": 271086780,
      "step": 4839
    },
    {
      "epoch": 10.779510022271715,
      "grad_norm": 22.08172607421875,
      "learning_rate": 1e-06,
      "loss": 0.6062,
      "num_input_tokens_seen": 271142040,
      "step": 4840
    },
    {
      "epoch": 10.779510022271715,
      "loss": 0.6195139288902283,
      "loss_ce": 0.00012916022387798876,
      "loss_iou": 0.267578125,
      "loss_num": 0.0166015625,
      "loss_xval": 0.62109375,
      "num_input_tokens_seen": 271142040,
      "step": 4840
    },
    {
      "epoch": 10.78173719376392,
      "grad_norm": 31.35270881652832,
      "learning_rate": 1e-06,
      "loss": 0.4893,
      "num_input_tokens_seen": 271198064,
      "step": 4841
    },
    {
      "epoch": 10.78173719376392,
      "loss": 0.6805669069290161,
      "loss_ce": 0.00014694841229356825,
      "loss_iou": 0.29296875,
      "loss_num": 0.019287109375,
      "loss_xval": 0.6796875,
      "num_input_tokens_seen": 271198064,
      "step": 4841
    },
    {
      "epoch": 10.783964365256125,
      "grad_norm": 19.86975860595703,
      "learning_rate": 1e-06,
      "loss": 0.5828,
      "num_input_tokens_seen": 271253592,
      "step": 4842
    },
    {
      "epoch": 10.783964365256125,
      "loss": 0.6472707390785217,
      "loss_ce": 0.00020657834829762578,
      "loss_iou": 0.2353515625,
      "loss_num": 0.035400390625,
      "loss_xval": 0.6484375,
      "num_input_tokens_seen": 271253592,
      "step": 4842
    },
    {
      "epoch": 10.78619153674833,
      "grad_norm": 21.59044075012207,
      "learning_rate": 1e-06,
      "loss": 0.7014,
      "num_input_tokens_seen": 271310024,
      "step": 4843
    },
    {
      "epoch": 10.78619153674833,
      "loss": 0.7749493718147278,
      "loss_ce": 0.0002301402564626187,
      "loss_iou": 0.306640625,
      "loss_num": 0.03271484375,
      "loss_xval": 0.7734375,
      "num_input_tokens_seen": 271310024,
      "step": 4843
    },
    {
      "epoch": 10.788418708240535,
      "grad_norm": 20.93609619140625,
      "learning_rate": 1e-06,
      "loss": 0.4648,
      "num_input_tokens_seen": 271361888,
      "step": 4844
    },
    {
      "epoch": 10.788418708240535,
      "loss": 0.5555073022842407,
      "loss_ce": 0.00020940190006513149,
      "loss_iou": 0.23046875,
      "loss_num": 0.0191650390625,
      "loss_xval": 0.5546875,
      "num_input_tokens_seen": 271361888,
      "step": 4844
    },
    {
      "epoch": 10.79064587973274,
      "grad_norm": 35.37983322143555,
      "learning_rate": 1e-06,
      "loss": 0.5951,
      "num_input_tokens_seen": 271417132,
      "step": 4845
    },
    {
      "epoch": 10.79064587973274,
      "loss": 0.44118914008140564,
      "loss_ce": 0.0001490934519097209,
      "loss_iou": 0.197265625,
      "loss_num": 0.00927734375,
      "loss_xval": 0.44140625,
      "num_input_tokens_seen": 271417132,
      "step": 4845
    },
    {
      "epoch": 10.792873051224944,
      "grad_norm": 19.222061157226562,
      "learning_rate": 1e-06,
      "loss": 0.5955,
      "num_input_tokens_seen": 271471928,
      "step": 4846
    },
    {
      "epoch": 10.792873051224944,
      "loss": 0.6026766300201416,
      "loss_ce": 0.00013759825378656387,
      "loss_iou": 0.26171875,
      "loss_num": 0.0155029296875,
      "loss_xval": 0.6015625,
      "num_input_tokens_seen": 271471928,
      "step": 4846
    },
    {
      "epoch": 10.79510022271715,
      "grad_norm": 17.895984649658203,
      "learning_rate": 1e-06,
      "loss": 0.5379,
      "num_input_tokens_seen": 271526128,
      "step": 4847
    },
    {
      "epoch": 10.79510022271715,
      "loss": 0.4887864291667938,
      "loss_ce": 0.00013897617463953793,
      "loss_iou": 0.21875,
      "loss_num": 0.01019287109375,
      "loss_xval": 0.48828125,
      "num_input_tokens_seen": 271526128,
      "step": 4847
    },
    {
      "epoch": 10.797327394209354,
      "grad_norm": 22.95305061340332,
      "learning_rate": 1e-06,
      "loss": 0.5451,
      "num_input_tokens_seen": 271581608,
      "step": 4848
    },
    {
      "epoch": 10.797327394209354,
      "loss": 0.42127272486686707,
      "loss_ce": 0.00013015880540478975,
      "loss_iou": 0.1923828125,
      "loss_num": 0.007415771484375,
      "loss_xval": 0.421875,
      "num_input_tokens_seen": 271581608,
      "step": 4848
    },
    {
      "epoch": 10.799554565701559,
      "grad_norm": 28.807262420654297,
      "learning_rate": 1e-06,
      "loss": 0.5903,
      "num_input_tokens_seen": 271638680,
      "step": 4849
    },
    {
      "epoch": 10.799554565701559,
      "loss": 0.8107374310493469,
      "loss_ce": 0.00019058278121519834,
      "loss_iou": 0.345703125,
      "loss_num": 0.024169921875,
      "loss_xval": 0.8125,
      "num_input_tokens_seen": 271638680,
      "step": 4849
    },
    {
      "epoch": 10.801781737193764,
      "grad_norm": 22.489822387695312,
      "learning_rate": 1e-06,
      "loss": 0.4126,
      "num_input_tokens_seen": 271694712,
      "step": 4850
    },
    {
      "epoch": 10.801781737193764,
      "loss": 0.43495360016822815,
      "loss_ce": 0.000139150011818856,
      "loss_iou": 0.1904296875,
      "loss_num": 0.01080322265625,
      "loss_xval": 0.435546875,
      "num_input_tokens_seen": 271694712,
      "step": 4850
    },
    {
      "epoch": 10.804008908685969,
      "grad_norm": 19.50384521484375,
      "learning_rate": 1e-06,
      "loss": 0.4934,
      "num_input_tokens_seen": 271750692,
      "step": 4851
    },
    {
      "epoch": 10.804008908685969,
      "loss": 0.36242765188217163,
      "loss_ce": 0.00012293207691982388,
      "loss_iou": 0.1611328125,
      "loss_num": 0.00799560546875,
      "loss_xval": 0.36328125,
      "num_input_tokens_seen": 271750692,
      "step": 4851
    },
    {
      "epoch": 10.806236080178174,
      "grad_norm": 20.99281883239746,
      "learning_rate": 1e-06,
      "loss": 0.5668,
      "num_input_tokens_seen": 271807552,
      "step": 4852
    },
    {
      "epoch": 10.806236080178174,
      "loss": 0.4213836193084717,
      "loss_ce": 0.00011898807133547962,
      "loss_iou": 0.1708984375,
      "loss_num": 0.0159912109375,
      "loss_xval": 0.421875,
      "num_input_tokens_seen": 271807552,
      "step": 4852
    },
    {
      "epoch": 10.808463251670378,
      "grad_norm": 14.47896957397461,
      "learning_rate": 1e-06,
      "loss": 0.5013,
      "num_input_tokens_seen": 271862320,
      "step": 4853
    },
    {
      "epoch": 10.808463251670378,
      "loss": 0.608984112739563,
      "loss_ce": 9.740462701302022e-05,
      "loss_iou": 0.21875,
      "loss_num": 0.0341796875,
      "loss_xval": 0.609375,
      "num_input_tokens_seen": 271862320,
      "step": 4853
    },
    {
      "epoch": 10.810690423162583,
      "grad_norm": 25.232343673706055,
      "learning_rate": 1e-06,
      "loss": 0.7236,
      "num_input_tokens_seen": 271916944,
      "step": 4854
    },
    {
      "epoch": 10.810690423162583,
      "loss": 0.6214773654937744,
      "loss_ce": 0.00013948752894066274,
      "loss_iou": 0.265625,
      "loss_num": 0.017578125,
      "loss_xval": 0.62109375,
      "num_input_tokens_seen": 271916944,
      "step": 4854
    },
    {
      "epoch": 10.812917594654788,
      "grad_norm": 24.822452545166016,
      "learning_rate": 1e-06,
      "loss": 0.4163,
      "num_input_tokens_seen": 271972016,
      "step": 4855
    },
    {
      "epoch": 10.812917594654788,
      "loss": 0.461017906665802,
      "loss_ce": 0.0004466102982405573,
      "loss_iou": 0.2080078125,
      "loss_num": 0.00897216796875,
      "loss_xval": 0.4609375,
      "num_input_tokens_seen": 271972016,
      "step": 4855
    },
    {
      "epoch": 10.815144766146993,
      "grad_norm": 17.059764862060547,
      "learning_rate": 1e-06,
      "loss": 0.5627,
      "num_input_tokens_seen": 272026808,
      "step": 4856
    },
    {
      "epoch": 10.815144766146993,
      "loss": 0.7428523898124695,
      "loss_ce": 0.0001765760825946927,
      "loss_iou": 0.3203125,
      "loss_num": 0.020263671875,
      "loss_xval": 0.7421875,
      "num_input_tokens_seen": 272026808,
      "step": 4856
    },
    {
      "epoch": 10.817371937639198,
      "grad_norm": 22.709232330322266,
      "learning_rate": 1e-06,
      "loss": 0.5341,
      "num_input_tokens_seen": 272080680,
      "step": 4857
    },
    {
      "epoch": 10.817371937639198,
      "loss": 0.6549853086471558,
      "loss_ce": 0.0002001727989409119,
      "loss_iou": 0.275390625,
      "loss_num": 0.0206298828125,
      "loss_xval": 0.65625,
      "num_input_tokens_seen": 272080680,
      "step": 4857
    },
    {
      "epoch": 10.819599109131403,
      "grad_norm": 19.562105178833008,
      "learning_rate": 1e-06,
      "loss": 0.5892,
      "num_input_tokens_seen": 272137808,
      "step": 4858
    },
    {
      "epoch": 10.819599109131403,
      "loss": 0.5621104836463928,
      "loss_ce": 0.00022084277588874102,
      "loss_iou": 0.2275390625,
      "loss_num": 0.0213623046875,
      "loss_xval": 0.5625,
      "num_input_tokens_seen": 272137808,
      "step": 4858
    },
    {
      "epoch": 10.821826280623608,
      "grad_norm": 11.724193572998047,
      "learning_rate": 1e-06,
      "loss": 0.4827,
      "num_input_tokens_seen": 272196140,
      "step": 4859
    },
    {
      "epoch": 10.821826280623608,
      "loss": 0.531132161617279,
      "loss_ce": 0.00024835762451402843,
      "loss_iou": 0.197265625,
      "loss_num": 0.0274658203125,
      "loss_xval": 0.53125,
      "num_input_tokens_seen": 272196140,
      "step": 4859
    },
    {
      "epoch": 10.824053452115812,
      "grad_norm": 26.051990509033203,
      "learning_rate": 1e-06,
      "loss": 0.7094,
      "num_input_tokens_seen": 272252688,
      "step": 4860
    },
    {
      "epoch": 10.824053452115812,
      "loss": 0.38587331771850586,
      "loss_ce": 0.0001311138621531427,
      "loss_iou": 0.17578125,
      "loss_num": 0.00677490234375,
      "loss_xval": 0.38671875,
      "num_input_tokens_seen": 272252688,
      "step": 4860
    },
    {
      "epoch": 10.826280623608017,
      "grad_norm": 19.309463500976562,
      "learning_rate": 1e-06,
      "loss": 0.4197,
      "num_input_tokens_seen": 272307848,
      "step": 4861
    },
    {
      "epoch": 10.826280623608017,
      "loss": 0.42674195766448975,
      "loss_ce": 0.00010623160051181912,
      "loss_iou": 0.177734375,
      "loss_num": 0.01416015625,
      "loss_xval": 0.42578125,
      "num_input_tokens_seen": 272307848,
      "step": 4861
    },
    {
      "epoch": 10.828507795100222,
      "grad_norm": 25.22977066040039,
      "learning_rate": 1e-06,
      "loss": 0.6507,
      "num_input_tokens_seen": 272364068,
      "step": 4862
    },
    {
      "epoch": 10.828507795100222,
      "loss": 0.6106005311012268,
      "loss_ce": 0.00024895532988011837,
      "loss_iou": 0.251953125,
      "loss_num": 0.0211181640625,
      "loss_xval": 0.609375,
      "num_input_tokens_seen": 272364068,
      "step": 4862
    },
    {
      "epoch": 10.830734966592427,
      "grad_norm": 13.005237579345703,
      "learning_rate": 1e-06,
      "loss": 0.6308,
      "num_input_tokens_seen": 272421856,
      "step": 4863
    },
    {
      "epoch": 10.830734966592427,
      "loss": 0.5312567949295044,
      "loss_ce": 0.00012890055950265378,
      "loss_iou": 0.208984375,
      "loss_num": 0.0225830078125,
      "loss_xval": 0.53125,
      "num_input_tokens_seen": 272421856,
      "step": 4863
    },
    {
      "epoch": 10.832962138084632,
      "grad_norm": 14.533258438110352,
      "learning_rate": 1e-06,
      "loss": 0.4557,
      "num_input_tokens_seen": 272479860,
      "step": 4864
    },
    {
      "epoch": 10.832962138084632,
      "loss": 0.4337289333343506,
      "loss_ce": 0.000135167152620852,
      "loss_iou": 0.193359375,
      "loss_num": 0.00933837890625,
      "loss_xval": 0.43359375,
      "num_input_tokens_seen": 272479860,
      "step": 4864
    },
    {
      "epoch": 10.835189309576837,
      "grad_norm": 16.71564292907715,
      "learning_rate": 1e-06,
      "loss": 0.4138,
      "num_input_tokens_seen": 272534472,
      "step": 4865
    },
    {
      "epoch": 10.835189309576837,
      "loss": 0.5382235646247864,
      "loss_ce": 0.0001376058644382283,
      "loss_iou": 0.244140625,
      "loss_num": 0.01007080078125,
      "loss_xval": 0.5390625,
      "num_input_tokens_seen": 272534472,
      "step": 4865
    },
    {
      "epoch": 10.837416481069042,
      "grad_norm": 35.24814987182617,
      "learning_rate": 1e-06,
      "loss": 0.5191,
      "num_input_tokens_seen": 272591164,
      "step": 4866
    },
    {
      "epoch": 10.837416481069042,
      "loss": 0.4814545214176178,
      "loss_ce": 0.00013124944234732538,
      "loss_iou": 0.197265625,
      "loss_num": 0.017333984375,
      "loss_xval": 0.48046875,
      "num_input_tokens_seen": 272591164,
      "step": 4866
    },
    {
      "epoch": 10.839643652561247,
      "grad_norm": 13.28064250946045,
      "learning_rate": 1e-06,
      "loss": 0.4789,
      "num_input_tokens_seen": 272647484,
      "step": 4867
    },
    {
      "epoch": 10.839643652561247,
      "loss": 0.4676069915294647,
      "loss_ce": 0.00013870664406567812,
      "loss_iou": 0.201171875,
      "loss_num": 0.01287841796875,
      "loss_xval": 0.466796875,
      "num_input_tokens_seen": 272647484,
      "step": 4867
    },
    {
      "epoch": 10.841870824053451,
      "grad_norm": 20.322200775146484,
      "learning_rate": 1e-06,
      "loss": 0.5039,
      "num_input_tokens_seen": 272703104,
      "step": 4868
    },
    {
      "epoch": 10.841870824053451,
      "loss": 0.5150094628334045,
      "loss_ce": 0.0005441223620437086,
      "loss_iou": 0.2138671875,
      "loss_num": 0.017333984375,
      "loss_xval": 0.515625,
      "num_input_tokens_seen": 272703104,
      "step": 4868
    },
    {
      "epoch": 10.844097995545656,
      "grad_norm": 24.41085433959961,
      "learning_rate": 1e-06,
      "loss": 0.5288,
      "num_input_tokens_seen": 272760632,
      "step": 4869
    },
    {
      "epoch": 10.844097995545656,
      "loss": 0.6352982521057129,
      "loss_ce": 0.00016645353753119707,
      "loss_iou": 0.25390625,
      "loss_num": 0.0252685546875,
      "loss_xval": 0.63671875,
      "num_input_tokens_seen": 272760632,
      "step": 4869
    },
    {
      "epoch": 10.846325167037861,
      "grad_norm": 26.662748336791992,
      "learning_rate": 1e-06,
      "loss": 0.5393,
      "num_input_tokens_seen": 272817712,
      "step": 4870
    },
    {
      "epoch": 10.846325167037861,
      "loss": 0.3460940420627594,
      "loss_ce": 0.00014678272418677807,
      "loss_iou": 0.158203125,
      "loss_num": 0.0059814453125,
      "loss_xval": 0.345703125,
      "num_input_tokens_seen": 272817712,
      "step": 4870
    },
    {
      "epoch": 10.848552338530066,
      "grad_norm": 24.583892822265625,
      "learning_rate": 1e-06,
      "loss": 0.7168,
      "num_input_tokens_seen": 272871604,
      "step": 4871
    },
    {
      "epoch": 10.848552338530066,
      "loss": 0.7562465667724609,
      "loss_ce": 0.0001430131815141067,
      "loss_iou": 0.3125,
      "loss_num": 0.0263671875,
      "loss_xval": 0.7578125,
      "num_input_tokens_seen": 272871604,
      "step": 4871
    },
    {
      "epoch": 10.85077951002227,
      "grad_norm": 21.29097557067871,
      "learning_rate": 1e-06,
      "loss": 0.4865,
      "num_input_tokens_seen": 272924804,
      "step": 4872
    },
    {
      "epoch": 10.85077951002227,
      "loss": 0.5399699211120605,
      "loss_ce": 0.00017501205729786307,
      "loss_iou": 0.2333984375,
      "loss_num": 0.014404296875,
      "loss_xval": 0.5390625,
      "num_input_tokens_seen": 272924804,
      "step": 4872
    },
    {
      "epoch": 10.853006681514476,
      "grad_norm": 24.785858154296875,
      "learning_rate": 1e-06,
      "loss": 0.7053,
      "num_input_tokens_seen": 272979376,
      "step": 4873
    },
    {
      "epoch": 10.853006681514476,
      "loss": 0.7309668660163879,
      "loss_ce": 0.0002539680281188339,
      "loss_iou": 0.322265625,
      "loss_num": 0.01708984375,
      "loss_xval": 0.73046875,
      "num_input_tokens_seen": 272979376,
      "step": 4873
    },
    {
      "epoch": 10.855233853006682,
      "grad_norm": 21.17921257019043,
      "learning_rate": 1e-06,
      "loss": 0.4353,
      "num_input_tokens_seen": 273037616,
      "step": 4874
    },
    {
      "epoch": 10.855233853006682,
      "loss": 0.5316353440284729,
      "loss_ce": 0.00014121364802122116,
      "loss_iou": 0.2255859375,
      "loss_num": 0.0162353515625,
      "loss_xval": 0.53125,
      "num_input_tokens_seen": 273037616,
      "step": 4874
    },
    {
      "epoch": 10.857461024498887,
      "grad_norm": 21.799673080444336,
      "learning_rate": 1e-06,
      "loss": 0.8208,
      "num_input_tokens_seen": 273097464,
      "step": 4875
    },
    {
      "epoch": 10.857461024498887,
      "loss": 0.9345442652702332,
      "loss_ce": 0.00021809733880218118,
      "loss_iou": 0.341796875,
      "loss_num": 0.0498046875,
      "loss_xval": 0.93359375,
      "num_input_tokens_seen": 273097464,
      "step": 4875
    },
    {
      "epoch": 10.859688195991092,
      "grad_norm": 20.746036529541016,
      "learning_rate": 1e-06,
      "loss": 0.5591,
      "num_input_tokens_seen": 273152000,
      "step": 4876
    },
    {
      "epoch": 10.859688195991092,
      "loss": 0.5373209118843079,
      "loss_ce": 0.00021149568783584982,
      "loss_iou": 0.2431640625,
      "loss_num": 0.01043701171875,
      "loss_xval": 0.5390625,
      "num_input_tokens_seen": 273152000,
      "step": 4876
    },
    {
      "epoch": 10.861915367483297,
      "grad_norm": 16.878854751586914,
      "learning_rate": 1e-06,
      "loss": 0.5007,
      "num_input_tokens_seen": 273209936,
      "step": 4877
    },
    {
      "epoch": 10.861915367483297,
      "loss": 0.5112770795822144,
      "loss_ce": 0.0001686862960923463,
      "loss_iou": 0.20703125,
      "loss_num": 0.019287109375,
      "loss_xval": 0.51171875,
      "num_input_tokens_seen": 273209936,
      "step": 4877
    },
    {
      "epoch": 10.864142538975502,
      "grad_norm": 17.051406860351562,
      "learning_rate": 1e-06,
      "loss": 0.4679,
      "num_input_tokens_seen": 273264832,
      "step": 4878
    },
    {
      "epoch": 10.864142538975502,
      "loss": 0.5240702629089355,
      "loss_ce": 0.00014447516878135502,
      "loss_iou": 0.220703125,
      "loss_num": 0.0166015625,
      "loss_xval": 0.5234375,
      "num_input_tokens_seen": 273264832,
      "step": 4878
    },
    {
      "epoch": 10.866369710467707,
      "grad_norm": 19.403690338134766,
      "learning_rate": 1e-06,
      "loss": 0.7818,
      "num_input_tokens_seen": 273321468,
      "step": 4879
    },
    {
      "epoch": 10.866369710467707,
      "loss": 0.8524916172027588,
      "loss_ce": 0.00013568377471528947,
      "loss_iou": 0.345703125,
      "loss_num": 0.0322265625,
      "loss_xval": 0.8515625,
      "num_input_tokens_seen": 273321468,
      "step": 4879
    },
    {
      "epoch": 10.868596881959911,
      "grad_norm": 17.883703231811523,
      "learning_rate": 1e-06,
      "loss": 0.589,
      "num_input_tokens_seen": 273378420,
      "step": 4880
    },
    {
      "epoch": 10.868596881959911,
      "loss": 0.49145275354385376,
      "loss_ce": 0.00024180561013054103,
      "loss_iou": 0.2236328125,
      "loss_num": 0.0089111328125,
      "loss_xval": 0.4921875,
      "num_input_tokens_seen": 273378420,
      "step": 4880
    },
    {
      "epoch": 10.870824053452116,
      "grad_norm": 29.07217025756836,
      "learning_rate": 1e-06,
      "loss": 0.4582,
      "num_input_tokens_seen": 273433924,
      "step": 4881
    },
    {
      "epoch": 10.870824053452116,
      "loss": 0.4086874723434448,
      "loss_ce": 0.00011811024160124362,
      "loss_iou": 0.1748046875,
      "loss_num": 0.01171875,
      "loss_xval": 0.408203125,
      "num_input_tokens_seen": 273433924,
      "step": 4881
    },
    {
      "epoch": 10.873051224944321,
      "grad_norm": 22.693309783935547,
      "learning_rate": 1e-06,
      "loss": 0.4058,
      "num_input_tokens_seen": 273488764,
      "step": 4882
    },
    {
      "epoch": 10.873051224944321,
      "loss": 0.5119615197181702,
      "loss_ce": 0.00012070569937350228,
      "loss_iou": 0.21875,
      "loss_num": 0.0147705078125,
      "loss_xval": 0.51171875,
      "num_input_tokens_seen": 273488764,
      "step": 4882
    },
    {
      "epoch": 10.875278396436526,
      "grad_norm": 16.45703125,
      "learning_rate": 1e-06,
      "loss": 0.5703,
      "num_input_tokens_seen": 273547644,
      "step": 4883
    },
    {
      "epoch": 10.875278396436526,
      "loss": 0.5592530965805054,
      "loss_ce": 0.00017106565064750612,
      "loss_iou": 0.234375,
      "loss_num": 0.017822265625,
      "loss_xval": 0.55859375,
      "num_input_tokens_seen": 273547644,
      "step": 4883
    },
    {
      "epoch": 10.877505567928731,
      "grad_norm": 16.91400909423828,
      "learning_rate": 1e-06,
      "loss": 0.4222,
      "num_input_tokens_seen": 273602376,
      "step": 4884
    },
    {
      "epoch": 10.877505567928731,
      "loss": 0.5984039306640625,
      "loss_ce": 0.00013728870544582605,
      "loss_iou": 0.224609375,
      "loss_num": 0.0299072265625,
      "loss_xval": 0.59765625,
      "num_input_tokens_seen": 273602376,
      "step": 4884
    },
    {
      "epoch": 10.879732739420936,
      "grad_norm": 15.582435607910156,
      "learning_rate": 1e-06,
      "loss": 0.736,
      "num_input_tokens_seen": 273658164,
      "step": 4885
    },
    {
      "epoch": 10.879732739420936,
      "loss": 0.728879451751709,
      "loss_ce": 0.00011967639875365421,
      "loss_iou": 0.296875,
      "loss_num": 0.02685546875,
      "loss_xval": 0.73046875,
      "num_input_tokens_seen": 273658164,
      "step": 4885
    },
    {
      "epoch": 10.88195991091314,
      "grad_norm": 21.012617111206055,
      "learning_rate": 1e-06,
      "loss": 0.5643,
      "num_input_tokens_seen": 273715456,
      "step": 4886
    },
    {
      "epoch": 10.88195991091314,
      "loss": 0.49945536255836487,
      "loss_ce": 0.00018778612138703465,
      "loss_iou": 0.208984375,
      "loss_num": 0.016357421875,
      "loss_xval": 0.5,
      "num_input_tokens_seen": 273715456,
      "step": 4886
    },
    {
      "epoch": 10.884187082405345,
      "grad_norm": 18.003925323486328,
      "learning_rate": 1e-06,
      "loss": 0.4371,
      "num_input_tokens_seen": 273766692,
      "step": 4887
    },
    {
      "epoch": 10.884187082405345,
      "loss": 0.45562559366226196,
      "loss_ce": 0.00033384724520146847,
      "loss_iou": 0.1669921875,
      "loss_num": 0.024169921875,
      "loss_xval": 0.455078125,
      "num_input_tokens_seen": 273766692,
      "step": 4887
    },
    {
      "epoch": 10.88641425389755,
      "grad_norm": 26.937620162963867,
      "learning_rate": 1e-06,
      "loss": 0.5724,
      "num_input_tokens_seen": 273822332,
      "step": 4888
    },
    {
      "epoch": 10.88641425389755,
      "loss": 0.5189422369003296,
      "loss_ce": 0.00014341062342282385,
      "loss_iou": 0.228515625,
      "loss_num": 0.01220703125,
      "loss_xval": 0.51953125,
      "num_input_tokens_seen": 273822332,
      "step": 4888
    },
    {
      "epoch": 10.888641425389755,
      "grad_norm": 25.740150451660156,
      "learning_rate": 1e-06,
      "loss": 0.5291,
      "num_input_tokens_seen": 273877776,
      "step": 4889
    },
    {
      "epoch": 10.888641425389755,
      "loss": 0.5222045183181763,
      "loss_ce": 0.00023185207101050764,
      "loss_iou": 0.2216796875,
      "loss_num": 0.015869140625,
      "loss_xval": 0.5234375,
      "num_input_tokens_seen": 273877776,
      "step": 4889
    },
    {
      "epoch": 10.89086859688196,
      "grad_norm": 13.944757461547852,
      "learning_rate": 1e-06,
      "loss": 0.5313,
      "num_input_tokens_seen": 273934432,
      "step": 4890
    },
    {
      "epoch": 10.89086859688196,
      "loss": 0.5552128553390503,
      "loss_ce": 0.00015907504712231457,
      "loss_iou": 0.2255859375,
      "loss_num": 0.020751953125,
      "loss_xval": 0.5546875,
      "num_input_tokens_seen": 273934432,
      "step": 4890
    },
    {
      "epoch": 10.893095768374165,
      "grad_norm": 32.77812194824219,
      "learning_rate": 1e-06,
      "loss": 0.5006,
      "num_input_tokens_seen": 273990888,
      "step": 4891
    },
    {
      "epoch": 10.893095768374165,
      "loss": 0.5784051418304443,
      "loss_ce": 0.0001580218377057463,
      "loss_iou": 0.2392578125,
      "loss_num": 0.02001953125,
      "loss_xval": 0.578125,
      "num_input_tokens_seen": 273990888,
      "step": 4891
    },
    {
      "epoch": 10.89532293986637,
      "grad_norm": 21.749361038208008,
      "learning_rate": 1e-06,
      "loss": 0.4988,
      "num_input_tokens_seen": 274045808,
      "step": 4892
    },
    {
      "epoch": 10.89532293986637,
      "loss": 0.47173595428466797,
      "loss_ce": 0.0017652796814218163,
      "loss_iou": 0.2060546875,
      "loss_num": 0.011474609375,
      "loss_xval": 0.470703125,
      "num_input_tokens_seen": 274045808,
      "step": 4892
    },
    {
      "epoch": 10.897550111358575,
      "grad_norm": 22.703105926513672,
      "learning_rate": 1e-06,
      "loss": 0.4889,
      "num_input_tokens_seen": 274101860,
      "step": 4893
    },
    {
      "epoch": 10.897550111358575,
      "loss": 0.5470762252807617,
      "loss_ce": 0.000323267828207463,
      "loss_iou": 0.2021484375,
      "loss_num": 0.0281982421875,
      "loss_xval": 0.546875,
      "num_input_tokens_seen": 274101860,
      "step": 4893
    },
    {
      "epoch": 10.89977728285078,
      "grad_norm": 19.098848342895508,
      "learning_rate": 1e-06,
      "loss": 0.5863,
      "num_input_tokens_seen": 274158588,
      "step": 4894
    },
    {
      "epoch": 10.89977728285078,
      "loss": 0.47424912452697754,
      "loss_ce": 0.0001280199212487787,
      "loss_iou": 0.20703125,
      "loss_num": 0.01190185546875,
      "loss_xval": 0.474609375,
      "num_input_tokens_seen": 274158588,
      "step": 4894
    },
    {
      "epoch": 10.902004454342984,
      "grad_norm": 25.0705509185791,
      "learning_rate": 1e-06,
      "loss": 0.5995,
      "num_input_tokens_seen": 274214796,
      "step": 4895
    },
    {
      "epoch": 10.902004454342984,
      "loss": 0.7026468515396118,
      "loss_ce": 0.00013221558765508235,
      "loss_iou": 0.294921875,
      "loss_num": 0.0223388671875,
      "loss_xval": 0.703125,
      "num_input_tokens_seen": 274214796,
      "step": 4895
    },
    {
      "epoch": 10.90423162583519,
      "grad_norm": 16.38619613647461,
      "learning_rate": 1e-06,
      "loss": 0.4676,
      "num_input_tokens_seen": 274270000,
      "step": 4896
    },
    {
      "epoch": 10.90423162583519,
      "loss": 0.5520434379577637,
      "loss_ce": 0.0002856444625649601,
      "loss_iou": 0.22265625,
      "loss_num": 0.0211181640625,
      "loss_xval": 0.55078125,
      "num_input_tokens_seen": 274270000,
      "step": 4896
    },
    {
      "epoch": 10.906458797327394,
      "grad_norm": 28.38006591796875,
      "learning_rate": 1e-06,
      "loss": 0.4696,
      "num_input_tokens_seen": 274324580,
      "step": 4897
    },
    {
      "epoch": 10.906458797327394,
      "loss": 0.513346791267395,
      "loss_ce": 0.00016320010763593018,
      "loss_iou": 0.2333984375,
      "loss_num": 0.009521484375,
      "loss_xval": 0.51171875,
      "num_input_tokens_seen": 274324580,
      "step": 4897
    },
    {
      "epoch": 10.908685968819599,
      "grad_norm": 34.72296142578125,
      "learning_rate": 1e-06,
      "loss": 0.7088,
      "num_input_tokens_seen": 274380456,
      "step": 4898
    },
    {
      "epoch": 10.908685968819599,
      "loss": 0.5927585363388062,
      "loss_ce": 0.0002292667340952903,
      "loss_iou": 0.251953125,
      "loss_num": 0.0174560546875,
      "loss_xval": 0.59375,
      "num_input_tokens_seen": 274380456,
      "step": 4898
    },
    {
      "epoch": 10.910913140311804,
      "grad_norm": 17.991031646728516,
      "learning_rate": 1e-06,
      "loss": 0.7028,
      "num_input_tokens_seen": 274436900,
      "step": 4899
    },
    {
      "epoch": 10.910913140311804,
      "loss": 1.07291579246521,
      "loss_ce": 0.0001619314862182364,
      "loss_iou": 0.451171875,
      "loss_num": 0.033935546875,
      "loss_xval": 1.0703125,
      "num_input_tokens_seen": 274436900,
      "step": 4899
    },
    {
      "epoch": 10.913140311804009,
      "grad_norm": 19.890304565429688,
      "learning_rate": 1e-06,
      "loss": 0.6784,
      "num_input_tokens_seen": 274492560,
      "step": 4900
    },
    {
      "epoch": 10.913140311804009,
      "loss": 0.7131062746047974,
      "loss_ce": 0.0007039305055513978,
      "loss_iou": 0.28515625,
      "loss_num": 0.028564453125,
      "loss_xval": 0.7109375,
      "num_input_tokens_seen": 274492560,
      "step": 4900
    },
    {
      "epoch": 10.915367483296214,
      "grad_norm": 19.241336822509766,
      "learning_rate": 1e-06,
      "loss": 0.4938,
      "num_input_tokens_seen": 274549672,
      "step": 4901
    },
    {
      "epoch": 10.915367483296214,
      "loss": 0.4737103581428528,
      "loss_ce": 0.00013855646830052137,
      "loss_iou": 0.2138671875,
      "loss_num": 0.00933837890625,
      "loss_xval": 0.47265625,
      "num_input_tokens_seen": 274549672,
      "step": 4901
    },
    {
      "epoch": 10.917594654788418,
      "grad_norm": 20.94312858581543,
      "learning_rate": 1e-06,
      "loss": 0.5268,
      "num_input_tokens_seen": 274606064,
      "step": 4902
    },
    {
      "epoch": 10.917594654788418,
      "loss": 0.6111065149307251,
      "loss_ce": 0.00014459769590757787,
      "loss_iou": 0.2451171875,
      "loss_num": 0.0244140625,
      "loss_xval": 0.609375,
      "num_input_tokens_seen": 274606064,
      "step": 4902
    },
    {
      "epoch": 10.919821826280623,
      "grad_norm": 55.303871154785156,
      "learning_rate": 1e-06,
      "loss": 0.4566,
      "num_input_tokens_seen": 274662744,
      "step": 4903
    },
    {
      "epoch": 10.919821826280623,
      "loss": 0.4556983709335327,
      "loss_ce": 0.0001320057053817436,
      "loss_iou": 0.2060546875,
      "loss_num": 0.0086669921875,
      "loss_xval": 0.455078125,
      "num_input_tokens_seen": 274662744,
      "step": 4903
    },
    {
      "epoch": 10.922048997772828,
      "grad_norm": 13.991397857666016,
      "learning_rate": 1e-06,
      "loss": 0.3496,
      "num_input_tokens_seen": 274719288,
      "step": 4904
    },
    {
      "epoch": 10.922048997772828,
      "loss": 0.35973674058914185,
      "loss_ce": 0.00011758786422433332,
      "loss_iou": 0.1513671875,
      "loss_num": 0.0113525390625,
      "loss_xval": 0.359375,
      "num_input_tokens_seen": 274719288,
      "step": 4904
    },
    {
      "epoch": 10.924276169265033,
      "grad_norm": 18.747949600219727,
      "learning_rate": 1e-06,
      "loss": 0.4155,
      "num_input_tokens_seen": 274774944,
      "step": 4905
    },
    {
      "epoch": 10.924276169265033,
      "loss": 0.44049468636512756,
      "loss_ce": 0.0002480950206518173,
      "loss_iou": 0.185546875,
      "loss_num": 0.013671875,
      "loss_xval": 0.439453125,
      "num_input_tokens_seen": 274774944,
      "step": 4905
    },
    {
      "epoch": 10.926503340757238,
      "grad_norm": 21.45561408996582,
      "learning_rate": 1e-06,
      "loss": 0.4906,
      "num_input_tokens_seen": 274831664,
      "step": 4906
    },
    {
      "epoch": 10.926503340757238,
      "loss": 0.5997252464294434,
      "loss_ce": 0.00011587167682591826,
      "loss_iou": 0.27734375,
      "loss_num": 0.0086669921875,
      "loss_xval": 0.6015625,
      "num_input_tokens_seen": 274831664,
      "step": 4906
    },
    {
      "epoch": 10.928730512249443,
      "grad_norm": 21.097562789916992,
      "learning_rate": 1e-06,
      "loss": 0.4488,
      "num_input_tokens_seen": 274888812,
      "step": 4907
    },
    {
      "epoch": 10.928730512249443,
      "loss": 0.4747345447540283,
      "loss_ce": 0.00012518178846221417,
      "loss_iou": 0.220703125,
      "loss_num": 0.00640869140625,
      "loss_xval": 0.474609375,
      "num_input_tokens_seen": 274888812,
      "step": 4907
    },
    {
      "epoch": 10.930957683741648,
      "grad_norm": 20.24143409729004,
      "learning_rate": 1e-06,
      "loss": 0.4771,
      "num_input_tokens_seen": 274945504,
      "step": 4908
    },
    {
      "epoch": 10.930957683741648,
      "loss": 0.39001041650772095,
      "loss_ce": 0.00011784063826780766,
      "loss_iou": 0.177734375,
      "loss_num": 0.00677490234375,
      "loss_xval": 0.390625,
      "num_input_tokens_seen": 274945504,
      "step": 4908
    },
    {
      "epoch": 10.933184855233852,
      "grad_norm": 20.678117752075195,
      "learning_rate": 1e-06,
      "loss": 0.4148,
      "num_input_tokens_seen": 275004552,
      "step": 4909
    },
    {
      "epoch": 10.933184855233852,
      "loss": 0.3625844419002533,
      "loss_ce": 0.0001576666400069371,
      "loss_iou": 0.1611328125,
      "loss_num": 0.00787353515625,
      "loss_xval": 0.36328125,
      "num_input_tokens_seen": 275004552,
      "step": 4909
    },
    {
      "epoch": 10.935412026726057,
      "grad_norm": 20.854366302490234,
      "learning_rate": 1e-06,
      "loss": 0.5889,
      "num_input_tokens_seen": 275062356,
      "step": 4910
    },
    {
      "epoch": 10.935412026726057,
      "loss": 0.4733467102050781,
      "loss_ce": 0.0002021462714765221,
      "loss_iou": 0.2158203125,
      "loss_num": 0.00811767578125,
      "loss_xval": 0.47265625,
      "num_input_tokens_seen": 275062356,
      "step": 4910
    },
    {
      "epoch": 10.937639198218262,
      "grad_norm": 23.64391326904297,
      "learning_rate": 1e-06,
      "loss": 0.7998,
      "num_input_tokens_seen": 275117932,
      "step": 4911
    },
    {
      "epoch": 10.937639198218262,
      "loss": 0.7572309970855713,
      "loss_ce": 0.0001508938439656049,
      "loss_iou": 0.322265625,
      "loss_num": 0.0220947265625,
      "loss_xval": 0.7578125,
      "num_input_tokens_seen": 275117932,
      "step": 4911
    },
    {
      "epoch": 10.939866369710467,
      "grad_norm": 22.301734924316406,
      "learning_rate": 1e-06,
      "loss": 0.4655,
      "num_input_tokens_seen": 275171980,
      "step": 4912
    },
    {
      "epoch": 10.939866369710467,
      "loss": 0.4408171772956848,
      "loss_ce": 0.00014336456661112607,
      "loss_iou": 0.208984375,
      "loss_num": 0.0045166015625,
      "loss_xval": 0.44140625,
      "num_input_tokens_seen": 275171980,
      "step": 4912
    },
    {
      "epoch": 10.942093541202672,
      "grad_norm": 18.19380760192871,
      "learning_rate": 1e-06,
      "loss": 0.6564,
      "num_input_tokens_seen": 275227324,
      "step": 4913
    },
    {
      "epoch": 10.942093541202672,
      "loss": 0.8009915351867676,
      "loss_ce": 0.00021027974435128272,
      "loss_iou": 0.3515625,
      "loss_num": 0.0194091796875,
      "loss_xval": 0.80078125,
      "num_input_tokens_seen": 275227324,
      "step": 4913
    },
    {
      "epoch": 10.944320712694877,
      "grad_norm": 14.776144981384277,
      "learning_rate": 1e-06,
      "loss": 0.5706,
      "num_input_tokens_seen": 275282508,
      "step": 4914
    },
    {
      "epoch": 10.944320712694877,
      "loss": 0.657973051071167,
      "loss_ce": 0.00013618965749628842,
      "loss_iou": 0.2734375,
      "loss_num": 0.022216796875,
      "loss_xval": 0.65625,
      "num_input_tokens_seen": 275282508,
      "step": 4914
    },
    {
      "epoch": 10.946547884187082,
      "grad_norm": 23.789960861206055,
      "learning_rate": 1e-06,
      "loss": 0.4952,
      "num_input_tokens_seen": 275338856,
      "step": 4915
    },
    {
      "epoch": 10.946547884187082,
      "loss": 0.48818719387054443,
      "loss_ce": 0.00015006719331722707,
      "loss_iou": 0.2158203125,
      "loss_num": 0.01141357421875,
      "loss_xval": 0.48828125,
      "num_input_tokens_seen": 275338856,
      "step": 4915
    },
    {
      "epoch": 10.948775055679288,
      "grad_norm": 19.028474807739258,
      "learning_rate": 1e-06,
      "loss": 0.5003,
      "num_input_tokens_seen": 275395740,
      "step": 4916
    },
    {
      "epoch": 10.948775055679288,
      "loss": 0.5374853610992432,
      "loss_ce": 0.0001318200957030058,
      "loss_iou": 0.2294921875,
      "loss_num": 0.01556396484375,
      "loss_xval": 0.5390625,
      "num_input_tokens_seen": 275395740,
      "step": 4916
    },
    {
      "epoch": 10.951002227171493,
      "grad_norm": 20.012557983398438,
      "learning_rate": 1e-06,
      "loss": 0.4823,
      "num_input_tokens_seen": 275454060,
      "step": 4917
    },
    {
      "epoch": 10.951002227171493,
      "loss": 0.466092973947525,
      "loss_ce": 0.0001506020489614457,
      "loss_iou": 0.2021484375,
      "loss_num": 0.01226806640625,
      "loss_xval": 0.466796875,
      "num_input_tokens_seen": 275454060,
      "step": 4917
    },
    {
      "epoch": 10.953229398663698,
      "grad_norm": 16.04680633544922,
      "learning_rate": 1e-06,
      "loss": 0.4923,
      "num_input_tokens_seen": 275511984,
      "step": 4918
    },
    {
      "epoch": 10.953229398663698,
      "loss": 0.4693390130996704,
      "loss_ce": 0.0001617702655494213,
      "loss_iou": 0.2138671875,
      "loss_num": 0.00836181640625,
      "loss_xval": 0.46875,
      "num_input_tokens_seen": 275511984,
      "step": 4918
    },
    {
      "epoch": 10.955456570155903,
      "grad_norm": 19.825653076171875,
      "learning_rate": 1e-06,
      "loss": 0.5232,
      "num_input_tokens_seen": 275568436,
      "step": 4919
    },
    {
      "epoch": 10.955456570155903,
      "loss": 0.561336100101471,
      "loss_ce": 0.00011784351227106526,
      "loss_iou": 0.2392578125,
      "loss_num": 0.0166015625,
      "loss_xval": 0.5625,
      "num_input_tokens_seen": 275568436,
      "step": 4919
    },
    {
      "epoch": 10.957683741648108,
      "grad_norm": 22.18250846862793,
      "learning_rate": 1e-06,
      "loss": 0.4804,
      "num_input_tokens_seen": 275625196,
      "step": 4920
    },
    {
      "epoch": 10.957683741648108,
      "loss": 0.43557554483413696,
      "loss_ce": 0.00015077181160449982,
      "loss_iou": 0.18359375,
      "loss_num": 0.0135498046875,
      "loss_xval": 0.435546875,
      "num_input_tokens_seen": 275625196,
      "step": 4920
    },
    {
      "epoch": 10.959910913140313,
      "grad_norm": 23.71367835998535,
      "learning_rate": 1e-06,
      "loss": 0.5365,
      "num_input_tokens_seen": 275681500,
      "step": 4921
    },
    {
      "epoch": 10.959910913140313,
      "loss": 0.5750996470451355,
      "loss_ce": 0.0002705350052565336,
      "loss_iou": 0.2353515625,
      "loss_num": 0.02099609375,
      "loss_xval": 0.57421875,
      "num_input_tokens_seen": 275681500,
      "step": 4921
    },
    {
      "epoch": 10.962138084632517,
      "grad_norm": 36.595542907714844,
      "learning_rate": 1e-06,
      "loss": 0.4071,
      "num_input_tokens_seen": 275736712,
      "step": 4922
    },
    {
      "epoch": 10.962138084632517,
      "loss": 0.35608160495758057,
      "loss_ce": 0.0001245439052581787,
      "loss_iou": 0.162109375,
      "loss_num": 0.0064697265625,
      "loss_xval": 0.35546875,
      "num_input_tokens_seen": 275736712,
      "step": 4922
    },
    {
      "epoch": 10.964365256124722,
      "grad_norm": 22.66077995300293,
      "learning_rate": 1e-06,
      "loss": 0.6015,
      "num_input_tokens_seen": 275792276,
      "step": 4923
    },
    {
      "epoch": 10.964365256124722,
      "loss": 0.39454376697540283,
      "loss_ce": 0.00013458277680911124,
      "loss_iou": 0.1748046875,
      "loss_num": 0.0089111328125,
      "loss_xval": 0.39453125,
      "num_input_tokens_seen": 275792276,
      "step": 4923
    },
    {
      "epoch": 10.966592427616927,
      "grad_norm": 18.849925994873047,
      "learning_rate": 1e-06,
      "loss": 0.4902,
      "num_input_tokens_seen": 275847012,
      "step": 4924
    },
    {
      "epoch": 10.966592427616927,
      "loss": 0.5802684426307678,
      "loss_ce": 0.0015330992173403502,
      "loss_iou": 0.2412109375,
      "loss_num": 0.019287109375,
      "loss_xval": 0.578125,
      "num_input_tokens_seen": 275847012,
      "step": 4924
    },
    {
      "epoch": 10.968819599109132,
      "grad_norm": 25.976478576660156,
      "learning_rate": 1e-06,
      "loss": 0.6866,
      "num_input_tokens_seen": 275905656,
      "step": 4925
    },
    {
      "epoch": 10.968819599109132,
      "loss": 0.5414575338363647,
      "loss_ce": 0.0001978027867153287,
      "loss_iou": 0.2236328125,
      "loss_num": 0.018798828125,
      "loss_xval": 0.54296875,
      "num_input_tokens_seen": 275905656,
      "step": 4925
    },
    {
      "epoch": 10.971046770601337,
      "grad_norm": 22.996116638183594,
      "learning_rate": 1e-06,
      "loss": 0.578,
      "num_input_tokens_seen": 275961284,
      "step": 4926
    },
    {
      "epoch": 10.971046770601337,
      "loss": 0.4320530593395233,
      "loss_ce": 0.00016829956439323723,
      "loss_iou": 0.1826171875,
      "loss_num": 0.01361083984375,
      "loss_xval": 0.431640625,
      "num_input_tokens_seen": 275961284,
      "step": 4926
    },
    {
      "epoch": 10.973273942093542,
      "grad_norm": 15.009370803833008,
      "learning_rate": 1e-06,
      "loss": 0.596,
      "num_input_tokens_seen": 276017624,
      "step": 4927
    },
    {
      "epoch": 10.973273942093542,
      "loss": 0.6515401601791382,
      "loss_ce": 0.0001729477517073974,
      "loss_iou": 0.2734375,
      "loss_num": 0.0208740234375,
      "loss_xval": 0.65234375,
      "num_input_tokens_seen": 276017624,
      "step": 4927
    },
    {
      "epoch": 10.975501113585747,
      "grad_norm": 17.324737548828125,
      "learning_rate": 1e-06,
      "loss": 0.4243,
      "num_input_tokens_seen": 276075216,
      "step": 4928
    },
    {
      "epoch": 10.975501113585747,
      "loss": 0.45728790760040283,
      "loss_ce": 0.00013462905189953744,
      "loss_iou": 0.193359375,
      "loss_num": 0.01397705078125,
      "loss_xval": 0.45703125,
      "num_input_tokens_seen": 276075216,
      "step": 4928
    },
    {
      "epoch": 10.977728285077951,
      "grad_norm": 77.1220932006836,
      "learning_rate": 1e-06,
      "loss": 0.4622,
      "num_input_tokens_seen": 276132700,
      "step": 4929
    },
    {
      "epoch": 10.977728285077951,
      "loss": 0.4627261161804199,
      "loss_ce": 0.000201685048523359,
      "loss_iou": 0.2021484375,
      "loss_num": 0.0115966796875,
      "loss_xval": 0.462890625,
      "num_input_tokens_seen": 276132700,
      "step": 4929
    },
    {
      "epoch": 10.979955456570156,
      "grad_norm": 21.860414505004883,
      "learning_rate": 1e-06,
      "loss": 0.4459,
      "num_input_tokens_seen": 276192092,
      "step": 4930
    },
    {
      "epoch": 10.979955456570156,
      "loss": 0.5067447423934937,
      "loss_ce": 0.00015294540207833052,
      "loss_iou": 0.2236328125,
      "loss_num": 0.01177978515625,
      "loss_xval": 0.5078125,
      "num_input_tokens_seen": 276192092,
      "step": 4930
    },
    {
      "epoch": 10.982182628062361,
      "grad_norm": 23.157732009887695,
      "learning_rate": 1e-06,
      "loss": 0.5643,
      "num_input_tokens_seen": 276248164,
      "step": 4931
    },
    {
      "epoch": 10.982182628062361,
      "loss": 0.5332338809967041,
      "loss_ce": 0.00013755704276263714,
      "loss_iou": 0.2392578125,
      "loss_num": 0.0111083984375,
      "loss_xval": 0.53125,
      "num_input_tokens_seen": 276248164,
      "step": 4931
    },
    {
      "epoch": 10.984409799554566,
      "grad_norm": 18.797258377075195,
      "learning_rate": 1e-06,
      "loss": 0.5975,
      "num_input_tokens_seen": 276304632,
      "step": 4932
    },
    {
      "epoch": 10.984409799554566,
      "loss": 0.6781335473060608,
      "loss_ce": 0.00027711730217561126,
      "loss_iou": 0.265625,
      "loss_num": 0.0296630859375,
      "loss_xval": 0.6796875,
      "num_input_tokens_seen": 276304632,
      "step": 4932
    },
    {
      "epoch": 10.98663697104677,
      "grad_norm": 23.31140899658203,
      "learning_rate": 1e-06,
      "loss": 0.4345,
      "num_input_tokens_seen": 276361908,
      "step": 4933
    },
    {
      "epoch": 10.98663697104677,
      "loss": 0.5331314206123352,
      "loss_ce": 0.00011140106653328985,
      "loss_iou": 0.2392578125,
      "loss_num": 0.010986328125,
      "loss_xval": 0.53125,
      "num_input_tokens_seen": 276361908,
      "step": 4933
    },
    {
      "epoch": 10.988864142538976,
      "grad_norm": 15.896737098693848,
      "learning_rate": 1e-06,
      "loss": 0.6221,
      "num_input_tokens_seen": 276420060,
      "step": 4934
    },
    {
      "epoch": 10.988864142538976,
      "loss": 0.7978502511978149,
      "loss_ce": 0.00012079518637619913,
      "loss_iou": 0.3515625,
      "loss_num": 0.0186767578125,
      "loss_xval": 0.796875,
      "num_input_tokens_seen": 276420060,
      "step": 4934
    },
    {
      "epoch": 10.99109131403118,
      "grad_norm": 17.5236759185791,
      "learning_rate": 1e-06,
      "loss": 0.6223,
      "num_input_tokens_seen": 276475372,
      "step": 4935
    },
    {
      "epoch": 10.99109131403118,
      "loss": 0.7966042160987854,
      "loss_ce": 0.00021749922598246485,
      "loss_iou": 0.30859375,
      "loss_num": 0.035888671875,
      "loss_xval": 0.796875,
      "num_input_tokens_seen": 276475372,
      "step": 4935
    },
    {
      "epoch": 10.993318485523385,
      "grad_norm": 16.806560516357422,
      "learning_rate": 1e-06,
      "loss": 0.5073,
      "num_input_tokens_seen": 276529704,
      "step": 4936
    },
    {
      "epoch": 10.993318485523385,
      "loss": 0.3875495195388794,
      "loss_ce": 0.00015937026182655245,
      "loss_iou": 0.1669921875,
      "loss_num": 0.0107421875,
      "loss_xval": 0.38671875,
      "num_input_tokens_seen": 276529704,
      "step": 4936
    },
    {
      "epoch": 10.99554565701559,
      "grad_norm": 15.568613052368164,
      "learning_rate": 1e-06,
      "loss": 0.4306,
      "num_input_tokens_seen": 276584256,
      "step": 4937
    },
    {
      "epoch": 10.99554565701559,
      "loss": 0.3633689284324646,
      "loss_ce": 0.0002707808162085712,
      "loss_iou": 0.1279296875,
      "loss_num": 0.021728515625,
      "loss_xval": 0.36328125,
      "num_input_tokens_seen": 276584256,
      "step": 4937
    },
    {
      "epoch": 10.997772828507795,
      "grad_norm": 25.16409683227539,
      "learning_rate": 1e-06,
      "loss": 0.3894,
      "num_input_tokens_seen": 276640164,
      "step": 4938
    },
    {
      "epoch": 10.997772828507795,
      "loss": 0.41700875759124756,
      "loss_ce": 0.00013862067135050893,
      "loss_iou": 0.1865234375,
      "loss_num": 0.00860595703125,
      "loss_xval": 0.416015625,
      "num_input_tokens_seen": 276640164,
      "step": 4938
    },
    {
      "epoch": 11.0,
      "grad_norm": 23.693870544433594,
      "learning_rate": 1e-06,
      "loss": 0.7586,
      "num_input_tokens_seen": 276695568,
      "step": 4939
    },
    {
      "epoch": 11.0,
      "loss": 0.8163493275642395,
      "loss_ce": 0.00018723538960330188,
      "loss_iou": 0.35546875,
      "loss_num": 0.020751953125,
      "loss_xval": 0.81640625,
      "num_input_tokens_seen": 276695568,
      "step": 4939
    },
    {
      "epoch": 11.002227171492205,
      "grad_norm": 14.375535011291504,
      "learning_rate": 1e-06,
      "loss": 0.408,
      "num_input_tokens_seen": 276751748,
      "step": 4940
    },
    {
      "epoch": 11.002227171492205,
      "loss": 0.34511420130729675,
      "loss_ce": 0.00020451581804081798,
      "loss_iou": 0.158203125,
      "loss_num": 0.005462646484375,
      "loss_xval": 0.345703125,
      "num_input_tokens_seen": 276751748,
      "step": 4940
    },
    {
      "epoch": 11.00445434298441,
      "grad_norm": 17.11101722717285,
      "learning_rate": 1e-06,
      "loss": 0.4981,
      "num_input_tokens_seen": 276808900,
      "step": 4941
    },
    {
      "epoch": 11.00445434298441,
      "loss": 0.5812661647796631,
      "loss_ce": 0.00021148948871996254,
      "loss_iou": 0.259765625,
      "loss_num": 0.0120849609375,
      "loss_xval": 0.58203125,
      "num_input_tokens_seen": 276808900,
      "step": 4941
    },
    {
      "epoch": 11.006681514476615,
      "grad_norm": 24.729581832885742,
      "learning_rate": 1e-06,
      "loss": 0.591,
      "num_input_tokens_seen": 276865236,
      "step": 4942
    },
    {
      "epoch": 11.006681514476615,
      "loss": 0.5612735748291016,
      "loss_ce": 0.00048253341810777783,
      "loss_iou": 0.248046875,
      "loss_num": 0.0130615234375,
      "loss_xval": 0.5625,
      "num_input_tokens_seen": 276865236,
      "step": 4942
    },
    {
      "epoch": 11.00890868596882,
      "grad_norm": 18.79891014099121,
      "learning_rate": 1e-06,
      "loss": 0.3387,
      "num_input_tokens_seen": 276923576,
      "step": 4943
    },
    {
      "epoch": 11.00890868596882,
      "loss": 0.26598143577575684,
      "loss_ce": 0.00011229849769733846,
      "loss_iou": 0.12109375,
      "loss_num": 0.004608154296875,
      "loss_xval": 0.265625,
      "num_input_tokens_seen": 276923576,
      "step": 4943
    },
    {
      "epoch": 11.011135857461024,
      "grad_norm": 19.149585723876953,
      "learning_rate": 1e-06,
      "loss": 0.3532,
      "num_input_tokens_seen": 276980108,
      "step": 4944
    },
    {
      "epoch": 11.011135857461024,
      "loss": 0.377750962972641,
      "loss_ce": 0.00012643022637348622,
      "loss_iou": 0.17578125,
      "loss_num": 0.0052490234375,
      "loss_xval": 0.376953125,
      "num_input_tokens_seen": 276980108,
      "step": 4944
    },
    {
      "epoch": 11.01336302895323,
      "grad_norm": 19.660892486572266,
      "learning_rate": 1e-06,
      "loss": 0.4988,
      "num_input_tokens_seen": 277035084,
      "step": 4945
    },
    {
      "epoch": 11.01336302895323,
      "loss": 0.49671003222465515,
      "loss_ce": 0.00012798060197383165,
      "loss_iou": 0.21875,
      "loss_num": 0.01171875,
      "loss_xval": 0.49609375,
      "num_input_tokens_seen": 277035084,
      "step": 4945
    },
    {
      "epoch": 11.015590200445434,
      "grad_norm": 17.99364471435547,
      "learning_rate": 1e-06,
      "loss": 0.5213,
      "num_input_tokens_seen": 277090600,
      "step": 4946
    },
    {
      "epoch": 11.015590200445434,
      "loss": 0.5547381043434143,
      "loss_ce": 0.0002947351022157818,
      "loss_iou": 0.2314453125,
      "loss_num": 0.018310546875,
      "loss_xval": 0.5546875,
      "num_input_tokens_seen": 277090600,
      "step": 4946
    },
    {
      "epoch": 11.017817371937639,
      "grad_norm": 26.641109466552734,
      "learning_rate": 1e-06,
      "loss": 0.4299,
      "num_input_tokens_seen": 277144644,
      "step": 4947
    },
    {
      "epoch": 11.017817371937639,
      "loss": 0.3615342974662781,
      "loss_ce": 0.0001451151620130986,
      "loss_iou": 0.1357421875,
      "loss_num": 0.0181884765625,
      "loss_xval": 0.361328125,
      "num_input_tokens_seen": 277144644,
      "step": 4947
    },
    {
      "epoch": 11.020044543429844,
      "grad_norm": 25.904264450073242,
      "learning_rate": 1e-06,
      "loss": 0.8089,
      "num_input_tokens_seen": 277198968,
      "step": 4948
    },
    {
      "epoch": 11.020044543429844,
      "loss": 0.8873788118362427,
      "loss_ce": 0.0001717885461403057,
      "loss_iou": 0.373046875,
      "loss_num": 0.028076171875,
      "loss_xval": 0.88671875,
      "num_input_tokens_seen": 277198968,
      "step": 4948
    },
    {
      "epoch": 11.022271714922049,
      "grad_norm": 18.756681442260742,
      "learning_rate": 1e-06,
      "loss": 0.6074,
      "num_input_tokens_seen": 277252344,
      "step": 4949
    },
    {
      "epoch": 11.022271714922049,
      "loss": 0.7791860103607178,
      "loss_ce": 0.00037738942774012685,
      "loss_iou": 0.302734375,
      "loss_num": 0.0341796875,
      "loss_xval": 0.77734375,
      "num_input_tokens_seen": 277252344,
      "step": 4949
    },
    {
      "epoch": 11.024498886414253,
      "grad_norm": 17.9628849029541,
      "learning_rate": 1e-06,
      "loss": 0.3852,
      "num_input_tokens_seen": 277308488,
      "step": 4950
    },
    {
      "epoch": 11.024498886414253,
      "loss": 0.34923800826072693,
      "loss_ce": 0.00011689884559018537,
      "loss_iou": 0.1513671875,
      "loss_num": 0.00927734375,
      "loss_xval": 0.349609375,
      "num_input_tokens_seen": 277308488,
      "step": 4950
    },
    {
      "epoch": 11.026726057906458,
      "grad_norm": 14.993110656738281,
      "learning_rate": 1e-06,
      "loss": 0.3982,
      "num_input_tokens_seen": 277364872,
      "step": 4951
    },
    {
      "epoch": 11.026726057906458,
      "loss": 0.3778972625732422,
      "loss_ce": 0.00015067942149471492,
      "loss_iou": 0.16796875,
      "loss_num": 0.00823974609375,
      "loss_xval": 0.376953125,
      "num_input_tokens_seen": 277364872,
      "step": 4951
    },
    {
      "epoch": 11.028953229398663,
      "grad_norm": 18.500215530395508,
      "learning_rate": 1e-06,
      "loss": 0.4492,
      "num_input_tokens_seen": 277422360,
      "step": 4952
    },
    {
      "epoch": 11.028953229398663,
      "loss": 0.517828643321991,
      "loss_ce": 0.00012847439211327583,
      "loss_iou": 0.2119140625,
      "loss_num": 0.0185546875,
      "loss_xval": 0.51953125,
      "num_input_tokens_seen": 277422360,
      "step": 4952
    },
    {
      "epoch": 11.031180400890868,
      "grad_norm": 28.1491641998291,
      "learning_rate": 1e-06,
      "loss": 0.6864,
      "num_input_tokens_seen": 277479676,
      "step": 4953
    },
    {
      "epoch": 11.031180400890868,
      "loss": 0.7513668537139893,
      "loss_ce": 0.00014616544649470598,
      "loss_iou": 0.2890625,
      "loss_num": 0.03466796875,
      "loss_xval": 0.75,
      "num_input_tokens_seen": 277479676,
      "step": 4953
    },
    {
      "epoch": 11.033407572383073,
      "grad_norm": 14.530891418457031,
      "learning_rate": 1e-06,
      "loss": 0.5625,
      "num_input_tokens_seen": 277535648,
      "step": 4954
    },
    {
      "epoch": 11.033407572383073,
      "loss": 0.6487134695053101,
      "loss_ce": 0.00015391816850751638,
      "loss_iou": 0.265625,
      "loss_num": 0.023681640625,
      "loss_xval": 0.6484375,
      "num_input_tokens_seen": 277535648,
      "step": 4954
    },
    {
      "epoch": 11.035634743875278,
      "grad_norm": 18.762638092041016,
      "learning_rate": 1e-06,
      "loss": 0.4931,
      "num_input_tokens_seen": 277593048,
      "step": 4955
    },
    {
      "epoch": 11.035634743875278,
      "loss": 0.5269228219985962,
      "loss_ce": 0.00012840772978961468,
      "loss_iou": 0.21875,
      "loss_num": 0.0177001953125,
      "loss_xval": 0.52734375,
      "num_input_tokens_seen": 277593048,
      "step": 4955
    },
    {
      "epoch": 11.037861915367483,
      "grad_norm": 14.296708106994629,
      "learning_rate": 1e-06,
      "loss": 0.6159,
      "num_input_tokens_seen": 277647512,
      "step": 4956
    },
    {
      "epoch": 11.037861915367483,
      "loss": 0.7726843357086182,
      "loss_ce": 0.00022338703274726868,
      "loss_iou": 0.296875,
      "loss_num": 0.03564453125,
      "loss_xval": 0.7734375,
      "num_input_tokens_seen": 277647512,
      "step": 4956
    },
    {
      "epoch": 11.040089086859687,
      "grad_norm": 16.444766998291016,
      "learning_rate": 1e-06,
      "loss": 0.476,
      "num_input_tokens_seen": 277705924,
      "step": 4957
    },
    {
      "epoch": 11.040089086859687,
      "loss": 0.48854783177375793,
      "loss_ce": 0.0001445203961338848,
      "loss_iou": 0.2119140625,
      "loss_num": 0.0126953125,
      "loss_xval": 0.48828125,
      "num_input_tokens_seen": 277705924,
      "step": 4957
    },
    {
      "epoch": 11.042316258351892,
      "grad_norm": 33.47507095336914,
      "learning_rate": 1e-06,
      "loss": 0.6862,
      "num_input_tokens_seen": 277763404,
      "step": 4958
    },
    {
      "epoch": 11.042316258351892,
      "loss": 0.7765988111495972,
      "loss_ce": 0.00010955688776448369,
      "loss_iou": 0.34765625,
      "loss_num": 0.0166015625,
      "loss_xval": 0.77734375,
      "num_input_tokens_seen": 277763404,
      "step": 4958
    },
    {
      "epoch": 11.044543429844097,
      "grad_norm": 23.200275421142578,
      "learning_rate": 1e-06,
      "loss": 0.5959,
      "num_input_tokens_seen": 277819472,
      "step": 4959
    },
    {
      "epoch": 11.044543429844097,
      "loss": 0.7127882838249207,
      "loss_ce": 0.00014180471771396697,
      "loss_iou": 0.32421875,
      "loss_num": 0.01287841796875,
      "loss_xval": 0.7109375,
      "num_input_tokens_seen": 277819472,
      "step": 4959
    },
    {
      "epoch": 11.046770601336302,
      "grad_norm": 19.549856185913086,
      "learning_rate": 1e-06,
      "loss": 0.6735,
      "num_input_tokens_seen": 277874920,
      "step": 4960
    },
    {
      "epoch": 11.046770601336302,
      "loss": 0.7458186149597168,
      "loss_ce": 0.00021311640739440918,
      "loss_iou": 0.294921875,
      "loss_num": 0.03125,
      "loss_xval": 0.74609375,
      "num_input_tokens_seen": 277874920,
      "step": 4960
    },
    {
      "epoch": 11.048997772828507,
      "grad_norm": 18.301462173461914,
      "learning_rate": 1e-06,
      "loss": 0.4467,
      "num_input_tokens_seen": 277932412,
      "step": 4961
    },
    {
      "epoch": 11.048997772828507,
      "loss": 0.38100382685661316,
      "loss_ce": 0.0001444508379790932,
      "loss_iou": 0.16796875,
      "loss_num": 0.0091552734375,
      "loss_xval": 0.380859375,
      "num_input_tokens_seen": 277932412,
      "step": 4961
    },
    {
      "epoch": 11.051224944320714,
      "grad_norm": 22.31706428527832,
      "learning_rate": 1e-06,
      "loss": 0.6524,
      "num_input_tokens_seen": 277990888,
      "step": 4962
    },
    {
      "epoch": 11.051224944320714,
      "loss": 0.6529885530471802,
      "loss_ce": 0.00015653553418815136,
      "loss_iou": 0.265625,
      "loss_num": 0.0242919921875,
      "loss_xval": 0.65234375,
      "num_input_tokens_seen": 277990888,
      "step": 4962
    },
    {
      "epoch": 11.053452115812918,
      "grad_norm": 48.24069595336914,
      "learning_rate": 1e-06,
      "loss": 0.5579,
      "num_input_tokens_seen": 278045352,
      "step": 4963
    },
    {
      "epoch": 11.053452115812918,
      "loss": 0.4462721645832062,
      "loss_ce": 0.00013570513692684472,
      "loss_iou": 0.171875,
      "loss_num": 0.0206298828125,
      "loss_xval": 0.4453125,
      "num_input_tokens_seen": 278045352,
      "step": 4963
    },
    {
      "epoch": 11.055679287305123,
      "grad_norm": 14.883129119873047,
      "learning_rate": 1e-06,
      "loss": 0.4838,
      "num_input_tokens_seen": 278100884,
      "step": 4964
    },
    {
      "epoch": 11.055679287305123,
      "loss": 0.5324569344520569,
      "loss_ce": 0.00016932294238358736,
      "loss_iou": 0.220703125,
      "loss_num": 0.01806640625,
      "loss_xval": 0.53125,
      "num_input_tokens_seen": 278100884,
      "step": 4964
    },
    {
      "epoch": 11.057906458797328,
      "grad_norm": 92.90229797363281,
      "learning_rate": 1e-06,
      "loss": 0.5606,
      "num_input_tokens_seen": 278155508,
      "step": 4965
    },
    {
      "epoch": 11.057906458797328,
      "loss": 0.6233317255973816,
      "loss_ce": 0.00016281349235214293,
      "loss_iou": 0.287109375,
      "loss_num": 0.00994873046875,
      "loss_xval": 0.625,
      "num_input_tokens_seen": 278155508,
      "step": 4965
    },
    {
      "epoch": 11.060133630289533,
      "grad_norm": 22.694913864135742,
      "learning_rate": 1e-06,
      "loss": 0.5463,
      "num_input_tokens_seen": 278210972,
      "step": 4966
    },
    {
      "epoch": 11.060133630289533,
      "loss": 0.5516934394836426,
      "loss_ce": 0.00017979381664190441,
      "loss_iou": 0.2294921875,
      "loss_num": 0.018310546875,
      "loss_xval": 0.55078125,
      "num_input_tokens_seen": 278210972,
      "step": 4966
    },
    {
      "epoch": 11.062360801781738,
      "grad_norm": 14.845329284667969,
      "learning_rate": 1e-06,
      "loss": 0.3656,
      "num_input_tokens_seen": 278266736,
      "step": 4967
    },
    {
      "epoch": 11.062360801781738,
      "loss": 0.3910008668899536,
      "loss_ce": 0.00019277536193840206,
      "loss_iou": 0.1767578125,
      "loss_num": 0.0074462890625,
      "loss_xval": 0.390625,
      "num_input_tokens_seen": 278266736,
      "step": 4967
    },
    {
      "epoch": 11.064587973273943,
      "grad_norm": 16.45245361328125,
      "learning_rate": 1e-06,
      "loss": 0.3967,
      "num_input_tokens_seen": 278321340,
      "step": 4968
    },
    {
      "epoch": 11.064587973273943,
      "loss": 0.4062022566795349,
      "loss_ce": 0.00013537969789467752,
      "loss_iou": 0.177734375,
      "loss_num": 0.00994873046875,
      "loss_xval": 0.40625,
      "num_input_tokens_seen": 278321340,
      "step": 4968
    },
    {
      "epoch": 11.066815144766148,
      "grad_norm": 13.446118354797363,
      "learning_rate": 1e-06,
      "loss": 0.4845,
      "num_input_tokens_seen": 278377700,
      "step": 4969
    },
    {
      "epoch": 11.066815144766148,
      "loss": 0.5192990303039551,
      "loss_ce": 0.00013396976282820106,
      "loss_iou": 0.2255859375,
      "loss_num": 0.01361083984375,
      "loss_xval": 0.51953125,
      "num_input_tokens_seen": 278377700,
      "step": 4969
    },
    {
      "epoch": 11.069042316258352,
      "grad_norm": 31.59632682800293,
      "learning_rate": 1e-06,
      "loss": 0.5488,
      "num_input_tokens_seen": 278431796,
      "step": 4970
    },
    {
      "epoch": 11.069042316258352,
      "loss": 0.5724695920944214,
      "loss_ce": 0.0001734097022563219,
      "loss_iou": 0.234375,
      "loss_num": 0.0206298828125,
      "loss_xval": 0.57421875,
      "num_input_tokens_seen": 278431796,
      "step": 4970
    },
    {
      "epoch": 11.071269487750557,
      "grad_norm": 21.374279022216797,
      "learning_rate": 1e-06,
      "loss": 0.5448,
      "num_input_tokens_seen": 278486652,
      "step": 4971
    },
    {
      "epoch": 11.071269487750557,
      "loss": 0.5677635073661804,
      "loss_ce": 0.00013656642113346606,
      "loss_iou": 0.2431640625,
      "loss_num": 0.0162353515625,
      "loss_xval": 0.56640625,
      "num_input_tokens_seen": 278486652,
      "step": 4971
    },
    {
      "epoch": 11.073496659242762,
      "grad_norm": 62.18361282348633,
      "learning_rate": 1e-06,
      "loss": 0.696,
      "num_input_tokens_seen": 278543052,
      "step": 4972
    },
    {
      "epoch": 11.073496659242762,
      "loss": 0.5261435508728027,
      "loss_ce": 0.00014258406008593738,
      "loss_iou": 0.232421875,
      "loss_num": 0.01220703125,
      "loss_xval": 0.52734375,
      "num_input_tokens_seen": 278543052,
      "step": 4972
    },
    {
      "epoch": 11.075723830734967,
      "grad_norm": 16.852285385131836,
      "learning_rate": 1e-06,
      "loss": 0.7198,
      "num_input_tokens_seen": 278598500,
      "step": 4973
    },
    {
      "epoch": 11.075723830734967,
      "loss": 0.7613972425460815,
      "loss_ce": 0.00016673312347847968,
      "loss_iou": 0.333984375,
      "loss_num": 0.0189208984375,
      "loss_xval": 0.76171875,
      "num_input_tokens_seen": 278598500,
      "step": 4973
    },
    {
      "epoch": 11.077951002227172,
      "grad_norm": 26.721847534179688,
      "learning_rate": 1e-06,
      "loss": 0.5459,
      "num_input_tokens_seen": 278652932,
      "step": 4974
    },
    {
      "epoch": 11.077951002227172,
      "loss": 0.6054716110229492,
      "loss_ce": 0.0001249103806912899,
      "loss_iou": 0.2314453125,
      "loss_num": 0.0289306640625,
      "loss_xval": 0.60546875,
      "num_input_tokens_seen": 278652932,
      "step": 4974
    },
    {
      "epoch": 11.080178173719377,
      "grad_norm": 28.737173080444336,
      "learning_rate": 1e-06,
      "loss": 0.5445,
      "num_input_tokens_seen": 278709224,
      "step": 4975
    },
    {
      "epoch": 11.080178173719377,
      "loss": 0.5305880904197693,
      "loss_ce": 0.00013153886538930237,
      "loss_iou": 0.208984375,
      "loss_num": 0.0224609375,
      "loss_xval": 0.53125,
      "num_input_tokens_seen": 278709224,
      "step": 4975
    },
    {
      "epoch": 11.082405345211582,
      "grad_norm": 22.68699836730957,
      "learning_rate": 1e-06,
      "loss": 0.4661,
      "num_input_tokens_seen": 278761808,
      "step": 4976
    },
    {
      "epoch": 11.082405345211582,
      "loss": 0.41276633739471436,
      "loss_ce": 0.00016868110105860978,
      "loss_iou": 0.181640625,
      "loss_num": 0.0096435546875,
      "loss_xval": 0.412109375,
      "num_input_tokens_seen": 278761808,
      "step": 4976
    },
    {
      "epoch": 11.084632516703786,
      "grad_norm": 23.453115463256836,
      "learning_rate": 1e-06,
      "loss": 0.6005,
      "num_input_tokens_seen": 278817304,
      "step": 4977
    },
    {
      "epoch": 11.084632516703786,
      "loss": 0.5315991044044495,
      "loss_ce": 0.00010497516632312909,
      "loss_iou": 0.220703125,
      "loss_num": 0.01806640625,
      "loss_xval": 0.53125,
      "num_input_tokens_seen": 278817304,
      "step": 4977
    },
    {
      "epoch": 11.086859688195991,
      "grad_norm": 22.387310028076172,
      "learning_rate": 1e-06,
      "loss": 0.5064,
      "num_input_tokens_seen": 278873696,
      "step": 4978
    },
    {
      "epoch": 11.086859688195991,
      "loss": 0.41907966136932373,
      "loss_ce": 0.00013435332220979035,
      "loss_iou": 0.1748046875,
      "loss_num": 0.01385498046875,
      "loss_xval": 0.41796875,
      "num_input_tokens_seen": 278873696,
      "step": 4978
    },
    {
      "epoch": 11.089086859688196,
      "grad_norm": 15.790838241577148,
      "learning_rate": 1e-06,
      "loss": 0.596,
      "num_input_tokens_seen": 278931180,
      "step": 4979
    },
    {
      "epoch": 11.089086859688196,
      "loss": 0.5526787042617798,
      "loss_ce": 0.00012747167784254998,
      "loss_iou": 0.248046875,
      "loss_num": 0.011474609375,
      "loss_xval": 0.55078125,
      "num_input_tokens_seen": 278931180,
      "step": 4979
    },
    {
      "epoch": 11.091314031180401,
      "grad_norm": 15.259247779846191,
      "learning_rate": 1e-06,
      "loss": 0.6733,
      "num_input_tokens_seen": 278988536,
      "step": 4980
    },
    {
      "epoch": 11.091314031180401,
      "loss": 0.7944159507751465,
      "loss_ce": 0.0004706614126916975,
      "loss_iou": 0.341796875,
      "loss_num": 0.0218505859375,
      "loss_xval": 0.79296875,
      "num_input_tokens_seen": 278988536,
      "step": 4980
    },
    {
      "epoch": 11.093541202672606,
      "grad_norm": 22.694530487060547,
      "learning_rate": 1e-06,
      "loss": 0.5318,
      "num_input_tokens_seen": 279045060,
      "step": 4981
    },
    {
      "epoch": 11.093541202672606,
      "loss": 0.5756011605262756,
      "loss_ce": 0.0001617103407625109,
      "loss_iou": 0.2490234375,
      "loss_num": 0.015380859375,
      "loss_xval": 0.57421875,
      "num_input_tokens_seen": 279045060,
      "step": 4981
    },
    {
      "epoch": 11.09576837416481,
      "grad_norm": 20.57561492919922,
      "learning_rate": 1e-06,
      "loss": 0.5533,
      "num_input_tokens_seen": 279101812,
      "step": 4982
    },
    {
      "epoch": 11.09576837416481,
      "loss": 0.5785654783248901,
      "loss_ce": 0.00019637157674878836,
      "loss_iou": 0.2392578125,
      "loss_num": 0.02001953125,
      "loss_xval": 0.578125,
      "num_input_tokens_seen": 279101812,
      "step": 4982
    },
    {
      "epoch": 11.097995545657016,
      "grad_norm": 20.943668365478516,
      "learning_rate": 1e-06,
      "loss": 0.4383,
      "num_input_tokens_seen": 279156604,
      "step": 4983
    },
    {
      "epoch": 11.097995545657016,
      "loss": 0.4315411448478699,
      "loss_ce": 0.0002667338994797319,
      "loss_iou": 0.1787109375,
      "loss_num": 0.0145263671875,
      "loss_xval": 0.431640625,
      "num_input_tokens_seen": 279156604,
      "step": 4983
    },
    {
      "epoch": 11.10022271714922,
      "grad_norm": 25.43724822998047,
      "learning_rate": 1e-06,
      "loss": 0.4991,
      "num_input_tokens_seen": 279212128,
      "step": 4984
    },
    {
      "epoch": 11.10022271714922,
      "loss": 0.560702919960022,
      "loss_ce": 0.00015607105160597712,
      "loss_iou": 0.248046875,
      "loss_num": 0.012939453125,
      "loss_xval": 0.5625,
      "num_input_tokens_seen": 279212128,
      "step": 4984
    },
    {
      "epoch": 11.102449888641425,
      "grad_norm": 30.57325553894043,
      "learning_rate": 1e-06,
      "loss": 0.6142,
      "num_input_tokens_seen": 279267056,
      "step": 4985
    },
    {
      "epoch": 11.102449888641425,
      "loss": 0.5670490860939026,
      "loss_ce": 0.00015456389519385993,
      "loss_iou": 0.2451171875,
      "loss_num": 0.0152587890625,
      "loss_xval": 0.56640625,
      "num_input_tokens_seen": 279267056,
      "step": 4985
    },
    {
      "epoch": 11.10467706013363,
      "grad_norm": 21.70029067993164,
      "learning_rate": 1e-06,
      "loss": 0.4192,
      "num_input_tokens_seen": 279322728,
      "step": 4986
    },
    {
      "epoch": 11.10467706013363,
      "loss": 0.4730498492717743,
      "loss_ce": 0.00014946176088415086,
      "loss_iou": 0.1982421875,
      "loss_num": 0.015380859375,
      "loss_xval": 0.47265625,
      "num_input_tokens_seen": 279322728,
      "step": 4986
    },
    {
      "epoch": 11.106904231625835,
      "grad_norm": 17.88077735900879,
      "learning_rate": 1e-06,
      "loss": 0.3673,
      "num_input_tokens_seen": 279379872,
      "step": 4987
    },
    {
      "epoch": 11.106904231625835,
      "loss": 0.3716070055961609,
      "loss_ce": 0.00014704751083627343,
      "loss_iou": 0.169921875,
      "loss_num": 0.006561279296875,
      "loss_xval": 0.37109375,
      "num_input_tokens_seen": 279379872,
      "step": 4987
    },
    {
      "epoch": 11.10913140311804,
      "grad_norm": 21.7382755279541,
      "learning_rate": 1e-06,
      "loss": 0.609,
      "num_input_tokens_seen": 279435852,
      "step": 4988
    },
    {
      "epoch": 11.10913140311804,
      "loss": 0.6673039197921753,
      "loss_ce": 0.00018964617629535496,
      "loss_iou": 0.265625,
      "loss_num": 0.0272216796875,
      "loss_xval": 0.66796875,
      "num_input_tokens_seen": 279435852,
      "step": 4988
    },
    {
      "epoch": 11.111358574610245,
      "grad_norm": 13.589532852172852,
      "learning_rate": 1e-06,
      "loss": 0.4054,
      "num_input_tokens_seen": 279492372,
      "step": 4989
    },
    {
      "epoch": 11.111358574610245,
      "loss": 0.5108894109725952,
      "loss_ce": 0.00014718393504153937,
      "loss_iou": 0.2255859375,
      "loss_num": 0.01190185546875,
      "loss_xval": 0.51171875,
      "num_input_tokens_seen": 279492372,
      "step": 4989
    },
    {
      "epoch": 11.11358574610245,
      "grad_norm": 30.877511978149414,
      "learning_rate": 1e-06,
      "loss": 0.4336,
      "num_input_tokens_seen": 279548732,
      "step": 4990
    },
    {
      "epoch": 11.11358574610245,
      "loss": 0.36500683426856995,
      "loss_ce": 0.0001386810909025371,
      "loss_iou": 0.1650390625,
      "loss_num": 0.007049560546875,
      "loss_xval": 0.365234375,
      "num_input_tokens_seen": 279548732,
      "step": 4990
    },
    {
      "epoch": 11.115812917594655,
      "grad_norm": 20.038894653320312,
      "learning_rate": 1e-06,
      "loss": 0.5222,
      "num_input_tokens_seen": 279605916,
      "step": 4991
    },
    {
      "epoch": 11.115812917594655,
      "loss": 0.6044121980667114,
      "loss_ce": 0.00016413633420597762,
      "loss_iou": 0.267578125,
      "loss_num": 0.0140380859375,
      "loss_xval": 0.60546875,
      "num_input_tokens_seen": 279605916,
      "step": 4991
    },
    {
      "epoch": 11.11804008908686,
      "grad_norm": 23.694477081298828,
      "learning_rate": 1e-06,
      "loss": 0.521,
      "num_input_tokens_seen": 279659500,
      "step": 4992
    },
    {
      "epoch": 11.11804008908686,
      "loss": 0.4539830684661865,
      "loss_ce": 0.00012564370990730822,
      "loss_iou": 0.201171875,
      "loss_num": 0.010498046875,
      "loss_xval": 0.453125,
      "num_input_tokens_seen": 279659500,
      "step": 4992
    },
    {
      "epoch": 11.120267260579064,
      "grad_norm": 16.13028907775879,
      "learning_rate": 1e-06,
      "loss": 0.5406,
      "num_input_tokens_seen": 279716076,
      "step": 4993
    },
    {
      "epoch": 11.120267260579064,
      "loss": 0.4950714707374573,
      "loss_ce": 0.00013739880523644388,
      "loss_iou": 0.2158203125,
      "loss_num": 0.01251220703125,
      "loss_xval": 0.494140625,
      "num_input_tokens_seen": 279716076,
      "step": 4993
    },
    {
      "epoch": 11.122494432071269,
      "grad_norm": 25.81940460205078,
      "learning_rate": 1e-06,
      "loss": 0.7181,
      "num_input_tokens_seen": 279772812,
      "step": 4994
    },
    {
      "epoch": 11.122494432071269,
      "loss": 0.6644489765167236,
      "loss_ce": 0.00014238519361242652,
      "loss_iou": 0.283203125,
      "loss_num": 0.02001953125,
      "loss_xval": 0.6640625,
      "num_input_tokens_seen": 279772812,
      "step": 4994
    },
    {
      "epoch": 11.124721603563474,
      "grad_norm": 21.302309036254883,
      "learning_rate": 1e-06,
      "loss": 0.5902,
      "num_input_tokens_seen": 279828292,
      "step": 4995
    },
    {
      "epoch": 11.124721603563474,
      "loss": 0.5325771570205688,
      "loss_ce": 0.00022852000256534666,
      "loss_iou": 0.240234375,
      "loss_num": 0.0106201171875,
      "loss_xval": 0.53125,
      "num_input_tokens_seen": 279828292,
      "step": 4995
    },
    {
      "epoch": 11.126948775055679,
      "grad_norm": 16.87586212158203,
      "learning_rate": 1e-06,
      "loss": 0.472,
      "num_input_tokens_seen": 279885472,
      "step": 4996
    },
    {
      "epoch": 11.126948775055679,
      "loss": 0.3971105217933655,
      "loss_ce": 0.0001378602028125897,
      "loss_iou": 0.166015625,
      "loss_num": 0.01312255859375,
      "loss_xval": 0.396484375,
      "num_input_tokens_seen": 279885472,
      "step": 4996
    },
    {
      "epoch": 11.129175946547884,
      "grad_norm": 20.097671508789062,
      "learning_rate": 1e-06,
      "loss": 0.6366,
      "num_input_tokens_seen": 279938956,
      "step": 4997
    },
    {
      "epoch": 11.129175946547884,
      "loss": 0.634530246257782,
      "loss_ce": 0.00013083560043014586,
      "loss_iou": 0.271484375,
      "loss_num": 0.01806640625,
      "loss_xval": 0.6328125,
      "num_input_tokens_seen": 279938956,
      "step": 4997
    },
    {
      "epoch": 11.131403118040089,
      "grad_norm": 16.503360748291016,
      "learning_rate": 1e-06,
      "loss": 0.5446,
      "num_input_tokens_seen": 279994336,
      "step": 4998
    },
    {
      "epoch": 11.131403118040089,
      "loss": 0.38283979892730713,
      "loss_ce": 0.0001493898016633466,
      "loss_iou": 0.16015625,
      "loss_num": 0.0123291015625,
      "loss_xval": 0.3828125,
      "num_input_tokens_seen": 279994336,
      "step": 4998
    },
    {
      "epoch": 11.133630289532293,
      "grad_norm": 13.981780052185059,
      "learning_rate": 1e-06,
      "loss": 0.606,
      "num_input_tokens_seen": 280048108,
      "step": 4999
    },
    {
      "epoch": 11.133630289532293,
      "loss": 0.5781067609786987,
      "loss_ce": 0.00010381722677266225,
      "loss_iou": 0.205078125,
      "loss_num": 0.03369140625,
      "loss_xval": 0.578125,
      "num_input_tokens_seen": 280048108,
      "step": 4999
    },
    {
      "epoch": 11.135857461024498,
      "grad_norm": 16.743953704833984,
      "learning_rate": 1e-06,
      "loss": 0.4433,
      "num_input_tokens_seen": 280103384,
      "step": 5000
    },
    {
      "epoch": 11.135857461024498,
      "eval_seeclick_web_CIoU": 0.5857278108596802,
      "eval_seeclick_web_GIoU": 0.5834327638149261,
      "eval_seeclick_web_IoU": 0.6043886542320251,
      "eval_seeclick_web_MAE_all": 0.015452081337571144,
      "eval_seeclick_web_MAE_h": 0.0073387217707931995,
      "eval_seeclick_web_MAE_w": 0.015388870611786842,
      "eval_seeclick_web_MAE_x_boxes": 0.009706755401566625,
      "eval_seeclick_web_MAE_y_boxes": 0.021358829457312822,
      "eval_seeclick_web_inside_bbox": 0.9166666567325592,
      "eval_seeclick_web_loss": 0.9223926067352295,
      "eval_seeclick_web_loss_ce": 0.0001983401962206699,
      "eval_seeclick_web_loss_iou": 0.4200439453125,
      "eval_seeclick_web_loss_num": 0.01239013671875,
      "eval_seeclick_web_loss_xval": 0.90185546875,
      "eval_seeclick_web_runtime": 21.9551,
      "eval_seeclick_web_samples_per_second": 2.277,
      "eval_seeclick_web_steps_per_second": 0.091,
      "num_input_tokens_seen": 280103384,
      "step": 5000
    },
    {
      "epoch": 11.135857461024498,
      "eval_icons_CIoU": 0.2814893424510956,
      "eval_icons_GIoU": 0.3012331575155258,
      "eval_icons_IoU": 0.3595561385154724,
      "eval_icons_MAE_all": 0.059540608897805214,
      "eval_icons_MAE_h": 0.03891334868967533,
      "eval_icons_MAE_w": 0.0580837675370276,
      "eval_icons_MAE_x_boxes": 0.05784302018582821,
      "eval_icons_MAE_y_boxes": 0.038528745993971825,
      "eval_icons_inside_bbox": 0.6059027910232544,
      "eval_icons_loss": 1.7233656644821167,
      "eval_icons_loss_ce": 0.00022988053387962282,
      "eval_icons_loss_iou": 0.678466796875,
      "eval_icons_loss_num": 0.058696746826171875,
      "eval_icons_loss_xval": 1.6513671875,
      "eval_icons_runtime": 20.195,
      "eval_icons_samples_per_second": 2.476,
      "eval_icons_steps_per_second": 0.099,
      "num_input_tokens_seen": 280103384,
      "step": 5000
    },
    {
      "epoch": 11.135857461024498,
      "eval_screenspot_CIoU": 0.3535158932209015,
      "eval_screenspot_GIoU": 0.36772539218266803,
      "eval_screenspot_IoU": 0.43389413754145306,
      "eval_screenspot_MAE_all": 0.05942438915371895,
      "eval_screenspot_MAE_h": 0.03856873946885268,
      "eval_screenspot_MAE_w": 0.0688897892832756,
      "eval_screenspot_MAE_x_boxes": 0.07383330973486106,
      "eval_screenspot_MAE_y_boxes": 0.03976897584895293,
      "eval_screenspot_inside_bbox": 0.6862499912579855,
      "eval_screenspot_loss": 1.6220872402191162,
      "eval_screenspot_loss_ce": 0.000261851722219338,
      "eval_screenspot_loss_iou": 0.67236328125,
      "eval_screenspot_loss_num": 0.06821314493815105,
      "eval_screenspot_loss_xval": 1.6868489583333333,
      "eval_screenspot_runtime": 33.7501,
      "eval_screenspot_samples_per_second": 2.637,
      "eval_screenspot_steps_per_second": 0.089,
      "num_input_tokens_seen": 280103384,
      "step": 5000
    },
    {
      "epoch": 11.135857461024498,
      "eval_compot_CIoU": 0.3433762341737747,
      "eval_compot_GIoU": 0.35395348072052,
      "eval_compot_IoU": 0.4032938480377197,
      "eval_compot_MAE_all": 0.01813736092299223,
      "eval_compot_MAE_h": 0.009242744650691748,
      "eval_compot_MAE_w": 0.02075517177581787,
      "eval_compot_MAE_x_boxes": 0.030595741234719753,
      "eval_compot_MAE_y_boxes": 0.007154007675126195,
      "eval_compot_inside_bbox": 0.6302083432674408,
      "eval_compot_loss": 1.3828521966934204,
      "eval_compot_loss_ce": 0.000194034306332469,
      "eval_compot_loss_iou": 0.6356201171875,
      "eval_compot_loss_num": 0.016706466674804688,
      "eval_compot_loss_xval": 1.35498046875,
      "eval_compot_runtime": 22.2096,
      "eval_compot_samples_per_second": 2.251,
      "eval_compot_steps_per_second": 0.09,
      "num_input_tokens_seen": 280103384,
      "step": 5000
    },
    {
      "epoch": 11.135857461024498,
      "eval_custom_ui_val_CIoU": 0.46921708765957093,
      "eval_custom_ui_val_GIoU": 0.48126794232262504,
      "eval_custom_ui_val_IoU": 0.5315591428014967,
      "eval_custom_ui_val_MAE_all": 0.030911025901635487,
      "eval_custom_ui_val_MAE_h": 0.016821589320898056,
      "eval_custom_ui_val_MAE_w": 0.0393333797239595,
      "eval_custom_ui_val_MAE_x_boxes": 0.03895654301676485,
      "eval_custom_ui_val_MAE_y_boxes": 0.015129889025249414,
      "eval_custom_ui_val_inside_bbox": 0.7527006202273898,
      "eval_custom_ui_val_loss": 1.2006866931915283,
      "eval_custom_ui_val_loss_ce": 0.00023336601023200073,
      "eval_custom_ui_val_loss_iou": 0.5094129774305556,
      "eval_custom_ui_val_loss_num": 0.028292549981011286,
      "eval_custom_ui_val_loss_xval": 1.1603190104166667,
      "eval_custom_ui_val_runtime": 65.5609,
      "eval_custom_ui_val_samples_per_second": 4.042,
      "eval_custom_ui_val_steps_per_second": 0.137,
      "num_input_tokens_seen": 280103384,
      "step": 5000
    },
    {
      "epoch": 11.135857461024498,
      "loss": 0.8996152877807617,
      "loss_ce": 0.00020119547843933105,
      "loss_iou": 0.38671875,
      "loss_num": 0.02490234375,
      "loss_xval": 0.8984375,
      "num_input_tokens_seen": 280103384,
      "step": 5000
    },
    {
      "epoch": 11.138084632516703,
      "grad_norm": 19.082584381103516,
      "learning_rate": 1e-06,
      "loss": 0.4929,
      "num_input_tokens_seen": 280158504,
      "step": 5001
    },
    {
      "epoch": 11.138084632516703,
      "loss": 0.39870956540107727,
      "loss_ce": 0.00015000064740888774,
      "loss_iou": 0.1728515625,
      "loss_num": 0.010498046875,
      "loss_xval": 0.3984375,
      "num_input_tokens_seen": 280158504,
      "step": 5001
    },
    {
      "epoch": 11.140311804008908,
      "grad_norm": 26.122913360595703,
      "learning_rate": 1e-06,
      "loss": 0.5198,
      "num_input_tokens_seen": 280214744,
      "step": 5002
    },
    {
      "epoch": 11.140311804008908,
      "loss": 0.45124655961990356,
      "loss_ce": 0.00013570950250141323,
      "loss_iou": 0.19921875,
      "loss_num": 0.0107421875,
      "loss_xval": 0.451171875,
      "num_input_tokens_seen": 280214744,
      "step": 5002
    },
    {
      "epoch": 11.142538975501113,
      "grad_norm": 27.3781681060791,
      "learning_rate": 1e-06,
      "loss": 0.4616,
      "num_input_tokens_seen": 280271032,
      "step": 5003
    },
    {
      "epoch": 11.142538975501113,
      "loss": 0.32605159282684326,
      "loss_ce": 0.00012386470916680992,
      "loss_iou": 0.125,
      "loss_num": 0.0150146484375,
      "loss_xval": 0.326171875,
      "num_input_tokens_seen": 280271032,
      "step": 5003
    },
    {
      "epoch": 11.144766146993318,
      "grad_norm": 19.879722595214844,
      "learning_rate": 1e-06,
      "loss": 0.531,
      "num_input_tokens_seen": 280327288,
      "step": 5004
    },
    {
      "epoch": 11.144766146993318,
      "loss": 0.43859922885894775,
      "loss_ce": 0.00012264544784557074,
      "loss_iou": 0.1982421875,
      "loss_num": 0.00823974609375,
      "loss_xval": 0.4375,
      "num_input_tokens_seen": 280327288,
      "step": 5004
    },
    {
      "epoch": 11.146993318485523,
      "grad_norm": 22.407188415527344,
      "learning_rate": 1e-06,
      "loss": 0.7016,
      "num_input_tokens_seen": 280383788,
      "step": 5005
    },
    {
      "epoch": 11.146993318485523,
      "loss": 0.49451902508735657,
      "loss_ce": 0.00013426925579551607,
      "loss_iou": 0.203125,
      "loss_num": 0.0177001953125,
      "loss_xval": 0.494140625,
      "num_input_tokens_seen": 280383788,
      "step": 5005
    },
    {
      "epoch": 11.14922048997773,
      "grad_norm": 17.652860641479492,
      "learning_rate": 1e-06,
      "loss": 0.4529,
      "num_input_tokens_seen": 280438888,
      "step": 5006
    },
    {
      "epoch": 11.14922048997773,
      "loss": 0.38709527254104614,
      "loss_ce": 0.00013238785322755575,
      "loss_iou": 0.1796875,
      "loss_num": 0.00537109375,
      "loss_xval": 0.38671875,
      "num_input_tokens_seen": 280438888,
      "step": 5006
    },
    {
      "epoch": 11.151447661469934,
      "grad_norm": 28.968141555786133,
      "learning_rate": 1e-06,
      "loss": 0.561,
      "num_input_tokens_seen": 280494244,
      "step": 5007
    },
    {
      "epoch": 11.151447661469934,
      "loss": 0.5118998289108276,
      "loss_ce": 0.0005473287310451269,
      "loss_iou": 0.2158203125,
      "loss_num": 0.0159912109375,
      "loss_xval": 0.51171875,
      "num_input_tokens_seen": 280494244,
      "step": 5007
    },
    {
      "epoch": 11.153674832962139,
      "grad_norm": 22.353212356567383,
      "learning_rate": 1e-06,
      "loss": 0.5322,
      "num_input_tokens_seen": 280547136,
      "step": 5008
    },
    {
      "epoch": 11.153674832962139,
      "loss": 0.6800668239593506,
      "loss_ce": 0.00013513598241843283,
      "loss_iou": 0.30078125,
      "loss_num": 0.0159912109375,
      "loss_xval": 0.6796875,
      "num_input_tokens_seen": 280547136,
      "step": 5008
    },
    {
      "epoch": 11.155902004454344,
      "grad_norm": 16.07948875427246,
      "learning_rate": 1e-06,
      "loss": 0.6479,
      "num_input_tokens_seen": 280605596,
      "step": 5009
    },
    {
      "epoch": 11.155902004454344,
      "loss": 0.7251046895980835,
      "loss_ce": 0.00012912959209643304,
      "loss_iou": 0.2890625,
      "loss_num": 0.029296875,
      "loss_xval": 0.7265625,
      "num_input_tokens_seen": 280605596,
      "step": 5009
    },
    {
      "epoch": 11.158129175946549,
      "grad_norm": 21.49631690979004,
      "learning_rate": 1e-06,
      "loss": 0.411,
      "num_input_tokens_seen": 280662056,
      "step": 5010
    },
    {
      "epoch": 11.158129175946549,
      "loss": 0.46879857778549194,
      "loss_ce": 0.000170662795426324,
      "loss_iou": 0.2138671875,
      "loss_num": 0.00836181640625,
      "loss_xval": 0.46875,
      "num_input_tokens_seen": 280662056,
      "step": 5010
    },
    {
      "epoch": 11.160356347438753,
      "grad_norm": 25.566015243530273,
      "learning_rate": 1e-06,
      "loss": 0.4451,
      "num_input_tokens_seen": 280718144,
      "step": 5011
    },
    {
      "epoch": 11.160356347438753,
      "loss": 0.3402805030345917,
      "loss_ce": 0.00019259938562754542,
      "loss_iou": 0.1533203125,
      "loss_num": 0.00677490234375,
      "loss_xval": 0.33984375,
      "num_input_tokens_seen": 280718144,
      "step": 5011
    },
    {
      "epoch": 11.162583518930958,
      "grad_norm": 19.456497192382812,
      "learning_rate": 1e-06,
      "loss": 0.5608,
      "num_input_tokens_seen": 280769956,
      "step": 5012
    },
    {
      "epoch": 11.162583518930958,
      "loss": 0.5276530981063843,
      "loss_ce": 0.00018727785209193826,
      "loss_iou": 0.228515625,
      "loss_num": 0.0140380859375,
      "loss_xval": 0.52734375,
      "num_input_tokens_seen": 280769956,
      "step": 5012
    },
    {
      "epoch": 11.164810690423163,
      "grad_norm": 19.570919036865234,
      "learning_rate": 1e-06,
      "loss": 0.4798,
      "num_input_tokens_seen": 280824936,
      "step": 5013
    },
    {
      "epoch": 11.164810690423163,
      "loss": 0.3978237807750702,
      "loss_ce": 0.00011871426249854267,
      "loss_iou": 0.15625,
      "loss_num": 0.016845703125,
      "loss_xval": 0.3984375,
      "num_input_tokens_seen": 280824936,
      "step": 5013
    },
    {
      "epoch": 11.167037861915368,
      "grad_norm": 26.35681915283203,
      "learning_rate": 1e-06,
      "loss": 0.5786,
      "num_input_tokens_seen": 280878960,
      "step": 5014
    },
    {
      "epoch": 11.167037861915368,
      "loss": 0.6183251142501831,
      "loss_ce": 0.00016105023678392172,
      "loss_iou": 0.255859375,
      "loss_num": 0.0213623046875,
      "loss_xval": 0.6171875,
      "num_input_tokens_seen": 280878960,
      "step": 5014
    },
    {
      "epoch": 11.169265033407573,
      "grad_norm": 20.287382125854492,
      "learning_rate": 1e-06,
      "loss": 0.5377,
      "num_input_tokens_seen": 280934428,
      "step": 5015
    },
    {
      "epoch": 11.169265033407573,
      "loss": 0.5026686191558838,
      "loss_ce": 0.00010515956091694534,
      "loss_iou": 0.2060546875,
      "loss_num": 0.01806640625,
      "loss_xval": 0.50390625,
      "num_input_tokens_seen": 280934428,
      "step": 5015
    },
    {
      "epoch": 11.171492204899778,
      "grad_norm": 27.965999603271484,
      "learning_rate": 1e-06,
      "loss": 0.514,
      "num_input_tokens_seen": 280992196,
      "step": 5016
    },
    {
      "epoch": 11.171492204899778,
      "loss": 0.521459698677063,
      "loss_ce": 0.00021950851078145206,
      "loss_iou": 0.2216796875,
      "loss_num": 0.015625,
      "loss_xval": 0.51953125,
      "num_input_tokens_seen": 280992196,
      "step": 5016
    },
    {
      "epoch": 11.173719376391983,
      "grad_norm": 16.570053100585938,
      "learning_rate": 1e-06,
      "loss": 0.4161,
      "num_input_tokens_seen": 281050828,
      "step": 5017
    },
    {
      "epoch": 11.173719376391983,
      "loss": 0.4344647526741028,
      "loss_ce": 0.00013857701560482383,
      "loss_iou": 0.1796875,
      "loss_num": 0.01513671875,
      "loss_xval": 0.43359375,
      "num_input_tokens_seen": 281050828,
      "step": 5017
    },
    {
      "epoch": 11.175946547884188,
      "grad_norm": 16.34454345703125,
      "learning_rate": 1e-06,
      "loss": 0.6104,
      "num_input_tokens_seen": 281106284,
      "step": 5018
    },
    {
      "epoch": 11.175946547884188,
      "loss": 0.7851086854934692,
      "loss_ce": 0.00019656957010738552,
      "loss_iou": 0.361328125,
      "loss_num": 0.01275634765625,
      "loss_xval": 0.78515625,
      "num_input_tokens_seen": 281106284,
      "step": 5018
    },
    {
      "epoch": 11.178173719376392,
      "grad_norm": 15.942395210266113,
      "learning_rate": 1e-06,
      "loss": 0.4458,
      "num_input_tokens_seen": 281164500,
      "step": 5019
    },
    {
      "epoch": 11.178173719376392,
      "loss": 0.3177483081817627,
      "loss_ce": 0.00012136220175307244,
      "loss_iou": 0.140625,
      "loss_num": 0.007080078125,
      "loss_xval": 0.318359375,
      "num_input_tokens_seen": 281164500,
      "step": 5019
    },
    {
      "epoch": 11.180400890868597,
      "grad_norm": 20.594022750854492,
      "learning_rate": 1e-06,
      "loss": 0.6501,
      "num_input_tokens_seen": 281222920,
      "step": 5020
    },
    {
      "epoch": 11.180400890868597,
      "loss": 0.4495760202407837,
      "loss_ce": 0.00011313259165035561,
      "loss_iou": 0.2041015625,
      "loss_num": 0.00836181640625,
      "loss_xval": 0.44921875,
      "num_input_tokens_seen": 281222920,
      "step": 5020
    },
    {
      "epoch": 11.182628062360802,
      "grad_norm": 20.63737678527832,
      "learning_rate": 1e-06,
      "loss": 0.7283,
      "num_input_tokens_seen": 281280520,
      "step": 5021
    },
    {
      "epoch": 11.182628062360802,
      "loss": 0.8178755640983582,
      "loss_ce": 0.00012653997691813856,
      "loss_iou": 0.330078125,
      "loss_num": 0.03173828125,
      "loss_xval": 0.81640625,
      "num_input_tokens_seen": 281280520,
      "step": 5021
    },
    {
      "epoch": 11.184855233853007,
      "grad_norm": 19.35364532470703,
      "learning_rate": 1e-06,
      "loss": 0.5065,
      "num_input_tokens_seen": 281338688,
      "step": 5022
    },
    {
      "epoch": 11.184855233853007,
      "loss": 0.5651026964187622,
      "loss_ce": 0.00016126442642416805,
      "loss_iou": 0.244140625,
      "loss_num": 0.01519775390625,
      "loss_xval": 0.56640625,
      "num_input_tokens_seen": 281338688,
      "step": 5022
    },
    {
      "epoch": 11.187082405345212,
      "grad_norm": 17.587987899780273,
      "learning_rate": 1e-06,
      "loss": 0.41,
      "num_input_tokens_seen": 281394900,
      "step": 5023
    },
    {
      "epoch": 11.187082405345212,
      "loss": 0.2979922890663147,
      "loss_ce": 0.00014070735778659582,
      "loss_iou": 0.119140625,
      "loss_num": 0.01202392578125,
      "loss_xval": 0.296875,
      "num_input_tokens_seen": 281394900,
      "step": 5023
    },
    {
      "epoch": 11.189309576837417,
      "grad_norm": 42.784358978271484,
      "learning_rate": 1e-06,
      "loss": 0.5281,
      "num_input_tokens_seen": 281448492,
      "step": 5024
    },
    {
      "epoch": 11.189309576837417,
      "loss": 0.5797094106674194,
      "loss_ce": 0.00011953900684602559,
      "loss_iou": 0.263671875,
      "loss_num": 0.01025390625,
      "loss_xval": 0.578125,
      "num_input_tokens_seen": 281448492,
      "step": 5024
    },
    {
      "epoch": 11.191536748329622,
      "grad_norm": 19.45455551147461,
      "learning_rate": 1e-06,
      "loss": 0.5473,
      "num_input_tokens_seen": 281505840,
      "step": 5025
    },
    {
      "epoch": 11.191536748329622,
      "loss": 0.4985758662223816,
      "loss_ce": 0.00016279635019600391,
      "loss_iou": 0.2177734375,
      "loss_num": 0.012451171875,
      "loss_xval": 0.498046875,
      "num_input_tokens_seen": 281505840,
      "step": 5025
    },
    {
      "epoch": 11.193763919821826,
      "grad_norm": 14.789341926574707,
      "learning_rate": 1e-06,
      "loss": 0.428,
      "num_input_tokens_seen": 281563968,
      "step": 5026
    },
    {
      "epoch": 11.193763919821826,
      "loss": 0.5229382514953613,
      "loss_ce": 0.00023318573948927224,
      "loss_iou": 0.2099609375,
      "loss_num": 0.0206298828125,
      "loss_xval": 0.5234375,
      "num_input_tokens_seen": 281563968,
      "step": 5026
    },
    {
      "epoch": 11.195991091314031,
      "grad_norm": 24.538976669311523,
      "learning_rate": 1e-06,
      "loss": 0.5299,
      "num_input_tokens_seen": 281618908,
      "step": 5027
    },
    {
      "epoch": 11.195991091314031,
      "loss": 0.4837605655193329,
      "loss_ce": 0.0001179840401164256,
      "loss_iou": 0.20703125,
      "loss_num": 0.01397705078125,
      "loss_xval": 0.484375,
      "num_input_tokens_seen": 281618908,
      "step": 5027
    },
    {
      "epoch": 11.198218262806236,
      "grad_norm": 16.29832649230957,
      "learning_rate": 1e-06,
      "loss": 0.4994,
      "num_input_tokens_seen": 281674612,
      "step": 5028
    },
    {
      "epoch": 11.198218262806236,
      "loss": 0.5437781810760498,
      "loss_ce": 0.00013804963964503258,
      "loss_iou": 0.2314453125,
      "loss_num": 0.0159912109375,
      "loss_xval": 0.54296875,
      "num_input_tokens_seen": 281674612,
      "step": 5028
    },
    {
      "epoch": 11.200445434298441,
      "grad_norm": 16.095746994018555,
      "learning_rate": 1e-06,
      "loss": 0.6377,
      "num_input_tokens_seen": 281731544,
      "step": 5029
    },
    {
      "epoch": 11.200445434298441,
      "loss": 0.584113359451294,
      "loss_ce": 0.0003731203032657504,
      "loss_iou": 0.224609375,
      "loss_num": 0.027099609375,
      "loss_xval": 0.58203125,
      "num_input_tokens_seen": 281731544,
      "step": 5029
    },
    {
      "epoch": 11.202672605790646,
      "grad_norm": 18.842683792114258,
      "learning_rate": 1e-06,
      "loss": 0.3468,
      "num_input_tokens_seen": 281788084,
      "step": 5030
    },
    {
      "epoch": 11.202672605790646,
      "loss": 0.38637834787368774,
      "loss_ce": 0.00014787877444177866,
      "loss_iou": 0.1611328125,
      "loss_num": 0.0125732421875,
      "loss_xval": 0.38671875,
      "num_input_tokens_seen": 281788084,
      "step": 5030
    },
    {
      "epoch": 11.20489977728285,
      "grad_norm": 17.219839096069336,
      "learning_rate": 1e-06,
      "loss": 0.655,
      "num_input_tokens_seen": 281846932,
      "step": 5031
    },
    {
      "epoch": 11.20489977728285,
      "loss": 0.9407340884208679,
      "loss_ce": 0.00018237490439787507,
      "loss_iou": 0.36328125,
      "loss_num": 0.042236328125,
      "loss_xval": 0.94140625,
      "num_input_tokens_seen": 281846932,
      "step": 5031
    },
    {
      "epoch": 11.207126948775056,
      "grad_norm": 22.804384231567383,
      "learning_rate": 1e-06,
      "loss": 0.452,
      "num_input_tokens_seen": 281905708,
      "step": 5032
    },
    {
      "epoch": 11.207126948775056,
      "loss": 0.5562899112701416,
      "loss_ce": 0.00013752697850577533,
      "loss_iou": 0.255859375,
      "loss_num": 0.00860595703125,
      "loss_xval": 0.5546875,
      "num_input_tokens_seen": 281905708,
      "step": 5032
    },
    {
      "epoch": 11.20935412026726,
      "grad_norm": 23.392213821411133,
      "learning_rate": 1e-06,
      "loss": 0.5097,
      "num_input_tokens_seen": 281963088,
      "step": 5033
    },
    {
      "epoch": 11.20935412026726,
      "loss": 0.4477725028991699,
      "loss_ce": 0.00014065910363569856,
      "loss_iou": 0.193359375,
      "loss_num": 0.01220703125,
      "loss_xval": 0.447265625,
      "num_input_tokens_seen": 281963088,
      "step": 5033
    },
    {
      "epoch": 11.211581291759465,
      "grad_norm": 26.24521255493164,
      "learning_rate": 1e-06,
      "loss": 0.673,
      "num_input_tokens_seen": 282019504,
      "step": 5034
    },
    {
      "epoch": 11.211581291759465,
      "loss": 0.5856134295463562,
      "loss_ce": 0.00016419796156696975,
      "loss_iou": 0.251953125,
      "loss_num": 0.0166015625,
      "loss_xval": 0.5859375,
      "num_input_tokens_seen": 282019504,
      "step": 5034
    },
    {
      "epoch": 11.21380846325167,
      "grad_norm": 13.377967834472656,
      "learning_rate": 1e-06,
      "loss": 0.5941,
      "num_input_tokens_seen": 282077584,
      "step": 5035
    },
    {
      "epoch": 11.21380846325167,
      "loss": 0.5731761455535889,
      "loss_ce": 0.00017806813411880285,
      "loss_iou": 0.263671875,
      "loss_num": 0.0091552734375,
      "loss_xval": 0.57421875,
      "num_input_tokens_seen": 282077584,
      "step": 5035
    },
    {
      "epoch": 11.216035634743875,
      "grad_norm": 23.46733283996582,
      "learning_rate": 1e-06,
      "loss": 0.555,
      "num_input_tokens_seen": 282135040,
      "step": 5036
    },
    {
      "epoch": 11.216035634743875,
      "loss": 0.5774731040000916,
      "loss_ce": 0.00020261471217963845,
      "loss_iou": 0.251953125,
      "loss_num": 0.014892578125,
      "loss_xval": 0.578125,
      "num_input_tokens_seen": 282135040,
      "step": 5036
    },
    {
      "epoch": 11.21826280623608,
      "grad_norm": 18.02016258239746,
      "learning_rate": 1e-06,
      "loss": 0.5232,
      "num_input_tokens_seen": 282192724,
      "step": 5037
    },
    {
      "epoch": 11.21826280623608,
      "loss": 0.5690996050834656,
      "loss_ce": 0.00012990040704607964,
      "loss_iou": 0.2158203125,
      "loss_num": 0.027587890625,
      "loss_xval": 0.5703125,
      "num_input_tokens_seen": 282192724,
      "step": 5037
    },
    {
      "epoch": 11.220489977728285,
      "grad_norm": 17.373300552368164,
      "learning_rate": 1e-06,
      "loss": 0.4308,
      "num_input_tokens_seen": 282245988,
      "step": 5038
    },
    {
      "epoch": 11.220489977728285,
      "loss": 0.4509298801422119,
      "loss_ce": 0.00012421750579960644,
      "loss_iou": 0.193359375,
      "loss_num": 0.0126953125,
      "loss_xval": 0.451171875,
      "num_input_tokens_seen": 282245988,
      "step": 5038
    },
    {
      "epoch": 11.22271714922049,
      "grad_norm": 16.49465560913086,
      "learning_rate": 1e-06,
      "loss": 0.6109,
      "num_input_tokens_seen": 282301460,
      "step": 5039
    },
    {
      "epoch": 11.22271714922049,
      "loss": 0.5691109895706177,
      "loss_ce": 0.00014124812150839716,
      "loss_iou": 0.2578125,
      "loss_num": 0.010498046875,
      "loss_xval": 0.5703125,
      "num_input_tokens_seen": 282301460,
      "step": 5039
    },
    {
      "epoch": 11.224944320712694,
      "grad_norm": 24.778156280517578,
      "learning_rate": 1e-06,
      "loss": 0.5473,
      "num_input_tokens_seen": 282356420,
      "step": 5040
    },
    {
      "epoch": 11.224944320712694,
      "loss": 0.5384612679481506,
      "loss_ce": 0.0001312288804911077,
      "loss_iou": 0.23828125,
      "loss_num": 0.01220703125,
      "loss_xval": 0.5390625,
      "num_input_tokens_seen": 282356420,
      "step": 5040
    },
    {
      "epoch": 11.2271714922049,
      "grad_norm": 17.579057693481445,
      "learning_rate": 1e-06,
      "loss": 0.5956,
      "num_input_tokens_seen": 282411144,
      "step": 5041
    },
    {
      "epoch": 11.2271714922049,
      "loss": 0.5568938851356506,
      "loss_ce": 0.00013120746007189155,
      "loss_iou": 0.236328125,
      "loss_num": 0.0167236328125,
      "loss_xval": 0.55859375,
      "num_input_tokens_seen": 282411144,
      "step": 5041
    },
    {
      "epoch": 11.229398663697104,
      "grad_norm": 20.767990112304688,
      "learning_rate": 1e-06,
      "loss": 0.5745,
      "num_input_tokens_seen": 282466800,
      "step": 5042
    },
    {
      "epoch": 11.229398663697104,
      "loss": 0.35937535762786865,
      "loss_ce": 0.00012243175297044218,
      "loss_iou": 0.162109375,
      "loss_num": 0.006927490234375,
      "loss_xval": 0.359375,
      "num_input_tokens_seen": 282466800,
      "step": 5042
    },
    {
      "epoch": 11.231625835189309,
      "grad_norm": 18.92509651184082,
      "learning_rate": 1e-06,
      "loss": 0.4772,
      "num_input_tokens_seen": 282524332,
      "step": 5043
    },
    {
      "epoch": 11.231625835189309,
      "loss": 0.47509121894836426,
      "loss_ce": 0.00011561952851479873,
      "loss_iou": 0.22265625,
      "loss_num": 0.006011962890625,
      "loss_xval": 0.474609375,
      "num_input_tokens_seen": 282524332,
      "step": 5043
    },
    {
      "epoch": 11.233853006681514,
      "grad_norm": 22.10957908630371,
      "learning_rate": 1e-06,
      "loss": 0.5585,
      "num_input_tokens_seen": 282580732,
      "step": 5044
    },
    {
      "epoch": 11.233853006681514,
      "loss": 0.7198188304901123,
      "loss_ce": 0.00033639208413660526,
      "loss_iou": 0.306640625,
      "loss_num": 0.0213623046875,
      "loss_xval": 0.71875,
      "num_input_tokens_seen": 282580732,
      "step": 5044
    },
    {
      "epoch": 11.236080178173719,
      "grad_norm": 20.335216522216797,
      "learning_rate": 1e-06,
      "loss": 0.3889,
      "num_input_tokens_seen": 282638292,
      "step": 5045
    },
    {
      "epoch": 11.236080178173719,
      "loss": 0.44411370158195496,
      "loss_ce": 0.00014397443737834692,
      "loss_iou": 0.2041015625,
      "loss_num": 0.00701904296875,
      "loss_xval": 0.443359375,
      "num_input_tokens_seen": 282638292,
      "step": 5045
    },
    {
      "epoch": 11.238307349665924,
      "grad_norm": 17.235029220581055,
      "learning_rate": 1e-06,
      "loss": 0.5289,
      "num_input_tokens_seen": 282695300,
      "step": 5046
    },
    {
      "epoch": 11.238307349665924,
      "loss": 0.5496923923492432,
      "loss_ce": 0.00013185839634388685,
      "loss_iou": 0.23828125,
      "loss_num": 0.014404296875,
      "loss_xval": 0.55078125,
      "num_input_tokens_seen": 282695300,
      "step": 5046
    },
    {
      "epoch": 11.240534521158128,
      "grad_norm": 15.562467575073242,
      "learning_rate": 1e-06,
      "loss": 0.7311,
      "num_input_tokens_seen": 282751756,
      "step": 5047
    },
    {
      "epoch": 11.240534521158128,
      "loss": 0.8263590335845947,
      "loss_ce": 0.00018715820624493062,
      "loss_iou": 0.33984375,
      "loss_num": 0.029052734375,
      "loss_xval": 0.828125,
      "num_input_tokens_seen": 282751756,
      "step": 5047
    },
    {
      "epoch": 11.242761692650333,
      "grad_norm": 12.08515453338623,
      "learning_rate": 1e-06,
      "loss": 0.5536,
      "num_input_tokens_seen": 282807936,
      "step": 5048
    },
    {
      "epoch": 11.242761692650333,
      "loss": 0.4205526113510132,
      "loss_ce": 0.00014246124192140996,
      "loss_iou": 0.1904296875,
      "loss_num": 0.0078125,
      "loss_xval": 0.419921875,
      "num_input_tokens_seen": 282807936,
      "step": 5048
    },
    {
      "epoch": 11.244988864142538,
      "grad_norm": 15.926555633544922,
      "learning_rate": 1e-06,
      "loss": 0.5036,
      "num_input_tokens_seen": 282863292,
      "step": 5049
    },
    {
      "epoch": 11.244988864142538,
      "loss": 0.6954518556594849,
      "loss_ce": 0.00013932373258285224,
      "loss_iou": 0.298828125,
      "loss_num": 0.0194091796875,
      "loss_xval": 0.6953125,
      "num_input_tokens_seen": 282863292,
      "step": 5049
    },
    {
      "epoch": 11.247216035634743,
      "grad_norm": 41.89063262939453,
      "learning_rate": 1e-06,
      "loss": 0.405,
      "num_input_tokens_seen": 282917964,
      "step": 5050
    },
    {
      "epoch": 11.247216035634743,
      "loss": 0.4797305464744568,
      "loss_ce": 0.00011627860658336431,
      "loss_iou": 0.216796875,
      "loss_num": 0.00909423828125,
      "loss_xval": 0.48046875,
      "num_input_tokens_seen": 282917964,
      "step": 5050
    },
    {
      "epoch": 11.249443207126948,
      "grad_norm": 20.080503463745117,
      "learning_rate": 1e-06,
      "loss": 0.5513,
      "num_input_tokens_seen": 282975304,
      "step": 5051
    },
    {
      "epoch": 11.249443207126948,
      "loss": 0.5874269604682922,
      "loss_ce": 0.0001466784015065059,
      "loss_iou": 0.255859375,
      "loss_num": 0.0150146484375,
      "loss_xval": 0.5859375,
      "num_input_tokens_seen": 282975304,
      "step": 5051
    },
    {
      "epoch": 11.251670378619155,
      "grad_norm": 18.951017379760742,
      "learning_rate": 1e-06,
      "loss": 0.4595,
      "num_input_tokens_seen": 283031220,
      "step": 5052
    },
    {
      "epoch": 11.251670378619155,
      "loss": 0.43687593936920166,
      "loss_ce": 0.00010837505396921188,
      "loss_iou": 0.1796875,
      "loss_num": 0.01556396484375,
      "loss_xval": 0.4375,
      "num_input_tokens_seen": 283031220,
      "step": 5052
    },
    {
      "epoch": 11.25389755011136,
      "grad_norm": 20.08397674560547,
      "learning_rate": 1e-06,
      "loss": 0.5126,
      "num_input_tokens_seen": 283088852,
      "step": 5053
    },
    {
      "epoch": 11.25389755011136,
      "loss": 0.47264423966407776,
      "loss_ce": 0.00011004768748534843,
      "loss_iou": 0.197265625,
      "loss_num": 0.015625,
      "loss_xval": 0.47265625,
      "num_input_tokens_seen": 283088852,
      "step": 5053
    },
    {
      "epoch": 11.256124721603564,
      "grad_norm": 23.235828399658203,
      "learning_rate": 1e-06,
      "loss": 0.6491,
      "num_input_tokens_seen": 283146176,
      "step": 5054
    },
    {
      "epoch": 11.256124721603564,
      "loss": 0.5091769695281982,
      "loss_ce": 0.000143796467455104,
      "loss_iou": 0.21484375,
      "loss_num": 0.015869140625,
      "loss_xval": 0.5078125,
      "num_input_tokens_seen": 283146176,
      "step": 5054
    },
    {
      "epoch": 11.25835189309577,
      "grad_norm": 18.110916137695312,
      "learning_rate": 1e-06,
      "loss": 0.4254,
      "num_input_tokens_seen": 283201764,
      "step": 5055
    },
    {
      "epoch": 11.25835189309577,
      "loss": 0.3657214045524597,
      "loss_ce": 0.00012084872287232429,
      "loss_iou": 0.1640625,
      "loss_num": 0.007568359375,
      "loss_xval": 0.365234375,
      "num_input_tokens_seen": 283201764,
      "step": 5055
    },
    {
      "epoch": 11.260579064587974,
      "grad_norm": 15.130694389343262,
      "learning_rate": 1e-06,
      "loss": 0.5045,
      "num_input_tokens_seen": 283259004,
      "step": 5056
    },
    {
      "epoch": 11.260579064587974,
      "loss": 0.401218056678772,
      "loss_ce": 9.503310866421089e-05,
      "loss_iou": 0.16796875,
      "loss_num": 0.012939453125,
      "loss_xval": 0.400390625,
      "num_input_tokens_seen": 283259004,
      "step": 5056
    },
    {
      "epoch": 11.262806236080179,
      "grad_norm": 20.411800384521484,
      "learning_rate": 1e-06,
      "loss": 0.6591,
      "num_input_tokens_seen": 283318136,
      "step": 5057
    },
    {
      "epoch": 11.262806236080179,
      "loss": 0.8671286702156067,
      "loss_ce": 0.00018532315152697265,
      "loss_iou": 0.34375,
      "loss_num": 0.0361328125,
      "loss_xval": 0.8671875,
      "num_input_tokens_seen": 283318136,
      "step": 5057
    },
    {
      "epoch": 11.265033407572384,
      "grad_norm": 15.510054588317871,
      "learning_rate": 1e-06,
      "loss": 0.5234,
      "num_input_tokens_seen": 283376952,
      "step": 5058
    },
    {
      "epoch": 11.265033407572384,
      "loss": 0.48036888241767883,
      "loss_ce": 0.00014426674169953912,
      "loss_iou": 0.2109375,
      "loss_num": 0.01165771484375,
      "loss_xval": 0.48046875,
      "num_input_tokens_seen": 283376952,
      "step": 5058
    },
    {
      "epoch": 11.267260579064589,
      "grad_norm": 16.76605224609375,
      "learning_rate": 1e-06,
      "loss": 0.5593,
      "num_input_tokens_seen": 283432408,
      "step": 5059
    },
    {
      "epoch": 11.267260579064589,
      "loss": 0.5044692754745483,
      "loss_ce": 0.000135770023916848,
      "loss_iou": 0.220703125,
      "loss_num": 0.012451171875,
      "loss_xval": 0.50390625,
      "num_input_tokens_seen": 283432408,
      "step": 5059
    },
    {
      "epoch": 11.269487750556793,
      "grad_norm": 21.036209106445312,
      "learning_rate": 1e-06,
      "loss": 0.5646,
      "num_input_tokens_seen": 283490212,
      "step": 5060
    },
    {
      "epoch": 11.269487750556793,
      "loss": 0.7949022054672241,
      "loss_ce": 0.00022446672664955258,
      "loss_iou": 0.337890625,
      "loss_num": 0.0238037109375,
      "loss_xval": 0.79296875,
      "num_input_tokens_seen": 283490212,
      "step": 5060
    },
    {
      "epoch": 11.271714922048998,
      "grad_norm": 14.880892753601074,
      "learning_rate": 1e-06,
      "loss": 0.5155,
      "num_input_tokens_seen": 283547208,
      "step": 5061
    },
    {
      "epoch": 11.271714922048998,
      "loss": 0.34775838255882263,
      "loss_ce": 0.00010212791676167399,
      "loss_iou": 0.126953125,
      "loss_num": 0.0186767578125,
      "loss_xval": 0.34765625,
      "num_input_tokens_seen": 283547208,
      "step": 5061
    },
    {
      "epoch": 11.273942093541203,
      "grad_norm": 11.523423194885254,
      "learning_rate": 1e-06,
      "loss": 0.4296,
      "num_input_tokens_seen": 283605480,
      "step": 5062
    },
    {
      "epoch": 11.273942093541203,
      "loss": 0.2592126131057739,
      "loss_ce": 0.00011838733917102218,
      "loss_iou": 0.103515625,
      "loss_num": 0.01043701171875,
      "loss_xval": 0.259765625,
      "num_input_tokens_seen": 283605480,
      "step": 5062
    },
    {
      "epoch": 11.276169265033408,
      "grad_norm": 20.444461822509766,
      "learning_rate": 1e-06,
      "loss": 0.4135,
      "num_input_tokens_seen": 283660960,
      "step": 5063
    },
    {
      "epoch": 11.276169265033408,
      "loss": 0.3837610185146332,
      "loss_ce": 0.0001245247694896534,
      "loss_iou": 0.1611328125,
      "loss_num": 0.01220703125,
      "loss_xval": 0.3828125,
      "num_input_tokens_seen": 283660960,
      "step": 5063
    },
    {
      "epoch": 11.278396436525613,
      "grad_norm": 16.973899841308594,
      "learning_rate": 1e-06,
      "loss": 0.5658,
      "num_input_tokens_seen": 283716348,
      "step": 5064
    },
    {
      "epoch": 11.278396436525613,
      "loss": 0.47617107629776,
      "loss_ce": 0.000585106376092881,
      "loss_iou": 0.216796875,
      "loss_num": 0.00848388671875,
      "loss_xval": 0.4765625,
      "num_input_tokens_seen": 283716348,
      "step": 5064
    },
    {
      "epoch": 11.280623608017818,
      "grad_norm": 15.647051811218262,
      "learning_rate": 1e-06,
      "loss": 0.5102,
      "num_input_tokens_seen": 283775620,
      "step": 5065
    },
    {
      "epoch": 11.280623608017818,
      "loss": 0.4950053095817566,
      "loss_ce": 0.00019331733346916735,
      "loss_iou": 0.20703125,
      "loss_num": 0.0162353515625,
      "loss_xval": 0.494140625,
      "num_input_tokens_seen": 283775620,
      "step": 5065
    },
    {
      "epoch": 11.282850779510023,
      "grad_norm": 14.765268325805664,
      "learning_rate": 1e-06,
      "loss": 0.4677,
      "num_input_tokens_seen": 283834152,
      "step": 5066
    },
    {
      "epoch": 11.282850779510023,
      "loss": 0.4379033148288727,
      "loss_ce": 0.00012866513861808926,
      "loss_iou": 0.193359375,
      "loss_num": 0.01007080078125,
      "loss_xval": 0.4375,
      "num_input_tokens_seen": 283834152,
      "step": 5066
    },
    {
      "epoch": 11.285077951002227,
      "grad_norm": 18.143211364746094,
      "learning_rate": 1e-06,
      "loss": 0.4485,
      "num_input_tokens_seen": 283891084,
      "step": 5067
    },
    {
      "epoch": 11.285077951002227,
      "loss": 0.4209212064743042,
      "loss_ce": 0.00014482939150184393,
      "loss_iou": 0.185546875,
      "loss_num": 0.010009765625,
      "loss_xval": 0.419921875,
      "num_input_tokens_seen": 283891084,
      "step": 5067
    },
    {
      "epoch": 11.287305122494432,
      "grad_norm": 14.855110168457031,
      "learning_rate": 1e-06,
      "loss": 0.3737,
      "num_input_tokens_seen": 283949732,
      "step": 5068
    },
    {
      "epoch": 11.287305122494432,
      "loss": 0.3620651960372925,
      "loss_ce": 0.0001267299521714449,
      "loss_iou": 0.166015625,
      "loss_num": 0.005828857421875,
      "loss_xval": 0.361328125,
      "num_input_tokens_seen": 283949732,
      "step": 5068
    },
    {
      "epoch": 11.289532293986637,
      "grad_norm": 43.02797317504883,
      "learning_rate": 1e-06,
      "loss": 0.5399,
      "num_input_tokens_seen": 284006896,
      "step": 5069
    },
    {
      "epoch": 11.289532293986637,
      "loss": 0.5649739503860474,
      "loss_ce": 0.00015458805137313902,
      "loss_iou": 0.23828125,
      "loss_num": 0.0174560546875,
      "loss_xval": 0.56640625,
      "num_input_tokens_seen": 284006896,
      "step": 5069
    },
    {
      "epoch": 11.291759465478842,
      "grad_norm": 25.057252883911133,
      "learning_rate": 1e-06,
      "loss": 0.5187,
      "num_input_tokens_seen": 284063816,
      "step": 5070
    },
    {
      "epoch": 11.291759465478842,
      "loss": 0.6080396175384521,
      "loss_ce": 0.0001294128887820989,
      "loss_iou": 0.287109375,
      "loss_num": 0.007080078125,
      "loss_xval": 0.609375,
      "num_input_tokens_seen": 284063816,
      "step": 5070
    },
    {
      "epoch": 11.293986636971047,
      "grad_norm": 18.99828338623047,
      "learning_rate": 1e-06,
      "loss": 0.4259,
      "num_input_tokens_seen": 284120160,
      "step": 5071
    },
    {
      "epoch": 11.293986636971047,
      "loss": 0.5792200565338135,
      "loss_ce": 0.00011846120469272137,
      "loss_iou": 0.26953125,
      "loss_num": 0.00787353515625,
      "loss_xval": 0.578125,
      "num_input_tokens_seen": 284120160,
      "step": 5071
    },
    {
      "epoch": 11.296213808463252,
      "grad_norm": 11.530062675476074,
      "learning_rate": 1e-06,
      "loss": 0.4847,
      "num_input_tokens_seen": 284178348,
      "step": 5072
    },
    {
      "epoch": 11.296213808463252,
      "loss": 0.4051365852355957,
      "loss_ce": 0.0001073086605174467,
      "loss_iou": 0.1787109375,
      "loss_num": 0.00970458984375,
      "loss_xval": 0.404296875,
      "num_input_tokens_seen": 284178348,
      "step": 5072
    },
    {
      "epoch": 11.298440979955457,
      "grad_norm": 30.209646224975586,
      "learning_rate": 1e-06,
      "loss": 0.512,
      "num_input_tokens_seen": 284237160,
      "step": 5073
    },
    {
      "epoch": 11.298440979955457,
      "loss": 0.583868682384491,
      "loss_ce": 0.00012847641482949257,
      "loss_iou": 0.248046875,
      "loss_num": 0.017333984375,
      "loss_xval": 0.58203125,
      "num_input_tokens_seen": 284237160,
      "step": 5073
    },
    {
      "epoch": 11.300668151447661,
      "grad_norm": 23.079030990600586,
      "learning_rate": 1e-06,
      "loss": 0.4586,
      "num_input_tokens_seen": 284293580,
      "step": 5074
    },
    {
      "epoch": 11.300668151447661,
      "loss": 0.5660747289657593,
      "loss_ce": 0.00015678048657719046,
      "loss_iou": 0.25,
      "loss_num": 0.01300048828125,
      "loss_xval": 0.56640625,
      "num_input_tokens_seen": 284293580,
      "step": 5074
    },
    {
      "epoch": 11.302895322939866,
      "grad_norm": 18.702524185180664,
      "learning_rate": 1e-06,
      "loss": 0.4264,
      "num_input_tokens_seen": 284350604,
      "step": 5075
    },
    {
      "epoch": 11.302895322939866,
      "loss": 0.38073596358299255,
      "loss_ce": 0.00012072990648448467,
      "loss_iou": 0.1630859375,
      "loss_num": 0.01092529296875,
      "loss_xval": 0.380859375,
      "num_input_tokens_seen": 284350604,
      "step": 5075
    },
    {
      "epoch": 11.305122494432071,
      "grad_norm": 72.22659301757812,
      "learning_rate": 1e-06,
      "loss": 0.7127,
      "num_input_tokens_seen": 284408168,
      "step": 5076
    },
    {
      "epoch": 11.305122494432071,
      "loss": 0.8336408138275146,
      "loss_ce": 0.00014475997886620462,
      "loss_iou": 0.349609375,
      "loss_num": 0.0269775390625,
      "loss_xval": 0.83203125,
      "num_input_tokens_seen": 284408168,
      "step": 5076
    },
    {
      "epoch": 11.307349665924276,
      "grad_norm": 27.653568267822266,
      "learning_rate": 1e-06,
      "loss": 0.4948,
      "num_input_tokens_seen": 284465572,
      "step": 5077
    },
    {
      "epoch": 11.307349665924276,
      "loss": 0.4449566602706909,
      "loss_ce": 0.0001324501063209027,
      "loss_iou": 0.197265625,
      "loss_num": 0.00994873046875,
      "loss_xval": 0.4453125,
      "num_input_tokens_seen": 284465572,
      "step": 5077
    },
    {
      "epoch": 11.309576837416481,
      "grad_norm": 29.11145782470703,
      "learning_rate": 1e-06,
      "loss": 0.4789,
      "num_input_tokens_seen": 284519732,
      "step": 5078
    },
    {
      "epoch": 11.309576837416481,
      "loss": 0.3955211043357849,
      "loss_ce": 0.0001353362895315513,
      "loss_iou": 0.162109375,
      "loss_num": 0.01416015625,
      "loss_xval": 0.39453125,
      "num_input_tokens_seen": 284519732,
      "step": 5078
    },
    {
      "epoch": 11.311804008908686,
      "grad_norm": 14.795896530151367,
      "learning_rate": 1e-06,
      "loss": 0.5233,
      "num_input_tokens_seen": 284578024,
      "step": 5079
    },
    {
      "epoch": 11.311804008908686,
      "loss": 0.3413148522377014,
      "loss_ce": 0.00012834850349463522,
      "loss_iou": 0.1591796875,
      "loss_num": 0.00457763671875,
      "loss_xval": 0.341796875,
      "num_input_tokens_seen": 284578024,
      "step": 5079
    },
    {
      "epoch": 11.31403118040089,
      "grad_norm": 20.22231101989746,
      "learning_rate": 1e-06,
      "loss": 0.3846,
      "num_input_tokens_seen": 284636628,
      "step": 5080
    },
    {
      "epoch": 11.31403118040089,
      "loss": 0.40492209792137146,
      "loss_ce": 0.0001369206584058702,
      "loss_iou": 0.1875,
      "loss_num": 0.006103515625,
      "loss_xval": 0.404296875,
      "num_input_tokens_seen": 284636628,
      "step": 5080
    },
    {
      "epoch": 11.316258351893095,
      "grad_norm": 14.39472484588623,
      "learning_rate": 1e-06,
      "loss": 0.5526,
      "num_input_tokens_seen": 284694192,
      "step": 5081
    },
    {
      "epoch": 11.316258351893095,
      "loss": 0.5521363019943237,
      "loss_ce": 0.00013434255379252136,
      "loss_iou": 0.236328125,
      "loss_num": 0.0159912109375,
      "loss_xval": 0.55078125,
      "num_input_tokens_seen": 284694192,
      "step": 5081
    },
    {
      "epoch": 11.3184855233853,
      "grad_norm": 15.9749755859375,
      "learning_rate": 1e-06,
      "loss": 0.4355,
      "num_input_tokens_seen": 284752428,
      "step": 5082
    },
    {
      "epoch": 11.3184855233853,
      "loss": 0.5515244603157043,
      "loss_ce": 0.00013286221656017005,
      "loss_iou": 0.236328125,
      "loss_num": 0.0157470703125,
      "loss_xval": 0.55078125,
      "num_input_tokens_seen": 284752428,
      "step": 5082
    },
    {
      "epoch": 11.320712694877505,
      "grad_norm": 17.713111877441406,
      "learning_rate": 1e-06,
      "loss": 0.4379,
      "num_input_tokens_seen": 284806252,
      "step": 5083
    },
    {
      "epoch": 11.320712694877505,
      "loss": 0.44035178422927856,
      "loss_ce": 0.0001662498980294913,
      "loss_iou": 0.2001953125,
      "loss_num": 0.008056640625,
      "loss_xval": 0.439453125,
      "num_input_tokens_seen": 284806252,
      "step": 5083
    },
    {
      "epoch": 11.32293986636971,
      "grad_norm": 16.1520938873291,
      "learning_rate": 1e-06,
      "loss": 0.5477,
      "num_input_tokens_seen": 284863332,
      "step": 5084
    },
    {
      "epoch": 11.32293986636971,
      "loss": 0.6391512155532837,
      "loss_ce": 0.000357240904122591,
      "loss_iou": 0.26953125,
      "loss_num": 0.020263671875,
      "loss_xval": 0.640625,
      "num_input_tokens_seen": 284863332,
      "step": 5084
    },
    {
      "epoch": 11.325167037861915,
      "grad_norm": 19.611209869384766,
      "learning_rate": 1e-06,
      "loss": 0.4462,
      "num_input_tokens_seen": 284919672,
      "step": 5085
    },
    {
      "epoch": 11.325167037861915,
      "loss": 0.37651515007019043,
      "loss_ce": 0.00017235177801921964,
      "loss_iou": 0.1650390625,
      "loss_num": 0.0093994140625,
      "loss_xval": 0.376953125,
      "num_input_tokens_seen": 284919672,
      "step": 5085
    },
    {
      "epoch": 11.32739420935412,
      "grad_norm": 17.526779174804688,
      "learning_rate": 1e-06,
      "loss": 0.6155,
      "num_input_tokens_seen": 284977908,
      "step": 5086
    },
    {
      "epoch": 11.32739420935412,
      "loss": 0.48550522327423096,
      "loss_ce": 0.00015366033767350018,
      "loss_iou": 0.220703125,
      "loss_num": 0.00885009765625,
      "loss_xval": 0.484375,
      "num_input_tokens_seen": 284977908,
      "step": 5086
    },
    {
      "epoch": 11.329621380846325,
      "grad_norm": 14.409747123718262,
      "learning_rate": 1e-06,
      "loss": 0.4557,
      "num_input_tokens_seen": 285033440,
      "step": 5087
    },
    {
      "epoch": 11.329621380846325,
      "loss": 0.44422948360443115,
      "loss_ce": 0.00013769487850368023,
      "loss_iou": 0.177734375,
      "loss_num": 0.0179443359375,
      "loss_xval": 0.443359375,
      "num_input_tokens_seen": 285033440,
      "step": 5087
    },
    {
      "epoch": 11.33184855233853,
      "grad_norm": 30.620845794677734,
      "learning_rate": 1e-06,
      "loss": 0.55,
      "num_input_tokens_seen": 285089588,
      "step": 5088
    },
    {
      "epoch": 11.33184855233853,
      "loss": 0.440792441368103,
      "loss_ce": 0.00011861581879202276,
      "loss_iou": 0.1865234375,
      "loss_num": 0.013427734375,
      "loss_xval": 0.44140625,
      "num_input_tokens_seen": 285089588,
      "step": 5088
    },
    {
      "epoch": 11.334075723830734,
      "grad_norm": 21.544681549072266,
      "learning_rate": 1e-06,
      "loss": 0.496,
      "num_input_tokens_seen": 285144700,
      "step": 5089
    },
    {
      "epoch": 11.334075723830734,
      "loss": 0.5140707492828369,
      "loss_ce": 0.00015477146371267736,
      "loss_iou": 0.2255859375,
      "loss_num": 0.01251220703125,
      "loss_xval": 0.515625,
      "num_input_tokens_seen": 285144700,
      "step": 5089
    },
    {
      "epoch": 11.33630289532294,
      "grad_norm": 26.6117000579834,
      "learning_rate": 1e-06,
      "loss": 0.5348,
      "num_input_tokens_seen": 285200876,
      "step": 5090
    },
    {
      "epoch": 11.33630289532294,
      "loss": 0.5355468392372131,
      "loss_ce": 0.00014644389739260077,
      "loss_iou": 0.2431640625,
      "loss_num": 0.0098876953125,
      "loss_xval": 0.53515625,
      "num_input_tokens_seen": 285200876,
      "step": 5090
    },
    {
      "epoch": 11.338530066815144,
      "grad_norm": 17.32594871520996,
      "learning_rate": 1e-06,
      "loss": 0.6661,
      "num_input_tokens_seen": 285258992,
      "step": 5091
    },
    {
      "epoch": 11.338530066815144,
      "loss": 0.8300636410713196,
      "loss_ce": 0.00022966302640270442,
      "loss_iou": 0.33203125,
      "loss_num": 0.032958984375,
      "loss_xval": 0.828125,
      "num_input_tokens_seen": 285258992,
      "step": 5091
    },
    {
      "epoch": 11.340757238307349,
      "grad_norm": 26.523305892944336,
      "learning_rate": 1e-06,
      "loss": 0.6046,
      "num_input_tokens_seen": 285314720,
      "step": 5092
    },
    {
      "epoch": 11.340757238307349,
      "loss": 0.6371076107025146,
      "loss_ce": 0.0001447718241252005,
      "loss_iou": 0.2734375,
      "loss_num": 0.017822265625,
      "loss_xval": 0.63671875,
      "num_input_tokens_seen": 285314720,
      "step": 5092
    },
    {
      "epoch": 11.342984409799554,
      "grad_norm": 23.49717140197754,
      "learning_rate": 1e-06,
      "loss": 0.4436,
      "num_input_tokens_seen": 285372940,
      "step": 5093
    },
    {
      "epoch": 11.342984409799554,
      "loss": 0.5071090459823608,
      "loss_ce": 0.00015102185716386884,
      "loss_iou": 0.220703125,
      "loss_num": 0.01318359375,
      "loss_xval": 0.5078125,
      "num_input_tokens_seen": 285372940,
      "step": 5093
    },
    {
      "epoch": 11.345211581291759,
      "grad_norm": 21.801851272583008,
      "learning_rate": 1e-06,
      "loss": 0.5209,
      "num_input_tokens_seen": 285429172,
      "step": 5094
    },
    {
      "epoch": 11.345211581291759,
      "loss": 0.526289701461792,
      "loss_ce": 0.00016666974988766015,
      "loss_iou": 0.2255859375,
      "loss_num": 0.01507568359375,
      "loss_xval": 0.52734375,
      "num_input_tokens_seen": 285429172,
      "step": 5094
    },
    {
      "epoch": 11.347438752783964,
      "grad_norm": 16.54652214050293,
      "learning_rate": 1e-06,
      "loss": 0.6023,
      "num_input_tokens_seen": 285484976,
      "step": 5095
    },
    {
      "epoch": 11.347438752783964,
      "loss": 0.5767943263053894,
      "loss_ce": 0.00013418751768767834,
      "loss_iou": 0.25390625,
      "loss_num": 0.01318359375,
      "loss_xval": 0.578125,
      "num_input_tokens_seen": 285484976,
      "step": 5095
    },
    {
      "epoch": 11.34966592427617,
      "grad_norm": 21.48624610900879,
      "learning_rate": 1e-06,
      "loss": 0.5497,
      "num_input_tokens_seen": 285540752,
      "step": 5096
    },
    {
      "epoch": 11.34966592427617,
      "loss": 0.6203560829162598,
      "loss_ce": 0.00023887879797257483,
      "loss_iou": 0.2578125,
      "loss_num": 0.0211181640625,
      "loss_xval": 0.62109375,
      "num_input_tokens_seen": 285540752,
      "step": 5096
    },
    {
      "epoch": 11.351893095768375,
      "grad_norm": 24.3822021484375,
      "learning_rate": 1e-06,
      "loss": 0.553,
      "num_input_tokens_seen": 285594876,
      "step": 5097
    },
    {
      "epoch": 11.351893095768375,
      "loss": 0.4341278076171875,
      "loss_ce": 0.00010682163701858371,
      "loss_iou": 0.1748046875,
      "loss_num": 0.0166015625,
      "loss_xval": 0.43359375,
      "num_input_tokens_seen": 285594876,
      "step": 5097
    },
    {
      "epoch": 11.35412026726058,
      "grad_norm": 16.19316864013672,
      "learning_rate": 1e-06,
      "loss": 0.5853,
      "num_input_tokens_seen": 285650340,
      "step": 5098
    },
    {
      "epoch": 11.35412026726058,
      "loss": 0.7105726003646851,
      "loss_ce": 0.00012335649807937443,
      "loss_iou": 0.306640625,
      "loss_num": 0.019287109375,
      "loss_xval": 0.7109375,
      "num_input_tokens_seen": 285650340,
      "step": 5098
    },
    {
      "epoch": 11.356347438752785,
      "grad_norm": 17.640634536743164,
      "learning_rate": 1e-06,
      "loss": 0.516,
      "num_input_tokens_seen": 285706040,
      "step": 5099
    },
    {
      "epoch": 11.356347438752785,
      "loss": 0.6544548273086548,
      "loss_ce": 0.0001579629024490714,
      "loss_iou": 0.287109375,
      "loss_num": 0.016357421875,
      "loss_xval": 0.65625,
      "num_input_tokens_seen": 285706040,
      "step": 5099
    },
    {
      "epoch": 11.35857461024499,
      "grad_norm": 22.37962532043457,
      "learning_rate": 1e-06,
      "loss": 0.7185,
      "num_input_tokens_seen": 285762072,
      "step": 5100
    },
    {
      "epoch": 11.35857461024499,
      "loss": 0.8059933185577393,
      "loss_ce": 0.00020718795713037252,
      "loss_iou": 0.306640625,
      "loss_num": 0.038330078125,
      "loss_xval": 0.8046875,
      "num_input_tokens_seen": 285762072,
      "step": 5100
    },
    {
      "epoch": 11.360801781737194,
      "grad_norm": 14.851436614990234,
      "learning_rate": 1e-06,
      "loss": 0.4088,
      "num_input_tokens_seen": 285818724,
      "step": 5101
    },
    {
      "epoch": 11.360801781737194,
      "loss": 0.40162622928619385,
      "loss_ce": 0.0001369684759993106,
      "loss_iou": 0.1806640625,
      "loss_num": 0.0081787109375,
      "loss_xval": 0.40234375,
      "num_input_tokens_seen": 285818724,
      "step": 5101
    },
    {
      "epoch": 11.3630289532294,
      "grad_norm": 16.314359664916992,
      "learning_rate": 1e-06,
      "loss": 0.4846,
      "num_input_tokens_seen": 285875484,
      "step": 5102
    },
    {
      "epoch": 11.3630289532294,
      "loss": 0.46129605174064636,
      "loss_ce": 0.00011440047819633037,
      "loss_iou": 0.208984375,
      "loss_num": 0.0084228515625,
      "loss_xval": 0.4609375,
      "num_input_tokens_seen": 285875484,
      "step": 5102
    },
    {
      "epoch": 11.365256124721604,
      "grad_norm": 23.65479278564453,
      "learning_rate": 1e-06,
      "loss": 0.5254,
      "num_input_tokens_seen": 285931436,
      "step": 5103
    },
    {
      "epoch": 11.365256124721604,
      "loss": 0.6119776964187622,
      "loss_ce": 0.00016128391143865883,
      "loss_iou": 0.267578125,
      "loss_num": 0.015380859375,
      "loss_xval": 0.61328125,
      "num_input_tokens_seen": 285931436,
      "step": 5103
    },
    {
      "epoch": 11.367483296213809,
      "grad_norm": 19.001941680908203,
      "learning_rate": 1e-06,
      "loss": 0.522,
      "num_input_tokens_seen": 285987400,
      "step": 5104
    },
    {
      "epoch": 11.367483296213809,
      "loss": 0.6559271216392517,
      "loss_ce": 0.0001653625804465264,
      "loss_iou": 0.28125,
      "loss_num": 0.0184326171875,
      "loss_xval": 0.65625,
      "num_input_tokens_seen": 285987400,
      "step": 5104
    },
    {
      "epoch": 11.369710467706014,
      "grad_norm": 22.59067153930664,
      "learning_rate": 1e-06,
      "loss": 0.5637,
      "num_input_tokens_seen": 286040144,
      "step": 5105
    },
    {
      "epoch": 11.369710467706014,
      "loss": 0.5408531427383423,
      "loss_ce": 0.00032578702666796744,
      "loss_iou": 0.2470703125,
      "loss_num": 0.009033203125,
      "loss_xval": 0.5390625,
      "num_input_tokens_seen": 286040144,
      "step": 5105
    },
    {
      "epoch": 11.371937639198219,
      "grad_norm": 13.022751808166504,
      "learning_rate": 1e-06,
      "loss": 0.4658,
      "num_input_tokens_seen": 286095652,
      "step": 5106
    },
    {
      "epoch": 11.371937639198219,
      "loss": 0.49904271960258484,
      "loss_ce": 0.0001413495047017932,
      "loss_iou": 0.2099609375,
      "loss_num": 0.015869140625,
      "loss_xval": 0.498046875,
      "num_input_tokens_seen": 286095652,
      "step": 5106
    },
    {
      "epoch": 11.374164810690424,
      "grad_norm": 17.726234436035156,
      "learning_rate": 1e-06,
      "loss": 0.4394,
      "num_input_tokens_seen": 286153252,
      "step": 5107
    },
    {
      "epoch": 11.374164810690424,
      "loss": 0.4727289080619812,
      "loss_ce": 0.00019473947759252042,
      "loss_iou": 0.189453125,
      "loss_num": 0.0186767578125,
      "loss_xval": 0.47265625,
      "num_input_tokens_seen": 286153252,
      "step": 5107
    },
    {
      "epoch": 11.376391982182628,
      "grad_norm": 20.799928665161133,
      "learning_rate": 1e-06,
      "loss": 0.5046,
      "num_input_tokens_seen": 286210432,
      "step": 5108
    },
    {
      "epoch": 11.376391982182628,
      "loss": 0.666250467300415,
      "loss_ce": 0.00023488188162446022,
      "loss_iou": 0.26953125,
      "loss_num": 0.02490234375,
      "loss_xval": 0.6640625,
      "num_input_tokens_seen": 286210432,
      "step": 5108
    },
    {
      "epoch": 11.378619153674833,
      "grad_norm": 21.947765350341797,
      "learning_rate": 1e-06,
      "loss": 0.4669,
      "num_input_tokens_seen": 286266940,
      "step": 5109
    },
    {
      "epoch": 11.378619153674833,
      "loss": 0.5005042552947998,
      "loss_ce": 0.00013799341104459018,
      "loss_iou": 0.22265625,
      "loss_num": 0.01129150390625,
      "loss_xval": 0.5,
      "num_input_tokens_seen": 286266940,
      "step": 5109
    },
    {
      "epoch": 11.380846325167038,
      "grad_norm": 15.626119613647461,
      "learning_rate": 1e-06,
      "loss": 0.5814,
      "num_input_tokens_seen": 286321024,
      "step": 5110
    },
    {
      "epoch": 11.380846325167038,
      "loss": 0.5123552680015564,
      "loss_ce": 0.0001482515363022685,
      "loss_iou": 0.228515625,
      "loss_num": 0.01129150390625,
      "loss_xval": 0.51171875,
      "num_input_tokens_seen": 286321024,
      "step": 5110
    },
    {
      "epoch": 11.383073496659243,
      "grad_norm": 20.23106575012207,
      "learning_rate": 1e-06,
      "loss": 0.4268,
      "num_input_tokens_seen": 286376468,
      "step": 5111
    },
    {
      "epoch": 11.383073496659243,
      "loss": 0.46500349044799805,
      "loss_ce": 0.00015975304995663464,
      "loss_iou": 0.21875,
      "loss_num": 0.005584716796875,
      "loss_xval": 0.46484375,
      "num_input_tokens_seen": 286376468,
      "step": 5111
    },
    {
      "epoch": 11.385300668151448,
      "grad_norm": 16.623905181884766,
      "learning_rate": 1e-06,
      "loss": 0.5151,
      "num_input_tokens_seen": 286429412,
      "step": 5112
    },
    {
      "epoch": 11.385300668151448,
      "loss": 0.46804624795913696,
      "loss_ce": 0.00015074793191161007,
      "loss_iou": 0.181640625,
      "loss_num": 0.0206298828125,
      "loss_xval": 0.46875,
      "num_input_tokens_seen": 286429412,
      "step": 5112
    },
    {
      "epoch": 11.387527839643653,
      "grad_norm": 20.689804077148438,
      "learning_rate": 1e-06,
      "loss": 0.3954,
      "num_input_tokens_seen": 286486412,
      "step": 5113
    },
    {
      "epoch": 11.387527839643653,
      "loss": 0.3609923720359802,
      "loss_ce": 0.00015251885633915663,
      "loss_iou": 0.1640625,
      "loss_num": 0.006500244140625,
      "loss_xval": 0.361328125,
      "num_input_tokens_seen": 286486412,
      "step": 5113
    },
    {
      "epoch": 11.389755011135858,
      "grad_norm": 28.57643699645996,
      "learning_rate": 1e-06,
      "loss": 0.5011,
      "num_input_tokens_seen": 286541332,
      "step": 5114
    },
    {
      "epoch": 11.389755011135858,
      "loss": 0.4327373206615448,
      "loss_ce": 0.00012012844672426581,
      "loss_iou": 0.1923828125,
      "loss_num": 0.009765625,
      "loss_xval": 0.43359375,
      "num_input_tokens_seen": 286541332,
      "step": 5114
    },
    {
      "epoch": 11.391982182628063,
      "grad_norm": 15.638270378112793,
      "learning_rate": 1e-06,
      "loss": 0.5359,
      "num_input_tokens_seen": 286599148,
      "step": 5115
    },
    {
      "epoch": 11.391982182628063,
      "loss": 0.4875502586364746,
      "loss_ce": 0.0001235070376424119,
      "loss_iou": 0.2119140625,
      "loss_num": 0.01263427734375,
      "loss_xval": 0.48828125,
      "num_input_tokens_seen": 286599148,
      "step": 5115
    },
    {
      "epoch": 11.394209354120267,
      "grad_norm": 22.461973190307617,
      "learning_rate": 1e-06,
      "loss": 0.5282,
      "num_input_tokens_seen": 286656460,
      "step": 5116
    },
    {
      "epoch": 11.394209354120267,
      "loss": 0.7095192074775696,
      "loss_ce": 0.00016860665346030146,
      "loss_iou": 0.28515625,
      "loss_num": 0.0277099609375,
      "loss_xval": 0.7109375,
      "num_input_tokens_seen": 286656460,
      "step": 5116
    },
    {
      "epoch": 11.396436525612472,
      "grad_norm": 18.777915954589844,
      "learning_rate": 1e-06,
      "loss": 0.5819,
      "num_input_tokens_seen": 286711268,
      "step": 5117
    },
    {
      "epoch": 11.396436525612472,
      "loss": 0.5128494501113892,
      "loss_ce": 0.00015413996879942715,
      "loss_iou": 0.2158203125,
      "loss_num": 0.0159912109375,
      "loss_xval": 0.51171875,
      "num_input_tokens_seen": 286711268,
      "step": 5117
    },
    {
      "epoch": 11.398663697104677,
      "grad_norm": 18.205013275146484,
      "learning_rate": 1e-06,
      "loss": 0.4723,
      "num_input_tokens_seen": 286767632,
      "step": 5118
    },
    {
      "epoch": 11.398663697104677,
      "loss": 0.44845569133758545,
      "loss_ce": 0.0001524692343082279,
      "loss_iou": 0.2060546875,
      "loss_num": 0.007232666015625,
      "loss_xval": 0.44921875,
      "num_input_tokens_seen": 286767632,
      "step": 5118
    },
    {
      "epoch": 11.400890868596882,
      "grad_norm": 14.385238647460938,
      "learning_rate": 1e-06,
      "loss": 0.5913,
      "num_input_tokens_seen": 286824064,
      "step": 5119
    },
    {
      "epoch": 11.400890868596882,
      "loss": 0.6861345767974854,
      "loss_ce": 9.943717304849997e-05,
      "loss_iou": 0.310546875,
      "loss_num": 0.013427734375,
      "loss_xval": 0.6875,
      "num_input_tokens_seen": 286824064,
      "step": 5119
    },
    {
      "epoch": 11.403118040089087,
      "grad_norm": 15.494654655456543,
      "learning_rate": 1e-06,
      "loss": 0.3787,
      "num_input_tokens_seen": 286881632,
      "step": 5120
    },
    {
      "epoch": 11.403118040089087,
      "loss": 0.36211222410202026,
      "loss_ce": 0.00011270066897850484,
      "loss_iou": 0.158203125,
      "loss_num": 0.0089111328125,
      "loss_xval": 0.361328125,
      "num_input_tokens_seen": 286881632,
      "step": 5120
    },
    {
      "epoch": 11.405345211581292,
      "grad_norm": 17.95199203491211,
      "learning_rate": 1e-06,
      "loss": 0.711,
      "num_input_tokens_seen": 286934124,
      "step": 5121
    },
    {
      "epoch": 11.405345211581292,
      "loss": 0.7193175554275513,
      "loss_ce": 0.00020134558144491166,
      "loss_iou": 0.31640625,
      "loss_num": 0.0172119140625,
      "loss_xval": 0.71875,
      "num_input_tokens_seen": 286934124,
      "step": 5121
    },
    {
      "epoch": 11.407572383073497,
      "grad_norm": 28.730737686157227,
      "learning_rate": 1e-06,
      "loss": 0.5827,
      "num_input_tokens_seen": 286990012,
      "step": 5122
    },
    {
      "epoch": 11.407572383073497,
      "loss": 0.39236217737197876,
      "loss_ce": 0.00019602719112299383,
      "loss_iou": 0.177734375,
      "loss_num": 0.00750732421875,
      "loss_xval": 0.392578125,
      "num_input_tokens_seen": 286990012,
      "step": 5122
    },
    {
      "epoch": 11.409799554565701,
      "grad_norm": 39.97563171386719,
      "learning_rate": 1e-06,
      "loss": 0.6908,
      "num_input_tokens_seen": 287044416,
      "step": 5123
    },
    {
      "epoch": 11.409799554565701,
      "loss": 0.6997367739677429,
      "loss_ce": 0.00015180771879386157,
      "loss_iou": 0.306640625,
      "loss_num": 0.016845703125,
      "loss_xval": 0.69921875,
      "num_input_tokens_seen": 287044416,
      "step": 5123
    },
    {
      "epoch": 11.412026726057906,
      "grad_norm": 21.006521224975586,
      "learning_rate": 1e-06,
      "loss": 0.5632,
      "num_input_tokens_seen": 287100240,
      "step": 5124
    },
    {
      "epoch": 11.412026726057906,
      "loss": 0.6319659352302551,
      "loss_ce": 0.00012999521277379245,
      "loss_iou": 0.279296875,
      "loss_num": 0.01458740234375,
      "loss_xval": 0.6328125,
      "num_input_tokens_seen": 287100240,
      "step": 5124
    },
    {
      "epoch": 11.414253897550111,
      "grad_norm": 17.86157989501953,
      "learning_rate": 1e-06,
      "loss": 0.5612,
      "num_input_tokens_seen": 287157868,
      "step": 5125
    },
    {
      "epoch": 11.414253897550111,
      "loss": 0.3876880407333374,
      "loss_ce": 0.00011479659588076174,
      "loss_iou": 0.162109375,
      "loss_num": 0.01263427734375,
      "loss_xval": 0.38671875,
      "num_input_tokens_seen": 287157868,
      "step": 5125
    },
    {
      "epoch": 11.416481069042316,
      "grad_norm": 16.256969451904297,
      "learning_rate": 1e-06,
      "loss": 0.6867,
      "num_input_tokens_seen": 287212888,
      "step": 5126
    },
    {
      "epoch": 11.416481069042316,
      "loss": 0.6945188045501709,
      "loss_ce": 0.00012188311666250229,
      "loss_iou": 0.28515625,
      "loss_num": 0.025146484375,
      "loss_xval": 0.6953125,
      "num_input_tokens_seen": 287212888,
      "step": 5126
    },
    {
      "epoch": 11.41870824053452,
      "grad_norm": 36.22967529296875,
      "learning_rate": 1e-06,
      "loss": 0.6743,
      "num_input_tokens_seen": 287268684,
      "step": 5127
    },
    {
      "epoch": 11.41870824053452,
      "loss": 0.8071569204330444,
      "loss_ce": 0.00015009319758974016,
      "loss_iou": 0.375,
      "loss_num": 0.01123046875,
      "loss_xval": 0.80859375,
      "num_input_tokens_seen": 287268684,
      "step": 5127
    },
    {
      "epoch": 11.420935412026726,
      "grad_norm": 18.855321884155273,
      "learning_rate": 1e-06,
      "loss": 0.6559,
      "num_input_tokens_seen": 287326204,
      "step": 5128
    },
    {
      "epoch": 11.420935412026726,
      "loss": 0.6211212873458862,
      "loss_ce": 0.00014961831038817763,
      "loss_iou": 0.265625,
      "loss_num": 0.017578125,
      "loss_xval": 0.62109375,
      "num_input_tokens_seen": 287326204,
      "step": 5128
    },
    {
      "epoch": 11.42316258351893,
      "grad_norm": 13.376533508300781,
      "learning_rate": 1e-06,
      "loss": 0.5103,
      "num_input_tokens_seen": 287381800,
      "step": 5129
    },
    {
      "epoch": 11.42316258351893,
      "loss": 0.5299174189567566,
      "loss_ce": 0.0001627803430892527,
      "loss_iou": 0.2294921875,
      "loss_num": 0.0142822265625,
      "loss_xval": 0.53125,
      "num_input_tokens_seen": 287381800,
      "step": 5129
    },
    {
      "epoch": 11.425389755011135,
      "grad_norm": 22.959978103637695,
      "learning_rate": 1e-06,
      "loss": 0.5615,
      "num_input_tokens_seen": 287435956,
      "step": 5130
    },
    {
      "epoch": 11.425389755011135,
      "loss": 0.45396700501441956,
      "loss_ce": 0.0001095635088859126,
      "loss_iou": 0.1923828125,
      "loss_num": 0.013916015625,
      "loss_xval": 0.453125,
      "num_input_tokens_seen": 287435956,
      "step": 5130
    },
    {
      "epoch": 11.42761692650334,
      "grad_norm": 25.492908477783203,
      "learning_rate": 1e-06,
      "loss": 0.5793,
      "num_input_tokens_seen": 287491780,
      "step": 5131
    },
    {
      "epoch": 11.42761692650334,
      "loss": 0.6335743069648743,
      "loss_ce": 0.0001514378236606717,
      "loss_iou": 0.259765625,
      "loss_num": 0.0230712890625,
      "loss_xval": 0.6328125,
      "num_input_tokens_seen": 287491780,
      "step": 5131
    },
    {
      "epoch": 11.429844097995545,
      "grad_norm": 24.56211280822754,
      "learning_rate": 1e-06,
      "loss": 0.6357,
      "num_input_tokens_seen": 287548600,
      "step": 5132
    },
    {
      "epoch": 11.429844097995545,
      "loss": 0.28725236654281616,
      "loss_ce": 0.000387123815016821,
      "loss_iou": 0.1162109375,
      "loss_num": 0.0108642578125,
      "loss_xval": 0.287109375,
      "num_input_tokens_seen": 287548600,
      "step": 5132
    },
    {
      "epoch": 11.43207126948775,
      "grad_norm": 19.3801212310791,
      "learning_rate": 1e-06,
      "loss": 0.6795,
      "num_input_tokens_seen": 287603800,
      "step": 5133
    },
    {
      "epoch": 11.43207126948775,
      "loss": 0.7802882194519043,
      "loss_ce": 0.00013687220052815974,
      "loss_iou": 0.3203125,
      "loss_num": 0.0277099609375,
      "loss_xval": 0.78125,
      "num_input_tokens_seen": 287603800,
      "step": 5133
    },
    {
      "epoch": 11.434298440979955,
      "grad_norm": 32.62427520751953,
      "learning_rate": 1e-06,
      "loss": 0.658,
      "num_input_tokens_seen": 287658344,
      "step": 5134
    },
    {
      "epoch": 11.434298440979955,
      "loss": 0.539149820804596,
      "loss_ce": 0.0001483388477936387,
      "loss_iou": 0.232421875,
      "loss_num": 0.01513671875,
      "loss_xval": 0.5390625,
      "num_input_tokens_seen": 287658344,
      "step": 5134
    },
    {
      "epoch": 11.43652561247216,
      "grad_norm": 36.44560623168945,
      "learning_rate": 1e-06,
      "loss": 0.6056,
      "num_input_tokens_seen": 287713276,
      "step": 5135
    },
    {
      "epoch": 11.43652561247216,
      "loss": 0.49588677287101746,
      "loss_ce": 0.0001592589687788859,
      "loss_iou": 0.2265625,
      "loss_num": 0.00848388671875,
      "loss_xval": 0.49609375,
      "num_input_tokens_seen": 287713276,
      "step": 5135
    },
    {
      "epoch": 11.438752783964365,
      "grad_norm": 29.658565521240234,
      "learning_rate": 1e-06,
      "loss": 0.3693,
      "num_input_tokens_seen": 287770612,
      "step": 5136
    },
    {
      "epoch": 11.438752783964365,
      "loss": 0.3307318091392517,
      "loss_ce": 0.000165409262990579,
      "loss_iou": 0.1494140625,
      "loss_num": 0.0064697265625,
      "loss_xval": 0.330078125,
      "num_input_tokens_seen": 287770612,
      "step": 5136
    },
    {
      "epoch": 11.44097995545657,
      "grad_norm": 16.48049545288086,
      "learning_rate": 1e-06,
      "loss": 0.508,
      "num_input_tokens_seen": 287824620,
      "step": 5137
    },
    {
      "epoch": 11.44097995545657,
      "loss": 0.45837414264678955,
      "loss_ce": 0.00012218940537422895,
      "loss_iou": 0.1953125,
      "loss_num": 0.01336669921875,
      "loss_xval": 0.458984375,
      "num_input_tokens_seen": 287824620,
      "step": 5137
    },
    {
      "epoch": 11.443207126948774,
      "grad_norm": 22.14146614074707,
      "learning_rate": 1e-06,
      "loss": 0.5379,
      "num_input_tokens_seen": 287882580,
      "step": 5138
    },
    {
      "epoch": 11.443207126948774,
      "loss": 0.5207144021987915,
      "loss_ce": 0.0001455346355214715,
      "loss_iou": 0.216796875,
      "loss_num": 0.017333984375,
      "loss_xval": 0.51953125,
      "num_input_tokens_seen": 287882580,
      "step": 5138
    },
    {
      "epoch": 11.44543429844098,
      "grad_norm": 19.23894500732422,
      "learning_rate": 1e-06,
      "loss": 0.3896,
      "num_input_tokens_seen": 287938492,
      "step": 5139
    },
    {
      "epoch": 11.44543429844098,
      "loss": 0.3553389310836792,
      "loss_ce": 0.0001143506815424189,
      "loss_iou": 0.154296875,
      "loss_num": 0.0093994140625,
      "loss_xval": 0.35546875,
      "num_input_tokens_seen": 287938492,
      "step": 5139
    },
    {
      "epoch": 11.447661469933184,
      "grad_norm": 17.079357147216797,
      "learning_rate": 1e-06,
      "loss": 0.5968,
      "num_input_tokens_seen": 287994108,
      "step": 5140
    },
    {
      "epoch": 11.447661469933184,
      "loss": 0.5203468203544617,
      "loss_ce": 0.00014417944476008415,
      "loss_iou": 0.236328125,
      "loss_num": 0.009521484375,
      "loss_xval": 0.51953125,
      "num_input_tokens_seen": 287994108,
      "step": 5140
    },
    {
      "epoch": 11.449888641425389,
      "grad_norm": 14.164678573608398,
      "learning_rate": 1e-06,
      "loss": 0.5193,
      "num_input_tokens_seen": 288049672,
      "step": 5141
    },
    {
      "epoch": 11.449888641425389,
      "loss": 0.4837890863418579,
      "loss_ce": 0.0001465213717892766,
      "loss_iou": 0.2060546875,
      "loss_num": 0.014404296875,
      "loss_xval": 0.484375,
      "num_input_tokens_seen": 288049672,
      "step": 5141
    },
    {
      "epoch": 11.452115812917596,
      "grad_norm": 20.210201263427734,
      "learning_rate": 1e-06,
      "loss": 0.7254,
      "num_input_tokens_seen": 288106140,
      "step": 5142
    },
    {
      "epoch": 11.452115812917596,
      "loss": 0.6080299615859985,
      "loss_ce": 0.00011977371468674392,
      "loss_iou": 0.267578125,
      "loss_num": 0.01470947265625,
      "loss_xval": 0.609375,
      "num_input_tokens_seen": 288106140,
      "step": 5142
    },
    {
      "epoch": 11.4543429844098,
      "grad_norm": 20.572181701660156,
      "learning_rate": 1e-06,
      "loss": 0.4897,
      "num_input_tokens_seen": 288163140,
      "step": 5143
    },
    {
      "epoch": 11.4543429844098,
      "loss": 0.49498531222343445,
      "loss_ce": 0.00011225108028156683,
      "loss_iou": 0.220703125,
      "loss_num": 0.01068115234375,
      "loss_xval": 0.494140625,
      "num_input_tokens_seen": 288163140,
      "step": 5143
    },
    {
      "epoch": 11.456570155902005,
      "grad_norm": 24.588138580322266,
      "learning_rate": 1e-06,
      "loss": 0.5276,
      "num_input_tokens_seen": 288219856,
      "step": 5144
    },
    {
      "epoch": 11.456570155902005,
      "loss": 0.5592126846313477,
      "loss_ce": 0.0003747770097106695,
      "loss_iou": 0.240234375,
      "loss_num": 0.015625,
      "loss_xval": 0.55859375,
      "num_input_tokens_seen": 288219856,
      "step": 5144
    },
    {
      "epoch": 11.45879732739421,
      "grad_norm": 16.87946319580078,
      "learning_rate": 1e-06,
      "loss": 0.3826,
      "num_input_tokens_seen": 288276876,
      "step": 5145
    },
    {
      "epoch": 11.45879732739421,
      "loss": 0.4041239619255066,
      "loss_ce": 0.0008036848739720881,
      "loss_iou": 0.1689453125,
      "loss_num": 0.01318359375,
      "loss_xval": 0.40234375,
      "num_input_tokens_seen": 288276876,
      "step": 5145
    },
    {
      "epoch": 11.461024498886415,
      "grad_norm": 34.00703811645508,
      "learning_rate": 1e-06,
      "loss": 0.5651,
      "num_input_tokens_seen": 288334908,
      "step": 5146
    },
    {
      "epoch": 11.461024498886415,
      "loss": 0.4293323755264282,
      "loss_ce": 0.0001331909152213484,
      "loss_iou": 0.2001953125,
      "loss_num": 0.0059814453125,
      "loss_xval": 0.4296875,
      "num_input_tokens_seen": 288334908,
      "step": 5146
    },
    {
      "epoch": 11.46325167037862,
      "grad_norm": 21.397933959960938,
      "learning_rate": 1e-06,
      "loss": 0.4555,
      "num_input_tokens_seen": 288393052,
      "step": 5147
    },
    {
      "epoch": 11.46325167037862,
      "loss": 0.4838143587112427,
      "loss_ce": 0.00017176388064399362,
      "loss_iou": 0.203125,
      "loss_num": 0.015380859375,
      "loss_xval": 0.484375,
      "num_input_tokens_seen": 288393052,
      "step": 5147
    },
    {
      "epoch": 11.465478841870825,
      "grad_norm": 26.195497512817383,
      "learning_rate": 1e-06,
      "loss": 0.4912,
      "num_input_tokens_seen": 288445524,
      "step": 5148
    },
    {
      "epoch": 11.465478841870825,
      "loss": 0.5275430679321289,
      "loss_ce": 0.0008096360834315419,
      "loss_iou": 0.2236328125,
      "loss_num": 0.0159912109375,
      "loss_xval": 0.52734375,
      "num_input_tokens_seen": 288445524,
      "step": 5148
    },
    {
      "epoch": 11.46770601336303,
      "grad_norm": 24.309276580810547,
      "learning_rate": 1e-06,
      "loss": 0.4972,
      "num_input_tokens_seen": 288500740,
      "step": 5149
    },
    {
      "epoch": 11.46770601336303,
      "loss": 0.623292088508606,
      "loss_ce": 0.00012314121704548597,
      "loss_iou": 0.263671875,
      "loss_num": 0.0191650390625,
      "loss_xval": 0.625,
      "num_input_tokens_seen": 288500740,
      "step": 5149
    },
    {
      "epoch": 11.469933184855234,
      "grad_norm": 16.2718448638916,
      "learning_rate": 1e-06,
      "loss": 0.6084,
      "num_input_tokens_seen": 288553616,
      "step": 5150
    },
    {
      "epoch": 11.469933184855234,
      "loss": 0.4980197846889496,
      "loss_ce": 9.498859435552731e-05,
      "loss_iou": 0.197265625,
      "loss_num": 0.0206298828125,
      "loss_xval": 0.498046875,
      "num_input_tokens_seen": 288553616,
      "step": 5150
    },
    {
      "epoch": 11.47216035634744,
      "grad_norm": 14.356439590454102,
      "learning_rate": 1e-06,
      "loss": 0.4675,
      "num_input_tokens_seen": 288611856,
      "step": 5151
    },
    {
      "epoch": 11.47216035634744,
      "loss": 0.6592938303947449,
      "loss_ce": 0.00011415663175284863,
      "loss_iou": 0.287109375,
      "loss_num": 0.0172119140625,
      "loss_xval": 0.66015625,
      "num_input_tokens_seen": 288611856,
      "step": 5151
    },
    {
      "epoch": 11.474387527839644,
      "grad_norm": 15.644416809082031,
      "learning_rate": 1e-06,
      "loss": 0.5468,
      "num_input_tokens_seen": 288669496,
      "step": 5152
    },
    {
      "epoch": 11.474387527839644,
      "loss": 0.4559364318847656,
      "loss_ce": 0.00012589515245053917,
      "loss_iou": 0.189453125,
      "loss_num": 0.01531982421875,
      "loss_xval": 0.455078125,
      "num_input_tokens_seen": 288669496,
      "step": 5152
    },
    {
      "epoch": 11.476614699331849,
      "grad_norm": 17.78461265563965,
      "learning_rate": 1e-06,
      "loss": 0.4928,
      "num_input_tokens_seen": 288727056,
      "step": 5153
    },
    {
      "epoch": 11.476614699331849,
      "loss": 0.6607722043991089,
      "loss_ce": 0.00012771939509548247,
      "loss_iou": 0.2578125,
      "loss_num": 0.0289306640625,
      "loss_xval": 0.66015625,
      "num_input_tokens_seen": 288727056,
      "step": 5153
    },
    {
      "epoch": 11.478841870824054,
      "grad_norm": 12.651546478271484,
      "learning_rate": 1e-06,
      "loss": 0.4647,
      "num_input_tokens_seen": 288782852,
      "step": 5154
    },
    {
      "epoch": 11.478841870824054,
      "loss": 0.3253457248210907,
      "loss_ce": 0.00015041950973682106,
      "loss_iou": 0.1484375,
      "loss_num": 0.0054931640625,
      "loss_xval": 0.32421875,
      "num_input_tokens_seen": 288782852,
      "step": 5154
    },
    {
      "epoch": 11.481069042316259,
      "grad_norm": 18.01323890686035,
      "learning_rate": 1e-06,
      "loss": 0.6355,
      "num_input_tokens_seen": 288835892,
      "step": 5155
    },
    {
      "epoch": 11.481069042316259,
      "loss": 0.5539816617965698,
      "loss_ce": 0.00014869558799546212,
      "loss_iou": 0.240234375,
      "loss_num": 0.01458740234375,
      "loss_xval": 0.5546875,
      "num_input_tokens_seen": 288835892,
      "step": 5155
    },
    {
      "epoch": 11.483296213808464,
      "grad_norm": 13.251718521118164,
      "learning_rate": 1e-06,
      "loss": 0.5962,
      "num_input_tokens_seen": 288891508,
      "step": 5156
    },
    {
      "epoch": 11.483296213808464,
      "loss": 0.8043664693832397,
      "loss_ce": 0.00016726629110053182,
      "loss_iou": 0.337890625,
      "loss_num": 0.02587890625,
      "loss_xval": 0.8046875,
      "num_input_tokens_seen": 288891508,
      "step": 5156
    },
    {
      "epoch": 11.485523385300668,
      "grad_norm": 13.777200698852539,
      "learning_rate": 1e-06,
      "loss": 0.4696,
      "num_input_tokens_seen": 288948688,
      "step": 5157
    },
    {
      "epoch": 11.485523385300668,
      "loss": 0.4466667175292969,
      "loss_ce": 0.00013350519293453544,
      "loss_iou": 0.201171875,
      "loss_num": 0.0089111328125,
      "loss_xval": 0.447265625,
      "num_input_tokens_seen": 288948688,
      "step": 5157
    },
    {
      "epoch": 11.487750556792873,
      "grad_norm": 19.02153968811035,
      "learning_rate": 1e-06,
      "loss": 0.3835,
      "num_input_tokens_seen": 289004432,
      "step": 5158
    },
    {
      "epoch": 11.487750556792873,
      "loss": 0.5000653266906738,
      "loss_ce": 0.00018739307415671647,
      "loss_iou": 0.19921875,
      "loss_num": 0.0205078125,
      "loss_xval": 0.5,
      "num_input_tokens_seen": 289004432,
      "step": 5158
    },
    {
      "epoch": 11.489977728285078,
      "grad_norm": 35.4432258605957,
      "learning_rate": 1e-06,
      "loss": 0.4898,
      "num_input_tokens_seen": 289058552,
      "step": 5159
    },
    {
      "epoch": 11.489977728285078,
      "loss": 0.3960947096347809,
      "loss_ce": 9.86138402367942e-05,
      "loss_iou": 0.162109375,
      "loss_num": 0.014404296875,
      "loss_xval": 0.396484375,
      "num_input_tokens_seen": 289058552,
      "step": 5159
    },
    {
      "epoch": 11.492204899777283,
      "grad_norm": 44.51033401489258,
      "learning_rate": 1e-06,
      "loss": 0.4918,
      "num_input_tokens_seen": 289115320,
      "step": 5160
    },
    {
      "epoch": 11.492204899777283,
      "loss": 0.5892336964607239,
      "loss_ce": 0.00012237070768605918,
      "loss_iou": 0.283203125,
      "loss_num": 0.00482177734375,
      "loss_xval": 0.58984375,
      "num_input_tokens_seen": 289115320,
      "step": 5160
    },
    {
      "epoch": 11.494432071269488,
      "grad_norm": 20.451549530029297,
      "learning_rate": 1e-06,
      "loss": 0.6798,
      "num_input_tokens_seen": 289171932,
      "step": 5161
    },
    {
      "epoch": 11.494432071269488,
      "loss": 0.5714766383171082,
      "loss_ce": 0.00018755605560727417,
      "loss_iou": 0.2421875,
      "loss_num": 0.01708984375,
      "loss_xval": 0.5703125,
      "num_input_tokens_seen": 289171932,
      "step": 5161
    },
    {
      "epoch": 11.496659242761693,
      "grad_norm": 17.223535537719727,
      "learning_rate": 1e-06,
      "loss": 0.3775,
      "num_input_tokens_seen": 289227384,
      "step": 5162
    },
    {
      "epoch": 11.496659242761693,
      "loss": 0.41967087984085083,
      "loss_ce": 0.00011523931607371196,
      "loss_iou": 0.1845703125,
      "loss_num": 0.010009765625,
      "loss_xval": 0.419921875,
      "num_input_tokens_seen": 289227384,
      "step": 5162
    },
    {
      "epoch": 11.498886414253898,
      "grad_norm": 18.763608932495117,
      "learning_rate": 1e-06,
      "loss": 0.3766,
      "num_input_tokens_seen": 289282336,
      "step": 5163
    },
    {
      "epoch": 11.498886414253898,
      "loss": 0.39795833826065063,
      "loss_ce": 0.0001312018430326134,
      "loss_iou": 0.17578125,
      "loss_num": 0.00909423828125,
      "loss_xval": 0.3984375,
      "num_input_tokens_seen": 289282336,
      "step": 5163
    },
    {
      "epoch": 11.501113585746102,
      "grad_norm": 25.63532829284668,
      "learning_rate": 1e-06,
      "loss": 0.6739,
      "num_input_tokens_seen": 289337816,
      "step": 5164
    },
    {
      "epoch": 11.501113585746102,
      "loss": 0.7168751358985901,
      "loss_ce": 0.00026139506371691823,
      "loss_iou": 0.271484375,
      "loss_num": 0.034912109375,
      "loss_xval": 0.71484375,
      "num_input_tokens_seen": 289337816,
      "step": 5164
    },
    {
      "epoch": 11.503340757238307,
      "grad_norm": 17.458229064941406,
      "learning_rate": 1e-06,
      "loss": 0.5773,
      "num_input_tokens_seen": 289394264,
      "step": 5165
    },
    {
      "epoch": 11.503340757238307,
      "loss": 0.570071816444397,
      "loss_ce": 0.00012553209671750665,
      "loss_iou": 0.2431640625,
      "loss_num": 0.0164794921875,
      "loss_xval": 0.5703125,
      "num_input_tokens_seen": 289394264,
      "step": 5165
    },
    {
      "epoch": 11.505567928730512,
      "grad_norm": 15.031347274780273,
      "learning_rate": 1e-06,
      "loss": 0.4042,
      "num_input_tokens_seen": 289452076,
      "step": 5166
    },
    {
      "epoch": 11.505567928730512,
      "loss": 0.32777801156044006,
      "loss_ce": 0.0002633580006659031,
      "loss_iou": 0.14453125,
      "loss_num": 0.0078125,
      "loss_xval": 0.328125,
      "num_input_tokens_seen": 289452076,
      "step": 5166
    },
    {
      "epoch": 11.507795100222717,
      "grad_norm": 17.256629943847656,
      "learning_rate": 1e-06,
      "loss": 0.5362,
      "num_input_tokens_seen": 289507452,
      "step": 5167
    },
    {
      "epoch": 11.507795100222717,
      "loss": 0.6130794286727905,
      "loss_ce": 0.00016442046035081148,
      "loss_iou": 0.25390625,
      "loss_num": 0.021484375,
      "loss_xval": 0.61328125,
      "num_input_tokens_seen": 289507452,
      "step": 5167
    },
    {
      "epoch": 11.510022271714922,
      "grad_norm": 23.37602424621582,
      "learning_rate": 1e-06,
      "loss": 0.5456,
      "num_input_tokens_seen": 289562412,
      "step": 5168
    },
    {
      "epoch": 11.510022271714922,
      "loss": 0.6275073289871216,
      "loss_ce": 0.0001879626652225852,
      "loss_iou": 0.28125,
      "loss_num": 0.01318359375,
      "loss_xval": 0.62890625,
      "num_input_tokens_seen": 289562412,
      "step": 5168
    },
    {
      "epoch": 11.512249443207127,
      "grad_norm": 28.662569046020508,
      "learning_rate": 1e-06,
      "loss": 0.5528,
      "num_input_tokens_seen": 289618920,
      "step": 5169
    },
    {
      "epoch": 11.512249443207127,
      "loss": 0.564295768737793,
      "loss_ce": 0.00020888610742986202,
      "loss_iou": 0.24609375,
      "loss_num": 0.01422119140625,
      "loss_xval": 0.5625,
      "num_input_tokens_seen": 289618920,
      "step": 5169
    },
    {
      "epoch": 11.514476614699332,
      "grad_norm": 28.88167381286621,
      "learning_rate": 1e-06,
      "loss": 0.5267,
      "num_input_tokens_seen": 289675588,
      "step": 5170
    },
    {
      "epoch": 11.514476614699332,
      "loss": 0.46462541818618774,
      "loss_ce": 0.0001478660269640386,
      "loss_iou": 0.212890625,
      "loss_num": 0.00787353515625,
      "loss_xval": 0.46484375,
      "num_input_tokens_seen": 289675588,
      "step": 5170
    },
    {
      "epoch": 11.516703786191536,
      "grad_norm": 12.050206184387207,
      "learning_rate": 1e-06,
      "loss": 0.4747,
      "num_input_tokens_seen": 289733692,
      "step": 5171
    },
    {
      "epoch": 11.516703786191536,
      "loss": 0.41063833236694336,
      "loss_ce": 0.00011587166227400303,
      "loss_iou": 0.18359375,
      "loss_num": 0.00860595703125,
      "loss_xval": 0.41015625,
      "num_input_tokens_seen": 289733692,
      "step": 5171
    },
    {
      "epoch": 11.518930957683741,
      "grad_norm": 13.707185745239258,
      "learning_rate": 1e-06,
      "loss": 0.5053,
      "num_input_tokens_seen": 289791396,
      "step": 5172
    },
    {
      "epoch": 11.518930957683741,
      "loss": 0.5518975257873535,
      "loss_ce": 0.00013974419562146068,
      "loss_iou": 0.2578125,
      "loss_num": 0.00677490234375,
      "loss_xval": 0.55078125,
      "num_input_tokens_seen": 289791396,
      "step": 5172
    },
    {
      "epoch": 11.521158129175946,
      "grad_norm": 17.915735244750977,
      "learning_rate": 1e-06,
      "loss": 0.6097,
      "num_input_tokens_seen": 289848220,
      "step": 5173
    },
    {
      "epoch": 11.521158129175946,
      "loss": 0.711284875869751,
      "loss_ce": 0.0003473775868769735,
      "loss_iou": 0.28515625,
      "loss_num": 0.028076171875,
      "loss_xval": 0.7109375,
      "num_input_tokens_seen": 289848220,
      "step": 5173
    },
    {
      "epoch": 11.523385300668151,
      "grad_norm": 19.404088973999023,
      "learning_rate": 1e-06,
      "loss": 0.4896,
      "num_input_tokens_seen": 289907036,
      "step": 5174
    },
    {
      "epoch": 11.523385300668151,
      "loss": 0.4599684178829193,
      "loss_ce": 0.00012956123100593686,
      "loss_iou": 0.203125,
      "loss_num": 0.0107421875,
      "loss_xval": 0.458984375,
      "num_input_tokens_seen": 289907036,
      "step": 5174
    },
    {
      "epoch": 11.525612472160356,
      "grad_norm": 14.955151557922363,
      "learning_rate": 1e-06,
      "loss": 0.4704,
      "num_input_tokens_seen": 289963580,
      "step": 5175
    },
    {
      "epoch": 11.525612472160356,
      "loss": 0.6151525974273682,
      "loss_ce": 0.0001623880089027807,
      "loss_iou": 0.271484375,
      "loss_num": 0.01409912109375,
      "loss_xval": 0.61328125,
      "num_input_tokens_seen": 289963580,
      "step": 5175
    },
    {
      "epoch": 11.52783964365256,
      "grad_norm": 30.26787567138672,
      "learning_rate": 1e-06,
      "loss": 0.4761,
      "num_input_tokens_seen": 290016656,
      "step": 5176
    },
    {
      "epoch": 11.52783964365256,
      "loss": 0.3829212784767151,
      "loss_ce": 0.00010878632019739598,
      "loss_iou": 0.15625,
      "loss_num": 0.0140380859375,
      "loss_xval": 0.3828125,
      "num_input_tokens_seen": 290016656,
      "step": 5176
    },
    {
      "epoch": 11.530066815144766,
      "grad_norm": 16.603086471557617,
      "learning_rate": 1e-06,
      "loss": 0.447,
      "num_input_tokens_seen": 290074272,
      "step": 5177
    },
    {
      "epoch": 11.530066815144766,
      "loss": 0.38126683235168457,
      "loss_ce": 0.00016332257655449212,
      "loss_iou": 0.16796875,
      "loss_num": 0.00909423828125,
      "loss_xval": 0.380859375,
      "num_input_tokens_seen": 290074272,
      "step": 5177
    },
    {
      "epoch": 11.53229398663697,
      "grad_norm": 13.765169143676758,
      "learning_rate": 1e-06,
      "loss": 0.6256,
      "num_input_tokens_seen": 290129748,
      "step": 5178
    },
    {
      "epoch": 11.53229398663697,
      "loss": 0.6682248115539551,
      "loss_ce": 0.00013396795839071274,
      "loss_iou": 0.294921875,
      "loss_num": 0.015869140625,
      "loss_xval": 0.66796875,
      "num_input_tokens_seen": 290129748,
      "step": 5178
    },
    {
      "epoch": 11.534521158129175,
      "grad_norm": 26.716320037841797,
      "learning_rate": 1e-06,
      "loss": 0.7209,
      "num_input_tokens_seen": 290182560,
      "step": 5179
    },
    {
      "epoch": 11.534521158129175,
      "loss": 0.7032727003097534,
      "loss_ce": 0.0001476738107157871,
      "loss_iou": 0.3046875,
      "loss_num": 0.0181884765625,
      "loss_xval": 0.703125,
      "num_input_tokens_seen": 290182560,
      "step": 5179
    },
    {
      "epoch": 11.53674832962138,
      "grad_norm": 23.010082244873047,
      "learning_rate": 1e-06,
      "loss": 0.7079,
      "num_input_tokens_seen": 290238272,
      "step": 5180
    },
    {
      "epoch": 11.53674832962138,
      "loss": 0.6293861865997314,
      "loss_ce": 0.00023578619584441185,
      "loss_iou": 0.28515625,
      "loss_num": 0.01171875,
      "loss_xval": 0.62890625,
      "num_input_tokens_seen": 290238272,
      "step": 5180
    },
    {
      "epoch": 11.538975501113585,
      "grad_norm": 12.722779273986816,
      "learning_rate": 1e-06,
      "loss": 0.4231,
      "num_input_tokens_seen": 290294832,
      "step": 5181
    },
    {
      "epoch": 11.538975501113585,
      "loss": 0.5143045783042908,
      "loss_ce": 0.00014442717656493187,
      "loss_iou": 0.2236328125,
      "loss_num": 0.01324462890625,
      "loss_xval": 0.515625,
      "num_input_tokens_seen": 290294832,
      "step": 5181
    },
    {
      "epoch": 11.54120267260579,
      "grad_norm": 17.865341186523438,
      "learning_rate": 1e-06,
      "loss": 0.599,
      "num_input_tokens_seen": 290351116,
      "step": 5182
    },
    {
      "epoch": 11.54120267260579,
      "loss": 0.6641983985900879,
      "loss_ce": 0.0001359161688014865,
      "loss_iou": 0.255859375,
      "loss_num": 0.03076171875,
      "loss_xval": 0.6640625,
      "num_input_tokens_seen": 290351116,
      "step": 5182
    },
    {
      "epoch": 11.543429844097995,
      "grad_norm": 16.24390983581543,
      "learning_rate": 1e-06,
      "loss": 0.4797,
      "num_input_tokens_seen": 290406636,
      "step": 5183
    },
    {
      "epoch": 11.543429844097995,
      "loss": 0.5366250276565552,
      "loss_ce": 0.0001260171557078138,
      "loss_iou": 0.2255859375,
      "loss_num": 0.01708984375,
      "loss_xval": 0.53515625,
      "num_input_tokens_seen": 290406636,
      "step": 5183
    },
    {
      "epoch": 11.5456570155902,
      "grad_norm": 33.181251525878906,
      "learning_rate": 1e-06,
      "loss": 0.5539,
      "num_input_tokens_seen": 290464236,
      "step": 5184
    },
    {
      "epoch": 11.5456570155902,
      "loss": 0.4681469798088074,
      "loss_ce": 0.0001293782697757706,
      "loss_iou": 0.2138671875,
      "loss_num": 0.008056640625,
      "loss_xval": 0.46875,
      "num_input_tokens_seen": 290464236,
      "step": 5184
    },
    {
      "epoch": 11.547884187082406,
      "grad_norm": 15.906049728393555,
      "learning_rate": 1e-06,
      "loss": 0.4499,
      "num_input_tokens_seen": 290519360,
      "step": 5185
    },
    {
      "epoch": 11.547884187082406,
      "loss": 0.5545018315315247,
      "loss_ce": 0.00018054830434266478,
      "loss_iou": 0.2294921875,
      "loss_num": 0.0191650390625,
      "loss_xval": 0.5546875,
      "num_input_tokens_seen": 290519360,
      "step": 5185
    },
    {
      "epoch": 11.550111358574611,
      "grad_norm": 25.81562042236328,
      "learning_rate": 1e-06,
      "loss": 0.5231,
      "num_input_tokens_seen": 290572196,
      "step": 5186
    },
    {
      "epoch": 11.550111358574611,
      "loss": 0.5178329944610596,
      "loss_ce": 0.00025483942590653896,
      "loss_iou": 0.23046875,
      "loss_num": 0.01141357421875,
      "loss_xval": 0.515625,
      "num_input_tokens_seen": 290572196,
      "step": 5186
    },
    {
      "epoch": 11.552338530066816,
      "grad_norm": 17.265830993652344,
      "learning_rate": 1e-06,
      "loss": 0.4436,
      "num_input_tokens_seen": 290629824,
      "step": 5187
    },
    {
      "epoch": 11.552338530066816,
      "loss": 0.4118680953979492,
      "loss_ce": 0.0001249448541784659,
      "loss_iou": 0.1728515625,
      "loss_num": 0.01318359375,
      "loss_xval": 0.412109375,
      "num_input_tokens_seen": 290629824,
      "step": 5187
    },
    {
      "epoch": 11.55456570155902,
      "grad_norm": 13.806538581848145,
      "learning_rate": 1e-06,
      "loss": 0.465,
      "num_input_tokens_seen": 290683676,
      "step": 5188
    },
    {
      "epoch": 11.55456570155902,
      "loss": 0.28974953293800354,
      "loss_ce": 0.00010721624130383134,
      "loss_iou": 0.1220703125,
      "loss_num": 0.00921630859375,
      "loss_xval": 0.2890625,
      "num_input_tokens_seen": 290683676,
      "step": 5188
    },
    {
      "epoch": 11.556792873051226,
      "grad_norm": 25.621070861816406,
      "learning_rate": 1e-06,
      "loss": 0.5772,
      "num_input_tokens_seen": 290738676,
      "step": 5189
    },
    {
      "epoch": 11.556792873051226,
      "loss": 0.5882642865180969,
      "loss_ce": 0.0001295104157179594,
      "loss_iou": 0.25,
      "loss_num": 0.0174560546875,
      "loss_xval": 0.58984375,
      "num_input_tokens_seen": 290738676,
      "step": 5189
    },
    {
      "epoch": 11.55902004454343,
      "grad_norm": 21.53620719909668,
      "learning_rate": 1e-06,
      "loss": 0.583,
      "num_input_tokens_seen": 290793796,
      "step": 5190
    },
    {
      "epoch": 11.55902004454343,
      "loss": 0.6559303998947144,
      "loss_ce": 0.000412814028095454,
      "loss_iou": 0.27734375,
      "loss_num": 0.0203857421875,
      "loss_xval": 0.65625,
      "num_input_tokens_seen": 290793796,
      "step": 5190
    },
    {
      "epoch": 11.561247216035635,
      "grad_norm": 16.96056365966797,
      "learning_rate": 1e-06,
      "loss": 0.5001,
      "num_input_tokens_seen": 290850768,
      "step": 5191
    },
    {
      "epoch": 11.561247216035635,
      "loss": 0.5135844349861145,
      "loss_ce": 0.00015668988635297865,
      "loss_iou": 0.23046875,
      "loss_num": 0.010498046875,
      "loss_xval": 0.51171875,
      "num_input_tokens_seen": 290850768,
      "step": 5191
    },
    {
      "epoch": 11.56347438752784,
      "grad_norm": 19.151756286621094,
      "learning_rate": 1e-06,
      "loss": 0.5346,
      "num_input_tokens_seen": 290905584,
      "step": 5192
    },
    {
      "epoch": 11.56347438752784,
      "loss": 0.3937605321407318,
      "loss_ce": 0.00014479970559477806,
      "loss_iou": 0.1640625,
      "loss_num": 0.0130615234375,
      "loss_xval": 0.39453125,
      "num_input_tokens_seen": 290905584,
      "step": 5192
    },
    {
      "epoch": 11.565701559020045,
      "grad_norm": 19.19948959350586,
      "learning_rate": 1e-06,
      "loss": 0.5524,
      "num_input_tokens_seen": 290963692,
      "step": 5193
    },
    {
      "epoch": 11.565701559020045,
      "loss": 0.6412444114685059,
      "loss_ce": 0.00013114791363477707,
      "loss_iou": 0.267578125,
      "loss_num": 0.0208740234375,
      "loss_xval": 0.640625,
      "num_input_tokens_seen": 290963692,
      "step": 5193
    },
    {
      "epoch": 11.56792873051225,
      "grad_norm": 18.386707305908203,
      "learning_rate": 1e-06,
      "loss": 0.4723,
      "num_input_tokens_seen": 291021232,
      "step": 5194
    },
    {
      "epoch": 11.56792873051225,
      "loss": 0.3769450783729553,
      "loss_ce": 0.0002360670769121498,
      "loss_iou": 0.1533203125,
      "loss_num": 0.01397705078125,
      "loss_xval": 0.376953125,
      "num_input_tokens_seen": 291021232,
      "step": 5194
    },
    {
      "epoch": 11.570155902004455,
      "grad_norm": 20.23085594177246,
      "learning_rate": 1e-06,
      "loss": 0.7242,
      "num_input_tokens_seen": 291075260,
      "step": 5195
    },
    {
      "epoch": 11.570155902004455,
      "loss": 0.5950526595115662,
      "loss_ce": 0.00014300120528787374,
      "loss_iou": 0.2734375,
      "loss_num": 0.00982666015625,
      "loss_xval": 0.59375,
      "num_input_tokens_seen": 291075260,
      "step": 5195
    },
    {
      "epoch": 11.57238307349666,
      "grad_norm": 23.331451416015625,
      "learning_rate": 1e-06,
      "loss": 0.7261,
      "num_input_tokens_seen": 291131208,
      "step": 5196
    },
    {
      "epoch": 11.57238307349666,
      "loss": 0.664703905582428,
      "loss_ce": 0.000153136788867414,
      "loss_iou": 0.28125,
      "loss_num": 0.020263671875,
      "loss_xval": 0.6640625,
      "num_input_tokens_seen": 291131208,
      "step": 5196
    },
    {
      "epoch": 11.574610244988865,
      "grad_norm": 25.03767967224121,
      "learning_rate": 1e-06,
      "loss": 0.567,
      "num_input_tokens_seen": 291187856,
      "step": 5197
    },
    {
      "epoch": 11.574610244988865,
      "loss": 0.38165202736854553,
      "loss_ce": 0.0001823049533413723,
      "loss_iou": 0.1611328125,
      "loss_num": 0.01177978515625,
      "loss_xval": 0.380859375,
      "num_input_tokens_seen": 291187856,
      "step": 5197
    },
    {
      "epoch": 11.57683741648107,
      "grad_norm": 18.827882766723633,
      "learning_rate": 1e-06,
      "loss": 0.7002,
      "num_input_tokens_seen": 291243456,
      "step": 5198
    },
    {
      "epoch": 11.57683741648107,
      "loss": 0.9664729237556458,
      "loss_ce": 0.0004084957472514361,
      "loss_iou": 0.44140625,
      "loss_num": 0.0164794921875,
      "loss_xval": 0.96484375,
      "num_input_tokens_seen": 291243456,
      "step": 5198
    },
    {
      "epoch": 11.579064587973274,
      "grad_norm": 15.689250946044922,
      "learning_rate": 1e-06,
      "loss": 0.5917,
      "num_input_tokens_seen": 291301168,
      "step": 5199
    },
    {
      "epoch": 11.579064587973274,
      "loss": 0.37210649251937866,
      "loss_ce": 0.00015823188005015254,
      "loss_iou": 0.1630859375,
      "loss_num": 0.009033203125,
      "loss_xval": 0.37109375,
      "num_input_tokens_seen": 291301168,
      "step": 5199
    },
    {
      "epoch": 11.58129175946548,
      "grad_norm": 26.7224178314209,
      "learning_rate": 1e-06,
      "loss": 0.6347,
      "num_input_tokens_seen": 291356144,
      "step": 5200
    },
    {
      "epoch": 11.58129175946548,
      "loss": 0.6205277442932129,
      "loss_ce": 0.00016642545233480632,
      "loss_iou": 0.2734375,
      "loss_num": 0.0147705078125,
      "loss_xval": 0.62109375,
      "num_input_tokens_seen": 291356144,
      "step": 5200
    },
    {
      "epoch": 11.583518930957684,
      "grad_norm": 17.09632110595703,
      "learning_rate": 1e-06,
      "loss": 0.431,
      "num_input_tokens_seen": 291412052,
      "step": 5201
    },
    {
      "epoch": 11.583518930957684,
      "loss": 0.3809802234172821,
      "loss_ce": 0.0001208461108035408,
      "loss_iou": 0.16796875,
      "loss_num": 0.0089111328125,
      "loss_xval": 0.380859375,
      "num_input_tokens_seen": 291412052,
      "step": 5201
    },
    {
      "epoch": 11.585746102449889,
      "grad_norm": 13.215712547302246,
      "learning_rate": 1e-06,
      "loss": 0.3926,
      "num_input_tokens_seen": 291468324,
      "step": 5202
    },
    {
      "epoch": 11.585746102449889,
      "loss": 0.35443952679634094,
      "loss_ce": 0.00013045519881416112,
      "loss_iou": 0.1533203125,
      "loss_num": 0.00946044921875,
      "loss_xval": 0.353515625,
      "num_input_tokens_seen": 291468324,
      "step": 5202
    },
    {
      "epoch": 11.587973273942094,
      "grad_norm": 15.096648216247559,
      "learning_rate": 1e-06,
      "loss": 0.4367,
      "num_input_tokens_seen": 291525764,
      "step": 5203
    },
    {
      "epoch": 11.587973273942094,
      "loss": 0.31994152069091797,
      "loss_ce": 0.00011730578989954665,
      "loss_iou": 0.1455078125,
      "loss_num": 0.00579833984375,
      "loss_xval": 0.3203125,
      "num_input_tokens_seen": 291525764,
      "step": 5203
    },
    {
      "epoch": 11.590200445434299,
      "grad_norm": 21.537437438964844,
      "learning_rate": 1e-06,
      "loss": 0.4525,
      "num_input_tokens_seen": 291581816,
      "step": 5204
    },
    {
      "epoch": 11.590200445434299,
      "loss": 0.4407280683517456,
      "loss_ce": 0.0001763150212354958,
      "loss_iou": 0.181640625,
      "loss_num": 0.0155029296875,
      "loss_xval": 0.44140625,
      "num_input_tokens_seen": 291581816,
      "step": 5204
    },
    {
      "epoch": 11.592427616926503,
      "grad_norm": 17.791162490844727,
      "learning_rate": 1e-06,
      "loss": 0.4189,
      "num_input_tokens_seen": 291640008,
      "step": 5205
    },
    {
      "epoch": 11.592427616926503,
      "loss": 0.45544835925102234,
      "loss_ce": 0.0001261141151189804,
      "loss_iou": 0.20703125,
      "loss_num": 0.00836181640625,
      "loss_xval": 0.455078125,
      "num_input_tokens_seen": 291640008,
      "step": 5205
    },
    {
      "epoch": 11.594654788418708,
      "grad_norm": 15.262166976928711,
      "learning_rate": 1e-06,
      "loss": 0.4908,
      "num_input_tokens_seen": 291696576,
      "step": 5206
    },
    {
      "epoch": 11.594654788418708,
      "loss": 0.5726983547210693,
      "loss_ce": 0.00012755353236570954,
      "loss_iou": 0.2470703125,
      "loss_num": 0.0157470703125,
      "loss_xval": 0.57421875,
      "num_input_tokens_seen": 291696576,
      "step": 5206
    },
    {
      "epoch": 11.596881959910913,
      "grad_norm": 18.53944969177246,
      "learning_rate": 1e-06,
      "loss": 0.4826,
      "num_input_tokens_seen": 291753164,
      "step": 5207
    },
    {
      "epoch": 11.596881959910913,
      "loss": 0.5484709143638611,
      "loss_ce": 0.0001310570805799216,
      "loss_iou": 0.2236328125,
      "loss_num": 0.0205078125,
      "loss_xval": 0.546875,
      "num_input_tokens_seen": 291753164,
      "step": 5207
    },
    {
      "epoch": 11.599109131403118,
      "grad_norm": 20.32893943786621,
      "learning_rate": 1e-06,
      "loss": 0.5119,
      "num_input_tokens_seen": 291807164,
      "step": 5208
    },
    {
      "epoch": 11.599109131403118,
      "loss": 0.6429668664932251,
      "loss_ce": 0.00014457208453677595,
      "loss_iou": 0.2734375,
      "loss_num": 0.019287109375,
      "loss_xval": 0.64453125,
      "num_input_tokens_seen": 291807164,
      "step": 5208
    },
    {
      "epoch": 11.601336302895323,
      "grad_norm": 16.207355499267578,
      "learning_rate": 1e-06,
      "loss": 0.3707,
      "num_input_tokens_seen": 291861832,
      "step": 5209
    },
    {
      "epoch": 11.601336302895323,
      "loss": 0.35583117604255676,
      "loss_ce": 0.00011830384755739942,
      "loss_iou": 0.1591796875,
      "loss_num": 0.007354736328125,
      "loss_xval": 0.35546875,
      "num_input_tokens_seen": 291861832,
      "step": 5209
    },
    {
      "epoch": 11.603563474387528,
      "grad_norm": 20.643909454345703,
      "learning_rate": 1e-06,
      "loss": 0.3916,
      "num_input_tokens_seen": 291919612,
      "step": 5210
    },
    {
      "epoch": 11.603563474387528,
      "loss": 0.40422919392585754,
      "loss_ce": 0.0001764479384291917,
      "loss_iou": 0.162109375,
      "loss_num": 0.015869140625,
      "loss_xval": 0.404296875,
      "num_input_tokens_seen": 291919612,
      "step": 5210
    },
    {
      "epoch": 11.605790645879733,
      "grad_norm": 18.431650161743164,
      "learning_rate": 1e-06,
      "loss": 0.5809,
      "num_input_tokens_seen": 291976832,
      "step": 5211
    },
    {
      "epoch": 11.605790645879733,
      "loss": 0.5501755475997925,
      "loss_ce": 0.00012670463183894753,
      "loss_iou": 0.2490234375,
      "loss_num": 0.010498046875,
      "loss_xval": 0.55078125,
      "num_input_tokens_seen": 291976832,
      "step": 5211
    },
    {
      "epoch": 11.608017817371937,
      "grad_norm": 24.48961639404297,
      "learning_rate": 1e-06,
      "loss": 0.5894,
      "num_input_tokens_seen": 292033284,
      "step": 5212
    },
    {
      "epoch": 11.608017817371937,
      "loss": 0.565216064453125,
      "loss_ce": 0.00015255842299666256,
      "loss_iou": 0.248046875,
      "loss_num": 0.0135498046875,
      "loss_xval": 0.56640625,
      "num_input_tokens_seen": 292033284,
      "step": 5212
    },
    {
      "epoch": 11.610244988864142,
      "grad_norm": 16.848508834838867,
      "learning_rate": 1e-06,
      "loss": 0.404,
      "num_input_tokens_seen": 292086472,
      "step": 5213
    },
    {
      "epoch": 11.610244988864142,
      "loss": 0.3769487142562866,
      "loss_ce": 0.0001176485966425389,
      "loss_iou": 0.169921875,
      "loss_num": 0.007354736328125,
      "loss_xval": 0.376953125,
      "num_input_tokens_seen": 292086472,
      "step": 5213
    },
    {
      "epoch": 11.612472160356347,
      "grad_norm": 26.47736167907715,
      "learning_rate": 1e-06,
      "loss": 0.5472,
      "num_input_tokens_seen": 292141628,
      "step": 5214
    },
    {
      "epoch": 11.612472160356347,
      "loss": 0.6076180338859558,
      "loss_ce": 0.00013512838631868362,
      "loss_iou": 0.26953125,
      "loss_num": 0.01361083984375,
      "loss_xval": 0.609375,
      "num_input_tokens_seen": 292141628,
      "step": 5214
    },
    {
      "epoch": 11.614699331848552,
      "grad_norm": 20.124786376953125,
      "learning_rate": 1e-06,
      "loss": 0.5394,
      "num_input_tokens_seen": 292195460,
      "step": 5215
    },
    {
      "epoch": 11.614699331848552,
      "loss": 0.746225118637085,
      "loss_ce": 0.00013138932990841568,
      "loss_iou": 0.33203125,
      "loss_num": 0.0166015625,
      "loss_xval": 0.74609375,
      "num_input_tokens_seen": 292195460,
      "step": 5215
    },
    {
      "epoch": 11.616926503340757,
      "grad_norm": 16.484128952026367,
      "learning_rate": 1e-06,
      "loss": 0.5526,
      "num_input_tokens_seen": 292251864,
      "step": 5216
    },
    {
      "epoch": 11.616926503340757,
      "loss": 0.5467870831489563,
      "loss_ce": 0.00015625265950802714,
      "loss_iou": 0.22265625,
      "loss_num": 0.0201416015625,
      "loss_xval": 0.546875,
      "num_input_tokens_seen": 292251864,
      "step": 5216
    },
    {
      "epoch": 11.619153674832962,
      "grad_norm": 24.076684951782227,
      "learning_rate": 1e-06,
      "loss": 0.6972,
      "num_input_tokens_seen": 292304412,
      "step": 5217
    },
    {
      "epoch": 11.619153674832962,
      "loss": 0.6573572158813477,
      "loss_ce": 0.00013063887308817357,
      "loss_iou": 0.2890625,
      "loss_num": 0.01611328125,
      "loss_xval": 0.65625,
      "num_input_tokens_seen": 292304412,
      "step": 5217
    },
    {
      "epoch": 11.621380846325167,
      "grad_norm": 20.732887268066406,
      "learning_rate": 1e-06,
      "loss": 0.5674,
      "num_input_tokens_seen": 292360272,
      "step": 5218
    },
    {
      "epoch": 11.621380846325167,
      "loss": 0.6395326256752014,
      "loss_ce": 0.00012831481581088156,
      "loss_iou": 0.27734375,
      "loss_num": 0.017333984375,
      "loss_xval": 0.640625,
      "num_input_tokens_seen": 292360272,
      "step": 5218
    },
    {
      "epoch": 11.623608017817372,
      "grad_norm": 19.97538948059082,
      "learning_rate": 1e-06,
      "loss": 0.4222,
      "num_input_tokens_seen": 292415788,
      "step": 5219
    },
    {
      "epoch": 11.623608017817372,
      "loss": 0.3983200788497925,
      "loss_ce": 0.00012667715782299638,
      "loss_iou": 0.1845703125,
      "loss_num": 0.0057373046875,
      "loss_xval": 0.3984375,
      "num_input_tokens_seen": 292415788,
      "step": 5219
    },
    {
      "epoch": 11.625835189309576,
      "grad_norm": 22.856096267700195,
      "learning_rate": 1e-06,
      "loss": 0.5726,
      "num_input_tokens_seen": 292467908,
      "step": 5220
    },
    {
      "epoch": 11.625835189309576,
      "loss": 0.6926459074020386,
      "loss_ce": 0.00014099694089964032,
      "loss_iou": 0.310546875,
      "loss_num": 0.0147705078125,
      "loss_xval": 0.69140625,
      "num_input_tokens_seen": 292467908,
      "step": 5220
    },
    {
      "epoch": 11.628062360801781,
      "grad_norm": 15.5626859664917,
      "learning_rate": 1e-06,
      "loss": 0.6509,
      "num_input_tokens_seen": 292523568,
      "step": 5221
    },
    {
      "epoch": 11.628062360801781,
      "loss": 0.6292951107025146,
      "loss_ce": 0.0001447499671485275,
      "loss_iou": 0.296875,
      "loss_num": 0.00738525390625,
      "loss_xval": 0.62890625,
      "num_input_tokens_seen": 292523568,
      "step": 5221
    },
    {
      "epoch": 11.630289532293986,
      "grad_norm": 38.288299560546875,
      "learning_rate": 1e-06,
      "loss": 0.5488,
      "num_input_tokens_seen": 292577496,
      "step": 5222
    },
    {
      "epoch": 11.630289532293986,
      "loss": 0.6104898452758789,
      "loss_ce": 0.00013826916983816773,
      "loss_iou": 0.2451171875,
      "loss_num": 0.02392578125,
      "loss_xval": 0.609375,
      "num_input_tokens_seen": 292577496,
      "step": 5222
    },
    {
      "epoch": 11.632516703786191,
      "grad_norm": 25.399972915649414,
      "learning_rate": 1e-06,
      "loss": 0.482,
      "num_input_tokens_seen": 292633528,
      "step": 5223
    },
    {
      "epoch": 11.632516703786191,
      "loss": 0.5387084484100342,
      "loss_ce": 0.00013427785597741604,
      "loss_iou": 0.2412109375,
      "loss_num": 0.010986328125,
      "loss_xval": 0.5390625,
      "num_input_tokens_seen": 292633528,
      "step": 5223
    },
    {
      "epoch": 11.634743875278396,
      "grad_norm": 85.37445831298828,
      "learning_rate": 1e-06,
      "loss": 0.5576,
      "num_input_tokens_seen": 292688036,
      "step": 5224
    },
    {
      "epoch": 11.634743875278396,
      "loss": 0.7559876441955566,
      "loss_ce": 0.00012828274338971823,
      "loss_iou": 0.298828125,
      "loss_num": 0.031494140625,
      "loss_xval": 0.7578125,
      "num_input_tokens_seen": 292688036,
      "step": 5224
    },
    {
      "epoch": 11.6369710467706,
      "grad_norm": 54.62841796875,
      "learning_rate": 1e-06,
      "loss": 0.4919,
      "num_input_tokens_seen": 292745052,
      "step": 5225
    },
    {
      "epoch": 11.6369710467706,
      "loss": 0.5994965434074402,
      "loss_ce": 0.00013131627929396927,
      "loss_iou": 0.265625,
      "loss_num": 0.0140380859375,
      "loss_xval": 0.59765625,
      "num_input_tokens_seen": 292745052,
      "step": 5225
    },
    {
      "epoch": 11.639198218262806,
      "grad_norm": 13.105633735656738,
      "learning_rate": 1e-06,
      "loss": 0.3966,
      "num_input_tokens_seen": 292803352,
      "step": 5226
    },
    {
      "epoch": 11.639198218262806,
      "loss": 0.29151397943496704,
      "loss_ce": 0.0001321233285125345,
      "loss_iou": 0.134765625,
      "loss_num": 0.00457763671875,
      "loss_xval": 0.291015625,
      "num_input_tokens_seen": 292803352,
      "step": 5226
    },
    {
      "epoch": 11.64142538975501,
      "grad_norm": 30.192888259887695,
      "learning_rate": 1e-06,
      "loss": 0.5925,
      "num_input_tokens_seen": 292859900,
      "step": 5227
    },
    {
      "epoch": 11.64142538975501,
      "loss": 0.59175705909729,
      "loss_ce": 0.00020430152653716505,
      "loss_iou": 0.236328125,
      "loss_num": 0.0238037109375,
      "loss_xval": 0.58984375,
      "num_input_tokens_seen": 292859900,
      "step": 5227
    },
    {
      "epoch": 11.643652561247215,
      "grad_norm": 21.47431755065918,
      "learning_rate": 1e-06,
      "loss": 0.4515,
      "num_input_tokens_seen": 292916668,
      "step": 5228
    },
    {
      "epoch": 11.643652561247215,
      "loss": 0.5364910364151001,
      "loss_ce": 0.00011407821148168296,
      "loss_iou": 0.228515625,
      "loss_num": 0.0159912109375,
      "loss_xval": 0.53515625,
      "num_input_tokens_seen": 292916668,
      "step": 5228
    },
    {
      "epoch": 11.64587973273942,
      "grad_norm": 29.797399520874023,
      "learning_rate": 1e-06,
      "loss": 0.4839,
      "num_input_tokens_seen": 292969696,
      "step": 5229
    },
    {
      "epoch": 11.64587973273942,
      "loss": 0.41816890239715576,
      "loss_ce": 0.00010860178736038506,
      "loss_iou": 0.1796875,
      "loss_num": 0.01177978515625,
      "loss_xval": 0.41796875,
      "num_input_tokens_seen": 292969696,
      "step": 5229
    },
    {
      "epoch": 11.648106904231625,
      "grad_norm": 11.707711219787598,
      "learning_rate": 1e-06,
      "loss": 0.456,
      "num_input_tokens_seen": 293027520,
      "step": 5230
    },
    {
      "epoch": 11.648106904231625,
      "loss": 0.5190684199333191,
      "loss_ce": 0.00014752443530596793,
      "loss_iou": 0.21875,
      "loss_num": 0.0162353515625,
      "loss_xval": 0.51953125,
      "num_input_tokens_seen": 293027520,
      "step": 5230
    },
    {
      "epoch": 11.65033407572383,
      "grad_norm": 29.20980453491211,
      "learning_rate": 1e-06,
      "loss": 0.53,
      "num_input_tokens_seen": 293085036,
      "step": 5231
    },
    {
      "epoch": 11.65033407572383,
      "loss": 0.42614778876304626,
      "loss_ce": 0.00012240000069141388,
      "loss_iou": 0.1787109375,
      "loss_num": 0.013671875,
      "loss_xval": 0.42578125,
      "num_input_tokens_seen": 293085036,
      "step": 5231
    },
    {
      "epoch": 11.652561247216035,
      "grad_norm": 39.99692916870117,
      "learning_rate": 1e-06,
      "loss": 0.7307,
      "num_input_tokens_seen": 293139624,
      "step": 5232
    },
    {
      "epoch": 11.652561247216035,
      "loss": 0.8740052580833435,
      "loss_ce": 0.00022594796610064805,
      "loss_iou": 0.361328125,
      "loss_num": 0.0301513671875,
      "loss_xval": 0.875,
      "num_input_tokens_seen": 293139624,
      "step": 5232
    },
    {
      "epoch": 11.654788418708241,
      "grad_norm": 13.025435447692871,
      "learning_rate": 1e-06,
      "loss": 0.3551,
      "num_input_tokens_seen": 293195124,
      "step": 5233
    },
    {
      "epoch": 11.654788418708241,
      "loss": 0.299808144569397,
      "loss_ce": 0.00018654628365766257,
      "loss_iou": 0.1279296875,
      "loss_num": 0.00872802734375,
      "loss_xval": 0.298828125,
      "num_input_tokens_seen": 293195124,
      "step": 5233
    },
    {
      "epoch": 11.657015590200446,
      "grad_norm": 18.018341064453125,
      "learning_rate": 1e-06,
      "loss": 0.4124,
      "num_input_tokens_seen": 293250064,
      "step": 5234
    },
    {
      "epoch": 11.657015590200446,
      "loss": 0.48889029026031494,
      "loss_ce": 0.00012073374819010496,
      "loss_iou": 0.2216796875,
      "loss_num": 0.009033203125,
      "loss_xval": 0.48828125,
      "num_input_tokens_seen": 293250064,
      "step": 5234
    },
    {
      "epoch": 11.659242761692651,
      "grad_norm": 17.814626693725586,
      "learning_rate": 1e-06,
      "loss": 0.5089,
      "num_input_tokens_seen": 293302392,
      "step": 5235
    },
    {
      "epoch": 11.659242761692651,
      "loss": 0.4006240665912628,
      "loss_ce": 0.00011137685942230746,
      "loss_iou": 0.162109375,
      "loss_num": 0.01507568359375,
      "loss_xval": 0.400390625,
      "num_input_tokens_seen": 293302392,
      "step": 5235
    },
    {
      "epoch": 11.661469933184856,
      "grad_norm": 25.53102684020996,
      "learning_rate": 1e-06,
      "loss": 0.3704,
      "num_input_tokens_seen": 293356928,
      "step": 5236
    },
    {
      "epoch": 11.661469933184856,
      "loss": 0.36693528294563293,
      "loss_ce": 0.00011399855429772288,
      "loss_iou": 0.146484375,
      "loss_num": 0.0146484375,
      "loss_xval": 0.3671875,
      "num_input_tokens_seen": 293356928,
      "step": 5236
    },
    {
      "epoch": 11.66369710467706,
      "grad_norm": 18.298686981201172,
      "learning_rate": 1e-06,
      "loss": 0.5776,
      "num_input_tokens_seen": 293410648,
      "step": 5237
    },
    {
      "epoch": 11.66369710467706,
      "loss": 0.4626512825489044,
      "loss_ce": 0.00012686576519627124,
      "loss_iou": 0.2021484375,
      "loss_num": 0.0115966796875,
      "loss_xval": 0.462890625,
      "num_input_tokens_seen": 293410648,
      "step": 5237
    },
    {
      "epoch": 11.665924276169266,
      "grad_norm": 11.563762664794922,
      "learning_rate": 1e-06,
      "loss": 0.3882,
      "num_input_tokens_seen": 293463284,
      "step": 5238
    },
    {
      "epoch": 11.665924276169266,
      "loss": 0.2963898777961731,
      "loss_ce": 0.00012523468467406929,
      "loss_iou": 0.11865234375,
      "loss_num": 0.01171875,
      "loss_xval": 0.296875,
      "num_input_tokens_seen": 293463284,
      "step": 5238
    },
    {
      "epoch": 11.66815144766147,
      "grad_norm": 16.8227596282959,
      "learning_rate": 1e-06,
      "loss": 0.6908,
      "num_input_tokens_seen": 293517264,
      "step": 5239
    },
    {
      "epoch": 11.66815144766147,
      "loss": 0.7157348394393921,
      "loss_ce": 0.0001587077567819506,
      "loss_iou": 0.30078125,
      "loss_num": 0.0233154296875,
      "loss_xval": 0.71484375,
      "num_input_tokens_seen": 293517264,
      "step": 5239
    },
    {
      "epoch": 11.670378619153675,
      "grad_norm": 17.42998695373535,
      "learning_rate": 1e-06,
      "loss": 0.4203,
      "num_input_tokens_seen": 293572600,
      "step": 5240
    },
    {
      "epoch": 11.670378619153675,
      "loss": 0.5567784905433655,
      "loss_ce": 0.0001378921151626855,
      "loss_iou": 0.2353515625,
      "loss_num": 0.0172119140625,
      "loss_xval": 0.5546875,
      "num_input_tokens_seen": 293572600,
      "step": 5240
    },
    {
      "epoch": 11.67260579064588,
      "grad_norm": 21.24955177307129,
      "learning_rate": 1e-06,
      "loss": 0.512,
      "num_input_tokens_seen": 293628344,
      "step": 5241
    },
    {
      "epoch": 11.67260579064588,
      "loss": 0.3846355080604553,
      "loss_ce": 0.00011400008224882185,
      "loss_iou": 0.1494140625,
      "loss_num": 0.0172119140625,
      "loss_xval": 0.384765625,
      "num_input_tokens_seen": 293628344,
      "step": 5241
    },
    {
      "epoch": 11.674832962138085,
      "grad_norm": 27.092111587524414,
      "learning_rate": 1e-06,
      "loss": 0.6355,
      "num_input_tokens_seen": 293684796,
      "step": 5242
    },
    {
      "epoch": 11.674832962138085,
      "loss": 0.6805571913719177,
      "loss_ce": 0.0001372963743051514,
      "loss_iou": 0.306640625,
      "loss_num": 0.01373291015625,
      "loss_xval": 0.6796875,
      "num_input_tokens_seen": 293684796,
      "step": 5242
    },
    {
      "epoch": 11.67706013363029,
      "grad_norm": 16.437114715576172,
      "learning_rate": 1e-06,
      "loss": 0.4618,
      "num_input_tokens_seen": 293740992,
      "step": 5243
    },
    {
      "epoch": 11.67706013363029,
      "loss": 0.3265749514102936,
      "loss_ce": 0.00012842280557379127,
      "loss_iou": 0.14453125,
      "loss_num": 0.00732421875,
      "loss_xval": 0.326171875,
      "num_input_tokens_seen": 293740992,
      "step": 5243
    },
    {
      "epoch": 11.679287305122495,
      "grad_norm": 23.96824836730957,
      "learning_rate": 1e-06,
      "loss": 0.6494,
      "num_input_tokens_seen": 293795756,
      "step": 5244
    },
    {
      "epoch": 11.679287305122495,
      "loss": 0.47252053022384644,
      "loss_ce": 0.00010840976028703153,
      "loss_iou": 0.2001953125,
      "loss_num": 0.01416015625,
      "loss_xval": 0.47265625,
      "num_input_tokens_seen": 293795756,
      "step": 5244
    },
    {
      "epoch": 11.6815144766147,
      "grad_norm": 24.439743041992188,
      "learning_rate": 1e-06,
      "loss": 0.5064,
      "num_input_tokens_seen": 293854164,
      "step": 5245
    },
    {
      "epoch": 11.6815144766147,
      "loss": 0.6212034821510315,
      "loss_ce": 0.00010969607683364302,
      "loss_iou": 0.27734375,
      "loss_num": 0.013671875,
      "loss_xval": 0.62109375,
      "num_input_tokens_seen": 293854164,
      "step": 5245
    },
    {
      "epoch": 11.683741648106905,
      "grad_norm": 14.390751838684082,
      "learning_rate": 1e-06,
      "loss": 0.5511,
      "num_input_tokens_seen": 293910360,
      "step": 5246
    },
    {
      "epoch": 11.683741648106905,
      "loss": 0.42605873942375183,
      "loss_ce": 0.00027748823049478233,
      "loss_iou": 0.2001953125,
      "loss_num": 0.005126953125,
      "loss_xval": 0.42578125,
      "num_input_tokens_seen": 293910360,
      "step": 5246
    },
    {
      "epoch": 11.68596881959911,
      "grad_norm": 24.848167419433594,
      "learning_rate": 1e-06,
      "loss": 0.4161,
      "num_input_tokens_seen": 293969128,
      "step": 5247
    },
    {
      "epoch": 11.68596881959911,
      "loss": 0.38915523886680603,
      "loss_ce": 0.0001171459662145935,
      "loss_iou": 0.171875,
      "loss_num": 0.009033203125,
      "loss_xval": 0.388671875,
      "num_input_tokens_seen": 293969128,
      "step": 5247
    },
    {
      "epoch": 11.688195991091314,
      "grad_norm": 18.111801147460938,
      "learning_rate": 1e-06,
      "loss": 0.4344,
      "num_input_tokens_seen": 294024516,
      "step": 5248
    },
    {
      "epoch": 11.688195991091314,
      "loss": 0.5797381401062012,
      "loss_ce": 0.0001482985680922866,
      "loss_iou": 0.244140625,
      "loss_num": 0.01806640625,
      "loss_xval": 0.578125,
      "num_input_tokens_seen": 294024516,
      "step": 5248
    },
    {
      "epoch": 11.690423162583519,
      "grad_norm": 18.03676414489746,
      "learning_rate": 1e-06,
      "loss": 0.543,
      "num_input_tokens_seen": 294076868,
      "step": 5249
    },
    {
      "epoch": 11.690423162583519,
      "loss": 0.41762179136276245,
      "loss_ce": 0.0001413249410688877,
      "loss_iou": 0.1845703125,
      "loss_num": 0.009521484375,
      "loss_xval": 0.41796875,
      "num_input_tokens_seen": 294076868,
      "step": 5249
    },
    {
      "epoch": 11.692650334075724,
      "grad_norm": 12.850332260131836,
      "learning_rate": 1e-06,
      "loss": 0.401,
      "num_input_tokens_seen": 294132900,
      "step": 5250
    },
    {
      "epoch": 11.692650334075724,
      "eval_seeclick_web_CIoU": 0.5772451758384705,
      "eval_seeclick_web_GIoU": 0.5742323994636536,
      "eval_seeclick_web_IoU": 0.5963829755783081,
      "eval_seeclick_web_MAE_all": 0.015578721649944782,
      "eval_seeclick_web_MAE_h": 0.007713136961683631,
      "eval_seeclick_web_MAE_w": 0.015674122143536806,
      "eval_seeclick_web_MAE_x_boxes": 0.009651401545852423,
      "eval_seeclick_web_MAE_y_boxes": 0.021532843122258782,
      "eval_seeclick_web_inside_bbox": 0.9166666567325592,
      "eval_seeclick_web_loss": 0.929573655128479,
      "eval_seeclick_web_loss_ce": 0.00019275367230875418,
      "eval_seeclick_web_loss_iou": 0.423828125,
      "eval_seeclick_web_loss_num": 0.012483596801757812,
      "eval_seeclick_web_loss_xval": 0.909912109375,
      "eval_seeclick_web_runtime": 21.994,
      "eval_seeclick_web_samples_per_second": 2.273,
      "eval_seeclick_web_steps_per_second": 0.091,
      "num_input_tokens_seen": 294132900,
      "step": 5250
    },
    {
      "epoch": 11.692650334075724,
      "eval_icons_CIoU": 0.2656567245721817,
      "eval_icons_GIoU": 0.2912362068891525,
      "eval_icons_IoU": 0.35016924142837524,
      "eval_icons_MAE_all": 0.06389489211142063,
      "eval_icons_MAE_h": 0.04006360750645399,
      "eval_icons_MAE_w": 0.0678694061934948,
      "eval_icons_MAE_x_boxes": 0.059559810906648636,
      "eval_icons_MAE_y_boxes": 0.03940416965633631,
      "eval_icons_inside_bbox": 0.578125,
      "eval_icons_loss": 1.754642128944397,
      "eval_icons_loss_ce": 0.000254698476055637,
      "eval_icons_loss_iou": 0.68310546875,
      "eval_icons_loss_num": 0.06172943115234375,
      "eval_icons_loss_xval": 1.67333984375,
      "eval_icons_runtime": 20.5945,
      "eval_icons_samples_per_second": 2.428,
      "eval_icons_steps_per_second": 0.097,
      "num_input_tokens_seen": 294132900,
      "step": 5250
    },
    {
      "epoch": 11.692650334075724,
      "eval_screenspot_CIoU": 0.3591058651606242,
      "eval_screenspot_GIoU": 0.3725058138370514,
      "eval_screenspot_IoU": 0.4368898868560791,
      "eval_screenspot_MAE_all": 0.05924547587831815,
      "eval_screenspot_MAE_h": 0.039297002057234444,
      "eval_screenspot_MAE_w": 0.06727503115932147,
      "eval_screenspot_MAE_x_boxes": 0.07256409463783105,
      "eval_screenspot_MAE_y_boxes": 0.04037608547757069,
      "eval_screenspot_inside_bbox": 0.6862499912579855,
      "eval_screenspot_loss": 1.6120883226394653,
      "eval_screenspot_loss_ce": 0.0002683925946863989,
      "eval_screenspot_loss_iou": 0.66650390625,
      "eval_screenspot_loss_num": 0.06815338134765625,
      "eval_screenspot_loss_xval": 1.6746419270833333,
      "eval_screenspot_runtime": 35.2255,
      "eval_screenspot_samples_per_second": 2.527,
      "eval_screenspot_steps_per_second": 0.085,
      "num_input_tokens_seen": 294132900,
      "step": 5250
    },
    {
      "epoch": 11.692650334075724,
      "eval_compot_CIoU": 0.34088848531246185,
      "eval_compot_GIoU": 0.3492155075073242,
      "eval_compot_IoU": 0.400636687874794,
      "eval_compot_MAE_all": 0.01801011897623539,
      "eval_compot_MAE_h": 0.009378379676491022,
      "eval_compot_MAE_w": 0.021228870376944542,
      "eval_compot_MAE_x_boxes": 0.02999929618090391,
      "eval_compot_MAE_y_boxes": 0.007140443194657564,
      "eval_compot_inside_bbox": 0.6302083432674408,
      "eval_compot_loss": 1.4145660400390625,
      "eval_compot_loss_ce": 0.00018989732052432373,
      "eval_compot_loss_iou": 0.651611328125,
      "eval_compot_loss_num": 0.016811370849609375,
      "eval_compot_loss_xval": 1.387451171875,
      "eval_compot_runtime": 22.0109,
      "eval_compot_samples_per_second": 2.272,
      "eval_compot_steps_per_second": 0.091,
      "num_input_tokens_seen": 294132900,
      "step": 5250
    },
    {
      "epoch": 11.692650334075724,
      "eval_custom_ui_val_CIoU": 0.4712279670768314,
      "eval_custom_ui_val_GIoU": 0.48230206304126316,
      "eval_custom_ui_val_IoU": 0.533691535393397,
      "eval_custom_ui_val_MAE_all": 0.030553390168481402,
      "eval_custom_ui_val_MAE_h": 0.016979538809715047,
      "eval_custom_ui_val_MAE_w": 0.039266514798833266,
      "eval_custom_ui_val_MAE_x_boxes": 0.03789379137257735,
      "eval_custom_ui_val_MAE_y_boxes": 0.015024640881973837,
      "eval_custom_ui_val_inside_bbox": 0.7685185207260979,
      "eval_custom_ui_val_loss": 1.2054945230484009,
      "eval_custom_ui_val_loss_ce": 0.0002350941342431017,
      "eval_custom_ui_val_loss_iou": 0.5133327907986112,
      "eval_custom_ui_val_loss_num": 0.027909808688693576,
      "eval_custom_ui_val_loss_xval": 1.1665581597222223,
      "eval_custom_ui_val_runtime": 59.6121,
      "eval_custom_ui_val_samples_per_second": 4.445,
      "eval_custom_ui_val_steps_per_second": 0.151,
      "num_input_tokens_seen": 294132900,
      "step": 5250
    },
    {
      "epoch": 11.692650334075724,
      "loss": 0.9047340154647827,
      "loss_ce": 0.0001930339785758406,
      "loss_iou": 0.390625,
      "loss_num": 0.0242919921875,
      "loss_xval": 0.90625,
      "num_input_tokens_seen": 294132900,
      "step": 5250
    },
    {
      "epoch": 11.694877505567929,
      "grad_norm": 22.63847541809082,
      "learning_rate": 1e-06,
      "loss": 0.4242,
      "num_input_tokens_seen": 294188952,
      "step": 5251
    },
    {
      "epoch": 11.694877505567929,
      "loss": 0.36535871028900146,
      "loss_ce": 0.00012433102529030293,
      "loss_iou": 0.1552734375,
      "loss_num": 0.0108642578125,
      "loss_xval": 0.365234375,
      "num_input_tokens_seen": 294188952,
      "step": 5251
    },
    {
      "epoch": 11.697104677060134,
      "grad_norm": 23.704172134399414,
      "learning_rate": 1e-06,
      "loss": 0.4967,
      "num_input_tokens_seen": 294246668,
      "step": 5252
    },
    {
      "epoch": 11.697104677060134,
      "loss": 0.47535350918769836,
      "loss_ce": 0.00013378591393120587,
      "loss_iou": 0.2197265625,
      "loss_num": 0.0069580078125,
      "loss_xval": 0.474609375,
      "num_input_tokens_seen": 294246668,
      "step": 5252
    },
    {
      "epoch": 11.699331848552339,
      "grad_norm": 64.23724365234375,
      "learning_rate": 1e-06,
      "loss": 0.5925,
      "num_input_tokens_seen": 294303032,
      "step": 5253
    },
    {
      "epoch": 11.699331848552339,
      "loss": 0.4765591025352478,
      "loss_ce": 0.00011863937834277749,
      "loss_iou": 0.185546875,
      "loss_num": 0.02099609375,
      "loss_xval": 0.4765625,
      "num_input_tokens_seen": 294303032,
      "step": 5253
    },
    {
      "epoch": 11.701559020044543,
      "grad_norm": 19.613231658935547,
      "learning_rate": 1e-06,
      "loss": 0.5125,
      "num_input_tokens_seen": 294360592,
      "step": 5254
    },
    {
      "epoch": 11.701559020044543,
      "loss": 0.5249412059783936,
      "loss_ce": 0.0001609077153261751,
      "loss_iou": 0.2216796875,
      "loss_num": 0.0162353515625,
      "loss_xval": 0.5234375,
      "num_input_tokens_seen": 294360592,
      "step": 5254
    },
    {
      "epoch": 11.703786191536748,
      "grad_norm": 18.052928924560547,
      "learning_rate": 1e-06,
      "loss": 0.4107,
      "num_input_tokens_seen": 294416216,
      "step": 5255
    },
    {
      "epoch": 11.703786191536748,
      "loss": 0.3549973666667938,
      "loss_ce": 0.00013897998724132776,
      "loss_iou": 0.158203125,
      "loss_num": 0.007659912109375,
      "loss_xval": 0.35546875,
      "num_input_tokens_seen": 294416216,
      "step": 5255
    },
    {
      "epoch": 11.706013363028953,
      "grad_norm": 20.044328689575195,
      "learning_rate": 1e-06,
      "loss": 0.6376,
      "num_input_tokens_seen": 294473684,
      "step": 5256
    },
    {
      "epoch": 11.706013363028953,
      "loss": 0.5567925572395325,
      "loss_ce": 0.00015196386084426194,
      "loss_iou": 0.2294921875,
      "loss_num": 0.019775390625,
      "loss_xval": 0.5546875,
      "num_input_tokens_seen": 294473684,
      "step": 5256
    },
    {
      "epoch": 11.708240534521158,
      "grad_norm": 32.992637634277344,
      "learning_rate": 1e-06,
      "loss": 0.4129,
      "num_input_tokens_seen": 294531200,
      "step": 5257
    },
    {
      "epoch": 11.708240534521158,
      "loss": 0.41336357593536377,
      "loss_ce": 0.00015554996207356453,
      "loss_iou": 0.19140625,
      "loss_num": 0.00604248046875,
      "loss_xval": 0.4140625,
      "num_input_tokens_seen": 294531200,
      "step": 5257
    },
    {
      "epoch": 11.710467706013363,
      "grad_norm": 17.366310119628906,
      "learning_rate": 1e-06,
      "loss": 0.3378,
      "num_input_tokens_seen": 294587048,
      "step": 5258
    },
    {
      "epoch": 11.710467706013363,
      "loss": 0.3781304359436035,
      "loss_ce": 0.00020075507927685976,
      "loss_iou": 0.1767578125,
      "loss_num": 0.00494384765625,
      "loss_xval": 0.37890625,
      "num_input_tokens_seen": 294587048,
      "step": 5258
    },
    {
      "epoch": 11.712694877505568,
      "grad_norm": 19.407751083374023,
      "learning_rate": 1e-06,
      "loss": 0.55,
      "num_input_tokens_seen": 294645188,
      "step": 5259
    },
    {
      "epoch": 11.712694877505568,
      "loss": 0.6682122349739075,
      "loss_ce": 0.00012143873027525842,
      "loss_iou": 0.283203125,
      "loss_num": 0.0203857421875,
      "loss_xval": 0.66796875,
      "num_input_tokens_seen": 294645188,
      "step": 5259
    },
    {
      "epoch": 11.714922048997773,
      "grad_norm": 27.0356388092041,
      "learning_rate": 1e-06,
      "loss": 0.5196,
      "num_input_tokens_seen": 294702920,
      "step": 5260
    },
    {
      "epoch": 11.714922048997773,
      "loss": 0.5554238557815552,
      "loss_ce": 0.00012605905067175627,
      "loss_iou": 0.240234375,
      "loss_num": 0.01507568359375,
      "loss_xval": 0.5546875,
      "num_input_tokens_seen": 294702920,
      "step": 5260
    },
    {
      "epoch": 11.717149220489977,
      "grad_norm": 25.279115676879883,
      "learning_rate": 1e-06,
      "loss": 0.5789,
      "num_input_tokens_seen": 294760020,
      "step": 5261
    },
    {
      "epoch": 11.717149220489977,
      "loss": 0.6674556732177734,
      "loss_ce": 0.0002193464315496385,
      "loss_iou": 0.283203125,
      "loss_num": 0.0205078125,
      "loss_xval": 0.66796875,
      "num_input_tokens_seen": 294760020,
      "step": 5261
    },
    {
      "epoch": 11.719376391982182,
      "grad_norm": 28.12050437927246,
      "learning_rate": 1e-06,
      "loss": 0.4775,
      "num_input_tokens_seen": 294815096,
      "step": 5262
    },
    {
      "epoch": 11.719376391982182,
      "loss": 0.4449352025985718,
      "loss_ce": 0.00011096900561824441,
      "loss_iou": 0.2001953125,
      "loss_num": 0.0089111328125,
      "loss_xval": 0.4453125,
      "num_input_tokens_seen": 294815096,
      "step": 5262
    },
    {
      "epoch": 11.721603563474387,
      "grad_norm": 15.876343727111816,
      "learning_rate": 1e-06,
      "loss": 0.6174,
      "num_input_tokens_seen": 294872920,
      "step": 5263
    },
    {
      "epoch": 11.721603563474387,
      "loss": 0.6857922077178955,
      "loss_ce": 0.0003674498002510518,
      "loss_iou": 0.30078125,
      "loss_num": 0.0166015625,
      "loss_xval": 0.68359375,
      "num_input_tokens_seen": 294872920,
      "step": 5263
    },
    {
      "epoch": 11.723830734966592,
      "grad_norm": 19.064332962036133,
      "learning_rate": 1e-06,
      "loss": 0.4169,
      "num_input_tokens_seen": 294929388,
      "step": 5264
    },
    {
      "epoch": 11.723830734966592,
      "loss": 0.2714940309524536,
      "loss_ce": 0.00013172137551009655,
      "loss_iou": 0.11279296875,
      "loss_num": 0.00909423828125,
      "loss_xval": 0.271484375,
      "num_input_tokens_seen": 294929388,
      "step": 5264
    },
    {
      "epoch": 11.726057906458797,
      "grad_norm": 19.456663131713867,
      "learning_rate": 1e-06,
      "loss": 0.4162,
      "num_input_tokens_seen": 294983048,
      "step": 5265
    },
    {
      "epoch": 11.726057906458797,
      "loss": 0.4633742868900299,
      "loss_ce": 0.00011746423842851073,
      "loss_iou": 0.2021484375,
      "loss_num": 0.01165771484375,
      "loss_xval": 0.462890625,
      "num_input_tokens_seen": 294983048,
      "step": 5265
    },
    {
      "epoch": 11.728285077951002,
      "grad_norm": 18.227783203125,
      "learning_rate": 1e-06,
      "loss": 0.5663,
      "num_input_tokens_seen": 295042996,
      "step": 5266
    },
    {
      "epoch": 11.728285077951002,
      "loss": 0.4463956952095032,
      "loss_ce": 0.00010662179556675255,
      "loss_iou": 0.1748046875,
      "loss_num": 0.0191650390625,
      "loss_xval": 0.4453125,
      "num_input_tokens_seen": 295042996,
      "step": 5266
    },
    {
      "epoch": 11.730512249443207,
      "grad_norm": 16.869722366333008,
      "learning_rate": 1e-06,
      "loss": 0.5432,
      "num_input_tokens_seen": 295096444,
      "step": 5267
    },
    {
      "epoch": 11.730512249443207,
      "loss": 0.6898572444915771,
      "loss_ce": 0.00015996204456314445,
      "loss_iou": 0.306640625,
      "loss_num": 0.01531982421875,
      "loss_xval": 0.69140625,
      "num_input_tokens_seen": 295096444,
      "step": 5267
    },
    {
      "epoch": 11.732739420935411,
      "grad_norm": 20.643428802490234,
      "learning_rate": 1e-06,
      "loss": 0.5397,
      "num_input_tokens_seen": 295151264,
      "step": 5268
    },
    {
      "epoch": 11.732739420935411,
      "loss": 0.45317167043685913,
      "loss_ce": 0.00010773130634333938,
      "loss_iou": 0.2060546875,
      "loss_num": 0.00799560546875,
      "loss_xval": 0.453125,
      "num_input_tokens_seen": 295151264,
      "step": 5268
    },
    {
      "epoch": 11.734966592427616,
      "grad_norm": 21.514039993286133,
      "learning_rate": 1e-06,
      "loss": 0.4523,
      "num_input_tokens_seen": 295208360,
      "step": 5269
    },
    {
      "epoch": 11.734966592427616,
      "loss": 0.41239604353904724,
      "loss_ce": 0.0001646070450078696,
      "loss_iou": 0.181640625,
      "loss_num": 0.0096435546875,
      "loss_xval": 0.412109375,
      "num_input_tokens_seen": 295208360,
      "step": 5269
    },
    {
      "epoch": 11.737193763919821,
      "grad_norm": 32.98282241821289,
      "learning_rate": 1e-06,
      "loss": 0.5202,
      "num_input_tokens_seen": 295262584,
      "step": 5270
    },
    {
      "epoch": 11.737193763919821,
      "loss": 0.4672822058200836,
      "loss_ce": 0.00011912808258784935,
      "loss_iou": 0.193359375,
      "loss_num": 0.0162353515625,
      "loss_xval": 0.466796875,
      "num_input_tokens_seen": 295262584,
      "step": 5270
    },
    {
      "epoch": 11.739420935412026,
      "grad_norm": 34.023441314697266,
      "learning_rate": 1e-06,
      "loss": 0.6488,
      "num_input_tokens_seen": 295318340,
      "step": 5271
    },
    {
      "epoch": 11.739420935412026,
      "loss": 0.686636209487915,
      "loss_ce": 0.00011279142927378416,
      "loss_iou": 0.3203125,
      "loss_num": 0.00927734375,
      "loss_xval": 0.6875,
      "num_input_tokens_seen": 295318340,
      "step": 5271
    },
    {
      "epoch": 11.74164810690423,
      "grad_norm": 26.27116584777832,
      "learning_rate": 1e-06,
      "loss": 0.6936,
      "num_input_tokens_seen": 295373900,
      "step": 5272
    },
    {
      "epoch": 11.74164810690423,
      "loss": 0.806282639503479,
      "loss_ce": 0.0001302346063312143,
      "loss_iou": 0.345703125,
      "loss_num": 0.0228271484375,
      "loss_xval": 0.8046875,
      "num_input_tokens_seen": 295373900,
      "step": 5272
    },
    {
      "epoch": 11.743875278396436,
      "grad_norm": 23.026466369628906,
      "learning_rate": 1e-06,
      "loss": 0.3724,
      "num_input_tokens_seen": 295429820,
      "step": 5273
    },
    {
      "epoch": 11.743875278396436,
      "loss": 0.42163270711898804,
      "loss_ce": 0.00012393189535941929,
      "loss_iou": 0.19921875,
      "loss_num": 0.0045166015625,
      "loss_xval": 0.421875,
      "num_input_tokens_seen": 295429820,
      "step": 5273
    },
    {
      "epoch": 11.74610244988864,
      "grad_norm": 22.444780349731445,
      "learning_rate": 1e-06,
      "loss": 0.6569,
      "num_input_tokens_seen": 295488184,
      "step": 5274
    },
    {
      "epoch": 11.74610244988864,
      "loss": 0.6977794766426086,
      "loss_ce": 0.00014764037041459233,
      "loss_iou": 0.298828125,
      "loss_num": 0.0201416015625,
      "loss_xval": 0.69921875,
      "num_input_tokens_seen": 295488184,
      "step": 5274
    },
    {
      "epoch": 11.748329621380847,
      "grad_norm": 13.779067993164062,
      "learning_rate": 1e-06,
      "loss": 0.4753,
      "num_input_tokens_seen": 295545832,
      "step": 5275
    },
    {
      "epoch": 11.748329621380847,
      "loss": 0.5852853059768677,
      "loss_ce": 0.0003243696701247245,
      "loss_iou": 0.26171875,
      "loss_num": 0.01275634765625,
      "loss_xval": 0.5859375,
      "num_input_tokens_seen": 295545832,
      "step": 5275
    },
    {
      "epoch": 11.750556792873052,
      "grad_norm": 28.335033416748047,
      "learning_rate": 1e-06,
      "loss": 0.8353,
      "num_input_tokens_seen": 295602708,
      "step": 5276
    },
    {
      "epoch": 11.750556792873052,
      "loss": 0.6632611155509949,
      "loss_ce": 0.0001752136304276064,
      "loss_iou": 0.283203125,
      "loss_num": 0.019287109375,
      "loss_xval": 0.6640625,
      "num_input_tokens_seen": 295602708,
      "step": 5276
    },
    {
      "epoch": 11.752783964365257,
      "grad_norm": 17.9951114654541,
      "learning_rate": 1e-06,
      "loss": 0.5694,
      "num_input_tokens_seen": 295661544,
      "step": 5277
    },
    {
      "epoch": 11.752783964365257,
      "loss": 0.4410516023635864,
      "loss_ce": 0.00013365020276978612,
      "loss_iou": 0.205078125,
      "loss_num": 0.00604248046875,
      "loss_xval": 0.44140625,
      "num_input_tokens_seen": 295661544,
      "step": 5277
    },
    {
      "epoch": 11.755011135857462,
      "grad_norm": 18.01840591430664,
      "learning_rate": 1e-06,
      "loss": 0.5552,
      "num_input_tokens_seen": 295718028,
      "step": 5278
    },
    {
      "epoch": 11.755011135857462,
      "loss": 0.5536874532699585,
      "loss_ce": 0.00015963352052494884,
      "loss_iou": 0.255859375,
      "loss_num": 0.00823974609375,
      "loss_xval": 0.5546875,
      "num_input_tokens_seen": 295718028,
      "step": 5278
    },
    {
      "epoch": 11.757238307349667,
      "grad_norm": 16.344501495361328,
      "learning_rate": 1e-06,
      "loss": 0.494,
      "num_input_tokens_seen": 295774608,
      "step": 5279
    },
    {
      "epoch": 11.757238307349667,
      "loss": 0.4308076500892639,
      "loss_ce": 0.000143599376315251,
      "loss_iou": 0.1982421875,
      "loss_num": 0.00677490234375,
      "loss_xval": 0.4296875,
      "num_input_tokens_seen": 295774608,
      "step": 5279
    },
    {
      "epoch": 11.759465478841872,
      "grad_norm": 16.951433181762695,
      "learning_rate": 1e-06,
      "loss": 0.4982,
      "num_input_tokens_seen": 295830944,
      "step": 5280
    },
    {
      "epoch": 11.759465478841872,
      "loss": 0.5626462697982788,
      "loss_ce": 0.00014624299365095794,
      "loss_iou": 0.2421875,
      "loss_num": 0.0157470703125,
      "loss_xval": 0.5625,
      "num_input_tokens_seen": 295830944,
      "step": 5280
    },
    {
      "epoch": 11.761692650334076,
      "grad_norm": 16.005573272705078,
      "learning_rate": 1e-06,
      "loss": 0.5351,
      "num_input_tokens_seen": 295887036,
      "step": 5281
    },
    {
      "epoch": 11.761692650334076,
      "loss": 0.3308425545692444,
      "loss_ce": 0.00015408273611683398,
      "loss_iou": 0.140625,
      "loss_num": 0.00982666015625,
      "loss_xval": 0.330078125,
      "num_input_tokens_seen": 295887036,
      "step": 5281
    },
    {
      "epoch": 11.763919821826281,
      "grad_norm": 24.683542251586914,
      "learning_rate": 1e-06,
      "loss": 0.4432,
      "num_input_tokens_seen": 295942760,
      "step": 5282
    },
    {
      "epoch": 11.763919821826281,
      "loss": 0.432136595249176,
      "loss_ce": 0.00012974410492461175,
      "loss_iou": 0.1943359375,
      "loss_num": 0.00860595703125,
      "loss_xval": 0.431640625,
      "num_input_tokens_seen": 295942760,
      "step": 5282
    },
    {
      "epoch": 11.766146993318486,
      "grad_norm": 14.518014907836914,
      "learning_rate": 1e-06,
      "loss": 0.4037,
      "num_input_tokens_seen": 295998632,
      "step": 5283
    },
    {
      "epoch": 11.766146993318486,
      "loss": 0.4797380268573761,
      "loss_ce": 0.00012378576502669603,
      "loss_iou": 0.20703125,
      "loss_num": 0.01300048828125,
      "loss_xval": 0.48046875,
      "num_input_tokens_seen": 295998632,
      "step": 5283
    },
    {
      "epoch": 11.768374164810691,
      "grad_norm": 21.187023162841797,
      "learning_rate": 1e-06,
      "loss": 0.4499,
      "num_input_tokens_seen": 296051864,
      "step": 5284
    },
    {
      "epoch": 11.768374164810691,
      "loss": 0.38384902477264404,
      "loss_ce": 0.00018203401123173535,
      "loss_iou": 0.17578125,
      "loss_num": 0.006195068359375,
      "loss_xval": 0.3828125,
      "num_input_tokens_seen": 296051864,
      "step": 5284
    },
    {
      "epoch": 11.770601336302896,
      "grad_norm": 34.56988525390625,
      "learning_rate": 1e-06,
      "loss": 0.4661,
      "num_input_tokens_seen": 296106776,
      "step": 5285
    },
    {
      "epoch": 11.770601336302896,
      "loss": 0.44788599014282227,
      "loss_ce": 0.00013207312440499663,
      "loss_iou": 0.1962890625,
      "loss_num": 0.010986328125,
      "loss_xval": 0.447265625,
      "num_input_tokens_seen": 296106776,
      "step": 5285
    },
    {
      "epoch": 11.7728285077951,
      "grad_norm": 15.454012870788574,
      "learning_rate": 1e-06,
      "loss": 0.7173,
      "num_input_tokens_seen": 296162332,
      "step": 5286
    },
    {
      "epoch": 11.7728285077951,
      "loss": 0.5844168066978455,
      "loss_ce": 0.0001882594224298373,
      "loss_iou": 0.23828125,
      "loss_num": 0.021728515625,
      "loss_xval": 0.5859375,
      "num_input_tokens_seen": 296162332,
      "step": 5286
    },
    {
      "epoch": 11.775055679287306,
      "grad_norm": 16.151887893676758,
      "learning_rate": 1e-06,
      "loss": 0.5455,
      "num_input_tokens_seen": 296219860,
      "step": 5287
    },
    {
      "epoch": 11.775055679287306,
      "loss": 0.5150243043899536,
      "loss_ce": 0.00013172382023185492,
      "loss_iou": 0.234375,
      "loss_num": 0.00933837890625,
      "loss_xval": 0.515625,
      "num_input_tokens_seen": 296219860,
      "step": 5287
    },
    {
      "epoch": 11.77728285077951,
      "grad_norm": 37.95972442626953,
      "learning_rate": 1e-06,
      "loss": 0.5543,
      "num_input_tokens_seen": 296273744,
      "step": 5288
    },
    {
      "epoch": 11.77728285077951,
      "loss": 0.5362709760665894,
      "loss_ce": 0.00013816248974762857,
      "loss_iou": 0.2333984375,
      "loss_num": 0.01385498046875,
      "loss_xval": 0.53515625,
      "num_input_tokens_seen": 296273744,
      "step": 5288
    },
    {
      "epoch": 11.779510022271715,
      "grad_norm": 30.687088012695312,
      "learning_rate": 1e-06,
      "loss": 0.465,
      "num_input_tokens_seen": 296330476,
      "step": 5289
    },
    {
      "epoch": 11.779510022271715,
      "loss": 0.575337290763855,
      "loss_ce": 0.0001419962791260332,
      "loss_iou": 0.25,
      "loss_num": 0.01495361328125,
      "loss_xval": 0.57421875,
      "num_input_tokens_seen": 296330476,
      "step": 5289
    },
    {
      "epoch": 11.78173719376392,
      "grad_norm": 22.51045799255371,
      "learning_rate": 1e-06,
      "loss": 0.6343,
      "num_input_tokens_seen": 296386848,
      "step": 5290
    },
    {
      "epoch": 11.78173719376392,
      "loss": 0.40053892135620117,
      "loss_ce": 0.00014829869905952364,
      "loss_iou": 0.1728515625,
      "loss_num": 0.01092529296875,
      "loss_xval": 0.400390625,
      "num_input_tokens_seen": 296386848,
      "step": 5290
    },
    {
      "epoch": 11.783964365256125,
      "grad_norm": 17.85065269470215,
      "learning_rate": 1e-06,
      "loss": 0.5055,
      "num_input_tokens_seen": 296442688,
      "step": 5291
    },
    {
      "epoch": 11.783964365256125,
      "loss": 0.36511754989624023,
      "loss_ce": 0.0001273062516702339,
      "loss_iou": 0.1484375,
      "loss_num": 0.013427734375,
      "loss_xval": 0.365234375,
      "num_input_tokens_seen": 296442688,
      "step": 5291
    },
    {
      "epoch": 11.78619153674833,
      "grad_norm": 16.581172943115234,
      "learning_rate": 1e-06,
      "loss": 0.5759,
      "num_input_tokens_seen": 296498192,
      "step": 5292
    },
    {
      "epoch": 11.78619153674833,
      "loss": 0.3235490620136261,
      "loss_ce": 0.0001237650285474956,
      "loss_iou": 0.1416015625,
      "loss_num": 0.008056640625,
      "loss_xval": 0.32421875,
      "num_input_tokens_seen": 296498192,
      "step": 5292
    },
    {
      "epoch": 11.788418708240535,
      "grad_norm": 14.886438369750977,
      "learning_rate": 1e-06,
      "loss": 0.425,
      "num_input_tokens_seen": 296555524,
      "step": 5293
    },
    {
      "epoch": 11.788418708240535,
      "loss": 0.44675683975219727,
      "loss_ce": 0.00016260198026429862,
      "loss_iou": 0.166015625,
      "loss_num": 0.0228271484375,
      "loss_xval": 0.447265625,
      "num_input_tokens_seen": 296555524,
      "step": 5293
    },
    {
      "epoch": 11.79064587973274,
      "grad_norm": 15.52322006225586,
      "learning_rate": 1e-06,
      "loss": 0.4785,
      "num_input_tokens_seen": 296613260,
      "step": 5294
    },
    {
      "epoch": 11.79064587973274,
      "loss": 0.3428999185562134,
      "loss_ce": 0.00012646152754314244,
      "loss_iou": 0.162109375,
      "loss_num": 0.0036163330078125,
      "loss_xval": 0.34375,
      "num_input_tokens_seen": 296613260,
      "step": 5294
    },
    {
      "epoch": 11.792873051224944,
      "grad_norm": 27.934289932250977,
      "learning_rate": 1e-06,
      "loss": 0.502,
      "num_input_tokens_seen": 296667928,
      "step": 5295
    },
    {
      "epoch": 11.792873051224944,
      "loss": 0.5954668521881104,
      "loss_ce": 0.00012995509314350784,
      "loss_iou": 0.25390625,
      "loss_num": 0.017578125,
      "loss_xval": 0.59375,
      "num_input_tokens_seen": 296667928,
      "step": 5295
    },
    {
      "epoch": 11.79510022271715,
      "grad_norm": 18.869325637817383,
      "learning_rate": 1e-06,
      "loss": 0.4571,
      "num_input_tokens_seen": 296723936,
      "step": 5296
    },
    {
      "epoch": 11.79510022271715,
      "loss": 0.5156769752502441,
      "loss_ce": 0.0003877178824041039,
      "loss_iou": 0.216796875,
      "loss_num": 0.0162353515625,
      "loss_xval": 0.515625,
      "num_input_tokens_seen": 296723936,
      "step": 5296
    },
    {
      "epoch": 11.797327394209354,
      "grad_norm": 19.03141975402832,
      "learning_rate": 1e-06,
      "loss": 0.5304,
      "num_input_tokens_seen": 296783132,
      "step": 5297
    },
    {
      "epoch": 11.797327394209354,
      "loss": 0.3304428160190582,
      "loss_ce": 0.00012054571561748162,
      "loss_iou": 0.1474609375,
      "loss_num": 0.0069580078125,
      "loss_xval": 0.330078125,
      "num_input_tokens_seen": 296783132,
      "step": 5297
    },
    {
      "epoch": 11.799554565701559,
      "grad_norm": 20.347238540649414,
      "learning_rate": 1e-06,
      "loss": 0.5689,
      "num_input_tokens_seen": 296839528,
      "step": 5298
    },
    {
      "epoch": 11.799554565701559,
      "loss": 0.5347905158996582,
      "loss_ce": 0.00012258738570380956,
      "loss_iou": 0.2421875,
      "loss_num": 0.01031494140625,
      "loss_xval": 0.53515625,
      "num_input_tokens_seen": 296839528,
      "step": 5298
    },
    {
      "epoch": 11.801781737193764,
      "grad_norm": 18.98991584777832,
      "learning_rate": 1e-06,
      "loss": 0.5361,
      "num_input_tokens_seen": 296899164,
      "step": 5299
    },
    {
      "epoch": 11.801781737193764,
      "loss": 0.6929968595504761,
      "loss_ce": 0.00018683550297282636,
      "loss_iou": 0.283203125,
      "loss_num": 0.025390625,
      "loss_xval": 0.69140625,
      "num_input_tokens_seen": 296899164,
      "step": 5299
    },
    {
      "epoch": 11.804008908685969,
      "grad_norm": 19.457019805908203,
      "learning_rate": 1e-06,
      "loss": 0.5461,
      "num_input_tokens_seen": 296954940,
      "step": 5300
    },
    {
      "epoch": 11.804008908685969,
      "loss": 0.5115100741386414,
      "loss_ce": 0.00015755894128233194,
      "loss_iou": 0.2265625,
      "loss_num": 0.01141357421875,
      "loss_xval": 0.51171875,
      "num_input_tokens_seen": 296954940,
      "step": 5300
    },
    {
      "epoch": 11.806236080178174,
      "grad_norm": 18.224740982055664,
      "learning_rate": 1e-06,
      "loss": 0.5393,
      "num_input_tokens_seen": 297010916,
      "step": 5301
    },
    {
      "epoch": 11.806236080178174,
      "loss": 0.6895302534103394,
      "loss_ce": 0.00019916013116016984,
      "loss_iou": 0.3046875,
      "loss_num": 0.0155029296875,
      "loss_xval": 0.6875,
      "num_input_tokens_seen": 297010916,
      "step": 5301
    },
    {
      "epoch": 11.808463251670378,
      "grad_norm": 22.27739715576172,
      "learning_rate": 1e-06,
      "loss": 0.5199,
      "num_input_tokens_seen": 297066748,
      "step": 5302
    },
    {
      "epoch": 11.808463251670378,
      "loss": 0.4963473081588745,
      "loss_ce": 0.00013145770935807377,
      "loss_iou": 0.2216796875,
      "loss_num": 0.010498046875,
      "loss_xval": 0.49609375,
      "num_input_tokens_seen": 297066748,
      "step": 5302
    },
    {
      "epoch": 11.810690423162583,
      "grad_norm": 13.744184494018555,
      "learning_rate": 1e-06,
      "loss": 0.3914,
      "num_input_tokens_seen": 297123360,
      "step": 5303
    },
    {
      "epoch": 11.810690423162583,
      "loss": 0.32909929752349854,
      "loss_ce": 0.00011983991134911776,
      "loss_iou": 0.146484375,
      "loss_num": 0.00732421875,
      "loss_xval": 0.328125,
      "num_input_tokens_seen": 297123360,
      "step": 5303
    },
    {
      "epoch": 11.812917594654788,
      "grad_norm": 17.198087692260742,
      "learning_rate": 1e-06,
      "loss": 0.6204,
      "num_input_tokens_seen": 297179212,
      "step": 5304
    },
    {
      "epoch": 11.812917594654788,
      "loss": 0.5738571882247925,
      "loss_ce": 0.0001267509942408651,
      "loss_iou": 0.22265625,
      "loss_num": 0.0257568359375,
      "loss_xval": 0.57421875,
      "num_input_tokens_seen": 297179212,
      "step": 5304
    },
    {
      "epoch": 11.815144766146993,
      "grad_norm": 21.798635482788086,
      "learning_rate": 1e-06,
      "loss": 0.4382,
      "num_input_tokens_seen": 297237840,
      "step": 5305
    },
    {
      "epoch": 11.815144766146993,
      "loss": 0.5149092674255371,
      "loss_ce": 0.00013874081196263433,
      "loss_iou": 0.2353515625,
      "loss_num": 0.0087890625,
      "loss_xval": 0.515625,
      "num_input_tokens_seen": 297237840,
      "step": 5305
    },
    {
      "epoch": 11.817371937639198,
      "grad_norm": 15.751426696777344,
      "learning_rate": 1e-06,
      "loss": 0.5304,
      "num_input_tokens_seen": 297295812,
      "step": 5306
    },
    {
      "epoch": 11.817371937639198,
      "loss": 0.5815914869308472,
      "loss_ce": 0.0001705837930785492,
      "loss_iou": 0.240234375,
      "loss_num": 0.020263671875,
      "loss_xval": 0.58203125,
      "num_input_tokens_seen": 297295812,
      "step": 5306
    },
    {
      "epoch": 11.819599109131403,
      "grad_norm": 30.267009735107422,
      "learning_rate": 1e-06,
      "loss": 0.6813,
      "num_input_tokens_seen": 297351380,
      "step": 5307
    },
    {
      "epoch": 11.819599109131403,
      "loss": 0.7589607238769531,
      "loss_ce": 0.000171684252563864,
      "loss_iou": 0.3046875,
      "loss_num": 0.0302734375,
      "loss_xval": 0.7578125,
      "num_input_tokens_seen": 297351380,
      "step": 5307
    },
    {
      "epoch": 11.821826280623608,
      "grad_norm": 16.236085891723633,
      "learning_rate": 1e-06,
      "loss": 0.5397,
      "num_input_tokens_seen": 297406340,
      "step": 5308
    },
    {
      "epoch": 11.821826280623608,
      "loss": 0.5382388830184937,
      "loss_ce": 0.00015292633906938136,
      "loss_iou": 0.2412109375,
      "loss_num": 0.01141357421875,
      "loss_xval": 0.5390625,
      "num_input_tokens_seen": 297406340,
      "step": 5308
    },
    {
      "epoch": 11.824053452115812,
      "grad_norm": 15.723315238952637,
      "learning_rate": 1e-06,
      "loss": 0.5465,
      "num_input_tokens_seen": 297464004,
      "step": 5309
    },
    {
      "epoch": 11.824053452115812,
      "loss": 0.6881057024002075,
      "loss_ce": 0.00011742223432520404,
      "loss_iou": 0.2890625,
      "loss_num": 0.021728515625,
      "loss_xval": 0.6875,
      "num_input_tokens_seen": 297464004,
      "step": 5309
    },
    {
      "epoch": 11.826280623608017,
      "grad_norm": 26.35066795349121,
      "learning_rate": 1e-06,
      "loss": 0.4958,
      "num_input_tokens_seen": 297520084,
      "step": 5310
    },
    {
      "epoch": 11.826280623608017,
      "loss": 0.4250958263874054,
      "loss_ce": 0.00016907165991142392,
      "loss_iou": 0.1748046875,
      "loss_num": 0.01495361328125,
      "loss_xval": 0.42578125,
      "num_input_tokens_seen": 297520084,
      "step": 5310
    },
    {
      "epoch": 11.828507795100222,
      "grad_norm": 17.730464935302734,
      "learning_rate": 1e-06,
      "loss": 0.6199,
      "num_input_tokens_seen": 297575516,
      "step": 5311
    },
    {
      "epoch": 11.828507795100222,
      "loss": 0.7365161180496216,
      "loss_ce": 0.0008594049722887576,
      "loss_iou": 0.291015625,
      "loss_num": 0.0306396484375,
      "loss_xval": 0.734375,
      "num_input_tokens_seen": 297575516,
      "step": 5311
    },
    {
      "epoch": 11.830734966592427,
      "grad_norm": 23.806507110595703,
      "learning_rate": 1e-06,
      "loss": 0.5492,
      "num_input_tokens_seen": 297631068,
      "step": 5312
    },
    {
      "epoch": 11.830734966592427,
      "loss": 0.6309938430786133,
      "loss_ce": 0.00013447852688841522,
      "loss_iou": 0.275390625,
      "loss_num": 0.016357421875,
      "loss_xval": 0.6328125,
      "num_input_tokens_seen": 297631068,
      "step": 5312
    },
    {
      "epoch": 11.832962138084632,
      "grad_norm": 14.093077659606934,
      "learning_rate": 1e-06,
      "loss": 0.4271,
      "num_input_tokens_seen": 297685616,
      "step": 5313
    },
    {
      "epoch": 11.832962138084632,
      "loss": 0.23368564248085022,
      "loss_ce": 0.00019564517424441874,
      "loss_iou": 0.10205078125,
      "loss_num": 0.00579833984375,
      "loss_xval": 0.2333984375,
      "num_input_tokens_seen": 297685616,
      "step": 5313
    },
    {
      "epoch": 11.835189309576837,
      "grad_norm": 22.90943717956543,
      "learning_rate": 1e-06,
      "loss": 0.4905,
      "num_input_tokens_seen": 297743540,
      "step": 5314
    },
    {
      "epoch": 11.835189309576837,
      "loss": 0.39391031861305237,
      "loss_ce": 0.00011150065256515518,
      "loss_iou": 0.166015625,
      "loss_num": 0.01239013671875,
      "loss_xval": 0.39453125,
      "num_input_tokens_seen": 297743540,
      "step": 5314
    },
    {
      "epoch": 11.837416481069042,
      "grad_norm": 16.595645904541016,
      "learning_rate": 1e-06,
      "loss": 0.4073,
      "num_input_tokens_seen": 297801180,
      "step": 5315
    },
    {
      "epoch": 11.837416481069042,
      "loss": 0.6017987728118896,
      "loss_ce": 0.00011419894872233272,
      "loss_iou": 0.267578125,
      "loss_num": 0.0135498046875,
      "loss_xval": 0.6015625,
      "num_input_tokens_seen": 297801180,
      "step": 5315
    },
    {
      "epoch": 11.839643652561247,
      "grad_norm": 17.16973876953125,
      "learning_rate": 1e-06,
      "loss": 0.4081,
      "num_input_tokens_seen": 297858728,
      "step": 5316
    },
    {
      "epoch": 11.839643652561247,
      "loss": 0.3546229302883148,
      "loss_ce": 0.00013074639718979597,
      "loss_iou": 0.162109375,
      "loss_num": 0.006072998046875,
      "loss_xval": 0.35546875,
      "num_input_tokens_seen": 297858728,
      "step": 5316
    },
    {
      "epoch": 11.841870824053451,
      "grad_norm": 18.856740951538086,
      "learning_rate": 1e-06,
      "loss": 0.318,
      "num_input_tokens_seen": 297915964,
      "step": 5317
    },
    {
      "epoch": 11.841870824053451,
      "loss": 0.23262880742549896,
      "loss_ce": 0.0001001132040983066,
      "loss_iou": 0.10009765625,
      "loss_num": 0.006439208984375,
      "loss_xval": 0.232421875,
      "num_input_tokens_seen": 297915964,
      "step": 5317
    },
    {
      "epoch": 11.844097995545656,
      "grad_norm": 16.652254104614258,
      "learning_rate": 1e-06,
      "loss": 0.5737,
      "num_input_tokens_seen": 297971460,
      "step": 5318
    },
    {
      "epoch": 11.844097995545656,
      "loss": 0.6619553565979004,
      "loss_ce": 0.00033427210291847587,
      "loss_iou": 0.27734375,
      "loss_num": 0.021484375,
      "loss_xval": 0.66015625,
      "num_input_tokens_seen": 297971460,
      "step": 5318
    },
    {
      "epoch": 11.846325167037861,
      "grad_norm": 27.506261825561523,
      "learning_rate": 1e-06,
      "loss": 0.4867,
      "num_input_tokens_seen": 298027772,
      "step": 5319
    },
    {
      "epoch": 11.846325167037861,
      "loss": 0.42390304803848267,
      "loss_ce": 0.00013593377661891282,
      "loss_iou": 0.1845703125,
      "loss_num": 0.0111083984375,
      "loss_xval": 0.423828125,
      "num_input_tokens_seen": 298027772,
      "step": 5319
    },
    {
      "epoch": 11.848552338530066,
      "grad_norm": 22.7269229888916,
      "learning_rate": 1e-06,
      "loss": 0.4718,
      "num_input_tokens_seen": 298084460,
      "step": 5320
    },
    {
      "epoch": 11.848552338530066,
      "loss": 0.469988614320755,
      "loss_ce": 0.0007503442466259003,
      "loss_iou": 0.2001953125,
      "loss_num": 0.01397705078125,
      "loss_xval": 0.46875,
      "num_input_tokens_seen": 298084460,
      "step": 5320
    },
    {
      "epoch": 11.85077951002227,
      "grad_norm": 18.758642196655273,
      "learning_rate": 1e-06,
      "loss": 0.4439,
      "num_input_tokens_seen": 298138912,
      "step": 5321
    },
    {
      "epoch": 11.85077951002227,
      "loss": 0.5238094925880432,
      "loss_ce": 0.00012785423314198852,
      "loss_iou": 0.220703125,
      "loss_num": 0.016357421875,
      "loss_xval": 0.5234375,
      "num_input_tokens_seen": 298138912,
      "step": 5321
    },
    {
      "epoch": 11.853006681514476,
      "grad_norm": 20.615978240966797,
      "learning_rate": 1e-06,
      "loss": 0.5517,
      "num_input_tokens_seen": 298193028,
      "step": 5322
    },
    {
      "epoch": 11.853006681514476,
      "loss": 0.5698904991149902,
      "loss_ce": 0.00018834380898624659,
      "loss_iou": 0.2431640625,
      "loss_num": 0.0166015625,
      "loss_xval": 0.5703125,
      "num_input_tokens_seen": 298193028,
      "step": 5322
    },
    {
      "epoch": 11.855233853006682,
      "grad_norm": 16.021848678588867,
      "learning_rate": 1e-06,
      "loss": 0.342,
      "num_input_tokens_seen": 298248720,
      "step": 5323
    },
    {
      "epoch": 11.855233853006682,
      "loss": 0.24690712988376617,
      "loss_ce": 0.00014198827557265759,
      "loss_iou": 0.10693359375,
      "loss_num": 0.00665283203125,
      "loss_xval": 0.2470703125,
      "num_input_tokens_seen": 298248720,
      "step": 5323
    },
    {
      "epoch": 11.857461024498887,
      "grad_norm": 23.891801834106445,
      "learning_rate": 1e-06,
      "loss": 0.561,
      "num_input_tokens_seen": 298303808,
      "step": 5324
    },
    {
      "epoch": 11.857461024498887,
      "loss": 0.6983004808425903,
      "loss_ce": 0.000302450789604336,
      "loss_iou": 0.255859375,
      "loss_num": 0.037109375,
      "loss_xval": 0.69921875,
      "num_input_tokens_seen": 298303808,
      "step": 5324
    },
    {
      "epoch": 11.859688195991092,
      "grad_norm": 18.63146209716797,
      "learning_rate": 1e-06,
      "loss": 0.414,
      "num_input_tokens_seen": 298358804,
      "step": 5325
    },
    {
      "epoch": 11.859688195991092,
      "loss": 0.35854852199554443,
      "loss_ce": 0.00015006760077085346,
      "loss_iou": 0.1572265625,
      "loss_num": 0.00897216796875,
      "loss_xval": 0.359375,
      "num_input_tokens_seen": 298358804,
      "step": 5325
    },
    {
      "epoch": 11.861915367483297,
      "grad_norm": 19.940717697143555,
      "learning_rate": 1e-06,
      "loss": 0.4274,
      "num_input_tokens_seen": 298418072,
      "step": 5326
    },
    {
      "epoch": 11.861915367483297,
      "loss": 0.2874617576599121,
      "loss_ce": 0.00010823093907674775,
      "loss_iou": 0.130859375,
      "loss_num": 0.00494384765625,
      "loss_xval": 0.287109375,
      "num_input_tokens_seen": 298418072,
      "step": 5326
    },
    {
      "epoch": 11.864142538975502,
      "grad_norm": 23.029714584350586,
      "learning_rate": 1e-06,
      "loss": 0.4766,
      "num_input_tokens_seen": 298476076,
      "step": 5327
    },
    {
      "epoch": 11.864142538975502,
      "loss": 0.5553493499755859,
      "loss_ce": 0.00017354739247821271,
      "loss_iou": 0.2392578125,
      "loss_num": 0.0155029296875,
      "loss_xval": 0.5546875,
      "num_input_tokens_seen": 298476076,
      "step": 5327
    },
    {
      "epoch": 11.866369710467707,
      "grad_norm": 13.412109375,
      "learning_rate": 1e-06,
      "loss": 0.4057,
      "num_input_tokens_seen": 298532080,
      "step": 5328
    },
    {
      "epoch": 11.866369710467707,
      "loss": 0.3781711161136627,
      "loss_ce": 0.00011935001384699717,
      "loss_iou": 0.1494140625,
      "loss_num": 0.0157470703125,
      "loss_xval": 0.37890625,
      "num_input_tokens_seen": 298532080,
      "step": 5328
    },
    {
      "epoch": 11.868596881959911,
      "grad_norm": 22.732973098754883,
      "learning_rate": 1e-06,
      "loss": 0.5291,
      "num_input_tokens_seen": 298585788,
      "step": 5329
    },
    {
      "epoch": 11.868596881959911,
      "loss": 0.5079842209815979,
      "loss_ce": 0.00017171379295177758,
      "loss_iou": 0.2314453125,
      "loss_num": 0.00885009765625,
      "loss_xval": 0.5078125,
      "num_input_tokens_seen": 298585788,
      "step": 5329
    },
    {
      "epoch": 11.870824053452116,
      "grad_norm": 22.004831314086914,
      "learning_rate": 1e-06,
      "loss": 0.555,
      "num_input_tokens_seen": 298641296,
      "step": 5330
    },
    {
      "epoch": 11.870824053452116,
      "loss": 0.7135499715805054,
      "loss_ce": 0.00017109722830355167,
      "loss_iou": 0.330078125,
      "loss_num": 0.01043701171875,
      "loss_xval": 0.71484375,
      "num_input_tokens_seen": 298641296,
      "step": 5330
    },
    {
      "epoch": 11.873051224944321,
      "grad_norm": 33.82167434692383,
      "learning_rate": 1e-06,
      "loss": 0.7149,
      "num_input_tokens_seen": 298697308,
      "step": 5331
    },
    {
      "epoch": 11.873051224944321,
      "loss": 0.9743344187736511,
      "loss_ce": 0.0002132941735908389,
      "loss_iou": 0.4296875,
      "loss_num": 0.0225830078125,
      "loss_xval": 0.97265625,
      "num_input_tokens_seen": 298697308,
      "step": 5331
    },
    {
      "epoch": 11.875278396436526,
      "grad_norm": 23.670555114746094,
      "learning_rate": 1e-06,
      "loss": 0.5041,
      "num_input_tokens_seen": 298751336,
      "step": 5332
    },
    {
      "epoch": 11.875278396436526,
      "loss": 0.5620195865631104,
      "loss_ce": 0.00012994898133911192,
      "loss_iou": 0.26171875,
      "loss_num": 0.007568359375,
      "loss_xval": 0.5625,
      "num_input_tokens_seen": 298751336,
      "step": 5332
    },
    {
      "epoch": 11.877505567928731,
      "grad_norm": 19.82380485534668,
      "learning_rate": 1e-06,
      "loss": 0.6098,
      "num_input_tokens_seen": 298808312,
      "step": 5333
    },
    {
      "epoch": 11.877505567928731,
      "loss": 0.33348560333251953,
      "loss_ce": 0.00011158882989548147,
      "loss_iou": 0.1416015625,
      "loss_num": 0.0101318359375,
      "loss_xval": 0.333984375,
      "num_input_tokens_seen": 298808312,
      "step": 5333
    },
    {
      "epoch": 11.879732739420936,
      "grad_norm": 13.22695541381836,
      "learning_rate": 1e-06,
      "loss": 0.3795,
      "num_input_tokens_seen": 298864232,
      "step": 5334
    },
    {
      "epoch": 11.879732739420936,
      "loss": 0.37341421842575073,
      "loss_ce": 0.00012321470421738923,
      "loss_iou": 0.146484375,
      "loss_num": 0.0162353515625,
      "loss_xval": 0.373046875,
      "num_input_tokens_seen": 298864232,
      "step": 5334
    },
    {
      "epoch": 11.88195991091314,
      "grad_norm": 19.371503829956055,
      "learning_rate": 1e-06,
      "loss": 0.5704,
      "num_input_tokens_seen": 298919880,
      "step": 5335
    },
    {
      "epoch": 11.88195991091314,
      "loss": 0.5620713829994202,
      "loss_ce": 0.00012070624507032335,
      "loss_iou": 0.2392578125,
      "loss_num": 0.0166015625,
      "loss_xval": 0.5625,
      "num_input_tokens_seen": 298919880,
      "step": 5335
    },
    {
      "epoch": 11.884187082405345,
      "grad_norm": 20.38959503173828,
      "learning_rate": 1e-06,
      "loss": 0.4472,
      "num_input_tokens_seen": 298971852,
      "step": 5336
    },
    {
      "epoch": 11.884187082405345,
      "loss": 0.3908699154853821,
      "loss_ce": 0.0001228695473400876,
      "loss_iou": 0.17578125,
      "loss_num": 0.00762939453125,
      "loss_xval": 0.390625,
      "num_input_tokens_seen": 298971852,
      "step": 5336
    },
    {
      "epoch": 11.88641425389755,
      "grad_norm": 57.669029235839844,
      "learning_rate": 1e-06,
      "loss": 0.6538,
      "num_input_tokens_seen": 299029644,
      "step": 5337
    },
    {
      "epoch": 11.88641425389755,
      "loss": 0.6255764365196228,
      "loss_ce": 0.0011867830762639642,
      "loss_iou": 0.265625,
      "loss_num": 0.018798828125,
      "loss_xval": 0.625,
      "num_input_tokens_seen": 299029644,
      "step": 5337
    },
    {
      "epoch": 11.888641425389755,
      "grad_norm": 18.04958724975586,
      "learning_rate": 1e-06,
      "loss": 0.306,
      "num_input_tokens_seen": 299085212,
      "step": 5338
    },
    {
      "epoch": 11.888641425389755,
      "loss": 0.3291063904762268,
      "loss_ce": 0.00012687721755355597,
      "loss_iou": 0.1552734375,
      "loss_num": 0.00384521484375,
      "loss_xval": 0.328125,
      "num_input_tokens_seen": 299085212,
      "step": 5338
    },
    {
      "epoch": 11.89086859688196,
      "grad_norm": 26.414382934570312,
      "learning_rate": 1e-06,
      "loss": 0.6099,
      "num_input_tokens_seen": 299142868,
      "step": 5339
    },
    {
      "epoch": 11.89086859688196,
      "loss": 0.7533276081085205,
      "loss_ce": 0.0001537687494419515,
      "loss_iou": 0.30078125,
      "loss_num": 0.030517578125,
      "loss_xval": 0.75390625,
      "num_input_tokens_seen": 299142868,
      "step": 5339
    },
    {
      "epoch": 11.893095768374165,
      "grad_norm": 53.97526168823242,
      "learning_rate": 1e-06,
      "loss": 0.6077,
      "num_input_tokens_seen": 299200428,
      "step": 5340
    },
    {
      "epoch": 11.893095768374165,
      "loss": 0.7335488796234131,
      "loss_ce": 0.0001504583196947351,
      "loss_iou": 0.3046875,
      "loss_num": 0.024658203125,
      "loss_xval": 0.734375,
      "num_input_tokens_seen": 299200428,
      "step": 5340
    },
    {
      "epoch": 11.89532293986637,
      "grad_norm": 20.818517684936523,
      "learning_rate": 1e-06,
      "loss": 0.4193,
      "num_input_tokens_seen": 299255160,
      "step": 5341
    },
    {
      "epoch": 11.89532293986637,
      "loss": 0.4669525623321533,
      "loss_ce": 0.0001404241193085909,
      "loss_iou": 0.20703125,
      "loss_num": 0.0103759765625,
      "loss_xval": 0.466796875,
      "num_input_tokens_seen": 299255160,
      "step": 5341
    },
    {
      "epoch": 11.897550111358575,
      "grad_norm": 24.498870849609375,
      "learning_rate": 1e-06,
      "loss": 0.5564,
      "num_input_tokens_seen": 299309776,
      "step": 5342
    },
    {
      "epoch": 11.897550111358575,
      "loss": 0.5995131731033325,
      "loss_ce": 0.00014790653949603438,
      "loss_iou": 0.26953125,
      "loss_num": 0.01202392578125,
      "loss_xval": 0.59765625,
      "num_input_tokens_seen": 299309776,
      "step": 5342
    },
    {
      "epoch": 11.89977728285078,
      "grad_norm": 23.035675048828125,
      "learning_rate": 1e-06,
      "loss": 0.5478,
      "num_input_tokens_seen": 299366904,
      "step": 5343
    },
    {
      "epoch": 11.89977728285078,
      "loss": 0.5480039715766907,
      "loss_ce": 0.00015243007510434836,
      "loss_iou": 0.228515625,
      "loss_num": 0.018310546875,
      "loss_xval": 0.546875,
      "num_input_tokens_seen": 299366904,
      "step": 5343
    },
    {
      "epoch": 11.902004454342984,
      "grad_norm": 25.75360679626465,
      "learning_rate": 1e-06,
      "loss": 0.526,
      "num_input_tokens_seen": 299421724,
      "step": 5344
    },
    {
      "epoch": 11.902004454342984,
      "loss": 0.39622652530670166,
      "loss_ce": 0.00010832876432687044,
      "loss_iou": 0.17578125,
      "loss_num": 0.00885009765625,
      "loss_xval": 0.396484375,
      "num_input_tokens_seen": 299421724,
      "step": 5344
    },
    {
      "epoch": 11.90423162583519,
      "grad_norm": 15.133755683898926,
      "learning_rate": 1e-06,
      "loss": 0.4379,
      "num_input_tokens_seen": 299478644,
      "step": 5345
    },
    {
      "epoch": 11.90423162583519,
      "loss": 0.515461802482605,
      "loss_ce": 0.00020301563199609518,
      "loss_iou": 0.224609375,
      "loss_num": 0.0130615234375,
      "loss_xval": 0.515625,
      "num_input_tokens_seen": 299478644,
      "step": 5345
    },
    {
      "epoch": 11.906458797327394,
      "grad_norm": 14.36051082611084,
      "learning_rate": 1e-06,
      "loss": 0.4898,
      "num_input_tokens_seen": 299534612,
      "step": 5346
    },
    {
      "epoch": 11.906458797327394,
      "loss": 0.5628827810287476,
      "loss_ce": 0.00013867080269847065,
      "loss_iou": 0.255859375,
      "loss_num": 0.01025390625,
      "loss_xval": 0.5625,
      "num_input_tokens_seen": 299534612,
      "step": 5346
    },
    {
      "epoch": 11.908685968819599,
      "grad_norm": 15.240983009338379,
      "learning_rate": 1e-06,
      "loss": 0.5047,
      "num_input_tokens_seen": 299590416,
      "step": 5347
    },
    {
      "epoch": 11.908685968819599,
      "loss": 0.558215856552124,
      "loss_ce": 0.00011038097727578133,
      "loss_iou": 0.2490234375,
      "loss_num": 0.01202392578125,
      "loss_xval": 0.55859375,
      "num_input_tokens_seen": 299590416,
      "step": 5347
    },
    {
      "epoch": 11.910913140311804,
      "grad_norm": 27.012542724609375,
      "learning_rate": 1e-06,
      "loss": 0.6381,
      "num_input_tokens_seen": 299642864,
      "step": 5348
    },
    {
      "epoch": 11.910913140311804,
      "loss": 0.6602959632873535,
      "loss_ce": 0.00013973098248243332,
      "loss_iou": 0.28515625,
      "loss_num": 0.017822265625,
      "loss_xval": 0.66015625,
      "num_input_tokens_seen": 299642864,
      "step": 5348
    },
    {
      "epoch": 11.913140311804009,
      "grad_norm": 23.465747833251953,
      "learning_rate": 1e-06,
      "loss": 0.6058,
      "num_input_tokens_seen": 299696992,
      "step": 5349
    },
    {
      "epoch": 11.913140311804009,
      "loss": 0.6304618716239929,
      "loss_ce": 0.00033489661291241646,
      "loss_iou": 0.279296875,
      "loss_num": 0.01397705078125,
      "loss_xval": 0.62890625,
      "num_input_tokens_seen": 299696992,
      "step": 5349
    },
    {
      "epoch": 11.915367483296214,
      "grad_norm": 18.98960304260254,
      "learning_rate": 1e-06,
      "loss": 0.5868,
      "num_input_tokens_seen": 299753692,
      "step": 5350
    },
    {
      "epoch": 11.915367483296214,
      "loss": 0.6957433223724365,
      "loss_ce": 0.00018664819072000682,
      "loss_iou": 0.29296875,
      "loss_num": 0.021728515625,
      "loss_xval": 0.6953125,
      "num_input_tokens_seen": 299753692,
      "step": 5350
    },
    {
      "epoch": 11.917594654788418,
      "grad_norm": 18.785682678222656,
      "learning_rate": 1e-06,
      "loss": 0.5032,
      "num_input_tokens_seen": 299807860,
      "step": 5351
    },
    {
      "epoch": 11.917594654788418,
      "loss": 0.48331549763679504,
      "loss_ce": 0.00010016474698204547,
      "loss_iou": 0.2060546875,
      "loss_num": 0.01409912109375,
      "loss_xval": 0.482421875,
      "num_input_tokens_seen": 299807860,
      "step": 5351
    },
    {
      "epoch": 11.919821826280623,
      "grad_norm": 17.613218307495117,
      "learning_rate": 1e-06,
      "loss": 0.6931,
      "num_input_tokens_seen": 299864664,
      "step": 5352
    },
    {
      "epoch": 11.919821826280623,
      "loss": 0.5709285736083984,
      "loss_ce": 0.00012783391866832972,
      "loss_iou": 0.2392578125,
      "loss_num": 0.0184326171875,
      "loss_xval": 0.5703125,
      "num_input_tokens_seen": 299864664,
      "step": 5352
    },
    {
      "epoch": 11.922048997772828,
      "grad_norm": 27.25409698486328,
      "learning_rate": 1e-06,
      "loss": 0.5266,
      "num_input_tokens_seen": 299919748,
      "step": 5353
    },
    {
      "epoch": 11.922048997772828,
      "loss": 0.41632139682769775,
      "loss_ce": 0.00030577427241951227,
      "loss_iou": 0.1806640625,
      "loss_num": 0.0108642578125,
      "loss_xval": 0.416015625,
      "num_input_tokens_seen": 299919748,
      "step": 5353
    },
    {
      "epoch": 11.924276169265033,
      "grad_norm": 23.497005462646484,
      "learning_rate": 1e-06,
      "loss": 0.4849,
      "num_input_tokens_seen": 299977172,
      "step": 5354
    },
    {
      "epoch": 11.924276169265033,
      "loss": 0.5234737396240234,
      "loss_ce": 0.00015833397628739476,
      "loss_iou": 0.2314453125,
      "loss_num": 0.01214599609375,
      "loss_xval": 0.5234375,
      "num_input_tokens_seen": 299977172,
      "step": 5354
    },
    {
      "epoch": 11.926503340757238,
      "grad_norm": 24.323163986206055,
      "learning_rate": 1e-06,
      "loss": 0.4479,
      "num_input_tokens_seen": 300033976,
      "step": 5355
    },
    {
      "epoch": 11.926503340757238,
      "loss": 0.4954897165298462,
      "loss_ce": 0.00025045976508408785,
      "loss_iou": 0.2216796875,
      "loss_num": 0.0103759765625,
      "loss_xval": 0.49609375,
      "num_input_tokens_seen": 300033976,
      "step": 5355
    },
    {
      "epoch": 11.928730512249443,
      "grad_norm": 12.777032852172852,
      "learning_rate": 1e-06,
      "loss": 0.3908,
      "num_input_tokens_seen": 300089880,
      "step": 5356
    },
    {
      "epoch": 11.928730512249443,
      "loss": 0.38708722591400146,
      "loss_ce": 0.00012434215750545263,
      "loss_iou": 0.1708984375,
      "loss_num": 0.00921630859375,
      "loss_xval": 0.38671875,
      "num_input_tokens_seen": 300089880,
      "step": 5356
    },
    {
      "epoch": 11.930957683741648,
      "grad_norm": 18.82436180114746,
      "learning_rate": 1e-06,
      "loss": 0.3538,
      "num_input_tokens_seen": 300144144,
      "step": 5357
    },
    {
      "epoch": 11.930957683741648,
      "loss": 0.4027044475078583,
      "loss_ce": 0.00011656976130325347,
      "loss_iou": 0.1796875,
      "loss_num": 0.00848388671875,
      "loss_xval": 0.40234375,
      "num_input_tokens_seen": 300144144,
      "step": 5357
    },
    {
      "epoch": 11.933184855233852,
      "grad_norm": 15.833600044250488,
      "learning_rate": 1e-06,
      "loss": 0.3947,
      "num_input_tokens_seen": 300199272,
      "step": 5358
    },
    {
      "epoch": 11.933184855233852,
      "loss": 0.4046657085418701,
      "loss_ce": 0.00012470208457671106,
      "loss_iou": 0.1767578125,
      "loss_num": 0.0103759765625,
      "loss_xval": 0.404296875,
      "num_input_tokens_seen": 300199272,
      "step": 5358
    },
    {
      "epoch": 11.935412026726057,
      "grad_norm": 19.36171531677246,
      "learning_rate": 1e-06,
      "loss": 0.7569,
      "num_input_tokens_seen": 300253388,
      "step": 5359
    },
    {
      "epoch": 11.935412026726057,
      "loss": 0.6293305158615112,
      "loss_ce": 0.00018012213695328683,
      "loss_iou": 0.28125,
      "loss_num": 0.01336669921875,
      "loss_xval": 0.62890625,
      "num_input_tokens_seen": 300253388,
      "step": 5359
    },
    {
      "epoch": 11.937639198218262,
      "grad_norm": 26.180601119995117,
      "learning_rate": 1e-06,
      "loss": 0.5059,
      "num_input_tokens_seen": 300307808,
      "step": 5360
    },
    {
      "epoch": 11.937639198218262,
      "loss": 0.5130758881568909,
      "loss_ce": 0.0001364394265692681,
      "loss_iou": 0.2373046875,
      "loss_num": 0.007659912109375,
      "loss_xval": 0.51171875,
      "num_input_tokens_seen": 300307808,
      "step": 5360
    },
    {
      "epoch": 11.939866369710467,
      "grad_norm": 15.848928451538086,
      "learning_rate": 1e-06,
      "loss": 0.523,
      "num_input_tokens_seen": 300363860,
      "step": 5361
    },
    {
      "epoch": 11.939866369710467,
      "loss": 0.4611893892288208,
      "loss_ce": 0.0001298237475566566,
      "loss_iou": 0.201171875,
      "loss_num": 0.01153564453125,
      "loss_xval": 0.4609375,
      "num_input_tokens_seen": 300363860,
      "step": 5361
    },
    {
      "epoch": 11.942093541202672,
      "grad_norm": 17.287405014038086,
      "learning_rate": 1e-06,
      "loss": 0.5096,
      "num_input_tokens_seen": 300421080,
      "step": 5362
    },
    {
      "epoch": 11.942093541202672,
      "loss": 0.33068329095840454,
      "loss_ce": 0.00011686367361107841,
      "loss_iou": 0.1376953125,
      "loss_num": 0.01092529296875,
      "loss_xval": 0.330078125,
      "num_input_tokens_seen": 300421080,
      "step": 5362
    },
    {
      "epoch": 11.944320712694877,
      "grad_norm": 18.37860870361328,
      "learning_rate": 1e-06,
      "loss": 0.5593,
      "num_input_tokens_seen": 300477264,
      "step": 5363
    },
    {
      "epoch": 11.944320712694877,
      "loss": 0.5681136846542358,
      "loss_ce": 0.00024256901815533638,
      "loss_iou": 0.25,
      "loss_num": 0.01318359375,
      "loss_xval": 0.56640625,
      "num_input_tokens_seen": 300477264,
      "step": 5363
    },
    {
      "epoch": 11.946547884187082,
      "grad_norm": 23.356395721435547,
      "learning_rate": 1e-06,
      "loss": 0.4799,
      "num_input_tokens_seen": 300535224,
      "step": 5364
    },
    {
      "epoch": 11.946547884187082,
      "loss": 0.4557165503501892,
      "loss_ce": 0.00015014578821137547,
      "loss_iou": 0.19921875,
      "loss_num": 0.01165771484375,
      "loss_xval": 0.455078125,
      "num_input_tokens_seen": 300535224,
      "step": 5364
    },
    {
      "epoch": 11.948775055679288,
      "grad_norm": 23.67485809326172,
      "learning_rate": 1e-06,
      "loss": 0.7032,
      "num_input_tokens_seen": 300589224,
      "step": 5365
    },
    {
      "epoch": 11.948775055679288,
      "loss": 0.7572053670883179,
      "loss_ce": 0.00012526212958618999,
      "loss_iou": 0.298828125,
      "loss_num": 0.031494140625,
      "loss_xval": 0.7578125,
      "num_input_tokens_seen": 300589224,
      "step": 5365
    },
    {
      "epoch": 11.951002227171493,
      "grad_norm": 18.73485565185547,
      "learning_rate": 1e-06,
      "loss": 0.7542,
      "num_input_tokens_seen": 300643164,
      "step": 5366
    },
    {
      "epoch": 11.951002227171493,
      "loss": 0.9482482075691223,
      "loss_ce": 0.0002501863054931164,
      "loss_iou": 0.357421875,
      "loss_num": 0.04638671875,
      "loss_xval": 0.94921875,
      "num_input_tokens_seen": 300643164,
      "step": 5366
    },
    {
      "epoch": 11.953229398663698,
      "grad_norm": 22.971054077148438,
      "learning_rate": 1e-06,
      "loss": 0.4646,
      "num_input_tokens_seen": 300700556,
      "step": 5367
    },
    {
      "epoch": 11.953229398663698,
      "loss": 0.35095998644828796,
      "loss_ce": 0.0001299169525736943,
      "loss_iou": 0.1630859375,
      "loss_num": 0.004913330078125,
      "loss_xval": 0.3515625,
      "num_input_tokens_seen": 300700556,
      "step": 5367
    },
    {
      "epoch": 11.955456570155903,
      "grad_norm": 20.057994842529297,
      "learning_rate": 1e-06,
      "loss": 0.4387,
      "num_input_tokens_seen": 300755556,
      "step": 5368
    },
    {
      "epoch": 11.955456570155903,
      "loss": 0.41150492429733276,
      "loss_ce": 0.00012799599790014327,
      "loss_iou": 0.18359375,
      "loss_num": 0.00885009765625,
      "loss_xval": 0.412109375,
      "num_input_tokens_seen": 300755556,
      "step": 5368
    },
    {
      "epoch": 11.957683741648108,
      "grad_norm": 17.977746963500977,
      "learning_rate": 1e-06,
      "loss": 0.5373,
      "num_input_tokens_seen": 300809448,
      "step": 5369
    },
    {
      "epoch": 11.957683741648108,
      "loss": 0.5350706577301025,
      "loss_ce": 0.00015853876539040357,
      "loss_iou": 0.248046875,
      "loss_num": 0.007659912109375,
      "loss_xval": 0.53515625,
      "num_input_tokens_seen": 300809448,
      "step": 5369
    },
    {
      "epoch": 11.959910913140313,
      "grad_norm": 31.153818130493164,
      "learning_rate": 1e-06,
      "loss": 0.4172,
      "num_input_tokens_seen": 300863900,
      "step": 5370
    },
    {
      "epoch": 11.959910913140313,
      "loss": 0.4988964796066284,
      "loss_ce": 0.00011718348105205223,
      "loss_iou": 0.185546875,
      "loss_num": 0.025390625,
      "loss_xval": 0.498046875,
      "num_input_tokens_seen": 300863900,
      "step": 5370
    },
    {
      "epoch": 11.962138084632517,
      "grad_norm": 20.06622314453125,
      "learning_rate": 1e-06,
      "loss": 0.385,
      "num_input_tokens_seen": 300919612,
      "step": 5371
    },
    {
      "epoch": 11.962138084632517,
      "loss": 0.37853488326072693,
      "loss_ce": 0.00011691114195855334,
      "loss_iou": 0.1728515625,
      "loss_num": 0.006683349609375,
      "loss_xval": 0.37890625,
      "num_input_tokens_seen": 300919612,
      "step": 5371
    },
    {
      "epoch": 11.964365256124722,
      "grad_norm": 22.610204696655273,
      "learning_rate": 1e-06,
      "loss": 0.4837,
      "num_input_tokens_seen": 300975088,
      "step": 5372
    },
    {
      "epoch": 11.964365256124722,
      "loss": 0.5133309364318848,
      "loss_ce": 0.00014736468438059092,
      "loss_iou": 0.2314453125,
      "loss_num": 0.01025390625,
      "loss_xval": 0.51171875,
      "num_input_tokens_seen": 300975088,
      "step": 5372
    },
    {
      "epoch": 11.966592427616927,
      "grad_norm": 23.854276657104492,
      "learning_rate": 1e-06,
      "loss": 0.5991,
      "num_input_tokens_seen": 301032024,
      "step": 5373
    },
    {
      "epoch": 11.966592427616927,
      "loss": 0.6122125387191772,
      "loss_ce": 0.00015198803157545626,
      "loss_iou": 0.25390625,
      "loss_num": 0.0205078125,
      "loss_xval": 0.61328125,
      "num_input_tokens_seen": 301032024,
      "step": 5373
    },
    {
      "epoch": 11.968819599109132,
      "grad_norm": 14.804436683654785,
      "learning_rate": 1e-06,
      "loss": 0.6603,
      "num_input_tokens_seen": 301087604,
      "step": 5374
    },
    {
      "epoch": 11.968819599109132,
      "loss": 0.8352023363113403,
      "loss_ce": 0.0001192896525026299,
      "loss_iou": 0.345703125,
      "loss_num": 0.0289306640625,
      "loss_xval": 0.8359375,
      "num_input_tokens_seen": 301087604,
      "step": 5374
    },
    {
      "epoch": 11.971046770601337,
      "grad_norm": 22.94840431213379,
      "learning_rate": 1e-06,
      "loss": 0.622,
      "num_input_tokens_seen": 301142676,
      "step": 5375
    },
    {
      "epoch": 11.971046770601337,
      "loss": 0.6568686962127686,
      "loss_ce": 0.0001304006582358852,
      "loss_iou": 0.279296875,
      "loss_num": 0.019287109375,
      "loss_xval": 0.65625,
      "num_input_tokens_seen": 301142676,
      "step": 5375
    },
    {
      "epoch": 11.973273942093542,
      "grad_norm": 12.997676849365234,
      "learning_rate": 1e-06,
      "loss": 0.4442,
      "num_input_tokens_seen": 301201488,
      "step": 5376
    },
    {
      "epoch": 11.973273942093542,
      "loss": 0.3666253089904785,
      "loss_ce": 0.00010921184730250388,
      "loss_iou": 0.1650390625,
      "loss_num": 0.00726318359375,
      "loss_xval": 0.3671875,
      "num_input_tokens_seen": 301201488,
      "step": 5376
    },
    {
      "epoch": 11.975501113585747,
      "grad_norm": 28.56017303466797,
      "learning_rate": 1e-06,
      "loss": 0.5158,
      "num_input_tokens_seen": 301259584,
      "step": 5377
    },
    {
      "epoch": 11.975501113585747,
      "loss": 0.3714648485183716,
      "loss_ce": 0.00012695527402684093,
      "loss_iou": 0.16796875,
      "loss_num": 0.006866455078125,
      "loss_xval": 0.37109375,
      "num_input_tokens_seen": 301259584,
      "step": 5377
    },
    {
      "epoch": 11.977728285077951,
      "grad_norm": 18.66181755065918,
      "learning_rate": 1e-06,
      "loss": 0.5058,
      "num_input_tokens_seen": 301316356,
      "step": 5378
    },
    {
      "epoch": 11.977728285077951,
      "loss": 0.5127632021903992,
      "loss_ce": 0.00018997653387486935,
      "loss_iou": 0.2353515625,
      "loss_num": 0.0084228515625,
      "loss_xval": 0.51171875,
      "num_input_tokens_seen": 301316356,
      "step": 5378
    },
    {
      "epoch": 11.979955456570156,
      "grad_norm": 21.85940170288086,
      "learning_rate": 1e-06,
      "loss": 0.5486,
      "num_input_tokens_seen": 301370076,
      "step": 5379
    },
    {
      "epoch": 11.979955456570156,
      "loss": 0.4443596601486206,
      "loss_ce": 0.00014578478294424713,
      "loss_iou": 0.2021484375,
      "loss_num": 0.00811767578125,
      "loss_xval": 0.443359375,
      "num_input_tokens_seen": 301370076,
      "step": 5379
    },
    {
      "epoch": 11.982182628062361,
      "grad_norm": 17.551218032836914,
      "learning_rate": 1e-06,
      "loss": 0.5728,
      "num_input_tokens_seen": 301427120,
      "step": 5380
    },
    {
      "epoch": 11.982182628062361,
      "loss": 0.7055622935295105,
      "loss_ce": 0.00011793218436650932,
      "loss_iou": 0.310546875,
      "loss_num": 0.0167236328125,
      "loss_xval": 0.70703125,
      "num_input_tokens_seen": 301427120,
      "step": 5380
    },
    {
      "epoch": 11.984409799554566,
      "grad_norm": 14.2471284866333,
      "learning_rate": 1e-06,
      "loss": 0.4307,
      "num_input_tokens_seen": 301480952,
      "step": 5381
    },
    {
      "epoch": 11.984409799554566,
      "loss": 0.4769345819950104,
      "loss_ce": 0.0001279542047996074,
      "loss_iou": 0.20703125,
      "loss_num": 0.0126953125,
      "loss_xval": 0.4765625,
      "num_input_tokens_seen": 301480952,
      "step": 5381
    },
    {
      "epoch": 11.98663697104677,
      "grad_norm": 23.872955322265625,
      "learning_rate": 1e-06,
      "loss": 0.6408,
      "num_input_tokens_seen": 301536748,
      "step": 5382
    },
    {
      "epoch": 11.98663697104677,
      "loss": 0.6590736508369446,
      "loss_ce": 0.00013808724179398268,
      "loss_iou": 0.29296875,
      "loss_num": 0.014404296875,
      "loss_xval": 0.66015625,
      "num_input_tokens_seen": 301536748,
      "step": 5382
    },
    {
      "epoch": 11.988864142538976,
      "grad_norm": 27.215002059936523,
      "learning_rate": 1e-06,
      "loss": 0.6737,
      "num_input_tokens_seen": 301591624,
      "step": 5383
    },
    {
      "epoch": 11.988864142538976,
      "loss": 0.7730184197425842,
      "loss_ce": 0.0003133205755148083,
      "loss_iou": 0.341796875,
      "loss_num": 0.01806640625,
      "loss_xval": 0.7734375,
      "num_input_tokens_seen": 301591624,
      "step": 5383
    },
    {
      "epoch": 11.99109131403118,
      "grad_norm": 16.59343910217285,
      "learning_rate": 1e-06,
      "loss": 0.4352,
      "num_input_tokens_seen": 301651368,
      "step": 5384
    },
    {
      "epoch": 11.99109131403118,
      "loss": 0.2988908886909485,
      "loss_ce": 0.00024585792561993003,
      "loss_iou": 0.1328125,
      "loss_num": 0.00677490234375,
      "loss_xval": 0.298828125,
      "num_input_tokens_seen": 301651368,
      "step": 5384
    },
    {
      "epoch": 11.993318485523385,
      "grad_norm": 17.6257266998291,
      "learning_rate": 1e-06,
      "loss": 0.5236,
      "num_input_tokens_seen": 301708848,
      "step": 5385
    },
    {
      "epoch": 11.993318485523385,
      "loss": 0.3951897919178009,
      "loss_ce": 0.000170250961673446,
      "loss_iou": 0.16796875,
      "loss_num": 0.01171875,
      "loss_xval": 0.39453125,
      "num_input_tokens_seen": 301708848,
      "step": 5385
    },
    {
      "epoch": 11.99554565701559,
      "grad_norm": 18.835487365722656,
      "learning_rate": 1e-06,
      "loss": 0.4832,
      "num_input_tokens_seen": 301766832,
      "step": 5386
    },
    {
      "epoch": 11.99554565701559,
      "loss": 0.45329028367996216,
      "loss_ce": 0.00016528656124137342,
      "loss_iou": 0.2060546875,
      "loss_num": 0.0081787109375,
      "loss_xval": 0.453125,
      "num_input_tokens_seen": 301766832,
      "step": 5386
    },
    {
      "epoch": 11.997772828507795,
      "grad_norm": 23.853605270385742,
      "learning_rate": 1e-06,
      "loss": 0.6195,
      "num_input_tokens_seen": 301822292,
      "step": 5387
    },
    {
      "epoch": 11.997772828507795,
      "loss": 0.6632393598556519,
      "loss_ce": 0.00015345893916673958,
      "loss_iou": 0.26953125,
      "loss_num": 0.0242919921875,
      "loss_xval": 0.6640625,
      "num_input_tokens_seen": 301822292,
      "step": 5387
    },
    {
      "epoch": 12.0,
      "grad_norm": 18.651371002197266,
      "learning_rate": 1e-06,
      "loss": 0.4679,
      "num_input_tokens_seen": 301876956,
      "step": 5388
    },
    {
      "epoch": 12.0,
      "loss": 0.5064892768859863,
      "loss_ce": 0.00014158777776174247,
      "loss_iou": 0.228515625,
      "loss_num": 0.00994873046875,
      "loss_xval": 0.5078125,
      "num_input_tokens_seen": 301876956,
      "step": 5388
    },
    {
      "epoch": 12.002227171492205,
      "grad_norm": 24.909053802490234,
      "learning_rate": 1e-06,
      "loss": 0.6428,
      "num_input_tokens_seen": 301930580,
      "step": 5389
    },
    {
      "epoch": 12.002227171492205,
      "loss": 0.7445248365402222,
      "loss_ce": 0.00014005119737703353,
      "loss_iou": 0.314453125,
      "loss_num": 0.0228271484375,
      "loss_xval": 0.74609375,
      "num_input_tokens_seen": 301930580,
      "step": 5389
    },
    {
      "epoch": 12.00445434298441,
      "grad_norm": 23.323347091674805,
      "learning_rate": 1e-06,
      "loss": 0.5544,
      "num_input_tokens_seen": 301986100,
      "step": 5390
    },
    {
      "epoch": 12.00445434298441,
      "loss": 0.45033586025238037,
      "loss_ce": 0.0001405369839631021,
      "loss_iou": 0.20703125,
      "loss_num": 0.00701904296875,
      "loss_xval": 0.44921875,
      "num_input_tokens_seen": 301986100,
      "step": 5390
    },
    {
      "epoch": 12.006681514476615,
      "grad_norm": 16.831626892089844,
      "learning_rate": 1e-06,
      "loss": 0.2852,
      "num_input_tokens_seen": 302043816,
      "step": 5391
    },
    {
      "epoch": 12.006681514476615,
      "loss": 0.2847111225128174,
      "loss_ce": 0.00010420664330013096,
      "loss_iou": 0.12158203125,
      "loss_num": 0.00823974609375,
      "loss_xval": 0.28515625,
      "num_input_tokens_seen": 302043816,
      "step": 5391
    },
    {
      "epoch": 12.00890868596882,
      "grad_norm": 37.528282165527344,
      "learning_rate": 1e-06,
      "loss": 0.6277,
      "num_input_tokens_seen": 302098732,
      "step": 5392
    },
    {
      "epoch": 12.00890868596882,
      "loss": 0.6959569454193115,
      "loss_ce": 0.00015614379663020372,
      "loss_iou": 0.30859375,
      "loss_num": 0.015869140625,
      "loss_xval": 0.6953125,
      "num_input_tokens_seen": 302098732,
      "step": 5392
    },
    {
      "epoch": 12.011135857461024,
      "grad_norm": 24.19828987121582,
      "learning_rate": 1e-06,
      "loss": 0.4908,
      "num_input_tokens_seen": 302157220,
      "step": 5393
    },
    {
      "epoch": 12.011135857461024,
      "loss": 0.5651944279670715,
      "loss_ce": 0.00013094657333567739,
      "loss_iou": 0.259765625,
      "loss_num": 0.00909423828125,
      "loss_xval": 0.56640625,
      "num_input_tokens_seen": 302157220,
      "step": 5393
    },
    {
      "epoch": 12.01336302895323,
      "grad_norm": 14.71508502960205,
      "learning_rate": 1e-06,
      "loss": 0.496,
      "num_input_tokens_seen": 302214052,
      "step": 5394
    },
    {
      "epoch": 12.01336302895323,
      "loss": 0.5302682518959045,
      "loss_ce": 0.00011686344078043476,
      "loss_iou": 0.2412109375,
      "loss_num": 0.00927734375,
      "loss_xval": 0.53125,
      "num_input_tokens_seen": 302214052,
      "step": 5394
    },
    {
      "epoch": 12.015590200445434,
      "grad_norm": 18.591922760009766,
      "learning_rate": 1e-06,
      "loss": 0.491,
      "num_input_tokens_seen": 302269552,
      "step": 5395
    },
    {
      "epoch": 12.015590200445434,
      "loss": 0.4772958755493164,
      "loss_ce": 0.00024510070215910673,
      "loss_iou": 0.2177734375,
      "loss_num": 0.00823974609375,
      "loss_xval": 0.4765625,
      "num_input_tokens_seen": 302269552,
      "step": 5395
    },
    {
      "epoch": 12.017817371937639,
      "grad_norm": 13.601188659667969,
      "learning_rate": 1e-06,
      "loss": 0.4128,
      "num_input_tokens_seen": 302326968,
      "step": 5396
    },
    {
      "epoch": 12.017817371937639,
      "loss": 0.4751969277858734,
      "loss_ce": 9.925015910994262e-05,
      "loss_iou": 0.197265625,
      "loss_num": 0.016357421875,
      "loss_xval": 0.474609375,
      "num_input_tokens_seen": 302326968,
      "step": 5396
    },
    {
      "epoch": 12.020044543429844,
      "grad_norm": 22.953596115112305,
      "learning_rate": 1e-06,
      "loss": 0.449,
      "num_input_tokens_seen": 302383496,
      "step": 5397
    },
    {
      "epoch": 12.020044543429844,
      "loss": 0.5418699979782104,
      "loss_ce": 0.0001219719197251834,
      "loss_iou": 0.2431640625,
      "loss_num": 0.0111083984375,
      "loss_xval": 0.54296875,
      "num_input_tokens_seen": 302383496,
      "step": 5397
    },
    {
      "epoch": 12.022271714922049,
      "grad_norm": 28.485471725463867,
      "learning_rate": 1e-06,
      "loss": 0.4552,
      "num_input_tokens_seen": 302438028,
      "step": 5398
    },
    {
      "epoch": 12.022271714922049,
      "loss": 0.46738266944885254,
      "loss_ce": 9.752875484991819e-05,
      "loss_iou": 0.177734375,
      "loss_num": 0.0224609375,
      "loss_xval": 0.466796875,
      "num_input_tokens_seen": 302438028,
      "step": 5398
    },
    {
      "epoch": 12.024498886414253,
      "grad_norm": 16.616348266601562,
      "learning_rate": 1e-06,
      "loss": 0.5826,
      "num_input_tokens_seen": 302494768,
      "step": 5399
    },
    {
      "epoch": 12.024498886414253,
      "loss": 0.5422226190567017,
      "loss_ce": 0.0001083296156139113,
      "loss_iou": 0.224609375,
      "loss_num": 0.0184326171875,
      "loss_xval": 0.54296875,
      "num_input_tokens_seen": 302494768,
      "step": 5399
    },
    {
      "epoch": 12.026726057906458,
      "grad_norm": 13.746217727661133,
      "learning_rate": 1e-06,
      "loss": 0.4764,
      "num_input_tokens_seen": 302550772,
      "step": 5400
    },
    {
      "epoch": 12.026726057906458,
      "loss": 0.4675309658050537,
      "loss_ce": 0.0001237354299519211,
      "loss_iou": 0.21484375,
      "loss_num": 0.0074462890625,
      "loss_xval": 0.466796875,
      "num_input_tokens_seen": 302550772,
      "step": 5400
    },
    {
      "epoch": 12.028953229398663,
      "grad_norm": 17.52446746826172,
      "learning_rate": 1e-06,
      "loss": 0.5146,
      "num_input_tokens_seen": 302608476,
      "step": 5401
    },
    {
      "epoch": 12.028953229398663,
      "loss": 0.5597883462905884,
      "loss_ce": 0.0008283640490844846,
      "loss_iou": 0.208984375,
      "loss_num": 0.0281982421875,
      "loss_xval": 0.55859375,
      "num_input_tokens_seen": 302608476,
      "step": 5401
    },
    {
      "epoch": 12.031180400890868,
      "grad_norm": 21.796451568603516,
      "learning_rate": 1e-06,
      "loss": 0.4571,
      "num_input_tokens_seen": 302663956,
      "step": 5402
    },
    {
      "epoch": 12.031180400890868,
      "loss": 0.5090689659118652,
      "loss_ce": 0.00021888897754251957,
      "loss_iou": 0.2099609375,
      "loss_num": 0.017822265625,
      "loss_xval": 0.5078125,
      "num_input_tokens_seen": 302663956,
      "step": 5402
    },
    {
      "epoch": 12.033407572383073,
      "grad_norm": 24.178424835205078,
      "learning_rate": 1e-06,
      "loss": 0.45,
      "num_input_tokens_seen": 302719060,
      "step": 5403
    },
    {
      "epoch": 12.033407572383073,
      "loss": 0.39098450541496277,
      "loss_ce": 0.00011537145473994315,
      "loss_iou": 0.18359375,
      "loss_num": 0.004669189453125,
      "loss_xval": 0.390625,
      "num_input_tokens_seen": 302719060,
      "step": 5403
    },
    {
      "epoch": 12.035634743875278,
      "grad_norm": 17.01597785949707,
      "learning_rate": 1e-06,
      "loss": 0.4611,
      "num_input_tokens_seen": 302773636,
      "step": 5404
    },
    {
      "epoch": 12.035634743875278,
      "loss": 0.3315487504005432,
      "loss_ce": 0.00012785526632796973,
      "loss_iou": 0.1298828125,
      "loss_num": 0.01422119140625,
      "loss_xval": 0.33203125,
      "num_input_tokens_seen": 302773636,
      "step": 5404
    },
    {
      "epoch": 12.037861915367483,
      "grad_norm": 14.340649604797363,
      "learning_rate": 1e-06,
      "loss": 0.4127,
      "num_input_tokens_seen": 302831388,
      "step": 5405
    },
    {
      "epoch": 12.037861915367483,
      "loss": 0.3785756528377533,
      "loss_ce": 0.00012716675701085478,
      "loss_iou": 0.171875,
      "loss_num": 0.0068359375,
      "loss_xval": 0.37890625,
      "num_input_tokens_seen": 302831388,
      "step": 5405
    },
    {
      "epoch": 12.040089086859687,
      "grad_norm": 25.321979522705078,
      "learning_rate": 1e-06,
      "loss": 0.5422,
      "num_input_tokens_seen": 302888920,
      "step": 5406
    },
    {
      "epoch": 12.040089086859687,
      "loss": 0.36657315492630005,
      "loss_ce": 0.00011808329145424068,
      "loss_iou": 0.1552734375,
      "loss_num": 0.01123046875,
      "loss_xval": 0.3671875,
      "num_input_tokens_seen": 302888920,
      "step": 5406
    },
    {
      "epoch": 12.042316258351892,
      "grad_norm": 18.01930809020996,
      "learning_rate": 1e-06,
      "loss": 0.4401,
      "num_input_tokens_seen": 302942712,
      "step": 5407
    },
    {
      "epoch": 12.042316258351892,
      "loss": 0.49123328924179077,
      "loss_ce": 0.00014440924860537052,
      "loss_iou": 0.21484375,
      "loss_num": 0.01239013671875,
      "loss_xval": 0.490234375,
      "num_input_tokens_seen": 302942712,
      "step": 5407
    },
    {
      "epoch": 12.044543429844097,
      "grad_norm": 35.77482604980469,
      "learning_rate": 1e-06,
      "loss": 0.6199,
      "num_input_tokens_seen": 302998400,
      "step": 5408
    },
    {
      "epoch": 12.044543429844097,
      "loss": 0.6408650875091553,
      "loss_ce": 0.00011807896953541785,
      "loss_iou": 0.275390625,
      "loss_num": 0.018310546875,
      "loss_xval": 0.640625,
      "num_input_tokens_seen": 302998400,
      "step": 5408
    },
    {
      "epoch": 12.046770601336302,
      "grad_norm": 40.30015563964844,
      "learning_rate": 1e-06,
      "loss": 0.4045,
      "num_input_tokens_seen": 303052680,
      "step": 5409
    },
    {
      "epoch": 12.046770601336302,
      "loss": 0.36034607887268066,
      "loss_ce": 0.00011658002040348947,
      "loss_iou": 0.1572265625,
      "loss_num": 0.00927734375,
      "loss_xval": 0.359375,
      "num_input_tokens_seen": 303052680,
      "step": 5409
    },
    {
      "epoch": 12.048997772828507,
      "grad_norm": 19.596263885498047,
      "learning_rate": 1e-06,
      "loss": 0.4598,
      "num_input_tokens_seen": 303110848,
      "step": 5410
    },
    {
      "epoch": 12.048997772828507,
      "loss": 0.3800050616264343,
      "loss_ce": 0.00012227029947098345,
      "loss_iou": 0.177734375,
      "loss_num": 0.005035400390625,
      "loss_xval": 0.37890625,
      "num_input_tokens_seen": 303110848,
      "step": 5410
    },
    {
      "epoch": 12.051224944320714,
      "grad_norm": 25.05034637451172,
      "learning_rate": 1e-06,
      "loss": 0.5697,
      "num_input_tokens_seen": 303167852,
      "step": 5411
    },
    {
      "epoch": 12.051224944320714,
      "loss": 0.6123285889625549,
      "loss_ce": 0.00014595050015486777,
      "loss_iou": 0.271484375,
      "loss_num": 0.0137939453125,
      "loss_xval": 0.61328125,
      "num_input_tokens_seen": 303167852,
      "step": 5411
    },
    {
      "epoch": 12.053452115812918,
      "grad_norm": 15.722366333007812,
      "learning_rate": 1e-06,
      "loss": 0.4301,
      "num_input_tokens_seen": 303226468,
      "step": 5412
    },
    {
      "epoch": 12.053452115812918,
      "loss": 0.44557318091392517,
      "loss_ce": 0.00026066991267725825,
      "loss_iou": 0.2001953125,
      "loss_num": 0.0089111328125,
      "loss_xval": 0.4453125,
      "num_input_tokens_seen": 303226468,
      "step": 5412
    },
    {
      "epoch": 12.055679287305123,
      "grad_norm": 30.098329544067383,
      "learning_rate": 1e-06,
      "loss": 0.5986,
      "num_input_tokens_seen": 303282616,
      "step": 5413
    },
    {
      "epoch": 12.055679287305123,
      "loss": 0.5686604976654053,
      "loss_ce": 0.0003011383814737201,
      "loss_iou": 0.2353515625,
      "loss_num": 0.019287109375,
      "loss_xval": 0.5703125,
      "num_input_tokens_seen": 303282616,
      "step": 5413
    },
    {
      "epoch": 12.057906458797328,
      "grad_norm": 11.368029594421387,
      "learning_rate": 1e-06,
      "loss": 0.618,
      "num_input_tokens_seen": 303340068,
      "step": 5414
    },
    {
      "epoch": 12.057906458797328,
      "loss": 0.6175664663314819,
      "loss_ce": 0.00013479686458595097,
      "loss_iou": 0.248046875,
      "loss_num": 0.0242919921875,
      "loss_xval": 0.6171875,
      "num_input_tokens_seen": 303340068,
      "step": 5414
    },
    {
      "epoch": 12.060133630289533,
      "grad_norm": 18.459564208984375,
      "learning_rate": 1e-06,
      "loss": 0.4627,
      "num_input_tokens_seen": 303396632,
      "step": 5415
    },
    {
      "epoch": 12.060133630289533,
      "loss": 0.5457392930984497,
      "loss_ce": 0.00014600652502849698,
      "loss_iou": 0.25,
      "loss_num": 0.00885009765625,
      "loss_xval": 0.546875,
      "num_input_tokens_seen": 303396632,
      "step": 5415
    },
    {
      "epoch": 12.062360801781738,
      "grad_norm": 27.545385360717773,
      "learning_rate": 1e-06,
      "loss": 0.4222,
      "num_input_tokens_seen": 303452996,
      "step": 5416
    },
    {
      "epoch": 12.062360801781738,
      "loss": 0.3862226605415344,
      "loss_ce": 0.00011425558477640152,
      "loss_iou": 0.1796875,
      "loss_num": 0.00531005859375,
      "loss_xval": 0.38671875,
      "num_input_tokens_seen": 303452996,
      "step": 5416
    },
    {
      "epoch": 12.064587973273943,
      "grad_norm": 17.189687728881836,
      "learning_rate": 1e-06,
      "loss": 0.6149,
      "num_input_tokens_seen": 303509476,
      "step": 5417
    },
    {
      "epoch": 12.064587973273943,
      "loss": 0.4504542648792267,
      "loss_ce": 0.0001368774683214724,
      "loss_iou": 0.201171875,
      "loss_num": 0.00970458984375,
      "loss_xval": 0.451171875,
      "num_input_tokens_seen": 303509476,
      "step": 5417
    },
    {
      "epoch": 12.066815144766148,
      "grad_norm": 13.223028182983398,
      "learning_rate": 1e-06,
      "loss": 0.4632,
      "num_input_tokens_seen": 303566016,
      "step": 5418
    },
    {
      "epoch": 12.066815144766148,
      "loss": 0.5367510318756104,
      "loss_ce": 0.00012994115240871906,
      "loss_iou": 0.2041015625,
      "loss_num": 0.0257568359375,
      "loss_xval": 0.53515625,
      "num_input_tokens_seen": 303566016,
      "step": 5418
    },
    {
      "epoch": 12.069042316258352,
      "grad_norm": 22.01005744934082,
      "learning_rate": 1e-06,
      "loss": 0.5708,
      "num_input_tokens_seen": 303621640,
      "step": 5419
    },
    {
      "epoch": 12.069042316258352,
      "loss": 0.33945903182029724,
      "loss_ce": 0.00010356900747865438,
      "loss_iou": 0.1474609375,
      "loss_num": 0.00872802734375,
      "loss_xval": 0.33984375,
      "num_input_tokens_seen": 303621640,
      "step": 5419
    },
    {
      "epoch": 12.071269487750557,
      "grad_norm": 16.25597381591797,
      "learning_rate": 1e-06,
      "loss": 0.4158,
      "num_input_tokens_seen": 303676132,
      "step": 5420
    },
    {
      "epoch": 12.071269487750557,
      "loss": 0.43920350074768066,
      "loss_ce": 0.00011659059964586049,
      "loss_iou": 0.1875,
      "loss_num": 0.0126953125,
      "loss_xval": 0.439453125,
      "num_input_tokens_seen": 303676132,
      "step": 5420
    },
    {
      "epoch": 12.073496659242762,
      "grad_norm": 26.435026168823242,
      "learning_rate": 1e-06,
      "loss": 0.4823,
      "num_input_tokens_seen": 303731772,
      "step": 5421
    },
    {
      "epoch": 12.073496659242762,
      "loss": 0.5533719658851624,
      "loss_ce": 0.00014934616046957672,
      "loss_iou": 0.2099609375,
      "loss_num": 0.0264892578125,
      "loss_xval": 0.5546875,
      "num_input_tokens_seen": 303731772,
      "step": 5421
    },
    {
      "epoch": 12.075723830734967,
      "grad_norm": 317.4898376464844,
      "learning_rate": 1e-06,
      "loss": 0.6007,
      "num_input_tokens_seen": 303788636,
      "step": 5422
    },
    {
      "epoch": 12.075723830734967,
      "loss": 0.8187873363494873,
      "loss_ce": 0.00018380230176262558,
      "loss_iou": 0.326171875,
      "loss_num": 0.033203125,
      "loss_xval": 0.8203125,
      "num_input_tokens_seen": 303788636,
      "step": 5422
    },
    {
      "epoch": 12.077951002227172,
      "grad_norm": 16.596609115600586,
      "learning_rate": 1e-06,
      "loss": 0.3791,
      "num_input_tokens_seen": 303844472,
      "step": 5423
    },
    {
      "epoch": 12.077951002227172,
      "loss": 0.29760003089904785,
      "loss_ce": 0.00011467649164842442,
      "loss_iou": 0.1259765625,
      "loss_num": 0.00921630859375,
      "loss_xval": 0.296875,
      "num_input_tokens_seen": 303844472,
      "step": 5423
    },
    {
      "epoch": 12.080178173719377,
      "grad_norm": 19.923240661621094,
      "learning_rate": 1e-06,
      "loss": 0.5823,
      "num_input_tokens_seen": 303899116,
      "step": 5424
    },
    {
      "epoch": 12.080178173719377,
      "loss": 0.7797752618789673,
      "loss_ce": 0.00011215673293918371,
      "loss_iou": 0.318359375,
      "loss_num": 0.028564453125,
      "loss_xval": 0.78125,
      "num_input_tokens_seen": 303899116,
      "step": 5424
    },
    {
      "epoch": 12.082405345211582,
      "grad_norm": 22.035127639770508,
      "learning_rate": 1e-06,
      "loss": 0.5462,
      "num_input_tokens_seen": 303952664,
      "step": 5425
    },
    {
      "epoch": 12.082405345211582,
      "loss": 0.6474908590316772,
      "loss_ce": 0.00015202595386654139,
      "loss_iou": 0.26953125,
      "loss_num": 0.0216064453125,
      "loss_xval": 0.6484375,
      "num_input_tokens_seen": 303952664,
      "step": 5425
    },
    {
      "epoch": 12.084632516703786,
      "grad_norm": 25.01907730102539,
      "learning_rate": 1e-06,
      "loss": 0.5354,
      "num_input_tokens_seen": 304009648,
      "step": 5426
    },
    {
      "epoch": 12.084632516703786,
      "loss": 0.660736083984375,
      "loss_ce": 0.00021362912957556546,
      "loss_iou": 0.263671875,
      "loss_num": 0.0267333984375,
      "loss_xval": 0.66015625,
      "num_input_tokens_seen": 304009648,
      "step": 5426
    },
    {
      "epoch": 12.086859688195991,
      "grad_norm": 14.428705215454102,
      "learning_rate": 1e-06,
      "loss": 0.4073,
      "num_input_tokens_seen": 304069088,
      "step": 5427
    },
    {
      "epoch": 12.086859688195991,
      "loss": 0.3368920683860779,
      "loss_ce": 0.0001000880729407072,
      "loss_iou": 0.1572265625,
      "loss_num": 0.0045166015625,
      "loss_xval": 0.3359375,
      "num_input_tokens_seen": 304069088,
      "step": 5427
    },
    {
      "epoch": 12.089086859688196,
      "grad_norm": 23.767419815063477,
      "learning_rate": 1e-06,
      "loss": 0.5659,
      "num_input_tokens_seen": 304125008,
      "step": 5428
    },
    {
      "epoch": 12.089086859688196,
      "loss": 0.5689517259597778,
      "loss_ce": 0.00010411132825538516,
      "loss_iou": 0.2275390625,
      "loss_num": 0.022705078125,
      "loss_xval": 0.5703125,
      "num_input_tokens_seen": 304125008,
      "step": 5428
    },
    {
      "epoch": 12.091314031180401,
      "grad_norm": 25.242143630981445,
      "learning_rate": 1e-06,
      "loss": 0.3432,
      "num_input_tokens_seen": 304182308,
      "step": 5429
    },
    {
      "epoch": 12.091314031180401,
      "loss": 0.38244450092315674,
      "loss_ce": 0.00012029155914206058,
      "loss_iou": 0.1796875,
      "loss_num": 0.00445556640625,
      "loss_xval": 0.3828125,
      "num_input_tokens_seen": 304182308,
      "step": 5429
    },
    {
      "epoch": 12.093541202672606,
      "grad_norm": 13.21595287322998,
      "learning_rate": 1e-06,
      "loss": 0.5233,
      "num_input_tokens_seen": 304239692,
      "step": 5430
    },
    {
      "epoch": 12.093541202672606,
      "loss": 0.6246329545974731,
      "loss_ce": 0.00012122253247071058,
      "loss_iou": 0.263671875,
      "loss_num": 0.0196533203125,
      "loss_xval": 0.625,
      "num_input_tokens_seen": 304239692,
      "step": 5430
    },
    {
      "epoch": 12.09576837416481,
      "grad_norm": 26.7417049407959,
      "learning_rate": 1e-06,
      "loss": 0.6778,
      "num_input_tokens_seen": 304296608,
      "step": 5431
    },
    {
      "epoch": 12.09576837416481,
      "loss": 0.5760781168937683,
      "loss_ce": 0.0001503834209870547,
      "loss_iou": 0.26171875,
      "loss_num": 0.010498046875,
      "loss_xval": 0.57421875,
      "num_input_tokens_seen": 304296608,
      "step": 5431
    },
    {
      "epoch": 12.097995545657016,
      "grad_norm": 14.140853881835938,
      "learning_rate": 1e-06,
      "loss": 0.5346,
      "num_input_tokens_seen": 304351720,
      "step": 5432
    },
    {
      "epoch": 12.097995545657016,
      "loss": 0.6522471308708191,
      "loss_ce": 0.00014752510469406843,
      "loss_iou": 0.265625,
      "loss_num": 0.0244140625,
      "loss_xval": 0.65234375,
      "num_input_tokens_seen": 304351720,
      "step": 5432
    },
    {
      "epoch": 12.10022271714922,
      "grad_norm": 17.195842742919922,
      "learning_rate": 1e-06,
      "loss": 0.4608,
      "num_input_tokens_seen": 304411140,
      "step": 5433
    },
    {
      "epoch": 12.10022271714922,
      "loss": 0.4692525267601013,
      "loss_ce": 0.00013632513582706451,
      "loss_iou": 0.212890625,
      "loss_num": 0.00848388671875,
      "loss_xval": 0.46875,
      "num_input_tokens_seen": 304411140,
      "step": 5433
    },
    {
      "epoch": 12.102449888641425,
      "grad_norm": 17.013219833374023,
      "learning_rate": 1e-06,
      "loss": 0.5863,
      "num_input_tokens_seen": 304466832,
      "step": 5434
    },
    {
      "epoch": 12.102449888641425,
      "loss": 0.46361833810806274,
      "loss_ce": 0.00011738392640836537,
      "loss_iou": 0.205078125,
      "loss_num": 0.01043701171875,
      "loss_xval": 0.462890625,
      "num_input_tokens_seen": 304466832,
      "step": 5434
    },
    {
      "epoch": 12.10467706013363,
      "grad_norm": 19.217731475830078,
      "learning_rate": 1e-06,
      "loss": 0.5909,
      "num_input_tokens_seen": 304524596,
      "step": 5435
    },
    {
      "epoch": 12.10467706013363,
      "loss": 0.5402824878692627,
      "loss_ce": 0.00012132612755522132,
      "loss_iou": 0.240234375,
      "loss_num": 0.01220703125,
      "loss_xval": 0.5390625,
      "num_input_tokens_seen": 304524596,
      "step": 5435
    },
    {
      "epoch": 12.106904231625835,
      "grad_norm": 12.651470184326172,
      "learning_rate": 1e-06,
      "loss": 0.4578,
      "num_input_tokens_seen": 304581824,
      "step": 5436
    },
    {
      "epoch": 12.106904231625835,
      "loss": 0.35941869020462036,
      "loss_ce": 0.00016576812777202576,
      "loss_iou": 0.1435546875,
      "loss_num": 0.0142822265625,
      "loss_xval": 0.359375,
      "num_input_tokens_seen": 304581824,
      "step": 5436
    },
    {
      "epoch": 12.10913140311804,
      "grad_norm": 15.489208221435547,
      "learning_rate": 1e-06,
      "loss": 0.3276,
      "num_input_tokens_seen": 304637832,
      "step": 5437
    },
    {
      "epoch": 12.10913140311804,
      "loss": 0.27845609188079834,
      "loss_ce": 0.00013578942161984742,
      "loss_iou": 0.1162109375,
      "loss_num": 0.00909423828125,
      "loss_xval": 0.27734375,
      "num_input_tokens_seen": 304637832,
      "step": 5437
    },
    {
      "epoch": 12.111358574610245,
      "grad_norm": 31.793598175048828,
      "learning_rate": 1e-06,
      "loss": 0.4174,
      "num_input_tokens_seen": 304695616,
      "step": 5438
    },
    {
      "epoch": 12.111358574610245,
      "loss": 0.40172773599624634,
      "loss_ce": 0.00011640525190159678,
      "loss_iou": 0.166015625,
      "loss_num": 0.01397705078125,
      "loss_xval": 0.40234375,
      "num_input_tokens_seen": 304695616,
      "step": 5438
    },
    {
      "epoch": 12.11358574610245,
      "grad_norm": 23.2586669921875,
      "learning_rate": 1e-06,
      "loss": 0.3909,
      "num_input_tokens_seen": 304751796,
      "step": 5439
    },
    {
      "epoch": 12.11358574610245,
      "loss": 0.3819471001625061,
      "loss_ce": 0.00011117332905996591,
      "loss_iou": 0.171875,
      "loss_num": 0.00762939453125,
      "loss_xval": 0.3828125,
      "num_input_tokens_seen": 304751796,
      "step": 5439
    },
    {
      "epoch": 12.115812917594655,
      "grad_norm": 15.429174423217773,
      "learning_rate": 1e-06,
      "loss": 0.3506,
      "num_input_tokens_seen": 304806936,
      "step": 5440
    },
    {
      "epoch": 12.115812917594655,
      "loss": 0.3502577543258667,
      "loss_ce": 0.0001600981195224449,
      "loss_iou": 0.16015625,
      "loss_num": 0.00616455078125,
      "loss_xval": 0.349609375,
      "num_input_tokens_seen": 304806936,
      "step": 5440
    },
    {
      "epoch": 12.11804008908686,
      "grad_norm": 16.58045196533203,
      "learning_rate": 1e-06,
      "loss": 0.408,
      "num_input_tokens_seen": 304864392,
      "step": 5441
    },
    {
      "epoch": 12.11804008908686,
      "loss": 0.4236249625682831,
      "loss_ce": 0.00016305723693221807,
      "loss_iou": 0.173828125,
      "loss_num": 0.01513671875,
      "loss_xval": 0.423828125,
      "num_input_tokens_seen": 304864392,
      "step": 5441
    },
    {
      "epoch": 12.120267260579064,
      "grad_norm": 13.274721145629883,
      "learning_rate": 1e-06,
      "loss": 0.4532,
      "num_input_tokens_seen": 304921888,
      "step": 5442
    },
    {
      "epoch": 12.120267260579064,
      "loss": 0.4875772297382355,
      "loss_ce": 0.00015048254863359034,
      "loss_iou": 0.21484375,
      "loss_num": 0.01171875,
      "loss_xval": 0.48828125,
      "num_input_tokens_seen": 304921888,
      "step": 5442
    },
    {
      "epoch": 12.122494432071269,
      "grad_norm": 17.956384658813477,
      "learning_rate": 1e-06,
      "loss": 0.6849,
      "num_input_tokens_seen": 304977556,
      "step": 5443
    },
    {
      "epoch": 12.122494432071269,
      "loss": 0.7030162811279297,
      "loss_ce": 0.00013541628140956163,
      "loss_iou": 0.263671875,
      "loss_num": 0.03515625,
      "loss_xval": 0.703125,
      "num_input_tokens_seen": 304977556,
      "step": 5443
    },
    {
      "epoch": 12.124721603563474,
      "grad_norm": 16.645193099975586,
      "learning_rate": 1e-06,
      "loss": 0.6395,
      "num_input_tokens_seen": 305033024,
      "step": 5444
    },
    {
      "epoch": 12.124721603563474,
      "loss": 0.5505350232124329,
      "loss_ce": 0.00011998764239251614,
      "loss_iou": 0.244140625,
      "loss_num": 0.01239013671875,
      "loss_xval": 0.55078125,
      "num_input_tokens_seen": 305033024,
      "step": 5444
    },
    {
      "epoch": 12.126948775055679,
      "grad_norm": 13.85136604309082,
      "learning_rate": 1e-06,
      "loss": 0.4585,
      "num_input_tokens_seen": 305089956,
      "step": 5445
    },
    {
      "epoch": 12.126948775055679,
      "loss": 0.44617876410484314,
      "loss_ce": 0.00013385072816163301,
      "loss_iou": 0.1650390625,
      "loss_num": 0.02294921875,
      "loss_xval": 0.4453125,
      "num_input_tokens_seen": 305089956,
      "step": 5445
    },
    {
      "epoch": 12.129175946547884,
      "grad_norm": 23.356822967529297,
      "learning_rate": 1e-06,
      "loss": 0.4678,
      "num_input_tokens_seen": 305143452,
      "step": 5446
    },
    {
      "epoch": 12.129175946547884,
      "loss": 0.4208873510360718,
      "loss_ce": 0.00011096645175712183,
      "loss_iou": 0.1904296875,
      "loss_num": 0.00811767578125,
      "loss_xval": 0.419921875,
      "num_input_tokens_seen": 305143452,
      "step": 5446
    },
    {
      "epoch": 12.131403118040089,
      "grad_norm": 24.38487434387207,
      "learning_rate": 1e-06,
      "loss": 0.488,
      "num_input_tokens_seen": 305196260,
      "step": 5447
    },
    {
      "epoch": 12.131403118040089,
      "loss": 0.4843854010105133,
      "loss_ce": 0.00013247507740743458,
      "loss_iou": 0.2119140625,
      "loss_num": 0.01202392578125,
      "loss_xval": 0.484375,
      "num_input_tokens_seen": 305196260,
      "step": 5447
    },
    {
      "epoch": 12.133630289532293,
      "grad_norm": 15.976398468017578,
      "learning_rate": 1e-06,
      "loss": 0.4716,
      "num_input_tokens_seen": 305253716,
      "step": 5448
    },
    {
      "epoch": 12.133630289532293,
      "loss": 0.5756908655166626,
      "loss_ce": 0.00012929181684739888,
      "loss_iou": 0.23046875,
      "loss_num": 0.02294921875,
      "loss_xval": 0.57421875,
      "num_input_tokens_seen": 305253716,
      "step": 5448
    },
    {
      "epoch": 12.135857461024498,
      "grad_norm": 17.87403678894043,
      "learning_rate": 1e-06,
      "loss": 0.5232,
      "num_input_tokens_seen": 305307148,
      "step": 5449
    },
    {
      "epoch": 12.135857461024498,
      "loss": 0.6177228689193726,
      "loss_ce": 0.00016915984451770782,
      "loss_iou": 0.26953125,
      "loss_num": 0.0159912109375,
      "loss_xval": 0.6171875,
      "num_input_tokens_seen": 305307148,
      "step": 5449
    },
    {
      "epoch": 12.138084632516703,
      "grad_norm": 18.055152893066406,
      "learning_rate": 1e-06,
      "loss": 0.5738,
      "num_input_tokens_seen": 305364540,
      "step": 5450
    },
    {
      "epoch": 12.138084632516703,
      "loss": 0.615594744682312,
      "loss_ce": 0.00011621808516792953,
      "loss_iou": 0.267578125,
      "loss_num": 0.0162353515625,
      "loss_xval": 0.6171875,
      "num_input_tokens_seen": 305364540,
      "step": 5450
    },
    {
      "epoch": 12.140311804008908,
      "grad_norm": 13.896754264831543,
      "learning_rate": 1e-06,
      "loss": 0.4093,
      "num_input_tokens_seen": 305419432,
      "step": 5451
    },
    {
      "epoch": 12.140311804008908,
      "loss": 0.4336770176887512,
      "loss_ce": 0.0009377308306284249,
      "loss_iou": 0.1826171875,
      "loss_num": 0.013671875,
      "loss_xval": 0.43359375,
      "num_input_tokens_seen": 305419432,
      "step": 5451
    },
    {
      "epoch": 12.142538975501113,
      "grad_norm": 70.83055114746094,
      "learning_rate": 1e-06,
      "loss": 0.4653,
      "num_input_tokens_seen": 305475840,
      "step": 5452
    },
    {
      "epoch": 12.142538975501113,
      "loss": 0.3804883062839508,
      "loss_ce": 0.00011720365728251636,
      "loss_iou": 0.166015625,
      "loss_num": 0.00970458984375,
      "loss_xval": 0.380859375,
      "num_input_tokens_seen": 305475840,
      "step": 5452
    },
    {
      "epoch": 12.144766146993318,
      "grad_norm": 33.42408752441406,
      "learning_rate": 1e-06,
      "loss": 0.661,
      "num_input_tokens_seen": 305532432,
      "step": 5453
    },
    {
      "epoch": 12.144766146993318,
      "loss": 0.8063072562217712,
      "loss_ce": 0.000154937180923298,
      "loss_iou": 0.337890625,
      "loss_num": 0.025634765625,
      "loss_xval": 0.8046875,
      "num_input_tokens_seen": 305532432,
      "step": 5453
    },
    {
      "epoch": 12.146993318485523,
      "grad_norm": 23.046737670898438,
      "learning_rate": 1e-06,
      "loss": 0.3551,
      "num_input_tokens_seen": 305589804,
      "step": 5454
    },
    {
      "epoch": 12.146993318485523,
      "loss": 0.37097907066345215,
      "loss_ce": 0.00012947215873282403,
      "loss_iou": 0.166015625,
      "loss_num": 0.0076904296875,
      "loss_xval": 0.37109375,
      "num_input_tokens_seen": 305589804,
      "step": 5454
    },
    {
      "epoch": 12.14922048997773,
      "grad_norm": 17.94960594177246,
      "learning_rate": 1e-06,
      "loss": 0.533,
      "num_input_tokens_seen": 305643368,
      "step": 5455
    },
    {
      "epoch": 12.14922048997773,
      "loss": 0.5467525720596313,
      "loss_ce": 0.00012171192065579817,
      "loss_iou": 0.255859375,
      "loss_num": 0.00701904296875,
      "loss_xval": 0.546875,
      "num_input_tokens_seen": 305643368,
      "step": 5455
    },
    {
      "epoch": 12.151447661469934,
      "grad_norm": 21.4966983795166,
      "learning_rate": 1e-06,
      "loss": 0.4453,
      "num_input_tokens_seen": 305698236,
      "step": 5456
    },
    {
      "epoch": 12.151447661469934,
      "loss": 0.5014622807502747,
      "loss_ce": 0.00011951071064686403,
      "loss_iou": 0.220703125,
      "loss_num": 0.01214599609375,
      "loss_xval": 0.5,
      "num_input_tokens_seen": 305698236,
      "step": 5456
    },
    {
      "epoch": 12.153674832962139,
      "grad_norm": 32.09659194946289,
      "learning_rate": 1e-06,
      "loss": 0.5381,
      "num_input_tokens_seen": 305755728,
      "step": 5457
    },
    {
      "epoch": 12.153674832962139,
      "loss": 0.6394698023796082,
      "loss_ce": 0.0001875544257927686,
      "loss_iou": 0.2890625,
      "loss_num": 0.01177978515625,
      "loss_xval": 0.640625,
      "num_input_tokens_seen": 305755728,
      "step": 5457
    },
    {
      "epoch": 12.155902004454344,
      "grad_norm": 28.464859008789062,
      "learning_rate": 1e-06,
      "loss": 0.4574,
      "num_input_tokens_seen": 305811120,
      "step": 5458
    },
    {
      "epoch": 12.155902004454344,
      "loss": 0.34215444326400757,
      "loss_ce": 0.00011341302888467908,
      "loss_iou": 0.1474609375,
      "loss_num": 0.00933837890625,
      "loss_xval": 0.341796875,
      "num_input_tokens_seen": 305811120,
      "step": 5458
    },
    {
      "epoch": 12.158129175946549,
      "grad_norm": 22.244796752929688,
      "learning_rate": 1e-06,
      "loss": 0.4736,
      "num_input_tokens_seen": 305867104,
      "step": 5459
    },
    {
      "epoch": 12.158129175946549,
      "loss": 0.4112476706504822,
      "loss_ce": 0.00011484247806947678,
      "loss_iou": 0.1884765625,
      "loss_num": 0.006866455078125,
      "loss_xval": 0.41015625,
      "num_input_tokens_seen": 305867104,
      "step": 5459
    },
    {
      "epoch": 12.160356347438753,
      "grad_norm": 29.14021873474121,
      "learning_rate": 1e-06,
      "loss": 0.5295,
      "num_input_tokens_seen": 305921188,
      "step": 5460
    },
    {
      "epoch": 12.160356347438753,
      "loss": 0.6369410753250122,
      "loss_ce": 0.0004054922901559621,
      "loss_iou": 0.283203125,
      "loss_num": 0.0140380859375,
      "loss_xval": 0.63671875,
      "num_input_tokens_seen": 305921188,
      "step": 5460
    },
    {
      "epoch": 12.162583518930958,
      "grad_norm": 16.301149368286133,
      "learning_rate": 1e-06,
      "loss": 0.4447,
      "num_input_tokens_seen": 305978144,
      "step": 5461
    },
    {
      "epoch": 12.162583518930958,
      "loss": 0.5294920802116394,
      "loss_ce": 0.0001952020829776302,
      "loss_iou": 0.2333984375,
      "loss_num": 0.01263427734375,
      "loss_xval": 0.53125,
      "num_input_tokens_seen": 305978144,
      "step": 5461
    },
    {
      "epoch": 12.164810690423163,
      "grad_norm": 17.781740188598633,
      "learning_rate": 1e-06,
      "loss": 0.2532,
      "num_input_tokens_seen": 306033640,
      "step": 5462
    },
    {
      "epoch": 12.164810690423163,
      "loss": 0.2828177809715271,
      "loss_ce": 0.0001639821712160483,
      "loss_iou": 0.1259765625,
      "loss_num": 0.00604248046875,
      "loss_xval": 0.283203125,
      "num_input_tokens_seen": 306033640,
      "step": 5462
    },
    {
      "epoch": 12.167037861915368,
      "grad_norm": 33.484806060791016,
      "learning_rate": 1e-06,
      "loss": 0.4501,
      "num_input_tokens_seen": 306091020,
      "step": 5463
    },
    {
      "epoch": 12.167037861915368,
      "loss": 0.46839091181755066,
      "loss_ce": 0.00012922003224957734,
      "loss_iou": 0.18359375,
      "loss_num": 0.020263671875,
      "loss_xval": 0.46875,
      "num_input_tokens_seen": 306091020,
      "step": 5463
    },
    {
      "epoch": 12.169265033407573,
      "grad_norm": 26.4747257232666,
      "learning_rate": 1e-06,
      "loss": 0.6015,
      "num_input_tokens_seen": 306147840,
      "step": 5464
    },
    {
      "epoch": 12.169265033407573,
      "loss": 0.5650879740715027,
      "loss_ce": 0.00014657803694717586,
      "loss_iou": 0.25390625,
      "loss_num": 0.01177978515625,
      "loss_xval": 0.56640625,
      "num_input_tokens_seen": 306147840,
      "step": 5464
    },
    {
      "epoch": 12.171492204899778,
      "grad_norm": 24.981590270996094,
      "learning_rate": 1e-06,
      "loss": 0.6129,
      "num_input_tokens_seen": 306203280,
      "step": 5465
    },
    {
      "epoch": 12.171492204899778,
      "loss": 0.8055815100669861,
      "loss_ce": 0.00016156808123923838,
      "loss_iou": 0.376953125,
      "loss_num": 0.010009765625,
      "loss_xval": 0.8046875,
      "num_input_tokens_seen": 306203280,
      "step": 5465
    },
    {
      "epoch": 12.173719376391983,
      "grad_norm": 14.754565238952637,
      "learning_rate": 1e-06,
      "loss": 0.5721,
      "num_input_tokens_seen": 306257676,
      "step": 5466
    },
    {
      "epoch": 12.173719376391983,
      "loss": 0.5629807710647583,
      "loss_ce": 0.00011456996435299516,
      "loss_iou": 0.25390625,
      "loss_num": 0.0108642578125,
      "loss_xval": 0.5625,
      "num_input_tokens_seen": 306257676,
      "step": 5466
    },
    {
      "epoch": 12.175946547884188,
      "grad_norm": 25.851285934448242,
      "learning_rate": 1e-06,
      "loss": 0.5937,
      "num_input_tokens_seen": 306314104,
      "step": 5467
    },
    {
      "epoch": 12.175946547884188,
      "loss": 0.5030755996704102,
      "loss_ce": 0.00020692951511591673,
      "loss_iou": 0.2109375,
      "loss_num": 0.0159912109375,
      "loss_xval": 0.50390625,
      "num_input_tokens_seen": 306314104,
      "step": 5467
    },
    {
      "epoch": 12.178173719376392,
      "grad_norm": 19.489961624145508,
      "learning_rate": 1e-06,
      "loss": 0.385,
      "num_input_tokens_seen": 306372532,
      "step": 5468
    },
    {
      "epoch": 12.178173719376392,
      "loss": 0.4690125584602356,
      "loss_ce": 0.00014049038873054087,
      "loss_iou": 0.2001953125,
      "loss_num": 0.0135498046875,
      "loss_xval": 0.46875,
      "num_input_tokens_seen": 306372532,
      "step": 5468
    },
    {
      "epoch": 12.180400890868597,
      "grad_norm": 14.486827850341797,
      "learning_rate": 1e-06,
      "loss": 0.416,
      "num_input_tokens_seen": 306428160,
      "step": 5469
    },
    {
      "epoch": 12.180400890868597,
      "loss": 0.4102684259414673,
      "loss_ce": 0.00011218419240321964,
      "loss_iou": 0.185546875,
      "loss_num": 0.0078125,
      "loss_xval": 0.41015625,
      "num_input_tokens_seen": 306428160,
      "step": 5469
    },
    {
      "epoch": 12.182628062360802,
      "grad_norm": 48.91545867919922,
      "learning_rate": 1e-06,
      "loss": 0.5029,
      "num_input_tokens_seen": 306484716,
      "step": 5470
    },
    {
      "epoch": 12.182628062360802,
      "loss": 0.4544667601585388,
      "loss_ce": 0.00012103513290639967,
      "loss_iou": 0.1982421875,
      "loss_num": 0.01165771484375,
      "loss_xval": 0.455078125,
      "num_input_tokens_seen": 306484716,
      "step": 5470
    },
    {
      "epoch": 12.184855233853007,
      "grad_norm": 15.201376914978027,
      "learning_rate": 1e-06,
      "loss": 0.6931,
      "num_input_tokens_seen": 306542220,
      "step": 5471
    },
    {
      "epoch": 12.184855233853007,
      "loss": 0.5883694887161255,
      "loss_ce": 0.00011262335465289652,
      "loss_iou": 0.26953125,
      "loss_num": 0.00946044921875,
      "loss_xval": 0.58984375,
      "num_input_tokens_seen": 306542220,
      "step": 5471
    },
    {
      "epoch": 12.187082405345212,
      "grad_norm": 17.70683479309082,
      "learning_rate": 1e-06,
      "loss": 0.6325,
      "num_input_tokens_seen": 306598764,
      "step": 5472
    },
    {
      "epoch": 12.187082405345212,
      "loss": 0.5439872741699219,
      "loss_ce": 0.00016404017515014857,
      "loss_iou": 0.234375,
      "loss_num": 0.01507568359375,
      "loss_xval": 0.54296875,
      "num_input_tokens_seen": 306598764,
      "step": 5472
    },
    {
      "epoch": 12.189309576837417,
      "grad_norm": 15.60269546508789,
      "learning_rate": 1e-06,
      "loss": 0.4718,
      "num_input_tokens_seen": 306655192,
      "step": 5473
    },
    {
      "epoch": 12.189309576837417,
      "loss": 0.5688609480857849,
      "loss_ce": 0.000135363225126639,
      "loss_iou": 0.2373046875,
      "loss_num": 0.0189208984375,
      "loss_xval": 0.5703125,
      "num_input_tokens_seen": 306655192,
      "step": 5473
    },
    {
      "epoch": 12.191536748329622,
      "grad_norm": 23.64163589477539,
      "learning_rate": 1e-06,
      "loss": 0.5773,
      "num_input_tokens_seen": 306711220,
      "step": 5474
    },
    {
      "epoch": 12.191536748329622,
      "loss": 0.4205397963523865,
      "loss_ce": 0.00012965156929567456,
      "loss_iou": 0.1796875,
      "loss_num": 0.0120849609375,
      "loss_xval": 0.419921875,
      "num_input_tokens_seen": 306711220,
      "step": 5474
    },
    {
      "epoch": 12.193763919821826,
      "grad_norm": 19.232250213623047,
      "learning_rate": 1e-06,
      "loss": 0.6147,
      "num_input_tokens_seen": 306765912,
      "step": 5475
    },
    {
      "epoch": 12.193763919821826,
      "loss": 0.6638138294219971,
      "loss_ce": 0.0001785356434993446,
      "loss_iou": 0.3046875,
      "loss_num": 0.01116943359375,
      "loss_xval": 0.6640625,
      "num_input_tokens_seen": 306765912,
      "step": 5475
    },
    {
      "epoch": 12.195991091314031,
      "grad_norm": 24.970657348632812,
      "learning_rate": 1e-06,
      "loss": 0.4975,
      "num_input_tokens_seen": 306823380,
      "step": 5476
    },
    {
      "epoch": 12.195991091314031,
      "loss": 0.5586141347885132,
      "loss_ce": 0.000142430275445804,
      "loss_iou": 0.255859375,
      "loss_num": 0.00927734375,
      "loss_xval": 0.55859375,
      "num_input_tokens_seen": 306823380,
      "step": 5476
    },
    {
      "epoch": 12.198218262806236,
      "grad_norm": 23.488998413085938,
      "learning_rate": 1e-06,
      "loss": 0.5034,
      "num_input_tokens_seen": 306880924,
      "step": 5477
    },
    {
      "epoch": 12.198218262806236,
      "loss": 0.5716678500175476,
      "loss_ce": 0.00013462700007949024,
      "loss_iou": 0.24609375,
      "loss_num": 0.015869140625,
      "loss_xval": 0.5703125,
      "num_input_tokens_seen": 306880924,
      "step": 5477
    },
    {
      "epoch": 12.200445434298441,
      "grad_norm": 13.182589530944824,
      "learning_rate": 1e-06,
      "loss": 0.4163,
      "num_input_tokens_seen": 306937384,
      "step": 5478
    },
    {
      "epoch": 12.200445434298441,
      "loss": 0.4369211792945862,
      "loss_ce": 0.00015359185636043549,
      "loss_iou": 0.1982421875,
      "loss_num": 0.0079345703125,
      "loss_xval": 0.4375,
      "num_input_tokens_seen": 306937384,
      "step": 5478
    },
    {
      "epoch": 12.202672605790646,
      "grad_norm": 24.044740676879883,
      "learning_rate": 1e-06,
      "loss": 0.365,
      "num_input_tokens_seen": 306990924,
      "step": 5479
    },
    {
      "epoch": 12.202672605790646,
      "loss": 0.3719392418861389,
      "loss_ce": 0.00011306589294690639,
      "loss_iou": 0.16796875,
      "loss_num": 0.00701904296875,
      "loss_xval": 0.37109375,
      "num_input_tokens_seen": 306990924,
      "step": 5479
    },
    {
      "epoch": 12.20489977728285,
      "grad_norm": 29.678030014038086,
      "learning_rate": 1e-06,
      "loss": 0.5995,
      "num_input_tokens_seen": 307047468,
      "step": 5480
    },
    {
      "epoch": 12.20489977728285,
      "loss": 0.4761919677257538,
      "loss_ce": 0.00011774588347179815,
      "loss_iou": 0.2060546875,
      "loss_num": 0.01275634765625,
      "loss_xval": 0.4765625,
      "num_input_tokens_seen": 307047468,
      "step": 5480
    },
    {
      "epoch": 12.207126948775056,
      "grad_norm": 11.851914405822754,
      "learning_rate": 1e-06,
      "loss": 0.4111,
      "num_input_tokens_seen": 307103488,
      "step": 5481
    },
    {
      "epoch": 12.207126948775056,
      "loss": 0.25440388917922974,
      "loss_ce": 0.00022297943360172212,
      "loss_iou": 0.11328125,
      "loss_num": 0.00567626953125,
      "loss_xval": 0.25390625,
      "num_input_tokens_seen": 307103488,
      "step": 5481
    },
    {
      "epoch": 12.20935412026726,
      "grad_norm": 17.849971771240234,
      "learning_rate": 1e-06,
      "loss": 0.3751,
      "num_input_tokens_seen": 307159768,
      "step": 5482
    },
    {
      "epoch": 12.20935412026726,
      "loss": 0.32383567094802856,
      "loss_ce": 0.00010521031072130427,
      "loss_iou": 0.13671875,
      "loss_num": 0.00994873046875,
      "loss_xval": 0.32421875,
      "num_input_tokens_seen": 307159768,
      "step": 5482
    },
    {
      "epoch": 12.211581291759465,
      "grad_norm": 24.031517028808594,
      "learning_rate": 1e-06,
      "loss": 0.3802,
      "num_input_tokens_seen": 307212860,
      "step": 5483
    },
    {
      "epoch": 12.211581291759465,
      "loss": 0.354860782623291,
      "loss_ce": 0.00012447501649148762,
      "loss_iou": 0.1630859375,
      "loss_num": 0.0054931640625,
      "loss_xval": 0.35546875,
      "num_input_tokens_seen": 307212860,
      "step": 5483
    },
    {
      "epoch": 12.21380846325167,
      "grad_norm": 25.485084533691406,
      "learning_rate": 1e-06,
      "loss": 0.4435,
      "num_input_tokens_seen": 307265808,
      "step": 5484
    },
    {
      "epoch": 12.21380846325167,
      "loss": 0.42531439661979675,
      "loss_ce": 0.00013584828411694616,
      "loss_iou": 0.1943359375,
      "loss_num": 0.00714111328125,
      "loss_xval": 0.42578125,
      "num_input_tokens_seen": 307265808,
      "step": 5484
    },
    {
      "epoch": 12.216035634743875,
      "grad_norm": 31.889545440673828,
      "learning_rate": 1e-06,
      "loss": 0.5601,
      "num_input_tokens_seen": 307318728,
      "step": 5485
    },
    {
      "epoch": 12.216035634743875,
      "loss": 0.5997210741043091,
      "loss_ce": 0.00011171124060638249,
      "loss_iou": 0.28125,
      "loss_num": 0.0078125,
      "loss_xval": 0.6015625,
      "num_input_tokens_seen": 307318728,
      "step": 5485
    },
    {
      "epoch": 12.21826280623608,
      "grad_norm": 13.831578254699707,
      "learning_rate": 1e-06,
      "loss": 0.4215,
      "num_input_tokens_seen": 307374860,
      "step": 5486
    },
    {
      "epoch": 12.21826280623608,
      "loss": 0.48658156394958496,
      "loss_ce": 0.00013140994997229427,
      "loss_iou": 0.2060546875,
      "loss_num": 0.014892578125,
      "loss_xval": 0.486328125,
      "num_input_tokens_seen": 307374860,
      "step": 5486
    },
    {
      "epoch": 12.220489977728285,
      "grad_norm": 19.31671714782715,
      "learning_rate": 1e-06,
      "loss": 0.4264,
      "num_input_tokens_seen": 307432404,
      "step": 5487
    },
    {
      "epoch": 12.220489977728285,
      "loss": 0.4271966814994812,
      "loss_ce": 0.00013369151565711945,
      "loss_iou": 0.1669921875,
      "loss_num": 0.018798828125,
      "loss_xval": 0.427734375,
      "num_input_tokens_seen": 307432404,
      "step": 5487
    },
    {
      "epoch": 12.22271714922049,
      "grad_norm": 27.503997802734375,
      "learning_rate": 1e-06,
      "loss": 0.4349,
      "num_input_tokens_seen": 307488920,
      "step": 5488
    },
    {
      "epoch": 12.22271714922049,
      "loss": 0.4608370065689087,
      "loss_ce": 0.00014365185052156448,
      "loss_iou": 0.1923828125,
      "loss_num": 0.01513671875,
      "loss_xval": 0.4609375,
      "num_input_tokens_seen": 307488920,
      "step": 5488
    },
    {
      "epoch": 12.224944320712694,
      "grad_norm": 20.619155883789062,
      "learning_rate": 1e-06,
      "loss": 0.5518,
      "num_input_tokens_seen": 307542936,
      "step": 5489
    },
    {
      "epoch": 12.224944320712694,
      "loss": 0.4091936945915222,
      "loss_ce": 0.00013606807624455541,
      "loss_iou": 0.1884765625,
      "loss_num": 0.006439208984375,
      "loss_xval": 0.408203125,
      "num_input_tokens_seen": 307542936,
      "step": 5489
    },
    {
      "epoch": 12.2271714922049,
      "grad_norm": 40.87490463256836,
      "learning_rate": 1e-06,
      "loss": 0.5842,
      "num_input_tokens_seen": 307599944,
      "step": 5490
    },
    {
      "epoch": 12.2271714922049,
      "loss": 0.5064796805381775,
      "loss_ce": 0.0001320538140134886,
      "loss_iou": 0.2216796875,
      "loss_num": 0.01263427734375,
      "loss_xval": 0.5078125,
      "num_input_tokens_seen": 307599944,
      "step": 5490
    },
    {
      "epoch": 12.229398663697104,
      "grad_norm": 19.404682159423828,
      "learning_rate": 1e-06,
      "loss": 0.3539,
      "num_input_tokens_seen": 307656088,
      "step": 5491
    },
    {
      "epoch": 12.229398663697104,
      "loss": 0.3526671230792999,
      "loss_ce": 0.00012805430742446333,
      "loss_iou": 0.158203125,
      "loss_num": 0.007232666015625,
      "loss_xval": 0.3515625,
      "num_input_tokens_seen": 307656088,
      "step": 5491
    },
    {
      "epoch": 12.231625835189309,
      "grad_norm": 15.24465560913086,
      "learning_rate": 1e-06,
      "loss": 0.6326,
      "num_input_tokens_seen": 307711560,
      "step": 5492
    },
    {
      "epoch": 12.231625835189309,
      "loss": 0.7728477120399475,
      "loss_ce": 0.00014267500955611467,
      "loss_iou": 0.306640625,
      "loss_num": 0.03173828125,
      "loss_xval": 0.7734375,
      "num_input_tokens_seen": 307711560,
      "step": 5492
    },
    {
      "epoch": 12.233853006681514,
      "grad_norm": 20.780717849731445,
      "learning_rate": 1e-06,
      "loss": 0.6387,
      "num_input_tokens_seen": 307767096,
      "step": 5493
    },
    {
      "epoch": 12.233853006681514,
      "loss": 0.6877992153167725,
      "loss_ce": 0.00011609600915107876,
      "loss_iou": 0.244140625,
      "loss_num": 0.0400390625,
      "loss_xval": 0.6875,
      "num_input_tokens_seen": 307767096,
      "step": 5493
    },
    {
      "epoch": 12.236080178173719,
      "grad_norm": 38.66435241699219,
      "learning_rate": 1e-06,
      "loss": 0.4721,
      "num_input_tokens_seen": 307824084,
      "step": 5494
    },
    {
      "epoch": 12.236080178173719,
      "loss": 0.4882151782512665,
      "loss_ce": 0.0001780918682925403,
      "loss_iou": 0.2080078125,
      "loss_num": 0.01422119140625,
      "loss_xval": 0.48828125,
      "num_input_tokens_seen": 307824084,
      "step": 5494
    },
    {
      "epoch": 12.238307349665924,
      "grad_norm": 169.19766235351562,
      "learning_rate": 1e-06,
      "loss": 0.5511,
      "num_input_tokens_seen": 307878876,
      "step": 5495
    },
    {
      "epoch": 12.238307349665924,
      "loss": 0.6929959058761597,
      "loss_ce": 0.00012478306598495692,
      "loss_iou": 0.26171875,
      "loss_num": 0.033935546875,
      "loss_xval": 0.69140625,
      "num_input_tokens_seen": 307878876,
      "step": 5495
    },
    {
      "epoch": 12.240534521158128,
      "grad_norm": 18.359050750732422,
      "learning_rate": 1e-06,
      "loss": 0.4092,
      "num_input_tokens_seen": 307936548,
      "step": 5496
    },
    {
      "epoch": 12.240534521158128,
      "loss": 0.40076589584350586,
      "loss_ce": 0.00013115604815538973,
      "loss_iou": 0.185546875,
      "loss_num": 0.005828857421875,
      "loss_xval": 0.400390625,
      "num_input_tokens_seen": 307936548,
      "step": 5496
    },
    {
      "epoch": 12.242761692650333,
      "grad_norm": 16.64657211303711,
      "learning_rate": 1e-06,
      "loss": 0.5673,
      "num_input_tokens_seen": 307993096,
      "step": 5497
    },
    {
      "epoch": 12.242761692650333,
      "loss": 0.6693442463874817,
      "loss_ce": 0.0001547938009025529,
      "loss_iou": 0.2890625,
      "loss_num": 0.01806640625,
      "loss_xval": 0.66796875,
      "num_input_tokens_seen": 307993096,
      "step": 5497
    },
    {
      "epoch": 12.244988864142538,
      "grad_norm": 23.77404022216797,
      "learning_rate": 1e-06,
      "loss": 0.6996,
      "num_input_tokens_seen": 308046960,
      "step": 5498
    },
    {
      "epoch": 12.244988864142538,
      "loss": 0.610221266746521,
      "loss_ce": 0.00011382724915165454,
      "loss_iou": 0.2578125,
      "loss_num": 0.0191650390625,
      "loss_xval": 0.609375,
      "num_input_tokens_seen": 308046960,
      "step": 5498
    },
    {
      "epoch": 12.247216035634743,
      "grad_norm": 19.345346450805664,
      "learning_rate": 1e-06,
      "loss": 0.3564,
      "num_input_tokens_seen": 308103812,
      "step": 5499
    },
    {
      "epoch": 12.247216035634743,
      "loss": 0.3957526683807373,
      "loss_ce": 0.00012281053932383657,
      "loss_iou": 0.1806640625,
      "loss_num": 0.006805419921875,
      "loss_xval": 0.396484375,
      "num_input_tokens_seen": 308103812,
      "step": 5499
    },
    {
      "epoch": 12.249443207126948,
      "grad_norm": 18.410499572753906,
      "learning_rate": 1e-06,
      "loss": 0.5145,
      "num_input_tokens_seen": 308158408,
      "step": 5500
    },
    {
      "epoch": 12.249443207126948,
      "eval_seeclick_web_CIoU": 0.585688054561615,
      "eval_seeclick_web_GIoU": 0.5826278626918793,
      "eval_seeclick_web_IoU": 0.6041875779628754,
      "eval_seeclick_web_MAE_all": 0.015673364512622356,
      "eval_seeclick_web_MAE_h": 0.007765157613903284,
      "eval_seeclick_web_MAE_w": 0.015952853485941887,
      "eval_seeclick_web_MAE_x_boxes": 0.009551033610478044,
      "eval_seeclick_web_MAE_y_boxes": 0.02141634118743241,
      "eval_seeclick_web_inside_bbox": 0.9166666567325592,
      "eval_seeclick_web_loss": 0.9183205962181091,
      "eval_seeclick_web_loss_ce": 0.00018939840811071917,
      "eval_seeclick_web_loss_iou": 0.4190673828125,
      "eval_seeclick_web_loss_num": 0.0124969482421875,
      "eval_seeclick_web_loss_xval": 0.900390625,
      "eval_seeclick_web_runtime": 24.0722,
      "eval_seeclick_web_samples_per_second": 2.077,
      "eval_seeclick_web_steps_per_second": 0.083,
      "num_input_tokens_seen": 308158408,
      "step": 5500
    },
    {
      "epoch": 12.249443207126948,
      "eval_icons_CIoU": 0.2844693139195442,
      "eval_icons_GIoU": 0.29983824491500854,
      "eval_icons_IoU": 0.3569711297750473,
      "eval_icons_MAE_all": 0.056874074041843414,
      "eval_icons_MAE_h": 0.039589228108525276,
      "eval_icons_MAE_w": 0.05369975045323372,
      "eval_icons_MAE_x_boxes": 0.051283443346619606,
      "eval_icons_MAE_y_boxes": 0.03893454186618328,
      "eval_icons_inside_bbox": 0.6215277910232544,
      "eval_icons_loss": 1.6849582195281982,
      "eval_icons_loss_ce": 0.00027994449192192405,
      "eval_icons_loss_iou": 0.662353515625,
      "eval_icons_loss_num": 0.04971122741699219,
      "eval_icons_loss_xval": 1.573486328125,
      "eval_icons_runtime": 23.9366,
      "eval_icons_samples_per_second": 2.089,
      "eval_icons_steps_per_second": 0.084,
      "num_input_tokens_seen": 308158408,
      "step": 5500
    },
    {
      "epoch": 12.249443207126948,
      "eval_screenspot_CIoU": 0.35463671882947284,
      "eval_screenspot_GIoU": 0.36913161476453143,
      "eval_screenspot_IoU": 0.43245549003283185,
      "eval_screenspot_MAE_all": 0.058157578110694885,
      "eval_screenspot_MAE_h": 0.038887947176893554,
      "eval_screenspot_MAE_w": 0.067777914305528,
      "eval_screenspot_MAE_x_boxes": 0.07268660329282284,
      "eval_screenspot_MAE_y_boxes": 0.037438808319469295,
      "eval_screenspot_inside_bbox": 0.6966666579246521,
      "eval_screenspot_loss": 1.6075245141983032,
      "eval_screenspot_loss_ce": 0.0002708134804076205,
      "eval_screenspot_loss_iou": 0.6668294270833334,
      "eval_screenspot_loss_num": 0.06592305501302083,
      "eval_screenspot_loss_xval": 1.66162109375,
      "eval_screenspot_runtime": 38.3047,
      "eval_screenspot_samples_per_second": 2.323,
      "eval_screenspot_steps_per_second": 0.078,
      "num_input_tokens_seen": 308158408,
      "step": 5500
    },
    {
      "epoch": 12.249443207126948,
      "eval_compot_CIoU": 0.34248843789100647,
      "eval_compot_GIoU": 0.3517104983329773,
      "eval_compot_IoU": 0.40232492983341217,
      "eval_compot_MAE_all": 0.017755805049091578,
      "eval_compot_MAE_h": 0.009214944671839476,
      "eval_compot_MAE_w": 0.02113647386431694,
      "eval_compot_MAE_x_boxes": 0.029819749295711517,
      "eval_compot_MAE_y_boxes": 0.0068535758182406425,
      "eval_compot_inside_bbox": 0.6302083432674408,
      "eval_compot_loss": 1.4059686660766602,
      "eval_compot_loss_ce": 0.00018623641517478973,
      "eval_compot_loss_iou": 0.65087890625,
      "eval_compot_loss_num": 0.016448974609375,
      "eval_compot_loss_xval": 1.3837890625,
      "eval_compot_runtime": 24.7377,
      "eval_compot_samples_per_second": 2.021,
      "eval_compot_steps_per_second": 0.081,
      "num_input_tokens_seen": 308158408,
      "step": 5500
    },
    {
      "epoch": 12.249443207126948,
      "eval_custom_ui_val_CIoU": 0.47646190888351864,
      "eval_custom_ui_val_GIoU": 0.4871201482084062,
      "eval_custom_ui_val_IoU": 0.5378114382425944,
      "eval_custom_ui_val_MAE_all": 0.02943614311516285,
      "eval_custom_ui_val_MAE_h": 0.016217040493049555,
      "eval_custom_ui_val_MAE_w": 0.03668393205023474,
      "eval_custom_ui_val_MAE_x_boxes": 0.03531148243281576,
      "eval_custom_ui_val_MAE_y_boxes": 0.014928720322334103,
      "eval_custom_ui_val_inside_bbox": 0.7719907429483202,
      "eval_custom_ui_val_loss": 1.2021739482879639,
      "eval_custom_ui_val_loss_ce": 0.00023076724998342493,
      "eval_custom_ui_val_loss_iou": 0.5134412977430556,
      "eval_custom_ui_val_loss_num": 0.027161492241753474,
      "eval_custom_ui_val_loss_xval": 1.1628689236111112,
      "eval_custom_ui_val_runtime": 76.3883,
      "eval_custom_ui_val_samples_per_second": 3.469,
      "eval_custom_ui_val_steps_per_second": 0.118,
      "num_input_tokens_seen": 308158408,
      "step": 5500
    },
    {
      "epoch": 12.249443207126948,
      "loss": 0.8959341049194336,
      "loss_ce": 0.00018216308671981096,
      "loss_iou": 0.390625,
      "loss_num": 0.0228271484375,
      "loss_xval": 0.89453125,
      "num_input_tokens_seen": 308158408,
      "step": 5500
    },
    {
      "epoch": 12.251670378619155,
      "grad_norm": 15.896366119384766,
      "learning_rate": 1e-06,
      "loss": 0.3427,
      "num_input_tokens_seen": 308216040,
      "step": 5501
    },
    {
      "epoch": 12.251670378619155,
      "loss": 0.28972911834716797,
      "loss_ce": 0.0001172900665551424,
      "loss_iou": 0.11962890625,
      "loss_num": 0.010009765625,
      "loss_xval": 0.2890625,
      "num_input_tokens_seen": 308216040,
      "step": 5501
    },
    {
      "epoch": 12.25389755011136,
      "grad_norm": 31.605810165405273,
      "learning_rate": 1e-06,
      "loss": 0.4781,
      "num_input_tokens_seen": 308272560,
      "step": 5502
    },
    {
      "epoch": 12.25389755011136,
      "loss": 0.5278470516204834,
      "loss_ce": 0.00013708796177525073,
      "loss_iou": 0.244140625,
      "loss_num": 0.00787353515625,
      "loss_xval": 0.52734375,
      "num_input_tokens_seen": 308272560,
      "step": 5502
    },
    {
      "epoch": 12.256124721603564,
      "grad_norm": 18.19232940673828,
      "learning_rate": 1e-06,
      "loss": 0.5417,
      "num_input_tokens_seen": 308330492,
      "step": 5503
    },
    {
      "epoch": 12.256124721603564,
      "loss": 0.5479559302330017,
      "loss_ce": 0.00010438320168759674,
      "loss_iou": 0.2421875,
      "loss_num": 0.01263427734375,
      "loss_xval": 0.546875,
      "num_input_tokens_seen": 308330492,
      "step": 5503
    },
    {
      "epoch": 12.25835189309577,
      "grad_norm": 14.01863956451416,
      "learning_rate": 1e-06,
      "loss": 0.6479,
      "num_input_tokens_seen": 308385352,
      "step": 5504
    },
    {
      "epoch": 12.25835189309577,
      "loss": 0.4697185158729553,
      "loss_ce": 0.00011405147233745083,
      "loss_iou": 0.193359375,
      "loss_num": 0.0166015625,
      "loss_xval": 0.46875,
      "num_input_tokens_seen": 308385352,
      "step": 5504
    },
    {
      "epoch": 12.260579064587974,
      "grad_norm": 21.31361198425293,
      "learning_rate": 1e-06,
      "loss": 0.7673,
      "num_input_tokens_seen": 308443100,
      "step": 5505
    },
    {
      "epoch": 12.260579064587974,
      "loss": 0.7160589694976807,
      "loss_ce": 0.00011650074156932533,
      "loss_iou": 0.29296875,
      "loss_num": 0.025634765625,
      "loss_xval": 0.71484375,
      "num_input_tokens_seen": 308443100,
      "step": 5505
    },
    {
      "epoch": 12.262806236080179,
      "grad_norm": 199.44406127929688,
      "learning_rate": 1e-06,
      "loss": 0.4544,
      "num_input_tokens_seen": 308499652,
      "step": 5506
    },
    {
      "epoch": 12.262806236080179,
      "loss": 0.556039035320282,
      "loss_ce": 0.0001308466453338042,
      "loss_iou": 0.251953125,
      "loss_num": 0.01007080078125,
      "loss_xval": 0.5546875,
      "num_input_tokens_seen": 308499652,
      "step": 5506
    },
    {
      "epoch": 12.265033407572384,
      "grad_norm": 28.171159744262695,
      "learning_rate": 1e-06,
      "loss": 0.4748,
      "num_input_tokens_seen": 308551468,
      "step": 5507
    },
    {
      "epoch": 12.265033407572384,
      "loss": 0.4873943626880646,
      "loss_ce": 0.00021174979337956756,
      "loss_iou": 0.212890625,
      "loss_num": 0.01220703125,
      "loss_xval": 0.486328125,
      "num_input_tokens_seen": 308551468,
      "step": 5507
    },
    {
      "epoch": 12.267260579064589,
      "grad_norm": 17.694232940673828,
      "learning_rate": 1e-06,
      "loss": 0.5427,
      "num_input_tokens_seen": 308607228,
      "step": 5508
    },
    {
      "epoch": 12.267260579064589,
      "loss": 0.3442959189414978,
      "loss_ce": 0.00011866106069646776,
      "loss_iou": 0.134765625,
      "loss_num": 0.01513671875,
      "loss_xval": 0.34375,
      "num_input_tokens_seen": 308607228,
      "step": 5508
    },
    {
      "epoch": 12.269487750556793,
      "grad_norm": 17.69139289855957,
      "learning_rate": 1e-06,
      "loss": 0.5362,
      "num_input_tokens_seen": 308664588,
      "step": 5509
    },
    {
      "epoch": 12.269487750556793,
      "loss": 0.4937598705291748,
      "loss_ce": 0.00010753308015409857,
      "loss_iou": 0.21875,
      "loss_num": 0.010986328125,
      "loss_xval": 0.494140625,
      "num_input_tokens_seen": 308664588,
      "step": 5509
    },
    {
      "epoch": 12.271714922048998,
      "grad_norm": 18.234146118164062,
      "learning_rate": 1e-06,
      "loss": 0.6428,
      "num_input_tokens_seen": 308716208,
      "step": 5510
    },
    {
      "epoch": 12.271714922048998,
      "loss": 0.5791558027267456,
      "loss_ce": 0.00014576420653611422,
      "loss_iou": 0.26171875,
      "loss_num": 0.0107421875,
      "loss_xval": 0.578125,
      "num_input_tokens_seen": 308716208,
      "step": 5510
    },
    {
      "epoch": 12.273942093541203,
      "grad_norm": 18.1691951751709,
      "learning_rate": 1e-06,
      "loss": 0.5535,
      "num_input_tokens_seen": 308772680,
      "step": 5511
    },
    {
      "epoch": 12.273942093541203,
      "loss": 0.5991677045822144,
      "loss_ce": 0.00016867450904101133,
      "loss_iou": 0.26171875,
      "loss_num": 0.0147705078125,
      "loss_xval": 0.59765625,
      "num_input_tokens_seen": 308772680,
      "step": 5511
    },
    {
      "epoch": 12.276169265033408,
      "grad_norm": 15.399046897888184,
      "learning_rate": 1e-06,
      "loss": 0.4188,
      "num_input_tokens_seen": 308829192,
      "step": 5512
    },
    {
      "epoch": 12.276169265033408,
      "loss": 0.5632211565971375,
      "loss_ce": 0.0001413593563484028,
      "loss_iou": 0.2001953125,
      "loss_num": 0.032470703125,
      "loss_xval": 0.5625,
      "num_input_tokens_seen": 308829192,
      "step": 5512
    },
    {
      "epoch": 12.278396436525613,
      "grad_norm": 15.511956214904785,
      "learning_rate": 1e-06,
      "loss": 0.5697,
      "num_input_tokens_seen": 308886748,
      "step": 5513
    },
    {
      "epoch": 12.278396436525613,
      "loss": 0.5418818593025208,
      "loss_ce": 0.00013380208110902458,
      "loss_iou": 0.240234375,
      "loss_num": 0.01239013671875,
      "loss_xval": 0.54296875,
      "num_input_tokens_seen": 308886748,
      "step": 5513
    },
    {
      "epoch": 12.280623608017818,
      "grad_norm": 14.026480674743652,
      "learning_rate": 1e-06,
      "loss": 0.6363,
      "num_input_tokens_seen": 308942736,
      "step": 5514
    },
    {
      "epoch": 12.280623608017818,
      "loss": 0.7164598703384399,
      "loss_ce": 0.0001512969029136002,
      "loss_iou": 0.259765625,
      "loss_num": 0.0390625,
      "loss_xval": 0.71484375,
      "num_input_tokens_seen": 308942736,
      "step": 5514
    },
    {
      "epoch": 12.282850779510023,
      "grad_norm": 16.546672821044922,
      "learning_rate": 1e-06,
      "loss": 0.4416,
      "num_input_tokens_seen": 308997492,
      "step": 5515
    },
    {
      "epoch": 12.282850779510023,
      "loss": 0.3985058069229126,
      "loss_ce": 0.00012933829566463828,
      "loss_iou": 0.166015625,
      "loss_num": 0.01318359375,
      "loss_xval": 0.3984375,
      "num_input_tokens_seen": 308997492,
      "step": 5515
    },
    {
      "epoch": 12.285077951002227,
      "grad_norm": 16.246435165405273,
      "learning_rate": 1e-06,
      "loss": 0.517,
      "num_input_tokens_seen": 309054160,
      "step": 5516
    },
    {
      "epoch": 12.285077951002227,
      "loss": 0.4222395122051239,
      "loss_ce": 0.00012038354179821908,
      "loss_iou": 0.1953125,
      "loss_num": 0.006256103515625,
      "loss_xval": 0.421875,
      "num_input_tokens_seen": 309054160,
      "step": 5516
    },
    {
      "epoch": 12.287305122494432,
      "grad_norm": 16.089317321777344,
      "learning_rate": 1e-06,
      "loss": 0.4671,
      "num_input_tokens_seen": 309109496,
      "step": 5517
    },
    {
      "epoch": 12.287305122494432,
      "loss": 0.48379087448120117,
      "loss_ce": 0.00014830243890173733,
      "loss_iou": 0.2109375,
      "loss_num": 0.01214599609375,
      "loss_xval": 0.484375,
      "num_input_tokens_seen": 309109496,
      "step": 5517
    },
    {
      "epoch": 12.289532293986637,
      "grad_norm": 14.831037521362305,
      "learning_rate": 1e-06,
      "loss": 0.4507,
      "num_input_tokens_seen": 309166044,
      "step": 5518
    },
    {
      "epoch": 12.289532293986637,
      "loss": 0.6069037914276123,
      "loss_ce": 0.00015331347822211683,
      "loss_iou": 0.275390625,
      "loss_num": 0.0115966796875,
      "loss_xval": 0.60546875,
      "num_input_tokens_seen": 309166044,
      "step": 5518
    },
    {
      "epoch": 12.291759465478842,
      "grad_norm": 16.727554321289062,
      "learning_rate": 1e-06,
      "loss": 0.5212,
      "num_input_tokens_seen": 309224472,
      "step": 5519
    },
    {
      "epoch": 12.291759465478842,
      "loss": 0.37400585412979126,
      "loss_ce": 0.00010447671229485422,
      "loss_iou": 0.1611328125,
      "loss_num": 0.0103759765625,
      "loss_xval": 0.373046875,
      "num_input_tokens_seen": 309224472,
      "step": 5519
    },
    {
      "epoch": 12.293986636971047,
      "grad_norm": 16.9213809967041,
      "learning_rate": 1e-06,
      "loss": 0.8055,
      "num_input_tokens_seen": 309281284,
      "step": 5520
    },
    {
      "epoch": 12.293986636971047,
      "loss": 0.9495736360549927,
      "loss_ce": 0.00011074876238126308,
      "loss_iou": 0.373046875,
      "loss_num": 0.040771484375,
      "loss_xval": 0.94921875,
      "num_input_tokens_seen": 309281284,
      "step": 5520
    },
    {
      "epoch": 12.296213808463252,
      "grad_norm": 24.03369140625,
      "learning_rate": 1e-06,
      "loss": 0.5665,
      "num_input_tokens_seen": 309336800,
      "step": 5521
    },
    {
      "epoch": 12.296213808463252,
      "loss": 0.5887036323547363,
      "loss_ce": 0.0003247222339268774,
      "loss_iou": 0.232421875,
      "loss_num": 0.0250244140625,
      "loss_xval": 0.58984375,
      "num_input_tokens_seen": 309336800,
      "step": 5521
    },
    {
      "epoch": 12.298440979955457,
      "grad_norm": 17.732921600341797,
      "learning_rate": 1e-06,
      "loss": 0.6271,
      "num_input_tokens_seen": 309390848,
      "step": 5522
    },
    {
      "epoch": 12.298440979955457,
      "loss": 0.858605146408081,
      "loss_ce": 0.000206666489248164,
      "loss_iou": 0.365234375,
      "loss_num": 0.025634765625,
      "loss_xval": 0.859375,
      "num_input_tokens_seen": 309390848,
      "step": 5522
    },
    {
      "epoch": 12.300668151447661,
      "grad_norm": 14.111288070678711,
      "learning_rate": 1e-06,
      "loss": 0.5704,
      "num_input_tokens_seen": 309447824,
      "step": 5523
    },
    {
      "epoch": 12.300668151447661,
      "loss": 0.3774692416191101,
      "loss_ce": 0.00014990594354458153,
      "loss_iou": 0.154296875,
      "loss_num": 0.0135498046875,
      "loss_xval": 0.376953125,
      "num_input_tokens_seen": 309447824,
      "step": 5523
    },
    {
      "epoch": 12.302895322939866,
      "grad_norm": 19.964794158935547,
      "learning_rate": 1e-06,
      "loss": 0.4378,
      "num_input_tokens_seen": 309504964,
      "step": 5524
    },
    {
      "epoch": 12.302895322939866,
      "loss": 0.3729360103607178,
      "loss_ce": 0.00013328931527212262,
      "loss_iou": 0.1669921875,
      "loss_num": 0.0076904296875,
      "loss_xval": 0.373046875,
      "num_input_tokens_seen": 309504964,
      "step": 5524
    },
    {
      "epoch": 12.305122494432071,
      "grad_norm": 14.157723426818848,
      "learning_rate": 1e-06,
      "loss": 0.4644,
      "num_input_tokens_seen": 309560456,
      "step": 5525
    },
    {
      "epoch": 12.305122494432071,
      "loss": 0.5831543207168579,
      "loss_ce": 0.00014654998085461557,
      "loss_iou": 0.2177734375,
      "loss_num": 0.0296630859375,
      "loss_xval": 0.58203125,
      "num_input_tokens_seen": 309560456,
      "step": 5525
    },
    {
      "epoch": 12.307349665924276,
      "grad_norm": 22.95296287536621,
      "learning_rate": 1e-06,
      "loss": 0.5372,
      "num_input_tokens_seen": 309619680,
      "step": 5526
    },
    {
      "epoch": 12.307349665924276,
      "loss": 0.5235176086425781,
      "loss_ce": 0.00020221451995894313,
      "loss_iou": 0.2041015625,
      "loss_num": 0.0230712890625,
      "loss_xval": 0.5234375,
      "num_input_tokens_seen": 309619680,
      "step": 5526
    },
    {
      "epoch": 12.309576837416481,
      "grad_norm": 14.134220123291016,
      "learning_rate": 1e-06,
      "loss": 0.4565,
      "num_input_tokens_seen": 309676328,
      "step": 5527
    },
    {
      "epoch": 12.309576837416481,
      "loss": 0.35591933131217957,
      "loss_ce": 0.00020644423784688115,
      "loss_iou": 0.1611328125,
      "loss_num": 0.00677490234375,
      "loss_xval": 0.35546875,
      "num_input_tokens_seen": 309676328,
      "step": 5527
    },
    {
      "epoch": 12.311804008908686,
      "grad_norm": 26.886777877807617,
      "learning_rate": 1e-06,
      "loss": 0.5358,
      "num_input_tokens_seen": 309727028,
      "step": 5528
    },
    {
      "epoch": 12.311804008908686,
      "loss": 0.5397249460220337,
      "loss_ce": 0.0001741335727274418,
      "loss_iou": 0.236328125,
      "loss_num": 0.0133056640625,
      "loss_xval": 0.5390625,
      "num_input_tokens_seen": 309727028,
      "step": 5528
    },
    {
      "epoch": 12.31403118040089,
      "grad_norm": 17.852651596069336,
      "learning_rate": 1e-06,
      "loss": 0.5016,
      "num_input_tokens_seen": 309783832,
      "step": 5529
    },
    {
      "epoch": 12.31403118040089,
      "loss": 0.4710628390312195,
      "loss_ce": 0.00011555143282748759,
      "loss_iou": 0.2060546875,
      "loss_num": 0.01165771484375,
      "loss_xval": 0.470703125,
      "num_input_tokens_seen": 309783832,
      "step": 5529
    },
    {
      "epoch": 12.316258351893095,
      "grad_norm": 14.421836853027344,
      "learning_rate": 1e-06,
      "loss": 0.4286,
      "num_input_tokens_seen": 309839264,
      "step": 5530
    },
    {
      "epoch": 12.316258351893095,
      "loss": 0.29735976457595825,
      "loss_ce": 0.00011855886259581894,
      "loss_iou": 0.1298828125,
      "loss_num": 0.00738525390625,
      "loss_xval": 0.296875,
      "num_input_tokens_seen": 309839264,
      "step": 5530
    },
    {
      "epoch": 12.3184855233853,
      "grad_norm": 41.36547088623047,
      "learning_rate": 1e-06,
      "loss": 0.5509,
      "num_input_tokens_seen": 309893232,
      "step": 5531
    },
    {
      "epoch": 12.3184855233853,
      "loss": 0.648200511932373,
      "loss_ce": 0.00012922041059937328,
      "loss_iou": 0.296875,
      "loss_num": 0.0107421875,
      "loss_xval": 0.6484375,
      "num_input_tokens_seen": 309893232,
      "step": 5531
    },
    {
      "epoch": 12.320712694877505,
      "grad_norm": 22.48734474182129,
      "learning_rate": 1e-06,
      "loss": 0.505,
      "num_input_tokens_seen": 309948492,
      "step": 5532
    },
    {
      "epoch": 12.320712694877505,
      "loss": 0.4604489803314209,
      "loss_ce": 0.00012181737110950053,
      "loss_iou": 0.2060546875,
      "loss_num": 0.009521484375,
      "loss_xval": 0.4609375,
      "num_input_tokens_seen": 309948492,
      "step": 5532
    },
    {
      "epoch": 12.32293986636971,
      "grad_norm": 17.8016414642334,
      "learning_rate": 1e-06,
      "loss": 0.3587,
      "num_input_tokens_seen": 310005616,
      "step": 5533
    },
    {
      "epoch": 12.32293986636971,
      "loss": 0.39793771505355835,
      "loss_ce": 0.0001105824630940333,
      "loss_iou": 0.1806640625,
      "loss_num": 0.00726318359375,
      "loss_xval": 0.3984375,
      "num_input_tokens_seen": 310005616,
      "step": 5533
    },
    {
      "epoch": 12.325167037861915,
      "grad_norm": 17.646106719970703,
      "learning_rate": 1e-06,
      "loss": 0.581,
      "num_input_tokens_seen": 310061764,
      "step": 5534
    },
    {
      "epoch": 12.325167037861915,
      "loss": 0.666999101638794,
      "loss_ce": 0.00012893178791273385,
      "loss_iou": 0.2734375,
      "loss_num": 0.024169921875,
      "loss_xval": 0.66796875,
      "num_input_tokens_seen": 310061764,
      "step": 5534
    },
    {
      "epoch": 12.32739420935412,
      "grad_norm": 19.61566925048828,
      "learning_rate": 1e-06,
      "loss": 0.5461,
      "num_input_tokens_seen": 310116556,
      "step": 5535
    },
    {
      "epoch": 12.32739420935412,
      "loss": 0.31176358461380005,
      "loss_ce": 0.00011805207759607583,
      "loss_iou": 0.1416015625,
      "loss_num": 0.005584716796875,
      "loss_xval": 0.3125,
      "num_input_tokens_seen": 310116556,
      "step": 5535
    },
    {
      "epoch": 12.329621380846325,
      "grad_norm": 14.809009552001953,
      "learning_rate": 1e-06,
      "loss": 0.4208,
      "num_input_tokens_seen": 310173568,
      "step": 5536
    },
    {
      "epoch": 12.329621380846325,
      "loss": 0.40904849767684937,
      "loss_ce": 0.00011295877629891038,
      "loss_iou": 0.1845703125,
      "loss_num": 0.00799560546875,
      "loss_xval": 0.408203125,
      "num_input_tokens_seen": 310173568,
      "step": 5536
    },
    {
      "epoch": 12.33184855233853,
      "grad_norm": 26.909616470336914,
      "learning_rate": 1e-06,
      "loss": 0.4644,
      "num_input_tokens_seen": 310228872,
      "step": 5537
    },
    {
      "epoch": 12.33184855233853,
      "loss": 0.580930233001709,
      "loss_ce": 0.00011961960990447551,
      "loss_iou": 0.240234375,
      "loss_num": 0.0198974609375,
      "loss_xval": 0.58203125,
      "num_input_tokens_seen": 310228872,
      "step": 5537
    },
    {
      "epoch": 12.334075723830734,
      "grad_norm": 18.16213035583496,
      "learning_rate": 1e-06,
      "loss": 0.6525,
      "num_input_tokens_seen": 310284508,
      "step": 5538
    },
    {
      "epoch": 12.334075723830734,
      "loss": 0.5219680070877075,
      "loss_ce": 0.00011739489855244756,
      "loss_iou": 0.2353515625,
      "loss_num": 0.01019287109375,
      "loss_xval": 0.5234375,
      "num_input_tokens_seen": 310284508,
      "step": 5538
    },
    {
      "epoch": 12.33630289532294,
      "grad_norm": 32.21243667602539,
      "learning_rate": 1e-06,
      "loss": 0.4215,
      "num_input_tokens_seen": 310340144,
      "step": 5539
    },
    {
      "epoch": 12.33630289532294,
      "loss": 0.31780463457107544,
      "loss_ce": 0.00011664302292047068,
      "loss_iou": 0.1318359375,
      "loss_num": 0.01092529296875,
      "loss_xval": 0.318359375,
      "num_input_tokens_seen": 310340144,
      "step": 5539
    },
    {
      "epoch": 12.338530066815144,
      "grad_norm": 23.456138610839844,
      "learning_rate": 1e-06,
      "loss": 0.3759,
      "num_input_tokens_seen": 310398476,
      "step": 5540
    },
    {
      "epoch": 12.338530066815144,
      "loss": 0.36424580216407776,
      "loss_ce": 0.00011004651605617255,
      "loss_iou": 0.158203125,
      "loss_num": 0.00946044921875,
      "loss_xval": 0.36328125,
      "num_input_tokens_seen": 310398476,
      "step": 5540
    },
    {
      "epoch": 12.340757238307349,
      "grad_norm": 18.373727798461914,
      "learning_rate": 1e-06,
      "loss": 0.4471,
      "num_input_tokens_seen": 310456252,
      "step": 5541
    },
    {
      "epoch": 12.340757238307349,
      "loss": 0.4610535502433777,
      "loss_ce": 0.0001160675601568073,
      "loss_iou": 0.2060546875,
      "loss_num": 0.009765625,
      "loss_xval": 0.4609375,
      "num_input_tokens_seen": 310456252,
      "step": 5541
    },
    {
      "epoch": 12.342984409799554,
      "grad_norm": 15.688360214233398,
      "learning_rate": 1e-06,
      "loss": 0.4345,
      "num_input_tokens_seen": 310509648,
      "step": 5542
    },
    {
      "epoch": 12.342984409799554,
      "loss": 0.33905208110809326,
      "loss_ce": 0.00012385296577122062,
      "loss_iou": 0.1484375,
      "loss_num": 0.0081787109375,
      "loss_xval": 0.33984375,
      "num_input_tokens_seen": 310509648,
      "step": 5542
    },
    {
      "epoch": 12.345211581291759,
      "grad_norm": 24.152406692504883,
      "learning_rate": 1e-06,
      "loss": 0.4911,
      "num_input_tokens_seen": 310565268,
      "step": 5543
    },
    {
      "epoch": 12.345211581291759,
      "loss": 0.45101678371429443,
      "loss_ce": 0.00015007876208983362,
      "loss_iou": 0.1923828125,
      "loss_num": 0.01312255859375,
      "loss_xval": 0.451171875,
      "num_input_tokens_seen": 310565268,
      "step": 5543
    },
    {
      "epoch": 12.347438752783964,
      "grad_norm": 18.96198081970215,
      "learning_rate": 1e-06,
      "loss": 0.3821,
      "num_input_tokens_seen": 310618196,
      "step": 5544
    },
    {
      "epoch": 12.347438752783964,
      "loss": 0.4492288827896118,
      "loss_ce": 0.0001322347525274381,
      "loss_iou": 0.2060546875,
      "loss_num": 0.007476806640625,
      "loss_xval": 0.44921875,
      "num_input_tokens_seen": 310618196,
      "step": 5544
    },
    {
      "epoch": 12.34966592427617,
      "grad_norm": 21.896156311035156,
      "learning_rate": 1e-06,
      "loss": 0.468,
      "num_input_tokens_seen": 310673820,
      "step": 5545
    },
    {
      "epoch": 12.34966592427617,
      "loss": 0.3710940182209015,
      "loss_ce": 0.00012234911264386028,
      "loss_iou": 0.1640625,
      "loss_num": 0.00848388671875,
      "loss_xval": 0.37109375,
      "num_input_tokens_seen": 310673820,
      "step": 5545
    },
    {
      "epoch": 12.351893095768375,
      "grad_norm": 33.179386138916016,
      "learning_rate": 1e-06,
      "loss": 0.5911,
      "num_input_tokens_seen": 310725344,
      "step": 5546
    },
    {
      "epoch": 12.351893095768375,
      "loss": 0.85118567943573,
      "loss_ce": 0.0008438542135991156,
      "loss_iou": 0.375,
      "loss_num": 0.02001953125,
      "loss_xval": 0.8515625,
      "num_input_tokens_seen": 310725344,
      "step": 5546
    },
    {
      "epoch": 12.35412026726058,
      "grad_norm": 24.786273956298828,
      "learning_rate": 1e-06,
      "loss": 0.5604,
      "num_input_tokens_seen": 310780808,
      "step": 5547
    },
    {
      "epoch": 12.35412026726058,
      "loss": 0.37451988458633423,
      "loss_ce": 0.00025231053587049246,
      "loss_iou": 0.16796875,
      "loss_num": 0.007720947265625,
      "loss_xval": 0.375,
      "num_input_tokens_seen": 310780808,
      "step": 5547
    },
    {
      "epoch": 12.356347438752785,
      "grad_norm": 22.51511573791504,
      "learning_rate": 1e-06,
      "loss": 0.5251,
      "num_input_tokens_seen": 310839704,
      "step": 5548
    },
    {
      "epoch": 12.356347438752785,
      "loss": 0.6323890089988708,
      "loss_ce": 0.00012582968338392675,
      "loss_iou": 0.25390625,
      "loss_num": 0.024658203125,
      "loss_xval": 0.6328125,
      "num_input_tokens_seen": 310839704,
      "step": 5548
    },
    {
      "epoch": 12.35857461024499,
      "grad_norm": 19.136016845703125,
      "learning_rate": 1e-06,
      "loss": 0.6059,
      "num_input_tokens_seen": 310896396,
      "step": 5549
    },
    {
      "epoch": 12.35857461024499,
      "loss": 0.5548136234283447,
      "loss_ce": 0.00012610982230398804,
      "loss_iou": 0.2412109375,
      "loss_num": 0.0142822265625,
      "loss_xval": 0.5546875,
      "num_input_tokens_seen": 310896396,
      "step": 5549
    },
    {
      "epoch": 12.360801781737194,
      "grad_norm": 18.60460090637207,
      "learning_rate": 1e-06,
      "loss": 0.4583,
      "num_input_tokens_seen": 310954004,
      "step": 5550
    },
    {
      "epoch": 12.360801781737194,
      "loss": 0.4283403754234314,
      "loss_ce": 0.00011774353333748877,
      "loss_iou": 0.1796875,
      "loss_num": 0.0137939453125,
      "loss_xval": 0.427734375,
      "num_input_tokens_seen": 310954004,
      "step": 5550
    },
    {
      "epoch": 12.3630289532294,
      "grad_norm": 21.40949249267578,
      "learning_rate": 1e-06,
      "loss": 0.6223,
      "num_input_tokens_seen": 311010500,
      "step": 5551
    },
    {
      "epoch": 12.3630289532294,
      "loss": 0.8038879632949829,
      "loss_ce": 0.00017696505528874695,
      "loss_iou": 0.314453125,
      "loss_num": 0.034912109375,
      "loss_xval": 0.8046875,
      "num_input_tokens_seen": 311010500,
      "step": 5551
    },
    {
      "epoch": 12.365256124721604,
      "grad_norm": 18.762351989746094,
      "learning_rate": 1e-06,
      "loss": 0.3948,
      "num_input_tokens_seen": 311070112,
      "step": 5552
    },
    {
      "epoch": 12.365256124721604,
      "loss": 0.27796292304992676,
      "loss_ce": 0.00013091039727441967,
      "loss_iou": 0.11865234375,
      "loss_num": 0.008056640625,
      "loss_xval": 0.27734375,
      "num_input_tokens_seen": 311070112,
      "step": 5552
    },
    {
      "epoch": 12.367483296213809,
      "grad_norm": 22.09250259399414,
      "learning_rate": 1e-06,
      "loss": 0.7311,
      "num_input_tokens_seen": 311125564,
      "step": 5553
    },
    {
      "epoch": 12.367483296213809,
      "loss": 0.7942242622375488,
      "loss_ce": 0.00015692379383835942,
      "loss_iou": 0.337890625,
      "loss_num": 0.023193359375,
      "loss_xval": 0.79296875,
      "num_input_tokens_seen": 311125564,
      "step": 5553
    },
    {
      "epoch": 12.369710467706014,
      "grad_norm": 20.51223373413086,
      "learning_rate": 1e-06,
      "loss": 0.5099,
      "num_input_tokens_seen": 311182804,
      "step": 5554
    },
    {
      "epoch": 12.369710467706014,
      "loss": 0.6075503826141357,
      "loss_ce": 0.00012847562902607024,
      "loss_iou": 0.265625,
      "loss_num": 0.0150146484375,
      "loss_xval": 0.609375,
      "num_input_tokens_seen": 311182804,
      "step": 5554
    },
    {
      "epoch": 12.371937639198219,
      "grad_norm": 27.149372100830078,
      "learning_rate": 1e-06,
      "loss": 0.6807,
      "num_input_tokens_seen": 311239764,
      "step": 5555
    },
    {
      "epoch": 12.371937639198219,
      "loss": 0.666401743888855,
      "loss_ce": 0.00014195009134709835,
      "loss_iou": 0.28125,
      "loss_num": 0.020751953125,
      "loss_xval": 0.66796875,
      "num_input_tokens_seen": 311239764,
      "step": 5555
    },
    {
      "epoch": 12.374164810690424,
      "grad_norm": 19.294330596923828,
      "learning_rate": 1e-06,
      "loss": 0.4761,
      "num_input_tokens_seen": 311295712,
      "step": 5556
    },
    {
      "epoch": 12.374164810690424,
      "loss": 0.3280073404312134,
      "loss_ce": 0.00012647973198909312,
      "loss_iou": 0.150390625,
      "loss_num": 0.005340576171875,
      "loss_xval": 0.328125,
      "num_input_tokens_seen": 311295712,
      "step": 5556
    },
    {
      "epoch": 12.376391982182628,
      "grad_norm": 30.496156692504883,
      "learning_rate": 1e-06,
      "loss": 0.5462,
      "num_input_tokens_seen": 311354216,
      "step": 5557
    },
    {
      "epoch": 12.376391982182628,
      "loss": 0.4644961357116699,
      "loss_ce": 0.00014063574781175703,
      "loss_iou": 0.2060546875,
      "loss_num": 0.0106201171875,
      "loss_xval": 0.46484375,
      "num_input_tokens_seen": 311354216,
      "step": 5557
    },
    {
      "epoch": 12.378619153674833,
      "grad_norm": 18.934402465820312,
      "learning_rate": 1e-06,
      "loss": 0.3934,
      "num_input_tokens_seen": 311409028,
      "step": 5558
    },
    {
      "epoch": 12.378619153674833,
      "loss": 0.39881080389022827,
      "loss_ce": 0.0001291544467676431,
      "loss_iou": 0.166015625,
      "loss_num": 0.01324462890625,
      "loss_xval": 0.3984375,
      "num_input_tokens_seen": 311409028,
      "step": 5558
    },
    {
      "epoch": 12.380846325167038,
      "grad_norm": 25.294477462768555,
      "learning_rate": 1e-06,
      "loss": 0.5882,
      "num_input_tokens_seen": 311465116,
      "step": 5559
    },
    {
      "epoch": 12.380846325167038,
      "loss": 0.6579493880271912,
      "loss_ce": 0.00011249056842643768,
      "loss_iou": 0.27734375,
      "loss_num": 0.0208740234375,
      "loss_xval": 0.65625,
      "num_input_tokens_seen": 311465116,
      "step": 5559
    },
    {
      "epoch": 12.383073496659243,
      "grad_norm": 21.457990646362305,
      "learning_rate": 1e-06,
      "loss": 0.4843,
      "num_input_tokens_seen": 311520028,
      "step": 5560
    },
    {
      "epoch": 12.383073496659243,
      "loss": 0.4148017168045044,
      "loss_ce": 0.00012887550110463053,
      "loss_iou": 0.1904296875,
      "loss_num": 0.00665283203125,
      "loss_xval": 0.4140625,
      "num_input_tokens_seen": 311520028,
      "step": 5560
    },
    {
      "epoch": 12.385300668151448,
      "grad_norm": 22.746858596801758,
      "learning_rate": 1e-06,
      "loss": 0.4334,
      "num_input_tokens_seen": 311572928,
      "step": 5561
    },
    {
      "epoch": 12.385300668151448,
      "loss": 0.2498151659965515,
      "loss_ce": 0.00018137965525966138,
      "loss_iou": 0.09619140625,
      "loss_num": 0.01153564453125,
      "loss_xval": 0.25,
      "num_input_tokens_seen": 311572928,
      "step": 5561
    },
    {
      "epoch": 12.387527839643653,
      "grad_norm": 28.644351959228516,
      "learning_rate": 1e-06,
      "loss": 0.586,
      "num_input_tokens_seen": 311630476,
      "step": 5562
    },
    {
      "epoch": 12.387527839643653,
      "loss": 0.6954451203346252,
      "loss_ce": 0.0001326243655057624,
      "loss_iou": 0.310546875,
      "loss_num": 0.01531982421875,
      "loss_xval": 0.6953125,
      "num_input_tokens_seen": 311630476,
      "step": 5562
    },
    {
      "epoch": 12.389755011135858,
      "grad_norm": 38.925472259521484,
      "learning_rate": 1e-06,
      "loss": 0.4525,
      "num_input_tokens_seen": 311687972,
      "step": 5563
    },
    {
      "epoch": 12.389755011135858,
      "loss": 0.43446385860443115,
      "loss_ce": 0.00013769854558631778,
      "loss_iou": 0.201171875,
      "loss_num": 0.00634765625,
      "loss_xval": 0.43359375,
      "num_input_tokens_seen": 311687972,
      "step": 5563
    },
    {
      "epoch": 12.391982182628063,
      "grad_norm": 19.39438247680664,
      "learning_rate": 1e-06,
      "loss": 0.5437,
      "num_input_tokens_seen": 311744768,
      "step": 5564
    },
    {
      "epoch": 12.391982182628063,
      "loss": 0.5501693487167358,
      "loss_ce": 0.00012056773994117975,
      "loss_iou": 0.251953125,
      "loss_num": 0.0096435546875,
      "loss_xval": 0.55078125,
      "num_input_tokens_seen": 311744768,
      "step": 5564
    },
    {
      "epoch": 12.394209354120267,
      "grad_norm": 13.309589385986328,
      "learning_rate": 1e-06,
      "loss": 0.4274,
      "num_input_tokens_seen": 311799736,
      "step": 5565
    },
    {
      "epoch": 12.394209354120267,
      "loss": 0.5023265480995178,
      "loss_ce": 0.0001292880333494395,
      "loss_iou": 0.212890625,
      "loss_num": 0.01513671875,
      "loss_xval": 0.50390625,
      "num_input_tokens_seen": 311799736,
      "step": 5565
    },
    {
      "epoch": 12.396436525612472,
      "grad_norm": 15.245659828186035,
      "learning_rate": 1e-06,
      "loss": 0.4462,
      "num_input_tokens_seen": 311856860,
      "step": 5566
    },
    {
      "epoch": 12.396436525612472,
      "loss": 0.5222017765045166,
      "loss_ce": 0.00010704126907512546,
      "loss_iou": 0.2294921875,
      "loss_num": 0.01263427734375,
      "loss_xval": 0.5234375,
      "num_input_tokens_seen": 311856860,
      "step": 5566
    },
    {
      "epoch": 12.398663697104677,
      "grad_norm": 12.112483024597168,
      "learning_rate": 1e-06,
      "loss": 0.4344,
      "num_input_tokens_seen": 311909940,
      "step": 5567
    },
    {
      "epoch": 12.398663697104677,
      "loss": 0.6277226805686951,
      "loss_ce": 0.00015922555758152157,
      "loss_iou": 0.287109375,
      "loss_num": 0.0111083984375,
      "loss_xval": 0.62890625,
      "num_input_tokens_seen": 311909940,
      "step": 5567
    },
    {
      "epoch": 12.400890868596882,
      "grad_norm": 35.967140197753906,
      "learning_rate": 1e-06,
      "loss": 0.5718,
      "num_input_tokens_seen": 311964724,
      "step": 5568
    },
    {
      "epoch": 12.400890868596882,
      "loss": 0.539790153503418,
      "loss_ce": 0.0001172933480120264,
      "loss_iou": 0.2294921875,
      "loss_num": 0.0159912109375,
      "loss_xval": 0.5390625,
      "num_input_tokens_seen": 311964724,
      "step": 5568
    },
    {
      "epoch": 12.403118040089087,
      "grad_norm": 12.642704963684082,
      "learning_rate": 1e-06,
      "loss": 0.4183,
      "num_input_tokens_seen": 312021624,
      "step": 5569
    },
    {
      "epoch": 12.403118040089087,
      "loss": 0.2902145981788635,
      "loss_ce": 0.00011449479643488303,
      "loss_iou": 0.125,
      "loss_num": 0.00787353515625,
      "loss_xval": 0.291015625,
      "num_input_tokens_seen": 312021624,
      "step": 5569
    },
    {
      "epoch": 12.405345211581292,
      "grad_norm": 24.059127807617188,
      "learning_rate": 1e-06,
      "loss": 0.4588,
      "num_input_tokens_seen": 312075924,
      "step": 5570
    },
    {
      "epoch": 12.405345211581292,
      "loss": 0.31721970438957214,
      "loss_ce": 0.00011153465311508626,
      "loss_iou": 0.1318359375,
      "loss_num": 0.0106201171875,
      "loss_xval": 0.31640625,
      "num_input_tokens_seen": 312075924,
      "step": 5570
    },
    {
      "epoch": 12.407572383073497,
      "grad_norm": 16.74148178100586,
      "learning_rate": 1e-06,
      "loss": 0.386,
      "num_input_tokens_seen": 312132284,
      "step": 5571
    },
    {
      "epoch": 12.407572383073497,
      "loss": 0.43999338150024414,
      "loss_ce": 0.0001740275911288336,
      "loss_iou": 0.1884765625,
      "loss_num": 0.0128173828125,
      "loss_xval": 0.439453125,
      "num_input_tokens_seen": 312132284,
      "step": 5571
    },
    {
      "epoch": 12.409799554565701,
      "grad_norm": 33.77167510986328,
      "learning_rate": 1e-06,
      "loss": 0.5518,
      "num_input_tokens_seen": 312190496,
      "step": 5572
    },
    {
      "epoch": 12.409799554565701,
      "loss": 0.3965058922767639,
      "loss_ce": 0.0001435701851733029,
      "loss_iou": 0.1826171875,
      "loss_num": 0.006378173828125,
      "loss_xval": 0.396484375,
      "num_input_tokens_seen": 312190496,
      "step": 5572
    },
    {
      "epoch": 12.412026726057906,
      "grad_norm": 45.48998260498047,
      "learning_rate": 1e-06,
      "loss": 0.5871,
      "num_input_tokens_seen": 312246352,
      "step": 5573
    },
    {
      "epoch": 12.412026726057906,
      "loss": 0.5653384327888489,
      "loss_ce": 0.00015288355643860996,
      "loss_iou": 0.240234375,
      "loss_num": 0.0169677734375,
      "loss_xval": 0.56640625,
      "num_input_tokens_seen": 312246352,
      "step": 5573
    },
    {
      "epoch": 12.414253897550111,
      "grad_norm": 15.57947063446045,
      "learning_rate": 1e-06,
      "loss": 0.5422,
      "num_input_tokens_seen": 312301816,
      "step": 5574
    },
    {
      "epoch": 12.414253897550111,
      "loss": 0.6265942454338074,
      "loss_ce": 0.00012941220484208316,
      "loss_iou": 0.259765625,
      "loss_num": 0.021240234375,
      "loss_xval": 0.625,
      "num_input_tokens_seen": 312301816,
      "step": 5574
    },
    {
      "epoch": 12.416481069042316,
      "grad_norm": 13.934830665588379,
      "learning_rate": 1e-06,
      "loss": 0.3072,
      "num_input_tokens_seen": 312361472,
      "step": 5575
    },
    {
      "epoch": 12.416481069042316,
      "loss": 0.23996922373771667,
      "loss_ce": 0.00010107570415129885,
      "loss_iou": 0.10595703125,
      "loss_num": 0.005645751953125,
      "loss_xval": 0.240234375,
      "num_input_tokens_seen": 312361472,
      "step": 5575
    },
    {
      "epoch": 12.41870824053452,
      "grad_norm": 22.51824951171875,
      "learning_rate": 1e-06,
      "loss": 0.5321,
      "num_input_tokens_seen": 312416660,
      "step": 5576
    },
    {
      "epoch": 12.41870824053452,
      "loss": 0.3880695104598999,
      "loss_ce": 0.00013007389497943223,
      "loss_iou": 0.1796875,
      "loss_num": 0.0054931640625,
      "loss_xval": 0.388671875,
      "num_input_tokens_seen": 312416660,
      "step": 5576
    },
    {
      "epoch": 12.420935412026726,
      "grad_norm": 12.614936828613281,
      "learning_rate": 1e-06,
      "loss": 0.3598,
      "num_input_tokens_seen": 312472660,
      "step": 5577
    },
    {
      "epoch": 12.420935412026726,
      "loss": 0.42401188611984253,
      "loss_ce": 0.00018374717910774052,
      "loss_iou": 0.1767578125,
      "loss_num": 0.0140380859375,
      "loss_xval": 0.423828125,
      "num_input_tokens_seen": 312472660,
      "step": 5577
    },
    {
      "epoch": 12.42316258351893,
      "grad_norm": 19.709209442138672,
      "learning_rate": 1e-06,
      "loss": 0.515,
      "num_input_tokens_seen": 312528764,
      "step": 5578
    },
    {
      "epoch": 12.42316258351893,
      "loss": 0.517235279083252,
      "loss_ce": 0.00014545858721248806,
      "loss_iou": 0.21875,
      "loss_num": 0.0159912109375,
      "loss_xval": 0.515625,
      "num_input_tokens_seen": 312528764,
      "step": 5578
    },
    {
      "epoch": 12.425389755011135,
      "grad_norm": 15.936797142028809,
      "learning_rate": 1e-06,
      "loss": 0.5646,
      "num_input_tokens_seen": 312586228,
      "step": 5579
    },
    {
      "epoch": 12.425389755011135,
      "loss": 0.44118812680244446,
      "loss_ce": 0.00014807441039010882,
      "loss_iou": 0.20703125,
      "loss_num": 0.005523681640625,
      "loss_xval": 0.44140625,
      "num_input_tokens_seen": 312586228,
      "step": 5579
    },
    {
      "epoch": 12.42761692650334,
      "grad_norm": 16.843250274658203,
      "learning_rate": 1e-06,
      "loss": 0.3889,
      "num_input_tokens_seen": 312640980,
      "step": 5580
    },
    {
      "epoch": 12.42761692650334,
      "loss": 0.3072529435157776,
      "loss_ce": 0.00012405663437675685,
      "loss_iou": 0.1318359375,
      "loss_num": 0.00872802734375,
      "loss_xval": 0.306640625,
      "num_input_tokens_seen": 312640980,
      "step": 5580
    },
    {
      "epoch": 12.429844097995545,
      "grad_norm": 29.384336471557617,
      "learning_rate": 1e-06,
      "loss": 0.5642,
      "num_input_tokens_seen": 312696640,
      "step": 5581
    },
    {
      "epoch": 12.429844097995545,
      "loss": 0.5222534537315369,
      "loss_ce": 0.00015875368262641132,
      "loss_iou": 0.2275390625,
      "loss_num": 0.01324462890625,
      "loss_xval": 0.5234375,
      "num_input_tokens_seen": 312696640,
      "step": 5581
    },
    {
      "epoch": 12.43207126948775,
      "grad_norm": 31.232393264770508,
      "learning_rate": 1e-06,
      "loss": 0.5349,
      "num_input_tokens_seen": 312751632,
      "step": 5582
    },
    {
      "epoch": 12.43207126948775,
      "loss": 0.4201045632362366,
      "loss_ce": 0.00012167952081654221,
      "loss_iou": 0.19140625,
      "loss_num": 0.00750732421875,
      "loss_xval": 0.419921875,
      "num_input_tokens_seen": 312751632,
      "step": 5582
    },
    {
      "epoch": 12.434298440979955,
      "grad_norm": 36.4710578918457,
      "learning_rate": 1e-06,
      "loss": 0.5033,
      "num_input_tokens_seen": 312802876,
      "step": 5583
    },
    {
      "epoch": 12.434298440979955,
      "loss": 0.44115835428237915,
      "loss_ce": 0.00011831161828013137,
      "loss_iou": 0.203125,
      "loss_num": 0.0068359375,
      "loss_xval": 0.44140625,
      "num_input_tokens_seen": 312802876,
      "step": 5583
    },
    {
      "epoch": 12.43652561247216,
      "grad_norm": 17.38640594482422,
      "learning_rate": 1e-06,
      "loss": 0.4538,
      "num_input_tokens_seen": 312861152,
      "step": 5584
    },
    {
      "epoch": 12.43652561247216,
      "loss": 0.40885454416275024,
      "loss_ce": 0.0001631533377803862,
      "loss_iou": 0.1884765625,
      "loss_num": 0.006378173828125,
      "loss_xval": 0.408203125,
      "num_input_tokens_seen": 312861152,
      "step": 5584
    },
    {
      "epoch": 12.438752783964365,
      "grad_norm": 18.26358985900879,
      "learning_rate": 1e-06,
      "loss": 0.4418,
      "num_input_tokens_seen": 312915440,
      "step": 5585
    },
    {
      "epoch": 12.438752783964365,
      "loss": 0.542262613773346,
      "loss_ce": 0.0001483731612097472,
      "loss_iou": 0.23828125,
      "loss_num": 0.0133056640625,
      "loss_xval": 0.54296875,
      "num_input_tokens_seen": 312915440,
      "step": 5585
    },
    {
      "epoch": 12.44097995545657,
      "grad_norm": 19.2590274810791,
      "learning_rate": 1e-06,
      "loss": 0.5392,
      "num_input_tokens_seen": 312969320,
      "step": 5586
    },
    {
      "epoch": 12.44097995545657,
      "loss": 0.5372472405433655,
      "loss_ce": 0.00013788053183816373,
      "loss_iou": 0.2412109375,
      "loss_num": 0.01092529296875,
      "loss_xval": 0.5390625,
      "num_input_tokens_seen": 312969320,
      "step": 5586
    },
    {
      "epoch": 12.443207126948774,
      "grad_norm": 15.812176704406738,
      "learning_rate": 1e-06,
      "loss": 0.454,
      "num_input_tokens_seen": 313026268,
      "step": 5587
    },
    {
      "epoch": 12.443207126948774,
      "loss": 0.504636287689209,
      "loss_ce": 0.00011972515494562685,
      "loss_iou": 0.2001953125,
      "loss_num": 0.0208740234375,
      "loss_xval": 0.50390625,
      "num_input_tokens_seen": 313026268,
      "step": 5587
    },
    {
      "epoch": 12.44543429844098,
      "grad_norm": 15.350566864013672,
      "learning_rate": 1e-06,
      "loss": 0.618,
      "num_input_tokens_seen": 313082880,
      "step": 5588
    },
    {
      "epoch": 12.44543429844098,
      "loss": 0.4841225743293762,
      "loss_ce": 0.00011377451301086694,
      "loss_iou": 0.1923828125,
      "loss_num": 0.0201416015625,
      "loss_xval": 0.484375,
      "num_input_tokens_seen": 313082880,
      "step": 5588
    },
    {
      "epoch": 12.447661469933184,
      "grad_norm": 17.000885009765625,
      "learning_rate": 1e-06,
      "loss": 0.6271,
      "num_input_tokens_seen": 313140436,
      "step": 5589
    },
    {
      "epoch": 12.447661469933184,
      "loss": 0.959378182888031,
      "loss_ce": 0.0001496242475695908,
      "loss_iou": 0.37109375,
      "loss_num": 0.04345703125,
      "loss_xval": 0.9609375,
      "num_input_tokens_seen": 313140436,
      "step": 5589
    },
    {
      "epoch": 12.449888641425389,
      "grad_norm": 24.725290298461914,
      "learning_rate": 1e-06,
      "loss": 0.4681,
      "num_input_tokens_seen": 313195976,
      "step": 5590
    },
    {
      "epoch": 12.449888641425389,
      "loss": 0.47022485733032227,
      "loss_ce": 0.00013207047595642507,
      "loss_iou": 0.2060546875,
      "loss_num": 0.01153564453125,
      "loss_xval": 0.470703125,
      "num_input_tokens_seen": 313195976,
      "step": 5590
    },
    {
      "epoch": 12.452115812917596,
      "grad_norm": 17.44597053527832,
      "learning_rate": 1e-06,
      "loss": 0.376,
      "num_input_tokens_seen": 313253104,
      "step": 5591
    },
    {
      "epoch": 12.452115812917596,
      "loss": 0.3353341221809387,
      "loss_ce": 0.00012904632603749633,
      "loss_iou": 0.1337890625,
      "loss_num": 0.01361083984375,
      "loss_xval": 0.3359375,
      "num_input_tokens_seen": 313253104,
      "step": 5591
    },
    {
      "epoch": 12.4543429844098,
      "grad_norm": 22.889427185058594,
      "learning_rate": 1e-06,
      "loss": 0.5976,
      "num_input_tokens_seen": 313309392,
      "step": 5592
    },
    {
      "epoch": 12.4543429844098,
      "loss": 0.6958190202713013,
      "loss_ce": 0.0001402627385687083,
      "loss_iou": 0.314453125,
      "loss_num": 0.01324462890625,
      "loss_xval": 0.6953125,
      "num_input_tokens_seen": 313309392,
      "step": 5592
    },
    {
      "epoch": 12.456570155902005,
      "grad_norm": 23.9648494720459,
      "learning_rate": 1e-06,
      "loss": 0.3895,
      "num_input_tokens_seen": 313365352,
      "step": 5593
    },
    {
      "epoch": 12.456570155902005,
      "loss": 0.4225989580154419,
      "loss_ce": 0.00011362304212525487,
      "loss_iou": 0.1943359375,
      "loss_num": 0.0069580078125,
      "loss_xval": 0.421875,
      "num_input_tokens_seen": 313365352,
      "step": 5593
    },
    {
      "epoch": 12.45879732739421,
      "grad_norm": 22.63255500793457,
      "learning_rate": 1e-06,
      "loss": 0.5079,
      "num_input_tokens_seen": 313422004,
      "step": 5594
    },
    {
      "epoch": 12.45879732739421,
      "loss": 0.37878933548927307,
      "loss_ce": 0.00012723426334559917,
      "loss_iou": 0.1728515625,
      "loss_num": 0.00665283203125,
      "loss_xval": 0.37890625,
      "num_input_tokens_seen": 313422004,
      "step": 5594
    },
    {
      "epoch": 12.461024498886415,
      "grad_norm": 12.69820785522461,
      "learning_rate": 1e-06,
      "loss": 0.454,
      "num_input_tokens_seen": 313480284,
      "step": 5595
    },
    {
      "epoch": 12.461024498886415,
      "loss": 0.34528595209121704,
      "loss_ce": 0.0001321507734246552,
      "loss_iou": 0.150390625,
      "loss_num": 0.0087890625,
      "loss_xval": 0.345703125,
      "num_input_tokens_seen": 313480284,
      "step": 5595
    },
    {
      "epoch": 12.46325167037862,
      "grad_norm": 155.06365966796875,
      "learning_rate": 1e-06,
      "loss": 0.4323,
      "num_input_tokens_seen": 313537320,
      "step": 5596
    },
    {
      "epoch": 12.46325167037862,
      "loss": 0.5743743181228638,
      "loss_ce": 0.0001555891940370202,
      "loss_iou": 0.23828125,
      "loss_num": 0.019775390625,
      "loss_xval": 0.57421875,
      "num_input_tokens_seen": 313537320,
      "step": 5596
    },
    {
      "epoch": 12.465478841870825,
      "grad_norm": 14.968194007873535,
      "learning_rate": 1e-06,
      "loss": 0.566,
      "num_input_tokens_seen": 313592364,
      "step": 5597
    },
    {
      "epoch": 12.465478841870825,
      "loss": 0.4135543406009674,
      "loss_ce": 0.00010219329124083742,
      "loss_iou": 0.1748046875,
      "loss_num": 0.0128173828125,
      "loss_xval": 0.4140625,
      "num_input_tokens_seen": 313592364,
      "step": 5597
    },
    {
      "epoch": 12.46770601336303,
      "grad_norm": 20.62652587890625,
      "learning_rate": 1e-06,
      "loss": 0.5161,
      "num_input_tokens_seen": 313649912,
      "step": 5598
    },
    {
      "epoch": 12.46770601336303,
      "loss": 0.6557765007019043,
      "loss_ce": 0.00025890767574310303,
      "loss_iou": 0.28125,
      "loss_num": 0.0186767578125,
      "loss_xval": 0.65625,
      "num_input_tokens_seen": 313649912,
      "step": 5598
    },
    {
      "epoch": 12.469933184855234,
      "grad_norm": 30.149272918701172,
      "learning_rate": 1e-06,
      "loss": 0.3283,
      "num_input_tokens_seen": 313707300,
      "step": 5599
    },
    {
      "epoch": 12.469933184855234,
      "loss": 0.28626585006713867,
      "loss_ce": 0.00010252131323795766,
      "loss_iou": 0.1220703125,
      "loss_num": 0.0084228515625,
      "loss_xval": 0.287109375,
      "num_input_tokens_seen": 313707300,
      "step": 5599
    },
    {
      "epoch": 12.47216035634744,
      "grad_norm": 23.769775390625,
      "learning_rate": 1e-06,
      "loss": 0.4423,
      "num_input_tokens_seen": 313761436,
      "step": 5600
    },
    {
      "epoch": 12.47216035634744,
      "loss": 0.4040485620498657,
      "loss_ce": 0.00011789177369792014,
      "loss_iou": 0.1787109375,
      "loss_num": 0.00946044921875,
      "loss_xval": 0.404296875,
      "num_input_tokens_seen": 313761436,
      "step": 5600
    },
    {
      "epoch": 12.474387527839644,
      "grad_norm": 15.069268226623535,
      "learning_rate": 1e-06,
      "loss": 0.3506,
      "num_input_tokens_seen": 313818588,
      "step": 5601
    },
    {
      "epoch": 12.474387527839644,
      "loss": 0.38050034642219543,
      "loss_ce": 0.00012924219481647015,
      "loss_iou": 0.1708984375,
      "loss_num": 0.00762939453125,
      "loss_xval": 0.380859375,
      "num_input_tokens_seen": 313818588,
      "step": 5601
    },
    {
      "epoch": 12.476614699331849,
      "grad_norm": 23.391334533691406,
      "learning_rate": 1e-06,
      "loss": 0.6686,
      "num_input_tokens_seen": 313875568,
      "step": 5602
    },
    {
      "epoch": 12.476614699331849,
      "loss": 1.0041569471359253,
      "loss_ce": 0.002203819341957569,
      "loss_iou": 0.35546875,
      "loss_num": 0.05810546875,
      "loss_xval": 1.0,
      "num_input_tokens_seen": 313875568,
      "step": 5602
    },
    {
      "epoch": 12.478841870824054,
      "grad_norm": 27.693687438964844,
      "learning_rate": 1e-06,
      "loss": 0.4511,
      "num_input_tokens_seen": 313932528,
      "step": 5603
    },
    {
      "epoch": 12.478841870824054,
      "loss": 0.4338412880897522,
      "loss_ce": 0.00012547013466246426,
      "loss_iou": 0.1796875,
      "loss_num": 0.0150146484375,
      "loss_xval": 0.43359375,
      "num_input_tokens_seen": 313932528,
      "step": 5603
    },
    {
      "epoch": 12.481069042316259,
      "grad_norm": 14.572260856628418,
      "learning_rate": 1e-06,
      "loss": 0.5831,
      "num_input_tokens_seen": 313990412,
      "step": 5604
    },
    {
      "epoch": 12.481069042316259,
      "loss": 0.47828370332717896,
      "loss_ce": 0.00013426817895378917,
      "loss_iou": 0.212890625,
      "loss_num": 0.0103759765625,
      "loss_xval": 0.478515625,
      "num_input_tokens_seen": 313990412,
      "step": 5604
    },
    {
      "epoch": 12.483296213808464,
      "grad_norm": 41.894989013671875,
      "learning_rate": 1e-06,
      "loss": 0.3697,
      "num_input_tokens_seen": 314045856,
      "step": 5605
    },
    {
      "epoch": 12.483296213808464,
      "loss": 0.3678470551967621,
      "loss_ce": 0.00011023304250556976,
      "loss_iou": 0.16015625,
      "loss_num": 0.00970458984375,
      "loss_xval": 0.3671875,
      "num_input_tokens_seen": 314045856,
      "step": 5605
    },
    {
      "epoch": 12.485523385300668,
      "grad_norm": 23.39073371887207,
      "learning_rate": 1e-06,
      "loss": 0.377,
      "num_input_tokens_seen": 314100568,
      "step": 5606
    },
    {
      "epoch": 12.485523385300668,
      "loss": 0.3292108476161957,
      "loss_ce": 0.00010928411938948557,
      "loss_iou": 0.1474609375,
      "loss_num": 0.00689697265625,
      "loss_xval": 0.328125,
      "num_input_tokens_seen": 314100568,
      "step": 5606
    },
    {
      "epoch": 12.487750556792873,
      "grad_norm": 26.50432014465332,
      "learning_rate": 1e-06,
      "loss": 0.7268,
      "num_input_tokens_seen": 314154296,
      "step": 5607
    },
    {
      "epoch": 12.487750556792873,
      "loss": 0.7526974081993103,
      "loss_ce": 0.00013394109555520117,
      "loss_iou": 0.318359375,
      "loss_num": 0.0233154296875,
      "loss_xval": 0.75390625,
      "num_input_tokens_seen": 314154296,
      "step": 5607
    },
    {
      "epoch": 12.489977728285078,
      "grad_norm": 16.59043312072754,
      "learning_rate": 1e-06,
      "loss": 0.4595,
      "num_input_tokens_seen": 314211988,
      "step": 5608
    },
    {
      "epoch": 12.489977728285078,
      "loss": 0.5807017683982849,
      "loss_ce": 0.00013538115308620036,
      "loss_iou": 0.24609375,
      "loss_num": 0.017578125,
      "loss_xval": 0.58203125,
      "num_input_tokens_seen": 314211988,
      "step": 5608
    },
    {
      "epoch": 12.492204899777283,
      "grad_norm": 15.27440071105957,
      "learning_rate": 1e-06,
      "loss": 0.4032,
      "num_input_tokens_seen": 314270680,
      "step": 5609
    },
    {
      "epoch": 12.492204899777283,
      "loss": 0.349552184343338,
      "loss_ce": 0.00012591719860211015,
      "loss_iou": 0.1533203125,
      "loss_num": 0.008544921875,
      "loss_xval": 0.349609375,
      "num_input_tokens_seen": 314270680,
      "step": 5609
    },
    {
      "epoch": 12.494432071269488,
      "grad_norm": 27.822227478027344,
      "learning_rate": 1e-06,
      "loss": 0.3882,
      "num_input_tokens_seen": 314326436,
      "step": 5610
    },
    {
      "epoch": 12.494432071269488,
      "loss": 0.28033125400543213,
      "loss_ce": 0.00011883871047757566,
      "loss_iou": 0.12890625,
      "loss_num": 0.004730224609375,
      "loss_xval": 0.279296875,
      "num_input_tokens_seen": 314326436,
      "step": 5610
    },
    {
      "epoch": 12.496659242761693,
      "grad_norm": 18.214462280273438,
      "learning_rate": 1e-06,
      "loss": 0.4729,
      "num_input_tokens_seen": 314383216,
      "step": 5611
    },
    {
      "epoch": 12.496659242761693,
      "loss": 0.5203236937522888,
      "loss_ce": 0.00012107704242225736,
      "loss_iou": 0.2412109375,
      "loss_num": 0.00762939453125,
      "loss_xval": 0.51953125,
      "num_input_tokens_seen": 314383216,
      "step": 5611
    },
    {
      "epoch": 12.498886414253898,
      "grad_norm": 17.27739715576172,
      "learning_rate": 1e-06,
      "loss": 0.412,
      "num_input_tokens_seen": 314438016,
      "step": 5612
    },
    {
      "epoch": 12.498886414253898,
      "loss": 0.4331158995628357,
      "loss_ce": 0.0001325118209933862,
      "loss_iou": 0.1982421875,
      "loss_num": 0.007293701171875,
      "loss_xval": 0.43359375,
      "num_input_tokens_seen": 314438016,
      "step": 5612
    },
    {
      "epoch": 12.501113585746102,
      "grad_norm": 25.569000244140625,
      "learning_rate": 1e-06,
      "loss": 0.5059,
      "num_input_tokens_seen": 314494216,
      "step": 5613
    },
    {
      "epoch": 12.501113585746102,
      "loss": 0.4582577347755432,
      "loss_ce": 0.00012786393926944584,
      "loss_iou": 0.1943359375,
      "loss_num": 0.0137939453125,
      "loss_xval": 0.458984375,
      "num_input_tokens_seen": 314494216,
      "step": 5613
    },
    {
      "epoch": 12.503340757238307,
      "grad_norm": 16.633384704589844,
      "learning_rate": 1e-06,
      "loss": 0.529,
      "num_input_tokens_seen": 314550764,
      "step": 5614
    },
    {
      "epoch": 12.503340757238307,
      "loss": 0.5543428659439087,
      "loss_ce": 0.00014360195200424641,
      "loss_iou": 0.2158203125,
      "loss_num": 0.024658203125,
      "loss_xval": 0.5546875,
      "num_input_tokens_seen": 314550764,
      "step": 5614
    },
    {
      "epoch": 12.505567928730512,
      "grad_norm": 39.219783782958984,
      "learning_rate": 1e-06,
      "loss": 0.6081,
      "num_input_tokens_seen": 314608284,
      "step": 5615
    },
    {
      "epoch": 12.505567928730512,
      "loss": 0.4896491467952728,
      "loss_ce": 0.00014716846635565162,
      "loss_iou": 0.212890625,
      "loss_num": 0.0125732421875,
      "loss_xval": 0.490234375,
      "num_input_tokens_seen": 314608284,
      "step": 5615
    },
    {
      "epoch": 12.507795100222717,
      "grad_norm": 30.928625106811523,
      "learning_rate": 1e-06,
      "loss": 0.4487,
      "num_input_tokens_seen": 314660020,
      "step": 5616
    },
    {
      "epoch": 12.507795100222717,
      "loss": 0.4297997057437897,
      "loss_ce": 0.00011219394946238026,
      "loss_iou": 0.1943359375,
      "loss_num": 0.00836181640625,
      "loss_xval": 0.4296875,
      "num_input_tokens_seen": 314660020,
      "step": 5616
    },
    {
      "epoch": 12.510022271714922,
      "grad_norm": 20.47149085998535,
      "learning_rate": 1e-06,
      "loss": 0.4247,
      "num_input_tokens_seen": 314715408,
      "step": 5617
    },
    {
      "epoch": 12.510022271714922,
      "loss": 0.3184962868690491,
      "loss_ce": 0.0001369007513858378,
      "loss_iou": 0.130859375,
      "loss_num": 0.01116943359375,
      "loss_xval": 0.318359375,
      "num_input_tokens_seen": 314715408,
      "step": 5617
    },
    {
      "epoch": 12.512249443207127,
      "grad_norm": 23.522342681884766,
      "learning_rate": 1e-06,
      "loss": 0.4706,
      "num_input_tokens_seen": 314768072,
      "step": 5618
    },
    {
      "epoch": 12.512249443207127,
      "loss": 0.62452632188797,
      "loss_ce": 0.00013670045882463455,
      "loss_iou": 0.271484375,
      "loss_num": 0.01611328125,
      "loss_xval": 0.625,
      "num_input_tokens_seen": 314768072,
      "step": 5618
    },
    {
      "epoch": 12.514476614699332,
      "grad_norm": 24.376148223876953,
      "learning_rate": 1e-06,
      "loss": 0.3809,
      "num_input_tokens_seen": 314824920,
      "step": 5619
    },
    {
      "epoch": 12.514476614699332,
      "loss": 0.43566471338272095,
      "loss_ce": 0.00011783596710301936,
      "loss_iou": 0.17578125,
      "loss_num": 0.016845703125,
      "loss_xval": 0.435546875,
      "num_input_tokens_seen": 314824920,
      "step": 5619
    },
    {
      "epoch": 12.516703786191536,
      "grad_norm": 21.62662124633789,
      "learning_rate": 1e-06,
      "loss": 0.4331,
      "num_input_tokens_seen": 314881352,
      "step": 5620
    },
    {
      "epoch": 12.516703786191536,
      "loss": 0.42295846343040466,
      "loss_ce": 0.00013743109593633562,
      "loss_iou": 0.185546875,
      "loss_num": 0.01031494140625,
      "loss_xval": 0.421875,
      "num_input_tokens_seen": 314881352,
      "step": 5620
    },
    {
      "epoch": 12.518930957683741,
      "grad_norm": 17.14795684814453,
      "learning_rate": 1e-06,
      "loss": 0.4885,
      "num_input_tokens_seen": 314938836,
      "step": 5621
    },
    {
      "epoch": 12.518930957683741,
      "loss": 0.5315926671028137,
      "loss_ce": 0.00022057080059312284,
      "loss_iou": 0.2451171875,
      "loss_num": 0.00830078125,
      "loss_xval": 0.53125,
      "num_input_tokens_seen": 314938836,
      "step": 5621
    },
    {
      "epoch": 12.521158129175946,
      "grad_norm": 32.41496658325195,
      "learning_rate": 1e-06,
      "loss": 0.5334,
      "num_input_tokens_seen": 314997228,
      "step": 5622
    },
    {
      "epoch": 12.521158129175946,
      "loss": 0.5393984317779541,
      "loss_ce": 0.00015282572712749243,
      "loss_iou": 0.22265625,
      "loss_num": 0.018798828125,
      "loss_xval": 0.5390625,
      "num_input_tokens_seen": 314997228,
      "step": 5622
    },
    {
      "epoch": 12.523385300668151,
      "grad_norm": 21.443031311035156,
      "learning_rate": 1e-06,
      "loss": 0.5174,
      "num_input_tokens_seen": 315052164,
      "step": 5623
    },
    {
      "epoch": 12.523385300668151,
      "loss": 0.49842369556427,
      "loss_ce": 0.00013268653128761798,
      "loss_iou": 0.2138671875,
      "loss_num": 0.0140380859375,
      "loss_xval": 0.498046875,
      "num_input_tokens_seen": 315052164,
      "step": 5623
    },
    {
      "epoch": 12.525612472160356,
      "grad_norm": 21.690479278564453,
      "learning_rate": 1e-06,
      "loss": 0.5104,
      "num_input_tokens_seen": 315108840,
      "step": 5624
    },
    {
      "epoch": 12.525612472160356,
      "loss": 0.3883495032787323,
      "loss_ce": 0.00016590597806498408,
      "loss_iou": 0.1748046875,
      "loss_num": 0.0079345703125,
      "loss_xval": 0.388671875,
      "num_input_tokens_seen": 315108840,
      "step": 5624
    },
    {
      "epoch": 12.52783964365256,
      "grad_norm": 32.13372039794922,
      "learning_rate": 1e-06,
      "loss": 0.4522,
      "num_input_tokens_seen": 315165896,
      "step": 5625
    },
    {
      "epoch": 12.52783964365256,
      "loss": 0.5064772367477417,
      "loss_ce": 0.00012959350715391338,
      "loss_iou": 0.2333984375,
      "loss_num": 0.008056640625,
      "loss_xval": 0.5078125,
      "num_input_tokens_seen": 315165896,
      "step": 5625
    },
    {
      "epoch": 12.530066815144766,
      "grad_norm": 80.0682144165039,
      "learning_rate": 1e-06,
      "loss": 0.5871,
      "num_input_tokens_seen": 315219956,
      "step": 5626
    },
    {
      "epoch": 12.530066815144766,
      "loss": 0.7462646961212158,
      "loss_ce": 0.0001709566276986152,
      "loss_iou": 0.32421875,
      "loss_num": 0.01953125,
      "loss_xval": 0.74609375,
      "num_input_tokens_seen": 315219956,
      "step": 5626
    },
    {
      "epoch": 12.53229398663697,
      "grad_norm": 20.835941314697266,
      "learning_rate": 1e-06,
      "loss": 0.4619,
      "num_input_tokens_seen": 315277464,
      "step": 5627
    },
    {
      "epoch": 12.53229398663697,
      "loss": 0.362972229719162,
      "loss_ce": 0.00017925890279002488,
      "loss_iou": 0.1494140625,
      "loss_num": 0.0128173828125,
      "loss_xval": 0.36328125,
      "num_input_tokens_seen": 315277464,
      "step": 5627
    },
    {
      "epoch": 12.534521158129175,
      "grad_norm": 15.360301971435547,
      "learning_rate": 1e-06,
      "loss": 0.3773,
      "num_input_tokens_seen": 315333928,
      "step": 5628
    },
    {
      "epoch": 12.534521158129175,
      "loss": 0.31863662600517273,
      "loss_ce": 0.00015518809959758073,
      "loss_iou": 0.1396484375,
      "loss_num": 0.007781982421875,
      "loss_xval": 0.318359375,
      "num_input_tokens_seen": 315333928,
      "step": 5628
    },
    {
      "epoch": 12.53674832962138,
      "grad_norm": 18.045223236083984,
      "learning_rate": 1e-06,
      "loss": 0.444,
      "num_input_tokens_seen": 315390648,
      "step": 5629
    },
    {
      "epoch": 12.53674832962138,
      "loss": 0.30956727266311646,
      "loss_ce": 0.00011902242840733379,
      "loss_iou": 0.1298828125,
      "loss_num": 0.0098876953125,
      "loss_xval": 0.30859375,
      "num_input_tokens_seen": 315390648,
      "step": 5629
    },
    {
      "epoch": 12.538975501113585,
      "grad_norm": 36.879791259765625,
      "learning_rate": 1e-06,
      "loss": 0.4498,
      "num_input_tokens_seen": 315447484,
      "step": 5630
    },
    {
      "epoch": 12.538975501113585,
      "loss": 0.33547037839889526,
      "loss_ce": 0.00011271548282820731,
      "loss_iou": 0.1337890625,
      "loss_num": 0.0135498046875,
      "loss_xval": 0.3359375,
      "num_input_tokens_seen": 315447484,
      "step": 5630
    },
    {
      "epoch": 12.54120267260579,
      "grad_norm": 18.903820037841797,
      "learning_rate": 1e-06,
      "loss": 0.4593,
      "num_input_tokens_seen": 315503840,
      "step": 5631
    },
    {
      "epoch": 12.54120267260579,
      "loss": 0.3675609529018402,
      "loss_ce": 0.0001293124514631927,
      "loss_iou": 0.1650390625,
      "loss_num": 0.007659912109375,
      "loss_xval": 0.3671875,
      "num_input_tokens_seen": 315503840,
      "step": 5631
    },
    {
      "epoch": 12.543429844097995,
      "grad_norm": 15.022705078125,
      "learning_rate": 1e-06,
      "loss": 0.5607,
      "num_input_tokens_seen": 315560960,
      "step": 5632
    },
    {
      "epoch": 12.543429844097995,
      "loss": 0.5480488538742065,
      "loss_ce": 0.00013630017929244787,
      "loss_iou": 0.2353515625,
      "loss_num": 0.015380859375,
      "loss_xval": 0.546875,
      "num_input_tokens_seen": 315560960,
      "step": 5632
    },
    {
      "epoch": 12.5456570155902,
      "grad_norm": 18.663776397705078,
      "learning_rate": 1e-06,
      "loss": 0.5616,
      "num_input_tokens_seen": 315616752,
      "step": 5633
    },
    {
      "epoch": 12.5456570155902,
      "loss": 0.5263513922691345,
      "loss_ce": 0.00010624493006616831,
      "loss_iou": 0.23046875,
      "loss_num": 0.01324462890625,
      "loss_xval": 0.52734375,
      "num_input_tokens_seen": 315616752,
      "step": 5633
    },
    {
      "epoch": 12.547884187082406,
      "grad_norm": 19.984249114990234,
      "learning_rate": 1e-06,
      "loss": 0.5842,
      "num_input_tokens_seen": 315670220,
      "step": 5634
    },
    {
      "epoch": 12.547884187082406,
      "loss": 0.47094982862472534,
      "loss_ce": 0.0001246376777999103,
      "loss_iou": 0.2177734375,
      "loss_num": 0.007110595703125,
      "loss_xval": 0.470703125,
      "num_input_tokens_seen": 315670220,
      "step": 5634
    },
    {
      "epoch": 12.550111358574611,
      "grad_norm": 22.835325241088867,
      "learning_rate": 1e-06,
      "loss": 0.5918,
      "num_input_tokens_seen": 315723144,
      "step": 5635
    },
    {
      "epoch": 12.550111358574611,
      "loss": 0.5138300657272339,
      "loss_ce": 0.00015820455155335367,
      "loss_iou": 0.2021484375,
      "loss_num": 0.02197265625,
      "loss_xval": 0.515625,
      "num_input_tokens_seen": 315723144,
      "step": 5635
    },
    {
      "epoch": 12.552338530066816,
      "grad_norm": 13.647897720336914,
      "learning_rate": 1e-06,
      "loss": 0.5345,
      "num_input_tokens_seen": 315778516,
      "step": 5636
    },
    {
      "epoch": 12.552338530066816,
      "loss": 0.5501181483268738,
      "loss_ce": 0.00013035524170845747,
      "loss_iou": 0.2373046875,
      "loss_num": 0.01507568359375,
      "loss_xval": 0.55078125,
      "num_input_tokens_seen": 315778516,
      "step": 5636
    },
    {
      "epoch": 12.55456570155902,
      "grad_norm": 17.706628799438477,
      "learning_rate": 1e-06,
      "loss": 0.8976,
      "num_input_tokens_seen": 315835116,
      "step": 5637
    },
    {
      "epoch": 12.55456570155902,
      "loss": 0.998259425163269,
      "loss_ce": 0.00015154268476180732,
      "loss_iou": 0.419921875,
      "loss_num": 0.031494140625,
      "loss_xval": 1.0,
      "num_input_tokens_seen": 315835116,
      "step": 5637
    },
    {
      "epoch": 12.556792873051226,
      "grad_norm": 16.907272338867188,
      "learning_rate": 1e-06,
      "loss": 0.5527,
      "num_input_tokens_seen": 315890436,
      "step": 5638
    },
    {
      "epoch": 12.556792873051226,
      "loss": 0.652888298034668,
      "loss_ce": 0.00011732908751582727,
      "loss_iou": 0.28515625,
      "loss_num": 0.0159912109375,
      "loss_xval": 0.65234375,
      "num_input_tokens_seen": 315890436,
      "step": 5638
    },
    {
      "epoch": 12.55902004454343,
      "grad_norm": 14.611957550048828,
      "learning_rate": 1e-06,
      "loss": 0.7071,
      "num_input_tokens_seen": 315944500,
      "step": 5639
    },
    {
      "epoch": 12.55902004454343,
      "loss": 0.6298529505729675,
      "loss_ce": 0.0015570501564070582,
      "loss_iou": 0.236328125,
      "loss_num": 0.03125,
      "loss_xval": 0.62890625,
      "num_input_tokens_seen": 315944500,
      "step": 5639
    },
    {
      "epoch": 12.561247216035635,
      "grad_norm": 39.95273208618164,
      "learning_rate": 1e-06,
      "loss": 0.7408,
      "num_input_tokens_seen": 315998188,
      "step": 5640
    },
    {
      "epoch": 12.561247216035635,
      "loss": 0.7573456168174744,
      "loss_ce": 0.0001434848236385733,
      "loss_iou": 0.298828125,
      "loss_num": 0.03173828125,
      "loss_xval": 0.7578125,
      "num_input_tokens_seen": 315998188,
      "step": 5640
    },
    {
      "epoch": 12.56347438752784,
      "grad_norm": 17.88589859008789,
      "learning_rate": 1e-06,
      "loss": 0.3115,
      "num_input_tokens_seen": 316054048,
      "step": 5641
    },
    {
      "epoch": 12.56347438752784,
      "loss": 0.3503269553184509,
      "loss_ce": 0.00010725160245783627,
      "loss_iou": 0.1611328125,
      "loss_num": 0.005523681640625,
      "loss_xval": 0.349609375,
      "num_input_tokens_seen": 316054048,
      "step": 5641
    },
    {
      "epoch": 12.565701559020045,
      "grad_norm": 76.38765716552734,
      "learning_rate": 1e-06,
      "loss": 0.7108,
      "num_input_tokens_seen": 316110144,
      "step": 5642
    },
    {
      "epoch": 12.565701559020045,
      "loss": 0.7477288842201233,
      "loss_ce": 0.00017026919522322714,
      "loss_iou": 0.33203125,
      "loss_num": 0.0169677734375,
      "loss_xval": 0.74609375,
      "num_input_tokens_seen": 316110144,
      "step": 5642
    },
    {
      "epoch": 12.56792873051225,
      "grad_norm": 13.18075180053711,
      "learning_rate": 1e-06,
      "loss": 0.6063,
      "num_input_tokens_seen": 316167200,
      "step": 5643
    },
    {
      "epoch": 12.56792873051225,
      "loss": 0.6305142641067505,
      "loss_ce": 0.00014317099703475833,
      "loss_iou": 0.2431640625,
      "loss_num": 0.0289306640625,
      "loss_xval": 0.62890625,
      "num_input_tokens_seen": 316167200,
      "step": 5643
    },
    {
      "epoch": 12.570155902004455,
      "grad_norm": 14.634928703308105,
      "learning_rate": 1e-06,
      "loss": 0.4603,
      "num_input_tokens_seen": 316224676,
      "step": 5644
    },
    {
      "epoch": 12.570155902004455,
      "loss": 0.34102046489715576,
      "loss_ce": 0.00020015044719912112,
      "loss_iou": 0.1552734375,
      "loss_num": 0.006256103515625,
      "loss_xval": 0.33984375,
      "num_input_tokens_seen": 316224676,
      "step": 5644
    },
    {
      "epoch": 12.57238307349666,
      "grad_norm": 21.283550262451172,
      "learning_rate": 1e-06,
      "loss": 0.5681,
      "num_input_tokens_seen": 316282056,
      "step": 5645
    },
    {
      "epoch": 12.57238307349666,
      "loss": 0.6777347326278687,
      "loss_ce": 0.0001223925792146474,
      "loss_iou": 0.296875,
      "loss_num": 0.0169677734375,
      "loss_xval": 0.67578125,
      "num_input_tokens_seen": 316282056,
      "step": 5645
    },
    {
      "epoch": 12.574610244988865,
      "grad_norm": 21.21981430053711,
      "learning_rate": 1e-06,
      "loss": 0.604,
      "num_input_tokens_seen": 316336632,
      "step": 5646
    },
    {
      "epoch": 12.574610244988865,
      "loss": 0.4394035339355469,
      "loss_ce": 0.00028606440173462033,
      "loss_iou": 0.1875,
      "loss_num": 0.0128173828125,
      "loss_xval": 0.439453125,
      "num_input_tokens_seen": 316336632,
      "step": 5646
    },
    {
      "epoch": 12.57683741648107,
      "grad_norm": 16.894546508789062,
      "learning_rate": 1e-06,
      "loss": 0.3813,
      "num_input_tokens_seen": 316392980,
      "step": 5647
    },
    {
      "epoch": 12.57683741648107,
      "loss": 0.40146303176879883,
      "loss_ce": 9.585732186678797e-05,
      "loss_iou": 0.1826171875,
      "loss_num": 0.00738525390625,
      "loss_xval": 0.40234375,
      "num_input_tokens_seen": 316392980,
      "step": 5647
    },
    {
      "epoch": 12.579064587973274,
      "grad_norm": 14.076600074768066,
      "learning_rate": 1e-06,
      "loss": 0.4109,
      "num_input_tokens_seen": 316450712,
      "step": 5648
    },
    {
      "epoch": 12.579064587973274,
      "loss": 0.4312838315963745,
      "loss_ce": 0.00013148068683221936,
      "loss_iou": 0.1875,
      "loss_num": 0.0111083984375,
      "loss_xval": 0.431640625,
      "num_input_tokens_seen": 316450712,
      "step": 5648
    },
    {
      "epoch": 12.58129175946548,
      "grad_norm": 18.681835174560547,
      "learning_rate": 1e-06,
      "loss": 0.5381,
      "num_input_tokens_seen": 316506272,
      "step": 5649
    },
    {
      "epoch": 12.58129175946548,
      "loss": 0.5017099380493164,
      "loss_ce": 0.00012300520029384643,
      "loss_iou": 0.212890625,
      "loss_num": 0.01519775390625,
      "loss_xval": 0.5,
      "num_input_tokens_seen": 316506272,
      "step": 5649
    },
    {
      "epoch": 12.583518930957684,
      "grad_norm": 19.394519805908203,
      "learning_rate": 1e-06,
      "loss": 0.5845,
      "num_input_tokens_seen": 316560448,
      "step": 5650
    },
    {
      "epoch": 12.583518930957684,
      "loss": 0.49112582206726074,
      "loss_ce": 0.00015903441817499697,
      "loss_iou": 0.197265625,
      "loss_num": 0.0194091796875,
      "loss_xval": 0.490234375,
      "num_input_tokens_seen": 316560448,
      "step": 5650
    },
    {
      "epoch": 12.585746102449889,
      "grad_norm": 19.638614654541016,
      "learning_rate": 1e-06,
      "loss": 0.3961,
      "num_input_tokens_seen": 316616300,
      "step": 5651
    },
    {
      "epoch": 12.585746102449889,
      "loss": 0.3637651801109314,
      "loss_ce": 0.00011775536404456943,
      "loss_iou": 0.1650390625,
      "loss_num": 0.00689697265625,
      "loss_xval": 0.36328125,
      "num_input_tokens_seen": 316616300,
      "step": 5651
    },
    {
      "epoch": 12.587973273942094,
      "grad_norm": 20.799516677856445,
      "learning_rate": 1e-06,
      "loss": 0.7011,
      "num_input_tokens_seen": 316672848,
      "step": 5652
    },
    {
      "epoch": 12.587973273942094,
      "loss": 0.555069088935852,
      "loss_ce": 0.00013745573232881725,
      "loss_iou": 0.255859375,
      "loss_num": 0.0084228515625,
      "loss_xval": 0.5546875,
      "num_input_tokens_seen": 316672848,
      "step": 5652
    },
    {
      "epoch": 12.590200445434299,
      "grad_norm": 16.250247955322266,
      "learning_rate": 1e-06,
      "loss": 0.5712,
      "num_input_tokens_seen": 316726660,
      "step": 5653
    },
    {
      "epoch": 12.590200445434299,
      "loss": 0.7183680534362793,
      "loss_ce": 0.00018261410878039896,
      "loss_iou": 0.271484375,
      "loss_num": 0.035400390625,
      "loss_xval": 0.71875,
      "num_input_tokens_seen": 316726660,
      "step": 5653
    },
    {
      "epoch": 12.592427616926503,
      "grad_norm": 16.84650993347168,
      "learning_rate": 1e-06,
      "loss": 0.4514,
      "num_input_tokens_seen": 316785652,
      "step": 5654
    },
    {
      "epoch": 12.592427616926503,
      "loss": 0.4916442036628723,
      "loss_ce": 0.00012808601604774594,
      "loss_iou": 0.220703125,
      "loss_num": 0.01025390625,
      "loss_xval": 0.4921875,
      "num_input_tokens_seen": 316785652,
      "step": 5654
    },
    {
      "epoch": 12.594654788418708,
      "grad_norm": 19.085908889770508,
      "learning_rate": 1e-06,
      "loss": 0.549,
      "num_input_tokens_seen": 316838728,
      "step": 5655
    },
    {
      "epoch": 12.594654788418708,
      "loss": 0.45017945766448975,
      "loss_ce": 0.0002282563509652391,
      "loss_iou": 0.201171875,
      "loss_num": 0.00970458984375,
      "loss_xval": 0.44921875,
      "num_input_tokens_seen": 316838728,
      "step": 5655
    },
    {
      "epoch": 12.596881959910913,
      "grad_norm": 18.604005813598633,
      "learning_rate": 1e-06,
      "loss": 0.4854,
      "num_input_tokens_seen": 316895116,
      "step": 5656
    },
    {
      "epoch": 12.596881959910913,
      "loss": 0.4594961106777191,
      "loss_ce": 0.00014551397180184722,
      "loss_iou": 0.201171875,
      "loss_num": 0.01116943359375,
      "loss_xval": 0.458984375,
      "num_input_tokens_seen": 316895116,
      "step": 5656
    },
    {
      "epoch": 12.599109131403118,
      "grad_norm": 19.779462814331055,
      "learning_rate": 1e-06,
      "loss": 0.442,
      "num_input_tokens_seen": 316951228,
      "step": 5657
    },
    {
      "epoch": 12.599109131403118,
      "loss": 0.43638354539871216,
      "loss_ce": 0.00010424081847304478,
      "loss_iou": 0.1884765625,
      "loss_num": 0.01190185546875,
      "loss_xval": 0.435546875,
      "num_input_tokens_seen": 316951228,
      "step": 5657
    },
    {
      "epoch": 12.601336302895323,
      "grad_norm": 17.714229583740234,
      "learning_rate": 1e-06,
      "loss": 0.6931,
      "num_input_tokens_seen": 317007992,
      "step": 5658
    },
    {
      "epoch": 12.601336302895323,
      "loss": 0.7979649305343628,
      "loss_ce": 0.00011335601448081434,
      "loss_iou": 0.337890625,
      "loss_num": 0.0242919921875,
      "loss_xval": 0.796875,
      "num_input_tokens_seen": 317007992,
      "step": 5658
    },
    {
      "epoch": 12.603563474387528,
      "grad_norm": 18.232986450195312,
      "learning_rate": 1e-06,
      "loss": 0.3883,
      "num_input_tokens_seen": 317062396,
      "step": 5659
    },
    {
      "epoch": 12.603563474387528,
      "loss": 0.3348948359489441,
      "loss_ce": 0.00011698143498506397,
      "loss_iou": 0.140625,
      "loss_num": 0.01055908203125,
      "loss_xval": 0.333984375,
      "num_input_tokens_seen": 317062396,
      "step": 5659
    },
    {
      "epoch": 12.605790645879733,
      "grad_norm": 22.718849182128906,
      "learning_rate": 1e-06,
      "loss": 0.519,
      "num_input_tokens_seen": 317118224,
      "step": 5660
    },
    {
      "epoch": 12.605790645879733,
      "loss": 0.58899986743927,
      "loss_ce": 0.00013271119678393006,
      "loss_iou": 0.263671875,
      "loss_num": 0.01239013671875,
      "loss_xval": 0.58984375,
      "num_input_tokens_seen": 317118224,
      "step": 5660
    },
    {
      "epoch": 12.608017817371937,
      "grad_norm": 29.809764862060547,
      "learning_rate": 1e-06,
      "loss": 0.5945,
      "num_input_tokens_seen": 317172604,
      "step": 5661
    },
    {
      "epoch": 12.608017817371937,
      "loss": 0.6281754374504089,
      "loss_ce": 0.0001236619718838483,
      "loss_iou": 0.271484375,
      "loss_num": 0.0169677734375,
      "loss_xval": 0.62890625,
      "num_input_tokens_seen": 317172604,
      "step": 5661
    },
    {
      "epoch": 12.610244988864142,
      "grad_norm": 19.71194076538086,
      "learning_rate": 1e-06,
      "loss": 0.4504,
      "num_input_tokens_seen": 317229076,
      "step": 5662
    },
    {
      "epoch": 12.610244988864142,
      "loss": 0.3518773019313812,
      "loss_ce": 0.00013169541489332914,
      "loss_iou": 0.134765625,
      "loss_num": 0.0164794921875,
      "loss_xval": 0.3515625,
      "num_input_tokens_seen": 317229076,
      "step": 5662
    },
    {
      "epoch": 12.612472160356347,
      "grad_norm": 24.877761840820312,
      "learning_rate": 1e-06,
      "loss": 0.5304,
      "num_input_tokens_seen": 317284080,
      "step": 5663
    },
    {
      "epoch": 12.612472160356347,
      "loss": 0.5929080247879028,
      "loss_ce": 0.0001345914788544178,
      "loss_iou": 0.25390625,
      "loss_num": 0.0172119140625,
      "loss_xval": 0.59375,
      "num_input_tokens_seen": 317284080,
      "step": 5663
    },
    {
      "epoch": 12.614699331848552,
      "grad_norm": 37.88312530517578,
      "learning_rate": 1e-06,
      "loss": 0.7225,
      "num_input_tokens_seen": 317338048,
      "step": 5664
    },
    {
      "epoch": 12.614699331848552,
      "loss": 0.859398365020752,
      "loss_ce": 0.0002674940915312618,
      "loss_iou": 0.359375,
      "loss_num": 0.028564453125,
      "loss_xval": 0.859375,
      "num_input_tokens_seen": 317338048,
      "step": 5664
    },
    {
      "epoch": 12.616926503340757,
      "grad_norm": 25.082626342773438,
      "learning_rate": 1e-06,
      "loss": 0.6775,
      "num_input_tokens_seen": 317396176,
      "step": 5665
    },
    {
      "epoch": 12.616926503340757,
      "loss": 0.6872255802154541,
      "loss_ce": 0.00033595875720493495,
      "loss_iou": 0.283203125,
      "loss_num": 0.024169921875,
      "loss_xval": 0.6875,
      "num_input_tokens_seen": 317396176,
      "step": 5665
    },
    {
      "epoch": 12.619153674832962,
      "grad_norm": 16.618837356567383,
      "learning_rate": 1e-06,
      "loss": 0.6738,
      "num_input_tokens_seen": 317453000,
      "step": 5666
    },
    {
      "epoch": 12.619153674832962,
      "loss": 0.7559080123901367,
      "loss_ce": 0.0001707018236629665,
      "loss_iou": 0.326171875,
      "loss_num": 0.020751953125,
      "loss_xval": 0.75390625,
      "num_input_tokens_seen": 317453000,
      "step": 5666
    },
    {
      "epoch": 12.621380846325167,
      "grad_norm": 13.969486236572266,
      "learning_rate": 1e-06,
      "loss": 0.3752,
      "num_input_tokens_seen": 317510472,
      "step": 5667
    },
    {
      "epoch": 12.621380846325167,
      "loss": 0.42377543449401855,
      "loss_ce": 0.0001303813187405467,
      "loss_iou": 0.193359375,
      "loss_num": 0.00732421875,
      "loss_xval": 0.423828125,
      "num_input_tokens_seen": 317510472,
      "step": 5667
    },
    {
      "epoch": 12.623608017817372,
      "grad_norm": 13.138300895690918,
      "learning_rate": 1e-06,
      "loss": 0.4098,
      "num_input_tokens_seen": 317567232,
      "step": 5668
    },
    {
      "epoch": 12.623608017817372,
      "loss": 0.3994289040565491,
      "loss_ce": 0.00013694007066078484,
      "loss_iou": 0.1904296875,
      "loss_num": 0.0035552978515625,
      "loss_xval": 0.3984375,
      "num_input_tokens_seen": 317567232,
      "step": 5668
    },
    {
      "epoch": 12.625835189309576,
      "grad_norm": 19.55143165588379,
      "learning_rate": 1e-06,
      "loss": 0.378,
      "num_input_tokens_seen": 317624192,
      "step": 5669
    },
    {
      "epoch": 12.625835189309576,
      "loss": 0.4045639634132385,
      "loss_ce": 0.00014502542035188526,
      "loss_iou": 0.1796875,
      "loss_num": 0.0089111328125,
      "loss_xval": 0.404296875,
      "num_input_tokens_seen": 317624192,
      "step": 5669
    },
    {
      "epoch": 12.628062360801781,
      "grad_norm": 33.738197326660156,
      "learning_rate": 1e-06,
      "loss": 0.4112,
      "num_input_tokens_seen": 317676972,
      "step": 5670
    },
    {
      "epoch": 12.628062360801781,
      "loss": 0.4077775180339813,
      "loss_ce": 0.00012370766489766538,
      "loss_iou": 0.1796875,
      "loss_num": 0.009765625,
      "loss_xval": 0.408203125,
      "num_input_tokens_seen": 317676972,
      "step": 5670
    },
    {
      "epoch": 12.630289532293986,
      "grad_norm": 26.519407272338867,
      "learning_rate": 1e-06,
      "loss": 0.4019,
      "num_input_tokens_seen": 317734016,
      "step": 5671
    },
    {
      "epoch": 12.630289532293986,
      "loss": 0.40515443682670593,
      "loss_ce": 0.00012512919784057885,
      "loss_iou": 0.1650390625,
      "loss_num": 0.0147705078125,
      "loss_xval": 0.404296875,
      "num_input_tokens_seen": 317734016,
      "step": 5671
    },
    {
      "epoch": 12.632516703786191,
      "grad_norm": 22.868947982788086,
      "learning_rate": 1e-06,
      "loss": 0.439,
      "num_input_tokens_seen": 317789104,
      "step": 5672
    },
    {
      "epoch": 12.632516703786191,
      "loss": 0.403316855430603,
      "loss_ce": 0.00011859952792292461,
      "loss_iou": 0.1884765625,
      "loss_num": 0.00518798828125,
      "loss_xval": 0.40234375,
      "num_input_tokens_seen": 317789104,
      "step": 5672
    },
    {
      "epoch": 12.634743875278396,
      "grad_norm": 23.70015525817871,
      "learning_rate": 1e-06,
      "loss": 0.4627,
      "num_input_tokens_seen": 317843592,
      "step": 5673
    },
    {
      "epoch": 12.634743875278396,
      "loss": 0.5965588092803955,
      "loss_ce": 0.00012326473370194435,
      "loss_iou": 0.25390625,
      "loss_num": 0.0179443359375,
      "loss_xval": 0.59765625,
      "num_input_tokens_seen": 317843592,
      "step": 5673
    },
    {
      "epoch": 12.6369710467706,
      "grad_norm": 19.277725219726562,
      "learning_rate": 1e-06,
      "loss": 0.4047,
      "num_input_tokens_seen": 317900916,
      "step": 5674
    },
    {
      "epoch": 12.6369710467706,
      "loss": 0.3441046476364136,
      "loss_ce": 0.00011049180466216058,
      "loss_iou": 0.1484375,
      "loss_num": 0.00927734375,
      "loss_xval": 0.34375,
      "num_input_tokens_seen": 317900916,
      "step": 5674
    },
    {
      "epoch": 12.639198218262806,
      "grad_norm": 28.94078826904297,
      "learning_rate": 1e-06,
      "loss": 0.4402,
      "num_input_tokens_seen": 317957468,
      "step": 5675
    },
    {
      "epoch": 12.639198218262806,
      "loss": 0.3978157639503479,
      "loss_ce": 0.00011067395826103166,
      "loss_iou": 0.1591796875,
      "loss_num": 0.015869140625,
      "loss_xval": 0.3984375,
      "num_input_tokens_seen": 317957468,
      "step": 5675
    },
    {
      "epoch": 12.64142538975501,
      "grad_norm": 19.98812484741211,
      "learning_rate": 1e-06,
      "loss": 0.6088,
      "num_input_tokens_seen": 318014916,
      "step": 5676
    },
    {
      "epoch": 12.64142538975501,
      "loss": 0.5992527008056641,
      "loss_ce": 0.0001316650304943323,
      "loss_iou": 0.263671875,
      "loss_num": 0.0140380859375,
      "loss_xval": 0.59765625,
      "num_input_tokens_seen": 318014916,
      "step": 5676
    },
    {
      "epoch": 12.643652561247215,
      "grad_norm": 29.456689834594727,
      "learning_rate": 1e-06,
      "loss": 0.4071,
      "num_input_tokens_seen": 318069856,
      "step": 5677
    },
    {
      "epoch": 12.643652561247215,
      "loss": 0.29436251521110535,
      "loss_ce": 0.00017306354129686952,
      "loss_iou": 0.1376953125,
      "loss_num": 0.003936767578125,
      "loss_xval": 0.294921875,
      "num_input_tokens_seen": 318069856,
      "step": 5677
    },
    {
      "epoch": 12.64587973273942,
      "grad_norm": 29.443206787109375,
      "learning_rate": 1e-06,
      "loss": 0.5605,
      "num_input_tokens_seen": 318127208,
      "step": 5678
    },
    {
      "epoch": 12.64587973273942,
      "loss": 0.6685495972633362,
      "loss_ce": 0.0003367103636264801,
      "loss_iou": 0.28125,
      "loss_num": 0.021240234375,
      "loss_xval": 0.66796875,
      "num_input_tokens_seen": 318127208,
      "step": 5678
    },
    {
      "epoch": 12.648106904231625,
      "grad_norm": 16.127708435058594,
      "learning_rate": 1e-06,
      "loss": 0.4754,
      "num_input_tokens_seen": 318183468,
      "step": 5679
    },
    {
      "epoch": 12.648106904231625,
      "loss": 0.43421459197998047,
      "loss_ce": 0.00013255488011054695,
      "loss_iou": 0.1884765625,
      "loss_num": 0.01153564453125,
      "loss_xval": 0.43359375,
      "num_input_tokens_seen": 318183468,
      "step": 5679
    },
    {
      "epoch": 12.65033407572383,
      "grad_norm": 36.4363899230957,
      "learning_rate": 1e-06,
      "loss": 0.5273,
      "num_input_tokens_seen": 318237944,
      "step": 5680
    },
    {
      "epoch": 12.65033407572383,
      "loss": 0.3793076276779175,
      "loss_ce": 0.00015721011732239276,
      "loss_iou": 0.17578125,
      "loss_num": 0.00555419921875,
      "loss_xval": 0.37890625,
      "num_input_tokens_seen": 318237944,
      "step": 5680
    },
    {
      "epoch": 12.652561247216035,
      "grad_norm": 15.695813179016113,
      "learning_rate": 1e-06,
      "loss": 0.5399,
      "num_input_tokens_seen": 318297084,
      "step": 5681
    },
    {
      "epoch": 12.652561247216035,
      "loss": 0.46922624111175537,
      "loss_ce": 0.00011001349048456177,
      "loss_iou": 0.208984375,
      "loss_num": 0.0101318359375,
      "loss_xval": 0.46875,
      "num_input_tokens_seen": 318297084,
      "step": 5681
    },
    {
      "epoch": 12.654788418708241,
      "grad_norm": 24.64415168762207,
      "learning_rate": 1e-06,
      "loss": 0.4352,
      "num_input_tokens_seen": 318350612,
      "step": 5682
    },
    {
      "epoch": 12.654788418708241,
      "loss": 0.5813356041908264,
      "loss_ce": 0.00015886628534644842,
      "loss_iou": 0.26171875,
      "loss_num": 0.01190185546875,
      "loss_xval": 0.58203125,
      "num_input_tokens_seen": 318350612,
      "step": 5682
    },
    {
      "epoch": 12.657015590200446,
      "grad_norm": 14.695323944091797,
      "learning_rate": 1e-06,
      "loss": 0.5865,
      "num_input_tokens_seen": 318406072,
      "step": 5683
    },
    {
      "epoch": 12.657015590200446,
      "loss": 0.4219498932361603,
      "loss_ce": 0.000120684111607261,
      "loss_iou": 0.19140625,
      "loss_num": 0.00762939453125,
      "loss_xval": 0.421875,
      "num_input_tokens_seen": 318406072,
      "step": 5683
    },
    {
      "epoch": 12.659242761692651,
      "grad_norm": 18.124658584594727,
      "learning_rate": 1e-06,
      "loss": 0.5347,
      "num_input_tokens_seen": 318463084,
      "step": 5684
    },
    {
      "epoch": 12.659242761692651,
      "loss": 0.6086758971214294,
      "loss_ce": 0.00015539961168542504,
      "loss_iou": 0.23046875,
      "loss_num": 0.0296630859375,
      "loss_xval": 0.609375,
      "num_input_tokens_seen": 318463084,
      "step": 5684
    },
    {
      "epoch": 12.661469933184856,
      "grad_norm": 14.113274574279785,
      "learning_rate": 1e-06,
      "loss": 0.466,
      "num_input_tokens_seen": 318520428,
      "step": 5685
    },
    {
      "epoch": 12.661469933184856,
      "loss": 0.5447894930839539,
      "loss_ce": 0.00011178314161952585,
      "loss_iou": 0.2265625,
      "loss_num": 0.018310546875,
      "loss_xval": 0.54296875,
      "num_input_tokens_seen": 318520428,
      "step": 5685
    },
    {
      "epoch": 12.66369710467706,
      "grad_norm": 12.743885040283203,
      "learning_rate": 1e-06,
      "loss": 0.7553,
      "num_input_tokens_seen": 318576972,
      "step": 5686
    },
    {
      "epoch": 12.66369710467706,
      "loss": 0.9210690259933472,
      "loss_ce": 0.0003536554577294737,
      "loss_iou": 0.37109375,
      "loss_num": 0.035888671875,
      "loss_xval": 0.921875,
      "num_input_tokens_seen": 318576972,
      "step": 5686
    },
    {
      "epoch": 12.665924276169266,
      "grad_norm": 14.964234352111816,
      "learning_rate": 1e-06,
      "loss": 0.4453,
      "num_input_tokens_seen": 318630160,
      "step": 5687
    },
    {
      "epoch": 12.665924276169266,
      "loss": 0.42542383074760437,
      "loss_ce": 0.00013086556282360107,
      "loss_iou": 0.1826171875,
      "loss_num": 0.01202392578125,
      "loss_xval": 0.42578125,
      "num_input_tokens_seen": 318630160,
      "step": 5687
    },
    {
      "epoch": 12.66815144766147,
      "grad_norm": 22.878787994384766,
      "learning_rate": 1e-06,
      "loss": 0.5036,
      "num_input_tokens_seen": 318683880,
      "step": 5688
    },
    {
      "epoch": 12.66815144766147,
      "loss": 0.47283750772476196,
      "loss_ce": 0.00012018828419968486,
      "loss_iou": 0.2119140625,
      "loss_num": 0.00982666015625,
      "loss_xval": 0.47265625,
      "num_input_tokens_seen": 318683880,
      "step": 5688
    },
    {
      "epoch": 12.670378619153675,
      "grad_norm": 16.93242073059082,
      "learning_rate": 1e-06,
      "loss": 0.4694,
      "num_input_tokens_seen": 318737980,
      "step": 5689
    },
    {
      "epoch": 12.670378619153675,
      "loss": 0.4250328838825226,
      "loss_ce": 0.00010613650374580175,
      "loss_iou": 0.1875,
      "loss_num": 0.00982666015625,
      "loss_xval": 0.42578125,
      "num_input_tokens_seen": 318737980,
      "step": 5689
    },
    {
      "epoch": 12.67260579064588,
      "grad_norm": 18.397140502929688,
      "learning_rate": 1e-06,
      "loss": 0.4985,
      "num_input_tokens_seen": 318793520,
      "step": 5690
    },
    {
      "epoch": 12.67260579064588,
      "loss": 0.5429917573928833,
      "loss_ce": 0.0001451352145522833,
      "loss_iou": 0.2314453125,
      "loss_num": 0.015869140625,
      "loss_xval": 0.54296875,
      "num_input_tokens_seen": 318793520,
      "step": 5690
    },
    {
      "epoch": 12.674832962138085,
      "grad_norm": 23.1961669921875,
      "learning_rate": 1e-06,
      "loss": 0.4391,
      "num_input_tokens_seen": 318848756,
      "step": 5691
    },
    {
      "epoch": 12.674832962138085,
      "loss": 0.45615941286087036,
      "loss_ce": 0.00010473289148649201,
      "loss_iou": 0.2001953125,
      "loss_num": 0.01141357421875,
      "loss_xval": 0.45703125,
      "num_input_tokens_seen": 318848756,
      "step": 5691
    },
    {
      "epoch": 12.67706013363029,
      "grad_norm": 14.611727714538574,
      "learning_rate": 1e-06,
      "loss": 0.4351,
      "num_input_tokens_seen": 318907276,
      "step": 5692
    },
    {
      "epoch": 12.67706013363029,
      "loss": 0.4373611807823181,
      "loss_ce": 0.0001053235464496538,
      "loss_iou": 0.189453125,
      "loss_num": 0.011474609375,
      "loss_xval": 0.4375,
      "num_input_tokens_seen": 318907276,
      "step": 5692
    },
    {
      "epoch": 12.679287305122495,
      "grad_norm": 16.515594482421875,
      "learning_rate": 1e-06,
      "loss": 0.4227,
      "num_input_tokens_seen": 318962132,
      "step": 5693
    },
    {
      "epoch": 12.679287305122495,
      "loss": 0.4306679964065552,
      "loss_ce": 0.00012597074965015054,
      "loss_iou": 0.1982421875,
      "loss_num": 0.0067138671875,
      "loss_xval": 0.4296875,
      "num_input_tokens_seen": 318962132,
      "step": 5693
    },
    {
      "epoch": 12.6815144766147,
      "grad_norm": 20.435245513916016,
      "learning_rate": 1e-06,
      "loss": 0.6693,
      "num_input_tokens_seen": 319018288,
      "step": 5694
    },
    {
      "epoch": 12.6815144766147,
      "loss": 0.9051839709281921,
      "loss_ce": 0.00015468656783923507,
      "loss_iou": 0.380859375,
      "loss_num": 0.029052734375,
      "loss_xval": 0.90625,
      "num_input_tokens_seen": 319018288,
      "step": 5694
    },
    {
      "epoch": 12.683741648106905,
      "grad_norm": 17.04014778137207,
      "learning_rate": 1e-06,
      "loss": 0.3592,
      "num_input_tokens_seen": 319073932,
      "step": 5695
    },
    {
      "epoch": 12.683741648106905,
      "loss": 0.45563840866088867,
      "loss_ce": 0.00019404885824769735,
      "loss_iou": 0.2060546875,
      "loss_num": 0.0084228515625,
      "loss_xval": 0.455078125,
      "num_input_tokens_seen": 319073932,
      "step": 5695
    },
    {
      "epoch": 12.68596881959911,
      "grad_norm": 18.900136947631836,
      "learning_rate": 1e-06,
      "loss": 0.4399,
      "num_input_tokens_seen": 319131752,
      "step": 5696
    },
    {
      "epoch": 12.68596881959911,
      "loss": 0.4523688554763794,
      "loss_ce": 9.835186938289553e-05,
      "loss_iou": 0.1884765625,
      "loss_num": 0.01495361328125,
      "loss_xval": 0.453125,
      "num_input_tokens_seen": 319131752,
      "step": 5696
    },
    {
      "epoch": 12.688195991091314,
      "grad_norm": 25.752071380615234,
      "learning_rate": 1e-06,
      "loss": 0.4534,
      "num_input_tokens_seen": 319187908,
      "step": 5697
    },
    {
      "epoch": 12.688195991091314,
      "loss": 0.42492321133613586,
      "loss_ce": 0.00011853590694954619,
      "loss_iou": 0.193359375,
      "loss_num": 0.007568359375,
      "loss_xval": 0.42578125,
      "num_input_tokens_seen": 319187908,
      "step": 5697
    },
    {
      "epoch": 12.690423162583519,
      "grad_norm": 19.532682418823242,
      "learning_rate": 1e-06,
      "loss": 0.538,
      "num_input_tokens_seen": 319244400,
      "step": 5698
    },
    {
      "epoch": 12.690423162583519,
      "loss": 0.5101226568222046,
      "loss_ce": 0.00011285034997854382,
      "loss_iou": 0.228515625,
      "loss_num": 0.01068115234375,
      "loss_xval": 0.51171875,
      "num_input_tokens_seen": 319244400,
      "step": 5698
    },
    {
      "epoch": 12.692650334075724,
      "grad_norm": 26.41732406616211,
      "learning_rate": 1e-06,
      "loss": 0.4874,
      "num_input_tokens_seen": 319297184,
      "step": 5699
    },
    {
      "epoch": 12.692650334075724,
      "loss": 0.57191401720047,
      "loss_ce": 0.00013665735605172813,
      "loss_iou": 0.251953125,
      "loss_num": 0.01385498046875,
      "loss_xval": 0.5703125,
      "num_input_tokens_seen": 319297184,
      "step": 5699
    },
    {
      "epoch": 12.694877505567929,
      "grad_norm": 18.265283584594727,
      "learning_rate": 1e-06,
      "loss": 0.4927,
      "num_input_tokens_seen": 319355800,
      "step": 5700
    },
    {
      "epoch": 12.694877505567929,
      "loss": 0.5474244356155396,
      "loss_ce": 0.00012223176599945873,
      "loss_iou": 0.2314453125,
      "loss_num": 0.0169677734375,
      "loss_xval": 0.546875,
      "num_input_tokens_seen": 319355800,
      "step": 5700
    },
    {
      "epoch": 12.697104677060134,
      "grad_norm": 26.21422576904297,
      "learning_rate": 1e-06,
      "loss": 0.7005,
      "num_input_tokens_seen": 319413272,
      "step": 5701
    },
    {
      "epoch": 12.697104677060134,
      "loss": 0.9590729475021362,
      "loss_ce": 0.00021064060274511576,
      "loss_iou": 0.359375,
      "loss_num": 0.04833984375,
      "loss_xval": 0.95703125,
      "num_input_tokens_seen": 319413272,
      "step": 5701
    },
    {
      "epoch": 12.699331848552339,
      "grad_norm": 24.68498992919922,
      "learning_rate": 1e-06,
      "loss": 0.4993,
      "num_input_tokens_seen": 319467372,
      "step": 5702
    },
    {
      "epoch": 12.699331848552339,
      "loss": 0.5787380933761597,
      "loss_ce": 0.0001248328626388684,
      "loss_iou": 0.224609375,
      "loss_num": 0.02587890625,
      "loss_xval": 0.578125,
      "num_input_tokens_seen": 319467372,
      "step": 5702
    },
    {
      "epoch": 12.701559020044543,
      "grad_norm": 20.236909866333008,
      "learning_rate": 1e-06,
      "loss": 0.5838,
      "num_input_tokens_seen": 319524172,
      "step": 5703
    },
    {
      "epoch": 12.701559020044543,
      "loss": 0.6807386875152588,
      "loss_ce": 0.0005628582439385355,
      "loss_iou": 0.2890625,
      "loss_num": 0.0203857421875,
      "loss_xval": 0.6796875,
      "num_input_tokens_seen": 319524172,
      "step": 5703
    },
    {
      "epoch": 12.703786191536748,
      "grad_norm": 27.707962036132812,
      "learning_rate": 1e-06,
      "loss": 0.6471,
      "num_input_tokens_seen": 319578916,
      "step": 5704
    },
    {
      "epoch": 12.703786191536748,
      "loss": 0.8126009702682495,
      "loss_ce": 0.00010092551383422688,
      "loss_iou": 0.357421875,
      "loss_num": 0.019775390625,
      "loss_xval": 0.8125,
      "num_input_tokens_seen": 319578916,
      "step": 5704
    },
    {
      "epoch": 12.706013363028953,
      "grad_norm": 21.496240615844727,
      "learning_rate": 1e-06,
      "loss": 0.449,
      "num_input_tokens_seen": 319634576,
      "step": 5705
    },
    {
      "epoch": 12.706013363028953,
      "loss": 0.4259480834007263,
      "loss_ce": 0.00010578571527730674,
      "loss_iou": 0.181640625,
      "loss_num": 0.01263427734375,
      "loss_xval": 0.42578125,
      "num_input_tokens_seen": 319634576,
      "step": 5705
    },
    {
      "epoch": 12.708240534521158,
      "grad_norm": 17.127723693847656,
      "learning_rate": 1e-06,
      "loss": 0.3865,
      "num_input_tokens_seen": 319689076,
      "step": 5706
    },
    {
      "epoch": 12.708240534521158,
      "loss": 0.3831561207771301,
      "loss_ce": 9.949406376108527e-05,
      "loss_iou": 0.15234375,
      "loss_num": 0.0157470703125,
      "loss_xval": 0.3828125,
      "num_input_tokens_seen": 319689076,
      "step": 5706
    },
    {
      "epoch": 12.710467706013363,
      "grad_norm": 19.543319702148438,
      "learning_rate": 1e-06,
      "loss": 0.7191,
      "num_input_tokens_seen": 319745764,
      "step": 5707
    },
    {
      "epoch": 12.710467706013363,
      "loss": 0.780562162399292,
      "loss_ce": 0.00028868275694549084,
      "loss_iou": 0.337890625,
      "loss_num": 0.02099609375,
      "loss_xval": 0.78125,
      "num_input_tokens_seen": 319745764,
      "step": 5707
    },
    {
      "epoch": 12.712694877505568,
      "grad_norm": 20.706933975219727,
      "learning_rate": 1e-06,
      "loss": 0.3687,
      "num_input_tokens_seen": 319803272,
      "step": 5708
    },
    {
      "epoch": 12.712694877505568,
      "loss": 0.45703423023223877,
      "loss_ce": 0.00018610812549013644,
      "loss_iou": 0.2119140625,
      "loss_num": 0.006805419921875,
      "loss_xval": 0.45703125,
      "num_input_tokens_seen": 319803272,
      "step": 5708
    },
    {
      "epoch": 12.714922048997773,
      "grad_norm": 13.944723129272461,
      "learning_rate": 1e-06,
      "loss": 0.3891,
      "num_input_tokens_seen": 319860864,
      "step": 5709
    },
    {
      "epoch": 12.714922048997773,
      "loss": 0.3766050636768341,
      "loss_ce": 0.00014021531387697905,
      "loss_iou": 0.16796875,
      "loss_num": 0.00830078125,
      "loss_xval": 0.376953125,
      "num_input_tokens_seen": 319860864,
      "step": 5709
    },
    {
      "epoch": 12.717149220489977,
      "grad_norm": 20.481975555419922,
      "learning_rate": 1e-06,
      "loss": 0.639,
      "num_input_tokens_seen": 319916192,
      "step": 5710
    },
    {
      "epoch": 12.717149220489977,
      "loss": 0.6568468809127808,
      "loss_ce": 0.0001086572592612356,
      "loss_iou": 0.265625,
      "loss_num": 0.024658203125,
      "loss_xval": 0.65625,
      "num_input_tokens_seen": 319916192,
      "step": 5710
    },
    {
      "epoch": 12.719376391982182,
      "grad_norm": 14.734237670898438,
      "learning_rate": 1e-06,
      "loss": 0.6377,
      "num_input_tokens_seen": 319974044,
      "step": 5711
    },
    {
      "epoch": 12.719376391982182,
      "loss": 0.8217337131500244,
      "loss_ce": 0.00020050689636263996,
      "loss_iou": 0.3515625,
      "loss_num": 0.0238037109375,
      "loss_xval": 0.8203125,
      "num_input_tokens_seen": 319974044,
      "step": 5711
    },
    {
      "epoch": 12.721603563474387,
      "grad_norm": 16.80459213256836,
      "learning_rate": 1e-06,
      "loss": 0.4424,
      "num_input_tokens_seen": 320027692,
      "step": 5712
    },
    {
      "epoch": 12.721603563474387,
      "loss": 0.46265923976898193,
      "loss_ce": 0.00013479188783094287,
      "loss_iou": 0.2109375,
      "loss_num": 0.0081787109375,
      "loss_xval": 0.462890625,
      "num_input_tokens_seen": 320027692,
      "step": 5712
    },
    {
      "epoch": 12.723830734966592,
      "grad_norm": 13.117379188537598,
      "learning_rate": 1e-06,
      "loss": 0.5347,
      "num_input_tokens_seen": 320084636,
      "step": 5713
    },
    {
      "epoch": 12.723830734966592,
      "loss": 0.4970431923866272,
      "loss_ce": 9.497034625383094e-05,
      "loss_iou": 0.2021484375,
      "loss_num": 0.0186767578125,
      "loss_xval": 0.49609375,
      "num_input_tokens_seen": 320084636,
      "step": 5713
    },
    {
      "epoch": 12.726057906458797,
      "grad_norm": 15.85624885559082,
      "learning_rate": 1e-06,
      "loss": 0.5115,
      "num_input_tokens_seen": 320141228,
      "step": 5714
    },
    {
      "epoch": 12.726057906458797,
      "loss": 0.5403798222541809,
      "loss_ce": 0.0002187040081480518,
      "loss_iou": 0.2294921875,
      "loss_num": 0.016357421875,
      "loss_xval": 0.5390625,
      "num_input_tokens_seen": 320141228,
      "step": 5714
    },
    {
      "epoch": 12.728285077951002,
      "grad_norm": 15.332324981689453,
      "learning_rate": 1e-06,
      "loss": 0.3932,
      "num_input_tokens_seen": 320196208,
      "step": 5715
    },
    {
      "epoch": 12.728285077951002,
      "loss": 0.4100413918495178,
      "loss_ce": 0.00012929215154144913,
      "loss_iou": 0.1708984375,
      "loss_num": 0.0137939453125,
      "loss_xval": 0.41015625,
      "num_input_tokens_seen": 320196208,
      "step": 5715
    },
    {
      "epoch": 12.730512249443207,
      "grad_norm": 21.221384048461914,
      "learning_rate": 1e-06,
      "loss": 0.3587,
      "num_input_tokens_seen": 320252508,
      "step": 5716
    },
    {
      "epoch": 12.730512249443207,
      "loss": 0.27532318234443665,
      "loss_ce": 0.00011566374450922012,
      "loss_iou": 0.12109375,
      "loss_num": 0.00653076171875,
      "loss_xval": 0.275390625,
      "num_input_tokens_seen": 320252508,
      "step": 5716
    },
    {
      "epoch": 12.732739420935411,
      "grad_norm": 19.24123191833496,
      "learning_rate": 1e-06,
      "loss": 0.4894,
      "num_input_tokens_seen": 320308308,
      "step": 5717
    },
    {
      "epoch": 12.732739420935411,
      "loss": 0.44945940375328064,
      "loss_ce": 0.00011856977653224021,
      "loss_iou": 0.1884765625,
      "loss_num": 0.0142822265625,
      "loss_xval": 0.44921875,
      "num_input_tokens_seen": 320308308,
      "step": 5717
    },
    {
      "epoch": 12.734966592427616,
      "grad_norm": 18.568979263305664,
      "learning_rate": 1e-06,
      "loss": 0.5369,
      "num_input_tokens_seen": 320363968,
      "step": 5718
    },
    {
      "epoch": 12.734966592427616,
      "loss": 0.5608214139938354,
      "loss_ce": 0.00015247752889990807,
      "loss_iou": 0.25390625,
      "loss_num": 0.01025390625,
      "loss_xval": 0.5625,
      "num_input_tokens_seen": 320363968,
      "step": 5718
    },
    {
      "epoch": 12.737193763919821,
      "grad_norm": 27.49472427368164,
      "learning_rate": 1e-06,
      "loss": 0.4842,
      "num_input_tokens_seen": 320420160,
      "step": 5719
    },
    {
      "epoch": 12.737193763919821,
      "loss": 0.44296640157699585,
      "loss_ce": 9.531481191515923e-05,
      "loss_iou": 0.173828125,
      "loss_num": 0.0189208984375,
      "loss_xval": 0.443359375,
      "num_input_tokens_seen": 320420160,
      "step": 5719
    },
    {
      "epoch": 12.739420935412026,
      "grad_norm": 35.03486633300781,
      "learning_rate": 1e-06,
      "loss": 0.6408,
      "num_input_tokens_seen": 320476520,
      "step": 5720
    },
    {
      "epoch": 12.739420935412026,
      "loss": 0.6617658734321594,
      "loss_ce": 0.000144752484629862,
      "loss_iou": 0.283203125,
      "loss_num": 0.019287109375,
      "loss_xval": 0.66015625,
      "num_input_tokens_seen": 320476520,
      "step": 5720
    },
    {
      "epoch": 12.74164810690423,
      "grad_norm": 19.232091903686523,
      "learning_rate": 1e-06,
      "loss": 0.3847,
      "num_input_tokens_seen": 320531336,
      "step": 5721
    },
    {
      "epoch": 12.74164810690423,
      "loss": 0.41272228956222534,
      "loss_ce": 0.00012462519225664437,
      "loss_iou": 0.185546875,
      "loss_num": 0.00811767578125,
      "loss_xval": 0.412109375,
      "num_input_tokens_seen": 320531336,
      "step": 5721
    },
    {
      "epoch": 12.743875278396436,
      "grad_norm": 19.980670928955078,
      "learning_rate": 1e-06,
      "loss": 0.4017,
      "num_input_tokens_seen": 320590096,
      "step": 5722
    },
    {
      "epoch": 12.743875278396436,
      "loss": 0.4484846591949463,
      "loss_ce": 0.00012038621207466349,
      "loss_iou": 0.1884765625,
      "loss_num": 0.01422119140625,
      "loss_xval": 0.44921875,
      "num_input_tokens_seen": 320590096,
      "step": 5722
    },
    {
      "epoch": 12.74610244988864,
      "grad_norm": 23.660293579101562,
      "learning_rate": 1e-06,
      "loss": 0.4983,
      "num_input_tokens_seen": 320644036,
      "step": 5723
    },
    {
      "epoch": 12.74610244988864,
      "loss": 0.5904831290245056,
      "loss_ce": 0.0001510925212642178,
      "loss_iou": 0.24609375,
      "loss_num": 0.019775390625,
      "loss_xval": 0.58984375,
      "num_input_tokens_seen": 320644036,
      "step": 5723
    },
    {
      "epoch": 12.748329621380847,
      "grad_norm": 16.64579963684082,
      "learning_rate": 1e-06,
      "loss": 0.4319,
      "num_input_tokens_seen": 320699152,
      "step": 5724
    },
    {
      "epoch": 12.748329621380847,
      "loss": 0.4705604612827301,
      "loss_ce": 0.00010146087151952088,
      "loss_iou": 0.21484375,
      "loss_num": 0.00830078125,
      "loss_xval": 0.470703125,
      "num_input_tokens_seen": 320699152,
      "step": 5724
    },
    {
      "epoch": 12.750556792873052,
      "grad_norm": 24.360746383666992,
      "learning_rate": 1e-06,
      "loss": 0.6527,
      "num_input_tokens_seen": 320751832,
      "step": 5725
    },
    {
      "epoch": 12.750556792873052,
      "loss": 0.7235379219055176,
      "loss_ce": 0.00014920823741704226,
      "loss_iou": 0.287109375,
      "loss_num": 0.0296630859375,
      "loss_xval": 0.72265625,
      "num_input_tokens_seen": 320751832,
      "step": 5725
    },
    {
      "epoch": 12.752783964365257,
      "grad_norm": 18.12118148803711,
      "learning_rate": 1e-06,
      "loss": 0.5865,
      "num_input_tokens_seen": 320807464,
      "step": 5726
    },
    {
      "epoch": 12.752783964365257,
      "loss": 0.45097601413726807,
      "loss_ce": 0.00010932252916973084,
      "loss_iou": 0.2041015625,
      "loss_num": 0.00872802734375,
      "loss_xval": 0.451171875,
      "num_input_tokens_seen": 320807464,
      "step": 5726
    },
    {
      "epoch": 12.755011135857462,
      "grad_norm": 104.34639739990234,
      "learning_rate": 1e-06,
      "loss": 0.6179,
      "num_input_tokens_seen": 320861640,
      "step": 5727
    },
    {
      "epoch": 12.755011135857462,
      "loss": 0.5127081871032715,
      "loss_ce": 0.000134985864860937,
      "loss_iou": 0.2119140625,
      "loss_num": 0.017578125,
      "loss_xval": 0.51171875,
      "num_input_tokens_seen": 320861640,
      "step": 5727
    },
    {
      "epoch": 12.757238307349667,
      "grad_norm": 71.3646011352539,
      "learning_rate": 1e-06,
      "loss": 0.6442,
      "num_input_tokens_seen": 320915204,
      "step": 5728
    },
    {
      "epoch": 12.757238307349667,
      "loss": 0.6957196593284607,
      "loss_ce": 0.00016302938456647098,
      "loss_iou": 0.306640625,
      "loss_num": 0.016357421875,
      "loss_xval": 0.6953125,
      "num_input_tokens_seen": 320915204,
      "step": 5728
    },
    {
      "epoch": 12.759465478841872,
      "grad_norm": 26.151466369628906,
      "learning_rate": 1e-06,
      "loss": 0.3624,
      "num_input_tokens_seen": 320971060,
      "step": 5729
    },
    {
      "epoch": 12.759465478841872,
      "loss": 0.34516897797584534,
      "loss_ce": 0.00010673723591025919,
      "loss_iou": 0.1416015625,
      "loss_num": 0.012451171875,
      "loss_xval": 0.345703125,
      "num_input_tokens_seen": 320971060,
      "step": 5729
    },
    {
      "epoch": 12.761692650334076,
      "grad_norm": 17.449047088623047,
      "learning_rate": 1e-06,
      "loss": 0.4778,
      "num_input_tokens_seen": 321024964,
      "step": 5730
    },
    {
      "epoch": 12.761692650334076,
      "loss": 0.5893593430519104,
      "loss_ce": 0.00012595132284332067,
      "loss_iou": 0.27734375,
      "loss_num": 0.00726318359375,
      "loss_xval": 0.58984375,
      "num_input_tokens_seen": 321024964,
      "step": 5730
    },
    {
      "epoch": 12.763919821826281,
      "grad_norm": 18.459321975708008,
      "learning_rate": 1e-06,
      "loss": 0.511,
      "num_input_tokens_seen": 321081864,
      "step": 5731
    },
    {
      "epoch": 12.763919821826281,
      "loss": 0.5054360628128052,
      "loss_ce": 0.0005532268551178277,
      "loss_iou": 0.2236328125,
      "loss_num": 0.011474609375,
      "loss_xval": 0.50390625,
      "num_input_tokens_seen": 321081864,
      "step": 5731
    },
    {
      "epoch": 12.766146993318486,
      "grad_norm": 14.343801498413086,
      "learning_rate": 1e-06,
      "loss": 0.4483,
      "num_input_tokens_seen": 321134864,
      "step": 5732
    },
    {
      "epoch": 12.766146993318486,
      "loss": 0.4396534860134125,
      "loss_ce": 0.00020037099602632225,
      "loss_iou": 0.1875,
      "loss_num": 0.0126953125,
      "loss_xval": 0.439453125,
      "num_input_tokens_seen": 321134864,
      "step": 5732
    },
    {
      "epoch": 12.768374164810691,
      "grad_norm": 25.014375686645508,
      "learning_rate": 1e-06,
      "loss": 0.4792,
      "num_input_tokens_seen": 321189340,
      "step": 5733
    },
    {
      "epoch": 12.768374164810691,
      "loss": 0.48544344305992126,
      "loss_ce": 9.186341048916802e-05,
      "loss_iou": 0.21484375,
      "loss_num": 0.0113525390625,
      "loss_xval": 0.484375,
      "num_input_tokens_seen": 321189340,
      "step": 5733
    },
    {
      "epoch": 12.770601336302896,
      "grad_norm": 21.6157283782959,
      "learning_rate": 1e-06,
      "loss": 0.6859,
      "num_input_tokens_seen": 321243252,
      "step": 5734
    },
    {
      "epoch": 12.770601336302896,
      "loss": 0.938653290271759,
      "loss_ce": 0.0001766829373082146,
      "loss_iou": 0.396484375,
      "loss_num": 0.02880859375,
      "loss_xval": 0.9375,
      "num_input_tokens_seen": 321243252,
      "step": 5734
    },
    {
      "epoch": 12.7728285077951,
      "grad_norm": 27.44350814819336,
      "learning_rate": 1e-06,
      "loss": 0.3906,
      "num_input_tokens_seen": 321297028,
      "step": 5735
    },
    {
      "epoch": 12.7728285077951,
      "loss": 0.39947906136512756,
      "loss_ce": 0.00012604435323737562,
      "loss_iou": 0.1708984375,
      "loss_num": 0.01153564453125,
      "loss_xval": 0.3984375,
      "num_input_tokens_seen": 321297028,
      "step": 5735
    },
    {
      "epoch": 12.775055679287306,
      "grad_norm": 18.349063873291016,
      "learning_rate": 1e-06,
      "loss": 0.4358,
      "num_input_tokens_seen": 321354732,
      "step": 5736
    },
    {
      "epoch": 12.775055679287306,
      "loss": 0.3706546723842621,
      "loss_ce": 0.00011024883133359253,
      "loss_iou": 0.1689453125,
      "loss_num": 0.006561279296875,
      "loss_xval": 0.37109375,
      "num_input_tokens_seen": 321354732,
      "step": 5736
    },
    {
      "epoch": 12.77728285077951,
      "grad_norm": 34.92224884033203,
      "learning_rate": 1e-06,
      "loss": 0.6155,
      "num_input_tokens_seen": 321408740,
      "step": 5737
    },
    {
      "epoch": 12.77728285077951,
      "loss": 0.7788517475128174,
      "loss_ce": 0.00016523349040653557,
      "loss_iou": 0.361328125,
      "loss_num": 0.0111083984375,
      "loss_xval": 0.77734375,
      "num_input_tokens_seen": 321408740,
      "step": 5737
    },
    {
      "epoch": 12.779510022271715,
      "grad_norm": 19.885868072509766,
      "learning_rate": 1e-06,
      "loss": 0.4942,
      "num_input_tokens_seen": 321465824,
      "step": 5738
    },
    {
      "epoch": 12.779510022271715,
      "loss": 0.49599504470825195,
      "loss_ce": 0.0001454516313970089,
      "loss_iou": 0.2021484375,
      "loss_num": 0.0184326171875,
      "loss_xval": 0.49609375,
      "num_input_tokens_seen": 321465824,
      "step": 5738
    },
    {
      "epoch": 12.78173719376392,
      "grad_norm": 24.478364944458008,
      "learning_rate": 1e-06,
      "loss": 0.4904,
      "num_input_tokens_seen": 321521952,
      "step": 5739
    },
    {
      "epoch": 12.78173719376392,
      "loss": 0.34351837635040283,
      "loss_ce": 0.00013458389730658382,
      "loss_iou": 0.134765625,
      "loss_num": 0.01483154296875,
      "loss_xval": 0.34375,
      "num_input_tokens_seen": 321521952,
      "step": 5739
    },
    {
      "epoch": 12.783964365256125,
      "grad_norm": 34.84294128417969,
      "learning_rate": 1e-06,
      "loss": 0.4395,
      "num_input_tokens_seen": 321577264,
      "step": 5740
    },
    {
      "epoch": 12.783964365256125,
      "loss": 0.396610826253891,
      "loss_ce": 0.00012645949027501047,
      "loss_iou": 0.181640625,
      "loss_num": 0.006591796875,
      "loss_xval": 0.396484375,
      "num_input_tokens_seen": 321577264,
      "step": 5740
    },
    {
      "epoch": 12.78619153674833,
      "grad_norm": 18.09517478942871,
      "learning_rate": 1e-06,
      "loss": 0.5365,
      "num_input_tokens_seen": 321634008,
      "step": 5741
    },
    {
      "epoch": 12.78619153674833,
      "loss": 0.7020202875137329,
      "loss_ce": 0.00011601038568187505,
      "loss_iou": 0.31640625,
      "loss_num": 0.01385498046875,
      "loss_xval": 0.703125,
      "num_input_tokens_seen": 321634008,
      "step": 5741
    },
    {
      "epoch": 12.788418708240535,
      "grad_norm": 20.708227157592773,
      "learning_rate": 1e-06,
      "loss": 0.4856,
      "num_input_tokens_seen": 321690796,
      "step": 5742
    },
    {
      "epoch": 12.788418708240535,
      "loss": 0.5002317428588867,
      "loss_ce": 0.00010964082321152091,
      "loss_iou": 0.20703125,
      "loss_num": 0.01708984375,
      "loss_xval": 0.5,
      "num_input_tokens_seen": 321690796,
      "step": 5742
    },
    {
      "epoch": 12.79064587973274,
      "grad_norm": 22.730310440063477,
      "learning_rate": 1e-06,
      "loss": 0.676,
      "num_input_tokens_seen": 321744736,
      "step": 5743
    },
    {
      "epoch": 12.79064587973274,
      "loss": 0.7551612854003906,
      "loss_ce": 0.00015634976443834603,
      "loss_iou": 0.3203125,
      "loss_num": 0.022705078125,
      "loss_xval": 0.75390625,
      "num_input_tokens_seen": 321744736,
      "step": 5743
    },
    {
      "epoch": 12.792873051224944,
      "grad_norm": 21.29202651977539,
      "learning_rate": 1e-06,
      "loss": 0.512,
      "num_input_tokens_seen": 321801344,
      "step": 5744
    },
    {
      "epoch": 12.792873051224944,
      "loss": 0.39646512269973755,
      "loss_ce": 0.00010285238386131823,
      "loss_iou": 0.171875,
      "loss_num": 0.01055908203125,
      "loss_xval": 0.396484375,
      "num_input_tokens_seen": 321801344,
      "step": 5744
    },
    {
      "epoch": 12.79510022271715,
      "grad_norm": 29.273391723632812,
      "learning_rate": 1e-06,
      "loss": 0.4766,
      "num_input_tokens_seen": 321857176,
      "step": 5745
    },
    {
      "epoch": 12.79510022271715,
      "loss": 0.6940093040466309,
      "loss_ce": 0.0001615973305888474,
      "loss_iou": 0.30078125,
      "loss_num": 0.0184326171875,
      "loss_xval": 0.6953125,
      "num_input_tokens_seen": 321857176,
      "step": 5745
    },
    {
      "epoch": 12.797327394209354,
      "grad_norm": 15.866458892822266,
      "learning_rate": 1e-06,
      "loss": 0.4327,
      "num_input_tokens_seen": 321913700,
      "step": 5746
    },
    {
      "epoch": 12.797327394209354,
      "loss": 0.3660312294960022,
      "loss_ce": 0.00012545731442514807,
      "loss_iou": 0.162109375,
      "loss_num": 0.00836181640625,
      "loss_xval": 0.365234375,
      "num_input_tokens_seen": 321913700,
      "step": 5746
    },
    {
      "epoch": 12.799554565701559,
      "grad_norm": 22.117265701293945,
      "learning_rate": 1e-06,
      "loss": 0.4373,
      "num_input_tokens_seen": 321968160,
      "step": 5747
    },
    {
      "epoch": 12.799554565701559,
      "loss": 0.40611228346824646,
      "loss_ce": 0.00010643507994245738,
      "loss_iou": 0.177734375,
      "loss_num": 0.0101318359375,
      "loss_xval": 0.40625,
      "num_input_tokens_seen": 321968160,
      "step": 5747
    },
    {
      "epoch": 12.801781737193764,
      "grad_norm": 17.11176300048828,
      "learning_rate": 1e-06,
      "loss": 0.6027,
      "num_input_tokens_seen": 322023584,
      "step": 5748
    },
    {
      "epoch": 12.801781737193764,
      "loss": 0.6447978615760803,
      "loss_ce": 0.00014456806820817292,
      "loss_iou": 0.2578125,
      "loss_num": 0.0252685546875,
      "loss_xval": 0.64453125,
      "num_input_tokens_seen": 322023584,
      "step": 5748
    },
    {
      "epoch": 12.804008908685969,
      "grad_norm": 31.402372360229492,
      "learning_rate": 1e-06,
      "loss": 0.4811,
      "num_input_tokens_seen": 322082440,
      "step": 5749
    },
    {
      "epoch": 12.804008908685969,
      "loss": 0.39622175693511963,
      "loss_ce": 0.00010357976134400815,
      "loss_iou": 0.1708984375,
      "loss_num": 0.010986328125,
      "loss_xval": 0.396484375,
      "num_input_tokens_seen": 322082440,
      "step": 5749
    },
    {
      "epoch": 12.806236080178174,
      "grad_norm": 15.063833236694336,
      "learning_rate": 1e-06,
      "loss": 0.4572,
      "num_input_tokens_seen": 322138808,
      "step": 5750
    },
    {
      "epoch": 12.806236080178174,
      "eval_seeclick_web_CIoU": 0.5856587886810303,
      "eval_seeclick_web_GIoU": 0.583136647939682,
      "eval_seeclick_web_IoU": 0.6040867269039154,
      "eval_seeclick_web_MAE_all": 0.0155873978510499,
      "eval_seeclick_web_MAE_h": 0.007719088811427355,
      "eval_seeclick_web_MAE_w": 0.015769829973578453,
      "eval_seeclick_web_MAE_x_boxes": 0.009643178898841143,
      "eval_seeclick_web_MAE_y_boxes": 0.021339073311537504,
      "eval_seeclick_web_inside_bbox": 0.9166666567325592,
      "eval_seeclick_web_loss": 0.9100306630134583,
      "eval_seeclick_web_loss_ce": 0.00018124158668797463,
      "eval_seeclick_web_loss_iou": 0.417724609375,
      "eval_seeclick_web_loss_num": 0.012420654296875,
      "eval_seeclick_web_loss_xval": 0.8966064453125,
      "eval_seeclick_web_runtime": 22.4296,
      "eval_seeclick_web_samples_per_second": 2.229,
      "eval_seeclick_web_steps_per_second": 0.089,
      "num_input_tokens_seen": 322138808,
      "step": 5750
    },
    {
      "epoch": 12.806236080178174,
      "eval_icons_CIoU": 0.29134996235370636,
      "eval_icons_GIoU": 0.31351229548454285,
      "eval_icons_IoU": 0.3615979105234146,
      "eval_icons_MAE_all": 0.05436762422323227,
      "eval_icons_MAE_h": 0.036054016556590796,
      "eval_icons_MAE_w": 0.045310807414352894,
      "eval_icons_MAE_x_boxes": 0.05722554586827755,
      "eval_icons_MAE_y_boxes": 0.03668802231550217,
      "eval_icons_inside_bbox": 0.6059027910232544,
      "eval_icons_loss": 1.6602177619934082,
      "eval_icons_loss_ce": 0.00021313664910849184,
      "eval_icons_loss_iou": 0.668701171875,
      "eval_icons_loss_num": 0.053501129150390625,
      "eval_icons_loss_xval": 1.60546875,
      "eval_icons_runtime": 18.5773,
      "eval_icons_samples_per_second": 2.691,
      "eval_icons_steps_per_second": 0.108,
      "num_input_tokens_seen": 322138808,
      "step": 5750
    },
    {
      "epoch": 12.806236080178174,
      "eval_screenspot_CIoU": 0.35812679926554364,
      "eval_screenspot_GIoU": 0.37103012204170227,
      "eval_screenspot_IoU": 0.43642645080884296,
      "eval_screenspot_MAE_all": 0.05802058925231298,
      "eval_screenspot_MAE_h": 0.03933714081843694,
      "eval_screenspot_MAE_w": 0.06783675154050191,
      "eval_screenspot_MAE_x_boxes": 0.07193685260911782,
      "eval_screenspot_MAE_y_boxes": 0.03876749984920025,
      "eval_screenspot_inside_bbox": 0.6966666579246521,
      "eval_screenspot_loss": 1.6010042428970337,
      "eval_screenspot_loss_ce": 0.00024160779624556503,
      "eval_screenspot_loss_iou": 0.664306640625,
      "eval_screenspot_loss_num": 0.06610488891601562,
      "eval_screenspot_loss_xval": 1.6591796875,
      "eval_screenspot_runtime": 32.6846,
      "eval_screenspot_samples_per_second": 2.723,
      "eval_screenspot_steps_per_second": 0.092,
      "num_input_tokens_seen": 322138808,
      "step": 5750
    },
    {
      "epoch": 12.806236080178174,
      "eval_compot_CIoU": 0.3403613865375519,
      "eval_compot_GIoU": 0.3508555740118027,
      "eval_compot_IoU": 0.3995762914419174,
      "eval_compot_MAE_all": 0.018100187182426453,
      "eval_compot_MAE_h": 0.009705259930342436,
      "eval_compot_MAE_w": 0.02139892429113388,
      "eval_compot_MAE_x_boxes": 0.0300004780292511,
      "eval_compot_MAE_y_boxes": 0.007140443893149495,
      "eval_compot_inside_bbox": 0.6302083432674408,
      "eval_compot_loss": 1.413923740386963,
      "eval_compot_loss_ce": 0.00017305201618000865,
      "eval_compot_loss_iou": 0.6578369140625,
      "eval_compot_loss_num": 0.016866683959960938,
      "eval_compot_loss_xval": 1.399169921875,
      "eval_compot_runtime": 19.9376,
      "eval_compot_samples_per_second": 2.508,
      "eval_compot_steps_per_second": 0.1,
      "num_input_tokens_seen": 322138808,
      "step": 5750
    },
    {
      "epoch": 12.806236080178174,
      "eval_custom_ui_val_CIoU": 0.47460226383474136,
      "eval_custom_ui_val_GIoU": 0.4856240898370743,
      "eval_custom_ui_val_IoU": 0.5360172192255656,
      "eval_custom_ui_val_MAE_all": 0.028572862967848778,
      "eval_custom_ui_val_MAE_h": 0.016446984217812616,
      "eval_custom_ui_val_MAE_w": 0.03571122243172593,
      "eval_custom_ui_val_MAE_x_boxes": 0.03412022265709109,
      "eval_custom_ui_val_MAE_y_boxes": 0.014356489810678694,
      "eval_custom_ui_val_inside_bbox": 0.7685185207260979,
      "eval_custom_ui_val_loss": 1.1905332803726196,
      "eval_custom_ui_val_loss_ce": 0.00019961439506409483,
      "eval_custom_ui_val_loss_iou": 0.5112169053819444,
      "eval_custom_ui_val_loss_num": 0.026023017035590276,
      "eval_custom_ui_val_loss_xval": 1.1525065104166667,
      "eval_custom_ui_val_runtime": 60.1958,
      "eval_custom_ui_val_samples_per_second": 4.402,
      "eval_custom_ui_val_steps_per_second": 0.15,
      "num_input_tokens_seen": 322138808,
      "step": 5750
    },
    {
      "epoch": 12.806236080178174,
      "loss": 0.9020213484764099,
      "loss_ce": 0.00016588816652074456,
      "loss_iou": 0.400390625,
      "loss_num": 0.0206298828125,
      "loss_xval": 0.90234375,
      "num_input_tokens_seen": 322138808,
      "step": 5750
    },
    {
      "epoch": 12.808463251670378,
      "grad_norm": 38.99467086791992,
      "learning_rate": 1e-06,
      "loss": 0.7047,
      "num_input_tokens_seen": 322196804,
      "step": 5751
    },
    {
      "epoch": 12.808463251670378,
      "loss": 0.594732940196991,
      "loss_ce": 0.0001284430327359587,
      "loss_iou": 0.26171875,
      "loss_num": 0.013671875,
      "loss_xval": 0.59375,
      "num_input_tokens_seen": 322196804,
      "step": 5751
    },
    {
      "epoch": 12.810690423162583,
      "grad_norm": 18.36638832092285,
      "learning_rate": 1e-06,
      "loss": 0.3216,
      "num_input_tokens_seen": 322252400,
      "step": 5752
    },
    {
      "epoch": 12.810690423162583,
      "loss": 0.3923182189464569,
      "loss_ce": 0.00010630811448208988,
      "loss_iou": 0.1708984375,
      "loss_num": 0.010009765625,
      "loss_xval": 0.392578125,
      "num_input_tokens_seen": 322252400,
      "step": 5752
    },
    {
      "epoch": 12.812917594654788,
      "grad_norm": 29.22452163696289,
      "learning_rate": 1e-06,
      "loss": 0.6287,
      "num_input_tokens_seen": 322308460,
      "step": 5753
    },
    {
      "epoch": 12.812917594654788,
      "loss": 0.6383031606674194,
      "loss_ce": 0.00011960881238337606,
      "loss_iou": 0.2578125,
      "loss_num": 0.0244140625,
      "loss_xval": 0.63671875,
      "num_input_tokens_seen": 322308460,
      "step": 5753
    },
    {
      "epoch": 12.815144766146993,
      "grad_norm": 21.79752540588379,
      "learning_rate": 1e-06,
      "loss": 0.4675,
      "num_input_tokens_seen": 322362036,
      "step": 5754
    },
    {
      "epoch": 12.815144766146993,
      "loss": 0.5063232183456421,
      "loss_ce": 9.764191054273397e-05,
      "loss_iou": 0.1904296875,
      "loss_num": 0.025146484375,
      "loss_xval": 0.5078125,
      "num_input_tokens_seen": 322362036,
      "step": 5754
    },
    {
      "epoch": 12.817371937639198,
      "grad_norm": 16.744384765625,
      "learning_rate": 1e-06,
      "loss": 0.6904,
      "num_input_tokens_seen": 322418496,
      "step": 5755
    },
    {
      "epoch": 12.817371937639198,
      "loss": 0.6048610806465149,
      "loss_ce": 0.000124765865621157,
      "loss_iou": 0.251953125,
      "loss_num": 0.0206298828125,
      "loss_xval": 0.60546875,
      "num_input_tokens_seen": 322418496,
      "step": 5755
    },
    {
      "epoch": 12.819599109131403,
      "grad_norm": 17.073532104492188,
      "learning_rate": 1e-06,
      "loss": 0.5932,
      "num_input_tokens_seen": 322473040,
      "step": 5756
    },
    {
      "epoch": 12.819599109131403,
      "loss": 0.733528733253479,
      "loss_ce": 0.00013029485126025975,
      "loss_iou": 0.3125,
      "loss_num": 0.0211181640625,
      "loss_xval": 0.734375,
      "num_input_tokens_seen": 322473040,
      "step": 5756
    },
    {
      "epoch": 12.821826280623608,
      "grad_norm": 20.584901809692383,
      "learning_rate": 1e-06,
      "loss": 0.7059,
      "num_input_tokens_seen": 322528604,
      "step": 5757
    },
    {
      "epoch": 12.821826280623608,
      "loss": 0.8982992768287659,
      "loss_ce": 0.00010590371675789356,
      "loss_iou": 0.38671875,
      "loss_num": 0.025146484375,
      "loss_xval": 0.8984375,
      "num_input_tokens_seen": 322528604,
      "step": 5757
    },
    {
      "epoch": 12.824053452115812,
      "grad_norm": 20.415708541870117,
      "learning_rate": 1e-06,
      "loss": 0.5264,
      "num_input_tokens_seen": 322583604,
      "step": 5758
    },
    {
      "epoch": 12.824053452115812,
      "loss": 0.4630866050720215,
      "loss_ce": 0.00013495869643520564,
      "loss_iou": 0.1962890625,
      "loss_num": 0.01409912109375,
      "loss_xval": 0.462890625,
      "num_input_tokens_seen": 322583604,
      "step": 5758
    },
    {
      "epoch": 12.826280623608017,
      "grad_norm": 15.553330421447754,
      "learning_rate": 1e-06,
      "loss": 0.3762,
      "num_input_tokens_seen": 322642080,
      "step": 5759
    },
    {
      "epoch": 12.826280623608017,
      "loss": 0.4416653513908386,
      "loss_ce": 0.0002591132652014494,
      "loss_iou": 0.2041015625,
      "loss_num": 0.006622314453125,
      "loss_xval": 0.44140625,
      "num_input_tokens_seen": 322642080,
      "step": 5759
    },
    {
      "epoch": 12.828507795100222,
      "grad_norm": 14.952034950256348,
      "learning_rate": 1e-06,
      "loss": 0.5567,
      "num_input_tokens_seen": 322697788,
      "step": 5760
    },
    {
      "epoch": 12.828507795100222,
      "loss": 0.548295259475708,
      "loss_ce": 0.00019955809693783522,
      "loss_iou": 0.2373046875,
      "loss_num": 0.0146484375,
      "loss_xval": 0.546875,
      "num_input_tokens_seen": 322697788,
      "step": 5760
    },
    {
      "epoch": 12.830734966592427,
      "grad_norm": 25.951000213623047,
      "learning_rate": 1e-06,
      "loss": 0.4949,
      "num_input_tokens_seen": 322752844,
      "step": 5761
    },
    {
      "epoch": 12.830734966592427,
      "loss": 0.4571736454963684,
      "loss_ce": 0.00014240591553971171,
      "loss_iou": 0.20703125,
      "loss_num": 0.0087890625,
      "loss_xval": 0.45703125,
      "num_input_tokens_seen": 322752844,
      "step": 5761
    },
    {
      "epoch": 12.832962138084632,
      "grad_norm": 43.291751861572266,
      "learning_rate": 1e-06,
      "loss": 0.5937,
      "num_input_tokens_seen": 322808240,
      "step": 5762
    },
    {
      "epoch": 12.832962138084632,
      "loss": 0.598831832408905,
      "loss_ce": 0.0001380076282657683,
      "loss_iou": 0.248046875,
      "loss_num": 0.0206298828125,
      "loss_xval": 0.59765625,
      "num_input_tokens_seen": 322808240,
      "step": 5762
    },
    {
      "epoch": 12.835189309576837,
      "grad_norm": 21.269405364990234,
      "learning_rate": 1e-06,
      "loss": 0.5252,
      "num_input_tokens_seen": 322863788,
      "step": 5763
    },
    {
      "epoch": 12.835189309576837,
      "loss": 0.5071128010749817,
      "loss_ce": 0.00015480615547858179,
      "loss_iou": 0.2373046875,
      "loss_num": 0.006744384765625,
      "loss_xval": 0.5078125,
      "num_input_tokens_seen": 322863788,
      "step": 5763
    },
    {
      "epoch": 12.837416481069042,
      "grad_norm": 25.716829299926758,
      "learning_rate": 1e-06,
      "loss": 0.5679,
      "num_input_tokens_seen": 322917120,
      "step": 5764
    },
    {
      "epoch": 12.837416481069042,
      "loss": 0.6708798408508301,
      "loss_ce": 0.0002255703293485567,
      "loss_iou": 0.306640625,
      "loss_num": 0.01141357421875,
      "loss_xval": 0.671875,
      "num_input_tokens_seen": 322917120,
      "step": 5764
    },
    {
      "epoch": 12.839643652561247,
      "grad_norm": 26.497522354125977,
      "learning_rate": 1e-06,
      "loss": 0.7658,
      "num_input_tokens_seen": 322972616,
      "step": 5765
    },
    {
      "epoch": 12.839643652561247,
      "loss": 0.7098448872566223,
      "loss_ce": 0.0001281315489904955,
      "loss_iou": 0.283203125,
      "loss_num": 0.0284423828125,
      "loss_xval": 0.7109375,
      "num_input_tokens_seen": 322972616,
      "step": 5765
    },
    {
      "epoch": 12.841870824053451,
      "grad_norm": 76.6500244140625,
      "learning_rate": 1e-06,
      "loss": 0.6196,
      "num_input_tokens_seen": 323026852,
      "step": 5766
    },
    {
      "epoch": 12.841870824053451,
      "loss": 0.8302084803581238,
      "loss_ce": 0.00013035557640250772,
      "loss_iou": 0.337890625,
      "loss_num": 0.03125,
      "loss_xval": 0.828125,
      "num_input_tokens_seen": 323026852,
      "step": 5766
    },
    {
      "epoch": 12.844097995545656,
      "grad_norm": 62.04082107543945,
      "learning_rate": 1e-06,
      "loss": 0.6541,
      "num_input_tokens_seen": 323081780,
      "step": 5767
    },
    {
      "epoch": 12.844097995545656,
      "loss": 0.5926365256309509,
      "loss_ce": 0.00010719904821598902,
      "loss_iou": 0.2734375,
      "loss_num": 0.00933837890625,
      "loss_xval": 0.59375,
      "num_input_tokens_seen": 323081780,
      "step": 5767
    },
    {
      "epoch": 12.846325167037861,
      "grad_norm": 15.04085636138916,
      "learning_rate": 1e-06,
      "loss": 0.3717,
      "num_input_tokens_seen": 323138412,
      "step": 5768
    },
    {
      "epoch": 12.846325167037861,
      "loss": 0.3711279630661011,
      "loss_ce": 0.00015629868721589446,
      "loss_iou": 0.162109375,
      "loss_num": 0.00921630859375,
      "loss_xval": 0.37109375,
      "num_input_tokens_seen": 323138412,
      "step": 5768
    },
    {
      "epoch": 12.848552338530066,
      "grad_norm": 16.58966064453125,
      "learning_rate": 1e-06,
      "loss": 0.468,
      "num_input_tokens_seen": 323194440,
      "step": 5769
    },
    {
      "epoch": 12.848552338530066,
      "loss": 0.47240063548088074,
      "loss_ce": 0.00011059406824642792,
      "loss_iou": 0.201171875,
      "loss_num": 0.013916015625,
      "loss_xval": 0.47265625,
      "num_input_tokens_seen": 323194440,
      "step": 5769
    },
    {
      "epoch": 12.85077951002227,
      "grad_norm": 15.409582138061523,
      "learning_rate": 1e-06,
      "loss": 0.6628,
      "num_input_tokens_seen": 323250432,
      "step": 5770
    },
    {
      "epoch": 12.85077951002227,
      "loss": 0.5492154955863953,
      "loss_ce": 0.0001432021672371775,
      "loss_iou": 0.2470703125,
      "loss_num": 0.0108642578125,
      "loss_xval": 0.55078125,
      "num_input_tokens_seen": 323250432,
      "step": 5770
    },
    {
      "epoch": 12.853006681514476,
      "grad_norm": 17.768123626708984,
      "learning_rate": 1e-06,
      "loss": 0.467,
      "num_input_tokens_seen": 323305916,
      "step": 5771
    },
    {
      "epoch": 12.853006681514476,
      "loss": 0.3561987280845642,
      "loss_ce": 0.00011960781557718292,
      "loss_iou": 0.1611328125,
      "loss_num": 0.0067138671875,
      "loss_xval": 0.35546875,
      "num_input_tokens_seen": 323305916,
      "step": 5771
    },
    {
      "epoch": 12.855233853006682,
      "grad_norm": 26.621932983398438,
      "learning_rate": 1e-06,
      "loss": 0.6905,
      "num_input_tokens_seen": 323363612,
      "step": 5772
    },
    {
      "epoch": 12.855233853006682,
      "loss": 0.621519923210144,
      "loss_ce": 0.000304146931739524,
      "loss_iou": 0.287109375,
      "loss_num": 0.0093994140625,
      "loss_xval": 0.62109375,
      "num_input_tokens_seen": 323363612,
      "step": 5772
    },
    {
      "epoch": 12.857461024498887,
      "grad_norm": 36.93680953979492,
      "learning_rate": 1e-06,
      "loss": 0.4199,
      "num_input_tokens_seen": 323418516,
      "step": 5773
    },
    {
      "epoch": 12.857461024498887,
      "loss": 0.4533904790878296,
      "loss_ce": 0.0001434021396562457,
      "loss_iou": 0.1923828125,
      "loss_num": 0.01385498046875,
      "loss_xval": 0.453125,
      "num_input_tokens_seen": 323418516,
      "step": 5773
    },
    {
      "epoch": 12.859688195991092,
      "grad_norm": 12.92249870300293,
      "learning_rate": 1e-06,
      "loss": 0.3578,
      "num_input_tokens_seen": 323475544,
      "step": 5774
    },
    {
      "epoch": 12.859688195991092,
      "loss": 0.3216310143470764,
      "loss_ce": 9.778769162949175e-05,
      "loss_iou": 0.1474609375,
      "loss_num": 0.005126953125,
      "loss_xval": 0.322265625,
      "num_input_tokens_seen": 323475544,
      "step": 5774
    },
    {
      "epoch": 12.861915367483297,
      "grad_norm": 15.084147453308105,
      "learning_rate": 1e-06,
      "loss": 0.5663,
      "num_input_tokens_seen": 323531332,
      "step": 5775
    },
    {
      "epoch": 12.861915367483297,
      "loss": 0.5575112104415894,
      "loss_ce": 0.0001381381880491972,
      "loss_iou": 0.255859375,
      "loss_num": 0.00933837890625,
      "loss_xval": 0.55859375,
      "num_input_tokens_seen": 323531332,
      "step": 5775
    },
    {
      "epoch": 12.864142538975502,
      "grad_norm": 23.849468231201172,
      "learning_rate": 1e-06,
      "loss": 0.4872,
      "num_input_tokens_seen": 323586272,
      "step": 5776
    },
    {
      "epoch": 12.864142538975502,
      "loss": 0.3724247217178345,
      "loss_ce": 0.00011026560969185084,
      "loss_iou": 0.16015625,
      "loss_num": 0.010498046875,
      "loss_xval": 0.373046875,
      "num_input_tokens_seen": 323586272,
      "step": 5776
    },
    {
      "epoch": 12.866369710467707,
      "grad_norm": 16.385759353637695,
      "learning_rate": 1e-06,
      "loss": 0.4698,
      "num_input_tokens_seen": 323642804,
      "step": 5777
    },
    {
      "epoch": 12.866369710467707,
      "loss": 0.6077675819396973,
      "loss_ce": 0.00010155436029890552,
      "loss_iou": 0.26953125,
      "loss_num": 0.01348876953125,
      "loss_xval": 0.609375,
      "num_input_tokens_seen": 323642804,
      "step": 5777
    },
    {
      "epoch": 12.868596881959911,
      "grad_norm": 13.454336166381836,
      "learning_rate": 1e-06,
      "loss": 0.3738,
      "num_input_tokens_seen": 323699508,
      "step": 5778
    },
    {
      "epoch": 12.868596881959911,
      "loss": 0.3336215317249298,
      "loss_ce": 0.00012544200581032783,
      "loss_iou": 0.1552734375,
      "loss_num": 0.004486083984375,
      "loss_xval": 0.333984375,
      "num_input_tokens_seen": 323699508,
      "step": 5778
    },
    {
      "epoch": 12.870824053452116,
      "grad_norm": 32.175724029541016,
      "learning_rate": 1e-06,
      "loss": 0.3957,
      "num_input_tokens_seen": 323753524,
      "step": 5779
    },
    {
      "epoch": 12.870824053452116,
      "loss": 0.38816550374031067,
      "loss_ce": 0.00010398947779322043,
      "loss_iou": 0.1787109375,
      "loss_num": 0.006317138671875,
      "loss_xval": 0.388671875,
      "num_input_tokens_seen": 323753524,
      "step": 5779
    },
    {
      "epoch": 12.873051224944321,
      "grad_norm": 16.53754234313965,
      "learning_rate": 1e-06,
      "loss": 0.3905,
      "num_input_tokens_seen": 323807584,
      "step": 5780
    },
    {
      "epoch": 12.873051224944321,
      "loss": 0.460315078496933,
      "loss_ce": 0.00011000905215041712,
      "loss_iou": 0.189453125,
      "loss_num": 0.0162353515625,
      "loss_xval": 0.4609375,
      "num_input_tokens_seen": 323807584,
      "step": 5780
    },
    {
      "epoch": 12.875278396436526,
      "grad_norm": 33.93476486206055,
      "learning_rate": 1e-06,
      "loss": 0.5997,
      "num_input_tokens_seen": 323865248,
      "step": 5781
    },
    {
      "epoch": 12.875278396436526,
      "loss": 0.8404668569564819,
      "loss_ce": 0.00013486588431987911,
      "loss_iou": 0.365234375,
      "loss_num": 0.021728515625,
      "loss_xval": 0.83984375,
      "num_input_tokens_seen": 323865248,
      "step": 5781
    },
    {
      "epoch": 12.877505567928731,
      "grad_norm": 24.43963623046875,
      "learning_rate": 1e-06,
      "loss": 0.4762,
      "num_input_tokens_seen": 323921064,
      "step": 5782
    },
    {
      "epoch": 12.877505567928731,
      "loss": 0.5513740181922913,
      "loss_ce": 0.00010449648834764957,
      "loss_iou": 0.2138671875,
      "loss_num": 0.0247802734375,
      "loss_xval": 0.55078125,
      "num_input_tokens_seen": 323921064,
      "step": 5782
    },
    {
      "epoch": 12.879732739420936,
      "grad_norm": 19.904743194580078,
      "learning_rate": 1e-06,
      "loss": 0.4667,
      "num_input_tokens_seen": 323979316,
      "step": 5783
    },
    {
      "epoch": 12.879732739420936,
      "loss": 0.4628022313117981,
      "loss_ce": 0.00015576645091641694,
      "loss_iou": 0.1923828125,
      "loss_num": 0.015625,
      "loss_xval": 0.462890625,
      "num_input_tokens_seen": 323979316,
      "step": 5783
    },
    {
      "epoch": 12.88195991091314,
      "grad_norm": 13.64018440246582,
      "learning_rate": 1e-06,
      "loss": 0.3038,
      "num_input_tokens_seen": 324037372,
      "step": 5784
    },
    {
      "epoch": 12.88195991091314,
      "loss": 0.361078143119812,
      "loss_ce": 0.00011623941827565432,
      "loss_iou": 0.15234375,
      "loss_num": 0.01116943359375,
      "loss_xval": 0.361328125,
      "num_input_tokens_seen": 324037372,
      "step": 5784
    },
    {
      "epoch": 12.884187082405345,
      "grad_norm": 75.058349609375,
      "learning_rate": 1e-06,
      "loss": 0.6451,
      "num_input_tokens_seen": 324095976,
      "step": 5785
    },
    {
      "epoch": 12.884187082405345,
      "loss": 0.6566053628921509,
      "loss_ce": 0.0011488739401102066,
      "loss_iou": 0.267578125,
      "loss_num": 0.023681640625,
      "loss_xval": 0.65625,
      "num_input_tokens_seen": 324095976,
      "step": 5785
    },
    {
      "epoch": 12.88641425389755,
      "grad_norm": 13.707096099853516,
      "learning_rate": 1e-06,
      "loss": 0.5832,
      "num_input_tokens_seen": 324150696,
      "step": 5786
    },
    {
      "epoch": 12.88641425389755,
      "loss": 0.3526431918144226,
      "loss_ce": 0.00010411690163891762,
      "loss_iou": 0.154296875,
      "loss_num": 0.00860595703125,
      "loss_xval": 0.3515625,
      "num_input_tokens_seen": 324150696,
      "step": 5786
    },
    {
      "epoch": 12.888641425389755,
      "grad_norm": 18.37203025817871,
      "learning_rate": 1e-06,
      "loss": 0.5096,
      "num_input_tokens_seen": 324204900,
      "step": 5787
    },
    {
      "epoch": 12.888641425389755,
      "loss": 0.5298916697502136,
      "loss_ce": 0.00010648852912709117,
      "loss_iou": 0.228515625,
      "loss_num": 0.01434326171875,
      "loss_xval": 0.53125,
      "num_input_tokens_seen": 324204900,
      "step": 5787
    },
    {
      "epoch": 12.89086859688196,
      "grad_norm": 18.70199203491211,
      "learning_rate": 1e-06,
      "loss": 0.4261,
      "num_input_tokens_seen": 324262452,
      "step": 5788
    },
    {
      "epoch": 12.89086859688196,
      "loss": 0.3846321105957031,
      "loss_ce": 0.00011065039871027693,
      "loss_iou": 0.1787109375,
      "loss_num": 0.005279541015625,
      "loss_xval": 0.384765625,
      "num_input_tokens_seen": 324262452,
      "step": 5788
    },
    {
      "epoch": 12.893095768374165,
      "grad_norm": 22.604270935058594,
      "learning_rate": 1e-06,
      "loss": 0.5116,
      "num_input_tokens_seen": 324313472,
      "step": 5789
    },
    {
      "epoch": 12.893095768374165,
      "loss": 0.5472476482391357,
      "loss_ce": 0.00012846475874539465,
      "loss_iou": 0.240234375,
      "loss_num": 0.01336669921875,
      "loss_xval": 0.546875,
      "num_input_tokens_seen": 324313472,
      "step": 5789
    },
    {
      "epoch": 12.89532293986637,
      "grad_norm": 19.753644943237305,
      "learning_rate": 1e-06,
      "loss": 0.4183,
      "num_input_tokens_seen": 324369044,
      "step": 5790
    },
    {
      "epoch": 12.89532293986637,
      "loss": 0.5045735836029053,
      "loss_ce": 0.00011799067578976974,
      "loss_iou": 0.21875,
      "loss_num": 0.01348876953125,
      "loss_xval": 0.50390625,
      "num_input_tokens_seen": 324369044,
      "step": 5790
    },
    {
      "epoch": 12.897550111358575,
      "grad_norm": 31.045259475708008,
      "learning_rate": 1e-06,
      "loss": 0.6084,
      "num_input_tokens_seen": 324424920,
      "step": 5791
    },
    {
      "epoch": 12.897550111358575,
      "loss": 0.7539187669754028,
      "loss_ce": 0.0001345592609141022,
      "loss_iou": 0.326171875,
      "loss_num": 0.0205078125,
      "loss_xval": 0.75390625,
      "num_input_tokens_seen": 324424920,
      "step": 5791
    },
    {
      "epoch": 12.89977728285078,
      "grad_norm": 23.0539493560791,
      "learning_rate": 1e-06,
      "loss": 0.5878,
      "num_input_tokens_seen": 324481128,
      "step": 5792
    },
    {
      "epoch": 12.89977728285078,
      "loss": 0.44552081823349,
      "loss_ce": 0.00014727559755556285,
      "loss_iou": 0.185546875,
      "loss_num": 0.0147705078125,
      "loss_xval": 0.4453125,
      "num_input_tokens_seen": 324481128,
      "step": 5792
    },
    {
      "epoch": 12.902004454342984,
      "grad_norm": 20.192691802978516,
      "learning_rate": 1e-06,
      "loss": 0.4366,
      "num_input_tokens_seen": 324537512,
      "step": 5793
    },
    {
      "epoch": 12.902004454342984,
      "loss": 0.4317473769187927,
      "loss_ce": 0.00010673434007912874,
      "loss_iou": 0.1875,
      "loss_num": 0.0111083984375,
      "loss_xval": 0.431640625,
      "num_input_tokens_seen": 324537512,
      "step": 5793
    },
    {
      "epoch": 12.90423162583519,
      "grad_norm": 23.165922164916992,
      "learning_rate": 1e-06,
      "loss": 0.5654,
      "num_input_tokens_seen": 324591436,
      "step": 5794
    },
    {
      "epoch": 12.90423162583519,
      "loss": 0.7123044729232788,
      "loss_ce": 0.000146272053825669,
      "loss_iou": 0.2890625,
      "loss_num": 0.026611328125,
      "loss_xval": 0.7109375,
      "num_input_tokens_seen": 324591436,
      "step": 5794
    },
    {
      "epoch": 12.906458797327394,
      "grad_norm": 19.26278305053711,
      "learning_rate": 1e-06,
      "loss": 0.6626,
      "num_input_tokens_seen": 324646868,
      "step": 5795
    },
    {
      "epoch": 12.906458797327394,
      "loss": 0.9051787257194519,
      "loss_ce": 0.0001494084281148389,
      "loss_iou": 0.373046875,
      "loss_num": 0.031494140625,
      "loss_xval": 0.90625,
      "num_input_tokens_seen": 324646868,
      "step": 5795
    },
    {
      "epoch": 12.908685968819599,
      "grad_norm": 18.52065658569336,
      "learning_rate": 1e-06,
      "loss": 0.4669,
      "num_input_tokens_seen": 324704652,
      "step": 5796
    },
    {
      "epoch": 12.908685968819599,
      "loss": 0.4736798405647278,
      "loss_ce": 0.0001691254146862775,
      "loss_iou": 0.18359375,
      "loss_num": 0.02099609375,
      "loss_xval": 0.47265625,
      "num_input_tokens_seen": 324704652,
      "step": 5796
    },
    {
      "epoch": 12.910913140311804,
      "grad_norm": 22.1650333404541,
      "learning_rate": 1e-06,
      "loss": 0.5009,
      "num_input_tokens_seen": 324762108,
      "step": 5797
    },
    {
      "epoch": 12.910913140311804,
      "loss": 0.44127357006073,
      "loss_ce": 0.00011143650044687092,
      "loss_iou": 0.1943359375,
      "loss_num": 0.010498046875,
      "loss_xval": 0.44140625,
      "num_input_tokens_seen": 324762108,
      "step": 5797
    },
    {
      "epoch": 12.913140311804009,
      "grad_norm": 18.897441864013672,
      "learning_rate": 1e-06,
      "loss": 0.4502,
      "num_input_tokens_seen": 324817916,
      "step": 5798
    },
    {
      "epoch": 12.913140311804009,
      "loss": 0.480578750371933,
      "loss_ce": 0.00010997521167155355,
      "loss_iou": 0.2158203125,
      "loss_num": 0.00958251953125,
      "loss_xval": 0.48046875,
      "num_input_tokens_seen": 324817916,
      "step": 5798
    },
    {
      "epoch": 12.915367483296214,
      "grad_norm": 13.311334609985352,
      "learning_rate": 1e-06,
      "loss": 0.5007,
      "num_input_tokens_seen": 324872960,
      "step": 5799
    },
    {
      "epoch": 12.915367483296214,
      "loss": 0.5006000995635986,
      "loss_ce": 0.00011186262418050319,
      "loss_iou": 0.2158203125,
      "loss_num": 0.013671875,
      "loss_xval": 0.5,
      "num_input_tokens_seen": 324872960,
      "step": 5799
    },
    {
      "epoch": 12.917594654788418,
      "grad_norm": 12.26301097869873,
      "learning_rate": 1e-06,
      "loss": 0.4789,
      "num_input_tokens_seen": 324927908,
      "step": 5800
    },
    {
      "epoch": 12.917594654788418,
      "loss": 0.47144681215286255,
      "loss_ce": 0.00013335124822333455,
      "loss_iou": 0.2109375,
      "loss_num": 0.0096435546875,
      "loss_xval": 0.470703125,
      "num_input_tokens_seen": 324927908,
      "step": 5800
    },
    {
      "epoch": 12.919821826280623,
      "grad_norm": 16.52623176574707,
      "learning_rate": 1e-06,
      "loss": 0.6662,
      "num_input_tokens_seen": 324986040,
      "step": 5801
    },
    {
      "epoch": 12.919821826280623,
      "loss": 0.5938632488250732,
      "loss_ce": 0.0012118464801460505,
      "loss_iou": 0.263671875,
      "loss_num": 0.01275634765625,
      "loss_xval": 0.59375,
      "num_input_tokens_seen": 324986040,
      "step": 5801
    },
    {
      "epoch": 12.922048997772828,
      "grad_norm": 29.222322463989258,
      "learning_rate": 1e-06,
      "loss": 0.4547,
      "num_input_tokens_seen": 325043160,
      "step": 5802
    },
    {
      "epoch": 12.922048997772828,
      "loss": 0.4256219267845154,
      "loss_ce": 8.482815610477701e-05,
      "loss_iou": 0.1787109375,
      "loss_num": 0.0135498046875,
      "loss_xval": 0.42578125,
      "num_input_tokens_seen": 325043160,
      "step": 5802
    },
    {
      "epoch": 12.924276169265033,
      "grad_norm": 19.258743286132812,
      "learning_rate": 1e-06,
      "loss": 0.3797,
      "num_input_tokens_seen": 325099556,
      "step": 5803
    },
    {
      "epoch": 12.924276169265033,
      "loss": 0.4342288076877594,
      "loss_ce": 0.00014675800048280507,
      "loss_iou": 0.1904296875,
      "loss_num": 0.01068115234375,
      "loss_xval": 0.43359375,
      "num_input_tokens_seen": 325099556,
      "step": 5803
    },
    {
      "epoch": 12.926503340757238,
      "grad_norm": 18.667795181274414,
      "learning_rate": 1e-06,
      "loss": 0.5903,
      "num_input_tokens_seen": 325154616,
      "step": 5804
    },
    {
      "epoch": 12.926503340757238,
      "loss": 0.5471016764640808,
      "loss_ce": 0.00016564295219723135,
      "loss_iou": 0.255859375,
      "loss_num": 0.007232666015625,
      "loss_xval": 0.546875,
      "num_input_tokens_seen": 325154616,
      "step": 5804
    },
    {
      "epoch": 12.928730512249443,
      "grad_norm": 19.413196563720703,
      "learning_rate": 1e-06,
      "loss": 0.6071,
      "num_input_tokens_seen": 325208228,
      "step": 5805
    },
    {
      "epoch": 12.928730512249443,
      "loss": 0.6254149079322815,
      "loss_ce": 0.00010969527647830546,
      "loss_iou": 0.25390625,
      "loss_num": 0.0235595703125,
      "loss_xval": 0.625,
      "num_input_tokens_seen": 325208228,
      "step": 5805
    },
    {
      "epoch": 12.930957683741648,
      "grad_norm": 16.120576858520508,
      "learning_rate": 1e-06,
      "loss": 0.5249,
      "num_input_tokens_seen": 325261784,
      "step": 5806
    },
    {
      "epoch": 12.930957683741648,
      "loss": 0.5421345829963684,
      "loss_ce": 0.0001423685171175748,
      "loss_iou": 0.234375,
      "loss_num": 0.01470947265625,
      "loss_xval": 0.54296875,
      "num_input_tokens_seen": 325261784,
      "step": 5806
    },
    {
      "epoch": 12.933184855233852,
      "grad_norm": 14.40259075164795,
      "learning_rate": 1e-06,
      "loss": 0.4978,
      "num_input_tokens_seen": 325318096,
      "step": 5807
    },
    {
      "epoch": 12.933184855233852,
      "loss": 0.33383339643478394,
      "loss_ce": 0.00015419043484143913,
      "loss_iou": 0.1494140625,
      "loss_num": 0.0069580078125,
      "loss_xval": 0.333984375,
      "num_input_tokens_seen": 325318096,
      "step": 5807
    },
    {
      "epoch": 12.935412026726057,
      "grad_norm": 29.450130462646484,
      "learning_rate": 1e-06,
      "loss": 0.5965,
      "num_input_tokens_seen": 325372384,
      "step": 5808
    },
    {
      "epoch": 12.935412026726057,
      "loss": 0.4921649694442749,
      "loss_ce": 9.954207052942365e-05,
      "loss_iou": 0.2158203125,
      "loss_num": 0.01190185546875,
      "loss_xval": 0.4921875,
      "num_input_tokens_seen": 325372384,
      "step": 5808
    },
    {
      "epoch": 12.937639198218262,
      "grad_norm": 14.617000579833984,
      "learning_rate": 1e-06,
      "loss": 0.3659,
      "num_input_tokens_seen": 325427388,
      "step": 5809
    },
    {
      "epoch": 12.937639198218262,
      "loss": 0.4301881492137909,
      "loss_ce": 0.0001344462507404387,
      "loss_iou": 0.1689453125,
      "loss_num": 0.0184326171875,
      "loss_xval": 0.4296875,
      "num_input_tokens_seen": 325427388,
      "step": 5809
    },
    {
      "epoch": 12.939866369710467,
      "grad_norm": 24.749897003173828,
      "learning_rate": 1e-06,
      "loss": 0.655,
      "num_input_tokens_seen": 325481232,
      "step": 5810
    },
    {
      "epoch": 12.939866369710467,
      "loss": 0.7332806587219238,
      "loss_ce": 0.0001263371086679399,
      "loss_iou": 0.314453125,
      "loss_num": 0.0208740234375,
      "loss_xval": 0.734375,
      "num_input_tokens_seen": 325481232,
      "step": 5810
    },
    {
      "epoch": 12.942093541202672,
      "grad_norm": 17.5683650970459,
      "learning_rate": 1e-06,
      "loss": 0.4979,
      "num_input_tokens_seen": 325535700,
      "step": 5811
    },
    {
      "epoch": 12.942093541202672,
      "loss": 0.6848030090332031,
      "loss_ce": 0.00011064937280025333,
      "loss_iou": 0.310546875,
      "loss_num": 0.01275634765625,
      "loss_xval": 0.68359375,
      "num_input_tokens_seen": 325535700,
      "step": 5811
    },
    {
      "epoch": 12.944320712694877,
      "grad_norm": 25.585346221923828,
      "learning_rate": 1e-06,
      "loss": 0.4126,
      "num_input_tokens_seen": 325594156,
      "step": 5812
    },
    {
      "epoch": 12.944320712694877,
      "loss": 0.37169232964515686,
      "loss_ce": 0.00011027594155166298,
      "loss_iou": 0.1513671875,
      "loss_num": 0.013671875,
      "loss_xval": 0.37109375,
      "num_input_tokens_seen": 325594156,
      "step": 5812
    },
    {
      "epoch": 12.946547884187082,
      "grad_norm": 18.788799285888672,
      "learning_rate": 1e-06,
      "loss": 0.6043,
      "num_input_tokens_seen": 325648944,
      "step": 5813
    },
    {
      "epoch": 12.946547884187082,
      "loss": 0.6141434907913208,
      "loss_ce": 0.0001298107672482729,
      "loss_iou": 0.25390625,
      "loss_num": 0.021240234375,
      "loss_xval": 0.61328125,
      "num_input_tokens_seen": 325648944,
      "step": 5813
    },
    {
      "epoch": 12.948775055679288,
      "grad_norm": 21.269174575805664,
      "learning_rate": 1e-06,
      "loss": 0.8081,
      "num_input_tokens_seen": 325706188,
      "step": 5814
    },
    {
      "epoch": 12.948775055679288,
      "loss": 0.8276509046554565,
      "loss_ce": 0.0002583618916105479,
      "loss_iou": 0.326171875,
      "loss_num": 0.034912109375,
      "loss_xval": 0.828125,
      "num_input_tokens_seen": 325706188,
      "step": 5814
    },
    {
      "epoch": 12.951002227171493,
      "grad_norm": 26.132108688354492,
      "learning_rate": 1e-06,
      "loss": 0.5818,
      "num_input_tokens_seen": 325759016,
      "step": 5815
    },
    {
      "epoch": 12.951002227171493,
      "loss": 0.7224629521369934,
      "loss_ce": 0.0004475700843613595,
      "loss_iou": 0.296875,
      "loss_num": 0.0257568359375,
      "loss_xval": 0.72265625,
      "num_input_tokens_seen": 325759016,
      "step": 5815
    },
    {
      "epoch": 12.953229398663698,
      "grad_norm": 19.72740364074707,
      "learning_rate": 1e-06,
      "loss": 0.3361,
      "num_input_tokens_seen": 325815532,
      "step": 5816
    },
    {
      "epoch": 12.953229398663698,
      "loss": 0.43498772382736206,
      "loss_ce": 0.00011223182082176208,
      "loss_iou": 0.166015625,
      "loss_num": 0.0205078125,
      "loss_xval": 0.435546875,
      "num_input_tokens_seen": 325815532,
      "step": 5816
    },
    {
      "epoch": 12.955456570155903,
      "grad_norm": 34.82518005371094,
      "learning_rate": 1e-06,
      "loss": 0.6078,
      "num_input_tokens_seen": 325871972,
      "step": 5817
    },
    {
      "epoch": 12.955456570155903,
      "loss": 0.743614912033081,
      "loss_ce": 0.00020671662059612572,
      "loss_iou": 0.29296875,
      "loss_num": 0.031494140625,
      "loss_xval": 0.7421875,
      "num_input_tokens_seen": 325871972,
      "step": 5817
    },
    {
      "epoch": 12.957683741648108,
      "grad_norm": 101.52168273925781,
      "learning_rate": 1e-06,
      "loss": 0.5968,
      "num_input_tokens_seen": 325928992,
      "step": 5818
    },
    {
      "epoch": 12.957683741648108,
      "loss": 0.7953479886054993,
      "loss_ce": 0.00012093935220036656,
      "loss_iou": 0.330078125,
      "loss_num": 0.027099609375,
      "loss_xval": 0.796875,
      "num_input_tokens_seen": 325928992,
      "step": 5818
    },
    {
      "epoch": 12.959910913140313,
      "grad_norm": 19.320589065551758,
      "learning_rate": 1e-06,
      "loss": 0.8007,
      "num_input_tokens_seen": 325984976,
      "step": 5819
    },
    {
      "epoch": 12.959910913140313,
      "loss": 0.8446329236030579,
      "loss_ce": 0.00015049244393594563,
      "loss_iou": 0.341796875,
      "loss_num": 0.0322265625,
      "loss_xval": 0.84375,
      "num_input_tokens_seen": 325984976,
      "step": 5819
    },
    {
      "epoch": 12.962138084632517,
      "grad_norm": 17.67026710510254,
      "learning_rate": 1e-06,
      "loss": 0.4272,
      "num_input_tokens_seen": 326041104,
      "step": 5820
    },
    {
      "epoch": 12.962138084632517,
      "loss": 0.3929480314254761,
      "loss_ce": 0.0001257585099665448,
      "loss_iou": 0.1748046875,
      "loss_num": 0.0086669921875,
      "loss_xval": 0.392578125,
      "num_input_tokens_seen": 326041104,
      "step": 5820
    },
    {
      "epoch": 12.964365256124722,
      "grad_norm": 17.830753326416016,
      "learning_rate": 1e-06,
      "loss": 0.4385,
      "num_input_tokens_seen": 326094916,
      "step": 5821
    },
    {
      "epoch": 12.964365256124722,
      "loss": 0.38365253806114197,
      "loss_ce": 0.00010762730380520225,
      "loss_iou": 0.1640625,
      "loss_num": 0.01104736328125,
      "loss_xval": 0.3828125,
      "num_input_tokens_seen": 326094916,
      "step": 5821
    },
    {
      "epoch": 12.966592427616927,
      "grad_norm": 28.51872444152832,
      "learning_rate": 1e-06,
      "loss": 0.495,
      "num_input_tokens_seen": 326151036,
      "step": 5822
    },
    {
      "epoch": 12.966592427616927,
      "loss": 0.4512762427330017,
      "loss_ce": 0.0007147122523747385,
      "loss_iou": 0.1904296875,
      "loss_num": 0.0140380859375,
      "loss_xval": 0.451171875,
      "num_input_tokens_seen": 326151036,
      "step": 5822
    },
    {
      "epoch": 12.968819599109132,
      "grad_norm": 19.198762893676758,
      "learning_rate": 1e-06,
      "loss": 0.7932,
      "num_input_tokens_seen": 326205816,
      "step": 5823
    },
    {
      "epoch": 12.968819599109132,
      "loss": 0.7585012316703796,
      "loss_ce": 0.00020045909332111478,
      "loss_iou": 0.3125,
      "loss_num": 0.02685546875,
      "loss_xval": 0.7578125,
      "num_input_tokens_seen": 326205816,
      "step": 5823
    },
    {
      "epoch": 12.971046770601337,
      "grad_norm": 18.618919372558594,
      "learning_rate": 1e-06,
      "loss": 0.5058,
      "num_input_tokens_seen": 326260604,
      "step": 5824
    },
    {
      "epoch": 12.971046770601337,
      "loss": 0.6246432662010193,
      "loss_ce": 0.00013153106556273997,
      "loss_iou": 0.279296875,
      "loss_num": 0.013427734375,
      "loss_xval": 0.625,
      "num_input_tokens_seen": 326260604,
      "step": 5824
    },
    {
      "epoch": 12.973273942093542,
      "grad_norm": 14.253881454467773,
      "learning_rate": 1e-06,
      "loss": 0.4857,
      "num_input_tokens_seen": 326318360,
      "step": 5825
    },
    {
      "epoch": 12.973273942093542,
      "loss": 0.4617213010787964,
      "loss_ce": 0.00017345792730338871,
      "loss_iou": 0.1875,
      "loss_num": 0.0172119140625,
      "loss_xval": 0.4609375,
      "num_input_tokens_seen": 326318360,
      "step": 5825
    },
    {
      "epoch": 12.975501113585747,
      "grad_norm": 21.891193389892578,
      "learning_rate": 1e-06,
      "loss": 0.5982,
      "num_input_tokens_seen": 326375376,
      "step": 5826
    },
    {
      "epoch": 12.975501113585747,
      "loss": 0.6038088202476501,
      "loss_ce": 0.00017113759531639516,
      "loss_iou": 0.236328125,
      "loss_num": 0.026123046875,
      "loss_xval": 0.60546875,
      "num_input_tokens_seen": 326375376,
      "step": 5826
    },
    {
      "epoch": 12.977728285077951,
      "grad_norm": 24.07076072692871,
      "learning_rate": 1e-06,
      "loss": 0.412,
      "num_input_tokens_seen": 326430360,
      "step": 5827
    },
    {
      "epoch": 12.977728285077951,
      "loss": 0.5371109247207642,
      "loss_ce": 0.00012366127339191735,
      "loss_iou": 0.2421875,
      "loss_num": 0.01055908203125,
      "loss_xval": 0.53515625,
      "num_input_tokens_seen": 326430360,
      "step": 5827
    },
    {
      "epoch": 12.979955456570156,
      "grad_norm": 24.092050552368164,
      "learning_rate": 1e-06,
      "loss": 0.5159,
      "num_input_tokens_seen": 326485148,
      "step": 5828
    },
    {
      "epoch": 12.979955456570156,
      "loss": 0.6072977781295776,
      "loss_ce": 0.00012003826850559562,
      "loss_iou": 0.26171875,
      "loss_num": 0.0167236328125,
      "loss_xval": 0.60546875,
      "num_input_tokens_seen": 326485148,
      "step": 5828
    },
    {
      "epoch": 12.982182628062361,
      "grad_norm": 15.061753273010254,
      "learning_rate": 1e-06,
      "loss": 0.4169,
      "num_input_tokens_seen": 326541436,
      "step": 5829
    },
    {
      "epoch": 12.982182628062361,
      "loss": 0.37990421056747437,
      "loss_ce": 0.00014345324598252773,
      "loss_iou": 0.15625,
      "loss_num": 0.01324462890625,
      "loss_xval": 0.37890625,
      "num_input_tokens_seen": 326541436,
      "step": 5829
    },
    {
      "epoch": 12.984409799554566,
      "grad_norm": 17.364686965942383,
      "learning_rate": 1e-06,
      "loss": 0.5024,
      "num_input_tokens_seen": 326598392,
      "step": 5830
    },
    {
      "epoch": 12.984409799554566,
      "loss": 0.47910645604133606,
      "loss_ce": 0.00010255920642521232,
      "loss_iou": 0.2041015625,
      "loss_num": 0.01416015625,
      "loss_xval": 0.478515625,
      "num_input_tokens_seen": 326598392,
      "step": 5830
    },
    {
      "epoch": 12.98663697104677,
      "grad_norm": 18.025129318237305,
      "learning_rate": 1e-06,
      "loss": 0.4841,
      "num_input_tokens_seen": 326654788,
      "step": 5831
    },
    {
      "epoch": 12.98663697104677,
      "loss": 0.39598119258880615,
      "loss_ce": 0.0001071398874046281,
      "loss_iou": 0.1787109375,
      "loss_num": 0.007781982421875,
      "loss_xval": 0.396484375,
      "num_input_tokens_seen": 326654788,
      "step": 5831
    },
    {
      "epoch": 12.988864142538976,
      "grad_norm": 43.578460693359375,
      "learning_rate": 1e-06,
      "loss": 0.5461,
      "num_input_tokens_seen": 326710960,
      "step": 5832
    },
    {
      "epoch": 12.988864142538976,
      "loss": 0.588495135307312,
      "loss_ce": 0.00011619819269981235,
      "loss_iou": 0.2392578125,
      "loss_num": 0.0218505859375,
      "loss_xval": 0.58984375,
      "num_input_tokens_seen": 326710960,
      "step": 5832
    },
    {
      "epoch": 12.99109131403118,
      "grad_norm": 16.58260154724121,
      "learning_rate": 1e-06,
      "loss": 0.5595,
      "num_input_tokens_seen": 326767560,
      "step": 5833
    },
    {
      "epoch": 12.99109131403118,
      "loss": 0.6758525371551514,
      "loss_ce": 0.00019332932424731553,
      "loss_iou": 0.265625,
      "loss_num": 0.029296875,
      "loss_xval": 0.67578125,
      "num_input_tokens_seen": 326767560,
      "step": 5833
    },
    {
      "epoch": 12.993318485523385,
      "grad_norm": 100.04027557373047,
      "learning_rate": 1e-06,
      "loss": 0.4254,
      "num_input_tokens_seen": 326824960,
      "step": 5834
    },
    {
      "epoch": 12.993318485523385,
      "loss": 0.4192938804626465,
      "loss_ce": 0.00010442556231282651,
      "loss_iou": 0.17578125,
      "loss_num": 0.01348876953125,
      "loss_xval": 0.419921875,
      "num_input_tokens_seen": 326824960,
      "step": 5834
    },
    {
      "epoch": 12.99554565701559,
      "grad_norm": 18.884227752685547,
      "learning_rate": 1e-06,
      "loss": 0.4109,
      "num_input_tokens_seen": 326882800,
      "step": 5835
    },
    {
      "epoch": 12.99554565701559,
      "loss": 0.41633379459381104,
      "loss_ce": 0.00013507826952263713,
      "loss_iou": 0.193359375,
      "loss_num": 0.00604248046875,
      "loss_xval": 0.416015625,
      "num_input_tokens_seen": 326882800,
      "step": 5835
    },
    {
      "epoch": 12.997772828507795,
      "grad_norm": 20.262374877929688,
      "learning_rate": 1e-06,
      "loss": 0.4506,
      "num_input_tokens_seen": 326940800,
      "step": 5836
    },
    {
      "epoch": 12.997772828507795,
      "loss": 0.48572030663490295,
      "loss_ce": 0.00012461420556064695,
      "loss_iou": 0.2099609375,
      "loss_num": 0.01300048828125,
      "loss_xval": 0.486328125,
      "num_input_tokens_seen": 326940800,
      "step": 5836
    },
    {
      "epoch": 13.0,
      "grad_norm": 17.950050354003906,
      "learning_rate": 1e-06,
      "loss": 0.4372,
      "num_input_tokens_seen": 326994644,
      "step": 5837
    },
    {
      "epoch": 13.0,
      "loss": 0.47044771909713745,
      "loss_ce": 0.00011081169941462576,
      "loss_iou": 0.181640625,
      "loss_num": 0.021728515625,
      "loss_xval": 0.470703125,
      "num_input_tokens_seen": 326994644,
      "step": 5837
    },
    {
      "epoch": 13.002227171492205,
      "grad_norm": 18.163402557373047,
      "learning_rate": 1e-06,
      "loss": 0.3952,
      "num_input_tokens_seen": 327051472,
      "step": 5838
    },
    {
      "epoch": 13.002227171492205,
      "loss": 0.45926615595817566,
      "loss_ce": 0.0001597180962562561,
      "loss_iou": 0.18359375,
      "loss_num": 0.0185546875,
      "loss_xval": 0.458984375,
      "num_input_tokens_seen": 327051472,
      "step": 5838
    },
    {
      "epoch": 13.00445434298441,
      "grad_norm": 17.97021484375,
      "learning_rate": 1e-06,
      "loss": 0.4858,
      "num_input_tokens_seen": 327107924,
      "step": 5839
    },
    {
      "epoch": 13.00445434298441,
      "loss": 0.6263518929481506,
      "loss_ce": 0.00013117733760736883,
      "loss_iou": 0.27734375,
      "loss_num": 0.014404296875,
      "loss_xval": 0.625,
      "num_input_tokens_seen": 327107924,
      "step": 5839
    },
    {
      "epoch": 13.006681514476615,
      "grad_norm": 28.137418746948242,
      "learning_rate": 1e-06,
      "loss": 0.4594,
      "num_input_tokens_seen": 327166408,
      "step": 5840
    },
    {
      "epoch": 13.006681514476615,
      "loss": 0.4613088369369507,
      "loss_ce": 0.0001271882065339014,
      "loss_iou": 0.2138671875,
      "loss_num": 0.0067138671875,
      "loss_xval": 0.4609375,
      "num_input_tokens_seen": 327166408,
      "step": 5840
    },
    {
      "epoch": 13.00890868596882,
      "grad_norm": 19.74036979675293,
      "learning_rate": 1e-06,
      "loss": 0.5521,
      "num_input_tokens_seen": 327223808,
      "step": 5841
    },
    {
      "epoch": 13.00890868596882,
      "loss": 0.39747354388237,
      "loss_ce": 0.00013468455290421844,
      "loss_iou": 0.173828125,
      "loss_num": 0.00970458984375,
      "loss_xval": 0.396484375,
      "num_input_tokens_seen": 327223808,
      "step": 5841
    },
    {
      "epoch": 13.011135857461024,
      "grad_norm": 40.84404754638672,
      "learning_rate": 1e-06,
      "loss": 0.5209,
      "num_input_tokens_seen": 327280344,
      "step": 5842
    },
    {
      "epoch": 13.011135857461024,
      "loss": 0.3576486110687256,
      "loss_ce": 0.0001046471661538817,
      "loss_iou": 0.162109375,
      "loss_num": 0.006500244140625,
      "loss_xval": 0.357421875,
      "num_input_tokens_seen": 327280344,
      "step": 5842
    },
    {
      "epoch": 13.01336302895323,
      "grad_norm": 25.01186180114746,
      "learning_rate": 1e-06,
      "loss": 0.4771,
      "num_input_tokens_seen": 327332880,
      "step": 5843
    },
    {
      "epoch": 13.01336302895323,
      "loss": 0.4530574679374695,
      "loss_ce": 0.00011555968376342207,
      "loss_iou": 0.201171875,
      "loss_num": 0.0101318359375,
      "loss_xval": 0.453125,
      "num_input_tokens_seen": 327332880,
      "step": 5843
    },
    {
      "epoch": 13.015590200445434,
      "grad_norm": 17.630598068237305,
      "learning_rate": 1e-06,
      "loss": 0.5489,
      "num_input_tokens_seen": 327390252,
      "step": 5844
    },
    {
      "epoch": 13.015590200445434,
      "loss": 0.5149204730987549,
      "loss_ce": 0.00011946188897127286,
      "loss_iou": 0.216796875,
      "loss_num": 0.0164794921875,
      "loss_xval": 0.515625,
      "num_input_tokens_seen": 327390252,
      "step": 5844
    },
    {
      "epoch": 13.017817371937639,
      "grad_norm": 16.314720153808594,
      "learning_rate": 1e-06,
      "loss": 0.4371,
      "num_input_tokens_seen": 327445668,
      "step": 5845
    },
    {
      "epoch": 13.017817371937639,
      "loss": 0.34429049491882324,
      "loss_ce": 0.0001132564211729914,
      "loss_iou": 0.142578125,
      "loss_num": 0.01190185546875,
      "loss_xval": 0.34375,
      "num_input_tokens_seen": 327445668,
      "step": 5845
    },
    {
      "epoch": 13.020044543429844,
      "grad_norm": 20.126976013183594,
      "learning_rate": 1e-06,
      "loss": 0.3745,
      "num_input_tokens_seen": 327502572,
      "step": 5846
    },
    {
      "epoch": 13.020044543429844,
      "loss": 0.3755548298358917,
      "loss_ce": 0.00012758253433275968,
      "loss_iou": 0.1787109375,
      "loss_num": 0.0035247802734375,
      "loss_xval": 0.375,
      "num_input_tokens_seen": 327502572,
      "step": 5846
    },
    {
      "epoch": 13.022271714922049,
      "grad_norm": 18.101938247680664,
      "learning_rate": 1e-06,
      "loss": 0.4179,
      "num_input_tokens_seen": 327560524,
      "step": 5847
    },
    {
      "epoch": 13.022271714922049,
      "loss": 0.4644816517829895,
      "loss_ce": 0.00012616706953849643,
      "loss_iou": 0.208984375,
      "loss_num": 0.0093994140625,
      "loss_xval": 0.46484375,
      "num_input_tokens_seen": 327560524,
      "step": 5847
    },
    {
      "epoch": 13.024498886414253,
      "grad_norm": 22.762983322143555,
      "learning_rate": 1e-06,
      "loss": 0.5214,
      "num_input_tokens_seen": 327615932,
      "step": 5848
    },
    {
      "epoch": 13.024498886414253,
      "loss": 0.6657699346542358,
      "loss_ce": 0.00012048571079503745,
      "loss_iou": 0.2890625,
      "loss_num": 0.0174560546875,
      "loss_xval": 0.6640625,
      "num_input_tokens_seen": 327615932,
      "step": 5848
    },
    {
      "epoch": 13.026726057906458,
      "grad_norm": 38.849609375,
      "learning_rate": 1e-06,
      "loss": 0.5811,
      "num_input_tokens_seen": 327669496,
      "step": 5849
    },
    {
      "epoch": 13.026726057906458,
      "loss": 0.3934794068336487,
      "loss_ce": 0.00010783905599964783,
      "loss_iou": 0.158203125,
      "loss_num": 0.0155029296875,
      "loss_xval": 0.392578125,
      "num_input_tokens_seen": 327669496,
      "step": 5849
    },
    {
      "epoch": 13.028953229398663,
      "grad_norm": 13.994296073913574,
      "learning_rate": 1e-06,
      "loss": 0.573,
      "num_input_tokens_seen": 327726764,
      "step": 5850
    },
    {
      "epoch": 13.028953229398663,
      "loss": 0.5137879252433777,
      "loss_ce": 0.00011606388579821214,
      "loss_iou": 0.22265625,
      "loss_num": 0.0137939453125,
      "loss_xval": 0.515625,
      "num_input_tokens_seen": 327726764,
      "step": 5850
    },
    {
      "epoch": 13.031180400890868,
      "grad_norm": 29.13859748840332,
      "learning_rate": 1e-06,
      "loss": 0.7501,
      "num_input_tokens_seen": 327783880,
      "step": 5851
    },
    {
      "epoch": 13.031180400890868,
      "loss": 0.9725739359855652,
      "loss_ce": 0.00016182669787667692,
      "loss_iou": 0.3828125,
      "loss_num": 0.041748046875,
      "loss_xval": 0.97265625,
      "num_input_tokens_seen": 327783880,
      "step": 5851
    },
    {
      "epoch": 13.033407572383073,
      "grad_norm": 15.132269859313965,
      "learning_rate": 1e-06,
      "loss": 0.4839,
      "num_input_tokens_seen": 327842228,
      "step": 5852
    },
    {
      "epoch": 13.033407572383073,
      "loss": 0.5506539344787598,
      "loss_ce": 0.00011678160808514804,
      "loss_iou": 0.240234375,
      "loss_num": 0.01416015625,
      "loss_xval": 0.55078125,
      "num_input_tokens_seen": 327842228,
      "step": 5852
    },
    {
      "epoch": 13.035634743875278,
      "grad_norm": 21.988449096679688,
      "learning_rate": 1e-06,
      "loss": 0.4047,
      "num_input_tokens_seen": 327899700,
      "step": 5853
    },
    {
      "epoch": 13.035634743875278,
      "loss": 0.37145259976387024,
      "loss_ce": 0.00011470088065834716,
      "loss_iou": 0.1689453125,
      "loss_num": 0.00653076171875,
      "loss_xval": 0.37109375,
      "num_input_tokens_seen": 327899700,
      "step": 5853
    },
    {
      "epoch": 13.037861915367483,
      "grad_norm": 18.117429733276367,
      "learning_rate": 1e-06,
      "loss": 0.5186,
      "num_input_tokens_seen": 327957944,
      "step": 5854
    },
    {
      "epoch": 13.037861915367483,
      "loss": 0.3974757194519043,
      "loss_ce": 0.00013685536396224052,
      "loss_iou": 0.1767578125,
      "loss_num": 0.00897216796875,
      "loss_xval": 0.396484375,
      "num_input_tokens_seen": 327957944,
      "step": 5854
    },
    {
      "epoch": 13.040089086859687,
      "grad_norm": 20.6132869720459,
      "learning_rate": 1e-06,
      "loss": 0.3864,
      "num_input_tokens_seen": 328014804,
      "step": 5855
    },
    {
      "epoch": 13.040089086859687,
      "loss": 0.411630779504776,
      "loss_ce": 0.00013175193453207612,
      "loss_iou": 0.189453125,
      "loss_num": 0.00665283203125,
      "loss_xval": 0.412109375,
      "num_input_tokens_seen": 328014804,
      "step": 5855
    },
    {
      "epoch": 13.042316258351892,
      "grad_norm": 15.854263305664062,
      "learning_rate": 1e-06,
      "loss": 0.5165,
      "num_input_tokens_seen": 328072992,
      "step": 5856
    },
    {
      "epoch": 13.042316258351892,
      "loss": 0.5808364748954773,
      "loss_ce": 0.00014802644727751613,
      "loss_iou": 0.24609375,
      "loss_num": 0.017578125,
      "loss_xval": 0.58203125,
      "num_input_tokens_seen": 328072992,
      "step": 5856
    },
    {
      "epoch": 13.044543429844097,
      "grad_norm": 14.205804824829102,
      "learning_rate": 1e-06,
      "loss": 0.4901,
      "num_input_tokens_seen": 328130500,
      "step": 5857
    },
    {
      "epoch": 13.044543429844097,
      "loss": 0.5825508832931519,
      "loss_ce": 0.00015347708540502936,
      "loss_iou": 0.2578125,
      "loss_num": 0.012939453125,
      "loss_xval": 0.58203125,
      "num_input_tokens_seen": 328130500,
      "step": 5857
    },
    {
      "epoch": 13.046770601336302,
      "grad_norm": 18.12480354309082,
      "learning_rate": 1e-06,
      "loss": 0.4782,
      "num_input_tokens_seen": 328185140,
      "step": 5858
    },
    {
      "epoch": 13.046770601336302,
      "loss": 0.5846756100654602,
      "loss_ce": 0.0002029853785643354,
      "loss_iou": 0.2470703125,
      "loss_num": 0.0179443359375,
      "loss_xval": 0.5859375,
      "num_input_tokens_seen": 328185140,
      "step": 5858
    },
    {
      "epoch": 13.048997772828507,
      "grad_norm": 18.047836303710938,
      "learning_rate": 1e-06,
      "loss": 0.4255,
      "num_input_tokens_seen": 328239304,
      "step": 5859
    },
    {
      "epoch": 13.048997772828507,
      "loss": 0.5025643110275269,
      "loss_ce": 0.00012291455641388893,
      "loss_iou": 0.224609375,
      "loss_num": 0.01068115234375,
      "loss_xval": 0.50390625,
      "num_input_tokens_seen": 328239304,
      "step": 5859
    },
    {
      "epoch": 13.051224944320714,
      "grad_norm": 19.500625610351562,
      "learning_rate": 1e-06,
      "loss": 0.5162,
      "num_input_tokens_seen": 328295948,
      "step": 5860
    },
    {
      "epoch": 13.051224944320714,
      "loss": 0.47790929675102234,
      "loss_ce": 0.0001260775316040963,
      "loss_iou": 0.212890625,
      "loss_num": 0.01031494140625,
      "loss_xval": 0.478515625,
      "num_input_tokens_seen": 328295948,
      "step": 5860
    },
    {
      "epoch": 13.053452115812918,
      "grad_norm": 30.18915557861328,
      "learning_rate": 1e-06,
      "loss": 0.7097,
      "num_input_tokens_seen": 328348972,
      "step": 5861
    },
    {
      "epoch": 13.053452115812918,
      "loss": 0.7575015425682068,
      "loss_ce": 0.0001773214025888592,
      "loss_iou": 0.330078125,
      "loss_num": 0.0194091796875,
      "loss_xval": 0.7578125,
      "num_input_tokens_seen": 328348972,
      "step": 5861
    },
    {
      "epoch": 13.055679287305123,
      "grad_norm": 15.668800354003906,
      "learning_rate": 1e-06,
      "loss": 0.3318,
      "num_input_tokens_seen": 328403404,
      "step": 5862
    },
    {
      "epoch": 13.055679287305123,
      "loss": 0.3245905935764313,
      "loss_ce": 0.0001277004776056856,
      "loss_iou": 0.138671875,
      "loss_num": 0.0093994140625,
      "loss_xval": 0.32421875,
      "num_input_tokens_seen": 328403404,
      "step": 5862
    },
    {
      "epoch": 13.057906458797328,
      "grad_norm": 15.234992027282715,
      "learning_rate": 1e-06,
      "loss": 0.4268,
      "num_input_tokens_seen": 328459968,
      "step": 5863
    },
    {
      "epoch": 13.057906458797328,
      "loss": 0.5279508829116821,
      "loss_ce": 0.00011885567801073194,
      "loss_iou": 0.2265625,
      "loss_num": 0.01483154296875,
      "loss_xval": 0.52734375,
      "num_input_tokens_seen": 328459968,
      "step": 5863
    },
    {
      "epoch": 13.060133630289533,
      "grad_norm": 19.412704467773438,
      "learning_rate": 1e-06,
      "loss": 0.5039,
      "num_input_tokens_seen": 328516632,
      "step": 5864
    },
    {
      "epoch": 13.060133630289533,
      "loss": 0.6012612581253052,
      "loss_ce": 0.0001870518026407808,
      "loss_iou": 0.2333984375,
      "loss_num": 0.0267333984375,
      "loss_xval": 0.6015625,
      "num_input_tokens_seen": 328516632,
      "step": 5864
    },
    {
      "epoch": 13.062360801781738,
      "grad_norm": 14.758916854858398,
      "learning_rate": 1e-06,
      "loss": 0.4313,
      "num_input_tokens_seen": 328571604,
      "step": 5865
    },
    {
      "epoch": 13.062360801781738,
      "loss": 0.37966063618659973,
      "loss_ce": 0.00014401419321075082,
      "loss_iou": 0.1708984375,
      "loss_num": 0.0074462890625,
      "loss_xval": 0.37890625,
      "num_input_tokens_seen": 328571604,
      "step": 5865
    },
    {
      "epoch": 13.064587973273943,
      "grad_norm": 35.47437286376953,
      "learning_rate": 1e-06,
      "loss": 0.5209,
      "num_input_tokens_seen": 328627528,
      "step": 5866
    },
    {
      "epoch": 13.064587973273943,
      "loss": 0.5552887916564941,
      "loss_ce": 0.00011297700984869152,
      "loss_iou": 0.2119140625,
      "loss_num": 0.0264892578125,
      "loss_xval": 0.5546875,
      "num_input_tokens_seen": 328627528,
      "step": 5866
    },
    {
      "epoch": 13.066815144766148,
      "grad_norm": 24.737062454223633,
      "learning_rate": 1e-06,
      "loss": 0.3764,
      "num_input_tokens_seen": 328684780,
      "step": 5867
    },
    {
      "epoch": 13.066815144766148,
      "loss": 0.33116066455841064,
      "loss_ce": 0.00010598442167975008,
      "loss_iou": 0.1572265625,
      "loss_num": 0.0034942626953125,
      "loss_xval": 0.33203125,
      "num_input_tokens_seen": 328684780,
      "step": 5867
    },
    {
      "epoch": 13.069042316258352,
      "grad_norm": 32.23481750488281,
      "learning_rate": 1e-06,
      "loss": 0.5629,
      "num_input_tokens_seen": 328738920,
      "step": 5868
    },
    {
      "epoch": 13.069042316258352,
      "loss": 0.42393577098846436,
      "loss_ce": 0.0001076542102964595,
      "loss_iou": 0.193359375,
      "loss_num": 0.007232666015625,
      "loss_xval": 0.423828125,
      "num_input_tokens_seen": 328738920,
      "step": 5868
    },
    {
      "epoch": 13.071269487750557,
      "grad_norm": 15.290202140808105,
      "learning_rate": 1e-06,
      "loss": 0.3362,
      "num_input_tokens_seen": 328796456,
      "step": 5869
    },
    {
      "epoch": 13.071269487750557,
      "loss": 0.33213841915130615,
      "loss_ce": 0.00022925705707166344,
      "loss_iou": 0.125,
      "loss_num": 0.0162353515625,
      "loss_xval": 0.33203125,
      "num_input_tokens_seen": 328796456,
      "step": 5869
    },
    {
      "epoch": 13.073496659242762,
      "grad_norm": 20.73664093017578,
      "learning_rate": 1e-06,
      "loss": 0.5255,
      "num_input_tokens_seen": 328851748,
      "step": 5870
    },
    {
      "epoch": 13.073496659242762,
      "loss": 0.41393500566482544,
      "loss_ce": 0.0001166543661383912,
      "loss_iou": 0.1796875,
      "loss_num": 0.0108642578125,
      "loss_xval": 0.4140625,
      "num_input_tokens_seen": 328851748,
      "step": 5870
    },
    {
      "epoch": 13.075723830734967,
      "grad_norm": 19.62030601501465,
      "learning_rate": 1e-06,
      "loss": 0.4958,
      "num_input_tokens_seen": 328907888,
      "step": 5871
    },
    {
      "epoch": 13.075723830734967,
      "loss": 0.5278200507164001,
      "loss_ce": 0.00011009951413143426,
      "loss_iou": 0.228515625,
      "loss_num": 0.0142822265625,
      "loss_xval": 0.52734375,
      "num_input_tokens_seen": 328907888,
      "step": 5871
    },
    {
      "epoch": 13.077951002227172,
      "grad_norm": 24.643142700195312,
      "learning_rate": 1e-06,
      "loss": 0.5101,
      "num_input_tokens_seen": 328964040,
      "step": 5872
    },
    {
      "epoch": 13.077951002227172,
      "loss": 0.5145541429519653,
      "loss_ce": 0.0001498450292274356,
      "loss_iou": 0.2294921875,
      "loss_num": 0.0111083984375,
      "loss_xval": 0.515625,
      "num_input_tokens_seen": 328964040,
      "step": 5872
    },
    {
      "epoch": 13.080178173719377,
      "grad_norm": 18.29912567138672,
      "learning_rate": 1e-06,
      "loss": 0.4219,
      "num_input_tokens_seen": 329019624,
      "step": 5873
    },
    {
      "epoch": 13.080178173719377,
      "loss": 0.4937739372253418,
      "loss_ce": 0.00012156983575550839,
      "loss_iou": 0.21484375,
      "loss_num": 0.01263427734375,
      "loss_xval": 0.494140625,
      "num_input_tokens_seen": 329019624,
      "step": 5873
    },
    {
      "epoch": 13.082405345211582,
      "grad_norm": 16.88625717163086,
      "learning_rate": 1e-06,
      "loss": 0.4574,
      "num_input_tokens_seen": 329077484,
      "step": 5874
    },
    {
      "epoch": 13.082405345211582,
      "loss": 0.3620465397834778,
      "loss_ce": 0.00010803519398905337,
      "loss_iou": 0.1611328125,
      "loss_num": 0.008056640625,
      "loss_xval": 0.361328125,
      "num_input_tokens_seen": 329077484,
      "step": 5874
    },
    {
      "epoch": 13.084632516703786,
      "grad_norm": 13.229454040527344,
      "learning_rate": 1e-06,
      "loss": 0.5683,
      "num_input_tokens_seen": 329135212,
      "step": 5875
    },
    {
      "epoch": 13.084632516703786,
      "loss": 0.6875127553939819,
      "loss_ce": 0.0001348142686765641,
      "loss_iou": 0.271484375,
      "loss_num": 0.02880859375,
      "loss_xval": 0.6875,
      "num_input_tokens_seen": 329135212,
      "step": 5875
    },
    {
      "epoch": 13.086859688195991,
      "grad_norm": 30.31863021850586,
      "learning_rate": 1e-06,
      "loss": 0.5017,
      "num_input_tokens_seen": 329190660,
      "step": 5876
    },
    {
      "epoch": 13.086859688195991,
      "loss": 0.5497363805770874,
      "loss_ce": 0.0001758422004058957,
      "loss_iou": 0.2353515625,
      "loss_num": 0.0157470703125,
      "loss_xval": 0.55078125,
      "num_input_tokens_seen": 329190660,
      "step": 5876
    },
    {
      "epoch": 13.089086859688196,
      "grad_norm": 16.870037078857422,
      "learning_rate": 1e-06,
      "loss": 0.3866,
      "num_input_tokens_seen": 329247060,
      "step": 5877
    },
    {
      "epoch": 13.089086859688196,
      "loss": 0.36009788513183594,
      "loss_ce": 0.00011254477431066334,
      "loss_iou": 0.162109375,
      "loss_num": 0.007293701171875,
      "loss_xval": 0.359375,
      "num_input_tokens_seen": 329247060,
      "step": 5877
    },
    {
      "epoch": 13.091314031180401,
      "grad_norm": 21.952919006347656,
      "learning_rate": 1e-06,
      "loss": 0.4696,
      "num_input_tokens_seen": 329301984,
      "step": 5878
    },
    {
      "epoch": 13.091314031180401,
      "loss": 0.4436010420322418,
      "loss_ce": 0.00011958822869928554,
      "loss_iou": 0.2001953125,
      "loss_num": 0.0087890625,
      "loss_xval": 0.443359375,
      "num_input_tokens_seen": 329301984,
      "step": 5878
    },
    {
      "epoch": 13.093541202672606,
      "grad_norm": 18.792381286621094,
      "learning_rate": 1e-06,
      "loss": 0.4403,
      "num_input_tokens_seen": 329360572,
      "step": 5879
    },
    {
      "epoch": 13.093541202672606,
      "loss": 0.5003446340560913,
      "loss_ce": 0.00010050551645690575,
      "loss_iou": 0.21484375,
      "loss_num": 0.01397705078125,
      "loss_xval": 0.5,
      "num_input_tokens_seen": 329360572,
      "step": 5879
    },
    {
      "epoch": 13.09576837416481,
      "grad_norm": 18.393442153930664,
      "learning_rate": 1e-06,
      "loss": 0.6429,
      "num_input_tokens_seen": 329417288,
      "step": 5880
    },
    {
      "epoch": 13.09576837416481,
      "loss": 0.7582250833511353,
      "loss_ce": 0.00016841033357195556,
      "loss_iou": 0.314453125,
      "loss_num": 0.0257568359375,
      "loss_xval": 0.7578125,
      "num_input_tokens_seen": 329417288,
      "step": 5880
    },
    {
      "epoch": 13.097995545657016,
      "grad_norm": 13.813261985778809,
      "learning_rate": 1e-06,
      "loss": 0.4882,
      "num_input_tokens_seen": 329473344,
      "step": 5881
    },
    {
      "epoch": 13.097995545657016,
      "loss": 0.6629891395568848,
      "loss_ce": 0.0001473267620895058,
      "loss_iou": 0.271484375,
      "loss_num": 0.0235595703125,
      "loss_xval": 0.6640625,
      "num_input_tokens_seen": 329473344,
      "step": 5881
    },
    {
      "epoch": 13.10022271714922,
      "grad_norm": 15.445968627929688,
      "learning_rate": 1e-06,
      "loss": 0.4329,
      "num_input_tokens_seen": 329531204,
      "step": 5882
    },
    {
      "epoch": 13.10022271714922,
      "loss": 0.43495747447013855,
      "loss_ce": 0.00014300504699349403,
      "loss_iou": 0.197265625,
      "loss_num": 0.008056640625,
      "loss_xval": 0.435546875,
      "num_input_tokens_seen": 329531204,
      "step": 5882
    },
    {
      "epoch": 13.102449888641425,
      "grad_norm": 14.054932594299316,
      "learning_rate": 1e-06,
      "loss": 0.4633,
      "num_input_tokens_seen": 329586744,
      "step": 5883
    },
    {
      "epoch": 13.102449888641425,
      "loss": 0.5273807644844055,
      "loss_ce": 0.0005252888076938689,
      "loss_iou": 0.2353515625,
      "loss_num": 0.01129150390625,
      "loss_xval": 0.52734375,
      "num_input_tokens_seen": 329586744,
      "step": 5883
    },
    {
      "epoch": 13.10467706013363,
      "grad_norm": 16.611223220825195,
      "learning_rate": 1e-06,
      "loss": 0.397,
      "num_input_tokens_seen": 329642216,
      "step": 5884
    },
    {
      "epoch": 13.10467706013363,
      "loss": 0.4468822479248047,
      "loss_ce": 0.00010488741099834442,
      "loss_iou": 0.19921875,
      "loss_num": 0.00946044921875,
      "loss_xval": 0.447265625,
      "num_input_tokens_seen": 329642216,
      "step": 5884
    },
    {
      "epoch": 13.106904231625835,
      "grad_norm": 18.158891677856445,
      "learning_rate": 1e-06,
      "loss": 0.324,
      "num_input_tokens_seen": 329698692,
      "step": 5885
    },
    {
      "epoch": 13.106904231625835,
      "loss": 0.3763335049152374,
      "loss_ce": 0.00011280621401965618,
      "loss_iou": 0.1630859375,
      "loss_num": 0.0101318359375,
      "loss_xval": 0.376953125,
      "num_input_tokens_seen": 329698692,
      "step": 5885
    },
    {
      "epoch": 13.10913140311804,
      "grad_norm": 20.2602481842041,
      "learning_rate": 1e-06,
      "loss": 0.4628,
      "num_input_tokens_seen": 329753408,
      "step": 5886
    },
    {
      "epoch": 13.10913140311804,
      "loss": 0.590545117855072,
      "loss_ce": 0.0005792796146124601,
      "loss_iou": 0.232421875,
      "loss_num": 0.0247802734375,
      "loss_xval": 0.58984375,
      "num_input_tokens_seen": 329753408,
      "step": 5886
    },
    {
      "epoch": 13.111358574610245,
      "grad_norm": 17.374492645263672,
      "learning_rate": 1e-06,
      "loss": 0.5085,
      "num_input_tokens_seen": 329811252,
      "step": 5887
    },
    {
      "epoch": 13.111358574610245,
      "loss": 0.6364624500274658,
      "loss_ce": 0.00010987733548972756,
      "loss_iou": 0.28125,
      "loss_num": 0.0150146484375,
      "loss_xval": 0.63671875,
      "num_input_tokens_seen": 329811252,
      "step": 5887
    },
    {
      "epoch": 13.11358574610245,
      "grad_norm": 24.767221450805664,
      "learning_rate": 1e-06,
      "loss": 0.6877,
      "num_input_tokens_seen": 329864808,
      "step": 5888
    },
    {
      "epoch": 13.11358574610245,
      "loss": 0.9313529133796692,
      "loss_ce": 0.0002005502174142748,
      "loss_iou": 0.384765625,
      "loss_num": 0.0322265625,
      "loss_xval": 0.9296875,
      "num_input_tokens_seen": 329864808,
      "step": 5888
    },
    {
      "epoch": 13.115812917594655,
      "grad_norm": 18.444570541381836,
      "learning_rate": 1e-06,
      "loss": 0.3468,
      "num_input_tokens_seen": 329923220,
      "step": 5889
    },
    {
      "epoch": 13.115812917594655,
      "loss": 0.38743722438812256,
      "loss_ce": 0.00010810734238475561,
      "loss_iou": 0.1806640625,
      "loss_num": 0.0052490234375,
      "loss_xval": 0.38671875,
      "num_input_tokens_seen": 329923220,
      "step": 5889
    },
    {
      "epoch": 13.11804008908686,
      "grad_norm": 14.813982009887695,
      "learning_rate": 1e-06,
      "loss": 0.5645,
      "num_input_tokens_seen": 329982272,
      "step": 5890
    },
    {
      "epoch": 13.11804008908686,
      "loss": 0.7087510824203491,
      "loss_ce": 0.00013293002848513424,
      "loss_iou": 0.296875,
      "loss_num": 0.02294921875,
      "loss_xval": 0.70703125,
      "num_input_tokens_seen": 329982272,
      "step": 5890
    },
    {
      "epoch": 13.120267260579064,
      "grad_norm": 17.98593521118164,
      "learning_rate": 1e-06,
      "loss": 0.6082,
      "num_input_tokens_seen": 330037796,
      "step": 5891
    },
    {
      "epoch": 13.120267260579064,
      "loss": 0.6543379426002502,
      "loss_ce": 0.00031572196166962385,
      "loss_iou": 0.265625,
      "loss_num": 0.0247802734375,
      "loss_xval": 0.65234375,
      "num_input_tokens_seen": 330037796,
      "step": 5891
    },
    {
      "epoch": 13.122494432071269,
      "grad_norm": 16.788930892944336,
      "learning_rate": 1e-06,
      "loss": 0.4808,
      "num_input_tokens_seen": 330092396,
      "step": 5892
    },
    {
      "epoch": 13.122494432071269,
      "loss": 0.5958355665206909,
      "loss_ce": 0.00013241068518254906,
      "loss_iou": 0.23828125,
      "loss_num": 0.0238037109375,
      "loss_xval": 0.59375,
      "num_input_tokens_seen": 330092396,
      "step": 5892
    },
    {
      "epoch": 13.124721603563474,
      "grad_norm": 17.42109489440918,
      "learning_rate": 1e-06,
      "loss": 0.405,
      "num_input_tokens_seen": 330147804,
      "step": 5893
    },
    {
      "epoch": 13.124721603563474,
      "loss": 0.4320144057273865,
      "loss_ce": 0.000129641528474167,
      "loss_iou": 0.193359375,
      "loss_num": 0.00933837890625,
      "loss_xval": 0.431640625,
      "num_input_tokens_seen": 330147804,
      "step": 5893
    },
    {
      "epoch": 13.126948775055679,
      "grad_norm": 17.75126838684082,
      "learning_rate": 1e-06,
      "loss": 0.5447,
      "num_input_tokens_seen": 330205316,
      "step": 5894
    },
    {
      "epoch": 13.126948775055679,
      "loss": 0.6002463698387146,
      "loss_ce": 0.00014872349856887013,
      "loss_iou": 0.2470703125,
      "loss_num": 0.0213623046875,
      "loss_xval": 0.6015625,
      "num_input_tokens_seen": 330205316,
      "step": 5894
    },
    {
      "epoch": 13.129175946547884,
      "grad_norm": 22.149547576904297,
      "learning_rate": 1e-06,
      "loss": 0.4957,
      "num_input_tokens_seen": 330261848,
      "step": 5895
    },
    {
      "epoch": 13.129175946547884,
      "loss": 0.6235384941101074,
      "loss_ce": 0.00012537876318674535,
      "loss_iou": 0.275390625,
      "loss_num": 0.0146484375,
      "loss_xval": 0.625,
      "num_input_tokens_seen": 330261848,
      "step": 5895
    },
    {
      "epoch": 13.131403118040089,
      "grad_norm": 18.989896774291992,
      "learning_rate": 1e-06,
      "loss": 0.5355,
      "num_input_tokens_seen": 330318236,
      "step": 5896
    },
    {
      "epoch": 13.131403118040089,
      "loss": 0.4923059344291687,
      "loss_ce": 0.00011842740059364587,
      "loss_iou": 0.208984375,
      "loss_num": 0.01495361328125,
      "loss_xval": 0.4921875,
      "num_input_tokens_seen": 330318236,
      "step": 5896
    },
    {
      "epoch": 13.133630289532293,
      "grad_norm": 15.263077735900879,
      "learning_rate": 1e-06,
      "loss": 0.486,
      "num_input_tokens_seen": 330377552,
      "step": 5897
    },
    {
      "epoch": 13.133630289532293,
      "loss": 0.4600605368614197,
      "loss_ce": 9.961408795788884e-05,
      "loss_iou": 0.1796875,
      "loss_num": 0.0201416015625,
      "loss_xval": 0.4609375,
      "num_input_tokens_seen": 330377552,
      "step": 5897
    },
    {
      "epoch": 13.135857461024498,
      "grad_norm": 17.812576293945312,
      "learning_rate": 1e-06,
      "loss": 0.5491,
      "num_input_tokens_seen": 330433484,
      "step": 5898
    },
    {
      "epoch": 13.135857461024498,
      "loss": 0.5521707534790039,
      "loss_ce": 0.00010780214506667107,
      "loss_iou": 0.234375,
      "loss_num": 0.0164794921875,
      "loss_xval": 0.55078125,
      "num_input_tokens_seen": 330433484,
      "step": 5898
    },
    {
      "epoch": 13.138084632516703,
      "grad_norm": 28.198577880859375,
      "learning_rate": 1e-06,
      "loss": 0.6077,
      "num_input_tokens_seen": 330489352,
      "step": 5899
    },
    {
      "epoch": 13.138084632516703,
      "loss": 0.6835901737213135,
      "loss_ce": 0.0001184626089525409,
      "loss_iou": 0.2890625,
      "loss_num": 0.0205078125,
      "loss_xval": 0.68359375,
      "num_input_tokens_seen": 330489352,
      "step": 5899
    },
    {
      "epoch": 13.140311804008908,
      "grad_norm": 16.04581069946289,
      "learning_rate": 1e-06,
      "loss": 0.5163,
      "num_input_tokens_seen": 330544888,
      "step": 5900
    },
    {
      "epoch": 13.140311804008908,
      "loss": 0.5014405846595764,
      "loss_ce": 0.00012069322110619396,
      "loss_iou": 0.21875,
      "loss_num": 0.01275634765625,
      "loss_xval": 0.5,
      "num_input_tokens_seen": 330544888,
      "step": 5900
    },
    {
      "epoch": 13.142538975501113,
      "grad_norm": 18.238014221191406,
      "learning_rate": 1e-06,
      "loss": 0.3431,
      "num_input_tokens_seen": 330602552,
      "step": 5901
    },
    {
      "epoch": 13.142538975501113,
      "loss": 0.3762207627296448,
      "loss_ce": 0.0001221080165123567,
      "loss_iou": 0.1708984375,
      "loss_num": 0.006927490234375,
      "loss_xval": 0.376953125,
      "num_input_tokens_seen": 330602552,
      "step": 5901
    },
    {
      "epoch": 13.144766146993318,
      "grad_norm": 32.46430206298828,
      "learning_rate": 1e-06,
      "loss": 0.5533,
      "num_input_tokens_seen": 330657020,
      "step": 5902
    },
    {
      "epoch": 13.144766146993318,
      "loss": 0.6620012521743774,
      "loss_ce": 0.00013605058484245092,
      "loss_iou": 0.27734375,
      "loss_num": 0.0216064453125,
      "loss_xval": 0.66015625,
      "num_input_tokens_seen": 330657020,
      "step": 5902
    },
    {
      "epoch": 13.146993318485523,
      "grad_norm": 30.36663246154785,
      "learning_rate": 1e-06,
      "loss": 0.4032,
      "num_input_tokens_seen": 330713536,
      "step": 5903
    },
    {
      "epoch": 13.146993318485523,
      "loss": 0.36765336990356445,
      "loss_ce": 0.0005879499949514866,
      "loss_iou": 0.146484375,
      "loss_num": 0.0147705078125,
      "loss_xval": 0.3671875,
      "num_input_tokens_seen": 330713536,
      "step": 5903
    },
    {
      "epoch": 13.14922048997773,
      "grad_norm": 15.999008178710938,
      "learning_rate": 1e-06,
      "loss": 0.4386,
      "num_input_tokens_seen": 330767572,
      "step": 5904
    },
    {
      "epoch": 13.14922048997773,
      "loss": 0.5433182716369629,
      "loss_ce": 0.00010540579387452453,
      "loss_iou": 0.2265625,
      "loss_num": 0.0181884765625,
      "loss_xval": 0.54296875,
      "num_input_tokens_seen": 330767572,
      "step": 5904
    },
    {
      "epoch": 13.151447661469934,
      "grad_norm": 17.65342903137207,
      "learning_rate": 1e-06,
      "loss": 0.3948,
      "num_input_tokens_seen": 330825640,
      "step": 5905
    },
    {
      "epoch": 13.151447661469934,
      "loss": 0.3161456882953644,
      "loss_ce": 0.000105641498521436,
      "loss_iou": 0.1396484375,
      "loss_num": 0.0074462890625,
      "loss_xval": 0.31640625,
      "num_input_tokens_seen": 330825640,
      "step": 5905
    },
    {
      "epoch": 13.153674832962139,
      "grad_norm": 31.29542350769043,
      "learning_rate": 1e-06,
      "loss": 0.7188,
      "num_input_tokens_seen": 330884276,
      "step": 5906
    },
    {
      "epoch": 13.153674832962139,
      "loss": 0.7189792990684509,
      "loss_ce": 0.00010728358756750822,
      "loss_iou": 0.31640625,
      "loss_num": 0.0174560546875,
      "loss_xval": 0.71875,
      "num_input_tokens_seen": 330884276,
      "step": 5906
    },
    {
      "epoch": 13.155902004454344,
      "grad_norm": 18.346426010131836,
      "learning_rate": 1e-06,
      "loss": 0.5431,
      "num_input_tokens_seen": 330938632,
      "step": 5907
    },
    {
      "epoch": 13.155902004454344,
      "loss": 0.4295383095741272,
      "loss_ce": 9.494097321294248e-05,
      "loss_iou": 0.169921875,
      "loss_num": 0.0179443359375,
      "loss_xval": 0.4296875,
      "num_input_tokens_seen": 330938632,
      "step": 5907
    },
    {
      "epoch": 13.158129175946549,
      "grad_norm": 22.875947952270508,
      "learning_rate": 1e-06,
      "loss": 0.6056,
      "num_input_tokens_seen": 330995628,
      "step": 5908
    },
    {
      "epoch": 13.158129175946549,
      "loss": 0.5698830485343933,
      "loss_ce": 0.00011983538570348173,
      "loss_iou": 0.23046875,
      "loss_num": 0.021728515625,
      "loss_xval": 0.5703125,
      "num_input_tokens_seen": 330995628,
      "step": 5908
    },
    {
      "epoch": 13.160356347438753,
      "grad_norm": 15.61497974395752,
      "learning_rate": 1e-06,
      "loss": 0.4997,
      "num_input_tokens_seen": 331049196,
      "step": 5909
    },
    {
      "epoch": 13.160356347438753,
      "loss": 0.37679868936538696,
      "loss_ce": 8.96861165529117e-05,
      "loss_iou": 0.1513671875,
      "loss_num": 0.0147705078125,
      "loss_xval": 0.376953125,
      "num_input_tokens_seen": 331049196,
      "step": 5909
    },
    {
      "epoch": 13.162583518930958,
      "grad_norm": 15.534133911132812,
      "learning_rate": 1e-06,
      "loss": 0.6097,
      "num_input_tokens_seen": 331104396,
      "step": 5910
    },
    {
      "epoch": 13.162583518930958,
      "loss": 0.8099589347839355,
      "loss_ce": 0.0001444965455448255,
      "loss_iou": 0.33984375,
      "loss_num": 0.0262451171875,
      "loss_xval": 0.80859375,
      "num_input_tokens_seen": 331104396,
      "step": 5910
    },
    {
      "epoch": 13.164810690423163,
      "grad_norm": 41.32310104370117,
      "learning_rate": 1e-06,
      "loss": 0.581,
      "num_input_tokens_seen": 331159788,
      "step": 5911
    },
    {
      "epoch": 13.164810690423163,
      "loss": 0.4186583161354065,
      "loss_ce": 0.00014023938274476677,
      "loss_iou": 0.181640625,
      "loss_num": 0.010986328125,
      "loss_xval": 0.41796875,
      "num_input_tokens_seen": 331159788,
      "step": 5911
    },
    {
      "epoch": 13.167037861915368,
      "grad_norm": 17.732685089111328,
      "learning_rate": 1e-06,
      "loss": 0.5119,
      "num_input_tokens_seen": 331216032,
      "step": 5912
    },
    {
      "epoch": 13.167037861915368,
      "loss": 0.588236391544342,
      "loss_ce": 0.00010163354454562068,
      "loss_iou": 0.2578125,
      "loss_num": 0.014892578125,
      "loss_xval": 0.58984375,
      "num_input_tokens_seen": 331216032,
      "step": 5912
    },
    {
      "epoch": 13.169265033407573,
      "grad_norm": 20.534914016723633,
      "learning_rate": 1e-06,
      "loss": 0.3891,
      "num_input_tokens_seen": 331273480,
      "step": 5913
    },
    {
      "epoch": 13.169265033407573,
      "loss": 0.3801114559173584,
      "loss_ce": 0.00010655220103217289,
      "loss_iou": 0.169921875,
      "loss_num": 0.0081787109375,
      "loss_xval": 0.380859375,
      "num_input_tokens_seen": 331273480,
      "step": 5913
    },
    {
      "epoch": 13.171492204899778,
      "grad_norm": 45.9957389831543,
      "learning_rate": 1e-06,
      "loss": 0.5419,
      "num_input_tokens_seen": 331328248,
      "step": 5914
    },
    {
      "epoch": 13.171492204899778,
      "loss": 0.45751655101776123,
      "loss_ce": 0.00011909787281183526,
      "loss_iou": 0.208984375,
      "loss_num": 0.0076904296875,
      "loss_xval": 0.45703125,
      "num_input_tokens_seen": 331328248,
      "step": 5914
    },
    {
      "epoch": 13.173719376391983,
      "grad_norm": 21.617551803588867,
      "learning_rate": 1e-06,
      "loss": 0.4976,
      "num_input_tokens_seen": 331383188,
      "step": 5915
    },
    {
      "epoch": 13.173719376391983,
      "loss": 0.5962991118431091,
      "loss_ce": 0.0010842570336535573,
      "loss_iou": 0.26171875,
      "loss_num": 0.01397705078125,
      "loss_xval": 0.59375,
      "num_input_tokens_seen": 331383188,
      "step": 5915
    },
    {
      "epoch": 13.175946547884188,
      "grad_norm": 18.615833282470703,
      "learning_rate": 1e-06,
      "loss": 0.442,
      "num_input_tokens_seen": 331439448,
      "step": 5916
    },
    {
      "epoch": 13.175946547884188,
      "loss": 0.6015082001686096,
      "loss_ce": 0.00022037216695025563,
      "loss_iou": 0.236328125,
      "loss_num": 0.02587890625,
      "loss_xval": 0.6015625,
      "num_input_tokens_seen": 331439448,
      "step": 5916
    },
    {
      "epoch": 13.178173719376392,
      "grad_norm": 26.171741485595703,
      "learning_rate": 1e-06,
      "loss": 0.5615,
      "num_input_tokens_seen": 331492752,
      "step": 5917
    },
    {
      "epoch": 13.178173719376392,
      "loss": 0.47891539335250854,
      "loss_ce": 0.00015565170906484127,
      "loss_iou": 0.224609375,
      "loss_num": 0.006011962890625,
      "loss_xval": 0.478515625,
      "num_input_tokens_seen": 331492752,
      "step": 5917
    },
    {
      "epoch": 13.180400890868597,
      "grad_norm": 29.380762100219727,
      "learning_rate": 1e-06,
      "loss": 0.4608,
      "num_input_tokens_seen": 331548328,
      "step": 5918
    },
    {
      "epoch": 13.180400890868597,
      "loss": 0.37973326444625854,
      "loss_ce": 9.459797001909465e-05,
      "loss_iou": 0.154296875,
      "loss_num": 0.01409912109375,
      "loss_xval": 0.37890625,
      "num_input_tokens_seen": 331548328,
      "step": 5918
    },
    {
      "epoch": 13.182628062360802,
      "grad_norm": 19.07701301574707,
      "learning_rate": 1e-06,
      "loss": 0.5202,
      "num_input_tokens_seen": 331603848,
      "step": 5919
    },
    {
      "epoch": 13.182628062360802,
      "loss": 0.5613878965377808,
      "loss_ce": 0.00010859225585591048,
      "loss_iou": 0.24609375,
      "loss_num": 0.013671875,
      "loss_xval": 0.5625,
      "num_input_tokens_seen": 331603848,
      "step": 5919
    },
    {
      "epoch": 13.184855233853007,
      "grad_norm": 23.228954315185547,
      "learning_rate": 1e-06,
      "loss": 0.4674,
      "num_input_tokens_seen": 331663044,
      "step": 5920
    },
    {
      "epoch": 13.184855233853007,
      "loss": 0.6362162828445435,
      "loss_ce": 0.00010789166844915599,
      "loss_iou": 0.27734375,
      "loss_num": 0.0167236328125,
      "loss_xval": 0.63671875,
      "num_input_tokens_seen": 331663044,
      "step": 5920
    },
    {
      "epoch": 13.187082405345212,
      "grad_norm": 24.631132125854492,
      "learning_rate": 1e-06,
      "loss": 0.7198,
      "num_input_tokens_seen": 331717048,
      "step": 5921
    },
    {
      "epoch": 13.187082405345212,
      "loss": 0.5893706679344177,
      "loss_ce": 0.00013725095777772367,
      "loss_iou": 0.2578125,
      "loss_num": 0.01470947265625,
      "loss_xval": 0.58984375,
      "num_input_tokens_seen": 331717048,
      "step": 5921
    },
    {
      "epoch": 13.189309576837417,
      "grad_norm": 27.25778579711914,
      "learning_rate": 1e-06,
      "loss": 0.4561,
      "num_input_tokens_seen": 331770924,
      "step": 5922
    },
    {
      "epoch": 13.189309576837417,
      "loss": 0.3740364909172058,
      "loss_ce": 0.00013512761506717652,
      "loss_iou": 0.1572265625,
      "loss_num": 0.0120849609375,
      "loss_xval": 0.373046875,
      "num_input_tokens_seen": 331770924,
      "step": 5922
    },
    {
      "epoch": 13.191536748329622,
      "grad_norm": 13.549928665161133,
      "learning_rate": 1e-06,
      "loss": 0.5417,
      "num_input_tokens_seen": 331826796,
      "step": 5923
    },
    {
      "epoch": 13.191536748329622,
      "loss": 0.3583794832229614,
      "loss_ce": 0.00010311156802345067,
      "loss_iou": 0.16015625,
      "loss_num": 0.00738525390625,
      "loss_xval": 0.357421875,
      "num_input_tokens_seen": 331826796,
      "step": 5923
    },
    {
      "epoch": 13.193763919821826,
      "grad_norm": 19.371231079101562,
      "learning_rate": 1e-06,
      "loss": 0.4902,
      "num_input_tokens_seen": 331882152,
      "step": 5924
    },
    {
      "epoch": 13.193763919821826,
      "loss": 0.6329140067100525,
      "loss_ce": 0.00010151312744710594,
      "loss_iou": 0.291015625,
      "loss_num": 0.01019287109375,
      "loss_xval": 0.6328125,
      "num_input_tokens_seen": 331882152,
      "step": 5924
    },
    {
      "epoch": 13.195991091314031,
      "grad_norm": 20.2581787109375,
      "learning_rate": 1e-06,
      "loss": 0.4575,
      "num_input_tokens_seen": 331939604,
      "step": 5925
    },
    {
      "epoch": 13.195991091314031,
      "loss": 0.3990633487701416,
      "loss_ce": 0.0001375791907776147,
      "loss_iou": 0.1708984375,
      "loss_num": 0.011474609375,
      "loss_xval": 0.3984375,
      "num_input_tokens_seen": 331939604,
      "step": 5925
    },
    {
      "epoch": 13.198218262806236,
      "grad_norm": 86.76207733154297,
      "learning_rate": 1e-06,
      "loss": 0.5783,
      "num_input_tokens_seen": 331993812,
      "step": 5926
    },
    {
      "epoch": 13.198218262806236,
      "loss": 0.5379533171653748,
      "loss_ce": 0.00011149346391903237,
      "loss_iou": 0.2392578125,
      "loss_num": 0.0118408203125,
      "loss_xval": 0.5390625,
      "num_input_tokens_seen": 331993812,
      "step": 5926
    },
    {
      "epoch": 13.200445434298441,
      "grad_norm": 19.799470901489258,
      "learning_rate": 1e-06,
      "loss": 0.5504,
      "num_input_tokens_seen": 332050244,
      "step": 5927
    },
    {
      "epoch": 13.200445434298441,
      "loss": 0.5327932834625244,
      "loss_ce": 0.00020055694039911032,
      "loss_iou": 0.23828125,
      "loss_num": 0.01129150390625,
      "loss_xval": 0.53125,
      "num_input_tokens_seen": 332050244,
      "step": 5927
    },
    {
      "epoch": 13.202672605790646,
      "grad_norm": 21.825603485107422,
      "learning_rate": 1e-06,
      "loss": 0.5413,
      "num_input_tokens_seen": 332107696,
      "step": 5928
    },
    {
      "epoch": 13.202672605790646,
      "loss": 0.4124048948287964,
      "loss_ce": 0.00011244384222663939,
      "loss_iou": 0.1845703125,
      "loss_num": 0.00885009765625,
      "loss_xval": 0.412109375,
      "num_input_tokens_seen": 332107696,
      "step": 5928
    },
    {
      "epoch": 13.20489977728285,
      "grad_norm": 22.338329315185547,
      "learning_rate": 1e-06,
      "loss": 0.4693,
      "num_input_tokens_seen": 332164076,
      "step": 5929
    },
    {
      "epoch": 13.20489977728285,
      "loss": 0.2473558485507965,
      "loss_ce": 0.00010242719145026058,
      "loss_iou": 0.10595703125,
      "loss_num": 0.007110595703125,
      "loss_xval": 0.2470703125,
      "num_input_tokens_seen": 332164076,
      "step": 5929
    },
    {
      "epoch": 13.207126948775056,
      "grad_norm": 13.758119583129883,
      "learning_rate": 1e-06,
      "loss": 0.3369,
      "num_input_tokens_seen": 332222060,
      "step": 5930
    },
    {
      "epoch": 13.207126948775056,
      "loss": 0.3643418550491333,
      "loss_ce": 0.00032818166073411703,
      "loss_iou": 0.140625,
      "loss_num": 0.0167236328125,
      "loss_xval": 0.36328125,
      "num_input_tokens_seen": 332222060,
      "step": 5930
    },
    {
      "epoch": 13.20935412026726,
      "grad_norm": 20.242292404174805,
      "learning_rate": 1e-06,
      "loss": 0.4062,
      "num_input_tokens_seen": 332280184,
      "step": 5931
    },
    {
      "epoch": 13.20935412026726,
      "loss": 0.3549681603908539,
      "loss_ce": 0.00023182205040939152,
      "loss_iou": 0.16015625,
      "loss_num": 0.007080078125,
      "loss_xval": 0.35546875,
      "num_input_tokens_seen": 332280184,
      "step": 5931
    },
    {
      "epoch": 13.211581291759465,
      "grad_norm": 57.8745002746582,
      "learning_rate": 1e-06,
      "loss": 0.5825,
      "num_input_tokens_seen": 332335648,
      "step": 5932
    },
    {
      "epoch": 13.211581291759465,
      "loss": 0.5452816486358643,
      "loss_ce": 0.00011567381443455815,
      "loss_iou": 0.216796875,
      "loss_num": 0.0224609375,
      "loss_xval": 0.546875,
      "num_input_tokens_seen": 332335648,
      "step": 5932
    },
    {
      "epoch": 13.21380846325167,
      "grad_norm": 23.37744140625,
      "learning_rate": 1e-06,
      "loss": 0.5968,
      "num_input_tokens_seen": 332391416,
      "step": 5933
    },
    {
      "epoch": 13.21380846325167,
      "loss": 0.5802435278892517,
      "loss_ce": 0.00034849648363888264,
      "loss_iou": 0.240234375,
      "loss_num": 0.0198974609375,
      "loss_xval": 0.578125,
      "num_input_tokens_seen": 332391416,
      "step": 5933
    },
    {
      "epoch": 13.216035634743875,
      "grad_norm": 24.591421127319336,
      "learning_rate": 1e-06,
      "loss": 0.4096,
      "num_input_tokens_seen": 332448788,
      "step": 5934
    },
    {
      "epoch": 13.216035634743875,
      "loss": 0.5153904557228088,
      "loss_ce": 0.00013165900600142777,
      "loss_iou": 0.23828125,
      "loss_num": 0.00787353515625,
      "loss_xval": 0.515625,
      "num_input_tokens_seen": 332448788,
      "step": 5934
    },
    {
      "epoch": 13.21826280623608,
      "grad_norm": 23.88517951965332,
      "learning_rate": 1e-06,
      "loss": 0.4557,
      "num_input_tokens_seen": 332505524,
      "step": 5935
    },
    {
      "epoch": 13.21826280623608,
      "loss": 0.4439461827278137,
      "loss_ce": 9.8556381999515e-05,
      "loss_iou": 0.1826171875,
      "loss_num": 0.01556396484375,
      "loss_xval": 0.443359375,
      "num_input_tokens_seen": 332505524,
      "step": 5935
    },
    {
      "epoch": 13.220489977728285,
      "grad_norm": 18.264699935913086,
      "learning_rate": 1e-06,
      "loss": 0.4795,
      "num_input_tokens_seen": 332562732,
      "step": 5936
    },
    {
      "epoch": 13.220489977728285,
      "loss": 0.5457608699798584,
      "loss_ce": 0.00010662610293366015,
      "loss_iou": 0.228515625,
      "loss_num": 0.017822265625,
      "loss_xval": 0.546875,
      "num_input_tokens_seen": 332562732,
      "step": 5936
    },
    {
      "epoch": 13.22271714922049,
      "grad_norm": 20.514163970947266,
      "learning_rate": 1e-06,
      "loss": 0.5394,
      "num_input_tokens_seen": 332617164,
      "step": 5937
    },
    {
      "epoch": 13.22271714922049,
      "loss": 0.5808427333831787,
      "loss_ce": 0.00015426846221089363,
      "loss_iou": 0.25390625,
      "loss_num": 0.0147705078125,
      "loss_xval": 0.58203125,
      "num_input_tokens_seen": 332617164,
      "step": 5937
    },
    {
      "epoch": 13.224944320712694,
      "grad_norm": 19.999271392822266,
      "learning_rate": 1e-06,
      "loss": 0.4511,
      "num_input_tokens_seen": 332672196,
      "step": 5938
    },
    {
      "epoch": 13.224944320712694,
      "loss": 0.5069501996040344,
      "loss_ce": 0.00017530655895825475,
      "loss_iou": 0.208984375,
      "loss_num": 0.0179443359375,
      "loss_xval": 0.5078125,
      "num_input_tokens_seen": 332672196,
      "step": 5938
    },
    {
      "epoch": 13.2271714922049,
      "grad_norm": 19.271041870117188,
      "learning_rate": 1e-06,
      "loss": 0.5313,
      "num_input_tokens_seen": 332729132,
      "step": 5939
    },
    {
      "epoch": 13.2271714922049,
      "loss": 0.5557813048362732,
      "loss_ce": 0.00011723280476871878,
      "loss_iou": 0.255859375,
      "loss_num": 0.00860595703125,
      "loss_xval": 0.5546875,
      "num_input_tokens_seen": 332729132,
      "step": 5939
    },
    {
      "epoch": 13.229398663697104,
      "grad_norm": 22.940664291381836,
      "learning_rate": 1e-06,
      "loss": 0.7251,
      "num_input_tokens_seen": 332783480,
      "step": 5940
    },
    {
      "epoch": 13.229398663697104,
      "loss": 0.8985084891319275,
      "loss_ce": 0.00019303051522001624,
      "loss_iou": 0.330078125,
      "loss_num": 0.0478515625,
      "loss_xval": 0.8984375,
      "num_input_tokens_seen": 332783480,
      "step": 5940
    },
    {
      "epoch": 13.231625835189309,
      "grad_norm": 17.761545181274414,
      "learning_rate": 1e-06,
      "loss": 0.4187,
      "num_input_tokens_seen": 332840084,
      "step": 5941
    },
    {
      "epoch": 13.231625835189309,
      "loss": 0.5540998578071594,
      "loss_ce": 0.00014480168465524912,
      "loss_iou": 0.244140625,
      "loss_num": 0.0133056640625,
      "loss_xval": 0.5546875,
      "num_input_tokens_seen": 332840084,
      "step": 5941
    },
    {
      "epoch": 13.233853006681514,
      "grad_norm": 13.681570053100586,
      "learning_rate": 1e-06,
      "loss": 0.36,
      "num_input_tokens_seen": 332896372,
      "step": 5942
    },
    {
      "epoch": 13.233853006681514,
      "loss": 0.34540247917175293,
      "loss_ce": 0.00018764848937280476,
      "loss_iou": 0.1494140625,
      "loss_num": 0.0091552734375,
      "loss_xval": 0.345703125,
      "num_input_tokens_seen": 332896372,
      "step": 5942
    },
    {
      "epoch": 13.236080178173719,
      "grad_norm": 17.690418243408203,
      "learning_rate": 1e-06,
      "loss": 0.5782,
      "num_input_tokens_seen": 332953536,
      "step": 5943
    },
    {
      "epoch": 13.236080178173719,
      "loss": 0.43799030780792236,
      "loss_ce": 0.00012410686758812517,
      "loss_iou": 0.20703125,
      "loss_num": 0.004608154296875,
      "loss_xval": 0.4375,
      "num_input_tokens_seen": 332953536,
      "step": 5943
    },
    {
      "epoch": 13.238307349665924,
      "grad_norm": 25.94896125793457,
      "learning_rate": 1e-06,
      "loss": 0.5025,
      "num_input_tokens_seen": 333008228,
      "step": 5944
    },
    {
      "epoch": 13.238307349665924,
      "loss": 0.5409551858901978,
      "loss_ce": 0.00012263350072316825,
      "loss_iou": 0.2314453125,
      "loss_num": 0.0157470703125,
      "loss_xval": 0.5390625,
      "num_input_tokens_seen": 333008228,
      "step": 5944
    },
    {
      "epoch": 13.240534521158128,
      "grad_norm": 20.952259063720703,
      "learning_rate": 1e-06,
      "loss": 0.4657,
      "num_input_tokens_seen": 333065924,
      "step": 5945
    },
    {
      "epoch": 13.240534521158128,
      "loss": 0.2911282777786255,
      "loss_ce": 0.00011265826469752938,
      "loss_iou": 0.1328125,
      "loss_num": 0.004913330078125,
      "loss_xval": 0.291015625,
      "num_input_tokens_seen": 333065924,
      "step": 5945
    },
    {
      "epoch": 13.242761692650333,
      "grad_norm": 117.26244354248047,
      "learning_rate": 1e-06,
      "loss": 0.6348,
      "num_input_tokens_seen": 333117748,
      "step": 5946
    },
    {
      "epoch": 13.242761692650333,
      "loss": 0.9416841864585876,
      "loss_ce": 0.00015579743194393814,
      "loss_iou": 0.421875,
      "loss_num": 0.0196533203125,
      "loss_xval": 0.94140625,
      "num_input_tokens_seen": 333117748,
      "step": 5946
    },
    {
      "epoch": 13.244988864142538,
      "grad_norm": 20.61402130126953,
      "learning_rate": 1e-06,
      "loss": 0.5212,
      "num_input_tokens_seen": 333175660,
      "step": 5947
    },
    {
      "epoch": 13.244988864142538,
      "loss": 0.5732554793357849,
      "loss_ce": 0.00025745650054886937,
      "loss_iou": 0.2470703125,
      "loss_num": 0.0157470703125,
      "loss_xval": 0.57421875,
      "num_input_tokens_seen": 333175660,
      "step": 5947
    },
    {
      "epoch": 13.247216035634743,
      "grad_norm": 18.954225540161133,
      "learning_rate": 1e-06,
      "loss": 0.621,
      "num_input_tokens_seen": 333231676,
      "step": 5948
    },
    {
      "epoch": 13.247216035634743,
      "loss": 0.6287169456481934,
      "loss_ce": 0.0002990136854350567,
      "loss_iou": 0.271484375,
      "loss_num": 0.016845703125,
      "loss_xval": 0.62890625,
      "num_input_tokens_seen": 333231676,
      "step": 5948
    },
    {
      "epoch": 13.249443207126948,
      "grad_norm": 16.13975715637207,
      "learning_rate": 1e-06,
      "loss": 0.5298,
      "num_input_tokens_seen": 333287440,
      "step": 5949
    },
    {
      "epoch": 13.249443207126948,
      "loss": 0.5134612917900085,
      "loss_ce": 0.00015562049520667642,
      "loss_iou": 0.224609375,
      "loss_num": 0.01275634765625,
      "loss_xval": 0.51171875,
      "num_input_tokens_seen": 333287440,
      "step": 5949
    },
    {
      "epoch": 13.251670378619155,
      "grad_norm": 21.270713806152344,
      "learning_rate": 1e-06,
      "loss": 0.4151,
      "num_input_tokens_seen": 333340480,
      "step": 5950
    },
    {
      "epoch": 13.251670378619155,
      "loss": 0.4000067412853241,
      "loss_ce": 0.00010438874596729875,
      "loss_iou": 0.1728515625,
      "loss_num": 0.01080322265625,
      "loss_xval": 0.400390625,
      "num_input_tokens_seen": 333340480,
      "step": 5950
    },
    {
      "epoch": 13.25389755011136,
      "grad_norm": 12.74732780456543,
      "learning_rate": 1e-06,
      "loss": 0.3358,
      "num_input_tokens_seen": 333396808,
      "step": 5951
    },
    {
      "epoch": 13.25389755011136,
      "loss": 0.2938316762447357,
      "loss_ce": 0.00013049585686530918,
      "loss_iou": 0.109375,
      "loss_num": 0.01507568359375,
      "loss_xval": 0.29296875,
      "num_input_tokens_seen": 333396808,
      "step": 5951
    },
    {
      "epoch": 13.256124721603564,
      "grad_norm": 16.515743255615234,
      "learning_rate": 1e-06,
      "loss": 0.4615,
      "num_input_tokens_seen": 333450628,
      "step": 5952
    },
    {
      "epoch": 13.256124721603564,
      "loss": 0.48923662304878235,
      "loss_ce": 0.00010087803093483672,
      "loss_iou": 0.201171875,
      "loss_num": 0.0172119140625,
      "loss_xval": 0.48828125,
      "num_input_tokens_seen": 333450628,
      "step": 5952
    },
    {
      "epoch": 13.25835189309577,
      "grad_norm": 29.306018829345703,
      "learning_rate": 1e-06,
      "loss": 0.3888,
      "num_input_tokens_seen": 333507188,
      "step": 5953
    },
    {
      "epoch": 13.25835189309577,
      "loss": 0.2866702973842621,
      "loss_ce": 0.00011024670675396919,
      "loss_iou": 0.1259765625,
      "loss_num": 0.006805419921875,
      "loss_xval": 0.287109375,
      "num_input_tokens_seen": 333507188,
      "step": 5953
    },
    {
      "epoch": 13.260579064587974,
      "grad_norm": 17.44719696044922,
      "learning_rate": 1e-06,
      "loss": 0.5703,
      "num_input_tokens_seen": 333560676,
      "step": 5954
    },
    {
      "epoch": 13.260579064587974,
      "loss": 0.6639593839645386,
      "loss_ce": 0.00014105206355452538,
      "loss_iou": 0.302734375,
      "loss_num": 0.01153564453125,
      "loss_xval": 0.6640625,
      "num_input_tokens_seen": 333560676,
      "step": 5954
    },
    {
      "epoch": 13.262806236080179,
      "grad_norm": 128.22251892089844,
      "learning_rate": 1e-06,
      "loss": 0.5567,
      "num_input_tokens_seen": 333618720,
      "step": 5955
    },
    {
      "epoch": 13.262806236080179,
      "loss": 0.6405112743377686,
      "loss_ce": 0.00013040687190368772,
      "loss_iou": 0.263671875,
      "loss_num": 0.02294921875,
      "loss_xval": 0.640625,
      "num_input_tokens_seen": 333618720,
      "step": 5955
    },
    {
      "epoch": 13.265033407572384,
      "grad_norm": 18.05571746826172,
      "learning_rate": 1e-06,
      "loss": 0.4476,
      "num_input_tokens_seen": 333676704,
      "step": 5956
    },
    {
      "epoch": 13.265033407572384,
      "loss": 0.4810963273048401,
      "loss_ce": 0.00013927766121923923,
      "loss_iou": 0.19921875,
      "loss_num": 0.0164794921875,
      "loss_xval": 0.48046875,
      "num_input_tokens_seen": 333676704,
      "step": 5956
    },
    {
      "epoch": 13.267260579064589,
      "grad_norm": 17.209970474243164,
      "learning_rate": 1e-06,
      "loss": 0.4548,
      "num_input_tokens_seen": 333732432,
      "step": 5957
    },
    {
      "epoch": 13.267260579064589,
      "loss": 0.4918234944343567,
      "loss_ce": 0.00012429816706571728,
      "loss_iou": 0.2041015625,
      "loss_num": 0.016845703125,
      "loss_xval": 0.4921875,
      "num_input_tokens_seen": 333732432,
      "step": 5957
    },
    {
      "epoch": 13.269487750556793,
      "grad_norm": 24.414487838745117,
      "learning_rate": 1e-06,
      "loss": 0.3929,
      "num_input_tokens_seen": 333790064,
      "step": 5958
    },
    {
      "epoch": 13.269487750556793,
      "loss": 0.2766364514827728,
      "loss_ce": 0.00014718460442963988,
      "loss_iou": 0.1171875,
      "loss_num": 0.00836181640625,
      "loss_xval": 0.27734375,
      "num_input_tokens_seen": 333790064,
      "step": 5958
    },
    {
      "epoch": 13.271714922048998,
      "grad_norm": 24.38360023498535,
      "learning_rate": 1e-06,
      "loss": 0.3817,
      "num_input_tokens_seen": 333845108,
      "step": 5959
    },
    {
      "epoch": 13.271714922048998,
      "loss": 0.34324610233306885,
      "loss_ce": 0.00047269114293158054,
      "loss_iou": 0.1572265625,
      "loss_num": 0.0057373046875,
      "loss_xval": 0.34375,
      "num_input_tokens_seen": 333845108,
      "step": 5959
    },
    {
      "epoch": 13.273942093541203,
      "grad_norm": 28.868574142456055,
      "learning_rate": 1e-06,
      "loss": 0.5365,
      "num_input_tokens_seen": 333899952,
      "step": 5960
    },
    {
      "epoch": 13.273942093541203,
      "loss": 0.45695704221725464,
      "loss_ce": 0.00010890320118051022,
      "loss_iou": 0.19140625,
      "loss_num": 0.0150146484375,
      "loss_xval": 0.45703125,
      "num_input_tokens_seen": 333899952,
      "step": 5960
    },
    {
      "epoch": 13.276169265033408,
      "grad_norm": 17.488313674926758,
      "learning_rate": 1e-06,
      "loss": 0.5148,
      "num_input_tokens_seen": 333955864,
      "step": 5961
    },
    {
      "epoch": 13.276169265033408,
      "loss": 0.64747554063797,
      "loss_ce": 0.00013665850565303117,
      "loss_iou": 0.275390625,
      "loss_num": 0.019287109375,
      "loss_xval": 0.6484375,
      "num_input_tokens_seen": 333955864,
      "step": 5961
    },
    {
      "epoch": 13.278396436525613,
      "grad_norm": 14.644098281860352,
      "learning_rate": 1e-06,
      "loss": 0.3249,
      "num_input_tokens_seen": 334012288,
      "step": 5962
    },
    {
      "epoch": 13.278396436525613,
      "loss": 0.32973378896713257,
      "loss_ce": 0.00014394470781553537,
      "loss_iou": 0.14453125,
      "loss_num": 0.00811767578125,
      "loss_xval": 0.330078125,
      "num_input_tokens_seen": 334012288,
      "step": 5962
    },
    {
      "epoch": 13.280623608017818,
      "grad_norm": 14.35937213897705,
      "learning_rate": 1e-06,
      "loss": 0.4478,
      "num_input_tokens_seen": 334070048,
      "step": 5963
    },
    {
      "epoch": 13.280623608017818,
      "loss": 0.5133100152015686,
      "loss_ce": 0.00012641007197089493,
      "loss_iou": 0.2236328125,
      "loss_num": 0.01318359375,
      "loss_xval": 0.51171875,
      "num_input_tokens_seen": 334070048,
      "step": 5963
    },
    {
      "epoch": 13.282850779510023,
      "grad_norm": 20.978622436523438,
      "learning_rate": 1e-06,
      "loss": 0.6486,
      "num_input_tokens_seen": 334125576,
      "step": 5964
    },
    {
      "epoch": 13.282850779510023,
      "loss": 0.6014399528503418,
      "loss_ce": 0.00012159476318629459,
      "loss_iou": 0.26171875,
      "loss_num": 0.015869140625,
      "loss_xval": 0.6015625,
      "num_input_tokens_seen": 334125576,
      "step": 5964
    },
    {
      "epoch": 13.285077951002227,
      "grad_norm": 17.045791625976562,
      "learning_rate": 1e-06,
      "loss": 0.4465,
      "num_input_tokens_seen": 334179024,
      "step": 5965
    },
    {
      "epoch": 13.285077951002227,
      "loss": 0.40493443608283997,
      "loss_ce": 0.00014927683514542878,
      "loss_iou": 0.18359375,
      "loss_num": 0.007415771484375,
      "loss_xval": 0.404296875,
      "num_input_tokens_seen": 334179024,
      "step": 5965
    },
    {
      "epoch": 13.287305122494432,
      "grad_norm": 17.58751678466797,
      "learning_rate": 1e-06,
      "loss": 0.5238,
      "num_input_tokens_seen": 334233464,
      "step": 5966
    },
    {
      "epoch": 13.287305122494432,
      "loss": 0.4632442593574524,
      "loss_ce": 0.00010949977149721235,
      "loss_iou": 0.2138671875,
      "loss_num": 0.007049560546875,
      "loss_xval": 0.462890625,
      "num_input_tokens_seen": 334233464,
      "step": 5966
    },
    {
      "epoch": 13.289532293986637,
      "grad_norm": 17.390804290771484,
      "learning_rate": 1e-06,
      "loss": 0.5365,
      "num_input_tokens_seen": 334288260,
      "step": 5967
    },
    {
      "epoch": 13.289532293986637,
      "loss": 0.6280704140663147,
      "loss_ce": 0.00014069479948375374,
      "loss_iou": 0.267578125,
      "loss_num": 0.018798828125,
      "loss_xval": 0.62890625,
      "num_input_tokens_seen": 334288260,
      "step": 5967
    },
    {
      "epoch": 13.291759465478842,
      "grad_norm": 20.9576358795166,
      "learning_rate": 1e-06,
      "loss": 0.3652,
      "num_input_tokens_seen": 334345852,
      "step": 5968
    },
    {
      "epoch": 13.291759465478842,
      "loss": 0.3794044554233551,
      "loss_ce": 0.0001319996954407543,
      "loss_iou": 0.1650390625,
      "loss_num": 0.0098876953125,
      "loss_xval": 0.37890625,
      "num_input_tokens_seen": 334345852,
      "step": 5968
    },
    {
      "epoch": 13.293986636971047,
      "grad_norm": 20.519437789916992,
      "learning_rate": 1e-06,
      "loss": 0.6077,
      "num_input_tokens_seen": 334402480,
      "step": 5969
    },
    {
      "epoch": 13.293986636971047,
      "loss": 0.8281785249710083,
      "loss_ce": 0.00011449880548752844,
      "loss_iou": 0.345703125,
      "loss_num": 0.02783203125,
      "loss_xval": 0.828125,
      "num_input_tokens_seen": 334402480,
      "step": 5969
    },
    {
      "epoch": 13.296213808463252,
      "grad_norm": 17.099870681762695,
      "learning_rate": 1e-06,
      "loss": 0.4134,
      "num_input_tokens_seen": 334459196,
      "step": 5970
    },
    {
      "epoch": 13.296213808463252,
      "loss": 0.35288766026496887,
      "loss_ce": 0.0001044573073158972,
      "loss_iou": 0.1533203125,
      "loss_num": 0.00921630859375,
      "loss_xval": 0.353515625,
      "num_input_tokens_seen": 334459196,
      "step": 5970
    },
    {
      "epoch": 13.298440979955457,
      "grad_norm": 16.155277252197266,
      "learning_rate": 1e-06,
      "loss": 0.4195,
      "num_input_tokens_seen": 334514964,
      "step": 5971
    },
    {
      "epoch": 13.298440979955457,
      "loss": 0.4192987084388733,
      "loss_ce": 0.00010926800314337015,
      "loss_iou": 0.1865234375,
      "loss_num": 0.00909423828125,
      "loss_xval": 0.419921875,
      "num_input_tokens_seen": 334514964,
      "step": 5971
    },
    {
      "epoch": 13.300668151447661,
      "grad_norm": 22.117937088012695,
      "learning_rate": 1e-06,
      "loss": 0.5437,
      "num_input_tokens_seen": 334570392,
      "step": 5972
    },
    {
      "epoch": 13.300668151447661,
      "loss": 0.524844765663147,
      "loss_ce": 0.0005832784809172153,
      "loss_iou": 0.205078125,
      "loss_num": 0.022705078125,
      "loss_xval": 0.5234375,
      "num_input_tokens_seen": 334570392,
      "step": 5972
    },
    {
      "epoch": 13.302895322939866,
      "grad_norm": 52.279701232910156,
      "learning_rate": 1e-06,
      "loss": 0.588,
      "num_input_tokens_seen": 334628400,
      "step": 5973
    },
    {
      "epoch": 13.302895322939866,
      "loss": 0.4728614389896393,
      "loss_ce": 0.001250399393029511,
      "loss_iou": 0.212890625,
      "loss_num": 0.00927734375,
      "loss_xval": 0.470703125,
      "num_input_tokens_seen": 334628400,
      "step": 5973
    },
    {
      "epoch": 13.305122494432071,
      "grad_norm": 42.509918212890625,
      "learning_rate": 1e-06,
      "loss": 0.7125,
      "num_input_tokens_seen": 334682692,
      "step": 5974
    },
    {
      "epoch": 13.305122494432071,
      "loss": 1.0085320472717285,
      "loss_ce": 0.0012078466825187206,
      "loss_iou": 0.41015625,
      "loss_num": 0.037109375,
      "loss_xval": 1.0078125,
      "num_input_tokens_seen": 334682692,
      "step": 5974
    },
    {
      "epoch": 13.307349665924276,
      "grad_norm": 23.17171287536621,
      "learning_rate": 1e-06,
      "loss": 0.5796,
      "num_input_tokens_seen": 334738240,
      "step": 5975
    },
    {
      "epoch": 13.307349665924276,
      "loss": 0.5287770628929138,
      "loss_ce": 9.056585986400023e-05,
      "loss_iou": 0.2060546875,
      "loss_num": 0.0230712890625,
      "loss_xval": 0.52734375,
      "num_input_tokens_seen": 334738240,
      "step": 5975
    },
    {
      "epoch": 13.309576837416481,
      "grad_norm": 15.882946014404297,
      "learning_rate": 1e-06,
      "loss": 0.449,
      "num_input_tokens_seen": 334795800,
      "step": 5976
    },
    {
      "epoch": 13.309576837416481,
      "loss": 0.51024329662323,
      "loss_ce": 0.00011146925680804998,
      "loss_iou": 0.224609375,
      "loss_num": 0.0123291015625,
      "loss_xval": 0.51171875,
      "num_input_tokens_seen": 334795800,
      "step": 5976
    },
    {
      "epoch": 13.311804008908686,
      "grad_norm": 16.047292709350586,
      "learning_rate": 1e-06,
      "loss": 0.525,
      "num_input_tokens_seen": 334853320,
      "step": 5977
    },
    {
      "epoch": 13.311804008908686,
      "loss": 0.42943665385246277,
      "loss_ce": 0.0001153927732957527,
      "loss_iou": 0.19140625,
      "loss_num": 0.0091552734375,
      "loss_xval": 0.4296875,
      "num_input_tokens_seen": 334853320,
      "step": 5977
    },
    {
      "epoch": 13.31403118040089,
      "grad_norm": 18.631088256835938,
      "learning_rate": 1e-06,
      "loss": 0.4337,
      "num_input_tokens_seen": 334908868,
      "step": 5978
    },
    {
      "epoch": 13.31403118040089,
      "loss": 0.3484327793121338,
      "loss_ce": 0.00010514883615542203,
      "loss_iou": 0.154296875,
      "loss_num": 0.0078125,
      "loss_xval": 0.34765625,
      "num_input_tokens_seen": 334908868,
      "step": 5978
    },
    {
      "epoch": 13.316258351893095,
      "grad_norm": 21.7668514251709,
      "learning_rate": 1e-06,
      "loss": 0.5039,
      "num_input_tokens_seen": 334966060,
      "step": 5979
    },
    {
      "epoch": 13.316258351893095,
      "loss": 0.5759820938110352,
      "loss_ce": 0.00011536551028257236,
      "loss_iou": 0.265625,
      "loss_num": 0.0091552734375,
      "loss_xval": 0.57421875,
      "num_input_tokens_seen": 334966060,
      "step": 5979
    },
    {
      "epoch": 13.3184855233853,
      "grad_norm": 18.8905029296875,
      "learning_rate": 1e-06,
      "loss": 0.3987,
      "num_input_tokens_seen": 335019804,
      "step": 5980
    },
    {
      "epoch": 13.3184855233853,
      "loss": 0.3916875720024109,
      "loss_ce": 0.00020808231784030795,
      "loss_iou": 0.1630859375,
      "loss_num": 0.01318359375,
      "loss_xval": 0.390625,
      "num_input_tokens_seen": 335019804,
      "step": 5980
    },
    {
      "epoch": 13.320712694877505,
      "grad_norm": 33.502071380615234,
      "learning_rate": 1e-06,
      "loss": 0.5049,
      "num_input_tokens_seen": 335077580,
      "step": 5981
    },
    {
      "epoch": 13.320712694877505,
      "loss": 0.5615602731704712,
      "loss_ce": 0.00015890991198830307,
      "loss_iou": 0.2265625,
      "loss_num": 0.021728515625,
      "loss_xval": 0.5625,
      "num_input_tokens_seen": 335077580,
      "step": 5981
    },
    {
      "epoch": 13.32293986636971,
      "grad_norm": 48.20892333984375,
      "learning_rate": 1e-06,
      "loss": 0.4792,
      "num_input_tokens_seen": 335131592,
      "step": 5982
    },
    {
      "epoch": 13.32293986636971,
      "loss": 0.5993452668190002,
      "loss_ce": 0.00010209815809503198,
      "loss_iou": 0.255859375,
      "loss_num": 0.01708984375,
      "loss_xval": 0.59765625,
      "num_input_tokens_seen": 335131592,
      "step": 5982
    },
    {
      "epoch": 13.325167037861915,
      "grad_norm": 14.08517837524414,
      "learning_rate": 1e-06,
      "loss": 0.4242,
      "num_input_tokens_seen": 335186972,
      "step": 5983
    },
    {
      "epoch": 13.325167037861915,
      "loss": 0.5378490090370178,
      "loss_ce": 0.0010219014948233962,
      "loss_iou": 0.203125,
      "loss_num": 0.0263671875,
      "loss_xval": 0.53515625,
      "num_input_tokens_seen": 335186972,
      "step": 5983
    },
    {
      "epoch": 13.32739420935412,
      "grad_norm": 15.391449928283691,
      "learning_rate": 1e-06,
      "loss": 0.6247,
      "num_input_tokens_seen": 335240660,
      "step": 5984
    },
    {
      "epoch": 13.32739420935412,
      "loss": 0.46067914366722107,
      "loss_ce": 0.00010786119673866779,
      "loss_iou": 0.201171875,
      "loss_num": 0.01177978515625,
      "loss_xval": 0.4609375,
      "num_input_tokens_seen": 335240660,
      "step": 5984
    },
    {
      "epoch": 13.329621380846325,
      "grad_norm": 25.606029510498047,
      "learning_rate": 1e-06,
      "loss": 0.4295,
      "num_input_tokens_seen": 335297500,
      "step": 5985
    },
    {
      "epoch": 13.329621380846325,
      "loss": 0.4411547780036926,
      "loss_ce": 0.00023681171296630055,
      "loss_iou": 0.1962890625,
      "loss_num": 0.00970458984375,
      "loss_xval": 0.44140625,
      "num_input_tokens_seen": 335297500,
      "step": 5985
    },
    {
      "epoch": 13.33184855233853,
      "grad_norm": 31.844512939453125,
      "learning_rate": 1e-06,
      "loss": 0.4698,
      "num_input_tokens_seen": 335350156,
      "step": 5986
    },
    {
      "epoch": 13.33184855233853,
      "loss": 0.41549986600875854,
      "loss_ce": 9.457894338993356e-05,
      "loss_iou": 0.166015625,
      "loss_num": 0.0167236328125,
      "loss_xval": 0.416015625,
      "num_input_tokens_seen": 335350156,
      "step": 5986
    },
    {
      "epoch": 13.334075723830734,
      "grad_norm": 24.63311195373535,
      "learning_rate": 1e-06,
      "loss": 0.5548,
      "num_input_tokens_seen": 335405788,
      "step": 5987
    },
    {
      "epoch": 13.334075723830734,
      "loss": 0.36372703313827515,
      "loss_ce": 0.00014061132969800383,
      "loss_iou": 0.1728515625,
      "loss_num": 0.0037994384765625,
      "loss_xval": 0.36328125,
      "num_input_tokens_seen": 335405788,
      "step": 5987
    },
    {
      "epoch": 13.33630289532294,
      "grad_norm": 13.776124000549316,
      "learning_rate": 1e-06,
      "loss": 0.4612,
      "num_input_tokens_seen": 335460976,
      "step": 5988
    },
    {
      "epoch": 13.33630289532294,
      "loss": 0.6514391899108887,
      "loss_ce": 0.00013307490735314786,
      "loss_iou": 0.29296875,
      "loss_num": 0.01263427734375,
      "loss_xval": 0.65234375,
      "num_input_tokens_seen": 335460976,
      "step": 5988
    },
    {
      "epoch": 13.338530066815144,
      "grad_norm": 15.759697914123535,
      "learning_rate": 1e-06,
      "loss": 0.5365,
      "num_input_tokens_seen": 335516704,
      "step": 5989
    },
    {
      "epoch": 13.338530066815144,
      "loss": 0.3756389021873474,
      "loss_ce": 0.00015063249156810343,
      "loss_iou": 0.138671875,
      "loss_num": 0.0196533203125,
      "loss_xval": 0.375,
      "num_input_tokens_seen": 335516704,
      "step": 5989
    },
    {
      "epoch": 13.340757238307349,
      "grad_norm": 14.048306465148926,
      "learning_rate": 1e-06,
      "loss": 0.2585,
      "num_input_tokens_seen": 335573700,
      "step": 5990
    },
    {
      "epoch": 13.340757238307349,
      "loss": 0.2530948221683502,
      "loss_ce": 0.00010410351387690753,
      "loss_iou": 0.11181640625,
      "loss_num": 0.00579833984375,
      "loss_xval": 0.25390625,
      "num_input_tokens_seen": 335573700,
      "step": 5990
    },
    {
      "epoch": 13.342984409799554,
      "grad_norm": 45.206886291503906,
      "learning_rate": 1e-06,
      "loss": 0.4845,
      "num_input_tokens_seen": 335628184,
      "step": 5991
    },
    {
      "epoch": 13.342984409799554,
      "loss": 0.6019352078437805,
      "loss_ce": 0.0001285744656343013,
      "loss_iou": 0.267578125,
      "loss_num": 0.01324462890625,
      "loss_xval": 0.6015625,
      "num_input_tokens_seen": 335628184,
      "step": 5991
    },
    {
      "epoch": 13.345211581291759,
      "grad_norm": 99.1800765991211,
      "learning_rate": 1e-06,
      "loss": 0.5862,
      "num_input_tokens_seen": 335684580,
      "step": 5992
    },
    {
      "epoch": 13.345211581291759,
      "loss": 0.6407464146614075,
      "loss_ce": 0.00012141239130869508,
      "loss_iou": 0.2890625,
      "loss_num": 0.0123291015625,
      "loss_xval": 0.640625,
      "num_input_tokens_seen": 335684580,
      "step": 5992
    },
    {
      "epoch": 13.347438752783964,
      "grad_norm": 16.95970916748047,
      "learning_rate": 1e-06,
      "loss": 0.4526,
      "num_input_tokens_seen": 335741296,
      "step": 5993
    },
    {
      "epoch": 13.347438752783964,
      "loss": 0.32951620221138,
      "loss_ce": 0.00010945653048111126,
      "loss_iou": 0.142578125,
      "loss_num": 0.0087890625,
      "loss_xval": 0.330078125,
      "num_input_tokens_seen": 335741296,
      "step": 5993
    },
    {
      "epoch": 13.34966592427617,
      "grad_norm": 40.8525505065918,
      "learning_rate": 1e-06,
      "loss": 0.4401,
      "num_input_tokens_seen": 335797804,
      "step": 5994
    },
    {
      "epoch": 13.34966592427617,
      "loss": 0.46965697407722473,
      "loss_ce": 0.00017455026682000607,
      "loss_iou": 0.1748046875,
      "loss_num": 0.0240478515625,
      "loss_xval": 0.46875,
      "num_input_tokens_seen": 335797804,
      "step": 5994
    },
    {
      "epoch": 13.351893095768375,
      "grad_norm": 23.179540634155273,
      "learning_rate": 1e-06,
      "loss": 0.6922,
      "num_input_tokens_seen": 335855320,
      "step": 5995
    },
    {
      "epoch": 13.351893095768375,
      "loss": 0.5909324288368225,
      "loss_ce": 0.00011209775402676314,
      "loss_iou": 0.271484375,
      "loss_num": 0.009521484375,
      "loss_xval": 0.58984375,
      "num_input_tokens_seen": 335855320,
      "step": 5995
    },
    {
      "epoch": 13.35412026726058,
      "grad_norm": 28.706586837768555,
      "learning_rate": 1e-06,
      "loss": 0.4458,
      "num_input_tokens_seen": 335911324,
      "step": 5996
    },
    {
      "epoch": 13.35412026726058,
      "loss": 0.4113472104072571,
      "loss_ce": 0.00021440400450956076,
      "loss_iou": 0.1904296875,
      "loss_num": 0.006195068359375,
      "loss_xval": 0.41015625,
      "num_input_tokens_seen": 335911324,
      "step": 5996
    },
    {
      "epoch": 13.356347438752785,
      "grad_norm": 14.33905029296875,
      "learning_rate": 1e-06,
      "loss": 0.3891,
      "num_input_tokens_seen": 335964808,
      "step": 5997
    },
    {
      "epoch": 13.356347438752785,
      "loss": 0.41698315739631653,
      "loss_ce": 0.00011303767678327858,
      "loss_iou": 0.1640625,
      "loss_num": 0.017578125,
      "loss_xval": 0.416015625,
      "num_input_tokens_seen": 335964808,
      "step": 5997
    },
    {
      "epoch": 13.35857461024499,
      "grad_norm": 15.864407539367676,
      "learning_rate": 1e-06,
      "loss": 0.3952,
      "num_input_tokens_seen": 336021256,
      "step": 5998
    },
    {
      "epoch": 13.35857461024499,
      "loss": 0.4110654890537262,
      "loss_ce": 0.00017680224846117198,
      "loss_iou": 0.1796875,
      "loss_num": 0.01025390625,
      "loss_xval": 0.41015625,
      "num_input_tokens_seen": 336021256,
      "step": 5998
    },
    {
      "epoch": 13.360801781737194,
      "grad_norm": 28.089126586914062,
      "learning_rate": 1e-06,
      "loss": 0.4279,
      "num_input_tokens_seen": 336074836,
      "step": 5999
    },
    {
      "epoch": 13.360801781737194,
      "loss": 0.4389985203742981,
      "loss_ce": 0.00017103503341786563,
      "loss_iou": 0.1689453125,
      "loss_num": 0.02001953125,
      "loss_xval": 0.439453125,
      "num_input_tokens_seen": 336074836,
      "step": 5999
    },
    {
      "epoch": 13.3630289532294,
      "grad_norm": 16.476625442504883,
      "learning_rate": 1e-06,
      "loss": 0.5683,
      "num_input_tokens_seen": 336130500,
      "step": 6000
    },
    {
      "epoch": 13.3630289532294,
      "eval_seeclick_web_CIoU": 0.589046448469162,
      "eval_seeclick_web_GIoU": 0.5870300531387329,
      "eval_seeclick_web_IoU": 0.6074443459510803,
      "eval_seeclick_web_MAE_all": 0.015537765808403492,
      "eval_seeclick_web_MAE_h": 0.007694335887208581,
      "eval_seeclick_web_MAE_w": 0.015699473209679127,
      "eval_seeclick_web_MAE_x_boxes": 0.009341908851638436,
      "eval_seeclick_web_MAE_y_boxes": 0.021628314396366477,
      "eval_seeclick_web_inside_bbox": 0.9166666567325592,
      "eval_seeclick_web_loss": 0.888150691986084,
      "eval_seeclick_web_loss_ce": 0.0001769300361047499,
      "eval_seeclick_web_loss_iou": 0.40771484375,
      "eval_seeclick_web_loss_num": 0.0123291015625,
      "eval_seeclick_web_loss_xval": 0.8765869140625,
      "eval_seeclick_web_runtime": 19.879,
      "eval_seeclick_web_samples_per_second": 2.515,
      "eval_seeclick_web_steps_per_second": 0.101,
      "num_input_tokens_seen": 336130500,
      "step": 6000
    },
    {
      "epoch": 13.3630289532294,
      "eval_icons_CIoU": 0.2701048105955124,
      "eval_icons_GIoU": 0.294067844748497,
      "eval_icons_IoU": 0.3528982400894165,
      "eval_icons_MAE_all": 0.06405339390039444,
      "eval_icons_MAE_h": 0.03851390350610018,
      "eval_icons_MAE_w": 0.06734280101954937,
      "eval_icons_MAE_x_boxes": 0.06041870452463627,
      "eval_icons_MAE_y_boxes": 0.03873829450458288,
      "eval_icons_inside_bbox": 0.6059027910232544,
      "eval_icons_loss": 1.748345971107483,
      "eval_icons_loss_ce": 0.00021618494793074206,
      "eval_icons_loss_iou": 0.6826171875,
      "eval_icons_loss_num": 0.061618804931640625,
      "eval_icons_loss_xval": 1.673828125,
      "eval_icons_runtime": 18.4453,
      "eval_icons_samples_per_second": 2.711,
      "eval_icons_steps_per_second": 0.108,
      "num_input_tokens_seen": 336130500,
      "step": 6000
    },
    {
      "epoch": 13.3630289532294,
      "eval_screenspot_CIoU": 0.35929131507873535,
      "eval_screenspot_GIoU": 0.3761854072411855,
      "eval_screenspot_IoU": 0.43523843089739483,
      "eval_screenspot_MAE_all": 0.05693357313672701,
      "eval_screenspot_MAE_h": 0.039716811850667,
      "eval_screenspot_MAE_w": 0.06415350238482158,
      "eval_screenspot_MAE_x_boxes": 0.069387707238396,
      "eval_screenspot_MAE_y_boxes": 0.03772336399803559,
      "eval_screenspot_inside_bbox": 0.6966666579246521,
      "eval_screenspot_loss": 1.5850136280059814,
      "eval_screenspot_loss_ce": 0.00022644254689415297,
      "eval_screenspot_loss_iou": 0.6593424479166666,
      "eval_screenspot_loss_num": 0.0646069844563802,
      "eval_screenspot_loss_xval": 1.640625,
      "eval_screenspot_runtime": 29.6001,
      "eval_screenspot_samples_per_second": 3.007,
      "eval_screenspot_steps_per_second": 0.101,
      "num_input_tokens_seen": 336130500,
      "step": 6000
    },
    {
      "epoch": 13.3630289532294,
      "eval_compot_CIoU": 0.3427671641111374,
      "eval_compot_GIoU": 0.3569800406694412,
      "eval_compot_IoU": 0.4003662168979645,
      "eval_compot_MAE_all": 0.019289949908852577,
      "eval_compot_MAE_h": 0.012674622237682343,
      "eval_compot_MAE_w": 0.021172930486500263,
      "eval_compot_MAE_x_boxes": 0.02997487783432007,
      "eval_compot_MAE_y_boxes": 0.007259466219693422,
      "eval_compot_inside_bbox": 0.6302083432674408,
      "eval_compot_loss": 1.4314215183258057,
      "eval_compot_loss_ce": 0.00017105540609918535,
      "eval_compot_loss_iou": 0.6610107421875,
      "eval_compot_loss_num": 0.018083572387695312,
      "eval_compot_loss_xval": 1.413330078125,
      "eval_compot_runtime": 19.5253,
      "eval_compot_samples_per_second": 2.561,
      "eval_compot_steps_per_second": 0.102,
      "num_input_tokens_seen": 336130500,
      "step": 6000
    },
    {
      "epoch": 13.3630289532294,
      "eval_custom_ui_val_CIoU": 0.48122422645489377,
      "eval_custom_ui_val_GIoU": 0.4910411420795653,
      "eval_custom_ui_val_IoU": 0.5414565900961558,
      "eval_custom_ui_val_MAE_all": 0.027185753505263064,
      "eval_custom_ui_val_MAE_h": 0.014914580983006291,
      "eval_custom_ui_val_MAE_w": 0.03390147609429227,
      "eval_custom_ui_val_MAE_x_boxes": 0.03341104726617535,
      "eval_custom_ui_val_MAE_y_boxes": 0.013283828376895852,
      "eval_custom_ui_val_inside_bbox": 0.7754629651705424,
      "eval_custom_ui_val_loss": 1.1734305620193481,
      "eval_custom_ui_val_loss_ce": 0.0001952857144513271,
      "eval_custom_ui_val_loss_iou": 0.5054389105902778,
      "eval_custom_ui_val_loss_num": 0.024320814344618056,
      "eval_custom_ui_val_loss_xval": 1.1326497395833333,
      "eval_custom_ui_val_runtime": 56.7538,
      "eval_custom_ui_val_samples_per_second": 4.669,
      "eval_custom_ui_val_steps_per_second": 0.159,
      "num_input_tokens_seen": 336130500,
      "step": 6000
    },
    {
      "epoch": 13.3630289532294,
      "loss": 0.8641664385795593,
      "loss_ce": 0.00015277693455573171,
      "loss_iou": 0.392578125,
      "loss_num": 0.015869140625,
      "loss_xval": 0.86328125,
      "num_input_tokens_seen": 336130500,
      "step": 6000
    },
    {
      "epoch": 13.365256124721604,
      "grad_norm": 16.492589950561523,
      "learning_rate": 1e-06,
      "loss": 0.3624,
      "num_input_tokens_seen": 336187644,
      "step": 6001
    },
    {
      "epoch": 13.365256124721604,
      "loss": 0.38099753856658936,
      "loss_ce": 0.00026025049737654626,
      "loss_iou": 0.1728515625,
      "loss_num": 0.00701904296875,
      "loss_xval": 0.380859375,
      "num_input_tokens_seen": 336187644,
      "step": 6001
    },
    {
      "epoch": 13.367483296213809,
      "grad_norm": 16.520002365112305,
      "learning_rate": 1e-06,
      "loss": 0.4096,
      "num_input_tokens_seen": 336244060,
      "step": 6002
    },
    {
      "epoch": 13.367483296213809,
      "loss": 0.2786784768104553,
      "loss_ce": 0.00011401639494579285,
      "loss_iou": 0.11181640625,
      "loss_num": 0.010986328125,
      "loss_xval": 0.279296875,
      "num_input_tokens_seen": 336244060,
      "step": 6002
    },
    {
      "epoch": 13.369710467706014,
      "grad_norm": 27.00398826599121,
      "learning_rate": 1e-06,
      "loss": 0.4073,
      "num_input_tokens_seen": 336298968,
      "step": 6003
    },
    {
      "epoch": 13.369710467706014,
      "loss": 0.3439924120903015,
      "loss_ce": 0.00012036073167109862,
      "loss_iou": 0.146484375,
      "loss_num": 0.0103759765625,
      "loss_xval": 0.34375,
      "num_input_tokens_seen": 336298968,
      "step": 6003
    },
    {
      "epoch": 13.371937639198219,
      "grad_norm": 14.05265998840332,
      "learning_rate": 1e-06,
      "loss": 0.3699,
      "num_input_tokens_seen": 336354760,
      "step": 6004
    },
    {
      "epoch": 13.371937639198219,
      "loss": 0.3934824466705322,
      "loss_ce": 0.0001108648139052093,
      "loss_iou": 0.1767578125,
      "loss_num": 0.0079345703125,
      "loss_xval": 0.392578125,
      "num_input_tokens_seen": 336354760,
      "step": 6004
    },
    {
      "epoch": 13.374164810690424,
      "grad_norm": 24.29564094543457,
      "learning_rate": 1e-06,
      "loss": 0.5934,
      "num_input_tokens_seen": 336412268,
      "step": 6005
    },
    {
      "epoch": 13.374164810690424,
      "loss": 0.372583270072937,
      "loss_ce": 0.00011621671728789806,
      "loss_iou": 0.166015625,
      "loss_num": 0.008056640625,
      "loss_xval": 0.373046875,
      "num_input_tokens_seen": 336412268,
      "step": 6005
    },
    {
      "epoch": 13.376391982182628,
      "grad_norm": 18.128660202026367,
      "learning_rate": 1e-06,
      "loss": 0.4611,
      "num_input_tokens_seen": 336467380,
      "step": 6006
    },
    {
      "epoch": 13.376391982182628,
      "loss": 0.4521748423576355,
      "loss_ce": 0.00014848702994640917,
      "loss_iou": 0.20703125,
      "loss_num": 0.007781982421875,
      "loss_xval": 0.451171875,
      "num_input_tokens_seen": 336467380,
      "step": 6006
    },
    {
      "epoch": 13.378619153674833,
      "grad_norm": 17.15667152404785,
      "learning_rate": 1e-06,
      "loss": 0.4673,
      "num_input_tokens_seen": 336522912,
      "step": 6007
    },
    {
      "epoch": 13.378619153674833,
      "loss": 0.4950922429561615,
      "loss_ce": 9.710421727504581e-05,
      "loss_iou": 0.2158203125,
      "loss_num": 0.012939453125,
      "loss_xval": 0.494140625,
      "num_input_tokens_seen": 336522912,
      "step": 6007
    },
    {
      "epoch": 13.380846325167038,
      "grad_norm": 36.65664291381836,
      "learning_rate": 1e-06,
      "loss": 0.4762,
      "num_input_tokens_seen": 336578132,
      "step": 6008
    },
    {
      "epoch": 13.380846325167038,
      "loss": 0.4739997982978821,
      "loss_ce": 0.0002449120220262557,
      "loss_iou": 0.1728515625,
      "loss_num": 0.025634765625,
      "loss_xval": 0.474609375,
      "num_input_tokens_seen": 336578132,
      "step": 6008
    },
    {
      "epoch": 13.383073496659243,
      "grad_norm": 17.556377410888672,
      "learning_rate": 1e-06,
      "loss": 0.4995,
      "num_input_tokens_seen": 336636856,
      "step": 6009
    },
    {
      "epoch": 13.383073496659243,
      "loss": 0.5950038433074951,
      "loss_ce": 0.0001551837776787579,
      "loss_iou": 0.25390625,
      "loss_num": 0.017578125,
      "loss_xval": 0.59375,
      "num_input_tokens_seen": 336636856,
      "step": 6009
    },
    {
      "epoch": 13.385300668151448,
      "grad_norm": 30.719839096069336,
      "learning_rate": 1e-06,
      "loss": 0.5576,
      "num_input_tokens_seen": 336694384,
      "step": 6010
    },
    {
      "epoch": 13.385300668151448,
      "loss": 0.5820613503456116,
      "loss_ce": 0.00015218451153486967,
      "loss_iou": 0.24609375,
      "loss_num": 0.0179443359375,
      "loss_xval": 0.58203125,
      "num_input_tokens_seen": 336694384,
      "step": 6010
    },
    {
      "epoch": 13.387527839643653,
      "grad_norm": 16.480695724487305,
      "learning_rate": 1e-06,
      "loss": 0.5837,
      "num_input_tokens_seen": 336751420,
      "step": 6011
    },
    {
      "epoch": 13.387527839643653,
      "loss": 0.4113236367702484,
      "loss_ce": 0.00012982796761207283,
      "loss_iou": 0.1796875,
      "loss_num": 0.010498046875,
      "loss_xval": 0.412109375,
      "num_input_tokens_seen": 336751420,
      "step": 6011
    },
    {
      "epoch": 13.389755011135858,
      "grad_norm": 27.67957305908203,
      "learning_rate": 1e-06,
      "loss": 0.4264,
      "num_input_tokens_seen": 336804964,
      "step": 6012
    },
    {
      "epoch": 13.389755011135858,
      "loss": 0.5694878101348877,
      "loss_ce": 0.00015187214012257755,
      "loss_iou": 0.265625,
      "loss_num": 0.00732421875,
      "loss_xval": 0.5703125,
      "num_input_tokens_seen": 336804964,
      "step": 6012
    },
    {
      "epoch": 13.391982182628063,
      "grad_norm": 15.864426612854004,
      "learning_rate": 1e-06,
      "loss": 0.6295,
      "num_input_tokens_seen": 336862960,
      "step": 6013
    },
    {
      "epoch": 13.391982182628063,
      "loss": 0.425285279750824,
      "loss_ce": 0.00011438117508077994,
      "loss_iou": 0.189453125,
      "loss_num": 0.00927734375,
      "loss_xval": 0.42578125,
      "num_input_tokens_seen": 336862960,
      "step": 6013
    },
    {
      "epoch": 13.394209354120267,
      "grad_norm": 13.584254264831543,
      "learning_rate": 1e-06,
      "loss": 0.6117,
      "num_input_tokens_seen": 336920076,
      "step": 6014
    },
    {
      "epoch": 13.394209354120267,
      "loss": 0.4187964200973511,
      "loss_ce": 9.524912456981838e-05,
      "loss_iou": 0.1865234375,
      "loss_num": 0.00927734375,
      "loss_xval": 0.41796875,
      "num_input_tokens_seen": 336920076,
      "step": 6014
    },
    {
      "epoch": 13.396436525612472,
      "grad_norm": 15.942584037780762,
      "learning_rate": 1e-06,
      "loss": 0.4851,
      "num_input_tokens_seen": 336975628,
      "step": 6015
    },
    {
      "epoch": 13.396436525612472,
      "loss": 0.5123189687728882,
      "loss_ce": 0.00011192913370905444,
      "loss_iou": 0.2294921875,
      "loss_num": 0.0107421875,
      "loss_xval": 0.51171875,
      "num_input_tokens_seen": 336975628,
      "step": 6015
    },
    {
      "epoch": 13.398663697104677,
      "grad_norm": 29.960575103759766,
      "learning_rate": 1e-06,
      "loss": 0.4581,
      "num_input_tokens_seen": 337031400,
      "step": 6016
    },
    {
      "epoch": 13.398663697104677,
      "loss": 0.4230641722679138,
      "loss_ce": 9.053543908521533e-05,
      "loss_iou": 0.1728515625,
      "loss_num": 0.015625,
      "loss_xval": 0.423828125,
      "num_input_tokens_seen": 337031400,
      "step": 6016
    },
    {
      "epoch": 13.400890868596882,
      "grad_norm": 17.33017921447754,
      "learning_rate": 1e-06,
      "loss": 0.3594,
      "num_input_tokens_seen": 337086372,
      "step": 6017
    },
    {
      "epoch": 13.400890868596882,
      "loss": 0.46412086486816406,
      "loss_ce": 0.00013159040827304125,
      "loss_iou": 0.2138671875,
      "loss_num": 0.0072021484375,
      "loss_xval": 0.46484375,
      "num_input_tokens_seen": 337086372,
      "step": 6017
    },
    {
      "epoch": 13.403118040089087,
      "grad_norm": 24.617557525634766,
      "learning_rate": 1e-06,
      "loss": 0.6371,
      "num_input_tokens_seen": 337141460,
      "step": 6018
    },
    {
      "epoch": 13.403118040089087,
      "loss": 0.4511684775352478,
      "loss_ce": 0.00011867978173540905,
      "loss_iou": 0.201171875,
      "loss_num": 0.0098876953125,
      "loss_xval": 0.451171875,
      "num_input_tokens_seen": 337141460,
      "step": 6018
    },
    {
      "epoch": 13.405345211581292,
      "grad_norm": 22.865720748901367,
      "learning_rate": 1e-06,
      "loss": 0.4544,
      "num_input_tokens_seen": 337199884,
      "step": 6019
    },
    {
      "epoch": 13.405345211581292,
      "loss": 0.4763756990432739,
      "loss_ce": 0.00011837005149573088,
      "loss_iou": 0.216796875,
      "loss_num": 0.0084228515625,
      "loss_xval": 0.4765625,
      "num_input_tokens_seen": 337199884,
      "step": 6019
    },
    {
      "epoch": 13.407572383073497,
      "grad_norm": 30.050506591796875,
      "learning_rate": 1e-06,
      "loss": 0.3933,
      "num_input_tokens_seen": 337255584,
      "step": 6020
    },
    {
      "epoch": 13.407572383073497,
      "loss": 0.5196518301963806,
      "loss_ce": 0.00012056018749717623,
      "loss_iou": 0.203125,
      "loss_num": 0.022705078125,
      "loss_xval": 0.51953125,
      "num_input_tokens_seen": 337255584,
      "step": 6020
    },
    {
      "epoch": 13.409799554565701,
      "grad_norm": 16.1778564453125,
      "learning_rate": 1e-06,
      "loss": 0.4133,
      "num_input_tokens_seen": 337314328,
      "step": 6021
    },
    {
      "epoch": 13.409799554565701,
      "loss": 0.4304283857345581,
      "loss_ce": 0.00013054994633421302,
      "loss_iou": 0.19921875,
      "loss_num": 0.006317138671875,
      "loss_xval": 0.4296875,
      "num_input_tokens_seen": 337314328,
      "step": 6021
    },
    {
      "epoch": 13.412026726057906,
      "grad_norm": 19.43562126159668,
      "learning_rate": 1e-06,
      "loss": 0.5453,
      "num_input_tokens_seen": 337371160,
      "step": 6022
    },
    {
      "epoch": 13.412026726057906,
      "loss": 0.8006750345230103,
      "loss_ce": 0.00013796251732856035,
      "loss_iou": 0.3125,
      "loss_num": 0.03515625,
      "loss_xval": 0.80078125,
      "num_input_tokens_seen": 337371160,
      "step": 6022
    },
    {
      "epoch": 13.414253897550111,
      "grad_norm": 20.63788414001465,
      "learning_rate": 1e-06,
      "loss": 0.3787,
      "num_input_tokens_seen": 337427584,
      "step": 6023
    },
    {
      "epoch": 13.414253897550111,
      "loss": 0.45666825771331787,
      "loss_ce": 0.00012530997628346086,
      "loss_iou": 0.2001953125,
      "loss_num": 0.0113525390625,
      "loss_xval": 0.45703125,
      "num_input_tokens_seen": 337427584,
      "step": 6023
    },
    {
      "epoch": 13.416481069042316,
      "grad_norm": 14.40611457824707,
      "learning_rate": 1e-06,
      "loss": 0.4986,
      "num_input_tokens_seen": 337486896,
      "step": 6024
    },
    {
      "epoch": 13.416481069042316,
      "loss": 0.4070853590965271,
      "loss_ce": 0.00010296083928551525,
      "loss_iou": 0.1875,
      "loss_num": 0.00653076171875,
      "loss_xval": 0.40625,
      "num_input_tokens_seen": 337486896,
      "step": 6024
    },
    {
      "epoch": 13.41870824053452,
      "grad_norm": 17.12459945678711,
      "learning_rate": 1e-06,
      "loss": 0.3375,
      "num_input_tokens_seen": 337544700,
      "step": 6025
    },
    {
      "epoch": 13.41870824053452,
      "loss": 0.28037208318710327,
      "loss_ce": 9.865299216471612e-05,
      "loss_iou": 0.12109375,
      "loss_num": 0.007476806640625,
      "loss_xval": 0.28125,
      "num_input_tokens_seen": 337544700,
      "step": 6025
    },
    {
      "epoch": 13.420935412026726,
      "grad_norm": 17.389301300048828,
      "learning_rate": 1e-06,
      "loss": 0.5866,
      "num_input_tokens_seen": 337602792,
      "step": 6026
    },
    {
      "epoch": 13.420935412026726,
      "loss": 0.34812402725219727,
      "loss_ce": 0.00010155315976589918,
      "loss_iou": 0.15234375,
      "loss_num": 0.00872802734375,
      "loss_xval": 0.34765625,
      "num_input_tokens_seen": 337602792,
      "step": 6026
    },
    {
      "epoch": 13.42316258351893,
      "grad_norm": 14.135052680969238,
      "learning_rate": 1e-06,
      "loss": 0.6344,
      "num_input_tokens_seen": 337657324,
      "step": 6027
    },
    {
      "epoch": 13.42316258351893,
      "loss": 0.9954932332038879,
      "loss_ce": 0.00013188435696065426,
      "loss_iou": 0.419921875,
      "loss_num": 0.031494140625,
      "loss_xval": 0.99609375,
      "num_input_tokens_seen": 337657324,
      "step": 6027
    },
    {
      "epoch": 13.425389755011135,
      "grad_norm": 17.855140686035156,
      "learning_rate": 1e-06,
      "loss": 0.4056,
      "num_input_tokens_seen": 337713028,
      "step": 6028
    },
    {
      "epoch": 13.425389755011135,
      "loss": 0.30981168150901794,
      "loss_ce": 0.00011929747415706515,
      "loss_iou": 0.1318359375,
      "loss_num": 0.00921630859375,
      "loss_xval": 0.310546875,
      "num_input_tokens_seen": 337713028,
      "step": 6028
    },
    {
      "epoch": 13.42761692650334,
      "grad_norm": 19.777183532714844,
      "learning_rate": 1e-06,
      "loss": 0.4961,
      "num_input_tokens_seen": 337767148,
      "step": 6029
    },
    {
      "epoch": 13.42761692650334,
      "loss": 0.45299914479255676,
      "loss_ce": 0.00011829864524770528,
      "loss_iou": 0.19921875,
      "loss_num": 0.01104736328125,
      "loss_xval": 0.453125,
      "num_input_tokens_seen": 337767148,
      "step": 6029
    },
    {
      "epoch": 13.429844097995545,
      "grad_norm": 28.941667556762695,
      "learning_rate": 1e-06,
      "loss": 0.3823,
      "num_input_tokens_seen": 337824252,
      "step": 6030
    },
    {
      "epoch": 13.429844097995545,
      "loss": 0.4544570744037628,
      "loss_ce": 0.00011135141539853066,
      "loss_iou": 0.2001953125,
      "loss_num": 0.0108642578125,
      "loss_xval": 0.455078125,
      "num_input_tokens_seen": 337824252,
      "step": 6030
    },
    {
      "epoch": 13.43207126948775,
      "grad_norm": 20.595233917236328,
      "learning_rate": 1e-06,
      "loss": 0.7976,
      "num_input_tokens_seen": 337879876,
      "step": 6031
    },
    {
      "epoch": 13.43207126948775,
      "loss": 0.8424028754234314,
      "loss_ce": 0.0001787315122783184,
      "loss_iou": 0.34765625,
      "loss_num": 0.029296875,
      "loss_xval": 0.84375,
      "num_input_tokens_seen": 337879876,
      "step": 6031
    },
    {
      "epoch": 13.434298440979955,
      "grad_norm": 37.23560333251953,
      "learning_rate": 1e-06,
      "loss": 0.5585,
      "num_input_tokens_seen": 337936932,
      "step": 6032
    },
    {
      "epoch": 13.434298440979955,
      "loss": 0.4548329710960388,
      "loss_ce": 0.00012105993664590642,
      "loss_iou": 0.181640625,
      "loss_num": 0.018310546875,
      "loss_xval": 0.455078125,
      "num_input_tokens_seen": 337936932,
      "step": 6032
    },
    {
      "epoch": 13.43652561247216,
      "grad_norm": 20.569580078125,
      "learning_rate": 1e-06,
      "loss": 0.4455,
      "num_input_tokens_seen": 337993396,
      "step": 6033
    },
    {
      "epoch": 13.43652561247216,
      "loss": 0.6414711475372314,
      "loss_ce": 0.00011370141146471724,
      "loss_iou": 0.2294921875,
      "loss_num": 0.036376953125,
      "loss_xval": 0.640625,
      "num_input_tokens_seen": 337993396,
      "step": 6033
    },
    {
      "epoch": 13.438752783964365,
      "grad_norm": 21.35686492919922,
      "learning_rate": 1e-06,
      "loss": 0.5028,
      "num_input_tokens_seen": 338048168,
      "step": 6034
    },
    {
      "epoch": 13.438752783964365,
      "loss": 0.27768754959106445,
      "loss_ce": 0.0001606793375685811,
      "loss_iou": 0.1279296875,
      "loss_num": 0.004180908203125,
      "loss_xval": 0.27734375,
      "num_input_tokens_seen": 338048168,
      "step": 6034
    },
    {
      "epoch": 13.44097995545657,
      "grad_norm": 14.87231159210205,
      "learning_rate": 1e-06,
      "loss": 0.6545,
      "num_input_tokens_seen": 338104776,
      "step": 6035
    },
    {
      "epoch": 13.44097995545657,
      "loss": 0.5546998381614685,
      "loss_ce": 0.00013442571798805147,
      "loss_iou": 0.2294921875,
      "loss_num": 0.0191650390625,
      "loss_xval": 0.5546875,
      "num_input_tokens_seen": 338104776,
      "step": 6035
    },
    {
      "epoch": 13.443207126948774,
      "grad_norm": 17.35746955871582,
      "learning_rate": 1e-06,
      "loss": 0.3714,
      "num_input_tokens_seen": 338160688,
      "step": 6036
    },
    {
      "epoch": 13.443207126948774,
      "loss": 0.2007206380367279,
      "loss_ce": 8.280624751932919e-05,
      "loss_iou": 0.080078125,
      "loss_num": 0.00799560546875,
      "loss_xval": 0.2001953125,
      "num_input_tokens_seen": 338160688,
      "step": 6036
    },
    {
      "epoch": 13.44543429844098,
      "grad_norm": 32.42356872558594,
      "learning_rate": 1e-06,
      "loss": 0.5975,
      "num_input_tokens_seen": 338216708,
      "step": 6037
    },
    {
      "epoch": 13.44543429844098,
      "loss": 0.4356258511543274,
      "loss_ce": 0.00020103095448575914,
      "loss_iou": 0.171875,
      "loss_num": 0.0185546875,
      "loss_xval": 0.435546875,
      "num_input_tokens_seen": 338216708,
      "step": 6037
    },
    {
      "epoch": 13.447661469933184,
      "grad_norm": 16.385221481323242,
      "learning_rate": 1e-06,
      "loss": 0.3499,
      "num_input_tokens_seen": 338273736,
      "step": 6038
    },
    {
      "epoch": 13.447661469933184,
      "loss": 0.3894416391849518,
      "loss_ce": 0.00015943063772283494,
      "loss_iou": 0.16015625,
      "loss_num": 0.01361083984375,
      "loss_xval": 0.388671875,
      "num_input_tokens_seen": 338273736,
      "step": 6038
    },
    {
      "epoch": 13.449888641425389,
      "grad_norm": 16.43487548828125,
      "learning_rate": 1e-06,
      "loss": 0.4453,
      "num_input_tokens_seen": 338329848,
      "step": 6039
    },
    {
      "epoch": 13.449888641425389,
      "loss": 0.4355180859565735,
      "loss_ce": 9.329354361398146e-05,
      "loss_iou": 0.18359375,
      "loss_num": 0.01373291015625,
      "loss_xval": 0.435546875,
      "num_input_tokens_seen": 338329848,
      "step": 6039
    },
    {
      "epoch": 13.452115812917596,
      "grad_norm": 10.888897895812988,
      "learning_rate": 1e-06,
      "loss": 0.2805,
      "num_input_tokens_seen": 338384060,
      "step": 6040
    },
    {
      "epoch": 13.452115812917596,
      "loss": 0.24022263288497925,
      "loss_ce": 0.0001103131435229443,
      "loss_iou": 0.09912109375,
      "loss_num": 0.0084228515625,
      "loss_xval": 0.240234375,
      "num_input_tokens_seen": 338384060,
      "step": 6040
    },
    {
      "epoch": 13.4543429844098,
      "grad_norm": 16.131216049194336,
      "learning_rate": 1e-06,
      "loss": 0.3434,
      "num_input_tokens_seen": 338439568,
      "step": 6041
    },
    {
      "epoch": 13.4543429844098,
      "loss": 0.44243180751800537,
      "loss_ce": 0.00011001640814356506,
      "loss_iou": 0.1884765625,
      "loss_num": 0.012939453125,
      "loss_xval": 0.44140625,
      "num_input_tokens_seen": 338439568,
      "step": 6041
    },
    {
      "epoch": 13.456570155902005,
      "grad_norm": 24.778518676757812,
      "learning_rate": 1e-06,
      "loss": 0.3777,
      "num_input_tokens_seen": 338495172,
      "step": 6042
    },
    {
      "epoch": 13.456570155902005,
      "loss": 0.5041686296463013,
      "loss_ce": 0.00035395825398154557,
      "loss_iou": 0.2197265625,
      "loss_num": 0.01287841796875,
      "loss_xval": 0.50390625,
      "num_input_tokens_seen": 338495172,
      "step": 6042
    },
    {
      "epoch": 13.45879732739421,
      "grad_norm": 25.524629592895508,
      "learning_rate": 1e-06,
      "loss": 0.5037,
      "num_input_tokens_seen": 338550088,
      "step": 6043
    },
    {
      "epoch": 13.45879732739421,
      "loss": 0.5563878417015076,
      "loss_ce": 0.00011343901860527694,
      "loss_iou": 0.248046875,
      "loss_num": 0.01202392578125,
      "loss_xval": 0.5546875,
      "num_input_tokens_seen": 338550088,
      "step": 6043
    },
    {
      "epoch": 13.461024498886415,
      "grad_norm": 19.04424285888672,
      "learning_rate": 1e-06,
      "loss": 0.583,
      "num_input_tokens_seen": 338605712,
      "step": 6044
    },
    {
      "epoch": 13.461024498886415,
      "loss": 0.6645296812057495,
      "loss_ce": 0.00022306920436676592,
      "loss_iou": 0.2578125,
      "loss_num": 0.0301513671875,
      "loss_xval": 0.6640625,
      "num_input_tokens_seen": 338605712,
      "step": 6044
    },
    {
      "epoch": 13.46325167037862,
      "grad_norm": 14.406312942504883,
      "learning_rate": 1e-06,
      "loss": 0.379,
      "num_input_tokens_seen": 338662752,
      "step": 6045
    },
    {
      "epoch": 13.46325167037862,
      "loss": 0.3758947253227234,
      "loss_ce": 0.00010126342385774478,
      "loss_iou": 0.1533203125,
      "loss_num": 0.013671875,
      "loss_xval": 0.375,
      "num_input_tokens_seen": 338662752,
      "step": 6045
    },
    {
      "epoch": 13.465478841870825,
      "grad_norm": 40.62894821166992,
      "learning_rate": 1e-06,
      "loss": 0.5142,
      "num_input_tokens_seen": 338716672,
      "step": 6046
    },
    {
      "epoch": 13.465478841870825,
      "loss": 0.471059650182724,
      "loss_ce": 0.00011239292507525533,
      "loss_iou": 0.2099609375,
      "loss_num": 0.01025390625,
      "loss_xval": 0.470703125,
      "num_input_tokens_seen": 338716672,
      "step": 6046
    },
    {
      "epoch": 13.46770601336303,
      "grad_norm": 22.999338150024414,
      "learning_rate": 1e-06,
      "loss": 0.6799,
      "num_input_tokens_seen": 338773908,
      "step": 6047
    },
    {
      "epoch": 13.46770601336303,
      "loss": 0.46885955333709717,
      "loss_ce": 0.00010957221093121916,
      "loss_iou": 0.2119140625,
      "loss_num": 0.00885009765625,
      "loss_xval": 0.46875,
      "num_input_tokens_seen": 338773908,
      "step": 6047
    },
    {
      "epoch": 13.469933184855234,
      "grad_norm": 17.78192138671875,
      "learning_rate": 1e-06,
      "loss": 0.4976,
      "num_input_tokens_seen": 338831676,
      "step": 6048
    },
    {
      "epoch": 13.469933184855234,
      "loss": 0.47100013494491577,
      "loss_ce": 0.00017495593056082726,
      "loss_iou": 0.1953125,
      "loss_num": 0.0159912109375,
      "loss_xval": 0.470703125,
      "num_input_tokens_seen": 338831676,
      "step": 6048
    },
    {
      "epoch": 13.47216035634744,
      "grad_norm": 31.006885528564453,
      "learning_rate": 1e-06,
      "loss": 0.5294,
      "num_input_tokens_seen": 338886208,
      "step": 6049
    },
    {
      "epoch": 13.47216035634744,
      "loss": 0.49864301085472107,
      "loss_ce": 0.00010783917969092727,
      "loss_iou": 0.2275390625,
      "loss_num": 0.00872802734375,
      "loss_xval": 0.498046875,
      "num_input_tokens_seen": 338886208,
      "step": 6049
    },
    {
      "epoch": 13.474387527839644,
      "grad_norm": 17.400957107543945,
      "learning_rate": 1e-06,
      "loss": 0.5978,
      "num_input_tokens_seen": 338941916,
      "step": 6050
    },
    {
      "epoch": 13.474387527839644,
      "loss": 0.5912978649139404,
      "loss_ce": 0.00017235292762052268,
      "loss_iou": 0.263671875,
      "loss_num": 0.01263427734375,
      "loss_xval": 0.58984375,
      "num_input_tokens_seen": 338941916,
      "step": 6050
    },
    {
      "epoch": 13.476614699331849,
      "grad_norm": 17.695465087890625,
      "learning_rate": 1e-06,
      "loss": 0.4803,
      "num_input_tokens_seen": 338998860,
      "step": 6051
    },
    {
      "epoch": 13.476614699331849,
      "loss": 0.4964655339717865,
      "loss_ce": 0.0001276445691473782,
      "loss_iou": 0.21875,
      "loss_num": 0.01190185546875,
      "loss_xval": 0.49609375,
      "num_input_tokens_seen": 338998860,
      "step": 6051
    },
    {
      "epoch": 13.478841870824054,
      "grad_norm": 21.691905975341797,
      "learning_rate": 1e-06,
      "loss": 0.6199,
      "num_input_tokens_seen": 339051660,
      "step": 6052
    },
    {
      "epoch": 13.478841870824054,
      "loss": 0.5180391073226929,
      "loss_ce": 9.476025297772139e-05,
      "loss_iou": 0.2197265625,
      "loss_num": 0.015869140625,
      "loss_xval": 0.51953125,
      "num_input_tokens_seen": 339051660,
      "step": 6052
    },
    {
      "epoch": 13.481069042316259,
      "grad_norm": 17.7266845703125,
      "learning_rate": 1e-06,
      "loss": 0.5722,
      "num_input_tokens_seen": 339107488,
      "step": 6053
    },
    {
      "epoch": 13.481069042316259,
      "loss": 0.4541052579879761,
      "loss_ce": 0.00012577215966302902,
      "loss_iou": 0.205078125,
      "loss_num": 0.0087890625,
      "loss_xval": 0.453125,
      "num_input_tokens_seen": 339107488,
      "step": 6053
    },
    {
      "epoch": 13.483296213808464,
      "grad_norm": 15.927001953125,
      "learning_rate": 1e-06,
      "loss": 0.4988,
      "num_input_tokens_seen": 339162448,
      "step": 6054
    },
    {
      "epoch": 13.483296213808464,
      "loss": 0.6007080078125,
      "loss_ce": 0.00012205771054141223,
      "loss_iou": 0.26171875,
      "loss_num": 0.0157470703125,
      "loss_xval": 0.6015625,
      "num_input_tokens_seen": 339162448,
      "step": 6054
    },
    {
      "epoch": 13.485523385300668,
      "grad_norm": 23.987932205200195,
      "learning_rate": 1e-06,
      "loss": 0.6223,
      "num_input_tokens_seen": 339219908,
      "step": 6055
    },
    {
      "epoch": 13.485523385300668,
      "loss": 0.4472610354423523,
      "loss_ce": 0.00011747775715775788,
      "loss_iou": 0.1953125,
      "loss_num": 0.011474609375,
      "loss_xval": 0.447265625,
      "num_input_tokens_seen": 339219908,
      "step": 6055
    },
    {
      "epoch": 13.487750556792873,
      "grad_norm": 26.649721145629883,
      "learning_rate": 1e-06,
      "loss": 0.3933,
      "num_input_tokens_seen": 339274728,
      "step": 6056
    },
    {
      "epoch": 13.487750556792873,
      "loss": 0.37924766540527344,
      "loss_ce": 9.728018630994484e-05,
      "loss_iou": 0.1640625,
      "loss_num": 0.0101318359375,
      "loss_xval": 0.37890625,
      "num_input_tokens_seen": 339274728,
      "step": 6056
    },
    {
      "epoch": 13.489977728285078,
      "grad_norm": 11.94345760345459,
      "learning_rate": 1e-06,
      "loss": 0.3739,
      "num_input_tokens_seen": 339332392,
      "step": 6057
    },
    {
      "epoch": 13.489977728285078,
      "loss": 0.36925819516181946,
      "loss_ce": 0.00011755910963984206,
      "loss_iou": 0.1728515625,
      "loss_num": 0.004791259765625,
      "loss_xval": 0.369140625,
      "num_input_tokens_seen": 339332392,
      "step": 6057
    },
    {
      "epoch": 13.492204899777283,
      "grad_norm": 19.75101089477539,
      "learning_rate": 1e-06,
      "loss": 0.5577,
      "num_input_tokens_seen": 339389392,
      "step": 6058
    },
    {
      "epoch": 13.492204899777283,
      "loss": 0.7065274715423584,
      "loss_ce": 0.00010661823762347922,
      "loss_iou": 0.279296875,
      "loss_num": 0.0294189453125,
      "loss_xval": 0.70703125,
      "num_input_tokens_seen": 339389392,
      "step": 6058
    },
    {
      "epoch": 13.494432071269488,
      "grad_norm": 15.584753036499023,
      "learning_rate": 1e-06,
      "loss": 0.3834,
      "num_input_tokens_seen": 339446776,
      "step": 6059
    },
    {
      "epoch": 13.494432071269488,
      "loss": 0.4769487977027893,
      "loss_ce": 0.00011162673763465136,
      "loss_iou": 0.1884765625,
      "loss_num": 0.02001953125,
      "loss_xval": 0.4765625,
      "num_input_tokens_seen": 339446776,
      "step": 6059
    },
    {
      "epoch": 13.496659242761693,
      "grad_norm": 17.79234504699707,
      "learning_rate": 1e-06,
      "loss": 0.504,
      "num_input_tokens_seen": 339503364,
      "step": 6060
    },
    {
      "epoch": 13.496659242761693,
      "loss": 0.5225712060928345,
      "loss_ce": 0.00011029178131138906,
      "loss_iou": 0.2001953125,
      "loss_num": 0.0242919921875,
      "loss_xval": 0.5234375,
      "num_input_tokens_seen": 339503364,
      "step": 6060
    },
    {
      "epoch": 13.498886414253898,
      "grad_norm": 16.18769073486328,
      "learning_rate": 1e-06,
      "loss": 0.4275,
      "num_input_tokens_seen": 339561052,
      "step": 6061
    },
    {
      "epoch": 13.498886414253898,
      "loss": 0.40653717517852783,
      "loss_ce": 0.00013457259046845138,
      "loss_iou": 0.166015625,
      "loss_num": 0.0150146484375,
      "loss_xval": 0.40625,
      "num_input_tokens_seen": 339561052,
      "step": 6061
    },
    {
      "epoch": 13.501113585746102,
      "grad_norm": 47.07794952392578,
      "learning_rate": 1e-06,
      "loss": 0.4221,
      "num_input_tokens_seen": 339615028,
      "step": 6062
    },
    {
      "epoch": 13.501113585746102,
      "loss": 0.5137805342674255,
      "loss_ce": 0.00010863743227673694,
      "loss_iou": 0.220703125,
      "loss_num": 0.0145263671875,
      "loss_xval": 0.515625,
      "num_input_tokens_seen": 339615028,
      "step": 6062
    },
    {
      "epoch": 13.503340757238307,
      "grad_norm": 23.532276153564453,
      "learning_rate": 1e-06,
      "loss": 0.5707,
      "num_input_tokens_seen": 339671324,
      "step": 6063
    },
    {
      "epoch": 13.503340757238307,
      "loss": 0.5221333503723145,
      "loss_ce": 0.00016070085985120386,
      "loss_iou": 0.2314453125,
      "loss_num": 0.01171875,
      "loss_xval": 0.5234375,
      "num_input_tokens_seen": 339671324,
      "step": 6063
    },
    {
      "epoch": 13.505567928730512,
      "grad_norm": 15.250123977661133,
      "learning_rate": 1e-06,
      "loss": 0.3295,
      "num_input_tokens_seen": 339726244,
      "step": 6064
    },
    {
      "epoch": 13.505567928730512,
      "loss": 0.3152201473712921,
      "loss_ce": 9.563060302753001e-05,
      "loss_iou": 0.12451171875,
      "loss_num": 0.01318359375,
      "loss_xval": 0.314453125,
      "num_input_tokens_seen": 339726244,
      "step": 6064
    },
    {
      "epoch": 13.507795100222717,
      "grad_norm": 30.90220832824707,
      "learning_rate": 1e-06,
      "loss": 0.6724,
      "num_input_tokens_seen": 339782572,
      "step": 6065
    },
    {
      "epoch": 13.507795100222717,
      "loss": 0.5396183729171753,
      "loss_ce": 0.00012867330224253237,
      "loss_iou": 0.234375,
      "loss_num": 0.0142822265625,
      "loss_xval": 0.5390625,
      "num_input_tokens_seen": 339782572,
      "step": 6065
    },
    {
      "epoch": 13.510022271714922,
      "grad_norm": 23.081607818603516,
      "learning_rate": 1e-06,
      "loss": 0.4508,
      "num_input_tokens_seen": 339836684,
      "step": 6066
    },
    {
      "epoch": 13.510022271714922,
      "loss": 0.37926536798477173,
      "loss_ce": 0.00011500762775540352,
      "loss_iou": 0.1611328125,
      "loss_num": 0.01123046875,
      "loss_xval": 0.37890625,
      "num_input_tokens_seen": 339836684,
      "step": 6066
    },
    {
      "epoch": 13.512249443207127,
      "grad_norm": 16.77637481689453,
      "learning_rate": 1e-06,
      "loss": 0.4764,
      "num_input_tokens_seen": 339895292,
      "step": 6067
    },
    {
      "epoch": 13.512249443207127,
      "loss": 0.5568765997886658,
      "loss_ce": 0.00011388568964321166,
      "loss_iou": 0.2451171875,
      "loss_num": 0.01318359375,
      "loss_xval": 0.55859375,
      "num_input_tokens_seen": 339895292,
      "step": 6067
    },
    {
      "epoch": 13.514476614699332,
      "grad_norm": 14.442283630371094,
      "learning_rate": 1e-06,
      "loss": 0.3909,
      "num_input_tokens_seen": 339952572,
      "step": 6068
    },
    {
      "epoch": 13.514476614699332,
      "loss": 0.33204570412635803,
      "loss_ce": 0.000136529139126651,
      "loss_iou": 0.13671875,
      "loss_num": 0.011474609375,
      "loss_xval": 0.33203125,
      "num_input_tokens_seen": 339952572,
      "step": 6068
    },
    {
      "epoch": 13.516703786191536,
      "grad_norm": 28.08944320678711,
      "learning_rate": 1e-06,
      "loss": 0.4879,
      "num_input_tokens_seen": 340010084,
      "step": 6069
    },
    {
      "epoch": 13.516703786191536,
      "loss": 0.3379353880882263,
      "loss_ce": 0.00010579422814771533,
      "loss_iou": 0.142578125,
      "loss_num": 0.01055908203125,
      "loss_xval": 0.337890625,
      "num_input_tokens_seen": 340010084,
      "step": 6069
    },
    {
      "epoch": 13.518930957683741,
      "grad_norm": 17.983135223388672,
      "learning_rate": 1e-06,
      "loss": 0.4794,
      "num_input_tokens_seen": 340068108,
      "step": 6070
    },
    {
      "epoch": 13.518930957683741,
      "loss": 0.3784530460834503,
      "loss_ce": 9.612538269720972e-05,
      "loss_iou": 0.1552734375,
      "loss_num": 0.01373291015625,
      "loss_xval": 0.37890625,
      "num_input_tokens_seen": 340068108,
      "step": 6070
    },
    {
      "epoch": 13.521158129175946,
      "grad_norm": 31.96721076965332,
      "learning_rate": 1e-06,
      "loss": 0.4647,
      "num_input_tokens_seen": 340125068,
      "step": 6071
    },
    {
      "epoch": 13.521158129175946,
      "loss": 0.4451667070388794,
      "loss_ce": 9.836716344580054e-05,
      "loss_iou": 0.1875,
      "loss_num": 0.0140380859375,
      "loss_xval": 0.4453125,
      "num_input_tokens_seen": 340125068,
      "step": 6071
    },
    {
      "epoch": 13.523385300668151,
      "grad_norm": 13.225302696228027,
      "learning_rate": 1e-06,
      "loss": 0.4289,
      "num_input_tokens_seen": 340183096,
      "step": 6072
    },
    {
      "epoch": 13.523385300668151,
      "loss": 0.4312567710876465,
      "loss_ce": 0.00010440793994348496,
      "loss_iou": 0.19140625,
      "loss_num": 0.00970458984375,
      "loss_xval": 0.431640625,
      "num_input_tokens_seen": 340183096,
      "step": 6072
    },
    {
      "epoch": 13.525612472160356,
      "grad_norm": 11.266600608825684,
      "learning_rate": 1e-06,
      "loss": 0.2725,
      "num_input_tokens_seen": 340239536,
      "step": 6073
    },
    {
      "epoch": 13.525612472160356,
      "loss": 0.353848934173584,
      "loss_ce": 0.00011968802573392168,
      "loss_iou": 0.1572265625,
      "loss_num": 0.00787353515625,
      "loss_xval": 0.353515625,
      "num_input_tokens_seen": 340239536,
      "step": 6073
    },
    {
      "epoch": 13.52783964365256,
      "grad_norm": 17.064437866210938,
      "learning_rate": 1e-06,
      "loss": 0.4341,
      "num_input_tokens_seen": 340295956,
      "step": 6074
    },
    {
      "epoch": 13.52783964365256,
      "loss": 0.6365495920181274,
      "loss_ce": 0.00013600349484477192,
      "loss_iou": 0.25390625,
      "loss_num": 0.0255126953125,
      "loss_xval": 0.63671875,
      "num_input_tokens_seen": 340295956,
      "step": 6074
    },
    {
      "epoch": 13.530066815144766,
      "grad_norm": 22.03349494934082,
      "learning_rate": 1e-06,
      "loss": 0.4328,
      "num_input_tokens_seen": 340353864,
      "step": 6075
    },
    {
      "epoch": 13.530066815144766,
      "loss": 0.44873154163360596,
      "loss_ce": 0.00012314703781157732,
      "loss_iou": 0.2119140625,
      "loss_num": 0.005126953125,
      "loss_xval": 0.44921875,
      "num_input_tokens_seen": 340353864,
      "step": 6075
    },
    {
      "epoch": 13.53229398663697,
      "grad_norm": 19.127866744995117,
      "learning_rate": 1e-06,
      "loss": 0.6748,
      "num_input_tokens_seen": 340407388,
      "step": 6076
    },
    {
      "epoch": 13.53229398663697,
      "loss": 0.7257640361785889,
      "loss_ce": 0.0001475829049013555,
      "loss_iou": 0.26953125,
      "loss_num": 0.03759765625,
      "loss_xval": 0.7265625,
      "num_input_tokens_seen": 340407388,
      "step": 6076
    },
    {
      "epoch": 13.534521158129175,
      "grad_norm": 21.562355041503906,
      "learning_rate": 1e-06,
      "loss": 0.5317,
      "num_input_tokens_seen": 340463944,
      "step": 6077
    },
    {
      "epoch": 13.534521158129175,
      "loss": 0.5379794836044312,
      "loss_ce": 0.00013768361532129347,
      "loss_iou": 0.2255859375,
      "loss_num": 0.017333984375,
      "loss_xval": 0.5390625,
      "num_input_tokens_seen": 340463944,
      "step": 6077
    },
    {
      "epoch": 13.53674832962138,
      "grad_norm": 18.547332763671875,
      "learning_rate": 1e-06,
      "loss": 0.4863,
      "num_input_tokens_seen": 340516692,
      "step": 6078
    },
    {
      "epoch": 13.53674832962138,
      "loss": 0.3476347327232361,
      "loss_ce": 0.0001005365265882574,
      "loss_iou": 0.16015625,
      "loss_num": 0.005523681640625,
      "loss_xval": 0.34765625,
      "num_input_tokens_seen": 340516692,
      "step": 6078
    },
    {
      "epoch": 13.538975501113585,
      "grad_norm": 29.104293823242188,
      "learning_rate": 1e-06,
      "loss": 0.5956,
      "num_input_tokens_seen": 340571300,
      "step": 6079
    },
    {
      "epoch": 13.538975501113585,
      "loss": 0.6296223402023315,
      "loss_ce": 0.00010570493031991646,
      "loss_iou": 0.2578125,
      "loss_num": 0.02294921875,
      "loss_xval": 0.62890625,
      "num_input_tokens_seen": 340571300,
      "step": 6079
    },
    {
      "epoch": 13.54120267260579,
      "grad_norm": 21.421527862548828,
      "learning_rate": 1e-06,
      "loss": 0.4915,
      "num_input_tokens_seen": 340628576,
      "step": 6080
    },
    {
      "epoch": 13.54120267260579,
      "loss": 0.39269521832466125,
      "loss_ce": 0.00011710192484315485,
      "loss_iou": 0.1826171875,
      "loss_num": 0.00543212890625,
      "loss_xval": 0.392578125,
      "num_input_tokens_seen": 340628576,
      "step": 6080
    },
    {
      "epoch": 13.543429844097995,
      "grad_norm": 20.787569046020508,
      "learning_rate": 1e-06,
      "loss": 0.5235,
      "num_input_tokens_seen": 340685408,
      "step": 6081
    },
    {
      "epoch": 13.543429844097995,
      "loss": 0.7552393078804016,
      "loss_ce": 0.0008448001462966204,
      "loss_iou": 0.30859375,
      "loss_num": 0.027099609375,
      "loss_xval": 0.75390625,
      "num_input_tokens_seen": 340685408,
      "step": 6081
    },
    {
      "epoch": 13.5456570155902,
      "grad_norm": 18.21969985961914,
      "learning_rate": 1e-06,
      "loss": 0.5217,
      "num_input_tokens_seen": 340741140,
      "step": 6082
    },
    {
      "epoch": 13.5456570155902,
      "loss": 0.579943060874939,
      "loss_ce": 0.00010910046694334596,
      "loss_iou": 0.25390625,
      "loss_num": 0.01416015625,
      "loss_xval": 0.578125,
      "num_input_tokens_seen": 340741140,
      "step": 6082
    },
    {
      "epoch": 13.547884187082406,
      "grad_norm": 22.53654670715332,
      "learning_rate": 1e-06,
      "loss": 0.4789,
      "num_input_tokens_seen": 340796904,
      "step": 6083
    },
    {
      "epoch": 13.547884187082406,
      "loss": 0.3791908621788025,
      "loss_ce": 0.00010148352885153145,
      "loss_iou": 0.171875,
      "loss_num": 0.006927490234375,
      "loss_xval": 0.37890625,
      "num_input_tokens_seen": 340796904,
      "step": 6083
    },
    {
      "epoch": 13.550111358574611,
      "grad_norm": 25.32193946838379,
      "learning_rate": 1e-06,
      "loss": 0.3296,
      "num_input_tokens_seen": 340852732,
      "step": 6084
    },
    {
      "epoch": 13.550111358574611,
      "loss": 0.39512139558792114,
      "loss_ce": 0.00010186532745137811,
      "loss_iou": 0.1689453125,
      "loss_num": 0.0115966796875,
      "loss_xval": 0.39453125,
      "num_input_tokens_seen": 340852732,
      "step": 6084
    },
    {
      "epoch": 13.552338530066816,
      "grad_norm": 16.27466583251953,
      "learning_rate": 1e-06,
      "loss": 0.3523,
      "num_input_tokens_seen": 340909556,
      "step": 6085
    },
    {
      "epoch": 13.552338530066816,
      "loss": 0.38939177989959717,
      "loss_ce": 0.00010956230107694864,
      "loss_iou": 0.1806640625,
      "loss_num": 0.005706787109375,
      "loss_xval": 0.388671875,
      "num_input_tokens_seen": 340909556,
      "step": 6085
    },
    {
      "epoch": 13.55456570155902,
      "grad_norm": 82.14557647705078,
      "learning_rate": 1e-06,
      "loss": 0.5106,
      "num_input_tokens_seen": 340967244,
      "step": 6086
    },
    {
      "epoch": 13.55456570155902,
      "loss": 0.40943825244903564,
      "loss_ce": 0.00013651512563228607,
      "loss_iou": 0.1728515625,
      "loss_num": 0.0126953125,
      "loss_xval": 0.41015625,
      "num_input_tokens_seen": 340967244,
      "step": 6086
    },
    {
      "epoch": 13.556792873051226,
      "grad_norm": 24.278966903686523,
      "learning_rate": 1e-06,
      "loss": 0.3729,
      "num_input_tokens_seen": 341022336,
      "step": 6087
    },
    {
      "epoch": 13.556792873051226,
      "loss": 0.3812282681465149,
      "loss_ce": 0.00012474997492972761,
      "loss_iou": 0.1689453125,
      "loss_num": 0.0086669921875,
      "loss_xval": 0.380859375,
      "num_input_tokens_seen": 341022336,
      "step": 6087
    },
    {
      "epoch": 13.55902004454343,
      "grad_norm": 17.74454689025879,
      "learning_rate": 1e-06,
      "loss": 0.5962,
      "num_input_tokens_seen": 341078164,
      "step": 6088
    },
    {
      "epoch": 13.55902004454343,
      "loss": 0.34824979305267334,
      "loss_ce": 0.00010525665857130662,
      "loss_iou": 0.150390625,
      "loss_num": 0.0096435546875,
      "loss_xval": 0.34765625,
      "num_input_tokens_seen": 341078164,
      "step": 6088
    },
    {
      "epoch": 13.561247216035635,
      "grad_norm": 15.242683410644531,
      "learning_rate": 1e-06,
      "loss": 0.3475,
      "num_input_tokens_seen": 341135916,
      "step": 6089
    },
    {
      "epoch": 13.561247216035635,
      "loss": 0.34324419498443604,
      "loss_ce": 0.00010453617142047733,
      "loss_iou": 0.158203125,
      "loss_num": 0.005462646484375,
      "loss_xval": 0.34375,
      "num_input_tokens_seen": 341135916,
      "step": 6089
    },
    {
      "epoch": 13.56347438752784,
      "grad_norm": 19.24591827392578,
      "learning_rate": 1e-06,
      "loss": 0.5041,
      "num_input_tokens_seen": 341190420,
      "step": 6090
    },
    {
      "epoch": 13.56347438752784,
      "loss": 0.36112260818481445,
      "loss_ce": 0.00016069506818894297,
      "loss_iou": 0.1708984375,
      "loss_num": 0.003875732421875,
      "loss_xval": 0.361328125,
      "num_input_tokens_seen": 341190420,
      "step": 6090
    },
    {
      "epoch": 13.565701559020045,
      "grad_norm": 20.001699447631836,
      "learning_rate": 1e-06,
      "loss": 0.7154,
      "num_input_tokens_seen": 341245680,
      "step": 6091
    },
    {
      "epoch": 13.565701559020045,
      "loss": 0.670586109161377,
      "loss_ce": 0.00017597324040252715,
      "loss_iou": 0.2890625,
      "loss_num": 0.0184326171875,
      "loss_xval": 0.671875,
      "num_input_tokens_seen": 341245680,
      "step": 6091
    },
    {
      "epoch": 13.56792873051225,
      "grad_norm": 21.868173599243164,
      "learning_rate": 1e-06,
      "loss": 0.4888,
      "num_input_tokens_seen": 341299816,
      "step": 6092
    },
    {
      "epoch": 13.56792873051225,
      "loss": 0.49984219670295715,
      "loss_ce": 0.00011683723278110847,
      "loss_iou": 0.21484375,
      "loss_num": 0.01397705078125,
      "loss_xval": 0.5,
      "num_input_tokens_seen": 341299816,
      "step": 6092
    },
    {
      "epoch": 13.570155902004455,
      "grad_norm": 26.928741455078125,
      "learning_rate": 1e-06,
      "loss": 0.4256,
      "num_input_tokens_seen": 341351004,
      "step": 6093
    },
    {
      "epoch": 13.570155902004455,
      "loss": 0.5890183448791504,
      "loss_ce": 0.00015114745474420488,
      "loss_iou": 0.2392578125,
      "loss_num": 0.0220947265625,
      "loss_xval": 0.58984375,
      "num_input_tokens_seen": 341351004,
      "step": 6093
    },
    {
      "epoch": 13.57238307349666,
      "grad_norm": 15.756858825683594,
      "learning_rate": 1e-06,
      "loss": 0.4628,
      "num_input_tokens_seen": 341406828,
      "step": 6094
    },
    {
      "epoch": 13.57238307349666,
      "loss": 0.608989953994751,
      "loss_ce": 0.00010319190187146887,
      "loss_iou": 0.259765625,
      "loss_num": 0.01806640625,
      "loss_xval": 0.609375,
      "num_input_tokens_seen": 341406828,
      "step": 6094
    },
    {
      "epoch": 13.574610244988865,
      "grad_norm": 16.31437110900879,
      "learning_rate": 1e-06,
      "loss": 0.5845,
      "num_input_tokens_seen": 341463276,
      "step": 6095
    },
    {
      "epoch": 13.574610244988865,
      "loss": 0.5452686548233032,
      "loss_ce": 0.0001026621539494954,
      "loss_iou": 0.2412109375,
      "loss_num": 0.01263427734375,
      "loss_xval": 0.546875,
      "num_input_tokens_seen": 341463276,
      "step": 6095
    },
    {
      "epoch": 13.57683741648107,
      "grad_norm": 22.215185165405273,
      "learning_rate": 1e-06,
      "loss": 0.4455,
      "num_input_tokens_seen": 341520420,
      "step": 6096
    },
    {
      "epoch": 13.57683741648107,
      "loss": 0.4676268398761749,
      "loss_ce": 9.754978236742318e-05,
      "loss_iou": 0.208984375,
      "loss_num": 0.00994873046875,
      "loss_xval": 0.466796875,
      "num_input_tokens_seen": 341520420,
      "step": 6096
    },
    {
      "epoch": 13.579064587973274,
      "grad_norm": 15.833555221557617,
      "learning_rate": 1e-06,
      "loss": 0.5093,
      "num_input_tokens_seen": 341577448,
      "step": 6097
    },
    {
      "epoch": 13.579064587973274,
      "loss": 0.5223226547241211,
      "loss_ce": 0.00010583880066405982,
      "loss_iou": 0.2314453125,
      "loss_num": 0.0118408203125,
      "loss_xval": 0.5234375,
      "num_input_tokens_seen": 341577448,
      "step": 6097
    },
    {
      "epoch": 13.58129175946548,
      "grad_norm": 22.064420700073242,
      "learning_rate": 1e-06,
      "loss": 0.4894,
      "num_input_tokens_seen": 341634480,
      "step": 6098
    },
    {
      "epoch": 13.58129175946548,
      "loss": 0.40562325716018677,
      "loss_ce": 0.00010569434380158782,
      "loss_iou": 0.1796875,
      "loss_num": 0.00909423828125,
      "loss_xval": 0.40625,
      "num_input_tokens_seen": 341634480,
      "step": 6098
    },
    {
      "epoch": 13.583518930957684,
      "grad_norm": 16.516891479492188,
      "learning_rate": 1e-06,
      "loss": 0.4845,
      "num_input_tokens_seen": 341690908,
      "step": 6099
    },
    {
      "epoch": 13.583518930957684,
      "loss": 0.36192750930786133,
      "loss_ce": 0.00011110490595456213,
      "loss_iou": 0.1591796875,
      "loss_num": 0.00885009765625,
      "loss_xval": 0.361328125,
      "num_input_tokens_seen": 341690908,
      "step": 6099
    },
    {
      "epoch": 13.585746102449889,
      "grad_norm": 30.61893081665039,
      "learning_rate": 1e-06,
      "loss": 0.4235,
      "num_input_tokens_seen": 341744072,
      "step": 6100
    },
    {
      "epoch": 13.585746102449889,
      "loss": 0.38622716069221497,
      "loss_ce": 0.0001187745074275881,
      "loss_iou": 0.1728515625,
      "loss_num": 0.008056640625,
      "loss_xval": 0.38671875,
      "num_input_tokens_seen": 341744072,
      "step": 6100
    },
    {
      "epoch": 13.587973273942094,
      "grad_norm": 14.904489517211914,
      "learning_rate": 1e-06,
      "loss": 0.3583,
      "num_input_tokens_seen": 341802688,
      "step": 6101
    },
    {
      "epoch": 13.587973273942094,
      "loss": 0.3524148464202881,
      "loss_ce": 0.00011993409134447575,
      "loss_iou": 0.1640625,
      "loss_num": 0.004791259765625,
      "loss_xval": 0.3515625,
      "num_input_tokens_seen": 341802688,
      "step": 6101
    },
    {
      "epoch": 13.590200445434299,
      "grad_norm": 24.03605842590332,
      "learning_rate": 1e-06,
      "loss": 0.4686,
      "num_input_tokens_seen": 341857712,
      "step": 6102
    },
    {
      "epoch": 13.590200445434299,
      "loss": 0.44786059856414795,
      "loss_ce": 0.00022875834838487208,
      "loss_iou": 0.1904296875,
      "loss_num": 0.0133056640625,
      "loss_xval": 0.447265625,
      "num_input_tokens_seen": 341857712,
      "step": 6102
    },
    {
      "epoch": 13.592427616926503,
      "grad_norm": 41.318817138671875,
      "learning_rate": 1e-06,
      "loss": 0.4597,
      "num_input_tokens_seen": 341911152,
      "step": 6103
    },
    {
      "epoch": 13.592427616926503,
      "loss": 0.3626706302165985,
      "loss_ce": 0.0001218135585077107,
      "loss_iou": 0.1650390625,
      "loss_num": 0.006591796875,
      "loss_xval": 0.36328125,
      "num_input_tokens_seen": 341911152,
      "step": 6103
    },
    {
      "epoch": 13.594654788418708,
      "grad_norm": 27.05071449279785,
      "learning_rate": 1e-06,
      "loss": 0.4711,
      "num_input_tokens_seen": 341965840,
      "step": 6104
    },
    {
      "epoch": 13.594654788418708,
      "loss": 0.4232081174850464,
      "loss_ce": 0.00011239905143156648,
      "loss_iou": 0.181640625,
      "loss_num": 0.0118408203125,
      "loss_xval": 0.423828125,
      "num_input_tokens_seen": 341965840,
      "step": 6104
    },
    {
      "epoch": 13.596881959910913,
      "grad_norm": 16.308061599731445,
      "learning_rate": 1e-06,
      "loss": 0.4512,
      "num_input_tokens_seen": 342024032,
      "step": 6105
    },
    {
      "epoch": 13.596881959910913,
      "loss": 0.41770923137664795,
      "loss_ce": 0.0001066841505235061,
      "loss_iou": 0.1845703125,
      "loss_num": 0.00958251953125,
      "loss_xval": 0.41796875,
      "num_input_tokens_seen": 342024032,
      "step": 6105
    },
    {
      "epoch": 13.599109131403118,
      "grad_norm": 15.531981468200684,
      "learning_rate": 1e-06,
      "loss": 0.3986,
      "num_input_tokens_seen": 342078468,
      "step": 6106
    },
    {
      "epoch": 13.599109131403118,
      "loss": 0.4702088534832001,
      "loss_ce": 0.00011607841588556767,
      "loss_iou": 0.20703125,
      "loss_num": 0.01104736328125,
      "loss_xval": 0.470703125,
      "num_input_tokens_seen": 342078468,
      "step": 6106
    },
    {
      "epoch": 13.601336302895323,
      "grad_norm": 21.346141815185547,
      "learning_rate": 1e-06,
      "loss": 0.4758,
      "num_input_tokens_seen": 342134352,
      "step": 6107
    },
    {
      "epoch": 13.601336302895323,
      "loss": 0.43420565128326416,
      "loss_ce": 0.00012361952394712716,
      "loss_iou": 0.1923828125,
      "loss_num": 0.009765625,
      "loss_xval": 0.43359375,
      "num_input_tokens_seen": 342134352,
      "step": 6107
    },
    {
      "epoch": 13.603563474387528,
      "grad_norm": 18.531774520874023,
      "learning_rate": 1e-06,
      "loss": 0.4012,
      "num_input_tokens_seen": 342192804,
      "step": 6108
    },
    {
      "epoch": 13.603563474387528,
      "loss": 0.3070923388004303,
      "loss_ce": 8.548818004783243e-05,
      "loss_iou": 0.13671875,
      "loss_num": 0.00689697265625,
      "loss_xval": 0.306640625,
      "num_input_tokens_seen": 342192804,
      "step": 6108
    },
    {
      "epoch": 13.605790645879733,
      "grad_norm": 26.71335792541504,
      "learning_rate": 1e-06,
      "loss": 0.5715,
      "num_input_tokens_seen": 342249992,
      "step": 6109
    },
    {
      "epoch": 13.605790645879733,
      "loss": 0.5036748051643372,
      "loss_ce": 0.00013476383173838258,
      "loss_iou": 0.203125,
      "loss_num": 0.0194091796875,
      "loss_xval": 0.50390625,
      "num_input_tokens_seen": 342249992,
      "step": 6109
    },
    {
      "epoch": 13.608017817371937,
      "grad_norm": 18.72128677368164,
      "learning_rate": 1e-06,
      "loss": 0.4366,
      "num_input_tokens_seen": 342306096,
      "step": 6110
    },
    {
      "epoch": 13.608017817371937,
      "loss": 0.39169156551361084,
      "loss_ce": 0.0001129057418438606,
      "loss_iou": 0.1787109375,
      "loss_num": 0.00665283203125,
      "loss_xval": 0.390625,
      "num_input_tokens_seen": 342306096,
      "step": 6110
    },
    {
      "epoch": 13.610244988864142,
      "grad_norm": 19.031667709350586,
      "learning_rate": 1e-06,
      "loss": 0.4382,
      "num_input_tokens_seen": 342360516,
      "step": 6111
    },
    {
      "epoch": 13.610244988864142,
      "loss": 0.417463481426239,
      "loss_ce": 0.00010505890531931072,
      "loss_iou": 0.181640625,
      "loss_num": 0.0108642578125,
      "loss_xval": 0.41796875,
      "num_input_tokens_seen": 342360516,
      "step": 6111
    },
    {
      "epoch": 13.612472160356347,
      "grad_norm": 16.581167221069336,
      "learning_rate": 1e-06,
      "loss": 0.4032,
      "num_input_tokens_seen": 342419136,
      "step": 6112
    },
    {
      "epoch": 13.612472160356347,
      "loss": 0.46885186433792114,
      "loss_ce": 0.00010187575389863923,
      "loss_iou": 0.19140625,
      "loss_num": 0.0174560546875,
      "loss_xval": 0.46875,
      "num_input_tokens_seen": 342419136,
      "step": 6112
    },
    {
      "epoch": 13.614699331848552,
      "grad_norm": 21.586606979370117,
      "learning_rate": 1e-06,
      "loss": 0.4784,
      "num_input_tokens_seen": 342474436,
      "step": 6113
    },
    {
      "epoch": 13.614699331848552,
      "loss": 0.6273195743560791,
      "loss_ce": 0.00024438908440060914,
      "loss_iou": 0.26171875,
      "loss_num": 0.02099609375,
      "loss_xval": 0.62890625,
      "num_input_tokens_seen": 342474436,
      "step": 6113
    },
    {
      "epoch": 13.616926503340757,
      "grad_norm": 19.1909122467041,
      "learning_rate": 1e-06,
      "loss": 0.4076,
      "num_input_tokens_seen": 342531496,
      "step": 6114
    },
    {
      "epoch": 13.616926503340757,
      "loss": 0.5540522933006287,
      "loss_ce": 9.722121467348188e-05,
      "loss_iou": 0.2451171875,
      "loss_num": 0.0125732421875,
      "loss_xval": 0.5546875,
      "num_input_tokens_seen": 342531496,
      "step": 6114
    },
    {
      "epoch": 13.619153674832962,
      "grad_norm": 15.85007381439209,
      "learning_rate": 1e-06,
      "loss": 0.4735,
      "num_input_tokens_seen": 342587152,
      "step": 6115
    },
    {
      "epoch": 13.619153674832962,
      "loss": 0.47263315320014954,
      "loss_ce": 0.000709333224222064,
      "loss_iou": 0.19921875,
      "loss_num": 0.014892578125,
      "loss_xval": 0.47265625,
      "num_input_tokens_seen": 342587152,
      "step": 6115
    },
    {
      "epoch": 13.621380846325167,
      "grad_norm": 18.876895904541016,
      "learning_rate": 1e-06,
      "loss": 0.4915,
      "num_input_tokens_seen": 342642656,
      "step": 6116
    },
    {
      "epoch": 13.621380846325167,
      "loss": 0.5616579651832581,
      "loss_ce": 0.0001345211494481191,
      "loss_iou": 0.2392578125,
      "loss_num": 0.016845703125,
      "loss_xval": 0.5625,
      "num_input_tokens_seen": 342642656,
      "step": 6116
    },
    {
      "epoch": 13.623608017817372,
      "grad_norm": 18.534420013427734,
      "learning_rate": 1e-06,
      "loss": 0.4426,
      "num_input_tokens_seen": 342696908,
      "step": 6117
    },
    {
      "epoch": 13.623608017817372,
      "loss": 0.55345219373703,
      "loss_ce": 0.00010746420593932271,
      "loss_iou": 0.2421875,
      "loss_num": 0.01373291015625,
      "loss_xval": 0.5546875,
      "num_input_tokens_seen": 342696908,
      "step": 6117
    },
    {
      "epoch": 13.625835189309576,
      "grad_norm": 13.372831344604492,
      "learning_rate": 1e-06,
      "loss": 0.4674,
      "num_input_tokens_seen": 342754328,
      "step": 6118
    },
    {
      "epoch": 13.625835189309576,
      "loss": 0.5626158714294434,
      "loss_ce": 0.00011584434105316177,
      "loss_iou": 0.216796875,
      "loss_num": 0.0257568359375,
      "loss_xval": 0.5625,
      "num_input_tokens_seen": 342754328,
      "step": 6118
    },
    {
      "epoch": 13.628062360801781,
      "grad_norm": 34.681114196777344,
      "learning_rate": 1e-06,
      "loss": 0.5469,
      "num_input_tokens_seen": 342812480,
      "step": 6119
    },
    {
      "epoch": 13.628062360801781,
      "loss": 0.5858219861984253,
      "loss_ce": 0.000128652696730569,
      "loss_iou": 0.259765625,
      "loss_num": 0.01300048828125,
      "loss_xval": 0.5859375,
      "num_input_tokens_seen": 342812480,
      "step": 6119
    },
    {
      "epoch": 13.630289532293986,
      "grad_norm": 20.8580265045166,
      "learning_rate": 1e-06,
      "loss": 0.5921,
      "num_input_tokens_seen": 342868916,
      "step": 6120
    },
    {
      "epoch": 13.630289532293986,
      "loss": 0.49220043420791626,
      "loss_ce": 0.00013499979104381055,
      "loss_iou": 0.2275390625,
      "loss_num": 0.00750732421875,
      "loss_xval": 0.4921875,
      "num_input_tokens_seen": 342868916,
      "step": 6120
    },
    {
      "epoch": 13.632516703786191,
      "grad_norm": 28.065649032592773,
      "learning_rate": 1e-06,
      "loss": 0.4325,
      "num_input_tokens_seen": 342923592,
      "step": 6121
    },
    {
      "epoch": 13.632516703786191,
      "loss": 0.41245564818382263,
      "loss_ce": 0.00010210397886112332,
      "loss_iou": 0.189453125,
      "loss_num": 0.006622314453125,
      "loss_xval": 0.412109375,
      "num_input_tokens_seen": 342923592,
      "step": 6121
    },
    {
      "epoch": 13.634743875278396,
      "grad_norm": 21.294170379638672,
      "learning_rate": 1e-06,
      "loss": 0.4094,
      "num_input_tokens_seen": 342978364,
      "step": 6122
    },
    {
      "epoch": 13.634743875278396,
      "loss": 0.43907153606414795,
      "loss_ce": 0.00010670385381672531,
      "loss_iou": 0.19921875,
      "loss_num": 0.00823974609375,
      "loss_xval": 0.439453125,
      "num_input_tokens_seen": 342978364,
      "step": 6122
    },
    {
      "epoch": 13.6369710467706,
      "grad_norm": 15.803836822509766,
      "learning_rate": 1e-06,
      "loss": 0.6057,
      "num_input_tokens_seen": 343034688,
      "step": 6123
    },
    {
      "epoch": 13.6369710467706,
      "loss": 0.6668506264686584,
      "loss_ce": 0.00010255983943352476,
      "loss_iou": 0.302734375,
      "loss_num": 0.0120849609375,
      "loss_xval": 0.66796875,
      "num_input_tokens_seen": 343034688,
      "step": 6123
    },
    {
      "epoch": 13.639198218262806,
      "grad_norm": 16.880170822143555,
      "learning_rate": 1e-06,
      "loss": 0.5102,
      "num_input_tokens_seen": 343091224,
      "step": 6124
    },
    {
      "epoch": 13.639198218262806,
      "loss": 0.4283756613731384,
      "loss_ce": 0.0001530120789539069,
      "loss_iou": 0.19140625,
      "loss_num": 0.009033203125,
      "loss_xval": 0.427734375,
      "num_input_tokens_seen": 343091224,
      "step": 6124
    },
    {
      "epoch": 13.64142538975501,
      "grad_norm": 22.758419036865234,
      "learning_rate": 1e-06,
      "loss": 0.5235,
      "num_input_tokens_seen": 343147108,
      "step": 6125
    },
    {
      "epoch": 13.64142538975501,
      "loss": 0.44180482625961304,
      "loss_ce": 0.00015441025607287884,
      "loss_iou": 0.1845703125,
      "loss_num": 0.01446533203125,
      "loss_xval": 0.44140625,
      "num_input_tokens_seen": 343147108,
      "step": 6125
    },
    {
      "epoch": 13.643652561247215,
      "grad_norm": 24.595998764038086,
      "learning_rate": 1e-06,
      "loss": 0.5212,
      "num_input_tokens_seen": 343202528,
      "step": 6126
    },
    {
      "epoch": 13.643652561247215,
      "loss": 0.6394776701927185,
      "loss_ce": 0.0001954361068783328,
      "loss_iou": 0.279296875,
      "loss_num": 0.0164794921875,
      "loss_xval": 0.640625,
      "num_input_tokens_seen": 343202528,
      "step": 6126
    },
    {
      "epoch": 13.64587973273942,
      "grad_norm": 13.369487762451172,
      "learning_rate": 1e-06,
      "loss": 0.383,
      "num_input_tokens_seen": 343259188,
      "step": 6127
    },
    {
      "epoch": 13.64587973273942,
      "loss": 0.451729953289032,
      "loss_ce": 0.0004359965678304434,
      "loss_iou": 0.1826171875,
      "loss_num": 0.01708984375,
      "loss_xval": 0.451171875,
      "num_input_tokens_seen": 343259188,
      "step": 6127
    },
    {
      "epoch": 13.648106904231625,
      "grad_norm": 20.52755355834961,
      "learning_rate": 1e-06,
      "loss": 0.4196,
      "num_input_tokens_seen": 343316188,
      "step": 6128
    },
    {
      "epoch": 13.648106904231625,
      "loss": 0.3270159065723419,
      "loss_ce": 0.00011160006397403777,
      "loss_iou": 0.154296875,
      "loss_num": 0.003692626953125,
      "loss_xval": 0.326171875,
      "num_input_tokens_seen": 343316188,
      "step": 6128
    },
    {
      "epoch": 13.65033407572383,
      "grad_norm": 21.22062873840332,
      "learning_rate": 1e-06,
      "loss": 0.4893,
      "num_input_tokens_seen": 343369164,
      "step": 6129
    },
    {
      "epoch": 13.65033407572383,
      "loss": 0.4859585165977478,
      "loss_ce": 0.00011869698209920898,
      "loss_iou": 0.2158203125,
      "loss_num": 0.01104736328125,
      "loss_xval": 0.486328125,
      "num_input_tokens_seen": 343369164,
      "step": 6129
    },
    {
      "epoch": 13.652561247216035,
      "grad_norm": 21.014507293701172,
      "learning_rate": 1e-06,
      "loss": 0.5389,
      "num_input_tokens_seen": 343425304,
      "step": 6130
    },
    {
      "epoch": 13.652561247216035,
      "loss": 0.4401944875717163,
      "loss_ce": 0.00013100114301778376,
      "loss_iou": 0.19921875,
      "loss_num": 0.0084228515625,
      "loss_xval": 0.439453125,
      "num_input_tokens_seen": 343425304,
      "step": 6130
    },
    {
      "epoch": 13.654788418708241,
      "grad_norm": 25.746784210205078,
      "learning_rate": 1e-06,
      "loss": 0.5463,
      "num_input_tokens_seen": 343483376,
      "step": 6131
    },
    {
      "epoch": 13.654788418708241,
      "loss": 0.7010675668716431,
      "loss_ce": 0.00013978403876535594,
      "loss_iou": 0.302734375,
      "loss_num": 0.0189208984375,
      "loss_xval": 0.69921875,
      "num_input_tokens_seen": 343483376,
      "step": 6131
    },
    {
      "epoch": 13.657015590200446,
      "grad_norm": 13.484420776367188,
      "learning_rate": 1e-06,
      "loss": 0.4436,
      "num_input_tokens_seen": 343540884,
      "step": 6132
    },
    {
      "epoch": 13.657015590200446,
      "loss": 0.44479644298553467,
      "loss_ce": 9.430477803107351e-05,
      "loss_iou": 0.201171875,
      "loss_num": 0.008544921875,
      "loss_xval": 0.4453125,
      "num_input_tokens_seen": 343540884,
      "step": 6132
    },
    {
      "epoch": 13.659242761692651,
      "grad_norm": 27.06049346923828,
      "learning_rate": 1e-06,
      "loss": 0.4747,
      "num_input_tokens_seen": 343592412,
      "step": 6133
    },
    {
      "epoch": 13.659242761692651,
      "loss": 0.40442535281181335,
      "loss_ce": 0.00012847778270952404,
      "loss_iou": 0.173828125,
      "loss_num": 0.0113525390625,
      "loss_xval": 0.404296875,
      "num_input_tokens_seen": 343592412,
      "step": 6133
    },
    {
      "epoch": 13.661469933184856,
      "grad_norm": 17.517444610595703,
      "learning_rate": 1e-06,
      "loss": 0.5275,
      "num_input_tokens_seen": 343646720,
      "step": 6134
    },
    {
      "epoch": 13.661469933184856,
      "loss": 0.44858014583587646,
      "loss_ce": 0.00012434981181286275,
      "loss_iou": 0.201171875,
      "loss_num": 0.009033203125,
      "loss_xval": 0.44921875,
      "num_input_tokens_seen": 343646720,
      "step": 6134
    },
    {
      "epoch": 13.66369710467706,
      "grad_norm": 13.671886444091797,
      "learning_rate": 1e-06,
      "loss": 0.4997,
      "num_input_tokens_seen": 343703044,
      "step": 6135
    },
    {
      "epoch": 13.66369710467706,
      "loss": 0.3881835341453552,
      "loss_ce": 0.00015255186008289456,
      "loss_iou": 0.173828125,
      "loss_num": 0.00799560546875,
      "loss_xval": 0.388671875,
      "num_input_tokens_seen": 343703044,
      "step": 6135
    },
    {
      "epoch": 13.665924276169266,
      "grad_norm": 28.485546112060547,
      "learning_rate": 1e-06,
      "loss": 0.4675,
      "num_input_tokens_seen": 343759188,
      "step": 6136
    },
    {
      "epoch": 13.665924276169266,
      "loss": 0.4605761170387268,
      "loss_ce": 0.00012687427806667984,
      "loss_iou": 0.1904296875,
      "loss_num": 0.0159912109375,
      "loss_xval": 0.4609375,
      "num_input_tokens_seen": 343759188,
      "step": 6136
    },
    {
      "epoch": 13.66815144766147,
      "grad_norm": 23.070680618286133,
      "learning_rate": 1e-06,
      "loss": 0.3344,
      "num_input_tokens_seen": 343815236,
      "step": 6137
    },
    {
      "epoch": 13.66815144766147,
      "loss": 0.29162734746932983,
      "loss_ce": 0.0001234151714015752,
      "loss_iou": 0.1103515625,
      "loss_num": 0.0140380859375,
      "loss_xval": 0.291015625,
      "num_input_tokens_seen": 343815236,
      "step": 6137
    },
    {
      "epoch": 13.670378619153675,
      "grad_norm": 19.009531021118164,
      "learning_rate": 1e-06,
      "loss": 0.6555,
      "num_input_tokens_seen": 343872764,
      "step": 6138
    },
    {
      "epoch": 13.670378619153675,
      "loss": 0.7084763050079346,
      "loss_ce": 0.00010225173900835216,
      "loss_iou": 0.27734375,
      "loss_num": 0.0303955078125,
      "loss_xval": 0.70703125,
      "num_input_tokens_seen": 343872764,
      "step": 6138
    },
    {
      "epoch": 13.67260579064588,
      "grad_norm": 15.859116554260254,
      "learning_rate": 1e-06,
      "loss": 0.5408,
      "num_input_tokens_seen": 343930516,
      "step": 6139
    },
    {
      "epoch": 13.67260579064588,
      "loss": 0.6228233575820923,
      "loss_ce": 0.0001426705566700548,
      "loss_iou": 0.26953125,
      "loss_num": 0.0164794921875,
      "loss_xval": 0.62109375,
      "num_input_tokens_seen": 343930516,
      "step": 6139
    },
    {
      "epoch": 13.674832962138085,
      "grad_norm": 16.675167083740234,
      "learning_rate": 1e-06,
      "loss": 0.6865,
      "num_input_tokens_seen": 343987948,
      "step": 6140
    },
    {
      "epoch": 13.674832962138085,
      "loss": 0.7435585856437683,
      "loss_ce": 0.0013711238279938698,
      "loss_iou": 0.296875,
      "loss_num": 0.0296630859375,
      "loss_xval": 0.7421875,
      "num_input_tokens_seen": 343987948,
      "step": 6140
    },
    {
      "epoch": 13.67706013363029,
      "grad_norm": 18.57615852355957,
      "learning_rate": 1e-06,
      "loss": 0.5157,
      "num_input_tokens_seen": 344043668,
      "step": 6141
    },
    {
      "epoch": 13.67706013363029,
      "loss": 0.5919273495674133,
      "loss_ce": 0.0001304733450524509,
      "loss_iou": 0.2470703125,
      "loss_num": 0.01953125,
      "loss_xval": 0.59375,
      "num_input_tokens_seen": 344043668,
      "step": 6141
    },
    {
      "epoch": 13.679287305122495,
      "grad_norm": 17.68915367126465,
      "learning_rate": 1e-06,
      "loss": 0.3544,
      "num_input_tokens_seen": 344098800,
      "step": 6142
    },
    {
      "epoch": 13.679287305122495,
      "loss": 0.3864895701408386,
      "loss_ce": 0.00013704363664146513,
      "loss_iou": 0.1748046875,
      "loss_num": 0.00738525390625,
      "loss_xval": 0.38671875,
      "num_input_tokens_seen": 344098800,
      "step": 6142
    },
    {
      "epoch": 13.6815144766147,
      "grad_norm": 25.44866371154785,
      "learning_rate": 1e-06,
      "loss": 0.3647,
      "num_input_tokens_seen": 344156784,
      "step": 6143
    },
    {
      "epoch": 13.6815144766147,
      "loss": 0.3855780065059662,
      "loss_ce": 0.00011047557200072333,
      "loss_iou": 0.171875,
      "loss_num": 0.00848388671875,
      "loss_xval": 0.384765625,
      "num_input_tokens_seen": 344156784,
      "step": 6143
    },
    {
      "epoch": 13.683741648106905,
      "grad_norm": 15.047554969787598,
      "learning_rate": 1e-06,
      "loss": 0.3801,
      "num_input_tokens_seen": 344215212,
      "step": 6144
    },
    {
      "epoch": 13.683741648106905,
      "loss": 0.2978753447532654,
      "loss_ce": 8.48290219437331e-05,
      "loss_iou": 0.1298828125,
      "loss_num": 0.007568359375,
      "loss_xval": 0.296875,
      "num_input_tokens_seen": 344215212,
      "step": 6144
    },
    {
      "epoch": 13.68596881959911,
      "grad_norm": 48.20288848876953,
      "learning_rate": 1e-06,
      "loss": 0.4122,
      "num_input_tokens_seen": 344271784,
      "step": 6145
    },
    {
      "epoch": 13.68596881959911,
      "loss": 0.40208396315574646,
      "loss_ce": 0.00010644205030985177,
      "loss_iou": 0.185546875,
      "loss_num": 0.006317138671875,
      "loss_xval": 0.40234375,
      "num_input_tokens_seen": 344271784,
      "step": 6145
    },
    {
      "epoch": 13.688195991091314,
      "grad_norm": 71.81108093261719,
      "learning_rate": 1e-06,
      "loss": 0.576,
      "num_input_tokens_seen": 344329204,
      "step": 6146
    },
    {
      "epoch": 13.688195991091314,
      "loss": 0.5167222023010254,
      "loss_ce": 0.00012068171781720594,
      "loss_iou": 0.2236328125,
      "loss_num": 0.0140380859375,
      "loss_xval": 0.515625,
      "num_input_tokens_seen": 344329204,
      "step": 6146
    },
    {
      "epoch": 13.690423162583519,
      "grad_norm": 12.681405067443848,
      "learning_rate": 1e-06,
      "loss": 0.4676,
      "num_input_tokens_seen": 344385616,
      "step": 6147
    },
    {
      "epoch": 13.690423162583519,
      "loss": 0.5067310929298401,
      "loss_ce": 0.0005054936627857387,
      "loss_iou": 0.1962890625,
      "loss_num": 0.0230712890625,
      "loss_xval": 0.5078125,
      "num_input_tokens_seen": 344385616,
      "step": 6147
    },
    {
      "epoch": 13.692650334075724,
      "grad_norm": 25.468963623046875,
      "learning_rate": 1e-06,
      "loss": 0.3891,
      "num_input_tokens_seen": 344442500,
      "step": 6148
    },
    {
      "epoch": 13.692650334075724,
      "loss": 0.443470299243927,
      "loss_ce": 0.0001109380682464689,
      "loss_iou": 0.1962890625,
      "loss_num": 0.0101318359375,
      "loss_xval": 0.443359375,
      "num_input_tokens_seen": 344442500,
      "step": 6148
    },
    {
      "epoch": 13.694877505567929,
      "grad_norm": 29.40053367614746,
      "learning_rate": 1e-06,
      "loss": 0.5509,
      "num_input_tokens_seen": 344499324,
      "step": 6149
    },
    {
      "epoch": 13.694877505567929,
      "loss": 0.5093941688537598,
      "loss_ce": 0.00011685908248182386,
      "loss_iou": 0.23046875,
      "loss_num": 0.0096435546875,
      "loss_xval": 0.5078125,
      "num_input_tokens_seen": 344499324,
      "step": 6149
    },
    {
      "epoch": 13.697104677060134,
      "grad_norm": 21.182100296020508,
      "learning_rate": 1e-06,
      "loss": 0.4615,
      "num_input_tokens_seen": 344556944,
      "step": 6150
    },
    {
      "epoch": 13.697104677060134,
      "loss": 0.568960428237915,
      "loss_ce": 0.00011279522004770115,
      "loss_iou": 0.23828125,
      "loss_num": 0.0185546875,
      "loss_xval": 0.5703125,
      "num_input_tokens_seen": 344556944,
      "step": 6150
    },
    {
      "epoch": 13.699331848552339,
      "grad_norm": 34.760746002197266,
      "learning_rate": 1e-06,
      "loss": 0.4418,
      "num_input_tokens_seen": 344613508,
      "step": 6151
    },
    {
      "epoch": 13.699331848552339,
      "loss": 0.49267372488975525,
      "loss_ce": 0.00012003096344415098,
      "loss_iou": 0.2138671875,
      "loss_num": 0.01312255859375,
      "loss_xval": 0.4921875,
      "num_input_tokens_seen": 344613508,
      "step": 6151
    },
    {
      "epoch": 13.701559020044543,
      "grad_norm": 34.04802322387695,
      "learning_rate": 1e-06,
      "loss": 0.4694,
      "num_input_tokens_seen": 344669656,
      "step": 6152
    },
    {
      "epoch": 13.701559020044543,
      "loss": 0.38548383116722107,
      "loss_ce": 0.00010786794882733375,
      "loss_iou": 0.1796875,
      "loss_num": 0.00506591796875,
      "loss_xval": 0.384765625,
      "num_input_tokens_seen": 344669656,
      "step": 6152
    },
    {
      "epoch": 13.703786191536748,
      "grad_norm": 19.769567489624023,
      "learning_rate": 1e-06,
      "loss": 0.4088,
      "num_input_tokens_seen": 344725372,
      "step": 6153
    },
    {
      "epoch": 13.703786191536748,
      "loss": 0.49809765815734863,
      "loss_ce": 0.00017283624038100243,
      "loss_iou": 0.216796875,
      "loss_num": 0.01318359375,
      "loss_xval": 0.498046875,
      "num_input_tokens_seen": 344725372,
      "step": 6153
    },
    {
      "epoch": 13.706013363028953,
      "grad_norm": 23.97338104248047,
      "learning_rate": 1e-06,
      "loss": 0.7272,
      "num_input_tokens_seen": 344781164,
      "step": 6154
    },
    {
      "epoch": 13.706013363028953,
      "loss": 0.5699473023414612,
      "loss_ce": 0.0001230589987244457,
      "loss_iou": 0.23828125,
      "loss_num": 0.0186767578125,
      "loss_xval": 0.5703125,
      "num_input_tokens_seen": 344781164,
      "step": 6154
    },
    {
      "epoch": 13.708240534521158,
      "grad_norm": 39.72063064575195,
      "learning_rate": 1e-06,
      "loss": 0.4243,
      "num_input_tokens_seen": 344836304,
      "step": 6155
    },
    {
      "epoch": 13.708240534521158,
      "loss": 0.37404966354370117,
      "loss_ce": 0.00014831856242381036,
      "loss_iou": 0.1630859375,
      "loss_num": 0.00970458984375,
      "loss_xval": 0.373046875,
      "num_input_tokens_seen": 344836304,
      "step": 6155
    },
    {
      "epoch": 13.710467706013363,
      "grad_norm": 23.376707077026367,
      "learning_rate": 1e-06,
      "loss": 0.5827,
      "num_input_tokens_seen": 344894388,
      "step": 6156
    },
    {
      "epoch": 13.710467706013363,
      "loss": 0.42637595534324646,
      "loss_ce": 0.00010641853441484272,
      "loss_iou": 0.189453125,
      "loss_num": 0.00933837890625,
      "loss_xval": 0.42578125,
      "num_input_tokens_seen": 344894388,
      "step": 6156
    },
    {
      "epoch": 13.712694877505568,
      "grad_norm": 13.803214073181152,
      "learning_rate": 1e-06,
      "loss": 0.3646,
      "num_input_tokens_seen": 344951236,
      "step": 6157
    },
    {
      "epoch": 13.712694877505568,
      "loss": 0.3642551600933075,
      "loss_ce": 0.00011941190314246342,
      "loss_iou": 0.1669921875,
      "loss_num": 0.006134033203125,
      "loss_xval": 0.36328125,
      "num_input_tokens_seen": 344951236,
      "step": 6157
    },
    {
      "epoch": 13.714922048997773,
      "grad_norm": 22.018762588500977,
      "learning_rate": 1e-06,
      "loss": 0.6011,
      "num_input_tokens_seen": 345007564,
      "step": 6158
    },
    {
      "epoch": 13.714922048997773,
      "loss": 0.6203541159629822,
      "loss_ce": 0.00011484955030027777,
      "loss_iou": 0.265625,
      "loss_num": 0.0179443359375,
      "loss_xval": 0.62109375,
      "num_input_tokens_seen": 345007564,
      "step": 6158
    },
    {
      "epoch": 13.717149220489977,
      "grad_norm": 17.11638069152832,
      "learning_rate": 1e-06,
      "loss": 0.4277,
      "num_input_tokens_seen": 345064000,
      "step": 6159
    },
    {
      "epoch": 13.717149220489977,
      "loss": 0.5495654344558716,
      "loss_ce": 0.00012692378368228674,
      "loss_iou": 0.2470703125,
      "loss_num": 0.01129150390625,
      "loss_xval": 0.55078125,
      "num_input_tokens_seen": 345064000,
      "step": 6159
    },
    {
      "epoch": 13.719376391982182,
      "grad_norm": 20.523475646972656,
      "learning_rate": 1e-06,
      "loss": 0.3536,
      "num_input_tokens_seen": 345120072,
      "step": 6160
    },
    {
      "epoch": 13.719376391982182,
      "loss": 0.37014520168304443,
      "loss_ce": 0.00027214642614126205,
      "loss_iou": 0.1650390625,
      "loss_num": 0.0081787109375,
      "loss_xval": 0.369140625,
      "num_input_tokens_seen": 345120072,
      "step": 6160
    },
    {
      "epoch": 13.721603563474387,
      "grad_norm": 32.90972900390625,
      "learning_rate": 1e-06,
      "loss": 0.5949,
      "num_input_tokens_seen": 345176544,
      "step": 6161
    },
    {
      "epoch": 13.721603563474387,
      "loss": 0.44534897804260254,
      "loss_ce": 9.752592450240627e-05,
      "loss_iou": 0.18359375,
      "loss_num": 0.0157470703125,
      "loss_xval": 0.4453125,
      "num_input_tokens_seen": 345176544,
      "step": 6161
    },
    {
      "epoch": 13.723830734966592,
      "grad_norm": 18.18287467956543,
      "learning_rate": 1e-06,
      "loss": 0.379,
      "num_input_tokens_seen": 345233380,
      "step": 6162
    },
    {
      "epoch": 13.723830734966592,
      "loss": 0.2666533589363098,
      "loss_ce": 8.995502139441669e-05,
      "loss_iou": 0.11279296875,
      "loss_num": 0.00830078125,
      "loss_xval": 0.265625,
      "num_input_tokens_seen": 345233380,
      "step": 6162
    },
    {
      "epoch": 13.726057906458797,
      "grad_norm": 15.033989906311035,
      "learning_rate": 1e-06,
      "loss": 0.6433,
      "num_input_tokens_seen": 345289992,
      "step": 6163
    },
    {
      "epoch": 13.726057906458797,
      "loss": 0.7077064514160156,
      "loss_ce": 0.0009193975711241364,
      "loss_iou": 0.3046875,
      "loss_num": 0.0194091796875,
      "loss_xval": 0.70703125,
      "num_input_tokens_seen": 345289992,
      "step": 6163
    },
    {
      "epoch": 13.728285077951002,
      "grad_norm": 12.950138092041016,
      "learning_rate": 1e-06,
      "loss": 0.4458,
      "num_input_tokens_seen": 345345848,
      "step": 6164
    },
    {
      "epoch": 13.728285077951002,
      "loss": 0.27292484045028687,
      "loss_ce": 9.76905066636391e-05,
      "loss_iou": 0.11669921875,
      "loss_num": 0.00787353515625,
      "loss_xval": 0.2734375,
      "num_input_tokens_seen": 345345848,
      "step": 6164
    },
    {
      "epoch": 13.730512249443207,
      "grad_norm": 17.856874465942383,
      "learning_rate": 1e-06,
      "loss": 0.4661,
      "num_input_tokens_seen": 345402492,
      "step": 6165
    },
    {
      "epoch": 13.730512249443207,
      "loss": 0.5341655015945435,
      "loss_ce": 0.00022992276353761554,
      "loss_iou": 0.2109375,
      "loss_num": 0.0225830078125,
      "loss_xval": 0.53515625,
      "num_input_tokens_seen": 345402492,
      "step": 6165
    },
    {
      "epoch": 13.732739420935411,
      "grad_norm": 17.676965713500977,
      "learning_rate": 1e-06,
      "loss": 0.4308,
      "num_input_tokens_seen": 345459516,
      "step": 6166
    },
    {
      "epoch": 13.732739420935411,
      "loss": 0.5320949554443359,
      "loss_ce": 0.00011253212142037228,
      "loss_iou": 0.2373046875,
      "loss_num": 0.0113525390625,
      "loss_xval": 0.53125,
      "num_input_tokens_seen": 345459516,
      "step": 6166
    },
    {
      "epoch": 13.734966592427616,
      "grad_norm": 19.01355743408203,
      "learning_rate": 1e-06,
      "loss": 0.4474,
      "num_input_tokens_seen": 345514608,
      "step": 6167
    },
    {
      "epoch": 13.734966592427616,
      "loss": 0.4031931757926941,
      "loss_ce": 0.0001169660608866252,
      "loss_iou": 0.1845703125,
      "loss_num": 0.006866455078125,
      "loss_xval": 0.40234375,
      "num_input_tokens_seen": 345514608,
      "step": 6167
    },
    {
      "epoch": 13.737193763919821,
      "grad_norm": 19.29640769958496,
      "learning_rate": 1e-06,
      "loss": 0.5236,
      "num_input_tokens_seen": 345571524,
      "step": 6168
    },
    {
      "epoch": 13.737193763919821,
      "loss": 0.491804838180542,
      "loss_ce": 0.00010560073133092374,
      "loss_iou": 0.2177734375,
      "loss_num": 0.01123046875,
      "loss_xval": 0.4921875,
      "num_input_tokens_seen": 345571524,
      "step": 6168
    },
    {
      "epoch": 13.739420935412026,
      "grad_norm": 21.25169563293457,
      "learning_rate": 1e-06,
      "loss": 0.4618,
      "num_input_tokens_seen": 345627688,
      "step": 6169
    },
    {
      "epoch": 13.739420935412026,
      "loss": 0.4485178589820862,
      "loss_ce": 0.00012308621080592275,
      "loss_iou": 0.1923828125,
      "loss_num": 0.0126953125,
      "loss_xval": 0.44921875,
      "num_input_tokens_seen": 345627688,
      "step": 6169
    },
    {
      "epoch": 13.74164810690423,
      "grad_norm": 16.865291595458984,
      "learning_rate": 1e-06,
      "loss": 0.5896,
      "num_input_tokens_seen": 345684380,
      "step": 6170
    },
    {
      "epoch": 13.74164810690423,
      "loss": 0.5811623334884644,
      "loss_ce": 0.00010766902414616197,
      "loss_iou": 0.26171875,
      "loss_num": 0.01123046875,
      "loss_xval": 0.58203125,
      "num_input_tokens_seen": 345684380,
      "step": 6170
    },
    {
      "epoch": 13.743875278396436,
      "grad_norm": 20.563676834106445,
      "learning_rate": 1e-06,
      "loss": 0.5899,
      "num_input_tokens_seen": 345740580,
      "step": 6171
    },
    {
      "epoch": 13.743875278396436,
      "loss": 0.44517427682876587,
      "loss_ce": 0.00010591248428681865,
      "loss_iou": 0.1923828125,
      "loss_num": 0.01202392578125,
      "loss_xval": 0.4453125,
      "num_input_tokens_seen": 345740580,
      "step": 6171
    },
    {
      "epoch": 13.74610244988864,
      "grad_norm": 16.448043823242188,
      "learning_rate": 1e-06,
      "loss": 0.5441,
      "num_input_tokens_seen": 345799396,
      "step": 6172
    },
    {
      "epoch": 13.74610244988864,
      "loss": 0.6715624332427979,
      "loss_ce": 0.00011470924073364586,
      "loss_iou": 0.2890625,
      "loss_num": 0.0181884765625,
      "loss_xval": 0.671875,
      "num_input_tokens_seen": 345799396,
      "step": 6172
    },
    {
      "epoch": 13.748329621380847,
      "grad_norm": 47.44057846069336,
      "learning_rate": 1e-06,
      "loss": 0.5772,
      "num_input_tokens_seen": 345855724,
      "step": 6173
    },
    {
      "epoch": 13.748329621380847,
      "loss": 0.7166056632995605,
      "loss_ce": 0.00017500200192444026,
      "loss_iou": 0.322265625,
      "loss_num": 0.0147705078125,
      "loss_xval": 0.71484375,
      "num_input_tokens_seen": 345855724,
      "step": 6173
    },
    {
      "epoch": 13.750556792873052,
      "grad_norm": 16.256362915039062,
      "learning_rate": 1e-06,
      "loss": 0.5361,
      "num_input_tokens_seen": 345912732,
      "step": 6174
    },
    {
      "epoch": 13.750556792873052,
      "loss": 0.4795916676521301,
      "loss_ce": 9.949602099368349e-05,
      "loss_iou": 0.1923828125,
      "loss_num": 0.01904296875,
      "loss_xval": 0.48046875,
      "num_input_tokens_seen": 345912732,
      "step": 6174
    },
    {
      "epoch": 13.752783964365257,
      "grad_norm": 21.9121036529541,
      "learning_rate": 1e-06,
      "loss": 0.495,
      "num_input_tokens_seen": 345968416,
      "step": 6175
    },
    {
      "epoch": 13.752783964365257,
      "loss": 0.3780425786972046,
      "loss_ce": 0.00011291421833448112,
      "loss_iou": 0.169921875,
      "loss_num": 0.007476806640625,
      "loss_xval": 0.37890625,
      "num_input_tokens_seen": 345968416,
      "step": 6175
    },
    {
      "epoch": 13.755011135857462,
      "grad_norm": 15.832091331481934,
      "learning_rate": 1e-06,
      "loss": 0.6457,
      "num_input_tokens_seen": 346026396,
      "step": 6176
    },
    {
      "epoch": 13.755011135857462,
      "loss": 0.4686300754547119,
      "loss_ce": 0.00012421910651028156,
      "loss_iou": 0.2138671875,
      "loss_num": 0.00830078125,
      "loss_xval": 0.46875,
      "num_input_tokens_seen": 346026396,
      "step": 6176
    },
    {
      "epoch": 13.757238307349667,
      "grad_norm": 21.13070297241211,
      "learning_rate": 1e-06,
      "loss": 0.5723,
      "num_input_tokens_seen": 346080884,
      "step": 6177
    },
    {
      "epoch": 13.757238307349667,
      "loss": 0.5670015811920166,
      "loss_ce": 0.00010699567792471498,
      "loss_iou": 0.244140625,
      "loss_num": 0.0159912109375,
      "loss_xval": 0.56640625,
      "num_input_tokens_seen": 346080884,
      "step": 6177
    },
    {
      "epoch": 13.759465478841872,
      "grad_norm": 23.448190689086914,
      "learning_rate": 1e-06,
      "loss": 0.5021,
      "num_input_tokens_seen": 346134556,
      "step": 6178
    },
    {
      "epoch": 13.759465478841872,
      "loss": 0.47252950072288513,
      "loss_ce": 0.00011740469199139625,
      "loss_iou": 0.1923828125,
      "loss_num": 0.017333984375,
      "loss_xval": 0.47265625,
      "num_input_tokens_seen": 346134556,
      "step": 6178
    },
    {
      "epoch": 13.761692650334076,
      "grad_norm": 18.187257766723633,
      "learning_rate": 1e-06,
      "loss": 0.3635,
      "num_input_tokens_seen": 346188980,
      "step": 6179
    },
    {
      "epoch": 13.761692650334076,
      "loss": 0.35398566722869873,
      "loss_ce": 0.00010384367487858981,
      "loss_iou": 0.1630859375,
      "loss_num": 0.005615234375,
      "loss_xval": 0.353515625,
      "num_input_tokens_seen": 346188980,
      "step": 6179
    },
    {
      "epoch": 13.763919821826281,
      "grad_norm": 31.72901153564453,
      "learning_rate": 1e-06,
      "loss": 0.3924,
      "num_input_tokens_seen": 346245600,
      "step": 6180
    },
    {
      "epoch": 13.763919821826281,
      "loss": 0.3832623362541199,
      "loss_ce": 8.365388202946633e-05,
      "loss_iou": 0.173828125,
      "loss_num": 0.007110595703125,
      "loss_xval": 0.3828125,
      "num_input_tokens_seen": 346245600,
      "step": 6180
    },
    {
      "epoch": 13.766146993318486,
      "grad_norm": 21.675914764404297,
      "learning_rate": 1e-06,
      "loss": 0.4688,
      "num_input_tokens_seen": 346299216,
      "step": 6181
    },
    {
      "epoch": 13.766146993318486,
      "loss": 0.5464923977851868,
      "loss_ce": 0.0001056903856806457,
      "loss_iou": 0.216796875,
      "loss_num": 0.0225830078125,
      "loss_xval": 0.546875,
      "num_input_tokens_seen": 346299216,
      "step": 6181
    },
    {
      "epoch": 13.768374164810691,
      "grad_norm": 24.515111923217773,
      "learning_rate": 1e-06,
      "loss": 0.5284,
      "num_input_tokens_seen": 346353404,
      "step": 6182
    },
    {
      "epoch": 13.768374164810691,
      "loss": 0.47796687483787537,
      "loss_ce": 0.00018368265591561794,
      "loss_iou": 0.197265625,
      "loss_num": 0.0164794921875,
      "loss_xval": 0.478515625,
      "num_input_tokens_seen": 346353404,
      "step": 6182
    },
    {
      "epoch": 13.770601336302896,
      "grad_norm": 13.62939453125,
      "learning_rate": 1e-06,
      "loss": 0.4113,
      "num_input_tokens_seen": 346409016,
      "step": 6183
    },
    {
      "epoch": 13.770601336302896,
      "loss": 0.395562082529068,
      "loss_ce": 0.00011531692871358246,
      "loss_iou": 0.1708984375,
      "loss_num": 0.0106201171875,
      "loss_xval": 0.39453125,
      "num_input_tokens_seen": 346409016,
      "step": 6183
    },
    {
      "epoch": 13.7728285077951,
      "grad_norm": 14.132533073425293,
      "learning_rate": 1e-06,
      "loss": 0.418,
      "num_input_tokens_seen": 346464900,
      "step": 6184
    },
    {
      "epoch": 13.7728285077951,
      "loss": 0.416792631149292,
      "loss_ce": 0.00016667114687152207,
      "loss_iou": 0.1923828125,
      "loss_num": 0.006439208984375,
      "loss_xval": 0.416015625,
      "num_input_tokens_seen": 346464900,
      "step": 6184
    },
    {
      "epoch": 13.775055679287306,
      "grad_norm": 78.39266967773438,
      "learning_rate": 1e-06,
      "loss": 0.6477,
      "num_input_tokens_seen": 346521172,
      "step": 6185
    },
    {
      "epoch": 13.775055679287306,
      "loss": 0.45030081272125244,
      "loss_ce": 0.00010552178719080985,
      "loss_iou": 0.197265625,
      "loss_num": 0.01116943359375,
      "loss_xval": 0.44921875,
      "num_input_tokens_seen": 346521172,
      "step": 6185
    },
    {
      "epoch": 13.77728285077951,
      "grad_norm": 23.97130584716797,
      "learning_rate": 1e-06,
      "loss": 0.5591,
      "num_input_tokens_seen": 346576376,
      "step": 6186
    },
    {
      "epoch": 13.77728285077951,
      "loss": 0.7201170921325684,
      "loss_ce": 0.00014645657211076468,
      "loss_iou": 0.322265625,
      "loss_num": 0.0155029296875,
      "loss_xval": 0.71875,
      "num_input_tokens_seen": 346576376,
      "step": 6186
    },
    {
      "epoch": 13.779510022271715,
      "grad_norm": 16.92420196533203,
      "learning_rate": 1e-06,
      "loss": 0.4977,
      "num_input_tokens_seen": 346633420,
      "step": 6187
    },
    {
      "epoch": 13.779510022271715,
      "loss": 0.633420467376709,
      "loss_ce": 0.00011968903709203005,
      "loss_iou": 0.26171875,
      "loss_num": 0.022216796875,
      "loss_xval": 0.6328125,
      "num_input_tokens_seen": 346633420,
      "step": 6187
    },
    {
      "epoch": 13.78173719376392,
      "grad_norm": 18.442733764648438,
      "learning_rate": 1e-06,
      "loss": 0.5796,
      "num_input_tokens_seen": 346689692,
      "step": 6188
    },
    {
      "epoch": 13.78173719376392,
      "loss": 0.7934539318084717,
      "loss_ce": 0.00011891472968272865,
      "loss_iou": 0.345703125,
      "loss_num": 0.020263671875,
      "loss_xval": 0.79296875,
      "num_input_tokens_seen": 346689692,
      "step": 6188
    },
    {
      "epoch": 13.783964365256125,
      "grad_norm": 21.14725685119629,
      "learning_rate": 1e-06,
      "loss": 0.3824,
      "num_input_tokens_seen": 346745212,
      "step": 6189
    },
    {
      "epoch": 13.783964365256125,
      "loss": 0.3489500880241394,
      "loss_ce": 0.00013415844296105206,
      "loss_iou": 0.15625,
      "loss_num": 0.007232666015625,
      "loss_xval": 0.349609375,
      "num_input_tokens_seen": 346745212,
      "step": 6189
    },
    {
      "epoch": 13.78619153674833,
      "grad_norm": 23.13970375061035,
      "learning_rate": 1e-06,
      "loss": 0.3716,
      "num_input_tokens_seen": 346803220,
      "step": 6190
    },
    {
      "epoch": 13.78619153674833,
      "loss": 0.43285852670669556,
      "loss_ce": 0.00011927761079277843,
      "loss_iou": 0.205078125,
      "loss_num": 0.00439453125,
      "loss_xval": 0.43359375,
      "num_input_tokens_seen": 346803220,
      "step": 6190
    },
    {
      "epoch": 13.788418708240535,
      "grad_norm": 17.715566635131836,
      "learning_rate": 1e-06,
      "loss": 0.49,
      "num_input_tokens_seen": 346858040,
      "step": 6191
    },
    {
      "epoch": 13.788418708240535,
      "loss": 0.29737383127212524,
      "loss_ce": 0.0001326015335507691,
      "loss_iou": 0.1328125,
      "loss_num": 0.0062255859375,
      "loss_xval": 0.296875,
      "num_input_tokens_seen": 346858040,
      "step": 6191
    },
    {
      "epoch": 13.79064587973274,
      "grad_norm": 25.321876525878906,
      "learning_rate": 1e-06,
      "loss": 0.5248,
      "num_input_tokens_seen": 346914080,
      "step": 6192
    },
    {
      "epoch": 13.79064587973274,
      "loss": 0.6016849279403687,
      "loss_ce": 0.00010718886915128678,
      "loss_iou": 0.25390625,
      "loss_num": 0.0185546875,
      "loss_xval": 0.6015625,
      "num_input_tokens_seen": 346914080,
      "step": 6192
    },
    {
      "epoch": 13.792873051224944,
      "grad_norm": 15.98505973815918,
      "learning_rate": 1e-06,
      "loss": 0.5626,
      "num_input_tokens_seen": 346967296,
      "step": 6193
    },
    {
      "epoch": 13.792873051224944,
      "loss": 0.677720844745636,
      "loss_ce": 0.0001085417898138985,
      "loss_iou": 0.30078125,
      "loss_num": 0.01507568359375,
      "loss_xval": 0.67578125,
      "num_input_tokens_seen": 346967296,
      "step": 6193
    },
    {
      "epoch": 13.79510022271715,
      "grad_norm": 16.50602149963379,
      "learning_rate": 1e-06,
      "loss": 0.6727,
      "num_input_tokens_seen": 347023416,
      "step": 6194
    },
    {
      "epoch": 13.79510022271715,
      "loss": 0.8356806039810181,
      "loss_ce": 0.00010931311408057809,
      "loss_iou": 0.33203125,
      "loss_num": 0.033935546875,
      "loss_xval": 0.8359375,
      "num_input_tokens_seen": 347023416,
      "step": 6194
    },
    {
      "epoch": 13.797327394209354,
      "grad_norm": 25.20470428466797,
      "learning_rate": 1e-06,
      "loss": 0.4481,
      "num_input_tokens_seen": 347078308,
      "step": 6195
    },
    {
      "epoch": 13.797327394209354,
      "loss": 0.39737868309020996,
      "loss_ce": 0.00016190031601581722,
      "loss_iou": 0.1787109375,
      "loss_num": 0.00799560546875,
      "loss_xval": 0.396484375,
      "num_input_tokens_seen": 347078308,
      "step": 6195
    },
    {
      "epoch": 13.799554565701559,
      "grad_norm": 19.322734832763672,
      "learning_rate": 1e-06,
      "loss": 0.5349,
      "num_input_tokens_seen": 347134956,
      "step": 6196
    },
    {
      "epoch": 13.799554565701559,
      "loss": 0.510852038860321,
      "loss_ce": 0.0007202195120044053,
      "loss_iou": 0.2236328125,
      "loss_num": 0.01239013671875,
      "loss_xval": 0.51171875,
      "num_input_tokens_seen": 347134956,
      "step": 6196
    },
    {
      "epoch": 13.801781737193764,
      "grad_norm": 25.328693389892578,
      "learning_rate": 1e-06,
      "loss": 0.7473,
      "num_input_tokens_seen": 347193060,
      "step": 6197
    },
    {
      "epoch": 13.801781737193764,
      "loss": 0.9841349720954895,
      "loss_ce": 0.00012613809667527676,
      "loss_iou": 0.40234375,
      "loss_num": 0.03564453125,
      "loss_xval": 0.984375,
      "num_input_tokens_seen": 347193060,
      "step": 6197
    },
    {
      "epoch": 13.804008908685969,
      "grad_norm": 17.876684188842773,
      "learning_rate": 1e-06,
      "loss": 0.4347,
      "num_input_tokens_seen": 347247788,
      "step": 6198
    },
    {
      "epoch": 13.804008908685969,
      "loss": 0.5318405628204346,
      "loss_ce": 0.00010229815234197304,
      "loss_iou": 0.2255859375,
      "loss_num": 0.01611328125,
      "loss_xval": 0.53125,
      "num_input_tokens_seen": 347247788,
      "step": 6198
    },
    {
      "epoch": 13.806236080178174,
      "grad_norm": 21.041126251220703,
      "learning_rate": 1e-06,
      "loss": 0.6075,
      "num_input_tokens_seen": 347302380,
      "step": 6199
    },
    {
      "epoch": 13.806236080178174,
      "loss": 0.7548277378082275,
      "loss_ce": 0.00025013202684931457,
      "loss_iou": 0.314453125,
      "loss_num": 0.0257568359375,
      "loss_xval": 0.75390625,
      "num_input_tokens_seen": 347302380,
      "step": 6199
    },
    {
      "epoch": 13.808463251670378,
      "grad_norm": 14.308113098144531,
      "learning_rate": 1e-06,
      "loss": 0.5091,
      "num_input_tokens_seen": 347355180,
      "step": 6200
    },
    {
      "epoch": 13.808463251670378,
      "loss": 0.6481593251228333,
      "loss_ce": 0.00014905582065694034,
      "loss_iou": 0.255859375,
      "loss_num": 0.027099609375,
      "loss_xval": 0.6484375,
      "num_input_tokens_seen": 347355180,
      "step": 6200
    },
    {
      "epoch": 13.810690423162583,
      "grad_norm": 24.352041244506836,
      "learning_rate": 1e-06,
      "loss": 0.4425,
      "num_input_tokens_seen": 347410936,
      "step": 6201
    },
    {
      "epoch": 13.810690423162583,
      "loss": 0.40238019824028015,
      "loss_ce": 9.74750000750646e-05,
      "loss_iou": 0.17578125,
      "loss_num": 0.01025390625,
      "loss_xval": 0.40234375,
      "num_input_tokens_seen": 347410936,
      "step": 6201
    },
    {
      "epoch": 13.812917594654788,
      "grad_norm": 12.734814643859863,
      "learning_rate": 1e-06,
      "loss": 0.4761,
      "num_input_tokens_seen": 347466652,
      "step": 6202
    },
    {
      "epoch": 13.812917594654788,
      "loss": 0.5395181179046631,
      "loss_ce": 0.00010463084618095309,
      "loss_iou": 0.23046875,
      "loss_num": 0.015380859375,
      "loss_xval": 0.5390625,
      "num_input_tokens_seen": 347466652,
      "step": 6202
    },
    {
      "epoch": 13.815144766146993,
      "grad_norm": 19.63954734802246,
      "learning_rate": 1e-06,
      "loss": 0.5178,
      "num_input_tokens_seen": 347523340,
      "step": 6203
    },
    {
      "epoch": 13.815144766146993,
      "loss": 0.6680713295936584,
      "loss_ce": 0.00010255551023874432,
      "loss_iou": 0.294921875,
      "loss_num": 0.015380859375,
      "loss_xval": 0.66796875,
      "num_input_tokens_seen": 347523340,
      "step": 6203
    },
    {
      "epoch": 13.817371937639198,
      "grad_norm": 27.84977149963379,
      "learning_rate": 1e-06,
      "loss": 0.4513,
      "num_input_tokens_seen": 347581752,
      "step": 6204
    },
    {
      "epoch": 13.817371937639198,
      "loss": 0.509273886680603,
      "loss_ce": 0.00011864411499118432,
      "loss_iou": 0.220703125,
      "loss_num": 0.0135498046875,
      "loss_xval": 0.5078125,
      "num_input_tokens_seen": 347581752,
      "step": 6204
    },
    {
      "epoch": 13.819599109131403,
      "grad_norm": 18.612722396850586,
      "learning_rate": 1e-06,
      "loss": 0.5813,
      "num_input_tokens_seen": 347641196,
      "step": 6205
    },
    {
      "epoch": 13.819599109131403,
      "loss": 0.6570059061050415,
      "loss_ce": 0.00011501011613290757,
      "loss_iou": 0.275390625,
      "loss_num": 0.0213623046875,
      "loss_xval": 0.65625,
      "num_input_tokens_seen": 347641196,
      "step": 6205
    },
    {
      "epoch": 13.821826280623608,
      "grad_norm": 17.84321403503418,
      "learning_rate": 1e-06,
      "loss": 0.3995,
      "num_input_tokens_seen": 347699640,
      "step": 6206
    },
    {
      "epoch": 13.821826280623608,
      "loss": 0.42600324749946594,
      "loss_ce": 9.991762635763735e-05,
      "loss_iou": 0.1884765625,
      "loss_num": 0.00982666015625,
      "loss_xval": 0.42578125,
      "num_input_tokens_seen": 347699640,
      "step": 6206
    },
    {
      "epoch": 13.824053452115812,
      "grad_norm": 17.020217895507812,
      "learning_rate": 1e-06,
      "loss": 0.5993,
      "num_input_tokens_seen": 347755124,
      "step": 6207
    },
    {
      "epoch": 13.824053452115812,
      "loss": 0.4133188724517822,
      "loss_ce": 0.00011088576866313815,
      "loss_iou": 0.1748046875,
      "loss_num": 0.0126953125,
      "loss_xval": 0.4140625,
      "num_input_tokens_seen": 347755124,
      "step": 6207
    },
    {
      "epoch": 13.826280623608017,
      "grad_norm": 16.579845428466797,
      "learning_rate": 1e-06,
      "loss": 0.5781,
      "num_input_tokens_seen": 347809184,
      "step": 6208
    },
    {
      "epoch": 13.826280623608017,
      "loss": 0.5595046281814575,
      "loss_ce": 0.00017845621914602816,
      "loss_iou": 0.2265625,
      "loss_num": 0.0213623046875,
      "loss_xval": 0.55859375,
      "num_input_tokens_seen": 347809184,
      "step": 6208
    },
    {
      "epoch": 13.828507795100222,
      "grad_norm": 21.396059036254883,
      "learning_rate": 1e-06,
      "loss": 0.5244,
      "num_input_tokens_seen": 347865520,
      "step": 6209
    },
    {
      "epoch": 13.828507795100222,
      "loss": 0.5646365880966187,
      "loss_ce": 0.00012240419164299965,
      "loss_iou": 0.2255859375,
      "loss_num": 0.02294921875,
      "loss_xval": 0.56640625,
      "num_input_tokens_seen": 347865520,
      "step": 6209
    },
    {
      "epoch": 13.830734966592427,
      "grad_norm": 12.559648513793945,
      "learning_rate": 1e-06,
      "loss": 0.4268,
      "num_input_tokens_seen": 347922004,
      "step": 6210
    },
    {
      "epoch": 13.830734966592427,
      "loss": 0.5307496786117554,
      "loss_ce": 0.00011002724932041019,
      "loss_iou": 0.2314453125,
      "loss_num": 0.01348876953125,
      "loss_xval": 0.53125,
      "num_input_tokens_seen": 347922004,
      "step": 6210
    },
    {
      "epoch": 13.832962138084632,
      "grad_norm": 20.28154945373535,
      "learning_rate": 1e-06,
      "loss": 0.5987,
      "num_input_tokens_seen": 347977316,
      "step": 6211
    },
    {
      "epoch": 13.832962138084632,
      "loss": 0.8992745876312256,
      "loss_ce": 0.00010463996295584366,
      "loss_iou": 0.357421875,
      "loss_num": 0.036865234375,
      "loss_xval": 0.8984375,
      "num_input_tokens_seen": 347977316,
      "step": 6211
    },
    {
      "epoch": 13.835189309576837,
      "grad_norm": 21.434589385986328,
      "learning_rate": 1e-06,
      "loss": 0.5093,
      "num_input_tokens_seen": 348031884,
      "step": 6212
    },
    {
      "epoch": 13.835189309576837,
      "loss": 0.5445865392684937,
      "loss_ce": 0.0001529275468783453,
      "loss_iou": 0.23046875,
      "loss_num": 0.0164794921875,
      "loss_xval": 0.54296875,
      "num_input_tokens_seen": 348031884,
      "step": 6212
    },
    {
      "epoch": 13.837416481069042,
      "grad_norm": 13.820694923400879,
      "learning_rate": 1e-06,
      "loss": 0.5881,
      "num_input_tokens_seen": 348088760,
      "step": 6213
    },
    {
      "epoch": 13.837416481069042,
      "loss": 0.5043725371360779,
      "loss_ce": 0.0001000510819721967,
      "loss_iou": 0.203125,
      "loss_num": 0.0196533203125,
      "loss_xval": 0.50390625,
      "num_input_tokens_seen": 348088760,
      "step": 6213
    },
    {
      "epoch": 13.839643652561247,
      "grad_norm": 16.25212287902832,
      "learning_rate": 1e-06,
      "loss": 0.4709,
      "num_input_tokens_seen": 348145232,
      "step": 6214
    },
    {
      "epoch": 13.839643652561247,
      "loss": 0.5301461219787598,
      "loss_ce": 0.0001167980080936104,
      "loss_iou": 0.23828125,
      "loss_num": 0.0106201171875,
      "loss_xval": 0.53125,
      "num_input_tokens_seen": 348145232,
      "step": 6214
    },
    {
      "epoch": 13.841870824053451,
      "grad_norm": 13.156241416931152,
      "learning_rate": 1e-06,
      "loss": 0.6487,
      "num_input_tokens_seen": 348203452,
      "step": 6215
    },
    {
      "epoch": 13.841870824053451,
      "loss": 0.9121133089065552,
      "loss_ce": 0.00012593530118465424,
      "loss_iou": 0.396484375,
      "loss_num": 0.02392578125,
      "loss_xval": 0.91015625,
      "num_input_tokens_seen": 348203452,
      "step": 6215
    },
    {
      "epoch": 13.844097995545656,
      "grad_norm": 19.261072158813477,
      "learning_rate": 1e-06,
      "loss": 0.5774,
      "num_input_tokens_seen": 348261088,
      "step": 6216
    },
    {
      "epoch": 13.844097995545656,
      "loss": 0.3862478733062744,
      "loss_ce": 7.846013613743708e-05,
      "loss_iou": 0.1708984375,
      "loss_num": 0.0087890625,
      "loss_xval": 0.38671875,
      "num_input_tokens_seen": 348261088,
      "step": 6216
    },
    {
      "epoch": 13.846325167037861,
      "grad_norm": 14.543188095092773,
      "learning_rate": 1e-06,
      "loss": 0.3715,
      "num_input_tokens_seen": 348316648,
      "step": 6217
    },
    {
      "epoch": 13.846325167037861,
      "loss": 0.398532897233963,
      "loss_ce": 9.540202154312283e-05,
      "loss_iou": 0.171875,
      "loss_num": 0.0108642578125,
      "loss_xval": 0.3984375,
      "num_input_tokens_seen": 348316648,
      "step": 6217
    },
    {
      "epoch": 13.848552338530066,
      "grad_norm": 16.337297439575195,
      "learning_rate": 1e-06,
      "loss": 0.6434,
      "num_input_tokens_seen": 348376184,
      "step": 6218
    },
    {
      "epoch": 13.848552338530066,
      "loss": 0.41022244095802307,
      "loss_ce": 0.00012721509847324342,
      "loss_iou": 0.177734375,
      "loss_num": 0.01080322265625,
      "loss_xval": 0.41015625,
      "num_input_tokens_seen": 348376184,
      "step": 6218
    },
    {
      "epoch": 13.85077951002227,
      "grad_norm": 34.131065368652344,
      "learning_rate": 1e-06,
      "loss": 0.6145,
      "num_input_tokens_seen": 348429572,
      "step": 6219
    },
    {
      "epoch": 13.85077951002227,
      "loss": 0.7063025236129761,
      "loss_ce": 0.0001257747644558549,
      "loss_iou": 0.306640625,
      "loss_num": 0.01904296875,
      "loss_xval": 0.70703125,
      "num_input_tokens_seen": 348429572,
      "step": 6219
    },
    {
      "epoch": 13.853006681514476,
      "grad_norm": 17.740169525146484,
      "learning_rate": 1e-06,
      "loss": 0.4083,
      "num_input_tokens_seen": 348487176,
      "step": 6220
    },
    {
      "epoch": 13.853006681514476,
      "loss": 0.4807821214199066,
      "loss_ce": 0.00013025107909925282,
      "loss_iou": 0.2041015625,
      "loss_num": 0.01458740234375,
      "loss_xval": 0.48046875,
      "num_input_tokens_seen": 348487176,
      "step": 6220
    },
    {
      "epoch": 13.855233853006682,
      "grad_norm": 25.03499984741211,
      "learning_rate": 1e-06,
      "loss": 0.5935,
      "num_input_tokens_seen": 348542216,
      "step": 6221
    },
    {
      "epoch": 13.855233853006682,
      "loss": 0.7307603359222412,
      "loss_ce": 0.0001694800885161385,
      "loss_iou": 0.30859375,
      "loss_num": 0.0228271484375,
      "loss_xval": 0.73046875,
      "num_input_tokens_seen": 348542216,
      "step": 6221
    },
    {
      "epoch": 13.857461024498887,
      "grad_norm": 22.183853149414062,
      "learning_rate": 1e-06,
      "loss": 0.4923,
      "num_input_tokens_seen": 348599416,
      "step": 6222
    },
    {
      "epoch": 13.857461024498887,
      "loss": 0.5290480852127075,
      "loss_ce": 0.00011743127834051847,
      "loss_iou": 0.24609375,
      "loss_num": 0.0074462890625,
      "loss_xval": 0.52734375,
      "num_input_tokens_seen": 348599416,
      "step": 6222
    },
    {
      "epoch": 13.859688195991092,
      "grad_norm": 16.716617584228516,
      "learning_rate": 1e-06,
      "loss": 0.4975,
      "num_input_tokens_seen": 348657108,
      "step": 6223
    },
    {
      "epoch": 13.859688195991092,
      "loss": 0.5516940355300903,
      "loss_ce": 0.00011933541827602312,
      "loss_iou": 0.2431640625,
      "loss_num": 0.012939453125,
      "loss_xval": 0.55078125,
      "num_input_tokens_seen": 348657108,
      "step": 6223
    },
    {
      "epoch": 13.861915367483297,
      "grad_norm": 24.67982292175293,
      "learning_rate": 1e-06,
      "loss": 0.4872,
      "num_input_tokens_seen": 348711952,
      "step": 6224
    },
    {
      "epoch": 13.861915367483297,
      "loss": 0.36566367745399475,
      "loss_ce": 0.0001241198042407632,
      "loss_iou": 0.158203125,
      "loss_num": 0.009765625,
      "loss_xval": 0.365234375,
      "num_input_tokens_seen": 348711952,
      "step": 6224
    },
    {
      "epoch": 13.864142538975502,
      "grad_norm": 23.012540817260742,
      "learning_rate": 1e-06,
      "loss": 0.4581,
      "num_input_tokens_seen": 348767728,
      "step": 6225
    },
    {
      "epoch": 13.864142538975502,
      "loss": 0.5795168280601501,
      "loss_ce": 0.00011010624439222738,
      "loss_iou": 0.263671875,
      "loss_num": 0.0103759765625,
      "loss_xval": 0.578125,
      "num_input_tokens_seen": 348767728,
      "step": 6225
    },
    {
      "epoch": 13.866369710467707,
      "grad_norm": 26.270526885986328,
      "learning_rate": 1e-06,
      "loss": 0.3405,
      "num_input_tokens_seen": 348821880,
      "step": 6226
    },
    {
      "epoch": 13.866369710467707,
      "loss": 0.3175492286682129,
      "loss_ce": 0.00010535772162256762,
      "loss_iou": 0.1259765625,
      "loss_num": 0.0130615234375,
      "loss_xval": 0.318359375,
      "num_input_tokens_seen": 348821880,
      "step": 6226
    },
    {
      "epoch": 13.868596881959911,
      "grad_norm": 19.418540954589844,
      "learning_rate": 1e-06,
      "loss": 0.3874,
      "num_input_tokens_seen": 348875740,
      "step": 6227
    },
    {
      "epoch": 13.868596881959911,
      "loss": 0.3931885063648224,
      "loss_ce": 0.00012211257126182318,
      "loss_iou": 0.173828125,
      "loss_num": 0.0089111328125,
      "loss_xval": 0.392578125,
      "num_input_tokens_seen": 348875740,
      "step": 6227
    },
    {
      "epoch": 13.870824053452116,
      "grad_norm": 19.871246337890625,
      "learning_rate": 1e-06,
      "loss": 0.6113,
      "num_input_tokens_seen": 348931196,
      "step": 6228
    },
    {
      "epoch": 13.870824053452116,
      "loss": 0.4289402961730957,
      "loss_ce": 0.000107303996628616,
      "loss_iou": 0.1982421875,
      "loss_num": 0.00634765625,
      "loss_xval": 0.4296875,
      "num_input_tokens_seen": 348931196,
      "step": 6228
    },
    {
      "epoch": 13.873051224944321,
      "grad_norm": 37.94038772583008,
      "learning_rate": 1e-06,
      "loss": 0.4865,
      "num_input_tokens_seen": 348988736,
      "step": 6229
    },
    {
      "epoch": 13.873051224944321,
      "loss": 0.48054730892181396,
      "loss_ce": 0.00013959786156192422,
      "loss_iou": 0.20703125,
      "loss_num": 0.01348876953125,
      "loss_xval": 0.48046875,
      "num_input_tokens_seen": 348988736,
      "step": 6229
    },
    {
      "epoch": 13.875278396436526,
      "grad_norm": 17.617862701416016,
      "learning_rate": 1e-06,
      "loss": 0.3775,
      "num_input_tokens_seen": 349044504,
      "step": 6230
    },
    {
      "epoch": 13.875278396436526,
      "loss": 0.3631332814693451,
      "loss_ce": 9.617961768526584e-05,
      "loss_iou": 0.1484375,
      "loss_num": 0.01336669921875,
      "loss_xval": 0.36328125,
      "num_input_tokens_seen": 349044504,
      "step": 6230
    },
    {
      "epoch": 13.877505567928731,
      "grad_norm": 19.312604904174805,
      "learning_rate": 1e-06,
      "loss": 0.533,
      "num_input_tokens_seen": 349101952,
      "step": 6231
    },
    {
      "epoch": 13.877505567928731,
      "loss": 0.4969325065612793,
      "loss_ce": 0.00010634450882207602,
      "loss_iou": 0.220703125,
      "loss_num": 0.01104736328125,
      "loss_xval": 0.49609375,
      "num_input_tokens_seen": 349101952,
      "step": 6231
    },
    {
      "epoch": 13.879732739420936,
      "grad_norm": 13.717912673950195,
      "learning_rate": 1e-06,
      "loss": 0.3243,
      "num_input_tokens_seen": 349155880,
      "step": 6232
    },
    {
      "epoch": 13.879732739420936,
      "loss": 0.3837384581565857,
      "loss_ce": 0.0001325202756561339,
      "loss_iou": 0.171875,
      "loss_num": 0.00787353515625,
      "loss_xval": 0.3828125,
      "num_input_tokens_seen": 349155880,
      "step": 6232
    },
    {
      "epoch": 13.88195991091314,
      "grad_norm": 22.239849090576172,
      "learning_rate": 1e-06,
      "loss": 0.4274,
      "num_input_tokens_seen": 349208756,
      "step": 6233
    },
    {
      "epoch": 13.88195991091314,
      "loss": 0.47713950276374817,
      "loss_ce": 8.873137267073616e-05,
      "loss_iou": 0.2177734375,
      "loss_num": 0.0081787109375,
      "loss_xval": 0.4765625,
      "num_input_tokens_seen": 349208756,
      "step": 6233
    },
    {
      "epoch": 13.884187082405345,
      "grad_norm": 17.712202072143555,
      "learning_rate": 1e-06,
      "loss": 0.4714,
      "num_input_tokens_seen": 349265684,
      "step": 6234
    },
    {
      "epoch": 13.884187082405345,
      "loss": 0.5098845958709717,
      "loss_ce": 0.00011895706120412797,
      "loss_iou": 0.2265625,
      "loss_num": 0.01129150390625,
      "loss_xval": 0.5078125,
      "num_input_tokens_seen": 349265684,
      "step": 6234
    },
    {
      "epoch": 13.88641425389755,
      "grad_norm": 20.694042205810547,
      "learning_rate": 1e-06,
      "loss": 0.5065,
      "num_input_tokens_seen": 349323304,
      "step": 6235
    },
    {
      "epoch": 13.88641425389755,
      "loss": 0.505508542060852,
      "loss_ce": 0.00013747680350206792,
      "loss_iou": 0.203125,
      "loss_num": 0.0196533203125,
      "loss_xval": 0.50390625,
      "num_input_tokens_seen": 349323304,
      "step": 6235
    },
    {
      "epoch": 13.888641425389755,
      "grad_norm": 17.659744262695312,
      "learning_rate": 1e-06,
      "loss": 0.6468,
      "num_input_tokens_seen": 349378928,
      "step": 6236
    },
    {
      "epoch": 13.888641425389755,
      "loss": 0.7527920603752136,
      "loss_ce": 0.00010650817421264946,
      "loss_iou": 0.326171875,
      "loss_num": 0.0205078125,
      "loss_xval": 0.75390625,
      "num_input_tokens_seen": 349378928,
      "step": 6236
    },
    {
      "epoch": 13.89086859688196,
      "grad_norm": 13.622602462768555,
      "learning_rate": 1e-06,
      "loss": 0.3588,
      "num_input_tokens_seen": 349436664,
      "step": 6237
    },
    {
      "epoch": 13.89086859688196,
      "loss": 0.42641109228134155,
      "loss_ce": 0.00014156656106933951,
      "loss_iou": 0.169921875,
      "loss_num": 0.01708984375,
      "loss_xval": 0.42578125,
      "num_input_tokens_seen": 349436664,
      "step": 6237
    },
    {
      "epoch": 13.893095768374165,
      "grad_norm": 16.968406677246094,
      "learning_rate": 1e-06,
      "loss": 0.4017,
      "num_input_tokens_seen": 349492432,
      "step": 6238
    },
    {
      "epoch": 13.893095768374165,
      "loss": 0.4099974036216736,
      "loss_ce": 8.531904313713312e-05,
      "loss_iou": 0.181640625,
      "loss_num": 0.00946044921875,
      "loss_xval": 0.41015625,
      "num_input_tokens_seen": 349492432,
      "step": 6238
    },
    {
      "epoch": 13.89532293986637,
      "grad_norm": 28.478656768798828,
      "learning_rate": 1e-06,
      "loss": 0.4145,
      "num_input_tokens_seen": 349550060,
      "step": 6239
    },
    {
      "epoch": 13.89532293986637,
      "loss": 0.37417882680892944,
      "loss_ce": 0.000277488463325426,
      "loss_iou": 0.1630859375,
      "loss_num": 0.0093994140625,
      "loss_xval": 0.373046875,
      "num_input_tokens_seen": 349550060,
      "step": 6239
    },
    {
      "epoch": 13.897550111358575,
      "grad_norm": 23.06316375732422,
      "learning_rate": 1e-06,
      "loss": 0.4668,
      "num_input_tokens_seen": 349603676,
      "step": 6240
    },
    {
      "epoch": 13.897550111358575,
      "loss": 0.5238842964172363,
      "loss_ce": 0.0001416070736013353,
      "loss_iou": 0.2236328125,
      "loss_num": 0.0152587890625,
      "loss_xval": 0.5234375,
      "num_input_tokens_seen": 349603676,
      "step": 6240
    },
    {
      "epoch": 13.89977728285078,
      "grad_norm": 15.570908546447754,
      "learning_rate": 1e-06,
      "loss": 0.6621,
      "num_input_tokens_seen": 349655680,
      "step": 6241
    },
    {
      "epoch": 13.89977728285078,
      "loss": 0.6655749678611755,
      "loss_ce": 0.00010867592936847359,
      "loss_iou": 0.263671875,
      "loss_num": 0.0272216796875,
      "loss_xval": 0.6640625,
      "num_input_tokens_seen": 349655680,
      "step": 6241
    },
    {
      "epoch": 13.902004454342984,
      "grad_norm": 20.07220458984375,
      "learning_rate": 1e-06,
      "loss": 0.4852,
      "num_input_tokens_seen": 349710520,
      "step": 6242
    },
    {
      "epoch": 13.902004454342984,
      "loss": 0.48412150144577026,
      "loss_ce": 0.00011269759124843404,
      "loss_iou": 0.22265625,
      "loss_num": 0.00787353515625,
      "loss_xval": 0.484375,
      "num_input_tokens_seen": 349710520,
      "step": 6242
    },
    {
      "epoch": 13.90423162583519,
      "grad_norm": 19.768949508666992,
      "learning_rate": 1e-06,
      "loss": 0.4714,
      "num_input_tokens_seen": 349766024,
      "step": 6243
    },
    {
      "epoch": 13.90423162583519,
      "loss": 0.5210692882537842,
      "loss_ce": 0.0003172849537804723,
      "loss_iou": 0.212890625,
      "loss_num": 0.0191650390625,
      "loss_xval": 0.51953125,
      "num_input_tokens_seen": 349766024,
      "step": 6243
    },
    {
      "epoch": 13.906458797327394,
      "grad_norm": 22.428667068481445,
      "learning_rate": 1e-06,
      "loss": 0.5109,
      "num_input_tokens_seen": 349821500,
      "step": 6244
    },
    {
      "epoch": 13.906458797327394,
      "loss": 0.5417461395263672,
      "loss_ce": 0.0001201537816086784,
      "loss_iou": 0.2236328125,
      "loss_num": 0.0189208984375,
      "loss_xval": 0.54296875,
      "num_input_tokens_seen": 349821500,
      "step": 6244
    },
    {
      "epoch": 13.908685968819599,
      "grad_norm": 14.843369483947754,
      "learning_rate": 1e-06,
      "loss": 0.4807,
      "num_input_tokens_seen": 349877108,
      "step": 6245
    },
    {
      "epoch": 13.908685968819599,
      "loss": 0.6390482187271118,
      "loss_ce": 0.0001321482559433207,
      "loss_iou": 0.263671875,
      "loss_num": 0.0223388671875,
      "loss_xval": 0.640625,
      "num_input_tokens_seen": 349877108,
      "step": 6245
    },
    {
      "epoch": 13.910913140311804,
      "grad_norm": 77.71258544921875,
      "learning_rate": 1e-06,
      "loss": 0.5519,
      "num_input_tokens_seen": 349930668,
      "step": 6246
    },
    {
      "epoch": 13.910913140311804,
      "loss": 0.7169502377510071,
      "loss_ce": 0.00015332532348111272,
      "loss_iou": 0.298828125,
      "loss_num": 0.024169921875,
      "loss_xval": 0.71875,
      "num_input_tokens_seen": 349930668,
      "step": 6246
    },
    {
      "epoch": 13.913140311804009,
      "grad_norm": 16.5616512298584,
      "learning_rate": 1e-06,
      "loss": 0.4834,
      "num_input_tokens_seen": 349988348,
      "step": 6247
    },
    {
      "epoch": 13.913140311804009,
      "loss": 0.39076149463653564,
      "loss_ce": 0.00013649219181388617,
      "loss_iou": 0.1640625,
      "loss_num": 0.0126953125,
      "loss_xval": 0.390625,
      "num_input_tokens_seen": 349988348,
      "step": 6247
    },
    {
      "epoch": 13.915367483296214,
      "grad_norm": 20.010635375976562,
      "learning_rate": 1e-06,
      "loss": 0.5611,
      "num_input_tokens_seen": 350046812,
      "step": 6248
    },
    {
      "epoch": 13.915367483296214,
      "loss": 0.6575692892074585,
      "loss_ce": 9.861437138170004e-05,
      "loss_iou": 0.296875,
      "loss_num": 0.01251220703125,
      "loss_xval": 0.65625,
      "num_input_tokens_seen": 350046812,
      "step": 6248
    },
    {
      "epoch": 13.917594654788418,
      "grad_norm": 15.879632949829102,
      "learning_rate": 1e-06,
      "loss": 0.4344,
      "num_input_tokens_seen": 350104192,
      "step": 6249
    },
    {
      "epoch": 13.917594654788418,
      "loss": 0.38620513677597046,
      "loss_ce": 9.672513988334686e-05,
      "loss_iou": 0.14453125,
      "loss_num": 0.0194091796875,
      "loss_xval": 0.38671875,
      "num_input_tokens_seen": 350104192,
      "step": 6249
    },
    {
      "epoch": 13.919821826280623,
      "grad_norm": 13.887146949768066,
      "learning_rate": 1e-06,
      "loss": 0.4196,
      "num_input_tokens_seen": 350161060,
      "step": 6250
    },
    {
      "epoch": 13.919821826280623,
      "eval_seeclick_web_CIoU": 0.5887904167175293,
      "eval_seeclick_web_GIoU": 0.5874381363391876,
      "eval_seeclick_web_IoU": 0.6068101227283478,
      "eval_seeclick_web_MAE_all": 0.015451115556061268,
      "eval_seeclick_web_MAE_h": 0.007658603135496378,
      "eval_seeclick_web_MAE_w": 0.015746516175568104,
      "eval_seeclick_web_MAE_x_boxes": 0.009071735199540854,
      "eval_seeclick_web_MAE_y_boxes": 0.02139524114318192,
      "eval_seeclick_web_inside_bbox": 0.9166666567325592,
      "eval_seeclick_web_loss": 0.8951788544654846,
      "eval_seeclick_web_loss_ce": 0.00017347045650240034,
      "eval_seeclick_web_loss_iou": 0.41064453125,
      "eval_seeclick_web_loss_num": 0.01229095458984375,
      "eval_seeclick_web_loss_xval": 0.88232421875,
      "eval_seeclick_web_runtime": 21.1304,
      "eval_seeclick_web_samples_per_second": 2.366,
      "eval_seeclick_web_steps_per_second": 0.095,
      "num_input_tokens_seen": 350161060,
      "step": 6250
    },
    {
      "epoch": 13.919821826280623,
      "eval_icons_CIoU": 0.2605983316898346,
      "eval_icons_GIoU": 0.29324978590011597,
      "eval_icons_IoU": 0.3456629067659378,
      "eval_icons_MAE_all": 0.06302645802497864,
      "eval_icons_MAE_h": 0.033296503126621246,
      "eval_icons_MAE_w": 0.07835287041962147,
      "eval_icons_MAE_x_boxes": 0.05176934972405434,
      "eval_icons_MAE_y_boxes": 0.03694954700767994,
      "eval_icons_inside_bbox": 0.6059027910232544,
      "eval_icons_loss": 1.7142837047576904,
      "eval_icons_loss_ce": 0.00022611367603531107,
      "eval_icons_loss_iou": 0.67041015625,
      "eval_icons_loss_num": 0.06027984619140625,
      "eval_icons_loss_xval": 1.642822265625,
      "eval_icons_runtime": 19.5905,
      "eval_icons_samples_per_second": 2.552,
      "eval_icons_steps_per_second": 0.102,
      "num_input_tokens_seen": 350161060,
      "step": 6250
    },
    {
      "epoch": 13.919821826280623,
      "eval_screenspot_CIoU": 0.38164886832237244,
      "eval_screenspot_GIoU": 0.39806386828422546,
      "eval_screenspot_IoU": 0.4498288830121358,
      "eval_screenspot_MAE_all": 0.05408271153767904,
      "eval_screenspot_MAE_h": 0.03916273762782415,
      "eval_screenspot_MAE_w": 0.056641269475221634,
      "eval_screenspot_MAE_x_boxes": 0.06717804819345474,
      "eval_screenspot_MAE_y_boxes": 0.036980644799768925,
      "eval_screenspot_inside_bbox": 0.7070833245913187,
      "eval_screenspot_loss": 1.5276081562042236,
      "eval_screenspot_loss_ce": 0.00023197308473754674,
      "eval_screenspot_loss_iou": 0.6381022135416666,
      "eval_screenspot_loss_num": 0.061542510986328125,
      "eval_screenspot_loss_xval": 1.5841471354166667,
      "eval_screenspot_runtime": 34.0345,
      "eval_screenspot_samples_per_second": 2.615,
      "eval_screenspot_steps_per_second": 0.088,
      "num_input_tokens_seen": 350161060,
      "step": 6250
    },
    {
      "epoch": 13.919821826280623,
      "eval_compot_CIoU": 0.3389701098203659,
      "eval_compot_GIoU": 0.3561352342367172,
      "eval_compot_IoU": 0.39824149012565613,
      "eval_compot_MAE_all": 0.019840517081320286,
      "eval_compot_MAE_h": 0.013703062664717436,
      "eval_compot_MAE_w": 0.021303851157426834,
      "eval_compot_MAE_x_boxes": 0.030402760952711105,
      "eval_compot_MAE_y_boxes": 0.006665045628324151,
      "eval_compot_inside_bbox": 0.6302083432674408,
      "eval_compot_loss": 1.4226690530776978,
      "eval_compot_loss_ce": 0.0001683719819993712,
      "eval_compot_loss_iou": 0.655029296875,
      "eval_compot_loss_num": 0.018640518188476562,
      "eval_compot_loss_xval": 1.403564453125,
      "eval_compot_runtime": 20.5568,
      "eval_compot_samples_per_second": 2.432,
      "eval_compot_steps_per_second": 0.097,
      "num_input_tokens_seen": 350161060,
      "step": 6250
    },
    {
      "epoch": 13.919821826280623,
      "eval_custom_ui_val_CIoU": 0.4785856149262852,
      "eval_custom_ui_val_GIoU": 0.4871194263299306,
      "eval_custom_ui_val_IoU": 0.5391448040803274,
      "eval_custom_ui_val_MAE_all": 0.027749659959226847,
      "eval_custom_ui_val_MAE_h": 0.015134843920047084,
      "eval_custom_ui_val_MAE_w": 0.03520229996906386,
      "eval_custom_ui_val_MAE_x_boxes": 0.03391206937117709,
      "eval_custom_ui_val_MAE_y_boxes": 0.01371948312347134,
      "eval_custom_ui_val_inside_bbox": 0.7650462985038757,
      "eval_custom_ui_val_loss": 1.1787360906600952,
      "eval_custom_ui_val_loss_ce": 0.00019364922385041913,
      "eval_custom_ui_val_loss_iou": 0.5048014322916666,
      "eval_custom_ui_val_loss_num": 0.024472342597113714,
      "eval_custom_ui_val_loss_xval": 1.1316189236111112,
      "eval_custom_ui_val_runtime": 61.6566,
      "eval_custom_ui_val_samples_per_second": 4.298,
      "eval_custom_ui_val_steps_per_second": 0.146,
      "num_input_tokens_seen": 350161060,
      "step": 6250
    },
    {
      "epoch": 13.919821826280623,
      "loss": 0.8202172517776489,
      "loss_ce": 0.00014893364277668297,
      "loss_iou": 0.369140625,
      "loss_num": 0.016357421875,
      "loss_xval": 0.8203125,
      "num_input_tokens_seen": 350161060,
      "step": 6250
    },
    {
      "epoch": 13.922048997772828,
      "grad_norm": 18.40821647644043,
      "learning_rate": 1e-06,
      "loss": 0.5081,
      "num_input_tokens_seen": 350218528,
      "step": 6251
    },
    {
      "epoch": 13.922048997772828,
      "loss": 0.4476110637187958,
      "loss_ce": 0.00010129214206244797,
      "loss_iou": 0.19140625,
      "loss_num": 0.01275634765625,
      "loss_xval": 0.447265625,
      "num_input_tokens_seen": 350218528,
      "step": 6251
    },
    {
      "epoch": 13.924276169265033,
      "grad_norm": 28.36144256591797,
      "learning_rate": 1e-06,
      "loss": 0.4074,
      "num_input_tokens_seen": 350276052,
      "step": 6252
    },
    {
      "epoch": 13.924276169265033,
      "loss": 0.3698493242263794,
      "loss_ce": 9.837304969551042e-05,
      "loss_iou": 0.171875,
      "loss_num": 0.005462646484375,
      "loss_xval": 0.369140625,
      "num_input_tokens_seen": 350276052,
      "step": 6252
    },
    {
      "epoch": 13.926503340757238,
      "grad_norm": 21.219388961791992,
      "learning_rate": 1e-06,
      "loss": 0.4852,
      "num_input_tokens_seen": 350333528,
      "step": 6253
    },
    {
      "epoch": 13.926503340757238,
      "loss": 0.43346133828163147,
      "loss_ce": 0.00011172999802511185,
      "loss_iou": 0.197265625,
      "loss_num": 0.00775146484375,
      "loss_xval": 0.43359375,
      "num_input_tokens_seen": 350333528,
      "step": 6253
    },
    {
      "epoch": 13.928730512249443,
      "grad_norm": 12.669755935668945,
      "learning_rate": 1e-06,
      "loss": 0.3754,
      "num_input_tokens_seen": 350391076,
      "step": 6254
    },
    {
      "epoch": 13.928730512249443,
      "loss": 0.30187392234802246,
      "loss_ce": 0.00011611805530264974,
      "loss_iou": 0.140625,
      "loss_num": 0.0042724609375,
      "loss_xval": 0.30078125,
      "num_input_tokens_seen": 350391076,
      "step": 6254
    },
    {
      "epoch": 13.930957683741648,
      "grad_norm": 13.667628288269043,
      "learning_rate": 1e-06,
      "loss": 0.2749,
      "num_input_tokens_seen": 350447892,
      "step": 6255
    },
    {
      "epoch": 13.930957683741648,
      "loss": 0.21279628574848175,
      "loss_ce": 8.877179789124057e-05,
      "loss_iou": 0.0859375,
      "loss_num": 0.00823974609375,
      "loss_xval": 0.212890625,
      "num_input_tokens_seen": 350447892,
      "step": 6255
    },
    {
      "epoch": 13.933184855233852,
      "grad_norm": 28.72197723388672,
      "learning_rate": 1e-06,
      "loss": 0.47,
      "num_input_tokens_seen": 350503796,
      "step": 6256
    },
    {
      "epoch": 13.933184855233852,
      "loss": 0.46128642559051514,
      "loss_ce": 0.00010479921184014529,
      "loss_iou": 0.212890625,
      "loss_num": 0.006866455078125,
      "loss_xval": 0.4609375,
      "num_input_tokens_seen": 350503796,
      "step": 6256
    },
    {
      "epoch": 13.935412026726057,
      "grad_norm": 21.46579933166504,
      "learning_rate": 1e-06,
      "loss": 0.5377,
      "num_input_tokens_seen": 350558404,
      "step": 6257
    },
    {
      "epoch": 13.935412026726057,
      "loss": 0.32118260860443115,
      "loss_ce": 0.00010715379903558642,
      "loss_iou": 0.1357421875,
      "loss_num": 0.010009765625,
      "loss_xval": 0.3203125,
      "num_input_tokens_seen": 350558404,
      "step": 6257
    },
    {
      "epoch": 13.937639198218262,
      "grad_norm": 21.574478149414062,
      "learning_rate": 1e-06,
      "loss": 0.5635,
      "num_input_tokens_seen": 350617276,
      "step": 6258
    },
    {
      "epoch": 13.937639198218262,
      "loss": 0.4228344261646271,
      "loss_ce": 0.00010491947614355013,
      "loss_iou": 0.1875,
      "loss_num": 0.00958251953125,
      "loss_xval": 0.421875,
      "num_input_tokens_seen": 350617276,
      "step": 6258
    },
    {
      "epoch": 13.939866369710467,
      "grad_norm": 23.370849609375,
      "learning_rate": 1e-06,
      "loss": 0.392,
      "num_input_tokens_seen": 350674816,
      "step": 6259
    },
    {
      "epoch": 13.939866369710467,
      "loss": 0.5362622737884521,
      "loss_ce": 0.0001294276735279709,
      "loss_iou": 0.2392578125,
      "loss_num": 0.0115966796875,
      "loss_xval": 0.53515625,
      "num_input_tokens_seen": 350674816,
      "step": 6259
    },
    {
      "epoch": 13.942093541202672,
      "grad_norm": 27.103878021240234,
      "learning_rate": 1e-06,
      "loss": 0.4583,
      "num_input_tokens_seen": 350729952,
      "step": 6260
    },
    {
      "epoch": 13.942093541202672,
      "loss": 0.41460520029067993,
      "loss_ce": 0.00011545630695763975,
      "loss_iou": 0.1884765625,
      "loss_num": 0.00750732421875,
      "loss_xval": 0.4140625,
      "num_input_tokens_seen": 350729952,
      "step": 6260
    },
    {
      "epoch": 13.944320712694877,
      "grad_norm": 17.797826766967773,
      "learning_rate": 1e-06,
      "loss": 0.3609,
      "num_input_tokens_seen": 350784792,
      "step": 6261
    },
    {
      "epoch": 13.944320712694877,
      "loss": 0.33117377758026123,
      "loss_ce": 0.00011909526074305177,
      "loss_iou": 0.1552734375,
      "loss_num": 0.00408935546875,
      "loss_xval": 0.33203125,
      "num_input_tokens_seen": 350784792,
      "step": 6261
    },
    {
      "epoch": 13.946547884187082,
      "grad_norm": 19.662729263305664,
      "learning_rate": 1e-06,
      "loss": 0.5463,
      "num_input_tokens_seen": 350838324,
      "step": 6262
    },
    {
      "epoch": 13.946547884187082,
      "loss": 0.5496599674224854,
      "loss_ce": 9.940941527020186e-05,
      "loss_iou": 0.2421875,
      "loss_num": 0.01318359375,
      "loss_xval": 0.55078125,
      "num_input_tokens_seen": 350838324,
      "step": 6262
    },
    {
      "epoch": 13.948775055679288,
      "grad_norm": 15.954992294311523,
      "learning_rate": 1e-06,
      "loss": 0.3793,
      "num_input_tokens_seen": 350897484,
      "step": 6263
    },
    {
      "epoch": 13.948775055679288,
      "loss": 0.46910548210144043,
      "loss_ce": 0.00011134070518892258,
      "loss_iou": 0.212890625,
      "loss_num": 0.00860595703125,
      "loss_xval": 0.46875,
      "num_input_tokens_seen": 350897484,
      "step": 6263
    },
    {
      "epoch": 13.951002227171493,
      "grad_norm": 28.188735961914062,
      "learning_rate": 1e-06,
      "loss": 0.606,
      "num_input_tokens_seen": 350953472,
      "step": 6264
    },
    {
      "epoch": 13.951002227171493,
      "loss": 0.5297653079032898,
      "loss_ce": 0.00010221092088613659,
      "loss_iou": 0.203125,
      "loss_num": 0.0247802734375,
      "loss_xval": 0.53125,
      "num_input_tokens_seen": 350953472,
      "step": 6264
    },
    {
      "epoch": 13.953229398663698,
      "grad_norm": 18.34393310546875,
      "learning_rate": 1e-06,
      "loss": 0.4006,
      "num_input_tokens_seen": 351008120,
      "step": 6265
    },
    {
      "epoch": 13.953229398663698,
      "loss": 0.5186876058578491,
      "loss_ce": 0.00013288386981002986,
      "loss_iou": 0.224609375,
      "loss_num": 0.013671875,
      "loss_xval": 0.51953125,
      "num_input_tokens_seen": 351008120,
      "step": 6265
    },
    {
      "epoch": 13.955456570155903,
      "grad_norm": 19.615419387817383,
      "learning_rate": 1e-06,
      "loss": 0.5483,
      "num_input_tokens_seen": 351064248,
      "step": 6266
    },
    {
      "epoch": 13.955456570155903,
      "loss": 0.7774511575698853,
      "loss_ce": 0.00010741624282673001,
      "loss_iou": 0.328125,
      "loss_num": 0.0242919921875,
      "loss_xval": 0.77734375,
      "num_input_tokens_seen": 351064248,
      "step": 6266
    },
    {
      "epoch": 13.957683741648108,
      "grad_norm": 25.06759262084961,
      "learning_rate": 1e-06,
      "loss": 0.5834,
      "num_input_tokens_seen": 351117008,
      "step": 6267
    },
    {
      "epoch": 13.957683741648108,
      "loss": 0.4678989052772522,
      "loss_ce": 0.00012549315579235554,
      "loss_iou": 0.20703125,
      "loss_num": 0.0108642578125,
      "loss_xval": 0.46875,
      "num_input_tokens_seen": 351117008,
      "step": 6267
    },
    {
      "epoch": 13.959910913140313,
      "grad_norm": 14.953537940979004,
      "learning_rate": 1e-06,
      "loss": 0.3766,
      "num_input_tokens_seen": 351171192,
      "step": 6268
    },
    {
      "epoch": 13.959910913140313,
      "loss": 0.3148655891418457,
      "loss_ce": 0.00010729426867328584,
      "loss_iou": 0.140625,
      "loss_num": 0.006866455078125,
      "loss_xval": 0.314453125,
      "num_input_tokens_seen": 351171192,
      "step": 6268
    },
    {
      "epoch": 13.962138084632517,
      "grad_norm": 17.59319305419922,
      "learning_rate": 1e-06,
      "loss": 0.4396,
      "num_input_tokens_seen": 351224944,
      "step": 6269
    },
    {
      "epoch": 13.962138084632517,
      "loss": 0.5811617374420166,
      "loss_ce": 0.00016810203669592738,
      "loss_iou": 0.267578125,
      "loss_num": 0.00897216796875,
      "loss_xval": 0.58203125,
      "num_input_tokens_seen": 351224944,
      "step": 6269
    },
    {
      "epoch": 13.964365256124722,
      "grad_norm": 12.526680946350098,
      "learning_rate": 1e-06,
      "loss": 0.4156,
      "num_input_tokens_seen": 351278692,
      "step": 6270
    },
    {
      "epoch": 13.964365256124722,
      "loss": 0.4158197343349457,
      "loss_ce": 0.0001092795209842734,
      "loss_iou": 0.177734375,
      "loss_num": 0.01190185546875,
      "loss_xval": 0.416015625,
      "num_input_tokens_seen": 351278692,
      "step": 6270
    },
    {
      "epoch": 13.966592427616927,
      "grad_norm": 15.219315528869629,
      "learning_rate": 1e-06,
      "loss": 0.4471,
      "num_input_tokens_seen": 351330732,
      "step": 6271
    },
    {
      "epoch": 13.966592427616927,
      "loss": 0.4622631371021271,
      "loss_ce": 0.0001049246930051595,
      "loss_iou": 0.2119140625,
      "loss_num": 0.0078125,
      "loss_xval": 0.462890625,
      "num_input_tokens_seen": 351330732,
      "step": 6271
    },
    {
      "epoch": 13.968819599109132,
      "grad_norm": 16.425418853759766,
      "learning_rate": 1e-06,
      "loss": 0.6101,
      "num_input_tokens_seen": 351389508,
      "step": 6272
    },
    {
      "epoch": 13.968819599109132,
      "loss": 0.4642532467842102,
      "loss_ce": 0.0001418821921106428,
      "loss_iou": 0.1708984375,
      "loss_num": 0.0245361328125,
      "loss_xval": 0.46484375,
      "num_input_tokens_seen": 351389508,
      "step": 6272
    },
    {
      "epoch": 13.971046770601337,
      "grad_norm": 15.943306922912598,
      "learning_rate": 1e-06,
      "loss": 0.5201,
      "num_input_tokens_seen": 351447368,
      "step": 6273
    },
    {
      "epoch": 13.971046770601337,
      "loss": 0.5003509521484375,
      "loss_ce": 0.00010678636317607015,
      "loss_iou": 0.2177734375,
      "loss_num": 0.01287841796875,
      "loss_xval": 0.5,
      "num_input_tokens_seen": 351447368,
      "step": 6273
    },
    {
      "epoch": 13.973273942093542,
      "grad_norm": 21.748048782348633,
      "learning_rate": 1e-06,
      "loss": 0.4717,
      "num_input_tokens_seen": 351506568,
      "step": 6274
    },
    {
      "epoch": 13.973273942093542,
      "loss": 0.5563892126083374,
      "loss_ce": 0.00011485861614346504,
      "loss_iou": 0.228515625,
      "loss_num": 0.0198974609375,
      "loss_xval": 0.5546875,
      "num_input_tokens_seen": 351506568,
      "step": 6274
    },
    {
      "epoch": 13.975501113585747,
      "grad_norm": 23.033185958862305,
      "learning_rate": 1e-06,
      "loss": 0.4948,
      "num_input_tokens_seen": 351562164,
      "step": 6275
    },
    {
      "epoch": 13.975501113585747,
      "loss": 0.662088930606842,
      "loss_ce": 0.00010161636600969359,
      "loss_iou": 0.294921875,
      "loss_num": 0.014404296875,
      "loss_xval": 0.66015625,
      "num_input_tokens_seen": 351562164,
      "step": 6275
    },
    {
      "epoch": 13.977728285077951,
      "grad_norm": 14.771820068359375,
      "learning_rate": 1e-06,
      "loss": 0.3768,
      "num_input_tokens_seen": 351616064,
      "step": 6276
    },
    {
      "epoch": 13.977728285077951,
      "loss": 0.5282983779907227,
      "loss_ce": 0.00022217544028535485,
      "loss_iou": 0.2353515625,
      "loss_num": 0.01153564453125,
      "loss_xval": 0.52734375,
      "num_input_tokens_seen": 351616064,
      "step": 6276
    },
    {
      "epoch": 13.979955456570156,
      "grad_norm": 15.470437049865723,
      "learning_rate": 1e-06,
      "loss": 0.4431,
      "num_input_tokens_seen": 351671208,
      "step": 6277
    },
    {
      "epoch": 13.979955456570156,
      "loss": 0.4118441939353943,
      "loss_ce": 0.00010104169632541016,
      "loss_iou": 0.1865234375,
      "loss_num": 0.0076904296875,
      "loss_xval": 0.412109375,
      "num_input_tokens_seen": 351671208,
      "step": 6277
    },
    {
      "epoch": 13.982182628062361,
      "grad_norm": 19.21230697631836,
      "learning_rate": 1e-06,
      "loss": 0.4367,
      "num_input_tokens_seen": 351725684,
      "step": 6278
    },
    {
      "epoch": 13.982182628062361,
      "loss": 0.27793729305267334,
      "loss_ce": 0.00010527193080633879,
      "loss_iou": 0.11669921875,
      "loss_num": 0.0089111328125,
      "loss_xval": 0.27734375,
      "num_input_tokens_seen": 351725684,
      "step": 6278
    },
    {
      "epoch": 13.984409799554566,
      "grad_norm": 21.042078018188477,
      "learning_rate": 1e-06,
      "loss": 0.3497,
      "num_input_tokens_seen": 351782280,
      "step": 6279
    },
    {
      "epoch": 13.984409799554566,
      "loss": 0.2832984924316406,
      "loss_ce": 9.53472190303728e-05,
      "loss_iou": 0.12451171875,
      "loss_num": 0.00677490234375,
      "loss_xval": 0.283203125,
      "num_input_tokens_seen": 351782280,
      "step": 6279
    },
    {
      "epoch": 13.98663697104677,
      "grad_norm": 19.898780822753906,
      "learning_rate": 1e-06,
      "loss": 0.6716,
      "num_input_tokens_seen": 351838828,
      "step": 6280
    },
    {
      "epoch": 13.98663697104677,
      "loss": 0.9190574288368225,
      "loss_ce": 0.00011211737000849098,
      "loss_iou": 0.36328125,
      "loss_num": 0.038818359375,
      "loss_xval": 0.91796875,
      "num_input_tokens_seen": 351838828,
      "step": 6280
    },
    {
      "epoch": 13.988864142538976,
      "grad_norm": 20.307798385620117,
      "learning_rate": 1e-06,
      "loss": 0.4632,
      "num_input_tokens_seen": 351894124,
      "step": 6281
    },
    {
      "epoch": 13.988864142538976,
      "loss": 0.3406025767326355,
      "loss_ce": 0.00014847918646410108,
      "loss_iou": 0.15234375,
      "loss_num": 0.006988525390625,
      "loss_xval": 0.33984375,
      "num_input_tokens_seen": 351894124,
      "step": 6281
    },
    {
      "epoch": 13.99109131403118,
      "grad_norm": 16.50027084350586,
      "learning_rate": 1e-06,
      "loss": 0.4563,
      "num_input_tokens_seen": 351949824,
      "step": 6282
    },
    {
      "epoch": 13.99109131403118,
      "loss": 0.38493967056274414,
      "loss_ce": 0.0001129888987634331,
      "loss_iou": 0.1806640625,
      "loss_num": 0.004547119140625,
      "loss_xval": 0.384765625,
      "num_input_tokens_seen": 351949824,
      "step": 6282
    },
    {
      "epoch": 13.993318485523385,
      "grad_norm": 30.10491180419922,
      "learning_rate": 1e-06,
      "loss": 0.5799,
      "num_input_tokens_seen": 352004840,
      "step": 6283
    },
    {
      "epoch": 13.993318485523385,
      "loss": 0.3842771351337433,
      "loss_ce": 0.00012185641389805824,
      "loss_iou": 0.16796875,
      "loss_num": 0.00982666015625,
      "loss_xval": 0.384765625,
      "num_input_tokens_seen": 352004840,
      "step": 6283
    },
    {
      "epoch": 13.99554565701559,
      "grad_norm": 26.573183059692383,
      "learning_rate": 1e-06,
      "loss": 0.6592,
      "num_input_tokens_seen": 352060908,
      "step": 6284
    },
    {
      "epoch": 13.99554565701559,
      "loss": 0.8157883882522583,
      "loss_ce": 0.00011452929174993187,
      "loss_iou": 0.33203125,
      "loss_num": 0.030029296875,
      "loss_xval": 0.81640625,
      "num_input_tokens_seen": 352060908,
      "step": 6284
    },
    {
      "epoch": 13.997772828507795,
      "grad_norm": 21.5324649810791,
      "learning_rate": 1e-06,
      "loss": 0.5067,
      "num_input_tokens_seen": 352115772,
      "step": 6285
    },
    {
      "epoch": 13.997772828507795,
      "loss": 0.6693138480186462,
      "loss_ce": 0.00012438424164429307,
      "loss_iou": 0.271484375,
      "loss_num": 0.0252685546875,
      "loss_xval": 0.66796875,
      "num_input_tokens_seen": 352115772,
      "step": 6285
    },
    {
      "epoch": 14.0,
      "grad_norm": 16.17325210571289,
      "learning_rate": 1e-06,
      "loss": 0.3972,
      "num_input_tokens_seen": 352173692,
      "step": 6286
    },
    {
      "epoch": 14.0,
      "loss": 0.43904364109039307,
      "loss_ce": 0.00010934725287370384,
      "loss_iou": 0.1845703125,
      "loss_num": 0.01409912109375,
      "loss_xval": 0.439453125,
      "num_input_tokens_seen": 352173692,
      "step": 6286
    },
    {
      "epoch": 14.002227171492205,
      "grad_norm": 22.501575469970703,
      "learning_rate": 1e-06,
      "loss": 0.4806,
      "num_input_tokens_seen": 352229596,
      "step": 6287
    },
    {
      "epoch": 14.002227171492205,
      "loss": 0.4168241024017334,
      "loss_ce": 0.00010654539801180363,
      "loss_iou": 0.1845703125,
      "loss_num": 0.00946044921875,
      "loss_xval": 0.416015625,
      "num_input_tokens_seen": 352229596,
      "step": 6287
    },
    {
      "epoch": 14.00445434298441,
      "grad_norm": 19.57029151916504,
      "learning_rate": 1e-06,
      "loss": 0.5103,
      "num_input_tokens_seen": 352285008,
      "step": 6288
    },
    {
      "epoch": 14.00445434298441,
      "loss": 0.5710501670837402,
      "loss_ce": 0.0008597684209235013,
      "loss_iou": 0.2470703125,
      "loss_num": 0.01519775390625,
      "loss_xval": 0.5703125,
      "num_input_tokens_seen": 352285008,
      "step": 6288
    },
    {
      "epoch": 14.006681514476615,
      "grad_norm": 13.75488567352295,
      "learning_rate": 1e-06,
      "loss": 0.3481,
      "num_input_tokens_seen": 352342884,
      "step": 6289
    },
    {
      "epoch": 14.006681514476615,
      "loss": 0.4669305682182312,
      "loss_ce": 0.0001337063149549067,
      "loss_iou": 0.2158203125,
      "loss_num": 0.0069580078125,
      "loss_xval": 0.466796875,
      "num_input_tokens_seen": 352342884,
      "step": 6289
    },
    {
      "epoch": 14.00890868596882,
      "grad_norm": 19.051847457885742,
      "learning_rate": 1e-06,
      "loss": 0.4405,
      "num_input_tokens_seen": 352397744,
      "step": 6290
    },
    {
      "epoch": 14.00890868596882,
      "loss": 0.5650573968887329,
      "loss_ce": 0.00011596856347750872,
      "loss_iou": 0.255859375,
      "loss_num": 0.01068115234375,
      "loss_xval": 0.56640625,
      "num_input_tokens_seen": 352397744,
      "step": 6290
    },
    {
      "epoch": 14.011135857461024,
      "grad_norm": 15.41951847076416,
      "learning_rate": 1e-06,
      "loss": 0.4502,
      "num_input_tokens_seen": 352454612,
      "step": 6291
    },
    {
      "epoch": 14.011135857461024,
      "loss": 0.5639560222625732,
      "loss_ce": 0.00011326879757689312,
      "loss_iou": 0.2431640625,
      "loss_num": 0.0155029296875,
      "loss_xval": 0.5625,
      "num_input_tokens_seen": 352454612,
      "step": 6291
    },
    {
      "epoch": 14.01336302895323,
      "grad_norm": 12.051553726196289,
      "learning_rate": 1e-06,
      "loss": 0.4497,
      "num_input_tokens_seen": 352511128,
      "step": 6292
    },
    {
      "epoch": 14.01336302895323,
      "loss": 0.2919390797615051,
      "loss_ce": 0.00013000219769310206,
      "loss_iou": 0.12060546875,
      "loss_num": 0.0101318359375,
      "loss_xval": 0.291015625,
      "num_input_tokens_seen": 352511128,
      "step": 6292
    },
    {
      "epoch": 14.015590200445434,
      "grad_norm": 22.342031478881836,
      "learning_rate": 1e-06,
      "loss": 0.3669,
      "num_input_tokens_seen": 352567572,
      "step": 6293
    },
    {
      "epoch": 14.015590200445434,
      "loss": 0.4516177773475647,
      "loss_ce": 0.00014069155440665781,
      "loss_iou": 0.1962890625,
      "loss_num": 0.01165771484375,
      "loss_xval": 0.451171875,
      "num_input_tokens_seen": 352567572,
      "step": 6293
    },
    {
      "epoch": 14.017817371937639,
      "grad_norm": 21.999927520751953,
      "learning_rate": 1e-06,
      "loss": 0.495,
      "num_input_tokens_seen": 352623720,
      "step": 6294
    },
    {
      "epoch": 14.017817371937639,
      "loss": 0.3385887145996094,
      "loss_ce": 0.00010300398571416736,
      "loss_iou": 0.15625,
      "loss_num": 0.005340576171875,
      "loss_xval": 0.337890625,
      "num_input_tokens_seen": 352623720,
      "step": 6294
    },
    {
      "epoch": 14.020044543429844,
      "grad_norm": 18.128568649291992,
      "learning_rate": 1e-06,
      "loss": 0.5133,
      "num_input_tokens_seen": 352679200,
      "step": 6295
    },
    {
      "epoch": 14.020044543429844,
      "loss": 0.4763103127479553,
      "loss_ce": 0.0001140405802289024,
      "loss_iou": 0.216796875,
      "loss_num": 0.00848388671875,
      "loss_xval": 0.4765625,
      "num_input_tokens_seen": 352679200,
      "step": 6295
    },
    {
      "epoch": 14.022271714922049,
      "grad_norm": 22.718978881835938,
      "learning_rate": 1e-06,
      "loss": 0.4883,
      "num_input_tokens_seen": 352737092,
      "step": 6296
    },
    {
      "epoch": 14.022271714922049,
      "loss": 0.6851996183395386,
      "loss_ce": 0.00014106131857261062,
      "loss_iou": 0.30859375,
      "loss_num": 0.013427734375,
      "loss_xval": 0.68359375,
      "num_input_tokens_seen": 352737092,
      "step": 6296
    },
    {
      "epoch": 14.024498886414253,
      "grad_norm": 23.775402069091797,
      "learning_rate": 1e-06,
      "loss": 0.5388,
      "num_input_tokens_seen": 352791712,
      "step": 6297
    },
    {
      "epoch": 14.024498886414253,
      "loss": 0.728790819644928,
      "loss_ce": 9.21202008612454e-05,
      "loss_iou": 0.3203125,
      "loss_num": 0.017578125,
      "loss_xval": 0.73046875,
      "num_input_tokens_seen": 352791712,
      "step": 6297
    },
    {
      "epoch": 14.026726057906458,
      "grad_norm": 27.953927993774414,
      "learning_rate": 1e-06,
      "loss": 0.4414,
      "num_input_tokens_seen": 352846396,
      "step": 6298
    },
    {
      "epoch": 14.026726057906458,
      "loss": 0.41728514432907104,
      "loss_ce": 0.00010984927939716727,
      "loss_iou": 0.1845703125,
      "loss_num": 0.0096435546875,
      "loss_xval": 0.41796875,
      "num_input_tokens_seen": 352846396,
      "step": 6298
    },
    {
      "epoch": 14.028953229398663,
      "grad_norm": 14.124860763549805,
      "learning_rate": 1e-06,
      "loss": 0.573,
      "num_input_tokens_seen": 352902036,
      "step": 6299
    },
    {
      "epoch": 14.028953229398663,
      "loss": 0.3383561968803406,
      "loss_ce": 9.933360706781968e-05,
      "loss_iou": 0.1591796875,
      "loss_num": 0.0040283203125,
      "loss_xval": 0.337890625,
      "num_input_tokens_seen": 352902036,
      "step": 6299
    },
    {
      "epoch": 14.031180400890868,
      "grad_norm": 22.924598693847656,
      "learning_rate": 1e-06,
      "loss": 0.4167,
      "num_input_tokens_seen": 352956388,
      "step": 6300
    },
    {
      "epoch": 14.031180400890868,
      "loss": 0.41771095991134644,
      "loss_ce": 0.00010839892638614401,
      "loss_iou": 0.1943359375,
      "loss_num": 0.0059814453125,
      "loss_xval": 0.41796875,
      "num_input_tokens_seen": 352956388,
      "step": 6300
    },
    {
      "epoch": 14.033407572383073,
      "grad_norm": 16.806791305541992,
      "learning_rate": 1e-06,
      "loss": 0.4694,
      "num_input_tokens_seen": 353012648,
      "step": 6301
    },
    {
      "epoch": 14.033407572383073,
      "loss": 0.6352797150611877,
      "loss_ce": 0.0001478988560847938,
      "loss_iou": 0.259765625,
      "loss_num": 0.0230712890625,
      "loss_xval": 0.63671875,
      "num_input_tokens_seen": 353012648,
      "step": 6301
    },
    {
      "epoch": 14.035634743875278,
      "grad_norm": 16.72489356994629,
      "learning_rate": 1e-06,
      "loss": 0.4935,
      "num_input_tokens_seen": 353071536,
      "step": 6302
    },
    {
      "epoch": 14.035634743875278,
      "loss": 0.5169607996940613,
      "loss_ce": 0.0001150843090726994,
      "loss_iou": 0.205078125,
      "loss_num": 0.0211181640625,
      "loss_xval": 0.515625,
      "num_input_tokens_seen": 353071536,
      "step": 6302
    },
    {
      "epoch": 14.037861915367483,
      "grad_norm": 19.627099990844727,
      "learning_rate": 1e-06,
      "loss": 0.4313,
      "num_input_tokens_seen": 353128896,
      "step": 6303
    },
    {
      "epoch": 14.037861915367483,
      "loss": 0.3792652487754822,
      "loss_ce": 0.00011487019946798682,
      "loss_iou": 0.1728515625,
      "loss_num": 0.006805419921875,
      "loss_xval": 0.37890625,
      "num_input_tokens_seen": 353128896,
      "step": 6303
    },
    {
      "epoch": 14.040089086859687,
      "grad_norm": 18.96303367614746,
      "learning_rate": 1e-06,
      "loss": 0.5243,
      "num_input_tokens_seen": 353185372,
      "step": 6304
    },
    {
      "epoch": 14.040089086859687,
      "loss": 0.5035980939865112,
      "loss_ce": 0.00011909975728485733,
      "loss_iou": 0.21875,
      "loss_num": 0.01324462890625,
      "loss_xval": 0.50390625,
      "num_input_tokens_seen": 353185372,
      "step": 6304
    },
    {
      "epoch": 14.042316258351892,
      "grad_norm": 24.597492218017578,
      "learning_rate": 1e-06,
      "loss": 0.5827,
      "num_input_tokens_seen": 353240952,
      "step": 6305
    },
    {
      "epoch": 14.042316258351892,
      "loss": 0.3702905774116516,
      "loss_ce": 0.0002954796073026955,
      "loss_iou": 0.1630859375,
      "loss_num": 0.0086669921875,
      "loss_xval": 0.369140625,
      "num_input_tokens_seen": 353240952,
      "step": 6305
    },
    {
      "epoch": 14.044543429844097,
      "grad_norm": 25.436721801757812,
      "learning_rate": 1e-06,
      "loss": 0.6157,
      "num_input_tokens_seen": 353297520,
      "step": 6306
    },
    {
      "epoch": 14.044543429844097,
      "loss": 0.5884321928024292,
      "loss_ce": 0.0001753760880092159,
      "loss_iou": 0.248046875,
      "loss_num": 0.0185546875,
      "loss_xval": 0.58984375,
      "num_input_tokens_seen": 353297520,
      "step": 6306
    },
    {
      "epoch": 14.046770601336302,
      "grad_norm": 18.7614688873291,
      "learning_rate": 1e-06,
      "loss": 0.5636,
      "num_input_tokens_seen": 353352068,
      "step": 6307
    },
    {
      "epoch": 14.046770601336302,
      "loss": 0.7366272807121277,
      "loss_ce": 0.00011604969040490687,
      "loss_iou": 0.28125,
      "loss_num": 0.034912109375,
      "loss_xval": 0.73828125,
      "num_input_tokens_seen": 353352068,
      "step": 6307
    },
    {
      "epoch": 14.048997772828507,
      "grad_norm": 15.662527084350586,
      "learning_rate": 1e-06,
      "loss": 0.4108,
      "num_input_tokens_seen": 353409180,
      "step": 6308
    },
    {
      "epoch": 14.048997772828507,
      "loss": 0.4494924545288086,
      "loss_ce": 9.055372356669977e-05,
      "loss_iou": 0.201171875,
      "loss_num": 0.00946044921875,
      "loss_xval": 0.44921875,
      "num_input_tokens_seen": 353409180,
      "step": 6308
    },
    {
      "epoch": 14.051224944320714,
      "grad_norm": 16.739437103271484,
      "learning_rate": 1e-06,
      "loss": 0.3383,
      "num_input_tokens_seen": 353465760,
      "step": 6309
    },
    {
      "epoch": 14.051224944320714,
      "loss": 0.4493185877799988,
      "loss_ce": 9.985115320887417e-05,
      "loss_iou": 0.1884765625,
      "loss_num": 0.01446533203125,
      "loss_xval": 0.44921875,
      "num_input_tokens_seen": 353465760,
      "step": 6309
    },
    {
      "epoch": 14.053452115812918,
      "grad_norm": 15.760394096374512,
      "learning_rate": 1e-06,
      "loss": 0.3149,
      "num_input_tokens_seen": 353519740,
      "step": 6310
    },
    {
      "epoch": 14.053452115812918,
      "loss": 0.28842777013778687,
      "loss_ce": 9.770273754838854e-05,
      "loss_iou": 0.1279296875,
      "loss_num": 0.006500244140625,
      "loss_xval": 0.2890625,
      "num_input_tokens_seen": 353519740,
      "step": 6310
    },
    {
      "epoch": 14.055679287305123,
      "grad_norm": 22.58936882019043,
      "learning_rate": 1e-06,
      "loss": 0.4747,
      "num_input_tokens_seen": 353575876,
      "step": 6311
    },
    {
      "epoch": 14.055679287305123,
      "loss": 0.4683777093887329,
      "loss_ce": 0.00011599110439419746,
      "loss_iou": 0.220703125,
      "loss_num": 0.00555419921875,
      "loss_xval": 0.46875,
      "num_input_tokens_seen": 353575876,
      "step": 6311
    },
    {
      "epoch": 14.057906458797328,
      "grad_norm": 15.249661445617676,
      "learning_rate": 1e-06,
      "loss": 0.4432,
      "num_input_tokens_seen": 353631208,
      "step": 6312
    },
    {
      "epoch": 14.057906458797328,
      "loss": 0.462121844291687,
      "loss_ce": 8.568944758735597e-05,
      "loss_iou": 0.205078125,
      "loss_num": 0.01025390625,
      "loss_xval": 0.462890625,
      "num_input_tokens_seen": 353631208,
      "step": 6312
    },
    {
      "epoch": 14.060133630289533,
      "grad_norm": 14.119874954223633,
      "learning_rate": 1e-06,
      "loss": 0.365,
      "num_input_tokens_seen": 353687380,
      "step": 6313
    },
    {
      "epoch": 14.060133630289533,
      "loss": 0.3032883107662201,
      "loss_ce": 9.618209878681228e-05,
      "loss_iou": 0.130859375,
      "loss_num": 0.00836181640625,
      "loss_xval": 0.302734375,
      "num_input_tokens_seen": 353687380,
      "step": 6313
    },
    {
      "epoch": 14.062360801781738,
      "grad_norm": 21.891582489013672,
      "learning_rate": 1e-06,
      "loss": 0.7123,
      "num_input_tokens_seen": 353742548,
      "step": 6314
    },
    {
      "epoch": 14.062360801781738,
      "loss": 0.7177752256393433,
      "loss_ce": 0.00012388010509312153,
      "loss_iou": 0.318359375,
      "loss_num": 0.0162353515625,
      "loss_xval": 0.71875,
      "num_input_tokens_seen": 353742548,
      "step": 6314
    },
    {
      "epoch": 14.064587973273943,
      "grad_norm": 18.22274398803711,
      "learning_rate": 1e-06,
      "loss": 0.4084,
      "num_input_tokens_seen": 353796396,
      "step": 6315
    },
    {
      "epoch": 14.064587973273943,
      "loss": 0.4653054475784302,
      "loss_ce": 9.54567440203391e-05,
      "loss_iou": 0.2060546875,
      "loss_num": 0.010498046875,
      "loss_xval": 0.46484375,
      "num_input_tokens_seen": 353796396,
      "step": 6315
    },
    {
      "epoch": 14.066815144766148,
      "grad_norm": 14.07210922241211,
      "learning_rate": 1e-06,
      "loss": 0.3755,
      "num_input_tokens_seen": 353852356,
      "step": 6316
    },
    {
      "epoch": 14.066815144766148,
      "loss": 0.3321226239204407,
      "loss_ce": 9.135504660662264e-05,
      "loss_iou": 0.142578125,
      "loss_num": 0.00958251953125,
      "loss_xval": 0.33203125,
      "num_input_tokens_seen": 353852356,
      "step": 6316
    },
    {
      "epoch": 14.069042316258352,
      "grad_norm": 12.254171371459961,
      "learning_rate": 1e-06,
      "loss": 0.3489,
      "num_input_tokens_seen": 353908400,
      "step": 6317
    },
    {
      "epoch": 14.069042316258352,
      "loss": 0.37034958600997925,
      "loss_ce": 0.00011033388000214472,
      "loss_iou": 0.171875,
      "loss_num": 0.00543212890625,
      "loss_xval": 0.37109375,
      "num_input_tokens_seen": 353908400,
      "step": 6317
    },
    {
      "epoch": 14.071269487750557,
      "grad_norm": 19.02251625061035,
      "learning_rate": 1e-06,
      "loss": 0.3671,
      "num_input_tokens_seen": 353963924,
      "step": 6318
    },
    {
      "epoch": 14.071269487750557,
      "loss": 0.32162177562713623,
      "loss_ce": 8.858899673214182e-05,
      "loss_iou": 0.138671875,
      "loss_num": 0.0087890625,
      "loss_xval": 0.322265625,
      "num_input_tokens_seen": 353963924,
      "step": 6318
    },
    {
      "epoch": 14.073496659242762,
      "grad_norm": 21.90892791748047,
      "learning_rate": 1e-06,
      "loss": 0.5865,
      "num_input_tokens_seen": 354021696,
      "step": 6319
    },
    {
      "epoch": 14.073496659242762,
      "loss": 0.7010501623153687,
      "loss_ce": 0.00012240943033248186,
      "loss_iou": 0.283203125,
      "loss_num": 0.027099609375,
      "loss_xval": 0.69921875,
      "num_input_tokens_seen": 354021696,
      "step": 6319
    },
    {
      "epoch": 14.075723830734967,
      "grad_norm": 22.8824520111084,
      "learning_rate": 1e-06,
      "loss": 0.3748,
      "num_input_tokens_seen": 354076836,
      "step": 6320
    },
    {
      "epoch": 14.075723830734967,
      "loss": 0.43087825179100037,
      "loss_ce": 9.210931602865458e-05,
      "loss_iou": 0.1767578125,
      "loss_num": 0.015380859375,
      "loss_xval": 0.431640625,
      "num_input_tokens_seen": 354076836,
      "step": 6320
    },
    {
      "epoch": 14.077951002227172,
      "grad_norm": 18.361591339111328,
      "learning_rate": 1e-06,
      "loss": 0.6328,
      "num_input_tokens_seen": 354131804,
      "step": 6321
    },
    {
      "epoch": 14.077951002227172,
      "loss": 0.9000133275985718,
      "loss_ce": 0.00011099971743533388,
      "loss_iou": 0.359375,
      "loss_num": 0.03662109375,
      "loss_xval": 0.8984375,
      "num_input_tokens_seen": 354131804,
      "step": 6321
    },
    {
      "epoch": 14.080178173719377,
      "grad_norm": 41.84800338745117,
      "learning_rate": 1e-06,
      "loss": 0.4503,
      "num_input_tokens_seen": 354187764,
      "step": 6322
    },
    {
      "epoch": 14.080178173719377,
      "loss": 0.4067497253417969,
      "loss_ce": 0.00013353029498830438,
      "loss_iou": 0.1650390625,
      "loss_num": 0.015380859375,
      "loss_xval": 0.40625,
      "num_input_tokens_seen": 354187764,
      "step": 6322
    },
    {
      "epoch": 14.082405345211582,
      "grad_norm": 24.2533016204834,
      "learning_rate": 1e-06,
      "loss": 0.5849,
      "num_input_tokens_seen": 354243268,
      "step": 6323
    },
    {
      "epoch": 14.082405345211582,
      "loss": 0.5544114708900452,
      "loss_ce": 9.021456935442984e-05,
      "loss_iou": 0.23828125,
      "loss_num": 0.0152587890625,
      "loss_xval": 0.5546875,
      "num_input_tokens_seen": 354243268,
      "step": 6323
    },
    {
      "epoch": 14.084632516703786,
      "grad_norm": 18.090354919433594,
      "learning_rate": 1e-06,
      "loss": 0.4398,
      "num_input_tokens_seen": 354301500,
      "step": 6324
    },
    {
      "epoch": 14.084632516703786,
      "loss": 0.4318804442882538,
      "loss_ce": 0.00011774900485761464,
      "loss_iou": 0.189453125,
      "loss_num": 0.01055908203125,
      "loss_xval": 0.431640625,
      "num_input_tokens_seen": 354301500,
      "step": 6324
    },
    {
      "epoch": 14.086859688195991,
      "grad_norm": 14.694714546203613,
      "learning_rate": 1e-06,
      "loss": 0.547,
      "num_input_tokens_seen": 354357332,
      "step": 6325
    },
    {
      "epoch": 14.086859688195991,
      "loss": 0.42843663692474365,
      "loss_ce": 9.19099838938564e-05,
      "loss_iou": 0.189453125,
      "loss_num": 0.009765625,
      "loss_xval": 0.427734375,
      "num_input_tokens_seen": 354357332,
      "step": 6325
    },
    {
      "epoch": 14.089086859688196,
      "grad_norm": 23.925561904907227,
      "learning_rate": 1e-06,
      "loss": 0.3976,
      "num_input_tokens_seen": 354411896,
      "step": 6326
    },
    {
      "epoch": 14.089086859688196,
      "loss": 0.47738510370254517,
      "loss_ce": 9.019082790473476e-05,
      "loss_iou": 0.1953125,
      "loss_num": 0.0174560546875,
      "loss_xval": 0.4765625,
      "num_input_tokens_seen": 354411896,
      "step": 6326
    },
    {
      "epoch": 14.091314031180401,
      "grad_norm": 17.225788116455078,
      "learning_rate": 1e-06,
      "loss": 0.5451,
      "num_input_tokens_seen": 354468544,
      "step": 6327
    },
    {
      "epoch": 14.091314031180401,
      "loss": 0.47155094146728516,
      "loss_ce": 0.00011540517152752727,
      "loss_iou": 0.2041015625,
      "loss_num": 0.01275634765625,
      "loss_xval": 0.470703125,
      "num_input_tokens_seen": 354468544,
      "step": 6327
    },
    {
      "epoch": 14.093541202672606,
      "grad_norm": 37.0084342956543,
      "learning_rate": 1e-06,
      "loss": 0.3675,
      "num_input_tokens_seen": 354527544,
      "step": 6328
    },
    {
      "epoch": 14.093541202672606,
      "loss": 0.3508872985839844,
      "loss_ce": 0.00011826444824691862,
      "loss_iou": 0.150390625,
      "loss_num": 0.0101318359375,
      "loss_xval": 0.3515625,
      "num_input_tokens_seen": 354527544,
      "step": 6328
    },
    {
      "epoch": 14.09576837416481,
      "grad_norm": 20.719806671142578,
      "learning_rate": 1e-06,
      "loss": 0.3541,
      "num_input_tokens_seen": 354581636,
      "step": 6329
    },
    {
      "epoch": 14.09576837416481,
      "loss": 0.2988535761833191,
      "loss_ce": 8.648804941913113e-05,
      "loss_iou": 0.134765625,
      "loss_num": 0.00592041015625,
      "loss_xval": 0.298828125,
      "num_input_tokens_seen": 354581636,
      "step": 6329
    },
    {
      "epoch": 14.097995545657016,
      "grad_norm": 23.697895050048828,
      "learning_rate": 1e-06,
      "loss": 0.4439,
      "num_input_tokens_seen": 354638068,
      "step": 6330
    },
    {
      "epoch": 14.097995545657016,
      "loss": 0.3387835621833801,
      "loss_ce": 9.947673243004829e-05,
      "loss_iou": 0.1474609375,
      "loss_num": 0.0086669921875,
      "loss_xval": 0.337890625,
      "num_input_tokens_seen": 354638068,
      "step": 6330
    },
    {
      "epoch": 14.10022271714922,
      "grad_norm": 13.052238464355469,
      "learning_rate": 1e-06,
      "loss": 0.3469,
      "num_input_tokens_seen": 354695576,
      "step": 6331
    },
    {
      "epoch": 14.10022271714922,
      "loss": 0.3428958058357239,
      "loss_ce": 0.00012233792222104967,
      "loss_iou": 0.138671875,
      "loss_num": 0.01287841796875,
      "loss_xval": 0.34375,
      "num_input_tokens_seen": 354695576,
      "step": 6331
    },
    {
      "epoch": 14.102449888641425,
      "grad_norm": 12.137063980102539,
      "learning_rate": 1e-06,
      "loss": 0.4448,
      "num_input_tokens_seen": 354750020,
      "step": 6332
    },
    {
      "epoch": 14.102449888641425,
      "loss": 0.5491877198219299,
      "loss_ce": 0.00011541788990143687,
      "loss_iou": 0.255859375,
      "loss_num": 0.007354736328125,
      "loss_xval": 0.55078125,
      "num_input_tokens_seen": 354750020,
      "step": 6332
    },
    {
      "epoch": 14.10467706013363,
      "grad_norm": 18.966337203979492,
      "learning_rate": 1e-06,
      "loss": 0.3768,
      "num_input_tokens_seen": 354808800,
      "step": 6333
    },
    {
      "epoch": 14.10467706013363,
      "loss": 0.30045419931411743,
      "loss_ce": 0.00010020330955740064,
      "loss_iou": 0.130859375,
      "loss_num": 0.007537841796875,
      "loss_xval": 0.30078125,
      "num_input_tokens_seen": 354808800,
      "step": 6333
    },
    {
      "epoch": 14.106904231625835,
      "grad_norm": 69.14574432373047,
      "learning_rate": 1e-06,
      "loss": 0.5943,
      "num_input_tokens_seen": 354863896,
      "step": 6334
    },
    {
      "epoch": 14.106904231625835,
      "loss": 0.6161885261535645,
      "loss_ce": 9.962108015315607e-05,
      "loss_iou": 0.248046875,
      "loss_num": 0.0240478515625,
      "loss_xval": 0.6171875,
      "num_input_tokens_seen": 354863896,
      "step": 6334
    },
    {
      "epoch": 14.10913140311804,
      "grad_norm": 29.898832321166992,
      "learning_rate": 1e-06,
      "loss": 0.5497,
      "num_input_tokens_seen": 354920516,
      "step": 6335
    },
    {
      "epoch": 14.10913140311804,
      "loss": 0.5754295587539673,
      "loss_ce": 0.00011214042024221271,
      "loss_iou": 0.271484375,
      "loss_num": 0.00689697265625,
      "loss_xval": 0.57421875,
      "num_input_tokens_seen": 354920516,
      "step": 6335
    },
    {
      "epoch": 14.111358574610245,
      "grad_norm": 17.41396141052246,
      "learning_rate": 1e-06,
      "loss": 0.2817,
      "num_input_tokens_seen": 354976696,
      "step": 6336
    },
    {
      "epoch": 14.111358574610245,
      "loss": 0.3508448898792267,
      "loss_ce": 0.0001368697703583166,
      "loss_iou": 0.1416015625,
      "loss_num": 0.0135498046875,
      "loss_xval": 0.3515625,
      "num_input_tokens_seen": 354976696,
      "step": 6336
    },
    {
      "epoch": 14.11358574610245,
      "grad_norm": 41.86484146118164,
      "learning_rate": 1e-06,
      "loss": 0.5174,
      "num_input_tokens_seen": 355034820,
      "step": 6337
    },
    {
      "epoch": 14.11358574610245,
      "loss": 0.38658392429351807,
      "loss_ce": 0.00010932501754723489,
      "loss_iou": 0.171875,
      "loss_num": 0.0084228515625,
      "loss_xval": 0.38671875,
      "num_input_tokens_seen": 355034820,
      "step": 6337
    },
    {
      "epoch": 14.115812917594655,
      "grad_norm": 22.32383918762207,
      "learning_rate": 1e-06,
      "loss": 0.6725,
      "num_input_tokens_seen": 355092456,
      "step": 6338
    },
    {
      "epoch": 14.115812917594655,
      "loss": 0.47444137930870056,
      "loss_ce": 0.00013718288391828537,
      "loss_iou": 0.19140625,
      "loss_num": 0.0184326171875,
      "loss_xval": 0.474609375,
      "num_input_tokens_seen": 355092456,
      "step": 6338
    },
    {
      "epoch": 14.11804008908686,
      "grad_norm": 15.70544147491455,
      "learning_rate": 1e-06,
      "loss": 0.3347,
      "num_input_tokens_seen": 355146944,
      "step": 6339
    },
    {
      "epoch": 14.11804008908686,
      "loss": 0.33699989318847656,
      "loss_ce": 8.58691637404263e-05,
      "loss_iou": 0.1396484375,
      "loss_num": 0.01171875,
      "loss_xval": 0.3359375,
      "num_input_tokens_seen": 355146944,
      "step": 6339
    },
    {
      "epoch": 14.120267260579064,
      "grad_norm": 16.518766403198242,
      "learning_rate": 1e-06,
      "loss": 0.4509,
      "num_input_tokens_seen": 355204304,
      "step": 6340
    },
    {
      "epoch": 14.120267260579064,
      "loss": 0.626310408115387,
      "loss_ce": 8.97101781447418e-05,
      "loss_iou": 0.2578125,
      "loss_num": 0.0224609375,
      "loss_xval": 0.625,
      "num_input_tokens_seen": 355204304,
      "step": 6340
    },
    {
      "epoch": 14.122494432071269,
      "grad_norm": 18.224716186523438,
      "learning_rate": 1e-06,
      "loss": 0.4472,
      "num_input_tokens_seen": 355260784,
      "step": 6341
    },
    {
      "epoch": 14.122494432071269,
      "loss": 0.504758894443512,
      "loss_ce": 0.00012020649592159316,
      "loss_iou": 0.2353515625,
      "loss_num": 0.00677490234375,
      "loss_xval": 0.50390625,
      "num_input_tokens_seen": 355260784,
      "step": 6341
    },
    {
      "epoch": 14.124721603563474,
      "grad_norm": 16.273900985717773,
      "learning_rate": 1e-06,
      "loss": 0.47,
      "num_input_tokens_seen": 355315728,
      "step": 6342
    },
    {
      "epoch": 14.124721603563474,
      "loss": 0.4144068956375122,
      "loss_ce": 0.0001002842909656465,
      "loss_iou": 0.19140625,
      "loss_num": 0.006256103515625,
      "loss_xval": 0.4140625,
      "num_input_tokens_seen": 355315728,
      "step": 6342
    },
    {
      "epoch": 14.126948775055679,
      "grad_norm": 13.019576072692871,
      "learning_rate": 1e-06,
      "loss": 0.3871,
      "num_input_tokens_seen": 355369944,
      "step": 6343
    },
    {
      "epoch": 14.126948775055679,
      "loss": 0.40841740369796753,
      "loss_ce": 9.22168546821922e-05,
      "loss_iou": 0.1728515625,
      "loss_num": 0.01275634765625,
      "loss_xval": 0.408203125,
      "num_input_tokens_seen": 355369944,
      "step": 6343
    },
    {
      "epoch": 14.129175946547884,
      "grad_norm": 20.190988540649414,
      "learning_rate": 1e-06,
      "loss": 0.4851,
      "num_input_tokens_seen": 355423788,
      "step": 6344
    },
    {
      "epoch": 14.129175946547884,
      "loss": 0.4641231596469879,
      "loss_ce": 0.00013389563537202775,
      "loss_iou": 0.21875,
      "loss_num": 0.00518798828125,
      "loss_xval": 0.46484375,
      "num_input_tokens_seen": 355423788,
      "step": 6344
    },
    {
      "epoch": 14.131403118040089,
      "grad_norm": 18.983137130737305,
      "learning_rate": 1e-06,
      "loss": 0.2741,
      "num_input_tokens_seen": 355483244,
      "step": 6345
    },
    {
      "epoch": 14.131403118040089,
      "loss": 0.33998721837997437,
      "loss_ce": 0.0001434668229194358,
      "loss_iou": 0.1494140625,
      "loss_num": 0.0081787109375,
      "loss_xval": 0.33984375,
      "num_input_tokens_seen": 355483244,
      "step": 6345
    },
    {
      "epoch": 14.133630289532293,
      "grad_norm": 15.164654731750488,
      "learning_rate": 1e-06,
      "loss": 0.4301,
      "num_input_tokens_seen": 355538616,
      "step": 6346
    },
    {
      "epoch": 14.133630289532293,
      "loss": 0.669284462928772,
      "loss_ce": 0.00015608461399096996,
      "loss_iou": 0.291015625,
      "loss_num": 0.0177001953125,
      "loss_xval": 0.66796875,
      "num_input_tokens_seen": 355538616,
      "step": 6346
    },
    {
      "epoch": 14.135857461024498,
      "grad_norm": 35.34605026245117,
      "learning_rate": 1e-06,
      "loss": 0.5471,
      "num_input_tokens_seen": 355594156,
      "step": 6347
    },
    {
      "epoch": 14.135857461024498,
      "loss": 0.6321955323219299,
      "loss_ce": 0.00011544384324224666,
      "loss_iou": 0.2734375,
      "loss_num": 0.0167236328125,
      "loss_xval": 0.6328125,
      "num_input_tokens_seen": 355594156,
      "step": 6347
    },
    {
      "epoch": 14.138084632516703,
      "grad_norm": 20.306251525878906,
      "learning_rate": 1e-06,
      "loss": 0.3952,
      "num_input_tokens_seen": 355649844,
      "step": 6348
    },
    {
      "epoch": 14.138084632516703,
      "loss": 0.3479854166507721,
      "loss_ce": 8.503998105879873e-05,
      "loss_iou": 0.1533203125,
      "loss_num": 0.00830078125,
      "loss_xval": 0.34765625,
      "num_input_tokens_seen": 355649844,
      "step": 6348
    },
    {
      "epoch": 14.140311804008908,
      "grad_norm": 15.569748878479004,
      "learning_rate": 1e-06,
      "loss": 0.4208,
      "num_input_tokens_seen": 355703872,
      "step": 6349
    },
    {
      "epoch": 14.140311804008908,
      "loss": 0.37485039234161377,
      "loss_ce": 9.451658843318e-05,
      "loss_iou": 0.162109375,
      "loss_num": 0.01007080078125,
      "loss_xval": 0.375,
      "num_input_tokens_seen": 355703872,
      "step": 6349
    },
    {
      "epoch": 14.142538975501113,
      "grad_norm": 26.381547927856445,
      "learning_rate": 1e-06,
      "loss": 0.561,
      "num_input_tokens_seen": 355758976,
      "step": 6350
    },
    {
      "epoch": 14.142538975501113,
      "loss": 0.6393318176269531,
      "loss_ce": 0.0001717099512461573,
      "loss_iou": 0.28515625,
      "loss_num": 0.01361083984375,
      "loss_xval": 0.640625,
      "num_input_tokens_seen": 355758976,
      "step": 6350
    },
    {
      "epoch": 14.144766146993318,
      "grad_norm": 18.737266540527344,
      "learning_rate": 1e-06,
      "loss": 0.5725,
      "num_input_tokens_seen": 355810544,
      "step": 6351
    },
    {
      "epoch": 14.144766146993318,
      "loss": 0.4250592589378357,
      "loss_ce": 0.00013248772302176803,
      "loss_iou": 0.1875,
      "loss_num": 0.0098876953125,
      "loss_xval": 0.42578125,
      "num_input_tokens_seen": 355810544,
      "step": 6351
    },
    {
      "epoch": 14.146993318485523,
      "grad_norm": 20.644193649291992,
      "learning_rate": 1e-06,
      "loss": 0.4436,
      "num_input_tokens_seen": 355865660,
      "step": 6352
    },
    {
      "epoch": 14.146993318485523,
      "loss": 0.3755820393562317,
      "loss_ce": 9.376133675687015e-05,
      "loss_iou": 0.1689453125,
      "loss_num": 0.0074462890625,
      "loss_xval": 0.375,
      "num_input_tokens_seen": 355865660,
      "step": 6352
    },
    {
      "epoch": 14.14922048997773,
      "grad_norm": 23.84221839904785,
      "learning_rate": 1e-06,
      "loss": 0.3666,
      "num_input_tokens_seen": 355921244,
      "step": 6353
    },
    {
      "epoch": 14.14922048997773,
      "loss": 0.3702180087566376,
      "loss_ce": 0.00010083305824082345,
      "loss_iou": 0.1689453125,
      "loss_num": 0.006500244140625,
      "loss_xval": 0.37109375,
      "num_input_tokens_seen": 355921244,
      "step": 6353
    },
    {
      "epoch": 14.151447661469934,
      "grad_norm": 28.195049285888672,
      "learning_rate": 1e-06,
      "loss": 0.4026,
      "num_input_tokens_seen": 355978880,
      "step": 6354
    },
    {
      "epoch": 14.151447661469934,
      "loss": 0.48697197437286377,
      "loss_ce": 0.00015556240396108478,
      "loss_iou": 0.2021484375,
      "loss_num": 0.016357421875,
      "loss_xval": 0.486328125,
      "num_input_tokens_seen": 355978880,
      "step": 6354
    },
    {
      "epoch": 14.153674832962139,
      "grad_norm": 19.111854553222656,
      "learning_rate": 1e-06,
      "loss": 0.529,
      "num_input_tokens_seen": 356033656,
      "step": 6355
    },
    {
      "epoch": 14.153674832962139,
      "loss": 0.5453453063964844,
      "loss_ce": 0.00011828625429188833,
      "loss_iou": 0.240234375,
      "loss_num": 0.01324462890625,
      "loss_xval": 0.546875,
      "num_input_tokens_seen": 356033656,
      "step": 6355
    },
    {
      "epoch": 14.155902004454344,
      "grad_norm": 17.10947036743164,
      "learning_rate": 1e-06,
      "loss": 0.4607,
      "num_input_tokens_seen": 356092412,
      "step": 6356
    },
    {
      "epoch": 14.155902004454344,
      "loss": 0.3630419373512268,
      "loss_ce": 0.00012691874871961772,
      "loss_iou": 0.1650390625,
      "loss_num": 0.0064697265625,
      "loss_xval": 0.36328125,
      "num_input_tokens_seen": 356092412,
      "step": 6356
    },
    {
      "epoch": 14.158129175946549,
      "grad_norm": 21.90672492980957,
      "learning_rate": 1e-06,
      "loss": 0.3746,
      "num_input_tokens_seen": 356149732,
      "step": 6357
    },
    {
      "epoch": 14.158129175946549,
      "loss": 0.35400235652923584,
      "loss_ce": 0.0002426167920930311,
      "loss_iou": 0.158203125,
      "loss_num": 0.007476806640625,
      "loss_xval": 0.353515625,
      "num_input_tokens_seen": 356149732,
      "step": 6357
    },
    {
      "epoch": 14.160356347438753,
      "grad_norm": 31.798358917236328,
      "learning_rate": 1e-06,
      "loss": 0.4167,
      "num_input_tokens_seen": 356204552,
      "step": 6358
    },
    {
      "epoch": 14.160356347438753,
      "loss": 0.5162287950515747,
      "loss_ce": 0.00011550244380487129,
      "loss_iou": 0.234375,
      "loss_num": 0.00933837890625,
      "loss_xval": 0.515625,
      "num_input_tokens_seen": 356204552,
      "step": 6358
    },
    {
      "epoch": 14.162583518930958,
      "grad_norm": 15.809988021850586,
      "learning_rate": 1e-06,
      "loss": 0.3727,
      "num_input_tokens_seen": 356262836,
      "step": 6359
    },
    {
      "epoch": 14.162583518930958,
      "loss": 0.4119860827922821,
      "loss_ce": 0.00012085679190931842,
      "loss_iou": 0.1875,
      "loss_num": 0.00726318359375,
      "loss_xval": 0.412109375,
      "num_input_tokens_seen": 356262836,
      "step": 6359
    },
    {
      "epoch": 14.164810690423163,
      "grad_norm": 14.865153312683105,
      "learning_rate": 1e-06,
      "loss": 0.4879,
      "num_input_tokens_seen": 356319092,
      "step": 6360
    },
    {
      "epoch": 14.164810690423163,
      "loss": 0.34744900465011597,
      "loss_ce": 9.792236960493028e-05,
      "loss_iou": 0.16015625,
      "loss_num": 0.005279541015625,
      "loss_xval": 0.34765625,
      "num_input_tokens_seen": 356319092,
      "step": 6360
    },
    {
      "epoch": 14.167037861915368,
      "grad_norm": 21.529083251953125,
      "learning_rate": 1e-06,
      "loss": 0.4694,
      "num_input_tokens_seen": 356375520,
      "step": 6361
    },
    {
      "epoch": 14.167037861915368,
      "loss": 0.3479962944984436,
      "loss_ce": 9.590189438313246e-05,
      "loss_iou": 0.1494140625,
      "loss_num": 0.010009765625,
      "loss_xval": 0.34765625,
      "num_input_tokens_seen": 356375520,
      "step": 6361
    },
    {
      "epoch": 14.169265033407573,
      "grad_norm": 15.021240234375,
      "learning_rate": 1e-06,
      "loss": 0.2992,
      "num_input_tokens_seen": 356434312,
      "step": 6362
    },
    {
      "epoch": 14.169265033407573,
      "loss": 0.27756401896476746,
      "loss_ce": 9.818993567023426e-05,
      "loss_iou": 0.11572265625,
      "loss_num": 0.00921630859375,
      "loss_xval": 0.27734375,
      "num_input_tokens_seen": 356434312,
      "step": 6362
    },
    {
      "epoch": 14.171492204899778,
      "grad_norm": 51.64773941040039,
      "learning_rate": 1e-06,
      "loss": 0.3894,
      "num_input_tokens_seen": 356489392,
      "step": 6363
    },
    {
      "epoch": 14.171492204899778,
      "loss": 0.5095750689506531,
      "loss_ce": 0.0002977301483042538,
      "loss_iou": 0.2001953125,
      "loss_num": 0.0218505859375,
      "loss_xval": 0.5078125,
      "num_input_tokens_seen": 356489392,
      "step": 6363
    },
    {
      "epoch": 14.173719376391983,
      "grad_norm": 16.491901397705078,
      "learning_rate": 1e-06,
      "loss": 0.371,
      "num_input_tokens_seen": 356548064,
      "step": 6364
    },
    {
      "epoch": 14.173719376391983,
      "loss": 0.3434831500053406,
      "loss_ce": 9.938179573509842e-05,
      "loss_iou": 0.15234375,
      "loss_num": 0.00762939453125,
      "loss_xval": 0.34375,
      "num_input_tokens_seen": 356548064,
      "step": 6364
    },
    {
      "epoch": 14.175946547884188,
      "grad_norm": 12.178879737854004,
      "learning_rate": 1e-06,
      "loss": 0.4276,
      "num_input_tokens_seen": 356606084,
      "step": 6365
    },
    {
      "epoch": 14.175946547884188,
      "loss": 0.39342576265335083,
      "loss_ce": 0.00011522185377543792,
      "loss_iou": 0.173828125,
      "loss_num": 0.0091552734375,
      "loss_xval": 0.392578125,
      "num_input_tokens_seen": 356606084,
      "step": 6365
    },
    {
      "epoch": 14.178173719376392,
      "grad_norm": 31.509164810180664,
      "learning_rate": 1e-06,
      "loss": 0.5393,
      "num_input_tokens_seen": 356662456,
      "step": 6366
    },
    {
      "epoch": 14.178173719376392,
      "loss": 0.522935152053833,
      "loss_ce": 0.00010799485608004034,
      "loss_iou": 0.232421875,
      "loss_num": 0.011474609375,
      "loss_xval": 0.5234375,
      "num_input_tokens_seen": 356662456,
      "step": 6366
    },
    {
      "epoch": 14.180400890868597,
      "grad_norm": 15.755391120910645,
      "learning_rate": 1e-06,
      "loss": 0.478,
      "num_input_tokens_seen": 356718496,
      "step": 6367
    },
    {
      "epoch": 14.180400890868597,
      "loss": 0.46932005882263184,
      "loss_ce": 8.176272240234539e-05,
      "loss_iou": 0.1962890625,
      "loss_num": 0.01513671875,
      "loss_xval": 0.46875,
      "num_input_tokens_seen": 356718496,
      "step": 6367
    },
    {
      "epoch": 14.182628062360802,
      "grad_norm": 20.124523162841797,
      "learning_rate": 1e-06,
      "loss": 0.443,
      "num_input_tokens_seen": 356772880,
      "step": 6368
    },
    {
      "epoch": 14.182628062360802,
      "loss": 0.4469994902610779,
      "loss_ce": 0.00010007787204813212,
      "loss_iou": 0.1787109375,
      "loss_num": 0.0179443359375,
      "loss_xval": 0.447265625,
      "num_input_tokens_seen": 356772880,
      "step": 6368
    },
    {
      "epoch": 14.184855233853007,
      "grad_norm": 14.184540748596191,
      "learning_rate": 1e-06,
      "loss": 0.364,
      "num_input_tokens_seen": 356829648,
      "step": 6369
    },
    {
      "epoch": 14.184855233853007,
      "loss": 0.49447864294052124,
      "loss_ce": 9.389698243467137e-05,
      "loss_iou": 0.2177734375,
      "loss_num": 0.01171875,
      "loss_xval": 0.494140625,
      "num_input_tokens_seen": 356829648,
      "step": 6369
    },
    {
      "epoch": 14.187082405345212,
      "grad_norm": 17.749513626098633,
      "learning_rate": 1e-06,
      "loss": 0.4139,
      "num_input_tokens_seen": 356886340,
      "step": 6370
    },
    {
      "epoch": 14.187082405345212,
      "loss": 0.38717857003211975,
      "loss_ce": 9.360718831885606e-05,
      "loss_iou": 0.171875,
      "loss_num": 0.008544921875,
      "loss_xval": 0.38671875,
      "num_input_tokens_seen": 356886340,
      "step": 6370
    },
    {
      "epoch": 14.189309576837417,
      "grad_norm": 49.11314392089844,
      "learning_rate": 1e-06,
      "loss": 0.6898,
      "num_input_tokens_seen": 356942864,
      "step": 6371
    },
    {
      "epoch": 14.189309576837417,
      "loss": 0.91033935546875,
      "loss_ce": 0.00018309304141439497,
      "loss_iou": 0.3984375,
      "loss_num": 0.0225830078125,
      "loss_xval": 0.91015625,
      "num_input_tokens_seen": 356942864,
      "step": 6371
    },
    {
      "epoch": 14.191536748329622,
      "grad_norm": 14.434630393981934,
      "learning_rate": 1e-06,
      "loss": 0.4182,
      "num_input_tokens_seen": 356998272,
      "step": 6372
    },
    {
      "epoch": 14.191536748329622,
      "loss": 0.3537047207355499,
      "loss_ce": 0.00012805727601516992,
      "loss_iou": 0.1611328125,
      "loss_num": 0.006195068359375,
      "loss_xval": 0.353515625,
      "num_input_tokens_seen": 356998272,
      "step": 6372
    },
    {
      "epoch": 14.193763919821826,
      "grad_norm": 26.62126350402832,
      "learning_rate": 1e-06,
      "loss": 0.3928,
      "num_input_tokens_seen": 357056092,
      "step": 6373
    },
    {
      "epoch": 14.193763919821826,
      "loss": 0.3926212787628174,
      "loss_ce": 0.00010417943849461153,
      "loss_iou": 0.1640625,
      "loss_num": 0.0130615234375,
      "loss_xval": 0.392578125,
      "num_input_tokens_seen": 357056092,
      "step": 6373
    },
    {
      "epoch": 14.195991091314031,
      "grad_norm": 19.854206085205078,
      "learning_rate": 1e-06,
      "loss": 0.4684,
      "num_input_tokens_seen": 357114944,
      "step": 6374
    },
    {
      "epoch": 14.195991091314031,
      "loss": 0.5363778471946716,
      "loss_ce": 0.00012294600310269743,
      "loss_iou": 0.2353515625,
      "loss_num": 0.01318359375,
      "loss_xval": 0.53515625,
      "num_input_tokens_seen": 357114944,
      "step": 6374
    },
    {
      "epoch": 14.198218262806236,
      "grad_norm": 17.966840744018555,
      "learning_rate": 1e-06,
      "loss": 0.4877,
      "num_input_tokens_seen": 357170380,
      "step": 6375
    },
    {
      "epoch": 14.198218262806236,
      "loss": 0.5723801851272583,
      "loss_ce": 0.00017562352877575904,
      "loss_iou": 0.2333984375,
      "loss_num": 0.0208740234375,
      "loss_xval": 0.5703125,
      "num_input_tokens_seen": 357170380,
      "step": 6375
    },
    {
      "epoch": 14.200445434298441,
      "grad_norm": 20.891889572143555,
      "learning_rate": 1e-06,
      "loss": 0.5619,
      "num_input_tokens_seen": 357227228,
      "step": 6376
    },
    {
      "epoch": 14.200445434298441,
      "loss": 0.6446791887283325,
      "loss_ce": 0.00011739273031707853,
      "loss_iou": 0.267578125,
      "loss_num": 0.021728515625,
      "loss_xval": 0.64453125,
      "num_input_tokens_seen": 357227228,
      "step": 6376
    },
    {
      "epoch": 14.202672605790646,
      "grad_norm": 18.76251792907715,
      "learning_rate": 1e-06,
      "loss": 0.3435,
      "num_input_tokens_seen": 357285012,
      "step": 6377
    },
    {
      "epoch": 14.202672605790646,
      "loss": 0.3950054347515106,
      "loss_ce": 0.00010797058348543942,
      "loss_iou": 0.173828125,
      "loss_num": 0.00927734375,
      "loss_xval": 0.39453125,
      "num_input_tokens_seen": 357285012,
      "step": 6377
    },
    {
      "epoch": 14.20489977728285,
      "grad_norm": 13.638835906982422,
      "learning_rate": 1e-06,
      "loss": 0.3753,
      "num_input_tokens_seen": 357341596,
      "step": 6378
    },
    {
      "epoch": 14.20489977728285,
      "loss": 0.4419354498386383,
      "loss_ce": 0.00010194515925832093,
      "loss_iou": 0.185546875,
      "loss_num": 0.01409912109375,
      "loss_xval": 0.44140625,
      "num_input_tokens_seen": 357341596,
      "step": 6378
    },
    {
      "epoch": 14.207126948775056,
      "grad_norm": 17.007076263427734,
      "learning_rate": 1e-06,
      "loss": 0.5073,
      "num_input_tokens_seen": 357397024,
      "step": 6379
    },
    {
      "epoch": 14.207126948775056,
      "loss": 0.2852746248245239,
      "loss_ce": 0.00011835141776828095,
      "loss_iou": 0.1240234375,
      "loss_num": 0.007415771484375,
      "loss_xval": 0.28515625,
      "num_input_tokens_seen": 357397024,
      "step": 6379
    },
    {
      "epoch": 14.20935412026726,
      "grad_norm": 15.55772590637207,
      "learning_rate": 1e-06,
      "loss": 0.4203,
      "num_input_tokens_seen": 357453348,
      "step": 6380
    },
    {
      "epoch": 14.20935412026726,
      "loss": 0.3796178698539734,
      "loss_ce": 0.00010125982225872576,
      "loss_iou": 0.17578125,
      "loss_num": 0.00567626953125,
      "loss_xval": 0.37890625,
      "num_input_tokens_seen": 357453348,
      "step": 6380
    },
    {
      "epoch": 14.211581291759465,
      "grad_norm": 21.045106887817383,
      "learning_rate": 1e-06,
      "loss": 0.4578,
      "num_input_tokens_seen": 357508568,
      "step": 6381
    },
    {
      "epoch": 14.211581291759465,
      "loss": 0.40396052598953247,
      "loss_ce": 0.00015192307182587683,
      "loss_iou": 0.16796875,
      "loss_num": 0.013671875,
      "loss_xval": 0.404296875,
      "num_input_tokens_seen": 357508568,
      "step": 6381
    },
    {
      "epoch": 14.21380846325167,
      "grad_norm": 27.14858627319336,
      "learning_rate": 1e-06,
      "loss": 0.4897,
      "num_input_tokens_seen": 357561732,
      "step": 6382
    },
    {
      "epoch": 14.21380846325167,
      "loss": 0.5612715482711792,
      "loss_ce": 0.00023641872394364327,
      "loss_iou": 0.24609375,
      "loss_num": 0.01385498046875,
      "loss_xval": 0.5625,
      "num_input_tokens_seen": 357561732,
      "step": 6382
    },
    {
      "epoch": 14.216035634743875,
      "grad_norm": 17.41556167602539,
      "learning_rate": 1e-06,
      "loss": 0.4793,
      "num_input_tokens_seen": 357618236,
      "step": 6383
    },
    {
      "epoch": 14.216035634743875,
      "loss": 0.5158074498176575,
      "loss_ce": 0.00012142492778366432,
      "loss_iou": 0.2138671875,
      "loss_num": 0.0177001953125,
      "loss_xval": 0.515625,
      "num_input_tokens_seen": 357618236,
      "step": 6383
    },
    {
      "epoch": 14.21826280623608,
      "grad_norm": 15.721260070800781,
      "learning_rate": 1e-06,
      "loss": 0.3657,
      "num_input_tokens_seen": 357676752,
      "step": 6384
    },
    {
      "epoch": 14.21826280623608,
      "loss": 0.4123426675796509,
      "loss_ce": 0.0001112363679567352,
      "loss_iou": 0.1826171875,
      "loss_num": 0.00927734375,
      "loss_xval": 0.412109375,
      "num_input_tokens_seen": 357676752,
      "step": 6384
    },
    {
      "epoch": 14.220489977728285,
      "grad_norm": 26.944326400756836,
      "learning_rate": 1e-06,
      "loss": 0.4681,
      "num_input_tokens_seen": 357734280,
      "step": 6385
    },
    {
      "epoch": 14.220489977728285,
      "loss": 0.46214836835861206,
      "loss_ce": 0.00011222571629332379,
      "loss_iou": 0.197265625,
      "loss_num": 0.013427734375,
      "loss_xval": 0.462890625,
      "num_input_tokens_seen": 357734280,
      "step": 6385
    },
    {
      "epoch": 14.22271714922049,
      "grad_norm": 19.070985794067383,
      "learning_rate": 1e-06,
      "loss": 0.3808,
      "num_input_tokens_seen": 357788536,
      "step": 6386
    },
    {
      "epoch": 14.22271714922049,
      "loss": 0.32470569014549255,
      "loss_ce": 0.00012072586105205119,
      "loss_iou": 0.1416015625,
      "loss_num": 0.00836181640625,
      "loss_xval": 0.32421875,
      "num_input_tokens_seen": 357788536,
      "step": 6386
    },
    {
      "epoch": 14.224944320712694,
      "grad_norm": 12.709535598754883,
      "learning_rate": 1e-06,
      "loss": 0.5202,
      "num_input_tokens_seen": 357844904,
      "step": 6387
    },
    {
      "epoch": 14.224944320712694,
      "loss": 0.35143405199050903,
      "loss_ce": 0.00011568302579689771,
      "loss_iou": 0.1611328125,
      "loss_num": 0.005706787109375,
      "loss_xval": 0.3515625,
      "num_input_tokens_seen": 357844904,
      "step": 6387
    },
    {
      "epoch": 14.2271714922049,
      "grad_norm": 15.428078651428223,
      "learning_rate": 1e-06,
      "loss": 0.4241,
      "num_input_tokens_seen": 357902220,
      "step": 6388
    },
    {
      "epoch": 14.2271714922049,
      "loss": 0.5605834126472473,
      "loss_ce": 0.00015861910651437938,
      "loss_iou": 0.26171875,
      "loss_num": 0.007354736328125,
      "loss_xval": 0.55859375,
      "num_input_tokens_seen": 357902220,
      "step": 6388
    },
    {
      "epoch": 14.229398663697104,
      "grad_norm": 25.415952682495117,
      "learning_rate": 1e-06,
      "loss": 0.4713,
      "num_input_tokens_seen": 357958080,
      "step": 6389
    },
    {
      "epoch": 14.229398663697104,
      "loss": 0.47960424423217773,
      "loss_ce": 0.00011205507325939834,
      "loss_iou": 0.224609375,
      "loss_num": 0.00604248046875,
      "loss_xval": 0.48046875,
      "num_input_tokens_seen": 357958080,
      "step": 6389
    },
    {
      "epoch": 14.231625835189309,
      "grad_norm": 14.51382827758789,
      "learning_rate": 1e-06,
      "loss": 0.4541,
      "num_input_tokens_seen": 358015476,
      "step": 6390
    },
    {
      "epoch": 14.231625835189309,
      "loss": 0.3873499631881714,
      "loss_ce": 0.00038707145722582936,
      "loss_iou": 0.1796875,
      "loss_num": 0.00543212890625,
      "loss_xval": 0.38671875,
      "num_input_tokens_seen": 358015476,
      "step": 6390
    },
    {
      "epoch": 14.233853006681514,
      "grad_norm": 20.579307556152344,
      "learning_rate": 1e-06,
      "loss": 0.592,
      "num_input_tokens_seen": 358073920,
      "step": 6391
    },
    {
      "epoch": 14.233853006681514,
      "loss": 0.554311215877533,
      "loss_ce": 0.00011194508260814473,
      "loss_iou": 0.234375,
      "loss_num": 0.016845703125,
      "loss_xval": 0.5546875,
      "num_input_tokens_seen": 358073920,
      "step": 6391
    },
    {
      "epoch": 14.236080178173719,
      "grad_norm": 18.920732498168945,
      "learning_rate": 1e-06,
      "loss": 0.4462,
      "num_input_tokens_seen": 358130704,
      "step": 6392
    },
    {
      "epoch": 14.236080178173719,
      "loss": 0.5425920486450195,
      "loss_ce": 0.00011164323223056272,
      "loss_iou": 0.2412109375,
      "loss_num": 0.01202392578125,
      "loss_xval": 0.54296875,
      "num_input_tokens_seen": 358130704,
      "step": 6392
    },
    {
      "epoch": 14.238307349665924,
      "grad_norm": 16.76801872253418,
      "learning_rate": 1e-06,
      "loss": 0.3899,
      "num_input_tokens_seen": 358189660,
      "step": 6393
    },
    {
      "epoch": 14.238307349665924,
      "loss": 0.345196008682251,
      "loss_ce": 0.00010322515299776569,
      "loss_iou": 0.150390625,
      "loss_num": 0.0087890625,
      "loss_xval": 0.345703125,
      "num_input_tokens_seen": 358189660,
      "step": 6393
    },
    {
      "epoch": 14.240534521158128,
      "grad_norm": 20.114219665527344,
      "learning_rate": 1e-06,
      "loss": 0.3621,
      "num_input_tokens_seen": 358247428,
      "step": 6394
    },
    {
      "epoch": 14.240534521158128,
      "loss": 0.3770490884780884,
      "loss_ce": 9.597234020475298e-05,
      "loss_iou": 0.17578125,
      "loss_num": 0.005096435546875,
      "loss_xval": 0.376953125,
      "num_input_tokens_seen": 358247428,
      "step": 6394
    },
    {
      "epoch": 14.242761692650333,
      "grad_norm": 20.959352493286133,
      "learning_rate": 1e-06,
      "loss": 0.5468,
      "num_input_tokens_seen": 358304388,
      "step": 6395
    },
    {
      "epoch": 14.242761692650333,
      "loss": 0.6914682388305664,
      "loss_ce": 0.00030615812283940613,
      "loss_iou": 0.298828125,
      "loss_num": 0.0185546875,
      "loss_xval": 0.69140625,
      "num_input_tokens_seen": 358304388,
      "step": 6395
    },
    {
      "epoch": 14.244988864142538,
      "grad_norm": 26.55893325805664,
      "learning_rate": 1e-06,
      "loss": 0.5794,
      "num_input_tokens_seen": 358360252,
      "step": 6396
    },
    {
      "epoch": 14.244988864142538,
      "loss": 0.46212613582611084,
      "loss_ce": 8.996979158837348e-05,
      "loss_iou": 0.2021484375,
      "loss_num": 0.01141357421875,
      "loss_xval": 0.462890625,
      "num_input_tokens_seen": 358360252,
      "step": 6396
    },
    {
      "epoch": 14.247216035634743,
      "grad_norm": 18.9809627532959,
      "learning_rate": 1e-06,
      "loss": 0.4802,
      "num_input_tokens_seen": 358418356,
      "step": 6397
    },
    {
      "epoch": 14.247216035634743,
      "loss": 0.3323664665222168,
      "loss_ce": 9.106392099056393e-05,
      "loss_iou": 0.1494140625,
      "loss_num": 0.006561279296875,
      "loss_xval": 0.33203125,
      "num_input_tokens_seen": 358418356,
      "step": 6397
    },
    {
      "epoch": 14.249443207126948,
      "grad_norm": 16.457611083984375,
      "learning_rate": 1e-06,
      "loss": 0.4579,
      "num_input_tokens_seen": 358475404,
      "step": 6398
    },
    {
      "epoch": 14.249443207126948,
      "loss": 0.6108676791191101,
      "loss_ce": 0.00014991118223406374,
      "loss_iou": 0.232421875,
      "loss_num": 0.029296875,
      "loss_xval": 0.609375,
      "num_input_tokens_seen": 358475404,
      "step": 6398
    },
    {
      "epoch": 14.251670378619155,
      "grad_norm": 19.050983428955078,
      "learning_rate": 1e-06,
      "loss": 0.4976,
      "num_input_tokens_seen": 358528924,
      "step": 6399
    },
    {
      "epoch": 14.251670378619155,
      "loss": 0.6439430713653564,
      "loss_ce": 0.0001442273351131007,
      "loss_iou": 0.2890625,
      "loss_num": 0.01312255859375,
      "loss_xval": 0.64453125,
      "num_input_tokens_seen": 358528924,
      "step": 6399
    },
    {
      "epoch": 14.25389755011136,
      "grad_norm": 14.289558410644531,
      "learning_rate": 1e-06,
      "loss": 0.46,
      "num_input_tokens_seen": 358585368,
      "step": 6400
    },
    {
      "epoch": 14.25389755011136,
      "loss": 0.6162996292114258,
      "loss_ce": 8.872315811458975e-05,
      "loss_iou": 0.255859375,
      "loss_num": 0.0206298828125,
      "loss_xval": 0.6171875,
      "num_input_tokens_seen": 358585368,
      "step": 6400
    },
    {
      "epoch": 14.256124721603564,
      "grad_norm": 22.632549285888672,
      "learning_rate": 1e-06,
      "loss": 0.4443,
      "num_input_tokens_seen": 358642216,
      "step": 6401
    },
    {
      "epoch": 14.256124721603564,
      "loss": 0.39599111676216125,
      "loss_ce": 0.00011710106628015637,
      "loss_iou": 0.1708984375,
      "loss_num": 0.01068115234375,
      "loss_xval": 0.396484375,
      "num_input_tokens_seen": 358642216,
      "step": 6401
    },
    {
      "epoch": 14.25835189309577,
      "grad_norm": 15.929319381713867,
      "learning_rate": 1e-06,
      "loss": 0.4255,
      "num_input_tokens_seen": 358698924,
      "step": 6402
    },
    {
      "epoch": 14.25835189309577,
      "loss": 0.35032692551612854,
      "loss_ce": 0.00010718655539676547,
      "loss_iou": 0.1533203125,
      "loss_num": 0.00860595703125,
      "loss_xval": 0.349609375,
      "num_input_tokens_seen": 358698924,
      "step": 6402
    },
    {
      "epoch": 14.260579064587974,
      "grad_norm": 13.306114196777344,
      "learning_rate": 1e-06,
      "loss": 0.3793,
      "num_input_tokens_seen": 358755212,
      "step": 6403
    },
    {
      "epoch": 14.260579064587974,
      "loss": 0.2552165389060974,
      "loss_ce": 8.956858073361218e-05,
      "loss_iou": 0.11181640625,
      "loss_num": 0.0062255859375,
      "loss_xval": 0.255859375,
      "num_input_tokens_seen": 358755212,
      "step": 6403
    },
    {
      "epoch": 14.262806236080179,
      "grad_norm": 17.36546516418457,
      "learning_rate": 1e-06,
      "loss": 0.4423,
      "num_input_tokens_seen": 358814232,
      "step": 6404
    },
    {
      "epoch": 14.262806236080179,
      "loss": 0.4097899794578552,
      "loss_ce": 0.00012203957885503769,
      "loss_iou": 0.1806640625,
      "loss_num": 0.0096435546875,
      "loss_xval": 0.41015625,
      "num_input_tokens_seen": 358814232,
      "step": 6404
    },
    {
      "epoch": 14.265033407572384,
      "grad_norm": 29.2259464263916,
      "learning_rate": 1e-06,
      "loss": 0.5395,
      "num_input_tokens_seen": 358871184,
      "step": 6405
    },
    {
      "epoch": 14.265033407572384,
      "loss": 0.682025671005249,
      "loss_ce": 0.00014085797010920942,
      "loss_iou": 0.28125,
      "loss_num": 0.0235595703125,
      "loss_xval": 0.68359375,
      "num_input_tokens_seen": 358871184,
      "step": 6405
    },
    {
      "epoch": 14.267260579064589,
      "grad_norm": 12.634069442749023,
      "learning_rate": 1e-06,
      "loss": 0.2714,
      "num_input_tokens_seen": 358928648,
      "step": 6406
    },
    {
      "epoch": 14.267260579064589,
      "loss": 0.2787477970123291,
      "loss_ce": 9.179921471513808e-05,
      "loss_iou": 0.1123046875,
      "loss_num": 0.01092529296875,
      "loss_xval": 0.279296875,
      "num_input_tokens_seen": 358928648,
      "step": 6406
    },
    {
      "epoch": 14.269487750556793,
      "grad_norm": 25.011402130126953,
      "learning_rate": 1e-06,
      "loss": 0.5663,
      "num_input_tokens_seen": 358987088,
      "step": 6407
    },
    {
      "epoch": 14.269487750556793,
      "loss": 0.4417472183704376,
      "loss_ce": 9.685405530035496e-05,
      "loss_iou": 0.1845703125,
      "loss_num": 0.01470947265625,
      "loss_xval": 0.44140625,
      "num_input_tokens_seen": 358987088,
      "step": 6407
    },
    {
      "epoch": 14.271714922048998,
      "grad_norm": 14.262114524841309,
      "learning_rate": 1e-06,
      "loss": 0.4572,
      "num_input_tokens_seen": 359043584,
      "step": 6408
    },
    {
      "epoch": 14.271714922048998,
      "loss": 0.40514233708381653,
      "loss_ce": 0.00011304439249215648,
      "loss_iou": 0.1669921875,
      "loss_num": 0.01446533203125,
      "loss_xval": 0.404296875,
      "num_input_tokens_seen": 359043584,
      "step": 6408
    },
    {
      "epoch": 14.273942093541203,
      "grad_norm": 19.64780616760254,
      "learning_rate": 1e-06,
      "loss": 0.3768,
      "num_input_tokens_seen": 359097740,
      "step": 6409
    },
    {
      "epoch": 14.273942093541203,
      "loss": 0.39305636286735535,
      "loss_ce": 0.00011201576853636652,
      "loss_iou": 0.1845703125,
      "loss_num": 0.004852294921875,
      "loss_xval": 0.392578125,
      "num_input_tokens_seen": 359097740,
      "step": 6409
    },
    {
      "epoch": 14.276169265033408,
      "grad_norm": 16.076786041259766,
      "learning_rate": 1e-06,
      "loss": 0.5821,
      "num_input_tokens_seen": 359153996,
      "step": 6410
    },
    {
      "epoch": 14.276169265033408,
      "loss": 0.5809260606765747,
      "loss_ce": 0.00011550600902410224,
      "loss_iou": 0.2470703125,
      "loss_num": 0.0174560546875,
      "loss_xval": 0.58203125,
      "num_input_tokens_seen": 359153996,
      "step": 6410
    },
    {
      "epoch": 14.278396436525613,
      "grad_norm": 14.925246238708496,
      "learning_rate": 1e-06,
      "loss": 0.3981,
      "num_input_tokens_seen": 359209560,
      "step": 6411
    },
    {
      "epoch": 14.278396436525613,
      "loss": 0.4395759403705597,
      "loss_ce": 0.00012282837997190654,
      "loss_iou": 0.19921875,
      "loss_num": 0.00830078125,
      "loss_xval": 0.439453125,
      "num_input_tokens_seen": 359209560,
      "step": 6411
    },
    {
      "epoch": 14.280623608017818,
      "grad_norm": 20.60141372680664,
      "learning_rate": 1e-06,
      "loss": 0.6134,
      "num_input_tokens_seen": 359263596,
      "step": 6412
    },
    {
      "epoch": 14.280623608017818,
      "loss": 0.5152615308761597,
      "loss_ce": 0.00012483607861213386,
      "loss_iou": 0.220703125,
      "loss_num": 0.01483154296875,
      "loss_xval": 0.515625,
      "num_input_tokens_seen": 359263596,
      "step": 6412
    },
    {
      "epoch": 14.282850779510023,
      "grad_norm": 43.90515899658203,
      "learning_rate": 1e-06,
      "loss": 0.6257,
      "num_input_tokens_seen": 359322224,
      "step": 6413
    },
    {
      "epoch": 14.282850779510023,
      "loss": 0.6065540313720703,
      "loss_ce": 0.00010870952974073589,
      "loss_iou": 0.26953125,
      "loss_num": 0.0135498046875,
      "loss_xval": 0.60546875,
      "num_input_tokens_seen": 359322224,
      "step": 6413
    },
    {
      "epoch": 14.285077951002227,
      "grad_norm": 22.319034576416016,
      "learning_rate": 1e-06,
      "loss": 0.3332,
      "num_input_tokens_seen": 359377716,
      "step": 6414
    },
    {
      "epoch": 14.285077951002227,
      "loss": 0.2222072184085846,
      "loss_ce": 0.00010028109682025388,
      "loss_iou": 0.091796875,
      "loss_num": 0.007568359375,
      "loss_xval": 0.2216796875,
      "num_input_tokens_seen": 359377716,
      "step": 6414
    },
    {
      "epoch": 14.287305122494432,
      "grad_norm": 16.046680450439453,
      "learning_rate": 1e-06,
      "loss": 0.4248,
      "num_input_tokens_seen": 359432444,
      "step": 6415
    },
    {
      "epoch": 14.287305122494432,
      "loss": 0.37368685007095337,
      "loss_ce": 9.067119390238076e-05,
      "loss_iou": 0.16015625,
      "loss_num": 0.01080322265625,
      "loss_xval": 0.373046875,
      "num_input_tokens_seen": 359432444,
      "step": 6415
    },
    {
      "epoch": 14.289532293986637,
      "grad_norm": 14.897786140441895,
      "learning_rate": 1e-06,
      "loss": 0.3186,
      "num_input_tokens_seen": 359489464,
      "step": 6416
    },
    {
      "epoch": 14.289532293986637,
      "loss": 0.36753690242767334,
      "loss_ce": 0.00010526920959819108,
      "loss_iou": 0.15625,
      "loss_num": 0.010986328125,
      "loss_xval": 0.3671875,
      "num_input_tokens_seen": 359489464,
      "step": 6416
    },
    {
      "epoch": 14.291759465478842,
      "grad_norm": 17.839426040649414,
      "learning_rate": 1e-06,
      "loss": 0.455,
      "num_input_tokens_seen": 359543156,
      "step": 6417
    },
    {
      "epoch": 14.291759465478842,
      "loss": 0.5617268085479736,
      "loss_ce": 0.0001423186477040872,
      "loss_iou": 0.2294921875,
      "loss_num": 0.0205078125,
      "loss_xval": 0.5625,
      "num_input_tokens_seen": 359543156,
      "step": 6417
    },
    {
      "epoch": 14.293986636971047,
      "grad_norm": 18.264699935913086,
      "learning_rate": 1e-06,
      "loss": 0.5459,
      "num_input_tokens_seen": 359597988,
      "step": 6418
    },
    {
      "epoch": 14.293986636971047,
      "loss": 0.5761609077453613,
      "loss_ce": 0.00011114442168036476,
      "loss_iou": 0.21484375,
      "loss_num": 0.0289306640625,
      "loss_xval": 0.57421875,
      "num_input_tokens_seen": 359597988,
      "step": 6418
    },
    {
      "epoch": 14.296213808463252,
      "grad_norm": 110.11897277832031,
      "learning_rate": 1e-06,
      "loss": 0.4713,
      "num_input_tokens_seen": 359655404,
      "step": 6419
    },
    {
      "epoch": 14.296213808463252,
      "loss": 0.3378644287586212,
      "loss_ce": 0.00015691184671595693,
      "loss_iou": 0.150390625,
      "loss_num": 0.007476806640625,
      "loss_xval": 0.337890625,
      "num_input_tokens_seen": 359655404,
      "step": 6419
    },
    {
      "epoch": 14.298440979955457,
      "grad_norm": 15.50516128540039,
      "learning_rate": 1e-06,
      "loss": 0.3962,
      "num_input_tokens_seen": 359712096,
      "step": 6420
    },
    {
      "epoch": 14.298440979955457,
      "loss": 0.3304358124732971,
      "loss_ce": 0.00011355809692759067,
      "loss_iou": 0.14453125,
      "loss_num": 0.00823974609375,
      "loss_xval": 0.330078125,
      "num_input_tokens_seen": 359712096,
      "step": 6420
    },
    {
      "epoch": 14.300668151447661,
      "grad_norm": 23.723712921142578,
      "learning_rate": 1e-06,
      "loss": 0.5297,
      "num_input_tokens_seen": 359769160,
      "step": 6421
    },
    {
      "epoch": 14.300668151447661,
      "loss": 0.6636959910392761,
      "loss_ce": 0.00012177543976576999,
      "loss_iou": 0.279296875,
      "loss_num": 0.021240234375,
      "loss_xval": 0.6640625,
      "num_input_tokens_seen": 359769160,
      "step": 6421
    },
    {
      "epoch": 14.302895322939866,
      "grad_norm": 13.204316139221191,
      "learning_rate": 1e-06,
      "loss": 0.7653,
      "num_input_tokens_seen": 359824584,
      "step": 6422
    },
    {
      "epoch": 14.302895322939866,
      "loss": 0.5646822452545166,
      "loss_ce": 0.00010707967157941312,
      "loss_iou": 0.2177734375,
      "loss_num": 0.02587890625,
      "loss_xval": 0.56640625,
      "num_input_tokens_seen": 359824584,
      "step": 6422
    },
    {
      "epoch": 14.305122494432071,
      "grad_norm": 25.105751037597656,
      "learning_rate": 1e-06,
      "loss": 0.5937,
      "num_input_tokens_seen": 359877240,
      "step": 6423
    },
    {
      "epoch": 14.305122494432071,
      "loss": 0.6058588027954102,
      "loss_ce": 0.00026802660431712866,
      "loss_iou": 0.25,
      "loss_num": 0.020751953125,
      "loss_xval": 0.60546875,
      "num_input_tokens_seen": 359877240,
      "step": 6423
    },
    {
      "epoch": 14.307349665924276,
      "grad_norm": 20.69352149963379,
      "learning_rate": 1e-06,
      "loss": 0.579,
      "num_input_tokens_seen": 359932336,
      "step": 6424
    },
    {
      "epoch": 14.307349665924276,
      "loss": 0.5356358289718628,
      "loss_ce": 0.00011333586007822305,
      "loss_iou": 0.248046875,
      "loss_num": 0.00811767578125,
      "loss_xval": 0.53515625,
      "num_input_tokens_seen": 359932336,
      "step": 6424
    },
    {
      "epoch": 14.309576837416481,
      "grad_norm": 21.304393768310547,
      "learning_rate": 1e-06,
      "loss": 0.3508,
      "num_input_tokens_seen": 359991588,
      "step": 6425
    },
    {
      "epoch": 14.309576837416481,
      "loss": 0.3918312191963196,
      "loss_ce": 0.0001076057887985371,
      "loss_iou": 0.173828125,
      "loss_num": 0.0089111328125,
      "loss_xval": 0.392578125,
      "num_input_tokens_seen": 359991588,
      "step": 6425
    },
    {
      "epoch": 14.311804008908686,
      "grad_norm": 16.4875545501709,
      "learning_rate": 1e-06,
      "loss": 0.3818,
      "num_input_tokens_seen": 360046080,
      "step": 6426
    },
    {
      "epoch": 14.311804008908686,
      "loss": 0.42865312099456787,
      "loss_ce": 9.47633889154531e-05,
      "loss_iou": 0.1845703125,
      "loss_num": 0.01165771484375,
      "loss_xval": 0.427734375,
      "num_input_tokens_seen": 360046080,
      "step": 6426
    },
    {
      "epoch": 14.31403118040089,
      "grad_norm": 27.263931274414062,
      "learning_rate": 1e-06,
      "loss": 0.4138,
      "num_input_tokens_seen": 360101588,
      "step": 6427
    },
    {
      "epoch": 14.31403118040089,
      "loss": 0.4932384490966797,
      "loss_ce": 0.00013543458771891892,
      "loss_iou": 0.212890625,
      "loss_num": 0.013427734375,
      "loss_xval": 0.4921875,
      "num_input_tokens_seen": 360101588,
      "step": 6427
    },
    {
      "epoch": 14.316258351893095,
      "grad_norm": 15.546148300170898,
      "learning_rate": 1e-06,
      "loss": 0.5737,
      "num_input_tokens_seen": 360157696,
      "step": 6428
    },
    {
      "epoch": 14.316258351893095,
      "loss": 0.774878740310669,
      "loss_ce": 0.00034262199187651277,
      "loss_iou": 0.291015625,
      "loss_num": 0.0380859375,
      "loss_xval": 0.7734375,
      "num_input_tokens_seen": 360157696,
      "step": 6428
    },
    {
      "epoch": 14.3184855233853,
      "grad_norm": 17.817882537841797,
      "learning_rate": 1e-06,
      "loss": 0.5475,
      "num_input_tokens_seen": 360211328,
      "step": 6429
    },
    {
      "epoch": 14.3184855233853,
      "loss": 0.4766749143600464,
      "loss_ce": 0.00011241542233619839,
      "loss_iou": 0.205078125,
      "loss_num": 0.013427734375,
      "loss_xval": 0.4765625,
      "num_input_tokens_seen": 360211328,
      "step": 6429
    },
    {
      "epoch": 14.320712694877505,
      "grad_norm": 26.077342987060547,
      "learning_rate": 1e-06,
      "loss": 0.5338,
      "num_input_tokens_seen": 360266996,
      "step": 6430
    },
    {
      "epoch": 14.320712694877505,
      "loss": 0.5293970704078674,
      "loss_ce": 0.00010019890760304406,
      "loss_iou": 0.224609375,
      "loss_num": 0.01611328125,
      "loss_xval": 0.53125,
      "num_input_tokens_seen": 360266996,
      "step": 6430
    },
    {
      "epoch": 14.32293986636971,
      "grad_norm": 21.88715362548828,
      "learning_rate": 1e-06,
      "loss": 0.5877,
      "num_input_tokens_seen": 360324620,
      "step": 6431
    },
    {
      "epoch": 14.32293986636971,
      "loss": 0.5899729132652283,
      "loss_ce": 0.00012914868420921266,
      "loss_iou": 0.2470703125,
      "loss_num": 0.0191650390625,
      "loss_xval": 0.58984375,
      "num_input_tokens_seen": 360324620,
      "step": 6431
    },
    {
      "epoch": 14.325167037861915,
      "grad_norm": 21.561594009399414,
      "learning_rate": 1e-06,
      "loss": 0.519,
      "num_input_tokens_seen": 360381216,
      "step": 6432
    },
    {
      "epoch": 14.325167037861915,
      "loss": 0.6034383773803711,
      "loss_ce": 0.00016689574113115668,
      "loss_iou": 0.2734375,
      "loss_num": 0.01116943359375,
      "loss_xval": 0.6015625,
      "num_input_tokens_seen": 360381216,
      "step": 6432
    },
    {
      "epoch": 14.32739420935412,
      "grad_norm": 34.43752670288086,
      "learning_rate": 1e-06,
      "loss": 0.8583,
      "num_input_tokens_seen": 360434988,
      "step": 6433
    },
    {
      "epoch": 14.32739420935412,
      "loss": 0.6987577080726624,
      "loss_ce": 0.00014932786871213466,
      "loss_iou": 0.275390625,
      "loss_num": 0.030029296875,
      "loss_xval": 0.69921875,
      "num_input_tokens_seen": 360434988,
      "step": 6433
    },
    {
      "epoch": 14.329621380846325,
      "grad_norm": 14.277132034301758,
      "learning_rate": 1e-06,
      "loss": 0.653,
      "num_input_tokens_seen": 360490608,
      "step": 6434
    },
    {
      "epoch": 14.329621380846325,
      "loss": 0.697412371635437,
      "loss_ce": 0.0001467484253225848,
      "loss_iou": 0.283203125,
      "loss_num": 0.026611328125,
      "loss_xval": 0.6953125,
      "num_input_tokens_seen": 360490608,
      "step": 6434
    },
    {
      "epoch": 14.33184855233853,
      "grad_norm": 18.22037696838379,
      "learning_rate": 1e-06,
      "loss": 0.4797,
      "num_input_tokens_seen": 360547056,
      "step": 6435
    },
    {
      "epoch": 14.33184855233853,
      "loss": 0.4767254590988159,
      "loss_ce": 0.00010193933121627197,
      "loss_iou": 0.2119140625,
      "loss_num": 0.0106201171875,
      "loss_xval": 0.4765625,
      "num_input_tokens_seen": 360547056,
      "step": 6435
    },
    {
      "epoch": 14.334075723830734,
      "grad_norm": 16.72195816040039,
      "learning_rate": 1e-06,
      "loss": 0.3839,
      "num_input_tokens_seen": 360604288,
      "step": 6436
    },
    {
      "epoch": 14.334075723830734,
      "loss": 0.38419631123542786,
      "loss_ce": 0.00010207582090515643,
      "loss_iou": 0.158203125,
      "loss_num": 0.01373291015625,
      "loss_xval": 0.384765625,
      "num_input_tokens_seen": 360604288,
      "step": 6436
    },
    {
      "epoch": 14.33630289532294,
      "grad_norm": 16.082204818725586,
      "learning_rate": 1e-06,
      "loss": 0.5829,
      "num_input_tokens_seen": 360662528,
      "step": 6437
    },
    {
      "epoch": 14.33630289532294,
      "loss": 0.446768581867218,
      "loss_ce": 0.00011331496352795511,
      "loss_iou": 0.19140625,
      "loss_num": 0.01275634765625,
      "loss_xval": 0.447265625,
      "num_input_tokens_seen": 360662528,
      "step": 6437
    },
    {
      "epoch": 14.338530066815144,
      "grad_norm": 19.101377487182617,
      "learning_rate": 1e-06,
      "loss": 0.3889,
      "num_input_tokens_seen": 360717448,
      "step": 6438
    },
    {
      "epoch": 14.338530066815144,
      "loss": 0.5185633897781372,
      "loss_ce": 0.00013074232265353203,
      "loss_iou": 0.216796875,
      "loss_num": 0.016845703125,
      "loss_xval": 0.51953125,
      "num_input_tokens_seen": 360717448,
      "step": 6438
    },
    {
      "epoch": 14.340757238307349,
      "grad_norm": 21.411943435668945,
      "learning_rate": 1e-06,
      "loss": 0.4566,
      "num_input_tokens_seen": 360773220,
      "step": 6439
    },
    {
      "epoch": 14.340757238307349,
      "loss": 0.4258883595466614,
      "loss_ce": 0.00010708505578804761,
      "loss_iou": 0.197265625,
      "loss_num": 0.006378173828125,
      "loss_xval": 0.42578125,
      "num_input_tokens_seen": 360773220,
      "step": 6439
    },
    {
      "epoch": 14.342984409799554,
      "grad_norm": 21.08489418029785,
      "learning_rate": 1e-06,
      "loss": 0.4344,
      "num_input_tokens_seen": 360829380,
      "step": 6440
    },
    {
      "epoch": 14.342984409799554,
      "loss": 0.45614272356033325,
      "loss_ce": 8.802305092103779e-05,
      "loss_iou": 0.185546875,
      "loss_num": 0.0169677734375,
      "loss_xval": 0.45703125,
      "num_input_tokens_seen": 360829380,
      "step": 6440
    },
    {
      "epoch": 14.345211581291759,
      "grad_norm": 19.268587112426758,
      "learning_rate": 1e-06,
      "loss": 0.4262,
      "num_input_tokens_seen": 360885632,
      "step": 6441
    },
    {
      "epoch": 14.345211581291759,
      "loss": 0.561874270439148,
      "loss_ce": 0.00010666967136785388,
      "loss_iou": 0.220703125,
      "loss_num": 0.0240478515625,
      "loss_xval": 0.5625,
      "num_input_tokens_seen": 360885632,
      "step": 6441
    },
    {
      "epoch": 14.347438752783964,
      "grad_norm": 16.14497947692871,
      "learning_rate": 1e-06,
      "loss": 0.4857,
      "num_input_tokens_seen": 360941352,
      "step": 6442
    },
    {
      "epoch": 14.347438752783964,
      "loss": 0.6506056189537048,
      "loss_ce": 9.296346252085641e-05,
      "loss_iou": 0.294921875,
      "loss_num": 0.01239013671875,
      "loss_xval": 0.65234375,
      "num_input_tokens_seen": 360941352,
      "step": 6442
    },
    {
      "epoch": 14.34966592427617,
      "grad_norm": 16.787885665893555,
      "learning_rate": 1e-06,
      "loss": 0.3944,
      "num_input_tokens_seen": 360996836,
      "step": 6443
    },
    {
      "epoch": 14.34966592427617,
      "loss": 0.44542229175567627,
      "loss_ce": 0.00010977737110806629,
      "loss_iou": 0.20703125,
      "loss_num": 0.005950927734375,
      "loss_xval": 0.4453125,
      "num_input_tokens_seen": 360996836,
      "step": 6443
    },
    {
      "epoch": 14.351893095768375,
      "grad_norm": 18.984052658081055,
      "learning_rate": 1e-06,
      "loss": 0.7604,
      "num_input_tokens_seen": 361050092,
      "step": 6444
    },
    {
      "epoch": 14.351893095768375,
      "loss": 0.9335750341415405,
      "loss_ce": 0.00010336375271435827,
      "loss_iou": 0.392578125,
      "loss_num": 0.0294189453125,
      "loss_xval": 0.93359375,
      "num_input_tokens_seen": 361050092,
      "step": 6444
    },
    {
      "epoch": 14.35412026726058,
      "grad_norm": 16.68486213684082,
      "learning_rate": 1e-06,
      "loss": 0.3219,
      "num_input_tokens_seen": 361106880,
      "step": 6445
    },
    {
      "epoch": 14.35412026726058,
      "loss": 0.3571700155735016,
      "loss_ce": 0.00011435095075285062,
      "loss_iou": 0.1630859375,
      "loss_num": 0.00628662109375,
      "loss_xval": 0.357421875,
      "num_input_tokens_seen": 361106880,
      "step": 6445
    },
    {
      "epoch": 14.356347438752785,
      "grad_norm": 31.47307014465332,
      "learning_rate": 1e-06,
      "loss": 0.4125,
      "num_input_tokens_seen": 361161316,
      "step": 6446
    },
    {
      "epoch": 14.356347438752785,
      "loss": 0.3598456084728241,
      "loss_ce": 0.00010440165351610631,
      "loss_iou": 0.1572265625,
      "loss_num": 0.00921630859375,
      "loss_xval": 0.359375,
      "num_input_tokens_seen": 361161316,
      "step": 6446
    },
    {
      "epoch": 14.35857461024499,
      "grad_norm": 20.589462280273438,
      "learning_rate": 1e-06,
      "loss": 0.425,
      "num_input_tokens_seen": 361216012,
      "step": 6447
    },
    {
      "epoch": 14.35857461024499,
      "loss": 0.4665713906288147,
      "loss_ce": 0.0001407517702318728,
      "loss_iou": 0.201171875,
      "loss_num": 0.0126953125,
      "loss_xval": 0.466796875,
      "num_input_tokens_seen": 361216012,
      "step": 6447
    },
    {
      "epoch": 14.360801781737194,
      "grad_norm": 31.11367416381836,
      "learning_rate": 1e-06,
      "loss": 0.5042,
      "num_input_tokens_seen": 361273420,
      "step": 6448
    },
    {
      "epoch": 14.360801781737194,
      "loss": 0.5230991244316101,
      "loss_ce": 0.00014988121984060854,
      "loss_iou": 0.2373046875,
      "loss_num": 0.00958251953125,
      "loss_xval": 0.5234375,
      "num_input_tokens_seen": 361273420,
      "step": 6448
    },
    {
      "epoch": 14.3630289532294,
      "grad_norm": 21.973804473876953,
      "learning_rate": 1e-06,
      "loss": 0.5315,
      "num_input_tokens_seen": 361327552,
      "step": 6449
    },
    {
      "epoch": 14.3630289532294,
      "loss": 0.5520008206367493,
      "loss_ce": 0.00012092564429622144,
      "loss_iou": 0.2490234375,
      "loss_num": 0.0107421875,
      "loss_xval": 0.55078125,
      "num_input_tokens_seen": 361327552,
      "step": 6449
    },
    {
      "epoch": 14.365256124721604,
      "grad_norm": 16.959115982055664,
      "learning_rate": 1e-06,
      "loss": 0.5215,
      "num_input_tokens_seen": 361384308,
      "step": 6450
    },
    {
      "epoch": 14.365256124721604,
      "loss": 0.5186026096343994,
      "loss_ce": 0.0001089318175218068,
      "loss_iou": 0.2158203125,
      "loss_num": 0.0172119140625,
      "loss_xval": 0.51953125,
      "num_input_tokens_seen": 361384308,
      "step": 6450
    },
    {
      "epoch": 14.367483296213809,
      "grad_norm": 18.288415908813477,
      "learning_rate": 1e-06,
      "loss": 0.3716,
      "num_input_tokens_seen": 361441164,
      "step": 6451
    },
    {
      "epoch": 14.367483296213809,
      "loss": 0.2685241103172302,
      "loss_ce": 9.149497782345861e-05,
      "loss_iou": 0.11181640625,
      "loss_num": 0.0089111328125,
      "loss_xval": 0.267578125,
      "num_input_tokens_seen": 361441164,
      "step": 6451
    },
    {
      "epoch": 14.369710467706014,
      "grad_norm": 22.0161190032959,
      "learning_rate": 1e-06,
      "loss": 0.4447,
      "num_input_tokens_seen": 361500032,
      "step": 6452
    },
    {
      "epoch": 14.369710467706014,
      "loss": 0.5349976420402527,
      "loss_ce": 0.0001465780078433454,
      "loss_iou": 0.2373046875,
      "loss_num": 0.01190185546875,
      "loss_xval": 0.53515625,
      "num_input_tokens_seen": 361500032,
      "step": 6452
    },
    {
      "epoch": 14.371937639198219,
      "grad_norm": 22.7355899810791,
      "learning_rate": 1e-06,
      "loss": 0.4947,
      "num_input_tokens_seen": 361555872,
      "step": 6453
    },
    {
      "epoch": 14.371937639198219,
      "loss": 0.46275314688682556,
      "loss_ce": 0.00010668374306987971,
      "loss_iou": 0.2021484375,
      "loss_num": 0.01171875,
      "loss_xval": 0.462890625,
      "num_input_tokens_seen": 361555872,
      "step": 6453
    },
    {
      "epoch": 14.374164810690424,
      "grad_norm": 25.038970947265625,
      "learning_rate": 1e-06,
      "loss": 0.4937,
      "num_input_tokens_seen": 361614468,
      "step": 6454
    },
    {
      "epoch": 14.374164810690424,
      "loss": 0.5680195093154907,
      "loss_ce": 8.73721728567034e-05,
      "loss_iou": 0.232421875,
      "loss_num": 0.0208740234375,
      "loss_xval": 0.56640625,
      "num_input_tokens_seen": 361614468,
      "step": 6454
    },
    {
      "epoch": 14.376391982182628,
      "grad_norm": 18.690160751342773,
      "learning_rate": 1e-06,
      "loss": 0.6027,
      "num_input_tokens_seen": 361670736,
      "step": 6455
    },
    {
      "epoch": 14.376391982182628,
      "loss": 0.7412841320037842,
      "loss_ce": 0.00043940101750195026,
      "loss_iou": 0.306640625,
      "loss_num": 0.0257568359375,
      "loss_xval": 0.7421875,
      "num_input_tokens_seen": 361670736,
      "step": 6455
    },
    {
      "epoch": 14.378619153674833,
      "grad_norm": 13.359357833862305,
      "learning_rate": 1e-06,
      "loss": 0.4412,
      "num_input_tokens_seen": 361725704,
      "step": 6456
    },
    {
      "epoch": 14.378619153674833,
      "loss": 0.3059990406036377,
      "loss_ce": 9.084792691282928e-05,
      "loss_iou": 0.134765625,
      "loss_num": 0.00738525390625,
      "loss_xval": 0.306640625,
      "num_input_tokens_seen": 361725704,
      "step": 6456
    },
    {
      "epoch": 14.380846325167038,
      "grad_norm": 18.930578231811523,
      "learning_rate": 1e-06,
      "loss": 0.4059,
      "num_input_tokens_seen": 361782364,
      "step": 6457
    },
    {
      "epoch": 14.380846325167038,
      "loss": 0.4163687527179718,
      "loss_ce": 0.00010898528853431344,
      "loss_iou": 0.19140625,
      "loss_num": 0.0068359375,
      "loss_xval": 0.416015625,
      "num_input_tokens_seen": 361782364,
      "step": 6457
    },
    {
      "epoch": 14.383073496659243,
      "grad_norm": 14.719660758972168,
      "learning_rate": 1e-06,
      "loss": 0.4225,
      "num_input_tokens_seen": 361836700,
      "step": 6458
    },
    {
      "epoch": 14.383073496659243,
      "loss": 0.41648274660110474,
      "loss_ce": 0.0001009054685709998,
      "loss_iou": 0.181640625,
      "loss_num": 0.0107421875,
      "loss_xval": 0.416015625,
      "num_input_tokens_seen": 361836700,
      "step": 6458
    },
    {
      "epoch": 14.385300668151448,
      "grad_norm": 17.509685516357422,
      "learning_rate": 1e-06,
      "loss": 0.4016,
      "num_input_tokens_seen": 361894212,
      "step": 6459
    },
    {
      "epoch": 14.385300668151448,
      "loss": 0.42406773567199707,
      "loss_ce": 0.0001785791973816231,
      "loss_iou": 0.16015625,
      "loss_num": 0.0206298828125,
      "loss_xval": 0.423828125,
      "num_input_tokens_seen": 361894212,
      "step": 6459
    },
    {
      "epoch": 14.387527839643653,
      "grad_norm": 21.18560218811035,
      "learning_rate": 1e-06,
      "loss": 0.6036,
      "num_input_tokens_seen": 361949500,
      "step": 6460
    },
    {
      "epoch": 14.387527839643653,
      "loss": 0.5647926926612854,
      "loss_ce": 9.54304778133519e-05,
      "loss_iou": 0.25390625,
      "loss_num": 0.01104736328125,
      "loss_xval": 0.56640625,
      "num_input_tokens_seen": 361949500,
      "step": 6460
    },
    {
      "epoch": 14.389755011135858,
      "grad_norm": 18.161174774169922,
      "learning_rate": 1e-06,
      "loss": 0.4029,
      "num_input_tokens_seen": 362002964,
      "step": 6461
    },
    {
      "epoch": 14.389755011135858,
      "loss": 0.3434833586215973,
      "loss_ce": 9.955540008377284e-05,
      "loss_iou": 0.138671875,
      "loss_num": 0.01336669921875,
      "loss_xval": 0.34375,
      "num_input_tokens_seen": 362002964,
      "step": 6461
    },
    {
      "epoch": 14.391982182628063,
      "grad_norm": 18.891141891479492,
      "learning_rate": 1e-06,
      "loss": 0.4224,
      "num_input_tokens_seen": 362056564,
      "step": 6462
    },
    {
      "epoch": 14.391982182628063,
      "loss": 0.3922403156757355,
      "loss_ce": 0.00015046796761453152,
      "loss_iou": 0.16796875,
      "loss_num": 0.0111083984375,
      "loss_xval": 0.392578125,
      "num_input_tokens_seen": 362056564,
      "step": 6462
    },
    {
      "epoch": 14.394209354120267,
      "grad_norm": 42.18777847290039,
      "learning_rate": 1e-06,
      "loss": 0.5473,
      "num_input_tokens_seen": 362113504,
      "step": 6463
    },
    {
      "epoch": 14.394209354120267,
      "loss": 0.6558715105056763,
      "loss_ce": 0.00010980549268424511,
      "loss_iou": 0.30078125,
      "loss_num": 0.01080322265625,
      "loss_xval": 0.65625,
      "num_input_tokens_seen": 362113504,
      "step": 6463
    },
    {
      "epoch": 14.396436525612472,
      "grad_norm": 17.04292869567871,
      "learning_rate": 1e-06,
      "loss": 0.5568,
      "num_input_tokens_seen": 362168292,
      "step": 6464
    },
    {
      "epoch": 14.396436525612472,
      "loss": 0.5546635985374451,
      "loss_ce": 9.817900718189776e-05,
      "loss_iou": 0.2490234375,
      "loss_num": 0.01104736328125,
      "loss_xval": 0.5546875,
      "num_input_tokens_seen": 362168292,
      "step": 6464
    },
    {
      "epoch": 14.398663697104677,
      "grad_norm": 18.7966251373291,
      "learning_rate": 1e-06,
      "loss": 0.4122,
      "num_input_tokens_seen": 362226676,
      "step": 6465
    },
    {
      "epoch": 14.398663697104677,
      "loss": 0.5053236484527588,
      "loss_ce": 0.0001357014843961224,
      "loss_iou": 0.2099609375,
      "loss_num": 0.01708984375,
      "loss_xval": 0.50390625,
      "num_input_tokens_seen": 362226676,
      "step": 6465
    },
    {
      "epoch": 14.400890868596882,
      "grad_norm": 17.50356674194336,
      "learning_rate": 1e-06,
      "loss": 0.4747,
      "num_input_tokens_seen": 362285324,
      "step": 6466
    },
    {
      "epoch": 14.400890868596882,
      "loss": 0.6701542735099792,
      "loss_ce": 0.00011035045463358983,
      "loss_iou": 0.283203125,
      "loss_num": 0.02099609375,
      "loss_xval": 0.671875,
      "num_input_tokens_seen": 362285324,
      "step": 6466
    },
    {
      "epoch": 14.403118040089087,
      "grad_norm": 18.688034057617188,
      "learning_rate": 1e-06,
      "loss": 0.4448,
      "num_input_tokens_seen": 362343140,
      "step": 6467
    },
    {
      "epoch": 14.403118040089087,
      "loss": 0.6092594861984253,
      "loss_ce": 0.00012863794108852744,
      "loss_iou": 0.26171875,
      "loss_num": 0.0172119140625,
      "loss_xval": 0.609375,
      "num_input_tokens_seen": 362343140,
      "step": 6467
    },
    {
      "epoch": 14.405345211581292,
      "grad_norm": 25.48725700378418,
      "learning_rate": 1e-06,
      "loss": 0.5325,
      "num_input_tokens_seen": 362401224,
      "step": 6468
    },
    {
      "epoch": 14.405345211581292,
      "loss": 0.5204082131385803,
      "loss_ce": 0.0001445315283490345,
      "loss_iou": 0.236328125,
      "loss_num": 0.009765625,
      "loss_xval": 0.51953125,
      "num_input_tokens_seen": 362401224,
      "step": 6468
    },
    {
      "epoch": 14.407572383073497,
      "grad_norm": 19.30179786682129,
      "learning_rate": 1e-06,
      "loss": 0.4949,
      "num_input_tokens_seen": 362459744,
      "step": 6469
    },
    {
      "epoch": 14.407572383073497,
      "loss": 0.3543410301208496,
      "loss_ce": 9.295364725403488e-05,
      "loss_iou": 0.15234375,
      "loss_num": 0.00994873046875,
      "loss_xval": 0.353515625,
      "num_input_tokens_seen": 362459744,
      "step": 6469
    },
    {
      "epoch": 14.409799554565701,
      "grad_norm": 19.212997436523438,
      "learning_rate": 1e-06,
      "loss": 0.5026,
      "num_input_tokens_seen": 362516148,
      "step": 6470
    },
    {
      "epoch": 14.409799554565701,
      "loss": 0.34019219875335693,
      "loss_ce": 0.00010428522364236414,
      "loss_iou": 0.1494140625,
      "loss_num": 0.0084228515625,
      "loss_xval": 0.33984375,
      "num_input_tokens_seen": 362516148,
      "step": 6470
    },
    {
      "epoch": 14.412026726057906,
      "grad_norm": 20.393836975097656,
      "learning_rate": 1e-06,
      "loss": 0.5051,
      "num_input_tokens_seen": 362571256,
      "step": 6471
    },
    {
      "epoch": 14.412026726057906,
      "loss": 0.49153202772140503,
      "loss_ce": 0.0003210945869795978,
      "loss_iou": 0.1923828125,
      "loss_num": 0.021240234375,
      "loss_xval": 0.4921875,
      "num_input_tokens_seen": 362571256,
      "step": 6471
    },
    {
      "epoch": 14.414253897550111,
      "grad_norm": 48.49162292480469,
      "learning_rate": 1e-06,
      "loss": 0.4618,
      "num_input_tokens_seen": 362627180,
      "step": 6472
    },
    {
      "epoch": 14.414253897550111,
      "loss": 0.5423309206962585,
      "loss_ce": 9.457490523345768e-05,
      "loss_iou": 0.2255859375,
      "loss_num": 0.0181884765625,
      "loss_xval": 0.54296875,
      "num_input_tokens_seen": 362627180,
      "step": 6472
    },
    {
      "epoch": 14.416481069042316,
      "grad_norm": 38.27447509765625,
      "learning_rate": 1e-06,
      "loss": 0.6114,
      "num_input_tokens_seen": 362681880,
      "step": 6473
    },
    {
      "epoch": 14.416481069042316,
      "loss": 0.6459391117095947,
      "loss_ce": 0.00012608422548510134,
      "loss_iou": 0.2734375,
      "loss_num": 0.020263671875,
      "loss_xval": 0.64453125,
      "num_input_tokens_seen": 362681880,
      "step": 6473
    },
    {
      "epoch": 14.41870824053452,
      "grad_norm": 25.192468643188477,
      "learning_rate": 1e-06,
      "loss": 0.5313,
      "num_input_tokens_seen": 362738364,
      "step": 6474
    },
    {
      "epoch": 14.41870824053452,
      "loss": 0.44662946462631226,
      "loss_ce": 9.624061931390315e-05,
      "loss_iou": 0.2060546875,
      "loss_num": 0.006927490234375,
      "loss_xval": 0.447265625,
      "num_input_tokens_seen": 362738364,
      "step": 6474
    },
    {
      "epoch": 14.420935412026726,
      "grad_norm": 19.721994400024414,
      "learning_rate": 1e-06,
      "loss": 0.5204,
      "num_input_tokens_seen": 362793976,
      "step": 6475
    },
    {
      "epoch": 14.420935412026726,
      "loss": 0.530852198600769,
      "loss_ce": 9.049595973920077e-05,
      "loss_iou": 0.208984375,
      "loss_num": 0.022705078125,
      "loss_xval": 0.53125,
      "num_input_tokens_seen": 362793976,
      "step": 6475
    },
    {
      "epoch": 14.42316258351893,
      "grad_norm": 25.936969757080078,
      "learning_rate": 1e-06,
      "loss": 0.5892,
      "num_input_tokens_seen": 362849184,
      "step": 6476
    },
    {
      "epoch": 14.42316258351893,
      "loss": 0.6352825164794922,
      "loss_ce": 0.00015069330402184278,
      "loss_iou": 0.27734375,
      "loss_num": 0.016357421875,
      "loss_xval": 0.63671875,
      "num_input_tokens_seen": 362849184,
      "step": 6476
    },
    {
      "epoch": 14.425389755011135,
      "grad_norm": 14.354238510131836,
      "learning_rate": 1e-06,
      "loss": 0.3588,
      "num_input_tokens_seen": 362902516,
      "step": 6477
    },
    {
      "epoch": 14.425389755011135,
      "loss": 0.4610298275947571,
      "loss_ce": 9.230234718415886e-05,
      "loss_iou": 0.1728515625,
      "loss_num": 0.02294921875,
      "loss_xval": 0.4609375,
      "num_input_tokens_seen": 362902516,
      "step": 6477
    },
    {
      "epoch": 14.42761692650334,
      "grad_norm": 16.179262161254883,
      "learning_rate": 1e-06,
      "loss": 0.4084,
      "num_input_tokens_seen": 362954364,
      "step": 6478
    },
    {
      "epoch": 14.42761692650334,
      "loss": 0.2978529930114746,
      "loss_ce": 9.299164230469614e-05,
      "loss_iou": 0.130859375,
      "loss_num": 0.00726318359375,
      "loss_xval": 0.296875,
      "num_input_tokens_seen": 362954364,
      "step": 6478
    },
    {
      "epoch": 14.429844097995545,
      "grad_norm": 38.88999557495117,
      "learning_rate": 1e-06,
      "loss": 0.4598,
      "num_input_tokens_seen": 363011612,
      "step": 6479
    },
    {
      "epoch": 14.429844097995545,
      "loss": 0.5600452423095703,
      "loss_ce": 0.00016974634490907192,
      "loss_iou": 0.26171875,
      "loss_num": 0.007354736328125,
      "loss_xval": 0.55859375,
      "num_input_tokens_seen": 363011612,
      "step": 6479
    },
    {
      "epoch": 14.43207126948775,
      "grad_norm": 34.526512145996094,
      "learning_rate": 1e-06,
      "loss": 0.4824,
      "num_input_tokens_seen": 363069500,
      "step": 6480
    },
    {
      "epoch": 14.43207126948775,
      "loss": 0.4800405502319336,
      "loss_ce": 0.00012108102964702994,
      "loss_iou": 0.193359375,
      "loss_num": 0.0184326171875,
      "loss_xval": 0.48046875,
      "num_input_tokens_seen": 363069500,
      "step": 6480
    },
    {
      "epoch": 14.434298440979955,
      "grad_norm": 24.97442626953125,
      "learning_rate": 1e-06,
      "loss": 0.5423,
      "num_input_tokens_seen": 363125684,
      "step": 6481
    },
    {
      "epoch": 14.434298440979955,
      "loss": 0.3962355852127075,
      "loss_ce": 0.00011745323718059808,
      "loss_iou": 0.1728515625,
      "loss_num": 0.0103759765625,
      "loss_xval": 0.396484375,
      "num_input_tokens_seen": 363125684,
      "step": 6481
    },
    {
      "epoch": 14.43652561247216,
      "grad_norm": 13.889195442199707,
      "learning_rate": 1e-06,
      "loss": 0.4004,
      "num_input_tokens_seen": 363181616,
      "step": 6482
    },
    {
      "epoch": 14.43652561247216,
      "loss": 0.4475998878479004,
      "loss_ce": 0.0001511801965534687,
      "loss_iou": 0.1943359375,
      "loss_num": 0.01190185546875,
      "loss_xval": 0.447265625,
      "num_input_tokens_seen": 363181616,
      "step": 6482
    },
    {
      "epoch": 14.438752783964365,
      "grad_norm": 16.7701358795166,
      "learning_rate": 1e-06,
      "loss": 0.3514,
      "num_input_tokens_seen": 363238068,
      "step": 6483
    },
    {
      "epoch": 14.438752783964365,
      "loss": 0.38326865434646606,
      "loss_ce": 8.997365512186661e-05,
      "loss_iou": 0.162109375,
      "loss_num": 0.0118408203125,
      "loss_xval": 0.3828125,
      "num_input_tokens_seen": 363238068,
      "step": 6483
    },
    {
      "epoch": 14.44097995545657,
      "grad_norm": 16.830076217651367,
      "learning_rate": 1e-06,
      "loss": 0.4811,
      "num_input_tokens_seen": 363293032,
      "step": 6484
    },
    {
      "epoch": 14.44097995545657,
      "loss": 0.4167139530181885,
      "loss_ce": 8.8007356680464e-05,
      "loss_iou": 0.1806640625,
      "loss_num": 0.01092529296875,
      "loss_xval": 0.416015625,
      "num_input_tokens_seen": 363293032,
      "step": 6484
    },
    {
      "epoch": 14.443207126948774,
      "grad_norm": 13.648219108581543,
      "learning_rate": 1e-06,
      "loss": 0.543,
      "num_input_tokens_seen": 363349520,
      "step": 6485
    },
    {
      "epoch": 14.443207126948774,
      "loss": 0.47776174545288086,
      "loss_ce": 0.0002226911747129634,
      "loss_iou": 0.1865234375,
      "loss_num": 0.0206298828125,
      "loss_xval": 0.4765625,
      "num_input_tokens_seen": 363349520,
      "step": 6485
    },
    {
      "epoch": 14.44543429844098,
      "grad_norm": 14.256153106689453,
      "learning_rate": 1e-06,
      "loss": 0.4144,
      "num_input_tokens_seen": 363407380,
      "step": 6486
    },
    {
      "epoch": 14.44543429844098,
      "loss": 0.21311235427856445,
      "loss_ce": 9.964955097530037e-05,
      "loss_iou": 0.0927734375,
      "loss_num": 0.00543212890625,
      "loss_xval": 0.212890625,
      "num_input_tokens_seen": 363407380,
      "step": 6486
    },
    {
      "epoch": 14.447661469933184,
      "grad_norm": 20.257829666137695,
      "learning_rate": 1e-06,
      "loss": 0.4631,
      "num_input_tokens_seen": 363463888,
      "step": 6487
    },
    {
      "epoch": 14.447661469933184,
      "loss": 0.4675000011920929,
      "loss_ce": 9.276315540773794e-05,
      "loss_iou": 0.193359375,
      "loss_num": 0.01611328125,
      "loss_xval": 0.466796875,
      "num_input_tokens_seen": 363463888,
      "step": 6487
    },
    {
      "epoch": 14.449888641425389,
      "grad_norm": 15.722443580627441,
      "learning_rate": 1e-06,
      "loss": 0.401,
      "num_input_tokens_seen": 363520880,
      "step": 6488
    },
    {
      "epoch": 14.449888641425389,
      "loss": 0.3268738389015198,
      "loss_ce": 9.159540786640719e-05,
      "loss_iou": 0.142578125,
      "loss_num": 0.0084228515625,
      "loss_xval": 0.326171875,
      "num_input_tokens_seen": 363520880,
      "step": 6488
    },
    {
      "epoch": 14.452115812917596,
      "grad_norm": 24.610681533813477,
      "learning_rate": 1e-06,
      "loss": 0.5481,
      "num_input_tokens_seen": 363575452,
      "step": 6489
    },
    {
      "epoch": 14.452115812917596,
      "loss": 0.42991262674331665,
      "loss_ce": 0.00010304449824616313,
      "loss_iou": 0.19140625,
      "loss_num": 0.00933837890625,
      "loss_xval": 0.4296875,
      "num_input_tokens_seen": 363575452,
      "step": 6489
    },
    {
      "epoch": 14.4543429844098,
      "grad_norm": 15.40800952911377,
      "learning_rate": 1e-06,
      "loss": 0.32,
      "num_input_tokens_seen": 363630124,
      "step": 6490
    },
    {
      "epoch": 14.4543429844098,
      "loss": 0.35164231061935425,
      "loss_ce": 9.507540380582213e-05,
      "loss_iou": 0.15625,
      "loss_num": 0.007659912109375,
      "loss_xval": 0.3515625,
      "num_input_tokens_seen": 363630124,
      "step": 6490
    },
    {
      "epoch": 14.456570155902005,
      "grad_norm": 27.49700927734375,
      "learning_rate": 1e-06,
      "loss": 0.4365,
      "num_input_tokens_seen": 363683792,
      "step": 6491
    },
    {
      "epoch": 14.456570155902005,
      "loss": 0.5234502553939819,
      "loss_ce": 0.00013481616042554379,
      "loss_iou": 0.2412109375,
      "loss_num": 0.00823974609375,
      "loss_xval": 0.5234375,
      "num_input_tokens_seen": 363683792,
      "step": 6491
    },
    {
      "epoch": 14.45879732739421,
      "grad_norm": 16.735675811767578,
      "learning_rate": 1e-06,
      "loss": 0.359,
      "num_input_tokens_seen": 363740708,
      "step": 6492
    },
    {
      "epoch": 14.45879732739421,
      "loss": 0.35196787118911743,
      "loss_ce": 0.00010018555622082204,
      "loss_iou": 0.13671875,
      "loss_num": 0.0157470703125,
      "loss_xval": 0.3515625,
      "num_input_tokens_seen": 363740708,
      "step": 6492
    },
    {
      "epoch": 14.461024498886415,
      "grad_norm": 27.95579719543457,
      "learning_rate": 1e-06,
      "loss": 0.5228,
      "num_input_tokens_seen": 363795664,
      "step": 6493
    },
    {
      "epoch": 14.461024498886415,
      "loss": 0.5091565847396851,
      "loss_ce": 0.00012336287181824446,
      "loss_iou": 0.2333984375,
      "loss_num": 0.00836181640625,
      "loss_xval": 0.5078125,
      "num_input_tokens_seen": 363795664,
      "step": 6493
    },
    {
      "epoch": 14.46325167037862,
      "grad_norm": 20.272802352905273,
      "learning_rate": 1e-06,
      "loss": 0.4311,
      "num_input_tokens_seen": 363853956,
      "step": 6494
    },
    {
      "epoch": 14.46325167037862,
      "loss": 0.4811999201774597,
      "loss_ce": 0.00012079046427970752,
      "loss_iou": 0.220703125,
      "loss_num": 0.00799560546875,
      "loss_xval": 0.48046875,
      "num_input_tokens_seen": 363853956,
      "step": 6494
    },
    {
      "epoch": 14.465478841870825,
      "grad_norm": 18.344619750976562,
      "learning_rate": 1e-06,
      "loss": 0.2613,
      "num_input_tokens_seen": 363911696,
      "step": 6495
    },
    {
      "epoch": 14.465478841870825,
      "loss": 0.237228661775589,
      "loss_ce": 0.00010707815818022937,
      "loss_iou": 0.1044921875,
      "loss_num": 0.005523681640625,
      "loss_xval": 0.2373046875,
      "num_input_tokens_seen": 363911696,
      "step": 6495
    },
    {
      "epoch": 14.46770601336303,
      "grad_norm": 35.97431564331055,
      "learning_rate": 1e-06,
      "loss": 0.8532,
      "num_input_tokens_seen": 363966144,
      "step": 6496
    },
    {
      "epoch": 14.46770601336303,
      "loss": 0.8004045486450195,
      "loss_ce": 0.00011163462477270514,
      "loss_iou": 0.34375,
      "loss_num": 0.0223388671875,
      "loss_xval": 0.80078125,
      "num_input_tokens_seen": 363966144,
      "step": 6496
    },
    {
      "epoch": 14.469933184855234,
      "grad_norm": 33.640872955322266,
      "learning_rate": 1e-06,
      "loss": 0.5273,
      "num_input_tokens_seen": 364021896,
      "step": 6497
    },
    {
      "epoch": 14.469933184855234,
      "loss": 0.3864748775959015,
      "loss_ce": 0.00012233102461323142,
      "loss_iou": 0.1669921875,
      "loss_num": 0.01055908203125,
      "loss_xval": 0.38671875,
      "num_input_tokens_seen": 364021896,
      "step": 6497
    },
    {
      "epoch": 14.47216035634744,
      "grad_norm": 18.709012985229492,
      "learning_rate": 1e-06,
      "loss": 0.4962,
      "num_input_tokens_seen": 364079360,
      "step": 6498
    },
    {
      "epoch": 14.47216035634744,
      "loss": 0.40141546726226807,
      "loss_ce": 0.00010935450700344518,
      "loss_iou": 0.1708984375,
      "loss_num": 0.0118408203125,
      "loss_xval": 0.400390625,
      "num_input_tokens_seen": 364079360,
      "step": 6498
    },
    {
      "epoch": 14.474387527839644,
      "grad_norm": 18.93471908569336,
      "learning_rate": 1e-06,
      "loss": 0.3134,
      "num_input_tokens_seen": 364136176,
      "step": 6499
    },
    {
      "epoch": 14.474387527839644,
      "loss": 0.36948075890541077,
      "loss_ce": 0.00021804316202178597,
      "loss_iou": 0.1591796875,
      "loss_num": 0.01007080078125,
      "loss_xval": 0.369140625,
      "num_input_tokens_seen": 364136176,
      "step": 6499
    },
    {
      "epoch": 14.476614699331849,
      "grad_norm": 21.971736907958984,
      "learning_rate": 1e-06,
      "loss": 0.4228,
      "num_input_tokens_seen": 364192972,
      "step": 6500
    },
    {
      "epoch": 14.476614699331849,
      "eval_seeclick_web_CIoU": 0.5850892961025238,
      "eval_seeclick_web_GIoU": 0.5826087892055511,
      "eval_seeclick_web_IoU": 0.6035565435886383,
      "eval_seeclick_web_MAE_all": 0.015261294320225716,
      "eval_seeclick_web_MAE_h": 0.007648690138012171,
      "eval_seeclick_web_MAE_w": 0.015442864038050175,
      "eval_seeclick_web_MAE_x_boxes": 0.009242635453119874,
      "eval_seeclick_web_MAE_y_boxes": 0.021023853914812207,
      "eval_seeclick_web_inside_bbox": 0.9166666567325592,
      "eval_seeclick_web_loss": 0.9148587584495544,
      "eval_seeclick_web_loss_ce": 0.0001663194634602405,
      "eval_seeclick_web_loss_iou": 0.4219970703125,
      "eval_seeclick_web_loss_num": 0.01227569580078125,
      "eval_seeclick_web_loss_xval": 0.9052734375,
      "eval_seeclick_web_runtime": 22.0534,
      "eval_seeclick_web_samples_per_second": 2.267,
      "eval_seeclick_web_steps_per_second": 0.091,
      "num_input_tokens_seen": 364192972,
      "step": 6500
    },
    {
      "epoch": 14.476614699331849,
      "eval_icons_CIoU": 0.294839546084404,
      "eval_icons_GIoU": 0.3122626394033432,
      "eval_icons_IoU": 0.36477692425251007,
      "eval_icons_MAE_all": 0.05375087633728981,
      "eval_icons_MAE_h": 0.032996498979628086,
      "eval_icons_MAE_w": 0.051369220949709415,
      "eval_icons_MAE_x_boxes": 0.05044420249760151,
      "eval_icons_MAE_y_boxes": 0.03664529975503683,
      "eval_icons_inside_bbox": 0.6649305522441864,
      "eval_icons_loss": 1.6395889520645142,
      "eval_icons_loss_ce": 0.00020623258751584217,
      "eval_icons_loss_iou": 0.659423828125,
      "eval_icons_loss_num": 0.05293846130371094,
      "eval_icons_loss_xval": 1.583984375,
      "eval_icons_runtime": 19.0398,
      "eval_icons_samples_per_second": 2.626,
      "eval_icons_steps_per_second": 0.105,
      "num_input_tokens_seen": 364192972,
      "step": 6500
    },
    {
      "epoch": 14.476614699331849,
      "eval_screenspot_CIoU": 0.37580615282058716,
      "eval_screenspot_GIoU": 0.3901708126068115,
      "eval_screenspot_IoU": 0.44902459780375165,
      "eval_screenspot_MAE_all": 0.054527596880992256,
      "eval_screenspot_MAE_h": 0.03962646176417669,
      "eval_screenspot_MAE_w": 0.061993442475795746,
      "eval_screenspot_MAE_x_boxes": 0.07007026796539624,
      "eval_screenspot_MAE_y_boxes": 0.037376622669398785,
      "eval_screenspot_inside_bbox": 0.7041666706403097,
      "eval_screenspot_loss": 1.547979474067688,
      "eval_screenspot_loss_ce": 0.00023582103798010698,
      "eval_screenspot_loss_iou": 0.646484375,
      "eval_screenspot_loss_num": 0.06189727783203125,
      "eval_screenspot_loss_xval": 1.60205078125,
      "eval_screenspot_runtime": 35.3553,
      "eval_screenspot_samples_per_second": 2.517,
      "eval_screenspot_steps_per_second": 0.085,
      "num_input_tokens_seen": 364192972,
      "step": 6500
    },
    {
      "epoch": 14.476614699331849,
      "eval_compot_CIoU": 0.3404065817594528,
      "eval_compot_GIoU": 0.35687774419784546,
      "eval_compot_IoU": 0.40112267434597015,
      "eval_compot_MAE_all": 0.01965143345296383,
      "eval_compot_MAE_h": 0.013854971155524254,
      "eval_compot_MAE_w": 0.02131012175232172,
      "eval_compot_MAE_x_boxes": 0.030210323631763458,
      "eval_compot_MAE_y_boxes": 0.006537551758810878,
      "eval_compot_inside_bbox": 0.6302083432674408,
      "eval_compot_loss": 1.4148516654968262,
      "eval_compot_loss_ce": 0.00016382582543883473,
      "eval_compot_loss_iou": 0.649169921875,
      "eval_compot_loss_num": 0.018445968627929688,
      "eval_compot_loss_xval": 1.390380859375,
      "eval_compot_runtime": 21.0154,
      "eval_compot_samples_per_second": 2.379,
      "eval_compot_steps_per_second": 0.095,
      "num_input_tokens_seen": 364192972,
      "step": 6500
    },
    {
      "epoch": 14.476614699331849,
      "eval_custom_ui_val_CIoU": 0.4791228680147065,
      "eval_custom_ui_val_GIoU": 0.4861221942636702,
      "eval_custom_ui_val_IoU": 0.5396713250213199,
      "eval_custom_ui_val_MAE_all": 0.02759691560640931,
      "eval_custom_ui_val_MAE_h": 0.015171076895462142,
      "eval_custom_ui_val_MAE_w": 0.03542729518893692,
      "eval_custom_ui_val_MAE_x_boxes": 0.03368757442674703,
      "eval_custom_ui_val_MAE_y_boxes": 0.013593513725532426,
      "eval_custom_ui_val_inside_bbox": 0.7789351873927646,
      "eval_custom_ui_val_loss": 1.167110800743103,
      "eval_custom_ui_val_loss_ce": 0.00018443458328773786,
      "eval_custom_ui_val_loss_iou": 0.5000135633680556,
      "eval_custom_ui_val_loss_num": 0.024277369181315105,
      "eval_custom_ui_val_loss_xval": 1.1207139756944444,
      "eval_custom_ui_val_runtime": 62.376,
      "eval_custom_ui_val_samples_per_second": 4.248,
      "eval_custom_ui_val_steps_per_second": 0.144,
      "num_input_tokens_seen": 364192972,
      "step": 6500
    },
    {
      "epoch": 14.476614699331849,
      "loss": 0.8343714475631714,
      "loss_ce": 0.00014291857951320708,
      "loss_iou": 0.375,
      "loss_num": 0.0164794921875,
      "loss_xval": 0.8359375,
      "num_input_tokens_seen": 364192972,
      "step": 6500
    },
    {
      "epoch": 14.478841870824054,
      "grad_norm": 12.816341400146484,
      "learning_rate": 1e-06,
      "loss": 0.4073,
      "num_input_tokens_seen": 364250380,
      "step": 6501
    },
    {
      "epoch": 14.478841870824054,
      "loss": 0.26279568672180176,
      "loss_ce": 0.00010036412277258933,
      "loss_iou": 0.109375,
      "loss_num": 0.00872802734375,
      "loss_xval": 0.26171875,
      "num_input_tokens_seen": 364250380,
      "step": 6501
    },
    {
      "epoch": 14.481069042316259,
      "grad_norm": 12.928550720214844,
      "learning_rate": 1e-06,
      "loss": 0.4519,
      "num_input_tokens_seen": 364306508,
      "step": 6502
    },
    {
      "epoch": 14.481069042316259,
      "loss": 0.47788378596305847,
      "loss_ce": 0.00010056734754471108,
      "loss_iou": 0.2080078125,
      "loss_num": 0.01220703125,
      "loss_xval": 0.478515625,
      "num_input_tokens_seen": 364306508,
      "step": 6502
    },
    {
      "epoch": 14.483296213808464,
      "grad_norm": 17.1734619140625,
      "learning_rate": 1e-06,
      "loss": 0.4719,
      "num_input_tokens_seen": 364362888,
      "step": 6503
    },
    {
      "epoch": 14.483296213808464,
      "loss": 0.48692283034324646,
      "loss_ce": 0.00010643218411132693,
      "loss_iou": 0.20703125,
      "loss_num": 0.014404296875,
      "loss_xval": 0.486328125,
      "num_input_tokens_seen": 364362888,
      "step": 6503
    },
    {
      "epoch": 14.485523385300668,
      "grad_norm": 24.356948852539062,
      "learning_rate": 1e-06,
      "loss": 0.5933,
      "num_input_tokens_seen": 364418476,
      "step": 6504
    },
    {
      "epoch": 14.485523385300668,
      "loss": 0.5144957900047302,
      "loss_ce": 9.150611003860831e-05,
      "loss_iou": 0.2275390625,
      "loss_num": 0.01202392578125,
      "loss_xval": 0.515625,
      "num_input_tokens_seen": 364418476,
      "step": 6504
    },
    {
      "epoch": 14.487750556792873,
      "grad_norm": 21.26868438720703,
      "learning_rate": 1e-06,
      "loss": 0.4377,
      "num_input_tokens_seen": 364474968,
      "step": 6505
    },
    {
      "epoch": 14.487750556792873,
      "loss": 0.3293471932411194,
      "loss_ce": 0.00012356144725345075,
      "loss_iou": 0.1376953125,
      "loss_num": 0.0106201171875,
      "loss_xval": 0.330078125,
      "num_input_tokens_seen": 364474968,
      "step": 6505
    },
    {
      "epoch": 14.489977728285078,
      "grad_norm": 12.99026107788086,
      "learning_rate": 1e-06,
      "loss": 0.4491,
      "num_input_tokens_seen": 364532716,
      "step": 6506
    },
    {
      "epoch": 14.489977728285078,
      "loss": 0.25777584314346313,
      "loss_ce": 8.539756527170539e-05,
      "loss_iou": 0.10595703125,
      "loss_num": 0.0091552734375,
      "loss_xval": 0.2578125,
      "num_input_tokens_seen": 364532716,
      "step": 6506
    },
    {
      "epoch": 14.492204899777283,
      "grad_norm": 15.61511516571045,
      "learning_rate": 1e-06,
      "loss": 0.4898,
      "num_input_tokens_seen": 364589292,
      "step": 6507
    },
    {
      "epoch": 14.492204899777283,
      "loss": 0.5754222869873047,
      "loss_ce": 0.00010490816930541769,
      "loss_iou": 0.2451171875,
      "loss_num": 0.0172119140625,
      "loss_xval": 0.57421875,
      "num_input_tokens_seen": 364589292,
      "step": 6507
    },
    {
      "epoch": 14.494432071269488,
      "grad_norm": 17.655942916870117,
      "learning_rate": 1e-06,
      "loss": 0.4318,
      "num_input_tokens_seen": 364644880,
      "step": 6508
    },
    {
      "epoch": 14.494432071269488,
      "loss": 0.3868103623390198,
      "loss_ce": 9.161405614577234e-05,
      "loss_iou": 0.1689453125,
      "loss_num": 0.00994873046875,
      "loss_xval": 0.38671875,
      "num_input_tokens_seen": 364644880,
      "step": 6508
    },
    {
      "epoch": 14.496659242761693,
      "grad_norm": 20.9801082611084,
      "learning_rate": 1e-06,
      "loss": 0.4926,
      "num_input_tokens_seen": 364702920,
      "step": 6509
    },
    {
      "epoch": 14.496659242761693,
      "loss": 0.48010796308517456,
      "loss_ce": 0.00012747629079967737,
      "loss_iou": 0.2255859375,
      "loss_num": 0.005859375,
      "loss_xval": 0.48046875,
      "num_input_tokens_seen": 364702920,
      "step": 6509
    },
    {
      "epoch": 14.498886414253898,
      "grad_norm": 20.405927658081055,
      "learning_rate": 1e-06,
      "loss": 0.439,
      "num_input_tokens_seen": 364757992,
      "step": 6510
    },
    {
      "epoch": 14.498886414253898,
      "loss": 0.642722487449646,
      "loss_ce": 0.00014438082871492952,
      "loss_iou": 0.255859375,
      "loss_num": 0.0257568359375,
      "loss_xval": 0.640625,
      "num_input_tokens_seen": 364757992,
      "step": 6510
    },
    {
      "epoch": 14.501113585746102,
      "grad_norm": 22.799161911010742,
      "learning_rate": 1e-06,
      "loss": 0.3945,
      "num_input_tokens_seen": 364815876,
      "step": 6511
    },
    {
      "epoch": 14.501113585746102,
      "loss": 0.5448106527328491,
      "loss_ce": 0.00013291001960169524,
      "loss_iou": 0.244140625,
      "loss_num": 0.01123046875,
      "loss_xval": 0.54296875,
      "num_input_tokens_seen": 364815876,
      "step": 6511
    },
    {
      "epoch": 14.503340757238307,
      "grad_norm": 18.602733612060547,
      "learning_rate": 1e-06,
      "loss": 0.533,
      "num_input_tokens_seen": 364871856,
      "step": 6512
    },
    {
      "epoch": 14.503340757238307,
      "loss": 0.3779085874557495,
      "loss_ce": 0.00010096091136801988,
      "loss_iou": 0.142578125,
      "loss_num": 0.018310546875,
      "loss_xval": 0.376953125,
      "num_input_tokens_seen": 364871856,
      "step": 6512
    },
    {
      "epoch": 14.505567928730512,
      "grad_norm": 35.201637268066406,
      "learning_rate": 1e-06,
      "loss": 0.6026,
      "num_input_tokens_seen": 364927504,
      "step": 6513
    },
    {
      "epoch": 14.505567928730512,
      "loss": 0.4659165143966675,
      "loss_ce": 9.620728815207258e-05,
      "loss_iou": 0.2109375,
      "loss_num": 0.00872802734375,
      "loss_xval": 0.46484375,
      "num_input_tokens_seen": 364927504,
      "step": 6513
    },
    {
      "epoch": 14.507795100222717,
      "grad_norm": 19.808265686035156,
      "learning_rate": 1e-06,
      "loss": 0.5087,
      "num_input_tokens_seen": 364985060,
      "step": 6514
    },
    {
      "epoch": 14.507795100222717,
      "loss": 0.5555931329727173,
      "loss_ce": 0.00011220310989301652,
      "loss_iou": 0.2314453125,
      "loss_num": 0.0184326171875,
      "loss_xval": 0.5546875,
      "num_input_tokens_seen": 364985060,
      "step": 6514
    },
    {
      "epoch": 14.510022271714922,
      "grad_norm": 14.76258373260498,
      "learning_rate": 1e-06,
      "loss": 0.3267,
      "num_input_tokens_seen": 365042232,
      "step": 6515
    },
    {
      "epoch": 14.510022271714922,
      "loss": 0.3308001160621643,
      "loss_ce": 0.00011161710426677018,
      "loss_iou": 0.154296875,
      "loss_num": 0.00445556640625,
      "loss_xval": 0.330078125,
      "num_input_tokens_seen": 365042232,
      "step": 6515
    },
    {
      "epoch": 14.512249443207127,
      "grad_norm": 15.700284004211426,
      "learning_rate": 1e-06,
      "loss": 0.5323,
      "num_input_tokens_seen": 365094928,
      "step": 6516
    },
    {
      "epoch": 14.512249443207127,
      "loss": 0.48436519503593445,
      "loss_ce": 0.00011227864888496697,
      "loss_iou": 0.1943359375,
      "loss_num": 0.019287109375,
      "loss_xval": 0.484375,
      "num_input_tokens_seen": 365094928,
      "step": 6516
    },
    {
      "epoch": 14.514476614699332,
      "grad_norm": 12.311739921569824,
      "learning_rate": 1e-06,
      "loss": 0.3459,
      "num_input_tokens_seen": 365152328,
      "step": 6517
    },
    {
      "epoch": 14.514476614699332,
      "loss": 0.41318339109420776,
      "loss_ce": 9.744486305862665e-05,
      "loss_iou": 0.177734375,
      "loss_num": 0.01177978515625,
      "loss_xval": 0.4140625,
      "num_input_tokens_seen": 365152328,
      "step": 6517
    },
    {
      "epoch": 14.516703786191536,
      "grad_norm": 16.08889389038086,
      "learning_rate": 1e-06,
      "loss": 0.4272,
      "num_input_tokens_seen": 365211004,
      "step": 6518
    },
    {
      "epoch": 14.516703786191536,
      "loss": 0.40454918146133423,
      "loss_ce": 0.0002522985450923443,
      "loss_iou": 0.1669921875,
      "loss_num": 0.01416015625,
      "loss_xval": 0.404296875,
      "num_input_tokens_seen": 365211004,
      "step": 6518
    },
    {
      "epoch": 14.518930957683741,
      "grad_norm": 15.88845157623291,
      "learning_rate": 1e-06,
      "loss": 0.474,
      "num_input_tokens_seen": 365267152,
      "step": 6519
    },
    {
      "epoch": 14.518930957683741,
      "loss": 0.4582583010196686,
      "loss_ce": 0.0001284035388380289,
      "loss_iou": 0.203125,
      "loss_num": 0.01043701171875,
      "loss_xval": 0.458984375,
      "num_input_tokens_seen": 365267152,
      "step": 6519
    },
    {
      "epoch": 14.521158129175946,
      "grad_norm": 22.944534301757812,
      "learning_rate": 1e-06,
      "loss": 0.3661,
      "num_input_tokens_seen": 365322816,
      "step": 6520
    },
    {
      "epoch": 14.521158129175946,
      "loss": 0.3829045295715332,
      "loss_ce": 0.00021408403699751943,
      "loss_iou": 0.166015625,
      "loss_num": 0.0101318359375,
      "loss_xval": 0.3828125,
      "num_input_tokens_seen": 365322816,
      "step": 6520
    },
    {
      "epoch": 14.523385300668151,
      "grad_norm": 15.910293579101562,
      "learning_rate": 1e-06,
      "loss": 0.5393,
      "num_input_tokens_seen": 365377560,
      "step": 6521
    },
    {
      "epoch": 14.523385300668151,
      "loss": 0.6288875937461853,
      "loss_ce": 0.0001033980370266363,
      "loss_iou": 0.27734375,
      "loss_num": 0.014404296875,
      "loss_xval": 0.62890625,
      "num_input_tokens_seen": 365377560,
      "step": 6521
    },
    {
      "epoch": 14.525612472160356,
      "grad_norm": 31.319820404052734,
      "learning_rate": 1e-06,
      "loss": 0.5413,
      "num_input_tokens_seen": 365431876,
      "step": 6522
    },
    {
      "epoch": 14.525612472160356,
      "loss": 0.45200419425964355,
      "loss_ce": 9.987234807340428e-05,
      "loss_iou": 0.201171875,
      "loss_num": 0.010009765625,
      "loss_xval": 0.451171875,
      "num_input_tokens_seen": 365431876,
      "step": 6522
    },
    {
      "epoch": 14.52783964365256,
      "grad_norm": 33.108680725097656,
      "learning_rate": 1e-06,
      "loss": 0.5389,
      "num_input_tokens_seen": 365487292,
      "step": 6523
    },
    {
      "epoch": 14.52783964365256,
      "loss": 0.4871612787246704,
      "loss_ce": 0.0001007293612929061,
      "loss_iou": 0.2138671875,
      "loss_num": 0.01202392578125,
      "loss_xval": 0.486328125,
      "num_input_tokens_seen": 365487292,
      "step": 6523
    },
    {
      "epoch": 14.530066815144766,
      "grad_norm": 26.078786849975586,
      "learning_rate": 1e-06,
      "loss": 0.3983,
      "num_input_tokens_seen": 365545116,
      "step": 6524
    },
    {
      "epoch": 14.530066815144766,
      "loss": 0.5119673013687134,
      "loss_ce": 0.00012648927804548293,
      "loss_iou": 0.21875,
      "loss_num": 0.01495361328125,
      "loss_xval": 0.51171875,
      "num_input_tokens_seen": 365545116,
      "step": 6524
    },
    {
      "epoch": 14.53229398663697,
      "grad_norm": 21.211854934692383,
      "learning_rate": 1e-06,
      "loss": 0.4073,
      "num_input_tokens_seen": 365603000,
      "step": 6525
    },
    {
      "epoch": 14.53229398663697,
      "loss": 0.5028167963027954,
      "loss_ce": 0.00013121790834702551,
      "loss_iou": 0.21484375,
      "loss_num": 0.014404296875,
      "loss_xval": 0.50390625,
      "num_input_tokens_seen": 365603000,
      "step": 6525
    },
    {
      "epoch": 14.534521158129175,
      "grad_norm": 13.118301391601562,
      "learning_rate": 1e-06,
      "loss": 0.3706,
      "num_input_tokens_seen": 365658648,
      "step": 6526
    },
    {
      "epoch": 14.534521158129175,
      "loss": 0.37536337971687317,
      "loss_ce": 0.00011921751865884289,
      "loss_iou": 0.16796875,
      "loss_num": 0.0079345703125,
      "loss_xval": 0.375,
      "num_input_tokens_seen": 365658648,
      "step": 6526
    },
    {
      "epoch": 14.53674832962138,
      "grad_norm": 12.883088111877441,
      "learning_rate": 1e-06,
      "loss": 0.4133,
      "num_input_tokens_seen": 365713996,
      "step": 6527
    },
    {
      "epoch": 14.53674832962138,
      "loss": 0.4605136811733246,
      "loss_ce": 0.0004306669579818845,
      "loss_iou": 0.2021484375,
      "loss_num": 0.01104736328125,
      "loss_xval": 0.4609375,
      "num_input_tokens_seen": 365713996,
      "step": 6527
    },
    {
      "epoch": 14.538975501113585,
      "grad_norm": 21.47534942626953,
      "learning_rate": 1e-06,
      "loss": 0.3516,
      "num_input_tokens_seen": 365769532,
      "step": 6528
    },
    {
      "epoch": 14.538975501113585,
      "loss": 0.3355334401130676,
      "loss_ce": 8.420874655712396e-05,
      "loss_iou": 0.150390625,
      "loss_num": 0.007110595703125,
      "loss_xval": 0.3359375,
      "num_input_tokens_seen": 365769532,
      "step": 6528
    },
    {
      "epoch": 14.54120267260579,
      "grad_norm": 17.586118698120117,
      "learning_rate": 1e-06,
      "loss": 0.549,
      "num_input_tokens_seen": 365826440,
      "step": 6529
    },
    {
      "epoch": 14.54120267260579,
      "loss": 0.5860310792922974,
      "loss_ce": 9.35560601647012e-05,
      "loss_iou": 0.2353515625,
      "loss_num": 0.02294921875,
      "loss_xval": 0.5859375,
      "num_input_tokens_seen": 365826440,
      "step": 6529
    },
    {
      "epoch": 14.543429844097995,
      "grad_norm": 19.080286026000977,
      "learning_rate": 1e-06,
      "loss": 0.4543,
      "num_input_tokens_seen": 365882108,
      "step": 6530
    },
    {
      "epoch": 14.543429844097995,
      "loss": 0.4774046540260315,
      "loss_ce": 0.00010971432493533939,
      "loss_iou": 0.21484375,
      "loss_num": 0.00958251953125,
      "loss_xval": 0.4765625,
      "num_input_tokens_seen": 365882108,
      "step": 6530
    },
    {
      "epoch": 14.5456570155902,
      "grad_norm": 18.724021911621094,
      "learning_rate": 1e-06,
      "loss": 0.5758,
      "num_input_tokens_seen": 365940436,
      "step": 6531
    },
    {
      "epoch": 14.5456570155902,
      "loss": 0.7265654802322388,
      "loss_ce": 0.00012504393816925585,
      "loss_iou": 0.30078125,
      "loss_num": 0.0247802734375,
      "loss_xval": 0.7265625,
      "num_input_tokens_seen": 365940436,
      "step": 6531
    },
    {
      "epoch": 14.547884187082406,
      "grad_norm": 16.194499969482422,
      "learning_rate": 1e-06,
      "loss": 0.4139,
      "num_input_tokens_seen": 365996348,
      "step": 6532
    },
    {
      "epoch": 14.547884187082406,
      "loss": 0.36508145928382874,
      "loss_ce": 9.123167546931654e-05,
      "loss_iou": 0.169921875,
      "loss_num": 0.004791259765625,
      "loss_xval": 0.365234375,
      "num_input_tokens_seen": 365996348,
      "step": 6532
    },
    {
      "epoch": 14.550111358574611,
      "grad_norm": 14.120255470275879,
      "learning_rate": 1e-06,
      "loss": 0.4066,
      "num_input_tokens_seen": 366050788,
      "step": 6533
    },
    {
      "epoch": 14.550111358574611,
      "loss": 0.380734384059906,
      "loss_ce": 0.0003022679884452373,
      "loss_iou": 0.1669921875,
      "loss_num": 0.00921630859375,
      "loss_xval": 0.380859375,
      "num_input_tokens_seen": 366050788,
      "step": 6533
    },
    {
      "epoch": 14.552338530066816,
      "grad_norm": 16.832948684692383,
      "learning_rate": 1e-06,
      "loss": 0.5344,
      "num_input_tokens_seen": 366103204,
      "step": 6534
    },
    {
      "epoch": 14.552338530066816,
      "loss": 0.49608922004699707,
      "loss_ce": 0.00011752717546187341,
      "loss_iou": 0.2138671875,
      "loss_num": 0.0135498046875,
      "loss_xval": 0.49609375,
      "num_input_tokens_seen": 366103204,
      "step": 6534
    },
    {
      "epoch": 14.55456570155902,
      "grad_norm": 20.420133590698242,
      "learning_rate": 1e-06,
      "loss": 0.4178,
      "num_input_tokens_seen": 366161640,
      "step": 6535
    },
    {
      "epoch": 14.55456570155902,
      "loss": 0.4828900098800659,
      "loss_ce": 0.00010191020555794239,
      "loss_iou": 0.2041015625,
      "loss_num": 0.01470947265625,
      "loss_xval": 0.482421875,
      "num_input_tokens_seen": 366161640,
      "step": 6535
    },
    {
      "epoch": 14.556792873051226,
      "grad_norm": 19.563575744628906,
      "learning_rate": 1e-06,
      "loss": 0.3978,
      "num_input_tokens_seen": 366216184,
      "step": 6536
    },
    {
      "epoch": 14.556792873051226,
      "loss": 0.5175689458847046,
      "loss_ce": 0.0001129416050389409,
      "loss_iou": 0.23046875,
      "loss_num": 0.01141357421875,
      "loss_xval": 0.515625,
      "num_input_tokens_seen": 366216184,
      "step": 6536
    },
    {
      "epoch": 14.55902004454343,
      "grad_norm": 14.907690048217773,
      "learning_rate": 1e-06,
      "loss": 0.5119,
      "num_input_tokens_seen": 366273176,
      "step": 6537
    },
    {
      "epoch": 14.55902004454343,
      "loss": 0.4903450906276703,
      "loss_ce": 0.00011072470806539059,
      "loss_iou": 0.212890625,
      "loss_num": 0.01318359375,
      "loss_xval": 0.490234375,
      "num_input_tokens_seen": 366273176,
      "step": 6537
    },
    {
      "epoch": 14.561247216035635,
      "grad_norm": 21.058177947998047,
      "learning_rate": 1e-06,
      "loss": 0.4864,
      "num_input_tokens_seen": 366329652,
      "step": 6538
    },
    {
      "epoch": 14.561247216035635,
      "loss": 0.38425058126449585,
      "loss_ce": 9.530353418085724e-05,
      "loss_iou": 0.1591796875,
      "loss_num": 0.01312255859375,
      "loss_xval": 0.384765625,
      "num_input_tokens_seen": 366329652,
      "step": 6538
    },
    {
      "epoch": 14.56347438752784,
      "grad_norm": 76.75809478759766,
      "learning_rate": 1e-06,
      "loss": 0.6636,
      "num_input_tokens_seen": 366388276,
      "step": 6539
    },
    {
      "epoch": 14.56347438752784,
      "loss": 0.8201836347579956,
      "loss_ce": 0.00011528450704645365,
      "loss_iou": 0.330078125,
      "loss_num": 0.03173828125,
      "loss_xval": 0.8203125,
      "num_input_tokens_seen": 366388276,
      "step": 6539
    },
    {
      "epoch": 14.565701559020045,
      "grad_norm": 18.305295944213867,
      "learning_rate": 1e-06,
      "loss": 0.5901,
      "num_input_tokens_seen": 366444384,
      "step": 6540
    },
    {
      "epoch": 14.565701559020045,
      "loss": 0.5785954594612122,
      "loss_ce": 0.00010423710045870394,
      "loss_iou": 0.25390625,
      "loss_num": 0.01385498046875,
      "loss_xval": 0.578125,
      "num_input_tokens_seen": 366444384,
      "step": 6540
    },
    {
      "epoch": 14.56792873051225,
      "grad_norm": 16.77396583557129,
      "learning_rate": 1e-06,
      "loss": 0.5024,
      "num_input_tokens_seen": 366500264,
      "step": 6541
    },
    {
      "epoch": 14.56792873051225,
      "loss": 0.47824224829673767,
      "loss_ce": 9.282668179366738e-05,
      "loss_iou": 0.1748046875,
      "loss_num": 0.025634765625,
      "loss_xval": 0.478515625,
      "num_input_tokens_seen": 366500264,
      "step": 6541
    },
    {
      "epoch": 14.570155902004455,
      "grad_norm": 22.724868774414062,
      "learning_rate": 1e-06,
      "loss": 0.4041,
      "num_input_tokens_seen": 366556940,
      "step": 6542
    },
    {
      "epoch": 14.570155902004455,
      "loss": 0.45686769485473633,
      "loss_ce": 8.05618183221668e-05,
      "loss_iou": 0.193359375,
      "loss_num": 0.0140380859375,
      "loss_xval": 0.45703125,
      "num_input_tokens_seen": 366556940,
      "step": 6542
    },
    {
      "epoch": 14.57238307349666,
      "grad_norm": 20.71229362487793,
      "learning_rate": 1e-06,
      "loss": 0.5195,
      "num_input_tokens_seen": 366610572,
      "step": 6543
    },
    {
      "epoch": 14.57238307349666,
      "loss": 0.5313507318496704,
      "loss_ce": 0.0001007560349535197,
      "loss_iou": 0.2119140625,
      "loss_num": 0.0213623046875,
      "loss_xval": 0.53125,
      "num_input_tokens_seen": 366610572,
      "step": 6543
    },
    {
      "epoch": 14.574610244988865,
      "grad_norm": 20.400714874267578,
      "learning_rate": 1e-06,
      "loss": 0.5844,
      "num_input_tokens_seen": 366665496,
      "step": 6544
    },
    {
      "epoch": 14.574610244988865,
      "loss": 0.46006056666374207,
      "loss_ce": 9.960292663890868e-05,
      "loss_iou": 0.2216796875,
      "loss_num": 0.0036163330078125,
      "loss_xval": 0.4609375,
      "num_input_tokens_seen": 366665496,
      "step": 6544
    },
    {
      "epoch": 14.57683741648107,
      "grad_norm": 16.26686668395996,
      "learning_rate": 1e-06,
      "loss": 0.6016,
      "num_input_tokens_seen": 366720024,
      "step": 6545
    },
    {
      "epoch": 14.57683741648107,
      "loss": 0.622795820236206,
      "loss_ce": 0.00011514931975398213,
      "loss_iou": 0.2578125,
      "loss_num": 0.021240234375,
      "loss_xval": 0.62109375,
      "num_input_tokens_seen": 366720024,
      "step": 6545
    },
    {
      "epoch": 14.579064587973274,
      "grad_norm": 14.509425163269043,
      "learning_rate": 1e-06,
      "loss": 0.4275,
      "num_input_tokens_seen": 366778568,
      "step": 6546
    },
    {
      "epoch": 14.579064587973274,
      "loss": 0.36485838890075684,
      "loss_ce": 0.00011229590745642781,
      "loss_iou": 0.169921875,
      "loss_num": 0.0050048828125,
      "loss_xval": 0.365234375,
      "num_input_tokens_seen": 366778568,
      "step": 6546
    },
    {
      "epoch": 14.58129175946548,
      "grad_norm": 26.61733627319336,
      "learning_rate": 1e-06,
      "loss": 0.5409,
      "num_input_tokens_seen": 366834208,
      "step": 6547
    },
    {
      "epoch": 14.58129175946548,
      "loss": 0.6287715435028076,
      "loss_ce": 0.0001094555773306638,
      "loss_iou": 0.28125,
      "loss_num": 0.01336669921875,
      "loss_xval": 0.62890625,
      "num_input_tokens_seen": 366834208,
      "step": 6547
    },
    {
      "epoch": 14.583518930957684,
      "grad_norm": 21.295969009399414,
      "learning_rate": 1e-06,
      "loss": 0.4005,
      "num_input_tokens_seen": 366889968,
      "step": 6548
    },
    {
      "epoch": 14.583518930957684,
      "loss": 0.4716094136238098,
      "loss_ce": 0.00011283693311270326,
      "loss_iou": 0.21484375,
      "loss_num": 0.00836181640625,
      "loss_xval": 0.470703125,
      "num_input_tokens_seen": 366889968,
      "step": 6548
    },
    {
      "epoch": 14.585746102449889,
      "grad_norm": 30.666751861572266,
      "learning_rate": 1e-06,
      "loss": 0.4064,
      "num_input_tokens_seen": 366945348,
      "step": 6549
    },
    {
      "epoch": 14.585746102449889,
      "loss": 0.4583427906036377,
      "loss_ce": 9.083442273549736e-05,
      "loss_iou": 0.1875,
      "loss_num": 0.0166015625,
      "loss_xval": 0.458984375,
      "num_input_tokens_seen": 366945348,
      "step": 6549
    },
    {
      "epoch": 14.587973273942094,
      "grad_norm": 36.7827262878418,
      "learning_rate": 1e-06,
      "loss": 0.5327,
      "num_input_tokens_seen": 367000992,
      "step": 6550
    },
    {
      "epoch": 14.587973273942094,
      "loss": 0.6207618713378906,
      "loss_ce": 0.0001563684199936688,
      "loss_iou": 0.28125,
      "loss_num": 0.01202392578125,
      "loss_xval": 0.62109375,
      "num_input_tokens_seen": 367000992,
      "step": 6550
    },
    {
      "epoch": 14.590200445434299,
      "grad_norm": 13.286869049072266,
      "learning_rate": 1e-06,
      "loss": 0.4203,
      "num_input_tokens_seen": 367057428,
      "step": 6551
    },
    {
      "epoch": 14.590200445434299,
      "loss": 0.3902452886104584,
      "loss_ce": 0.00010857303277589381,
      "loss_iou": 0.171875,
      "loss_num": 0.00933837890625,
      "loss_xval": 0.390625,
      "num_input_tokens_seen": 367057428,
      "step": 6551
    },
    {
      "epoch": 14.592427616926503,
      "grad_norm": 25.443744659423828,
      "learning_rate": 1e-06,
      "loss": 0.486,
      "num_input_tokens_seen": 367110136,
      "step": 6552
    },
    {
      "epoch": 14.592427616926503,
      "loss": 0.612711489200592,
      "loss_ce": 0.0016885376535356045,
      "loss_iou": 0.267578125,
      "loss_num": 0.015625,
      "loss_xval": 0.609375,
      "num_input_tokens_seen": 367110136,
      "step": 6552
    },
    {
      "epoch": 14.594654788418708,
      "grad_norm": 25.125425338745117,
      "learning_rate": 1e-06,
      "loss": 0.4382,
      "num_input_tokens_seen": 367167548,
      "step": 6553
    },
    {
      "epoch": 14.594654788418708,
      "loss": 0.3994516134262085,
      "loss_ce": 9.859049168881029e-05,
      "loss_iou": 0.1767578125,
      "loss_num": 0.009033203125,
      "loss_xval": 0.3984375,
      "num_input_tokens_seen": 367167548,
      "step": 6553
    },
    {
      "epoch": 14.596881959910913,
      "grad_norm": 16.59556770324707,
      "learning_rate": 1e-06,
      "loss": 0.4072,
      "num_input_tokens_seen": 367223212,
      "step": 6554
    },
    {
      "epoch": 14.596881959910913,
      "loss": 0.3758706748485565,
      "loss_ce": 0.00010771742381621152,
      "loss_iou": 0.16796875,
      "loss_num": 0.008056640625,
      "loss_xval": 0.375,
      "num_input_tokens_seen": 367223212,
      "step": 6554
    },
    {
      "epoch": 14.599109131403118,
      "grad_norm": 12.429368019104004,
      "learning_rate": 1e-06,
      "loss": 0.4022,
      "num_input_tokens_seen": 367279056,
      "step": 6555
    },
    {
      "epoch": 14.599109131403118,
      "loss": 0.5174868702888489,
      "loss_ce": 9.187131217913702e-05,
      "loss_iou": 0.236328125,
      "loss_num": 0.00909423828125,
      "loss_xval": 0.515625,
      "num_input_tokens_seen": 367279056,
      "step": 6555
    },
    {
      "epoch": 14.601336302895323,
      "grad_norm": 18.785852432250977,
      "learning_rate": 1e-06,
      "loss": 0.4512,
      "num_input_tokens_seen": 367336108,
      "step": 6556
    },
    {
      "epoch": 14.601336302895323,
      "loss": 0.5070604085922241,
      "loss_ce": 0.00010236204252578318,
      "loss_iou": 0.21484375,
      "loss_num": 0.015380859375,
      "loss_xval": 0.5078125,
      "num_input_tokens_seen": 367336108,
      "step": 6556
    },
    {
      "epoch": 14.603563474387528,
      "grad_norm": 15.28122329711914,
      "learning_rate": 1e-06,
      "loss": 0.3388,
      "num_input_tokens_seen": 367393664,
      "step": 6557
    },
    {
      "epoch": 14.603563474387528,
      "loss": 0.4066086709499359,
      "loss_ce": 0.00011453506886027753,
      "loss_iou": 0.17578125,
      "loss_num": 0.01080322265625,
      "loss_xval": 0.40625,
      "num_input_tokens_seen": 367393664,
      "step": 6557
    },
    {
      "epoch": 14.605790645879733,
      "grad_norm": 12.419171333312988,
      "learning_rate": 1e-06,
      "loss": 0.4405,
      "num_input_tokens_seen": 367451044,
      "step": 6558
    },
    {
      "epoch": 14.605790645879733,
      "loss": 0.4510771334171295,
      "loss_ce": 0.00014940105029381812,
      "loss_iou": 0.1923828125,
      "loss_num": 0.0133056640625,
      "loss_xval": 0.451171875,
      "num_input_tokens_seen": 367451044,
      "step": 6558
    },
    {
      "epoch": 14.608017817371937,
      "grad_norm": 14.014966011047363,
      "learning_rate": 1e-06,
      "loss": 0.5277,
      "num_input_tokens_seen": 367508844,
      "step": 6559
    },
    {
      "epoch": 14.608017817371937,
      "loss": 0.5670410990715027,
      "loss_ce": 0.00014657452993560582,
      "loss_iou": 0.21875,
      "loss_num": 0.025634765625,
      "loss_xval": 0.56640625,
      "num_input_tokens_seen": 367508844,
      "step": 6559
    },
    {
      "epoch": 14.610244988864142,
      "grad_norm": 15.982457160949707,
      "learning_rate": 1e-06,
      "loss": 0.3984,
      "num_input_tokens_seen": 367566664,
      "step": 6560
    },
    {
      "epoch": 14.610244988864142,
      "loss": 0.400331974029541,
      "loss_ce": 0.0001854914880823344,
      "loss_iou": 0.1728515625,
      "loss_num": 0.0108642578125,
      "loss_xval": 0.400390625,
      "num_input_tokens_seen": 367566664,
      "step": 6560
    },
    {
      "epoch": 14.612472160356347,
      "grad_norm": 20.686161041259766,
      "learning_rate": 1e-06,
      "loss": 0.4902,
      "num_input_tokens_seen": 367619140,
      "step": 6561
    },
    {
      "epoch": 14.612472160356347,
      "loss": 0.6208482384681702,
      "loss_ce": 0.00012072353274561465,
      "loss_iou": 0.259765625,
      "loss_num": 0.02001953125,
      "loss_xval": 0.62109375,
      "num_input_tokens_seen": 367619140,
      "step": 6561
    },
    {
      "epoch": 14.614699331848552,
      "grad_norm": 22.73474884033203,
      "learning_rate": 1e-06,
      "loss": 0.4558,
      "num_input_tokens_seen": 367675996,
      "step": 6562
    },
    {
      "epoch": 14.614699331848552,
      "loss": 0.4799794554710388,
      "loss_ce": 0.00012105887435609475,
      "loss_iou": 0.1845703125,
      "loss_num": 0.0220947265625,
      "loss_xval": 0.48046875,
      "num_input_tokens_seen": 367675996,
      "step": 6562
    },
    {
      "epoch": 14.616926503340757,
      "grad_norm": 19.15255355834961,
      "learning_rate": 1e-06,
      "loss": 0.4012,
      "num_input_tokens_seen": 367733188,
      "step": 6563
    },
    {
      "epoch": 14.616926503340757,
      "loss": 0.38610512018203735,
      "loss_ce": 0.00011879668454639614,
      "loss_iou": 0.1689453125,
      "loss_num": 0.009521484375,
      "loss_xval": 0.38671875,
      "num_input_tokens_seen": 367733188,
      "step": 6563
    },
    {
      "epoch": 14.619153674832962,
      "grad_norm": 19.479084014892578,
      "learning_rate": 1e-06,
      "loss": 0.5865,
      "num_input_tokens_seen": 367791252,
      "step": 6564
    },
    {
      "epoch": 14.619153674832962,
      "loss": 0.6352872848510742,
      "loss_ce": 0.0001554101036163047,
      "loss_iou": 0.28125,
      "loss_num": 0.0147705078125,
      "loss_xval": 0.63671875,
      "num_input_tokens_seen": 367791252,
      "step": 6564
    },
    {
      "epoch": 14.621380846325167,
      "grad_norm": 33.37403869628906,
      "learning_rate": 1e-06,
      "loss": 0.494,
      "num_input_tokens_seen": 367845888,
      "step": 6565
    },
    {
      "epoch": 14.621380846325167,
      "loss": 0.3272702097892761,
      "loss_ce": 0.00012174884614069015,
      "loss_iou": 0.1416015625,
      "loss_num": 0.00885009765625,
      "loss_xval": 0.328125,
      "num_input_tokens_seen": 367845888,
      "step": 6565
    },
    {
      "epoch": 14.623608017817372,
      "grad_norm": 14.532173156738281,
      "learning_rate": 1e-06,
      "loss": 0.5322,
      "num_input_tokens_seen": 367902024,
      "step": 6566
    },
    {
      "epoch": 14.623608017817372,
      "loss": 0.6220463514328003,
      "loss_ce": 9.808540198719129e-05,
      "loss_iou": 0.259765625,
      "loss_num": 0.0208740234375,
      "loss_xval": 0.62109375,
      "num_input_tokens_seen": 367902024,
      "step": 6566
    },
    {
      "epoch": 14.625835189309576,
      "grad_norm": 25.17264747619629,
      "learning_rate": 1e-06,
      "loss": 0.4745,
      "num_input_tokens_seen": 367956996,
      "step": 6567
    },
    {
      "epoch": 14.625835189309576,
      "loss": 0.46531087160110474,
      "loss_ce": 0.00010092130105476826,
      "loss_iou": 0.193359375,
      "loss_num": 0.0157470703125,
      "loss_xval": 0.46484375,
      "num_input_tokens_seen": 367956996,
      "step": 6567
    },
    {
      "epoch": 14.628062360801781,
      "grad_norm": 21.86960792541504,
      "learning_rate": 1e-06,
      "loss": 0.3833,
      "num_input_tokens_seen": 368011628,
      "step": 6568
    },
    {
      "epoch": 14.628062360801781,
      "loss": 0.41635486483573914,
      "loss_ce": 9.509771916782483e-05,
      "loss_iou": 0.19140625,
      "loss_num": 0.006561279296875,
      "loss_xval": 0.416015625,
      "num_input_tokens_seen": 368011628,
      "step": 6568
    },
    {
      "epoch": 14.630289532293986,
      "grad_norm": 20.430923461914062,
      "learning_rate": 1e-06,
      "loss": 0.7204,
      "num_input_tokens_seen": 368065324,
      "step": 6569
    },
    {
      "epoch": 14.630289532293986,
      "loss": 0.7128270268440247,
      "loss_ce": 0.0001805600186344236,
      "loss_iou": 0.306640625,
      "loss_num": 0.0198974609375,
      "loss_xval": 0.7109375,
      "num_input_tokens_seen": 368065324,
      "step": 6569
    },
    {
      "epoch": 14.632516703786191,
      "grad_norm": 23.498016357421875,
      "learning_rate": 1e-06,
      "loss": 0.5559,
      "num_input_tokens_seen": 368120640,
      "step": 6570
    },
    {
      "epoch": 14.632516703786191,
      "loss": 0.6478054523468018,
      "loss_ce": 0.00010038249456556514,
      "loss_iou": 0.263671875,
      "loss_num": 0.0242919921875,
      "loss_xval": 0.6484375,
      "num_input_tokens_seen": 368120640,
      "step": 6570
    },
    {
      "epoch": 14.634743875278396,
      "grad_norm": 20.486783981323242,
      "learning_rate": 1e-06,
      "loss": 0.5641,
      "num_input_tokens_seen": 368176708,
      "step": 6571
    },
    {
      "epoch": 14.634743875278396,
      "loss": 0.49949222803115845,
      "loss_ce": 0.00010254726657876745,
      "loss_iou": 0.208984375,
      "loss_num": 0.0162353515625,
      "loss_xval": 0.5,
      "num_input_tokens_seen": 368176708,
      "step": 6571
    },
    {
      "epoch": 14.6369710467706,
      "grad_norm": 18.985647201538086,
      "learning_rate": 1e-06,
      "loss": 0.5617,
      "num_input_tokens_seen": 368231812,
      "step": 6572
    },
    {
      "epoch": 14.6369710467706,
      "loss": 0.6410832405090332,
      "loss_ce": 9.200733620673418e-05,
      "loss_iou": 0.298828125,
      "loss_num": 0.00897216796875,
      "loss_xval": 0.640625,
      "num_input_tokens_seen": 368231812,
      "step": 6572
    },
    {
      "epoch": 14.639198218262806,
      "grad_norm": 14.87316608428955,
      "learning_rate": 1e-06,
      "loss": 0.6231,
      "num_input_tokens_seen": 368285028,
      "step": 6573
    },
    {
      "epoch": 14.639198218262806,
      "loss": 0.7178068161010742,
      "loss_ce": 0.00015544812777079642,
      "loss_iou": 0.302734375,
      "loss_num": 0.0224609375,
      "loss_xval": 0.71875,
      "num_input_tokens_seen": 368285028,
      "step": 6573
    },
    {
      "epoch": 14.64142538975501,
      "grad_norm": 18.456830978393555,
      "learning_rate": 1e-06,
      "loss": 0.4465,
      "num_input_tokens_seen": 368341996,
      "step": 6574
    },
    {
      "epoch": 14.64142538975501,
      "loss": 0.5098215937614441,
      "loss_ce": 0.0015208279946818948,
      "loss_iou": 0.2265625,
      "loss_num": 0.01104736328125,
      "loss_xval": 0.5078125,
      "num_input_tokens_seen": 368341996,
      "step": 6574
    },
    {
      "epoch": 14.643652561247215,
      "grad_norm": 15.073392868041992,
      "learning_rate": 1e-06,
      "loss": 0.5345,
      "num_input_tokens_seen": 368397584,
      "step": 6575
    },
    {
      "epoch": 14.643652561247215,
      "loss": 0.49510324001312256,
      "loss_ce": 0.00010811796528287232,
      "loss_iou": 0.208984375,
      "loss_num": 0.01556396484375,
      "loss_xval": 0.494140625,
      "num_input_tokens_seen": 368397584,
      "step": 6575
    },
    {
      "epoch": 14.64587973273942,
      "grad_norm": 22.879812240600586,
      "learning_rate": 1e-06,
      "loss": 0.4901,
      "num_input_tokens_seen": 368452996,
      "step": 6576
    },
    {
      "epoch": 14.64587973273942,
      "loss": 0.6450592279434204,
      "loss_ce": 0.00016175321070477366,
      "loss_iou": 0.2578125,
      "loss_num": 0.0257568359375,
      "loss_xval": 0.64453125,
      "num_input_tokens_seen": 368452996,
      "step": 6576
    },
    {
      "epoch": 14.648106904231625,
      "grad_norm": 36.055259704589844,
      "learning_rate": 1e-06,
      "loss": 0.4346,
      "num_input_tokens_seen": 368511036,
      "step": 6577
    },
    {
      "epoch": 14.648106904231625,
      "loss": 0.457261860370636,
      "loss_ce": 0.00010854535503312945,
      "loss_iou": 0.1953125,
      "loss_num": 0.01318359375,
      "loss_xval": 0.45703125,
      "num_input_tokens_seen": 368511036,
      "step": 6577
    },
    {
      "epoch": 14.65033407572383,
      "grad_norm": 22.31118392944336,
      "learning_rate": 1e-06,
      "loss": 0.4259,
      "num_input_tokens_seen": 368564932,
      "step": 6578
    },
    {
      "epoch": 14.65033407572383,
      "loss": 0.39772558212280273,
      "loss_ce": 0.00014260000898502767,
      "loss_iou": 0.162109375,
      "loss_num": 0.0146484375,
      "loss_xval": 0.3984375,
      "num_input_tokens_seen": 368564932,
      "step": 6578
    },
    {
      "epoch": 14.652561247216035,
      "grad_norm": 17.83094596862793,
      "learning_rate": 1e-06,
      "loss": 0.5154,
      "num_input_tokens_seen": 368621580,
      "step": 6579
    },
    {
      "epoch": 14.652561247216035,
      "loss": 0.5228110551834106,
      "loss_ce": 0.0002280529006384313,
      "loss_iou": 0.21875,
      "loss_num": 0.0172119140625,
      "loss_xval": 0.5234375,
      "num_input_tokens_seen": 368621580,
      "step": 6579
    },
    {
      "epoch": 14.654788418708241,
      "grad_norm": 24.0405216217041,
      "learning_rate": 1e-06,
      "loss": 0.4541,
      "num_input_tokens_seen": 368674988,
      "step": 6580
    },
    {
      "epoch": 14.654788418708241,
      "loss": 0.4888764023780823,
      "loss_ce": 0.00010689307237043977,
      "loss_iou": 0.2158203125,
      "loss_num": 0.0113525390625,
      "loss_xval": 0.48828125,
      "num_input_tokens_seen": 368674988,
      "step": 6580
    },
    {
      "epoch": 14.657015590200446,
      "grad_norm": 17.233251571655273,
      "learning_rate": 1e-06,
      "loss": 0.3594,
      "num_input_tokens_seen": 368731232,
      "step": 6581
    },
    {
      "epoch": 14.657015590200446,
      "loss": 0.319793164730072,
      "loss_ce": 9.099852468352765e-05,
      "loss_iou": 0.138671875,
      "loss_num": 0.00848388671875,
      "loss_xval": 0.3203125,
      "num_input_tokens_seen": 368731232,
      "step": 6581
    },
    {
      "epoch": 14.659242761692651,
      "grad_norm": 31.06590461730957,
      "learning_rate": 1e-06,
      "loss": 0.6744,
      "num_input_tokens_seen": 368787912,
      "step": 6582
    },
    {
      "epoch": 14.659242761692651,
      "loss": 0.6104459762573242,
      "loss_ce": 9.442644659429789e-05,
      "loss_iou": 0.23828125,
      "loss_num": 0.0269775390625,
      "loss_xval": 0.609375,
      "num_input_tokens_seen": 368787912,
      "step": 6582
    },
    {
      "epoch": 14.661469933184856,
      "grad_norm": 16.3990535736084,
      "learning_rate": 1e-06,
      "loss": 0.3871,
      "num_input_tokens_seen": 368844564,
      "step": 6583
    },
    {
      "epoch": 14.661469933184856,
      "loss": 0.4364955723285675,
      "loss_ce": 9.420434071216732e-05,
      "loss_iou": 0.1953125,
      "loss_num": 0.00921630859375,
      "loss_xval": 0.435546875,
      "num_input_tokens_seen": 368844564,
      "step": 6583
    },
    {
      "epoch": 14.66369710467706,
      "grad_norm": 16.442039489746094,
      "learning_rate": 1e-06,
      "loss": 0.5207,
      "num_input_tokens_seen": 368903620,
      "step": 6584
    },
    {
      "epoch": 14.66369710467706,
      "loss": 0.8181473612785339,
      "loss_ce": 0.00015422608703374863,
      "loss_iou": 0.33203125,
      "loss_num": 0.0308837890625,
      "loss_xval": 0.81640625,
      "num_input_tokens_seen": 368903620,
      "step": 6584
    },
    {
      "epoch": 14.665924276169266,
      "grad_norm": 18.313331604003906,
      "learning_rate": 1e-06,
      "loss": 0.5139,
      "num_input_tokens_seen": 368961612,
      "step": 6585
    },
    {
      "epoch": 14.665924276169266,
      "loss": 0.6265621185302734,
      "loss_ce": 9.728968143463135e-05,
      "loss_iou": 0.26953125,
      "loss_num": 0.017578125,
      "loss_xval": 0.625,
      "num_input_tokens_seen": 368961612,
      "step": 6585
    },
    {
      "epoch": 14.66815144766147,
      "grad_norm": 27.344846725463867,
      "learning_rate": 1e-06,
      "loss": 0.4549,
      "num_input_tokens_seen": 369014408,
      "step": 6586
    },
    {
      "epoch": 14.66815144766147,
      "loss": 0.5841131806373596,
      "loss_ce": 0.00012881754082627594,
      "loss_iou": 0.234375,
      "loss_num": 0.023193359375,
      "loss_xval": 0.5859375,
      "num_input_tokens_seen": 369014408,
      "step": 6586
    },
    {
      "epoch": 14.670378619153675,
      "grad_norm": 19.41844367980957,
      "learning_rate": 1e-06,
      "loss": 0.4142,
      "num_input_tokens_seen": 369068716,
      "step": 6587
    },
    {
      "epoch": 14.670378619153675,
      "loss": 0.4551857113838196,
      "loss_ce": 0.00010757872951216996,
      "loss_iou": 0.2080078125,
      "loss_num": 0.0078125,
      "loss_xval": 0.455078125,
      "num_input_tokens_seen": 369068716,
      "step": 6587
    },
    {
      "epoch": 14.67260579064588,
      "grad_norm": 17.168302536010742,
      "learning_rate": 1e-06,
      "loss": 0.5873,
      "num_input_tokens_seen": 369125348,
      "step": 6588
    },
    {
      "epoch": 14.67260579064588,
      "loss": 0.46933096647262573,
      "loss_ce": 9.269404108636081e-05,
      "loss_iou": 0.20703125,
      "loss_num": 0.010986328125,
      "loss_xval": 0.46875,
      "num_input_tokens_seen": 369125348,
      "step": 6588
    },
    {
      "epoch": 14.674832962138085,
      "grad_norm": 24.008495330810547,
      "learning_rate": 1e-06,
      "loss": 0.5045,
      "num_input_tokens_seen": 369181304,
      "step": 6589
    },
    {
      "epoch": 14.674832962138085,
      "loss": 0.46079450845718384,
      "loss_ce": 0.00010113501048181206,
      "loss_iou": 0.216796875,
      "loss_num": 0.00567626953125,
      "loss_xval": 0.4609375,
      "num_input_tokens_seen": 369181304,
      "step": 6589
    },
    {
      "epoch": 14.67706013363029,
      "grad_norm": 15.958958625793457,
      "learning_rate": 1e-06,
      "loss": 0.365,
      "num_input_tokens_seen": 369237872,
      "step": 6590
    },
    {
      "epoch": 14.67706013363029,
      "loss": 0.4678671360015869,
      "loss_ce": 9.370467159897089e-05,
      "loss_iou": 0.2001953125,
      "loss_num": 0.01361083984375,
      "loss_xval": 0.46875,
      "num_input_tokens_seen": 369237872,
      "step": 6590
    },
    {
      "epoch": 14.679287305122495,
      "grad_norm": 14.867953300476074,
      "learning_rate": 1e-06,
      "loss": 0.364,
      "num_input_tokens_seen": 369294824,
      "step": 6591
    },
    {
      "epoch": 14.679287305122495,
      "loss": 0.4817339777946472,
      "loss_ce": 0.0001055731117958203,
      "loss_iou": 0.2197265625,
      "loss_num": 0.0084228515625,
      "loss_xval": 0.482421875,
      "num_input_tokens_seen": 369294824,
      "step": 6591
    },
    {
      "epoch": 14.6815144766147,
      "grad_norm": 30.02640724182129,
      "learning_rate": 1e-06,
      "loss": 0.4666,
      "num_input_tokens_seen": 369352304,
      "step": 6592
    },
    {
      "epoch": 14.6815144766147,
      "loss": 0.5130610466003418,
      "loss_ce": 0.00012160430196672678,
      "loss_iou": 0.21875,
      "loss_num": 0.01531982421875,
      "loss_xval": 0.51171875,
      "num_input_tokens_seen": 369352304,
      "step": 6592
    },
    {
      "epoch": 14.683741648106905,
      "grad_norm": 29.63843536376953,
      "learning_rate": 1e-06,
      "loss": 0.4834,
      "num_input_tokens_seen": 369407840,
      "step": 6593
    },
    {
      "epoch": 14.683741648106905,
      "loss": 0.5572299957275391,
      "loss_ce": 0.00010112335439771414,
      "loss_iou": 0.2197265625,
      "loss_num": 0.0234375,
      "loss_xval": 0.55859375,
      "num_input_tokens_seen": 369407840,
      "step": 6593
    },
    {
      "epoch": 14.68596881959911,
      "grad_norm": 21.299360275268555,
      "learning_rate": 1e-06,
      "loss": 0.5773,
      "num_input_tokens_seen": 369461952,
      "step": 6594
    },
    {
      "epoch": 14.68596881959911,
      "loss": 0.6562404632568359,
      "loss_ce": 0.0003566770756151527,
      "loss_iou": 0.2890625,
      "loss_num": 0.0155029296875,
      "loss_xval": 0.65625,
      "num_input_tokens_seen": 369461952,
      "step": 6594
    },
    {
      "epoch": 14.688195991091314,
      "grad_norm": 14.860562324523926,
      "learning_rate": 1e-06,
      "loss": 0.6268,
      "num_input_tokens_seen": 369517888,
      "step": 6595
    },
    {
      "epoch": 14.688195991091314,
      "loss": 0.745347797870636,
      "loss_ce": 0.00010857224697247148,
      "loss_iou": 0.333984375,
      "loss_num": 0.01556396484375,
      "loss_xval": 0.74609375,
      "num_input_tokens_seen": 369517888,
      "step": 6595
    },
    {
      "epoch": 14.690423162583519,
      "grad_norm": 18.80927085876465,
      "learning_rate": 1e-06,
      "loss": 0.4626,
      "num_input_tokens_seen": 369575536,
      "step": 6596
    },
    {
      "epoch": 14.690423162583519,
      "loss": 0.5963727831840515,
      "loss_ce": 0.00012036593398079276,
      "loss_iou": 0.263671875,
      "loss_num": 0.01397705078125,
      "loss_xval": 0.59765625,
      "num_input_tokens_seen": 369575536,
      "step": 6596
    },
    {
      "epoch": 14.692650334075724,
      "grad_norm": 19.051189422607422,
      "learning_rate": 1e-06,
      "loss": 0.3497,
      "num_input_tokens_seen": 369633808,
      "step": 6597
    },
    {
      "epoch": 14.692650334075724,
      "loss": 0.4147190749645233,
      "loss_ce": 0.0002903687418438494,
      "loss_iou": 0.1767578125,
      "loss_num": 0.0120849609375,
      "loss_xval": 0.4140625,
      "num_input_tokens_seen": 369633808,
      "step": 6597
    },
    {
      "epoch": 14.694877505567929,
      "grad_norm": 18.837825775146484,
      "learning_rate": 1e-06,
      "loss": 0.4103,
      "num_input_tokens_seen": 369691260,
      "step": 6598
    },
    {
      "epoch": 14.694877505567929,
      "loss": 0.49033933877944946,
      "loss_ce": 0.00010496500908629969,
      "loss_iou": 0.2001953125,
      "loss_num": 0.0179443359375,
      "loss_xval": 0.490234375,
      "num_input_tokens_seen": 369691260,
      "step": 6598
    },
    {
      "epoch": 14.697104677060134,
      "grad_norm": 21.672534942626953,
      "learning_rate": 1e-06,
      "loss": 0.4892,
      "num_input_tokens_seen": 369748488,
      "step": 6599
    },
    {
      "epoch": 14.697104677060134,
      "loss": 0.5086407661437988,
      "loss_ce": 9.587158274371177e-05,
      "loss_iou": 0.224609375,
      "loss_num": 0.01177978515625,
      "loss_xval": 0.5078125,
      "num_input_tokens_seen": 369748488,
      "step": 6599
    },
    {
      "epoch": 14.699331848552339,
      "grad_norm": 17.431072235107422,
      "learning_rate": 1e-06,
      "loss": 0.369,
      "num_input_tokens_seen": 369805384,
      "step": 6600
    },
    {
      "epoch": 14.699331848552339,
      "loss": 0.3214370012283325,
      "loss_ce": 8.691436960361898e-05,
      "loss_iou": 0.1494140625,
      "loss_num": 0.004364013671875,
      "loss_xval": 0.322265625,
      "num_input_tokens_seen": 369805384,
      "step": 6600
    },
    {
      "epoch": 14.701559020044543,
      "grad_norm": 18.672565460205078,
      "learning_rate": 1e-06,
      "loss": 0.4749,
      "num_input_tokens_seen": 369860176,
      "step": 6601
    },
    {
      "epoch": 14.701559020044543,
      "loss": 0.5711730718612671,
      "loss_ce": 0.0001281223667319864,
      "loss_iou": 0.2392578125,
      "loss_num": 0.0185546875,
      "loss_xval": 0.5703125,
      "num_input_tokens_seen": 369860176,
      "step": 6601
    },
    {
      "epoch": 14.703786191536748,
      "grad_norm": 17.336938858032227,
      "learning_rate": 1e-06,
      "loss": 0.4992,
      "num_input_tokens_seen": 369916752,
      "step": 6602
    },
    {
      "epoch": 14.703786191536748,
      "loss": 0.290561705827713,
      "loss_ce": 9.540806786390021e-05,
      "loss_iou": 0.1337890625,
      "loss_num": 0.004669189453125,
      "loss_xval": 0.291015625,
      "num_input_tokens_seen": 369916752,
      "step": 6602
    },
    {
      "epoch": 14.706013363028953,
      "grad_norm": 19.08681297302246,
      "learning_rate": 1e-06,
      "loss": 0.417,
      "num_input_tokens_seen": 369973148,
      "step": 6603
    },
    {
      "epoch": 14.706013363028953,
      "loss": 0.4227031469345093,
      "loss_ce": 9.572567796567455e-05,
      "loss_iou": 0.1845703125,
      "loss_num": 0.0106201171875,
      "loss_xval": 0.421875,
      "num_input_tokens_seen": 369973148,
      "step": 6603
    },
    {
      "epoch": 14.708240534521158,
      "grad_norm": 19.571962356567383,
      "learning_rate": 1e-06,
      "loss": 0.3897,
      "num_input_tokens_seen": 370030248,
      "step": 6604
    },
    {
      "epoch": 14.708240534521158,
      "loss": 0.3606414496898651,
      "loss_ce": 0.00022884068312123418,
      "loss_iou": 0.154296875,
      "loss_num": 0.01043701171875,
      "loss_xval": 0.361328125,
      "num_input_tokens_seen": 370030248,
      "step": 6604
    },
    {
      "epoch": 14.710467706013363,
      "grad_norm": 25.896663665771484,
      "learning_rate": 1e-06,
      "loss": 0.444,
      "num_input_tokens_seen": 370086860,
      "step": 6605
    },
    {
      "epoch": 14.710467706013363,
      "loss": 0.4243563711643219,
      "loss_ce": 0.00010100057988893241,
      "loss_iou": 0.193359375,
      "loss_num": 0.00762939453125,
      "loss_xval": 0.423828125,
      "num_input_tokens_seen": 370086860,
      "step": 6605
    },
    {
      "epoch": 14.712694877505568,
      "grad_norm": 11.470439910888672,
      "learning_rate": 1e-06,
      "loss": 0.4257,
      "num_input_tokens_seen": 370140792,
      "step": 6606
    },
    {
      "epoch": 14.712694877505568,
      "loss": 0.4308934211730957,
      "loss_ce": 0.00010726226173574105,
      "loss_iou": 0.1865234375,
      "loss_num": 0.0115966796875,
      "loss_xval": 0.431640625,
      "num_input_tokens_seen": 370140792,
      "step": 6606
    },
    {
      "epoch": 14.714922048997773,
      "grad_norm": 37.93767547607422,
      "learning_rate": 1e-06,
      "loss": 0.4366,
      "num_input_tokens_seen": 370198188,
      "step": 6607
    },
    {
      "epoch": 14.714922048997773,
      "loss": 0.3961656987667084,
      "loss_ce": 0.000108551379526034,
      "loss_iou": 0.15625,
      "loss_num": 0.0167236328125,
      "loss_xval": 0.396484375,
      "num_input_tokens_seen": 370198188,
      "step": 6607
    },
    {
      "epoch": 14.717149220489977,
      "grad_norm": 19.189268112182617,
      "learning_rate": 1e-06,
      "loss": 0.5345,
      "num_input_tokens_seen": 370256068,
      "step": 6608
    },
    {
      "epoch": 14.717149220489977,
      "loss": 0.5142604112625122,
      "loss_ce": 0.00010023896174971014,
      "loss_iou": 0.2373046875,
      "loss_num": 0.008056640625,
      "loss_xval": 0.515625,
      "num_input_tokens_seen": 370256068,
      "step": 6608
    },
    {
      "epoch": 14.719376391982182,
      "grad_norm": 27.949832916259766,
      "learning_rate": 1e-06,
      "loss": 0.4502,
      "num_input_tokens_seen": 370310132,
      "step": 6609
    },
    {
      "epoch": 14.719376391982182,
      "loss": 0.41979140043258667,
      "loss_ce": 0.000540890556294471,
      "loss_iou": 0.177734375,
      "loss_num": 0.01287841796875,
      "loss_xval": 0.419921875,
      "num_input_tokens_seen": 370310132,
      "step": 6609
    },
    {
      "epoch": 14.721603563474387,
      "grad_norm": 19.57186508178711,
      "learning_rate": 1e-06,
      "loss": 0.4912,
      "num_input_tokens_seen": 370365772,
      "step": 6610
    },
    {
      "epoch": 14.721603563474387,
      "loss": 0.4778703451156616,
      "loss_ce": 8.710511610843241e-05,
      "loss_iou": 0.2109375,
      "loss_num": 0.01129150390625,
      "loss_xval": 0.478515625,
      "num_input_tokens_seen": 370365772,
      "step": 6610
    },
    {
      "epoch": 14.723830734966592,
      "grad_norm": 22.238262176513672,
      "learning_rate": 1e-06,
      "loss": 0.4615,
      "num_input_tokens_seen": 370422604,
      "step": 6611
    },
    {
      "epoch": 14.723830734966592,
      "loss": 0.536961019039154,
      "loss_ce": 0.00021785832359455526,
      "loss_iou": 0.236328125,
      "loss_num": 0.01275634765625,
      "loss_xval": 0.53515625,
      "num_input_tokens_seen": 370422604,
      "step": 6611
    },
    {
      "epoch": 14.726057906458797,
      "grad_norm": 19.965530395507812,
      "learning_rate": 1e-06,
      "loss": 0.4523,
      "num_input_tokens_seen": 370477032,
      "step": 6612
    },
    {
      "epoch": 14.726057906458797,
      "loss": 0.44715508818626404,
      "loss_ce": 0.00013360095908865333,
      "loss_iou": 0.20703125,
      "loss_num": 0.00640869140625,
      "loss_xval": 0.447265625,
      "num_input_tokens_seen": 370477032,
      "step": 6612
    },
    {
      "epoch": 14.728285077951002,
      "grad_norm": 18.242788314819336,
      "learning_rate": 1e-06,
      "loss": 0.5108,
      "num_input_tokens_seen": 370532400,
      "step": 6613
    },
    {
      "epoch": 14.728285077951002,
      "loss": 0.606315553188324,
      "loss_ce": 0.00011437687498982996,
      "loss_iou": 0.265625,
      "loss_num": 0.01507568359375,
      "loss_xval": 0.60546875,
      "num_input_tokens_seen": 370532400,
      "step": 6613
    },
    {
      "epoch": 14.730512249443207,
      "grad_norm": 14.153312683105469,
      "learning_rate": 1e-06,
      "loss": 0.484,
      "num_input_tokens_seen": 370589516,
      "step": 6614
    },
    {
      "epoch": 14.730512249443207,
      "loss": 0.6295005083084106,
      "loss_ce": 0.0001059654459822923,
      "loss_iou": 0.251953125,
      "loss_num": 0.0252685546875,
      "loss_xval": 0.62890625,
      "num_input_tokens_seen": 370589516,
      "step": 6614
    },
    {
      "epoch": 14.732739420935411,
      "grad_norm": 19.10258674621582,
      "learning_rate": 1e-06,
      "loss": 0.465,
      "num_input_tokens_seen": 370647412,
      "step": 6615
    },
    {
      "epoch": 14.732739420935411,
      "loss": 0.6935760974884033,
      "loss_ce": 9.462784510105848e-05,
      "loss_iou": 0.30859375,
      "loss_num": 0.014892578125,
      "loss_xval": 0.6953125,
      "num_input_tokens_seen": 370647412,
      "step": 6615
    },
    {
      "epoch": 14.734966592427616,
      "grad_norm": 14.81308650970459,
      "learning_rate": 1e-06,
      "loss": 0.4206,
      "num_input_tokens_seen": 370706064,
      "step": 6616
    },
    {
      "epoch": 14.734966592427616,
      "loss": 0.5853002071380615,
      "loss_ce": 9.509964729659259e-05,
      "loss_iou": 0.240234375,
      "loss_num": 0.0208740234375,
      "loss_xval": 0.5859375,
      "num_input_tokens_seen": 370706064,
      "step": 6616
    },
    {
      "epoch": 14.737193763919821,
      "grad_norm": 44.200828552246094,
      "learning_rate": 1e-06,
      "loss": 0.4139,
      "num_input_tokens_seen": 370762332,
      "step": 6617
    },
    {
      "epoch": 14.737193763919821,
      "loss": 0.5105452537536621,
      "loss_ce": 0.00016923280782066286,
      "loss_iou": 0.2197265625,
      "loss_num": 0.01409912109375,
      "loss_xval": 0.51171875,
      "num_input_tokens_seen": 370762332,
      "step": 6617
    },
    {
      "epoch": 14.739420935412026,
      "grad_norm": 15.740882873535156,
      "learning_rate": 1e-06,
      "loss": 0.3973,
      "num_input_tokens_seen": 370817860,
      "step": 6618
    },
    {
      "epoch": 14.739420935412026,
      "loss": 0.43589556217193604,
      "loss_ce": 0.00010457850294187665,
      "loss_iou": 0.18359375,
      "loss_num": 0.01373291015625,
      "loss_xval": 0.435546875,
      "num_input_tokens_seen": 370817860,
      "step": 6618
    },
    {
      "epoch": 14.74164810690423,
      "grad_norm": 18.04197120666504,
      "learning_rate": 1e-06,
      "loss": 0.4308,
      "num_input_tokens_seen": 370872024,
      "step": 6619
    },
    {
      "epoch": 14.74164810690423,
      "loss": 0.29858967661857605,
      "loss_ce": 9.725069685373455e-05,
      "loss_iou": 0.1259765625,
      "loss_num": 0.00921630859375,
      "loss_xval": 0.298828125,
      "num_input_tokens_seen": 370872024,
      "step": 6619
    },
    {
      "epoch": 14.743875278396436,
      "grad_norm": 17.093969345092773,
      "learning_rate": 1e-06,
      "loss": 0.4089,
      "num_input_tokens_seen": 370928616,
      "step": 6620
    },
    {
      "epoch": 14.743875278396436,
      "loss": 0.37071430683135986,
      "loss_ce": 0.00010881570779019967,
      "loss_iou": 0.1728515625,
      "loss_num": 0.005035400390625,
      "loss_xval": 0.37109375,
      "num_input_tokens_seen": 370928616,
      "step": 6620
    },
    {
      "epoch": 14.74610244988864,
      "grad_norm": 19.94365882873535,
      "learning_rate": 1e-06,
      "loss": 0.5362,
      "num_input_tokens_seen": 370984480,
      "step": 6621
    },
    {
      "epoch": 14.74610244988864,
      "loss": 0.7782934904098511,
      "loss_ce": 9.5245341071859e-05,
      "loss_iou": 0.35546875,
      "loss_num": 0.01385498046875,
      "loss_xval": 0.77734375,
      "num_input_tokens_seen": 370984480,
      "step": 6621
    },
    {
      "epoch": 14.748329621380847,
      "grad_norm": 19.63141632080078,
      "learning_rate": 1e-06,
      "loss": 0.451,
      "num_input_tokens_seen": 371042128,
      "step": 6622
    },
    {
      "epoch": 14.748329621380847,
      "loss": 0.46288466453552246,
      "loss_ce": 0.00011610893852775916,
      "loss_iou": 0.1962890625,
      "loss_num": 0.01416015625,
      "loss_xval": 0.462890625,
      "num_input_tokens_seen": 371042128,
      "step": 6622
    },
    {
      "epoch": 14.750556792873052,
      "grad_norm": 23.052513122558594,
      "learning_rate": 1e-06,
      "loss": 0.4397,
      "num_input_tokens_seen": 371099580,
      "step": 6623
    },
    {
      "epoch": 14.750556792873052,
      "loss": 0.5583330392837524,
      "loss_ce": 0.00010552619642112404,
      "loss_iou": 0.255859375,
      "loss_num": 0.00970458984375,
      "loss_xval": 0.55859375,
      "num_input_tokens_seen": 371099580,
      "step": 6623
    },
    {
      "epoch": 14.752783964365257,
      "grad_norm": 12.574701309204102,
      "learning_rate": 1e-06,
      "loss": 0.2786,
      "num_input_tokens_seen": 371154964,
      "step": 6624
    },
    {
      "epoch": 14.752783964365257,
      "loss": 0.29406648874282837,
      "loss_ce": 9.068727376870811e-05,
      "loss_iou": 0.1298828125,
      "loss_num": 0.00689697265625,
      "loss_xval": 0.294921875,
      "num_input_tokens_seen": 371154964,
      "step": 6624
    },
    {
      "epoch": 14.755011135857462,
      "grad_norm": 16.99627113342285,
      "learning_rate": 1e-06,
      "loss": 0.5441,
      "num_input_tokens_seen": 371211736,
      "step": 6625
    },
    {
      "epoch": 14.755011135857462,
      "loss": 0.5980323553085327,
      "loss_ce": 0.00013196782674640417,
      "loss_iou": 0.2373046875,
      "loss_num": 0.0245361328125,
      "loss_xval": 0.59765625,
      "num_input_tokens_seen": 371211736,
      "step": 6625
    },
    {
      "epoch": 14.757238307349667,
      "grad_norm": 25.345653533935547,
      "learning_rate": 1e-06,
      "loss": 0.6329,
      "num_input_tokens_seen": 371267888,
      "step": 6626
    },
    {
      "epoch": 14.757238307349667,
      "loss": 0.5716487765312195,
      "loss_ce": 0.00011558398546185344,
      "loss_iou": 0.2578125,
      "loss_num": 0.0113525390625,
      "loss_xval": 0.5703125,
      "num_input_tokens_seen": 371267888,
      "step": 6626
    },
    {
      "epoch": 14.759465478841872,
      "grad_norm": 35.48616027832031,
      "learning_rate": 1e-06,
      "loss": 0.5655,
      "num_input_tokens_seen": 371324636,
      "step": 6627
    },
    {
      "epoch": 14.759465478841872,
      "loss": 0.6881007552146912,
      "loss_ce": 0.00011247480870224535,
      "loss_iou": 0.279296875,
      "loss_num": 0.025634765625,
      "loss_xval": 0.6875,
      "num_input_tokens_seen": 371324636,
      "step": 6627
    },
    {
      "epoch": 14.761692650334076,
      "grad_norm": 17.21990394592285,
      "learning_rate": 1e-06,
      "loss": 0.7006,
      "num_input_tokens_seen": 371381504,
      "step": 6628
    },
    {
      "epoch": 14.761692650334076,
      "loss": 0.7638993263244629,
      "loss_ce": 0.00010532997839618474,
      "loss_iou": 0.302734375,
      "loss_num": 0.03125,
      "loss_xval": 0.765625,
      "num_input_tokens_seen": 371381504,
      "step": 6628
    },
    {
      "epoch": 14.763919821826281,
      "grad_norm": 14.376852035522461,
      "learning_rate": 1e-06,
      "loss": 0.3699,
      "num_input_tokens_seen": 371435912,
      "step": 6629
    },
    {
      "epoch": 14.763919821826281,
      "loss": 0.43231716752052307,
      "loss_ce": 0.00012721640814561397,
      "loss_iou": 0.1767578125,
      "loss_num": 0.0155029296875,
      "loss_xval": 0.431640625,
      "num_input_tokens_seen": 371435912,
      "step": 6629
    },
    {
      "epoch": 14.766146993318486,
      "grad_norm": 18.436325073242188,
      "learning_rate": 1e-06,
      "loss": 0.3864,
      "num_input_tokens_seen": 371491144,
      "step": 6630
    },
    {
      "epoch": 14.766146993318486,
      "loss": 0.47628775238990784,
      "loss_ce": 9.145887452177703e-05,
      "loss_iou": 0.2138671875,
      "loss_num": 0.009765625,
      "loss_xval": 0.4765625,
      "num_input_tokens_seen": 371491144,
      "step": 6630
    },
    {
      "epoch": 14.768374164810691,
      "grad_norm": 19.47183609008789,
      "learning_rate": 1e-06,
      "loss": 0.4151,
      "num_input_tokens_seen": 371548080,
      "step": 6631
    },
    {
      "epoch": 14.768374164810691,
      "loss": 0.4063809812068939,
      "loss_ce": 0.0001309568469878286,
      "loss_iou": 0.17578125,
      "loss_num": 0.01092529296875,
      "loss_xval": 0.40625,
      "num_input_tokens_seen": 371548080,
      "step": 6631
    },
    {
      "epoch": 14.770601336302896,
      "grad_norm": 14.947174072265625,
      "learning_rate": 1e-06,
      "loss": 0.5903,
      "num_input_tokens_seen": 371605028,
      "step": 6632
    },
    {
      "epoch": 14.770601336302896,
      "loss": 0.6663827896118164,
      "loss_ce": 0.00012304184201639146,
      "loss_iou": 0.27734375,
      "loss_num": 0.0223388671875,
      "loss_xval": 0.66796875,
      "num_input_tokens_seen": 371605028,
      "step": 6632
    },
    {
      "epoch": 14.7728285077951,
      "grad_norm": 13.995532989501953,
      "learning_rate": 1e-06,
      "loss": 0.387,
      "num_input_tokens_seen": 371660708,
      "step": 6633
    },
    {
      "epoch": 14.7728285077951,
      "loss": 0.29860347509384155,
      "loss_ce": 8.051018812693655e-05,
      "loss_iou": 0.1259765625,
      "loss_num": 0.00927734375,
      "loss_xval": 0.298828125,
      "num_input_tokens_seen": 371660708,
      "step": 6633
    },
    {
      "epoch": 14.775055679287306,
      "grad_norm": 20.504301071166992,
      "learning_rate": 1e-06,
      "loss": 0.5561,
      "num_input_tokens_seen": 371717676,
      "step": 6634
    },
    {
      "epoch": 14.775055679287306,
      "loss": 0.6223191022872925,
      "loss_ce": 0.0002488004101905972,
      "loss_iou": 0.271484375,
      "loss_num": 0.015625,
      "loss_xval": 0.62109375,
      "num_input_tokens_seen": 371717676,
      "step": 6634
    },
    {
      "epoch": 14.77728285077951,
      "grad_norm": 14.648569107055664,
      "learning_rate": 1e-06,
      "loss": 0.3508,
      "num_input_tokens_seen": 371772420,
      "step": 6635
    },
    {
      "epoch": 14.77728285077951,
      "loss": 0.4088403880596161,
      "loss_ce": 0.000148980543599464,
      "loss_iou": 0.166015625,
      "loss_num": 0.01556396484375,
      "loss_xval": 0.408203125,
      "num_input_tokens_seen": 371772420,
      "step": 6635
    },
    {
      "epoch": 14.779510022271715,
      "grad_norm": 17.389978408813477,
      "learning_rate": 1e-06,
      "loss": 0.6203,
      "num_input_tokens_seen": 371827592,
      "step": 6636
    },
    {
      "epoch": 14.779510022271715,
      "loss": 0.5807891488075256,
      "loss_ce": 0.00010066662798635662,
      "loss_iou": 0.2578125,
      "loss_num": 0.0130615234375,
      "loss_xval": 0.58203125,
      "num_input_tokens_seen": 371827592,
      "step": 6636
    },
    {
      "epoch": 14.78173719376392,
      "grad_norm": 15.41741943359375,
      "learning_rate": 1e-06,
      "loss": 0.3496,
      "num_input_tokens_seen": 371883112,
      "step": 6637
    },
    {
      "epoch": 14.78173719376392,
      "loss": 0.36627358198165894,
      "loss_ce": 0.0001541778037790209,
      "loss_iou": 0.1513671875,
      "loss_num": 0.0125732421875,
      "loss_xval": 0.365234375,
      "num_input_tokens_seen": 371883112,
      "step": 6637
    },
    {
      "epoch": 14.783964365256125,
      "grad_norm": 11.838160514831543,
      "learning_rate": 1e-06,
      "loss": 0.4474,
      "num_input_tokens_seen": 371939696,
      "step": 6638
    },
    {
      "epoch": 14.783964365256125,
      "loss": 0.49210840463638306,
      "loss_ce": 0.0001039962880895473,
      "loss_iou": 0.203125,
      "loss_num": 0.0174560546875,
      "loss_xval": 0.4921875,
      "num_input_tokens_seen": 371939696,
      "step": 6638
    },
    {
      "epoch": 14.78619153674833,
      "grad_norm": 20.356592178344727,
      "learning_rate": 1e-06,
      "loss": 0.3329,
      "num_input_tokens_seen": 371998200,
      "step": 6639
    },
    {
      "epoch": 14.78619153674833,
      "loss": 0.32651287317276,
      "loss_ce": 9.684590622782707e-05,
      "loss_iou": 0.1474609375,
      "loss_num": 0.00616455078125,
      "loss_xval": 0.326171875,
      "num_input_tokens_seen": 371998200,
      "step": 6639
    },
    {
      "epoch": 14.788418708240535,
      "grad_norm": 16.34804344177246,
      "learning_rate": 1e-06,
      "loss": 0.4732,
      "num_input_tokens_seen": 372054520,
      "step": 6640
    },
    {
      "epoch": 14.788418708240535,
      "loss": 0.386675626039505,
      "loss_ce": 7.893913425505161e-05,
      "loss_iou": 0.158203125,
      "loss_num": 0.01409912109375,
      "loss_xval": 0.38671875,
      "num_input_tokens_seen": 372054520,
      "step": 6640
    },
    {
      "epoch": 14.79064587973274,
      "grad_norm": 31.840667724609375,
      "learning_rate": 1e-06,
      "loss": 0.5544,
      "num_input_tokens_seen": 372109948,
      "step": 6641
    },
    {
      "epoch": 14.79064587973274,
      "loss": 0.5972850322723389,
      "loss_ce": 0.00011702888878062367,
      "loss_iou": 0.25,
      "loss_num": 0.019287109375,
      "loss_xval": 0.59765625,
      "num_input_tokens_seen": 372109948,
      "step": 6641
    },
    {
      "epoch": 14.792873051224944,
      "grad_norm": 16.975744247436523,
      "learning_rate": 1e-06,
      "loss": 0.4895,
      "num_input_tokens_seen": 372165020,
      "step": 6642
    },
    {
      "epoch": 14.792873051224944,
      "loss": 0.4539491832256317,
      "loss_ce": 9.175058221444488e-05,
      "loss_iou": 0.2041015625,
      "loss_num": 0.0091552734375,
      "loss_xval": 0.453125,
      "num_input_tokens_seen": 372165020,
      "step": 6642
    },
    {
      "epoch": 14.79510022271715,
      "grad_norm": 17.83343505859375,
      "learning_rate": 1e-06,
      "loss": 0.7496,
      "num_input_tokens_seen": 372220692,
      "step": 6643
    },
    {
      "epoch": 14.79510022271715,
      "loss": 0.9950196743011475,
      "loss_ce": 0.0001466473040636629,
      "loss_iou": 0.412109375,
      "loss_num": 0.034423828125,
      "loss_xval": 0.99609375,
      "num_input_tokens_seen": 372220692,
      "step": 6643
    },
    {
      "epoch": 14.797327394209354,
      "grad_norm": 13.327498435974121,
      "learning_rate": 1e-06,
      "loss": 0.4146,
      "num_input_tokens_seen": 372275868,
      "step": 6644
    },
    {
      "epoch": 14.797327394209354,
      "loss": 0.3324364423751831,
      "loss_ce": 0.00010003681381931528,
      "loss_iou": 0.1533203125,
      "loss_num": 0.005035400390625,
      "loss_xval": 0.33203125,
      "num_input_tokens_seen": 372275868,
      "step": 6644
    },
    {
      "epoch": 14.799554565701559,
      "grad_norm": 20.737871170043945,
      "learning_rate": 1e-06,
      "loss": 0.5179,
      "num_input_tokens_seen": 372327744,
      "step": 6645
    },
    {
      "epoch": 14.799554565701559,
      "loss": 0.3846180737018585,
      "loss_ce": 9.659097850089893e-05,
      "loss_iou": 0.171875,
      "loss_num": 0.008056640625,
      "loss_xval": 0.384765625,
      "num_input_tokens_seen": 372327744,
      "step": 6645
    },
    {
      "epoch": 14.801781737193764,
      "grad_norm": 19.62179946899414,
      "learning_rate": 1e-06,
      "loss": 0.5129,
      "num_input_tokens_seen": 372384104,
      "step": 6646
    },
    {
      "epoch": 14.801781737193764,
      "loss": 0.31222450733184814,
      "loss_ce": 9.073612454812974e-05,
      "loss_iou": 0.11669921875,
      "loss_num": 0.0157470703125,
      "loss_xval": 0.3125,
      "num_input_tokens_seen": 372384104,
      "step": 6646
    },
    {
      "epoch": 14.804008908685969,
      "grad_norm": 19.51956558227539,
      "learning_rate": 1e-06,
      "loss": 0.4317,
      "num_input_tokens_seen": 372440792,
      "step": 6647
    },
    {
      "epoch": 14.804008908685969,
      "loss": 0.40453433990478516,
      "loss_ce": 0.00011538183025550097,
      "loss_iou": 0.1806640625,
      "loss_num": 0.0084228515625,
      "loss_xval": 0.404296875,
      "num_input_tokens_seen": 372440792,
      "step": 6647
    },
    {
      "epoch": 14.806236080178174,
      "grad_norm": 18.005043029785156,
      "learning_rate": 1e-06,
      "loss": 0.5158,
      "num_input_tokens_seen": 372496276,
      "step": 6648
    },
    {
      "epoch": 14.806236080178174,
      "loss": 0.7478663921356201,
      "loss_ce": 0.00030782315297983587,
      "loss_iou": 0.333984375,
      "loss_num": 0.0159912109375,
      "loss_xval": 0.74609375,
      "num_input_tokens_seen": 372496276,
      "step": 6648
    },
    {
      "epoch": 14.808463251670378,
      "grad_norm": 15.16850757598877,
      "learning_rate": 1e-06,
      "loss": 0.6658,
      "num_input_tokens_seen": 372554104,
      "step": 6649
    },
    {
      "epoch": 14.808463251670378,
      "loss": 0.7384415864944458,
      "loss_ce": 0.00016030135157052428,
      "loss_iou": 0.26171875,
      "loss_num": 0.043212890625,
      "loss_xval": 0.73828125,
      "num_input_tokens_seen": 372554104,
      "step": 6649
    },
    {
      "epoch": 14.810690423162583,
      "grad_norm": 15.439560890197754,
      "learning_rate": 1e-06,
      "loss": 0.4836,
      "num_input_tokens_seen": 372609652,
      "step": 6650
    },
    {
      "epoch": 14.810690423162583,
      "loss": 0.499762624502182,
      "loss_ce": 0.0002508952165953815,
      "loss_iou": 0.201171875,
      "loss_num": 0.0191650390625,
      "loss_xval": 0.5,
      "num_input_tokens_seen": 372609652,
      "step": 6650
    },
    {
      "epoch": 14.812917594654788,
      "grad_norm": 17.865459442138672,
      "learning_rate": 1e-06,
      "loss": 0.5121,
      "num_input_tokens_seen": 372665168,
      "step": 6651
    },
    {
      "epoch": 14.812917594654788,
      "loss": 0.5034142136573792,
      "loss_ce": 0.00011833346798084676,
      "loss_iou": 0.2099609375,
      "loss_num": 0.016845703125,
      "loss_xval": 0.50390625,
      "num_input_tokens_seen": 372665168,
      "step": 6651
    },
    {
      "epoch": 14.815144766146993,
      "grad_norm": 24.064001083374023,
      "learning_rate": 1e-06,
      "loss": 0.4675,
      "num_input_tokens_seen": 372719608,
      "step": 6652
    },
    {
      "epoch": 14.815144766146993,
      "loss": 0.550445020198822,
      "loss_ce": 0.00015209712728392333,
      "loss_iou": 0.255859375,
      "loss_num": 0.00732421875,
      "loss_xval": 0.55078125,
      "num_input_tokens_seen": 372719608,
      "step": 6652
    },
    {
      "epoch": 14.817371937639198,
      "grad_norm": 19.275604248046875,
      "learning_rate": 1e-06,
      "loss": 0.3122,
      "num_input_tokens_seen": 372777240,
      "step": 6653
    },
    {
      "epoch": 14.817371937639198,
      "loss": 0.3434959650039673,
      "loss_ce": 0.00011215943231945857,
      "loss_iou": 0.140625,
      "loss_num": 0.0125732421875,
      "loss_xval": 0.34375,
      "num_input_tokens_seen": 372777240,
      "step": 6653
    },
    {
      "epoch": 14.819599109131403,
      "grad_norm": 24.118017196655273,
      "learning_rate": 1e-06,
      "loss": 0.5053,
      "num_input_tokens_seen": 372828860,
      "step": 6654
    },
    {
      "epoch": 14.819599109131403,
      "loss": 0.48031556606292725,
      "loss_ce": 9.092504478758201e-05,
      "loss_iou": 0.189453125,
      "loss_num": 0.0201416015625,
      "loss_xval": 0.48046875,
      "num_input_tokens_seen": 372828860,
      "step": 6654
    },
    {
      "epoch": 14.821826280623608,
      "grad_norm": 15.35105037689209,
      "learning_rate": 1e-06,
      "loss": 0.4081,
      "num_input_tokens_seen": 372886876,
      "step": 6655
    },
    {
      "epoch": 14.821826280623608,
      "loss": 0.3949987590312958,
      "loss_ce": 0.00010130574082722887,
      "loss_iou": 0.1650390625,
      "loss_num": 0.01287841796875,
      "loss_xval": 0.39453125,
      "num_input_tokens_seen": 372886876,
      "step": 6655
    },
    {
      "epoch": 14.824053452115812,
      "grad_norm": 22.87786293029785,
      "learning_rate": 1e-06,
      "loss": 0.5704,
      "num_input_tokens_seen": 372943852,
      "step": 6656
    },
    {
      "epoch": 14.824053452115812,
      "loss": 0.5047581195831299,
      "loss_ce": 0.00011940376134589314,
      "loss_iou": 0.203125,
      "loss_num": 0.01953125,
      "loss_xval": 0.50390625,
      "num_input_tokens_seen": 372943852,
      "step": 6656
    },
    {
      "epoch": 14.826280623608017,
      "grad_norm": 14.802397727966309,
      "learning_rate": 1e-06,
      "loss": 0.2924,
      "num_input_tokens_seen": 373000588,
      "step": 6657
    },
    {
      "epoch": 14.826280623608017,
      "loss": 0.2866537570953369,
      "loss_ce": 9.368563769385219e-05,
      "loss_iou": 0.12890625,
      "loss_num": 0.00555419921875,
      "loss_xval": 0.287109375,
      "num_input_tokens_seen": 373000588,
      "step": 6657
    },
    {
      "epoch": 14.828507795100222,
      "grad_norm": 24.41092872619629,
      "learning_rate": 1e-06,
      "loss": 0.4099,
      "num_input_tokens_seen": 373058656,
      "step": 6658
    },
    {
      "epoch": 14.828507795100222,
      "loss": 0.4437221884727478,
      "loss_ce": 0.00011863959662150592,
      "loss_iou": 0.1923828125,
      "loss_num": 0.01190185546875,
      "loss_xval": 0.443359375,
      "num_input_tokens_seen": 373058656,
      "step": 6658
    },
    {
      "epoch": 14.830734966592427,
      "grad_norm": 30.71437644958496,
      "learning_rate": 1e-06,
      "loss": 0.4298,
      "num_input_tokens_seen": 373112252,
      "step": 6659
    },
    {
      "epoch": 14.830734966592427,
      "loss": 0.3821251690387726,
      "loss_ce": 0.00010613477934384719,
      "loss_iou": 0.166015625,
      "loss_num": 0.010009765625,
      "loss_xval": 0.3828125,
      "num_input_tokens_seen": 373112252,
      "step": 6659
    },
    {
      "epoch": 14.832962138084632,
      "grad_norm": 14.227256774902344,
      "learning_rate": 1e-06,
      "loss": 0.4838,
      "num_input_tokens_seen": 373169236,
      "step": 6660
    },
    {
      "epoch": 14.832962138084632,
      "loss": 0.4358832836151123,
      "loss_ce": 9.228321141563356e-05,
      "loss_iou": 0.1953125,
      "loss_num": 0.00909423828125,
      "loss_xval": 0.435546875,
      "num_input_tokens_seen": 373169236,
      "step": 6660
    },
    {
      "epoch": 14.835189309576837,
      "grad_norm": 17.108369827270508,
      "learning_rate": 1e-06,
      "loss": 0.6556,
      "num_input_tokens_seen": 373227608,
      "step": 6661
    },
    {
      "epoch": 14.835189309576837,
      "loss": 0.8442061543464661,
      "loss_ce": 0.00015093988622538745,
      "loss_iou": 0.333984375,
      "loss_num": 0.034912109375,
      "loss_xval": 0.84375,
      "num_input_tokens_seen": 373227608,
      "step": 6661
    },
    {
      "epoch": 14.837416481069042,
      "grad_norm": 15.365765571594238,
      "learning_rate": 1e-06,
      "loss": 0.5749,
      "num_input_tokens_seen": 373280432,
      "step": 6662
    },
    {
      "epoch": 14.837416481069042,
      "loss": 0.6416932940483093,
      "loss_ce": 9.174088336294517e-05,
      "loss_iou": 0.265625,
      "loss_num": 0.021728515625,
      "loss_xval": 0.640625,
      "num_input_tokens_seen": 373280432,
      "step": 6662
    },
    {
      "epoch": 14.839643652561247,
      "grad_norm": 20.231231689453125,
      "learning_rate": 1e-06,
      "loss": 0.5829,
      "num_input_tokens_seen": 373335292,
      "step": 6663
    },
    {
      "epoch": 14.839643652561247,
      "loss": 0.5822228193283081,
      "loss_ce": 0.00013056171883363277,
      "loss_iou": 0.248046875,
      "loss_num": 0.01708984375,
      "loss_xval": 0.58203125,
      "num_input_tokens_seen": 373335292,
      "step": 6663
    },
    {
      "epoch": 14.841870824053451,
      "grad_norm": 14.344491004943848,
      "learning_rate": 1e-06,
      "loss": 0.3425,
      "num_input_tokens_seen": 373391224,
      "step": 6664
    },
    {
      "epoch": 14.841870824053451,
      "loss": 0.4869232773780823,
      "loss_ce": 0.00010687689064070582,
      "loss_iou": 0.201171875,
      "loss_num": 0.0167236328125,
      "loss_xval": 0.486328125,
      "num_input_tokens_seen": 373391224,
      "step": 6664
    },
    {
      "epoch": 14.844097995545656,
      "grad_norm": 18.94646644592285,
      "learning_rate": 1e-06,
      "loss": 0.4103,
      "num_input_tokens_seen": 373448636,
      "step": 6665
    },
    {
      "epoch": 14.844097995545656,
      "loss": 0.34946849942207336,
      "loss_ce": 0.00010326325718779117,
      "loss_iou": 0.1640625,
      "loss_num": 0.004180908203125,
      "loss_xval": 0.349609375,
      "num_input_tokens_seen": 373448636,
      "step": 6665
    },
    {
      "epoch": 14.846325167037861,
      "grad_norm": 26.247119903564453,
      "learning_rate": 1e-06,
      "loss": 0.4845,
      "num_input_tokens_seen": 373505896,
      "step": 6666
    },
    {
      "epoch": 14.846325167037861,
      "loss": 0.44346243143081665,
      "loss_ce": 0.0001030644925776869,
      "loss_iou": 0.18359375,
      "loss_num": 0.01513671875,
      "loss_xval": 0.443359375,
      "num_input_tokens_seen": 373505896,
      "step": 6666
    },
    {
      "epoch": 14.848552338530066,
      "grad_norm": 16.330686569213867,
      "learning_rate": 1e-06,
      "loss": 0.3428,
      "num_input_tokens_seen": 373562748,
      "step": 6667
    },
    {
      "epoch": 14.848552338530066,
      "loss": 0.3223639130592346,
      "loss_ce": 9.82833735179156e-05,
      "loss_iou": 0.1337890625,
      "loss_num": 0.01080322265625,
      "loss_xval": 0.322265625,
      "num_input_tokens_seen": 373562748,
      "step": 6667
    },
    {
      "epoch": 14.85077951002227,
      "grad_norm": 17.0013484954834,
      "learning_rate": 1e-06,
      "loss": 0.4217,
      "num_input_tokens_seen": 373620644,
      "step": 6668
    },
    {
      "epoch": 14.85077951002227,
      "loss": 0.4391821026802063,
      "loss_ce": 9.516206046100706e-05,
      "loss_iou": 0.1962890625,
      "loss_num": 0.00927734375,
      "loss_xval": 0.439453125,
      "num_input_tokens_seen": 373620644,
      "step": 6668
    },
    {
      "epoch": 14.853006681514476,
      "grad_norm": 18.03797721862793,
      "learning_rate": 1e-06,
      "loss": 0.5505,
      "num_input_tokens_seen": 373675496,
      "step": 6669
    },
    {
      "epoch": 14.853006681514476,
      "loss": 0.6275442242622375,
      "loss_ce": 0.00010283813753630966,
      "loss_iou": 0.279296875,
      "loss_num": 0.013427734375,
      "loss_xval": 0.62890625,
      "num_input_tokens_seen": 373675496,
      "step": 6669
    },
    {
      "epoch": 14.855233853006682,
      "grad_norm": 18.613082885742188,
      "learning_rate": 1e-06,
      "loss": 0.4396,
      "num_input_tokens_seen": 373729180,
      "step": 6670
    },
    {
      "epoch": 14.855233853006682,
      "loss": 0.28043514490127563,
      "loss_ce": 0.00010065691458294168,
      "loss_iou": 0.119140625,
      "loss_num": 0.00830078125,
      "loss_xval": 0.28125,
      "num_input_tokens_seen": 373729180,
      "step": 6670
    },
    {
      "epoch": 14.857461024498887,
      "grad_norm": 15.190927505493164,
      "learning_rate": 1e-06,
      "loss": 0.632,
      "num_input_tokens_seen": 373785848,
      "step": 6671
    },
    {
      "epoch": 14.857461024498887,
      "loss": 0.4553018808364868,
      "loss_ce": 0.00010172194743063301,
      "loss_iou": 0.1884765625,
      "loss_num": 0.01556396484375,
      "loss_xval": 0.455078125,
      "num_input_tokens_seen": 373785848,
      "step": 6671
    },
    {
      "epoch": 14.859688195991092,
      "grad_norm": 14.331643104553223,
      "learning_rate": 1e-06,
      "loss": 0.5239,
      "num_input_tokens_seen": 373840668,
      "step": 6672
    },
    {
      "epoch": 14.859688195991092,
      "loss": 0.386011004447937,
      "loss_ce": 8.569139026803896e-05,
      "loss_iou": 0.1494140625,
      "loss_num": 0.017333984375,
      "loss_xval": 0.38671875,
      "num_input_tokens_seen": 373840668,
      "step": 6672
    },
    {
      "epoch": 14.861915367483297,
      "grad_norm": 14.409646987915039,
      "learning_rate": 1e-06,
      "loss": 0.5363,
      "num_input_tokens_seen": 373895844,
      "step": 6673
    },
    {
      "epoch": 14.861915367483297,
      "loss": 0.6987284421920776,
      "loss_ce": 0.00012002349831163883,
      "loss_iou": 0.3203125,
      "loss_num": 0.0118408203125,
      "loss_xval": 0.69921875,
      "num_input_tokens_seen": 373895844,
      "step": 6673
    },
    {
      "epoch": 14.864142538975502,
      "grad_norm": 18.301753997802734,
      "learning_rate": 1e-06,
      "loss": 0.4477,
      "num_input_tokens_seen": 373949004,
      "step": 6674
    },
    {
      "epoch": 14.864142538975502,
      "loss": 0.45737865567207336,
      "loss_ce": 0.00010326381016056985,
      "loss_iou": 0.1982421875,
      "loss_num": 0.01220703125,
      "loss_xval": 0.45703125,
      "num_input_tokens_seen": 373949004,
      "step": 6674
    },
    {
      "epoch": 14.866369710467707,
      "grad_norm": 12.914517402648926,
      "learning_rate": 1e-06,
      "loss": 0.4054,
      "num_input_tokens_seen": 374005364,
      "step": 6675
    },
    {
      "epoch": 14.866369710467707,
      "loss": 0.3800061047077179,
      "loss_ce": 0.0001232977374456823,
      "loss_iou": 0.1748046875,
      "loss_num": 0.006072998046875,
      "loss_xval": 0.37890625,
      "num_input_tokens_seen": 374005364,
      "step": 6675
    },
    {
      "epoch": 14.868596881959911,
      "grad_norm": 17.27826690673828,
      "learning_rate": 1e-06,
      "loss": 0.3994,
      "num_input_tokens_seen": 374060220,
      "step": 6676
    },
    {
      "epoch": 14.868596881959911,
      "loss": 0.3738711476325989,
      "loss_ce": 9.184792725136504e-05,
      "loss_iou": 0.1611328125,
      "loss_num": 0.01025390625,
      "loss_xval": 0.373046875,
      "num_input_tokens_seen": 374060220,
      "step": 6676
    },
    {
      "epoch": 14.870824053452116,
      "grad_norm": 19.53494644165039,
      "learning_rate": 1e-06,
      "loss": 0.2948,
      "num_input_tokens_seen": 374117248,
      "step": 6677
    },
    {
      "epoch": 14.870824053452116,
      "loss": 0.31820207834243774,
      "loss_ce": 8.682158659212291e-05,
      "loss_iou": 0.134765625,
      "loss_num": 0.009521484375,
      "loss_xval": 0.318359375,
      "num_input_tokens_seen": 374117248,
      "step": 6677
    },
    {
      "epoch": 14.873051224944321,
      "grad_norm": 25.144716262817383,
      "learning_rate": 1e-06,
      "loss": 0.5595,
      "num_input_tokens_seen": 374170888,
      "step": 6678
    },
    {
      "epoch": 14.873051224944321,
      "loss": 0.7696267366409302,
      "loss_ce": 9.549126116326079e-05,
      "loss_iou": 0.287109375,
      "loss_num": 0.038818359375,
      "loss_xval": 0.76953125,
      "num_input_tokens_seen": 374170888,
      "step": 6678
    },
    {
      "epoch": 14.875278396436526,
      "grad_norm": 26.434158325195312,
      "learning_rate": 1e-06,
      "loss": 0.3463,
      "num_input_tokens_seen": 374229660,
      "step": 6679
    },
    {
      "epoch": 14.875278396436526,
      "loss": 0.5142629146575928,
      "loss_ce": 0.00010273464431520551,
      "loss_iou": 0.236328125,
      "loss_num": 0.00823974609375,
      "loss_xval": 0.515625,
      "num_input_tokens_seen": 374229660,
      "step": 6679
    },
    {
      "epoch": 14.877505567928731,
      "grad_norm": 18.97011947631836,
      "learning_rate": 1e-06,
      "loss": 0.6984,
      "num_input_tokens_seen": 374285568,
      "step": 6680
    },
    {
      "epoch": 14.877505567928731,
      "loss": 0.852816641330719,
      "loss_ce": 0.00015551211254205555,
      "loss_iou": 0.33984375,
      "loss_num": 0.0341796875,
      "loss_xval": 0.8515625,
      "num_input_tokens_seen": 374285568,
      "step": 6680
    },
    {
      "epoch": 14.879732739420936,
      "grad_norm": 23.420482635498047,
      "learning_rate": 1e-06,
      "loss": 0.5408,
      "num_input_tokens_seen": 374343228,
      "step": 6681
    },
    {
      "epoch": 14.879732739420936,
      "loss": 0.7148338556289673,
      "loss_ce": 0.0004784108605235815,
      "loss_iou": 0.296875,
      "loss_num": 0.02392578125,
      "loss_xval": 0.71484375,
      "num_input_tokens_seen": 374343228,
      "step": 6681
    },
    {
      "epoch": 14.88195991091314,
      "grad_norm": 19.06096076965332,
      "learning_rate": 1e-06,
      "loss": 0.4187,
      "num_input_tokens_seen": 374399024,
      "step": 6682
    },
    {
      "epoch": 14.88195991091314,
      "loss": 0.29825353622436523,
      "loss_ce": 9.67753876466304e-05,
      "loss_iou": 0.1337890625,
      "loss_num": 0.0059814453125,
      "loss_xval": 0.298828125,
      "num_input_tokens_seen": 374399024,
      "step": 6682
    },
    {
      "epoch": 14.884187082405345,
      "grad_norm": 25.65113639831543,
      "learning_rate": 1e-06,
      "loss": 0.5878,
      "num_input_tokens_seen": 374456632,
      "step": 6683
    },
    {
      "epoch": 14.884187082405345,
      "loss": 0.6034575700759888,
      "loss_ce": 9.450462675886229e-05,
      "loss_iou": 0.26953125,
      "loss_num": 0.01287841796875,
      "loss_xval": 0.6015625,
      "num_input_tokens_seen": 374456632,
      "step": 6683
    },
    {
      "epoch": 14.88641425389755,
      "grad_norm": 18.01433753967285,
      "learning_rate": 1e-06,
      "loss": 0.4982,
      "num_input_tokens_seen": 374511092,
      "step": 6684
    },
    {
      "epoch": 14.88641425389755,
      "loss": 0.5666952133178711,
      "loss_ce": 0.00010584105621092021,
      "loss_iou": 0.232421875,
      "loss_num": 0.020263671875,
      "loss_xval": 0.56640625,
      "num_input_tokens_seen": 374511092,
      "step": 6684
    },
    {
      "epoch": 14.888641425389755,
      "grad_norm": 19.60915756225586,
      "learning_rate": 1e-06,
      "loss": 0.4411,
      "num_input_tokens_seen": 374566352,
      "step": 6685
    },
    {
      "epoch": 14.888641425389755,
      "loss": 0.5250272154808044,
      "loss_ce": 0.00012483444879762828,
      "loss_iou": 0.240234375,
      "loss_num": 0.009033203125,
      "loss_xval": 0.5234375,
      "num_input_tokens_seen": 374566352,
      "step": 6685
    },
    {
      "epoch": 14.89086859688196,
      "grad_norm": 22.95705795288086,
      "learning_rate": 1e-06,
      "loss": 0.454,
      "num_input_tokens_seen": 374624096,
      "step": 6686
    },
    {
      "epoch": 14.89086859688196,
      "loss": 0.3750418722629547,
      "loss_ce": 0.00010290060890838504,
      "loss_iou": 0.15625,
      "loss_num": 0.01251220703125,
      "loss_xval": 0.375,
      "num_input_tokens_seen": 374624096,
      "step": 6686
    },
    {
      "epoch": 14.893095768374165,
      "grad_norm": 21.94954490661621,
      "learning_rate": 1e-06,
      "loss": 0.5847,
      "num_input_tokens_seen": 374678944,
      "step": 6687
    },
    {
      "epoch": 14.893095768374165,
      "loss": 0.6209622621536255,
      "loss_ce": 0.0001126552015193738,
      "loss_iou": 0.263671875,
      "loss_num": 0.0186767578125,
      "loss_xval": 0.62109375,
      "num_input_tokens_seen": 374678944,
      "step": 6687
    },
    {
      "epoch": 14.89532293986637,
      "grad_norm": 21.50327491760254,
      "learning_rate": 1e-06,
      "loss": 0.5005,
      "num_input_tokens_seen": 374736156,
      "step": 6688
    },
    {
      "epoch": 14.89532293986637,
      "loss": 0.420645534992218,
      "loss_ce": 0.00011328914115438238,
      "loss_iou": 0.201171875,
      "loss_num": 0.0038299560546875,
      "loss_xval": 0.419921875,
      "num_input_tokens_seen": 374736156,
      "step": 6688
    },
    {
      "epoch": 14.897550111358575,
      "grad_norm": 25.155494689941406,
      "learning_rate": 1e-06,
      "loss": 0.4771,
      "num_input_tokens_seen": 374792696,
      "step": 6689
    },
    {
      "epoch": 14.897550111358575,
      "loss": 0.4023246765136719,
      "loss_ce": 0.00010299000859959051,
      "loss_iou": 0.173828125,
      "loss_num": 0.01116943359375,
      "loss_xval": 0.40234375,
      "num_input_tokens_seen": 374792696,
      "step": 6689
    },
    {
      "epoch": 14.89977728285078,
      "grad_norm": 15.196331024169922,
      "learning_rate": 1e-06,
      "loss": 0.2636,
      "num_input_tokens_seen": 374849908,
      "step": 6690
    },
    {
      "epoch": 14.89977728285078,
      "loss": 0.24606722593307495,
      "loss_ce": 0.0009500437881797552,
      "loss_iou": 0.1123046875,
      "loss_num": 0.003997802734375,
      "loss_xval": 0.2451171875,
      "num_input_tokens_seen": 374849908,
      "step": 6690
    },
    {
      "epoch": 14.902004454342984,
      "grad_norm": 26.573570251464844,
      "learning_rate": 1e-06,
      "loss": 0.4118,
      "num_input_tokens_seen": 374904924,
      "step": 6691
    },
    {
      "epoch": 14.902004454342984,
      "loss": 0.3819239139556885,
      "loss_ce": 8.79746803548187e-05,
      "loss_iou": 0.1611328125,
      "loss_num": 0.0118408203125,
      "loss_xval": 0.3828125,
      "num_input_tokens_seen": 374904924,
      "step": 6691
    },
    {
      "epoch": 14.90423162583519,
      "grad_norm": 29.359655380249023,
      "learning_rate": 1e-06,
      "loss": 0.4543,
      "num_input_tokens_seen": 374959508,
      "step": 6692
    },
    {
      "epoch": 14.90423162583519,
      "loss": 0.28562378883361816,
      "loss_ce": 0.00010133428440894932,
      "loss_iou": 0.11474609375,
      "loss_num": 0.01116943359375,
      "loss_xval": 0.28515625,
      "num_input_tokens_seen": 374959508,
      "step": 6692
    },
    {
      "epoch": 14.906458797327394,
      "grad_norm": 16.882619857788086,
      "learning_rate": 1e-06,
      "loss": 0.5222,
      "num_input_tokens_seen": 375016108,
      "step": 6693
    },
    {
      "epoch": 14.906458797327394,
      "loss": 0.4062420725822449,
      "loss_ce": 0.00011414792970754206,
      "loss_iou": 0.18359375,
      "loss_num": 0.007781982421875,
      "loss_xval": 0.40625,
      "num_input_tokens_seen": 375016108,
      "step": 6693
    },
    {
      "epoch": 14.908685968819599,
      "grad_norm": 15.009581565856934,
      "learning_rate": 1e-06,
      "loss": 0.4768,
      "num_input_tokens_seen": 375073380,
      "step": 6694
    },
    {
      "epoch": 14.908685968819599,
      "loss": 0.5022239685058594,
      "loss_ce": 0.00014879369700793177,
      "loss_iou": 0.2255859375,
      "loss_num": 0.01025390625,
      "loss_xval": 0.50390625,
      "num_input_tokens_seen": 375073380,
      "step": 6694
    },
    {
      "epoch": 14.910913140311804,
      "grad_norm": 16.225059509277344,
      "learning_rate": 1e-06,
      "loss": 0.3288,
      "num_input_tokens_seen": 375129260,
      "step": 6695
    },
    {
      "epoch": 14.910913140311804,
      "loss": 0.23318275809288025,
      "loss_ce": 8.948949835030362e-05,
      "loss_iou": 0.10009765625,
      "loss_num": 0.006500244140625,
      "loss_xval": 0.2333984375,
      "num_input_tokens_seen": 375129260,
      "step": 6695
    },
    {
      "epoch": 14.913140311804009,
      "grad_norm": 18.727340698242188,
      "learning_rate": 1e-06,
      "loss": 0.4627,
      "num_input_tokens_seen": 375184036,
      "step": 6696
    },
    {
      "epoch": 14.913140311804009,
      "loss": 0.5411036610603333,
      "loss_ce": 8.803869422990829e-05,
      "loss_iou": 0.2158203125,
      "loss_num": 0.0218505859375,
      "loss_xval": 0.5390625,
      "num_input_tokens_seen": 375184036,
      "step": 6696
    },
    {
      "epoch": 14.915367483296214,
      "grad_norm": 14.84618854522705,
      "learning_rate": 1e-06,
      "loss": 0.3687,
      "num_input_tokens_seen": 375240860,
      "step": 6697
    },
    {
      "epoch": 14.915367483296214,
      "loss": 0.3764420747756958,
      "loss_ce": 9.930084343068302e-05,
      "loss_iou": 0.1748046875,
      "loss_num": 0.00543212890625,
      "loss_xval": 0.376953125,
      "num_input_tokens_seen": 375240860,
      "step": 6697
    },
    {
      "epoch": 14.917594654788418,
      "grad_norm": 14.689440727233887,
      "learning_rate": 1e-06,
      "loss": 0.5437,
      "num_input_tokens_seen": 375297676,
      "step": 6698
    },
    {
      "epoch": 14.917594654788418,
      "loss": 0.5835915803909302,
      "loss_ce": 9.553229756420478e-05,
      "loss_iou": 0.2578125,
      "loss_num": 0.0133056640625,
      "loss_xval": 0.58203125,
      "num_input_tokens_seen": 375297676,
      "step": 6698
    },
    {
      "epoch": 14.919821826280623,
      "grad_norm": 17.56962013244629,
      "learning_rate": 1e-06,
      "loss": 0.5139,
      "num_input_tokens_seen": 375353512,
      "step": 6699
    },
    {
      "epoch": 14.919821826280623,
      "loss": 0.5415977239608765,
      "loss_ce": 9.384811710333452e-05,
      "loss_iou": 0.2177734375,
      "loss_num": 0.021240234375,
      "loss_xval": 0.54296875,
      "num_input_tokens_seen": 375353512,
      "step": 6699
    },
    {
      "epoch": 14.922048997772828,
      "grad_norm": 17.380096435546875,
      "learning_rate": 1e-06,
      "loss": 0.329,
      "num_input_tokens_seen": 375408140,
      "step": 6700
    },
    {
      "epoch": 14.922048997772828,
      "loss": 0.3042938709259033,
      "loss_ce": 9.468305506743491e-05,
      "loss_iou": 0.126953125,
      "loss_num": 0.00994873046875,
      "loss_xval": 0.3046875,
      "num_input_tokens_seen": 375408140,
      "step": 6700
    },
    {
      "epoch": 14.924276169265033,
      "grad_norm": 19.394245147705078,
      "learning_rate": 1e-06,
      "loss": 0.3533,
      "num_input_tokens_seen": 375462496,
      "step": 6701
    },
    {
      "epoch": 14.924276169265033,
      "loss": 0.3009084463119507,
      "loss_ce": 9.669142309576273e-05,
      "loss_iou": 0.12890625,
      "loss_num": 0.00836181640625,
      "loss_xval": 0.30078125,
      "num_input_tokens_seen": 375462496,
      "step": 6701
    },
    {
      "epoch": 14.926503340757238,
      "grad_norm": 46.145076751708984,
      "learning_rate": 1e-06,
      "loss": 0.5536,
      "num_input_tokens_seen": 375515212,
      "step": 6702
    },
    {
      "epoch": 14.926503340757238,
      "loss": 0.44150978326797485,
      "loss_ce": 0.00010355835547670722,
      "loss_iou": 0.1826171875,
      "loss_num": 0.01519775390625,
      "loss_xval": 0.44140625,
      "num_input_tokens_seen": 375515212,
      "step": 6702
    },
    {
      "epoch": 14.928730512249443,
      "grad_norm": 17.22191619873047,
      "learning_rate": 1e-06,
      "loss": 0.6082,
      "num_input_tokens_seen": 375573600,
      "step": 6703
    },
    {
      "epoch": 14.928730512249443,
      "loss": 0.49972373247146606,
      "loss_ce": 8.993092342279851e-05,
      "loss_iou": 0.2314453125,
      "loss_num": 0.00738525390625,
      "loss_xval": 0.5,
      "num_input_tokens_seen": 375573600,
      "step": 6703
    },
    {
      "epoch": 14.930957683741648,
      "grad_norm": 17.4095458984375,
      "learning_rate": 1e-06,
      "loss": 0.4707,
      "num_input_tokens_seen": 375627936,
      "step": 6704
    },
    {
      "epoch": 14.930957683741648,
      "loss": 0.35287582874298096,
      "loss_ce": 9.260718070436269e-05,
      "loss_iou": 0.1474609375,
      "loss_num": 0.0115966796875,
      "loss_xval": 0.353515625,
      "num_input_tokens_seen": 375627936,
      "step": 6704
    },
    {
      "epoch": 14.933184855233852,
      "grad_norm": 14.413276672363281,
      "learning_rate": 1e-06,
      "loss": 0.502,
      "num_input_tokens_seen": 375684652,
      "step": 6705
    },
    {
      "epoch": 14.933184855233852,
      "loss": 0.6157550811767578,
      "loss_ce": 9.349231550004333e-05,
      "loss_iou": 0.267578125,
      "loss_num": 0.0162353515625,
      "loss_xval": 0.6171875,
      "num_input_tokens_seen": 375684652,
      "step": 6705
    },
    {
      "epoch": 14.935412026726057,
      "grad_norm": 28.362884521484375,
      "learning_rate": 1e-06,
      "loss": 0.4897,
      "num_input_tokens_seen": 375741700,
      "step": 6706
    },
    {
      "epoch": 14.935412026726057,
      "loss": 0.3510599732398987,
      "loss_ce": 0.00010782700701383874,
      "loss_iou": 0.15625,
      "loss_num": 0.00787353515625,
      "loss_xval": 0.3515625,
      "num_input_tokens_seen": 375741700,
      "step": 6706
    },
    {
      "epoch": 14.937639198218262,
      "grad_norm": 24.491363525390625,
      "learning_rate": 1e-06,
      "loss": 0.4178,
      "num_input_tokens_seen": 375797580,
      "step": 6707
    },
    {
      "epoch": 14.937639198218262,
      "loss": 0.2931881546974182,
      "loss_ce": 9.731938189361244e-05,
      "loss_iou": 0.1298828125,
      "loss_num": 0.006805419921875,
      "loss_xval": 0.29296875,
      "num_input_tokens_seen": 375797580,
      "step": 6707
    },
    {
      "epoch": 14.939866369710467,
      "grad_norm": 18.060317993164062,
      "learning_rate": 1e-06,
      "loss": 0.5446,
      "num_input_tokens_seen": 375851716,
      "step": 6708
    },
    {
      "epoch": 14.939866369710467,
      "loss": 0.40395689010620117,
      "loss_ce": 0.00014827345148660243,
      "loss_iou": 0.185546875,
      "loss_num": 0.006622314453125,
      "loss_xval": 0.404296875,
      "num_input_tokens_seen": 375851716,
      "step": 6708
    },
    {
      "epoch": 14.942093541202672,
      "grad_norm": 29.22158432006836,
      "learning_rate": 1e-06,
      "loss": 0.4788,
      "num_input_tokens_seen": 375907252,
      "step": 6709
    },
    {
      "epoch": 14.942093541202672,
      "loss": 0.43601399660110474,
      "loss_ce": 0.00010091814328916371,
      "loss_iou": 0.19921875,
      "loss_num": 0.007659912109375,
      "loss_xval": 0.435546875,
      "num_input_tokens_seen": 375907252,
      "step": 6709
    },
    {
      "epoch": 14.944320712694877,
      "grad_norm": 23.212482452392578,
      "learning_rate": 1e-06,
      "loss": 0.5383,
      "num_input_tokens_seen": 375962688,
      "step": 6710
    },
    {
      "epoch": 14.944320712694877,
      "loss": 0.6079156398773193,
      "loss_ce": 0.0001275775139220059,
      "loss_iou": 0.263671875,
      "loss_num": 0.0162353515625,
      "loss_xval": 0.609375,
      "num_input_tokens_seen": 375962688,
      "step": 6710
    },
    {
      "epoch": 14.946547884187082,
      "grad_norm": 17.981061935424805,
      "learning_rate": 1e-06,
      "loss": 0.3841,
      "num_input_tokens_seen": 376018616,
      "step": 6711
    },
    {
      "epoch": 14.946547884187082,
      "loss": 0.2717045545578003,
      "loss_ce": 9.8096948931925e-05,
      "loss_iou": 0.11572265625,
      "loss_num": 0.00799560546875,
      "loss_xval": 0.271484375,
      "num_input_tokens_seen": 376018616,
      "step": 6711
    },
    {
      "epoch": 14.948775055679288,
      "grad_norm": 19.030765533447266,
      "learning_rate": 1e-06,
      "loss": 0.3394,
      "num_input_tokens_seen": 376074472,
      "step": 6712
    },
    {
      "epoch": 14.948775055679288,
      "loss": 0.31912869215011597,
      "loss_ce": 9.793009667191654e-05,
      "loss_iou": 0.1484375,
      "loss_num": 0.004241943359375,
      "loss_xval": 0.318359375,
      "num_input_tokens_seen": 376074472,
      "step": 6712
    },
    {
      "epoch": 14.951002227171493,
      "grad_norm": 18.102794647216797,
      "learning_rate": 1e-06,
      "loss": 0.5757,
      "num_input_tokens_seen": 376129896,
      "step": 6713
    },
    {
      "epoch": 14.951002227171493,
      "loss": 0.7676899433135986,
      "loss_ce": 0.00011178333807038143,
      "loss_iou": 0.294921875,
      "loss_num": 0.03515625,
      "loss_xval": 0.765625,
      "num_input_tokens_seen": 376129896,
      "step": 6713
    },
    {
      "epoch": 14.953229398663698,
      "grad_norm": 28.493032455444336,
      "learning_rate": 1e-06,
      "loss": 0.5445,
      "num_input_tokens_seen": 376188248,
      "step": 6714
    },
    {
      "epoch": 14.953229398663698,
      "loss": 0.6533517837524414,
      "loss_ce": 0.00015353792696259916,
      "loss_iou": 0.267578125,
      "loss_num": 0.023681640625,
      "loss_xval": 0.65234375,
      "num_input_tokens_seen": 376188248,
      "step": 6714
    },
    {
      "epoch": 14.955456570155903,
      "grad_norm": 36.27149200439453,
      "learning_rate": 1e-06,
      "loss": 0.5764,
      "num_input_tokens_seen": 376241592,
      "step": 6715
    },
    {
      "epoch": 14.955456570155903,
      "loss": 0.3418195843696594,
      "loss_ce": 0.00014476115757133812,
      "loss_iou": 0.1484375,
      "loss_num": 0.00909423828125,
      "loss_xval": 0.341796875,
      "num_input_tokens_seen": 376241592,
      "step": 6715
    },
    {
      "epoch": 14.957683741648108,
      "grad_norm": 16.67228889465332,
      "learning_rate": 1e-06,
      "loss": 0.4621,
      "num_input_tokens_seen": 376298924,
      "step": 6716
    },
    {
      "epoch": 14.957683741648108,
      "loss": 0.6529318690299988,
      "loss_ce": 9.987308294512331e-05,
      "loss_iou": 0.267578125,
      "loss_num": 0.0233154296875,
      "loss_xval": 0.65234375,
      "num_input_tokens_seen": 376298924,
      "step": 6716
    },
    {
      "epoch": 14.959910913140313,
      "grad_norm": 14.509023666381836,
      "learning_rate": 1e-06,
      "loss": 0.4821,
      "num_input_tokens_seen": 376356404,
      "step": 6717
    },
    {
      "epoch": 14.959910913140313,
      "loss": 0.4666660726070404,
      "loss_ce": 0.00011332825670251623,
      "loss_iou": 0.208984375,
      "loss_num": 0.0096435546875,
      "loss_xval": 0.466796875,
      "num_input_tokens_seen": 376356404,
      "step": 6717
    },
    {
      "epoch": 14.962138084632517,
      "grad_norm": 18.535402297973633,
      "learning_rate": 1e-06,
      "loss": 0.5056,
      "num_input_tokens_seen": 376413404,
      "step": 6718
    },
    {
      "epoch": 14.962138084632517,
      "loss": 0.4525294899940491,
      "loss_ce": 0.0013728843769058585,
      "loss_iou": 0.171875,
      "loss_num": 0.0213623046875,
      "loss_xval": 0.451171875,
      "num_input_tokens_seen": 376413404,
      "step": 6718
    },
    {
      "epoch": 14.964365256124722,
      "grad_norm": 23.715373992919922,
      "learning_rate": 1e-06,
      "loss": 0.3654,
      "num_input_tokens_seen": 376471520,
      "step": 6719
    },
    {
      "epoch": 14.964365256124722,
      "loss": 0.3640136122703552,
      "loss_ce": 0.00012200840865261853,
      "loss_iou": 0.16015625,
      "loss_num": 0.0087890625,
      "loss_xval": 0.36328125,
      "num_input_tokens_seen": 376471520,
      "step": 6719
    },
    {
      "epoch": 14.966592427616927,
      "grad_norm": 22.593597412109375,
      "learning_rate": 1e-06,
      "loss": 0.5501,
      "num_input_tokens_seen": 376527260,
      "step": 6720
    },
    {
      "epoch": 14.966592427616927,
      "loss": 0.5731148719787598,
      "loss_ce": 0.00011685446224873886,
      "loss_iou": 0.2294921875,
      "loss_num": 0.022705078125,
      "loss_xval": 0.57421875,
      "num_input_tokens_seen": 376527260,
      "step": 6720
    },
    {
      "epoch": 14.968819599109132,
      "grad_norm": 14.375116348266602,
      "learning_rate": 1e-06,
      "loss": 0.5558,
      "num_input_tokens_seen": 376585680,
      "step": 6721
    },
    {
      "epoch": 14.968819599109132,
      "loss": 0.45606058835983276,
      "loss_ce": 0.0006162732024677098,
      "loss_iou": 0.1982421875,
      "loss_num": 0.01165771484375,
      "loss_xval": 0.455078125,
      "num_input_tokens_seen": 376585680,
      "step": 6721
    },
    {
      "epoch": 14.971046770601337,
      "grad_norm": 15.256864547729492,
      "learning_rate": 1e-06,
      "loss": 0.4053,
      "num_input_tokens_seen": 376641292,
      "step": 6722
    },
    {
      "epoch": 14.971046770601337,
      "loss": 0.44297972321510315,
      "loss_ce": 0.00010863250645343214,
      "loss_iou": 0.181640625,
      "loss_num": 0.0159912109375,
      "loss_xval": 0.443359375,
      "num_input_tokens_seen": 376641292,
      "step": 6722
    },
    {
      "epoch": 14.973273942093542,
      "grad_norm": 20.12420082092285,
      "learning_rate": 1e-06,
      "loss": 0.5,
      "num_input_tokens_seen": 376696188,
      "step": 6723
    },
    {
      "epoch": 14.973273942093542,
      "loss": 0.42197349667549133,
      "loss_ce": 9.847040200838819e-05,
      "loss_iou": 0.171875,
      "loss_num": 0.0157470703125,
      "loss_xval": 0.421875,
      "num_input_tokens_seen": 376696188,
      "step": 6723
    },
    {
      "epoch": 14.975501113585747,
      "grad_norm": 21.817781448364258,
      "learning_rate": 1e-06,
      "loss": 0.5414,
      "num_input_tokens_seen": 376752988,
      "step": 6724
    },
    {
      "epoch": 14.975501113585747,
      "loss": 0.5460734367370605,
      "loss_ce": 0.00017501445836387575,
      "loss_iou": 0.22265625,
      "loss_num": 0.0205078125,
      "loss_xval": 0.546875,
      "num_input_tokens_seen": 376752988,
      "step": 6724
    },
    {
      "epoch": 14.977728285077951,
      "grad_norm": 16.294130325317383,
      "learning_rate": 1e-06,
      "loss": 0.3279,
      "num_input_tokens_seen": 376810460,
      "step": 6725
    },
    {
      "epoch": 14.977728285077951,
      "loss": 0.28637370467185974,
      "loss_ce": 0.00011881387035828084,
      "loss_iou": 0.109375,
      "loss_num": 0.01348876953125,
      "loss_xval": 0.287109375,
      "num_input_tokens_seen": 376810460,
      "step": 6725
    },
    {
      "epoch": 14.979955456570156,
      "grad_norm": 17.604066848754883,
      "learning_rate": 1e-06,
      "loss": 0.4209,
      "num_input_tokens_seen": 376866340,
      "step": 6726
    },
    {
      "epoch": 14.979955456570156,
      "loss": 0.4271067678928375,
      "loss_ce": 0.00010479907359695062,
      "loss_iou": 0.193359375,
      "loss_num": 0.00823974609375,
      "loss_xval": 0.427734375,
      "num_input_tokens_seen": 376866340,
      "step": 6726
    },
    {
      "epoch": 14.982182628062361,
      "grad_norm": 18.081811904907227,
      "learning_rate": 1e-06,
      "loss": 0.4397,
      "num_input_tokens_seen": 376921200,
      "step": 6727
    },
    {
      "epoch": 14.982182628062361,
      "loss": 0.43722379207611084,
      "loss_ce": 9.000849240692332e-05,
      "loss_iou": 0.1826171875,
      "loss_num": 0.01446533203125,
      "loss_xval": 0.4375,
      "num_input_tokens_seen": 376921200,
      "step": 6727
    },
    {
      "epoch": 14.984409799554566,
      "grad_norm": 18.38675308227539,
      "learning_rate": 1e-06,
      "loss": 0.3883,
      "num_input_tokens_seen": 376979648,
      "step": 6728
    },
    {
      "epoch": 14.984409799554566,
      "loss": 0.36479222774505615,
      "loss_ce": 0.00010718008707044646,
      "loss_iou": 0.146484375,
      "loss_num": 0.0145263671875,
      "loss_xval": 0.365234375,
      "num_input_tokens_seen": 376979648,
      "step": 6728
    },
    {
      "epoch": 14.98663697104677,
      "grad_norm": 24.232728958129883,
      "learning_rate": 1e-06,
      "loss": 0.521,
      "num_input_tokens_seen": 377035696,
      "step": 6729
    },
    {
      "epoch": 14.98663697104677,
      "loss": 0.47929060459136963,
      "loss_ce": 0.00016459994367323816,
      "loss_iou": 0.1923828125,
      "loss_num": 0.01904296875,
      "loss_xval": 0.478515625,
      "num_input_tokens_seen": 377035696,
      "step": 6729
    },
    {
      "epoch": 14.988864142538976,
      "grad_norm": 16.893199920654297,
      "learning_rate": 1e-06,
      "loss": 0.4381,
      "num_input_tokens_seen": 377094444,
      "step": 6730
    },
    {
      "epoch": 14.988864142538976,
      "loss": 0.32279184460639954,
      "loss_ce": 0.00016000941104721278,
      "loss_iou": 0.150390625,
      "loss_num": 0.00457763671875,
      "loss_xval": 0.322265625,
      "num_input_tokens_seen": 377094444,
      "step": 6730
    },
    {
      "epoch": 14.99109131403118,
      "grad_norm": 19.161134719848633,
      "learning_rate": 1e-06,
      "loss": 0.3922,
      "num_input_tokens_seen": 377154056,
      "step": 6731
    },
    {
      "epoch": 14.99109131403118,
      "loss": 0.3056414723396301,
      "loss_ce": 9.948984370566905e-05,
      "loss_iou": 0.140625,
      "loss_num": 0.0048828125,
      "loss_xval": 0.3046875,
      "num_input_tokens_seen": 377154056,
      "step": 6731
    },
    {
      "epoch": 14.993318485523385,
      "grad_norm": 19.205812454223633,
      "learning_rate": 1e-06,
      "loss": 0.3231,
      "num_input_tokens_seen": 377211084,
      "step": 6732
    },
    {
      "epoch": 14.993318485523385,
      "loss": 0.34961456060409546,
      "loss_ce": 0.00012724896077997983,
      "loss_iou": 0.150390625,
      "loss_num": 0.0096435546875,
      "loss_xval": 0.349609375,
      "num_input_tokens_seen": 377211084,
      "step": 6732
    },
    {
      "epoch": 14.99554565701559,
      "grad_norm": 19.172298431396484,
      "learning_rate": 1e-06,
      "loss": 0.5644,
      "num_input_tokens_seen": 377269112,
      "step": 6733
    },
    {
      "epoch": 14.99554565701559,
      "loss": 0.5044995546340942,
      "loss_ce": 0.00010500279313419014,
      "loss_iou": 0.21875,
      "loss_num": 0.0133056640625,
      "loss_xval": 0.50390625,
      "num_input_tokens_seen": 377269112,
      "step": 6733
    },
    {
      "epoch": 14.997772828507795,
      "grad_norm": 18.66171646118164,
      "learning_rate": 1e-06,
      "loss": 0.4576,
      "num_input_tokens_seen": 377325276,
      "step": 6734
    },
    {
      "epoch": 14.997772828507795,
      "loss": 0.29831087589263916,
      "loss_ce": 9.311985195381567e-05,
      "loss_iou": 0.1220703125,
      "loss_num": 0.01080322265625,
      "loss_xval": 0.298828125,
      "num_input_tokens_seen": 377325276,
      "step": 6734
    },
    {
      "epoch": 15.0,
      "grad_norm": 23.2895565032959,
      "learning_rate": 1e-06,
      "loss": 0.3809,
      "num_input_tokens_seen": 377382228,
      "step": 6735
    },
    {
      "epoch": 15.0,
      "loss": 0.4358910620212555,
      "loss_ce": 0.00010004001524066553,
      "loss_iou": 0.1953125,
      "loss_num": 0.0089111328125,
      "loss_xval": 0.435546875,
      "num_input_tokens_seen": 377382228,
      "step": 6735
    },
    {
      "epoch": 15.002227171492205,
      "grad_norm": 15.336739540100098,
      "learning_rate": 1e-06,
      "loss": 0.4672,
      "num_input_tokens_seen": 377438264,
      "step": 6736
    },
    {
      "epoch": 15.002227171492205,
      "loss": 0.25816991925239563,
      "loss_ce": 0.00011327323591103777,
      "loss_iou": 0.11474609375,
      "loss_num": 0.00579833984375,
      "loss_xval": 0.2578125,
      "num_input_tokens_seen": 377438264,
      "step": 6736
    },
    {
      "epoch": 15.00445434298441,
      "grad_norm": 16.66543197631836,
      "learning_rate": 1e-06,
      "loss": 0.5834,
      "num_input_tokens_seen": 377490272,
      "step": 6737
    },
    {
      "epoch": 15.00445434298441,
      "loss": 0.7062318325042725,
      "loss_ce": 0.0001771746901795268,
      "loss_iou": 0.3046875,
      "loss_num": 0.019775390625,
      "loss_xval": 0.70703125,
      "num_input_tokens_seen": 377490272,
      "step": 6737
    },
    {
      "epoch": 15.006681514476615,
      "grad_norm": 16.386180877685547,
      "learning_rate": 1e-06,
      "loss": 0.3941,
      "num_input_tokens_seen": 377547544,
      "step": 6738
    },
    {
      "epoch": 15.006681514476615,
      "loss": 0.4180727005004883,
      "loss_ce": 8.869110752129927e-05,
      "loss_iou": 0.169921875,
      "loss_num": 0.015625,
      "loss_xval": 0.41796875,
      "num_input_tokens_seen": 377547544,
      "step": 6738
    },
    {
      "epoch": 15.00890868596882,
      "grad_norm": 13.663156509399414,
      "learning_rate": 1e-06,
      "loss": 0.3605,
      "num_input_tokens_seen": 377605960,
      "step": 6739
    },
    {
      "epoch": 15.00890868596882,
      "loss": 0.29490476846694946,
      "loss_ce": 0.00010495680180611089,
      "loss_iou": 0.130859375,
      "loss_num": 0.00653076171875,
      "loss_xval": 0.294921875,
      "num_input_tokens_seen": 377605960,
      "step": 6739
    },
    {
      "epoch": 15.011135857461024,
      "grad_norm": 20.333576202392578,
      "learning_rate": 1e-06,
      "loss": 0.3858,
      "num_input_tokens_seen": 377663176,
      "step": 6740
    },
    {
      "epoch": 15.011135857461024,
      "loss": 0.34868597984313965,
      "loss_ce": 0.00011420654482208192,
      "loss_iou": 0.15625,
      "loss_num": 0.00726318359375,
      "loss_xval": 0.34765625,
      "num_input_tokens_seen": 377663176,
      "step": 6740
    },
    {
      "epoch": 15.01336302895323,
      "grad_norm": 13.948286056518555,
      "learning_rate": 1e-06,
      "loss": 0.3959,
      "num_input_tokens_seen": 377717832,
      "step": 6741
    },
    {
      "epoch": 15.01336302895323,
      "loss": 0.42416131496429443,
      "loss_ce": 8.905168215278536e-05,
      "loss_iou": 0.173828125,
      "loss_num": 0.0155029296875,
      "loss_xval": 0.423828125,
      "num_input_tokens_seen": 377717832,
      "step": 6741
    },
    {
      "epoch": 15.015590200445434,
      "grad_norm": 19.52324867248535,
      "learning_rate": 1e-06,
      "loss": 0.3894,
      "num_input_tokens_seen": 377775816,
      "step": 6742
    },
    {
      "epoch": 15.015590200445434,
      "loss": 0.43786442279815674,
      "loss_ce": 0.00012025667092530057,
      "loss_iou": 0.177734375,
      "loss_num": 0.0162353515625,
      "loss_xval": 0.4375,
      "num_input_tokens_seen": 377775816,
      "step": 6742
    },
    {
      "epoch": 15.017817371937639,
      "grad_norm": 26.31039047241211,
      "learning_rate": 1e-06,
      "loss": 0.4081,
      "num_input_tokens_seen": 377834092,
      "step": 6743
    },
    {
      "epoch": 15.017817371937639,
      "loss": 0.4304569363594055,
      "loss_ce": 9.805826266529039e-05,
      "loss_iou": 0.189453125,
      "loss_num": 0.01025390625,
      "loss_xval": 0.4296875,
      "num_input_tokens_seen": 377834092,
      "step": 6743
    },
    {
      "epoch": 15.020044543429844,
      "grad_norm": 13.118270874023438,
      "learning_rate": 1e-06,
      "loss": 0.6506,
      "num_input_tokens_seen": 377890548,
      "step": 6744
    },
    {
      "epoch": 15.020044543429844,
      "loss": 0.8540452718734741,
      "loss_ce": 0.0001634818036109209,
      "loss_iou": 0.349609375,
      "loss_num": 0.03125,
      "loss_xval": 0.85546875,
      "num_input_tokens_seen": 377890548,
      "step": 6744
    },
    {
      "epoch": 15.022271714922049,
      "grad_norm": 16.432491302490234,
      "learning_rate": 1e-06,
      "loss": 0.394,
      "num_input_tokens_seen": 377948476,
      "step": 6745
    },
    {
      "epoch": 15.022271714922049,
      "loss": 0.39914602041244507,
      "loss_ce": 9.820661216508597e-05,
      "loss_iou": 0.181640625,
      "loss_num": 0.00738525390625,
      "loss_xval": 0.3984375,
      "num_input_tokens_seen": 377948476,
      "step": 6745
    },
    {
      "epoch": 15.024498886414253,
      "grad_norm": 17.178516387939453,
      "learning_rate": 1e-06,
      "loss": 0.435,
      "num_input_tokens_seen": 378003020,
      "step": 6746
    },
    {
      "epoch": 15.024498886414253,
      "loss": 0.5235254168510437,
      "loss_ce": 0.0004541404196061194,
      "loss_iou": 0.232421875,
      "loss_num": 0.01171875,
      "loss_xval": 0.5234375,
      "num_input_tokens_seen": 378003020,
      "step": 6746
    },
    {
      "epoch": 15.026726057906458,
      "grad_norm": 28.950069427490234,
      "learning_rate": 1e-06,
      "loss": 0.327,
      "num_input_tokens_seen": 378060880,
      "step": 6747
    },
    {
      "epoch": 15.026726057906458,
      "loss": 0.35339534282684326,
      "loss_ce": 0.0001238558324985206,
      "loss_iou": 0.16015625,
      "loss_num": 0.006622314453125,
      "loss_xval": 0.353515625,
      "num_input_tokens_seen": 378060880,
      "step": 6747
    },
    {
      "epoch": 15.028953229398663,
      "grad_norm": 30.91240882873535,
      "learning_rate": 1e-06,
      "loss": 0.6332,
      "num_input_tokens_seen": 378113388,
      "step": 6748
    },
    {
      "epoch": 15.028953229398663,
      "loss": 0.6445518136024475,
      "loss_ce": 0.0016074487939476967,
      "loss_iou": 0.283203125,
      "loss_num": 0.0150146484375,
      "loss_xval": 0.64453125,
      "num_input_tokens_seen": 378113388,
      "step": 6748
    },
    {
      "epoch": 15.031180400890868,
      "grad_norm": 110.00398254394531,
      "learning_rate": 1e-06,
      "loss": 0.3879,
      "num_input_tokens_seen": 378168524,
      "step": 6749
    },
    {
      "epoch": 15.031180400890868,
      "loss": 0.41330039501190186,
      "loss_ce": 9.242070518666878e-05,
      "loss_iou": 0.1904296875,
      "loss_num": 0.006500244140625,
      "loss_xval": 0.4140625,
      "num_input_tokens_seen": 378168524,
      "step": 6749
    },
    {
      "epoch": 15.033407572383073,
      "grad_norm": 36.06410217285156,
      "learning_rate": 1e-06,
      "loss": 0.6565,
      "num_input_tokens_seen": 378223964,
      "step": 6750
    },
    {
      "epoch": 15.033407572383073,
      "eval_seeclick_web_CIoU": 0.5852765142917633,
      "eval_seeclick_web_GIoU": 0.5835680663585663,
      "eval_seeclick_web_IoU": 0.6037326455116272,
      "eval_seeclick_web_MAE_all": 0.015391120221465826,
      "eval_seeclick_web_MAE_h": 0.007827216759324074,
      "eval_seeclick_web_MAE_w": 0.015470336191356182,
      "eval_seeclick_web_MAE_x_boxes": 0.008828274440020323,
      "eval_seeclick_web_MAE_y_boxes": 0.021092181792482734,
      "eval_seeclick_web_inside_bbox": 0.9166666567325592,
      "eval_seeclick_web_loss": 0.9029772877693176,
      "eval_seeclick_web_loss_ce": 0.00016041052003856748,
      "eval_seeclick_web_loss_iou": 0.4168701171875,
      "eval_seeclick_web_loss_num": 0.012361526489257812,
      "eval_seeclick_web_loss_xval": 0.895263671875,
      "eval_seeclick_web_runtime": 25.9811,
      "eval_seeclick_web_samples_per_second": 1.924,
      "eval_seeclick_web_steps_per_second": 0.077,
      "num_input_tokens_seen": 378223964,
      "step": 6750
    },
    {
      "epoch": 15.033407572383073,
      "eval_icons_CIoU": 0.28028304874897003,
      "eval_icons_GIoU": 0.30072087049484253,
      "eval_icons_IoU": 0.3571899086236954,
      "eval_icons_MAE_all": 0.057889632880687714,
      "eval_icons_MAE_h": 0.03522232733666897,
      "eval_icons_MAE_w": 0.05322147347033024,
      "eval_icons_MAE_x_boxes": 0.05888655222952366,
      "eval_icons_MAE_y_boxes": 0.03719859570264816,
      "eval_icons_inside_bbox": 0.6371527910232544,
      "eval_icons_loss": 1.6711426973342896,
      "eval_icons_loss_ce": 0.00020551962370518595,
      "eval_icons_loss_iou": 0.650146484375,
      "eval_icons_loss_num": 0.04911231994628906,
      "eval_icons_loss_xval": 1.546142578125,
      "eval_icons_runtime": 25.2226,
      "eval_icons_samples_per_second": 1.982,
      "eval_icons_steps_per_second": 0.079,
      "num_input_tokens_seen": 378223964,
      "step": 6750
    },
    {
      "epoch": 15.033407572383073,
      "eval_screenspot_CIoU": 0.37860554456710815,
      "eval_screenspot_GIoU": 0.39325101176897687,
      "eval_screenspot_IoU": 0.4482241968313853,
      "eval_screenspot_MAE_all": 0.05375574777523676,
      "eval_screenspot_MAE_h": 0.03931415639817715,
      "eval_screenspot_MAE_w": 0.05822847535212835,
      "eval_screenspot_MAE_x_boxes": 0.06611844276388486,
      "eval_screenspot_MAE_y_boxes": 0.036580439967413746,
      "eval_screenspot_inside_bbox": 0.7041666706403097,
      "eval_screenspot_loss": 1.5432301759719849,
      "eval_screenspot_loss_ce": 0.00022206837699438134,
      "eval_screenspot_loss_iou": 0.644775390625,
      "eval_screenspot_loss_num": 0.061370849609375,
      "eval_screenspot_loss_xval": 1.5965169270833333,
      "eval_screenspot_runtime": 42.0263,
      "eval_screenspot_samples_per_second": 2.118,
      "eval_screenspot_steps_per_second": 0.071,
      "num_input_tokens_seen": 378223964,
      "step": 6750
    },
    {
      "epoch": 15.033407572383073,
      "eval_compot_CIoU": 0.34727177023887634,
      "eval_compot_GIoU": 0.3595842719078064,
      "eval_compot_IoU": 0.4042260944843292,
      "eval_compot_MAE_all": 0.018605078104883432,
      "eval_compot_MAE_h": 0.011235271580517292,
      "eval_compot_MAE_w": 0.020991048775613308,
      "eval_compot_MAE_x_boxes": 0.030081474222242832,
      "eval_compot_MAE_y_boxes": 0.006614114856347442,
      "eval_compot_inside_bbox": 0.6302083432674408,
      "eval_compot_loss": 1.4067184925079346,
      "eval_compot_loss_ce": 0.0001557795621920377,
      "eval_compot_loss_iou": 0.650146484375,
      "eval_compot_loss_num": 0.0174102783203125,
      "eval_compot_loss_xval": 1.3876953125,
      "eval_compot_runtime": 25.4621,
      "eval_compot_samples_per_second": 1.964,
      "eval_compot_steps_per_second": 0.079,
      "num_input_tokens_seen": 378223964,
      "step": 6750
    },
    {
      "epoch": 15.033407572383073,
      "eval_custom_ui_val_CIoU": 0.48026987661918,
      "eval_custom_ui_val_GIoU": 0.48731139302253723,
      "eval_custom_ui_val_IoU": 0.5409456855720944,
      "eval_custom_ui_val_MAE_all": 0.027578827666325703,
      "eval_custom_ui_val_MAE_h": 0.015306917821160622,
      "eval_custom_ui_val_MAE_w": 0.03556989893938104,
      "eval_custom_ui_val_MAE_x_boxes": 0.033483781981178455,
      "eval_custom_ui_val_MAE_y_boxes": 0.013460160000249743,
      "eval_custom_ui_val_inside_bbox": 0.7719907429483202,
      "eval_custom_ui_val_loss": 1.1606297492980957,
      "eval_custom_ui_val_loss_ce": 0.0001816005743522611,
      "eval_custom_ui_val_loss_iou": 0.4960530598958333,
      "eval_custom_ui_val_loss_num": 0.02418337927924262,
      "eval_custom_ui_val_loss_xval": 1.1128472222222223,
      "eval_custom_ui_val_runtime": 78.0431,
      "eval_custom_ui_val_samples_per_second": 3.396,
      "eval_custom_ui_val_steps_per_second": 0.115,
      "num_input_tokens_seen": 378223964,
      "step": 6750
    },
    {
      "epoch": 15.033407572383073,
      "loss": 0.8121492862701416,
      "loss_ce": 0.00013749845675192773,
      "loss_iou": 0.365234375,
      "loss_num": 0.0166015625,
      "loss_xval": 0.8125,
      "num_input_tokens_seen": 378223964,
      "step": 6750
    },
    {
      "epoch": 15.035634743875278,
      "grad_norm": 23.943416595458984,
      "learning_rate": 1e-06,
      "loss": 0.3742,
      "num_input_tokens_seen": 378281124,
      "step": 6751
    },
    {
      "epoch": 15.035634743875278,
      "loss": 0.33616477251052856,
      "loss_ce": 0.00010521589138079435,
      "loss_iou": 0.1474609375,
      "loss_num": 0.00811767578125,
      "loss_xval": 0.3359375,
      "num_input_tokens_seen": 378281124,
      "step": 6751
    },
    {
      "epoch": 15.037861915367483,
      "grad_norm": 16.794620513916016,
      "learning_rate": 1e-06,
      "loss": 0.5123,
      "num_input_tokens_seen": 378335736,
      "step": 6752
    },
    {
      "epoch": 15.037861915367483,
      "loss": 0.42880141735076904,
      "loss_ce": 9.046800551004708e-05,
      "loss_iou": 0.1923828125,
      "loss_num": 0.0089111328125,
      "loss_xval": 0.4296875,
      "num_input_tokens_seen": 378335736,
      "step": 6752
    },
    {
      "epoch": 15.040089086859687,
      "grad_norm": 23.620935440063477,
      "learning_rate": 1e-06,
      "loss": 0.6279,
      "num_input_tokens_seen": 378391368,
      "step": 6753
    },
    {
      "epoch": 15.040089086859687,
      "loss": 0.624180793762207,
      "loss_ce": 0.00015735081979073584,
      "loss_iou": 0.279296875,
      "loss_num": 0.01312255859375,
      "loss_xval": 0.625,
      "num_input_tokens_seen": 378391368,
      "step": 6753
    },
    {
      "epoch": 15.042316258351892,
      "grad_norm": 16.139259338378906,
      "learning_rate": 1e-06,
      "loss": 0.4466,
      "num_input_tokens_seen": 378446784,
      "step": 6754
    },
    {
      "epoch": 15.042316258351892,
      "loss": 0.5067760348320007,
      "loss_ce": 9.269756264984608e-05,
      "loss_iou": 0.20703125,
      "loss_num": 0.018310546875,
      "loss_xval": 0.5078125,
      "num_input_tokens_seen": 378446784,
      "step": 6754
    },
    {
      "epoch": 15.044543429844097,
      "grad_norm": 17.31475257873535,
      "learning_rate": 1e-06,
      "loss": 0.3974,
      "num_input_tokens_seen": 378504372,
      "step": 6755
    },
    {
      "epoch": 15.044543429844097,
      "loss": 0.44474291801452637,
      "loss_ce": 0.00022385823831427842,
      "loss_iou": 0.177734375,
      "loss_num": 0.0177001953125,
      "loss_xval": 0.4453125,
      "num_input_tokens_seen": 378504372,
      "step": 6755
    },
    {
      "epoch": 15.046770601336302,
      "grad_norm": 22.26161766052246,
      "learning_rate": 1e-06,
      "loss": 0.6522,
      "num_input_tokens_seen": 378560752,
      "step": 6756
    },
    {
      "epoch": 15.046770601336302,
      "loss": 0.6529574990272522,
      "loss_ce": 0.00036961075966246426,
      "loss_iou": 0.28515625,
      "loss_num": 0.0162353515625,
      "loss_xval": 0.65234375,
      "num_input_tokens_seen": 378560752,
      "step": 6756
    },
    {
      "epoch": 15.048997772828507,
      "grad_norm": 18.25050163269043,
      "learning_rate": 1e-06,
      "loss": 0.5268,
      "num_input_tokens_seen": 378619176,
      "step": 6757
    },
    {
      "epoch": 15.048997772828507,
      "loss": 0.43017441034317017,
      "loss_ce": 0.00012070426600985229,
      "loss_iou": 0.18359375,
      "loss_num": 0.0125732421875,
      "loss_xval": 0.4296875,
      "num_input_tokens_seen": 378619176,
      "step": 6757
    },
    {
      "epoch": 15.051224944320714,
      "grad_norm": 16.298351287841797,
      "learning_rate": 1e-06,
      "loss": 0.3852,
      "num_input_tokens_seen": 378673764,
      "step": 6758
    },
    {
      "epoch": 15.051224944320714,
      "loss": 0.3430963158607483,
      "loss_ce": 0.00012450873327907175,
      "loss_iou": 0.1376953125,
      "loss_num": 0.0137939453125,
      "loss_xval": 0.34375,
      "num_input_tokens_seen": 378673764,
      "step": 6758
    },
    {
      "epoch": 15.053452115812918,
      "grad_norm": 15.573384284973145,
      "learning_rate": 1e-06,
      "loss": 0.3511,
      "num_input_tokens_seen": 378729620,
      "step": 6759
    },
    {
      "epoch": 15.053452115812918,
      "loss": 0.36082327365875244,
      "loss_ce": 0.00010551903687883168,
      "loss_iou": 0.166015625,
      "loss_num": 0.00579833984375,
      "loss_xval": 0.361328125,
      "num_input_tokens_seen": 378729620,
      "step": 6759
    },
    {
      "epoch": 15.055679287305123,
      "grad_norm": 16.886507034301758,
      "learning_rate": 1e-06,
      "loss": 0.5778,
      "num_input_tokens_seen": 378784440,
      "step": 6760
    },
    {
      "epoch": 15.055679287305123,
      "loss": 0.5877401828765869,
      "loss_ce": 9.368160681333393e-05,
      "loss_iou": 0.2275390625,
      "loss_num": 0.0264892578125,
      "loss_xval": 0.5859375,
      "num_input_tokens_seen": 378784440,
      "step": 6760
    },
    {
      "epoch": 15.057906458797328,
      "grad_norm": 18.894807815551758,
      "learning_rate": 1e-06,
      "loss": 0.4008,
      "num_input_tokens_seen": 378839964,
      "step": 6761
    },
    {
      "epoch": 15.057906458797328,
      "loss": 0.3778371512889862,
      "loss_ce": 9.055679402081296e-05,
      "loss_iou": 0.169921875,
      "loss_num": 0.007598876953125,
      "loss_xval": 0.376953125,
      "num_input_tokens_seen": 378839964,
      "step": 6761
    },
    {
      "epoch": 15.060133630289533,
      "grad_norm": 22.449024200439453,
      "learning_rate": 1e-06,
      "loss": 0.5987,
      "num_input_tokens_seen": 378897408,
      "step": 6762
    },
    {
      "epoch": 15.060133630289533,
      "loss": 0.6189790368080139,
      "loss_ce": 8.254876593127847e-05,
      "loss_iou": 0.255859375,
      "loss_num": 0.0216064453125,
      "loss_xval": 0.6171875,
      "num_input_tokens_seen": 378897408,
      "step": 6762
    },
    {
      "epoch": 15.062360801781738,
      "grad_norm": 21.661102294921875,
      "learning_rate": 1e-06,
      "loss": 0.4368,
      "num_input_tokens_seen": 378951352,
      "step": 6763
    },
    {
      "epoch": 15.062360801781738,
      "loss": 0.39707666635513306,
      "loss_ce": 0.00010398250014986843,
      "loss_iou": 0.1826171875,
      "loss_num": 0.006439208984375,
      "loss_xval": 0.396484375,
      "num_input_tokens_seen": 378951352,
      "step": 6763
    },
    {
      "epoch": 15.064587973273943,
      "grad_norm": 14.383728981018066,
      "learning_rate": 1e-06,
      "loss": 0.5996,
      "num_input_tokens_seen": 379007704,
      "step": 6764
    },
    {
      "epoch": 15.064587973273943,
      "loss": 0.6301254034042358,
      "loss_ce": 0.00012049202632624656,
      "loss_iou": 0.265625,
      "loss_num": 0.0194091796875,
      "loss_xval": 0.62890625,
      "num_input_tokens_seen": 379007704,
      "step": 6764
    },
    {
      "epoch": 15.066815144766148,
      "grad_norm": 16.69063377380371,
      "learning_rate": 1e-06,
      "loss": 0.5299,
      "num_input_tokens_seen": 379060260,
      "step": 6765
    },
    {
      "epoch": 15.066815144766148,
      "loss": 0.49555858969688416,
      "loss_ce": 0.00019726283790078014,
      "loss_iou": 0.1953125,
      "loss_num": 0.0211181640625,
      "loss_xval": 0.49609375,
      "num_input_tokens_seen": 379060260,
      "step": 6765
    },
    {
      "epoch": 15.069042316258352,
      "grad_norm": 18.389291763305664,
      "learning_rate": 1e-06,
      "loss": 0.6429,
      "num_input_tokens_seen": 379115936,
      "step": 6766
    },
    {
      "epoch": 15.069042316258352,
      "loss": 0.9386324882507324,
      "loss_ce": 0.00015596779121551663,
      "loss_iou": 0.41015625,
      "loss_num": 0.0240478515625,
      "loss_xval": 0.9375,
      "num_input_tokens_seen": 379115936,
      "step": 6766
    },
    {
      "epoch": 15.071269487750557,
      "grad_norm": 33.009193420410156,
      "learning_rate": 1e-06,
      "loss": 0.4666,
      "num_input_tokens_seen": 379171576,
      "step": 6767
    },
    {
      "epoch": 15.071269487750557,
      "loss": 0.5097314119338989,
      "loss_ce": 8.78519203979522e-05,
      "loss_iou": 0.2275390625,
      "loss_num": 0.0108642578125,
      "loss_xval": 0.5078125,
      "num_input_tokens_seen": 379171576,
      "step": 6767
    },
    {
      "epoch": 15.073496659242762,
      "grad_norm": 20.735618591308594,
      "learning_rate": 1e-06,
      "loss": 0.3335,
      "num_input_tokens_seen": 379231148,
      "step": 6768
    },
    {
      "epoch": 15.073496659242762,
      "loss": 0.4110013246536255,
      "loss_ce": 0.00011265121429460123,
      "loss_iou": 0.1962890625,
      "loss_num": 0.0038299560546875,
      "loss_xval": 0.41015625,
      "num_input_tokens_seen": 379231148,
      "step": 6768
    },
    {
      "epoch": 15.075723830734967,
      "grad_norm": 13.399028778076172,
      "learning_rate": 1e-06,
      "loss": 0.4072,
      "num_input_tokens_seen": 379288424,
      "step": 6769
    },
    {
      "epoch": 15.075723830734967,
      "loss": 0.2551080882549286,
      "loss_ce": 0.00010320404544472694,
      "loss_iou": 0.11474609375,
      "loss_num": 0.00506591796875,
      "loss_xval": 0.255859375,
      "num_input_tokens_seen": 379288424,
      "step": 6769
    },
    {
      "epoch": 15.077951002227172,
      "grad_norm": 15.024955749511719,
      "learning_rate": 1e-06,
      "loss": 0.5097,
      "num_input_tokens_seen": 379341648,
      "step": 6770
    },
    {
      "epoch": 15.077951002227172,
      "loss": 0.5870286226272583,
      "loss_ce": 0.00011455308413133025,
      "loss_iou": 0.25,
      "loss_num": 0.0172119140625,
      "loss_xval": 0.5859375,
      "num_input_tokens_seen": 379341648,
      "step": 6770
    },
    {
      "epoch": 15.080178173719377,
      "grad_norm": 57.792747497558594,
      "learning_rate": 1e-06,
      "loss": 0.5396,
      "num_input_tokens_seen": 379396880,
      "step": 6771
    },
    {
      "epoch": 15.080178173719377,
      "loss": 0.6382882595062256,
      "loss_ce": 0.00010471623681951314,
      "loss_iou": 0.251953125,
      "loss_num": 0.02685546875,
      "loss_xval": 0.63671875,
      "num_input_tokens_seen": 379396880,
      "step": 6771
    },
    {
      "epoch": 15.082405345211582,
      "grad_norm": 17.591594696044922,
      "learning_rate": 1e-06,
      "loss": 0.4073,
      "num_input_tokens_seen": 379451464,
      "step": 6772
    },
    {
      "epoch": 15.082405345211582,
      "loss": 0.41464707255363464,
      "loss_ce": 9.628412954043597e-05,
      "loss_iou": 0.1884765625,
      "loss_num": 0.00762939453125,
      "loss_xval": 0.4140625,
      "num_input_tokens_seen": 379451464,
      "step": 6772
    },
    {
      "epoch": 15.084632516703786,
      "grad_norm": 19.806339263916016,
      "learning_rate": 1e-06,
      "loss": 0.3308,
      "num_input_tokens_seen": 379507000,
      "step": 6773
    },
    {
      "epoch": 15.084632516703786,
      "loss": 0.32612937688827515,
      "loss_ce": 7.959181675687432e-05,
      "loss_iou": 0.12890625,
      "loss_num": 0.01348876953125,
      "loss_xval": 0.326171875,
      "num_input_tokens_seen": 379507000,
      "step": 6773
    },
    {
      "epoch": 15.086859688195991,
      "grad_norm": 17.055814743041992,
      "learning_rate": 1e-06,
      "loss": 0.3506,
      "num_input_tokens_seen": 379562540,
      "step": 6774
    },
    {
      "epoch": 15.086859688195991,
      "loss": 0.31467652320861816,
      "loss_ce": 0.00010132987517863512,
      "loss_iou": 0.13671875,
      "loss_num": 0.00799560546875,
      "loss_xval": 0.314453125,
      "num_input_tokens_seen": 379562540,
      "step": 6774
    },
    {
      "epoch": 15.089086859688196,
      "grad_norm": 24.606420516967773,
      "learning_rate": 1e-06,
      "loss": 0.401,
      "num_input_tokens_seen": 379620760,
      "step": 6775
    },
    {
      "epoch": 15.089086859688196,
      "loss": 0.3719760477542877,
      "loss_ce": 8.884338603820652e-05,
      "loss_iou": 0.169921875,
      "loss_num": 0.006256103515625,
      "loss_xval": 0.37109375,
      "num_input_tokens_seen": 379620760,
      "step": 6775
    },
    {
      "epoch": 15.091314031180401,
      "grad_norm": 20.017690658569336,
      "learning_rate": 1e-06,
      "loss": 0.4856,
      "num_input_tokens_seen": 379677336,
      "step": 6776
    },
    {
      "epoch": 15.091314031180401,
      "loss": 0.45812952518463135,
      "loss_ce": 0.00012173528375569731,
      "loss_iou": 0.18359375,
      "loss_num": 0.018310546875,
      "loss_xval": 0.45703125,
      "num_input_tokens_seen": 379677336,
      "step": 6776
    },
    {
      "epoch": 15.093541202672606,
      "grad_norm": 16.497709274291992,
      "learning_rate": 1e-06,
      "loss": 0.4482,
      "num_input_tokens_seen": 379734720,
      "step": 6777
    },
    {
      "epoch": 15.093541202672606,
      "loss": 0.34024685621261597,
      "loss_ce": 9.794162178877741e-05,
      "loss_iou": 0.15234375,
      "loss_num": 0.007080078125,
      "loss_xval": 0.33984375,
      "num_input_tokens_seen": 379734720,
      "step": 6777
    },
    {
      "epoch": 15.09576837416481,
      "grad_norm": 15.530109405517578,
      "learning_rate": 1e-06,
      "loss": 0.4104,
      "num_input_tokens_seen": 379790532,
      "step": 6778
    },
    {
      "epoch": 15.09576837416481,
      "loss": 0.30709517002105713,
      "loss_ce": 8.832212188281119e-05,
      "loss_iou": 0.13671875,
      "loss_num": 0.006805419921875,
      "loss_xval": 0.306640625,
      "num_input_tokens_seen": 379790532,
      "step": 6778
    },
    {
      "epoch": 15.097995545657016,
      "grad_norm": 16.189306259155273,
      "learning_rate": 1e-06,
      "loss": 0.4835,
      "num_input_tokens_seen": 379847060,
      "step": 6779
    },
    {
      "epoch": 15.097995545657016,
      "loss": 0.5428781509399414,
      "loss_ce": 0.0001535568735562265,
      "loss_iou": 0.224609375,
      "loss_num": 0.0186767578125,
      "loss_xval": 0.54296875,
      "num_input_tokens_seen": 379847060,
      "step": 6779
    },
    {
      "epoch": 15.10022271714922,
      "grad_norm": 15.896251678466797,
      "learning_rate": 1e-06,
      "loss": 0.4458,
      "num_input_tokens_seen": 379902864,
      "step": 6780
    },
    {
      "epoch": 15.10022271714922,
      "loss": 0.38522452116012573,
      "loss_ce": 9.271127782994881e-05,
      "loss_iou": 0.1748046875,
      "loss_num": 0.0069580078125,
      "loss_xval": 0.384765625,
      "num_input_tokens_seen": 379902864,
      "step": 6780
    },
    {
      "epoch": 15.102449888641425,
      "grad_norm": 52.78928756713867,
      "learning_rate": 1e-06,
      "loss": 0.3965,
      "num_input_tokens_seen": 379959096,
      "step": 6781
    },
    {
      "epoch": 15.102449888641425,
      "loss": 0.4485187530517578,
      "loss_ce": 0.0002765837707556784,
      "loss_iou": 0.19140625,
      "loss_num": 0.01318359375,
      "loss_xval": 0.44921875,
      "num_input_tokens_seen": 379959096,
      "step": 6781
    },
    {
      "epoch": 15.10467706013363,
      "grad_norm": 25.405174255371094,
      "learning_rate": 1e-06,
      "loss": 0.362,
      "num_input_tokens_seen": 380015168,
      "step": 6782
    },
    {
      "epoch": 15.10467706013363,
      "loss": 0.42562615871429443,
      "loss_ce": 8.907825394999236e-05,
      "loss_iou": 0.185546875,
      "loss_num": 0.01104736328125,
      "loss_xval": 0.42578125,
      "num_input_tokens_seen": 380015168,
      "step": 6782
    },
    {
      "epoch": 15.106904231625835,
      "grad_norm": 11.471501350402832,
      "learning_rate": 1e-06,
      "loss": 0.4449,
      "num_input_tokens_seen": 380071884,
      "step": 6783
    },
    {
      "epoch": 15.106904231625835,
      "loss": 0.4200718402862549,
      "loss_ce": 9.654099267208949e-05,
      "loss_iou": 0.1953125,
      "loss_num": 0.00592041015625,
      "loss_xval": 0.419921875,
      "num_input_tokens_seen": 380071884,
      "step": 6783
    },
    {
      "epoch": 15.10913140311804,
      "grad_norm": 15.025222778320312,
      "learning_rate": 1e-06,
      "loss": 0.4441,
      "num_input_tokens_seen": 380129560,
      "step": 6784
    },
    {
      "epoch": 15.10913140311804,
      "loss": 0.4893530011177063,
      "loss_ce": 9.517707803752273e-05,
      "loss_iou": 0.203125,
      "loss_num": 0.0166015625,
      "loss_xval": 0.48828125,
      "num_input_tokens_seen": 380129560,
      "step": 6784
    },
    {
      "epoch": 15.111358574610245,
      "grad_norm": 29.484661102294922,
      "learning_rate": 1e-06,
      "loss": 0.4307,
      "num_input_tokens_seen": 380185012,
      "step": 6785
    },
    {
      "epoch": 15.111358574610245,
      "loss": 0.314787894487381,
      "loss_ce": 9.060885349754244e-05,
      "loss_iou": 0.140625,
      "loss_num": 0.00677490234375,
      "loss_xval": 0.314453125,
      "num_input_tokens_seen": 380185012,
      "step": 6785
    },
    {
      "epoch": 15.11358574610245,
      "grad_norm": 14.522575378417969,
      "learning_rate": 1e-06,
      "loss": 0.5448,
      "num_input_tokens_seen": 380241104,
      "step": 6786
    },
    {
      "epoch": 15.11358574610245,
      "loss": 0.576589822769165,
      "loss_ce": 0.00011273389827692881,
      "loss_iou": 0.2490234375,
      "loss_num": 0.015625,
      "loss_xval": 0.578125,
      "num_input_tokens_seen": 380241104,
      "step": 6786
    },
    {
      "epoch": 15.115812917594655,
      "grad_norm": 13.87673568725586,
      "learning_rate": 1e-06,
      "loss": 0.5374,
      "num_input_tokens_seen": 380297004,
      "step": 6787
    },
    {
      "epoch": 15.115812917594655,
      "loss": 0.30384352803230286,
      "loss_ce": 0.0001325808116234839,
      "loss_iou": 0.1376953125,
      "loss_num": 0.005615234375,
      "loss_xval": 0.3046875,
      "num_input_tokens_seen": 380297004,
      "step": 6787
    },
    {
      "epoch": 15.11804008908686,
      "grad_norm": 23.66053581237793,
      "learning_rate": 1e-06,
      "loss": 0.5607,
      "num_input_tokens_seen": 380352836,
      "step": 6788
    },
    {
      "epoch": 15.11804008908686,
      "loss": 0.36867833137512207,
      "loss_ce": 8.703065395820886e-05,
      "loss_iou": 0.166015625,
      "loss_num": 0.00738525390625,
      "loss_xval": 0.369140625,
      "num_input_tokens_seen": 380352836,
      "step": 6788
    },
    {
      "epoch": 15.120267260579064,
      "grad_norm": 17.455827713012695,
      "learning_rate": 1e-06,
      "loss": 0.3954,
      "num_input_tokens_seen": 380406124,
      "step": 6789
    },
    {
      "epoch": 15.120267260579064,
      "loss": 0.3701331913471222,
      "loss_ce": 7.704219751758501e-05,
      "loss_iou": 0.15234375,
      "loss_num": 0.01312255859375,
      "loss_xval": 0.369140625,
      "num_input_tokens_seen": 380406124,
      "step": 6789
    },
    {
      "epoch": 15.122494432071269,
      "grad_norm": 33.90388870239258,
      "learning_rate": 1e-06,
      "loss": 0.5768,
      "num_input_tokens_seen": 380459684,
      "step": 6790
    },
    {
      "epoch": 15.122494432071269,
      "loss": 0.4020087718963623,
      "loss_ce": 9.228185808751732e-05,
      "loss_iou": 0.166015625,
      "loss_num": 0.01416015625,
      "loss_xval": 0.40234375,
      "num_input_tokens_seen": 380459684,
      "step": 6790
    },
    {
      "epoch": 15.124721603563474,
      "grad_norm": 16.799598693847656,
      "learning_rate": 1e-06,
      "loss": 0.4023,
      "num_input_tokens_seen": 380516820,
      "step": 6791
    },
    {
      "epoch": 15.124721603563474,
      "loss": 0.44541066884994507,
      "loss_ce": 9.819894330576062e-05,
      "loss_iou": 0.1943359375,
      "loss_num": 0.0113525390625,
      "loss_xval": 0.4453125,
      "num_input_tokens_seen": 380516820,
      "step": 6791
    },
    {
      "epoch": 15.126948775055679,
      "grad_norm": 15.393778800964355,
      "learning_rate": 1e-06,
      "loss": 0.6106,
      "num_input_tokens_seen": 380575020,
      "step": 6792
    },
    {
      "epoch": 15.126948775055679,
      "loss": 0.702007532119751,
      "loss_ce": 0.00010328319331165403,
      "loss_iou": 0.2890625,
      "loss_num": 0.0250244140625,
      "loss_xval": 0.703125,
      "num_input_tokens_seen": 380575020,
      "step": 6792
    },
    {
      "epoch": 15.129175946547884,
      "grad_norm": 12.515711784362793,
      "learning_rate": 1e-06,
      "loss": 0.6247,
      "num_input_tokens_seen": 380631440,
      "step": 6793
    },
    {
      "epoch": 15.129175946547884,
      "loss": 0.8331481218338013,
      "loss_ce": 0.00014029248268343508,
      "loss_iou": 0.31640625,
      "loss_num": 0.0400390625,
      "loss_xval": 0.83203125,
      "num_input_tokens_seen": 380631440,
      "step": 6793
    },
    {
      "epoch": 15.131403118040089,
      "grad_norm": 14.25368595123291,
      "learning_rate": 1e-06,
      "loss": 0.2824,
      "num_input_tokens_seen": 380690028,
      "step": 6794
    },
    {
      "epoch": 15.131403118040089,
      "loss": 0.3566052317619324,
      "loss_ce": 0.00015993951819837093,
      "loss_iou": 0.1552734375,
      "loss_num": 0.00927734375,
      "loss_xval": 0.35546875,
      "num_input_tokens_seen": 380690028,
      "step": 6794
    },
    {
      "epoch": 15.133630289532293,
      "grad_norm": 12.242840766906738,
      "learning_rate": 1e-06,
      "loss": 0.4051,
      "num_input_tokens_seen": 380747684,
      "step": 6795
    },
    {
      "epoch": 15.133630289532293,
      "loss": 0.33505433797836304,
      "loss_ce": 9.341866825707257e-05,
      "loss_iou": 0.158203125,
      "loss_num": 0.00372314453125,
      "loss_xval": 0.3359375,
      "num_input_tokens_seen": 380747684,
      "step": 6795
    },
    {
      "epoch": 15.135857461024498,
      "grad_norm": 15.275101661682129,
      "learning_rate": 1e-06,
      "loss": 0.3874,
      "num_input_tokens_seen": 380803156,
      "step": 6796
    },
    {
      "epoch": 15.135857461024498,
      "loss": 0.4023858308792114,
      "loss_ce": 0.0001641570997890085,
      "loss_iou": 0.17578125,
      "loss_num": 0.0103759765625,
      "loss_xval": 0.40234375,
      "num_input_tokens_seen": 380803156,
      "step": 6796
    },
    {
      "epoch": 15.138084632516703,
      "grad_norm": 16.61083221435547,
      "learning_rate": 1e-06,
      "loss": 0.5824,
      "num_input_tokens_seen": 380860000,
      "step": 6797
    },
    {
      "epoch": 15.138084632516703,
      "loss": 0.5763627290725708,
      "loss_ce": 0.00012979336315765977,
      "loss_iou": 0.251953125,
      "loss_num": 0.014404296875,
      "loss_xval": 0.578125,
      "num_input_tokens_seen": 380860000,
      "step": 6797
    },
    {
      "epoch": 15.140311804008908,
      "grad_norm": 17.85312271118164,
      "learning_rate": 1e-06,
      "loss": 0.3495,
      "num_input_tokens_seen": 380914548,
      "step": 6798
    },
    {
      "epoch": 15.140311804008908,
      "loss": 0.2694869935512543,
      "loss_ce": 7.780264422763139e-05,
      "loss_iou": 0.11669921875,
      "loss_num": 0.00714111328125,
      "loss_xval": 0.26953125,
      "num_input_tokens_seen": 380914548,
      "step": 6798
    },
    {
      "epoch": 15.142538975501113,
      "grad_norm": 26.615489959716797,
      "learning_rate": 1e-06,
      "loss": 0.4413,
      "num_input_tokens_seen": 380972508,
      "step": 6799
    },
    {
      "epoch": 15.142538975501113,
      "loss": 0.4014506936073303,
      "loss_ce": 8.351253200089559e-05,
      "loss_iou": 0.162109375,
      "loss_num": 0.01531982421875,
      "loss_xval": 0.40234375,
      "num_input_tokens_seen": 380972508,
      "step": 6799
    },
    {
      "epoch": 15.144766146993318,
      "grad_norm": 15.07624626159668,
      "learning_rate": 1e-06,
      "loss": 0.3529,
      "num_input_tokens_seen": 381027844,
      "step": 6800
    },
    {
      "epoch": 15.144766146993318,
      "loss": 0.41293323040008545,
      "loss_ce": 9.14369520614855e-05,
      "loss_iou": 0.1875,
      "loss_num": 0.00775146484375,
      "loss_xval": 0.412109375,
      "num_input_tokens_seen": 381027844,
      "step": 6800
    },
    {
      "epoch": 15.146993318485523,
      "grad_norm": 18.466833114624023,
      "learning_rate": 1e-06,
      "loss": 0.489,
      "num_input_tokens_seen": 381084744,
      "step": 6801
    },
    {
      "epoch": 15.146993318485523,
      "loss": 0.37544137239456177,
      "loss_ce": 7.517053745687008e-05,
      "loss_iou": 0.1689453125,
      "loss_num": 0.007598876953125,
      "loss_xval": 0.375,
      "num_input_tokens_seen": 381084744,
      "step": 6801
    },
    {
      "epoch": 15.14922048997773,
      "grad_norm": 18.854787826538086,
      "learning_rate": 1e-06,
      "loss": 0.3958,
      "num_input_tokens_seen": 381139260,
      "step": 6802
    },
    {
      "epoch": 15.14922048997773,
      "loss": 0.386452853679657,
      "loss_ce": 0.00010031522833742201,
      "loss_iou": 0.1689453125,
      "loss_num": 0.009521484375,
      "loss_xval": 0.38671875,
      "num_input_tokens_seen": 381139260,
      "step": 6802
    },
    {
      "epoch": 15.151447661469934,
      "grad_norm": 28.44632339477539,
      "learning_rate": 1e-06,
      "loss": 0.3858,
      "num_input_tokens_seen": 381197604,
      "step": 6803
    },
    {
      "epoch": 15.151447661469934,
      "loss": 0.4399263262748718,
      "loss_ce": 0.0002290659467689693,
      "loss_iou": 0.1962890625,
      "loss_num": 0.0096435546875,
      "loss_xval": 0.439453125,
      "num_input_tokens_seen": 381197604,
      "step": 6803
    },
    {
      "epoch": 15.153674832962139,
      "grad_norm": 36.481109619140625,
      "learning_rate": 1e-06,
      "loss": 0.4761,
      "num_input_tokens_seen": 381253820,
      "step": 6804
    },
    {
      "epoch": 15.153674832962139,
      "loss": 0.38608303666114807,
      "loss_ce": 9.673433669377118e-05,
      "loss_iou": 0.1689453125,
      "loss_num": 0.00982666015625,
      "loss_xval": 0.38671875,
      "num_input_tokens_seen": 381253820,
      "step": 6804
    },
    {
      "epoch": 15.155902004454344,
      "grad_norm": 98.84033203125,
      "learning_rate": 1e-06,
      "loss": 0.3979,
      "num_input_tokens_seen": 381309320,
      "step": 6805
    },
    {
      "epoch": 15.155902004454344,
      "loss": 0.35694801807403564,
      "loss_ce": 0.00016701644926797599,
      "loss_iou": 0.1552734375,
      "loss_num": 0.00921630859375,
      "loss_xval": 0.357421875,
      "num_input_tokens_seen": 381309320,
      "step": 6805
    },
    {
      "epoch": 15.158129175946549,
      "grad_norm": 18.540637969970703,
      "learning_rate": 1e-06,
      "loss": 0.3822,
      "num_input_tokens_seen": 381365092,
      "step": 6806
    },
    {
      "epoch": 15.158129175946549,
      "loss": 0.40906059741973877,
      "loss_ce": 0.00012505166523624212,
      "loss_iou": 0.181640625,
      "loss_num": 0.00909423828125,
      "loss_xval": 0.408203125,
      "num_input_tokens_seen": 381365092,
      "step": 6806
    },
    {
      "epoch": 15.160356347438753,
      "grad_norm": 25.038162231445312,
      "learning_rate": 1e-06,
      "loss": 0.6632,
      "num_input_tokens_seen": 381421476,
      "step": 6807
    },
    {
      "epoch": 15.160356347438753,
      "loss": 0.6734679937362671,
      "loss_ce": 0.00012814889487344772,
      "loss_iou": 0.30078125,
      "loss_num": 0.014404296875,
      "loss_xval": 0.671875,
      "num_input_tokens_seen": 381421476,
      "step": 6807
    },
    {
      "epoch": 15.162583518930958,
      "grad_norm": 19.82026481628418,
      "learning_rate": 1e-06,
      "loss": 0.538,
      "num_input_tokens_seen": 381479976,
      "step": 6808
    },
    {
      "epoch": 15.162583518930958,
      "loss": 0.40109795331954956,
      "loss_ce": 9.697194036561996e-05,
      "loss_iou": 0.1875,
      "loss_num": 0.0052490234375,
      "loss_xval": 0.400390625,
      "num_input_tokens_seen": 381479976,
      "step": 6808
    },
    {
      "epoch": 15.164810690423163,
      "grad_norm": 23.628137588500977,
      "learning_rate": 1e-06,
      "loss": 0.448,
      "num_input_tokens_seen": 381534776,
      "step": 6809
    },
    {
      "epoch": 15.164810690423163,
      "loss": 0.5230967402458191,
      "loss_ce": 0.0001475035387557,
      "loss_iou": 0.216796875,
      "loss_num": 0.01806640625,
      "loss_xval": 0.5234375,
      "num_input_tokens_seen": 381534776,
      "step": 6809
    },
    {
      "epoch": 15.167037861915368,
      "grad_norm": 21.607646942138672,
      "learning_rate": 1e-06,
      "loss": 0.504,
      "num_input_tokens_seen": 381592192,
      "step": 6810
    },
    {
      "epoch": 15.167037861915368,
      "loss": 0.5715093612670898,
      "loss_ce": 9.827417670749128e-05,
      "loss_iou": 0.22265625,
      "loss_num": 0.0255126953125,
      "loss_xval": 0.5703125,
      "num_input_tokens_seen": 381592192,
      "step": 6810
    },
    {
      "epoch": 15.169265033407573,
      "grad_norm": 17.767423629760742,
      "learning_rate": 1e-06,
      "loss": 0.4062,
      "num_input_tokens_seen": 381650596,
      "step": 6811
    },
    {
      "epoch": 15.169265033407573,
      "loss": 0.3915339708328247,
      "loss_ce": 0.00011550119234016165,
      "loss_iou": 0.1787109375,
      "loss_num": 0.006805419921875,
      "loss_xval": 0.390625,
      "num_input_tokens_seen": 381650596,
      "step": 6811
    },
    {
      "epoch": 15.171492204899778,
      "grad_norm": 13.659879684448242,
      "learning_rate": 1e-06,
      "loss": 0.4005,
      "num_input_tokens_seen": 381706044,
      "step": 6812
    },
    {
      "epoch": 15.171492204899778,
      "loss": 0.37923288345336914,
      "loss_ce": 8.25146198621951e-05,
      "loss_iou": 0.1572265625,
      "loss_num": 0.0128173828125,
      "loss_xval": 0.37890625,
      "num_input_tokens_seen": 381706044,
      "step": 6812
    },
    {
      "epoch": 15.173719376391983,
      "grad_norm": 14.41109561920166,
      "learning_rate": 1e-06,
      "loss": 0.4001,
      "num_input_tokens_seen": 381759412,
      "step": 6813
    },
    {
      "epoch": 15.173719376391983,
      "loss": 0.38645416498184204,
      "loss_ce": 0.00010162763646803796,
      "loss_iou": 0.166015625,
      "loss_num": 0.01080322265625,
      "loss_xval": 0.38671875,
      "num_input_tokens_seen": 381759412,
      "step": 6813
    },
    {
      "epoch": 15.175946547884188,
      "grad_norm": 15.983702659606934,
      "learning_rate": 1e-06,
      "loss": 0.4464,
      "num_input_tokens_seen": 381817464,
      "step": 6814
    },
    {
      "epoch": 15.175946547884188,
      "loss": 0.45959311723709106,
      "loss_ce": 0.00012045353651046753,
      "loss_iou": 0.20703125,
      "loss_num": 0.00921630859375,
      "loss_xval": 0.458984375,
      "num_input_tokens_seen": 381817464,
      "step": 6814
    },
    {
      "epoch": 15.178173719376392,
      "grad_norm": 22.976959228515625,
      "learning_rate": 1e-06,
      "loss": 0.5359,
      "num_input_tokens_seen": 381870696,
      "step": 6815
    },
    {
      "epoch": 15.178173719376392,
      "loss": 0.3837730288505554,
      "loss_ce": 0.0001670640049269423,
      "loss_iou": 0.16015625,
      "loss_num": 0.01275634765625,
      "loss_xval": 0.3828125,
      "num_input_tokens_seen": 381870696,
      "step": 6815
    },
    {
      "epoch": 15.180400890868597,
      "grad_norm": 23.866098403930664,
      "learning_rate": 1e-06,
      "loss": 0.5447,
      "num_input_tokens_seen": 381925464,
      "step": 6816
    },
    {
      "epoch": 15.180400890868597,
      "loss": 0.3768770694732666,
      "loss_ce": 0.0001070558573701419,
      "loss_iou": 0.1767578125,
      "loss_num": 0.0047607421875,
      "loss_xval": 0.376953125,
      "num_input_tokens_seen": 381925464,
      "step": 6816
    },
    {
      "epoch": 15.182628062360802,
      "grad_norm": 21.5544490814209,
      "learning_rate": 1e-06,
      "loss": 0.5746,
      "num_input_tokens_seen": 381979292,
      "step": 6817
    },
    {
      "epoch": 15.182628062360802,
      "loss": 0.6533318758010864,
      "loss_ce": 0.0001337053836323321,
      "loss_iou": 0.283203125,
      "loss_num": 0.017822265625,
      "loss_xval": 0.65234375,
      "num_input_tokens_seen": 381979292,
      "step": 6817
    },
    {
      "epoch": 15.184855233853007,
      "grad_norm": 21.73903465270996,
      "learning_rate": 1e-06,
      "loss": 0.4286,
      "num_input_tokens_seen": 382034532,
      "step": 6818
    },
    {
      "epoch": 15.184855233853007,
      "loss": 0.4159244894981384,
      "loss_ce": 0.00015300727682188153,
      "loss_iou": 0.1806640625,
      "loss_num": 0.0111083984375,
      "loss_xval": 0.416015625,
      "num_input_tokens_seen": 382034532,
      "step": 6818
    },
    {
      "epoch": 15.187082405345212,
      "grad_norm": 14.082677841186523,
      "learning_rate": 1e-06,
      "loss": 0.345,
      "num_input_tokens_seen": 382089388,
      "step": 6819
    },
    {
      "epoch": 15.187082405345212,
      "loss": 0.3454124331474304,
      "loss_ce": 7.550232112407684e-05,
      "loss_iou": 0.14453125,
      "loss_num": 0.0111083984375,
      "loss_xval": 0.345703125,
      "num_input_tokens_seen": 382089388,
      "step": 6819
    },
    {
      "epoch": 15.189309576837417,
      "grad_norm": 13.859386444091797,
      "learning_rate": 1e-06,
      "loss": 0.3842,
      "num_input_tokens_seen": 382148896,
      "step": 6820
    },
    {
      "epoch": 15.189309576837417,
      "loss": 0.33224329352378845,
      "loss_ce": 8.99658989510499e-05,
      "loss_iou": 0.1484375,
      "loss_num": 0.0069580078125,
      "loss_xval": 0.33203125,
      "num_input_tokens_seen": 382148896,
      "step": 6820
    },
    {
      "epoch": 15.191536748329622,
      "grad_norm": 21.40119171142578,
      "learning_rate": 1e-06,
      "loss": 0.5285,
      "num_input_tokens_seen": 382204260,
      "step": 6821
    },
    {
      "epoch": 15.191536748329622,
      "loss": 0.38690587878227234,
      "loss_ce": 9.558402234688401e-05,
      "loss_iou": 0.171875,
      "loss_num": 0.00848388671875,
      "loss_xval": 0.38671875,
      "num_input_tokens_seen": 382204260,
      "step": 6821
    },
    {
      "epoch": 15.193763919821826,
      "grad_norm": 23.956090927124023,
      "learning_rate": 1e-06,
      "loss": 0.2846,
      "num_input_tokens_seen": 382261104,
      "step": 6822
    },
    {
      "epoch": 15.193763919821826,
      "loss": 0.33018404245376587,
      "loss_ce": 0.0001059087153407745,
      "loss_iou": 0.1533203125,
      "loss_num": 0.00469970703125,
      "loss_xval": 0.330078125,
      "num_input_tokens_seen": 382261104,
      "step": 6822
    },
    {
      "epoch": 15.195991091314031,
      "grad_norm": 29.63949203491211,
      "learning_rate": 1e-06,
      "loss": 0.4216,
      "num_input_tokens_seen": 382316244,
      "step": 6823
    },
    {
      "epoch": 15.195991091314031,
      "loss": 0.27267590165138245,
      "loss_ce": 9.288682485930622e-05,
      "loss_iou": 0.125,
      "loss_num": 0.00439453125,
      "loss_xval": 0.2734375,
      "num_input_tokens_seen": 382316244,
      "step": 6823
    },
    {
      "epoch": 15.198218262806236,
      "grad_norm": 25.926671981811523,
      "learning_rate": 1e-06,
      "loss": 0.5995,
      "num_input_tokens_seen": 382367780,
      "step": 6824
    },
    {
      "epoch": 15.198218262806236,
      "loss": 0.7526465654373169,
      "loss_ce": 0.0001441580825485289,
      "loss_iou": 0.3203125,
      "loss_num": 0.0225830078125,
      "loss_xval": 0.75390625,
      "num_input_tokens_seen": 382367780,
      "step": 6824
    },
    {
      "epoch": 15.200445434298441,
      "grad_norm": 32.10568618774414,
      "learning_rate": 1e-06,
      "loss": 0.5171,
      "num_input_tokens_seen": 382423572,
      "step": 6825
    },
    {
      "epoch": 15.200445434298441,
      "loss": 0.5426996946334839,
      "loss_ce": 9.716699423734099e-05,
      "loss_iou": 0.2392578125,
      "loss_num": 0.0128173828125,
      "loss_xval": 0.54296875,
      "num_input_tokens_seen": 382423572,
      "step": 6825
    },
    {
      "epoch": 15.202672605790646,
      "grad_norm": 30.0792293548584,
      "learning_rate": 1e-06,
      "loss": 0.4723,
      "num_input_tokens_seen": 382476324,
      "step": 6826
    },
    {
      "epoch": 15.202672605790646,
      "loss": 0.47158297896385193,
      "loss_ce": 0.0001474320306442678,
      "loss_iou": 0.2060546875,
      "loss_num": 0.01165771484375,
      "loss_xval": 0.470703125,
      "num_input_tokens_seen": 382476324,
      "step": 6826
    },
    {
      "epoch": 15.20489977728285,
      "grad_norm": 25.993913650512695,
      "learning_rate": 1e-06,
      "loss": 0.3837,
      "num_input_tokens_seen": 382531860,
      "step": 6827
    },
    {
      "epoch": 15.20489977728285,
      "loss": 0.37444430589675903,
      "loss_ce": 0.00011569406342459843,
      "loss_iou": 0.15625,
      "loss_num": 0.012451171875,
      "loss_xval": 0.375,
      "num_input_tokens_seen": 382531860,
      "step": 6827
    },
    {
      "epoch": 15.207126948775056,
      "grad_norm": 20.49637794494629,
      "learning_rate": 1e-06,
      "loss": 0.5971,
      "num_input_tokens_seen": 382587412,
      "step": 6828
    },
    {
      "epoch": 15.207126948775056,
      "loss": 0.459941565990448,
      "loss_ce": 0.00010270022903569043,
      "loss_iou": 0.2041015625,
      "loss_num": 0.010498046875,
      "loss_xval": 0.458984375,
      "num_input_tokens_seen": 382587412,
      "step": 6828
    },
    {
      "epoch": 15.20935412026726,
      "grad_norm": 16.51508331298828,
      "learning_rate": 1e-06,
      "loss": 0.4748,
      "num_input_tokens_seen": 382642572,
      "step": 6829
    },
    {
      "epoch": 15.20935412026726,
      "loss": 0.5448847413063049,
      "loss_ce": 8.491726475767791e-05,
      "loss_iou": 0.2333984375,
      "loss_num": 0.01556396484375,
      "loss_xval": 0.54296875,
      "num_input_tokens_seen": 382642572,
      "step": 6829
    },
    {
      "epoch": 15.211581291759465,
      "grad_norm": 17.06146240234375,
      "learning_rate": 1e-06,
      "loss": 0.4023,
      "num_input_tokens_seen": 382699556,
      "step": 6830
    },
    {
      "epoch": 15.211581291759465,
      "loss": 0.424904465675354,
      "loss_ce": 0.00022184928820934147,
      "loss_iou": 0.1904296875,
      "loss_num": 0.0086669921875,
      "loss_xval": 0.423828125,
      "num_input_tokens_seen": 382699556,
      "step": 6830
    },
    {
      "epoch": 15.21380846325167,
      "grad_norm": 18.930295944213867,
      "learning_rate": 1e-06,
      "loss": 0.4568,
      "num_input_tokens_seen": 382757356,
      "step": 6831
    },
    {
      "epoch": 15.21380846325167,
      "loss": 0.5095734596252441,
      "loss_ce": 0.0004182119155302644,
      "loss_iou": 0.201171875,
      "loss_num": 0.0211181640625,
      "loss_xval": 0.5078125,
      "num_input_tokens_seen": 382757356,
      "step": 6831
    },
    {
      "epoch": 15.216035634743875,
      "grad_norm": 21.91002655029297,
      "learning_rate": 1e-06,
      "loss": 0.375,
      "num_input_tokens_seen": 382811564,
      "step": 6832
    },
    {
      "epoch": 15.216035634743875,
      "loss": 0.40269356966018677,
      "loss_ce": 0.00010566625132923946,
      "loss_iou": 0.1826171875,
      "loss_num": 0.0076904296875,
      "loss_xval": 0.40234375,
      "num_input_tokens_seen": 382811564,
      "step": 6832
    },
    {
      "epoch": 15.21826280623608,
      "grad_norm": 26.341665267944336,
      "learning_rate": 1e-06,
      "loss": 0.4079,
      "num_input_tokens_seen": 382865496,
      "step": 6833
    },
    {
      "epoch": 15.21826280623608,
      "loss": 0.4928325414657593,
      "loss_ce": 9.574595605954528e-05,
      "loss_iou": 0.1962890625,
      "loss_num": 0.02001953125,
      "loss_xval": 0.4921875,
      "num_input_tokens_seen": 382865496,
      "step": 6833
    },
    {
      "epoch": 15.220489977728285,
      "grad_norm": 21.50320816040039,
      "learning_rate": 1e-06,
      "loss": 0.355,
      "num_input_tokens_seen": 382922332,
      "step": 6834
    },
    {
      "epoch": 15.220489977728285,
      "loss": 0.38630521297454834,
      "loss_ce": 0.0007308792555704713,
      "loss_iou": 0.162109375,
      "loss_num": 0.01226806640625,
      "loss_xval": 0.384765625,
      "num_input_tokens_seen": 382922332,
      "step": 6834
    },
    {
      "epoch": 15.22271714922049,
      "grad_norm": 22.518569946289062,
      "learning_rate": 1e-06,
      "loss": 0.4254,
      "num_input_tokens_seen": 382980300,
      "step": 6835
    },
    {
      "epoch": 15.22271714922049,
      "loss": 0.3738847076892853,
      "loss_ce": 0.00010540773655520752,
      "loss_iou": 0.16796875,
      "loss_num": 0.00750732421875,
      "loss_xval": 0.373046875,
      "num_input_tokens_seen": 382980300,
      "step": 6835
    },
    {
      "epoch": 15.224944320712694,
      "grad_norm": 20.747737884521484,
      "learning_rate": 1e-06,
      "loss": 0.4773,
      "num_input_tokens_seen": 383038560,
      "step": 6836
    },
    {
      "epoch": 15.224944320712694,
      "loss": 0.6503742337226868,
      "loss_ce": 0.000105653190985322,
      "loss_iou": 0.267578125,
      "loss_num": 0.0228271484375,
      "loss_xval": 0.6484375,
      "num_input_tokens_seen": 383038560,
      "step": 6836
    },
    {
      "epoch": 15.2271714922049,
      "grad_norm": 31.612144470214844,
      "learning_rate": 1e-06,
      "loss": 0.3559,
      "num_input_tokens_seen": 383093332,
      "step": 6837
    },
    {
      "epoch": 15.2271714922049,
      "loss": 0.3483448326587677,
      "loss_ce": 7.823983469279483e-05,
      "loss_iou": 0.1494140625,
      "loss_num": 0.010009765625,
      "loss_xval": 0.34765625,
      "num_input_tokens_seen": 383093332,
      "step": 6837
    },
    {
      "epoch": 15.229398663697104,
      "grad_norm": 27.15004539489746,
      "learning_rate": 1e-06,
      "loss": 0.3923,
      "num_input_tokens_seen": 383150088,
      "step": 6838
    },
    {
      "epoch": 15.229398663697104,
      "loss": 0.3999464511871338,
      "loss_ce": 0.00028826179914176464,
      "loss_iou": 0.1787109375,
      "loss_num": 0.00836181640625,
      "loss_xval": 0.400390625,
      "num_input_tokens_seen": 383150088,
      "step": 6838
    },
    {
      "epoch": 15.231625835189309,
      "grad_norm": 20.663650512695312,
      "learning_rate": 1e-06,
      "loss": 0.5478,
      "num_input_tokens_seen": 383208636,
      "step": 6839
    },
    {
      "epoch": 15.231625835189309,
      "loss": 0.5868840217590332,
      "loss_ce": 9.200301428791136e-05,
      "loss_iou": 0.2392578125,
      "loss_num": 0.021728515625,
      "loss_xval": 0.5859375,
      "num_input_tokens_seen": 383208636,
      "step": 6839
    },
    {
      "epoch": 15.233853006681514,
      "grad_norm": 15.867166519165039,
      "learning_rate": 1e-06,
      "loss": 0.4572,
      "num_input_tokens_seen": 383262456,
      "step": 6840
    },
    {
      "epoch": 15.233853006681514,
      "loss": 0.5546283721923828,
      "loss_ce": 0.0018329141894355416,
      "loss_iou": 0.244140625,
      "loss_num": 0.01251220703125,
      "loss_xval": 0.5546875,
      "num_input_tokens_seen": 383262456,
      "step": 6840
    },
    {
      "epoch": 15.236080178173719,
      "grad_norm": 22.131988525390625,
      "learning_rate": 1e-06,
      "loss": 0.3841,
      "num_input_tokens_seen": 383319500,
      "step": 6841
    },
    {
      "epoch": 15.236080178173719,
      "loss": 0.3393481373786926,
      "loss_ce": 0.00011472964251879603,
      "loss_iou": 0.1396484375,
      "loss_num": 0.0120849609375,
      "loss_xval": 0.33984375,
      "num_input_tokens_seen": 383319500,
      "step": 6841
    },
    {
      "epoch": 15.238307349665924,
      "grad_norm": 17.503231048583984,
      "learning_rate": 1e-06,
      "loss": 0.5277,
      "num_input_tokens_seen": 383373476,
      "step": 6842
    },
    {
      "epoch": 15.238307349665924,
      "loss": 0.6524415612220764,
      "loss_ce": 9.780684194993228e-05,
      "loss_iou": 0.267578125,
      "loss_num": 0.023681640625,
      "loss_xval": 0.65234375,
      "num_input_tokens_seen": 383373476,
      "step": 6842
    },
    {
      "epoch": 15.240534521158128,
      "grad_norm": 27.421920776367188,
      "learning_rate": 1e-06,
      "loss": 0.3908,
      "num_input_tokens_seen": 383430396,
      "step": 6843
    },
    {
      "epoch": 15.240534521158128,
      "loss": 0.2873254120349884,
      "loss_ce": 9.397394023835659e-05,
      "loss_iou": 0.1279296875,
      "loss_num": 0.0064697265625,
      "loss_xval": 0.287109375,
      "num_input_tokens_seen": 383430396,
      "step": 6843
    },
    {
      "epoch": 15.242761692650333,
      "grad_norm": 18.538227081298828,
      "learning_rate": 1e-06,
      "loss": 0.8737,
      "num_input_tokens_seen": 383486732,
      "step": 6844
    },
    {
      "epoch": 15.242761692650333,
      "loss": 0.9400348663330078,
      "loss_ce": 9.340968244941905e-05,
      "loss_iou": 0.3984375,
      "loss_num": 0.028564453125,
      "loss_xval": 0.94140625,
      "num_input_tokens_seen": 383486732,
      "step": 6844
    },
    {
      "epoch": 15.244988864142538,
      "grad_norm": 18.570964813232422,
      "learning_rate": 1e-06,
      "loss": 0.4564,
      "num_input_tokens_seen": 383544740,
      "step": 6845
    },
    {
      "epoch": 15.244988864142538,
      "loss": 0.507795512676239,
      "loss_ce": 0.00010509089042898268,
      "loss_iou": 0.1962890625,
      "loss_num": 0.0230712890625,
      "loss_xval": 0.5078125,
      "num_input_tokens_seen": 383544740,
      "step": 6845
    },
    {
      "epoch": 15.247216035634743,
      "grad_norm": 25.994709014892578,
      "learning_rate": 1e-06,
      "loss": 0.3808,
      "num_input_tokens_seen": 383601352,
      "step": 6846
    },
    {
      "epoch": 15.247216035634743,
      "loss": 0.4001128673553467,
      "loss_ce": 8.847533172229305e-05,
      "loss_iou": 0.1806640625,
      "loss_num": 0.0076904296875,
      "loss_xval": 0.400390625,
      "num_input_tokens_seen": 383601352,
      "step": 6846
    },
    {
      "epoch": 15.249443207126948,
      "grad_norm": 30.125736236572266,
      "learning_rate": 1e-06,
      "loss": 0.3879,
      "num_input_tokens_seen": 383656492,
      "step": 6847
    },
    {
      "epoch": 15.249443207126948,
      "loss": 0.33361074328422546,
      "loss_ce": 0.00011466215801192448,
      "loss_iou": 0.1494140625,
      "loss_num": 0.00677490234375,
      "loss_xval": 0.333984375,
      "num_input_tokens_seen": 383656492,
      "step": 6847
    },
    {
      "epoch": 15.251670378619155,
      "grad_norm": 16.290237426757812,
      "learning_rate": 1e-06,
      "loss": 0.5262,
      "num_input_tokens_seen": 383713968,
      "step": 6848
    },
    {
      "epoch": 15.251670378619155,
      "loss": 0.4467582702636719,
      "loss_ce": 0.00010298557754140347,
      "loss_iou": 0.1904296875,
      "loss_num": 0.01300048828125,
      "loss_xval": 0.447265625,
      "num_input_tokens_seen": 383713968,
      "step": 6848
    },
    {
      "epoch": 15.25389755011136,
      "grad_norm": 32.09817886352539,
      "learning_rate": 1e-06,
      "loss": 0.5307,
      "num_input_tokens_seen": 383769468,
      "step": 6849
    },
    {
      "epoch": 15.25389755011136,
      "loss": 0.5073274970054626,
      "loss_ce": 0.0001253713999176398,
      "loss_iou": 0.2021484375,
      "loss_num": 0.0203857421875,
      "loss_xval": 0.5078125,
      "num_input_tokens_seen": 383769468,
      "step": 6849
    },
    {
      "epoch": 15.256124721603564,
      "grad_norm": 19.521888732910156,
      "learning_rate": 1e-06,
      "loss": 0.4386,
      "num_input_tokens_seen": 383826068,
      "step": 6850
    },
    {
      "epoch": 15.256124721603564,
      "loss": 0.3947462737560272,
      "loss_ce": 9.295322524849325e-05,
      "loss_iou": 0.1494140625,
      "loss_num": 0.0191650390625,
      "loss_xval": 0.39453125,
      "num_input_tokens_seen": 383826068,
      "step": 6850
    },
    {
      "epoch": 15.25835189309577,
      "grad_norm": 18.782377243041992,
      "learning_rate": 1e-06,
      "loss": 0.2575,
      "num_input_tokens_seen": 383883012,
      "step": 6851
    },
    {
      "epoch": 15.25835189309577,
      "loss": 0.2647382318973541,
      "loss_ce": 8.9788663899526e-05,
      "loss_iou": 0.109375,
      "loss_num": 0.009033203125,
      "loss_xval": 0.265625,
      "num_input_tokens_seen": 383883012,
      "step": 6851
    },
    {
      "epoch": 15.260579064587974,
      "grad_norm": 21.034971237182617,
      "learning_rate": 1e-06,
      "loss": 0.4045,
      "num_input_tokens_seen": 383936956,
      "step": 6852
    },
    {
      "epoch": 15.260579064587974,
      "loss": 0.4515724182128906,
      "loss_ce": 9.537780715618283e-05,
      "loss_iou": 0.185546875,
      "loss_num": 0.015869140625,
      "loss_xval": 0.451171875,
      "num_input_tokens_seen": 383936956,
      "step": 6852
    },
    {
      "epoch": 15.262806236080179,
      "grad_norm": 32.299102783203125,
      "learning_rate": 1e-06,
      "loss": 0.3529,
      "num_input_tokens_seen": 383993404,
      "step": 6853
    },
    {
      "epoch": 15.262806236080179,
      "loss": 0.31591275334358215,
      "loss_ce": 0.00011685363278957084,
      "loss_iou": 0.142578125,
      "loss_num": 0.00628662109375,
      "loss_xval": 0.31640625,
      "num_input_tokens_seen": 383993404,
      "step": 6853
    },
    {
      "epoch": 15.265033407572384,
      "grad_norm": 17.330768585205078,
      "learning_rate": 1e-06,
      "loss": 0.485,
      "num_input_tokens_seen": 384048856,
      "step": 6854
    },
    {
      "epoch": 15.265033407572384,
      "loss": 0.5928666591644287,
      "loss_ce": 0.00021531574020627886,
      "loss_iou": 0.2255859375,
      "loss_num": 0.0281982421875,
      "loss_xval": 0.59375,
      "num_input_tokens_seen": 384048856,
      "step": 6854
    },
    {
      "epoch": 15.267260579064589,
      "grad_norm": 19.941410064697266,
      "learning_rate": 1e-06,
      "loss": 0.4413,
      "num_input_tokens_seen": 384104048,
      "step": 6855
    },
    {
      "epoch": 15.267260579064589,
      "loss": 0.40443968772888184,
      "loss_ce": 8.178211282938719e-05,
      "loss_iou": 0.1748046875,
      "loss_num": 0.0111083984375,
      "loss_xval": 0.404296875,
      "num_input_tokens_seen": 384104048,
      "step": 6855
    },
    {
      "epoch": 15.269487750556793,
      "grad_norm": 19.013338088989258,
      "learning_rate": 1e-06,
      "loss": 0.52,
      "num_input_tokens_seen": 384155284,
      "step": 6856
    },
    {
      "epoch": 15.269487750556793,
      "loss": 0.3980572819709778,
      "loss_ce": 0.00010806175851030275,
      "loss_iou": 0.169921875,
      "loss_num": 0.01177978515625,
      "loss_xval": 0.3984375,
      "num_input_tokens_seen": 384155284,
      "step": 6856
    },
    {
      "epoch": 15.271714922048998,
      "grad_norm": 17.63640594482422,
      "learning_rate": 1e-06,
      "loss": 0.4304,
      "num_input_tokens_seen": 384212084,
      "step": 6857
    },
    {
      "epoch": 15.271714922048998,
      "loss": 0.46271538734436035,
      "loss_ce": 0.00025199330411851406,
      "loss_iou": 0.1845703125,
      "loss_num": 0.0184326171875,
      "loss_xval": 0.462890625,
      "num_input_tokens_seen": 384212084,
      "step": 6857
    },
    {
      "epoch": 15.273942093541203,
      "grad_norm": 17.39361572265625,
      "learning_rate": 1e-06,
      "loss": 0.5069,
      "num_input_tokens_seen": 384266828,
      "step": 6858
    },
    {
      "epoch": 15.273942093541203,
      "loss": 0.4388231039047241,
      "loss_ce": 0.0001023877048282884,
      "loss_iou": 0.18359375,
      "loss_num": 0.01422119140625,
      "loss_xval": 0.439453125,
      "num_input_tokens_seen": 384266828,
      "step": 6858
    },
    {
      "epoch": 15.276169265033408,
      "grad_norm": 20.657392501831055,
      "learning_rate": 1e-06,
      "loss": 0.3784,
      "num_input_tokens_seen": 384320760,
      "step": 6859
    },
    {
      "epoch": 15.276169265033408,
      "loss": 0.3058285713195801,
      "loss_ce": 0.00010348795331083238,
      "loss_iou": 0.140625,
      "loss_num": 0.004791259765625,
      "loss_xval": 0.306640625,
      "num_input_tokens_seen": 384320760,
      "step": 6859
    },
    {
      "epoch": 15.278396436525613,
      "grad_norm": 17.122583389282227,
      "learning_rate": 1e-06,
      "loss": 0.3612,
      "num_input_tokens_seen": 384377020,
      "step": 6860
    },
    {
      "epoch": 15.278396436525613,
      "loss": 0.33143696188926697,
      "loss_ce": 0.0001076041953638196,
      "loss_iou": 0.138671875,
      "loss_num": 0.0107421875,
      "loss_xval": 0.33203125,
      "num_input_tokens_seen": 384377020,
      "step": 6860
    },
    {
      "epoch": 15.280623608017818,
      "grad_norm": 24.28658676147461,
      "learning_rate": 1e-06,
      "loss": 0.4031,
      "num_input_tokens_seen": 384431080,
      "step": 6861
    },
    {
      "epoch": 15.280623608017818,
      "loss": 0.518648624420166,
      "loss_ce": 9.391328057972714e-05,
      "loss_iou": 0.21875,
      "loss_num": 0.0162353515625,
      "loss_xval": 0.51953125,
      "num_input_tokens_seen": 384431080,
      "step": 6861
    },
    {
      "epoch": 15.282850779510023,
      "grad_norm": 11.215580940246582,
      "learning_rate": 1e-06,
      "loss": 0.304,
      "num_input_tokens_seen": 384487584,
      "step": 6862
    },
    {
      "epoch": 15.282850779510023,
      "loss": 0.31601682305336,
      "loss_ce": 9.885276085697114e-05,
      "loss_iou": 0.1357421875,
      "loss_num": 0.00885009765625,
      "loss_xval": 0.31640625,
      "num_input_tokens_seen": 384487584,
      "step": 6862
    },
    {
      "epoch": 15.285077951002227,
      "grad_norm": 22.595476150512695,
      "learning_rate": 1e-06,
      "loss": 0.3841,
      "num_input_tokens_seen": 384544196,
      "step": 6863
    },
    {
      "epoch": 15.285077951002227,
      "loss": 0.49600616097450256,
      "loss_ce": 0.00015653966693207622,
      "loss_iou": 0.2021484375,
      "loss_num": 0.018310546875,
      "loss_xval": 0.49609375,
      "num_input_tokens_seen": 384544196,
      "step": 6863
    },
    {
      "epoch": 15.287305122494432,
      "grad_norm": 65.94528198242188,
      "learning_rate": 1e-06,
      "loss": 0.4559,
      "num_input_tokens_seen": 384598184,
      "step": 6864
    },
    {
      "epoch": 15.287305122494432,
      "loss": 0.5419918894767761,
      "loss_ce": 0.00012176702875876799,
      "loss_iou": 0.248046875,
      "loss_num": 0.009033203125,
      "loss_xval": 0.54296875,
      "num_input_tokens_seen": 384598184,
      "step": 6864
    },
    {
      "epoch": 15.289532293986637,
      "grad_norm": 18.129741668701172,
      "learning_rate": 1e-06,
      "loss": 0.4641,
      "num_input_tokens_seen": 384655212,
      "step": 6865
    },
    {
      "epoch": 15.289532293986637,
      "loss": 0.5154638290405273,
      "loss_ce": 8.296032319776714e-05,
      "loss_iou": 0.2138671875,
      "loss_num": 0.0177001953125,
      "loss_xval": 0.515625,
      "num_input_tokens_seen": 384655212,
      "step": 6865
    },
    {
      "epoch": 15.291759465478842,
      "grad_norm": 33.346805572509766,
      "learning_rate": 1e-06,
      "loss": 0.4791,
      "num_input_tokens_seen": 384711224,
      "step": 6866
    },
    {
      "epoch": 15.291759465478842,
      "loss": 0.5016932487487793,
      "loss_ce": 0.00010639546962920576,
      "loss_iou": 0.2294921875,
      "loss_num": 0.0086669921875,
      "loss_xval": 0.5,
      "num_input_tokens_seen": 384711224,
      "step": 6866
    },
    {
      "epoch": 15.293986636971047,
      "grad_norm": 17.753189086914062,
      "learning_rate": 1e-06,
      "loss": 0.507,
      "num_input_tokens_seen": 384770584,
      "step": 6867
    },
    {
      "epoch": 15.293986636971047,
      "loss": 0.47503072023391724,
      "loss_ce": 0.00011617955169640481,
      "loss_iou": 0.1953125,
      "loss_num": 0.0169677734375,
      "loss_xval": 0.474609375,
      "num_input_tokens_seen": 384770584,
      "step": 6867
    },
    {
      "epoch": 15.296213808463252,
      "grad_norm": 21.835477828979492,
      "learning_rate": 1e-06,
      "loss": 0.3904,
      "num_input_tokens_seen": 384827512,
      "step": 6868
    },
    {
      "epoch": 15.296213808463252,
      "loss": 0.3441203534603119,
      "loss_ce": 8.04224400781095e-05,
      "loss_iou": 0.142578125,
      "loss_num": 0.0118408203125,
      "loss_xval": 0.34375,
      "num_input_tokens_seen": 384827512,
      "step": 6868
    },
    {
      "epoch": 15.298440979955457,
      "grad_norm": 11.53410816192627,
      "learning_rate": 1e-06,
      "loss": 0.3595,
      "num_input_tokens_seen": 384884920,
      "step": 6869
    },
    {
      "epoch": 15.298440979955457,
      "loss": 0.42830324172973633,
      "loss_ce": 8.056573278736323e-05,
      "loss_iou": 0.189453125,
      "loss_num": 0.01007080078125,
      "loss_xval": 0.427734375,
      "num_input_tokens_seen": 384884920,
      "step": 6869
    },
    {
      "epoch": 15.300668151447661,
      "grad_norm": 24.92241668701172,
      "learning_rate": 1e-06,
      "loss": 0.4735,
      "num_input_tokens_seen": 384940092,
      "step": 6870
    },
    {
      "epoch": 15.300668151447661,
      "loss": 0.30489498376846313,
      "loss_ce": 8.541756687918678e-05,
      "loss_iou": 0.1328125,
      "loss_num": 0.0078125,
      "loss_xval": 0.3046875,
      "num_input_tokens_seen": 384940092,
      "step": 6870
    },
    {
      "epoch": 15.302895322939866,
      "grad_norm": 24.968305587768555,
      "learning_rate": 1e-06,
      "loss": 0.48,
      "num_input_tokens_seen": 384996776,
      "step": 6871
    },
    {
      "epoch": 15.302895322939866,
      "loss": 0.4336788058280945,
      "loss_ce": 8.504305151291192e-05,
      "loss_iou": 0.181640625,
      "loss_num": 0.01385498046875,
      "loss_xval": 0.43359375,
      "num_input_tokens_seen": 384996776,
      "step": 6871
    },
    {
      "epoch": 15.305122494432071,
      "grad_norm": 15.246042251586914,
      "learning_rate": 1e-06,
      "loss": 0.4207,
      "num_input_tokens_seen": 385056728,
      "step": 6872
    },
    {
      "epoch": 15.305122494432071,
      "loss": 0.35528939962387085,
      "loss_ce": 9.531193063594401e-05,
      "loss_iou": 0.150390625,
      "loss_num": 0.01080322265625,
      "loss_xval": 0.35546875,
      "num_input_tokens_seen": 385056728,
      "step": 6872
    },
    {
      "epoch": 15.307349665924276,
      "grad_norm": 18.476036071777344,
      "learning_rate": 1e-06,
      "loss": 0.3138,
      "num_input_tokens_seen": 385113780,
      "step": 6873
    },
    {
      "epoch": 15.307349665924276,
      "loss": 0.3595203459262848,
      "loss_ce": 0.00011481612455099821,
      "loss_iou": 0.1640625,
      "loss_num": 0.006439208984375,
      "loss_xval": 0.359375,
      "num_input_tokens_seen": 385113780,
      "step": 6873
    },
    {
      "epoch": 15.309576837416481,
      "grad_norm": 17.357139587402344,
      "learning_rate": 1e-06,
      "loss": 0.3486,
      "num_input_tokens_seen": 385169428,
      "step": 6874
    },
    {
      "epoch": 15.309576837416481,
      "loss": 0.45047512650489807,
      "loss_ce": 9.672968735685572e-05,
      "loss_iou": 0.205078125,
      "loss_num": 0.0079345703125,
      "loss_xval": 0.451171875,
      "num_input_tokens_seen": 385169428,
      "step": 6874
    },
    {
      "epoch": 15.311804008908686,
      "grad_norm": 16.00808334350586,
      "learning_rate": 1e-06,
      "loss": 0.4884,
      "num_input_tokens_seen": 385226272,
      "step": 6875
    },
    {
      "epoch": 15.311804008908686,
      "loss": 0.48656389117240906,
      "loss_ce": 0.00011366438411641866,
      "loss_iou": 0.2109375,
      "loss_num": 0.01300048828125,
      "loss_xval": 0.486328125,
      "num_input_tokens_seen": 385226272,
      "step": 6875
    },
    {
      "epoch": 15.31403118040089,
      "grad_norm": 22.285694122314453,
      "learning_rate": 1e-06,
      "loss": 0.3616,
      "num_input_tokens_seen": 385281308,
      "step": 6876
    },
    {
      "epoch": 15.31403118040089,
      "loss": 0.30913233757019043,
      "loss_ce": 8.081403211690485e-05,
      "loss_iou": 0.11767578125,
      "loss_num": 0.01470947265625,
      "loss_xval": 0.30859375,
      "num_input_tokens_seen": 385281308,
      "step": 6876
    },
    {
      "epoch": 15.316258351893095,
      "grad_norm": 406.5437927246094,
      "learning_rate": 1e-06,
      "loss": 0.4264,
      "num_input_tokens_seen": 385338752,
      "step": 6877
    },
    {
      "epoch": 15.316258351893095,
      "loss": 0.3918250799179077,
      "loss_ce": 0.00010145184933207929,
      "loss_iou": 0.17578125,
      "loss_num": 0.007781982421875,
      "loss_xval": 0.392578125,
      "num_input_tokens_seen": 385338752,
      "step": 6877
    },
    {
      "epoch": 15.3184855233853,
      "grad_norm": 17.710416793823242,
      "learning_rate": 1e-06,
      "loss": 0.3374,
      "num_input_tokens_seen": 385396576,
      "step": 6878
    },
    {
      "epoch": 15.3184855233853,
      "loss": 0.35166865587234497,
      "loss_ce": 0.00010614388156682253,
      "loss_iou": 0.1552734375,
      "loss_num": 0.00836181640625,
      "loss_xval": 0.3515625,
      "num_input_tokens_seen": 385396576,
      "step": 6878
    },
    {
      "epoch": 15.320712694877505,
      "grad_norm": 16.165138244628906,
      "learning_rate": 1e-06,
      "loss": 0.4187,
      "num_input_tokens_seen": 385454284,
      "step": 6879
    },
    {
      "epoch": 15.320712694877505,
      "loss": 0.4211636483669281,
      "loss_ce": 0.00014315356384031475,
      "loss_iou": 0.185546875,
      "loss_num": 0.0098876953125,
      "loss_xval": 0.421875,
      "num_input_tokens_seen": 385454284,
      "step": 6879
    },
    {
      "epoch": 15.32293986636971,
      "grad_norm": 19.196983337402344,
      "learning_rate": 1e-06,
      "loss": 0.4114,
      "num_input_tokens_seen": 385511284,
      "step": 6880
    },
    {
      "epoch": 15.32293986636971,
      "loss": 0.42685467004776,
      "loss_ce": 9.687192505225539e-05,
      "loss_iou": 0.181640625,
      "loss_num": 0.01263427734375,
      "loss_xval": 0.42578125,
      "num_input_tokens_seen": 385511284,
      "step": 6880
    },
    {
      "epoch": 15.325167037861915,
      "grad_norm": 25.625106811523438,
      "learning_rate": 1e-06,
      "loss": 0.3899,
      "num_input_tokens_seen": 385565236,
      "step": 6881
    },
    {
      "epoch": 15.325167037861915,
      "loss": 0.3682980537414551,
      "loss_ce": 0.00013401404430624098,
      "loss_iou": 0.1640625,
      "loss_num": 0.0078125,
      "loss_xval": 0.3671875,
      "num_input_tokens_seen": 385565236,
      "step": 6881
    },
    {
      "epoch": 15.32739420935412,
      "grad_norm": 21.3586368560791,
      "learning_rate": 1e-06,
      "loss": 0.4805,
      "num_input_tokens_seen": 385619852,
      "step": 6882
    },
    {
      "epoch": 15.32739420935412,
      "loss": 0.5712690353393555,
      "loss_ce": 0.00010209472384303808,
      "loss_iou": 0.26953125,
      "loss_num": 0.006378173828125,
      "loss_xval": 0.5703125,
      "num_input_tokens_seen": 385619852,
      "step": 6882
    },
    {
      "epoch": 15.329621380846325,
      "grad_norm": 20.074663162231445,
      "learning_rate": 1e-06,
      "loss": 0.4422,
      "num_input_tokens_seen": 385675444,
      "step": 6883
    },
    {
      "epoch": 15.329621380846325,
      "loss": 0.44748926162719727,
      "loss_ce": 0.00010156280768569559,
      "loss_iou": 0.1943359375,
      "loss_num": 0.0115966796875,
      "loss_xval": 0.447265625,
      "num_input_tokens_seen": 385675444,
      "step": 6883
    },
    {
      "epoch": 15.33184855233853,
      "grad_norm": 19.188108444213867,
      "learning_rate": 1e-06,
      "loss": 0.5017,
      "num_input_tokens_seen": 385731136,
      "step": 6884
    },
    {
      "epoch": 15.33184855233853,
      "loss": 0.269458532333374,
      "loss_ce": 0.0001714447425911203,
      "loss_iou": 0.123046875,
      "loss_num": 0.004638671875,
      "loss_xval": 0.26953125,
      "num_input_tokens_seen": 385731136,
      "step": 6884
    },
    {
      "epoch": 15.334075723830734,
      "grad_norm": 16.712444305419922,
      "learning_rate": 1e-06,
      "loss": 0.4189,
      "num_input_tokens_seen": 385788012,
      "step": 6885
    },
    {
      "epoch": 15.334075723830734,
      "loss": 0.4538283050060272,
      "loss_ce": 9.298422082792968e-05,
      "loss_iou": 0.2021484375,
      "loss_num": 0.00982666015625,
      "loss_xval": 0.453125,
      "num_input_tokens_seen": 385788012,
      "step": 6885
    },
    {
      "epoch": 15.33630289532294,
      "grad_norm": 20.65199089050293,
      "learning_rate": 1e-06,
      "loss": 0.5626,
      "num_input_tokens_seen": 385846652,
      "step": 6886
    },
    {
      "epoch": 15.33630289532294,
      "loss": 0.44176751375198364,
      "loss_ce": 0.00011710204125847667,
      "loss_iou": 0.2001953125,
      "loss_num": 0.00823974609375,
      "loss_xval": 0.44140625,
      "num_input_tokens_seen": 385846652,
      "step": 6886
    },
    {
      "epoch": 15.338530066815144,
      "grad_norm": 36.08283996582031,
      "learning_rate": 1e-06,
      "loss": 0.4439,
      "num_input_tokens_seen": 385903884,
      "step": 6887
    },
    {
      "epoch": 15.338530066815144,
      "loss": 0.3505267798900604,
      "loss_ce": 0.00018498269491828978,
      "loss_iou": 0.1513671875,
      "loss_num": 0.0093994140625,
      "loss_xval": 0.349609375,
      "num_input_tokens_seen": 385903884,
      "step": 6887
    },
    {
      "epoch": 15.340757238307349,
      "grad_norm": 27.132768630981445,
      "learning_rate": 1e-06,
      "loss": 0.4734,
      "num_input_tokens_seen": 385958580,
      "step": 6888
    },
    {
      "epoch": 15.340757238307349,
      "loss": 0.42540186643600464,
      "loss_ce": 0.00010891577403526753,
      "loss_iou": 0.19140625,
      "loss_num": 0.00848388671875,
      "loss_xval": 0.42578125,
      "num_input_tokens_seen": 385958580,
      "step": 6888
    },
    {
      "epoch": 15.342984409799554,
      "grad_norm": 18.235618591308594,
      "learning_rate": 1e-06,
      "loss": 0.4809,
      "num_input_tokens_seen": 386016000,
      "step": 6889
    },
    {
      "epoch": 15.342984409799554,
      "loss": 0.4169900715351105,
      "loss_ce": 0.00011995389650110155,
      "loss_iou": 0.1982421875,
      "loss_num": 0.004058837890625,
      "loss_xval": 0.416015625,
      "num_input_tokens_seen": 386016000,
      "step": 6889
    },
    {
      "epoch": 15.345211581291759,
      "grad_norm": 19.752094268798828,
      "learning_rate": 1e-06,
      "loss": 0.6642,
      "num_input_tokens_seen": 386071424,
      "step": 6890
    },
    {
      "epoch": 15.345211581291759,
      "loss": 0.9156758189201355,
      "loss_ce": 0.00014846479461994022,
      "loss_iou": 0.380859375,
      "loss_num": 0.03076171875,
      "loss_xval": 0.9140625,
      "num_input_tokens_seen": 386071424,
      "step": 6890
    },
    {
      "epoch": 15.347438752783964,
      "grad_norm": 20.173500061035156,
      "learning_rate": 1e-06,
      "loss": 0.5695,
      "num_input_tokens_seen": 386121956,
      "step": 6891
    },
    {
      "epoch": 15.347438752783964,
      "loss": 0.2851613163948059,
      "loss_ce": 9.661800868343562e-05,
      "loss_iou": 0.126953125,
      "loss_num": 0.006134033203125,
      "loss_xval": 0.28515625,
      "num_input_tokens_seen": 386121956,
      "step": 6891
    },
    {
      "epoch": 15.34966592427617,
      "grad_norm": 42.235107421875,
      "learning_rate": 1e-06,
      "loss": 0.5524,
      "num_input_tokens_seen": 386174436,
      "step": 6892
    },
    {
      "epoch": 15.34966592427617,
      "loss": 0.5490976572036743,
      "loss_ce": 0.00014746160013601184,
      "loss_iou": 0.236328125,
      "loss_num": 0.01519775390625,
      "loss_xval": 0.55078125,
      "num_input_tokens_seen": 386174436,
      "step": 6892
    },
    {
      "epoch": 15.351893095768375,
      "grad_norm": 62.30870056152344,
      "learning_rate": 1e-06,
      "loss": 0.4912,
      "num_input_tokens_seen": 386229148,
      "step": 6893
    },
    {
      "epoch": 15.351893095768375,
      "loss": 0.4774854779243469,
      "loss_ce": 0.00012952039833180606,
      "loss_iou": 0.203125,
      "loss_num": 0.014404296875,
      "loss_xval": 0.4765625,
      "num_input_tokens_seen": 386229148,
      "step": 6893
    },
    {
      "epoch": 15.35412026726058,
      "grad_norm": 16.55988883972168,
      "learning_rate": 1e-06,
      "loss": 0.3568,
      "num_input_tokens_seen": 386284964,
      "step": 6894
    },
    {
      "epoch": 15.35412026726058,
      "loss": 0.4656769633293152,
      "loss_ce": 9.315234638052061e-05,
      "loss_iou": 0.185546875,
      "loss_num": 0.018798828125,
      "loss_xval": 0.46484375,
      "num_input_tokens_seen": 386284964,
      "step": 6894
    },
    {
      "epoch": 15.356347438752785,
      "grad_norm": 13.955078125,
      "learning_rate": 1e-06,
      "loss": 0.3558,
      "num_input_tokens_seen": 386342464,
      "step": 6895
    },
    {
      "epoch": 15.356347438752785,
      "loss": 0.3424103856086731,
      "loss_ce": 0.00012523065379355103,
      "loss_iou": 0.154296875,
      "loss_num": 0.00665283203125,
      "loss_xval": 0.341796875,
      "num_input_tokens_seen": 386342464,
      "step": 6895
    },
    {
      "epoch": 15.35857461024499,
      "grad_norm": 17.421804428100586,
      "learning_rate": 1e-06,
      "loss": 0.707,
      "num_input_tokens_seen": 386396108,
      "step": 6896
    },
    {
      "epoch": 15.35857461024499,
      "loss": 0.33755582571029663,
      "loss_ce": 9.246024274034426e-05,
      "loss_iou": 0.150390625,
      "loss_num": 0.00732421875,
      "loss_xval": 0.337890625,
      "num_input_tokens_seen": 386396108,
      "step": 6896
    },
    {
      "epoch": 15.360801781737194,
      "grad_norm": 20.638612747192383,
      "learning_rate": 1e-06,
      "loss": 0.3819,
      "num_input_tokens_seen": 386452760,
      "step": 6897
    },
    {
      "epoch": 15.360801781737194,
      "loss": 0.267736554145813,
      "loss_ce": 9.742352995090187e-05,
      "loss_iou": 0.11962890625,
      "loss_num": 0.00567626953125,
      "loss_xval": 0.267578125,
      "num_input_tokens_seen": 386452760,
      "step": 6897
    },
    {
      "epoch": 15.3630289532294,
      "grad_norm": 17.11157989501953,
      "learning_rate": 1e-06,
      "loss": 0.4,
      "num_input_tokens_seen": 386511592,
      "step": 6898
    },
    {
      "epoch": 15.3630289532294,
      "loss": 0.3749087154865265,
      "loss_ce": 9.182207577396184e-05,
      "loss_iou": 0.154296875,
      "loss_num": 0.0130615234375,
      "loss_xval": 0.375,
      "num_input_tokens_seen": 386511592,
      "step": 6898
    },
    {
      "epoch": 15.365256124721604,
      "grad_norm": 12.453452110290527,
      "learning_rate": 1e-06,
      "loss": 0.3186,
      "num_input_tokens_seen": 386568560,
      "step": 6899
    },
    {
      "epoch": 15.365256124721604,
      "loss": 0.2840937376022339,
      "loss_ce": 9.713226609164849e-05,
      "loss_iou": 0.130859375,
      "loss_num": 0.00445556640625,
      "loss_xval": 0.283203125,
      "num_input_tokens_seen": 386568560,
      "step": 6899
    },
    {
      "epoch": 15.367483296213809,
      "grad_norm": 19.97893524169922,
      "learning_rate": 1e-06,
      "loss": 0.3577,
      "num_input_tokens_seen": 386623992,
      "step": 6900
    },
    {
      "epoch": 15.367483296213809,
      "loss": 0.3777737617492676,
      "loss_ce": 8.821256778901443e-05,
      "loss_iou": 0.16796875,
      "loss_num": 0.00799560546875,
      "loss_xval": 0.376953125,
      "num_input_tokens_seen": 386623992,
      "step": 6900
    },
    {
      "epoch": 15.369710467706014,
      "grad_norm": 22.45269775390625,
      "learning_rate": 1e-06,
      "loss": 0.3661,
      "num_input_tokens_seen": 386681452,
      "step": 6901
    },
    {
      "epoch": 15.369710467706014,
      "loss": 0.2848760187625885,
      "loss_ce": 8.596775296609849e-05,
      "loss_iou": 0.1240234375,
      "loss_num": 0.00726318359375,
      "loss_xval": 0.28515625,
      "num_input_tokens_seen": 386681452,
      "step": 6901
    },
    {
      "epoch": 15.371937639198219,
      "grad_norm": 14.801541328430176,
      "learning_rate": 1e-06,
      "loss": 0.4713,
      "num_input_tokens_seen": 386737060,
      "step": 6902
    },
    {
      "epoch": 15.371937639198219,
      "loss": 0.5325612425804138,
      "loss_ce": 9.053810936165974e-05,
      "loss_iou": 0.232421875,
      "loss_num": 0.0133056640625,
      "loss_xval": 0.53125,
      "num_input_tokens_seen": 386737060,
      "step": 6902
    },
    {
      "epoch": 15.374164810690424,
      "grad_norm": 41.93655776977539,
      "learning_rate": 1e-06,
      "loss": 0.3774,
      "num_input_tokens_seen": 386792788,
      "step": 6903
    },
    {
      "epoch": 15.374164810690424,
      "loss": 0.4628770351409912,
      "loss_ce": 0.00010845393990166485,
      "loss_iou": 0.2177734375,
      "loss_num": 0.00543212890625,
      "loss_xval": 0.462890625,
      "num_input_tokens_seen": 386792788,
      "step": 6903
    },
    {
      "epoch": 15.376391982182628,
      "grad_norm": 16.000699996948242,
      "learning_rate": 1e-06,
      "loss": 0.3095,
      "num_input_tokens_seen": 386851464,
      "step": 6904
    },
    {
      "epoch": 15.376391982182628,
      "loss": 0.2996518611907959,
      "loss_ce": 9.130668331636116e-05,
      "loss_iou": 0.1337890625,
      "loss_num": 0.006317138671875,
      "loss_xval": 0.298828125,
      "num_input_tokens_seen": 386851464,
      "step": 6904
    },
    {
      "epoch": 15.378619153674833,
      "grad_norm": 19.54730987548828,
      "learning_rate": 1e-06,
      "loss": 0.36,
      "num_input_tokens_seen": 386903512,
      "step": 6905
    },
    {
      "epoch": 15.378619153674833,
      "loss": 0.5413779020309448,
      "loss_ce": 0.0004843665228690952,
      "loss_iou": 0.203125,
      "loss_num": 0.0269775390625,
      "loss_xval": 0.5390625,
      "num_input_tokens_seen": 386903512,
      "step": 6905
    },
    {
      "epoch": 15.380846325167038,
      "grad_norm": 14.333488464355469,
      "learning_rate": 1e-06,
      "loss": 0.3866,
      "num_input_tokens_seen": 386958488,
      "step": 6906
    },
    {
      "epoch": 15.380846325167038,
      "loss": 0.4710484445095062,
      "loss_ce": 0.0001011955610010773,
      "loss_iou": 0.197265625,
      "loss_num": 0.01519775390625,
      "loss_xval": 0.470703125,
      "num_input_tokens_seen": 386958488,
      "step": 6906
    },
    {
      "epoch": 15.383073496659243,
      "grad_norm": 32.25919723510742,
      "learning_rate": 1e-06,
      "loss": 0.3692,
      "num_input_tokens_seen": 387014204,
      "step": 6907
    },
    {
      "epoch": 15.383073496659243,
      "loss": 0.31186944246292114,
      "loss_ce": 0.0001018614784698002,
      "loss_iou": 0.140625,
      "loss_num": 0.00616455078125,
      "loss_xval": 0.3125,
      "num_input_tokens_seen": 387014204,
      "step": 6907
    },
    {
      "epoch": 15.385300668151448,
      "grad_norm": 22.024864196777344,
      "learning_rate": 1e-06,
      "loss": 0.307,
      "num_input_tokens_seen": 387067160,
      "step": 6908
    },
    {
      "epoch": 15.385300668151448,
      "loss": 0.329712450504303,
      "loss_ce": 9.209779818775132e-05,
      "loss_iou": 0.1533203125,
      "loss_num": 0.004852294921875,
      "loss_xval": 0.330078125,
      "num_input_tokens_seen": 387067160,
      "step": 6908
    },
    {
      "epoch": 15.387527839643653,
      "grad_norm": 19.91106414794922,
      "learning_rate": 1e-06,
      "loss": 0.4446,
      "num_input_tokens_seen": 387124480,
      "step": 6909
    },
    {
      "epoch": 15.387527839643653,
      "loss": 0.3715466856956482,
      "loss_ce": 8.674498531036079e-05,
      "loss_iou": 0.1708984375,
      "loss_num": 0.005767822265625,
      "loss_xval": 0.37109375,
      "num_input_tokens_seen": 387124480,
      "step": 6909
    },
    {
      "epoch": 15.389755011135858,
      "grad_norm": 23.40995979309082,
      "learning_rate": 1e-06,
      "loss": 0.349,
      "num_input_tokens_seen": 387183660,
      "step": 6910
    },
    {
      "epoch": 15.389755011135858,
      "loss": 0.41477489471435547,
      "loss_ce": 0.00010205684520769864,
      "loss_iou": 0.1943359375,
      "loss_num": 0.00518798828125,
      "loss_xval": 0.4140625,
      "num_input_tokens_seen": 387183660,
      "step": 6910
    },
    {
      "epoch": 15.391982182628063,
      "grad_norm": 20.379980087280273,
      "learning_rate": 1e-06,
      "loss": 0.4204,
      "num_input_tokens_seen": 387243208,
      "step": 6911
    },
    {
      "epoch": 15.391982182628063,
      "loss": 0.4162381589412689,
      "loss_ce": 0.0001004426449071616,
      "loss_iou": 0.1953125,
      "loss_num": 0.00494384765625,
      "loss_xval": 0.416015625,
      "num_input_tokens_seen": 387243208,
      "step": 6911
    },
    {
      "epoch": 15.394209354120267,
      "grad_norm": 13.906294822692871,
      "learning_rate": 1e-06,
      "loss": 0.3458,
      "num_input_tokens_seen": 387298652,
      "step": 6912
    },
    {
      "epoch": 15.394209354120267,
      "loss": 0.3168086111545563,
      "loss_ce": 9.718221554066986e-05,
      "loss_iou": 0.1474609375,
      "loss_num": 0.004180908203125,
      "loss_xval": 0.31640625,
      "num_input_tokens_seen": 387298652,
      "step": 6912
    },
    {
      "epoch": 15.396436525612472,
      "grad_norm": 18.725072860717773,
      "learning_rate": 1e-06,
      "loss": 0.5747,
      "num_input_tokens_seen": 387353036,
      "step": 6913
    },
    {
      "epoch": 15.396436525612472,
      "loss": 0.6614874601364136,
      "loss_ce": 0.00011045205610571429,
      "loss_iou": 0.28515625,
      "loss_num": 0.01806640625,
      "loss_xval": 0.66015625,
      "num_input_tokens_seen": 387353036,
      "step": 6913
    },
    {
      "epoch": 15.398663697104677,
      "grad_norm": 17.18587875366211,
      "learning_rate": 1e-06,
      "loss": 0.3727,
      "num_input_tokens_seen": 387408840,
      "step": 6914
    },
    {
      "epoch": 15.398663697104677,
      "loss": 0.35583341121673584,
      "loss_ce": 0.00012054783292114735,
      "loss_iou": 0.16796875,
      "loss_num": 0.00390625,
      "loss_xval": 0.35546875,
      "num_input_tokens_seen": 387408840,
      "step": 6914
    },
    {
      "epoch": 15.400890868596882,
      "grad_norm": 33.79663848876953,
      "learning_rate": 1e-06,
      "loss": 0.4239,
      "num_input_tokens_seen": 387464996,
      "step": 6915
    },
    {
      "epoch": 15.400890868596882,
      "loss": 0.4479781687259674,
      "loss_ce": 0.00010222237324342132,
      "loss_iou": 0.2041015625,
      "loss_num": 0.007781982421875,
      "loss_xval": 0.447265625,
      "num_input_tokens_seen": 387464996,
      "step": 6915
    },
    {
      "epoch": 15.403118040089087,
      "grad_norm": 28.106069564819336,
      "learning_rate": 1e-06,
      "loss": 0.6411,
      "num_input_tokens_seen": 387520852,
      "step": 6916
    },
    {
      "epoch": 15.403118040089087,
      "loss": 0.5481947660446167,
      "loss_ce": 9.903394675347954e-05,
      "loss_iou": 0.234375,
      "loss_num": 0.015869140625,
      "loss_xval": 0.546875,
      "num_input_tokens_seen": 387520852,
      "step": 6916
    },
    {
      "epoch": 15.405345211581292,
      "grad_norm": 14.857207298278809,
      "learning_rate": 1e-06,
      "loss": 0.4853,
      "num_input_tokens_seen": 387573816,
      "step": 6917
    },
    {
      "epoch": 15.405345211581292,
      "loss": 0.49757710099220276,
      "loss_ce": 0.00011006787826772779,
      "loss_iou": 0.2138671875,
      "loss_num": 0.01385498046875,
      "loss_xval": 0.498046875,
      "num_input_tokens_seen": 387573816,
      "step": 6917
    },
    {
      "epoch": 15.407572383073497,
      "grad_norm": 25.532848358154297,
      "learning_rate": 1e-06,
      "loss": 0.3972,
      "num_input_tokens_seen": 387631328,
      "step": 6918
    },
    {
      "epoch": 15.407572383073497,
      "loss": 0.32285118103027344,
      "loss_ce": 9.72603156697005e-05,
      "loss_iou": 0.146484375,
      "loss_num": 0.00604248046875,
      "loss_xval": 0.322265625,
      "num_input_tokens_seen": 387631328,
      "step": 6918
    },
    {
      "epoch": 15.409799554565701,
      "grad_norm": 32.86465072631836,
      "learning_rate": 1e-06,
      "loss": 0.5478,
      "num_input_tokens_seen": 387688500,
      "step": 6919
    },
    {
      "epoch": 15.409799554565701,
      "loss": 0.7613215446472168,
      "loss_ce": 9.104014316108078e-05,
      "loss_iou": 0.314453125,
      "loss_num": 0.0262451171875,
      "loss_xval": 0.76171875,
      "num_input_tokens_seen": 387688500,
      "step": 6919
    },
    {
      "epoch": 15.412026726057906,
      "grad_norm": 30.64493179321289,
      "learning_rate": 1e-06,
      "loss": 0.4451,
      "num_input_tokens_seen": 387743892,
      "step": 6920
    },
    {
      "epoch": 15.412026726057906,
      "loss": 0.4576044976711273,
      "loss_ce": 8.494246867485344e-05,
      "loss_iou": 0.1904296875,
      "loss_num": 0.015380859375,
      "loss_xval": 0.45703125,
      "num_input_tokens_seen": 387743892,
      "step": 6920
    },
    {
      "epoch": 15.414253897550111,
      "grad_norm": 21.50493621826172,
      "learning_rate": 1e-06,
      "loss": 0.6023,
      "num_input_tokens_seen": 387800900,
      "step": 6921
    },
    {
      "epoch": 15.414253897550111,
      "loss": 0.6438992023468018,
      "loss_ce": 0.00010035550076281652,
      "loss_iou": 0.28125,
      "loss_num": 0.0166015625,
      "loss_xval": 0.64453125,
      "num_input_tokens_seen": 387800900,
      "step": 6921
    },
    {
      "epoch": 15.416481069042316,
      "grad_norm": 20.963998794555664,
      "learning_rate": 1e-06,
      "loss": 0.3538,
      "num_input_tokens_seen": 387858872,
      "step": 6922
    },
    {
      "epoch": 15.416481069042316,
      "loss": 0.4122142493724823,
      "loss_ce": 0.00010487253894098103,
      "loss_iou": 0.1796875,
      "loss_num": 0.01068115234375,
      "loss_xval": 0.412109375,
      "num_input_tokens_seen": 387858872,
      "step": 6922
    },
    {
      "epoch": 15.41870824053452,
      "grad_norm": 20.77920913696289,
      "learning_rate": 1e-06,
      "loss": 0.4124,
      "num_input_tokens_seen": 387913556,
      "step": 6923
    },
    {
      "epoch": 15.41870824053452,
      "loss": 0.2547425329685211,
      "loss_ce": 0.00022592084133066237,
      "loss_iou": 0.11181640625,
      "loss_num": 0.006134033203125,
      "loss_xval": 0.25390625,
      "num_input_tokens_seen": 387913556,
      "step": 6923
    },
    {
      "epoch": 15.420935412026726,
      "grad_norm": 14.842677116394043,
      "learning_rate": 1e-06,
      "loss": 0.3154,
      "num_input_tokens_seen": 387967992,
      "step": 6924
    },
    {
      "epoch": 15.420935412026726,
      "loss": 0.35966259241104126,
      "loss_ce": 0.00010450358968228102,
      "loss_iou": 0.16015625,
      "loss_num": 0.0078125,
      "loss_xval": 0.359375,
      "num_input_tokens_seen": 387967992,
      "step": 6924
    },
    {
      "epoch": 15.42316258351893,
      "grad_norm": 19.513763427734375,
      "learning_rate": 1e-06,
      "loss": 0.449,
      "num_input_tokens_seen": 388025440,
      "step": 6925
    },
    {
      "epoch": 15.42316258351893,
      "loss": 0.4793444275856018,
      "loss_ce": 9.635625610826537e-05,
      "loss_iou": 0.203125,
      "loss_num": 0.0147705078125,
      "loss_xval": 0.478515625,
      "num_input_tokens_seen": 388025440,
      "step": 6925
    },
    {
      "epoch": 15.425389755011135,
      "grad_norm": 14.279714584350586,
      "learning_rate": 1e-06,
      "loss": 0.4925,
      "num_input_tokens_seen": 388082980,
      "step": 6926
    },
    {
      "epoch": 15.425389755011135,
      "loss": 0.5041942596435547,
      "loss_ce": 0.00041012922883965075,
      "loss_iou": 0.2041015625,
      "loss_num": 0.0189208984375,
      "loss_xval": 0.50390625,
      "num_input_tokens_seen": 388082980,
      "step": 6926
    },
    {
      "epoch": 15.42761692650334,
      "grad_norm": 44.215213775634766,
      "learning_rate": 1e-06,
      "loss": 0.4468,
      "num_input_tokens_seen": 388138776,
      "step": 6927
    },
    {
      "epoch": 15.42761692650334,
      "loss": 0.4352813959121704,
      "loss_ce": 0.00010072036093333736,
      "loss_iou": 0.1884765625,
      "loss_num": 0.01153564453125,
      "loss_xval": 0.435546875,
      "num_input_tokens_seen": 388138776,
      "step": 6927
    },
    {
      "epoch": 15.429844097995545,
      "grad_norm": 14.542059898376465,
      "learning_rate": 1e-06,
      "loss": 0.428,
      "num_input_tokens_seen": 388193644,
      "step": 6928
    },
    {
      "epoch": 15.429844097995545,
      "loss": 0.5707679986953735,
      "loss_ce": 8.928313036449254e-05,
      "loss_iou": 0.2421875,
      "loss_num": 0.01708984375,
      "loss_xval": 0.5703125,
      "num_input_tokens_seen": 388193644,
      "step": 6928
    },
    {
      "epoch": 15.43207126948775,
      "grad_norm": 21.284255981445312,
      "learning_rate": 1e-06,
      "loss": 0.5289,
      "num_input_tokens_seen": 388250780,
      "step": 6929
    },
    {
      "epoch": 15.43207126948775,
      "loss": 0.6608084440231323,
      "loss_ce": 0.0001029052073135972,
      "loss_iou": 0.287109375,
      "loss_num": 0.0172119140625,
      "loss_xval": 0.66015625,
      "num_input_tokens_seen": 388250780,
      "step": 6929
    },
    {
      "epoch": 15.434298440979955,
      "grad_norm": 11.952805519104004,
      "learning_rate": 1e-06,
      "loss": 0.3343,
      "num_input_tokens_seen": 388305896,
      "step": 6930
    },
    {
      "epoch": 15.434298440979955,
      "loss": 0.4233115315437317,
      "loss_ce": 9.375169611303136e-05,
      "loss_iou": 0.1650390625,
      "loss_num": 0.018310546875,
      "loss_xval": 0.423828125,
      "num_input_tokens_seen": 388305896,
      "step": 6930
    },
    {
      "epoch": 15.43652561247216,
      "grad_norm": 21.607667922973633,
      "learning_rate": 1e-06,
      "loss": 0.4612,
      "num_input_tokens_seen": 388362324,
      "step": 6931
    },
    {
      "epoch": 15.43652561247216,
      "loss": 0.3633829951286316,
      "loss_ce": 0.0001628020836506039,
      "loss_iou": 0.1650390625,
      "loss_num": 0.006500244140625,
      "loss_xval": 0.36328125,
      "num_input_tokens_seen": 388362324,
      "step": 6931
    },
    {
      "epoch": 15.438752783964365,
      "grad_norm": 26.418800354003906,
      "learning_rate": 1e-06,
      "loss": 0.4414,
      "num_input_tokens_seen": 388417860,
      "step": 6932
    },
    {
      "epoch": 15.438752783964365,
      "loss": 0.38741356134414673,
      "loss_ce": 8.447450818493962e-05,
      "loss_iou": 0.1728515625,
      "loss_num": 0.00848388671875,
      "loss_xval": 0.38671875,
      "num_input_tokens_seen": 388417860,
      "step": 6932
    },
    {
      "epoch": 15.44097995545657,
      "grad_norm": 21.265588760375977,
      "learning_rate": 1e-06,
      "loss": 0.4136,
      "num_input_tokens_seen": 388472732,
      "step": 6933
    },
    {
      "epoch": 15.44097995545657,
      "loss": 0.42595887184143066,
      "loss_ce": 0.00011659698066068813,
      "loss_iou": 0.201171875,
      "loss_num": 0.004638671875,
      "loss_xval": 0.42578125,
      "num_input_tokens_seen": 388472732,
      "step": 6933
    },
    {
      "epoch": 15.443207126948774,
      "grad_norm": 18.719799041748047,
      "learning_rate": 1e-06,
      "loss": 0.4502,
      "num_input_tokens_seen": 388531568,
      "step": 6934
    },
    {
      "epoch": 15.443207126948774,
      "loss": 0.4275781214237213,
      "loss_ce": 8.788651030045003e-05,
      "loss_iou": 0.181640625,
      "loss_num": 0.01263427734375,
      "loss_xval": 0.427734375,
      "num_input_tokens_seen": 388531568,
      "step": 6934
    },
    {
      "epoch": 15.44543429844098,
      "grad_norm": 13.221081733703613,
      "learning_rate": 1e-06,
      "loss": 0.4555,
      "num_input_tokens_seen": 388587640,
      "step": 6935
    },
    {
      "epoch": 15.44543429844098,
      "loss": 0.4827903211116791,
      "loss_ce": 0.0001243370061274618,
      "loss_iou": 0.2119140625,
      "loss_num": 0.011962890625,
      "loss_xval": 0.482421875,
      "num_input_tokens_seen": 388587640,
      "step": 6935
    },
    {
      "epoch": 15.447661469933184,
      "grad_norm": 15.271649360656738,
      "learning_rate": 1e-06,
      "loss": 0.3251,
      "num_input_tokens_seen": 388645528,
      "step": 6936
    },
    {
      "epoch": 15.447661469933184,
      "loss": 0.35192036628723145,
      "loss_ce": 8.32095684017986e-05,
      "loss_iou": 0.154296875,
      "loss_num": 0.0087890625,
      "loss_xval": 0.3515625,
      "num_input_tokens_seen": 388645528,
      "step": 6936
    },
    {
      "epoch": 15.449888641425389,
      "grad_norm": 18.44422149658203,
      "learning_rate": 1e-06,
      "loss": 0.5482,
      "num_input_tokens_seen": 388698432,
      "step": 6937
    },
    {
      "epoch": 15.449888641425389,
      "loss": 0.569794774055481,
      "loss_ce": 9.262157982448116e-05,
      "loss_iou": 0.2578125,
      "loss_num": 0.010986328125,
      "loss_xval": 0.5703125,
      "num_input_tokens_seen": 388698432,
      "step": 6937
    },
    {
      "epoch": 15.452115812917596,
      "grad_norm": 23.565710067749023,
      "learning_rate": 1e-06,
      "loss": 0.3753,
      "num_input_tokens_seen": 388757224,
      "step": 6938
    },
    {
      "epoch": 15.452115812917596,
      "loss": 0.30654728412628174,
      "loss_ce": 8.977011020760983e-05,
      "loss_iou": 0.140625,
      "loss_num": 0.00518798828125,
      "loss_xval": 0.306640625,
      "num_input_tokens_seen": 388757224,
      "step": 6938
    },
    {
      "epoch": 15.4543429844098,
      "grad_norm": 13.924105644226074,
      "learning_rate": 1e-06,
      "loss": 0.3849,
      "num_input_tokens_seen": 388812656,
      "step": 6939
    },
    {
      "epoch": 15.4543429844098,
      "loss": 0.21841883659362793,
      "loss_ce": 9.608666005078703e-05,
      "loss_iou": 0.09619140625,
      "loss_num": 0.005157470703125,
      "loss_xval": 0.21875,
      "num_input_tokens_seen": 388812656,
      "step": 6939
    },
    {
      "epoch": 15.456570155902005,
      "grad_norm": 21.45891571044922,
      "learning_rate": 1e-06,
      "loss": 0.432,
      "num_input_tokens_seen": 388867552,
      "step": 6940
    },
    {
      "epoch": 15.456570155902005,
      "loss": 0.4227088689804077,
      "loss_ce": 0.00010145263513550162,
      "loss_iou": 0.1767578125,
      "loss_num": 0.013671875,
      "loss_xval": 0.421875,
      "num_input_tokens_seen": 388867552,
      "step": 6940
    },
    {
      "epoch": 15.45879732739421,
      "grad_norm": 19.371652603149414,
      "learning_rate": 1e-06,
      "loss": 0.631,
      "num_input_tokens_seen": 388923044,
      "step": 6941
    },
    {
      "epoch": 15.45879732739421,
      "loss": 0.6530581712722778,
      "loss_ce": 0.00010402440966572613,
      "loss_iou": 0.29296875,
      "loss_num": 0.01324462890625,
      "loss_xval": 0.65234375,
      "num_input_tokens_seen": 388923044,
      "step": 6941
    },
    {
      "epoch": 15.461024498886415,
      "grad_norm": 24.440471649169922,
      "learning_rate": 1e-06,
      "loss": 0.425,
      "num_input_tokens_seen": 388976632,
      "step": 6942
    },
    {
      "epoch": 15.461024498886415,
      "loss": 0.4324702322483063,
      "loss_ce": 9.719458466861397e-05,
      "loss_iou": 0.1904296875,
      "loss_num": 0.010498046875,
      "loss_xval": 0.431640625,
      "num_input_tokens_seen": 388976632,
      "step": 6942
    },
    {
      "epoch": 15.46325167037862,
      "grad_norm": 18.36454963684082,
      "learning_rate": 1e-06,
      "loss": 0.3942,
      "num_input_tokens_seen": 389031268,
      "step": 6943
    },
    {
      "epoch": 15.46325167037862,
      "loss": 0.39659687876701355,
      "loss_ce": 0.00011250589159317315,
      "loss_iou": 0.177734375,
      "loss_num": 0.00823974609375,
      "loss_xval": 0.396484375,
      "num_input_tokens_seen": 389031268,
      "step": 6943
    },
    {
      "epoch": 15.465478841870825,
      "grad_norm": 17.621511459350586,
      "learning_rate": 1e-06,
      "loss": 0.4569,
      "num_input_tokens_seen": 389089160,
      "step": 6944
    },
    {
      "epoch": 15.465478841870825,
      "loss": 0.33285778760910034,
      "loss_ce": 9.410581697011366e-05,
      "loss_iou": 0.15625,
      "loss_num": 0.00421142578125,
      "loss_xval": 0.33203125,
      "num_input_tokens_seen": 389089160,
      "step": 6944
    },
    {
      "epoch": 15.46770601336303,
      "grad_norm": 16.278667449951172,
      "learning_rate": 1e-06,
      "loss": 0.3584,
      "num_input_tokens_seen": 389144868,
      "step": 6945
    },
    {
      "epoch": 15.46770601336303,
      "loss": 0.3341924548149109,
      "loss_ce": 8.60134678077884e-05,
      "loss_iou": 0.14453125,
      "loss_num": 0.009033203125,
      "loss_xval": 0.333984375,
      "num_input_tokens_seen": 389144868,
      "step": 6945
    },
    {
      "epoch": 15.469933184855234,
      "grad_norm": 14.054006576538086,
      "learning_rate": 1e-06,
      "loss": 0.4937,
      "num_input_tokens_seen": 389202724,
      "step": 6946
    },
    {
      "epoch": 15.469933184855234,
      "loss": 0.4199259281158447,
      "loss_ce": 0.00012615637388080359,
      "loss_iou": 0.1689453125,
      "loss_num": 0.016357421875,
      "loss_xval": 0.419921875,
      "num_input_tokens_seen": 389202724,
      "step": 6946
    },
    {
      "epoch": 15.47216035634744,
      "grad_norm": 16.913305282592773,
      "learning_rate": 1e-06,
      "loss": 0.5997,
      "num_input_tokens_seen": 389260720,
      "step": 6947
    },
    {
      "epoch": 15.47216035634744,
      "loss": 0.46518251299858093,
      "loss_ce": 9.461917215958238e-05,
      "loss_iou": 0.203125,
      "loss_num": 0.0115966796875,
      "loss_xval": 0.46484375,
      "num_input_tokens_seen": 389260720,
      "step": 6947
    },
    {
      "epoch": 15.474387527839644,
      "grad_norm": 18.17692756652832,
      "learning_rate": 1e-06,
      "loss": 0.5238,
      "num_input_tokens_seen": 389315804,
      "step": 6948
    },
    {
      "epoch": 15.474387527839644,
      "loss": 0.5708150863647461,
      "loss_ce": 7.536027260357514e-05,
      "loss_iou": 0.244140625,
      "loss_num": 0.0166015625,
      "loss_xval": 0.5703125,
      "num_input_tokens_seen": 389315804,
      "step": 6948
    },
    {
      "epoch": 15.476614699331849,
      "grad_norm": 30.645221710205078,
      "learning_rate": 1e-06,
      "loss": 0.393,
      "num_input_tokens_seen": 389373028,
      "step": 6949
    },
    {
      "epoch": 15.476614699331849,
      "loss": 0.37821051478385925,
      "loss_ce": 9.771598706720397e-05,
      "loss_iou": 0.162109375,
      "loss_num": 0.01080322265625,
      "loss_xval": 0.37890625,
      "num_input_tokens_seen": 389373028,
      "step": 6949
    },
    {
      "epoch": 15.478841870824054,
      "grad_norm": 23.542387008666992,
      "learning_rate": 1e-06,
      "loss": 0.5772,
      "num_input_tokens_seen": 389429224,
      "step": 6950
    },
    {
      "epoch": 15.478841870824054,
      "loss": 0.7504802942276001,
      "loss_ce": 0.001334758591838181,
      "loss_iou": 0.306640625,
      "loss_num": 0.0272216796875,
      "loss_xval": 0.75,
      "num_input_tokens_seen": 389429224,
      "step": 6950
    },
    {
      "epoch": 15.481069042316259,
      "grad_norm": 27.925491333007812,
      "learning_rate": 1e-06,
      "loss": 0.3852,
      "num_input_tokens_seen": 389486864,
      "step": 6951
    },
    {
      "epoch": 15.481069042316259,
      "loss": 0.48157191276550293,
      "loss_ce": 0.00012661112123169005,
      "loss_iou": 0.166015625,
      "loss_num": 0.029541015625,
      "loss_xval": 0.48046875,
      "num_input_tokens_seen": 389486864,
      "step": 6951
    },
    {
      "epoch": 15.483296213808464,
      "grad_norm": 15.1088285446167,
      "learning_rate": 1e-06,
      "loss": 0.3681,
      "num_input_tokens_seen": 389542096,
      "step": 6952
    },
    {
      "epoch": 15.483296213808464,
      "loss": 0.41208964586257935,
      "loss_ce": 0.00010235629451926798,
      "loss_iou": 0.1943359375,
      "loss_num": 0.004608154296875,
      "loss_xval": 0.412109375,
      "num_input_tokens_seen": 389542096,
      "step": 6952
    },
    {
      "epoch": 15.485523385300668,
      "grad_norm": 20.11035919189453,
      "learning_rate": 1e-06,
      "loss": 0.5138,
      "num_input_tokens_seen": 389596724,
      "step": 6953
    },
    {
      "epoch": 15.485523385300668,
      "loss": 0.37680959701538086,
      "loss_ce": 0.00010061028297059238,
      "loss_iou": 0.1669921875,
      "loss_num": 0.0086669921875,
      "loss_xval": 0.376953125,
      "num_input_tokens_seen": 389596724,
      "step": 6953
    },
    {
      "epoch": 15.487750556792873,
      "grad_norm": 13.750123977661133,
      "learning_rate": 1e-06,
      "loss": 0.5883,
      "num_input_tokens_seen": 389652700,
      "step": 6954
    },
    {
      "epoch": 15.487750556792873,
      "loss": 0.6634094715118408,
      "loss_ce": 0.0001404019130859524,
      "loss_iou": 0.26171875,
      "loss_num": 0.028076171875,
      "loss_xval": 0.6640625,
      "num_input_tokens_seen": 389652700,
      "step": 6954
    },
    {
      "epoch": 15.489977728285078,
      "grad_norm": 13.668987274169922,
      "learning_rate": 1e-06,
      "loss": 0.3403,
      "num_input_tokens_seen": 389710304,
      "step": 6955
    },
    {
      "epoch": 15.489977728285078,
      "loss": 0.33821815252304077,
      "loss_ce": 8.338829502463341e-05,
      "loss_iou": 0.1484375,
      "loss_num": 0.00811767578125,
      "loss_xval": 0.337890625,
      "num_input_tokens_seen": 389710304,
      "step": 6955
    },
    {
      "epoch": 15.492204899777283,
      "grad_norm": 20.135528564453125,
      "learning_rate": 1e-06,
      "loss": 0.4608,
      "num_input_tokens_seen": 389764472,
      "step": 6956
    },
    {
      "epoch": 15.492204899777283,
      "loss": 0.46713533997535706,
      "loss_ce": 9.433674858883023e-05,
      "loss_iou": 0.208984375,
      "loss_num": 0.01007080078125,
      "loss_xval": 0.466796875,
      "num_input_tokens_seen": 389764472,
      "step": 6956
    },
    {
      "epoch": 15.494432071269488,
      "grad_norm": 19.5330867767334,
      "learning_rate": 1e-06,
      "loss": 0.3886,
      "num_input_tokens_seen": 389820184,
      "step": 6957
    },
    {
      "epoch": 15.494432071269488,
      "loss": 0.46810320019721985,
      "loss_ce": 8.56061524245888e-05,
      "loss_iou": 0.2099609375,
      "loss_num": 0.00958251953125,
      "loss_xval": 0.46875,
      "num_input_tokens_seen": 389820184,
      "step": 6957
    },
    {
      "epoch": 15.496659242761693,
      "grad_norm": 15.649205207824707,
      "learning_rate": 1e-06,
      "loss": 0.3468,
      "num_input_tokens_seen": 389875816,
      "step": 6958
    },
    {
      "epoch": 15.496659242761693,
      "loss": 0.3824237287044525,
      "loss_ce": 9.9498953204602e-05,
      "loss_iou": 0.1787109375,
      "loss_num": 0.005157470703125,
      "loss_xval": 0.3828125,
      "num_input_tokens_seen": 389875816,
      "step": 6958
    },
    {
      "epoch": 15.498886414253898,
      "grad_norm": 17.48607063293457,
      "learning_rate": 1e-06,
      "loss": 0.4428,
      "num_input_tokens_seen": 389934296,
      "step": 6959
    },
    {
      "epoch": 15.498886414253898,
      "loss": 0.45505648851394653,
      "loss_ce": 0.00010044968803413212,
      "loss_iou": 0.1953125,
      "loss_num": 0.0126953125,
      "loss_xval": 0.455078125,
      "num_input_tokens_seen": 389934296,
      "step": 6959
    },
    {
      "epoch": 15.501113585746102,
      "grad_norm": 23.723533630371094,
      "learning_rate": 1e-06,
      "loss": 0.3997,
      "num_input_tokens_seen": 389988784,
      "step": 6960
    },
    {
      "epoch": 15.501113585746102,
      "loss": 0.5000880360603333,
      "loss_ce": 8.804388926364481e-05,
      "loss_iou": 0.224609375,
      "loss_num": 0.01025390625,
      "loss_xval": 0.5,
      "num_input_tokens_seen": 389988784,
      "step": 6960
    },
    {
      "epoch": 15.503340757238307,
      "grad_norm": 24.47084617614746,
      "learning_rate": 1e-06,
      "loss": 0.3908,
      "num_input_tokens_seen": 390042176,
      "step": 6961
    },
    {
      "epoch": 15.503340757238307,
      "loss": 0.3538658022880554,
      "loss_ce": 0.00010604046110529453,
      "loss_iou": 0.1650390625,
      "loss_num": 0.00494384765625,
      "loss_xval": 0.353515625,
      "num_input_tokens_seen": 390042176,
      "step": 6961
    },
    {
      "epoch": 15.505567928730512,
      "grad_norm": 19.05753517150879,
      "learning_rate": 1e-06,
      "loss": 0.3267,
      "num_input_tokens_seen": 390098484,
      "step": 6962
    },
    {
      "epoch": 15.505567928730512,
      "loss": 0.3786473274230957,
      "loss_ce": 0.00010729986388469115,
      "loss_iou": 0.166015625,
      "loss_num": 0.00921630859375,
      "loss_xval": 0.37890625,
      "num_input_tokens_seen": 390098484,
      "step": 6962
    },
    {
      "epoch": 15.507795100222717,
      "grad_norm": 10.61585807800293,
      "learning_rate": 1e-06,
      "loss": 0.3648,
      "num_input_tokens_seen": 390156300,
      "step": 6963
    },
    {
      "epoch": 15.507795100222717,
      "loss": 0.378567099571228,
      "loss_ce": 8.807641279418021e-05,
      "loss_iou": 0.16015625,
      "loss_num": 0.01165771484375,
      "loss_xval": 0.37890625,
      "num_input_tokens_seen": 390156300,
      "step": 6963
    },
    {
      "epoch": 15.510022271714922,
      "grad_norm": 25.818944931030273,
      "learning_rate": 1e-06,
      "loss": 0.3777,
      "num_input_tokens_seen": 390210808,
      "step": 6964
    },
    {
      "epoch": 15.510022271714922,
      "loss": 0.42782527208328247,
      "loss_ce": 9.0899906354025e-05,
      "loss_iou": 0.1708984375,
      "loss_num": 0.017333984375,
      "loss_xval": 0.427734375,
      "num_input_tokens_seen": 390210808,
      "step": 6964
    },
    {
      "epoch": 15.512249443207127,
      "grad_norm": 17.772178649902344,
      "learning_rate": 1e-06,
      "loss": 0.3301,
      "num_input_tokens_seen": 390266472,
      "step": 6965
    },
    {
      "epoch": 15.512249443207127,
      "loss": 0.40292441844940186,
      "loss_ce": 9.23790066735819e-05,
      "loss_iou": 0.1865234375,
      "loss_num": 0.006011962890625,
      "loss_xval": 0.40234375,
      "num_input_tokens_seen": 390266472,
      "step": 6965
    },
    {
      "epoch": 15.514476614699332,
      "grad_norm": 29.70581817626953,
      "learning_rate": 1e-06,
      "loss": 0.4923,
      "num_input_tokens_seen": 390323712,
      "step": 6966
    },
    {
      "epoch": 15.514476614699332,
      "loss": 0.4616038203239441,
      "loss_ce": 0.00011701375478878617,
      "loss_iou": 0.208984375,
      "loss_num": 0.008544921875,
      "loss_xval": 0.4609375,
      "num_input_tokens_seen": 390323712,
      "step": 6966
    },
    {
      "epoch": 15.516703786191536,
      "grad_norm": 50.45325469970703,
      "learning_rate": 1e-06,
      "loss": 0.3747,
      "num_input_tokens_seen": 390380440,
      "step": 6967
    },
    {
      "epoch": 15.516703786191536,
      "loss": 0.3414144814014435,
      "loss_ce": 0.00010588267468847334,
      "loss_iou": 0.150390625,
      "loss_num": 0.00811767578125,
      "loss_xval": 0.341796875,
      "num_input_tokens_seen": 390380440,
      "step": 6967
    },
    {
      "epoch": 15.518930957683741,
      "grad_norm": 18.55625343322754,
      "learning_rate": 1e-06,
      "loss": 0.4821,
      "num_input_tokens_seen": 390437412,
      "step": 6968
    },
    {
      "epoch": 15.518930957683741,
      "loss": 0.48657822608947754,
      "loss_ce": 0.00012801533739548177,
      "loss_iou": 0.2265625,
      "loss_num": 0.006378173828125,
      "loss_xval": 0.486328125,
      "num_input_tokens_seen": 390437412,
      "step": 6968
    },
    {
      "epoch": 15.521158129175946,
      "grad_norm": 25.324066162109375,
      "learning_rate": 1e-06,
      "loss": 0.4971,
      "num_input_tokens_seen": 390491840,
      "step": 6969
    },
    {
      "epoch": 15.521158129175946,
      "loss": 0.536230206489563,
      "loss_ce": 9.739406232256442e-05,
      "loss_iou": 0.23046875,
      "loss_num": 0.01483154296875,
      "loss_xval": 0.53515625,
      "num_input_tokens_seen": 390491840,
      "step": 6969
    },
    {
      "epoch": 15.523385300668151,
      "grad_norm": 18.16643714904785,
      "learning_rate": 1e-06,
      "loss": 0.4849,
      "num_input_tokens_seen": 390546924,
      "step": 6970
    },
    {
      "epoch": 15.523385300668151,
      "loss": 0.54595947265625,
      "loss_ce": 9.153402788797393e-05,
      "loss_iou": 0.2431640625,
      "loss_num": 0.01153564453125,
      "loss_xval": 0.546875,
      "num_input_tokens_seen": 390546924,
      "step": 6970
    },
    {
      "epoch": 15.525612472160356,
      "grad_norm": 15.818894386291504,
      "learning_rate": 1e-06,
      "loss": 0.5592,
      "num_input_tokens_seen": 390603308,
      "step": 6971
    },
    {
      "epoch": 15.525612472160356,
      "loss": 0.597541093826294,
      "loss_ce": 0.00012892311497125775,
      "loss_iou": 0.265625,
      "loss_num": 0.0135498046875,
      "loss_xval": 0.59765625,
      "num_input_tokens_seen": 390603308,
      "step": 6971
    },
    {
      "epoch": 15.52783964365256,
      "grad_norm": 30.140596389770508,
      "learning_rate": 1e-06,
      "loss": 0.5127,
      "num_input_tokens_seen": 390657280,
      "step": 6972
    },
    {
      "epoch": 15.52783964365256,
      "loss": 0.6534337401390076,
      "loss_ce": 0.00011343492224114016,
      "loss_iou": 0.294921875,
      "loss_num": 0.01251220703125,
      "loss_xval": 0.65234375,
      "num_input_tokens_seen": 390657280,
      "step": 6972
    },
    {
      "epoch": 15.530066815144766,
      "grad_norm": 19.85612678527832,
      "learning_rate": 1e-06,
      "loss": 0.5059,
      "num_input_tokens_seen": 390714956,
      "step": 6973
    },
    {
      "epoch": 15.530066815144766,
      "loss": 0.764450192451477,
      "loss_ce": 0.00010690266208257526,
      "loss_iou": 0.34765625,
      "loss_num": 0.01385498046875,
      "loss_xval": 0.765625,
      "num_input_tokens_seen": 390714956,
      "step": 6973
    },
    {
      "epoch": 15.53229398663697,
      "grad_norm": 12.298739433288574,
      "learning_rate": 1e-06,
      "loss": 0.5395,
      "num_input_tokens_seen": 390772920,
      "step": 6974
    },
    {
      "epoch": 15.53229398663697,
      "loss": 0.7317103147506714,
      "loss_ce": 0.00026501319371163845,
      "loss_iou": 0.296875,
      "loss_num": 0.0277099609375,
      "loss_xval": 0.73046875,
      "num_input_tokens_seen": 390772920,
      "step": 6974
    },
    {
      "epoch": 15.534521158129175,
      "grad_norm": 22.31592559814453,
      "learning_rate": 1e-06,
      "loss": 0.4543,
      "num_input_tokens_seen": 390826668,
      "step": 6975
    },
    {
      "epoch": 15.534521158129175,
      "loss": 0.5942932367324829,
      "loss_ce": 0.00017704666242934763,
      "loss_iou": 0.267578125,
      "loss_num": 0.01165771484375,
      "loss_xval": 0.59375,
      "num_input_tokens_seen": 390826668,
      "step": 6975
    },
    {
      "epoch": 15.53674832962138,
      "grad_norm": 18.439233779907227,
      "learning_rate": 1e-06,
      "loss": 0.4188,
      "num_input_tokens_seen": 390882348,
      "step": 6976
    },
    {
      "epoch": 15.53674832962138,
      "loss": 0.2939029932022095,
      "loss_ce": 9.499942825641483e-05,
      "loss_iou": 0.1318359375,
      "loss_num": 0.006103515625,
      "loss_xval": 0.29296875,
      "num_input_tokens_seen": 390882348,
      "step": 6976
    },
    {
      "epoch": 15.538975501113585,
      "grad_norm": 16.87866973876953,
      "learning_rate": 1e-06,
      "loss": 0.3277,
      "num_input_tokens_seen": 390939420,
      "step": 6977
    },
    {
      "epoch": 15.538975501113585,
      "loss": 0.36519938707351685,
      "loss_ce": 8.707845699973404e-05,
      "loss_iou": 0.15234375,
      "loss_num": 0.01214599609375,
      "loss_xval": 0.365234375,
      "num_input_tokens_seen": 390939420,
      "step": 6977
    },
    {
      "epoch": 15.54120267260579,
      "grad_norm": 14.497117042541504,
      "learning_rate": 1e-06,
      "loss": 0.3234,
      "num_input_tokens_seen": 390992964,
      "step": 6978
    },
    {
      "epoch": 15.54120267260579,
      "loss": 0.36044928431510925,
      "loss_ce": 9.772789780981839e-05,
      "loss_iou": 0.171875,
      "loss_num": 0.00347900390625,
      "loss_xval": 0.359375,
      "num_input_tokens_seen": 390992964,
      "step": 6978
    },
    {
      "epoch": 15.543429844097995,
      "grad_norm": 18.7347412109375,
      "learning_rate": 1e-06,
      "loss": 0.3939,
      "num_input_tokens_seen": 391049216,
      "step": 6979
    },
    {
      "epoch": 15.543429844097995,
      "loss": 0.4501742124557495,
      "loss_ce": 0.00010095632751472294,
      "loss_iou": 0.19921875,
      "loss_num": 0.01043701171875,
      "loss_xval": 0.44921875,
      "num_input_tokens_seen": 391049216,
      "step": 6979
    },
    {
      "epoch": 15.5456570155902,
      "grad_norm": 18.551483154296875,
      "learning_rate": 1e-06,
      "loss": 0.4831,
      "num_input_tokens_seen": 391105972,
      "step": 6980
    },
    {
      "epoch": 15.5456570155902,
      "loss": 0.5650371313095093,
      "loss_ce": 9.57687952904962e-05,
      "loss_iou": 0.255859375,
      "loss_num": 0.0106201171875,
      "loss_xval": 0.56640625,
      "num_input_tokens_seen": 391105972,
      "step": 6980
    },
    {
      "epoch": 15.547884187082406,
      "grad_norm": 18.95163345336914,
      "learning_rate": 1e-06,
      "loss": 0.4954,
      "num_input_tokens_seen": 391159136,
      "step": 6981
    },
    {
      "epoch": 15.547884187082406,
      "loss": 0.5984832048416138,
      "loss_ce": 9.453899110667408e-05,
      "loss_iou": 0.2734375,
      "loss_num": 0.01025390625,
      "loss_xval": 0.59765625,
      "num_input_tokens_seen": 391159136,
      "step": 6981
    },
    {
      "epoch": 15.550111358574611,
      "grad_norm": 15.434839248657227,
      "learning_rate": 1e-06,
      "loss": 0.3884,
      "num_input_tokens_seen": 391215408,
      "step": 6982
    },
    {
      "epoch": 15.550111358574611,
      "loss": 0.46562230587005615,
      "loss_ce": 0.00010716939868871123,
      "loss_iou": 0.208984375,
      "loss_num": 0.00946044921875,
      "loss_xval": 0.46484375,
      "num_input_tokens_seen": 391215408,
      "step": 6982
    },
    {
      "epoch": 15.552338530066816,
      "grad_norm": 56.02677536010742,
      "learning_rate": 1e-06,
      "loss": 0.5136,
      "num_input_tokens_seen": 391270184,
      "step": 6983
    },
    {
      "epoch": 15.552338530066816,
      "loss": 0.5048550367355347,
      "loss_ce": 9.426505130250007e-05,
      "loss_iou": 0.2294921875,
      "loss_num": 0.00921630859375,
      "loss_xval": 0.50390625,
      "num_input_tokens_seen": 391270184,
      "step": 6983
    },
    {
      "epoch": 15.55456570155902,
      "grad_norm": 37.09487533569336,
      "learning_rate": 1e-06,
      "loss": 0.4122,
      "num_input_tokens_seen": 391323132,
      "step": 6984
    },
    {
      "epoch": 15.55456570155902,
      "loss": 0.43672704696655273,
      "loss_ce": 8.155644172802567e-05,
      "loss_iou": 0.19140625,
      "loss_num": 0.01055908203125,
      "loss_xval": 0.4375,
      "num_input_tokens_seen": 391323132,
      "step": 6984
    },
    {
      "epoch": 15.556792873051226,
      "grad_norm": 29.41912078857422,
      "learning_rate": 1e-06,
      "loss": 0.4368,
      "num_input_tokens_seen": 391378804,
      "step": 6985
    },
    {
      "epoch": 15.556792873051226,
      "loss": 0.45505693554878235,
      "loss_ce": 0.00010087570990435779,
      "loss_iou": 0.2021484375,
      "loss_num": 0.00994873046875,
      "loss_xval": 0.455078125,
      "num_input_tokens_seen": 391378804,
      "step": 6985
    },
    {
      "epoch": 15.55902004454343,
      "grad_norm": 22.739253997802734,
      "learning_rate": 1e-06,
      "loss": 0.5131,
      "num_input_tokens_seen": 391434260,
      "step": 6986
    },
    {
      "epoch": 15.55902004454343,
      "loss": 0.5328000783920288,
      "loss_ce": 8.528250327799469e-05,
      "loss_iou": 0.232421875,
      "loss_num": 0.013671875,
      "loss_xval": 0.53125,
      "num_input_tokens_seen": 391434260,
      "step": 6986
    },
    {
      "epoch": 15.561247216035635,
      "grad_norm": 15.653422355651855,
      "learning_rate": 1e-06,
      "loss": 0.5016,
      "num_input_tokens_seen": 391493312,
      "step": 6987
    },
    {
      "epoch": 15.561247216035635,
      "loss": 0.4914650619029999,
      "loss_ce": 0.0001320439187111333,
      "loss_iou": 0.22265625,
      "loss_num": 0.0093994140625,
      "loss_xval": 0.4921875,
      "num_input_tokens_seen": 391493312,
      "step": 6987
    },
    {
      "epoch": 15.56347438752784,
      "grad_norm": 21.306421279907227,
      "learning_rate": 1e-06,
      "loss": 0.5295,
      "num_input_tokens_seen": 391548072,
      "step": 6988
    },
    {
      "epoch": 15.56347438752784,
      "loss": 0.6857566833496094,
      "loss_ce": 8.777156472206116e-05,
      "loss_iou": 0.287109375,
      "loss_num": 0.0225830078125,
      "loss_xval": 0.6875,
      "num_input_tokens_seen": 391548072,
      "step": 6988
    },
    {
      "epoch": 15.565701559020045,
      "grad_norm": 18.79554557800293,
      "learning_rate": 1e-06,
      "loss": 0.5781,
      "num_input_tokens_seen": 391602092,
      "step": 6989
    },
    {
      "epoch": 15.565701559020045,
      "loss": 0.6327000856399536,
      "loss_ce": 0.00013169870362617075,
      "loss_iou": 0.279296875,
      "loss_num": 0.01470947265625,
      "loss_xval": 0.6328125,
      "num_input_tokens_seen": 391602092,
      "step": 6989
    },
    {
      "epoch": 15.56792873051225,
      "grad_norm": 19.208559036254883,
      "learning_rate": 1e-06,
      "loss": 0.4444,
      "num_input_tokens_seen": 391659220,
      "step": 6990
    },
    {
      "epoch": 15.56792873051225,
      "loss": 0.40756696462631226,
      "loss_ce": 9.625191160012037e-05,
      "loss_iou": 0.1767578125,
      "loss_num": 0.01092529296875,
      "loss_xval": 0.408203125,
      "num_input_tokens_seen": 391659220,
      "step": 6990
    },
    {
      "epoch": 15.570155902004455,
      "grad_norm": 25.114105224609375,
      "learning_rate": 1e-06,
      "loss": 0.5278,
      "num_input_tokens_seen": 391713976,
      "step": 6991
    },
    {
      "epoch": 15.570155902004455,
      "loss": 0.760166347026825,
      "loss_ce": 0.0001565346319694072,
      "loss_iou": 0.29296875,
      "loss_num": 0.034912109375,
      "loss_xval": 0.76171875,
      "num_input_tokens_seen": 391713976,
      "step": 6991
    },
    {
      "epoch": 15.57238307349666,
      "grad_norm": 19.96893310546875,
      "learning_rate": 1e-06,
      "loss": 0.4927,
      "num_input_tokens_seen": 391768912,
      "step": 6992
    },
    {
      "epoch": 15.57238307349666,
      "loss": 0.5373374223709106,
      "loss_ce": 0.00010592768376227468,
      "loss_iou": 0.244140625,
      "loss_num": 0.00994873046875,
      "loss_xval": 0.5390625,
      "num_input_tokens_seen": 391768912,
      "step": 6992
    },
    {
      "epoch": 15.574610244988865,
      "grad_norm": 16.1031494140625,
      "learning_rate": 1e-06,
      "loss": 0.6124,
      "num_input_tokens_seen": 391825536,
      "step": 6993
    },
    {
      "epoch": 15.574610244988865,
      "loss": 0.762206494808197,
      "loss_ce": 0.00012153427815064788,
      "loss_iou": 0.296875,
      "loss_num": 0.0341796875,
      "loss_xval": 0.76171875,
      "num_input_tokens_seen": 391825536,
      "step": 6993
    },
    {
      "epoch": 15.57683741648107,
      "grad_norm": 20.991905212402344,
      "learning_rate": 1e-06,
      "loss": 0.6052,
      "num_input_tokens_seen": 391879656,
      "step": 6994
    },
    {
      "epoch": 15.57683741648107,
      "loss": 0.7651537656784058,
      "loss_ce": 0.00020015102927573025,
      "loss_iou": 0.283203125,
      "loss_num": 0.039794921875,
      "loss_xval": 0.765625,
      "num_input_tokens_seen": 391879656,
      "step": 6994
    },
    {
      "epoch": 15.579064587973274,
      "grad_norm": 16.341785430908203,
      "learning_rate": 1e-06,
      "loss": 0.4781,
      "num_input_tokens_seen": 391935488,
      "step": 6995
    },
    {
      "epoch": 15.579064587973274,
      "loss": 0.5291576385498047,
      "loss_ce": 0.00010491380817256868,
      "loss_iou": 0.220703125,
      "loss_num": 0.0174560546875,
      "loss_xval": 0.52734375,
      "num_input_tokens_seen": 391935488,
      "step": 6995
    },
    {
      "epoch": 15.58129175946548,
      "grad_norm": 19.71142578125,
      "learning_rate": 1e-06,
      "loss": 0.5139,
      "num_input_tokens_seen": 391991892,
      "step": 6996
    },
    {
      "epoch": 15.58129175946548,
      "loss": 0.44407549500465393,
      "loss_ce": 0.00010574980115052313,
      "loss_iou": 0.203125,
      "loss_num": 0.00750732421875,
      "loss_xval": 0.443359375,
      "num_input_tokens_seen": 391991892,
      "step": 6996
    },
    {
      "epoch": 15.583518930957684,
      "grad_norm": 22.070716857910156,
      "learning_rate": 1e-06,
      "loss": 0.3774,
      "num_input_tokens_seen": 392046924,
      "step": 6997
    },
    {
      "epoch": 15.583518930957684,
      "loss": 0.2757876515388489,
      "loss_ce": 9.184512600768358e-05,
      "loss_iou": 0.1162109375,
      "loss_num": 0.0086669921875,
      "loss_xval": 0.275390625,
      "num_input_tokens_seen": 392046924,
      "step": 6997
    },
    {
      "epoch": 15.585746102449889,
      "grad_norm": 20.057531356811523,
      "learning_rate": 1e-06,
      "loss": 0.4721,
      "num_input_tokens_seen": 392104208,
      "step": 6998
    },
    {
      "epoch": 15.585746102449889,
      "loss": 0.5038794279098511,
      "loss_ce": 9.52341069933027e-05,
      "loss_iou": 0.2138671875,
      "loss_num": 0.0150146484375,
      "loss_xval": 0.50390625,
      "num_input_tokens_seen": 392104208,
      "step": 6998
    },
    {
      "epoch": 15.587973273942094,
      "grad_norm": 25.237815856933594,
      "learning_rate": 1e-06,
      "loss": 0.5194,
      "num_input_tokens_seen": 392161208,
      "step": 6999
    },
    {
      "epoch": 15.587973273942094,
      "loss": 0.46581047773361206,
      "loss_ce": 0.00011221040040254593,
      "loss_iou": 0.2060546875,
      "loss_num": 0.01080322265625,
      "loss_xval": 0.46484375,
      "num_input_tokens_seen": 392161208,
      "step": 6999
    },
    {
      "epoch": 15.590200445434299,
      "grad_norm": 15.100346565246582,
      "learning_rate": 1e-06,
      "loss": 0.4739,
      "num_input_tokens_seen": 392214772,
      "step": 7000
    },
    {
      "epoch": 15.590200445434299,
      "eval_seeclick_web_CIoU": 0.5901551246643066,
      "eval_seeclick_web_GIoU": 0.5888436436653137,
      "eval_seeclick_web_IoU": 0.6081466972827911,
      "eval_seeclick_web_MAE_all": 0.015473631210625172,
      "eval_seeclick_web_MAE_h": 0.0076336238998919725,
      "eval_seeclick_web_MAE_w": 0.015680983662605286,
      "eval_seeclick_web_MAE_x_boxes": 0.008724939078092575,
      "eval_seeclick_web_MAE_y_boxes": 0.021282089641317725,
      "eval_seeclick_web_inside_bbox": 0.9166666567325592,
      "eval_seeclick_web_loss": 0.907034695148468,
      "eval_seeclick_web_loss_ce": 0.00015554412675555795,
      "eval_seeclick_web_loss_iou": 0.4166259765625,
      "eval_seeclick_web_loss_num": 0.01241302490234375,
      "eval_seeclick_web_loss_xval": 0.895263671875,
      "eval_seeclick_web_runtime": 24.1637,
      "eval_seeclick_web_samples_per_second": 2.069,
      "eval_seeclick_web_steps_per_second": 0.083,
      "num_input_tokens_seen": 392214772,
      "step": 7000
    },
    {
      "epoch": 15.590200445434299,
      "eval_icons_CIoU": 0.2777545750141144,
      "eval_icons_GIoU": 0.30234697461128235,
      "eval_icons_IoU": 0.3519442528486252,
      "eval_icons_MAE_all": 0.05883444473147392,
      "eval_icons_MAE_h": 0.031461546663194895,
      "eval_icons_MAE_w": 0.058418434113264084,
      "eval_icons_MAE_x_boxes": 0.06006164848804474,
      "eval_icons_MAE_y_boxes": 0.03738272096961737,
      "eval_icons_inside_bbox": 0.59375,
      "eval_icons_loss": 1.7067599296569824,
      "eval_icons_loss_ce": 0.00019056371820624918,
      "eval_icons_loss_iou": 0.668701171875,
      "eval_icons_loss_num": 0.05805206298828125,
      "eval_icons_loss_xval": 1.628662109375,
      "eval_icons_runtime": 19.373,
      "eval_icons_samples_per_second": 2.581,
      "eval_icons_steps_per_second": 0.103,
      "num_input_tokens_seen": 392214772,
      "step": 7000
    },
    {
      "epoch": 15.590200445434299,
      "eval_screenspot_CIoU": 0.371360719203949,
      "eval_screenspot_GIoU": 0.3886072834332784,
      "eval_screenspot_IoU": 0.4512639542420705,
      "eval_screenspot_MAE_all": 0.055713951587677,
      "eval_screenspot_MAE_h": 0.039317984133958817,
      "eval_screenspot_MAE_w": 0.06650510802865028,
      "eval_screenspot_MAE_x_boxes": 0.07384055045743783,
      "eval_screenspot_MAE_y_boxes": 0.03706817328929901,
      "eval_screenspot_inside_bbox": 0.7145833373069763,
      "eval_screenspot_loss": 1.5692362785339355,
      "eval_screenspot_loss_ce": 0.00024274400008531907,
      "eval_screenspot_loss_iou": 0.6532389322916666,
      "eval_screenspot_loss_num": 0.06329091389973958,
      "eval_screenspot_loss_xval": 1.6232096354166667,
      "eval_screenspot_runtime": 37.4679,
      "eval_screenspot_samples_per_second": 2.375,
      "eval_screenspot_steps_per_second": 0.08,
      "num_input_tokens_seen": 392214772,
      "step": 7000
    },
    {
      "epoch": 15.590200445434299,
      "eval_compot_CIoU": 0.3413173258304596,
      "eval_compot_GIoU": 0.35517509281635284,
      "eval_compot_IoU": 0.401169016957283,
      "eval_compot_MAE_all": 0.019693247973918915,
      "eval_compot_MAE_h": 0.013937031384557486,
      "eval_compot_MAE_w": 0.021615090779960155,
      "eval_compot_MAE_x_boxes": 0.02992706559598446,
      "eval_compot_MAE_y_boxes": 0.0066331722773611546,
      "eval_compot_inside_bbox": 0.6302083432674408,
      "eval_compot_loss": 1.4276500940322876,
      "eval_compot_loss_ce": 0.00015015306416898966,
      "eval_compot_loss_iou": 0.6561279296875,
      "eval_compot_loss_num": 0.018407821655273438,
      "eval_compot_loss_xval": 1.404296875,
      "eval_compot_runtime": 24.1014,
      "eval_compot_samples_per_second": 2.075,
      "eval_compot_steps_per_second": 0.083,
      "num_input_tokens_seen": 392214772,
      "step": 7000
    },
    {
      "epoch": 15.590200445434299,
      "eval_custom_ui_val_CIoU": 0.4745243142048518,
      "eval_custom_ui_val_GIoU": 0.4812171955903371,
      "eval_custom_ui_val_IoU": 0.535526971022288,
      "eval_custom_ui_val_MAE_all": 0.027974836269600525,
      "eval_custom_ui_val_MAE_h": 0.015499611799087789,
      "eval_custom_ui_val_MAE_w": 0.03617638742758168,
      "eval_custom_ui_val_MAE_x_boxes": 0.033723721021993294,
      "eval_custom_ui_val_MAE_y_boxes": 0.013699741387325857,
      "eval_custom_ui_val_inside_bbox": 0.7685185207260979,
      "eval_custom_ui_val_loss": 1.1788626909255981,
      "eval_custom_ui_val_loss_ce": 0.00017328551944552196,
      "eval_custom_ui_val_loss_iou": 0.5056423611111112,
      "eval_custom_ui_val_loss_num": 0.02455192142062717,
      "eval_custom_ui_val_loss_xval": 1.1339518229166667,
      "eval_custom_ui_val_runtime": 72.2545,
      "eval_custom_ui_val_samples_per_second": 3.668,
      "eval_custom_ui_val_steps_per_second": 0.125,
      "num_input_tokens_seen": 392214772,
      "step": 7000
    },
    {
      "epoch": 15.590200445434299,
      "loss": 0.8509615063667297,
      "loss_ce": 0.0001314119144808501,
      "loss_iou": 0.3828125,
      "loss_num": 0.01708984375,
      "loss_xval": 0.8515625,
      "num_input_tokens_seen": 392214772,
      "step": 7000
    },
    {
      "epoch": 15.592427616926503,
      "grad_norm": 15.370309829711914,
      "learning_rate": 1e-06,
      "loss": 0.5216,
      "num_input_tokens_seen": 392269336,
      "step": 7001
    },
    {
      "epoch": 15.592427616926503,
      "loss": 0.47645553946495056,
      "loss_ce": 0.00013718298578169197,
      "loss_iou": 0.2099609375,
      "loss_num": 0.01123046875,
      "loss_xval": 0.4765625,
      "num_input_tokens_seen": 392269336,
      "step": 7001
    },
    {
      "epoch": 15.594654788418708,
      "grad_norm": 14.073122024536133,
      "learning_rate": 1e-06,
      "loss": 0.4764,
      "num_input_tokens_seen": 392324300,
      "step": 7002
    },
    {
      "epoch": 15.594654788418708,
      "loss": 0.3416048288345337,
      "loss_ce": 0.00011310819536447525,
      "loss_iou": 0.1552734375,
      "loss_num": 0.006195068359375,
      "loss_xval": 0.341796875,
      "num_input_tokens_seen": 392324300,
      "step": 7002
    },
    {
      "epoch": 15.596881959910913,
      "grad_norm": 23.41605567932129,
      "learning_rate": 1e-06,
      "loss": 0.4436,
      "num_input_tokens_seen": 392381052,
      "step": 7003
    },
    {
      "epoch": 15.596881959910913,
      "loss": 0.4482382535934448,
      "loss_ce": 0.00036228023236617446,
      "loss_iou": 0.19921875,
      "loss_num": 0.0098876953125,
      "loss_xval": 0.447265625,
      "num_input_tokens_seen": 392381052,
      "step": 7003
    },
    {
      "epoch": 15.599109131403118,
      "grad_norm": 18.33258628845215,
      "learning_rate": 1e-06,
      "loss": 0.3708,
      "num_input_tokens_seen": 392438500,
      "step": 7004
    },
    {
      "epoch": 15.599109131403118,
      "loss": 0.49753135442733765,
      "loss_ce": 9.481675806455314e-05,
      "loss_iou": 0.2001953125,
      "loss_num": 0.019287109375,
      "loss_xval": 0.498046875,
      "num_input_tokens_seen": 392438500,
      "step": 7004
    },
    {
      "epoch": 15.601336302895323,
      "grad_norm": 40.55979537963867,
      "learning_rate": 1e-06,
      "loss": 0.4622,
      "num_input_tokens_seen": 392491684,
      "step": 7005
    },
    {
      "epoch": 15.601336302895323,
      "loss": 0.2954270541667938,
      "loss_ce": 0.00013897480675950646,
      "loss_iou": 0.1318359375,
      "loss_num": 0.006195068359375,
      "loss_xval": 0.294921875,
      "num_input_tokens_seen": 392491684,
      "step": 7005
    },
    {
      "epoch": 15.603563474387528,
      "grad_norm": 10.46609115600586,
      "learning_rate": 1e-06,
      "loss": 0.4256,
      "num_input_tokens_seen": 392550124,
      "step": 7006
    },
    {
      "epoch": 15.603563474387528,
      "loss": 0.4542039632797241,
      "loss_ce": 0.00010239450784865767,
      "loss_iou": 0.2001953125,
      "loss_num": 0.01080322265625,
      "loss_xval": 0.453125,
      "num_input_tokens_seen": 392550124,
      "step": 7006
    },
    {
      "epoch": 15.605790645879733,
      "grad_norm": 15.158596992492676,
      "learning_rate": 1e-06,
      "loss": 0.3997,
      "num_input_tokens_seen": 392606256,
      "step": 7007
    },
    {
      "epoch": 15.605790645879733,
      "loss": 0.473002552986145,
      "loss_ce": 0.00010216711962129921,
      "loss_iou": 0.1865234375,
      "loss_num": 0.02001953125,
      "loss_xval": 0.47265625,
      "num_input_tokens_seen": 392606256,
      "step": 7007
    },
    {
      "epoch": 15.608017817371937,
      "grad_norm": 23.731393814086914,
      "learning_rate": 1e-06,
      "loss": 0.4986,
      "num_input_tokens_seen": 392661192,
      "step": 7008
    },
    {
      "epoch": 15.608017817371937,
      "loss": 0.4049649238586426,
      "loss_ce": 8.820135553833097e-05,
      "loss_iou": 0.1845703125,
      "loss_num": 0.00732421875,
      "loss_xval": 0.404296875,
      "num_input_tokens_seen": 392661192,
      "step": 7008
    },
    {
      "epoch": 15.610244988864142,
      "grad_norm": 15.854848861694336,
      "learning_rate": 1e-06,
      "loss": 0.4375,
      "num_input_tokens_seen": 392716600,
      "step": 7009
    },
    {
      "epoch": 15.610244988864142,
      "loss": 0.5096275806427002,
      "loss_ce": 0.00010606721480144188,
      "loss_iou": 0.20703125,
      "loss_num": 0.01904296875,
      "loss_xval": 0.5078125,
      "num_input_tokens_seen": 392716600,
      "step": 7009
    },
    {
      "epoch": 15.612472160356347,
      "grad_norm": 13.901849746704102,
      "learning_rate": 1e-06,
      "loss": 0.4194,
      "num_input_tokens_seen": 392774900,
      "step": 7010
    },
    {
      "epoch": 15.612472160356347,
      "loss": 0.4252671003341675,
      "loss_ce": 9.621331264497712e-05,
      "loss_iou": 0.19140625,
      "loss_num": 0.0084228515625,
      "loss_xval": 0.42578125,
      "num_input_tokens_seen": 392774900,
      "step": 7010
    },
    {
      "epoch": 15.614699331848552,
      "grad_norm": 13.334481239318848,
      "learning_rate": 1e-06,
      "loss": 0.4398,
      "num_input_tokens_seen": 392830348,
      "step": 7011
    },
    {
      "epoch": 15.614699331848552,
      "loss": 0.6181551218032837,
      "loss_ce": 0.00011313124559819698,
      "loss_iou": 0.244140625,
      "loss_num": 0.02587890625,
      "loss_xval": 0.6171875,
      "num_input_tokens_seen": 392830348,
      "step": 7011
    },
    {
      "epoch": 15.616926503340757,
      "grad_norm": 17.49168586730957,
      "learning_rate": 1e-06,
      "loss": 0.6052,
      "num_input_tokens_seen": 392887200,
      "step": 7012
    },
    {
      "epoch": 15.616926503340757,
      "loss": 0.7159167528152466,
      "loss_ce": 9.638856136007234e-05,
      "loss_iou": 0.310546875,
      "loss_num": 0.0186767578125,
      "loss_xval": 0.71484375,
      "num_input_tokens_seen": 392887200,
      "step": 7012
    },
    {
      "epoch": 15.619153674832962,
      "grad_norm": 18.257720947265625,
      "learning_rate": 1e-06,
      "loss": 0.3736,
      "num_input_tokens_seen": 392945796,
      "step": 7013
    },
    {
      "epoch": 15.619153674832962,
      "loss": 0.4207872748374939,
      "loss_ce": 0.000132970220874995,
      "loss_iou": 0.1796875,
      "loss_num": 0.012451171875,
      "loss_xval": 0.419921875,
      "num_input_tokens_seen": 392945796,
      "step": 7013
    },
    {
      "epoch": 15.621380846325167,
      "grad_norm": 21.57740020751953,
      "learning_rate": 1e-06,
      "loss": 0.3268,
      "num_input_tokens_seen": 393004392,
      "step": 7014
    },
    {
      "epoch": 15.621380846325167,
      "loss": 0.26535099744796753,
      "loss_ce": 9.219862113241106e-05,
      "loss_iou": 0.12109375,
      "loss_num": 0.00457763671875,
      "loss_xval": 0.265625,
      "num_input_tokens_seen": 393004392,
      "step": 7014
    },
    {
      "epoch": 15.623608017817372,
      "grad_norm": 17.83563232421875,
      "learning_rate": 1e-06,
      "loss": 0.3994,
      "num_input_tokens_seen": 393060176,
      "step": 7015
    },
    {
      "epoch": 15.623608017817372,
      "loss": 0.3122119903564453,
      "loss_ce": 7.820721657481045e-05,
      "loss_iou": 0.1357421875,
      "loss_num": 0.00799560546875,
      "loss_xval": 0.3125,
      "num_input_tokens_seen": 393060176,
      "step": 7015
    },
    {
      "epoch": 15.625835189309576,
      "grad_norm": 29.299760818481445,
      "learning_rate": 1e-06,
      "loss": 0.5213,
      "num_input_tokens_seen": 393118056,
      "step": 7016
    },
    {
      "epoch": 15.625835189309576,
      "loss": 0.37060514092445374,
      "loss_ce": 0.00012174376752227545,
      "loss_iou": 0.16796875,
      "loss_num": 0.006927490234375,
      "loss_xval": 0.37109375,
      "num_input_tokens_seen": 393118056,
      "step": 7016
    },
    {
      "epoch": 15.628062360801781,
      "grad_norm": 12.648885726928711,
      "learning_rate": 1e-06,
      "loss": 0.5216,
      "num_input_tokens_seen": 393177148,
      "step": 7017
    },
    {
      "epoch": 15.628062360801781,
      "loss": 0.40695855021476746,
      "loss_ce": 9.820509876590222e-05,
      "loss_iou": 0.189453125,
      "loss_num": 0.00543212890625,
      "loss_xval": 0.40625,
      "num_input_tokens_seen": 393177148,
      "step": 7017
    },
    {
      "epoch": 15.630289532293986,
      "grad_norm": 24.881315231323242,
      "learning_rate": 1e-06,
      "loss": 0.3718,
      "num_input_tokens_seen": 393231232,
      "step": 7018
    },
    {
      "epoch": 15.630289532293986,
      "loss": 0.41211479902267456,
      "loss_ce": 0.00012749881716445088,
      "loss_iou": 0.1689453125,
      "loss_num": 0.0147705078125,
      "loss_xval": 0.412109375,
      "num_input_tokens_seen": 393231232,
      "step": 7018
    },
    {
      "epoch": 15.632516703786191,
      "grad_norm": 21.462350845336914,
      "learning_rate": 1e-06,
      "loss": 0.5006,
      "num_input_tokens_seen": 393285416,
      "step": 7019
    },
    {
      "epoch": 15.632516703786191,
      "loss": 0.383044958114624,
      "loss_ce": 0.00023245607735589147,
      "loss_iou": 0.1767578125,
      "loss_num": 0.005767822265625,
      "loss_xval": 0.3828125,
      "num_input_tokens_seen": 393285416,
      "step": 7019
    },
    {
      "epoch": 15.634743875278396,
      "grad_norm": 18.3259220123291,
      "learning_rate": 1e-06,
      "loss": 0.5313,
      "num_input_tokens_seen": 393342488,
      "step": 7020
    },
    {
      "epoch": 15.634743875278396,
      "loss": 0.6249858140945435,
      "loss_ce": 0.00016893941210582852,
      "loss_iou": 0.259765625,
      "loss_num": 0.02099609375,
      "loss_xval": 0.625,
      "num_input_tokens_seen": 393342488,
      "step": 7020
    },
    {
      "epoch": 15.6369710467706,
      "grad_norm": 33.112945556640625,
      "learning_rate": 1e-06,
      "loss": 0.5863,
      "num_input_tokens_seen": 393400080,
      "step": 7021
    },
    {
      "epoch": 15.6369710467706,
      "loss": 0.4414959251880646,
      "loss_ce": 8.966858149506152e-05,
      "loss_iou": 0.1953125,
      "loss_num": 0.01025390625,
      "loss_xval": 0.44140625,
      "num_input_tokens_seen": 393400080,
      "step": 7021
    },
    {
      "epoch": 15.639198218262806,
      "grad_norm": 16.2619571685791,
      "learning_rate": 1e-06,
      "loss": 0.3958,
      "num_input_tokens_seen": 393456744,
      "step": 7022
    },
    {
      "epoch": 15.639198218262806,
      "loss": 0.37460190057754517,
      "loss_ce": 9.020163997774944e-05,
      "loss_iou": 0.16796875,
      "loss_num": 0.007720947265625,
      "loss_xval": 0.375,
      "num_input_tokens_seen": 393456744,
      "step": 7022
    },
    {
      "epoch": 15.64142538975501,
      "grad_norm": 20.06168556213379,
      "learning_rate": 1e-06,
      "loss": 0.3711,
      "num_input_tokens_seen": 393511708,
      "step": 7023
    },
    {
      "epoch": 15.64142538975501,
      "loss": 0.4522230625152588,
      "loss_ce": 7.460695633199066e-05,
      "loss_iou": 0.1875,
      "loss_num": 0.0155029296875,
      "loss_xval": 0.453125,
      "num_input_tokens_seen": 393511708,
      "step": 7023
    },
    {
      "epoch": 15.643652561247215,
      "grad_norm": 13.850775718688965,
      "learning_rate": 1e-06,
      "loss": 0.4639,
      "num_input_tokens_seen": 393569980,
      "step": 7024
    },
    {
      "epoch": 15.643652561247215,
      "loss": 0.4485822916030884,
      "loss_ce": 9.598617180017754e-05,
      "loss_iou": 0.1904296875,
      "loss_num": 0.01348876953125,
      "loss_xval": 0.44921875,
      "num_input_tokens_seen": 393569980,
      "step": 7024
    },
    {
      "epoch": 15.64587973273942,
      "grad_norm": 24.194141387939453,
      "learning_rate": 1e-06,
      "loss": 0.6029,
      "num_input_tokens_seen": 393623688,
      "step": 7025
    },
    {
      "epoch": 15.64587973273942,
      "loss": 0.7536381483078003,
      "loss_ce": 9.815217345021665e-05,
      "loss_iou": 0.283203125,
      "loss_num": 0.037353515625,
      "loss_xval": 0.75390625,
      "num_input_tokens_seen": 393623688,
      "step": 7025
    },
    {
      "epoch": 15.648106904231625,
      "grad_norm": 149.88299560546875,
      "learning_rate": 1e-06,
      "loss": 0.4324,
      "num_input_tokens_seen": 393678112,
      "step": 7026
    },
    {
      "epoch": 15.648106904231625,
      "loss": 0.4286256432533264,
      "loss_ce": 9.779322135727853e-05,
      "loss_iou": 0.1748046875,
      "loss_num": 0.0157470703125,
      "loss_xval": 0.427734375,
      "num_input_tokens_seen": 393678112,
      "step": 7026
    },
    {
      "epoch": 15.65033407572383,
      "grad_norm": 18.4648494720459,
      "learning_rate": 1e-06,
      "loss": 0.5376,
      "num_input_tokens_seen": 393735444,
      "step": 7027
    },
    {
      "epoch": 15.65033407572383,
      "loss": 0.5531010627746582,
      "loss_ce": 0.00012253447494003922,
      "loss_iou": 0.2158203125,
      "loss_num": 0.0245361328125,
      "loss_xval": 0.5546875,
      "num_input_tokens_seen": 393735444,
      "step": 7027
    },
    {
      "epoch": 15.652561247216035,
      "grad_norm": 23.74418830871582,
      "learning_rate": 1e-06,
      "loss": 0.4233,
      "num_input_tokens_seen": 393789508,
      "step": 7028
    },
    {
      "epoch": 15.652561247216035,
      "loss": 0.3642292618751526,
      "loss_ce": 9.35211282921955e-05,
      "loss_iou": 0.1552734375,
      "loss_num": 0.01080322265625,
      "loss_xval": 0.36328125,
      "num_input_tokens_seen": 393789508,
      "step": 7028
    },
    {
      "epoch": 15.654788418708241,
      "grad_norm": 18.24901580810547,
      "learning_rate": 1e-06,
      "loss": 0.4294,
      "num_input_tokens_seen": 393845576,
      "step": 7029
    },
    {
      "epoch": 15.654788418708241,
      "loss": 0.337370365858078,
      "loss_ce": 9.009480709210038e-05,
      "loss_iou": 0.150390625,
      "loss_num": 0.007293701171875,
      "loss_xval": 0.337890625,
      "num_input_tokens_seen": 393845576,
      "step": 7029
    },
    {
      "epoch": 15.657015590200446,
      "grad_norm": 18.35692596435547,
      "learning_rate": 1e-06,
      "loss": 0.3768,
      "num_input_tokens_seen": 393901840,
      "step": 7030
    },
    {
      "epoch": 15.657015590200446,
      "loss": 0.3451303243637085,
      "loss_ce": 0.00015963747864589095,
      "loss_iou": 0.158203125,
      "loss_num": 0.005462646484375,
      "loss_xval": 0.345703125,
      "num_input_tokens_seen": 393901840,
      "step": 7030
    },
    {
      "epoch": 15.659242761692651,
      "grad_norm": 19.633066177368164,
      "learning_rate": 1e-06,
      "loss": 0.3587,
      "num_input_tokens_seen": 393956624,
      "step": 7031
    },
    {
      "epoch": 15.659242761692651,
      "loss": 0.23722299933433533,
      "loss_ce": 0.00010140843369299546,
      "loss_iou": 0.1083984375,
      "loss_num": 0.004058837890625,
      "loss_xval": 0.2373046875,
      "num_input_tokens_seen": 393956624,
      "step": 7031
    },
    {
      "epoch": 15.661469933184856,
      "grad_norm": 13.26457691192627,
      "learning_rate": 1e-06,
      "loss": 0.4878,
      "num_input_tokens_seen": 394015648,
      "step": 7032
    },
    {
      "epoch": 15.661469933184856,
      "loss": 0.5062072277069092,
      "loss_ce": 0.0003479141159914434,
      "loss_iou": 0.2041015625,
      "loss_num": 0.01953125,
      "loss_xval": 0.5078125,
      "num_input_tokens_seen": 394015648,
      "step": 7032
    },
    {
      "epoch": 15.66369710467706,
      "grad_norm": 15.56800651550293,
      "learning_rate": 1e-06,
      "loss": 0.4825,
      "num_input_tokens_seen": 394068676,
      "step": 7033
    },
    {
      "epoch": 15.66369710467706,
      "loss": 0.38187873363494873,
      "loss_ce": 0.0001038167392835021,
      "loss_iou": 0.1552734375,
      "loss_num": 0.01409912109375,
      "loss_xval": 0.380859375,
      "num_input_tokens_seen": 394068676,
      "step": 7033
    },
    {
      "epoch": 15.665924276169266,
      "grad_norm": 13.747230529785156,
      "learning_rate": 1e-06,
      "loss": 0.502,
      "num_input_tokens_seen": 394125676,
      "step": 7034
    },
    {
      "epoch": 15.665924276169266,
      "loss": 0.47450119256973267,
      "loss_ce": 0.0001359964517178014,
      "loss_iou": 0.208984375,
      "loss_num": 0.01129150390625,
      "loss_xval": 0.474609375,
      "num_input_tokens_seen": 394125676,
      "step": 7034
    },
    {
      "epoch": 15.66815144766147,
      "grad_norm": 13.67151165008545,
      "learning_rate": 1e-06,
      "loss": 0.3227,
      "num_input_tokens_seen": 394182648,
      "step": 7035
    },
    {
      "epoch": 15.66815144766147,
      "loss": 0.3714517056941986,
      "loss_ce": 0.0001138186635216698,
      "loss_iou": 0.1689453125,
      "loss_num": 0.006683349609375,
      "loss_xval": 0.37109375,
      "num_input_tokens_seen": 394182648,
      "step": 7035
    },
    {
      "epoch": 15.670378619153675,
      "grad_norm": 27.722789764404297,
      "learning_rate": 1e-06,
      "loss": 0.4209,
      "num_input_tokens_seen": 394236828,
      "step": 7036
    },
    {
      "epoch": 15.670378619153675,
      "loss": 0.4359535276889801,
      "loss_ce": 0.00010151089372811839,
      "loss_iou": 0.189453125,
      "loss_num": 0.011474609375,
      "loss_xval": 0.435546875,
      "num_input_tokens_seen": 394236828,
      "step": 7036
    },
    {
      "epoch": 15.67260579064588,
      "grad_norm": 18.144771575927734,
      "learning_rate": 1e-06,
      "loss": 0.4676,
      "num_input_tokens_seen": 394291552,
      "step": 7037
    },
    {
      "epoch": 15.67260579064588,
      "loss": 0.4619390070438385,
      "loss_ce": 8.598440763307735e-05,
      "loss_iou": 0.201171875,
      "loss_num": 0.011962890625,
      "loss_xval": 0.4609375,
      "num_input_tokens_seen": 394291552,
      "step": 7037
    },
    {
      "epoch": 15.674832962138085,
      "grad_norm": 15.883332252502441,
      "learning_rate": 1e-06,
      "loss": 0.3619,
      "num_input_tokens_seen": 394349988,
      "step": 7038
    },
    {
      "epoch": 15.674832962138085,
      "loss": 0.3775450587272644,
      "loss_ce": 0.00010363116598455235,
      "loss_iou": 0.166015625,
      "loss_num": 0.00897216796875,
      "loss_xval": 0.376953125,
      "num_input_tokens_seen": 394349988,
      "step": 7038
    },
    {
      "epoch": 15.67706013363029,
      "grad_norm": 18.735139846801758,
      "learning_rate": 1e-06,
      "loss": 0.3232,
      "num_input_tokens_seen": 394405816,
      "step": 7039
    },
    {
      "epoch": 15.67706013363029,
      "loss": 0.30405598878860474,
      "loss_ce": 0.0001009251645882614,
      "loss_iou": 0.12890625,
      "loss_num": 0.0093994140625,
      "loss_xval": 0.3046875,
      "num_input_tokens_seen": 394405816,
      "step": 7039
    },
    {
      "epoch": 15.679287305122495,
      "grad_norm": 32.7242546081543,
      "learning_rate": 1e-06,
      "loss": 0.5001,
      "num_input_tokens_seen": 394458988,
      "step": 7040
    },
    {
      "epoch": 15.679287305122495,
      "loss": 0.6007954478263855,
      "loss_ce": 8.743096987018362e-05,
      "loss_iou": 0.265625,
      "loss_num": 0.0137939453125,
      "loss_xval": 0.6015625,
      "num_input_tokens_seen": 394458988,
      "step": 7040
    },
    {
      "epoch": 15.6815144766147,
      "grad_norm": 17.01080322265625,
      "learning_rate": 1e-06,
      "loss": 0.4883,
      "num_input_tokens_seen": 394517044,
      "step": 7041
    },
    {
      "epoch": 15.6815144766147,
      "loss": 0.6097177267074585,
      "loss_ce": 9.863986633718014e-05,
      "loss_iou": 0.26953125,
      "loss_num": 0.014404296875,
      "loss_xval": 0.609375,
      "num_input_tokens_seen": 394517044,
      "step": 7041
    },
    {
      "epoch": 15.683741648106905,
      "grad_norm": 23.347501754760742,
      "learning_rate": 1e-06,
      "loss": 0.4004,
      "num_input_tokens_seen": 394575972,
      "step": 7042
    },
    {
      "epoch": 15.683741648106905,
      "loss": 0.424541175365448,
      "loss_ce": 0.00010270069469697773,
      "loss_iou": 0.193359375,
      "loss_num": 0.007659912109375,
      "loss_xval": 0.423828125,
      "num_input_tokens_seen": 394575972,
      "step": 7042
    },
    {
      "epoch": 15.68596881959911,
      "grad_norm": 17.75396156311035,
      "learning_rate": 1e-06,
      "loss": 0.4391,
      "num_input_tokens_seen": 394631332,
      "step": 7043
    },
    {
      "epoch": 15.68596881959911,
      "loss": 0.5316604375839233,
      "loss_ce": 0.0010207871673628688,
      "loss_iou": 0.2275390625,
      "loss_num": 0.01519775390625,
      "loss_xval": 0.53125,
      "num_input_tokens_seen": 394631332,
      "step": 7043
    },
    {
      "epoch": 15.688195991091314,
      "grad_norm": 20.462841033935547,
      "learning_rate": 1e-06,
      "loss": 0.2736,
      "num_input_tokens_seen": 394687344,
      "step": 7044
    },
    {
      "epoch": 15.688195991091314,
      "loss": 0.30839377641677856,
      "loss_ce": 0.0001052148436428979,
      "loss_iou": 0.12890625,
      "loss_num": 0.0101318359375,
      "loss_xval": 0.30859375,
      "num_input_tokens_seen": 394687344,
      "step": 7044
    },
    {
      "epoch": 15.690423162583519,
      "grad_norm": 17.69515037536621,
      "learning_rate": 1e-06,
      "loss": 0.5623,
      "num_input_tokens_seen": 394742256,
      "step": 7045
    },
    {
      "epoch": 15.690423162583519,
      "loss": 0.6370512247085571,
      "loss_ce": 0.00011880889360327274,
      "loss_iou": 0.279296875,
      "loss_num": 0.0155029296875,
      "loss_xval": 0.63671875,
      "num_input_tokens_seen": 394742256,
      "step": 7045
    },
    {
      "epoch": 15.692650334075724,
      "grad_norm": 19.867359161376953,
      "learning_rate": 1e-06,
      "loss": 0.4736,
      "num_input_tokens_seen": 394796212,
      "step": 7046
    },
    {
      "epoch": 15.692650334075724,
      "loss": 0.5707098245620728,
      "loss_ce": 0.00015316563076339662,
      "loss_iou": 0.2353515625,
      "loss_num": 0.0198974609375,
      "loss_xval": 0.5703125,
      "num_input_tokens_seen": 394796212,
      "step": 7046
    },
    {
      "epoch": 15.694877505567929,
      "grad_norm": 17.22439193725586,
      "learning_rate": 1e-06,
      "loss": 0.5096,
      "num_input_tokens_seen": 394849460,
      "step": 7047
    },
    {
      "epoch": 15.694877505567929,
      "loss": 0.4576007127761841,
      "loss_ce": 8.11615027487278e-05,
      "loss_iou": 0.2060546875,
      "loss_num": 0.0091552734375,
      "loss_xval": 0.45703125,
      "num_input_tokens_seen": 394849460,
      "step": 7047
    },
    {
      "epoch": 15.697104677060134,
      "grad_norm": 19.11503791809082,
      "learning_rate": 1e-06,
      "loss": 0.3946,
      "num_input_tokens_seen": 394907576,
      "step": 7048
    },
    {
      "epoch": 15.697104677060134,
      "loss": 0.29610675573349,
      "loss_ce": 8.626521594123915e-05,
      "loss_iou": 0.1328125,
      "loss_num": 0.006256103515625,
      "loss_xval": 0.296875,
      "num_input_tokens_seen": 394907576,
      "step": 7048
    },
    {
      "epoch": 15.699331848552339,
      "grad_norm": 14.611241340637207,
      "learning_rate": 1e-06,
      "loss": 0.3639,
      "num_input_tokens_seen": 394963124,
      "step": 7049
    },
    {
      "epoch": 15.699331848552339,
      "loss": 0.4972745180130005,
      "loss_ce": 8.216071000788361e-05,
      "loss_iou": 0.2099609375,
      "loss_num": 0.015625,
      "loss_xval": 0.498046875,
      "num_input_tokens_seen": 394963124,
      "step": 7049
    },
    {
      "epoch": 15.701559020044543,
      "grad_norm": 22.508859634399414,
      "learning_rate": 1e-06,
      "loss": 0.4025,
      "num_input_tokens_seen": 395020276,
      "step": 7050
    },
    {
      "epoch": 15.701559020044543,
      "loss": 0.35203513503074646,
      "loss_ce": 0.00010644704889273271,
      "loss_iou": 0.154296875,
      "loss_num": 0.0086669921875,
      "loss_xval": 0.3515625,
      "num_input_tokens_seen": 395020276,
      "step": 7050
    },
    {
      "epoch": 15.703786191536748,
      "grad_norm": 24.426239013671875,
      "learning_rate": 1e-06,
      "loss": 0.5065,
      "num_input_tokens_seen": 395078896,
      "step": 7051
    },
    {
      "epoch": 15.703786191536748,
      "loss": 0.530193030834198,
      "loss_ce": 0.00010265262972097844,
      "loss_iou": 0.2333984375,
      "loss_num": 0.0128173828125,
      "loss_xval": 0.53125,
      "num_input_tokens_seen": 395078896,
      "step": 7051
    },
    {
      "epoch": 15.706013363028953,
      "grad_norm": 21.90961456298828,
      "learning_rate": 1e-06,
      "loss": 0.6311,
      "num_input_tokens_seen": 395137768,
      "step": 7052
    },
    {
      "epoch": 15.706013363028953,
      "loss": 0.6914348006248474,
      "loss_ce": 8.963010623119771e-05,
      "loss_iou": 0.267578125,
      "loss_num": 0.03125,
      "loss_xval": 0.69140625,
      "num_input_tokens_seen": 395137768,
      "step": 7052
    },
    {
      "epoch": 15.708240534521158,
      "grad_norm": 17.655078887939453,
      "learning_rate": 1e-06,
      "loss": 0.4446,
      "num_input_tokens_seen": 395194468,
      "step": 7053
    },
    {
      "epoch": 15.708240534521158,
      "loss": 0.49838685989379883,
      "loss_ce": 9.587279055267572e-05,
      "loss_iou": 0.208984375,
      "loss_num": 0.0162353515625,
      "loss_xval": 0.498046875,
      "num_input_tokens_seen": 395194468,
      "step": 7053
    },
    {
      "epoch": 15.710467706013363,
      "grad_norm": 16.85371971130371,
      "learning_rate": 1e-06,
      "loss": 0.7519,
      "num_input_tokens_seen": 395248904,
      "step": 7054
    },
    {
      "epoch": 15.710467706013363,
      "loss": 0.6303013563156128,
      "loss_ce": 0.0001133399928221479,
      "loss_iou": 0.248046875,
      "loss_num": 0.027099609375,
      "loss_xval": 0.62890625,
      "num_input_tokens_seen": 395248904,
      "step": 7054
    },
    {
      "epoch": 15.712694877505568,
      "grad_norm": 18.303388595581055,
      "learning_rate": 1e-06,
      "loss": 0.349,
      "num_input_tokens_seen": 395306728,
      "step": 7055
    },
    {
      "epoch": 15.712694877505568,
      "loss": 0.32140427827835083,
      "loss_ce": 0.00011521350825205445,
      "loss_iou": 0.13671875,
      "loss_num": 0.009765625,
      "loss_xval": 0.3203125,
      "num_input_tokens_seen": 395306728,
      "step": 7055
    },
    {
      "epoch": 15.714922048997773,
      "grad_norm": 16.185035705566406,
      "learning_rate": 1e-06,
      "loss": 0.4021,
      "num_input_tokens_seen": 395362652,
      "step": 7056
    },
    {
      "epoch": 15.714922048997773,
      "loss": 0.24985191226005554,
      "loss_ce": 9.60544275585562e-05,
      "loss_iou": 0.1103515625,
      "loss_num": 0.005706787109375,
      "loss_xval": 0.25,
      "num_input_tokens_seen": 395362652,
      "step": 7056
    },
    {
      "epoch": 15.717149220489977,
      "grad_norm": 12.218123435974121,
      "learning_rate": 1e-06,
      "loss": 0.4379,
      "num_input_tokens_seen": 395419380,
      "step": 7057
    },
    {
      "epoch": 15.717149220489977,
      "loss": 0.4651813805103302,
      "loss_ce": 9.347945888293907e-05,
      "loss_iou": 0.203125,
      "loss_num": 0.0115966796875,
      "loss_xval": 0.46484375,
      "num_input_tokens_seen": 395419380,
      "step": 7057
    },
    {
      "epoch": 15.719376391982182,
      "grad_norm": 19.603456497192383,
      "learning_rate": 1e-06,
      "loss": 0.3471,
      "num_input_tokens_seen": 395475324,
      "step": 7058
    },
    {
      "epoch": 15.719376391982182,
      "loss": 0.33774223923683167,
      "loss_ce": 9.575536387274042e-05,
      "loss_iou": 0.154296875,
      "loss_num": 0.005828857421875,
      "loss_xval": 0.337890625,
      "num_input_tokens_seen": 395475324,
      "step": 7058
    },
    {
      "epoch": 15.721603563474387,
      "grad_norm": 25.467702865600586,
      "learning_rate": 1e-06,
      "loss": 0.379,
      "num_input_tokens_seen": 395532144,
      "step": 7059
    },
    {
      "epoch": 15.721603563474387,
      "loss": 0.3512808680534363,
      "loss_ce": 8.456782961729914e-05,
      "loss_iou": 0.1494140625,
      "loss_num": 0.010498046875,
      "loss_xval": 0.3515625,
      "num_input_tokens_seen": 395532144,
      "step": 7059
    },
    {
      "epoch": 15.723830734966592,
      "grad_norm": 18.630399703979492,
      "learning_rate": 1e-06,
      "loss": 0.3926,
      "num_input_tokens_seen": 395587848,
      "step": 7060
    },
    {
      "epoch": 15.723830734966592,
      "loss": 0.4616337716579437,
      "loss_ce": 8.594746032031253e-05,
      "loss_iou": 0.1953125,
      "loss_num": 0.01422119140625,
      "loss_xval": 0.4609375,
      "num_input_tokens_seen": 395587848,
      "step": 7060
    },
    {
      "epoch": 15.726057906458797,
      "grad_norm": 14.84992504119873,
      "learning_rate": 1e-06,
      "loss": 0.6171,
      "num_input_tokens_seen": 395642736,
      "step": 7061
    },
    {
      "epoch": 15.726057906458797,
      "loss": 0.6872535943984985,
      "loss_ce": 0.0001198096142616123,
      "loss_iou": 0.265625,
      "loss_num": 0.031494140625,
      "loss_xval": 0.6875,
      "num_input_tokens_seen": 395642736,
      "step": 7061
    },
    {
      "epoch": 15.728285077951002,
      "grad_norm": 23.57808494567871,
      "learning_rate": 1e-06,
      "loss": 0.4608,
      "num_input_tokens_seen": 395699808,
      "step": 7062
    },
    {
      "epoch": 15.728285077951002,
      "loss": 0.6477620601654053,
      "loss_ce": 0.00011803221423178911,
      "loss_iou": 0.298828125,
      "loss_num": 0.01031494140625,
      "loss_xval": 0.6484375,
      "num_input_tokens_seen": 395699808,
      "step": 7062
    },
    {
      "epoch": 15.730512249443207,
      "grad_norm": 15.21978759765625,
      "learning_rate": 1e-06,
      "loss": 0.4271,
      "num_input_tokens_seen": 395757824,
      "step": 7063
    },
    {
      "epoch": 15.730512249443207,
      "loss": 0.3883220851421356,
      "loss_ce": 7.74528380134143e-05,
      "loss_iou": 0.173828125,
      "loss_num": 0.00799560546875,
      "loss_xval": 0.388671875,
      "num_input_tokens_seen": 395757824,
      "step": 7063
    },
    {
      "epoch": 15.732739420935411,
      "grad_norm": 29.952592849731445,
      "learning_rate": 1e-06,
      "loss": 0.3861,
      "num_input_tokens_seen": 395815416,
      "step": 7064
    },
    {
      "epoch": 15.732739420935411,
      "loss": 0.37809085845947266,
      "loss_ce": 0.00010012884740717709,
      "loss_iou": 0.1728515625,
      "loss_num": 0.006500244140625,
      "loss_xval": 0.37890625,
      "num_input_tokens_seen": 395815416,
      "step": 7064
    },
    {
      "epoch": 15.734966592427616,
      "grad_norm": 16.03706169128418,
      "learning_rate": 1e-06,
      "loss": 0.4989,
      "num_input_tokens_seen": 395868684,
      "step": 7065
    },
    {
      "epoch": 15.734966592427616,
      "loss": 0.523518443107605,
      "loss_ce": 8.08899276307784e-05,
      "loss_iou": 0.2255859375,
      "loss_num": 0.0142822265625,
      "loss_xval": 0.5234375,
      "num_input_tokens_seen": 395868684,
      "step": 7065
    },
    {
      "epoch": 15.737193763919821,
      "grad_norm": 16.18170928955078,
      "learning_rate": 1e-06,
      "loss": 0.3421,
      "num_input_tokens_seen": 395925000,
      "step": 7066
    },
    {
      "epoch": 15.737193763919821,
      "loss": 0.4338296949863434,
      "loss_ce": 0.00011389805149519816,
      "loss_iou": 0.1982421875,
      "loss_num": 0.007568359375,
      "loss_xval": 0.43359375,
      "num_input_tokens_seen": 395925000,
      "step": 7066
    },
    {
      "epoch": 15.739420935412026,
      "grad_norm": 11.560032844543457,
      "learning_rate": 1e-06,
      "loss": 0.3395,
      "num_input_tokens_seen": 395979500,
      "step": 7067
    },
    {
      "epoch": 15.739420935412026,
      "loss": 0.26005110144615173,
      "loss_ce": 0.00010237214155495167,
      "loss_iou": 0.107421875,
      "loss_num": 0.009033203125,
      "loss_xval": 0.259765625,
      "num_input_tokens_seen": 395979500,
      "step": 7067
    },
    {
      "epoch": 15.74164810690423,
      "grad_norm": 24.783843994140625,
      "learning_rate": 1e-06,
      "loss": 0.3423,
      "num_input_tokens_seen": 396037980,
      "step": 7068
    },
    {
      "epoch": 15.74164810690423,
      "loss": 0.2767605781555176,
      "loss_ce": 8.823502867016941e-05,
      "loss_iou": 0.10791015625,
      "loss_num": 0.01214599609375,
      "loss_xval": 0.27734375,
      "num_input_tokens_seen": 396037980,
      "step": 7068
    },
    {
      "epoch": 15.743875278396436,
      "grad_norm": 13.664560317993164,
      "learning_rate": 1e-06,
      "loss": 0.3566,
      "num_input_tokens_seen": 396095480,
      "step": 7069
    },
    {
      "epoch": 15.743875278396436,
      "loss": 0.30287209153175354,
      "loss_ce": 7.669955084566027e-05,
      "loss_iou": 0.1328125,
      "loss_num": 0.00750732421875,
      "loss_xval": 0.302734375,
      "num_input_tokens_seen": 396095480,
      "step": 7069
    },
    {
      "epoch": 15.74610244988864,
      "grad_norm": 17.73408317565918,
      "learning_rate": 1e-06,
      "loss": 0.4481,
      "num_input_tokens_seen": 396152320,
      "step": 7070
    },
    {
      "epoch": 15.74610244988864,
      "loss": 0.5743053555488586,
      "loss_ce": 8.658809383632615e-05,
      "loss_iou": 0.2392578125,
      "loss_num": 0.019287109375,
      "loss_xval": 0.57421875,
      "num_input_tokens_seen": 396152320,
      "step": 7070
    },
    {
      "epoch": 15.748329621380847,
      "grad_norm": 16.201440811157227,
      "learning_rate": 1e-06,
      "loss": 0.4466,
      "num_input_tokens_seen": 396212012,
      "step": 7071
    },
    {
      "epoch": 15.748329621380847,
      "loss": 0.5690320730209351,
      "loss_ce": 0.00018442686996422708,
      "loss_iou": 0.232421875,
      "loss_num": 0.0211181640625,
      "loss_xval": 0.5703125,
      "num_input_tokens_seen": 396212012,
      "step": 7071
    },
    {
      "epoch": 15.750556792873052,
      "grad_norm": 15.789632797241211,
      "learning_rate": 1e-06,
      "loss": 0.3196,
      "num_input_tokens_seen": 396266208,
      "step": 7072
    },
    {
      "epoch": 15.750556792873052,
      "loss": 0.30217671394348145,
      "loss_ce": 0.00011373275629011914,
      "loss_iou": 0.126953125,
      "loss_num": 0.00970458984375,
      "loss_xval": 0.302734375,
      "num_input_tokens_seen": 396266208,
      "step": 7072
    },
    {
      "epoch": 15.752783964365257,
      "grad_norm": 18.404560089111328,
      "learning_rate": 1e-06,
      "loss": 0.4347,
      "num_input_tokens_seen": 396325024,
      "step": 7073
    },
    {
      "epoch": 15.752783964365257,
      "loss": 0.2929561734199524,
      "loss_ce": 7.899131742306054e-05,
      "loss_iou": 0.12060546875,
      "loss_num": 0.0103759765625,
      "loss_xval": 0.29296875,
      "num_input_tokens_seen": 396325024,
      "step": 7073
    },
    {
      "epoch": 15.755011135857462,
      "grad_norm": 24.089750289916992,
      "learning_rate": 1e-06,
      "loss": 0.5156,
      "num_input_tokens_seen": 396382244,
      "step": 7074
    },
    {
      "epoch": 15.755011135857462,
      "loss": 0.2893972396850586,
      "loss_ce": 9.059577132575214e-05,
      "loss_iou": 0.125,
      "loss_num": 0.00799560546875,
      "loss_xval": 0.2890625,
      "num_input_tokens_seen": 396382244,
      "step": 7074
    },
    {
      "epoch": 15.757238307349667,
      "grad_norm": 13.719793319702148,
      "learning_rate": 1e-06,
      "loss": 0.3745,
      "num_input_tokens_seen": 396438428,
      "step": 7075
    },
    {
      "epoch": 15.757238307349667,
      "loss": 0.41726016998291016,
      "loss_ce": 0.00014590269711334258,
      "loss_iou": 0.1650390625,
      "loss_num": 0.017333984375,
      "loss_xval": 0.41796875,
      "num_input_tokens_seen": 396438428,
      "step": 7075
    },
    {
      "epoch": 15.759465478841872,
      "grad_norm": 17.886260986328125,
      "learning_rate": 1e-06,
      "loss": 0.5086,
      "num_input_tokens_seen": 396492232,
      "step": 7076
    },
    {
      "epoch": 15.759465478841872,
      "loss": 0.2809308171272278,
      "loss_ce": 0.00010807791841216385,
      "loss_iou": 0.11083984375,
      "loss_num": 0.0118408203125,
      "loss_xval": 0.28125,
      "num_input_tokens_seen": 396492232,
      "step": 7076
    },
    {
      "epoch": 15.761692650334076,
      "grad_norm": 27.26051902770996,
      "learning_rate": 1e-06,
      "loss": 0.4125,
      "num_input_tokens_seen": 396546276,
      "step": 7077
    },
    {
      "epoch": 15.761692650334076,
      "loss": 0.391597181558609,
      "loss_ce": 0.00011767500836867839,
      "loss_iou": 0.1552734375,
      "loss_num": 0.0162353515625,
      "loss_xval": 0.390625,
      "num_input_tokens_seen": 396546276,
      "step": 7077
    },
    {
      "epoch": 15.763919821826281,
      "grad_norm": 19.251379013061523,
      "learning_rate": 1e-06,
      "loss": 0.6037,
      "num_input_tokens_seen": 396600784,
      "step": 7078
    },
    {
      "epoch": 15.763919821826281,
      "loss": 0.5241034030914307,
      "loss_ce": 0.00011657152936095372,
      "loss_iou": 0.234375,
      "loss_num": 0.01080322265625,
      "loss_xval": 0.5234375,
      "num_input_tokens_seen": 396600784,
      "step": 7078
    },
    {
      "epoch": 15.766146993318486,
      "grad_norm": 15.035192489624023,
      "learning_rate": 1e-06,
      "loss": 0.579,
      "num_input_tokens_seen": 396656868,
      "step": 7079
    },
    {
      "epoch": 15.766146993318486,
      "loss": 0.699531078338623,
      "loss_ce": 0.00031230467720888555,
      "loss_iou": 0.296875,
      "loss_num": 0.021240234375,
      "loss_xval": 0.69921875,
      "num_input_tokens_seen": 396656868,
      "step": 7079
    },
    {
      "epoch": 15.768374164810691,
      "grad_norm": 23.482166290283203,
      "learning_rate": 1e-06,
      "loss": 0.4239,
      "num_input_tokens_seen": 396711564,
      "step": 7080
    },
    {
      "epoch": 15.768374164810691,
      "loss": 0.4484688639640808,
      "loss_ce": 0.00010461645433679223,
      "loss_iou": 0.18359375,
      "loss_num": 0.0162353515625,
      "loss_xval": 0.44921875,
      "num_input_tokens_seen": 396711564,
      "step": 7080
    },
    {
      "epoch": 15.770601336302896,
      "grad_norm": 15.677742958068848,
      "learning_rate": 1e-06,
      "loss": 0.4709,
      "num_input_tokens_seen": 396769364,
      "step": 7081
    },
    {
      "epoch": 15.770601336302896,
      "loss": 0.4059576392173767,
      "loss_ce": 0.00016540827346034348,
      "loss_iou": 0.177734375,
      "loss_num": 0.01019287109375,
      "loss_xval": 0.40625,
      "num_input_tokens_seen": 396769364,
      "step": 7081
    },
    {
      "epoch": 15.7728285077951,
      "grad_norm": 28.17963409423828,
      "learning_rate": 1e-06,
      "loss": 0.4652,
      "num_input_tokens_seen": 396826500,
      "step": 7082
    },
    {
      "epoch": 15.7728285077951,
      "loss": 0.4379936456680298,
      "loss_ce": 0.00012745258572977036,
      "loss_iou": 0.1728515625,
      "loss_num": 0.018310546875,
      "loss_xval": 0.4375,
      "num_input_tokens_seen": 396826500,
      "step": 7082
    },
    {
      "epoch": 15.775055679287306,
      "grad_norm": 15.786062240600586,
      "learning_rate": 1e-06,
      "loss": 0.4024,
      "num_input_tokens_seen": 396881984,
      "step": 7083
    },
    {
      "epoch": 15.775055679287306,
      "loss": 0.3337669372558594,
      "loss_ce": 0.00014876520435791463,
      "loss_iou": 0.1533203125,
      "loss_num": 0.00537109375,
      "loss_xval": 0.333984375,
      "num_input_tokens_seen": 396881984,
      "step": 7083
    },
    {
      "epoch": 15.77728285077951,
      "grad_norm": 18.75821304321289,
      "learning_rate": 1e-06,
      "loss": 0.5366,
      "num_input_tokens_seen": 396937636,
      "step": 7084
    },
    {
      "epoch": 15.77728285077951,
      "loss": 0.43820372223854065,
      "loss_ce": 9.338198287878186e-05,
      "loss_iou": 0.197265625,
      "loss_num": 0.00872802734375,
      "loss_xval": 0.4375,
      "num_input_tokens_seen": 396937636,
      "step": 7084
    },
    {
      "epoch": 15.779510022271715,
      "grad_norm": 21.541553497314453,
      "learning_rate": 1e-06,
      "loss": 0.3054,
      "num_input_tokens_seen": 396991964,
      "step": 7085
    },
    {
      "epoch": 15.779510022271715,
      "loss": 0.25417301058769226,
      "loss_ce": 0.0002667623048182577,
      "loss_iou": 0.10693359375,
      "loss_num": 0.008056640625,
      "loss_xval": 0.25390625,
      "num_input_tokens_seen": 396991964,
      "step": 7085
    },
    {
      "epoch": 15.78173719376392,
      "grad_norm": 21.87734603881836,
      "learning_rate": 1e-06,
      "loss": 0.3177,
      "num_input_tokens_seen": 397048488,
      "step": 7086
    },
    {
      "epoch": 15.78173719376392,
      "loss": 0.32540494203567505,
      "loss_ce": 8.75709592946805e-05,
      "loss_iou": 0.134765625,
      "loss_num": 0.01116943359375,
      "loss_xval": 0.326171875,
      "num_input_tokens_seen": 397048488,
      "step": 7086
    },
    {
      "epoch": 15.783964365256125,
      "grad_norm": 21.914344787597656,
      "learning_rate": 1e-06,
      "loss": 0.3939,
      "num_input_tokens_seen": 397103460,
      "step": 7087
    },
    {
      "epoch": 15.783964365256125,
      "loss": 0.4146553874015808,
      "loss_ce": 0.00010459712939336896,
      "loss_iou": 0.166015625,
      "loss_num": 0.0167236328125,
      "loss_xval": 0.4140625,
      "num_input_tokens_seen": 397103460,
      "step": 7087
    },
    {
      "epoch": 15.78619153674833,
      "grad_norm": 21.92269515991211,
      "learning_rate": 1e-06,
      "loss": 0.3736,
      "num_input_tokens_seen": 397158120,
      "step": 7088
    },
    {
      "epoch": 15.78619153674833,
      "loss": 0.3409271240234375,
      "loss_ce": 0.00010680149716790766,
      "loss_iou": 0.1611328125,
      "loss_num": 0.003692626953125,
      "loss_xval": 0.33984375,
      "num_input_tokens_seen": 397158120,
      "step": 7088
    },
    {
      "epoch": 15.788418708240535,
      "grad_norm": 19.19719696044922,
      "learning_rate": 1e-06,
      "loss": 0.4235,
      "num_input_tokens_seen": 397215380,
      "step": 7089
    },
    {
      "epoch": 15.788418708240535,
      "loss": 0.45488834381103516,
      "loss_ce": 0.00017644368926994503,
      "loss_iou": 0.2041015625,
      "loss_num": 0.00946044921875,
      "loss_xval": 0.455078125,
      "num_input_tokens_seen": 397215380,
      "step": 7089
    },
    {
      "epoch": 15.79064587973274,
      "grad_norm": 21.13319969177246,
      "learning_rate": 1e-06,
      "loss": 0.4486,
      "num_input_tokens_seen": 397270500,
      "step": 7090
    },
    {
      "epoch": 15.79064587973274,
      "loss": 0.3929722011089325,
      "loss_ce": 0.00014992158685345203,
      "loss_iou": 0.17578125,
      "loss_num": 0.0081787109375,
      "loss_xval": 0.392578125,
      "num_input_tokens_seen": 397270500,
      "step": 7090
    },
    {
      "epoch": 15.792873051224944,
      "grad_norm": 13.024203300476074,
      "learning_rate": 1e-06,
      "loss": 0.4421,
      "num_input_tokens_seen": 397328104,
      "step": 7091
    },
    {
      "epoch": 15.792873051224944,
      "loss": 0.5068001747131348,
      "loss_ce": 8.632005483377725e-05,
      "loss_iou": 0.2236328125,
      "loss_num": 0.01177978515625,
      "loss_xval": 0.5078125,
      "num_input_tokens_seen": 397328104,
      "step": 7091
    },
    {
      "epoch": 15.79510022271715,
      "grad_norm": 17.045452117919922,
      "learning_rate": 1e-06,
      "loss": 0.3977,
      "num_input_tokens_seen": 397384764,
      "step": 7092
    },
    {
      "epoch": 15.79510022271715,
      "loss": 0.3076792359352112,
      "loss_ce": 9.255674376618117e-05,
      "loss_iou": 0.138671875,
      "loss_num": 0.00592041015625,
      "loss_xval": 0.306640625,
      "num_input_tokens_seen": 397384764,
      "step": 7092
    },
    {
      "epoch": 15.797327394209354,
      "grad_norm": 15.998059272766113,
      "learning_rate": 1e-06,
      "loss": 0.3722,
      "num_input_tokens_seen": 397438400,
      "step": 7093
    },
    {
      "epoch": 15.797327394209354,
      "loss": 0.3463239073753357,
      "loss_ce": 0.00013250944903120399,
      "loss_iou": 0.1474609375,
      "loss_num": 0.01007080078125,
      "loss_xval": 0.345703125,
      "num_input_tokens_seen": 397438400,
      "step": 7093
    },
    {
      "epoch": 15.799554565701559,
      "grad_norm": 19.02997589111328,
      "learning_rate": 1e-06,
      "loss": 0.4442,
      "num_input_tokens_seen": 397496316,
      "step": 7094
    },
    {
      "epoch": 15.799554565701559,
      "loss": 0.39536595344543457,
      "loss_ce": 0.0001022747892420739,
      "loss_iou": 0.1796875,
      "loss_num": 0.0069580078125,
      "loss_xval": 0.39453125,
      "num_input_tokens_seen": 397496316,
      "step": 7094
    },
    {
      "epoch": 15.801781737193764,
      "grad_norm": 22.2495059967041,
      "learning_rate": 1e-06,
      "loss": 0.482,
      "num_input_tokens_seen": 397551668,
      "step": 7095
    },
    {
      "epoch": 15.801781737193764,
      "loss": 0.5242829918861389,
      "loss_ce": 0.00029618252301588655,
      "loss_iou": 0.2138671875,
      "loss_num": 0.0194091796875,
      "loss_xval": 0.5234375,
      "num_input_tokens_seen": 397551668,
      "step": 7095
    },
    {
      "epoch": 15.804008908685969,
      "grad_norm": 14.987038612365723,
      "learning_rate": 1e-06,
      "loss": 0.4245,
      "num_input_tokens_seen": 397609464,
      "step": 7096
    },
    {
      "epoch": 15.804008908685969,
      "loss": 0.41660401225090027,
      "loss_ce": 0.00010012378334067762,
      "loss_iou": 0.1748046875,
      "loss_num": 0.01336669921875,
      "loss_xval": 0.416015625,
      "num_input_tokens_seen": 397609464,
      "step": 7096
    },
    {
      "epoch": 15.806236080178174,
      "grad_norm": 14.618922233581543,
      "learning_rate": 1e-06,
      "loss": 0.4714,
      "num_input_tokens_seen": 397667576,
      "step": 7097
    },
    {
      "epoch": 15.806236080178174,
      "loss": 0.5953065156936646,
      "loss_ce": 9.167171083390713e-05,
      "loss_iou": 0.275390625,
      "loss_num": 0.00848388671875,
      "loss_xval": 0.59375,
      "num_input_tokens_seen": 397667576,
      "step": 7097
    },
    {
      "epoch": 15.808463251670378,
      "grad_norm": 14.448540687561035,
      "learning_rate": 1e-06,
      "loss": 0.4018,
      "num_input_tokens_seen": 397722704,
      "step": 7098
    },
    {
      "epoch": 15.808463251670378,
      "loss": 0.486581951379776,
      "loss_ce": 0.00010122812818735838,
      "loss_iou": 0.1943359375,
      "loss_num": 0.0196533203125,
      "loss_xval": 0.486328125,
      "num_input_tokens_seen": 397722704,
      "step": 7098
    },
    {
      "epoch": 15.810690423162583,
      "grad_norm": 15.517051696777344,
      "learning_rate": 1e-06,
      "loss": 0.3963,
      "num_input_tokens_seen": 397776580,
      "step": 7099
    },
    {
      "epoch": 15.810690423162583,
      "loss": 0.36538606882095337,
      "loss_ce": 0.00012117931328248233,
      "loss_iou": 0.158203125,
      "loss_num": 0.00982666015625,
      "loss_xval": 0.365234375,
      "num_input_tokens_seen": 397776580,
      "step": 7099
    },
    {
      "epoch": 15.812917594654788,
      "grad_norm": 23.810832977294922,
      "learning_rate": 1e-06,
      "loss": 0.4988,
      "num_input_tokens_seen": 397832360,
      "step": 7100
    },
    {
      "epoch": 15.812917594654788,
      "loss": 0.6032587885856628,
      "loss_ce": 0.00010939198546111584,
      "loss_iou": 0.265625,
      "loss_num": 0.01385498046875,
      "loss_xval": 0.6015625,
      "num_input_tokens_seen": 397832360,
      "step": 7100
    },
    {
      "epoch": 15.815144766146993,
      "grad_norm": 15.830489158630371,
      "learning_rate": 1e-06,
      "loss": 0.5576,
      "num_input_tokens_seen": 397888856,
      "step": 7101
    },
    {
      "epoch": 15.815144766146993,
      "loss": 0.607633113861084,
      "loss_ce": 8.912877819966525e-05,
      "loss_iou": 0.2470703125,
      "loss_num": 0.0228271484375,
      "loss_xval": 0.609375,
      "num_input_tokens_seen": 397888856,
      "step": 7101
    },
    {
      "epoch": 15.817371937639198,
      "grad_norm": 19.18913459777832,
      "learning_rate": 1e-06,
      "loss": 0.4931,
      "num_input_tokens_seen": 397941060,
      "step": 7102
    },
    {
      "epoch": 15.817371937639198,
      "loss": 0.3700322210788727,
      "loss_ce": 9.815287921810523e-05,
      "loss_iou": 0.1484375,
      "loss_num": 0.0145263671875,
      "loss_xval": 0.369140625,
      "num_input_tokens_seen": 397941060,
      "step": 7102
    },
    {
      "epoch": 15.819599109131403,
      "grad_norm": 13.465466499328613,
      "learning_rate": 1e-06,
      "loss": 0.3357,
      "num_input_tokens_seen": 397996392,
      "step": 7103
    },
    {
      "epoch": 15.819599109131403,
      "loss": 0.395857036113739,
      "loss_ce": 0.00010508089326322079,
      "loss_iou": 0.177734375,
      "loss_num": 0.00811767578125,
      "loss_xval": 0.396484375,
      "num_input_tokens_seen": 397996392,
      "step": 7103
    },
    {
      "epoch": 15.821826280623608,
      "grad_norm": 17.05833625793457,
      "learning_rate": 1e-06,
      "loss": 0.4311,
      "num_input_tokens_seen": 398053248,
      "step": 7104
    },
    {
      "epoch": 15.821826280623608,
      "loss": 0.4999108910560608,
      "loss_ce": 0.0001550541928736493,
      "loss_iou": 0.2060546875,
      "loss_num": 0.0177001953125,
      "loss_xval": 0.5,
      "num_input_tokens_seen": 398053248,
      "step": 7104
    },
    {
      "epoch": 15.824053452115812,
      "grad_norm": 20.213592529296875,
      "learning_rate": 1e-06,
      "loss": 0.4103,
      "num_input_tokens_seen": 398110088,
      "step": 7105
    },
    {
      "epoch": 15.824053452115812,
      "loss": 0.5278196334838867,
      "loss_ce": 0.00010968356218654662,
      "loss_iou": 0.224609375,
      "loss_num": 0.015625,
      "loss_xval": 0.52734375,
      "num_input_tokens_seen": 398110088,
      "step": 7105
    },
    {
      "epoch": 15.826280623608017,
      "grad_norm": 17.705123901367188,
      "learning_rate": 1e-06,
      "loss": 0.515,
      "num_input_tokens_seen": 398168524,
      "step": 7106
    },
    {
      "epoch": 15.826280623608017,
      "loss": 0.39392590522766113,
      "loss_ce": 0.0007374440901912749,
      "loss_iou": 0.1630859375,
      "loss_num": 0.0133056640625,
      "loss_xval": 0.392578125,
      "num_input_tokens_seen": 398168524,
      "step": 7106
    },
    {
      "epoch": 15.828507795100222,
      "grad_norm": 11.036231994628906,
      "learning_rate": 1e-06,
      "loss": 0.5384,
      "num_input_tokens_seen": 398226688,
      "step": 7107
    },
    {
      "epoch": 15.828507795100222,
      "loss": 0.5651922225952148,
      "loss_ce": 0.00012873421655967832,
      "loss_iou": 0.2314453125,
      "loss_num": 0.0205078125,
      "loss_xval": 0.56640625,
      "num_input_tokens_seen": 398226688,
      "step": 7107
    },
    {
      "epoch": 15.830734966592427,
      "grad_norm": 19.0125789642334,
      "learning_rate": 1e-06,
      "loss": 0.4158,
      "num_input_tokens_seen": 398282356,
      "step": 7108
    },
    {
      "epoch": 15.830734966592427,
      "loss": 0.28864729404449463,
      "loss_ce": 7.308388012461364e-05,
      "loss_iou": 0.12451171875,
      "loss_num": 0.00799560546875,
      "loss_xval": 0.2890625,
      "num_input_tokens_seen": 398282356,
      "step": 7108
    },
    {
      "epoch": 15.832962138084632,
      "grad_norm": 21.866071701049805,
      "learning_rate": 1e-06,
      "loss": 0.4139,
      "num_input_tokens_seen": 398339436,
      "step": 7109
    },
    {
      "epoch": 15.832962138084632,
      "loss": 0.5395206809043884,
      "loss_ce": 9.19756930670701e-05,
      "loss_iou": 0.2421875,
      "loss_num": 0.0108642578125,
      "loss_xval": 0.5390625,
      "num_input_tokens_seen": 398339436,
      "step": 7109
    },
    {
      "epoch": 15.835189309576837,
      "grad_norm": 25.225812911987305,
      "learning_rate": 1e-06,
      "loss": 0.5883,
      "num_input_tokens_seen": 398395324,
      "step": 7110
    },
    {
      "epoch": 15.835189309576837,
      "loss": 0.5729167461395264,
      "loss_ce": 0.00016287056496366858,
      "loss_iou": 0.25390625,
      "loss_num": 0.01318359375,
      "loss_xval": 0.57421875,
      "num_input_tokens_seen": 398395324,
      "step": 7110
    },
    {
      "epoch": 15.837416481069042,
      "grad_norm": 17.75248908996582,
      "learning_rate": 1e-06,
      "loss": 0.3587,
      "num_input_tokens_seen": 398450532,
      "step": 7111
    },
    {
      "epoch": 15.837416481069042,
      "loss": 0.3196753263473511,
      "loss_ce": 9.522202162770554e-05,
      "loss_iou": 0.142578125,
      "loss_num": 0.00677490234375,
      "loss_xval": 0.3203125,
      "num_input_tokens_seen": 398450532,
      "step": 7111
    },
    {
      "epoch": 15.839643652561247,
      "grad_norm": 18.968854904174805,
      "learning_rate": 1e-06,
      "loss": 0.3703,
      "num_input_tokens_seen": 398505144,
      "step": 7112
    },
    {
      "epoch": 15.839643652561247,
      "loss": 0.3004349172115326,
      "loss_ce": 8.092600910458714e-05,
      "loss_iou": 0.134765625,
      "loss_num": 0.006195068359375,
      "loss_xval": 0.30078125,
      "num_input_tokens_seen": 398505144,
      "step": 7112
    },
    {
      "epoch": 15.841870824053451,
      "grad_norm": 18.018905639648438,
      "learning_rate": 1e-06,
      "loss": 0.3754,
      "num_input_tokens_seen": 398560608,
      "step": 7113
    },
    {
      "epoch": 15.841870824053451,
      "loss": 0.3709629476070404,
      "loss_ce": 0.0001133316254708916,
      "loss_iou": 0.1611328125,
      "loss_num": 0.00958251953125,
      "loss_xval": 0.37109375,
      "num_input_tokens_seen": 398560608,
      "step": 7113
    },
    {
      "epoch": 15.844097995545656,
      "grad_norm": 24.227947235107422,
      "learning_rate": 1e-06,
      "loss": 0.4965,
      "num_input_tokens_seen": 398617888,
      "step": 7114
    },
    {
      "epoch": 15.844097995545656,
      "loss": 0.5535033941268921,
      "loss_ce": 9.764648712007329e-05,
      "loss_iou": 0.2421875,
      "loss_num": 0.013671875,
      "loss_xval": 0.5546875,
      "num_input_tokens_seen": 398617888,
      "step": 7114
    },
    {
      "epoch": 15.846325167037861,
      "grad_norm": 23.27136993408203,
      "learning_rate": 1e-06,
      "loss": 0.58,
      "num_input_tokens_seen": 398671732,
      "step": 7115
    },
    {
      "epoch": 15.846325167037861,
      "loss": 0.7010478973388672,
      "loss_ce": 0.00010489902342669666,
      "loss_iou": 0.3046875,
      "loss_num": 0.018310546875,
      "loss_xval": 0.69921875,
      "num_input_tokens_seen": 398671732,
      "step": 7115
    },
    {
      "epoch": 15.848552338530066,
      "grad_norm": 18.09553337097168,
      "learning_rate": 1e-06,
      "loss": 0.4632,
      "num_input_tokens_seen": 398728176,
      "step": 7116
    },
    {
      "epoch": 15.848552338530066,
      "loss": 0.4016414284706116,
      "loss_ce": 9.112786210607737e-05,
      "loss_iou": 0.1689453125,
      "loss_num": 0.0125732421875,
      "loss_xval": 0.40234375,
      "num_input_tokens_seen": 398728176,
      "step": 7116
    },
    {
      "epoch": 15.85077951002227,
      "grad_norm": 17.273834228515625,
      "learning_rate": 1e-06,
      "loss": 0.6624,
      "num_input_tokens_seen": 398783004,
      "step": 7117
    },
    {
      "epoch": 15.85077951002227,
      "loss": 0.753657341003418,
      "loss_ce": 0.00011729233665391803,
      "loss_iou": 0.26953125,
      "loss_num": 0.042724609375,
      "loss_xval": 0.75390625,
      "num_input_tokens_seen": 398783004,
      "step": 7117
    },
    {
      "epoch": 15.853006681514476,
      "grad_norm": 12.48859977722168,
      "learning_rate": 1e-06,
      "loss": 0.4666,
      "num_input_tokens_seen": 398842220,
      "step": 7118
    },
    {
      "epoch": 15.853006681514476,
      "loss": 0.3523826599121094,
      "loss_ce": 8.77102866070345e-05,
      "loss_iou": 0.138671875,
      "loss_num": 0.01513671875,
      "loss_xval": 0.3515625,
      "num_input_tokens_seen": 398842220,
      "step": 7118
    },
    {
      "epoch": 15.855233853006682,
      "grad_norm": 13.259708404541016,
      "learning_rate": 1e-06,
      "loss": 0.4148,
      "num_input_tokens_seen": 398899104,
      "step": 7119
    },
    {
      "epoch": 15.855233853006682,
      "loss": 0.5045324563980103,
      "loss_ce": 0.00013790541561320424,
      "loss_iou": 0.1923828125,
      "loss_num": 0.0240478515625,
      "loss_xval": 0.50390625,
      "num_input_tokens_seen": 398899104,
      "step": 7119
    },
    {
      "epoch": 15.857461024498887,
      "grad_norm": 18.383527755737305,
      "learning_rate": 1e-06,
      "loss": 0.4292,
      "num_input_tokens_seen": 398952944,
      "step": 7120
    },
    {
      "epoch": 15.857461024498887,
      "loss": 0.42771950364112854,
      "loss_ce": 0.00010719949204940349,
      "loss_iou": 0.2001953125,
      "loss_num": 0.005645751953125,
      "loss_xval": 0.427734375,
      "num_input_tokens_seen": 398952944,
      "step": 7120
    },
    {
      "epoch": 15.859688195991092,
      "grad_norm": 44.14472579956055,
      "learning_rate": 1e-06,
      "loss": 0.4647,
      "num_input_tokens_seen": 399007520,
      "step": 7121
    },
    {
      "epoch": 15.859688195991092,
      "loss": 0.501062273979187,
      "loss_ce": 8.574766980018467e-05,
      "loss_iou": 0.201171875,
      "loss_num": 0.0198974609375,
      "loss_xval": 0.5,
      "num_input_tokens_seen": 399007520,
      "step": 7121
    },
    {
      "epoch": 15.861915367483297,
      "grad_norm": 19.604982376098633,
      "learning_rate": 1e-06,
      "loss": 0.5346,
      "num_input_tokens_seen": 399062428,
      "step": 7122
    },
    {
      "epoch": 15.861915367483297,
      "loss": 0.5388393998146057,
      "loss_ce": 8.2079553976655e-05,
      "loss_iou": 0.220703125,
      "loss_num": 0.0194091796875,
      "loss_xval": 0.5390625,
      "num_input_tokens_seen": 399062428,
      "step": 7122
    },
    {
      "epoch": 15.864142538975502,
      "grad_norm": 15.576171875,
      "learning_rate": 1e-06,
      "loss": 0.4607,
      "num_input_tokens_seen": 399117136,
      "step": 7123
    },
    {
      "epoch": 15.864142538975502,
      "loss": 0.6172924637794495,
      "loss_ce": 0.00010496602772036567,
      "loss_iou": 0.28515625,
      "loss_num": 0.00933837890625,
      "loss_xval": 0.6171875,
      "num_input_tokens_seen": 399117136,
      "step": 7123
    },
    {
      "epoch": 15.866369710467707,
      "grad_norm": 20.8918514251709,
      "learning_rate": 1e-06,
      "loss": 0.4582,
      "num_input_tokens_seen": 399172568,
      "step": 7124
    },
    {
      "epoch": 15.866369710467707,
      "loss": 0.3362141251564026,
      "loss_ce": 9.351145854452625e-05,
      "loss_iou": 0.142578125,
      "loss_num": 0.01025390625,
      "loss_xval": 0.3359375,
      "num_input_tokens_seen": 399172568,
      "step": 7124
    },
    {
      "epoch": 15.868596881959911,
      "grad_norm": 14.206762313842773,
      "learning_rate": 1e-06,
      "loss": 0.3939,
      "num_input_tokens_seen": 399229784,
      "step": 7125
    },
    {
      "epoch": 15.868596881959911,
      "loss": 0.3726486265659332,
      "loss_ce": 9.005493484437466e-05,
      "loss_iou": 0.154296875,
      "loss_num": 0.01300048828125,
      "loss_xval": 0.373046875,
      "num_input_tokens_seen": 399229784,
      "step": 7125
    },
    {
      "epoch": 15.870824053452116,
      "grad_norm": 40.32673263549805,
      "learning_rate": 1e-06,
      "loss": 0.4629,
      "num_input_tokens_seen": 399285780,
      "step": 7126
    },
    {
      "epoch": 15.870824053452116,
      "loss": 0.49388349056243896,
      "loss_ce": 0.00010904869122896343,
      "loss_iou": 0.234375,
      "loss_num": 0.004852294921875,
      "loss_xval": 0.494140625,
      "num_input_tokens_seen": 399285780,
      "step": 7126
    },
    {
      "epoch": 15.873051224944321,
      "grad_norm": 13.4733304977417,
      "learning_rate": 1e-06,
      "loss": 0.5213,
      "num_input_tokens_seen": 399342376,
      "step": 7127
    },
    {
      "epoch": 15.873051224944321,
      "loss": 0.5921471118927002,
      "loss_ce": 0.00010613157064653933,
      "loss_iou": 0.265625,
      "loss_num": 0.01226806640625,
      "loss_xval": 0.59375,
      "num_input_tokens_seen": 399342376,
      "step": 7127
    },
    {
      "epoch": 15.875278396436526,
      "grad_norm": 15.447736740112305,
      "learning_rate": 1e-06,
      "loss": 0.4606,
      "num_input_tokens_seen": 399400404,
      "step": 7128
    },
    {
      "epoch": 15.875278396436526,
      "loss": 0.5045484304428101,
      "loss_ce": 0.00015390958287753165,
      "loss_iou": 0.2158203125,
      "loss_num": 0.0146484375,
      "loss_xval": 0.50390625,
      "num_input_tokens_seen": 399400404,
      "step": 7128
    },
    {
      "epoch": 15.877505567928731,
      "grad_norm": 15.931106567382812,
      "learning_rate": 1e-06,
      "loss": 0.4239,
      "num_input_tokens_seen": 399455484,
      "step": 7129
    },
    {
      "epoch": 15.877505567928731,
      "loss": 0.4633306860923767,
      "loss_ce": 0.00031800862052477896,
      "loss_iou": 0.19921875,
      "loss_num": 0.0128173828125,
      "loss_xval": 0.462890625,
      "num_input_tokens_seen": 399455484,
      "step": 7129
    },
    {
      "epoch": 15.879732739420936,
      "grad_norm": 17.251869201660156,
      "learning_rate": 1e-06,
      "loss": 0.3316,
      "num_input_tokens_seen": 399510620,
      "step": 7130
    },
    {
      "epoch": 15.879732739420936,
      "loss": 0.31160733103752136,
      "loss_ce": 8.388744026888162e-05,
      "loss_iou": 0.1318359375,
      "loss_num": 0.00946044921875,
      "loss_xval": 0.3125,
      "num_input_tokens_seen": 399510620,
      "step": 7130
    },
    {
      "epoch": 15.88195991091314,
      "grad_norm": 20.6621150970459,
      "learning_rate": 1e-06,
      "loss": 0.367,
      "num_input_tokens_seen": 399567148,
      "step": 7131
    },
    {
      "epoch": 15.88195991091314,
      "loss": 0.39595168828964233,
      "loss_ce": 7.766317867208272e-05,
      "loss_iou": 0.18359375,
      "loss_num": 0.005767822265625,
      "loss_xval": 0.396484375,
      "num_input_tokens_seen": 399567148,
      "step": 7131
    },
    {
      "epoch": 15.884187082405345,
      "grad_norm": 20.26247215270996,
      "learning_rate": 1e-06,
      "loss": 0.5707,
      "num_input_tokens_seen": 399619636,
      "step": 7132
    },
    {
      "epoch": 15.884187082405345,
      "loss": 0.6330655217170715,
      "loss_ce": 0.000130962478579022,
      "loss_iou": 0.267578125,
      "loss_num": 0.02001953125,
      "loss_xval": 0.6328125,
      "num_input_tokens_seen": 399619636,
      "step": 7132
    },
    {
      "epoch": 15.88641425389755,
      "grad_norm": 19.18486213684082,
      "learning_rate": 1e-06,
      "loss": 0.5153,
      "num_input_tokens_seen": 399674064,
      "step": 7133
    },
    {
      "epoch": 15.88641425389755,
      "loss": 0.5503383874893188,
      "loss_ce": 0.00010647171438904479,
      "loss_iou": 0.228515625,
      "loss_num": 0.0185546875,
      "loss_xval": 0.55078125,
      "num_input_tokens_seen": 399674064,
      "step": 7133
    },
    {
      "epoch": 15.888641425389755,
      "grad_norm": 16.185976028442383,
      "learning_rate": 1e-06,
      "loss": 0.3956,
      "num_input_tokens_seen": 399728452,
      "step": 7134
    },
    {
      "epoch": 15.888641425389755,
      "loss": 0.24313417077064514,
      "loss_ce": 9.21574974199757e-05,
      "loss_iou": 0.10888671875,
      "loss_num": 0.005096435546875,
      "loss_xval": 0.2431640625,
      "num_input_tokens_seen": 399728452,
      "step": 7134
    },
    {
      "epoch": 15.89086859688196,
      "grad_norm": 18.77434539794922,
      "learning_rate": 1e-06,
      "loss": 0.5478,
      "num_input_tokens_seen": 399783984,
      "step": 7135
    },
    {
      "epoch": 15.89086859688196,
      "loss": 0.4603026211261749,
      "loss_ce": 9.756326471688226e-05,
      "loss_iou": 0.203125,
      "loss_num": 0.0106201171875,
      "loss_xval": 0.4609375,
      "num_input_tokens_seen": 399783984,
      "step": 7135
    },
    {
      "epoch": 15.893095768374165,
      "grad_norm": 17.734636306762695,
      "learning_rate": 1e-06,
      "loss": 0.53,
      "num_input_tokens_seen": 399839268,
      "step": 7136
    },
    {
      "epoch": 15.893095768374165,
      "loss": 0.6124453544616699,
      "loss_ce": 0.00014065181312616915,
      "loss_iou": 0.251953125,
      "loss_num": 0.0216064453125,
      "loss_xval": 0.61328125,
      "num_input_tokens_seen": 399839268,
      "step": 7136
    },
    {
      "epoch": 15.89532293986637,
      "grad_norm": 20.12199592590332,
      "learning_rate": 1e-06,
      "loss": 0.4646,
      "num_input_tokens_seen": 399896836,
      "step": 7137
    },
    {
      "epoch": 15.89532293986637,
      "loss": 0.5352360010147095,
      "loss_ce": 7.976061169756576e-05,
      "loss_iou": 0.2333984375,
      "loss_num": 0.013427734375,
      "loss_xval": 0.53515625,
      "num_input_tokens_seen": 399896836,
      "step": 7137
    },
    {
      "epoch": 15.897550111358575,
      "grad_norm": 20.401044845581055,
      "learning_rate": 1e-06,
      "loss": 0.5685,
      "num_input_tokens_seen": 399954484,
      "step": 7138
    },
    {
      "epoch": 15.897550111358575,
      "loss": 0.512428879737854,
      "loss_ce": 9.979259630199522e-05,
      "loss_iou": 0.212890625,
      "loss_num": 0.0172119140625,
      "loss_xval": 0.51171875,
      "num_input_tokens_seen": 399954484,
      "step": 7138
    },
    {
      "epoch": 15.89977728285078,
      "grad_norm": 33.50958251953125,
      "learning_rate": 1e-06,
      "loss": 0.4488,
      "num_input_tokens_seen": 400013840,
      "step": 7139
    },
    {
      "epoch": 15.89977728285078,
      "loss": 0.44518011808395386,
      "loss_ce": 0.00015754257037770003,
      "loss_iou": 0.19140625,
      "loss_num": 0.01263427734375,
      "loss_xval": 0.4453125,
      "num_input_tokens_seen": 400013840,
      "step": 7139
    },
    {
      "epoch": 15.902004454342984,
      "grad_norm": 14.90358829498291,
      "learning_rate": 1e-06,
      "loss": 0.6293,
      "num_input_tokens_seen": 400068508,
      "step": 7140
    },
    {
      "epoch": 15.902004454342984,
      "loss": 0.7256982326507568,
      "loss_ce": 0.00011228243238292634,
      "loss_iou": 0.294921875,
      "loss_num": 0.0277099609375,
      "loss_xval": 0.7265625,
      "num_input_tokens_seen": 400068508,
      "step": 7140
    },
    {
      "epoch": 15.90423162583519,
      "grad_norm": 15.052915573120117,
      "learning_rate": 1e-06,
      "loss": 0.3924,
      "num_input_tokens_seen": 400125004,
      "step": 7141
    },
    {
      "epoch": 15.90423162583519,
      "loss": 0.49386632442474365,
      "loss_ce": 0.00015298393554985523,
      "loss_iou": 0.1884765625,
      "loss_num": 0.0235595703125,
      "loss_xval": 0.494140625,
      "num_input_tokens_seen": 400125004,
      "step": 7141
    },
    {
      "epoch": 15.906458797327394,
      "grad_norm": 21.69573402404785,
      "learning_rate": 1e-06,
      "loss": 0.4718,
      "num_input_tokens_seen": 400179284,
      "step": 7142
    },
    {
      "epoch": 15.906458797327394,
      "loss": 0.2900165319442749,
      "loss_ce": 9.953400876838714e-05,
      "loss_iou": 0.1279296875,
      "loss_num": 0.0067138671875,
      "loss_xval": 0.2890625,
      "num_input_tokens_seen": 400179284,
      "step": 7142
    },
    {
      "epoch": 15.908685968819599,
      "grad_norm": 17.714357376098633,
      "learning_rate": 1e-06,
      "loss": 0.5839,
      "num_input_tokens_seen": 400233780,
      "step": 7143
    },
    {
      "epoch": 15.908685968819599,
      "loss": 0.5741137862205505,
      "loss_ce": 0.0001391808473272249,
      "loss_iou": 0.265625,
      "loss_num": 0.0089111328125,
      "loss_xval": 0.57421875,
      "num_input_tokens_seen": 400233780,
      "step": 7143
    },
    {
      "epoch": 15.910913140311804,
      "grad_norm": 32.93324279785156,
      "learning_rate": 1e-06,
      "loss": 0.5251,
      "num_input_tokens_seen": 400291408,
      "step": 7144
    },
    {
      "epoch": 15.910913140311804,
      "loss": 0.4064697325229645,
      "loss_ce": 9.767108713276684e-05,
      "loss_iou": 0.1806640625,
      "loss_num": 0.00897216796875,
      "loss_xval": 0.40625,
      "num_input_tokens_seen": 400291408,
      "step": 7144
    },
    {
      "epoch": 15.913140311804009,
      "grad_norm": 19.784011840820312,
      "learning_rate": 1e-06,
      "loss": 0.5329,
      "num_input_tokens_seen": 400348664,
      "step": 7145
    },
    {
      "epoch": 15.913140311804009,
      "loss": 0.4993619918823242,
      "loss_ce": 9.443063754588366e-05,
      "loss_iou": 0.216796875,
      "loss_num": 0.01300048828125,
      "loss_xval": 0.5,
      "num_input_tokens_seen": 400348664,
      "step": 7145
    },
    {
      "epoch": 15.915367483296214,
      "grad_norm": 17.10857391357422,
      "learning_rate": 1e-06,
      "loss": 0.4308,
      "num_input_tokens_seen": 400402800,
      "step": 7146
    },
    {
      "epoch": 15.915367483296214,
      "loss": 0.5981404781341553,
      "loss_ce": 0.0002400614321231842,
      "loss_iou": 0.251953125,
      "loss_num": 0.0191650390625,
      "loss_xval": 0.59765625,
      "num_input_tokens_seen": 400402800,
      "step": 7146
    },
    {
      "epoch": 15.917594654788418,
      "grad_norm": 18.01358413696289,
      "learning_rate": 1e-06,
      "loss": 0.4396,
      "num_input_tokens_seen": 400460196,
      "step": 7147
    },
    {
      "epoch": 15.917594654788418,
      "loss": 0.524014949798584,
      "loss_ce": 8.919274841900915e-05,
      "loss_iou": 0.208984375,
      "loss_num": 0.0211181640625,
      "loss_xval": 0.5234375,
      "num_input_tokens_seen": 400460196,
      "step": 7147
    },
    {
      "epoch": 15.919821826280623,
      "grad_norm": 16.308366775512695,
      "learning_rate": 1e-06,
      "loss": 0.5253,
      "num_input_tokens_seen": 400517080,
      "step": 7148
    },
    {
      "epoch": 15.919821826280623,
      "loss": 0.4072136878967285,
      "loss_ce": 0.00010919298802036792,
      "loss_iou": 0.1591796875,
      "loss_num": 0.017578125,
      "loss_xval": 0.40625,
      "num_input_tokens_seen": 400517080,
      "step": 7148
    },
    {
      "epoch": 15.922048997772828,
      "grad_norm": 14.491548538208008,
      "learning_rate": 1e-06,
      "loss": 0.4755,
      "num_input_tokens_seen": 400572672,
      "step": 7149
    },
    {
      "epoch": 15.922048997772828,
      "loss": 0.4918016791343689,
      "loss_ce": 0.0001024569064611569,
      "loss_iou": 0.224609375,
      "loss_num": 0.0084228515625,
      "loss_xval": 0.4921875,
      "num_input_tokens_seen": 400572672,
      "step": 7149
    },
    {
      "epoch": 15.924276169265033,
      "grad_norm": 25.39553451538086,
      "learning_rate": 1e-06,
      "loss": 0.5337,
      "num_input_tokens_seen": 400628244,
      "step": 7150
    },
    {
      "epoch": 15.924276169265033,
      "loss": 0.423916757106781,
      "loss_ce": 8.863602124620229e-05,
      "loss_iou": 0.1728515625,
      "loss_num": 0.015625,
      "loss_xval": 0.423828125,
      "num_input_tokens_seen": 400628244,
      "step": 7150
    },
    {
      "epoch": 15.926503340757238,
      "grad_norm": 18.745763778686523,
      "learning_rate": 1e-06,
      "loss": 0.4148,
      "num_input_tokens_seen": 400685080,
      "step": 7151
    },
    {
      "epoch": 15.926503340757238,
      "loss": 0.4842875599861145,
      "loss_ce": 0.00012620513734873384,
      "loss_iou": 0.189453125,
      "loss_num": 0.021240234375,
      "loss_xval": 0.484375,
      "num_input_tokens_seen": 400685080,
      "step": 7151
    },
    {
      "epoch": 15.928730512249443,
      "grad_norm": 21.89429473876953,
      "learning_rate": 1e-06,
      "loss": 0.4732,
      "num_input_tokens_seen": 400736332,
      "step": 7152
    },
    {
      "epoch": 15.928730512249443,
      "loss": 0.4724288582801819,
      "loss_ce": 7.779937004670501e-05,
      "loss_iou": 0.197265625,
      "loss_num": 0.015869140625,
      "loss_xval": 0.47265625,
      "num_input_tokens_seen": 400736332,
      "step": 7152
    },
    {
      "epoch": 15.930957683741648,
      "grad_norm": 22.908029556274414,
      "learning_rate": 1e-06,
      "loss": 0.2874,
      "num_input_tokens_seen": 400792008,
      "step": 7153
    },
    {
      "epoch": 15.930957683741648,
      "loss": 0.2497827112674713,
      "loss_ce": 8.78690043464303e-05,
      "loss_iou": 0.11376953125,
      "loss_num": 0.004364013671875,
      "loss_xval": 0.25,
      "num_input_tokens_seen": 400792008,
      "step": 7153
    },
    {
      "epoch": 15.933184855233852,
      "grad_norm": 16.189254760742188,
      "learning_rate": 1e-06,
      "loss": 0.4473,
      "num_input_tokens_seen": 400849396,
      "step": 7154
    },
    {
      "epoch": 15.933184855233852,
      "loss": 0.4695594310760498,
      "loss_ce": 7.704535528318956e-05,
      "loss_iou": 0.208984375,
      "loss_num": 0.0103759765625,
      "loss_xval": 0.46875,
      "num_input_tokens_seen": 400849396,
      "step": 7154
    },
    {
      "epoch": 15.935412026726057,
      "grad_norm": 19.59444236755371,
      "learning_rate": 1e-06,
      "loss": 0.4609,
      "num_input_tokens_seen": 400905600,
      "step": 7155
    },
    {
      "epoch": 15.935412026726057,
      "loss": 0.26319292187690735,
      "loss_ce": 0.0001313859538640827,
      "loss_iou": 0.1103515625,
      "loss_num": 0.00836181640625,
      "loss_xval": 0.263671875,
      "num_input_tokens_seen": 400905600,
      "step": 7155
    },
    {
      "epoch": 15.937639198218262,
      "grad_norm": 14.831480979919434,
      "learning_rate": 1e-06,
      "loss": 0.5633,
      "num_input_tokens_seen": 400964476,
      "step": 7156
    },
    {
      "epoch": 15.937639198218262,
      "loss": 0.6740528345108032,
      "loss_ce": 0.00010258887050440535,
      "loss_iou": 0.275390625,
      "loss_num": 0.0247802734375,
      "loss_xval": 0.67578125,
      "num_input_tokens_seen": 400964476,
      "step": 7156
    },
    {
      "epoch": 15.939866369710467,
      "grad_norm": 22.283842086791992,
      "learning_rate": 1e-06,
      "loss": 0.4229,
      "num_input_tokens_seen": 401020968,
      "step": 7157
    },
    {
      "epoch": 15.939866369710467,
      "loss": 0.48083391785621643,
      "loss_ce": 9.051487722899765e-05,
      "loss_iou": 0.2138671875,
      "loss_num": 0.0107421875,
      "loss_xval": 0.48046875,
      "num_input_tokens_seen": 401020968,
      "step": 7157
    },
    {
      "epoch": 15.942093541202672,
      "grad_norm": 13.97571849822998,
      "learning_rate": 1e-06,
      "loss": 0.4721,
      "num_input_tokens_seen": 401075944,
      "step": 7158
    },
    {
      "epoch": 15.942093541202672,
      "loss": 0.3242437243461609,
      "loss_ce": 8.600985893281177e-05,
      "loss_iou": 0.1484375,
      "loss_num": 0.00555419921875,
      "loss_xval": 0.32421875,
      "num_input_tokens_seen": 401075944,
      "step": 7158
    },
    {
      "epoch": 15.944320712694877,
      "grad_norm": 25.421710968017578,
      "learning_rate": 1e-06,
      "loss": 0.7803,
      "num_input_tokens_seen": 401130976,
      "step": 7159
    },
    {
      "epoch": 15.944320712694877,
      "loss": 0.8276197910308838,
      "loss_ce": 0.00010511695290915668,
      "loss_iou": 0.3359375,
      "loss_num": 0.031494140625,
      "loss_xval": 0.828125,
      "num_input_tokens_seen": 401130976,
      "step": 7159
    },
    {
      "epoch": 15.946547884187082,
      "grad_norm": 39.614234924316406,
      "learning_rate": 1e-06,
      "loss": 0.4332,
      "num_input_tokens_seen": 401186464,
      "step": 7160
    },
    {
      "epoch": 15.946547884187082,
      "loss": 0.483007550239563,
      "loss_ce": 9.739773668115959e-05,
      "loss_iou": 0.19921875,
      "loss_num": 0.0167236328125,
      "loss_xval": 0.482421875,
      "num_input_tokens_seen": 401186464,
      "step": 7160
    },
    {
      "epoch": 15.948775055679288,
      "grad_norm": 16.9848690032959,
      "learning_rate": 1e-06,
      "loss": 0.5029,
      "num_input_tokens_seen": 401242136,
      "step": 7161
    },
    {
      "epoch": 15.948775055679288,
      "loss": 0.5525789260864258,
      "loss_ce": 8.865697600413114e-05,
      "loss_iou": 0.232421875,
      "loss_num": 0.0174560546875,
      "loss_xval": 0.55078125,
      "num_input_tokens_seen": 401242136,
      "step": 7161
    },
    {
      "epoch": 15.951002227171493,
      "grad_norm": 19.457195281982422,
      "learning_rate": 1e-06,
      "loss": 0.4659,
      "num_input_tokens_seen": 401297980,
      "step": 7162
    },
    {
      "epoch": 15.951002227171493,
      "loss": 0.28521448373794556,
      "loss_ce": 8.872566104400903e-05,
      "loss_iou": 0.1298828125,
      "loss_num": 0.005218505859375,
      "loss_xval": 0.28515625,
      "num_input_tokens_seen": 401297980,
      "step": 7162
    },
    {
      "epoch": 15.953229398663698,
      "grad_norm": 13.675963401794434,
      "learning_rate": 1e-06,
      "loss": 0.3121,
      "num_input_tokens_seen": 401355992,
      "step": 7163
    },
    {
      "epoch": 15.953229398663698,
      "loss": 0.300504207611084,
      "loss_ce": 8.916326623875648e-05,
      "loss_iou": 0.1376953125,
      "loss_num": 0.005035400390625,
      "loss_xval": 0.30078125,
      "num_input_tokens_seen": 401355992,
      "step": 7163
    },
    {
      "epoch": 15.955456570155903,
      "grad_norm": 17.08403205871582,
      "learning_rate": 1e-06,
      "loss": 0.6135,
      "num_input_tokens_seen": 401410732,
      "step": 7164
    },
    {
      "epoch": 15.955456570155903,
      "loss": 0.6625217199325562,
      "loss_ce": 0.0001681805297266692,
      "loss_iou": 0.27734375,
      "loss_num": 0.021484375,
      "loss_xval": 0.6640625,
      "num_input_tokens_seen": 401410732,
      "step": 7164
    },
    {
      "epoch": 15.957683741648108,
      "grad_norm": 16.065011978149414,
      "learning_rate": 1e-06,
      "loss": 0.4816,
      "num_input_tokens_seen": 401466736,
      "step": 7165
    },
    {
      "epoch": 15.957683741648108,
      "loss": 0.605895459651947,
      "loss_ce": 0.00018256741168443114,
      "loss_iou": 0.2392578125,
      "loss_num": 0.0255126953125,
      "loss_xval": 0.60546875,
      "num_input_tokens_seen": 401466736,
      "step": 7165
    },
    {
      "epoch": 15.959910913140313,
      "grad_norm": 15.503355979919434,
      "learning_rate": 1e-06,
      "loss": 0.5208,
      "num_input_tokens_seen": 401525152,
      "step": 7166
    },
    {
      "epoch": 15.959910913140313,
      "loss": 0.6843163967132568,
      "loss_ce": 0.00011229477968299761,
      "loss_iou": 0.306640625,
      "loss_num": 0.01446533203125,
      "loss_xval": 0.68359375,
      "num_input_tokens_seen": 401525152,
      "step": 7166
    },
    {
      "epoch": 15.962138084632517,
      "grad_norm": 23.097681045532227,
      "learning_rate": 1e-06,
      "loss": 0.3613,
      "num_input_tokens_seen": 401576820,
      "step": 7167
    },
    {
      "epoch": 15.962138084632517,
      "loss": 0.289186954498291,
      "loss_ce": 0.0001244572049472481,
      "loss_iou": 0.11083984375,
      "loss_num": 0.01348876953125,
      "loss_xval": 0.2890625,
      "num_input_tokens_seen": 401576820,
      "step": 7167
    },
    {
      "epoch": 15.964365256124722,
      "grad_norm": 13.968826293945312,
      "learning_rate": 1e-06,
      "loss": 0.3522,
      "num_input_tokens_seen": 401630452,
      "step": 7168
    },
    {
      "epoch": 15.964365256124722,
      "loss": 0.37191227078437805,
      "loss_ce": 8.611210068920627e-05,
      "loss_iou": 0.1611328125,
      "loss_num": 0.00994873046875,
      "loss_xval": 0.37109375,
      "num_input_tokens_seen": 401630452,
      "step": 7168
    },
    {
      "epoch": 15.966592427616927,
      "grad_norm": 16.292312622070312,
      "learning_rate": 1e-06,
      "loss": 0.3667,
      "num_input_tokens_seen": 401686368,
      "step": 7169
    },
    {
      "epoch": 15.966592427616927,
      "loss": 0.37168073654174805,
      "loss_ce": 9.872023656498641e-05,
      "loss_iou": 0.16796875,
      "loss_num": 0.006988525390625,
      "loss_xval": 0.37109375,
      "num_input_tokens_seen": 401686368,
      "step": 7169
    },
    {
      "epoch": 15.968819599109132,
      "grad_norm": 30.845420837402344,
      "learning_rate": 1e-06,
      "loss": 0.4289,
      "num_input_tokens_seen": 401743020,
      "step": 7170
    },
    {
      "epoch": 15.968819599109132,
      "loss": 0.4194309413433075,
      "loss_ce": 0.00011940038530156016,
      "loss_iou": 0.1787109375,
      "loss_num": 0.01220703125,
      "loss_xval": 0.419921875,
      "num_input_tokens_seen": 401743020,
      "step": 7170
    },
    {
      "epoch": 15.971046770601337,
      "grad_norm": 24.887876510620117,
      "learning_rate": 1e-06,
      "loss": 0.5927,
      "num_input_tokens_seen": 401795184,
      "step": 7171
    },
    {
      "epoch": 15.971046770601337,
      "loss": 0.7264291048049927,
      "loss_ce": 0.00011069556057918817,
      "loss_iou": 0.30859375,
      "loss_num": 0.021484375,
      "loss_xval": 0.7265625,
      "num_input_tokens_seen": 401795184,
      "step": 7171
    },
    {
      "epoch": 15.973273942093542,
      "grad_norm": 14.26672077178955,
      "learning_rate": 1e-06,
      "loss": 0.4417,
      "num_input_tokens_seen": 401852640,
      "step": 7172
    },
    {
      "epoch": 15.973273942093542,
      "loss": 0.4775257706642151,
      "loss_ce": 0.00010878611647058278,
      "loss_iou": 0.2138671875,
      "loss_num": 0.00994873046875,
      "loss_xval": 0.4765625,
      "num_input_tokens_seen": 401852640,
      "step": 7172
    },
    {
      "epoch": 15.975501113585747,
      "grad_norm": 20.04558753967285,
      "learning_rate": 1e-06,
      "loss": 0.3823,
      "num_input_tokens_seen": 401909140,
      "step": 7173
    },
    {
      "epoch": 15.975501113585747,
      "loss": 0.35957685112953186,
      "loss_ce": 7.978198118507862e-05,
      "loss_iou": 0.1650390625,
      "loss_num": 0.00579833984375,
      "loss_xval": 0.359375,
      "num_input_tokens_seen": 401909140,
      "step": 7173
    },
    {
      "epoch": 15.977728285077951,
      "grad_norm": 14.882387161254883,
      "learning_rate": 1e-06,
      "loss": 0.4257,
      "num_input_tokens_seen": 401968400,
      "step": 7174
    },
    {
      "epoch": 15.977728285077951,
      "loss": 0.48474133014678955,
      "loss_ce": 0.00012219653581269085,
      "loss_iou": 0.220703125,
      "loss_num": 0.0086669921875,
      "loss_xval": 0.484375,
      "num_input_tokens_seen": 401968400,
      "step": 7174
    },
    {
      "epoch": 15.979955456570156,
      "grad_norm": 12.847021102905273,
      "learning_rate": 1e-06,
      "loss": 0.3644,
      "num_input_tokens_seen": 402023480,
      "step": 7175
    },
    {
      "epoch": 15.979955456570156,
      "loss": 0.4103115499019623,
      "loss_ce": 9.426410542801023e-05,
      "loss_iou": 0.1728515625,
      "loss_num": 0.01287841796875,
      "loss_xval": 0.41015625,
      "num_input_tokens_seen": 402023480,
      "step": 7175
    },
    {
      "epoch": 15.982182628062361,
      "grad_norm": 13.715554237365723,
      "learning_rate": 1e-06,
      "loss": 0.3894,
      "num_input_tokens_seen": 402080180,
      "step": 7176
    },
    {
      "epoch": 15.982182628062361,
      "loss": 0.47568291425704956,
      "loss_ce": 9.699161455500871e-05,
      "loss_iou": 0.2080078125,
      "loss_num": 0.01202392578125,
      "loss_xval": 0.4765625,
      "num_input_tokens_seen": 402080180,
      "step": 7176
    },
    {
      "epoch": 15.984409799554566,
      "grad_norm": 16.425464630126953,
      "learning_rate": 1e-06,
      "loss": 0.3162,
      "num_input_tokens_seen": 402135848,
      "step": 7177
    },
    {
      "epoch": 15.984409799554566,
      "loss": 0.38852375745773315,
      "loss_ce": 9.600124030839652e-05,
      "loss_iou": 0.1767578125,
      "loss_num": 0.007171630859375,
      "loss_xval": 0.388671875,
      "num_input_tokens_seen": 402135848,
      "step": 7177
    },
    {
      "epoch": 15.98663697104677,
      "grad_norm": 15.072603225708008,
      "learning_rate": 1e-06,
      "loss": 0.4141,
      "num_input_tokens_seen": 402191316,
      "step": 7178
    },
    {
      "epoch": 15.98663697104677,
      "loss": 0.48317813873291016,
      "loss_ce": 0.00014590806677006185,
      "loss_iou": 0.2080078125,
      "loss_num": 0.01336669921875,
      "loss_xval": 0.482421875,
      "num_input_tokens_seen": 402191316,
      "step": 7178
    },
    {
      "epoch": 15.988864142538976,
      "grad_norm": 26.326980590820312,
      "learning_rate": 1e-06,
      "loss": 0.4093,
      "num_input_tokens_seen": 402246396,
      "step": 7179
    },
    {
      "epoch": 15.988864142538976,
      "loss": 0.3864772319793701,
      "loss_ce": 0.0001246822066605091,
      "loss_iou": 0.1708984375,
      "loss_num": 0.00872802734375,
      "loss_xval": 0.38671875,
      "num_input_tokens_seen": 402246396,
      "step": 7179
    },
    {
      "epoch": 15.99109131403118,
      "grad_norm": 14.615377426147461,
      "learning_rate": 1e-06,
      "loss": 0.3313,
      "num_input_tokens_seen": 402303500,
      "step": 7180
    },
    {
      "epoch": 15.99109131403118,
      "loss": 0.2933087646961212,
      "loss_ce": 0.00015690606960561126,
      "loss_iou": 0.12060546875,
      "loss_num": 0.01025390625,
      "loss_xval": 0.29296875,
      "num_input_tokens_seen": 402303500,
      "step": 7180
    },
    {
      "epoch": 15.993318485523385,
      "grad_norm": 20.349151611328125,
      "learning_rate": 1e-06,
      "loss": 0.3356,
      "num_input_tokens_seen": 402358536,
      "step": 7181
    },
    {
      "epoch": 15.993318485523385,
      "loss": 0.3299248516559601,
      "loss_ce": 9.087211219593883e-05,
      "loss_iou": 0.1435546875,
      "loss_num": 0.00860595703125,
      "loss_xval": 0.330078125,
      "num_input_tokens_seen": 402358536,
      "step": 7181
    },
    {
      "epoch": 15.99554565701559,
      "grad_norm": 14.283751487731934,
      "learning_rate": 1e-06,
      "loss": 0.3803,
      "num_input_tokens_seen": 402414524,
      "step": 7182
    },
    {
      "epoch": 15.99554565701559,
      "loss": 0.2790136933326721,
      "loss_ce": 8.300953777506948e-05,
      "loss_iou": 0.1025390625,
      "loss_num": 0.01470947265625,
      "loss_xval": 0.279296875,
      "num_input_tokens_seen": 402414524,
      "step": 7182
    },
    {
      "epoch": 15.997772828507795,
      "grad_norm": 53.64034652709961,
      "learning_rate": 1e-06,
      "loss": 0.4319,
      "num_input_tokens_seen": 402473320,
      "step": 7183
    },
    {
      "epoch": 15.997772828507795,
      "loss": 0.478466272354126,
      "loss_ce": 0.00019476463785395026,
      "loss_iou": 0.2236328125,
      "loss_num": 0.006195068359375,
      "loss_xval": 0.478515625,
      "num_input_tokens_seen": 402473320,
      "step": 7183
    },
    {
      "epoch": 16.0,
      "grad_norm": 29.848583221435547,
      "learning_rate": 1e-06,
      "loss": 0.4615,
      "num_input_tokens_seen": 402529284,
      "step": 7184
    },
    {
      "epoch": 16.0,
      "loss": 0.5135668516159058,
      "loss_ce": 0.00013914526789449155,
      "loss_iou": 0.220703125,
      "loss_num": 0.01422119140625,
      "loss_xval": 0.51171875,
      "num_input_tokens_seen": 402529284,
      "step": 7184
    },
    {
      "epoch": 16.002227171492205,
      "grad_norm": 27.619600296020508,
      "learning_rate": 1e-06,
      "loss": 0.4962,
      "num_input_tokens_seen": 402585352,
      "step": 7185
    },
    {
      "epoch": 16.002227171492205,
      "loss": 0.6742956042289734,
      "loss_ce": 0.00010129276779480278,
      "loss_iou": 0.279296875,
      "loss_num": 0.0233154296875,
      "loss_xval": 0.67578125,
      "num_input_tokens_seen": 402585352,
      "step": 7185
    },
    {
      "epoch": 16.00445434298441,
      "grad_norm": 20.31178855895996,
      "learning_rate": 1e-06,
      "loss": 0.5351,
      "num_input_tokens_seen": 402638996,
      "step": 7186
    },
    {
      "epoch": 16.00445434298441,
      "loss": 0.5318279266357422,
      "loss_ce": 8.96200945135206e-05,
      "loss_iou": 0.23828125,
      "loss_num": 0.0113525390625,
      "loss_xval": 0.53125,
      "num_input_tokens_seen": 402638996,
      "step": 7186
    },
    {
      "epoch": 16.006681514476615,
      "grad_norm": 17.070863723754883,
      "learning_rate": 1e-06,
      "loss": 0.448,
      "num_input_tokens_seen": 402695996,
      "step": 7187
    },
    {
      "epoch": 16.006681514476615,
      "loss": 0.4115889072418213,
      "loss_ce": 8.989499474409968e-05,
      "loss_iou": 0.1708984375,
      "loss_num": 0.0140380859375,
      "loss_xval": 0.412109375,
      "num_input_tokens_seen": 402695996,
      "step": 7187
    },
    {
      "epoch": 16.00890868596882,
      "grad_norm": 12.187263488769531,
      "learning_rate": 1e-06,
      "loss": 0.2493,
      "num_input_tokens_seen": 402753988,
      "step": 7188
    },
    {
      "epoch": 16.00890868596882,
      "loss": 0.24472574889659882,
      "loss_ce": 9.685206168796867e-05,
      "loss_iou": 0.10498046875,
      "loss_num": 0.006927490234375,
      "loss_xval": 0.244140625,
      "num_input_tokens_seen": 402753988,
      "step": 7188
    },
    {
      "epoch": 16.011135857461024,
      "grad_norm": 13.688761711120605,
      "learning_rate": 1e-06,
      "loss": 0.3247,
      "num_input_tokens_seen": 402810468,
      "step": 7189
    },
    {
      "epoch": 16.011135857461024,
      "loss": 0.42870235443115234,
      "loss_ce": 0.00011348059342708439,
      "loss_iou": 0.1826171875,
      "loss_num": 0.01263427734375,
      "loss_xval": 0.427734375,
      "num_input_tokens_seen": 402810468,
      "step": 7189
    },
    {
      "epoch": 16.01336302895323,
      "grad_norm": 17.964651107788086,
      "learning_rate": 1e-06,
      "loss": 0.5188,
      "num_input_tokens_seen": 402862988,
      "step": 7190
    },
    {
      "epoch": 16.01336302895323,
      "loss": 0.5445123314857483,
      "loss_ce": 7.876359450165182e-05,
      "loss_iou": 0.2001953125,
      "loss_num": 0.029052734375,
      "loss_xval": 0.54296875,
      "num_input_tokens_seen": 402862988,
      "step": 7190
    },
    {
      "epoch": 16.015590200445434,
      "grad_norm": 22.054534912109375,
      "learning_rate": 1e-06,
      "loss": 0.3839,
      "num_input_tokens_seen": 402919908,
      "step": 7191
    },
    {
      "epoch": 16.015590200445434,
      "loss": 0.3673054575920105,
      "loss_ce": 0.00011793937301263213,
      "loss_iou": 0.171875,
      "loss_num": 0.004669189453125,
      "loss_xval": 0.3671875,
      "num_input_tokens_seen": 402919908,
      "step": 7191
    },
    {
      "epoch": 16.01781737193764,
      "grad_norm": 16.937606811523438,
      "learning_rate": 1e-06,
      "loss": 0.4729,
      "num_input_tokens_seen": 402976004,
      "step": 7192
    },
    {
      "epoch": 16.01781737193764,
      "loss": 0.5479812026023865,
      "loss_ce": 0.0001907070109155029,
      "loss_iou": 0.2421875,
      "loss_num": 0.0126953125,
      "loss_xval": 0.546875,
      "num_input_tokens_seen": 402976004,
      "step": 7192
    },
    {
      "epoch": 16.020044543429844,
      "grad_norm": 21.24249839782715,
      "learning_rate": 1e-06,
      "loss": 0.4644,
      "num_input_tokens_seen": 403032388,
      "step": 7193
    },
    {
      "epoch": 16.020044543429844,
      "loss": 0.3727763891220093,
      "loss_ce": 0.00015674906899221241,
      "loss_iou": 0.1708984375,
      "loss_num": 0.006072998046875,
      "loss_xval": 0.373046875,
      "num_input_tokens_seen": 403032388,
      "step": 7193
    },
    {
      "epoch": 16.02227171492205,
      "grad_norm": 15.852849960327148,
      "learning_rate": 1e-06,
      "loss": 0.4715,
      "num_input_tokens_seen": 403091080,
      "step": 7194
    },
    {
      "epoch": 16.02227171492205,
      "loss": 0.5445390939712524,
      "loss_ce": 0.00010548095451667905,
      "loss_iou": 0.232421875,
      "loss_num": 0.01611328125,
      "loss_xval": 0.54296875,
      "num_input_tokens_seen": 403091080,
      "step": 7194
    },
    {
      "epoch": 16.024498886414253,
      "grad_norm": 22.950151443481445,
      "learning_rate": 1e-06,
      "loss": 0.648,
      "num_input_tokens_seen": 403148524,
      "step": 7195
    },
    {
      "epoch": 16.024498886414253,
      "loss": 0.6838928461074829,
      "loss_ce": 0.00017702819604892284,
      "loss_iou": 0.2734375,
      "loss_num": 0.02783203125,
      "loss_xval": 0.68359375,
      "num_input_tokens_seen": 403148524,
      "step": 7195
    },
    {
      "epoch": 16.02672605790646,
      "grad_norm": 21.878934860229492,
      "learning_rate": 1e-06,
      "loss": 0.5133,
      "num_input_tokens_seen": 403205004,
      "step": 7196
    },
    {
      "epoch": 16.02672605790646,
      "loss": 0.5003616809844971,
      "loss_ce": 0.0003617034526541829,
      "loss_iou": 0.2255859375,
      "loss_num": 0.0096435546875,
      "loss_xval": 0.5,
      "num_input_tokens_seen": 403205004,
      "step": 7196
    },
    {
      "epoch": 16.028953229398663,
      "grad_norm": 24.652055740356445,
      "learning_rate": 1e-06,
      "loss": 0.5203,
      "num_input_tokens_seen": 403256684,
      "step": 7197
    },
    {
      "epoch": 16.028953229398663,
      "loss": 0.38059282302856445,
      "loss_ce": 9.965993376681581e-05,
      "loss_iou": 0.15625,
      "loss_num": 0.01361083984375,
      "loss_xval": 0.380859375,
      "num_input_tokens_seen": 403256684,
      "step": 7197
    },
    {
      "epoch": 16.031180400890868,
      "grad_norm": 14.730175018310547,
      "learning_rate": 1e-06,
      "loss": 0.5587,
      "num_input_tokens_seen": 403313248,
      "step": 7198
    },
    {
      "epoch": 16.031180400890868,
      "loss": 0.6153490543365479,
      "loss_ce": 0.00011463207920314744,
      "loss_iou": 0.259765625,
      "loss_num": 0.019287109375,
      "loss_xval": 0.6171875,
      "num_input_tokens_seen": 403313248,
      "step": 7198
    },
    {
      "epoch": 16.033407572383073,
      "grad_norm": 18.867799758911133,
      "learning_rate": 1e-06,
      "loss": 0.4701,
      "num_input_tokens_seen": 403371832,
      "step": 7199
    },
    {
      "epoch": 16.033407572383073,
      "loss": 0.6135093569755554,
      "loss_ce": 0.0002280865446664393,
      "loss_iou": 0.26953125,
      "loss_num": 0.01519775390625,
      "loss_xval": 0.61328125,
      "num_input_tokens_seen": 403371832,
      "step": 7199
    },
    {
      "epoch": 16.035634743875278,
      "grad_norm": 15.737879753112793,
      "learning_rate": 1e-06,
      "loss": 0.6626,
      "num_input_tokens_seen": 403428312,
      "step": 7200
    },
    {
      "epoch": 16.035634743875278,
      "loss": 0.7531107068061829,
      "loss_ce": 0.00018101301975548267,
      "loss_iou": 0.337890625,
      "loss_num": 0.01519775390625,
      "loss_xval": 0.75390625,
      "num_input_tokens_seen": 403428312,
      "step": 7200
    },
    {
      "epoch": 16.037861915367483,
      "grad_norm": 16.80666160583496,
      "learning_rate": 1e-06,
      "loss": 0.4065,
      "num_input_tokens_seen": 403486724,
      "step": 7201
    },
    {
      "epoch": 16.037861915367483,
      "loss": 0.47910380363464355,
      "loss_ce": 9.990914259105921e-05,
      "loss_iou": 0.22265625,
      "loss_num": 0.006500244140625,
      "loss_xval": 0.478515625,
      "num_input_tokens_seen": 403486724,
      "step": 7201
    },
    {
      "epoch": 16.040089086859687,
      "grad_norm": 11.938071250915527,
      "learning_rate": 1e-06,
      "loss": 0.4147,
      "num_input_tokens_seen": 403541852,
      "step": 7202
    },
    {
      "epoch": 16.040089086859687,
      "loss": 0.4437464475631714,
      "loss_ce": 8.189181244233623e-05,
      "loss_iou": 0.1884765625,
      "loss_num": 0.0133056640625,
      "loss_xval": 0.443359375,
      "num_input_tokens_seen": 403541852,
      "step": 7202
    },
    {
      "epoch": 16.042316258351892,
      "grad_norm": 33.16254806518555,
      "learning_rate": 1e-06,
      "loss": 0.4485,
      "num_input_tokens_seen": 403598276,
      "step": 7203
    },
    {
      "epoch": 16.042316258351892,
      "loss": 0.3794906437397003,
      "loss_ce": 9.610810957383364e-05,
      "loss_iou": 0.16796875,
      "loss_num": 0.0087890625,
      "loss_xval": 0.37890625,
      "num_input_tokens_seen": 403598276,
      "step": 7203
    },
    {
      "epoch": 16.044543429844097,
      "grad_norm": 28.809633255004883,
      "learning_rate": 1e-06,
      "loss": 0.4134,
      "num_input_tokens_seen": 403657000,
      "step": 7204
    },
    {
      "epoch": 16.044543429844097,
      "loss": 0.3428630530834198,
      "loss_ce": 8.960704144556075e-05,
      "loss_iou": 0.1552734375,
      "loss_num": 0.006378173828125,
      "loss_xval": 0.34375,
      "num_input_tokens_seen": 403657000,
      "step": 7204
    },
    {
      "epoch": 16.046770601336302,
      "grad_norm": 25.093759536743164,
      "learning_rate": 1e-06,
      "loss": 0.6919,
      "num_input_tokens_seen": 403711572,
      "step": 7205
    },
    {
      "epoch": 16.046770601336302,
      "loss": 0.38008594512939453,
      "loss_ce": 8.106790482997894e-05,
      "loss_iou": 0.1611328125,
      "loss_num": 0.01165771484375,
      "loss_xval": 0.380859375,
      "num_input_tokens_seen": 403711572,
      "step": 7205
    },
    {
      "epoch": 16.048997772828507,
      "grad_norm": 19.109130859375,
      "learning_rate": 1e-06,
      "loss": 0.2554,
      "num_input_tokens_seen": 403767064,
      "step": 7206
    },
    {
      "epoch": 16.048997772828507,
      "loss": 0.22988614439964294,
      "loss_ce": 8.877603249857202e-05,
      "loss_iou": 0.10009765625,
      "loss_num": 0.005889892578125,
      "loss_xval": 0.2294921875,
      "num_input_tokens_seen": 403767064,
      "step": 7206
    },
    {
      "epoch": 16.051224944320712,
      "grad_norm": 14.356725692749023,
      "learning_rate": 1e-06,
      "loss": 0.4188,
      "num_input_tokens_seen": 403823508,
      "step": 7207
    },
    {
      "epoch": 16.051224944320712,
      "loss": 0.4483864903450012,
      "loss_ce": 0.0001443219225620851,
      "loss_iou": 0.1923828125,
      "loss_num": 0.0126953125,
      "loss_xval": 0.44921875,
      "num_input_tokens_seen": 403823508,
      "step": 7207
    },
    {
      "epoch": 16.053452115812917,
      "grad_norm": 18.274364471435547,
      "learning_rate": 1e-06,
      "loss": 0.4482,
      "num_input_tokens_seen": 403878140,
      "step": 7208
    },
    {
      "epoch": 16.053452115812917,
      "loss": 0.4601691961288452,
      "loss_ce": 8.618818537797779e-05,
      "loss_iou": 0.1982421875,
      "loss_num": 0.01263427734375,
      "loss_xval": 0.4609375,
      "num_input_tokens_seen": 403878140,
      "step": 7208
    },
    {
      "epoch": 16.05567928730512,
      "grad_norm": 19.279592514038086,
      "learning_rate": 1e-06,
      "loss": 0.5412,
      "num_input_tokens_seen": 403932144,
      "step": 7209
    },
    {
      "epoch": 16.05567928730512,
      "loss": 0.6549313068389893,
      "loss_ce": 0.00014613490202464163,
      "loss_iou": 0.267578125,
      "loss_num": 0.0242919921875,
      "loss_xval": 0.65625,
      "num_input_tokens_seen": 403932144,
      "step": 7209
    },
    {
      "epoch": 16.057906458797326,
      "grad_norm": 15.034238815307617,
      "learning_rate": 1e-06,
      "loss": 0.471,
      "num_input_tokens_seen": 403987788,
      "step": 7210
    },
    {
      "epoch": 16.057906458797326,
      "loss": 0.43674468994140625,
      "loss_ce": 9.918860450852662e-05,
      "loss_iou": 0.201171875,
      "loss_num": 0.006988525390625,
      "loss_xval": 0.4375,
      "num_input_tokens_seen": 403987788,
      "step": 7210
    },
    {
      "epoch": 16.06013363028953,
      "grad_norm": 13.465814590454102,
      "learning_rate": 1e-06,
      "loss": 0.3559,
      "num_input_tokens_seen": 404044384,
      "step": 7211
    },
    {
      "epoch": 16.06013363028953,
      "loss": 0.3224736154079437,
      "loss_ce": 8.591696678195149e-05,
      "loss_iou": 0.1416015625,
      "loss_num": 0.007781982421875,
      "loss_xval": 0.322265625,
      "num_input_tokens_seen": 404044384,
      "step": 7211
    },
    {
      "epoch": 16.062360801781736,
      "grad_norm": 18.50025749206543,
      "learning_rate": 1e-06,
      "loss": 0.369,
      "num_input_tokens_seen": 404099392,
      "step": 7212
    },
    {
      "epoch": 16.062360801781736,
      "loss": 0.35518187284469604,
      "loss_ce": 7.935409666970372e-05,
      "loss_iou": 0.1494140625,
      "loss_num": 0.01123046875,
      "loss_xval": 0.35546875,
      "num_input_tokens_seen": 404099392,
      "step": 7212
    },
    {
      "epoch": 16.06458797327394,
      "grad_norm": 12.4852294921875,
      "learning_rate": 1e-06,
      "loss": 0.4552,
      "num_input_tokens_seen": 404153904,
      "step": 7213
    },
    {
      "epoch": 16.06458797327394,
      "loss": 0.5071730017662048,
      "loss_ce": 9.29402667679824e-05,
      "loss_iou": 0.2138671875,
      "loss_num": 0.0157470703125,
      "loss_xval": 0.5078125,
      "num_input_tokens_seen": 404153904,
      "step": 7213
    },
    {
      "epoch": 16.066815144766146,
      "grad_norm": 22.292587280273438,
      "learning_rate": 1e-06,
      "loss": 0.5293,
      "num_input_tokens_seen": 404207456,
      "step": 7214
    },
    {
      "epoch": 16.066815144766146,
      "loss": 0.3804709315299988,
      "loss_ce": 9.985938959289342e-05,
      "loss_iou": 0.1708984375,
      "loss_num": 0.007537841796875,
      "loss_xval": 0.380859375,
      "num_input_tokens_seen": 404207456,
      "step": 7214
    },
    {
      "epoch": 16.06904231625835,
      "grad_norm": 21.88534927368164,
      "learning_rate": 1e-06,
      "loss": 0.4697,
      "num_input_tokens_seen": 404262880,
      "step": 7215
    },
    {
      "epoch": 16.06904231625835,
      "loss": 0.6787058115005493,
      "loss_ce": 0.0001169023453257978,
      "loss_iou": 0.314453125,
      "loss_num": 0.009765625,
      "loss_xval": 0.6796875,
      "num_input_tokens_seen": 404262880,
      "step": 7215
    },
    {
      "epoch": 16.071269487750556,
      "grad_norm": 29.59630584716797,
      "learning_rate": 1e-06,
      "loss": 0.5467,
      "num_input_tokens_seen": 404316780,
      "step": 7216
    },
    {
      "epoch": 16.071269487750556,
      "loss": 0.505081057548523,
      "loss_ce": 7.61750852689147e-05,
      "loss_iou": 0.220703125,
      "loss_num": 0.012451171875,
      "loss_xval": 0.50390625,
      "num_input_tokens_seen": 404316780,
      "step": 7216
    },
    {
      "epoch": 16.07349665924276,
      "grad_norm": 14.112740516662598,
      "learning_rate": 1e-06,
      "loss": 0.4883,
      "num_input_tokens_seen": 404374304,
      "step": 7217
    },
    {
      "epoch": 16.07349665924276,
      "loss": 0.29110169410705566,
      "loss_ce": 8.60571744851768e-05,
      "loss_iou": 0.1328125,
      "loss_num": 0.00518798828125,
      "loss_xval": 0.291015625,
      "num_input_tokens_seen": 404374304,
      "step": 7217
    },
    {
      "epoch": 16.075723830734965,
      "grad_norm": 17.42337417602539,
      "learning_rate": 1e-06,
      "loss": 0.3548,
      "num_input_tokens_seen": 404430784,
      "step": 7218
    },
    {
      "epoch": 16.075723830734965,
      "loss": 0.3221060037612915,
      "loss_ce": 8.452765177935362e-05,
      "loss_iou": 0.142578125,
      "loss_num": 0.007537841796875,
      "loss_xval": 0.322265625,
      "num_input_tokens_seen": 404430784,
      "step": 7218
    },
    {
      "epoch": 16.07795100222717,
      "grad_norm": 20.392663955688477,
      "learning_rate": 1e-06,
      "loss": 0.5018,
      "num_input_tokens_seen": 404486928,
      "step": 7219
    },
    {
      "epoch": 16.07795100222717,
      "loss": 0.45973044633865356,
      "loss_ce": 7.468648254871368e-05,
      "loss_iou": 0.205078125,
      "loss_num": 0.0098876953125,
      "loss_xval": 0.458984375,
      "num_input_tokens_seen": 404486928,
      "step": 7219
    },
    {
      "epoch": 16.080178173719375,
      "grad_norm": 21.25473403930664,
      "learning_rate": 1e-06,
      "loss": 0.6253,
      "num_input_tokens_seen": 404542168,
      "step": 7220
    },
    {
      "epoch": 16.080178173719375,
      "loss": 0.576141357421875,
      "loss_ce": 9.15601704036817e-05,
      "loss_iou": 0.2412109375,
      "loss_num": 0.0186767578125,
      "loss_xval": 0.57421875,
      "num_input_tokens_seen": 404542168,
      "step": 7220
    },
    {
      "epoch": 16.08240534521158,
      "grad_norm": 15.591520309448242,
      "learning_rate": 1e-06,
      "loss": 0.4269,
      "num_input_tokens_seen": 404597716,
      "step": 7221
    },
    {
      "epoch": 16.08240534521158,
      "loss": 0.38363730907440186,
      "loss_ce": 9.239626524504274e-05,
      "loss_iou": 0.1650390625,
      "loss_num": 0.01068115234375,
      "loss_xval": 0.3828125,
      "num_input_tokens_seen": 404597716,
      "step": 7221
    },
    {
      "epoch": 16.084632516703785,
      "grad_norm": 34.75109100341797,
      "learning_rate": 1e-06,
      "loss": 0.4051,
      "num_input_tokens_seen": 404652108,
      "step": 7222
    },
    {
      "epoch": 16.084632516703785,
      "loss": 0.3343104124069214,
      "loss_ce": 8.188547508325428e-05,
      "loss_iou": 0.140625,
      "loss_num": 0.0106201171875,
      "loss_xval": 0.333984375,
      "num_input_tokens_seen": 404652108,
      "step": 7222
    },
    {
      "epoch": 16.08685968819599,
      "grad_norm": 17.89012908935547,
      "learning_rate": 1e-06,
      "loss": 0.5697,
      "num_input_tokens_seen": 404708964,
      "step": 7223
    },
    {
      "epoch": 16.08685968819599,
      "loss": 0.3971855044364929,
      "loss_ce": 9.079407755052671e-05,
      "loss_iou": 0.18359375,
      "loss_num": 0.005859375,
      "loss_xval": 0.396484375,
      "num_input_tokens_seen": 404708964,
      "step": 7223
    },
    {
      "epoch": 16.089086859688194,
      "grad_norm": 18.168272018432617,
      "learning_rate": 1e-06,
      "loss": 0.4312,
      "num_input_tokens_seen": 404760744,
      "step": 7224
    },
    {
      "epoch": 16.089086859688194,
      "loss": 0.4721256494522095,
      "loss_ce": 7.975117478054017e-05,
      "loss_iou": 0.19921875,
      "loss_num": 0.0147705078125,
      "loss_xval": 0.47265625,
      "num_input_tokens_seen": 404760744,
      "step": 7224
    },
    {
      "epoch": 16.0913140311804,
      "grad_norm": 22.09956169128418,
      "learning_rate": 1e-06,
      "loss": 0.2769,
      "num_input_tokens_seen": 404818936,
      "step": 7225
    },
    {
      "epoch": 16.0913140311804,
      "loss": 0.19356948137283325,
      "loss_ce": 8.803060336504132e-05,
      "loss_iou": 0.087890625,
      "loss_num": 0.0035400390625,
      "loss_xval": 0.193359375,
      "num_input_tokens_seen": 404818936,
      "step": 7225
    },
    {
      "epoch": 16.093541202672604,
      "grad_norm": 18.174219131469727,
      "learning_rate": 1e-06,
      "loss": 0.3931,
      "num_input_tokens_seen": 404874408,
      "step": 7226
    },
    {
      "epoch": 16.093541202672604,
      "loss": 0.40315836668014526,
      "loss_ce": 8.220285963034257e-05,
      "loss_iou": 0.1708984375,
      "loss_num": 0.01214599609375,
      "loss_xval": 0.40234375,
      "num_input_tokens_seen": 404874408,
      "step": 7226
    },
    {
      "epoch": 16.09576837416481,
      "grad_norm": 21.083255767822266,
      "learning_rate": 1e-06,
      "loss": 0.2451,
      "num_input_tokens_seen": 404932184,
      "step": 7227
    },
    {
      "epoch": 16.09576837416481,
      "loss": 0.2461136281490326,
      "loss_ce": 8.089626498986036e-05,
      "loss_iou": 0.10693359375,
      "loss_num": 0.0064697265625,
      "loss_xval": 0.24609375,
      "num_input_tokens_seen": 404932184,
      "step": 7227
    },
    {
      "epoch": 16.097995545657014,
      "grad_norm": 15.489051818847656,
      "learning_rate": 1e-06,
      "loss": 0.4338,
      "num_input_tokens_seen": 404989408,
      "step": 7228
    },
    {
      "epoch": 16.097995545657014,
      "loss": 0.46629413962364197,
      "loss_ce": 0.00010758035205071792,
      "loss_iou": 0.216796875,
      "loss_num": 0.00634765625,
      "loss_xval": 0.466796875,
      "num_input_tokens_seen": 404989408,
      "step": 7228
    },
    {
      "epoch": 16.100222717149222,
      "grad_norm": 28.974552154541016,
      "learning_rate": 1e-06,
      "loss": 0.5686,
      "num_input_tokens_seen": 405044892,
      "step": 7229
    },
    {
      "epoch": 16.100222717149222,
      "loss": 0.4830133318901062,
      "loss_ce": 0.00010318079148419201,
      "loss_iou": 0.220703125,
      "loss_num": 0.00823974609375,
      "loss_xval": 0.482421875,
      "num_input_tokens_seen": 405044892,
      "step": 7229
    },
    {
      "epoch": 16.102449888641427,
      "grad_norm": 27.26142692565918,
      "learning_rate": 1e-06,
      "loss": 0.5382,
      "num_input_tokens_seen": 405102960,
      "step": 7230
    },
    {
      "epoch": 16.102449888641427,
      "loss": 0.6526839137077332,
      "loss_ce": 9.603158105164766e-05,
      "loss_iou": 0.28125,
      "loss_num": 0.0179443359375,
      "loss_xval": 0.65234375,
      "num_input_tokens_seen": 405102960,
      "step": 7230
    },
    {
      "epoch": 16.104677060133632,
      "grad_norm": 33.493282318115234,
      "learning_rate": 1e-06,
      "loss": 0.4217,
      "num_input_tokens_seen": 405159344,
      "step": 7231
    },
    {
      "epoch": 16.104677060133632,
      "loss": 0.47552353143692017,
      "loss_ce": 0.00012070016236975789,
      "loss_iou": 0.1884765625,
      "loss_num": 0.0196533203125,
      "loss_xval": 0.474609375,
      "num_input_tokens_seen": 405159344,
      "step": 7231
    },
    {
      "epoch": 16.106904231625837,
      "grad_norm": 16.54621124267578,
      "learning_rate": 1e-06,
      "loss": 0.3254,
      "num_input_tokens_seen": 405215920,
      "step": 7232
    },
    {
      "epoch": 16.106904231625837,
      "loss": 0.3362829387187958,
      "loss_ce": 0.00010127984569408,
      "loss_iou": 0.1494140625,
      "loss_num": 0.00750732421875,
      "loss_xval": 0.3359375,
      "num_input_tokens_seen": 405215920,
      "step": 7232
    },
    {
      "epoch": 16.10913140311804,
      "grad_norm": 15.172420501708984,
      "learning_rate": 1e-06,
      "loss": 0.271,
      "num_input_tokens_seen": 405274772,
      "step": 7233
    },
    {
      "epoch": 16.10913140311804,
      "loss": 0.1989382654428482,
      "loss_ce": 8.573340164730325e-05,
      "loss_iou": 0.08642578125,
      "loss_num": 0.005126953125,
      "loss_xval": 0.19921875,
      "num_input_tokens_seen": 405274772,
      "step": 7233
    },
    {
      "epoch": 16.111358574610247,
      "grad_norm": 25.89063835144043,
      "learning_rate": 1e-06,
      "loss": 0.428,
      "num_input_tokens_seen": 405328804,
      "step": 7234
    },
    {
      "epoch": 16.111358574610247,
      "loss": 0.48993584513664246,
      "loss_ce": 0.00012872781371697783,
      "loss_iou": 0.1982421875,
      "loss_num": 0.0185546875,
      "loss_xval": 0.490234375,
      "num_input_tokens_seen": 405328804,
      "step": 7234
    },
    {
      "epoch": 16.11358574610245,
      "grad_norm": 17.315288543701172,
      "learning_rate": 1e-06,
      "loss": 0.3599,
      "num_input_tokens_seen": 405384920,
      "step": 7235
    },
    {
      "epoch": 16.11358574610245,
      "loss": 0.40499159693717957,
      "loss_ce": 8.437626820523292e-05,
      "loss_iou": 0.181640625,
      "loss_num": 0.00836181640625,
      "loss_xval": 0.404296875,
      "num_input_tokens_seen": 405384920,
      "step": 7235
    },
    {
      "epoch": 16.115812917594656,
      "grad_norm": 11.827547073364258,
      "learning_rate": 1e-06,
      "loss": 0.4208,
      "num_input_tokens_seen": 405440816,
      "step": 7236
    },
    {
      "epoch": 16.115812917594656,
      "loss": 0.43206048011779785,
      "loss_ce": 0.00011468880984466523,
      "loss_iou": 0.1962890625,
      "loss_num": 0.007720947265625,
      "loss_xval": 0.431640625,
      "num_input_tokens_seen": 405440816,
      "step": 7236
    },
    {
      "epoch": 16.11804008908686,
      "grad_norm": 16.36489486694336,
      "learning_rate": 1e-06,
      "loss": 0.4765,
      "num_input_tokens_seen": 405497324,
      "step": 7237
    },
    {
      "epoch": 16.11804008908686,
      "loss": 0.4784888029098511,
      "loss_ce": 9.523934568278491e-05,
      "loss_iou": 0.1787109375,
      "loss_num": 0.0242919921875,
      "loss_xval": 0.478515625,
      "num_input_tokens_seen": 405497324,
      "step": 7237
    },
    {
      "epoch": 16.120267260579066,
      "grad_norm": 19.48402214050293,
      "learning_rate": 1e-06,
      "loss": 0.521,
      "num_input_tokens_seen": 405551216,
      "step": 7238
    },
    {
      "epoch": 16.120267260579066,
      "loss": 0.6264359951019287,
      "loss_ce": 9.326158760813996e-05,
      "loss_iou": 0.2431640625,
      "loss_num": 0.028076171875,
      "loss_xval": 0.625,
      "num_input_tokens_seen": 405551216,
      "step": 7238
    },
    {
      "epoch": 16.12249443207127,
      "grad_norm": 20.337753295898438,
      "learning_rate": 1e-06,
      "loss": 0.3115,
      "num_input_tokens_seen": 405608940,
      "step": 7239
    },
    {
      "epoch": 16.12249443207127,
      "loss": 0.27264925837516785,
      "loss_ce": 0.000310403760522604,
      "loss_iou": 0.11376953125,
      "loss_num": 0.009033203125,
      "loss_xval": 0.271484375,
      "num_input_tokens_seen": 405608940,
      "step": 7239
    },
    {
      "epoch": 16.124721603563476,
      "grad_norm": 66.91169738769531,
      "learning_rate": 1e-06,
      "loss": 0.4479,
      "num_input_tokens_seen": 405666224,
      "step": 7240
    },
    {
      "epoch": 16.124721603563476,
      "loss": 0.5517283082008362,
      "loss_ce": 9.25497297430411e-05,
      "loss_iou": 0.26171875,
      "loss_num": 0.00592041015625,
      "loss_xval": 0.55078125,
      "num_input_tokens_seen": 405666224,
      "step": 7240
    },
    {
      "epoch": 16.12694877505568,
      "grad_norm": 14.807357788085938,
      "learning_rate": 1e-06,
      "loss": 0.4308,
      "num_input_tokens_seen": 405722588,
      "step": 7241
    },
    {
      "epoch": 16.12694877505568,
      "loss": 0.47714054584503174,
      "loss_ce": 8.976385288406163e-05,
      "loss_iou": 0.220703125,
      "loss_num": 0.00738525390625,
      "loss_xval": 0.4765625,
      "num_input_tokens_seen": 405722588,
      "step": 7241
    },
    {
      "epoch": 16.129175946547885,
      "grad_norm": 40.85698318481445,
      "learning_rate": 1e-06,
      "loss": 0.3856,
      "num_input_tokens_seen": 405778132,
      "step": 7242
    },
    {
      "epoch": 16.129175946547885,
      "loss": 0.40902432799339294,
      "loss_ce": 8.878795779310167e-05,
      "loss_iou": 0.1845703125,
      "loss_num": 0.008056640625,
      "loss_xval": 0.408203125,
      "num_input_tokens_seen": 405778132,
      "step": 7242
    },
    {
      "epoch": 16.13140311804009,
      "grad_norm": 19.070831298828125,
      "learning_rate": 1e-06,
      "loss": 0.4482,
      "num_input_tokens_seen": 405834844,
      "step": 7243
    },
    {
      "epoch": 16.13140311804009,
      "loss": 0.5067603588104248,
      "loss_ce": 0.00010757060954347253,
      "loss_iou": 0.216796875,
      "loss_num": 0.01458740234375,
      "loss_xval": 0.5078125,
      "num_input_tokens_seen": 405834844,
      "step": 7243
    },
    {
      "epoch": 16.133630289532295,
      "grad_norm": 22.71406364440918,
      "learning_rate": 1e-06,
      "loss": 0.5471,
      "num_input_tokens_seen": 405889304,
      "step": 7244
    },
    {
      "epoch": 16.133630289532295,
      "loss": 0.4746703505516052,
      "loss_ce": 9.150059486273676e-05,
      "loss_iou": 0.2138671875,
      "loss_num": 0.009521484375,
      "loss_xval": 0.474609375,
      "num_input_tokens_seen": 405889304,
      "step": 7244
    },
    {
      "epoch": 16.1358574610245,
      "grad_norm": 26.008378982543945,
      "learning_rate": 1e-06,
      "loss": 0.4208,
      "num_input_tokens_seen": 405948244,
      "step": 7245
    },
    {
      "epoch": 16.1358574610245,
      "loss": 0.44539159536361694,
      "loss_ce": 7.910738349892199e-05,
      "loss_iou": 0.2060546875,
      "loss_num": 0.006561279296875,
      "loss_xval": 0.4453125,
      "num_input_tokens_seen": 405948244,
      "step": 7245
    },
    {
      "epoch": 16.138084632516705,
      "grad_norm": 17.593692779541016,
      "learning_rate": 1e-06,
      "loss": 0.6091,
      "num_input_tokens_seen": 406004756,
      "step": 7246
    },
    {
      "epoch": 16.138084632516705,
      "loss": 0.672180712223053,
      "loss_ce": 0.00018365512369200587,
      "loss_iou": 0.314453125,
      "loss_num": 0.009033203125,
      "loss_xval": 0.671875,
      "num_input_tokens_seen": 406004756,
      "step": 7246
    },
    {
      "epoch": 16.14031180400891,
      "grad_norm": 31.90231704711914,
      "learning_rate": 1e-06,
      "loss": 0.6659,
      "num_input_tokens_seen": 406061152,
      "step": 7247
    },
    {
      "epoch": 16.14031180400891,
      "loss": 0.34200403094291687,
      "loss_ce": 8.508679457008839e-05,
      "loss_iou": 0.150390625,
      "loss_num": 0.00799560546875,
      "loss_xval": 0.341796875,
      "num_input_tokens_seen": 406061152,
      "step": 7247
    },
    {
      "epoch": 16.142538975501115,
      "grad_norm": 13.510699272155762,
      "learning_rate": 1e-06,
      "loss": 0.2774,
      "num_input_tokens_seen": 406116672,
      "step": 7248
    },
    {
      "epoch": 16.142538975501115,
      "loss": 0.29828941822052,
      "loss_ce": 7.161758549045771e-05,
      "loss_iou": 0.125,
      "loss_num": 0.00970458984375,
      "loss_xval": 0.298828125,
      "num_input_tokens_seen": 406116672,
      "step": 7248
    },
    {
      "epoch": 16.14476614699332,
      "grad_norm": 13.294876098632812,
      "learning_rate": 1e-06,
      "loss": 0.3414,
      "num_input_tokens_seen": 406175172,
      "step": 7249
    },
    {
      "epoch": 16.14476614699332,
      "loss": 0.3227793574333191,
      "loss_ce": 8.648384391563013e-05,
      "loss_iou": 0.1533203125,
      "loss_num": 0.0031280517578125,
      "loss_xval": 0.322265625,
      "num_input_tokens_seen": 406175172,
      "step": 7249
    },
    {
      "epoch": 16.146993318485524,
      "grad_norm": 18.024934768676758,
      "learning_rate": 1e-06,
      "loss": 0.4407,
      "num_input_tokens_seen": 406234160,
      "step": 7250
    },
    {
      "epoch": 16.146993318485524,
      "eval_seeclick_web_CIoU": 0.588862270116806,
      "eval_seeclick_web_GIoU": 0.587228536605835,
      "eval_seeclick_web_IoU": 0.6071678400039673,
      "eval_seeclick_web_MAE_all": 0.015188148012384772,
      "eval_seeclick_web_MAE_h": 0.007499874569475651,
      "eval_seeclick_web_MAE_w": 0.015260101296007633,
      "eval_seeclick_web_MAE_x_boxes": 0.008470539702102542,
      "eval_seeclick_web_MAE_y_boxes": 0.02111952123232186,
      "eval_seeclick_web_inside_bbox": 0.9166666567325592,
      "eval_seeclick_web_loss": 0.9167152643203735,
      "eval_seeclick_web_loss_ce": 0.00014891428872942924,
      "eval_seeclick_web_loss_iou": 0.4229736328125,
      "eval_seeclick_web_loss_num": 0.01216888427734375,
      "eval_seeclick_web_loss_xval": 0.906982421875,
      "eval_seeclick_web_runtime": 20.8185,
      "eval_seeclick_web_samples_per_second": 2.402,
      "eval_seeclick_web_steps_per_second": 0.096,
      "num_input_tokens_seen": 406234160,
      "step": 7250
    },
    {
      "epoch": 16.146993318485524,
      "eval_icons_CIoU": 0.26575663685798645,
      "eval_icons_GIoU": 0.2966430187225342,
      "eval_icons_IoU": 0.3484746217727661,
      "eval_icons_MAE_all": 0.06336861476302147,
      "eval_icons_MAE_h": 0.03217336814850569,
      "eval_icons_MAE_w": 0.07706875540316105,
      "eval_icons_MAE_x_boxes": 0.05414869636297226,
      "eval_icons_MAE_y_boxes": 0.037681372836232185,
      "eval_icons_inside_bbox": 0.59375,
      "eval_icons_loss": 1.7173901796340942,
      "eval_icons_loss_ce": 0.00019168824655935168,
      "eval_icons_loss_iou": 0.66650390625,
      "eval_icons_loss_num": 0.061092376708984375,
      "eval_icons_loss_xval": 1.638671875,
      "eval_icons_runtime": 20.0031,
      "eval_icons_samples_per_second": 2.5,
      "eval_icons_steps_per_second": 0.1,
      "num_input_tokens_seen": 406234160,
      "step": 7250
    },
    {
      "epoch": 16.146993318485524,
      "eval_screenspot_CIoU": 0.38737640778223675,
      "eval_screenspot_GIoU": 0.40602253874142963,
      "eval_screenspot_IoU": 0.45554816722869873,
      "eval_screenspot_MAE_all": 0.05349355190992355,
      "eval_screenspot_MAE_h": 0.039430550610025726,
      "eval_screenspot_MAE_w": 0.0571071021258831,
      "eval_screenspot_MAE_x_boxes": 0.0624094990392526,
      "eval_screenspot_MAE_y_boxes": 0.0382879643390576,
      "eval_screenspot_inside_bbox": 0.7041666706403097,
      "eval_screenspot_loss": 1.5262062549591064,
      "eval_screenspot_loss_ce": 0.00022643499445014945,
      "eval_screenspot_loss_iou": 0.6376953125,
      "eval_screenspot_loss_num": 0.06162389119466146,
      "eval_screenspot_loss_xval": 1.58349609375,
      "eval_screenspot_runtime": 33.0274,
      "eval_screenspot_samples_per_second": 2.695,
      "eval_screenspot_steps_per_second": 0.091,
      "num_input_tokens_seen": 406234160,
      "step": 7250
    },
    {
      "epoch": 16.146993318485524,
      "eval_compot_CIoU": 0.34477105736732483,
      "eval_compot_GIoU": 0.3559461981058121,
      "eval_compot_IoU": 0.40127159655094147,
      "eval_compot_MAE_all": 0.01874891249462962,
      "eval_compot_MAE_h": 0.011277861427515745,
      "eval_compot_MAE_w": 0.02132485620677471,
      "eval_compot_MAE_x_boxes": 0.029761233367025852,
      "eval_compot_MAE_y_boxes": 0.0067885443568229675,
      "eval_compot_inside_bbox": 0.6302083432674408,
      "eval_compot_loss": 1.402956247329712,
      "eval_compot_loss_ce": 0.00014377458137460053,
      "eval_compot_loss_iou": 0.6480712890625,
      "eval_compot_loss_num": 0.017492294311523438,
      "eval_compot_loss_xval": 1.3837890625,
      "eval_compot_runtime": 20.4358,
      "eval_compot_samples_per_second": 2.447,
      "eval_compot_steps_per_second": 0.098,
      "num_input_tokens_seen": 406234160,
      "step": 7250
    },
    {
      "epoch": 16.146993318485524,
      "eval_custom_ui_val_CIoU": 0.4759764571984609,
      "eval_custom_ui_val_GIoU": 0.48175321850511765,
      "eval_custom_ui_val_IoU": 0.5367199348078834,
      "eval_custom_ui_val_MAE_all": 0.027078964850968786,
      "eval_custom_ui_val_MAE_h": 0.014058138916475905,
      "eval_custom_ui_val_MAE_w": 0.036136620212346315,
      "eval_custom_ui_val_MAE_x_boxes": 0.0336780981419401,
      "eval_custom_ui_val_MAE_y_boxes": 0.012756779815794693,
      "eval_custom_ui_val_inside_bbox": 0.7719907429483202,
      "eval_custom_ui_val_loss": 1.1681312322616577,
      "eval_custom_ui_val_loss_ce": 0.00017250773412848098,
      "eval_custom_ui_val_loss_iou": 0.5040554470486112,
      "eval_custom_ui_val_loss_num": 0.023714171515570745,
      "eval_custom_ui_val_loss_xval": 1.1267903645833333,
      "eval_custom_ui_val_runtime": 56.1097,
      "eval_custom_ui_val_samples_per_second": 4.723,
      "eval_custom_ui_val_steps_per_second": 0.16,
      "num_input_tokens_seen": 406234160,
      "step": 7250
    },
    {
      "epoch": 16.146993318485524,
      "loss": 0.8809858560562134,
      "loss_ce": 0.0001264033926418051,
      "loss_iou": 0.3984375,
      "loss_num": 0.016845703125,
      "loss_xval": 0.8828125,
      "num_input_tokens_seen": 406234160,
      "step": 7250
    },
    {
      "epoch": 16.14922048997773,
      "grad_norm": 13.051247596740723,
      "learning_rate": 1e-06,
      "loss": 0.2912,
      "num_input_tokens_seen": 406287840,
      "step": 7251
    },
    {
      "epoch": 16.14922048997773,
      "loss": 0.3833409249782562,
      "loss_ce": 0.00010119502258021384,
      "loss_iou": 0.1728515625,
      "loss_num": 0.00738525390625,
      "loss_xval": 0.3828125,
      "num_input_tokens_seen": 406287840,
      "step": 7251
    },
    {
      "epoch": 16.151447661469934,
      "grad_norm": 17.345809936523438,
      "learning_rate": 1e-06,
      "loss": 0.4775,
      "num_input_tokens_seen": 406344524,
      "step": 7252
    },
    {
      "epoch": 16.151447661469934,
      "loss": 0.3954039514064789,
      "loss_ce": 7.925261161290109e-05,
      "loss_iou": 0.1787109375,
      "loss_num": 0.00775146484375,
      "loss_xval": 0.39453125,
      "num_input_tokens_seen": 406344524,
      "step": 7252
    },
    {
      "epoch": 16.15367483296214,
      "grad_norm": 17.00496482849121,
      "learning_rate": 1e-06,
      "loss": 0.389,
      "num_input_tokens_seen": 406399856,
      "step": 7253
    },
    {
      "epoch": 16.15367483296214,
      "loss": 0.35250192880630493,
      "loss_ce": 8.492142660543323e-05,
      "loss_iou": 0.1533203125,
      "loss_num": 0.00927734375,
      "loss_xval": 0.3515625,
      "num_input_tokens_seen": 406399856,
      "step": 7253
    },
    {
      "epoch": 16.155902004454344,
      "grad_norm": 13.97536563873291,
      "learning_rate": 1e-06,
      "loss": 0.3508,
      "num_input_tokens_seen": 406457204,
      "step": 7254
    },
    {
      "epoch": 16.155902004454344,
      "loss": 0.3879046142101288,
      "loss_ce": 8.720727055333555e-05,
      "loss_iou": 0.1767578125,
      "loss_num": 0.0067138671875,
      "loss_xval": 0.388671875,
      "num_input_tokens_seen": 406457204,
      "step": 7254
    },
    {
      "epoch": 16.15812917594655,
      "grad_norm": 14.464372634887695,
      "learning_rate": 1e-06,
      "loss": 0.4449,
      "num_input_tokens_seen": 406513904,
      "step": 7255
    },
    {
      "epoch": 16.15812917594655,
      "loss": 0.37365224957466125,
      "loss_ce": 0.00023914946359582245,
      "loss_iou": 0.1533203125,
      "loss_num": 0.01348876953125,
      "loss_xval": 0.373046875,
      "num_input_tokens_seen": 406513904,
      "step": 7255
    },
    {
      "epoch": 16.160356347438753,
      "grad_norm": 14.928121566772461,
      "learning_rate": 1e-06,
      "loss": 0.4961,
      "num_input_tokens_seen": 406570348,
      "step": 7256
    },
    {
      "epoch": 16.160356347438753,
      "loss": 0.7330120801925659,
      "loss_ce": 0.00010193722846452147,
      "loss_iou": 0.259765625,
      "loss_num": 0.04248046875,
      "loss_xval": 0.734375,
      "num_input_tokens_seen": 406570348,
      "step": 7256
    },
    {
      "epoch": 16.16258351893096,
      "grad_norm": 20.846715927124023,
      "learning_rate": 1e-06,
      "loss": 0.6857,
      "num_input_tokens_seen": 406624784,
      "step": 7257
    },
    {
      "epoch": 16.16258351893096,
      "loss": 0.9515559673309326,
      "loss_ce": 0.00013993156608194113,
      "loss_iou": 0.369140625,
      "loss_num": 0.042724609375,
      "loss_xval": 0.953125,
      "num_input_tokens_seen": 406624784,
      "step": 7257
    },
    {
      "epoch": 16.164810690423163,
      "grad_norm": 14.39743423461914,
      "learning_rate": 1e-06,
      "loss": 0.5994,
      "num_input_tokens_seen": 406679488,
      "step": 7258
    },
    {
      "epoch": 16.164810690423163,
      "loss": 0.450466126203537,
      "loss_ce": 8.771609282121062e-05,
      "loss_iou": 0.19921875,
      "loss_num": 0.01025390625,
      "loss_xval": 0.451171875,
      "num_input_tokens_seen": 406679488,
      "step": 7258
    },
    {
      "epoch": 16.167037861915368,
      "grad_norm": 31.99239158630371,
      "learning_rate": 1e-06,
      "loss": 0.3736,
      "num_input_tokens_seen": 406737108,
      "step": 7259
    },
    {
      "epoch": 16.167037861915368,
      "loss": 0.25009024143218994,
      "loss_ce": 9.024170140037313e-05,
      "loss_iou": 0.1083984375,
      "loss_num": 0.006744384765625,
      "loss_xval": 0.25,
      "num_input_tokens_seen": 406737108,
      "step": 7259
    },
    {
      "epoch": 16.169265033407573,
      "grad_norm": 28.246858596801758,
      "learning_rate": 1e-06,
      "loss": 0.3793,
      "num_input_tokens_seen": 406794236,
      "step": 7260
    },
    {
      "epoch": 16.169265033407573,
      "loss": 0.3895048499107361,
      "loss_ce": 0.00010054315498564392,
      "loss_iou": 0.17578125,
      "loss_num": 0.007659912109375,
      "loss_xval": 0.388671875,
      "num_input_tokens_seen": 406794236,
      "step": 7260
    },
    {
      "epoch": 16.171492204899778,
      "grad_norm": 13.259971618652344,
      "learning_rate": 1e-06,
      "loss": 0.3625,
      "num_input_tokens_seen": 406851408,
      "step": 7261
    },
    {
      "epoch": 16.171492204899778,
      "loss": 0.39652132987976074,
      "loss_ce": 9.797551319934428e-05,
      "loss_iou": 0.17578125,
      "loss_num": 0.00897216796875,
      "loss_xval": 0.396484375,
      "num_input_tokens_seen": 406851408,
      "step": 7261
    },
    {
      "epoch": 16.173719376391983,
      "grad_norm": 18.107431411743164,
      "learning_rate": 1e-06,
      "loss": 0.4601,
      "num_input_tokens_seen": 406909764,
      "step": 7262
    },
    {
      "epoch": 16.173719376391983,
      "loss": 0.5472245216369629,
      "loss_ce": 0.0001053560699801892,
      "loss_iou": 0.236328125,
      "loss_num": 0.01507568359375,
      "loss_xval": 0.546875,
      "num_input_tokens_seen": 406909764,
      "step": 7262
    },
    {
      "epoch": 16.175946547884188,
      "grad_norm": 19.771875381469727,
      "learning_rate": 1e-06,
      "loss": 0.4511,
      "num_input_tokens_seen": 406967348,
      "step": 7263
    },
    {
      "epoch": 16.175946547884188,
      "loss": 0.3307103216648102,
      "loss_ce": 8.286767115350813e-05,
      "loss_iou": 0.150390625,
      "loss_num": 0.005889892578125,
      "loss_xval": 0.330078125,
      "num_input_tokens_seen": 406967348,
      "step": 7263
    },
    {
      "epoch": 16.178173719376392,
      "grad_norm": 20.80531883239746,
      "learning_rate": 1e-06,
      "loss": 0.4105,
      "num_input_tokens_seen": 407023044,
      "step": 7264
    },
    {
      "epoch": 16.178173719376392,
      "loss": 0.40901893377304077,
      "loss_ce": 8.336683094967157e-05,
      "loss_iou": 0.189453125,
      "loss_num": 0.006011962890625,
      "loss_xval": 0.408203125,
      "num_input_tokens_seen": 407023044,
      "step": 7264
    },
    {
      "epoch": 16.180400890868597,
      "grad_norm": 15.4229736328125,
      "learning_rate": 1e-06,
      "loss": 0.366,
      "num_input_tokens_seen": 407078844,
      "step": 7265
    },
    {
      "epoch": 16.180400890868597,
      "loss": 0.22904685139656067,
      "loss_ce": 0.0001039912604028359,
      "loss_iou": 0.09228515625,
      "loss_num": 0.00885009765625,
      "loss_xval": 0.228515625,
      "num_input_tokens_seen": 407078844,
      "step": 7265
    },
    {
      "epoch": 16.182628062360802,
      "grad_norm": 15.534099578857422,
      "learning_rate": 1e-06,
      "loss": 0.4986,
      "num_input_tokens_seen": 407135112,
      "step": 7266
    },
    {
      "epoch": 16.182628062360802,
      "loss": 0.41431134939193726,
      "loss_ce": 0.00012679336941801012,
      "loss_iou": 0.1826171875,
      "loss_num": 0.00994873046875,
      "loss_xval": 0.4140625,
      "num_input_tokens_seen": 407135112,
      "step": 7266
    },
    {
      "epoch": 16.184855233853007,
      "grad_norm": 11.986660957336426,
      "learning_rate": 1e-06,
      "loss": 0.3914,
      "num_input_tokens_seen": 407193344,
      "step": 7267
    },
    {
      "epoch": 16.184855233853007,
      "loss": 0.3952281177043915,
      "loss_ce": 8.651654206914827e-05,
      "loss_iou": 0.1650390625,
      "loss_num": 0.01300048828125,
      "loss_xval": 0.39453125,
      "num_input_tokens_seen": 407193344,
      "step": 7267
    },
    {
      "epoch": 16.187082405345212,
      "grad_norm": 14.020734786987305,
      "learning_rate": 1e-06,
      "loss": 0.4324,
      "num_input_tokens_seen": 407249716,
      "step": 7268
    },
    {
      "epoch": 16.187082405345212,
      "loss": 0.48159003257751465,
      "loss_ce": 0.0001447216491214931,
      "loss_iou": 0.2060546875,
      "loss_num": 0.013671875,
      "loss_xval": 0.48046875,
      "num_input_tokens_seen": 407249716,
      "step": 7268
    },
    {
      "epoch": 16.189309576837417,
      "grad_norm": 11.799369812011719,
      "learning_rate": 1e-06,
      "loss": 0.3835,
      "num_input_tokens_seen": 407306452,
      "step": 7269
    },
    {
      "epoch": 16.189309576837417,
      "loss": 0.3408535122871399,
      "loss_ce": 9.422671428183094e-05,
      "loss_iou": 0.1552734375,
      "loss_num": 0.005950927734375,
      "loss_xval": 0.33984375,
      "num_input_tokens_seen": 407306452,
      "step": 7269
    },
    {
      "epoch": 16.19153674832962,
      "grad_norm": 14.987945556640625,
      "learning_rate": 1e-06,
      "loss": 0.3707,
      "num_input_tokens_seen": 407363312,
      "step": 7270
    },
    {
      "epoch": 16.19153674832962,
      "loss": 0.2843882739543915,
      "loss_ce": 8.651558891870081e-05,
      "loss_iou": 0.12353515625,
      "loss_num": 0.007476806640625,
      "loss_xval": 0.28515625,
      "num_input_tokens_seen": 407363312,
      "step": 7270
    },
    {
      "epoch": 16.193763919821826,
      "grad_norm": 14.705431938171387,
      "learning_rate": 1e-06,
      "loss": 0.3561,
      "num_input_tokens_seen": 407417848,
      "step": 7271
    },
    {
      "epoch": 16.193763919821826,
      "loss": 0.30038201808929443,
      "loss_ce": 8.906603034120053e-05,
      "loss_iou": 0.123046875,
      "loss_num": 0.01080322265625,
      "loss_xval": 0.30078125,
      "num_input_tokens_seen": 407417848,
      "step": 7271
    },
    {
      "epoch": 16.19599109131403,
      "grad_norm": 33.002174377441406,
      "learning_rate": 1e-06,
      "loss": 0.5219,
      "num_input_tokens_seen": 407475472,
      "step": 7272
    },
    {
      "epoch": 16.19599109131403,
      "loss": 0.510102391242981,
      "loss_ce": 9.262951789423823e-05,
      "loss_iou": 0.2197265625,
      "loss_num": 0.01409912109375,
      "loss_xval": 0.51171875,
      "num_input_tokens_seen": 407475472,
      "step": 7272
    },
    {
      "epoch": 16.198218262806236,
      "grad_norm": 29.614242553710938,
      "learning_rate": 1e-06,
      "loss": 0.3678,
      "num_input_tokens_seen": 407531476,
      "step": 7273
    },
    {
      "epoch": 16.198218262806236,
      "loss": 0.37849289178848267,
      "loss_ce": 7.491311407648027e-05,
      "loss_iou": 0.1650390625,
      "loss_num": 0.00958251953125,
      "loss_xval": 0.37890625,
      "num_input_tokens_seen": 407531476,
      "step": 7273
    },
    {
      "epoch": 16.20044543429844,
      "grad_norm": 16.283767700195312,
      "learning_rate": 1e-06,
      "loss": 0.4892,
      "num_input_tokens_seen": 407587100,
      "step": 7274
    },
    {
      "epoch": 16.20044543429844,
      "loss": 0.5279887914657593,
      "loss_ce": 9.569715621182695e-05,
      "loss_iou": 0.22265625,
      "loss_num": 0.0166015625,
      "loss_xval": 0.52734375,
      "num_input_tokens_seen": 407587100,
      "step": 7274
    },
    {
      "epoch": 16.202672605790646,
      "grad_norm": 16.31856346130371,
      "learning_rate": 1e-06,
      "loss": 0.4131,
      "num_input_tokens_seen": 407643272,
      "step": 7275
    },
    {
      "epoch": 16.202672605790646,
      "loss": 0.40553590655326843,
      "loss_ce": 0.00010989317524945363,
      "loss_iou": 0.17578125,
      "loss_num": 0.01080322265625,
      "loss_xval": 0.40625,
      "num_input_tokens_seen": 407643272,
      "step": 7275
    },
    {
      "epoch": 16.20489977728285,
      "grad_norm": 18.60124969482422,
      "learning_rate": 1e-06,
      "loss": 0.3314,
      "num_input_tokens_seen": 407699796,
      "step": 7276
    },
    {
      "epoch": 16.20489977728285,
      "loss": 0.28377124667167664,
      "loss_ce": 7.983081741258502e-05,
      "loss_iou": 0.125,
      "loss_num": 0.006591796875,
      "loss_xval": 0.283203125,
      "num_input_tokens_seen": 407699796,
      "step": 7276
    },
    {
      "epoch": 16.207126948775056,
      "grad_norm": 22.093128204345703,
      "learning_rate": 1e-06,
      "loss": 0.4329,
      "num_input_tokens_seen": 407753556,
      "step": 7277
    },
    {
      "epoch": 16.207126948775056,
      "loss": 0.4864369034767151,
      "loss_ce": 0.00010880557238124311,
      "loss_iou": 0.2138671875,
      "loss_num": 0.0118408203125,
      "loss_xval": 0.486328125,
      "num_input_tokens_seen": 407753556,
      "step": 7277
    },
    {
      "epoch": 16.20935412026726,
      "grad_norm": 17.99799156188965,
      "learning_rate": 1e-06,
      "loss": 0.3275,
      "num_input_tokens_seen": 407811196,
      "step": 7278
    },
    {
      "epoch": 16.20935412026726,
      "loss": 0.2680291533470154,
      "loss_ce": 8.48027557367459e-05,
      "loss_iou": 0.1201171875,
      "loss_num": 0.0054931640625,
      "loss_xval": 0.267578125,
      "num_input_tokens_seen": 407811196,
      "step": 7278
    },
    {
      "epoch": 16.211581291759465,
      "grad_norm": 62.915122985839844,
      "learning_rate": 1e-06,
      "loss": 0.4762,
      "num_input_tokens_seen": 407866648,
      "step": 7279
    },
    {
      "epoch": 16.211581291759465,
      "loss": 0.3208872675895691,
      "loss_ce": 8.647728100186214e-05,
      "loss_iou": 0.1494140625,
      "loss_num": 0.0042724609375,
      "loss_xval": 0.3203125,
      "num_input_tokens_seen": 407866648,
      "step": 7279
    },
    {
      "epoch": 16.21380846325167,
      "grad_norm": 24.828845977783203,
      "learning_rate": 1e-06,
      "loss": 0.5674,
      "num_input_tokens_seen": 407922120,
      "step": 7280
    },
    {
      "epoch": 16.21380846325167,
      "loss": 0.4975318908691406,
      "loss_ce": 9.53997514443472e-05,
      "loss_iou": 0.2138671875,
      "loss_num": 0.01409912109375,
      "loss_xval": 0.498046875,
      "num_input_tokens_seen": 407922120,
      "step": 7280
    },
    {
      "epoch": 16.216035634743875,
      "grad_norm": 16.575681686401367,
      "learning_rate": 1e-06,
      "loss": 0.4067,
      "num_input_tokens_seen": 407974588,
      "step": 7281
    },
    {
      "epoch": 16.216035634743875,
      "loss": 0.5100204944610596,
      "loss_ce": 0.00104833475779742,
      "loss_iou": 0.208984375,
      "loss_num": 0.01806640625,
      "loss_xval": 0.5078125,
      "num_input_tokens_seen": 407974588,
      "step": 7281
    },
    {
      "epoch": 16.21826280623608,
      "grad_norm": 16.462749481201172,
      "learning_rate": 1e-06,
      "loss": 0.3674,
      "num_input_tokens_seen": 408030196,
      "step": 7282
    },
    {
      "epoch": 16.21826280623608,
      "loss": 0.43087995052337646,
      "loss_ce": 9.384381701238453e-05,
      "loss_iou": 0.1787109375,
      "loss_num": 0.014892578125,
      "loss_xval": 0.431640625,
      "num_input_tokens_seen": 408030196,
      "step": 7282
    },
    {
      "epoch": 16.220489977728285,
      "grad_norm": 18.638999938964844,
      "learning_rate": 1e-06,
      "loss": 0.4374,
      "num_input_tokens_seen": 408084800,
      "step": 7283
    },
    {
      "epoch": 16.220489977728285,
      "loss": 0.4949721395969391,
      "loss_ce": 9.908679930958897e-05,
      "loss_iou": 0.22265625,
      "loss_num": 0.00982666015625,
      "loss_xval": 0.494140625,
      "num_input_tokens_seen": 408084800,
      "step": 7283
    },
    {
      "epoch": 16.22271714922049,
      "grad_norm": 13.85960578918457,
      "learning_rate": 1e-06,
      "loss": 0.3035,
      "num_input_tokens_seen": 408142404,
      "step": 7284
    },
    {
      "epoch": 16.22271714922049,
      "loss": 0.3523852527141571,
      "loss_ce": 9.034241520566866e-05,
      "loss_iou": 0.1513671875,
      "loss_num": 0.00970458984375,
      "loss_xval": 0.3515625,
      "num_input_tokens_seen": 408142404,
      "step": 7284
    },
    {
      "epoch": 16.224944320712694,
      "grad_norm": 23.113548278808594,
      "learning_rate": 1e-06,
      "loss": 0.5333,
      "num_input_tokens_seen": 408197056,
      "step": 7285
    },
    {
      "epoch": 16.224944320712694,
      "loss": 0.5909112691879272,
      "loss_ce": 9.095118002733216e-05,
      "loss_iou": 0.271484375,
      "loss_num": 0.00946044921875,
      "loss_xval": 0.58984375,
      "num_input_tokens_seen": 408197056,
      "step": 7285
    },
    {
      "epoch": 16.2271714922049,
      "grad_norm": 19.409629821777344,
      "learning_rate": 1e-06,
      "loss": 0.3651,
      "num_input_tokens_seen": 408249604,
      "step": 7286
    },
    {
      "epoch": 16.2271714922049,
      "loss": 0.36336296796798706,
      "loss_ce": 8.17275868030265e-05,
      "loss_iou": 0.1650390625,
      "loss_num": 0.0067138671875,
      "loss_xval": 0.36328125,
      "num_input_tokens_seen": 408249604,
      "step": 7286
    },
    {
      "epoch": 16.229398663697104,
      "grad_norm": 17.79453468322754,
      "learning_rate": 1e-06,
      "loss": 0.328,
      "num_input_tokens_seen": 408306396,
      "step": 7287
    },
    {
      "epoch": 16.229398663697104,
      "loss": 0.34055426716804504,
      "loss_ce": 0.00010017414024332538,
      "loss_iou": 0.158203125,
      "loss_num": 0.004669189453125,
      "loss_xval": 0.33984375,
      "num_input_tokens_seen": 408306396,
      "step": 7287
    },
    {
      "epoch": 16.23162583518931,
      "grad_norm": 18.441938400268555,
      "learning_rate": 1e-06,
      "loss": 0.4376,
      "num_input_tokens_seen": 408362596,
      "step": 7288
    },
    {
      "epoch": 16.23162583518931,
      "loss": 0.470440149307251,
      "loss_ce": 0.00010323309834348038,
      "loss_iou": 0.2158203125,
      "loss_num": 0.00787353515625,
      "loss_xval": 0.470703125,
      "num_input_tokens_seen": 408362596,
      "step": 7288
    },
    {
      "epoch": 16.233853006681514,
      "grad_norm": 24.354867935180664,
      "learning_rate": 1e-06,
      "loss": 0.3503,
      "num_input_tokens_seen": 408417028,
      "step": 7289
    },
    {
      "epoch": 16.233853006681514,
      "loss": 0.3560473918914795,
      "loss_ce": 9.036185656441376e-05,
      "loss_iou": 0.158203125,
      "loss_num": 0.00799560546875,
      "loss_xval": 0.35546875,
      "num_input_tokens_seen": 408417028,
      "step": 7289
    },
    {
      "epoch": 16.23608017817372,
      "grad_norm": 15.969454765319824,
      "learning_rate": 1e-06,
      "loss": 0.5287,
      "num_input_tokens_seen": 408472732,
      "step": 7290
    },
    {
      "epoch": 16.23608017817372,
      "loss": 0.5947001576423645,
      "loss_ce": 9.568793757352978e-05,
      "loss_iou": 0.255859375,
      "loss_num": 0.0167236328125,
      "loss_xval": 0.59375,
      "num_input_tokens_seen": 408472732,
      "step": 7290
    },
    {
      "epoch": 16.238307349665924,
      "grad_norm": 19.128751754760742,
      "learning_rate": 1e-06,
      "loss": 0.3783,
      "num_input_tokens_seen": 408531592,
      "step": 7291
    },
    {
      "epoch": 16.238307349665924,
      "loss": 0.37436580657958984,
      "loss_ce": 9.823799337027594e-05,
      "loss_iou": 0.16015625,
      "loss_num": 0.01092529296875,
      "loss_xval": 0.375,
      "num_input_tokens_seen": 408531592,
      "step": 7291
    },
    {
      "epoch": 16.24053452115813,
      "grad_norm": 20.96180534362793,
      "learning_rate": 1e-06,
      "loss": 0.4015,
      "num_input_tokens_seen": 408587248,
      "step": 7292
    },
    {
      "epoch": 16.24053452115813,
      "loss": 0.3449326753616333,
      "loss_ce": 8.403346146224067e-05,
      "loss_iou": 0.1572265625,
      "loss_num": 0.006011962890625,
      "loss_xval": 0.345703125,
      "num_input_tokens_seen": 408587248,
      "step": 7292
    },
    {
      "epoch": 16.242761692650333,
      "grad_norm": 16.29045867919922,
      "learning_rate": 1e-06,
      "loss": 0.3866,
      "num_input_tokens_seen": 408643652,
      "step": 7293
    },
    {
      "epoch": 16.242761692650333,
      "loss": 0.461269736289978,
      "loss_ce": 8.806748519418761e-05,
      "loss_iou": 0.208984375,
      "loss_num": 0.00872802734375,
      "loss_xval": 0.4609375,
      "num_input_tokens_seen": 408643652,
      "step": 7293
    },
    {
      "epoch": 16.244988864142538,
      "grad_norm": 16.806655883789062,
      "learning_rate": 1e-06,
      "loss": 0.4739,
      "num_input_tokens_seen": 408696544,
      "step": 7294
    },
    {
      "epoch": 16.244988864142538,
      "loss": 0.5039190053939819,
      "loss_ce": 0.00013484872761182487,
      "loss_iou": 0.212890625,
      "loss_num": 0.0157470703125,
      "loss_xval": 0.50390625,
      "num_input_tokens_seen": 408696544,
      "step": 7294
    },
    {
      "epoch": 16.247216035634743,
      "grad_norm": 19.150453567504883,
      "learning_rate": 1e-06,
      "loss": 0.5361,
      "num_input_tokens_seen": 408752192,
      "step": 7295
    },
    {
      "epoch": 16.247216035634743,
      "loss": 0.4750003218650818,
      "loss_ce": 8.576853724662215e-05,
      "loss_iou": 0.2099609375,
      "loss_num": 0.0111083984375,
      "loss_xval": 0.474609375,
      "num_input_tokens_seen": 408752192,
      "step": 7295
    },
    {
      "epoch": 16.249443207126948,
      "grad_norm": 14.885661125183105,
      "learning_rate": 1e-06,
      "loss": 0.4392,
      "num_input_tokens_seen": 408807428,
      "step": 7296
    },
    {
      "epoch": 16.249443207126948,
      "loss": 0.3516565263271332,
      "loss_ce": 9.402677096659318e-05,
      "loss_iou": 0.1591796875,
      "loss_num": 0.00653076171875,
      "loss_xval": 0.3515625,
      "num_input_tokens_seen": 408807428,
      "step": 7296
    },
    {
      "epoch": 16.251670378619153,
      "grad_norm": 20.185546875,
      "learning_rate": 1e-06,
      "loss": 0.6554,
      "num_input_tokens_seen": 408862660,
      "step": 7297
    },
    {
      "epoch": 16.251670378619153,
      "loss": 0.6360622644424438,
      "loss_ce": 7.591473695356399e-05,
      "loss_iou": 0.275390625,
      "loss_num": 0.017333984375,
      "loss_xval": 0.63671875,
      "num_input_tokens_seen": 408862660,
      "step": 7297
    },
    {
      "epoch": 16.253897550111358,
      "grad_norm": 24.526891708374023,
      "learning_rate": 1e-06,
      "loss": 0.5942,
      "num_input_tokens_seen": 408915492,
      "step": 7298
    },
    {
      "epoch": 16.253897550111358,
      "loss": 0.4688360095024109,
      "loss_ce": 8.602187153883278e-05,
      "loss_iou": 0.18359375,
      "loss_num": 0.0203857421875,
      "loss_xval": 0.46875,
      "num_input_tokens_seen": 408915492,
      "step": 7298
    },
    {
      "epoch": 16.256124721603562,
      "grad_norm": 28.41362953186035,
      "learning_rate": 1e-06,
      "loss": 0.3349,
      "num_input_tokens_seen": 408971936,
      "step": 7299
    },
    {
      "epoch": 16.256124721603562,
      "loss": 0.3919934928417206,
      "loss_ce": 0.00014778405602555722,
      "loss_iou": 0.1806640625,
      "loss_num": 0.005950927734375,
      "loss_xval": 0.392578125,
      "num_input_tokens_seen": 408971936,
      "step": 7299
    },
    {
      "epoch": 16.258351893095767,
      "grad_norm": 16.66225814819336,
      "learning_rate": 1e-06,
      "loss": 0.4285,
      "num_input_tokens_seen": 409026252,
      "step": 7300
    },
    {
      "epoch": 16.258351893095767,
      "loss": 0.4181760251522064,
      "loss_ce": 8.519048424204811e-05,
      "loss_iou": 0.1845703125,
      "loss_num": 0.0098876953125,
      "loss_xval": 0.41796875,
      "num_input_tokens_seen": 409026252,
      "step": 7300
    },
    {
      "epoch": 16.260579064587972,
      "grad_norm": 19.2691593170166,
      "learning_rate": 1e-06,
      "loss": 0.67,
      "num_input_tokens_seen": 409084124,
      "step": 7301
    },
    {
      "epoch": 16.260579064587972,
      "loss": 0.6911599636077881,
      "loss_ce": 0.00011992135114269331,
      "loss_iou": 0.275390625,
      "loss_num": 0.0279541015625,
      "loss_xval": 0.69140625,
      "num_input_tokens_seen": 409084124,
      "step": 7301
    },
    {
      "epoch": 16.262806236080177,
      "grad_norm": 14.751253128051758,
      "learning_rate": 1e-06,
      "loss": 0.3899,
      "num_input_tokens_seen": 409141304,
      "step": 7302
    },
    {
      "epoch": 16.262806236080177,
      "loss": 0.47578924894332886,
      "loss_ce": 8.124149462673813e-05,
      "loss_iou": 0.1982421875,
      "loss_num": 0.01611328125,
      "loss_xval": 0.4765625,
      "num_input_tokens_seen": 409141304,
      "step": 7302
    },
    {
      "epoch": 16.265033407572382,
      "grad_norm": 14.404290199279785,
      "learning_rate": 1e-06,
      "loss": 0.3463,
      "num_input_tokens_seen": 409197796,
      "step": 7303
    },
    {
      "epoch": 16.265033407572382,
      "loss": 0.26949194073677063,
      "loss_ce": 8.277669257950038e-05,
      "loss_iou": 0.1259765625,
      "loss_num": 0.003570556640625,
      "loss_xval": 0.26953125,
      "num_input_tokens_seen": 409197796,
      "step": 7303
    },
    {
      "epoch": 16.267260579064587,
      "grad_norm": 20.53440284729004,
      "learning_rate": 1e-06,
      "loss": 0.3827,
      "num_input_tokens_seen": 409252580,
      "step": 7304
    },
    {
      "epoch": 16.267260579064587,
      "loss": 0.4455221891403198,
      "loss_ce": 8.760465425439179e-05,
      "loss_iou": 0.1904296875,
      "loss_num": 0.01263427734375,
      "loss_xval": 0.4453125,
      "num_input_tokens_seen": 409252580,
      "step": 7304
    },
    {
      "epoch": 16.26948775055679,
      "grad_norm": 28.248842239379883,
      "learning_rate": 1e-06,
      "loss": 0.6714,
      "num_input_tokens_seen": 409309884,
      "step": 7305
    },
    {
      "epoch": 16.26948775055679,
      "loss": 0.570942759513855,
      "loss_ce": 0.0006302678375504911,
      "loss_iou": 0.2412109375,
      "loss_num": 0.017822265625,
      "loss_xval": 0.5703125,
      "num_input_tokens_seen": 409309884,
      "step": 7305
    },
    {
      "epoch": 16.271714922048996,
      "grad_norm": 15.996672630310059,
      "learning_rate": 1e-06,
      "loss": 0.5986,
      "num_input_tokens_seen": 409366592,
      "step": 7306
    },
    {
      "epoch": 16.271714922048996,
      "loss": 0.5386756658554077,
      "loss_ce": 0.000101389319752343,
      "loss_iou": 0.2216796875,
      "loss_num": 0.0189208984375,
      "loss_xval": 0.5390625,
      "num_input_tokens_seen": 409366592,
      "step": 7306
    },
    {
      "epoch": 16.2739420935412,
      "grad_norm": 25.504066467285156,
      "learning_rate": 1e-06,
      "loss": 0.4661,
      "num_input_tokens_seen": 409423384,
      "step": 7307
    },
    {
      "epoch": 16.2739420935412,
      "loss": 0.5826142430305481,
      "loss_ce": 9.470840450376272e-05,
      "loss_iou": 0.263671875,
      "loss_num": 0.0107421875,
      "loss_xval": 0.58203125,
      "num_input_tokens_seen": 409423384,
      "step": 7307
    },
    {
      "epoch": 16.276169265033406,
      "grad_norm": 16.374908447265625,
      "learning_rate": 1e-06,
      "loss": 0.5241,
      "num_input_tokens_seen": 409479828,
      "step": 7308
    },
    {
      "epoch": 16.276169265033406,
      "loss": 0.6848694086074829,
      "loss_ce": 0.00017701313481666148,
      "loss_iou": 0.30859375,
      "loss_num": 0.01336669921875,
      "loss_xval": 0.68359375,
      "num_input_tokens_seen": 409479828,
      "step": 7308
    },
    {
      "epoch": 16.27839643652561,
      "grad_norm": 17.140676498413086,
      "learning_rate": 1e-06,
      "loss": 0.5788,
      "num_input_tokens_seen": 409536380,
      "step": 7309
    },
    {
      "epoch": 16.27839643652561,
      "loss": 0.7437722682952881,
      "loss_ce": 0.00011992130021099001,
      "loss_iou": 0.29296875,
      "loss_num": 0.031005859375,
      "loss_xval": 0.7421875,
      "num_input_tokens_seen": 409536380,
      "step": 7309
    },
    {
      "epoch": 16.280623608017816,
      "grad_norm": 14.995177268981934,
      "learning_rate": 1e-06,
      "loss": 0.4511,
      "num_input_tokens_seen": 409594828,
      "step": 7310
    },
    {
      "epoch": 16.280623608017816,
      "loss": 0.3734976649284363,
      "loss_ce": 8.45702743390575e-05,
      "loss_iou": 0.1630859375,
      "loss_num": 0.00933837890625,
      "loss_xval": 0.373046875,
      "num_input_tokens_seen": 409594828,
      "step": 7310
    },
    {
      "epoch": 16.28285077951002,
      "grad_norm": 16.360715866088867,
      "learning_rate": 1e-06,
      "loss": 0.4679,
      "num_input_tokens_seen": 409650308,
      "step": 7311
    },
    {
      "epoch": 16.28285077951002,
      "loss": 0.4848405718803406,
      "loss_ce": 9.936667629517615e-05,
      "loss_iou": 0.2177734375,
      "loss_num": 0.0098876953125,
      "loss_xval": 0.484375,
      "num_input_tokens_seen": 409650308,
      "step": 7311
    },
    {
      "epoch": 16.285077951002226,
      "grad_norm": 21.59230613708496,
      "learning_rate": 1e-06,
      "loss": 0.4159,
      "num_input_tokens_seen": 409706112,
      "step": 7312
    },
    {
      "epoch": 16.285077951002226,
      "loss": 0.5259721875190735,
      "loss_ce": 9.327764564659446e-05,
      "loss_iou": 0.251953125,
      "loss_num": 0.004608154296875,
      "loss_xval": 0.52734375,
      "num_input_tokens_seen": 409706112,
      "step": 7312
    },
    {
      "epoch": 16.28730512249443,
      "grad_norm": 15.556495666503906,
      "learning_rate": 1e-06,
      "loss": 0.3937,
      "num_input_tokens_seen": 409763944,
      "step": 7313
    },
    {
      "epoch": 16.28730512249443,
      "loss": 0.3481227159500122,
      "loss_ce": 0.00010025159281212837,
      "loss_iou": 0.1611328125,
      "loss_num": 0.0052490234375,
      "loss_xval": 0.34765625,
      "num_input_tokens_seen": 409763944,
      "step": 7313
    },
    {
      "epoch": 16.289532293986635,
      "grad_norm": 17.400310516357422,
      "learning_rate": 1e-06,
      "loss": 0.4426,
      "num_input_tokens_seen": 409815968,
      "step": 7314
    },
    {
      "epoch": 16.289532293986635,
      "loss": 0.4380715787410736,
      "loss_ce": 8.330351556651294e-05,
      "loss_iou": 0.18359375,
      "loss_num": 0.01416015625,
      "loss_xval": 0.4375,
      "num_input_tokens_seen": 409815968,
      "step": 7314
    },
    {
      "epoch": 16.29175946547884,
      "grad_norm": 17.484792709350586,
      "learning_rate": 1e-06,
      "loss": 0.3899,
      "num_input_tokens_seen": 409870240,
      "step": 7315
    },
    {
      "epoch": 16.29175946547884,
      "loss": 0.42489010095596313,
      "loss_ce": 8.540091221220791e-05,
      "loss_iou": 0.1708984375,
      "loss_num": 0.0164794921875,
      "loss_xval": 0.42578125,
      "num_input_tokens_seen": 409870240,
      "step": 7315
    },
    {
      "epoch": 16.293986636971045,
      "grad_norm": 14.591642379760742,
      "learning_rate": 1e-06,
      "loss": 0.3274,
      "num_input_tokens_seen": 409926680,
      "step": 7316
    },
    {
      "epoch": 16.293986636971045,
      "loss": 0.2797589600086212,
      "loss_ce": 9.588080865796655e-05,
      "loss_iou": 0.126953125,
      "loss_num": 0.005126953125,
      "loss_xval": 0.279296875,
      "num_input_tokens_seen": 409926680,
      "step": 7316
    },
    {
      "epoch": 16.29621380846325,
      "grad_norm": 13.230761528015137,
      "learning_rate": 1e-06,
      "loss": 0.3727,
      "num_input_tokens_seen": 409985004,
      "step": 7317
    },
    {
      "epoch": 16.29621380846325,
      "loss": 0.3750907778739929,
      "loss_ce": 9.078568109543994e-05,
      "loss_iou": 0.15625,
      "loss_num": 0.0125732421875,
      "loss_xval": 0.375,
      "num_input_tokens_seen": 409985004,
      "step": 7317
    },
    {
      "epoch": 16.29844097995546,
      "grad_norm": 32.36854934692383,
      "learning_rate": 1e-06,
      "loss": 0.4765,
      "num_input_tokens_seen": 410040396,
      "step": 7318
    },
    {
      "epoch": 16.29844097995546,
      "loss": 0.43508851528167725,
      "loss_ce": 9.096259600482881e-05,
      "loss_iou": 0.1962890625,
      "loss_num": 0.008544921875,
      "loss_xval": 0.435546875,
      "num_input_tokens_seen": 410040396,
      "step": 7318
    },
    {
      "epoch": 16.30066815144766,
      "grad_norm": 20.330373764038086,
      "learning_rate": 1e-06,
      "loss": 0.3185,
      "num_input_tokens_seen": 410093968,
      "step": 7319
    },
    {
      "epoch": 16.30066815144766,
      "loss": 0.3120901584625244,
      "loss_ce": 7.842134800739586e-05,
      "loss_iou": 0.134765625,
      "loss_num": 0.008544921875,
      "loss_xval": 0.3125,
      "num_input_tokens_seen": 410093968,
      "step": 7319
    },
    {
      "epoch": 16.302895322939868,
      "grad_norm": 15.776129722595215,
      "learning_rate": 1e-06,
      "loss": 0.3339,
      "num_input_tokens_seen": 410149384,
      "step": 7320
    },
    {
      "epoch": 16.302895322939868,
      "loss": 0.47568702697753906,
      "loss_ce": 0.00010110878793057054,
      "loss_iou": 0.2041015625,
      "loss_num": 0.01348876953125,
      "loss_xval": 0.4765625,
      "num_input_tokens_seen": 410149384,
      "step": 7320
    },
    {
      "epoch": 16.305122494432073,
      "grad_norm": 21.19259262084961,
      "learning_rate": 1e-06,
      "loss": 0.5786,
      "num_input_tokens_seen": 410206596,
      "step": 7321
    },
    {
      "epoch": 16.305122494432073,
      "loss": 0.36239707469940186,
      "loss_ce": 9.236540063284338e-05,
      "loss_iou": 0.1640625,
      "loss_num": 0.006866455078125,
      "loss_xval": 0.36328125,
      "num_input_tokens_seen": 410206596,
      "step": 7321
    },
    {
      "epoch": 16.307349665924278,
      "grad_norm": 22.16942596435547,
      "learning_rate": 1e-06,
      "loss": 0.4556,
      "num_input_tokens_seen": 410261328,
      "step": 7322
    },
    {
      "epoch": 16.307349665924278,
      "loss": 0.5388376712799072,
      "loss_ce": 8.029842138057575e-05,
      "loss_iou": 0.2099609375,
      "loss_num": 0.023681640625,
      "loss_xval": 0.5390625,
      "num_input_tokens_seen": 410261328,
      "step": 7322
    },
    {
      "epoch": 16.309576837416483,
      "grad_norm": 21.34149169921875,
      "learning_rate": 1e-06,
      "loss": 0.4369,
      "num_input_tokens_seen": 410317892,
      "step": 7323
    },
    {
      "epoch": 16.309576837416483,
      "loss": 0.5326899886131287,
      "loss_ce": 9.718516957946122e-05,
      "loss_iou": 0.2333984375,
      "loss_num": 0.01318359375,
      "loss_xval": 0.53125,
      "num_input_tokens_seen": 410317892,
      "step": 7323
    },
    {
      "epoch": 16.311804008908688,
      "grad_norm": 22.547149658203125,
      "learning_rate": 1e-06,
      "loss": 0.3742,
      "num_input_tokens_seen": 410372852,
      "step": 7324
    },
    {
      "epoch": 16.311804008908688,
      "loss": 0.3052588105201721,
      "loss_ce": 8.303693175548688e-05,
      "loss_iou": 0.1337890625,
      "loss_num": 0.007659912109375,
      "loss_xval": 0.3046875,
      "num_input_tokens_seen": 410372852,
      "step": 7324
    },
    {
      "epoch": 16.314031180400892,
      "grad_norm": 24.133289337158203,
      "learning_rate": 1e-06,
      "loss": 0.3867,
      "num_input_tokens_seen": 410428208,
      "step": 7325
    },
    {
      "epoch": 16.314031180400892,
      "loss": 0.3308970034122467,
      "loss_ce": 8.646737842354923e-05,
      "loss_iou": 0.1494140625,
      "loss_num": 0.006256103515625,
      "loss_xval": 0.330078125,
      "num_input_tokens_seen": 410428208,
      "step": 7325
    },
    {
      "epoch": 16.316258351893097,
      "grad_norm": 27.495380401611328,
      "learning_rate": 1e-06,
      "loss": 0.5133,
      "num_input_tokens_seen": 410483916,
      "step": 7326
    },
    {
      "epoch": 16.316258351893097,
      "loss": 0.5182612538337708,
      "loss_ce": 7.278715202119201e-05,
      "loss_iou": 0.2421875,
      "loss_num": 0.0068359375,
      "loss_xval": 0.51953125,
      "num_input_tokens_seen": 410483916,
      "step": 7326
    },
    {
      "epoch": 16.318485523385302,
      "grad_norm": 26.30192756652832,
      "learning_rate": 1e-06,
      "loss": 0.3812,
      "num_input_tokens_seen": 410539204,
      "step": 7327
    },
    {
      "epoch": 16.318485523385302,
      "loss": 0.47011250257492065,
      "loss_ce": 0.00014180486323311925,
      "loss_iou": 0.1982421875,
      "loss_num": 0.01483154296875,
      "loss_xval": 0.470703125,
      "num_input_tokens_seen": 410539204,
      "step": 7327
    },
    {
      "epoch": 16.320712694877507,
      "grad_norm": 17.40894317626953,
      "learning_rate": 1e-06,
      "loss": 0.4687,
      "num_input_tokens_seen": 410594636,
      "step": 7328
    },
    {
      "epoch": 16.320712694877507,
      "loss": 0.44930049777030945,
      "loss_ce": 8.17526743048802e-05,
      "loss_iou": 0.1943359375,
      "loss_num": 0.01226806640625,
      "loss_xval": 0.44921875,
      "num_input_tokens_seen": 410594636,
      "step": 7328
    },
    {
      "epoch": 16.322939866369712,
      "grad_norm": 21.12818145751953,
      "learning_rate": 1e-06,
      "loss": 0.3845,
      "num_input_tokens_seen": 410651876,
      "step": 7329
    },
    {
      "epoch": 16.322939866369712,
      "loss": 0.3459153175354004,
      "loss_ce": 9.013438830152154e-05,
      "loss_iou": 0.1630859375,
      "loss_num": 0.003936767578125,
      "loss_xval": 0.345703125,
      "num_input_tokens_seen": 410651876,
      "step": 7329
    },
    {
      "epoch": 16.325167037861917,
      "grad_norm": 10.89534854888916,
      "learning_rate": 1e-06,
      "loss": 0.5001,
      "num_input_tokens_seen": 410706904,
      "step": 7330
    },
    {
      "epoch": 16.325167037861917,
      "loss": 0.46097275614738464,
      "loss_ce": 9.627400140743703e-05,
      "loss_iou": 0.205078125,
      "loss_num": 0.01019287109375,
      "loss_xval": 0.4609375,
      "num_input_tokens_seen": 410706904,
      "step": 7330
    },
    {
      "epoch": 16.32739420935412,
      "grad_norm": 20.05289077758789,
      "learning_rate": 1e-06,
      "loss": 0.4095,
      "num_input_tokens_seen": 410763772,
      "step": 7331
    },
    {
      "epoch": 16.32739420935412,
      "loss": 0.3861430883407593,
      "loss_ce": 9.571410191711038e-05,
      "loss_iou": 0.1689453125,
      "loss_num": 0.0096435546875,
      "loss_xval": 0.38671875,
      "num_input_tokens_seen": 410763772,
      "step": 7331
    },
    {
      "epoch": 16.329621380846326,
      "grad_norm": 15.384933471679688,
      "learning_rate": 1e-06,
      "loss": 0.3819,
      "num_input_tokens_seen": 410823420,
      "step": 7332
    },
    {
      "epoch": 16.329621380846326,
      "loss": 0.4961845278739929,
      "loss_ce": 9.074924309970811e-05,
      "loss_iou": 0.2255859375,
      "loss_num": 0.00909423828125,
      "loss_xval": 0.49609375,
      "num_input_tokens_seen": 410823420,
      "step": 7332
    },
    {
      "epoch": 16.33184855233853,
      "grad_norm": 24.686939239501953,
      "learning_rate": 1e-06,
      "loss": 0.4537,
      "num_input_tokens_seen": 410877940,
      "step": 7333
    },
    {
      "epoch": 16.33184855233853,
      "loss": 0.5593637824058533,
      "loss_ce": 0.00015970882668625563,
      "loss_iou": 0.234375,
      "loss_num": 0.018310546875,
      "loss_xval": 0.55859375,
      "num_input_tokens_seen": 410877940,
      "step": 7333
    },
    {
      "epoch": 16.334075723830736,
      "grad_norm": 22.968791961669922,
      "learning_rate": 1e-06,
      "loss": 0.5139,
      "num_input_tokens_seen": 410932476,
      "step": 7334
    },
    {
      "epoch": 16.334075723830736,
      "loss": 0.39212775230407715,
      "loss_ce": 9.897441486828029e-05,
      "loss_iou": 0.171875,
      "loss_num": 0.00958251953125,
      "loss_xval": 0.392578125,
      "num_input_tokens_seen": 410932476,
      "step": 7334
    },
    {
      "epoch": 16.33630289532294,
      "grad_norm": 15.322009086608887,
      "learning_rate": 1e-06,
      "loss": 0.4757,
      "num_input_tokens_seen": 410990488,
      "step": 7335
    },
    {
      "epoch": 16.33630289532294,
      "loss": 0.5929901003837585,
      "loss_ce": 9.460109868086874e-05,
      "loss_iou": 0.240234375,
      "loss_num": 0.0223388671875,
      "loss_xval": 0.59375,
      "num_input_tokens_seen": 410990488,
      "step": 7335
    },
    {
      "epoch": 16.338530066815146,
      "grad_norm": 34.48351287841797,
      "learning_rate": 1e-06,
      "loss": 0.5019,
      "num_input_tokens_seen": 411046180,
      "step": 7336
    },
    {
      "epoch": 16.338530066815146,
      "loss": 0.581402063369751,
      "loss_ce": 0.00010323138849344105,
      "loss_iou": 0.2578125,
      "loss_num": 0.0135498046875,
      "loss_xval": 0.58203125,
      "num_input_tokens_seen": 411046180,
      "step": 7336
    },
    {
      "epoch": 16.34075723830735,
      "grad_norm": 16.476621627807617,
      "learning_rate": 1e-06,
      "loss": 0.3056,
      "num_input_tokens_seen": 411102856,
      "step": 7337
    },
    {
      "epoch": 16.34075723830735,
      "loss": 0.1867981106042862,
      "loss_ce": 9.157234308077022e-05,
      "loss_iou": 0.08251953125,
      "loss_num": 0.00439453125,
      "loss_xval": 0.1865234375,
      "num_input_tokens_seen": 411102856,
      "step": 7337
    },
    {
      "epoch": 16.342984409799556,
      "grad_norm": 19.494869232177734,
      "learning_rate": 1e-06,
      "loss": 0.3802,
      "num_input_tokens_seen": 411159860,
      "step": 7338
    },
    {
      "epoch": 16.342984409799556,
      "loss": 0.3621300756931305,
      "loss_ce": 0.0003136666491627693,
      "loss_iou": 0.1455078125,
      "loss_num": 0.01422119140625,
      "loss_xval": 0.361328125,
      "num_input_tokens_seen": 411159860,
      "step": 7338
    },
    {
      "epoch": 16.34521158129176,
      "grad_norm": 19.032352447509766,
      "learning_rate": 1e-06,
      "loss": 0.4447,
      "num_input_tokens_seen": 411215444,
      "step": 7339
    },
    {
      "epoch": 16.34521158129176,
      "loss": 0.3872949481010437,
      "loss_ce": 8.788703416939825e-05,
      "loss_iou": 0.1611328125,
      "loss_num": 0.0130615234375,
      "loss_xval": 0.38671875,
      "num_input_tokens_seen": 411215444,
      "step": 7339
    },
    {
      "epoch": 16.347438752783965,
      "grad_norm": 18.541751861572266,
      "learning_rate": 1e-06,
      "loss": 0.4312,
      "num_input_tokens_seen": 411270292,
      "step": 7340
    },
    {
      "epoch": 16.347438752783965,
      "loss": 0.5562987923622131,
      "loss_ce": 8.540366252418607e-05,
      "loss_iou": 0.2392578125,
      "loss_num": 0.015380859375,
      "loss_xval": 0.5546875,
      "num_input_tokens_seen": 411270292,
      "step": 7340
    },
    {
      "epoch": 16.34966592427617,
      "grad_norm": 13.023123741149902,
      "learning_rate": 1e-06,
      "loss": 0.5157,
      "num_input_tokens_seen": 411327140,
      "step": 7341
    },
    {
      "epoch": 16.34966592427617,
      "loss": 0.4642055630683899,
      "loss_ce": 9.424821473658085e-05,
      "loss_iou": 0.1923828125,
      "loss_num": 0.0157470703125,
      "loss_xval": 0.46484375,
      "num_input_tokens_seen": 411327140,
      "step": 7341
    },
    {
      "epoch": 16.351893095768375,
      "grad_norm": 18.090694427490234,
      "learning_rate": 1e-06,
      "loss": 0.428,
      "num_input_tokens_seen": 411383152,
      "step": 7342
    },
    {
      "epoch": 16.351893095768375,
      "loss": 0.46582120656967163,
      "loss_ce": 0.0009774666978046298,
      "loss_iou": 0.2021484375,
      "loss_num": 0.011962890625,
      "loss_xval": 0.46484375,
      "num_input_tokens_seen": 411383152,
      "step": 7342
    },
    {
      "epoch": 16.35412026726058,
      "grad_norm": 16.27545928955078,
      "learning_rate": 1e-06,
      "loss": 0.3941,
      "num_input_tokens_seen": 411438268,
      "step": 7343
    },
    {
      "epoch": 16.35412026726058,
      "loss": 0.41550177335739136,
      "loss_ce": 9.647855767980218e-05,
      "loss_iou": 0.1904296875,
      "loss_num": 0.0068359375,
      "loss_xval": 0.416015625,
      "num_input_tokens_seen": 411438268,
      "step": 7343
    },
    {
      "epoch": 16.356347438752785,
      "grad_norm": 22.559267044067383,
      "learning_rate": 1e-06,
      "loss": 0.4177,
      "num_input_tokens_seen": 411493464,
      "step": 7344
    },
    {
      "epoch": 16.356347438752785,
      "loss": 0.4500234127044678,
      "loss_ce": 7.222830026876181e-05,
      "loss_iou": 0.19140625,
      "loss_num": 0.01348876953125,
      "loss_xval": 0.44921875,
      "num_input_tokens_seen": 411493464,
      "step": 7344
    },
    {
      "epoch": 16.35857461024499,
      "grad_norm": 17.649221420288086,
      "learning_rate": 1e-06,
      "loss": 0.3428,
      "num_input_tokens_seen": 411550844,
      "step": 7345
    },
    {
      "epoch": 16.35857461024499,
      "loss": 0.46036189794540405,
      "loss_ce": 9.580078767612576e-05,
      "loss_iou": 0.1962890625,
      "loss_num": 0.013671875,
      "loss_xval": 0.4609375,
      "num_input_tokens_seen": 411550844,
      "step": 7345
    },
    {
      "epoch": 16.360801781737194,
      "grad_norm": 19.68630027770996,
      "learning_rate": 1e-06,
      "loss": 0.3448,
      "num_input_tokens_seen": 411606484,
      "step": 7346
    },
    {
      "epoch": 16.360801781737194,
      "loss": 0.31284722685813904,
      "loss_ce": 0.00010307719639968127,
      "loss_iou": 0.14453125,
      "loss_num": 0.00482177734375,
      "loss_xval": 0.3125,
      "num_input_tokens_seen": 411606484,
      "step": 7346
    },
    {
      "epoch": 16.3630289532294,
      "grad_norm": 23.66335678100586,
      "learning_rate": 1e-06,
      "loss": 0.4744,
      "num_input_tokens_seen": 411662964,
      "step": 7347
    },
    {
      "epoch": 16.3630289532294,
      "loss": 0.38772842288017273,
      "loss_ce": 0.00015515368431806564,
      "loss_iou": 0.171875,
      "loss_num": 0.00885009765625,
      "loss_xval": 0.38671875,
      "num_input_tokens_seen": 411662964,
      "step": 7347
    },
    {
      "epoch": 16.365256124721604,
      "grad_norm": 16.446657180786133,
      "learning_rate": 1e-06,
      "loss": 0.4439,
      "num_input_tokens_seen": 411718656,
      "step": 7348
    },
    {
      "epoch": 16.365256124721604,
      "loss": 0.40172719955444336,
      "loss_ce": 0.00011588144116103649,
      "loss_iou": 0.16796875,
      "loss_num": 0.01300048828125,
      "loss_xval": 0.40234375,
      "num_input_tokens_seen": 411718656,
      "step": 7348
    },
    {
      "epoch": 16.36748329621381,
      "grad_norm": 12.772607803344727,
      "learning_rate": 1e-06,
      "loss": 0.354,
      "num_input_tokens_seen": 411774636,
      "step": 7349
    },
    {
      "epoch": 16.36748329621381,
      "loss": 0.3649415373802185,
      "loss_ce": 7.335752889048308e-05,
      "loss_iou": 0.154296875,
      "loss_num": 0.0113525390625,
      "loss_xval": 0.365234375,
      "num_input_tokens_seen": 411774636,
      "step": 7349
    },
    {
      "epoch": 16.369710467706014,
      "grad_norm": 199.87049865722656,
      "learning_rate": 1e-06,
      "loss": 0.4038,
      "num_input_tokens_seen": 411828580,
      "step": 7350
    },
    {
      "epoch": 16.369710467706014,
      "loss": 0.5851800441741943,
      "loss_ce": 9.696916822576895e-05,
      "loss_iou": 0.26171875,
      "loss_num": 0.01251220703125,
      "loss_xval": 0.5859375,
      "num_input_tokens_seen": 411828580,
      "step": 7350
    },
    {
      "epoch": 16.37193763919822,
      "grad_norm": 13.645878791809082,
      "learning_rate": 1e-06,
      "loss": 0.3657,
      "num_input_tokens_seen": 411886588,
      "step": 7351
    },
    {
      "epoch": 16.37193763919822,
      "loss": 0.5355052947998047,
      "loss_ce": 0.00010485852544661611,
      "loss_iou": 0.2138671875,
      "loss_num": 0.021728515625,
      "loss_xval": 0.53515625,
      "num_input_tokens_seen": 411886588,
      "step": 7351
    },
    {
      "epoch": 16.374164810690424,
      "grad_norm": 16.982749938964844,
      "learning_rate": 1e-06,
      "loss": 0.4847,
      "num_input_tokens_seen": 411944540,
      "step": 7352
    },
    {
      "epoch": 16.374164810690424,
      "loss": 0.48775994777679443,
      "loss_ce": 8.901581168174744e-05,
      "loss_iou": 0.212890625,
      "loss_num": 0.01220703125,
      "loss_xval": 0.48828125,
      "num_input_tokens_seen": 411944540,
      "step": 7352
    },
    {
      "epoch": 16.37639198218263,
      "grad_norm": 19.0021915435791,
      "learning_rate": 1e-06,
      "loss": 0.5334,
      "num_input_tokens_seen": 412002580,
      "step": 7353
    },
    {
      "epoch": 16.37639198218263,
      "loss": 0.7117968797683716,
      "loss_ce": 0.00012699057697318494,
      "loss_iou": 0.32421875,
      "loss_num": 0.0123291015625,
      "loss_xval": 0.7109375,
      "num_input_tokens_seen": 412002580,
      "step": 7353
    },
    {
      "epoch": 16.378619153674833,
      "grad_norm": 16.590665817260742,
      "learning_rate": 1e-06,
      "loss": 0.4343,
      "num_input_tokens_seen": 412059804,
      "step": 7354
    },
    {
      "epoch": 16.378619153674833,
      "loss": 0.41806820034980774,
      "loss_ce": 9.94568836176768e-05,
      "loss_iou": 0.1796875,
      "loss_num": 0.01165771484375,
      "loss_xval": 0.41796875,
      "num_input_tokens_seen": 412059804,
      "step": 7354
    },
    {
      "epoch": 16.380846325167038,
      "grad_norm": 14.91476821899414,
      "learning_rate": 1e-06,
      "loss": 0.389,
      "num_input_tokens_seen": 412117424,
      "step": 7355
    },
    {
      "epoch": 16.380846325167038,
      "loss": 0.429162859916687,
      "loss_ce": 8.569219789933413e-05,
      "loss_iou": 0.17578125,
      "loss_num": 0.01531982421875,
      "loss_xval": 0.4296875,
      "num_input_tokens_seen": 412117424,
      "step": 7355
    },
    {
      "epoch": 16.383073496659243,
      "grad_norm": 21.37236976623535,
      "learning_rate": 1e-06,
      "loss": 0.3955,
      "num_input_tokens_seen": 412175316,
      "step": 7356
    },
    {
      "epoch": 16.383073496659243,
      "loss": 0.471897691488266,
      "loss_ce": 9.593518188921735e-05,
      "loss_iou": 0.2119140625,
      "loss_num": 0.009521484375,
      "loss_xval": 0.47265625,
      "num_input_tokens_seen": 412175316,
      "step": 7356
    },
    {
      "epoch": 16.385300668151448,
      "grad_norm": 23.095491409301758,
      "learning_rate": 1e-06,
      "loss": 0.5033,
      "num_input_tokens_seen": 412232752,
      "step": 7357
    },
    {
      "epoch": 16.385300668151448,
      "loss": 0.42098525166511536,
      "loss_ce": 8.681518374942243e-05,
      "loss_iou": 0.1787109375,
      "loss_num": 0.0125732421875,
      "loss_xval": 0.421875,
      "num_input_tokens_seen": 412232752,
      "step": 7357
    },
    {
      "epoch": 16.387527839643653,
      "grad_norm": 73.42839813232422,
      "learning_rate": 1e-06,
      "loss": 0.6915,
      "num_input_tokens_seen": 412284664,
      "step": 7358
    },
    {
      "epoch": 16.387527839643653,
      "loss": 0.7482743859291077,
      "loss_ce": 0.00010545070108491927,
      "loss_iou": 0.318359375,
      "loss_num": 0.02197265625,
      "loss_xval": 0.75,
      "num_input_tokens_seen": 412284664,
      "step": 7358
    },
    {
      "epoch": 16.389755011135858,
      "grad_norm": 24.8407039642334,
      "learning_rate": 1e-06,
      "loss": 0.5937,
      "num_input_tokens_seen": 412342800,
      "step": 7359
    },
    {
      "epoch": 16.389755011135858,
      "loss": 0.5182784795761108,
      "loss_ce": 9.002141450764611e-05,
      "loss_iou": 0.2392578125,
      "loss_num": 0.00799560546875,
      "loss_xval": 0.51953125,
      "num_input_tokens_seen": 412342800,
      "step": 7359
    },
    {
      "epoch": 16.391982182628063,
      "grad_norm": 19.556819915771484,
      "learning_rate": 1e-06,
      "loss": 0.4151,
      "num_input_tokens_seen": 412397900,
      "step": 7360
    },
    {
      "epoch": 16.391982182628063,
      "loss": 0.3638564348220825,
      "loss_ce": 8.691055700182915e-05,
      "loss_iou": 0.15625,
      "loss_num": 0.010498046875,
      "loss_xval": 0.36328125,
      "num_input_tokens_seen": 412397900,
      "step": 7360
    },
    {
      "epoch": 16.394209354120267,
      "grad_norm": 27.42656707763672,
      "learning_rate": 1e-06,
      "loss": 0.4955,
      "num_input_tokens_seen": 412454180,
      "step": 7361
    },
    {
      "epoch": 16.394209354120267,
      "loss": 0.5466644763946533,
      "loss_ce": 9.469907672610134e-05,
      "loss_iou": 0.2421875,
      "loss_num": 0.01226806640625,
      "loss_xval": 0.546875,
      "num_input_tokens_seen": 412454180,
      "step": 7361
    },
    {
      "epoch": 16.396436525612472,
      "grad_norm": 12.27857780456543,
      "learning_rate": 1e-06,
      "loss": 0.3187,
      "num_input_tokens_seen": 412511772,
      "step": 7362
    },
    {
      "epoch": 16.396436525612472,
      "loss": 0.31820547580718994,
      "loss_ce": 9.021394362207502e-05,
      "loss_iou": 0.13671875,
      "loss_num": 0.00885009765625,
      "loss_xval": 0.318359375,
      "num_input_tokens_seen": 412511772,
      "step": 7362
    },
    {
      "epoch": 16.398663697104677,
      "grad_norm": 15.365641593933105,
      "learning_rate": 1e-06,
      "loss": 0.4709,
      "num_input_tokens_seen": 412565312,
      "step": 7363
    },
    {
      "epoch": 16.398663697104677,
      "loss": 0.44320639967918396,
      "loss_ce": 9.116313594859093e-05,
      "loss_iou": 0.1787109375,
      "loss_num": 0.0172119140625,
      "loss_xval": 0.443359375,
      "num_input_tokens_seen": 412565312,
      "step": 7363
    },
    {
      "epoch": 16.400890868596882,
      "grad_norm": 16.197721481323242,
      "learning_rate": 1e-06,
      "loss": 0.4165,
      "num_input_tokens_seen": 412619724,
      "step": 7364
    },
    {
      "epoch": 16.400890868596882,
      "loss": 0.3934442400932312,
      "loss_ce": 0.00013366495841182768,
      "loss_iou": 0.177734375,
      "loss_num": 0.0074462890625,
      "loss_xval": 0.392578125,
      "num_input_tokens_seen": 412619724,
      "step": 7364
    },
    {
      "epoch": 16.403118040089087,
      "grad_norm": 19.257944107055664,
      "learning_rate": 1e-06,
      "loss": 0.4529,
      "num_input_tokens_seen": 412672252,
      "step": 7365
    },
    {
      "epoch": 16.403118040089087,
      "loss": 0.3346201777458191,
      "loss_ce": 8.649235678603873e-05,
      "loss_iou": 0.150390625,
      "loss_num": 0.00689697265625,
      "loss_xval": 0.333984375,
      "num_input_tokens_seen": 412672252,
      "step": 7365
    },
    {
      "epoch": 16.40534521158129,
      "grad_norm": 35.989013671875,
      "learning_rate": 1e-06,
      "loss": 0.4365,
      "num_input_tokens_seen": 412727280,
      "step": 7366
    },
    {
      "epoch": 16.40534521158129,
      "loss": 0.46420300006866455,
      "loss_ce": 9.16688732104376e-05,
      "loss_iou": 0.197265625,
      "loss_num": 0.0137939453125,
      "loss_xval": 0.46484375,
      "num_input_tokens_seen": 412727280,
      "step": 7366
    },
    {
      "epoch": 16.407572383073497,
      "grad_norm": 19.286439895629883,
      "learning_rate": 1e-06,
      "loss": 0.4871,
      "num_input_tokens_seen": 412782212,
      "step": 7367
    },
    {
      "epoch": 16.407572383073497,
      "loss": 0.4057268500328064,
      "loss_ce": 8.719813922652975e-05,
      "loss_iou": 0.189453125,
      "loss_num": 0.00531005859375,
      "loss_xval": 0.40625,
      "num_input_tokens_seen": 412782212,
      "step": 7367
    },
    {
      "epoch": 16.4097995545657,
      "grad_norm": 12.866009712219238,
      "learning_rate": 1e-06,
      "loss": 0.473,
      "num_input_tokens_seen": 412836960,
      "step": 7368
    },
    {
      "epoch": 16.4097995545657,
      "loss": 0.5923727750778198,
      "loss_ce": 8.767165127210319e-05,
      "loss_iou": 0.24609375,
      "loss_num": 0.02001953125,
      "loss_xval": 0.59375,
      "num_input_tokens_seen": 412836960,
      "step": 7368
    },
    {
      "epoch": 16.412026726057906,
      "grad_norm": 18.949840545654297,
      "learning_rate": 1e-06,
      "loss": 0.4169,
      "num_input_tokens_seen": 412894060,
      "step": 7369
    },
    {
      "epoch": 16.412026726057906,
      "loss": 0.5453881025314331,
      "loss_ce": 0.00022210404858924448,
      "loss_iou": 0.2373046875,
      "loss_num": 0.01397705078125,
      "loss_xval": 0.546875,
      "num_input_tokens_seen": 412894060,
      "step": 7369
    },
    {
      "epoch": 16.41425389755011,
      "grad_norm": 16.93563461303711,
      "learning_rate": 1e-06,
      "loss": 0.4276,
      "num_input_tokens_seen": 412950112,
      "step": 7370
    },
    {
      "epoch": 16.41425389755011,
      "loss": 0.383827805519104,
      "loss_ce": 0.00013031261914875358,
      "loss_iou": 0.1669921875,
      "loss_num": 0.01007080078125,
      "loss_xval": 0.3828125,
      "num_input_tokens_seen": 412950112,
      "step": 7370
    },
    {
      "epoch": 16.416481069042316,
      "grad_norm": 20.606613159179688,
      "learning_rate": 1e-06,
      "loss": 0.4678,
      "num_input_tokens_seen": 413005572,
      "step": 7371
    },
    {
      "epoch": 16.416481069042316,
      "loss": 0.5181921720504761,
      "loss_ce": 0.00012574487482197583,
      "loss_iou": 0.22265625,
      "loss_num": 0.0142822265625,
      "loss_xval": 0.51953125,
      "num_input_tokens_seen": 413005572,
      "step": 7371
    },
    {
      "epoch": 16.41870824053452,
      "grad_norm": 24.132949829101562,
      "learning_rate": 1e-06,
      "loss": 0.5933,
      "num_input_tokens_seen": 413061068,
      "step": 7372
    },
    {
      "epoch": 16.41870824053452,
      "loss": 0.5776008367538452,
      "loss_ce": 8.622092718724161e-05,
      "loss_iou": 0.263671875,
      "loss_num": 0.01007080078125,
      "loss_xval": 0.578125,
      "num_input_tokens_seen": 413061068,
      "step": 7372
    },
    {
      "epoch": 16.420935412026726,
      "grad_norm": 15.339935302734375,
      "learning_rate": 1e-06,
      "loss": 0.5645,
      "num_input_tokens_seen": 413118180,
      "step": 7373
    },
    {
      "epoch": 16.420935412026726,
      "loss": 0.5817098617553711,
      "loss_ce": 0.0001058535126503557,
      "loss_iou": 0.25390625,
      "loss_num": 0.014892578125,
      "loss_xval": 0.58203125,
      "num_input_tokens_seen": 413118180,
      "step": 7373
    },
    {
      "epoch": 16.42316258351893,
      "grad_norm": 17.126455307006836,
      "learning_rate": 1e-06,
      "loss": 0.4166,
      "num_input_tokens_seen": 413176160,
      "step": 7374
    },
    {
      "epoch": 16.42316258351893,
      "loss": 0.3593493700027466,
      "loss_ce": 9.64459904935211e-05,
      "loss_iou": 0.158203125,
      "loss_num": 0.008544921875,
      "loss_xval": 0.359375,
      "num_input_tokens_seen": 413176160,
      "step": 7374
    },
    {
      "epoch": 16.425389755011135,
      "grad_norm": 13.592398643493652,
      "learning_rate": 1e-06,
      "loss": 0.2656,
      "num_input_tokens_seen": 413233164,
      "step": 7375
    },
    {
      "epoch": 16.425389755011135,
      "loss": 0.23620259761810303,
      "loss_ce": 8.810235158307478e-05,
      "loss_iou": 0.10498046875,
      "loss_num": 0.00506591796875,
      "loss_xval": 0.236328125,
      "num_input_tokens_seen": 413233164,
      "step": 7375
    },
    {
      "epoch": 16.42761692650334,
      "grad_norm": 13.771684646606445,
      "learning_rate": 1e-06,
      "loss": 0.4194,
      "num_input_tokens_seen": 413290152,
      "step": 7376
    },
    {
      "epoch": 16.42761692650334,
      "loss": 0.410991370677948,
      "loss_ce": 0.00010270239727105945,
      "loss_iou": 0.171875,
      "loss_num": 0.013427734375,
      "loss_xval": 0.41015625,
      "num_input_tokens_seen": 413290152,
      "step": 7376
    },
    {
      "epoch": 16.429844097995545,
      "grad_norm": 12.945503234863281,
      "learning_rate": 1e-06,
      "loss": 0.2917,
      "num_input_tokens_seen": 413346308,
      "step": 7377
    },
    {
      "epoch": 16.429844097995545,
      "loss": 0.2603330910205841,
      "loss_ce": 7.917418406577781e-05,
      "loss_iou": 0.11962890625,
      "loss_num": 0.004119873046875,
      "loss_xval": 0.259765625,
      "num_input_tokens_seen": 413346308,
      "step": 7377
    },
    {
      "epoch": 16.43207126948775,
      "grad_norm": 31.2889404296875,
      "learning_rate": 1e-06,
      "loss": 0.5393,
      "num_input_tokens_seen": 413400924,
      "step": 7378
    },
    {
      "epoch": 16.43207126948775,
      "loss": 0.45716169476509094,
      "loss_ce": 9.228027192875743e-05,
      "loss_iou": 0.208984375,
      "loss_num": 0.0078125,
      "loss_xval": 0.45703125,
      "num_input_tokens_seen": 413400924,
      "step": 7378
    },
    {
      "epoch": 16.434298440979955,
      "grad_norm": 18.560880661010742,
      "learning_rate": 1e-06,
      "loss": 0.4677,
      "num_input_tokens_seen": 413456024,
      "step": 7379
    },
    {
      "epoch": 16.434298440979955,
      "loss": 0.580654501914978,
      "loss_ce": 8.805579273030162e-05,
      "loss_iou": 0.2451171875,
      "loss_num": 0.01806640625,
      "loss_xval": 0.58203125,
      "num_input_tokens_seen": 413456024,
      "step": 7379
    },
    {
      "epoch": 16.43652561247216,
      "grad_norm": 16.95804786682129,
      "learning_rate": 1e-06,
      "loss": 0.6411,
      "num_input_tokens_seen": 413513556,
      "step": 7380
    },
    {
      "epoch": 16.43652561247216,
      "loss": 0.973118245601654,
      "loss_ce": 9.582037455402315e-05,
      "loss_iou": 0.37890625,
      "loss_num": 0.04296875,
      "loss_xval": 0.97265625,
      "num_input_tokens_seen": 413513556,
      "step": 7380
    },
    {
      "epoch": 16.438752783964365,
      "grad_norm": 17.053752899169922,
      "learning_rate": 1e-06,
      "loss": 0.4378,
      "num_input_tokens_seen": 413569980,
      "step": 7381
    },
    {
      "epoch": 16.438752783964365,
      "loss": 0.42026323080062866,
      "loss_ce": 9.722512913867831e-05,
      "loss_iou": 0.1796875,
      "loss_num": 0.012451171875,
      "loss_xval": 0.419921875,
      "num_input_tokens_seen": 413569980,
      "step": 7381
    },
    {
      "epoch": 16.44097995545657,
      "grad_norm": 18.919458389282227,
      "learning_rate": 1e-06,
      "loss": 0.4147,
      "num_input_tokens_seen": 413627584,
      "step": 7382
    },
    {
      "epoch": 16.44097995545657,
      "loss": 0.27885180711746216,
      "loss_ce": 0.00010426974768051878,
      "loss_iou": 0.11865234375,
      "loss_num": 0.0081787109375,
      "loss_xval": 0.279296875,
      "num_input_tokens_seen": 413627584,
      "step": 7382
    },
    {
      "epoch": 16.443207126948774,
      "grad_norm": 14.903229713439941,
      "learning_rate": 1e-06,
      "loss": 0.3204,
      "num_input_tokens_seen": 413684008,
      "step": 7383
    },
    {
      "epoch": 16.443207126948774,
      "loss": 0.30172598361968994,
      "loss_ce": 9.026764746522531e-05,
      "loss_iou": 0.1318359375,
      "loss_num": 0.007720947265625,
      "loss_xval": 0.30078125,
      "num_input_tokens_seen": 413684008,
      "step": 7383
    },
    {
      "epoch": 16.44543429844098,
      "grad_norm": 20.07790184020996,
      "learning_rate": 1e-06,
      "loss": 0.4495,
      "num_input_tokens_seen": 413740804,
      "step": 7384
    },
    {
      "epoch": 16.44543429844098,
      "loss": 0.4181979298591614,
      "loss_ce": 0.00010710630158428103,
      "loss_iou": 0.1953125,
      "loss_num": 0.00555419921875,
      "loss_xval": 0.41796875,
      "num_input_tokens_seen": 413740804,
      "step": 7384
    },
    {
      "epoch": 16.447661469933184,
      "grad_norm": 26.779462814331055,
      "learning_rate": 1e-06,
      "loss": 0.4354,
      "num_input_tokens_seen": 413795972,
      "step": 7385
    },
    {
      "epoch": 16.447661469933184,
      "loss": 0.5660018920898438,
      "loss_ce": 8.394767064601183e-05,
      "loss_iou": 0.251953125,
      "loss_num": 0.01214599609375,
      "loss_xval": 0.56640625,
      "num_input_tokens_seen": 413795972,
      "step": 7385
    },
    {
      "epoch": 16.44988864142539,
      "grad_norm": 33.413516998291016,
      "learning_rate": 1e-06,
      "loss": 0.3646,
      "num_input_tokens_seen": 413850008,
      "step": 7386
    },
    {
      "epoch": 16.44988864142539,
      "loss": 0.35230499505996704,
      "loss_ce": 0.00019320733554195613,
      "loss_iou": 0.16015625,
      "loss_num": 0.006378173828125,
      "loss_xval": 0.3515625,
      "num_input_tokens_seen": 413850008,
      "step": 7386
    },
    {
      "epoch": 16.452115812917594,
      "grad_norm": 16.137666702270508,
      "learning_rate": 1e-06,
      "loss": 0.3793,
      "num_input_tokens_seen": 413908860,
      "step": 7387
    },
    {
      "epoch": 16.452115812917594,
      "loss": 0.37570327520370483,
      "loss_ce": 9.29224188439548e-05,
      "loss_iou": 0.15625,
      "loss_num": 0.012451171875,
      "loss_xval": 0.375,
      "num_input_tokens_seen": 413908860,
      "step": 7387
    },
    {
      "epoch": 16.4543429844098,
      "grad_norm": 27.6522274017334,
      "learning_rate": 1e-06,
      "loss": 0.3894,
      "num_input_tokens_seen": 413964468,
      "step": 7388
    },
    {
      "epoch": 16.4543429844098,
      "loss": 0.36471468210220337,
      "loss_ce": 9.06552595552057e-05,
      "loss_iou": 0.16796875,
      "loss_num": 0.005859375,
      "loss_xval": 0.365234375,
      "num_input_tokens_seen": 413964468,
      "step": 7388
    },
    {
      "epoch": 16.456570155902003,
      "grad_norm": 19.11590003967285,
      "learning_rate": 1e-06,
      "loss": 0.4466,
      "num_input_tokens_seen": 414020268,
      "step": 7389
    },
    {
      "epoch": 16.456570155902003,
      "loss": 0.39550697803497314,
      "loss_ce": 0.00012124592467444018,
      "loss_iou": 0.16796875,
      "loss_num": 0.01220703125,
      "loss_xval": 0.39453125,
      "num_input_tokens_seen": 414020268,
      "step": 7389
    },
    {
      "epoch": 16.45879732739421,
      "grad_norm": 12.807024002075195,
      "learning_rate": 1e-06,
      "loss": 0.3725,
      "num_input_tokens_seen": 414079048,
      "step": 7390
    },
    {
      "epoch": 16.45879732739421,
      "loss": 0.3609822988510132,
      "loss_ce": 8.14046070445329e-05,
      "loss_iou": 0.1513671875,
      "loss_num": 0.01153564453125,
      "loss_xval": 0.361328125,
      "num_input_tokens_seen": 414079048,
      "step": 7390
    },
    {
      "epoch": 16.461024498886413,
      "grad_norm": 15.505331039428711,
      "learning_rate": 1e-06,
      "loss": 0.3684,
      "num_input_tokens_seen": 414134828,
      "step": 7391
    },
    {
      "epoch": 16.461024498886413,
      "loss": 0.37520790100097656,
      "loss_ce": 8.581792644690722e-05,
      "loss_iou": 0.1611328125,
      "loss_num": 0.01031494140625,
      "loss_xval": 0.375,
      "num_input_tokens_seen": 414134828,
      "step": 7391
    },
    {
      "epoch": 16.463251670378618,
      "grad_norm": 52.602256774902344,
      "learning_rate": 1e-06,
      "loss": 0.4731,
      "num_input_tokens_seen": 414188716,
      "step": 7392
    },
    {
      "epoch": 16.463251670378618,
      "loss": 0.5707922577857971,
      "loss_ce": 8.304757648147643e-05,
      "loss_iou": 0.2333984375,
      "loss_num": 0.0208740234375,
      "loss_xval": 0.5703125,
      "num_input_tokens_seen": 414188716,
      "step": 7392
    },
    {
      "epoch": 16.465478841870823,
      "grad_norm": 17.48988151550293,
      "learning_rate": 1e-06,
      "loss": 0.5544,
      "num_input_tokens_seen": 414247204,
      "step": 7393
    },
    {
      "epoch": 16.465478841870823,
      "loss": 0.6307286024093628,
      "loss_ce": 0.00011338148033246398,
      "loss_iou": 0.263671875,
      "loss_num": 0.0205078125,
      "loss_xval": 0.62890625,
      "num_input_tokens_seen": 414247204,
      "step": 7393
    },
    {
      "epoch": 16.467706013363028,
      "grad_norm": 20.05725860595703,
      "learning_rate": 1e-06,
      "loss": 0.4437,
      "num_input_tokens_seen": 414302244,
      "step": 7394
    },
    {
      "epoch": 16.467706013363028,
      "loss": 0.4451483488082886,
      "loss_ce": 7.999275112524629e-05,
      "loss_iou": 0.19921875,
      "loss_num": 0.00933837890625,
      "loss_xval": 0.4453125,
      "num_input_tokens_seen": 414302244,
      "step": 7394
    },
    {
      "epoch": 16.469933184855233,
      "grad_norm": 21.719507217407227,
      "learning_rate": 1e-06,
      "loss": 0.4859,
      "num_input_tokens_seen": 414358148,
      "step": 7395
    },
    {
      "epoch": 16.469933184855233,
      "loss": 0.5905551910400391,
      "loss_ce": 0.00019262763089500368,
      "loss_iou": 0.23046875,
      "loss_num": 0.0257568359375,
      "loss_xval": 0.58984375,
      "num_input_tokens_seen": 414358148,
      "step": 7395
    },
    {
      "epoch": 16.472160356347437,
      "grad_norm": 18.241609573364258,
      "learning_rate": 1e-06,
      "loss": 0.396,
      "num_input_tokens_seen": 414413656,
      "step": 7396
    },
    {
      "epoch": 16.472160356347437,
      "loss": 0.4245295226573944,
      "loss_ce": 9.104832133743912e-05,
      "loss_iou": 0.18359375,
      "loss_num": 0.01153564453125,
      "loss_xval": 0.423828125,
      "num_input_tokens_seen": 414413656,
      "step": 7396
    },
    {
      "epoch": 16.474387527839642,
      "grad_norm": 16.02747344970703,
      "learning_rate": 1e-06,
      "loss": 0.3917,
      "num_input_tokens_seen": 414471236,
      "step": 7397
    },
    {
      "epoch": 16.474387527839642,
      "loss": 0.4211219251155853,
      "loss_ce": 0.00010141961683984846,
      "loss_iou": 0.19140625,
      "loss_num": 0.007781982421875,
      "loss_xval": 0.421875,
      "num_input_tokens_seen": 414471236,
      "step": 7397
    },
    {
      "epoch": 16.476614699331847,
      "grad_norm": 26.210834503173828,
      "learning_rate": 1e-06,
      "loss": 0.3714,
      "num_input_tokens_seen": 414526900,
      "step": 7398
    },
    {
      "epoch": 16.476614699331847,
      "loss": 0.36032408475875854,
      "loss_ce": 9.459605644224212e-05,
      "loss_iou": 0.1630859375,
      "loss_num": 0.0068359375,
      "loss_xval": 0.359375,
      "num_input_tokens_seen": 414526900,
      "step": 7398
    },
    {
      "epoch": 16.478841870824052,
      "grad_norm": 21.00611114501953,
      "learning_rate": 1e-06,
      "loss": 0.3868,
      "num_input_tokens_seen": 414582196,
      "step": 7399
    },
    {
      "epoch": 16.478841870824052,
      "loss": 0.3853411376476288,
      "loss_ce": 8.721851918380708e-05,
      "loss_iou": 0.1513671875,
      "loss_num": 0.0167236328125,
      "loss_xval": 0.384765625,
      "num_input_tokens_seen": 414582196,
      "step": 7399
    },
    {
      "epoch": 16.481069042316257,
      "grad_norm": 17.40221405029297,
      "learning_rate": 1e-06,
      "loss": 0.5933,
      "num_input_tokens_seen": 414638744,
      "step": 7400
    },
    {
      "epoch": 16.481069042316257,
      "loss": 0.7026082277297974,
      "loss_ce": 9.35388816287741e-05,
      "loss_iou": 0.28125,
      "loss_num": 0.02734375,
      "loss_xval": 0.703125,
      "num_input_tokens_seen": 414638744,
      "step": 7400
    },
    {
      "epoch": 16.48329621380846,
      "grad_norm": 17.026206970214844,
      "learning_rate": 1e-06,
      "loss": 0.5015,
      "num_input_tokens_seen": 414695508,
      "step": 7401
    },
    {
      "epoch": 16.48329621380846,
      "loss": 0.5965702533721924,
      "loss_ce": 0.00013468590623233467,
      "loss_iou": 0.2353515625,
      "loss_num": 0.02490234375,
      "loss_xval": 0.59765625,
      "num_input_tokens_seen": 414695508,
      "step": 7401
    },
    {
      "epoch": 16.485523385300667,
      "grad_norm": 20.803653717041016,
      "learning_rate": 1e-06,
      "loss": 0.4458,
      "num_input_tokens_seen": 414750064,
      "step": 7402
    },
    {
      "epoch": 16.485523385300667,
      "loss": 0.4915461540222168,
      "loss_ce": 0.00015212838479783386,
      "loss_iou": 0.216796875,
      "loss_num": 0.01165771484375,
      "loss_xval": 0.4921875,
      "num_input_tokens_seen": 414750064,
      "step": 7402
    },
    {
      "epoch": 16.48775055679287,
      "grad_norm": 18.564687728881836,
      "learning_rate": 1e-06,
      "loss": 0.4463,
      "num_input_tokens_seen": 414807600,
      "step": 7403
    },
    {
      "epoch": 16.48775055679287,
      "loss": 0.4674866199493408,
      "loss_ce": 7.938436465337873e-05,
      "loss_iou": 0.20703125,
      "loss_num": 0.01092529296875,
      "loss_xval": 0.466796875,
      "num_input_tokens_seen": 414807600,
      "step": 7403
    },
    {
      "epoch": 16.489977728285076,
      "grad_norm": 19.705589294433594,
      "learning_rate": 1e-06,
      "loss": 0.4589,
      "num_input_tokens_seen": 414862628,
      "step": 7404
    },
    {
      "epoch": 16.489977728285076,
      "loss": 0.524620771408081,
      "loss_ce": 8.46216207719408e-05,
      "loss_iou": 0.22265625,
      "loss_num": 0.0159912109375,
      "loss_xval": 0.5234375,
      "num_input_tokens_seen": 414862628,
      "step": 7404
    },
    {
      "epoch": 16.49220489977728,
      "grad_norm": 15.944623947143555,
      "learning_rate": 1e-06,
      "loss": 0.2974,
      "num_input_tokens_seen": 414921268,
      "step": 7405
    },
    {
      "epoch": 16.49220489977728,
      "loss": 0.3903467059135437,
      "loss_ce": 8.790481660980731e-05,
      "loss_iou": 0.15625,
      "loss_num": 0.0155029296875,
      "loss_xval": 0.390625,
      "num_input_tokens_seen": 414921268,
      "step": 7405
    },
    {
      "epoch": 16.494432071269486,
      "grad_norm": 16.492647171020508,
      "learning_rate": 1e-06,
      "loss": 0.3803,
      "num_input_tokens_seen": 414979472,
      "step": 7406
    },
    {
      "epoch": 16.494432071269486,
      "loss": 0.46630802750587463,
      "loss_ce": 0.00012149002577643842,
      "loss_iou": 0.212890625,
      "loss_num": 0.00811767578125,
      "loss_xval": 0.466796875,
      "num_input_tokens_seen": 414979472,
      "step": 7406
    },
    {
      "epoch": 16.49665924276169,
      "grad_norm": 19.485671997070312,
      "learning_rate": 1e-06,
      "loss": 0.4414,
      "num_input_tokens_seen": 415036420,
      "step": 7407
    },
    {
      "epoch": 16.49665924276169,
      "loss": 0.3690652847290039,
      "loss_ce": 7.723318412899971e-05,
      "loss_iou": 0.16015625,
      "loss_num": 0.0096435546875,
      "loss_xval": 0.369140625,
      "num_input_tokens_seen": 415036420,
      "step": 7407
    },
    {
      "epoch": 16.498886414253896,
      "grad_norm": 29.99024200439453,
      "learning_rate": 1e-06,
      "loss": 0.4787,
      "num_input_tokens_seen": 415092432,
      "step": 7408
    },
    {
      "epoch": 16.498886414253896,
      "loss": 0.40274959802627563,
      "loss_ce": 0.00010066555842058733,
      "loss_iou": 0.1787109375,
      "loss_num": 0.00909423828125,
      "loss_xval": 0.40234375,
      "num_input_tokens_seen": 415092432,
      "step": 7408
    },
    {
      "epoch": 16.501113585746104,
      "grad_norm": 19.012744903564453,
      "learning_rate": 1e-06,
      "loss": 0.2874,
      "num_input_tokens_seen": 415148876,
      "step": 7409
    },
    {
      "epoch": 16.501113585746104,
      "loss": 0.2875651717185974,
      "loss_ce": 8.95664852578193e-05,
      "loss_iou": 0.1328125,
      "loss_num": 0.00421142578125,
      "loss_xval": 0.287109375,
      "num_input_tokens_seen": 415148876,
      "step": 7409
    },
    {
      "epoch": 16.50334075723831,
      "grad_norm": 14.734273910522461,
      "learning_rate": 1e-06,
      "loss": 0.6548,
      "num_input_tokens_seen": 415204496,
      "step": 7410
    },
    {
      "epoch": 16.50334075723831,
      "loss": 0.7657289505004883,
      "loss_ce": 0.00010389051749370992,
      "loss_iou": 0.326171875,
      "loss_num": 0.0228271484375,
      "loss_xval": 0.765625,
      "num_input_tokens_seen": 415204496,
      "step": 7410
    },
    {
      "epoch": 16.505567928730514,
      "grad_norm": 16.262977600097656,
      "learning_rate": 1e-06,
      "loss": 0.4765,
      "num_input_tokens_seen": 415260244,
      "step": 7411
    },
    {
      "epoch": 16.505567928730514,
      "loss": 0.28719890117645264,
      "loss_ce": 8.953356882557273e-05,
      "loss_iou": 0.126953125,
      "loss_num": 0.006439208984375,
      "loss_xval": 0.287109375,
      "num_input_tokens_seen": 415260244,
      "step": 7411
    },
    {
      "epoch": 16.50779510022272,
      "grad_norm": 21.31691551208496,
      "learning_rate": 1e-06,
      "loss": 0.3835,
      "num_input_tokens_seen": 415314104,
      "step": 7412
    },
    {
      "epoch": 16.50779510022272,
      "loss": 0.3603422939777374,
      "loss_ce": 0.00011278928286628798,
      "loss_iou": 0.1630859375,
      "loss_num": 0.00689697265625,
      "loss_xval": 0.359375,
      "num_input_tokens_seen": 415314104,
      "step": 7412
    },
    {
      "epoch": 16.510022271714924,
      "grad_norm": 20.083385467529297,
      "learning_rate": 1e-06,
      "loss": 0.4469,
      "num_input_tokens_seen": 415370528,
      "step": 7413
    },
    {
      "epoch": 16.510022271714924,
      "loss": 0.42683807015419006,
      "loss_ce": 8.025132410693914e-05,
      "loss_iou": 0.197265625,
      "loss_num": 0.00653076171875,
      "loss_xval": 0.42578125,
      "num_input_tokens_seen": 415370528,
      "step": 7413
    },
    {
      "epoch": 16.51224944320713,
      "grad_norm": 22.564624786376953,
      "learning_rate": 1e-06,
      "loss": 0.388,
      "num_input_tokens_seen": 415430264,
      "step": 7414
    },
    {
      "epoch": 16.51224944320713,
      "loss": 0.38656115531921387,
      "loss_ce": 8.655471901874989e-05,
      "loss_iou": 0.17578125,
      "loss_num": 0.007110595703125,
      "loss_xval": 0.38671875,
      "num_input_tokens_seen": 415430264,
      "step": 7414
    },
    {
      "epoch": 16.514476614699333,
      "grad_norm": 17.342039108276367,
      "learning_rate": 1e-06,
      "loss": 0.4182,
      "num_input_tokens_seen": 415486144,
      "step": 7415
    },
    {
      "epoch": 16.514476614699333,
      "loss": 0.3627013564109802,
      "loss_ce": 9.151514677796513e-05,
      "loss_iou": 0.1533203125,
      "loss_num": 0.01123046875,
      "loss_xval": 0.36328125,
      "num_input_tokens_seen": 415486144,
      "step": 7415
    },
    {
      "epoch": 16.51670378619154,
      "grad_norm": 34.493228912353516,
      "learning_rate": 1e-06,
      "loss": 0.4162,
      "num_input_tokens_seen": 415543736,
      "step": 7416
    },
    {
      "epoch": 16.51670378619154,
      "loss": 0.42243343591690063,
      "loss_ce": 0.00031430544913746417,
      "loss_iou": 0.185546875,
      "loss_num": 0.010009765625,
      "loss_xval": 0.421875,
      "num_input_tokens_seen": 415543736,
      "step": 7416
    },
    {
      "epoch": 16.518930957683743,
      "grad_norm": 26.808746337890625,
      "learning_rate": 1e-06,
      "loss": 0.4272,
      "num_input_tokens_seen": 415601828,
      "step": 7417
    },
    {
      "epoch": 16.518930957683743,
      "loss": 0.5273232460021973,
      "loss_ce": 0.00010160428064409643,
      "loss_iou": 0.2041015625,
      "loss_num": 0.023681640625,
      "loss_xval": 0.52734375,
      "num_input_tokens_seen": 415601828,
      "step": 7417
    },
    {
      "epoch": 16.521158129175948,
      "grad_norm": 9.941266059875488,
      "learning_rate": 1e-06,
      "loss": 0.3009,
      "num_input_tokens_seen": 415660184,
      "step": 7418
    },
    {
      "epoch": 16.521158129175948,
      "loss": 0.23453950881958008,
      "loss_ce": 7.294982788152993e-05,
      "loss_iou": 0.09912109375,
      "loss_num": 0.00732421875,
      "loss_xval": 0.234375,
      "num_input_tokens_seen": 415660184,
      "step": 7418
    },
    {
      "epoch": 16.523385300668153,
      "grad_norm": 26.528013229370117,
      "learning_rate": 1e-06,
      "loss": 0.6055,
      "num_input_tokens_seen": 415715484,
      "step": 7419
    },
    {
      "epoch": 16.523385300668153,
      "loss": 0.8102438449859619,
      "loss_ce": 0.000429451436502859,
      "loss_iou": 0.3125,
      "loss_num": 0.036865234375,
      "loss_xval": 0.80859375,
      "num_input_tokens_seen": 415715484,
      "step": 7419
    },
    {
      "epoch": 16.525612472160358,
      "grad_norm": 19.867149353027344,
      "learning_rate": 1e-06,
      "loss": 0.3117,
      "num_input_tokens_seen": 415770232,
      "step": 7420
    },
    {
      "epoch": 16.525612472160358,
      "loss": 0.26326221227645874,
      "loss_ce": 7.859835750423372e-05,
      "loss_iou": 0.12158203125,
      "loss_num": 0.0040283203125,
      "loss_xval": 0.263671875,
      "num_input_tokens_seen": 415770232,
      "step": 7420
    },
    {
      "epoch": 16.527839643652563,
      "grad_norm": 19.17835807800293,
      "learning_rate": 1e-06,
      "loss": 0.4112,
      "num_input_tokens_seen": 415825868,
      "step": 7421
    },
    {
      "epoch": 16.527839643652563,
      "loss": 0.43073534965515137,
      "loss_ce": 7.126775744836777e-05,
      "loss_iou": 0.1787109375,
      "loss_num": 0.01458740234375,
      "loss_xval": 0.4296875,
      "num_input_tokens_seen": 415825868,
      "step": 7421
    },
    {
      "epoch": 16.530066815144767,
      "grad_norm": 17.873729705810547,
      "learning_rate": 1e-06,
      "loss": 0.3862,
      "num_input_tokens_seen": 415882708,
      "step": 7422
    },
    {
      "epoch": 16.530066815144767,
      "loss": 0.26814356446266174,
      "loss_ce": 7.717055268585682e-05,
      "loss_iou": 0.119140625,
      "loss_num": 0.006072998046875,
      "loss_xval": 0.267578125,
      "num_input_tokens_seen": 415882708,
      "step": 7422
    },
    {
      "epoch": 16.532293986636972,
      "grad_norm": 16.114519119262695,
      "learning_rate": 1e-06,
      "loss": 0.5852,
      "num_input_tokens_seen": 415938224,
      "step": 7423
    },
    {
      "epoch": 16.532293986636972,
      "loss": 0.6861782073974609,
      "loss_ce": 0.000265107664745301,
      "loss_iou": 0.29296875,
      "loss_num": 0.0196533203125,
      "loss_xval": 0.6875,
      "num_input_tokens_seen": 415938224,
      "step": 7423
    },
    {
      "epoch": 16.534521158129177,
      "grad_norm": 21.42132568359375,
      "learning_rate": 1e-06,
      "loss": 0.3615,
      "num_input_tokens_seen": 415992012,
      "step": 7424
    },
    {
      "epoch": 16.534521158129177,
      "loss": 0.25483426451683044,
      "loss_ce": 7.351664680754766e-05,
      "loss_iou": 0.10546875,
      "loss_num": 0.00885009765625,
      "loss_xval": 0.25390625,
      "num_input_tokens_seen": 415992012,
      "step": 7424
    },
    {
      "epoch": 16.536748329621382,
      "grad_norm": 28.71778678894043,
      "learning_rate": 1e-06,
      "loss": 0.556,
      "num_input_tokens_seen": 416048204,
      "step": 7425
    },
    {
      "epoch": 16.536748329621382,
      "loss": 0.48399072885513306,
      "loss_ce": 0.00010403442865936086,
      "loss_iou": 0.2255859375,
      "loss_num": 0.00640869140625,
      "loss_xval": 0.484375,
      "num_input_tokens_seen": 416048204,
      "step": 7425
    },
    {
      "epoch": 16.538975501113587,
      "grad_norm": 19.75933265686035,
      "learning_rate": 1e-06,
      "loss": 0.5605,
      "num_input_tokens_seen": 416101644,
      "step": 7426
    },
    {
      "epoch": 16.538975501113587,
      "loss": 0.74393230676651,
      "loss_ce": 0.00018841384735424072,
      "loss_iou": 0.294921875,
      "loss_num": 0.0306396484375,
      "loss_xval": 0.7421875,
      "num_input_tokens_seen": 416101644,
      "step": 7426
    },
    {
      "epoch": 16.54120267260579,
      "grad_norm": 22.49873924255371,
      "learning_rate": 1e-06,
      "loss": 0.3687,
      "num_input_tokens_seen": 416160076,
      "step": 7427
    },
    {
      "epoch": 16.54120267260579,
      "loss": 0.33309027552604675,
      "loss_ce": 8.246669312939048e-05,
      "loss_iou": 0.150390625,
      "loss_num": 0.006500244140625,
      "loss_xval": 0.33203125,
      "num_input_tokens_seen": 416160076,
      "step": 7427
    },
    {
      "epoch": 16.543429844097997,
      "grad_norm": 13.900813102722168,
      "learning_rate": 1e-06,
      "loss": 0.4714,
      "num_input_tokens_seen": 416217540,
      "step": 7428
    },
    {
      "epoch": 16.543429844097997,
      "loss": 0.5082305669784546,
      "loss_ce": 0.00017388846026733518,
      "loss_iou": 0.2021484375,
      "loss_num": 0.0206298828125,
      "loss_xval": 0.5078125,
      "num_input_tokens_seen": 416217540,
      "step": 7428
    },
    {
      "epoch": 16.5456570155902,
      "grad_norm": 21.1329402923584,
      "learning_rate": 1e-06,
      "loss": 0.4258,
      "num_input_tokens_seen": 416274740,
      "step": 7429
    },
    {
      "epoch": 16.5456570155902,
      "loss": 0.3279839754104614,
      "loss_ce": 0.00010313085658708587,
      "loss_iou": 0.1318359375,
      "loss_num": 0.01300048828125,
      "loss_xval": 0.328125,
      "num_input_tokens_seen": 416274740,
      "step": 7429
    },
    {
      "epoch": 16.547884187082406,
      "grad_norm": 20.564817428588867,
      "learning_rate": 1e-06,
      "loss": 0.3677,
      "num_input_tokens_seen": 416330560,
      "step": 7430
    },
    {
      "epoch": 16.547884187082406,
      "loss": 0.31831276416778564,
      "loss_ce": 7.547263521701097e-05,
      "loss_iou": 0.1396484375,
      "loss_num": 0.00762939453125,
      "loss_xval": 0.318359375,
      "num_input_tokens_seen": 416330560,
      "step": 7430
    },
    {
      "epoch": 16.55011135857461,
      "grad_norm": 19.32866859436035,
      "learning_rate": 1e-06,
      "loss": 0.3478,
      "num_input_tokens_seen": 416384232,
      "step": 7431
    },
    {
      "epoch": 16.55011135857461,
      "loss": 0.2748854160308838,
      "loss_ce": 0.00010513007873669267,
      "loss_iou": 0.1240234375,
      "loss_num": 0.00537109375,
      "loss_xval": 0.275390625,
      "num_input_tokens_seen": 416384232,
      "step": 7431
    },
    {
      "epoch": 16.552338530066816,
      "grad_norm": 16.105804443359375,
      "learning_rate": 1e-06,
      "loss": 0.4259,
      "num_input_tokens_seen": 416442036,
      "step": 7432
    },
    {
      "epoch": 16.552338530066816,
      "loss": 0.5960502624511719,
      "loss_ce": 0.00010297551489202306,
      "loss_iou": 0.248046875,
      "loss_num": 0.0198974609375,
      "loss_xval": 0.59765625,
      "num_input_tokens_seen": 416442036,
      "step": 7432
    },
    {
      "epoch": 16.55456570155902,
      "grad_norm": 26.98290252685547,
      "learning_rate": 1e-06,
      "loss": 0.4073,
      "num_input_tokens_seen": 416495444,
      "step": 7433
    },
    {
      "epoch": 16.55456570155902,
      "loss": 0.2754194438457489,
      "loss_ce": 8.982772123999894e-05,
      "loss_iou": 0.12109375,
      "loss_num": 0.006500244140625,
      "loss_xval": 0.275390625,
      "num_input_tokens_seen": 416495444,
      "step": 7433
    },
    {
      "epoch": 16.556792873051226,
      "grad_norm": 23.57514762878418,
      "learning_rate": 1e-06,
      "loss": 0.5139,
      "num_input_tokens_seen": 416550332,
      "step": 7434
    },
    {
      "epoch": 16.556792873051226,
      "loss": 0.44662949442863464,
      "loss_ce": 9.629656415199861e-05,
      "loss_iou": 0.1875,
      "loss_num": 0.01409912109375,
      "loss_xval": 0.447265625,
      "num_input_tokens_seen": 416550332,
      "step": 7434
    },
    {
      "epoch": 16.55902004454343,
      "grad_norm": 15.292926788330078,
      "learning_rate": 1e-06,
      "loss": 0.5265,
      "num_input_tokens_seen": 416605772,
      "step": 7435
    },
    {
      "epoch": 16.55902004454343,
      "loss": 0.6938190460205078,
      "loss_ce": 0.00015451818762812763,
      "loss_iou": 0.259765625,
      "loss_num": 0.034912109375,
      "loss_xval": 0.6953125,
      "num_input_tokens_seen": 416605772,
      "step": 7435
    },
    {
      "epoch": 16.561247216035635,
      "grad_norm": 29.28976058959961,
      "learning_rate": 1e-06,
      "loss": 0.5638,
      "num_input_tokens_seen": 416662900,
      "step": 7436
    },
    {
      "epoch": 16.561247216035635,
      "loss": 0.43191277980804443,
      "loss_ce": 8.905126014724374e-05,
      "loss_iou": 0.1865234375,
      "loss_num": 0.011474609375,
      "loss_xval": 0.431640625,
      "num_input_tokens_seen": 416662900,
      "step": 7436
    },
    {
      "epoch": 16.56347438752784,
      "grad_norm": 17.69676971435547,
      "learning_rate": 1e-06,
      "loss": 0.4027,
      "num_input_tokens_seen": 416718660,
      "step": 7437
    },
    {
      "epoch": 16.56347438752784,
      "loss": 0.368381142616272,
      "loss_ce": 9.499008592683822e-05,
      "loss_iou": 0.1748046875,
      "loss_num": 0.0035552978515625,
      "loss_xval": 0.369140625,
      "num_input_tokens_seen": 416718660,
      "step": 7437
    },
    {
      "epoch": 16.565701559020045,
      "grad_norm": 21.888330459594727,
      "learning_rate": 1e-06,
      "loss": 0.4108,
      "num_input_tokens_seen": 416775236,
      "step": 7438
    },
    {
      "epoch": 16.565701559020045,
      "loss": 0.48281651735305786,
      "loss_ce": 8.945601439336315e-05,
      "loss_iou": 0.2197265625,
      "loss_num": 0.00872802734375,
      "loss_xval": 0.482421875,
      "num_input_tokens_seen": 416775236,
      "step": 7438
    },
    {
      "epoch": 16.56792873051225,
      "grad_norm": 15.91091537475586,
      "learning_rate": 1e-06,
      "loss": 0.4277,
      "num_input_tokens_seen": 416831260,
      "step": 7439
    },
    {
      "epoch": 16.56792873051225,
      "loss": 0.4805620312690735,
      "loss_ce": 9.329438034910709e-05,
      "loss_iou": 0.21875,
      "loss_num": 0.00836181640625,
      "loss_xval": 0.48046875,
      "num_input_tokens_seen": 416831260,
      "step": 7439
    },
    {
      "epoch": 16.570155902004455,
      "grad_norm": 32.6405143737793,
      "learning_rate": 1e-06,
      "loss": 0.4557,
      "num_input_tokens_seen": 416887804,
      "step": 7440
    },
    {
      "epoch": 16.570155902004455,
      "loss": 0.47945916652679443,
      "loss_ce": 8.904725837055594e-05,
      "loss_iou": 0.21875,
      "loss_num": 0.00848388671875,
      "loss_xval": 0.478515625,
      "num_input_tokens_seen": 416887804,
      "step": 7440
    },
    {
      "epoch": 16.57238307349666,
      "grad_norm": 17.98710060119629,
      "learning_rate": 1e-06,
      "loss": 0.5698,
      "num_input_tokens_seen": 416944556,
      "step": 7441
    },
    {
      "epoch": 16.57238307349666,
      "loss": 0.5035039782524109,
      "loss_ce": 8.599211287219077e-05,
      "loss_iou": 0.2021484375,
      "loss_num": 0.0201416015625,
      "loss_xval": 0.50390625,
      "num_input_tokens_seen": 416944556,
      "step": 7441
    },
    {
      "epoch": 16.574610244988865,
      "grad_norm": 17.1207332611084,
      "learning_rate": 1e-06,
      "loss": 0.5436,
      "num_input_tokens_seen": 417000276,
      "step": 7442
    },
    {
      "epoch": 16.574610244988865,
      "loss": 0.4915321469306946,
      "loss_ce": 7.706407632213086e-05,
      "loss_iou": 0.208984375,
      "loss_num": 0.01458740234375,
      "loss_xval": 0.4921875,
      "num_input_tokens_seen": 417000276,
      "step": 7442
    },
    {
      "epoch": 16.57683741648107,
      "grad_norm": 12.052204132080078,
      "learning_rate": 1e-06,
      "loss": 0.3561,
      "num_input_tokens_seen": 417056968,
      "step": 7443
    },
    {
      "epoch": 16.57683741648107,
      "loss": 0.3951103687286377,
      "loss_ce": 9.083513577934355e-05,
      "loss_iou": 0.150390625,
      "loss_num": 0.0186767578125,
      "loss_xval": 0.39453125,
      "num_input_tokens_seen": 417056968,
      "step": 7443
    },
    {
      "epoch": 16.579064587973274,
      "grad_norm": 29.809144973754883,
      "learning_rate": 1e-06,
      "loss": 0.3865,
      "num_input_tokens_seen": 417111816,
      "step": 7444
    },
    {
      "epoch": 16.579064587973274,
      "loss": 0.3363809287548065,
      "loss_ce": 7.720827125012875e-05,
      "loss_iou": 0.1513671875,
      "loss_num": 0.006866455078125,
      "loss_xval": 0.3359375,
      "num_input_tokens_seen": 417111816,
      "step": 7444
    },
    {
      "epoch": 16.58129175946548,
      "grad_norm": 43.40556716918945,
      "learning_rate": 1e-06,
      "loss": 0.3145,
      "num_input_tokens_seen": 417170836,
      "step": 7445
    },
    {
      "epoch": 16.58129175946548,
      "loss": 0.2581978440284729,
      "loss_ce": 8.018691733013839e-05,
      "loss_iou": 0.1123046875,
      "loss_num": 0.00653076171875,
      "loss_xval": 0.2578125,
      "num_input_tokens_seen": 417170836,
      "step": 7445
    },
    {
      "epoch": 16.583518930957684,
      "grad_norm": 19.96676254272461,
      "learning_rate": 1e-06,
      "loss": 0.2797,
      "num_input_tokens_seen": 417226100,
      "step": 7446
    },
    {
      "epoch": 16.583518930957684,
      "loss": 0.307026207447052,
      "loss_ce": 8.040900866035372e-05,
      "loss_iou": 0.13671875,
      "loss_num": 0.006744384765625,
      "loss_xval": 0.306640625,
      "num_input_tokens_seen": 417226100,
      "step": 7446
    },
    {
      "epoch": 16.58574610244989,
      "grad_norm": 18.28860855102539,
      "learning_rate": 1e-06,
      "loss": 0.4555,
      "num_input_tokens_seen": 417282572,
      "step": 7447
    },
    {
      "epoch": 16.58574610244989,
      "loss": 0.434162974357605,
      "loss_ce": 8.092315692920238e-05,
      "loss_iou": 0.185546875,
      "loss_num": 0.01263427734375,
      "loss_xval": 0.43359375,
      "num_input_tokens_seen": 417282572,
      "step": 7447
    },
    {
      "epoch": 16.587973273942094,
      "grad_norm": 19.922115325927734,
      "learning_rate": 1e-06,
      "loss": 0.3285,
      "num_input_tokens_seen": 417336348,
      "step": 7448
    },
    {
      "epoch": 16.587973273942094,
      "loss": 0.2561919093132019,
      "loss_ce": 8.838169014779851e-05,
      "loss_iou": 0.1162109375,
      "loss_num": 0.00482177734375,
      "loss_xval": 0.255859375,
      "num_input_tokens_seen": 417336348,
      "step": 7448
    },
    {
      "epoch": 16.5902004454343,
      "grad_norm": 15.854660034179688,
      "learning_rate": 1e-06,
      "loss": 0.3569,
      "num_input_tokens_seen": 417393976,
      "step": 7449
    },
    {
      "epoch": 16.5902004454343,
      "loss": 0.4261277914047241,
      "loss_ce": 0.00010242144344374537,
      "loss_iou": 0.1962890625,
      "loss_num": 0.006561279296875,
      "loss_xval": 0.42578125,
      "num_input_tokens_seen": 417393976,
      "step": 7449
    },
    {
      "epoch": 16.592427616926503,
      "grad_norm": 15.823163986206055,
      "learning_rate": 1e-06,
      "loss": 0.4953,
      "num_input_tokens_seen": 417453068,
      "step": 7450
    },
    {
      "epoch": 16.592427616926503,
      "loss": 0.43979281187057495,
      "loss_ce": 9.553506970405579e-05,
      "loss_iou": 0.1689453125,
      "loss_num": 0.0205078125,
      "loss_xval": 0.439453125,
      "num_input_tokens_seen": 417453068,
      "step": 7450
    },
    {
      "epoch": 16.59465478841871,
      "grad_norm": 15.928594589233398,
      "learning_rate": 1e-06,
      "loss": 0.5031,
      "num_input_tokens_seen": 417509392,
      "step": 7451
    },
    {
      "epoch": 16.59465478841871,
      "loss": 0.5577445030212402,
      "loss_ce": 0.00012731979950331151,
      "loss_iou": 0.25,
      "loss_num": 0.01123046875,
      "loss_xval": 0.55859375,
      "num_input_tokens_seen": 417509392,
      "step": 7451
    },
    {
      "epoch": 16.596881959910913,
      "grad_norm": 20.19134521484375,
      "learning_rate": 1e-06,
      "loss": 0.3192,
      "num_input_tokens_seen": 417565928,
      "step": 7452
    },
    {
      "epoch": 16.596881959910913,
      "loss": 0.2749907076358795,
      "loss_ce": 8.836391498334706e-05,
      "loss_iou": 0.10107421875,
      "loss_num": 0.01458740234375,
      "loss_xval": 0.275390625,
      "num_input_tokens_seen": 417565928,
      "step": 7452
    },
    {
      "epoch": 16.599109131403118,
      "grad_norm": 28.828327178955078,
      "learning_rate": 1e-06,
      "loss": 0.4603,
      "num_input_tokens_seen": 417621032,
      "step": 7453
    },
    {
      "epoch": 16.599109131403118,
      "loss": 0.46700233221054077,
      "loss_ce": 8.33997328300029e-05,
      "loss_iou": 0.220703125,
      "loss_num": 0.00494384765625,
      "loss_xval": 0.466796875,
      "num_input_tokens_seen": 417621032,
      "step": 7453
    },
    {
      "epoch": 16.601336302895323,
      "grad_norm": 42.66580581665039,
      "learning_rate": 1e-06,
      "loss": 0.5839,
      "num_input_tokens_seen": 417678064,
      "step": 7454
    },
    {
      "epoch": 16.601336302895323,
      "loss": 0.4696599841117859,
      "loss_ce": 0.00042174485861323774,
      "loss_iou": 0.2041015625,
      "loss_num": 0.01214599609375,
      "loss_xval": 0.46875,
      "num_input_tokens_seen": 417678064,
      "step": 7454
    },
    {
      "epoch": 16.603563474387528,
      "grad_norm": 13.862714767456055,
      "learning_rate": 1e-06,
      "loss": 0.2689,
      "num_input_tokens_seen": 417734624,
      "step": 7455
    },
    {
      "epoch": 16.603563474387528,
      "loss": 0.2475118190050125,
      "loss_ce": 7.529581489507109e-05,
      "loss_iou": 0.1064453125,
      "loss_num": 0.006805419921875,
      "loss_xval": 0.2470703125,
      "num_input_tokens_seen": 417734624,
      "step": 7455
    },
    {
      "epoch": 16.605790645879733,
      "grad_norm": 18.104511260986328,
      "learning_rate": 1e-06,
      "loss": 0.3205,
      "num_input_tokens_seen": 417790132,
      "step": 7456
    },
    {
      "epoch": 16.605790645879733,
      "loss": 0.2718583941459656,
      "loss_ce": 0.00012988239177502692,
      "loss_iou": 0.109375,
      "loss_num": 0.0106201171875,
      "loss_xval": 0.271484375,
      "num_input_tokens_seen": 417790132,
      "step": 7456
    },
    {
      "epoch": 16.608017817371937,
      "grad_norm": 15.675362586975098,
      "learning_rate": 1e-06,
      "loss": 0.455,
      "num_input_tokens_seen": 417844852,
      "step": 7457
    },
    {
      "epoch": 16.608017817371937,
      "loss": 0.5134113430976868,
      "loss_ce": 0.00010569683945504948,
      "loss_iou": 0.212890625,
      "loss_num": 0.017578125,
      "loss_xval": 0.51171875,
      "num_input_tokens_seen": 417844852,
      "step": 7457
    },
    {
      "epoch": 16.610244988864142,
      "grad_norm": 18.084177017211914,
      "learning_rate": 1e-06,
      "loss": 0.5179,
      "num_input_tokens_seen": 417901268,
      "step": 7458
    },
    {
      "epoch": 16.610244988864142,
      "loss": 0.5478125810623169,
      "loss_ce": 8.308600808959454e-05,
      "loss_iou": 0.2392578125,
      "loss_num": 0.013916015625,
      "loss_xval": 0.546875,
      "num_input_tokens_seen": 417901268,
      "step": 7458
    },
    {
      "epoch": 16.612472160356347,
      "grad_norm": 15.700066566467285,
      "learning_rate": 1e-06,
      "loss": 0.4414,
      "num_input_tokens_seen": 417957084,
      "step": 7459
    },
    {
      "epoch": 16.612472160356347,
      "loss": 0.4971623420715332,
      "loss_ce": 9.201420471072197e-05,
      "loss_iou": 0.20703125,
      "loss_num": 0.016357421875,
      "loss_xval": 0.49609375,
      "num_input_tokens_seen": 417957084,
      "step": 7459
    },
    {
      "epoch": 16.614699331848552,
      "grad_norm": 19.968868255615234,
      "learning_rate": 1e-06,
      "loss": 0.4322,
      "num_input_tokens_seen": 418013952,
      "step": 7460
    },
    {
      "epoch": 16.614699331848552,
      "loss": 0.4334394335746765,
      "loss_ce": 8.979369886219501e-05,
      "loss_iou": 0.1982421875,
      "loss_num": 0.0074462890625,
      "loss_xval": 0.43359375,
      "num_input_tokens_seen": 418013952,
      "step": 7460
    },
    {
      "epoch": 16.616926503340757,
      "grad_norm": 15.141510009765625,
      "learning_rate": 1e-06,
      "loss": 0.3758,
      "num_input_tokens_seen": 418069596,
      "step": 7461
    },
    {
      "epoch": 16.616926503340757,
      "loss": 0.35867738723754883,
      "loss_ce": 9.585694351699203e-05,
      "loss_iou": 0.1552734375,
      "loss_num": 0.009521484375,
      "loss_xval": 0.359375,
      "num_input_tokens_seen": 418069596,
      "step": 7461
    },
    {
      "epoch": 16.619153674832962,
      "grad_norm": 16.423025131225586,
      "learning_rate": 1e-06,
      "loss": 0.5082,
      "num_input_tokens_seen": 418125016,
      "step": 7462
    },
    {
      "epoch": 16.619153674832962,
      "loss": 0.7706419229507446,
      "loss_ce": 0.00013412557018455118,
      "loss_iou": 0.30078125,
      "loss_num": 0.033935546875,
      "loss_xval": 0.76953125,
      "num_input_tokens_seen": 418125016,
      "step": 7462
    },
    {
      "epoch": 16.621380846325167,
      "grad_norm": 21.270771026611328,
      "learning_rate": 1e-06,
      "loss": 0.5281,
      "num_input_tokens_seen": 418180920,
      "step": 7463
    },
    {
      "epoch": 16.621380846325167,
      "loss": 0.5241183042526245,
      "loss_ce": 7.047592953313142e-05,
      "loss_iou": 0.203125,
      "loss_num": 0.023681640625,
      "loss_xval": 0.5234375,
      "num_input_tokens_seen": 418180920,
      "step": 7463
    },
    {
      "epoch": 16.62360801781737,
      "grad_norm": 18.848526000976562,
      "learning_rate": 1e-06,
      "loss": 0.4706,
      "num_input_tokens_seen": 418236248,
      "step": 7464
    },
    {
      "epoch": 16.62360801781737,
      "loss": 0.5589928030967712,
      "loss_ce": 0.0002769696293398738,
      "loss_iou": 0.232421875,
      "loss_num": 0.018798828125,
      "loss_xval": 0.55859375,
      "num_input_tokens_seen": 418236248,
      "step": 7464
    },
    {
      "epoch": 16.625835189309576,
      "grad_norm": 18.260950088500977,
      "learning_rate": 1e-06,
      "loss": 0.4122,
      "num_input_tokens_seen": 418292596,
      "step": 7465
    },
    {
      "epoch": 16.625835189309576,
      "loss": 0.5881445407867432,
      "loss_ce": 0.00013190052413847297,
      "loss_iou": 0.255859375,
      "loss_num": 0.015380859375,
      "loss_xval": 0.58984375,
      "num_input_tokens_seen": 418292596,
      "step": 7465
    },
    {
      "epoch": 16.62806236080178,
      "grad_norm": 19.902740478515625,
      "learning_rate": 1e-06,
      "loss": 0.4294,
      "num_input_tokens_seen": 418348084,
      "step": 7466
    },
    {
      "epoch": 16.62806236080178,
      "loss": 0.4913012981414795,
      "loss_ce": 9.037186828209087e-05,
      "loss_iou": 0.2275390625,
      "loss_num": 0.00726318359375,
      "loss_xval": 0.4921875,
      "num_input_tokens_seen": 418348084,
      "step": 7466
    },
    {
      "epoch": 16.630289532293986,
      "grad_norm": 23.93308448791504,
      "learning_rate": 1e-06,
      "loss": 0.3016,
      "num_input_tokens_seen": 418404308,
      "step": 7467
    },
    {
      "epoch": 16.630289532293986,
      "loss": 0.2502018213272095,
      "loss_ce": 7.976325287017971e-05,
      "loss_iou": 0.11572265625,
      "loss_num": 0.0037689208984375,
      "loss_xval": 0.25,
      "num_input_tokens_seen": 418404308,
      "step": 7467
    },
    {
      "epoch": 16.63251670378619,
      "grad_norm": 16.425756454467773,
      "learning_rate": 1e-06,
      "loss": 0.4722,
      "num_input_tokens_seen": 418460596,
      "step": 7468
    },
    {
      "epoch": 16.63251670378619,
      "loss": 0.5028359889984131,
      "loss_ce": 8.939744293456897e-05,
      "loss_iou": 0.23046875,
      "loss_num": 0.00860595703125,
      "loss_xval": 0.50390625,
      "num_input_tokens_seen": 418460596,
      "step": 7468
    },
    {
      "epoch": 16.634743875278396,
      "grad_norm": 15.644196510314941,
      "learning_rate": 1e-06,
      "loss": 0.5605,
      "num_input_tokens_seen": 418516840,
      "step": 7469
    },
    {
      "epoch": 16.634743875278396,
      "loss": 0.8118726015090942,
      "loss_ce": 0.00010501188808120787,
      "loss_iou": 0.306640625,
      "loss_num": 0.03955078125,
      "loss_xval": 0.8125,
      "num_input_tokens_seen": 418516840,
      "step": 7469
    },
    {
      "epoch": 16.6369710467706,
      "grad_norm": 21.61148452758789,
      "learning_rate": 1e-06,
      "loss": 0.5139,
      "num_input_tokens_seen": 418575128,
      "step": 7470
    },
    {
      "epoch": 16.6369710467706,
      "loss": 0.6065396070480347,
      "loss_ce": 9.430487989448011e-05,
      "loss_iou": 0.271484375,
      "loss_num": 0.01239013671875,
      "loss_xval": 0.60546875,
      "num_input_tokens_seen": 418575128,
      "step": 7470
    },
    {
      "epoch": 16.639198218262806,
      "grad_norm": 17.30372428894043,
      "learning_rate": 1e-06,
      "loss": 0.5143,
      "num_input_tokens_seen": 418631736,
      "step": 7471
    },
    {
      "epoch": 16.639198218262806,
      "loss": 0.32015693187713623,
      "loss_ce": 8.858228102326393e-05,
      "loss_iou": 0.134765625,
      "loss_num": 0.0098876953125,
      "loss_xval": 0.3203125,
      "num_input_tokens_seen": 418631736,
      "step": 7471
    },
    {
      "epoch": 16.64142538975501,
      "grad_norm": 20.17923355102539,
      "learning_rate": 1e-06,
      "loss": 0.3716,
      "num_input_tokens_seen": 418685912,
      "step": 7472
    },
    {
      "epoch": 16.64142538975501,
      "loss": 0.39355170726776123,
      "loss_ce": 0.0007294651586562395,
      "loss_iou": 0.1826171875,
      "loss_num": 0.005645751953125,
      "loss_xval": 0.392578125,
      "num_input_tokens_seen": 418685912,
      "step": 7472
    },
    {
      "epoch": 16.643652561247215,
      "grad_norm": 21.51079559326172,
      "learning_rate": 1e-06,
      "loss": 0.386,
      "num_input_tokens_seen": 418739856,
      "step": 7473
    },
    {
      "epoch": 16.643652561247215,
      "loss": 0.3053746223449707,
      "loss_ce": 7.678715337533504e-05,
      "loss_iou": 0.1298828125,
      "loss_num": 0.00897216796875,
      "loss_xval": 0.3046875,
      "num_input_tokens_seen": 418739856,
      "step": 7473
    },
    {
      "epoch": 16.64587973273942,
      "grad_norm": 18.511350631713867,
      "learning_rate": 1e-06,
      "loss": 0.5517,
      "num_input_tokens_seen": 418796664,
      "step": 7474
    },
    {
      "epoch": 16.64587973273942,
      "loss": 0.4539787769317627,
      "loss_ce": 0.0003654793545138091,
      "loss_iou": 0.19140625,
      "loss_num": 0.0142822265625,
      "loss_xval": 0.453125,
      "num_input_tokens_seen": 418796664,
      "step": 7474
    },
    {
      "epoch": 16.648106904231625,
      "grad_norm": 18.879634857177734,
      "learning_rate": 1e-06,
      "loss": 0.5122,
      "num_input_tokens_seen": 418853068,
      "step": 7475
    },
    {
      "epoch": 16.648106904231625,
      "loss": 0.6420133113861084,
      "loss_ce": 0.00010659612598828971,
      "loss_iou": 0.296875,
      "loss_num": 0.00970458984375,
      "loss_xval": 0.640625,
      "num_input_tokens_seen": 418853068,
      "step": 7475
    },
    {
      "epoch": 16.65033407572383,
      "grad_norm": 22.095844268798828,
      "learning_rate": 1e-06,
      "loss": 0.4475,
      "num_input_tokens_seen": 418907716,
      "step": 7476
    },
    {
      "epoch": 16.65033407572383,
      "loss": 0.5320791006088257,
      "loss_ce": 9.662572119850665e-05,
      "loss_iou": 0.251953125,
      "loss_num": 0.00555419921875,
      "loss_xval": 0.53125,
      "num_input_tokens_seen": 418907716,
      "step": 7476
    },
    {
      "epoch": 16.652561247216035,
      "grad_norm": 22.207090377807617,
      "learning_rate": 1e-06,
      "loss": 0.4054,
      "num_input_tokens_seen": 418963728,
      "step": 7477
    },
    {
      "epoch": 16.652561247216035,
      "loss": 0.3793591856956482,
      "loss_ce": 8.672128024045378e-05,
      "loss_iou": 0.1728515625,
      "loss_num": 0.00677490234375,
      "loss_xval": 0.37890625,
      "num_input_tokens_seen": 418963728,
      "step": 7477
    },
    {
      "epoch": 16.65478841870824,
      "grad_norm": 16.67801856994629,
      "learning_rate": 1e-06,
      "loss": 0.6013,
      "num_input_tokens_seen": 419016604,
      "step": 7478
    },
    {
      "epoch": 16.65478841870824,
      "loss": 0.6391165256500244,
      "loss_ce": 7.844362698961049e-05,
      "loss_iou": 0.279296875,
      "loss_num": 0.016357421875,
      "loss_xval": 0.640625,
      "num_input_tokens_seen": 419016604,
      "step": 7478
    },
    {
      "epoch": 16.657015590200444,
      "grad_norm": 14.033241271972656,
      "learning_rate": 1e-06,
      "loss": 0.4418,
      "num_input_tokens_seen": 419074052,
      "step": 7479
    },
    {
      "epoch": 16.657015590200444,
      "loss": 0.3826572299003601,
      "loss_ce": 8.888993761502206e-05,
      "loss_iou": 0.177734375,
      "loss_num": 0.0054931640625,
      "loss_xval": 0.3828125,
      "num_input_tokens_seen": 419074052,
      "step": 7479
    },
    {
      "epoch": 16.65924276169265,
      "grad_norm": 13.08681869506836,
      "learning_rate": 1e-06,
      "loss": 0.3011,
      "num_input_tokens_seen": 419133076,
      "step": 7480
    },
    {
      "epoch": 16.65924276169265,
      "loss": 0.18371227383613586,
      "loss_ce": 8.800254727248102e-05,
      "loss_iou": 0.07373046875,
      "loss_num": 0.007110595703125,
      "loss_xval": 0.18359375,
      "num_input_tokens_seen": 419133076,
      "step": 7480
    },
    {
      "epoch": 16.661469933184854,
      "grad_norm": 19.100345611572266,
      "learning_rate": 1e-06,
      "loss": 0.3935,
      "num_input_tokens_seen": 419186412,
      "step": 7481
    },
    {
      "epoch": 16.661469933184854,
      "loss": 0.3841213583946228,
      "loss_ce": 0.00021022261353209615,
      "loss_iou": 0.1748046875,
      "loss_num": 0.006591796875,
      "loss_xval": 0.384765625,
      "num_input_tokens_seen": 419186412,
      "step": 7481
    },
    {
      "epoch": 16.66369710467706,
      "grad_norm": 15.811746597290039,
      "learning_rate": 1e-06,
      "loss": 0.3697,
      "num_input_tokens_seen": 419244064,
      "step": 7482
    },
    {
      "epoch": 16.66369710467706,
      "loss": 0.36989980936050415,
      "loss_ce": 8.779930794844404e-05,
      "loss_iou": 0.169921875,
      "loss_num": 0.0059814453125,
      "loss_xval": 0.369140625,
      "num_input_tokens_seen": 419244064,
      "step": 7482
    },
    {
      "epoch": 16.665924276169264,
      "grad_norm": 19.878259658813477,
      "learning_rate": 1e-06,
      "loss": 0.4357,
      "num_input_tokens_seen": 419300828,
      "step": 7483
    },
    {
      "epoch": 16.665924276169264,
      "loss": 0.3201012909412384,
      "loss_ce": 9.395321103511378e-05,
      "loss_iou": 0.134765625,
      "loss_num": 0.01007080078125,
      "loss_xval": 0.3203125,
      "num_input_tokens_seen": 419300828,
      "step": 7483
    },
    {
      "epoch": 16.66815144766147,
      "grad_norm": 18.946975708007812,
      "learning_rate": 1e-06,
      "loss": 0.4566,
      "num_input_tokens_seen": 419355872,
      "step": 7484
    },
    {
      "epoch": 16.66815144766147,
      "loss": 0.48397648334503174,
      "loss_ce": 8.974310185294598e-05,
      "loss_iou": 0.2236328125,
      "loss_num": 0.00750732421875,
      "loss_xval": 0.484375,
      "num_input_tokens_seen": 419355872,
      "step": 7484
    },
    {
      "epoch": 16.670378619153674,
      "grad_norm": 30.613773345947266,
      "learning_rate": 1e-06,
      "loss": 0.4782,
      "num_input_tokens_seen": 419411584,
      "step": 7485
    },
    {
      "epoch": 16.670378619153674,
      "loss": 0.42331117391586304,
      "loss_ce": 0.00021546825882978737,
      "loss_iou": 0.16796875,
      "loss_num": 0.0174560546875,
      "loss_xval": 0.423828125,
      "num_input_tokens_seen": 419411584,
      "step": 7485
    },
    {
      "epoch": 16.67260579064588,
      "grad_norm": 12.629964828491211,
      "learning_rate": 1e-06,
      "loss": 0.4128,
      "num_input_tokens_seen": 419468052,
      "step": 7486
    },
    {
      "epoch": 16.67260579064588,
      "loss": 0.5439175367355347,
      "loss_ce": 9.432664228370413e-05,
      "loss_iou": 0.2314453125,
      "loss_num": 0.01611328125,
      "loss_xval": 0.54296875,
      "num_input_tokens_seen": 419468052,
      "step": 7486
    },
    {
      "epoch": 16.674832962138083,
      "grad_norm": 21.88045883178711,
      "learning_rate": 1e-06,
      "loss": 0.3252,
      "num_input_tokens_seen": 419522816,
      "step": 7487
    },
    {
      "epoch": 16.674832962138083,
      "loss": 0.3819087743759155,
      "loss_ce": 0.00013384762860368937,
      "loss_iou": 0.1640625,
      "loss_num": 0.01080322265625,
      "loss_xval": 0.380859375,
      "num_input_tokens_seen": 419522816,
      "step": 7487
    },
    {
      "epoch": 16.677060133630288,
      "grad_norm": 20.802846908569336,
      "learning_rate": 1e-06,
      "loss": 0.4882,
      "num_input_tokens_seen": 419576664,
      "step": 7488
    },
    {
      "epoch": 16.677060133630288,
      "loss": 0.3609151840209961,
      "loss_ce": 0.00010585598647594452,
      "loss_iou": 0.171875,
      "loss_num": 0.003204345703125,
      "loss_xval": 0.361328125,
      "num_input_tokens_seen": 419576664,
      "step": 7488
    },
    {
      "epoch": 16.679287305122493,
      "grad_norm": 30.43656349182129,
      "learning_rate": 1e-06,
      "loss": 0.4603,
      "num_input_tokens_seen": 419631152,
      "step": 7489
    },
    {
      "epoch": 16.679287305122493,
      "loss": 0.5530772805213928,
      "loss_ce": 9.875621617538854e-05,
      "loss_iou": 0.2490234375,
      "loss_num": 0.01104736328125,
      "loss_xval": 0.5546875,
      "num_input_tokens_seen": 419631152,
      "step": 7489
    },
    {
      "epoch": 16.681514476614698,
      "grad_norm": 24.13094139099121,
      "learning_rate": 1e-06,
      "loss": 0.4435,
      "num_input_tokens_seen": 419687488,
      "step": 7490
    },
    {
      "epoch": 16.681514476614698,
      "loss": 0.37974900007247925,
      "loss_ce": 0.00011031976464437321,
      "loss_iou": 0.1708984375,
      "loss_num": 0.007476806640625,
      "loss_xval": 0.37890625,
      "num_input_tokens_seen": 419687488,
      "step": 7490
    },
    {
      "epoch": 16.683741648106903,
      "grad_norm": 15.836715698242188,
      "learning_rate": 1e-06,
      "loss": 0.4628,
      "num_input_tokens_seen": 419745188,
      "step": 7491
    },
    {
      "epoch": 16.683741648106903,
      "loss": 0.4252764582633972,
      "loss_ce": 0.0001055425891536288,
      "loss_iou": 0.169921875,
      "loss_num": 0.01708984375,
      "loss_xval": 0.42578125,
      "num_input_tokens_seen": 419745188,
      "step": 7491
    },
    {
      "epoch": 16.685968819599108,
      "grad_norm": 13.860366821289062,
      "learning_rate": 1e-06,
      "loss": 0.7326,
      "num_input_tokens_seen": 419798892,
      "step": 7492
    },
    {
      "epoch": 16.685968819599108,
      "loss": 0.8469661474227905,
      "loss_ce": 0.00010327581549063325,
      "loss_iou": 0.33984375,
      "loss_num": 0.033935546875,
      "loss_xval": 0.84765625,
      "num_input_tokens_seen": 419798892,
      "step": 7492
    },
    {
      "epoch": 16.688195991091312,
      "grad_norm": 18.5252742767334,
      "learning_rate": 1e-06,
      "loss": 0.4905,
      "num_input_tokens_seen": 419852848,
      "step": 7493
    },
    {
      "epoch": 16.688195991091312,
      "loss": 0.4520108699798584,
      "loss_ce": 0.00012183596845716238,
      "loss_iou": 0.193359375,
      "loss_num": 0.01287841796875,
      "loss_xval": 0.451171875,
      "num_input_tokens_seen": 419852848,
      "step": 7493
    },
    {
      "epoch": 16.690423162583517,
      "grad_norm": 33.927978515625,
      "learning_rate": 1e-06,
      "loss": 0.4021,
      "num_input_tokens_seen": 419909276,
      "step": 7494
    },
    {
      "epoch": 16.690423162583517,
      "loss": 0.4526256322860718,
      "loss_ce": 0.00013765764015261084,
      "loss_iou": 0.2060546875,
      "loss_num": 0.0081787109375,
      "loss_xval": 0.453125,
      "num_input_tokens_seen": 419909276,
      "step": 7494
    },
    {
      "epoch": 16.692650334075722,
      "grad_norm": 17.114513397216797,
      "learning_rate": 1e-06,
      "loss": 0.4049,
      "num_input_tokens_seen": 419965156,
      "step": 7495
    },
    {
      "epoch": 16.692650334075722,
      "loss": 0.5079430341720581,
      "loss_ce": 0.00019155110931023955,
      "loss_iou": 0.23046875,
      "loss_num": 0.00933837890625,
      "loss_xval": 0.5078125,
      "num_input_tokens_seen": 419965156,
      "step": 7495
    },
    {
      "epoch": 16.694877505567927,
      "grad_norm": 20.878402709960938,
      "learning_rate": 1e-06,
      "loss": 0.3769,
      "num_input_tokens_seen": 420021460,
      "step": 7496
    },
    {
      "epoch": 16.694877505567927,
      "loss": 0.5149946212768555,
      "loss_ce": 0.00010204176942352206,
      "loss_iou": 0.19921875,
      "loss_num": 0.0235595703125,
      "loss_xval": 0.515625,
      "num_input_tokens_seen": 420021460,
      "step": 7496
    },
    {
      "epoch": 16.697104677060132,
      "grad_norm": 11.510066032409668,
      "learning_rate": 1e-06,
      "loss": 0.3932,
      "num_input_tokens_seen": 420077988,
      "step": 7497
    },
    {
      "epoch": 16.697104677060132,
      "loss": 0.4444170892238617,
      "loss_ce": 8.116405660985038e-05,
      "loss_iou": 0.19140625,
      "loss_num": 0.01214599609375,
      "loss_xval": 0.4453125,
      "num_input_tokens_seen": 420077988,
      "step": 7497
    },
    {
      "epoch": 16.69933184855234,
      "grad_norm": 21.444026947021484,
      "learning_rate": 1e-06,
      "loss": 0.4323,
      "num_input_tokens_seen": 420135408,
      "step": 7498
    },
    {
      "epoch": 16.69933184855234,
      "loss": 0.34426337480545044,
      "loss_ce": 8.61354055814445e-05,
      "loss_iou": 0.16015625,
      "loss_num": 0.004638671875,
      "loss_xval": 0.34375,
      "num_input_tokens_seen": 420135408,
      "step": 7498
    },
    {
      "epoch": 16.70155902004454,
      "grad_norm": 18.71256446838379,
      "learning_rate": 1e-06,
      "loss": 0.3837,
      "num_input_tokens_seen": 420191188,
      "step": 7499
    },
    {
      "epoch": 16.70155902004454,
      "loss": 0.5224635601043701,
      "loss_ce": 0.00012473194510675967,
      "loss_iou": 0.2333984375,
      "loss_num": 0.01129150390625,
      "loss_xval": 0.5234375,
      "num_input_tokens_seen": 420191188,
      "step": 7499
    },
    {
      "epoch": 16.70378619153675,
      "grad_norm": 29.126888275146484,
      "learning_rate": 1e-06,
      "loss": 0.3387,
      "num_input_tokens_seen": 420248812,
      "step": 7500
    },
    {
      "epoch": 16.70378619153675,
      "eval_seeclick_web_CIoU": 0.5869028568267822,
      "eval_seeclick_web_GIoU": 0.5851646363735199,
      "eval_seeclick_web_IoU": 0.6060521006584167,
      "eval_seeclick_web_MAE_all": 0.015174605650827289,
      "eval_seeclick_web_MAE_h": 0.007236489560455084,
      "eval_seeclick_web_MAE_w": 0.015413875225931406,
      "eval_seeclick_web_MAE_x_boxes": 0.008164346683770418,
      "eval_seeclick_web_MAE_y_boxes": 0.02128999726846814,
      "eval_seeclick_web_inside_bbox": 0.9010416567325592,
      "eval_seeclick_web_loss": 0.9088981747627258,
      "eval_seeclick_web_loss_ce": 0.00014426549023482949,
      "eval_seeclick_web_loss_iou": 0.419921875,
      "eval_seeclick_web_loss_num": 0.01212310791015625,
      "eval_seeclick_web_loss_xval": 0.900146484375,
      "eval_seeclick_web_runtime": 24.9651,
      "eval_seeclick_web_samples_per_second": 2.003,
      "eval_seeclick_web_steps_per_second": 0.08,
      "num_input_tokens_seen": 420248812,
      "step": 7500
    },
    {
      "epoch": 16.70378619153675,
      "eval_icons_CIoU": 0.2704924941062927,
      "eval_icons_GIoU": 0.2915039211511612,
      "eval_icons_IoU": 0.34834006428718567,
      "eval_icons_MAE_all": 0.059374475851655006,
      "eval_icons_MAE_h": 0.03465741407126188,
      "eval_icons_MAE_w": 0.058007813058793545,
      "eval_icons_MAE_x_boxes": 0.05980631522834301,
      "eval_icons_MAE_y_boxes": 0.03711246699094772,
      "eval_icons_inside_bbox": 0.59375,
      "eval_icons_loss": 1.7317432165145874,
      "eval_icons_loss_ce": 0.0001763724285410717,
      "eval_icons_loss_iou": 0.680908203125,
      "eval_icons_loss_num": 0.0583648681640625,
      "eval_icons_loss_xval": 1.654541015625,
      "eval_icons_runtime": 24.5042,
      "eval_icons_samples_per_second": 2.04,
      "eval_icons_steps_per_second": 0.082,
      "num_input_tokens_seen": 420248812,
      "step": 7500
    },
    {
      "epoch": 16.70378619153675,
      "eval_screenspot_CIoU": 0.37447473406791687,
      "eval_screenspot_GIoU": 0.392286479473114,
      "eval_screenspot_IoU": 0.4470125635464986,
      "eval_screenspot_MAE_all": 0.057117752730846405,
      "eval_screenspot_MAE_h": 0.039663772409160934,
      "eval_screenspot_MAE_w": 0.06271877388159434,
      "eval_screenspot_MAE_x_boxes": 0.06692921556532383,
      "eval_screenspot_MAE_y_boxes": 0.03935454785823822,
      "eval_screenspot_inside_bbox": 0.693750003973643,
      "eval_screenspot_loss": 1.5693501234054565,
      "eval_screenspot_loss_ce": 0.0001859396434156224,
      "eval_screenspot_loss_iou": 0.651611328125,
      "eval_screenspot_loss_num": 0.06552505493164062,
      "eval_screenspot_loss_xval": 1.6298828125,
      "eval_screenspot_runtime": 40.939,
      "eval_screenspot_samples_per_second": 2.174,
      "eval_screenspot_steps_per_second": 0.073,
      "num_input_tokens_seen": 420248812,
      "step": 7500
    },
    {
      "epoch": 16.70378619153675,
      "eval_compot_CIoU": 0.346510648727417,
      "eval_compot_GIoU": 0.35510683059692383,
      "eval_compot_IoU": 0.4062563627958298,
      "eval_compot_MAE_all": 0.01756852399557829,
      "eval_compot_MAE_h": 0.00829399167560041,
      "eval_compot_MAE_w": 0.02114854846149683,
      "eval_compot_MAE_x_boxes": 0.029594723135232925,
      "eval_compot_MAE_y_boxes": 0.0067301481030881405,
      "eval_compot_inside_bbox": 0.6302083432674408,
      "eval_compot_loss": 1.3873258829116821,
      "eval_compot_loss_ce": 0.00013870510883862153,
      "eval_compot_loss_iou": 0.6431884765625,
      "eval_compot_loss_num": 0.016324996948242188,
      "eval_compot_loss_xval": 1.366943359375,
      "eval_compot_runtime": 25.1596,
      "eval_compot_samples_per_second": 1.987,
      "eval_compot_steps_per_second": 0.079,
      "num_input_tokens_seen": 420248812,
      "step": 7500
    },
    {
      "epoch": 16.70378619153675,
      "eval_custom_ui_val_CIoU": 0.47488271362251705,
      "eval_custom_ui_val_GIoU": 0.4806961201959186,
      "eval_custom_ui_val_IoU": 0.5364741186300913,
      "eval_custom_ui_val_MAE_all": 0.027154105249792337,
      "eval_custom_ui_val_MAE_h": 0.014837595815252926,
      "eval_custom_ui_val_MAE_w": 0.036442533497595124,
      "eval_custom_ui_val_MAE_x_boxes": 0.03312135862910913,
      "eval_custom_ui_val_MAE_y_boxes": 0.012457883761574825,
      "eval_custom_ui_val_inside_bbox": 0.7754629651705424,
      "eval_custom_ui_val_loss": 1.1677404642105103,
      "eval_custom_ui_val_loss_ce": 0.00016540858794340037,
      "eval_custom_ui_val_loss_iou": 0.5015055338541666,
      "eval_custom_ui_val_loss_num": 0.023720741271972656,
      "eval_custom_ui_val_loss_xval": 1.1214735243055556,
      "eval_custom_ui_val_runtime": 78.9911,
      "eval_custom_ui_val_samples_per_second": 3.355,
      "eval_custom_ui_val_steps_per_second": 0.114,
      "num_input_tokens_seen": 420248812,
      "step": 7500
    },
    {
      "epoch": 16.70378619153675,
      "loss": 0.8367924690246582,
      "loss_ce": 0.00012252142187207937,
      "loss_iou": 0.376953125,
      "loss_num": 0.016845703125,
      "loss_xval": 0.8359375,
      "num_input_tokens_seen": 420248812,
      "step": 7500
    },
    {
      "epoch": 16.706013363028955,
      "grad_norm": 16.65875244140625,
      "learning_rate": 1e-06,
      "loss": 0.4408,
      "num_input_tokens_seen": 420303256,
      "step": 7501
    },
    {
      "epoch": 16.706013363028955,
      "loss": 0.3736181855201721,
      "loss_ce": 8.305851952172816e-05,
      "loss_iou": 0.1591796875,
      "loss_num": 0.01116943359375,
      "loss_xval": 0.373046875,
      "num_input_tokens_seen": 420303256,
      "step": 7501
    },
    {
      "epoch": 16.70824053452116,
      "grad_norm": 17.692901611328125,
      "learning_rate": 1e-06,
      "loss": 0.5898,
      "num_input_tokens_seen": 420356912,
      "step": 7502
    },
    {
      "epoch": 16.70824053452116,
      "loss": 0.33262205123901367,
      "loss_ce": 0.00010252789070364088,
      "loss_iou": 0.142578125,
      "loss_num": 0.00958251953125,
      "loss_xval": 0.33203125,
      "num_input_tokens_seen": 420356912,
      "step": 7502
    },
    {
      "epoch": 16.710467706013365,
      "grad_norm": 19.661197662353516,
      "learning_rate": 1e-06,
      "loss": 0.4288,
      "num_input_tokens_seen": 420414376,
      "step": 7503
    },
    {
      "epoch": 16.710467706013365,
      "loss": 0.4624839723110199,
      "loss_ce": 8.163749589584768e-05,
      "loss_iou": 0.181640625,
      "loss_num": 0.0196533203125,
      "loss_xval": 0.462890625,
      "num_input_tokens_seen": 420414376,
      "step": 7503
    },
    {
      "epoch": 16.71269487750557,
      "grad_norm": 16.957685470581055,
      "learning_rate": 1e-06,
      "loss": 0.4464,
      "num_input_tokens_seen": 420470196,
      "step": 7504
    },
    {
      "epoch": 16.71269487750557,
      "loss": 0.39802345633506775,
      "loss_ce": 7.423176430165768e-05,
      "loss_iou": 0.166015625,
      "loss_num": 0.0130615234375,
      "loss_xval": 0.3984375,
      "num_input_tokens_seen": 420470196,
      "step": 7504
    },
    {
      "epoch": 16.714922048997774,
      "grad_norm": 18.1544246673584,
      "learning_rate": 1e-06,
      "loss": 0.4431,
      "num_input_tokens_seen": 420525872,
      "step": 7505
    },
    {
      "epoch": 16.714922048997774,
      "loss": 0.48571863770484924,
      "loss_ce": 0.00024499627761542797,
      "loss_iou": 0.2001953125,
      "loss_num": 0.016845703125,
      "loss_xval": 0.486328125,
      "num_input_tokens_seen": 420525872,
      "step": 7505
    },
    {
      "epoch": 16.71714922048998,
      "grad_norm": 14.577059745788574,
      "learning_rate": 1e-06,
      "loss": 0.5386,
      "num_input_tokens_seen": 420582216,
      "step": 7506
    },
    {
      "epoch": 16.71714922048998,
      "loss": 0.6150659918785095,
      "loss_ce": 0.00013681976997759193,
      "loss_iou": 0.2451171875,
      "loss_num": 0.0247802734375,
      "loss_xval": 0.61328125,
      "num_input_tokens_seen": 420582216,
      "step": 7506
    },
    {
      "epoch": 16.719376391982184,
      "grad_norm": 16.228076934814453,
      "learning_rate": 1e-06,
      "loss": 0.3875,
      "num_input_tokens_seen": 420636876,
      "step": 7507
    },
    {
      "epoch": 16.719376391982184,
      "loss": 0.5354748964309692,
      "loss_ce": 7.452783756889403e-05,
      "loss_iou": 0.240234375,
      "loss_num": 0.01080322265625,
      "loss_xval": 0.53515625,
      "num_input_tokens_seen": 420636876,
      "step": 7507
    },
    {
      "epoch": 16.72160356347439,
      "grad_norm": 17.355379104614258,
      "learning_rate": 1e-06,
      "loss": 0.3967,
      "num_input_tokens_seen": 420692648,
      "step": 7508
    },
    {
      "epoch": 16.72160356347439,
      "loss": 0.5319057703018188,
      "loss_ce": 0.00010642388951964676,
      "loss_iou": 0.2216796875,
      "loss_num": 0.017578125,
      "loss_xval": 0.53125,
      "num_input_tokens_seen": 420692648,
      "step": 7508
    },
    {
      "epoch": 16.723830734966594,
      "grad_norm": 22.888402938842773,
      "learning_rate": 1e-06,
      "loss": 0.3304,
      "num_input_tokens_seen": 420749380,
      "step": 7509
    },
    {
      "epoch": 16.723830734966594,
      "loss": 0.2873430550098419,
      "loss_ce": 0.00011160832218592986,
      "loss_iou": 0.1142578125,
      "loss_num": 0.01171875,
      "loss_xval": 0.287109375,
      "num_input_tokens_seen": 420749380,
      "step": 7509
    },
    {
      "epoch": 16.7260579064588,
      "grad_norm": 23.103485107421875,
      "learning_rate": 1e-06,
      "loss": 0.5022,
      "num_input_tokens_seen": 420805372,
      "step": 7510
    },
    {
      "epoch": 16.7260579064588,
      "loss": 0.6403497457504272,
      "loss_ce": 9.102303010877222e-05,
      "loss_iou": 0.296875,
      "loss_num": 0.00958251953125,
      "loss_xval": 0.640625,
      "num_input_tokens_seen": 420805372,
      "step": 7510
    },
    {
      "epoch": 16.728285077951004,
      "grad_norm": 26.94756317138672,
      "learning_rate": 1e-06,
      "loss": 0.4418,
      "num_input_tokens_seen": 420859376,
      "step": 7511
    },
    {
      "epoch": 16.728285077951004,
      "loss": 0.3712967336177826,
      "loss_ce": 8.091152994893491e-05,
      "loss_iou": 0.166015625,
      "loss_num": 0.00787353515625,
      "loss_xval": 0.37109375,
      "num_input_tokens_seen": 420859376,
      "step": 7511
    },
    {
      "epoch": 16.73051224944321,
      "grad_norm": 29.820947647094727,
      "learning_rate": 1e-06,
      "loss": 0.4807,
      "num_input_tokens_seen": 420916012,
      "step": 7512
    },
    {
      "epoch": 16.73051224944321,
      "loss": 0.5431552529335022,
      "loss_ce": 0.00018649011326488107,
      "loss_iou": 0.2265625,
      "loss_num": 0.0179443359375,
      "loss_xval": 0.54296875,
      "num_input_tokens_seen": 420916012,
      "step": 7512
    },
    {
      "epoch": 16.732739420935413,
      "grad_norm": 16.320466995239258,
      "learning_rate": 1e-06,
      "loss": 0.7292,
      "num_input_tokens_seen": 420970600,
      "step": 7513
    },
    {
      "epoch": 16.732739420935413,
      "loss": 0.877778172492981,
      "loss_ce": 9.26764914765954e-05,
      "loss_iou": 0.34765625,
      "loss_num": 0.036376953125,
      "loss_xval": 0.87890625,
      "num_input_tokens_seen": 420970600,
      "step": 7513
    },
    {
      "epoch": 16.734966592427618,
      "grad_norm": 19.527677536010742,
      "learning_rate": 1e-06,
      "loss": 0.3925,
      "num_input_tokens_seen": 421024580,
      "step": 7514
    },
    {
      "epoch": 16.734966592427618,
      "loss": 0.29793840646743774,
      "loss_ce": 8.684221393195912e-05,
      "loss_iou": 0.1376953125,
      "loss_num": 0.0045166015625,
      "loss_xval": 0.296875,
      "num_input_tokens_seen": 421024580,
      "step": 7514
    },
    {
      "epoch": 16.737193763919823,
      "grad_norm": 16.6092472076416,
      "learning_rate": 1e-06,
      "loss": 0.4416,
      "num_input_tokens_seen": 421080592,
      "step": 7515
    },
    {
      "epoch": 16.737193763919823,
      "loss": 0.32314634323120117,
      "loss_ce": 8.72400269145146e-05,
      "loss_iou": 0.1328125,
      "loss_num": 0.01141357421875,
      "loss_xval": 0.322265625,
      "num_input_tokens_seen": 421080592,
      "step": 7515
    },
    {
      "epoch": 16.739420935412028,
      "grad_norm": 18.05164337158203,
      "learning_rate": 1e-06,
      "loss": 0.4574,
      "num_input_tokens_seen": 421138628,
      "step": 7516
    },
    {
      "epoch": 16.739420935412028,
      "loss": 0.44442254304885864,
      "loss_ce": 8.662776235723868e-05,
      "loss_iou": 0.1884765625,
      "loss_num": 0.013671875,
      "loss_xval": 0.4453125,
      "num_input_tokens_seen": 421138628,
      "step": 7516
    },
    {
      "epoch": 16.741648106904233,
      "grad_norm": 34.90769958496094,
      "learning_rate": 1e-06,
      "loss": 0.3832,
      "num_input_tokens_seen": 421194844,
      "step": 7517
    },
    {
      "epoch": 16.741648106904233,
      "loss": 0.4219837486743927,
      "loss_ce": 0.00010874809231609106,
      "loss_iou": 0.197265625,
      "loss_num": 0.00555419921875,
      "loss_xval": 0.421875,
      "num_input_tokens_seen": 421194844,
      "step": 7517
    },
    {
      "epoch": 16.743875278396438,
      "grad_norm": 19.655269622802734,
      "learning_rate": 1e-06,
      "loss": 0.404,
      "num_input_tokens_seen": 421248904,
      "step": 7518
    },
    {
      "epoch": 16.743875278396438,
      "loss": 0.48480457067489624,
      "loss_ce": 0.00012440019054338336,
      "loss_iou": 0.216796875,
      "loss_num": 0.01019287109375,
      "loss_xval": 0.484375,
      "num_input_tokens_seen": 421248904,
      "step": 7518
    },
    {
      "epoch": 16.746102449888642,
      "grad_norm": 22.647844314575195,
      "learning_rate": 1e-06,
      "loss": 0.671,
      "num_input_tokens_seen": 421304408,
      "step": 7519
    },
    {
      "epoch": 16.746102449888642,
      "loss": 0.8310278654098511,
      "loss_ce": 9.524515189696103e-05,
      "loss_iou": 0.353515625,
      "loss_num": 0.025146484375,
      "loss_xval": 0.83203125,
      "num_input_tokens_seen": 421304408,
      "step": 7519
    },
    {
      "epoch": 16.748329621380847,
      "grad_norm": 12.694985389709473,
      "learning_rate": 1e-06,
      "loss": 0.3328,
      "num_input_tokens_seen": 421359956,
      "step": 7520
    },
    {
      "epoch": 16.748329621380847,
      "loss": 0.3517601191997528,
      "loss_ce": 7.55441069486551e-05,
      "loss_iou": 0.158203125,
      "loss_num": 0.007232666015625,
      "loss_xval": 0.3515625,
      "num_input_tokens_seen": 421359956,
      "step": 7520
    },
    {
      "epoch": 16.750556792873052,
      "grad_norm": 26.37984275817871,
      "learning_rate": 1e-06,
      "loss": 0.635,
      "num_input_tokens_seen": 421418924,
      "step": 7521
    },
    {
      "epoch": 16.750556792873052,
      "loss": 0.6089690923690796,
      "loss_ce": 8.241091563832015e-05,
      "loss_iou": 0.255859375,
      "loss_num": 0.019775390625,
      "loss_xval": 0.609375,
      "num_input_tokens_seen": 421418924,
      "step": 7521
    },
    {
      "epoch": 16.752783964365257,
      "grad_norm": 19.50495719909668,
      "learning_rate": 1e-06,
      "loss": 0.3985,
      "num_input_tokens_seen": 421474016,
      "step": 7522
    },
    {
      "epoch": 16.752783964365257,
      "loss": 0.46183595061302185,
      "loss_ce": 0.00010498787742108107,
      "loss_iou": 0.203125,
      "loss_num": 0.01123046875,
      "loss_xval": 0.4609375,
      "num_input_tokens_seen": 421474016,
      "step": 7522
    },
    {
      "epoch": 16.755011135857462,
      "grad_norm": 19.55943489074707,
      "learning_rate": 1e-06,
      "loss": 0.5932,
      "num_input_tokens_seen": 421528604,
      "step": 7523
    },
    {
      "epoch": 16.755011135857462,
      "loss": 0.672928512096405,
      "loss_ce": 0.00019901388441212475,
      "loss_iou": 0.279296875,
      "loss_num": 0.02294921875,
      "loss_xval": 0.671875,
      "num_input_tokens_seen": 421528604,
      "step": 7523
    },
    {
      "epoch": 16.757238307349667,
      "grad_norm": 18.727272033691406,
      "learning_rate": 1e-06,
      "loss": 0.4604,
      "num_input_tokens_seen": 421582348,
      "step": 7524
    },
    {
      "epoch": 16.757238307349667,
      "loss": 0.43595296144485474,
      "loss_ce": 0.00016193735064007342,
      "loss_iou": 0.1962890625,
      "loss_num": 0.008544921875,
      "loss_xval": 0.435546875,
      "num_input_tokens_seen": 421582348,
      "step": 7524
    },
    {
      "epoch": 16.75946547884187,
      "grad_norm": 14.18976879119873,
      "learning_rate": 1e-06,
      "loss": 0.3096,
      "num_input_tokens_seen": 421637456,
      "step": 7525
    },
    {
      "epoch": 16.75946547884187,
      "loss": 0.2345210313796997,
      "loss_ce": 8.500301919411868e-05,
      "loss_iou": 0.10888671875,
      "loss_num": 0.0033416748046875,
      "loss_xval": 0.234375,
      "num_input_tokens_seen": 421637456,
      "step": 7525
    },
    {
      "epoch": 16.761692650334076,
      "grad_norm": 14.92958927154541,
      "learning_rate": 1e-06,
      "loss": 0.4751,
      "num_input_tokens_seen": 421694176,
      "step": 7526
    },
    {
      "epoch": 16.761692650334076,
      "loss": 0.3878576159477234,
      "loss_ce": 0.00010129041038453579,
      "loss_iou": 0.1591796875,
      "loss_num": 0.0140380859375,
      "loss_xval": 0.388671875,
      "num_input_tokens_seen": 421694176,
      "step": 7526
    },
    {
      "epoch": 16.76391982182628,
      "grad_norm": 23.31275749206543,
      "learning_rate": 1e-06,
      "loss": 0.5295,
      "num_input_tokens_seen": 421751032,
      "step": 7527
    },
    {
      "epoch": 16.76391982182628,
      "loss": 0.5836684703826904,
      "loss_ce": 8.08104668976739e-05,
      "loss_iou": 0.224609375,
      "loss_num": 0.02685546875,
      "loss_xval": 0.58203125,
      "num_input_tokens_seen": 421751032,
      "step": 7527
    },
    {
      "epoch": 16.766146993318486,
      "grad_norm": 24.81498146057129,
      "learning_rate": 1e-06,
      "loss": 0.4622,
      "num_input_tokens_seen": 421806836,
      "step": 7528
    },
    {
      "epoch": 16.766146993318486,
      "loss": 0.37533727288246155,
      "loss_ce": 9.314088674727827e-05,
      "loss_iou": 0.1728515625,
      "loss_num": 0.005767822265625,
      "loss_xval": 0.375,
      "num_input_tokens_seen": 421806836,
      "step": 7528
    },
    {
      "epoch": 16.76837416481069,
      "grad_norm": 23.45658302307129,
      "learning_rate": 1e-06,
      "loss": 0.4903,
      "num_input_tokens_seen": 421863004,
      "step": 7529
    },
    {
      "epoch": 16.76837416481069,
      "loss": 0.6420989036560059,
      "loss_ce": 0.00013110964209772646,
      "loss_iou": 0.291015625,
      "loss_num": 0.011962890625,
      "loss_xval": 0.640625,
      "num_input_tokens_seen": 421863004,
      "step": 7529
    },
    {
      "epoch": 16.770601336302896,
      "grad_norm": 17.918804168701172,
      "learning_rate": 1e-06,
      "loss": 0.3386,
      "num_input_tokens_seen": 421918756,
      "step": 7530
    },
    {
      "epoch": 16.770601336302896,
      "loss": 0.3590957224369049,
      "loss_ce": 8.693411655258387e-05,
      "loss_iou": 0.150390625,
      "loss_num": 0.0115966796875,
      "loss_xval": 0.359375,
      "num_input_tokens_seen": 421918756,
      "step": 7530
    },
    {
      "epoch": 16.7728285077951,
      "grad_norm": 21.4240665435791,
      "learning_rate": 1e-06,
      "loss": 0.4388,
      "num_input_tokens_seen": 421975608,
      "step": 7531
    },
    {
      "epoch": 16.7728285077951,
      "loss": 0.3268599510192871,
      "loss_ce": 7.773490506224334e-05,
      "loss_iou": 0.138671875,
      "loss_num": 0.009765625,
      "loss_xval": 0.326171875,
      "num_input_tokens_seen": 421975608,
      "step": 7531
    },
    {
      "epoch": 16.775055679287306,
      "grad_norm": 22.135826110839844,
      "learning_rate": 1e-06,
      "loss": 0.2985,
      "num_input_tokens_seen": 422029336,
      "step": 7532
    },
    {
      "epoch": 16.775055679287306,
      "loss": 0.2817379832267761,
      "loss_ce": 9.12583782337606e-05,
      "loss_iou": 0.12109375,
      "loss_num": 0.00799560546875,
      "loss_xval": 0.28125,
      "num_input_tokens_seen": 422029336,
      "step": 7532
    },
    {
      "epoch": 16.77728285077951,
      "grad_norm": 12.753730773925781,
      "learning_rate": 1e-06,
      "loss": 0.4948,
      "num_input_tokens_seen": 422087488,
      "step": 7533
    },
    {
      "epoch": 16.77728285077951,
      "loss": 0.6205785274505615,
      "loss_ce": 0.00046129614929668605,
      "loss_iou": 0.2734375,
      "loss_num": 0.01483154296875,
      "loss_xval": 0.62109375,
      "num_input_tokens_seen": 422087488,
      "step": 7533
    },
    {
      "epoch": 16.779510022271715,
      "grad_norm": 17.831968307495117,
      "learning_rate": 1e-06,
      "loss": 0.3904,
      "num_input_tokens_seen": 422144960,
      "step": 7534
    },
    {
      "epoch": 16.779510022271715,
      "loss": 0.3965558111667633,
      "loss_ce": 7.143749098759145e-05,
      "loss_iou": 0.150390625,
      "loss_num": 0.0194091796875,
      "loss_xval": 0.396484375,
      "num_input_tokens_seen": 422144960,
      "step": 7534
    },
    {
      "epoch": 16.78173719376392,
      "grad_norm": 19.048486709594727,
      "learning_rate": 1e-06,
      "loss": 0.2356,
      "num_input_tokens_seen": 422202632,
      "step": 7535
    },
    {
      "epoch": 16.78173719376392,
      "loss": 0.19454200565814972,
      "loss_ce": 8.399138459935784e-05,
      "loss_iou": 0.0888671875,
      "loss_num": 0.003265380859375,
      "loss_xval": 0.1943359375,
      "num_input_tokens_seen": 422202632,
      "step": 7535
    },
    {
      "epoch": 16.783964365256125,
      "grad_norm": 15.023216247558594,
      "learning_rate": 1e-06,
      "loss": 0.3777,
      "num_input_tokens_seen": 422258820,
      "step": 7536
    },
    {
      "epoch": 16.783964365256125,
      "loss": 0.25117045640945435,
      "loss_ce": 7.181320688687265e-05,
      "loss_iou": 0.1162109375,
      "loss_num": 0.003814697265625,
      "loss_xval": 0.251953125,
      "num_input_tokens_seen": 422258820,
      "step": 7536
    },
    {
      "epoch": 16.78619153674833,
      "grad_norm": 12.973251342773438,
      "learning_rate": 1e-06,
      "loss": 0.421,
      "num_input_tokens_seen": 422317024,
      "step": 7537
    },
    {
      "epoch": 16.78619153674833,
      "loss": 0.37338966131210327,
      "loss_ce": 9.861498983809724e-05,
      "loss_iou": 0.1591796875,
      "loss_num": 0.01092529296875,
      "loss_xval": 0.373046875,
      "num_input_tokens_seen": 422317024,
      "step": 7537
    },
    {
      "epoch": 16.788418708240535,
      "grad_norm": 26.241365432739258,
      "learning_rate": 1e-06,
      "loss": 0.3026,
      "num_input_tokens_seen": 422372340,
      "step": 7538
    },
    {
      "epoch": 16.788418708240535,
      "loss": 0.31962519884109497,
      "loss_ce": 0.00010615789506118745,
      "loss_iou": 0.1513671875,
      "loss_num": 0.0032806396484375,
      "loss_xval": 0.3203125,
      "num_input_tokens_seen": 422372340,
      "step": 7538
    },
    {
      "epoch": 16.79064587973274,
      "grad_norm": 17.4132080078125,
      "learning_rate": 1e-06,
      "loss": 0.3208,
      "num_input_tokens_seen": 422427068,
      "step": 7539
    },
    {
      "epoch": 16.79064587973274,
      "loss": 0.4059646427631378,
      "loss_ce": 8.086010348051786e-05,
      "loss_iou": 0.1875,
      "loss_num": 0.006317138671875,
      "loss_xval": 0.40625,
      "num_input_tokens_seen": 422427068,
      "step": 7539
    },
    {
      "epoch": 16.792873051224944,
      "grad_norm": 18.924129486083984,
      "learning_rate": 1e-06,
      "loss": 0.4015,
      "num_input_tokens_seen": 422485208,
      "step": 7540
    },
    {
      "epoch": 16.792873051224944,
      "loss": 0.549044668674469,
      "loss_ce": 9.448261698707938e-05,
      "loss_iou": 0.2294921875,
      "loss_num": 0.0181884765625,
      "loss_xval": 0.55078125,
      "num_input_tokens_seen": 422485208,
      "step": 7540
    },
    {
      "epoch": 16.79510022271715,
      "grad_norm": 21.377471923828125,
      "learning_rate": 1e-06,
      "loss": 0.2579,
      "num_input_tokens_seen": 422541288,
      "step": 7541
    },
    {
      "epoch": 16.79510022271715,
      "loss": 0.22633783519268036,
      "loss_ce": 8.05138552095741e-05,
      "loss_iou": 0.09619140625,
      "loss_num": 0.006805419921875,
      "loss_xval": 0.2265625,
      "num_input_tokens_seen": 422541288,
      "step": 7541
    },
    {
      "epoch": 16.797327394209354,
      "grad_norm": 18.096935272216797,
      "learning_rate": 1e-06,
      "loss": 0.4972,
      "num_input_tokens_seen": 422598728,
      "step": 7542
    },
    {
      "epoch": 16.797327394209354,
      "loss": 0.598351240158081,
      "loss_ce": 8.468693704344332e-05,
      "loss_iou": 0.2353515625,
      "loss_num": 0.025146484375,
      "loss_xval": 0.59765625,
      "num_input_tokens_seen": 422598728,
      "step": 7542
    },
    {
      "epoch": 16.79955456570156,
      "grad_norm": 18.19603157043457,
      "learning_rate": 1e-06,
      "loss": 0.463,
      "num_input_tokens_seen": 422653832,
      "step": 7543
    },
    {
      "epoch": 16.79955456570156,
      "loss": 0.5140941143035889,
      "loss_ce": 8.656126010464504e-05,
      "loss_iou": 0.205078125,
      "loss_num": 0.0205078125,
      "loss_xval": 0.515625,
      "num_input_tokens_seen": 422653832,
      "step": 7543
    },
    {
      "epoch": 16.801781737193764,
      "grad_norm": 20.3741397857666,
      "learning_rate": 1e-06,
      "loss": 0.3391,
      "num_input_tokens_seen": 422711052,
      "step": 7544
    },
    {
      "epoch": 16.801781737193764,
      "loss": 0.37486082315444946,
      "loss_ce": 0.00010497802577447146,
      "loss_iou": 0.173828125,
      "loss_num": 0.00518798828125,
      "loss_xval": 0.375,
      "num_input_tokens_seen": 422711052,
      "step": 7544
    },
    {
      "epoch": 16.80400890868597,
      "grad_norm": 15.383708953857422,
      "learning_rate": 1e-06,
      "loss": 0.4458,
      "num_input_tokens_seen": 422768836,
      "step": 7545
    },
    {
      "epoch": 16.80400890868597,
      "loss": 0.2621135711669922,
      "loss_ce": 8.963444997789338e-05,
      "loss_iou": 0.1201171875,
      "loss_num": 0.004302978515625,
      "loss_xval": 0.26171875,
      "num_input_tokens_seen": 422768836,
      "step": 7545
    },
    {
      "epoch": 16.806236080178174,
      "grad_norm": 19.3375186920166,
      "learning_rate": 1e-06,
      "loss": 0.3409,
      "num_input_tokens_seen": 422826720,
      "step": 7546
    },
    {
      "epoch": 16.806236080178174,
      "loss": 0.23085373640060425,
      "loss_ce": 7.98294713604264e-05,
      "loss_iou": 0.107421875,
      "loss_num": 0.003204345703125,
      "loss_xval": 0.23046875,
      "num_input_tokens_seen": 422826720,
      "step": 7546
    },
    {
      "epoch": 16.80846325167038,
      "grad_norm": 27.87537956237793,
      "learning_rate": 1e-06,
      "loss": 0.3236,
      "num_input_tokens_seen": 422881060,
      "step": 7547
    },
    {
      "epoch": 16.80846325167038,
      "loss": 0.2520201802253723,
      "loss_ce": 6.706906424369663e-05,
      "loss_iou": 0.08837890625,
      "loss_num": 0.01513671875,
      "loss_xval": 0.251953125,
      "num_input_tokens_seen": 422881060,
      "step": 7547
    },
    {
      "epoch": 16.810690423162583,
      "grad_norm": 9.865744590759277,
      "learning_rate": 1e-06,
      "loss": 0.3187,
      "num_input_tokens_seen": 422938468,
      "step": 7548
    },
    {
      "epoch": 16.810690423162583,
      "loss": 0.39917415380477905,
      "loss_ce": 0.00012633406731765717,
      "loss_iou": 0.1787109375,
      "loss_num": 0.00830078125,
      "loss_xval": 0.3984375,
      "num_input_tokens_seen": 422938468,
      "step": 7548
    },
    {
      "epoch": 16.812917594654788,
      "grad_norm": 18.907991409301758,
      "learning_rate": 1e-06,
      "loss": 0.4061,
      "num_input_tokens_seen": 422996068,
      "step": 7549
    },
    {
      "epoch": 16.812917594654788,
      "loss": 0.5208408832550049,
      "loss_ce": 8.891297329682857e-05,
      "loss_iou": 0.21875,
      "loss_num": 0.0167236328125,
      "loss_xval": 0.51953125,
      "num_input_tokens_seen": 422996068,
      "step": 7549
    },
    {
      "epoch": 16.815144766146993,
      "grad_norm": 14.58835506439209,
      "learning_rate": 1e-06,
      "loss": 0.3215,
      "num_input_tokens_seen": 423051768,
      "step": 7550
    },
    {
      "epoch": 16.815144766146993,
      "loss": 0.4090169668197632,
      "loss_ce": 8.143430750351399e-05,
      "loss_iou": 0.1884765625,
      "loss_num": 0.00628662109375,
      "loss_xval": 0.408203125,
      "num_input_tokens_seen": 423051768,
      "step": 7550
    },
    {
      "epoch": 16.817371937639198,
      "grad_norm": 14.23434066772461,
      "learning_rate": 1e-06,
      "loss": 0.3621,
      "num_input_tokens_seen": 423107452,
      "step": 7551
    },
    {
      "epoch": 16.817371937639198,
      "loss": 0.2738141119480133,
      "loss_ce": 7.142306276364252e-05,
      "loss_iou": 0.119140625,
      "loss_num": 0.00701904296875,
      "loss_xval": 0.2734375,
      "num_input_tokens_seen": 423107452,
      "step": 7551
    },
    {
      "epoch": 16.819599109131403,
      "grad_norm": 17.120817184448242,
      "learning_rate": 1e-06,
      "loss": 0.4439,
      "num_input_tokens_seen": 423165408,
      "step": 7552
    },
    {
      "epoch": 16.819599109131403,
      "loss": 0.466147780418396,
      "loss_ce": 8.328772673849016e-05,
      "loss_iou": 0.1865234375,
      "loss_num": 0.0185546875,
      "loss_xval": 0.466796875,
      "num_input_tokens_seen": 423165408,
      "step": 7552
    },
    {
      "epoch": 16.821826280623608,
      "grad_norm": 16.54876708984375,
      "learning_rate": 1e-06,
      "loss": 0.5828,
      "num_input_tokens_seen": 423220424,
      "step": 7553
    },
    {
      "epoch": 16.821826280623608,
      "loss": 0.656104326248169,
      "loss_ce": 9.84652797342278e-05,
      "loss_iou": 0.28515625,
      "loss_num": 0.017333984375,
      "loss_xval": 0.65625,
      "num_input_tokens_seen": 423220424,
      "step": 7553
    },
    {
      "epoch": 16.824053452115812,
      "grad_norm": 29.920957565307617,
      "learning_rate": 1e-06,
      "loss": 0.5121,
      "num_input_tokens_seen": 423275472,
      "step": 7554
    },
    {
      "epoch": 16.824053452115812,
      "loss": 0.4944070279598236,
      "loss_ce": 8.32905643619597e-05,
      "loss_iou": 0.1943359375,
      "loss_num": 0.02099609375,
      "loss_xval": 0.494140625,
      "num_input_tokens_seen": 423275472,
      "step": 7554
    },
    {
      "epoch": 16.826280623608017,
      "grad_norm": 14.726850509643555,
      "learning_rate": 1e-06,
      "loss": 0.3757,
      "num_input_tokens_seen": 423332856,
      "step": 7555
    },
    {
      "epoch": 16.826280623608017,
      "loss": 0.4061638116836548,
      "loss_ce": 9.694288019090891e-05,
      "loss_iou": 0.16796875,
      "loss_num": 0.013916015625,
      "loss_xval": 0.40625,
      "num_input_tokens_seen": 423332856,
      "step": 7555
    },
    {
      "epoch": 16.828507795100222,
      "grad_norm": 18.766006469726562,
      "learning_rate": 1e-06,
      "loss": 0.5588,
      "num_input_tokens_seen": 423388552,
      "step": 7556
    },
    {
      "epoch": 16.828507795100222,
      "loss": 0.5013834834098816,
      "loss_ce": 0.00010174483759328723,
      "loss_iou": 0.205078125,
      "loss_num": 0.018310546875,
      "loss_xval": 0.5,
      "num_input_tokens_seen": 423388552,
      "step": 7556
    },
    {
      "epoch": 16.830734966592427,
      "grad_norm": 16.07285499572754,
      "learning_rate": 1e-06,
      "loss": 0.4426,
      "num_input_tokens_seen": 423443352,
      "step": 7557
    },
    {
      "epoch": 16.830734966592427,
      "loss": 0.531631588935852,
      "loss_ce": 0.00013745043543167412,
      "loss_iou": 0.236328125,
      "loss_num": 0.01177978515625,
      "loss_xval": 0.53125,
      "num_input_tokens_seen": 423443352,
      "step": 7557
    },
    {
      "epoch": 16.832962138084632,
      "grad_norm": 22.96206283569336,
      "learning_rate": 1e-06,
      "loss": 0.2828,
      "num_input_tokens_seen": 423497072,
      "step": 7558
    },
    {
      "epoch": 16.832962138084632,
      "loss": 0.2979274392127991,
      "loss_ce": 7.588087464682758e-05,
      "loss_iou": 0.12353515625,
      "loss_num": 0.01025390625,
      "loss_xval": 0.296875,
      "num_input_tokens_seen": 423497072,
      "step": 7558
    },
    {
      "epoch": 16.835189309576837,
      "grad_norm": 14.747753143310547,
      "learning_rate": 1e-06,
      "loss": 0.3194,
      "num_input_tokens_seen": 423554452,
      "step": 7559
    },
    {
      "epoch": 16.835189309576837,
      "loss": 0.41189420223236084,
      "loss_ce": 9.000718273455277e-05,
      "loss_iou": 0.1923828125,
      "loss_num": 0.005218505859375,
      "loss_xval": 0.412109375,
      "num_input_tokens_seen": 423554452,
      "step": 7559
    },
    {
      "epoch": 16.83741648106904,
      "grad_norm": 13.279129981994629,
      "learning_rate": 1e-06,
      "loss": 0.3928,
      "num_input_tokens_seen": 423612016,
      "step": 7560
    },
    {
      "epoch": 16.83741648106904,
      "loss": 0.32673531770706177,
      "loss_ce": 7.517023186665028e-05,
      "loss_iou": 0.1376953125,
      "loss_num": 0.01025390625,
      "loss_xval": 0.326171875,
      "num_input_tokens_seen": 423612016,
      "step": 7560
    },
    {
      "epoch": 16.839643652561247,
      "grad_norm": 15.924846649169922,
      "learning_rate": 1e-06,
      "loss": 0.392,
      "num_input_tokens_seen": 423668608,
      "step": 7561
    },
    {
      "epoch": 16.839643652561247,
      "loss": 0.2862039804458618,
      "loss_ce": 7.118898065527901e-05,
      "loss_iou": 0.12353515625,
      "loss_num": 0.00775146484375,
      "loss_xval": 0.28515625,
      "num_input_tokens_seen": 423668608,
      "step": 7561
    },
    {
      "epoch": 16.84187082405345,
      "grad_norm": 19.908687591552734,
      "learning_rate": 1e-06,
      "loss": 0.5825,
      "num_input_tokens_seen": 423725484,
      "step": 7562
    },
    {
      "epoch": 16.84187082405345,
      "loss": 0.5031726360321045,
      "loss_ce": 0.00012085959315299988,
      "loss_iou": 0.2236328125,
      "loss_num": 0.01116943359375,
      "loss_xval": 0.50390625,
      "num_input_tokens_seen": 423725484,
      "step": 7562
    },
    {
      "epoch": 16.844097995545656,
      "grad_norm": 25.447425842285156,
      "learning_rate": 1e-06,
      "loss": 0.3432,
      "num_input_tokens_seen": 423778384,
      "step": 7563
    },
    {
      "epoch": 16.844097995545656,
      "loss": 0.3695857524871826,
      "loss_ce": 7.891730638220906e-05,
      "loss_iou": 0.1591796875,
      "loss_num": 0.01019287109375,
      "loss_xval": 0.369140625,
      "num_input_tokens_seen": 423778384,
      "step": 7563
    },
    {
      "epoch": 16.84632516703786,
      "grad_norm": 13.935490608215332,
      "learning_rate": 1e-06,
      "loss": 0.4838,
      "num_input_tokens_seen": 423834376,
      "step": 7564
    },
    {
      "epoch": 16.84632516703786,
      "loss": 0.4895142912864685,
      "loss_ce": 0.00013441329065244645,
      "loss_iou": 0.1845703125,
      "loss_num": 0.024169921875,
      "loss_xval": 0.490234375,
      "num_input_tokens_seen": 423834376,
      "step": 7564
    },
    {
      "epoch": 16.848552338530066,
      "grad_norm": 16.72061538696289,
      "learning_rate": 1e-06,
      "loss": 0.3445,
      "num_input_tokens_seen": 423890956,
      "step": 7565
    },
    {
      "epoch": 16.848552338530066,
      "loss": 0.4084116220474243,
      "loss_ce": 0.00014747484237886965,
      "loss_iou": 0.1826171875,
      "loss_num": 0.00848388671875,
      "loss_xval": 0.408203125,
      "num_input_tokens_seen": 423890956,
      "step": 7565
    },
    {
      "epoch": 16.85077951002227,
      "grad_norm": 26.282730102539062,
      "learning_rate": 1e-06,
      "loss": 0.519,
      "num_input_tokens_seen": 423947064,
      "step": 7566
    },
    {
      "epoch": 16.85077951002227,
      "loss": 0.6066428422927856,
      "loss_ce": 7.544091204181314e-05,
      "loss_iou": 0.25390625,
      "loss_num": 0.0196533203125,
      "loss_xval": 0.60546875,
      "num_input_tokens_seen": 423947064,
      "step": 7566
    },
    {
      "epoch": 16.853006681514476,
      "grad_norm": 15.286888122558594,
      "learning_rate": 1e-06,
      "loss": 0.4031,
      "num_input_tokens_seen": 424005024,
      "step": 7567
    },
    {
      "epoch": 16.853006681514476,
      "loss": 0.29945024847984314,
      "loss_ce": 7.280143472598866e-05,
      "loss_iou": 0.1318359375,
      "loss_num": 0.007171630859375,
      "loss_xval": 0.298828125,
      "num_input_tokens_seen": 424005024,
      "step": 7567
    },
    {
      "epoch": 16.85523385300668,
      "grad_norm": 22.391944885253906,
      "learning_rate": 1e-06,
      "loss": 0.3159,
      "num_input_tokens_seen": 424061520,
      "step": 7568
    },
    {
      "epoch": 16.85523385300668,
      "loss": 0.3320120573043823,
      "loss_ce": 0.00010290154023095965,
      "loss_iou": 0.15234375,
      "loss_num": 0.005401611328125,
      "loss_xval": 0.33203125,
      "num_input_tokens_seen": 424061520,
      "step": 7568
    },
    {
      "epoch": 16.857461024498885,
      "grad_norm": 28.07231903076172,
      "learning_rate": 1e-06,
      "loss": 0.512,
      "num_input_tokens_seen": 424115184,
      "step": 7569
    },
    {
      "epoch": 16.857461024498885,
      "loss": 0.4680614471435547,
      "loss_ce": 0.00010489917622180656,
      "loss_iou": 0.212890625,
      "loss_num": 0.00860595703125,
      "loss_xval": 0.46875,
      "num_input_tokens_seen": 424115184,
      "step": 7569
    },
    {
      "epoch": 16.85968819599109,
      "grad_norm": 23.56254768371582,
      "learning_rate": 1e-06,
      "loss": 0.3468,
      "num_input_tokens_seen": 424169644,
      "step": 7570
    },
    {
      "epoch": 16.85968819599109,
      "loss": 0.41865983605384827,
      "loss_ce": 8.075643563643098e-05,
      "loss_iou": 0.1953125,
      "loss_num": 0.0054931640625,
      "loss_xval": 0.41796875,
      "num_input_tokens_seen": 424169644,
      "step": 7570
    },
    {
      "epoch": 16.861915367483295,
      "grad_norm": 25.315942764282227,
      "learning_rate": 1e-06,
      "loss": 0.4083,
      "num_input_tokens_seen": 424227580,
      "step": 7571
    },
    {
      "epoch": 16.861915367483295,
      "loss": 0.37649667263031006,
      "loss_ce": 9.28444933379069e-05,
      "loss_iou": 0.166015625,
      "loss_num": 0.00872802734375,
      "loss_xval": 0.376953125,
      "num_input_tokens_seen": 424227580,
      "step": 7571
    },
    {
      "epoch": 16.8641425389755,
      "grad_norm": 15.719208717346191,
      "learning_rate": 1e-06,
      "loss": 0.4909,
      "num_input_tokens_seen": 424283408,
      "step": 7572
    },
    {
      "epoch": 16.8641425389755,
      "loss": 0.4299257695674896,
      "loss_ce": 0.00011618290591286495,
      "loss_iou": 0.185546875,
      "loss_num": 0.01165771484375,
      "loss_xval": 0.4296875,
      "num_input_tokens_seen": 424283408,
      "step": 7572
    },
    {
      "epoch": 16.866369710467705,
      "grad_norm": 28.304363250732422,
      "learning_rate": 1e-06,
      "loss": 0.2948,
      "num_input_tokens_seen": 424338840,
      "step": 7573
    },
    {
      "epoch": 16.866369710467705,
      "loss": 0.3035458028316498,
      "loss_ce": 7.902114157332107e-05,
      "loss_iou": 0.1318359375,
      "loss_num": 0.0078125,
      "loss_xval": 0.302734375,
      "num_input_tokens_seen": 424338840,
      "step": 7573
    },
    {
      "epoch": 16.86859688195991,
      "grad_norm": 13.032916069030762,
      "learning_rate": 1e-06,
      "loss": 0.5414,
      "num_input_tokens_seen": 424394268,
      "step": 7574
    },
    {
      "epoch": 16.86859688195991,
      "loss": 0.5727271437644958,
      "loss_ce": 9.532412514090538e-05,
      "loss_iou": 0.2373046875,
      "loss_num": 0.0196533203125,
      "loss_xval": 0.57421875,
      "num_input_tokens_seen": 424394268,
      "step": 7574
    },
    {
      "epoch": 16.870824053452115,
      "grad_norm": 16.96046257019043,
      "learning_rate": 1e-06,
      "loss": 0.4261,
      "num_input_tokens_seen": 424453868,
      "step": 7575
    },
    {
      "epoch": 16.870824053452115,
      "loss": 0.38206472992897034,
      "loss_ce": 0.0001067298071575351,
      "loss_iou": 0.173828125,
      "loss_num": 0.007080078125,
      "loss_xval": 0.3828125,
      "num_input_tokens_seen": 424453868,
      "step": 7575
    },
    {
      "epoch": 16.87305122494432,
      "grad_norm": 23.169174194335938,
      "learning_rate": 1e-06,
      "loss": 0.3393,
      "num_input_tokens_seen": 424510884,
      "step": 7576
    },
    {
      "epoch": 16.87305122494432,
      "loss": 0.33035528659820557,
      "loss_ce": 9.405257151229307e-05,
      "loss_iou": 0.1416015625,
      "loss_num": 0.00927734375,
      "loss_xval": 0.330078125,
      "num_input_tokens_seen": 424510884,
      "step": 7576
    },
    {
      "epoch": 16.875278396436524,
      "grad_norm": 20.488079071044922,
      "learning_rate": 1e-06,
      "loss": 0.3433,
      "num_input_tokens_seen": 424566384,
      "step": 7577
    },
    {
      "epoch": 16.875278396436524,
      "loss": 0.32358628511428833,
      "loss_ce": 9.996739390771836e-05,
      "loss_iou": 0.142578125,
      "loss_num": 0.00750732421875,
      "loss_xval": 0.32421875,
      "num_input_tokens_seen": 424566384,
      "step": 7577
    },
    {
      "epoch": 16.87750556792873,
      "grad_norm": 13.715553283691406,
      "learning_rate": 1e-06,
      "loss": 0.3461,
      "num_input_tokens_seen": 424623228,
      "step": 7578
    },
    {
      "epoch": 16.87750556792873,
      "loss": 0.3440733253955841,
      "loss_ce": 7.916930189821869e-05,
      "loss_iou": 0.1455078125,
      "loss_num": 0.0107421875,
      "loss_xval": 0.34375,
      "num_input_tokens_seen": 424623228,
      "step": 7578
    },
    {
      "epoch": 16.879732739420934,
      "grad_norm": 16.320846557617188,
      "learning_rate": 1e-06,
      "loss": 0.5924,
      "num_input_tokens_seen": 424682444,
      "step": 7579
    },
    {
      "epoch": 16.879732739420934,
      "loss": 0.6377947330474854,
      "loss_ce": 9.944755584001541e-05,
      "loss_iou": 0.2734375,
      "loss_num": 0.0181884765625,
      "loss_xval": 0.63671875,
      "num_input_tokens_seen": 424682444,
      "step": 7579
    },
    {
      "epoch": 16.88195991091314,
      "grad_norm": 18.41837501525879,
      "learning_rate": 1e-06,
      "loss": 0.6497,
      "num_input_tokens_seen": 424740464,
      "step": 7580
    },
    {
      "epoch": 16.88195991091314,
      "loss": 0.6376847624778748,
      "loss_ce": 0.00011151684884680435,
      "loss_iou": 0.26953125,
      "loss_num": 0.0196533203125,
      "loss_xval": 0.63671875,
      "num_input_tokens_seen": 424740464,
      "step": 7580
    },
    {
      "epoch": 16.884187082405344,
      "grad_norm": 16.372474670410156,
      "learning_rate": 1e-06,
      "loss": 0.3357,
      "num_input_tokens_seen": 424794412,
      "step": 7581
    },
    {
      "epoch": 16.884187082405344,
      "loss": 0.28870633244514465,
      "loss_ce": 7.107005512807518e-05,
      "loss_iou": 0.1181640625,
      "loss_num": 0.01043701171875,
      "loss_xval": 0.2890625,
      "num_input_tokens_seen": 424794412,
      "step": 7581
    },
    {
      "epoch": 16.88641425389755,
      "grad_norm": 23.682388305664062,
      "learning_rate": 1e-06,
      "loss": 0.4056,
      "num_input_tokens_seen": 424852012,
      "step": 7582
    },
    {
      "epoch": 16.88641425389755,
      "loss": 0.34498876333236694,
      "loss_ce": 7.910602289484814e-05,
      "loss_iou": 0.15234375,
      "loss_num": 0.0079345703125,
      "loss_xval": 0.345703125,
      "num_input_tokens_seen": 424852012,
      "step": 7582
    },
    {
      "epoch": 16.888641425389753,
      "grad_norm": 22.36638069152832,
      "learning_rate": 1e-06,
      "loss": 0.4973,
      "num_input_tokens_seen": 424903972,
      "step": 7583
    },
    {
      "epoch": 16.888641425389753,
      "loss": 0.41586142778396606,
      "loss_ce": 8.992112998384982e-05,
      "loss_iou": 0.1865234375,
      "loss_num": 0.008544921875,
      "loss_xval": 0.416015625,
      "num_input_tokens_seen": 424903972,
      "step": 7583
    },
    {
      "epoch": 16.89086859688196,
      "grad_norm": 28.00014305114746,
      "learning_rate": 1e-06,
      "loss": 0.5041,
      "num_input_tokens_seen": 424960808,
      "step": 7584
    },
    {
      "epoch": 16.89086859688196,
      "loss": 0.5798003077507019,
      "loss_ce": 8.836777124088258e-05,
      "loss_iou": 0.251953125,
      "loss_num": 0.0155029296875,
      "loss_xval": 0.578125,
      "num_input_tokens_seen": 424960808,
      "step": 7584
    },
    {
      "epoch": 16.893095768374163,
      "grad_norm": 18.74024200439453,
      "learning_rate": 1e-06,
      "loss": 0.4247,
      "num_input_tokens_seen": 425018468,
      "step": 7585
    },
    {
      "epoch": 16.893095768374163,
      "loss": 0.477266788482666,
      "loss_ce": 9.394106746185571e-05,
      "loss_iou": 0.2119140625,
      "loss_num": 0.01043701171875,
      "loss_xval": 0.4765625,
      "num_input_tokens_seen": 425018468,
      "step": 7585
    },
    {
      "epoch": 16.895322939866368,
      "grad_norm": 20.79447364807129,
      "learning_rate": 1e-06,
      "loss": 0.4927,
      "num_input_tokens_seen": 425074148,
      "step": 7586
    },
    {
      "epoch": 16.895322939866368,
      "loss": 0.7077399492263794,
      "loss_ce": 9.840876737143844e-05,
      "loss_iou": 0.279296875,
      "loss_num": 0.029541015625,
      "loss_xval": 0.70703125,
      "num_input_tokens_seen": 425074148,
      "step": 7586
    },
    {
      "epoch": 16.897550111358576,
      "grad_norm": 23.163110733032227,
      "learning_rate": 1e-06,
      "loss": 0.3244,
      "num_input_tokens_seen": 425129428,
      "step": 7587
    },
    {
      "epoch": 16.897550111358576,
      "loss": 0.3205108642578125,
      "loss_ce": 7.629128958797082e-05,
      "loss_iou": 0.142578125,
      "loss_num": 0.007049560546875,
      "loss_xval": 0.3203125,
      "num_input_tokens_seen": 425129428,
      "step": 7587
    },
    {
      "epoch": 16.899777282850778,
      "grad_norm": 23.743364334106445,
      "learning_rate": 1e-06,
      "loss": 0.4087,
      "num_input_tokens_seen": 425184424,
      "step": 7588
    },
    {
      "epoch": 16.899777282850778,
      "loss": 0.5007613897323608,
      "loss_ce": 0.00015102376346476376,
      "loss_iou": 0.2255859375,
      "loss_num": 0.00994873046875,
      "loss_xval": 0.5,
      "num_input_tokens_seen": 425184424,
      "step": 7588
    },
    {
      "epoch": 16.902004454342986,
      "grad_norm": 22.105934143066406,
      "learning_rate": 1e-06,
      "loss": 0.3156,
      "num_input_tokens_seen": 425241216,
      "step": 7589
    },
    {
      "epoch": 16.902004454342986,
      "loss": 0.30233582854270935,
      "loss_ce": 8.974589582066983e-05,
      "loss_iou": 0.1357421875,
      "loss_num": 0.00604248046875,
      "loss_xval": 0.302734375,
      "num_input_tokens_seen": 425241216,
      "step": 7589
    },
    {
      "epoch": 16.90423162583519,
      "grad_norm": 26.043792724609375,
      "learning_rate": 1e-06,
      "loss": 0.6462,
      "num_input_tokens_seen": 425294740,
      "step": 7590
    },
    {
      "epoch": 16.90423162583519,
      "loss": 0.3308909833431244,
      "loss_ce": 8.044774585869163e-05,
      "loss_iou": 0.140625,
      "loss_num": 0.00982666015625,
      "loss_xval": 0.330078125,
      "num_input_tokens_seen": 425294740,
      "step": 7590
    },
    {
      "epoch": 16.906458797327396,
      "grad_norm": 25.259113311767578,
      "learning_rate": 1e-06,
      "loss": 0.4507,
      "num_input_tokens_seen": 425352476,
      "step": 7591
    },
    {
      "epoch": 16.906458797327396,
      "loss": 0.44027650356292725,
      "loss_ce": 9.096147550735623e-05,
      "loss_iou": 0.201171875,
      "loss_num": 0.007568359375,
      "loss_xval": 0.439453125,
      "num_input_tokens_seen": 425352476,
      "step": 7591
    },
    {
      "epoch": 16.9086859688196,
      "grad_norm": 18.93207359313965,
      "learning_rate": 1e-06,
      "loss": 0.4337,
      "num_input_tokens_seen": 425410420,
      "step": 7592
    },
    {
      "epoch": 16.9086859688196,
      "loss": 0.556984543800354,
      "loss_ce": 9.982170013245195e-05,
      "loss_iou": 0.2177734375,
      "loss_num": 0.0242919921875,
      "loss_xval": 0.55859375,
      "num_input_tokens_seen": 425410420,
      "step": 7592
    },
    {
      "epoch": 16.910913140311806,
      "grad_norm": 17.402809143066406,
      "learning_rate": 1e-06,
      "loss": 0.3179,
      "num_input_tokens_seen": 425468280,
      "step": 7593
    },
    {
      "epoch": 16.910913140311806,
      "loss": 0.3924216330051422,
      "loss_ce": 8.76395424711518e-05,
      "loss_iou": 0.177734375,
      "loss_num": 0.007568359375,
      "loss_xval": 0.392578125,
      "num_input_tokens_seen": 425468280,
      "step": 7593
    },
    {
      "epoch": 16.91314031180401,
      "grad_norm": 14.503034591674805,
      "learning_rate": 1e-06,
      "loss": 0.6439,
      "num_input_tokens_seen": 425525096,
      "step": 7594
    },
    {
      "epoch": 16.91314031180401,
      "loss": 0.9615048170089722,
      "loss_ce": 0.0005673256237059832,
      "loss_iou": 0.365234375,
      "loss_num": 0.04638671875,
      "loss_xval": 0.9609375,
      "num_input_tokens_seen": 425525096,
      "step": 7594
    },
    {
      "epoch": 16.915367483296215,
      "grad_norm": 21.804555892944336,
      "learning_rate": 1e-06,
      "loss": 0.4446,
      "num_input_tokens_seen": 425579188,
      "step": 7595
    },
    {
      "epoch": 16.915367483296215,
      "loss": 0.41501274704933167,
      "loss_ce": 9.576005686540157e-05,
      "loss_iou": 0.19140625,
      "loss_num": 0.00616455078125,
      "loss_xval": 0.4140625,
      "num_input_tokens_seen": 425579188,
      "step": 7595
    },
    {
      "epoch": 16.91759465478842,
      "grad_norm": 17.96055793762207,
      "learning_rate": 1e-06,
      "loss": 0.435,
      "num_input_tokens_seen": 425635356,
      "step": 7596
    },
    {
      "epoch": 16.91759465478842,
      "loss": 0.44575822353363037,
      "loss_ce": 7.95404048403725e-05,
      "loss_iou": 0.1982421875,
      "loss_num": 0.0096435546875,
      "loss_xval": 0.4453125,
      "num_input_tokens_seen": 425635356,
      "step": 7596
    },
    {
      "epoch": 16.919821826280625,
      "grad_norm": 22.727903366088867,
      "learning_rate": 1e-06,
      "loss": 0.4284,
      "num_input_tokens_seen": 425691988,
      "step": 7597
    },
    {
      "epoch": 16.919821826280625,
      "loss": 0.4949197769165039,
      "loss_ce": 0.0004129420267418027,
      "loss_iou": 0.2197265625,
      "loss_num": 0.01092529296875,
      "loss_xval": 0.494140625,
      "num_input_tokens_seen": 425691988,
      "step": 7597
    },
    {
      "epoch": 16.92204899777283,
      "grad_norm": 19.787315368652344,
      "learning_rate": 1e-06,
      "loss": 0.3273,
      "num_input_tokens_seen": 425749968,
      "step": 7598
    },
    {
      "epoch": 16.92204899777283,
      "loss": 0.2782631516456604,
      "loss_ce": 6.490422674687579e-05,
      "loss_iou": 0.1123046875,
      "loss_num": 0.0107421875,
      "loss_xval": 0.27734375,
      "num_input_tokens_seen": 425749968,
      "step": 7598
    },
    {
      "epoch": 16.924276169265035,
      "grad_norm": 19.99041748046875,
      "learning_rate": 1e-06,
      "loss": 0.5476,
      "num_input_tokens_seen": 425805264,
      "step": 7599
    },
    {
      "epoch": 16.924276169265035,
      "loss": 0.7404612302780151,
      "loss_ce": 0.00010478113836143166,
      "loss_iou": 0.30078125,
      "loss_num": 0.02783203125,
      "loss_xval": 0.7421875,
      "num_input_tokens_seen": 425805264,
      "step": 7599
    },
    {
      "epoch": 16.92650334075724,
      "grad_norm": 20.051441192626953,
      "learning_rate": 1e-06,
      "loss": 0.3816,
      "num_input_tokens_seen": 425860344,
      "step": 7600
    },
    {
      "epoch": 16.92650334075724,
      "loss": 0.4353218674659729,
      "loss_ce": 8.015791536308825e-05,
      "loss_iou": 0.1962890625,
      "loss_num": 0.0084228515625,
      "loss_xval": 0.435546875,
      "num_input_tokens_seen": 425860344,
      "step": 7600
    },
    {
      "epoch": 16.928730512249444,
      "grad_norm": 19.339773178100586,
      "learning_rate": 1e-06,
      "loss": 0.5078,
      "num_input_tokens_seen": 425916088,
      "step": 7601
    },
    {
      "epoch": 16.928730512249444,
      "loss": 0.4659002423286438,
      "loss_ce": 7.990589074324816e-05,
      "loss_iou": 0.21484375,
      "loss_num": 0.007171630859375,
      "loss_xval": 0.46484375,
      "num_input_tokens_seen": 425916088,
      "step": 7601
    },
    {
      "epoch": 16.93095768374165,
      "grad_norm": 11.72046947479248,
      "learning_rate": 1e-06,
      "loss": 0.271,
      "num_input_tokens_seen": 425971676,
      "step": 7602
    },
    {
      "epoch": 16.93095768374165,
      "loss": 0.3369971513748169,
      "loss_ce": 8.307769894599915e-05,
      "loss_iou": 0.142578125,
      "loss_num": 0.010498046875,
      "loss_xval": 0.3359375,
      "num_input_tokens_seen": 425971676,
      "step": 7602
    },
    {
      "epoch": 16.933184855233854,
      "grad_norm": 26.19060516357422,
      "learning_rate": 1e-06,
      "loss": 0.43,
      "num_input_tokens_seen": 426026612,
      "step": 7603
    },
    {
      "epoch": 16.933184855233854,
      "loss": 0.33956053853034973,
      "loss_ce": 8.299553883261979e-05,
      "loss_iou": 0.1416015625,
      "loss_num": 0.01123046875,
      "loss_xval": 0.33984375,
      "num_input_tokens_seen": 426026612,
      "step": 7603
    },
    {
      "epoch": 16.93541202672606,
      "grad_norm": 36.85425567626953,
      "learning_rate": 1e-06,
      "loss": 0.3744,
      "num_input_tokens_seen": 426084004,
      "step": 7604
    },
    {
      "epoch": 16.93541202672606,
      "loss": 0.4844852089881897,
      "loss_ce": 0.0004153858171775937,
      "loss_iou": 0.21484375,
      "loss_num": 0.01116943359375,
      "loss_xval": 0.484375,
      "num_input_tokens_seen": 426084004,
      "step": 7604
    },
    {
      "epoch": 16.937639198218264,
      "grad_norm": 22.670175552368164,
      "learning_rate": 1e-06,
      "loss": 0.3386,
      "num_input_tokens_seen": 426139888,
      "step": 7605
    },
    {
      "epoch": 16.937639198218264,
      "loss": 0.3556758165359497,
      "loss_ce": 8.5016421508044e-05,
      "loss_iou": 0.162109375,
      "loss_num": 0.006103515625,
      "loss_xval": 0.35546875,
      "num_input_tokens_seen": 426139888,
      "step": 7605
    },
    {
      "epoch": 16.93986636971047,
      "grad_norm": 18.98232650756836,
      "learning_rate": 1e-06,
      "loss": 0.4427,
      "num_input_tokens_seen": 426196256,
      "step": 7606
    },
    {
      "epoch": 16.93986636971047,
      "loss": 0.4649594724178314,
      "loss_ce": 0.00011571809591259807,
      "loss_iou": 0.205078125,
      "loss_num": 0.01104736328125,
      "loss_xval": 0.46484375,
      "num_input_tokens_seen": 426196256,
      "step": 7606
    },
    {
      "epoch": 16.942093541202674,
      "grad_norm": 774.5324096679688,
      "learning_rate": 1e-06,
      "loss": 0.3919,
      "num_input_tokens_seen": 426253332,
      "step": 7607
    },
    {
      "epoch": 16.942093541202674,
      "loss": 0.3354228436946869,
      "loss_ce": 9.569604299031198e-05,
      "loss_iou": 0.12890625,
      "loss_num": 0.015625,
      "loss_xval": 0.3359375,
      "num_input_tokens_seen": 426253332,
      "step": 7607
    },
    {
      "epoch": 16.94432071269488,
      "grad_norm": 53.58465576171875,
      "learning_rate": 1e-06,
      "loss": 0.4068,
      "num_input_tokens_seen": 426310116,
      "step": 7608
    },
    {
      "epoch": 16.94432071269488,
      "loss": 0.2681652903556824,
      "loss_ce": 9.887861961033195e-05,
      "loss_iou": 0.11669921875,
      "loss_num": 0.006988525390625,
      "loss_xval": 0.267578125,
      "num_input_tokens_seen": 426310116,
      "step": 7608
    },
    {
      "epoch": 16.946547884187083,
      "grad_norm": 19.653953552246094,
      "learning_rate": 1e-06,
      "loss": 0.3449,
      "num_input_tokens_seen": 426367920,
      "step": 7609
    },
    {
      "epoch": 16.946547884187083,
      "loss": 0.2864606976509094,
      "loss_ce": 0.0001142588589573279,
      "loss_iou": 0.126953125,
      "loss_num": 0.00640869140625,
      "loss_xval": 0.287109375,
      "num_input_tokens_seen": 426367920,
      "step": 7609
    },
    {
      "epoch": 16.948775055679288,
      "grad_norm": 15.588455200195312,
      "learning_rate": 1e-06,
      "loss": 0.5204,
      "num_input_tokens_seen": 426423652,
      "step": 7610
    },
    {
      "epoch": 16.948775055679288,
      "loss": 0.5305978655815125,
      "loss_ce": 8.031211473280564e-05,
      "loss_iou": 0.240234375,
      "loss_num": 0.01019287109375,
      "loss_xval": 0.53125,
      "num_input_tokens_seen": 426423652,
      "step": 7610
    },
    {
      "epoch": 16.951002227171493,
      "grad_norm": 14.135087013244629,
      "learning_rate": 1e-06,
      "loss": 0.3808,
      "num_input_tokens_seen": 426480624,
      "step": 7611
    },
    {
      "epoch": 16.951002227171493,
      "loss": 0.4105345606803894,
      "loss_ce": 7.310426735784858e-05,
      "loss_iou": 0.1767578125,
      "loss_num": 0.01123046875,
      "loss_xval": 0.41015625,
      "num_input_tokens_seen": 426480624,
      "step": 7611
    },
    {
      "epoch": 16.953229398663698,
      "grad_norm": 14.242335319519043,
      "learning_rate": 1e-06,
      "loss": 0.3675,
      "num_input_tokens_seen": 426537756,
      "step": 7612
    },
    {
      "epoch": 16.953229398663698,
      "loss": 0.37502074241638184,
      "loss_ce": 8.180327131412923e-05,
      "loss_iou": 0.171875,
      "loss_num": 0.00628662109375,
      "loss_xval": 0.375,
      "num_input_tokens_seen": 426537756,
      "step": 7612
    },
    {
      "epoch": 16.955456570155903,
      "grad_norm": 17.119190216064453,
      "learning_rate": 1e-06,
      "loss": 0.3499,
      "num_input_tokens_seen": 426592748,
      "step": 7613
    },
    {
      "epoch": 16.955456570155903,
      "loss": 0.3910788893699646,
      "loss_ce": 8.76820704434067e-05,
      "loss_iou": 0.158203125,
      "loss_num": 0.0146484375,
      "loss_xval": 0.390625,
      "num_input_tokens_seen": 426592748,
      "step": 7613
    },
    {
      "epoch": 16.957683741648108,
      "grad_norm": 19.266357421875,
      "learning_rate": 1e-06,
      "loss": 0.3833,
      "num_input_tokens_seen": 426648636,
      "step": 7614
    },
    {
      "epoch": 16.957683741648108,
      "loss": 0.27193689346313477,
      "loss_ce": 8.629226795164868e-05,
      "loss_iou": 0.12353515625,
      "loss_num": 0.004852294921875,
      "loss_xval": 0.271484375,
      "num_input_tokens_seen": 426648636,
      "step": 7614
    },
    {
      "epoch": 16.959910913140313,
      "grad_norm": 14.205963134765625,
      "learning_rate": 1e-06,
      "loss": 0.3386,
      "num_input_tokens_seen": 426706392,
      "step": 7615
    },
    {
      "epoch": 16.959910913140313,
      "loss": 0.3684968948364258,
      "loss_ce": 8.871590398484841e-05,
      "loss_iou": 0.166015625,
      "loss_num": 0.00726318359375,
      "loss_xval": 0.369140625,
      "num_input_tokens_seen": 426706392,
      "step": 7615
    },
    {
      "epoch": 16.962138084632517,
      "grad_norm": 31.090105056762695,
      "learning_rate": 1e-06,
      "loss": 0.4142,
      "num_input_tokens_seen": 426762144,
      "step": 7616
    },
    {
      "epoch": 16.962138084632517,
      "loss": 0.2914935052394867,
      "loss_ce": 8.114362572086975e-05,
      "loss_iou": 0.11962890625,
      "loss_num": 0.0106201171875,
      "loss_xval": 0.291015625,
      "num_input_tokens_seen": 426762144,
      "step": 7616
    },
    {
      "epoch": 16.964365256124722,
      "grad_norm": 16.874919891357422,
      "learning_rate": 1e-06,
      "loss": 0.3053,
      "num_input_tokens_seen": 426820804,
      "step": 7617
    },
    {
      "epoch": 16.964365256124722,
      "loss": 0.3068510890007019,
      "loss_ce": 8.837327914079651e-05,
      "loss_iou": 0.140625,
      "loss_num": 0.00506591796875,
      "loss_xval": 0.306640625,
      "num_input_tokens_seen": 426820804,
      "step": 7617
    },
    {
      "epoch": 16.966592427616927,
      "grad_norm": 21.651851654052734,
      "learning_rate": 1e-06,
      "loss": 0.6437,
      "num_input_tokens_seen": 426878344,
      "step": 7618
    },
    {
      "epoch": 16.966592427616927,
      "loss": 0.6224051713943481,
      "loss_ce": 9.070623491425067e-05,
      "loss_iou": 0.26953125,
      "loss_num": 0.0164794921875,
      "loss_xval": 0.62109375,
      "num_input_tokens_seen": 426878344,
      "step": 7618
    },
    {
      "epoch": 16.968819599109132,
      "grad_norm": 15.368985176086426,
      "learning_rate": 1e-06,
      "loss": 0.4305,
      "num_input_tokens_seen": 426936372,
      "step": 7619
    },
    {
      "epoch": 16.968819599109132,
      "loss": 0.4007796049118042,
      "loss_ce": 8.378856000490487e-05,
      "loss_iou": 0.162109375,
      "loss_num": 0.01544189453125,
      "loss_xval": 0.400390625,
      "num_input_tokens_seen": 426936372,
      "step": 7619
    },
    {
      "epoch": 16.971046770601337,
      "grad_norm": 16.223318099975586,
      "learning_rate": 1e-06,
      "loss": 0.2849,
      "num_input_tokens_seen": 426990840,
      "step": 7620
    },
    {
      "epoch": 16.971046770601337,
      "loss": 0.3703581690788269,
      "loss_ce": 0.00017992404173128307,
      "loss_iou": 0.16796875,
      "loss_num": 0.0069580078125,
      "loss_xval": 0.37109375,
      "num_input_tokens_seen": 426990840,
      "step": 7620
    },
    {
      "epoch": 16.97327394209354,
      "grad_norm": 42.12100601196289,
      "learning_rate": 1e-06,
      "loss": 0.4774,
      "num_input_tokens_seen": 427049316,
      "step": 7621
    },
    {
      "epoch": 16.97327394209354,
      "loss": 0.4062255620956421,
      "loss_ce": 9.765510185388848e-05,
      "loss_iou": 0.1767578125,
      "loss_num": 0.010498046875,
      "loss_xval": 0.40625,
      "num_input_tokens_seen": 427049316,
      "step": 7621
    },
    {
      "epoch": 16.975501113585747,
      "grad_norm": 15.246018409729004,
      "learning_rate": 1e-06,
      "loss": 0.4491,
      "num_input_tokens_seen": 427105040,
      "step": 7622
    },
    {
      "epoch": 16.975501113585747,
      "loss": 0.4461430311203003,
      "loss_ce": 9.8112752311863e-05,
      "loss_iou": 0.1982421875,
      "loss_num": 0.01007080078125,
      "loss_xval": 0.4453125,
      "num_input_tokens_seen": 427105040,
      "step": 7622
    },
    {
      "epoch": 16.97772828507795,
      "grad_norm": 74.06937408447266,
      "learning_rate": 1e-06,
      "loss": 0.5622,
      "num_input_tokens_seen": 427159736,
      "step": 7623
    },
    {
      "epoch": 16.97772828507795,
      "loss": 0.5899186730384827,
      "loss_ce": 7.488045957870781e-05,
      "loss_iou": 0.259765625,
      "loss_num": 0.0140380859375,
      "loss_xval": 0.58984375,
      "num_input_tokens_seen": 427159736,
      "step": 7623
    },
    {
      "epoch": 16.979955456570156,
      "grad_norm": 17.075176239013672,
      "learning_rate": 1e-06,
      "loss": 0.4932,
      "num_input_tokens_seen": 427216528,
      "step": 7624
    },
    {
      "epoch": 16.979955456570156,
      "loss": 0.5001233816146851,
      "loss_ce": 0.00012338865781202912,
      "loss_iou": 0.234375,
      "loss_num": 0.00616455078125,
      "loss_xval": 0.5,
      "num_input_tokens_seen": 427216528,
      "step": 7624
    },
    {
      "epoch": 16.98218262806236,
      "grad_norm": 12.647735595703125,
      "learning_rate": 1e-06,
      "loss": 0.3257,
      "num_input_tokens_seen": 427274456,
      "step": 7625
    },
    {
      "epoch": 16.98218262806236,
      "loss": 0.2577294707298279,
      "loss_ce": 0.00010009224934037775,
      "loss_iou": 0.1142578125,
      "loss_num": 0.005889892578125,
      "loss_xval": 0.2578125,
      "num_input_tokens_seen": 427274456,
      "step": 7625
    },
    {
      "epoch": 16.984409799554566,
      "grad_norm": 14.19373607635498,
      "learning_rate": 1e-06,
      "loss": 0.5021,
      "num_input_tokens_seen": 427328280,
      "step": 7626
    },
    {
      "epoch": 16.984409799554566,
      "loss": 0.5511077642440796,
      "loss_ce": 8.234484994318336e-05,
      "loss_iou": 0.2314453125,
      "loss_num": 0.0174560546875,
      "loss_xval": 0.55078125,
      "num_input_tokens_seen": 427328280,
      "step": 7626
    },
    {
      "epoch": 16.98663697104677,
      "grad_norm": 14.599117279052734,
      "learning_rate": 1e-06,
      "loss": 0.4076,
      "num_input_tokens_seen": 427384024,
      "step": 7627
    },
    {
      "epoch": 16.98663697104677,
      "loss": 0.40987348556518555,
      "loss_ce": 8.344671368831769e-05,
      "loss_iou": 0.1845703125,
      "loss_num": 0.00811767578125,
      "loss_xval": 0.41015625,
      "num_input_tokens_seen": 427384024,
      "step": 7627
    },
    {
      "epoch": 16.988864142538976,
      "grad_norm": 34.7947998046875,
      "learning_rate": 1e-06,
      "loss": 0.391,
      "num_input_tokens_seen": 427439656,
      "step": 7628
    },
    {
      "epoch": 16.988864142538976,
      "loss": 0.4776158928871155,
      "loss_ce": 7.684988668188453e-05,
      "loss_iou": 0.2138671875,
      "loss_num": 0.00994873046875,
      "loss_xval": 0.4765625,
      "num_input_tokens_seen": 427439656,
      "step": 7628
    },
    {
      "epoch": 16.99109131403118,
      "grad_norm": 14.100838661193848,
      "learning_rate": 1e-06,
      "loss": 0.3771,
      "num_input_tokens_seen": 427495292,
      "step": 7629
    },
    {
      "epoch": 16.99109131403118,
      "loss": 0.41444265842437744,
      "loss_ce": 0.000105498475022614,
      "loss_iou": 0.1845703125,
      "loss_num": 0.0089111328125,
      "loss_xval": 0.4140625,
      "num_input_tokens_seen": 427495292,
      "step": 7629
    },
    {
      "epoch": 16.993318485523385,
      "grad_norm": 25.862396240234375,
      "learning_rate": 1e-06,
      "loss": 0.3754,
      "num_input_tokens_seen": 427552112,
      "step": 7630
    },
    {
      "epoch": 16.993318485523385,
      "loss": 0.4190499782562256,
      "loss_ce": 0.00010467211541254073,
      "loss_iou": 0.1884765625,
      "loss_num": 0.00848388671875,
      "loss_xval": 0.41796875,
      "num_input_tokens_seen": 427552112,
      "step": 7630
    },
    {
      "epoch": 16.99554565701559,
      "grad_norm": 16.033910751342773,
      "learning_rate": 1e-06,
      "loss": 0.2646,
      "num_input_tokens_seen": 427609520,
      "step": 7631
    },
    {
      "epoch": 16.99554565701559,
      "loss": 0.28566277027130127,
      "loss_ce": 7.926442776806653e-05,
      "loss_iou": 0.1240234375,
      "loss_num": 0.007354736328125,
      "loss_xval": 0.28515625,
      "num_input_tokens_seen": 427609520,
      "step": 7631
    },
    {
      "epoch": 16.997772828507795,
      "grad_norm": 20.488983154296875,
      "learning_rate": 1e-06,
      "loss": 0.4234,
      "num_input_tokens_seen": 427661616,
      "step": 7632
    },
    {
      "epoch": 16.997772828507795,
      "loss": 0.40573614835739136,
      "loss_ce": 9.650552237872034e-05,
      "loss_iou": 0.1494140625,
      "loss_num": 0.021484375,
      "loss_xval": 0.40625,
      "num_input_tokens_seen": 427661616,
      "step": 7632
    },
    {
      "epoch": 17.0,
      "grad_norm": 23.362516403198242,
      "learning_rate": 1e-06,
      "loss": 0.3575,
      "num_input_tokens_seen": 427716580,
      "step": 7633
    },
    {
      "epoch": 17.0,
      "loss": 0.32436951994895935,
      "loss_ce": 8.972680370789021e-05,
      "loss_iou": 0.1484375,
      "loss_num": 0.0052490234375,
      "loss_xval": 0.32421875,
      "num_input_tokens_seen": 427716580,
      "step": 7633
    },
    {
      "epoch": 17.002227171492205,
      "grad_norm": 17.92319107055664,
      "learning_rate": 1e-06,
      "loss": 0.4558,
      "num_input_tokens_seen": 427773004,
      "step": 7634
    },
    {
      "epoch": 17.002227171492205,
      "loss": 0.4817723333835602,
      "loss_ce": 8.288519165944308e-05,
      "loss_iou": 0.2080078125,
      "loss_num": 0.012939453125,
      "loss_xval": 0.482421875,
      "num_input_tokens_seen": 427773004,
      "step": 7634
    },
    {
      "epoch": 17.00445434298441,
      "grad_norm": 26.28542709350586,
      "learning_rate": 1e-06,
      "loss": 0.4383,
      "num_input_tokens_seen": 427828620,
      "step": 7635
    },
    {
      "epoch": 17.00445434298441,
      "loss": 0.4154224097728729,
      "loss_ce": 7.814820855855942e-05,
      "loss_iou": 0.185546875,
      "loss_num": 0.00885009765625,
      "loss_xval": 0.416015625,
      "num_input_tokens_seen": 427828620,
      "step": 7635
    },
    {
      "epoch": 17.006681514476615,
      "grad_norm": 18.24226188659668,
      "learning_rate": 1e-06,
      "loss": 0.3023,
      "num_input_tokens_seen": 427885192,
      "step": 7636
    },
    {
      "epoch": 17.006681514476615,
      "loss": 0.2849215865135193,
      "loss_ce": 7.052010187180713e-05,
      "loss_iou": 0.130859375,
      "loss_num": 0.0045166015625,
      "loss_xval": 0.28515625,
      "num_input_tokens_seen": 427885192,
      "step": 7636
    },
    {
      "epoch": 17.00890868596882,
      "grad_norm": 30.52312660217285,
      "learning_rate": 1e-06,
      "loss": 0.4783,
      "num_input_tokens_seen": 427939364,
      "step": 7637
    },
    {
      "epoch": 17.00890868596882,
      "loss": 0.46554726362228394,
      "loss_ce": 9.315512579632923e-05,
      "loss_iou": 0.19921875,
      "loss_num": 0.01348876953125,
      "loss_xval": 0.46484375,
      "num_input_tokens_seen": 427939364,
      "step": 7637
    },
    {
      "epoch": 17.011135857461024,
      "grad_norm": 66.4842300415039,
      "learning_rate": 1e-06,
      "loss": 0.3289,
      "num_input_tokens_seen": 427998324,
      "step": 7638
    },
    {
      "epoch": 17.011135857461024,
      "loss": 0.23030278086662292,
      "loss_ce": 7.816310971975327e-05,
      "loss_iou": 0.1044921875,
      "loss_num": 0.00433349609375,
      "loss_xval": 0.23046875,
      "num_input_tokens_seen": 427998324,
      "step": 7638
    },
    {
      "epoch": 17.01336302895323,
      "grad_norm": 20.621551513671875,
      "learning_rate": 1e-06,
      "loss": 0.4443,
      "num_input_tokens_seen": 428052968,
      "step": 7639
    },
    {
      "epoch": 17.01336302895323,
      "loss": 0.502518355846405,
      "loss_ce": 7.69448815844953e-05,
      "loss_iou": 0.1943359375,
      "loss_num": 0.022705078125,
      "loss_xval": 0.50390625,
      "num_input_tokens_seen": 428052968,
      "step": 7639
    },
    {
      "epoch": 17.015590200445434,
      "grad_norm": 16.662851333618164,
      "learning_rate": 1e-06,
      "loss": 0.6814,
      "num_input_tokens_seen": 428110384,
      "step": 7640
    },
    {
      "epoch": 17.015590200445434,
      "loss": 0.931036114692688,
      "loss_ce": 0.00015838468971196562,
      "loss_iou": 0.349609375,
      "loss_num": 0.04638671875,
      "loss_xval": 0.9296875,
      "num_input_tokens_seen": 428110384,
      "step": 7640
    },
    {
      "epoch": 17.01781737193764,
      "grad_norm": 22.56868553161621,
      "learning_rate": 1e-06,
      "loss": 0.2991,
      "num_input_tokens_seen": 428166924,
      "step": 7641
    },
    {
      "epoch": 17.01781737193764,
      "loss": 0.22560018301010132,
      "loss_ce": 7.529689173679799e-05,
      "loss_iou": 0.09423828125,
      "loss_num": 0.0074462890625,
      "loss_xval": 0.2255859375,
      "num_input_tokens_seen": 428166924,
      "step": 7641
    },
    {
      "epoch": 17.020044543429844,
      "grad_norm": 19.790267944335938,
      "learning_rate": 1e-06,
      "loss": 0.4237,
      "num_input_tokens_seen": 428223112,
      "step": 7642
    },
    {
      "epoch": 17.020044543429844,
      "loss": 0.424101322889328,
      "loss_ce": 0.00015111861284822226,
      "loss_iou": 0.185546875,
      "loss_num": 0.0107421875,
      "loss_xval": 0.423828125,
      "num_input_tokens_seen": 428223112,
      "step": 7642
    },
    {
      "epoch": 17.02227171492205,
      "grad_norm": 22.081754684448242,
      "learning_rate": 1e-06,
      "loss": 0.5015,
      "num_input_tokens_seen": 428279696,
      "step": 7643
    },
    {
      "epoch": 17.02227171492205,
      "loss": 0.5504837036132812,
      "loss_ce": 0.00019074320152867585,
      "loss_iou": 0.2265625,
      "loss_num": 0.0196533203125,
      "loss_xval": 0.55078125,
      "num_input_tokens_seen": 428279696,
      "step": 7643
    },
    {
      "epoch": 17.024498886414253,
      "grad_norm": 12.306164741516113,
      "learning_rate": 1e-06,
      "loss": 0.4179,
      "num_input_tokens_seen": 428335044,
      "step": 7644
    },
    {
      "epoch": 17.024498886414253,
      "loss": 0.5076683163642883,
      "loss_ce": 9.997590677812696e-05,
      "loss_iou": 0.2392578125,
      "loss_num": 0.005889892578125,
      "loss_xval": 0.5078125,
      "num_input_tokens_seen": 428335044,
      "step": 7644
    },
    {
      "epoch": 17.02672605790646,
      "grad_norm": 21.696956634521484,
      "learning_rate": 1e-06,
      "loss": 0.48,
      "num_input_tokens_seen": 428390008,
      "step": 7645
    },
    {
      "epoch": 17.02672605790646,
      "loss": 0.4708196818828583,
      "loss_ce": 0.00011655631533358246,
      "loss_iou": 0.1982421875,
      "loss_num": 0.01470947265625,
      "loss_xval": 0.470703125,
      "num_input_tokens_seen": 428390008,
      "step": 7645
    },
    {
      "epoch": 17.028953229398663,
      "grad_norm": 36.35521697998047,
      "learning_rate": 1e-06,
      "loss": 0.3621,
      "num_input_tokens_seen": 428447080,
      "step": 7646
    },
    {
      "epoch": 17.028953229398663,
      "loss": 0.2672528028488159,
      "loss_ce": 0.00016677916573826224,
      "loss_iou": 0.10546875,
      "loss_num": 0.0111083984375,
      "loss_xval": 0.267578125,
      "num_input_tokens_seen": 428447080,
      "step": 7646
    },
    {
      "epoch": 17.031180400890868,
      "grad_norm": 21.043344497680664,
      "learning_rate": 1e-06,
      "loss": 0.3144,
      "num_input_tokens_seen": 428503564,
      "step": 7647
    },
    {
      "epoch": 17.031180400890868,
      "loss": 0.36540961265563965,
      "loss_ce": 8.367877308046445e-05,
      "loss_iou": 0.1728515625,
      "loss_num": 0.003997802734375,
      "loss_xval": 0.365234375,
      "num_input_tokens_seen": 428503564,
      "step": 7647
    },
    {
      "epoch": 17.033407572383073,
      "grad_norm": 24.031978607177734,
      "learning_rate": 1e-06,
      "loss": 0.334,
      "num_input_tokens_seen": 428559008,
      "step": 7648
    },
    {
      "epoch": 17.033407572383073,
      "loss": 0.2920740842819214,
      "loss_ce": 8.189551590476185e-05,
      "loss_iou": 0.1220703125,
      "loss_num": 0.00958251953125,
      "loss_xval": 0.29296875,
      "num_input_tokens_seen": 428559008,
      "step": 7648
    },
    {
      "epoch": 17.035634743875278,
      "grad_norm": 21.48931121826172,
      "learning_rate": 1e-06,
      "loss": 0.4046,
      "num_input_tokens_seen": 428613808,
      "step": 7649
    },
    {
      "epoch": 17.035634743875278,
      "loss": 0.21085509657859802,
      "loss_ce": 8.544648881070316e-05,
      "loss_iou": 0.0947265625,
      "loss_num": 0.0042724609375,
      "loss_xval": 0.2109375,
      "num_input_tokens_seen": 428613808,
      "step": 7649
    },
    {
      "epoch": 17.037861915367483,
      "grad_norm": 13.954534530639648,
      "learning_rate": 1e-06,
      "loss": 0.4201,
      "num_input_tokens_seen": 428672352,
      "step": 7650
    },
    {
      "epoch": 17.037861915367483,
      "loss": 0.25243866443634033,
      "loss_ce": 8.880048699211329e-05,
      "loss_iou": 0.10986328125,
      "loss_num": 0.006561279296875,
      "loss_xval": 0.251953125,
      "num_input_tokens_seen": 428672352,
      "step": 7650
    },
    {
      "epoch": 17.040089086859687,
      "grad_norm": 16.805583953857422,
      "learning_rate": 1e-06,
      "loss": 0.4008,
      "num_input_tokens_seen": 428726452,
      "step": 7651
    },
    {
      "epoch": 17.040089086859687,
      "loss": 0.3927674889564514,
      "loss_ce": 6.728603329975158e-05,
      "loss_iou": 0.1630859375,
      "loss_num": 0.01318359375,
      "loss_xval": 0.392578125,
      "num_input_tokens_seen": 428726452,
      "step": 7651
    },
    {
      "epoch": 17.042316258351892,
      "grad_norm": 22.515275955200195,
      "learning_rate": 1e-06,
      "loss": 0.3941,
      "num_input_tokens_seen": 428783940,
      "step": 7652
    },
    {
      "epoch": 17.042316258351892,
      "loss": 0.4546872675418854,
      "loss_ce": 9.743101691128686e-05,
      "loss_iou": 0.19921875,
      "loss_num": 0.010986328125,
      "loss_xval": 0.455078125,
      "num_input_tokens_seen": 428783940,
      "step": 7652
    },
    {
      "epoch": 17.044543429844097,
      "grad_norm": 23.46599769592285,
      "learning_rate": 1e-06,
      "loss": 0.3476,
      "num_input_tokens_seen": 428839472,
      "step": 7653
    },
    {
      "epoch": 17.044543429844097,
      "loss": 0.33765748143196106,
      "loss_ce": 7.202455890364945e-05,
      "loss_iou": 0.1328125,
      "loss_num": 0.01458740234375,
      "loss_xval": 0.337890625,
      "num_input_tokens_seen": 428839472,
      "step": 7653
    },
    {
      "epoch": 17.046770601336302,
      "grad_norm": 21.571443557739258,
      "learning_rate": 1e-06,
      "loss": 0.3301,
      "num_input_tokens_seen": 428896740,
      "step": 7654
    },
    {
      "epoch": 17.046770601336302,
      "loss": 0.3378799855709076,
      "loss_ce": 0.00011143009760417044,
      "loss_iou": 0.1474609375,
      "loss_num": 0.00836181640625,
      "loss_xval": 0.337890625,
      "num_input_tokens_seen": 428896740,
      "step": 7654
    },
    {
      "epoch": 17.048997772828507,
      "grad_norm": 20.132381439208984,
      "learning_rate": 1e-06,
      "loss": 0.2645,
      "num_input_tokens_seen": 428951072,
      "step": 7655
    },
    {
      "epoch": 17.048997772828507,
      "loss": 0.24837635457515717,
      "loss_ce": 8.534367952961475e-05,
      "loss_iou": 0.1103515625,
      "loss_num": 0.00543212890625,
      "loss_xval": 0.248046875,
      "num_input_tokens_seen": 428951072,
      "step": 7655
    },
    {
      "epoch": 17.051224944320712,
      "grad_norm": 20.21780014038086,
      "learning_rate": 1e-06,
      "loss": 0.4544,
      "num_input_tokens_seen": 429007504,
      "step": 7656
    },
    {
      "epoch": 17.051224944320712,
      "loss": 0.4101518392562866,
      "loss_ce": 0.00011766105308197439,
      "loss_iou": 0.1904296875,
      "loss_num": 0.005859375,
      "loss_xval": 0.41015625,
      "num_input_tokens_seen": 429007504,
      "step": 7656
    },
    {
      "epoch": 17.053452115812917,
      "grad_norm": 17.552833557128906,
      "learning_rate": 1e-06,
      "loss": 0.4551,
      "num_input_tokens_seen": 429061384,
      "step": 7657
    },
    {
      "epoch": 17.053452115812917,
      "loss": 0.42342638969421387,
      "loss_ce": 8.654448902234435e-05,
      "loss_iou": 0.1943359375,
      "loss_num": 0.0069580078125,
      "loss_xval": 0.423828125,
      "num_input_tokens_seen": 429061384,
      "step": 7657
    },
    {
      "epoch": 17.05567928730512,
      "grad_norm": 16.712745666503906,
      "learning_rate": 1e-06,
      "loss": 0.4366,
      "num_input_tokens_seen": 429117780,
      "step": 7658
    },
    {
      "epoch": 17.05567928730512,
      "loss": 0.5296164155006409,
      "loss_ce": 7.536636258009821e-05,
      "loss_iou": 0.2412109375,
      "loss_num": 0.0091552734375,
      "loss_xval": 0.53125,
      "num_input_tokens_seen": 429117780,
      "step": 7658
    },
    {
      "epoch": 17.057906458797326,
      "grad_norm": 17.775182723999023,
      "learning_rate": 1e-06,
      "loss": 0.4586,
      "num_input_tokens_seen": 429174152,
      "step": 7659
    },
    {
      "epoch": 17.057906458797326,
      "loss": 0.34088775515556335,
      "loss_ce": 6.743887206539512e-05,
      "loss_iou": 0.146484375,
      "loss_num": 0.00946044921875,
      "loss_xval": 0.33984375,
      "num_input_tokens_seen": 429174152,
      "step": 7659
    },
    {
      "epoch": 17.06013363028953,
      "grad_norm": 18.0156192779541,
      "learning_rate": 1e-06,
      "loss": 0.3511,
      "num_input_tokens_seen": 429231128,
      "step": 7660
    },
    {
      "epoch": 17.06013363028953,
      "loss": 0.3127681612968445,
      "loss_ce": 8.506246376782656e-05,
      "loss_iou": 0.130859375,
      "loss_num": 0.01025390625,
      "loss_xval": 0.3125,
      "num_input_tokens_seen": 429231128,
      "step": 7660
    },
    {
      "epoch": 17.062360801781736,
      "grad_norm": 52.67565155029297,
      "learning_rate": 1e-06,
      "loss": 0.6399,
      "num_input_tokens_seen": 429284628,
      "step": 7661
    },
    {
      "epoch": 17.062360801781736,
      "loss": 0.8818075060844421,
      "loss_ce": 9.362171840621158e-05,
      "loss_iou": 0.35546875,
      "loss_num": 0.03466796875,
      "loss_xval": 0.8828125,
      "num_input_tokens_seen": 429284628,
      "step": 7661
    },
    {
      "epoch": 17.06458797327394,
      "grad_norm": 27.630023956298828,
      "learning_rate": 1e-06,
      "loss": 0.4078,
      "num_input_tokens_seen": 429338128,
      "step": 7662
    },
    {
      "epoch": 17.06458797327394,
      "loss": 0.40474581718444824,
      "loss_ce": 8.270953549072146e-05,
      "loss_iou": 0.1875,
      "loss_num": 0.005828857421875,
      "loss_xval": 0.404296875,
      "num_input_tokens_seen": 429338128,
      "step": 7662
    },
    {
      "epoch": 17.066815144766146,
      "grad_norm": 22.74899673461914,
      "learning_rate": 1e-06,
      "loss": 0.5521,
      "num_input_tokens_seen": 429390620,
      "step": 7663
    },
    {
      "epoch": 17.066815144766146,
      "loss": 0.45908236503601074,
      "loss_ce": 9.800391853787005e-05,
      "loss_iou": 0.19921875,
      "loss_num": 0.011962890625,
      "loss_xval": 0.458984375,
      "num_input_tokens_seen": 429390620,
      "step": 7663
    },
    {
      "epoch": 17.06904231625835,
      "grad_norm": 16.609214782714844,
      "learning_rate": 1e-06,
      "loss": 0.4099,
      "num_input_tokens_seen": 429445244,
      "step": 7664
    },
    {
      "epoch": 17.06904231625835,
      "loss": 0.3943771719932556,
      "loss_ce": 9.003834566101432e-05,
      "loss_iou": 0.162109375,
      "loss_num": 0.01409912109375,
      "loss_xval": 0.39453125,
      "num_input_tokens_seen": 429445244,
      "step": 7664
    },
    {
      "epoch": 17.071269487750556,
      "grad_norm": 21.4134578704834,
      "learning_rate": 1e-06,
      "loss": 0.5788,
      "num_input_tokens_seen": 429502620,
      "step": 7665
    },
    {
      "epoch": 17.071269487750556,
      "loss": 0.5071187615394592,
      "loss_ce": 9.971446706913412e-05,
      "loss_iou": 0.2236328125,
      "loss_num": 0.0118408203125,
      "loss_xval": 0.5078125,
      "num_input_tokens_seen": 429502620,
      "step": 7665
    },
    {
      "epoch": 17.07349665924276,
      "grad_norm": 18.723894119262695,
      "learning_rate": 1e-06,
      "loss": 0.4943,
      "num_input_tokens_seen": 429562224,
      "step": 7666
    },
    {
      "epoch": 17.07349665924276,
      "loss": 0.47562456130981445,
      "loss_ce": 0.0001607102749403566,
      "loss_iou": 0.2080078125,
      "loss_num": 0.011962890625,
      "loss_xval": 0.474609375,
      "num_input_tokens_seen": 429562224,
      "step": 7666
    },
    {
      "epoch": 17.075723830734965,
      "grad_norm": 19.977685928344727,
      "learning_rate": 1e-06,
      "loss": 0.3345,
      "num_input_tokens_seen": 429620100,
      "step": 7667
    },
    {
      "epoch": 17.075723830734965,
      "loss": 0.31252235174179077,
      "loss_ce": 8.34053716971539e-05,
      "loss_iou": 0.134765625,
      "loss_num": 0.00836181640625,
      "loss_xval": 0.3125,
      "num_input_tokens_seen": 429620100,
      "step": 7667
    },
    {
      "epoch": 17.07795100222717,
      "grad_norm": 19.599905014038086,
      "learning_rate": 1e-06,
      "loss": 0.4729,
      "num_input_tokens_seen": 429677708,
      "step": 7668
    },
    {
      "epoch": 17.07795100222717,
      "loss": 0.6072638034820557,
      "loss_ce": 8.606135088484734e-05,
      "loss_iou": 0.28125,
      "loss_num": 0.0091552734375,
      "loss_xval": 0.60546875,
      "num_input_tokens_seen": 429677708,
      "step": 7668
    },
    {
      "epoch": 17.080178173719375,
      "grad_norm": 14.583081245422363,
      "learning_rate": 1e-06,
      "loss": 0.4426,
      "num_input_tokens_seen": 429735204,
      "step": 7669
    },
    {
      "epoch": 17.080178173719375,
      "loss": 0.43759340047836304,
      "loss_ce": 9.339496318716556e-05,
      "loss_iou": 0.2060546875,
      "loss_num": 0.005096435546875,
      "loss_xval": 0.4375,
      "num_input_tokens_seen": 429735204,
      "step": 7669
    },
    {
      "epoch": 17.08240534521158,
      "grad_norm": 34.837467193603516,
      "learning_rate": 1e-06,
      "loss": 0.4082,
      "num_input_tokens_seen": 429789724,
      "step": 7670
    },
    {
      "epoch": 17.08240534521158,
      "loss": 0.3650968670845032,
      "loss_ce": 0.00010662678687367588,
      "loss_iou": 0.1689453125,
      "loss_num": 0.0054931640625,
      "loss_xval": 0.365234375,
      "num_input_tokens_seen": 429789724,
      "step": 7670
    },
    {
      "epoch": 17.084632516703785,
      "grad_norm": 16.056995391845703,
      "learning_rate": 1e-06,
      "loss": 0.332,
      "num_input_tokens_seen": 429846636,
      "step": 7671
    },
    {
      "epoch": 17.084632516703785,
      "loss": 0.30904102325439453,
      "loss_ce": 8.104251901386306e-05,
      "loss_iou": 0.13671875,
      "loss_num": 0.006988525390625,
      "loss_xval": 0.30859375,
      "num_input_tokens_seen": 429846636,
      "step": 7671
    },
    {
      "epoch": 17.08685968819599,
      "grad_norm": 19.151451110839844,
      "learning_rate": 1e-06,
      "loss": 0.4416,
      "num_input_tokens_seen": 429905284,
      "step": 7672
    },
    {
      "epoch": 17.08685968819599,
      "loss": 0.47447669506073,
      "loss_ce": 0.0001114873739425093,
      "loss_iou": 0.224609375,
      "loss_num": 0.00518798828125,
      "loss_xval": 0.474609375,
      "num_input_tokens_seen": 429905284,
      "step": 7672
    },
    {
      "epoch": 17.089086859688194,
      "grad_norm": 14.329705238342285,
      "learning_rate": 1e-06,
      "loss": 0.6205,
      "num_input_tokens_seen": 429960820,
      "step": 7673
    },
    {
      "epoch": 17.089086859688194,
      "loss": 0.6159265041351318,
      "loss_ce": 8.180980512406677e-05,
      "loss_iou": 0.255859375,
      "loss_num": 0.020751953125,
      "loss_xval": 0.6171875,
      "num_input_tokens_seen": 429960820,
      "step": 7673
    },
    {
      "epoch": 17.0913140311804,
      "grad_norm": 12.72126579284668,
      "learning_rate": 1e-06,
      "loss": 0.2597,
      "num_input_tokens_seen": 430014656,
      "step": 7674
    },
    {
      "epoch": 17.0913140311804,
      "loss": 0.32735052704811096,
      "loss_ce": 8.001940295798704e-05,
      "loss_iou": 0.123046875,
      "loss_num": 0.0162353515625,
      "loss_xval": 0.328125,
      "num_input_tokens_seen": 430014656,
      "step": 7674
    },
    {
      "epoch": 17.093541202672604,
      "grad_norm": 15.150796890258789,
      "learning_rate": 1e-06,
      "loss": 0.414,
      "num_input_tokens_seen": 430071736,
      "step": 7675
    },
    {
      "epoch": 17.093541202672604,
      "loss": 0.316550612449646,
      "loss_ce": 8.332452125614509e-05,
      "loss_iou": 0.146484375,
      "loss_num": 0.00482177734375,
      "loss_xval": 0.31640625,
      "num_input_tokens_seen": 430071736,
      "step": 7675
    },
    {
      "epoch": 17.09576837416481,
      "grad_norm": 22.740732192993164,
      "learning_rate": 1e-06,
      "loss": 0.4941,
      "num_input_tokens_seen": 430127552,
      "step": 7676
    },
    {
      "epoch": 17.09576837416481,
      "loss": 0.688327968120575,
      "loss_ce": 9.552988194627687e-05,
      "loss_iou": 0.283203125,
      "loss_num": 0.024169921875,
      "loss_xval": 0.6875,
      "num_input_tokens_seen": 430127552,
      "step": 7676
    },
    {
      "epoch": 17.097995545657014,
      "grad_norm": 15.583344459533691,
      "learning_rate": 1e-06,
      "loss": 0.3403,
      "num_input_tokens_seen": 430183984,
      "step": 7677
    },
    {
      "epoch": 17.097995545657014,
      "loss": 0.2791343033313751,
      "loss_ce": 8.15772291389294e-05,
      "loss_iou": 0.11865234375,
      "loss_num": 0.00823974609375,
      "loss_xval": 0.279296875,
      "num_input_tokens_seen": 430183984,
      "step": 7677
    },
    {
      "epoch": 17.100222717149222,
      "grad_norm": 15.863219261169434,
      "learning_rate": 1e-06,
      "loss": 0.506,
      "num_input_tokens_seen": 430239088,
      "step": 7678
    },
    {
      "epoch": 17.100222717149222,
      "loss": 0.4502248167991638,
      "loss_ce": 9.05133638298139e-05,
      "loss_iou": 0.197265625,
      "loss_num": 0.010986328125,
      "loss_xval": 0.44921875,
      "num_input_tokens_seen": 430239088,
      "step": 7678
    },
    {
      "epoch": 17.102449888641427,
      "grad_norm": 17.146371841430664,
      "learning_rate": 1e-06,
      "loss": 0.4628,
      "num_input_tokens_seen": 430293988,
      "step": 7679
    },
    {
      "epoch": 17.102449888641427,
      "loss": 0.5290303826332092,
      "loss_ce": 9.971238614525646e-05,
      "loss_iou": 0.20703125,
      "loss_num": 0.02294921875,
      "loss_xval": 0.52734375,
      "num_input_tokens_seen": 430293988,
      "step": 7679
    },
    {
      "epoch": 17.104677060133632,
      "grad_norm": 21.910921096801758,
      "learning_rate": 1e-06,
      "loss": 0.4983,
      "num_input_tokens_seen": 430350332,
      "step": 7680
    },
    {
      "epoch": 17.104677060133632,
      "loss": 0.520658552646637,
      "loss_ce": 8.970967610366642e-05,
      "loss_iou": 0.203125,
      "loss_num": 0.023193359375,
      "loss_xval": 0.51953125,
      "num_input_tokens_seen": 430350332,
      "step": 7680
    },
    {
      "epoch": 17.106904231625837,
      "grad_norm": 21.317134857177734,
      "learning_rate": 1e-06,
      "loss": 0.4361,
      "num_input_tokens_seen": 430407436,
      "step": 7681
    },
    {
      "epoch": 17.106904231625837,
      "loss": 0.5610302090644836,
      "loss_ce": 0.00011710502440109849,
      "loss_iou": 0.234375,
      "loss_num": 0.018310546875,
      "loss_xval": 0.5625,
      "num_input_tokens_seen": 430407436,
      "step": 7681
    },
    {
      "epoch": 17.10913140311804,
      "grad_norm": 15.316725730895996,
      "learning_rate": 1e-06,
      "loss": 0.3235,
      "num_input_tokens_seen": 430464132,
      "step": 7682
    },
    {
      "epoch": 17.10913140311804,
      "loss": 0.3164796233177185,
      "loss_ce": 7.337974238907918e-05,
      "loss_iou": 0.1337890625,
      "loss_num": 0.0098876953125,
      "loss_xval": 0.31640625,
      "num_input_tokens_seen": 430464132,
      "step": 7682
    },
    {
      "epoch": 17.111358574610247,
      "grad_norm": 17.729398727416992,
      "learning_rate": 1e-06,
      "loss": 0.6469,
      "num_input_tokens_seen": 430518768,
      "step": 7683
    },
    {
      "epoch": 17.111358574610247,
      "loss": 0.49617958068847656,
      "loss_ce": 8.584219176555052e-05,
      "loss_iou": 0.216796875,
      "loss_num": 0.012451171875,
      "loss_xval": 0.49609375,
      "num_input_tokens_seen": 430518768,
      "step": 7683
    },
    {
      "epoch": 17.11358574610245,
      "grad_norm": 25.321056365966797,
      "learning_rate": 1e-06,
      "loss": 0.3942,
      "num_input_tokens_seen": 430574312,
      "step": 7684
    },
    {
      "epoch": 17.11358574610245,
      "loss": 0.2743665277957916,
      "loss_ce": 8.980404527392238e-05,
      "loss_iou": 0.1201171875,
      "loss_num": 0.006866455078125,
      "loss_xval": 0.2734375,
      "num_input_tokens_seen": 430574312,
      "step": 7684
    },
    {
      "epoch": 17.115812917594656,
      "grad_norm": 33.18670654296875,
      "learning_rate": 1e-06,
      "loss": 0.4818,
      "num_input_tokens_seen": 430629924,
      "step": 7685
    },
    {
      "epoch": 17.115812917594656,
      "loss": 0.4744425117969513,
      "loss_ce": 7.726340845692903e-05,
      "loss_iou": 0.201171875,
      "loss_num": 0.01434326171875,
      "loss_xval": 0.474609375,
      "num_input_tokens_seen": 430629924,
      "step": 7685
    },
    {
      "epoch": 17.11804008908686,
      "grad_norm": 18.0744571685791,
      "learning_rate": 1e-06,
      "loss": 0.4743,
      "num_input_tokens_seen": 430684496,
      "step": 7686
    },
    {
      "epoch": 17.11804008908686,
      "loss": 0.4138874113559723,
      "loss_ce": 6.904240581206977e-05,
      "loss_iou": 0.1875,
      "loss_num": 0.007537841796875,
      "loss_xval": 0.4140625,
      "num_input_tokens_seen": 430684496,
      "step": 7686
    },
    {
      "epoch": 17.120267260579066,
      "grad_norm": 12.516131401062012,
      "learning_rate": 1e-06,
      "loss": 0.4325,
      "num_input_tokens_seen": 430740936,
      "step": 7687
    },
    {
      "epoch": 17.120267260579066,
      "loss": 0.41224995255470276,
      "loss_ce": 0.00011005052510881796,
      "loss_iou": 0.1923828125,
      "loss_num": 0.00555419921875,
      "loss_xval": 0.412109375,
      "num_input_tokens_seen": 430740936,
      "step": 7687
    },
    {
      "epoch": 17.12249443207127,
      "grad_norm": 14.209662437438965,
      "learning_rate": 1e-06,
      "loss": 0.3043,
      "num_input_tokens_seen": 430797872,
      "step": 7688
    },
    {
      "epoch": 17.12249443207127,
      "loss": 0.36360567808151245,
      "loss_ce": 8.030241588130593e-05,
      "loss_iou": 0.1650390625,
      "loss_num": 0.00665283203125,
      "loss_xval": 0.36328125,
      "num_input_tokens_seen": 430797872,
      "step": 7688
    },
    {
      "epoch": 17.124721603563476,
      "grad_norm": 24.326807022094727,
      "learning_rate": 1e-06,
      "loss": 0.4336,
      "num_input_tokens_seen": 430856816,
      "step": 7689
    },
    {
      "epoch": 17.124721603563476,
      "loss": 0.40657997131347656,
      "loss_ce": 8.581295696785673e-05,
      "loss_iou": 0.1650390625,
      "loss_num": 0.01507568359375,
      "loss_xval": 0.40625,
      "num_input_tokens_seen": 430856816,
      "step": 7689
    },
    {
      "epoch": 17.12694877505568,
      "grad_norm": 15.567830085754395,
      "learning_rate": 1e-06,
      "loss": 0.4065,
      "num_input_tokens_seen": 430913996,
      "step": 7690
    },
    {
      "epoch": 17.12694877505568,
      "loss": 0.32644379138946533,
      "loss_ce": 8.880942186806351e-05,
      "loss_iou": 0.1474609375,
      "loss_num": 0.006256103515625,
      "loss_xval": 0.326171875,
      "num_input_tokens_seen": 430913996,
      "step": 7690
    },
    {
      "epoch": 17.129175946547885,
      "grad_norm": 27.735408782958984,
      "learning_rate": 1e-06,
      "loss": 0.3461,
      "num_input_tokens_seen": 430970364,
      "step": 7691
    },
    {
      "epoch": 17.129175946547885,
      "loss": 0.3675483167171478,
      "loss_ce": 0.00011668518709484488,
      "loss_iou": 0.1591796875,
      "loss_num": 0.00994873046875,
      "loss_xval": 0.3671875,
      "num_input_tokens_seen": 430970364,
      "step": 7691
    },
    {
      "epoch": 17.13140311804009,
      "grad_norm": 13.408978462219238,
      "learning_rate": 1e-06,
      "loss": 0.435,
      "num_input_tokens_seen": 431026824,
      "step": 7692
    },
    {
      "epoch": 17.13140311804009,
      "loss": 0.49011173844337463,
      "loss_ce": 0.00012149102985858917,
      "loss_iou": 0.2021484375,
      "loss_num": 0.0169677734375,
      "loss_xval": 0.490234375,
      "num_input_tokens_seen": 431026824,
      "step": 7692
    },
    {
      "epoch": 17.133630289532295,
      "grad_norm": 19.401525497436523,
      "learning_rate": 1e-06,
      "loss": 0.5539,
      "num_input_tokens_seen": 431082556,
      "step": 7693
    },
    {
      "epoch": 17.133630289532295,
      "loss": 0.5180162787437439,
      "loss_ce": 7.190792530309409e-05,
      "loss_iou": 0.22265625,
      "loss_num": 0.014404296875,
      "loss_xval": 0.51953125,
      "num_input_tokens_seen": 431082556,
      "step": 7693
    },
    {
      "epoch": 17.1358574610245,
      "grad_norm": 17.272777557373047,
      "learning_rate": 1e-06,
      "loss": 0.4486,
      "num_input_tokens_seen": 431140336,
      "step": 7694
    },
    {
      "epoch": 17.1358574610245,
      "loss": 0.39839935302734375,
      "loss_ce": 8.391607843805104e-05,
      "loss_iou": 0.1650390625,
      "loss_num": 0.01361083984375,
      "loss_xval": 0.3984375,
      "num_input_tokens_seen": 431140336,
      "step": 7694
    },
    {
      "epoch": 17.138084632516705,
      "grad_norm": 13.000822067260742,
      "learning_rate": 1e-06,
      "loss": 0.4472,
      "num_input_tokens_seen": 431196616,
      "step": 7695
    },
    {
      "epoch": 17.138084632516705,
      "loss": 0.39486390352249146,
      "loss_ce": 8.850420999806374e-05,
      "loss_iou": 0.1806640625,
      "loss_num": 0.00677490234375,
      "loss_xval": 0.39453125,
      "num_input_tokens_seen": 431196616,
      "step": 7695
    },
    {
      "epoch": 17.14031180400891,
      "grad_norm": 12.154135704040527,
      "learning_rate": 1e-06,
      "loss": 0.3779,
      "num_input_tokens_seen": 431253236,
      "step": 7696
    },
    {
      "epoch": 17.14031180400891,
      "loss": 0.3711104691028595,
      "loss_ce": 7.775596168357879e-05,
      "loss_iou": 0.1708984375,
      "loss_num": 0.005950927734375,
      "loss_xval": 0.37109375,
      "num_input_tokens_seen": 431253236,
      "step": 7696
    },
    {
      "epoch": 17.142538975501115,
      "grad_norm": 19.33846092224121,
      "learning_rate": 1e-06,
      "loss": 0.5077,
      "num_input_tokens_seen": 431311832,
      "step": 7697
    },
    {
      "epoch": 17.142538975501115,
      "loss": 0.5122911930084229,
      "loss_ce": 8.418020297540352e-05,
      "loss_iou": 0.2197265625,
      "loss_num": 0.0145263671875,
      "loss_xval": 0.51171875,
      "num_input_tokens_seen": 431311832,
      "step": 7697
    },
    {
      "epoch": 17.14476614699332,
      "grad_norm": 13.823887825012207,
      "learning_rate": 1e-06,
      "loss": 0.304,
      "num_input_tokens_seen": 431367724,
      "step": 7698
    },
    {
      "epoch": 17.14476614699332,
      "loss": 0.37503981590270996,
      "loss_ce": 0.00010085223766509444,
      "loss_iou": 0.16015625,
      "loss_num": 0.01080322265625,
      "loss_xval": 0.375,
      "num_input_tokens_seen": 431367724,
      "step": 7698
    },
    {
      "epoch": 17.146993318485524,
      "grad_norm": 20.612548828125,
      "learning_rate": 1e-06,
      "loss": 0.4507,
      "num_input_tokens_seen": 431423316,
      "step": 7699
    },
    {
      "epoch": 17.146993318485524,
      "loss": 0.5613565444946289,
      "loss_ce": 7.721249130554497e-05,
      "loss_iou": 0.205078125,
      "loss_num": 0.0301513671875,
      "loss_xval": 0.5625,
      "num_input_tokens_seen": 431423316,
      "step": 7699
    },
    {
      "epoch": 17.14922048997773,
      "grad_norm": 23.749818801879883,
      "learning_rate": 1e-06,
      "loss": 0.4074,
      "num_input_tokens_seen": 431478884,
      "step": 7700
    },
    {
      "epoch": 17.14922048997773,
      "loss": 0.3793688118457794,
      "loss_ce": 9.636204777052626e-05,
      "loss_iou": 0.1513671875,
      "loss_num": 0.0152587890625,
      "loss_xval": 0.37890625,
      "num_input_tokens_seen": 431478884,
      "step": 7700
    },
    {
      "epoch": 17.151447661469934,
      "grad_norm": 17.16455841064453,
      "learning_rate": 1e-06,
      "loss": 0.3772,
      "num_input_tokens_seen": 431537440,
      "step": 7701
    },
    {
      "epoch": 17.151447661469934,
      "loss": 0.4580909013748169,
      "loss_ce": 8.308385440614074e-05,
      "loss_iou": 0.185546875,
      "loss_num": 0.01708984375,
      "loss_xval": 0.45703125,
      "num_input_tokens_seen": 431537440,
      "step": 7701
    },
    {
      "epoch": 17.15367483296214,
      "grad_norm": 19.00710105895996,
      "learning_rate": 1e-06,
      "loss": 0.4474,
      "num_input_tokens_seen": 431592904,
      "step": 7702
    },
    {
      "epoch": 17.15367483296214,
      "loss": 0.4547297954559326,
      "loss_ce": 7.894145528553054e-05,
      "loss_iou": 0.1904296875,
      "loss_num": 0.01483154296875,
      "loss_xval": 0.455078125,
      "num_input_tokens_seen": 431592904,
      "step": 7702
    },
    {
      "epoch": 17.155902004454344,
      "grad_norm": 14.55823040008545,
      "learning_rate": 1e-06,
      "loss": 0.2756,
      "num_input_tokens_seen": 431651384,
      "step": 7703
    },
    {
      "epoch": 17.155902004454344,
      "loss": 0.23866218328475952,
      "loss_ce": 7.575111521873623e-05,
      "loss_iou": 0.10546875,
      "loss_num": 0.005462646484375,
      "loss_xval": 0.23828125,
      "num_input_tokens_seen": 431651384,
      "step": 7703
    },
    {
      "epoch": 17.15812917594655,
      "grad_norm": 13.731185913085938,
      "learning_rate": 1e-06,
      "loss": 0.37,
      "num_input_tokens_seen": 431708828,
      "step": 7704
    },
    {
      "epoch": 17.15812917594655,
      "loss": 0.347605437040329,
      "loss_ce": 7.126451964722946e-05,
      "loss_iou": 0.1474609375,
      "loss_num": 0.01043701171875,
      "loss_xval": 0.34765625,
      "num_input_tokens_seen": 431708828,
      "step": 7704
    },
    {
      "epoch": 17.160356347438753,
      "grad_norm": 21.002777099609375,
      "learning_rate": 1e-06,
      "loss": 0.5383,
      "num_input_tokens_seen": 431765240,
      "step": 7705
    },
    {
      "epoch": 17.160356347438753,
      "loss": 0.4782218933105469,
      "loss_ce": 7.249362533912063e-05,
      "loss_iou": 0.1943359375,
      "loss_num": 0.017822265625,
      "loss_xval": 0.478515625,
      "num_input_tokens_seen": 431765240,
      "step": 7705
    },
    {
      "epoch": 17.16258351893096,
      "grad_norm": 21.7739315032959,
      "learning_rate": 1e-06,
      "loss": 0.6177,
      "num_input_tokens_seen": 431820024,
      "step": 7706
    },
    {
      "epoch": 17.16258351893096,
      "loss": 0.7398481369018555,
      "loss_ce": 0.00010208625462837517,
      "loss_iou": 0.330078125,
      "loss_num": 0.015625,
      "loss_xval": 0.73828125,
      "num_input_tokens_seen": 431820024,
      "step": 7706
    },
    {
      "epoch": 17.164810690423163,
      "grad_norm": 18.27759552001953,
      "learning_rate": 1e-06,
      "loss": 0.3539,
      "num_input_tokens_seen": 431873500,
      "step": 7707
    },
    {
      "epoch": 17.164810690423163,
      "loss": 0.3486563265323639,
      "loss_ce": 8.456400246359408e-05,
      "loss_iou": 0.1630859375,
      "loss_num": 0.004669189453125,
      "loss_xval": 0.34765625,
      "num_input_tokens_seen": 431873500,
      "step": 7707
    },
    {
      "epoch": 17.167037861915368,
      "grad_norm": 32.455810546875,
      "learning_rate": 1e-06,
      "loss": 0.4563,
      "num_input_tokens_seen": 431928992,
      "step": 7708
    },
    {
      "epoch": 17.167037861915368,
      "loss": 0.5565398931503296,
      "loss_ce": 8.237551082856953e-05,
      "loss_iou": 0.2060546875,
      "loss_num": 0.029052734375,
      "loss_xval": 0.5546875,
      "num_input_tokens_seen": 431928992,
      "step": 7708
    },
    {
      "epoch": 17.169265033407573,
      "grad_norm": 24.227785110473633,
      "learning_rate": 1e-06,
      "loss": 0.4164,
      "num_input_tokens_seen": 431985988,
      "step": 7709
    },
    {
      "epoch": 17.169265033407573,
      "loss": 0.21856805682182312,
      "loss_ce": 9.271766612073407e-05,
      "loss_iou": 0.10205078125,
      "loss_num": 0.00274658203125,
      "loss_xval": 0.21875,
      "num_input_tokens_seen": 431985988,
      "step": 7709
    },
    {
      "epoch": 17.171492204899778,
      "grad_norm": 11.984196662902832,
      "learning_rate": 1e-06,
      "loss": 0.2504,
      "num_input_tokens_seen": 432040140,
      "step": 7710
    },
    {
      "epoch": 17.171492204899778,
      "loss": 0.2045476734638214,
      "loss_ce": 7.990330050233752e-05,
      "loss_iou": 0.087890625,
      "loss_num": 0.0057373046875,
      "loss_xval": 0.2041015625,
      "num_input_tokens_seen": 432040140,
      "step": 7710
    },
    {
      "epoch": 17.173719376391983,
      "grad_norm": 23.284042358398438,
      "learning_rate": 1e-06,
      "loss": 0.3402,
      "num_input_tokens_seen": 432093984,
      "step": 7711
    },
    {
      "epoch": 17.173719376391983,
      "loss": 0.3212023973464966,
      "loss_ce": 6.591899727936834e-05,
      "loss_iou": 0.138671875,
      "loss_num": 0.00872802734375,
      "loss_xval": 0.3203125,
      "num_input_tokens_seen": 432093984,
      "step": 7711
    },
    {
      "epoch": 17.175946547884188,
      "grad_norm": 15.671782493591309,
      "learning_rate": 1e-06,
      "loss": 0.3822,
      "num_input_tokens_seen": 432149584,
      "step": 7712
    },
    {
      "epoch": 17.175946547884188,
      "loss": 0.2187042385339737,
      "loss_ce": 7.631281914655119e-05,
      "loss_iou": 0.0869140625,
      "loss_num": 0.0089111328125,
      "loss_xval": 0.21875,
      "num_input_tokens_seen": 432149584,
      "step": 7712
    },
    {
      "epoch": 17.178173719376392,
      "grad_norm": 17.098852157592773,
      "learning_rate": 1e-06,
      "loss": 0.3588,
      "num_input_tokens_seen": 432206804,
      "step": 7713
    },
    {
      "epoch": 17.178173719376392,
      "loss": 0.4782329201698303,
      "loss_ce": 8.348520350409672e-05,
      "loss_iou": 0.2197265625,
      "loss_num": 0.007537841796875,
      "loss_xval": 0.478515625,
      "num_input_tokens_seen": 432206804,
      "step": 7713
    },
    {
      "epoch": 17.180400890868597,
      "grad_norm": 23.695720672607422,
      "learning_rate": 1e-06,
      "loss": 0.6842,
      "num_input_tokens_seen": 432259452,
      "step": 7714
    },
    {
      "epoch": 17.180400890868597,
      "loss": 0.6622397899627686,
      "loss_ce": 0.00025244534481316805,
      "loss_iou": 0.29296875,
      "loss_num": 0.0155029296875,
      "loss_xval": 0.66015625,
      "num_input_tokens_seen": 432259452,
      "step": 7714
    },
    {
      "epoch": 17.182628062360802,
      "grad_norm": 16.1234073638916,
      "learning_rate": 1e-06,
      "loss": 0.3404,
      "num_input_tokens_seen": 432317140,
      "step": 7715
    },
    {
      "epoch": 17.182628062360802,
      "loss": 0.2790215313434601,
      "loss_ce": 9.086594945983961e-05,
      "loss_iou": 0.12109375,
      "loss_num": 0.007476806640625,
      "loss_xval": 0.279296875,
      "num_input_tokens_seen": 432317140,
      "step": 7715
    },
    {
      "epoch": 17.184855233853007,
      "grad_norm": 21.128246307373047,
      "learning_rate": 1e-06,
      "loss": 0.4791,
      "num_input_tokens_seen": 432375580,
      "step": 7716
    },
    {
      "epoch": 17.184855233853007,
      "loss": 0.7009245753288269,
      "loss_ce": 8.839939255267382e-05,
      "loss_iou": 0.294921875,
      "loss_num": 0.0224609375,
      "loss_xval": 0.69921875,
      "num_input_tokens_seen": 432375580,
      "step": 7716
    },
    {
      "epoch": 17.187082405345212,
      "grad_norm": 18.01925277709961,
      "learning_rate": 1e-06,
      "loss": 0.3742,
      "num_input_tokens_seen": 432428196,
      "step": 7717
    },
    {
      "epoch": 17.187082405345212,
      "loss": 0.48470473289489746,
      "loss_ce": 8.559299749322236e-05,
      "loss_iou": 0.2138671875,
      "loss_num": 0.0115966796875,
      "loss_xval": 0.484375,
      "num_input_tokens_seen": 432428196,
      "step": 7717
    },
    {
      "epoch": 17.189309576837417,
      "grad_norm": 19.113862991333008,
      "learning_rate": 1e-06,
      "loss": 0.4242,
      "num_input_tokens_seen": 432482660,
      "step": 7718
    },
    {
      "epoch": 17.189309576837417,
      "loss": 0.41426295042037964,
      "loss_ce": 7.838521560188383e-05,
      "loss_iou": 0.18359375,
      "loss_num": 0.009521484375,
      "loss_xval": 0.4140625,
      "num_input_tokens_seen": 432482660,
      "step": 7718
    },
    {
      "epoch": 17.19153674832962,
      "grad_norm": 18.660619735717773,
      "learning_rate": 1e-06,
      "loss": 0.3095,
      "num_input_tokens_seen": 432539144,
      "step": 7719
    },
    {
      "epoch": 17.19153674832962,
      "loss": 0.2771776020526886,
      "loss_ce": 7.799692684784532e-05,
      "loss_iou": 0.111328125,
      "loss_num": 0.0108642578125,
      "loss_xval": 0.27734375,
      "num_input_tokens_seen": 432539144,
      "step": 7719
    },
    {
      "epoch": 17.193763919821826,
      "grad_norm": 18.885120391845703,
      "learning_rate": 1e-06,
      "loss": 0.3698,
      "num_input_tokens_seen": 432596392,
      "step": 7720
    },
    {
      "epoch": 17.193763919821826,
      "loss": 0.4347517788410187,
      "loss_ce": 0.00030352562316693366,
      "loss_iou": 0.1962890625,
      "loss_num": 0.0086669921875,
      "loss_xval": 0.43359375,
      "num_input_tokens_seen": 432596392,
      "step": 7720
    },
    {
      "epoch": 17.19599109131403,
      "grad_norm": 11.850611686706543,
      "learning_rate": 1e-06,
      "loss": 0.5695,
      "num_input_tokens_seen": 432650932,
      "step": 7721
    },
    {
      "epoch": 17.19599109131403,
      "loss": 0.5424901247024536,
      "loss_ce": 7.069206185406074e-05,
      "loss_iou": 0.22265625,
      "loss_num": 0.019287109375,
      "loss_xval": 0.54296875,
      "num_input_tokens_seen": 432650932,
      "step": 7721
    },
    {
      "epoch": 17.198218262806236,
      "grad_norm": 19.987720489501953,
      "learning_rate": 1e-06,
      "loss": 0.3179,
      "num_input_tokens_seen": 432706504,
      "step": 7722
    },
    {
      "epoch": 17.198218262806236,
      "loss": 0.3480537533760071,
      "loss_ce": 9.233770833816379e-05,
      "loss_iou": 0.1572265625,
      "loss_num": 0.006866455078125,
      "loss_xval": 0.34765625,
      "num_input_tokens_seen": 432706504,
      "step": 7722
    },
    {
      "epoch": 17.20044543429844,
      "grad_norm": 18.74901008605957,
      "learning_rate": 1e-06,
      "loss": 0.4127,
      "num_input_tokens_seen": 432765232,
      "step": 7723
    },
    {
      "epoch": 17.20044543429844,
      "loss": 0.4682392477989197,
      "loss_ce": 9.959404997061938e-05,
      "loss_iou": 0.208984375,
      "loss_num": 0.0098876953125,
      "loss_xval": 0.46875,
      "num_input_tokens_seen": 432765232,
      "step": 7723
    },
    {
      "epoch": 17.202672605790646,
      "grad_norm": 18.15958023071289,
      "learning_rate": 1e-06,
      "loss": 0.3961,
      "num_input_tokens_seen": 432820076,
      "step": 7724
    },
    {
      "epoch": 17.202672605790646,
      "loss": 0.41411644220352173,
      "loss_ce": 8.445019193459302e-05,
      "loss_iou": 0.185546875,
      "loss_num": 0.008544921875,
      "loss_xval": 0.4140625,
      "num_input_tokens_seen": 432820076,
      "step": 7724
    },
    {
      "epoch": 17.20489977728285,
      "grad_norm": 27.001567840576172,
      "learning_rate": 1e-06,
      "loss": 0.4505,
      "num_input_tokens_seen": 432877112,
      "step": 7725
    },
    {
      "epoch": 17.20489977728285,
      "loss": 0.46639156341552734,
      "loss_ce": 8.295044244732708e-05,
      "loss_iou": 0.2001953125,
      "loss_num": 0.01300048828125,
      "loss_xval": 0.466796875,
      "num_input_tokens_seen": 432877112,
      "step": 7725
    },
    {
      "epoch": 17.207126948775056,
      "grad_norm": 30.94824981689453,
      "learning_rate": 1e-06,
      "loss": 0.3738,
      "num_input_tokens_seen": 432934028,
      "step": 7726
    },
    {
      "epoch": 17.207126948775056,
      "loss": 0.31847047805786133,
      "loss_ce": 8.058200182858855e-05,
      "loss_iou": 0.142578125,
      "loss_num": 0.006591796875,
      "loss_xval": 0.318359375,
      "num_input_tokens_seen": 432934028,
      "step": 7726
    },
    {
      "epoch": 17.20935412026726,
      "grad_norm": 17.023353576660156,
      "learning_rate": 1e-06,
      "loss": 0.4936,
      "num_input_tokens_seen": 432991524,
      "step": 7727
    },
    {
      "epoch": 17.20935412026726,
      "loss": 0.4206371307373047,
      "loss_ce": 0.0002880272513721138,
      "loss_iou": 0.1669921875,
      "loss_num": 0.0174560546875,
      "loss_xval": 0.419921875,
      "num_input_tokens_seen": 432991524,
      "step": 7727
    },
    {
      "epoch": 17.211581291759465,
      "grad_norm": 17.618732452392578,
      "learning_rate": 1e-06,
      "loss": 0.3742,
      "num_input_tokens_seen": 433050212,
      "step": 7728
    },
    {
      "epoch": 17.211581291759465,
      "loss": 0.3367721438407898,
      "loss_ce": 0.0001022043579723686,
      "loss_iou": 0.15234375,
      "loss_num": 0.00653076171875,
      "loss_xval": 0.3359375,
      "num_input_tokens_seen": 433050212,
      "step": 7728
    },
    {
      "epoch": 17.21380846325167,
      "grad_norm": 17.69324493408203,
      "learning_rate": 1e-06,
      "loss": 0.4379,
      "num_input_tokens_seen": 433106864,
      "step": 7729
    },
    {
      "epoch": 17.21380846325167,
      "loss": 0.5612530708312988,
      "loss_ce": 9.585064981365576e-05,
      "loss_iou": 0.26171875,
      "loss_num": 0.007171630859375,
      "loss_xval": 0.5625,
      "num_input_tokens_seen": 433106864,
      "step": 7729
    },
    {
      "epoch": 17.216035634743875,
      "grad_norm": 20.909269332885742,
      "learning_rate": 1e-06,
      "loss": 0.3521,
      "num_input_tokens_seen": 433163416,
      "step": 7730
    },
    {
      "epoch": 17.216035634743875,
      "loss": 0.40844836831092834,
      "loss_ce": 9.26676148083061e-05,
      "loss_iou": 0.1748046875,
      "loss_num": 0.01190185546875,
      "loss_xval": 0.408203125,
      "num_input_tokens_seen": 433163416,
      "step": 7730
    },
    {
      "epoch": 17.21826280623608,
      "grad_norm": 17.81989860534668,
      "learning_rate": 1e-06,
      "loss": 0.3125,
      "num_input_tokens_seen": 433218392,
      "step": 7731
    },
    {
      "epoch": 17.21826280623608,
      "loss": 0.32392922043800354,
      "loss_ce": 7.668677426408976e-05,
      "loss_iou": 0.1298828125,
      "loss_num": 0.01300048828125,
      "loss_xval": 0.32421875,
      "num_input_tokens_seen": 433218392,
      "step": 7731
    },
    {
      "epoch": 17.220489977728285,
      "grad_norm": 17.176774978637695,
      "learning_rate": 1e-06,
      "loss": 0.3715,
      "num_input_tokens_seen": 433273928,
      "step": 7732
    },
    {
      "epoch": 17.220489977728285,
      "loss": 0.35421961545944214,
      "loss_ce": 9.36458382057026e-05,
      "loss_iou": 0.1630859375,
      "loss_num": 0.005584716796875,
      "loss_xval": 0.353515625,
      "num_input_tokens_seen": 433273928,
      "step": 7732
    },
    {
      "epoch": 17.22271714922049,
      "grad_norm": 13.050132751464844,
      "learning_rate": 1e-06,
      "loss": 0.534,
      "num_input_tokens_seen": 433330336,
      "step": 7733
    },
    {
      "epoch": 17.22271714922049,
      "loss": 0.5098897814750671,
      "loss_ce": 0.00012416887329891324,
      "loss_iou": 0.21484375,
      "loss_num": 0.015869140625,
      "loss_xval": 0.5078125,
      "num_input_tokens_seen": 433330336,
      "step": 7733
    },
    {
      "epoch": 17.224944320712694,
      "grad_norm": 20.13163948059082,
      "learning_rate": 1e-06,
      "loss": 0.4318,
      "num_input_tokens_seen": 433384892,
      "step": 7734
    },
    {
      "epoch": 17.224944320712694,
      "loss": 0.34920066595077515,
      "loss_ce": 7.957725756568834e-05,
      "loss_iou": 0.1552734375,
      "loss_num": 0.007720947265625,
      "loss_xval": 0.349609375,
      "num_input_tokens_seen": 433384892,
      "step": 7734
    },
    {
      "epoch": 17.2271714922049,
      "grad_norm": 25.29994773864746,
      "learning_rate": 1e-06,
      "loss": 0.2885,
      "num_input_tokens_seen": 433440724,
      "step": 7735
    },
    {
      "epoch": 17.2271714922049,
      "loss": 0.22060546278953552,
      "loss_ce": 8.54411773616448e-05,
      "loss_iou": 0.1005859375,
      "loss_num": 0.0038299560546875,
      "loss_xval": 0.220703125,
      "num_input_tokens_seen": 433440724,
      "step": 7735
    },
    {
      "epoch": 17.229398663697104,
      "grad_norm": 19.234081268310547,
      "learning_rate": 1e-06,
      "loss": 0.2669,
      "num_input_tokens_seen": 433497228,
      "step": 7736
    },
    {
      "epoch": 17.229398663697104,
      "loss": 0.3394434154033661,
      "loss_ce": 8.795637404546142e-05,
      "loss_iou": 0.142578125,
      "loss_num": 0.0107421875,
      "loss_xval": 0.33984375,
      "num_input_tokens_seen": 433497228,
      "step": 7736
    },
    {
      "epoch": 17.23162583518931,
      "grad_norm": 25.836366653442383,
      "learning_rate": 1e-06,
      "loss": 0.4823,
      "num_input_tokens_seen": 433553476,
      "step": 7737
    },
    {
      "epoch": 17.23162583518931,
      "loss": 0.45820939540863037,
      "loss_ce": 7.951643783599138e-05,
      "loss_iou": 0.1865234375,
      "loss_num": 0.0172119140625,
      "loss_xval": 0.458984375,
      "num_input_tokens_seen": 433553476,
      "step": 7737
    },
    {
      "epoch": 17.233853006681514,
      "grad_norm": 34.01211166381836,
      "learning_rate": 1e-06,
      "loss": 0.4521,
      "num_input_tokens_seen": 433608588,
      "step": 7738
    },
    {
      "epoch": 17.233853006681514,
      "loss": 0.5483601093292236,
      "loss_ce": 8.125473686959594e-05,
      "loss_iou": 0.244140625,
      "loss_num": 0.011962890625,
      "loss_xval": 0.546875,
      "num_input_tokens_seen": 433608588,
      "step": 7738
    },
    {
      "epoch": 17.23608017817372,
      "grad_norm": 26.347332000732422,
      "learning_rate": 1e-06,
      "loss": 0.403,
      "num_input_tokens_seen": 433665036,
      "step": 7739
    },
    {
      "epoch": 17.23608017817372,
      "loss": 0.4297289252281189,
      "loss_ce": 0.00010247422324027866,
      "loss_iou": 0.1806640625,
      "loss_num": 0.01373291015625,
      "loss_xval": 0.4296875,
      "num_input_tokens_seen": 433665036,
      "step": 7739
    },
    {
      "epoch": 17.238307349665924,
      "grad_norm": 16.483013153076172,
      "learning_rate": 1e-06,
      "loss": 0.4577,
      "num_input_tokens_seen": 433721384,
      "step": 7740
    },
    {
      "epoch": 17.238307349665924,
      "loss": 0.40596240758895874,
      "loss_ce": 7.860736513976008e-05,
      "loss_iou": 0.1787109375,
      "loss_num": 0.0098876953125,
      "loss_xval": 0.40625,
      "num_input_tokens_seen": 433721384,
      "step": 7740
    },
    {
      "epoch": 17.24053452115813,
      "grad_norm": 16.415891647338867,
      "learning_rate": 1e-06,
      "loss": 0.34,
      "num_input_tokens_seen": 433779720,
      "step": 7741
    },
    {
      "epoch": 17.24053452115813,
      "loss": 0.3139420747756958,
      "loss_ce": 9.928335202857852e-05,
      "loss_iou": 0.1376953125,
      "loss_num": 0.007598876953125,
      "loss_xval": 0.314453125,
      "num_input_tokens_seen": 433779720,
      "step": 7741
    },
    {
      "epoch": 17.242761692650333,
      "grad_norm": 13.877442359924316,
      "learning_rate": 1e-06,
      "loss": 0.335,
      "num_input_tokens_seen": 433835132,
      "step": 7742
    },
    {
      "epoch": 17.242761692650333,
      "loss": 0.32295700907707214,
      "loss_ce": 8.10332567198202e-05,
      "loss_iou": 0.14453125,
      "loss_num": 0.00677490234375,
      "loss_xval": 0.322265625,
      "num_input_tokens_seen": 433835132,
      "step": 7742
    },
    {
      "epoch": 17.244988864142538,
      "grad_norm": 25.894933700561523,
      "learning_rate": 1e-06,
      "loss": 0.4914,
      "num_input_tokens_seen": 433890280,
      "step": 7743
    },
    {
      "epoch": 17.244988864142538,
      "loss": 0.47230130434036255,
      "loss_ce": 8.75686964718625e-05,
      "loss_iou": 0.2109375,
      "loss_num": 0.0103759765625,
      "loss_xval": 0.47265625,
      "num_input_tokens_seen": 433890280,
      "step": 7743
    },
    {
      "epoch": 17.247216035634743,
      "grad_norm": 28.23295021057129,
      "learning_rate": 1e-06,
      "loss": 0.4699,
      "num_input_tokens_seen": 433945752,
      "step": 7744
    },
    {
      "epoch": 17.247216035634743,
      "loss": 0.6319034695625305,
      "loss_ce": 0.00018957318388856947,
      "loss_iou": 0.2578125,
      "loss_num": 0.0234375,
      "loss_xval": 0.6328125,
      "num_input_tokens_seen": 433945752,
      "step": 7744
    },
    {
      "epoch": 17.249443207126948,
      "grad_norm": 19.394866943359375,
      "learning_rate": 1e-06,
      "loss": 0.4551,
      "num_input_tokens_seen": 434001564,
      "step": 7745
    },
    {
      "epoch": 17.249443207126948,
      "loss": 0.531867504119873,
      "loss_ce": 0.0002207824436482042,
      "loss_iou": 0.2001953125,
      "loss_num": 0.0263671875,
      "loss_xval": 0.53125,
      "num_input_tokens_seen": 434001564,
      "step": 7745
    },
    {
      "epoch": 17.251670378619153,
      "grad_norm": 31.168285369873047,
      "learning_rate": 1e-06,
      "loss": 0.4663,
      "num_input_tokens_seen": 434058448,
      "step": 7746
    },
    {
      "epoch": 17.251670378619153,
      "loss": 0.6216709613800049,
      "loss_ce": 8.895625069271773e-05,
      "loss_iou": 0.2734375,
      "loss_num": 0.014892578125,
      "loss_xval": 0.62109375,
      "num_input_tokens_seen": 434058448,
      "step": 7746
    },
    {
      "epoch": 17.253897550111358,
      "grad_norm": 15.39581298828125,
      "learning_rate": 1e-06,
      "loss": 0.4586,
      "num_input_tokens_seen": 434115584,
      "step": 7747
    },
    {
      "epoch": 17.253897550111358,
      "loss": 0.5104674100875854,
      "loss_ce": 9.140316979028285e-05,
      "loss_iou": 0.2158203125,
      "loss_num": 0.0155029296875,
      "loss_xval": 0.51171875,
      "num_input_tokens_seen": 434115584,
      "step": 7747
    },
    {
      "epoch": 17.256124721603562,
      "grad_norm": 26.95697593688965,
      "learning_rate": 1e-06,
      "loss": 0.4361,
      "num_input_tokens_seen": 434169820,
      "step": 7748
    },
    {
      "epoch": 17.256124721603562,
      "loss": 0.35939961671829224,
      "loss_ce": 8.56606347952038e-05,
      "loss_iou": 0.162109375,
      "loss_num": 0.006988525390625,
      "loss_xval": 0.359375,
      "num_input_tokens_seen": 434169820,
      "step": 7748
    },
    {
      "epoch": 17.258351893095767,
      "grad_norm": 41.06592559814453,
      "learning_rate": 1e-06,
      "loss": 0.3006,
      "num_input_tokens_seen": 434228656,
      "step": 7749
    },
    {
      "epoch": 17.258351893095767,
      "loss": 0.29035407304763794,
      "loss_ce": 0.0008032863843254745,
      "loss_iou": 0.126953125,
      "loss_num": 0.00726318359375,
      "loss_xval": 0.2890625,
      "num_input_tokens_seen": 434228656,
      "step": 7749
    },
    {
      "epoch": 17.260579064587972,
      "grad_norm": 12.836146354675293,
      "learning_rate": 1e-06,
      "loss": 0.2803,
      "num_input_tokens_seen": 434285972,
      "step": 7750
    },
    {
      "epoch": 17.260579064587972,
      "eval_seeclick_web_CIoU": 0.5904313921928406,
      "eval_seeclick_web_GIoU": 0.5887089371681213,
      "eval_seeclick_web_IoU": 0.6092908382415771,
      "eval_seeclick_web_MAE_all": 0.015212189638987184,
      "eval_seeclick_web_MAE_h": 0.007458887062966824,
      "eval_seeclick_web_MAE_w": 0.01523585431277752,
      "eval_seeclick_web_MAE_x_boxes": 0.007938009221106768,
      "eval_seeclick_web_MAE_y_boxes": 0.021370060741901398,
      "eval_seeclick_web_inside_bbox": 0.9010416567325592,
      "eval_seeclick_web_loss": 0.8967059254646301,
      "eval_seeclick_web_loss_ce": 0.0001397554951836355,
      "eval_seeclick_web_loss_iou": 0.413818359375,
      "eval_seeclick_web_loss_num": 0.012087821960449219,
      "eval_seeclick_web_loss_xval": 0.887451171875,
      "eval_seeclick_web_runtime": 21.8766,
      "eval_seeclick_web_samples_per_second": 2.286,
      "eval_seeclick_web_steps_per_second": 0.091,
      "num_input_tokens_seen": 434285972,
      "step": 7750
    },
    {
      "epoch": 17.260579064587972,
      "eval_icons_CIoU": 0.25680967420339584,
      "eval_icons_GIoU": 0.2801186218857765,
      "eval_icons_IoU": 0.3349318951368332,
      "eval_icons_MAE_all": 0.05978231504559517,
      "eval_icons_MAE_h": 0.0324998227879405,
      "eval_icons_MAE_w": 0.06001891568303108,
      "eval_icons_MAE_x_boxes": 0.06087409518659115,
      "eval_icons_MAE_y_boxes": 0.03790356032550335,
      "eval_icons_inside_bbox": 0.59375,
      "eval_icons_loss": 1.7364193201065063,
      "eval_icons_loss_ce": 0.00016331803635694087,
      "eval_icons_loss_iou": 0.6763916015625,
      "eval_icons_loss_num": 0.052509307861328125,
      "eval_icons_loss_xval": 1.615966796875,
      "eval_icons_runtime": 21.8048,
      "eval_icons_samples_per_second": 2.293,
      "eval_icons_steps_per_second": 0.092,
      "num_input_tokens_seen": 434285972,
      "step": 7750
    },
    {
      "epoch": 17.260579064587972,
      "eval_screenspot_CIoU": 0.3815999726454417,
      "eval_screenspot_GIoU": 0.3997166156768799,
      "eval_screenspot_IoU": 0.45147113005320233,
      "eval_screenspot_MAE_all": 0.055647388100624084,
      "eval_screenspot_MAE_h": 0.03920063997308413,
      "eval_screenspot_MAE_w": 0.06165233999490738,
      "eval_screenspot_MAE_x_boxes": 0.06318879996736844,
      "eval_screenspot_MAE_y_boxes": 0.03922058828175068,
      "eval_screenspot_inside_bbox": 0.7145833373069763,
      "eval_screenspot_loss": 1.546026349067688,
      "eval_screenspot_loss_ce": 0.00020924270696317157,
      "eval_screenspot_loss_iou": 0.6417643229166666,
      "eval_screenspot_loss_num": 0.06403223673502605,
      "eval_screenspot_loss_xval": 1.6028645833333333,
      "eval_screenspot_runtime": 37.6194,
      "eval_screenspot_samples_per_second": 2.366,
      "eval_screenspot_steps_per_second": 0.08,
      "num_input_tokens_seen": 434285972,
      "step": 7750
    },
    {
      "epoch": 17.260579064587972,
      "eval_compot_CIoU": 0.3500685393810272,
      "eval_compot_GIoU": 0.36062242090702057,
      "eval_compot_IoU": 0.4054069072008133,
      "eval_compot_MAE_all": 0.0187700055539608,
      "eval_compot_MAE_h": 0.010957159101963043,
      "eval_compot_MAE_w": 0.02098592184484005,
      "eval_compot_MAE_x_boxes": 0.03000013902783394,
      "eval_compot_MAE_y_boxes": 0.006979038938879967,
      "eval_compot_inside_bbox": 0.6302083432674408,
      "eval_compot_loss": 1.3751345872879028,
      "eval_compot_loss_ce": 0.00013413318083621562,
      "eval_compot_loss_iou": 0.6326904296875,
      "eval_compot_loss_num": 0.017452239990234375,
      "eval_compot_loss_xval": 1.35302734375,
      "eval_compot_runtime": 23.5427,
      "eval_compot_samples_per_second": 2.124,
      "eval_compot_steps_per_second": 0.085,
      "num_input_tokens_seen": 434285972,
      "step": 7750
    },
    {
      "epoch": 17.260579064587972,
      "eval_custom_ui_val_CIoU": 0.47184327410327065,
      "eval_custom_ui_val_GIoU": 0.4758959710597992,
      "eval_custom_ui_val_IoU": 0.5333681570159065,
      "eval_custom_ui_val_MAE_all": 0.027272911483628884,
      "eval_custom_ui_val_MAE_h": 0.014620246106965674,
      "eval_custom_ui_val_MAE_w": 0.03669780415172378,
      "eval_custom_ui_val_MAE_x_boxes": 0.03337866820705434,
      "eval_custom_ui_val_MAE_y_boxes": 0.012776042127774822,
      "eval_custom_ui_val_inside_bbox": 0.7754629651705424,
      "eval_custom_ui_val_loss": 1.177891492843628,
      "eval_custom_ui_val_loss_ce": 0.00015998236550432112,
      "eval_custom_ui_val_loss_iou": 0.5063612196180556,
      "eval_custom_ui_val_loss_num": 0.023825857374403212,
      "eval_custom_ui_val_loss_xval": 1.1314561631944444,
      "eval_custom_ui_val_runtime": 65.7899,
      "eval_custom_ui_val_samples_per_second": 4.028,
      "eval_custom_ui_val_steps_per_second": 0.137,
      "num_input_tokens_seen": 434285972,
      "step": 7750
    },
    {
      "epoch": 17.260579064587972,
      "loss": 0.8485028743743896,
      "loss_ce": 0.00011421847739256918,
      "loss_iou": 0.3828125,
      "loss_num": 0.016845703125,
      "loss_xval": 0.84765625,
      "num_input_tokens_seen": 434285972,
      "step": 7750
    },
    {
      "epoch": 17.262806236080177,
      "grad_norm": 40.981712341308594,
      "learning_rate": 1e-06,
      "loss": 0.4852,
      "num_input_tokens_seen": 434344348,
      "step": 7751
    },
    {
      "epoch": 17.262806236080177,
      "loss": 0.6921710968017578,
      "loss_ce": 0.00021551080862991512,
      "loss_iou": 0.263671875,
      "loss_num": 0.03271484375,
      "loss_xval": 0.69140625,
      "num_input_tokens_seen": 434344348,
      "step": 7751
    },
    {
      "epoch": 17.265033407572382,
      "grad_norm": 21.000991821289062,
      "learning_rate": 1e-06,
      "loss": 0.3312,
      "num_input_tokens_seen": 434397948,
      "step": 7752
    },
    {
      "epoch": 17.265033407572382,
      "loss": 0.3175837993621826,
      "loss_ce": 7.892360736150295e-05,
      "loss_iou": 0.1455078125,
      "loss_num": 0.005340576171875,
      "loss_xval": 0.318359375,
      "num_input_tokens_seen": 434397948,
      "step": 7752
    },
    {
      "epoch": 17.267260579064587,
      "grad_norm": 19.446720123291016,
      "learning_rate": 1e-06,
      "loss": 0.4803,
      "num_input_tokens_seen": 434452904,
      "step": 7753
    },
    {
      "epoch": 17.267260579064587,
      "loss": 0.5616533756256104,
      "loss_ce": 9.945797501131892e-05,
      "loss_iou": 0.2314453125,
      "loss_num": 0.0196533203125,
      "loss_xval": 0.5625,
      "num_input_tokens_seen": 434452904,
      "step": 7753
    },
    {
      "epoch": 17.26948775055679,
      "grad_norm": 18.333148956298828,
      "learning_rate": 1e-06,
      "loss": 0.4172,
      "num_input_tokens_seen": 434508660,
      "step": 7754
    },
    {
      "epoch": 17.26948775055679,
      "loss": 0.34176281094551086,
      "loss_ce": 8.800373325357214e-05,
      "loss_iou": 0.138671875,
      "loss_num": 0.012939453125,
      "loss_xval": 0.341796875,
      "num_input_tokens_seen": 434508660,
      "step": 7754
    },
    {
      "epoch": 17.271714922048996,
      "grad_norm": 19.104904174804688,
      "learning_rate": 1e-06,
      "loss": 0.3037,
      "num_input_tokens_seen": 434565492,
      "step": 7755
    },
    {
      "epoch": 17.271714922048996,
      "loss": 0.27949851751327515,
      "loss_ce": 7.9582110629417e-05,
      "loss_iou": 0.11669921875,
      "loss_num": 0.0091552734375,
      "loss_xval": 0.279296875,
      "num_input_tokens_seen": 434565492,
      "step": 7755
    },
    {
      "epoch": 17.2739420935412,
      "grad_norm": 16.93346405029297,
      "learning_rate": 1e-06,
      "loss": 0.3254,
      "num_input_tokens_seen": 434619016,
      "step": 7756
    },
    {
      "epoch": 17.2739420935412,
      "loss": 0.2303803563117981,
      "loss_ce": 9.471694647800177e-05,
      "loss_iou": 0.09326171875,
      "loss_num": 0.00872802734375,
      "loss_xval": 0.23046875,
      "num_input_tokens_seen": 434619016,
      "step": 7756
    },
    {
      "epoch": 17.276169265033406,
      "grad_norm": 15.693754196166992,
      "learning_rate": 1e-06,
      "loss": 0.3854,
      "num_input_tokens_seen": 434674412,
      "step": 7757
    },
    {
      "epoch": 17.276169265033406,
      "loss": 0.5041153430938721,
      "loss_ce": 8.705217624083161e-05,
      "loss_iou": 0.2041015625,
      "loss_num": 0.0191650390625,
      "loss_xval": 0.50390625,
      "num_input_tokens_seen": 434674412,
      "step": 7757
    },
    {
      "epoch": 17.27839643652561,
      "grad_norm": 44.67047882080078,
      "learning_rate": 1e-06,
      "loss": 0.3596,
      "num_input_tokens_seen": 434728900,
      "step": 7758
    },
    {
      "epoch": 17.27839643652561,
      "loss": 0.4251922070980072,
      "loss_ce": 8.232867548940703e-05,
      "loss_iou": 0.1953125,
      "loss_num": 0.006744384765625,
      "loss_xval": 0.42578125,
      "num_input_tokens_seen": 434728900,
      "step": 7758
    },
    {
      "epoch": 17.280623608017816,
      "grad_norm": 49.442901611328125,
      "learning_rate": 1e-06,
      "loss": 0.3936,
      "num_input_tokens_seen": 434784116,
      "step": 7759
    },
    {
      "epoch": 17.280623608017816,
      "loss": 0.41439324617385864,
      "loss_ce": 8.657870057504624e-05,
      "loss_iou": 0.1728515625,
      "loss_num": 0.0137939453125,
      "loss_xval": 0.4140625,
      "num_input_tokens_seen": 434784116,
      "step": 7759
    },
    {
      "epoch": 17.28285077951002,
      "grad_norm": 19.590133666992188,
      "learning_rate": 1e-06,
      "loss": 0.4671,
      "num_input_tokens_seen": 434838652,
      "step": 7760
    },
    {
      "epoch": 17.28285077951002,
      "loss": 0.31147128343582153,
      "loss_ce": 6.993389979470521e-05,
      "loss_iou": 0.1376953125,
      "loss_num": 0.0072021484375,
      "loss_xval": 0.310546875,
      "num_input_tokens_seen": 434838652,
      "step": 7760
    },
    {
      "epoch": 17.285077951002226,
      "grad_norm": 18.076194763183594,
      "learning_rate": 1e-06,
      "loss": 0.408,
      "num_input_tokens_seen": 434892168,
      "step": 7761
    },
    {
      "epoch": 17.285077951002226,
      "loss": 0.35101962089538574,
      "loss_ce": 6.74750772304833e-05,
      "loss_iou": 0.15234375,
      "loss_num": 0.009033203125,
      "loss_xval": 0.3515625,
      "num_input_tokens_seen": 434892168,
      "step": 7761
    },
    {
      "epoch": 17.28730512249443,
      "grad_norm": 20.00171661376953,
      "learning_rate": 1e-06,
      "loss": 0.3366,
      "num_input_tokens_seen": 434949224,
      "step": 7762
    },
    {
      "epoch": 17.28730512249443,
      "loss": 0.38625282049179077,
      "loss_ce": 8.34054226288572e-05,
      "loss_iou": 0.17578125,
      "loss_num": 0.007110595703125,
      "loss_xval": 0.38671875,
      "num_input_tokens_seen": 434949224,
      "step": 7762
    },
    {
      "epoch": 17.289532293986635,
      "grad_norm": 24.69287109375,
      "learning_rate": 1e-06,
      "loss": 0.35,
      "num_input_tokens_seen": 435001896,
      "step": 7763
    },
    {
      "epoch": 17.289532293986635,
      "loss": 0.408402681350708,
      "loss_ce": 7.747893687337637e-05,
      "loss_iou": 0.173828125,
      "loss_num": 0.01220703125,
      "loss_xval": 0.408203125,
      "num_input_tokens_seen": 435001896,
      "step": 7763
    },
    {
      "epoch": 17.29175946547884,
      "grad_norm": 11.8654203414917,
      "learning_rate": 1e-06,
      "loss": 0.2978,
      "num_input_tokens_seen": 435057448,
      "step": 7764
    },
    {
      "epoch": 17.29175946547884,
      "loss": 0.35043925046920776,
      "loss_ce": 9.744616545503959e-05,
      "loss_iou": 0.1455078125,
      "loss_num": 0.01171875,
      "loss_xval": 0.349609375,
      "num_input_tokens_seen": 435057448,
      "step": 7764
    },
    {
      "epoch": 17.293986636971045,
      "grad_norm": 15.643658638000488,
      "learning_rate": 1e-06,
      "loss": 0.3667,
      "num_input_tokens_seen": 435114576,
      "step": 7765
    },
    {
      "epoch": 17.293986636971045,
      "loss": 0.2830941081047058,
      "loss_ce": 7.410335820168257e-05,
      "loss_iou": 0.1279296875,
      "loss_num": 0.005615234375,
      "loss_xval": 0.283203125,
      "num_input_tokens_seen": 435114576,
      "step": 7765
    },
    {
      "epoch": 17.29621380846325,
      "grad_norm": 28.67169189453125,
      "learning_rate": 1e-06,
      "loss": 0.3936,
      "num_input_tokens_seen": 435170508,
      "step": 7766
    },
    {
      "epoch": 17.29621380846325,
      "loss": 0.4348151683807373,
      "loss_ce": 0.00012277913629077375,
      "loss_iou": 0.189453125,
      "loss_num": 0.010986328125,
      "loss_xval": 0.435546875,
      "num_input_tokens_seen": 435170508,
      "step": 7766
    },
    {
      "epoch": 17.29844097995546,
      "grad_norm": 40.479034423828125,
      "learning_rate": 1e-06,
      "loss": 0.5411,
      "num_input_tokens_seen": 435225040,
      "step": 7767
    },
    {
      "epoch": 17.29844097995546,
      "loss": 0.44416916370391846,
      "loss_ce": 7.737807754892856e-05,
      "loss_iou": 0.1826171875,
      "loss_num": 0.015625,
      "loss_xval": 0.443359375,
      "num_input_tokens_seen": 435225040,
      "step": 7767
    },
    {
      "epoch": 17.30066815144766,
      "grad_norm": 19.786476135253906,
      "learning_rate": 1e-06,
      "loss": 0.275,
      "num_input_tokens_seen": 435280060,
      "step": 7768
    },
    {
      "epoch": 17.30066815144766,
      "loss": 0.25245094299316406,
      "loss_ce": 7.053982699289918e-05,
      "loss_iou": 0.1162109375,
      "loss_num": 0.0040283203125,
      "loss_xval": 0.251953125,
      "num_input_tokens_seen": 435280060,
      "step": 7768
    },
    {
      "epoch": 17.302895322939868,
      "grad_norm": 25.475353240966797,
      "learning_rate": 1e-06,
      "loss": 0.4287,
      "num_input_tokens_seen": 435337596,
      "step": 7769
    },
    {
      "epoch": 17.302895322939868,
      "loss": 0.4136672914028168,
      "loss_ce": 9.308782318839803e-05,
      "loss_iou": 0.1875,
      "loss_num": 0.0076904296875,
      "loss_xval": 0.4140625,
      "num_input_tokens_seen": 435337596,
      "step": 7769
    },
    {
      "epoch": 17.305122494432073,
      "grad_norm": 13.328644752502441,
      "learning_rate": 1e-06,
      "loss": 0.3073,
      "num_input_tokens_seen": 435394108,
      "step": 7770
    },
    {
      "epoch": 17.305122494432073,
      "loss": 0.34184500575065613,
      "loss_ce": 7.864900544518605e-05,
      "loss_iou": 0.15234375,
      "loss_num": 0.007568359375,
      "loss_xval": 0.341796875,
      "num_input_tokens_seen": 435394108,
      "step": 7770
    },
    {
      "epoch": 17.307349665924278,
      "grad_norm": 19.182100296020508,
      "learning_rate": 1e-06,
      "loss": 0.4608,
      "num_input_tokens_seen": 435448756,
      "step": 7771
    },
    {
      "epoch": 17.307349665924278,
      "loss": 0.577645480632782,
      "loss_ce": 6.979001045692712e-05,
      "loss_iou": 0.2470703125,
      "loss_num": 0.0169677734375,
      "loss_xval": 0.578125,
      "num_input_tokens_seen": 435448756,
      "step": 7771
    },
    {
      "epoch": 17.309576837416483,
      "grad_norm": 22.650524139404297,
      "learning_rate": 1e-06,
      "loss": 0.3382,
      "num_input_tokens_seen": 435505496,
      "step": 7772
    },
    {
      "epoch": 17.309576837416483,
      "loss": 0.413726806640625,
      "loss_ce": 9.157357271760702e-05,
      "loss_iou": 0.1865234375,
      "loss_num": 0.0081787109375,
      "loss_xval": 0.4140625,
      "num_input_tokens_seen": 435505496,
      "step": 7772
    },
    {
      "epoch": 17.311804008908688,
      "grad_norm": 22.297222137451172,
      "learning_rate": 1e-06,
      "loss": 0.5371,
      "num_input_tokens_seen": 435561892,
      "step": 7773
    },
    {
      "epoch": 17.311804008908688,
      "loss": 0.3261268436908722,
      "loss_ce": 7.704827294219285e-05,
      "loss_iou": 0.1376953125,
      "loss_num": 0.01007080078125,
      "loss_xval": 0.326171875,
      "num_input_tokens_seen": 435561892,
      "step": 7773
    },
    {
      "epoch": 17.314031180400892,
      "grad_norm": 23.894563674926758,
      "learning_rate": 1e-06,
      "loss": 0.38,
      "num_input_tokens_seen": 435617896,
      "step": 7774
    },
    {
      "epoch": 17.314031180400892,
      "loss": 0.3799806237220764,
      "loss_ce": 9.779801621334627e-05,
      "loss_iou": 0.1748046875,
      "loss_num": 0.0062255859375,
      "loss_xval": 0.37890625,
      "num_input_tokens_seen": 435617896,
      "step": 7774
    },
    {
      "epoch": 17.316258351893097,
      "grad_norm": 21.17926597595215,
      "learning_rate": 1e-06,
      "loss": 0.4776,
      "num_input_tokens_seen": 435674376,
      "step": 7775
    },
    {
      "epoch": 17.316258351893097,
      "loss": 0.47175133228302,
      "loss_ce": 7.16680588084273e-05,
      "loss_iou": 0.21484375,
      "loss_num": 0.00848388671875,
      "loss_xval": 0.47265625,
      "num_input_tokens_seen": 435674376,
      "step": 7775
    },
    {
      "epoch": 17.318485523385302,
      "grad_norm": 18.95273208618164,
      "learning_rate": 1e-06,
      "loss": 0.3432,
      "num_input_tokens_seen": 435734040,
      "step": 7776
    },
    {
      "epoch": 17.318485523385302,
      "loss": 0.3788660764694214,
      "loss_ce": 8.189848449546844e-05,
      "loss_iou": 0.16796875,
      "loss_num": 0.0084228515625,
      "loss_xval": 0.37890625,
      "num_input_tokens_seen": 435734040,
      "step": 7776
    },
    {
      "epoch": 17.320712694877507,
      "grad_norm": 12.807289123535156,
      "learning_rate": 1e-06,
      "loss": 0.3866,
      "num_input_tokens_seen": 435790640,
      "step": 7777
    },
    {
      "epoch": 17.320712694877507,
      "loss": 0.34432125091552734,
      "loss_ce": 8.297587919514626e-05,
      "loss_iou": 0.150390625,
      "loss_num": 0.00860595703125,
      "loss_xval": 0.34375,
      "num_input_tokens_seen": 435790640,
      "step": 7777
    },
    {
      "epoch": 17.322939866369712,
      "grad_norm": 19.11941146850586,
      "learning_rate": 1e-06,
      "loss": 0.4882,
      "num_input_tokens_seen": 435847892,
      "step": 7778
    },
    {
      "epoch": 17.322939866369712,
      "loss": 0.5650254487991333,
      "loss_ce": 8.401433296967298e-05,
      "loss_iou": 0.228515625,
      "loss_num": 0.021728515625,
      "loss_xval": 0.56640625,
      "num_input_tokens_seen": 435847892,
      "step": 7778
    },
    {
      "epoch": 17.325167037861917,
      "grad_norm": 19.643787384033203,
      "learning_rate": 1e-06,
      "loss": 0.4666,
      "num_input_tokens_seen": 435903812,
      "step": 7779
    },
    {
      "epoch": 17.325167037861917,
      "loss": 0.39589911699295044,
      "loss_ce": 8.612501551397145e-05,
      "loss_iou": 0.166015625,
      "loss_num": 0.0126953125,
      "loss_xval": 0.396484375,
      "num_input_tokens_seen": 435903812,
      "step": 7779
    },
    {
      "epoch": 17.32739420935412,
      "grad_norm": 22.365102767944336,
      "learning_rate": 1e-06,
      "loss": 0.4378,
      "num_input_tokens_seen": 435960584,
      "step": 7780
    },
    {
      "epoch": 17.32739420935412,
      "loss": 0.39949724078178406,
      "loss_ce": 8.315553714055568e-05,
      "loss_iou": 0.1728515625,
      "loss_num": 0.01092529296875,
      "loss_xval": 0.3984375,
      "num_input_tokens_seen": 435960584,
      "step": 7780
    },
    {
      "epoch": 17.329621380846326,
      "grad_norm": 27.921226501464844,
      "learning_rate": 1e-06,
      "loss": 0.4324,
      "num_input_tokens_seen": 436016512,
      "step": 7781
    },
    {
      "epoch": 17.329621380846326,
      "loss": 0.5156280994415283,
      "loss_ce": 0.00012516917195171118,
      "loss_iou": 0.224609375,
      "loss_num": 0.01312255859375,
      "loss_xval": 0.515625,
      "num_input_tokens_seen": 436016512,
      "step": 7781
    },
    {
      "epoch": 17.33184855233853,
      "grad_norm": 24.685932159423828,
      "learning_rate": 1e-06,
      "loss": 0.4589,
      "num_input_tokens_seen": 436072952,
      "step": 7782
    },
    {
      "epoch": 17.33184855233853,
      "loss": 0.41859763860702515,
      "loss_ce": 7.957972411531955e-05,
      "loss_iou": 0.1953125,
      "loss_num": 0.0054931640625,
      "loss_xval": 0.41796875,
      "num_input_tokens_seen": 436072952,
      "step": 7782
    },
    {
      "epoch": 17.334075723830736,
      "grad_norm": 30.725080490112305,
      "learning_rate": 1e-06,
      "loss": 0.5439,
      "num_input_tokens_seen": 436127800,
      "step": 7783
    },
    {
      "epoch": 17.334075723830736,
      "loss": 0.4986262917518616,
      "loss_ce": 9.114194836001843e-05,
      "loss_iou": 0.2314453125,
      "loss_num": 0.007232666015625,
      "loss_xval": 0.498046875,
      "num_input_tokens_seen": 436127800,
      "step": 7783
    },
    {
      "epoch": 17.33630289532294,
      "grad_norm": 20.326061248779297,
      "learning_rate": 1e-06,
      "loss": 0.3683,
      "num_input_tokens_seen": 436187140,
      "step": 7784
    },
    {
      "epoch": 17.33630289532294,
      "loss": 0.24855035543441772,
      "loss_ce": 7.622801786055788e-05,
      "loss_iou": 0.11083984375,
      "loss_num": 0.00537109375,
      "loss_xval": 0.248046875,
      "num_input_tokens_seen": 436187140,
      "step": 7784
    },
    {
      "epoch": 17.338530066815146,
      "grad_norm": 21.3463077545166,
      "learning_rate": 1e-06,
      "loss": 0.4456,
      "num_input_tokens_seen": 436242216,
      "step": 7785
    },
    {
      "epoch": 17.338530066815146,
      "loss": 0.3916824460029602,
      "loss_ce": 8.088418690022081e-05,
      "loss_iou": 0.169921875,
      "loss_num": 0.01025390625,
      "loss_xval": 0.390625,
      "num_input_tokens_seen": 436242216,
      "step": 7785
    },
    {
      "epoch": 17.34075723830735,
      "grad_norm": 17.23738670349121,
      "learning_rate": 1e-06,
      "loss": 0.467,
      "num_input_tokens_seen": 436301260,
      "step": 7786
    },
    {
      "epoch": 17.34075723830735,
      "loss": 0.549955427646637,
      "loss_ce": 8.968937618192285e-05,
      "loss_iou": 0.21875,
      "loss_num": 0.0223388671875,
      "loss_xval": 0.55078125,
      "num_input_tokens_seen": 436301260,
      "step": 7786
    },
    {
      "epoch": 17.342984409799556,
      "grad_norm": 15.01907730102539,
      "learning_rate": 1e-06,
      "loss": 0.3399,
      "num_input_tokens_seen": 436357676,
      "step": 7787
    },
    {
      "epoch": 17.342984409799556,
      "loss": 0.34413158893585205,
      "loss_ce": 7.641033153049648e-05,
      "loss_iou": 0.1455078125,
      "loss_num": 0.010498046875,
      "loss_xval": 0.34375,
      "num_input_tokens_seen": 436357676,
      "step": 7787
    },
    {
      "epoch": 17.34521158129176,
      "grad_norm": 17.24225425720215,
      "learning_rate": 1e-06,
      "loss": 0.3079,
      "num_input_tokens_seen": 436412808,
      "step": 7788
    },
    {
      "epoch": 17.34521158129176,
      "loss": 0.2603484094142914,
      "loss_ce": 7.925922545837238e-05,
      "loss_iou": 0.12060546875,
      "loss_num": 0.0037994384765625,
      "loss_xval": 0.259765625,
      "num_input_tokens_seen": 436412808,
      "step": 7788
    },
    {
      "epoch": 17.347438752783965,
      "grad_norm": 17.269451141357422,
      "learning_rate": 1e-06,
      "loss": 0.457,
      "num_input_tokens_seen": 436468436,
      "step": 7789
    },
    {
      "epoch": 17.347438752783965,
      "loss": 0.6248146891593933,
      "loss_ce": 0.00011986246681772172,
      "loss_iou": 0.275390625,
      "loss_num": 0.0146484375,
      "loss_xval": 0.625,
      "num_input_tokens_seen": 436468436,
      "step": 7789
    },
    {
      "epoch": 17.34966592427617,
      "grad_norm": 28.061927795410156,
      "learning_rate": 1e-06,
      "loss": 0.3785,
      "num_input_tokens_seen": 436526876,
      "step": 7790
    },
    {
      "epoch": 17.34966592427617,
      "loss": 0.4654099941253662,
      "loss_ce": 7.796321006026119e-05,
      "loss_iou": 0.205078125,
      "loss_num": 0.0111083984375,
      "loss_xval": 0.46484375,
      "num_input_tokens_seen": 436526876,
      "step": 7790
    },
    {
      "epoch": 17.351893095768375,
      "grad_norm": 10.565998077392578,
      "learning_rate": 1e-06,
      "loss": 0.3896,
      "num_input_tokens_seen": 436583952,
      "step": 7791
    },
    {
      "epoch": 17.351893095768375,
      "loss": 0.30404287576675415,
      "loss_ce": 8.778244227869436e-05,
      "loss_iou": 0.134765625,
      "loss_num": 0.006805419921875,
      "loss_xval": 0.3046875,
      "num_input_tokens_seen": 436583952,
      "step": 7791
    },
    {
      "epoch": 17.35412026726058,
      "grad_norm": 15.32834243774414,
      "learning_rate": 1e-06,
      "loss": 0.4977,
      "num_input_tokens_seen": 436640224,
      "step": 7792
    },
    {
      "epoch": 17.35412026726058,
      "loss": 0.5239353179931641,
      "loss_ce": 7.061338692437857e-05,
      "loss_iou": 0.2216796875,
      "loss_num": 0.0162353515625,
      "loss_xval": 0.5234375,
      "num_input_tokens_seen": 436640224,
      "step": 7792
    },
    {
      "epoch": 17.356347438752785,
      "grad_norm": 34.02385711669922,
      "learning_rate": 1e-06,
      "loss": 0.6217,
      "num_input_tokens_seen": 436698680,
      "step": 7793
    },
    {
      "epoch": 17.356347438752785,
      "loss": 0.467492938041687,
      "loss_ce": 8.570231148041785e-05,
      "loss_iou": 0.21484375,
      "loss_num": 0.007537841796875,
      "loss_xval": 0.466796875,
      "num_input_tokens_seen": 436698680,
      "step": 7793
    },
    {
      "epoch": 17.35857461024499,
      "grad_norm": 20.18070411682129,
      "learning_rate": 1e-06,
      "loss": 0.408,
      "num_input_tokens_seen": 436753572,
      "step": 7794
    },
    {
      "epoch": 17.35857461024499,
      "loss": 0.2802870273590088,
      "loss_ce": 7.461066707037389e-05,
      "loss_iou": 0.119140625,
      "loss_num": 0.00823974609375,
      "loss_xval": 0.279296875,
      "num_input_tokens_seen": 436753572,
      "step": 7794
    },
    {
      "epoch": 17.360801781737194,
      "grad_norm": 18.094005584716797,
      "learning_rate": 1e-06,
      "loss": 0.4583,
      "num_input_tokens_seen": 436807624,
      "step": 7795
    },
    {
      "epoch": 17.360801781737194,
      "loss": 0.48153021931648254,
      "loss_ce": 8.488957246299833e-05,
      "loss_iou": 0.1982421875,
      "loss_num": 0.0172119140625,
      "loss_xval": 0.48046875,
      "num_input_tokens_seen": 436807624,
      "step": 7795
    },
    {
      "epoch": 17.3630289532294,
      "grad_norm": 13.45447826385498,
      "learning_rate": 1e-06,
      "loss": 0.3153,
      "num_input_tokens_seen": 436863496,
      "step": 7796
    },
    {
      "epoch": 17.3630289532294,
      "loss": 0.3097211420536041,
      "loss_ce": 8.980316488305107e-05,
      "loss_iou": 0.1328125,
      "loss_num": 0.00872802734375,
      "loss_xval": 0.310546875,
      "num_input_tokens_seen": 436863496,
      "step": 7796
    },
    {
      "epoch": 17.365256124721604,
      "grad_norm": 16.23236083984375,
      "learning_rate": 1e-06,
      "loss": 0.3146,
      "num_input_tokens_seen": 436920076,
      "step": 7797
    },
    {
      "epoch": 17.365256124721604,
      "loss": 0.3300950825214386,
      "loss_ce": 7.799094601068646e-05,
      "loss_iou": 0.1396484375,
      "loss_num": 0.010009765625,
      "loss_xval": 0.330078125,
      "num_input_tokens_seen": 436920076,
      "step": 7797
    },
    {
      "epoch": 17.36748329621381,
      "grad_norm": 16.27332305908203,
      "learning_rate": 1e-06,
      "loss": 0.4944,
      "num_input_tokens_seen": 436975300,
      "step": 7798
    },
    {
      "epoch": 17.36748329621381,
      "loss": 0.3892417550086975,
      "loss_ce": 8.160505967680365e-05,
      "loss_iou": 0.171875,
      "loss_num": 0.0091552734375,
      "loss_xval": 0.388671875,
      "num_input_tokens_seen": 436975300,
      "step": 7798
    },
    {
      "epoch": 17.369710467706014,
      "grad_norm": 13.5247163772583,
      "learning_rate": 1e-06,
      "loss": 0.508,
      "num_input_tokens_seen": 437030760,
      "step": 7799
    },
    {
      "epoch": 17.369710467706014,
      "loss": 0.5794344544410706,
      "loss_ce": 8.873187471181154e-05,
      "loss_iou": 0.259765625,
      "loss_num": 0.01171875,
      "loss_xval": 0.578125,
      "num_input_tokens_seen": 437030760,
      "step": 7799
    },
    {
      "epoch": 17.37193763919822,
      "grad_norm": 16.678621292114258,
      "learning_rate": 1e-06,
      "loss": 0.4149,
      "num_input_tokens_seen": 437086492,
      "step": 7800
    },
    {
      "epoch": 17.37193763919822,
      "loss": 0.6766193509101868,
      "loss_ce": 0.00010563358955550939,
      "loss_iou": 0.283203125,
      "loss_num": 0.0220947265625,
      "loss_xval": 0.67578125,
      "num_input_tokens_seen": 437086492,
      "step": 7800
    },
    {
      "epoch": 17.374164810690424,
      "grad_norm": 16.469213485717773,
      "learning_rate": 1e-06,
      "loss": 0.3798,
      "num_input_tokens_seen": 437142412,
      "step": 7801
    },
    {
      "epoch": 17.374164810690424,
      "loss": 0.413549542427063,
      "loss_ce": 9.739689994603395e-05,
      "loss_iou": 0.1962890625,
      "loss_num": 0.004150390625,
      "loss_xval": 0.4140625,
      "num_input_tokens_seen": 437142412,
      "step": 7801
    },
    {
      "epoch": 17.37639198218263,
      "grad_norm": 27.44245719909668,
      "learning_rate": 1e-06,
      "loss": 0.4157,
      "num_input_tokens_seen": 437199560,
      "step": 7802
    },
    {
      "epoch": 17.37639198218263,
      "loss": 0.3337703347206116,
      "loss_ce": 0.00027421663980931044,
      "loss_iou": 0.1494140625,
      "loss_num": 0.00689697265625,
      "loss_xval": 0.333984375,
      "num_input_tokens_seen": 437199560,
      "step": 7802
    },
    {
      "epoch": 17.378619153674833,
      "grad_norm": 24.417043685913086,
      "learning_rate": 1e-06,
      "loss": 0.3968,
      "num_input_tokens_seen": 437253536,
      "step": 7803
    },
    {
      "epoch": 17.378619153674833,
      "loss": 0.3829765319824219,
      "loss_ce": 7.25047430023551e-05,
      "loss_iou": 0.16015625,
      "loss_num": 0.01251220703125,
      "loss_xval": 0.3828125,
      "num_input_tokens_seen": 437253536,
      "step": 7803
    },
    {
      "epoch": 17.380846325167038,
      "grad_norm": 18.77859878540039,
      "learning_rate": 1e-06,
      "loss": 0.4349,
      "num_input_tokens_seen": 437311192,
      "step": 7804
    },
    {
      "epoch": 17.380846325167038,
      "loss": 0.4672635793685913,
      "loss_ce": 0.00010047997056972235,
      "loss_iou": 0.220703125,
      "loss_num": 0.005401611328125,
      "loss_xval": 0.466796875,
      "num_input_tokens_seen": 437311192,
      "step": 7804
    },
    {
      "epoch": 17.383073496659243,
      "grad_norm": 14.215685844421387,
      "learning_rate": 1e-06,
      "loss": 0.5266,
      "num_input_tokens_seen": 437368228,
      "step": 7805
    },
    {
      "epoch": 17.383073496659243,
      "loss": 0.6981062293052673,
      "loss_ce": 0.00035229395143687725,
      "loss_iou": 0.306640625,
      "loss_num": 0.0167236328125,
      "loss_xval": 0.69921875,
      "num_input_tokens_seen": 437368228,
      "step": 7805
    },
    {
      "epoch": 17.385300668151448,
      "grad_norm": 13.988162994384766,
      "learning_rate": 1e-06,
      "loss": 0.3419,
      "num_input_tokens_seen": 437426224,
      "step": 7806
    },
    {
      "epoch": 17.385300668151448,
      "loss": 0.3153742551803589,
      "loss_ce": 6.663544627372175e-05,
      "loss_iou": 0.1396484375,
      "loss_num": 0.0072021484375,
      "loss_xval": 0.314453125,
      "num_input_tokens_seen": 437426224,
      "step": 7806
    },
    {
      "epoch": 17.387527839643653,
      "grad_norm": 15.999502182006836,
      "learning_rate": 1e-06,
      "loss": 0.4759,
      "num_input_tokens_seen": 437483276,
      "step": 7807
    },
    {
      "epoch": 17.387527839643653,
      "loss": 0.4618630111217499,
      "loss_ce": 7.104083488229662e-05,
      "loss_iou": 0.185546875,
      "loss_num": 0.0181884765625,
      "loss_xval": 0.4609375,
      "num_input_tokens_seen": 437483276,
      "step": 7807
    },
    {
      "epoch": 17.389755011135858,
      "grad_norm": 19.859189987182617,
      "learning_rate": 1e-06,
      "loss": 0.5076,
      "num_input_tokens_seen": 437539780,
      "step": 7808
    },
    {
      "epoch": 17.389755011135858,
      "loss": 0.4627245366573334,
      "loss_ce": 7.806568464729935e-05,
      "loss_iou": 0.201171875,
      "loss_num": 0.01214599609375,
      "loss_xval": 0.462890625,
      "num_input_tokens_seen": 437539780,
      "step": 7808
    },
    {
      "epoch": 17.391982182628063,
      "grad_norm": 18.35259437561035,
      "learning_rate": 1e-06,
      "loss": 0.4693,
      "num_input_tokens_seen": 437593776,
      "step": 7809
    },
    {
      "epoch": 17.391982182628063,
      "loss": 0.6174068450927734,
      "loss_ce": 9.7297815955244e-05,
      "loss_iou": 0.275390625,
      "loss_num": 0.01336669921875,
      "loss_xval": 0.6171875,
      "num_input_tokens_seen": 437593776,
      "step": 7809
    },
    {
      "epoch": 17.394209354120267,
      "grad_norm": 22.22394371032715,
      "learning_rate": 1e-06,
      "loss": 0.4799,
      "num_input_tokens_seen": 437648068,
      "step": 7810
    },
    {
      "epoch": 17.394209354120267,
      "loss": 0.5823743939399719,
      "loss_ce": 0.00016007671365514398,
      "loss_iou": 0.255859375,
      "loss_num": 0.0137939453125,
      "loss_xval": 0.58203125,
      "num_input_tokens_seen": 437648068,
      "step": 7810
    },
    {
      "epoch": 17.396436525612472,
      "grad_norm": 44.06281280517578,
      "learning_rate": 1e-06,
      "loss": 0.5151,
      "num_input_tokens_seen": 437703580,
      "step": 7811
    },
    {
      "epoch": 17.396436525612472,
      "loss": 0.674892783164978,
      "loss_ce": 8.813235035631806e-05,
      "loss_iou": 0.28125,
      "loss_num": 0.022705078125,
      "loss_xval": 0.67578125,
      "num_input_tokens_seen": 437703580,
      "step": 7811
    },
    {
      "epoch": 17.398663697104677,
      "grad_norm": 23.47769546508789,
      "learning_rate": 1e-06,
      "loss": 0.5547,
      "num_input_tokens_seen": 437757572,
      "step": 7812
    },
    {
      "epoch": 17.398663697104677,
      "loss": 0.5415530204772949,
      "loss_ce": 0.0001712157973088324,
      "loss_iou": 0.240234375,
      "loss_num": 0.0120849609375,
      "loss_xval": 0.54296875,
      "num_input_tokens_seen": 437757572,
      "step": 7812
    },
    {
      "epoch": 17.400890868596882,
      "grad_norm": 17.036544799804688,
      "learning_rate": 1e-06,
      "loss": 0.4651,
      "num_input_tokens_seen": 437810472,
      "step": 7813
    },
    {
      "epoch": 17.400890868596882,
      "loss": 0.3841232657432556,
      "loss_ce": 9.007104381453246e-05,
      "loss_iou": 0.166015625,
      "loss_num": 0.0103759765625,
      "loss_xval": 0.384765625,
      "num_input_tokens_seen": 437810472,
      "step": 7813
    },
    {
      "epoch": 17.403118040089087,
      "grad_norm": 17.07404327392578,
      "learning_rate": 1e-06,
      "loss": 0.3801,
      "num_input_tokens_seen": 437868112,
      "step": 7814
    },
    {
      "epoch": 17.403118040089087,
      "loss": 0.4032171964645386,
      "loss_ce": 7.997561624506488e-05,
      "loss_iou": 0.1689453125,
      "loss_num": 0.01300048828125,
      "loss_xval": 0.40234375,
      "num_input_tokens_seen": 437868112,
      "step": 7814
    },
    {
      "epoch": 17.40534521158129,
      "grad_norm": 13.92210578918457,
      "learning_rate": 1e-06,
      "loss": 0.3186,
      "num_input_tokens_seen": 437923616,
      "step": 7815
    },
    {
      "epoch": 17.40534521158129,
      "loss": 0.40229788422584534,
      "loss_ce": 7.62125855544582e-05,
      "loss_iou": 0.16796875,
      "loss_num": 0.01312255859375,
      "loss_xval": 0.40234375,
      "num_input_tokens_seen": 437923616,
      "step": 7815
    },
    {
      "epoch": 17.407572383073497,
      "grad_norm": 19.43670082092285,
      "learning_rate": 1e-06,
      "loss": 0.4878,
      "num_input_tokens_seen": 437979980,
      "step": 7816
    },
    {
      "epoch": 17.407572383073497,
      "loss": 0.7343358397483826,
      "loss_ce": 8.290271944133565e-05,
      "loss_iou": 0.306640625,
      "loss_num": 0.02392578125,
      "loss_xval": 0.734375,
      "num_input_tokens_seen": 437979980,
      "step": 7816
    },
    {
      "epoch": 17.4097995545657,
      "grad_norm": 13.963506698608398,
      "learning_rate": 1e-06,
      "loss": 0.4524,
      "num_input_tokens_seen": 438033348,
      "step": 7817
    },
    {
      "epoch": 17.4097995545657,
      "loss": 0.41097211837768555,
      "loss_ce": 8.345420064870268e-05,
      "loss_iou": 0.1748046875,
      "loss_num": 0.0123291015625,
      "loss_xval": 0.41015625,
      "num_input_tokens_seen": 438033348,
      "step": 7817
    },
    {
      "epoch": 17.412026726057906,
      "grad_norm": 18.236024856567383,
      "learning_rate": 1e-06,
      "loss": 0.3729,
      "num_input_tokens_seen": 438091792,
      "step": 7818
    },
    {
      "epoch": 17.412026726057906,
      "loss": 0.31764712929725647,
      "loss_ce": 8.121843711705878e-05,
      "loss_iou": 0.1376953125,
      "loss_num": 0.00848388671875,
      "loss_xval": 0.318359375,
      "num_input_tokens_seen": 438091792,
      "step": 7818
    },
    {
      "epoch": 17.41425389755011,
      "grad_norm": 19.653343200683594,
      "learning_rate": 1e-06,
      "loss": 0.2597,
      "num_input_tokens_seen": 438148088,
      "step": 7819
    },
    {
      "epoch": 17.41425389755011,
      "loss": 0.3163628578186035,
      "loss_ce": 7.866387022659183e-05,
      "loss_iou": 0.14453125,
      "loss_num": 0.00531005859375,
      "loss_xval": 0.31640625,
      "num_input_tokens_seen": 438148088,
      "step": 7819
    },
    {
      "epoch": 17.416481069042316,
      "grad_norm": 16.326854705810547,
      "learning_rate": 1e-06,
      "loss": 0.4654,
      "num_input_tokens_seen": 438205184,
      "step": 7820
    },
    {
      "epoch": 17.416481069042316,
      "loss": 0.4878746271133423,
      "loss_ce": 8.16921892692335e-05,
      "loss_iou": 0.2119140625,
      "loss_num": 0.0130615234375,
      "loss_xval": 0.48828125,
      "num_input_tokens_seen": 438205184,
      "step": 7820
    },
    {
      "epoch": 17.41870824053452,
      "grad_norm": 29.167247772216797,
      "learning_rate": 1e-06,
      "loss": 0.3659,
      "num_input_tokens_seen": 438260072,
      "step": 7821
    },
    {
      "epoch": 17.41870824053452,
      "loss": 0.40425050258636475,
      "loss_ce": 7.568299042759463e-05,
      "loss_iou": 0.185546875,
      "loss_num": 0.00665283203125,
      "loss_xval": 0.404296875,
      "num_input_tokens_seen": 438260072,
      "step": 7821
    },
    {
      "epoch": 17.420935412026726,
      "grad_norm": 18.79923439025879,
      "learning_rate": 1e-06,
      "loss": 0.4383,
      "num_input_tokens_seen": 438314384,
      "step": 7822
    },
    {
      "epoch": 17.420935412026726,
      "loss": 0.5045883655548096,
      "loss_ce": 7.173811900429428e-05,
      "loss_iou": 0.1962890625,
      "loss_num": 0.022216796875,
      "loss_xval": 0.50390625,
      "num_input_tokens_seen": 438314384,
      "step": 7822
    },
    {
      "epoch": 17.42316258351893,
      "grad_norm": 15.445906639099121,
      "learning_rate": 1e-06,
      "loss": 0.3509,
      "num_input_tokens_seen": 438370632,
      "step": 7823
    },
    {
      "epoch": 17.42316258351893,
      "loss": 0.43685096502304077,
      "loss_ce": 8.341444481629878e-05,
      "loss_iou": 0.1982421875,
      "loss_num": 0.00830078125,
      "loss_xval": 0.4375,
      "num_input_tokens_seen": 438370632,
      "step": 7823
    },
    {
      "epoch": 17.425389755011135,
      "grad_norm": 32.27328109741211,
      "learning_rate": 1e-06,
      "loss": 0.3304,
      "num_input_tokens_seen": 438423192,
      "step": 7824
    },
    {
      "epoch": 17.425389755011135,
      "loss": 0.33272212743759155,
      "loss_ce": 8.052479824982584e-05,
      "loss_iou": 0.146484375,
      "loss_num": 0.00811767578125,
      "loss_xval": 0.33203125,
      "num_input_tokens_seen": 438423192,
      "step": 7824
    },
    {
      "epoch": 17.42761692650334,
      "grad_norm": 23.52182960510254,
      "learning_rate": 1e-06,
      "loss": 0.6017,
      "num_input_tokens_seen": 438477292,
      "step": 7825
    },
    {
      "epoch": 17.42761692650334,
      "loss": 0.5533384680747986,
      "loss_ce": 0.00011578819248825312,
      "loss_iou": 0.2578125,
      "loss_num": 0.007781982421875,
      "loss_xval": 0.5546875,
      "num_input_tokens_seen": 438477292,
      "step": 7825
    },
    {
      "epoch": 17.429844097995545,
      "grad_norm": 26.976110458374023,
      "learning_rate": 1e-06,
      "loss": 0.4182,
      "num_input_tokens_seen": 438531796,
      "step": 7826
    },
    {
      "epoch": 17.429844097995545,
      "loss": 0.4023064076900482,
      "loss_ce": 8.469966996926814e-05,
      "loss_iou": 0.17578125,
      "loss_num": 0.0101318359375,
      "loss_xval": 0.40234375,
      "num_input_tokens_seen": 438531796,
      "step": 7826
    },
    {
      "epoch": 17.43207126948775,
      "grad_norm": 19.38138198852539,
      "learning_rate": 1e-06,
      "loss": 0.4998,
      "num_input_tokens_seen": 438589904,
      "step": 7827
    },
    {
      "epoch": 17.43207126948775,
      "loss": 0.3881371021270752,
      "loss_ce": 7.558944344054908e-05,
      "loss_iou": 0.1611328125,
      "loss_num": 0.0133056640625,
      "loss_xval": 0.388671875,
      "num_input_tokens_seen": 438589904,
      "step": 7827
    },
    {
      "epoch": 17.434298440979955,
      "grad_norm": 31.262649536132812,
      "learning_rate": 1e-06,
      "loss": 0.3436,
      "num_input_tokens_seen": 438643728,
      "step": 7828
    },
    {
      "epoch": 17.434298440979955,
      "loss": 0.3436218798160553,
      "loss_ce": 0.00011602583253988996,
      "loss_iou": 0.1552734375,
      "loss_num": 0.006744384765625,
      "loss_xval": 0.34375,
      "num_input_tokens_seen": 438643728,
      "step": 7828
    },
    {
      "epoch": 17.43652561247216,
      "grad_norm": 14.571344375610352,
      "learning_rate": 1e-06,
      "loss": 0.3795,
      "num_input_tokens_seen": 438701432,
      "step": 7829
    },
    {
      "epoch": 17.43652561247216,
      "loss": 0.4284363389015198,
      "loss_ce": 9.162020432995632e-05,
      "loss_iou": 0.1845703125,
      "loss_num": 0.0120849609375,
      "loss_xval": 0.427734375,
      "num_input_tokens_seen": 438701432,
      "step": 7829
    },
    {
      "epoch": 17.438752783964365,
      "grad_norm": 25.686256408691406,
      "learning_rate": 1e-06,
      "loss": 0.3957,
      "num_input_tokens_seen": 438755012,
      "step": 7830
    },
    {
      "epoch": 17.438752783964365,
      "loss": 0.42792049050331116,
      "loss_ce": 0.00012508549843914807,
      "loss_iou": 0.181640625,
      "loss_num": 0.01312255859375,
      "loss_xval": 0.427734375,
      "num_input_tokens_seen": 438755012,
      "step": 7830
    },
    {
      "epoch": 17.44097995545657,
      "grad_norm": 17.335966110229492,
      "learning_rate": 1e-06,
      "loss": 0.4715,
      "num_input_tokens_seen": 438807680,
      "step": 7831
    },
    {
      "epoch": 17.44097995545657,
      "loss": 0.4803728461265564,
      "loss_ce": 8.723569044377655e-05,
      "loss_iou": 0.203125,
      "loss_num": 0.01483154296875,
      "loss_xval": 0.48046875,
      "num_input_tokens_seen": 438807680,
      "step": 7831
    },
    {
      "epoch": 17.443207126948774,
      "grad_norm": 24.54792022705078,
      "learning_rate": 1e-06,
      "loss": 0.4759,
      "num_input_tokens_seen": 438864696,
      "step": 7832
    },
    {
      "epoch": 17.443207126948774,
      "loss": 0.5594247579574585,
      "loss_ce": 9.856373799266294e-05,
      "loss_iou": 0.2470703125,
      "loss_num": 0.01318359375,
      "loss_xval": 0.55859375,
      "num_input_tokens_seen": 438864696,
      "step": 7832
    },
    {
      "epoch": 17.44543429844098,
      "grad_norm": 21.406185150146484,
      "learning_rate": 1e-06,
      "loss": 0.4039,
      "num_input_tokens_seen": 438918952,
      "step": 7833
    },
    {
      "epoch": 17.44543429844098,
      "loss": 0.5075805187225342,
      "loss_ce": 0.00013420640607364476,
      "loss_iou": 0.220703125,
      "loss_num": 0.01348876953125,
      "loss_xval": 0.5078125,
      "num_input_tokens_seen": 438918952,
      "step": 7833
    },
    {
      "epoch": 17.447661469933184,
      "grad_norm": 21.712177276611328,
      "learning_rate": 1e-06,
      "loss": 0.264,
      "num_input_tokens_seen": 438976448,
      "step": 7834
    },
    {
      "epoch": 17.447661469933184,
      "loss": 0.1915939748287201,
      "loss_ce": 6.564041541423649e-05,
      "loss_iou": 0.08203125,
      "loss_num": 0.00555419921875,
      "loss_xval": 0.19140625,
      "num_input_tokens_seen": 438976448,
      "step": 7834
    },
    {
      "epoch": 17.44988864142539,
      "grad_norm": 16.437280654907227,
      "learning_rate": 1e-06,
      "loss": 0.3436,
      "num_input_tokens_seen": 439030240,
      "step": 7835
    },
    {
      "epoch": 17.44988864142539,
      "loss": 0.2427883744239807,
      "loss_ce": 8.206407801480964e-05,
      "loss_iou": 0.10791015625,
      "loss_num": 0.005340576171875,
      "loss_xval": 0.2431640625,
      "num_input_tokens_seen": 439030240,
      "step": 7835
    },
    {
      "epoch": 17.452115812917594,
      "grad_norm": 16.297969818115234,
      "learning_rate": 1e-06,
      "loss": 0.3661,
      "num_input_tokens_seen": 439087720,
      "step": 7836
    },
    {
      "epoch": 17.452115812917594,
      "loss": 0.3819359540939331,
      "loss_ce": 0.00010002277849707752,
      "loss_iou": 0.1767578125,
      "loss_num": 0.005767822265625,
      "loss_xval": 0.3828125,
      "num_input_tokens_seen": 439087720,
      "step": 7836
    },
    {
      "epoch": 17.4543429844098,
      "grad_norm": 14.235069274902344,
      "learning_rate": 1e-06,
      "loss": 0.4091,
      "num_input_tokens_seen": 439142152,
      "step": 7837
    },
    {
      "epoch": 17.4543429844098,
      "loss": 0.363969087600708,
      "loss_ce": 7.749867654638365e-05,
      "loss_iou": 0.150390625,
      "loss_num": 0.01251220703125,
      "loss_xval": 0.36328125,
      "num_input_tokens_seen": 439142152,
      "step": 7837
    },
    {
      "epoch": 17.456570155902003,
      "grad_norm": 18.100412368774414,
      "learning_rate": 1e-06,
      "loss": 0.4347,
      "num_input_tokens_seen": 439198644,
      "step": 7838
    },
    {
      "epoch": 17.456570155902003,
      "loss": 0.47042351961135864,
      "loss_ce": 8.66147456690669e-05,
      "loss_iou": 0.2001953125,
      "loss_num": 0.01422119140625,
      "loss_xval": 0.470703125,
      "num_input_tokens_seen": 439198644,
      "step": 7838
    },
    {
      "epoch": 17.45879732739421,
      "grad_norm": 17.923023223876953,
      "learning_rate": 1e-06,
      "loss": 0.5361,
      "num_input_tokens_seen": 439253288,
      "step": 7839
    },
    {
      "epoch": 17.45879732739421,
      "loss": 0.31425806879997253,
      "loss_ce": 0.0004763224278576672,
      "loss_iou": 0.140625,
      "loss_num": 0.006500244140625,
      "loss_xval": 0.314453125,
      "num_input_tokens_seen": 439253288,
      "step": 7839
    },
    {
      "epoch": 17.461024498886413,
      "grad_norm": 19.96112632751465,
      "learning_rate": 1e-06,
      "loss": 0.4533,
      "num_input_tokens_seen": 439308972,
      "step": 7840
    },
    {
      "epoch": 17.461024498886413,
      "loss": 0.40413880348205566,
      "loss_ce": 8.605894981883466e-05,
      "loss_iou": 0.1875,
      "loss_num": 0.00567626953125,
      "loss_xval": 0.404296875,
      "num_input_tokens_seen": 439308972,
      "step": 7840
    },
    {
      "epoch": 17.463251670378618,
      "grad_norm": 15.366341590881348,
      "learning_rate": 1e-06,
      "loss": 0.2831,
      "num_input_tokens_seen": 439367272,
      "step": 7841
    },
    {
      "epoch": 17.463251670378618,
      "loss": 0.33577966690063477,
      "loss_ce": 8.628957584733143e-05,
      "loss_iou": 0.138671875,
      "loss_num": 0.0115966796875,
      "loss_xval": 0.3359375,
      "num_input_tokens_seen": 439367272,
      "step": 7841
    },
    {
      "epoch": 17.465478841870823,
      "grad_norm": 15.852892875671387,
      "learning_rate": 1e-06,
      "loss": 0.5239,
      "num_input_tokens_seen": 439424584,
      "step": 7842
    },
    {
      "epoch": 17.465478841870823,
      "loss": 0.6492735147476196,
      "loss_ce": 0.00010354960249969736,
      "loss_iou": 0.279296875,
      "loss_num": 0.018310546875,
      "loss_xval": 0.6484375,
      "num_input_tokens_seen": 439424584,
      "step": 7842
    },
    {
      "epoch": 17.467706013363028,
      "grad_norm": 14.707501411437988,
      "learning_rate": 1e-06,
      "loss": 0.3847,
      "num_input_tokens_seen": 439479788,
      "step": 7843
    },
    {
      "epoch": 17.467706013363028,
      "loss": 0.31831464171409607,
      "loss_ce": 7.733616075711325e-05,
      "loss_iou": 0.1357421875,
      "loss_num": 0.00958251953125,
      "loss_xval": 0.318359375,
      "num_input_tokens_seen": 439479788,
      "step": 7843
    },
    {
      "epoch": 17.469933184855233,
      "grad_norm": 42.824195861816406,
      "learning_rate": 1e-06,
      "loss": 0.4873,
      "num_input_tokens_seen": 439536572,
      "step": 7844
    },
    {
      "epoch": 17.469933184855233,
      "loss": 0.3835739493370056,
      "loss_ce": 9.006427717395127e-05,
      "loss_iou": 0.177734375,
      "loss_num": 0.005767822265625,
      "loss_xval": 0.3828125,
      "num_input_tokens_seen": 439536572,
      "step": 7844
    },
    {
      "epoch": 17.472160356347437,
      "grad_norm": 14.055753707885742,
      "learning_rate": 1e-06,
      "loss": 0.564,
      "num_input_tokens_seen": 439594084,
      "step": 7845
    },
    {
      "epoch": 17.472160356347437,
      "loss": 0.6532701253890991,
      "loss_ce": 7.185361755546182e-05,
      "loss_iou": 0.265625,
      "loss_num": 0.0245361328125,
      "loss_xval": 0.65234375,
      "num_input_tokens_seen": 439594084,
      "step": 7845
    },
    {
      "epoch": 17.474387527839642,
      "grad_norm": 24.9837589263916,
      "learning_rate": 1e-06,
      "loss": 0.4575,
      "num_input_tokens_seen": 439649312,
      "step": 7846
    },
    {
      "epoch": 17.474387527839642,
      "loss": 0.45571112632751465,
      "loss_ce": 0.000144733494380489,
      "loss_iou": 0.1845703125,
      "loss_num": 0.0174560546875,
      "loss_xval": 0.455078125,
      "num_input_tokens_seen": 439649312,
      "step": 7846
    },
    {
      "epoch": 17.476614699331847,
      "grad_norm": 12.236130714416504,
      "learning_rate": 1e-06,
      "loss": 0.3456,
      "num_input_tokens_seen": 439705312,
      "step": 7847
    },
    {
      "epoch": 17.476614699331847,
      "loss": 0.4908040165901184,
      "loss_ce": 8.134340168908238e-05,
      "loss_iou": 0.2158203125,
      "loss_num": 0.011962890625,
      "loss_xval": 0.490234375,
      "num_input_tokens_seen": 439705312,
      "step": 7847
    },
    {
      "epoch": 17.478841870824052,
      "grad_norm": 15.341350555419922,
      "learning_rate": 1e-06,
      "loss": 0.2637,
      "num_input_tokens_seen": 439762020,
      "step": 7848
    },
    {
      "epoch": 17.478841870824052,
      "loss": 0.2432081252336502,
      "loss_ce": 0.00010509882849873975,
      "loss_iou": 0.10302734375,
      "loss_num": 0.00750732421875,
      "loss_xval": 0.2431640625,
      "num_input_tokens_seen": 439762020,
      "step": 7848
    },
    {
      "epoch": 17.481069042316257,
      "grad_norm": 20.993581771850586,
      "learning_rate": 1e-06,
      "loss": 0.3698,
      "num_input_tokens_seen": 439818004,
      "step": 7849
    },
    {
      "epoch": 17.481069042316257,
      "loss": 0.41051971912384033,
      "loss_ce": 0.00011930659093195572,
      "loss_iou": 0.1904296875,
      "loss_num": 0.00567626953125,
      "loss_xval": 0.41015625,
      "num_input_tokens_seen": 439818004,
      "step": 7849
    },
    {
      "epoch": 17.48329621380846,
      "grad_norm": 26.052579879760742,
      "learning_rate": 1e-06,
      "loss": 0.4034,
      "num_input_tokens_seen": 439876168,
      "step": 7850
    },
    {
      "epoch": 17.48329621380846,
      "loss": 0.42060598731040955,
      "loss_ce": 7.376498251687735e-05,
      "loss_iou": 0.1875,
      "loss_num": 0.0091552734375,
      "loss_xval": 0.419921875,
      "num_input_tokens_seen": 439876168,
      "step": 7850
    },
    {
      "epoch": 17.485523385300667,
      "grad_norm": 15.98454475402832,
      "learning_rate": 1e-06,
      "loss": 0.4183,
      "num_input_tokens_seen": 439929952,
      "step": 7851
    },
    {
      "epoch": 17.485523385300667,
      "loss": 0.3435441553592682,
      "loss_ce": 6.880345608806238e-05,
      "loss_iou": 0.142578125,
      "loss_num": 0.01171875,
      "loss_xval": 0.34375,
      "num_input_tokens_seen": 439929952,
      "step": 7851
    },
    {
      "epoch": 17.48775055679287,
      "grad_norm": 27.797683715820312,
      "learning_rate": 1e-06,
      "loss": 0.3736,
      "num_input_tokens_seen": 439986488,
      "step": 7852
    },
    {
      "epoch": 17.48775055679287,
      "loss": 0.3528550863265991,
      "loss_ce": 7.186994480434805e-05,
      "loss_iou": 0.162109375,
      "loss_num": 0.005645751953125,
      "loss_xval": 0.353515625,
      "num_input_tokens_seen": 439986488,
      "step": 7852
    },
    {
      "epoch": 17.489977728285076,
      "grad_norm": 17.55838966369629,
      "learning_rate": 1e-06,
      "loss": 0.4012,
      "num_input_tokens_seen": 440043948,
      "step": 7853
    },
    {
      "epoch": 17.489977728285076,
      "loss": 0.28866440057754517,
      "loss_ce": 9.020272409543395e-05,
      "loss_iou": 0.12890625,
      "loss_num": 0.006134033203125,
      "loss_xval": 0.2890625,
      "num_input_tokens_seen": 440043948,
      "step": 7853
    },
    {
      "epoch": 17.49220489977728,
      "grad_norm": 16.255395889282227,
      "learning_rate": 1e-06,
      "loss": 0.5026,
      "num_input_tokens_seen": 440101172,
      "step": 7854
    },
    {
      "epoch": 17.49220489977728,
      "loss": 0.4820084571838379,
      "loss_ce": 7.487165566999465e-05,
      "loss_iou": 0.220703125,
      "loss_num": 0.00836181640625,
      "loss_xval": 0.482421875,
      "num_input_tokens_seen": 440101172,
      "step": 7854
    },
    {
      "epoch": 17.494432071269486,
      "grad_norm": 13.98792839050293,
      "learning_rate": 1e-06,
      "loss": 0.3042,
      "num_input_tokens_seen": 440158728,
      "step": 7855
    },
    {
      "epoch": 17.494432071269486,
      "loss": 0.27856212854385376,
      "loss_ce": 8.1585232692305e-05,
      "loss_iou": 0.1279296875,
      "loss_num": 0.00457763671875,
      "loss_xval": 0.279296875,
      "num_input_tokens_seen": 440158728,
      "step": 7855
    },
    {
      "epoch": 17.49665924276169,
      "grad_norm": 15.809954643249512,
      "learning_rate": 1e-06,
      "loss": 0.4814,
      "num_input_tokens_seen": 440214632,
      "step": 7856
    },
    {
      "epoch": 17.49665924276169,
      "loss": 0.5641576647758484,
      "loss_ce": 7.077249756548554e-05,
      "loss_iou": 0.20703125,
      "loss_num": 0.030029296875,
      "loss_xval": 0.5625,
      "num_input_tokens_seen": 440214632,
      "step": 7856
    },
    {
      "epoch": 17.498886414253896,
      "grad_norm": 19.271411895751953,
      "learning_rate": 1e-06,
      "loss": 0.3269,
      "num_input_tokens_seen": 440273540,
      "step": 7857
    },
    {
      "epoch": 17.498886414253896,
      "loss": 0.36080411076545715,
      "loss_ce": 8.634180994704366e-05,
      "loss_iou": 0.15234375,
      "loss_num": 0.01123046875,
      "loss_xval": 0.361328125,
      "num_input_tokens_seen": 440273540,
      "step": 7857
    },
    {
      "epoch": 17.501113585746104,
      "grad_norm": 28.240005493164062,
      "learning_rate": 1e-06,
      "loss": 0.3847,
      "num_input_tokens_seen": 440328284,
      "step": 7858
    },
    {
      "epoch": 17.501113585746104,
      "loss": 0.3158857226371765,
      "loss_ce": 8.983007137430832e-05,
      "loss_iou": 0.1416015625,
      "loss_num": 0.006378173828125,
      "loss_xval": 0.31640625,
      "num_input_tokens_seen": 440328284,
      "step": 7858
    },
    {
      "epoch": 17.50334075723831,
      "grad_norm": 21.047122955322266,
      "learning_rate": 1e-06,
      "loss": 0.566,
      "num_input_tokens_seen": 440384876,
      "step": 7859
    },
    {
      "epoch": 17.50334075723831,
      "loss": 0.5007518529891968,
      "loss_ce": 0.00014154997188597918,
      "loss_iou": 0.1943359375,
      "loss_num": 0.0225830078125,
      "loss_xval": 0.5,
      "num_input_tokens_seen": 440384876,
      "step": 7859
    },
    {
      "epoch": 17.505567928730514,
      "grad_norm": 14.648688316345215,
      "learning_rate": 1e-06,
      "loss": 0.3644,
      "num_input_tokens_seen": 440440236,
      "step": 7860
    },
    {
      "epoch": 17.505567928730514,
      "loss": 0.33874061703681946,
      "loss_ce": 0.00020912080071866512,
      "loss_iou": 0.1484375,
      "loss_num": 0.0081787109375,
      "loss_xval": 0.337890625,
      "num_input_tokens_seen": 440440236,
      "step": 7860
    },
    {
      "epoch": 17.50779510022272,
      "grad_norm": 14.066313743591309,
      "learning_rate": 1e-06,
      "loss": 0.3047,
      "num_input_tokens_seen": 440495832,
      "step": 7861
    },
    {
      "epoch": 17.50779510022272,
      "loss": 0.2872577905654907,
      "loss_ce": 8.740053453948349e-05,
      "loss_iou": 0.11474609375,
      "loss_num": 0.011474609375,
      "loss_xval": 0.287109375,
      "num_input_tokens_seen": 440495832,
      "step": 7861
    },
    {
      "epoch": 17.510022271714924,
      "grad_norm": 14.307774543762207,
      "learning_rate": 1e-06,
      "loss": 0.3997,
      "num_input_tokens_seen": 440551848,
      "step": 7862
    },
    {
      "epoch": 17.510022271714924,
      "loss": 0.47919386625289917,
      "loss_ce": 6.788225437048823e-05,
      "loss_iou": 0.197265625,
      "loss_num": 0.01708984375,
      "loss_xval": 0.478515625,
      "num_input_tokens_seen": 440551848,
      "step": 7862
    },
    {
      "epoch": 17.51224944320713,
      "grad_norm": 21.205509185791016,
      "learning_rate": 1e-06,
      "loss": 0.359,
      "num_input_tokens_seen": 440605772,
      "step": 7863
    },
    {
      "epoch": 17.51224944320713,
      "loss": 0.2875545918941498,
      "loss_ce": 7.898983312770724e-05,
      "loss_iou": 0.125,
      "loss_num": 0.007415771484375,
      "loss_xval": 0.287109375,
      "num_input_tokens_seen": 440605772,
      "step": 7863
    },
    {
      "epoch": 17.514476614699333,
      "grad_norm": 17.232234954833984,
      "learning_rate": 1e-06,
      "loss": 0.4147,
      "num_input_tokens_seen": 440664340,
      "step": 7864
    },
    {
      "epoch": 17.514476614699333,
      "loss": 0.43732959032058716,
      "loss_ce": 7.375919085461646e-05,
      "loss_iou": 0.193359375,
      "loss_num": 0.01025390625,
      "loss_xval": 0.4375,
      "num_input_tokens_seen": 440664340,
      "step": 7864
    },
    {
      "epoch": 17.51670378619154,
      "grad_norm": 13.662554740905762,
      "learning_rate": 1e-06,
      "loss": 0.351,
      "num_input_tokens_seen": 440720316,
      "step": 7865
    },
    {
      "epoch": 17.51670378619154,
      "loss": 0.3087965250015259,
      "loss_ce": 8.073220669757575e-05,
      "loss_iou": 0.1279296875,
      "loss_num": 0.010498046875,
      "loss_xval": 0.30859375,
      "num_input_tokens_seen": 440720316,
      "step": 7865
    },
    {
      "epoch": 17.518930957683743,
      "grad_norm": 27.160282135009766,
      "learning_rate": 1e-06,
      "loss": 0.4684,
      "num_input_tokens_seen": 440777844,
      "step": 7866
    },
    {
      "epoch": 17.518930957683743,
      "loss": 0.5026195049285889,
      "loss_ce": 8.654060366097838e-05,
      "loss_iou": 0.2158203125,
      "loss_num": 0.01409912109375,
      "loss_xval": 0.50390625,
      "num_input_tokens_seen": 440777844,
      "step": 7866
    },
    {
      "epoch": 17.521158129175948,
      "grad_norm": 16.337854385375977,
      "learning_rate": 1e-06,
      "loss": 0.4732,
      "num_input_tokens_seen": 440835848,
      "step": 7867
    },
    {
      "epoch": 17.521158129175948,
      "loss": 0.5095036625862122,
      "loss_ce": 0.00010422736522741616,
      "loss_iou": 0.2138671875,
      "loss_num": 0.016357421875,
      "loss_xval": 0.5078125,
      "num_input_tokens_seen": 440835848,
      "step": 7867
    },
    {
      "epoch": 17.523385300668153,
      "grad_norm": 16.995683670043945,
      "learning_rate": 1e-06,
      "loss": 0.4265,
      "num_input_tokens_seen": 440894060,
      "step": 7868
    },
    {
      "epoch": 17.523385300668153,
      "loss": 0.2940109968185425,
      "loss_ce": 6.568758544744924e-05,
      "loss_iou": 0.126953125,
      "loss_num": 0.0081787109375,
      "loss_xval": 0.29296875,
      "num_input_tokens_seen": 440894060,
      "step": 7868
    },
    {
      "epoch": 17.525612472160358,
      "grad_norm": 19.353191375732422,
      "learning_rate": 1e-06,
      "loss": 0.4849,
      "num_input_tokens_seen": 440950212,
      "step": 7869
    },
    {
      "epoch": 17.525612472160358,
      "loss": 0.3159940838813782,
      "loss_ce": 7.612561603309587e-05,
      "loss_iou": 0.125,
      "loss_num": 0.01318359375,
      "loss_xval": 0.31640625,
      "num_input_tokens_seen": 440950212,
      "step": 7869
    },
    {
      "epoch": 17.527839643652563,
      "grad_norm": 29.26142692565918,
      "learning_rate": 1e-06,
      "loss": 0.5345,
      "num_input_tokens_seen": 441005288,
      "step": 7870
    },
    {
      "epoch": 17.527839643652563,
      "loss": 0.500335693359375,
      "loss_ce": 9.152606799034402e-05,
      "loss_iou": 0.2109375,
      "loss_num": 0.015625,
      "loss_xval": 0.5,
      "num_input_tokens_seen": 441005288,
      "step": 7870
    },
    {
      "epoch": 17.530066815144767,
      "grad_norm": 21.974380493164062,
      "learning_rate": 1e-06,
      "loss": 0.4412,
      "num_input_tokens_seen": 441060244,
      "step": 7871
    },
    {
      "epoch": 17.530066815144767,
      "loss": 0.42531251907348633,
      "loss_ce": 8.057255035964772e-05,
      "loss_iou": 0.19140625,
      "loss_num": 0.00848388671875,
      "loss_xval": 0.42578125,
      "num_input_tokens_seen": 441060244,
      "step": 7871
    },
    {
      "epoch": 17.532293986636972,
      "grad_norm": 12.20707893371582,
      "learning_rate": 1e-06,
      "loss": 0.3677,
      "num_input_tokens_seen": 441116740,
      "step": 7872
    },
    {
      "epoch": 17.532293986636972,
      "loss": 0.22715777158737183,
      "loss_ce": 6.883872265461832e-05,
      "loss_iou": 0.0908203125,
      "loss_num": 0.0091552734375,
      "loss_xval": 0.2275390625,
      "num_input_tokens_seen": 441116740,
      "step": 7872
    },
    {
      "epoch": 17.534521158129177,
      "grad_norm": 55.13029479980469,
      "learning_rate": 1e-06,
      "loss": 0.4155,
      "num_input_tokens_seen": 441171904,
      "step": 7873
    },
    {
      "epoch": 17.534521158129177,
      "loss": 0.3325577676296234,
      "loss_ce": 9.927057544700801e-05,
      "loss_iou": 0.140625,
      "loss_num": 0.01025390625,
      "loss_xval": 0.33203125,
      "num_input_tokens_seen": 441171904,
      "step": 7873
    },
    {
      "epoch": 17.536748329621382,
      "grad_norm": 16.75377082824707,
      "learning_rate": 1e-06,
      "loss": 0.4681,
      "num_input_tokens_seen": 441228244,
      "step": 7874
    },
    {
      "epoch": 17.536748329621382,
      "loss": 0.4394185543060303,
      "loss_ce": 8.750054985284805e-05,
      "loss_iou": 0.181640625,
      "loss_num": 0.01544189453125,
      "loss_xval": 0.439453125,
      "num_input_tokens_seen": 441228244,
      "step": 7874
    },
    {
      "epoch": 17.538975501113587,
      "grad_norm": 15.606276512145996,
      "learning_rate": 1e-06,
      "loss": 0.6399,
      "num_input_tokens_seen": 441283016,
      "step": 7875
    },
    {
      "epoch": 17.538975501113587,
      "loss": 0.5754181742668152,
      "loss_ce": 0.0001007895843940787,
      "loss_iou": 0.255859375,
      "loss_num": 0.01251220703125,
      "loss_xval": 0.57421875,
      "num_input_tokens_seen": 441283016,
      "step": 7875
    },
    {
      "epoch": 17.54120267260579,
      "grad_norm": 12.660006523132324,
      "learning_rate": 1e-06,
      "loss": 0.4726,
      "num_input_tokens_seen": 441340588,
      "step": 7876
    },
    {
      "epoch": 17.54120267260579,
      "loss": 0.37886592745780945,
      "loss_ce": 8.173068636097014e-05,
      "loss_iou": 0.15625,
      "loss_num": 0.0133056640625,
      "loss_xval": 0.37890625,
      "num_input_tokens_seen": 441340588,
      "step": 7876
    },
    {
      "epoch": 17.543429844097997,
      "grad_norm": 28.50706672668457,
      "learning_rate": 1e-06,
      "loss": 0.5147,
      "num_input_tokens_seen": 441396552,
      "step": 7877
    },
    {
      "epoch": 17.543429844097997,
      "loss": 0.40114733576774597,
      "loss_ce": 8.532906213076785e-05,
      "loss_iou": 0.1796875,
      "loss_num": 0.00823974609375,
      "loss_xval": 0.400390625,
      "num_input_tokens_seen": 441396552,
      "step": 7877
    },
    {
      "epoch": 17.5456570155902,
      "grad_norm": 18.953811645507812,
      "learning_rate": 1e-06,
      "loss": 0.3756,
      "num_input_tokens_seen": 441454428,
      "step": 7878
    },
    {
      "epoch": 17.5456570155902,
      "loss": 0.40304529666900635,
      "loss_ce": 9.119759488385171e-05,
      "loss_iou": 0.1708984375,
      "loss_num": 0.01226806640625,
      "loss_xval": 0.40234375,
      "num_input_tokens_seen": 441454428,
      "step": 7878
    },
    {
      "epoch": 17.547884187082406,
      "grad_norm": 40.04744338989258,
      "learning_rate": 1e-06,
      "loss": 0.3407,
      "num_input_tokens_seen": 441509888,
      "step": 7879
    },
    {
      "epoch": 17.547884187082406,
      "loss": 0.3022136688232422,
      "loss_ce": 8.962298306869343e-05,
      "loss_iou": 0.1318359375,
      "loss_num": 0.00775146484375,
      "loss_xval": 0.302734375,
      "num_input_tokens_seen": 441509888,
      "step": 7879
    },
    {
      "epoch": 17.55011135857461,
      "grad_norm": 23.62949562072754,
      "learning_rate": 1e-06,
      "loss": 0.3179,
      "num_input_tokens_seen": 441567796,
      "step": 7880
    },
    {
      "epoch": 17.55011135857461,
      "loss": 0.26015806198120117,
      "loss_ce": 8.726240776013583e-05,
      "loss_iou": 0.11328125,
      "loss_num": 0.006805419921875,
      "loss_xval": 0.259765625,
      "num_input_tokens_seen": 441567796,
      "step": 7880
    },
    {
      "epoch": 17.552338530066816,
      "grad_norm": 11.53111743927002,
      "learning_rate": 1e-06,
      "loss": 0.5086,
      "num_input_tokens_seen": 441624796,
      "step": 7881
    },
    {
      "epoch": 17.552338530066816,
      "loss": 0.4687040448188782,
      "loss_ce": 7.612221816089004e-05,
      "loss_iou": 0.1953125,
      "loss_num": 0.01544189453125,
      "loss_xval": 0.46875,
      "num_input_tokens_seen": 441624796,
      "step": 7881
    },
    {
      "epoch": 17.55456570155902,
      "grad_norm": 18.621740341186523,
      "learning_rate": 1e-06,
      "loss": 0.4156,
      "num_input_tokens_seen": 441681052,
      "step": 7882
    },
    {
      "epoch": 17.55456570155902,
      "loss": 0.4146304726600647,
      "loss_ce": 7.969325815793127e-05,
      "loss_iou": 0.1904296875,
      "loss_num": 0.006622314453125,
      "loss_xval": 0.4140625,
      "num_input_tokens_seen": 441681052,
      "step": 7882
    },
    {
      "epoch": 17.556792873051226,
      "grad_norm": 19.41095542907715,
      "learning_rate": 1e-06,
      "loss": 0.426,
      "num_input_tokens_seen": 441736492,
      "step": 7883
    },
    {
      "epoch": 17.556792873051226,
      "loss": 0.48913177847862244,
      "loss_ce": 0.00011811777949333191,
      "loss_iou": 0.1943359375,
      "loss_num": 0.0198974609375,
      "loss_xval": 0.48828125,
      "num_input_tokens_seen": 441736492,
      "step": 7883
    },
    {
      "epoch": 17.55902004454343,
      "grad_norm": 225.51150512695312,
      "learning_rate": 1e-06,
      "loss": 0.4115,
      "num_input_tokens_seen": 441791384,
      "step": 7884
    },
    {
      "epoch": 17.55902004454343,
      "loss": 0.42270350456237793,
      "loss_ce": 9.607183164916933e-05,
      "loss_iou": 0.1943359375,
      "loss_num": 0.006683349609375,
      "loss_xval": 0.421875,
      "num_input_tokens_seen": 441791384,
      "step": 7884
    },
    {
      "epoch": 17.561247216035635,
      "grad_norm": 22.389850616455078,
      "learning_rate": 1e-06,
      "loss": 0.4788,
      "num_input_tokens_seen": 441846172,
      "step": 7885
    },
    {
      "epoch": 17.561247216035635,
      "loss": 0.44719523191452026,
      "loss_ce": 0.00011274641292402521,
      "loss_iou": 0.19140625,
      "loss_num": 0.01275634765625,
      "loss_xval": 0.447265625,
      "num_input_tokens_seen": 441846172,
      "step": 7885
    },
    {
      "epoch": 17.56347438752784,
      "grad_norm": 25.177894592285156,
      "learning_rate": 1e-06,
      "loss": 0.4156,
      "num_input_tokens_seen": 441901144,
      "step": 7886
    },
    {
      "epoch": 17.56347438752784,
      "loss": 0.5812680721282959,
      "loss_ce": 9.13147086976096e-05,
      "loss_iou": 0.255859375,
      "loss_num": 0.014404296875,
      "loss_xval": 0.58203125,
      "num_input_tokens_seen": 441901144,
      "step": 7886
    },
    {
      "epoch": 17.565701559020045,
      "grad_norm": 20.27315902709961,
      "learning_rate": 1e-06,
      "loss": 0.3407,
      "num_input_tokens_seen": 441956844,
      "step": 7887
    },
    {
      "epoch": 17.565701559020045,
      "loss": 0.32326167821884155,
      "loss_ce": 8.051642362261191e-05,
      "loss_iou": 0.142578125,
      "loss_num": 0.007537841796875,
      "loss_xval": 0.322265625,
      "num_input_tokens_seen": 441956844,
      "step": 7887
    },
    {
      "epoch": 17.56792873051225,
      "grad_norm": 15.661298751831055,
      "learning_rate": 1e-06,
      "loss": 0.359,
      "num_input_tokens_seen": 442013600,
      "step": 7888
    },
    {
      "epoch": 17.56792873051225,
      "loss": 0.3030818700790405,
      "loss_ce": 0.00010335086699342355,
      "loss_iou": 0.1337890625,
      "loss_num": 0.0069580078125,
      "loss_xval": 0.302734375,
      "num_input_tokens_seen": 442013600,
      "step": 7888
    },
    {
      "epoch": 17.570155902004455,
      "grad_norm": 16.606544494628906,
      "learning_rate": 1e-06,
      "loss": 0.3319,
      "num_input_tokens_seen": 442070068,
      "step": 7889
    },
    {
      "epoch": 17.570155902004455,
      "loss": 0.34052973985671997,
      "loss_ce": 7.563138206023723e-05,
      "loss_iou": 0.1396484375,
      "loss_num": 0.01226806640625,
      "loss_xval": 0.33984375,
      "num_input_tokens_seen": 442070068,
      "step": 7889
    },
    {
      "epoch": 17.57238307349666,
      "grad_norm": 17.5957088470459,
      "learning_rate": 1e-06,
      "loss": 0.307,
      "num_input_tokens_seen": 442125372,
      "step": 7890
    },
    {
      "epoch": 17.57238307349666,
      "loss": 0.29927778244018555,
      "loss_ce": 8.342567889485508e-05,
      "loss_iou": 0.125,
      "loss_num": 0.00970458984375,
      "loss_xval": 0.298828125,
      "num_input_tokens_seen": 442125372,
      "step": 7890
    },
    {
      "epoch": 17.574610244988865,
      "grad_norm": 22.874229431152344,
      "learning_rate": 1e-06,
      "loss": 0.3021,
      "num_input_tokens_seen": 442180908,
      "step": 7891
    },
    {
      "epoch": 17.574610244988865,
      "loss": 0.34810495376586914,
      "loss_ce": 8.246798097388819e-05,
      "loss_iou": 0.1484375,
      "loss_num": 0.01007080078125,
      "loss_xval": 0.34765625,
      "num_input_tokens_seen": 442180908,
      "step": 7891
    },
    {
      "epoch": 17.57683741648107,
      "grad_norm": 12.980911254882812,
      "learning_rate": 1e-06,
      "loss": 0.5465,
      "num_input_tokens_seen": 442237344,
      "step": 7892
    },
    {
      "epoch": 17.57683741648107,
      "loss": 0.6626755595207214,
      "loss_ce": 7.78659014031291e-05,
      "loss_iou": 0.267578125,
      "loss_num": 0.025390625,
      "loss_xval": 0.6640625,
      "num_input_tokens_seen": 442237344,
      "step": 7892
    },
    {
      "epoch": 17.579064587973274,
      "grad_norm": 16.26838493347168,
      "learning_rate": 1e-06,
      "loss": 0.3547,
      "num_input_tokens_seen": 442291704,
      "step": 7893
    },
    {
      "epoch": 17.579064587973274,
      "loss": 0.18787457048892975,
      "loss_ce": 6.940308230696246e-05,
      "loss_iou": 0.07763671875,
      "loss_num": 0.006500244140625,
      "loss_xval": 0.1875,
      "num_input_tokens_seen": 442291704,
      "step": 7893
    },
    {
      "epoch": 17.58129175946548,
      "grad_norm": 19.98689079284668,
      "learning_rate": 1e-06,
      "loss": 0.5964,
      "num_input_tokens_seen": 442348288,
      "step": 7894
    },
    {
      "epoch": 17.58129175946548,
      "loss": 0.5989159345626831,
      "loss_ce": 0.00028316857060417533,
      "loss_iou": 0.263671875,
      "loss_num": 0.0142822265625,
      "loss_xval": 0.59765625,
      "num_input_tokens_seen": 442348288,
      "step": 7894
    },
    {
      "epoch": 17.583518930957684,
      "grad_norm": 16.321231842041016,
      "learning_rate": 1e-06,
      "loss": 0.354,
      "num_input_tokens_seen": 442404364,
      "step": 7895
    },
    {
      "epoch": 17.583518930957684,
      "loss": 0.3744066655635834,
      "loss_ce": 7.803218613844365e-05,
      "loss_iou": 0.1689453125,
      "loss_num": 0.00750732421875,
      "loss_xval": 0.375,
      "num_input_tokens_seen": 442404364,
      "step": 7895
    },
    {
      "epoch": 17.58574610244989,
      "grad_norm": 18.746238708496094,
      "learning_rate": 1e-06,
      "loss": 0.3205,
      "num_input_tokens_seen": 442461344,
      "step": 7896
    },
    {
      "epoch": 17.58574610244989,
      "loss": 0.2867514491081238,
      "loss_ce": 6.934157863724977e-05,
      "loss_iou": 0.1259765625,
      "loss_num": 0.007110595703125,
      "loss_xval": 0.287109375,
      "num_input_tokens_seen": 442461344,
      "step": 7896
    },
    {
      "epoch": 17.587973273942094,
      "grad_norm": 25.406570434570312,
      "learning_rate": 1e-06,
      "loss": 0.4222,
      "num_input_tokens_seen": 442516584,
      "step": 7897
    },
    {
      "epoch": 17.587973273942094,
      "loss": 0.33216458559036255,
      "loss_ce": 7.232959615066648e-05,
      "loss_iou": 0.150390625,
      "loss_num": 0.006195068359375,
      "loss_xval": 0.33203125,
      "num_input_tokens_seen": 442516584,
      "step": 7897
    },
    {
      "epoch": 17.5902004454343,
      "grad_norm": 18.10183334350586,
      "learning_rate": 1e-06,
      "loss": 0.3157,
      "num_input_tokens_seen": 442573824,
      "step": 7898
    },
    {
      "epoch": 17.5902004454343,
      "loss": 0.22795072197914124,
      "loss_ce": 7.596083742100745e-05,
      "loss_iou": 0.0966796875,
      "loss_num": 0.0068359375,
      "loss_xval": 0.2275390625,
      "num_input_tokens_seen": 442573824,
      "step": 7898
    },
    {
      "epoch": 17.592427616926503,
      "grad_norm": 38.653526306152344,
      "learning_rate": 1e-06,
      "loss": 0.5042,
      "num_input_tokens_seen": 442629168,
      "step": 7899
    },
    {
      "epoch": 17.592427616926503,
      "loss": 0.3851657509803772,
      "loss_ce": 9.493608376942575e-05,
      "loss_iou": 0.17578125,
      "loss_num": 0.00677490234375,
      "loss_xval": 0.384765625,
      "num_input_tokens_seen": 442629168,
      "step": 7899
    },
    {
      "epoch": 17.59465478841871,
      "grad_norm": 32.544410705566406,
      "learning_rate": 1e-06,
      "loss": 0.5607,
      "num_input_tokens_seen": 442683144,
      "step": 7900
    },
    {
      "epoch": 17.59465478841871,
      "loss": 0.556228756904602,
      "loss_ce": 7.639994146302342e-05,
      "loss_iou": 0.25390625,
      "loss_num": 0.010009765625,
      "loss_xval": 0.5546875,
      "num_input_tokens_seen": 442683144,
      "step": 7900
    },
    {
      "epoch": 17.596881959910913,
      "grad_norm": 19.76438331604004,
      "learning_rate": 1e-06,
      "loss": 0.5191,
      "num_input_tokens_seen": 442737092,
      "step": 7901
    },
    {
      "epoch": 17.596881959910913,
      "loss": 0.436367928981781,
      "loss_ce": 7.337699935305864e-05,
      "loss_iou": 0.1982421875,
      "loss_num": 0.008056640625,
      "loss_xval": 0.435546875,
      "num_input_tokens_seen": 442737092,
      "step": 7901
    },
    {
      "epoch": 17.599109131403118,
      "grad_norm": 25.040283203125,
      "learning_rate": 1e-06,
      "loss": 0.5854,
      "num_input_tokens_seen": 442790552,
      "step": 7902
    },
    {
      "epoch": 17.599109131403118,
      "loss": 0.5836904048919678,
      "loss_ce": 7.226417073979974e-05,
      "loss_iou": 0.26953125,
      "loss_num": 0.00897216796875,
      "loss_xval": 0.58203125,
      "num_input_tokens_seen": 442790552,
      "step": 7902
    },
    {
      "epoch": 17.601336302895323,
      "grad_norm": 25.69877052307129,
      "learning_rate": 1e-06,
      "loss": 0.4327,
      "num_input_tokens_seen": 442845024,
      "step": 7903
    },
    {
      "epoch": 17.601336302895323,
      "loss": 0.28219154477119446,
      "loss_ce": 8.705261279828846e-05,
      "loss_iou": 0.12158203125,
      "loss_num": 0.00787353515625,
      "loss_xval": 0.28125,
      "num_input_tokens_seen": 442845024,
      "step": 7903
    },
    {
      "epoch": 17.603563474387528,
      "grad_norm": 22.128719329833984,
      "learning_rate": 1e-06,
      "loss": 0.3273,
      "num_input_tokens_seen": 442901976,
      "step": 7904
    },
    {
      "epoch": 17.603563474387528,
      "loss": 0.382587730884552,
      "loss_ce": 8.040809188969433e-05,
      "loss_iou": 0.169921875,
      "loss_num": 0.00848388671875,
      "loss_xval": 0.3828125,
      "num_input_tokens_seen": 442901976,
      "step": 7904
    },
    {
      "epoch": 17.605790645879733,
      "grad_norm": 21.455659866333008,
      "learning_rate": 1e-06,
      "loss": 0.5433,
      "num_input_tokens_seen": 442959192,
      "step": 7905
    },
    {
      "epoch": 17.605790645879733,
      "loss": 0.4209858179092407,
      "loss_ce": 8.739200711715966e-05,
      "loss_iou": 0.1904296875,
      "loss_num": 0.00799560546875,
      "loss_xval": 0.421875,
      "num_input_tokens_seen": 442959192,
      "step": 7905
    },
    {
      "epoch": 17.608017817371937,
      "grad_norm": 19.420513153076172,
      "learning_rate": 1e-06,
      "loss": 0.3687,
      "num_input_tokens_seen": 443016136,
      "step": 7906
    },
    {
      "epoch": 17.608017817371937,
      "loss": 0.3827753961086273,
      "loss_ce": 8.494692883687094e-05,
      "loss_iou": 0.1572265625,
      "loss_num": 0.0135498046875,
      "loss_xval": 0.3828125,
      "num_input_tokens_seen": 443016136,
      "step": 7906
    },
    {
      "epoch": 17.610244988864142,
      "grad_norm": 15.976387023925781,
      "learning_rate": 1e-06,
      "loss": 0.6103,
      "num_input_tokens_seen": 443072296,
      "step": 7907
    },
    {
      "epoch": 17.610244988864142,
      "loss": 0.8074872493743896,
      "loss_ce": 0.0001141867833212018,
      "loss_iou": 0.310546875,
      "loss_num": 0.037353515625,
      "loss_xval": 0.80859375,
      "num_input_tokens_seen": 443072296,
      "step": 7907
    },
    {
      "epoch": 17.612472160356347,
      "grad_norm": 17.13764190673828,
      "learning_rate": 1e-06,
      "loss": 0.3673,
      "num_input_tokens_seen": 443128500,
      "step": 7908
    },
    {
      "epoch": 17.612472160356347,
      "loss": 0.4594786763191223,
      "loss_ce": 0.00012808736937586218,
      "loss_iou": 0.181640625,
      "loss_num": 0.019287109375,
      "loss_xval": 0.458984375,
      "num_input_tokens_seen": 443128500,
      "step": 7908
    },
    {
      "epoch": 17.614699331848552,
      "grad_norm": 17.32769203186035,
      "learning_rate": 1e-06,
      "loss": 0.3735,
      "num_input_tokens_seen": 443183760,
      "step": 7909
    },
    {
      "epoch": 17.614699331848552,
      "loss": 0.44984328746795654,
      "loss_ce": 7.522152736783028e-05,
      "loss_iou": 0.1982421875,
      "loss_num": 0.01068115234375,
      "loss_xval": 0.44921875,
      "num_input_tokens_seen": 443183760,
      "step": 7909
    },
    {
      "epoch": 17.616926503340757,
      "grad_norm": 22.6795711517334,
      "learning_rate": 1e-06,
      "loss": 0.3399,
      "num_input_tokens_seen": 443237420,
      "step": 7910
    },
    {
      "epoch": 17.616926503340757,
      "loss": 0.4001952111721039,
      "loss_ce": 0.0001708113995846361,
      "loss_iou": 0.1787109375,
      "loss_num": 0.00836181640625,
      "loss_xval": 0.400390625,
      "num_input_tokens_seen": 443237420,
      "step": 7910
    },
    {
      "epoch": 17.619153674832962,
      "grad_norm": 74.46725463867188,
      "learning_rate": 1e-06,
      "loss": 0.504,
      "num_input_tokens_seen": 443295776,
      "step": 7911
    },
    {
      "epoch": 17.619153674832962,
      "loss": 0.5105748772621155,
      "loss_ce": 7.682420255150646e-05,
      "loss_iou": 0.2216796875,
      "loss_num": 0.013427734375,
      "loss_xval": 0.51171875,
      "num_input_tokens_seen": 443295776,
      "step": 7911
    },
    {
      "epoch": 17.621380846325167,
      "grad_norm": 14.628600120544434,
      "learning_rate": 1e-06,
      "loss": 0.3959,
      "num_input_tokens_seen": 443352484,
      "step": 7912
    },
    {
      "epoch": 17.621380846325167,
      "loss": 0.5070550441741943,
      "loss_ce": 9.699978545540944e-05,
      "loss_iou": 0.2275390625,
      "loss_num": 0.01043701171875,
      "loss_xval": 0.5078125,
      "num_input_tokens_seen": 443352484,
      "step": 7912
    },
    {
      "epoch": 17.62360801781737,
      "grad_norm": 19.935930252075195,
      "learning_rate": 1e-06,
      "loss": 0.4519,
      "num_input_tokens_seen": 443406844,
      "step": 7913
    },
    {
      "epoch": 17.62360801781737,
      "loss": 0.47246021032333374,
      "loss_ce": 7.862341590225697e-05,
      "loss_iou": 0.2109375,
      "loss_num": 0.01007080078125,
      "loss_xval": 0.47265625,
      "num_input_tokens_seen": 443406844,
      "step": 7913
    },
    {
      "epoch": 17.625835189309576,
      "grad_norm": 18.648386001586914,
      "learning_rate": 1e-06,
      "loss": 0.37,
      "num_input_tokens_seen": 443463412,
      "step": 7914
    },
    {
      "epoch": 17.625835189309576,
      "loss": 0.4554082155227661,
      "loss_ce": 8.59397478052415e-05,
      "loss_iou": 0.2021484375,
      "loss_num": 0.01043701171875,
      "loss_xval": 0.455078125,
      "num_input_tokens_seen": 443463412,
      "step": 7914
    },
    {
      "epoch": 17.62806236080178,
      "grad_norm": 11.401290893554688,
      "learning_rate": 1e-06,
      "loss": 0.4383,
      "num_input_tokens_seen": 443517876,
      "step": 7915
    },
    {
      "epoch": 17.62806236080178,
      "loss": 0.4097170829772949,
      "loss_ce": 7.962982635945082e-05,
      "loss_iou": 0.1767578125,
      "loss_num": 0.0113525390625,
      "loss_xval": 0.41015625,
      "num_input_tokens_seen": 443517876,
      "step": 7915
    },
    {
      "epoch": 17.630289532293986,
      "grad_norm": 15.960351943969727,
      "learning_rate": 1e-06,
      "loss": 0.4956,
      "num_input_tokens_seen": 443575536,
      "step": 7916
    },
    {
      "epoch": 17.630289532293986,
      "loss": 0.5167785882949829,
      "loss_ce": 0.00011598135461099446,
      "loss_iou": 0.2158203125,
      "loss_num": 0.01708984375,
      "loss_xval": 0.515625,
      "num_input_tokens_seen": 443575536,
      "step": 7916
    },
    {
      "epoch": 17.63251670378619,
      "grad_norm": 21.701749801635742,
      "learning_rate": 1e-06,
      "loss": 0.4341,
      "num_input_tokens_seen": 443631704,
      "step": 7917
    },
    {
      "epoch": 17.63251670378619,
      "loss": 0.501663863658905,
      "loss_ce": 7.692172948736697e-05,
      "loss_iou": 0.2265625,
      "loss_num": 0.009765625,
      "loss_xval": 0.5,
      "num_input_tokens_seen": 443631704,
      "step": 7917
    },
    {
      "epoch": 17.634743875278396,
      "grad_norm": 14.116578102111816,
      "learning_rate": 1e-06,
      "loss": 0.4556,
      "num_input_tokens_seen": 443688408,
      "step": 7918
    },
    {
      "epoch": 17.634743875278396,
      "loss": 0.4800468385219574,
      "loss_ce": 6.637441401835531e-05,
      "loss_iou": 0.197265625,
      "loss_num": 0.016845703125,
      "loss_xval": 0.48046875,
      "num_input_tokens_seen": 443688408,
      "step": 7918
    },
    {
      "epoch": 17.6369710467706,
      "grad_norm": 83.376708984375,
      "learning_rate": 1e-06,
      "loss": 0.4785,
      "num_input_tokens_seen": 443745244,
      "step": 7919
    },
    {
      "epoch": 17.6369710467706,
      "loss": 0.38171082735061646,
      "loss_ce": 0.00011904077837243676,
      "loss_iou": 0.1650390625,
      "loss_num": 0.01031494140625,
      "loss_xval": 0.380859375,
      "num_input_tokens_seen": 443745244,
      "step": 7919
    },
    {
      "epoch": 17.639198218262806,
      "grad_norm": 18.27570152282715,
      "learning_rate": 1e-06,
      "loss": 0.4711,
      "num_input_tokens_seen": 443804444,
      "step": 7920
    },
    {
      "epoch": 17.639198218262806,
      "loss": 0.5395940542221069,
      "loss_ce": 0.00010429859685245901,
      "loss_iou": 0.2265625,
      "loss_num": 0.01708984375,
      "loss_xval": 0.5390625,
      "num_input_tokens_seen": 443804444,
      "step": 7920
    },
    {
      "epoch": 17.64142538975501,
      "grad_norm": 18.891740798950195,
      "learning_rate": 1e-06,
      "loss": 0.4308,
      "num_input_tokens_seen": 443860448,
      "step": 7921
    },
    {
      "epoch": 17.64142538975501,
      "loss": 0.2655089795589447,
      "loss_ce": 6.710184970870614e-05,
      "loss_iou": 0.11962890625,
      "loss_num": 0.005157470703125,
      "loss_xval": 0.265625,
      "num_input_tokens_seen": 443860448,
      "step": 7921
    },
    {
      "epoch": 17.643652561247215,
      "grad_norm": 21.501697540283203,
      "learning_rate": 1e-06,
      "loss": 0.2732,
      "num_input_tokens_seen": 443917392,
      "step": 7922
    },
    {
      "epoch": 17.643652561247215,
      "loss": 0.34542766213417053,
      "loss_ce": 9.074142144527286e-05,
      "loss_iou": 0.1572265625,
      "loss_num": 0.006317138671875,
      "loss_xval": 0.345703125,
      "num_input_tokens_seen": 443917392,
      "step": 7922
    },
    {
      "epoch": 17.64587973273942,
      "grad_norm": 25.702478408813477,
      "learning_rate": 1e-06,
      "loss": 0.5289,
      "num_input_tokens_seen": 443970952,
      "step": 7923
    },
    {
      "epoch": 17.64587973273942,
      "loss": 0.6104388236999512,
      "loss_ce": 8.727777458261698e-05,
      "loss_iou": 0.26953125,
      "loss_num": 0.01458740234375,
      "loss_xval": 0.609375,
      "num_input_tokens_seen": 443970952,
      "step": 7923
    },
    {
      "epoch": 17.648106904231625,
      "grad_norm": 17.689945220947266,
      "learning_rate": 1e-06,
      "loss": 0.4139,
      "num_input_tokens_seen": 444025404,
      "step": 7924
    },
    {
      "epoch": 17.648106904231625,
      "loss": 0.3999115824699402,
      "loss_ce": 8.553590305382386e-05,
      "loss_iou": 0.1796875,
      "loss_num": 0.00799560546875,
      "loss_xval": 0.400390625,
      "num_input_tokens_seen": 444025404,
      "step": 7924
    },
    {
      "epoch": 17.65033407572383,
      "grad_norm": 17.883432388305664,
      "learning_rate": 1e-06,
      "loss": 0.4983,
      "num_input_tokens_seen": 444081664,
      "step": 7925
    },
    {
      "epoch": 17.65033407572383,
      "loss": 0.4602789878845215,
      "loss_ce": 7.389666279777884e-05,
      "loss_iou": 0.1884765625,
      "loss_num": 0.0167236328125,
      "loss_xval": 0.4609375,
      "num_input_tokens_seen": 444081664,
      "step": 7925
    },
    {
      "epoch": 17.652561247216035,
      "grad_norm": 16.05899429321289,
      "learning_rate": 1e-06,
      "loss": 0.472,
      "num_input_tokens_seen": 444141192,
      "step": 7926
    },
    {
      "epoch": 17.652561247216035,
      "loss": 0.3934362232685089,
      "loss_ce": 9.515189594822004e-05,
      "loss_iou": 0.1708984375,
      "loss_num": 0.0103759765625,
      "loss_xval": 0.392578125,
      "num_input_tokens_seen": 444141192,
      "step": 7926
    },
    {
      "epoch": 17.65478841870824,
      "grad_norm": 26.469680786132812,
      "learning_rate": 1e-06,
      "loss": 0.4919,
      "num_input_tokens_seen": 444197160,
      "step": 7927
    },
    {
      "epoch": 17.65478841870824,
      "loss": 0.4767637252807617,
      "loss_ce": 7.915783498901874e-05,
      "loss_iou": 0.1962890625,
      "loss_num": 0.0167236328125,
      "loss_xval": 0.4765625,
      "num_input_tokens_seen": 444197160,
      "step": 7927
    },
    {
      "epoch": 17.657015590200444,
      "grad_norm": 13.821784973144531,
      "learning_rate": 1e-06,
      "loss": 0.3612,
      "num_input_tokens_seen": 444252964,
      "step": 7928
    },
    {
      "epoch": 17.657015590200444,
      "loss": 0.23647625744342804,
      "loss_ce": 8.710073598194867e-05,
      "loss_iou": 0.10302734375,
      "loss_num": 0.006103515625,
      "loss_xval": 0.236328125,
      "num_input_tokens_seen": 444252964,
      "step": 7928
    },
    {
      "epoch": 17.65924276169265,
      "grad_norm": 16.62885284423828,
      "learning_rate": 1e-06,
      "loss": 0.386,
      "num_input_tokens_seen": 444307512,
      "step": 7929
    },
    {
      "epoch": 17.65924276169265,
      "loss": 0.4383907914161682,
      "loss_ce": 6.681391096208245e-05,
      "loss_iou": 0.1689453125,
      "loss_num": 0.020263671875,
      "loss_xval": 0.4375,
      "num_input_tokens_seen": 444307512,
      "step": 7929
    },
    {
      "epoch": 17.661469933184854,
      "grad_norm": 20.293445587158203,
      "learning_rate": 1e-06,
      "loss": 0.4577,
      "num_input_tokens_seen": 444363292,
      "step": 7930
    },
    {
      "epoch": 17.661469933184854,
      "loss": 0.3440844714641571,
      "loss_ce": 9.034699178300798e-05,
      "loss_iou": 0.146484375,
      "loss_num": 0.01025390625,
      "loss_xval": 0.34375,
      "num_input_tokens_seen": 444363292,
      "step": 7930
    },
    {
      "epoch": 17.66369710467706,
      "grad_norm": 19.445138931274414,
      "learning_rate": 1e-06,
      "loss": 0.4071,
      "num_input_tokens_seen": 444417980,
      "step": 7931
    },
    {
      "epoch": 17.66369710467706,
      "loss": 0.35793596506118774,
      "loss_ce": 8.685909415362403e-05,
      "loss_iou": 0.1611328125,
      "loss_num": 0.0069580078125,
      "loss_xval": 0.357421875,
      "num_input_tokens_seen": 444417980,
      "step": 7931
    },
    {
      "epoch": 17.665924276169264,
      "grad_norm": 21.812707901000977,
      "learning_rate": 1e-06,
      "loss": 0.4571,
      "num_input_tokens_seen": 444475180,
      "step": 7932
    },
    {
      "epoch": 17.665924276169264,
      "loss": 0.48831433057785034,
      "loss_ce": 9.407131437910721e-05,
      "loss_iou": 0.2216796875,
      "loss_num": 0.0089111328125,
      "loss_xval": 0.48828125,
      "num_input_tokens_seen": 444475180,
      "step": 7932
    },
    {
      "epoch": 17.66815144766147,
      "grad_norm": 21.94272804260254,
      "learning_rate": 1e-06,
      "loss": 0.4738,
      "num_input_tokens_seen": 444530092,
      "step": 7933
    },
    {
      "epoch": 17.66815144766147,
      "loss": 0.4068112075328827,
      "loss_ce": 7.29096500435844e-05,
      "loss_iou": 0.1708984375,
      "loss_num": 0.0128173828125,
      "loss_xval": 0.40625,
      "num_input_tokens_seen": 444530092,
      "step": 7933
    },
    {
      "epoch": 17.670378619153674,
      "grad_norm": 21.521142959594727,
      "learning_rate": 1e-06,
      "loss": 0.5135,
      "num_input_tokens_seen": 444585348,
      "step": 7934
    },
    {
      "epoch": 17.670378619153674,
      "loss": 0.5115725994110107,
      "loss_ce": 9.7946947789751e-05,
      "loss_iou": 0.2197265625,
      "loss_num": 0.01434326171875,
      "loss_xval": 0.51171875,
      "num_input_tokens_seen": 444585348,
      "step": 7934
    },
    {
      "epoch": 17.67260579064588,
      "grad_norm": 14.11042308807373,
      "learning_rate": 1e-06,
      "loss": 0.2493,
      "num_input_tokens_seen": 444641960,
      "step": 7935
    },
    {
      "epoch": 17.67260579064588,
      "loss": 0.23591431975364685,
      "loss_ce": 7.447184179909527e-05,
      "loss_iou": 0.1064453125,
      "loss_num": 0.004547119140625,
      "loss_xval": 0.236328125,
      "num_input_tokens_seen": 444641960,
      "step": 7935
    },
    {
      "epoch": 17.674832962138083,
      "grad_norm": 23.0136775970459,
      "learning_rate": 1e-06,
      "loss": 0.4793,
      "num_input_tokens_seen": 444695372,
      "step": 7936
    },
    {
      "epoch": 17.674832962138083,
      "loss": 0.3692181408405304,
      "loss_ce": 7.750838994979858e-05,
      "loss_iou": 0.1630859375,
      "loss_num": 0.008544921875,
      "loss_xval": 0.369140625,
      "num_input_tokens_seen": 444695372,
      "step": 7936
    },
    {
      "epoch": 17.677060133630288,
      "grad_norm": 20.372121810913086,
      "learning_rate": 1e-06,
      "loss": 0.414,
      "num_input_tokens_seen": 444750148,
      "step": 7937
    },
    {
      "epoch": 17.677060133630288,
      "loss": 0.34128010272979736,
      "loss_ce": 9.356189548270777e-05,
      "loss_iou": 0.1591796875,
      "loss_num": 0.004547119140625,
      "loss_xval": 0.341796875,
      "num_input_tokens_seen": 444750148,
      "step": 7937
    },
    {
      "epoch": 17.679287305122493,
      "grad_norm": 23.466459274291992,
      "learning_rate": 1e-06,
      "loss": 0.3234,
      "num_input_tokens_seen": 444809496,
      "step": 7938
    },
    {
      "epoch": 17.679287305122493,
      "loss": 0.36702263355255127,
      "loss_ce": 7.92567734606564e-05,
      "loss_iou": 0.1630859375,
      "loss_num": 0.008056640625,
      "loss_xval": 0.3671875,
      "num_input_tokens_seen": 444809496,
      "step": 7938
    },
    {
      "epoch": 17.681514476614698,
      "grad_norm": 18.88533592224121,
      "learning_rate": 1e-06,
      "loss": 0.5686,
      "num_input_tokens_seen": 444865560,
      "step": 7939
    },
    {
      "epoch": 17.681514476614698,
      "loss": 0.4967142343521118,
      "loss_ce": 0.00013218897220212966,
      "loss_iou": 0.2060546875,
      "loss_num": 0.0169677734375,
      "loss_xval": 0.49609375,
      "num_input_tokens_seen": 444865560,
      "step": 7939
    },
    {
      "epoch": 17.683741648106903,
      "grad_norm": 31.180578231811523,
      "learning_rate": 1e-06,
      "loss": 0.5194,
      "num_input_tokens_seen": 444921728,
      "step": 7940
    },
    {
      "epoch": 17.683741648106903,
      "loss": 0.3380870223045349,
      "loss_ce": 0.0001658698165556416,
      "loss_iou": 0.1455078125,
      "loss_num": 0.00921630859375,
      "loss_xval": 0.337890625,
      "num_input_tokens_seen": 444921728,
      "step": 7940
    },
    {
      "epoch": 17.685968819599108,
      "grad_norm": 28.334543228149414,
      "learning_rate": 1e-06,
      "loss": 0.4694,
      "num_input_tokens_seen": 444976224,
      "step": 7941
    },
    {
      "epoch": 17.685968819599108,
      "loss": 0.44913750886917114,
      "loss_ce": 0.00010187811858486384,
      "loss_iou": 0.1953125,
      "loss_num": 0.01177978515625,
      "loss_xval": 0.44921875,
      "num_input_tokens_seen": 444976224,
      "step": 7941
    },
    {
      "epoch": 17.688195991091312,
      "grad_norm": 19.259824752807617,
      "learning_rate": 1e-06,
      "loss": 0.368,
      "num_input_tokens_seen": 445031004,
      "step": 7942
    },
    {
      "epoch": 17.688195991091312,
      "loss": 0.35163217782974243,
      "loss_ce": 6.968076922930777e-05,
      "loss_iou": 0.1572265625,
      "loss_num": 0.00750732421875,
      "loss_xval": 0.3515625,
      "num_input_tokens_seen": 445031004,
      "step": 7942
    },
    {
      "epoch": 17.690423162583517,
      "grad_norm": 15.660816192626953,
      "learning_rate": 1e-06,
      "loss": 0.4662,
      "num_input_tokens_seen": 445088476,
      "step": 7943
    },
    {
      "epoch": 17.690423162583517,
      "loss": 0.40098705887794495,
      "loss_ce": 0.00010816691064974293,
      "loss_iou": 0.185546875,
      "loss_num": 0.005859375,
      "loss_xval": 0.400390625,
      "num_input_tokens_seen": 445088476,
      "step": 7943
    },
    {
      "epoch": 17.692650334075722,
      "grad_norm": 63.67619705200195,
      "learning_rate": 1e-06,
      "loss": 0.5689,
      "num_input_tokens_seen": 445146920,
      "step": 7944
    },
    {
      "epoch": 17.692650334075722,
      "loss": 0.620599091053009,
      "loss_ce": 0.00011569763591978699,
      "loss_iou": 0.275390625,
      "loss_num": 0.01373291015625,
      "loss_xval": 0.62109375,
      "num_input_tokens_seen": 445146920,
      "step": 7944
    },
    {
      "epoch": 17.694877505567927,
      "grad_norm": 29.446853637695312,
      "learning_rate": 1e-06,
      "loss": 0.4434,
      "num_input_tokens_seen": 445199824,
      "step": 7945
    },
    {
      "epoch": 17.694877505567927,
      "loss": 0.44784268736839294,
      "loss_ce": 8.878282096702605e-05,
      "loss_iou": 0.1962890625,
      "loss_num": 0.0108642578125,
      "loss_xval": 0.447265625,
      "num_input_tokens_seen": 445199824,
      "step": 7945
    },
    {
      "epoch": 17.697104677060132,
      "grad_norm": 25.64476203918457,
      "learning_rate": 1e-06,
      "loss": 0.4774,
      "num_input_tokens_seen": 445257296,
      "step": 7946
    },
    {
      "epoch": 17.697104677060132,
      "loss": 0.3815593123435974,
      "loss_ce": 8.9571054559201e-05,
      "loss_iou": 0.1474609375,
      "loss_num": 0.0172119140625,
      "loss_xval": 0.380859375,
      "num_input_tokens_seen": 445257296,
      "step": 7946
    },
    {
      "epoch": 17.69933184855234,
      "grad_norm": 16.938840866088867,
      "learning_rate": 1e-06,
      "loss": 0.3748,
      "num_input_tokens_seen": 445314848,
      "step": 7947
    },
    {
      "epoch": 17.69933184855234,
      "loss": 0.32621848583221436,
      "loss_ce": 7.714293315075338e-05,
      "loss_iou": 0.1376953125,
      "loss_num": 0.0098876953125,
      "loss_xval": 0.326171875,
      "num_input_tokens_seen": 445314848,
      "step": 7947
    },
    {
      "epoch": 17.70155902004454,
      "grad_norm": 15.301907539367676,
      "learning_rate": 1e-06,
      "loss": 0.2937,
      "num_input_tokens_seen": 445372204,
      "step": 7948
    },
    {
      "epoch": 17.70155902004454,
      "loss": 0.30735015869140625,
      "loss_ce": 9.919532749336213e-05,
      "loss_iou": 0.138671875,
      "loss_num": 0.006103515625,
      "loss_xval": 0.306640625,
      "num_input_tokens_seen": 445372204,
      "step": 7948
    },
    {
      "epoch": 17.70378619153675,
      "grad_norm": 16.031461715698242,
      "learning_rate": 1e-06,
      "loss": 0.4855,
      "num_input_tokens_seen": 445430708,
      "step": 7949
    },
    {
      "epoch": 17.70378619153675,
      "loss": 0.49899452924728394,
      "loss_ce": 9.31676768232137e-05,
      "loss_iou": 0.18359375,
      "loss_num": 0.0262451171875,
      "loss_xval": 0.498046875,
      "num_input_tokens_seen": 445430708,
      "step": 7949
    },
    {
      "epoch": 17.706013363028955,
      "grad_norm": 13.996493339538574,
      "learning_rate": 1e-06,
      "loss": 0.4567,
      "num_input_tokens_seen": 445485268,
      "step": 7950
    },
    {
      "epoch": 17.706013363028955,
      "loss": 0.4317079484462738,
      "loss_ce": 6.729987217113376e-05,
      "loss_iou": 0.1796875,
      "loss_num": 0.01434326171875,
      "loss_xval": 0.431640625,
      "num_input_tokens_seen": 445485268,
      "step": 7950
    },
    {
      "epoch": 17.70824053452116,
      "grad_norm": 19.51962661743164,
      "learning_rate": 1e-06,
      "loss": 0.5369,
      "num_input_tokens_seen": 445538736,
      "step": 7951
    },
    {
      "epoch": 17.70824053452116,
      "loss": 0.7260493636131287,
      "loss_ce": 9.723611583467573e-05,
      "loss_iou": 0.310546875,
      "loss_num": 0.020751953125,
      "loss_xval": 0.7265625,
      "num_input_tokens_seen": 445538736,
      "step": 7951
    },
    {
      "epoch": 17.710467706013365,
      "grad_norm": 34.36812210083008,
      "learning_rate": 1e-06,
      "loss": 0.6776,
      "num_input_tokens_seen": 445593288,
      "step": 7952
    },
    {
      "epoch": 17.710467706013365,
      "loss": 0.8874142169952393,
      "loss_ce": 8.507441089022905e-05,
      "loss_iou": 0.3984375,
      "loss_num": 0.0185546875,
      "loss_xval": 0.88671875,
      "num_input_tokens_seen": 445593288,
      "step": 7952
    },
    {
      "epoch": 17.71269487750557,
      "grad_norm": 19.97995948791504,
      "learning_rate": 1e-06,
      "loss": 0.3888,
      "num_input_tokens_seen": 445651940,
      "step": 7953
    },
    {
      "epoch": 17.71269487750557,
      "loss": 0.35836881399154663,
      "loss_ce": 9.244780812878162e-05,
      "loss_iou": 0.1533203125,
      "loss_num": 0.01019287109375,
      "loss_xval": 0.357421875,
      "num_input_tokens_seen": 445651940,
      "step": 7953
    },
    {
      "epoch": 17.714922048997774,
      "grad_norm": 18.99198341369629,
      "learning_rate": 1e-06,
      "loss": 0.3816,
      "num_input_tokens_seen": 445707972,
      "step": 7954
    },
    {
      "epoch": 17.714922048997774,
      "loss": 0.43836966156959534,
      "loss_ce": 7.619359530508518e-05,
      "loss_iou": 0.193359375,
      "loss_num": 0.01043701171875,
      "loss_xval": 0.4375,
      "num_input_tokens_seen": 445707972,
      "step": 7954
    },
    {
      "epoch": 17.71714922048998,
      "grad_norm": 24.05302619934082,
      "learning_rate": 1e-06,
      "loss": 0.4304,
      "num_input_tokens_seen": 445765012,
      "step": 7955
    },
    {
      "epoch": 17.71714922048998,
      "loss": 0.5598816871643066,
      "loss_ce": 0.00012829234765376896,
      "loss_iou": 0.24609375,
      "loss_num": 0.013427734375,
      "loss_xval": 0.55859375,
      "num_input_tokens_seen": 445765012,
      "step": 7955
    },
    {
      "epoch": 17.719376391982184,
      "grad_norm": 22.349224090576172,
      "learning_rate": 1e-06,
      "loss": 0.4478,
      "num_input_tokens_seen": 445820252,
      "step": 7956
    },
    {
      "epoch": 17.719376391982184,
      "loss": 0.5640408992767334,
      "loss_ce": 7.6019496191293e-05,
      "loss_iou": 0.2265625,
      "loss_num": 0.0220947265625,
      "loss_xval": 0.5625,
      "num_input_tokens_seen": 445820252,
      "step": 7956
    },
    {
      "epoch": 17.72160356347439,
      "grad_norm": 33.075538635253906,
      "learning_rate": 1e-06,
      "loss": 0.399,
      "num_input_tokens_seen": 445875052,
      "step": 7957
    },
    {
      "epoch": 17.72160356347439,
      "loss": 0.39026761054992676,
      "loss_ce": 6.987876258790493e-05,
      "loss_iou": 0.173828125,
      "loss_num": 0.00860595703125,
      "loss_xval": 0.390625,
      "num_input_tokens_seen": 445875052,
      "step": 7957
    },
    {
      "epoch": 17.723830734966594,
      "grad_norm": 16.910858154296875,
      "learning_rate": 1e-06,
      "loss": 0.4148,
      "num_input_tokens_seen": 445928940,
      "step": 7958
    },
    {
      "epoch": 17.723830734966594,
      "loss": 0.2792437970638275,
      "loss_ce": 6.90067681716755e-05,
      "loss_iou": 0.11669921875,
      "loss_num": 0.0091552734375,
      "loss_xval": 0.279296875,
      "num_input_tokens_seen": 445928940,
      "step": 7958
    },
    {
      "epoch": 17.7260579064588,
      "grad_norm": 26.71137046813965,
      "learning_rate": 1e-06,
      "loss": 0.3061,
      "num_input_tokens_seen": 445983828,
      "step": 7959
    },
    {
      "epoch": 17.7260579064588,
      "loss": 0.20442020893096924,
      "loss_ce": 7.449374243151397e-05,
      "loss_iou": 0.0908203125,
      "loss_num": 0.004547119140625,
      "loss_xval": 0.2041015625,
      "num_input_tokens_seen": 445983828,
      "step": 7959
    },
    {
      "epoch": 17.728285077951004,
      "grad_norm": 18.750337600708008,
      "learning_rate": 1e-06,
      "loss": 0.4311,
      "num_input_tokens_seen": 446040604,
      "step": 7960
    },
    {
      "epoch": 17.728285077951004,
      "loss": 0.4378310739994049,
      "loss_ce": 8.695643191458657e-05,
      "loss_iou": 0.181640625,
      "loss_num": 0.0147705078125,
      "loss_xval": 0.4375,
      "num_input_tokens_seen": 446040604,
      "step": 7960
    },
    {
      "epoch": 17.73051224944321,
      "grad_norm": 21.329652786254883,
      "learning_rate": 1e-06,
      "loss": 0.6707,
      "num_input_tokens_seen": 446096288,
      "step": 7961
    },
    {
      "epoch": 17.73051224944321,
      "loss": 0.8002760410308838,
      "loss_ce": 0.00034930004039779305,
      "loss_iou": 0.32421875,
      "loss_num": 0.0299072265625,
      "loss_xval": 0.80078125,
      "num_input_tokens_seen": 446096288,
      "step": 7961
    },
    {
      "epoch": 17.732739420935413,
      "grad_norm": 19.80221176147461,
      "learning_rate": 1e-06,
      "loss": 0.4496,
      "num_input_tokens_seen": 446153240,
      "step": 7962
    },
    {
      "epoch": 17.732739420935413,
      "loss": 0.37191396951675415,
      "loss_ce": 8.780106145422906e-05,
      "loss_iou": 0.1484375,
      "loss_num": 0.0150146484375,
      "loss_xval": 0.37109375,
      "num_input_tokens_seen": 446153240,
      "step": 7962
    },
    {
      "epoch": 17.734966592427618,
      "grad_norm": 23.22903823852539,
      "learning_rate": 1e-06,
      "loss": 0.5215,
      "num_input_tokens_seen": 446209048,
      "step": 7963
    },
    {
      "epoch": 17.734966592427618,
      "loss": 0.5368286967277527,
      "loss_ce": 8.552963117836043e-05,
      "loss_iou": 0.2421875,
      "loss_num": 0.01055908203125,
      "loss_xval": 0.53515625,
      "num_input_tokens_seen": 446209048,
      "step": 7963
    },
    {
      "epoch": 17.737193763919823,
      "grad_norm": 22.78981590270996,
      "learning_rate": 1e-06,
      "loss": 0.3933,
      "num_input_tokens_seen": 446266012,
      "step": 7964
    },
    {
      "epoch": 17.737193763919823,
      "loss": 0.48521602153778076,
      "loss_ce": 0.00010859415488084778,
      "loss_iou": 0.2236328125,
      "loss_num": 0.007659912109375,
      "loss_xval": 0.484375,
      "num_input_tokens_seen": 446266012,
      "step": 7964
    },
    {
      "epoch": 17.739420935412028,
      "grad_norm": 19.937816619873047,
      "learning_rate": 1e-06,
      "loss": 0.2245,
      "num_input_tokens_seen": 446323072,
      "step": 7965
    },
    {
      "epoch": 17.739420935412028,
      "loss": 0.2009565830230713,
      "loss_ce": 8.989279740490019e-05,
      "loss_iou": 0.087890625,
      "loss_num": 0.005126953125,
      "loss_xval": 0.201171875,
      "num_input_tokens_seen": 446323072,
      "step": 7965
    },
    {
      "epoch": 17.741648106904233,
      "grad_norm": 15.368204116821289,
      "learning_rate": 1e-06,
      "loss": 0.4583,
      "num_input_tokens_seen": 446378856,
      "step": 7966
    },
    {
      "epoch": 17.741648106904233,
      "loss": 0.5300416350364685,
      "loss_ce": 7.336361159104854e-05,
      "loss_iou": 0.2275390625,
      "loss_num": 0.01507568359375,
      "loss_xval": 0.53125,
      "num_input_tokens_seen": 446378856,
      "step": 7966
    },
    {
      "epoch": 17.743875278396438,
      "grad_norm": 22.195009231567383,
      "learning_rate": 1e-06,
      "loss": 0.3353,
      "num_input_tokens_seen": 446434664,
      "step": 7967
    },
    {
      "epoch": 17.743875278396438,
      "loss": 0.3601817786693573,
      "loss_ce": 7.436297892127186e-05,
      "loss_iou": 0.1484375,
      "loss_num": 0.01251220703125,
      "loss_xval": 0.359375,
      "num_input_tokens_seen": 446434664,
      "step": 7967
    },
    {
      "epoch": 17.746102449888642,
      "grad_norm": 13.607110977172852,
      "learning_rate": 1e-06,
      "loss": 0.4167,
      "num_input_tokens_seen": 446490256,
      "step": 7968
    },
    {
      "epoch": 17.746102449888642,
      "loss": 0.5662601590156555,
      "loss_ce": 9.807750029722229e-05,
      "loss_iou": 0.2353515625,
      "loss_num": 0.0194091796875,
      "loss_xval": 0.56640625,
      "num_input_tokens_seen": 446490256,
      "step": 7968
    },
    {
      "epoch": 17.748329621380847,
      "grad_norm": 14.89394474029541,
      "learning_rate": 1e-06,
      "loss": 0.3675,
      "num_input_tokens_seen": 446545948,
      "step": 7969
    },
    {
      "epoch": 17.748329621380847,
      "loss": 0.38515520095825195,
      "loss_ce": 8.440592500846833e-05,
      "loss_iou": 0.166015625,
      "loss_num": 0.01055908203125,
      "loss_xval": 0.384765625,
      "num_input_tokens_seen": 446545948,
      "step": 7969
    },
    {
      "epoch": 17.750556792873052,
      "grad_norm": 18.065866470336914,
      "learning_rate": 1e-06,
      "loss": 0.3868,
      "num_input_tokens_seen": 446601844,
      "step": 7970
    },
    {
      "epoch": 17.750556792873052,
      "loss": 0.3480460047721863,
      "loss_ce": 8.455889474134892e-05,
      "loss_iou": 0.1513671875,
      "loss_num": 0.00909423828125,
      "loss_xval": 0.34765625,
      "num_input_tokens_seen": 446601844,
      "step": 7970
    },
    {
      "epoch": 17.752783964365257,
      "grad_norm": 17.822830200195312,
      "learning_rate": 1e-06,
      "loss": 0.4991,
      "num_input_tokens_seen": 446657308,
      "step": 7971
    },
    {
      "epoch": 17.752783964365257,
      "loss": 0.5141241550445557,
      "loss_ce": 8.604546019341797e-05,
      "loss_iou": 0.2314453125,
      "loss_num": 0.01007080078125,
      "loss_xval": 0.515625,
      "num_input_tokens_seen": 446657308,
      "step": 7971
    },
    {
      "epoch": 17.755011135857462,
      "grad_norm": 19.482011795043945,
      "learning_rate": 1e-06,
      "loss": 0.4116,
      "num_input_tokens_seen": 446714312,
      "step": 7972
    },
    {
      "epoch": 17.755011135857462,
      "loss": 0.3692222833633423,
      "loss_ce": 8.163502207025886e-05,
      "loss_iou": 0.173828125,
      "loss_num": 0.004425048828125,
      "loss_xval": 0.369140625,
      "num_input_tokens_seen": 446714312,
      "step": 7972
    },
    {
      "epoch": 17.757238307349667,
      "grad_norm": 17.875993728637695,
      "learning_rate": 1e-06,
      "loss": 0.3371,
      "num_input_tokens_seen": 446768140,
      "step": 7973
    },
    {
      "epoch": 17.757238307349667,
      "loss": 0.3051214814186096,
      "loss_ce": 6.776393274776638e-05,
      "loss_iou": 0.130859375,
      "loss_num": 0.00860595703125,
      "loss_xval": 0.3046875,
      "num_input_tokens_seen": 446768140,
      "step": 7973
    },
    {
      "epoch": 17.75946547884187,
      "grad_norm": 16.569456100463867,
      "learning_rate": 1e-06,
      "loss": 0.617,
      "num_input_tokens_seen": 446822136,
      "step": 7974
    },
    {
      "epoch": 17.75946547884187,
      "loss": 0.8183131814002991,
      "loss_ce": 0.0001368634111713618,
      "loss_iou": 0.275390625,
      "loss_num": 0.052978515625,
      "loss_xval": 0.81640625,
      "num_input_tokens_seen": 446822136,
      "step": 7974
    },
    {
      "epoch": 17.761692650334076,
      "grad_norm": 18.004579544067383,
      "learning_rate": 1e-06,
      "loss": 0.3327,
      "num_input_tokens_seen": 446878228,
      "step": 7975
    },
    {
      "epoch": 17.761692650334076,
      "loss": 0.2270025908946991,
      "loss_ce": 7.388347876258194e-05,
      "loss_iou": 0.09814453125,
      "loss_num": 0.006072998046875,
      "loss_xval": 0.2265625,
      "num_input_tokens_seen": 446878228,
      "step": 7975
    },
    {
      "epoch": 17.76391982182628,
      "grad_norm": 45.05828857421875,
      "learning_rate": 1e-06,
      "loss": 0.4351,
      "num_input_tokens_seen": 446934920,
      "step": 7976
    },
    {
      "epoch": 17.76391982182628,
      "loss": 0.36751073598861694,
      "loss_ce": 0.00010961330553982407,
      "loss_iou": 0.1396484375,
      "loss_num": 0.0174560546875,
      "loss_xval": 0.3671875,
      "num_input_tokens_seen": 446934920,
      "step": 7976
    },
    {
      "epoch": 17.766146993318486,
      "grad_norm": 19.841493606567383,
      "learning_rate": 1e-06,
      "loss": 0.5039,
      "num_input_tokens_seen": 446989424,
      "step": 7977
    },
    {
      "epoch": 17.766146993318486,
      "loss": 0.4566991627216339,
      "loss_ce": 9.516975842416286e-05,
      "loss_iou": 0.19140625,
      "loss_num": 0.0147705078125,
      "loss_xval": 0.45703125,
      "num_input_tokens_seen": 446989424,
      "step": 7977
    },
    {
      "epoch": 17.76837416481069,
      "grad_norm": 18.41941261291504,
      "learning_rate": 1e-06,
      "loss": 0.4241,
      "num_input_tokens_seen": 447046508,
      "step": 7978
    },
    {
      "epoch": 17.76837416481069,
      "loss": 0.3700698912143707,
      "loss_ce": 7.478601764887571e-05,
      "loss_iou": 0.1669921875,
      "loss_num": 0.00732421875,
      "loss_xval": 0.369140625,
      "num_input_tokens_seen": 447046508,
      "step": 7978
    },
    {
      "epoch": 17.770601336302896,
      "grad_norm": 19.740684509277344,
      "learning_rate": 1e-06,
      "loss": 0.3984,
      "num_input_tokens_seen": 447104012,
      "step": 7979
    },
    {
      "epoch": 17.770601336302896,
      "loss": 0.27607256174087524,
      "loss_ce": 7.156394713092595e-05,
      "loss_iou": 0.1279296875,
      "loss_num": 0.003875732421875,
      "loss_xval": 0.275390625,
      "num_input_tokens_seen": 447104012,
      "step": 7979
    },
    {
      "epoch": 17.7728285077951,
      "grad_norm": 15.659636497497559,
      "learning_rate": 1e-06,
      "loss": 0.2204,
      "num_input_tokens_seen": 447159184,
      "step": 7980
    },
    {
      "epoch": 17.7728285077951,
      "loss": 0.18346147239208221,
      "loss_ce": 8.134550444083288e-05,
      "loss_iou": 0.0732421875,
      "loss_num": 0.007476806640625,
      "loss_xval": 0.18359375,
      "num_input_tokens_seen": 447159184,
      "step": 7980
    },
    {
      "epoch": 17.775055679287306,
      "grad_norm": 18.540023803710938,
      "learning_rate": 1e-06,
      "loss": 0.3167,
      "num_input_tokens_seen": 447212524,
      "step": 7981
    },
    {
      "epoch": 17.775055679287306,
      "loss": 0.33039259910583496,
      "loss_ce": 7.033500878605992e-05,
      "loss_iou": 0.1435546875,
      "loss_num": 0.00860595703125,
      "loss_xval": 0.330078125,
      "num_input_tokens_seen": 447212524,
      "step": 7981
    },
    {
      "epoch": 17.77728285077951,
      "grad_norm": 26.151935577392578,
      "learning_rate": 1e-06,
      "loss": 0.4296,
      "num_input_tokens_seen": 447268040,
      "step": 7982
    },
    {
      "epoch": 17.77728285077951,
      "loss": 0.48384368419647217,
      "loss_ce": 7.902232027845457e-05,
      "loss_iou": 0.2197265625,
      "loss_num": 0.00909423828125,
      "loss_xval": 0.484375,
      "num_input_tokens_seen": 447268040,
      "step": 7982
    },
    {
      "epoch": 17.779510022271715,
      "grad_norm": 19.188148498535156,
      "learning_rate": 1e-06,
      "loss": 0.4998,
      "num_input_tokens_seen": 447323576,
      "step": 7983
    },
    {
      "epoch": 17.779510022271715,
      "loss": 0.35879120230674744,
      "loss_ce": 8.758992771618068e-05,
      "loss_iou": 0.16015625,
      "loss_num": 0.00787353515625,
      "loss_xval": 0.359375,
      "num_input_tokens_seen": 447323576,
      "step": 7983
    },
    {
      "epoch": 17.78173719376392,
      "grad_norm": 20.830575942993164,
      "learning_rate": 1e-06,
      "loss": 0.5727,
      "num_input_tokens_seen": 447379108,
      "step": 7984
    },
    {
      "epoch": 17.78173719376392,
      "loss": 0.5355324745178223,
      "loss_ce": 0.00013207507436163723,
      "loss_iou": 0.228515625,
      "loss_num": 0.015869140625,
      "loss_xval": 0.53515625,
      "num_input_tokens_seen": 447379108,
      "step": 7984
    },
    {
      "epoch": 17.783964365256125,
      "grad_norm": 24.485782623291016,
      "learning_rate": 1e-06,
      "loss": 0.3293,
      "num_input_tokens_seen": 447436116,
      "step": 7985
    },
    {
      "epoch": 17.783964365256125,
      "loss": 0.38479992747306824,
      "loss_ce": 9.53392154769972e-05,
      "loss_iou": 0.1689453125,
      "loss_num": 0.00909423828125,
      "loss_xval": 0.384765625,
      "num_input_tokens_seen": 447436116,
      "step": 7985
    },
    {
      "epoch": 17.78619153674833,
      "grad_norm": 27.353883743286133,
      "learning_rate": 1e-06,
      "loss": 0.3948,
      "num_input_tokens_seen": 447492812,
      "step": 7986
    },
    {
      "epoch": 17.78619153674833,
      "loss": 0.2935987710952759,
      "loss_ce": 8.068819442996755e-05,
      "loss_iou": 0.1240234375,
      "loss_num": 0.00909423828125,
      "loss_xval": 0.29296875,
      "num_input_tokens_seen": 447492812,
      "step": 7986
    },
    {
      "epoch": 17.788418708240535,
      "grad_norm": 17.87607192993164,
      "learning_rate": 1e-06,
      "loss": 0.35,
      "num_input_tokens_seen": 447549280,
      "step": 7987
    },
    {
      "epoch": 17.788418708240535,
      "loss": 0.3173452615737915,
      "loss_ce": 8.452979091089219e-05,
      "loss_iou": 0.13671875,
      "loss_num": 0.00897216796875,
      "loss_xval": 0.31640625,
      "num_input_tokens_seen": 447549280,
      "step": 7987
    },
    {
      "epoch": 17.79064587973274,
      "grad_norm": 21.554283142089844,
      "learning_rate": 1e-06,
      "loss": 0.4563,
      "num_input_tokens_seen": 447603876,
      "step": 7988
    },
    {
      "epoch": 17.79064587973274,
      "loss": 0.5782496929168701,
      "loss_ce": 0.00012468949717003852,
      "loss_iou": 0.25,
      "loss_num": 0.01513671875,
      "loss_xval": 0.578125,
      "num_input_tokens_seen": 447603876,
      "step": 7988
    },
    {
      "epoch": 17.792873051224944,
      "grad_norm": 17.834577560424805,
      "learning_rate": 1e-06,
      "loss": 0.3622,
      "num_input_tokens_seen": 447660636,
      "step": 7989
    },
    {
      "epoch": 17.792873051224944,
      "loss": 0.22425898909568787,
      "loss_ce": 7.684796582907438e-05,
      "loss_iou": 0.103515625,
      "loss_num": 0.0034332275390625,
      "loss_xval": 0.224609375,
      "num_input_tokens_seen": 447660636,
      "step": 7989
    },
    {
      "epoch": 17.79510022271715,
      "grad_norm": 16.187517166137695,
      "learning_rate": 1e-06,
      "loss": 0.3169,
      "num_input_tokens_seen": 447715244,
      "step": 7990
    },
    {
      "epoch": 17.79510022271715,
      "loss": 0.3078160881996155,
      "loss_ce": 7.682182331336662e-05,
      "loss_iou": 0.1298828125,
      "loss_num": 0.009521484375,
      "loss_xval": 0.30859375,
      "num_input_tokens_seen": 447715244,
      "step": 7990
    },
    {
      "epoch": 17.797327394209354,
      "grad_norm": 30.0116024017334,
      "learning_rate": 1e-06,
      "loss": 0.3553,
      "num_input_tokens_seen": 447771124,
      "step": 7991
    },
    {
      "epoch": 17.797327394209354,
      "loss": 0.2718275189399719,
      "loss_ce": 9.136189328273758e-05,
      "loss_iou": 0.123046875,
      "loss_num": 0.005126953125,
      "loss_xval": 0.271484375,
      "num_input_tokens_seen": 447771124,
      "step": 7991
    },
    {
      "epoch": 17.79955456570156,
      "grad_norm": 20.42539405822754,
      "learning_rate": 1e-06,
      "loss": 0.4136,
      "num_input_tokens_seen": 447826024,
      "step": 7992
    },
    {
      "epoch": 17.79955456570156,
      "loss": 0.4224761426448822,
      "loss_ce": 0.00011284490756224841,
      "loss_iou": 0.1826171875,
      "loss_num": 0.01141357421875,
      "loss_xval": 0.421875,
      "num_input_tokens_seen": 447826024,
      "step": 7992
    },
    {
      "epoch": 17.801781737193764,
      "grad_norm": 15.1818208694458,
      "learning_rate": 1e-06,
      "loss": 0.494,
      "num_input_tokens_seen": 447881836,
      "step": 7993
    },
    {
      "epoch": 17.801781737193764,
      "loss": 0.37318992614746094,
      "loss_ce": 8.200886804843321e-05,
      "loss_iou": 0.1494140625,
      "loss_num": 0.014892578125,
      "loss_xval": 0.373046875,
      "num_input_tokens_seen": 447881836,
      "step": 7993
    },
    {
      "epoch": 17.80400890868597,
      "grad_norm": 11.075129508972168,
      "learning_rate": 1e-06,
      "loss": 0.5868,
      "num_input_tokens_seen": 447938492,
      "step": 7994
    },
    {
      "epoch": 17.80400890868597,
      "loss": 0.8123493790626526,
      "loss_ce": 0.0001545300183352083,
      "loss_iou": 0.341796875,
      "loss_num": 0.0260009765625,
      "loss_xval": 0.8125,
      "num_input_tokens_seen": 447938492,
      "step": 7994
    },
    {
      "epoch": 17.806236080178174,
      "grad_norm": 26.124393463134766,
      "learning_rate": 1e-06,
      "loss": 0.5909,
      "num_input_tokens_seen": 447992224,
      "step": 7995
    },
    {
      "epoch": 17.806236080178174,
      "loss": 0.5432973504066467,
      "loss_ce": 8.448238804703578e-05,
      "loss_iou": 0.2412109375,
      "loss_num": 0.0123291015625,
      "loss_xval": 0.54296875,
      "num_input_tokens_seen": 447992224,
      "step": 7995
    },
    {
      "epoch": 17.80846325167038,
      "grad_norm": 21.8154296875,
      "learning_rate": 1e-06,
      "loss": 0.3901,
      "num_input_tokens_seen": 448051428,
      "step": 7996
    },
    {
      "epoch": 17.80846325167038,
      "loss": 0.4403107762336731,
      "loss_ce": 0.0001252087822649628,
      "loss_iou": 0.1953125,
      "loss_num": 0.009765625,
      "loss_xval": 0.439453125,
      "num_input_tokens_seen": 448051428,
      "step": 7996
    },
    {
      "epoch": 17.810690423162583,
      "grad_norm": 29.165897369384766,
      "learning_rate": 1e-06,
      "loss": 0.4263,
      "num_input_tokens_seen": 448108228,
      "step": 7997
    },
    {
      "epoch": 17.810690423162583,
      "loss": 0.35651570558547974,
      "loss_ce": 7.038043258944526e-05,
      "loss_iou": 0.1572265625,
      "loss_num": 0.0084228515625,
      "loss_xval": 0.35546875,
      "num_input_tokens_seen": 448108228,
      "step": 7997
    },
    {
      "epoch": 17.812917594654788,
      "grad_norm": 26.357572555541992,
      "learning_rate": 1e-06,
      "loss": 0.3326,
      "num_input_tokens_seen": 448163460,
      "step": 7998
    },
    {
      "epoch": 17.812917594654788,
      "loss": 0.2939828336238861,
      "loss_ce": 9.854609379544854e-05,
      "loss_iou": 0.134765625,
      "loss_num": 0.004852294921875,
      "loss_xval": 0.29296875,
      "num_input_tokens_seen": 448163460,
      "step": 7998
    },
    {
      "epoch": 17.815144766146993,
      "grad_norm": 17.523059844970703,
      "learning_rate": 1e-06,
      "loss": 0.3924,
      "num_input_tokens_seen": 448220148,
      "step": 7999
    },
    {
      "epoch": 17.815144766146993,
      "loss": 0.29866713285446167,
      "loss_ce": 8.312883437611163e-05,
      "loss_iou": 0.138671875,
      "loss_num": 0.00439453125,
      "loss_xval": 0.298828125,
      "num_input_tokens_seen": 448220148,
      "step": 7999
    },
    {
      "epoch": 17.817371937639198,
      "grad_norm": 21.25289535522461,
      "learning_rate": 1e-06,
      "loss": 0.5296,
      "num_input_tokens_seen": 448275732,
      "step": 8000
    },
    {
      "epoch": 17.817371937639198,
      "eval_seeclick_web_CIoU": 0.5856780111789703,
      "eval_seeclick_web_GIoU": 0.5835351943969727,
      "eval_seeclick_web_IoU": 0.6051326394081116,
      "eval_seeclick_web_MAE_all": 0.015136118279770017,
      "eval_seeclick_web_MAE_h": 0.007184438407421112,
      "eval_seeclick_web_MAE_w": 0.015148711390793324,
      "eval_seeclick_web_MAE_x_boxes": 0.008295744191855192,
      "eval_seeclick_web_MAE_y_boxes": 0.021377818658947945,
      "eval_seeclick_web_inside_bbox": 0.9010416567325592,
      "eval_seeclick_web_loss": 0.9110763072967529,
      "eval_seeclick_web_loss_ce": 0.00013513932935893536,
      "eval_seeclick_web_loss_iou": 0.4219970703125,
      "eval_seeclick_web_loss_num": 0.012152671813964844,
      "eval_seeclick_web_loss_xval": 0.905029296875,
      "eval_seeclick_web_runtime": 23.0578,
      "eval_seeclick_web_samples_per_second": 2.168,
      "eval_seeclick_web_steps_per_second": 0.087,
      "num_input_tokens_seen": 448275732,
      "step": 8000
    },
    {
      "epoch": 17.817371937639198,
      "eval_icons_CIoU": 0.2600770592689514,
      "eval_icons_GIoU": 0.28505839407444,
      "eval_icons_IoU": 0.3360164016485214,
      "eval_icons_MAE_all": 0.05904686264693737,
      "eval_icons_MAE_h": 0.03290587291121483,
      "eval_icons_MAE_w": 0.06054178345948458,
      "eval_icons_MAE_x_boxes": 0.05734286643564701,
      "eval_icons_MAE_y_boxes": 0.03782099112868309,
      "eval_icons_inside_bbox": 0.59375,
      "eval_icons_loss": 1.714854121208191,
      "eval_icons_loss_ce": 0.0001619036338524893,
      "eval_icons_loss_iou": 0.6759033203125,
      "eval_icons_loss_num": 0.058238983154296875,
      "eval_icons_loss_xval": 1.643798828125,
      "eval_icons_runtime": 21.0775,
      "eval_icons_samples_per_second": 2.372,
      "eval_icons_steps_per_second": 0.095,
      "num_input_tokens_seen": 448275732,
      "step": 8000
    },
    {
      "epoch": 17.817371937639198,
      "eval_screenspot_CIoU": 0.3892778257528941,
      "eval_screenspot_GIoU": 0.4072861274083455,
      "eval_screenspot_IoU": 0.45707400639851886,
      "eval_screenspot_MAE_all": 0.053338187436262764,
      "eval_screenspot_MAE_h": 0.03920335695147514,
      "eval_screenspot_MAE_w": 0.05588290343681971,
      "eval_screenspot_MAE_x_boxes": 0.062214924643437065,
      "eval_screenspot_MAE_y_boxes": 0.03902036137878895,
      "eval_screenspot_inside_bbox": 0.725000003973643,
      "eval_screenspot_loss": 1.5115993022918701,
      "eval_screenspot_loss_ce": 0.00019147307709014663,
      "eval_screenspot_loss_iou": 0.6321614583333334,
      "eval_screenspot_loss_num": 0.061505635579427086,
      "eval_screenspot_loss_xval": 1.5716145833333333,
      "eval_screenspot_runtime": 38.7561,
      "eval_screenspot_samples_per_second": 2.296,
      "eval_screenspot_steps_per_second": 0.077,
      "num_input_tokens_seen": 448275732,
      "step": 8000
    },
    {
      "epoch": 17.817371937639198,
      "eval_compot_CIoU": 0.3504429906606674,
      "eval_compot_GIoU": 0.35775288939476013,
      "eval_compot_IoU": 0.4085061550140381,
      "eval_compot_MAE_all": 0.01769328536465764,
      "eval_compot_MAE_h": 0.008335361024364829,
      "eval_compot_MAE_w": 0.020572240464389324,
      "eval_compot_MAE_x_boxes": 0.029930624179542065,
      "eval_compot_MAE_y_boxes": 0.007017011754214764,
      "eval_compot_inside_bbox": 0.6302083432674408,
      "eval_compot_loss": 1.3885656595230103,
      "eval_compot_loss_ce": 0.0001280211581615731,
      "eval_compot_loss_iou": 0.6407470703125,
      "eval_compot_loss_num": 0.016317367553710938,
      "eval_compot_loss_xval": 1.364013671875,
      "eval_compot_runtime": 20.8794,
      "eval_compot_samples_per_second": 2.395,
      "eval_compot_steps_per_second": 0.096,
      "num_input_tokens_seen": 448275732,
      "step": 8000
    },
    {
      "epoch": 17.817371937639198,
      "eval_custom_ui_val_CIoU": 0.4733108580112457,
      "eval_custom_ui_val_GIoU": 0.47808146807882523,
      "eval_custom_ui_val_IoU": 0.535189237859514,
      "eval_custom_ui_val_MAE_all": 0.027121951182683308,
      "eval_custom_ui_val_MAE_h": 0.014743615692067478,
      "eval_custom_ui_val_MAE_w": 0.03644685043642918,
      "eval_custom_ui_val_MAE_x_boxes": 0.033436041299460664,
      "eval_custom_ui_val_MAE_y_boxes": 0.012357140529072948,
      "eval_custom_ui_val_inside_bbox": 0.7685185207260979,
      "eval_custom_ui_val_loss": 1.1683473587036133,
      "eval_custom_ui_val_loss_ce": 0.00015340095301831348,
      "eval_custom_ui_val_loss_iou": 0.5008816189236112,
      "eval_custom_ui_val_loss_num": 0.0236766603257921,
      "eval_custom_ui_val_loss_xval": 1.1200358072916667,
      "eval_custom_ui_val_runtime": 70.4441,
      "eval_custom_ui_val_samples_per_second": 3.762,
      "eval_custom_ui_val_steps_per_second": 0.128,
      "num_input_tokens_seen": 448275732,
      "step": 8000
    },
    {
      "epoch": 17.817371937639198,
      "loss": 0.8243292570114136,
      "loss_ce": 0.00011052313493564725,
      "loss_iou": 0.37109375,
      "loss_num": 0.0167236328125,
      "loss_xval": 0.82421875,
      "num_input_tokens_seen": 448275732,
      "step": 8000
    },
    {
      "epoch": 17.819599109131403,
      "grad_norm": 16.163515090942383,
      "learning_rate": 1e-06,
      "loss": 0.3863,
      "num_input_tokens_seen": 448332804,
      "step": 8001
    },
    {
      "epoch": 17.819599109131403,
      "loss": 0.4817669987678528,
      "loss_ce": 7.758761057630181e-05,
      "loss_iou": 0.2080078125,
      "loss_num": 0.013427734375,
      "loss_xval": 0.482421875,
      "num_input_tokens_seen": 448332804,
      "step": 8001
    },
    {
      "epoch": 17.821826280623608,
      "grad_norm": 21.57859992980957,
      "learning_rate": 1e-06,
      "loss": 0.4429,
      "num_input_tokens_seen": 448389324,
      "step": 8002
    },
    {
      "epoch": 17.821826280623608,
      "loss": 0.41340628266334534,
      "loss_ce": 7.620293035870418e-05,
      "loss_iou": 0.19140625,
      "loss_num": 0.005950927734375,
      "loss_xval": 0.4140625,
      "num_input_tokens_seen": 448389324,
      "step": 8002
    },
    {
      "epoch": 17.824053452115812,
      "grad_norm": 21.757307052612305,
      "learning_rate": 1e-06,
      "loss": 0.4946,
      "num_input_tokens_seen": 448446192,
      "step": 8003
    },
    {
      "epoch": 17.824053452115812,
      "loss": 0.5204148292541504,
      "loss_ce": 0.00015119729505386204,
      "loss_iou": 0.224609375,
      "loss_num": 0.0140380859375,
      "loss_xval": 0.51953125,
      "num_input_tokens_seen": 448446192,
      "step": 8003
    },
    {
      "epoch": 17.826280623608017,
      "grad_norm": 18.847604751586914,
      "learning_rate": 1e-06,
      "loss": 0.3709,
      "num_input_tokens_seen": 448503164,
      "step": 8004
    },
    {
      "epoch": 17.826280623608017,
      "loss": 0.2775651812553406,
      "loss_ce": 6.884684989927337e-05,
      "loss_iou": 0.111328125,
      "loss_num": 0.0108642578125,
      "loss_xval": 0.27734375,
      "num_input_tokens_seen": 448503164,
      "step": 8004
    },
    {
      "epoch": 17.828507795100222,
      "grad_norm": 20.84021759033203,
      "learning_rate": 1e-06,
      "loss": 0.4736,
      "num_input_tokens_seen": 448558900,
      "step": 8005
    },
    {
      "epoch": 17.828507795100222,
      "loss": 0.31135812401771545,
      "loss_ce": 7.882342470111325e-05,
      "loss_iou": 0.146484375,
      "loss_num": 0.0038299560546875,
      "loss_xval": 0.310546875,
      "num_input_tokens_seen": 448558900,
      "step": 8005
    },
    {
      "epoch": 17.830734966592427,
      "grad_norm": 16.282339096069336,
      "learning_rate": 1e-06,
      "loss": 0.398,
      "num_input_tokens_seen": 448614056,
      "step": 8006
    },
    {
      "epoch": 17.830734966592427,
      "loss": 0.30350208282470703,
      "loss_ce": 9.633424633648247e-05,
      "loss_iou": 0.13671875,
      "loss_num": 0.006103515625,
      "loss_xval": 0.302734375,
      "num_input_tokens_seen": 448614056,
      "step": 8006
    },
    {
      "epoch": 17.832962138084632,
      "grad_norm": 16.245391845703125,
      "learning_rate": 1e-06,
      "loss": 0.3856,
      "num_input_tokens_seen": 448668056,
      "step": 8007
    },
    {
      "epoch": 17.832962138084632,
      "loss": 0.30444949865341187,
      "loss_ce": 6.718316581100225e-05,
      "loss_iou": 0.13671875,
      "loss_num": 0.006195068359375,
      "loss_xval": 0.3046875,
      "num_input_tokens_seen": 448668056,
      "step": 8007
    },
    {
      "epoch": 17.835189309576837,
      "grad_norm": 21.52396011352539,
      "learning_rate": 1e-06,
      "loss": 0.4856,
      "num_input_tokens_seen": 448724780,
      "step": 8008
    },
    {
      "epoch": 17.835189309576837,
      "loss": 0.36812153458595276,
      "loss_ce": 7.954495958983898e-05,
      "loss_iou": 0.158203125,
      "loss_num": 0.01007080078125,
      "loss_xval": 0.3671875,
      "num_input_tokens_seen": 448724780,
      "step": 8008
    },
    {
      "epoch": 17.83741648106904,
      "grad_norm": 15.395381927490234,
      "learning_rate": 1e-06,
      "loss": 0.4704,
      "num_input_tokens_seen": 448781896,
      "step": 8009
    },
    {
      "epoch": 17.83741648106904,
      "loss": 0.42657172679901123,
      "loss_ce": 8.857337525114417e-05,
      "loss_iou": 0.189453125,
      "loss_num": 0.0093994140625,
      "loss_xval": 0.42578125,
      "num_input_tokens_seen": 448781896,
      "step": 8009
    },
    {
      "epoch": 17.839643652561247,
      "grad_norm": 14.230578422546387,
      "learning_rate": 1e-06,
      "loss": 0.5784,
      "num_input_tokens_seen": 448835332,
      "step": 8010
    },
    {
      "epoch": 17.839643652561247,
      "loss": 0.7218839526176453,
      "loss_ce": 8.216348942369223e-05,
      "loss_iou": 0.3125,
      "loss_num": 0.0191650390625,
      "loss_xval": 0.72265625,
      "num_input_tokens_seen": 448835332,
      "step": 8010
    },
    {
      "epoch": 17.84187082405345,
      "grad_norm": 25.663774490356445,
      "learning_rate": 1e-06,
      "loss": 0.2717,
      "num_input_tokens_seen": 448890712,
      "step": 8011
    },
    {
      "epoch": 17.84187082405345,
      "loss": 0.25164851546287537,
      "loss_ce": 6.160917837405577e-05,
      "loss_iou": 0.1142578125,
      "loss_num": 0.00469970703125,
      "loss_xval": 0.251953125,
      "num_input_tokens_seen": 448890712,
      "step": 8011
    },
    {
      "epoch": 17.844097995545656,
      "grad_norm": 16.865924835205078,
      "learning_rate": 1e-06,
      "loss": 0.4874,
      "num_input_tokens_seen": 448948344,
      "step": 8012
    },
    {
      "epoch": 17.844097995545656,
      "loss": 0.7613823413848877,
      "loss_ce": 9.086247882805765e-05,
      "loss_iou": 0.328125,
      "loss_num": 0.0205078125,
      "loss_xval": 0.76171875,
      "num_input_tokens_seen": 448948344,
      "step": 8012
    },
    {
      "epoch": 17.84632516703786,
      "grad_norm": 31.958927154541016,
      "learning_rate": 1e-06,
      "loss": 0.3505,
      "num_input_tokens_seen": 449001604,
      "step": 8013
    },
    {
      "epoch": 17.84632516703786,
      "loss": 0.2753995656967163,
      "loss_ce": 6.995358853600919e-05,
      "loss_iou": 0.11279296875,
      "loss_num": 0.009765625,
      "loss_xval": 0.275390625,
      "num_input_tokens_seen": 449001604,
      "step": 8013
    },
    {
      "epoch": 17.848552338530066,
      "grad_norm": 19.097925186157227,
      "learning_rate": 1e-06,
      "loss": 0.4448,
      "num_input_tokens_seen": 449057784,
      "step": 8014
    },
    {
      "epoch": 17.848552338530066,
      "loss": 0.42127907276153564,
      "loss_ce": 7.547304267063737e-05,
      "loss_iou": 0.17578125,
      "loss_num": 0.01373291015625,
      "loss_xval": 0.421875,
      "num_input_tokens_seen": 449057784,
      "step": 8014
    },
    {
      "epoch": 17.85077951002227,
      "grad_norm": 22.88896942138672,
      "learning_rate": 1e-06,
      "loss": 0.4073,
      "num_input_tokens_seen": 449115216,
      "step": 8015
    },
    {
      "epoch": 17.85077951002227,
      "loss": 0.40796226263046265,
      "loss_ce": 0.0004915721947327256,
      "loss_iou": 0.1611328125,
      "loss_num": 0.0172119140625,
      "loss_xval": 0.408203125,
      "num_input_tokens_seen": 449115216,
      "step": 8015
    },
    {
      "epoch": 17.853006681514476,
      "grad_norm": 17.308486938476562,
      "learning_rate": 1e-06,
      "loss": 0.3195,
      "num_input_tokens_seen": 449171056,
      "step": 8016
    },
    {
      "epoch": 17.853006681514476,
      "loss": 0.23818965256214142,
      "loss_ce": 9.149865218205377e-05,
      "loss_iou": 0.10888671875,
      "loss_num": 0.004119873046875,
      "loss_xval": 0.23828125,
      "num_input_tokens_seen": 449171056,
      "step": 8016
    },
    {
      "epoch": 17.85523385300668,
      "grad_norm": 26.236623764038086,
      "learning_rate": 1e-06,
      "loss": 0.3454,
      "num_input_tokens_seen": 449226700,
      "step": 8017
    },
    {
      "epoch": 17.85523385300668,
      "loss": 0.30879464745521545,
      "loss_ce": 7.88304241723381e-05,
      "loss_iou": 0.140625,
      "loss_num": 0.005523681640625,
      "loss_xval": 0.30859375,
      "num_input_tokens_seen": 449226700,
      "step": 8017
    },
    {
      "epoch": 17.857461024498885,
      "grad_norm": 20.232860565185547,
      "learning_rate": 1e-06,
      "loss": 0.4247,
      "num_input_tokens_seen": 449282184,
      "step": 8018
    },
    {
      "epoch": 17.857461024498885,
      "loss": 0.5520839691162109,
      "loss_ce": 8.206926577258855e-05,
      "loss_iou": 0.2265625,
      "loss_num": 0.0198974609375,
      "loss_xval": 0.55078125,
      "num_input_tokens_seen": 449282184,
      "step": 8018
    },
    {
      "epoch": 17.85968819599109,
      "grad_norm": 16.037681579589844,
      "learning_rate": 1e-06,
      "loss": 0.4071,
      "num_input_tokens_seen": 449338720,
      "step": 8019
    },
    {
      "epoch": 17.85968819599109,
      "loss": 0.3350412845611572,
      "loss_ce": 8.032634650589898e-05,
      "loss_iou": 0.1484375,
      "loss_num": 0.00750732421875,
      "loss_xval": 0.3359375,
      "num_input_tokens_seen": 449338720,
      "step": 8019
    },
    {
      "epoch": 17.861915367483295,
      "grad_norm": 15.602432250976562,
      "learning_rate": 1e-06,
      "loss": 0.3929,
      "num_input_tokens_seen": 449394160,
      "step": 8020
    },
    {
      "epoch": 17.861915367483295,
      "loss": 0.3082042932510376,
      "loss_ce": 9.883566235657781e-05,
      "loss_iou": 0.1376953125,
      "loss_num": 0.00665283203125,
      "loss_xval": 0.30859375,
      "num_input_tokens_seen": 449394160,
      "step": 8020
    },
    {
      "epoch": 17.8641425389755,
      "grad_norm": 20.469484329223633,
      "learning_rate": 1e-06,
      "loss": 0.2791,
      "num_input_tokens_seen": 449448936,
      "step": 8021
    },
    {
      "epoch": 17.8641425389755,
      "loss": 0.3228263556957245,
      "loss_ce": 7.245552114909515e-05,
      "loss_iou": 0.1298828125,
      "loss_num": 0.01263427734375,
      "loss_xval": 0.322265625,
      "num_input_tokens_seen": 449448936,
      "step": 8021
    },
    {
      "epoch": 17.866369710467705,
      "grad_norm": 11.62727165222168,
      "learning_rate": 1e-06,
      "loss": 0.3996,
      "num_input_tokens_seen": 449505480,
      "step": 8022
    },
    {
      "epoch": 17.866369710467705,
      "loss": 0.4349091947078705,
      "loss_ce": 9.47638473007828e-05,
      "loss_iou": 0.1875,
      "loss_num": 0.01202392578125,
      "loss_xval": 0.435546875,
      "num_input_tokens_seen": 449505480,
      "step": 8022
    },
    {
      "epoch": 17.86859688195991,
      "grad_norm": 24.577762603759766,
      "learning_rate": 1e-06,
      "loss": 0.4352,
      "num_input_tokens_seen": 449561952,
      "step": 8023
    },
    {
      "epoch": 17.86859688195991,
      "loss": 0.5243306756019592,
      "loss_ce": 0.00016074010636657476,
      "loss_iou": 0.2373046875,
      "loss_num": 0.009765625,
      "loss_xval": 0.5234375,
      "num_input_tokens_seen": 449561952,
      "step": 8023
    },
    {
      "epoch": 17.870824053452115,
      "grad_norm": 24.447031021118164,
      "learning_rate": 1e-06,
      "loss": 0.3999,
      "num_input_tokens_seen": 449614092,
      "step": 8024
    },
    {
      "epoch": 17.870824053452115,
      "loss": 0.44069501757621765,
      "loss_ce": 8.22499132482335e-05,
      "loss_iou": 0.1904296875,
      "loss_num": 0.01214599609375,
      "loss_xval": 0.44140625,
      "num_input_tokens_seen": 449614092,
      "step": 8024
    },
    {
      "epoch": 17.87305122494432,
      "grad_norm": 22.536909103393555,
      "learning_rate": 1e-06,
      "loss": 0.5064,
      "num_input_tokens_seen": 449669248,
      "step": 8025
    },
    {
      "epoch": 17.87305122494432,
      "loss": 0.4587002098560333,
      "loss_ce": 8.205789345083758e-05,
      "loss_iou": 0.1962890625,
      "loss_num": 0.012939453125,
      "loss_xval": 0.458984375,
      "num_input_tokens_seen": 449669248,
      "step": 8025
    },
    {
      "epoch": 17.875278396436524,
      "grad_norm": 16.079544067382812,
      "learning_rate": 1e-06,
      "loss": 0.4124,
      "num_input_tokens_seen": 449726032,
      "step": 8026
    },
    {
      "epoch": 17.875278396436524,
      "loss": 0.402201771736145,
      "loss_ce": 0.00010215782094746828,
      "loss_iou": 0.173828125,
      "loss_num": 0.0108642578125,
      "loss_xval": 0.40234375,
      "num_input_tokens_seen": 449726032,
      "step": 8026
    },
    {
      "epoch": 17.87750556792873,
      "grad_norm": 16.75348472595215,
      "learning_rate": 1e-06,
      "loss": 0.589,
      "num_input_tokens_seen": 449782292,
      "step": 8027
    },
    {
      "epoch": 17.87750556792873,
      "loss": 0.5953076481819153,
      "loss_ce": 9.280974336434156e-05,
      "loss_iou": 0.244140625,
      "loss_num": 0.021484375,
      "loss_xval": 0.59375,
      "num_input_tokens_seen": 449782292,
      "step": 8027
    },
    {
      "epoch": 17.879732739420934,
      "grad_norm": 54.10028839111328,
      "learning_rate": 1e-06,
      "loss": 0.4516,
      "num_input_tokens_seen": 449837808,
      "step": 8028
    },
    {
      "epoch": 17.879732739420934,
      "loss": 0.6945455074310303,
      "loss_ce": 8.752994472160935e-05,
      "loss_iou": 0.26953125,
      "loss_num": 0.031494140625,
      "loss_xval": 0.6953125,
      "num_input_tokens_seen": 449837808,
      "step": 8028
    },
    {
      "epoch": 17.88195991091314,
      "grad_norm": 18.08180046081543,
      "learning_rate": 1e-06,
      "loss": 0.4964,
      "num_input_tokens_seen": 449896016,
      "step": 8029
    },
    {
      "epoch": 17.88195991091314,
      "loss": 0.36804813146591187,
      "loss_ce": 0.00049442338058725,
      "loss_iou": 0.169921875,
      "loss_num": 0.005706787109375,
      "loss_xval": 0.3671875,
      "num_input_tokens_seen": 449896016,
      "step": 8029
    },
    {
      "epoch": 17.884187082405344,
      "grad_norm": 20.480897903442383,
      "learning_rate": 1e-06,
      "loss": 0.4051,
      "num_input_tokens_seen": 449952976,
      "step": 8030
    },
    {
      "epoch": 17.884187082405344,
      "loss": 0.34723442792892456,
      "loss_ce": 6.646300607826561e-05,
      "loss_iou": 0.154296875,
      "loss_num": 0.007598876953125,
      "loss_xval": 0.34765625,
      "num_input_tokens_seen": 449952976,
      "step": 8030
    },
    {
      "epoch": 17.88641425389755,
      "grad_norm": 38.36183166503906,
      "learning_rate": 1e-06,
      "loss": 0.4273,
      "num_input_tokens_seen": 450010648,
      "step": 8031
    },
    {
      "epoch": 17.88641425389755,
      "loss": 0.3772750496864319,
      "loss_ce": 7.779923907946795e-05,
      "loss_iou": 0.162109375,
      "loss_num": 0.01068115234375,
      "loss_xval": 0.376953125,
      "num_input_tokens_seen": 450010648,
      "step": 8031
    },
    {
      "epoch": 17.888641425389753,
      "grad_norm": 21.646387100219727,
      "learning_rate": 1e-06,
      "loss": 0.3668,
      "num_input_tokens_seen": 450065440,
      "step": 8032
    },
    {
      "epoch": 17.888641425389753,
      "loss": 0.363721638917923,
      "loss_ce": 7.417659799102694e-05,
      "loss_iou": 0.142578125,
      "loss_num": 0.0155029296875,
      "loss_xval": 0.36328125,
      "num_input_tokens_seen": 450065440,
      "step": 8032
    },
    {
      "epoch": 17.89086859688196,
      "grad_norm": 14.319405555725098,
      "learning_rate": 1e-06,
      "loss": 0.3845,
      "num_input_tokens_seen": 450122044,
      "step": 8033
    },
    {
      "epoch": 17.89086859688196,
      "loss": 0.41428041458129883,
      "loss_ce": 9.586406667949632e-05,
      "loss_iou": 0.1787109375,
      "loss_num": 0.01129150390625,
      "loss_xval": 0.4140625,
      "num_input_tokens_seen": 450122044,
      "step": 8033
    },
    {
      "epoch": 17.893095768374163,
      "grad_norm": 35.73308563232422,
      "learning_rate": 1e-06,
      "loss": 0.4115,
      "num_input_tokens_seen": 450178668,
      "step": 8034
    },
    {
      "epoch": 17.893095768374163,
      "loss": 0.4911773204803467,
      "loss_ce": 8.844825788401067e-05,
      "loss_iou": 0.203125,
      "loss_num": 0.016845703125,
      "loss_xval": 0.490234375,
      "num_input_tokens_seen": 450178668,
      "step": 8034
    },
    {
      "epoch": 17.895322939866368,
      "grad_norm": 20.128206253051758,
      "learning_rate": 1e-06,
      "loss": 0.2906,
      "num_input_tokens_seen": 450235868,
      "step": 8035
    },
    {
      "epoch": 17.895322939866368,
      "loss": 0.31039196252822876,
      "loss_ce": 8.923389395931736e-05,
      "loss_iou": 0.13671875,
      "loss_num": 0.00732421875,
      "loss_xval": 0.310546875,
      "num_input_tokens_seen": 450235868,
      "step": 8035
    },
    {
      "epoch": 17.897550111358576,
      "grad_norm": 16.04209327697754,
      "learning_rate": 1e-06,
      "loss": 0.3579,
      "num_input_tokens_seen": 450292616,
      "step": 8036
    },
    {
      "epoch": 17.897550111358576,
      "loss": 0.3482053279876709,
      "loss_ce": 0.00018285455007571727,
      "loss_iou": 0.16015625,
      "loss_num": 0.00579833984375,
      "loss_xval": 0.34765625,
      "num_input_tokens_seen": 450292616,
      "step": 8036
    },
    {
      "epoch": 17.899777282850778,
      "grad_norm": 25.872989654541016,
      "learning_rate": 1e-06,
      "loss": 0.4616,
      "num_input_tokens_seen": 450348424,
      "step": 8037
    },
    {
      "epoch": 17.899777282850778,
      "loss": 0.4966566562652588,
      "loss_ce": 7.462559005944058e-05,
      "loss_iou": 0.208984375,
      "loss_num": 0.0157470703125,
      "loss_xval": 0.49609375,
      "num_input_tokens_seen": 450348424,
      "step": 8037
    },
    {
      "epoch": 17.902004454342986,
      "grad_norm": 17.968185424804688,
      "learning_rate": 1e-06,
      "loss": 0.397,
      "num_input_tokens_seen": 450404344,
      "step": 8038
    },
    {
      "epoch": 17.902004454342986,
      "loss": 0.43942493200302124,
      "loss_ce": 9.388932812726125e-05,
      "loss_iou": 0.1962890625,
      "loss_num": 0.00927734375,
      "loss_xval": 0.439453125,
      "num_input_tokens_seen": 450404344,
      "step": 8038
    },
    {
      "epoch": 17.90423162583519,
      "grad_norm": 22.768774032592773,
      "learning_rate": 1e-06,
      "loss": 0.4194,
      "num_input_tokens_seen": 450457360,
      "step": 8039
    },
    {
      "epoch": 17.90423162583519,
      "loss": 0.4151099920272827,
      "loss_ce": 7.091661973390728e-05,
      "loss_iou": 0.1806640625,
      "loss_num": 0.01080322265625,
      "loss_xval": 0.4140625,
      "num_input_tokens_seen": 450457360,
      "step": 8039
    },
    {
      "epoch": 17.906458797327396,
      "grad_norm": 11.735021591186523,
      "learning_rate": 1e-06,
      "loss": 0.4589,
      "num_input_tokens_seen": 450512100,
      "step": 8040
    },
    {
      "epoch": 17.906458797327396,
      "loss": 0.5093483924865723,
      "loss_ce": 0.0001931376027641818,
      "loss_iou": 0.224609375,
      "loss_num": 0.011962890625,
      "loss_xval": 0.5078125,
      "num_input_tokens_seen": 450512100,
      "step": 8040
    },
    {
      "epoch": 17.9086859688196,
      "grad_norm": 16.40501594543457,
      "learning_rate": 1e-06,
      "loss": 0.4686,
      "num_input_tokens_seen": 450569724,
      "step": 8041
    },
    {
      "epoch": 17.9086859688196,
      "loss": 0.3620176613330841,
      "loss_ce": 7.919156632851809e-05,
      "loss_iou": 0.16796875,
      "loss_num": 0.0050048828125,
      "loss_xval": 0.361328125,
      "num_input_tokens_seen": 450569724,
      "step": 8041
    },
    {
      "epoch": 17.910913140311806,
      "grad_norm": 18.876176834106445,
      "learning_rate": 1e-06,
      "loss": 0.336,
      "num_input_tokens_seen": 450626440,
      "step": 8042
    },
    {
      "epoch": 17.910913140311806,
      "loss": 0.3145880401134491,
      "loss_ce": 7.385006756521761e-05,
      "loss_iou": 0.14453125,
      "loss_num": 0.005096435546875,
      "loss_xval": 0.314453125,
      "num_input_tokens_seen": 450626440,
      "step": 8042
    },
    {
      "epoch": 17.91314031180401,
      "grad_norm": 12.777106285095215,
      "learning_rate": 1e-06,
      "loss": 0.3849,
      "num_input_tokens_seen": 450684836,
      "step": 8043
    },
    {
      "epoch": 17.91314031180401,
      "loss": 0.5031352043151855,
      "loss_ce": 8.34753445815295e-05,
      "loss_iou": 0.2294921875,
      "loss_num": 0.00885009765625,
      "loss_xval": 0.50390625,
      "num_input_tokens_seen": 450684836,
      "step": 8043
    },
    {
      "epoch": 17.915367483296215,
      "grad_norm": 30.933420181274414,
      "learning_rate": 1e-06,
      "loss": 0.357,
      "num_input_tokens_seen": 450739768,
      "step": 8044
    },
    {
      "epoch": 17.915367483296215,
      "loss": 0.3753129243850708,
      "loss_ce": 6.877203122712672e-05,
      "loss_iou": 0.1640625,
      "loss_num": 0.009521484375,
      "loss_xval": 0.375,
      "num_input_tokens_seen": 450739768,
      "step": 8044
    },
    {
      "epoch": 17.91759465478842,
      "grad_norm": 21.37887191772461,
      "learning_rate": 1e-06,
      "loss": 0.3017,
      "num_input_tokens_seen": 450796720,
      "step": 8045
    },
    {
      "epoch": 17.91759465478842,
      "loss": 0.34577932953834534,
      "loss_ce": 7.619416282977909e-05,
      "loss_iou": 0.1572265625,
      "loss_num": 0.006256103515625,
      "loss_xval": 0.345703125,
      "num_input_tokens_seen": 450796720,
      "step": 8045
    },
    {
      "epoch": 17.919821826280625,
      "grad_norm": 21.42198944091797,
      "learning_rate": 1e-06,
      "loss": 0.5496,
      "num_input_tokens_seen": 450852760,
      "step": 8046
    },
    {
      "epoch": 17.919821826280625,
      "loss": 0.5246258974075317,
      "loss_ce": 8.982230065157637e-05,
      "loss_iou": 0.23046875,
      "loss_num": 0.01263427734375,
      "loss_xval": 0.5234375,
      "num_input_tokens_seen": 450852760,
      "step": 8046
    },
    {
      "epoch": 17.92204899777283,
      "grad_norm": 19.944995880126953,
      "learning_rate": 1e-06,
      "loss": 0.3975,
      "num_input_tokens_seen": 450909692,
      "step": 8047
    },
    {
      "epoch": 17.92204899777283,
      "loss": 0.4114391803741455,
      "loss_ce": 6.225903052836657e-05,
      "loss_iou": 0.185546875,
      "loss_num": 0.0081787109375,
      "loss_xval": 0.412109375,
      "num_input_tokens_seen": 450909692,
      "step": 8047
    },
    {
      "epoch": 17.924276169265035,
      "grad_norm": 22.188114166259766,
      "learning_rate": 1e-06,
      "loss": 0.4788,
      "num_input_tokens_seen": 450968200,
      "step": 8048
    },
    {
      "epoch": 17.924276169265035,
      "loss": 0.4510143995285034,
      "loss_ce": 8.66201298777014e-05,
      "loss_iou": 0.1923828125,
      "loss_num": 0.01312255859375,
      "loss_xval": 0.451171875,
      "num_input_tokens_seen": 450968200,
      "step": 8048
    },
    {
      "epoch": 17.92650334075724,
      "grad_norm": 16.4459228515625,
      "learning_rate": 1e-06,
      "loss": 0.3947,
      "num_input_tokens_seen": 451022056,
      "step": 8049
    },
    {
      "epoch": 17.92650334075724,
      "loss": 0.35028839111328125,
      "loss_ce": 6.865533214295283e-05,
      "loss_iou": 0.142578125,
      "loss_num": 0.01318359375,
      "loss_xval": 0.349609375,
      "num_input_tokens_seen": 451022056,
      "step": 8049
    },
    {
      "epoch": 17.928730512249444,
      "grad_norm": 14.635376930236816,
      "learning_rate": 1e-06,
      "loss": 0.3044,
      "num_input_tokens_seen": 451078512,
      "step": 8050
    },
    {
      "epoch": 17.928730512249444,
      "loss": 0.28396010398864746,
      "loss_ce": 8.557151886634529e-05,
      "loss_iou": 0.1240234375,
      "loss_num": 0.0072021484375,
      "loss_xval": 0.283203125,
      "num_input_tokens_seen": 451078512,
      "step": 8050
    },
    {
      "epoch": 17.93095768374165,
      "grad_norm": 24.16907501220703,
      "learning_rate": 1e-06,
      "loss": 0.4527,
      "num_input_tokens_seen": 451133112,
      "step": 8051
    },
    {
      "epoch": 17.93095768374165,
      "loss": 0.5191126465797424,
      "loss_ce": 0.0003138238680548966,
      "loss_iou": 0.1923828125,
      "loss_num": 0.026611328125,
      "loss_xval": 0.51953125,
      "num_input_tokens_seen": 451133112,
      "step": 8051
    },
    {
      "epoch": 17.933184855233854,
      "grad_norm": 13.90976333618164,
      "learning_rate": 1e-06,
      "loss": 0.3535,
      "num_input_tokens_seen": 451190024,
      "step": 8052
    },
    {
      "epoch": 17.933184855233854,
      "loss": 0.3013458847999573,
      "loss_ce": 7.635784277226776e-05,
      "loss_iou": 0.140625,
      "loss_num": 0.004180908203125,
      "loss_xval": 0.30078125,
      "num_input_tokens_seen": 451190024,
      "step": 8052
    },
    {
      "epoch": 17.93541202672606,
      "grad_norm": 17.68203353881836,
      "learning_rate": 1e-06,
      "loss": 0.3295,
      "num_input_tokens_seen": 451243952,
      "step": 8053
    },
    {
      "epoch": 17.93541202672606,
      "loss": 0.3571292757987976,
      "loss_ce": 7.360937888734043e-05,
      "loss_iou": 0.1611328125,
      "loss_num": 0.00701904296875,
      "loss_xval": 0.357421875,
      "num_input_tokens_seen": 451243952,
      "step": 8053
    },
    {
      "epoch": 17.937639198218264,
      "grad_norm": 24.68354034423828,
      "learning_rate": 1e-06,
      "loss": 0.4696,
      "num_input_tokens_seen": 451299776,
      "step": 8054
    },
    {
      "epoch": 17.937639198218264,
      "loss": 0.3794766664505005,
      "loss_ce": 8.214540139306337e-05,
      "loss_iou": 0.1689453125,
      "loss_num": 0.00836181640625,
      "loss_xval": 0.37890625,
      "num_input_tokens_seen": 451299776,
      "step": 8054
    },
    {
      "epoch": 17.93986636971047,
      "grad_norm": 12.557443618774414,
      "learning_rate": 1e-06,
      "loss": 0.3637,
      "num_input_tokens_seen": 451354904,
      "step": 8055
    },
    {
      "epoch": 17.93986636971047,
      "loss": 0.26801323890686035,
      "loss_ce": 6.891635712236166e-05,
      "loss_iou": 0.12109375,
      "loss_num": 0.005157470703125,
      "loss_xval": 0.267578125,
      "num_input_tokens_seen": 451354904,
      "step": 8055
    },
    {
      "epoch": 17.942093541202674,
      "grad_norm": 31.136751174926758,
      "learning_rate": 1e-06,
      "loss": 0.4939,
      "num_input_tokens_seen": 451406784,
      "step": 8056
    },
    {
      "epoch": 17.942093541202674,
      "loss": 0.5043526291847229,
      "loss_ce": 8.016474748728797e-05,
      "loss_iou": 0.2265625,
      "loss_num": 0.010498046875,
      "loss_xval": 0.50390625,
      "num_input_tokens_seen": 451406784,
      "step": 8056
    },
    {
      "epoch": 17.94432071269488,
      "grad_norm": 12.46948528289795,
      "learning_rate": 1e-06,
      "loss": 0.3284,
      "num_input_tokens_seen": 451464456,
      "step": 8057
    },
    {
      "epoch": 17.94432071269488,
      "loss": 0.29316914081573486,
      "loss_ce": 7.832865230739117e-05,
      "loss_iou": 0.1357421875,
      "loss_num": 0.004425048828125,
      "loss_xval": 0.29296875,
      "num_input_tokens_seen": 451464456,
      "step": 8057
    },
    {
      "epoch": 17.946547884187083,
      "grad_norm": 26.139951705932617,
      "learning_rate": 1e-06,
      "loss": 0.3216,
      "num_input_tokens_seen": 451517900,
      "step": 8058
    },
    {
      "epoch": 17.946547884187083,
      "loss": 0.3385404348373413,
      "loss_ce": 0.00010048142576124519,
      "loss_iou": 0.1572265625,
      "loss_num": 0.004852294921875,
      "loss_xval": 0.337890625,
      "num_input_tokens_seen": 451517900,
      "step": 8058
    },
    {
      "epoch": 17.948775055679288,
      "grad_norm": 20.38810920715332,
      "learning_rate": 1e-06,
      "loss": 0.3234,
      "num_input_tokens_seen": 451574048,
      "step": 8059
    },
    {
      "epoch": 17.948775055679288,
      "loss": 0.2848048210144043,
      "loss_ce": 7.584408012917265e-05,
      "loss_iou": 0.1298828125,
      "loss_num": 0.005096435546875,
      "loss_xval": 0.28515625,
      "num_input_tokens_seen": 451574048,
      "step": 8059
    },
    {
      "epoch": 17.951002227171493,
      "grad_norm": 17.65831756591797,
      "learning_rate": 1e-06,
      "loss": 0.5641,
      "num_input_tokens_seen": 451632452,
      "step": 8060
    },
    {
      "epoch": 17.951002227171493,
      "loss": 0.38423237204551697,
      "loss_ce": 7.70928745623678e-05,
      "loss_iou": 0.1748046875,
      "loss_num": 0.006805419921875,
      "loss_xval": 0.384765625,
      "num_input_tokens_seen": 451632452,
      "step": 8060
    },
    {
      "epoch": 17.953229398663698,
      "grad_norm": 10.756651878356934,
      "learning_rate": 1e-06,
      "loss": 0.362,
      "num_input_tokens_seen": 451687680,
      "step": 8061
    },
    {
      "epoch": 17.953229398663698,
      "loss": 0.46239495277404785,
      "loss_ce": 8.412712486460805e-05,
      "loss_iou": 0.171875,
      "loss_num": 0.0238037109375,
      "loss_xval": 0.462890625,
      "num_input_tokens_seen": 451687680,
      "step": 8061
    },
    {
      "epoch": 17.955456570155903,
      "grad_norm": 22.19445037841797,
      "learning_rate": 1e-06,
      "loss": 0.4,
      "num_input_tokens_seen": 451746496,
      "step": 8062
    },
    {
      "epoch": 17.955456570155903,
      "loss": 0.3811803460121155,
      "loss_ce": 7.682880095671862e-05,
      "loss_iou": 0.1669921875,
      "loss_num": 0.009521484375,
      "loss_xval": 0.380859375,
      "num_input_tokens_seen": 451746496,
      "step": 8062
    },
    {
      "epoch": 17.957683741648108,
      "grad_norm": 14.64289665222168,
      "learning_rate": 1e-06,
      "loss": 0.3557,
      "num_input_tokens_seen": 451802292,
      "step": 8063
    },
    {
      "epoch": 17.957683741648108,
      "loss": 0.3765408396720886,
      "loss_ce": 0.00010646959708537906,
      "loss_iou": 0.1572265625,
      "loss_num": 0.01239013671875,
      "loss_xval": 0.376953125,
      "num_input_tokens_seen": 451802292,
      "step": 8063
    },
    {
      "epoch": 17.959910913140313,
      "grad_norm": 20.15620994567871,
      "learning_rate": 1e-06,
      "loss": 0.5471,
      "num_input_tokens_seen": 451854968,
      "step": 8064
    },
    {
      "epoch": 17.959910913140313,
      "loss": 0.5528594255447388,
      "loss_ce": 6.400723214028403e-05,
      "loss_iou": 0.2421875,
      "loss_num": 0.01373291015625,
      "loss_xval": 0.5546875,
      "num_input_tokens_seen": 451854968,
      "step": 8064
    },
    {
      "epoch": 17.962138084632517,
      "grad_norm": 16.74308967590332,
      "learning_rate": 1e-06,
      "loss": 0.3935,
      "num_input_tokens_seen": 451913180,
      "step": 8065
    },
    {
      "epoch": 17.962138084632517,
      "loss": 0.5324406623840332,
      "loss_ce": 9.207165567204356e-05,
      "loss_iou": 0.2353515625,
      "loss_num": 0.01239013671875,
      "loss_xval": 0.53125,
      "num_input_tokens_seen": 451913180,
      "step": 8065
    },
    {
      "epoch": 17.964365256124722,
      "grad_norm": 17.335187911987305,
      "learning_rate": 1e-06,
      "loss": 0.5185,
      "num_input_tokens_seen": 451969652,
      "step": 8066
    },
    {
      "epoch": 17.964365256124722,
      "loss": 0.47579044103622437,
      "loss_ce": 8.243897173088044e-05,
      "loss_iou": 0.189453125,
      "loss_num": 0.0194091796875,
      "loss_xval": 0.4765625,
      "num_input_tokens_seen": 451969652,
      "step": 8066
    },
    {
      "epoch": 17.966592427616927,
      "grad_norm": 20.39455795288086,
      "learning_rate": 1e-06,
      "loss": 0.4586,
      "num_input_tokens_seen": 452024588,
      "step": 8067
    },
    {
      "epoch": 17.966592427616927,
      "loss": 0.6492767333984375,
      "loss_ce": 0.00010679697152227163,
      "loss_iou": 0.26171875,
      "loss_num": 0.0247802734375,
      "loss_xval": 0.6484375,
      "num_input_tokens_seen": 452024588,
      "step": 8067
    },
    {
      "epoch": 17.968819599109132,
      "grad_norm": 21.565893173217773,
      "learning_rate": 1e-06,
      "loss": 0.3403,
      "num_input_tokens_seen": 452079184,
      "step": 8068
    },
    {
      "epoch": 17.968819599109132,
      "loss": 0.3689712882041931,
      "loss_ce": 7.478379120584577e-05,
      "loss_iou": 0.1650390625,
      "loss_num": 0.007598876953125,
      "loss_xval": 0.369140625,
      "num_input_tokens_seen": 452079184,
      "step": 8068
    },
    {
      "epoch": 17.971046770601337,
      "grad_norm": 53.53007507324219,
      "learning_rate": 1e-06,
      "loss": 0.4302,
      "num_input_tokens_seen": 452132824,
      "step": 8069
    },
    {
      "epoch": 17.971046770601337,
      "loss": 0.3132534623146057,
      "loss_ce": 0.00014310533879324794,
      "loss_iou": 0.146484375,
      "loss_num": 0.004058837890625,
      "loss_xval": 0.3125,
      "num_input_tokens_seen": 452132824,
      "step": 8069
    },
    {
      "epoch": 17.97327394209354,
      "grad_norm": 25.093297958374023,
      "learning_rate": 1e-06,
      "loss": 0.3963,
      "num_input_tokens_seen": 452188908,
      "step": 8070
    },
    {
      "epoch": 17.97327394209354,
      "loss": 0.3288014531135559,
      "loss_ce": 6.609824777115136e-05,
      "loss_iou": 0.1435546875,
      "loss_num": 0.0084228515625,
      "loss_xval": 0.328125,
      "num_input_tokens_seen": 452188908,
      "step": 8070
    },
    {
      "epoch": 17.975501113585747,
      "grad_norm": 14.360114097595215,
      "learning_rate": 1e-06,
      "loss": 0.4598,
      "num_input_tokens_seen": 452245072,
      "step": 8071
    },
    {
      "epoch": 17.975501113585747,
      "loss": 0.49251696467399597,
      "loss_ce": 8.531761704944074e-05,
      "loss_iou": 0.203125,
      "loss_num": 0.01708984375,
      "loss_xval": 0.4921875,
      "num_input_tokens_seen": 452245072,
      "step": 8071
    },
    {
      "epoch": 17.97772828507795,
      "grad_norm": 19.300607681274414,
      "learning_rate": 1e-06,
      "loss": 0.4567,
      "num_input_tokens_seen": 452302856,
      "step": 8072
    },
    {
      "epoch": 17.97772828507795,
      "loss": 0.47664445638656616,
      "loss_ce": 8.195844566216692e-05,
      "loss_iou": 0.2060546875,
      "loss_num": 0.0125732421875,
      "loss_xval": 0.4765625,
      "num_input_tokens_seen": 452302856,
      "step": 8072
    },
    {
      "epoch": 17.979955456570156,
      "grad_norm": 22.800689697265625,
      "learning_rate": 1e-06,
      "loss": 0.3213,
      "num_input_tokens_seen": 452358036,
      "step": 8073
    },
    {
      "epoch": 17.979955456570156,
      "loss": 0.30914774537086487,
      "loss_ce": 6.572413258254528e-05,
      "loss_iou": 0.1259765625,
      "loss_num": 0.01153564453125,
      "loss_xval": 0.30859375,
      "num_input_tokens_seen": 452358036,
      "step": 8073
    },
    {
      "epoch": 17.98218262806236,
      "grad_norm": 18.20235824584961,
      "learning_rate": 1e-06,
      "loss": 0.3886,
      "num_input_tokens_seen": 452412620,
      "step": 8074
    },
    {
      "epoch": 17.98218262806236,
      "loss": 0.3233228921890259,
      "loss_ce": 8.070748299360275e-05,
      "loss_iou": 0.1435546875,
      "loss_num": 0.0072021484375,
      "loss_xval": 0.32421875,
      "num_input_tokens_seen": 452412620,
      "step": 8074
    },
    {
      "epoch": 17.984409799554566,
      "grad_norm": 14.406597137451172,
      "learning_rate": 1e-06,
      "loss": 0.4968,
      "num_input_tokens_seen": 452469772,
      "step": 8075
    },
    {
      "epoch": 17.984409799554566,
      "loss": 0.5883782505989075,
      "loss_ce": 0.00012145160872023553,
      "loss_iou": 0.25,
      "loss_num": 0.017333984375,
      "loss_xval": 0.58984375,
      "num_input_tokens_seen": 452469772,
      "step": 8075
    },
    {
      "epoch": 17.98663697104677,
      "grad_norm": 26.771181106567383,
      "learning_rate": 1e-06,
      "loss": 0.4041,
      "num_input_tokens_seen": 452523692,
      "step": 8076
    },
    {
      "epoch": 17.98663697104677,
      "loss": 0.5332375764846802,
      "loss_ce": 9.549761307425797e-05,
      "loss_iou": 0.2109375,
      "loss_num": 0.0220947265625,
      "loss_xval": 0.53125,
      "num_input_tokens_seen": 452523692,
      "step": 8076
    },
    {
      "epoch": 17.988864142538976,
      "grad_norm": 15.690217971801758,
      "learning_rate": 1e-06,
      "loss": 0.3856,
      "num_input_tokens_seen": 452581640,
      "step": 8077
    },
    {
      "epoch": 17.988864142538976,
      "loss": 0.3996119499206543,
      "loss_ce": 0.0001978981599677354,
      "loss_iou": 0.166015625,
      "loss_num": 0.01348876953125,
      "loss_xval": 0.3984375,
      "num_input_tokens_seen": 452581640,
      "step": 8077
    },
    {
      "epoch": 17.99109131403118,
      "grad_norm": 20.969528198242188,
      "learning_rate": 1e-06,
      "loss": 0.3346,
      "num_input_tokens_seen": 452634292,
      "step": 8078
    },
    {
      "epoch": 17.99109131403118,
      "loss": 0.3364979922771454,
      "loss_ce": 7.221752457553521e-05,
      "loss_iou": 0.1552734375,
      "loss_num": 0.005035400390625,
      "loss_xval": 0.3359375,
      "num_input_tokens_seen": 452634292,
      "step": 8078
    },
    {
      "epoch": 17.993318485523385,
      "grad_norm": 22.376567840576172,
      "learning_rate": 1e-06,
      "loss": 0.4623,
      "num_input_tokens_seen": 452691208,
      "step": 8079
    },
    {
      "epoch": 17.993318485523385,
      "loss": 0.38116979598999023,
      "loss_ce": 6.630241841776296e-05,
      "loss_iou": 0.1728515625,
      "loss_num": 0.007232666015625,
      "loss_xval": 0.380859375,
      "num_input_tokens_seen": 452691208,
      "step": 8079
    },
    {
      "epoch": 17.99554565701559,
      "grad_norm": 25.332250595092773,
      "learning_rate": 1e-06,
      "loss": 0.38,
      "num_input_tokens_seen": 452744012,
      "step": 8080
    },
    {
      "epoch": 17.99554565701559,
      "loss": 0.5036153793334961,
      "loss_ce": 0.00019747592159546912,
      "loss_iou": 0.21875,
      "loss_num": 0.0133056640625,
      "loss_xval": 0.50390625,
      "num_input_tokens_seen": 452744012,
      "step": 8080
    },
    {
      "epoch": 17.997772828507795,
      "grad_norm": 21.98955726623535,
      "learning_rate": 1e-06,
      "loss": 0.3497,
      "num_input_tokens_seen": 452801908,
      "step": 8081
    },
    {
      "epoch": 17.997772828507795,
      "loss": 0.3676411509513855,
      "loss_ce": 8.742515638004988e-05,
      "loss_iou": 0.166015625,
      "loss_num": 0.007080078125,
      "loss_xval": 0.3671875,
      "num_input_tokens_seen": 452801908,
      "step": 8081
    },
    {
      "epoch": 18.0,
      "grad_norm": 16.604522705078125,
      "learning_rate": 1e-06,
      "loss": 0.4416,
      "num_input_tokens_seen": 452859412,
      "step": 8082
    },
    {
      "epoch": 18.0,
      "loss": 0.5428001880645752,
      "loss_ce": 7.557860226370394e-05,
      "loss_iou": 0.216796875,
      "loss_num": 0.02197265625,
      "loss_xval": 0.54296875,
      "num_input_tokens_seen": 452859412,
      "step": 8082
    },
    {
      "epoch": 18.002227171492205,
      "grad_norm": 13.554404258728027,
      "learning_rate": 1e-06,
      "loss": 0.3327,
      "num_input_tokens_seen": 452915892,
      "step": 8083
    },
    {
      "epoch": 18.002227171492205,
      "loss": 0.16937774419784546,
      "loss_ce": 6.622253567911685e-05,
      "loss_iou": 0.0712890625,
      "loss_num": 0.005401611328125,
      "loss_xval": 0.1689453125,
      "num_input_tokens_seen": 452915892,
      "step": 8083
    },
    {
      "epoch": 18.00445434298441,
      "grad_norm": 21.166494369506836,
      "learning_rate": 1e-06,
      "loss": 0.4786,
      "num_input_tokens_seen": 452973948,
      "step": 8084
    },
    {
      "epoch": 18.00445434298441,
      "loss": 0.3788573145866394,
      "loss_ce": 7.313516107387841e-05,
      "loss_iou": 0.17578125,
      "loss_num": 0.005584716796875,
      "loss_xval": 0.37890625,
      "num_input_tokens_seen": 452973948,
      "step": 8084
    },
    {
      "epoch": 18.006681514476615,
      "grad_norm": 20.805810928344727,
      "learning_rate": 1e-06,
      "loss": 0.4782,
      "num_input_tokens_seen": 453030260,
      "step": 8085
    },
    {
      "epoch": 18.006681514476615,
      "loss": 0.5902899503707886,
      "loss_ce": 8.001519017852843e-05,
      "loss_iou": 0.2373046875,
      "loss_num": 0.0230712890625,
      "loss_xval": 0.58984375,
      "num_input_tokens_seen": 453030260,
      "step": 8085
    },
    {
      "epoch": 18.00890868596882,
      "grad_norm": 16.5931396484375,
      "learning_rate": 1e-06,
      "loss": 0.4174,
      "num_input_tokens_seen": 453085656,
      "step": 8086
    },
    {
      "epoch": 18.00890868596882,
      "loss": 0.4264695942401886,
      "loss_ce": 7.800738967489451e-05,
      "loss_iou": 0.185546875,
      "loss_num": 0.0108642578125,
      "loss_xval": 0.42578125,
      "num_input_tokens_seen": 453085656,
      "step": 8086
    },
    {
      "epoch": 18.011135857461024,
      "grad_norm": 20.476335525512695,
      "learning_rate": 1e-06,
      "loss": 0.3997,
      "num_input_tokens_seen": 453141140,
      "step": 8087
    },
    {
      "epoch": 18.011135857461024,
      "loss": 0.40523386001586914,
      "loss_ce": 0.0002045718429144472,
      "loss_iou": 0.1884765625,
      "loss_num": 0.005462646484375,
      "loss_xval": 0.404296875,
      "num_input_tokens_seen": 453141140,
      "step": 8087
    },
    {
      "epoch": 18.01336302895323,
      "grad_norm": 26.6048641204834,
      "learning_rate": 1e-06,
      "loss": 0.3262,
      "num_input_tokens_seen": 453199108,
      "step": 8088
    },
    {
      "epoch": 18.01336302895323,
      "loss": 0.3548569679260254,
      "loss_ce": 5.9605521528283134e-05,
      "loss_iou": 0.1572265625,
      "loss_num": 0.008056640625,
      "loss_xval": 0.35546875,
      "num_input_tokens_seen": 453199108,
      "step": 8088
    },
    {
      "epoch": 18.015590200445434,
      "grad_norm": 28.440725326538086,
      "learning_rate": 1e-06,
      "loss": 0.56,
      "num_input_tokens_seen": 453254600,
      "step": 8089
    },
    {
      "epoch": 18.015590200445434,
      "loss": 0.6332626342773438,
      "loss_ce": 8.392542076762766e-05,
      "loss_iou": 0.283203125,
      "loss_num": 0.0137939453125,
      "loss_xval": 0.6328125,
      "num_input_tokens_seen": 453254600,
      "step": 8089
    },
    {
      "epoch": 18.01781737193764,
      "grad_norm": 17.789857864379883,
      "learning_rate": 1e-06,
      "loss": 0.5431,
      "num_input_tokens_seen": 453311020,
      "step": 8090
    },
    {
      "epoch": 18.01781737193764,
      "loss": 0.7038158774375916,
      "loss_ce": 8.052912016864866e-05,
      "loss_iou": 0.30078125,
      "loss_num": 0.0201416015625,
      "loss_xval": 0.703125,
      "num_input_tokens_seen": 453311020,
      "step": 8090
    },
    {
      "epoch": 18.020044543429844,
      "grad_norm": 19.747709274291992,
      "learning_rate": 1e-06,
      "loss": 0.378,
      "num_input_tokens_seen": 453364652,
      "step": 8091
    },
    {
      "epoch": 18.020044543429844,
      "loss": 0.47260355949401855,
      "loss_ce": 6.938957085367292e-05,
      "loss_iou": 0.2119140625,
      "loss_num": 0.00970458984375,
      "loss_xval": 0.47265625,
      "num_input_tokens_seen": 453364652,
      "step": 8091
    },
    {
      "epoch": 18.02227171492205,
      "grad_norm": 24.440204620361328,
      "learning_rate": 1e-06,
      "loss": 0.4914,
      "num_input_tokens_seen": 453419676,
      "step": 8092
    },
    {
      "epoch": 18.02227171492205,
      "loss": 0.43805378675460815,
      "loss_ce": 6.548190140165389e-05,
      "loss_iou": 0.173828125,
      "loss_num": 0.01806640625,
      "loss_xval": 0.4375,
      "num_input_tokens_seen": 453419676,
      "step": 8092
    },
    {
      "epoch": 18.024498886414253,
      "grad_norm": 14.660680770874023,
      "learning_rate": 1e-06,
      "loss": 0.4239,
      "num_input_tokens_seen": 453476280,
      "step": 8093
    },
    {
      "epoch": 18.024498886414253,
      "loss": 0.3206265866756439,
      "loss_ce": 6.993835268076509e-05,
      "loss_iou": 0.1376953125,
      "loss_num": 0.009033203125,
      "loss_xval": 0.3203125,
      "num_input_tokens_seen": 453476280,
      "step": 8093
    },
    {
      "epoch": 18.02672605790646,
      "grad_norm": 22.54802894592285,
      "learning_rate": 1e-06,
      "loss": 0.4878,
      "num_input_tokens_seen": 453529916,
      "step": 8094
    },
    {
      "epoch": 18.02672605790646,
      "loss": 0.37210169434547424,
      "loss_ce": 9.240545477950945e-05,
      "loss_iou": 0.16796875,
      "loss_num": 0.007354736328125,
      "loss_xval": 0.37109375,
      "num_input_tokens_seen": 453529916,
      "step": 8094
    },
    {
      "epoch": 18.028953229398663,
      "grad_norm": 20.00684928894043,
      "learning_rate": 1e-06,
      "loss": 0.3406,
      "num_input_tokens_seen": 453583004,
      "step": 8095
    },
    {
      "epoch": 18.028953229398663,
      "loss": 0.39649391174316406,
      "loss_ce": 7.059978815959767e-05,
      "loss_iou": 0.171875,
      "loss_num": 0.0106201171875,
      "loss_xval": 0.396484375,
      "num_input_tokens_seen": 453583004,
      "step": 8095
    },
    {
      "epoch": 18.031180400890868,
      "grad_norm": 25.416080474853516,
      "learning_rate": 1e-06,
      "loss": 0.5202,
      "num_input_tokens_seen": 453637924,
      "step": 8096
    },
    {
      "epoch": 18.031180400890868,
      "loss": 0.5738502740859985,
      "loss_ce": 0.00011981255374848843,
      "loss_iou": 0.255859375,
      "loss_num": 0.0128173828125,
      "loss_xval": 0.57421875,
      "num_input_tokens_seen": 453637924,
      "step": 8096
    },
    {
      "epoch": 18.033407572383073,
      "grad_norm": 17.065658569335938,
      "learning_rate": 1e-06,
      "loss": 0.4977,
      "num_input_tokens_seen": 453693352,
      "step": 8097
    },
    {
      "epoch": 18.033407572383073,
      "loss": 0.32081741094589233,
      "loss_ce": 7.766792987240478e-05,
      "loss_iou": 0.12890625,
      "loss_num": 0.0126953125,
      "loss_xval": 0.3203125,
      "num_input_tokens_seen": 453693352,
      "step": 8097
    },
    {
      "epoch": 18.035634743875278,
      "grad_norm": 18.961402893066406,
      "learning_rate": 1e-06,
      "loss": 0.4149,
      "num_input_tokens_seen": 453748036,
      "step": 8098
    },
    {
      "epoch": 18.035634743875278,
      "loss": 0.41537126898765564,
      "loss_ce": 8.809100836515427e-05,
      "loss_iou": 0.169921875,
      "loss_num": 0.014892578125,
      "loss_xval": 0.416015625,
      "num_input_tokens_seen": 453748036,
      "step": 8098
    },
    {
      "epoch": 18.037861915367483,
      "grad_norm": 19.813962936401367,
      "learning_rate": 1e-06,
      "loss": 0.2732,
      "num_input_tokens_seen": 453804100,
      "step": 8099
    },
    {
      "epoch": 18.037861915367483,
      "loss": 0.27010267972946167,
      "loss_ce": 8.316422463394701e-05,
      "loss_iou": 0.1142578125,
      "loss_num": 0.00823974609375,
      "loss_xval": 0.26953125,
      "num_input_tokens_seen": 453804100,
      "step": 8099
    },
    {
      "epoch": 18.040089086859687,
      "grad_norm": 21.94256019592285,
      "learning_rate": 1e-06,
      "loss": 0.3819,
      "num_input_tokens_seen": 453862180,
      "step": 8100
    },
    {
      "epoch": 18.040089086859687,
      "loss": 0.40962302684783936,
      "loss_ce": 7.715914398431778e-05,
      "loss_iou": 0.189453125,
      "loss_num": 0.00628662109375,
      "loss_xval": 0.41015625,
      "num_input_tokens_seen": 453862180,
      "step": 8100
    },
    {
      "epoch": 18.042316258351892,
      "grad_norm": 26.7554988861084,
      "learning_rate": 1e-06,
      "loss": 0.5004,
      "num_input_tokens_seen": 453918752,
      "step": 8101
    },
    {
      "epoch": 18.042316258351892,
      "loss": 0.33896124362945557,
      "loss_ce": 9.407360630575567e-05,
      "loss_iou": 0.1484375,
      "loss_num": 0.00830078125,
      "loss_xval": 0.33984375,
      "num_input_tokens_seen": 453918752,
      "step": 8101
    },
    {
      "epoch": 18.044543429844097,
      "grad_norm": 15.414385795593262,
      "learning_rate": 1e-06,
      "loss": 0.4937,
      "num_input_tokens_seen": 453970240,
      "step": 8102
    },
    {
      "epoch": 18.044543429844097,
      "loss": 0.5033870935440063,
      "loss_ce": 9.12140094442293e-05,
      "loss_iou": 0.203125,
      "loss_num": 0.019287109375,
      "loss_xval": 0.50390625,
      "num_input_tokens_seen": 453970240,
      "step": 8102
    },
    {
      "epoch": 18.046770601336302,
      "grad_norm": 19.02543067932129,
      "learning_rate": 1e-06,
      "loss": 0.3916,
      "num_input_tokens_seen": 454026744,
      "step": 8103
    },
    {
      "epoch": 18.046770601336302,
      "loss": 0.46980106830596924,
      "loss_ce": 7.446320523740724e-05,
      "loss_iou": 0.1953125,
      "loss_num": 0.015869140625,
      "loss_xval": 0.46875,
      "num_input_tokens_seen": 454026744,
      "step": 8103
    },
    {
      "epoch": 18.048997772828507,
      "grad_norm": 14.910806655883789,
      "learning_rate": 1e-06,
      "loss": 0.3787,
      "num_input_tokens_seen": 454084836,
      "step": 8104
    },
    {
      "epoch": 18.048997772828507,
      "loss": 0.37930968403816223,
      "loss_ce": 9.826038149185479e-05,
      "loss_iou": 0.17578125,
      "loss_num": 0.005462646484375,
      "loss_xval": 0.37890625,
      "num_input_tokens_seen": 454084836,
      "step": 8104
    },
    {
      "epoch": 18.051224944320712,
      "grad_norm": 15.848285675048828,
      "learning_rate": 1e-06,
      "loss": 0.512,
      "num_input_tokens_seen": 454140536,
      "step": 8105
    },
    {
      "epoch": 18.051224944320712,
      "loss": 0.5733276009559631,
      "loss_ce": 8.542699652025476e-05,
      "loss_iou": 0.251953125,
      "loss_num": 0.01416015625,
      "loss_xval": 0.57421875,
      "num_input_tokens_seen": 454140536,
      "step": 8105
    },
    {
      "epoch": 18.053452115812917,
      "grad_norm": 18.083738327026367,
      "learning_rate": 1e-06,
      "loss": 0.3539,
      "num_input_tokens_seen": 454198948,
      "step": 8106
    },
    {
      "epoch": 18.053452115812917,
      "loss": 0.46540844440460205,
      "loss_ce": 7.641864067409188e-05,
      "loss_iou": 0.2119140625,
      "loss_num": 0.0081787109375,
      "loss_xval": 0.46484375,
      "num_input_tokens_seen": 454198948,
      "step": 8106
    },
    {
      "epoch": 18.05567928730512,
      "grad_norm": 19.562829971313477,
      "learning_rate": 1e-06,
      "loss": 0.3211,
      "num_input_tokens_seen": 454254564,
      "step": 8107
    },
    {
      "epoch": 18.05567928730512,
      "loss": 0.32736125588417053,
      "loss_ce": 9.073851106222719e-05,
      "loss_iou": 0.140625,
      "loss_num": 0.0093994140625,
      "loss_xval": 0.328125,
      "num_input_tokens_seen": 454254564,
      "step": 8107
    },
    {
      "epoch": 18.057906458797326,
      "grad_norm": 28.230247497558594,
      "learning_rate": 1e-06,
      "loss": 0.516,
      "num_input_tokens_seen": 454310376,
      "step": 8108
    },
    {
      "epoch": 18.057906458797326,
      "loss": 0.46714556217193604,
      "loss_ce": 0.00010453614231664687,
      "loss_iou": 0.16015625,
      "loss_num": 0.029296875,
      "loss_xval": 0.466796875,
      "num_input_tokens_seen": 454310376,
      "step": 8108
    },
    {
      "epoch": 18.06013363028953,
      "grad_norm": 12.480415344238281,
      "learning_rate": 1e-06,
      "loss": 0.2964,
      "num_input_tokens_seen": 454367776,
      "step": 8109
    },
    {
      "epoch": 18.06013363028953,
      "loss": 0.356157511472702,
      "loss_ce": 7.839675527065992e-05,
      "loss_iou": 0.14453125,
      "loss_num": 0.01336669921875,
      "loss_xval": 0.35546875,
      "num_input_tokens_seen": 454367776,
      "step": 8109
    },
    {
      "epoch": 18.062360801781736,
      "grad_norm": 22.18619728088379,
      "learning_rate": 1e-06,
      "loss": 0.4995,
      "num_input_tokens_seen": 454423244,
      "step": 8110
    },
    {
      "epoch": 18.062360801781736,
      "loss": 0.5585031509399414,
      "loss_ce": 0.00015356890799012035,
      "loss_iou": 0.232421875,
      "loss_num": 0.0189208984375,
      "loss_xval": 0.55859375,
      "num_input_tokens_seen": 454423244,
      "step": 8110
    },
    {
      "epoch": 18.06458797327394,
      "grad_norm": 25.902963638305664,
      "learning_rate": 1e-06,
      "loss": 0.522,
      "num_input_tokens_seen": 454480660,
      "step": 8111
    },
    {
      "epoch": 18.06458797327394,
      "loss": 0.49017781019210815,
      "loss_ce": 6.550169200636446e-05,
      "loss_iou": 0.216796875,
      "loss_num": 0.01153564453125,
      "loss_xval": 0.490234375,
      "num_input_tokens_seen": 454480660,
      "step": 8111
    },
    {
      "epoch": 18.066815144766146,
      "grad_norm": 27.04472541809082,
      "learning_rate": 1e-06,
      "loss": 0.3989,
      "num_input_tokens_seen": 454539016,
      "step": 8112
    },
    {
      "epoch": 18.066815144766146,
      "loss": 0.26980409026145935,
      "loss_ce": 8.019209781195968e-05,
      "loss_iou": 0.12158203125,
      "loss_num": 0.005340576171875,
      "loss_xval": 0.26953125,
      "num_input_tokens_seen": 454539016,
      "step": 8112
    },
    {
      "epoch": 18.06904231625835,
      "grad_norm": 15.210087776184082,
      "learning_rate": 1e-06,
      "loss": 0.5414,
      "num_input_tokens_seen": 454595484,
      "step": 8113
    },
    {
      "epoch": 18.06904231625835,
      "loss": 0.652846097946167,
      "loss_ce": 7.511470903409645e-05,
      "loss_iou": 0.255859375,
      "loss_num": 0.0284423828125,
      "loss_xval": 0.65234375,
      "num_input_tokens_seen": 454595484,
      "step": 8113
    },
    {
      "epoch": 18.071269487750556,
      "grad_norm": 24.391822814941406,
      "learning_rate": 1e-06,
      "loss": 0.5438,
      "num_input_tokens_seen": 454647108,
      "step": 8114
    },
    {
      "epoch": 18.071269487750556,
      "loss": 0.5746660828590393,
      "loss_ce": 8.115639502648264e-05,
      "loss_iou": 0.271484375,
      "loss_num": 0.006683349609375,
      "loss_xval": 0.57421875,
      "num_input_tokens_seen": 454647108,
      "step": 8114
    },
    {
      "epoch": 18.07349665924276,
      "grad_norm": 13.966802597045898,
      "learning_rate": 1e-06,
      "loss": 0.3106,
      "num_input_tokens_seen": 454704844,
      "step": 8115
    },
    {
      "epoch": 18.07349665924276,
      "loss": 0.31044334173202515,
      "loss_ce": 7.955444743856788e-05,
      "loss_iou": 0.1298828125,
      "loss_num": 0.0103759765625,
      "loss_xval": 0.310546875,
      "num_input_tokens_seen": 454704844,
      "step": 8115
    },
    {
      "epoch": 18.075723830734965,
      "grad_norm": 18.81356430053711,
      "learning_rate": 1e-06,
      "loss": 0.4646,
      "num_input_tokens_seen": 454762604,
      "step": 8116
    },
    {
      "epoch": 18.075723830734965,
      "loss": 0.4306223690509796,
      "loss_ce": 8.036733197513968e-05,
      "loss_iou": 0.1982421875,
      "loss_num": 0.00701904296875,
      "loss_xval": 0.4296875,
      "num_input_tokens_seen": 454762604,
      "step": 8116
    },
    {
      "epoch": 18.07795100222717,
      "grad_norm": 13.788688659667969,
      "learning_rate": 1e-06,
      "loss": 0.249,
      "num_input_tokens_seen": 454818024,
      "step": 8117
    },
    {
      "epoch": 18.07795100222717,
      "loss": 0.23960940539836884,
      "loss_ce": 7.69186153775081e-05,
      "loss_iou": 0.1025390625,
      "loss_num": 0.00689697265625,
      "loss_xval": 0.2392578125,
      "num_input_tokens_seen": 454818024,
      "step": 8117
    },
    {
      "epoch": 18.080178173719375,
      "grad_norm": 14.488852500915527,
      "learning_rate": 1e-06,
      "loss": 0.3516,
      "num_input_tokens_seen": 454873348,
      "step": 8118
    },
    {
      "epoch": 18.080178173719375,
      "loss": 0.40378397703170776,
      "loss_ce": 9.744857379700989e-05,
      "loss_iou": 0.1669921875,
      "loss_num": 0.0137939453125,
      "loss_xval": 0.404296875,
      "num_input_tokens_seen": 454873348,
      "step": 8118
    },
    {
      "epoch": 18.08240534521158,
      "grad_norm": 27.27067756652832,
      "learning_rate": 1e-06,
      "loss": 0.4236,
      "num_input_tokens_seen": 454929640,
      "step": 8119
    },
    {
      "epoch": 18.08240534521158,
      "loss": 0.3224753439426422,
      "loss_ce": 8.763029472902417e-05,
      "loss_iou": 0.1494140625,
      "loss_num": 0.00469970703125,
      "loss_xval": 0.322265625,
      "num_input_tokens_seen": 454929640,
      "step": 8119
    },
    {
      "epoch": 18.084632516703785,
      "grad_norm": 18.79537010192871,
      "learning_rate": 1e-06,
      "loss": 0.4925,
      "num_input_tokens_seen": 454984244,
      "step": 8120
    },
    {
      "epoch": 18.084632516703785,
      "loss": 0.47800394892692566,
      "loss_ce": 0.00015972901019267738,
      "loss_iou": 0.203125,
      "loss_num": 0.0142822265625,
      "loss_xval": 0.478515625,
      "num_input_tokens_seen": 454984244,
      "step": 8120
    },
    {
      "epoch": 18.08685968819599,
      "grad_norm": 15.249739646911621,
      "learning_rate": 1e-06,
      "loss": 0.3372,
      "num_input_tokens_seen": 455041144,
      "step": 8121
    },
    {
      "epoch": 18.08685968819599,
      "loss": 0.2875575125217438,
      "loss_ce": 8.190819789888337e-05,
      "loss_iou": 0.1318359375,
      "loss_num": 0.004669189453125,
      "loss_xval": 0.287109375,
      "num_input_tokens_seen": 455041144,
      "step": 8121
    },
    {
      "epoch": 18.089086859688194,
      "grad_norm": 14.420731544494629,
      "learning_rate": 1e-06,
      "loss": 0.4776,
      "num_input_tokens_seen": 455097144,
      "step": 8122
    },
    {
      "epoch": 18.089086859688194,
      "loss": 0.3784327805042267,
      "loss_ce": 0.00013688384206034243,
      "loss_iou": 0.1572265625,
      "loss_num": 0.01263427734375,
      "loss_xval": 0.37890625,
      "num_input_tokens_seen": 455097144,
      "step": 8122
    },
    {
      "epoch": 18.0913140311804,
      "grad_norm": 19.662689208984375,
      "learning_rate": 1e-06,
      "loss": 0.4817,
      "num_input_tokens_seen": 455151808,
      "step": 8123
    },
    {
      "epoch": 18.0913140311804,
      "loss": 0.567717969417572,
      "loss_ce": 9.098585724132136e-05,
      "loss_iou": 0.244140625,
      "loss_num": 0.0159912109375,
      "loss_xval": 0.56640625,
      "num_input_tokens_seen": 455151808,
      "step": 8123
    },
    {
      "epoch": 18.093541202672604,
      "grad_norm": 19.106225967407227,
      "learning_rate": 1e-06,
      "loss": 0.2999,
      "num_input_tokens_seen": 455207776,
      "step": 8124
    },
    {
      "epoch": 18.093541202672604,
      "loss": 0.33509403467178345,
      "loss_ce": 7.206852023955435e-05,
      "loss_iou": 0.1513671875,
      "loss_num": 0.006591796875,
      "loss_xval": 0.3359375,
      "num_input_tokens_seen": 455207776,
      "step": 8124
    },
    {
      "epoch": 18.09576837416481,
      "grad_norm": 16.473522186279297,
      "learning_rate": 1e-06,
      "loss": 0.4367,
      "num_input_tokens_seen": 455264208,
      "step": 8125
    },
    {
      "epoch": 18.09576837416481,
      "loss": 0.4178709387779236,
      "loss_ce": 8.530144987162203e-05,
      "loss_iou": 0.1728515625,
      "loss_num": 0.014404296875,
      "loss_xval": 0.41796875,
      "num_input_tokens_seen": 455264208,
      "step": 8125
    },
    {
      "epoch": 18.097995545657014,
      "grad_norm": 26.5972900390625,
      "learning_rate": 1e-06,
      "loss": 0.3408,
      "num_input_tokens_seen": 455323352,
      "step": 8126
    },
    {
      "epoch": 18.097995545657014,
      "loss": 0.3901003897190094,
      "loss_ce": 8.57616396388039e-05,
      "loss_iou": 0.1796875,
      "loss_num": 0.00616455078125,
      "loss_xval": 0.390625,
      "num_input_tokens_seen": 455323352,
      "step": 8126
    },
    {
      "epoch": 18.100222717149222,
      "grad_norm": 21.1645565032959,
      "learning_rate": 1e-06,
      "loss": 0.4185,
      "num_input_tokens_seen": 455377804,
      "step": 8127
    },
    {
      "epoch": 18.100222717149222,
      "loss": 0.5210694074630737,
      "loss_ce": 7.333118992391974e-05,
      "loss_iou": 0.2265625,
      "loss_num": 0.01361083984375,
      "loss_xval": 0.51953125,
      "num_input_tokens_seen": 455377804,
      "step": 8127
    },
    {
      "epoch": 18.102449888641427,
      "grad_norm": 13.508529663085938,
      "learning_rate": 1e-06,
      "loss": 0.5152,
      "num_input_tokens_seen": 455436200,
      "step": 8128
    },
    {
      "epoch": 18.102449888641427,
      "loss": 0.6549245119094849,
      "loss_ce": 6.30651629762724e-05,
      "loss_iou": 0.265625,
      "loss_num": 0.0247802734375,
      "loss_xval": 0.65625,
      "num_input_tokens_seen": 455436200,
      "step": 8128
    },
    {
      "epoch": 18.104677060133632,
      "grad_norm": 36.7231330871582,
      "learning_rate": 1e-06,
      "loss": 0.3776,
      "num_input_tokens_seen": 455492220,
      "step": 8129
    },
    {
      "epoch": 18.104677060133632,
      "loss": 0.3046456277370453,
      "loss_ce": 8.02054419182241e-05,
      "loss_iou": 0.1337890625,
      "loss_num": 0.007354736328125,
      "loss_xval": 0.3046875,
      "num_input_tokens_seen": 455492220,
      "step": 8129
    },
    {
      "epoch": 18.106904231625837,
      "grad_norm": 19.406326293945312,
      "learning_rate": 1e-06,
      "loss": 0.4491,
      "num_input_tokens_seen": 455548928,
      "step": 8130
    },
    {
      "epoch": 18.106904231625837,
      "loss": 0.3191646337509155,
      "loss_ce": 7.285462925210595e-05,
      "loss_iou": 0.1455078125,
      "loss_num": 0.0057373046875,
      "loss_xval": 0.318359375,
      "num_input_tokens_seen": 455548928,
      "step": 8130
    },
    {
      "epoch": 18.10913140311804,
      "grad_norm": 12.066944122314453,
      "learning_rate": 1e-06,
      "loss": 0.4965,
      "num_input_tokens_seen": 455605732,
      "step": 8131
    },
    {
      "epoch": 18.10913140311804,
      "loss": 0.49812638759613037,
      "loss_ce": 7.951979932840914e-05,
      "loss_iou": 0.1962890625,
      "loss_num": 0.02099609375,
      "loss_xval": 0.498046875,
      "num_input_tokens_seen": 455605732,
      "step": 8131
    },
    {
      "epoch": 18.111358574610247,
      "grad_norm": 12.560493469238281,
      "learning_rate": 1e-06,
      "loss": 0.3447,
      "num_input_tokens_seen": 455660540,
      "step": 8132
    },
    {
      "epoch": 18.111358574610247,
      "loss": 0.4943099915981293,
      "loss_ce": 0.00010833313717739657,
      "loss_iou": 0.212890625,
      "loss_num": 0.01361083984375,
      "loss_xval": 0.494140625,
      "num_input_tokens_seen": 455660540,
      "step": 8132
    },
    {
      "epoch": 18.11358574610245,
      "grad_norm": 23.26740264892578,
      "learning_rate": 1e-06,
      "loss": 0.4059,
      "num_input_tokens_seen": 455712784,
      "step": 8133
    },
    {
      "epoch": 18.11358574610245,
      "loss": 0.533281683921814,
      "loss_ce": 7.859165634727105e-05,
      "loss_iou": 0.234375,
      "loss_num": 0.01275634765625,
      "loss_xval": 0.53125,
      "num_input_tokens_seen": 455712784,
      "step": 8133
    },
    {
      "epoch": 18.115812917594656,
      "grad_norm": 23.393814086914062,
      "learning_rate": 1e-06,
      "loss": 0.3933,
      "num_input_tokens_seen": 455769952,
      "step": 8134
    },
    {
      "epoch": 18.115812917594656,
      "loss": 0.4792192578315735,
      "loss_ce": 9.330162720289081e-05,
      "loss_iou": 0.2216796875,
      "loss_num": 0.007080078125,
      "loss_xval": 0.478515625,
      "num_input_tokens_seen": 455769952,
      "step": 8134
    },
    {
      "epoch": 18.11804008908686,
      "grad_norm": 23.41754913330078,
      "learning_rate": 1e-06,
      "loss": 0.3536,
      "num_input_tokens_seen": 455826364,
      "step": 8135
    },
    {
      "epoch": 18.11804008908686,
      "loss": 0.39287176728248596,
      "loss_ce": 7.998933142516762e-05,
      "loss_iou": 0.1728515625,
      "loss_num": 0.00933837890625,
      "loss_xval": 0.392578125,
      "num_input_tokens_seen": 455826364,
      "step": 8135
    },
    {
      "epoch": 18.120267260579066,
      "grad_norm": 18.5450382232666,
      "learning_rate": 1e-06,
      "loss": 0.5557,
      "num_input_tokens_seen": 455882520,
      "step": 8136
    },
    {
      "epoch": 18.120267260579066,
      "loss": 0.4983789920806885,
      "loss_ce": 8.797197369858623e-05,
      "loss_iou": 0.2021484375,
      "loss_num": 0.0189208984375,
      "loss_xval": 0.498046875,
      "num_input_tokens_seen": 455882520,
      "step": 8136
    },
    {
      "epoch": 18.12249443207127,
      "grad_norm": 21.666677474975586,
      "learning_rate": 1e-06,
      "loss": 0.6017,
      "num_input_tokens_seen": 455937676,
      "step": 8137
    },
    {
      "epoch": 18.12249443207127,
      "loss": 0.7193117141723633,
      "loss_ce": 7.343379547819495e-05,
      "loss_iou": 0.31640625,
      "loss_num": 0.017333984375,
      "loss_xval": 0.71875,
      "num_input_tokens_seen": 455937676,
      "step": 8137
    },
    {
      "epoch": 18.124721603563476,
      "grad_norm": 16.64803695678711,
      "learning_rate": 1e-06,
      "loss": 0.2924,
      "num_input_tokens_seen": 455993536,
      "step": 8138
    },
    {
      "epoch": 18.124721603563476,
      "loss": 0.39221036434173584,
      "loss_ce": 0.00018156672012992203,
      "loss_iou": 0.1533203125,
      "loss_num": 0.016845703125,
      "loss_xval": 0.392578125,
      "num_input_tokens_seen": 455993536,
      "step": 8138
    },
    {
      "epoch": 18.12694877505568,
      "grad_norm": 19.38474464416504,
      "learning_rate": 1e-06,
      "loss": 0.3963,
      "num_input_tokens_seen": 456050968,
      "step": 8139
    },
    {
      "epoch": 18.12694877505568,
      "loss": 0.3352212905883789,
      "loss_ce": 7.726570765953511e-05,
      "loss_iou": 0.1474609375,
      "loss_num": 0.0079345703125,
      "loss_xval": 0.3359375,
      "num_input_tokens_seen": 456050968,
      "step": 8139
    },
    {
      "epoch": 18.129175946547885,
      "grad_norm": 19.141530990600586,
      "learning_rate": 1e-06,
      "loss": 0.3805,
      "num_input_tokens_seen": 456106612,
      "step": 8140
    },
    {
      "epoch": 18.129175946547885,
      "loss": 0.3776403069496155,
      "loss_ce": 7.682391151320189e-05,
      "loss_iou": 0.16796875,
      "loss_num": 0.0084228515625,
      "loss_xval": 0.376953125,
      "num_input_tokens_seen": 456106612,
      "step": 8140
    },
    {
      "epoch": 18.13140311804009,
      "grad_norm": 26.25833511352539,
      "learning_rate": 1e-06,
      "loss": 0.4945,
      "num_input_tokens_seen": 456162008,
      "step": 8141
    },
    {
      "epoch": 18.13140311804009,
      "loss": 0.3847208023071289,
      "loss_ce": 7.726027979515493e-05,
      "loss_iou": 0.177734375,
      "loss_num": 0.005889892578125,
      "loss_xval": 0.384765625,
      "num_input_tokens_seen": 456162008,
      "step": 8141
    },
    {
      "epoch": 18.133630289532295,
      "grad_norm": 16.26892852783203,
      "learning_rate": 1e-06,
      "loss": 0.3777,
      "num_input_tokens_seen": 456218592,
      "step": 8142
    },
    {
      "epoch": 18.133630289532295,
      "loss": 0.29285314679145813,
      "loss_ce": 9.801473061088473e-05,
      "loss_iou": 0.1142578125,
      "loss_num": 0.0128173828125,
      "loss_xval": 0.29296875,
      "num_input_tokens_seen": 456218592,
      "step": 8142
    },
    {
      "epoch": 18.1358574610245,
      "grad_norm": 13.948265075683594,
      "learning_rate": 1e-06,
      "loss": 0.3755,
      "num_input_tokens_seen": 456276108,
      "step": 8143
    },
    {
      "epoch": 18.1358574610245,
      "loss": 0.2600770592689514,
      "loss_ce": 6.72862006467767e-05,
      "loss_iou": 0.1181640625,
      "loss_num": 0.0047607421875,
      "loss_xval": 0.259765625,
      "num_input_tokens_seen": 456276108,
      "step": 8143
    },
    {
      "epoch": 18.138084632516705,
      "grad_norm": 14.333640098571777,
      "learning_rate": 1e-06,
      "loss": 0.3341,
      "num_input_tokens_seen": 456331636,
      "step": 8144
    },
    {
      "epoch": 18.138084632516705,
      "loss": 0.37764212489128113,
      "loss_ce": 0.0008721151389181614,
      "loss_iou": 0.169921875,
      "loss_num": 0.00738525390625,
      "loss_xval": 0.376953125,
      "num_input_tokens_seen": 456331636,
      "step": 8144
    },
    {
      "epoch": 18.14031180400891,
      "grad_norm": 16.1193904876709,
      "learning_rate": 1e-06,
      "loss": 0.2635,
      "num_input_tokens_seen": 456388988,
      "step": 8145
    },
    {
      "epoch": 18.14031180400891,
      "loss": 0.30690258741378784,
      "loss_ce": 7.886123057687655e-05,
      "loss_iou": 0.1318359375,
      "loss_num": 0.0084228515625,
      "loss_xval": 0.306640625,
      "num_input_tokens_seen": 456388988,
      "step": 8145
    },
    {
      "epoch": 18.142538975501115,
      "grad_norm": 16.353914260864258,
      "learning_rate": 1e-06,
      "loss": 0.371,
      "num_input_tokens_seen": 456447512,
      "step": 8146
    },
    {
      "epoch": 18.142538975501115,
      "loss": 0.4683547616004944,
      "loss_ce": 9.304599370807409e-05,
      "loss_iou": 0.19140625,
      "loss_num": 0.0169677734375,
      "loss_xval": 0.46875,
      "num_input_tokens_seen": 456447512,
      "step": 8146
    },
    {
      "epoch": 18.14476614699332,
      "grad_norm": 21.104829788208008,
      "learning_rate": 1e-06,
      "loss": 0.3263,
      "num_input_tokens_seen": 456506516,
      "step": 8147
    },
    {
      "epoch": 18.14476614699332,
      "loss": 0.24555513262748718,
      "loss_ce": 7.17366419848986e-05,
      "loss_iou": 0.111328125,
      "loss_num": 0.00457763671875,
      "loss_xval": 0.2451171875,
      "num_input_tokens_seen": 456506516,
      "step": 8147
    },
    {
      "epoch": 18.146993318485524,
      "grad_norm": 16.034555435180664,
      "learning_rate": 1e-06,
      "loss": 0.4458,
      "num_input_tokens_seen": 456563372,
      "step": 8148
    },
    {
      "epoch": 18.146993318485524,
      "loss": 0.2350102663040161,
      "loss_ce": 8.594193059252575e-05,
      "loss_iou": 0.10888671875,
      "loss_num": 0.00335693359375,
      "loss_xval": 0.2353515625,
      "num_input_tokens_seen": 456563372,
      "step": 8148
    },
    {
      "epoch": 18.14922048997773,
      "grad_norm": 21.92203140258789,
      "learning_rate": 1e-06,
      "loss": 0.5644,
      "num_input_tokens_seen": 456620288,
      "step": 8149
    },
    {
      "epoch": 18.14922048997773,
      "loss": 0.2608814239501953,
      "loss_ce": 7.819505117367953e-05,
      "loss_iou": 0.12353515625,
      "loss_num": 0.0027313232421875,
      "loss_xval": 0.26171875,
      "num_input_tokens_seen": 456620288,
      "step": 8149
    },
    {
      "epoch": 18.151447661469934,
      "grad_norm": 14.489221572875977,
      "learning_rate": 1e-06,
      "loss": 0.306,
      "num_input_tokens_seen": 456677872,
      "step": 8150
    },
    {
      "epoch": 18.151447661469934,
      "loss": 0.31026989221572876,
      "loss_ce": 8.923574932850897e-05,
      "loss_iou": 0.1376953125,
      "loss_num": 0.00677490234375,
      "loss_xval": 0.310546875,
      "num_input_tokens_seen": 456677872,
      "step": 8150
    },
    {
      "epoch": 18.15367483296214,
      "grad_norm": 15.693221092224121,
      "learning_rate": 1e-06,
      "loss": 0.3977,
      "num_input_tokens_seen": 456730576,
      "step": 8151
    },
    {
      "epoch": 18.15367483296214,
      "loss": 0.35370901226997375,
      "loss_ce": 7.132487371563911e-05,
      "loss_iou": 0.15234375,
      "loss_num": 0.00970458984375,
      "loss_xval": 0.353515625,
      "num_input_tokens_seen": 456730576,
      "step": 8151
    },
    {
      "epoch": 18.155902004454344,
      "grad_norm": 21.944501876831055,
      "learning_rate": 1e-06,
      "loss": 0.4647,
      "num_input_tokens_seen": 456787084,
      "step": 8152
    },
    {
      "epoch": 18.155902004454344,
      "loss": 0.48086851835250854,
      "loss_ce": 9.459961438551545e-05,
      "loss_iou": 0.1904296875,
      "loss_num": 0.0201416015625,
      "loss_xval": 0.48046875,
      "num_input_tokens_seen": 456787084,
      "step": 8152
    },
    {
      "epoch": 18.15812917594655,
      "grad_norm": 17.893390655517578,
      "learning_rate": 1e-06,
      "loss": 0.409,
      "num_input_tokens_seen": 456845316,
      "step": 8153
    },
    {
      "epoch": 18.15812917594655,
      "loss": 0.4042748212814331,
      "loss_ce": 0.00010002141789300367,
      "loss_iou": 0.1875,
      "loss_num": 0.005706787109375,
      "loss_xval": 0.404296875,
      "num_input_tokens_seen": 456845316,
      "step": 8153
    },
    {
      "epoch": 18.160356347438753,
      "grad_norm": 18.064180374145508,
      "learning_rate": 1e-06,
      "loss": 0.3617,
      "num_input_tokens_seen": 456902888,
      "step": 8154
    },
    {
      "epoch": 18.160356347438753,
      "loss": 0.41487082839012146,
      "loss_ce": 7.58875539759174e-05,
      "loss_iou": 0.1669921875,
      "loss_num": 0.016357421875,
      "loss_xval": 0.4140625,
      "num_input_tokens_seen": 456902888,
      "step": 8154
    },
    {
      "epoch": 18.16258351893096,
      "grad_norm": 16.817781448364258,
      "learning_rate": 1e-06,
      "loss": 0.4589,
      "num_input_tokens_seen": 456955400,
      "step": 8155
    },
    {
      "epoch": 18.16258351893096,
      "loss": 0.702592134475708,
      "loss_ce": 0.00013854095595888793,
      "loss_iou": 0.287109375,
      "loss_num": 0.025390625,
      "loss_xval": 0.703125,
      "num_input_tokens_seen": 456955400,
      "step": 8155
    },
    {
      "epoch": 18.164810690423163,
      "grad_norm": 16.054969787597656,
      "learning_rate": 1e-06,
      "loss": 0.3455,
      "num_input_tokens_seen": 457011064,
      "step": 8156
    },
    {
      "epoch": 18.164810690423163,
      "loss": 0.3856889605522156,
      "loss_ce": 6.883072637720034e-05,
      "loss_iou": 0.1728515625,
      "loss_num": 0.008056640625,
      "loss_xval": 0.384765625,
      "num_input_tokens_seen": 457011064,
      "step": 8156
    },
    {
      "epoch": 18.167037861915368,
      "grad_norm": 16.399335861206055,
      "learning_rate": 1e-06,
      "loss": 0.3965,
      "num_input_tokens_seen": 457064464,
      "step": 8157
    },
    {
      "epoch": 18.167037861915368,
      "loss": 0.4999640882015228,
      "loss_ce": 8.612703823018819e-05,
      "loss_iou": 0.2060546875,
      "loss_num": 0.0177001953125,
      "loss_xval": 0.5,
      "num_input_tokens_seen": 457064464,
      "step": 8157
    },
    {
      "epoch": 18.169265033407573,
      "grad_norm": 26.273801803588867,
      "learning_rate": 1e-06,
      "loss": 0.2832,
      "num_input_tokens_seen": 457120308,
      "step": 8158
    },
    {
      "epoch": 18.169265033407573,
      "loss": 0.3040543496608734,
      "loss_ce": 9.926770871970803e-05,
      "loss_iou": 0.1396484375,
      "loss_num": 0.0048828125,
      "loss_xval": 0.3046875,
      "num_input_tokens_seen": 457120308,
      "step": 8158
    },
    {
      "epoch": 18.171492204899778,
      "grad_norm": 21.627939224243164,
      "learning_rate": 1e-06,
      "loss": 0.5024,
      "num_input_tokens_seen": 457175336,
      "step": 8159
    },
    {
      "epoch": 18.171492204899778,
      "loss": 0.5851208567619324,
      "loss_ce": 9.887100895866752e-05,
      "loss_iou": 0.267578125,
      "loss_num": 0.01031494140625,
      "loss_xval": 0.5859375,
      "num_input_tokens_seen": 457175336,
      "step": 8159
    },
    {
      "epoch": 18.173719376391983,
      "grad_norm": 36.85986328125,
      "learning_rate": 1e-06,
      "loss": 0.5207,
      "num_input_tokens_seen": 457232364,
      "step": 8160
    },
    {
      "epoch": 18.173719376391983,
      "loss": 0.4390491247177124,
      "loss_ce": 8.427293505519629e-05,
      "loss_iou": 0.197265625,
      "loss_num": 0.0086669921875,
      "loss_xval": 0.439453125,
      "num_input_tokens_seen": 457232364,
      "step": 8160
    },
    {
      "epoch": 18.175946547884188,
      "grad_norm": 27.42588996887207,
      "learning_rate": 1e-06,
      "loss": 0.4134,
      "num_input_tokens_seen": 457287852,
      "step": 8161
    },
    {
      "epoch": 18.175946547884188,
      "loss": 0.3699635863304138,
      "loss_ce": 9.052493987837806e-05,
      "loss_iou": 0.1533203125,
      "loss_num": 0.01251220703125,
      "loss_xval": 0.369140625,
      "num_input_tokens_seen": 457287852,
      "step": 8161
    },
    {
      "epoch": 18.178173719376392,
      "grad_norm": 19.563459396362305,
      "learning_rate": 1e-06,
      "loss": 0.4402,
      "num_input_tokens_seen": 457342640,
      "step": 8162
    },
    {
      "epoch": 18.178173719376392,
      "loss": 0.5288839936256409,
      "loss_ce": 7.537108467658982e-05,
      "loss_iou": 0.2216796875,
      "loss_num": 0.01708984375,
      "loss_xval": 0.52734375,
      "num_input_tokens_seen": 457342640,
      "step": 8162
    },
    {
      "epoch": 18.180400890868597,
      "grad_norm": 17.342567443847656,
      "learning_rate": 1e-06,
      "loss": 0.3396,
      "num_input_tokens_seen": 457396200,
      "step": 8163
    },
    {
      "epoch": 18.180400890868597,
      "loss": 0.36558669805526733,
      "loss_ce": 7.767054921714589e-05,
      "loss_iou": 0.1689453125,
      "loss_num": 0.005340576171875,
      "loss_xval": 0.365234375,
      "num_input_tokens_seen": 457396200,
      "step": 8163
    },
    {
      "epoch": 18.182628062360802,
      "grad_norm": 21.52007293701172,
      "learning_rate": 1e-06,
      "loss": 0.3781,
      "num_input_tokens_seen": 457453456,
      "step": 8164
    },
    {
      "epoch": 18.182628062360802,
      "loss": 0.2819899618625641,
      "loss_ce": 6.859673885628581e-05,
      "loss_iou": 0.11767578125,
      "loss_num": 0.00933837890625,
      "loss_xval": 0.28125,
      "num_input_tokens_seen": 457453456,
      "step": 8164
    },
    {
      "epoch": 18.184855233853007,
      "grad_norm": 23.580726623535156,
      "learning_rate": 1e-06,
      "loss": 0.4272,
      "num_input_tokens_seen": 457510128,
      "step": 8165
    },
    {
      "epoch": 18.184855233853007,
      "loss": 0.5174942016601562,
      "loss_ce": 0.00016021478222683072,
      "loss_iou": 0.2353515625,
      "loss_num": 0.00921630859375,
      "loss_xval": 0.515625,
      "num_input_tokens_seen": 457510128,
      "step": 8165
    },
    {
      "epoch": 18.187082405345212,
      "grad_norm": 17.68349266052246,
      "learning_rate": 1e-06,
      "loss": 0.3081,
      "num_input_tokens_seen": 457566892,
      "step": 8166
    },
    {
      "epoch": 18.187082405345212,
      "loss": 0.4296377897262573,
      "loss_ce": 7.235370139824226e-05,
      "loss_iou": 0.173828125,
      "loss_num": 0.016357421875,
      "loss_xval": 0.4296875,
      "num_input_tokens_seen": 457566892,
      "step": 8166
    },
    {
      "epoch": 18.189309576837417,
      "grad_norm": 15.33013916015625,
      "learning_rate": 1e-06,
      "loss": 0.3858,
      "num_input_tokens_seen": 457624908,
      "step": 8167
    },
    {
      "epoch": 18.189309576837417,
      "loss": 0.34201037883758545,
      "loss_ce": 9.140933980233967e-05,
      "loss_iou": 0.154296875,
      "loss_num": 0.00677490234375,
      "loss_xval": 0.341796875,
      "num_input_tokens_seen": 457624908,
      "step": 8167
    },
    {
      "epoch": 18.19153674832962,
      "grad_norm": 25.988941192626953,
      "learning_rate": 1e-06,
      "loss": 0.2713,
      "num_input_tokens_seen": 457681264,
      "step": 8168
    },
    {
      "epoch": 18.19153674832962,
      "loss": 0.2663070559501648,
      "loss_ce": 7.169348828028888e-05,
      "loss_iou": 0.1220703125,
      "loss_num": 0.004364013671875,
      "loss_xval": 0.265625,
      "num_input_tokens_seen": 457681264,
      "step": 8168
    },
    {
      "epoch": 18.193763919821826,
      "grad_norm": 17.212726593017578,
      "learning_rate": 1e-06,
      "loss": 0.3666,
      "num_input_tokens_seen": 457737384,
      "step": 8169
    },
    {
      "epoch": 18.193763919821826,
      "loss": 0.3459562063217163,
      "loss_ce": 8.524451550329104e-05,
      "loss_iou": 0.142578125,
      "loss_num": 0.01214599609375,
      "loss_xval": 0.345703125,
      "num_input_tokens_seen": 457737384,
      "step": 8169
    },
    {
      "epoch": 18.19599109131403,
      "grad_norm": 42.238746643066406,
      "learning_rate": 1e-06,
      "loss": 0.4758,
      "num_input_tokens_seen": 457792120,
      "step": 8170
    },
    {
      "epoch": 18.19599109131403,
      "loss": 0.571884036064148,
      "loss_ce": 0.00010671824566088617,
      "loss_iou": 0.2578125,
      "loss_num": 0.0115966796875,
      "loss_xval": 0.5703125,
      "num_input_tokens_seen": 457792120,
      "step": 8170
    },
    {
      "epoch": 18.198218262806236,
      "grad_norm": 18.594070434570312,
      "learning_rate": 1e-06,
      "loss": 0.4922,
      "num_input_tokens_seen": 457846880,
      "step": 8171
    },
    {
      "epoch": 18.198218262806236,
      "loss": 0.5872501134872437,
      "loss_ce": 9.18953082873486e-05,
      "loss_iou": 0.263671875,
      "loss_num": 0.01165771484375,
      "loss_xval": 0.5859375,
      "num_input_tokens_seen": 457846880,
      "step": 8171
    },
    {
      "epoch": 18.20044543429844,
      "grad_norm": 16.824237823486328,
      "learning_rate": 1e-06,
      "loss": 0.5248,
      "num_input_tokens_seen": 457901312,
      "step": 8172
    },
    {
      "epoch": 18.20044543429844,
      "loss": 0.3980518877506256,
      "loss_ce": 0.00010268213372910395,
      "loss_iou": 0.1796875,
      "loss_num": 0.0079345703125,
      "loss_xval": 0.3984375,
      "num_input_tokens_seen": 457901312,
      "step": 8172
    },
    {
      "epoch": 18.202672605790646,
      "grad_norm": 14.367027282714844,
      "learning_rate": 1e-06,
      "loss": 0.5356,
      "num_input_tokens_seen": 457956652,
      "step": 8173
    },
    {
      "epoch": 18.202672605790646,
      "loss": 0.544747531414032,
      "loss_ce": 6.98043149895966e-05,
      "loss_iou": 0.2119140625,
      "loss_num": 0.0242919921875,
      "loss_xval": 0.54296875,
      "num_input_tokens_seen": 457956652,
      "step": 8173
    },
    {
      "epoch": 18.20489977728285,
      "grad_norm": 16.34149742126465,
      "learning_rate": 1e-06,
      "loss": 0.4287,
      "num_input_tokens_seen": 458012632,
      "step": 8174
    },
    {
      "epoch": 18.20489977728285,
      "loss": 0.5932860374450684,
      "loss_ce": 0.0001463656226405874,
      "loss_iou": 0.240234375,
      "loss_num": 0.022705078125,
      "loss_xval": 0.59375,
      "num_input_tokens_seen": 458012632,
      "step": 8174
    },
    {
      "epoch": 18.207126948775056,
      "grad_norm": 14.004996299743652,
      "learning_rate": 1e-06,
      "loss": 0.3419,
      "num_input_tokens_seen": 458068888,
      "step": 8175
    },
    {
      "epoch": 18.207126948775056,
      "loss": 0.31001418828964233,
      "loss_ce": 7.767122588120401e-05,
      "loss_iou": 0.134765625,
      "loss_num": 0.008056640625,
      "loss_xval": 0.310546875,
      "num_input_tokens_seen": 458068888,
      "step": 8175
    },
    {
      "epoch": 18.20935412026726,
      "grad_norm": 15.622814178466797,
      "learning_rate": 1e-06,
      "loss": 0.7315,
      "num_input_tokens_seen": 458126280,
      "step": 8176
    },
    {
      "epoch": 18.20935412026726,
      "loss": 0.942706286907196,
      "loss_ce": 7.930257561383769e-05,
      "loss_iou": 0.39453125,
      "loss_num": 0.0302734375,
      "loss_xval": 0.94140625,
      "num_input_tokens_seen": 458126280,
      "step": 8176
    },
    {
      "epoch": 18.211581291759465,
      "grad_norm": 19.05216407775879,
      "learning_rate": 1e-06,
      "loss": 0.4613,
      "num_input_tokens_seen": 458184116,
      "step": 8177
    },
    {
      "epoch": 18.211581291759465,
      "loss": 0.4942248463630676,
      "loss_ce": 8.419141522608697e-05,
      "loss_iou": 0.224609375,
      "loss_num": 0.009033203125,
      "loss_xval": 0.494140625,
      "num_input_tokens_seen": 458184116,
      "step": 8177
    },
    {
      "epoch": 18.21380846325167,
      "grad_norm": 14.203112602233887,
      "learning_rate": 1e-06,
      "loss": 0.2897,
      "num_input_tokens_seen": 458239480,
      "step": 8178
    },
    {
      "epoch": 18.21380846325167,
      "loss": 0.18030065298080444,
      "loss_ce": 7.909360283520073e-05,
      "loss_iou": 0.06982421875,
      "loss_num": 0.0081787109375,
      "loss_xval": 0.1806640625,
      "num_input_tokens_seen": 458239480,
      "step": 8178
    },
    {
      "epoch": 18.216035634743875,
      "grad_norm": 18.163633346557617,
      "learning_rate": 1e-06,
      "loss": 0.3865,
      "num_input_tokens_seen": 458296128,
      "step": 8179
    },
    {
      "epoch": 18.216035634743875,
      "loss": 0.39704078435897827,
      "loss_ce": 6.810402555856854e-05,
      "loss_iou": 0.173828125,
      "loss_num": 0.00994873046875,
      "loss_xval": 0.396484375,
      "num_input_tokens_seen": 458296128,
      "step": 8179
    },
    {
      "epoch": 18.21826280623608,
      "grad_norm": 11.550317764282227,
      "learning_rate": 1e-06,
      "loss": 0.2529,
      "num_input_tokens_seen": 458351960,
      "step": 8180
    },
    {
      "epoch": 18.21826280623608,
      "loss": 0.25586456060409546,
      "loss_ce": 6.619263149332255e-05,
      "loss_iou": 0.1142578125,
      "loss_num": 0.005523681640625,
      "loss_xval": 0.255859375,
      "num_input_tokens_seen": 458351960,
      "step": 8180
    },
    {
      "epoch": 18.220489977728285,
      "grad_norm": 22.90071678161621,
      "learning_rate": 1e-06,
      "loss": 0.2964,
      "num_input_tokens_seen": 458407868,
      "step": 8181
    },
    {
      "epoch": 18.220489977728285,
      "loss": 0.27848535776138306,
      "loss_ce": 7.347905193455517e-05,
      "loss_iou": 0.1259765625,
      "loss_num": 0.005279541015625,
      "loss_xval": 0.279296875,
      "num_input_tokens_seen": 458407868,
      "step": 8181
    },
    {
      "epoch": 18.22271714922049,
      "grad_norm": 19.092378616333008,
      "learning_rate": 1e-06,
      "loss": 0.2401,
      "num_input_tokens_seen": 458464468,
      "step": 8182
    },
    {
      "epoch": 18.22271714922049,
      "loss": 0.25201308727264404,
      "loss_ce": 5.996804611640982e-05,
      "loss_iou": 0.11181640625,
      "loss_num": 0.00567626953125,
      "loss_xval": 0.251953125,
      "num_input_tokens_seen": 458464468,
      "step": 8182
    },
    {
      "epoch": 18.224944320712694,
      "grad_norm": 21.963666915893555,
      "learning_rate": 1e-06,
      "loss": 0.4337,
      "num_input_tokens_seen": 458520888,
      "step": 8183
    },
    {
      "epoch": 18.224944320712694,
      "loss": 0.37996935844421387,
      "loss_ce": 8.6537329480052e-05,
      "loss_iou": 0.158203125,
      "loss_num": 0.01251220703125,
      "loss_xval": 0.37890625,
      "num_input_tokens_seen": 458520888,
      "step": 8183
    },
    {
      "epoch": 18.2271714922049,
      "grad_norm": 25.58934211730957,
      "learning_rate": 1e-06,
      "loss": 0.5241,
      "num_input_tokens_seen": 458575140,
      "step": 8184
    },
    {
      "epoch": 18.2271714922049,
      "loss": 0.3485790491104126,
      "loss_ce": 6.830115307820961e-05,
      "loss_iou": 0.1416015625,
      "loss_num": 0.01324462890625,
      "loss_xval": 0.34765625,
      "num_input_tokens_seen": 458575140,
      "step": 8184
    },
    {
      "epoch": 18.229398663697104,
      "grad_norm": 12.794227600097656,
      "learning_rate": 1e-06,
      "loss": 0.3202,
      "num_input_tokens_seen": 458631632,
      "step": 8185
    },
    {
      "epoch": 18.229398663697104,
      "loss": 0.21777865290641785,
      "loss_ce": 6.625376408919692e-05,
      "loss_iou": 0.07861328125,
      "loss_num": 0.0120849609375,
      "loss_xval": 0.2177734375,
      "num_input_tokens_seen": 458631632,
      "step": 8185
    },
    {
      "epoch": 18.23162583518931,
      "grad_norm": 18.660388946533203,
      "learning_rate": 1e-06,
      "loss": 0.4079,
      "num_input_tokens_seen": 458688408,
      "step": 8186
    },
    {
      "epoch": 18.23162583518931,
      "loss": 0.34663307666778564,
      "loss_ce": 7.544427353423089e-05,
      "loss_iou": 0.1513671875,
      "loss_num": 0.008544921875,
      "loss_xval": 0.345703125,
      "num_input_tokens_seen": 458688408,
      "step": 8186
    },
    {
      "epoch": 18.233853006681514,
      "grad_norm": 23.186193466186523,
      "learning_rate": 1e-06,
      "loss": 0.5892,
      "num_input_tokens_seen": 458743956,
      "step": 8187
    },
    {
      "epoch": 18.233853006681514,
      "loss": 0.42682188749313354,
      "loss_ce": 6.407788896467537e-05,
      "loss_iou": 0.193359375,
      "loss_num": 0.0081787109375,
      "loss_xval": 0.42578125,
      "num_input_tokens_seen": 458743956,
      "step": 8187
    },
    {
      "epoch": 18.23608017817372,
      "grad_norm": 21.697927474975586,
      "learning_rate": 1e-06,
      "loss": 0.3607,
      "num_input_tokens_seen": 458798376,
      "step": 8188
    },
    {
      "epoch": 18.23608017817372,
      "loss": 0.3312641382217407,
      "loss_ce": 8.736809832043946e-05,
      "loss_iou": 0.1435546875,
      "loss_num": 0.0087890625,
      "loss_xval": 0.33203125,
      "num_input_tokens_seen": 458798376,
      "step": 8188
    },
    {
      "epoch": 18.238307349665924,
      "grad_norm": 14.134407997131348,
      "learning_rate": 1e-06,
      "loss": 0.4647,
      "num_input_tokens_seen": 458852484,
      "step": 8189
    },
    {
      "epoch": 18.238307349665924,
      "loss": 0.35396501421928406,
      "loss_ce": 8.317639003507793e-05,
      "loss_iou": 0.1630859375,
      "loss_num": 0.00567626953125,
      "loss_xval": 0.353515625,
      "num_input_tokens_seen": 458852484,
      "step": 8189
    },
    {
      "epoch": 18.24053452115813,
      "grad_norm": 38.25514602661133,
      "learning_rate": 1e-06,
      "loss": 0.613,
      "num_input_tokens_seen": 458906996,
      "step": 8190
    },
    {
      "epoch": 18.24053452115813,
      "loss": 0.5307399034500122,
      "loss_ce": 0.00022236474615056068,
      "loss_iou": 0.23828125,
      "loss_num": 0.0108642578125,
      "loss_xval": 0.53125,
      "num_input_tokens_seen": 458906996,
      "step": 8190
    },
    {
      "epoch": 18.242761692650333,
      "grad_norm": 15.433220863342285,
      "learning_rate": 1e-06,
      "loss": 0.3704,
      "num_input_tokens_seen": 458964332,
      "step": 8191
    },
    {
      "epoch": 18.242761692650333,
      "loss": 0.35981857776641846,
      "loss_ce": 7.736931729596108e-05,
      "loss_iou": 0.162109375,
      "loss_num": 0.0072021484375,
      "loss_xval": 0.359375,
      "num_input_tokens_seen": 458964332,
      "step": 8191
    },
    {
      "epoch": 18.244988864142538,
      "grad_norm": 20.43923568725586,
      "learning_rate": 1e-06,
      "loss": 0.3444,
      "num_input_tokens_seen": 459017136,
      "step": 8192
    },
    {
      "epoch": 18.244988864142538,
      "loss": 0.4209337830543518,
      "loss_ce": 9.635718015488237e-05,
      "loss_iou": 0.189453125,
      "loss_num": 0.00823974609375,
      "loss_xval": 0.419921875,
      "num_input_tokens_seen": 459017136,
      "step": 8192
    },
    {
      "epoch": 18.247216035634743,
      "grad_norm": 17.763324737548828,
      "learning_rate": 1e-06,
      "loss": 0.373,
      "num_input_tokens_seen": 459070944,
      "step": 8193
    },
    {
      "epoch": 18.247216035634743,
      "loss": 0.2321740686893463,
      "loss_ce": 8.788384002400562e-05,
      "loss_iou": 0.10400390625,
      "loss_num": 0.004730224609375,
      "loss_xval": 0.232421875,
      "num_input_tokens_seen": 459070944,
      "step": 8193
    },
    {
      "epoch": 18.249443207126948,
      "grad_norm": 25.71845245361328,
      "learning_rate": 1e-06,
      "loss": 0.2999,
      "num_input_tokens_seen": 459126108,
      "step": 8194
    },
    {
      "epoch": 18.249443207126948,
      "loss": 0.31814372539520264,
      "loss_ce": 8.950403571361676e-05,
      "loss_iou": 0.1435546875,
      "loss_num": 0.00634765625,
      "loss_xval": 0.318359375,
      "num_input_tokens_seen": 459126108,
      "step": 8194
    },
    {
      "epoch": 18.251670378619153,
      "grad_norm": 36.80335235595703,
      "learning_rate": 1e-06,
      "loss": 0.5035,
      "num_input_tokens_seen": 459183928,
      "step": 8195
    },
    {
      "epoch": 18.251670378619153,
      "loss": 0.5821275115013123,
      "loss_ce": 9.623746882425621e-05,
      "loss_iou": 0.2578125,
      "loss_num": 0.01324462890625,
      "loss_xval": 0.58203125,
      "num_input_tokens_seen": 459183928,
      "step": 8195
    },
    {
      "epoch": 18.253897550111358,
      "grad_norm": 19.333406448364258,
      "learning_rate": 1e-06,
      "loss": 0.3609,
      "num_input_tokens_seen": 459241312,
      "step": 8196
    },
    {
      "epoch": 18.253897550111358,
      "loss": 0.4230514466762543,
      "loss_ce": 7.78458925196901e-05,
      "loss_iou": 0.1806640625,
      "loss_num": 0.0123291015625,
      "loss_xval": 0.423828125,
      "num_input_tokens_seen": 459241312,
      "step": 8196
    },
    {
      "epoch": 18.256124721603562,
      "grad_norm": 22.73419189453125,
      "learning_rate": 1e-06,
      "loss": 0.5667,
      "num_input_tokens_seen": 459298740,
      "step": 8197
    },
    {
      "epoch": 18.256124721603562,
      "loss": 0.6623133420944214,
      "loss_ce": 8.191996312234551e-05,
      "loss_iou": 0.306640625,
      "loss_num": 0.0096435546875,
      "loss_xval": 0.6640625,
      "num_input_tokens_seen": 459298740,
      "step": 8197
    },
    {
      "epoch": 18.258351893095767,
      "grad_norm": 19.6635684967041,
      "learning_rate": 1e-06,
      "loss": 0.35,
      "num_input_tokens_seen": 459354172,
      "step": 8198
    },
    {
      "epoch": 18.258351893095767,
      "loss": 0.2642420530319214,
      "loss_ce": 8.189349318854511e-05,
      "loss_iou": 0.1220703125,
      "loss_num": 0.004058837890625,
      "loss_xval": 0.263671875,
      "num_input_tokens_seen": 459354172,
      "step": 8198
    },
    {
      "epoch": 18.260579064587972,
      "grad_norm": 14.02831745147705,
      "learning_rate": 1e-06,
      "loss": 0.3704,
      "num_input_tokens_seen": 459409124,
      "step": 8199
    },
    {
      "epoch": 18.260579064587972,
      "loss": 0.37897345423698425,
      "loss_ce": 6.719774683006108e-05,
      "loss_iou": 0.1728515625,
      "loss_num": 0.006561279296875,
      "loss_xval": 0.37890625,
      "num_input_tokens_seen": 459409124,
      "step": 8199
    },
    {
      "epoch": 18.262806236080177,
      "grad_norm": 21.54979133605957,
      "learning_rate": 1e-06,
      "loss": 0.469,
      "num_input_tokens_seen": 459463080,
      "step": 8200
    },
    {
      "epoch": 18.262806236080177,
      "loss": 0.46509498357772827,
      "loss_ce": 6.810689228586853e-05,
      "loss_iou": 0.18359375,
      "loss_num": 0.01953125,
      "loss_xval": 0.46484375,
      "num_input_tokens_seen": 459463080,
      "step": 8200
    },
    {
      "epoch": 18.265033407572382,
      "grad_norm": 15.60781192779541,
      "learning_rate": 1e-06,
      "loss": 0.3833,
      "num_input_tokens_seen": 459518640,
      "step": 8201
    },
    {
      "epoch": 18.265033407572382,
      "loss": 0.31758350133895874,
      "loss_ce": 7.863431528676301e-05,
      "loss_iou": 0.1376953125,
      "loss_num": 0.008544921875,
      "loss_xval": 0.318359375,
      "num_input_tokens_seen": 459518640,
      "step": 8201
    },
    {
      "epoch": 18.267260579064587,
      "grad_norm": 25.999467849731445,
      "learning_rate": 1e-06,
      "loss": 0.4274,
      "num_input_tokens_seen": 459574496,
      "step": 8202
    },
    {
      "epoch": 18.267260579064587,
      "loss": 0.3373531103134155,
      "loss_ce": 7.284604362212121e-05,
      "loss_iou": 0.150390625,
      "loss_num": 0.0072021484375,
      "loss_xval": 0.337890625,
      "num_input_tokens_seen": 459574496,
      "step": 8202
    },
    {
      "epoch": 18.26948775055679,
      "grad_norm": 23.961406707763672,
      "learning_rate": 1e-06,
      "loss": 0.2495,
      "num_input_tokens_seen": 459631000,
      "step": 8203
    },
    {
      "epoch": 18.26948775055679,
      "loss": 0.2549530863761902,
      "loss_ce": 7.027012179605663e-05,
      "loss_iou": 0.1162109375,
      "loss_num": 0.004547119140625,
      "loss_xval": 0.25390625,
      "num_input_tokens_seen": 459631000,
      "step": 8203
    },
    {
      "epoch": 18.271714922048996,
      "grad_norm": 13.854894638061523,
      "learning_rate": 1e-06,
      "loss": 0.4288,
      "num_input_tokens_seen": 459689184,
      "step": 8204
    },
    {
      "epoch": 18.271714922048996,
      "loss": 0.32967978715896606,
      "loss_ce": 8.993731171358377e-05,
      "loss_iou": 0.150390625,
      "loss_num": 0.005767822265625,
      "loss_xval": 0.330078125,
      "num_input_tokens_seen": 459689184,
      "step": 8204
    },
    {
      "epoch": 18.2739420935412,
      "grad_norm": 15.130828857421875,
      "learning_rate": 1e-06,
      "loss": 0.331,
      "num_input_tokens_seen": 459744348,
      "step": 8205
    },
    {
      "epoch": 18.2739420935412,
      "loss": 0.3028114438056946,
      "loss_ce": 0.0002601708984002471,
      "loss_iou": 0.1357421875,
      "loss_num": 0.00628662109375,
      "loss_xval": 0.302734375,
      "num_input_tokens_seen": 459744348,
      "step": 8205
    },
    {
      "epoch": 18.276169265033406,
      "grad_norm": 22.3679256439209,
      "learning_rate": 1e-06,
      "loss": 0.4245,
      "num_input_tokens_seen": 459800880,
      "step": 8206
    },
    {
      "epoch": 18.276169265033406,
      "loss": 0.3924823999404907,
      "loss_ce": 8.740788325667381e-05,
      "loss_iou": 0.173828125,
      "loss_num": 0.00921630859375,
      "loss_xval": 0.392578125,
      "num_input_tokens_seen": 459800880,
      "step": 8206
    },
    {
      "epoch": 18.27839643652561,
      "grad_norm": 18.131832122802734,
      "learning_rate": 1e-06,
      "loss": 0.4461,
      "num_input_tokens_seen": 459856516,
      "step": 8207
    },
    {
      "epoch": 18.27839643652561,
      "loss": 0.4801885783672333,
      "loss_ce": 8.601776062278077e-05,
      "loss_iou": 0.189453125,
      "loss_num": 0.0203857421875,
      "loss_xval": 0.48046875,
      "num_input_tokens_seen": 459856516,
      "step": 8207
    },
    {
      "epoch": 18.280623608017816,
      "grad_norm": 17.009628295898438,
      "learning_rate": 1e-06,
      "loss": 0.4747,
      "num_input_tokens_seen": 459912960,
      "step": 8208
    },
    {
      "epoch": 18.280623608017816,
      "loss": 0.41185104846954346,
      "loss_ce": 0.00010787278006318957,
      "loss_iou": 0.181640625,
      "loss_num": 0.0096435546875,
      "loss_xval": 0.412109375,
      "num_input_tokens_seen": 459912960,
      "step": 8208
    },
    {
      "epoch": 18.28285077951002,
      "grad_norm": 16.9517765045166,
      "learning_rate": 1e-06,
      "loss": 0.4285,
      "num_input_tokens_seen": 459971620,
      "step": 8209
    },
    {
      "epoch": 18.28285077951002,
      "loss": 0.43293631076812744,
      "loss_ce": 7.499181083403528e-05,
      "loss_iou": 0.181640625,
      "loss_num": 0.01409912109375,
      "loss_xval": 0.43359375,
      "num_input_tokens_seen": 459971620,
      "step": 8209
    },
    {
      "epoch": 18.285077951002226,
      "grad_norm": 19.251338958740234,
      "learning_rate": 1e-06,
      "loss": 0.3578,
      "num_input_tokens_seen": 460028300,
      "step": 8210
    },
    {
      "epoch": 18.285077951002226,
      "loss": 0.38544806838035583,
      "loss_ce": 7.210190233308822e-05,
      "loss_iou": 0.162109375,
      "loss_num": 0.01214599609375,
      "loss_xval": 0.384765625,
      "num_input_tokens_seen": 460028300,
      "step": 8210
    },
    {
      "epoch": 18.28730512249443,
      "grad_norm": 18.597429275512695,
      "learning_rate": 1e-06,
      "loss": 0.3693,
      "num_input_tokens_seen": 460084228,
      "step": 8211
    },
    {
      "epoch": 18.28730512249443,
      "loss": 0.2983230948448181,
      "loss_ce": 0.00022741041902918369,
      "loss_iou": 0.12890625,
      "loss_num": 0.00811767578125,
      "loss_xval": 0.298828125,
      "num_input_tokens_seen": 460084228,
      "step": 8211
    },
    {
      "epoch": 18.289532293986635,
      "grad_norm": 17.22545623779297,
      "learning_rate": 1e-06,
      "loss": 0.264,
      "num_input_tokens_seen": 460139016,
      "step": 8212
    },
    {
      "epoch": 18.289532293986635,
      "loss": 0.2883983552455902,
      "loss_ce": 6.828906043665484e-05,
      "loss_iou": 0.1298828125,
      "loss_num": 0.005767822265625,
      "loss_xval": 0.2890625,
      "num_input_tokens_seen": 460139016,
      "step": 8212
    },
    {
      "epoch": 18.29175946547884,
      "grad_norm": 34.40812301635742,
      "learning_rate": 1e-06,
      "loss": 0.4567,
      "num_input_tokens_seen": 460194692,
      "step": 8213
    },
    {
      "epoch": 18.29175946547884,
      "loss": 0.3748340904712677,
      "loss_ce": 7.823290070518851e-05,
      "loss_iou": 0.1650390625,
      "loss_num": 0.00909423828125,
      "loss_xval": 0.375,
      "num_input_tokens_seen": 460194692,
      "step": 8213
    },
    {
      "epoch": 18.293986636971045,
      "grad_norm": 21.1818904876709,
      "learning_rate": 1e-06,
      "loss": 0.51,
      "num_input_tokens_seen": 460251096,
      "step": 8214
    },
    {
      "epoch": 18.293986636971045,
      "loss": 0.6246699690818787,
      "loss_ce": 0.00015822870773263276,
      "loss_iou": 0.25390625,
      "loss_num": 0.0234375,
      "loss_xval": 0.625,
      "num_input_tokens_seen": 460251096,
      "step": 8214
    },
    {
      "epoch": 18.29621380846325,
      "grad_norm": 20.41522216796875,
      "learning_rate": 1e-06,
      "loss": 0.4014,
      "num_input_tokens_seen": 460309636,
      "step": 8215
    },
    {
      "epoch": 18.29621380846325,
      "loss": 0.46027785539627075,
      "loss_ce": 7.279778947122395e-05,
      "loss_iou": 0.2119140625,
      "loss_num": 0.007171630859375,
      "loss_xval": 0.4609375,
      "num_input_tokens_seen": 460309636,
      "step": 8215
    },
    {
      "epoch": 18.29844097995546,
      "grad_norm": 17.60694694519043,
      "learning_rate": 1e-06,
      "loss": 0.3979,
      "num_input_tokens_seen": 460365224,
      "step": 8216
    },
    {
      "epoch": 18.29844097995546,
      "loss": 0.3157610297203064,
      "loss_ce": 8.721975609660149e-05,
      "loss_iou": 0.125,
      "loss_num": 0.012939453125,
      "loss_xval": 0.31640625,
      "num_input_tokens_seen": 460365224,
      "step": 8216
    },
    {
      "epoch": 18.30066815144766,
      "grad_norm": 27.434608459472656,
      "learning_rate": 1e-06,
      "loss": 0.4209,
      "num_input_tokens_seen": 460423160,
      "step": 8217
    },
    {
      "epoch": 18.30066815144766,
      "loss": 0.5255805850028992,
      "loss_ce": 6.787220627302304e-05,
      "loss_iou": 0.23828125,
      "loss_num": 0.00982666015625,
      "loss_xval": 0.52734375,
      "num_input_tokens_seen": 460423160,
      "step": 8217
    },
    {
      "epoch": 18.302895322939868,
      "grad_norm": 15.852324485778809,
      "learning_rate": 1e-06,
      "loss": 0.3734,
      "num_input_tokens_seen": 460480832,
      "step": 8218
    },
    {
      "epoch": 18.302895322939868,
      "loss": 0.3597148060798645,
      "loss_ce": 9.567091183271259e-05,
      "loss_iou": 0.16015625,
      "loss_num": 0.00787353515625,
      "loss_xval": 0.359375,
      "num_input_tokens_seen": 460480832,
      "step": 8218
    },
    {
      "epoch": 18.305122494432073,
      "grad_norm": 16.711715698242188,
      "learning_rate": 1e-06,
      "loss": 0.5244,
      "num_input_tokens_seen": 460534584,
      "step": 8219
    },
    {
      "epoch": 18.305122494432073,
      "loss": 0.605729341506958,
      "loss_ce": 7.750350050628185e-05,
      "loss_iou": 0.2275390625,
      "loss_num": 0.0301513671875,
      "loss_xval": 0.60546875,
      "num_input_tokens_seen": 460534584,
      "step": 8219
    },
    {
      "epoch": 18.307349665924278,
      "grad_norm": 24.398696899414062,
      "learning_rate": 1e-06,
      "loss": 0.4038,
      "num_input_tokens_seen": 460591852,
      "step": 8220
    },
    {
      "epoch": 18.307349665924278,
      "loss": 0.3758271336555481,
      "loss_ce": 9.471758676227182e-05,
      "loss_iou": 0.1591796875,
      "loss_num": 0.0113525390625,
      "loss_xval": 0.375,
      "num_input_tokens_seen": 460591852,
      "step": 8220
    },
    {
      "epoch": 18.309576837416483,
      "grad_norm": 15.600603103637695,
      "learning_rate": 1e-06,
      "loss": 0.4869,
      "num_input_tokens_seen": 460650120,
      "step": 8221
    },
    {
      "epoch": 18.309576837416483,
      "loss": 0.6405990123748779,
      "loss_ce": 9.607061656424776e-05,
      "loss_iou": 0.287109375,
      "loss_num": 0.01348876953125,
      "loss_xval": 0.640625,
      "num_input_tokens_seen": 460650120,
      "step": 8221
    },
    {
      "epoch": 18.311804008908688,
      "grad_norm": 14.608988761901855,
      "learning_rate": 1e-06,
      "loss": 0.3083,
      "num_input_tokens_seen": 460708680,
      "step": 8222
    },
    {
      "epoch": 18.311804008908688,
      "loss": 0.31593137979507446,
      "loss_ce": 7.447078678524122e-05,
      "loss_iou": 0.12890625,
      "loss_num": 0.01153564453125,
      "loss_xval": 0.31640625,
      "num_input_tokens_seen": 460708680,
      "step": 8222
    },
    {
      "epoch": 18.314031180400892,
      "grad_norm": 12.612223625183105,
      "learning_rate": 1e-06,
      "loss": 0.431,
      "num_input_tokens_seen": 460762612,
      "step": 8223
    },
    {
      "epoch": 18.314031180400892,
      "loss": 0.5143847465515137,
      "loss_ce": 0.00010250776540488005,
      "loss_iou": 0.216796875,
      "loss_num": 0.015869140625,
      "loss_xval": 0.515625,
      "num_input_tokens_seen": 460762612,
      "step": 8223
    },
    {
      "epoch": 18.316258351893097,
      "grad_norm": 19.262920379638672,
      "learning_rate": 1e-06,
      "loss": 0.5283,
      "num_input_tokens_seen": 460814596,
      "step": 8224
    },
    {
      "epoch": 18.316258351893097,
      "loss": 0.6533957123756409,
      "loss_ce": 7.540466322097927e-05,
      "loss_iou": 0.28125,
      "loss_num": 0.0185546875,
      "loss_xval": 0.65234375,
      "num_input_tokens_seen": 460814596,
      "step": 8224
    },
    {
      "epoch": 18.318485523385302,
      "grad_norm": 17.134294509887695,
      "learning_rate": 1e-06,
      "loss": 0.3269,
      "num_input_tokens_seen": 460871852,
      "step": 8225
    },
    {
      "epoch": 18.318485523385302,
      "loss": 0.2810301184654236,
      "loss_ce": 8.527821046300232e-05,
      "loss_iou": 0.1318359375,
      "loss_num": 0.003570556640625,
      "loss_xval": 0.28125,
      "num_input_tokens_seen": 460871852,
      "step": 8225
    },
    {
      "epoch": 18.320712694877507,
      "grad_norm": 16.73043441772461,
      "learning_rate": 1e-06,
      "loss": 0.4727,
      "num_input_tokens_seen": 460928820,
      "step": 8226
    },
    {
      "epoch": 18.320712694877507,
      "loss": 0.4601028263568878,
      "loss_ce": 8.087064634310082e-05,
      "loss_iou": 0.177734375,
      "loss_num": 0.0208740234375,
      "loss_xval": 0.4609375,
      "num_input_tokens_seen": 460928820,
      "step": 8226
    },
    {
      "epoch": 18.322939866369712,
      "grad_norm": 13.59985065460205,
      "learning_rate": 1e-06,
      "loss": 0.4533,
      "num_input_tokens_seen": 460985356,
      "step": 8227
    },
    {
      "epoch": 18.322939866369712,
      "loss": 0.43281298875808716,
      "loss_ce": 0.00019579757645260543,
      "loss_iou": 0.201171875,
      "loss_num": 0.006256103515625,
      "loss_xval": 0.43359375,
      "num_input_tokens_seen": 460985356,
      "step": 8227
    },
    {
      "epoch": 18.325167037861917,
      "grad_norm": 12.674905776977539,
      "learning_rate": 1e-06,
      "loss": 0.3026,
      "num_input_tokens_seen": 461042168,
      "step": 8228
    },
    {
      "epoch": 18.325167037861917,
      "loss": 0.3006775975227356,
      "loss_ce": 7.942065712995827e-05,
      "loss_iou": 0.1298828125,
      "loss_num": 0.00811767578125,
      "loss_xval": 0.30078125,
      "num_input_tokens_seen": 461042168,
      "step": 8228
    },
    {
      "epoch": 18.32739420935412,
      "grad_norm": 17.03211784362793,
      "learning_rate": 1e-06,
      "loss": 0.4486,
      "num_input_tokens_seen": 461097504,
      "step": 8229
    },
    {
      "epoch": 18.32739420935412,
      "loss": 0.50324547290802,
      "loss_ce": 7.163839472923428e-05,
      "loss_iou": 0.21875,
      "loss_num": 0.01324462890625,
      "loss_xval": 0.50390625,
      "num_input_tokens_seen": 461097504,
      "step": 8229
    },
    {
      "epoch": 18.329621380846326,
      "grad_norm": 12.527713775634766,
      "learning_rate": 1e-06,
      "loss": 0.3258,
      "num_input_tokens_seen": 461152312,
      "step": 8230
    },
    {
      "epoch": 18.329621380846326,
      "loss": 0.2726093530654907,
      "loss_ce": 7.59223330533132e-05,
      "loss_iou": 0.10693359375,
      "loss_num": 0.01177978515625,
      "loss_xval": 0.2734375,
      "num_input_tokens_seen": 461152312,
      "step": 8230
    },
    {
      "epoch": 18.33184855233853,
      "grad_norm": 17.93193244934082,
      "learning_rate": 1e-06,
      "loss": 0.3163,
      "num_input_tokens_seen": 461207068,
      "step": 8231
    },
    {
      "epoch": 18.33184855233853,
      "loss": 0.2764345407485962,
      "loss_ce": 6.737266085110605e-05,
      "loss_iou": 0.1123046875,
      "loss_num": 0.0103759765625,
      "loss_xval": 0.27734375,
      "num_input_tokens_seen": 461207068,
      "step": 8231
    },
    {
      "epoch": 18.334075723830736,
      "grad_norm": 15.471948623657227,
      "learning_rate": 1e-06,
      "loss": 0.3574,
      "num_input_tokens_seen": 461262992,
      "step": 8232
    },
    {
      "epoch": 18.334075723830736,
      "loss": 0.38583964109420776,
      "loss_ce": 6.69407527311705e-05,
      "loss_iou": 0.1748046875,
      "loss_num": 0.007293701171875,
      "loss_xval": 0.38671875,
      "num_input_tokens_seen": 461262992,
      "step": 8232
    },
    {
      "epoch": 18.33630289532294,
      "grad_norm": 19.41752815246582,
      "learning_rate": 1e-06,
      "loss": 0.2833,
      "num_input_tokens_seen": 461317216,
      "step": 8233
    },
    {
      "epoch": 18.33630289532294,
      "loss": 0.2918124198913574,
      "loss_ce": 6.43762614345178e-05,
      "loss_iou": 0.125,
      "loss_num": 0.0081787109375,
      "loss_xval": 0.291015625,
      "num_input_tokens_seen": 461317216,
      "step": 8233
    },
    {
      "epoch": 18.338530066815146,
      "grad_norm": 21.109203338623047,
      "learning_rate": 1e-06,
      "loss": 0.5718,
      "num_input_tokens_seen": 461372448,
      "step": 8234
    },
    {
      "epoch": 18.338530066815146,
      "loss": 0.7163882851600647,
      "loss_ce": 7.966908742673695e-05,
      "loss_iou": 0.330078125,
      "loss_num": 0.01116943359375,
      "loss_xval": 0.71484375,
      "num_input_tokens_seen": 461372448,
      "step": 8234
    },
    {
      "epoch": 18.34075723830735,
      "grad_norm": 11.767685890197754,
      "learning_rate": 1e-06,
      "loss": 0.3292,
      "num_input_tokens_seen": 461426004,
      "step": 8235
    },
    {
      "epoch": 18.34075723830735,
      "loss": 0.2593403160572052,
      "loss_ce": 0.0002003002300625667,
      "loss_iou": 0.11328125,
      "loss_num": 0.0064697265625,
      "loss_xval": 0.259765625,
      "num_input_tokens_seen": 461426004,
      "step": 8235
    },
    {
      "epoch": 18.342984409799556,
      "grad_norm": 17.519113540649414,
      "learning_rate": 1e-06,
      "loss": 0.3103,
      "num_input_tokens_seen": 461482340,
      "step": 8236
    },
    {
      "epoch": 18.342984409799556,
      "loss": 0.20156915485858917,
      "loss_ce": 9.21003520488739e-05,
      "loss_iou": 0.091796875,
      "loss_num": 0.003631591796875,
      "loss_xval": 0.201171875,
      "num_input_tokens_seen": 461482340,
      "step": 8236
    },
    {
      "epoch": 18.34521158129176,
      "grad_norm": 33.767704010009766,
      "learning_rate": 1e-06,
      "loss": 0.3042,
      "num_input_tokens_seen": 461537340,
      "step": 8237
    },
    {
      "epoch": 18.34521158129176,
      "loss": 0.31417620182037354,
      "loss_ce": 8.927624730858952e-05,
      "loss_iou": 0.1376953125,
      "loss_num": 0.007720947265625,
      "loss_xval": 0.314453125,
      "num_input_tokens_seen": 461537340,
      "step": 8237
    },
    {
      "epoch": 18.347438752783965,
      "grad_norm": 12.72602653503418,
      "learning_rate": 1e-06,
      "loss": 0.2971,
      "num_input_tokens_seen": 461593824,
      "step": 8238
    },
    {
      "epoch": 18.347438752783965,
      "loss": 0.3771233558654785,
      "loss_ce": 7.865474617574364e-05,
      "loss_iou": 0.1513671875,
      "loss_num": 0.014892578125,
      "loss_xval": 0.376953125,
      "num_input_tokens_seen": 461593824,
      "step": 8238
    },
    {
      "epoch": 18.34966592427617,
      "grad_norm": 19.39862632751465,
      "learning_rate": 1e-06,
      "loss": 0.3585,
      "num_input_tokens_seen": 461648568,
      "step": 8239
    },
    {
      "epoch": 18.34966592427617,
      "loss": 0.33659249544143677,
      "loss_ce": 7.516345067415386e-05,
      "loss_iou": 0.1484375,
      "loss_num": 0.007720947265625,
      "loss_xval": 0.3359375,
      "num_input_tokens_seen": 461648568,
      "step": 8239
    },
    {
      "epoch": 18.351893095768375,
      "grad_norm": 19.25311851501465,
      "learning_rate": 1e-06,
      "loss": 0.6112,
      "num_input_tokens_seen": 461705024,
      "step": 8240
    },
    {
      "epoch": 18.351893095768375,
      "loss": 0.6639153361320496,
      "loss_ce": 9.700231021270156e-05,
      "loss_iou": 0.279296875,
      "loss_num": 0.0211181640625,
      "loss_xval": 0.6640625,
      "num_input_tokens_seen": 461705024,
      "step": 8240
    },
    {
      "epoch": 18.35412026726058,
      "grad_norm": 13.526534080505371,
      "learning_rate": 1e-06,
      "loss": 0.2696,
      "num_input_tokens_seen": 461762520,
      "step": 8241
    },
    {
      "epoch": 18.35412026726058,
      "loss": 0.23994842171669006,
      "loss_ce": 8.790172432782128e-05,
      "loss_iou": 0.109375,
      "loss_num": 0.004119873046875,
      "loss_xval": 0.240234375,
      "num_input_tokens_seen": 461762520,
      "step": 8241
    },
    {
      "epoch": 18.356347438752785,
      "grad_norm": 16.30290412902832,
      "learning_rate": 1e-06,
      "loss": 0.3445,
      "num_input_tokens_seen": 461820448,
      "step": 8242
    },
    {
      "epoch": 18.356347438752785,
      "loss": 0.34260737895965576,
      "loss_ce": 7.808158989064395e-05,
      "loss_iou": 0.1572265625,
      "loss_num": 0.0057373046875,
      "loss_xval": 0.341796875,
      "num_input_tokens_seen": 461820448,
      "step": 8242
    },
    {
      "epoch": 18.35857461024499,
      "grad_norm": 16.620193481445312,
      "learning_rate": 1e-06,
      "loss": 0.2799,
      "num_input_tokens_seen": 461876008,
      "step": 8243
    },
    {
      "epoch": 18.35857461024499,
      "loss": 0.2714713215827942,
      "loss_ce": 0.00010899978951783851,
      "loss_iou": 0.0947265625,
      "loss_num": 0.016357421875,
      "loss_xval": 0.271484375,
      "num_input_tokens_seen": 461876008,
      "step": 8243
    },
    {
      "epoch": 18.360801781737194,
      "grad_norm": 15.673192977905273,
      "learning_rate": 1e-06,
      "loss": 0.4081,
      "num_input_tokens_seen": 461932492,
      "step": 8244
    },
    {
      "epoch": 18.360801781737194,
      "loss": 0.37256646156311035,
      "loss_ce": 6.890152872074395e-05,
      "loss_iou": 0.15625,
      "loss_num": 0.01220703125,
      "loss_xval": 0.373046875,
      "num_input_tokens_seen": 461932492,
      "step": 8244
    },
    {
      "epoch": 18.3630289532294,
      "grad_norm": 15.485980987548828,
      "learning_rate": 1e-06,
      "loss": 0.4505,
      "num_input_tokens_seen": 461989756,
      "step": 8245
    },
    {
      "epoch": 18.3630289532294,
      "loss": 0.4861689805984497,
      "loss_ce": 8.499999967170879e-05,
      "loss_iou": 0.2041015625,
      "loss_num": 0.015625,
      "loss_xval": 0.486328125,
      "num_input_tokens_seen": 461989756,
      "step": 8245
    },
    {
      "epoch": 18.365256124721604,
      "grad_norm": 31.88662338256836,
      "learning_rate": 1e-06,
      "loss": 0.5343,
      "num_input_tokens_seen": 462045504,
      "step": 8246
    },
    {
      "epoch": 18.365256124721604,
      "loss": 0.636799156665802,
      "loss_ce": 8.0388635979034e-05,
      "loss_iou": 0.255859375,
      "loss_num": 0.024658203125,
      "loss_xval": 0.63671875,
      "num_input_tokens_seen": 462045504,
      "step": 8246
    },
    {
      "epoch": 18.36748329621381,
      "grad_norm": 17.961257934570312,
      "learning_rate": 1e-06,
      "loss": 0.3355,
      "num_input_tokens_seen": 462101768,
      "step": 8247
    },
    {
      "epoch": 18.36748329621381,
      "loss": 0.29511559009552,
      "loss_ce": 7.165400893427432e-05,
      "loss_iou": 0.11767578125,
      "loss_num": 0.011962890625,
      "loss_xval": 0.294921875,
      "num_input_tokens_seen": 462101768,
      "step": 8247
    },
    {
      "epoch": 18.369710467706014,
      "grad_norm": 17.18083381652832,
      "learning_rate": 1e-06,
      "loss": 0.5334,
      "num_input_tokens_seen": 462158008,
      "step": 8248
    },
    {
      "epoch": 18.369710467706014,
      "loss": 0.5379419922828674,
      "loss_ce": 0.00010018746252171695,
      "loss_iou": 0.234375,
      "loss_num": 0.01361083984375,
      "loss_xval": 0.5390625,
      "num_input_tokens_seen": 462158008,
      "step": 8248
    },
    {
      "epoch": 18.37193763919822,
      "grad_norm": 20.73891258239746,
      "learning_rate": 1e-06,
      "loss": 0.3165,
      "num_input_tokens_seen": 462213512,
      "step": 8249
    },
    {
      "epoch": 18.37193763919822,
      "loss": 0.3529818058013916,
      "loss_ce": 7.652836939087138e-05,
      "loss_iou": 0.154296875,
      "loss_num": 0.009033203125,
      "loss_xval": 0.353515625,
      "num_input_tokens_seen": 462213512,
      "step": 8249
    },
    {
      "epoch": 18.374164810690424,
      "grad_norm": 18.682357788085938,
      "learning_rate": 1e-06,
      "loss": 0.3058,
      "num_input_tokens_seen": 462271092,
      "step": 8250
    },
    {
      "epoch": 18.374164810690424,
      "eval_seeclick_web_CIoU": 0.5856335759162903,
      "eval_seeclick_web_GIoU": 0.5846323072910309,
      "eval_seeclick_web_IoU": 0.604880303144455,
      "eval_seeclick_web_MAE_all": 0.015254544327035546,
      "eval_seeclick_web_MAE_h": 0.0071105193346738815,
      "eval_seeclick_web_MAE_w": 0.015218731015920639,
      "eval_seeclick_web_MAE_x_boxes": 0.008255395339801908,
      "eval_seeclick_web_MAE_y_boxes": 0.02131559420377016,
      "eval_seeclick_web_inside_bbox": 0.9010416567325592,
      "eval_seeclick_web_loss": 0.9020078778266907,
      "eval_seeclick_web_loss_ce": 0.00012946181232109666,
      "eval_seeclick_web_loss_iou": 0.41748046875,
      "eval_seeclick_web_loss_num": 0.012150764465332031,
      "eval_seeclick_web_loss_xval": 0.896484375,
      "eval_seeclick_web_runtime": 24.3274,
      "eval_seeclick_web_samples_per_second": 2.055,
      "eval_seeclick_web_steps_per_second": 0.082,
      "num_input_tokens_seen": 462271092,
      "step": 8250
    },
    {
      "epoch": 18.374164810690424,
      "eval_icons_CIoU": 0.2648079916834831,
      "eval_icons_GIoU": 0.2893849313259125,
      "eval_icons_IoU": 0.341605544090271,
      "eval_icons_MAE_all": 0.06054504215717316,
      "eval_icons_MAE_h": 0.03238860424607992,
      "eval_icons_MAE_w": 0.06985406205058098,
      "eval_icons_MAE_x_boxes": 0.05330132879316807,
      "eval_icons_MAE_y_boxes": 0.03792322427034378,
      "eval_icons_inside_bbox": 0.59375,
      "eval_icons_loss": 1.7429778575897217,
      "eval_icons_loss_ce": 0.00015769631136208773,
      "eval_icons_loss_iou": 0.6812744140625,
      "eval_icons_loss_num": 0.059600830078125,
      "eval_icons_loss_xval": 1.66064453125,
      "eval_icons_runtime": 23.8883,
      "eval_icons_samples_per_second": 2.093,
      "eval_icons_steps_per_second": 0.084,
      "num_input_tokens_seen": 462271092,
      "step": 8250
    },
    {
      "epoch": 18.374164810690424,
      "eval_screenspot_CIoU": 0.3744619091351827,
      "eval_screenspot_GIoU": 0.391864409049352,
      "eval_screenspot_IoU": 0.4460846583048503,
      "eval_screenspot_MAE_all": 0.056168291717767715,
      "eval_screenspot_MAE_h": 0.03892662810782591,
      "eval_screenspot_MAE_w": 0.0636625016729037,
      "eval_screenspot_MAE_x_boxes": 0.06623268065353234,
      "eval_screenspot_MAE_y_boxes": 0.03843462746590376,
      "eval_screenspot_inside_bbox": 0.7145833373069763,
      "eval_screenspot_loss": 1.5575822591781616,
      "eval_screenspot_loss_ce": 0.00017815509151356915,
      "eval_screenspot_loss_iou": 0.6473795572916666,
      "eval_screenspot_loss_num": 0.06395212809244792,
      "eval_screenspot_loss_xval": 1.6150716145833333,
      "eval_screenspot_runtime": 40.5304,
      "eval_screenspot_samples_per_second": 2.196,
      "eval_screenspot_steps_per_second": 0.074,
      "num_input_tokens_seen": 462271092,
      "step": 8250
    },
    {
      "epoch": 18.374164810690424,
      "eval_compot_CIoU": 0.3456965386867523,
      "eval_compot_GIoU": 0.3530429005622864,
      "eval_compot_IoU": 0.4042937010526657,
      "eval_compot_MAE_all": 0.017928121611475945,
      "eval_compot_MAE_h": 0.008343497524037957,
      "eval_compot_MAE_w": 0.020461782813072205,
      "eval_compot_MAE_x_boxes": 0.030165866017341614,
      "eval_compot_MAE_y_boxes": 0.006972244009375572,
      "eval_compot_inside_bbox": 0.6302083432674408,
      "eval_compot_loss": 1.4054356813430786,
      "eval_compot_loss_ce": 0.00012335985229583457,
      "eval_compot_loss_iou": 0.6524658203125,
      "eval_compot_loss_num": 0.016485214233398438,
      "eval_compot_loss_xval": 1.385986328125,
      "eval_compot_runtime": 25.2618,
      "eval_compot_samples_per_second": 1.979,
      "eval_compot_steps_per_second": 0.079,
      "num_input_tokens_seen": 462271092,
      "step": 8250
    },
    {
      "epoch": 18.374164810690424,
      "eval_custom_ui_val_CIoU": 0.4731468657652537,
      "eval_custom_ui_val_GIoU": 0.4772413820028305,
      "eval_custom_ui_val_IoU": 0.5357410940859053,
      "eval_custom_ui_val_MAE_all": 0.02722639176580641,
      "eval_custom_ui_val_MAE_h": 0.013777732486940093,
      "eval_custom_ui_val_MAE_w": 0.03672494749642081,
      "eval_custom_ui_val_MAE_x_boxes": 0.03293217242591911,
      "eval_custom_ui_val_MAE_y_boxes": 0.013356222900458507,
      "eval_custom_ui_val_inside_bbox": 0.7719907429483202,
      "eval_custom_ui_val_loss": 1.1706430912017822,
      "eval_custom_ui_val_loss_ce": 0.0001427585027866169,
      "eval_custom_ui_val_loss_iou": 0.501953125,
      "eval_custom_ui_val_loss_num": 0.0238491694132487,
      "eval_custom_ui_val_loss_xval": 1.1229383680555556,
      "eval_custom_ui_val_runtime": 76.2841,
      "eval_custom_ui_val_samples_per_second": 3.474,
      "eval_custom_ui_val_steps_per_second": 0.118,
      "num_input_tokens_seen": 462271092,
      "step": 8250
    },
    {
      "epoch": 18.374164810690424,
      "loss": 0.8289626836776733,
      "loss_ce": 0.00010525943071115762,
      "loss_iou": 0.373046875,
      "loss_num": 0.0166015625,
      "loss_xval": 0.828125,
      "num_input_tokens_seen": 462271092,
      "step": 8250
    },
    {
      "epoch": 18.37639198218263,
      "grad_norm": 33.0701789855957,
      "learning_rate": 1e-06,
      "loss": 0.4336,
      "num_input_tokens_seen": 462326668,
      "step": 8251
    },
    {
      "epoch": 18.37639198218263,
      "loss": 0.485931396484375,
      "loss_ce": 9.15604323381558e-05,
      "loss_iou": 0.228515625,
      "loss_num": 0.0057373046875,
      "loss_xval": 0.486328125,
      "num_input_tokens_seen": 462326668,
      "step": 8251
    },
    {
      "epoch": 18.378619153674833,
      "grad_norm": 18.959814071655273,
      "learning_rate": 1e-06,
      "loss": 0.3617,
      "num_input_tokens_seen": 462381308,
      "step": 8252
    },
    {
      "epoch": 18.378619153674833,
      "loss": 0.35047417879104614,
      "loss_ce": 7.13385088602081e-05,
      "loss_iou": 0.1669921875,
      "loss_num": 0.003143310546875,
      "loss_xval": 0.349609375,
      "num_input_tokens_seen": 462381308,
      "step": 8252
    },
    {
      "epoch": 18.380846325167038,
      "grad_norm": 18.871074676513672,
      "learning_rate": 1e-06,
      "loss": 0.3387,
      "num_input_tokens_seen": 462439476,
      "step": 8253
    },
    {
      "epoch": 18.380846325167038,
      "loss": 0.4407457113265991,
      "loss_ce": 7.18739174772054e-05,
      "loss_iou": 0.1982421875,
      "loss_num": 0.0087890625,
      "loss_xval": 0.44140625,
      "num_input_tokens_seen": 462439476,
      "step": 8253
    },
    {
      "epoch": 18.383073496659243,
      "grad_norm": 17.004552841186523,
      "learning_rate": 1e-06,
      "loss": 0.3689,
      "num_input_tokens_seen": 462492792,
      "step": 8254
    },
    {
      "epoch": 18.383073496659243,
      "loss": 0.34425753355026245,
      "loss_ce": 8.024969429243356e-05,
      "loss_iou": 0.150390625,
      "loss_num": 0.00848388671875,
      "loss_xval": 0.34375,
      "num_input_tokens_seen": 462492792,
      "step": 8254
    },
    {
      "epoch": 18.385300668151448,
      "grad_norm": 20.56110382080078,
      "learning_rate": 1e-06,
      "loss": 0.3938,
      "num_input_tokens_seen": 462549204,
      "step": 8255
    },
    {
      "epoch": 18.385300668151448,
      "loss": 0.4753004312515259,
      "loss_ce": 8.071900811046362e-05,
      "loss_iou": 0.1923828125,
      "loss_num": 0.0181884765625,
      "loss_xval": 0.474609375,
      "num_input_tokens_seen": 462549204,
      "step": 8255
    },
    {
      "epoch": 18.387527839643653,
      "grad_norm": 21.641469955444336,
      "learning_rate": 1e-06,
      "loss": 0.3759,
      "num_input_tokens_seen": 462604868,
      "step": 8256
    },
    {
      "epoch": 18.387527839643653,
      "loss": 0.42918264865875244,
      "loss_ce": 7.497478509321809e-05,
      "loss_iou": 0.162109375,
      "loss_num": 0.0208740234375,
      "loss_xval": 0.4296875,
      "num_input_tokens_seen": 462604868,
      "step": 8256
    },
    {
      "epoch": 18.389755011135858,
      "grad_norm": 10.594487190246582,
      "learning_rate": 1e-06,
      "loss": 0.4945,
      "num_input_tokens_seen": 462662936,
      "step": 8257
    },
    {
      "epoch": 18.389755011135858,
      "loss": 0.47706377506256104,
      "loss_ce": 7.405664655379951e-05,
      "loss_iou": 0.2060546875,
      "loss_num": 0.012939453125,
      "loss_xval": 0.4765625,
      "num_input_tokens_seen": 462662936,
      "step": 8257
    },
    {
      "epoch": 18.391982182628063,
      "grad_norm": 19.134235382080078,
      "learning_rate": 1e-06,
      "loss": 0.4228,
      "num_input_tokens_seen": 462717820,
      "step": 8258
    },
    {
      "epoch": 18.391982182628063,
      "loss": 0.30488812923431396,
      "loss_ce": 7.853595889173448e-05,
      "loss_iou": 0.1328125,
      "loss_num": 0.0079345703125,
      "loss_xval": 0.3046875,
      "num_input_tokens_seen": 462717820,
      "step": 8258
    },
    {
      "epoch": 18.394209354120267,
      "grad_norm": 20.989978790283203,
      "learning_rate": 1e-06,
      "loss": 0.3927,
      "num_input_tokens_seen": 462772384,
      "step": 8259
    },
    {
      "epoch": 18.394209354120267,
      "loss": 0.435162216424942,
      "loss_ce": 7.693507359363139e-05,
      "loss_iou": 0.185546875,
      "loss_num": 0.012939453125,
      "loss_xval": 0.435546875,
      "num_input_tokens_seen": 462772384,
      "step": 8259
    },
    {
      "epoch": 18.396436525612472,
      "grad_norm": 14.269499778747559,
      "learning_rate": 1e-06,
      "loss": 0.355,
      "num_input_tokens_seen": 462827028,
      "step": 8260
    },
    {
      "epoch": 18.396436525612472,
      "loss": 0.31690651178359985,
      "loss_ce": 7.30217871023342e-05,
      "loss_iou": 0.1455078125,
      "loss_num": 0.00506591796875,
      "loss_xval": 0.31640625,
      "num_input_tokens_seen": 462827028,
      "step": 8260
    },
    {
      "epoch": 18.398663697104677,
      "grad_norm": 15.133282661437988,
      "learning_rate": 1e-06,
      "loss": 0.55,
      "num_input_tokens_seen": 462883484,
      "step": 8261
    },
    {
      "epoch": 18.398663697104677,
      "loss": 0.5562236905097961,
      "loss_ce": 0.0002544423332437873,
      "loss_iou": 0.2041015625,
      "loss_num": 0.029541015625,
      "loss_xval": 0.5546875,
      "num_input_tokens_seen": 462883484,
      "step": 8261
    },
    {
      "epoch": 18.400890868596882,
      "grad_norm": 18.654014587402344,
      "learning_rate": 1e-06,
      "loss": 0.3458,
      "num_input_tokens_seen": 462941596,
      "step": 8262
    },
    {
      "epoch": 18.400890868596882,
      "loss": 0.4173450469970703,
      "loss_ce": 0.00010873953578993678,
      "loss_iou": 0.17578125,
      "loss_num": 0.01312255859375,
      "loss_xval": 0.41796875,
      "num_input_tokens_seen": 462941596,
      "step": 8262
    },
    {
      "epoch": 18.403118040089087,
      "grad_norm": 21.48236656188965,
      "learning_rate": 1e-06,
      "loss": 0.3443,
      "num_input_tokens_seen": 462997916,
      "step": 8263
    },
    {
      "epoch": 18.403118040089087,
      "loss": 0.29500818252563477,
      "loss_ce": 8.628673094790429e-05,
      "loss_iou": 0.1376953125,
      "loss_num": 0.003997802734375,
      "loss_xval": 0.294921875,
      "num_input_tokens_seen": 462997916,
      "step": 8263
    },
    {
      "epoch": 18.40534521158129,
      "grad_norm": 29.181659698486328,
      "learning_rate": 1e-06,
      "loss": 0.4467,
      "num_input_tokens_seen": 463053976,
      "step": 8264
    },
    {
      "epoch": 18.40534521158129,
      "loss": 0.5064201354980469,
      "loss_ce": 7.253700459841639e-05,
      "loss_iou": 0.236328125,
      "loss_num": 0.00689697265625,
      "loss_xval": 0.5078125,
      "num_input_tokens_seen": 463053976,
      "step": 8264
    },
    {
      "epoch": 18.407572383073497,
      "grad_norm": 18.721237182617188,
      "learning_rate": 1e-06,
      "loss": 0.4064,
      "num_input_tokens_seen": 463111504,
      "step": 8265
    },
    {
      "epoch": 18.407572383073497,
      "loss": 0.45783811807632446,
      "loss_ce": 7.442037167493254e-05,
      "loss_iou": 0.2099609375,
      "loss_num": 0.007568359375,
      "loss_xval": 0.45703125,
      "num_input_tokens_seen": 463111504,
      "step": 8265
    },
    {
      "epoch": 18.4097995545657,
      "grad_norm": 57.14765548706055,
      "learning_rate": 1e-06,
      "loss": 0.7487,
      "num_input_tokens_seen": 463168420,
      "step": 8266
    },
    {
      "epoch": 18.4097995545657,
      "loss": 0.6589952707290649,
      "loss_ce": 5.971048813080415e-05,
      "loss_iou": 0.29296875,
      "loss_num": 0.01495361328125,
      "loss_xval": 0.66015625,
      "num_input_tokens_seen": 463168420,
      "step": 8266
    },
    {
      "epoch": 18.412026726057906,
      "grad_norm": 18.461252212524414,
      "learning_rate": 1e-06,
      "loss": 0.3474,
      "num_input_tokens_seen": 463225216,
      "step": 8267
    },
    {
      "epoch": 18.412026726057906,
      "loss": 0.38068366050720215,
      "loss_ce": 6.842733273515478e-05,
      "loss_iou": 0.1767578125,
      "loss_num": 0.005462646484375,
      "loss_xval": 0.380859375,
      "num_input_tokens_seen": 463225216,
      "step": 8267
    },
    {
      "epoch": 18.41425389755011,
      "grad_norm": 23.637659072875977,
      "learning_rate": 1e-06,
      "loss": 0.5465,
      "num_input_tokens_seen": 463279816,
      "step": 8268
    },
    {
      "epoch": 18.41425389755011,
      "loss": 0.513511061668396,
      "loss_ce": 8.326944953296334e-05,
      "loss_iou": 0.232421875,
      "loss_num": 0.00958251953125,
      "loss_xval": 0.51171875,
      "num_input_tokens_seen": 463279816,
      "step": 8268
    },
    {
      "epoch": 18.416481069042316,
      "grad_norm": 19.79255485534668,
      "learning_rate": 1e-06,
      "loss": 0.5626,
      "num_input_tokens_seen": 463336080,
      "step": 8269
    },
    {
      "epoch": 18.416481069042316,
      "loss": 0.5551245808601379,
      "loss_ce": 7.088375423336402e-05,
      "loss_iou": 0.2255859375,
      "loss_num": 0.0208740234375,
      "loss_xval": 0.5546875,
      "num_input_tokens_seen": 463336080,
      "step": 8269
    },
    {
      "epoch": 18.41870824053452,
      "grad_norm": 22.363147735595703,
      "learning_rate": 1e-06,
      "loss": 0.3449,
      "num_input_tokens_seen": 463392188,
      "step": 8270
    },
    {
      "epoch": 18.41870824053452,
      "loss": 0.27490442991256714,
      "loss_ce": 6.311971810646355e-05,
      "loss_iou": 0.1181640625,
      "loss_num": 0.0078125,
      "loss_xval": 0.275390625,
      "num_input_tokens_seen": 463392188,
      "step": 8270
    },
    {
      "epoch": 18.420935412026726,
      "grad_norm": 12.046845436096191,
      "learning_rate": 1e-06,
      "loss": 0.3125,
      "num_input_tokens_seen": 463450560,
      "step": 8271
    },
    {
      "epoch": 18.420935412026726,
      "loss": 0.33826929330825806,
      "loss_ce": 7.349113002419472e-05,
      "loss_iou": 0.1396484375,
      "loss_num": 0.01190185546875,
      "loss_xval": 0.337890625,
      "num_input_tokens_seen": 463450560,
      "step": 8271
    },
    {
      "epoch": 18.42316258351893,
      "grad_norm": 14.040665626525879,
      "learning_rate": 1e-06,
      "loss": 0.3608,
      "num_input_tokens_seen": 463504208,
      "step": 8272
    },
    {
      "epoch": 18.42316258351893,
      "loss": 0.2680038809776306,
      "loss_ce": 0.00012056898412993178,
      "loss_iou": 0.10986328125,
      "loss_num": 0.00970458984375,
      "loss_xval": 0.267578125,
      "num_input_tokens_seen": 463504208,
      "step": 8272
    },
    {
      "epoch": 18.425389755011135,
      "grad_norm": 20.083757400512695,
      "learning_rate": 1e-06,
      "loss": 0.3636,
      "num_input_tokens_seen": 463562476,
      "step": 8273
    },
    {
      "epoch": 18.425389755011135,
      "loss": 0.23639193177223206,
      "loss_ce": 6.381591811077669e-05,
      "loss_iou": 0.107421875,
      "loss_num": 0.0042724609375,
      "loss_xval": 0.236328125,
      "num_input_tokens_seen": 463562476,
      "step": 8273
    },
    {
      "epoch": 18.42761692650334,
      "grad_norm": 22.100793838500977,
      "learning_rate": 1e-06,
      "loss": 0.427,
      "num_input_tokens_seen": 463617160,
      "step": 8274
    },
    {
      "epoch": 18.42761692650334,
      "loss": 0.23444892466068268,
      "loss_ce": 5.8669171266956255e-05,
      "loss_iou": 0.0869140625,
      "loss_num": 0.01214599609375,
      "loss_xval": 0.234375,
      "num_input_tokens_seen": 463617160,
      "step": 8274
    },
    {
      "epoch": 18.429844097995545,
      "grad_norm": 20.41813087463379,
      "learning_rate": 1e-06,
      "loss": 0.4329,
      "num_input_tokens_seen": 463673804,
      "step": 8275
    },
    {
      "epoch": 18.429844097995545,
      "loss": 0.38324350118637085,
      "loss_ce": 6.480171578004956e-05,
      "loss_iou": 0.1572265625,
      "loss_num": 0.0135498046875,
      "loss_xval": 0.3828125,
      "num_input_tokens_seen": 463673804,
      "step": 8275
    },
    {
      "epoch": 18.43207126948775,
      "grad_norm": 17.835351943969727,
      "learning_rate": 1e-06,
      "loss": 0.4192,
      "num_input_tokens_seen": 463731528,
      "step": 8276
    },
    {
      "epoch": 18.43207126948775,
      "loss": 0.3969186246395111,
      "loss_ce": 6.806287274230272e-05,
      "loss_iou": 0.166015625,
      "loss_num": 0.0130615234375,
      "loss_xval": 0.396484375,
      "num_input_tokens_seen": 463731528,
      "step": 8276
    },
    {
      "epoch": 18.434298440979955,
      "grad_norm": 20.909772872924805,
      "learning_rate": 1e-06,
      "loss": 0.3829,
      "num_input_tokens_seen": 463786092,
      "step": 8277
    },
    {
      "epoch": 18.434298440979955,
      "loss": 0.35701119899749756,
      "loss_ce": 7.761004235362634e-05,
      "loss_iou": 0.14453125,
      "loss_num": 0.01348876953125,
      "loss_xval": 0.357421875,
      "num_input_tokens_seen": 463786092,
      "step": 8277
    },
    {
      "epoch": 18.43652561247216,
      "grad_norm": 15.374547958374023,
      "learning_rate": 1e-06,
      "loss": 0.4143,
      "num_input_tokens_seen": 463838012,
      "step": 8278
    },
    {
      "epoch": 18.43652561247216,
      "loss": 0.4003612995147705,
      "loss_ce": 9.272516763303429e-05,
      "loss_iou": 0.1689453125,
      "loss_num": 0.0126953125,
      "loss_xval": 0.400390625,
      "num_input_tokens_seen": 463838012,
      "step": 8278
    },
    {
      "epoch": 18.438752783964365,
      "grad_norm": 12.226360321044922,
      "learning_rate": 1e-06,
      "loss": 0.4797,
      "num_input_tokens_seen": 463892444,
      "step": 8279
    },
    {
      "epoch": 18.438752783964365,
      "loss": 0.42820942401885986,
      "loss_ce": 7.8319848398678e-05,
      "loss_iou": 0.1884765625,
      "loss_num": 0.01007080078125,
      "loss_xval": 0.427734375,
      "num_input_tokens_seen": 463892444,
      "step": 8279
    },
    {
      "epoch": 18.44097995545657,
      "grad_norm": 17.95158576965332,
      "learning_rate": 1e-06,
      "loss": 0.3166,
      "num_input_tokens_seen": 463947952,
      "step": 8280
    },
    {
      "epoch": 18.44097995545657,
      "loss": 0.3516421616077423,
      "loss_ce": 7.965406985022128e-05,
      "loss_iou": 0.146484375,
      "loss_num": 0.01153564453125,
      "loss_xval": 0.3515625,
      "num_input_tokens_seen": 463947952,
      "step": 8280
    },
    {
      "epoch": 18.443207126948774,
      "grad_norm": 16.93501853942871,
      "learning_rate": 1e-06,
      "loss": 0.4472,
      "num_input_tokens_seen": 464005712,
      "step": 8281
    },
    {
      "epoch": 18.443207126948774,
      "loss": 0.47952741384506226,
      "loss_ce": 9.625754319131374e-05,
      "loss_iou": 0.2060546875,
      "loss_num": 0.01324462890625,
      "loss_xval": 0.478515625,
      "num_input_tokens_seen": 464005712,
      "step": 8281
    },
    {
      "epoch": 18.44543429844098,
      "grad_norm": 20.220458984375,
      "learning_rate": 1e-06,
      "loss": 0.4457,
      "num_input_tokens_seen": 464062920,
      "step": 8282
    },
    {
      "epoch": 18.44543429844098,
      "loss": 0.4217246472835541,
      "loss_ce": 9.378813410876319e-05,
      "loss_iou": 0.17578125,
      "loss_num": 0.0137939453125,
      "loss_xval": 0.421875,
      "num_input_tokens_seen": 464062920,
      "step": 8282
    },
    {
      "epoch": 18.447661469933184,
      "grad_norm": 15.873303413391113,
      "learning_rate": 1e-06,
      "loss": 0.2138,
      "num_input_tokens_seen": 464116484,
      "step": 8283
    },
    {
      "epoch": 18.447661469933184,
      "loss": 0.14678719639778137,
      "loss_ce": 6.631132418988273e-05,
      "loss_iou": 0.06298828125,
      "loss_num": 0.004241943359375,
      "loss_xval": 0.146484375,
      "num_input_tokens_seen": 464116484,
      "step": 8283
    },
    {
      "epoch": 18.44988864142539,
      "grad_norm": 18.4737548828125,
      "learning_rate": 1e-06,
      "loss": 0.41,
      "num_input_tokens_seen": 464172992,
      "step": 8284
    },
    {
      "epoch": 18.44988864142539,
      "loss": 0.3817276358604431,
      "loss_ce": 7.478396582882851e-05,
      "loss_iou": 0.1689453125,
      "loss_num": 0.00872802734375,
      "loss_xval": 0.380859375,
      "num_input_tokens_seen": 464172992,
      "step": 8284
    },
    {
      "epoch": 18.452115812917594,
      "grad_norm": 19.414098739624023,
      "learning_rate": 1e-06,
      "loss": 0.4462,
      "num_input_tokens_seen": 464225548,
      "step": 8285
    },
    {
      "epoch": 18.452115812917594,
      "loss": 0.42627277970314026,
      "loss_ce": 0.00012532366963569075,
      "loss_iou": 0.1865234375,
      "loss_num": 0.01068115234375,
      "loss_xval": 0.42578125,
      "num_input_tokens_seen": 464225548,
      "step": 8285
    },
    {
      "epoch": 18.4543429844098,
      "grad_norm": 16.32335090637207,
      "learning_rate": 1e-06,
      "loss": 0.3318,
      "num_input_tokens_seen": 464281668,
      "step": 8286
    },
    {
      "epoch": 18.4543429844098,
      "loss": 0.3537209630012512,
      "loss_ce": 8.327406249009073e-05,
      "loss_iou": 0.1572265625,
      "loss_num": 0.00787353515625,
      "loss_xval": 0.353515625,
      "num_input_tokens_seen": 464281668,
      "step": 8286
    },
    {
      "epoch": 18.456570155902003,
      "grad_norm": 15.517791748046875,
      "learning_rate": 1e-06,
      "loss": 0.3719,
      "num_input_tokens_seen": 464337160,
      "step": 8287
    },
    {
      "epoch": 18.456570155902003,
      "loss": 0.3465074300765991,
      "loss_ce": 0.0001023836521198973,
      "loss_iou": 0.1376953125,
      "loss_num": 0.01434326171875,
      "loss_xval": 0.345703125,
      "num_input_tokens_seen": 464337160,
      "step": 8287
    },
    {
      "epoch": 18.45879732739421,
      "grad_norm": 18.08324432373047,
      "learning_rate": 1e-06,
      "loss": 0.5502,
      "num_input_tokens_seen": 464392772,
      "step": 8288
    },
    {
      "epoch": 18.45879732739421,
      "loss": 0.6127073168754578,
      "loss_ce": 0.00015847003669478,
      "loss_iou": 0.23828125,
      "loss_num": 0.0272216796875,
      "loss_xval": 0.61328125,
      "num_input_tokens_seen": 464392772,
      "step": 8288
    },
    {
      "epoch": 18.461024498886413,
      "grad_norm": 18.83432960510254,
      "learning_rate": 1e-06,
      "loss": 0.3693,
      "num_input_tokens_seen": 464448784,
      "step": 8289
    },
    {
      "epoch": 18.461024498886413,
      "loss": 0.4016864001750946,
      "loss_ce": 7.508075213991106e-05,
      "loss_iou": 0.1845703125,
      "loss_num": 0.006378173828125,
      "loss_xval": 0.40234375,
      "num_input_tokens_seen": 464448784,
      "step": 8289
    },
    {
      "epoch": 18.463251670378618,
      "grad_norm": 13.436317443847656,
      "learning_rate": 1e-06,
      "loss": 0.3396,
      "num_input_tokens_seen": 464505648,
      "step": 8290
    },
    {
      "epoch": 18.463251670378618,
      "loss": 0.35365384817123413,
      "loss_ce": 7.717913103988394e-05,
      "loss_iou": 0.15234375,
      "loss_num": 0.009765625,
      "loss_xval": 0.353515625,
      "num_input_tokens_seen": 464505648,
      "step": 8290
    },
    {
      "epoch": 18.465478841870823,
      "grad_norm": 19.951303482055664,
      "learning_rate": 1e-06,
      "loss": 0.3453,
      "num_input_tokens_seen": 464559060,
      "step": 8291
    },
    {
      "epoch": 18.465478841870823,
      "loss": 0.38332536816596985,
      "loss_ce": 0.00014667093637399375,
      "loss_iou": 0.1708984375,
      "loss_num": 0.008056640625,
      "loss_xval": 0.3828125,
      "num_input_tokens_seen": 464559060,
      "step": 8291
    },
    {
      "epoch": 18.467706013363028,
      "grad_norm": 228.0105438232422,
      "learning_rate": 1e-06,
      "loss": 0.4727,
      "num_input_tokens_seen": 464614584,
      "step": 8292
    },
    {
      "epoch": 18.467706013363028,
      "loss": 0.3697110116481781,
      "loss_ce": 8.208431245293468e-05,
      "loss_iou": 0.150390625,
      "loss_num": 0.013916015625,
      "loss_xval": 0.369140625,
      "num_input_tokens_seen": 464614584,
      "step": 8292
    },
    {
      "epoch": 18.469933184855233,
      "grad_norm": 14.540281295776367,
      "learning_rate": 1e-06,
      "loss": 0.3425,
      "num_input_tokens_seen": 464668316,
      "step": 8293
    },
    {
      "epoch": 18.469933184855233,
      "loss": 0.34370023012161255,
      "loss_ce": 7.22910335753113e-05,
      "loss_iou": 0.15625,
      "loss_num": 0.006072998046875,
      "loss_xval": 0.34375,
      "num_input_tokens_seen": 464668316,
      "step": 8293
    },
    {
      "epoch": 18.472160356347437,
      "grad_norm": 20.030624389648438,
      "learning_rate": 1e-06,
      "loss": 0.4044,
      "num_input_tokens_seen": 464723240,
      "step": 8294
    },
    {
      "epoch": 18.472160356347437,
      "loss": 0.40011394023895264,
      "loss_ce": 8.954846998676658e-05,
      "loss_iou": 0.18359375,
      "loss_num": 0.00677490234375,
      "loss_xval": 0.400390625,
      "num_input_tokens_seen": 464723240,
      "step": 8294
    },
    {
      "epoch": 18.474387527839642,
      "grad_norm": 13.285386085510254,
      "learning_rate": 1e-06,
      "loss": 0.4026,
      "num_input_tokens_seen": 464779508,
      "step": 8295
    },
    {
      "epoch": 18.474387527839642,
      "loss": 0.223904550075531,
      "loss_ce": 8.862309186952189e-05,
      "loss_iou": 0.10498046875,
      "loss_num": 0.0027008056640625,
      "loss_xval": 0.2236328125,
      "num_input_tokens_seen": 464779508,
      "step": 8295
    },
    {
      "epoch": 18.476614699331847,
      "grad_norm": 18.240877151489258,
      "learning_rate": 1e-06,
      "loss": 0.3385,
      "num_input_tokens_seen": 464834748,
      "step": 8296
    },
    {
      "epoch": 18.476614699331847,
      "loss": 0.29583436250686646,
      "loss_ce": 8.852417522575706e-05,
      "loss_iou": 0.1376953125,
      "loss_num": 0.003936767578125,
      "loss_xval": 0.294921875,
      "num_input_tokens_seen": 464834748,
      "step": 8296
    },
    {
      "epoch": 18.478841870824052,
      "grad_norm": 15.406899452209473,
      "learning_rate": 1e-06,
      "loss": 0.3808,
      "num_input_tokens_seen": 464892396,
      "step": 8297
    },
    {
      "epoch": 18.478841870824052,
      "loss": 0.379098504781723,
      "loss_ce": 7.018136966507882e-05,
      "loss_iou": 0.1650390625,
      "loss_num": 0.00994873046875,
      "loss_xval": 0.37890625,
      "num_input_tokens_seen": 464892396,
      "step": 8297
    },
    {
      "epoch": 18.481069042316257,
      "grad_norm": 22.26421546936035,
      "learning_rate": 1e-06,
      "loss": 0.3392,
      "num_input_tokens_seen": 464949808,
      "step": 8298
    },
    {
      "epoch": 18.481069042316257,
      "loss": 0.21200445294380188,
      "loss_ce": 9.038990538101643e-05,
      "loss_iou": 0.091796875,
      "loss_num": 0.005645751953125,
      "loss_xval": 0.2119140625,
      "num_input_tokens_seen": 464949808,
      "step": 8298
    },
    {
      "epoch": 18.48329621380846,
      "grad_norm": 15.141512870788574,
      "learning_rate": 1e-06,
      "loss": 0.403,
      "num_input_tokens_seen": 465003316,
      "step": 8299
    },
    {
      "epoch": 18.48329621380846,
      "loss": 0.4960575997829437,
      "loss_ce": 8.59203573781997e-05,
      "loss_iou": 0.193359375,
      "loss_num": 0.0216064453125,
      "loss_xval": 0.49609375,
      "num_input_tokens_seen": 465003316,
      "step": 8299
    },
    {
      "epoch": 18.485523385300667,
      "grad_norm": 29.124523162841797,
      "learning_rate": 1e-06,
      "loss": 0.5493,
      "num_input_tokens_seen": 465058828,
      "step": 8300
    },
    {
      "epoch": 18.485523385300667,
      "loss": 0.6504954695701599,
      "loss_ce": 0.0001048904232447967,
      "loss_iou": 0.298828125,
      "loss_num": 0.01068115234375,
      "loss_xval": 0.6484375,
      "num_input_tokens_seen": 465058828,
      "step": 8300
    },
    {
      "epoch": 18.48775055679287,
      "grad_norm": 17.59904670715332,
      "learning_rate": 1e-06,
      "loss": 0.3864,
      "num_input_tokens_seen": 465114628,
      "step": 8301
    },
    {
      "epoch": 18.48775055679287,
      "loss": 0.263241708278656,
      "loss_ce": 0.00011917389929294586,
      "loss_iou": 0.115234375,
      "loss_num": 0.00653076171875,
      "loss_xval": 0.263671875,
      "num_input_tokens_seen": 465114628,
      "step": 8301
    },
    {
      "epoch": 18.489977728285076,
      "grad_norm": 22.050037384033203,
      "learning_rate": 1e-06,
      "loss": 0.4467,
      "num_input_tokens_seen": 465169408,
      "step": 8302
    },
    {
      "epoch": 18.489977728285076,
      "loss": 0.29093292355537415,
      "loss_ce": 6.987818778725341e-05,
      "loss_iou": 0.130859375,
      "loss_num": 0.005889892578125,
      "loss_xval": 0.291015625,
      "num_input_tokens_seen": 465169408,
      "step": 8302
    },
    {
      "epoch": 18.49220489977728,
      "grad_norm": 20.6636962890625,
      "learning_rate": 1e-06,
      "loss": 0.3993,
      "num_input_tokens_seen": 465224880,
      "step": 8303
    },
    {
      "epoch": 18.49220489977728,
      "loss": 0.41500934958457947,
      "loss_ce": 9.234360186383128e-05,
      "loss_iou": 0.173828125,
      "loss_num": 0.013671875,
      "loss_xval": 0.4140625,
      "num_input_tokens_seen": 465224880,
      "step": 8303
    },
    {
      "epoch": 18.494432071269486,
      "grad_norm": 23.35464859008789,
      "learning_rate": 1e-06,
      "loss": 0.3545,
      "num_input_tokens_seen": 465281588,
      "step": 8304
    },
    {
      "epoch": 18.494432071269486,
      "loss": 0.4025862514972687,
      "loss_ce": 7.466854003723711e-05,
      "loss_iou": 0.1708984375,
      "loss_num": 0.01226806640625,
      "loss_xval": 0.40234375,
      "num_input_tokens_seen": 465281588,
      "step": 8304
    },
    {
      "epoch": 18.49665924276169,
      "grad_norm": 18.54194450378418,
      "learning_rate": 1e-06,
      "loss": 0.4179,
      "num_input_tokens_seen": 465336308,
      "step": 8305
    },
    {
      "epoch": 18.49665924276169,
      "loss": 0.4852302670478821,
      "loss_ce": 6.180583295645192e-05,
      "loss_iou": 0.2021484375,
      "loss_num": 0.0159912109375,
      "loss_xval": 0.484375,
      "num_input_tokens_seen": 465336308,
      "step": 8305
    },
    {
      "epoch": 18.498886414253896,
      "grad_norm": 16.24866485595703,
      "learning_rate": 1e-06,
      "loss": 0.2652,
      "num_input_tokens_seen": 465392124,
      "step": 8306
    },
    {
      "epoch": 18.498886414253896,
      "loss": 0.3223702311515808,
      "loss_ce": 0.00010461719648446888,
      "loss_iou": 0.1474609375,
      "loss_num": 0.005462646484375,
      "loss_xval": 0.322265625,
      "num_input_tokens_seen": 465392124,
      "step": 8306
    },
    {
      "epoch": 18.501113585746104,
      "grad_norm": 23.2437801361084,
      "learning_rate": 1e-06,
      "loss": 0.4014,
      "num_input_tokens_seen": 465448524,
      "step": 8307
    },
    {
      "epoch": 18.501113585746104,
      "loss": 0.32942575216293335,
      "loss_ce": 8.003541734069586e-05,
      "loss_iou": 0.1494140625,
      "loss_num": 0.006103515625,
      "loss_xval": 0.330078125,
      "num_input_tokens_seen": 465448524,
      "step": 8307
    },
    {
      "epoch": 18.50334075723831,
      "grad_norm": 17.066225051879883,
      "learning_rate": 1e-06,
      "loss": 0.5872,
      "num_input_tokens_seen": 465504656,
      "step": 8308
    },
    {
      "epoch": 18.50334075723831,
      "loss": 0.6132311820983887,
      "loss_ce": 7.201532571343705e-05,
      "loss_iou": 0.2412109375,
      "loss_num": 0.0260009765625,
      "loss_xval": 0.61328125,
      "num_input_tokens_seen": 465504656,
      "step": 8308
    },
    {
      "epoch": 18.505567928730514,
      "grad_norm": 37.251121520996094,
      "learning_rate": 1e-06,
      "loss": 0.325,
      "num_input_tokens_seen": 465560192,
      "step": 8309
    },
    {
      "epoch": 18.505567928730514,
      "loss": 0.24585197865962982,
      "loss_ce": 6.340416439343244e-05,
      "loss_iou": 0.109375,
      "loss_num": 0.00543212890625,
      "loss_xval": 0.24609375,
      "num_input_tokens_seen": 465560192,
      "step": 8309
    },
    {
      "epoch": 18.50779510022272,
      "grad_norm": 16.032716751098633,
      "learning_rate": 1e-06,
      "loss": 0.5305,
      "num_input_tokens_seen": 465618100,
      "step": 8310
    },
    {
      "epoch": 18.50779510022272,
      "loss": 0.629966139793396,
      "loss_ce": 8.333367441082373e-05,
      "loss_iou": 0.25390625,
      "loss_num": 0.024169921875,
      "loss_xval": 0.62890625,
      "num_input_tokens_seen": 465618100,
      "step": 8310
    },
    {
      "epoch": 18.510022271714924,
      "grad_norm": 15.98995304107666,
      "learning_rate": 1e-06,
      "loss": 0.2552,
      "num_input_tokens_seen": 465674400,
      "step": 8311
    },
    {
      "epoch": 18.510022271714924,
      "loss": 0.27437126636505127,
      "loss_ce": 7.928490958875045e-05,
      "loss_iou": 0.1298828125,
      "loss_num": 0.0030364990234375,
      "loss_xval": 0.2734375,
      "num_input_tokens_seen": 465674400,
      "step": 8311
    },
    {
      "epoch": 18.51224944320713,
      "grad_norm": 16.019752502441406,
      "learning_rate": 1e-06,
      "loss": 0.3386,
      "num_input_tokens_seen": 465732188,
      "step": 8312
    },
    {
      "epoch": 18.51224944320713,
      "loss": 0.5247637033462524,
      "loss_ce": 0.00010548259160714224,
      "loss_iou": 0.21484375,
      "loss_num": 0.01904296875,
      "loss_xval": 0.5234375,
      "num_input_tokens_seen": 465732188,
      "step": 8312
    },
    {
      "epoch": 18.514476614699333,
      "grad_norm": 17.876802444458008,
      "learning_rate": 1e-06,
      "loss": 0.3636,
      "num_input_tokens_seen": 465786628,
      "step": 8313
    },
    {
      "epoch": 18.514476614699333,
      "loss": 0.35969775915145874,
      "loss_ce": 7.861069752834737e-05,
      "loss_iou": 0.16796875,
      "loss_num": 0.004791259765625,
      "loss_xval": 0.359375,
      "num_input_tokens_seen": 465786628,
      "step": 8313
    },
    {
      "epoch": 18.51670378619154,
      "grad_norm": 15.022839546203613,
      "learning_rate": 1e-06,
      "loss": 0.3146,
      "num_input_tokens_seen": 465842532,
      "step": 8314
    },
    {
      "epoch": 18.51670378619154,
      "loss": 0.35521650314331055,
      "loss_ce": 0.00011394656030461192,
      "loss_iou": 0.14453125,
      "loss_num": 0.01318359375,
      "loss_xval": 0.35546875,
      "num_input_tokens_seen": 465842532,
      "step": 8314
    },
    {
      "epoch": 18.518930957683743,
      "grad_norm": 13.363021850585938,
      "learning_rate": 1e-06,
      "loss": 0.4604,
      "num_input_tokens_seen": 465900596,
      "step": 8315
    },
    {
      "epoch": 18.518930957683743,
      "loss": 0.583387017250061,
      "loss_ce": 0.00013512340956367552,
      "loss_iou": 0.224609375,
      "loss_num": 0.02685546875,
      "loss_xval": 0.58203125,
      "num_input_tokens_seen": 465900596,
      "step": 8315
    },
    {
      "epoch": 18.521158129175948,
      "grad_norm": 15.131711959838867,
      "learning_rate": 1e-06,
      "loss": 0.4687,
      "num_input_tokens_seen": 465958308,
      "step": 8316
    },
    {
      "epoch": 18.521158129175948,
      "loss": 0.5298267602920532,
      "loss_ce": 7.216697849798948e-05,
      "loss_iou": 0.2158203125,
      "loss_num": 0.01953125,
      "loss_xval": 0.53125,
      "num_input_tokens_seen": 465958308,
      "step": 8316
    },
    {
      "epoch": 18.523385300668153,
      "grad_norm": 12.049201965332031,
      "learning_rate": 1e-06,
      "loss": 0.3896,
      "num_input_tokens_seen": 466015604,
      "step": 8317
    },
    {
      "epoch": 18.523385300668153,
      "loss": 0.2444000393152237,
      "loss_ce": 7.62987692723982e-05,
      "loss_iou": 0.09130859375,
      "loss_num": 0.01239013671875,
      "loss_xval": 0.244140625,
      "num_input_tokens_seen": 466015604,
      "step": 8317
    },
    {
      "epoch": 18.525612472160358,
      "grad_norm": 25.494447708129883,
      "learning_rate": 1e-06,
      "loss": 0.4866,
      "num_input_tokens_seen": 466072012,
      "step": 8318
    },
    {
      "epoch": 18.525612472160358,
      "loss": 0.5554917454719543,
      "loss_ce": 7.183050183812156e-05,
      "loss_iou": 0.2255859375,
      "loss_num": 0.0206298828125,
      "loss_xval": 0.5546875,
      "num_input_tokens_seen": 466072012,
      "step": 8318
    },
    {
      "epoch": 18.527839643652563,
      "grad_norm": 32.9737548828125,
      "learning_rate": 1e-06,
      "loss": 0.5518,
      "num_input_tokens_seen": 466126616,
      "step": 8319
    },
    {
      "epoch": 18.527839643652563,
      "loss": 0.5966952443122864,
      "loss_ce": 7.660294068045914e-05,
      "loss_iou": 0.21484375,
      "loss_num": 0.033447265625,
      "loss_xval": 0.59765625,
      "num_input_tokens_seen": 466126616,
      "step": 8319
    },
    {
      "epoch": 18.530066815144767,
      "grad_norm": 12.370501518249512,
      "learning_rate": 1e-06,
      "loss": 0.299,
      "num_input_tokens_seen": 466181408,
      "step": 8320
    },
    {
      "epoch": 18.530066815144767,
      "loss": 0.357611745595932,
      "loss_ce": 6.781626143492758e-05,
      "loss_iou": 0.162109375,
      "loss_num": 0.006805419921875,
      "loss_xval": 0.357421875,
      "num_input_tokens_seen": 466181408,
      "step": 8320
    },
    {
      "epoch": 18.532293986636972,
      "grad_norm": 15.927129745483398,
      "learning_rate": 1e-06,
      "loss": 0.383,
      "num_input_tokens_seen": 466238792,
      "step": 8321
    },
    {
      "epoch": 18.532293986636972,
      "loss": 0.40339329838752747,
      "loss_ce": 7.296907278941944e-05,
      "loss_iou": 0.1630859375,
      "loss_num": 0.01556396484375,
      "loss_xval": 0.40234375,
      "num_input_tokens_seen": 466238792,
      "step": 8321
    },
    {
      "epoch": 18.534521158129177,
      "grad_norm": 15.46563720703125,
      "learning_rate": 1e-06,
      "loss": 0.342,
      "num_input_tokens_seen": 466291660,
      "step": 8322
    },
    {
      "epoch": 18.534521158129177,
      "loss": 0.2837243676185608,
      "loss_ce": 0.00015505831106565893,
      "loss_iou": 0.12060546875,
      "loss_num": 0.0084228515625,
      "loss_xval": 0.283203125,
      "num_input_tokens_seen": 466291660,
      "step": 8322
    },
    {
      "epoch": 18.536748329621382,
      "grad_norm": 16.087427139282227,
      "learning_rate": 1e-06,
      "loss": 0.5815,
      "num_input_tokens_seen": 466344276,
      "step": 8323
    },
    {
      "epoch": 18.536748329621382,
      "loss": 0.5009580850601196,
      "loss_ce": 0.00010357146675232798,
      "loss_iou": 0.2158203125,
      "loss_num": 0.01385498046875,
      "loss_xval": 0.5,
      "num_input_tokens_seen": 466344276,
      "step": 8323
    },
    {
      "epoch": 18.538975501113587,
      "grad_norm": 20.600358963012695,
      "learning_rate": 1e-06,
      "loss": 0.5411,
      "num_input_tokens_seen": 466400420,
      "step": 8324
    },
    {
      "epoch": 18.538975501113587,
      "loss": 0.47846150398254395,
      "loss_ce": 6.796044181101024e-05,
      "loss_iou": 0.212890625,
      "loss_num": 0.0103759765625,
      "loss_xval": 0.478515625,
      "num_input_tokens_seen": 466400420,
      "step": 8324
    },
    {
      "epoch": 18.54120267260579,
      "grad_norm": 26.053192138671875,
      "learning_rate": 1e-06,
      "loss": 0.3268,
      "num_input_tokens_seen": 466453920,
      "step": 8325
    },
    {
      "epoch": 18.54120267260579,
      "loss": 0.4151201546192169,
      "loss_ce": 8.109994087135419e-05,
      "loss_iou": 0.18359375,
      "loss_num": 0.0093994140625,
      "loss_xval": 0.4140625,
      "num_input_tokens_seen": 466453920,
      "step": 8325
    },
    {
      "epoch": 18.543429844097997,
      "grad_norm": 12.725341796875,
      "learning_rate": 1e-06,
      "loss": 0.3043,
      "num_input_tokens_seen": 466510280,
      "step": 8326
    },
    {
      "epoch": 18.543429844097997,
      "loss": 0.19257915019989014,
      "loss_ce": 7.426021329592913e-05,
      "loss_iou": 0.0888671875,
      "loss_num": 0.0030517578125,
      "loss_xval": 0.1923828125,
      "num_input_tokens_seen": 466510280,
      "step": 8326
    },
    {
      "epoch": 18.5456570155902,
      "grad_norm": 19.8201904296875,
      "learning_rate": 1e-06,
      "loss": 0.6099,
      "num_input_tokens_seen": 466565908,
      "step": 8327
    },
    {
      "epoch": 18.5456570155902,
      "loss": 0.7313508987426758,
      "loss_ce": 0.00014971070049796253,
      "loss_iou": 0.296875,
      "loss_num": 0.0277099609375,
      "loss_xval": 0.73046875,
      "num_input_tokens_seen": 466565908,
      "step": 8327
    },
    {
      "epoch": 18.547884187082406,
      "grad_norm": 24.441503524780273,
      "learning_rate": 1e-06,
      "loss": 0.3867,
      "num_input_tokens_seen": 466624756,
      "step": 8328
    },
    {
      "epoch": 18.547884187082406,
      "loss": 0.46570825576782227,
      "loss_ce": 7.106846169335768e-05,
      "loss_iou": 0.2099609375,
      "loss_num": 0.00909423828125,
      "loss_xval": 0.46484375,
      "num_input_tokens_seen": 466624756,
      "step": 8328
    },
    {
      "epoch": 18.55011135857461,
      "grad_norm": 16.419023513793945,
      "learning_rate": 1e-06,
      "loss": 0.3623,
      "num_input_tokens_seen": 466682572,
      "step": 8329
    },
    {
      "epoch": 18.55011135857461,
      "loss": 0.24810969829559326,
      "loss_ce": 0.00036799628287553787,
      "loss_iou": 0.11279296875,
      "loss_num": 0.004425048828125,
      "loss_xval": 0.248046875,
      "num_input_tokens_seen": 466682572,
      "step": 8329
    },
    {
      "epoch": 18.552338530066816,
      "grad_norm": 29.954700469970703,
      "learning_rate": 1e-06,
      "loss": 0.2989,
      "num_input_tokens_seen": 466739668,
      "step": 8330
    },
    {
      "epoch": 18.552338530066816,
      "loss": 0.264358252286911,
      "loss_ce": 7.602167170261964e-05,
      "loss_iou": 0.1201171875,
      "loss_num": 0.004852294921875,
      "loss_xval": 0.263671875,
      "num_input_tokens_seen": 466739668,
      "step": 8330
    },
    {
      "epoch": 18.55456570155902,
      "grad_norm": 21.188688278198242,
      "learning_rate": 1e-06,
      "loss": 0.4046,
      "num_input_tokens_seen": 466798292,
      "step": 8331
    },
    {
      "epoch": 18.55456570155902,
      "loss": 0.2884581685066223,
      "loss_ce": 6.705736450385302e-05,
      "loss_iou": 0.1259765625,
      "loss_num": 0.007415771484375,
      "loss_xval": 0.2890625,
      "num_input_tokens_seen": 466798292,
      "step": 8331
    },
    {
      "epoch": 18.556792873051226,
      "grad_norm": 13.528090476989746,
      "learning_rate": 1e-06,
      "loss": 0.4685,
      "num_input_tokens_seen": 466856176,
      "step": 8332
    },
    {
      "epoch": 18.556792873051226,
      "loss": 0.5392540693283081,
      "loss_ce": 6.949950329726562e-05,
      "loss_iou": 0.22265625,
      "loss_num": 0.0186767578125,
      "loss_xval": 0.5390625,
      "num_input_tokens_seen": 466856176,
      "step": 8332
    },
    {
      "epoch": 18.55902004454343,
      "grad_norm": 18.641672134399414,
      "learning_rate": 1e-06,
      "loss": 0.4738,
      "num_input_tokens_seen": 466908128,
      "step": 8333
    },
    {
      "epoch": 18.55902004454343,
      "loss": 0.4274500608444214,
      "loss_ce": 8.191791857825592e-05,
      "loss_iou": 0.1787109375,
      "loss_num": 0.0140380859375,
      "loss_xval": 0.427734375,
      "num_input_tokens_seen": 466908128,
      "step": 8333
    },
    {
      "epoch": 18.561247216035635,
      "grad_norm": 29.179141998291016,
      "learning_rate": 1e-06,
      "loss": 0.3837,
      "num_input_tokens_seen": 466965624,
      "step": 8334
    },
    {
      "epoch": 18.561247216035635,
      "loss": 0.3577282428741455,
      "loss_ce": 6.22493025730364e-05,
      "loss_iou": 0.1591796875,
      "loss_num": 0.0078125,
      "loss_xval": 0.357421875,
      "num_input_tokens_seen": 466965624,
      "step": 8334
    },
    {
      "epoch": 18.56347438752784,
      "grad_norm": 15.943428993225098,
      "learning_rate": 1e-06,
      "loss": 0.4853,
      "num_input_tokens_seen": 467019384,
      "step": 8335
    },
    {
      "epoch": 18.56347438752784,
      "loss": 0.24585062265396118,
      "loss_ce": 6.203760858625174e-05,
      "loss_iou": 0.099609375,
      "loss_num": 0.0093994140625,
      "loss_xval": 0.24609375,
      "num_input_tokens_seen": 467019384,
      "step": 8335
    },
    {
      "epoch": 18.565701559020045,
      "grad_norm": 17.852943420410156,
      "learning_rate": 1e-06,
      "loss": 0.4954,
      "num_input_tokens_seen": 467076952,
      "step": 8336
    },
    {
      "epoch": 18.565701559020045,
      "loss": 0.46540504693984985,
      "loss_ce": 7.300135621335357e-05,
      "loss_iou": 0.20703125,
      "loss_num": 0.01031494140625,
      "loss_xval": 0.46484375,
      "num_input_tokens_seen": 467076952,
      "step": 8336
    },
    {
      "epoch": 18.56792873051225,
      "grad_norm": 16.556804656982422,
      "learning_rate": 1e-06,
      "loss": 0.4165,
      "num_input_tokens_seen": 467130340,
      "step": 8337
    },
    {
      "epoch": 18.56792873051225,
      "loss": 0.4885514974594116,
      "loss_ce": 8.713564602658153e-05,
      "loss_iou": 0.1796875,
      "loss_num": 0.0260009765625,
      "loss_xval": 0.48828125,
      "num_input_tokens_seen": 467130340,
      "step": 8337
    },
    {
      "epoch": 18.570155902004455,
      "grad_norm": 18.44285011291504,
      "learning_rate": 1e-06,
      "loss": 0.4634,
      "num_input_tokens_seen": 467187616,
      "step": 8338
    },
    {
      "epoch": 18.570155902004455,
      "loss": 0.5214405059814453,
      "loss_ce": 7.816489960532635e-05,
      "loss_iou": 0.205078125,
      "loss_num": 0.0223388671875,
      "loss_xval": 0.51953125,
      "num_input_tokens_seen": 467187616,
      "step": 8338
    },
    {
      "epoch": 18.57238307349666,
      "grad_norm": 16.12818717956543,
      "learning_rate": 1e-06,
      "loss": 0.5461,
      "num_input_tokens_seen": 467242808,
      "step": 8339
    },
    {
      "epoch": 18.57238307349666,
      "loss": 0.49351948499679565,
      "loss_ce": 0.00011128241749247536,
      "loss_iou": 0.201171875,
      "loss_num": 0.0181884765625,
      "loss_xval": 0.494140625,
      "num_input_tokens_seen": 467242808,
      "step": 8339
    },
    {
      "epoch": 18.574610244988865,
      "grad_norm": 13.428854942321777,
      "learning_rate": 1e-06,
      "loss": 0.3354,
      "num_input_tokens_seen": 467299608,
      "step": 8340
    },
    {
      "epoch": 18.574610244988865,
      "loss": 0.2194262444972992,
      "loss_ce": 6.589026452274993e-05,
      "loss_iou": 0.07470703125,
      "loss_num": 0.01397705078125,
      "loss_xval": 0.2197265625,
      "num_input_tokens_seen": 467299608,
      "step": 8340
    },
    {
      "epoch": 18.57683741648107,
      "grad_norm": 341.0340881347656,
      "learning_rate": 1e-06,
      "loss": 0.552,
      "num_input_tokens_seen": 467354420,
      "step": 8341
    },
    {
      "epoch": 18.57683741648107,
      "loss": 0.6670901775360107,
      "loss_ce": 9.794821380637586e-05,
      "loss_iou": 0.287109375,
      "loss_num": 0.018798828125,
      "loss_xval": 0.66796875,
      "num_input_tokens_seen": 467354420,
      "step": 8341
    },
    {
      "epoch": 18.579064587973274,
      "grad_norm": 13.796717643737793,
      "learning_rate": 1e-06,
      "loss": 0.4222,
      "num_input_tokens_seen": 467409192,
      "step": 8342
    },
    {
      "epoch": 18.579064587973274,
      "loss": 0.46046194434165955,
      "loss_ce": 7.377237488981336e-05,
      "loss_iou": 0.2041015625,
      "loss_num": 0.01055908203125,
      "loss_xval": 0.4609375,
      "num_input_tokens_seen": 467409192,
      "step": 8342
    },
    {
      "epoch": 18.58129175946548,
      "grad_norm": 24.182491302490234,
      "learning_rate": 1e-06,
      "loss": 0.4398,
      "num_input_tokens_seen": 467464804,
      "step": 8343
    },
    {
      "epoch": 18.58129175946548,
      "loss": 0.4643157720565796,
      "loss_ce": 8.239349699579179e-05,
      "loss_iou": 0.216796875,
      "loss_num": 0.006134033203125,
      "loss_xval": 0.46484375,
      "num_input_tokens_seen": 467464804,
      "step": 8343
    },
    {
      "epoch": 18.583518930957684,
      "grad_norm": 29.6402645111084,
      "learning_rate": 1e-06,
      "loss": 0.5611,
      "num_input_tokens_seen": 467520644,
      "step": 8344
    },
    {
      "epoch": 18.583518930957684,
      "loss": 0.660957932472229,
      "loss_ce": 6.922356260474771e-05,
      "loss_iou": 0.2734375,
      "loss_num": 0.023193359375,
      "loss_xval": 0.66015625,
      "num_input_tokens_seen": 467520644,
      "step": 8344
    },
    {
      "epoch": 18.58574610244989,
      "grad_norm": 22.357826232910156,
      "learning_rate": 1e-06,
      "loss": 0.2284,
      "num_input_tokens_seen": 467578040,
      "step": 8345
    },
    {
      "epoch": 18.58574610244989,
      "loss": 0.16835345327854156,
      "loss_ce": 7.953326712595299e-05,
      "loss_iou": 0.06005859375,
      "loss_num": 0.00970458984375,
      "loss_xval": 0.16796875,
      "num_input_tokens_seen": 467578040,
      "step": 8345
    },
    {
      "epoch": 18.587973273942094,
      "grad_norm": 42.021976470947266,
      "learning_rate": 1e-06,
      "loss": 0.4582,
      "num_input_tokens_seen": 467633620,
      "step": 8346
    },
    {
      "epoch": 18.587973273942094,
      "loss": 0.41333815455436707,
      "loss_ce": 9.962193144019693e-05,
      "loss_iou": 0.189453125,
      "loss_num": 0.00701904296875,
      "loss_xval": 0.4140625,
      "num_input_tokens_seen": 467633620,
      "step": 8346
    },
    {
      "epoch": 18.5902004454343,
      "grad_norm": 21.736515045166016,
      "learning_rate": 1e-06,
      "loss": 0.374,
      "num_input_tokens_seen": 467691248,
      "step": 8347
    },
    {
      "epoch": 18.5902004454343,
      "loss": 0.37519749999046326,
      "loss_ce": 7.541398372268304e-05,
      "loss_iou": 0.1708984375,
      "loss_num": 0.00665283203125,
      "loss_xval": 0.375,
      "num_input_tokens_seen": 467691248,
      "step": 8347
    },
    {
      "epoch": 18.592427616926503,
      "grad_norm": 25.333066940307617,
      "learning_rate": 1e-06,
      "loss": 0.4718,
      "num_input_tokens_seen": 467744092,
      "step": 8348
    },
    {
      "epoch": 18.592427616926503,
      "loss": 0.39171791076660156,
      "loss_ce": 0.00011635862028924748,
      "loss_iou": 0.1748046875,
      "loss_num": 0.00836181640625,
      "loss_xval": 0.390625,
      "num_input_tokens_seen": 467744092,
      "step": 8348
    },
    {
      "epoch": 18.59465478841871,
      "grad_norm": 19.491870880126953,
      "learning_rate": 1e-06,
      "loss": 0.2949,
      "num_input_tokens_seen": 467800704,
      "step": 8349
    },
    {
      "epoch": 18.59465478841871,
      "loss": 0.2856776714324951,
      "loss_ce": 9.417271940037608e-05,
      "loss_iou": 0.11669921875,
      "loss_num": 0.0103759765625,
      "loss_xval": 0.28515625,
      "num_input_tokens_seen": 467800704,
      "step": 8349
    },
    {
      "epoch": 18.596881959910913,
      "grad_norm": 11.345683097839355,
      "learning_rate": 1e-06,
      "loss": 0.2598,
      "num_input_tokens_seen": 467858852,
      "step": 8350
    },
    {
      "epoch": 18.596881959910913,
      "loss": 0.23634421825408936,
      "loss_ce": 7.713548257015646e-05,
      "loss_iou": 0.1044921875,
      "loss_num": 0.005340576171875,
      "loss_xval": 0.236328125,
      "num_input_tokens_seen": 467858852,
      "step": 8350
    },
    {
      "epoch": 18.599109131403118,
      "grad_norm": 16.46609115600586,
      "learning_rate": 1e-06,
      "loss": 0.3114,
      "num_input_tokens_seen": 467914508,
      "step": 8351
    },
    {
      "epoch": 18.599109131403118,
      "loss": 0.2579895257949829,
      "loss_ce": 0.00011599328718148172,
      "loss_iou": 0.109375,
      "loss_num": 0.007781982421875,
      "loss_xval": 0.2578125,
      "num_input_tokens_seen": 467914508,
      "step": 8351
    },
    {
      "epoch": 18.601336302895323,
      "grad_norm": 21.424537658691406,
      "learning_rate": 1e-06,
      "loss": 0.4388,
      "num_input_tokens_seen": 467969980,
      "step": 8352
    },
    {
      "epoch": 18.601336302895323,
      "loss": 0.31832072138786316,
      "loss_ce": 8.341785724041983e-05,
      "loss_iou": 0.14453125,
      "loss_num": 0.005767822265625,
      "loss_xval": 0.318359375,
      "num_input_tokens_seen": 467969980,
      "step": 8352
    },
    {
      "epoch": 18.603563474387528,
      "grad_norm": 16.22272491455078,
      "learning_rate": 1e-06,
      "loss": 0.3837,
      "num_input_tokens_seen": 468026936,
      "step": 8353
    },
    {
      "epoch": 18.603563474387528,
      "loss": 0.42490488290786743,
      "loss_ce": 0.00010020087211159989,
      "loss_iou": 0.189453125,
      "loss_num": 0.009033203125,
      "loss_xval": 0.42578125,
      "num_input_tokens_seen": 468026936,
      "step": 8353
    },
    {
      "epoch": 18.605790645879733,
      "grad_norm": 19.997718811035156,
      "learning_rate": 1e-06,
      "loss": 0.3883,
      "num_input_tokens_seen": 468082576,
      "step": 8354
    },
    {
      "epoch": 18.605790645879733,
      "loss": 0.41608962416648865,
      "loss_ce": 7.400992035400122e-05,
      "loss_iou": 0.1953125,
      "loss_num": 0.004974365234375,
      "loss_xval": 0.416015625,
      "num_input_tokens_seen": 468082576,
      "step": 8354
    },
    {
      "epoch": 18.608017817371937,
      "grad_norm": 18.959848403930664,
      "learning_rate": 1e-06,
      "loss": 0.3883,
      "num_input_tokens_seen": 468139516,
      "step": 8355
    },
    {
      "epoch": 18.608017817371937,
      "loss": 0.3627595603466034,
      "loss_ce": 8.86675261426717e-05,
      "loss_iou": 0.1630859375,
      "loss_num": 0.00732421875,
      "loss_xval": 0.36328125,
      "num_input_tokens_seen": 468139516,
      "step": 8355
    },
    {
      "epoch": 18.610244988864142,
      "grad_norm": 10.21224308013916,
      "learning_rate": 1e-06,
      "loss": 0.4558,
      "num_input_tokens_seen": 468196392,
      "step": 8356
    },
    {
      "epoch": 18.610244988864142,
      "loss": 0.3999983072280884,
      "loss_ce": 9.597405733074993e-05,
      "loss_iou": 0.15234375,
      "loss_num": 0.01904296875,
      "loss_xval": 0.400390625,
      "num_input_tokens_seen": 468196392,
      "step": 8356
    },
    {
      "epoch": 18.612472160356347,
      "grad_norm": 12.426268577575684,
      "learning_rate": 1e-06,
      "loss": 0.551,
      "num_input_tokens_seen": 468252860,
      "step": 8357
    },
    {
      "epoch": 18.612472160356347,
      "loss": 0.8520680069923401,
      "loss_ce": 7.824598287697881e-05,
      "loss_iou": 0.326171875,
      "loss_num": 0.039306640625,
      "loss_xval": 0.8515625,
      "num_input_tokens_seen": 468252860,
      "step": 8357
    },
    {
      "epoch": 18.614699331848552,
      "grad_norm": 19.070262908935547,
      "learning_rate": 1e-06,
      "loss": 0.3966,
      "num_input_tokens_seen": 468308700,
      "step": 8358
    },
    {
      "epoch": 18.614699331848552,
      "loss": 0.5038906931877136,
      "loss_ce": 0.00010653713979991153,
      "loss_iou": 0.201171875,
      "loss_num": 0.0203857421875,
      "loss_xval": 0.50390625,
      "num_input_tokens_seen": 468308700,
      "step": 8358
    },
    {
      "epoch": 18.616926503340757,
      "grad_norm": 29.98044776916504,
      "learning_rate": 1e-06,
      "loss": 0.6309,
      "num_input_tokens_seen": 468363020,
      "step": 8359
    },
    {
      "epoch": 18.616926503340757,
      "loss": 0.703240156173706,
      "loss_ce": 0.00011515267397044227,
      "loss_iou": 0.296875,
      "loss_num": 0.021728515625,
      "loss_xval": 0.703125,
      "num_input_tokens_seen": 468363020,
      "step": 8359
    },
    {
      "epoch": 18.619153674832962,
      "grad_norm": 23.00288963317871,
      "learning_rate": 1e-06,
      "loss": 0.5153,
      "num_input_tokens_seen": 468419740,
      "step": 8360
    },
    {
      "epoch": 18.619153674832962,
      "loss": 0.5316950678825378,
      "loss_ce": 7.883799844421446e-05,
      "loss_iou": 0.220703125,
      "loss_num": 0.0179443359375,
      "loss_xval": 0.53125,
      "num_input_tokens_seen": 468419740,
      "step": 8360
    },
    {
      "epoch": 18.621380846325167,
      "grad_norm": 26.114784240722656,
      "learning_rate": 1e-06,
      "loss": 0.4365,
      "num_input_tokens_seen": 468475752,
      "step": 8361
    },
    {
      "epoch": 18.621380846325167,
      "loss": 0.40563201904296875,
      "loss_ce": 0.00011443370021879673,
      "loss_iou": 0.1708984375,
      "loss_num": 0.0126953125,
      "loss_xval": 0.40625,
      "num_input_tokens_seen": 468475752,
      "step": 8361
    },
    {
      "epoch": 18.62360801781737,
      "grad_norm": 25.014408111572266,
      "learning_rate": 1e-06,
      "loss": 0.3513,
      "num_input_tokens_seen": 468529432,
      "step": 8362
    },
    {
      "epoch": 18.62360801781737,
      "loss": 0.3612205982208252,
      "loss_ce": 7.560283120255917e-05,
      "loss_iou": 0.1591796875,
      "loss_num": 0.0084228515625,
      "loss_xval": 0.361328125,
      "num_input_tokens_seen": 468529432,
      "step": 8362
    },
    {
      "epoch": 18.625835189309576,
      "grad_norm": 16.121501922607422,
      "learning_rate": 1e-06,
      "loss": 0.264,
      "num_input_tokens_seen": 468587328,
      "step": 8363
    },
    {
      "epoch": 18.625835189309576,
      "loss": 0.21666020154953003,
      "loss_ce": 9.220950596500188e-05,
      "loss_iou": 0.09326171875,
      "loss_num": 0.00592041015625,
      "loss_xval": 0.216796875,
      "num_input_tokens_seen": 468587328,
      "step": 8363
    },
    {
      "epoch": 18.62806236080178,
      "grad_norm": 17.978118896484375,
      "learning_rate": 1e-06,
      "loss": 0.4242,
      "num_input_tokens_seen": 468644720,
      "step": 8364
    },
    {
      "epoch": 18.62806236080178,
      "loss": 0.6226370334625244,
      "loss_ce": 7.844380888855085e-05,
      "loss_iou": 0.2578125,
      "loss_num": 0.021484375,
      "loss_xval": 0.62109375,
      "num_input_tokens_seen": 468644720,
      "step": 8364
    },
    {
      "epoch": 18.630289532293986,
      "grad_norm": 19.74319076538086,
      "learning_rate": 1e-06,
      "loss": 0.3705,
      "num_input_tokens_seen": 468700752,
      "step": 8365
    },
    {
      "epoch": 18.630289532293986,
      "loss": 0.3942309021949768,
      "loss_ce": 6.587227107957006e-05,
      "loss_iou": 0.14453125,
      "loss_num": 0.0213623046875,
      "loss_xval": 0.39453125,
      "num_input_tokens_seen": 468700752,
      "step": 8365
    },
    {
      "epoch": 18.63251670378619,
      "grad_norm": 17.06964683532715,
      "learning_rate": 1e-06,
      "loss": 0.4573,
      "num_input_tokens_seen": 468755944,
      "step": 8366
    },
    {
      "epoch": 18.63251670378619,
      "loss": 0.28583765029907227,
      "loss_ce": 7.103722600731999e-05,
      "loss_iou": 0.1318359375,
      "loss_num": 0.004302978515625,
      "loss_xval": 0.28515625,
      "num_input_tokens_seen": 468755944,
      "step": 8366
    },
    {
      "epoch": 18.634743875278396,
      "grad_norm": 13.701447486877441,
      "learning_rate": 1e-06,
      "loss": 0.4467,
      "num_input_tokens_seen": 468811364,
      "step": 8367
    },
    {
      "epoch": 18.634743875278396,
      "loss": 0.3969283103942871,
      "loss_ce": 7.772387471050024e-05,
      "loss_iou": 0.1806640625,
      "loss_num": 0.00714111328125,
      "loss_xval": 0.396484375,
      "num_input_tokens_seen": 468811364,
      "step": 8367
    },
    {
      "epoch": 18.6369710467706,
      "grad_norm": 20.248186111450195,
      "learning_rate": 1e-06,
      "loss": 0.3801,
      "num_input_tokens_seen": 468869972,
      "step": 8368
    },
    {
      "epoch": 18.6369710467706,
      "loss": 0.4675615429878235,
      "loss_ce": 9.328986925538629e-05,
      "loss_iou": 0.203125,
      "loss_num": 0.011962890625,
      "loss_xval": 0.466796875,
      "num_input_tokens_seen": 468869972,
      "step": 8368
    },
    {
      "epoch": 18.639198218262806,
      "grad_norm": 25.159513473510742,
      "learning_rate": 1e-06,
      "loss": 0.4678,
      "num_input_tokens_seen": 468926332,
      "step": 8369
    },
    {
      "epoch": 18.639198218262806,
      "loss": 0.600066065788269,
      "loss_ce": 9.047788626048714e-05,
      "loss_iou": 0.26953125,
      "loss_num": 0.01239013671875,
      "loss_xval": 0.6015625,
      "num_input_tokens_seen": 468926332,
      "step": 8369
    },
    {
      "epoch": 18.64142538975501,
      "grad_norm": 14.643291473388672,
      "learning_rate": 1e-06,
      "loss": 0.3295,
      "num_input_tokens_seen": 468983940,
      "step": 8370
    },
    {
      "epoch": 18.64142538975501,
      "loss": 0.25062063336372375,
      "loss_ce": 7.132052996894345e-05,
      "loss_iou": 0.09130859375,
      "loss_num": 0.01361083984375,
      "loss_xval": 0.25,
      "num_input_tokens_seen": 468983940,
      "step": 8370
    },
    {
      "epoch": 18.643652561247215,
      "grad_norm": 15.840713500976562,
      "learning_rate": 1e-06,
      "loss": 0.3142,
      "num_input_tokens_seen": 469040776,
      "step": 8371
    },
    {
      "epoch": 18.643652561247215,
      "loss": 0.3157404363155365,
      "loss_ce": 6.659841164946556e-05,
      "loss_iou": 0.1416015625,
      "loss_num": 0.006439208984375,
      "loss_xval": 0.31640625,
      "num_input_tokens_seen": 469040776,
      "step": 8371
    },
    {
      "epoch": 18.64587973273942,
      "grad_norm": 40.83860397338867,
      "learning_rate": 1e-06,
      "loss": 0.4205,
      "num_input_tokens_seen": 469098340,
      "step": 8372
    },
    {
      "epoch": 18.64587973273942,
      "loss": 0.46250006556510925,
      "loss_ce": 9.772660268936306e-05,
      "loss_iou": 0.2119140625,
      "loss_num": 0.0078125,
      "loss_xval": 0.462890625,
      "num_input_tokens_seen": 469098340,
      "step": 8372
    },
    {
      "epoch": 18.648106904231625,
      "grad_norm": 16.642513275146484,
      "learning_rate": 1e-06,
      "loss": 0.3309,
      "num_input_tokens_seen": 469155160,
      "step": 8373
    },
    {
      "epoch": 18.648106904231625,
      "loss": 0.38674429059028625,
      "loss_ce": 8.656815043650568e-05,
      "loss_iou": 0.177734375,
      "loss_num": 0.006134033203125,
      "loss_xval": 0.38671875,
      "num_input_tokens_seen": 469155160,
      "step": 8373
    },
    {
      "epoch": 18.65033407572383,
      "grad_norm": 16.026294708251953,
      "learning_rate": 1e-06,
      "loss": 0.4616,
      "num_input_tokens_seen": 469211288,
      "step": 8374
    },
    {
      "epoch": 18.65033407572383,
      "loss": 0.40791958570480347,
      "loss_ce": 8.267381781479344e-05,
      "loss_iou": 0.1865234375,
      "loss_num": 0.00677490234375,
      "loss_xval": 0.408203125,
      "num_input_tokens_seen": 469211288,
      "step": 8374
    },
    {
      "epoch": 18.652561247216035,
      "grad_norm": 26.638465881347656,
      "learning_rate": 1e-06,
      "loss": 0.3375,
      "num_input_tokens_seen": 469267800,
      "step": 8375
    },
    {
      "epoch": 18.652561247216035,
      "loss": 0.3067091703414917,
      "loss_ce": 6.856051186332479e-05,
      "loss_iou": 0.13671875,
      "loss_num": 0.00653076171875,
      "loss_xval": 0.306640625,
      "num_input_tokens_seen": 469267800,
      "step": 8375
    },
    {
      "epoch": 18.65478841870824,
      "grad_norm": 12.635597229003906,
      "learning_rate": 1e-06,
      "loss": 0.4255,
      "num_input_tokens_seen": 469325572,
      "step": 8376
    },
    {
      "epoch": 18.65478841870824,
      "loss": 0.5839434266090393,
      "loss_ce": 8.110229828162119e-05,
      "loss_iou": 0.259765625,
      "loss_num": 0.01312255859375,
      "loss_xval": 0.58203125,
      "num_input_tokens_seen": 469325572,
      "step": 8376
    },
    {
      "epoch": 18.657015590200444,
      "grad_norm": 15.969237327575684,
      "learning_rate": 1e-06,
      "loss": 0.3131,
      "num_input_tokens_seen": 469381652,
      "step": 8377
    },
    {
      "epoch": 18.657015590200444,
      "loss": 0.3782818019390106,
      "loss_ce": 7.744554022792727e-05,
      "loss_iou": 0.173828125,
      "loss_num": 0.006195068359375,
      "loss_xval": 0.37890625,
      "num_input_tokens_seen": 469381652,
      "step": 8377
    },
    {
      "epoch": 18.65924276169265,
      "grad_norm": 16.6336727142334,
      "learning_rate": 1e-06,
      "loss": 0.3071,
      "num_input_tokens_seen": 469437196,
      "step": 8378
    },
    {
      "epoch": 18.65924276169265,
      "loss": 0.32802021503448486,
      "loss_ce": 7.833550625946373e-05,
      "loss_iou": 0.126953125,
      "loss_num": 0.01507568359375,
      "loss_xval": 0.328125,
      "num_input_tokens_seen": 469437196,
      "step": 8378
    },
    {
      "epoch": 18.661469933184854,
      "grad_norm": 80.5313491821289,
      "learning_rate": 1e-06,
      "loss": 0.3585,
      "num_input_tokens_seen": 469493580,
      "step": 8379
    },
    {
      "epoch": 18.661469933184854,
      "loss": 0.3814215064048767,
      "loss_ce": 7.384659693343565e-05,
      "loss_iou": 0.169921875,
      "loss_num": 0.0081787109375,
      "loss_xval": 0.380859375,
      "num_input_tokens_seen": 469493580,
      "step": 8379
    },
    {
      "epoch": 18.66369710467706,
      "grad_norm": 16.55701446533203,
      "learning_rate": 1e-06,
      "loss": 0.5018,
      "num_input_tokens_seen": 469551092,
      "step": 8380
    },
    {
      "epoch": 18.66369710467706,
      "loss": 0.4989837408065796,
      "loss_ce": 8.234399138018489e-05,
      "loss_iou": 0.2099609375,
      "loss_num": 0.0159912109375,
      "loss_xval": 0.498046875,
      "num_input_tokens_seen": 469551092,
      "step": 8380
    },
    {
      "epoch": 18.665924276169264,
      "grad_norm": 23.459165573120117,
      "learning_rate": 1e-06,
      "loss": 0.5548,
      "num_input_tokens_seen": 469611232,
      "step": 8381
    },
    {
      "epoch": 18.665924276169264,
      "loss": 0.6222162246704102,
      "loss_ce": 8.488097955705598e-05,
      "loss_iou": 0.2734375,
      "loss_num": 0.0150146484375,
      "loss_xval": 0.62109375,
      "num_input_tokens_seen": 469611232,
      "step": 8381
    },
    {
      "epoch": 18.66815144766147,
      "grad_norm": 17.257272720336914,
      "learning_rate": 1e-06,
      "loss": 0.3304,
      "num_input_tokens_seen": 469664948,
      "step": 8382
    },
    {
      "epoch": 18.66815144766147,
      "loss": 0.2271890938282013,
      "loss_ce": 7.72664716350846e-05,
      "loss_iou": 0.1005859375,
      "loss_num": 0.0052490234375,
      "loss_xval": 0.2275390625,
      "num_input_tokens_seen": 469664948,
      "step": 8382
    },
    {
      "epoch": 18.670378619153674,
      "grad_norm": 20.108436584472656,
      "learning_rate": 1e-06,
      "loss": 0.405,
      "num_input_tokens_seen": 469718928,
      "step": 8383
    },
    {
      "epoch": 18.670378619153674,
      "loss": 0.42782682180404663,
      "loss_ce": 9.24727282836102e-05,
      "loss_iou": 0.189453125,
      "loss_num": 0.0096435546875,
      "loss_xval": 0.427734375,
      "num_input_tokens_seen": 469718928,
      "step": 8383
    },
    {
      "epoch": 18.67260579064588,
      "grad_norm": 17.33344841003418,
      "learning_rate": 1e-06,
      "loss": 0.4995,
      "num_input_tokens_seen": 469777728,
      "step": 8384
    },
    {
      "epoch": 18.67260579064588,
      "loss": 0.449005663394928,
      "loss_ce": 9.207701805280522e-05,
      "loss_iou": 0.1884765625,
      "loss_num": 0.014404296875,
      "loss_xval": 0.44921875,
      "num_input_tokens_seen": 469777728,
      "step": 8384
    },
    {
      "epoch": 18.674832962138083,
      "grad_norm": 19.488948822021484,
      "learning_rate": 1e-06,
      "loss": 0.5887,
      "num_input_tokens_seen": 469832624,
      "step": 8385
    },
    {
      "epoch": 18.674832962138083,
      "loss": 0.3829535245895386,
      "loss_ce": 7.999550871318206e-05,
      "loss_iou": 0.1640625,
      "loss_num": 0.0107421875,
      "loss_xval": 0.3828125,
      "num_input_tokens_seen": 469832624,
      "step": 8385
    },
    {
      "epoch": 18.677060133630288,
      "grad_norm": 13.389050483703613,
      "learning_rate": 1e-06,
      "loss": 0.4056,
      "num_input_tokens_seen": 469891116,
      "step": 8386
    },
    {
      "epoch": 18.677060133630288,
      "loss": 0.35558634996414185,
      "loss_ce": 8.708509267307818e-05,
      "loss_iou": 0.154296875,
      "loss_num": 0.00946044921875,
      "loss_xval": 0.35546875,
      "num_input_tokens_seen": 469891116,
      "step": 8386
    },
    {
      "epoch": 18.679287305122493,
      "grad_norm": 14.323442459106445,
      "learning_rate": 1e-06,
      "loss": 0.3495,
      "num_input_tokens_seen": 469946560,
      "step": 8387
    },
    {
      "epoch": 18.679287305122493,
      "loss": 0.21974313259124756,
      "loss_ce": 7.760837615933269e-05,
      "loss_iou": 0.095703125,
      "loss_num": 0.005584716796875,
      "loss_xval": 0.2197265625,
      "num_input_tokens_seen": 469946560,
      "step": 8387
    },
    {
      "epoch": 18.681514476614698,
      "grad_norm": 28.984878540039062,
      "learning_rate": 1e-06,
      "loss": 0.2812,
      "num_input_tokens_seen": 470002268,
      "step": 8388
    },
    {
      "epoch": 18.681514476614698,
      "loss": 0.24085469543933868,
      "loss_ce": 7.100608490873128e-05,
      "loss_iou": 0.107421875,
      "loss_num": 0.00518798828125,
      "loss_xval": 0.2412109375,
      "num_input_tokens_seen": 470002268,
      "step": 8388
    },
    {
      "epoch": 18.683741648106903,
      "grad_norm": 15.400225639343262,
      "learning_rate": 1e-06,
      "loss": 0.4376,
      "num_input_tokens_seen": 470058180,
      "step": 8389
    },
    {
      "epoch": 18.683741648106903,
      "loss": 0.5878542065620422,
      "loss_ce": 8.563668234273791e-05,
      "loss_iou": 0.25390625,
      "loss_num": 0.016357421875,
      "loss_xval": 0.5859375,
      "num_input_tokens_seen": 470058180,
      "step": 8389
    },
    {
      "epoch": 18.685968819599108,
      "grad_norm": 23.56451988220215,
      "learning_rate": 1e-06,
      "loss": 0.5334,
      "num_input_tokens_seen": 470113364,
      "step": 8390
    },
    {
      "epoch": 18.685968819599108,
      "loss": 0.6698175668716431,
      "loss_ce": 7.881040801294148e-05,
      "loss_iou": 0.2734375,
      "loss_num": 0.024658203125,
      "loss_xval": 0.66796875,
      "num_input_tokens_seen": 470113364,
      "step": 8390
    },
    {
      "epoch": 18.688195991091312,
      "grad_norm": 20.264080047607422,
      "learning_rate": 1e-06,
      "loss": 0.4449,
      "num_input_tokens_seen": 470171080,
      "step": 8391
    },
    {
      "epoch": 18.688195991091312,
      "loss": 0.6437484622001648,
      "loss_ce": 7.168846786953509e-05,
      "loss_iou": 0.2578125,
      "loss_num": 0.025146484375,
      "loss_xval": 0.64453125,
      "num_input_tokens_seen": 470171080,
      "step": 8391
    },
    {
      "epoch": 18.690423162583517,
      "grad_norm": 18.708011627197266,
      "learning_rate": 1e-06,
      "loss": 0.4008,
      "num_input_tokens_seen": 470227556,
      "step": 8392
    },
    {
      "epoch": 18.690423162583517,
      "loss": 0.3530987799167633,
      "loss_ce": 7.14488051016815e-05,
      "loss_iou": 0.1640625,
      "loss_num": 0.005126953125,
      "loss_xval": 0.353515625,
      "num_input_tokens_seen": 470227556,
      "step": 8392
    },
    {
      "epoch": 18.692650334075722,
      "grad_norm": 33.3140754699707,
      "learning_rate": 1e-06,
      "loss": 0.4063,
      "num_input_tokens_seen": 470282536,
      "step": 8393
    },
    {
      "epoch": 18.692650334075722,
      "loss": 0.36571255326271057,
      "loss_ce": 0.0001119803127949126,
      "loss_iou": 0.166015625,
      "loss_num": 0.0067138671875,
      "loss_xval": 0.365234375,
      "num_input_tokens_seen": 470282536,
      "step": 8393
    },
    {
      "epoch": 18.694877505567927,
      "grad_norm": 14.095598220825195,
      "learning_rate": 1e-06,
      "loss": 0.4011,
      "num_input_tokens_seen": 470339848,
      "step": 8394
    },
    {
      "epoch": 18.694877505567927,
      "loss": 0.37506699562072754,
      "loss_ce": 6.698207289446145e-05,
      "loss_iou": 0.166015625,
      "loss_num": 0.00848388671875,
      "loss_xval": 0.375,
      "num_input_tokens_seen": 470339848,
      "step": 8394
    },
    {
      "epoch": 18.697104677060132,
      "grad_norm": 22.200273513793945,
      "learning_rate": 1e-06,
      "loss": 0.4252,
      "num_input_tokens_seen": 470393704,
      "step": 8395
    },
    {
      "epoch": 18.697104677060132,
      "loss": 0.5692760348320007,
      "loss_ce": 6.21723011136055e-05,
      "loss_iou": 0.25390625,
      "loss_num": 0.01177978515625,
      "loss_xval": 0.5703125,
      "num_input_tokens_seen": 470393704,
      "step": 8395
    },
    {
      "epoch": 18.69933184855234,
      "grad_norm": 19.075931549072266,
      "learning_rate": 1e-06,
      "loss": 0.4925,
      "num_input_tokens_seen": 470450140,
      "step": 8396
    },
    {
      "epoch": 18.69933184855234,
      "loss": 0.811962902545929,
      "loss_ce": 7.329651270993054e-05,
      "loss_iou": 0.318359375,
      "loss_num": 0.03466796875,
      "loss_xval": 0.8125,
      "num_input_tokens_seen": 470450140,
      "step": 8396
    },
    {
      "epoch": 18.70155902004454,
      "grad_norm": 15.641639709472656,
      "learning_rate": 1e-06,
      "loss": 0.4462,
      "num_input_tokens_seen": 470505936,
      "step": 8397
    },
    {
      "epoch": 18.70155902004454,
      "loss": 0.47692984342575073,
      "loss_ce": 0.0001232139766216278,
      "loss_iou": 0.1982421875,
      "loss_num": 0.0162353515625,
      "loss_xval": 0.4765625,
      "num_input_tokens_seen": 470505936,
      "step": 8397
    },
    {
      "epoch": 18.70378619153675,
      "grad_norm": 19.46721839904785,
      "learning_rate": 1e-06,
      "loss": 0.3472,
      "num_input_tokens_seen": 470560592,
      "step": 8398
    },
    {
      "epoch": 18.70378619153675,
      "loss": 0.20130589604377747,
      "loss_ce": 7.29848543414846e-05,
      "loss_iou": 0.080078125,
      "loss_num": 0.00830078125,
      "loss_xval": 0.201171875,
      "num_input_tokens_seen": 470560592,
      "step": 8398
    },
    {
      "epoch": 18.706013363028955,
      "grad_norm": 22.994003295898438,
      "learning_rate": 1e-06,
      "loss": 0.3366,
      "num_input_tokens_seen": 470616532,
      "step": 8399
    },
    {
      "epoch": 18.706013363028955,
      "loss": 0.40258532762527466,
      "loss_ce": 0.00011949414329137653,
      "loss_iou": 0.1923828125,
      "loss_num": 0.0036773681640625,
      "loss_xval": 0.40234375,
      "num_input_tokens_seen": 470616532,
      "step": 8399
    },
    {
      "epoch": 18.70824053452116,
      "grad_norm": 15.002291679382324,
      "learning_rate": 1e-06,
      "loss": 0.3363,
      "num_input_tokens_seen": 470673684,
      "step": 8400
    },
    {
      "epoch": 18.70824053452116,
      "loss": 0.3424663245677948,
      "loss_ce": 0.00012012013758067042,
      "loss_iou": 0.138671875,
      "loss_num": 0.0128173828125,
      "loss_xval": 0.341796875,
      "num_input_tokens_seen": 470673684,
      "step": 8400
    },
    {
      "epoch": 18.710467706013365,
      "grad_norm": 17.72799301147461,
      "learning_rate": 1e-06,
      "loss": 0.2912,
      "num_input_tokens_seen": 470732336,
      "step": 8401
    },
    {
      "epoch": 18.710467706013365,
      "loss": 0.3603472411632538,
      "loss_ce": 7.197496597655118e-05,
      "loss_iou": 0.16796875,
      "loss_num": 0.0047607421875,
      "loss_xval": 0.359375,
      "num_input_tokens_seen": 470732336,
      "step": 8401
    },
    {
      "epoch": 18.71269487750557,
      "grad_norm": 19.49424171447754,
      "learning_rate": 1e-06,
      "loss": 0.4016,
      "num_input_tokens_seen": 470789100,
      "step": 8402
    },
    {
      "epoch": 18.71269487750557,
      "loss": 0.5133872628211975,
      "loss_ce": 8.159891876857728e-05,
      "loss_iou": 0.23828125,
      "loss_num": 0.00750732421875,
      "loss_xval": 0.51171875,
      "num_input_tokens_seen": 470789100,
      "step": 8402
    },
    {
      "epoch": 18.714922048997774,
      "grad_norm": 18.081327438354492,
      "learning_rate": 1e-06,
      "loss": 0.4933,
      "num_input_tokens_seen": 470845216,
      "step": 8403
    },
    {
      "epoch": 18.714922048997774,
      "loss": 0.4872537851333618,
      "loss_ce": 7.114657637430355e-05,
      "loss_iou": 0.1875,
      "loss_num": 0.0223388671875,
      "loss_xval": 0.486328125,
      "num_input_tokens_seen": 470845216,
      "step": 8403
    },
    {
      "epoch": 18.71714922048998,
      "grad_norm": 17.4000186920166,
      "learning_rate": 1e-06,
      "loss": 0.3173,
      "num_input_tokens_seen": 470902088,
      "step": 8404
    },
    {
      "epoch": 18.71714922048998,
      "loss": 0.2885332405567169,
      "loss_ce": 8.110511407721788e-05,
      "loss_iou": 0.125,
      "loss_num": 0.00750732421875,
      "loss_xval": 0.2890625,
      "num_input_tokens_seen": 470902088,
      "step": 8404
    },
    {
      "epoch": 18.719376391982184,
      "grad_norm": 17.23605728149414,
      "learning_rate": 1e-06,
      "loss": 0.523,
      "num_input_tokens_seen": 470956888,
      "step": 8405
    },
    {
      "epoch": 18.719376391982184,
      "loss": 0.21879133582115173,
      "loss_ce": 0.0001786548673408106,
      "loss_iou": 0.09326171875,
      "loss_num": 0.00640869140625,
      "loss_xval": 0.21875,
      "num_input_tokens_seen": 470956888,
      "step": 8405
    },
    {
      "epoch": 18.72160356347439,
      "grad_norm": 22.435617446899414,
      "learning_rate": 1e-06,
      "loss": 0.4289,
      "num_input_tokens_seen": 471010248,
      "step": 8406
    },
    {
      "epoch": 18.72160356347439,
      "loss": 0.4550238251686096,
      "loss_ce": 0.00018985335191246122,
      "loss_iou": 0.1875,
      "loss_num": 0.015869140625,
      "loss_xval": 0.455078125,
      "num_input_tokens_seen": 471010248,
      "step": 8406
    },
    {
      "epoch": 18.723830734966594,
      "grad_norm": 16.684246063232422,
      "learning_rate": 1e-06,
      "loss": 0.431,
      "num_input_tokens_seen": 471067932,
      "step": 8407
    },
    {
      "epoch": 18.723830734966594,
      "loss": 0.3406444191932678,
      "loss_ce": 6.82445170241408e-05,
      "loss_iou": 0.1513671875,
      "loss_num": 0.00762939453125,
      "loss_xval": 0.33984375,
      "num_input_tokens_seen": 471067932,
      "step": 8407
    },
    {
      "epoch": 18.7260579064588,
      "grad_norm": 22.295700073242188,
      "learning_rate": 1e-06,
      "loss": 0.3284,
      "num_input_tokens_seen": 471122864,
      "step": 8408
    },
    {
      "epoch": 18.7260579064588,
      "loss": 0.30706241726875305,
      "loss_ce": 8.610197255620733e-05,
      "loss_iou": 0.123046875,
      "loss_num": 0.01226806640625,
      "loss_xval": 0.306640625,
      "num_input_tokens_seen": 471122864,
      "step": 8408
    },
    {
      "epoch": 18.728285077951004,
      "grad_norm": 17.58847999572754,
      "learning_rate": 1e-06,
      "loss": 0.3384,
      "num_input_tokens_seen": 471180772,
      "step": 8409
    },
    {
      "epoch": 18.728285077951004,
      "loss": 0.3137807548046112,
      "loss_ce": 9.056746785063297e-05,
      "loss_iou": 0.1240234375,
      "loss_num": 0.01324462890625,
      "loss_xval": 0.314453125,
      "num_input_tokens_seen": 471180772,
      "step": 8409
    },
    {
      "epoch": 18.73051224944321,
      "grad_norm": 24.846115112304688,
      "learning_rate": 1e-06,
      "loss": 0.3556,
      "num_input_tokens_seen": 471237380,
      "step": 8410
    },
    {
      "epoch": 18.73051224944321,
      "loss": 0.3991147577762604,
      "loss_ce": 6.691899034194648e-05,
      "loss_iou": 0.185546875,
      "loss_num": 0.00555419921875,
      "loss_xval": 0.3984375,
      "num_input_tokens_seen": 471237380,
      "step": 8410
    },
    {
      "epoch": 18.732739420935413,
      "grad_norm": 18.676179885864258,
      "learning_rate": 1e-06,
      "loss": 0.3904,
      "num_input_tokens_seen": 471293424,
      "step": 8411
    },
    {
      "epoch": 18.732739420935413,
      "loss": 0.28310564160346985,
      "loss_ce": 8.561325375922024e-05,
      "loss_iou": 0.1220703125,
      "loss_num": 0.00787353515625,
      "loss_xval": 0.283203125,
      "num_input_tokens_seen": 471293424,
      "step": 8411
    },
    {
      "epoch": 18.734966592427618,
      "grad_norm": 20.800790786743164,
      "learning_rate": 1e-06,
      "loss": 0.4449,
      "num_input_tokens_seen": 471349792,
      "step": 8412
    },
    {
      "epoch": 18.734966592427618,
      "loss": 0.3075633645057678,
      "loss_ce": 6.822836439823732e-05,
      "loss_iou": 0.1416015625,
      "loss_num": 0.004638671875,
      "loss_xval": 0.306640625,
      "num_input_tokens_seen": 471349792,
      "step": 8412
    },
    {
      "epoch": 18.737193763919823,
      "grad_norm": 23.218751907348633,
      "learning_rate": 1e-06,
      "loss": 0.3355,
      "num_input_tokens_seen": 471407216,
      "step": 8413
    },
    {
      "epoch": 18.737193763919823,
      "loss": 0.4214501678943634,
      "loss_ce": 6.347508315229788e-05,
      "loss_iou": 0.1865234375,
      "loss_num": 0.00970458984375,
      "loss_xval": 0.421875,
      "num_input_tokens_seen": 471407216,
      "step": 8413
    },
    {
      "epoch": 18.739420935412028,
      "grad_norm": 18.05634117126465,
      "learning_rate": 1e-06,
      "loss": 0.3551,
      "num_input_tokens_seen": 471464544,
      "step": 8414
    },
    {
      "epoch": 18.739420935412028,
      "loss": 0.388986736536026,
      "loss_ce": 7.072095468174666e-05,
      "loss_iou": 0.1572265625,
      "loss_num": 0.01507568359375,
      "loss_xval": 0.388671875,
      "num_input_tokens_seen": 471464544,
      "step": 8414
    },
    {
      "epoch": 18.741648106904233,
      "grad_norm": 22.533185958862305,
      "learning_rate": 1e-06,
      "loss": 0.4586,
      "num_input_tokens_seen": 471520660,
      "step": 8415
    },
    {
      "epoch": 18.741648106904233,
      "loss": 0.46280378103256226,
      "loss_ce": 8.86318739503622e-05,
      "loss_iou": 0.2060546875,
      "loss_num": 0.0103759765625,
      "loss_xval": 0.462890625,
      "num_input_tokens_seen": 471520660,
      "step": 8415
    },
    {
      "epoch": 18.743875278396438,
      "grad_norm": 20.770368576049805,
      "learning_rate": 1e-06,
      "loss": 0.329,
      "num_input_tokens_seen": 471579084,
      "step": 8416
    },
    {
      "epoch": 18.743875278396438,
      "loss": 0.31473100185394287,
      "loss_ce": 9.477273124502972e-05,
      "loss_iou": 0.1435546875,
      "loss_num": 0.00567626953125,
      "loss_xval": 0.314453125,
      "num_input_tokens_seen": 471579084,
      "step": 8416
    },
    {
      "epoch": 18.746102449888642,
      "grad_norm": 48.90117645263672,
      "learning_rate": 1e-06,
      "loss": 0.2945,
      "num_input_tokens_seen": 471633636,
      "step": 8417
    },
    {
      "epoch": 18.746102449888642,
      "loss": 0.25058692693710327,
      "loss_ce": 6.0503818531287834e-05,
      "loss_iou": 0.11474609375,
      "loss_num": 0.004241943359375,
      "loss_xval": 0.25,
      "num_input_tokens_seen": 471633636,
      "step": 8417
    },
    {
      "epoch": 18.748329621380847,
      "grad_norm": 15.029791831970215,
      "learning_rate": 1e-06,
      "loss": 0.3173,
      "num_input_tokens_seen": 471690856,
      "step": 8418
    },
    {
      "epoch": 18.748329621380847,
      "loss": 0.32165291905403137,
      "loss_ce": 5.868007428944111e-05,
      "loss_iou": 0.14453125,
      "loss_num": 0.006500244140625,
      "loss_xval": 0.322265625,
      "num_input_tokens_seen": 471690856,
      "step": 8418
    },
    {
      "epoch": 18.750556792873052,
      "grad_norm": 18.00300407409668,
      "learning_rate": 1e-06,
      "loss": 0.3783,
      "num_input_tokens_seen": 471749244,
      "step": 8419
    },
    {
      "epoch": 18.750556792873052,
      "loss": 0.325858473777771,
      "loss_ce": 0.0001748991635395214,
      "loss_iou": 0.1455078125,
      "loss_num": 0.00689697265625,
      "loss_xval": 0.326171875,
      "num_input_tokens_seen": 471749244,
      "step": 8419
    },
    {
      "epoch": 18.752783964365257,
      "grad_norm": 12.98948860168457,
      "learning_rate": 1e-06,
      "loss": 0.4111,
      "num_input_tokens_seen": 471804068,
      "step": 8420
    },
    {
      "epoch": 18.752783964365257,
      "loss": 0.4239133596420288,
      "loss_ce": 8.521559357177466e-05,
      "loss_iou": 0.15625,
      "loss_num": 0.0223388671875,
      "loss_xval": 0.423828125,
      "num_input_tokens_seen": 471804068,
      "step": 8420
    },
    {
      "epoch": 18.755011135857462,
      "grad_norm": 13.578185081481934,
      "learning_rate": 1e-06,
      "loss": 0.3306,
      "num_input_tokens_seen": 471859564,
      "step": 8421
    },
    {
      "epoch": 18.755011135857462,
      "loss": 0.285760760307312,
      "loss_ce": 5.517357203643769e-05,
      "loss_iou": 0.11083984375,
      "loss_num": 0.0126953125,
      "loss_xval": 0.28515625,
      "num_input_tokens_seen": 471859564,
      "step": 8421
    },
    {
      "epoch": 18.757238307349667,
      "grad_norm": 13.726969718933105,
      "learning_rate": 1e-06,
      "loss": 0.5174,
      "num_input_tokens_seen": 471918184,
      "step": 8422
    },
    {
      "epoch": 18.757238307349667,
      "loss": 0.4225544333457947,
      "loss_ce": 6.908080104039982e-05,
      "loss_iou": 0.1953125,
      "loss_num": 0.0064697265625,
      "loss_xval": 0.421875,
      "num_input_tokens_seen": 471918184,
      "step": 8422
    },
    {
      "epoch": 18.75946547884187,
      "grad_norm": 27.957782745361328,
      "learning_rate": 1e-06,
      "loss": 0.4461,
      "num_input_tokens_seen": 471972848,
      "step": 8423
    },
    {
      "epoch": 18.75946547884187,
      "loss": 0.5910878777503967,
      "loss_ce": 8.448412700090557e-05,
      "loss_iou": 0.265625,
      "loss_num": 0.0118408203125,
      "loss_xval": 0.58984375,
      "num_input_tokens_seen": 471972848,
      "step": 8423
    },
    {
      "epoch": 18.761692650334076,
      "grad_norm": 20.952011108398438,
      "learning_rate": 1e-06,
      "loss": 0.6627,
      "num_input_tokens_seen": 472030196,
      "step": 8424
    },
    {
      "epoch": 18.761692650334076,
      "loss": 0.3828308880329132,
      "loss_ce": 7.944751996546984e-05,
      "loss_iou": 0.171875,
      "loss_num": 0.007781982421875,
      "loss_xval": 0.3828125,
      "num_input_tokens_seen": 472030196,
      "step": 8424
    },
    {
      "epoch": 18.76391982182628,
      "grad_norm": 17.825992584228516,
      "learning_rate": 1e-06,
      "loss": 0.3981,
      "num_input_tokens_seen": 472086448,
      "step": 8425
    },
    {
      "epoch": 18.76391982182628,
      "loss": 0.3020787835121155,
      "loss_ce": 7.681577699258924e-05,
      "loss_iou": 0.134765625,
      "loss_num": 0.006439208984375,
      "loss_xval": 0.302734375,
      "num_input_tokens_seen": 472086448,
      "step": 8425
    },
    {
      "epoch": 18.766146993318486,
      "grad_norm": 15.120186805725098,
      "learning_rate": 1e-06,
      "loss": 0.6746,
      "num_input_tokens_seen": 472144096,
      "step": 8426
    },
    {
      "epoch": 18.766146993318486,
      "loss": 0.4926231801509857,
      "loss_ce": 6.947731162654236e-05,
      "loss_iou": 0.236328125,
      "loss_num": 0.00396728515625,
      "loss_xval": 0.4921875,
      "num_input_tokens_seen": 472144096,
      "step": 8426
    },
    {
      "epoch": 18.76837416481069,
      "grad_norm": 60.89303207397461,
      "learning_rate": 1e-06,
      "loss": 0.6013,
      "num_input_tokens_seen": 472197860,
      "step": 8427
    },
    {
      "epoch": 18.76837416481069,
      "loss": 0.546029269695282,
      "loss_ce": 6.977089651627466e-05,
      "loss_iou": 0.201171875,
      "loss_num": 0.0286865234375,
      "loss_xval": 0.546875,
      "num_input_tokens_seen": 472197860,
      "step": 8427
    },
    {
      "epoch": 18.770601336302896,
      "grad_norm": 23.926664352416992,
      "learning_rate": 1e-06,
      "loss": 0.4401,
      "num_input_tokens_seen": 472254716,
      "step": 8428
    },
    {
      "epoch": 18.770601336302896,
      "loss": 0.444293737411499,
      "loss_ce": 7.984477269928902e-05,
      "loss_iou": 0.1953125,
      "loss_num": 0.01055908203125,
      "loss_xval": 0.443359375,
      "num_input_tokens_seen": 472254716,
      "step": 8428
    },
    {
      "epoch": 18.7728285077951,
      "grad_norm": 20.06789779663086,
      "learning_rate": 1e-06,
      "loss": 0.6177,
      "num_input_tokens_seen": 472312756,
      "step": 8429
    },
    {
      "epoch": 18.7728285077951,
      "loss": 0.5740514993667603,
      "loss_ce": 7.688780897296965e-05,
      "loss_iou": 0.2236328125,
      "loss_num": 0.0255126953125,
      "loss_xval": 0.57421875,
      "num_input_tokens_seen": 472312756,
      "step": 8429
    },
    {
      "epoch": 18.775055679287306,
      "grad_norm": 32.13956069946289,
      "learning_rate": 1e-06,
      "loss": 0.3333,
      "num_input_tokens_seen": 472367428,
      "step": 8430
    },
    {
      "epoch": 18.775055679287306,
      "loss": 0.2963276207447052,
      "loss_ce": 6.297710933722556e-05,
      "loss_iou": 0.1357421875,
      "loss_num": 0.00482177734375,
      "loss_xval": 0.296875,
      "num_input_tokens_seen": 472367428,
      "step": 8430
    },
    {
      "epoch": 18.77728285077951,
      "grad_norm": 16.847000122070312,
      "learning_rate": 1e-06,
      "loss": 0.329,
      "num_input_tokens_seen": 472424024,
      "step": 8431
    },
    {
      "epoch": 18.77728285077951,
      "loss": 0.3400919437408447,
      "loss_ce": 6.50831643724814e-05,
      "loss_iou": 0.1357421875,
      "loss_num": 0.01348876953125,
      "loss_xval": 0.33984375,
      "num_input_tokens_seen": 472424024,
      "step": 8431
    },
    {
      "epoch": 18.779510022271715,
      "grad_norm": 27.27733039855957,
      "learning_rate": 1e-06,
      "loss": 0.3656,
      "num_input_tokens_seen": 472478816,
      "step": 8432
    },
    {
      "epoch": 18.779510022271715,
      "loss": 0.36883819103240967,
      "loss_ce": 6.378746184054762e-05,
      "loss_iou": 0.162109375,
      "loss_num": 0.0087890625,
      "loss_xval": 0.369140625,
      "num_input_tokens_seen": 472478816,
      "step": 8432
    },
    {
      "epoch": 18.78173719376392,
      "grad_norm": 42.84598922729492,
      "learning_rate": 1e-06,
      "loss": 0.2843,
      "num_input_tokens_seen": 472534048,
      "step": 8433
    },
    {
      "epoch": 18.78173719376392,
      "loss": 0.2845653295516968,
      "loss_ce": 8.045761933317408e-05,
      "loss_iou": 0.12158203125,
      "loss_num": 0.00830078125,
      "loss_xval": 0.28515625,
      "num_input_tokens_seen": 472534048,
      "step": 8433
    },
    {
      "epoch": 18.783964365256125,
      "grad_norm": 13.706661224365234,
      "learning_rate": 1e-06,
      "loss": 0.4059,
      "num_input_tokens_seen": 472588848,
      "step": 8434
    },
    {
      "epoch": 18.783964365256125,
      "loss": 0.5129237771034241,
      "loss_ce": 0.00010639386164257303,
      "loss_iou": 0.2021484375,
      "loss_num": 0.0218505859375,
      "loss_xval": 0.51171875,
      "num_input_tokens_seen": 472588848,
      "step": 8434
    },
    {
      "epoch": 18.78619153674833,
      "grad_norm": 17.942852020263672,
      "learning_rate": 1e-06,
      "loss": 0.3433,
      "num_input_tokens_seen": 472646472,
      "step": 8435
    },
    {
      "epoch": 18.78619153674833,
      "loss": 0.323493629693985,
      "loss_ce": 6.834171654190868e-05,
      "loss_iou": 0.142578125,
      "loss_num": 0.007415771484375,
      "loss_xval": 0.32421875,
      "num_input_tokens_seen": 472646472,
      "step": 8435
    },
    {
      "epoch": 18.788418708240535,
      "grad_norm": 19.692903518676758,
      "learning_rate": 1e-06,
      "loss": 0.5043,
      "num_input_tokens_seen": 472705540,
      "step": 8436
    },
    {
      "epoch": 18.788418708240535,
      "loss": 0.5412243604660034,
      "loss_ce": 8.667838119436055e-05,
      "loss_iou": 0.2265625,
      "loss_num": 0.0174560546875,
      "loss_xval": 0.54296875,
      "num_input_tokens_seen": 472705540,
      "step": 8436
    },
    {
      "epoch": 18.79064587973274,
      "grad_norm": 16.23405647277832,
      "learning_rate": 1e-06,
      "loss": 0.4239,
      "num_input_tokens_seen": 472761744,
      "step": 8437
    },
    {
      "epoch": 18.79064587973274,
      "loss": 0.44612982869148254,
      "loss_ce": 8.490896289004013e-05,
      "loss_iou": 0.1982421875,
      "loss_num": 0.0101318359375,
      "loss_xval": 0.4453125,
      "num_input_tokens_seen": 472761744,
      "step": 8437
    },
    {
      "epoch": 18.792873051224944,
      "grad_norm": 23.637725830078125,
      "learning_rate": 1e-06,
      "loss": 0.3022,
      "num_input_tokens_seen": 472817648,
      "step": 8438
    },
    {
      "epoch": 18.792873051224944,
      "loss": 0.30181801319122314,
      "loss_ce": 6.019488864694722e-05,
      "loss_iou": 0.1298828125,
      "loss_num": 0.00848388671875,
      "loss_xval": 0.30078125,
      "num_input_tokens_seen": 472817648,
      "step": 8438
    },
    {
      "epoch": 18.79510022271715,
      "grad_norm": 16.10801887512207,
      "learning_rate": 1e-06,
      "loss": 0.5896,
      "num_input_tokens_seen": 472873080,
      "step": 8439
    },
    {
      "epoch": 18.79510022271715,
      "loss": 0.7904741168022156,
      "loss_ce": 6.880288128741086e-05,
      "loss_iou": 0.33203125,
      "loss_num": 0.0250244140625,
      "loss_xval": 0.7890625,
      "num_input_tokens_seen": 472873080,
      "step": 8439
    },
    {
      "epoch": 18.797327394209354,
      "grad_norm": 29.073570251464844,
      "learning_rate": 1e-06,
      "loss": 0.2354,
      "num_input_tokens_seen": 472930596,
      "step": 8440
    },
    {
      "epoch": 18.797327394209354,
      "loss": 0.18952415883541107,
      "loss_ce": 7.103992538759485e-05,
      "loss_iou": 0.07666015625,
      "loss_num": 0.00726318359375,
      "loss_xval": 0.189453125,
      "num_input_tokens_seen": 472930596,
      "step": 8440
    },
    {
      "epoch": 18.79955456570156,
      "grad_norm": 19.053442001342773,
      "learning_rate": 1e-06,
      "loss": 0.5383,
      "num_input_tokens_seen": 472984744,
      "step": 8441
    },
    {
      "epoch": 18.79955456570156,
      "loss": 0.39138782024383545,
      "loss_ce": 9.141798363998532e-05,
      "loss_iou": 0.1591796875,
      "loss_num": 0.01470947265625,
      "loss_xval": 0.390625,
      "num_input_tokens_seen": 472984744,
      "step": 8441
    },
    {
      "epoch": 18.801781737193764,
      "grad_norm": 16.559202194213867,
      "learning_rate": 1e-06,
      "loss": 0.4231,
      "num_input_tokens_seen": 473041436,
      "step": 8442
    },
    {
      "epoch": 18.801781737193764,
      "loss": 0.47130343317985535,
      "loss_ce": 8.148739289026707e-05,
      "loss_iou": 0.1630859375,
      "loss_num": 0.0291748046875,
      "loss_xval": 0.470703125,
      "num_input_tokens_seen": 473041436,
      "step": 8442
    },
    {
      "epoch": 18.80400890868597,
      "grad_norm": 25.658018112182617,
      "learning_rate": 1e-06,
      "loss": 0.3696,
      "num_input_tokens_seen": 473093136,
      "step": 8443
    },
    {
      "epoch": 18.80400890868597,
      "loss": 0.31416118144989014,
      "loss_ce": 7.427769742207602e-05,
      "loss_iou": 0.14453125,
      "loss_num": 0.005096435546875,
      "loss_xval": 0.314453125,
      "num_input_tokens_seen": 473093136,
      "step": 8443
    },
    {
      "epoch": 18.806236080178174,
      "grad_norm": 21.92131996154785,
      "learning_rate": 1e-06,
      "loss": 0.3055,
      "num_input_tokens_seen": 473149008,
      "step": 8444
    },
    {
      "epoch": 18.806236080178174,
      "loss": 0.33100777864456177,
      "loss_ce": 7.515733886975795e-05,
      "loss_iou": 0.150390625,
      "loss_num": 0.006195068359375,
      "loss_xval": 0.330078125,
      "num_input_tokens_seen": 473149008,
      "step": 8444
    },
    {
      "epoch": 18.80846325167038,
      "grad_norm": 21.062211990356445,
      "learning_rate": 1e-06,
      "loss": 0.3811,
      "num_input_tokens_seen": 473203320,
      "step": 8445
    },
    {
      "epoch": 18.80846325167038,
      "loss": 0.36316657066345215,
      "loss_ce": 6.842101720394567e-05,
      "loss_iou": 0.1669921875,
      "loss_num": 0.005859375,
      "loss_xval": 0.36328125,
      "num_input_tokens_seen": 473203320,
      "step": 8445
    },
    {
      "epoch": 18.810690423162583,
      "grad_norm": 24.50238800048828,
      "learning_rate": 1e-06,
      "loss": 0.3116,
      "num_input_tokens_seen": 473259040,
      "step": 8446
    },
    {
      "epoch": 18.810690423162583,
      "loss": 0.31474488973617554,
      "loss_ce": 7.813816046109423e-05,
      "loss_iou": 0.1396484375,
      "loss_num": 0.00726318359375,
      "loss_xval": 0.314453125,
      "num_input_tokens_seen": 473259040,
      "step": 8446
    },
    {
      "epoch": 18.812917594654788,
      "grad_norm": 18.850893020629883,
      "learning_rate": 1e-06,
      "loss": 0.5647,
      "num_input_tokens_seen": 473312764,
      "step": 8447
    },
    {
      "epoch": 18.812917594654788,
      "loss": 0.6031259298324585,
      "loss_ce": 9.859535930445418e-05,
      "loss_iou": 0.236328125,
      "loss_num": 0.025634765625,
      "loss_xval": 0.6015625,
      "num_input_tokens_seen": 473312764,
      "step": 8447
    },
    {
      "epoch": 18.815144766146993,
      "grad_norm": 122.17961883544922,
      "learning_rate": 1e-06,
      "loss": 0.3084,
      "num_input_tokens_seen": 473371280,
      "step": 8448
    },
    {
      "epoch": 18.815144766146993,
      "loss": 0.32418200373649597,
      "loss_ce": 8.531348430551589e-05,
      "loss_iou": 0.14453125,
      "loss_num": 0.006805419921875,
      "loss_xval": 0.32421875,
      "num_input_tokens_seen": 473371280,
      "step": 8448
    },
    {
      "epoch": 18.817371937639198,
      "grad_norm": 26.249561309814453,
      "learning_rate": 1e-06,
      "loss": 0.336,
      "num_input_tokens_seen": 473426996,
      "step": 8449
    },
    {
      "epoch": 18.817371937639198,
      "loss": 0.44048649072647095,
      "loss_ce": 5.6789031077641994e-05,
      "loss_iou": 0.1943359375,
      "loss_num": 0.01025390625,
      "loss_xval": 0.44140625,
      "num_input_tokens_seen": 473426996,
      "step": 8449
    },
    {
      "epoch": 18.819599109131403,
      "grad_norm": 13.715222358703613,
      "learning_rate": 1e-06,
      "loss": 0.2667,
      "num_input_tokens_seen": 473482228,
      "step": 8450
    },
    {
      "epoch": 18.819599109131403,
      "loss": 0.30467987060546875,
      "loss_ce": 0.00011442266986705363,
      "loss_iou": 0.1396484375,
      "loss_num": 0.005218505859375,
      "loss_xval": 0.3046875,
      "num_input_tokens_seen": 473482228,
      "step": 8450
    },
    {
      "epoch": 18.821826280623608,
      "grad_norm": 37.36369705200195,
      "learning_rate": 1e-06,
      "loss": 0.4166,
      "num_input_tokens_seen": 473535108,
      "step": 8451
    },
    {
      "epoch": 18.821826280623608,
      "loss": 0.44977593421936035,
      "loss_ce": 6.89039152348414e-05,
      "loss_iou": 0.19140625,
      "loss_num": 0.013427734375,
      "loss_xval": 0.44921875,
      "num_input_tokens_seen": 473535108,
      "step": 8451
    },
    {
      "epoch": 18.824053452115812,
      "grad_norm": 20.77572250366211,
      "learning_rate": 1e-06,
      "loss": 0.2944,
      "num_input_tokens_seen": 473592996,
      "step": 8452
    },
    {
      "epoch": 18.824053452115812,
      "loss": 0.3387192189693451,
      "loss_ce": 9.616896568331867e-05,
      "loss_iou": 0.1416015625,
      "loss_num": 0.010986328125,
      "loss_xval": 0.337890625,
      "num_input_tokens_seen": 473592996,
      "step": 8452
    },
    {
      "epoch": 18.826280623608017,
      "grad_norm": 17.34552001953125,
      "learning_rate": 1e-06,
      "loss": 0.4778,
      "num_input_tokens_seen": 473649796,
      "step": 8453
    },
    {
      "epoch": 18.826280623608017,
      "loss": 0.4614381790161133,
      "loss_ce": 7.342195021919906e-05,
      "loss_iou": 0.1904296875,
      "loss_num": 0.016357421875,
      "loss_xval": 0.4609375,
      "num_input_tokens_seen": 473649796,
      "step": 8453
    },
    {
      "epoch": 18.828507795100222,
      "grad_norm": 12.931282997131348,
      "learning_rate": 1e-06,
      "loss": 0.3362,
      "num_input_tokens_seen": 473706892,
      "step": 8454
    },
    {
      "epoch": 18.828507795100222,
      "loss": 0.2970700263977051,
      "loss_ce": 7.297511911019683e-05,
      "loss_iou": 0.1279296875,
      "loss_num": 0.0081787109375,
      "loss_xval": 0.296875,
      "num_input_tokens_seen": 473706892,
      "step": 8454
    },
    {
      "epoch": 18.830734966592427,
      "grad_norm": 40.271270751953125,
      "learning_rate": 1e-06,
      "loss": 0.5246,
      "num_input_tokens_seen": 473763124,
      "step": 8455
    },
    {
      "epoch": 18.830734966592427,
      "loss": 0.5076436996459961,
      "loss_ce": 7.534350879723206e-05,
      "loss_iou": 0.21875,
      "loss_num": 0.01416015625,
      "loss_xval": 0.5078125,
      "num_input_tokens_seen": 473763124,
      "step": 8455
    },
    {
      "epoch": 18.832962138084632,
      "grad_norm": 49.25373840332031,
      "learning_rate": 1e-06,
      "loss": 0.4866,
      "num_input_tokens_seen": 473818376,
      "step": 8456
    },
    {
      "epoch": 18.832962138084632,
      "loss": 0.4683504104614258,
      "loss_ce": 8.872270700521767e-05,
      "loss_iou": 0.2138671875,
      "loss_num": 0.00823974609375,
      "loss_xval": 0.46875,
      "num_input_tokens_seen": 473818376,
      "step": 8456
    },
    {
      "epoch": 18.835189309576837,
      "grad_norm": 16.492551803588867,
      "learning_rate": 1e-06,
      "loss": 0.5093,
      "num_input_tokens_seen": 473873400,
      "step": 8457
    },
    {
      "epoch": 18.835189309576837,
      "loss": 0.5018711686134338,
      "loss_ce": 8.585450996179134e-05,
      "loss_iou": 0.2314453125,
      "loss_num": 0.00799560546875,
      "loss_xval": 0.5,
      "num_input_tokens_seen": 473873400,
      "step": 8457
    },
    {
      "epoch": 18.83741648106904,
      "grad_norm": 18.73830223083496,
      "learning_rate": 1e-06,
      "loss": 0.4191,
      "num_input_tokens_seen": 473928576,
      "step": 8458
    },
    {
      "epoch": 18.83741648106904,
      "loss": 0.3809613585472107,
      "loss_ce": 0.00010200223186984658,
      "loss_iou": 0.1611328125,
      "loss_num": 0.01153564453125,
      "loss_xval": 0.380859375,
      "num_input_tokens_seen": 473928576,
      "step": 8458
    },
    {
      "epoch": 18.839643652561247,
      "grad_norm": 21.033058166503906,
      "learning_rate": 1e-06,
      "loss": 0.3665,
      "num_input_tokens_seen": 473982784,
      "step": 8459
    },
    {
      "epoch": 18.839643652561247,
      "loss": 0.28056997060775757,
      "loss_ce": 5.241744656814262e-05,
      "loss_iou": 0.123046875,
      "loss_num": 0.00689697265625,
      "loss_xval": 0.28125,
      "num_input_tokens_seen": 473982784,
      "step": 8459
    },
    {
      "epoch": 18.84187082405345,
      "grad_norm": 18.85097312927246,
      "learning_rate": 1e-06,
      "loss": 0.4517,
      "num_input_tokens_seen": 474036980,
      "step": 8460
    },
    {
      "epoch": 18.84187082405345,
      "loss": 0.4336727559566498,
      "loss_ce": 7.89997138781473e-05,
      "loss_iou": 0.1767578125,
      "loss_num": 0.0162353515625,
      "loss_xval": 0.43359375,
      "num_input_tokens_seen": 474036980,
      "step": 8460
    },
    {
      "epoch": 18.844097995545656,
      "grad_norm": 22.325136184692383,
      "learning_rate": 1e-06,
      "loss": 0.3886,
      "num_input_tokens_seen": 474091556,
      "step": 8461
    },
    {
      "epoch": 18.844097995545656,
      "loss": 0.43452924489974976,
      "loss_ce": 0.00014204179751686752,
      "loss_iou": 0.1875,
      "loss_num": 0.01190185546875,
      "loss_xval": 0.43359375,
      "num_input_tokens_seen": 474091556,
      "step": 8461
    },
    {
      "epoch": 18.84632516703786,
      "grad_norm": 14.574409484863281,
      "learning_rate": 1e-06,
      "loss": 0.3643,
      "num_input_tokens_seen": 474147028,
      "step": 8462
    },
    {
      "epoch": 18.84632516703786,
      "loss": 0.22980637848377228,
      "loss_ce": 7.004974031588063e-05,
      "loss_iou": 0.10302734375,
      "loss_num": 0.004608154296875,
      "loss_xval": 0.2294921875,
      "num_input_tokens_seen": 474147028,
      "step": 8462
    },
    {
      "epoch": 18.848552338530066,
      "grad_norm": 31.346830368041992,
      "learning_rate": 1e-06,
      "loss": 0.4338,
      "num_input_tokens_seen": 474203072,
      "step": 8463
    },
    {
      "epoch": 18.848552338530066,
      "loss": 0.4625985324382782,
      "loss_ce": 7.410830585286021e-05,
      "loss_iou": 0.2099609375,
      "loss_num": 0.00860595703125,
      "loss_xval": 0.462890625,
      "num_input_tokens_seen": 474203072,
      "step": 8463
    },
    {
      "epoch": 18.85077951002227,
      "grad_norm": 15.782477378845215,
      "learning_rate": 1e-06,
      "loss": 0.3963,
      "num_input_tokens_seen": 474259464,
      "step": 8464
    },
    {
      "epoch": 18.85077951002227,
      "loss": 0.4502849578857422,
      "loss_ce": 8.967139001470059e-05,
      "loss_iou": 0.1875,
      "loss_num": 0.01519775390625,
      "loss_xval": 0.44921875,
      "num_input_tokens_seen": 474259464,
      "step": 8464
    },
    {
      "epoch": 18.853006681514476,
      "grad_norm": 19.337142944335938,
      "learning_rate": 1e-06,
      "loss": 0.3895,
      "num_input_tokens_seen": 474316776,
      "step": 8465
    },
    {
      "epoch": 18.853006681514476,
      "loss": 0.47631609439849854,
      "loss_ce": 0.00011980824638158083,
      "loss_iou": 0.1923828125,
      "loss_num": 0.01806640625,
      "loss_xval": 0.4765625,
      "num_input_tokens_seen": 474316776,
      "step": 8465
    },
    {
      "epoch": 18.85523385300668,
      "grad_norm": 16.090877532958984,
      "learning_rate": 1e-06,
      "loss": 0.4098,
      "num_input_tokens_seen": 474372780,
      "step": 8466
    },
    {
      "epoch": 18.85523385300668,
      "loss": 0.3744850754737854,
      "loss_ce": 9.543487976770848e-05,
      "loss_iou": 0.169921875,
      "loss_num": 0.006683349609375,
      "loss_xval": 0.375,
      "num_input_tokens_seen": 474372780,
      "step": 8466
    },
    {
      "epoch": 18.857461024498885,
      "grad_norm": 19.818513870239258,
      "learning_rate": 1e-06,
      "loss": 0.4536,
      "num_input_tokens_seen": 474429796,
      "step": 8467
    },
    {
      "epoch": 18.857461024498885,
      "loss": 0.6125023365020752,
      "loss_ce": 7.554069452453405e-05,
      "loss_iou": 0.27734375,
      "loss_num": 0.01123046875,
      "loss_xval": 0.61328125,
      "num_input_tokens_seen": 474429796,
      "step": 8467
    },
    {
      "epoch": 18.85968819599109,
      "grad_norm": 18.213647842407227,
      "learning_rate": 1e-06,
      "loss": 0.3334,
      "num_input_tokens_seen": 474489184,
      "step": 8468
    },
    {
      "epoch": 18.85968819599109,
      "loss": 0.26028338074684143,
      "loss_ce": 7.525723776780069e-05,
      "loss_iou": 0.12109375,
      "loss_num": 0.003570556640625,
      "loss_xval": 0.259765625,
      "num_input_tokens_seen": 474489184,
      "step": 8468
    },
    {
      "epoch": 18.861915367483295,
      "grad_norm": 23.33075523376465,
      "learning_rate": 1e-06,
      "loss": 0.3492,
      "num_input_tokens_seen": 474546056,
      "step": 8469
    },
    {
      "epoch": 18.861915367483295,
      "loss": 0.4007795751094818,
      "loss_ce": 8.379328937735409e-05,
      "loss_iou": 0.1708984375,
      "loss_num": 0.01171875,
      "loss_xval": 0.400390625,
      "num_input_tokens_seen": 474546056,
      "step": 8469
    },
    {
      "epoch": 18.8641425389755,
      "grad_norm": 30.516542434692383,
      "learning_rate": 1e-06,
      "loss": 0.3336,
      "num_input_tokens_seen": 474601092,
      "step": 8470
    },
    {
      "epoch": 18.8641425389755,
      "loss": 0.2731455862522125,
      "loss_ce": 7.429896504618227e-05,
      "loss_iou": 0.126953125,
      "loss_num": 0.003814697265625,
      "loss_xval": 0.2734375,
      "num_input_tokens_seen": 474601092,
      "step": 8470
    },
    {
      "epoch": 18.866369710467705,
      "grad_norm": 17.464170455932617,
      "learning_rate": 1e-06,
      "loss": 0.4543,
      "num_input_tokens_seen": 474657324,
      "step": 8471
    },
    {
      "epoch": 18.866369710467705,
      "loss": 0.48732370138168335,
      "loss_ce": 8.003832772374153e-05,
      "loss_iou": 0.2119140625,
      "loss_num": 0.0125732421875,
      "loss_xval": 0.486328125,
      "num_input_tokens_seen": 474657324,
      "step": 8471
    },
    {
      "epoch": 18.86859688195991,
      "grad_norm": 19.651233673095703,
      "learning_rate": 1e-06,
      "loss": 0.3527,
      "num_input_tokens_seen": 474713160,
      "step": 8472
    },
    {
      "epoch": 18.86859688195991,
      "loss": 0.3799467086791992,
      "loss_ce": 6.391612987499684e-05,
      "loss_iou": 0.1689453125,
      "loss_num": 0.00830078125,
      "loss_xval": 0.37890625,
      "num_input_tokens_seen": 474713160,
      "step": 8472
    },
    {
      "epoch": 18.870824053452115,
      "grad_norm": 41.639102935791016,
      "learning_rate": 1e-06,
      "loss": 0.6776,
      "num_input_tokens_seen": 474771492,
      "step": 8473
    },
    {
      "epoch": 18.870824053452115,
      "loss": 0.556826651096344,
      "loss_ce": 6.396578100975603e-05,
      "loss_iou": 0.259765625,
      "loss_num": 0.007720947265625,
      "loss_xval": 0.55859375,
      "num_input_tokens_seen": 474771492,
      "step": 8473
    },
    {
      "epoch": 18.87305122494432,
      "grad_norm": 20.428213119506836,
      "learning_rate": 1e-06,
      "loss": 0.4747,
      "num_input_tokens_seen": 474825884,
      "step": 8474
    },
    {
      "epoch": 18.87305122494432,
      "loss": 0.5070443153381348,
      "loss_ce": 8.636478742118925e-05,
      "loss_iou": 0.23828125,
      "loss_num": 0.006256103515625,
      "loss_xval": 0.5078125,
      "num_input_tokens_seen": 474825884,
      "step": 8474
    },
    {
      "epoch": 18.875278396436524,
      "grad_norm": 16.10317039489746,
      "learning_rate": 1e-06,
      "loss": 0.2764,
      "num_input_tokens_seen": 474880660,
      "step": 8475
    },
    {
      "epoch": 18.875278396436524,
      "loss": 0.1743912398815155,
      "loss_ce": 7.483765512006357e-05,
      "loss_iou": 0.0791015625,
      "loss_num": 0.0032806396484375,
      "loss_xval": 0.173828125,
      "num_input_tokens_seen": 474880660,
      "step": 8475
    },
    {
      "epoch": 18.87750556792873,
      "grad_norm": 21.885726928710938,
      "learning_rate": 1e-06,
      "loss": 0.5585,
      "num_input_tokens_seen": 474936820,
      "step": 8476
    },
    {
      "epoch": 18.87750556792873,
      "loss": 0.6800100803375244,
      "loss_ce": 7.844208448659629e-05,
      "loss_iou": 0.30078125,
      "loss_num": 0.0157470703125,
      "loss_xval": 0.6796875,
      "num_input_tokens_seen": 474936820,
      "step": 8476
    },
    {
      "epoch": 18.879732739420934,
      "grad_norm": 26.458858489990234,
      "learning_rate": 1e-06,
      "loss": 0.487,
      "num_input_tokens_seen": 474995968,
      "step": 8477
    },
    {
      "epoch": 18.879732739420934,
      "loss": 0.4402797818183899,
      "loss_ce": 9.42522456170991e-05,
      "loss_iou": 0.1865234375,
      "loss_num": 0.01336669921875,
      "loss_xval": 0.439453125,
      "num_input_tokens_seen": 474995968,
      "step": 8477
    },
    {
      "epoch": 18.88195991091314,
      "grad_norm": 21.456289291381836,
      "learning_rate": 1e-06,
      "loss": 0.4094,
      "num_input_tokens_seen": 475052128,
      "step": 8478
    },
    {
      "epoch": 18.88195991091314,
      "loss": 0.3907051384449005,
      "loss_ce": 8.013312617549673e-05,
      "loss_iou": 0.173828125,
      "loss_num": 0.0086669921875,
      "loss_xval": 0.390625,
      "num_input_tokens_seen": 475052128,
      "step": 8478
    },
    {
      "epoch": 18.884187082405344,
      "grad_norm": 19.8070068359375,
      "learning_rate": 1e-06,
      "loss": 0.5119,
      "num_input_tokens_seen": 475109916,
      "step": 8479
    },
    {
      "epoch": 18.884187082405344,
      "loss": 0.33552417159080505,
      "loss_ce": 7.495496538467705e-05,
      "loss_iou": 0.1494140625,
      "loss_num": 0.007537841796875,
      "loss_xval": 0.3359375,
      "num_input_tokens_seen": 475109916,
      "step": 8479
    },
    {
      "epoch": 18.88641425389755,
      "grad_norm": 18.59421157836914,
      "learning_rate": 1e-06,
      "loss": 0.3461,
      "num_input_tokens_seen": 475165660,
      "step": 8480
    },
    {
      "epoch": 18.88641425389755,
      "loss": 0.34626448154449463,
      "loss_ce": 7.30836036382243e-05,
      "loss_iou": 0.162109375,
      "loss_num": 0.00445556640625,
      "loss_xval": 0.345703125,
      "num_input_tokens_seen": 475165660,
      "step": 8480
    },
    {
      "epoch": 18.888641425389753,
      "grad_norm": 35.3235969543457,
      "learning_rate": 1e-06,
      "loss": 0.3462,
      "num_input_tokens_seen": 475223024,
      "step": 8481
    },
    {
      "epoch": 18.888641425389753,
      "loss": 0.3160410225391388,
      "loss_ce": 6.200766074471176e-05,
      "loss_iou": 0.1396484375,
      "loss_num": 0.00726318359375,
      "loss_xval": 0.31640625,
      "num_input_tokens_seen": 475223024,
      "step": 8481
    },
    {
      "epoch": 18.89086859688196,
      "grad_norm": 17.297746658325195,
      "learning_rate": 1e-06,
      "loss": 0.3611,
      "num_input_tokens_seen": 475278500,
      "step": 8482
    },
    {
      "epoch": 18.89086859688196,
      "loss": 0.2839501202106476,
      "loss_ce": 7.561447273474187e-05,
      "loss_iou": 0.134765625,
      "loss_num": 0.00262451171875,
      "loss_xval": 0.283203125,
      "num_input_tokens_seen": 475278500,
      "step": 8482
    },
    {
      "epoch": 18.893095768374163,
      "grad_norm": 26.896717071533203,
      "learning_rate": 1e-06,
      "loss": 0.5017,
      "num_input_tokens_seen": 475332444,
      "step": 8483
    },
    {
      "epoch": 18.893095768374163,
      "loss": 0.4822501540184021,
      "loss_ce": 7.242616266012192e-05,
      "loss_iou": 0.1796875,
      "loss_num": 0.0244140625,
      "loss_xval": 0.482421875,
      "num_input_tokens_seen": 475332444,
      "step": 8483
    },
    {
      "epoch": 18.895322939866368,
      "grad_norm": 27.973371505737305,
      "learning_rate": 1e-06,
      "loss": 0.3581,
      "num_input_tokens_seen": 475389244,
      "step": 8484
    },
    {
      "epoch": 18.895322939866368,
      "loss": 0.31134331226348877,
      "loss_ce": 6.400147685781121e-05,
      "loss_iou": 0.134765625,
      "loss_num": 0.00823974609375,
      "loss_xval": 0.310546875,
      "num_input_tokens_seen": 475389244,
      "step": 8484
    },
    {
      "epoch": 18.897550111358576,
      "grad_norm": 16.82830810546875,
      "learning_rate": 1e-06,
      "loss": 0.3729,
      "num_input_tokens_seen": 475446276,
      "step": 8485
    },
    {
      "epoch": 18.897550111358576,
      "loss": 0.3148788809776306,
      "loss_ce": 5.956278619123623e-05,
      "loss_iou": 0.142578125,
      "loss_num": 0.006072998046875,
      "loss_xval": 0.314453125,
      "num_input_tokens_seen": 475446276,
      "step": 8485
    },
    {
      "epoch": 18.899777282850778,
      "grad_norm": 16.516233444213867,
      "learning_rate": 1e-06,
      "loss": 0.2496,
      "num_input_tokens_seen": 475501428,
      "step": 8486
    },
    {
      "epoch": 18.899777282850778,
      "loss": 0.33766478300094604,
      "loss_ce": 7.93259241618216e-05,
      "loss_iou": 0.1591796875,
      "loss_num": 0.00396728515625,
      "loss_xval": 0.337890625,
      "num_input_tokens_seen": 475501428,
      "step": 8486
    },
    {
      "epoch": 18.902004454342986,
      "grad_norm": 19.306852340698242,
      "learning_rate": 1e-06,
      "loss": 0.3633,
      "num_input_tokens_seen": 475557696,
      "step": 8487
    },
    {
      "epoch": 18.902004454342986,
      "loss": 0.2621913552284241,
      "loss_ce": 7.588176958961412e-05,
      "loss_iou": 0.11474609375,
      "loss_num": 0.006439208984375,
      "loss_xval": 0.26171875,
      "num_input_tokens_seen": 475557696,
      "step": 8487
    },
    {
      "epoch": 18.90423162583519,
      "grad_norm": 17.466773986816406,
      "learning_rate": 1e-06,
      "loss": 0.346,
      "num_input_tokens_seen": 475614528,
      "step": 8488
    },
    {
      "epoch": 18.90423162583519,
      "loss": 0.5012964010238647,
      "loss_ce": 7.571464811917394e-05,
      "loss_iou": 0.2109375,
      "loss_num": 0.0159912109375,
      "loss_xval": 0.5,
      "num_input_tokens_seen": 475614528,
      "step": 8488
    },
    {
      "epoch": 18.906458797327396,
      "grad_norm": 16.38589096069336,
      "learning_rate": 1e-06,
      "loss": 0.4567,
      "num_input_tokens_seen": 475671076,
      "step": 8489
    },
    {
      "epoch": 18.906458797327396,
      "loss": 0.43280452489852905,
      "loss_ce": 6.528291123686358e-05,
      "loss_iou": 0.1982421875,
      "loss_num": 0.007110595703125,
      "loss_xval": 0.43359375,
      "num_input_tokens_seen": 475671076,
      "step": 8489
    },
    {
      "epoch": 18.9086859688196,
      "grad_norm": 22.471385955810547,
      "learning_rate": 1e-06,
      "loss": 0.6548,
      "num_input_tokens_seen": 475723864,
      "step": 8490
    },
    {
      "epoch": 18.9086859688196,
      "loss": 0.7026335000991821,
      "loss_ce": 0.0003630055289249867,
      "loss_iou": 0.328125,
      "loss_num": 0.00872802734375,
      "loss_xval": 0.703125,
      "num_input_tokens_seen": 475723864,
      "step": 8490
    },
    {
      "epoch": 18.910913140311806,
      "grad_norm": 16.447473526000977,
      "learning_rate": 1e-06,
      "loss": 0.3351,
      "num_input_tokens_seen": 475779860,
      "step": 8491
    },
    {
      "epoch": 18.910913140311806,
      "loss": 0.29600396752357483,
      "loss_ce": 7.501787331420928e-05,
      "loss_iou": 0.12890625,
      "loss_num": 0.007537841796875,
      "loss_xval": 0.296875,
      "num_input_tokens_seen": 475779860,
      "step": 8491
    },
    {
      "epoch": 18.91314031180401,
      "grad_norm": 16.438947677612305,
      "learning_rate": 1e-06,
      "loss": 0.4427,
      "num_input_tokens_seen": 475835912,
      "step": 8492
    },
    {
      "epoch": 18.91314031180401,
      "loss": 0.4535049796104431,
      "loss_ce": 7.47992453398183e-05,
      "loss_iou": 0.2001953125,
      "loss_num": 0.01055908203125,
      "loss_xval": 0.453125,
      "num_input_tokens_seen": 475835912,
      "step": 8492
    },
    {
      "epoch": 18.915367483296215,
      "grad_norm": 16.084348678588867,
      "learning_rate": 1e-06,
      "loss": 0.7689,
      "num_input_tokens_seen": 475892356,
      "step": 8493
    },
    {
      "epoch": 18.915367483296215,
      "loss": 0.9672378301620483,
      "loss_ce": 7.476539758499712e-05,
      "loss_iou": 0.359375,
      "loss_num": 0.04931640625,
      "loss_xval": 0.96875,
      "num_input_tokens_seen": 475892356,
      "step": 8493
    },
    {
      "epoch": 18.91759465478842,
      "grad_norm": 21.32638168334961,
      "learning_rate": 1e-06,
      "loss": 0.4226,
      "num_input_tokens_seen": 475948980,
      "step": 8494
    },
    {
      "epoch": 18.91759465478842,
      "loss": 0.461247056722641,
      "loss_ce": 0.0001875017478596419,
      "loss_iou": 0.2080078125,
      "loss_num": 0.00921630859375,
      "loss_xval": 0.4609375,
      "num_input_tokens_seen": 475948980,
      "step": 8494
    },
    {
      "epoch": 18.919821826280625,
      "grad_norm": 36.91980743408203,
      "learning_rate": 1e-06,
      "loss": 0.4261,
      "num_input_tokens_seen": 476005300,
      "step": 8495
    },
    {
      "epoch": 18.919821826280625,
      "loss": 0.39789801836013794,
      "loss_ce": 7.085979450494051e-05,
      "loss_iou": 0.1826171875,
      "loss_num": 0.006683349609375,
      "loss_xval": 0.3984375,
      "num_input_tokens_seen": 476005300,
      "step": 8495
    },
    {
      "epoch": 18.92204899777283,
      "grad_norm": 28.42441749572754,
      "learning_rate": 1e-06,
      "loss": 0.3562,
      "num_input_tokens_seen": 476059068,
      "step": 8496
    },
    {
      "epoch": 18.92204899777283,
      "loss": 0.35445064306259155,
      "loss_ce": 8.05156523711048e-05,
      "loss_iou": 0.1630859375,
      "loss_num": 0.00555419921875,
      "loss_xval": 0.353515625,
      "num_input_tokens_seen": 476059068,
      "step": 8496
    },
    {
      "epoch": 18.924276169265035,
      "grad_norm": 15.65682601928711,
      "learning_rate": 1e-06,
      "loss": 0.3527,
      "num_input_tokens_seen": 476116536,
      "step": 8497
    },
    {
      "epoch": 18.924276169265035,
      "loss": 0.347369909286499,
      "loss_ce": 7.986459240783006e-05,
      "loss_iou": 0.15234375,
      "loss_num": 0.00848388671875,
      "loss_xval": 0.34765625,
      "num_input_tokens_seen": 476116536,
      "step": 8497
    },
    {
      "epoch": 18.92650334075724,
      "grad_norm": 14.354292869567871,
      "learning_rate": 1e-06,
      "loss": 0.5194,
      "num_input_tokens_seen": 476172064,
      "step": 8498
    },
    {
      "epoch": 18.92650334075724,
      "loss": 0.49591702222824097,
      "loss_ce": 6.739451055182144e-05,
      "loss_iou": 0.228515625,
      "loss_num": 0.007598876953125,
      "loss_xval": 0.49609375,
      "num_input_tokens_seen": 476172064,
      "step": 8498
    },
    {
      "epoch": 18.928730512249444,
      "grad_norm": 26.137008666992188,
      "learning_rate": 1e-06,
      "loss": 0.4098,
      "num_input_tokens_seen": 476226344,
      "step": 8499
    },
    {
      "epoch": 18.928730512249444,
      "loss": 0.36395373940467834,
      "loss_ce": 6.214019958861172e-05,
      "loss_iou": 0.1591796875,
      "loss_num": 0.00921630859375,
      "loss_xval": 0.36328125,
      "num_input_tokens_seen": 476226344,
      "step": 8499
    },
    {
      "epoch": 18.93095768374165,
      "grad_norm": 15.024909973144531,
      "learning_rate": 1e-06,
      "loss": 0.2597,
      "num_input_tokens_seen": 476279788,
      "step": 8500
    },
    {
      "epoch": 18.93095768374165,
      "eval_seeclick_web_CIoU": 0.5848062932491302,
      "eval_seeclick_web_GIoU": 0.583672821521759,
      "eval_seeclick_web_IoU": 0.6038743853569031,
      "eval_seeclick_web_MAE_all": 0.015243555419147015,
      "eval_seeclick_web_MAE_h": 0.007093099411576986,
      "eval_seeclick_web_MAE_w": 0.015268665738403797,
      "eval_seeclick_web_MAE_x_boxes": 0.00826921034604311,
      "eval_seeclick_web_MAE_y_boxes": 0.02124078758060932,
      "eval_seeclick_web_inside_bbox": 0.9010416567325592,
      "eval_seeclick_web_loss": 0.9087226986885071,
      "eval_seeclick_web_loss_ce": 0.00012506780331023037,
      "eval_seeclick_web_loss_iou": 0.42181396484375,
      "eval_seeclick_web_loss_num": 0.012174606323242188,
      "eval_seeclick_web_loss_xval": 0.9036865234375,
      "eval_seeclick_web_runtime": 22.9412,
      "eval_seeclick_web_samples_per_second": 2.179,
      "eval_seeclick_web_steps_per_second": 0.087,
      "num_input_tokens_seen": 476279788,
      "step": 8500
    },
    {
      "epoch": 18.93095768374165,
      "eval_icons_CIoU": 0.2560933604836464,
      "eval_icons_GIoU": 0.290867879986763,
      "eval_icons_IoU": 0.3409384936094284,
      "eval_icons_MAE_all": 0.06184186786413193,
      "eval_icons_MAE_h": 0.029924143571406603,
      "eval_icons_MAE_w": 0.06795705109834671,
      "eval_icons_MAE_x_boxes": 0.06149038299918175,
      "eval_icons_MAE_y_boxes": 0.037362379021942616,
      "eval_icons_inside_bbox": 0.59375,
      "eval_icons_loss": 1.7201489210128784,
      "eval_icons_loss_ce": 0.0001441572530893609,
      "eval_icons_loss_iou": 0.67724609375,
      "eval_icons_loss_num": 0.05980682373046875,
      "eval_icons_loss_xval": 1.65380859375,
      "eval_icons_runtime": 22.6756,
      "eval_icons_samples_per_second": 2.205,
      "eval_icons_steps_per_second": 0.088,
      "num_input_tokens_seen": 476279788,
      "step": 8500
    },
    {
      "epoch": 18.93095768374165,
      "eval_screenspot_CIoU": 0.3761156400044759,
      "eval_screenspot_GIoU": 0.3939984142780304,
      "eval_screenspot_IoU": 0.44694022337595624,
      "eval_screenspot_MAE_all": 0.055543094873428345,
      "eval_screenspot_MAE_h": 0.039369805405537285,
      "eval_screenspot_MAE_w": 0.05557269603013992,
      "eval_screenspot_MAE_x_boxes": 0.06785313226282597,
      "eval_screenspot_MAE_y_boxes": 0.04004095122218132,
      "eval_screenspot_inside_bbox": 0.7145833373069763,
      "eval_screenspot_loss": 1.5496809482574463,
      "eval_screenspot_loss_ce": 0.00017849090121065578,
      "eval_screenspot_loss_iou": 0.644775390625,
      "eval_screenspot_loss_num": 0.0635693868001302,
      "eval_screenspot_loss_xval": 1.6062825520833333,
      "eval_screenspot_runtime": 38.5423,
      "eval_screenspot_samples_per_second": 2.309,
      "eval_screenspot_steps_per_second": 0.078,
      "num_input_tokens_seen": 476279788,
      "step": 8500
    },
    {
      "epoch": 18.93095768374165,
      "eval_compot_CIoU": 0.3419719487428665,
      "eval_compot_GIoU": 0.3519115000963211,
      "eval_compot_IoU": 0.40101131796836853,
      "eval_compot_MAE_all": 0.018997764214873314,
      "eval_compot_MAE_h": 0.011835724115371704,
      "eval_compot_MAE_w": 0.02075411193072796,
      "eval_compot_MAE_x_boxes": 0.03001493401825428,
      "eval_compot_MAE_y_boxes": 0.006946815177798271,
      "eval_compot_inside_bbox": 0.6302083432674408,
      "eval_compot_loss": 1.3982415199279785,
      "eval_compot_loss_ce": 0.00011685657591442578,
      "eval_compot_loss_iou": 0.646240234375,
      "eval_compot_loss_num": 0.017612457275390625,
      "eval_compot_loss_xval": 1.3798828125,
      "eval_compot_runtime": 24.3361,
      "eval_compot_samples_per_second": 2.055,
      "eval_compot_steps_per_second": 0.082,
      "num_input_tokens_seen": 476279788,
      "step": 8500
    },
    {
      "epoch": 18.93095768374165,
      "eval_custom_ui_val_CIoU": 0.47164858794874615,
      "eval_custom_ui_val_GIoU": 0.47631729145844776,
      "eval_custom_ui_val_IoU": 0.5336444907718234,
      "eval_custom_ui_val_MAE_all": 0.026907082750565477,
      "eval_custom_ui_val_MAE_h": 0.013476735032680962,
      "eval_custom_ui_val_MAE_w": 0.03637263929057452,
      "eval_custom_ui_val_MAE_x_boxes": 0.03278244765371912,
      "eval_custom_ui_val_MAE_y_boxes": 0.013371242494839761,
      "eval_custom_ui_val_inside_bbox": 0.7685185207260979,
      "eval_custom_ui_val_loss": 1.1731735467910767,
      "eval_custom_ui_val_loss_ce": 0.0001309802391915582,
      "eval_custom_ui_val_loss_iou": 0.5032552083333334,
      "eval_custom_ui_val_loss_num": 0.023616578843858507,
      "eval_custom_ui_val_loss_xval": 1.1244574652777777,
      "eval_custom_ui_val_runtime": 72.5876,
      "eval_custom_ui_val_samples_per_second": 3.651,
      "eval_custom_ui_val_steps_per_second": 0.124,
      "num_input_tokens_seen": 476279788,
      "step": 8500
    },
    {
      "epoch": 18.93095768374165,
      "loss": 0.8257832527160645,
      "loss_ce": 9.972714178729802e-05,
      "loss_iou": 0.37109375,
      "loss_num": 0.016357421875,
      "loss_xval": 0.82421875,
      "num_input_tokens_seen": 476279788,
      "step": 8500
    },
    {
      "epoch": 18.933184855233854,
      "grad_norm": 16.40049934387207,
      "learning_rate": 1e-06,
      "loss": 0.3533,
      "num_input_tokens_seen": 476335896,
      "step": 8501
    },
    {
      "epoch": 18.933184855233854,
      "loss": 0.35964030027389526,
      "loss_ce": 0.00011269970855209976,
      "loss_iou": 0.1611328125,
      "loss_num": 0.007568359375,
      "loss_xval": 0.359375,
      "num_input_tokens_seen": 476335896,
      "step": 8501
    },
    {
      "epoch": 18.93541202672606,
      "grad_norm": 15.26424503326416,
      "learning_rate": 1e-06,
      "loss": 0.2743,
      "num_input_tokens_seen": 476390700,
      "step": 8502
    },
    {
      "epoch": 18.93541202672606,
      "loss": 0.33460482954978943,
      "loss_ce": 7.113382162060589e-05,
      "loss_iou": 0.146484375,
      "loss_num": 0.00823974609375,
      "loss_xval": 0.333984375,
      "num_input_tokens_seen": 476390700,
      "step": 8502
    },
    {
      "epoch": 18.937639198218264,
      "grad_norm": 19.776226043701172,
      "learning_rate": 1e-06,
      "loss": 0.2961,
      "num_input_tokens_seen": 476444976,
      "step": 8503
    },
    {
      "epoch": 18.937639198218264,
      "loss": 0.34564700722694397,
      "loss_ce": 6.596426828764379e-05,
      "loss_iou": 0.15234375,
      "loss_num": 0.0084228515625,
      "loss_xval": 0.345703125,
      "num_input_tokens_seen": 476444976,
      "step": 8503
    },
    {
      "epoch": 18.93986636971047,
      "grad_norm": 13.866682052612305,
      "learning_rate": 1e-06,
      "loss": 0.3222,
      "num_input_tokens_seen": 476500844,
      "step": 8504
    },
    {
      "epoch": 18.93986636971047,
      "loss": 0.31439658999443054,
      "loss_ce": 6.556001608259976e-05,
      "loss_iou": 0.1240234375,
      "loss_num": 0.01312255859375,
      "loss_xval": 0.314453125,
      "num_input_tokens_seen": 476500844,
      "step": 8504
    },
    {
      "epoch": 18.942093541202674,
      "grad_norm": 19.28264045715332,
      "learning_rate": 1e-06,
      "loss": 0.3066,
      "num_input_tokens_seen": 476556968,
      "step": 8505
    },
    {
      "epoch": 18.942093541202674,
      "loss": 0.3727559447288513,
      "loss_ce": 7.529569120379165e-05,
      "loss_iou": 0.166015625,
      "loss_num": 0.00823974609375,
      "loss_xval": 0.373046875,
      "num_input_tokens_seen": 476556968,
      "step": 8505
    },
    {
      "epoch": 18.94432071269488,
      "grad_norm": 20.56502342224121,
      "learning_rate": 1e-06,
      "loss": 0.3223,
      "num_input_tokens_seen": 476614640,
      "step": 8506
    },
    {
      "epoch": 18.94432071269488,
      "loss": 0.41475239396095276,
      "loss_ce": 7.952825399115682e-05,
      "loss_iou": 0.1787109375,
      "loss_num": 0.01141357421875,
      "loss_xval": 0.4140625,
      "num_input_tokens_seen": 476614640,
      "step": 8506
    },
    {
      "epoch": 18.946547884187083,
      "grad_norm": 34.558292388916016,
      "learning_rate": 1e-06,
      "loss": 0.5155,
      "num_input_tokens_seen": 476670132,
      "step": 8507
    },
    {
      "epoch": 18.946547884187083,
      "loss": 0.4488101005554199,
      "loss_ce": 7.96294043539092e-05,
      "loss_iou": 0.197265625,
      "loss_num": 0.0107421875,
      "loss_xval": 0.44921875,
      "num_input_tokens_seen": 476670132,
      "step": 8507
    },
    {
      "epoch": 18.948775055679288,
      "grad_norm": 18.674837112426758,
      "learning_rate": 1e-06,
      "loss": 0.4771,
      "num_input_tokens_seen": 476726448,
      "step": 8508
    },
    {
      "epoch": 18.948775055679288,
      "loss": 0.5774888396263123,
      "loss_ce": 9.624622180126607e-05,
      "loss_iou": 0.267578125,
      "loss_num": 0.00823974609375,
      "loss_xval": 0.578125,
      "num_input_tokens_seen": 476726448,
      "step": 8508
    },
    {
      "epoch": 18.951002227171493,
      "grad_norm": 15.41340160369873,
      "learning_rate": 1e-06,
      "loss": 0.4142,
      "num_input_tokens_seen": 476784916,
      "step": 8509
    },
    {
      "epoch": 18.951002227171493,
      "loss": 0.4316607713699341,
      "loss_ce": 8.12113648862578e-05,
      "loss_iou": 0.1953125,
      "loss_num": 0.00811767578125,
      "loss_xval": 0.431640625,
      "num_input_tokens_seen": 476784916,
      "step": 8509
    },
    {
      "epoch": 18.953229398663698,
      "grad_norm": 14.191808700561523,
      "learning_rate": 1e-06,
      "loss": 0.422,
      "num_input_tokens_seen": 476843120,
      "step": 8510
    },
    {
      "epoch": 18.953229398663698,
      "loss": 0.4266318678855896,
      "loss_ce": 0.00011820608051493764,
      "loss_iou": 0.169921875,
      "loss_num": 0.0174560546875,
      "loss_xval": 0.42578125,
      "num_input_tokens_seen": 476843120,
      "step": 8510
    },
    {
      "epoch": 18.955456570155903,
      "grad_norm": 16.592500686645508,
      "learning_rate": 1e-06,
      "loss": 0.4383,
      "num_input_tokens_seen": 476899824,
      "step": 8511
    },
    {
      "epoch": 18.955456570155903,
      "loss": 0.40235602855682373,
      "loss_ce": 7.331261440413073e-05,
      "loss_iou": 0.166015625,
      "loss_num": 0.01385498046875,
      "loss_xval": 0.40234375,
      "num_input_tokens_seen": 476899824,
      "step": 8511
    },
    {
      "epoch": 18.957683741648108,
      "grad_norm": 12.043611526489258,
      "learning_rate": 1e-06,
      "loss": 0.3217,
      "num_input_tokens_seen": 476956480,
      "step": 8512
    },
    {
      "epoch": 18.957683741648108,
      "loss": 0.3974224627017975,
      "loss_ce": 8.359744242625311e-05,
      "loss_iou": 0.1669921875,
      "loss_num": 0.01275634765625,
      "loss_xval": 0.396484375,
      "num_input_tokens_seen": 476956480,
      "step": 8512
    },
    {
      "epoch": 18.959910913140313,
      "grad_norm": 20.1372013092041,
      "learning_rate": 1e-06,
      "loss": 0.3755,
      "num_input_tokens_seen": 477011216,
      "step": 8513
    },
    {
      "epoch": 18.959910913140313,
      "loss": 0.3341856598854065,
      "loss_ce": 7.918624032754451e-05,
      "loss_iou": 0.142578125,
      "loss_num": 0.0096435546875,
      "loss_xval": 0.333984375,
      "num_input_tokens_seen": 477011216,
      "step": 8513
    },
    {
      "epoch": 18.962138084632517,
      "grad_norm": 23.11919593811035,
      "learning_rate": 1e-06,
      "loss": 0.3368,
      "num_input_tokens_seen": 477068100,
      "step": 8514
    },
    {
      "epoch": 18.962138084632517,
      "loss": 0.4352511167526245,
      "loss_ce": 7.042582728900015e-05,
      "loss_iou": 0.16015625,
      "loss_num": 0.02294921875,
      "loss_xval": 0.435546875,
      "num_input_tokens_seen": 477068100,
      "step": 8514
    },
    {
      "epoch": 18.964365256124722,
      "grad_norm": 15.498306274414062,
      "learning_rate": 1e-06,
      "loss": 0.4073,
      "num_input_tokens_seen": 477124952,
      "step": 8515
    },
    {
      "epoch": 18.964365256124722,
      "loss": 0.4004598557949066,
      "loss_ce": 6.92239118507132e-05,
      "loss_iou": 0.17578125,
      "loss_num": 0.00982666015625,
      "loss_xval": 0.400390625,
      "num_input_tokens_seen": 477124952,
      "step": 8515
    },
    {
      "epoch": 18.966592427616927,
      "grad_norm": 16.32062530517578,
      "learning_rate": 1e-06,
      "loss": 0.3115,
      "num_input_tokens_seen": 477178756,
      "step": 8516
    },
    {
      "epoch": 18.966592427616927,
      "loss": 0.38423728942871094,
      "loss_ce": 8.200977754313499e-05,
      "loss_iou": 0.16796875,
      "loss_num": 0.00958251953125,
      "loss_xval": 0.384765625,
      "num_input_tokens_seen": 477178756,
      "step": 8516
    },
    {
      "epoch": 18.968819599109132,
      "grad_norm": 16.013669967651367,
      "learning_rate": 1e-06,
      "loss": 0.4225,
      "num_input_tokens_seen": 477234552,
      "step": 8517
    },
    {
      "epoch": 18.968819599109132,
      "loss": 0.45508986711502075,
      "loss_ce": 7.279004785232246e-05,
      "loss_iou": 0.1943359375,
      "loss_num": 0.0133056640625,
      "loss_xval": 0.455078125,
      "num_input_tokens_seen": 477234552,
      "step": 8517
    },
    {
      "epoch": 18.971046770601337,
      "grad_norm": 15.03101921081543,
      "learning_rate": 1e-06,
      "loss": 0.5013,
      "num_input_tokens_seen": 477291364,
      "step": 8518
    },
    {
      "epoch": 18.971046770601337,
      "loss": 0.4453286826610565,
      "loss_ce": 7.720896974205971e-05,
      "loss_iou": 0.2041015625,
      "loss_num": 0.007476806640625,
      "loss_xval": 0.4453125,
      "num_input_tokens_seen": 477291364,
      "step": 8518
    },
    {
      "epoch": 18.97327394209354,
      "grad_norm": 16.71440887451172,
      "learning_rate": 1e-06,
      "loss": 0.3699,
      "num_input_tokens_seen": 477346348,
      "step": 8519
    },
    {
      "epoch": 18.97327394209354,
      "loss": 0.3506791889667511,
      "loss_ce": 9.326588042313233e-05,
      "loss_iou": 0.16015625,
      "loss_num": 0.00604248046875,
      "loss_xval": 0.3515625,
      "num_input_tokens_seen": 477346348,
      "step": 8519
    },
    {
      "epoch": 18.975501113585747,
      "grad_norm": 29.49169158935547,
      "learning_rate": 1e-06,
      "loss": 0.4555,
      "num_input_tokens_seen": 477402184,
      "step": 8520
    },
    {
      "epoch": 18.975501113585747,
      "loss": 0.31228724122047424,
      "loss_ce": 6.190109706949443e-05,
      "loss_iou": 0.1298828125,
      "loss_num": 0.01043701171875,
      "loss_xval": 0.3125,
      "num_input_tokens_seen": 477402184,
      "step": 8520
    },
    {
      "epoch": 18.97772828507795,
      "grad_norm": 17.541460037231445,
      "learning_rate": 1e-06,
      "loss": 0.4045,
      "num_input_tokens_seen": 477458060,
      "step": 8521
    },
    {
      "epoch": 18.97772828507795,
      "loss": 0.30591824650764465,
      "loss_ce": 7.108383579179645e-05,
      "loss_iou": 0.1396484375,
      "loss_num": 0.005279541015625,
      "loss_xval": 0.306640625,
      "num_input_tokens_seen": 477458060,
      "step": 8521
    },
    {
      "epoch": 18.979955456570156,
      "grad_norm": 17.852941513061523,
      "learning_rate": 1e-06,
      "loss": 0.5856,
      "num_input_tokens_seen": 477511612,
      "step": 8522
    },
    {
      "epoch": 18.979955456570156,
      "loss": 0.4428853988647461,
      "loss_ce": 7.535393524449319e-05,
      "loss_iou": 0.1806640625,
      "loss_num": 0.0164794921875,
      "loss_xval": 0.443359375,
      "num_input_tokens_seen": 477511612,
      "step": 8522
    },
    {
      "epoch": 18.98218262806236,
      "grad_norm": 18.045700073242188,
      "learning_rate": 1e-06,
      "loss": 0.3042,
      "num_input_tokens_seen": 477566404,
      "step": 8523
    },
    {
      "epoch": 18.98218262806236,
      "loss": 0.3434655964374542,
      "loss_ce": 8.181348675861955e-05,
      "loss_iou": 0.1396484375,
      "loss_num": 0.01300048828125,
      "loss_xval": 0.34375,
      "num_input_tokens_seen": 477566404,
      "step": 8523
    },
    {
      "epoch": 18.984409799554566,
      "grad_norm": 21.54753875732422,
      "learning_rate": 1e-06,
      "loss": 0.4359,
      "num_input_tokens_seen": 477623140,
      "step": 8524
    },
    {
      "epoch": 18.984409799554566,
      "loss": 0.4005663990974426,
      "loss_ce": 0.0004199253162369132,
      "loss_iou": 0.1640625,
      "loss_num": 0.014404296875,
      "loss_xval": 0.400390625,
      "num_input_tokens_seen": 477623140,
      "step": 8524
    },
    {
      "epoch": 18.98663697104677,
      "grad_norm": 14.066349983215332,
      "learning_rate": 1e-06,
      "loss": 0.3884,
      "num_input_tokens_seen": 477678528,
      "step": 8525
    },
    {
      "epoch": 18.98663697104677,
      "loss": 0.42566099762916565,
      "loss_ce": 6.285572453634813e-05,
      "loss_iou": 0.197265625,
      "loss_num": 0.00616455078125,
      "loss_xval": 0.42578125,
      "num_input_tokens_seen": 477678528,
      "step": 8525
    },
    {
      "epoch": 18.988864142538976,
      "grad_norm": 17.08536720275879,
      "learning_rate": 1e-06,
      "loss": 0.3323,
      "num_input_tokens_seen": 477735340,
      "step": 8526
    },
    {
      "epoch": 18.988864142538976,
      "loss": 0.3974015414714813,
      "loss_ce": 0.00012369919568300247,
      "loss_iou": 0.17578125,
      "loss_num": 0.009033203125,
      "loss_xval": 0.396484375,
      "num_input_tokens_seen": 477735340,
      "step": 8526
    },
    {
      "epoch": 18.99109131403118,
      "grad_norm": 17.082090377807617,
      "learning_rate": 1e-06,
      "loss": 0.3237,
      "num_input_tokens_seen": 477792108,
      "step": 8527
    },
    {
      "epoch": 18.99109131403118,
      "loss": 0.27058377861976624,
      "loss_ce": 7.597178046125919e-05,
      "loss_iou": 0.1103515625,
      "loss_num": 0.010009765625,
      "loss_xval": 0.26953125,
      "num_input_tokens_seen": 477792108,
      "step": 8527
    },
    {
      "epoch": 18.993318485523385,
      "grad_norm": 13.515477180480957,
      "learning_rate": 1e-06,
      "loss": 0.35,
      "num_input_tokens_seen": 477846252,
      "step": 8528
    },
    {
      "epoch": 18.993318485523385,
      "loss": 0.24356822669506073,
      "loss_ce": 6.846075120847672e-05,
      "loss_iou": 0.10693359375,
      "loss_num": 0.006011962890625,
      "loss_xval": 0.2431640625,
      "num_input_tokens_seen": 477846252,
      "step": 8528
    },
    {
      "epoch": 18.99554565701559,
      "grad_norm": 12.548789978027344,
      "learning_rate": 1e-06,
      "loss": 0.4884,
      "num_input_tokens_seen": 477901432,
      "step": 8529
    },
    {
      "epoch": 18.99554565701559,
      "loss": 0.44635307788848877,
      "loss_ce": 6.401098653441295e-05,
      "loss_iou": 0.1806640625,
      "loss_num": 0.01708984375,
      "loss_xval": 0.4453125,
      "num_input_tokens_seen": 477901432,
      "step": 8529
    },
    {
      "epoch": 18.997772828507795,
      "grad_norm": 19.36771011352539,
      "learning_rate": 1e-06,
      "loss": 0.4848,
      "num_input_tokens_seen": 477959164,
      "step": 8530
    },
    {
      "epoch": 18.997772828507795,
      "loss": 0.4359922409057617,
      "loss_ce": 0.00014018421643413603,
      "loss_iou": 0.185546875,
      "loss_num": 0.012939453125,
      "loss_xval": 0.435546875,
      "num_input_tokens_seen": 477959164,
      "step": 8530
    },
    {
      "epoch": 19.0,
      "grad_norm": 24.555925369262695,
      "learning_rate": 1e-06,
      "loss": 0.4317,
      "num_input_tokens_seen": 478014600,
      "step": 8531
    },
    {
      "epoch": 19.0,
      "loss": 0.5009192228317261,
      "loss_ce": 6.472200766438618e-05,
      "loss_iou": 0.2236328125,
      "loss_num": 0.0108642578125,
      "loss_xval": 0.5,
      "num_input_tokens_seen": 478014600,
      "step": 8531
    },
    {
      "epoch": 19.002227171492205,
      "grad_norm": 19.39182472229004,
      "learning_rate": 1e-06,
      "loss": 0.3046,
      "num_input_tokens_seen": 478070168,
      "step": 8532
    },
    {
      "epoch": 19.002227171492205,
      "loss": 0.2748461961746216,
      "loss_ce": 6.590808334294707e-05,
      "loss_iou": 0.12255859375,
      "loss_num": 0.005859375,
      "loss_xval": 0.275390625,
      "num_input_tokens_seen": 478070168,
      "step": 8532
    },
    {
      "epoch": 19.00445434298441,
      "grad_norm": 15.633817672729492,
      "learning_rate": 1e-06,
      "loss": 0.2743,
      "num_input_tokens_seen": 478124420,
      "step": 8533
    },
    {
      "epoch": 19.00445434298441,
      "loss": 0.2822956442832947,
      "loss_ce": 6.907560600666329e-05,
      "loss_iou": 0.11865234375,
      "loss_num": 0.00897216796875,
      "loss_xval": 0.28125,
      "num_input_tokens_seen": 478124420,
      "step": 8533
    },
    {
      "epoch": 19.006681514476615,
      "grad_norm": 20.823501586914062,
      "learning_rate": 1e-06,
      "loss": 0.4119,
      "num_input_tokens_seen": 478179792,
      "step": 8534
    },
    {
      "epoch": 19.006681514476615,
      "loss": 0.4808357357978821,
      "loss_ce": 6.183330697240308e-05,
      "loss_iou": 0.2001953125,
      "loss_num": 0.01611328125,
      "loss_xval": 0.48046875,
      "num_input_tokens_seen": 478179792,
      "step": 8534
    },
    {
      "epoch": 19.00890868596882,
      "grad_norm": 23.914974212646484,
      "learning_rate": 1e-06,
      "loss": 0.433,
      "num_input_tokens_seen": 478236068,
      "step": 8535
    },
    {
      "epoch": 19.00890868596882,
      "loss": 0.5453649759292603,
      "loss_ce": 7.688709592912346e-05,
      "loss_iou": 0.2412109375,
      "loss_num": 0.0125732421875,
      "loss_xval": 0.546875,
      "num_input_tokens_seen": 478236068,
      "step": 8535
    },
    {
      "epoch": 19.011135857461024,
      "grad_norm": 13.552783966064453,
      "learning_rate": 1e-06,
      "loss": 0.2469,
      "num_input_tokens_seen": 478293348,
      "step": 8536
    },
    {
      "epoch": 19.011135857461024,
      "loss": 0.2078023999929428,
      "loss_ce": 6.924466288182884e-05,
      "loss_iou": 0.09375,
      "loss_num": 0.0040283203125,
      "loss_xval": 0.2080078125,
      "num_input_tokens_seen": 478293348,
      "step": 8536
    },
    {
      "epoch": 19.01336302895323,
      "grad_norm": 16.82329750061035,
      "learning_rate": 1e-06,
      "loss": 0.4532,
      "num_input_tokens_seen": 478352520,
      "step": 8537
    },
    {
      "epoch": 19.01336302895323,
      "loss": 0.4592329263687134,
      "loss_ce": 6.543014023918658e-05,
      "loss_iou": 0.205078125,
      "loss_num": 0.010009765625,
      "loss_xval": 0.458984375,
      "num_input_tokens_seen": 478352520,
      "step": 8537
    },
    {
      "epoch": 19.015590200445434,
      "grad_norm": 18.402179718017578,
      "learning_rate": 1e-06,
      "loss": 0.4807,
      "num_input_tokens_seen": 478409540,
      "step": 8538
    },
    {
      "epoch": 19.015590200445434,
      "loss": 0.5504905581474304,
      "loss_ce": 7.548804569523782e-05,
      "loss_iou": 0.240234375,
      "loss_num": 0.0140380859375,
      "loss_xval": 0.55078125,
      "num_input_tokens_seen": 478409540,
      "step": 8538
    },
    {
      "epoch": 19.01781737193764,
      "grad_norm": 18.57283592224121,
      "learning_rate": 1e-06,
      "loss": 0.2209,
      "num_input_tokens_seen": 478465060,
      "step": 8539
    },
    {
      "epoch": 19.01781737193764,
      "loss": 0.2368859052658081,
      "loss_ce": 6.950320675969124e-05,
      "loss_iou": 0.09912109375,
      "loss_num": 0.00762939453125,
      "loss_xval": 0.236328125,
      "num_input_tokens_seen": 478465060,
      "step": 8539
    },
    {
      "epoch": 19.020044543429844,
      "grad_norm": 20.981054306030273,
      "learning_rate": 1e-06,
      "loss": 0.4838,
      "num_input_tokens_seen": 478520264,
      "step": 8540
    },
    {
      "epoch": 19.020044543429844,
      "loss": 0.39203161001205444,
      "loss_ce": 6.387151370290667e-05,
      "loss_iou": 0.1650390625,
      "loss_num": 0.01239013671875,
      "loss_xval": 0.392578125,
      "num_input_tokens_seen": 478520264,
      "step": 8540
    },
    {
      "epoch": 19.02227171492205,
      "grad_norm": 22.715524673461914,
      "learning_rate": 1e-06,
      "loss": 0.3701,
      "num_input_tokens_seen": 478577056,
      "step": 8541
    },
    {
      "epoch": 19.02227171492205,
      "loss": 0.37838345766067505,
      "loss_ce": 8.756719762459397e-05,
      "loss_iou": 0.158203125,
      "loss_num": 0.01239013671875,
      "loss_xval": 0.37890625,
      "num_input_tokens_seen": 478577056,
      "step": 8541
    },
    {
      "epoch": 19.024498886414253,
      "grad_norm": 19.18893051147461,
      "learning_rate": 1e-06,
      "loss": 0.3867,
      "num_input_tokens_seen": 478633512,
      "step": 8542
    },
    {
      "epoch": 19.024498886414253,
      "loss": 0.3619902729988098,
      "loss_ce": 6.705922714900225e-05,
      "loss_iou": 0.1640625,
      "loss_num": 0.00689697265625,
      "loss_xval": 0.361328125,
      "num_input_tokens_seen": 478633512,
      "step": 8542
    },
    {
      "epoch": 19.02672605790646,
      "grad_norm": 16.043140411376953,
      "learning_rate": 1e-06,
      "loss": 0.5777,
      "num_input_tokens_seen": 478689620,
      "step": 8543
    },
    {
      "epoch": 19.02672605790646,
      "loss": 0.8731194734573364,
      "loss_ce": 7.26049329387024e-05,
      "loss_iou": 0.375,
      "loss_num": 0.02490234375,
      "loss_xval": 0.875,
      "num_input_tokens_seen": 478689620,
      "step": 8543
    },
    {
      "epoch": 19.028953229398663,
      "grad_norm": 22.03814697265625,
      "learning_rate": 1e-06,
      "loss": 0.3668,
      "num_input_tokens_seen": 478745996,
      "step": 8544
    },
    {
      "epoch": 19.028953229398663,
      "loss": 0.3608505129814148,
      "loss_ce": 7.170035678427666e-05,
      "loss_iou": 0.158203125,
      "loss_num": 0.00909423828125,
      "loss_xval": 0.361328125,
      "num_input_tokens_seen": 478745996,
      "step": 8544
    },
    {
      "epoch": 19.031180400890868,
      "grad_norm": 18.709949493408203,
      "learning_rate": 1e-06,
      "loss": 0.3843,
      "num_input_tokens_seen": 478802344,
      "step": 8545
    },
    {
      "epoch": 19.031180400890868,
      "loss": 0.3520601689815521,
      "loss_ce": 7.041079516056925e-05,
      "loss_iou": 0.1328125,
      "loss_num": 0.017333984375,
      "loss_xval": 0.3515625,
      "num_input_tokens_seen": 478802344,
      "step": 8545
    },
    {
      "epoch": 19.033407572383073,
      "grad_norm": 14.813765525817871,
      "learning_rate": 1e-06,
      "loss": 0.463,
      "num_input_tokens_seen": 478860080,
      "step": 8546
    },
    {
      "epoch": 19.033407572383073,
      "loss": 0.5505503416061401,
      "loss_ce": 7.423260831274092e-05,
      "loss_iou": 0.2060546875,
      "loss_num": 0.027587890625,
      "loss_xval": 0.55078125,
      "num_input_tokens_seen": 478860080,
      "step": 8546
    },
    {
      "epoch": 19.035634743875278,
      "grad_norm": 23.64231300354004,
      "learning_rate": 1e-06,
      "loss": 0.3234,
      "num_input_tokens_seen": 478919644,
      "step": 8547
    },
    {
      "epoch": 19.035634743875278,
      "loss": 0.3638024926185608,
      "loss_ce": 9.396575478604063e-05,
      "loss_iou": 0.1220703125,
      "loss_num": 0.02392578125,
      "loss_xval": 0.36328125,
      "num_input_tokens_seen": 478919644,
      "step": 8547
    },
    {
      "epoch": 19.037861915367483,
      "grad_norm": 22.59861183166504,
      "learning_rate": 1e-06,
      "loss": 0.2975,
      "num_input_tokens_seen": 478974500,
      "step": 8548
    },
    {
      "epoch": 19.037861915367483,
      "loss": 0.27411675453186035,
      "loss_ce": 6.891056546010077e-05,
      "loss_iou": 0.119140625,
      "loss_num": 0.00701904296875,
      "loss_xval": 0.2734375,
      "num_input_tokens_seen": 478974500,
      "step": 8548
    },
    {
      "epoch": 19.040089086859687,
      "grad_norm": 21.548479080200195,
      "learning_rate": 1e-06,
      "loss": 0.5299,
      "num_input_tokens_seen": 479029432,
      "step": 8549
    },
    {
      "epoch": 19.040089086859687,
      "loss": 0.5104723572731018,
      "loss_ce": 9.636413597036153e-05,
      "loss_iou": 0.20703125,
      "loss_num": 0.019287109375,
      "loss_xval": 0.51171875,
      "num_input_tokens_seen": 479029432,
      "step": 8549
    },
    {
      "epoch": 19.042316258351892,
      "grad_norm": 22.90362548828125,
      "learning_rate": 1e-06,
      "loss": 0.6133,
      "num_input_tokens_seen": 479086928,
      "step": 8550
    },
    {
      "epoch": 19.042316258351892,
      "loss": 0.6397278904914856,
      "loss_ce": 7.949080463731661e-05,
      "loss_iou": 0.263671875,
      "loss_num": 0.0228271484375,
      "loss_xval": 0.640625,
      "num_input_tokens_seen": 479086928,
      "step": 8550
    },
    {
      "epoch": 19.044543429844097,
      "grad_norm": 13.663710594177246,
      "learning_rate": 1e-06,
      "loss": 0.4714,
      "num_input_tokens_seen": 479143420,
      "step": 8551
    },
    {
      "epoch": 19.044543429844097,
      "loss": 0.41580140590667725,
      "loss_ce": 9.095711720874533e-05,
      "loss_iou": 0.1640625,
      "loss_num": 0.017333984375,
      "loss_xval": 0.416015625,
      "num_input_tokens_seen": 479143420,
      "step": 8551
    },
    {
      "epoch": 19.046770601336302,
      "grad_norm": 26.639122009277344,
      "learning_rate": 1e-06,
      "loss": 0.4388,
      "num_input_tokens_seen": 479199488,
      "step": 8552
    },
    {
      "epoch": 19.046770601336302,
      "loss": 0.4706590175628662,
      "loss_ce": 7.797098805895075e-05,
      "loss_iou": 0.2060546875,
      "loss_num": 0.01190185546875,
      "loss_xval": 0.470703125,
      "num_input_tokens_seen": 479199488,
      "step": 8552
    },
    {
      "epoch": 19.048997772828507,
      "grad_norm": 37.29257583618164,
      "learning_rate": 1e-06,
      "loss": 0.4645,
      "num_input_tokens_seen": 479255004,
      "step": 8553
    },
    {
      "epoch": 19.048997772828507,
      "loss": 0.6149426698684692,
      "loss_ce": 7.449123950209469e-05,
      "loss_iou": 0.248046875,
      "loss_num": 0.023681640625,
      "loss_xval": 0.61328125,
      "num_input_tokens_seen": 479255004,
      "step": 8553
    },
    {
      "epoch": 19.051224944320712,
      "grad_norm": 187.8523712158203,
      "learning_rate": 1e-06,
      "loss": 0.4137,
      "num_input_tokens_seen": 479310756,
      "step": 8554
    },
    {
      "epoch": 19.051224944320712,
      "loss": 0.35520029067993164,
      "loss_ce": 9.777834929991513e-05,
      "loss_iou": 0.14453125,
      "loss_num": 0.01336669921875,
      "loss_xval": 0.35546875,
      "num_input_tokens_seen": 479310756,
      "step": 8554
    },
    {
      "epoch": 19.053452115812917,
      "grad_norm": 73.10868835449219,
      "learning_rate": 1e-06,
      "loss": 0.4949,
      "num_input_tokens_seen": 479365280,
      "step": 8555
    },
    {
      "epoch": 19.053452115812917,
      "loss": 0.5539067387580872,
      "loss_ce": 7.372665277216583e-05,
      "loss_iou": 0.2451171875,
      "loss_num": 0.01275634765625,
      "loss_xval": 0.5546875,
      "num_input_tokens_seen": 479365280,
      "step": 8555
    },
    {
      "epoch": 19.05567928730512,
      "grad_norm": 25.378646850585938,
      "learning_rate": 1e-06,
      "loss": 0.261,
      "num_input_tokens_seen": 479421220,
      "step": 8556
    },
    {
      "epoch": 19.05567928730512,
      "loss": 0.2171689122915268,
      "loss_ce": 6.685940024908632e-05,
      "loss_iou": 0.09033203125,
      "loss_num": 0.007293701171875,
      "loss_xval": 0.216796875,
      "num_input_tokens_seen": 479421220,
      "step": 8556
    },
    {
      "epoch": 19.057906458797326,
      "grad_norm": 14.335795402526855,
      "learning_rate": 1e-06,
      "loss": 0.289,
      "num_input_tokens_seen": 479477700,
      "step": 8557
    },
    {
      "epoch": 19.057906458797326,
      "loss": 0.18457838892936707,
      "loss_ce": 6.912185926921666e-05,
      "loss_iou": 0.07666015625,
      "loss_num": 0.00616455078125,
      "loss_xval": 0.1845703125,
      "num_input_tokens_seen": 479477700,
      "step": 8557
    },
    {
      "epoch": 19.06013363028953,
      "grad_norm": 14.203603744506836,
      "learning_rate": 1e-06,
      "loss": 0.4151,
      "num_input_tokens_seen": 479536192,
      "step": 8558
    },
    {
      "epoch": 19.06013363028953,
      "loss": 0.47530919313430786,
      "loss_ce": 8.947977039497346e-05,
      "loss_iou": 0.208984375,
      "loss_num": 0.01153564453125,
      "loss_xval": 0.474609375,
      "num_input_tokens_seen": 479536192,
      "step": 8558
    },
    {
      "epoch": 19.062360801781736,
      "grad_norm": 15.753239631652832,
      "learning_rate": 1e-06,
      "loss": 0.2125,
      "num_input_tokens_seen": 479591752,
      "step": 8559
    },
    {
      "epoch": 19.062360801781736,
      "loss": 0.19928745925426483,
      "loss_ce": 6.870189099572599e-05,
      "loss_iou": 0.07568359375,
      "loss_num": 0.0096435546875,
      "loss_xval": 0.19921875,
      "num_input_tokens_seen": 479591752,
      "step": 8559
    },
    {
      "epoch": 19.06458797327394,
      "grad_norm": 13.365279197692871,
      "learning_rate": 1e-06,
      "loss": 0.2772,
      "num_input_tokens_seen": 479650200,
      "step": 8560
    },
    {
      "epoch": 19.06458797327394,
      "loss": 0.31562650203704834,
      "loss_ce": 7.472425932064652e-05,
      "loss_iou": 0.140625,
      "loss_num": 0.00689697265625,
      "loss_xval": 0.31640625,
      "num_input_tokens_seen": 479650200,
      "step": 8560
    },
    {
      "epoch": 19.066815144766146,
      "grad_norm": 16.33476448059082,
      "learning_rate": 1e-06,
      "loss": 0.3525,
      "num_input_tokens_seen": 479705436,
      "step": 8561
    },
    {
      "epoch": 19.066815144766146,
      "loss": 0.3400951027870178,
      "loss_ce": 6.827242759754881e-05,
      "loss_iou": 0.140625,
      "loss_num": 0.01177978515625,
      "loss_xval": 0.33984375,
      "num_input_tokens_seen": 479705436,
      "step": 8561
    },
    {
      "epoch": 19.06904231625835,
      "grad_norm": 11.422565460205078,
      "learning_rate": 1e-06,
      "loss": 0.2924,
      "num_input_tokens_seen": 479760176,
      "step": 8562
    },
    {
      "epoch": 19.06904231625835,
      "loss": 0.21368908882141113,
      "loss_ce": 6.604377267649397e-05,
      "loss_iou": 0.08740234375,
      "loss_num": 0.00775146484375,
      "loss_xval": 0.2138671875,
      "num_input_tokens_seen": 479760176,
      "step": 8562
    },
    {
      "epoch": 19.071269487750556,
      "grad_norm": 14.924948692321777,
      "learning_rate": 1e-06,
      "loss": 0.3429,
      "num_input_tokens_seen": 479815112,
      "step": 8563
    },
    {
      "epoch": 19.071269487750556,
      "loss": 0.42559611797332764,
      "loss_ce": 5.89884803048335e-05,
      "loss_iou": 0.15234375,
      "loss_num": 0.0240478515625,
      "loss_xval": 0.42578125,
      "num_input_tokens_seen": 479815112,
      "step": 8563
    },
    {
      "epoch": 19.07349665924276,
      "grad_norm": 26.38053321838379,
      "learning_rate": 1e-06,
      "loss": 0.3599,
      "num_input_tokens_seen": 479870284,
      "step": 8564
    },
    {
      "epoch": 19.07349665924276,
      "loss": 0.36798515915870667,
      "loss_ce": 6.524256605189294e-05,
      "loss_iou": 0.1708984375,
      "loss_num": 0.00537109375,
      "loss_xval": 0.3671875,
      "num_input_tokens_seen": 479870284,
      "step": 8564
    },
    {
      "epoch": 19.075723830734965,
      "grad_norm": 13.398744583129883,
      "learning_rate": 1e-06,
      "loss": 0.3668,
      "num_input_tokens_seen": 479927696,
      "step": 8565
    },
    {
      "epoch": 19.075723830734965,
      "loss": 0.4121933877468109,
      "loss_ce": 8.401382365263999e-05,
      "loss_iou": 0.1884765625,
      "loss_num": 0.007171630859375,
      "loss_xval": 0.412109375,
      "num_input_tokens_seen": 479927696,
      "step": 8565
    },
    {
      "epoch": 19.07795100222717,
      "grad_norm": 23.407733917236328,
      "learning_rate": 1e-06,
      "loss": 0.5092,
      "num_input_tokens_seen": 479984008,
      "step": 8566
    },
    {
      "epoch": 19.07795100222717,
      "loss": 0.536881685256958,
      "loss_ce": 0.00013851752737537026,
      "loss_iou": 0.220703125,
      "loss_num": 0.0189208984375,
      "loss_xval": 0.53515625,
      "num_input_tokens_seen": 479984008,
      "step": 8566
    },
    {
      "epoch": 19.080178173719375,
      "grad_norm": 26.368122100830078,
      "learning_rate": 1e-06,
      "loss": 0.4519,
      "num_input_tokens_seen": 480038472,
      "step": 8567
    },
    {
      "epoch": 19.080178173719375,
      "loss": 0.5067821741104126,
      "loss_ce": 0.00031240255339071155,
      "loss_iou": 0.21484375,
      "loss_num": 0.01531982421875,
      "loss_xval": 0.5078125,
      "num_input_tokens_seen": 480038472,
      "step": 8567
    },
    {
      "epoch": 19.08240534521158,
      "grad_norm": 16.010215759277344,
      "learning_rate": 1e-06,
      "loss": 0.3062,
      "num_input_tokens_seen": 480091528,
      "step": 8568
    },
    {
      "epoch": 19.08240534521158,
      "loss": 0.3443114161491394,
      "loss_ce": 7.312708476092666e-05,
      "loss_iou": 0.1611328125,
      "loss_num": 0.0045166015625,
      "loss_xval": 0.34375,
      "num_input_tokens_seen": 480091528,
      "step": 8568
    },
    {
      "epoch": 19.084632516703785,
      "grad_norm": 15.22715950012207,
      "learning_rate": 1e-06,
      "loss": 0.3562,
      "num_input_tokens_seen": 480147780,
      "step": 8569
    },
    {
      "epoch": 19.084632516703785,
      "loss": 0.4474610388278961,
      "loss_ce": 7.335421105381101e-05,
      "loss_iou": 0.19140625,
      "loss_num": 0.01312255859375,
      "loss_xval": 0.447265625,
      "num_input_tokens_seen": 480147780,
      "step": 8569
    },
    {
      "epoch": 19.08685968819599,
      "grad_norm": 12.806982040405273,
      "learning_rate": 1e-06,
      "loss": 0.3955,
      "num_input_tokens_seen": 480203968,
      "step": 8570
    },
    {
      "epoch": 19.08685968819599,
      "loss": 0.2964867353439331,
      "loss_ce": 0.00010003504576161504,
      "loss_iou": 0.13671875,
      "loss_num": 0.004669189453125,
      "loss_xval": 0.296875,
      "num_input_tokens_seen": 480203968,
      "step": 8570
    },
    {
      "epoch": 19.089086859688194,
      "grad_norm": 21.179973602294922,
      "learning_rate": 1e-06,
      "loss": 0.356,
      "num_input_tokens_seen": 480260560,
      "step": 8571
    },
    {
      "epoch": 19.089086859688194,
      "loss": 0.29164212942123413,
      "loss_ce": 7.720879511907697e-05,
      "loss_iou": 0.1318359375,
      "loss_num": 0.005706787109375,
      "loss_xval": 0.291015625,
      "num_input_tokens_seen": 480260560,
      "step": 8571
    },
    {
      "epoch": 19.0913140311804,
      "grad_norm": 17.90711784362793,
      "learning_rate": 1e-06,
      "loss": 0.4287,
      "num_input_tokens_seen": 480317040,
      "step": 8572
    },
    {
      "epoch": 19.0913140311804,
      "loss": 0.5795639753341675,
      "loss_ce": 0.00015723146498203278,
      "loss_iou": 0.2421875,
      "loss_num": 0.0191650390625,
      "loss_xval": 0.578125,
      "num_input_tokens_seen": 480317040,
      "step": 8572
    },
    {
      "epoch": 19.093541202672604,
      "grad_norm": 26.034122467041016,
      "learning_rate": 1e-06,
      "loss": 0.3087,
      "num_input_tokens_seen": 480371384,
      "step": 8573
    },
    {
      "epoch": 19.093541202672604,
      "loss": 0.33735281229019165,
      "loss_ce": 7.252431532833725e-05,
      "loss_iou": 0.1474609375,
      "loss_num": 0.00830078125,
      "loss_xval": 0.337890625,
      "num_input_tokens_seen": 480371384,
      "step": 8573
    },
    {
      "epoch": 19.09576837416481,
      "grad_norm": 17.473873138427734,
      "learning_rate": 1e-06,
      "loss": 0.3811,
      "num_input_tokens_seen": 480426792,
      "step": 8574
    },
    {
      "epoch": 19.09576837416481,
      "loss": 0.4617387652397156,
      "loss_ce": 6.881446461193264e-05,
      "loss_iou": 0.1904296875,
      "loss_num": 0.0159912109375,
      "loss_xval": 0.4609375,
      "num_input_tokens_seen": 480426792,
      "step": 8574
    },
    {
      "epoch": 19.097995545657014,
      "grad_norm": 16.091129302978516,
      "learning_rate": 1e-06,
      "loss": 0.4135,
      "num_input_tokens_seen": 480483268,
      "step": 8575
    },
    {
      "epoch": 19.097995545657014,
      "loss": 0.29927489161491394,
      "loss_ce": 8.053722558543086e-05,
      "loss_iou": 0.1357421875,
      "loss_num": 0.005462646484375,
      "loss_xval": 0.298828125,
      "num_input_tokens_seen": 480483268,
      "step": 8575
    },
    {
      "epoch": 19.100222717149222,
      "grad_norm": 22.522878646850586,
      "learning_rate": 1e-06,
      "loss": 0.595,
      "num_input_tokens_seen": 480537020,
      "step": 8576
    },
    {
      "epoch": 19.100222717149222,
      "loss": 0.5641788244247437,
      "loss_ce": 9.193105506710708e-05,
      "loss_iou": 0.2470703125,
      "loss_num": 0.01409912109375,
      "loss_xval": 0.5625,
      "num_input_tokens_seen": 480537020,
      "step": 8576
    },
    {
      "epoch": 19.102449888641427,
      "grad_norm": 24.5418701171875,
      "learning_rate": 1e-06,
      "loss": 0.2987,
      "num_input_tokens_seen": 480592488,
      "step": 8577
    },
    {
      "epoch": 19.102449888641427,
      "loss": 0.2232351005077362,
      "loss_ce": 7.150069723138586e-05,
      "loss_iou": 0.1025390625,
      "loss_num": 0.003631591796875,
      "loss_xval": 0.2236328125,
      "num_input_tokens_seen": 480592488,
      "step": 8577
    },
    {
      "epoch": 19.104677060133632,
      "grad_norm": 27.310413360595703,
      "learning_rate": 1e-06,
      "loss": 0.4059,
      "num_input_tokens_seen": 480647028,
      "step": 8578
    },
    {
      "epoch": 19.104677060133632,
      "loss": 0.4668709635734558,
      "loss_ce": 7.407699013128877e-05,
      "loss_iou": 0.19921875,
      "loss_num": 0.013671875,
      "loss_xval": 0.466796875,
      "num_input_tokens_seen": 480647028,
      "step": 8578
    },
    {
      "epoch": 19.106904231625837,
      "grad_norm": 19.787166595458984,
      "learning_rate": 1e-06,
      "loss": 0.3002,
      "num_input_tokens_seen": 480702564,
      "step": 8579
    },
    {
      "epoch": 19.106904231625837,
      "loss": 0.2308284193277359,
      "loss_ce": 8.500830153934658e-05,
      "loss_iou": 0.107421875,
      "loss_num": 0.0031280517578125,
      "loss_xval": 0.23046875,
      "num_input_tokens_seen": 480702564,
      "step": 8579
    },
    {
      "epoch": 19.10913140311804,
      "grad_norm": 20.583070755004883,
      "learning_rate": 1e-06,
      "loss": 0.3846,
      "num_input_tokens_seen": 480758448,
      "step": 8580
    },
    {
      "epoch": 19.10913140311804,
      "loss": 0.39584484696388245,
      "loss_ce": 6.236594344954938e-05,
      "loss_iou": 0.1591796875,
      "loss_num": 0.0155029296875,
      "loss_xval": 0.396484375,
      "num_input_tokens_seen": 480758448,
      "step": 8580
    },
    {
      "epoch": 19.111358574610247,
      "grad_norm": 20.13869285583496,
      "learning_rate": 1e-06,
      "loss": 0.3753,
      "num_input_tokens_seen": 480810532,
      "step": 8581
    },
    {
      "epoch": 19.111358574610247,
      "loss": 0.4490373134613037,
      "loss_ce": 6.272210157476366e-05,
      "loss_iou": 0.2060546875,
      "loss_num": 0.007476806640625,
      "loss_xval": 0.44921875,
      "num_input_tokens_seen": 480810532,
      "step": 8581
    },
    {
      "epoch": 19.11358574610245,
      "grad_norm": 27.062620162963867,
      "learning_rate": 1e-06,
      "loss": 0.4051,
      "num_input_tokens_seen": 480867128,
      "step": 8582
    },
    {
      "epoch": 19.11358574610245,
      "loss": 0.388627290725708,
      "loss_ce": 7.747422205284238e-05,
      "loss_iou": 0.181640625,
      "loss_num": 0.005279541015625,
      "loss_xval": 0.388671875,
      "num_input_tokens_seen": 480867128,
      "step": 8582
    },
    {
      "epoch": 19.115812917594656,
      "grad_norm": 14.657607078552246,
      "learning_rate": 1e-06,
      "loss": 0.2535,
      "num_input_tokens_seen": 480921476,
      "step": 8583
    },
    {
      "epoch": 19.115812917594656,
      "loss": 0.35783708095550537,
      "loss_ce": 0.00017106709128711373,
      "loss_iou": 0.1572265625,
      "loss_num": 0.008544921875,
      "loss_xval": 0.357421875,
      "num_input_tokens_seen": 480921476,
      "step": 8583
    },
    {
      "epoch": 19.11804008908686,
      "grad_norm": 21.15728759765625,
      "learning_rate": 1e-06,
      "loss": 0.4817,
      "num_input_tokens_seen": 480977352,
      "step": 8584
    },
    {
      "epoch": 19.11804008908686,
      "loss": 0.541458010673523,
      "loss_ce": 7.615622598677874e-05,
      "loss_iou": 0.212890625,
      "loss_num": 0.023193359375,
      "loss_xval": 0.54296875,
      "num_input_tokens_seen": 480977352,
      "step": 8584
    },
    {
      "epoch": 19.120267260579066,
      "grad_norm": 20.03176498413086,
      "learning_rate": 1e-06,
      "loss": 0.4498,
      "num_input_tokens_seen": 481031908,
      "step": 8585
    },
    {
      "epoch": 19.120267260579066,
      "loss": 0.45001912117004395,
      "loss_ce": 6.798194954171777e-05,
      "loss_iou": 0.169921875,
      "loss_num": 0.0218505859375,
      "loss_xval": 0.44921875,
      "num_input_tokens_seen": 481031908,
      "step": 8585
    },
    {
      "epoch": 19.12249443207127,
      "grad_norm": 20.93001937866211,
      "learning_rate": 1e-06,
      "loss": 0.3745,
      "num_input_tokens_seen": 481088708,
      "step": 8586
    },
    {
      "epoch": 19.12249443207127,
      "loss": 0.41656631231307983,
      "loss_ce": 6.239269714569673e-05,
      "loss_iou": 0.189453125,
      "loss_num": 0.00738525390625,
      "loss_xval": 0.416015625,
      "num_input_tokens_seen": 481088708,
      "step": 8586
    },
    {
      "epoch": 19.124721603563476,
      "grad_norm": 15.11413860321045,
      "learning_rate": 1e-06,
      "loss": 0.3466,
      "num_input_tokens_seen": 481146480,
      "step": 8587
    },
    {
      "epoch": 19.124721603563476,
      "loss": 0.3888677954673767,
      "loss_ce": 7.386014476651326e-05,
      "loss_iou": 0.1630859375,
      "loss_num": 0.01226806640625,
      "loss_xval": 0.388671875,
      "num_input_tokens_seen": 481146480,
      "step": 8587
    },
    {
      "epoch": 19.12694877505568,
      "grad_norm": 15.431100845336914,
      "learning_rate": 1e-06,
      "loss": 0.2699,
      "num_input_tokens_seen": 481202468,
      "step": 8588
    },
    {
      "epoch": 19.12694877505568,
      "loss": 0.22169449925422668,
      "loss_ce": 7.583482511108741e-05,
      "loss_iou": 0.09765625,
      "loss_num": 0.00537109375,
      "loss_xval": 0.2216796875,
      "num_input_tokens_seen": 481202468,
      "step": 8588
    },
    {
      "epoch": 19.129175946547885,
      "grad_norm": 14.637504577636719,
      "learning_rate": 1e-06,
      "loss": 0.2738,
      "num_input_tokens_seen": 481258796,
      "step": 8589
    },
    {
      "epoch": 19.129175946547885,
      "loss": 0.2992823123931885,
      "loss_ce": 8.799122588243335e-05,
      "loss_iou": 0.1298828125,
      "loss_num": 0.008056640625,
      "loss_xval": 0.298828125,
      "num_input_tokens_seen": 481258796,
      "step": 8589
    },
    {
      "epoch": 19.13140311804009,
      "grad_norm": 18.58306312561035,
      "learning_rate": 1e-06,
      "loss": 0.4947,
      "num_input_tokens_seen": 481314720,
      "step": 8590
    },
    {
      "epoch": 19.13140311804009,
      "loss": 0.6162921786308289,
      "loss_ce": 8.123279258143157e-05,
      "loss_iou": 0.279296875,
      "loss_num": 0.0113525390625,
      "loss_xval": 0.6171875,
      "num_input_tokens_seen": 481314720,
      "step": 8590
    },
    {
      "epoch": 19.133630289532295,
      "grad_norm": 24.67803382873535,
      "learning_rate": 1e-06,
      "loss": 0.3293,
      "num_input_tokens_seen": 481371188,
      "step": 8591
    },
    {
      "epoch": 19.133630289532295,
      "loss": 0.3968676030635834,
      "loss_ce": 7.80346745159477e-05,
      "loss_iou": 0.173828125,
      "loss_num": 0.0098876953125,
      "loss_xval": 0.396484375,
      "num_input_tokens_seen": 481371188,
      "step": 8591
    },
    {
      "epoch": 19.1358574610245,
      "grad_norm": 10.766166687011719,
      "learning_rate": 1e-06,
      "loss": 0.3724,
      "num_input_tokens_seen": 481427332,
      "step": 8592
    },
    {
      "epoch": 19.1358574610245,
      "loss": 0.4355580806732178,
      "loss_ce": 7.223740976769477e-05,
      "loss_iou": 0.193359375,
      "loss_num": 0.0096435546875,
      "loss_xval": 0.435546875,
      "num_input_tokens_seen": 481427332,
      "step": 8592
    },
    {
      "epoch": 19.138084632516705,
      "grad_norm": 18.193437576293945,
      "learning_rate": 1e-06,
      "loss": 0.3822,
      "num_input_tokens_seen": 481484316,
      "step": 8593
    },
    {
      "epoch": 19.138084632516705,
      "loss": 0.3628009557723999,
      "loss_ce": 6.902994937263429e-05,
      "loss_iou": 0.1669921875,
      "loss_num": 0.005859375,
      "loss_xval": 0.36328125,
      "num_input_tokens_seen": 481484316,
      "step": 8593
    },
    {
      "epoch": 19.14031180400891,
      "grad_norm": 20.437171936035156,
      "learning_rate": 1e-06,
      "loss": 0.3256,
      "num_input_tokens_seen": 481538132,
      "step": 8594
    },
    {
      "epoch": 19.14031180400891,
      "loss": 0.22638912498950958,
      "loss_ce": 7.076388283167034e-05,
      "loss_iou": 0.10107421875,
      "loss_num": 0.0048828125,
      "loss_xval": 0.2265625,
      "num_input_tokens_seen": 481538132,
      "step": 8594
    },
    {
      "epoch": 19.142538975501115,
      "grad_norm": 17.62116050720215,
      "learning_rate": 1e-06,
      "loss": 0.414,
      "num_input_tokens_seen": 481596580,
      "step": 8595
    },
    {
      "epoch": 19.142538975501115,
      "loss": 0.3615182042121887,
      "loss_ce": 6.798798858653754e-05,
      "loss_iou": 0.1572265625,
      "loss_num": 0.0093994140625,
      "loss_xval": 0.361328125,
      "num_input_tokens_seen": 481596580,
      "step": 8595
    },
    {
      "epoch": 19.14476614699332,
      "grad_norm": 16.3442440032959,
      "learning_rate": 1e-06,
      "loss": 0.3613,
      "num_input_tokens_seen": 481653060,
      "step": 8596
    },
    {
      "epoch": 19.14476614699332,
      "loss": 0.39320868253707886,
      "loss_ce": 0.00014228782674763352,
      "loss_iou": 0.171875,
      "loss_num": 0.010009765625,
      "loss_xval": 0.392578125,
      "num_input_tokens_seen": 481653060,
      "step": 8596
    },
    {
      "epoch": 19.146993318485524,
      "grad_norm": 11.166417121887207,
      "learning_rate": 1e-06,
      "loss": 0.5462,
      "num_input_tokens_seen": 481709604,
      "step": 8597
    },
    {
      "epoch": 19.146993318485524,
      "loss": 0.8146294355392456,
      "loss_ce": 0.00025259278481826186,
      "loss_iou": 0.271484375,
      "loss_num": 0.05419921875,
      "loss_xval": 0.8125,
      "num_input_tokens_seen": 481709604,
      "step": 8597
    },
    {
      "epoch": 19.14922048997773,
      "grad_norm": 24.46466827392578,
      "learning_rate": 1e-06,
      "loss": 0.5409,
      "num_input_tokens_seen": 481766076,
      "step": 8598
    },
    {
      "epoch": 19.14922048997773,
      "loss": 0.37435752153396606,
      "loss_ce": 8.992976654553786e-05,
      "loss_iou": 0.1669921875,
      "loss_num": 0.008056640625,
      "loss_xval": 0.375,
      "num_input_tokens_seen": 481766076,
      "step": 8598
    },
    {
      "epoch": 19.151447661469934,
      "grad_norm": 13.304176330566406,
      "learning_rate": 1e-06,
      "loss": 0.3356,
      "num_input_tokens_seen": 481822160,
      "step": 8599
    },
    {
      "epoch": 19.151447661469934,
      "loss": 0.24649234116077423,
      "loss_ce": 7.815843855496496e-05,
      "loss_iou": 0.11083984375,
      "loss_num": 0.0050048828125,
      "loss_xval": 0.24609375,
      "num_input_tokens_seen": 481822160,
      "step": 8599
    },
    {
      "epoch": 19.15367483296214,
      "grad_norm": 31.582448959350586,
      "learning_rate": 1e-06,
      "loss": 0.2843,
      "num_input_tokens_seen": 481878668,
      "step": 8600
    },
    {
      "epoch": 19.15367483296214,
      "loss": 0.27753129601478577,
      "loss_ce": 6.547504017362371e-05,
      "loss_iou": 0.1259765625,
      "loss_num": 0.00518798828125,
      "loss_xval": 0.27734375,
      "num_input_tokens_seen": 481878668,
      "step": 8600
    },
    {
      "epoch": 19.155902004454344,
      "grad_norm": 16.38020896911621,
      "learning_rate": 1e-06,
      "loss": 0.3844,
      "num_input_tokens_seen": 481936496,
      "step": 8601
    },
    {
      "epoch": 19.155902004454344,
      "loss": 0.43722057342529297,
      "loss_ce": 8.677801815792918e-05,
      "loss_iou": 0.193359375,
      "loss_num": 0.01025390625,
      "loss_xval": 0.4375,
      "num_input_tokens_seen": 481936496,
      "step": 8601
    },
    {
      "epoch": 19.15812917594655,
      "grad_norm": 79.99256896972656,
      "learning_rate": 1e-06,
      "loss": 0.4235,
      "num_input_tokens_seen": 481993192,
      "step": 8602
    },
    {
      "epoch": 19.15812917594655,
      "loss": 0.4538060426712036,
      "loss_ce": 7.070847641443834e-05,
      "loss_iou": 0.1884765625,
      "loss_num": 0.015380859375,
      "loss_xval": 0.453125,
      "num_input_tokens_seen": 481993192,
      "step": 8602
    },
    {
      "epoch": 19.160356347438753,
      "grad_norm": 13.313328742980957,
      "learning_rate": 1e-06,
      "loss": 0.3876,
      "num_input_tokens_seen": 482048636,
      "step": 8603
    },
    {
      "epoch": 19.160356347438753,
      "loss": 0.39753034710884094,
      "loss_ce": 6.941702304175124e-05,
      "loss_iou": 0.177734375,
      "loss_num": 0.0084228515625,
      "loss_xval": 0.3984375,
      "num_input_tokens_seen": 482048636,
      "step": 8603
    },
    {
      "epoch": 19.16258351893096,
      "grad_norm": 27.735868453979492,
      "learning_rate": 1e-06,
      "loss": 0.5142,
      "num_input_tokens_seen": 482099436,
      "step": 8604
    },
    {
      "epoch": 19.16258351893096,
      "loss": 0.3438310921192169,
      "loss_ce": 8.108046313282102e-05,
      "loss_iou": 0.1611328125,
      "loss_num": 0.00421142578125,
      "loss_xval": 0.34375,
      "num_input_tokens_seen": 482099436,
      "step": 8604
    },
    {
      "epoch": 19.164810690423163,
      "grad_norm": 17.522174835205078,
      "learning_rate": 1e-06,
      "loss": 0.4765,
      "num_input_tokens_seen": 482154916,
      "step": 8605
    },
    {
      "epoch": 19.164810690423163,
      "loss": 0.4795961380004883,
      "loss_ce": 0.00010396166180726141,
      "loss_iou": 0.1953125,
      "loss_num": 0.017578125,
      "loss_xval": 0.48046875,
      "num_input_tokens_seen": 482154916,
      "step": 8605
    },
    {
      "epoch": 19.167037861915368,
      "grad_norm": 17.66177749633789,
      "learning_rate": 1e-06,
      "loss": 0.5927,
      "num_input_tokens_seen": 482211392,
      "step": 8606
    },
    {
      "epoch": 19.167037861915368,
      "loss": 0.5756547451019287,
      "loss_ce": 0.00021524931071326137,
      "loss_iou": 0.259765625,
      "loss_num": 0.0113525390625,
      "loss_xval": 0.57421875,
      "num_input_tokens_seen": 482211392,
      "step": 8606
    },
    {
      "epoch": 19.169265033407573,
      "grad_norm": 19.507108688354492,
      "learning_rate": 1e-06,
      "loss": 0.2847,
      "num_input_tokens_seen": 482270844,
      "step": 8607
    },
    {
      "epoch": 19.169265033407573,
      "loss": 0.2603249251842499,
      "loss_ce": 7.102765084709972e-05,
      "loss_iou": 0.1220703125,
      "loss_num": 0.00323486328125,
      "loss_xval": 0.259765625,
      "num_input_tokens_seen": 482270844,
      "step": 8607
    },
    {
      "epoch": 19.171492204899778,
      "grad_norm": 28.656322479248047,
      "learning_rate": 1e-06,
      "loss": 0.4528,
      "num_input_tokens_seen": 482325476,
      "step": 8608
    },
    {
      "epoch": 19.171492204899778,
      "loss": 0.43023812770843506,
      "loss_ce": 0.00012339533714111894,
      "loss_iou": 0.1923828125,
      "loss_num": 0.00909423828125,
      "loss_xval": 0.4296875,
      "num_input_tokens_seen": 482325476,
      "step": 8608
    },
    {
      "epoch": 19.173719376391983,
      "grad_norm": 23.096620559692383,
      "learning_rate": 1e-06,
      "loss": 0.3489,
      "num_input_tokens_seen": 482380532,
      "step": 8609
    },
    {
      "epoch": 19.173719376391983,
      "loss": 0.48346471786499023,
      "loss_ce": 6.630421557929367e-05,
      "loss_iou": 0.2109375,
      "loss_num": 0.01226806640625,
      "loss_xval": 0.484375,
      "num_input_tokens_seen": 482380532,
      "step": 8609
    },
    {
      "epoch": 19.175946547884188,
      "grad_norm": 21.371700286865234,
      "learning_rate": 1e-06,
      "loss": 0.4853,
      "num_input_tokens_seen": 482433260,
      "step": 8610
    },
    {
      "epoch": 19.175946547884188,
      "loss": 0.4420054256916046,
      "loss_ce": 0.00023296594736166298,
      "loss_iou": 0.1865234375,
      "loss_num": 0.01385498046875,
      "loss_xval": 0.44140625,
      "num_input_tokens_seen": 482433260,
      "step": 8610
    },
    {
      "epoch": 19.178173719376392,
      "grad_norm": 22.586750030517578,
      "learning_rate": 1e-06,
      "loss": 0.4347,
      "num_input_tokens_seen": 482489688,
      "step": 8611
    },
    {
      "epoch": 19.178173719376392,
      "loss": 0.42078396677970886,
      "loss_ce": 6.861792644485831e-05,
      "loss_iou": 0.1904296875,
      "loss_num": 0.008056640625,
      "loss_xval": 0.419921875,
      "num_input_tokens_seen": 482489688,
      "step": 8611
    },
    {
      "epoch": 19.180400890868597,
      "grad_norm": 14.678589820861816,
      "learning_rate": 1e-06,
      "loss": 0.2694,
      "num_input_tokens_seen": 482546336,
      "step": 8612
    },
    {
      "epoch": 19.180400890868597,
      "loss": 0.22610211372375488,
      "loss_ce": 8.891787729226053e-05,
      "loss_iou": 0.099609375,
      "loss_num": 0.00543212890625,
      "loss_xval": 0.2255859375,
      "num_input_tokens_seen": 482546336,
      "step": 8612
    },
    {
      "epoch": 19.182628062360802,
      "grad_norm": 16.697389602661133,
      "learning_rate": 1e-06,
      "loss": 0.3748,
      "num_input_tokens_seen": 482603824,
      "step": 8613
    },
    {
      "epoch": 19.182628062360802,
      "loss": 0.37165093421936035,
      "loss_ce": 6.890263466630131e-05,
      "loss_iou": 0.1630859375,
      "loss_num": 0.00909423828125,
      "loss_xval": 0.37109375,
      "num_input_tokens_seen": 482603824,
      "step": 8613
    },
    {
      "epoch": 19.184855233853007,
      "grad_norm": 20.081663131713867,
      "learning_rate": 1e-06,
      "loss": 0.259,
      "num_input_tokens_seen": 482662992,
      "step": 8614
    },
    {
      "epoch": 19.184855233853007,
      "loss": 0.2634428143501282,
      "loss_ce": 7.611673936480656e-05,
      "loss_iou": 0.1103515625,
      "loss_num": 0.00848388671875,
      "loss_xval": 0.263671875,
      "num_input_tokens_seen": 482662992,
      "step": 8614
    },
    {
      "epoch": 19.187082405345212,
      "grad_norm": 15.964731216430664,
      "learning_rate": 1e-06,
      "loss": 0.3832,
      "num_input_tokens_seen": 482716832,
      "step": 8615
    },
    {
      "epoch": 19.187082405345212,
      "loss": 0.31769198179244995,
      "loss_ce": 6.505024066427723e-05,
      "loss_iou": 0.138671875,
      "loss_num": 0.00787353515625,
      "loss_xval": 0.318359375,
      "num_input_tokens_seen": 482716832,
      "step": 8615
    },
    {
      "epoch": 19.189309576837417,
      "grad_norm": 17.872699737548828,
      "learning_rate": 1e-06,
      "loss": 0.3562,
      "num_input_tokens_seen": 482771736,
      "step": 8616
    },
    {
      "epoch": 19.189309576837417,
      "loss": 0.36840832233428955,
      "loss_ce": 6.113122071838006e-05,
      "loss_iou": 0.1591796875,
      "loss_num": 0.00994873046875,
      "loss_xval": 0.369140625,
      "num_input_tokens_seen": 482771736,
      "step": 8616
    },
    {
      "epoch": 19.19153674832962,
      "grad_norm": 17.340930938720703,
      "learning_rate": 1e-06,
      "loss": 0.3884,
      "num_input_tokens_seen": 482826724,
      "step": 8617
    },
    {
      "epoch": 19.19153674832962,
      "loss": 0.4055081009864807,
      "loss_ce": 8.206171332858503e-05,
      "loss_iou": 0.1796875,
      "loss_num": 0.00946044921875,
      "loss_xval": 0.40625,
      "num_input_tokens_seen": 482826724,
      "step": 8617
    },
    {
      "epoch": 19.193763919821826,
      "grad_norm": 17.151872634887695,
      "learning_rate": 1e-06,
      "loss": 0.303,
      "num_input_tokens_seen": 482883452,
      "step": 8618
    },
    {
      "epoch": 19.193763919821826,
      "loss": 0.4231913685798645,
      "loss_ce": 9.566033986629918e-05,
      "loss_iou": 0.1875,
      "loss_num": 0.0096435546875,
      "loss_xval": 0.423828125,
      "num_input_tokens_seen": 482883452,
      "step": 8618
    },
    {
      "epoch": 19.19599109131403,
      "grad_norm": 17.975557327270508,
      "learning_rate": 1e-06,
      "loss": 0.3269,
      "num_input_tokens_seen": 482941656,
      "step": 8619
    },
    {
      "epoch": 19.19599109131403,
      "loss": 0.3718949258327484,
      "loss_ce": 6.874144310131669e-05,
      "loss_iou": 0.1611328125,
      "loss_num": 0.00982666015625,
      "loss_xval": 0.37109375,
      "num_input_tokens_seen": 482941656,
      "step": 8619
    },
    {
      "epoch": 19.198218262806236,
      "grad_norm": 20.601882934570312,
      "learning_rate": 1e-06,
      "loss": 0.3974,
      "num_input_tokens_seen": 482997328,
      "step": 8620
    },
    {
      "epoch": 19.198218262806236,
      "loss": 0.32132768630981445,
      "loss_ce": 5.390634396462701e-05,
      "loss_iou": 0.13671875,
      "loss_num": 0.00970458984375,
      "loss_xval": 0.3203125,
      "num_input_tokens_seen": 482997328,
      "step": 8620
    },
    {
      "epoch": 19.20044543429844,
      "grad_norm": 23.428096771240234,
      "learning_rate": 1e-06,
      "loss": 0.4218,
      "num_input_tokens_seen": 483051884,
      "step": 8621
    },
    {
      "epoch": 19.20044543429844,
      "loss": 0.28058868646621704,
      "loss_ce": 7.111984450602904e-05,
      "loss_iou": 0.1240234375,
      "loss_num": 0.006500244140625,
      "loss_xval": 0.28125,
      "num_input_tokens_seen": 483051884,
      "step": 8621
    },
    {
      "epoch": 19.202672605790646,
      "grad_norm": 24.112133026123047,
      "learning_rate": 1e-06,
      "loss": 0.3654,
      "num_input_tokens_seen": 483111104,
      "step": 8622
    },
    {
      "epoch": 19.202672605790646,
      "loss": 0.35199394822120667,
      "loss_ce": 6.522829062305391e-05,
      "loss_iou": 0.16015625,
      "loss_num": 0.0064697265625,
      "loss_xval": 0.3515625,
      "num_input_tokens_seen": 483111104,
      "step": 8622
    },
    {
      "epoch": 19.20489977728285,
      "grad_norm": 21.558853149414062,
      "learning_rate": 1e-06,
      "loss": 0.2897,
      "num_input_tokens_seen": 483165588,
      "step": 8623
    },
    {
      "epoch": 19.20489977728285,
      "loss": 0.2628226578235626,
      "loss_ce": 6.632453005295247e-05,
      "loss_iou": 0.11669921875,
      "loss_num": 0.00592041015625,
      "loss_xval": 0.263671875,
      "num_input_tokens_seen": 483165588,
      "step": 8623
    },
    {
      "epoch": 19.207126948775056,
      "grad_norm": 17.29518699645996,
      "learning_rate": 1e-06,
      "loss": 0.278,
      "num_input_tokens_seen": 483217036,
      "step": 8624
    },
    {
      "epoch": 19.207126948775056,
      "loss": 0.24366407096385956,
      "loss_ce": 7.275763346115127e-05,
      "loss_iou": 0.10302734375,
      "loss_num": 0.007415771484375,
      "loss_xval": 0.2431640625,
      "num_input_tokens_seen": 483217036,
      "step": 8624
    },
    {
      "epoch": 19.20935412026726,
      "grad_norm": 17.489004135131836,
      "learning_rate": 1e-06,
      "loss": 0.4856,
      "num_input_tokens_seen": 483274096,
      "step": 8625
    },
    {
      "epoch": 19.20935412026726,
      "loss": 0.37572720646858215,
      "loss_ce": 0.0001168523303931579,
      "loss_iou": 0.15625,
      "loss_num": 0.01251220703125,
      "loss_xval": 0.375,
      "num_input_tokens_seen": 483274096,
      "step": 8625
    },
    {
      "epoch": 19.211581291759465,
      "grad_norm": 17.45191192626953,
      "learning_rate": 1e-06,
      "loss": 0.5543,
      "num_input_tokens_seen": 483330112,
      "step": 8626
    },
    {
      "epoch": 19.211581291759465,
      "loss": 0.36878734827041626,
      "loss_ce": 7.397578156087548e-05,
      "loss_iou": 0.158203125,
      "loss_num": 0.01068115234375,
      "loss_xval": 0.369140625,
      "num_input_tokens_seen": 483330112,
      "step": 8626
    },
    {
      "epoch": 19.21380846325167,
      "grad_norm": 18.69688606262207,
      "learning_rate": 1e-06,
      "loss": 0.3022,
      "num_input_tokens_seen": 483388400,
      "step": 8627
    },
    {
      "epoch": 19.21380846325167,
      "loss": 0.24286232888698578,
      "loss_ce": 6.447385385399684e-05,
      "loss_iou": 0.103515625,
      "loss_num": 0.00701904296875,
      "loss_xval": 0.2431640625,
      "num_input_tokens_seen": 483388400,
      "step": 8627
    },
    {
      "epoch": 19.216035634743875,
      "grad_norm": 18.915224075317383,
      "learning_rate": 1e-06,
      "loss": 0.3235,
      "num_input_tokens_seen": 483444252,
      "step": 8628
    },
    {
      "epoch": 19.216035634743875,
      "loss": 0.40314409136772156,
      "loss_ce": 6.792263593524694e-05,
      "loss_iou": 0.1669921875,
      "loss_num": 0.013671875,
      "loss_xval": 0.40234375,
      "num_input_tokens_seen": 483444252,
      "step": 8628
    },
    {
      "epoch": 19.21826280623608,
      "grad_norm": 11.416555404663086,
      "learning_rate": 1e-06,
      "loss": 0.2597,
      "num_input_tokens_seen": 483503672,
      "step": 8629
    },
    {
      "epoch": 19.21826280623608,
      "loss": 0.2568409740924835,
      "loss_ce": 6.605684757232666e-05,
      "loss_iou": 0.1171875,
      "loss_num": 0.00457763671875,
      "loss_xval": 0.255859375,
      "num_input_tokens_seen": 483503672,
      "step": 8629
    },
    {
      "epoch": 19.220489977728285,
      "grad_norm": 21.57809066772461,
      "learning_rate": 1e-06,
      "loss": 0.4147,
      "num_input_tokens_seen": 483560072,
      "step": 8630
    },
    {
      "epoch": 19.220489977728285,
      "loss": 0.3244105577468872,
      "loss_ce": 6.974257121328264e-05,
      "loss_iou": 0.1298828125,
      "loss_num": 0.01300048828125,
      "loss_xval": 0.32421875,
      "num_input_tokens_seen": 483560072,
      "step": 8630
    },
    {
      "epoch": 19.22271714922049,
      "grad_norm": 16.324159622192383,
      "learning_rate": 1e-06,
      "loss": 0.2693,
      "num_input_tokens_seen": 483614548,
      "step": 8631
    },
    {
      "epoch": 19.22271714922049,
      "loss": 0.2832646369934082,
      "loss_ce": 6.152082642074674e-05,
      "loss_iou": 0.12060546875,
      "loss_num": 0.0084228515625,
      "loss_xval": 0.283203125,
      "num_input_tokens_seen": 483614548,
      "step": 8631
    },
    {
      "epoch": 19.224944320712694,
      "grad_norm": 29.336238861083984,
      "learning_rate": 1e-06,
      "loss": 0.3302,
      "num_input_tokens_seen": 483672292,
      "step": 8632
    },
    {
      "epoch": 19.224944320712694,
      "loss": 0.32135581970214844,
      "loss_ce": 6.674337782897055e-05,
      "loss_iou": 0.1494140625,
      "loss_num": 0.00439453125,
      "loss_xval": 0.3203125,
      "num_input_tokens_seen": 483672292,
      "step": 8632
    },
    {
      "epoch": 19.2271714922049,
      "grad_norm": 18.221179962158203,
      "learning_rate": 1e-06,
      "loss": 0.5706,
      "num_input_tokens_seen": 483729908,
      "step": 8633
    },
    {
      "epoch": 19.2271714922049,
      "loss": 0.6349157691001892,
      "loss_ce": 0.00015011939103715122,
      "loss_iou": 0.27734375,
      "loss_num": 0.016357421875,
      "loss_xval": 0.6328125,
      "num_input_tokens_seen": 483729908,
      "step": 8633
    },
    {
      "epoch": 19.229398663697104,
      "grad_norm": 18.27716636657715,
      "learning_rate": 1e-06,
      "loss": 0.4169,
      "num_input_tokens_seen": 483786140,
      "step": 8634
    },
    {
      "epoch": 19.229398663697104,
      "loss": 0.3237491250038147,
      "loss_ce": 7.969920261530206e-05,
      "loss_iou": 0.140625,
      "loss_num": 0.008544921875,
      "loss_xval": 0.32421875,
      "num_input_tokens_seen": 483786140,
      "step": 8634
    },
    {
      "epoch": 19.23162583518931,
      "grad_norm": 33.526123046875,
      "learning_rate": 1e-06,
      "loss": 0.3883,
      "num_input_tokens_seen": 483842292,
      "step": 8635
    },
    {
      "epoch": 19.23162583518931,
      "loss": 0.27155420184135437,
      "loss_ce": 6.981080514378846e-05,
      "loss_iou": 0.1201171875,
      "loss_num": 0.00616455078125,
      "loss_xval": 0.271484375,
      "num_input_tokens_seen": 483842292,
      "step": 8635
    },
    {
      "epoch": 19.233853006681514,
      "grad_norm": 12.381620407104492,
      "learning_rate": 1e-06,
      "loss": 0.3823,
      "num_input_tokens_seen": 483898136,
      "step": 8636
    },
    {
      "epoch": 19.233853006681514,
      "loss": 0.510744571685791,
      "loss_ce": 0.0001244788581971079,
      "loss_iou": 0.220703125,
      "loss_num": 0.0137939453125,
      "loss_xval": 0.51171875,
      "num_input_tokens_seen": 483898136,
      "step": 8636
    },
    {
      "epoch": 19.23608017817372,
      "grad_norm": 15.86150074005127,
      "learning_rate": 1e-06,
      "loss": 0.4379,
      "num_input_tokens_seen": 483952128,
      "step": 8637
    },
    {
      "epoch": 19.23608017817372,
      "loss": 0.4342408776283264,
      "loss_ce": 8.635166886961088e-05,
      "loss_iou": 0.189453125,
      "loss_num": 0.01092529296875,
      "loss_xval": 0.43359375,
      "num_input_tokens_seen": 483952128,
      "step": 8637
    },
    {
      "epoch": 19.238307349665924,
      "grad_norm": 22.681493759155273,
      "learning_rate": 1e-06,
      "loss": 0.347,
      "num_input_tokens_seen": 484007556,
      "step": 8638
    },
    {
      "epoch": 19.238307349665924,
      "loss": 0.26111987233161926,
      "loss_ce": 7.25125937606208e-05,
      "loss_iou": 0.1220703125,
      "loss_num": 0.0033721923828125,
      "loss_xval": 0.26171875,
      "num_input_tokens_seen": 484007556,
      "step": 8638
    },
    {
      "epoch": 19.24053452115813,
      "grad_norm": 16.731151580810547,
      "learning_rate": 1e-06,
      "loss": 0.3764,
      "num_input_tokens_seen": 484061948,
      "step": 8639
    },
    {
      "epoch": 19.24053452115813,
      "loss": 0.40643227100372314,
      "loss_ce": 6.019037391524762e-05,
      "loss_iou": 0.1884765625,
      "loss_num": 0.005859375,
      "loss_xval": 0.40625,
      "num_input_tokens_seen": 484061948,
      "step": 8639
    },
    {
      "epoch": 19.242761692650333,
      "grad_norm": 17.2427978515625,
      "learning_rate": 1e-06,
      "loss": 0.3552,
      "num_input_tokens_seen": 484118452,
      "step": 8640
    },
    {
      "epoch": 19.242761692650333,
      "loss": 0.3023113012313843,
      "loss_ce": 6.518626469187438e-05,
      "loss_iou": 0.130859375,
      "loss_num": 0.00799560546875,
      "loss_xval": 0.302734375,
      "num_input_tokens_seen": 484118452,
      "step": 8640
    },
    {
      "epoch": 19.244988864142538,
      "grad_norm": 48.943878173828125,
      "learning_rate": 1e-06,
      "loss": 0.3616,
      "num_input_tokens_seen": 484172552,
      "step": 8641
    },
    {
      "epoch": 19.244988864142538,
      "loss": 0.34272468090057373,
      "loss_ce": 7.331671076826751e-05,
      "loss_iou": 0.1416015625,
      "loss_num": 0.0115966796875,
      "loss_xval": 0.341796875,
      "num_input_tokens_seen": 484172552,
      "step": 8641
    },
    {
      "epoch": 19.247216035634743,
      "grad_norm": 14.787582397460938,
      "learning_rate": 1e-06,
      "loss": 0.3113,
      "num_input_tokens_seen": 484228180,
      "step": 8642
    },
    {
      "epoch": 19.247216035634743,
      "loss": 0.34266364574432373,
      "loss_ce": 7.330120570259169e-05,
      "loss_iou": 0.16015625,
      "loss_num": 0.00457763671875,
      "loss_xval": 0.341796875,
      "num_input_tokens_seen": 484228180,
      "step": 8642
    },
    {
      "epoch": 19.249443207126948,
      "grad_norm": 18.2416934967041,
      "learning_rate": 1e-06,
      "loss": 0.4346,
      "num_input_tokens_seen": 484283708,
      "step": 8643
    },
    {
      "epoch": 19.249443207126948,
      "loss": 0.37335968017578125,
      "loss_ce": 6.869265052955598e-05,
      "loss_iou": 0.1474609375,
      "loss_num": 0.015625,
      "loss_xval": 0.373046875,
      "num_input_tokens_seen": 484283708,
      "step": 8643
    },
    {
      "epoch": 19.251670378619153,
      "grad_norm": 18.17558479309082,
      "learning_rate": 1e-06,
      "loss": 0.3687,
      "num_input_tokens_seen": 484337716,
      "step": 8644
    },
    {
      "epoch": 19.251670378619153,
      "loss": 0.259723424911499,
      "loss_ce": 6.458204006776214e-05,
      "loss_iou": 0.11962890625,
      "loss_num": 0.004119873046875,
      "loss_xval": 0.259765625,
      "num_input_tokens_seen": 484337716,
      "step": 8644
    },
    {
      "epoch": 19.253897550111358,
      "grad_norm": 14.284744262695312,
      "learning_rate": 1e-06,
      "loss": 0.6006,
      "num_input_tokens_seen": 484392308,
      "step": 8645
    },
    {
      "epoch": 19.253897550111358,
      "loss": 0.6434794664382935,
      "loss_ce": 7.73589126765728e-05,
      "loss_iou": 0.25390625,
      "loss_num": 0.02734375,
      "loss_xval": 0.64453125,
      "num_input_tokens_seen": 484392308,
      "step": 8645
    },
    {
      "epoch": 19.256124721603562,
      "grad_norm": 12.745560646057129,
      "learning_rate": 1e-06,
      "loss": 0.2475,
      "num_input_tokens_seen": 484446740,
      "step": 8646
    },
    {
      "epoch": 19.256124721603562,
      "loss": 0.2596558630466461,
      "loss_ce": 7.333945541176945e-05,
      "loss_iou": 0.119140625,
      "loss_num": 0.0042724609375,
      "loss_xval": 0.259765625,
      "num_input_tokens_seen": 484446740,
      "step": 8646
    },
    {
      "epoch": 19.258351893095767,
      "grad_norm": 19.38439178466797,
      "learning_rate": 1e-06,
      "loss": 0.4594,
      "num_input_tokens_seen": 484501160,
      "step": 8647
    },
    {
      "epoch": 19.258351893095767,
      "loss": 0.4618569016456604,
      "loss_ce": 6.492799002444372e-05,
      "loss_iou": 0.20703125,
      "loss_num": 0.009765625,
      "loss_xval": 0.4609375,
      "num_input_tokens_seen": 484501160,
      "step": 8647
    },
    {
      "epoch": 19.260579064587972,
      "grad_norm": 14.868861198425293,
      "learning_rate": 1e-06,
      "loss": 0.5964,
      "num_input_tokens_seen": 484556848,
      "step": 8648
    },
    {
      "epoch": 19.260579064587972,
      "loss": 0.8090330362319946,
      "loss_ce": 7.303664460778236e-05,
      "loss_iou": 0.3359375,
      "loss_num": 0.02783203125,
      "loss_xval": 0.80859375,
      "num_input_tokens_seen": 484556848,
      "step": 8648
    },
    {
      "epoch": 19.262806236080177,
      "grad_norm": 17.061763763427734,
      "learning_rate": 1e-06,
      "loss": 0.3803,
      "num_input_tokens_seen": 484614472,
      "step": 8649
    },
    {
      "epoch": 19.262806236080177,
      "loss": 0.4142407476902008,
      "loss_ce": 5.6169934396166354e-05,
      "loss_iou": 0.1669921875,
      "loss_num": 0.0159912109375,
      "loss_xval": 0.4140625,
      "num_input_tokens_seen": 484614472,
      "step": 8649
    },
    {
      "epoch": 19.265033407572382,
      "grad_norm": 19.03131675720215,
      "learning_rate": 1e-06,
      "loss": 0.283,
      "num_input_tokens_seen": 484667276,
      "step": 8650
    },
    {
      "epoch": 19.265033407572382,
      "loss": 0.3287457227706909,
      "loss_ce": 0.0001019124174490571,
      "loss_iou": 0.1474609375,
      "loss_num": 0.0068359375,
      "loss_xval": 0.328125,
      "num_input_tokens_seen": 484667276,
      "step": 8650
    },
    {
      "epoch": 19.267260579064587,
      "grad_norm": 42.60155487060547,
      "learning_rate": 1e-06,
      "loss": 0.4224,
      "num_input_tokens_seen": 484723500,
      "step": 8651
    },
    {
      "epoch": 19.267260579064587,
      "loss": 0.37189292907714844,
      "loss_ce": 6.674042379017919e-05,
      "loss_iou": 0.154296875,
      "loss_num": 0.01275634765625,
      "loss_xval": 0.37109375,
      "num_input_tokens_seen": 484723500,
      "step": 8651
    },
    {
      "epoch": 19.26948775055679,
      "grad_norm": 17.580562591552734,
      "learning_rate": 1e-06,
      "loss": 0.363,
      "num_input_tokens_seen": 484779672,
      "step": 8652
    },
    {
      "epoch": 19.26948775055679,
      "loss": 0.2955964207649231,
      "loss_ce": 6.419201235985383e-05,
      "loss_iou": 0.130859375,
      "loss_num": 0.00701904296875,
      "loss_xval": 0.294921875,
      "num_input_tokens_seen": 484779672,
      "step": 8652
    },
    {
      "epoch": 19.271714922048996,
      "grad_norm": 17.12259292602539,
      "learning_rate": 1e-06,
      "loss": 0.4862,
      "num_input_tokens_seen": 484836188,
      "step": 8653
    },
    {
      "epoch": 19.271714922048996,
      "loss": 0.3255062699317932,
      "loss_ce": 6.683074752800167e-05,
      "loss_iou": 0.146484375,
      "loss_num": 0.00628662109375,
      "loss_xval": 0.326171875,
      "num_input_tokens_seen": 484836188,
      "step": 8653
    },
    {
      "epoch": 19.2739420935412,
      "grad_norm": 21.847190856933594,
      "learning_rate": 1e-06,
      "loss": 0.2709,
      "num_input_tokens_seen": 484895900,
      "step": 8654
    },
    {
      "epoch": 19.2739420935412,
      "loss": 0.25446265935897827,
      "loss_ce": 6.814206426497549e-05,
      "loss_iou": 0.115234375,
      "loss_num": 0.0047607421875,
      "loss_xval": 0.25390625,
      "num_input_tokens_seen": 484895900,
      "step": 8654
    },
    {
      "epoch": 19.276169265033406,
      "grad_norm": 13.837854385375977,
      "learning_rate": 1e-06,
      "loss": 0.3591,
      "num_input_tokens_seen": 484953396,
      "step": 8655
    },
    {
      "epoch": 19.276169265033406,
      "loss": 0.28464746475219727,
      "loss_ce": 0.00010157265205634758,
      "loss_iou": 0.123046875,
      "loss_num": 0.00775146484375,
      "loss_xval": 0.28515625,
      "num_input_tokens_seen": 484953396,
      "step": 8655
    },
    {
      "epoch": 19.27839643652561,
      "grad_norm": 18.63173484802246,
      "learning_rate": 1e-06,
      "loss": 0.5888,
      "num_input_tokens_seen": 485004972,
      "step": 8656
    },
    {
      "epoch": 19.27839643652561,
      "loss": 0.5696535706520081,
      "loss_ce": 7.348039071075618e-05,
      "loss_iou": 0.25390625,
      "loss_num": 0.0126953125,
      "loss_xval": 0.5703125,
      "num_input_tokens_seen": 485004972,
      "step": 8656
    },
    {
      "epoch": 19.280623608017816,
      "grad_norm": 23.69566535949707,
      "learning_rate": 1e-06,
      "loss": 0.4021,
      "num_input_tokens_seen": 485061276,
      "step": 8657
    },
    {
      "epoch": 19.280623608017816,
      "loss": 0.4415389895439148,
      "loss_ce": 0.00013270846102386713,
      "loss_iou": 0.2021484375,
      "loss_num": 0.00738525390625,
      "loss_xval": 0.44140625,
      "num_input_tokens_seen": 485061276,
      "step": 8657
    },
    {
      "epoch": 19.28285077951002,
      "grad_norm": 16.324665069580078,
      "learning_rate": 1e-06,
      "loss": 0.5027,
      "num_input_tokens_seen": 485116212,
      "step": 8658
    },
    {
      "epoch": 19.28285077951002,
      "loss": 0.656080961227417,
      "loss_ce": 7.512497541029006e-05,
      "loss_iou": 0.283203125,
      "loss_num": 0.0177001953125,
      "loss_xval": 0.65625,
      "num_input_tokens_seen": 485116212,
      "step": 8658
    },
    {
      "epoch": 19.285077951002226,
      "grad_norm": 25.834796905517578,
      "learning_rate": 1e-06,
      "loss": 0.5209,
      "num_input_tokens_seen": 485171484,
      "step": 8659
    },
    {
      "epoch": 19.285077951002226,
      "loss": 0.42792704701423645,
      "loss_ce": 7.059163181111217e-05,
      "loss_iou": 0.189453125,
      "loss_num": 0.009765625,
      "loss_xval": 0.427734375,
      "num_input_tokens_seen": 485171484,
      "step": 8659
    },
    {
      "epoch": 19.28730512249443,
      "grad_norm": 14.427014350891113,
      "learning_rate": 1e-06,
      "loss": 0.434,
      "num_input_tokens_seen": 485229056,
      "step": 8660
    },
    {
      "epoch": 19.28730512249443,
      "loss": 0.596036434173584,
      "loss_ce": 8.911389159038663e-05,
      "loss_iou": 0.236328125,
      "loss_num": 0.0247802734375,
      "loss_xval": 0.59765625,
      "num_input_tokens_seen": 485229056,
      "step": 8660
    },
    {
      "epoch": 19.289532293986635,
      "grad_norm": 56.55402374267578,
      "learning_rate": 1e-06,
      "loss": 0.5668,
      "num_input_tokens_seen": 485286492,
      "step": 8661
    },
    {
      "epoch": 19.289532293986635,
      "loss": 0.6791332960128784,
      "loss_ce": 5.614275869447738e-05,
      "loss_iou": 0.267578125,
      "loss_num": 0.0284423828125,
      "loss_xval": 0.6796875,
      "num_input_tokens_seen": 485286492,
      "step": 8661
    },
    {
      "epoch": 19.29175946547884,
      "grad_norm": 16.25436019897461,
      "learning_rate": 1e-06,
      "loss": 0.3742,
      "num_input_tokens_seen": 485343172,
      "step": 8662
    },
    {
      "epoch": 19.29175946547884,
      "loss": 0.30635008215904236,
      "loss_ce": 7.566015119664371e-05,
      "loss_iou": 0.1357421875,
      "loss_num": 0.007049560546875,
      "loss_xval": 0.306640625,
      "num_input_tokens_seen": 485343172,
      "step": 8662
    },
    {
      "epoch": 19.293986636971045,
      "grad_norm": 17.356847763061523,
      "learning_rate": 1e-06,
      "loss": 0.3913,
      "num_input_tokens_seen": 485399704,
      "step": 8663
    },
    {
      "epoch": 19.293986636971045,
      "loss": 0.4219398498535156,
      "loss_ce": 6.486824713647366e-05,
      "loss_iou": 0.173828125,
      "loss_num": 0.01470947265625,
      "loss_xval": 0.421875,
      "num_input_tokens_seen": 485399704,
      "step": 8663
    },
    {
      "epoch": 19.29621380846325,
      "grad_norm": 21.03586769104004,
      "learning_rate": 1e-06,
      "loss": 0.2966,
      "num_input_tokens_seen": 485455500,
      "step": 8664
    },
    {
      "epoch": 19.29621380846325,
      "loss": 0.22834303975105286,
      "loss_ce": 7.155879575293511e-05,
      "loss_iou": 0.103515625,
      "loss_num": 0.0042724609375,
      "loss_xval": 0.228515625,
      "num_input_tokens_seen": 485455500,
      "step": 8664
    },
    {
      "epoch": 19.29844097995546,
      "grad_norm": 25.01719093322754,
      "learning_rate": 1e-06,
      "loss": 0.4601,
      "num_input_tokens_seen": 485511580,
      "step": 8665
    },
    {
      "epoch": 19.29844097995546,
      "loss": 0.5397475361824036,
      "loss_ce": 7.467882824130356e-05,
      "loss_iou": 0.259765625,
      "loss_num": 0.004241943359375,
      "loss_xval": 0.5390625,
      "num_input_tokens_seen": 485511580,
      "step": 8665
    },
    {
      "epoch": 19.30066815144766,
      "grad_norm": 15.293524742126465,
      "learning_rate": 1e-06,
      "loss": 0.4385,
      "num_input_tokens_seen": 485567524,
      "step": 8666
    },
    {
      "epoch": 19.30066815144766,
      "loss": 0.29761672019958496,
      "loss_ce": 7.03330006217584e-05,
      "loss_iou": 0.1279296875,
      "loss_num": 0.00830078125,
      "loss_xval": 0.296875,
      "num_input_tokens_seen": 485567524,
      "step": 8666
    },
    {
      "epoch": 19.302895322939868,
      "grad_norm": 24.937509536743164,
      "learning_rate": 1e-06,
      "loss": 0.4014,
      "num_input_tokens_seen": 485623744,
      "step": 8667
    },
    {
      "epoch": 19.302895322939868,
      "loss": 0.5748975872993469,
      "loss_ce": 6.84759725118056e-05,
      "loss_iou": 0.251953125,
      "loss_num": 0.013916015625,
      "loss_xval": 0.57421875,
      "num_input_tokens_seen": 485623744,
      "step": 8667
    },
    {
      "epoch": 19.305122494432073,
      "grad_norm": 16.04351234436035,
      "learning_rate": 1e-06,
      "loss": 0.4303,
      "num_input_tokens_seen": 485680980,
      "step": 8668
    },
    {
      "epoch": 19.305122494432073,
      "loss": 0.310863733291626,
      "loss_ce": 7.271443610079587e-05,
      "loss_iou": 0.12890625,
      "loss_num": 0.01043701171875,
      "loss_xval": 0.310546875,
      "num_input_tokens_seen": 485680980,
      "step": 8668
    },
    {
      "epoch": 19.307349665924278,
      "grad_norm": 34.41288757324219,
      "learning_rate": 1e-06,
      "loss": 0.5049,
      "num_input_tokens_seen": 485738648,
      "step": 8669
    },
    {
      "epoch": 19.307349665924278,
      "loss": 0.6703619956970215,
      "loss_ce": 7.39232636988163e-05,
      "loss_iou": 0.302734375,
      "loss_num": 0.01287841796875,
      "loss_xval": 0.671875,
      "num_input_tokens_seen": 485738648,
      "step": 8669
    },
    {
      "epoch": 19.309576837416483,
      "grad_norm": 16.024730682373047,
      "learning_rate": 1e-06,
      "loss": 0.4571,
      "num_input_tokens_seen": 485795136,
      "step": 8670
    },
    {
      "epoch": 19.309576837416483,
      "loss": 0.3097517490386963,
      "loss_ce": 5.93543445575051e-05,
      "loss_iou": 0.1318359375,
      "loss_num": 0.00921630859375,
      "loss_xval": 0.310546875,
      "num_input_tokens_seen": 485795136,
      "step": 8670
    },
    {
      "epoch": 19.311804008908688,
      "grad_norm": 22.233224868774414,
      "learning_rate": 1e-06,
      "loss": 0.3574,
      "num_input_tokens_seen": 485853408,
      "step": 8671
    },
    {
      "epoch": 19.311804008908688,
      "loss": 0.45661628246307373,
      "loss_ce": 7.331735105253756e-05,
      "loss_iou": 0.21875,
      "loss_num": 0.00396728515625,
      "loss_xval": 0.45703125,
      "num_input_tokens_seen": 485853408,
      "step": 8671
    },
    {
      "epoch": 19.314031180400892,
      "grad_norm": 21.287254333496094,
      "learning_rate": 1e-06,
      "loss": 0.2966,
      "num_input_tokens_seen": 485908808,
      "step": 8672
    },
    {
      "epoch": 19.314031180400892,
      "loss": 0.31123119592666626,
      "loss_ce": 7.398641901090741e-05,
      "loss_iou": 0.146484375,
      "loss_num": 0.003875732421875,
      "loss_xval": 0.310546875,
      "num_input_tokens_seen": 485908808,
      "step": 8672
    },
    {
      "epoch": 19.316258351893097,
      "grad_norm": 17.1425838470459,
      "learning_rate": 1e-06,
      "loss": 0.4107,
      "num_input_tokens_seen": 485965732,
      "step": 8673
    },
    {
      "epoch": 19.316258351893097,
      "loss": 0.4722324013710022,
      "loss_ce": 6.443218444474041e-05,
      "loss_iou": 0.1982421875,
      "loss_num": 0.01495361328125,
      "loss_xval": 0.47265625,
      "num_input_tokens_seen": 485965732,
      "step": 8673
    },
    {
      "epoch": 19.318485523385302,
      "grad_norm": 23.668569564819336,
      "learning_rate": 1e-06,
      "loss": 0.4553,
      "num_input_tokens_seen": 486019256,
      "step": 8674
    },
    {
      "epoch": 19.318485523385302,
      "loss": 0.5221450328826904,
      "loss_ce": 8.080543193500489e-05,
      "loss_iou": 0.2138671875,
      "loss_num": 0.0189208984375,
      "loss_xval": 0.5234375,
      "num_input_tokens_seen": 486019256,
      "step": 8674
    },
    {
      "epoch": 19.320712694877507,
      "grad_norm": 19.682819366455078,
      "learning_rate": 1e-06,
      "loss": 0.2918,
      "num_input_tokens_seen": 486077160,
      "step": 8675
    },
    {
      "epoch": 19.320712694877507,
      "loss": 0.2836154103279114,
      "loss_ce": 7.660247501917183e-05,
      "loss_iou": 0.12158203125,
      "loss_num": 0.0079345703125,
      "loss_xval": 0.283203125,
      "num_input_tokens_seen": 486077160,
      "step": 8675
    },
    {
      "epoch": 19.322939866369712,
      "grad_norm": 19.56645965576172,
      "learning_rate": 1e-06,
      "loss": 0.4519,
      "num_input_tokens_seen": 486132592,
      "step": 8676
    },
    {
      "epoch": 19.322939866369712,
      "loss": 0.4211920499801636,
      "loss_ce": 0.00011051179171772674,
      "loss_iou": 0.166015625,
      "loss_num": 0.0177001953125,
      "loss_xval": 0.421875,
      "num_input_tokens_seen": 486132592,
      "step": 8676
    },
    {
      "epoch": 19.325167037861917,
      "grad_norm": 10.575581550598145,
      "learning_rate": 1e-06,
      "loss": 0.3449,
      "num_input_tokens_seen": 486190616,
      "step": 8677
    },
    {
      "epoch": 19.325167037861917,
      "loss": 0.2473614513874054,
      "loss_ce": 6.224659591680393e-05,
      "loss_iou": 0.1123046875,
      "loss_num": 0.0045166015625,
      "loss_xval": 0.2470703125,
      "num_input_tokens_seen": 486190616,
      "step": 8677
    },
    {
      "epoch": 19.32739420935412,
      "grad_norm": 16.035282135009766,
      "learning_rate": 1e-06,
      "loss": 0.4254,
      "num_input_tokens_seen": 486245220,
      "step": 8678
    },
    {
      "epoch": 19.32739420935412,
      "loss": 0.3443076014518738,
      "loss_ce": 6.930766539881006e-05,
      "loss_iou": 0.15625,
      "loss_num": 0.006195068359375,
      "loss_xval": 0.34375,
      "num_input_tokens_seen": 486245220,
      "step": 8678
    },
    {
      "epoch": 19.329621380846326,
      "grad_norm": 17.30709457397461,
      "learning_rate": 1e-06,
      "loss": 0.4013,
      "num_input_tokens_seen": 486302496,
      "step": 8679
    },
    {
      "epoch": 19.329621380846326,
      "loss": 0.3727584779262543,
      "loss_ce": 7.78008543420583e-05,
      "loss_iou": 0.16796875,
      "loss_num": 0.007110595703125,
      "loss_xval": 0.373046875,
      "num_input_tokens_seen": 486302496,
      "step": 8679
    },
    {
      "epoch": 19.33184855233853,
      "grad_norm": 20.775163650512695,
      "learning_rate": 1e-06,
      "loss": 0.415,
      "num_input_tokens_seen": 486356500,
      "step": 8680
    },
    {
      "epoch": 19.33184855233853,
      "loss": 0.4452681541442871,
      "loss_ce": 7.77466339059174e-05,
      "loss_iou": 0.2099609375,
      "loss_num": 0.005218505859375,
      "loss_xval": 0.4453125,
      "num_input_tokens_seen": 486356500,
      "step": 8680
    },
    {
      "epoch": 19.334075723830736,
      "grad_norm": 35.834449768066406,
      "learning_rate": 1e-06,
      "loss": 0.2574,
      "num_input_tokens_seen": 486413916,
      "step": 8681
    },
    {
      "epoch": 19.334075723830736,
      "loss": 0.27562782168388367,
      "loss_ce": 6.935855344636366e-05,
      "loss_iou": 0.11669921875,
      "loss_num": 0.0084228515625,
      "loss_xval": 0.275390625,
      "num_input_tokens_seen": 486413916,
      "step": 8681
    },
    {
      "epoch": 19.33630289532294,
      "grad_norm": 14.752124786376953,
      "learning_rate": 1e-06,
      "loss": 0.3223,
      "num_input_tokens_seen": 486471340,
      "step": 8682
    },
    {
      "epoch": 19.33630289532294,
      "loss": 0.43706992268562317,
      "loss_ce": 5.819134821649641e-05,
      "loss_iou": 0.1826171875,
      "loss_num": 0.01446533203125,
      "loss_xval": 0.4375,
      "num_input_tokens_seen": 486471340,
      "step": 8682
    },
    {
      "epoch": 19.338530066815146,
      "grad_norm": 19.141727447509766,
      "learning_rate": 1e-06,
      "loss": 0.3472,
      "num_input_tokens_seen": 486529184,
      "step": 8683
    },
    {
      "epoch": 19.338530066815146,
      "loss": 0.2464015781879425,
      "loss_ce": 6.367703463183716e-05,
      "loss_iou": 0.109375,
      "loss_num": 0.005645751953125,
      "loss_xval": 0.24609375,
      "num_input_tokens_seen": 486529184,
      "step": 8683
    },
    {
      "epoch": 19.34075723830735,
      "grad_norm": 19.265857696533203,
      "learning_rate": 1e-06,
      "loss": 0.2633,
      "num_input_tokens_seen": 486584540,
      "step": 8684
    },
    {
      "epoch": 19.34075723830735,
      "loss": 0.24787020683288574,
      "loss_ce": 6.746564758941531e-05,
      "loss_iou": 0.11181640625,
      "loss_num": 0.0048828125,
      "loss_xval": 0.248046875,
      "num_input_tokens_seen": 486584540,
      "step": 8684
    },
    {
      "epoch": 19.342984409799556,
      "grad_norm": 64.30094909667969,
      "learning_rate": 1e-06,
      "loss": 0.4085,
      "num_input_tokens_seen": 486639632,
      "step": 8685
    },
    {
      "epoch": 19.342984409799556,
      "loss": 0.3603741526603699,
      "loss_ce": 8.36373510537669e-05,
      "loss_iou": 0.1689453125,
      "loss_num": 0.004547119140625,
      "loss_xval": 0.359375,
      "num_input_tokens_seen": 486639632,
      "step": 8685
    },
    {
      "epoch": 19.34521158129176,
      "grad_norm": 22.54205322265625,
      "learning_rate": 1e-06,
      "loss": 0.4647,
      "num_input_tokens_seen": 486693520,
      "step": 8686
    },
    {
      "epoch": 19.34521158129176,
      "loss": 0.39545318484306335,
      "loss_ce": 6.7443739681039e-05,
      "loss_iou": 0.173828125,
      "loss_num": 0.00970458984375,
      "loss_xval": 0.39453125,
      "num_input_tokens_seen": 486693520,
      "step": 8686
    },
    {
      "epoch": 19.347438752783965,
      "grad_norm": 19.74683380126953,
      "learning_rate": 1e-06,
      "loss": 0.5212,
      "num_input_tokens_seen": 486747984,
      "step": 8687
    },
    {
      "epoch": 19.347438752783965,
      "loss": 0.47200244665145874,
      "loss_ce": 0.00044482320663519204,
      "loss_iou": 0.1962890625,
      "loss_num": 0.01556396484375,
      "loss_xval": 0.470703125,
      "num_input_tokens_seen": 486747984,
      "step": 8687
    },
    {
      "epoch": 19.34966592427617,
      "grad_norm": 26.414714813232422,
      "learning_rate": 1e-06,
      "loss": 0.3239,
      "num_input_tokens_seen": 486803160,
      "step": 8688
    },
    {
      "epoch": 19.34966592427617,
      "loss": 0.30175623297691345,
      "loss_ce": 5.944071017438546e-05,
      "loss_iou": 0.1259765625,
      "loss_num": 0.00982666015625,
      "loss_xval": 0.30078125,
      "num_input_tokens_seen": 486803160,
      "step": 8688
    },
    {
      "epoch": 19.351893095768375,
      "grad_norm": 17.674320220947266,
      "learning_rate": 1e-06,
      "loss": 0.4622,
      "num_input_tokens_seen": 486861400,
      "step": 8689
    },
    {
      "epoch": 19.351893095768375,
      "loss": 0.45283421874046326,
      "loss_ce": 7.542921230196953e-05,
      "loss_iou": 0.181640625,
      "loss_num": 0.017822265625,
      "loss_xval": 0.453125,
      "num_input_tokens_seen": 486861400,
      "step": 8689
    },
    {
      "epoch": 19.35412026726058,
      "grad_norm": 24.226484298706055,
      "learning_rate": 1e-06,
      "loss": 0.5093,
      "num_input_tokens_seen": 486913828,
      "step": 8690
    },
    {
      "epoch": 19.35412026726058,
      "loss": 0.6187217235565186,
      "loss_ce": 6.936366116860881e-05,
      "loss_iou": 0.283203125,
      "loss_num": 0.01068115234375,
      "loss_xval": 0.6171875,
      "num_input_tokens_seen": 486913828,
      "step": 8690
    },
    {
      "epoch": 19.356347438752785,
      "grad_norm": 22.290658950805664,
      "learning_rate": 1e-06,
      "loss": 0.4345,
      "num_input_tokens_seen": 486968612,
      "step": 8691
    },
    {
      "epoch": 19.356347438752785,
      "loss": 0.3384547233581543,
      "loss_ce": 7.58438982302323e-05,
      "loss_iou": 0.1552734375,
      "loss_num": 0.00567626953125,
      "loss_xval": 0.337890625,
      "num_input_tokens_seen": 486968612,
      "step": 8691
    },
    {
      "epoch": 19.35857461024499,
      "grad_norm": 17.18849754333496,
      "learning_rate": 1e-06,
      "loss": 0.3679,
      "num_input_tokens_seen": 487024476,
      "step": 8692
    },
    {
      "epoch": 19.35857461024499,
      "loss": 0.46119123697280884,
      "loss_ce": 0.00013169522571843117,
      "loss_iou": 0.2041015625,
      "loss_num": 0.01068115234375,
      "loss_xval": 0.4609375,
      "num_input_tokens_seen": 487024476,
      "step": 8692
    },
    {
      "epoch": 19.360801781737194,
      "grad_norm": 21.173423767089844,
      "learning_rate": 1e-06,
      "loss": 0.3566,
      "num_input_tokens_seen": 487081268,
      "step": 8693
    },
    {
      "epoch": 19.360801781737194,
      "loss": 0.43475276231765747,
      "loss_ce": 6.03883781877812e-05,
      "loss_iou": 0.1796875,
      "loss_num": 0.01519775390625,
      "loss_xval": 0.435546875,
      "num_input_tokens_seen": 487081268,
      "step": 8693
    },
    {
      "epoch": 19.3630289532294,
      "grad_norm": 19.107248306274414,
      "learning_rate": 1e-06,
      "loss": 0.3304,
      "num_input_tokens_seen": 487133808,
      "step": 8694
    },
    {
      "epoch": 19.3630289532294,
      "loss": 0.30717021226882935,
      "loss_ce": 5.656223220285028e-05,
      "loss_iou": 0.1396484375,
      "loss_num": 0.005706787109375,
      "loss_xval": 0.306640625,
      "num_input_tokens_seen": 487133808,
      "step": 8694
    },
    {
      "epoch": 19.365256124721604,
      "grad_norm": 18.290464401245117,
      "learning_rate": 1e-06,
      "loss": 0.3325,
      "num_input_tokens_seen": 487191508,
      "step": 8695
    },
    {
      "epoch": 19.365256124721604,
      "loss": 0.41119277477264404,
      "loss_ce": 5.9931116993539035e-05,
      "loss_iou": 0.1884765625,
      "loss_num": 0.0067138671875,
      "loss_xval": 0.41015625,
      "num_input_tokens_seen": 487191508,
      "step": 8695
    },
    {
      "epoch": 19.36748329621381,
      "grad_norm": 20.532798767089844,
      "learning_rate": 1e-06,
      "loss": 0.4025,
      "num_input_tokens_seen": 487248060,
      "step": 8696
    },
    {
      "epoch": 19.36748329621381,
      "loss": 0.24934418499469757,
      "loss_ce": 7.660340634174645e-05,
      "loss_iou": 0.11328125,
      "loss_num": 0.004486083984375,
      "loss_xval": 0.2490234375,
      "num_input_tokens_seen": 487248060,
      "step": 8696
    },
    {
      "epoch": 19.369710467706014,
      "grad_norm": 21.59139633178711,
      "learning_rate": 1e-06,
      "loss": 0.4072,
      "num_input_tokens_seen": 487304516,
      "step": 8697
    },
    {
      "epoch": 19.369710467706014,
      "loss": 0.503758430480957,
      "loss_ce": 9.630977001506835e-05,
      "loss_iou": 0.2060546875,
      "loss_num": 0.0181884765625,
      "loss_xval": 0.50390625,
      "num_input_tokens_seen": 487304516,
      "step": 8697
    },
    {
      "epoch": 19.37193763919822,
      "grad_norm": 18.260494232177734,
      "learning_rate": 1e-06,
      "loss": 0.3184,
      "num_input_tokens_seen": 487360040,
      "step": 8698
    },
    {
      "epoch": 19.37193763919822,
      "loss": 0.3482136130332947,
      "loss_ce": 6.908161594765261e-05,
      "loss_iou": 0.1494140625,
      "loss_num": 0.009765625,
      "loss_xval": 0.34765625,
      "num_input_tokens_seen": 487360040,
      "step": 8698
    },
    {
      "epoch": 19.374164810690424,
      "grad_norm": 13.985121726989746,
      "learning_rate": 1e-06,
      "loss": 0.2731,
      "num_input_tokens_seen": 487413428,
      "step": 8699
    },
    {
      "epoch": 19.374164810690424,
      "loss": 0.23252707719802856,
      "loss_ce": 7.46890582377091e-05,
      "loss_iou": 0.107421875,
      "loss_num": 0.00347900390625,
      "loss_xval": 0.232421875,
      "num_input_tokens_seen": 487413428,
      "step": 8699
    },
    {
      "epoch": 19.37639198218263,
      "grad_norm": 23.42201805114746,
      "learning_rate": 1e-06,
      "loss": 0.4771,
      "num_input_tokens_seen": 487468832,
      "step": 8700
    },
    {
      "epoch": 19.37639198218263,
      "loss": 0.4661286175251007,
      "loss_ce": 6.414036033675075e-05,
      "loss_iou": 0.2060546875,
      "loss_num": 0.0107421875,
      "loss_xval": 0.466796875,
      "num_input_tokens_seen": 487468832,
      "step": 8700
    },
    {
      "epoch": 19.378619153674833,
      "grad_norm": 18.14457893371582,
      "learning_rate": 1e-06,
      "loss": 0.3369,
      "num_input_tokens_seen": 487525960,
      "step": 8701
    },
    {
      "epoch": 19.378619153674833,
      "loss": 0.3606569170951843,
      "loss_ce": 0.00012223681551404297,
      "loss_iou": 0.14453125,
      "loss_num": 0.01446533203125,
      "loss_xval": 0.361328125,
      "num_input_tokens_seen": 487525960,
      "step": 8701
    },
    {
      "epoch": 19.380846325167038,
      "grad_norm": 11.536683082580566,
      "learning_rate": 1e-06,
      "loss": 0.3009,
      "num_input_tokens_seen": 487583988,
      "step": 8702
    },
    {
      "epoch": 19.380846325167038,
      "loss": 0.3344113230705261,
      "loss_ce": 6.077022408135235e-05,
      "loss_iou": 0.1337890625,
      "loss_num": 0.01318359375,
      "loss_xval": 0.333984375,
      "num_input_tokens_seen": 487583988,
      "step": 8702
    },
    {
      "epoch": 19.383073496659243,
      "grad_norm": 21.613056182861328,
      "learning_rate": 1e-06,
      "loss": 0.3386,
      "num_input_tokens_seen": 487636664,
      "step": 8703
    },
    {
      "epoch": 19.383073496659243,
      "loss": 0.3694436550140381,
      "loss_ce": 8.942555723479018e-05,
      "loss_iou": 0.14453125,
      "loss_num": 0.0157470703125,
      "loss_xval": 0.369140625,
      "num_input_tokens_seen": 487636664,
      "step": 8703
    },
    {
      "epoch": 19.385300668151448,
      "grad_norm": 22.000041961669922,
      "learning_rate": 1e-06,
      "loss": 0.3079,
      "num_input_tokens_seen": 487694000,
      "step": 8704
    },
    {
      "epoch": 19.385300668151448,
      "loss": 0.23017969727516174,
      "loss_ce": 7.716739492025226e-05,
      "loss_iou": 0.10693359375,
      "loss_num": 0.003326416015625,
      "loss_xval": 0.23046875,
      "num_input_tokens_seen": 487694000,
      "step": 8704
    },
    {
      "epoch": 19.387527839643653,
      "grad_norm": 13.974677085876465,
      "learning_rate": 1e-06,
      "loss": 0.3866,
      "num_input_tokens_seen": 487749452,
      "step": 8705
    },
    {
      "epoch": 19.387527839643653,
      "loss": 0.27674055099487305,
      "loss_ce": 6.818344991188496e-05,
      "loss_iou": 0.12890625,
      "loss_num": 0.003692626953125,
      "loss_xval": 0.27734375,
      "num_input_tokens_seen": 487749452,
      "step": 8705
    },
    {
      "epoch": 19.389755011135858,
      "grad_norm": 24.663671493530273,
      "learning_rate": 1e-06,
      "loss": 0.3754,
      "num_input_tokens_seen": 487803040,
      "step": 8706
    },
    {
      "epoch": 19.389755011135858,
      "loss": 0.4255967140197754,
      "loss_ce": 5.9576763305813074e-05,
      "loss_iou": 0.19140625,
      "loss_num": 0.00836181640625,
      "loss_xval": 0.42578125,
      "num_input_tokens_seen": 487803040,
      "step": 8706
    },
    {
      "epoch": 19.391982182628063,
      "grad_norm": 15.723613739013672,
      "learning_rate": 1e-06,
      "loss": 0.385,
      "num_input_tokens_seen": 487861660,
      "step": 8707
    },
    {
      "epoch": 19.391982182628063,
      "loss": 0.502314031124115,
      "loss_ce": 0.0001167724549304694,
      "loss_iou": 0.2216796875,
      "loss_num": 0.0115966796875,
      "loss_xval": 0.50390625,
      "num_input_tokens_seen": 487861660,
      "step": 8707
    },
    {
      "epoch": 19.394209354120267,
      "grad_norm": 18.70146369934082,
      "learning_rate": 1e-06,
      "loss": 0.4727,
      "num_input_tokens_seen": 487917728,
      "step": 8708
    },
    {
      "epoch": 19.394209354120267,
      "loss": 0.4696964621543884,
      "loss_ce": 9.19793383218348e-05,
      "loss_iou": 0.197265625,
      "loss_num": 0.0150146484375,
      "loss_xval": 0.46875,
      "num_input_tokens_seen": 487917728,
      "step": 8708
    },
    {
      "epoch": 19.396436525612472,
      "grad_norm": 18.742565155029297,
      "learning_rate": 1e-06,
      "loss": 0.5473,
      "num_input_tokens_seen": 487972684,
      "step": 8709
    },
    {
      "epoch": 19.396436525612472,
      "loss": 0.47677797079086304,
      "loss_ce": 9.340511314803734e-05,
      "loss_iou": 0.2119140625,
      "loss_num": 0.01068115234375,
      "loss_xval": 0.4765625,
      "num_input_tokens_seen": 487972684,
      "step": 8709
    },
    {
      "epoch": 19.398663697104677,
      "grad_norm": 79.24798583984375,
      "learning_rate": 1e-06,
      "loss": 0.477,
      "num_input_tokens_seen": 488027984,
      "step": 8710
    },
    {
      "epoch": 19.398663697104677,
      "loss": 0.7484385967254639,
      "loss_ce": 8.648347284179181e-05,
      "loss_iou": 0.271484375,
      "loss_num": 0.040771484375,
      "loss_xval": 0.75,
      "num_input_tokens_seen": 488027984,
      "step": 8710
    },
    {
      "epoch": 19.400890868596882,
      "grad_norm": 23.097761154174805,
      "learning_rate": 1e-06,
      "loss": 0.4981,
      "num_input_tokens_seen": 488084492,
      "step": 8711
    },
    {
      "epoch": 19.400890868596882,
      "loss": 0.5120543241500854,
      "loss_ce": 9.138373570749536e-05,
      "loss_iou": 0.2021484375,
      "loss_num": 0.021728515625,
      "loss_xval": 0.51171875,
      "num_input_tokens_seen": 488084492,
      "step": 8711
    },
    {
      "epoch": 19.403118040089087,
      "grad_norm": 26.856395721435547,
      "learning_rate": 1e-06,
      "loss": 0.5109,
      "num_input_tokens_seen": 488139008,
      "step": 8712
    },
    {
      "epoch": 19.403118040089087,
      "loss": 0.4314088821411133,
      "loss_ce": 7.342306344071403e-05,
      "loss_iou": 0.193359375,
      "loss_num": 0.00885009765625,
      "loss_xval": 0.431640625,
      "num_input_tokens_seen": 488139008,
      "step": 8712
    },
    {
      "epoch": 19.40534521158129,
      "grad_norm": 17.314449310302734,
      "learning_rate": 1e-06,
      "loss": 0.443,
      "num_input_tokens_seen": 488193600,
      "step": 8713
    },
    {
      "epoch": 19.40534521158129,
      "loss": 0.39160478115081787,
      "loss_ce": 6.424939056159928e-05,
      "loss_iou": 0.1748046875,
      "loss_num": 0.00823974609375,
      "loss_xval": 0.390625,
      "num_input_tokens_seen": 488193600,
      "step": 8713
    },
    {
      "epoch": 19.407572383073497,
      "grad_norm": 25.597942352294922,
      "learning_rate": 1e-06,
      "loss": 0.2987,
      "num_input_tokens_seen": 488251452,
      "step": 8714
    },
    {
      "epoch": 19.407572383073497,
      "loss": 0.3611512780189514,
      "loss_ce": 6.729191227350384e-05,
      "loss_iou": 0.16015625,
      "loss_num": 0.00811767578125,
      "loss_xval": 0.361328125,
      "num_input_tokens_seen": 488251452,
      "step": 8714
    },
    {
      "epoch": 19.4097995545657,
      "grad_norm": 25.324542999267578,
      "learning_rate": 1e-06,
      "loss": 0.4442,
      "num_input_tokens_seen": 488307252,
      "step": 8715
    },
    {
      "epoch": 19.4097995545657,
      "loss": 0.5760635137557983,
      "loss_ce": 7.472602010238916e-05,
      "loss_iou": 0.2392578125,
      "loss_num": 0.019287109375,
      "loss_xval": 0.57421875,
      "num_input_tokens_seen": 488307252,
      "step": 8715
    },
    {
      "epoch": 19.412026726057906,
      "grad_norm": 17.964107513427734,
      "learning_rate": 1e-06,
      "loss": 0.3626,
      "num_input_tokens_seen": 488363092,
      "step": 8716
    },
    {
      "epoch": 19.412026726057906,
      "loss": 0.3571329712867737,
      "loss_ce": 7.73169522290118e-05,
      "loss_iou": 0.1572265625,
      "loss_num": 0.00836181640625,
      "loss_xval": 0.357421875,
      "num_input_tokens_seen": 488363092,
      "step": 8716
    },
    {
      "epoch": 19.41425389755011,
      "grad_norm": 16.54916763305664,
      "learning_rate": 1e-06,
      "loss": 0.2583,
      "num_input_tokens_seen": 488420924,
      "step": 8717
    },
    {
      "epoch": 19.41425389755011,
      "loss": 0.1723698377609253,
      "loss_ce": 6.760148971807212e-05,
      "loss_iou": 0.068359375,
      "loss_num": 0.00714111328125,
      "loss_xval": 0.171875,
      "num_input_tokens_seen": 488420924,
      "step": 8717
    },
    {
      "epoch": 19.416481069042316,
      "grad_norm": 31.5208683013916,
      "learning_rate": 1e-06,
      "loss": 0.5982,
      "num_input_tokens_seen": 488473012,
      "step": 8718
    },
    {
      "epoch": 19.416481069042316,
      "loss": 0.48655349016189575,
      "loss_ce": 0.00010326504707336426,
      "loss_iou": 0.20703125,
      "loss_num": 0.01434326171875,
      "loss_xval": 0.486328125,
      "num_input_tokens_seen": 488473012,
      "step": 8718
    },
    {
      "epoch": 19.41870824053452,
      "grad_norm": 14.7725248336792,
      "learning_rate": 1e-06,
      "loss": 0.3455,
      "num_input_tokens_seen": 488529484,
      "step": 8719
    },
    {
      "epoch": 19.41870824053452,
      "loss": 0.34601306915283203,
      "loss_ce": 6.580428453162313e-05,
      "loss_iou": 0.16015625,
      "loss_num": 0.005126953125,
      "loss_xval": 0.345703125,
      "num_input_tokens_seen": 488529484,
      "step": 8719
    },
    {
      "epoch": 19.420935412026726,
      "grad_norm": 18.437286376953125,
      "learning_rate": 1e-06,
      "loss": 0.384,
      "num_input_tokens_seen": 488587032,
      "step": 8720
    },
    {
      "epoch": 19.420935412026726,
      "loss": 0.30788394808769226,
      "loss_ce": 0.0001446868700440973,
      "loss_iou": 0.1259765625,
      "loss_num": 0.0113525390625,
      "loss_xval": 0.30859375,
      "num_input_tokens_seen": 488587032,
      "step": 8720
    },
    {
      "epoch": 19.42316258351893,
      "grad_norm": 18.742013931274414,
      "learning_rate": 1e-06,
      "loss": 0.6236,
      "num_input_tokens_seen": 488643196,
      "step": 8721
    },
    {
      "epoch": 19.42316258351893,
      "loss": 0.742760181427002,
      "loss_ce": 8.442148100584745e-05,
      "loss_iou": 0.298828125,
      "loss_num": 0.029052734375,
      "loss_xval": 0.7421875,
      "num_input_tokens_seen": 488643196,
      "step": 8721
    },
    {
      "epoch": 19.425389755011135,
      "grad_norm": 20.891321182250977,
      "learning_rate": 1e-06,
      "loss": 0.4322,
      "num_input_tokens_seen": 488699628,
      "step": 8722
    },
    {
      "epoch": 19.425389755011135,
      "loss": 0.44647514820098877,
      "loss_ce": 6.399239646270871e-05,
      "loss_iou": 0.20703125,
      "loss_num": 0.006561279296875,
      "loss_xval": 0.447265625,
      "num_input_tokens_seen": 488699628,
      "step": 8722
    },
    {
      "epoch": 19.42761692650334,
      "grad_norm": 11.5533447265625,
      "learning_rate": 1e-06,
      "loss": 0.4345,
      "num_input_tokens_seen": 488756332,
      "step": 8723
    },
    {
      "epoch": 19.42761692650334,
      "loss": 0.4649144411087036,
      "loss_ce": 7.067194383125752e-05,
      "loss_iou": 0.1884765625,
      "loss_num": 0.0174560546875,
      "loss_xval": 0.46484375,
      "num_input_tokens_seen": 488756332,
      "step": 8723
    },
    {
      "epoch": 19.429844097995545,
      "grad_norm": 27.625247955322266,
      "learning_rate": 1e-06,
      "loss": 0.3431,
      "num_input_tokens_seen": 488812856,
      "step": 8724
    },
    {
      "epoch": 19.429844097995545,
      "loss": 0.4067014753818512,
      "loss_ce": 8.526656165486202e-05,
      "loss_iou": 0.17578125,
      "loss_num": 0.01080322265625,
      "loss_xval": 0.40625,
      "num_input_tokens_seen": 488812856,
      "step": 8724
    },
    {
      "epoch": 19.43207126948775,
      "grad_norm": 17.141159057617188,
      "learning_rate": 1e-06,
      "loss": 0.3232,
      "num_input_tokens_seen": 488866352,
      "step": 8725
    },
    {
      "epoch": 19.43207126948775,
      "loss": 0.34912025928497314,
      "loss_ce": 6.021084118401632e-05,
      "loss_iou": 0.150390625,
      "loss_num": 0.00982666015625,
      "loss_xval": 0.349609375,
      "num_input_tokens_seen": 488866352,
      "step": 8725
    },
    {
      "epoch": 19.434298440979955,
      "grad_norm": 18.241275787353516,
      "learning_rate": 1e-06,
      "loss": 0.36,
      "num_input_tokens_seen": 488925556,
      "step": 8726
    },
    {
      "epoch": 19.434298440979955,
      "loss": 0.3274751305580139,
      "loss_ce": 8.253618580056354e-05,
      "loss_iou": 0.1513671875,
      "loss_num": 0.004791259765625,
      "loss_xval": 0.328125,
      "num_input_tokens_seen": 488925556,
      "step": 8726
    },
    {
      "epoch": 19.43652561247216,
      "grad_norm": 61.344459533691406,
      "learning_rate": 1e-06,
      "loss": 0.3267,
      "num_input_tokens_seen": 488983256,
      "step": 8727
    },
    {
      "epoch": 19.43652561247216,
      "loss": 0.4625932276248932,
      "loss_ce": 6.8813213147223e-05,
      "loss_iou": 0.197265625,
      "loss_num": 0.01336669921875,
      "loss_xval": 0.462890625,
      "num_input_tokens_seen": 488983256,
      "step": 8727
    },
    {
      "epoch": 19.438752783964365,
      "grad_norm": 24.339759826660156,
      "learning_rate": 1e-06,
      "loss": 0.4207,
      "num_input_tokens_seen": 489039964,
      "step": 8728
    },
    {
      "epoch": 19.438752783964365,
      "loss": 0.4445319175720215,
      "loss_ce": 7.392082625301555e-05,
      "loss_iou": 0.2099609375,
      "loss_num": 0.0047607421875,
      "loss_xval": 0.4453125,
      "num_input_tokens_seen": 489039964,
      "step": 8728
    },
    {
      "epoch": 19.44097995545657,
      "grad_norm": 14.037641525268555,
      "learning_rate": 1e-06,
      "loss": 0.4557,
      "num_input_tokens_seen": 489095200,
      "step": 8729
    },
    {
      "epoch": 19.44097995545657,
      "loss": 0.36377742886543274,
      "loss_ce": 6.892348756082356e-05,
      "loss_iou": 0.140625,
      "loss_num": 0.016357421875,
      "loss_xval": 0.36328125,
      "num_input_tokens_seen": 489095200,
      "step": 8729
    },
    {
      "epoch": 19.443207126948774,
      "grad_norm": 30.27907943725586,
      "learning_rate": 1e-06,
      "loss": 0.3208,
      "num_input_tokens_seen": 489150700,
      "step": 8730
    },
    {
      "epoch": 19.443207126948774,
      "loss": 0.3462543785572052,
      "loss_ce": 6.298153311945498e-05,
      "loss_iou": 0.1572265625,
      "loss_num": 0.006317138671875,
      "loss_xval": 0.345703125,
      "num_input_tokens_seen": 489150700,
      "step": 8730
    },
    {
      "epoch": 19.44543429844098,
      "grad_norm": 13.437280654907227,
      "learning_rate": 1e-06,
      "loss": 0.512,
      "num_input_tokens_seen": 489206188,
      "step": 8731
    },
    {
      "epoch": 19.44543429844098,
      "loss": 0.36061540246009827,
      "loss_ce": 8.075297955656424e-05,
      "loss_iou": 0.1552734375,
      "loss_num": 0.01007080078125,
      "loss_xval": 0.361328125,
      "num_input_tokens_seen": 489206188,
      "step": 8731
    },
    {
      "epoch": 19.447661469933184,
      "grad_norm": 23.991952896118164,
      "learning_rate": 1e-06,
      "loss": 0.3874,
      "num_input_tokens_seen": 489260872,
      "step": 8732
    },
    {
      "epoch": 19.447661469933184,
      "loss": 0.3317917287349701,
      "loss_ce": 6.565573858097196e-05,
      "loss_iou": 0.14453125,
      "loss_num": 0.00848388671875,
      "loss_xval": 0.33203125,
      "num_input_tokens_seen": 489260872,
      "step": 8732
    },
    {
      "epoch": 19.44988864142539,
      "grad_norm": 14.908693313598633,
      "learning_rate": 1e-06,
      "loss": 0.3623,
      "num_input_tokens_seen": 489317248,
      "step": 8733
    },
    {
      "epoch": 19.44988864142539,
      "loss": 0.30453062057495117,
      "loss_ce": 8.725294173927978e-05,
      "loss_iou": 0.134765625,
      "loss_num": 0.007080078125,
      "loss_xval": 0.3046875,
      "num_input_tokens_seen": 489317248,
      "step": 8733
    },
    {
      "epoch": 19.452115812917594,
      "grad_norm": 72.13795471191406,
      "learning_rate": 1e-06,
      "loss": 0.2718,
      "num_input_tokens_seen": 489374240,
      "step": 8734
    },
    {
      "epoch": 19.452115812917594,
      "loss": 0.3349186182022095,
      "loss_ce": 7.976653432706371e-05,
      "loss_iou": 0.1572265625,
      "loss_num": 0.004302978515625,
      "loss_xval": 0.333984375,
      "num_input_tokens_seen": 489374240,
      "step": 8734
    },
    {
      "epoch": 19.4543429844098,
      "grad_norm": 15.769805908203125,
      "learning_rate": 1e-06,
      "loss": 0.3651,
      "num_input_tokens_seen": 489432756,
      "step": 8735
    },
    {
      "epoch": 19.4543429844098,
      "loss": 0.3257424235343933,
      "loss_ce": 5.8816825912799686e-05,
      "loss_iou": 0.1328125,
      "loss_num": 0.01202392578125,
      "loss_xval": 0.326171875,
      "num_input_tokens_seen": 489432756,
      "step": 8735
    },
    {
      "epoch": 19.456570155902003,
      "grad_norm": 134.32081604003906,
      "learning_rate": 1e-06,
      "loss": 0.3294,
      "num_input_tokens_seen": 489488532,
      "step": 8736
    },
    {
      "epoch": 19.456570155902003,
      "loss": 0.30285748839378357,
      "loss_ce": 6.207545811776072e-05,
      "loss_iou": 0.1337890625,
      "loss_num": 0.0068359375,
      "loss_xval": 0.302734375,
      "num_input_tokens_seen": 489488532,
      "step": 8736
    },
    {
      "epoch": 19.45879732739421,
      "grad_norm": 19.14018440246582,
      "learning_rate": 1e-06,
      "loss": 0.4107,
      "num_input_tokens_seen": 489541320,
      "step": 8737
    },
    {
      "epoch": 19.45879732739421,
      "loss": 0.36432066559791565,
      "loss_ce": 6.283754191827029e-05,
      "loss_iou": 0.1611328125,
      "loss_num": 0.00836181640625,
      "loss_xval": 0.36328125,
      "num_input_tokens_seen": 489541320,
      "step": 8737
    },
    {
      "epoch": 19.461024498886413,
      "grad_norm": 17.099512100219727,
      "learning_rate": 1e-06,
      "loss": 0.3985,
      "num_input_tokens_seen": 489597048,
      "step": 8738
    },
    {
      "epoch": 19.461024498886413,
      "loss": 0.45394012331962585,
      "loss_ce": 8.270963735412806e-05,
      "loss_iou": 0.1884765625,
      "loss_num": 0.01544189453125,
      "loss_xval": 0.453125,
      "num_input_tokens_seen": 489597048,
      "step": 8738
    },
    {
      "epoch": 19.463251670378618,
      "grad_norm": 17.23607063293457,
      "learning_rate": 1e-06,
      "loss": 0.44,
      "num_input_tokens_seen": 489653544,
      "step": 8739
    },
    {
      "epoch": 19.463251670378618,
      "loss": 0.47176241874694824,
      "loss_ce": 8.272690320154652e-05,
      "loss_iou": 0.212890625,
      "loss_num": 0.009033203125,
      "loss_xval": 0.47265625,
      "num_input_tokens_seen": 489653544,
      "step": 8739
    },
    {
      "epoch": 19.465478841870823,
      "grad_norm": 25.947065353393555,
      "learning_rate": 1e-06,
      "loss": 0.5785,
      "num_input_tokens_seen": 489710516,
      "step": 8740
    },
    {
      "epoch": 19.465478841870823,
      "loss": 0.40167826414108276,
      "loss_ce": 6.691856833640486e-05,
      "loss_iou": 0.1806640625,
      "loss_num": 0.00811767578125,
      "loss_xval": 0.40234375,
      "num_input_tokens_seen": 489710516,
      "step": 8740
    },
    {
      "epoch": 19.467706013363028,
      "grad_norm": 15.991206169128418,
      "learning_rate": 1e-06,
      "loss": 0.2839,
      "num_input_tokens_seen": 489766984,
      "step": 8741
    },
    {
      "epoch": 19.467706013363028,
      "loss": 0.25690385699272156,
      "loss_ce": 6.792375643271953e-05,
      "loss_iou": 0.1142578125,
      "loss_num": 0.0057373046875,
      "loss_xval": 0.2578125,
      "num_input_tokens_seen": 489766984,
      "step": 8741
    },
    {
      "epoch": 19.469933184855233,
      "grad_norm": 12.759182929992676,
      "learning_rate": 1e-06,
      "loss": 0.3062,
      "num_input_tokens_seen": 489823144,
      "step": 8742
    },
    {
      "epoch": 19.469933184855233,
      "loss": 0.3527446389198303,
      "loss_ce": 0.0001140242675319314,
      "loss_iou": 0.15625,
      "loss_num": 0.0079345703125,
      "loss_xval": 0.353515625,
      "num_input_tokens_seen": 489823144,
      "step": 8742
    },
    {
      "epoch": 19.472160356347437,
      "grad_norm": 18.29888343811035,
      "learning_rate": 1e-06,
      "loss": 0.336,
      "num_input_tokens_seen": 489878156,
      "step": 8743
    },
    {
      "epoch": 19.472160356347437,
      "loss": 0.3941093683242798,
      "loss_ce": 6.639507773797959e-05,
      "loss_iou": 0.1796875,
      "loss_num": 0.007232666015625,
      "loss_xval": 0.39453125,
      "num_input_tokens_seen": 489878156,
      "step": 8743
    },
    {
      "epoch": 19.474387527839642,
      "grad_norm": 27.575145721435547,
      "learning_rate": 1e-06,
      "loss": 0.3742,
      "num_input_tokens_seen": 489932784,
      "step": 8744
    },
    {
      "epoch": 19.474387527839642,
      "loss": 0.45381438732147217,
      "loss_ce": 7.904722588136792e-05,
      "loss_iou": 0.1962890625,
      "loss_num": 0.01214599609375,
      "loss_xval": 0.453125,
      "num_input_tokens_seen": 489932784,
      "step": 8744
    },
    {
      "epoch": 19.476614699331847,
      "grad_norm": 16.64528465270996,
      "learning_rate": 1e-06,
      "loss": 0.3247,
      "num_input_tokens_seen": 489988008,
      "step": 8745
    },
    {
      "epoch": 19.476614699331847,
      "loss": 0.37039196491241455,
      "loss_ce": 7.641837873961776e-05,
      "loss_iou": 0.12353515625,
      "loss_num": 0.024658203125,
      "loss_xval": 0.37109375,
      "num_input_tokens_seen": 489988008,
      "step": 8745
    },
    {
      "epoch": 19.478841870824052,
      "grad_norm": 12.457158088684082,
      "learning_rate": 1e-06,
      "loss": 0.4706,
      "num_input_tokens_seen": 490045656,
      "step": 8746
    },
    {
      "epoch": 19.478841870824052,
      "loss": 0.576514482498169,
      "loss_ce": 9.850240166997537e-05,
      "loss_iou": 0.212890625,
      "loss_num": 0.0301513671875,
      "loss_xval": 0.578125,
      "num_input_tokens_seen": 490045656,
      "step": 8746
    },
    {
      "epoch": 19.481069042316257,
      "grad_norm": 29.269418716430664,
      "learning_rate": 1e-06,
      "loss": 0.5014,
      "num_input_tokens_seen": 490100304,
      "step": 8747
    },
    {
      "epoch": 19.481069042316257,
      "loss": 0.4481244385242462,
      "loss_ce": 6.533482519444078e-05,
      "loss_iou": 0.1982421875,
      "loss_num": 0.01025390625,
      "loss_xval": 0.447265625,
      "num_input_tokens_seen": 490100304,
      "step": 8747
    },
    {
      "epoch": 19.48329621380846,
      "grad_norm": 17.77977752685547,
      "learning_rate": 1e-06,
      "loss": 0.3385,
      "num_input_tokens_seen": 490155632,
      "step": 8748
    },
    {
      "epoch": 19.48329621380846,
      "loss": 0.24175959825515747,
      "loss_ce": 6.037494677002542e-05,
      "loss_iou": 0.1015625,
      "loss_num": 0.00762939453125,
      "loss_xval": 0.2421875,
      "num_input_tokens_seen": 490155632,
      "step": 8748
    },
    {
      "epoch": 19.485523385300667,
      "grad_norm": 16.200801849365234,
      "learning_rate": 1e-06,
      "loss": 0.3294,
      "num_input_tokens_seen": 490213296,
      "step": 8749
    },
    {
      "epoch": 19.485523385300667,
      "loss": 0.3568875789642334,
      "loss_ce": 7.605206337757409e-05,
      "loss_iou": 0.1669921875,
      "loss_num": 0.0047607421875,
      "loss_xval": 0.357421875,
      "num_input_tokens_seen": 490213296,
      "step": 8749
    },
    {
      "epoch": 19.48775055679287,
      "grad_norm": 15.862672805786133,
      "learning_rate": 1e-06,
      "loss": 0.3418,
      "num_input_tokens_seen": 490266248,
      "step": 8750
    },
    {
      "epoch": 19.48775055679287,
      "eval_seeclick_web_CIoU": 0.5831195414066315,
      "eval_seeclick_web_GIoU": 0.5824593603610992,
      "eval_seeclick_web_IoU": 0.6025447845458984,
      "eval_seeclick_web_MAE_all": 0.0153358387760818,
      "eval_seeclick_web_MAE_h": 0.0069596245884895325,
      "eval_seeclick_web_MAE_w": 0.015038002282381058,
      "eval_seeclick_web_MAE_x_boxes": 0.008386612171307206,
      "eval_seeclick_web_MAE_y_boxes": 0.02156838239170611,
      "eval_seeclick_web_inside_bbox": 0.9010416567325592,
      "eval_seeclick_web_loss": 0.9198112487792969,
      "eval_seeclick_web_loss_ce": 0.00012025472460663877,
      "eval_seeclick_web_loss_iou": 0.4251708984375,
      "eval_seeclick_web_loss_num": 0.012319564819335938,
      "eval_seeclick_web_loss_xval": 0.912109375,
      "eval_seeclick_web_runtime": 21.1555,
      "eval_seeclick_web_samples_per_second": 2.363,
      "eval_seeclick_web_steps_per_second": 0.095,
      "num_input_tokens_seen": 490266248,
      "step": 8750
    },
    {
      "epoch": 19.48775055679287,
      "eval_icons_CIoU": 0.26400046050548553,
      "eval_icons_GIoU": 0.299440860748291,
      "eval_icons_IoU": 0.34618473052978516,
      "eval_icons_MAE_all": 0.062128059566020966,
      "eval_icons_MAE_h": 0.029708989895880222,
      "eval_icons_MAE_w": 0.06808132492005825,
      "eval_icons_MAE_x_boxes": 0.06109212152659893,
      "eval_icons_MAE_y_boxes": 0.03715028055012226,
      "eval_icons_inside_bbox": 0.59375,
      "eval_icons_loss": 1.7176454067230225,
      "eval_icons_loss_ce": 0.00013828581722918898,
      "eval_icons_loss_iou": 0.673095703125,
      "eval_icons_loss_num": 0.06012725830078125,
      "eval_icons_loss_xval": 1.6474609375,
      "eval_icons_runtime": 18.377,
      "eval_icons_samples_per_second": 2.721,
      "eval_icons_steps_per_second": 0.109,
      "num_input_tokens_seen": 490266248,
      "step": 8750
    },
    {
      "epoch": 19.48775055679287,
      "eval_screenspot_CIoU": 0.37087904413541156,
      "eval_screenspot_GIoU": 0.3902689814567566,
      "eval_screenspot_IoU": 0.4464185933272044,
      "eval_screenspot_MAE_all": 0.058752829829851784,
      "eval_screenspot_MAE_h": 0.039602997402350106,
      "eval_screenspot_MAE_w": 0.06322793414195378,
      "eval_screenspot_MAE_x_boxes": 0.0708691483984391,
      "eval_screenspot_MAE_y_boxes": 0.04212713334709406,
      "eval_screenspot_inside_bbox": 0.7041666706403097,
      "eval_screenspot_loss": 1.5793300867080688,
      "eval_screenspot_loss_ce": 0.00015701642648006478,
      "eval_screenspot_loss_iou": 0.6509602864583334,
      "eval_screenspot_loss_num": 0.06783040364583333,
      "eval_screenspot_loss_xval": 1.640625,
      "eval_screenspot_runtime": 31.0289,
      "eval_screenspot_samples_per_second": 2.868,
      "eval_screenspot_steps_per_second": 0.097,
      "num_input_tokens_seen": 490266248,
      "step": 8750
    },
    {
      "epoch": 19.48775055679287,
      "eval_compot_CIoU": 0.34746964275836945,
      "eval_compot_GIoU": 0.3572629541158676,
      "eval_compot_IoU": 0.40467870235443115,
      "eval_compot_MAE_all": 0.018142362125217915,
      "eval_compot_MAE_h": 0.009449589531868696,
      "eval_compot_MAE_w": 0.020473646000027657,
      "eval_compot_MAE_x_boxes": 0.030018706806004047,
      "eval_compot_MAE_y_boxes": 0.006989206187427044,
      "eval_compot_inside_bbox": 0.6302083432674408,
      "eval_compot_loss": 1.3744853734970093,
      "eval_compot_loss_ce": 0.00010999454752891324,
      "eval_compot_loss_iou": 0.6365966796875,
      "eval_compot_loss_num": 0.016666412353515625,
      "eval_compot_loss_xval": 1.356689453125,
      "eval_compot_runtime": 19.0377,
      "eval_compot_samples_per_second": 2.626,
      "eval_compot_steps_per_second": 0.105,
      "num_input_tokens_seen": 490266248,
      "step": 8750
    },
    {
      "epoch": 19.48775055679287,
      "eval_custom_ui_val_CIoU": 0.4767077672812674,
      "eval_custom_ui_val_GIoU": 0.481793655289544,
      "eval_custom_ui_val_IoU": 0.5382914178901248,
      "eval_custom_ui_val_MAE_all": 0.026706857451548178,
      "eval_custom_ui_val_MAE_h": 0.013321074657142162,
      "eval_custom_ui_val_MAE_w": 0.03640370096804367,
      "eval_custom_ui_val_MAE_x_boxes": 0.032115884642634124,
      "eval_custom_ui_val_MAE_y_boxes": 0.013279020553454757,
      "eval_custom_ui_val_inside_bbox": 0.7754629651705424,
      "eval_custom_ui_val_loss": 1.1639938354492188,
      "eval_custom_ui_val_loss_ce": 0.00012037252761527068,
      "eval_custom_ui_val_loss_iou": 0.5000542534722222,
      "eval_custom_ui_val_loss_num": 0.023473739624023438,
      "eval_custom_ui_val_loss_xval": 1.1174858940972223,
      "eval_custom_ui_val_runtime": 57.2523,
      "eval_custom_ui_val_samples_per_second": 4.629,
      "eval_custom_ui_val_steps_per_second": 0.157,
      "num_input_tokens_seen": 490266248,
      "step": 8750
    },
    {
      "epoch": 19.48775055679287,
      "loss": 0.8289517164230347,
      "loss_ce": 9.427011536899954e-05,
      "loss_iou": 0.373046875,
      "loss_num": 0.0164794921875,
      "loss_xval": 0.828125,
      "num_input_tokens_seen": 490266248,
      "step": 8750
    },
    {
      "epoch": 19.489977728285076,
      "grad_norm": 16.564908981323242,
      "learning_rate": 1e-06,
      "loss": 0.3765,
      "num_input_tokens_seen": 490324400,
      "step": 8751
    },
    {
      "epoch": 19.489977728285076,
      "loss": 0.39094218611717224,
      "loss_ce": 7.304349128389731e-05,
      "loss_iou": 0.1787109375,
      "loss_num": 0.0068359375,
      "loss_xval": 0.390625,
      "num_input_tokens_seen": 490324400,
      "step": 8751
    },
    {
      "epoch": 19.49220489977728,
      "grad_norm": 18.350833892822266,
      "learning_rate": 1e-06,
      "loss": 0.4225,
      "num_input_tokens_seen": 490380904,
      "step": 8752
    },
    {
      "epoch": 19.49220489977728,
      "loss": 0.43350207805633545,
      "loss_ce": 6.09189628448803e-05,
      "loss_iou": 0.1826171875,
      "loss_num": 0.01361083984375,
      "loss_xval": 0.43359375,
      "num_input_tokens_seen": 490380904,
      "step": 8752
    },
    {
      "epoch": 19.494432071269486,
      "grad_norm": 21.1101016998291,
      "learning_rate": 1e-06,
      "loss": 0.5412,
      "num_input_tokens_seen": 490436388,
      "step": 8753
    },
    {
      "epoch": 19.494432071269486,
      "loss": 0.5999248027801514,
      "loss_ce": 7.128326979000121e-05,
      "loss_iou": 0.25,
      "loss_num": 0.0196533203125,
      "loss_xval": 0.6015625,
      "num_input_tokens_seen": 490436388,
      "step": 8753
    },
    {
      "epoch": 19.49665924276169,
      "grad_norm": 20.924692153930664,
      "learning_rate": 1e-06,
      "loss": 0.3887,
      "num_input_tokens_seen": 490491856,
      "step": 8754
    },
    {
      "epoch": 19.49665924276169,
      "loss": 0.5576558113098145,
      "loss_ce": 0.00028280686819925904,
      "loss_iou": 0.2470703125,
      "loss_num": 0.0128173828125,
      "loss_xval": 0.55859375,
      "num_input_tokens_seen": 490491856,
      "step": 8754
    },
    {
      "epoch": 19.498886414253896,
      "grad_norm": 16.07146644592285,
      "learning_rate": 1e-06,
      "loss": 0.3163,
      "num_input_tokens_seen": 490547508,
      "step": 8755
    },
    {
      "epoch": 19.498886414253896,
      "loss": 0.35533058643341064,
      "loss_ce": 7.546801498392597e-05,
      "loss_iou": 0.1630859375,
      "loss_num": 0.0059814453125,
      "loss_xval": 0.35546875,
      "num_input_tokens_seen": 490547508,
      "step": 8755
    },
    {
      "epoch": 19.501113585746104,
      "grad_norm": 17.8480167388916,
      "learning_rate": 1e-06,
      "loss": 0.3092,
      "num_input_tokens_seen": 490602244,
      "step": 8756
    },
    {
      "epoch": 19.501113585746104,
      "loss": 0.3217475414276123,
      "loss_ce": 9.226159454556182e-05,
      "loss_iou": 0.1416015625,
      "loss_num": 0.007598876953125,
      "loss_xval": 0.322265625,
      "num_input_tokens_seen": 490602244,
      "step": 8756
    },
    {
      "epoch": 19.50334075723831,
      "grad_norm": 20.913305282592773,
      "learning_rate": 1e-06,
      "loss": 0.4527,
      "num_input_tokens_seen": 490659044,
      "step": 8757
    },
    {
      "epoch": 19.50334075723831,
      "loss": 0.49548420310020447,
      "loss_ce": 6.180736818350852e-05,
      "loss_iou": 0.19140625,
      "loss_num": 0.0224609375,
      "loss_xval": 0.49609375,
      "num_input_tokens_seen": 490659044,
      "step": 8757
    },
    {
      "epoch": 19.505567928730514,
      "grad_norm": 18.693899154663086,
      "learning_rate": 1e-06,
      "loss": 0.2973,
      "num_input_tokens_seen": 490715388,
      "step": 8758
    },
    {
      "epoch": 19.505567928730514,
      "loss": 0.19214215874671936,
      "loss_ce": 6.452086381614208e-05,
      "loss_iou": 0.08056640625,
      "loss_num": 0.006256103515625,
      "loss_xval": 0.1923828125,
      "num_input_tokens_seen": 490715388,
      "step": 8758
    },
    {
      "epoch": 19.50779510022272,
      "grad_norm": 22.483903884887695,
      "learning_rate": 1e-06,
      "loss": 0.4002,
      "num_input_tokens_seen": 490771780,
      "step": 8759
    },
    {
      "epoch": 19.50779510022272,
      "loss": 0.40626299381256104,
      "loss_ce": 7.403266499750316e-05,
      "loss_iou": 0.181640625,
      "loss_num": 0.00860595703125,
      "loss_xval": 0.40625,
      "num_input_tokens_seen": 490771780,
      "step": 8759
    },
    {
      "epoch": 19.510022271714924,
      "grad_norm": 14.928621292114258,
      "learning_rate": 1e-06,
      "loss": 0.3059,
      "num_input_tokens_seen": 490829264,
      "step": 8760
    },
    {
      "epoch": 19.510022271714924,
      "loss": 0.32245999574661255,
      "loss_ce": 7.230706978589296e-05,
      "loss_iou": 0.1376953125,
      "loss_num": 0.00927734375,
      "loss_xval": 0.322265625,
      "num_input_tokens_seen": 490829264,
      "step": 8760
    },
    {
      "epoch": 19.51224944320713,
      "grad_norm": 40.57841491699219,
      "learning_rate": 1e-06,
      "loss": 0.4451,
      "num_input_tokens_seen": 490885112,
      "step": 8761
    },
    {
      "epoch": 19.51224944320713,
      "loss": 0.2985847294330597,
      "loss_ce": 6.179253978189081e-05,
      "loss_iou": 0.1279296875,
      "loss_num": 0.00830078125,
      "loss_xval": 0.298828125,
      "num_input_tokens_seen": 490885112,
      "step": 8761
    },
    {
      "epoch": 19.514476614699333,
      "grad_norm": 26.560928344726562,
      "learning_rate": 1e-06,
      "loss": 0.376,
      "num_input_tokens_seen": 490940464,
      "step": 8762
    },
    {
      "epoch": 19.514476614699333,
      "loss": 0.35675495862960815,
      "loss_ce": 6.548444798681885e-05,
      "loss_iou": 0.1640625,
      "loss_num": 0.00579833984375,
      "loss_xval": 0.357421875,
      "num_input_tokens_seen": 490940464,
      "step": 8762
    },
    {
      "epoch": 19.51670378619154,
      "grad_norm": 14.288774490356445,
      "learning_rate": 1e-06,
      "loss": 0.4027,
      "num_input_tokens_seen": 490995184,
      "step": 8763
    },
    {
      "epoch": 19.51670378619154,
      "loss": 0.2923039197921753,
      "loss_ce": 6.761118129361421e-05,
      "loss_iou": 0.12890625,
      "loss_num": 0.00677490234375,
      "loss_xval": 0.29296875,
      "num_input_tokens_seen": 490995184,
      "step": 8763
    },
    {
      "epoch": 19.518930957683743,
      "grad_norm": 23.1356201171875,
      "learning_rate": 1e-06,
      "loss": 0.3756,
      "num_input_tokens_seen": 491051012,
      "step": 8764
    },
    {
      "epoch": 19.518930957683743,
      "loss": 0.41393959522247314,
      "loss_ce": 6.0199585277587175e-05,
      "loss_iou": 0.16796875,
      "loss_num": 0.015380859375,
      "loss_xval": 0.4140625,
      "num_input_tokens_seen": 491051012,
      "step": 8764
    },
    {
      "epoch": 19.521158129175948,
      "grad_norm": 18.624935150146484,
      "learning_rate": 1e-06,
      "loss": 0.3166,
      "num_input_tokens_seen": 491107176,
      "step": 8765
    },
    {
      "epoch": 19.521158129175948,
      "loss": 0.34882843494415283,
      "loss_ce": 7.35686844564043e-05,
      "loss_iou": 0.16015625,
      "loss_num": 0.005462646484375,
      "loss_xval": 0.349609375,
      "num_input_tokens_seen": 491107176,
      "step": 8765
    },
    {
      "epoch": 19.523385300668153,
      "grad_norm": 20.93644905090332,
      "learning_rate": 1e-06,
      "loss": 0.4879,
      "num_input_tokens_seen": 491162216,
      "step": 8766
    },
    {
      "epoch": 19.523385300668153,
      "loss": 0.4118029773235321,
      "loss_ce": 5.981113645248115e-05,
      "loss_iou": 0.1875,
      "loss_num": 0.0074462890625,
      "loss_xval": 0.412109375,
      "num_input_tokens_seen": 491162216,
      "step": 8766
    },
    {
      "epoch": 19.525612472160358,
      "grad_norm": 15.544754028320312,
      "learning_rate": 1e-06,
      "loss": 0.4786,
      "num_input_tokens_seen": 491217424,
      "step": 8767
    },
    {
      "epoch": 19.525612472160358,
      "loss": 0.5115818381309509,
      "loss_ce": 0.00010719275451265275,
      "loss_iou": 0.2080078125,
      "loss_num": 0.0189208984375,
      "loss_xval": 0.51171875,
      "num_input_tokens_seen": 491217424,
      "step": 8767
    },
    {
      "epoch": 19.527839643652563,
      "grad_norm": 16.272380828857422,
      "learning_rate": 1e-06,
      "loss": 0.3691,
      "num_input_tokens_seen": 491269164,
      "step": 8768
    },
    {
      "epoch": 19.527839643652563,
      "loss": 0.326735258102417,
      "loss_ce": 7.508813723688945e-05,
      "loss_iou": 0.1435546875,
      "loss_num": 0.0078125,
      "loss_xval": 0.326171875,
      "num_input_tokens_seen": 491269164,
      "step": 8768
    },
    {
      "epoch": 19.530066815144767,
      "grad_norm": 19.87215232849121,
      "learning_rate": 1e-06,
      "loss": 0.276,
      "num_input_tokens_seen": 491328140,
      "step": 8769
    },
    {
      "epoch": 19.530066815144767,
      "loss": 0.25873589515686035,
      "loss_ce": 6.88969885231927e-05,
      "loss_iou": 0.1220703125,
      "loss_num": 0.002838134765625,
      "loss_xval": 0.2578125,
      "num_input_tokens_seen": 491328140,
      "step": 8769
    },
    {
      "epoch": 19.532293986636972,
      "grad_norm": 16.369861602783203,
      "learning_rate": 1e-06,
      "loss": 0.5755,
      "num_input_tokens_seen": 491381900,
      "step": 8770
    },
    {
      "epoch": 19.532293986636972,
      "loss": 0.7968376874923706,
      "loss_ce": 8.477165829390287e-05,
      "loss_iou": 0.310546875,
      "loss_num": 0.035400390625,
      "loss_xval": 0.796875,
      "num_input_tokens_seen": 491381900,
      "step": 8770
    },
    {
      "epoch": 19.534521158129177,
      "grad_norm": 29.245031356811523,
      "learning_rate": 1e-06,
      "loss": 0.4296,
      "num_input_tokens_seen": 491437320,
      "step": 8771
    },
    {
      "epoch": 19.534521158129177,
      "loss": 0.2833980321884155,
      "loss_ce": 7.282086880877614e-05,
      "loss_iou": 0.1318359375,
      "loss_num": 0.00396728515625,
      "loss_xval": 0.283203125,
      "num_input_tokens_seen": 491437320,
      "step": 8771
    },
    {
      "epoch": 19.536748329621382,
      "grad_norm": 27.237791061401367,
      "learning_rate": 1e-06,
      "loss": 0.4367,
      "num_input_tokens_seen": 491493612,
      "step": 8772
    },
    {
      "epoch": 19.536748329621382,
      "loss": 0.3653065860271454,
      "loss_ce": 7.220754923764616e-05,
      "loss_iou": 0.1689453125,
      "loss_num": 0.00543212890625,
      "loss_xval": 0.365234375,
      "num_input_tokens_seen": 491493612,
      "step": 8772
    },
    {
      "epoch": 19.538975501113587,
      "grad_norm": 52.4660758972168,
      "learning_rate": 1e-06,
      "loss": 0.3087,
      "num_input_tokens_seen": 491546616,
      "step": 8773
    },
    {
      "epoch": 19.538975501113587,
      "loss": 0.2954094111919403,
      "loss_ce": 6.026863775332458e-05,
      "loss_iou": 0.134765625,
      "loss_num": 0.00537109375,
      "loss_xval": 0.294921875,
      "num_input_tokens_seen": 491546616,
      "step": 8773
    },
    {
      "epoch": 19.54120267260579,
      "grad_norm": 20.992326736450195,
      "learning_rate": 1e-06,
      "loss": 0.4035,
      "num_input_tokens_seen": 491598516,
      "step": 8774
    },
    {
      "epoch": 19.54120267260579,
      "loss": 0.37043464183807373,
      "loss_ce": 7.328739593503997e-05,
      "loss_iou": 0.16015625,
      "loss_num": 0.010009765625,
      "loss_xval": 0.37109375,
      "num_input_tokens_seen": 491598516,
      "step": 8774
    },
    {
      "epoch": 19.543429844097997,
      "grad_norm": 17.51434898376465,
      "learning_rate": 1e-06,
      "loss": 0.3259,
      "num_input_tokens_seen": 491655988,
      "step": 8775
    },
    {
      "epoch": 19.543429844097997,
      "loss": 0.3194010257720947,
      "loss_ce": 6.507511716336012e-05,
      "loss_iou": 0.1455078125,
      "loss_num": 0.00555419921875,
      "loss_xval": 0.3203125,
      "num_input_tokens_seen": 491655988,
      "step": 8775
    },
    {
      "epoch": 19.5456570155902,
      "grad_norm": 19.03451919555664,
      "learning_rate": 1e-06,
      "loss": 0.3281,
      "num_input_tokens_seen": 491711688,
      "step": 8776
    },
    {
      "epoch": 19.5456570155902,
      "loss": 0.26713547110557556,
      "loss_ce": 7.614593778271228e-05,
      "loss_iou": 0.1220703125,
      "loss_num": 0.0045166015625,
      "loss_xval": 0.267578125,
      "num_input_tokens_seen": 491711688,
      "step": 8776
    },
    {
      "epoch": 19.547884187082406,
      "grad_norm": 26.791309356689453,
      "learning_rate": 1e-06,
      "loss": 0.4318,
      "num_input_tokens_seen": 491765368,
      "step": 8777
    },
    {
      "epoch": 19.547884187082406,
      "loss": 0.45965927839279175,
      "loss_ce": 6.455044058384374e-05,
      "loss_iou": 0.201171875,
      "loss_num": 0.011474609375,
      "loss_xval": 0.458984375,
      "num_input_tokens_seen": 491765368,
      "step": 8777
    },
    {
      "epoch": 19.55011135857461,
      "grad_norm": 39.71892166137695,
      "learning_rate": 1e-06,
      "loss": 0.4227,
      "num_input_tokens_seen": 491822976,
      "step": 8778
    },
    {
      "epoch": 19.55011135857461,
      "loss": 0.439960241317749,
      "loss_ce": 7.986063428688794e-05,
      "loss_iou": 0.1962890625,
      "loss_num": 0.0093994140625,
      "loss_xval": 0.439453125,
      "num_input_tokens_seen": 491822976,
      "step": 8778
    },
    {
      "epoch": 19.552338530066816,
      "grad_norm": 14.03285026550293,
      "learning_rate": 1e-06,
      "loss": 0.3264,
      "num_input_tokens_seen": 491879560,
      "step": 8779
    },
    {
      "epoch": 19.552338530066816,
      "loss": 0.28052568435668945,
      "loss_ce": 6.915038829902187e-05,
      "loss_iou": 0.12451171875,
      "loss_num": 0.006256103515625,
      "loss_xval": 0.28125,
      "num_input_tokens_seen": 491879560,
      "step": 8779
    },
    {
      "epoch": 19.55456570155902,
      "grad_norm": 18.03692626953125,
      "learning_rate": 1e-06,
      "loss": 0.6105,
      "num_input_tokens_seen": 491932316,
      "step": 8780
    },
    {
      "epoch": 19.55456570155902,
      "loss": 0.5739569067955017,
      "loss_ce": 0.00010437377204652876,
      "loss_iou": 0.23046875,
      "loss_num": 0.022705078125,
      "loss_xval": 0.57421875,
      "num_input_tokens_seen": 491932316,
      "step": 8780
    },
    {
      "epoch": 19.556792873051226,
      "grad_norm": 32.511112213134766,
      "learning_rate": 1e-06,
      "loss": 0.4244,
      "num_input_tokens_seen": 491988108,
      "step": 8781
    },
    {
      "epoch": 19.556792873051226,
      "loss": 0.32041823863983154,
      "loss_ce": 7.521534280385822e-05,
      "loss_iou": 0.138671875,
      "loss_num": 0.008544921875,
      "loss_xval": 0.3203125,
      "num_input_tokens_seen": 491988108,
      "step": 8781
    },
    {
      "epoch": 19.55902004454343,
      "grad_norm": 16.41408920288086,
      "learning_rate": 1e-06,
      "loss": 0.3266,
      "num_input_tokens_seen": 492045888,
      "step": 8782
    },
    {
      "epoch": 19.55902004454343,
      "loss": 0.24403327703475952,
      "loss_ce": 7.575724157504737e-05,
      "loss_iou": 0.11083984375,
      "loss_num": 0.004486083984375,
      "loss_xval": 0.244140625,
      "num_input_tokens_seen": 492045888,
      "step": 8782
    },
    {
      "epoch": 19.561247216035635,
      "grad_norm": 19.35165786743164,
      "learning_rate": 1e-06,
      "loss": 0.4494,
      "num_input_tokens_seen": 492102372,
      "step": 8783
    },
    {
      "epoch": 19.561247216035635,
      "loss": 0.42511671781539917,
      "loss_ce": 6.788804603274912e-05,
      "loss_iou": 0.1845703125,
      "loss_num": 0.01129150390625,
      "loss_xval": 0.42578125,
      "num_input_tokens_seen": 492102372,
      "step": 8783
    },
    {
      "epoch": 19.56347438752784,
      "grad_norm": 28.635347366333008,
      "learning_rate": 1e-06,
      "loss": 0.4119,
      "num_input_tokens_seen": 492158548,
      "step": 8784
    },
    {
      "epoch": 19.56347438752784,
      "loss": 0.319175660610199,
      "loss_ce": 8.385031105717644e-05,
      "loss_iou": 0.138671875,
      "loss_num": 0.00823974609375,
      "loss_xval": 0.318359375,
      "num_input_tokens_seen": 492158548,
      "step": 8784
    },
    {
      "epoch": 19.565701559020045,
      "grad_norm": 20.49407386779785,
      "learning_rate": 1e-06,
      "loss": 0.3606,
      "num_input_tokens_seen": 492214608,
      "step": 8785
    },
    {
      "epoch": 19.565701559020045,
      "loss": 0.32773077487945557,
      "loss_ce": 9.405831951880828e-05,
      "loss_iou": 0.1513671875,
      "loss_num": 0.00482177734375,
      "loss_xval": 0.328125,
      "num_input_tokens_seen": 492214608,
      "step": 8785
    },
    {
      "epoch": 19.56792873051225,
      "grad_norm": 19.137483596801758,
      "learning_rate": 1e-06,
      "loss": 0.6018,
      "num_input_tokens_seen": 492271240,
      "step": 8786
    },
    {
      "epoch": 19.56792873051225,
      "loss": 0.5808820128440857,
      "loss_ce": 7.148716395022348e-05,
      "loss_iou": 0.244140625,
      "loss_num": 0.0185546875,
      "loss_xval": 0.58203125,
      "num_input_tokens_seen": 492271240,
      "step": 8786
    },
    {
      "epoch": 19.570155902004455,
      "grad_norm": 13.811071395874023,
      "learning_rate": 1e-06,
      "loss": 0.3109,
      "num_input_tokens_seen": 492323760,
      "step": 8787
    },
    {
      "epoch": 19.570155902004455,
      "loss": 0.32258814573287964,
      "loss_ce": 7.836205622879788e-05,
      "loss_iou": 0.1435546875,
      "loss_num": 0.006988525390625,
      "loss_xval": 0.322265625,
      "num_input_tokens_seen": 492323760,
      "step": 8787
    },
    {
      "epoch": 19.57238307349666,
      "grad_norm": 17.646892547607422,
      "learning_rate": 1e-06,
      "loss": 0.3744,
      "num_input_tokens_seen": 492377292,
      "step": 8788
    },
    {
      "epoch": 19.57238307349666,
      "loss": 0.48769521713256836,
      "loss_ce": 8.538780093658715e-05,
      "loss_iou": 0.203125,
      "loss_num": 0.0164794921875,
      "loss_xval": 0.48828125,
      "num_input_tokens_seen": 492377292,
      "step": 8788
    },
    {
      "epoch": 19.574610244988865,
      "grad_norm": 32.08982467651367,
      "learning_rate": 1e-06,
      "loss": 0.339,
      "num_input_tokens_seen": 492432440,
      "step": 8789
    },
    {
      "epoch": 19.574610244988865,
      "loss": 0.32432878017425537,
      "loss_ce": 6.424468301702291e-05,
      "loss_iou": 0.1484375,
      "loss_num": 0.0054931640625,
      "loss_xval": 0.32421875,
      "num_input_tokens_seen": 492432440,
      "step": 8789
    },
    {
      "epoch": 19.57683741648107,
      "grad_norm": 10.528325080871582,
      "learning_rate": 1e-06,
      "loss": 0.2973,
      "num_input_tokens_seen": 492490824,
      "step": 8790
    },
    {
      "epoch": 19.57683741648107,
      "loss": 0.3674495816230774,
      "loss_ce": 7.898532930994406e-05,
      "loss_iou": 0.1513671875,
      "loss_num": 0.0130615234375,
      "loss_xval": 0.3671875,
      "num_input_tokens_seen": 492490824,
      "step": 8790
    },
    {
      "epoch": 19.579064587973274,
      "grad_norm": 18.546995162963867,
      "learning_rate": 1e-06,
      "loss": 0.3658,
      "num_input_tokens_seen": 492549284,
      "step": 8791
    },
    {
      "epoch": 19.579064587973274,
      "loss": 0.2795357406139374,
      "loss_ce": 5.5770447943359613e-05,
      "loss_iou": 0.11669921875,
      "loss_num": 0.0091552734375,
      "loss_xval": 0.279296875,
      "num_input_tokens_seen": 492549284,
      "step": 8791
    },
    {
      "epoch": 19.58129175946548,
      "grad_norm": 19.887773513793945,
      "learning_rate": 1e-06,
      "loss": 0.376,
      "num_input_tokens_seen": 492605324,
      "step": 8792
    },
    {
      "epoch": 19.58129175946548,
      "loss": 0.35456758737564087,
      "loss_ce": 7.540988008258864e-05,
      "loss_iou": 0.1572265625,
      "loss_num": 0.0081787109375,
      "loss_xval": 0.35546875,
      "num_input_tokens_seen": 492605324,
      "step": 8792
    },
    {
      "epoch": 19.583518930957684,
      "grad_norm": 19.090816497802734,
      "learning_rate": 1e-06,
      "loss": 0.3357,
      "num_input_tokens_seen": 492661580,
      "step": 8793
    },
    {
      "epoch": 19.583518930957684,
      "loss": 0.36834782361984253,
      "loss_ce": 6.167573155835271e-05,
      "loss_iou": 0.15234375,
      "loss_num": 0.0125732421875,
      "loss_xval": 0.369140625,
      "num_input_tokens_seen": 492661580,
      "step": 8793
    },
    {
      "epoch": 19.58574610244989,
      "grad_norm": 18.095928192138672,
      "learning_rate": 1e-06,
      "loss": 0.3908,
      "num_input_tokens_seen": 492719140,
      "step": 8794
    },
    {
      "epoch": 19.58574610244989,
      "loss": 0.2904747724533081,
      "loss_ce": 6.948962982278317e-05,
      "loss_iou": 0.1259765625,
      "loss_num": 0.007720947265625,
      "loss_xval": 0.291015625,
      "num_input_tokens_seen": 492719140,
      "step": 8794
    },
    {
      "epoch": 19.587973273942094,
      "grad_norm": 15.511141777038574,
      "learning_rate": 1e-06,
      "loss": 0.4641,
      "num_input_tokens_seen": 492775612,
      "step": 8795
    },
    {
      "epoch": 19.587973273942094,
      "loss": 0.42014381289482117,
      "loss_ce": 9.987192606786266e-05,
      "loss_iou": 0.1904296875,
      "loss_num": 0.00775146484375,
      "loss_xval": 0.419921875,
      "num_input_tokens_seen": 492775612,
      "step": 8795
    },
    {
      "epoch": 19.5902004454343,
      "grad_norm": 19.146825790405273,
      "learning_rate": 1e-06,
      "loss": 0.4842,
      "num_input_tokens_seen": 492833836,
      "step": 8796
    },
    {
      "epoch": 19.5902004454343,
      "loss": 0.5711435079574585,
      "loss_ce": 9.86100931186229e-05,
      "loss_iou": 0.2412109375,
      "loss_num": 0.0177001953125,
      "loss_xval": 0.5703125,
      "num_input_tokens_seen": 492833836,
      "step": 8796
    },
    {
      "epoch": 19.592427616926503,
      "grad_norm": 14.747178077697754,
      "learning_rate": 1e-06,
      "loss": 0.444,
      "num_input_tokens_seen": 492887604,
      "step": 8797
    },
    {
      "epoch": 19.592427616926503,
      "loss": 0.29975390434265137,
      "loss_ce": 7.127322896849364e-05,
      "loss_iou": 0.140625,
      "loss_num": 0.00384521484375,
      "loss_xval": 0.298828125,
      "num_input_tokens_seen": 492887604,
      "step": 8797
    },
    {
      "epoch": 19.59465478841871,
      "grad_norm": 18.54738426208496,
      "learning_rate": 1e-06,
      "loss": 0.4126,
      "num_input_tokens_seen": 492944056,
      "step": 8798
    },
    {
      "epoch": 19.59465478841871,
      "loss": 0.259499728679657,
      "loss_ce": 6.978299643378705e-05,
      "loss_iou": 0.1171875,
      "loss_num": 0.00494384765625,
      "loss_xval": 0.259765625,
      "num_input_tokens_seen": 492944056,
      "step": 8798
    },
    {
      "epoch": 19.596881959910913,
      "grad_norm": 22.788713455200195,
      "learning_rate": 1e-06,
      "loss": 0.4753,
      "num_input_tokens_seen": 493001668,
      "step": 8799
    },
    {
      "epoch": 19.596881959910913,
      "loss": 0.5575045347213745,
      "loss_ce": 7.041562639642507e-05,
      "loss_iou": 0.2333984375,
      "loss_num": 0.0179443359375,
      "loss_xval": 0.55859375,
      "num_input_tokens_seen": 493001668,
      "step": 8799
    },
    {
      "epoch": 19.599109131403118,
      "grad_norm": 16.851640701293945,
      "learning_rate": 1e-06,
      "loss": 0.2476,
      "num_input_tokens_seen": 493060028,
      "step": 8800
    },
    {
      "epoch": 19.599109131403118,
      "loss": 0.2806518077850342,
      "loss_ce": 7.319905853364617e-05,
      "loss_iou": 0.125,
      "loss_num": 0.00628662109375,
      "loss_xval": 0.28125,
      "num_input_tokens_seen": 493060028,
      "step": 8800
    },
    {
      "epoch": 19.601336302895323,
      "grad_norm": 17.70904541015625,
      "learning_rate": 1e-06,
      "loss": 0.3181,
      "num_input_tokens_seen": 493117108,
      "step": 8801
    },
    {
      "epoch": 19.601336302895323,
      "loss": 0.2125520408153534,
      "loss_ce": 5.813151801703498e-05,
      "loss_iou": 0.095703125,
      "loss_num": 0.004150390625,
      "loss_xval": 0.212890625,
      "num_input_tokens_seen": 493117108,
      "step": 8801
    },
    {
      "epoch": 19.603563474387528,
      "grad_norm": 15.63475513458252,
      "learning_rate": 1e-06,
      "loss": 0.2397,
      "num_input_tokens_seen": 493170576,
      "step": 8802
    },
    {
      "epoch": 19.603563474387528,
      "loss": 0.2356223165988922,
      "loss_ce": 5.7125496823573485e-05,
      "loss_iou": 0.1044921875,
      "loss_num": 0.005218505859375,
      "loss_xval": 0.2353515625,
      "num_input_tokens_seen": 493170576,
      "step": 8802
    },
    {
      "epoch": 19.605790645879733,
      "grad_norm": 12.706445693969727,
      "learning_rate": 1e-06,
      "loss": 0.5809,
      "num_input_tokens_seen": 493227004,
      "step": 8803
    },
    {
      "epoch": 19.605790645879733,
      "loss": 0.6226192712783813,
      "loss_ce": 6.07035071880091e-05,
      "loss_iou": 0.2451171875,
      "loss_num": 0.0264892578125,
      "loss_xval": 0.62109375,
      "num_input_tokens_seen": 493227004,
      "step": 8803
    },
    {
      "epoch": 19.608017817371937,
      "grad_norm": 12.90841293334961,
      "learning_rate": 1e-06,
      "loss": 0.5149,
      "num_input_tokens_seen": 493284640,
      "step": 8804
    },
    {
      "epoch": 19.608017817371937,
      "loss": 0.36311158537864685,
      "loss_ce": 7.448208634741604e-05,
      "loss_iou": 0.1630859375,
      "loss_num": 0.00738525390625,
      "loss_xval": 0.36328125,
      "num_input_tokens_seen": 493284640,
      "step": 8804
    },
    {
      "epoch": 19.610244988864142,
      "grad_norm": 20.88025665283203,
      "learning_rate": 1e-06,
      "loss": 0.5289,
      "num_input_tokens_seen": 493343304,
      "step": 8805
    },
    {
      "epoch": 19.610244988864142,
      "loss": 0.4120563268661499,
      "loss_ce": 6.90398010192439e-05,
      "loss_iou": 0.1728515625,
      "loss_num": 0.0133056640625,
      "loss_xval": 0.412109375,
      "num_input_tokens_seen": 493343304,
      "step": 8805
    },
    {
      "epoch": 19.612472160356347,
      "grad_norm": 18.58097267150879,
      "learning_rate": 1e-06,
      "loss": 0.3529,
      "num_input_tokens_seen": 493401548,
      "step": 8806
    },
    {
      "epoch": 19.612472160356347,
      "loss": 0.3655468821525574,
      "loss_ce": 6.83803009451367e-05,
      "loss_iou": 0.1455078125,
      "loss_num": 0.01513671875,
      "loss_xval": 0.365234375,
      "num_input_tokens_seen": 493401548,
      "step": 8806
    },
    {
      "epoch": 19.614699331848552,
      "grad_norm": 23.805185317993164,
      "learning_rate": 1e-06,
      "loss": 0.3137,
      "num_input_tokens_seen": 493456364,
      "step": 8807
    },
    {
      "epoch": 19.614699331848552,
      "loss": 0.414080947637558,
      "loss_ce": 7.948270649649203e-05,
      "loss_iou": 0.1796875,
      "loss_num": 0.0108642578125,
      "loss_xval": 0.4140625,
      "num_input_tokens_seen": 493456364,
      "step": 8807
    },
    {
      "epoch": 19.616926503340757,
      "grad_norm": 14.647207260131836,
      "learning_rate": 1e-06,
      "loss": 0.3448,
      "num_input_tokens_seen": 493514088,
      "step": 8808
    },
    {
      "epoch": 19.616926503340757,
      "loss": 0.3253961503505707,
      "loss_ce": 7.876554445829242e-05,
      "loss_iou": 0.1474609375,
      "loss_num": 0.005950927734375,
      "loss_xval": 0.326171875,
      "num_input_tokens_seen": 493514088,
      "step": 8808
    },
    {
      "epoch": 19.619153674832962,
      "grad_norm": 16.693212509155273,
      "learning_rate": 1e-06,
      "loss": 0.3789,
      "num_input_tokens_seen": 493571508,
      "step": 8809
    },
    {
      "epoch": 19.619153674832962,
      "loss": 0.20014190673828125,
      "loss_ce": 6.866748299216852e-05,
      "loss_iou": 0.08251953125,
      "loss_num": 0.00701904296875,
      "loss_xval": 0.2001953125,
      "num_input_tokens_seen": 493571508,
      "step": 8809
    },
    {
      "epoch": 19.621380846325167,
      "grad_norm": 24.3143367767334,
      "learning_rate": 1e-06,
      "loss": 0.3707,
      "num_input_tokens_seen": 493625268,
      "step": 8810
    },
    {
      "epoch": 19.621380846325167,
      "loss": 0.4873785972595215,
      "loss_ce": 7.392482802970335e-05,
      "loss_iou": 0.208984375,
      "loss_num": 0.01373291015625,
      "loss_xval": 0.48828125,
      "num_input_tokens_seen": 493625268,
      "step": 8810
    },
    {
      "epoch": 19.62360801781737,
      "grad_norm": 23.970718383789062,
      "learning_rate": 1e-06,
      "loss": 0.5419,
      "num_input_tokens_seen": 493682744,
      "step": 8811
    },
    {
      "epoch": 19.62360801781737,
      "loss": 0.5781893730163574,
      "loss_ce": 6.439335993491113e-05,
      "loss_iou": 0.24609375,
      "loss_num": 0.0172119140625,
      "loss_xval": 0.578125,
      "num_input_tokens_seen": 493682744,
      "step": 8811
    },
    {
      "epoch": 19.625835189309576,
      "grad_norm": 16.08030128479004,
      "learning_rate": 1e-06,
      "loss": 0.3674,
      "num_input_tokens_seen": 493738420,
      "step": 8812
    },
    {
      "epoch": 19.625835189309576,
      "loss": 0.36957836151123047,
      "loss_ce": 7.152724720072001e-05,
      "loss_iou": 0.1494140625,
      "loss_num": 0.013916015625,
      "loss_xval": 0.369140625,
      "num_input_tokens_seen": 493738420,
      "step": 8812
    },
    {
      "epoch": 19.62806236080178,
      "grad_norm": 16.537208557128906,
      "learning_rate": 1e-06,
      "loss": 0.4184,
      "num_input_tokens_seen": 493792444,
      "step": 8813
    },
    {
      "epoch": 19.62806236080178,
      "loss": 0.27265116572380066,
      "loss_ce": 6.81565361446701e-05,
      "loss_iou": 0.1142578125,
      "loss_num": 0.0087890625,
      "loss_xval": 0.2734375,
      "num_input_tokens_seen": 493792444,
      "step": 8813
    },
    {
      "epoch": 19.630289532293986,
      "grad_norm": 16.97813606262207,
      "learning_rate": 1e-06,
      "loss": 0.3923,
      "num_input_tokens_seen": 493851536,
      "step": 8814
    },
    {
      "epoch": 19.630289532293986,
      "loss": 0.22918465733528137,
      "loss_ce": 5.868993321200833e-05,
      "loss_iou": 0.10205078125,
      "loss_num": 0.004974365234375,
      "loss_xval": 0.2294921875,
      "num_input_tokens_seen": 493851536,
      "step": 8814
    },
    {
      "epoch": 19.63251670378619,
      "grad_norm": 18.847007751464844,
      "learning_rate": 1e-06,
      "loss": 0.4566,
      "num_input_tokens_seen": 493909260,
      "step": 8815
    },
    {
      "epoch": 19.63251670378619,
      "loss": 0.3699719309806824,
      "loss_ce": 9.890169894788414e-05,
      "loss_iou": 0.1640625,
      "loss_num": 0.0084228515625,
      "loss_xval": 0.369140625,
      "num_input_tokens_seen": 493909260,
      "step": 8815
    },
    {
      "epoch": 19.634743875278396,
      "grad_norm": 12.239165306091309,
      "learning_rate": 1e-06,
      "loss": 0.2959,
      "num_input_tokens_seen": 493964284,
      "step": 8816
    },
    {
      "epoch": 19.634743875278396,
      "loss": 0.36634206771850586,
      "loss_ce": 7.009244291111827e-05,
      "loss_iou": 0.15625,
      "loss_num": 0.01080322265625,
      "loss_xval": 0.3671875,
      "num_input_tokens_seen": 493964284,
      "step": 8816
    },
    {
      "epoch": 19.6369710467706,
      "grad_norm": 20.558364868164062,
      "learning_rate": 1e-06,
      "loss": 0.4118,
      "num_input_tokens_seen": 494020864,
      "step": 8817
    },
    {
      "epoch": 19.6369710467706,
      "loss": 0.4873617887496948,
      "loss_ce": 5.70821066503413e-05,
      "loss_iou": 0.2138671875,
      "loss_num": 0.0120849609375,
      "loss_xval": 0.48828125,
      "num_input_tokens_seen": 494020864,
      "step": 8817
    },
    {
      "epoch": 19.639198218262806,
      "grad_norm": 20.281597137451172,
      "learning_rate": 1e-06,
      "loss": 0.3563,
      "num_input_tokens_seen": 494076352,
      "step": 8818
    },
    {
      "epoch": 19.639198218262806,
      "loss": 0.2713017463684082,
      "loss_ce": 6.150568515295163e-05,
      "loss_iou": 0.125,
      "loss_num": 0.00408935546875,
      "loss_xval": 0.271484375,
      "num_input_tokens_seen": 494076352,
      "step": 8818
    },
    {
      "epoch": 19.64142538975501,
      "grad_norm": 11.944673538208008,
      "learning_rate": 1e-06,
      "loss": 0.2912,
      "num_input_tokens_seen": 494132992,
      "step": 8819
    },
    {
      "epoch": 19.64142538975501,
      "loss": 0.2255510687828064,
      "loss_ce": 8.72044765856117e-05,
      "loss_iou": 0.09912109375,
      "loss_num": 0.0054931640625,
      "loss_xval": 0.2255859375,
      "num_input_tokens_seen": 494132992,
      "step": 8819
    },
    {
      "epoch": 19.643652561247215,
      "grad_norm": 17.89927864074707,
      "learning_rate": 1e-06,
      "loss": 0.4362,
      "num_input_tokens_seen": 494187992,
      "step": 8820
    },
    {
      "epoch": 19.643652561247215,
      "loss": 0.3235388398170471,
      "loss_ce": 6.778095848858356e-05,
      "loss_iou": 0.1396484375,
      "loss_num": 0.0086669921875,
      "loss_xval": 0.32421875,
      "num_input_tokens_seen": 494187992,
      "step": 8820
    },
    {
      "epoch": 19.64587973273942,
      "grad_norm": 22.77928924560547,
      "learning_rate": 1e-06,
      "loss": 0.3405,
      "num_input_tokens_seen": 494243788,
      "step": 8821
    },
    {
      "epoch": 19.64587973273942,
      "loss": 0.24566447734832764,
      "loss_ce": 5.902632256038487e-05,
      "loss_iou": 0.10888671875,
      "loss_num": 0.005645751953125,
      "loss_xval": 0.24609375,
      "num_input_tokens_seen": 494243788,
      "step": 8821
    },
    {
      "epoch": 19.648106904231625,
      "grad_norm": 17.578380584716797,
      "learning_rate": 1e-06,
      "loss": 0.4421,
      "num_input_tokens_seen": 494300640,
      "step": 8822
    },
    {
      "epoch": 19.648106904231625,
      "loss": 0.6269207000732422,
      "loss_ce": 8.963636355474591e-05,
      "loss_iou": 0.2734375,
      "loss_num": 0.0162353515625,
      "loss_xval": 0.625,
      "num_input_tokens_seen": 494300640,
      "step": 8822
    },
    {
      "epoch": 19.65033407572383,
      "grad_norm": 18.126558303833008,
      "learning_rate": 1e-06,
      "loss": 0.3305,
      "num_input_tokens_seen": 494357432,
      "step": 8823
    },
    {
      "epoch": 19.65033407572383,
      "loss": 0.2599582076072693,
      "loss_ce": 7.050028216326609e-05,
      "loss_iou": 0.11474609375,
      "loss_num": 0.006134033203125,
      "loss_xval": 0.259765625,
      "num_input_tokens_seen": 494357432,
      "step": 8823
    },
    {
      "epoch": 19.652561247216035,
      "grad_norm": 14.803417205810547,
      "learning_rate": 1e-06,
      "loss": 0.6651,
      "num_input_tokens_seen": 494412312,
      "step": 8824
    },
    {
      "epoch": 19.652561247216035,
      "loss": 0.7808718681335449,
      "loss_ce": 0.00011010345770046115,
      "loss_iou": 0.3125,
      "loss_num": 0.031494140625,
      "loss_xval": 0.78125,
      "num_input_tokens_seen": 494412312,
      "step": 8824
    },
    {
      "epoch": 19.65478841870824,
      "grad_norm": 28.97988510131836,
      "learning_rate": 1e-06,
      "loss": 0.5272,
      "num_input_tokens_seen": 494465128,
      "step": 8825
    },
    {
      "epoch": 19.65478841870824,
      "loss": 0.629551112651825,
      "loss_ce": 0.0005227710935287178,
      "loss_iou": 0.287109375,
      "loss_num": 0.0113525390625,
      "loss_xval": 0.62890625,
      "num_input_tokens_seen": 494465128,
      "step": 8825
    },
    {
      "epoch": 19.657015590200444,
      "grad_norm": 14.80961799621582,
      "learning_rate": 1e-06,
      "loss": 0.3384,
      "num_input_tokens_seen": 494520212,
      "step": 8826
    },
    {
      "epoch": 19.657015590200444,
      "loss": 0.37103530764579773,
      "loss_ce": 6.362018757499754e-05,
      "loss_iou": 0.1640625,
      "loss_num": 0.0084228515625,
      "loss_xval": 0.37109375,
      "num_input_tokens_seen": 494520212,
      "step": 8826
    },
    {
      "epoch": 19.65924276169265,
      "grad_norm": 11.278867721557617,
      "learning_rate": 1e-06,
      "loss": 0.4748,
      "num_input_tokens_seen": 494576024,
      "step": 8827
    },
    {
      "epoch": 19.65924276169265,
      "loss": 0.3697161376476288,
      "loss_ce": 8.720482583157718e-05,
      "loss_iou": 0.16796875,
      "loss_num": 0.0067138671875,
      "loss_xval": 0.369140625,
      "num_input_tokens_seen": 494576024,
      "step": 8827
    },
    {
      "epoch": 19.661469933184854,
      "grad_norm": 21.07610321044922,
      "learning_rate": 1e-06,
      "loss": 0.4067,
      "num_input_tokens_seen": 494631812,
      "step": 8828
    },
    {
      "epoch": 19.661469933184854,
      "loss": 0.46012747287750244,
      "loss_ce": 7.498678314732388e-05,
      "loss_iou": 0.1796875,
      "loss_num": 0.02001953125,
      "loss_xval": 0.4609375,
      "num_input_tokens_seen": 494631812,
      "step": 8828
    },
    {
      "epoch": 19.66369710467706,
      "grad_norm": 29.130268096923828,
      "learning_rate": 1e-06,
      "loss": 0.5162,
      "num_input_tokens_seen": 494688008,
      "step": 8829
    },
    {
      "epoch": 19.66369710467706,
      "loss": 0.34423893690109253,
      "loss_ce": 6.168825348140672e-05,
      "loss_iou": 0.146484375,
      "loss_num": 0.0103759765625,
      "loss_xval": 0.34375,
      "num_input_tokens_seen": 494688008,
      "step": 8829
    },
    {
      "epoch": 19.665924276169264,
      "grad_norm": 15.006963729858398,
      "learning_rate": 1e-06,
      "loss": 0.2145,
      "num_input_tokens_seen": 494743180,
      "step": 8830
    },
    {
      "epoch": 19.665924276169264,
      "loss": 0.2853913903236389,
      "loss_ce": 0.0004792909894604236,
      "loss_iou": 0.1220703125,
      "loss_num": 0.008056640625,
      "loss_xval": 0.28515625,
      "num_input_tokens_seen": 494743180,
      "step": 8830
    },
    {
      "epoch": 19.66815144766147,
      "grad_norm": 18.46776580810547,
      "learning_rate": 1e-06,
      "loss": 0.3155,
      "num_input_tokens_seen": 494800836,
      "step": 8831
    },
    {
      "epoch": 19.66815144766147,
      "loss": 0.3410201072692871,
      "loss_ce": 7.773660036036745e-05,
      "loss_iou": 0.15234375,
      "loss_num": 0.0072021484375,
      "loss_xval": 0.341796875,
      "num_input_tokens_seen": 494800836,
      "step": 8831
    },
    {
      "epoch": 19.670378619153674,
      "grad_norm": 23.41823387145996,
      "learning_rate": 1e-06,
      "loss": 0.4599,
      "num_input_tokens_seen": 494858464,
      "step": 8832
    },
    {
      "epoch": 19.670378619153674,
      "loss": 0.5210896730422974,
      "loss_ce": 9.359928662888706e-05,
      "loss_iou": 0.224609375,
      "loss_num": 0.01422119140625,
      "loss_xval": 0.51953125,
      "num_input_tokens_seen": 494858464,
      "step": 8832
    },
    {
      "epoch": 19.67260579064588,
      "grad_norm": 17.398420333862305,
      "learning_rate": 1e-06,
      "loss": 0.3252,
      "num_input_tokens_seen": 494912524,
      "step": 8833
    },
    {
      "epoch": 19.67260579064588,
      "loss": 0.5025025606155396,
      "loss_ce": 6.115515134297311e-05,
      "loss_iou": 0.19921875,
      "loss_num": 0.0208740234375,
      "loss_xval": 0.50390625,
      "num_input_tokens_seen": 494912524,
      "step": 8833
    },
    {
      "epoch": 19.674832962138083,
      "grad_norm": 12.045059204101562,
      "learning_rate": 1e-06,
      "loss": 0.4081,
      "num_input_tokens_seen": 494968308,
      "step": 8834
    },
    {
      "epoch": 19.674832962138083,
      "loss": 0.38177552819252014,
      "loss_ce": 6.166419188957661e-05,
      "loss_iou": 0.173828125,
      "loss_num": 0.0067138671875,
      "loss_xval": 0.380859375,
      "num_input_tokens_seen": 494968308,
      "step": 8834
    },
    {
      "epoch": 19.677060133630288,
      "grad_norm": 24.122774124145508,
      "learning_rate": 1e-06,
      "loss": 0.4293,
      "num_input_tokens_seen": 495022460,
      "step": 8835
    },
    {
      "epoch": 19.677060133630288,
      "loss": 0.5443927049636841,
      "loss_ce": 8.117854304146022e-05,
      "loss_iou": 0.220703125,
      "loss_num": 0.0206298828125,
      "loss_xval": 0.54296875,
      "num_input_tokens_seen": 495022460,
      "step": 8835
    },
    {
      "epoch": 19.679287305122493,
      "grad_norm": 30.17648696899414,
      "learning_rate": 1e-06,
      "loss": 0.3063,
      "num_input_tokens_seen": 495079416,
      "step": 8836
    },
    {
      "epoch": 19.679287305122493,
      "loss": 0.363957941532135,
      "loss_ce": 6.632816075580195e-05,
      "loss_iou": 0.1630859375,
      "loss_num": 0.007720947265625,
      "loss_xval": 0.36328125,
      "num_input_tokens_seen": 495079416,
      "step": 8836
    },
    {
      "epoch": 19.681514476614698,
      "grad_norm": 28.630796432495117,
      "learning_rate": 1e-06,
      "loss": 0.4824,
      "num_input_tokens_seen": 495136332,
      "step": 8837
    },
    {
      "epoch": 19.681514476614698,
      "loss": 0.3295344114303589,
      "loss_ce": 6.66461419314146e-05,
      "loss_iou": 0.1435546875,
      "loss_num": 0.0084228515625,
      "loss_xval": 0.330078125,
      "num_input_tokens_seen": 495136332,
      "step": 8837
    },
    {
      "epoch": 19.683741648106903,
      "grad_norm": 16.524477005004883,
      "learning_rate": 1e-06,
      "loss": 0.3898,
      "num_input_tokens_seen": 495192408,
      "step": 8838
    },
    {
      "epoch": 19.683741648106903,
      "loss": 0.47682133316993713,
      "loss_ce": 7.57328380132094e-05,
      "loss_iou": 0.1943359375,
      "loss_num": 0.017578125,
      "loss_xval": 0.4765625,
      "num_input_tokens_seen": 495192408,
      "step": 8838
    },
    {
      "epoch": 19.685968819599108,
      "grad_norm": 15.386173248291016,
      "learning_rate": 1e-06,
      "loss": 0.487,
      "num_input_tokens_seen": 495248052,
      "step": 8839
    },
    {
      "epoch": 19.685968819599108,
      "loss": 0.6375049948692322,
      "loss_ce": 0.0001758981088642031,
      "loss_iou": 0.251953125,
      "loss_num": 0.027099609375,
      "loss_xval": 0.63671875,
      "num_input_tokens_seen": 495248052,
      "step": 8839
    },
    {
      "epoch": 19.688195991091312,
      "grad_norm": 13.274645805358887,
      "learning_rate": 1e-06,
      "loss": 0.4217,
      "num_input_tokens_seen": 495304220,
      "step": 8840
    },
    {
      "epoch": 19.688195991091312,
      "loss": 0.40014198422431946,
      "loss_ce": 5.652975960401818e-05,
      "loss_iou": 0.1708984375,
      "loss_num": 0.01177978515625,
      "loss_xval": 0.400390625,
      "num_input_tokens_seen": 495304220,
      "step": 8840
    },
    {
      "epoch": 19.690423162583517,
      "grad_norm": 19.099824905395508,
      "learning_rate": 1e-06,
      "loss": 0.4126,
      "num_input_tokens_seen": 495360180,
      "step": 8841
    },
    {
      "epoch": 19.690423162583517,
      "loss": 0.408029705286026,
      "loss_ce": 7.075037865433842e-05,
      "loss_iou": 0.1865234375,
      "loss_num": 0.007171630859375,
      "loss_xval": 0.408203125,
      "num_input_tokens_seen": 495360180,
      "step": 8841
    },
    {
      "epoch": 19.692650334075722,
      "grad_norm": 14.30420207977295,
      "learning_rate": 1e-06,
      "loss": 0.3627,
      "num_input_tokens_seen": 495417216,
      "step": 8842
    },
    {
      "epoch": 19.692650334075722,
      "loss": 0.5124090909957886,
      "loss_ce": 7.99886038294062e-05,
      "loss_iou": 0.2197265625,
      "loss_num": 0.01458740234375,
      "loss_xval": 0.51171875,
      "num_input_tokens_seen": 495417216,
      "step": 8842
    },
    {
      "epoch": 19.694877505567927,
      "grad_norm": 16.062725067138672,
      "learning_rate": 1e-06,
      "loss": 0.4026,
      "num_input_tokens_seen": 495473692,
      "step": 8843
    },
    {
      "epoch": 19.694877505567927,
      "loss": 0.44783011078834534,
      "loss_ce": 0.0001982695539481938,
      "loss_iou": 0.1728515625,
      "loss_num": 0.0205078125,
      "loss_xval": 0.447265625,
      "num_input_tokens_seen": 495473692,
      "step": 8843
    },
    {
      "epoch": 19.697104677060132,
      "grad_norm": 13.419224739074707,
      "learning_rate": 1e-06,
      "loss": 0.3158,
      "num_input_tokens_seen": 495529692,
      "step": 8844
    },
    {
      "epoch": 19.697104677060132,
      "loss": 0.41748589277267456,
      "loss_ce": 6.646426481893286e-05,
      "loss_iou": 0.1552734375,
      "loss_num": 0.021484375,
      "loss_xval": 0.41796875,
      "num_input_tokens_seen": 495529692,
      "step": 8844
    },
    {
      "epoch": 19.69933184855234,
      "grad_norm": 20.578535079956055,
      "learning_rate": 1e-06,
      "loss": 0.4155,
      "num_input_tokens_seen": 495583708,
      "step": 8845
    },
    {
      "epoch": 19.69933184855234,
      "loss": 0.42976486682891846,
      "loss_ce": 7.74070795159787e-05,
      "loss_iou": 0.19140625,
      "loss_num": 0.00946044921875,
      "loss_xval": 0.4296875,
      "num_input_tokens_seen": 495583708,
      "step": 8845
    },
    {
      "epoch": 19.70155902004454,
      "grad_norm": 16.558269500732422,
      "learning_rate": 1e-06,
      "loss": 0.4006,
      "num_input_tokens_seen": 495636772,
      "step": 8846
    },
    {
      "epoch": 19.70155902004454,
      "loss": 0.49481481313705444,
      "loss_ce": 6.387023313436657e-05,
      "loss_iou": 0.185546875,
      "loss_num": 0.0247802734375,
      "loss_xval": 0.494140625,
      "num_input_tokens_seen": 495636772,
      "step": 8846
    },
    {
      "epoch": 19.70378619153675,
      "grad_norm": 18.609947204589844,
      "learning_rate": 1e-06,
      "loss": 0.3799,
      "num_input_tokens_seen": 495690168,
      "step": 8847
    },
    {
      "epoch": 19.70378619153675,
      "loss": 0.3060356378555298,
      "loss_ce": 6.63895916659385e-05,
      "loss_iou": 0.1357421875,
      "loss_num": 0.006927490234375,
      "loss_xval": 0.306640625,
      "num_input_tokens_seen": 495690168,
      "step": 8847
    },
    {
      "epoch": 19.706013363028955,
      "grad_norm": 19.02481460571289,
      "learning_rate": 1e-06,
      "loss": 0.4229,
      "num_input_tokens_seen": 495748480,
      "step": 8848
    },
    {
      "epoch": 19.706013363028955,
      "loss": 0.520011305809021,
      "loss_ce": 0.00011385796096874401,
      "loss_iou": 0.220703125,
      "loss_num": 0.01556396484375,
      "loss_xval": 0.51953125,
      "num_input_tokens_seen": 495748480,
      "step": 8848
    },
    {
      "epoch": 19.70824053452116,
      "grad_norm": 13.990358352661133,
      "learning_rate": 1e-06,
      "loss": 0.3866,
      "num_input_tokens_seen": 495805712,
      "step": 8849
    },
    {
      "epoch": 19.70824053452116,
      "loss": 0.49600470066070557,
      "loss_ce": 9.406625758856535e-05,
      "loss_iou": 0.203125,
      "loss_num": 0.017822265625,
      "loss_xval": 0.49609375,
      "num_input_tokens_seen": 495805712,
      "step": 8849
    },
    {
      "epoch": 19.710467706013365,
      "grad_norm": 15.833937644958496,
      "learning_rate": 1e-06,
      "loss": 0.4138,
      "num_input_tokens_seen": 495862696,
      "step": 8850
    },
    {
      "epoch": 19.710467706013365,
      "loss": 0.44215184450149536,
      "loss_ce": 7.422211638186127e-05,
      "loss_iou": 0.1982421875,
      "loss_num": 0.00921630859375,
      "loss_xval": 0.44140625,
      "num_input_tokens_seen": 495862696,
      "step": 8850
    },
    {
      "epoch": 19.71269487750557,
      "grad_norm": 24.503477096557617,
      "learning_rate": 1e-06,
      "loss": 0.412,
      "num_input_tokens_seen": 495913924,
      "step": 8851
    },
    {
      "epoch": 19.71269487750557,
      "loss": 0.329414427280426,
      "loss_ce": 6.872846279293299e-05,
      "loss_iou": 0.146484375,
      "loss_num": 0.007354736328125,
      "loss_xval": 0.330078125,
      "num_input_tokens_seen": 495913924,
      "step": 8851
    },
    {
      "epoch": 19.714922048997774,
      "grad_norm": 20.467147827148438,
      "learning_rate": 1e-06,
      "loss": 0.4762,
      "num_input_tokens_seen": 495969596,
      "step": 8852
    },
    {
      "epoch": 19.714922048997774,
      "loss": 0.4412245452404022,
      "loss_ce": 6.245089753065258e-05,
      "loss_iou": 0.193359375,
      "loss_num": 0.01068115234375,
      "loss_xval": 0.44140625,
      "num_input_tokens_seen": 495969596,
      "step": 8852
    },
    {
      "epoch": 19.71714922048998,
      "grad_norm": 16.596433639526367,
      "learning_rate": 1e-06,
      "loss": 0.3338,
      "num_input_tokens_seen": 496022692,
      "step": 8853
    },
    {
      "epoch": 19.71714922048998,
      "loss": 0.3814437985420227,
      "loss_ce": 6.563542410731316e-05,
      "loss_iou": 0.1708984375,
      "loss_num": 0.007781982421875,
      "loss_xval": 0.380859375,
      "num_input_tokens_seen": 496022692,
      "step": 8853
    },
    {
      "epoch": 19.719376391982184,
      "grad_norm": 14.351813316345215,
      "learning_rate": 1e-06,
      "loss": 0.2733,
      "num_input_tokens_seen": 496077624,
      "step": 8854
    },
    {
      "epoch": 19.719376391982184,
      "loss": 0.30036044120788574,
      "loss_ce": 6.749367457814515e-05,
      "loss_iou": 0.130859375,
      "loss_num": 0.00775146484375,
      "loss_xval": 0.30078125,
      "num_input_tokens_seen": 496077624,
      "step": 8854
    },
    {
      "epoch": 19.72160356347439,
      "grad_norm": 21.87047004699707,
      "learning_rate": 1e-06,
      "loss": 0.5035,
      "num_input_tokens_seen": 496133564,
      "step": 8855
    },
    {
      "epoch": 19.72160356347439,
      "loss": 0.36122041940689087,
      "loss_ce": 7.540646765846759e-05,
      "loss_iou": 0.16015625,
      "loss_num": 0.0081787109375,
      "loss_xval": 0.361328125,
      "num_input_tokens_seen": 496133564,
      "step": 8855
    },
    {
      "epoch": 19.723830734966594,
      "grad_norm": 18.896873474121094,
      "learning_rate": 1e-06,
      "loss": 0.4041,
      "num_input_tokens_seen": 496190924,
      "step": 8856
    },
    {
      "epoch": 19.723830734966594,
      "loss": 0.38140565156936646,
      "loss_ce": 5.800734652439132e-05,
      "loss_iou": 0.1572265625,
      "loss_num": 0.0135498046875,
      "loss_xval": 0.380859375,
      "num_input_tokens_seen": 496190924,
      "step": 8856
    },
    {
      "epoch": 19.7260579064588,
      "grad_norm": 20.9360408782959,
      "learning_rate": 1e-06,
      "loss": 0.3515,
      "num_input_tokens_seen": 496247316,
      "step": 8857
    },
    {
      "epoch": 19.7260579064588,
      "loss": 0.49714395403862,
      "loss_ce": 7.367074431385845e-05,
      "loss_iou": 0.224609375,
      "loss_num": 0.009521484375,
      "loss_xval": 0.49609375,
      "num_input_tokens_seen": 496247316,
      "step": 8857
    },
    {
      "epoch": 19.728285077951004,
      "grad_norm": 22.762691497802734,
      "learning_rate": 1e-06,
      "loss": 0.3882,
      "num_input_tokens_seen": 496305148,
      "step": 8858
    },
    {
      "epoch": 19.728285077951004,
      "loss": 0.37469351291656494,
      "loss_ce": 5.972578583168797e-05,
      "loss_iou": 0.154296875,
      "loss_num": 0.013427734375,
      "loss_xval": 0.375,
      "num_input_tokens_seen": 496305148,
      "step": 8858
    },
    {
      "epoch": 19.73051224944321,
      "grad_norm": 17.930313110351562,
      "learning_rate": 1e-06,
      "loss": 0.4452,
      "num_input_tokens_seen": 496360372,
      "step": 8859
    },
    {
      "epoch": 19.73051224944321,
      "loss": 0.4869701862335205,
      "loss_ce": 6.224210665095598e-05,
      "loss_iou": 0.228515625,
      "loss_num": 0.005859375,
      "loss_xval": 0.486328125,
      "num_input_tokens_seen": 496360372,
      "step": 8859
    },
    {
      "epoch": 19.732739420935413,
      "grad_norm": 20.98902702331543,
      "learning_rate": 1e-06,
      "loss": 0.3769,
      "num_input_tokens_seen": 496418148,
      "step": 8860
    },
    {
      "epoch": 19.732739420935413,
      "loss": 0.367613285779953,
      "loss_ce": 5.957091343589127e-05,
      "loss_iou": 0.16015625,
      "loss_num": 0.00927734375,
      "loss_xval": 0.3671875,
      "num_input_tokens_seen": 496418148,
      "step": 8860
    },
    {
      "epoch": 19.734966592427618,
      "grad_norm": 27.783918380737305,
      "learning_rate": 1e-06,
      "loss": 0.3618,
      "num_input_tokens_seen": 496475176,
      "step": 8861
    },
    {
      "epoch": 19.734966592427618,
      "loss": 0.31055858731269836,
      "loss_ce": 7.274634845089167e-05,
      "loss_iou": 0.1240234375,
      "loss_num": 0.012451171875,
      "loss_xval": 0.310546875,
      "num_input_tokens_seen": 496475176,
      "step": 8861
    },
    {
      "epoch": 19.737193763919823,
      "grad_norm": 59.55124282836914,
      "learning_rate": 1e-06,
      "loss": 0.3794,
      "num_input_tokens_seen": 496530672,
      "step": 8862
    },
    {
      "epoch": 19.737193763919823,
      "loss": 0.5371950268745422,
      "loss_ce": 8.565981988795102e-05,
      "loss_iou": 0.251953125,
      "loss_num": 0.00640869140625,
      "loss_xval": 0.5390625,
      "num_input_tokens_seen": 496530672,
      "step": 8862
    },
    {
      "epoch": 19.739420935412028,
      "grad_norm": 20.648197174072266,
      "learning_rate": 1e-06,
      "loss": 0.4529,
      "num_input_tokens_seen": 496587424,
      "step": 8863
    },
    {
      "epoch": 19.739420935412028,
      "loss": 0.3034108877182007,
      "loss_ce": 6.618791667278856e-05,
      "loss_iou": 0.1357421875,
      "loss_num": 0.00653076171875,
      "loss_xval": 0.302734375,
      "num_input_tokens_seen": 496587424,
      "step": 8863
    },
    {
      "epoch": 19.741648106904233,
      "grad_norm": 20.388151168823242,
      "learning_rate": 1e-06,
      "loss": 0.5532,
      "num_input_tokens_seen": 496645436,
      "step": 8864
    },
    {
      "epoch": 19.741648106904233,
      "loss": 0.7695181369781494,
      "loss_ce": 0.00010891164129134268,
      "loss_iou": 0.2890625,
      "loss_num": 0.037841796875,
      "loss_xval": 0.76953125,
      "num_input_tokens_seen": 496645436,
      "step": 8864
    },
    {
      "epoch": 19.743875278396438,
      "grad_norm": 13.303028106689453,
      "learning_rate": 1e-06,
      "loss": 0.3442,
      "num_input_tokens_seen": 496701972,
      "step": 8865
    },
    {
      "epoch": 19.743875278396438,
      "loss": 0.2887694835662842,
      "loss_ce": 7.317646668525413e-05,
      "loss_iou": 0.12890625,
      "loss_num": 0.006439208984375,
      "loss_xval": 0.2890625,
      "num_input_tokens_seen": 496701972,
      "step": 8865
    },
    {
      "epoch": 19.746102449888642,
      "grad_norm": 63.60599899291992,
      "learning_rate": 1e-06,
      "loss": 0.3638,
      "num_input_tokens_seen": 496755828,
      "step": 8866
    },
    {
      "epoch": 19.746102449888642,
      "loss": 0.4621775150299072,
      "loss_ce": 8.034885104279965e-05,
      "loss_iou": 0.1904296875,
      "loss_num": 0.0159912109375,
      "loss_xval": 0.462890625,
      "num_input_tokens_seen": 496755828,
      "step": 8866
    },
    {
      "epoch": 19.748329621380847,
      "grad_norm": 22.44493293762207,
      "learning_rate": 1e-06,
      "loss": 0.3002,
      "num_input_tokens_seen": 496812708,
      "step": 8867
    },
    {
      "epoch": 19.748329621380847,
      "loss": 0.3059879541397095,
      "loss_ce": 7.977118366397917e-05,
      "loss_iou": 0.1298828125,
      "loss_num": 0.00921630859375,
      "loss_xval": 0.306640625,
      "num_input_tokens_seen": 496812708,
      "step": 8867
    },
    {
      "epoch": 19.750556792873052,
      "grad_norm": 21.83323860168457,
      "learning_rate": 1e-06,
      "loss": 0.385,
      "num_input_tokens_seen": 496868560,
      "step": 8868
    },
    {
      "epoch": 19.750556792873052,
      "loss": 0.30378103256225586,
      "loss_ce": 7.010510307736695e-05,
      "loss_iou": 0.1416015625,
      "loss_num": 0.004364013671875,
      "loss_xval": 0.3046875,
      "num_input_tokens_seen": 496868560,
      "step": 8868
    },
    {
      "epoch": 19.752783964365257,
      "grad_norm": 32.918434143066406,
      "learning_rate": 1e-06,
      "loss": 0.4902,
      "num_input_tokens_seen": 496922692,
      "step": 8869
    },
    {
      "epoch": 19.752783964365257,
      "loss": 0.5124045014381409,
      "loss_ce": 7.53938511479646e-05,
      "loss_iou": 0.2333984375,
      "loss_num": 0.0091552734375,
      "loss_xval": 0.51171875,
      "num_input_tokens_seen": 496922692,
      "step": 8869
    },
    {
      "epoch": 19.755011135857462,
      "grad_norm": 25.483102798461914,
      "learning_rate": 1e-06,
      "loss": 0.3786,
      "num_input_tokens_seen": 496980516,
      "step": 8870
    },
    {
      "epoch": 19.755011135857462,
      "loss": 0.3339354991912842,
      "loss_ce": 7.317691051866859e-05,
      "loss_iou": 0.154296875,
      "loss_num": 0.004913330078125,
      "loss_xval": 0.333984375,
      "num_input_tokens_seen": 496980516,
      "step": 8870
    },
    {
      "epoch": 19.757238307349667,
      "grad_norm": 15.504528045654297,
      "learning_rate": 1e-06,
      "loss": 0.4814,
      "num_input_tokens_seen": 497037204,
      "step": 8871
    },
    {
      "epoch": 19.757238307349667,
      "loss": 0.6138589382171631,
      "loss_ce": 7.417285814881325e-05,
      "loss_iou": 0.26953125,
      "loss_num": 0.0145263671875,
      "loss_xval": 0.61328125,
      "num_input_tokens_seen": 497037204,
      "step": 8871
    },
    {
      "epoch": 19.75946547884187,
      "grad_norm": 25.176443099975586,
      "learning_rate": 1e-06,
      "loss": 0.3316,
      "num_input_tokens_seen": 497089592,
      "step": 8872
    },
    {
      "epoch": 19.75946547884187,
      "loss": 0.2530931532382965,
      "loss_ce": 7.191546319518238e-05,
      "loss_iou": 0.1123046875,
      "loss_num": 0.005584716796875,
      "loss_xval": 0.25390625,
      "num_input_tokens_seen": 497089592,
      "step": 8872
    },
    {
      "epoch": 19.761692650334076,
      "grad_norm": 15.639261245727539,
      "learning_rate": 1e-06,
      "loss": 0.3225,
      "num_input_tokens_seen": 497143680,
      "step": 8873
    },
    {
      "epoch": 19.761692650334076,
      "loss": 0.2651139497756958,
      "loss_ce": 9.929385123541579e-05,
      "loss_iou": 0.11767578125,
      "loss_num": 0.005950927734375,
      "loss_xval": 0.265625,
      "num_input_tokens_seen": 497143680,
      "step": 8873
    },
    {
      "epoch": 19.76391982182628,
      "grad_norm": 23.403366088867188,
      "learning_rate": 1e-06,
      "loss": 0.3117,
      "num_input_tokens_seen": 497198980,
      "step": 8874
    },
    {
      "epoch": 19.76391982182628,
      "loss": 0.3706684112548828,
      "loss_ce": 6.295710045378655e-05,
      "loss_iou": 0.171875,
      "loss_num": 0.005462646484375,
      "loss_xval": 0.37109375,
      "num_input_tokens_seen": 497198980,
      "step": 8874
    },
    {
      "epoch": 19.766146993318486,
      "grad_norm": 14.233652114868164,
      "learning_rate": 1e-06,
      "loss": 0.3257,
      "num_input_tokens_seen": 497257808,
      "step": 8875
    },
    {
      "epoch": 19.766146993318486,
      "loss": 0.41550278663635254,
      "loss_ce": 0.0003416564140934497,
      "loss_iou": 0.1669921875,
      "loss_num": 0.0164794921875,
      "loss_xval": 0.416015625,
      "num_input_tokens_seen": 497257808,
      "step": 8875
    },
    {
      "epoch": 19.76837416481069,
      "grad_norm": 13.950034141540527,
      "learning_rate": 1e-06,
      "loss": 0.3255,
      "num_input_tokens_seen": 497314976,
      "step": 8876
    },
    {
      "epoch": 19.76837416481069,
      "loss": 0.2907865047454834,
      "loss_ce": 7.603740959893912e-05,
      "loss_iou": 0.12890625,
      "loss_num": 0.006561279296875,
      "loss_xval": 0.291015625,
      "num_input_tokens_seen": 497314976,
      "step": 8876
    },
    {
      "epoch": 19.770601336302896,
      "grad_norm": 12.77400016784668,
      "learning_rate": 1e-06,
      "loss": 0.3122,
      "num_input_tokens_seen": 497371316,
      "step": 8877
    },
    {
      "epoch": 19.770601336302896,
      "loss": 0.302242249250412,
      "loss_ce": 5.7184963225154206e-05,
      "loss_iou": 0.1328125,
      "loss_num": 0.00732421875,
      "loss_xval": 0.302734375,
      "num_input_tokens_seen": 497371316,
      "step": 8877
    },
    {
      "epoch": 19.7728285077951,
      "grad_norm": 16.136669158935547,
      "learning_rate": 1e-06,
      "loss": 0.4607,
      "num_input_tokens_seen": 497428012,
      "step": 8878
    },
    {
      "epoch": 19.7728285077951,
      "loss": 0.7113887667655945,
      "loss_ce": 8.501914271619171e-05,
      "loss_iou": 0.314453125,
      "loss_num": 0.0159912109375,
      "loss_xval": 0.7109375,
      "num_input_tokens_seen": 497428012,
      "step": 8878
    },
    {
      "epoch": 19.775055679287306,
      "grad_norm": 26.28825569152832,
      "learning_rate": 1e-06,
      "loss": 0.2925,
      "num_input_tokens_seen": 497485468,
      "step": 8879
    },
    {
      "epoch": 19.775055679287306,
      "loss": 0.346199095249176,
      "loss_ce": 6.87175925122574e-05,
      "loss_iou": 0.1611328125,
      "loss_num": 0.00469970703125,
      "loss_xval": 0.345703125,
      "num_input_tokens_seen": 497485468,
      "step": 8879
    },
    {
      "epoch": 19.77728285077951,
      "grad_norm": 17.712745666503906,
      "learning_rate": 1e-06,
      "loss": 0.3412,
      "num_input_tokens_seen": 497543128,
      "step": 8880
    },
    {
      "epoch": 19.77728285077951,
      "loss": 0.32294753193855286,
      "loss_ce": 7.154815102694556e-05,
      "loss_iou": 0.1474609375,
      "loss_num": 0.00537109375,
      "loss_xval": 0.322265625,
      "num_input_tokens_seen": 497543128,
      "step": 8880
    },
    {
      "epoch": 19.779510022271715,
      "grad_norm": 16.877283096313477,
      "learning_rate": 1e-06,
      "loss": 0.3093,
      "num_input_tokens_seen": 497601184,
      "step": 8881
    },
    {
      "epoch": 19.779510022271715,
      "loss": 0.26191645860671997,
      "loss_ce": 7.561576785519719e-05,
      "loss_iou": 0.10791015625,
      "loss_num": 0.0091552734375,
      "loss_xval": 0.26171875,
      "num_input_tokens_seen": 497601184,
      "step": 8881
    },
    {
      "epoch": 19.78173719376392,
      "grad_norm": 15.67979621887207,
      "learning_rate": 1e-06,
      "loss": 0.3683,
      "num_input_tokens_seen": 497658448,
      "step": 8882
    },
    {
      "epoch": 19.78173719376392,
      "loss": 0.4387954771518707,
      "loss_ce": 7.479546184185892e-05,
      "loss_iou": 0.171875,
      "loss_num": 0.0189208984375,
      "loss_xval": 0.439453125,
      "num_input_tokens_seen": 497658448,
      "step": 8882
    },
    {
      "epoch": 19.783964365256125,
      "grad_norm": 15.870912551879883,
      "learning_rate": 1e-06,
      "loss": 0.4725,
      "num_input_tokens_seen": 497714064,
      "step": 8883
    },
    {
      "epoch": 19.783964365256125,
      "loss": 0.47480326890945435,
      "loss_ce": 7.185728463809937e-05,
      "loss_iou": 0.212890625,
      "loss_num": 0.0096435546875,
      "loss_xval": 0.474609375,
      "num_input_tokens_seen": 497714064,
      "step": 8883
    },
    {
      "epoch": 19.78619153674833,
      "grad_norm": 22.020280838012695,
      "learning_rate": 1e-06,
      "loss": 0.5039,
      "num_input_tokens_seen": 497768280,
      "step": 8884
    },
    {
      "epoch": 19.78619153674833,
      "loss": 0.38886767625808716,
      "loss_ce": 7.375521818175912e-05,
      "loss_iou": 0.1865234375,
      "loss_num": 0.0032501220703125,
      "loss_xval": 0.388671875,
      "num_input_tokens_seen": 497768280,
      "step": 8884
    },
    {
      "epoch": 19.788418708240535,
      "grad_norm": 28.759016036987305,
      "learning_rate": 1e-06,
      "loss": 0.4566,
      "num_input_tokens_seen": 497822296,
      "step": 8885
    },
    {
      "epoch": 19.788418708240535,
      "loss": 0.5814878940582275,
      "loss_ce": 6.704413681291044e-05,
      "loss_iou": 0.236328125,
      "loss_num": 0.021728515625,
      "loss_xval": 0.58203125,
      "num_input_tokens_seen": 497822296,
      "step": 8885
    },
    {
      "epoch": 19.79064587973274,
      "grad_norm": 19.10303497314453,
      "learning_rate": 1e-06,
      "loss": 0.6282,
      "num_input_tokens_seen": 497880436,
      "step": 8886
    },
    {
      "epoch": 19.79064587973274,
      "loss": 0.586006760597229,
      "loss_ce": 6.921241583768278e-05,
      "loss_iou": 0.2275390625,
      "loss_num": 0.0263671875,
      "loss_xval": 0.5859375,
      "num_input_tokens_seen": 497880436,
      "step": 8886
    },
    {
      "epoch": 19.792873051224944,
      "grad_norm": 24.136104583740234,
      "learning_rate": 1e-06,
      "loss": 0.3318,
      "num_input_tokens_seen": 497934040,
      "step": 8887
    },
    {
      "epoch": 19.792873051224944,
      "loss": 0.3225743770599365,
      "loss_ce": 6.460573058575392e-05,
      "loss_iou": 0.140625,
      "loss_num": 0.00830078125,
      "loss_xval": 0.322265625,
      "num_input_tokens_seen": 497934040,
      "step": 8887
    },
    {
      "epoch": 19.79510022271715,
      "grad_norm": 15.562583923339844,
      "learning_rate": 1e-06,
      "loss": 0.4173,
      "num_input_tokens_seen": 497989992,
      "step": 8888
    },
    {
      "epoch": 19.79510022271715,
      "loss": 0.4295352101325989,
      "loss_ce": 6.133929127827287e-05,
      "loss_iou": 0.16796875,
      "loss_num": 0.0185546875,
      "loss_xval": 0.4296875,
      "num_input_tokens_seen": 497989992,
      "step": 8888
    },
    {
      "epoch": 19.797327394209354,
      "grad_norm": 12.647050857543945,
      "learning_rate": 1e-06,
      "loss": 0.3252,
      "num_input_tokens_seen": 498045484,
      "step": 8889
    },
    {
      "epoch": 19.797327394209354,
      "loss": 0.26227039098739624,
      "loss_ce": 6.334840873023495e-05,
      "loss_iou": 0.107421875,
      "loss_num": 0.009521484375,
      "loss_xval": 0.26171875,
      "num_input_tokens_seen": 498045484,
      "step": 8889
    },
    {
      "epoch": 19.79955456570156,
      "grad_norm": 17.201618194580078,
      "learning_rate": 1e-06,
      "loss": 0.3303,
      "num_input_tokens_seen": 498102048,
      "step": 8890
    },
    {
      "epoch": 19.79955456570156,
      "loss": 0.31794556975364685,
      "loss_ce": 7.44800636311993e-05,
      "loss_iou": 0.1279296875,
      "loss_num": 0.01220703125,
      "loss_xval": 0.318359375,
      "num_input_tokens_seen": 498102048,
      "step": 8890
    },
    {
      "epoch": 19.801781737193764,
      "grad_norm": 22.599042892456055,
      "learning_rate": 1e-06,
      "loss": 0.3699,
      "num_input_tokens_seen": 498159040,
      "step": 8891
    },
    {
      "epoch": 19.801781737193764,
      "loss": 0.2396983951330185,
      "loss_ce": 7.437313615810126e-05,
      "loss_iou": 0.107421875,
      "loss_num": 0.00494384765625,
      "loss_xval": 0.2392578125,
      "num_input_tokens_seen": 498159040,
      "step": 8891
    },
    {
      "epoch": 19.80400890868597,
      "grad_norm": 26.07795524597168,
      "learning_rate": 1e-06,
      "loss": 0.4677,
      "num_input_tokens_seen": 498212312,
      "step": 8892
    },
    {
      "epoch": 19.80400890868597,
      "loss": 0.5568862557411194,
      "loss_ce": 0.00012353430793154985,
      "loss_iou": 0.2333984375,
      "loss_num": 0.017822265625,
      "loss_xval": 0.55859375,
      "num_input_tokens_seen": 498212312,
      "step": 8892
    },
    {
      "epoch": 19.806236080178174,
      "grad_norm": 22.389713287353516,
      "learning_rate": 1e-06,
      "loss": 0.3661,
      "num_input_tokens_seen": 498267900,
      "step": 8893
    },
    {
      "epoch": 19.806236080178174,
      "loss": 0.4324456751346588,
      "loss_ce": 7.261322753038257e-05,
      "loss_iou": 0.201171875,
      "loss_num": 0.005950927734375,
      "loss_xval": 0.431640625,
      "num_input_tokens_seen": 498267900,
      "step": 8893
    },
    {
      "epoch": 19.80846325167038,
      "grad_norm": 16.400392532348633,
      "learning_rate": 1e-06,
      "loss": 0.2981,
      "num_input_tokens_seen": 498324320,
      "step": 8894
    },
    {
      "epoch": 19.80846325167038,
      "loss": 0.27405545115470886,
      "loss_ce": 6.862355803605169e-05,
      "loss_iou": 0.1279296875,
      "loss_num": 0.00360107421875,
      "loss_xval": 0.2734375,
      "num_input_tokens_seen": 498324320,
      "step": 8894
    },
    {
      "epoch": 19.810690423162583,
      "grad_norm": 21.506378173828125,
      "learning_rate": 1e-06,
      "loss": 0.3227,
      "num_input_tokens_seen": 498378944,
      "step": 8895
    },
    {
      "epoch": 19.810690423162583,
      "loss": 0.3589167296886444,
      "loss_ce": 6.0540056438185275e-05,
      "loss_iou": 0.1533203125,
      "loss_num": 0.01043701171875,
      "loss_xval": 0.359375,
      "num_input_tokens_seen": 498378944,
      "step": 8895
    },
    {
      "epoch": 19.812917594654788,
      "grad_norm": 23.304874420166016,
      "learning_rate": 1e-06,
      "loss": 0.2775,
      "num_input_tokens_seen": 498432504,
      "step": 8896
    },
    {
      "epoch": 19.812917594654788,
      "loss": 0.27460119128227234,
      "loss_ce": 6.50603833491914e-05,
      "loss_iou": 0.1240234375,
      "loss_num": 0.005218505859375,
      "loss_xval": 0.275390625,
      "num_input_tokens_seen": 498432504,
      "step": 8896
    },
    {
      "epoch": 19.815144766146993,
      "grad_norm": 15.28256893157959,
      "learning_rate": 1e-06,
      "loss": 0.334,
      "num_input_tokens_seen": 498489036,
      "step": 8897
    },
    {
      "epoch": 19.815144766146993,
      "loss": 0.3256245255470276,
      "loss_ce": 6.301122630247846e-05,
      "loss_iou": 0.150390625,
      "loss_num": 0.00518798828125,
      "loss_xval": 0.326171875,
      "num_input_tokens_seen": 498489036,
      "step": 8897
    },
    {
      "epoch": 19.817371937639198,
      "grad_norm": 37.541343688964844,
      "learning_rate": 1e-06,
      "loss": 0.556,
      "num_input_tokens_seen": 498544072,
      "step": 8898
    },
    {
      "epoch": 19.817371937639198,
      "loss": 0.7319257259368896,
      "loss_ce": 0.0001142127366620116,
      "loss_iou": 0.26953125,
      "loss_num": 0.038818359375,
      "loss_xval": 0.73046875,
      "num_input_tokens_seen": 498544072,
      "step": 8898
    },
    {
      "epoch": 19.819599109131403,
      "grad_norm": 35.13840866088867,
      "learning_rate": 1e-06,
      "loss": 0.4129,
      "num_input_tokens_seen": 498600592,
      "step": 8899
    },
    {
      "epoch": 19.819599109131403,
      "loss": 0.2181350737810135,
      "loss_ce": 5.645793498842977e-05,
      "loss_iou": 0.08837890625,
      "loss_num": 0.00823974609375,
      "loss_xval": 0.2177734375,
      "num_input_tokens_seen": 498600592,
      "step": 8899
    },
    {
      "epoch": 19.821826280623608,
      "grad_norm": 18.837566375732422,
      "learning_rate": 1e-06,
      "loss": 0.3815,
      "num_input_tokens_seen": 498655692,
      "step": 8900
    },
    {
      "epoch": 19.821826280623608,
      "loss": 0.21076907217502594,
      "loss_ce": 6.046070120646618e-05,
      "loss_iou": 0.09521484375,
      "loss_num": 0.004058837890625,
      "loss_xval": 0.2109375,
      "num_input_tokens_seen": 498655692,
      "step": 8900
    },
    {
      "epoch": 19.824053452115812,
      "grad_norm": 14.442410469055176,
      "learning_rate": 1e-06,
      "loss": 0.4286,
      "num_input_tokens_seen": 498710648,
      "step": 8901
    },
    {
      "epoch": 19.824053452115812,
      "loss": 0.48480677604675293,
      "loss_ce": 6.559376924997196e-05,
      "loss_iou": 0.205078125,
      "loss_num": 0.0150146484375,
      "loss_xval": 0.484375,
      "num_input_tokens_seen": 498710648,
      "step": 8901
    },
    {
      "epoch": 19.826280623608017,
      "grad_norm": 15.92866039276123,
      "learning_rate": 1e-06,
      "loss": 0.3134,
      "num_input_tokens_seen": 498766436,
      "step": 8902
    },
    {
      "epoch": 19.826280623608017,
      "loss": 0.35358044505119324,
      "loss_ce": 6.481447053374723e-05,
      "loss_iou": 0.1328125,
      "loss_num": 0.017578125,
      "loss_xval": 0.353515625,
      "num_input_tokens_seen": 498766436,
      "step": 8902
    },
    {
      "epoch": 19.828507795100222,
      "grad_norm": 18.898021697998047,
      "learning_rate": 1e-06,
      "loss": 0.5045,
      "num_input_tokens_seen": 498821800,
      "step": 8903
    },
    {
      "epoch": 19.828507795100222,
      "loss": 0.49323803186416626,
      "loss_ce": 7.400984031846747e-05,
      "loss_iou": 0.2119140625,
      "loss_num": 0.01385498046875,
      "loss_xval": 0.4921875,
      "num_input_tokens_seen": 498821800,
      "step": 8903
    },
    {
      "epoch": 19.830734966592427,
      "grad_norm": 11.007697105407715,
      "learning_rate": 1e-06,
      "loss": 0.2899,
      "num_input_tokens_seen": 498876356,
      "step": 8904
    },
    {
      "epoch": 19.830734966592427,
      "loss": 0.2739826440811157,
      "loss_ce": 0.00017892879259306937,
      "loss_iou": 0.11669921875,
      "loss_num": 0.0081787109375,
      "loss_xval": 0.2734375,
      "num_input_tokens_seen": 498876356,
      "step": 8904
    },
    {
      "epoch": 19.832962138084632,
      "grad_norm": 11.861038208007812,
      "learning_rate": 1e-06,
      "loss": 0.388,
      "num_input_tokens_seen": 498934096,
      "step": 8905
    },
    {
      "epoch": 19.832962138084632,
      "loss": 0.34637463092803955,
      "loss_ce": 6.114893039921299e-05,
      "loss_iou": 0.154296875,
      "loss_num": 0.00738525390625,
      "loss_xval": 0.345703125,
      "num_input_tokens_seen": 498934096,
      "step": 8905
    },
    {
      "epoch": 19.835189309576837,
      "grad_norm": 18.919658660888672,
      "learning_rate": 1e-06,
      "loss": 0.3537,
      "num_input_tokens_seen": 498989812,
      "step": 8906
    },
    {
      "epoch": 19.835189309576837,
      "loss": 0.30976182222366333,
      "loss_ce": 6.946048233658075e-05,
      "loss_iou": 0.1435546875,
      "loss_num": 0.004669189453125,
      "loss_xval": 0.310546875,
      "num_input_tokens_seen": 498989812,
      "step": 8906
    },
    {
      "epoch": 19.83741648106904,
      "grad_norm": 15.281343460083008,
      "learning_rate": 1e-06,
      "loss": 0.3773,
      "num_input_tokens_seen": 499047512,
      "step": 8907
    },
    {
      "epoch": 19.83741648106904,
      "loss": 0.3993797302246094,
      "loss_ce": 8.777561015449464e-05,
      "loss_iou": 0.1650390625,
      "loss_num": 0.01385498046875,
      "loss_xval": 0.3984375,
      "num_input_tokens_seen": 499047512,
      "step": 8907
    },
    {
      "epoch": 19.839643652561247,
      "grad_norm": 14.046529769897461,
      "learning_rate": 1e-06,
      "loss": 0.431,
      "num_input_tokens_seen": 499105152,
      "step": 8908
    },
    {
      "epoch": 19.839643652561247,
      "loss": 0.6798744201660156,
      "loss_ce": 6.483346078312024e-05,
      "loss_iou": 0.279296875,
      "loss_num": 0.024169921875,
      "loss_xval": 0.6796875,
      "num_input_tokens_seen": 499105152,
      "step": 8908
    },
    {
      "epoch": 19.84187082405345,
      "grad_norm": 19.07868766784668,
      "learning_rate": 1e-06,
      "loss": 0.362,
      "num_input_tokens_seen": 499160952,
      "step": 8909
    },
    {
      "epoch": 19.84187082405345,
      "loss": 0.5317122936248779,
      "loss_ce": 9.61277837632224e-05,
      "loss_iou": 0.205078125,
      "loss_num": 0.024169921875,
      "loss_xval": 0.53125,
      "num_input_tokens_seen": 499160952,
      "step": 8909
    },
    {
      "epoch": 19.844097995545656,
      "grad_norm": 17.901517868041992,
      "learning_rate": 1e-06,
      "loss": 0.4618,
      "num_input_tokens_seen": 499217672,
      "step": 8910
    },
    {
      "epoch": 19.844097995545656,
      "loss": 0.3117324709892273,
      "loss_ce": 8.69435680215247e-05,
      "loss_iou": 0.12158203125,
      "loss_num": 0.0135498046875,
      "loss_xval": 0.3125,
      "num_input_tokens_seen": 499217672,
      "step": 8910
    },
    {
      "epoch": 19.84632516703786,
      "grad_norm": 28.3020076751709,
      "learning_rate": 1e-06,
      "loss": 0.433,
      "num_input_tokens_seen": 499274752,
      "step": 8911
    },
    {
      "epoch": 19.84632516703786,
      "loss": 0.3874030113220215,
      "loss_ce": 7.39007446100004e-05,
      "loss_iou": 0.1787109375,
      "loss_num": 0.0059814453125,
      "loss_xval": 0.38671875,
      "num_input_tokens_seen": 499274752,
      "step": 8911
    },
    {
      "epoch": 19.848552338530066,
      "grad_norm": 29.61766242980957,
      "learning_rate": 1e-06,
      "loss": 0.3846,
      "num_input_tokens_seen": 499327256,
      "step": 8912
    },
    {
      "epoch": 19.848552338530066,
      "loss": 0.24732181429862976,
      "loss_ce": 6.841085996711627e-05,
      "loss_iou": 0.111328125,
      "loss_num": 0.004974365234375,
      "loss_xval": 0.2470703125,
      "num_input_tokens_seen": 499327256,
      "step": 8912
    },
    {
      "epoch": 19.85077951002227,
      "grad_norm": 25.11189079284668,
      "learning_rate": 1e-06,
      "loss": 0.3364,
      "num_input_tokens_seen": 499379248,
      "step": 8913
    },
    {
      "epoch": 19.85077951002227,
      "loss": 0.40423035621643066,
      "loss_ce": 5.5550888646394014e-05,
      "loss_iou": 0.1640625,
      "loss_num": 0.01513671875,
      "loss_xval": 0.404296875,
      "num_input_tokens_seen": 499379248,
      "step": 8913
    },
    {
      "epoch": 19.853006681514476,
      "grad_norm": 18.35516929626465,
      "learning_rate": 1e-06,
      "loss": 0.4902,
      "num_input_tokens_seen": 499436244,
      "step": 8914
    },
    {
      "epoch": 19.853006681514476,
      "loss": 0.41144680976867676,
      "loss_ce": 6.984162610024214e-05,
      "loss_iou": 0.18359375,
      "loss_num": 0.00885009765625,
      "loss_xval": 0.412109375,
      "num_input_tokens_seen": 499436244,
      "step": 8914
    },
    {
      "epoch": 19.85523385300668,
      "grad_norm": 14.630391120910645,
      "learning_rate": 1e-06,
      "loss": 0.3681,
      "num_input_tokens_seen": 499493472,
      "step": 8915
    },
    {
      "epoch": 19.85523385300668,
      "loss": 0.3745167851448059,
      "loss_ce": 6.610387936234474e-05,
      "loss_iou": 0.16796875,
      "loss_num": 0.0078125,
      "loss_xval": 0.375,
      "num_input_tokens_seen": 499493472,
      "step": 8915
    },
    {
      "epoch": 19.857461024498885,
      "grad_norm": 15.143143653869629,
      "learning_rate": 1e-06,
      "loss": 0.4017,
      "num_input_tokens_seen": 499548600,
      "step": 8916
    },
    {
      "epoch": 19.857461024498885,
      "loss": 0.5689959526062012,
      "loss_ce": 5.67264505662024e-05,
      "loss_iou": 0.2197265625,
      "loss_num": 0.026123046875,
      "loss_xval": 0.5703125,
      "num_input_tokens_seen": 499548600,
      "step": 8916
    },
    {
      "epoch": 19.85968819599109,
      "grad_norm": 25.465965270996094,
      "learning_rate": 1e-06,
      "loss": 0.3066,
      "num_input_tokens_seen": 499601500,
      "step": 8917
    },
    {
      "epoch": 19.85968819599109,
      "loss": 0.31936201453208923,
      "loss_ce": 8.711648843018338e-05,
      "loss_iou": 0.1328125,
      "loss_num": 0.01055908203125,
      "loss_xval": 0.318359375,
      "num_input_tokens_seen": 499601500,
      "step": 8917
    },
    {
      "epoch": 19.861915367483295,
      "grad_norm": 21.30182647705078,
      "learning_rate": 1e-06,
      "loss": 0.4865,
      "num_input_tokens_seen": 499657012,
      "step": 8918
    },
    {
      "epoch": 19.861915367483295,
      "loss": 0.38416028022766113,
      "loss_ce": 6.603718065889552e-05,
      "loss_iou": 0.154296875,
      "loss_num": 0.0152587890625,
      "loss_xval": 0.384765625,
      "num_input_tokens_seen": 499657012,
      "step": 8918
    },
    {
      "epoch": 19.8641425389755,
      "grad_norm": 19.05086898803711,
      "learning_rate": 1e-06,
      "loss": 0.5037,
      "num_input_tokens_seen": 499713108,
      "step": 8919
    },
    {
      "epoch": 19.8641425389755,
      "loss": 0.5627080202102661,
      "loss_ce": 8.592565427534282e-05,
      "loss_iou": 0.244140625,
      "loss_num": 0.01513671875,
      "loss_xval": 0.5625,
      "num_input_tokens_seen": 499713108,
      "step": 8919
    },
    {
      "epoch": 19.866369710467705,
      "grad_norm": 21.24616813659668,
      "learning_rate": 1e-06,
      "loss": 0.3316,
      "num_input_tokens_seen": 499771624,
      "step": 8920
    },
    {
      "epoch": 19.866369710467705,
      "loss": 0.3205050528049469,
      "loss_ce": 7.047764665912837e-05,
      "loss_iou": 0.1376953125,
      "loss_num": 0.0089111328125,
      "loss_xval": 0.3203125,
      "num_input_tokens_seen": 499771624,
      "step": 8920
    },
    {
      "epoch": 19.86859688195991,
      "grad_norm": 18.13334846496582,
      "learning_rate": 1e-06,
      "loss": 0.3272,
      "num_input_tokens_seen": 499825620,
      "step": 8921
    },
    {
      "epoch": 19.86859688195991,
      "loss": 0.3275926113128662,
      "loss_ce": 7.797064608894289e-05,
      "loss_iou": 0.1416015625,
      "loss_num": 0.0087890625,
      "loss_xval": 0.328125,
      "num_input_tokens_seen": 499825620,
      "step": 8921
    },
    {
      "epoch": 19.870824053452115,
      "grad_norm": 13.680047988891602,
      "learning_rate": 1e-06,
      "loss": 0.7128,
      "num_input_tokens_seen": 499881464,
      "step": 8922
    },
    {
      "epoch": 19.870824053452115,
      "loss": 0.5840435028076172,
      "loss_ce": 5.915117799304426e-05,
      "loss_iou": 0.2431640625,
      "loss_num": 0.0194091796875,
      "loss_xval": 0.5859375,
      "num_input_tokens_seen": 499881464,
      "step": 8922
    },
    {
      "epoch": 19.87305122494432,
      "grad_norm": 17.974990844726562,
      "learning_rate": 1e-06,
      "loss": 0.3335,
      "num_input_tokens_seen": 499935500,
      "step": 8923
    },
    {
      "epoch": 19.87305122494432,
      "loss": 0.39403825998306274,
      "loss_ce": 5.633125692838803e-05,
      "loss_iou": 0.1533203125,
      "loss_num": 0.017333984375,
      "loss_xval": 0.39453125,
      "num_input_tokens_seen": 499935500,
      "step": 8923
    },
    {
      "epoch": 19.875278396436524,
      "grad_norm": 16.1640625,
      "learning_rate": 1e-06,
      "loss": 0.5226,
      "num_input_tokens_seen": 499991676,
      "step": 8924
    },
    {
      "epoch": 19.875278396436524,
      "loss": 0.3693428039550781,
      "loss_ce": 0.00020217550627421588,
      "loss_iou": 0.1669921875,
      "loss_num": 0.0069580078125,
      "loss_xval": 0.369140625,
      "num_input_tokens_seen": 499991676,
      "step": 8924
    },
    {
      "epoch": 19.87750556792873,
      "grad_norm": 17.915504455566406,
      "learning_rate": 1e-06,
      "loss": 0.4869,
      "num_input_tokens_seen": 500049156,
      "step": 8925
    },
    {
      "epoch": 19.87750556792873,
      "loss": 0.44209641218185425,
      "loss_ce": 7.978252688189968e-05,
      "loss_iou": 0.2001953125,
      "loss_num": 0.00811767578125,
      "loss_xval": 0.44140625,
      "num_input_tokens_seen": 500049156,
      "step": 8925
    },
    {
      "epoch": 19.879732739420934,
      "grad_norm": 14.218931198120117,
      "learning_rate": 1e-06,
      "loss": 0.3029,
      "num_input_tokens_seen": 500105296,
      "step": 8926
    },
    {
      "epoch": 19.879732739420934,
      "loss": 0.33782947063446045,
      "loss_ce": 6.0907594161108136e-05,
      "loss_iou": 0.154296875,
      "loss_num": 0.005950927734375,
      "loss_xval": 0.337890625,
      "num_input_tokens_seen": 500105296,
      "step": 8926
    },
    {
      "epoch": 19.88195991091314,
      "grad_norm": 14.327959060668945,
      "learning_rate": 1e-06,
      "loss": 0.3458,
      "num_input_tokens_seen": 500161776,
      "step": 8927
    },
    {
      "epoch": 19.88195991091314,
      "loss": 0.3156734108924866,
      "loss_ce": 6.0611731896642596e-05,
      "loss_iou": 0.134765625,
      "loss_num": 0.00897216796875,
      "loss_xval": 0.31640625,
      "num_input_tokens_seen": 500161776,
      "step": 8927
    },
    {
      "epoch": 19.884187082405344,
      "grad_norm": 19.522199630737305,
      "learning_rate": 1e-06,
      "loss": 0.5647,
      "num_input_tokens_seen": 500216288,
      "step": 8928
    },
    {
      "epoch": 19.884187082405344,
      "loss": 0.6144946813583374,
      "loss_ce": 5.3751304221805185e-05,
      "loss_iou": 0.2158203125,
      "loss_num": 0.03662109375,
      "loss_xval": 0.61328125,
      "num_input_tokens_seen": 500216288,
      "step": 8928
    },
    {
      "epoch": 19.88641425389755,
      "grad_norm": 17.130699157714844,
      "learning_rate": 1e-06,
      "loss": 0.3437,
      "num_input_tokens_seen": 500271968,
      "step": 8929
    },
    {
      "epoch": 19.88641425389755,
      "loss": 0.2988947629928589,
      "loss_ce": 6.664128886768594e-05,
      "loss_iou": 0.1318359375,
      "loss_num": 0.006866455078125,
      "loss_xval": 0.298828125,
      "num_input_tokens_seen": 500271968,
      "step": 8929
    },
    {
      "epoch": 19.888641425389753,
      "grad_norm": 29.030588150024414,
      "learning_rate": 1e-06,
      "loss": 0.3789,
      "num_input_tokens_seen": 500329792,
      "step": 8930
    },
    {
      "epoch": 19.888641425389753,
      "loss": 0.40844833850860596,
      "loss_ce": 6.208484410308301e-05,
      "loss_iou": 0.1708984375,
      "loss_num": 0.013427734375,
      "loss_xval": 0.408203125,
      "num_input_tokens_seen": 500329792,
      "step": 8930
    },
    {
      "epoch": 19.89086859688196,
      "grad_norm": 20.700912475585938,
      "learning_rate": 1e-06,
      "loss": 0.3864,
      "num_input_tokens_seen": 500384792,
      "step": 8931
    },
    {
      "epoch": 19.89086859688196,
      "loss": 0.46894556283950806,
      "loss_ce": 7.346599886659533e-05,
      "loss_iou": 0.203125,
      "loss_num": 0.0126953125,
      "loss_xval": 0.46875,
      "num_input_tokens_seen": 500384792,
      "step": 8931
    },
    {
      "epoch": 19.893095768374163,
      "grad_norm": 20.326370239257812,
      "learning_rate": 1e-06,
      "loss": 0.2839,
      "num_input_tokens_seen": 500439812,
      "step": 8932
    },
    {
      "epoch": 19.893095768374163,
      "loss": 0.25635194778442383,
      "loss_ce": 6.53410388622433e-05,
      "loss_iou": 0.1162109375,
      "loss_num": 0.004791259765625,
      "loss_xval": 0.255859375,
      "num_input_tokens_seen": 500439812,
      "step": 8932
    },
    {
      "epoch": 19.895322939866368,
      "grad_norm": 18.51610565185547,
      "learning_rate": 1e-06,
      "loss": 0.4605,
      "num_input_tokens_seen": 500497212,
      "step": 8933
    },
    {
      "epoch": 19.895322939866368,
      "loss": 0.3897075355052948,
      "loss_ce": 5.907195736654103e-05,
      "loss_iou": 0.1669921875,
      "loss_num": 0.01123046875,
      "loss_xval": 0.390625,
      "num_input_tokens_seen": 500497212,
      "step": 8933
    },
    {
      "epoch": 19.897550111358576,
      "grad_norm": 15.211495399475098,
      "learning_rate": 1e-06,
      "loss": 0.3421,
      "num_input_tokens_seen": 500552316,
      "step": 8934
    },
    {
      "epoch": 19.897550111358576,
      "loss": 0.3298904299736023,
      "loss_ce": 5.644091288559139e-05,
      "loss_iou": 0.1376953125,
      "loss_num": 0.01080322265625,
      "loss_xval": 0.330078125,
      "num_input_tokens_seen": 500552316,
      "step": 8934
    },
    {
      "epoch": 19.899777282850778,
      "grad_norm": 20.869205474853516,
      "learning_rate": 1e-06,
      "loss": 0.2746,
      "num_input_tokens_seen": 500606956,
      "step": 8935
    },
    {
      "epoch": 19.899777282850778,
      "loss": 0.28754234313964844,
      "loss_ce": 6.67754648020491e-05,
      "loss_iou": 0.126953125,
      "loss_num": 0.006866455078125,
      "loss_xval": 0.287109375,
      "num_input_tokens_seen": 500606956,
      "step": 8935
    },
    {
      "epoch": 19.902004454342986,
      "grad_norm": 24.170188903808594,
      "learning_rate": 1e-06,
      "loss": 0.389,
      "num_input_tokens_seen": 500662136,
      "step": 8936
    },
    {
      "epoch": 19.902004454342986,
      "loss": 0.32582342624664307,
      "loss_ce": 7.880674093030393e-05,
      "loss_iou": 0.15234375,
      "loss_num": 0.004119873046875,
      "loss_xval": 0.326171875,
      "num_input_tokens_seen": 500662136,
      "step": 8936
    },
    {
      "epoch": 19.90423162583519,
      "grad_norm": 26.785816192626953,
      "learning_rate": 1e-06,
      "loss": 0.3826,
      "num_input_tokens_seen": 500716584,
      "step": 8937
    },
    {
      "epoch": 19.90423162583519,
      "loss": 0.41937169432640076,
      "loss_ce": 6.017117266310379e-05,
      "loss_iou": 0.1787109375,
      "loss_num": 0.0125732421875,
      "loss_xval": 0.419921875,
      "num_input_tokens_seen": 500716584,
      "step": 8937
    },
    {
      "epoch": 19.906458797327396,
      "grad_norm": 15.786996841430664,
      "learning_rate": 1e-06,
      "loss": 0.3471,
      "num_input_tokens_seen": 500772304,
      "step": 8938
    },
    {
      "epoch": 19.906458797327396,
      "loss": 0.26117777824401855,
      "loss_ce": 6.9360678025987e-05,
      "loss_iou": 0.11962890625,
      "loss_num": 0.004364013671875,
      "loss_xval": 0.26171875,
      "num_input_tokens_seen": 500772304,
      "step": 8938
    },
    {
      "epoch": 19.9086859688196,
      "grad_norm": 32.816749572753906,
      "learning_rate": 1e-06,
      "loss": 0.542,
      "num_input_tokens_seen": 500825560,
      "step": 8939
    },
    {
      "epoch": 19.9086859688196,
      "loss": 0.6478848457336426,
      "loss_ce": 5.768300252384506e-05,
      "loss_iou": 0.27734375,
      "loss_num": 0.0189208984375,
      "loss_xval": 0.6484375,
      "num_input_tokens_seen": 500825560,
      "step": 8939
    },
    {
      "epoch": 19.910913140311806,
      "grad_norm": 23.55243492126465,
      "learning_rate": 1e-06,
      "loss": 0.3671,
      "num_input_tokens_seen": 500881272,
      "step": 8940
    },
    {
      "epoch": 19.910913140311806,
      "loss": 0.28195321559906006,
      "loss_ce": 9.285648411605507e-05,
      "loss_iou": 0.1298828125,
      "loss_num": 0.004364013671875,
      "loss_xval": 0.28125,
      "num_input_tokens_seen": 500881272,
      "step": 8940
    },
    {
      "epoch": 19.91314031180401,
      "grad_norm": 19.66559600830078,
      "learning_rate": 1e-06,
      "loss": 0.49,
      "num_input_tokens_seen": 500937444,
      "step": 8941
    },
    {
      "epoch": 19.91314031180401,
      "loss": 0.621654748916626,
      "loss_ce": 7.26881917216815e-05,
      "loss_iou": 0.267578125,
      "loss_num": 0.0177001953125,
      "loss_xval": 0.62109375,
      "num_input_tokens_seen": 500937444,
      "step": 8941
    },
    {
      "epoch": 19.915367483296215,
      "grad_norm": 18.567909240722656,
      "learning_rate": 1e-06,
      "loss": 0.4054,
      "num_input_tokens_seen": 500995428,
      "step": 8942
    },
    {
      "epoch": 19.915367483296215,
      "loss": 0.36586493253707886,
      "loss_ce": 0.0001422582136001438,
      "loss_iou": 0.1630859375,
      "loss_num": 0.00799560546875,
      "loss_xval": 0.365234375,
      "num_input_tokens_seen": 500995428,
      "step": 8942
    },
    {
      "epoch": 19.91759465478842,
      "grad_norm": 37.899871826171875,
      "learning_rate": 1e-06,
      "loss": 0.3518,
      "num_input_tokens_seen": 501051864,
      "step": 8943
    },
    {
      "epoch": 19.91759465478842,
      "loss": 0.374164342880249,
      "loss_ce": 7.989482401171699e-05,
      "loss_iou": 0.1708984375,
      "loss_num": 0.00653076171875,
      "loss_xval": 0.375,
      "num_input_tokens_seen": 501051864,
      "step": 8943
    },
    {
      "epoch": 19.919821826280625,
      "grad_norm": 14.561594009399414,
      "learning_rate": 1e-06,
      "loss": 0.3799,
      "num_input_tokens_seen": 501107716,
      "step": 8944
    },
    {
      "epoch": 19.919821826280625,
      "loss": 0.3280653953552246,
      "loss_ce": 6.246700650081038e-05,
      "loss_iou": 0.146484375,
      "loss_num": 0.006866455078125,
      "loss_xval": 0.328125,
      "num_input_tokens_seen": 501107716,
      "step": 8944
    },
    {
      "epoch": 19.92204899777283,
      "grad_norm": 23.4508056640625,
      "learning_rate": 1e-06,
      "loss": 0.5593,
      "num_input_tokens_seen": 501162352,
      "step": 8945
    },
    {
      "epoch": 19.92204899777283,
      "loss": 0.4726704955101013,
      "loss_ce": 7.527978596044704e-05,
      "loss_iou": 0.1806640625,
      "loss_num": 0.02197265625,
      "loss_xval": 0.47265625,
      "num_input_tokens_seen": 501162352,
      "step": 8945
    },
    {
      "epoch": 19.924276169265035,
      "grad_norm": 17.324832916259766,
      "learning_rate": 1e-06,
      "loss": 0.3169,
      "num_input_tokens_seen": 501219728,
      "step": 8946
    },
    {
      "epoch": 19.924276169265035,
      "loss": 0.36799356341362,
      "loss_ce": 7.363170152530074e-05,
      "loss_iou": 0.1689453125,
      "loss_num": 0.00592041015625,
      "loss_xval": 0.3671875,
      "num_input_tokens_seen": 501219728,
      "step": 8946
    },
    {
      "epoch": 19.92650334075724,
      "grad_norm": 21.649171829223633,
      "learning_rate": 1e-06,
      "loss": 0.244,
      "num_input_tokens_seen": 501272792,
      "step": 8947
    },
    {
      "epoch": 19.92650334075724,
      "loss": 0.23100095987319946,
      "loss_ce": 0.00016598933143541217,
      "loss_iou": 0.103515625,
      "loss_num": 0.00482177734375,
      "loss_xval": 0.23046875,
      "num_input_tokens_seen": 501272792,
      "step": 8947
    },
    {
      "epoch": 19.928730512249444,
      "grad_norm": 14.748282432556152,
      "learning_rate": 1e-06,
      "loss": 0.326,
      "num_input_tokens_seen": 501330904,
      "step": 8948
    },
    {
      "epoch": 19.928730512249444,
      "loss": 0.3653227686882019,
      "loss_ce": 8.841823728289455e-05,
      "loss_iou": 0.1669921875,
      "loss_num": 0.006134033203125,
      "loss_xval": 0.365234375,
      "num_input_tokens_seen": 501330904,
      "step": 8948
    },
    {
      "epoch": 19.93095768374165,
      "grad_norm": 30.326812744140625,
      "learning_rate": 1e-06,
      "loss": 0.4558,
      "num_input_tokens_seen": 501387276,
      "step": 8949
    },
    {
      "epoch": 19.93095768374165,
      "loss": 0.4978707432746887,
      "loss_ce": 6.799891707487404e-05,
      "loss_iou": 0.17578125,
      "loss_num": 0.029052734375,
      "loss_xval": 0.498046875,
      "num_input_tokens_seen": 501387276,
      "step": 8949
    },
    {
      "epoch": 19.933184855233854,
      "grad_norm": 25.002771377563477,
      "learning_rate": 1e-06,
      "loss": 0.291,
      "num_input_tokens_seen": 501445204,
      "step": 8950
    },
    {
      "epoch": 19.933184855233854,
      "loss": 0.24606147408485413,
      "loss_ce": 5.927299935137853e-05,
      "loss_iou": 0.10595703125,
      "loss_num": 0.0068359375,
      "loss_xval": 0.24609375,
      "num_input_tokens_seen": 501445204,
      "step": 8950
    },
    {
      "epoch": 19.93541202672606,
      "grad_norm": 15.425994873046875,
      "learning_rate": 1e-06,
      "loss": 0.3445,
      "num_input_tokens_seen": 501501064,
      "step": 8951
    },
    {
      "epoch": 19.93541202672606,
      "loss": 0.29687991738319397,
      "loss_ce": 6.595243030460551e-05,
      "loss_iou": 0.1328125,
      "loss_num": 0.006072998046875,
      "loss_xval": 0.296875,
      "num_input_tokens_seen": 501501064,
      "step": 8951
    },
    {
      "epoch": 19.937639198218264,
      "grad_norm": 18.998098373413086,
      "learning_rate": 1e-06,
      "loss": 0.526,
      "num_input_tokens_seen": 501558712,
      "step": 8952
    },
    {
      "epoch": 19.937639198218264,
      "loss": 0.7495875358581543,
      "loss_ce": 7.577068754471838e-05,
      "loss_iou": 0.302734375,
      "loss_num": 0.028564453125,
      "loss_xval": 0.75,
      "num_input_tokens_seen": 501558712,
      "step": 8952
    },
    {
      "epoch": 19.93986636971047,
      "grad_norm": 13.607953071594238,
      "learning_rate": 1e-06,
      "loss": 0.2764,
      "num_input_tokens_seen": 501614280,
      "step": 8953
    },
    {
      "epoch": 19.93986636971047,
      "loss": 0.32002097368240356,
      "loss_ce": 7.468256808351725e-05,
      "loss_iou": 0.1416015625,
      "loss_num": 0.00750732421875,
      "loss_xval": 0.3203125,
      "num_input_tokens_seen": 501614280,
      "step": 8953
    },
    {
      "epoch": 19.942093541202674,
      "grad_norm": 17.92012596130371,
      "learning_rate": 1e-06,
      "loss": 0.5354,
      "num_input_tokens_seen": 501667972,
      "step": 8954
    },
    {
      "epoch": 19.942093541202674,
      "loss": 0.3914256989955902,
      "loss_ce": 6.828343612141907e-05,
      "loss_iou": 0.1767578125,
      "loss_num": 0.0078125,
      "loss_xval": 0.390625,
      "num_input_tokens_seen": 501667972,
      "step": 8954
    },
    {
      "epoch": 19.94432071269488,
      "grad_norm": 15.410776138305664,
      "learning_rate": 1e-06,
      "loss": 0.6504,
      "num_input_tokens_seen": 501724172,
      "step": 8955
    },
    {
      "epoch": 19.94432071269488,
      "loss": 0.7376750707626343,
      "loss_ce": 0.00012627900287043303,
      "loss_iou": 0.31640625,
      "loss_num": 0.0205078125,
      "loss_xval": 0.73828125,
      "num_input_tokens_seen": 501724172,
      "step": 8955
    },
    {
      "epoch": 19.946547884187083,
      "grad_norm": 14.882509231567383,
      "learning_rate": 1e-06,
      "loss": 0.2864,
      "num_input_tokens_seen": 501778848,
      "step": 8956
    },
    {
      "epoch": 19.946547884187083,
      "loss": 0.330025315284729,
      "loss_ce": 6.925308844074607e-05,
      "loss_iou": 0.1298828125,
      "loss_num": 0.0140380859375,
      "loss_xval": 0.330078125,
      "num_input_tokens_seen": 501778848,
      "step": 8956
    },
    {
      "epoch": 19.948775055679288,
      "grad_norm": 26.640323638916016,
      "learning_rate": 1e-06,
      "loss": 0.2967,
      "num_input_tokens_seen": 501833784,
      "step": 8957
    },
    {
      "epoch": 19.948775055679288,
      "loss": 0.23518048226833344,
      "loss_ce": 7.305956387426704e-05,
      "loss_iou": 0.0986328125,
      "loss_num": 0.00750732421875,
      "loss_xval": 0.2353515625,
      "num_input_tokens_seen": 501833784,
      "step": 8957
    },
    {
      "epoch": 19.951002227171493,
      "grad_norm": 23.93553352355957,
      "learning_rate": 1e-06,
      "loss": 0.4451,
      "num_input_tokens_seen": 501889872,
      "step": 8958
    },
    {
      "epoch": 19.951002227171493,
      "loss": 0.29492634534835815,
      "loss_ce": 6.549932732013986e-05,
      "loss_iou": 0.1328125,
      "loss_num": 0.006011962890625,
      "loss_xval": 0.294921875,
      "num_input_tokens_seen": 501889872,
      "step": 8958
    },
    {
      "epoch": 19.953229398663698,
      "grad_norm": 26.314388275146484,
      "learning_rate": 1e-06,
      "loss": 0.3367,
      "num_input_tokens_seen": 501946172,
      "step": 8959
    },
    {
      "epoch": 19.953229398663698,
      "loss": 0.3797050416469574,
      "loss_ce": 6.637527258135378e-05,
      "loss_iou": 0.17578125,
      "loss_num": 0.00543212890625,
      "loss_xval": 0.37890625,
      "num_input_tokens_seen": 501946172,
      "step": 8959
    },
    {
      "epoch": 19.955456570155903,
      "grad_norm": 12.816929817199707,
      "learning_rate": 1e-06,
      "loss": 0.4444,
      "num_input_tokens_seen": 502002296,
      "step": 8960
    },
    {
      "epoch": 19.955456570155903,
      "loss": 0.5041437149047852,
      "loss_ce": 6.582704372704029e-05,
      "loss_iou": 0.1904296875,
      "loss_num": 0.0244140625,
      "loss_xval": 0.50390625,
      "num_input_tokens_seen": 502002296,
      "step": 8960
    },
    {
      "epoch": 19.957683741648108,
      "grad_norm": 22.86423110961914,
      "learning_rate": 1e-06,
      "loss": 0.2911,
      "num_input_tokens_seen": 502056424,
      "step": 8961
    },
    {
      "epoch": 19.957683741648108,
      "loss": 0.2608652412891388,
      "loss_ce": 6.202118674991652e-05,
      "loss_iou": 0.1025390625,
      "loss_num": 0.01116943359375,
      "loss_xval": 0.26171875,
      "num_input_tokens_seen": 502056424,
      "step": 8961
    },
    {
      "epoch": 19.959910913140313,
      "grad_norm": 14.811808586120605,
      "learning_rate": 1e-06,
      "loss": 0.3608,
      "num_input_tokens_seen": 502112304,
      "step": 8962
    },
    {
      "epoch": 19.959910913140313,
      "loss": 0.4184660315513611,
      "loss_ce": 7.004072540439665e-05,
      "loss_iou": 0.1708984375,
      "loss_num": 0.01519775390625,
      "loss_xval": 0.41796875,
      "num_input_tokens_seen": 502112304,
      "step": 8962
    },
    {
      "epoch": 19.962138084632517,
      "grad_norm": 19.64674949645996,
      "learning_rate": 1e-06,
      "loss": 0.2859,
      "num_input_tokens_seen": 502167776,
      "step": 8963
    },
    {
      "epoch": 19.962138084632517,
      "loss": 0.2639721632003784,
      "loss_ce": 5.613587563857436e-05,
      "loss_iou": 0.10546875,
      "loss_num": 0.01055908203125,
      "loss_xval": 0.263671875,
      "num_input_tokens_seen": 502167776,
      "step": 8963
    },
    {
      "epoch": 19.964365256124722,
      "grad_norm": 14.231873512268066,
      "learning_rate": 1e-06,
      "loss": 0.3257,
      "num_input_tokens_seen": 502224072,
      "step": 8964
    },
    {
      "epoch": 19.964365256124722,
      "loss": 0.29962459206581116,
      "loss_ce": 6.405658496078104e-05,
      "loss_iou": 0.1328125,
      "loss_num": 0.00689697265625,
      "loss_xval": 0.298828125,
      "num_input_tokens_seen": 502224072,
      "step": 8964
    },
    {
      "epoch": 19.966592427616927,
      "grad_norm": 14.061171531677246,
      "learning_rate": 1e-06,
      "loss": 0.3948,
      "num_input_tokens_seen": 502281656,
      "step": 8965
    },
    {
      "epoch": 19.966592427616927,
      "loss": 0.36503365635871887,
      "loss_ce": 0.0002570544893387705,
      "loss_iou": 0.146484375,
      "loss_num": 0.01416015625,
      "loss_xval": 0.365234375,
      "num_input_tokens_seen": 502281656,
      "step": 8965
    },
    {
      "epoch": 19.968819599109132,
      "grad_norm": 39.687034606933594,
      "learning_rate": 1e-06,
      "loss": 0.2981,
      "num_input_tokens_seen": 502338012,
      "step": 8966
    },
    {
      "epoch": 19.968819599109132,
      "loss": 0.26977571845054626,
      "loss_ce": 6.136279262136668e-05,
      "loss_iou": 0.11865234375,
      "loss_num": 0.006622314453125,
      "loss_xval": 0.26953125,
      "num_input_tokens_seen": 502338012,
      "step": 8966
    },
    {
      "epoch": 19.971046770601337,
      "grad_norm": 17.423816680908203,
      "learning_rate": 1e-06,
      "loss": 0.3524,
      "num_input_tokens_seen": 502393216,
      "step": 8967
    },
    {
      "epoch": 19.971046770601337,
      "loss": 0.3833910822868347,
      "loss_ce": 0.00015138789603952318,
      "loss_iou": 0.16015625,
      "loss_num": 0.01263427734375,
      "loss_xval": 0.3828125,
      "num_input_tokens_seen": 502393216,
      "step": 8967
    },
    {
      "epoch": 19.97327394209354,
      "grad_norm": 15.628920555114746,
      "learning_rate": 1e-06,
      "loss": 0.6525,
      "num_input_tokens_seen": 502450296,
      "step": 8968
    },
    {
      "epoch": 19.97327394209354,
      "loss": 0.8214820027351379,
      "loss_ce": 7.082229421939701e-05,
      "loss_iou": 0.3046875,
      "loss_num": 0.04248046875,
      "loss_xval": 0.8203125,
      "num_input_tokens_seen": 502450296,
      "step": 8968
    },
    {
      "epoch": 19.975501113585747,
      "grad_norm": 19.731292724609375,
      "learning_rate": 1e-06,
      "loss": 0.3995,
      "num_input_tokens_seen": 502504404,
      "step": 8969
    },
    {
      "epoch": 19.975501113585747,
      "loss": 0.3412543833255768,
      "loss_ce": 6.785897130612284e-05,
      "loss_iou": 0.1435546875,
      "loss_num": 0.0107421875,
      "loss_xval": 0.341796875,
      "num_input_tokens_seen": 502504404,
      "step": 8969
    },
    {
      "epoch": 19.97772828507795,
      "grad_norm": 17.818803787231445,
      "learning_rate": 1e-06,
      "loss": 0.5062,
      "num_input_tokens_seen": 502561012,
      "step": 8970
    },
    {
      "epoch": 19.97772828507795,
      "loss": 0.46796897053718567,
      "loss_ce": 7.346458733081818e-05,
      "loss_iou": 0.1923828125,
      "loss_num": 0.0166015625,
      "loss_xval": 0.46875,
      "num_input_tokens_seen": 502561012,
      "step": 8970
    },
    {
      "epoch": 19.979955456570156,
      "grad_norm": 22.212177276611328,
      "learning_rate": 1e-06,
      "loss": 0.6205,
      "num_input_tokens_seen": 502615788,
      "step": 8971
    },
    {
      "epoch": 19.979955456570156,
      "loss": 0.680034875869751,
      "loss_ce": 0.00022534003073815256,
      "loss_iou": 0.28515625,
      "loss_num": 0.0220947265625,
      "loss_xval": 0.6796875,
      "num_input_tokens_seen": 502615788,
      "step": 8971
    },
    {
      "epoch": 19.98218262806236,
      "grad_norm": 28.786479949951172,
      "learning_rate": 1e-06,
      "loss": 0.4148,
      "num_input_tokens_seen": 502669668,
      "step": 8972
    },
    {
      "epoch": 19.98218262806236,
      "loss": 0.234740749001503,
      "loss_ce": 6.057398422854021e-05,
      "loss_iou": 0.1083984375,
      "loss_num": 0.0035400390625,
      "loss_xval": 0.234375,
      "num_input_tokens_seen": 502669668,
      "step": 8972
    },
    {
      "epoch": 19.984409799554566,
      "grad_norm": 22.56906509399414,
      "learning_rate": 1e-06,
      "loss": 0.3357,
      "num_input_tokens_seen": 502727112,
      "step": 8973
    },
    {
      "epoch": 19.984409799554566,
      "loss": 0.32478001713752747,
      "loss_ce": 7.298550917766988e-05,
      "loss_iou": 0.1435546875,
      "loss_num": 0.007415771484375,
      "loss_xval": 0.32421875,
      "num_input_tokens_seen": 502727112,
      "step": 8973
    },
    {
      "epoch": 19.98663697104677,
      "grad_norm": 12.906547546386719,
      "learning_rate": 1e-06,
      "loss": 0.6115,
      "num_input_tokens_seen": 502781448,
      "step": 8974
    },
    {
      "epoch": 19.98663697104677,
      "loss": 0.7082492113113403,
      "loss_ce": 5.82421307626646e-05,
      "loss_iou": 0.265625,
      "loss_num": 0.035400390625,
      "loss_xval": 0.70703125,
      "num_input_tokens_seen": 502781448,
      "step": 8974
    },
    {
      "epoch": 19.988864142538976,
      "grad_norm": 20.916061401367188,
      "learning_rate": 1e-06,
      "loss": 0.5555,
      "num_input_tokens_seen": 502834424,
      "step": 8975
    },
    {
      "epoch": 19.988864142538976,
      "loss": 0.562214195728302,
      "loss_ce": 8.040100510697812e-05,
      "loss_iou": 0.251953125,
      "loss_num": 0.01190185546875,
      "loss_xval": 0.5625,
      "num_input_tokens_seen": 502834424,
      "step": 8975
    },
    {
      "epoch": 19.99109131403118,
      "grad_norm": 27.778152465820312,
      "learning_rate": 1e-06,
      "loss": 0.5195,
      "num_input_tokens_seen": 502888060,
      "step": 8976
    },
    {
      "epoch": 19.99109131403118,
      "loss": 0.708865761756897,
      "loss_ce": 6.44870669930242e-05,
      "loss_iou": 0.3203125,
      "loss_num": 0.0135498046875,
      "loss_xval": 0.70703125,
      "num_input_tokens_seen": 502888060,
      "step": 8976
    },
    {
      "epoch": 19.993318485523385,
      "grad_norm": 16.12858772277832,
      "learning_rate": 1e-06,
      "loss": 0.3678,
      "num_input_tokens_seen": 502947760,
      "step": 8977
    },
    {
      "epoch": 19.993318485523385,
      "loss": 0.4400678873062134,
      "loss_ce": 6.544531788676977e-05,
      "loss_iou": 0.189453125,
      "loss_num": 0.01251220703125,
      "loss_xval": 0.439453125,
      "num_input_tokens_seen": 502947760,
      "step": 8977
    },
    {
      "epoch": 19.99554565701559,
      "grad_norm": 14.300686836242676,
      "learning_rate": 1e-06,
      "loss": 0.4194,
      "num_input_tokens_seen": 503003892,
      "step": 8978
    },
    {
      "epoch": 19.99554565701559,
      "loss": 0.554030179977417,
      "loss_ce": 7.511145668104291e-05,
      "loss_iou": 0.240234375,
      "loss_num": 0.0147705078125,
      "loss_xval": 0.5546875,
      "num_input_tokens_seen": 503003892,
      "step": 8978
    },
    {
      "epoch": 19.997772828507795,
      "grad_norm": 48.166603088378906,
      "learning_rate": 1e-06,
      "loss": 0.3181,
      "num_input_tokens_seen": 503060696,
      "step": 8979
    },
    {
      "epoch": 19.997772828507795,
      "loss": 0.36199867725372314,
      "loss_ce": 6.0221587773412466e-05,
      "loss_iou": 0.1533203125,
      "loss_num": 0.01116943359375,
      "loss_xval": 0.361328125,
      "num_input_tokens_seen": 503060696,
      "step": 8979
    },
    {
      "epoch": 20.0,
      "grad_norm": 19.31943130493164,
      "learning_rate": 1e-06,
      "loss": 0.3377,
      "num_input_tokens_seen": 503116608,
      "step": 8980
    },
    {
      "epoch": 20.0,
      "loss": 0.40692806243896484,
      "loss_ce": 6.770974141545594e-05,
      "loss_iou": 0.1796875,
      "loss_num": 0.00946044921875,
      "loss_xval": 0.40625,
      "num_input_tokens_seen": 503116608,
      "step": 8980
    },
    {
      "epoch": 20.002227171492205,
      "grad_norm": 19.004310607910156,
      "learning_rate": 1e-06,
      "loss": 0.3899,
      "num_input_tokens_seen": 503171700,
      "step": 8981
    },
    {
      "epoch": 20.002227171492205,
      "loss": 0.31977057456970215,
      "loss_ce": 6.842036236776039e-05,
      "loss_iou": 0.150390625,
      "loss_num": 0.0036163330078125,
      "loss_xval": 0.3203125,
      "num_input_tokens_seen": 503171700,
      "step": 8981
    },
    {
      "epoch": 20.00445434298441,
      "grad_norm": 16.16302490234375,
      "learning_rate": 1e-06,
      "loss": 0.3615,
      "num_input_tokens_seen": 503227328,
      "step": 8982
    },
    {
      "epoch": 20.00445434298441,
      "loss": 0.43536823987960815,
      "loss_ce": 6.547638622578233e-05,
      "loss_iou": 0.193359375,
      "loss_num": 0.00970458984375,
      "loss_xval": 0.435546875,
      "num_input_tokens_seen": 503227328,
      "step": 8982
    },
    {
      "epoch": 20.006681514476615,
      "grad_norm": 28.54294204711914,
      "learning_rate": 1e-06,
      "loss": 0.4652,
      "num_input_tokens_seen": 503280804,
      "step": 8983
    },
    {
      "epoch": 20.006681514476615,
      "loss": 0.3693348169326782,
      "loss_ce": 7.21029209671542e-05,
      "loss_iou": 0.16015625,
      "loss_num": 0.00982666015625,
      "loss_xval": 0.369140625,
      "num_input_tokens_seen": 503280804,
      "step": 8983
    },
    {
      "epoch": 20.00890868596882,
      "grad_norm": 13.636581420898438,
      "learning_rate": 1e-06,
      "loss": 0.3736,
      "num_input_tokens_seen": 503337904,
      "step": 8984
    },
    {
      "epoch": 20.00890868596882,
      "loss": 0.45746809244155884,
      "loss_ce": 7.06176069797948e-05,
      "loss_iou": 0.1962890625,
      "loss_num": 0.01312255859375,
      "loss_xval": 0.45703125,
      "num_input_tokens_seen": 503337904,
      "step": 8984
    },
    {
      "epoch": 20.011135857461024,
      "grad_norm": 23.064783096313477,
      "learning_rate": 1e-06,
      "loss": 0.4764,
      "num_input_tokens_seen": 503393424,
      "step": 8985
    },
    {
      "epoch": 20.011135857461024,
      "loss": 0.5955382585525513,
      "loss_ce": 7.930370338726789e-05,
      "loss_iou": 0.255859375,
      "loss_num": 0.0169677734375,
      "loss_xval": 0.59375,
      "num_input_tokens_seen": 503393424,
      "step": 8985
    },
    {
      "epoch": 20.01336302895323,
      "grad_norm": 20.41514015197754,
      "learning_rate": 1e-06,
      "loss": 0.3805,
      "num_input_tokens_seen": 503448520,
      "step": 8986
    },
    {
      "epoch": 20.01336302895323,
      "loss": 0.3952462673187256,
      "loss_ce": 0.00010465873492648825,
      "loss_iou": 0.1787109375,
      "loss_num": 0.007537841796875,
      "loss_xval": 0.39453125,
      "num_input_tokens_seen": 503448520,
      "step": 8986
    },
    {
      "epoch": 20.015590200445434,
      "grad_norm": 24.895496368408203,
      "learning_rate": 1e-06,
      "loss": 0.4096,
      "num_input_tokens_seen": 503504024,
      "step": 8987
    },
    {
      "epoch": 20.015590200445434,
      "loss": 0.43148207664489746,
      "loss_ce": 8.557141700293869e-05,
      "loss_iou": 0.189453125,
      "loss_num": 0.01031494140625,
      "loss_xval": 0.431640625,
      "num_input_tokens_seen": 503504024,
      "step": 8987
    },
    {
      "epoch": 20.01781737193764,
      "grad_norm": 23.964984893798828,
      "learning_rate": 1e-06,
      "loss": 0.3723,
      "num_input_tokens_seen": 503561476,
      "step": 8988
    },
    {
      "epoch": 20.01781737193764,
      "loss": 0.49469244480133057,
      "loss_ce": 6.352404307108372e-05,
      "loss_iou": 0.22265625,
      "loss_num": 0.010009765625,
      "loss_xval": 0.494140625,
      "num_input_tokens_seen": 503561476,
      "step": 8988
    },
    {
      "epoch": 20.020044543429844,
      "grad_norm": 23.97158432006836,
      "learning_rate": 1e-06,
      "loss": 0.6079,
      "num_input_tokens_seen": 503618636,
      "step": 8989
    },
    {
      "epoch": 20.020044543429844,
      "loss": 0.6602572202682495,
      "loss_ce": 0.00010094826575368643,
      "loss_iou": 0.255859375,
      "loss_num": 0.02978515625,
      "loss_xval": 0.66015625,
      "num_input_tokens_seen": 503618636,
      "step": 8989
    },
    {
      "epoch": 20.02227171492205,
      "grad_norm": 28.734207153320312,
      "learning_rate": 1e-06,
      "loss": 0.3533,
      "num_input_tokens_seen": 503674208,
      "step": 8990
    },
    {
      "epoch": 20.02227171492205,
      "loss": 0.34147197008132935,
      "loss_ce": 0.00019389839144423604,
      "loss_iou": 0.138671875,
      "loss_num": 0.01287841796875,
      "loss_xval": 0.341796875,
      "num_input_tokens_seen": 503674208,
      "step": 8990
    },
    {
      "epoch": 20.024498886414253,
      "grad_norm": 24.618515014648438,
      "learning_rate": 1e-06,
      "loss": 0.4778,
      "num_input_tokens_seen": 503728444,
      "step": 8991
    },
    {
      "epoch": 20.024498886414253,
      "loss": 0.24243876338005066,
      "loss_ce": 6.816858513047919e-05,
      "loss_iou": 0.10205078125,
      "loss_num": 0.00762939453125,
      "loss_xval": 0.2421875,
      "num_input_tokens_seen": 503728444,
      "step": 8991
    },
    {
      "epoch": 20.02672605790646,
      "grad_norm": 17.8887996673584,
      "learning_rate": 1e-06,
      "loss": 0.3836,
      "num_input_tokens_seen": 503784444,
      "step": 8992
    },
    {
      "epoch": 20.02672605790646,
      "loss": 0.4087654948234558,
      "loss_ce": 7.405409269267693e-05,
      "loss_iou": 0.177734375,
      "loss_num": 0.01043701171875,
      "loss_xval": 0.408203125,
      "num_input_tokens_seen": 503784444,
      "step": 8992
    },
    {
      "epoch": 20.028953229398663,
      "grad_norm": 19.861093521118164,
      "learning_rate": 1e-06,
      "loss": 0.5203,
      "num_input_tokens_seen": 503837876,
      "step": 8993
    },
    {
      "epoch": 20.028953229398663,
      "loss": 0.7142700552940369,
      "loss_ce": 6.721451791236177e-05,
      "loss_iou": 0.265625,
      "loss_num": 0.036865234375,
      "loss_xval": 0.71484375,
      "num_input_tokens_seen": 503837876,
      "step": 8993
    },
    {
      "epoch": 20.031180400890868,
      "grad_norm": 37.61943817138672,
      "learning_rate": 1e-06,
      "loss": 0.4567,
      "num_input_tokens_seen": 503894880,
      "step": 8994
    },
    {
      "epoch": 20.031180400890868,
      "loss": 0.4224729835987091,
      "loss_ce": 0.00010970777657348663,
      "loss_iou": 0.181640625,
      "loss_num": 0.01177978515625,
      "loss_xval": 0.421875,
      "num_input_tokens_seen": 503894880,
      "step": 8994
    },
    {
      "epoch": 20.033407572383073,
      "grad_norm": 21.20833969116211,
      "learning_rate": 1e-06,
      "loss": 0.4741,
      "num_input_tokens_seen": 503952424,
      "step": 8995
    },
    {
      "epoch": 20.033407572383073,
      "loss": 0.4196312725543976,
      "loss_ce": 7.561713573522866e-05,
      "loss_iou": 0.1640625,
      "loss_num": 0.0181884765625,
      "loss_xval": 0.419921875,
      "num_input_tokens_seen": 503952424,
      "step": 8995
    },
    {
      "epoch": 20.035634743875278,
      "grad_norm": 22.4307861328125,
      "learning_rate": 1e-06,
      "loss": 0.4745,
      "num_input_tokens_seen": 504008156,
      "step": 8996
    },
    {
      "epoch": 20.035634743875278,
      "loss": 0.4712446331977844,
      "loss_ce": 5.3244577429722995e-05,
      "loss_iou": 0.21875,
      "loss_num": 0.00677490234375,
      "loss_xval": 0.470703125,
      "num_input_tokens_seen": 504008156,
      "step": 8996
    },
    {
      "epoch": 20.037861915367483,
      "grad_norm": 35.790035247802734,
      "learning_rate": 1e-06,
      "loss": 0.3054,
      "num_input_tokens_seen": 504066160,
      "step": 8997
    },
    {
      "epoch": 20.037861915367483,
      "loss": 0.4005992114543915,
      "loss_ce": 8.653838449390605e-05,
      "loss_iou": 0.18359375,
      "loss_num": 0.006683349609375,
      "loss_xval": 0.400390625,
      "num_input_tokens_seen": 504066160,
      "step": 8997
    },
    {
      "epoch": 20.040089086859687,
      "grad_norm": 30.35886001586914,
      "learning_rate": 1e-06,
      "loss": 0.4581,
      "num_input_tokens_seen": 504123536,
      "step": 8998
    },
    {
      "epoch": 20.040089086859687,
      "loss": 0.4870605170726776,
      "loss_ce": 0.00012202416110085323,
      "loss_iou": 0.1865234375,
      "loss_num": 0.022705078125,
      "loss_xval": 0.486328125,
      "num_input_tokens_seen": 504123536,
      "step": 8998
    },
    {
      "epoch": 20.042316258351892,
      "grad_norm": 11.284721374511719,
      "learning_rate": 1e-06,
      "loss": 0.3595,
      "num_input_tokens_seen": 504179652,
      "step": 8999
    },
    {
      "epoch": 20.042316258351892,
      "loss": 0.37006866931915283,
      "loss_ce": 7.358190487138927e-05,
      "loss_iou": 0.158203125,
      "loss_num": 0.01092529296875,
      "loss_xval": 0.369140625,
      "num_input_tokens_seen": 504179652,
      "step": 8999
    },
    {
      "epoch": 20.044543429844097,
      "grad_norm": 14.649779319763184,
      "learning_rate": 1e-06,
      "loss": 0.4503,
      "num_input_tokens_seen": 504234948,
      "step": 9000
    },
    {
      "epoch": 20.044543429844097,
      "eval_seeclick_web_CIoU": 0.5833241045475006,
      "eval_seeclick_web_GIoU": 0.5827835202217102,
      "eval_seeclick_web_IoU": 0.602398693561554,
      "eval_seeclick_web_MAE_all": 0.015278099570423365,
      "eval_seeclick_web_MAE_h": 0.007092589279636741,
      "eval_seeclick_web_MAE_w": 0.015399211086332798,
      "eval_seeclick_web_MAE_x_boxes": 0.007994883926585317,
      "eval_seeclick_web_MAE_y_boxes": 0.021222305251285434,
      "eval_seeclick_web_inside_bbox": 0.9010416567325592,
      "eval_seeclick_web_loss": 0.9251190423965454,
      "eval_seeclick_web_loss_ce": 0.00011575160169741139,
      "eval_seeclick_web_loss_iou": 0.428955078125,
      "eval_seeclick_web_loss_num": 0.012242317199707031,
      "eval_seeclick_web_loss_xval": 0.9189453125,
      "eval_seeclick_web_runtime": 24.8776,
      "eval_seeclick_web_samples_per_second": 2.01,
      "eval_seeclick_web_steps_per_second": 0.08,
      "num_input_tokens_seen": 504234948,
      "step": 9000
    },
    {
      "epoch": 20.044543429844097,
      "eval_icons_CIoU": 0.26180562376976013,
      "eval_icons_GIoU": 0.2955218553543091,
      "eval_icons_IoU": 0.3424162417650223,
      "eval_icons_MAE_all": 0.057900117710232735,
      "eval_icons_MAE_h": 0.029276233166456223,
      "eval_icons_MAE_w": 0.06031078938394785,
      "eval_icons_MAE_x_boxes": 0.058521781116724014,
      "eval_icons_MAE_y_boxes": 0.03748843166977167,
      "eval_icons_inside_bbox": 0.59375,
      "eval_icons_loss": 1.7113921642303467,
      "eval_icons_loss_ce": 0.0001341397837677505,
      "eval_icons_loss_iou": 0.6754150390625,
      "eval_icons_loss_num": 0.057514190673828125,
      "eval_icons_loss_xval": 1.6396484375,
      "eval_icons_runtime": 24.3524,
      "eval_icons_samples_per_second": 2.053,
      "eval_icons_steps_per_second": 0.082,
      "num_input_tokens_seen": 504234948,
      "step": 9000
    },
    {
      "epoch": 20.044543429844097,
      "eval_screenspot_CIoU": 0.3642461995283763,
      "eval_screenspot_GIoU": 0.38147173325220746,
      "eval_screenspot_IoU": 0.4438290496667226,
      "eval_screenspot_MAE_all": 0.058075872560342155,
      "eval_screenspot_MAE_h": 0.039860475808382034,
      "eval_screenspot_MAE_w": 0.060762856155633926,
      "eval_screenspot_MAE_x_boxes": 0.07081380176047485,
      "eval_screenspot_MAE_y_boxes": 0.04531909463306268,
      "eval_screenspot_inside_bbox": 0.7041666706403097,
      "eval_screenspot_loss": 1.5968748331069946,
      "eval_screenspot_loss_ce": 0.00014579095780694237,
      "eval_screenspot_loss_iou": 0.6578776041666666,
      "eval_screenspot_loss_num": 0.06808090209960938,
      "eval_screenspot_loss_xval": 1.6564127604166667,
      "eval_screenspot_runtime": 41.1503,
      "eval_screenspot_samples_per_second": 2.163,
      "eval_screenspot_steps_per_second": 0.073,
      "num_input_tokens_seen": 504234948,
      "step": 9000
    },
    {
      "epoch": 20.044543429844097,
      "eval_compot_CIoU": 0.3458597809076309,
      "eval_compot_GIoU": 0.35780732333660126,
      "eval_compot_IoU": 0.40467947721481323,
      "eval_compot_MAE_all": 0.01937184017151594,
      "eval_compot_MAE_h": 0.01292520109564066,
      "eval_compot_MAE_w": 0.02040032297372818,
      "eval_compot_MAE_x_boxes": 0.030146288685500622,
      "eval_compot_MAE_y_boxes": 0.006630459800362587,
      "eval_compot_inside_bbox": 0.6302083432674408,
      "eval_compot_loss": 1.361356258392334,
      "eval_compot_loss_ce": 0.0001064865973603446,
      "eval_compot_loss_iou": 0.6224365234375,
      "eval_compot_loss_num": 0.017818450927734375,
      "eval_compot_loss_xval": 1.33349609375,
      "eval_compot_runtime": 25.482,
      "eval_compot_samples_per_second": 1.962,
      "eval_compot_steps_per_second": 0.078,
      "num_input_tokens_seen": 504234948,
      "step": 9000
    },
    {
      "epoch": 20.044543429844097,
      "eval_custom_ui_val_CIoU": 0.4760332836045159,
      "eval_custom_ui_val_GIoU": 0.48059673772917855,
      "eval_custom_ui_val_IoU": 0.5367237561278873,
      "eval_custom_ui_val_MAE_all": 0.026100213070296578,
      "eval_custom_ui_val_MAE_h": 0.013512549611429373,
      "eval_custom_ui_val_MAE_w": 0.036708953945587076,
      "eval_custom_ui_val_MAE_x_boxes": 0.03128902590833604,
      "eval_custom_ui_val_MAE_y_boxes": 0.012070265650335286,
      "eval_custom_ui_val_inside_bbox": 0.7719907429483202,
      "eval_custom_ui_val_loss": 1.1528477668762207,
      "eval_custom_ui_val_loss_ce": 0.00011129721032274474,
      "eval_custom_ui_val_loss_iou": 0.4967041015625,
      "eval_custom_ui_val_loss_num": 0.022805955674913194,
      "eval_custom_ui_val_loss_xval": 1.1072048611111112,
      "eval_custom_ui_val_runtime": 76.4297,
      "eval_custom_ui_val_samples_per_second": 3.467,
      "eval_custom_ui_val_steps_per_second": 0.118,
      "num_input_tokens_seen": 504234948,
      "step": 9000
    },
    {
      "epoch": 20.044543429844097,
      "loss": 0.8328539729118347,
      "loss_ce": 9.030340879689902e-05,
      "loss_iou": 0.375,
      "loss_num": 0.0166015625,
      "loss_xval": 0.83203125,
      "num_input_tokens_seen": 504234948,
      "step": 9000
    },
    {
      "epoch": 20.046770601336302,
      "grad_norm": 13.546182632446289,
      "learning_rate": 1e-06,
      "loss": 0.3243,
      "num_input_tokens_seen": 504290540,
      "step": 9001
    },
    {
      "epoch": 20.046770601336302,
      "loss": 0.26092007756233215,
      "loss_ce": 5.582950689131394e-05,
      "loss_iou": 0.1171875,
      "loss_num": 0.005340576171875,
      "loss_xval": 0.26171875,
      "num_input_tokens_seen": 504290540,
      "step": 9001
    },
    {
      "epoch": 20.048997772828507,
      "grad_norm": 11.733550071716309,
      "learning_rate": 1e-06,
      "loss": 0.3229,
      "num_input_tokens_seen": 504346220,
      "step": 9002
    },
    {
      "epoch": 20.048997772828507,
      "loss": 0.3032246530056,
      "loss_ce": 6.304614362306893e-05,
      "loss_iou": 0.1201171875,
      "loss_num": 0.01251220703125,
      "loss_xval": 0.302734375,
      "num_input_tokens_seen": 504346220,
      "step": 9002
    },
    {
      "epoch": 20.051224944320712,
      "grad_norm": 22.342601776123047,
      "learning_rate": 1e-06,
      "loss": 0.3593,
      "num_input_tokens_seen": 504402204,
      "step": 9003
    },
    {
      "epoch": 20.051224944320712,
      "loss": 0.26656997203826904,
      "loss_ce": 9.050035441759974e-05,
      "loss_iou": 0.119140625,
      "loss_num": 0.005706787109375,
      "loss_xval": 0.265625,
      "num_input_tokens_seen": 504402204,
      "step": 9003
    },
    {
      "epoch": 20.053452115812917,
      "grad_norm": 17.765623092651367,
      "learning_rate": 1e-06,
      "loss": 0.5938,
      "num_input_tokens_seen": 504458700,
      "step": 9004
    },
    {
      "epoch": 20.053452115812917,
      "loss": 0.5125864148139954,
      "loss_ce": 7.418861787300557e-05,
      "loss_iou": 0.2060546875,
      "loss_num": 0.0201416015625,
      "loss_xval": 0.51171875,
      "num_input_tokens_seen": 504458700,
      "step": 9004
    },
    {
      "epoch": 20.05567928730512,
      "grad_norm": 18.589075088500977,
      "learning_rate": 1e-06,
      "loss": 0.3828,
      "num_input_tokens_seen": 504514400,
      "step": 9005
    },
    {
      "epoch": 20.05567928730512,
      "loss": 0.3567635416984558,
      "loss_ce": 7.408991950796917e-05,
      "loss_iou": 0.14453125,
      "loss_num": 0.0135498046875,
      "loss_xval": 0.357421875,
      "num_input_tokens_seen": 504514400,
      "step": 9005
    },
    {
      "epoch": 20.057906458797326,
      "grad_norm": 19.300148010253906,
      "learning_rate": 1e-06,
      "loss": 0.5256,
      "num_input_tokens_seen": 504570592,
      "step": 9006
    },
    {
      "epoch": 20.057906458797326,
      "loss": 0.47822603583335876,
      "loss_ce": 7.661766721867025e-05,
      "loss_iou": 0.193359375,
      "loss_num": 0.0181884765625,
      "loss_xval": 0.478515625,
      "num_input_tokens_seen": 504570592,
      "step": 9006
    },
    {
      "epoch": 20.06013363028953,
      "grad_norm": 17.056005477905273,
      "learning_rate": 1e-06,
      "loss": 0.5382,
      "num_input_tokens_seen": 504623976,
      "step": 9007
    },
    {
      "epoch": 20.06013363028953,
      "loss": 0.6177802085876465,
      "loss_ce": 0.00010440404003020376,
      "loss_iou": 0.267578125,
      "loss_num": 0.0166015625,
      "loss_xval": 0.6171875,
      "num_input_tokens_seen": 504623976,
      "step": 9007
    },
    {
      "epoch": 20.062360801781736,
      "grad_norm": 12.326020240783691,
      "learning_rate": 1e-06,
      "loss": 0.2692,
      "num_input_tokens_seen": 504680516,
      "step": 9008
    },
    {
      "epoch": 20.062360801781736,
      "loss": 0.2412145882844925,
      "loss_ce": 6.467088678618893e-05,
      "loss_iou": 0.109375,
      "loss_num": 0.00439453125,
      "loss_xval": 0.2412109375,
      "num_input_tokens_seen": 504680516,
      "step": 9008
    },
    {
      "epoch": 20.06458797327394,
      "grad_norm": 15.348241806030273,
      "learning_rate": 1e-06,
      "loss": 0.2844,
      "num_input_tokens_seen": 504732616,
      "step": 9009
    },
    {
      "epoch": 20.06458797327394,
      "loss": 0.3364948332309723,
      "loss_ce": 6.905219197506085e-05,
      "loss_iou": 0.1328125,
      "loss_num": 0.01416015625,
      "loss_xval": 0.3359375,
      "num_input_tokens_seen": 504732616,
      "step": 9009
    },
    {
      "epoch": 20.066815144766146,
      "grad_norm": 274.50628662109375,
      "learning_rate": 1e-06,
      "loss": 0.3171,
      "num_input_tokens_seen": 504789792,
      "step": 9010
    },
    {
      "epoch": 20.066815144766146,
      "loss": 0.3943229615688324,
      "loss_ce": 6.636053149122745e-05,
      "loss_iou": 0.173828125,
      "loss_num": 0.00946044921875,
      "loss_xval": 0.39453125,
      "num_input_tokens_seen": 504789792,
      "step": 9010
    },
    {
      "epoch": 20.06904231625835,
      "grad_norm": 12.248230934143066,
      "learning_rate": 1e-06,
      "loss": 0.2955,
      "num_input_tokens_seen": 504845860,
      "step": 9011
    },
    {
      "epoch": 20.06904231625835,
      "loss": 0.2785142660140991,
      "loss_ce": 7.186854054452851e-05,
      "loss_iou": 0.10595703125,
      "loss_num": 0.0133056640625,
      "loss_xval": 0.279296875,
      "num_input_tokens_seen": 504845860,
      "step": 9011
    },
    {
      "epoch": 20.071269487750556,
      "grad_norm": 16.346586227416992,
      "learning_rate": 1e-06,
      "loss": 0.3333,
      "num_input_tokens_seen": 504897560,
      "step": 9012
    },
    {
      "epoch": 20.071269487750556,
      "loss": 0.25494903326034546,
      "loss_ce": 6.623686931561679e-05,
      "loss_iou": 0.10546875,
      "loss_num": 0.0086669921875,
      "loss_xval": 0.25390625,
      "num_input_tokens_seen": 504897560,
      "step": 9012
    },
    {
      "epoch": 20.07349665924276,
      "grad_norm": 14.807296752929688,
      "learning_rate": 1e-06,
      "loss": 0.2808,
      "num_input_tokens_seen": 504953152,
      "step": 9013
    },
    {
      "epoch": 20.07349665924276,
      "loss": 0.27936726808547974,
      "loss_ce": 7.039310003165156e-05,
      "loss_iou": 0.11865234375,
      "loss_num": 0.00836181640625,
      "loss_xval": 0.279296875,
      "num_input_tokens_seen": 504953152,
      "step": 9013
    },
    {
      "epoch": 20.075723830734965,
      "grad_norm": 17.577438354492188,
      "learning_rate": 1e-06,
      "loss": 0.3015,
      "num_input_tokens_seen": 505011868,
      "step": 9014
    },
    {
      "epoch": 20.075723830734965,
      "loss": 0.23732471466064453,
      "loss_ce": 5.0541042583063245e-05,
      "loss_iou": 0.10546875,
      "loss_num": 0.005279541015625,
      "loss_xval": 0.2373046875,
      "num_input_tokens_seen": 505011868,
      "step": 9014
    },
    {
      "epoch": 20.07795100222717,
      "grad_norm": 14.317474365234375,
      "learning_rate": 1e-06,
      "loss": 0.44,
      "num_input_tokens_seen": 505068340,
      "step": 9015
    },
    {
      "epoch": 20.07795100222717,
      "loss": 0.4671054482460022,
      "loss_ce": 6.44456158624962e-05,
      "loss_iou": 0.1884765625,
      "loss_num": 0.017822265625,
      "loss_xval": 0.466796875,
      "num_input_tokens_seen": 505068340,
      "step": 9015
    },
    {
      "epoch": 20.080178173719375,
      "grad_norm": 16.756528854370117,
      "learning_rate": 1e-06,
      "loss": 0.2364,
      "num_input_tokens_seen": 505126536,
      "step": 9016
    },
    {
      "epoch": 20.080178173719375,
      "loss": 0.22590871155261993,
      "loss_ce": 7.862491474952549e-05,
      "loss_iou": 0.10205078125,
      "loss_num": 0.004302978515625,
      "loss_xval": 0.2255859375,
      "num_input_tokens_seen": 505126536,
      "step": 9016
    },
    {
      "epoch": 20.08240534521158,
      "grad_norm": 13.3543062210083,
      "learning_rate": 1e-06,
      "loss": 0.3849,
      "num_input_tokens_seen": 505183652,
      "step": 9017
    },
    {
      "epoch": 20.08240534521158,
      "loss": 0.4101763963699341,
      "loss_ce": 8.116650133160874e-05,
      "loss_iou": 0.185546875,
      "loss_num": 0.007720947265625,
      "loss_xval": 0.41015625,
      "num_input_tokens_seen": 505183652,
      "step": 9017
    },
    {
      "epoch": 20.084632516703785,
      "grad_norm": 15.893471717834473,
      "learning_rate": 1e-06,
      "loss": 0.3159,
      "num_input_tokens_seen": 505238776,
      "step": 9018
    },
    {
      "epoch": 20.084632516703785,
      "loss": 0.3582761287689209,
      "loss_ce": 6.079444574424997e-05,
      "loss_iou": 0.162109375,
      "loss_num": 0.0067138671875,
      "loss_xval": 0.357421875,
      "num_input_tokens_seen": 505238776,
      "step": 9018
    },
    {
      "epoch": 20.08685968819599,
      "grad_norm": 26.983814239501953,
      "learning_rate": 1e-06,
      "loss": 0.502,
      "num_input_tokens_seen": 505295268,
      "step": 9019
    },
    {
      "epoch": 20.08685968819599,
      "loss": 0.48191970586776733,
      "loss_ce": 0.00010815928544616327,
      "loss_iou": 0.2158203125,
      "loss_num": 0.009765625,
      "loss_xval": 0.482421875,
      "num_input_tokens_seen": 505295268,
      "step": 9019
    },
    {
      "epoch": 20.089086859688194,
      "grad_norm": 13.190807342529297,
      "learning_rate": 1e-06,
      "loss": 0.327,
      "num_input_tokens_seen": 505351820,
      "step": 9020
    },
    {
      "epoch": 20.089086859688194,
      "loss": 0.3194577693939209,
      "loss_ce": 6.07890251558274e-05,
      "loss_iou": 0.1416015625,
      "loss_num": 0.007110595703125,
      "loss_xval": 0.3203125,
      "num_input_tokens_seen": 505351820,
      "step": 9020
    },
    {
      "epoch": 20.0913140311804,
      "grad_norm": 18.049283981323242,
      "learning_rate": 1e-06,
      "loss": 0.3301,
      "num_input_tokens_seen": 505409524,
      "step": 9021
    },
    {
      "epoch": 20.0913140311804,
      "loss": 0.3330759108066559,
      "loss_ce": 6.808601028751582e-05,
      "loss_iou": 0.1513671875,
      "loss_num": 0.006072998046875,
      "loss_xval": 0.33203125,
      "num_input_tokens_seen": 505409524,
      "step": 9021
    },
    {
      "epoch": 20.093541202672604,
      "grad_norm": 20.265382766723633,
      "learning_rate": 1e-06,
      "loss": 0.3425,
      "num_input_tokens_seen": 505466160,
      "step": 9022
    },
    {
      "epoch": 20.093541202672604,
      "loss": 0.2662951648235321,
      "loss_ce": 5.9811318351421505e-05,
      "loss_iou": 0.1171875,
      "loss_num": 0.006439208984375,
      "loss_xval": 0.265625,
      "num_input_tokens_seen": 505466160,
      "step": 9022
    },
    {
      "epoch": 20.09576837416481,
      "grad_norm": 18.630359649658203,
      "learning_rate": 1e-06,
      "loss": 0.459,
      "num_input_tokens_seen": 505521684,
      "step": 9023
    },
    {
      "epoch": 20.09576837416481,
      "loss": 0.34796595573425293,
      "loss_ce": 6.557940650964156e-05,
      "loss_iou": 0.1640625,
      "loss_num": 0.004180908203125,
      "loss_xval": 0.34765625,
      "num_input_tokens_seen": 505521684,
      "step": 9023
    },
    {
      "epoch": 20.097995545657014,
      "grad_norm": 18.471445083618164,
      "learning_rate": 1e-06,
      "loss": 0.44,
      "num_input_tokens_seen": 505579020,
      "step": 9024
    },
    {
      "epoch": 20.097995545657014,
      "loss": 0.4241413474082947,
      "loss_ce": 6.90808956278488e-05,
      "loss_iou": 0.185546875,
      "loss_num": 0.0107421875,
      "loss_xval": 0.423828125,
      "num_input_tokens_seen": 505579020,
      "step": 9024
    },
    {
      "epoch": 20.100222717149222,
      "grad_norm": 13.083913803100586,
      "learning_rate": 1e-06,
      "loss": 0.4995,
      "num_input_tokens_seen": 505636064,
      "step": 9025
    },
    {
      "epoch": 20.100222717149222,
      "loss": 0.3323401212692261,
      "loss_ce": 6.473006214946508e-05,
      "loss_iou": 0.15234375,
      "loss_num": 0.00537109375,
      "loss_xval": 0.33203125,
      "num_input_tokens_seen": 505636064,
      "step": 9025
    },
    {
      "epoch": 20.102449888641427,
      "grad_norm": 13.375629425048828,
      "learning_rate": 1e-06,
      "loss": 0.3271,
      "num_input_tokens_seen": 505694548,
      "step": 9026
    },
    {
      "epoch": 20.102449888641427,
      "loss": 0.261178195476532,
      "loss_ce": 6.979504541959614e-05,
      "loss_iou": 0.12255859375,
      "loss_num": 0.003265380859375,
      "loss_xval": 0.26171875,
      "num_input_tokens_seen": 505694548,
      "step": 9026
    },
    {
      "epoch": 20.104677060133632,
      "grad_norm": 15.511109352111816,
      "learning_rate": 1e-06,
      "loss": 0.4012,
      "num_input_tokens_seen": 505752036,
      "step": 9027
    },
    {
      "epoch": 20.104677060133632,
      "loss": 0.5342494249343872,
      "loss_ce": 6.970732647459954e-05,
      "loss_iou": 0.240234375,
      "loss_num": 0.0106201171875,
      "loss_xval": 0.53515625,
      "num_input_tokens_seen": 505752036,
      "step": 9027
    },
    {
      "epoch": 20.106904231625837,
      "grad_norm": 18.12278175354004,
      "learning_rate": 1e-06,
      "loss": 0.3609,
      "num_input_tokens_seen": 505806088,
      "step": 9028
    },
    {
      "epoch": 20.106904231625837,
      "loss": 0.21917679905891418,
      "loss_ce": 6.0579972341656685e-05,
      "loss_iou": 0.0859375,
      "loss_num": 0.0093994140625,
      "loss_xval": 0.21875,
      "num_input_tokens_seen": 505806088,
      "step": 9028
    },
    {
      "epoch": 20.10913140311804,
      "grad_norm": 19.46178436279297,
      "learning_rate": 1e-06,
      "loss": 0.3676,
      "num_input_tokens_seen": 505863524,
      "step": 9029
    },
    {
      "epoch": 20.10913140311804,
      "loss": 0.2694757878780365,
      "loss_ce": 6.661626684945077e-05,
      "loss_iou": 0.109375,
      "loss_num": 0.01007080078125,
      "loss_xval": 0.26953125,
      "num_input_tokens_seen": 505863524,
      "step": 9029
    },
    {
      "epoch": 20.111358574610247,
      "grad_norm": 25.316984176635742,
      "learning_rate": 1e-06,
      "loss": 0.3969,
      "num_input_tokens_seen": 505919056,
      "step": 9030
    },
    {
      "epoch": 20.111358574610247,
      "loss": 0.39313435554504395,
      "loss_ce": 6.794696673750877e-05,
      "loss_iou": 0.173828125,
      "loss_num": 0.00909423828125,
      "loss_xval": 0.392578125,
      "num_input_tokens_seen": 505919056,
      "step": 9030
    },
    {
      "epoch": 20.11358574610245,
      "grad_norm": 17.775178909301758,
      "learning_rate": 1e-06,
      "loss": 0.5433,
      "num_input_tokens_seen": 505973188,
      "step": 9031
    },
    {
      "epoch": 20.11358574610245,
      "loss": 0.5688655972480774,
      "loss_ce": 0.00014000045484863222,
      "loss_iou": 0.24609375,
      "loss_num": 0.0152587890625,
      "loss_xval": 0.5703125,
      "num_input_tokens_seen": 505973188,
      "step": 9031
    },
    {
      "epoch": 20.115812917594656,
      "grad_norm": 18.306743621826172,
      "learning_rate": 1e-06,
      "loss": 0.3148,
      "num_input_tokens_seen": 506026908,
      "step": 9032
    },
    {
      "epoch": 20.115812917594656,
      "loss": 0.25726398825645447,
      "loss_ce": 6.184067751746625e-05,
      "loss_iou": 0.107421875,
      "loss_num": 0.00848388671875,
      "loss_xval": 0.2578125,
      "num_input_tokens_seen": 506026908,
      "step": 9032
    },
    {
      "epoch": 20.11804008908686,
      "grad_norm": 18.657054901123047,
      "learning_rate": 1e-06,
      "loss": 0.4274,
      "num_input_tokens_seen": 506084560,
      "step": 9033
    },
    {
      "epoch": 20.11804008908686,
      "loss": 0.4233429729938507,
      "loss_ce": 6.415171810658649e-05,
      "loss_iou": 0.1865234375,
      "loss_num": 0.0101318359375,
      "loss_xval": 0.423828125,
      "num_input_tokens_seen": 506084560,
      "step": 9033
    },
    {
      "epoch": 20.120267260579066,
      "grad_norm": 20.036205291748047,
      "learning_rate": 1e-06,
      "loss": 0.3367,
      "num_input_tokens_seen": 506141540,
      "step": 9034
    },
    {
      "epoch": 20.120267260579066,
      "loss": 0.2808334231376648,
      "loss_ce": 7.170556636992842e-05,
      "loss_iou": 0.130859375,
      "loss_num": 0.0036773681640625,
      "loss_xval": 0.28125,
      "num_input_tokens_seen": 506141540,
      "step": 9034
    },
    {
      "epoch": 20.12249443207127,
      "grad_norm": 14.675041198730469,
      "learning_rate": 1e-06,
      "loss": 0.4575,
      "num_input_tokens_seen": 506195448,
      "step": 9035
    },
    {
      "epoch": 20.12249443207127,
      "loss": 0.43460237979888916,
      "loss_ce": 9.311149187851697e-05,
      "loss_iou": 0.2060546875,
      "loss_num": 0.00439453125,
      "loss_xval": 0.43359375,
      "num_input_tokens_seen": 506195448,
      "step": 9035
    },
    {
      "epoch": 20.124721603563476,
      "grad_norm": 22.29570770263672,
      "learning_rate": 1e-06,
      "loss": 0.4673,
      "num_input_tokens_seen": 506249464,
      "step": 9036
    },
    {
      "epoch": 20.124721603563476,
      "loss": 0.30451393127441406,
      "loss_ce": 7.058979826979339e-05,
      "loss_iou": 0.130859375,
      "loss_num": 0.008544921875,
      "loss_xval": 0.3046875,
      "num_input_tokens_seen": 506249464,
      "step": 9036
    },
    {
      "epoch": 20.12694877505568,
      "grad_norm": 18.38787841796875,
      "learning_rate": 1e-06,
      "loss": 0.3275,
      "num_input_tokens_seen": 506305972,
      "step": 9037
    },
    {
      "epoch": 20.12694877505568,
      "loss": 0.2641042470932007,
      "loss_ce": 6.616647442569956e-05,
      "loss_iou": 0.11328125,
      "loss_num": 0.007476806640625,
      "loss_xval": 0.263671875,
      "num_input_tokens_seen": 506305972,
      "step": 9037
    },
    {
      "epoch": 20.129175946547885,
      "grad_norm": 24.06353759765625,
      "learning_rate": 1e-06,
      "loss": 0.3628,
      "num_input_tokens_seen": 506363376,
      "step": 9038
    },
    {
      "epoch": 20.129175946547885,
      "loss": 0.3895137310028076,
      "loss_ce": 0.00010943982488242909,
      "loss_iou": 0.177734375,
      "loss_num": 0.006561279296875,
      "loss_xval": 0.388671875,
      "num_input_tokens_seen": 506363376,
      "step": 9038
    },
    {
      "epoch": 20.13140311804009,
      "grad_norm": 24.388277053833008,
      "learning_rate": 1e-06,
      "loss": 0.2923,
      "num_input_tokens_seen": 506419872,
      "step": 9039
    },
    {
      "epoch": 20.13140311804009,
      "loss": 0.33582985401153564,
      "loss_ce": 0.00013651110930368304,
      "loss_iou": 0.15234375,
      "loss_num": 0.00634765625,
      "loss_xval": 0.3359375,
      "num_input_tokens_seen": 506419872,
      "step": 9039
    },
    {
      "epoch": 20.133630289532295,
      "grad_norm": 12.003957748413086,
      "learning_rate": 1e-06,
      "loss": 0.3221,
      "num_input_tokens_seen": 506474320,
      "step": 9040
    },
    {
      "epoch": 20.133630289532295,
      "loss": 0.19839268922805786,
      "loss_ce": 5.894018613616936e-05,
      "loss_iou": 0.09130859375,
      "loss_num": 0.0030670166015625,
      "loss_xval": 0.1982421875,
      "num_input_tokens_seen": 506474320,
      "step": 9040
    },
    {
      "epoch": 20.1358574610245,
      "grad_norm": 20.20316505432129,
      "learning_rate": 1e-06,
      "loss": 0.4102,
      "num_input_tokens_seen": 506531432,
      "step": 9041
    },
    {
      "epoch": 20.1358574610245,
      "loss": 0.3831365704536438,
      "loss_ce": 6.466210470534861e-05,
      "loss_iou": 0.169921875,
      "loss_num": 0.008544921875,
      "loss_xval": 0.3828125,
      "num_input_tokens_seen": 506531432,
      "step": 9041
    },
    {
      "epoch": 20.138084632516705,
      "grad_norm": 19.908742904663086,
      "learning_rate": 1e-06,
      "loss": 0.4735,
      "num_input_tokens_seen": 506586268,
      "step": 9042
    },
    {
      "epoch": 20.138084632516705,
      "loss": 0.5597177743911743,
      "loss_ce": 0.0001474099699407816,
      "loss_iou": 0.24609375,
      "loss_num": 0.0135498046875,
      "loss_xval": 0.55859375,
      "num_input_tokens_seen": 506586268,
      "step": 9042
    },
    {
      "epoch": 20.14031180400891,
      "grad_norm": 15.677102088928223,
      "learning_rate": 1e-06,
      "loss": 0.3571,
      "num_input_tokens_seen": 506642036,
      "step": 9043
    },
    {
      "epoch": 20.14031180400891,
      "loss": 0.41627195477485657,
      "loss_ce": 7.322065357584506e-05,
      "loss_iou": 0.1884765625,
      "loss_num": 0.0078125,
      "loss_xval": 0.416015625,
      "num_input_tokens_seen": 506642036,
      "step": 9043
    },
    {
      "epoch": 20.142538975501115,
      "grad_norm": 21.689987182617188,
      "learning_rate": 1e-06,
      "loss": 0.4071,
      "num_input_tokens_seen": 506698484,
      "step": 9044
    },
    {
      "epoch": 20.142538975501115,
      "loss": 0.40728476643562317,
      "loss_ce": 5.823103856528178e-05,
      "loss_iou": 0.181640625,
      "loss_num": 0.008544921875,
      "loss_xval": 0.40625,
      "num_input_tokens_seen": 506698484,
      "step": 9044
    },
    {
      "epoch": 20.14476614699332,
      "grad_norm": 27.1544132232666,
      "learning_rate": 1e-06,
      "loss": 0.3355,
      "num_input_tokens_seen": 506752068,
      "step": 9045
    },
    {
      "epoch": 20.14476614699332,
      "loss": 0.3848227858543396,
      "loss_ce": 5.7126497267745435e-05,
      "loss_iou": 0.1591796875,
      "loss_num": 0.0133056640625,
      "loss_xval": 0.384765625,
      "num_input_tokens_seen": 506752068,
      "step": 9045
    },
    {
      "epoch": 20.146993318485524,
      "grad_norm": 15.306382179260254,
      "learning_rate": 1e-06,
      "loss": 0.4094,
      "num_input_tokens_seen": 506810520,
      "step": 9046
    },
    {
      "epoch": 20.146993318485524,
      "loss": 0.44698041677474976,
      "loss_ce": 8.10260244179517e-05,
      "loss_iou": 0.18359375,
      "loss_num": 0.0159912109375,
      "loss_xval": 0.447265625,
      "num_input_tokens_seen": 506810520,
      "step": 9046
    },
    {
      "epoch": 20.14922048997773,
      "grad_norm": 17.21749496459961,
      "learning_rate": 1e-06,
      "loss": 0.5436,
      "num_input_tokens_seen": 506865016,
      "step": 9047
    },
    {
      "epoch": 20.14922048997773,
      "loss": 0.689301609992981,
      "loss_ce": 9.259382932214066e-05,
      "loss_iou": 0.291015625,
      "loss_num": 0.0216064453125,
      "loss_xval": 0.6875,
      "num_input_tokens_seen": 506865016,
      "step": 9047
    },
    {
      "epoch": 20.151447661469934,
      "grad_norm": 28.808147430419922,
      "learning_rate": 1e-06,
      "loss": 0.477,
      "num_input_tokens_seen": 506918928,
      "step": 9048
    },
    {
      "epoch": 20.151447661469934,
      "loss": 0.46686720848083496,
      "loss_ce": 7.035446469672024e-05,
      "loss_iou": 0.19140625,
      "loss_num": 0.016845703125,
      "loss_xval": 0.466796875,
      "num_input_tokens_seen": 506918928,
      "step": 9048
    },
    {
      "epoch": 20.15367483296214,
      "grad_norm": 32.62353515625,
      "learning_rate": 1e-06,
      "loss": 0.3442,
      "num_input_tokens_seen": 506975208,
      "step": 9049
    },
    {
      "epoch": 20.15367483296214,
      "loss": 0.505931556224823,
      "loss_ce": 7.217634993139654e-05,
      "loss_iou": 0.2138671875,
      "loss_num": 0.0157470703125,
      "loss_xval": 0.5078125,
      "num_input_tokens_seen": 506975208,
      "step": 9049
    },
    {
      "epoch": 20.155902004454344,
      "grad_norm": 21.908447265625,
      "learning_rate": 1e-06,
      "loss": 0.4323,
      "num_input_tokens_seen": 507031872,
      "step": 9050
    },
    {
      "epoch": 20.155902004454344,
      "loss": 0.5479224324226379,
      "loss_ce": 7.087649282766506e-05,
      "loss_iou": 0.21875,
      "loss_num": 0.0220947265625,
      "loss_xval": 0.546875,
      "num_input_tokens_seen": 507031872,
      "step": 9050
    },
    {
      "epoch": 20.15812917594655,
      "grad_norm": 20.468429565429688,
      "learning_rate": 1e-06,
      "loss": 0.3044,
      "num_input_tokens_seen": 507089620,
      "step": 9051
    },
    {
      "epoch": 20.15812917594655,
      "loss": 0.3056070804595947,
      "loss_ce": 6.508109072456136e-05,
      "loss_iou": 0.142578125,
      "loss_num": 0.003997802734375,
      "loss_xval": 0.3046875,
      "num_input_tokens_seen": 507089620,
      "step": 9051
    },
    {
      "epoch": 20.160356347438753,
      "grad_norm": 22.60247802734375,
      "learning_rate": 1e-06,
      "loss": 0.3752,
      "num_input_tokens_seen": 507147488,
      "step": 9052
    },
    {
      "epoch": 20.160356347438753,
      "loss": 0.27081358432769775,
      "loss_ce": 6.161894270917401e-05,
      "loss_iou": 0.11962890625,
      "loss_num": 0.006317138671875,
      "loss_xval": 0.271484375,
      "num_input_tokens_seen": 507147488,
      "step": 9052
    },
    {
      "epoch": 20.16258351893096,
      "grad_norm": 27.35213851928711,
      "learning_rate": 1e-06,
      "loss": 0.6975,
      "num_input_tokens_seen": 507203384,
      "step": 9053
    },
    {
      "epoch": 20.16258351893096,
      "loss": 0.46700623631477356,
      "loss_ce": 8.729432738618925e-05,
      "loss_iou": 0.20703125,
      "loss_num": 0.0103759765625,
      "loss_xval": 0.466796875,
      "num_input_tokens_seen": 507203384,
      "step": 9053
    },
    {
      "epoch": 20.164810690423163,
      "grad_norm": 23.201065063476562,
      "learning_rate": 1e-06,
      "loss": 0.2998,
      "num_input_tokens_seen": 507256920,
      "step": 9054
    },
    {
      "epoch": 20.164810690423163,
      "loss": 0.1558593213558197,
      "loss_ce": 6.707788270432502e-05,
      "loss_iou": 0.06640625,
      "loss_num": 0.004638671875,
      "loss_xval": 0.15625,
      "num_input_tokens_seen": 507256920,
      "step": 9054
    },
    {
      "epoch": 20.167037861915368,
      "grad_norm": 17.185827255249023,
      "learning_rate": 1e-06,
      "loss": 0.2036,
      "num_input_tokens_seen": 507311420,
      "step": 9055
    },
    {
      "epoch": 20.167037861915368,
      "loss": 0.24561916291713715,
      "loss_ce": 7.47331214370206e-05,
      "loss_iou": 0.11474609375,
      "loss_num": 0.003173828125,
      "loss_xval": 0.2451171875,
      "num_input_tokens_seen": 507311420,
      "step": 9055
    },
    {
      "epoch": 20.169265033407573,
      "grad_norm": 12.79721450805664,
      "learning_rate": 1e-06,
      "loss": 0.3995,
      "num_input_tokens_seen": 507364720,
      "step": 9056
    },
    {
      "epoch": 20.169265033407573,
      "loss": 0.397408664226532,
      "loss_ce": 6.977266457397491e-05,
      "loss_iou": 0.1611328125,
      "loss_num": 0.014892578125,
      "loss_xval": 0.396484375,
      "num_input_tokens_seen": 507364720,
      "step": 9056
    },
    {
      "epoch": 20.171492204899778,
      "grad_norm": 21.816648483276367,
      "learning_rate": 1e-06,
      "loss": 0.4521,
      "num_input_tokens_seen": 507420248,
      "step": 9057
    },
    {
      "epoch": 20.171492204899778,
      "loss": 0.3353777527809143,
      "loss_ce": 6.583896174561232e-05,
      "loss_iou": 0.154296875,
      "loss_num": 0.005523681640625,
      "loss_xval": 0.3359375,
      "num_input_tokens_seen": 507420248,
      "step": 9057
    },
    {
      "epoch": 20.173719376391983,
      "grad_norm": 18.13716697692871,
      "learning_rate": 1e-06,
      "loss": 0.2414,
      "num_input_tokens_seen": 507477140,
      "step": 9058
    },
    {
      "epoch": 20.173719376391983,
      "loss": 0.22375068068504333,
      "loss_ce": 5.682064875145443e-05,
      "loss_iou": 0.09375,
      "loss_num": 0.0072021484375,
      "loss_xval": 0.2236328125,
      "num_input_tokens_seen": 507477140,
      "step": 9058
    },
    {
      "epoch": 20.175946547884188,
      "grad_norm": 15.739980697631836,
      "learning_rate": 1e-06,
      "loss": 0.3639,
      "num_input_tokens_seen": 507534024,
      "step": 9059
    },
    {
      "epoch": 20.175946547884188,
      "loss": 0.436111718416214,
      "loss_ce": 7.6580501627177e-05,
      "loss_iou": 0.201171875,
      "loss_num": 0.00665283203125,
      "loss_xval": 0.435546875,
      "num_input_tokens_seen": 507534024,
      "step": 9059
    },
    {
      "epoch": 20.178173719376392,
      "grad_norm": 25.104169845581055,
      "learning_rate": 1e-06,
      "loss": 0.5474,
      "num_input_tokens_seen": 507590760,
      "step": 9060
    },
    {
      "epoch": 20.178173719376392,
      "loss": 0.6292134523391724,
      "loss_ce": 0.00018516569980420172,
      "loss_iou": 0.2265625,
      "loss_num": 0.03515625,
      "loss_xval": 0.62890625,
      "num_input_tokens_seen": 507590760,
      "step": 9060
    },
    {
      "epoch": 20.180400890868597,
      "grad_norm": 14.639026641845703,
      "learning_rate": 1e-06,
      "loss": 0.3374,
      "num_input_tokens_seen": 507647528,
      "step": 9061
    },
    {
      "epoch": 20.180400890868597,
      "loss": 0.26568013429641724,
      "loss_ce": 5.5125230574049056e-05,
      "loss_iou": 0.1142578125,
      "loss_num": 0.007415771484375,
      "loss_xval": 0.265625,
      "num_input_tokens_seen": 507647528,
      "step": 9061
    },
    {
      "epoch": 20.182628062360802,
      "grad_norm": 25.42153549194336,
      "learning_rate": 1e-06,
      "loss": 0.4638,
      "num_input_tokens_seen": 507699664,
      "step": 9062
    },
    {
      "epoch": 20.182628062360802,
      "loss": 0.4531900882720947,
      "loss_ce": 6.505924829980358e-05,
      "loss_iou": 0.208984375,
      "loss_num": 0.007110595703125,
      "loss_xval": 0.453125,
      "num_input_tokens_seen": 507699664,
      "step": 9062
    },
    {
      "epoch": 20.184855233853007,
      "grad_norm": 20.597827911376953,
      "learning_rate": 1e-06,
      "loss": 0.2994,
      "num_input_tokens_seen": 507755092,
      "step": 9063
    },
    {
      "epoch": 20.184855233853007,
      "loss": 0.3056027889251709,
      "loss_ce": 6.079179729567841e-05,
      "loss_iou": 0.1328125,
      "loss_num": 0.0079345703125,
      "loss_xval": 0.3046875,
      "num_input_tokens_seen": 507755092,
      "step": 9063
    },
    {
      "epoch": 20.187082405345212,
      "grad_norm": 20.325477600097656,
      "learning_rate": 1e-06,
      "loss": 0.4699,
      "num_input_tokens_seen": 507810656,
      "step": 9064
    },
    {
      "epoch": 20.187082405345212,
      "loss": 0.5754441022872925,
      "loss_ce": 0.00012668846466112882,
      "loss_iou": 0.232421875,
      "loss_num": 0.0220947265625,
      "loss_xval": 0.57421875,
      "num_input_tokens_seen": 507810656,
      "step": 9064
    },
    {
      "epoch": 20.189309576837417,
      "grad_norm": 14.4949369430542,
      "learning_rate": 1e-06,
      "loss": 0.2323,
      "num_input_tokens_seen": 507868900,
      "step": 9065
    },
    {
      "epoch": 20.189309576837417,
      "loss": 0.29057085514068604,
      "loss_ce": 0.0002876527141779661,
      "loss_iou": 0.1201171875,
      "loss_num": 0.010009765625,
      "loss_xval": 0.291015625,
      "num_input_tokens_seen": 507868900,
      "step": 9065
    },
    {
      "epoch": 20.19153674832962,
      "grad_norm": 20.822650909423828,
      "learning_rate": 1e-06,
      "loss": 0.5267,
      "num_input_tokens_seen": 507927360,
      "step": 9066
    },
    {
      "epoch": 20.19153674832962,
      "loss": 0.4537545144557953,
      "loss_ce": 0.0001412302954122424,
      "loss_iou": 0.2109375,
      "loss_num": 0.00616455078125,
      "loss_xval": 0.453125,
      "num_input_tokens_seen": 507927360,
      "step": 9066
    },
    {
      "epoch": 20.193763919821826,
      "grad_norm": 25.914817810058594,
      "learning_rate": 1e-06,
      "loss": 0.4188,
      "num_input_tokens_seen": 507981276,
      "step": 9067
    },
    {
      "epoch": 20.193763919821826,
      "loss": 0.4693218767642975,
      "loss_ce": 8.358562627108768e-05,
      "loss_iou": 0.1845703125,
      "loss_num": 0.020263671875,
      "loss_xval": 0.46875,
      "num_input_tokens_seen": 507981276,
      "step": 9067
    },
    {
      "epoch": 20.19599109131403,
      "grad_norm": 21.08448028564453,
      "learning_rate": 1e-06,
      "loss": 0.2869,
      "num_input_tokens_seen": 508037936,
      "step": 9068
    },
    {
      "epoch": 20.19599109131403,
      "loss": 0.3300858736038208,
      "loss_ce": 6.879281136207283e-05,
      "loss_iou": 0.142578125,
      "loss_num": 0.00885009765625,
      "loss_xval": 0.330078125,
      "num_input_tokens_seen": 508037936,
      "step": 9068
    },
    {
      "epoch": 20.198218262806236,
      "grad_norm": 28.94850730895996,
      "learning_rate": 1e-06,
      "loss": 0.3374,
      "num_input_tokens_seen": 508094604,
      "step": 9069
    },
    {
      "epoch": 20.198218262806236,
      "loss": 0.311705619096756,
      "loss_ce": 6.010363722452894e-05,
      "loss_iou": 0.1416015625,
      "loss_num": 0.00555419921875,
      "loss_xval": 0.3125,
      "num_input_tokens_seen": 508094604,
      "step": 9069
    },
    {
      "epoch": 20.20044543429844,
      "grad_norm": 27.5219783782959,
      "learning_rate": 1e-06,
      "loss": 0.491,
      "num_input_tokens_seen": 508152580,
      "step": 9070
    },
    {
      "epoch": 20.20044543429844,
      "loss": 0.6100476384162903,
      "loss_ce": 6.228317215573043e-05,
      "loss_iou": 0.25390625,
      "loss_num": 0.02001953125,
      "loss_xval": 0.609375,
      "num_input_tokens_seen": 508152580,
      "step": 9070
    },
    {
      "epoch": 20.202672605790646,
      "grad_norm": 23.4427433013916,
      "learning_rate": 1e-06,
      "loss": 0.2314,
      "num_input_tokens_seen": 508206192,
      "step": 9071
    },
    {
      "epoch": 20.202672605790646,
      "loss": 0.2371232807636261,
      "loss_ce": 6.273827602853999e-05,
      "loss_iou": 0.10302734375,
      "loss_num": 0.006103515625,
      "loss_xval": 0.2373046875,
      "num_input_tokens_seen": 508206192,
      "step": 9071
    },
    {
      "epoch": 20.20489977728285,
      "grad_norm": 17.554397583007812,
      "learning_rate": 1e-06,
      "loss": 0.3968,
      "num_input_tokens_seen": 508262040,
      "step": 9072
    },
    {
      "epoch": 20.20489977728285,
      "loss": 0.4413501024246216,
      "loss_ce": 6.59393408568576e-05,
      "loss_iou": 0.19921875,
      "loss_num": 0.00872802734375,
      "loss_xval": 0.44140625,
      "num_input_tokens_seen": 508262040,
      "step": 9072
    },
    {
      "epoch": 20.207126948775056,
      "grad_norm": 15.516806602478027,
      "learning_rate": 1e-06,
      "loss": 0.3465,
      "num_input_tokens_seen": 508317992,
      "step": 9073
    },
    {
      "epoch": 20.207126948775056,
      "loss": 0.41605615615844727,
      "loss_ce": 0.0001626217272132635,
      "loss_iou": 0.1923828125,
      "loss_num": 0.00616455078125,
      "loss_xval": 0.416015625,
      "num_input_tokens_seen": 508317992,
      "step": 9073
    },
    {
      "epoch": 20.20935412026726,
      "grad_norm": 26.189729690551758,
      "learning_rate": 1e-06,
      "loss": 0.3646,
      "num_input_tokens_seen": 508373568,
      "step": 9074
    },
    {
      "epoch": 20.20935412026726,
      "loss": 0.3610216975212097,
      "loss_ce": 5.978911212878302e-05,
      "loss_iou": 0.15234375,
      "loss_num": 0.0111083984375,
      "loss_xval": 0.361328125,
      "num_input_tokens_seen": 508373568,
      "step": 9074
    },
    {
      "epoch": 20.211581291759465,
      "grad_norm": 16.33150863647461,
      "learning_rate": 1e-06,
      "loss": 0.4738,
      "num_input_tokens_seen": 508429440,
      "step": 9075
    },
    {
      "epoch": 20.211581291759465,
      "loss": 0.3367946445941925,
      "loss_ce": 6.368501635733992e-05,
      "loss_iou": 0.1337890625,
      "loss_num": 0.013916015625,
      "loss_xval": 0.3359375,
      "num_input_tokens_seen": 508429440,
      "step": 9075
    },
    {
      "epoch": 20.21380846325167,
      "grad_norm": 16.59500503540039,
      "learning_rate": 1e-06,
      "loss": 0.5425,
      "num_input_tokens_seen": 508486396,
      "step": 9076
    },
    {
      "epoch": 20.21380846325167,
      "loss": 0.4951022267341614,
      "loss_ce": 0.00010707708133850247,
      "loss_iou": 0.2158203125,
      "loss_num": 0.0125732421875,
      "loss_xval": 0.494140625,
      "num_input_tokens_seen": 508486396,
      "step": 9076
    },
    {
      "epoch": 20.216035634743875,
      "grad_norm": 16.87767791748047,
      "learning_rate": 1e-06,
      "loss": 0.6097,
      "num_input_tokens_seen": 508540776,
      "step": 9077
    },
    {
      "epoch": 20.216035634743875,
      "loss": 0.6077404022216797,
      "loss_ce": 7.441250636475161e-05,
      "loss_iou": 0.248046875,
      "loss_num": 0.0224609375,
      "loss_xval": 0.609375,
      "num_input_tokens_seen": 508540776,
      "step": 9077
    },
    {
      "epoch": 20.21826280623608,
      "grad_norm": 12.694762229919434,
      "learning_rate": 1e-06,
      "loss": 0.3561,
      "num_input_tokens_seen": 508597792,
      "step": 9078
    },
    {
      "epoch": 20.21826280623608,
      "loss": 0.22015753388404846,
      "loss_ce": 6.474574183812365e-05,
      "loss_iou": 0.099609375,
      "loss_num": 0.004119873046875,
      "loss_xval": 0.2197265625,
      "num_input_tokens_seen": 508597792,
      "step": 9078
    },
    {
      "epoch": 20.220489977728285,
      "grad_norm": 20.9544734954834,
      "learning_rate": 1e-06,
      "loss": 0.4849,
      "num_input_tokens_seen": 508655992,
      "step": 9079
    },
    {
      "epoch": 20.220489977728285,
      "loss": 0.7146784067153931,
      "loss_ce": 7.875564915593714e-05,
      "loss_iou": 0.3203125,
      "loss_num": 0.0147705078125,
      "loss_xval": 0.71484375,
      "num_input_tokens_seen": 508655992,
      "step": 9079
    },
    {
      "epoch": 20.22271714922049,
      "grad_norm": 9.648091316223145,
      "learning_rate": 1e-06,
      "loss": 0.401,
      "num_input_tokens_seen": 508711672,
      "step": 9080
    },
    {
      "epoch": 20.22271714922049,
      "loss": 0.2577533423900604,
      "loss_ce": 6.290937017183751e-05,
      "loss_iou": 0.1162109375,
      "loss_num": 0.005035400390625,
      "loss_xval": 0.2578125,
      "num_input_tokens_seen": 508711672,
      "step": 9080
    },
    {
      "epoch": 20.224944320712694,
      "grad_norm": 21.80451202392578,
      "learning_rate": 1e-06,
      "loss": 0.2682,
      "num_input_tokens_seen": 508769188,
      "step": 9081
    },
    {
      "epoch": 20.224944320712694,
      "loss": 0.2104550302028656,
      "loss_ce": 6.68463617330417e-05,
      "loss_iou": 0.087890625,
      "loss_num": 0.006866455078125,
      "loss_xval": 0.2099609375,
      "num_input_tokens_seen": 508769188,
      "step": 9081
    },
    {
      "epoch": 20.2271714922049,
      "grad_norm": 21.215343475341797,
      "learning_rate": 1e-06,
      "loss": 0.3385,
      "num_input_tokens_seen": 508824548,
      "step": 9082
    },
    {
      "epoch": 20.2271714922049,
      "loss": 0.4378107190132141,
      "loss_ce": 6.65808329358697e-05,
      "loss_iou": 0.1923828125,
      "loss_num": 0.01031494140625,
      "loss_xval": 0.4375,
      "num_input_tokens_seen": 508824548,
      "step": 9082
    },
    {
      "epoch": 20.229398663697104,
      "grad_norm": 17.058391571044922,
      "learning_rate": 1e-06,
      "loss": 0.2433,
      "num_input_tokens_seen": 508879452,
      "step": 9083
    },
    {
      "epoch": 20.229398663697104,
      "loss": 0.23303891718387604,
      "loss_ce": 6.771212792955339e-05,
      "loss_iou": 0.10302734375,
      "loss_num": 0.00543212890625,
      "loss_xval": 0.2333984375,
      "num_input_tokens_seen": 508879452,
      "step": 9083
    },
    {
      "epoch": 20.23162583518931,
      "grad_norm": 16.908109664916992,
      "learning_rate": 1e-06,
      "loss": 0.5078,
      "num_input_tokens_seen": 508935480,
      "step": 9084
    },
    {
      "epoch": 20.23162583518931,
      "loss": 0.29596656560897827,
      "loss_ce": 6.8145505792927e-05,
      "loss_iou": 0.1376953125,
      "loss_num": 0.00433349609375,
      "loss_xval": 0.296875,
      "num_input_tokens_seen": 508935480,
      "step": 9084
    },
    {
      "epoch": 20.233853006681514,
      "grad_norm": 20.34218978881836,
      "learning_rate": 1e-06,
      "loss": 0.3869,
      "num_input_tokens_seen": 508993000,
      "step": 9085
    },
    {
      "epoch": 20.233853006681514,
      "loss": 0.3470456302165985,
      "loss_ce": 6.076861609471962e-05,
      "loss_iou": 0.1513671875,
      "loss_num": 0.0089111328125,
      "loss_xval": 0.34765625,
      "num_input_tokens_seen": 508993000,
      "step": 9085
    },
    {
      "epoch": 20.23608017817372,
      "grad_norm": 23.703197479248047,
      "learning_rate": 1e-06,
      "loss": 0.3962,
      "num_input_tokens_seen": 509047736,
      "step": 9086
    },
    {
      "epoch": 20.23608017817372,
      "loss": 0.3868297338485718,
      "loss_ce": 0.00011096424714196473,
      "loss_iou": 0.1748046875,
      "loss_num": 0.007568359375,
      "loss_xval": 0.38671875,
      "num_input_tokens_seen": 509047736,
      "step": 9086
    },
    {
      "epoch": 20.238307349665924,
      "grad_norm": 17.429872512817383,
      "learning_rate": 1e-06,
      "loss": 0.3348,
      "num_input_tokens_seen": 509102768,
      "step": 9087
    },
    {
      "epoch": 20.238307349665924,
      "loss": 0.4313143193721771,
      "loss_ce": 0.00010093137825606391,
      "loss_iou": 0.1953125,
      "loss_num": 0.00823974609375,
      "loss_xval": 0.431640625,
      "num_input_tokens_seen": 509102768,
      "step": 9087
    },
    {
      "epoch": 20.24053452115813,
      "grad_norm": 24.075519561767578,
      "learning_rate": 1e-06,
      "loss": 0.3229,
      "num_input_tokens_seen": 509159036,
      "step": 9088
    },
    {
      "epoch": 20.24053452115813,
      "loss": 0.3190591335296631,
      "loss_ce": 5.887117004022002e-05,
      "loss_iou": 0.140625,
      "loss_num": 0.007568359375,
      "loss_xval": 0.318359375,
      "num_input_tokens_seen": 509159036,
      "step": 9088
    },
    {
      "epoch": 20.242761692650333,
      "grad_norm": 15.556453704833984,
      "learning_rate": 1e-06,
      "loss": 0.2766,
      "num_input_tokens_seen": 509215604,
      "step": 9089
    },
    {
      "epoch": 20.242761692650333,
      "loss": 0.2935373783111572,
      "loss_ce": 8.033675840124488e-05,
      "loss_iou": 0.1337890625,
      "loss_num": 0.005035400390625,
      "loss_xval": 0.29296875,
      "num_input_tokens_seen": 509215604,
      "step": 9089
    },
    {
      "epoch": 20.244988864142538,
      "grad_norm": 26.041152954101562,
      "learning_rate": 1e-06,
      "loss": 0.4012,
      "num_input_tokens_seen": 509271624,
      "step": 9090
    },
    {
      "epoch": 20.244988864142538,
      "loss": 0.28717702627182007,
      "loss_ce": 6.765717989765108e-05,
      "loss_iou": 0.1103515625,
      "loss_num": 0.0133056640625,
      "loss_xval": 0.287109375,
      "num_input_tokens_seen": 509271624,
      "step": 9090
    },
    {
      "epoch": 20.247216035634743,
      "grad_norm": 15.528565406799316,
      "learning_rate": 1e-06,
      "loss": 0.5706,
      "num_input_tokens_seen": 509327216,
      "step": 9091
    },
    {
      "epoch": 20.247216035634743,
      "loss": 0.7028372287750244,
      "loss_ce": 7.844123319955543e-05,
      "loss_iou": 0.30078125,
      "loss_num": 0.0203857421875,
      "loss_xval": 0.703125,
      "num_input_tokens_seen": 509327216,
      "step": 9091
    },
    {
      "epoch": 20.249443207126948,
      "grad_norm": 31.673660278320312,
      "learning_rate": 1e-06,
      "loss": 0.2802,
      "num_input_tokens_seen": 509384560,
      "step": 9092
    },
    {
      "epoch": 20.249443207126948,
      "loss": 0.2221912145614624,
      "loss_ce": 5.375898035708815e-05,
      "loss_iou": 0.09814453125,
      "loss_num": 0.005157470703125,
      "loss_xval": 0.2216796875,
      "num_input_tokens_seen": 509384560,
      "step": 9092
    },
    {
      "epoch": 20.251670378619153,
      "grad_norm": 17.960433959960938,
      "learning_rate": 1e-06,
      "loss": 0.2977,
      "num_input_tokens_seen": 509439108,
      "step": 9093
    },
    {
      "epoch": 20.251670378619153,
      "loss": 0.3416157364845276,
      "loss_ce": 6.299982487689704e-05,
      "loss_iou": 0.14453125,
      "loss_num": 0.010498046875,
      "loss_xval": 0.341796875,
      "num_input_tokens_seen": 509439108,
      "step": 9093
    },
    {
      "epoch": 20.253897550111358,
      "grad_norm": 15.661357879638672,
      "learning_rate": 1e-06,
      "loss": 0.3546,
      "num_input_tokens_seen": 509495464,
      "step": 9094
    },
    {
      "epoch": 20.253897550111358,
      "loss": 0.4557662904262543,
      "loss_ce": 7.779739098623395e-05,
      "loss_iou": 0.193359375,
      "loss_num": 0.01373291015625,
      "loss_xval": 0.455078125,
      "num_input_tokens_seen": 509495464,
      "step": 9094
    },
    {
      "epoch": 20.256124721603562,
      "grad_norm": 20.898950576782227,
      "learning_rate": 1e-06,
      "loss": 0.3629,
      "num_input_tokens_seen": 509552144,
      "step": 9095
    },
    {
      "epoch": 20.256124721603562,
      "loss": 0.3690887987613678,
      "loss_ce": 7.023525540716946e-05,
      "loss_iou": 0.1689453125,
      "loss_num": 0.006195068359375,
      "loss_xval": 0.369140625,
      "num_input_tokens_seen": 509552144,
      "step": 9095
    },
    {
      "epoch": 20.258351893095767,
      "grad_norm": 14.072615623474121,
      "learning_rate": 1e-06,
      "loss": 0.3833,
      "num_input_tokens_seen": 509607736,
      "step": 9096
    },
    {
      "epoch": 20.258351893095767,
      "loss": 0.41329166293144226,
      "loss_ce": 5.3150470193941146e-05,
      "loss_iou": 0.1845703125,
      "loss_num": 0.0087890625,
      "loss_xval": 0.4140625,
      "num_input_tokens_seen": 509607736,
      "step": 9096
    },
    {
      "epoch": 20.260579064587972,
      "grad_norm": 23.365612030029297,
      "learning_rate": 1e-06,
      "loss": 0.3804,
      "num_input_tokens_seen": 509662676,
      "step": 9097
    },
    {
      "epoch": 20.260579064587972,
      "loss": 0.42209792137145996,
      "loss_ce": 7.031915447441861e-05,
      "loss_iou": 0.158203125,
      "loss_num": 0.0211181640625,
      "loss_xval": 0.421875,
      "num_input_tokens_seen": 509662676,
      "step": 9097
    },
    {
      "epoch": 20.262806236080177,
      "grad_norm": 29.634624481201172,
      "learning_rate": 1e-06,
      "loss": 0.3842,
      "num_input_tokens_seen": 509720920,
      "step": 9098
    },
    {
      "epoch": 20.262806236080177,
      "loss": 0.4519707262516022,
      "loss_ce": 6.642582593485713e-05,
      "loss_iou": 0.1787109375,
      "loss_num": 0.01904296875,
      "loss_xval": 0.451171875,
      "num_input_tokens_seen": 509720920,
      "step": 9098
    },
    {
      "epoch": 20.265033407572382,
      "grad_norm": 15.21391773223877,
      "learning_rate": 1e-06,
      "loss": 0.4699,
      "num_input_tokens_seen": 509777720,
      "step": 9099
    },
    {
      "epoch": 20.265033407572382,
      "loss": 0.5803358554840088,
      "loss_ce": 7.468961121048778e-05,
      "loss_iou": 0.2431640625,
      "loss_num": 0.0185546875,
      "loss_xval": 0.58203125,
      "num_input_tokens_seen": 509777720,
      "step": 9099
    },
    {
      "epoch": 20.267260579064587,
      "grad_norm": 19.919464111328125,
      "learning_rate": 1e-06,
      "loss": 0.2832,
      "num_input_tokens_seen": 509834440,
      "step": 9100
    },
    {
      "epoch": 20.267260579064587,
      "loss": 0.2911492586135864,
      "loss_ce": 7.257152174133807e-05,
      "loss_iou": 0.12060546875,
      "loss_num": 0.01007080078125,
      "loss_xval": 0.291015625,
      "num_input_tokens_seen": 509834440,
      "step": 9100
    },
    {
      "epoch": 20.26948775055679,
      "grad_norm": 17.073530197143555,
      "learning_rate": 1e-06,
      "loss": 0.4515,
      "num_input_tokens_seen": 509891612,
      "step": 9101
    },
    {
      "epoch": 20.26948775055679,
      "loss": 0.4402480721473694,
      "loss_ce": 6.253210449358448e-05,
      "loss_iou": 0.169921875,
      "loss_num": 0.02001953125,
      "loss_xval": 0.439453125,
      "num_input_tokens_seen": 509891612,
      "step": 9101
    },
    {
      "epoch": 20.271714922048996,
      "grad_norm": 14.897684097290039,
      "learning_rate": 1e-06,
      "loss": 0.376,
      "num_input_tokens_seen": 509949728,
      "step": 9102
    },
    {
      "epoch": 20.271714922048996,
      "loss": 0.4393659234046936,
      "loss_ce": 6.541327456943691e-05,
      "loss_iou": 0.203125,
      "loss_num": 0.00640869140625,
      "loss_xval": 0.439453125,
      "num_input_tokens_seen": 509949728,
      "step": 9102
    },
    {
      "epoch": 20.2739420935412,
      "grad_norm": 17.08948516845703,
      "learning_rate": 1e-06,
      "loss": 0.2489,
      "num_input_tokens_seen": 510003660,
      "step": 9103
    },
    {
      "epoch": 20.2739420935412,
      "loss": 0.2572115659713745,
      "loss_ce": 7.0465452154167e-05,
      "loss_iou": 0.10009765625,
      "loss_num": 0.0113525390625,
      "loss_xval": 0.2578125,
      "num_input_tokens_seen": 510003660,
      "step": 9103
    },
    {
      "epoch": 20.276169265033406,
      "grad_norm": 19.830888748168945,
      "learning_rate": 1e-06,
      "loss": 0.3028,
      "num_input_tokens_seen": 510059112,
      "step": 9104
    },
    {
      "epoch": 20.276169265033406,
      "loss": 0.4022126793861389,
      "loss_ce": 0.00011306728993076831,
      "loss_iou": 0.1875,
      "loss_num": 0.0054931640625,
      "loss_xval": 0.40234375,
      "num_input_tokens_seen": 510059112,
      "step": 9104
    },
    {
      "epoch": 20.27839643652561,
      "grad_norm": 21.524456024169922,
      "learning_rate": 1e-06,
      "loss": 0.3969,
      "num_input_tokens_seen": 510113856,
      "step": 9105
    },
    {
      "epoch": 20.27839643652561,
      "loss": 0.2504650354385376,
      "loss_ce": 5.308254185365513e-05,
      "loss_iou": 0.10791015625,
      "loss_num": 0.006988525390625,
      "loss_xval": 0.25,
      "num_input_tokens_seen": 510113856,
      "step": 9105
    },
    {
      "epoch": 20.280623608017816,
      "grad_norm": 20.933208465576172,
      "learning_rate": 1e-06,
      "loss": 0.5396,
      "num_input_tokens_seen": 510171268,
      "step": 9106
    },
    {
      "epoch": 20.280623608017816,
      "loss": 0.566622257232666,
      "loss_ce": 9.393271466251463e-05,
      "loss_iou": 0.255859375,
      "loss_num": 0.0111083984375,
      "loss_xval": 0.56640625,
      "num_input_tokens_seen": 510171268,
      "step": 9106
    },
    {
      "epoch": 20.28285077951002,
      "grad_norm": 20.48546028137207,
      "learning_rate": 1e-06,
      "loss": 0.3572,
      "num_input_tokens_seen": 510224736,
      "step": 9107
    },
    {
      "epoch": 20.28285077951002,
      "loss": 0.33814841508865356,
      "loss_ce": 6.32288574706763e-05,
      "loss_iou": 0.146484375,
      "loss_num": 0.00897216796875,
      "loss_xval": 0.337890625,
      "num_input_tokens_seen": 510224736,
      "step": 9107
    },
    {
      "epoch": 20.285077951002226,
      "grad_norm": 11.746217727661133,
      "learning_rate": 1e-06,
      "loss": 0.3038,
      "num_input_tokens_seen": 510279584,
      "step": 9108
    },
    {
      "epoch": 20.285077951002226,
      "loss": 0.21575751900672913,
      "loss_ce": 5.92783690080978e-05,
      "loss_iou": 0.08740234375,
      "loss_num": 0.00811767578125,
      "loss_xval": 0.2158203125,
      "num_input_tokens_seen": 510279584,
      "step": 9108
    },
    {
      "epoch": 20.28730512249443,
      "grad_norm": 14.660599708557129,
      "learning_rate": 1e-06,
      "loss": 0.3486,
      "num_input_tokens_seen": 510337512,
      "step": 9109
    },
    {
      "epoch": 20.28730512249443,
      "loss": 0.3341745138168335,
      "loss_ce": 6.808016041759402e-05,
      "loss_iou": 0.150390625,
      "loss_num": 0.006591796875,
      "loss_xval": 0.333984375,
      "num_input_tokens_seen": 510337512,
      "step": 9109
    },
    {
      "epoch": 20.289532293986635,
      "grad_norm": 13.905699729919434,
      "learning_rate": 1e-06,
      "loss": 0.3323,
      "num_input_tokens_seen": 510393672,
      "step": 9110
    },
    {
      "epoch": 20.289532293986635,
      "loss": 0.4626014828681946,
      "loss_ce": 7.70436818129383e-05,
      "loss_iou": 0.1787109375,
      "loss_num": 0.02099609375,
      "loss_xval": 0.462890625,
      "num_input_tokens_seen": 510393672,
      "step": 9110
    },
    {
      "epoch": 20.29175946547884,
      "grad_norm": 22.228431701660156,
      "learning_rate": 1e-06,
      "loss": 0.3862,
      "num_input_tokens_seen": 510449288,
      "step": 9111
    },
    {
      "epoch": 20.29175946547884,
      "loss": 0.3664637804031372,
      "loss_ce": 6.975741416681558e-05,
      "loss_iou": 0.1630859375,
      "loss_num": 0.0081787109375,
      "loss_xval": 0.3671875,
      "num_input_tokens_seen": 510449288,
      "step": 9111
    },
    {
      "epoch": 20.293986636971045,
      "grad_norm": 21.240266799926758,
      "learning_rate": 1e-06,
      "loss": 0.323,
      "num_input_tokens_seen": 510504764,
      "step": 9112
    },
    {
      "epoch": 20.293986636971045,
      "loss": 0.31800100207328796,
      "loss_ce": 6.887991185067222e-05,
      "loss_iou": 0.1376953125,
      "loss_num": 0.0084228515625,
      "loss_xval": 0.318359375,
      "num_input_tokens_seen": 510504764,
      "step": 9112
    },
    {
      "epoch": 20.29621380846325,
      "grad_norm": 13.980697631835938,
      "learning_rate": 1e-06,
      "loss": 0.4736,
      "num_input_tokens_seen": 510560660,
      "step": 9113
    },
    {
      "epoch": 20.29621380846325,
      "loss": 0.4085160195827484,
      "loss_ce": 6.87609426677227e-05,
      "loss_iou": 0.181640625,
      "loss_num": 0.00909423828125,
      "loss_xval": 0.408203125,
      "num_input_tokens_seen": 510560660,
      "step": 9113
    },
    {
      "epoch": 20.29844097995546,
      "grad_norm": 13.33857536315918,
      "learning_rate": 1e-06,
      "loss": 0.3764,
      "num_input_tokens_seen": 510613008,
      "step": 9114
    },
    {
      "epoch": 20.29844097995546,
      "loss": 0.36590009927749634,
      "loss_ce": 8.587302727391943e-05,
      "loss_iou": 0.1552734375,
      "loss_num": 0.0111083984375,
      "loss_xval": 0.365234375,
      "num_input_tokens_seen": 510613008,
      "step": 9114
    },
    {
      "epoch": 20.30066815144766,
      "grad_norm": 18.67300033569336,
      "learning_rate": 1e-06,
      "loss": 0.2536,
      "num_input_tokens_seen": 510670700,
      "step": 9115
    },
    {
      "epoch": 20.30066815144766,
      "loss": 0.20209243893623352,
      "loss_ce": 6.605993257835507e-05,
      "loss_iou": 0.08935546875,
      "loss_num": 0.004608154296875,
      "loss_xval": 0.2021484375,
      "num_input_tokens_seen": 510670700,
      "step": 9115
    },
    {
      "epoch": 20.302895322939868,
      "grad_norm": 21.35281753540039,
      "learning_rate": 1e-06,
      "loss": 0.5114,
      "num_input_tokens_seen": 510721468,
      "step": 9116
    },
    {
      "epoch": 20.302895322939868,
      "loss": 0.388251394033432,
      "loss_ce": 6.779546674806625e-05,
      "loss_iou": 0.1728515625,
      "loss_num": 0.00848388671875,
      "loss_xval": 0.388671875,
      "num_input_tokens_seen": 510721468,
      "step": 9116
    },
    {
      "epoch": 20.305122494432073,
      "grad_norm": 17.087766647338867,
      "learning_rate": 1e-06,
      "loss": 0.5297,
      "num_input_tokens_seen": 510776316,
      "step": 9117
    },
    {
      "epoch": 20.305122494432073,
      "loss": 0.5188660621643066,
      "loss_ce": 6.725148705299944e-05,
      "loss_iou": 0.228515625,
      "loss_num": 0.01251220703125,
      "loss_xval": 0.51953125,
      "num_input_tokens_seen": 510776316,
      "step": 9117
    },
    {
      "epoch": 20.307349665924278,
      "grad_norm": 14.230613708496094,
      "learning_rate": 1e-06,
      "loss": 0.3324,
      "num_input_tokens_seen": 510833816,
      "step": 9118
    },
    {
      "epoch": 20.307349665924278,
      "loss": 0.2973046898841858,
      "loss_ce": 6.350380135700107e-05,
      "loss_iou": 0.138671875,
      "loss_num": 0.00390625,
      "loss_xval": 0.296875,
      "num_input_tokens_seen": 510833816,
      "step": 9118
    },
    {
      "epoch": 20.309576837416483,
      "grad_norm": 25.817169189453125,
      "learning_rate": 1e-06,
      "loss": 0.3926,
      "num_input_tokens_seen": 510890252,
      "step": 9119
    },
    {
      "epoch": 20.309576837416483,
      "loss": 0.3149741291999817,
      "loss_ce": 6.32612791378051e-05,
      "loss_iou": 0.1396484375,
      "loss_num": 0.007110595703125,
      "loss_xval": 0.314453125,
      "num_input_tokens_seen": 510890252,
      "step": 9119
    },
    {
      "epoch": 20.311804008908688,
      "grad_norm": 17.687280654907227,
      "learning_rate": 1e-06,
      "loss": 0.4531,
      "num_input_tokens_seen": 510946368,
      "step": 9120
    },
    {
      "epoch": 20.311804008908688,
      "loss": 0.5778282284736633,
      "loss_ce": 6.946115900063887e-05,
      "loss_iou": 0.234375,
      "loss_num": 0.02197265625,
      "loss_xval": 0.578125,
      "num_input_tokens_seen": 510946368,
      "step": 9120
    },
    {
      "epoch": 20.314031180400892,
      "grad_norm": 17.3331356048584,
      "learning_rate": 1e-06,
      "loss": 0.5977,
      "num_input_tokens_seen": 511002084,
      "step": 9121
    },
    {
      "epoch": 20.314031180400892,
      "loss": 0.4655322730541229,
      "loss_ce": 7.816165452823043e-05,
      "loss_iou": 0.201171875,
      "loss_num": 0.01275634765625,
      "loss_xval": 0.46484375,
      "num_input_tokens_seen": 511002084,
      "step": 9121
    },
    {
      "epoch": 20.316258351893097,
      "grad_norm": 28.032072067260742,
      "learning_rate": 1e-06,
      "loss": 0.3289,
      "num_input_tokens_seen": 511060036,
      "step": 9122
    },
    {
      "epoch": 20.316258351893097,
      "loss": 0.4448232352733612,
      "loss_ce": 6.0018668591510504e-05,
      "loss_iou": 0.19921875,
      "loss_num": 0.00933837890625,
      "loss_xval": 0.4453125,
      "num_input_tokens_seen": 511060036,
      "step": 9122
    },
    {
      "epoch": 20.318485523385302,
      "grad_norm": 31.768814086914062,
      "learning_rate": 1e-06,
      "loss": 0.407,
      "num_input_tokens_seen": 511115920,
      "step": 9123
    },
    {
      "epoch": 20.318485523385302,
      "loss": 0.32061654329299927,
      "loss_ce": 5.989862256683409e-05,
      "loss_iou": 0.1435546875,
      "loss_num": 0.0067138671875,
      "loss_xval": 0.3203125,
      "num_input_tokens_seen": 511115920,
      "step": 9123
    },
    {
      "epoch": 20.320712694877507,
      "grad_norm": 20.859983444213867,
      "learning_rate": 1e-06,
      "loss": 0.3574,
      "num_input_tokens_seen": 511173348,
      "step": 9124
    },
    {
      "epoch": 20.320712694877507,
      "loss": 0.36715149879455566,
      "loss_ce": 8.607155177742243e-05,
      "loss_iou": 0.15234375,
      "loss_num": 0.01239013671875,
      "loss_xval": 0.3671875,
      "num_input_tokens_seen": 511173348,
      "step": 9124
    },
    {
      "epoch": 20.322939866369712,
      "grad_norm": 14.776050567626953,
      "learning_rate": 1e-06,
      "loss": 0.4476,
      "num_input_tokens_seen": 511230864,
      "step": 9125
    },
    {
      "epoch": 20.322939866369712,
      "loss": 0.49763861298561096,
      "loss_ce": 6.477968418039382e-05,
      "loss_iou": 0.21875,
      "loss_num": 0.01220703125,
      "loss_xval": 0.498046875,
      "num_input_tokens_seen": 511230864,
      "step": 9125
    },
    {
      "epoch": 20.325167037861917,
      "grad_norm": 25.55546760559082,
      "learning_rate": 1e-06,
      "loss": 0.3811,
      "num_input_tokens_seen": 511285816,
      "step": 9126
    },
    {
      "epoch": 20.325167037861917,
      "loss": 0.32178765535354614,
      "loss_ce": 7.134034967748448e-05,
      "loss_iou": 0.130859375,
      "loss_num": 0.0118408203125,
      "loss_xval": 0.322265625,
      "num_input_tokens_seen": 511285816,
      "step": 9126
    },
    {
      "epoch": 20.32739420935412,
      "grad_norm": 21.569284439086914,
      "learning_rate": 1e-06,
      "loss": 0.2597,
      "num_input_tokens_seen": 511343148,
      "step": 9127
    },
    {
      "epoch": 20.32739420935412,
      "loss": 0.262149453163147,
      "loss_ce": 6.447016494348645e-05,
      "loss_iou": 0.11767578125,
      "loss_num": 0.00531005859375,
      "loss_xval": 0.26171875,
      "num_input_tokens_seen": 511343148,
      "step": 9127
    },
    {
      "epoch": 20.329621380846326,
      "grad_norm": 14.305514335632324,
      "learning_rate": 1e-06,
      "loss": 0.3907,
      "num_input_tokens_seen": 511400796,
      "step": 9128
    },
    {
      "epoch": 20.329621380846326,
      "loss": 0.32069259881973267,
      "loss_ce": 7.491336873499677e-05,
      "loss_iou": 0.142578125,
      "loss_num": 0.0072021484375,
      "loss_xval": 0.3203125,
      "num_input_tokens_seen": 511400796,
      "step": 9128
    },
    {
      "epoch": 20.33184855233853,
      "grad_norm": 15.63278865814209,
      "learning_rate": 1e-06,
      "loss": 0.2036,
      "num_input_tokens_seen": 511456720,
      "step": 9129
    },
    {
      "epoch": 20.33184855233853,
      "loss": 0.2301052063703537,
      "loss_ce": 6.369603215716779e-05,
      "loss_iou": 0.10693359375,
      "loss_num": 0.0032501220703125,
      "loss_xval": 0.23046875,
      "num_input_tokens_seen": 511456720,
      "step": 9129
    },
    {
      "epoch": 20.334075723830736,
      "grad_norm": 93.94779205322266,
      "learning_rate": 1e-06,
      "loss": 0.3359,
      "num_input_tokens_seen": 511513624,
      "step": 9130
    },
    {
      "epoch": 20.334075723830736,
      "loss": 0.24669964611530304,
      "loss_ce": 5.657020301441662e-05,
      "loss_iou": 0.11083984375,
      "loss_num": 0.00506591796875,
      "loss_xval": 0.2470703125,
      "num_input_tokens_seen": 511513624,
      "step": 9130
    },
    {
      "epoch": 20.33630289532294,
      "grad_norm": 33.43038558959961,
      "learning_rate": 1e-06,
      "loss": 0.3538,
      "num_input_tokens_seen": 511570664,
      "step": 9131
    },
    {
      "epoch": 20.33630289532294,
      "loss": 0.4840731620788574,
      "loss_ce": 0.00012543509365059435,
      "loss_iou": 0.21875,
      "loss_num": 0.00921630859375,
      "loss_xval": 0.484375,
      "num_input_tokens_seen": 511570664,
      "step": 9131
    },
    {
      "epoch": 20.338530066815146,
      "grad_norm": 11.925836563110352,
      "learning_rate": 1e-06,
      "loss": 0.2765,
      "num_input_tokens_seen": 511626512,
      "step": 9132
    },
    {
      "epoch": 20.338530066815146,
      "loss": 0.27660778164863586,
      "loss_ce": 5.7503631978761405e-05,
      "loss_iou": 0.12451171875,
      "loss_num": 0.00543212890625,
      "loss_xval": 0.27734375,
      "num_input_tokens_seen": 511626512,
      "step": 9132
    },
    {
      "epoch": 20.34075723830735,
      "grad_norm": 19.892284393310547,
      "learning_rate": 1e-06,
      "loss": 0.2753,
      "num_input_tokens_seen": 511684156,
      "step": 9133
    },
    {
      "epoch": 20.34075723830735,
      "loss": 0.2991905212402344,
      "loss_ce": 5.724587390432134e-05,
      "loss_iou": 0.13671875,
      "loss_num": 0.005279541015625,
      "loss_xval": 0.298828125,
      "num_input_tokens_seen": 511684156,
      "step": 9133
    },
    {
      "epoch": 20.342984409799556,
      "grad_norm": 14.584566116333008,
      "learning_rate": 1e-06,
      "loss": 0.3131,
      "num_input_tokens_seen": 511743004,
      "step": 9134
    },
    {
      "epoch": 20.342984409799556,
      "loss": 0.31488001346588135,
      "loss_ce": 6.067375943530351e-05,
      "loss_iou": 0.1435546875,
      "loss_num": 0.005615234375,
      "loss_xval": 0.314453125,
      "num_input_tokens_seen": 511743004,
      "step": 9134
    },
    {
      "epoch": 20.34521158129176,
      "grad_norm": 24.601272583007812,
      "learning_rate": 1e-06,
      "loss": 0.5077,
      "num_input_tokens_seen": 511794760,
      "step": 9135
    },
    {
      "epoch": 20.34521158129176,
      "loss": 0.6481292247772217,
      "loss_ce": 5.792171214125119e-05,
      "loss_iou": 0.263671875,
      "loss_num": 0.0238037109375,
      "loss_xval": 0.6484375,
      "num_input_tokens_seen": 511794760,
      "step": 9135
    },
    {
      "epoch": 20.347438752783965,
      "grad_norm": 13.779706954956055,
      "learning_rate": 1e-06,
      "loss": 0.4595,
      "num_input_tokens_seen": 511852632,
      "step": 9136
    },
    {
      "epoch": 20.347438752783965,
      "loss": 0.4468442499637604,
      "loss_ce": 6.692019815091044e-05,
      "loss_iou": 0.1884765625,
      "loss_num": 0.01409912109375,
      "loss_xval": 0.447265625,
      "num_input_tokens_seen": 511852632,
      "step": 9136
    },
    {
      "epoch": 20.34966592427617,
      "grad_norm": 17.18348503112793,
      "learning_rate": 1e-06,
      "loss": 0.3995,
      "num_input_tokens_seen": 511910288,
      "step": 9137
    },
    {
      "epoch": 20.34966592427617,
      "loss": 0.5219651460647583,
      "loss_ce": 0.00011460046516731381,
      "loss_iou": 0.1943359375,
      "loss_num": 0.026611328125,
      "loss_xval": 0.5234375,
      "num_input_tokens_seen": 511910288,
      "step": 9137
    },
    {
      "epoch": 20.351893095768375,
      "grad_norm": 16.120128631591797,
      "learning_rate": 1e-06,
      "loss": 0.434,
      "num_input_tokens_seen": 511965500,
      "step": 9138
    },
    {
      "epoch": 20.351893095768375,
      "loss": 0.4480791389942169,
      "loss_ce": 8.108362089842558e-05,
      "loss_iou": 0.1650390625,
      "loss_num": 0.0235595703125,
      "loss_xval": 0.447265625,
      "num_input_tokens_seen": 511965500,
      "step": 9138
    },
    {
      "epoch": 20.35412026726058,
      "grad_norm": 14.766048431396484,
      "learning_rate": 1e-06,
      "loss": 0.3227,
      "num_input_tokens_seen": 512020400,
      "step": 9139
    },
    {
      "epoch": 20.35412026726058,
      "loss": 0.17573511600494385,
      "loss_ce": 6.067653885111213e-05,
      "loss_iou": 0.07080078125,
      "loss_num": 0.006805419921875,
      "loss_xval": 0.17578125,
      "num_input_tokens_seen": 512020400,
      "step": 9139
    },
    {
      "epoch": 20.356347438752785,
      "grad_norm": 23.5654354095459,
      "learning_rate": 1e-06,
      "loss": 0.3828,
      "num_input_tokens_seen": 512079228,
      "step": 9140
    },
    {
      "epoch": 20.356347438752785,
      "loss": 0.4413471221923828,
      "loss_ce": 6.295171624515206e-05,
      "loss_iou": 0.19140625,
      "loss_num": 0.01177978515625,
      "loss_xval": 0.44140625,
      "num_input_tokens_seen": 512079228,
      "step": 9140
    },
    {
      "epoch": 20.35857461024499,
      "grad_norm": 17.19645118713379,
      "learning_rate": 1e-06,
      "loss": 0.4033,
      "num_input_tokens_seen": 512136932,
      "step": 9141
    },
    {
      "epoch": 20.35857461024499,
      "loss": 0.3532071113586426,
      "loss_ce": 5.770741336164065e-05,
      "loss_iou": 0.1435546875,
      "loss_num": 0.01312255859375,
      "loss_xval": 0.353515625,
      "num_input_tokens_seen": 512136932,
      "step": 9141
    },
    {
      "epoch": 20.360801781737194,
      "grad_norm": 15.467480659484863,
      "learning_rate": 1e-06,
      "loss": 0.3074,
      "num_input_tokens_seen": 512194392,
      "step": 9142
    },
    {
      "epoch": 20.360801781737194,
      "loss": 0.2540320158004761,
      "loss_ce": 0.00012579177564475685,
      "loss_iou": 0.10791015625,
      "loss_num": 0.007568359375,
      "loss_xval": 0.25390625,
      "num_input_tokens_seen": 512194392,
      "step": 9142
    },
    {
      "epoch": 20.3630289532294,
      "grad_norm": 21.110055923461914,
      "learning_rate": 1e-06,
      "loss": 0.4215,
      "num_input_tokens_seen": 512251156,
      "step": 9143
    },
    {
      "epoch": 20.3630289532294,
      "loss": 0.4504120349884033,
      "loss_ce": 0.00021672958973795176,
      "loss_iou": 0.2099609375,
      "loss_num": 0.00592041015625,
      "loss_xval": 0.44921875,
      "num_input_tokens_seen": 512251156,
      "step": 9143
    },
    {
      "epoch": 20.365256124721604,
      "grad_norm": 13.63264274597168,
      "learning_rate": 1e-06,
      "loss": 0.356,
      "num_input_tokens_seen": 512307892,
      "step": 9144
    },
    {
      "epoch": 20.365256124721604,
      "loss": 0.4380814731121063,
      "loss_ce": 6.268207653192803e-05,
      "loss_iou": 0.1953125,
      "loss_num": 0.00921630859375,
      "loss_xval": 0.4375,
      "num_input_tokens_seen": 512307892,
      "step": 9144
    },
    {
      "epoch": 20.36748329621381,
      "grad_norm": 37.342010498046875,
      "learning_rate": 1e-06,
      "loss": 0.3701,
      "num_input_tokens_seen": 512361460,
      "step": 9145
    },
    {
      "epoch": 20.36748329621381,
      "loss": 0.4177306294441223,
      "loss_ce": 6.703598774038255e-05,
      "loss_iou": 0.1748046875,
      "loss_num": 0.013671875,
      "loss_xval": 0.41796875,
      "num_input_tokens_seen": 512361460,
      "step": 9145
    },
    {
      "epoch": 20.369710467706014,
      "grad_norm": 17.764732360839844,
      "learning_rate": 1e-06,
      "loss": 0.3428,
      "num_input_tokens_seen": 512418536,
      "step": 9146
    },
    {
      "epoch": 20.369710467706014,
      "loss": 0.33477145433425903,
      "loss_ce": 5.4668213124386966e-05,
      "loss_iou": 0.150390625,
      "loss_num": 0.0068359375,
      "loss_xval": 0.333984375,
      "num_input_tokens_seen": 512418536,
      "step": 9146
    },
    {
      "epoch": 20.37193763919822,
      "grad_norm": 25.832767486572266,
      "learning_rate": 1e-06,
      "loss": 0.5058,
      "num_input_tokens_seen": 512473128,
      "step": 9147
    },
    {
      "epoch": 20.37193763919822,
      "loss": 0.4780276417732239,
      "loss_ce": 6.131519330665469e-05,
      "loss_iou": 0.185546875,
      "loss_num": 0.0213623046875,
      "loss_xval": 0.478515625,
      "num_input_tokens_seen": 512473128,
      "step": 9147
    },
    {
      "epoch": 20.374164810690424,
      "grad_norm": 21.50016212463379,
      "learning_rate": 1e-06,
      "loss": 0.4721,
      "num_input_tokens_seen": 512527572,
      "step": 9148
    },
    {
      "epoch": 20.374164810690424,
      "loss": 0.42840418219566345,
      "loss_ce": 5.943184805801138e-05,
      "loss_iou": 0.1845703125,
      "loss_num": 0.0115966796875,
      "loss_xval": 0.427734375,
      "num_input_tokens_seen": 512527572,
      "step": 9148
    },
    {
      "epoch": 20.37639198218263,
      "grad_norm": 13.383490562438965,
      "learning_rate": 1e-06,
      "loss": 0.3591,
      "num_input_tokens_seen": 512584280,
      "step": 9149
    },
    {
      "epoch": 20.37639198218263,
      "loss": 0.345126748085022,
      "loss_ce": 6.446812767535448e-05,
      "loss_iou": 0.15625,
      "loss_num": 0.006500244140625,
      "loss_xval": 0.345703125,
      "num_input_tokens_seen": 512584280,
      "step": 9149
    },
    {
      "epoch": 20.378619153674833,
      "grad_norm": 17.600801467895508,
      "learning_rate": 1e-06,
      "loss": 0.3535,
      "num_input_tokens_seen": 512640636,
      "step": 9150
    },
    {
      "epoch": 20.378619153674833,
      "loss": 0.45195910334587097,
      "loss_ce": 5.4796357289887965e-05,
      "loss_iou": 0.1806640625,
      "loss_num": 0.01806640625,
      "loss_xval": 0.451171875,
      "num_input_tokens_seen": 512640636,
      "step": 9150
    },
    {
      "epoch": 20.380846325167038,
      "grad_norm": 15.987733840942383,
      "learning_rate": 1e-06,
      "loss": 0.3294,
      "num_input_tokens_seen": 512697364,
      "step": 9151
    },
    {
      "epoch": 20.380846325167038,
      "loss": 0.42148369550704956,
      "loss_ce": 9.698521898826584e-05,
      "loss_iou": 0.18359375,
      "loss_num": 0.0107421875,
      "loss_xval": 0.421875,
      "num_input_tokens_seen": 512697364,
      "step": 9151
    },
    {
      "epoch": 20.383073496659243,
      "grad_norm": 17.254024505615234,
      "learning_rate": 1e-06,
      "loss": 0.4233,
      "num_input_tokens_seen": 512753732,
      "step": 9152
    },
    {
      "epoch": 20.383073496659243,
      "loss": 0.43444541096687317,
      "loss_ce": 0.00011922699195565656,
      "loss_iou": 0.2001953125,
      "loss_num": 0.00665283203125,
      "loss_xval": 0.43359375,
      "num_input_tokens_seen": 512753732,
      "step": 9152
    },
    {
      "epoch": 20.385300668151448,
      "grad_norm": 32.762237548828125,
      "learning_rate": 1e-06,
      "loss": 0.5156,
      "num_input_tokens_seen": 512808448,
      "step": 9153
    },
    {
      "epoch": 20.385300668151448,
      "loss": 0.7034420371055603,
      "loss_ce": 7.288139750016853e-05,
      "loss_iou": 0.283203125,
      "loss_num": 0.02734375,
      "loss_xval": 0.703125,
      "num_input_tokens_seen": 512808448,
      "step": 9153
    },
    {
      "epoch": 20.387527839643653,
      "grad_norm": 34.91286087036133,
      "learning_rate": 1e-06,
      "loss": 0.4527,
      "num_input_tokens_seen": 512864528,
      "step": 9154
    },
    {
      "epoch": 20.387527839643653,
      "loss": 0.48030704259872437,
      "loss_ce": 8.245596836786717e-05,
      "loss_iou": 0.181640625,
      "loss_num": 0.023193359375,
      "loss_xval": 0.48046875,
      "num_input_tokens_seen": 512864528,
      "step": 9154
    },
    {
      "epoch": 20.389755011135858,
      "grad_norm": 16.32798194885254,
      "learning_rate": 1e-06,
      "loss": 0.3353,
      "num_input_tokens_seen": 512921932,
      "step": 9155
    },
    {
      "epoch": 20.389755011135858,
      "loss": 0.3125593066215515,
      "loss_ce": 5.93056101934053e-05,
      "loss_iou": 0.1435546875,
      "loss_num": 0.005279541015625,
      "loss_xval": 0.3125,
      "num_input_tokens_seen": 512921932,
      "step": 9155
    },
    {
      "epoch": 20.391982182628063,
      "grad_norm": 22.385419845581055,
      "learning_rate": 1e-06,
      "loss": 0.4722,
      "num_input_tokens_seen": 512977824,
      "step": 9156
    },
    {
      "epoch": 20.391982182628063,
      "loss": 0.4710025191307068,
      "loss_ce": 5.522385981748812e-05,
      "loss_iou": 0.2109375,
      "loss_num": 0.00958251953125,
      "loss_xval": 0.470703125,
      "num_input_tokens_seen": 512977824,
      "step": 9156
    },
    {
      "epoch": 20.394209354120267,
      "grad_norm": 22.299945831298828,
      "learning_rate": 1e-06,
      "loss": 0.332,
      "num_input_tokens_seen": 513036572,
      "step": 9157
    },
    {
      "epoch": 20.394209354120267,
      "loss": 0.27536553144454956,
      "loss_ce": 6.644198583671823e-05,
      "loss_iou": 0.1259765625,
      "loss_num": 0.00482177734375,
      "loss_xval": 0.275390625,
      "num_input_tokens_seen": 513036572,
      "step": 9157
    },
    {
      "epoch": 20.396436525612472,
      "grad_norm": 18.22890853881836,
      "learning_rate": 1e-06,
      "loss": 0.5649,
      "num_input_tokens_seen": 513092372,
      "step": 9158
    },
    {
      "epoch": 20.396436525612472,
      "loss": 0.5173371434211731,
      "loss_ce": 6.420467980206013e-05,
      "loss_iou": 0.21875,
      "loss_num": 0.0162353515625,
      "loss_xval": 0.515625,
      "num_input_tokens_seen": 513092372,
      "step": 9158
    },
    {
      "epoch": 20.398663697104677,
      "grad_norm": 13.26450252532959,
      "learning_rate": 1e-06,
      "loss": 0.2702,
      "num_input_tokens_seen": 513146088,
      "step": 9159
    },
    {
      "epoch": 20.398663697104677,
      "loss": 0.24542546272277832,
      "loss_ce": 6.411702634068206e-05,
      "loss_iou": 0.111328125,
      "loss_num": 0.004547119140625,
      "loss_xval": 0.2451171875,
      "num_input_tokens_seen": 513146088,
      "step": 9159
    },
    {
      "epoch": 20.400890868596882,
      "grad_norm": 13.41020393371582,
      "learning_rate": 1e-06,
      "loss": 0.2748,
      "num_input_tokens_seen": 513201040,
      "step": 9160
    },
    {
      "epoch": 20.400890868596882,
      "loss": 0.32405906915664673,
      "loss_ce": 8.446039282716811e-05,
      "loss_iou": 0.1435546875,
      "loss_num": 0.007537841796875,
      "loss_xval": 0.32421875,
      "num_input_tokens_seen": 513201040,
      "step": 9160
    },
    {
      "epoch": 20.403118040089087,
      "grad_norm": 26.860549926757812,
      "learning_rate": 1e-06,
      "loss": 0.4289,
      "num_input_tokens_seen": 513259616,
      "step": 9161
    },
    {
      "epoch": 20.403118040089087,
      "loss": 0.5175533294677734,
      "loss_ce": 9.72962225205265e-05,
      "loss_iou": 0.21484375,
      "loss_num": 0.017822265625,
      "loss_xval": 0.515625,
      "num_input_tokens_seen": 513259616,
      "step": 9161
    },
    {
      "epoch": 20.40534521158129,
      "grad_norm": 16.444419860839844,
      "learning_rate": 1e-06,
      "loss": 0.3839,
      "num_input_tokens_seen": 513315256,
      "step": 9162
    },
    {
      "epoch": 20.40534521158129,
      "loss": 0.21068307757377625,
      "loss_ce": 5.074591172160581e-05,
      "loss_iou": 0.0888671875,
      "loss_num": 0.006622314453125,
      "loss_xval": 0.2109375,
      "num_input_tokens_seen": 513315256,
      "step": 9162
    },
    {
      "epoch": 20.407572383073497,
      "grad_norm": 118.25879669189453,
      "learning_rate": 1e-06,
      "loss": 0.3403,
      "num_input_tokens_seen": 513371288,
      "step": 9163
    },
    {
      "epoch": 20.407572383073497,
      "loss": 0.3745216727256775,
      "loss_ce": 7.099170761648566e-05,
      "loss_iou": 0.1513671875,
      "loss_num": 0.0145263671875,
      "loss_xval": 0.375,
      "num_input_tokens_seen": 513371288,
      "step": 9163
    },
    {
      "epoch": 20.4097995545657,
      "grad_norm": 23.073387145996094,
      "learning_rate": 1e-06,
      "loss": 0.298,
      "num_input_tokens_seen": 513429104,
      "step": 9164
    },
    {
      "epoch": 20.4097995545657,
      "loss": 0.2354094237089157,
      "loss_ce": 8.837871428113431e-05,
      "loss_iou": 0.1064453125,
      "loss_num": 0.004364013671875,
      "loss_xval": 0.2353515625,
      "num_input_tokens_seen": 513429104,
      "step": 9164
    },
    {
      "epoch": 20.412026726057906,
      "grad_norm": 16.92924690246582,
      "learning_rate": 1e-06,
      "loss": 0.4139,
      "num_input_tokens_seen": 513486220,
      "step": 9165
    },
    {
      "epoch": 20.412026726057906,
      "loss": 0.30585503578186035,
      "loss_ce": 6.89055013936013e-05,
      "loss_iou": 0.1337890625,
      "loss_num": 0.007476806640625,
      "loss_xval": 0.306640625,
      "num_input_tokens_seen": 513486220,
      "step": 9165
    },
    {
      "epoch": 20.41425389755011,
      "grad_norm": 19.239347457885742,
      "learning_rate": 1e-06,
      "loss": 0.2636,
      "num_input_tokens_seen": 513540880,
      "step": 9166
    },
    {
      "epoch": 20.41425389755011,
      "loss": 0.24502648413181305,
      "loss_ce": 6.187942926771939e-05,
      "loss_iou": 0.103515625,
      "loss_num": 0.007598876953125,
      "loss_xval": 0.2451171875,
      "num_input_tokens_seen": 513540880,
      "step": 9166
    },
    {
      "epoch": 20.416481069042316,
      "grad_norm": 18.908649444580078,
      "learning_rate": 1e-06,
      "loss": 0.2436,
      "num_input_tokens_seen": 513596724,
      "step": 9167
    },
    {
      "epoch": 20.416481069042316,
      "loss": 0.274596244096756,
      "loss_ce": 6.010765355313197e-05,
      "loss_iou": 0.12060546875,
      "loss_num": 0.0067138671875,
      "loss_xval": 0.275390625,
      "num_input_tokens_seen": 513596724,
      "step": 9167
    },
    {
      "epoch": 20.41870824053452,
      "grad_norm": 15.285927772521973,
      "learning_rate": 1e-06,
      "loss": 0.3752,
      "num_input_tokens_seen": 513652904,
      "step": 9168
    },
    {
      "epoch": 20.41870824053452,
      "loss": 0.34748250246047974,
      "loss_ce": 7.040443597361445e-05,
      "loss_iou": 0.1513671875,
      "loss_num": 0.0089111328125,
      "loss_xval": 0.34765625,
      "num_input_tokens_seen": 513652904,
      "step": 9168
    },
    {
      "epoch": 20.420935412026726,
      "grad_norm": 20.112258911132812,
      "learning_rate": 1e-06,
      "loss": 0.3643,
      "num_input_tokens_seen": 513708628,
      "step": 9169
    },
    {
      "epoch": 20.420935412026726,
      "loss": 0.43531090021133423,
      "loss_ce": 6.918917642906308e-05,
      "loss_iou": 0.2021484375,
      "loss_num": 0.00628662109375,
      "loss_xval": 0.435546875,
      "num_input_tokens_seen": 513708628,
      "step": 9169
    },
    {
      "epoch": 20.42316258351893,
      "grad_norm": 12.775040626525879,
      "learning_rate": 1e-06,
      "loss": 0.357,
      "num_input_tokens_seen": 513766376,
      "step": 9170
    },
    {
      "epoch": 20.42316258351893,
      "loss": 0.4500158429145813,
      "loss_ce": 6.470449443440884e-05,
      "loss_iou": 0.1591796875,
      "loss_num": 0.0263671875,
      "loss_xval": 0.44921875,
      "num_input_tokens_seen": 513766376,
      "step": 9170
    },
    {
      "epoch": 20.425389755011135,
      "grad_norm": 22.044282913208008,
      "learning_rate": 1e-06,
      "loss": 0.413,
      "num_input_tokens_seen": 513821636,
      "step": 9171
    },
    {
      "epoch": 20.425389755011135,
      "loss": 0.5979858040809631,
      "loss_ce": 8.540972339687869e-05,
      "loss_iou": 0.251953125,
      "loss_num": 0.0184326171875,
      "loss_xval": 0.59765625,
      "num_input_tokens_seen": 513821636,
      "step": 9171
    },
    {
      "epoch": 20.42761692650334,
      "grad_norm": 16.4615478515625,
      "learning_rate": 1e-06,
      "loss": 0.4864,
      "num_input_tokens_seen": 513878332,
      "step": 9172
    },
    {
      "epoch": 20.42761692650334,
      "loss": 0.3729918897151947,
      "loss_ce": 6.709150329697877e-05,
      "loss_iou": 0.17578125,
      "loss_num": 0.004119873046875,
      "loss_xval": 0.373046875,
      "num_input_tokens_seen": 513878332,
      "step": 9172
    },
    {
      "epoch": 20.429844097995545,
      "grad_norm": 17.841854095458984,
      "learning_rate": 1e-06,
      "loss": 0.329,
      "num_input_tokens_seen": 513934092,
      "step": 9173
    },
    {
      "epoch": 20.429844097995545,
      "loss": 0.3353223204612732,
      "loss_ce": 5.619797593681142e-05,
      "loss_iou": 0.154296875,
      "loss_num": 0.00537109375,
      "loss_xval": 0.3359375,
      "num_input_tokens_seen": 513934092,
      "step": 9173
    },
    {
      "epoch": 20.43207126948775,
      "grad_norm": 18.19709587097168,
      "learning_rate": 1e-06,
      "loss": 0.3155,
      "num_input_tokens_seen": 513989020,
      "step": 9174
    },
    {
      "epoch": 20.43207126948775,
      "loss": 0.2181463986635208,
      "loss_ce": 6.778672832297161e-05,
      "loss_iou": 0.09375,
      "loss_num": 0.00616455078125,
      "loss_xval": 0.2177734375,
      "num_input_tokens_seen": 513989020,
      "step": 9174
    },
    {
      "epoch": 20.434298440979955,
      "grad_norm": 35.7059440612793,
      "learning_rate": 1e-06,
      "loss": 0.5612,
      "num_input_tokens_seen": 514040600,
      "step": 9175
    },
    {
      "epoch": 20.434298440979955,
      "loss": 0.6007704734802246,
      "loss_ce": 6.243239477043971e-05,
      "loss_iou": 0.25390625,
      "loss_num": 0.018310546875,
      "loss_xval": 0.6015625,
      "num_input_tokens_seen": 514040600,
      "step": 9175
    },
    {
      "epoch": 20.43652561247216,
      "grad_norm": 17.803630828857422,
      "learning_rate": 1e-06,
      "loss": 0.2915,
      "num_input_tokens_seen": 514095804,
      "step": 9176
    },
    {
      "epoch": 20.43652561247216,
      "loss": 0.27951860427856445,
      "loss_ce": 9.964508353732526e-05,
      "loss_iou": 0.10791015625,
      "loss_num": 0.0128173828125,
      "loss_xval": 0.279296875,
      "num_input_tokens_seen": 514095804,
      "step": 9176
    },
    {
      "epoch": 20.438752783964365,
      "grad_norm": 15.95635986328125,
      "learning_rate": 1e-06,
      "loss": 0.3912,
      "num_input_tokens_seen": 514152436,
      "step": 9177
    },
    {
      "epoch": 20.438752783964365,
      "loss": 0.45222991704940796,
      "loss_ce": 8.150480425683782e-05,
      "loss_iou": 0.16796875,
      "loss_num": 0.0234375,
      "loss_xval": 0.453125,
      "num_input_tokens_seen": 514152436,
      "step": 9177
    },
    {
      "epoch": 20.44097995545657,
      "grad_norm": 30.868133544921875,
      "learning_rate": 1e-06,
      "loss": 0.5745,
      "num_input_tokens_seen": 514210436,
      "step": 9178
    },
    {
      "epoch": 20.44097995545657,
      "loss": 0.45710474252700806,
      "loss_ce": 7.350958185270429e-05,
      "loss_iou": 0.2177734375,
      "loss_num": 0.00421142578125,
      "loss_xval": 0.45703125,
      "num_input_tokens_seen": 514210436,
      "step": 9178
    },
    {
      "epoch": 20.443207126948774,
      "grad_norm": 28.9227352142334,
      "learning_rate": 1e-06,
      "loss": 0.3003,
      "num_input_tokens_seen": 514268436,
      "step": 9179
    },
    {
      "epoch": 20.443207126948774,
      "loss": 0.27464067935943604,
      "loss_ce": 4.350075323600322e-05,
      "loss_iou": 0.1181640625,
      "loss_num": 0.00762939453125,
      "loss_xval": 0.275390625,
      "num_input_tokens_seen": 514268436,
      "step": 9179
    },
    {
      "epoch": 20.44543429844098,
      "grad_norm": 18.41109848022461,
      "learning_rate": 1e-06,
      "loss": 0.2609,
      "num_input_tokens_seen": 514326408,
      "step": 9180
    },
    {
      "epoch": 20.44543429844098,
      "loss": 0.2086607664823532,
      "loss_ce": 0.00010363436012994498,
      "loss_iou": 0.08642578125,
      "loss_num": 0.007080078125,
      "loss_xval": 0.208984375,
      "num_input_tokens_seen": 514326408,
      "step": 9180
    },
    {
      "epoch": 20.447661469933184,
      "grad_norm": 22.37228012084961,
      "learning_rate": 1e-06,
      "loss": 0.2734,
      "num_input_tokens_seen": 514381776,
      "step": 9181
    },
    {
      "epoch": 20.447661469933184,
      "loss": 0.2675103545188904,
      "loss_ce": 5.431023964774795e-05,
      "loss_iou": 0.111328125,
      "loss_num": 0.00897216796875,
      "loss_xval": 0.267578125,
      "num_input_tokens_seen": 514381776,
      "step": 9181
    },
    {
      "epoch": 20.44988864142539,
      "grad_norm": 20.06549072265625,
      "learning_rate": 1e-06,
      "loss": 0.2565,
      "num_input_tokens_seen": 514435556,
      "step": 9182
    },
    {
      "epoch": 20.44988864142539,
      "loss": 0.2752624452114105,
      "loss_ce": 0.00011597707634791732,
      "loss_iou": 0.10546875,
      "loss_num": 0.012939453125,
      "loss_xval": 0.275390625,
      "num_input_tokens_seen": 514435556,
      "step": 9182
    },
    {
      "epoch": 20.452115812917594,
      "grad_norm": 14.442758560180664,
      "learning_rate": 1e-06,
      "loss": 0.4209,
      "num_input_tokens_seen": 514491956,
      "step": 9183
    },
    {
      "epoch": 20.452115812917594,
      "loss": 0.4563639163970947,
      "loss_ce": 6.508714432129636e-05,
      "loss_iou": 0.189453125,
      "loss_num": 0.0155029296875,
      "loss_xval": 0.45703125,
      "num_input_tokens_seen": 514491956,
      "step": 9183
    },
    {
      "epoch": 20.4543429844098,
      "grad_norm": 18.204957962036133,
      "learning_rate": 1e-06,
      "loss": 0.3314,
      "num_input_tokens_seen": 514548608,
      "step": 9184
    },
    {
      "epoch": 20.4543429844098,
      "loss": 0.2957186698913574,
      "loss_ce": 6.437343108700588e-05,
      "loss_iou": 0.1376953125,
      "loss_num": 0.00408935546875,
      "loss_xval": 0.294921875,
      "num_input_tokens_seen": 514548608,
      "step": 9184
    },
    {
      "epoch": 20.456570155902003,
      "grad_norm": 19.535131454467773,
      "learning_rate": 1e-06,
      "loss": 0.3625,
      "num_input_tokens_seen": 514604368,
      "step": 9185
    },
    {
      "epoch": 20.456570155902003,
      "loss": 0.4076521098613739,
      "loss_ce": 5.9345908084651455e-05,
      "loss_iou": 0.1796875,
      "loss_num": 0.00946044921875,
      "loss_xval": 0.408203125,
      "num_input_tokens_seen": 514604368,
      "step": 9185
    },
    {
      "epoch": 20.45879732739421,
      "grad_norm": 19.040164947509766,
      "learning_rate": 1e-06,
      "loss": 0.3052,
      "num_input_tokens_seen": 514661100,
      "step": 9186
    },
    {
      "epoch": 20.45879732739421,
      "loss": 0.29315710067749023,
      "loss_ce": 6.629424751736224e-05,
      "loss_iou": 0.130859375,
      "loss_num": 0.006103515625,
      "loss_xval": 0.29296875,
      "num_input_tokens_seen": 514661100,
      "step": 9186
    },
    {
      "epoch": 20.461024498886413,
      "grad_norm": 26.235673904418945,
      "learning_rate": 1e-06,
      "loss": 0.6357,
      "num_input_tokens_seen": 514715804,
      "step": 9187
    },
    {
      "epoch": 20.461024498886413,
      "loss": 0.4242648184299469,
      "loss_ce": 7.046835526125506e-05,
      "loss_iou": 0.1884765625,
      "loss_num": 0.00927734375,
      "loss_xval": 0.423828125,
      "num_input_tokens_seen": 514715804,
      "step": 9187
    },
    {
      "epoch": 20.463251670378618,
      "grad_norm": 20.59581756591797,
      "learning_rate": 1e-06,
      "loss": 0.2953,
      "num_input_tokens_seen": 514772648,
      "step": 9188
    },
    {
      "epoch": 20.463251670378618,
      "loss": 0.297730028629303,
      "loss_ce": 6.155186565592885e-05,
      "loss_iou": 0.1416015625,
      "loss_num": 0.0030364990234375,
      "loss_xval": 0.296875,
      "num_input_tokens_seen": 514772648,
      "step": 9188
    },
    {
      "epoch": 20.465478841870823,
      "grad_norm": 21.08500099182129,
      "learning_rate": 1e-06,
      "loss": 0.4619,
      "num_input_tokens_seen": 514827056,
      "step": 9189
    },
    {
      "epoch": 20.465478841870823,
      "loss": 0.5222786664962769,
      "loss_ce": 6.190245039761066e-05,
      "loss_iou": 0.2294921875,
      "loss_num": 0.01251220703125,
      "loss_xval": 0.5234375,
      "num_input_tokens_seen": 514827056,
      "step": 9189
    },
    {
      "epoch": 20.467706013363028,
      "grad_norm": 24.425458908081055,
      "learning_rate": 1e-06,
      "loss": 0.4498,
      "num_input_tokens_seen": 514878540,
      "step": 9190
    },
    {
      "epoch": 20.467706013363028,
      "loss": 0.27809709310531616,
      "loss_ce": 8.197416900657117e-05,
      "loss_iou": 0.115234375,
      "loss_num": 0.00933837890625,
      "loss_xval": 0.27734375,
      "num_input_tokens_seen": 514878540,
      "step": 9190
    },
    {
      "epoch": 20.469933184855233,
      "grad_norm": 20.29898452758789,
      "learning_rate": 1e-06,
      "loss": 0.3158,
      "num_input_tokens_seen": 514933080,
      "step": 9191
    },
    {
      "epoch": 20.469933184855233,
      "loss": 0.2902224063873291,
      "loss_ce": 6.12850344623439e-05,
      "loss_iou": 0.130859375,
      "loss_num": 0.005889892578125,
      "loss_xval": 0.291015625,
      "num_input_tokens_seen": 514933080,
      "step": 9191
    },
    {
      "epoch": 20.472160356347437,
      "grad_norm": 23.6481876373291,
      "learning_rate": 1e-06,
      "loss": 0.2375,
      "num_input_tokens_seen": 514987504,
      "step": 9192
    },
    {
      "epoch": 20.472160356347437,
      "loss": 0.16320902109146118,
      "loss_ce": 9.255468467017636e-05,
      "loss_iou": 0.06982421875,
      "loss_num": 0.004638671875,
      "loss_xval": 0.1630859375,
      "num_input_tokens_seen": 514987504,
      "step": 9192
    },
    {
      "epoch": 20.474387527839642,
      "grad_norm": 16.435392379760742,
      "learning_rate": 1e-06,
      "loss": 0.3885,
      "num_input_tokens_seen": 515043248,
      "step": 9193
    },
    {
      "epoch": 20.474387527839642,
      "loss": 0.4120558798313141,
      "loss_ce": 6.857035623397678e-05,
      "loss_iou": 0.1787109375,
      "loss_num": 0.01092529296875,
      "loss_xval": 0.412109375,
      "num_input_tokens_seen": 515043248,
      "step": 9193
    },
    {
      "epoch": 20.476614699331847,
      "grad_norm": 18.2564697265625,
      "learning_rate": 1e-06,
      "loss": 0.3537,
      "num_input_tokens_seen": 515098600,
      "step": 9194
    },
    {
      "epoch": 20.476614699331847,
      "loss": 0.3193051218986511,
      "loss_ce": 9.127189696300775e-05,
      "loss_iou": 0.1484375,
      "loss_num": 0.004547119140625,
      "loss_xval": 0.318359375,
      "num_input_tokens_seen": 515098600,
      "step": 9194
    },
    {
      "epoch": 20.478841870824052,
      "grad_norm": 47.8402099609375,
      "learning_rate": 1e-06,
      "loss": 0.2802,
      "num_input_tokens_seen": 515157696,
      "step": 9195
    },
    {
      "epoch": 20.478841870824052,
      "loss": 0.31334012746810913,
      "loss_ce": 0.00010771671804832295,
      "loss_iou": 0.1435546875,
      "loss_num": 0.005035400390625,
      "loss_xval": 0.3125,
      "num_input_tokens_seen": 515157696,
      "step": 9195
    },
    {
      "epoch": 20.481069042316257,
      "grad_norm": 16.30150604248047,
      "learning_rate": 1e-06,
      "loss": 0.2492,
      "num_input_tokens_seen": 515213144,
      "step": 9196
    },
    {
      "epoch": 20.481069042316257,
      "loss": 0.20818519592285156,
      "loss_ce": 5.53104946448002e-05,
      "loss_iou": 0.080078125,
      "loss_num": 0.00946044921875,
      "loss_xval": 0.2080078125,
      "num_input_tokens_seen": 515213144,
      "step": 9196
    },
    {
      "epoch": 20.48329621380846,
      "grad_norm": 12.833916664123535,
      "learning_rate": 1e-06,
      "loss": 0.2618,
      "num_input_tokens_seen": 515271576,
      "step": 9197
    },
    {
      "epoch": 20.48329621380846,
      "loss": 0.3264831006526947,
      "loss_ce": 6.709274020977318e-05,
      "loss_iou": 0.1357421875,
      "loss_num": 0.01092529296875,
      "loss_xval": 0.326171875,
      "num_input_tokens_seen": 515271576,
      "step": 9197
    },
    {
      "epoch": 20.485523385300667,
      "grad_norm": 124.69937896728516,
      "learning_rate": 1e-06,
      "loss": 0.3763,
      "num_input_tokens_seen": 515327292,
      "step": 9198
    },
    {
      "epoch": 20.485523385300667,
      "loss": 0.25111645460128784,
      "loss_ce": 7.887819083407521e-05,
      "loss_iou": 0.10595703125,
      "loss_num": 0.0079345703125,
      "loss_xval": 0.251953125,
      "num_input_tokens_seen": 515327292,
      "step": 9198
    },
    {
      "epoch": 20.48775055679287,
      "grad_norm": 30.976198196411133,
      "learning_rate": 1e-06,
      "loss": 0.5357,
      "num_input_tokens_seen": 515382716,
      "step": 9199
    },
    {
      "epoch": 20.48775055679287,
      "loss": 0.6795259118080139,
      "loss_ce": 9.781922562979162e-05,
      "loss_iou": 0.298828125,
      "loss_num": 0.016357421875,
      "loss_xval": 0.6796875,
      "num_input_tokens_seen": 515382716,
      "step": 9199
    },
    {
      "epoch": 20.489977728285076,
      "grad_norm": 37.82209396362305,
      "learning_rate": 1e-06,
      "loss": 0.3135,
      "num_input_tokens_seen": 515438296,
      "step": 9200
    },
    {
      "epoch": 20.489977728285076,
      "loss": 0.38873451948165894,
      "loss_ce": 6.264903640840203e-05,
      "loss_iou": 0.177734375,
      "loss_num": 0.006622314453125,
      "loss_xval": 0.388671875,
      "num_input_tokens_seen": 515438296,
      "step": 9200
    },
    {
      "epoch": 20.49220489977728,
      "grad_norm": 15.100608825683594,
      "learning_rate": 1e-06,
      "loss": 0.2894,
      "num_input_tokens_seen": 515495380,
      "step": 9201
    },
    {
      "epoch": 20.49220489977728,
      "loss": 0.3832995295524597,
      "loss_ce": 5.9795893321279436e-05,
      "loss_iou": 0.17578125,
      "loss_num": 0.00653076171875,
      "loss_xval": 0.3828125,
      "num_input_tokens_seen": 515495380,
      "step": 9201
    },
    {
      "epoch": 20.494432071269486,
      "grad_norm": 26.354969024658203,
      "learning_rate": 1e-06,
      "loss": 0.3353,
      "num_input_tokens_seen": 515552492,
      "step": 9202
    },
    {
      "epoch": 20.494432071269486,
      "loss": 0.3944174647331238,
      "loss_ce": 6.930494419066235e-05,
      "loss_iou": 0.162109375,
      "loss_num": 0.01409912109375,
      "loss_xval": 0.39453125,
      "num_input_tokens_seen": 515552492,
      "step": 9202
    },
    {
      "epoch": 20.49665924276169,
      "grad_norm": 19.706466674804688,
      "learning_rate": 1e-06,
      "loss": 0.3413,
      "num_input_tokens_seen": 515607424,
      "step": 9203
    },
    {
      "epoch": 20.49665924276169,
      "loss": 0.3399580419063568,
      "loss_ce": 5.32609956280794e-05,
      "loss_iou": 0.1298828125,
      "loss_num": 0.0159912109375,
      "loss_xval": 0.33984375,
      "num_input_tokens_seen": 515607424,
      "step": 9203
    },
    {
      "epoch": 20.498886414253896,
      "grad_norm": 24.811115264892578,
      "learning_rate": 1e-06,
      "loss": 0.4147,
      "num_input_tokens_seen": 515661964,
      "step": 9204
    },
    {
      "epoch": 20.498886414253896,
      "loss": 0.4160168766975403,
      "loss_ce": 9.280447557102889e-05,
      "loss_iou": 0.1806640625,
      "loss_num": 0.0107421875,
      "loss_xval": 0.416015625,
      "num_input_tokens_seen": 515661964,
      "step": 9204
    },
    {
      "epoch": 20.501113585746104,
      "grad_norm": 27.92363929748535,
      "learning_rate": 1e-06,
      "loss": 0.3021,
      "num_input_tokens_seen": 515717596,
      "step": 9205
    },
    {
      "epoch": 20.501113585746104,
      "loss": 0.39837294816970825,
      "loss_ce": 5.751060962211341e-05,
      "loss_iou": 0.169921875,
      "loss_num": 0.01171875,
      "loss_xval": 0.3984375,
      "num_input_tokens_seen": 515717596,
      "step": 9205
    },
    {
      "epoch": 20.50334075723831,
      "grad_norm": 14.264836311340332,
      "learning_rate": 1e-06,
      "loss": 0.4384,
      "num_input_tokens_seen": 515773496,
      "step": 9206
    },
    {
      "epoch": 20.50334075723831,
      "loss": 0.35908016562461853,
      "loss_ce": 7.139322406146675e-05,
      "loss_iou": 0.1494140625,
      "loss_num": 0.0120849609375,
      "loss_xval": 0.359375,
      "num_input_tokens_seen": 515773496,
      "step": 9206
    },
    {
      "epoch": 20.505567928730514,
      "grad_norm": 17.64116668701172,
      "learning_rate": 1e-06,
      "loss": 0.432,
      "num_input_tokens_seen": 515827180,
      "step": 9207
    },
    {
      "epoch": 20.505567928730514,
      "loss": 0.4668075442314148,
      "loss_ce": 7.171083416324109e-05,
      "loss_iou": 0.1962890625,
      "loss_num": 0.01495361328125,
      "loss_xval": 0.466796875,
      "num_input_tokens_seen": 515827180,
      "step": 9207
    },
    {
      "epoch": 20.50779510022272,
      "grad_norm": 19.3404541015625,
      "learning_rate": 1e-06,
      "loss": 0.3338,
      "num_input_tokens_seen": 515881772,
      "step": 9208
    },
    {
      "epoch": 20.50779510022272,
      "loss": 0.31878164410591125,
      "loss_ce": 5.6054937886074185e-05,
      "loss_iou": 0.146484375,
      "loss_num": 0.0050048828125,
      "loss_xval": 0.318359375,
      "num_input_tokens_seen": 515881772,
      "step": 9208
    },
    {
      "epoch": 20.510022271714924,
      "grad_norm": 19.894765853881836,
      "learning_rate": 1e-06,
      "loss": 0.3582,
      "num_input_tokens_seen": 515938532,
      "step": 9209
    },
    {
      "epoch": 20.510022271714924,
      "loss": 0.3078060746192932,
      "loss_ce": 6.68221473461017e-05,
      "loss_iou": 0.1279296875,
      "loss_num": 0.0106201171875,
      "loss_xval": 0.30859375,
      "num_input_tokens_seen": 515938532,
      "step": 9209
    },
    {
      "epoch": 20.51224944320713,
      "grad_norm": 20.052772521972656,
      "learning_rate": 1e-06,
      "loss": 0.4125,
      "num_input_tokens_seen": 515996532,
      "step": 9210
    },
    {
      "epoch": 20.51224944320713,
      "loss": 0.5830047130584717,
      "loss_ce": 5.793360833195038e-05,
      "loss_iou": 0.248046875,
      "loss_num": 0.017333984375,
      "loss_xval": 0.58203125,
      "num_input_tokens_seen": 515996532,
      "step": 9210
    },
    {
      "epoch": 20.514476614699333,
      "grad_norm": 24.21659278869629,
      "learning_rate": 1e-06,
      "loss": 0.313,
      "num_input_tokens_seen": 516052064,
      "step": 9211
    },
    {
      "epoch": 20.514476614699333,
      "loss": 0.323422372341156,
      "loss_ce": 5.810863513033837e-05,
      "loss_iou": 0.1396484375,
      "loss_num": 0.00872802734375,
      "loss_xval": 0.32421875,
      "num_input_tokens_seen": 516052064,
      "step": 9211
    },
    {
      "epoch": 20.51670378619154,
      "grad_norm": 18.89595603942871,
      "learning_rate": 1e-06,
      "loss": 0.2474,
      "num_input_tokens_seen": 516109116,
      "step": 9212
    },
    {
      "epoch": 20.51670378619154,
      "loss": 0.18618465960025787,
      "loss_ce": 5.79521874897182e-05,
      "loss_iou": 0.08203125,
      "loss_num": 0.004364013671875,
      "loss_xval": 0.1865234375,
      "num_input_tokens_seen": 516109116,
      "step": 9212
    },
    {
      "epoch": 20.518930957683743,
      "grad_norm": 22.48140525817871,
      "learning_rate": 1e-06,
      "loss": 0.3139,
      "num_input_tokens_seen": 516162824,
      "step": 9213
    },
    {
      "epoch": 20.518930957683743,
      "loss": 0.28497621417045593,
      "loss_ce": 6.412078801076859e-05,
      "loss_iou": 0.1259765625,
      "loss_num": 0.0067138671875,
      "loss_xval": 0.28515625,
      "num_input_tokens_seen": 516162824,
      "step": 9213
    },
    {
      "epoch": 20.521158129175948,
      "grad_norm": 20.399208068847656,
      "learning_rate": 1e-06,
      "loss": 0.4476,
      "num_input_tokens_seen": 516216872,
      "step": 9214
    },
    {
      "epoch": 20.521158129175948,
      "loss": 0.6123678684234619,
      "loss_ce": 6.322791159618646e-05,
      "loss_iou": 0.2734375,
      "loss_num": 0.012939453125,
      "loss_xval": 0.61328125,
      "num_input_tokens_seen": 516216872,
      "step": 9214
    },
    {
      "epoch": 20.523385300668153,
      "grad_norm": 23.317928314208984,
      "learning_rate": 1e-06,
      "loss": 0.4519,
      "num_input_tokens_seen": 516269896,
      "step": 9215
    },
    {
      "epoch": 20.523385300668153,
      "loss": 0.4913950562477112,
      "loss_ce": 6.203198427101597e-05,
      "loss_iou": 0.2099609375,
      "loss_num": 0.01434326171875,
      "loss_xval": 0.4921875,
      "num_input_tokens_seen": 516269896,
      "step": 9215
    },
    {
      "epoch": 20.525612472160358,
      "grad_norm": 24.348512649536133,
      "learning_rate": 1e-06,
      "loss": 0.5282,
      "num_input_tokens_seen": 516324752,
      "step": 9216
    },
    {
      "epoch": 20.525612472160358,
      "loss": 0.30804580450057983,
      "loss_ce": 6.23964297119528e-05,
      "loss_iou": 0.146484375,
      "loss_num": 0.0029449462890625,
      "loss_xval": 0.30859375,
      "num_input_tokens_seen": 516324752,
      "step": 9216
    },
    {
      "epoch": 20.527839643652563,
      "grad_norm": 19.019367218017578,
      "learning_rate": 1e-06,
      "loss": 0.3523,
      "num_input_tokens_seen": 516381168,
      "step": 9217
    },
    {
      "epoch": 20.527839643652563,
      "loss": 0.34173130989074707,
      "loss_ce": 5.6528966524638236e-05,
      "loss_iou": 0.15625,
      "loss_num": 0.005767822265625,
      "loss_xval": 0.341796875,
      "num_input_tokens_seen": 516381168,
      "step": 9217
    },
    {
      "epoch": 20.530066815144767,
      "grad_norm": 14.334078788757324,
      "learning_rate": 1e-06,
      "loss": 0.4507,
      "num_input_tokens_seen": 516438516,
      "step": 9218
    },
    {
      "epoch": 20.530066815144767,
      "loss": 0.6355735063552856,
      "loss_ce": 7.543759420514107e-05,
      "loss_iou": 0.26953125,
      "loss_num": 0.0196533203125,
      "loss_xval": 0.63671875,
      "num_input_tokens_seen": 516438516,
      "step": 9218
    },
    {
      "epoch": 20.532293986636972,
      "grad_norm": 24.67681312561035,
      "learning_rate": 1e-06,
      "loss": 0.5983,
      "num_input_tokens_seen": 516492580,
      "step": 9219
    },
    {
      "epoch": 20.532293986636972,
      "loss": 0.5057117938995361,
      "loss_ce": 0.00015756976790726185,
      "loss_iou": 0.1962890625,
      "loss_num": 0.0224609375,
      "loss_xval": 0.50390625,
      "num_input_tokens_seen": 516492580,
      "step": 9219
    },
    {
      "epoch": 20.534521158129177,
      "grad_norm": 25.80486297607422,
      "learning_rate": 1e-06,
      "loss": 0.4122,
      "num_input_tokens_seen": 516546048,
      "step": 9220
    },
    {
      "epoch": 20.534521158129177,
      "loss": 0.5275466442108154,
      "loss_ce": 8.086032175924629e-05,
      "loss_iou": 0.2333984375,
      "loss_num": 0.01214599609375,
      "loss_xval": 0.52734375,
      "num_input_tokens_seen": 516546048,
      "step": 9220
    },
    {
      "epoch": 20.536748329621382,
      "grad_norm": 19.290876388549805,
      "learning_rate": 1e-06,
      "loss": 0.3358,
      "num_input_tokens_seen": 516600172,
      "step": 9221
    },
    {
      "epoch": 20.536748329621382,
      "loss": 0.40956592559814453,
      "loss_ce": 6.581825437024236e-05,
      "loss_iou": 0.1904296875,
      "loss_num": 0.005767822265625,
      "loss_xval": 0.41015625,
      "num_input_tokens_seen": 516600172,
      "step": 9221
    },
    {
      "epoch": 20.538975501113587,
      "grad_norm": 14.446426391601562,
      "learning_rate": 1e-06,
      "loss": 0.3625,
      "num_input_tokens_seen": 516655008,
      "step": 9222
    },
    {
      "epoch": 20.538975501113587,
      "loss": 0.4629659950733185,
      "loss_ce": 7.537858618889004e-05,
      "loss_iou": 0.19140625,
      "loss_num": 0.0162353515625,
      "loss_xval": 0.462890625,
      "num_input_tokens_seen": 516655008,
      "step": 9222
    },
    {
      "epoch": 20.54120267260579,
      "grad_norm": 20.61759376525879,
      "learning_rate": 1e-06,
      "loss": 0.3554,
      "num_input_tokens_seen": 516713012,
      "step": 9223
    },
    {
      "epoch": 20.54120267260579,
      "loss": 0.2745492458343506,
      "loss_ce": 7.413580169668421e-05,
      "loss_iou": 0.12109375,
      "loss_num": 0.006561279296875,
      "loss_xval": 0.275390625,
      "num_input_tokens_seen": 516713012,
      "step": 9223
    },
    {
      "epoch": 20.543429844097997,
      "grad_norm": 15.4387788772583,
      "learning_rate": 1e-06,
      "loss": 0.3891,
      "num_input_tokens_seen": 516771928,
      "step": 9224
    },
    {
      "epoch": 20.543429844097997,
      "loss": 0.21673524379730225,
      "loss_ce": 6.0440983361331746e-05,
      "loss_iou": 0.09912109375,
      "loss_num": 0.003662109375,
      "loss_xval": 0.216796875,
      "num_input_tokens_seen": 516771928,
      "step": 9224
    },
    {
      "epoch": 20.5456570155902,
      "grad_norm": 73.51461791992188,
      "learning_rate": 1e-06,
      "loss": 0.3232,
      "num_input_tokens_seen": 516827052,
      "step": 9225
    },
    {
      "epoch": 20.5456570155902,
      "loss": 0.37042248249053955,
      "loss_ce": 6.116791337262839e-05,
      "loss_iou": 0.17578125,
      "loss_num": 0.0037994384765625,
      "loss_xval": 0.37109375,
      "num_input_tokens_seen": 516827052,
      "step": 9225
    },
    {
      "epoch": 20.547884187082406,
      "grad_norm": 18.046146392822266,
      "learning_rate": 1e-06,
      "loss": 0.3615,
      "num_input_tokens_seen": 516883920,
      "step": 9226
    },
    {
      "epoch": 20.547884187082406,
      "loss": 0.335880845785141,
      "loss_ce": 6.541566108353436e-05,
      "loss_iou": 0.1513671875,
      "loss_num": 0.0064697265625,
      "loss_xval": 0.3359375,
      "num_input_tokens_seen": 516883920,
      "step": 9226
    },
    {
      "epoch": 20.55011135857461,
      "grad_norm": 24.70758628845215,
      "learning_rate": 1e-06,
      "loss": 0.3517,
      "num_input_tokens_seen": 516939448,
      "step": 9227
    },
    {
      "epoch": 20.55011135857461,
      "loss": 0.3225664794445038,
      "loss_ce": 5.6713775848038495e-05,
      "loss_iou": 0.134765625,
      "loss_num": 0.01055908203125,
      "loss_xval": 0.322265625,
      "num_input_tokens_seen": 516939448,
      "step": 9227
    },
    {
      "epoch": 20.552338530066816,
      "grad_norm": 18.065723419189453,
      "learning_rate": 1e-06,
      "loss": 0.4389,
      "num_input_tokens_seen": 516994964,
      "step": 9228
    },
    {
      "epoch": 20.552338530066816,
      "loss": 0.35278570652008057,
      "loss_ce": 6.352636410156265e-05,
      "loss_iou": 0.16015625,
      "loss_num": 0.0064697265625,
      "loss_xval": 0.353515625,
      "num_input_tokens_seen": 516994964,
      "step": 9228
    },
    {
      "epoch": 20.55456570155902,
      "grad_norm": 17.124879837036133,
      "learning_rate": 1e-06,
      "loss": 0.4484,
      "num_input_tokens_seen": 517051364,
      "step": 9229
    },
    {
      "epoch": 20.55456570155902,
      "loss": 0.5044651031494141,
      "loss_ce": 7.060293137328699e-05,
      "loss_iou": 0.2265625,
      "loss_num": 0.01043701171875,
      "loss_xval": 0.50390625,
      "num_input_tokens_seen": 517051364,
      "step": 9229
    },
    {
      "epoch": 20.556792873051226,
      "grad_norm": 19.566333770751953,
      "learning_rate": 1e-06,
      "loss": 0.3415,
      "num_input_tokens_seen": 517106692,
      "step": 9230
    },
    {
      "epoch": 20.556792873051226,
      "loss": 0.3661907911300659,
      "loss_ce": 7.13969930075109e-05,
      "loss_iou": 0.1640625,
      "loss_num": 0.007415771484375,
      "loss_xval": 0.365234375,
      "num_input_tokens_seen": 517106692,
      "step": 9230
    },
    {
      "epoch": 20.55902004454343,
      "grad_norm": 16.50565528869629,
      "learning_rate": 1e-06,
      "loss": 0.3832,
      "num_input_tokens_seen": 517162332,
      "step": 9231
    },
    {
      "epoch": 20.55902004454343,
      "loss": 0.3715935945510864,
      "loss_ce": 7.258763071149588e-05,
      "loss_iou": 0.1630859375,
      "loss_num": 0.00933837890625,
      "loss_xval": 0.37109375,
      "num_input_tokens_seen": 517162332,
      "step": 9231
    },
    {
      "epoch": 20.561247216035635,
      "grad_norm": 26.36023712158203,
      "learning_rate": 1e-06,
      "loss": 0.3381,
      "num_input_tokens_seen": 517219120,
      "step": 9232
    },
    {
      "epoch": 20.561247216035635,
      "loss": 0.26105207204818726,
      "loss_ce": 6.5757536503952e-05,
      "loss_iou": 0.119140625,
      "loss_num": 0.004547119140625,
      "loss_xval": 0.26171875,
      "num_input_tokens_seen": 517219120,
      "step": 9232
    },
    {
      "epoch": 20.56347438752784,
      "grad_norm": 40.432926177978516,
      "learning_rate": 1e-06,
      "loss": 0.5472,
      "num_input_tokens_seen": 517273304,
      "step": 9233
    },
    {
      "epoch": 20.56347438752784,
      "loss": 0.3474746346473694,
      "loss_ce": 6.25449392828159e-05,
      "loss_iou": 0.162109375,
      "loss_num": 0.00482177734375,
      "loss_xval": 0.34765625,
      "num_input_tokens_seen": 517273304,
      "step": 9233
    },
    {
      "epoch": 20.565701559020045,
      "grad_norm": 37.93396759033203,
      "learning_rate": 1e-06,
      "loss": 0.3976,
      "num_input_tokens_seen": 517327196,
      "step": 9234
    },
    {
      "epoch": 20.565701559020045,
      "loss": 0.4561777412891388,
      "loss_ce": 6.202464282978326e-05,
      "loss_iou": 0.1962890625,
      "loss_num": 0.01263427734375,
      "loss_xval": 0.45703125,
      "num_input_tokens_seen": 517327196,
      "step": 9234
    },
    {
      "epoch": 20.56792873051225,
      "grad_norm": 16.984933853149414,
      "learning_rate": 1e-06,
      "loss": 0.3733,
      "num_input_tokens_seen": 517384080,
      "step": 9235
    },
    {
      "epoch": 20.56792873051225,
      "loss": 0.4556872844696045,
      "loss_ce": 5.983790470054373e-05,
      "loss_iou": 0.1826171875,
      "loss_num": 0.0179443359375,
      "loss_xval": 0.455078125,
      "num_input_tokens_seen": 517384080,
      "step": 9235
    },
    {
      "epoch": 20.570155902004455,
      "grad_norm": 9.283608436584473,
      "learning_rate": 1e-06,
      "loss": 0.3774,
      "num_input_tokens_seen": 517438320,
      "step": 9236
    },
    {
      "epoch": 20.570155902004455,
      "loss": 0.4296599328517914,
      "loss_ce": 0.00011740654008463025,
      "loss_iou": 0.18359375,
      "loss_num": 0.01220703125,
      "loss_xval": 0.4296875,
      "num_input_tokens_seen": 517438320,
      "step": 9236
    },
    {
      "epoch": 20.57238307349666,
      "grad_norm": 16.504417419433594,
      "learning_rate": 1e-06,
      "loss": 0.3546,
      "num_input_tokens_seen": 517496136,
      "step": 9237
    },
    {
      "epoch": 20.57238307349666,
      "loss": 0.3683554530143738,
      "loss_ce": 6.933379336260259e-05,
      "loss_iou": 0.16015625,
      "loss_num": 0.00958251953125,
      "loss_xval": 0.369140625,
      "num_input_tokens_seen": 517496136,
      "step": 9237
    },
    {
      "epoch": 20.574610244988865,
      "grad_norm": 19.25970458984375,
      "learning_rate": 1e-06,
      "loss": 0.2776,
      "num_input_tokens_seen": 517553096,
      "step": 9238
    },
    {
      "epoch": 20.574610244988865,
      "loss": 0.3049623966217041,
      "loss_ce": 6.127238884801045e-05,
      "loss_iou": 0.14453125,
      "loss_num": 0.0030517578125,
      "loss_xval": 0.3046875,
      "num_input_tokens_seen": 517553096,
      "step": 9238
    },
    {
      "epoch": 20.57683741648107,
      "grad_norm": 19.565448760986328,
      "learning_rate": 1e-06,
      "loss": 0.3708,
      "num_input_tokens_seen": 517608736,
      "step": 9239
    },
    {
      "epoch": 20.57683741648107,
      "loss": 0.23846621811389923,
      "loss_ce": 6.2903571233619e-05,
      "loss_iou": 0.09814453125,
      "loss_num": 0.0084228515625,
      "loss_xval": 0.23828125,
      "num_input_tokens_seen": 517608736,
      "step": 9239
    },
    {
      "epoch": 20.579064587973274,
      "grad_norm": 15.315186500549316,
      "learning_rate": 1e-06,
      "loss": 0.4186,
      "num_input_tokens_seen": 517664492,
      "step": 9240
    },
    {
      "epoch": 20.579064587973274,
      "loss": 0.5499961972236633,
      "loss_ce": 6.946978101041168e-05,
      "loss_iou": 0.2060546875,
      "loss_num": 0.0277099609375,
      "loss_xval": 0.55078125,
      "num_input_tokens_seen": 517664492,
      "step": 9240
    },
    {
      "epoch": 20.58129175946548,
      "grad_norm": 19.74601173400879,
      "learning_rate": 1e-06,
      "loss": 0.4442,
      "num_input_tokens_seen": 517723296,
      "step": 9241
    },
    {
      "epoch": 20.58129175946548,
      "loss": 0.39636969566345215,
      "loss_ce": 6.84402184560895e-05,
      "loss_iou": 0.1552734375,
      "loss_num": 0.0169677734375,
      "loss_xval": 0.396484375,
      "num_input_tokens_seen": 517723296,
      "step": 9241
    },
    {
      "epoch": 20.583518930957684,
      "grad_norm": 13.046077728271484,
      "learning_rate": 1e-06,
      "loss": 0.2947,
      "num_input_tokens_seen": 517783636,
      "step": 9242
    },
    {
      "epoch": 20.583518930957684,
      "loss": 0.2621099352836609,
      "loss_ce": 5.5477234127465636e-05,
      "loss_iou": 0.1201171875,
      "loss_num": 0.004302978515625,
      "loss_xval": 0.26171875,
      "num_input_tokens_seen": 517783636,
      "step": 9242
    },
    {
      "epoch": 20.58574610244989,
      "grad_norm": 15.847582817077637,
      "learning_rate": 1e-06,
      "loss": 0.2867,
      "num_input_tokens_seen": 517837348,
      "step": 9243
    },
    {
      "epoch": 20.58574610244989,
      "loss": 0.36224034428596497,
      "loss_ce": 5.771276482846588e-05,
      "loss_iou": 0.1591796875,
      "loss_num": 0.00860595703125,
      "loss_xval": 0.361328125,
      "num_input_tokens_seen": 517837348,
      "step": 9243
    },
    {
      "epoch": 20.587973273942094,
      "grad_norm": 18.27277374267578,
      "learning_rate": 1e-06,
      "loss": 0.4209,
      "num_input_tokens_seen": 517892936,
      "step": 9244
    },
    {
      "epoch": 20.587973273942094,
      "loss": 0.3434508144855499,
      "loss_ce": 6.703373946947977e-05,
      "loss_iou": 0.158203125,
      "loss_num": 0.005523681640625,
      "loss_xval": 0.34375,
      "num_input_tokens_seen": 517892936,
      "step": 9244
    },
    {
      "epoch": 20.5902004454343,
      "grad_norm": 18.745811462402344,
      "learning_rate": 1e-06,
      "loss": 0.3061,
      "num_input_tokens_seen": 517948732,
      "step": 9245
    },
    {
      "epoch": 20.5902004454343,
      "loss": 0.3197273313999176,
      "loss_ce": 0.0001472493022447452,
      "loss_iou": 0.134765625,
      "loss_num": 0.010009765625,
      "loss_xval": 0.3203125,
      "num_input_tokens_seen": 517948732,
      "step": 9245
    },
    {
      "epoch": 20.592427616926503,
      "grad_norm": 25.77054214477539,
      "learning_rate": 1e-06,
      "loss": 0.5403,
      "num_input_tokens_seen": 518005172,
      "step": 9246
    },
    {
      "epoch": 20.592427616926503,
      "loss": 0.6770814657211304,
      "loss_ce": 7.9466997704003e-05,
      "loss_iou": 0.291015625,
      "loss_num": 0.0189208984375,
      "loss_xval": 0.67578125,
      "num_input_tokens_seen": 518005172,
      "step": 9246
    },
    {
      "epoch": 20.59465478841871,
      "grad_norm": 14.424224853515625,
      "learning_rate": 1e-06,
      "loss": 0.5092,
      "num_input_tokens_seen": 518059824,
      "step": 9247
    },
    {
      "epoch": 20.59465478841871,
      "loss": 0.6334977746009827,
      "loss_ce": 7.492607983294874e-05,
      "loss_iou": 0.283203125,
      "loss_num": 0.0135498046875,
      "loss_xval": 0.6328125,
      "num_input_tokens_seen": 518059824,
      "step": 9247
    },
    {
      "epoch": 20.596881959910913,
      "grad_norm": 22.720426559448242,
      "learning_rate": 1e-06,
      "loss": 0.3461,
      "num_input_tokens_seen": 518116568,
      "step": 9248
    },
    {
      "epoch": 20.596881959910913,
      "loss": 0.3957892954349518,
      "loss_ce": 9.837795369094238e-05,
      "loss_iou": 0.1611328125,
      "loss_num": 0.01470947265625,
      "loss_xval": 0.396484375,
      "num_input_tokens_seen": 518116568,
      "step": 9248
    },
    {
      "epoch": 20.599109131403118,
      "grad_norm": 24.825748443603516,
      "learning_rate": 1e-06,
      "loss": 0.4304,
      "num_input_tokens_seen": 518171416,
      "step": 9249
    },
    {
      "epoch": 20.599109131403118,
      "loss": 0.44866153597831726,
      "loss_ce": 5.314835289027542e-05,
      "loss_iou": 0.197265625,
      "loss_num": 0.0108642578125,
      "loss_xval": 0.44921875,
      "num_input_tokens_seen": 518171416,
      "step": 9249
    },
    {
      "epoch": 20.601336302895323,
      "grad_norm": 16.588918685913086,
      "learning_rate": 1e-06,
      "loss": 0.2945,
      "num_input_tokens_seen": 518230356,
      "step": 9250
    },
    {
      "epoch": 20.601336302895323,
      "eval_seeclick_web_CIoU": 0.583829253911972,
      "eval_seeclick_web_GIoU": 0.5839782357215881,
      "eval_seeclick_web_IoU": 0.6033029854297638,
      "eval_seeclick_web_MAE_all": 0.015166566707193851,
      "eval_seeclick_web_MAE_h": 0.006818295689299703,
      "eval_seeclick_web_MAE_w": 0.015144134871661663,
      "eval_seeclick_web_MAE_x_boxes": 0.008378135273233056,
      "eval_seeclick_web_MAE_y_boxes": 0.021190070547163486,
      "eval_seeclick_web_inside_bbox": 0.9010416567325592,
      "eval_seeclick_web_loss": 0.9226139783859253,
      "eval_seeclick_web_loss_ce": 0.00011066030492656864,
      "eval_seeclick_web_loss_iou": 0.42724609375,
      "eval_seeclick_web_loss_num": 0.0122528076171875,
      "eval_seeclick_web_loss_xval": 0.9150390625,
      "eval_seeclick_web_runtime": 20.8356,
      "eval_seeclick_web_samples_per_second": 2.4,
      "eval_seeclick_web_steps_per_second": 0.096,
      "num_input_tokens_seen": 518230356,
      "step": 9250
    },
    {
      "epoch": 20.601336302895323,
      "eval_icons_CIoU": 0.25548499822616577,
      "eval_icons_GIoU": 0.2894558012485504,
      "eval_icons_IoU": 0.33956870436668396,
      "eval_icons_MAE_all": 0.062356239184737206,
      "eval_icons_MAE_h": 0.029669910669326782,
      "eval_icons_MAE_w": 0.07730034179985523,
      "eval_icons_MAE_x_boxes": 0.05503878928720951,
      "eval_icons_MAE_y_boxes": 0.037660352885723114,
      "eval_icons_inside_bbox": 0.59375,
      "eval_icons_loss": 1.7357631921768188,
      "eval_icons_loss_ce": 0.00013154443513485603,
      "eval_icons_loss_iou": 0.6806640625,
      "eval_icons_loss_num": 0.06060218811035156,
      "eval_icons_loss_xval": 1.6640625,
      "eval_icons_runtime": 20.933,
      "eval_icons_samples_per_second": 2.389,
      "eval_icons_steps_per_second": 0.096,
      "num_input_tokens_seen": 518230356,
      "step": 9250
    },
    {
      "epoch": 20.601336302895323,
      "eval_screenspot_CIoU": 0.3744088610013326,
      "eval_screenspot_GIoU": 0.3934180239836375,
      "eval_screenspot_IoU": 0.44850159684817,
      "eval_screenspot_MAE_all": 0.058480776846408844,
      "eval_screenspot_MAE_h": 0.03973913627366225,
      "eval_screenspot_MAE_w": 0.06190794085462888,
      "eval_screenspot_MAE_x_boxes": 0.06778114537398021,
      "eval_screenspot_MAE_y_boxes": 0.043080691869060196,
      "eval_screenspot_inside_bbox": 0.7041666706403097,
      "eval_screenspot_loss": 1.5582458972930908,
      "eval_screenspot_loss_ce": 0.00013941208211084208,
      "eval_screenspot_loss_iou": 0.640625,
      "eval_screenspot_loss_num": 0.06737772623697917,
      "eval_screenspot_loss_xval": 1.6188151041666667,
      "eval_screenspot_runtime": 29.263,
      "eval_screenspot_samples_per_second": 3.041,
      "eval_screenspot_steps_per_second": 0.103,
      "num_input_tokens_seen": 518230356,
      "step": 9250
    },
    {
      "epoch": 20.601336302895323,
      "eval_compot_CIoU": 0.3450569659471512,
      "eval_compot_GIoU": 0.3522755652666092,
      "eval_compot_IoU": 0.40288856625556946,
      "eval_compot_MAE_all": 0.018017619848251343,
      "eval_compot_MAE_h": 0.008548983139917254,
      "eval_compot_MAE_w": 0.020043439231812954,
      "eval_compot_MAE_x_boxes": 0.03034346178174019,
      "eval_compot_MAE_y_boxes": 0.007265899330377579,
      "eval_compot_inside_bbox": 0.6302083432674408,
      "eval_compot_loss": 1.3769763708114624,
      "eval_compot_loss_ce": 0.00010162792386836372,
      "eval_compot_loss_iou": 0.633544921875,
      "eval_compot_loss_num": 0.016551971435546875,
      "eval_compot_loss_xval": 1.34912109375,
      "eval_compot_runtime": 22.9916,
      "eval_compot_samples_per_second": 2.175,
      "eval_compot_steps_per_second": 0.087,
      "num_input_tokens_seen": 518230356,
      "step": 9250
    },
    {
      "epoch": 20.601336302895323,
      "eval_custom_ui_val_CIoU": 0.47359417213333976,
      "eval_custom_ui_val_GIoU": 0.4787476923730638,
      "eval_custom_ui_val_IoU": 0.5349448025226593,
      "eval_custom_ui_val_MAE_all": 0.02652052231132984,
      "eval_custom_ui_val_MAE_h": 0.013495955590365661,
      "eval_custom_ui_val_MAE_w": 0.036366066274543606,
      "eval_custom_ui_val_MAE_x_boxes": 0.03154980060127047,
      "eval_custom_ui_val_MAE_y_boxes": 0.012846555560827255,
      "eval_custom_ui_val_inside_bbox": 0.7596450646718343,
      "eval_custom_ui_val_loss": 1.1579389572143555,
      "eval_custom_ui_val_loss_ce": 0.00010260268269727628,
      "eval_custom_ui_val_loss_iou": 0.4972601996527778,
      "eval_custom_ui_val_loss_num": 0.023241043090820312,
      "eval_custom_ui_val_loss_xval": 1.1111111111111112,
      "eval_custom_ui_val_runtime": 58.4246,
      "eval_custom_ui_val_samples_per_second": 4.536,
      "eval_custom_ui_val_steps_per_second": 0.154,
      "num_input_tokens_seen": 518230356,
      "step": 9250
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 22450,
  "num_input_tokens_seen": 518230356,
  "num_train_epochs": 50,
  "save_steps": 250,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 2.410904613476211e+19,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}