{
  "best_metric": 0.9130434782608695,
  "best_model_checkpoint": "CTMAE2_CS_V7_1/checkpoint-13209",
  "epoch": 49.02,
  "eval_steps": 500,
  "global_step": 38850,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0002574002574002574,
      "grad_norm": 5.2134599685668945,
      "learning_rate": 2.5740025740025742e-08,
      "loss": 0.6904,
      "step": 10
    },
    {
      "epoch": 0.0005148005148005148,
      "grad_norm": 5.708760738372803,
      "learning_rate": 5.1480051480051484e-08,
      "loss": 0.6958,
      "step": 20
    },
    {
      "epoch": 0.0007722007722007722,
      "grad_norm": 5.368264675140381,
      "learning_rate": 7.722007722007723e-08,
      "loss": 0.696,
      "step": 30
    },
    {
      "epoch": 0.0010296010296010295,
      "grad_norm": 6.501317977905273,
      "learning_rate": 1.0296010296010297e-07,
      "loss": 0.7008,
      "step": 40
    },
    {
      "epoch": 0.001287001287001287,
      "grad_norm": 6.664251804351807,
      "learning_rate": 1.287001287001287e-07,
      "loss": 0.6545,
      "step": 50
    },
    {
      "epoch": 0.0015444015444015444,
      "grad_norm": 5.930378437042236,
      "learning_rate": 1.5444015444015445e-07,
      "loss": 0.6797,
      "step": 60
    },
    {
      "epoch": 0.0018018018018018018,
      "grad_norm": 8.338624954223633,
      "learning_rate": 1.801801801801802e-07,
      "loss": 0.6907,
      "step": 70
    },
    {
      "epoch": 0.002059202059202059,
      "grad_norm": 5.39778995513916,
      "learning_rate": 2.0592020592020594e-07,
      "loss": 0.6943,
      "step": 80
    },
    {
      "epoch": 0.0023166023166023165,
      "grad_norm": 6.76181173324585,
      "learning_rate": 2.3166023166023168e-07,
      "loss": 0.6661,
      "step": 90
    },
    {
      "epoch": 0.002574002574002574,
      "grad_norm": 6.561943054199219,
      "learning_rate": 2.574002574002574e-07,
      "loss": 0.6456,
      "step": 100
    },
    {
      "epoch": 0.0028314028314028314,
      "grad_norm": 9.402544975280762,
      "learning_rate": 2.8314028314028316e-07,
      "loss": 0.6365,
      "step": 110
    },
    {
      "epoch": 0.003088803088803089,
      "grad_norm": 6.95511531829834,
      "learning_rate": 3.088803088803089e-07,
      "loss": 0.6886,
      "step": 120
    },
    {
      "epoch": 0.0033462033462033462,
      "grad_norm": 12.216158866882324,
      "learning_rate": 3.3462033462033465e-07,
      "loss": 0.7131,
      "step": 130
    },
    {
      "epoch": 0.0036036036036036037,
      "grad_norm": 7.642661094665527,
      "learning_rate": 3.603603603603604e-07,
      "loss": 0.5935,
      "step": 140
    },
    {
      "epoch": 0.003861003861003861,
      "grad_norm": 6.88040018081665,
      "learning_rate": 3.8610038610038613e-07,
      "loss": 0.6159,
      "step": 150
    },
    {
      "epoch": 0.004118404118404118,
      "grad_norm": 7.177238941192627,
      "learning_rate": 4.1184041184041187e-07,
      "loss": 0.6105,
      "step": 160
    },
    {
      "epoch": 0.004375804375804376,
      "grad_norm": 7.751633644104004,
      "learning_rate": 4.375804375804376e-07,
      "loss": 0.5116,
      "step": 170
    },
    {
      "epoch": 0.004633204633204633,
      "grad_norm": 5.686230659484863,
      "learning_rate": 4.6332046332046336e-07,
      "loss": 0.7037,
      "step": 180
    },
    {
      "epoch": 0.004890604890604891,
      "grad_norm": 7.677026271820068,
      "learning_rate": 4.890604890604891e-07,
      "loss": 0.7483,
      "step": 190
    },
    {
      "epoch": 0.005148005148005148,
      "grad_norm": 9.224215507507324,
      "learning_rate": 5.148005148005148e-07,
      "loss": 0.6478,
      "step": 200
    },
    {
      "epoch": 0.005405405405405406,
      "grad_norm": 15.320013046264648,
      "learning_rate": 5.405405405405406e-07,
      "loss": 0.6244,
      "step": 210
    },
    {
      "epoch": 0.005662805662805663,
      "grad_norm": 4.737372875213623,
      "learning_rate": 5.662805662805663e-07,
      "loss": 0.5908,
      "step": 220
    },
    {
      "epoch": 0.005920205920205921,
      "grad_norm": 7.637610912322998,
      "learning_rate": 5.920205920205921e-07,
      "loss": 0.7024,
      "step": 230
    },
    {
      "epoch": 0.006177606177606178,
      "grad_norm": 60.20543670654297,
      "learning_rate": 6.177606177606178e-07,
      "loss": 0.6378,
      "step": 240
    },
    {
      "epoch": 0.006435006435006435,
      "grad_norm": 24.914642333984375,
      "learning_rate": 6.435006435006435e-07,
      "loss": 0.6039,
      "step": 250
    },
    {
      "epoch": 0.0066924066924066924,
      "grad_norm": 28.513675689697266,
      "learning_rate": 6.692406692406693e-07,
      "loss": 0.7308,
      "step": 260
    },
    {
      "epoch": 0.0069498069498069494,
      "grad_norm": 6.873077869415283,
      "learning_rate": 6.94980694980695e-07,
      "loss": 0.8579,
      "step": 270
    },
    {
      "epoch": 0.007207207207207207,
      "grad_norm": 7.60503625869751,
      "learning_rate": 7.207207207207208e-07,
      "loss": 0.7404,
      "step": 280
    },
    {
      "epoch": 0.007464607464607464,
      "grad_norm": 25.57830810546875,
      "learning_rate": 7.464607464607465e-07,
      "loss": 0.4125,
      "step": 290
    },
    {
      "epoch": 0.007722007722007722,
      "grad_norm": 5.907918930053711,
      "learning_rate": 7.722007722007723e-07,
      "loss": 0.7342,
      "step": 300
    },
    {
      "epoch": 0.00797940797940798,
      "grad_norm": 6.313519477844238,
      "learning_rate": 7.97940797940798e-07,
      "loss": 0.7723,
      "step": 310
    },
    {
      "epoch": 0.008236808236808236,
      "grad_norm": 30.61429214477539,
      "learning_rate": 8.236808236808237e-07,
      "loss": 0.831,
      "step": 320
    },
    {
      "epoch": 0.008494208494208495,
      "grad_norm": 5.046879768371582,
      "learning_rate": 8.494208494208495e-07,
      "loss": 0.5865,
      "step": 330
    },
    {
      "epoch": 0.008751608751608752,
      "grad_norm": 3.990790367126465,
      "learning_rate": 8.751608751608752e-07,
      "loss": 0.994,
      "step": 340
    },
    {
      "epoch": 0.009009009009009009,
      "grad_norm": 5.086954593658447,
      "learning_rate": 9.00900900900901e-07,
      "loss": 0.6809,
      "step": 350
    },
    {
      "epoch": 0.009266409266409266,
      "grad_norm": 28.828678131103516,
      "learning_rate": 9.266409266409267e-07,
      "loss": 0.8819,
      "step": 360
    },
    {
      "epoch": 0.009523809523809525,
      "grad_norm": 5.523406028747559,
      "learning_rate": 9.523809523809525e-07,
      "loss": 0.9103,
      "step": 370
    },
    {
      "epoch": 0.009781209781209782,
      "grad_norm": 80.04660034179688,
      "learning_rate": 9.781209781209782e-07,
      "loss": 1.1861,
      "step": 380
    },
    {
      "epoch": 0.010038610038610039,
      "grad_norm": 32.42625045776367,
      "learning_rate": 1.0038610038610038e-06,
      "loss": 0.5041,
      "step": 390
    },
    {
      "epoch": 0.010296010296010296,
      "grad_norm": 8.370193481445312,
      "learning_rate": 1.0296010296010297e-06,
      "loss": 1.3096,
      "step": 400
    },
    {
      "epoch": 0.010553410553410553,
      "grad_norm": 97.4281005859375,
      "learning_rate": 1.0553410553410555e-06,
      "loss": 0.8997,
      "step": 410
    },
    {
      "epoch": 0.010810810810810811,
      "grad_norm": 9.947352409362793,
      "learning_rate": 1.0810810810810812e-06,
      "loss": 1.1698,
      "step": 420
    },
    {
      "epoch": 0.011068211068211068,
      "grad_norm": 154.74952697753906,
      "learning_rate": 1.106821106821107e-06,
      "loss": 0.7863,
      "step": 430
    },
    {
      "epoch": 0.011325611325611325,
      "grad_norm": 8.229863166809082,
      "learning_rate": 1.1325611325611326e-06,
      "loss": 0.7175,
      "step": 440
    },
    {
      "epoch": 0.011583011583011582,
      "grad_norm": 7.129251956939697,
      "learning_rate": 1.1583011583011585e-06,
      "loss": 0.7535,
      "step": 450
    },
    {
      "epoch": 0.011840411840411841,
      "grad_norm": 6.870824337005615,
      "learning_rate": 1.1840411840411841e-06,
      "loss": 0.2594,
      "step": 460
    },
    {
      "epoch": 0.012097812097812098,
      "grad_norm": 40.79051971435547,
      "learning_rate": 1.20978120978121e-06,
      "loss": 0.7685,
      "step": 470
    },
    {
      "epoch": 0.012355212355212355,
      "grad_norm": 72.0915298461914,
      "learning_rate": 1.2355212355212356e-06,
      "loss": 0.9368,
      "step": 480
    },
    {
      "epoch": 0.012612612612612612,
      "grad_norm": 93.46292114257812,
      "learning_rate": 1.2612612612612613e-06,
      "loss": 0.8962,
      "step": 490
    },
    {
      "epoch": 0.01287001287001287,
      "grad_norm": 56.67453384399414,
      "learning_rate": 1.287001287001287e-06,
      "loss": 1.1966,
      "step": 500
    },
    {
      "epoch": 0.013127413127413128,
      "grad_norm": 3.2957065105438232,
      "learning_rate": 1.3127413127413127e-06,
      "loss": 0.7705,
      "step": 510
    },
    {
      "epoch": 0.013384813384813385,
      "grad_norm": 0.5807011723518372,
      "learning_rate": 1.3384813384813386e-06,
      "loss": 1.4432,
      "step": 520
    },
    {
      "epoch": 0.013642213642213642,
      "grad_norm": 2.716315746307373,
      "learning_rate": 1.3642213642213642e-06,
      "loss": 0.764,
      "step": 530
    },
    {
      "epoch": 0.013899613899613899,
      "grad_norm": 97.65186309814453,
      "learning_rate": 1.38996138996139e-06,
      "loss": 0.8481,
      "step": 540
    },
    {
      "epoch": 0.014157014157014158,
      "grad_norm": 0.27565112709999084,
      "learning_rate": 1.415701415701416e-06,
      "loss": 0.4895,
      "step": 550
    },
    {
      "epoch": 0.014414414414414415,
      "grad_norm": 0.5415809750556946,
      "learning_rate": 1.4414414414414416e-06,
      "loss": 0.8587,
      "step": 560
    },
    {
      "epoch": 0.014671814671814672,
      "grad_norm": 0.5196000933647156,
      "learning_rate": 1.4671814671814674e-06,
      "loss": 1.4641,
      "step": 570
    },
    {
      "epoch": 0.014929214929214929,
      "grad_norm": 59.97761917114258,
      "learning_rate": 1.492921492921493e-06,
      "loss": 2.4882,
      "step": 580
    },
    {
      "epoch": 0.015186615186615187,
      "grad_norm": 139.7994842529297,
      "learning_rate": 1.5186615186615189e-06,
      "loss": 1.3921,
      "step": 590
    },
    {
      "epoch": 0.015444015444015444,
      "grad_norm": 130.2068328857422,
      "learning_rate": 1.5444015444015445e-06,
      "loss": 2.1871,
      "step": 600
    },
    {
      "epoch": 0.015701415701415703,
      "grad_norm": 149.9769744873047,
      "learning_rate": 1.5701415701415704e-06,
      "loss": 1.6264,
      "step": 610
    },
    {
      "epoch": 0.01595881595881596,
      "grad_norm": 3.2275736331939697,
      "learning_rate": 1.595881595881596e-06,
      "loss": 1.2513,
      "step": 620
    },
    {
      "epoch": 0.016216216216216217,
      "grad_norm": 100.76615905761719,
      "learning_rate": 1.6216216216216219e-06,
      "loss": 2.0362,
      "step": 630
    },
    {
      "epoch": 0.016473616473616472,
      "grad_norm": 0.8999217748641968,
      "learning_rate": 1.6473616473616475e-06,
      "loss": 1.23,
      "step": 640
    },
    {
      "epoch": 0.01673101673101673,
      "grad_norm": 0.44757452607154846,
      "learning_rate": 1.6731016731016733e-06,
      "loss": 0.5529,
      "step": 650
    },
    {
      "epoch": 0.01698841698841699,
      "grad_norm": 0.33385616540908813,
      "learning_rate": 1.698841698841699e-06,
      "loss": 0.5156,
      "step": 660
    },
    {
      "epoch": 0.017245817245817245,
      "grad_norm": 0.26416975259780884,
      "learning_rate": 1.7245817245817248e-06,
      "loss": 1.5602,
      "step": 670
    },
    {
      "epoch": 0.017503217503217504,
      "grad_norm": 0.3138328790664673,
      "learning_rate": 1.7503217503217505e-06,
      "loss": 1.4557,
      "step": 680
    },
    {
      "epoch": 0.01776061776061776,
      "grad_norm": 52.74371337890625,
      "learning_rate": 1.7760617760617763e-06,
      "loss": 1.5494,
      "step": 690
    },
    {
      "epoch": 0.018018018018018018,
      "grad_norm": 50.65621566772461,
      "learning_rate": 1.801801801801802e-06,
      "loss": 1.0334,
      "step": 700
    },
    {
      "epoch": 0.018275418275418277,
      "grad_norm": 0.3436495363712311,
      "learning_rate": 1.8275418275418278e-06,
      "loss": 1.9683,
      "step": 710
    },
    {
      "epoch": 0.018532818532818532,
      "grad_norm": 0.81874018907547,
      "learning_rate": 1.8532818532818534e-06,
      "loss": 1.778,
      "step": 720
    },
    {
      "epoch": 0.01879021879021879,
      "grad_norm": 0.46802252531051636,
      "learning_rate": 1.8790218790218793e-06,
      "loss": 1.3622,
      "step": 730
    },
    {
      "epoch": 0.01904761904761905,
      "grad_norm": 63.82124328613281,
      "learning_rate": 1.904761904761905e-06,
      "loss": 2.0994,
      "step": 740
    },
    {
      "epoch": 0.019305019305019305,
      "grad_norm": 18.252229690551758,
      "learning_rate": 1.9305019305019305e-06,
      "loss": 1.2912,
      "step": 750
    },
    {
      "epoch": 0.019562419562419563,
      "grad_norm": 0.2647660970687866,
      "learning_rate": 1.9562419562419564e-06,
      "loss": 1.5519,
      "step": 760
    },
    {
      "epoch": 0.01981981981981982,
      "grad_norm": 0.2828594148159027,
      "learning_rate": 1.9819819819819822e-06,
      "loss": 1.0024,
      "step": 770
    },
    {
      "epoch": 0.02,
      "eval_accuracy": 0.45652173913043476,
      "eval_loss": 2.8390824794769287,
      "eval_runtime": 17.7944,
      "eval_samples_per_second": 2.585,
      "eval_steps_per_second": 2.585,
      "step": 777
    },
    {
      "epoch": 1.00007722007722,
      "grad_norm": 79.82716369628906,
      "learning_rate": 2.0077220077220077e-06,
      "loss": 0.7829,
      "step": 780
    },
    {
      "epoch": 1.0003346203346204,
      "grad_norm": 0.16030679643154144,
      "learning_rate": 2.0334620334620335e-06,
      "loss": 1.0346,
      "step": 790
    },
    {
      "epoch": 1.0005920205920207,
      "grad_norm": 57.7988166809082,
      "learning_rate": 2.0592020592020594e-06,
      "loss": 3.4014,
      "step": 800
    },
    {
      "epoch": 1.0008494208494207,
      "grad_norm": 0.7285271286964417,
      "learning_rate": 2.084942084942085e-06,
      "loss": 1.16,
      "step": 810
    },
    {
      "epoch": 1.001106821106821,
      "grad_norm": 0.50090491771698,
      "learning_rate": 2.110682110682111e-06,
      "loss": 1.1828,
      "step": 820
    },
    {
      "epoch": 1.0013642213642213,
      "grad_norm": 61.112510681152344,
      "learning_rate": 2.1364221364221365e-06,
      "loss": 1.7173,
      "step": 830
    },
    {
      "epoch": 1.0016216216216216,
      "grad_norm": 3.5773725509643555,
      "learning_rate": 2.1621621621621623e-06,
      "loss": 1.2737,
      "step": 840
    },
    {
      "epoch": 1.001879021879022,
      "grad_norm": 0.7043317556381226,
      "learning_rate": 2.187902187902188e-06,
      "loss": 1.804,
      "step": 850
    },
    {
      "epoch": 1.0021364221364222,
      "grad_norm": 0.6512853503227234,
      "learning_rate": 2.213642213642214e-06,
      "loss": 2.3513,
      "step": 860
    },
    {
      "epoch": 1.0023938223938225,
      "grad_norm": 41.23583221435547,
      "learning_rate": 2.2393822393822394e-06,
      "loss": 1.359,
      "step": 870
    },
    {
      "epoch": 1.0026512226512228,
      "grad_norm": 2.1360208988189697,
      "learning_rate": 2.2651222651222653e-06,
      "loss": 1.4268,
      "step": 880
    },
    {
      "epoch": 1.0029086229086228,
      "grad_norm": 0.1973283439874649,
      "learning_rate": 2.290862290862291e-06,
      "loss": 0.842,
      "step": 890
    },
    {
      "epoch": 1.003166023166023,
      "grad_norm": 80.92622375488281,
      "learning_rate": 2.316602316602317e-06,
      "loss": 1.9044,
      "step": 900
    },
    {
      "epoch": 1.0034234234234234,
      "grad_norm": 1.2733196020126343,
      "learning_rate": 2.3423423423423424e-06,
      "loss": 3.3045,
      "step": 910
    },
    {
      "epoch": 1.0036808236808237,
      "grad_norm": 79.40094757080078,
      "learning_rate": 2.3680823680823683e-06,
      "loss": 1.1075,
      "step": 920
    },
    {
      "epoch": 1.003938223938224,
      "grad_norm": 3.9844746589660645,
      "learning_rate": 2.393822393822394e-06,
      "loss": 1.1913,
      "step": 930
    },
    {
      "epoch": 1.0041956241956242,
      "grad_norm": 1.1007922887802124,
      "learning_rate": 2.41956241956242e-06,
      "loss": 0.9658,
      "step": 940
    },
    {
      "epoch": 1.0044530244530245,
      "grad_norm": 0.7749175429344177,
      "learning_rate": 2.4453024453024454e-06,
      "loss": 0.9729,
      "step": 950
    },
    {
      "epoch": 1.0047104247104248,
      "grad_norm": 58.21550369262695,
      "learning_rate": 2.4710424710424712e-06,
      "loss": 2.7515,
      "step": 960
    },
    {
      "epoch": 1.0049678249678249,
      "grad_norm": 0.8086639642715454,
      "learning_rate": 2.496782496782497e-06,
      "loss": 1.7731,
      "step": 970
    },
    {
      "epoch": 1.0052252252252252,
      "grad_norm": 0.3528954088687897,
      "learning_rate": 2.5225225225225225e-06,
      "loss": 1.5213,
      "step": 980
    },
    {
      "epoch": 1.0054826254826255,
      "grad_norm": 1.0501034259796143,
      "learning_rate": 2.5482625482625484e-06,
      "loss": 0.8826,
      "step": 990
    },
    {
      "epoch": 1.0057400257400257,
      "grad_norm": 64.3667984008789,
      "learning_rate": 2.574002574002574e-06,
      "loss": 2.4557,
      "step": 1000
    },
    {
      "epoch": 1.005997425997426,
      "grad_norm": 0.4949258267879486,
      "learning_rate": 2.5997425997426e-06,
      "loss": 1.8877,
      "step": 1010
    },
    {
      "epoch": 1.0062548262548263,
      "grad_norm": 44.34648895263672,
      "learning_rate": 2.6254826254826255e-06,
      "loss": 2.103,
      "step": 1020
    },
    {
      "epoch": 1.0065122265122266,
      "grad_norm": 1.014761209487915,
      "learning_rate": 2.6512226512226513e-06,
      "loss": 0.7794,
      "step": 1030
    },
    {
      "epoch": 1.0067696267696267,
      "grad_norm": 0.24838252365589142,
      "learning_rate": 2.676962676962677e-06,
      "loss": 0.0121,
      "step": 1040
    },
    {
      "epoch": 1.007027027027027,
      "grad_norm": 0.22434230148792267,
      "learning_rate": 2.702702702702703e-06,
      "loss": 1.5669,
      "step": 1050
    },
    {
      "epoch": 1.0072844272844272,
      "grad_norm": 0.5810171961784363,
      "learning_rate": 2.7284427284427284e-06,
      "loss": 1.016,
      "step": 1060
    },
    {
      "epoch": 1.0075418275418275,
      "grad_norm": 68.941650390625,
      "learning_rate": 2.7541827541827543e-06,
      "loss": 2.4654,
      "step": 1070
    },
    {
      "epoch": 1.0077992277992278,
      "grad_norm": 45.62387466430664,
      "learning_rate": 2.77992277992278e-06,
      "loss": 2.2378,
      "step": 1080
    },
    {
      "epoch": 1.008056628056628,
      "grad_norm": 51.002010345458984,
      "learning_rate": 2.805662805662806e-06,
      "loss": 1.5515,
      "step": 1090
    },
    {
      "epoch": 1.0083140283140284,
      "grad_norm": 1.32701575756073,
      "learning_rate": 2.831402831402832e-06,
      "loss": 1.6134,
      "step": 1100
    },
    {
      "epoch": 1.0085714285714287,
      "grad_norm": 3.001387357711792,
      "learning_rate": 2.8571428571428573e-06,
      "loss": 2.4355,
      "step": 1110
    },
    {
      "epoch": 1.0088288288288287,
      "grad_norm": 2.806532144546509,
      "learning_rate": 2.882882882882883e-06,
      "loss": 0.5442,
      "step": 1120
    },
    {
      "epoch": 1.009086229086229,
      "grad_norm": 33.64170455932617,
      "learning_rate": 2.908622908622909e-06,
      "loss": 1.8864,
      "step": 1130
    },
    {
      "epoch": 1.0093436293436293,
      "grad_norm": 46.05695724487305,
      "learning_rate": 2.934362934362935e-06,
      "loss": 1.3197,
      "step": 1140
    },
    {
      "epoch": 1.0096010296010296,
      "grad_norm": 0.5466468334197998,
      "learning_rate": 2.9601029601029602e-06,
      "loss": 0.8938,
      "step": 1150
    },
    {
      "epoch": 1.0098584298584299,
      "grad_norm": 0.12830115854740143,
      "learning_rate": 2.985842985842986e-06,
      "loss": 0.3857,
      "step": 1160
    },
    {
      "epoch": 1.0101158301158302,
      "grad_norm": 0.15344300866127014,
      "learning_rate": 3.011583011583012e-06,
      "loss": 0.5299,
      "step": 1170
    },
    {
      "epoch": 1.0103732303732305,
      "grad_norm": 0.11964306980371475,
      "learning_rate": 3.0373230373230378e-06,
      "loss": 0.0049,
      "step": 1180
    },
    {
      "epoch": 1.0106306306306305,
      "grad_norm": 46.51392364501953,
      "learning_rate": 3.063063063063063e-06,
      "loss": 3.7267,
      "step": 1190
    },
    {
      "epoch": 1.0108880308880308,
      "grad_norm": 1.6805133819580078,
      "learning_rate": 3.088803088803089e-06,
      "loss": 1.7001,
      "step": 1200
    },
    {
      "epoch": 1.011145431145431,
      "grad_norm": 47.035987854003906,
      "learning_rate": 3.114543114543115e-06,
      "loss": 1.5832,
      "step": 1210
    },
    {
      "epoch": 1.0114028314028314,
      "grad_norm": 0.3172648251056671,
      "learning_rate": 3.1402831402831407e-06,
      "loss": 1.4229,
      "step": 1220
    },
    {
      "epoch": 1.0116602316602317,
      "grad_norm": 0.3255147337913513,
      "learning_rate": 3.166023166023166e-06,
      "loss": 1.6876,
      "step": 1230
    },
    {
      "epoch": 1.011917631917632,
      "grad_norm": 1.1769987344741821,
      "learning_rate": 3.191763191763192e-06,
      "loss": 2.4022,
      "step": 1240
    },
    {
      "epoch": 1.0121750321750322,
      "grad_norm": 0.6336804628372192,
      "learning_rate": 3.217503217503218e-06,
      "loss": 0.6996,
      "step": 1250
    },
    {
      "epoch": 1.0124324324324325,
      "grad_norm": 47.805816650390625,
      "learning_rate": 3.2432432432432437e-06,
      "loss": 1.6999,
      "step": 1260
    },
    {
      "epoch": 1.0126898326898326,
      "grad_norm": 0.20486073195934296,
      "learning_rate": 3.268983268983269e-06,
      "loss": 0.9389,
      "step": 1270
    },
    {
      "epoch": 1.0129472329472329,
      "grad_norm": 39.7364387512207,
      "learning_rate": 3.294723294723295e-06,
      "loss": 1.3614,
      "step": 1280
    },
    {
      "epoch": 1.0132046332046332,
      "grad_norm": 0.8014540672302246,
      "learning_rate": 3.320463320463321e-06,
      "loss": 1.3432,
      "step": 1290
    },
    {
      "epoch": 1.0134620334620335,
      "grad_norm": 0.6335378289222717,
      "learning_rate": 3.3462033462033467e-06,
      "loss": 1.2641,
      "step": 1300
    },
    {
      "epoch": 1.0137194337194337,
      "grad_norm": 0.7638051509857178,
      "learning_rate": 3.371943371943372e-06,
      "loss": 1.7139,
      "step": 1310
    },
    {
      "epoch": 1.013976833976834,
      "grad_norm": 49.99736785888672,
      "learning_rate": 3.397683397683398e-06,
      "loss": 1.675,
      "step": 1320
    },
    {
      "epoch": 1.0142342342342343,
      "grad_norm": 59.96963882446289,
      "learning_rate": 3.423423423423424e-06,
      "loss": 2.2552,
      "step": 1330
    },
    {
      "epoch": 1.0144916344916346,
      "grad_norm": 0.4418865144252777,
      "learning_rate": 3.4491634491634496e-06,
      "loss": 1.1082,
      "step": 1340
    },
    {
      "epoch": 1.0147490347490347,
      "grad_norm": 1.3214060068130493,
      "learning_rate": 3.4749034749034755e-06,
      "loss": 1.2404,
      "step": 1350
    },
    {
      "epoch": 1.015006435006435,
      "grad_norm": 0.9876627922058105,
      "learning_rate": 3.500643500643501e-06,
      "loss": 1.9432,
      "step": 1360
    },
    {
      "epoch": 1.0152638352638352,
      "grad_norm": 50.07007598876953,
      "learning_rate": 3.5263835263835268e-06,
      "loss": 1.6856,
      "step": 1370
    },
    {
      "epoch": 1.0155212355212355,
      "grad_norm": 0.20070776343345642,
      "learning_rate": 3.5521235521235526e-06,
      "loss": 0.7285,
      "step": 1380
    },
    {
      "epoch": 1.0157786357786358,
      "grad_norm": 43.47827911376953,
      "learning_rate": 3.5778635778635785e-06,
      "loss": 2.393,
      "step": 1390
    },
    {
      "epoch": 1.016036036036036,
      "grad_norm": 51.43101501464844,
      "learning_rate": 3.603603603603604e-06,
      "loss": 1.4497,
      "step": 1400
    },
    {
      "epoch": 1.0162934362934364,
      "grad_norm": 40.5467529296875,
      "learning_rate": 3.6293436293436297e-06,
      "loss": 1.2371,
      "step": 1410
    },
    {
      "epoch": 1.0165508365508364,
      "grad_norm": 0.8846995830535889,
      "learning_rate": 3.6550836550836556e-06,
      "loss": 0.4215,
      "step": 1420
    },
    {
      "epoch": 1.0168082368082367,
      "grad_norm": 0.22001178562641144,
      "learning_rate": 3.6808236808236814e-06,
      "loss": 1.9439,
      "step": 1430
    },
    {
      "epoch": 1.017065637065637,
      "grad_norm": 156.90634155273438,
      "learning_rate": 3.706563706563707e-06,
      "loss": 2.5799,
      "step": 1440
    },
    {
      "epoch": 1.0173230373230373,
      "grad_norm": 2.8918557167053223,
      "learning_rate": 3.7323037323037327e-06,
      "loss": 0.479,
      "step": 1450
    },
    {
      "epoch": 1.0175804375804376,
      "grad_norm": 49.26845169067383,
      "learning_rate": 3.7580437580437585e-06,
      "loss": 1.8774,
      "step": 1460
    },
    {
      "epoch": 1.0178378378378379,
      "grad_norm": 0.13710249960422516,
      "learning_rate": 3.7837837837837844e-06,
      "loss": 0.4813,
      "step": 1470
    },
    {
      "epoch": 1.0180952380952382,
      "grad_norm": 0.31490692496299744,
      "learning_rate": 3.80952380952381e-06,
      "loss": 1.5386,
      "step": 1480
    },
    {
      "epoch": 1.0183526383526385,
      "grad_norm": 0.6874269247055054,
      "learning_rate": 3.835263835263835e-06,
      "loss": 0.4951,
      "step": 1490
    },
    {
      "epoch": 1.0186100386100385,
      "grad_norm": 0.10348591953516006,
      "learning_rate": 3.861003861003861e-06,
      "loss": 1.0332,
      "step": 1500
    },
    {
      "epoch": 1.0188674388674388,
      "grad_norm": 2.6218488216400146,
      "learning_rate": 3.886743886743887e-06,
      "loss": 2.4594,
      "step": 1510
    },
    {
      "epoch": 1.019124839124839,
      "grad_norm": 0.5357375741004944,
      "learning_rate": 3.912483912483913e-06,
      "loss": 0.864,
      "step": 1520
    },
    {
      "epoch": 1.0193822393822394,
      "grad_norm": 36.62129211425781,
      "learning_rate": 3.938223938223939e-06,
      "loss": 2.2063,
      "step": 1530
    },
    {
      "epoch": 1.0196396396396397,
      "grad_norm": 0.9374207854270935,
      "learning_rate": 3.9639639639639645e-06,
      "loss": 0.4099,
      "step": 1540
    },
    {
      "epoch": 1.01989703989704,
      "grad_norm": 44.0263671875,
      "learning_rate": 3.98970398970399e-06,
      "loss": 1.8322,
      "step": 1550
    },
    {
      "epoch": 1.02,
      "eval_accuracy": 0.45652173913043476,
      "eval_loss": 2.382408857345581,
      "eval_runtime": 16.4743,
      "eval_samples_per_second": 2.792,
      "eval_steps_per_second": 2.792,
      "step": 1554
    },
    {
      "epoch": 2.00015444015444,
      "grad_norm": 39.1295166015625,
      "learning_rate": 4.015444015444015e-06,
      "loss": 1.4183,
      "step": 1560
    },
    {
      "epoch": 2.0004118404118403,
      "grad_norm": 0.612335741519928,
      "learning_rate": 4.041184041184041e-06,
      "loss": 1.6604,
      "step": 1570
    },
    {
      "epoch": 2.000669240669241,
      "grad_norm": 0.7441545724868774,
      "learning_rate": 4.066924066924067e-06,
      "loss": 0.4295,
      "step": 1580
    },
    {
      "epoch": 2.000926640926641,
      "grad_norm": 43.12367248535156,
      "learning_rate": 4.092664092664093e-06,
      "loss": 1.8161,
      "step": 1590
    },
    {
      "epoch": 2.0011840411840414,
      "grad_norm": 81.20256805419922,
      "learning_rate": 4.118404118404119e-06,
      "loss": 0.9656,
      "step": 1600
    },
    {
      "epoch": 2.0014414414414414,
      "grad_norm": 0.589714765548706,
      "learning_rate": 4.1441441441441446e-06,
      "loss": 2.417,
      "step": 1610
    },
    {
      "epoch": 2.0016988416988415,
      "grad_norm": 2.8620784282684326,
      "learning_rate": 4.16988416988417e-06,
      "loss": 1.7992,
      "step": 1620
    },
    {
      "epoch": 2.001956241956242,
      "grad_norm": 28.455774307250977,
      "learning_rate": 4.195624195624196e-06,
      "loss": 1.4184,
      "step": 1630
    },
    {
      "epoch": 2.002213642213642,
      "grad_norm": 1.0315122604370117,
      "learning_rate": 4.221364221364222e-06,
      "loss": 1.5888,
      "step": 1640
    },
    {
      "epoch": 2.0024710424710426,
      "grad_norm": 0.40549129247665405,
      "learning_rate": 4.247104247104247e-06,
      "loss": 0.3045,
      "step": 1650
    },
    {
      "epoch": 2.0027284427284426,
      "grad_norm": 51.82611846923828,
      "learning_rate": 4.272844272844273e-06,
      "loss": 0.9216,
      "step": 1660
    },
    {
      "epoch": 2.002985842985843,
      "grad_norm": 55.43302536010742,
      "learning_rate": 4.298584298584299e-06,
      "loss": 1.9498,
      "step": 1670
    },
    {
      "epoch": 2.003243243243243,
      "grad_norm": 117.26138305664062,
      "learning_rate": 4.324324324324325e-06,
      "loss": 2.0472,
      "step": 1680
    },
    {
      "epoch": 2.0035006435006437,
      "grad_norm": 0.6769011616706848,
      "learning_rate": 4.3500643500643505e-06,
      "loss": 0.9034,
      "step": 1690
    },
    {
      "epoch": 2.003758043758044,
      "grad_norm": 2.4109325408935547,
      "learning_rate": 4.375804375804376e-06,
      "loss": 0.5953,
      "step": 1700
    },
    {
      "epoch": 2.004015444015444,
      "grad_norm": 41.65446472167969,
      "learning_rate": 4.401544401544402e-06,
      "loss": 2.514,
      "step": 1710
    },
    {
      "epoch": 2.0042728442728444,
      "grad_norm": 0.39666905999183655,
      "learning_rate": 4.427284427284428e-06,
      "loss": 1.5683,
      "step": 1720
    },
    {
      "epoch": 2.0045302445302444,
      "grad_norm": 1.444018006324768,
      "learning_rate": 4.453024453024453e-06,
      "loss": 0.508,
      "step": 1730
    },
    {
      "epoch": 2.004787644787645,
      "grad_norm": 0.13277959823608398,
      "learning_rate": 4.478764478764479e-06,
      "loss": 0.385,
      "step": 1740
    },
    {
      "epoch": 2.005045045045045,
      "grad_norm": 1.3261194229125977,
      "learning_rate": 4.504504504504505e-06,
      "loss": 1.6581,
      "step": 1750
    },
    {
      "epoch": 2.0053024453024455,
      "grad_norm": 81.54496002197266,
      "learning_rate": 4.530244530244531e-06,
      "loss": 3.6516,
      "step": 1760
    },
    {
      "epoch": 2.0055598455598456,
      "grad_norm": 84.0995864868164,
      "learning_rate": 4.5559845559845564e-06,
      "loss": 1.1699,
      "step": 1770
    },
    {
      "epoch": 2.0058172458172456,
      "grad_norm": 0.608340859413147,
      "learning_rate": 4.581724581724582e-06,
      "loss": 0.9071,
      "step": 1780
    },
    {
      "epoch": 2.006074646074646,
      "grad_norm": 0.5994428992271423,
      "learning_rate": 4.607464607464608e-06,
      "loss": 1.7902,
      "step": 1790
    },
    {
      "epoch": 2.006332046332046,
      "grad_norm": 0.7909229397773743,
      "learning_rate": 4.633204633204634e-06,
      "loss": 2.3984,
      "step": 1800
    },
    {
      "epoch": 2.0065894465894467,
      "grad_norm": 1.1406596899032593,
      "learning_rate": 4.658944658944659e-06,
      "loss": 0.9614,
      "step": 1810
    },
    {
      "epoch": 2.0068468468468468,
      "grad_norm": 0.39752158522605896,
      "learning_rate": 4.684684684684685e-06,
      "loss": 0.9369,
      "step": 1820
    },
    {
      "epoch": 2.0071042471042473,
      "grad_norm": 34.1922607421875,
      "learning_rate": 4.710424710424711e-06,
      "loss": 1.55,
      "step": 1830
    },
    {
      "epoch": 2.0073616473616473,
      "grad_norm": 34.1176643371582,
      "learning_rate": 4.7361647361647365e-06,
      "loss": 1.9668,
      "step": 1840
    },
    {
      "epoch": 2.0076190476190474,
      "grad_norm": 2.4576947689056396,
      "learning_rate": 4.761904761904762e-06,
      "loss": 1.8566,
      "step": 1850
    },
    {
      "epoch": 2.007876447876448,
      "grad_norm": 0.6945159435272217,
      "learning_rate": 4.787644787644788e-06,
      "loss": 0.7815,
      "step": 1860
    },
    {
      "epoch": 2.008133848133848,
      "grad_norm": 0.2894461154937744,
      "learning_rate": 4.813384813384814e-06,
      "loss": 1.2931,
      "step": 1870
    },
    {
      "epoch": 2.0083912483912485,
      "grad_norm": 2.950002431869507,
      "learning_rate": 4.83912483912484e-06,
      "loss": 1.9119,
      "step": 1880
    },
    {
      "epoch": 2.0086486486486486,
      "grad_norm": 1.8771700859069824,
      "learning_rate": 4.864864864864866e-06,
      "loss": 1.8962,
      "step": 1890
    },
    {
      "epoch": 2.008906048906049,
      "grad_norm": 30.816640853881836,
      "learning_rate": 4.890604890604891e-06,
      "loss": 0.8887,
      "step": 1900
    },
    {
      "epoch": 2.009163449163449,
      "grad_norm": 0.21323242783546448,
      "learning_rate": 4.916344916344917e-06,
      "loss": 0.9562,
      "step": 1910
    },
    {
      "epoch": 2.0094208494208496,
      "grad_norm": 0.3352386951446533,
      "learning_rate": 4.9420849420849425e-06,
      "loss": 1.3885,
      "step": 1920
    },
    {
      "epoch": 2.0096782496782497,
      "grad_norm": 0.2881447374820709,
      "learning_rate": 4.967824967824968e-06,
      "loss": 0.8771,
      "step": 1930
    },
    {
      "epoch": 2.0099356499356498,
      "grad_norm": 0.4859983026981354,
      "learning_rate": 4.993564993564994e-06,
      "loss": 2.1814,
      "step": 1940
    },
    {
      "epoch": 2.0101930501930503,
      "grad_norm": 0.3629135489463806,
      "learning_rate": 5.01930501930502e-06,
      "loss": 1.2579,
      "step": 1950
    },
    {
      "epoch": 2.0104504504504503,
      "grad_norm": 32.26611328125,
      "learning_rate": 5.045045045045045e-06,
      "loss": 1.1767,
      "step": 1960
    },
    {
      "epoch": 2.010707850707851,
      "grad_norm": 0.28480803966522217,
      "learning_rate": 5.070785070785072e-06,
      "loss": 0.4105,
      "step": 1970
    },
    {
      "epoch": 2.010965250965251,
      "grad_norm": 0.5369265079498291,
      "learning_rate": 5.096525096525097e-06,
      "loss": 1.3313,
      "step": 1980
    },
    {
      "epoch": 2.0112226512226514,
      "grad_norm": 0.269264817237854,
      "learning_rate": 5.122265122265123e-06,
      "loss": 0.9514,
      "step": 1990
    },
    {
      "epoch": 2.0114800514800515,
      "grad_norm": 0.19434118270874023,
      "learning_rate": 5.148005148005148e-06,
      "loss": 0.9533,
      "step": 2000
    },
    {
      "epoch": 2.0117374517374516,
      "grad_norm": 32.72032165527344,
      "learning_rate": 5.173745173745173e-06,
      "loss": 2.4204,
      "step": 2010
    },
    {
      "epoch": 2.011994851994852,
      "grad_norm": 1.2949801683425903,
      "learning_rate": 5.1994851994852e-06,
      "loss": 2.0395,
      "step": 2020
    },
    {
      "epoch": 2.012252252252252,
      "grad_norm": 32.453060150146484,
      "learning_rate": 5.225225225225226e-06,
      "loss": 1.7792,
      "step": 2030
    },
    {
      "epoch": 2.0125096525096526,
      "grad_norm": 33.515010833740234,
      "learning_rate": 5.250965250965251e-06,
      "loss": 1.3822,
      "step": 2040
    },
    {
      "epoch": 2.0127670527670527,
      "grad_norm": 1.7297950983047485,
      "learning_rate": 5.276705276705278e-06,
      "loss": 1.2671,
      "step": 2050
    },
    {
      "epoch": 2.013024453024453,
      "grad_norm": 44.402427673339844,
      "learning_rate": 5.302445302445303e-06,
      "loss": 1.9451,
      "step": 2060
    },
    {
      "epoch": 2.0132818532818533,
      "grad_norm": 44.23021697998047,
      "learning_rate": 5.328185328185329e-06,
      "loss": 1.3187,
      "step": 2070
    },
    {
      "epoch": 2.0135392535392533,
      "grad_norm": 0.1740846484899521,
      "learning_rate": 5.353925353925354e-06,
      "loss": 0.7822,
      "step": 2080
    },
    {
      "epoch": 2.013796653796654,
      "grad_norm": 42.054649353027344,
      "learning_rate": 5.379665379665379e-06,
      "loss": 1.4139,
      "step": 2090
    },
    {
      "epoch": 2.014054054054054,
      "grad_norm": 0.2841881215572357,
      "learning_rate": 5.405405405405406e-06,
      "loss": 1.3738,
      "step": 2100
    },
    {
      "epoch": 2.0143114543114544,
      "grad_norm": 0.4734525680541992,
      "learning_rate": 5.431145431145432e-06,
      "loss": 1.9621,
      "step": 2110
    },
    {
      "epoch": 2.0145688545688545,
      "grad_norm": 0.4707898795604706,
      "learning_rate": 5.456885456885457e-06,
      "loss": 0.6735,
      "step": 2120
    },
    {
      "epoch": 2.014826254826255,
      "grad_norm": 0.5470343232154846,
      "learning_rate": 5.4826254826254836e-06,
      "loss": 0.9792,
      "step": 2130
    },
    {
      "epoch": 2.015083655083655,
      "grad_norm": 0.32799121737480164,
      "learning_rate": 5.5083655083655086e-06,
      "loss": 1.867,
      "step": 2140
    },
    {
      "epoch": 2.015341055341055,
      "grad_norm": 0.09990080446004868,
      "learning_rate": 5.534105534105535e-06,
      "loss": 0.376,
      "step": 2150
    },
    {
      "epoch": 2.0155984555984556,
      "grad_norm": 39.74260711669922,
      "learning_rate": 5.55984555984556e-06,
      "loss": 2.8961,
      "step": 2160
    },
    {
      "epoch": 2.0158558558558557,
      "grad_norm": 2.921924352645874,
      "learning_rate": 5.585585585585585e-06,
      "loss": 0.3304,
      "step": 2170
    },
    {
      "epoch": 2.016113256113256,
      "grad_norm": 1.005566120147705,
      "learning_rate": 5.611325611325612e-06,
      "loss": 0.4781,
      "step": 2180
    },
    {
      "epoch": 2.0163706563706563,
      "grad_norm": 43.54568099975586,
      "learning_rate": 5.637065637065637e-06,
      "loss": 1.4212,
      "step": 2190
    },
    {
      "epoch": 2.0166280566280568,
      "grad_norm": 0.2699664235115051,
      "learning_rate": 5.662805662805664e-06,
      "loss": 2.2329,
      "step": 2200
    },
    {
      "epoch": 2.016885456885457,
      "grad_norm": 1.8670907020568848,
      "learning_rate": 5.6885456885456895e-06,
      "loss": 1.4189,
      "step": 2210
    },
    {
      "epoch": 2.0171428571428573,
      "grad_norm": 0.6736155152320862,
      "learning_rate": 5.7142857142857145e-06,
      "loss": 1.6321,
      "step": 2220
    },
    {
      "epoch": 2.0174002574002574,
      "grad_norm": 0.12274114787578583,
      "learning_rate": 5.740025740025741e-06,
      "loss": 1.6314,
      "step": 2230
    },
    {
      "epoch": 2.0176576576576575,
      "grad_norm": 1.2040308713912964,
      "learning_rate": 5.765765765765766e-06,
      "loss": 0.6115,
      "step": 2240
    },
    {
      "epoch": 2.017915057915058,
      "grad_norm": 2.5511398315429688,
      "learning_rate": 5.791505791505791e-06,
      "loss": 0.7289,
      "step": 2250
    },
    {
      "epoch": 2.018172458172458,
      "grad_norm": 0.15484865009784698,
      "learning_rate": 5.817245817245818e-06,
      "loss": 1.058,
      "step": 2260
    },
    {
      "epoch": 2.0184298584298586,
      "grad_norm": 40.67464828491211,
      "learning_rate": 5.842985842985843e-06,
      "loss": 1.3452,
      "step": 2270
    },
    {
      "epoch": 2.0186872586872586,
      "grad_norm": 0.10303802043199539,
      "learning_rate": 5.86872586872587e-06,
      "loss": 1.1189,
      "step": 2280
    },
    {
      "epoch": 2.018944658944659,
      "grad_norm": 1.1726830005645752,
      "learning_rate": 5.894465894465895e-06,
      "loss": 2.4109,
      "step": 2290
    },
    {
      "epoch": 2.019202059202059,
      "grad_norm": 0.34925177693367004,
      "learning_rate": 5.9202059202059204e-06,
      "loss": 0.6467,
      "step": 2300
    },
    {
      "epoch": 2.0194594594594593,
      "grad_norm": 36.52827072143555,
      "learning_rate": 5.945945945945947e-06,
      "loss": 1.2837,
      "step": 2310
    },
    {
      "epoch": 2.0197168597168598,
      "grad_norm": 58.13482666015625,
      "learning_rate": 5.971685971685972e-06,
      "loss": 2.3565,
      "step": 2320
    },
    {
      "epoch": 2.01997425997426,
      "grad_norm": 0.45640629529953003,
      "learning_rate": 5.997425997425997e-06,
      "loss": 0.8493,
      "step": 2330
    },
    {
      "epoch": 2.02,
      "eval_accuracy": 0.45652173913043476,
      "eval_loss": 1.1137661933898926,
      "eval_runtime": 16.6817,
      "eval_samples_per_second": 2.758,
      "eval_steps_per_second": 2.758,
      "step": 2331
    },
    {
      "epoch": 3.0002316602316603,
      "grad_norm": 33.54939651489258,
      "learning_rate": 6.023166023166024e-06,
      "loss": 0.4011,
      "step": 2340
    },
    {
      "epoch": 3.0004890604890604,
      "grad_norm": 3.8214738368988037,
      "learning_rate": 6.048906048906049e-06,
      "loss": 1.5293,
      "step": 2350
    },
    {
      "epoch": 3.000746460746461,
      "grad_norm": 4.559745788574219,
      "learning_rate": 6.0746460746460755e-06,
      "loss": 0.6424,
      "step": 2360
    },
    {
      "epoch": 3.001003861003861,
      "grad_norm": 0.11879268288612366,
      "learning_rate": 6.1003861003861005e-06,
      "loss": 0.7711,
      "step": 2370
    },
    {
      "epoch": 3.0012612612612615,
      "grad_norm": 62.36249542236328,
      "learning_rate": 6.126126126126126e-06,
      "loss": 1.7833,
      "step": 2380
    },
    {
      "epoch": 3.0015186615186615,
      "grad_norm": 15.011298179626465,
      "learning_rate": 6.151866151866153e-06,
      "loss": 0.8435,
      "step": 2390
    },
    {
      "epoch": 3.0017760617760616,
      "grad_norm": 21.581642150878906,
      "learning_rate": 6.177606177606178e-06,
      "loss": 0.5523,
      "step": 2400
    },
    {
      "epoch": 3.002033462033462,
      "grad_norm": 29.18716049194336,
      "learning_rate": 6.203346203346203e-06,
      "loss": 1.2024,
      "step": 2410
    },
    {
      "epoch": 3.002290862290862,
      "grad_norm": 0.08250798285007477,
      "learning_rate": 6.22908622908623e-06,
      "loss": 2.0342,
      "step": 2420
    },
    {
      "epoch": 3.0025482625482627,
      "grad_norm": 0.7816446423530579,
      "learning_rate": 6.254826254826255e-06,
      "loss": 1.2861,
      "step": 2430
    },
    {
      "epoch": 3.0028056628056627,
      "grad_norm": 70.71495056152344,
      "learning_rate": 6.2805662805662815e-06,
      "loss": 1.8372,
      "step": 2440
    },
    {
      "epoch": 3.0030630630630633,
      "grad_norm": 57.512603759765625,
      "learning_rate": 6.3063063063063065e-06,
      "loss": 1.9071,
      "step": 2450
    },
    {
      "epoch": 3.0033204633204633,
      "grad_norm": 1.428179144859314,
      "learning_rate": 6.332046332046332e-06,
      "loss": 1.127,
      "step": 2460
    },
    {
      "epoch": 3.0035778635778634,
      "grad_norm": 124.6517333984375,
      "learning_rate": 6.357786357786358e-06,
      "loss": 1.237,
      "step": 2470
    },
    {
      "epoch": 3.003835263835264,
      "grad_norm": 2.761348247528076,
      "learning_rate": 6.383526383526384e-06,
      "loss": 0.9261,
      "step": 2480
    },
    {
      "epoch": 3.004092664092664,
      "grad_norm": 0.9238646626472473,
      "learning_rate": 6.409266409266411e-06,
      "loss": 1.2299,
      "step": 2490
    },
    {
      "epoch": 3.0043500643500645,
      "grad_norm": 128.46592712402344,
      "learning_rate": 6.435006435006436e-06,
      "loss": 1.2574,
      "step": 2500
    },
    {
      "epoch": 3.0046074646074645,
      "grad_norm": 0.5994557738304138,
      "learning_rate": 6.460746460746461e-06,
      "loss": 1.0424,
      "step": 2510
    },
    {
      "epoch": 3.004864864864865,
      "grad_norm": 0.7990912795066833,
      "learning_rate": 6.486486486486487e-06,
      "loss": 1.3062,
      "step": 2520
    },
    {
      "epoch": 3.005122265122265,
      "grad_norm": 48.199005126953125,
      "learning_rate": 6.512226512226512e-06,
      "loss": 2.3465,
      "step": 2530
    },
    {
      "epoch": 3.005379665379665,
      "grad_norm": 1.4259648323059082,
      "learning_rate": 6.537966537966538e-06,
      "loss": 1.165,
      "step": 2540
    },
    {
      "epoch": 3.0056370656370657,
      "grad_norm": 0.08506966382265091,
      "learning_rate": 6.563706563706564e-06,
      "loss": 0.0076,
      "step": 2550
    },
    {
      "epoch": 3.0058944658944657,
      "grad_norm": 0.06992141902446747,
      "learning_rate": 6.58944658944659e-06,
      "loss": 1.07,
      "step": 2560
    },
    {
      "epoch": 3.0061518661518662,
      "grad_norm": 0.048462025821208954,
      "learning_rate": 6.615186615186616e-06,
      "loss": 0.9731,
      "step": 2570
    },
    {
      "epoch": 3.0064092664092663,
      "grad_norm": 0.43919050693511963,
      "learning_rate": 6.640926640926642e-06,
      "loss": 1.4364,
      "step": 2580
    },
    {
      "epoch": 3.006666666666667,
      "grad_norm": 0.037599463015794754,
      "learning_rate": 6.666666666666667e-06,
      "loss": 0.698,
      "step": 2590
    },
    {
      "epoch": 3.006924066924067,
      "grad_norm": 0.03720109537243843,
      "learning_rate": 6.692406692406693e-06,
      "loss": 0.0058,
      "step": 2600
    },
    {
      "epoch": 3.0071814671814674,
      "grad_norm": 0.0219118595123291,
      "learning_rate": 6.718146718146718e-06,
      "loss": 0.5981,
      "step": 2610
    },
    {
      "epoch": 3.0074388674388675,
      "grad_norm": 46.0898551940918,
      "learning_rate": 6.743886743886744e-06,
      "loss": 1.8813,
      "step": 2620
    },
    {
      "epoch": 3.0076962676962675,
      "grad_norm": 84.8902359008789,
      "learning_rate": 6.76962676962677e-06,
      "loss": 2.6947,
      "step": 2630
    },
    {
      "epoch": 3.007953667953668,
      "grad_norm": 0.8112554550170898,
      "learning_rate": 6.795366795366796e-06,
      "loss": 1.4373,
      "step": 2640
    },
    {
      "epoch": 3.008211068211068,
      "grad_norm": 4.990564823150635,
      "learning_rate": 6.821106821106822e-06,
      "loss": 1.9307,
      "step": 2650
    },
    {
      "epoch": 3.0084684684684686,
      "grad_norm": 25.768632888793945,
      "learning_rate": 6.846846846846848e-06,
      "loss": 1.3059,
      "step": 2660
    },
    {
      "epoch": 3.0087258687258687,
      "grad_norm": 54.58412170410156,
      "learning_rate": 6.872586872586873e-06,
      "loss": 0.9245,
      "step": 2670
    },
    {
      "epoch": 3.008983268983269,
      "grad_norm": 47.7413330078125,
      "learning_rate": 6.898326898326899e-06,
      "loss": 1.2352,
      "step": 2680
    },
    {
      "epoch": 3.0092406692406692,
      "grad_norm": 1.7431989908218384,
      "learning_rate": 6.924066924066924e-06,
      "loss": 1.2804,
      "step": 2690
    },
    {
      "epoch": 3.0094980694980693,
      "grad_norm": 0.4777579605579376,
      "learning_rate": 6.949806949806951e-06,
      "loss": 0.7333,
      "step": 2700
    },
    {
      "epoch": 3.00975546975547,
      "grad_norm": 56.5815544128418,
      "learning_rate": 6.975546975546976e-06,
      "loss": 1.6311,
      "step": 2710
    },
    {
      "epoch": 3.01001287001287,
      "grad_norm": 84.84012603759766,
      "learning_rate": 7.001287001287002e-06,
      "loss": 3.0436,
      "step": 2720
    },
    {
      "epoch": 3.0102702702702704,
      "grad_norm": 2.14959716796875,
      "learning_rate": 7.027027027027028e-06,
      "loss": 0.6488,
      "step": 2730
    },
    {
      "epoch": 3.0105276705276705,
      "grad_norm": 0.05591985583305359,
      "learning_rate": 7.0527670527670535e-06,
      "loss": 0.8272,
      "step": 2740
    },
    {
      "epoch": 3.010785070785071,
      "grad_norm": 0.3322344422340393,
      "learning_rate": 7.0785070785070785e-06,
      "loss": 0.5012,
      "step": 2750
    },
    {
      "epoch": 3.011042471042471,
      "grad_norm": 0.042405061423778534,
      "learning_rate": 7.104247104247105e-06,
      "loss": 1.7783,
      "step": 2760
    },
    {
      "epoch": 3.011299871299871,
      "grad_norm": 0.10109010338783264,
      "learning_rate": 7.12998712998713e-06,
      "loss": 1.5399,
      "step": 2770
    },
    {
      "epoch": 3.0115572715572716,
      "grad_norm": 1.0357915163040161,
      "learning_rate": 7.155727155727157e-06,
      "loss": 1.998,
      "step": 2780
    },
    {
      "epoch": 3.0118146718146717,
      "grad_norm": 3.6725423336029053,
      "learning_rate": 7.181467181467182e-06,
      "loss": 1.5494,
      "step": 2790
    },
    {
      "epoch": 3.012072072072072,
      "grad_norm": 0.21130582690238953,
      "learning_rate": 7.207207207207208e-06,
      "loss": 0.8854,
      "step": 2800
    },
    {
      "epoch": 3.0123294723294722,
      "grad_norm": 80.49689483642578,
      "learning_rate": 7.232947232947234e-06,
      "loss": 0.7072,
      "step": 2810
    },
    {
      "epoch": 3.0125868725868727,
      "grad_norm": 0.08989892154932022,
      "learning_rate": 7.2586872586872595e-06,
      "loss": 1.1194,
      "step": 2820
    },
    {
      "epoch": 3.012844272844273,
      "grad_norm": 2.3026134967803955,
      "learning_rate": 7.2844272844272845e-06,
      "loss": 1.0552,
      "step": 2830
    },
    {
      "epoch": 3.0131016731016733,
      "grad_norm": 26.29865264892578,
      "learning_rate": 7.310167310167311e-06,
      "loss": 0.5594,
      "step": 2840
    },
    {
      "epoch": 3.0133590733590734,
      "grad_norm": 22.02301597595215,
      "learning_rate": 7.335907335907336e-06,
      "loss": 0.4249,
      "step": 2850
    },
    {
      "epoch": 3.0136164736164734,
      "grad_norm": 118.84080505371094,
      "learning_rate": 7.361647361647363e-06,
      "loss": 0.9131,
      "step": 2860
    },
    {
      "epoch": 3.013873873873874,
      "grad_norm": 109.1220474243164,
      "learning_rate": 7.387387387387388e-06,
      "loss": 1.1904,
      "step": 2870
    },
    {
      "epoch": 3.014131274131274,
      "grad_norm": 9.534757614135742,
      "learning_rate": 7.413127413127414e-06,
      "loss": 1.3307,
      "step": 2880
    },
    {
      "epoch": 3.0143886743886745,
      "grad_norm": 108.12349700927734,
      "learning_rate": 7.4388674388674395e-06,
      "loss": 0.3219,
      "step": 2890
    },
    {
      "epoch": 3.0146460746460746,
      "grad_norm": 3.933692693710327,
      "learning_rate": 7.464607464607465e-06,
      "loss": 0.9342,
      "step": 2900
    },
    {
      "epoch": 3.014903474903475,
      "grad_norm": 164.84881591796875,
      "learning_rate": 7.49034749034749e-06,
      "loss": 1.1618,
      "step": 2910
    },
    {
      "epoch": 3.015160875160875,
      "grad_norm": 0.16784685850143433,
      "learning_rate": 7.516087516087517e-06,
      "loss": 0.9138,
      "step": 2920
    },
    {
      "epoch": 3.0154182754182752,
      "grad_norm": 0.22646842896938324,
      "learning_rate": 7.541827541827542e-06,
      "loss": 0.6134,
      "step": 2930
    },
    {
      "epoch": 3.0156756756756757,
      "grad_norm": 18.39739227294922,
      "learning_rate": 7.567567567567569e-06,
      "loss": 0.2646,
      "step": 2940
    },
    {
      "epoch": 3.015933075933076,
      "grad_norm": 0.6532003283500671,
      "learning_rate": 7.593307593307594e-06,
      "loss": 0.7952,
      "step": 2950
    },
    {
      "epoch": 3.0161904761904763,
      "grad_norm": 1.6719454526901245,
      "learning_rate": 7.61904761904762e-06,
      "loss": 2.2333,
      "step": 2960
    },
    {
      "epoch": 3.0164478764478764,
      "grad_norm": 0.13571570813655853,
      "learning_rate": 7.644787644787645e-06,
      "loss": 1.1252,
      "step": 2970
    },
    {
      "epoch": 3.016705276705277,
      "grad_norm": 0.48596158623695374,
      "learning_rate": 7.67052767052767e-06,
      "loss": 1.1305,
      "step": 2980
    },
    {
      "epoch": 3.016962676962677,
      "grad_norm": 2.5286593437194824,
      "learning_rate": 7.696267696267697e-06,
      "loss": 1.0435,
      "step": 2990
    },
    {
      "epoch": 3.017220077220077,
      "grad_norm": 40.78852081298828,
      "learning_rate": 7.722007722007722e-06,
      "loss": 1.9916,
      "step": 3000
    },
    {
      "epoch": 3.0174774774774775,
      "grad_norm": 0.17999140918254852,
      "learning_rate": 7.747747747747749e-06,
      "loss": 0.9862,
      "step": 3010
    },
    {
      "epoch": 3.0177348777348776,
      "grad_norm": 2.1499030590057373,
      "learning_rate": 7.773487773487774e-06,
      "loss": 1.7006,
      "step": 3020
    },
    {
      "epoch": 3.017992277992278,
      "grad_norm": 0.820077121257782,
      "learning_rate": 7.7992277992278e-06,
      "loss": 1.6318,
      "step": 3030
    },
    {
      "epoch": 3.018249678249678,
      "grad_norm": 0.7950751781463623,
      "learning_rate": 7.824967824967826e-06,
      "loss": 1.7085,
      "step": 3040
    },
    {
      "epoch": 3.0185070785070787,
      "grad_norm": 65.81083679199219,
      "learning_rate": 7.850707850707852e-06,
      "loss": 0.9623,
      "step": 3050
    },
    {
      "epoch": 3.0187644787644787,
      "grad_norm": 0.8133057355880737,
      "learning_rate": 7.876447876447877e-06,
      "loss": 2.0652,
      "step": 3060
    },
    {
      "epoch": 3.0190218790218792,
      "grad_norm": 0.6378155946731567,
      "learning_rate": 7.902187902187904e-06,
      "loss": 0.3423,
      "step": 3070
    },
    {
      "epoch": 3.0192792792792793,
      "grad_norm": 1.3878402709960938,
      "learning_rate": 7.927927927927929e-06,
      "loss": 0.751,
      "step": 3080
    },
    {
      "epoch": 3.0195366795366794,
      "grad_norm": 0.09455589950084686,
      "learning_rate": 7.953667953667954e-06,
      "loss": 2.3321,
      "step": 3090
    },
    {
      "epoch": 3.01979407979408,
      "grad_norm": 0.12332921475172043,
      "learning_rate": 7.97940797940798e-06,
      "loss": 0.4768,
      "step": 3100
    },
    {
      "epoch": 3.02,
      "eval_accuracy": 0.5,
      "eval_loss": 1.7404745817184448,
      "eval_runtime": 15.4534,
      "eval_samples_per_second": 2.977,
      "eval_steps_per_second": 2.977,
      "step": 3108
    },
    {
      "epoch": 4.00005148005148,
      "grad_norm": 0.7716472744941711,
      "learning_rate": 8.005148005148006e-06,
      "loss": 0.7142,
      "step": 3110
    },
    {
      "epoch": 4.00030888030888,
      "grad_norm": 136.4551239013672,
      "learning_rate": 8.03088803088803e-06,
      "loss": 1.2605,
      "step": 3120
    },
    {
      "epoch": 4.0005662805662805,
      "grad_norm": 0.08112074434757233,
      "learning_rate": 8.056628056628057e-06,
      "loss": 0.3466,
      "step": 3130
    },
    {
      "epoch": 4.000823680823681,
      "grad_norm": 0.0227906946092844,
      "learning_rate": 8.082368082368082e-06,
      "loss": 0.0025,
      "step": 3140
    },
    {
      "epoch": 4.0010810810810815,
      "grad_norm": 0.3076695203781128,
      "learning_rate": 8.108108108108109e-06,
      "loss": 2.6437,
      "step": 3150
    },
    {
      "epoch": 4.001338481338482,
      "grad_norm": 31.70879554748535,
      "learning_rate": 8.133848133848134e-06,
      "loss": 1.7806,
      "step": 3160
    },
    {
      "epoch": 4.001595881595882,
      "grad_norm": 0.3824242949485779,
      "learning_rate": 8.159588159588159e-06,
      "loss": 0.9026,
      "step": 3170
    },
    {
      "epoch": 4.001853281853282,
      "grad_norm": 1.8017970323562622,
      "learning_rate": 8.185328185328186e-06,
      "loss": 1.5216,
      "step": 3180
    },
    {
      "epoch": 4.002110682110682,
      "grad_norm": 0.25629478693008423,
      "learning_rate": 8.211068211068212e-06,
      "loss": 0.3092,
      "step": 3190
    },
    {
      "epoch": 4.002368082368083,
      "grad_norm": 0.0877305120229721,
      "learning_rate": 8.236808236808237e-06,
      "loss": 1.7358,
      "step": 3200
    },
    {
      "epoch": 4.002625482625483,
      "grad_norm": 87.11286926269531,
      "learning_rate": 8.262548262548264e-06,
      "loss": 1.9197,
      "step": 3210
    },
    {
      "epoch": 4.002882882882883,
      "grad_norm": 52.01739501953125,
      "learning_rate": 8.288288288288289e-06,
      "loss": 1.8059,
      "step": 3220
    },
    {
      "epoch": 4.003140283140283,
      "grad_norm": 1.063254952430725,
      "learning_rate": 8.314028314028316e-06,
      "loss": 0.2557,
      "step": 3230
    },
    {
      "epoch": 4.003397683397683,
      "grad_norm": 33.61056137084961,
      "learning_rate": 8.33976833976834e-06,
      "loss": 2.4951,
      "step": 3240
    },
    {
      "epoch": 4.003655083655084,
      "grad_norm": 40.32854461669922,
      "learning_rate": 8.365508365508366e-06,
      "loss": 1.8026,
      "step": 3250
    },
    {
      "epoch": 4.003912483912484,
      "grad_norm": 1.654442548751831,
      "learning_rate": 8.391248391248393e-06,
      "loss": 0.595,
      "step": 3260
    },
    {
      "epoch": 4.004169884169884,
      "grad_norm": 7.549476146697998,
      "learning_rate": 8.416988416988418e-06,
      "loss": 0.9348,
      "step": 3270
    },
    {
      "epoch": 4.004427284427284,
      "grad_norm": 0.09671440720558167,
      "learning_rate": 8.442728442728444e-06,
      "loss": 0.4664,
      "step": 3280
    },
    {
      "epoch": 4.004684684684685,
      "grad_norm": 0.10881602019071579,
      "learning_rate": 8.46846846846847e-06,
      "loss": 1.231,
      "step": 3290
    },
    {
      "epoch": 4.004942084942085,
      "grad_norm": 0.24008144438266754,
      "learning_rate": 8.494208494208494e-06,
      "loss": 0.7618,
      "step": 3300
    },
    {
      "epoch": 4.005199485199485,
      "grad_norm": 42.40687561035156,
      "learning_rate": 8.519948519948521e-06,
      "loss": 2.8717,
      "step": 3310
    },
    {
      "epoch": 4.005456885456885,
      "grad_norm": 59.698265075683594,
      "learning_rate": 8.545688545688546e-06,
      "loss": 1.3768,
      "step": 3320
    },
    {
      "epoch": 4.005714285714285,
      "grad_norm": 30.52323341369629,
      "learning_rate": 8.571428571428571e-06,
      "loss": 0.449,
      "step": 3330
    },
    {
      "epoch": 4.005971685971686,
      "grad_norm": 0.7098885774612427,
      "learning_rate": 8.597168597168598e-06,
      "loss": 0.3891,
      "step": 3340
    },
    {
      "epoch": 4.006229086229086,
      "grad_norm": 0.10432929545640945,
      "learning_rate": 8.622908622908623e-06,
      "loss": 1.038,
      "step": 3350
    },
    {
      "epoch": 4.006486486486486,
      "grad_norm": 35.44326400756836,
      "learning_rate": 8.64864864864865e-06,
      "loss": 2.8441,
      "step": 3360
    },
    {
      "epoch": 4.0067438867438865,
      "grad_norm": 0.09243518859148026,
      "learning_rate": 8.674388674388674e-06,
      "loss": 0.5229,
      "step": 3370
    },
    {
      "epoch": 4.007001287001287,
      "grad_norm": 59.27909469604492,
      "learning_rate": 8.700128700128701e-06,
      "loss": 1.4812,
      "step": 3380
    },
    {
      "epoch": 4.0072586872586875,
      "grad_norm": 0.09579314291477203,
      "learning_rate": 8.725868725868728e-06,
      "loss": 0.9448,
      "step": 3390
    },
    {
      "epoch": 4.007516087516088,
      "grad_norm": 78.7119140625,
      "learning_rate": 8.751608751608753e-06,
      "loss": 1.3743,
      "step": 3400
    },
    {
      "epoch": 4.007773487773488,
      "grad_norm": 17.389902114868164,
      "learning_rate": 8.777348777348778e-06,
      "loss": 0.3665,
      "step": 3410
    },
    {
      "epoch": 4.008030888030888,
      "grad_norm": 0.03131213039159775,
      "learning_rate": 8.803088803088804e-06,
      "loss": 0.95,
      "step": 3420
    },
    {
      "epoch": 4.008288288288289,
      "grad_norm": 0.07040321081876755,
      "learning_rate": 8.82882882882883e-06,
      "loss": 1.618,
      "step": 3430
    },
    {
      "epoch": 4.008545688545689,
      "grad_norm": 0.0936579629778862,
      "learning_rate": 8.854568854568856e-06,
      "loss": 1.8872,
      "step": 3440
    },
    {
      "epoch": 4.008803088803089,
      "grad_norm": 0.3549942970275879,
      "learning_rate": 8.880308880308881e-06,
      "loss": 0.8224,
      "step": 3450
    },
    {
      "epoch": 4.009060489060489,
      "grad_norm": 51.08911895751953,
      "learning_rate": 8.906048906048906e-06,
      "loss": 0.9038,
      "step": 3460
    },
    {
      "epoch": 4.009317889317889,
      "grad_norm": 96.68467712402344,
      "learning_rate": 8.931788931788933e-06,
      "loss": 1.0418,
      "step": 3470
    },
    {
      "epoch": 4.00957528957529,
      "grad_norm": 2.3784525394439697,
      "learning_rate": 8.957528957528958e-06,
      "loss": 0.4836,
      "step": 3480
    },
    {
      "epoch": 4.00983268983269,
      "grad_norm": 0.043230555951595306,
      "learning_rate": 8.983268983268984e-06,
      "loss": 2.1446,
      "step": 3490
    },
    {
      "epoch": 4.01009009009009,
      "grad_norm": 0.10071835666894913,
      "learning_rate": 9.00900900900901e-06,
      "loss": 0.8607,
      "step": 3500
    },
    {
      "epoch": 4.01034749034749,
      "grad_norm": 151.3938751220703,
      "learning_rate": 9.034749034749034e-06,
      "loss": 1.0037,
      "step": 3510
    },
    {
      "epoch": 4.010604890604891,
      "grad_norm": 1.0010333061218262,
      "learning_rate": 9.060489060489061e-06,
      "loss": 1.6101,
      "step": 3520
    },
    {
      "epoch": 4.010862290862291,
      "grad_norm": 139.9682159423828,
      "learning_rate": 9.086229086229086e-06,
      "loss": 0.8691,
      "step": 3530
    },
    {
      "epoch": 4.011119691119691,
      "grad_norm": 50.917869567871094,
      "learning_rate": 9.111969111969113e-06,
      "loss": 1.0019,
      "step": 3540
    },
    {
      "epoch": 4.011377091377091,
      "grad_norm": 31.801929473876953,
      "learning_rate": 9.137709137709138e-06,
      "loss": 1.0467,
      "step": 3550
    },
    {
      "epoch": 4.011634491634491,
      "grad_norm": 1.3450016975402832,
      "learning_rate": 9.163449163449165e-06,
      "loss": 0.5833,
      "step": 3560
    },
    {
      "epoch": 4.011891891891892,
      "grad_norm": 0.3178759217262268,
      "learning_rate": 9.189189189189191e-06,
      "loss": 0.2232,
      "step": 3570
    },
    {
      "epoch": 4.012149292149292,
      "grad_norm": 85.76224517822266,
      "learning_rate": 9.214929214929216e-06,
      "loss": 0.9149,
      "step": 3580
    },
    {
      "epoch": 4.012406692406692,
      "grad_norm": 0.0713811069726944,
      "learning_rate": 9.240669240669241e-06,
      "loss": 1.2048,
      "step": 3590
    },
    {
      "epoch": 4.012664092664092,
      "grad_norm": 261.92388916015625,
      "learning_rate": 9.266409266409268e-06,
      "loss": 0.9989,
      "step": 3600
    },
    {
      "epoch": 4.012921492921493,
      "grad_norm": 0.05194203555583954,
      "learning_rate": 9.292149292149293e-06,
      "loss": 1.0579,
      "step": 3610
    },
    {
      "epoch": 4.013178893178893,
      "grad_norm": 0.05077645182609558,
      "learning_rate": 9.317889317889318e-06,
      "loss": 0.6101,
      "step": 3620
    },
    {
      "epoch": 4.0134362934362935,
      "grad_norm": 0.11396818608045578,
      "learning_rate": 9.343629343629345e-06,
      "loss": 2.4917,
      "step": 3630
    },
    {
      "epoch": 4.0136936936936936,
      "grad_norm": 66.50660705566406,
      "learning_rate": 9.36936936936937e-06,
      "loss": 0.9722,
      "step": 3640
    },
    {
      "epoch": 4.013951093951094,
      "grad_norm": 91.65625762939453,
      "learning_rate": 9.395109395109396e-06,
      "loss": 1.8263,
      "step": 3650
    },
    {
      "epoch": 4.014208494208495,
      "grad_norm": 0.4844493567943573,
      "learning_rate": 9.420849420849421e-06,
      "loss": 1.0256,
      "step": 3660
    },
    {
      "epoch": 4.014465894465895,
      "grad_norm": 0.10221810638904572,
      "learning_rate": 9.446589446589446e-06,
      "loss": 0.8021,
      "step": 3670
    },
    {
      "epoch": 4.014723294723295,
      "grad_norm": 133.58883666992188,
      "learning_rate": 9.472329472329473e-06,
      "loss": 1.4417,
      "step": 3680
    },
    {
      "epoch": 4.014980694980695,
      "grad_norm": 105.06199645996094,
      "learning_rate": 9.498069498069498e-06,
      "loss": 1.2516,
      "step": 3690
    },
    {
      "epoch": 4.015238095238095,
      "grad_norm": 0.04352974891662598,
      "learning_rate": 9.523809523809525e-06,
      "loss": 0.9863,
      "step": 3700
    },
    {
      "epoch": 4.015495495495496,
      "grad_norm": 0.12624229490756989,
      "learning_rate": 9.54954954954955e-06,
      "loss": 0.5863,
      "step": 3710
    },
    {
      "epoch": 4.015752895752896,
      "grad_norm": 0.05648575723171234,
      "learning_rate": 9.575289575289576e-06,
      "loss": 1.0716,
      "step": 3720
    },
    {
      "epoch": 4.016010296010296,
      "grad_norm": 1.002746820449829,
      "learning_rate": 9.601029601029601e-06,
      "loss": 1.0243,
      "step": 3730
    },
    {
      "epoch": 4.016267696267696,
      "grad_norm": 0.12015024572610855,
      "learning_rate": 9.626769626769628e-06,
      "loss": 1.0932,
      "step": 3740
    },
    {
      "epoch": 4.016525096525097,
      "grad_norm": 86.13858032226562,
      "learning_rate": 9.652509652509653e-06,
      "loss": 1.452,
      "step": 3750
    },
    {
      "epoch": 4.016782496782497,
      "grad_norm": 378.3923034667969,
      "learning_rate": 9.67824967824968e-06,
      "loss": 1.5833,
      "step": 3760
    },
    {
      "epoch": 4.017039897039897,
      "grad_norm": 0.2414206564426422,
      "learning_rate": 9.703989703989705e-06,
      "loss": 0.6792,
      "step": 3770
    },
    {
      "epoch": 4.017297297297297,
      "grad_norm": 0.05312476307153702,
      "learning_rate": 9.729729729729732e-06,
      "loss": 0.2052,
      "step": 3780
    },
    {
      "epoch": 4.017554697554697,
      "grad_norm": 435.47015380859375,
      "learning_rate": 9.755469755469757e-06,
      "loss": 2.0852,
      "step": 3790
    },
    {
      "epoch": 4.017812097812098,
      "grad_norm": 94.0691909790039,
      "learning_rate": 9.781209781209782e-06,
      "loss": 0.6855,
      "step": 3800
    },
    {
      "epoch": 4.018069498069498,
      "grad_norm": 39.46242904663086,
      "learning_rate": 9.806949806949808e-06,
      "loss": 1.4943,
      "step": 3810
    },
    {
      "epoch": 4.018326898326898,
      "grad_norm": 0.44313281774520874,
      "learning_rate": 9.832689832689833e-06,
      "loss": 1.179,
      "step": 3820
    },
    {
      "epoch": 4.018584298584298,
      "grad_norm": 0.845443069934845,
      "learning_rate": 9.858429858429858e-06,
      "loss": 0.7002,
      "step": 3830
    },
    {
      "epoch": 4.018841698841699,
      "grad_norm": 0.0821545198559761,
      "learning_rate": 9.884169884169885e-06,
      "loss": 0.8125,
      "step": 3840
    },
    {
      "epoch": 4.019099099099099,
      "grad_norm": 20.311683654785156,
      "learning_rate": 9.90990990990991e-06,
      "loss": 1.363,
      "step": 3850
    },
    {
      "epoch": 4.019356499356499,
      "grad_norm": 0.4300023317337036,
      "learning_rate": 9.935649935649937e-06,
      "loss": 0.0073,
      "step": 3860
    },
    {
      "epoch": 4.0196138996138995,
      "grad_norm": 0.026708662509918213,
      "learning_rate": 9.961389961389962e-06,
      "loss": 0.6242,
      "step": 3870
    },
    {
      "epoch": 4.0198712998712995,
      "grad_norm": 0.08312055468559265,
      "learning_rate": 9.987129987129988e-06,
      "loss": 2.0611,
      "step": 3880
    },
    {
      "epoch": 4.02,
      "eval_accuracy": 0.782608695652174,
      "eval_loss": 1.0069572925567627,
      "eval_runtime": 15.454,
      "eval_samples_per_second": 2.977,
      "eval_steps_per_second": 2.977,
      "step": 3885
    },
    {
      "epoch": 5.0001287001287,
      "grad_norm": 0.08920583873987198,
      "learning_rate": 9.998569998569999e-06,
      "loss": 1.1077,
      "step": 3890
    },
    {
      "epoch": 5.0003861003861,
      "grad_norm": 0.2523305416107178,
      "learning_rate": 9.995709995709997e-06,
      "loss": 1.4093,
      "step": 3900
    },
    {
      "epoch": 5.000643500643501,
      "grad_norm": 124.69507598876953,
      "learning_rate": 9.992849992849994e-06,
      "loss": 0.9293,
      "step": 3910
    },
    {
      "epoch": 5.000900900900901,
      "grad_norm": 0.04359631985425949,
      "learning_rate": 9.989989989989992e-06,
      "loss": 0.6195,
      "step": 3920
    },
    {
      "epoch": 5.001158301158301,
      "grad_norm": 0.5823188424110413,
      "learning_rate": 9.987129987129988e-06,
      "loss": 1.2996,
      "step": 3930
    },
    {
      "epoch": 5.001415701415701,
      "grad_norm": 8.539837837219238,
      "learning_rate": 9.984269984269985e-06,
      "loss": 1.1665,
      "step": 3940
    },
    {
      "epoch": 5.001673101673101,
      "grad_norm": 48.99810028076172,
      "learning_rate": 9.981409981409981e-06,
      "loss": 0.9725,
      "step": 3950
    },
    {
      "epoch": 5.001930501930502,
      "grad_norm": 7.8650803565979,
      "learning_rate": 9.97854997854998e-06,
      "loss": 0.4707,
      "step": 3960
    },
    {
      "epoch": 5.002187902187902,
      "grad_norm": 0.041543882340192795,
      "learning_rate": 9.975689975689976e-06,
      "loss": 1.2242,
      "step": 3970
    },
    {
      "epoch": 5.002445302445302,
      "grad_norm": 0.11458486318588257,
      "learning_rate": 9.972829972829974e-06,
      "loss": 0.5848,
      "step": 3980
    },
    {
      "epoch": 5.0027027027027025,
      "grad_norm": 0.022899646311998367,
      "learning_rate": 9.96996996996997e-06,
      "loss": 1.6402,
      "step": 3990
    },
    {
      "epoch": 5.0029601029601025,
      "grad_norm": 0.3953354060649872,
      "learning_rate": 9.967109967109969e-06,
      "loss": 2.1309,
      "step": 4000
    },
    {
      "epoch": 5.0032175032175035,
      "grad_norm": 1.6589573621749878,
      "learning_rate": 9.964249964249965e-06,
      "loss": 0.7909,
      "step": 4010
    },
    {
      "epoch": 5.0034749034749035,
      "grad_norm": 59.53203201293945,
      "learning_rate": 9.961389961389962e-06,
      "loss": 1.8867,
      "step": 4020
    },
    {
      "epoch": 5.003732303732304,
      "grad_norm": 9.76834774017334,
      "learning_rate": 9.958529958529958e-06,
      "loss": 0.8285,
      "step": 4030
    },
    {
      "epoch": 5.003989703989704,
      "grad_norm": 0.16586628556251526,
      "learning_rate": 9.955669955669956e-06,
      "loss": 0.3786,
      "step": 4040
    },
    {
      "epoch": 5.004247104247105,
      "grad_norm": 0.14669224619865417,
      "learning_rate": 9.952809952809953e-06,
      "loss": 0.4211,
      "step": 4050
    },
    {
      "epoch": 5.004504504504505,
      "grad_norm": 158.1543731689453,
      "learning_rate": 9.949949949949951e-06,
      "loss": 0.1001,
      "step": 4060
    },
    {
      "epoch": 5.004761904761905,
      "grad_norm": 2.3664519786834717,
      "learning_rate": 9.947089947089947e-06,
      "loss": 0.4621,
      "step": 4070
    },
    {
      "epoch": 5.005019305019305,
      "grad_norm": 48.814117431640625,
      "learning_rate": 9.944229944229946e-06,
      "loss": 2.4297,
      "step": 4080
    },
    {
      "epoch": 5.005276705276705,
      "grad_norm": 46.63951873779297,
      "learning_rate": 9.941369941369942e-06,
      "loss": 1.5123,
      "step": 4090
    },
    {
      "epoch": 5.005534105534106,
      "grad_norm": 0.18840543925762177,
      "learning_rate": 9.938509938509938e-06,
      "loss": 0.542,
      "step": 4100
    },
    {
      "epoch": 5.005791505791506,
      "grad_norm": 0.28926903009414673,
      "learning_rate": 9.935649935649937e-06,
      "loss": 1.427,
      "step": 4110
    },
    {
      "epoch": 5.006048906048906,
      "grad_norm": 0.14491912722587585,
      "learning_rate": 9.932789932789933e-06,
      "loss": 0.9667,
      "step": 4120
    },
    {
      "epoch": 5.006306306306306,
      "grad_norm": 0.11946146190166473,
      "learning_rate": 9.929929929929931e-06,
      "loss": 0.4551,
      "step": 4130
    },
    {
      "epoch": 5.006563706563707,
      "grad_norm": 15.428871154785156,
      "learning_rate": 9.927069927069928e-06,
      "loss": 1.3655,
      "step": 4140
    },
    {
      "epoch": 5.006821106821107,
      "grad_norm": 126.84295654296875,
      "learning_rate": 9.924209924209926e-06,
      "loss": 0.9207,
      "step": 4150
    },
    {
      "epoch": 5.007078507078507,
      "grad_norm": 0.029610566794872284,
      "learning_rate": 9.921349921349922e-06,
      "loss": 0.004,
      "step": 4160
    },
    {
      "epoch": 5.007335907335907,
      "grad_norm": 0.1009824275970459,
      "learning_rate": 9.91848991848992e-06,
      "loss": 3.1177,
      "step": 4170
    },
    {
      "epoch": 5.007593307593307,
      "grad_norm": 0.4252803325653076,
      "learning_rate": 9.915629915629917e-06,
      "loss": 1.466,
      "step": 4180
    },
    {
      "epoch": 5.007850707850708,
      "grad_norm": 0.5266346335411072,
      "learning_rate": 9.912769912769913e-06,
      "loss": 0.7741,
      "step": 4190
    },
    {
      "epoch": 5.008108108108108,
      "grad_norm": 0.19914963841438293,
      "learning_rate": 9.90990990990991e-06,
      "loss": 0.9405,
      "step": 4200
    },
    {
      "epoch": 5.008365508365508,
      "grad_norm": 0.14336754381656647,
      "learning_rate": 9.907049907049908e-06,
      "loss": 0.5279,
      "step": 4210
    },
    {
      "epoch": 5.008622908622908,
      "grad_norm": 0.16508668661117554,
      "learning_rate": 9.904189904189905e-06,
      "loss": 1.5353,
      "step": 4220
    },
    {
      "epoch": 5.008880308880308,
      "grad_norm": 33.556251525878906,
      "learning_rate": 9.901329901329903e-06,
      "loss": 1.7293,
      "step": 4230
    },
    {
      "epoch": 5.009137709137709,
      "grad_norm": 0.7371513843536377,
      "learning_rate": 9.8984698984699e-06,
      "loss": 1.4752,
      "step": 4240
    },
    {
      "epoch": 5.0093951093951095,
      "grad_norm": 33.60354995727539,
      "learning_rate": 9.895609895609897e-06,
      "loss": 1.7302,
      "step": 4250
    },
    {
      "epoch": 5.0096525096525095,
      "grad_norm": 2.943221092224121,
      "learning_rate": 9.892749892749894e-06,
      "loss": 0.2602,
      "step": 4260
    },
    {
      "epoch": 5.00990990990991,
      "grad_norm": 0.029436398297548294,
      "learning_rate": 9.88988988988989e-06,
      "loss": 0.7335,
      "step": 4270
    },
    {
      "epoch": 5.0101673101673105,
      "grad_norm": 1.5921765565872192,
      "learning_rate": 9.887029887029887e-06,
      "loss": 1.551,
      "step": 4280
    },
    {
      "epoch": 5.010424710424711,
      "grad_norm": 0.15461337566375732,
      "learning_rate": 9.884169884169885e-06,
      "loss": 0.9979,
      "step": 4290
    },
    {
      "epoch": 5.010682110682111,
      "grad_norm": 0.40406131744384766,
      "learning_rate": 9.881309881309881e-06,
      "loss": 1.3223,
      "step": 4300
    },
    {
      "epoch": 5.010939510939511,
      "grad_norm": 154.21099853515625,
      "learning_rate": 9.87844987844988e-06,
      "loss": 1.8915,
      "step": 4310
    },
    {
      "epoch": 5.011196911196911,
      "grad_norm": 0.4178062081336975,
      "learning_rate": 9.875589875589876e-06,
      "loss": 0.2505,
      "step": 4320
    },
    {
      "epoch": 5.011454311454312,
      "grad_norm": 39.81876754760742,
      "learning_rate": 9.872729872729874e-06,
      "loss": 1.4891,
      "step": 4330
    },
    {
      "epoch": 5.011711711711712,
      "grad_norm": 0.10712433606386185,
      "learning_rate": 9.86986986986987e-06,
      "loss": 1.139,
      "step": 4340
    },
    {
      "epoch": 5.011969111969112,
      "grad_norm": 0.09909055382013321,
      "learning_rate": 9.867009867009867e-06,
      "loss": 0.4782,
      "step": 4350
    },
    {
      "epoch": 5.012226512226512,
      "grad_norm": 0.033393606543540955,
      "learning_rate": 9.864149864149865e-06,
      "loss": 1.3209,
      "step": 4360
    },
    {
      "epoch": 5.012483912483913,
      "grad_norm": 178.87274169921875,
      "learning_rate": 9.861289861289862e-06,
      "loss": 1.907,
      "step": 4370
    },
    {
      "epoch": 5.012741312741313,
      "grad_norm": 41.26455307006836,
      "learning_rate": 9.858429858429858e-06,
      "loss": 2.6711,
      "step": 4380
    },
    {
      "epoch": 5.012998712998713,
      "grad_norm": 0.3698756992816925,
      "learning_rate": 9.855569855569856e-06,
      "loss": 0.7055,
      "step": 4390
    },
    {
      "epoch": 5.013256113256113,
      "grad_norm": 240.4568328857422,
      "learning_rate": 9.852709852709853e-06,
      "loss": 0.5967,
      "step": 4400
    },
    {
      "epoch": 5.013513513513513,
      "grad_norm": 0.27150461077690125,
      "learning_rate": 9.849849849849851e-06,
      "loss": 1.4454,
      "step": 4410
    },
    {
      "epoch": 5.013770913770914,
      "grad_norm": 0.4859977960586548,
      "learning_rate": 9.846989846989847e-06,
      "loss": 0.5113,
      "step": 4420
    },
    {
      "epoch": 5.014028314028314,
      "grad_norm": 0.10837042331695557,
      "learning_rate": 9.844129844129846e-06,
      "loss": 1.0488,
      "step": 4430
    },
    {
      "epoch": 5.014285714285714,
      "grad_norm": 0.09987702965736389,
      "learning_rate": 9.841269841269842e-06,
      "loss": 0.2134,
      "step": 4440
    },
    {
      "epoch": 5.014543114543114,
      "grad_norm": 0.07767640054225922,
      "learning_rate": 9.838409838409839e-06,
      "loss": 1.6714,
      "step": 4450
    },
    {
      "epoch": 5.014800514800514,
      "grad_norm": 0.2725449502468109,
      "learning_rate": 9.835549835549837e-06,
      "loss": 0.8876,
      "step": 4460
    },
    {
      "epoch": 5.015057915057915,
      "grad_norm": 20.469709396362305,
      "learning_rate": 9.832689832689833e-06,
      "loss": 1.102,
      "step": 4470
    },
    {
      "epoch": 5.015315315315315,
      "grad_norm": 0.8056849241256714,
      "learning_rate": 9.829829829829831e-06,
      "loss": 0.9803,
      "step": 4480
    },
    {
      "epoch": 5.0155727155727154,
      "grad_norm": 0.11724364757537842,
      "learning_rate": 9.826969826969828e-06,
      "loss": 1.1717,
      "step": 4490
    },
    {
      "epoch": 5.0158301158301155,
      "grad_norm": 47.51169204711914,
      "learning_rate": 9.824109824109826e-06,
      "loss": 1.3517,
      "step": 4500
    },
    {
      "epoch": 5.0160875160875165,
      "grad_norm": 0.021014723926782608,
      "learning_rate": 9.821249821249822e-06,
      "loss": 1.8496,
      "step": 4510
    },
    {
      "epoch": 5.0163449163449165,
      "grad_norm": 109.34514617919922,
      "learning_rate": 9.818389818389819e-06,
      "loss": 0.5541,
      "step": 4520
    },
    {
      "epoch": 5.016602316602317,
      "grad_norm": 47.571590423583984,
      "learning_rate": 9.815529815529815e-06,
      "loss": 0.9756,
      "step": 4530
    },
    {
      "epoch": 5.016859716859717,
      "grad_norm": 0.08140508085489273,
      "learning_rate": 9.812669812669814e-06,
      "loss": 0.9688,
      "step": 4540
    },
    {
      "epoch": 5.017117117117117,
      "grad_norm": 300.04425048828125,
      "learning_rate": 9.80980980980981e-06,
      "loss": 0.9426,
      "step": 4550
    },
    {
      "epoch": 5.017374517374518,
      "grad_norm": 68.44358825683594,
      "learning_rate": 9.806949806949808e-06,
      "loss": 0.8523,
      "step": 4560
    },
    {
      "epoch": 5.017631917631918,
      "grad_norm": 0.3220089077949524,
      "learning_rate": 9.804089804089805e-06,
      "loss": 0.8468,
      "step": 4570
    },
    {
      "epoch": 5.017889317889318,
      "grad_norm": 101.19426727294922,
      "learning_rate": 9.801229801229803e-06,
      "loss": 0.557,
      "step": 4580
    },
    {
      "epoch": 5.018146718146718,
      "grad_norm": 0.06881493330001831,
      "learning_rate": 9.7983697983698e-06,
      "loss": 1.4231,
      "step": 4590
    },
    {
      "epoch": 5.018404118404119,
      "grad_norm": 0.06560848653316498,
      "learning_rate": 9.795509795509796e-06,
      "loss": 1.1197,
      "step": 4600
    },
    {
      "epoch": 5.018661518661519,
      "grad_norm": 43.2938117980957,
      "learning_rate": 9.792649792649794e-06,
      "loss": 1.691,
      "step": 4610
    },
    {
      "epoch": 5.018918918918919,
      "grad_norm": 227.61483764648438,
      "learning_rate": 9.78978978978979e-06,
      "loss": 1.1375,
      "step": 4620
    },
    {
      "epoch": 5.019176319176319,
      "grad_norm": 0.2670377790927887,
      "learning_rate": 9.786929786929787e-06,
      "loss": 1.6694,
      "step": 4630
    },
    {
      "epoch": 5.019433719433719,
      "grad_norm": 2.1211440563201904,
      "learning_rate": 9.784069784069785e-06,
      "loss": 1.3324,
      "step": 4640
    },
    {
      "epoch": 5.01969111969112,
      "grad_norm": 667.440185546875,
      "learning_rate": 9.781209781209782e-06,
      "loss": 1.2619,
      "step": 4650
    },
    {
      "epoch": 5.01994851994852,
      "grad_norm": 0.2644244432449341,
      "learning_rate": 9.77834977834978e-06,
      "loss": 0.4943,
      "step": 4660
    },
    {
      "epoch": 5.02,
      "eval_accuracy": 0.45652173913043476,
      "eval_loss": 3.257457971572876,
      "eval_runtime": 16.4996,
      "eval_samples_per_second": 2.788,
      "eval_steps_per_second": 2.788,
      "step": 4662
    },
    {
      "epoch": 6.000205920205921,
      "grad_norm": 0.05469132587313652,
      "learning_rate": 9.775489775489776e-06,
      "loss": 1.8461,
      "step": 4670
    },
    {
      "epoch": 6.000463320463321,
      "grad_norm": 0.45225292444229126,
      "learning_rate": 9.772629772629774e-06,
      "loss": 1.9015,
      "step": 4680
    },
    {
      "epoch": 6.000720720720721,
      "grad_norm": 33.26803970336914,
      "learning_rate": 9.76976976976977e-06,
      "loss": 2.2968,
      "step": 4690
    },
    {
      "epoch": 6.000978120978121,
      "grad_norm": 29.261432647705078,
      "learning_rate": 9.766909766909767e-06,
      "loss": 1.1083,
      "step": 4700
    },
    {
      "epoch": 6.001235521235521,
      "grad_norm": 0.2970881462097168,
      "learning_rate": 9.764049764049764e-06,
      "loss": 1.5378,
      "step": 4710
    },
    {
      "epoch": 6.001492921492922,
      "grad_norm": 0.820044755935669,
      "learning_rate": 9.761189761189762e-06,
      "loss": 0.7441,
      "step": 4720
    },
    {
      "epoch": 6.001750321750322,
      "grad_norm": 0.10087312757968903,
      "learning_rate": 9.758329758329758e-06,
      "loss": 0.3919,
      "step": 4730
    },
    {
      "epoch": 6.002007722007722,
      "grad_norm": 0.0689639076590538,
      "learning_rate": 9.755469755469757e-06,
      "loss": 1.4962,
      "step": 4740
    },
    {
      "epoch": 6.002265122265122,
      "grad_norm": 0.08053558319807053,
      "learning_rate": 9.752609752609753e-06,
      "loss": 1.8286,
      "step": 4750
    },
    {
      "epoch": 6.002522522522523,
      "grad_norm": 0.11258360743522644,
      "learning_rate": 9.749749749749751e-06,
      "loss": 0.9527,
      "step": 4760
    },
    {
      "epoch": 6.002779922779923,
      "grad_norm": 2.33036470413208,
      "learning_rate": 9.746889746889748e-06,
      "loss": 1.0479,
      "step": 4770
    },
    {
      "epoch": 6.003037323037323,
      "grad_norm": 2.5247790813446045,
      "learning_rate": 9.744029744029744e-06,
      "loss": 0.889,
      "step": 4780
    },
    {
      "epoch": 6.003294723294723,
      "grad_norm": 0.057729437947273254,
      "learning_rate": 9.741169741169742e-06,
      "loss": 1.0201,
      "step": 4790
    },
    {
      "epoch": 6.003552123552123,
      "grad_norm": 70.08110046386719,
      "learning_rate": 9.738309738309739e-06,
      "loss": 1.1606,
      "step": 4800
    },
    {
      "epoch": 6.003809523809524,
      "grad_norm": 0.26452142000198364,
      "learning_rate": 9.735449735449735e-06,
      "loss": 0.7754,
      "step": 4810
    },
    {
      "epoch": 6.004066924066924,
      "grad_norm": 0.436287522315979,
      "learning_rate": 9.732589732589733e-06,
      "loss": 0.9025,
      "step": 4820
    },
    {
      "epoch": 6.004324324324324,
      "grad_norm": 80.23675537109375,
      "learning_rate": 9.729729729729732e-06,
      "loss": 0.8798,
      "step": 4830
    },
    {
      "epoch": 6.004581724581724,
      "grad_norm": 187.31350708007812,
      "learning_rate": 9.726869726869728e-06,
      "loss": 1.0004,
      "step": 4840
    },
    {
      "epoch": 6.004839124839124,
      "grad_norm": 0.10163071006536484,
      "learning_rate": 9.724009724009724e-06,
      "loss": 1.1963,
      "step": 4850
    },
    {
      "epoch": 6.005096525096525,
      "grad_norm": 0.16315636038780212,
      "learning_rate": 9.721149721149723e-06,
      "loss": 1.364,
      "step": 4860
    },
    {
      "epoch": 6.005353925353925,
      "grad_norm": 0.1200859546661377,
      "learning_rate": 9.718289718289719e-06,
      "loss": 2.1892,
      "step": 4870
    },
    {
      "epoch": 6.0056113256113255,
      "grad_norm": 118.95958709716797,
      "learning_rate": 9.715429715429716e-06,
      "loss": 1.5139,
      "step": 4880
    },
    {
      "epoch": 6.0058687258687256,
      "grad_norm": 0.24599742889404297,
      "learning_rate": 9.712569712569714e-06,
      "loss": 0.8105,
      "step": 4890
    },
    {
      "epoch": 6.0061261261261265,
      "grad_norm": 1.2496775388717651,
      "learning_rate": 9.70970970970971e-06,
      "loss": 0.3488,
      "step": 4900
    },
    {
      "epoch": 6.006383526383527,
      "grad_norm": 32.44328689575195,
      "learning_rate": 9.706849706849708e-06,
      "loss": 0.8115,
      "step": 4910
    },
    {
      "epoch": 6.006640926640927,
      "grad_norm": 0.39588698744773865,
      "learning_rate": 9.703989703989705e-06,
      "loss": 1.2346,
      "step": 4920
    },
    {
      "epoch": 6.006898326898327,
      "grad_norm": 95.88199615478516,
      "learning_rate": 9.701129701129703e-06,
      "loss": 1.3044,
      "step": 4930
    },
    {
      "epoch": 6.007155727155727,
      "grad_norm": 0.2991234064102173,
      "learning_rate": 9.6982696982697e-06,
      "loss": 0.5547,
      "step": 4940
    },
    {
      "epoch": 6.007413127413128,
      "grad_norm": 0.5006702542304993,
      "learning_rate": 9.695409695409696e-06,
      "loss": 0.7136,
      "step": 4950
    },
    {
      "epoch": 6.007670527670528,
      "grad_norm": 0.5361593961715698,
      "learning_rate": 9.692549692549692e-06,
      "loss": 1.1746,
      "step": 4960
    },
    {
      "epoch": 6.007927927927928,
      "grad_norm": 0.11017898470163345,
      "learning_rate": 9.68968968968969e-06,
      "loss": 1.8545,
      "step": 4970
    },
    {
      "epoch": 6.008185328185328,
      "grad_norm": 0.18768690526485443,
      "learning_rate": 9.686829686829687e-06,
      "loss": 1.9508,
      "step": 4980
    },
    {
      "epoch": 6.008442728442729,
      "grad_norm": 0.10016202181577682,
      "learning_rate": 9.683969683969685e-06,
      "loss": 0.9797,
      "step": 4990
    },
    {
      "epoch": 6.008700128700129,
      "grad_norm": 309.5798645019531,
      "learning_rate": 9.681109681109682e-06,
      "loss": 1.8029,
      "step": 5000
    },
    {
      "epoch": 6.008957528957529,
      "grad_norm": 0.2282564789056778,
      "learning_rate": 9.67824967824968e-06,
      "loss": 1.4809,
      "step": 5010
    },
    {
      "epoch": 6.009214929214929,
      "grad_norm": 25.585124969482422,
      "learning_rate": 9.675389675389676e-06,
      "loss": 0.7676,
      "step": 5020
    },
    {
      "epoch": 6.009472329472329,
      "grad_norm": 614.1220092773438,
      "learning_rate": 9.672529672529673e-06,
      "loss": 1.0285,
      "step": 5030
    },
    {
      "epoch": 6.00972972972973,
      "grad_norm": 1062.7408447265625,
      "learning_rate": 9.669669669669671e-06,
      "loss": 1.3201,
      "step": 5040
    },
    {
      "epoch": 6.00998712998713,
      "grad_norm": 1.943227767944336,
      "learning_rate": 9.666809666809667e-06,
      "loss": 0.4388,
      "step": 5050
    },
    {
      "epoch": 6.01024453024453,
      "grad_norm": 0.21532581746578217,
      "learning_rate": 9.663949663949664e-06,
      "loss": 0.7339,
      "step": 5060
    },
    {
      "epoch": 6.01050193050193,
      "grad_norm": 0.018828408792614937,
      "learning_rate": 9.661089661089662e-06,
      "loss": 1.2573,
      "step": 5070
    },
    {
      "epoch": 6.01075933075933,
      "grad_norm": 0.5531067252159119,
      "learning_rate": 9.658229658229659e-06,
      "loss": 1.4139,
      "step": 5080
    },
    {
      "epoch": 6.011016731016731,
      "grad_norm": 0.6515811085700989,
      "learning_rate": 9.655369655369657e-06,
      "loss": 0.4051,
      "step": 5090
    },
    {
      "epoch": 6.011274131274131,
      "grad_norm": 0.48128044605255127,
      "learning_rate": 9.652509652509653e-06,
      "loss": 0.8952,
      "step": 5100
    },
    {
      "epoch": 6.011531531531531,
      "grad_norm": 0.38696253299713135,
      "learning_rate": 9.649649649649651e-06,
      "loss": 0.4374,
      "step": 5110
    },
    {
      "epoch": 6.0117889317889315,
      "grad_norm": 4.012783050537109,
      "learning_rate": 9.646789646789648e-06,
      "loss": 2.2389,
      "step": 5120
    },
    {
      "epoch": 6.012046332046332,
      "grad_norm": 569.133056640625,
      "learning_rate": 9.643929643929644e-06,
      "loss": 0.286,
      "step": 5130
    },
    {
      "epoch": 6.0123037323037325,
      "grad_norm": 27.13798713684082,
      "learning_rate": 9.64106964106964e-06,
      "loss": 1.8604,
      "step": 5140
    },
    {
      "epoch": 6.012561132561133,
      "grad_norm": 70.15827941894531,
      "learning_rate": 9.638209638209639e-06,
      "loss": 1.1294,
      "step": 5150
    },
    {
      "epoch": 6.012818532818533,
      "grad_norm": 43.90142059326172,
      "learning_rate": 9.635349635349635e-06,
      "loss": 1.5029,
      "step": 5160
    },
    {
      "epoch": 6.013075933075933,
      "grad_norm": 1.4846750497817993,
      "learning_rate": 9.632489632489634e-06,
      "loss": 0.3423,
      "step": 5170
    },
    {
      "epoch": 6.013333333333334,
      "grad_norm": 1.0613987445831299,
      "learning_rate": 9.62962962962963e-06,
      "loss": 0.3034,
      "step": 5180
    },
    {
      "epoch": 6.013590733590734,
      "grad_norm": 17.084869384765625,
      "learning_rate": 9.626769626769628e-06,
      "loss": 0.5023,
      "step": 5190
    },
    {
      "epoch": 6.013848133848134,
      "grad_norm": 0.020237158983945847,
      "learning_rate": 9.623909623909625e-06,
      "loss": 0.602,
      "step": 5200
    },
    {
      "epoch": 6.014105534105534,
      "grad_norm": 0.02380361221730709,
      "learning_rate": 9.621049621049621e-06,
      "loss": 0.8313,
      "step": 5210
    },
    {
      "epoch": 6.014362934362935,
      "grad_norm": 0.016284339129924774,
      "learning_rate": 9.61818961818962e-06,
      "loss": 1.5708,
      "step": 5220
    },
    {
      "epoch": 6.014620334620335,
      "grad_norm": 615.75,
      "learning_rate": 9.615329615329616e-06,
      "loss": 2.1004,
      "step": 5230
    },
    {
      "epoch": 6.014877734877735,
      "grad_norm": 0.3784148395061493,
      "learning_rate": 9.612469612469614e-06,
      "loss": 1.5562,
      "step": 5240
    },
    {
      "epoch": 6.015135135135135,
      "grad_norm": 0.0682976171374321,
      "learning_rate": 9.60960960960961e-06,
      "loss": 1.2095,
      "step": 5250
    },
    {
      "epoch": 6.015392535392535,
      "grad_norm": 0.5212209820747375,
      "learning_rate": 9.606749606749609e-06,
      "loss": 0.5701,
      "step": 5260
    },
    {
      "epoch": 6.015649935649936,
      "grad_norm": 458.25054931640625,
      "learning_rate": 9.603889603889605e-06,
      "loss": 1.0626,
      "step": 5270
    },
    {
      "epoch": 6.015907335907336,
      "grad_norm": 50.43803024291992,
      "learning_rate": 9.601029601029601e-06,
      "loss": 2.7078,
      "step": 5280
    },
    {
      "epoch": 6.016164736164736,
      "grad_norm": 0.5311137437820435,
      "learning_rate": 9.5981695981696e-06,
      "loss": 0.8123,
      "step": 5290
    },
    {
      "epoch": 6.016422136422136,
      "grad_norm": 0.015996312722563744,
      "learning_rate": 9.595309595309596e-06,
      "loss": 0.5862,
      "step": 5300
    },
    {
      "epoch": 6.016679536679536,
      "grad_norm": 0.4782085716724396,
      "learning_rate": 9.592449592449593e-06,
      "loss": 0.0025,
      "step": 5310
    },
    {
      "epoch": 6.016936936936937,
      "grad_norm": 0.11970794200897217,
      "learning_rate": 9.58958958958959e-06,
      "loss": 2.5401,
      "step": 5320
    },
    {
      "epoch": 6.017194337194337,
      "grad_norm": 134.60182189941406,
      "learning_rate": 9.586729586729587e-06,
      "loss": 1.5374,
      "step": 5330
    },
    {
      "epoch": 6.017451737451737,
      "grad_norm": 0.026819629594683647,
      "learning_rate": 9.583869583869585e-06,
      "loss": 0.2047,
      "step": 5340
    },
    {
      "epoch": 6.017709137709137,
      "grad_norm": 30.610950469970703,
      "learning_rate": 9.581009581009582e-06,
      "loss": 1.9448,
      "step": 5350
    },
    {
      "epoch": 6.017966537966538,
      "grad_norm": 23.234464645385742,
      "learning_rate": 9.57814957814958e-06,
      "loss": 2.3533,
      "step": 5360
    },
    {
      "epoch": 6.018223938223938,
      "grad_norm": 3.35427188873291,
      "learning_rate": 9.575289575289576e-06,
      "loss": 0.6203,
      "step": 5370
    },
    {
      "epoch": 6.0184813384813385,
      "grad_norm": 0.10274845361709595,
      "learning_rate": 9.572429572429573e-06,
      "loss": 0.8995,
      "step": 5380
    },
    {
      "epoch": 6.0187387387387385,
      "grad_norm": 49.04452896118164,
      "learning_rate": 9.56956956956957e-06,
      "loss": 2.0313,
      "step": 5390
    },
    {
      "epoch": 6.018996138996139,
      "grad_norm": 0.2745307385921478,
      "learning_rate": 9.566709566709568e-06,
      "loss": 1.3825,
      "step": 5400
    },
    {
      "epoch": 6.01925353925354,
      "grad_norm": 34.342376708984375,
      "learning_rate": 9.563849563849564e-06,
      "loss": 1.4412,
      "step": 5410
    },
    {
      "epoch": 6.01951093951094,
      "grad_norm": 0.6465485095977783,
      "learning_rate": 9.560989560989562e-06,
      "loss": 0.6112,
      "step": 5420
    },
    {
      "epoch": 6.01976833976834,
      "grad_norm": 0.03590095415711403,
      "learning_rate": 9.558129558129559e-06,
      "loss": 0.708,
      "step": 5430
    },
    {
      "epoch": 6.02,
      "eval_accuracy": 0.45652173913043476,
      "eval_loss": 2.1190643310546875,
      "eval_runtime": 15.4643,
      "eval_samples_per_second": 2.975,
      "eval_steps_per_second": 2.975,
      "step": 5439
    },
    {
      "epoch": 7.00002574002574,
      "grad_norm": 0.48325133323669434,
      "learning_rate": 9.555269555269557e-06,
      "loss": 2.154,
      "step": 5440
    },
    {
      "epoch": 7.00028314028314,
      "grad_norm": 0.226426899433136,
      "learning_rate": 9.552409552409553e-06,
      "loss": 0.6806,
      "step": 5450
    },
    {
      "epoch": 7.00054054054054,
      "grad_norm": 296.4757995605469,
      "learning_rate": 9.54954954954955e-06,
      "loss": 0.646,
      "step": 5460
    },
    {
      "epoch": 7.00079794079794,
      "grad_norm": 26.78754997253418,
      "learning_rate": 9.546689546689546e-06,
      "loss": 0.6618,
      "step": 5470
    },
    {
      "epoch": 7.001055341055341,
      "grad_norm": 0.030151739716529846,
      "learning_rate": 9.543829543829544e-06,
      "loss": 1.7167,
      "step": 5480
    },
    {
      "epoch": 7.001312741312741,
      "grad_norm": 179.15652465820312,
      "learning_rate": 9.540969540969541e-06,
      "loss": 1.1285,
      "step": 5490
    },
    {
      "epoch": 7.0015701415701415,
      "grad_norm": 0.09001732617616653,
      "learning_rate": 9.538109538109539e-06,
      "loss": 0.4645,
      "step": 5500
    },
    {
      "epoch": 7.0018275418275415,
      "grad_norm": 173.9577178955078,
      "learning_rate": 9.535249535249535e-06,
      "loss": 1.4252,
      "step": 5510
    },
    {
      "epoch": 7.0020849420849425,
      "grad_norm": 12.78044319152832,
      "learning_rate": 9.532389532389534e-06,
      "loss": 0.8229,
      "step": 5520
    },
    {
      "epoch": 7.0023423423423425,
      "grad_norm": 0.49420520663261414,
      "learning_rate": 9.52952952952953e-06,
      "loss": 0.023,
      "step": 5530
    },
    {
      "epoch": 7.002599742599743,
      "grad_norm": 34.80965042114258,
      "learning_rate": 9.526669526669528e-06,
      "loss": 0.6871,
      "step": 5540
    },
    {
      "epoch": 7.002857142857143,
      "grad_norm": 2.7141544818878174,
      "learning_rate": 9.523809523809525e-06,
      "loss": 2.8692,
      "step": 5550
    },
    {
      "epoch": 7.003114543114543,
      "grad_norm": 27.715356826782227,
      "learning_rate": 9.520949520949521e-06,
      "loss": 2.216,
      "step": 5560
    },
    {
      "epoch": 7.003371943371944,
      "grad_norm": 0.8110296726226807,
      "learning_rate": 9.51808951808952e-06,
      "loss": 0.6849,
      "step": 5570
    },
    {
      "epoch": 7.003629343629344,
      "grad_norm": 11.079747200012207,
      "learning_rate": 9.515229515229516e-06,
      "loss": 1.0239,
      "step": 5580
    },
    {
      "epoch": 7.003886743886744,
      "grad_norm": 0.3448202311992645,
      "learning_rate": 9.512369512369514e-06,
      "loss": 0.67,
      "step": 5590
    },
    {
      "epoch": 7.004144144144144,
      "grad_norm": 0.313346266746521,
      "learning_rate": 9.50950950950951e-06,
      "loss": 0.4556,
      "step": 5600
    },
    {
      "epoch": 7.004401544401545,
      "grad_norm": 24.44046401977539,
      "learning_rate": 9.506649506649509e-06,
      "loss": 1.7268,
      "step": 5610
    },
    {
      "epoch": 7.004658944658945,
      "grad_norm": 23.968717575073242,
      "learning_rate": 9.503789503789505e-06,
      "loss": 1.6574,
      "step": 5620
    },
    {
      "epoch": 7.004916344916345,
      "grad_norm": 1.9536049365997314,
      "learning_rate": 9.500929500929502e-06,
      "loss": 0.6711,
      "step": 5630
    },
    {
      "epoch": 7.005173745173745,
      "grad_norm": 0.11860974878072739,
      "learning_rate": 9.498069498069498e-06,
      "loss": 0.0218,
      "step": 5640
    },
    {
      "epoch": 7.005431145431145,
      "grad_norm": 3.811457872390747,
      "learning_rate": 9.495209495209496e-06,
      "loss": 0.4929,
      "step": 5650
    },
    {
      "epoch": 7.005688545688546,
      "grad_norm": 0.047364648431539536,
      "learning_rate": 9.492349492349493e-06,
      "loss": 0.5272,
      "step": 5660
    },
    {
      "epoch": 7.005945945945946,
      "grad_norm": 0.2796438932418823,
      "learning_rate": 9.489489489489491e-06,
      "loss": 1.3004,
      "step": 5670
    },
    {
      "epoch": 7.006203346203346,
      "grad_norm": 82.55836486816406,
      "learning_rate": 9.486629486629487e-06,
      "loss": 0.6588,
      "step": 5680
    },
    {
      "epoch": 7.006460746460746,
      "grad_norm": 0.2848183512687683,
      "learning_rate": 9.483769483769485e-06,
      "loss": 1.5836,
      "step": 5690
    },
    {
      "epoch": 7.006718146718146,
      "grad_norm": 60.738807678222656,
      "learning_rate": 9.480909480909482e-06,
      "loss": 0.8493,
      "step": 5700
    },
    {
      "epoch": 7.006975546975547,
      "grad_norm": 0.10891953110694885,
      "learning_rate": 9.478049478049478e-06,
      "loss": 0.2529,
      "step": 5710
    },
    {
      "epoch": 7.007232947232947,
      "grad_norm": 393.4809875488281,
      "learning_rate": 9.475189475189477e-06,
      "loss": 0.5602,
      "step": 5720
    },
    {
      "epoch": 7.007490347490347,
      "grad_norm": 0.00841040350496769,
      "learning_rate": 9.472329472329473e-06,
      "loss": 1.9119,
      "step": 5730
    },
    {
      "epoch": 7.0077477477477474,
      "grad_norm": 312.6628723144531,
      "learning_rate": 9.46946946946947e-06,
      "loss": 1.7177,
      "step": 5740
    },
    {
      "epoch": 7.008005148005148,
      "grad_norm": 2.4715847969055176,
      "learning_rate": 9.466609466609468e-06,
      "loss": 1.079,
      "step": 5750
    },
    {
      "epoch": 7.0082625482625485,
      "grad_norm": 21.34333610534668,
      "learning_rate": 9.463749463749464e-06,
      "loss": 0.3107,
      "step": 5760
    },
    {
      "epoch": 7.0085199485199485,
      "grad_norm": 359.7504577636719,
      "learning_rate": 9.460889460889462e-06,
      "loss": 1.2887,
      "step": 5770
    },
    {
      "epoch": 7.008777348777349,
      "grad_norm": 80.45604705810547,
      "learning_rate": 9.458029458029459e-06,
      "loss": 0.8233,
      "step": 5780
    },
    {
      "epoch": 7.009034749034749,
      "grad_norm": 0.030902400612831116,
      "learning_rate": 9.455169455169457e-06,
      "loss": 0.1018,
      "step": 5790
    },
    {
      "epoch": 7.00929214929215,
      "grad_norm": 25.676687240600586,
      "learning_rate": 9.452309452309453e-06,
      "loss": 1.71,
      "step": 5800
    },
    {
      "epoch": 7.00954954954955,
      "grad_norm": 2.5907270908355713,
      "learning_rate": 9.44944944944945e-06,
      "loss": 2.2116,
      "step": 5810
    },
    {
      "epoch": 7.00980694980695,
      "grad_norm": 3.035196542739868,
      "learning_rate": 9.446589446589446e-06,
      "loss": 0.181,
      "step": 5820
    },
    {
      "epoch": 7.01006435006435,
      "grad_norm": 0.8250662088394165,
      "learning_rate": 9.443729443729445e-06,
      "loss": 1.0846,
      "step": 5830
    },
    {
      "epoch": 7.010321750321751,
      "grad_norm": 0.5744351148605347,
      "learning_rate": 9.440869440869441e-06,
      "loss": 1.3217,
      "step": 5840
    },
    {
      "epoch": 7.010579150579151,
      "grad_norm": 33.13734436035156,
      "learning_rate": 9.43800943800944e-06,
      "loss": 1.1113,
      "step": 5850
    },
    {
      "epoch": 7.010836550836551,
      "grad_norm": 0.7868608832359314,
      "learning_rate": 9.435149435149436e-06,
      "loss": 2.172,
      "step": 5860
    },
    {
      "epoch": 7.011093951093951,
      "grad_norm": 1.6519296169281006,
      "learning_rate": 9.432289432289434e-06,
      "loss": 0.0257,
      "step": 5870
    },
    {
      "epoch": 7.011351351351351,
      "grad_norm": 396.9693908691406,
      "learning_rate": 9.42942942942943e-06,
      "loss": 0.5133,
      "step": 5880
    },
    {
      "epoch": 7.011608751608752,
      "grad_norm": 22.835020065307617,
      "learning_rate": 9.426569426569427e-06,
      "loss": 2.0565,
      "step": 5890
    },
    {
      "epoch": 7.011866151866152,
      "grad_norm": 54.88117980957031,
      "learning_rate": 9.423709423709423e-06,
      "loss": 1.8865,
      "step": 5900
    },
    {
      "epoch": 7.012123552123552,
      "grad_norm": 0.0457146093249321,
      "learning_rate": 9.420849420849421e-06,
      "loss": 0.0163,
      "step": 5910
    },
    {
      "epoch": 7.012380952380952,
      "grad_norm": 41.79448318481445,
      "learning_rate": 9.417989417989418e-06,
      "loss": 1.0209,
      "step": 5920
    },
    {
      "epoch": 7.012638352638352,
      "grad_norm": 270.59869384765625,
      "learning_rate": 9.415129415129416e-06,
      "loss": 1.266,
      "step": 5930
    },
    {
      "epoch": 7.012895752895753,
      "grad_norm": 3.5031347274780273,
      "learning_rate": 9.412269412269412e-06,
      "loss": 0.6264,
      "step": 5940
    },
    {
      "epoch": 7.013153153153153,
      "grad_norm": 0.2792441248893738,
      "learning_rate": 9.40940940940941e-06,
      "loss": 1.3217,
      "step": 5950
    },
    {
      "epoch": 7.013410553410553,
      "grad_norm": 938.5770874023438,
      "learning_rate": 9.406549406549407e-06,
      "loss": 0.7044,
      "step": 5960
    },
    {
      "epoch": 7.013667953667953,
      "grad_norm": 0.05712044611573219,
      "learning_rate": 9.403689403689405e-06,
      "loss": 0.4378,
      "step": 5970
    },
    {
      "epoch": 7.013925353925354,
      "grad_norm": 87.4327621459961,
      "learning_rate": 9.400829400829402e-06,
      "loss": 1.763,
      "step": 5980
    },
    {
      "epoch": 7.014182754182754,
      "grad_norm": 0.3371102511882782,
      "learning_rate": 9.397969397969398e-06,
      "loss": 0.9069,
      "step": 5990
    },
    {
      "epoch": 7.0144401544401545,
      "grad_norm": 0.3371520936489105,
      "learning_rate": 9.395109395109396e-06,
      "loss": 0.8725,
      "step": 6000
    },
    {
      "epoch": 7.0146975546975545,
      "grad_norm": 96.59081268310547,
      "learning_rate": 9.392249392249393e-06,
      "loss": 1.4305,
      "step": 6010
    },
    {
      "epoch": 7.014954954954955,
      "grad_norm": 0.3758425712585449,
      "learning_rate": 9.389389389389391e-06,
      "loss": 1.7905,
      "step": 6020
    },
    {
      "epoch": 7.0152123552123555,
      "grad_norm": 0.03138767182826996,
      "learning_rate": 9.386529386529387e-06,
      "loss": 2.0007,
      "step": 6030
    },
    {
      "epoch": 7.015469755469756,
      "grad_norm": 33.08185958862305,
      "learning_rate": 9.383669383669386e-06,
      "loss": 0.9456,
      "step": 6040
    },
    {
      "epoch": 7.015727155727156,
      "grad_norm": 0.5954800248146057,
      "learning_rate": 9.380809380809382e-06,
      "loss": 1.2856,
      "step": 6050
    },
    {
      "epoch": 7.015984555984556,
      "grad_norm": 0.17369569838047028,
      "learning_rate": 9.377949377949379e-06,
      "loss": 0.5541,
      "step": 6060
    },
    {
      "epoch": 7.016241956241957,
      "grad_norm": 0.3233203887939453,
      "learning_rate": 9.375089375089375e-06,
      "loss": 1.2611,
      "step": 6070
    },
    {
      "epoch": 7.016499356499357,
      "grad_norm": 40.826045989990234,
      "learning_rate": 9.372229372229373e-06,
      "loss": 1.7143,
      "step": 6080
    },
    {
      "epoch": 7.016756756756757,
      "grad_norm": 56.034400939941406,
      "learning_rate": 9.36936936936937e-06,
      "loss": 0.9582,
      "step": 6090
    },
    {
      "epoch": 7.017014157014157,
      "grad_norm": 8.84603214263916,
      "learning_rate": 9.366509366509368e-06,
      "loss": 0.018,
      "step": 6100
    },
    {
      "epoch": 7.017271557271557,
      "grad_norm": 0.7207472920417786,
      "learning_rate": 9.363649363649364e-06,
      "loss": 0.507,
      "step": 6110
    },
    {
      "epoch": 7.017528957528958,
      "grad_norm": 0.1129591315984726,
      "learning_rate": 9.360789360789362e-06,
      "loss": 0.4114,
      "step": 6120
    },
    {
      "epoch": 7.017786357786358,
      "grad_norm": 0.03248320892453194,
      "learning_rate": 9.357929357929359e-06,
      "loss": 1.3072,
      "step": 6130
    },
    {
      "epoch": 7.018043758043758,
      "grad_norm": 41.5610237121582,
      "learning_rate": 9.355069355069355e-06,
      "loss": 1.705,
      "step": 6140
    },
    {
      "epoch": 7.018301158301158,
      "grad_norm": 0.6427077651023865,
      "learning_rate": 9.352209352209352e-06,
      "loss": 0.9994,
      "step": 6150
    },
    {
      "epoch": 7.018558558558558,
      "grad_norm": 0.4536170959472656,
      "learning_rate": 9.34934934934935e-06,
      "loss": 0.4102,
      "step": 6160
    },
    {
      "epoch": 7.018815958815959,
      "grad_norm": 0.4406795799732208,
      "learning_rate": 9.346489346489346e-06,
      "loss": 0.4422,
      "step": 6170
    },
    {
      "epoch": 7.019073359073359,
      "grad_norm": 0.02585803158581257,
      "learning_rate": 9.343629343629345e-06,
      "loss": 0.4257,
      "step": 6180
    },
    {
      "epoch": 7.019330759330759,
      "grad_norm": 0.07975375652313232,
      "learning_rate": 9.340769340769341e-06,
      "loss": 0.5653,
      "step": 6190
    },
    {
      "epoch": 7.019588159588159,
      "grad_norm": 0.0948067232966423,
      "learning_rate": 9.33790933790934e-06,
      "loss": 1.3287,
      "step": 6200
    },
    {
      "epoch": 7.01984555984556,
      "grad_norm": 0.33225017786026,
      "learning_rate": 9.335049335049336e-06,
      "loss": 0.5518,
      "step": 6210
    },
    {
      "epoch": 7.02,
      "eval_accuracy": 0.7608695652173914,
      "eval_loss": 1.4169539213180542,
      "eval_runtime": 15.4728,
      "eval_samples_per_second": 2.973,
      "eval_steps_per_second": 2.973,
      "step": 6216
    },
    {
      "epoch": 8.00010296010296,
      "grad_norm": 0.03941204398870468,
      "learning_rate": 9.332189332189334e-06,
      "loss": 1.5208,
      "step": 6220
    },
    {
      "epoch": 8.00036036036036,
      "grad_norm": 0.024762960150837898,
      "learning_rate": 9.32932932932933e-06,
      "loss": 0.002,
      "step": 6230
    },
    {
      "epoch": 8.00061776061776,
      "grad_norm": 398.88885498046875,
      "learning_rate": 9.326469326469327e-06,
      "loss": 0.6583,
      "step": 6240
    },
    {
      "epoch": 8.000875160875161,
      "grad_norm": 0.023645753040909767,
      "learning_rate": 9.323609323609323e-06,
      "loss": 1.3758,
      "step": 6250
    },
    {
      "epoch": 8.001132561132561,
      "grad_norm": 0.012583833187818527,
      "learning_rate": 9.320749320749321e-06,
      "loss": 0.8099,
      "step": 6260
    },
    {
      "epoch": 8.001389961389961,
      "grad_norm": 0.020370107144117355,
      "learning_rate": 9.317889317889318e-06,
      "loss": 1.0169,
      "step": 6270
    },
    {
      "epoch": 8.001647361647361,
      "grad_norm": 0.051698509603738785,
      "learning_rate": 9.315029315029316e-06,
      "loss": 0.4926,
      "step": 6280
    },
    {
      "epoch": 8.001904761904761,
      "grad_norm": 0.9180232882499695,
      "learning_rate": 9.312169312169313e-06,
      "loss": 1.2576,
      "step": 6290
    },
    {
      "epoch": 8.002162162162163,
      "grad_norm": 1456.8536376953125,
      "learning_rate": 9.30930930930931e-06,
      "loss": 0.2036,
      "step": 6300
    },
    {
      "epoch": 8.002419562419563,
      "grad_norm": 164.44483947753906,
      "learning_rate": 9.306449306449307e-06,
      "loss": 2.9082,
      "step": 6310
    },
    {
      "epoch": 8.002676962676963,
      "grad_norm": 1.0428372621536255,
      "learning_rate": 9.303589303589304e-06,
      "loss": 0.3385,
      "step": 6320
    },
    {
      "epoch": 8.002934362934363,
      "grad_norm": 0.04050123319029808,
      "learning_rate": 9.300729300729302e-06,
      "loss": 0.4214,
      "step": 6330
    },
    {
      "epoch": 8.003191763191763,
      "grad_norm": 0.06089348345994949,
      "learning_rate": 9.297869297869298e-06,
      "loss": 0.8934,
      "step": 6340
    },
    {
      "epoch": 8.003449163449163,
      "grad_norm": 0.10277870297431946,
      "learning_rate": 9.295009295009296e-06,
      "loss": 1.4394,
      "step": 6350
    },
    {
      "epoch": 8.003706563706563,
      "grad_norm": 0.1762549728155136,
      "learning_rate": 9.292149292149293e-06,
      "loss": 2.8751,
      "step": 6360
    },
    {
      "epoch": 8.003963963963963,
      "grad_norm": 9.25487232208252,
      "learning_rate": 9.289289289289291e-06,
      "loss": 0.4291,
      "step": 6370
    },
    {
      "epoch": 8.004221364221364,
      "grad_norm": 0.05675046518445015,
      "learning_rate": 9.286429286429288e-06,
      "loss": 0.3791,
      "step": 6380
    },
    {
      "epoch": 8.004478764478764,
      "grad_norm": 0.13098298013210297,
      "learning_rate": 9.283569283569284e-06,
      "loss": 1.0009,
      "step": 6390
    },
    {
      "epoch": 8.004736164736165,
      "grad_norm": 0.7045866847038269,
      "learning_rate": 9.28070928070928e-06,
      "loss": 1.6997,
      "step": 6400
    },
    {
      "epoch": 8.004993564993566,
      "grad_norm": 0.40885546803474426,
      "learning_rate": 9.277849277849279e-06,
      "loss": 1.2252,
      "step": 6410
    },
    {
      "epoch": 8.005250965250966,
      "grad_norm": 0.27662187814712524,
      "learning_rate": 9.274989274989275e-06,
      "loss": 0.927,
      "step": 6420
    },
    {
      "epoch": 8.005508365508366,
      "grad_norm": 0.2577447295188904,
      "learning_rate": 9.272129272129273e-06,
      "loss": 0.4666,
      "step": 6430
    },
    {
      "epoch": 8.005765765765766,
      "grad_norm": 0.04225856065750122,
      "learning_rate": 9.26926926926927e-06,
      "loss": 1.812,
      "step": 6440
    },
    {
      "epoch": 8.006023166023166,
      "grad_norm": 0.5405942797660828,
      "learning_rate": 9.266409266409268e-06,
      "loss": 1.5169,
      "step": 6450
    },
    {
      "epoch": 8.006280566280566,
      "grad_norm": 40.016326904296875,
      "learning_rate": 9.263549263549264e-06,
      "loss": 2.1544,
      "step": 6460
    },
    {
      "epoch": 8.006537966537966,
      "grad_norm": 0.7114521861076355,
      "learning_rate": 9.260689260689263e-06,
      "loss": 0.3712,
      "step": 6470
    },
    {
      "epoch": 8.006795366795366,
      "grad_norm": 0.04546615853905678,
      "learning_rate": 9.257829257829259e-06,
      "loss": 0.2992,
      "step": 6480
    },
    {
      "epoch": 8.007052767052768,
      "grad_norm": 230.9104766845703,
      "learning_rate": 9.254969254969256e-06,
      "loss": 0.3405,
      "step": 6490
    },
    {
      "epoch": 8.007310167310168,
      "grad_norm": 0.03966987505555153,
      "learning_rate": 9.252109252109252e-06,
      "loss": 1.7531,
      "step": 6500
    },
    {
      "epoch": 8.007567567567568,
      "grad_norm": 0.1251051276922226,
      "learning_rate": 9.24924924924925e-06,
      "loss": 1.1201,
      "step": 6510
    },
    {
      "epoch": 8.007824967824968,
      "grad_norm": 0.15064413845539093,
      "learning_rate": 9.246389246389247e-06,
      "loss": 0.8073,
      "step": 6520
    },
    {
      "epoch": 8.008082368082368,
      "grad_norm": 27.61880874633789,
      "learning_rate": 9.243529243529245e-06,
      "loss": 0.8102,
      "step": 6530
    },
    {
      "epoch": 8.008339768339768,
      "grad_norm": 0.06560596078634262,
      "learning_rate": 9.240669240669241e-06,
      "loss": 1.1757,
      "step": 6540
    },
    {
      "epoch": 8.008597168597168,
      "grad_norm": 0.17579619586467743,
      "learning_rate": 9.23780923780924e-06,
      "loss": 0.8287,
      "step": 6550
    },
    {
      "epoch": 8.008854568854568,
      "grad_norm": 0.3297277092933655,
      "learning_rate": 9.234949234949236e-06,
      "loss": 0.4357,
      "step": 6560
    },
    {
      "epoch": 8.009111969111968,
      "grad_norm": 3.586346387863159,
      "learning_rate": 9.232089232089232e-06,
      "loss": 1.4158,
      "step": 6570
    },
    {
      "epoch": 8.00936936936937,
      "grad_norm": 94.88516998291016,
      "learning_rate": 9.229229229229229e-06,
      "loss": 1.24,
      "step": 6580
    },
    {
      "epoch": 8.00962676962677,
      "grad_norm": 112.45343780517578,
      "learning_rate": 9.226369226369227e-06,
      "loss": 1.1322,
      "step": 6590
    },
    {
      "epoch": 8.00988416988417,
      "grad_norm": 0.0703471377491951,
      "learning_rate": 9.223509223509223e-06,
      "loss": 1.4622,
      "step": 6600
    },
    {
      "epoch": 8.01014157014157,
      "grad_norm": 0.048765502870082855,
      "learning_rate": 9.220649220649222e-06,
      "loss": 1.6374,
      "step": 6610
    },
    {
      "epoch": 8.01039897039897,
      "grad_norm": 0.2779279053211212,
      "learning_rate": 9.217789217789218e-06,
      "loss": 1.0281,
      "step": 6620
    },
    {
      "epoch": 8.01065637065637,
      "grad_norm": 33.617103576660156,
      "learning_rate": 9.214929214929216e-06,
      "loss": 1.1899,
      "step": 6630
    },
    {
      "epoch": 8.01091377091377,
      "grad_norm": 54.08098602294922,
      "learning_rate": 9.212069212069213e-06,
      "loss": 1.3617,
      "step": 6640
    },
    {
      "epoch": 8.01117117117117,
      "grad_norm": 26.711761474609375,
      "learning_rate": 9.20920920920921e-06,
      "loss": 1.1732,
      "step": 6650
    },
    {
      "epoch": 8.01142857142857,
      "grad_norm": 79.53500366210938,
      "learning_rate": 9.206349206349207e-06,
      "loss": 0.6164,
      "step": 6660
    },
    {
      "epoch": 8.011685971685973,
      "grad_norm": 0.0222012996673584,
      "learning_rate": 9.203489203489204e-06,
      "loss": 0.904,
      "step": 6670
    },
    {
      "epoch": 8.011943371943373,
      "grad_norm": 37.63016128540039,
      "learning_rate": 9.2006292006292e-06,
      "loss": 1.8474,
      "step": 6680
    },
    {
      "epoch": 8.012200772200773,
      "grad_norm": 0.025222521275281906,
      "learning_rate": 9.197769197769198e-06,
      "loss": 0.8185,
      "step": 6690
    },
    {
      "epoch": 8.012458172458173,
      "grad_norm": 0.039707981050014496,
      "learning_rate": 9.194909194909197e-06,
      "loss": 1.7299,
      "step": 6700
    },
    {
      "epoch": 8.012715572715573,
      "grad_norm": 45.59052276611328,
      "learning_rate": 9.192049192049193e-06,
      "loss": 0.3896,
      "step": 6710
    },
    {
      "epoch": 8.012972972972973,
      "grad_norm": 2.0984888076782227,
      "learning_rate": 9.189189189189191e-06,
      "loss": 0.8229,
      "step": 6720
    },
    {
      "epoch": 8.013230373230373,
      "grad_norm": 0.26229918003082275,
      "learning_rate": 9.186329186329188e-06,
      "loss": 1.2765,
      "step": 6730
    },
    {
      "epoch": 8.013487773487773,
      "grad_norm": 0.5436154007911682,
      "learning_rate": 9.183469183469184e-06,
      "loss": 0.0315,
      "step": 6740
    },
    {
      "epoch": 8.013745173745173,
      "grad_norm": 0.10348271578550339,
      "learning_rate": 9.18060918060918e-06,
      "loss": 0.8857,
      "step": 6750
    },
    {
      "epoch": 8.014002574002575,
      "grad_norm": 0.26006609201431274,
      "learning_rate": 9.177749177749179e-06,
      "loss": 0.4398,
      "step": 6760
    },
    {
      "epoch": 8.014259974259975,
      "grad_norm": 0.019766027107834816,
      "learning_rate": 9.174889174889175e-06,
      "loss": 0.4339,
      "step": 6770
    },
    {
      "epoch": 8.014517374517375,
      "grad_norm": 0.3928743600845337,
      "learning_rate": 9.172029172029173e-06,
      "loss": 1.7112,
      "step": 6780
    },
    {
      "epoch": 8.014774774774775,
      "grad_norm": 76.59400939941406,
      "learning_rate": 9.16916916916917e-06,
      "loss": 1.4805,
      "step": 6790
    },
    {
      "epoch": 8.015032175032175,
      "grad_norm": 0.12240134179592133,
      "learning_rate": 9.166309166309168e-06,
      "loss": 1.3292,
      "step": 6800
    },
    {
      "epoch": 8.015289575289575,
      "grad_norm": 44.478981018066406,
      "learning_rate": 9.163449163449165e-06,
      "loss": 0.8538,
      "step": 6810
    },
    {
      "epoch": 8.015546975546975,
      "grad_norm": 0.02366860955953598,
      "learning_rate": 9.160589160589161e-06,
      "loss": 2.4329,
      "step": 6820
    },
    {
      "epoch": 8.015804375804375,
      "grad_norm": 0.42346879839897156,
      "learning_rate": 9.157729157729158e-06,
      "loss": 0.5564,
      "step": 6830
    },
    {
      "epoch": 8.016061776061775,
      "grad_norm": 208.15951538085938,
      "learning_rate": 9.154869154869156e-06,
      "loss": 0.8429,
      "step": 6840
    },
    {
      "epoch": 8.016319176319175,
      "grad_norm": 0.2946467995643616,
      "learning_rate": 9.152009152009152e-06,
      "loss": 1.0465,
      "step": 6850
    },
    {
      "epoch": 8.016576576576577,
      "grad_norm": 1.0983664989471436,
      "learning_rate": 9.14914914914915e-06,
      "loss": 1.2315,
      "step": 6860
    },
    {
      "epoch": 8.016833976833977,
      "grad_norm": 0.42272210121154785,
      "learning_rate": 9.146289146289147e-06,
      "loss": 0.5185,
      "step": 6870
    },
    {
      "epoch": 8.017091377091377,
      "grad_norm": 0.46459901332855225,
      "learning_rate": 9.143429143429145e-06,
      "loss": 2.1266,
      "step": 6880
    },
    {
      "epoch": 8.017348777348777,
      "grad_norm": 0.15095072984695435,
      "learning_rate": 9.140569140569141e-06,
      "loss": 0.5774,
      "step": 6890
    },
    {
      "epoch": 8.017606177606178,
      "grad_norm": 0.029569724574685097,
      "learning_rate": 9.137709137709138e-06,
      "loss": 0.3594,
      "step": 6900
    },
    {
      "epoch": 8.017863577863578,
      "grad_norm": 112.70240783691406,
      "learning_rate": 9.134849134849136e-06,
      "loss": 0.9142,
      "step": 6910
    },
    {
      "epoch": 8.018120978120978,
      "grad_norm": 0.057099491357803345,
      "learning_rate": 9.131989131989133e-06,
      "loss": 1.3048,
      "step": 6920
    },
    {
      "epoch": 8.018378378378378,
      "grad_norm": 231.97354125976562,
      "learning_rate": 9.129129129129129e-06,
      "loss": 0.0491,
      "step": 6930
    },
    {
      "epoch": 8.018635778635778,
      "grad_norm": 47.82123565673828,
      "learning_rate": 9.126269126269127e-06,
      "loss": 0.9094,
      "step": 6940
    },
    {
      "epoch": 8.01889317889318,
      "grad_norm": 27.74190902709961,
      "learning_rate": 9.123409123409124e-06,
      "loss": 0.9677,
      "step": 6950
    },
    {
      "epoch": 8.01915057915058,
      "grad_norm": 0.03225734457373619,
      "learning_rate": 9.120549120549122e-06,
      "loss": 1.1965,
      "step": 6960
    },
    {
      "epoch": 8.01940797940798,
      "grad_norm": 0.7187479138374329,
      "learning_rate": 9.117689117689118e-06,
      "loss": 0.9899,
      "step": 6970
    },
    {
      "epoch": 8.01966537966538,
      "grad_norm": 0.08666837960481644,
      "learning_rate": 9.114829114829116e-06,
      "loss": 0.6683,
      "step": 6980
    },
    {
      "epoch": 8.01992277992278,
      "grad_norm": 0.019565997645258904,
      "learning_rate": 9.111969111969113e-06,
      "loss": 1.2084,
      "step": 6990
    },
    {
      "epoch": 8.02,
      "eval_accuracy": 0.8478260869565217,
      "eval_loss": 0.6193599104881287,
      "eval_runtime": 16.653,
      "eval_samples_per_second": 2.762,
      "eval_steps_per_second": 2.762,
      "step": 6993
    },
    {
      "epoch": 9.00018018018018,
      "grad_norm": 0.5390567183494568,
      "learning_rate": 9.10910910910911e-06,
      "loss": 1.2249,
      "step": 7000
    },
    {
      "epoch": 9.00043758043758,
      "grad_norm": 0.24792684614658356,
      "learning_rate": 9.106249106249106e-06,
      "loss": 0.3936,
      "step": 7010
    },
    {
      "epoch": 9.00069498069498,
      "grad_norm": 0.17737895250320435,
      "learning_rate": 9.103389103389104e-06,
      "loss": 1.5842,
      "step": 7020
    },
    {
      "epoch": 9.00095238095238,
      "grad_norm": 0.08357981592416763,
      "learning_rate": 9.1005291005291e-06,
      "loss": 0.2975,
      "step": 7030
    },
    {
      "epoch": 9.00120978120978,
      "grad_norm": 0.15179596841335297,
      "learning_rate": 9.097669097669099e-06,
      "loss": 0.5302,
      "step": 7040
    },
    {
      "epoch": 9.00146718146718,
      "grad_norm": 0.021750060841441154,
      "learning_rate": 9.094809094809095e-06,
      "loss": 0.2233,
      "step": 7050
    },
    {
      "epoch": 9.001724581724583,
      "grad_norm": 357.4986267089844,
      "learning_rate": 9.091949091949093e-06,
      "loss": 1.7797,
      "step": 7060
    },
    {
      "epoch": 9.001981981981983,
      "grad_norm": 0.08269771188497543,
      "learning_rate": 9.08908908908909e-06,
      "loss": 1.0481,
      "step": 7070
    },
    {
      "epoch": 9.002239382239383,
      "grad_norm": 0.02010294981300831,
      "learning_rate": 9.086229086229086e-06,
      "loss": 1.2008,
      "step": 7080
    },
    {
      "epoch": 9.002496782496783,
      "grad_norm": 45.67758560180664,
      "learning_rate": 9.083369083369084e-06,
      "loss": 2.2412,
      "step": 7090
    },
    {
      "epoch": 9.002754182754183,
      "grad_norm": 24.944141387939453,
      "learning_rate": 9.08050908050908e-06,
      "loss": 2.3046,
      "step": 7100
    },
    {
      "epoch": 9.003011583011583,
      "grad_norm": 27.399066925048828,
      "learning_rate": 9.077649077649079e-06,
      "loss": 0.9692,
      "step": 7110
    },
    {
      "epoch": 9.003268983268983,
      "grad_norm": 2.160198450088501,
      "learning_rate": 9.074789074789075e-06,
      "loss": 0.4344,
      "step": 7120
    },
    {
      "epoch": 9.003526383526383,
      "grad_norm": 0.816569983959198,
      "learning_rate": 9.071929071929074e-06,
      "loss": 0.8706,
      "step": 7130
    },
    {
      "epoch": 9.003783783783783,
      "grad_norm": 1.0084000825881958,
      "learning_rate": 9.06906906906907e-06,
      "loss": 0.5086,
      "step": 7140
    },
    {
      "epoch": 9.004041184041185,
      "grad_norm": 0.053576670587062836,
      "learning_rate": 9.066209066209067e-06,
      "loss": 0.8812,
      "step": 7150
    },
    {
      "epoch": 9.004298584298585,
      "grad_norm": 25.146881103515625,
      "learning_rate": 9.063349063349065e-06,
      "loss": 0.5933,
      "step": 7160
    },
    {
      "epoch": 9.004555984555985,
      "grad_norm": 0.5635517239570618,
      "learning_rate": 9.060489060489061e-06,
      "loss": 1.2934,
      "step": 7170
    },
    {
      "epoch": 9.004813384813385,
      "grad_norm": 0.36832961440086365,
      "learning_rate": 9.057629057629058e-06,
      "loss": 1.8208,
      "step": 7180
    },
    {
      "epoch": 9.005070785070785,
      "grad_norm": 0.1600521355867386,
      "learning_rate": 9.054769054769056e-06,
      "loss": 0.3754,
      "step": 7190
    },
    {
      "epoch": 9.005328185328185,
      "grad_norm": 96.4287338256836,
      "learning_rate": 9.051909051909052e-06,
      "loss": 1.0538,
      "step": 7200
    },
    {
      "epoch": 9.005585585585585,
      "grad_norm": 4.262108325958252,
      "learning_rate": 9.04904904904905e-06,
      "loss": 0.4425,
      "step": 7210
    },
    {
      "epoch": 9.005842985842985,
      "grad_norm": 152.94366455078125,
      "learning_rate": 9.046189046189047e-06,
      "loss": 1.0066,
      "step": 7220
    },
    {
      "epoch": 9.006100386100385,
      "grad_norm": 0.028285371139645576,
      "learning_rate": 9.043329043329045e-06,
      "loss": 0.5249,
      "step": 7230
    },
    {
      "epoch": 9.006357786357785,
      "grad_norm": 0.15451128780841827,
      "learning_rate": 9.040469040469042e-06,
      "loss": 1.171,
      "step": 7240
    },
    {
      "epoch": 9.006615186615187,
      "grad_norm": 0.056364186108112335,
      "learning_rate": 9.037609037609038e-06,
      "loss": 0.5535,
      "step": 7250
    },
    {
      "epoch": 9.006872586872587,
      "grad_norm": 29.896148681640625,
      "learning_rate": 9.034749034749034e-06,
      "loss": 1.2453,
      "step": 7260
    },
    {
      "epoch": 9.007129987129987,
      "grad_norm": 0.19092273712158203,
      "learning_rate": 9.031889031889033e-06,
      "loss": 2.4563,
      "step": 7270
    },
    {
      "epoch": 9.007387387387388,
      "grad_norm": 2.0076355934143066,
      "learning_rate": 9.029029029029029e-06,
      "loss": 1.0114,
      "step": 7280
    },
    {
      "epoch": 9.007644787644788,
      "grad_norm": 29.020702362060547,
      "learning_rate": 9.026169026169027e-06,
      "loss": 1.0298,
      "step": 7290
    },
    {
      "epoch": 9.007902187902188,
      "grad_norm": 0.1018686518073082,
      "learning_rate": 9.023309023309024e-06,
      "loss": 1.2876,
      "step": 7300
    },
    {
      "epoch": 9.008159588159588,
      "grad_norm": 36.953407287597656,
      "learning_rate": 9.020449020449022e-06,
      "loss": 0.8832,
      "step": 7310
    },
    {
      "epoch": 9.008416988416988,
      "grad_norm": 0.04455059394240379,
      "learning_rate": 9.017589017589018e-06,
      "loss": 1.0117,
      "step": 7320
    },
    {
      "epoch": 9.008674388674388,
      "grad_norm": 34.155006408691406,
      "learning_rate": 9.014729014729015e-06,
      "loss": 0.6126,
      "step": 7330
    },
    {
      "epoch": 9.00893178893179,
      "grad_norm": 30.061325073242188,
      "learning_rate": 9.011869011869013e-06,
      "loss": 0.6105,
      "step": 7340
    },
    {
      "epoch": 9.00918918918919,
      "grad_norm": 27.52150535583496,
      "learning_rate": 9.00900900900901e-06,
      "loss": 2.3751,
      "step": 7350
    },
    {
      "epoch": 9.00944658944659,
      "grad_norm": 0.4952925443649292,
      "learning_rate": 9.006149006149006e-06,
      "loss": 0.6956,
      "step": 7360
    },
    {
      "epoch": 9.00970398970399,
      "grad_norm": 25.885974884033203,
      "learning_rate": 9.003289003289004e-06,
      "loss": 0.7876,
      "step": 7370
    },
    {
      "epoch": 9.00996138996139,
      "grad_norm": 0.8748680949211121,
      "learning_rate": 9.000429000429e-06,
      "loss": 0.9017,
      "step": 7380
    },
    {
      "epoch": 9.01021879021879,
      "grad_norm": 24.5634708404541,
      "learning_rate": 8.997568997568999e-06,
      "loss": 0.6759,
      "step": 7390
    },
    {
      "epoch": 9.01047619047619,
      "grad_norm": 24.236270904541016,
      "learning_rate": 8.994708994708995e-06,
      "loss": 0.9211,
      "step": 7400
    },
    {
      "epoch": 9.01073359073359,
      "grad_norm": 23.703262329101562,
      "learning_rate": 8.991848991848993e-06,
      "loss": 0.6189,
      "step": 7410
    },
    {
      "epoch": 9.01099099099099,
      "grad_norm": 241.41310119628906,
      "learning_rate": 8.98898898898899e-06,
      "loss": 0.8321,
      "step": 7420
    },
    {
      "epoch": 9.011248391248392,
      "grad_norm": 0.09685607999563217,
      "learning_rate": 8.986128986128986e-06,
      "loss": 0.4043,
      "step": 7430
    },
    {
      "epoch": 9.011505791505792,
      "grad_norm": 0.21560928225517273,
      "learning_rate": 8.983268983268984e-06,
      "loss": 0.2176,
      "step": 7440
    },
    {
      "epoch": 9.011763191763192,
      "grad_norm": 0.07047346979379654,
      "learning_rate": 8.980408980408981e-06,
      "loss": 0.618,
      "step": 7450
    },
    {
      "epoch": 9.012020592020592,
      "grad_norm": 0.04795054346323013,
      "learning_rate": 8.977548977548979e-06,
      "loss": 0.3734,
      "step": 7460
    },
    {
      "epoch": 9.012277992277992,
      "grad_norm": 0.0800858736038208,
      "learning_rate": 8.974688974688976e-06,
      "loss": 0.5159,
      "step": 7470
    },
    {
      "epoch": 9.012535392535392,
      "grad_norm": 0.08157502114772797,
      "learning_rate": 8.971828971828974e-06,
      "loss": 0.0621,
      "step": 7480
    },
    {
      "epoch": 9.012792792792792,
      "grad_norm": 27.266658782958984,
      "learning_rate": 8.96896896896897e-06,
      "loss": 1.2765,
      "step": 7490
    },
    {
      "epoch": 9.013050193050193,
      "grad_norm": 0.17528657615184784,
      "learning_rate": 8.966108966108967e-06,
      "loss": 2.575,
      "step": 7500
    },
    {
      "epoch": 9.013307593307593,
      "grad_norm": 0.031745266169309616,
      "learning_rate": 8.963248963248963e-06,
      "loss": 0.7197,
      "step": 7510
    },
    {
      "epoch": 9.013564993564994,
      "grad_norm": 241.27267456054688,
      "learning_rate": 8.960388960388961e-06,
      "loss": 0.671,
      "step": 7520
    },
    {
      "epoch": 9.013822393822394,
      "grad_norm": 0.04264303669333458,
      "learning_rate": 8.957528957528958e-06,
      "loss": 1.5703,
      "step": 7530
    },
    {
      "epoch": 9.014079794079795,
      "grad_norm": 0.016448037698864937,
      "learning_rate": 8.954668954668956e-06,
      "loss": 0.4535,
      "step": 7540
    },
    {
      "epoch": 9.014337194337195,
      "grad_norm": 0.028359560295939445,
      "learning_rate": 8.951808951808952e-06,
      "loss": 0.6602,
      "step": 7550
    },
    {
      "epoch": 9.014594594594595,
      "grad_norm": 0.04676564037799835,
      "learning_rate": 8.94894894894895e-06,
      "loss": 0.6422,
      "step": 7560
    },
    {
      "epoch": 9.014851994851995,
      "grad_norm": 0.31510064005851746,
      "learning_rate": 8.946088946088947e-06,
      "loss": 0.0245,
      "step": 7570
    },
    {
      "epoch": 9.015109395109395,
      "grad_norm": 0.02477765455842018,
      "learning_rate": 8.943228943228944e-06,
      "loss": 1.7651,
      "step": 7580
    },
    {
      "epoch": 9.015366795366795,
      "grad_norm": 0.023003634065389633,
      "learning_rate": 8.940368940368942e-06,
      "loss": 1.3478,
      "step": 7590
    },
    {
      "epoch": 9.015624195624195,
      "grad_norm": 0.03384900093078613,
      "learning_rate": 8.937508937508938e-06,
      "loss": 0.7807,
      "step": 7600
    },
    {
      "epoch": 9.015881595881597,
      "grad_norm": 35.01763153076172,
      "learning_rate": 8.934648934648935e-06,
      "loss": 0.6444,
      "step": 7610
    },
    {
      "epoch": 9.016138996138997,
      "grad_norm": 0.8073970675468445,
      "learning_rate": 8.931788931788933e-06,
      "loss": 0.0087,
      "step": 7620
    },
    {
      "epoch": 9.016396396396397,
      "grad_norm": 0.2138083279132843,
      "learning_rate": 8.92892892892893e-06,
      "loss": 1.3422,
      "step": 7630
    },
    {
      "epoch": 9.016653796653797,
      "grad_norm": 0.03585595265030861,
      "learning_rate": 8.926068926068927e-06,
      "loss": 0.452,
      "step": 7640
    },
    {
      "epoch": 9.016911196911197,
      "grad_norm": 0.04055719077587128,
      "learning_rate": 8.923208923208924e-06,
      "loss": 0.7551,
      "step": 7650
    },
    {
      "epoch": 9.017168597168597,
      "grad_norm": 25.380233764648438,
      "learning_rate": 8.920348920348922e-06,
      "loss": 0.7222,
      "step": 7660
    },
    {
      "epoch": 9.017425997425997,
      "grad_norm": 0.0834406390786171,
      "learning_rate": 8.917488917488919e-06,
      "loss": 0.8602,
      "step": 7670
    },
    {
      "epoch": 9.017683397683397,
      "grad_norm": 56.41606903076172,
      "learning_rate": 8.914628914628915e-06,
      "loss": 1.9781,
      "step": 7680
    },
    {
      "epoch": 9.017940797940797,
      "grad_norm": 0.08189455419778824,
      "learning_rate": 8.911768911768911e-06,
      "loss": 0.3519,
      "step": 7690
    },
    {
      "epoch": 9.018198198198197,
      "grad_norm": 0.034165192395448685,
      "learning_rate": 8.90890890890891e-06,
      "loss": 1.345,
      "step": 7700
    },
    {
      "epoch": 9.0184555984556,
      "grad_norm": 0.14859578013420105,
      "learning_rate": 8.906048906048906e-06,
      "loss": 0.2301,
      "step": 7710
    },
    {
      "epoch": 9.018712998713,
      "grad_norm": 0.040978193283081055,
      "learning_rate": 8.903188903188904e-06,
      "loss": 0.5242,
      "step": 7720
    },
    {
      "epoch": 9.0189703989704,
      "grad_norm": 1.3178437948226929,
      "learning_rate": 8.9003289003289e-06,
      "loss": 1.1051,
      "step": 7730
    },
    {
      "epoch": 9.0192277992278,
      "grad_norm": 0.22424042224884033,
      "learning_rate": 8.897468897468899e-06,
      "loss": 0.899,
      "step": 7740
    },
    {
      "epoch": 9.0194851994852,
      "grad_norm": 0.023189270868897438,
      "learning_rate": 8.894608894608895e-06,
      "loss": 1.1294,
      "step": 7750
    },
    {
      "epoch": 9.0197425997426,
      "grad_norm": 0.5692568421363831,
      "learning_rate": 8.891748891748892e-06,
      "loss": 1.3751,
      "step": 7760
    },
    {
      "epoch": 9.02,
      "grad_norm": 25.165239334106445,
      "learning_rate": 8.888888888888888e-06,
      "loss": 1.4958,
      "step": 7770
    },
    {
      "epoch": 9.02,
      "eval_accuracy": 0.8478260869565217,
      "eval_loss": 0.6078376173973083,
      "eval_runtime": 13.4317,
      "eval_samples_per_second": 3.425,
      "eval_steps_per_second": 3.425,
      "step": 7770
    },
    {
      "epoch": 10.0002574002574,
      "grad_norm": 0.1204485297203064,
      "learning_rate": 8.886028886028886e-06,
      "loss": 0.5537,
      "step": 7780
    },
    {
      "epoch": 10.0005148005148,
      "grad_norm": 0.06360988318920135,
      "learning_rate": 8.883168883168883e-06,
      "loss": 1.546,
      "step": 7790
    },
    {
      "epoch": 10.0007722007722,
      "grad_norm": 120.32550048828125,
      "learning_rate": 8.880308880308881e-06,
      "loss": 0.3902,
      "step": 7800
    },
    {
      "epoch": 10.0010296010296,
      "grad_norm": 0.15769629180431366,
      "learning_rate": 8.877448877448878e-06,
      "loss": 0.0061,
      "step": 7810
    },
    {
      "epoch": 10.001287001287002,
      "grad_norm": 0.497676819562912,
      "learning_rate": 8.874588874588876e-06,
      "loss": 0.95,
      "step": 7820
    },
    {
      "epoch": 10.001544401544402,
      "grad_norm": 6.817390441894531,
      "learning_rate": 8.871728871728872e-06,
      "loss": 0.7795,
      "step": 7830
    },
    {
      "epoch": 10.001801801801802,
      "grad_norm": 25.054075241088867,
      "learning_rate": 8.86886886886887e-06,
      "loss": 1.6924,
      "step": 7840
    },
    {
      "epoch": 10.002059202059202,
      "grad_norm": 0.5180888772010803,
      "learning_rate": 8.866008866008867e-06,
      "loss": 1.7818,
      "step": 7850
    },
    {
      "epoch": 10.002316602316602,
      "grad_norm": 96.92232513427734,
      "learning_rate": 8.863148863148863e-06,
      "loss": 1.1024,
      "step": 7860
    },
    {
      "epoch": 10.002574002574002,
      "grad_norm": 23.87799644470215,
      "learning_rate": 8.860288860288861e-06,
      "loss": 0.3728,
      "step": 7870
    },
    {
      "epoch": 10.002831402831402,
      "grad_norm": 0.5069870948791504,
      "learning_rate": 8.857428857428858e-06,
      "loss": 0.8391,
      "step": 7880
    },
    {
      "epoch": 10.003088803088803,
      "grad_norm": 0.02661108784377575,
      "learning_rate": 8.854568854568856e-06,
      "loss": 0.8022,
      "step": 7890
    },
    {
      "epoch": 10.003346203346203,
      "grad_norm": 0.21478380262851715,
      "learning_rate": 8.851708851708853e-06,
      "loss": 0.0109,
      "step": 7900
    },
    {
      "epoch": 10.003603603603604,
      "grad_norm": 3.073247194290161,
      "learning_rate": 8.84884884884885e-06,
      "loss": 1.2947,
      "step": 7910
    },
    {
      "epoch": 10.003861003861005,
      "grad_norm": 274.8877868652344,
      "learning_rate": 8.845988845988847e-06,
      "loss": 0.5697,
      "step": 7920
    },
    {
      "epoch": 10.004118404118405,
      "grad_norm": 37.88632583618164,
      "learning_rate": 8.843128843128844e-06,
      "loss": 1.3985,
      "step": 7930
    },
    {
      "epoch": 10.004375804375805,
      "grad_norm": 0.012297092005610466,
      "learning_rate": 8.84026884026884e-06,
      "loss": 0.4317,
      "step": 7940
    },
    {
      "epoch": 10.004633204633205,
      "grad_norm": 0.17633600533008575,
      "learning_rate": 8.837408837408838e-06,
      "loss": 0.6907,
      "step": 7950
    },
    {
      "epoch": 10.004890604890605,
      "grad_norm": 0.00784577988088131,
      "learning_rate": 8.834548834548835e-06,
      "loss": 1.0013,
      "step": 7960
    },
    {
      "epoch": 10.005148005148005,
      "grad_norm": 0.46426498889923096,
      "learning_rate": 8.831688831688833e-06,
      "loss": 1.081,
      "step": 7970
    },
    {
      "epoch": 10.005405405405405,
      "grad_norm": 0.21582862734794617,
      "learning_rate": 8.82882882882883e-06,
      "loss": 1.0164,
      "step": 7980
    },
    {
      "epoch": 10.005662805662805,
      "grad_norm": 159.87257385253906,
      "learning_rate": 8.825968825968828e-06,
      "loss": 1.0832,
      "step": 7990
    },
    {
      "epoch": 10.005920205920205,
      "grad_norm": 0.01719667576253414,
      "learning_rate": 8.823108823108824e-06,
      "loss": 0.9401,
      "step": 8000
    },
    {
      "epoch": 10.006177606177607,
      "grad_norm": 43.231407165527344,
      "learning_rate": 8.82024882024882e-06,
      "loss": 1.3899,
      "step": 8010
    },
    {
      "epoch": 10.006435006435007,
      "grad_norm": 0.17739209532737732,
      "learning_rate": 8.817388817388817e-06,
      "loss": 0.1192,
      "step": 8020
    },
    {
      "epoch": 10.006692406692407,
      "grad_norm": 44.295108795166016,
      "learning_rate": 8.814528814528815e-06,
      "loss": 1.9272,
      "step": 8030
    },
    {
      "epoch": 10.006949806949807,
      "grad_norm": 0.03208237141370773,
      "learning_rate": 8.811668811668812e-06,
      "loss": 0.9453,
      "step": 8040
    },
    {
      "epoch": 10.007207207207207,
      "grad_norm": 246.54135131835938,
      "learning_rate": 8.80880880880881e-06,
      "loss": 0.586,
      "step": 8050
    },
    {
      "epoch": 10.007464607464607,
      "grad_norm": 0.024292565882205963,
      "learning_rate": 8.805948805948806e-06,
      "loss": 0.306,
      "step": 8060
    },
    {
      "epoch": 10.007722007722007,
      "grad_norm": 0.043888118118047714,
      "learning_rate": 8.803088803088804e-06,
      "loss": 0.6646,
      "step": 8070
    },
    {
      "epoch": 10.007979407979407,
      "grad_norm": 0.05276341363787651,
      "learning_rate": 8.800228800228801e-06,
      "loss": 1.8646,
      "step": 8080
    },
    {
      "epoch": 10.008236808236807,
      "grad_norm": 3.0538644790649414,
      "learning_rate": 8.797368797368799e-06,
      "loss": 1.7883,
      "step": 8090
    },
    {
      "epoch": 10.00849420849421,
      "grad_norm": 0.43052423000335693,
      "learning_rate": 8.794508794508795e-06,
      "loss": 1.537,
      "step": 8100
    },
    {
      "epoch": 10.00875160875161,
      "grad_norm": 0.2308303564786911,
      "learning_rate": 8.791648791648792e-06,
      "loss": 0.6922,
      "step": 8110
    },
    {
      "epoch": 10.00900900900901,
      "grad_norm": 0.9070534706115723,
      "learning_rate": 8.788788788788788e-06,
      "loss": 0.7481,
      "step": 8120
    },
    {
      "epoch": 10.00926640926641,
      "grad_norm": 0.32801300287246704,
      "learning_rate": 8.785928785928787e-06,
      "loss": 0.6439,
      "step": 8130
    },
    {
      "epoch": 10.00952380952381,
      "grad_norm": 220.7642822265625,
      "learning_rate": 8.783068783068783e-06,
      "loss": 1.4445,
      "step": 8140
    },
    {
      "epoch": 10.00978120978121,
      "grad_norm": 0.18322208523750305,
      "learning_rate": 8.780208780208781e-06,
      "loss": 0.5422,
      "step": 8150
    },
    {
      "epoch": 10.01003861003861,
      "grad_norm": 0.15416328608989716,
      "learning_rate": 8.777348777348778e-06,
      "loss": 1.7991,
      "step": 8160
    },
    {
      "epoch": 10.01029601029601,
      "grad_norm": 0.02595771849155426,
      "learning_rate": 8.774488774488776e-06,
      "loss": 0.163,
      "step": 8170
    },
    {
      "epoch": 10.01055341055341,
      "grad_norm": 131.53567504882812,
      "learning_rate": 8.771628771628772e-06,
      "loss": 1.3368,
      "step": 8180
    },
    {
      "epoch": 10.010810810810812,
      "grad_norm": 0.39858779311180115,
      "learning_rate": 8.768768768768769e-06,
      "loss": 0.4107,
      "step": 8190
    },
    {
      "epoch": 10.011068211068212,
      "grad_norm": 0.34023207426071167,
      "learning_rate": 8.765908765908767e-06,
      "loss": 0.6566,
      "step": 8200
    },
    {
      "epoch": 10.011325611325612,
      "grad_norm": 0.4535696506500244,
      "learning_rate": 8.763048763048763e-06,
      "loss": 1.5843,
      "step": 8210
    },
    {
      "epoch": 10.011583011583012,
      "grad_norm": 0.04021540284156799,
      "learning_rate": 8.760188760188762e-06,
      "loss": 0.3503,
      "step": 8220
    },
    {
      "epoch": 10.011840411840412,
      "grad_norm": 46.56422424316406,
      "learning_rate": 8.757328757328758e-06,
      "loss": 1.2619,
      "step": 8230
    },
    {
      "epoch": 10.012097812097812,
      "grad_norm": 40.27321243286133,
      "learning_rate": 8.754468754468756e-06,
      "loss": 0.7487,
      "step": 8240
    },
    {
      "epoch": 10.012355212355212,
      "grad_norm": 25.20823097229004,
      "learning_rate": 8.751608751608753e-06,
      "loss": 1.2994,
      "step": 8250
    },
    {
      "epoch": 10.012612612612612,
      "grad_norm": 108.85746002197266,
      "learning_rate": 8.74874874874875e-06,
      "loss": 2.4518,
      "step": 8260
    },
    {
      "epoch": 10.012870012870012,
      "grad_norm": 0.0417940616607666,
      "learning_rate": 8.745888745888746e-06,
      "loss": 0.7443,
      "step": 8270
    },
    {
      "epoch": 10.013127413127414,
      "grad_norm": 0.008936642669141293,
      "learning_rate": 8.743028743028744e-06,
      "loss": 0.7142,
      "step": 8280
    },
    {
      "epoch": 10.013384813384814,
      "grad_norm": 0.036207396537065506,
      "learning_rate": 8.74016874016874e-06,
      "loss": 1.2926,
      "step": 8290
    },
    {
      "epoch": 10.013642213642214,
      "grad_norm": 0.016090108081698418,
      "learning_rate": 8.737308737308738e-06,
      "loss": 0.7859,
      "step": 8300
    },
    {
      "epoch": 10.013899613899614,
      "grad_norm": 0.013586858287453651,
      "learning_rate": 8.734448734448735e-06,
      "loss": 0.4006,
      "step": 8310
    },
    {
      "epoch": 10.014157014157014,
      "grad_norm": 0.01622426137328148,
      "learning_rate": 8.731588731588733e-06,
      "loss": 1.5877,
      "step": 8320
    },
    {
      "epoch": 10.014414414414414,
      "grad_norm": 0.1665509194135666,
      "learning_rate": 8.72872872872873e-06,
      "loss": 0.8227,
      "step": 8330
    },
    {
      "epoch": 10.014671814671814,
      "grad_norm": 0.13879388570785522,
      "learning_rate": 8.725868725868728e-06,
      "loss": 1.5522,
      "step": 8340
    },
    {
      "epoch": 10.014929214929214,
      "grad_norm": 44.89026641845703,
      "learning_rate": 8.723008723008724e-06,
      "loss": 1.3654,
      "step": 8350
    },
    {
      "epoch": 10.015186615186614,
      "grad_norm": 0.5491271018981934,
      "learning_rate": 8.72014872014872e-06,
      "loss": 1.0478,
      "step": 8360
    },
    {
      "epoch": 10.015444015444016,
      "grad_norm": 10.410693168640137,
      "learning_rate": 8.717288717288717e-06,
      "loss": 0.7527,
      "step": 8370
    },
    {
      "epoch": 10.015701415701416,
      "grad_norm": 0.08085234463214874,
      "learning_rate": 8.714428714428715e-06,
      "loss": 1.1404,
      "step": 8380
    },
    {
      "epoch": 10.015958815958816,
      "grad_norm": 0.05885207653045654,
      "learning_rate": 8.711568711568712e-06,
      "loss": 0.6541,
      "step": 8390
    },
    {
      "epoch": 10.016216216216216,
      "grad_norm": 0.12580057978630066,
      "learning_rate": 8.70870870870871e-06,
      "loss": 1.6544,
      "step": 8400
    },
    {
      "epoch": 10.016473616473617,
      "grad_norm": 0.2117607593536377,
      "learning_rate": 8.705848705848706e-06,
      "loss": 0.9909,
      "step": 8410
    },
    {
      "epoch": 10.016731016731017,
      "grad_norm": 0.22397179901599884,
      "learning_rate": 8.702988702988705e-06,
      "loss": 0.8806,
      "step": 8420
    },
    {
      "epoch": 10.016988416988417,
      "grad_norm": 25.2739200592041,
      "learning_rate": 8.700128700128701e-06,
      "loss": 1.1715,
      "step": 8430
    },
    {
      "epoch": 10.017245817245817,
      "grad_norm": 36.23960876464844,
      "learning_rate": 8.697268697268697e-06,
      "loss": 1.3008,
      "step": 8440
    },
    {
      "epoch": 10.017503217503217,
      "grad_norm": 0.24737899005413055,
      "learning_rate": 8.694408694408694e-06,
      "loss": 0.7906,
      "step": 8450
    },
    {
      "epoch": 10.017760617760617,
      "grad_norm": 37.60052490234375,
      "learning_rate": 8.691548691548692e-06,
      "loss": 1.56,
      "step": 8460
    },
    {
      "epoch": 10.018018018018019,
      "grad_norm": 56.47618865966797,
      "learning_rate": 8.688688688688689e-06,
      "loss": 1.2833,
      "step": 8470
    },
    {
      "epoch": 10.018275418275419,
      "grad_norm": 6.734436511993408,
      "learning_rate": 8.685828685828687e-06,
      "loss": 0.9528,
      "step": 8480
    },
    {
      "epoch": 10.018532818532819,
      "grad_norm": 9.789285659790039,
      "learning_rate": 8.682968682968683e-06,
      "loss": 1.0823,
      "step": 8490
    },
    {
      "epoch": 10.018790218790219,
      "grad_norm": 69.39130401611328,
      "learning_rate": 8.680108680108681e-06,
      "loss": 0.7756,
      "step": 8500
    },
    {
      "epoch": 10.019047619047619,
      "grad_norm": 0.3788280189037323,
      "learning_rate": 8.677248677248678e-06,
      "loss": 1.2168,
      "step": 8510
    },
    {
      "epoch": 10.019305019305019,
      "grad_norm": 32.36293411254883,
      "learning_rate": 8.674388674388674e-06,
      "loss": 1.4968,
      "step": 8520
    },
    {
      "epoch": 10.019562419562419,
      "grad_norm": 120.9354476928711,
      "learning_rate": 8.671528671528672e-06,
      "loss": 1.799,
      "step": 8530
    },
    {
      "epoch": 10.01981981981982,
      "grad_norm": 0.5419692993164062,
      "learning_rate": 8.668668668668669e-06,
      "loss": 1.7393,
      "step": 8540
    },
    {
      "epoch": 10.02,
      "eval_accuracy": 0.7608695652173914,
      "eval_loss": 0.7950884699821472,
      "eval_runtime": 13.4666,
      "eval_samples_per_second": 3.416,
      "eval_steps_per_second": 3.416,
      "step": 8547
    },
    {
      "epoch": 11.00007722007722,
      "grad_norm": 0.28460660576820374,
      "learning_rate": 8.665808665808665e-06,
      "loss": 0.878,
      "step": 8550
    },
    {
      "epoch": 11.00033462033462,
      "grad_norm": 24.105052947998047,
      "learning_rate": 8.662948662948664e-06,
      "loss": 1.2981,
      "step": 8560
    },
    {
      "epoch": 11.00059202059202,
      "grad_norm": 39.04997634887695,
      "learning_rate": 8.660088660088662e-06,
      "loss": 0.4637,
      "step": 8570
    },
    {
      "epoch": 11.000849420849422,
      "grad_norm": 0.8172537088394165,
      "learning_rate": 8.657228657228658e-06,
      "loss": 0.9727,
      "step": 8580
    },
    {
      "epoch": 11.001106821106822,
      "grad_norm": 0.018035829067230225,
      "learning_rate": 8.654368654368656e-06,
      "loss": 0.8219,
      "step": 8590
    },
    {
      "epoch": 11.001364221364222,
      "grad_norm": 100.51970672607422,
      "learning_rate": 8.651508651508653e-06,
      "loss": 0.4128,
      "step": 8600
    },
    {
      "epoch": 11.001621621621622,
      "grad_norm": 0.02792496792972088,
      "learning_rate": 8.64864864864865e-06,
      "loss": 1.9699,
      "step": 8610
    },
    {
      "epoch": 11.001879021879022,
      "grad_norm": 54.30400085449219,
      "learning_rate": 8.645788645788646e-06,
      "loss": 2.0642,
      "step": 8620
    },
    {
      "epoch": 11.002136422136422,
      "grad_norm": 72.56423950195312,
      "learning_rate": 8.642928642928644e-06,
      "loss": 0.299,
      "step": 8630
    },
    {
      "epoch": 11.002393822393822,
      "grad_norm": 0.7110329866409302,
      "learning_rate": 8.64006864006864e-06,
      "loss": 0.1974,
      "step": 8640
    },
    {
      "epoch": 11.002651222651222,
      "grad_norm": 587.3732299804688,
      "learning_rate": 8.637208637208639e-06,
      "loss": 0.9534,
      "step": 8650
    },
    {
      "epoch": 11.002908622908622,
      "grad_norm": 6.628912448883057,
      "learning_rate": 8.634348634348635e-06,
      "loss": 1.8159,
      "step": 8660
    },
    {
      "epoch": 11.003166023166024,
      "grad_norm": 35.30387878417969,
      "learning_rate": 8.631488631488633e-06,
      "loss": 0.3854,
      "step": 8670
    },
    {
      "epoch": 11.003423423423424,
      "grad_norm": 0.421171098947525,
      "learning_rate": 8.62862862862863e-06,
      "loss": 1.1167,
      "step": 8680
    },
    {
      "epoch": 11.003680823680824,
      "grad_norm": 0.05929442122578621,
      "learning_rate": 8.625768625768626e-06,
      "loss": 0.8298,
      "step": 8690
    },
    {
      "epoch": 11.003938223938224,
      "grad_norm": 0.041819799691438675,
      "learning_rate": 8.622908622908623e-06,
      "loss": 1.2108,
      "step": 8700
    },
    {
      "epoch": 11.004195624195624,
      "grad_norm": 0.06778862327337265,
      "learning_rate": 8.62004862004862e-06,
      "loss": 0.7713,
      "step": 8710
    },
    {
      "epoch": 11.004453024453024,
      "grad_norm": 40.76213455200195,
      "learning_rate": 8.617188617188617e-06,
      "loss": 1.4024,
      "step": 8720
    },
    {
      "epoch": 11.004710424710424,
      "grad_norm": 0.07441040128469467,
      "learning_rate": 8.614328614328615e-06,
      "loss": 0.7176,
      "step": 8730
    },
    {
      "epoch": 11.004967824967824,
      "grad_norm": 0.007878212258219719,
      "learning_rate": 8.611468611468612e-06,
      "loss": 0.0048,
      "step": 8740
    },
    {
      "epoch": 11.005225225225225,
      "grad_norm": 0.04463992267847061,
      "learning_rate": 8.60860860860861e-06,
      "loss": 0.4363,
      "step": 8750
    },
    {
      "epoch": 11.005482625482626,
      "grad_norm": 87.40447235107422,
      "learning_rate": 8.605748605748607e-06,
      "loss": 0.7171,
      "step": 8760
    },
    {
      "epoch": 11.005740025740026,
      "grad_norm": 0.3318406343460083,
      "learning_rate": 8.602888602888605e-06,
      "loss": 1.4153,
      "step": 8770
    },
    {
      "epoch": 11.005997425997426,
      "grad_norm": 0.25054681301116943,
      "learning_rate": 8.600028600028601e-06,
      "loss": 0.4377,
      "step": 8780
    },
    {
      "epoch": 11.006254826254827,
      "grad_norm": 0.2337009757757187,
      "learning_rate": 8.597168597168598e-06,
      "loss": 0.0044,
      "step": 8790
    },
    {
      "epoch": 11.006512226512227,
      "grad_norm": 0.26338788866996765,
      "learning_rate": 8.594308594308594e-06,
      "loss": 0.4798,
      "step": 8800
    },
    {
      "epoch": 11.006769626769627,
      "grad_norm": 24.581886291503906,
      "learning_rate": 8.591448591448592e-06,
      "loss": 2.3055,
      "step": 8810
    },
    {
      "epoch": 11.007027027027027,
      "grad_norm": 0.10813351720571518,
      "learning_rate": 8.588588588588589e-06,
      "loss": 0.1178,
      "step": 8820
    },
    {
      "epoch": 11.007284427284427,
      "grad_norm": 313.4660949707031,
      "learning_rate": 8.585728585728587e-06,
      "loss": 2.1168,
      "step": 8830
    },
    {
      "epoch": 11.007541827541827,
      "grad_norm": 3.0175836086273193,
      "learning_rate": 8.582868582868583e-06,
      "loss": 1.697,
      "step": 8840
    },
    {
      "epoch": 11.007799227799227,
      "grad_norm": 0.008230429142713547,
      "learning_rate": 8.580008580008582e-06,
      "loss": 1.0399,
      "step": 8850
    },
    {
      "epoch": 11.008056628056629,
      "grad_norm": 24.00910758972168,
      "learning_rate": 8.577148577148578e-06,
      "loss": 0.7707,
      "step": 8860
    },
    {
      "epoch": 11.008314028314029,
      "grad_norm": 160.81443786621094,
      "learning_rate": 8.574288574288574e-06,
      "loss": 1.0067,
      "step": 8870
    },
    {
      "epoch": 11.008571428571429,
      "grad_norm": 250.05177307128906,
      "learning_rate": 8.571428571428571e-06,
      "loss": 0.5634,
      "step": 8880
    },
    {
      "epoch": 11.008828828828829,
      "grad_norm": 0.2900729179382324,
      "learning_rate": 8.568568568568569e-06,
      "loss": 1.7788,
      "step": 8890
    },
    {
      "epoch": 11.009086229086229,
      "grad_norm": 0.0074234092608094215,
      "learning_rate": 8.565708565708566e-06,
      "loss": 1.3323,
      "step": 8900
    },
    {
      "epoch": 11.009343629343629,
      "grad_norm": 0.32027724385261536,
      "learning_rate": 8.562848562848564e-06,
      "loss": 0.5043,
      "step": 8910
    },
    {
      "epoch": 11.00960102960103,
      "grad_norm": 2.96531343460083,
      "learning_rate": 8.55998855998856e-06,
      "loss": 0.6105,
      "step": 8920
    },
    {
      "epoch": 11.00985842985843,
      "grad_norm": 0.34536412358283997,
      "learning_rate": 8.557128557128558e-06,
      "loss": 0.9915,
      "step": 8930
    },
    {
      "epoch": 11.01011583011583,
      "grad_norm": 5.191471099853516,
      "learning_rate": 8.554268554268555e-06,
      "loss": 0.5698,
      "step": 8940
    },
    {
      "epoch": 11.010373230373231,
      "grad_norm": 0.3449559509754181,
      "learning_rate": 8.551408551408551e-06,
      "loss": 0.004,
      "step": 8950
    },
    {
      "epoch": 11.010630630630631,
      "grad_norm": 0.012626473791897297,
      "learning_rate": 8.54854854854855e-06,
      "loss": 1.3268,
      "step": 8960
    },
    {
      "epoch": 11.010888030888031,
      "grad_norm": 0.15182413160800934,
      "learning_rate": 8.545688545688546e-06,
      "loss": 1.576,
      "step": 8970
    },
    {
      "epoch": 11.011145431145431,
      "grad_norm": 0.5726296305656433,
      "learning_rate": 8.542828542828544e-06,
      "loss": 0.606,
      "step": 8980
    },
    {
      "epoch": 11.011402831402831,
      "grad_norm": 0.0075800674967467785,
      "learning_rate": 8.53996853996854e-06,
      "loss": 0.4473,
      "step": 8990
    },
    {
      "epoch": 11.011660231660231,
      "grad_norm": 25.251062393188477,
      "learning_rate": 8.537108537108539e-06,
      "loss": 2.1039,
      "step": 9000
    },
    {
      "epoch": 11.011917631917632,
      "grad_norm": 22.96194076538086,
      "learning_rate": 8.534248534248535e-06,
      "loss": 2.052,
      "step": 9010
    },
    {
      "epoch": 11.012175032175032,
      "grad_norm": 75.685791015625,
      "learning_rate": 8.531388531388533e-06,
      "loss": 1.1469,
      "step": 9020
    },
    {
      "epoch": 11.012432432432432,
      "grad_norm": 0.6149288415908813,
      "learning_rate": 8.52852852852853e-06,
      "loss": 0.4877,
      "step": 9030
    },
    {
      "epoch": 11.012689832689833,
      "grad_norm": 0.15368354320526123,
      "learning_rate": 8.525668525668526e-06,
      "loss": 0.6674,
      "step": 9040
    },
    {
      "epoch": 11.012947232947234,
      "grad_norm": 277.51312255859375,
      "learning_rate": 8.522808522808523e-06,
      "loss": 2.0748,
      "step": 9050
    },
    {
      "epoch": 11.013204633204634,
      "grad_norm": 0.009949028491973877,
      "learning_rate": 8.519948519948521e-06,
      "loss": 0.382,
      "step": 9060
    },
    {
      "epoch": 11.013462033462034,
      "grad_norm": 33.68278121948242,
      "learning_rate": 8.517088517088517e-06,
      "loss": 0.9996,
      "step": 9070
    },
    {
      "epoch": 11.013719433719434,
      "grad_norm": 0.07025768607854843,
      "learning_rate": 8.514228514228516e-06,
      "loss": 0.7772,
      "step": 9080
    },
    {
      "epoch": 11.013976833976834,
      "grad_norm": 0.5183196663856506,
      "learning_rate": 8.511368511368512e-06,
      "loss": 0.0227,
      "step": 9090
    },
    {
      "epoch": 11.014234234234234,
      "grad_norm": 0.3555748462677002,
      "learning_rate": 8.50850850850851e-06,
      "loss": 2.1808,
      "step": 9100
    },
    {
      "epoch": 11.014491634491634,
      "grad_norm": 383.48785400390625,
      "learning_rate": 8.505648505648507e-06,
      "loss": 1.5736,
      "step": 9110
    },
    {
      "epoch": 11.014749034749034,
      "grad_norm": 0.4379149377346039,
      "learning_rate": 8.502788502788503e-06,
      "loss": 0.3884,
      "step": 9120
    },
    {
      "epoch": 11.015006435006436,
      "grad_norm": 0.4699688255786896,
      "learning_rate": 8.4999284999285e-06,
      "loss": 0.4054,
      "step": 9130
    },
    {
      "epoch": 11.015263835263836,
      "grad_norm": 1.8073198795318604,
      "learning_rate": 8.497068497068498e-06,
      "loss": 0.4486,
      "step": 9140
    },
    {
      "epoch": 11.015521235521236,
      "grad_norm": 297.4031066894531,
      "learning_rate": 8.494208494208494e-06,
      "loss": 1.0507,
      "step": 9150
    },
    {
      "epoch": 11.015778635778636,
      "grad_norm": 0.2800242602825165,
      "learning_rate": 8.491348491348492e-06,
      "loss": 0.8891,
      "step": 9160
    },
    {
      "epoch": 11.016036036036036,
      "grad_norm": 0.012133017182350159,
      "learning_rate": 8.488488488488489e-06,
      "loss": 0.4816,
      "step": 9170
    },
    {
      "epoch": 11.016293436293436,
      "grad_norm": 0.12698222696781158,
      "learning_rate": 8.485628485628487e-06,
      "loss": 1.0889,
      "step": 9180
    },
    {
      "epoch": 11.016550836550836,
      "grad_norm": 0.012583434581756592,
      "learning_rate": 8.482768482768483e-06,
      "loss": 1.4182,
      "step": 9190
    },
    {
      "epoch": 11.016808236808236,
      "grad_norm": 0.011629793792963028,
      "learning_rate": 8.47990847990848e-06,
      "loss": 0.3926,
      "step": 9200
    },
    {
      "epoch": 11.017065637065636,
      "grad_norm": 297.43707275390625,
      "learning_rate": 8.477048477048478e-06,
      "loss": 1.2325,
      "step": 9210
    },
    {
      "epoch": 11.017323037323038,
      "grad_norm": 1184.72265625,
      "learning_rate": 8.474188474188475e-06,
      "loss": 1.7258,
      "step": 9220
    },
    {
      "epoch": 11.017580437580438,
      "grad_norm": 0.061885394155979156,
      "learning_rate": 8.471328471328471e-06,
      "loss": 0.7547,
      "step": 9230
    },
    {
      "epoch": 11.017837837837838,
      "grad_norm": 101.12828826904297,
      "learning_rate": 8.46846846846847e-06,
      "loss": 1.0483,
      "step": 9240
    },
    {
      "epoch": 11.018095238095238,
      "grad_norm": 0.5181973576545715,
      "learning_rate": 8.465608465608466e-06,
      "loss": 0.4266,
      "step": 9250
    },
    {
      "epoch": 11.018352638352638,
      "grad_norm": 1.9151122570037842,
      "learning_rate": 8.462748462748464e-06,
      "loss": 1.0087,
      "step": 9260
    },
    {
      "epoch": 11.018610038610039,
      "grad_norm": 0.022047946229577065,
      "learning_rate": 8.45988845988846e-06,
      "loss": 0.6823,
      "step": 9270
    },
    {
      "epoch": 11.018867438867439,
      "grad_norm": 417.81298828125,
      "learning_rate": 8.457028457028458e-06,
      "loss": 0.3463,
      "step": 9280
    },
    {
      "epoch": 11.019124839124839,
      "grad_norm": 10.910456657409668,
      "learning_rate": 8.454168454168455e-06,
      "loss": 0.8723,
      "step": 9290
    },
    {
      "epoch": 11.019382239382239,
      "grad_norm": 0.32493460178375244,
      "learning_rate": 8.451308451308451e-06,
      "loss": 0.6945,
      "step": 9300
    },
    {
      "epoch": 11.019639639639639,
      "grad_norm": 0.010908570140600204,
      "learning_rate": 8.44844844844845e-06,
      "loss": 0.0103,
      "step": 9310
    },
    {
      "epoch": 11.01989703989704,
      "grad_norm": 0.07723493129014969,
      "learning_rate": 8.445588445588446e-06,
      "loss": 1.0987,
      "step": 9320
    },
    {
      "epoch": 11.02,
      "eval_accuracy": 0.8260869565217391,
      "eval_loss": 0.7859626412391663,
      "eval_runtime": 13.4764,
      "eval_samples_per_second": 3.413,
      "eval_steps_per_second": 3.413,
      "step": 9324
    },
    {
      "epoch": 12.00015444015444,
      "grad_norm": 0.47433146834373474,
      "learning_rate": 8.442728442728444e-06,
      "loss": 1.2386,
      "step": 9330
    },
    {
      "epoch": 12.000411840411841,
      "grad_norm": 0.011013006791472435,
      "learning_rate": 8.43986843986844e-06,
      "loss": 0.378,
      "step": 9340
    },
    {
      "epoch": 12.000669240669241,
      "grad_norm": 0.00891246646642685,
      "learning_rate": 8.437008437008439e-06,
      "loss": 0.3884,
      "step": 9350
    },
    {
      "epoch": 12.000926640926641,
      "grad_norm": 0.01927793025970459,
      "learning_rate": 8.434148434148435e-06,
      "loss": 0.5555,
      "step": 9360
    },
    {
      "epoch": 12.001184041184041,
      "grad_norm": 2.8038787841796875,
      "learning_rate": 8.431288431288432e-06,
      "loss": 1.0693,
      "step": 9370
    },
    {
      "epoch": 12.001441441441441,
      "grad_norm": 802.3114013671875,
      "learning_rate": 8.428428428428428e-06,
      "loss": 1.8102,
      "step": 9380
    },
    {
      "epoch": 12.001698841698841,
      "grad_norm": 0.3516940176486969,
      "learning_rate": 8.425568425568426e-06,
      "loss": 0.9108,
      "step": 9390
    },
    {
      "epoch": 12.001956241956242,
      "grad_norm": 0.00919923186302185,
      "learning_rate": 8.422708422708423e-06,
      "loss": 0.8663,
      "step": 9400
    },
    {
      "epoch": 12.002213642213642,
      "grad_norm": 0.17423267662525177,
      "learning_rate": 8.419848419848421e-06,
      "loss": 1.6467,
      "step": 9410
    },
    {
      "epoch": 12.002471042471042,
      "grad_norm": 0.010266111232340336,
      "learning_rate": 8.416988416988418e-06,
      "loss": 0.4335,
      "step": 9420
    },
    {
      "epoch": 12.002728442728444,
      "grad_norm": 0.29146385192871094,
      "learning_rate": 8.414128414128416e-06,
      "loss": 0.7404,
      "step": 9430
    },
    {
      "epoch": 12.002985842985844,
      "grad_norm": 0.3227342963218689,
      "learning_rate": 8.411268411268412e-06,
      "loss": 0.4085,
      "step": 9440
    },
    {
      "epoch": 12.003243243243244,
      "grad_norm": 46.19145965576172,
      "learning_rate": 8.408408408408409e-06,
      "loss": 1.2482,
      "step": 9450
    },
    {
      "epoch": 12.003500643500644,
      "grad_norm": 0.007554527837783098,
      "learning_rate": 8.405548405548407e-06,
      "loss": 1.2104,
      "step": 9460
    },
    {
      "epoch": 12.003758043758044,
      "grad_norm": 0.6184163689613342,
      "learning_rate": 8.402688402688403e-06,
      "loss": 0.7161,
      "step": 9470
    },
    {
      "epoch": 12.004015444015444,
      "grad_norm": 1685.7156982421875,
      "learning_rate": 8.3998283998284e-06,
      "loss": 1.4936,
      "step": 9480
    },
    {
      "epoch": 12.004272844272844,
      "grad_norm": 410.31732177734375,
      "learning_rate": 8.396968396968398e-06,
      "loss": 1.1704,
      "step": 9490
    },
    {
      "epoch": 12.004530244530244,
      "grad_norm": 15.203624725341797,
      "learning_rate": 8.394108394108394e-06,
      "loss": 0.0113,
      "step": 9500
    },
    {
      "epoch": 12.004787644787644,
      "grad_norm": 0.34546852111816406,
      "learning_rate": 8.391248391248393e-06,
      "loss": 0.9388,
      "step": 9510
    },
    {
      "epoch": 12.005045045045046,
      "grad_norm": 0.10535982996225357,
      "learning_rate": 8.388388388388389e-06,
      "loss": 0.4243,
      "step": 9520
    },
    {
      "epoch": 12.005302445302446,
      "grad_norm": 0.005769877228885889,
      "learning_rate": 8.385528385528387e-06,
      "loss": 0.8682,
      "step": 9530
    },
    {
      "epoch": 12.005559845559846,
      "grad_norm": 0.0826544538140297,
      "learning_rate": 8.382668382668384e-06,
      "loss": 1.2218,
      "step": 9540
    },
    {
      "epoch": 12.005817245817246,
      "grad_norm": 0.014597934670746326,
      "learning_rate": 8.37980837980838e-06,
      "loss": 0.8302,
      "step": 9550
    },
    {
      "epoch": 12.006074646074646,
      "grad_norm": 40.39971160888672,
      "learning_rate": 8.376948376948377e-06,
      "loss": 1.5042,
      "step": 9560
    },
    {
      "epoch": 12.006332046332046,
      "grad_norm": 98.84669494628906,
      "learning_rate": 8.374088374088375e-06,
      "loss": 0.9766,
      "step": 9570
    },
    {
      "epoch": 12.006589446589446,
      "grad_norm": 0.027279607951641083,
      "learning_rate": 8.371228371228371e-06,
      "loss": 0.7949,
      "step": 9580
    },
    {
      "epoch": 12.006846846846846,
      "grad_norm": 0.5903051495552063,
      "learning_rate": 8.36836836836837e-06,
      "loss": 1.9912,
      "step": 9590
    },
    {
      "epoch": 12.007104247104246,
      "grad_norm": 0.3788576126098633,
      "learning_rate": 8.365508365508366e-06,
      "loss": 0.0129,
      "step": 9600
    },
    {
      "epoch": 12.007361647361648,
      "grad_norm": 0.021588651463389397,
      "learning_rate": 8.362648362648364e-06,
      "loss": 2.1048,
      "step": 9610
    },
    {
      "epoch": 12.007619047619048,
      "grad_norm": 0.36952582001686096,
      "learning_rate": 8.35978835978836e-06,
      "loss": 1.2766,
      "step": 9620
    },
    {
      "epoch": 12.007876447876448,
      "grad_norm": 0.09042210876941681,
      "learning_rate": 8.356928356928357e-06,
      "loss": 0.7546,
      "step": 9630
    },
    {
      "epoch": 12.008133848133848,
      "grad_norm": 2171.077880859375,
      "learning_rate": 8.354068354068355e-06,
      "loss": 1.81,
      "step": 9640
    },
    {
      "epoch": 12.008391248391248,
      "grad_norm": 0.14514465630054474,
      "learning_rate": 8.351208351208352e-06,
      "loss": 1.5139,
      "step": 9650
    },
    {
      "epoch": 12.008648648648649,
      "grad_norm": 0.049092844128608704,
      "learning_rate": 8.348348348348348e-06,
      "loss": 1.7561,
      "step": 9660
    },
    {
      "epoch": 12.008906048906049,
      "grad_norm": 30.588136672973633,
      "learning_rate": 8.345488345488346e-06,
      "loss": 0.4941,
      "step": 9670
    },
    {
      "epoch": 12.009163449163449,
      "grad_norm": 0.8432363271713257,
      "learning_rate": 8.342628342628343e-06,
      "loss": 0.7714,
      "step": 9680
    },
    {
      "epoch": 12.009420849420849,
      "grad_norm": 1.5549473762512207,
      "learning_rate": 8.33976833976834e-06,
      "loss": 1.1233,
      "step": 9690
    },
    {
      "epoch": 12.009678249678249,
      "grad_norm": 1.1150774955749512,
      "learning_rate": 8.336908336908337e-06,
      "loss": 0.7609,
      "step": 9700
    },
    {
      "epoch": 12.00993564993565,
      "grad_norm": 0.13375124335289001,
      "learning_rate": 8.334048334048335e-06,
      "loss": 0.3168,
      "step": 9710
    },
    {
      "epoch": 12.01019305019305,
      "grad_norm": 0.010085425339639187,
      "learning_rate": 8.331188331188332e-06,
      "loss": 0.123,
      "step": 9720
    },
    {
      "epoch": 12.01045045045045,
      "grad_norm": 0.4398421347141266,
      "learning_rate": 8.328328328328328e-06,
      "loss": 0.0059,
      "step": 9730
    },
    {
      "epoch": 12.01070785070785,
      "grad_norm": 0.03640660271048546,
      "learning_rate": 8.325468325468327e-06,
      "loss": 1.4955,
      "step": 9740
    },
    {
      "epoch": 12.010965250965251,
      "grad_norm": 0.0063200341537594795,
      "learning_rate": 8.322608322608323e-06,
      "loss": 1.0175,
      "step": 9750
    },
    {
      "epoch": 12.011222651222651,
      "grad_norm": 0.2818882167339325,
      "learning_rate": 8.319748319748321e-06,
      "loss": 0.6537,
      "step": 9760
    },
    {
      "epoch": 12.011480051480051,
      "grad_norm": 30.326845169067383,
      "learning_rate": 8.316888316888318e-06,
      "loss": 0.8632,
      "step": 9770
    },
    {
      "epoch": 12.011737451737451,
      "grad_norm": 0.115543894469738,
      "learning_rate": 8.314028314028316e-06,
      "loss": 0.3636,
      "step": 9780
    },
    {
      "epoch": 12.011994851994851,
      "grad_norm": 1787.333740234375,
      "learning_rate": 8.311168311168312e-06,
      "loss": 1.3063,
      "step": 9790
    },
    {
      "epoch": 12.012252252252253,
      "grad_norm": 0.19999530911445618,
      "learning_rate": 8.308308308308309e-06,
      "loss": 0.8213,
      "step": 9800
    },
    {
      "epoch": 12.012509652509653,
      "grad_norm": 4.1851654052734375,
      "learning_rate": 8.305448305448305e-06,
      "loss": 0.0428,
      "step": 9810
    },
    {
      "epoch": 12.012767052767053,
      "grad_norm": 0.09847444295883179,
      "learning_rate": 8.302588302588303e-06,
      "loss": 0.4367,
      "step": 9820
    },
    {
      "epoch": 12.013024453024453,
      "grad_norm": 0.3852860629558563,
      "learning_rate": 8.2997282997283e-06,
      "loss": 1.4293,
      "step": 9830
    },
    {
      "epoch": 12.013281853281853,
      "grad_norm": 0.13400162756443024,
      "learning_rate": 8.296868296868298e-06,
      "loss": 0.0083,
      "step": 9840
    },
    {
      "epoch": 12.013539253539253,
      "grad_norm": 0.18309134244918823,
      "learning_rate": 8.294008294008295e-06,
      "loss": 0.114,
      "step": 9850
    },
    {
      "epoch": 12.013796653796653,
      "grad_norm": 0.12962332367897034,
      "learning_rate": 8.291148291148293e-06,
      "loss": 1.1698,
      "step": 9860
    },
    {
      "epoch": 12.014054054054053,
      "grad_norm": 0.711496114730835,
      "learning_rate": 8.288288288288289e-06,
      "loss": 1.2569,
      "step": 9870
    },
    {
      "epoch": 12.014311454311454,
      "grad_norm": 33.74251937866211,
      "learning_rate": 8.285428285428286e-06,
      "loss": 2.4428,
      "step": 9880
    },
    {
      "epoch": 12.014568854568855,
      "grad_norm": 0.04531510919332504,
      "learning_rate": 8.282568282568284e-06,
      "loss": 1.7002,
      "step": 9890
    },
    {
      "epoch": 12.014826254826255,
      "grad_norm": 55.16579818725586,
      "learning_rate": 8.27970827970828e-06,
      "loss": 0.3482,
      "step": 9900
    },
    {
      "epoch": 12.015083655083656,
      "grad_norm": 0.34218963980674744,
      "learning_rate": 8.276848276848277e-06,
      "loss": 0.7469,
      "step": 9910
    },
    {
      "epoch": 12.015341055341056,
      "grad_norm": 39.31129837036133,
      "learning_rate": 8.273988273988275e-06,
      "loss": 1.3403,
      "step": 9920
    },
    {
      "epoch": 12.015598455598456,
      "grad_norm": 0.025866910815238953,
      "learning_rate": 8.271128271128271e-06,
      "loss": 0.6076,
      "step": 9930
    },
    {
      "epoch": 12.015855855855856,
      "grad_norm": 58.11125946044922,
      "learning_rate": 8.26826826826827e-06,
      "loss": 1.2356,
      "step": 9940
    },
    {
      "epoch": 12.016113256113256,
      "grad_norm": 0.6136714816093445,
      "learning_rate": 8.265408265408266e-06,
      "loss": 1.3545,
      "step": 9950
    },
    {
      "epoch": 12.016370656370656,
      "grad_norm": 0.12369683384895325,
      "learning_rate": 8.262548262548264e-06,
      "loss": 0.7099,
      "step": 9960
    },
    {
      "epoch": 12.016628056628056,
      "grad_norm": 0.0038368524983525276,
      "learning_rate": 8.25968825968826e-06,
      "loss": 0.5042,
      "step": 9970
    },
    {
      "epoch": 12.016885456885458,
      "grad_norm": 33.09714889526367,
      "learning_rate": 8.256828256828257e-06,
      "loss": 1.0448,
      "step": 9980
    },
    {
      "epoch": 12.017142857142858,
      "grad_norm": 0.4139602780342102,
      "learning_rate": 8.253968253968254e-06,
      "loss": 0.015,
      "step": 9990
    },
    {
      "epoch": 12.017400257400258,
      "grad_norm": 0.0036292793229222298,
      "learning_rate": 8.251108251108252e-06,
      "loss": 0.0043,
      "step": 10000
    },
    {
      "epoch": 12.017657657657658,
      "grad_norm": 0.04275906831026077,
      "learning_rate": 8.248248248248248e-06,
      "loss": 1.3682,
      "step": 10010
    },
    {
      "epoch": 12.017915057915058,
      "grad_norm": 0.4722258150577545,
      "learning_rate": 8.245388245388246e-06,
      "loss": 1.6312,
      "step": 10020
    },
    {
      "epoch": 12.018172458172458,
      "grad_norm": 24.255712509155273,
      "learning_rate": 8.242528242528243e-06,
      "loss": 1.44,
      "step": 10030
    },
    {
      "epoch": 12.018429858429858,
      "grad_norm": 1.5887428522109985,
      "learning_rate": 8.239668239668241e-06,
      "loss": 0.3472,
      "step": 10040
    },
    {
      "epoch": 12.018687258687258,
      "grad_norm": 0.007095601875334978,
      "learning_rate": 8.236808236808237e-06,
      "loss": 1.1121,
      "step": 10050
    },
    {
      "epoch": 12.018944658944658,
      "grad_norm": 0.10487360507249832,
      "learning_rate": 8.233948233948234e-06,
      "loss": 0.0206,
      "step": 10060
    },
    {
      "epoch": 12.01920205920206,
      "grad_norm": 528.3002319335938,
      "learning_rate": 8.231088231088232e-06,
      "loss": 0.8335,
      "step": 10070
    },
    {
      "epoch": 12.01945945945946,
      "grad_norm": 0.003817453049123287,
      "learning_rate": 8.228228228228229e-06,
      "loss": 1.8427,
      "step": 10080
    },
    {
      "epoch": 12.01971685971686,
      "grad_norm": 41.397342681884766,
      "learning_rate": 8.225368225368227e-06,
      "loss": 2.466,
      "step": 10090
    },
    {
      "epoch": 12.01997425997426,
      "grad_norm": 1.2678889036178589,
      "learning_rate": 8.222508222508223e-06,
      "loss": 0.8703,
      "step": 10100
    },
    {
      "epoch": 12.02,
      "eval_accuracy": 0.782608695652174,
      "eval_loss": 0.9878782033920288,
      "eval_runtime": 13.4438,
      "eval_samples_per_second": 3.422,
      "eval_steps_per_second": 3.422,
      "step": 10101
    },
    {
      "epoch": 13.00023166023166,
      "grad_norm": 34.533180236816406,
      "learning_rate": 8.219648219648221e-06,
      "loss": 1.276,
      "step": 10110
    },
    {
      "epoch": 13.00048906048906,
      "grad_norm": 0.05692519620060921,
      "learning_rate": 8.216788216788218e-06,
      "loss": 0.3495,
      "step": 10120
    },
    {
      "epoch": 13.000746460746461,
      "grad_norm": 39.88087844848633,
      "learning_rate": 8.213928213928214e-06,
      "loss": 0.8957,
      "step": 10130
    },
    {
      "epoch": 13.001003861003861,
      "grad_norm": 0.8153846859931946,
      "learning_rate": 8.211068211068212e-06,
      "loss": 0.4844,
      "step": 10140
    },
    {
      "epoch": 13.001261261261261,
      "grad_norm": 0.43959328532218933,
      "learning_rate": 8.208208208208209e-06,
      "loss": 0.3933,
      "step": 10150
    },
    {
      "epoch": 13.001518661518661,
      "grad_norm": 0.044305652379989624,
      "learning_rate": 8.205348205348205e-06,
      "loss": 0.8691,
      "step": 10160
    },
    {
      "epoch": 13.001776061776061,
      "grad_norm": 0.014474362134933472,
      "learning_rate": 8.202488202488204e-06,
      "loss": 1.1859,
      "step": 10170
    },
    {
      "epoch": 13.002033462033461,
      "grad_norm": 0.018979106098413467,
      "learning_rate": 8.1996281996282e-06,
      "loss": 0.0387,
      "step": 10180
    },
    {
      "epoch": 13.002290862290863,
      "grad_norm": 0.03073548711836338,
      "learning_rate": 8.196768196768198e-06,
      "loss": 0.9004,
      "step": 10190
    },
    {
      "epoch": 13.002548262548263,
      "grad_norm": 0.06650511175394058,
      "learning_rate": 8.193908193908195e-06,
      "loss": 1.8394,
      "step": 10200
    },
    {
      "epoch": 13.002805662805663,
      "grad_norm": 0.024794111028313637,
      "learning_rate": 8.191048191048193e-06,
      "loss": 0.8927,
      "step": 10210
    },
    {
      "epoch": 13.003063063063063,
      "grad_norm": 0.039084263145923615,
      "learning_rate": 8.18818818818819e-06,
      "loss": 0.4123,
      "step": 10220
    },
    {
      "epoch": 13.003320463320463,
      "grad_norm": 0.06459786742925644,
      "learning_rate": 8.185328185328186e-06,
      "loss": 0.3527,
      "step": 10230
    },
    {
      "epoch": 13.003577863577863,
      "grad_norm": 44.55921173095703,
      "learning_rate": 8.182468182468182e-06,
      "loss": 0.0077,
      "step": 10240
    },
    {
      "epoch": 13.003835263835263,
      "grad_norm": 0.22598184645175934,
      "learning_rate": 8.17960817960818e-06,
      "loss": 1.4898,
      "step": 10250
    },
    {
      "epoch": 13.004092664092664,
      "grad_norm": 204.12083435058594,
      "learning_rate": 8.176748176748177e-06,
      "loss": 0.0211,
      "step": 10260
    },
    {
      "epoch": 13.004350064350064,
      "grad_norm": 213.62789916992188,
      "learning_rate": 8.173888173888175e-06,
      "loss": 0.7785,
      "step": 10270
    },
    {
      "epoch": 13.004607464607465,
      "grad_norm": 0.017117006704211235,
      "learning_rate": 8.171028171028171e-06,
      "loss": 0.2757,
      "step": 10280
    },
    {
      "epoch": 13.004864864864865,
      "grad_norm": 0.24598395824432373,
      "learning_rate": 8.16816816816817e-06,
      "loss": 1.5201,
      "step": 10290
    },
    {
      "epoch": 13.005122265122266,
      "grad_norm": 0.04848215728998184,
      "learning_rate": 8.165308165308166e-06,
      "loss": 0.8018,
      "step": 10300
    },
    {
      "epoch": 13.005379665379666,
      "grad_norm": 0.005087207071483135,
      "learning_rate": 8.162448162448163e-06,
      "loss": 1.3702,
      "step": 10310
    },
    {
      "epoch": 13.005637065637066,
      "grad_norm": 0.22931011021137238,
      "learning_rate": 8.159588159588159e-06,
      "loss": 1.4593,
      "step": 10320
    },
    {
      "epoch": 13.005894465894466,
      "grad_norm": 0.7556596994400024,
      "learning_rate": 8.156728156728157e-06,
      "loss": 0.6623,
      "step": 10330
    },
    {
      "epoch": 13.006151866151866,
      "grad_norm": 178.25360107421875,
      "learning_rate": 8.153868153868154e-06,
      "loss": 0.9403,
      "step": 10340
    },
    {
      "epoch": 13.006409266409266,
      "grad_norm": 0.3808182179927826,
      "learning_rate": 8.151008151008152e-06,
      "loss": 0.9131,
      "step": 10350
    },
    {
      "epoch": 13.006666666666666,
      "grad_norm": 0.014495433308184147,
      "learning_rate": 8.148148148148148e-06,
      "loss": 0.4056,
      "step": 10360
    },
    {
      "epoch": 13.006924066924068,
      "grad_norm": 27.801916122436523,
      "learning_rate": 8.145288145288146e-06,
      "loss": 1.3077,
      "step": 10370
    },
    {
      "epoch": 13.007181467181468,
      "grad_norm": 55.3273811340332,
      "learning_rate": 8.142428142428143e-06,
      "loss": 0.6712,
      "step": 10380
    },
    {
      "epoch": 13.007438867438868,
      "grad_norm": 0.006372170057147741,
      "learning_rate": 8.139568139568141e-06,
      "loss": 1.3903,
      "step": 10390
    },
    {
      "epoch": 13.007696267696268,
      "grad_norm": 24.87656593322754,
      "learning_rate": 8.136708136708138e-06,
      "loss": 2.7086,
      "step": 10400
    },
    {
      "epoch": 13.007953667953668,
      "grad_norm": 0.33655011653900146,
      "learning_rate": 8.133848133848134e-06,
      "loss": 0.9958,
      "step": 10410
    },
    {
      "epoch": 13.008211068211068,
      "grad_norm": 0.038330886512994766,
      "learning_rate": 8.13098813098813e-06,
      "loss": 1.4135,
      "step": 10420
    },
    {
      "epoch": 13.008468468468468,
      "grad_norm": 0.3756619393825531,
      "learning_rate": 8.128128128128129e-06,
      "loss": 0.7018,
      "step": 10430
    },
    {
      "epoch": 13.008725868725868,
      "grad_norm": 0.04136725515127182,
      "learning_rate": 8.125268125268127e-06,
      "loss": 0.7366,
      "step": 10440
    },
    {
      "epoch": 13.008983268983268,
      "grad_norm": 27.85120391845703,
      "learning_rate": 8.122408122408123e-06,
      "loss": 0.8004,
      "step": 10450
    },
    {
      "epoch": 13.009240669240668,
      "grad_norm": 0.12497317045927048,
      "learning_rate": 8.119548119548121e-06,
      "loss": 0.789,
      "step": 10460
    },
    {
      "epoch": 13.00949806949807,
      "grad_norm": 0.3569711148738861,
      "learning_rate": 8.116688116688118e-06,
      "loss": 0.0059,
      "step": 10470
    },
    {
      "epoch": 13.00975546975547,
      "grad_norm": 0.012415600009262562,
      "learning_rate": 8.113828113828114e-06,
      "loss": 1.2201,
      "step": 10480
    },
    {
      "epoch": 13.01001287001287,
      "grad_norm": 0.012238414026796818,
      "learning_rate": 8.110968110968111e-06,
      "loss": 0.0005,
      "step": 10490
    },
    {
      "epoch": 13.01027027027027,
      "grad_norm": 65.99483489990234,
      "learning_rate": 8.108108108108109e-06,
      "loss": 1.8716,
      "step": 10500
    },
    {
      "epoch": 13.01052767052767,
      "grad_norm": 0.11564770340919495,
      "learning_rate": 8.105248105248106e-06,
      "loss": 1.0245,
      "step": 10510
    },
    {
      "epoch": 13.01078507078507,
      "grad_norm": 35.921592712402344,
      "learning_rate": 8.102388102388104e-06,
      "loss": 1.9181,
      "step": 10520
    },
    {
      "epoch": 13.01104247104247,
      "grad_norm": 84.63827514648438,
      "learning_rate": 8.0995280995281e-06,
      "loss": 0.584,
      "step": 10530
    },
    {
      "epoch": 13.01129987129987,
      "grad_norm": 177.37327575683594,
      "learning_rate": 8.096668096668098e-06,
      "loss": 0.7746,
      "step": 10540
    },
    {
      "epoch": 13.01155727155727,
      "grad_norm": 31.49327850341797,
      "learning_rate": 8.093808093808095e-06,
      "loss": 2.6965,
      "step": 10550
    },
    {
      "epoch": 13.011814671814673,
      "grad_norm": 0.9470750689506531,
      "learning_rate": 8.090948090948091e-06,
      "loss": 0.212,
      "step": 10560
    },
    {
      "epoch": 13.012072072072073,
      "grad_norm": 0.012930578552186489,
      "learning_rate": 8.088088088088088e-06,
      "loss": 0.7271,
      "step": 10570
    },
    {
      "epoch": 13.012329472329473,
      "grad_norm": 0.01820276491343975,
      "learning_rate": 8.085228085228086e-06,
      "loss": 0.7162,
      "step": 10580
    },
    {
      "epoch": 13.012586872586873,
      "grad_norm": 21.15532684326172,
      "learning_rate": 8.082368082368082e-06,
      "loss": 2.9765,
      "step": 10590
    },
    {
      "epoch": 13.012844272844273,
      "grad_norm": 0.6806768178939819,
      "learning_rate": 8.07950807950808e-06,
      "loss": 0.7837,
      "step": 10600
    },
    {
      "epoch": 13.013101673101673,
      "grad_norm": 15.736740112304688,
      "learning_rate": 8.076648076648077e-06,
      "loss": 1.3596,
      "step": 10610
    },
    {
      "epoch": 13.013359073359073,
      "grad_norm": 0.6283915042877197,
      "learning_rate": 8.073788073788075e-06,
      "loss": 0.6883,
      "step": 10620
    },
    {
      "epoch": 13.013616473616473,
      "grad_norm": 0.961954653263092,
      "learning_rate": 8.070928070928072e-06,
      "loss": 2.2455,
      "step": 10630
    },
    {
      "epoch": 13.013873873873873,
      "grad_norm": 0.627880334854126,
      "learning_rate": 8.06806806806807e-06,
      "loss": 0.7301,
      "step": 10640
    },
    {
      "epoch": 13.014131274131275,
      "grad_norm": 0.23738040030002594,
      "learning_rate": 8.065208065208066e-06,
      "loss": 1.0691,
      "step": 10650
    },
    {
      "epoch": 13.014388674388675,
      "grad_norm": 0.4037235677242279,
      "learning_rate": 8.062348062348063e-06,
      "loss": 0.4116,
      "step": 10660
    },
    {
      "epoch": 13.014646074646075,
      "grad_norm": 72.83759307861328,
      "learning_rate": 8.05948805948806e-06,
      "loss": 1.5558,
      "step": 10670
    },
    {
      "epoch": 13.014903474903475,
      "grad_norm": 0.330087274312973,
      "learning_rate": 8.056628056628057e-06,
      "loss": 1.1148,
      "step": 10680
    },
    {
      "epoch": 13.015160875160875,
      "grad_norm": 0.4726085364818573,
      "learning_rate": 8.053768053768054e-06,
      "loss": 1.1219,
      "step": 10690
    },
    {
      "epoch": 13.015418275418275,
      "grad_norm": 52.33597946166992,
      "learning_rate": 8.050908050908052e-06,
      "loss": 1.2237,
      "step": 10700
    },
    {
      "epoch": 13.015675675675675,
      "grad_norm": 0.02517028898000717,
      "learning_rate": 8.048048048048048e-06,
      "loss": 0.889,
      "step": 10710
    },
    {
      "epoch": 13.015933075933075,
      "grad_norm": 0.34835392236709595,
      "learning_rate": 8.045188045188047e-06,
      "loss": 0.408,
      "step": 10720
    },
    {
      "epoch": 13.016190476190475,
      "grad_norm": 87.9798583984375,
      "learning_rate": 8.042328042328043e-06,
      "loss": 1.3656,
      "step": 10730
    },
    {
      "epoch": 13.016447876447877,
      "grad_norm": 0.05181077867746353,
      "learning_rate": 8.03946803946804e-06,
      "loss": 0.9934,
      "step": 10740
    },
    {
      "epoch": 13.016705276705277,
      "grad_norm": 0.011249941773712635,
      "learning_rate": 8.036608036608036e-06,
      "loss": 0.7905,
      "step": 10750
    },
    {
      "epoch": 13.016962676962677,
      "grad_norm": 0.6939520239830017,
      "learning_rate": 8.033748033748034e-06,
      "loss": 0.8866,
      "step": 10760
    },
    {
      "epoch": 13.017220077220077,
      "grad_norm": 0.22471930086612701,
      "learning_rate": 8.03088803088803e-06,
      "loss": 0.0061,
      "step": 10770
    },
    {
      "epoch": 13.017477477477478,
      "grad_norm": 2.8579299449920654,
      "learning_rate": 8.028028028028029e-06,
      "loss": 0.6723,
      "step": 10780
    },
    {
      "epoch": 13.017734877734878,
      "grad_norm": 0.15794941782951355,
      "learning_rate": 8.025168025168025e-06,
      "loss": 1.1582,
      "step": 10790
    },
    {
      "epoch": 13.017992277992278,
      "grad_norm": 0.09103313833475113,
      "learning_rate": 8.022308022308023e-06,
      "loss": 0.4761,
      "step": 10800
    },
    {
      "epoch": 13.018249678249678,
      "grad_norm": 2.296825408935547,
      "learning_rate": 8.01944801944802e-06,
      "loss": 0.8791,
      "step": 10810
    },
    {
      "epoch": 13.018507078507078,
      "grad_norm": 0.6113599538803101,
      "learning_rate": 8.016588016588016e-06,
      "loss": 1.0967,
      "step": 10820
    },
    {
      "epoch": 13.01876447876448,
      "grad_norm": 0.026831205934286118,
      "learning_rate": 8.013728013728015e-06,
      "loss": 1.3861,
      "step": 10830
    },
    {
      "epoch": 13.01902187902188,
      "grad_norm": 0.035754233598709106,
      "learning_rate": 8.010868010868011e-06,
      "loss": 0.9776,
      "step": 10840
    },
    {
      "epoch": 13.01927927927928,
      "grad_norm": 0.055288445204496384,
      "learning_rate": 8.00800800800801e-06,
      "loss": 0.8417,
      "step": 10850
    },
    {
      "epoch": 13.01953667953668,
      "grad_norm": 5.07427978515625,
      "learning_rate": 8.005148005148006e-06,
      "loss": 0.872,
      "step": 10860
    },
    {
      "epoch": 13.01979407979408,
      "grad_norm": 0.008023238740861416,
      "learning_rate": 8.002288002288004e-06,
      "loss": 1.0987,
      "step": 10870
    },
    {
      "epoch": 13.02,
      "eval_accuracy": 0.782608695652174,
      "eval_loss": 1.0684303045272827,
      "eval_runtime": 13.5415,
      "eval_samples_per_second": 3.397,
      "eval_steps_per_second": 3.397,
      "step": 10878
    },
    {
      "epoch": 14.00005148005148,
      "grad_norm": 0.0345095694065094,
      "learning_rate": 7.999427999428e-06,
      "loss": 1.0408,
      "step": 10880
    },
    {
      "epoch": 14.00030888030888,
      "grad_norm": 567.0715942382812,
      "learning_rate": 7.996567996567998e-06,
      "loss": 0.2114,
      "step": 10890
    },
    {
      "epoch": 14.00056628056628,
      "grad_norm": 0.04684921354055405,
      "learning_rate": 7.993707993707995e-06,
      "loss": 0.4128,
      "step": 10900
    },
    {
      "epoch": 14.00082368082368,
      "grad_norm": 0.021159229800105095,
      "learning_rate": 7.990847990847991e-06,
      "loss": 1.9883,
      "step": 10910
    },
    {
      "epoch": 14.00108108108108,
      "grad_norm": 0.7736424803733826,
      "learning_rate": 7.987987987987988e-06,
      "loss": 0.8038,
      "step": 10920
    },
    {
      "epoch": 14.00133848133848,
      "grad_norm": 25.88015365600586,
      "learning_rate": 7.985127985127986e-06,
      "loss": 1.3199,
      "step": 10930
    },
    {
      "epoch": 14.00159588159588,
      "grad_norm": 27.008317947387695,
      "learning_rate": 7.982267982267982e-06,
      "loss": 1.3514,
      "step": 10940
    },
    {
      "epoch": 14.001853281853283,
      "grad_norm": 0.025671614333987236,
      "learning_rate": 7.97940797940798e-06,
      "loss": 1.9131,
      "step": 10950
    },
    {
      "epoch": 14.002110682110683,
      "grad_norm": 0.06644091010093689,
      "learning_rate": 7.976547976547977e-06,
      "loss": 0.6798,
      "step": 10960
    },
    {
      "epoch": 14.002368082368083,
      "grad_norm": 0.0720016285777092,
      "learning_rate": 7.973687973687975e-06,
      "loss": 1.4921,
      "step": 10970
    },
    {
      "epoch": 14.002625482625483,
      "grad_norm": 0.058085646480321884,
      "learning_rate": 7.970827970827972e-06,
      "loss": 1.1381,
      "step": 10980
    },
    {
      "epoch": 14.002882882882883,
      "grad_norm": 28.230207443237305,
      "learning_rate": 7.967967967967968e-06,
      "loss": 0.5997,
      "step": 10990
    },
    {
      "epoch": 14.003140283140283,
      "grad_norm": 0.015309504233300686,
      "learning_rate": 7.965107965107965e-06,
      "loss": 0.0371,
      "step": 11000
    },
    {
      "epoch": 14.003397683397683,
      "grad_norm": 42.59065246582031,
      "learning_rate": 7.962247962247963e-06,
      "loss": 0.4222,
      "step": 11010
    },
    {
      "epoch": 14.003655083655083,
      "grad_norm": 0.2624385952949524,
      "learning_rate": 7.95938795938796e-06,
      "loss": 0.7029,
      "step": 11020
    },
    {
      "epoch": 14.003912483912483,
      "grad_norm": 0.036210574209690094,
      "learning_rate": 7.956527956527957e-06,
      "loss": 1.0643,
      "step": 11030
    },
    {
      "epoch": 14.004169884169885,
      "grad_norm": 0.25612005591392517,
      "learning_rate": 7.953667953667954e-06,
      "loss": 1.8172,
      "step": 11040
    },
    {
      "epoch": 14.004427284427285,
      "grad_norm": 0.010674877092242241,
      "learning_rate": 7.950807950807952e-06,
      "loss": 1.4193,
      "step": 11050
    },
    {
      "epoch": 14.004684684684685,
      "grad_norm": 0.021124469116330147,
      "learning_rate": 7.947947947947949e-06,
      "loss": 1.3414,
      "step": 11060
    },
    {
      "epoch": 14.004942084942085,
      "grad_norm": 3.517850637435913,
      "learning_rate": 7.945087945087945e-06,
      "loss": 0.9322,
      "step": 11070
    },
    {
      "epoch": 14.005199485199485,
      "grad_norm": 0.025426078587770462,
      "learning_rate": 7.942227942227943e-06,
      "loss": 0.0065,
      "step": 11080
    },
    {
      "epoch": 14.005456885456885,
      "grad_norm": 0.015744155272841454,
      "learning_rate": 7.93936793936794e-06,
      "loss": 0.0105,
      "step": 11090
    },
    {
      "epoch": 14.005714285714285,
      "grad_norm": 0.05212106183171272,
      "learning_rate": 7.936507936507936e-06,
      "loss": 0.3777,
      "step": 11100
    },
    {
      "epoch": 14.005971685971685,
      "grad_norm": 0.008490210399031639,
      "learning_rate": 7.933647933647934e-06,
      "loss": 1.4914,
      "step": 11110
    },
    {
      "epoch": 14.006229086229085,
      "grad_norm": 0.16355255246162415,
      "learning_rate": 7.93078793078793e-06,
      "loss": 0.9386,
      "step": 11120
    },
    {
      "epoch": 14.006486486486487,
      "grad_norm": 0.5630903244018555,
      "learning_rate": 7.927927927927929e-06,
      "loss": 0.9168,
      "step": 11130
    },
    {
      "epoch": 14.006743886743887,
      "grad_norm": 27.17743682861328,
      "learning_rate": 7.925067925067925e-06,
      "loss": 0.4545,
      "step": 11140
    },
    {
      "epoch": 14.007001287001287,
      "grad_norm": 0.007702608127146959,
      "learning_rate": 7.922207922207924e-06,
      "loss": 1.1431,
      "step": 11150
    },
    {
      "epoch": 14.007258687258688,
      "grad_norm": 1.3509392738342285,
      "learning_rate": 7.91934791934792e-06,
      "loss": 0.3952,
      "step": 11160
    },
    {
      "epoch": 14.007516087516088,
      "grad_norm": 1.47977614402771,
      "learning_rate": 7.916487916487917e-06,
      "loss": 1.6739,
      "step": 11170
    },
    {
      "epoch": 14.007773487773488,
      "grad_norm": 0.06250022351741791,
      "learning_rate": 7.913627913627915e-06,
      "loss": 1.11,
      "step": 11180
    },
    {
      "epoch": 14.008030888030888,
      "grad_norm": 42.10331726074219,
      "learning_rate": 7.910767910767911e-06,
      "loss": 1.5249,
      "step": 11190
    },
    {
      "epoch": 14.008288288288288,
      "grad_norm": 0.4262247383594513,
      "learning_rate": 7.90790790790791e-06,
      "loss": 0.5517,
      "step": 11200
    },
    {
      "epoch": 14.008545688545688,
      "grad_norm": 341.49090576171875,
      "learning_rate": 7.905047905047906e-06,
      "loss": 0.8725,
      "step": 11210
    },
    {
      "epoch": 14.00880308880309,
      "grad_norm": 111.06766510009766,
      "learning_rate": 7.902187902187904e-06,
      "loss": 0.5028,
      "step": 11220
    },
    {
      "epoch": 14.00906048906049,
      "grad_norm": 0.021268639713525772,
      "learning_rate": 7.8993278993279e-06,
      "loss": 0.675,
      "step": 11230
    },
    {
      "epoch": 14.00931788931789,
      "grad_norm": 0.00978196132928133,
      "learning_rate": 7.896467896467897e-06,
      "loss": 1.7491,
      "step": 11240
    },
    {
      "epoch": 14.00957528957529,
      "grad_norm": 0.13062025606632233,
      "learning_rate": 7.893607893607893e-06,
      "loss": 0.0045,
      "step": 11250
    },
    {
      "epoch": 14.00983268983269,
      "grad_norm": 0.40098121762275696,
      "learning_rate": 7.890747890747892e-06,
      "loss": 1.331,
      "step": 11260
    },
    {
      "epoch": 14.01009009009009,
      "grad_norm": 99.57262420654297,
      "learning_rate": 7.887887887887888e-06,
      "loss": 0.8479,
      "step": 11270
    },
    {
      "epoch": 14.01034749034749,
      "grad_norm": 0.15203261375427246,
      "learning_rate": 7.885027885027886e-06,
      "loss": 0.5903,
      "step": 11280
    },
    {
      "epoch": 14.01060489060489,
      "grad_norm": 0.16804452240467072,
      "learning_rate": 7.882167882167883e-06,
      "loss": 1.2317,
      "step": 11290
    },
    {
      "epoch": 14.01086229086229,
      "grad_norm": 0.05834130197763443,
      "learning_rate": 7.87930787930788e-06,
      "loss": 0.2802,
      "step": 11300
    },
    {
      "epoch": 14.01111969111969,
      "grad_norm": 0.24336637556552887,
      "learning_rate": 7.876447876447877e-06,
      "loss": 0.891,
      "step": 11310
    },
    {
      "epoch": 14.011377091377092,
      "grad_norm": 27.7568359375,
      "learning_rate": 7.873587873587874e-06,
      "loss": 1.3452,
      "step": 11320
    },
    {
      "epoch": 14.011634491634492,
      "grad_norm": 59.793251037597656,
      "learning_rate": 7.870727870727872e-06,
      "loss": 0.5921,
      "step": 11330
    },
    {
      "epoch": 14.011891891891892,
      "grad_norm": 3.0434534549713135,
      "learning_rate": 7.867867867867868e-06,
      "loss": 1.6081,
      "step": 11340
    },
    {
      "epoch": 14.012149292149292,
      "grad_norm": 0.012520610354840755,
      "learning_rate": 7.865007865007865e-06,
      "loss": 0.5631,
      "step": 11350
    },
    {
      "epoch": 14.012406692406692,
      "grad_norm": 0.03452794998884201,
      "learning_rate": 7.862147862147863e-06,
      "loss": 0.4335,
      "step": 11360
    },
    {
      "epoch": 14.012664092664092,
      "grad_norm": 0.013125480152666569,
      "learning_rate": 7.85928785928786e-06,
      "loss": 0.6082,
      "step": 11370
    },
    {
      "epoch": 14.012921492921492,
      "grad_norm": 0.07402829825878143,
      "learning_rate": 7.856427856427858e-06,
      "loss": 1.1006,
      "step": 11380
    },
    {
      "epoch": 14.013178893178893,
      "grad_norm": 5.510627269744873,
      "learning_rate": 7.853567853567854e-06,
      "loss": 0.2657,
      "step": 11390
    },
    {
      "epoch": 14.013436293436293,
      "grad_norm": 0.4086941182613373,
      "learning_rate": 7.850707850707852e-06,
      "loss": 0.4711,
      "step": 11400
    },
    {
      "epoch": 14.013693693693694,
      "grad_norm": 49.80360412597656,
      "learning_rate": 7.847847847847849e-06,
      "loss": 0.5284,
      "step": 11410
    },
    {
      "epoch": 14.013951093951095,
      "grad_norm": 137.60855102539062,
      "learning_rate": 7.844987844987845e-06,
      "loss": 1.0677,
      "step": 11420
    },
    {
      "epoch": 14.014208494208495,
      "grad_norm": 0.006826744880527258,
      "learning_rate": 7.842127842127842e-06,
      "loss": 2.5411,
      "step": 11430
    },
    {
      "epoch": 14.014465894465895,
      "grad_norm": 0.35207125544548035,
      "learning_rate": 7.83926783926784e-06,
      "loss": 1.3689,
      "step": 11440
    },
    {
      "epoch": 14.014723294723295,
      "grad_norm": 1.623186469078064,
      "learning_rate": 7.836407836407836e-06,
      "loss": 0.5639,
      "step": 11450
    },
    {
      "epoch": 14.014980694980695,
      "grad_norm": 0.17725925147533417,
      "learning_rate": 7.833547833547834e-06,
      "loss": 0.4686,
      "step": 11460
    },
    {
      "epoch": 14.015238095238095,
      "grad_norm": 0.007551025133579969,
      "learning_rate": 7.830687830687831e-06,
      "loss": 0.4924,
      "step": 11470
    },
    {
      "epoch": 14.015495495495495,
      "grad_norm": 46.443031311035156,
      "learning_rate": 7.827827827827829e-06,
      "loss": 1.5643,
      "step": 11480
    },
    {
      "epoch": 14.015752895752895,
      "grad_norm": 0.04052784666419029,
      "learning_rate": 7.824967824967826e-06,
      "loss": 0.854,
      "step": 11490
    },
    {
      "epoch": 14.016010296010297,
      "grad_norm": 33.286373138427734,
      "learning_rate": 7.822107822107822e-06,
      "loss": 0.8945,
      "step": 11500
    },
    {
      "epoch": 14.016267696267697,
      "grad_norm": 30.311275482177734,
      "learning_rate": 7.81924781924782e-06,
      "loss": 0.7156,
      "step": 11510
    },
    {
      "epoch": 14.016525096525097,
      "grad_norm": 0.5423368215560913,
      "learning_rate": 7.816387816387817e-06,
      "loss": 0.5232,
      "step": 11520
    },
    {
      "epoch": 14.016782496782497,
      "grad_norm": 55.021217346191406,
      "learning_rate": 7.813527813527813e-06,
      "loss": 0.9088,
      "step": 11530
    },
    {
      "epoch": 14.017039897039897,
      "grad_norm": 0.35494714975357056,
      "learning_rate": 7.810667810667811e-06,
      "loss": 0.551,
      "step": 11540
    },
    {
      "epoch": 14.017297297297297,
      "grad_norm": 0.24058228731155396,
      "learning_rate": 7.807807807807808e-06,
      "loss": 0.6426,
      "step": 11550
    },
    {
      "epoch": 14.017554697554697,
      "grad_norm": 1.1704317331314087,
      "learning_rate": 7.804947804947806e-06,
      "loss": 1.029,
      "step": 11560
    },
    {
      "epoch": 14.017812097812097,
      "grad_norm": 0.269205242395401,
      "learning_rate": 7.802087802087804e-06,
      "loss": 1.8044,
      "step": 11570
    },
    {
      "epoch": 14.018069498069497,
      "grad_norm": 0.7438356280326843,
      "learning_rate": 7.7992277992278e-06,
      "loss": 0.851,
      "step": 11580
    },
    {
      "epoch": 14.0183268983269,
      "grad_norm": 0.6309325695037842,
      "learning_rate": 7.796367796367797e-06,
      "loss": 1.1659,
      "step": 11590
    },
    {
      "epoch": 14.0185842985843,
      "grad_norm": 0.8117626905441284,
      "learning_rate": 7.793507793507794e-06,
      "loss": 0.8181,
      "step": 11600
    },
    {
      "epoch": 14.0188416988417,
      "grad_norm": 0.028226610273122787,
      "learning_rate": 7.790647790647792e-06,
      "loss": 1.0875,
      "step": 11610
    },
    {
      "epoch": 14.0190990990991,
      "grad_norm": 1.9404858350753784,
      "learning_rate": 7.787787787787788e-06,
      "loss": 1.3949,
      "step": 11620
    },
    {
      "epoch": 14.0193564993565,
      "grad_norm": 7.443615913391113,
      "learning_rate": 7.784927784927786e-06,
      "loss": 0.7276,
      "step": 11630
    },
    {
      "epoch": 14.0196138996139,
      "grad_norm": 0.020177453756332397,
      "learning_rate": 7.782067782067783e-06,
      "loss": 0.9539,
      "step": 11640
    },
    {
      "epoch": 14.0198712998713,
      "grad_norm": 0.0387568436563015,
      "learning_rate": 7.779207779207781e-06,
      "loss": 0.0074,
      "step": 11650
    },
    {
      "epoch": 14.02,
      "eval_accuracy": 0.7391304347826086,
      "eval_loss": 1.2900571823120117,
      "eval_runtime": 13.4677,
      "eval_samples_per_second": 3.416,
      "eval_steps_per_second": 3.416,
      "step": 11655
    },
    {
      "epoch": 15.0001287001287,
      "grad_norm": 0.21021750569343567,
      "learning_rate": 7.776347776347777e-06,
      "loss": 0.0063,
      "step": 11660
    },
    {
      "epoch": 15.0003861003861,
      "grad_norm": 0.18497781455516815,
      "learning_rate": 7.773487773487774e-06,
      "loss": 1.5418,
      "step": 11670
    },
    {
      "epoch": 15.0006435006435,
      "grad_norm": 0.07883092761039734,
      "learning_rate": 7.77062777062777e-06,
      "loss": 1.1415,
      "step": 11680
    },
    {
      "epoch": 15.0009009009009,
      "grad_norm": 66.12890625,
      "learning_rate": 7.767767767767769e-06,
      "loss": 0.6305,
      "step": 11690
    },
    {
      "epoch": 15.0011583011583,
      "grad_norm": 78.24878692626953,
      "learning_rate": 7.764907764907765e-06,
      "loss": 0.3121,
      "step": 11700
    },
    {
      "epoch": 15.001415701415702,
      "grad_norm": 0.011892065405845642,
      "learning_rate": 7.762047762047763e-06,
      "loss": 0.0114,
      "step": 11710
    },
    {
      "epoch": 15.001673101673102,
      "grad_norm": 0.012777644209563732,
      "learning_rate": 7.75918775918776e-06,
      "loss": 1.8192,
      "step": 11720
    },
    {
      "epoch": 15.001930501930502,
      "grad_norm": 0.17456497251987457,
      "learning_rate": 7.756327756327758e-06,
      "loss": 1.0947,
      "step": 11730
    },
    {
      "epoch": 15.002187902187902,
      "grad_norm": 95.4760971069336,
      "learning_rate": 7.753467753467754e-06,
      "loss": 0.5226,
      "step": 11740
    },
    {
      "epoch": 15.002445302445302,
      "grad_norm": 334.6659851074219,
      "learning_rate": 7.75060775060775e-06,
      "loss": 1.6629,
      "step": 11750
    },
    {
      "epoch": 15.002702702702702,
      "grad_norm": 0.3331473171710968,
      "learning_rate": 7.747747747747749e-06,
      "loss": 0.5419,
      "step": 11760
    },
    {
      "epoch": 15.002960102960103,
      "grad_norm": 0.0954216867685318,
      "learning_rate": 7.744887744887745e-06,
      "loss": 0.0035,
      "step": 11770
    },
    {
      "epoch": 15.003217503217503,
      "grad_norm": 28.08690643310547,
      "learning_rate": 7.742027742027742e-06,
      "loss": 0.4808,
      "step": 11780
    },
    {
      "epoch": 15.003474903474903,
      "grad_norm": 10.32856559753418,
      "learning_rate": 7.73916773916774e-06,
      "loss": 0.0072,
      "step": 11790
    },
    {
      "epoch": 15.003732303732304,
      "grad_norm": 79.47557830810547,
      "learning_rate": 7.736307736307736e-06,
      "loss": 0.6229,
      "step": 11800
    },
    {
      "epoch": 15.003989703989705,
      "grad_norm": 0.05898858606815338,
      "learning_rate": 7.733447733447735e-06,
      "loss": 0.6668,
      "step": 11810
    },
    {
      "epoch": 15.004247104247105,
      "grad_norm": 0.010334447957575321,
      "learning_rate": 7.730587730587731e-06,
      "loss": 0.5522,
      "step": 11820
    },
    {
      "epoch": 15.004504504504505,
      "grad_norm": 0.014395112171769142,
      "learning_rate": 7.72772772772773e-06,
      "loss": 0.6832,
      "step": 11830
    },
    {
      "epoch": 15.004761904761905,
      "grad_norm": 0.40451979637145996,
      "learning_rate": 7.724867724867726e-06,
      "loss": 0.4752,
      "step": 11840
    },
    {
      "epoch": 15.005019305019305,
      "grad_norm": 0.09375791996717453,
      "learning_rate": 7.722007722007722e-06,
      "loss": 0.7253,
      "step": 11850
    },
    {
      "epoch": 15.005276705276705,
      "grad_norm": 0.011127919889986515,
      "learning_rate": 7.719147719147719e-06,
      "loss": 1.563,
      "step": 11860
    },
    {
      "epoch": 15.005534105534105,
      "grad_norm": 0.10237417370080948,
      "learning_rate": 7.716287716287717e-06,
      "loss": 0.4648,
      "step": 11870
    },
    {
      "epoch": 15.005791505791505,
      "grad_norm": 0.13473637402057648,
      "learning_rate": 7.713427713427713e-06,
      "loss": 0.006,
      "step": 11880
    },
    {
      "epoch": 15.006048906048907,
      "grad_norm": 0.09255123883485794,
      "learning_rate": 7.710567710567711e-06,
      "loss": 0.004,
      "step": 11890
    },
    {
      "epoch": 15.006306306306307,
      "grad_norm": 0.41176626086235046,
      "learning_rate": 7.707707707707708e-06,
      "loss": 1.3119,
      "step": 11900
    },
    {
      "epoch": 15.006563706563707,
      "grad_norm": 0.0077965897507965565,
      "learning_rate": 7.704847704847706e-06,
      "loss": 1.5889,
      "step": 11910
    },
    {
      "epoch": 15.006821106821107,
      "grad_norm": 0.24158236384391785,
      "learning_rate": 7.701987701987703e-06,
      "loss": 1.7263,
      "step": 11920
    },
    {
      "epoch": 15.007078507078507,
      "grad_norm": 5.002826690673828,
      "learning_rate": 7.699127699127699e-06,
      "loss": 0.8271,
      "step": 11930
    },
    {
      "epoch": 15.007335907335907,
      "grad_norm": 39.31734848022461,
      "learning_rate": 7.696267696267697e-06,
      "loss": 2.0124,
      "step": 11940
    },
    {
      "epoch": 15.007593307593307,
      "grad_norm": 0.05357268825173378,
      "learning_rate": 7.693407693407694e-06,
      "loss": 0.434,
      "step": 11950
    },
    {
      "epoch": 15.007850707850707,
      "grad_norm": 0.13800211250782013,
      "learning_rate": 7.690547690547692e-06,
      "loss": 0.5399,
      "step": 11960
    },
    {
      "epoch": 15.008108108108107,
      "grad_norm": 201.25430297851562,
      "learning_rate": 7.687687687687688e-06,
      "loss": 1.66,
      "step": 11970
    },
    {
      "epoch": 15.00836550836551,
      "grad_norm": 0.3996284008026123,
      "learning_rate": 7.684827684827686e-06,
      "loss": 0.645,
      "step": 11980
    },
    {
      "epoch": 15.00862290862291,
      "grad_norm": 95.54779052734375,
      "learning_rate": 7.681967681967683e-06,
      "loss": 1.0521,
      "step": 11990
    },
    {
      "epoch": 15.00888030888031,
      "grad_norm": 3.103736162185669,
      "learning_rate": 7.67910767910768e-06,
      "loss": 0.649,
      "step": 12000
    },
    {
      "epoch": 15.00913770913771,
      "grad_norm": 99.35127258300781,
      "learning_rate": 7.676247676247678e-06,
      "loss": 0.9866,
      "step": 12010
    },
    {
      "epoch": 15.00939510939511,
      "grad_norm": 0.011812890879809856,
      "learning_rate": 7.673387673387674e-06,
      "loss": 0.0066,
      "step": 12020
    },
    {
      "epoch": 15.00965250965251,
      "grad_norm": 0.24866342544555664,
      "learning_rate": 7.67052767052767e-06,
      "loss": 1.5628,
      "step": 12030
    },
    {
      "epoch": 15.00990990990991,
      "grad_norm": 0.011373716406524181,
      "learning_rate": 7.667667667667669e-06,
      "loss": 0.2936,
      "step": 12040
    },
    {
      "epoch": 15.01016731016731,
      "grad_norm": 0.16900913417339325,
      "learning_rate": 7.664807664807665e-06,
      "loss": 0.9559,
      "step": 12050
    },
    {
      "epoch": 15.01042471042471,
      "grad_norm": 0.31922098994255066,
      "learning_rate": 7.661947661947663e-06,
      "loss": 0.151,
      "step": 12060
    },
    {
      "epoch": 15.010682110682112,
      "grad_norm": 352.00823974609375,
      "learning_rate": 7.65908765908766e-06,
      "loss": 0.5267,
      "step": 12070
    },
    {
      "epoch": 15.010939510939512,
      "grad_norm": 457.71478271484375,
      "learning_rate": 7.656227656227658e-06,
      "loss": 0.5103,
      "step": 12080
    },
    {
      "epoch": 15.011196911196912,
      "grad_norm": 0.18350684642791748,
      "learning_rate": 7.653367653367654e-06,
      "loss": 1.0546,
      "step": 12090
    },
    {
      "epoch": 15.011454311454312,
      "grad_norm": 81.0344009399414,
      "learning_rate": 7.65050765050765e-06,
      "loss": 1.6585,
      "step": 12100
    },
    {
      "epoch": 15.011711711711712,
      "grad_norm": 34.371742248535156,
      "learning_rate": 7.647647647647647e-06,
      "loss": 0.6012,
      "step": 12110
    },
    {
      "epoch": 15.011969111969112,
      "grad_norm": 54.86530303955078,
      "learning_rate": 7.644787644787645e-06,
      "loss": 1.5436,
      "step": 12120
    },
    {
      "epoch": 15.012226512226512,
      "grad_norm": 0.05258643254637718,
      "learning_rate": 7.641927641927642e-06,
      "loss": 2.2123,
      "step": 12130
    },
    {
      "epoch": 15.012483912483912,
      "grad_norm": 486.4405212402344,
      "learning_rate": 7.63906763906764e-06,
      "loss": 0.941,
      "step": 12140
    },
    {
      "epoch": 15.012741312741312,
      "grad_norm": 33.052574157714844,
      "learning_rate": 7.636207636207637e-06,
      "loss": 0.4,
      "step": 12150
    },
    {
      "epoch": 15.012998712998712,
      "grad_norm": 0.40217721462249756,
      "learning_rate": 7.633347633347635e-06,
      "loss": 1.2872,
      "step": 12160
    },
    {
      "epoch": 15.013256113256114,
      "grad_norm": 0.026713982224464417,
      "learning_rate": 7.630487630487631e-06,
      "loss": 1.5662,
      "step": 12170
    },
    {
      "epoch": 15.013513513513514,
      "grad_norm": 25.906904220581055,
      "learning_rate": 7.6276276276276285e-06,
      "loss": 1.0581,
      "step": 12180
    },
    {
      "epoch": 15.013770913770914,
      "grad_norm": 0.21586193144321442,
      "learning_rate": 7.624767624767625e-06,
      "loss": 0.5713,
      "step": 12190
    },
    {
      "epoch": 15.014028314028314,
      "grad_norm": 53.68928909301758,
      "learning_rate": 7.621907621907622e-06,
      "loss": 0.3419,
      "step": 12200
    },
    {
      "epoch": 15.014285714285714,
      "grad_norm": 0.16979816555976868,
      "learning_rate": 7.61904761904762e-06,
      "loss": 0.0113,
      "step": 12210
    },
    {
      "epoch": 15.014543114543114,
      "grad_norm": 0.515741229057312,
      "learning_rate": 7.616187616187617e-06,
      "loss": 0.5549,
      "step": 12220
    },
    {
      "epoch": 15.014800514800514,
      "grad_norm": 110.49329376220703,
      "learning_rate": 7.613327613327613e-06,
      "loss": 0.9444,
      "step": 12230
    },
    {
      "epoch": 15.015057915057914,
      "grad_norm": 58.470523834228516,
      "learning_rate": 7.610467610467612e-06,
      "loss": 1.3908,
      "step": 12240
    },
    {
      "epoch": 15.015315315315314,
      "grad_norm": 65.19776916503906,
      "learning_rate": 7.607607607607608e-06,
      "loss": 1.0306,
      "step": 12250
    },
    {
      "epoch": 15.015572715572716,
      "grad_norm": 45.314945220947266,
      "learning_rate": 7.604747604747605e-06,
      "loss": 0.5072,
      "step": 12260
    },
    {
      "epoch": 15.015830115830116,
      "grad_norm": 0.008456357754766941,
      "learning_rate": 7.601887601887602e-06,
      "loss": 2.315,
      "step": 12270
    },
    {
      "epoch": 15.016087516087516,
      "grad_norm": 0.0190535020083189,
      "learning_rate": 7.5990275990276e-06,
      "loss": 1.7028,
      "step": 12280
    },
    {
      "epoch": 15.016344916344917,
      "grad_norm": 25.07335662841797,
      "learning_rate": 7.5961675961675965e-06,
      "loss": 2.0921,
      "step": 12290
    },
    {
      "epoch": 15.016602316602317,
      "grad_norm": 1.3651502132415771,
      "learning_rate": 7.593307593307594e-06,
      "loss": 0.3876,
      "step": 12300
    },
    {
      "epoch": 15.016859716859717,
      "grad_norm": 332.7787170410156,
      "learning_rate": 7.590447590447592e-06,
      "loss": 0.8509,
      "step": 12310
    },
    {
      "epoch": 15.017117117117117,
      "grad_norm": 0.7402872443199158,
      "learning_rate": 7.587587587587588e-06,
      "loss": 0.4219,
      "step": 12320
    },
    {
      "epoch": 15.017374517374517,
      "grad_norm": 1.146626353263855,
      "learning_rate": 7.584727584727586e-06,
      "loss": 0.6808,
      "step": 12330
    },
    {
      "epoch": 15.017631917631917,
      "grad_norm": 59.793434143066406,
      "learning_rate": 7.581867581867582e-06,
      "loss": 0.8229,
      "step": 12340
    },
    {
      "epoch": 15.017889317889319,
      "grad_norm": 0.011528403498232365,
      "learning_rate": 7.57900757900758e-06,
      "loss": 0.2795,
      "step": 12350
    },
    {
      "epoch": 15.018146718146719,
      "grad_norm": 74.38329315185547,
      "learning_rate": 7.576147576147577e-06,
      "loss": 0.5573,
      "step": 12360
    },
    {
      "epoch": 15.018404118404119,
      "grad_norm": 39.21977233886719,
      "learning_rate": 7.573287573287574e-06,
      "loss": 1.7136,
      "step": 12370
    },
    {
      "epoch": 15.018661518661519,
      "grad_norm": 87.86930847167969,
      "learning_rate": 7.570427570427571e-06,
      "loss": 1.151,
      "step": 12380
    },
    {
      "epoch": 15.018918918918919,
      "grad_norm": 0.3435974419116974,
      "learning_rate": 7.567567567567569e-06,
      "loss": 0.5449,
      "step": 12390
    },
    {
      "epoch": 15.019176319176319,
      "grad_norm": 0.020228708162903786,
      "learning_rate": 7.564707564707565e-06,
      "loss": 1.1051,
      "step": 12400
    },
    {
      "epoch": 15.019433719433719,
      "grad_norm": 48.28805923461914,
      "learning_rate": 7.5618475618475626e-06,
      "loss": 0.9372,
      "step": 12410
    },
    {
      "epoch": 15.01969111969112,
      "grad_norm": 0.9852522611618042,
      "learning_rate": 7.558987558987559e-06,
      "loss": 0.6727,
      "step": 12420
    },
    {
      "epoch": 15.01994851994852,
      "grad_norm": 0.5902912020683289,
      "learning_rate": 7.556127556127557e-06,
      "loss": 1.39,
      "step": 12430
    },
    {
      "epoch": 15.02,
      "eval_accuracy": 0.6304347826086957,
      "eval_loss": 1.6627668142318726,
      "eval_runtime": 13.4992,
      "eval_samples_per_second": 3.408,
      "eval_steps_per_second": 3.408,
      "step": 12432
    },
    {
      "epoch": 16.00020592020592,
      "grad_norm": 0.04411958158016205,
      "learning_rate": 7.553267553267554e-06,
      "loss": 1.2937,
      "step": 12440
    },
    {
      "epoch": 16.00046332046332,
      "grad_norm": 0.03254665061831474,
      "learning_rate": 7.550407550407551e-06,
      "loss": 0.6591,
      "step": 12450
    },
    {
      "epoch": 16.00072072072072,
      "grad_norm": 0.3264172375202179,
      "learning_rate": 7.547547547547548e-06,
      "loss": 0.3905,
      "step": 12460
    },
    {
      "epoch": 16.00097812097812,
      "grad_norm": 0.06069876626133919,
      "learning_rate": 7.544687544687546e-06,
      "loss": 1.8253,
      "step": 12470
    },
    {
      "epoch": 16.00123552123552,
      "grad_norm": 0.23206213116645813,
      "learning_rate": 7.541827541827542e-06,
      "loss": 0.1752,
      "step": 12480
    },
    {
      "epoch": 16.001492921492922,
      "grad_norm": 0.4681251049041748,
      "learning_rate": 7.53896753896754e-06,
      "loss": 0.5147,
      "step": 12490
    },
    {
      "epoch": 16.001750321750322,
      "grad_norm": 0.06292621046304703,
      "learning_rate": 7.536107536107537e-06,
      "loss": 1.005,
      "step": 12500
    },
    {
      "epoch": 16.002007722007722,
      "grad_norm": 30.542930603027344,
      "learning_rate": 7.533247533247534e-06,
      "loss": 0.8872,
      "step": 12510
    },
    {
      "epoch": 16.002265122265122,
      "grad_norm": 0.456661581993103,
      "learning_rate": 7.5303875303875305e-06,
      "loss": 0.517,
      "step": 12520
    },
    {
      "epoch": 16.002522522522522,
      "grad_norm": 0.6798164248466492,
      "learning_rate": 7.527527527527529e-06,
      "loss": 1.2443,
      "step": 12530
    },
    {
      "epoch": 16.002779922779922,
      "grad_norm": 3.348841905593872,
      "learning_rate": 7.524667524667525e-06,
      "loss": 0.4661,
      "step": 12540
    },
    {
      "epoch": 16.003037323037322,
      "grad_norm": 0.005561114754527807,
      "learning_rate": 7.5218075218075225e-06,
      "loss": 0.4401,
      "step": 12550
    },
    {
      "epoch": 16.003294723294722,
      "grad_norm": 0.14631526172161102,
      "learning_rate": 7.518947518947519e-06,
      "loss": 0.4638,
      "step": 12560
    },
    {
      "epoch": 16.003552123552122,
      "grad_norm": 0.02118610218167305,
      "learning_rate": 7.516087516087517e-06,
      "loss": 0.678,
      "step": 12570
    },
    {
      "epoch": 16.003809523809522,
      "grad_norm": 0.13729602098464966,
      "learning_rate": 7.5132275132275136e-06,
      "loss": 0.4831,
      "step": 12580
    },
    {
      "epoch": 16.004066924066922,
      "grad_norm": 0.3196340501308441,
      "learning_rate": 7.510367510367511e-06,
      "loss": 0.9561,
      "step": 12590
    },
    {
      "epoch": 16.004324324324326,
      "grad_norm": 0.006972214672714472,
      "learning_rate": 7.507507507507507e-06,
      "loss": 1.2584,
      "step": 12600
    },
    {
      "epoch": 16.004581724581726,
      "grad_norm": 26.036048889160156,
      "learning_rate": 7.5046475046475055e-06,
      "loss": 2.6442,
      "step": 12610
    },
    {
      "epoch": 16.004839124839126,
      "grad_norm": 58.6983757019043,
      "learning_rate": 7.501787501787502e-06,
      "loss": 2.0255,
      "step": 12620
    },
    {
      "epoch": 16.005096525096526,
      "grad_norm": 243.01780700683594,
      "learning_rate": 7.498927498927499e-06,
      "loss": 0.2647,
      "step": 12630
    },
    {
      "epoch": 16.005353925353926,
      "grad_norm": 1.2858657836914062,
      "learning_rate": 7.496067496067496e-06,
      "loss": 0.3808,
      "step": 12640
    },
    {
      "epoch": 16.005611325611326,
      "grad_norm": 1.131397008895874,
      "learning_rate": 7.493207493207494e-06,
      "loss": 0.4255,
      "step": 12650
    },
    {
      "epoch": 16.005868725868726,
      "grad_norm": 0.0808020830154419,
      "learning_rate": 7.49034749034749e-06,
      "loss": 1.2511,
      "step": 12660
    },
    {
      "epoch": 16.006126126126127,
      "grad_norm": 52.445899963378906,
      "learning_rate": 7.487487487487488e-06,
      "loss": 0.9333,
      "step": 12670
    },
    {
      "epoch": 16.006383526383527,
      "grad_norm": 0.38577455282211304,
      "learning_rate": 7.484627484627486e-06,
      "loss": 0.0374,
      "step": 12680
    },
    {
      "epoch": 16.006640926640927,
      "grad_norm": 58.25796890258789,
      "learning_rate": 7.481767481767482e-06,
      "loss": 1.0307,
      "step": 12690
    },
    {
      "epoch": 16.006898326898327,
      "grad_norm": 0.05789424851536751,
      "learning_rate": 7.47890747890748e-06,
      "loss": 0.0427,
      "step": 12700
    },
    {
      "epoch": 16.007155727155727,
      "grad_norm": 0.046346213668584824,
      "learning_rate": 7.476047476047477e-06,
      "loss": 0.4548,
      "step": 12710
    },
    {
      "epoch": 16.007413127413127,
      "grad_norm": 0.00563997495919466,
      "learning_rate": 7.473187473187474e-06,
      "loss": 0.0034,
      "step": 12720
    },
    {
      "epoch": 16.007670527670527,
      "grad_norm": 0.01808410882949829,
      "learning_rate": 7.470327470327471e-06,
      "loss": 1.1605,
      "step": 12730
    },
    {
      "epoch": 16.007927927927927,
      "grad_norm": 0.005591370165348053,
      "learning_rate": 7.467467467467469e-06,
      "loss": 1.2435,
      "step": 12740
    },
    {
      "epoch": 16.008185328185327,
      "grad_norm": 0.14639759063720703,
      "learning_rate": 7.464607464607465e-06,
      "loss": 1.8811,
      "step": 12750
    },
    {
      "epoch": 16.008442728442727,
      "grad_norm": 0.0185764841735363,
      "learning_rate": 7.461747461747463e-06,
      "loss": 0.5857,
      "step": 12760
    },
    {
      "epoch": 16.008700128700127,
      "grad_norm": 61.76352310180664,
      "learning_rate": 7.458887458887459e-06,
      "loss": 0.6318,
      "step": 12770
    },
    {
      "epoch": 16.008957528957527,
      "grad_norm": 0.20087188482284546,
      "learning_rate": 7.456027456027457e-06,
      "loss": 0.003,
      "step": 12780
    },
    {
      "epoch": 16.00921492921493,
      "grad_norm": 0.11291741579771042,
      "learning_rate": 7.453167453167454e-06,
      "loss": 0.592,
      "step": 12790
    },
    {
      "epoch": 16.00947232947233,
      "grad_norm": 25.189674377441406,
      "learning_rate": 7.450307450307451e-06,
      "loss": 0.9249,
      "step": 12800
    },
    {
      "epoch": 16.00972972972973,
      "grad_norm": 0.01021770853549242,
      "learning_rate": 7.447447447447448e-06,
      "loss": 0.4188,
      "step": 12810
    },
    {
      "epoch": 16.00998712998713,
      "grad_norm": 0.7061002850532532,
      "learning_rate": 7.444587444587446e-06,
      "loss": 0.2926,
      "step": 12820
    },
    {
      "epoch": 16.01024453024453,
      "grad_norm": 0.2825552821159363,
      "learning_rate": 7.441727441727442e-06,
      "loss": 0.0139,
      "step": 12830
    },
    {
      "epoch": 16.01050193050193,
      "grad_norm": 0.2040746808052063,
      "learning_rate": 7.4388674388674395e-06,
      "loss": 0.8342,
      "step": 12840
    },
    {
      "epoch": 16.01075933075933,
      "grad_norm": 127.69252014160156,
      "learning_rate": 7.436007436007436e-06,
      "loss": 0.6337,
      "step": 12850
    },
    {
      "epoch": 16.01101673101673,
      "grad_norm": 284.5594787597656,
      "learning_rate": 7.433147433147434e-06,
      "loss": 1.0989,
      "step": 12860
    },
    {
      "epoch": 16.01127413127413,
      "grad_norm": 0.0073196436278522015,
      "learning_rate": 7.430287430287431e-06,
      "loss": 1.6559,
      "step": 12870
    },
    {
      "epoch": 16.01153153153153,
      "grad_norm": 0.018638137727975845,
      "learning_rate": 7.427427427427428e-06,
      "loss": 1.8099,
      "step": 12880
    },
    {
      "epoch": 16.01178893178893,
      "grad_norm": 0.12463497370481491,
      "learning_rate": 7.4245674245674244e-06,
      "loss": 1.7454,
      "step": 12890
    },
    {
      "epoch": 16.01204633204633,
      "grad_norm": 0.014725065790116787,
      "learning_rate": 7.421707421707423e-06,
      "loss": 1.511,
      "step": 12900
    },
    {
      "epoch": 16.01230373230373,
      "grad_norm": 80.54376220703125,
      "learning_rate": 7.418847418847419e-06,
      "loss": 1.1794,
      "step": 12910
    },
    {
      "epoch": 16.01256113256113,
      "grad_norm": 0.05689245089888573,
      "learning_rate": 7.415987415987416e-06,
      "loss": 0.5597,
      "step": 12920
    },
    {
      "epoch": 16.01281853281853,
      "grad_norm": 63.525150299072266,
      "learning_rate": 7.413127413127414e-06,
      "loss": 1.1223,
      "step": 12930
    },
    {
      "epoch": 16.013075933075932,
      "grad_norm": 0.011247380636632442,
      "learning_rate": 7.410267410267411e-06,
      "loss": 0.6782,
      "step": 12940
    },
    {
      "epoch": 16.013333333333332,
      "grad_norm": 0.007309725042432547,
      "learning_rate": 7.4074074074074075e-06,
      "loss": 0.0069,
      "step": 12950
    },
    {
      "epoch": 16.013590733590732,
      "grad_norm": 0.19261324405670166,
      "learning_rate": 7.404547404547406e-06,
      "loss": 0.4312,
      "step": 12960
    },
    {
      "epoch": 16.013848133848136,
      "grad_norm": 0.2645350992679596,
      "learning_rate": 7.401687401687402e-06,
      "loss": 1.1204,
      "step": 12970
    },
    {
      "epoch": 16.014105534105536,
      "grad_norm": 0.005993553902953863,
      "learning_rate": 7.3988273988273994e-06,
      "loss": 0.9521,
      "step": 12980
    },
    {
      "epoch": 16.014362934362936,
      "grad_norm": 0.12547120451927185,
      "learning_rate": 7.395967395967396e-06,
      "loss": 0.0541,
      "step": 12990
    },
    {
      "epoch": 16.014620334620336,
      "grad_norm": 148.83920288085938,
      "learning_rate": 7.393107393107394e-06,
      "loss": 1.1941,
      "step": 13000
    },
    {
      "epoch": 16.014877734877736,
      "grad_norm": 0.0067084734328091145,
      "learning_rate": 7.3902473902473905e-06,
      "loss": 0.4661,
      "step": 13010
    },
    {
      "epoch": 16.015135135135136,
      "grad_norm": 0.14274343848228455,
      "learning_rate": 7.387387387387388e-06,
      "loss": 0.2892,
      "step": 13020
    },
    {
      "epoch": 16.015392535392536,
      "grad_norm": 284.9540100097656,
      "learning_rate": 7.384527384527384e-06,
      "loss": 0.3677,
      "step": 13030
    },
    {
      "epoch": 16.015649935649936,
      "grad_norm": 0.053901880979537964,
      "learning_rate": 7.3816673816673825e-06,
      "loss": 0.5037,
      "step": 13040
    },
    {
      "epoch": 16.015907335907336,
      "grad_norm": 0.137607142329216,
      "learning_rate": 7.37880737880738e-06,
      "loss": 1.7523,
      "step": 13050
    },
    {
      "epoch": 16.016164736164736,
      "grad_norm": 0.3642958700656891,
      "learning_rate": 7.375947375947376e-06,
      "loss": 1.2588,
      "step": 13060
    },
    {
      "epoch": 16.016422136422136,
      "grad_norm": 28.83527183532715,
      "learning_rate": 7.3730873730873744e-06,
      "loss": 2.3998,
      "step": 13070
    },
    {
      "epoch": 16.016679536679536,
      "grad_norm": 0.5148364901542664,
      "learning_rate": 7.370227370227371e-06,
      "loss": 0.1868,
      "step": 13080
    },
    {
      "epoch": 16.016936936936936,
      "grad_norm": 1.1156384944915771,
      "learning_rate": 7.367367367367368e-06,
      "loss": 0.5132,
      "step": 13090
    },
    {
      "epoch": 16.017194337194336,
      "grad_norm": 219.80577087402344,
      "learning_rate": 7.364507364507365e-06,
      "loss": 0.4853,
      "step": 13100
    },
    {
      "epoch": 16.017451737451736,
      "grad_norm": 0.015949252992868423,
      "learning_rate": 7.361647361647363e-06,
      "loss": 0.1669,
      "step": 13110
    },
    {
      "epoch": 16.017709137709137,
      "grad_norm": 0.011382266879081726,
      "learning_rate": 7.358787358787359e-06,
      "loss": 0.0042,
      "step": 13120
    },
    {
      "epoch": 16.017966537966537,
      "grad_norm": 0.01774335838854313,
      "learning_rate": 7.355927355927357e-06,
      "loss": 0.3558,
      "step": 13130
    },
    {
      "epoch": 16.018223938223937,
      "grad_norm": 0.006449257023632526,
      "learning_rate": 7.353067353067354e-06,
      "loss": 0.6681,
      "step": 13140
    },
    {
      "epoch": 16.018481338481337,
      "grad_norm": 1302.005126953125,
      "learning_rate": 7.350207350207351e-06,
      "loss": 0.9604,
      "step": 13150
    },
    {
      "epoch": 16.01873873873874,
      "grad_norm": 0.00848193746060133,
      "learning_rate": 7.347347347347348e-06,
      "loss": 1.6639,
      "step": 13160
    },
    {
      "epoch": 16.01899613899614,
      "grad_norm": 0.8369548916816711,
      "learning_rate": 7.344487344487345e-06,
      "loss": 0.532,
      "step": 13170
    },
    {
      "epoch": 16.01925353925354,
      "grad_norm": 0.06056103855371475,
      "learning_rate": 7.341627341627342e-06,
      "loss": 0.539,
      "step": 13180
    },
    {
      "epoch": 16.01951093951094,
      "grad_norm": 0.15646892786026,
      "learning_rate": 7.33876733876734e-06,
      "loss": 0.0049,
      "step": 13190
    },
    {
      "epoch": 16.01976833976834,
      "grad_norm": 0.008946855552494526,
      "learning_rate": 7.335907335907336e-06,
      "loss": 0.2423,
      "step": 13200
    },
    {
      "epoch": 16.02,
      "eval_accuracy": 0.9130434782608695,
      "eval_loss": 0.42057546973228455,
      "eval_runtime": 13.3647,
      "eval_samples_per_second": 3.442,
      "eval_steps_per_second": 3.442,
      "step": 13209
    },
    {
      "epoch": 17.00002574002574,
      "grad_norm": 133.33091735839844,
      "learning_rate": 7.333047333047334e-06,
      "loss": 0.813,
      "step": 13210
    },
    {
      "epoch": 17.00028314028314,
      "grad_norm": 0.12156759202480316,
      "learning_rate": 7.330187330187331e-06,
      "loss": 1.0484,
      "step": 13220
    },
    {
      "epoch": 17.00054054054054,
      "grad_norm": 0.07529735565185547,
      "learning_rate": 7.327327327327328e-06,
      "loss": 1.0572,
      "step": 13230
    },
    {
      "epoch": 17.00079794079794,
      "grad_norm": 0.008459187112748623,
      "learning_rate": 7.3244673244673246e-06,
      "loss": 0.4062,
      "step": 13240
    },
    {
      "epoch": 17.00105534105534,
      "grad_norm": 0.029139036312699318,
      "learning_rate": 7.321607321607323e-06,
      "loss": 0.5629,
      "step": 13250
    },
    {
      "epoch": 17.00131274131274,
      "grad_norm": 24.8751220703125,
      "learning_rate": 7.318747318747319e-06,
      "loss": 0.9865,
      "step": 13260
    },
    {
      "epoch": 17.00157014157014,
      "grad_norm": 0.34304794669151306,
      "learning_rate": 7.3158873158873165e-06,
      "loss": 1.1491,
      "step": 13270
    },
    {
      "epoch": 17.00182754182754,
      "grad_norm": 0.24863027036190033,
      "learning_rate": 7.313027313027313e-06,
      "loss": 1.0053,
      "step": 13280
    },
    {
      "epoch": 17.00208494208494,
      "grad_norm": 0.015837261453270912,
      "learning_rate": 7.310167310167311e-06,
      "loss": 0.4432,
      "step": 13290
    },
    {
      "epoch": 17.00234234234234,
      "grad_norm": 0.12219315022230148,
      "learning_rate": 7.307307307307308e-06,
      "loss": 0.5438,
      "step": 13300
    },
    {
      "epoch": 17.00259974259974,
      "grad_norm": 13.343441009521484,
      "learning_rate": 7.304447304447305e-06,
      "loss": 0.0071,
      "step": 13310
    },
    {
      "epoch": 17.002857142857142,
      "grad_norm": 0.2364991009235382,
      "learning_rate": 7.301587301587301e-06,
      "loss": 0.8329,
      "step": 13320
    },
    {
      "epoch": 17.003114543114542,
      "grad_norm": 0.008193550631403923,
      "learning_rate": 7.2987272987272996e-06,
      "loss": 0.0023,
      "step": 13330
    },
    {
      "epoch": 17.003371943371942,
      "grad_norm": 72.55078887939453,
      "learning_rate": 7.295867295867296e-06,
      "loss": 0.5952,
      "step": 13340
    },
    {
      "epoch": 17.003629343629342,
      "grad_norm": 0.3132573664188385,
      "learning_rate": 7.293007293007293e-06,
      "loss": 2.4965,
      "step": 13350
    },
    {
      "epoch": 17.003886743886746,
      "grad_norm": 0.3048330843448639,
      "learning_rate": 7.290147290147291e-06,
      "loss": 1.5725,
      "step": 13360
    },
    {
      "epoch": 17.004144144144146,
      "grad_norm": 0.006918249651789665,
      "learning_rate": 7.287287287287288e-06,
      "loss": 1.2566,
      "step": 13370
    },
    {
      "epoch": 17.004401544401546,
      "grad_norm": 87.47795867919922,
      "learning_rate": 7.2844272844272845e-06,
      "loss": 0.7258,
      "step": 13380
    },
    {
      "epoch": 17.004658944658946,
      "grad_norm": 691.7349243164062,
      "learning_rate": 7.281567281567283e-06,
      "loss": 0.2839,
      "step": 13390
    },
    {
      "epoch": 17.004916344916346,
      "grad_norm": 45.06636047363281,
      "learning_rate": 7.278707278707279e-06,
      "loss": 1.7674,
      "step": 13400
    },
    {
      "epoch": 17.005173745173746,
      "grad_norm": 0.014170120470225811,
      "learning_rate": 7.275847275847276e-06,
      "loss": 0.0174,
      "step": 13410
    },
    {
      "epoch": 17.005431145431146,
      "grad_norm": 0.00938112661242485,
      "learning_rate": 7.272987272987273e-06,
      "loss": 0.6797,
      "step": 13420
    },
    {
      "epoch": 17.005688545688546,
      "grad_norm": 0.02455279231071472,
      "learning_rate": 7.270127270127271e-06,
      "loss": 0.3925,
      "step": 13430
    },
    {
      "epoch": 17.005945945945946,
      "grad_norm": 0.3047299087047577,
      "learning_rate": 7.267267267267268e-06,
      "loss": 0.8212,
      "step": 13440
    },
    {
      "epoch": 17.006203346203346,
      "grad_norm": 0.0063845436088740826,
      "learning_rate": 7.264407264407265e-06,
      "loss": 0.426,
      "step": 13450
    },
    {
      "epoch": 17.006460746460746,
      "grad_norm": 0.745879590511322,
      "learning_rate": 7.261547261547263e-06,
      "loss": 0.6011,
      "step": 13460
    },
    {
      "epoch": 17.006718146718146,
      "grad_norm": 113.57637786865234,
      "learning_rate": 7.2586872586872595e-06,
      "loss": 0.8187,
      "step": 13470
    },
    {
      "epoch": 17.006975546975546,
      "grad_norm": 0.03586834669113159,
      "learning_rate": 7.255827255827257e-06,
      "loss": 0.4669,
      "step": 13480
    },
    {
      "epoch": 17.007232947232946,
      "grad_norm": 0.31438571214675903,
      "learning_rate": 7.252967252967253e-06,
      "loss": 0.2039,
      "step": 13490
    },
    {
      "epoch": 17.007490347490346,
      "grad_norm": 0.03270833194255829,
      "learning_rate": 7.250107250107251e-06,
      "loss": 1.9521,
      "step": 13500
    },
    {
      "epoch": 17.007747747747747,
      "grad_norm": 98.77244567871094,
      "learning_rate": 7.247247247247248e-06,
      "loss": 0.8431,
      "step": 13510
    },
    {
      "epoch": 17.008005148005147,
      "grad_norm": 0.7995626330375671,
      "learning_rate": 7.244387244387245e-06,
      "loss": 0.0078,
      "step": 13520
    },
    {
      "epoch": 17.008262548262547,
      "grad_norm": 24.94700813293457,
      "learning_rate": 7.241527241527242e-06,
      "loss": 1.761,
      "step": 13530
    },
    {
      "epoch": 17.008519948519947,
      "grad_norm": 0.008624196983873844,
      "learning_rate": 7.23866723866724e-06,
      "loss": 0.8168,
      "step": 13540
    },
    {
      "epoch": 17.00877734877735,
      "grad_norm": 0.05530587583780289,
      "learning_rate": 7.235807235807236e-06,
      "loss": 1.7042,
      "step": 13550
    },
    {
      "epoch": 17.00903474903475,
      "grad_norm": 0.15424524247646332,
      "learning_rate": 7.232947232947234e-06,
      "loss": 0.0123,
      "step": 13560
    },
    {
      "epoch": 17.00929214929215,
      "grad_norm": 0.36576831340789795,
      "learning_rate": 7.23008723008723e-06,
      "loss": 0.5522,
      "step": 13570
    },
    {
      "epoch": 17.00954954954955,
      "grad_norm": 0.6042425632476807,
      "learning_rate": 7.227227227227228e-06,
      "loss": 0.9601,
      "step": 13580
    },
    {
      "epoch": 17.00980694980695,
      "grad_norm": 0.17106890678405762,
      "learning_rate": 7.224367224367225e-06,
      "loss": 1.7365,
      "step": 13590
    },
    {
      "epoch": 17.01006435006435,
      "grad_norm": 152.11839294433594,
      "learning_rate": 7.221507221507222e-06,
      "loss": 0.5564,
      "step": 13600
    },
    {
      "epoch": 17.01032175032175,
      "grad_norm": 0.12229585647583008,
      "learning_rate": 7.218647218647219e-06,
      "loss": 1.4063,
      "step": 13610
    },
    {
      "epoch": 17.01057915057915,
      "grad_norm": 428.216796875,
      "learning_rate": 7.215787215787217e-06,
      "loss": 1.8207,
      "step": 13620
    },
    {
      "epoch": 17.01083655083655,
      "grad_norm": 0.4224972128868103,
      "learning_rate": 7.212927212927213e-06,
      "loss": 0.6798,
      "step": 13630
    },
    {
      "epoch": 17.01109395109395,
      "grad_norm": 73.0058364868164,
      "learning_rate": 7.210067210067211e-06,
      "loss": 1.6111,
      "step": 13640
    },
    {
      "epoch": 17.01135135135135,
      "grad_norm": 0.010296664200723171,
      "learning_rate": 7.207207207207208e-06,
      "loss": 0.6907,
      "step": 13650
    },
    {
      "epoch": 17.01160875160875,
      "grad_norm": 0.022636495530605316,
      "learning_rate": 7.204347204347205e-06,
      "loss": 0.0496,
      "step": 13660
    },
    {
      "epoch": 17.01186615186615,
      "grad_norm": 0.02433607541024685,
      "learning_rate": 7.2014872014872015e-06,
      "loss": 0.4299,
      "step": 13670
    },
    {
      "epoch": 17.01212355212355,
      "grad_norm": 0.007355282548815012,
      "learning_rate": 7.1986271986272e-06,
      "loss": 0.4069,
      "step": 13680
    },
    {
      "epoch": 17.01238095238095,
      "grad_norm": 0.00920641515403986,
      "learning_rate": 7.195767195767196e-06,
      "loss": 1.1034,
      "step": 13690
    },
    {
      "epoch": 17.01263835263835,
      "grad_norm": 37.67753601074219,
      "learning_rate": 7.1929071929071935e-06,
      "loss": 2.2347,
      "step": 13700
    },
    {
      "epoch": 17.01289575289575,
      "grad_norm": 0.05537370219826698,
      "learning_rate": 7.19004719004719e-06,
      "loss": 0.2241,
      "step": 13710
    },
    {
      "epoch": 17.01315315315315,
      "grad_norm": 1.393071174621582,
      "learning_rate": 7.187187187187188e-06,
      "loss": 0.5425,
      "step": 13720
    },
    {
      "epoch": 17.013410553410555,
      "grad_norm": 0.037601619958877563,
      "learning_rate": 7.184327184327185e-06,
      "loss": 1.6061,
      "step": 13730
    },
    {
      "epoch": 17.013667953667955,
      "grad_norm": 0.01806103065609932,
      "learning_rate": 7.181467181467182e-06,
      "loss": 1.0232,
      "step": 13740
    },
    {
      "epoch": 17.013925353925355,
      "grad_norm": 0.01816631108522415,
      "learning_rate": 7.178607178607178e-06,
      "loss": 0.0538,
      "step": 13750
    },
    {
      "epoch": 17.014182754182755,
      "grad_norm": 0.005556888412684202,
      "learning_rate": 7.1757471757471765e-06,
      "loss": 0.8444,
      "step": 13760
    },
    {
      "epoch": 17.014440154440155,
      "grad_norm": 16.42436981201172,
      "learning_rate": 7.172887172887173e-06,
      "loss": 0.4929,
      "step": 13770
    },
    {
      "epoch": 17.014697554697555,
      "grad_norm": 2.0494089126586914,
      "learning_rate": 7.17002717002717e-06,
      "loss": 0.5265,
      "step": 13780
    },
    {
      "epoch": 17.014954954954955,
      "grad_norm": 0.4712642431259155,
      "learning_rate": 7.167167167167167e-06,
      "loss": 0.0059,
      "step": 13790
    },
    {
      "epoch": 17.015212355212356,
      "grad_norm": 47.848602294921875,
      "learning_rate": 7.164307164307165e-06,
      "loss": 0.8093,
      "step": 13800
    },
    {
      "epoch": 17.015469755469756,
      "grad_norm": 0.009627053514122963,
      "learning_rate": 7.161447161447162e-06,
      "loss": 0.0029,
      "step": 13810
    },
    {
      "epoch": 17.015727155727156,
      "grad_norm": 0.01003243587911129,
      "learning_rate": 7.158587158587159e-06,
      "loss": 0.0871,
      "step": 13820
    },
    {
      "epoch": 17.015984555984556,
      "grad_norm": 0.07373271137475967,
      "learning_rate": 7.155727155727157e-06,
      "loss": 0.6212,
      "step": 13830
    },
    {
      "epoch": 17.016241956241956,
      "grad_norm": 0.007202493958175182,
      "learning_rate": 7.152867152867153e-06,
      "loss": 0.4612,
      "step": 13840
    },
    {
      "epoch": 17.016499356499356,
      "grad_norm": 0.01294265128672123,
      "learning_rate": 7.150007150007151e-06,
      "loss": 0.7214,
      "step": 13850
    },
    {
      "epoch": 17.016756756756756,
      "grad_norm": 0.005408270284533501,
      "learning_rate": 7.147147147147148e-06,
      "loss": 0.6311,
      "step": 13860
    },
    {
      "epoch": 17.017014157014156,
      "grad_norm": 215.44699096679688,
      "learning_rate": 7.144287144287145e-06,
      "loss": 1.4436,
      "step": 13870
    },
    {
      "epoch": 17.017271557271556,
      "grad_norm": 0.01759917289018631,
      "learning_rate": 7.141427141427142e-06,
      "loss": 0.5106,
      "step": 13880
    },
    {
      "epoch": 17.017528957528956,
      "grad_norm": 0.02081468515098095,
      "learning_rate": 7.13856713856714e-06,
      "loss": 1.7547,
      "step": 13890
    },
    {
      "epoch": 17.017786357786356,
      "grad_norm": 0.6049190163612366,
      "learning_rate": 7.1357071357071364e-06,
      "loss": 1.6229,
      "step": 13900
    },
    {
      "epoch": 17.01804375804376,
      "grad_norm": 0.9745853543281555,
      "learning_rate": 7.132847132847134e-06,
      "loss": 0.484,
      "step": 13910
    },
    {
      "epoch": 17.01830115830116,
      "grad_norm": 0.10427520424127579,
      "learning_rate": 7.12998712998713e-06,
      "loss": 0.0048,
      "step": 13920
    },
    {
      "epoch": 17.01855855855856,
      "grad_norm": 0.09744715690612793,
      "learning_rate": 7.127127127127128e-06,
      "loss": 0.8102,
      "step": 13930
    },
    {
      "epoch": 17.01881595881596,
      "grad_norm": 0.26503968238830566,
      "learning_rate": 7.124267124267125e-06,
      "loss": 2.1186,
      "step": 13940
    },
    {
      "epoch": 17.01907335907336,
      "grad_norm": 0.0076754712499678135,
      "learning_rate": 7.121407121407122e-06,
      "loss": 0.5523,
      "step": 13950
    },
    {
      "epoch": 17.01933075933076,
      "grad_norm": 0.008721827529370785,
      "learning_rate": 7.118547118547119e-06,
      "loss": 0.9198,
      "step": 13960
    },
    {
      "epoch": 17.01958815958816,
      "grad_norm": 0.272495836019516,
      "learning_rate": 7.115687115687117e-06,
      "loss": 1.0507,
      "step": 13970
    },
    {
      "epoch": 17.01984555984556,
      "grad_norm": 0.43529975414276123,
      "learning_rate": 7.112827112827113e-06,
      "loss": 1.4765,
      "step": 13980
    },
    {
      "epoch": 17.02,
      "eval_accuracy": 0.8478260869565217,
      "eval_loss": 0.5210566520690918,
      "eval_runtime": 13.421,
      "eval_samples_per_second": 3.427,
      "eval_steps_per_second": 3.427,
      "step": 13986
    },
    {
      "epoch": 18.00010296010296,
      "grad_norm": 138.43173217773438,
      "learning_rate": 7.109967109967111e-06,
      "loss": 2.2081,
      "step": 13990
    },
    {
      "epoch": 18.00036036036036,
      "grad_norm": 0.004014770966023207,
      "learning_rate": 7.107107107107107e-06,
      "loss": 0.3443,
      "step": 14000
    },
    {
      "epoch": 18.00061776061776,
      "grad_norm": 0.5814266204833984,
      "learning_rate": 7.104247104247105e-06,
      "loss": 0.2336,
      "step": 14010
    },
    {
      "epoch": 18.00087516087516,
      "grad_norm": 2.1018807888031006,
      "learning_rate": 7.101387101387102e-06,
      "loss": 0.4295,
      "step": 14020
    },
    {
      "epoch": 18.00113256113256,
      "grad_norm": 0.5535208582878113,
      "learning_rate": 7.098527098527099e-06,
      "loss": 0.1623,
      "step": 14030
    },
    {
      "epoch": 18.00138996138996,
      "grad_norm": 0.011099116876721382,
      "learning_rate": 7.0956670956670955e-06,
      "loss": 0.0023,
      "step": 14040
    },
    {
      "epoch": 18.00164736164736,
      "grad_norm": 0.01819237321615219,
      "learning_rate": 7.092807092807094e-06,
      "loss": 0.0015,
      "step": 14050
    },
    {
      "epoch": 18.00190476190476,
      "grad_norm": 0.003633988555520773,
      "learning_rate": 7.08994708994709e-06,
      "loss": 0.9,
      "step": 14060
    },
    {
      "epoch": 18.00216216216216,
      "grad_norm": 551.4736328125,
      "learning_rate": 7.087087087087087e-06,
      "loss": 0.7453,
      "step": 14070
    },
    {
      "epoch": 18.00241956241956,
      "grad_norm": 0.003984920680522919,
      "learning_rate": 7.084227084227085e-06,
      "loss": 0.5739,
      "step": 14080
    },
    {
      "epoch": 18.00267696267696,
      "grad_norm": 51.20891189575195,
      "learning_rate": 7.081367081367082e-06,
      "loss": 0.6431,
      "step": 14090
    },
    {
      "epoch": 18.00293436293436,
      "grad_norm": 0.01216127909719944,
      "learning_rate": 7.0785070785070785e-06,
      "loss": 0.4871,
      "step": 14100
    },
    {
      "epoch": 18.00319176319176,
      "grad_norm": 0.3225768506526947,
      "learning_rate": 7.075647075647077e-06,
      "loss": 0.0077,
      "step": 14110
    },
    {
      "epoch": 18.003449163449165,
      "grad_norm": 809.6256103515625,
      "learning_rate": 7.072787072787073e-06,
      "loss": 2.0674,
      "step": 14120
    },
    {
      "epoch": 18.003706563706565,
      "grad_norm": 6.774144649505615,
      "learning_rate": 7.0699270699270705e-06,
      "loss": 0.8586,
      "step": 14130
    },
    {
      "epoch": 18.003963963963965,
      "grad_norm": 62.32297897338867,
      "learning_rate": 7.067067067067067e-06,
      "loss": 2.386,
      "step": 14140
    },
    {
      "epoch": 18.004221364221365,
      "grad_norm": 0.009228847920894623,
      "learning_rate": 7.064207064207065e-06,
      "loss": 1.2169,
      "step": 14150
    },
    {
      "epoch": 18.004478764478765,
      "grad_norm": 0.3130154013633728,
      "learning_rate": 7.0613470613470616e-06,
      "loss": 0.0056,
      "step": 14160
    },
    {
      "epoch": 18.004736164736165,
      "grad_norm": 0.15906237065792084,
      "learning_rate": 7.058487058487059e-06,
      "loss": 0.8068,
      "step": 14170
    },
    {
      "epoch": 18.004993564993566,
      "grad_norm": 0.21492519974708557,
      "learning_rate": 7.055627055627057e-06,
      "loss": 0.8975,
      "step": 14180
    },
    {
      "epoch": 18.005250965250966,
      "grad_norm": 972.6399536132812,
      "learning_rate": 7.0527670527670535e-06,
      "loss": 0.9294,
      "step": 14190
    },
    {
      "epoch": 18.005508365508366,
      "grad_norm": 0.2181262969970703,
      "learning_rate": 7.049907049907051e-06,
      "loss": 0.3442,
      "step": 14200
    },
    {
      "epoch": 18.005765765765766,
      "grad_norm": 0.010371779091656208,
      "learning_rate": 7.047047047047047e-06,
      "loss": 0.9992,
      "step": 14210
    },
    {
      "epoch": 18.006023166023166,
      "grad_norm": 0.16542059183120728,
      "learning_rate": 7.0441870441870455e-06,
      "loss": 1.0675,
      "step": 14220
    },
    {
      "epoch": 18.006280566280566,
      "grad_norm": 0.0024146682117134333,
      "learning_rate": 7.041327041327042e-06,
      "loss": 0.8035,
      "step": 14230
    },
    {
      "epoch": 18.006537966537966,
      "grad_norm": 0.00679363775998354,
      "learning_rate": 7.038467038467039e-06,
      "loss": 1.4576,
      "step": 14240
    },
    {
      "epoch": 18.006795366795366,
      "grad_norm": 42.69426345825195,
      "learning_rate": 7.035607035607036e-06,
      "loss": 1.4004,
      "step": 14250
    },
    {
      "epoch": 18.007052767052766,
      "grad_norm": 0.24903038144111633,
      "learning_rate": 7.032747032747034e-06,
      "loss": 0.0173,
      "step": 14260
    },
    {
      "epoch": 18.007310167310166,
      "grad_norm": 0.251482754945755,
      "learning_rate": 7.02988702988703e-06,
      "loss": 0.8903,
      "step": 14270
    },
    {
      "epoch": 18.007567567567566,
      "grad_norm": 0.15955966711044312,
      "learning_rate": 7.027027027027028e-06,
      "loss": 0.5083,
      "step": 14280
    },
    {
      "epoch": 18.007824967824966,
      "grad_norm": 0.08874122053384781,
      "learning_rate": 7.024167024167024e-06,
      "loss": 0.0515,
      "step": 14290
    },
    {
      "epoch": 18.00808236808237,
      "grad_norm": 193.19786071777344,
      "learning_rate": 7.021307021307022e-06,
      "loss": 1.0134,
      "step": 14300
    },
    {
      "epoch": 18.00833976833977,
      "grad_norm": 0.0486479327082634,
      "learning_rate": 7.018447018447019e-06,
      "loss": 1.2563,
      "step": 14310
    },
    {
      "epoch": 18.00859716859717,
      "grad_norm": 0.09023340046405792,
      "learning_rate": 7.015587015587016e-06,
      "loss": 0.8956,
      "step": 14320
    },
    {
      "epoch": 18.00885456885457,
      "grad_norm": 942.4277954101562,
      "learning_rate": 7.012727012727013e-06,
      "loss": 0.6395,
      "step": 14330
    },
    {
      "epoch": 18.00911196911197,
      "grad_norm": 0.028225857764482498,
      "learning_rate": 7.009867009867011e-06,
      "loss": 1.329,
      "step": 14340
    },
    {
      "epoch": 18.00936936936937,
      "grad_norm": 0.1975196748971939,
      "learning_rate": 7.007007007007007e-06,
      "loss": 0.009,
      "step": 14350
    },
    {
      "epoch": 18.00962676962677,
      "grad_norm": 0.29792848229408264,
      "learning_rate": 7.004147004147005e-06,
      "loss": 0.4711,
      "step": 14360
    },
    {
      "epoch": 18.00988416988417,
      "grad_norm": 0.003600794356316328,
      "learning_rate": 7.001287001287002e-06,
      "loss": 0.9534,
      "step": 14370
    },
    {
      "epoch": 18.01014157014157,
      "grad_norm": 0.4418034851551056,
      "learning_rate": 6.998426998426999e-06,
      "loss": 0.0036,
      "step": 14380
    },
    {
      "epoch": 18.01039897039897,
      "grad_norm": 53.621925354003906,
      "learning_rate": 6.995566995566996e-06,
      "loss": 1.7642,
      "step": 14390
    },
    {
      "epoch": 18.01065637065637,
      "grad_norm": 0.22013112902641296,
      "learning_rate": 6.992706992706994e-06,
      "loss": 1.1519,
      "step": 14400
    },
    {
      "epoch": 18.01091377091377,
      "grad_norm": 0.009073288179934025,
      "learning_rate": 6.98984698984699e-06,
      "loss": 1.3366,
      "step": 14410
    },
    {
      "epoch": 18.01117117117117,
      "grad_norm": 0.24850936233997345,
      "learning_rate": 6.9869869869869876e-06,
      "loss": 0.0021,
      "step": 14420
    },
    {
      "epoch": 18.01142857142857,
      "grad_norm": 354.8047790527344,
      "learning_rate": 6.984126984126984e-06,
      "loss": 0.0833,
      "step": 14430
    },
    {
      "epoch": 18.01168597168597,
      "grad_norm": 0.001540736760944128,
      "learning_rate": 6.981266981266982e-06,
      "loss": 0.0097,
      "step": 14440
    },
    {
      "epoch": 18.01194337194337,
      "grad_norm": 23.982343673706055,
      "learning_rate": 6.978406978406979e-06,
      "loss": 3.2159,
      "step": 14450
    },
    {
      "epoch": 18.01220077220077,
      "grad_norm": 0.686680793762207,
      "learning_rate": 6.975546975546976e-06,
      "loss": 0.4192,
      "step": 14460
    },
    {
      "epoch": 18.01245817245817,
      "grad_norm": 0.17515744268894196,
      "learning_rate": 6.9726869726869724e-06,
      "loss": 0.0089,
      "step": 14470
    },
    {
      "epoch": 18.01271557271557,
      "grad_norm": 0.6502960920333862,
      "learning_rate": 6.969826969826971e-06,
      "loss": 0.8773,
      "step": 14480
    },
    {
      "epoch": 18.012972972972975,
      "grad_norm": 81.23369598388672,
      "learning_rate": 6.966966966966967e-06,
      "loss": 0.9279,
      "step": 14490
    },
    {
      "epoch": 18.013230373230375,
      "grad_norm": 0.2982556223869324,
      "learning_rate": 6.964106964106964e-06,
      "loss": 0.4003,
      "step": 14500
    },
    {
      "epoch": 18.013487773487775,
      "grad_norm": 0.008642779663205147,
      "learning_rate": 6.961246961246962e-06,
      "loss": 0.8602,
      "step": 14510
    },
    {
      "epoch": 18.013745173745175,
      "grad_norm": 0.18318037688732147,
      "learning_rate": 6.958386958386959e-06,
      "loss": 0.005,
      "step": 14520
    },
    {
      "epoch": 18.014002574002575,
      "grad_norm": 0.3546335697174072,
      "learning_rate": 6.9555269555269555e-06,
      "loss": 0.0034,
      "step": 14530
    },
    {
      "epoch": 18.014259974259975,
      "grad_norm": 0.041229523718357086,
      "learning_rate": 6.952666952666954e-06,
      "loss": 0.0029,
      "step": 14540
    },
    {
      "epoch": 18.014517374517375,
      "grad_norm": 0.09673618525266647,
      "learning_rate": 6.949806949806951e-06,
      "loss": 0.0661,
      "step": 14550
    },
    {
      "epoch": 18.014774774774775,
      "grad_norm": 0.030932040885090828,
      "learning_rate": 6.9469469469469474e-06,
      "loss": 1.5543,
      "step": 14560
    },
    {
      "epoch": 18.015032175032175,
      "grad_norm": 0.1227671205997467,
      "learning_rate": 6.944086944086945e-06,
      "loss": 0.0012,
      "step": 14570
    },
    {
      "epoch": 18.015289575289575,
      "grad_norm": 0.15442447364330292,
      "learning_rate": 6.941226941226942e-06,
      "loss": 0.4028,
      "step": 14580
    },
    {
      "epoch": 18.015546975546975,
      "grad_norm": 525.80224609375,
      "learning_rate": 6.938366938366939e-06,
      "loss": 0.3612,
      "step": 14590
    },
    {
      "epoch": 18.015804375804375,
      "grad_norm": 1.3030730485916138,
      "learning_rate": 6.935506935506936e-06,
      "loss": 0.0142,
      "step": 14600
    },
    {
      "epoch": 18.016061776061775,
      "grad_norm": 0.0016896520974114537,
      "learning_rate": 6.932646932646934e-06,
      "loss": 1.0281,
      "step": 14610
    },
    {
      "epoch": 18.016319176319175,
      "grad_norm": 479.7268981933594,
      "learning_rate": 6.9297869297869305e-06,
      "loss": 0.3177,
      "step": 14620
    },
    {
      "epoch": 18.016576576576576,
      "grad_norm": 0.004493164364248514,
      "learning_rate": 6.926926926926928e-06,
      "loss": 0.6365,
      "step": 14630
    },
    {
      "epoch": 18.016833976833976,
      "grad_norm": 0.15360552072525024,
      "learning_rate": 6.924066924066924e-06,
      "loss": 0.0011,
      "step": 14640
    },
    {
      "epoch": 18.017091377091376,
      "grad_norm": 28.01856231689453,
      "learning_rate": 6.9212069212069224e-06,
      "loss": 0.9447,
      "step": 14650
    },
    {
      "epoch": 18.017348777348776,
      "grad_norm": 77.52731323242188,
      "learning_rate": 6.918346918346919e-06,
      "loss": 1.6479,
      "step": 14660
    },
    {
      "epoch": 18.01760617760618,
      "grad_norm": 0.0018540214514359832,
      "learning_rate": 6.915486915486916e-06,
      "loss": 0.0014,
      "step": 14670
    },
    {
      "epoch": 18.01786357786358,
      "grad_norm": 5.362928867340088,
      "learning_rate": 6.912626912626913e-06,
      "loss": 1.5786,
      "step": 14680
    },
    {
      "epoch": 18.01812097812098,
      "grad_norm": 0.010288232006132603,
      "learning_rate": 6.909766909766911e-06,
      "loss": 1.0865,
      "step": 14690
    },
    {
      "epoch": 18.01837837837838,
      "grad_norm": 0.00446264399215579,
      "learning_rate": 6.906906906906907e-06,
      "loss": 0.8953,
      "step": 14700
    },
    {
      "epoch": 18.01863577863578,
      "grad_norm": 0.036626916378736496,
      "learning_rate": 6.904046904046905e-06,
      "loss": 1.5741,
      "step": 14710
    },
    {
      "epoch": 18.01889317889318,
      "grad_norm": 105.58182525634766,
      "learning_rate": 6.901186901186901e-06,
      "loss": 0.4651,
      "step": 14720
    },
    {
      "epoch": 18.01915057915058,
      "grad_norm": 1.4560140371322632,
      "learning_rate": 6.898326898326899e-06,
      "loss": 1.017,
      "step": 14730
    },
    {
      "epoch": 18.01940797940798,
      "grad_norm": 0.0017882263055071235,
      "learning_rate": 6.895466895466896e-06,
      "loss": 0.853,
      "step": 14740
    },
    {
      "epoch": 18.01966537966538,
      "grad_norm": 0.20019206404685974,
      "learning_rate": 6.892606892606893e-06,
      "loss": 1.3635,
      "step": 14750
    },
    {
      "epoch": 18.01992277992278,
      "grad_norm": 0.003446888877078891,
      "learning_rate": 6.88974688974689e-06,
      "loss": 0.4732,
      "step": 14760
    },
    {
      "epoch": 18.02,
      "eval_accuracy": 0.8260869565217391,
      "eval_loss": 0.8912200927734375,
      "eval_runtime": 13.4927,
      "eval_samples_per_second": 3.409,
      "eval_steps_per_second": 3.409,
      "step": 14763
    },
    {
      "epoch": 19.00018018018018,
      "grad_norm": 0.819979727268219,
      "learning_rate": 6.886886886886888e-06,
      "loss": 0.479,
      "step": 14770
    },
    {
      "epoch": 19.00043758043758,
      "grad_norm": 0.3846934735774994,
      "learning_rate": 6.884026884026884e-06,
      "loss": 0.8816,
      "step": 14780
    },
    {
      "epoch": 19.00069498069498,
      "grad_norm": 0.28570055961608887,
      "learning_rate": 6.881166881166882e-06,
      "loss": 1.1393,
      "step": 14790
    },
    {
      "epoch": 19.00095238095238,
      "grad_norm": 0.0016494010342285037,
      "learning_rate": 6.878306878306879e-06,
      "loss": 0.0054,
      "step": 14800
    },
    {
      "epoch": 19.00120978120978,
      "grad_norm": 0.13023339211940765,
      "learning_rate": 6.875446875446876e-06,
      "loss": 0.0038,
      "step": 14810
    },
    {
      "epoch": 19.00146718146718,
      "grad_norm": 0.001983684953302145,
      "learning_rate": 6.872586872586873e-06,
      "loss": 0.4917,
      "step": 14820
    },
    {
      "epoch": 19.00172458172458,
      "grad_norm": 0.16467513144016266,
      "learning_rate": 6.869726869726871e-06,
      "loss": 0.4895,
      "step": 14830
    },
    {
      "epoch": 19.00198198198198,
      "grad_norm": 0.004174495581537485,
      "learning_rate": 6.866866866866867e-06,
      "loss": 0.0456,
      "step": 14840
    },
    {
      "epoch": 19.00223938223938,
      "grad_norm": 96.41185760498047,
      "learning_rate": 6.8640068640068645e-06,
      "loss": 0.9856,
      "step": 14850
    },
    {
      "epoch": 19.00249678249678,
      "grad_norm": 0.012012992985546589,
      "learning_rate": 6.861146861146861e-06,
      "loss": 0.7848,
      "step": 14860
    },
    {
      "epoch": 19.00275418275418,
      "grad_norm": 0.17765195667743683,
      "learning_rate": 6.858286858286859e-06,
      "loss": 0.5042,
      "step": 14870
    },
    {
      "epoch": 19.003011583011585,
      "grad_norm": 0.0029575249645859003,
      "learning_rate": 6.855426855426856e-06,
      "loss": 0.499,
      "step": 14880
    },
    {
      "epoch": 19.003268983268985,
      "grad_norm": 81.89859771728516,
      "learning_rate": 6.852566852566853e-06,
      "loss": 0.0079,
      "step": 14890
    },
    {
      "epoch": 19.003526383526385,
      "grad_norm": 0.0032242753077298403,
      "learning_rate": 6.849706849706849e-06,
      "loss": 0.0018,
      "step": 14900
    },
    {
      "epoch": 19.003783783783785,
      "grad_norm": 0.15290415287017822,
      "learning_rate": 6.846846846846848e-06,
      "loss": 1.2286,
      "step": 14910
    },
    {
      "epoch": 19.004041184041185,
      "grad_norm": 0.001420488115400076,
      "learning_rate": 6.843986843986845e-06,
      "loss": 0.478,
      "step": 14920
    },
    {
      "epoch": 19.004298584298585,
      "grad_norm": 0.009055113419890404,
      "learning_rate": 6.841126841126841e-06,
      "loss": 0.8538,
      "step": 14930
    },
    {
      "epoch": 19.004555984555985,
      "grad_norm": 0.001777785480953753,
      "learning_rate": 6.8382668382668395e-06,
      "loss": 0.0001,
      "step": 14940
    },
    {
      "epoch": 19.004813384813385,
      "grad_norm": 0.33212217688560486,
      "learning_rate": 6.835406835406836e-06,
      "loss": 1.249,
      "step": 14950
    },
    {
      "epoch": 19.005070785070785,
      "grad_norm": 0.17998294532299042,
      "learning_rate": 6.832546832546833e-06,
      "loss": 0.0113,
      "step": 14960
    },
    {
      "epoch": 19.005328185328185,
      "grad_norm": 0.004125115927308798,
      "learning_rate": 6.82968682968683e-06,
      "loss": 1.0992,
      "step": 14970
    },
    {
      "epoch": 19.005585585585585,
      "grad_norm": 0.06852870434522629,
      "learning_rate": 6.826826826826828e-06,
      "loss": 0.4464,
      "step": 14980
    },
    {
      "epoch": 19.005842985842985,
      "grad_norm": 0.272107869386673,
      "learning_rate": 6.823966823966824e-06,
      "loss": 1.0914,
      "step": 14990
    },
    {
      "epoch": 19.006100386100385,
      "grad_norm": 0.035318005830049515,
      "learning_rate": 6.821106821106822e-06,
      "loss": 0.9878,
      "step": 15000
    },
    {
      "epoch": 19.006357786357785,
      "grad_norm": 2.790893793106079,
      "learning_rate": 6.818246818246819e-06,
      "loss": 1.3259,
      "step": 15010
    },
    {
      "epoch": 19.006615186615186,
      "grad_norm": 79.72868347167969,
      "learning_rate": 6.815386815386816e-06,
      "loss": 0.5364,
      "step": 15020
    },
    {
      "epoch": 19.006872586872586,
      "grad_norm": 0.45355862379074097,
      "learning_rate": 6.812526812526813e-06,
      "loss": 0.4649,
      "step": 15030
    },
    {
      "epoch": 19.007129987129986,
      "grad_norm": 954.1507568359375,
      "learning_rate": 6.809666809666811e-06,
      "loss": 1.2683,
      "step": 15040
    },
    {
      "epoch": 19.007387387387386,
      "grad_norm": 143.90093994140625,
      "learning_rate": 6.8068068068068075e-06,
      "loss": 0.5476,
      "step": 15050
    },
    {
      "epoch": 19.00764478764479,
      "grad_norm": 0.19214926660060883,
      "learning_rate": 6.803946803946805e-06,
      "loss": 0.3,
      "step": 15060
    },
    {
      "epoch": 19.00790218790219,
      "grad_norm": 0.002590177347883582,
      "learning_rate": 6.801086801086801e-06,
      "loss": 0.002,
      "step": 15070
    },
    {
      "epoch": 19.00815958815959,
      "grad_norm": 0.030673101544380188,
      "learning_rate": 6.798226798226799e-06,
      "loss": 1.4,
      "step": 15080
    },
    {
      "epoch": 19.00841698841699,
      "grad_norm": 0.09079337120056152,
      "learning_rate": 6.795366795366796e-06,
      "loss": 0.0028,
      "step": 15090
    },
    {
      "epoch": 19.00867438867439,
      "grad_norm": 0.034733597189188004,
      "learning_rate": 6.792506792506793e-06,
      "loss": 1.5467,
      "step": 15100
    },
    {
      "epoch": 19.00893178893179,
      "grad_norm": 0.010993434116244316,
      "learning_rate": 6.78964678964679e-06,
      "loss": 0.9666,
      "step": 15110
    },
    {
      "epoch": 19.00918918918919,
      "grad_norm": 0.07968287169933319,
      "learning_rate": 6.786786786786788e-06,
      "loss": 0.9137,
      "step": 15120
    },
    {
      "epoch": 19.00944658944659,
      "grad_norm": 0.09925149381160736,
      "learning_rate": 6.783926783926784e-06,
      "loss": 0.1951,
      "step": 15130
    },
    {
      "epoch": 19.00970398970399,
      "grad_norm": 247.01406860351562,
      "learning_rate": 6.781066781066782e-06,
      "loss": 0.7136,
      "step": 15140
    },
    {
      "epoch": 19.00996138996139,
      "grad_norm": 0.009178612381219864,
      "learning_rate": 6.778206778206778e-06,
      "loss": 1.4312,
      "step": 15150
    },
    {
      "epoch": 19.01021879021879,
      "grad_norm": 0.0027663582004606724,
      "learning_rate": 6.775346775346776e-06,
      "loss": 0.7116,
      "step": 15160
    },
    {
      "epoch": 19.01047619047619,
      "grad_norm": 0.1210724487900734,
      "learning_rate": 6.772486772486773e-06,
      "loss": 0.5382,
      "step": 15170
    },
    {
      "epoch": 19.01073359073359,
      "grad_norm": 28.404563903808594,
      "learning_rate": 6.76962676962677e-06,
      "loss": 0.7048,
      "step": 15180
    },
    {
      "epoch": 19.01099099099099,
      "grad_norm": 0.0018646479584276676,
      "learning_rate": 6.7667667667667665e-06,
      "loss": 0.7556,
      "step": 15190
    },
    {
      "epoch": 19.01124839124839,
      "grad_norm": 0.003445609239861369,
      "learning_rate": 6.763906763906765e-06,
      "loss": 1.1341,
      "step": 15200
    },
    {
      "epoch": 19.01150579150579,
      "grad_norm": 0.0021248559933155775,
      "learning_rate": 6.761046761046761e-06,
      "loss": 1.7684,
      "step": 15210
    },
    {
      "epoch": 19.01176319176319,
      "grad_norm": 0.014535117894411087,
      "learning_rate": 6.7581867581867585e-06,
      "loss": 0.0021,
      "step": 15220
    },
    {
      "epoch": 19.01202059202059,
      "grad_norm": 0.07674351334571838,
      "learning_rate": 6.755326755326756e-06,
      "loss": 1.07,
      "step": 15230
    },
    {
      "epoch": 19.01227799227799,
      "grad_norm": 37.96657943725586,
      "learning_rate": 6.752466752466753e-06,
      "loss": 1.5324,
      "step": 15240
    },
    {
      "epoch": 19.012535392535394,
      "grad_norm": 0.5922670960426331,
      "learning_rate": 6.7496067496067496e-06,
      "loss": 0.5106,
      "step": 15250
    },
    {
      "epoch": 19.012792792792794,
      "grad_norm": 0.2684486210346222,
      "learning_rate": 6.746746746746748e-06,
      "loss": 0.82,
      "step": 15260
    },
    {
      "epoch": 19.013050193050194,
      "grad_norm": 0.003213370218873024,
      "learning_rate": 6.743886743886744e-06,
      "loss": 0.0033,
      "step": 15270
    },
    {
      "epoch": 19.013307593307594,
      "grad_norm": 0.004637459293007851,
      "learning_rate": 6.7410267410267415e-06,
      "loss": 0.5989,
      "step": 15280
    },
    {
      "epoch": 19.013564993564994,
      "grad_norm": 22.795841217041016,
      "learning_rate": 6.738166738166738e-06,
      "loss": 0.6718,
      "step": 15290
    },
    {
      "epoch": 19.013822393822394,
      "grad_norm": 30.51113510131836,
      "learning_rate": 6.735306735306736e-06,
      "loss": 0.9449,
      "step": 15300
    },
    {
      "epoch": 19.014079794079795,
      "grad_norm": 0.10669000446796417,
      "learning_rate": 6.7324467324467335e-06,
      "loss": 0.0043,
      "step": 15310
    },
    {
      "epoch": 19.014337194337195,
      "grad_norm": 0.10424339026212692,
      "learning_rate": 6.72958672958673e-06,
      "loss": 1.5502,
      "step": 15320
    },
    {
      "epoch": 19.014594594594595,
      "grad_norm": 0.010130718350410461,
      "learning_rate": 6.726726726726728e-06,
      "loss": 0.9745,
      "step": 15330
    },
    {
      "epoch": 19.014851994851995,
      "grad_norm": 0.04613509774208069,
      "learning_rate": 6.7238667238667246e-06,
      "loss": 1.0258,
      "step": 15340
    },
    {
      "epoch": 19.015109395109395,
      "grad_norm": 0.014101505279541016,
      "learning_rate": 6.721006721006722e-06,
      "loss": 0.4591,
      "step": 15350
    },
    {
      "epoch": 19.015366795366795,
      "grad_norm": 0.2920877933502197,
      "learning_rate": 6.718146718146718e-06,
      "loss": 0.8987,
      "step": 15360
    },
    {
      "epoch": 19.015624195624195,
      "grad_norm": 0.003366332734003663,
      "learning_rate": 6.7152867152867165e-06,
      "loss": 0.5684,
      "step": 15370
    },
    {
      "epoch": 19.015881595881595,
      "grad_norm": 0.014356138184666634,
      "learning_rate": 6.712426712426713e-06,
      "loss": 0.9558,
      "step": 15380
    },
    {
      "epoch": 19.016138996138995,
      "grad_norm": 0.033906999975442886,
      "learning_rate": 6.70956670956671e-06,
      "loss": 1.3507,
      "step": 15390
    },
    {
      "epoch": 19.016396396396395,
      "grad_norm": 0.18974842131137848,
      "learning_rate": 6.706706706706707e-06,
      "loss": 0.0257,
      "step": 15400
    },
    {
      "epoch": 19.016653796653795,
      "grad_norm": 0.008604479022324085,
      "learning_rate": 6.703846703846705e-06,
      "loss": 1.7851,
      "step": 15410
    },
    {
      "epoch": 19.016911196911195,
      "grad_norm": 0.2468060553073883,
      "learning_rate": 6.700986700986701e-06,
      "loss": 0.8377,
      "step": 15420
    },
    {
      "epoch": 19.0171685971686,
      "grad_norm": 0.025349579751491547,
      "learning_rate": 6.698126698126699e-06,
      "loss": 0.4866,
      "step": 15430
    },
    {
      "epoch": 19.017425997426,
      "grad_norm": 0.39254313707351685,
      "learning_rate": 6.695266695266695e-06,
      "loss": 0.4546,
      "step": 15440
    },
    {
      "epoch": 19.0176833976834,
      "grad_norm": 25.649742126464844,
      "learning_rate": 6.692406692406693e-06,
      "loss": 0.8261,
      "step": 15450
    },
    {
      "epoch": 19.0179407979408,
      "grad_norm": 0.19370250403881073,
      "learning_rate": 6.68954668954669e-06,
      "loss": 0.3301,
      "step": 15460
    },
    {
      "epoch": 19.0181981981982,
      "grad_norm": 70.06546783447266,
      "learning_rate": 6.686686686686687e-06,
      "loss": 1.0178,
      "step": 15470
    },
    {
      "epoch": 19.0184555984556,
      "grad_norm": 0.29361337423324585,
      "learning_rate": 6.6838266838266844e-06,
      "loss": 0.8067,
      "step": 15480
    },
    {
      "epoch": 19.018712998713,
      "grad_norm": 0.00261647067964077,
      "learning_rate": 6.680966680966682e-06,
      "loss": 1.6859,
      "step": 15490
    },
    {
      "epoch": 19.0189703989704,
      "grad_norm": 0.1030469685792923,
      "learning_rate": 6.678106678106678e-06,
      "loss": 1.0168,
      "step": 15500
    },
    {
      "epoch": 19.0192277992278,
      "grad_norm": 42.0329475402832,
      "learning_rate": 6.675246675246676e-06,
      "loss": 1.385,
      "step": 15510
    },
    {
      "epoch": 19.0194851994852,
      "grad_norm": 0.21290437877178192,
      "learning_rate": 6.672386672386673e-06,
      "loss": 0.2358,
      "step": 15520
    },
    {
      "epoch": 19.0197425997426,
      "grad_norm": 10.05051326751709,
      "learning_rate": 6.66952666952667e-06,
      "loss": 1.1642,
      "step": 15530
    },
    {
      "epoch": 19.02,
      "grad_norm": 0.021836817264556885,
      "learning_rate": 6.666666666666667e-06,
      "loss": 0.8663,
      "step": 15540
    },
    {
      "epoch": 19.02,
      "eval_accuracy": 0.8695652173913043,
      "eval_loss": 0.4560748338699341,
      "eval_runtime": 13.467,
      "eval_samples_per_second": 3.416,
      "eval_steps_per_second": 3.416,
      "step": 15540
    },
    {
      "epoch": 20.0002574002574,
      "grad_norm": 0.010793312452733517,
      "learning_rate": 6.663806663806665e-06,
      "loss": 0.0118,
      "step": 15550
    },
    {
      "epoch": 20.0005148005148,
      "grad_norm": 0.004275476094335318,
      "learning_rate": 6.660946660946661e-06,
      "loss": 0.0026,
      "step": 15560
    },
    {
      "epoch": 20.0007722007722,
      "grad_norm": 0.0035458989441394806,
      "learning_rate": 6.658086658086659e-06,
      "loss": 0.8338,
      "step": 15570
    },
    {
      "epoch": 20.0010296010296,
      "grad_norm": 0.002316601574420929,
      "learning_rate": 6.655226655226655e-06,
      "loss": 1.4684,
      "step": 15580
    },
    {
      "epoch": 20.001287001287,
      "grad_norm": 0.003117610700428486,
      "learning_rate": 6.652366652366653e-06,
      "loss": 0.7738,
      "step": 15590
    },
    {
      "epoch": 20.0015444015444,
      "grad_norm": 0.004058063495904207,
      "learning_rate": 6.64950664950665e-06,
      "loss": 1.7848,
      "step": 15600
    },
    {
      "epoch": 20.0018018018018,
      "grad_norm": 3.553769111633301,
      "learning_rate": 6.646646646646647e-06,
      "loss": 1.1946,
      "step": 15610
    },
    {
      "epoch": 20.0020592020592,
      "grad_norm": 0.03767959401011467,
      "learning_rate": 6.6437866437866435e-06,
      "loss": 0.9582,
      "step": 15620
    },
    {
      "epoch": 20.0023166023166,
      "grad_norm": 0.02251305803656578,
      "learning_rate": 6.640926640926642e-06,
      "loss": 0.0018,
      "step": 15630
    },
    {
      "epoch": 20.002574002574004,
      "grad_norm": 0.14382117986679077,
      "learning_rate": 6.638066638066638e-06,
      "loss": 0.4451,
      "step": 15640
    },
    {
      "epoch": 20.002831402831404,
      "grad_norm": 0.0074038137681782246,
      "learning_rate": 6.6352066352066354e-06,
      "loss": 0.0175,
      "step": 15650
    },
    {
      "epoch": 20.003088803088804,
      "grad_norm": 309.4783630371094,
      "learning_rate": 6.632346632346633e-06,
      "loss": 1.088,
      "step": 15660
    },
    {
      "epoch": 20.003346203346204,
      "grad_norm": 0.0028939915355294943,
      "learning_rate": 6.62948662948663e-06,
      "loss": 0.7427,
      "step": 15670
    },
    {
      "epoch": 20.003603603603604,
      "grad_norm": 0.01108844019472599,
      "learning_rate": 6.626626626626627e-06,
      "loss": 0.0041,
      "step": 15680
    },
    {
      "epoch": 20.003861003861005,
      "grad_norm": 0.23344767093658447,
      "learning_rate": 6.623766623766624e-06,
      "loss": 0.6241,
      "step": 15690
    },
    {
      "epoch": 20.004118404118405,
      "grad_norm": 382.5850830078125,
      "learning_rate": 6.620906620906622e-06,
      "loss": 0.5393,
      "step": 15700
    },
    {
      "epoch": 20.004375804375805,
      "grad_norm": 0.17710717022418976,
      "learning_rate": 6.6180466180466185e-06,
      "loss": 0.0036,
      "step": 15710
    },
    {
      "epoch": 20.004633204633205,
      "grad_norm": 0.8400039076805115,
      "learning_rate": 6.615186615186616e-06,
      "loss": 0.5744,
      "step": 15720
    },
    {
      "epoch": 20.004890604890605,
      "grad_norm": 0.10408028960227966,
      "learning_rate": 6.612326612326613e-06,
      "loss": 0.1466,
      "step": 15730
    },
    {
      "epoch": 20.005148005148005,
      "grad_norm": 0.0018540999153628945,
      "learning_rate": 6.6094666094666104e-06,
      "loss": 0.4497,
      "step": 15740
    },
    {
      "epoch": 20.005405405405405,
      "grad_norm": 0.004669906571507454,
      "learning_rate": 6.606606606606607e-06,
      "loss": 0.443,
      "step": 15750
    },
    {
      "epoch": 20.005662805662805,
      "grad_norm": 91.52081298828125,
      "learning_rate": 6.603746603746605e-06,
      "loss": 2.0061,
      "step": 15760
    },
    {
      "epoch": 20.005920205920205,
      "grad_norm": 0.014538582414388657,
      "learning_rate": 6.6008866008866015e-06,
      "loss": 0.6473,
      "step": 15770
    },
    {
      "epoch": 20.006177606177605,
      "grad_norm": 0.09253435581922531,
      "learning_rate": 6.598026598026599e-06,
      "loss": 0.563,
      "step": 15780
    },
    {
      "epoch": 20.006435006435005,
      "grad_norm": 0.005112254060804844,
      "learning_rate": 6.595166595166595e-06,
      "loss": 0.0031,
      "step": 15790
    },
    {
      "epoch": 20.006692406692405,
      "grad_norm": 0.01782931201159954,
      "learning_rate": 6.5923065923065935e-06,
      "loss": 0.0068,
      "step": 15800
    },
    {
      "epoch": 20.006949806949805,
      "grad_norm": 3.834109306335449,
      "learning_rate": 6.58944658944659e-06,
      "loss": 0.6368,
      "step": 15810
    },
    {
      "epoch": 20.00720720720721,
      "grad_norm": 0.3963877856731415,
      "learning_rate": 6.586586586586587e-06,
      "loss": 0.5123,
      "step": 15820
    },
    {
      "epoch": 20.00746460746461,
      "grad_norm": 0.0013604898704215884,
      "learning_rate": 6.583726583726584e-06,
      "loss": 0.7435,
      "step": 15830
    },
    {
      "epoch": 20.00772200772201,
      "grad_norm": 0.37704506516456604,
      "learning_rate": 6.580866580866582e-06,
      "loss": 0.0023,
      "step": 15840
    },
    {
      "epoch": 20.00797940797941,
      "grad_norm": 26.52029037475586,
      "learning_rate": 6.578006578006578e-06,
      "loss": 0.1471,
      "step": 15850
    },
    {
      "epoch": 20.00823680823681,
      "grad_norm": 0.026596808806061745,
      "learning_rate": 6.575146575146576e-06,
      "loss": 0.0049,
      "step": 15860
    },
    {
      "epoch": 20.00849420849421,
      "grad_norm": 0.055981434881687164,
      "learning_rate": 6.572286572286572e-06,
      "loss": 0.5553,
      "step": 15870
    },
    {
      "epoch": 20.00875160875161,
      "grad_norm": 0.00963540282100439,
      "learning_rate": 6.56942656942657e-06,
      "loss": 0.0018,
      "step": 15880
    },
    {
      "epoch": 20.00900900900901,
      "grad_norm": 0.06317219883203506,
      "learning_rate": 6.566566566566567e-06,
      "loss": 0.4827,
      "step": 15890
    },
    {
      "epoch": 20.00926640926641,
      "grad_norm": 0.05564889311790466,
      "learning_rate": 6.563706563706564e-06,
      "loss": 1.7282,
      "step": 15900
    },
    {
      "epoch": 20.00952380952381,
      "grad_norm": 58.55592346191406,
      "learning_rate": 6.560846560846561e-06,
      "loss": 0.786,
      "step": 15910
    },
    {
      "epoch": 20.00978120978121,
      "grad_norm": 0.07347892969846725,
      "learning_rate": 6.557986557986559e-06,
      "loss": 0.3896,
      "step": 15920
    },
    {
      "epoch": 20.01003861003861,
      "grad_norm": 0.11479922384023666,
      "learning_rate": 6.555126555126555e-06,
      "loss": 1.1231,
      "step": 15930
    },
    {
      "epoch": 20.01029601029601,
      "grad_norm": 0.0013371037784963846,
      "learning_rate": 6.552266552266553e-06,
      "loss": 1.6537,
      "step": 15940
    },
    {
      "epoch": 20.01055341055341,
      "grad_norm": 0.024909818544983864,
      "learning_rate": 6.54940654940655e-06,
      "loss": 0.5062,
      "step": 15950
    },
    {
      "epoch": 20.01081081081081,
      "grad_norm": 0.26621049642562866,
      "learning_rate": 6.546546546546547e-06,
      "loss": 1.1278,
      "step": 15960
    },
    {
      "epoch": 20.01106821106821,
      "grad_norm": 0.27428120374679565,
      "learning_rate": 6.543686543686544e-06,
      "loss": 0.0021,
      "step": 15970
    },
    {
      "epoch": 20.01132561132561,
      "grad_norm": 0.19413171708583832,
      "learning_rate": 6.540826540826542e-06,
      "loss": 1.3931,
      "step": 15980
    },
    {
      "epoch": 20.01158301158301,
      "grad_norm": 0.9698719382286072,
      "learning_rate": 6.537966537966538e-06,
      "loss": 0.0043,
      "step": 15990
    },
    {
      "epoch": 20.01184041184041,
      "grad_norm": 0.0036435227375477552,
      "learning_rate": 6.5351065351065356e-06,
      "loss": 0.0025,
      "step": 16000
    },
    {
      "epoch": 20.012097812097814,
      "grad_norm": 0.005679155234247446,
      "learning_rate": 6.532246532246532e-06,
      "loss": 0.3591,
      "step": 16010
    },
    {
      "epoch": 20.012355212355214,
      "grad_norm": 0.2528136372566223,
      "learning_rate": 6.52938652938653e-06,
      "loss": 1.0381,
      "step": 16020
    },
    {
      "epoch": 20.012612612612614,
      "grad_norm": 0.008972964249551296,
      "learning_rate": 6.526526526526527e-06,
      "loss": 0.5617,
      "step": 16030
    },
    {
      "epoch": 20.012870012870014,
      "grad_norm": 0.437963604927063,
      "learning_rate": 6.523666523666524e-06,
      "loss": 1.372,
      "step": 16040
    },
    {
      "epoch": 20.013127413127414,
      "grad_norm": 0.0023576742969453335,
      "learning_rate": 6.520806520806522e-06,
      "loss": 0.7603,
      "step": 16050
    },
    {
      "epoch": 20.013384813384814,
      "grad_norm": 0.002086960943415761,
      "learning_rate": 6.517946517946519e-06,
      "loss": 1.0576,
      "step": 16060
    },
    {
      "epoch": 20.013642213642214,
      "grad_norm": 0.03679662570357323,
      "learning_rate": 6.515086515086516e-06,
      "loss": 0.6061,
      "step": 16070
    },
    {
      "epoch": 20.013899613899614,
      "grad_norm": 0.0061900317668914795,
      "learning_rate": 6.512226512226512e-06,
      "loss": 0.9334,
      "step": 16080
    },
    {
      "epoch": 20.014157014157014,
      "grad_norm": 0.041533537209033966,
      "learning_rate": 6.5093665093665106e-06,
      "loss": 1.0999,
      "step": 16090
    },
    {
      "epoch": 20.014414414414414,
      "grad_norm": 0.003179677063599229,
      "learning_rate": 6.506506506506507e-06,
      "loss": 0.9324,
      "step": 16100
    },
    {
      "epoch": 20.014671814671814,
      "grad_norm": 51.39768600463867,
      "learning_rate": 6.503646503646504e-06,
      "loss": 1.2931,
      "step": 16110
    },
    {
      "epoch": 20.014929214929214,
      "grad_norm": 31.660776138305664,
      "learning_rate": 6.500786500786501e-06,
      "loss": 0.9663,
      "step": 16120
    },
    {
      "epoch": 20.015186615186614,
      "grad_norm": 27.631328582763672,
      "learning_rate": 6.497926497926499e-06,
      "loss": 0.5043,
      "step": 16130
    },
    {
      "epoch": 20.015444015444015,
      "grad_norm": 0.10719513893127441,
      "learning_rate": 6.4950664950664955e-06,
      "loss": 1.2169,
      "step": 16140
    },
    {
      "epoch": 20.015701415701415,
      "grad_norm": 0.5144920349121094,
      "learning_rate": 6.492206492206493e-06,
      "loss": 1.0045,
      "step": 16150
    },
    {
      "epoch": 20.015958815958815,
      "grad_norm": 124.62886810302734,
      "learning_rate": 6.48934648934649e-06,
      "loss": 1.8065,
      "step": 16160
    },
    {
      "epoch": 20.016216216216215,
      "grad_norm": 47.27402114868164,
      "learning_rate": 6.486486486486487e-06,
      "loss": 0.3963,
      "step": 16170
    },
    {
      "epoch": 20.016473616473615,
      "grad_norm": 0.2665756344795227,
      "learning_rate": 6.483626483626484e-06,
      "loss": 0.7758,
      "step": 16180
    },
    {
      "epoch": 20.01673101673102,
      "grad_norm": 31.503211975097656,
      "learning_rate": 6.480766480766482e-06,
      "loss": 1.632,
      "step": 16190
    },
    {
      "epoch": 20.01698841698842,
      "grad_norm": 0.003470065537840128,
      "learning_rate": 6.4779064779064785e-06,
      "loss": 0.9137,
      "step": 16200
    },
    {
      "epoch": 20.01724581724582,
      "grad_norm": 0.24733567237854004,
      "learning_rate": 6.475046475046476e-06,
      "loss": 0.0986,
      "step": 16210
    },
    {
      "epoch": 20.01750321750322,
      "grad_norm": 2.922898292541504,
      "learning_rate": 6.472186472186472e-06,
      "loss": 0.4414,
      "step": 16220
    },
    {
      "epoch": 20.01776061776062,
      "grad_norm": 0.6893411874771118,
      "learning_rate": 6.4693264693264705e-06,
      "loss": 0.0069,
      "step": 16230
    },
    {
      "epoch": 20.01801801801802,
      "grad_norm": 0.3383210599422455,
      "learning_rate": 6.466466466466467e-06,
      "loss": 0.4328,
      "step": 16240
    },
    {
      "epoch": 20.01827541827542,
      "grad_norm": 368.6217346191406,
      "learning_rate": 6.463606463606464e-06,
      "loss": 0.9819,
      "step": 16250
    },
    {
      "epoch": 20.01853281853282,
      "grad_norm": 0.19034291803836823,
      "learning_rate": 6.460746460746461e-06,
      "loss": 0.9037,
      "step": 16260
    },
    {
      "epoch": 20.01879021879022,
      "grad_norm": 0.08844771236181259,
      "learning_rate": 6.457886457886459e-06,
      "loss": 1.7082,
      "step": 16270
    },
    {
      "epoch": 20.01904761904762,
      "grad_norm": 0.32688507437705994,
      "learning_rate": 6.455026455026455e-06,
      "loss": 0.851,
      "step": 16280
    },
    {
      "epoch": 20.01930501930502,
      "grad_norm": 1.21635901927948,
      "learning_rate": 6.452166452166453e-06,
      "loss": 0.7596,
      "step": 16290
    },
    {
      "epoch": 20.01956241956242,
      "grad_norm": 0.3342961370944977,
      "learning_rate": 6.449306449306449e-06,
      "loss": 0.0038,
      "step": 16300
    },
    {
      "epoch": 20.01981981981982,
      "grad_norm": 0.06980375945568085,
      "learning_rate": 6.446446446446447e-06,
      "loss": 1.0169,
      "step": 16310
    },
    {
      "epoch": 20.02,
      "eval_accuracy": 0.8695652173913043,
      "eval_loss": 0.6228641271591187,
      "eval_runtime": 13.4391,
      "eval_samples_per_second": 3.423,
      "eval_steps_per_second": 3.423,
      "step": 16317
    },
    {
      "epoch": 21.00007722007722,
      "grad_norm": 0.05950063839554787,
      "learning_rate": 6.443586443586444e-06,
      "loss": 0.7334,
      "step": 16320
    },
    {
      "epoch": 21.00033462033462,
      "grad_norm": 0.0011386877158656716,
      "learning_rate": 6.440726440726441e-06,
      "loss": 0.6521,
      "step": 16330
    },
    {
      "epoch": 21.00059202059202,
      "grad_norm": 0.11157349497079849,
      "learning_rate": 6.4378664378664375e-06,
      "loss": 0.7863,
      "step": 16340
    },
    {
      "epoch": 21.00084942084942,
      "grad_norm": 0.13690587878227234,
      "learning_rate": 6.435006435006436e-06,
      "loss": 0.0157,
      "step": 16350
    },
    {
      "epoch": 21.00110682110682,
      "grad_norm": 0.1480686217546463,
      "learning_rate": 6.432146432146432e-06,
      "loss": 0.4204,
      "step": 16360
    },
    {
      "epoch": 21.00136422136422,
      "grad_norm": 0.0624036081135273,
      "learning_rate": 6.4292864292864295e-06,
      "loss": 1.0578,
      "step": 16370
    },
    {
      "epoch": 21.00162162162162,
      "grad_norm": 0.44504156708717346,
      "learning_rate": 6.426426426426427e-06,
      "loss": 0.3797,
      "step": 16380
    },
    {
      "epoch": 21.00187902187902,
      "grad_norm": 0.14833220839500427,
      "learning_rate": 6.423566423566424e-06,
      "loss": 0.5096,
      "step": 16390
    },
    {
      "epoch": 21.002136422136424,
      "grad_norm": 0.07981384545564651,
      "learning_rate": 6.420706420706421e-06,
      "loss": 0.4623,
      "step": 16400
    },
    {
      "epoch": 21.002393822393824,
      "grad_norm": 132.79852294921875,
      "learning_rate": 6.417846417846419e-06,
      "loss": 0.9159,
      "step": 16410
    },
    {
      "epoch": 21.002651222651224,
      "grad_norm": 0.0031174507457762957,
      "learning_rate": 6.414986414986416e-06,
      "loss": 1.6705,
      "step": 16420
    },
    {
      "epoch": 21.002908622908624,
      "grad_norm": 0.33982014656066895,
      "learning_rate": 6.4121264121264125e-06,
      "loss": 0.0024,
      "step": 16430
    },
    {
      "epoch": 21.003166023166024,
      "grad_norm": 0.06785816699266434,
      "learning_rate": 6.409266409266411e-06,
      "loss": 0.0103,
      "step": 16440
    },
    {
      "epoch": 21.003423423423424,
      "grad_norm": 0.0009388201870024204,
      "learning_rate": 6.406406406406407e-06,
      "loss": 0.604,
      "step": 16450
    },
    {
      "epoch": 21.003680823680824,
      "grad_norm": 0.3058651387691498,
      "learning_rate": 6.4035464035464045e-06,
      "loss": 0.7045,
      "step": 16460
    },
    {
      "epoch": 21.003938223938224,
      "grad_norm": 0.19235381484031677,
      "learning_rate": 6.400686400686401e-06,
      "loss": 1.2,
      "step": 16470
    },
    {
      "epoch": 21.004195624195624,
      "grad_norm": 0.18182158470153809,
      "learning_rate": 6.397826397826399e-06,
      "loss": 0.5073,
      "step": 16480
    },
    {
      "epoch": 21.004453024453024,
      "grad_norm": 0.0032803218346089125,
      "learning_rate": 6.394966394966396e-06,
      "loss": 1.8055,
      "step": 16490
    },
    {
      "epoch": 21.004710424710424,
      "grad_norm": 374.3335266113281,
      "learning_rate": 6.392106392106393e-06,
      "loss": 0.0291,
      "step": 16500
    },
    {
      "epoch": 21.004967824967824,
      "grad_norm": 0.002655749674886465,
      "learning_rate": 6.389246389246389e-06,
      "loss": 0.7674,
      "step": 16510
    },
    {
      "epoch": 21.005225225225225,
      "grad_norm": 0.035156842321157455,
      "learning_rate": 6.3863863863863875e-06,
      "loss": 0.9229,
      "step": 16520
    },
    {
      "epoch": 21.005482625482625,
      "grad_norm": 0.012515634298324585,
      "learning_rate": 6.383526383526384e-06,
      "loss": 0.5184,
      "step": 16530
    },
    {
      "epoch": 21.005740025740025,
      "grad_norm": 0.226984903216362,
      "learning_rate": 6.380666380666381e-06,
      "loss": 0.4388,
      "step": 16540
    },
    {
      "epoch": 21.005997425997425,
      "grad_norm": 0.13279147446155548,
      "learning_rate": 6.377806377806378e-06,
      "loss": 0.9475,
      "step": 16550
    },
    {
      "epoch": 21.006254826254825,
      "grad_norm": 0.13161540031433105,
      "learning_rate": 6.374946374946376e-06,
      "loss": 0.5151,
      "step": 16560
    },
    {
      "epoch": 21.006512226512225,
      "grad_norm": 0.018993711099028587,
      "learning_rate": 6.3720863720863724e-06,
      "loss": 0.4802,
      "step": 16570
    },
    {
      "epoch": 21.00676962676963,
      "grad_norm": 0.002260217210277915,
      "learning_rate": 6.36922636922637e-06,
      "loss": 0.5331,
      "step": 16580
    },
    {
      "epoch": 21.00702702702703,
      "grad_norm": 0.002179916715249419,
      "learning_rate": 6.366366366366366e-06,
      "loss": 1.0489,
      "step": 16590
    },
    {
      "epoch": 21.00728442728443,
      "grad_norm": 0.41448646783828735,
      "learning_rate": 6.363506363506364e-06,
      "loss": 0.9842,
      "step": 16600
    },
    {
      "epoch": 21.00754182754183,
      "grad_norm": 2809.028076171875,
      "learning_rate": 6.360646360646361e-06,
      "loss": 0.4465,
      "step": 16610
    },
    {
      "epoch": 21.00779922779923,
      "grad_norm": 3.043079376220703,
      "learning_rate": 6.357786357786358e-06,
      "loss": 1.0932,
      "step": 16620
    },
    {
      "epoch": 21.00805662805663,
      "grad_norm": 0.0025238539092242718,
      "learning_rate": 6.3549263549263555e-06,
      "loss": 0.7486,
      "step": 16630
    },
    {
      "epoch": 21.00831402831403,
      "grad_norm": 0.0024203872308135033,
      "learning_rate": 6.352066352066353e-06,
      "loss": 1.4116,
      "step": 16640
    },
    {
      "epoch": 21.00857142857143,
      "grad_norm": 0.18933537602424622,
      "learning_rate": 6.349206349206349e-06,
      "loss": 1.9212,
      "step": 16650
    },
    {
      "epoch": 21.00882882882883,
      "grad_norm": 0.00907233264297247,
      "learning_rate": 6.3463463463463474e-06,
      "loss": 1.2356,
      "step": 16660
    },
    {
      "epoch": 21.00908622908623,
      "grad_norm": 0.43250593543052673,
      "learning_rate": 6.343486343486344e-06,
      "loss": 0.0086,
      "step": 16670
    },
    {
      "epoch": 21.00934362934363,
      "grad_norm": 0.23632937669754028,
      "learning_rate": 6.340626340626341e-06,
      "loss": 0.2125,
      "step": 16680
    },
    {
      "epoch": 21.00960102960103,
      "grad_norm": 0.4456353187561035,
      "learning_rate": 6.337766337766338e-06,
      "loss": 0.5689,
      "step": 16690
    },
    {
      "epoch": 21.00985842985843,
      "grad_norm": 0.003951109945774078,
      "learning_rate": 6.334906334906336e-06,
      "loss": 1.4499,
      "step": 16700
    },
    {
      "epoch": 21.01011583011583,
      "grad_norm": 0.005860066507011652,
      "learning_rate": 6.332046332046332e-06,
      "loss": 0.6452,
      "step": 16710
    },
    {
      "epoch": 21.01037323037323,
      "grad_norm": 0.677861213684082,
      "learning_rate": 6.32918632918633e-06,
      "loss": 0.0304,
      "step": 16720
    },
    {
      "epoch": 21.01063063063063,
      "grad_norm": 0.22151070833206177,
      "learning_rate": 6.326326326326326e-06,
      "loss": 0.3219,
      "step": 16730
    },
    {
      "epoch": 21.01088803088803,
      "grad_norm": 0.002829265082255006,
      "learning_rate": 6.323466323466324e-06,
      "loss": 1.018,
      "step": 16740
    },
    {
      "epoch": 21.01114543114543,
      "grad_norm": 0.0027092285454273224,
      "learning_rate": 6.320606320606321e-06,
      "loss": 0.5489,
      "step": 16750
    },
    {
      "epoch": 21.011402831402833,
      "grad_norm": 0.011305739171802998,
      "learning_rate": 6.317746317746318e-06,
      "loss": 0.5165,
      "step": 16760
    },
    {
      "epoch": 21.011660231660233,
      "grad_norm": 2.433239221572876,
      "learning_rate": 6.3148863148863145e-06,
      "loss": 0.0064,
      "step": 16770
    },
    {
      "epoch": 21.011917631917633,
      "grad_norm": 107.0201416015625,
      "learning_rate": 6.312026312026313e-06,
      "loss": 1.519,
      "step": 16780
    },
    {
      "epoch": 21.012175032175033,
      "grad_norm": 0.11884047091007233,
      "learning_rate": 6.30916630916631e-06,
      "loss": 1.4826,
      "step": 16790
    },
    {
      "epoch": 21.012432432432433,
      "grad_norm": 1.756311297416687,
      "learning_rate": 6.3063063063063065e-06,
      "loss": 1.1079,
      "step": 16800
    },
    {
      "epoch": 21.012689832689833,
      "grad_norm": 0.4935521185398102,
      "learning_rate": 6.303446303446305e-06,
      "loss": 0.5533,
      "step": 16810
    },
    {
      "epoch": 21.012947232947234,
      "grad_norm": 0.14828822016716003,
      "learning_rate": 6.300586300586301e-06,
      "loss": 1.0428,
      "step": 16820
    },
    {
      "epoch": 21.013204633204634,
      "grad_norm": 0.10324635356664658,
      "learning_rate": 6.297726297726298e-06,
      "loss": 0.0046,
      "step": 16830
    },
    {
      "epoch": 21.013462033462034,
      "grad_norm": 0.1749248206615448,
      "learning_rate": 6.294866294866295e-06,
      "loss": 0.877,
      "step": 16840
    },
    {
      "epoch": 21.013719433719434,
      "grad_norm": 0.0036808913573622704,
      "learning_rate": 6.292006292006293e-06,
      "loss": 0.5626,
      "step": 16850
    },
    {
      "epoch": 21.013976833976834,
      "grad_norm": 0.007130262907594442,
      "learning_rate": 6.2891462891462895e-06,
      "loss": 0.1482,
      "step": 16860
    },
    {
      "epoch": 21.014234234234234,
      "grad_norm": 0.02497122436761856,
      "learning_rate": 6.286286286286287e-06,
      "loss": 0.4806,
      "step": 16870
    },
    {
      "epoch": 21.014491634491634,
      "grad_norm": 346.6941833496094,
      "learning_rate": 6.283426283426284e-06,
      "loss": 1.3267,
      "step": 16880
    },
    {
      "epoch": 21.014749034749034,
      "grad_norm": 0.004221619106829166,
      "learning_rate": 6.2805662805662815e-06,
      "loss": 1.259,
      "step": 16890
    },
    {
      "epoch": 21.015006435006434,
      "grad_norm": 0.0013160128146409988,
      "learning_rate": 6.277706277706278e-06,
      "loss": 0.715,
      "step": 16900
    },
    {
      "epoch": 21.015263835263834,
      "grad_norm": 0.0038280917797237635,
      "learning_rate": 6.274846274846276e-06,
      "loss": 0.5407,
      "step": 16910
    },
    {
      "epoch": 21.015521235521234,
      "grad_norm": 35.85282897949219,
      "learning_rate": 6.2719862719862726e-06,
      "loss": 0.957,
      "step": 16920
    },
    {
      "epoch": 21.015778635778634,
      "grad_norm": 0.06263106316328049,
      "learning_rate": 6.26912626912627e-06,
      "loss": 0.9689,
      "step": 16930
    },
    {
      "epoch": 21.016036036036034,
      "grad_norm": 0.09454566240310669,
      "learning_rate": 6.266266266266266e-06,
      "loss": 0.5719,
      "step": 16940
    },
    {
      "epoch": 21.016293436293438,
      "grad_norm": 60.81951904296875,
      "learning_rate": 6.2634062634062645e-06,
      "loss": 1.0382,
      "step": 16950
    },
    {
      "epoch": 21.016550836550838,
      "grad_norm": 29.10282325744629,
      "learning_rate": 6.260546260546261e-06,
      "loss": 1.8424,
      "step": 16960
    },
    {
      "epoch": 21.016808236808238,
      "grad_norm": 0.004924100823700428,
      "learning_rate": 6.257686257686258e-06,
      "loss": 0.0054,
      "step": 16970
    },
    {
      "epoch": 21.017065637065638,
      "grad_norm": 0.08713381737470627,
      "learning_rate": 6.254826254826255e-06,
      "loss": 0.0118,
      "step": 16980
    },
    {
      "epoch": 21.017323037323038,
      "grad_norm": 0.0026128473691642284,
      "learning_rate": 6.251966251966253e-06,
      "loss": 0.0031,
      "step": 16990
    },
    {
      "epoch": 21.01758043758044,
      "grad_norm": 0.0017107778694480658,
      "learning_rate": 6.249106249106249e-06,
      "loss": 1.7852,
      "step": 17000
    },
    {
      "epoch": 21.01783783783784,
      "grad_norm": 57.05978012084961,
      "learning_rate": 6.246246246246247e-06,
      "loss": 1.4597,
      "step": 17010
    },
    {
      "epoch": 21.01809523809524,
      "grad_norm": 6.505259037017822,
      "learning_rate": 6.243386243386243e-06,
      "loss": 0.449,
      "step": 17020
    },
    {
      "epoch": 21.01835263835264,
      "grad_norm": 0.7023016214370728,
      "learning_rate": 6.240526240526241e-06,
      "loss": 1.3402,
      "step": 17030
    },
    {
      "epoch": 21.01861003861004,
      "grad_norm": 0.07186837494373322,
      "learning_rate": 6.237666237666238e-06,
      "loss": 0.6516,
      "step": 17040
    },
    {
      "epoch": 21.01886743886744,
      "grad_norm": 0.006106872111558914,
      "learning_rate": 6.234806234806235e-06,
      "loss": 0.0604,
      "step": 17050
    },
    {
      "epoch": 21.01912483912484,
      "grad_norm": 0.001679259818047285,
      "learning_rate": 6.2319462319462325e-06,
      "loss": 0.4562,
      "step": 17060
    },
    {
      "epoch": 21.01938223938224,
      "grad_norm": 0.20496806502342224,
      "learning_rate": 6.22908622908623e-06,
      "loss": 1.5803,
      "step": 17070
    },
    {
      "epoch": 21.01963963963964,
      "grad_norm": 0.002377886790782213,
      "learning_rate": 6.226226226226226e-06,
      "loss": 0.4701,
      "step": 17080
    },
    {
      "epoch": 21.01989703989704,
      "grad_norm": 15.251885414123535,
      "learning_rate": 6.2233662233662236e-06,
      "loss": 0.0059,
      "step": 17090
    },
    {
      "epoch": 21.02,
      "eval_accuracy": 0.8478260869565217,
      "eval_loss": 0.7053625583648682,
      "eval_runtime": 13.4557,
      "eval_samples_per_second": 3.419,
      "eval_steps_per_second": 3.419,
      "step": 17094
    },
    {
      "epoch": 22.00015444015444,
      "grad_norm": 0.002505769720301032,
      "learning_rate": 6.220506220506221e-06,
      "loss": 0.5072,
      "step": 17100
    },
    {
      "epoch": 22.00041184041184,
      "grad_norm": 0.08272158354520798,
      "learning_rate": 6.217646217646218e-06,
      "loss": 0.5502,
      "step": 17110
    },
    {
      "epoch": 22.00066924066924,
      "grad_norm": 0.002082453342154622,
      "learning_rate": 6.214786214786215e-06,
      "loss": 1.0242,
      "step": 17120
    },
    {
      "epoch": 22.00092664092664,
      "grad_norm": 0.18459756672382355,
      "learning_rate": 6.211926211926213e-06,
      "loss": 0.4925,
      "step": 17130
    },
    {
      "epoch": 22.00118404118404,
      "grad_norm": 0.00420813262462616,
      "learning_rate": 6.209066209066209e-06,
      "loss": 0.8276,
      "step": 17140
    },
    {
      "epoch": 22.001441441441443,
      "grad_norm": 0.21216662228107452,
      "learning_rate": 6.206206206206207e-06,
      "loss": 0.2827,
      "step": 17150
    },
    {
      "epoch": 22.001698841698843,
      "grad_norm": 59.495052337646484,
      "learning_rate": 6.203346203346203e-06,
      "loss": 0.1756,
      "step": 17160
    },
    {
      "epoch": 22.001956241956243,
      "grad_norm": 0.07312590628862381,
      "learning_rate": 6.200486200486201e-06,
      "loss": 0.0012,
      "step": 17170
    },
    {
      "epoch": 22.002213642213643,
      "grad_norm": 0.2232220619916916,
      "learning_rate": 6.1976261976261986e-06,
      "loss": 0.5605,
      "step": 17180
    },
    {
      "epoch": 22.002471042471043,
      "grad_norm": 0.00914843287318945,
      "learning_rate": 6.194766194766195e-06,
      "loss": 0.5424,
      "step": 17190
    },
    {
      "epoch": 22.002728442728444,
      "grad_norm": 0.004124210216104984,
      "learning_rate": 6.191906191906193e-06,
      "loss": 0.013,
      "step": 17200
    },
    {
      "epoch": 22.002985842985844,
      "grad_norm": 133.94290161132812,
      "learning_rate": 6.18904618904619e-06,
      "loss": 1.0607,
      "step": 17210
    },
    {
      "epoch": 22.003243243243244,
      "grad_norm": 15.545845031738281,
      "learning_rate": 6.186186186186187e-06,
      "loss": 0.5845,
      "step": 17220
    },
    {
      "epoch": 22.003500643500644,
      "grad_norm": 0.2141716480255127,
      "learning_rate": 6.1833261833261834e-06,
      "loss": 1.2612,
      "step": 17230
    },
    {
      "epoch": 22.003758043758044,
      "grad_norm": 0.11907975375652313,
      "learning_rate": 6.180466180466182e-06,
      "loss": 0.5358,
      "step": 17240
    },
    {
      "epoch": 22.004015444015444,
      "grad_norm": 0.1688116192817688,
      "learning_rate": 6.177606177606178e-06,
      "loss": 0.0027,
      "step": 17250
    },
    {
      "epoch": 22.004272844272844,
      "grad_norm": 29.69304847717285,
      "learning_rate": 6.174746174746175e-06,
      "loss": 1.9443,
      "step": 17260
    },
    {
      "epoch": 22.004530244530244,
      "grad_norm": 0.001488303067162633,
      "learning_rate": 6.171886171886172e-06,
      "loss": 0.4166,
      "step": 17270
    },
    {
      "epoch": 22.004787644787644,
      "grad_norm": 0.003606772283092141,
      "learning_rate": 6.16902616902617e-06,
      "loss": 0.0036,
      "step": 17280
    },
    {
      "epoch": 22.005045045045044,
      "grad_norm": 0.23906883597373962,
      "learning_rate": 6.1661661661661665e-06,
      "loss": 0.8975,
      "step": 17290
    },
    {
      "epoch": 22.005302445302444,
      "grad_norm": 0.0013485512463375926,
      "learning_rate": 6.163306163306164e-06,
      "loss": 0.4444,
      "step": 17300
    },
    {
      "epoch": 22.005559845559844,
      "grad_norm": 0.0031310829799622297,
      "learning_rate": 6.160446160446161e-06,
      "loss": 0.0087,
      "step": 17310
    },
    {
      "epoch": 22.005817245817244,
      "grad_norm": 0.0013559467624872923,
      "learning_rate": 6.1575861575861584e-06,
      "loss": 1.025,
      "step": 17320
    },
    {
      "epoch": 22.006074646074644,
      "grad_norm": 0.11612682044506073,
      "learning_rate": 6.154726154726155e-06,
      "loss": 0.5893,
      "step": 17330
    },
    {
      "epoch": 22.006332046332048,
      "grad_norm": 0.001688244054093957,
      "learning_rate": 6.151866151866153e-06,
      "loss": 0.0039,
      "step": 17340
    },
    {
      "epoch": 22.006589446589448,
      "grad_norm": 0.017765646800398827,
      "learning_rate": 6.1490061490061495e-06,
      "loss": 1.5407,
      "step": 17350
    },
    {
      "epoch": 22.006846846846848,
      "grad_norm": 0.09817246347665787,
      "learning_rate": 6.146146146146147e-06,
      "loss": 0.0038,
      "step": 17360
    },
    {
      "epoch": 22.007104247104248,
      "grad_norm": 0.2186000794172287,
      "learning_rate": 6.143286143286143e-06,
      "loss": 0.4914,
      "step": 17370
    },
    {
      "epoch": 22.00736164736165,
      "grad_norm": 0.0013634796487167478,
      "learning_rate": 6.1404261404261415e-06,
      "loss": 0.0024,
      "step": 17380
    },
    {
      "epoch": 22.00761904761905,
      "grad_norm": 0.12472115457057953,
      "learning_rate": 6.137566137566138e-06,
      "loss": 0.615,
      "step": 17390
    },
    {
      "epoch": 22.00787644787645,
      "grad_norm": 40.63284683227539,
      "learning_rate": 6.134706134706135e-06,
      "loss": 1.4173,
      "step": 17400
    },
    {
      "epoch": 22.00813384813385,
      "grad_norm": 249.48081970214844,
      "learning_rate": 6.131846131846132e-06,
      "loss": 0.4317,
      "step": 17410
    },
    {
      "epoch": 22.00839124839125,
      "grad_norm": 0.1064755842089653,
      "learning_rate": 6.12898612898613e-06,
      "loss": 0.9979,
      "step": 17420
    },
    {
      "epoch": 22.00864864864865,
      "grad_norm": 0.002664172323420644,
      "learning_rate": 6.126126126126126e-06,
      "loss": 1.0715,
      "step": 17430
    },
    {
      "epoch": 22.00890604890605,
      "grad_norm": 0.11099614202976227,
      "learning_rate": 6.123266123266124e-06,
      "loss": 1.1395,
      "step": 17440
    },
    {
      "epoch": 22.00916344916345,
      "grad_norm": 451.1719665527344,
      "learning_rate": 6.12040612040612e-06,
      "loss": 0.595,
      "step": 17450
    },
    {
      "epoch": 22.00942084942085,
      "grad_norm": 1.3345633745193481,
      "learning_rate": 6.117546117546118e-06,
      "loss": 0.4494,
      "step": 17460
    },
    {
      "epoch": 22.00967824967825,
      "grad_norm": 0.8723010420799255,
      "learning_rate": 6.114686114686115e-06,
      "loss": 0.0143,
      "step": 17470
    },
    {
      "epoch": 22.00993564993565,
      "grad_norm": 164.72413635253906,
      "learning_rate": 6.111826111826112e-06,
      "loss": 1.8803,
      "step": 17480
    },
    {
      "epoch": 22.01019305019305,
      "grad_norm": 0.0019611294846981764,
      "learning_rate": 6.108966108966109e-06,
      "loss": 0.0016,
      "step": 17490
    },
    {
      "epoch": 22.01045045045045,
      "grad_norm": 29.908557891845703,
      "learning_rate": 6.106106106106107e-06,
      "loss": 1.4703,
      "step": 17500
    },
    {
      "epoch": 22.01070785070785,
      "grad_norm": 0.449646532535553,
      "learning_rate": 6.103246103246103e-06,
      "loss": 0.6162,
      "step": 17510
    },
    {
      "epoch": 22.010965250965253,
      "grad_norm": 0.007094630040228367,
      "learning_rate": 6.1003861003861005e-06,
      "loss": 0.5787,
      "step": 17520
    },
    {
      "epoch": 22.011222651222653,
      "grad_norm": 0.0014509069733321667,
      "learning_rate": 6.097526097526098e-06,
      "loss": 0.0015,
      "step": 17530
    },
    {
      "epoch": 22.011480051480053,
      "grad_norm": 0.011663383804261684,
      "learning_rate": 6.094666094666095e-06,
      "loss": 0.5842,
      "step": 17540
    },
    {
      "epoch": 22.011737451737453,
      "grad_norm": 0.1782633364200592,
      "learning_rate": 6.0918060918060925e-06,
      "loss": 1.0637,
      "step": 17550
    },
    {
      "epoch": 22.011994851994853,
      "grad_norm": 0.013621443882584572,
      "learning_rate": 6.08894608894609e-06,
      "loss": 0.3769,
      "step": 17560
    },
    {
      "epoch": 22.012252252252253,
      "grad_norm": 0.08306556195020676,
      "learning_rate": 6.086086086086087e-06,
      "loss": 0.5647,
      "step": 17570
    },
    {
      "epoch": 22.012509652509653,
      "grad_norm": 200.47479248046875,
      "learning_rate": 6.083226083226084e-06,
      "loss": 1.4816,
      "step": 17580
    },
    {
      "epoch": 22.012767052767053,
      "grad_norm": 0.008521323092281818,
      "learning_rate": 6.080366080366082e-06,
      "loss": 0.9749,
      "step": 17590
    },
    {
      "epoch": 22.013024453024453,
      "grad_norm": 0.0009275858174078166,
      "learning_rate": 6.077506077506078e-06,
      "loss": 2.1086,
      "step": 17600
    },
    {
      "epoch": 22.013281853281853,
      "grad_norm": 0.10126996785402298,
      "learning_rate": 6.0746460746460755e-06,
      "loss": 0.4938,
      "step": 17610
    },
    {
      "epoch": 22.013539253539253,
      "grad_norm": 0.0797620639204979,
      "learning_rate": 6.071786071786072e-06,
      "loss": 1.5423,
      "step": 17620
    },
    {
      "epoch": 22.013796653796653,
      "grad_norm": 0.19221942126750946,
      "learning_rate": 6.06892606892607e-06,
      "loss": 0.0867,
      "step": 17630
    },
    {
      "epoch": 22.014054054054053,
      "grad_norm": 0.538609504699707,
      "learning_rate": 6.066066066066067e-06,
      "loss": 0.3392,
      "step": 17640
    },
    {
      "epoch": 22.014311454311454,
      "grad_norm": 0.6996312141418457,
      "learning_rate": 6.063206063206064e-06,
      "loss": 1.3827,
      "step": 17650
    },
    {
      "epoch": 22.014568854568854,
      "grad_norm": 0.003910961095243692,
      "learning_rate": 6.06034606034606e-06,
      "loss": 0.459,
      "step": 17660
    },
    {
      "epoch": 22.014826254826254,
      "grad_norm": 0.0005794151220470667,
      "learning_rate": 6.057486057486059e-06,
      "loss": 0.5004,
      "step": 17670
    },
    {
      "epoch": 22.015083655083654,
      "grad_norm": 0.0007814746350049973,
      "learning_rate": 6.054626054626055e-06,
      "loss": 0.007,
      "step": 17680
    },
    {
      "epoch": 22.015341055341054,
      "grad_norm": 0.00357211846858263,
      "learning_rate": 6.051766051766052e-06,
      "loss": 0.0053,
      "step": 17690
    },
    {
      "epoch": 22.015598455598454,
      "grad_norm": 516.8284912109375,
      "learning_rate": 6.048906048906049e-06,
      "loss": 1.0764,
      "step": 17700
    },
    {
      "epoch": 22.015855855855857,
      "grad_norm": 16.29941177368164,
      "learning_rate": 6.046046046046047e-06,
      "loss": 1.5809,
      "step": 17710
    },
    {
      "epoch": 22.016113256113258,
      "grad_norm": 0.0014146931935101748,
      "learning_rate": 6.0431860431860435e-06,
      "loss": 0.3413,
      "step": 17720
    },
    {
      "epoch": 22.016370656370658,
      "grad_norm": 376.0869140625,
      "learning_rate": 6.040326040326041e-06,
      "loss": 2.2279,
      "step": 17730
    },
    {
      "epoch": 22.016628056628058,
      "grad_norm": 0.005802272353321314,
      "learning_rate": 6.037466037466037e-06,
      "loss": 0.6852,
      "step": 17740
    },
    {
      "epoch": 22.016885456885458,
      "grad_norm": 0.07146825641393661,
      "learning_rate": 6.034606034606035e-06,
      "loss": 0.8904,
      "step": 17750
    },
    {
      "epoch": 22.017142857142858,
      "grad_norm": 0.005169512704014778,
      "learning_rate": 6.031746031746032e-06,
      "loss": 0.9166,
      "step": 17760
    },
    {
      "epoch": 22.017400257400258,
      "grad_norm": 0.0047473483718931675,
      "learning_rate": 6.028886028886029e-06,
      "loss": 2.7498,
      "step": 17770
    },
    {
      "epoch": 22.017657657657658,
      "grad_norm": 0.001801575068384409,
      "learning_rate": 6.0260260260260265e-06,
      "loss": 0.4033,
      "step": 17780
    },
    {
      "epoch": 22.017915057915058,
      "grad_norm": 0.19776633381843567,
      "learning_rate": 6.023166023166024e-06,
      "loss": 1.0597,
      "step": 17790
    },
    {
      "epoch": 22.018172458172458,
      "grad_norm": 0.6332172751426697,
      "learning_rate": 6.02030602030602e-06,
      "loss": 0.4597,
      "step": 17800
    },
    {
      "epoch": 22.018429858429858,
      "grad_norm": 33.11475372314453,
      "learning_rate": 6.0174460174460185e-06,
      "loss": 1.4584,
      "step": 17810
    },
    {
      "epoch": 22.018687258687258,
      "grad_norm": 0.4599050283432007,
      "learning_rate": 6.014586014586015e-06,
      "loss": 0.015,
      "step": 17820
    },
    {
      "epoch": 22.01894465894466,
      "grad_norm": 0.21705041825771332,
      "learning_rate": 6.011726011726012e-06,
      "loss": 0.5132,
      "step": 17830
    },
    {
      "epoch": 22.01920205920206,
      "grad_norm": 0.8084737062454224,
      "learning_rate": 6.008866008866009e-06,
      "loss": 1.9364,
      "step": 17840
    },
    {
      "epoch": 22.01945945945946,
      "grad_norm": 0.15941214561462402,
      "learning_rate": 6.006006006006007e-06,
      "loss": 0.4626,
      "step": 17850
    },
    {
      "epoch": 22.01971685971686,
      "grad_norm": 0.001215206808410585,
      "learning_rate": 6.003146003146003e-06,
      "loss": 1.1089,
      "step": 17860
    },
    {
      "epoch": 22.01997425997426,
      "grad_norm": 0.029706159606575966,
      "learning_rate": 6.000286000286001e-06,
      "loss": 0.8826,
      "step": 17870
    },
    {
      "epoch": 22.02,
      "eval_accuracy": 0.7391304347826086,
      "eval_loss": 1.0915201902389526,
      "eval_runtime": 13.4617,
      "eval_samples_per_second": 3.417,
      "eval_steps_per_second": 3.417,
      "step": 17871
    },
    {
      "epoch": 23.00023166023166,
      "grad_norm": 0.4556454122066498,
      "learning_rate": 5.997425997425997e-06,
      "loss": 0.3958,
      "step": 17880
    },
    {
      "epoch": 23.00048906048906,
      "grad_norm": 0.3868144452571869,
      "learning_rate": 5.994565994565995e-06,
      "loss": 0.5247,
      "step": 17890
    },
    {
      "epoch": 23.00074646074646,
      "grad_norm": 0.009736313484609127,
      "learning_rate": 5.991705991705992e-06,
      "loss": 0.4699,
      "step": 17900
    },
    {
      "epoch": 23.001003861003863,
      "grad_norm": 4.145432472229004,
      "learning_rate": 5.988845988845989e-06,
      "loss": 0.442,
      "step": 17910
    },
    {
      "epoch": 23.001261261261263,
      "grad_norm": 0.5545969605445862,
      "learning_rate": 5.985985985985987e-06,
      "loss": 0.004,
      "step": 17920
    },
    {
      "epoch": 23.001518661518663,
      "grad_norm": 0.016343528404831886,
      "learning_rate": 5.983125983125984e-06,
      "loss": 0.002,
      "step": 17930
    },
    {
      "epoch": 23.001776061776063,
      "grad_norm": 0.0030438120011240244,
      "learning_rate": 5.980265980265981e-06,
      "loss": 1.1311,
      "step": 17940
    },
    {
      "epoch": 23.002033462033463,
      "grad_norm": 0.13489748537540436,
      "learning_rate": 5.9774059774059775e-06,
      "loss": 0.5768,
      "step": 17950
    },
    {
      "epoch": 23.002290862290863,
      "grad_norm": 72.87088012695312,
      "learning_rate": 5.974545974545976e-06,
      "loss": 0.9573,
      "step": 17960
    },
    {
      "epoch": 23.002548262548263,
      "grad_norm": 0.05114838853478432,
      "learning_rate": 5.971685971685972e-06,
      "loss": 0.923,
      "step": 17970
    },
    {
      "epoch": 23.002805662805663,
      "grad_norm": 0.38993966579437256,
      "learning_rate": 5.9688259688259695e-06,
      "loss": 0.0118,
      "step": 17980
    },
    {
      "epoch": 23.003063063063063,
      "grad_norm": 0.0010464813094586134,
      "learning_rate": 5.965965965965966e-06,
      "loss": 0.4669,
      "step": 17990
    },
    {
      "epoch": 23.003320463320463,
      "grad_norm": 0.0008095205412246287,
      "learning_rate": 5.963105963105964e-06,
      "loss": 0.0063,
      "step": 18000
    },
    {
      "epoch": 23.003577863577863,
      "grad_norm": 0.002685692859813571,
      "learning_rate": 5.9602459602459606e-06,
      "loss": 0.002,
      "step": 18010
    },
    {
      "epoch": 23.003835263835263,
      "grad_norm": 0.0549185648560524,
      "learning_rate": 5.957385957385958e-06,
      "loss": 0.0077,
      "step": 18020
    },
    {
      "epoch": 23.004092664092664,
      "grad_norm": 0.0013326237676665187,
      "learning_rate": 5.954525954525955e-06,
      "loss": 0.0023,
      "step": 18030
    },
    {
      "epoch": 23.004350064350064,
      "grad_norm": 0.002016945742070675,
      "learning_rate": 5.9516659516659525e-06,
      "loss": 0.4834,
      "step": 18040
    },
    {
      "epoch": 23.004607464607464,
      "grad_norm": 0.00044487594277597964,
      "learning_rate": 5.948805948805949e-06,
      "loss": 1.4307,
      "step": 18050
    },
    {
      "epoch": 23.004864864864864,
      "grad_norm": 0.09762829542160034,
      "learning_rate": 5.945945945945947e-06,
      "loss": 1.1566,
      "step": 18060
    },
    {
      "epoch": 23.005122265122264,
      "grad_norm": 124.17290496826172,
      "learning_rate": 5.943085943085944e-06,
      "loss": 1.1647,
      "step": 18070
    },
    {
      "epoch": 23.005379665379664,
      "grad_norm": 0.00485936738550663,
      "learning_rate": 5.940225940225941e-06,
      "loss": 1.4024,
      "step": 18080
    },
    {
      "epoch": 23.005637065637064,
      "grad_norm": 0.20460176467895508,
      "learning_rate": 5.937365937365937e-06,
      "loss": 0.8586,
      "step": 18090
    },
    {
      "epoch": 23.005894465894468,
      "grad_norm": 0.4110458493232727,
      "learning_rate": 5.9345059345059356e-06,
      "loss": 0.4524,
      "step": 18100
    },
    {
      "epoch": 23.006151866151868,
      "grad_norm": 0.7026017904281616,
      "learning_rate": 5.931645931645932e-06,
      "loss": 0.46,
      "step": 18110
    },
    {
      "epoch": 23.006409266409268,
      "grad_norm": 0.053139615803956985,
      "learning_rate": 5.928785928785929e-06,
      "loss": 1.507,
      "step": 18120
    },
    {
      "epoch": 23.006666666666668,
      "grad_norm": 0.06488683074712753,
      "learning_rate": 5.925925925925926e-06,
      "loss": 1.1119,
      "step": 18130
    },
    {
      "epoch": 23.006924066924068,
      "grad_norm": 0.0023442870005965233,
      "learning_rate": 5.923065923065924e-06,
      "loss": 0.4408,
      "step": 18140
    },
    {
      "epoch": 23.007181467181468,
      "grad_norm": 0.0014166425680741668,
      "learning_rate": 5.9202059202059204e-06,
      "loss": 0.604,
      "step": 18150
    },
    {
      "epoch": 23.007438867438868,
      "grad_norm": 0.05282068997621536,
      "learning_rate": 5.917345917345918e-06,
      "loss": 0.6589,
      "step": 18160
    },
    {
      "epoch": 23.007696267696268,
      "grad_norm": 0.16524852812290192,
      "learning_rate": 5.914485914485914e-06,
      "loss": 0.6691,
      "step": 18170
    },
    {
      "epoch": 23.007953667953668,
      "grad_norm": 0.19815614819526672,
      "learning_rate": 5.911625911625912e-06,
      "loss": 0.5881,
      "step": 18180
    },
    {
      "epoch": 23.008211068211068,
      "grad_norm": 0.0784997045993805,
      "learning_rate": 5.908765908765909e-06,
      "loss": 0.6896,
      "step": 18190
    },
    {
      "epoch": 23.008468468468468,
      "grad_norm": 88.85114288330078,
      "learning_rate": 5.905905905905906e-06,
      "loss": 0.5887,
      "step": 18200
    },
    {
      "epoch": 23.00872586872587,
      "grad_norm": 0.0565468929708004,
      "learning_rate": 5.9030459030459035e-06,
      "loss": 0.0018,
      "step": 18210
    },
    {
      "epoch": 23.00898326898327,
      "grad_norm": 0.5154985785484314,
      "learning_rate": 5.900185900185901e-06,
      "loss": 0.0018,
      "step": 18220
    },
    {
      "epoch": 23.00924066924067,
      "grad_norm": 6.401853084564209,
      "learning_rate": 5.897325897325897e-06,
      "loss": 0.004,
      "step": 18230
    },
    {
      "epoch": 23.00949806949807,
      "grad_norm": 0.0013837914448231459,
      "learning_rate": 5.894465894465895e-06,
      "loss": 0.9277,
      "step": 18240
    },
    {
      "epoch": 23.00975546975547,
      "grad_norm": 0.5020064115524292,
      "learning_rate": 5.891605891605892e-06,
      "loss": 0.2748,
      "step": 18250
    },
    {
      "epoch": 23.01001287001287,
      "grad_norm": 0.00070767110446468,
      "learning_rate": 5.888745888745889e-06,
      "loss": 0.8912,
      "step": 18260
    },
    {
      "epoch": 23.01027027027027,
      "grad_norm": 113.7410659790039,
      "learning_rate": 5.885885885885886e-06,
      "loss": 1.8436,
      "step": 18270
    },
    {
      "epoch": 23.010527670527672,
      "grad_norm": 0.0004945895634591579,
      "learning_rate": 5.883025883025884e-06,
      "loss": 1.3281,
      "step": 18280
    },
    {
      "epoch": 23.010785070785072,
      "grad_norm": 0.05701057240366936,
      "learning_rate": 5.880165880165881e-06,
      "loss": 1.4695,
      "step": 18290
    },
    {
      "epoch": 23.011042471042472,
      "grad_norm": 66.18766021728516,
      "learning_rate": 5.877305877305878e-06,
      "loss": 1.8692,
      "step": 18300
    },
    {
      "epoch": 23.011299871299872,
      "grad_norm": 42.15882110595703,
      "learning_rate": 5.874445874445876e-06,
      "loss": 0.6605,
      "step": 18310
    },
    {
      "epoch": 23.011557271557272,
      "grad_norm": 0.0012678230414167047,
      "learning_rate": 5.871585871585872e-06,
      "loss": 0.4896,
      "step": 18320
    },
    {
      "epoch": 23.011814671814673,
      "grad_norm": 0.0016251800116151571,
      "learning_rate": 5.86872586872587e-06,
      "loss": 0.4608,
      "step": 18330
    },
    {
      "epoch": 23.012072072072073,
      "grad_norm": 0.0016299422131851315,
      "learning_rate": 5.865865865865866e-06,
      "loss": 0.5679,
      "step": 18340
    },
    {
      "epoch": 23.012329472329473,
      "grad_norm": 68.07299041748047,
      "learning_rate": 5.863005863005864e-06,
      "loss": 1.4937,
      "step": 18350
    },
    {
      "epoch": 23.012586872586873,
      "grad_norm": 85.31804656982422,
      "learning_rate": 5.860145860145861e-06,
      "loss": 0.5221,
      "step": 18360
    },
    {
      "epoch": 23.012844272844273,
      "grad_norm": 0.05515197291970253,
      "learning_rate": 5.857285857285858e-06,
      "loss": 0.507,
      "step": 18370
    },
    {
      "epoch": 23.013101673101673,
      "grad_norm": 0.0015318605583161116,
      "learning_rate": 5.8544258544258545e-06,
      "loss": 0.5488,
      "step": 18380
    },
    {
      "epoch": 23.013359073359073,
      "grad_norm": 105.14557647705078,
      "learning_rate": 5.851565851565853e-06,
      "loss": 0.5017,
      "step": 18390
    },
    {
      "epoch": 23.013616473616473,
      "grad_norm": 0.0008998190169222653,
      "learning_rate": 5.848705848705849e-06,
      "loss": 0.4514,
      "step": 18400
    },
    {
      "epoch": 23.013873873873873,
      "grad_norm": 32.71900939941406,
      "learning_rate": 5.8458458458458464e-06,
      "loss": 1.1053,
      "step": 18410
    },
    {
      "epoch": 23.014131274131273,
      "grad_norm": 47.391929626464844,
      "learning_rate": 5.842985842985843e-06,
      "loss": 0.5652,
      "step": 18420
    },
    {
      "epoch": 23.014388674388673,
      "grad_norm": 0.0010785746853798628,
      "learning_rate": 5.840125840125841e-06,
      "loss": 0.5836,
      "step": 18430
    },
    {
      "epoch": 23.014646074646073,
      "grad_norm": 0.09901858866214752,
      "learning_rate": 5.8372658372658375e-06,
      "loss": 0.9411,
      "step": 18440
    },
    {
      "epoch": 23.014903474903473,
      "grad_norm": 0.08682543784379959,
      "learning_rate": 5.834405834405835e-06,
      "loss": 0.0068,
      "step": 18450
    },
    {
      "epoch": 23.015160875160873,
      "grad_norm": 0.0029455472249537706,
      "learning_rate": 5.831545831545832e-06,
      "loss": 0.001,
      "step": 18460
    },
    {
      "epoch": 23.015418275418277,
      "grad_norm": 0.01022744458168745,
      "learning_rate": 5.8286858286858295e-06,
      "loss": 0.6507,
      "step": 18470
    },
    {
      "epoch": 23.015675675675677,
      "grad_norm": 0.3779960870742798,
      "learning_rate": 5.825825825825826e-06,
      "loss": 0.5542,
      "step": 18480
    },
    {
      "epoch": 23.015933075933077,
      "grad_norm": 8.886568069458008,
      "learning_rate": 5.822965822965823e-06,
      "loss": 1.0689,
      "step": 18490
    },
    {
      "epoch": 23.016190476190477,
      "grad_norm": 0.0006304460694082081,
      "learning_rate": 5.820105820105821e-06,
      "loss": 0.54,
      "step": 18500
    },
    {
      "epoch": 23.016447876447877,
      "grad_norm": 122.1751937866211,
      "learning_rate": 5.817245817245818e-06,
      "loss": 0.6934,
      "step": 18510
    },
    {
      "epoch": 23.016705276705277,
      "grad_norm": 0.020981237292289734,
      "learning_rate": 5.814385814385814e-06,
      "loss": 0.0029,
      "step": 18520
    },
    {
      "epoch": 23.016962676962677,
      "grad_norm": 0.0008985757594928145,
      "learning_rate": 5.8115258115258125e-06,
      "loss": 0.0038,
      "step": 18530
    },
    {
      "epoch": 23.017220077220077,
      "grad_norm": 0.0008014594204723835,
      "learning_rate": 5.808665808665809e-06,
      "loss": 1.7014,
      "step": 18540
    },
    {
      "epoch": 23.017477477477478,
      "grad_norm": 1.0590686798095703,
      "learning_rate": 5.805805805805806e-06,
      "loss": 0.461,
      "step": 18550
    },
    {
      "epoch": 23.017734877734878,
      "grad_norm": 0.003036707639694214,
      "learning_rate": 5.802945802945803e-06,
      "loss": 0.0031,
      "step": 18560
    },
    {
      "epoch": 23.017992277992278,
      "grad_norm": 0.194211944937706,
      "learning_rate": 5.800085800085801e-06,
      "loss": 1.5044,
      "step": 18570
    },
    {
      "epoch": 23.018249678249678,
      "grad_norm": 53.92101287841797,
      "learning_rate": 5.797225797225797e-06,
      "loss": 0.4982,
      "step": 18580
    },
    {
      "epoch": 23.018507078507078,
      "grad_norm": 0.11296827346086502,
      "learning_rate": 5.794365794365795e-06,
      "loss": 0.0015,
      "step": 18590
    },
    {
      "epoch": 23.018764478764478,
      "grad_norm": 0.0718424916267395,
      "learning_rate": 5.791505791505791e-06,
      "loss": 0.0019,
      "step": 18600
    },
    {
      "epoch": 23.019021879021878,
      "grad_norm": 0.05146656185388565,
      "learning_rate": 5.788645788645789e-06,
      "loss": 1.7993,
      "step": 18610
    },
    {
      "epoch": 23.019279279279278,
      "grad_norm": 107.91058349609375,
      "learning_rate": 5.785785785785786e-06,
      "loss": 1.0372,
      "step": 18620
    },
    {
      "epoch": 23.019536679536678,
      "grad_norm": 94.91177368164062,
      "learning_rate": 5.782925782925783e-06,
      "loss": 1.1122,
      "step": 18630
    },
    {
      "epoch": 23.019794079794078,
      "grad_norm": 0.0015893593663349748,
      "learning_rate": 5.78006578006578e-06,
      "loss": 0.4632,
      "step": 18640
    },
    {
      "epoch": 23.02,
      "eval_accuracy": 0.8913043478260869,
      "eval_loss": 0.6939725875854492,
      "eval_runtime": 13.4416,
      "eval_samples_per_second": 3.422,
      "eval_steps_per_second": 3.422,
      "step": 18648
    },
    {
      "epoch": 24.00005148005148,
      "grad_norm": 0.05807127431035042,
      "learning_rate": 5.777205777205778e-06,
      "loss": 1.1037,
      "step": 18650
    },
    {
      "epoch": 24.00030888030888,
      "grad_norm": 92.4770278930664,
      "learning_rate": 5.774345774345775e-06,
      "loss": 1.2418,
      "step": 18660
    },
    {
      "epoch": 24.000566280566282,
      "grad_norm": 0.07446417212486267,
      "learning_rate": 5.7714857714857716e-06,
      "loss": 0.5503,
      "step": 18670
    },
    {
      "epoch": 24.000823680823682,
      "grad_norm": 0.0009795395890250802,
      "learning_rate": 5.76862576862577e-06,
      "loss": 0.0079,
      "step": 18680
    },
    {
      "epoch": 24.001081081081082,
      "grad_norm": 0.5915771126747131,
      "learning_rate": 5.765765765765766e-06,
      "loss": 0.0016,
      "step": 18690
    },
    {
      "epoch": 24.001338481338482,
      "grad_norm": 1.0889242887496948,
      "learning_rate": 5.7629057629057635e-06,
      "loss": 0.2133,
      "step": 18700
    },
    {
      "epoch": 24.001595881595883,
      "grad_norm": 0.0007270134519785643,
      "learning_rate": 5.760045760045761e-06,
      "loss": 0.425,
      "step": 18710
    },
    {
      "epoch": 24.001853281853283,
      "grad_norm": 0.0018572467379271984,
      "learning_rate": 5.757185757185758e-06,
      "loss": 1.1419,
      "step": 18720
    },
    {
      "epoch": 24.002110682110683,
      "grad_norm": 0.02151697687804699,
      "learning_rate": 5.754325754325755e-06,
      "loss": 0.0007,
      "step": 18730
    },
    {
      "epoch": 24.002368082368083,
      "grad_norm": 0.003686880227178335,
      "learning_rate": 5.751465751465753e-06,
      "loss": 0.0021,
      "step": 18740
    },
    {
      "epoch": 24.002625482625483,
      "grad_norm": 0.0470479391515255,
      "learning_rate": 5.748605748605749e-06,
      "loss": 0.0016,
      "step": 18750
    },
    {
      "epoch": 24.002882882882883,
      "grad_norm": 0.06280643492937088,
      "learning_rate": 5.7457457457457466e-06,
      "loss": 0.1671,
      "step": 18760
    },
    {
      "epoch": 24.003140283140283,
      "grad_norm": 0.0023778968024998903,
      "learning_rate": 5.742885742885743e-06,
      "loss": 0.5986,
      "step": 18770
    },
    {
      "epoch": 24.003397683397683,
      "grad_norm": 0.14014458656311035,
      "learning_rate": 5.740025740025741e-06,
      "loss": 0.716,
      "step": 18780
    },
    {
      "epoch": 24.003655083655083,
      "grad_norm": 0.009725264273583889,
      "learning_rate": 5.737165737165738e-06,
      "loss": 0.0008,
      "step": 18790
    },
    {
      "epoch": 24.003912483912483,
      "grad_norm": 0.16021865606307983,
      "learning_rate": 5.734305734305735e-06,
      "loss": 0.0109,
      "step": 18800
    },
    {
      "epoch": 24.004169884169883,
      "grad_norm": 0.0022326132748275995,
      "learning_rate": 5.7314457314457315e-06,
      "loss": 0.4593,
      "step": 18810
    },
    {
      "epoch": 24.004427284427283,
      "grad_norm": 48.16182327270508,
      "learning_rate": 5.72858572858573e-06,
      "loss": 0.5535,
      "step": 18820
    },
    {
      "epoch": 24.004684684684683,
      "grad_norm": 0.034395068883895874,
      "learning_rate": 5.725725725725726e-06,
      "loss": 0.6363,
      "step": 18830
    },
    {
      "epoch": 24.004942084942083,
      "grad_norm": 93.80135345458984,
      "learning_rate": 5.722865722865723e-06,
      "loss": 0.7659,
      "step": 18840
    },
    {
      "epoch": 24.005199485199483,
      "grad_norm": 0.12201082706451416,
      "learning_rate": 5.72000572000572e-06,
      "loss": 1.012,
      "step": 18850
    },
    {
      "epoch": 24.005456885456887,
      "grad_norm": 44.175174713134766,
      "learning_rate": 5.717145717145718e-06,
      "loss": 1.1665,
      "step": 18860
    },
    {
      "epoch": 24.005714285714287,
      "grad_norm": 0.003605901263654232,
      "learning_rate": 5.7142857142857145e-06,
      "loss": 1.3019,
      "step": 18870
    },
    {
      "epoch": 24.005971685971687,
      "grad_norm": 0.02306421287357807,
      "learning_rate": 5.711425711425712e-06,
      "loss": 0.0016,
      "step": 18880
    },
    {
      "epoch": 24.006229086229087,
      "grad_norm": 0.0023764725774526596,
      "learning_rate": 5.708565708565708e-06,
      "loss": 0.3606,
      "step": 18890
    },
    {
      "epoch": 24.006486486486487,
      "grad_norm": 105.87065124511719,
      "learning_rate": 5.7057057057057065e-06,
      "loss": 2.0477,
      "step": 18900
    },
    {
      "epoch": 24.006743886743887,
      "grad_norm": 33.09611511230469,
      "learning_rate": 5.702845702845703e-06,
      "loss": 0.589,
      "step": 18910
    },
    {
      "epoch": 24.007001287001287,
      "grad_norm": 0.10097678750753403,
      "learning_rate": 5.6999856999857e-06,
      "loss": 0.9933,
      "step": 18920
    },
    {
      "epoch": 24.007258687258688,
      "grad_norm": 0.0016176817007362843,
      "learning_rate": 5.6971256971256976e-06,
      "loss": 1.1839,
      "step": 18930
    },
    {
      "epoch": 24.007516087516088,
      "grad_norm": 0.006734101101756096,
      "learning_rate": 5.694265694265695e-06,
      "loss": 1.6423,
      "step": 18940
    },
    {
      "epoch": 24.007773487773488,
      "grad_norm": 0.015379363670945168,
      "learning_rate": 5.691405691405691e-06,
      "loss": 0.6437,
      "step": 18950
    },
    {
      "epoch": 24.008030888030888,
      "grad_norm": 0.16267436742782593,
      "learning_rate": 5.6885456885456895e-06,
      "loss": 0.5995,
      "step": 18960
    },
    {
      "epoch": 24.008288288288288,
      "grad_norm": 0.11775770038366318,
      "learning_rate": 5.685685685685686e-06,
      "loss": 0.5405,
      "step": 18970
    },
    {
      "epoch": 24.008545688545688,
      "grad_norm": 0.007473708596080542,
      "learning_rate": 5.682825682825683e-06,
      "loss": 0.5841,
      "step": 18980
    },
    {
      "epoch": 24.008803088803088,
      "grad_norm": 0.02219763956964016,
      "learning_rate": 5.67996567996568e-06,
      "loss": 0.6271,
      "step": 18990
    },
    {
      "epoch": 24.009060489060488,
      "grad_norm": 0.03645698353648186,
      "learning_rate": 5.677105677105678e-06,
      "loss": 0.5254,
      "step": 19000
    },
    {
      "epoch": 24.009317889317888,
      "grad_norm": 0.13147737085819244,
      "learning_rate": 5.674245674245674e-06,
      "loss": 0.5281,
      "step": 19010
    },
    {
      "epoch": 24.009575289575288,
      "grad_norm": 0.0017814674647524953,
      "learning_rate": 5.671385671385672e-06,
      "loss": 0.5026,
      "step": 19020
    },
    {
      "epoch": 24.009832689832688,
      "grad_norm": 0.5326687097549438,
      "learning_rate": 5.668525668525668e-06,
      "loss": 0.0525,
      "step": 19030
    },
    {
      "epoch": 24.01009009009009,
      "grad_norm": 0.003802061080932617,
      "learning_rate": 5.665665665665666e-06,
      "loss": 0.2906,
      "step": 19040
    },
    {
      "epoch": 24.010347490347492,
      "grad_norm": 0.12540924549102783,
      "learning_rate": 5.662805662805664e-06,
      "loss": 0.6137,
      "step": 19050
    },
    {
      "epoch": 24.010604890604892,
      "grad_norm": 0.002418956020846963,
      "learning_rate": 5.65994565994566e-06,
      "loss": 0.2951,
      "step": 19060
    },
    {
      "epoch": 24.010862290862292,
      "grad_norm": 0.000946474785450846,
      "learning_rate": 5.657085657085658e-06,
      "loss": 0.0026,
      "step": 19070
    },
    {
      "epoch": 24.011119691119692,
      "grad_norm": 0.8769616484642029,
      "learning_rate": 5.654225654225655e-06,
      "loss": 1.3478,
      "step": 19080
    },
    {
      "epoch": 24.011377091377092,
      "grad_norm": 66.09345245361328,
      "learning_rate": 5.651365651365652e-06,
      "loss": 0.4886,
      "step": 19090
    },
    {
      "epoch": 24.011634491634492,
      "grad_norm": 60.73049545288086,
      "learning_rate": 5.6485056485056485e-06,
      "loss": 3.3388,
      "step": 19100
    },
    {
      "epoch": 24.011891891891892,
      "grad_norm": 0.6036048531532288,
      "learning_rate": 5.645645645645647e-06,
      "loss": 1.0743,
      "step": 19110
    },
    {
      "epoch": 24.012149292149292,
      "grad_norm": 0.00891997292637825,
      "learning_rate": 5.642785642785643e-06,
      "loss": 1.5969,
      "step": 19120
    },
    {
      "epoch": 24.012406692406692,
      "grad_norm": 0.007564838510006666,
      "learning_rate": 5.6399256399256405e-06,
      "loss": 1.0353,
      "step": 19130
    },
    {
      "epoch": 24.012664092664092,
      "grad_norm": 0.0008338902262039483,
      "learning_rate": 5.637065637065637e-06,
      "loss": 0.2451,
      "step": 19140
    },
    {
      "epoch": 24.012921492921492,
      "grad_norm": 66.34471893310547,
      "learning_rate": 5.634205634205635e-06,
      "loss": 3.2594,
      "step": 19150
    },
    {
      "epoch": 24.013178893178893,
      "grad_norm": 0.5221831798553467,
      "learning_rate": 5.631345631345632e-06,
      "loss": 0.4028,
      "step": 19160
    },
    {
      "epoch": 24.013436293436293,
      "grad_norm": 0.0011873157927766442,
      "learning_rate": 5.628485628485629e-06,
      "loss": 0.9757,
      "step": 19170
    },
    {
      "epoch": 24.013693693693693,
      "grad_norm": 0.011443239636719227,
      "learning_rate": 5.625625625625626e-06,
      "loss": 0.8103,
      "step": 19180
    },
    {
      "epoch": 24.013951093951093,
      "grad_norm": 0.7135658860206604,
      "learning_rate": 5.6227656227656235e-06,
      "loss": 0.0047,
      "step": 19190
    },
    {
      "epoch": 24.014208494208493,
      "grad_norm": 0.09131744503974915,
      "learning_rate": 5.61990561990562e-06,
      "loss": 0.529,
      "step": 19200
    },
    {
      "epoch": 24.014465894465893,
      "grad_norm": 2919.614501953125,
      "learning_rate": 5.617045617045618e-06,
      "loss": 0.5351,
      "step": 19210
    },
    {
      "epoch": 24.014723294723296,
      "grad_norm": 0.0010455803712829947,
      "learning_rate": 5.614185614185615e-06,
      "loss": 1.8551,
      "step": 19220
    },
    {
      "epoch": 24.014980694980697,
      "grad_norm": 0.0007820248720236123,
      "learning_rate": 5.611325611325612e-06,
      "loss": 0.3175,
      "step": 19230
    },
    {
      "epoch": 24.015238095238097,
      "grad_norm": 7.954380512237549,
      "learning_rate": 5.6084656084656084e-06,
      "loss": 0.6042,
      "step": 19240
    },
    {
      "epoch": 24.015495495495497,
      "grad_norm": 0.33642780780792236,
      "learning_rate": 5.605605605605607e-06,
      "loss": 2.7028,
      "step": 19250
    },
    {
      "epoch": 24.015752895752897,
      "grad_norm": 0.17822858691215515,
      "learning_rate": 5.602745602745603e-06,
      "loss": 0.0028,
      "step": 19260
    },
    {
      "epoch": 24.016010296010297,
      "grad_norm": 0.2161465287208557,
      "learning_rate": 5.5998855998856e-06,
      "loss": 0.0047,
      "step": 19270
    },
    {
      "epoch": 24.016267696267697,
      "grad_norm": 0.16744586825370789,
      "learning_rate": 5.597025597025597e-06,
      "loss": 0.4083,
      "step": 19280
    },
    {
      "epoch": 24.016525096525097,
      "grad_norm": 0.001031736028380692,
      "learning_rate": 5.594165594165595e-06,
      "loss": 1.0542,
      "step": 19290
    },
    {
      "epoch": 24.016782496782497,
      "grad_norm": 0.45700398087501526,
      "learning_rate": 5.5913055913055915e-06,
      "loss": 0.0029,
      "step": 19300
    },
    {
      "epoch": 24.017039897039897,
      "grad_norm": 0.12631818652153015,
      "learning_rate": 5.588445588445589e-06,
      "loss": 0.3904,
      "step": 19310
    },
    {
      "epoch": 24.017297297297297,
      "grad_norm": 0.0006879584980197251,
      "learning_rate": 5.585585585585585e-06,
      "loss": 0.3659,
      "step": 19320
    },
    {
      "epoch": 24.017554697554697,
      "grad_norm": 633.3937377929688,
      "learning_rate": 5.5827255827255834e-06,
      "loss": 1.9248,
      "step": 19330
    },
    {
      "epoch": 24.017812097812097,
      "grad_norm": 0.5299988985061646,
      "learning_rate": 5.57986557986558e-06,
      "loss": 0.0055,
      "step": 19340
    },
    {
      "epoch": 24.018069498069497,
      "grad_norm": 0.06562308967113495,
      "learning_rate": 5.577005577005577e-06,
      "loss": 0.5882,
      "step": 19350
    },
    {
      "epoch": 24.018326898326897,
      "grad_norm": 0.07744754105806351,
      "learning_rate": 5.574145574145574e-06,
      "loss": 0.9198,
      "step": 19360
    },
    {
      "epoch": 24.018584298584297,
      "grad_norm": 0.8179073929786682,
      "learning_rate": 5.571285571285572e-06,
      "loss": 1.6893,
      "step": 19370
    },
    {
      "epoch": 24.018841698841698,
      "grad_norm": 0.12629270553588867,
      "learning_rate": 5.568425568425568e-06,
      "loss": 0.4581,
      "step": 19380
    },
    {
      "epoch": 24.019099099099098,
      "grad_norm": 0.1680009514093399,
      "learning_rate": 5.565565565565566e-06,
      "loss": 0.0179,
      "step": 19390
    },
    {
      "epoch": 24.019356499356498,
      "grad_norm": 0.0008770914864726365,
      "learning_rate": 5.562705562705563e-06,
      "loss": 1.0297,
      "step": 19400
    },
    {
      "epoch": 24.0196138996139,
      "grad_norm": 0.01934773102402687,
      "learning_rate": 5.55984555984556e-06,
      "loss": 0.0036,
      "step": 19410
    },
    {
      "epoch": 24.0198712998713,
      "grad_norm": 0.0009819872211664915,
      "learning_rate": 5.556985556985558e-06,
      "loss": 0.4901,
      "step": 19420
    },
    {
      "epoch": 24.02,
      "eval_accuracy": 0.8260869565217391,
      "eval_loss": 1.0198997259140015,
      "eval_runtime": 13.4522,
      "eval_samples_per_second": 3.42,
      "eval_steps_per_second": 3.42,
      "step": 19425
    },
    {
      "epoch": 25.000128700128702,
      "grad_norm": 0.0006284094415605068,
      "learning_rate": 5.554125554125555e-06,
      "loss": 0.0018,
      "step": 19430
    },
    {
      "epoch": 25.000386100386102,
      "grad_norm": 0.20130889117717743,
      "learning_rate": 5.551265551265552e-06,
      "loss": 0.4494,
      "step": 19440
    },
    {
      "epoch": 25.000643500643502,
      "grad_norm": 0.2460191547870636,
      "learning_rate": 5.548405548405549e-06,
      "loss": 0.4706,
      "step": 19450
    },
    {
      "epoch": 25.000900900900902,
      "grad_norm": 0.23789767920970917,
      "learning_rate": 5.545545545545547e-06,
      "loss": 0.0023,
      "step": 19460
    },
    {
      "epoch": 25.001158301158302,
      "grad_norm": 0.0007821908802725375,
      "learning_rate": 5.542685542685543e-06,
      "loss": 0.4147,
      "step": 19470
    },
    {
      "epoch": 25.001415701415702,
      "grad_norm": 0.10670550167560577,
      "learning_rate": 5.539825539825541e-06,
      "loss": 0.4267,
      "step": 19480
    },
    {
      "epoch": 25.001673101673102,
      "grad_norm": 0.05941932648420334,
      "learning_rate": 5.536965536965537e-06,
      "loss": 0.0058,
      "step": 19490
    },
    {
      "epoch": 25.001930501930502,
      "grad_norm": 0.06344068050384521,
      "learning_rate": 5.534105534105535e-06,
      "loss": 0.0027,
      "step": 19500
    },
    {
      "epoch": 25.002187902187902,
      "grad_norm": 0.06859828531742096,
      "learning_rate": 5.531245531245532e-06,
      "loss": 0.5773,
      "step": 19510
    },
    {
      "epoch": 25.002445302445302,
      "grad_norm": 0.14273127913475037,
      "learning_rate": 5.528385528385529e-06,
      "loss": 0.0028,
      "step": 19520
    },
    {
      "epoch": 25.002702702702702,
      "grad_norm": 0.0016670461045578122,
      "learning_rate": 5.5255255255255255e-06,
      "loss": 1.1759,
      "step": 19530
    },
    {
      "epoch": 25.002960102960103,
      "grad_norm": 0.5187317132949829,
      "learning_rate": 5.522665522665524e-06,
      "loss": 0.5372,
      "step": 19540
    },
    {
      "epoch": 25.003217503217503,
      "grad_norm": 0.0007036603637970984,
      "learning_rate": 5.51980551980552e-06,
      "loss": 1.282,
      "step": 19550
    },
    {
      "epoch": 25.003474903474903,
      "grad_norm": 899.7644653320312,
      "learning_rate": 5.5169455169455175e-06,
      "loss": 0.7085,
      "step": 19560
    },
    {
      "epoch": 25.003732303732303,
      "grad_norm": 0.0008057368686422706,
      "learning_rate": 5.514085514085514e-06,
      "loss": 0.0022,
      "step": 19570
    },
    {
      "epoch": 25.003989703989703,
      "grad_norm": 83.44464111328125,
      "learning_rate": 5.511225511225512e-06,
      "loss": 0.4203,
      "step": 19580
    },
    {
      "epoch": 25.004247104247103,
      "grad_norm": 0.32760846614837646,
      "learning_rate": 5.5083655083655086e-06,
      "loss": 0.8353,
      "step": 19590
    },
    {
      "epoch": 25.004504504504503,
      "grad_norm": 8.652959823608398,
      "learning_rate": 5.505505505505506e-06,
      "loss": 0.1742,
      "step": 19600
    },
    {
      "epoch": 25.004761904761907,
      "grad_norm": 0.0009207252878695726,
      "learning_rate": 5.502645502645503e-06,
      "loss": 0.4249,
      "step": 19610
    },
    {
      "epoch": 25.005019305019307,
      "grad_norm": 0.0004057708429172635,
      "learning_rate": 5.4997854997855005e-06,
      "loss": 0.0023,
      "step": 19620
    },
    {
      "epoch": 25.005276705276707,
      "grad_norm": 0.046492137014865875,
      "learning_rate": 5.496925496925497e-06,
      "loss": 0.7061,
      "step": 19630
    },
    {
      "epoch": 25.005534105534107,
      "grad_norm": 0.03466697037220001,
      "learning_rate": 5.494065494065494e-06,
      "loss": 0.5153,
      "step": 19640
    },
    {
      "epoch": 25.005791505791507,
      "grad_norm": 0.2154761403799057,
      "learning_rate": 5.491205491205492e-06,
      "loss": 0.5921,
      "step": 19650
    },
    {
      "epoch": 25.006048906048907,
      "grad_norm": 98.8327865600586,
      "learning_rate": 5.488345488345489e-06,
      "loss": 1.7348,
      "step": 19660
    },
    {
      "epoch": 25.006306306306307,
      "grad_norm": 0.031177762895822525,
      "learning_rate": 5.485485485485485e-06,
      "loss": 1.097,
      "step": 19670
    },
    {
      "epoch": 25.006563706563707,
      "grad_norm": 0.03502604365348816,
      "learning_rate": 5.4826254826254836e-06,
      "loss": 0.6449,
      "step": 19680
    },
    {
      "epoch": 25.006821106821107,
      "grad_norm": 0.0010554458713158965,
      "learning_rate": 5.47976547976548e-06,
      "loss": 0.0009,
      "step": 19690
    },
    {
      "epoch": 25.007078507078507,
      "grad_norm": 135.35574340820312,
      "learning_rate": 5.476905476905477e-06,
      "loss": 0.6102,
      "step": 19700
    },
    {
      "epoch": 25.007335907335907,
      "grad_norm": 0.005914623849093914,
      "learning_rate": 5.474045474045474e-06,
      "loss": 1.36,
      "step": 19710
    },
    {
      "epoch": 25.007593307593307,
      "grad_norm": 0.011718814261257648,
      "learning_rate": 5.471185471185472e-06,
      "loss": 0.0047,
      "step": 19720
    },
    {
      "epoch": 25.007850707850707,
      "grad_norm": 32.16627883911133,
      "learning_rate": 5.4683254683254685e-06,
      "loss": 1.2207,
      "step": 19730
    },
    {
      "epoch": 25.008108108108107,
      "grad_norm": 0.0004114433249924332,
      "learning_rate": 5.465465465465466e-06,
      "loss": 1.9784,
      "step": 19740
    },
    {
      "epoch": 25.008365508365507,
      "grad_norm": 0.0007554457988590002,
      "learning_rate": 5.462605462605462e-06,
      "loss": 0.4903,
      "step": 19750
    },
    {
      "epoch": 25.008622908622907,
      "grad_norm": 0.09954472631216049,
      "learning_rate": 5.45974545974546e-06,
      "loss": 0.0039,
      "step": 19760
    },
    {
      "epoch": 25.008880308880308,
      "grad_norm": 0.000810124387498945,
      "learning_rate": 5.456885456885457e-06,
      "loss": 0.5262,
      "step": 19770
    },
    {
      "epoch": 25.009137709137708,
      "grad_norm": 0.000533245038241148,
      "learning_rate": 5.454025454025454e-06,
      "loss": 0.0051,
      "step": 19780
    },
    {
      "epoch": 25.009395109395108,
      "grad_norm": 33.438385009765625,
      "learning_rate": 5.451165451165452e-06,
      "loss": 0.537,
      "step": 19790
    },
    {
      "epoch": 25.00965250965251,
      "grad_norm": 187.0354766845703,
      "learning_rate": 5.448305448305449e-06,
      "loss": 0.8539,
      "step": 19800
    },
    {
      "epoch": 25.00990990990991,
      "grad_norm": 0.11020729690790176,
      "learning_rate": 5.445445445445446e-06,
      "loss": 0.6905,
      "step": 19810
    },
    {
      "epoch": 25.01016731016731,
      "grad_norm": 0.1317060887813568,
      "learning_rate": 5.442585442585443e-06,
      "loss": 0.498,
      "step": 19820
    },
    {
      "epoch": 25.01042471042471,
      "grad_norm": 0.028832558542490005,
      "learning_rate": 5.439725439725441e-06,
      "loss": 0.004,
      "step": 19830
    },
    {
      "epoch": 25.01068211068211,
      "grad_norm": 36.885658264160156,
      "learning_rate": 5.436865436865437e-06,
      "loss": 1.3712,
      "step": 19840
    },
    {
      "epoch": 25.01093951093951,
      "grad_norm": 0.062064073979854584,
      "learning_rate": 5.4340054340054346e-06,
      "loss": 0.0027,
      "step": 19850
    },
    {
      "epoch": 25.01119691119691,
      "grad_norm": 0.003508289810270071,
      "learning_rate": 5.431145431145432e-06,
      "loss": 1.2743,
      "step": 19860
    },
    {
      "epoch": 25.01145431145431,
      "grad_norm": 0.21302658319473267,
      "learning_rate": 5.428285428285429e-06,
      "loss": 0.0033,
      "step": 19870
    },
    {
      "epoch": 25.011711711711712,
      "grad_norm": 304.8260192871094,
      "learning_rate": 5.425425425425426e-06,
      "loss": 0.4843,
      "step": 19880
    },
    {
      "epoch": 25.011969111969112,
      "grad_norm": 0.0794137567281723,
      "learning_rate": 5.422565422565423e-06,
      "loss": 0.5961,
      "step": 19890
    },
    {
      "epoch": 25.012226512226512,
      "grad_norm": 1.3407971858978271,
      "learning_rate": 5.41970541970542e-06,
      "loss": 0.3193,
      "step": 19900
    },
    {
      "epoch": 25.012483912483912,
      "grad_norm": 0.0008371305884793401,
      "learning_rate": 5.416845416845418e-06,
      "loss": 0.0019,
      "step": 19910
    },
    {
      "epoch": 25.012741312741312,
      "grad_norm": 0.00039880804251879454,
      "learning_rate": 5.413985413985414e-06,
      "loss": 1.143,
      "step": 19920
    },
    {
      "epoch": 25.012998712998712,
      "grad_norm": 0.002268696902319789,
      "learning_rate": 5.411125411125412e-06,
      "loss": 0.002,
      "step": 19930
    },
    {
      "epoch": 25.013256113256112,
      "grad_norm": 140.1703643798828,
      "learning_rate": 5.408265408265409e-06,
      "loss": 0.021,
      "step": 19940
    },
    {
      "epoch": 25.013513513513512,
      "grad_norm": 26.61545181274414,
      "learning_rate": 5.405405405405406e-06,
      "loss": 0.0069,
      "step": 19950
    },
    {
      "epoch": 25.013770913770912,
      "grad_norm": 0.00035982835106551647,
      "learning_rate": 5.4025454025454025e-06,
      "loss": 0.5916,
      "step": 19960
    },
    {
      "epoch": 25.014028314028312,
      "grad_norm": 0.11984622478485107,
      "learning_rate": 5.399685399685401e-06,
      "loss": 0.5778,
      "step": 19970
    },
    {
      "epoch": 25.014285714285716,
      "grad_norm": 0.22817088663578033,
      "learning_rate": 5.396825396825397e-06,
      "loss": 0.0021,
      "step": 19980
    },
    {
      "epoch": 25.014543114543116,
      "grad_norm": 0.2587432861328125,
      "learning_rate": 5.3939653939653944e-06,
      "loss": 0.8712,
      "step": 19990
    },
    {
      "epoch": 25.014800514800516,
      "grad_norm": 0.007341683376580477,
      "learning_rate": 5.391105391105391e-06,
      "loss": 0.6829,
      "step": 20000
    },
    {
      "epoch": 25.015057915057916,
      "grad_norm": 0.0007906981627456844,
      "learning_rate": 5.388245388245389e-06,
      "loss": 0.5817,
      "step": 20010
    },
    {
      "epoch": 25.015315315315316,
      "grad_norm": 0.0988348200917244,
      "learning_rate": 5.3853853853853856e-06,
      "loss": 1.7047,
      "step": 20020
    },
    {
      "epoch": 25.015572715572716,
      "grad_norm": 0.0006614571902900934,
      "learning_rate": 5.382525382525383e-06,
      "loss": 0.4903,
      "step": 20030
    },
    {
      "epoch": 25.015830115830116,
      "grad_norm": 0.2373206615447998,
      "learning_rate": 5.379665379665379e-06,
      "loss": 0.002,
      "step": 20040
    },
    {
      "epoch": 25.016087516087516,
      "grad_norm": 0.016258088871836662,
      "learning_rate": 5.3768053768053775e-06,
      "loss": 0.0274,
      "step": 20050
    },
    {
      "epoch": 25.016344916344917,
      "grad_norm": 148.20025634765625,
      "learning_rate": 5.373945373945374e-06,
      "loss": 0.9257,
      "step": 20060
    },
    {
      "epoch": 25.016602316602317,
      "grad_norm": 0.030567055568099022,
      "learning_rate": 5.371085371085371e-06,
      "loss": 0.0024,
      "step": 20070
    },
    {
      "epoch": 25.016859716859717,
      "grad_norm": 0.007402033545076847,
      "learning_rate": 5.368225368225369e-06,
      "loss": 0.6018,
      "step": 20080
    },
    {
      "epoch": 25.017117117117117,
      "grad_norm": 0.0003875165421050042,
      "learning_rate": 5.365365365365366e-06,
      "loss": 1.2368,
      "step": 20090
    },
    {
      "epoch": 25.017374517374517,
      "grad_norm": 0.0308153685182333,
      "learning_rate": 5.362505362505362e-06,
      "loss": 0.0537,
      "step": 20100
    },
    {
      "epoch": 25.017631917631917,
      "grad_norm": 1280.4652099609375,
      "learning_rate": 5.3596453596453605e-06,
      "loss": 2.5197,
      "step": 20110
    },
    {
      "epoch": 25.017889317889317,
      "grad_norm": 2.078805685043335,
      "learning_rate": 5.356785356785357e-06,
      "loss": 0.8314,
      "step": 20120
    },
    {
      "epoch": 25.018146718146717,
      "grad_norm": 0.001093574333935976,
      "learning_rate": 5.353925353925354e-06,
      "loss": 0.6905,
      "step": 20130
    },
    {
      "epoch": 25.018404118404117,
      "grad_norm": 0.0019031533738598228,
      "learning_rate": 5.351065351065351e-06,
      "loss": 0.0019,
      "step": 20140
    },
    {
      "epoch": 25.018661518661517,
      "grad_norm": 0.07940136641263962,
      "learning_rate": 5.348205348205349e-06,
      "loss": 1.0073,
      "step": 20150
    },
    {
      "epoch": 25.018918918918917,
      "grad_norm": 0.004731150344014168,
      "learning_rate": 5.345345345345346e-06,
      "loss": 0.0022,
      "step": 20160
    },
    {
      "epoch": 25.01917631917632,
      "grad_norm": 0.007168655749410391,
      "learning_rate": 5.342485342485343e-06,
      "loss": 1.4627,
      "step": 20170
    },
    {
      "epoch": 25.01943371943372,
      "grad_norm": 0.0005902776028960943,
      "learning_rate": 5.339625339625341e-06,
      "loss": 0.0157,
      "step": 20180
    },
    {
      "epoch": 25.01969111969112,
      "grad_norm": 1.8064913749694824,
      "learning_rate": 5.336765336765337e-06,
      "loss": 1.1586,
      "step": 20190
    },
    {
      "epoch": 25.01994851994852,
      "grad_norm": 0.08647841215133667,
      "learning_rate": 5.333905333905335e-06,
      "loss": 0.3927,
      "step": 20200
    },
    {
      "epoch": 25.02,
      "eval_accuracy": 0.8260869565217391,
      "eval_loss": 0.8669257760047913,
      "eval_runtime": 13.4055,
      "eval_samples_per_second": 3.431,
      "eval_steps_per_second": 3.431,
      "step": 20202
    },
    {
      "epoch": 26.00020592020592,
      "grad_norm": 41.40488815307617,
      "learning_rate": 5.331045331045331e-06,
      "loss": 1.0282,
      "step": 20210
    },
    {
      "epoch": 26.00046332046332,
      "grad_norm": 1.0805187225341797,
      "learning_rate": 5.328185328185329e-06,
      "loss": 1.0007,
      "step": 20220
    },
    {
      "epoch": 26.00072072072072,
      "grad_norm": 0.017646795138716698,
      "learning_rate": 5.325325325325326e-06,
      "loss": 0.4639,
      "step": 20230
    },
    {
      "epoch": 26.00097812097812,
      "grad_norm": 0.08881297707557678,
      "learning_rate": 5.322465322465323e-06,
      "loss": 0.0305,
      "step": 20240
    },
    {
      "epoch": 26.00123552123552,
      "grad_norm": 0.09640834480524063,
      "learning_rate": 5.31960531960532e-06,
      "loss": 0.494,
      "step": 20250
    },
    {
      "epoch": 26.001492921492922,
      "grad_norm": 87.36897277832031,
      "learning_rate": 5.316745316745318e-06,
      "loss": 1.564,
      "step": 20260
    },
    {
      "epoch": 26.001750321750322,
      "grad_norm": 0.0012800369877368212,
      "learning_rate": 5.313885313885314e-06,
      "loss": 0.0018,
      "step": 20270
    },
    {
      "epoch": 26.002007722007722,
      "grad_norm": 0.09953822195529938,
      "learning_rate": 5.3110253110253115e-06,
      "loss": 0.0041,
      "step": 20280
    },
    {
      "epoch": 26.002265122265122,
      "grad_norm": 39.34917449951172,
      "learning_rate": 5.308165308165308e-06,
      "loss": 1.5296,
      "step": 20290
    },
    {
      "epoch": 26.002522522522522,
      "grad_norm": 0.09963509440422058,
      "learning_rate": 5.305305305305306e-06,
      "loss": 0.0017,
      "step": 20300
    },
    {
      "epoch": 26.002779922779922,
      "grad_norm": 0.16209664940834045,
      "learning_rate": 5.302445302445303e-06,
      "loss": 0.494,
      "step": 20310
    },
    {
      "epoch": 26.003037323037322,
      "grad_norm": 0.0013167433207854629,
      "learning_rate": 5.2995852995853e-06,
      "loss": 0.0025,
      "step": 20320
    },
    {
      "epoch": 26.003294723294722,
      "grad_norm": 0.0038293786346912384,
      "learning_rate": 5.296725296725297e-06,
      "loss": 0.0017,
      "step": 20330
    },
    {
      "epoch": 26.003552123552122,
      "grad_norm": 0.0007873664726503193,
      "learning_rate": 5.293865293865295e-06,
      "loss": 0.5308,
      "step": 20340
    },
    {
      "epoch": 26.003809523809522,
      "grad_norm": 0.07472074031829834,
      "learning_rate": 5.291005291005291e-06,
      "loss": 1.9437,
      "step": 20350
    },
    {
      "epoch": 26.004066924066922,
      "grad_norm": 0.0006641810177825391,
      "learning_rate": 5.288145288145289e-06,
      "loss": 0.0143,
      "step": 20360
    },
    {
      "epoch": 26.004324324324326,
      "grad_norm": 0.2543509304523468,
      "learning_rate": 5.285285285285286e-06,
      "loss": 0.3686,
      "step": 20370
    },
    {
      "epoch": 26.004581724581726,
      "grad_norm": 0.0021348209120333195,
      "learning_rate": 5.282425282425283e-06,
      "loss": 0.6354,
      "step": 20380
    },
    {
      "epoch": 26.004839124839126,
      "grad_norm": 0.0008653805707581341,
      "learning_rate": 5.2795652795652795e-06,
      "loss": 0.3066,
      "step": 20390
    },
    {
      "epoch": 26.005096525096526,
      "grad_norm": 0.0004059734055772424,
      "learning_rate": 5.276705276705278e-06,
      "loss": 0.0022,
      "step": 20400
    },
    {
      "epoch": 26.005353925353926,
      "grad_norm": 0.0010638394160196185,
      "learning_rate": 5.273845273845274e-06,
      "loss": 0.5565,
      "step": 20410
    },
    {
      "epoch": 26.005611325611326,
      "grad_norm": 0.0026838104240596294,
      "learning_rate": 5.2709852709852714e-06,
      "loss": 0.5538,
      "step": 20420
    },
    {
      "epoch": 26.005868725868726,
      "grad_norm": 0.0007530002039857209,
      "learning_rate": 5.268125268125268e-06,
      "loss": 0.8306,
      "step": 20430
    },
    {
      "epoch": 26.006126126126127,
      "grad_norm": 0.04654620960354805,
      "learning_rate": 5.265265265265266e-06,
      "loss": 0.0011,
      "step": 20440
    },
    {
      "epoch": 26.006383526383527,
      "grad_norm": 1473.2542724609375,
      "learning_rate": 5.2624052624052625e-06,
      "loss": 0.1808,
      "step": 20450
    },
    {
      "epoch": 26.006640926640927,
      "grad_norm": 0.08300590515136719,
      "learning_rate": 5.25954525954526e-06,
      "loss": 0.9119,
      "step": 20460
    },
    {
      "epoch": 26.006898326898327,
      "grad_norm": 0.047874584794044495,
      "learning_rate": 5.256685256685256e-06,
      "loss": 0.0016,
      "step": 20470
    },
    {
      "epoch": 26.007155727155727,
      "grad_norm": 0.4152180254459381,
      "learning_rate": 5.2538252538252545e-06,
      "loss": 1.0339,
      "step": 20480
    },
    {
      "epoch": 26.007413127413127,
      "grad_norm": 0.000853995734360069,
      "learning_rate": 5.250965250965251e-06,
      "loss": 0.9289,
      "step": 20490
    },
    {
      "epoch": 26.007670527670527,
      "grad_norm": 0.08005805313587189,
      "learning_rate": 5.248105248105248e-06,
      "loss": 1.8532,
      "step": 20500
    },
    {
      "epoch": 26.007927927927927,
      "grad_norm": 1.376071810722351,
      "learning_rate": 5.245245245245245e-06,
      "loss": 0.6187,
      "step": 20510
    },
    {
      "epoch": 26.008185328185327,
      "grad_norm": 4.983120441436768,
      "learning_rate": 5.242385242385243e-06,
      "loss": 0.5194,
      "step": 20520
    },
    {
      "epoch": 26.008442728442727,
      "grad_norm": 0.01302572712302208,
      "learning_rate": 5.23952523952524e-06,
      "loss": 0.4392,
      "step": 20530
    },
    {
      "epoch": 26.008700128700127,
      "grad_norm": 0.05677972361445427,
      "learning_rate": 5.236665236665237e-06,
      "loss": 0.4977,
      "step": 20540
    },
    {
      "epoch": 26.008957528957527,
      "grad_norm": 0.18432343006134033,
      "learning_rate": 5.233805233805235e-06,
      "loss": 1.5045,
      "step": 20550
    },
    {
      "epoch": 26.00921492921493,
      "grad_norm": 0.0020320096518844366,
      "learning_rate": 5.230945230945231e-06,
      "loss": 0.0009,
      "step": 20560
    },
    {
      "epoch": 26.00947232947233,
      "grad_norm": 0.07179369032382965,
      "learning_rate": 5.228085228085229e-06,
      "loss": 1.7177,
      "step": 20570
    },
    {
      "epoch": 26.00972972972973,
      "grad_norm": 0.043726976960897446,
      "learning_rate": 5.225225225225226e-06,
      "loss": 0.9099,
      "step": 20580
    },
    {
      "epoch": 26.00998712998713,
      "grad_norm": 0.019464250653982162,
      "learning_rate": 5.222365222365223e-06,
      "loss": 1.0313,
      "step": 20590
    },
    {
      "epoch": 26.01024453024453,
      "grad_norm": 0.3122895061969757,
      "learning_rate": 5.21950521950522e-06,
      "loss": 1.1684,
      "step": 20600
    },
    {
      "epoch": 26.01050193050193,
      "grad_norm": 0.0006997102173045278,
      "learning_rate": 5.216645216645218e-06,
      "loss": 0.0393,
      "step": 20610
    },
    {
      "epoch": 26.01075933075933,
      "grad_norm": 39.65142059326172,
      "learning_rate": 5.213785213785214e-06,
      "loss": 0.5572,
      "step": 20620
    },
    {
      "epoch": 26.01101673101673,
      "grad_norm": 0.11646867543458939,
      "learning_rate": 5.210925210925212e-06,
      "loss": 0.0028,
      "step": 20630
    },
    {
      "epoch": 26.01127413127413,
      "grad_norm": 0.06890599429607391,
      "learning_rate": 5.208065208065208e-06,
      "loss": 0.0033,
      "step": 20640
    },
    {
      "epoch": 26.01153153153153,
      "grad_norm": 0.08268891274929047,
      "learning_rate": 5.205205205205206e-06,
      "loss": 0.5089,
      "step": 20650
    },
    {
      "epoch": 26.01178893178893,
      "grad_norm": 0.03577331081032753,
      "learning_rate": 5.202345202345203e-06,
      "loss": 1.2679,
      "step": 20660
    },
    {
      "epoch": 26.01204633204633,
      "grad_norm": 0.07769101113080978,
      "learning_rate": 5.1994851994852e-06,
      "loss": 0.0016,
      "step": 20670
    },
    {
      "epoch": 26.01230373230373,
      "grad_norm": 33.56912612915039,
      "learning_rate": 5.1966251966251966e-06,
      "loss": 1.2153,
      "step": 20680
    },
    {
      "epoch": 26.01256113256113,
      "grad_norm": 0.003670428181067109,
      "learning_rate": 5.193765193765195e-06,
      "loss": 0.9139,
      "step": 20690
    },
    {
      "epoch": 26.01281853281853,
      "grad_norm": 713.4417724609375,
      "learning_rate": 5.190905190905191e-06,
      "loss": 0.4005,
      "step": 20700
    },
    {
      "epoch": 26.013075933075932,
      "grad_norm": 0.0020978457760065794,
      "learning_rate": 5.1880451880451885e-06,
      "loss": 0.4827,
      "step": 20710
    },
    {
      "epoch": 26.013333333333332,
      "grad_norm": 0.0007727963966317475,
      "learning_rate": 5.185185185185185e-06,
      "loss": 0.4697,
      "step": 20720
    },
    {
      "epoch": 26.013590733590732,
      "grad_norm": 0.014731266535818577,
      "learning_rate": 5.182325182325183e-06,
      "loss": 0.4416,
      "step": 20730
    },
    {
      "epoch": 26.013848133848136,
      "grad_norm": 0.051897432655096054,
      "learning_rate": 5.17946517946518e-06,
      "loss": 0.0006,
      "step": 20740
    },
    {
      "epoch": 26.014105534105536,
      "grad_norm": 0.0008702686754986644,
      "learning_rate": 5.176605176605177e-06,
      "loss": 2.079,
      "step": 20750
    },
    {
      "epoch": 26.014362934362936,
      "grad_norm": 0.18641412258148193,
      "learning_rate": 5.173745173745173e-06,
      "loss": 0.0022,
      "step": 20760
    },
    {
      "epoch": 26.014620334620336,
      "grad_norm": 0.0020233727991580963,
      "learning_rate": 5.1708851708851716e-06,
      "loss": 0.0086,
      "step": 20770
    },
    {
      "epoch": 26.014877734877736,
      "grad_norm": 0.013192682527005672,
      "learning_rate": 5.168025168025168e-06,
      "loss": 0.9466,
      "step": 20780
    },
    {
      "epoch": 26.015135135135136,
      "grad_norm": 1.5512306690216064,
      "learning_rate": 5.165165165165165e-06,
      "loss": 0.0028,
      "step": 20790
    },
    {
      "epoch": 26.015392535392536,
      "grad_norm": 0.25116217136383057,
      "learning_rate": 5.162305162305163e-06,
      "loss": 0.0018,
      "step": 20800
    },
    {
      "epoch": 26.015649935649936,
      "grad_norm": 0.003960348200052977,
      "learning_rate": 5.15944515944516e-06,
      "loss": 0.0007,
      "step": 20810
    },
    {
      "epoch": 26.015907335907336,
      "grad_norm": 0.002162356860935688,
      "learning_rate": 5.1565851565851565e-06,
      "loss": 0.002,
      "step": 20820
    },
    {
      "epoch": 26.016164736164736,
      "grad_norm": 0.07155408710241318,
      "learning_rate": 5.153725153725155e-06,
      "loss": 0.027,
      "step": 20830
    },
    {
      "epoch": 26.016422136422136,
      "grad_norm": 0.10129252076148987,
      "learning_rate": 5.150865150865151e-06,
      "loss": 0.0012,
      "step": 20840
    },
    {
      "epoch": 26.016679536679536,
      "grad_norm": 0.004315822850912809,
      "learning_rate": 5.148005148005148e-06,
      "loss": 0.0052,
      "step": 20850
    },
    {
      "epoch": 26.016936936936936,
      "grad_norm": 0.001369721838273108,
      "learning_rate": 5.145145145145145e-06,
      "loss": 0.0012,
      "step": 20860
    },
    {
      "epoch": 26.017194337194336,
      "grad_norm": 0.006657337304204702,
      "learning_rate": 5.142285142285143e-06,
      "loss": 0.6222,
      "step": 20870
    },
    {
      "epoch": 26.017451737451736,
      "grad_norm": 35.71303939819336,
      "learning_rate": 5.1394251394251395e-06,
      "loss": 1.2577,
      "step": 20880
    },
    {
      "epoch": 26.017709137709137,
      "grad_norm": 0.0017888675210997462,
      "learning_rate": 5.136565136565137e-06,
      "loss": 0.0088,
      "step": 20890
    },
    {
      "epoch": 26.017966537966537,
      "grad_norm": 0.025619618594646454,
      "learning_rate": 5.133705133705133e-06,
      "loss": 1.1787,
      "step": 20900
    },
    {
      "epoch": 26.018223938223937,
      "grad_norm": 557.7401123046875,
      "learning_rate": 5.1308451308451314e-06,
      "loss": 0.8761,
      "step": 20910
    },
    {
      "epoch": 26.018481338481337,
      "grad_norm": 0.19152088463306427,
      "learning_rate": 5.127985127985129e-06,
      "loss": 0.001,
      "step": 20920
    },
    {
      "epoch": 26.01873873873874,
      "grad_norm": 47.91631317138672,
      "learning_rate": 5.125125125125125e-06,
      "loss": 0.6359,
      "step": 20930
    },
    {
      "epoch": 26.01899613899614,
      "grad_norm": 0.0006033479585312307,
      "learning_rate": 5.122265122265123e-06,
      "loss": 0.4703,
      "step": 20940
    },
    {
      "epoch": 26.01925353925354,
      "grad_norm": 0.07352429628372192,
      "learning_rate": 5.11940511940512e-06,
      "loss": 0.0185,
      "step": 20950
    },
    {
      "epoch": 26.01951093951094,
      "grad_norm": 0.08161481469869614,
      "learning_rate": 5.116545116545117e-06,
      "loss": 1.2037,
      "step": 20960
    },
    {
      "epoch": 26.01976833976834,
      "grad_norm": 0.12737224996089935,
      "learning_rate": 5.113685113685114e-06,
      "loss": 0.3564,
      "step": 20970
    },
    {
      "epoch": 26.02,
      "eval_accuracy": 0.8695652173913043,
      "eval_loss": 0.7219700813293457,
      "eval_runtime": 13.359,
      "eval_samples_per_second": 3.443,
      "eval_steps_per_second": 3.443,
      "step": 20979
    },
    {
      "epoch": 27.00002574002574,
      "grad_norm": 0.4492325186729431,
      "learning_rate": 5.110825110825112e-06,
      "loss": 0.0428,
      "step": 20980
    },
    {
      "epoch": 27.00028314028314,
      "grad_norm": 0.10700222104787827,
      "learning_rate": 5.107965107965108e-06,
      "loss": 0.0011,
      "step": 20990
    },
    {
      "epoch": 27.00054054054054,
      "grad_norm": 0.4348405599594116,
      "learning_rate": 5.105105105105106e-06,
      "loss": 0.2933,
      "step": 21000
    },
    {
      "epoch": 27.00079794079794,
      "grad_norm": 0.0012466449989005923,
      "learning_rate": 5.102245102245103e-06,
      "loss": 0.0769,
      "step": 21010
    },
    {
      "epoch": 27.00105534105534,
      "grad_norm": 0.041840698570013046,
      "learning_rate": 5.0993850993851e-06,
      "loss": 1.6616,
      "step": 21020
    },
    {
      "epoch": 27.00131274131274,
      "grad_norm": 0.002618017140775919,
      "learning_rate": 5.096525096525097e-06,
      "loss": 0.0011,
      "step": 21030
    },
    {
      "epoch": 27.00157014157014,
      "grad_norm": 0.13534986972808838,
      "learning_rate": 5.093665093665094e-06,
      "loss": 0.0007,
      "step": 21040
    },
    {
      "epoch": 27.00182754182754,
      "grad_norm": 37.55997848510742,
      "learning_rate": 5.090805090805091e-06,
      "loss": 0.6014,
      "step": 21050
    },
    {
      "epoch": 27.00208494208494,
      "grad_norm": 348.0312194824219,
      "learning_rate": 5.087945087945089e-06,
      "loss": 0.8582,
      "step": 21060
    },
    {
      "epoch": 27.00234234234234,
      "grad_norm": 0.5563425421714783,
      "learning_rate": 5.085085085085085e-06,
      "loss": 0.726,
      "step": 21070
    },
    {
      "epoch": 27.00259974259974,
      "grad_norm": 28.312299728393555,
      "learning_rate": 5.082225082225083e-06,
      "loss": 1.2086,
      "step": 21080
    },
    {
      "epoch": 27.002857142857142,
      "grad_norm": 0.06256058067083359,
      "learning_rate": 5.07936507936508e-06,
      "loss": 0.4858,
      "step": 21090
    },
    {
      "epoch": 27.003114543114542,
      "grad_norm": 0.005022318102419376,
      "learning_rate": 5.076505076505077e-06,
      "loss": 0.0021,
      "step": 21100
    },
    {
      "epoch": 27.003371943371942,
      "grad_norm": 0.212180495262146,
      "learning_rate": 5.0736450736450735e-06,
      "loss": 1.107,
      "step": 21110
    },
    {
      "epoch": 27.003629343629342,
      "grad_norm": 0.0004522952949628234,
      "learning_rate": 5.070785070785072e-06,
      "loss": 1.1168,
      "step": 21120
    },
    {
      "epoch": 27.003886743886746,
      "grad_norm": 0.1687791496515274,
      "learning_rate": 5.067925067925068e-06,
      "loss": 0.5336,
      "step": 21130
    },
    {
      "epoch": 27.004144144144146,
      "grad_norm": 0.008968188427388668,
      "learning_rate": 5.0650650650650655e-06,
      "loss": 0.4525,
      "step": 21140
    },
    {
      "epoch": 27.004401544401546,
      "grad_norm": 0.14755140244960785,
      "learning_rate": 5.062205062205062e-06,
      "loss": 0.0031,
      "step": 21150
    },
    {
      "epoch": 27.004658944658946,
      "grad_norm": 36.4864501953125,
      "learning_rate": 5.05934505934506e-06,
      "loss": 0.4908,
      "step": 21160
    },
    {
      "epoch": 27.004916344916346,
      "grad_norm": 0.00807279348373413,
      "learning_rate": 5.056485056485057e-06,
      "loss": 0.7936,
      "step": 21170
    },
    {
      "epoch": 27.005173745173746,
      "grad_norm": 0.4851839244365692,
      "learning_rate": 5.053625053625054e-06,
      "loss": 0.6647,
      "step": 21180
    },
    {
      "epoch": 27.005431145431146,
      "grad_norm": 27.190128326416016,
      "learning_rate": 5.05076505076505e-06,
      "loss": 1.577,
      "step": 21190
    },
    {
      "epoch": 27.005688545688546,
      "grad_norm": 0.0018694992177188396,
      "learning_rate": 5.0479050479050485e-06,
      "loss": 1.2588,
      "step": 21200
    },
    {
      "epoch": 27.005945945945946,
      "grad_norm": 0.05228858068585396,
      "learning_rate": 5.045045045045045e-06,
      "loss": 0.4246,
      "step": 21210
    },
    {
      "epoch": 27.006203346203346,
      "grad_norm": 0.20007288455963135,
      "learning_rate": 5.042185042185042e-06,
      "loss": 0.01,
      "step": 21220
    },
    {
      "epoch": 27.006460746460746,
      "grad_norm": 305.2531433105469,
      "learning_rate": 5.03932503932504e-06,
      "loss": 0.2956,
      "step": 21230
    },
    {
      "epoch": 27.006718146718146,
      "grad_norm": 0.00089081289479509,
      "learning_rate": 5.036465036465037e-06,
      "loss": 0.5486,
      "step": 21240
    },
    {
      "epoch": 27.006975546975546,
      "grad_norm": 0.48225823044776917,
      "learning_rate": 5.0336050336050334e-06,
      "loss": 0.4939,
      "step": 21250
    },
    {
      "epoch": 27.007232947232946,
      "grad_norm": 0.19505609571933746,
      "learning_rate": 5.030745030745032e-06,
      "loss": 0.9111,
      "step": 21260
    },
    {
      "epoch": 27.007490347490346,
      "grad_norm": 0.0040650321170687675,
      "learning_rate": 5.027885027885028e-06,
      "loss": 0.146,
      "step": 21270
    },
    {
      "epoch": 27.007747747747747,
      "grad_norm": 0.986337423324585,
      "learning_rate": 5.025025025025025e-06,
      "loss": 0.0111,
      "step": 21280
    },
    {
      "epoch": 27.008005148005147,
      "grad_norm": 0.1955615133047104,
      "learning_rate": 5.022165022165023e-06,
      "loss": 0.3612,
      "step": 21290
    },
    {
      "epoch": 27.008262548262547,
      "grad_norm": 0.0011161335278302431,
      "learning_rate": 5.01930501930502e-06,
      "loss": 0.0009,
      "step": 21300
    },
    {
      "epoch": 27.008519948519947,
      "grad_norm": 0.02377697266638279,
      "learning_rate": 5.016445016445017e-06,
      "loss": 0.5797,
      "step": 21310
    },
    {
      "epoch": 27.00877734877735,
      "grad_norm": 447.38330078125,
      "learning_rate": 5.013585013585014e-06,
      "loss": 0.8126,
      "step": 21320
    },
    {
      "epoch": 27.00903474903475,
      "grad_norm": 0.01540659461170435,
      "learning_rate": 5.010725010725012e-06,
      "loss": 0.5582,
      "step": 21330
    },
    {
      "epoch": 27.00929214929215,
      "grad_norm": 0.0029602760914713144,
      "learning_rate": 5.0078650078650084e-06,
      "loss": 0.4369,
      "step": 21340
    },
    {
      "epoch": 27.00954954954955,
      "grad_norm": 310.1184387207031,
      "learning_rate": 5.005005005005006e-06,
      "loss": 0.5738,
      "step": 21350
    },
    {
      "epoch": 27.00980694980695,
      "grad_norm": 0.08122654259204865,
      "learning_rate": 5.002145002145002e-06,
      "loss": 0.0019,
      "step": 21360
    },
    {
      "epoch": 27.01006435006435,
      "grad_norm": 0.0014425546396523714,
      "learning_rate": 4.9992849992849995e-06,
      "loss": 1.5169,
      "step": 21370
    },
    {
      "epoch": 27.01032175032175,
      "grad_norm": 0.01068554725497961,
      "learning_rate": 4.996424996424997e-06,
      "loss": 0.0114,
      "step": 21380
    },
    {
      "epoch": 27.01057915057915,
      "grad_norm": 0.001254943199455738,
      "learning_rate": 4.993564993564994e-06,
      "loss": 1.7242,
      "step": 21390
    },
    {
      "epoch": 27.01083655083655,
      "grad_norm": 0.010038587264716625,
      "learning_rate": 4.990704990704991e-06,
      "loss": 0.0024,
      "step": 21400
    },
    {
      "epoch": 27.01109395109395,
      "grad_norm": 0.004530706908553839,
      "learning_rate": 4.987844987844988e-06,
      "loss": 0.508,
      "step": 21410
    },
    {
      "epoch": 27.01135135135135,
      "grad_norm": 0.10551851242780685,
      "learning_rate": 4.984984984984985e-06,
      "loss": 0.5313,
      "step": 21420
    },
    {
      "epoch": 27.01160875160875,
      "grad_norm": 0.11509599536657333,
      "learning_rate": 4.982124982124983e-06,
      "loss": 0.3598,
      "step": 21430
    },
    {
      "epoch": 27.01186615186615,
      "grad_norm": 0.0032966965809464455,
      "learning_rate": 4.979264979264979e-06,
      "loss": 0.6849,
      "step": 21440
    },
    {
      "epoch": 27.01212355212355,
      "grad_norm": 0.0011607161723077297,
      "learning_rate": 4.976404976404976e-06,
      "loss": 0.2958,
      "step": 21450
    },
    {
      "epoch": 27.01238095238095,
      "grad_norm": 0.11378136277198792,
      "learning_rate": 4.973544973544974e-06,
      "loss": 0.0032,
      "step": 21460
    },
    {
      "epoch": 27.01263835263835,
      "grad_norm": 0.0004367120563983917,
      "learning_rate": 4.970684970684971e-06,
      "loss": 0.0012,
      "step": 21470
    },
    {
      "epoch": 27.01289575289575,
      "grad_norm": 0.003959926310926676,
      "learning_rate": 4.967824967824968e-06,
      "loss": 1.2426,
      "step": 21480
    },
    {
      "epoch": 27.01315315315315,
      "grad_norm": 0.3200651705265045,
      "learning_rate": 4.964964964964966e-06,
      "loss": 0.8776,
      "step": 21490
    },
    {
      "epoch": 27.013410553410555,
      "grad_norm": 28.075603485107422,
      "learning_rate": 4.962104962104963e-06,
      "loss": 0.4647,
      "step": 21500
    },
    {
      "epoch": 27.013667953667955,
      "grad_norm": 20.377717971801758,
      "learning_rate": 4.95924495924496e-06,
      "loss": 1.3318,
      "step": 21510
    },
    {
      "epoch": 27.013925353925355,
      "grad_norm": 0.09059357643127441,
      "learning_rate": 4.956384956384957e-06,
      "loss": 0.0026,
      "step": 21520
    },
    {
      "epoch": 27.014182754182755,
      "grad_norm": 0.9779305458068848,
      "learning_rate": 4.953524953524954e-06,
      "loss": 0.0012,
      "step": 21530
    },
    {
      "epoch": 27.014440154440155,
      "grad_norm": 0.045352060347795486,
      "learning_rate": 4.950664950664951e-06,
      "loss": 0.6597,
      "step": 21540
    },
    {
      "epoch": 27.014697554697555,
      "grad_norm": 0.03662479668855667,
      "learning_rate": 4.947804947804949e-06,
      "loss": 0.9173,
      "step": 21550
    },
    {
      "epoch": 27.014954954954955,
      "grad_norm": 0.10112226754426956,
      "learning_rate": 4.944944944944945e-06,
      "loss": 0.6186,
      "step": 21560
    },
    {
      "epoch": 27.015212355212356,
      "grad_norm": 0.03256876394152641,
      "learning_rate": 4.9420849420849425e-06,
      "loss": 1.1788,
      "step": 21570
    },
    {
      "epoch": 27.015469755469756,
      "grad_norm": 0.2824130654335022,
      "learning_rate": 4.93922493922494e-06,
      "loss": 1.1216,
      "step": 21580
    },
    {
      "epoch": 27.015727155727156,
      "grad_norm": 0.0007668504840694368,
      "learning_rate": 4.936364936364937e-06,
      "loss": 0.4967,
      "step": 21590
    },
    {
      "epoch": 27.015984555984556,
      "grad_norm": 0.17322510480880737,
      "learning_rate": 4.9335049335049336e-06,
      "loss": 0.6486,
      "step": 21600
    },
    {
      "epoch": 27.016241956241956,
      "grad_norm": 0.14373072981834412,
      "learning_rate": 4.930644930644931e-06,
      "loss": 1.4562,
      "step": 21610
    },
    {
      "epoch": 27.016499356499356,
      "grad_norm": 0.002229807898402214,
      "learning_rate": 4.927784927784928e-06,
      "loss": 0.5973,
      "step": 21620
    },
    {
      "epoch": 27.016756756756756,
      "grad_norm": 1.7468568086624146,
      "learning_rate": 4.9249249249249255e-06,
      "loss": 0.0018,
      "step": 21630
    },
    {
      "epoch": 27.017014157014156,
      "grad_norm": 1.5412856340408325,
      "learning_rate": 4.922064922064923e-06,
      "loss": 0.5743,
      "step": 21640
    },
    {
      "epoch": 27.017271557271556,
      "grad_norm": 9.415769577026367,
      "learning_rate": 4.919204919204919e-06,
      "loss": 0.4402,
      "step": 21650
    },
    {
      "epoch": 27.017528957528956,
      "grad_norm": 0.00046175954048521817,
      "learning_rate": 4.916344916344917e-06,
      "loss": 0.5361,
      "step": 21660
    },
    {
      "epoch": 27.017786357786356,
      "grad_norm": 6.175512313842773,
      "learning_rate": 4.913484913484914e-06,
      "loss": 0.007,
      "step": 21670
    },
    {
      "epoch": 27.01804375804376,
      "grad_norm": 0.09624309092760086,
      "learning_rate": 4.910624910624911e-06,
      "loss": 0.473,
      "step": 21680
    },
    {
      "epoch": 27.01830115830116,
      "grad_norm": 0.0006763459532521665,
      "learning_rate": 4.907764907764908e-06,
      "loss": 0.0005,
      "step": 21690
    },
    {
      "epoch": 27.01855855855856,
      "grad_norm": 0.00037406483897939324,
      "learning_rate": 4.904904904904905e-06,
      "loss": 0.8364,
      "step": 21700
    },
    {
      "epoch": 27.01881595881596,
      "grad_norm": 0.23709970712661743,
      "learning_rate": 4.902044902044902e-06,
      "loss": 0.5329,
      "step": 21710
    },
    {
      "epoch": 27.01907335907336,
      "grad_norm": 0.0009810479823499918,
      "learning_rate": 4.8991848991849e-06,
      "loss": 1.3556,
      "step": 21720
    },
    {
      "epoch": 27.01933075933076,
      "grad_norm": 0.02598871849477291,
      "learning_rate": 4.896324896324897e-06,
      "loss": 0.5565,
      "step": 21730
    },
    {
      "epoch": 27.01958815958816,
      "grad_norm": 112.14535522460938,
      "learning_rate": 4.8934648934648935e-06,
      "loss": 1.6131,
      "step": 21740
    },
    {
      "epoch": 27.01984555984556,
      "grad_norm": 0.008748788386583328,
      "learning_rate": 4.890604890604891e-06,
      "loss": 0.0017,
      "step": 21750
    },
    {
      "epoch": 27.02,
      "eval_accuracy": 0.782608695652174,
      "eval_loss": 1.2440866231918335,
      "eval_runtime": 14.5083,
      "eval_samples_per_second": 3.171,
      "eval_steps_per_second": 3.171,
      "step": 21756
    },
    {
      "epoch": 28.00010296010296,
      "grad_norm": 0.0006102527840994298,
      "learning_rate": 4.887744887744888e-06,
      "loss": 0.4715,
      "step": 21760
    },
    {
      "epoch": 28.00036036036036,
      "grad_norm": 0.0769384428858757,
      "learning_rate": 4.884884884884885e-06,
      "loss": 0.002,
      "step": 21770
    },
    {
      "epoch": 28.00061776061776,
      "grad_norm": 0.0811397060751915,
      "learning_rate": 4.882024882024882e-06,
      "loss": 0.4448,
      "step": 21780
    },
    {
      "epoch": 28.00087516087516,
      "grad_norm": 0.002516659675166011,
      "learning_rate": 4.879164879164879e-06,
      "loss": 1.1136,
      "step": 21790
    },
    {
      "epoch": 28.00113256113256,
      "grad_norm": 0.06139345467090607,
      "learning_rate": 4.8763048763048765e-06,
      "loss": 0.0014,
      "step": 21800
    },
    {
      "epoch": 28.00138996138996,
      "grad_norm": 35.70652770996094,
      "learning_rate": 4.873444873444874e-06,
      "loss": 1.0875,
      "step": 21810
    },
    {
      "epoch": 28.00164736164736,
      "grad_norm": 128.56268310546875,
      "learning_rate": 4.870584870584871e-06,
      "loss": 0.4463,
      "step": 21820
    },
    {
      "epoch": 28.00190476190476,
      "grad_norm": 0.00042608179501257837,
      "learning_rate": 4.867724867724868e-06,
      "loss": 0.4507,
      "step": 21830
    },
    {
      "epoch": 28.00216216216216,
      "grad_norm": 0.05981049686670303,
      "learning_rate": 4.864864864864866e-06,
      "loss": 1.251,
      "step": 21840
    },
    {
      "epoch": 28.00241956241956,
      "grad_norm": 0.029568631201982498,
      "learning_rate": 4.862004862004862e-06,
      "loss": 0.498,
      "step": 21850
    },
    {
      "epoch": 28.00267696267696,
      "grad_norm": 41.19403839111328,
      "learning_rate": 4.8591448591448596e-06,
      "loss": 0.8085,
      "step": 21860
    },
    {
      "epoch": 28.00293436293436,
      "grad_norm": 0.035941943526268005,
      "learning_rate": 4.856284856284857e-06,
      "loss": 0.0015,
      "step": 21870
    },
    {
      "epoch": 28.00319176319176,
      "grad_norm": 0.01107192412018776,
      "learning_rate": 4.853424853424854e-06,
      "loss": 0.8204,
      "step": 21880
    },
    {
      "epoch": 28.003449163449165,
      "grad_norm": 0.0007803246262483299,
      "learning_rate": 4.8505648505648515e-06,
      "loss": 0.0033,
      "step": 21890
    },
    {
      "epoch": 28.003706563706565,
      "grad_norm": 39.610023498535156,
      "learning_rate": 4.847704847704848e-06,
      "loss": 0.5976,
      "step": 21900
    },
    {
      "epoch": 28.003963963963965,
      "grad_norm": 0.4089675843715668,
      "learning_rate": 4.844844844844845e-06,
      "loss": 0.0006,
      "step": 21910
    },
    {
      "epoch": 28.004221364221365,
      "grad_norm": 0.0013353163376450539,
      "learning_rate": 4.841984841984843e-06,
      "loss": 0.0025,
      "step": 21920
    },
    {
      "epoch": 28.004478764478765,
      "grad_norm": 0.0005668631056323647,
      "learning_rate": 4.83912483912484e-06,
      "loss": 1.0795,
      "step": 21930
    },
    {
      "epoch": 28.004736164736165,
      "grad_norm": 0.0004358472360763699,
      "learning_rate": 4.836264836264836e-06,
      "loss": 0.8763,
      "step": 21940
    },
    {
      "epoch": 28.004993564993566,
      "grad_norm": 0.34297168254852295,
      "learning_rate": 4.833404833404834e-06,
      "loss": 0.5577,
      "step": 21950
    },
    {
      "epoch": 28.005250965250966,
      "grad_norm": 0.10034578293561935,
      "learning_rate": 4.830544830544831e-06,
      "loss": 2.3757,
      "step": 21960
    },
    {
      "epoch": 28.005508365508366,
      "grad_norm": 0.12659744918346405,
      "learning_rate": 4.827684827684828e-06,
      "loss": 0.5404,
      "step": 21970
    },
    {
      "epoch": 28.005765765765766,
      "grad_norm": 0.06474514305591583,
      "learning_rate": 4.824824824824826e-06,
      "loss": 0.4778,
      "step": 21980
    },
    {
      "epoch": 28.006023166023166,
      "grad_norm": 0.03867054730653763,
      "learning_rate": 4.821964821964822e-06,
      "loss": 0.5697,
      "step": 21990
    },
    {
      "epoch": 28.006280566280566,
      "grad_norm": 0.003705323673784733,
      "learning_rate": 4.8191048191048194e-06,
      "loss": 0.53,
      "step": 22000
    },
    {
      "epoch": 28.006537966537966,
      "grad_norm": 0.014577334746718407,
      "learning_rate": 4.816244816244817e-06,
      "loss": 0.0034,
      "step": 22010
    },
    {
      "epoch": 28.006795366795366,
      "grad_norm": 0.07273665815591812,
      "learning_rate": 4.813384813384814e-06,
      "loss": 0.5665,
      "step": 22020
    },
    {
      "epoch": 28.007052767052766,
      "grad_norm": 0.14401213824748993,
      "learning_rate": 4.8105248105248105e-06,
      "loss": 0.0013,
      "step": 22030
    },
    {
      "epoch": 28.007310167310166,
      "grad_norm": 0.0013334174873307347,
      "learning_rate": 4.807664807664808e-06,
      "loss": 0.2891,
      "step": 22040
    },
    {
      "epoch": 28.007567567567566,
      "grad_norm": 0.02023886889219284,
      "learning_rate": 4.804804804804805e-06,
      "loss": 0.0011,
      "step": 22050
    },
    {
      "epoch": 28.007824967824966,
      "grad_norm": 0.14436078071594238,
      "learning_rate": 4.8019448019448025e-06,
      "loss": 0.0016,
      "step": 22060
    },
    {
      "epoch": 28.00808236808237,
      "grad_norm": 0.03297024220228195,
      "learning_rate": 4.7990847990848e-06,
      "loss": 0.5523,
      "step": 22070
    },
    {
      "epoch": 28.00833976833977,
      "grad_norm": 507.6208801269531,
      "learning_rate": 4.796224796224796e-06,
      "loss": 0.4174,
      "step": 22080
    },
    {
      "epoch": 28.00859716859717,
      "grad_norm": 0.016587285324931145,
      "learning_rate": 4.793364793364794e-06,
      "loss": 0.6593,
      "step": 22090
    },
    {
      "epoch": 28.00885456885457,
      "grad_norm": 0.0026779465842992067,
      "learning_rate": 4.790504790504791e-06,
      "loss": 0.6397,
      "step": 22100
    },
    {
      "epoch": 28.00911196911197,
      "grad_norm": 0.14133141934871674,
      "learning_rate": 4.787644787644788e-06,
      "loss": 0.1802,
      "step": 22110
    },
    {
      "epoch": 28.00936936936937,
      "grad_norm": 0.16812646389007568,
      "learning_rate": 4.784784784784785e-06,
      "loss": 0.4667,
      "step": 22120
    },
    {
      "epoch": 28.00962676962677,
      "grad_norm": 0.08822160959243774,
      "learning_rate": 4.781924781924782e-06,
      "loss": 1.18,
      "step": 22130
    },
    {
      "epoch": 28.00988416988417,
      "grad_norm": 0.17704392969608307,
      "learning_rate": 4.779064779064779e-06,
      "loss": 0.0016,
      "step": 22140
    },
    {
      "epoch": 28.01014157014157,
      "grad_norm": 0.4867551028728485,
      "learning_rate": 4.776204776204777e-06,
      "loss": 1.0967,
      "step": 22150
    },
    {
      "epoch": 28.01039897039897,
      "grad_norm": 0.54176926612854,
      "learning_rate": 4.773344773344773e-06,
      "loss": 0.6071,
      "step": 22160
    },
    {
      "epoch": 28.01065637065637,
      "grad_norm": 0.0038197352550923824,
      "learning_rate": 4.7704847704847704e-06,
      "loss": 0.442,
      "step": 22170
    },
    {
      "epoch": 28.01091377091377,
      "grad_norm": 0.08110248297452927,
      "learning_rate": 4.767624767624768e-06,
      "loss": 1.2637,
      "step": 22180
    },
    {
      "epoch": 28.01117117117117,
      "grad_norm": 0.12915049493312836,
      "learning_rate": 4.764764764764765e-06,
      "loss": 0.3288,
      "step": 22190
    },
    {
      "epoch": 28.01142857142857,
      "grad_norm": 50.06999588012695,
      "learning_rate": 4.761904761904762e-06,
      "loss": 0.3762,
      "step": 22200
    },
    {
      "epoch": 28.01168597168597,
      "grad_norm": 0.0005307564279064536,
      "learning_rate": 4.75904475904476e-06,
      "loss": 0.1265,
      "step": 22210
    },
    {
      "epoch": 28.01194337194337,
      "grad_norm": 0.03496575355529785,
      "learning_rate": 4.756184756184757e-06,
      "loss": 0.0023,
      "step": 22220
    },
    {
      "epoch": 28.01220077220077,
      "grad_norm": 0.0009349982719868422,
      "learning_rate": 4.753324753324754e-06,
      "loss": 0.0194,
      "step": 22230
    },
    {
      "epoch": 28.01245817245817,
      "grad_norm": 0.0008056398364715278,
      "learning_rate": 4.750464750464751e-06,
      "loss": 0.0018,
      "step": 22240
    },
    {
      "epoch": 28.01271557271557,
      "grad_norm": 0.06082630529999733,
      "learning_rate": 4.747604747604748e-06,
      "loss": 0.727,
      "step": 22250
    },
    {
      "epoch": 28.012972972972975,
      "grad_norm": 0.11555928736925125,
      "learning_rate": 4.7447447447447454e-06,
      "loss": 0.0015,
      "step": 22260
    },
    {
      "epoch": 28.013230373230375,
      "grad_norm": 0.012335697188973427,
      "learning_rate": 4.741884741884743e-06,
      "loss": 0.3981,
      "step": 22270
    },
    {
      "epoch": 28.013487773487775,
      "grad_norm": 0.00791518110781908,
      "learning_rate": 4.739024739024739e-06,
      "loss": 0.5654,
      "step": 22280
    },
    {
      "epoch": 28.013745173745175,
      "grad_norm": 708.4443359375,
      "learning_rate": 4.7361647361647365e-06,
      "loss": 0.4047,
      "step": 22290
    },
    {
      "epoch": 28.014002574002575,
      "grad_norm": 0.008482186123728752,
      "learning_rate": 4.733304733304734e-06,
      "loss": 0.0004,
      "step": 22300
    },
    {
      "epoch": 28.014259974259975,
      "grad_norm": 0.08238717913627625,
      "learning_rate": 4.730444730444731e-06,
      "loss": 0.0009,
      "step": 22310
    },
    {
      "epoch": 28.014517374517375,
      "grad_norm": 0.00034206442069262266,
      "learning_rate": 4.7275847275847285e-06,
      "loss": 0.196,
      "step": 22320
    },
    {
      "epoch": 28.014774774774775,
      "grad_norm": 0.02654080092906952,
      "learning_rate": 4.724724724724725e-06,
      "loss": 0.0011,
      "step": 22330
    },
    {
      "epoch": 28.015032175032175,
      "grad_norm": 42.99850845336914,
      "learning_rate": 4.721864721864722e-06,
      "loss": 1.2377,
      "step": 22340
    },
    {
      "epoch": 28.015289575289575,
      "grad_norm": 0.653175413608551,
      "learning_rate": 4.71900471900472e-06,
      "loss": 1.3387,
      "step": 22350
    },
    {
      "epoch": 28.015546975546975,
      "grad_norm": 0.40929099917411804,
      "learning_rate": 4.716144716144717e-06,
      "loss": 0.001,
      "step": 22360
    },
    {
      "epoch": 28.015804375804375,
      "grad_norm": 0.17331182956695557,
      "learning_rate": 4.713284713284713e-06,
      "loss": 0.0152,
      "step": 22370
    },
    {
      "epoch": 28.016061776061775,
      "grad_norm": 0.05048976093530655,
      "learning_rate": 4.710424710424711e-06,
      "loss": 1.2726,
      "step": 22380
    },
    {
      "epoch": 28.016319176319175,
      "grad_norm": 0.24569039046764374,
      "learning_rate": 4.707564707564708e-06,
      "loss": 0.6923,
      "step": 22390
    },
    {
      "epoch": 28.016576576576576,
      "grad_norm": 0.011856620199978352,
      "learning_rate": 4.704704704704705e-06,
      "loss": 0.4904,
      "step": 22400
    },
    {
      "epoch": 28.016833976833976,
      "grad_norm": 0.0016780027654021978,
      "learning_rate": 4.701844701844703e-06,
      "loss": 0.0008,
      "step": 22410
    },
    {
      "epoch": 28.017091377091376,
      "grad_norm": 0.0035610671620815992,
      "learning_rate": 4.698984698984699e-06,
      "loss": 0.0015,
      "step": 22420
    },
    {
      "epoch": 28.017348777348776,
      "grad_norm": 0.0016747131012380123,
      "learning_rate": 4.696124696124696e-06,
      "loss": 0.0011,
      "step": 22430
    },
    {
      "epoch": 28.01760617760618,
      "grad_norm": 0.020335784181952477,
      "learning_rate": 4.693264693264694e-06,
      "loss": 1.5259,
      "step": 22440
    },
    {
      "epoch": 28.01786357786358,
      "grad_norm": 0.004793055355548859,
      "learning_rate": 4.690404690404691e-06,
      "loss": 0.9107,
      "step": 22450
    },
    {
      "epoch": 28.01812097812098,
      "grad_norm": 3.8998618125915527,
      "learning_rate": 4.6875446875446875e-06,
      "loss": 0.9888,
      "step": 22460
    },
    {
      "epoch": 28.01837837837838,
      "grad_norm": 0.003152688266709447,
      "learning_rate": 4.684684684684685e-06,
      "loss": 0.8276,
      "step": 22470
    },
    {
      "epoch": 28.01863577863578,
      "grad_norm": 0.07769756019115448,
      "learning_rate": 4.681824681824682e-06,
      "loss": 0.0053,
      "step": 22480
    },
    {
      "epoch": 28.01889317889318,
      "grad_norm": 0.0015963425394147635,
      "learning_rate": 4.6789646789646795e-06,
      "loss": 0.4891,
      "step": 22490
    },
    {
      "epoch": 28.01915057915058,
      "grad_norm": 0.0019973055459558964,
      "learning_rate": 4.676104676104676e-06,
      "loss": 0.0014,
      "step": 22500
    },
    {
      "epoch": 28.01940797940798,
      "grad_norm": 0.18019846081733704,
      "learning_rate": 4.673244673244673e-06,
      "loss": 1.7923,
      "step": 22510
    },
    {
      "epoch": 28.01966537966538,
      "grad_norm": 0.0022248122841119766,
      "learning_rate": 4.6703846703846706e-06,
      "loss": 0.7992,
      "step": 22520
    },
    {
      "epoch": 28.01992277992278,
      "grad_norm": 171.22250366210938,
      "learning_rate": 4.667524667524668e-06,
      "loss": 1.1658,
      "step": 22530
    },
    {
      "epoch": 28.02,
      "eval_accuracy": 0.8695652173913043,
      "eval_loss": 0.6606096625328064,
      "eval_runtime": 13.404,
      "eval_samples_per_second": 3.432,
      "eval_steps_per_second": 3.432,
      "step": 22533
    },
    {
      "epoch": 29.00018018018018,
      "grad_norm": 0.1187930479645729,
      "learning_rate": 4.664664664664665e-06,
      "loss": 0.0024,
      "step": 22540
    },
    {
      "epoch": 29.00043758043758,
      "grad_norm": 0.03540468215942383,
      "learning_rate": 4.661804661804662e-06,
      "loss": 0.0007,
      "step": 22550
    },
    {
      "epoch": 29.00069498069498,
      "grad_norm": 0.008411556482315063,
      "learning_rate": 4.658944658944659e-06,
      "loss": 0.1653,
      "step": 22560
    },
    {
      "epoch": 29.00095238095238,
      "grad_norm": 208.8349151611328,
      "learning_rate": 4.656084656084656e-06,
      "loss": 0.1932,
      "step": 22570
    },
    {
      "epoch": 29.00120978120978,
      "grad_norm": 0.5143762826919556,
      "learning_rate": 4.653224653224654e-06,
      "loss": 0.0046,
      "step": 22580
    },
    {
      "epoch": 29.00146718146718,
      "grad_norm": 0.7926735281944275,
      "learning_rate": 4.650364650364651e-06,
      "loss": 0.0034,
      "step": 22590
    },
    {
      "epoch": 29.00172458172458,
      "grad_norm": 0.000883811735548079,
      "learning_rate": 4.647504647504648e-06,
      "loss": 0.3767,
      "step": 22600
    },
    {
      "epoch": 29.00198198198198,
      "grad_norm": 686.049072265625,
      "learning_rate": 4.6446446446446456e-06,
      "loss": 0.4603,
      "step": 22610
    },
    {
      "epoch": 29.00223938223938,
      "grad_norm": 0.14210084080696106,
      "learning_rate": 4.641784641784642e-06,
      "loss": 0.0021,
      "step": 22620
    },
    {
      "epoch": 29.00249678249678,
      "grad_norm": 0.0504668764770031,
      "learning_rate": 4.638924638924639e-06,
      "loss": 0.4215,
      "step": 22630
    },
    {
      "epoch": 29.00275418275418,
      "grad_norm": 0.03664263337850571,
      "learning_rate": 4.636064636064637e-06,
      "loss": 0.0006,
      "step": 22640
    },
    {
      "epoch": 29.003011583011585,
      "grad_norm": 464.8655700683594,
      "learning_rate": 4.633204633204634e-06,
      "loss": 0.0139,
      "step": 22650
    },
    {
      "epoch": 29.003268983268985,
      "grad_norm": 0.00034964046790264547,
      "learning_rate": 4.630344630344631e-06,
      "loss": 1.1042,
      "step": 22660
    },
    {
      "epoch": 29.003526383526385,
      "grad_norm": 0.0003824385639745742,
      "learning_rate": 4.627484627484628e-06,
      "loss": 0.0007,
      "step": 22670
    },
    {
      "epoch": 29.003783783783785,
      "grad_norm": 0.0016709021292626858,
      "learning_rate": 4.624624624624625e-06,
      "loss": 1.7071,
      "step": 22680
    },
    {
      "epoch": 29.004041184041185,
      "grad_norm": 0.07814698666334152,
      "learning_rate": 4.621764621764622e-06,
      "loss": 0.5468,
      "step": 22690
    },
    {
      "epoch": 29.004298584298585,
      "grad_norm": 0.06099068745970726,
      "learning_rate": 4.61890461890462e-06,
      "loss": 0.7175,
      "step": 22700
    },
    {
      "epoch": 29.004555984555985,
      "grad_norm": 0.02045220322906971,
      "learning_rate": 4.616044616044616e-06,
      "loss": 0.5742,
      "step": 22710
    },
    {
      "epoch": 29.004813384813385,
      "grad_norm": 0.029201224446296692,
      "learning_rate": 4.6131846131846135e-06,
      "loss": 0.0006,
      "step": 22720
    },
    {
      "epoch": 29.005070785070785,
      "grad_norm": 6.332117557525635,
      "learning_rate": 4.610324610324611e-06,
      "loss": 0.006,
      "step": 22730
    },
    {
      "epoch": 29.005328185328185,
      "grad_norm": 0.12955205142498016,
      "learning_rate": 4.607464607464608e-06,
      "loss": 1.337,
      "step": 22740
    },
    {
      "epoch": 29.005585585585585,
      "grad_norm": 208.47537231445312,
      "learning_rate": 4.604604604604605e-06,
      "loss": 1.1339,
      "step": 22750
    },
    {
      "epoch": 29.005842985842985,
      "grad_norm": 0.03571253642439842,
      "learning_rate": 4.601744601744602e-06,
      "loss": 0.0006,
      "step": 22760
    },
    {
      "epoch": 29.006100386100385,
      "grad_norm": 0.018318068236112595,
      "learning_rate": 4.598884598884599e-06,
      "loss": 0.0017,
      "step": 22770
    },
    {
      "epoch": 29.006357786357785,
      "grad_norm": 0.0016119711799547076,
      "learning_rate": 4.5960245960245966e-06,
      "loss": 0.272,
      "step": 22780
    },
    {
      "epoch": 29.006615186615186,
      "grad_norm": 0.030253881588578224,
      "learning_rate": 4.593164593164594e-06,
      "loss": 0.6145,
      "step": 22790
    },
    {
      "epoch": 29.006872586872586,
      "grad_norm": 1.5183641910552979,
      "learning_rate": 4.59030459030459e-06,
      "loss": 0.0013,
      "step": 22800
    },
    {
      "epoch": 29.007129987129986,
      "grad_norm": 0.010023567825555801,
      "learning_rate": 4.587444587444588e-06,
      "loss": 0.5396,
      "step": 22810
    },
    {
      "epoch": 29.007387387387386,
      "grad_norm": 1.5276458263397217,
      "learning_rate": 4.584584584584585e-06,
      "loss": 0.0042,
      "step": 22820
    },
    {
      "epoch": 29.00764478764479,
      "grad_norm": 0.007271212060004473,
      "learning_rate": 4.581724581724582e-06,
      "loss": 0.6234,
      "step": 22830
    },
    {
      "epoch": 29.00790218790219,
      "grad_norm": 0.039576660841703415,
      "learning_rate": 4.578864578864579e-06,
      "loss": 0.0006,
      "step": 22840
    },
    {
      "epoch": 29.00815958815959,
      "grad_norm": 0.0008184545440599322,
      "learning_rate": 4.576004576004576e-06,
      "loss": 0.3065,
      "step": 22850
    },
    {
      "epoch": 29.00841698841699,
      "grad_norm": 0.012131177820265293,
      "learning_rate": 4.573144573144573e-06,
      "loss": 1.2567,
      "step": 22860
    },
    {
      "epoch": 29.00867438867439,
      "grad_norm": 0.06175015866756439,
      "learning_rate": 4.570284570284571e-06,
      "loss": 0.6189,
      "step": 22870
    },
    {
      "epoch": 29.00893178893179,
      "grad_norm": 0.03581232950091362,
      "learning_rate": 4.567424567424568e-06,
      "loss": 0.0004,
      "step": 22880
    },
    {
      "epoch": 29.00918918918919,
      "grad_norm": 54.22113037109375,
      "learning_rate": 4.5645645645645645e-06,
      "loss": 1.1175,
      "step": 22890
    },
    {
      "epoch": 29.00944658944659,
      "grad_norm": 24.692176818847656,
      "learning_rate": 4.561704561704562e-06,
      "loss": 0.8245,
      "step": 22900
    },
    {
      "epoch": 29.00970398970399,
      "grad_norm": 0.0129568325355649,
      "learning_rate": 4.558844558844559e-06,
      "loss": 0.0181,
      "step": 22910
    },
    {
      "epoch": 29.00996138996139,
      "grad_norm": 0.22810792922973633,
      "learning_rate": 4.5559845559845564e-06,
      "loss": 0.4374,
      "step": 22920
    },
    {
      "epoch": 29.01021879021879,
      "grad_norm": 0.030504779890179634,
      "learning_rate": 4.553124553124553e-06,
      "loss": 0.0009,
      "step": 22930
    },
    {
      "epoch": 29.01047619047619,
      "grad_norm": 0.010466916486620903,
      "learning_rate": 4.55026455026455e-06,
      "loss": 1.5878,
      "step": 22940
    },
    {
      "epoch": 29.01073359073359,
      "grad_norm": 0.005396629683673382,
      "learning_rate": 4.5474045474045475e-06,
      "loss": 0.0026,
      "step": 22950
    },
    {
      "epoch": 29.01099099099099,
      "grad_norm": 0.20146198570728302,
      "learning_rate": 4.544544544544545e-06,
      "loss": 0.4596,
      "step": 22960
    },
    {
      "epoch": 29.01124839124839,
      "grad_norm": 139.57632446289062,
      "learning_rate": 4.541684541684542e-06,
      "loss": 1.0697,
      "step": 22970
    },
    {
      "epoch": 29.01150579150579,
      "grad_norm": 85.71121215820312,
      "learning_rate": 4.5388245388245395e-06,
      "loss": 1.3387,
      "step": 22980
    },
    {
      "epoch": 29.01176319176319,
      "grad_norm": 0.12364278733730316,
      "learning_rate": 4.535964535964537e-06,
      "loss": 0.3547,
      "step": 22990
    },
    {
      "epoch": 29.01202059202059,
      "grad_norm": 0.0017210529185831547,
      "learning_rate": 4.533104533104533e-06,
      "loss": 0.0014,
      "step": 23000
    },
    {
      "epoch": 29.01227799227799,
      "grad_norm": 32.765933990478516,
      "learning_rate": 4.530244530244531e-06,
      "loss": 0.0297,
      "step": 23010
    },
    {
      "epoch": 29.012535392535394,
      "grad_norm": 38.97309875488281,
      "learning_rate": 4.527384527384528e-06,
      "loss": 0.5779,
      "step": 23020
    },
    {
      "epoch": 29.012792792792794,
      "grad_norm": 0.22549696266651154,
      "learning_rate": 4.524524524524525e-06,
      "loss": 0.0118,
      "step": 23030
    },
    {
      "epoch": 29.013050193050194,
      "grad_norm": 162.48883056640625,
      "learning_rate": 4.5216645216645225e-06,
      "loss": 1.2432,
      "step": 23040
    },
    {
      "epoch": 29.013307593307594,
      "grad_norm": 0.01860671490430832,
      "learning_rate": 4.518804518804519e-06,
      "loss": 1.0974,
      "step": 23050
    },
    {
      "epoch": 29.013564993564994,
      "grad_norm": 0.13944114744663239,
      "learning_rate": 4.515944515944516e-06,
      "loss": 0.0012,
      "step": 23060
    },
    {
      "epoch": 29.013822393822394,
      "grad_norm": 0.0077326660975813866,
      "learning_rate": 4.513084513084514e-06,
      "loss": 0.4396,
      "step": 23070
    },
    {
      "epoch": 29.014079794079795,
      "grad_norm": 0.0013942186487838626,
      "learning_rate": 4.510224510224511e-06,
      "loss": 0.0016,
      "step": 23080
    },
    {
      "epoch": 29.014337194337195,
      "grad_norm": 0.1694478839635849,
      "learning_rate": 4.5073645073645074e-06,
      "loss": 0.9909,
      "step": 23090
    },
    {
      "epoch": 29.014594594594595,
      "grad_norm": 156.31834411621094,
      "learning_rate": 4.504504504504505e-06,
      "loss": 0.0849,
      "step": 23100
    },
    {
      "epoch": 29.014851994851995,
      "grad_norm": 0.18722569942474365,
      "learning_rate": 4.501644501644502e-06,
      "loss": 0.4487,
      "step": 23110
    },
    {
      "epoch": 29.015109395109395,
      "grad_norm": 0.0004421714984346181,
      "learning_rate": 4.498784498784499e-06,
      "loss": 0.5625,
      "step": 23120
    },
    {
      "epoch": 29.015366795366795,
      "grad_norm": 279.7695617675781,
      "learning_rate": 4.495924495924497e-06,
      "loss": 0.4133,
      "step": 23130
    },
    {
      "epoch": 29.015624195624195,
      "grad_norm": 0.001477337907999754,
      "learning_rate": 4.493064493064493e-06,
      "loss": 2.0714,
      "step": 23140
    },
    {
      "epoch": 29.015881595881595,
      "grad_norm": 0.051959581673145294,
      "learning_rate": 4.4902044902044905e-06,
      "loss": 0.6586,
      "step": 23150
    },
    {
      "epoch": 29.016138996138995,
      "grad_norm": 59.180824279785156,
      "learning_rate": 4.487344487344488e-06,
      "loss": 1.1272,
      "step": 23160
    },
    {
      "epoch": 29.016396396396395,
      "grad_norm": 0.1820637434720993,
      "learning_rate": 4.484484484484485e-06,
      "loss": 0.5311,
      "step": 23170
    },
    {
      "epoch": 29.016653796653795,
      "grad_norm": 0.001803033985197544,
      "learning_rate": 4.481624481624482e-06,
      "loss": 2.0791,
      "step": 23180
    },
    {
      "epoch": 29.016911196911195,
      "grad_norm": 0.20301368832588196,
      "learning_rate": 4.478764478764479e-06,
      "loss": 0.5542,
      "step": 23190
    },
    {
      "epoch": 29.0171685971686,
      "grad_norm": 0.002093987073749304,
      "learning_rate": 4.475904475904476e-06,
      "loss": 0.0012,
      "step": 23200
    },
    {
      "epoch": 29.017425997426,
      "grad_norm": 1884.36181640625,
      "learning_rate": 4.4730444730444735e-06,
      "loss": 0.5681,
      "step": 23210
    },
    {
      "epoch": 29.0176833976834,
      "grad_norm": 0.0037956798914819956,
      "learning_rate": 4.470184470184471e-06,
      "loss": 0.0032,
      "step": 23220
    },
    {
      "epoch": 29.0179407979408,
      "grad_norm": 2.3361432552337646,
      "learning_rate": 4.467324467324467e-06,
      "loss": 0.4418,
      "step": 23230
    },
    {
      "epoch": 29.0181981981982,
      "grad_norm": 0.03866471350193024,
      "learning_rate": 4.464464464464465e-06,
      "loss": 1.0058,
      "step": 23240
    },
    {
      "epoch": 29.0184555984556,
      "grad_norm": 0.10158059746026993,
      "learning_rate": 4.461604461604462e-06,
      "loss": 0.367,
      "step": 23250
    },
    {
      "epoch": 29.018712998713,
      "grad_norm": 0.29946258664131165,
      "learning_rate": 4.458744458744459e-06,
      "loss": 0.5789,
      "step": 23260
    },
    {
      "epoch": 29.0189703989704,
      "grad_norm": 0.004636166617274284,
      "learning_rate": 4.455884455884456e-06,
      "loss": 0.6314,
      "step": 23270
    },
    {
      "epoch": 29.0192277992278,
      "grad_norm": 0.025164945051074028,
      "learning_rate": 4.453024453024453e-06,
      "loss": 0.0007,
      "step": 23280
    },
    {
      "epoch": 29.0194851994852,
      "grad_norm": 0.3000508248806,
      "learning_rate": 4.45016445016445e-06,
      "loss": 0.9031,
      "step": 23290
    },
    {
      "epoch": 29.0197425997426,
      "grad_norm": 0.002804351970553398,
      "learning_rate": 4.447304447304448e-06,
      "loss": 0.4417,
      "step": 23300
    },
    {
      "epoch": 29.02,
      "grad_norm": 0.04149908944964409,
      "learning_rate": 4.444444444444444e-06,
      "loss": 0.6756,
      "step": 23310
    },
    {
      "epoch": 29.02,
      "eval_accuracy": 0.717391304347826,
      "eval_loss": 1.6570743322372437,
      "eval_runtime": 13.3509,
      "eval_samples_per_second": 3.445,
      "eval_steps_per_second": 3.445,
      "step": 23310
    },
    {
      "epoch": 30.0002574002574,
      "grad_norm": 0.21637240052223206,
      "learning_rate": 4.4415844415844415e-06,
      "loss": 1.6747,
      "step": 23320
    },
    {
      "epoch": 30.0005148005148,
      "grad_norm": 0.8607198596000671,
      "learning_rate": 4.438724438724439e-06,
      "loss": 0.6147,
      "step": 23330
    },
    {
      "epoch": 30.0007722007722,
      "grad_norm": 0.0019639742095023394,
      "learning_rate": 4.435864435864436e-06,
      "loss": 0.0006,
      "step": 23340
    },
    {
      "epoch": 30.0010296010296,
      "grad_norm": 1.0920131206512451,
      "learning_rate": 4.433004433004433e-06,
      "loss": 1.0313,
      "step": 23350
    },
    {
      "epoch": 30.001287001287,
      "grad_norm": 0.0037155456375330687,
      "learning_rate": 4.430144430144431e-06,
      "loss": 0.5143,
      "step": 23360
    },
    {
      "epoch": 30.0015444015444,
      "grad_norm": 0.0018675295868888497,
      "learning_rate": 4.427284427284428e-06,
      "loss": 0.0016,
      "step": 23370
    },
    {
      "epoch": 30.0018018018018,
      "grad_norm": 0.008687763474881649,
      "learning_rate": 4.424424424424425e-06,
      "loss": 0.0012,
      "step": 23380
    },
    {
      "epoch": 30.0020592020592,
      "grad_norm": 0.03639661893248558,
      "learning_rate": 4.421564421564422e-06,
      "loss": 1.2112,
      "step": 23390
    },
    {
      "epoch": 30.0023166023166,
      "grad_norm": 0.12909619510173798,
      "learning_rate": 4.418704418704419e-06,
      "loss": 0.586,
      "step": 23400
    },
    {
      "epoch": 30.002574002574004,
      "grad_norm": 0.3514103591442108,
      "learning_rate": 4.4158444158444165e-06,
      "loss": 0.4516,
      "step": 23410
    },
    {
      "epoch": 30.002831402831404,
      "grad_norm": 37.94804000854492,
      "learning_rate": 4.412984412984414e-06,
      "loss": 0.2859,
      "step": 23420
    },
    {
      "epoch": 30.003088803088804,
      "grad_norm": 0.00035340170143172145,
      "learning_rate": 4.41012441012441e-06,
      "loss": 0.1207,
      "step": 23430
    },
    {
      "epoch": 30.003346203346204,
      "grad_norm": 0.00048636618885211647,
      "learning_rate": 4.4072644072644076e-06,
      "loss": 0.5357,
      "step": 23440
    },
    {
      "epoch": 30.003603603603604,
      "grad_norm": 0.00040685260319150984,
      "learning_rate": 4.404404404404405e-06,
      "loss": 0.0143,
      "step": 23450
    },
    {
      "epoch": 30.003861003861005,
      "grad_norm": 0.0009621041244827211,
      "learning_rate": 4.401544401544402e-06,
      "loss": 0.7266,
      "step": 23460
    },
    {
      "epoch": 30.004118404118405,
      "grad_norm": 0.04314970225095749,
      "learning_rate": 4.3986843986843995e-06,
      "loss": 1.1009,
      "step": 23470
    },
    {
      "epoch": 30.004375804375805,
      "grad_norm": 0.007667004596441984,
      "learning_rate": 4.395824395824396e-06,
      "loss": 0.0003,
      "step": 23480
    },
    {
      "epoch": 30.004633204633205,
      "grad_norm": 0.038383617997169495,
      "learning_rate": 4.392964392964393e-06,
      "loss": 0.002,
      "step": 23490
    },
    {
      "epoch": 30.004890604890605,
      "grad_norm": 0.03447192534804344,
      "learning_rate": 4.390104390104391e-06,
      "loss": 0.6983,
      "step": 23500
    },
    {
      "epoch": 30.005148005148005,
      "grad_norm": 0.03259385749697685,
      "learning_rate": 4.387244387244388e-06,
      "loss": 0.4584,
      "step": 23510
    },
    {
      "epoch": 30.005405405405405,
      "grad_norm": 0.005583514925092459,
      "learning_rate": 4.384384384384384e-06,
      "loss": 0.0099,
      "step": 23520
    },
    {
      "epoch": 30.005662805662805,
      "grad_norm": 0.01332216989248991,
      "learning_rate": 4.381524381524382e-06,
      "loss": 1.1229,
      "step": 23530
    },
    {
      "epoch": 30.005920205920205,
      "grad_norm": 0.0005782814114354551,
      "learning_rate": 4.378664378664379e-06,
      "loss": 0.5327,
      "step": 23540
    },
    {
      "epoch": 30.006177606177605,
      "grad_norm": 0.10082386434078217,
      "learning_rate": 4.375804375804376e-06,
      "loss": 0.4731,
      "step": 23550
    },
    {
      "epoch": 30.006435006435005,
      "grad_norm": 0.09667433053255081,
      "learning_rate": 4.372944372944373e-06,
      "loss": 0.4831,
      "step": 23560
    },
    {
      "epoch": 30.006692406692405,
      "grad_norm": 0.06305726617574692,
      "learning_rate": 4.37008437008437e-06,
      "loss": 1.1478,
      "step": 23570
    },
    {
      "epoch": 30.006949806949805,
      "grad_norm": 0.0013677224051207304,
      "learning_rate": 4.3672243672243675e-06,
      "loss": 0.2762,
      "step": 23580
    },
    {
      "epoch": 30.00720720720721,
      "grad_norm": 0.0005206306814216077,
      "learning_rate": 4.364364364364365e-06,
      "loss": 0.0009,
      "step": 23590
    },
    {
      "epoch": 30.00746460746461,
      "grad_norm": 0.06642114371061325,
      "learning_rate": 4.361504361504362e-06,
      "loss": 0.9313,
      "step": 23600
    },
    {
      "epoch": 30.00772200772201,
      "grad_norm": 0.20846591889858246,
      "learning_rate": 4.3586443586443586e-06,
      "loss": 0.5189,
      "step": 23610
    },
    {
      "epoch": 30.00797940797941,
      "grad_norm": 0.000537847401574254,
      "learning_rate": 4.355784355784356e-06,
      "loss": 0.0006,
      "step": 23620
    },
    {
      "epoch": 30.00823680823681,
      "grad_norm": 0.052646271884441376,
      "learning_rate": 4.352924352924353e-06,
      "loss": 0.0002,
      "step": 23630
    },
    {
      "epoch": 30.00849420849421,
      "grad_norm": 0.11752662807703018,
      "learning_rate": 4.3500643500643505e-06,
      "loss": 0.6548,
      "step": 23640
    },
    {
      "epoch": 30.00875160875161,
      "grad_norm": 170.06265258789062,
      "learning_rate": 4.347204347204347e-06,
      "loss": 0.7438,
      "step": 23650
    },
    {
      "epoch": 30.00900900900901,
      "grad_norm": 0.0005917979287914932,
      "learning_rate": 4.344344344344344e-06,
      "loss": 1.3001,
      "step": 23660
    },
    {
      "epoch": 30.00926640926641,
      "grad_norm": 0.014821925200521946,
      "learning_rate": 4.341484341484342e-06,
      "loss": 0.1067,
      "step": 23670
    },
    {
      "epoch": 30.00952380952381,
      "grad_norm": 0.0017121240962296724,
      "learning_rate": 4.338624338624339e-06,
      "loss": 0.5524,
      "step": 23680
    },
    {
      "epoch": 30.00978120978121,
      "grad_norm": 0.023720256984233856,
      "learning_rate": 4.335764335764336e-06,
      "loss": 0.0051,
      "step": 23690
    },
    {
      "epoch": 30.01003861003861,
      "grad_norm": 0.5435638427734375,
      "learning_rate": 4.332904332904333e-06,
      "loss": 0.0102,
      "step": 23700
    },
    {
      "epoch": 30.01029601029601,
      "grad_norm": 0.012425548397004604,
      "learning_rate": 4.330044330044331e-06,
      "loss": 0.1018,
      "step": 23710
    },
    {
      "epoch": 30.01055341055341,
      "grad_norm": 0.03509173542261124,
      "learning_rate": 4.327184327184328e-06,
      "loss": 0.1394,
      "step": 23720
    },
    {
      "epoch": 30.01081081081081,
      "grad_norm": 92.84154510498047,
      "learning_rate": 4.324324324324325e-06,
      "loss": 2.0118,
      "step": 23730
    },
    {
      "epoch": 30.01106821106821,
      "grad_norm": 254.40797424316406,
      "learning_rate": 4.321464321464322e-06,
      "loss": 1.1617,
      "step": 23740
    },
    {
      "epoch": 30.01132561132561,
      "grad_norm": 0.0050951456651091576,
      "learning_rate": 4.318604318604319e-06,
      "loss": 0.1634,
      "step": 23750
    },
    {
      "epoch": 30.01158301158301,
      "grad_norm": 0.00045341532677412033,
      "learning_rate": 4.315744315744317e-06,
      "loss": 0.0071,
      "step": 23760
    },
    {
      "epoch": 30.01184041184041,
      "grad_norm": 0.08173135668039322,
      "learning_rate": 4.312884312884313e-06,
      "loss": 0.0273,
      "step": 23770
    },
    {
      "epoch": 30.012097812097814,
      "grad_norm": 0.0005044002318754792,
      "learning_rate": 4.31002431002431e-06,
      "loss": 0.1153,
      "step": 23780
    },
    {
      "epoch": 30.012355212355214,
      "grad_norm": 0.11497832834720612,
      "learning_rate": 4.307164307164308e-06,
      "loss": 0.0008,
      "step": 23790
    },
    {
      "epoch": 30.012612612612614,
      "grad_norm": 0.12187304347753525,
      "learning_rate": 4.304304304304305e-06,
      "loss": 1.3989,
      "step": 23800
    },
    {
      "epoch": 30.012870012870014,
      "grad_norm": 5.552525520324707,
      "learning_rate": 4.301444301444302e-06,
      "loss": 0.136,
      "step": 23810
    },
    {
      "epoch": 30.013127413127414,
      "grad_norm": 0.0017980223055928946,
      "learning_rate": 4.298584298584299e-06,
      "loss": 0.001,
      "step": 23820
    },
    {
      "epoch": 30.013384813384814,
      "grad_norm": 74.36280822753906,
      "learning_rate": 4.295724295724296e-06,
      "loss": 1.3003,
      "step": 23830
    },
    {
      "epoch": 30.013642213642214,
      "grad_norm": 0.01991387829184532,
      "learning_rate": 4.2928642928642934e-06,
      "loss": 0.0004,
      "step": 23840
    },
    {
      "epoch": 30.013899613899614,
      "grad_norm": 0.020584069192409515,
      "learning_rate": 4.290004290004291e-06,
      "loss": 0.6308,
      "step": 23850
    },
    {
      "epoch": 30.014157014157014,
      "grad_norm": 0.0006809376063756645,
      "learning_rate": 4.287144287144287e-06,
      "loss": 0.0012,
      "step": 23860
    },
    {
      "epoch": 30.014414414414414,
      "grad_norm": 0.00031621515518054366,
      "learning_rate": 4.2842842842842845e-06,
      "loss": 0.0014,
      "step": 23870
    },
    {
      "epoch": 30.014671814671814,
      "grad_norm": 190.67312622070312,
      "learning_rate": 4.281424281424282e-06,
      "loss": 1.3941,
      "step": 23880
    },
    {
      "epoch": 30.014929214929214,
      "grad_norm": 0.16922950744628906,
      "learning_rate": 4.278564278564279e-06,
      "loss": 0.7061,
      "step": 23890
    },
    {
      "epoch": 30.015186615186614,
      "grad_norm": 0.010776959359645844,
      "learning_rate": 4.275704275704276e-06,
      "loss": 0.0011,
      "step": 23900
    },
    {
      "epoch": 30.015444015444015,
      "grad_norm": 8.569811820983887,
      "learning_rate": 4.272844272844273e-06,
      "loss": 1.355,
      "step": 23910
    },
    {
      "epoch": 30.015701415701415,
      "grad_norm": 0.21295268833637238,
      "learning_rate": 4.26998426998427e-06,
      "loss": 0.6669,
      "step": 23920
    },
    {
      "epoch": 30.015958815958815,
      "grad_norm": 0.0388345941901207,
      "learning_rate": 4.267124267124268e-06,
      "loss": 0.6833,
      "step": 23930
    },
    {
      "epoch": 30.016216216216215,
      "grad_norm": 0.03433572128415108,
      "learning_rate": 4.264264264264265e-06,
      "loss": 0.0011,
      "step": 23940
    },
    {
      "epoch": 30.016473616473615,
      "grad_norm": 0.02409375086426735,
      "learning_rate": 4.261404261404261e-06,
      "loss": 0.5623,
      "step": 23950
    },
    {
      "epoch": 30.01673101673102,
      "grad_norm": 0.0004545144329313189,
      "learning_rate": 4.258544258544259e-06,
      "loss": 0.0004,
      "step": 23960
    },
    {
      "epoch": 30.01698841698842,
      "grad_norm": 0.10565972328186035,
      "learning_rate": 4.255684255684256e-06,
      "loss": 0.0009,
      "step": 23970
    },
    {
      "epoch": 30.01724581724582,
      "grad_norm": 0.0003957377339247614,
      "learning_rate": 4.252824252824253e-06,
      "loss": 0.7128,
      "step": 23980
    },
    {
      "epoch": 30.01750321750322,
      "grad_norm": 0.06979605555534363,
      "learning_rate": 4.24996424996425e-06,
      "loss": 0.0011,
      "step": 23990
    },
    {
      "epoch": 30.01776061776062,
      "grad_norm": 0.014289547689259052,
      "learning_rate": 4.247104247104247e-06,
      "loss": 0.0015,
      "step": 24000
    },
    {
      "epoch": 30.01801801801802,
      "grad_norm": 0.03242909535765648,
      "learning_rate": 4.2442442442442444e-06,
      "loss": 0.5711,
      "step": 24010
    },
    {
      "epoch": 30.01827541827542,
      "grad_norm": 0.01781926117837429,
      "learning_rate": 4.241384241384242e-06,
      "loss": 0.1299,
      "step": 24020
    },
    {
      "epoch": 30.01853281853282,
      "grad_norm": 0.030324866995215416,
      "learning_rate": 4.238524238524239e-06,
      "loss": 0.6162,
      "step": 24030
    },
    {
      "epoch": 30.01879021879022,
      "grad_norm": 103.36428833007812,
      "learning_rate": 4.2356642356642355e-06,
      "loss": 0.7377,
      "step": 24040
    },
    {
      "epoch": 30.01904761904762,
      "grad_norm": 0.0003896548296324909,
      "learning_rate": 4.232804232804233e-06,
      "loss": 0.0011,
      "step": 24050
    },
    {
      "epoch": 30.01930501930502,
      "grad_norm": 0.11918792128562927,
      "learning_rate": 4.22994422994423e-06,
      "loss": 1.2608,
      "step": 24060
    },
    {
      "epoch": 30.01956241956242,
      "grad_norm": 0.00038182828575372696,
      "learning_rate": 4.2270842270842275e-06,
      "loss": 1.147,
      "step": 24070
    },
    {
      "epoch": 30.01981981981982,
      "grad_norm": 0.1870720386505127,
      "learning_rate": 4.224224224224225e-06,
      "loss": 0.0165,
      "step": 24080
    },
    {
      "epoch": 30.02,
      "eval_accuracy": 0.8695652173913043,
      "eval_loss": 0.7047258615493774,
      "eval_runtime": 13.4178,
      "eval_samples_per_second": 3.428,
      "eval_steps_per_second": 3.428,
      "step": 24087
    },
    {
      "epoch": 31.00007722007722,
      "grad_norm": 0.0019550484139472246,
      "learning_rate": 4.221364221364222e-06,
      "loss": 0.0009,
      "step": 24090
    },
    {
      "epoch": 31.00033462033462,
      "grad_norm": 0.004212299361824989,
      "learning_rate": 4.2185042185042194e-06,
      "loss": 0.4331,
      "step": 24100
    },
    {
      "epoch": 31.00059202059202,
      "grad_norm": 0.0004280452849343419,
      "learning_rate": 4.215644215644216e-06,
      "loss": 0.001,
      "step": 24110
    },
    {
      "epoch": 31.00084942084942,
      "grad_norm": 0.00045957849943079054,
      "learning_rate": 4.212784212784213e-06,
      "loss": 0.5633,
      "step": 24120
    },
    {
      "epoch": 31.00110682110682,
      "grad_norm": 0.006528822239488363,
      "learning_rate": 4.2099242099242105e-06,
      "loss": 0.0008,
      "step": 24130
    },
    {
      "epoch": 31.00136422136422,
      "grad_norm": 0.023158643394708633,
      "learning_rate": 4.207064207064208e-06,
      "loss": 0.4536,
      "step": 24140
    },
    {
      "epoch": 31.00162162162162,
      "grad_norm": 0.04101202264428139,
      "learning_rate": 4.204204204204204e-06,
      "loss": 0.9761,
      "step": 24150
    },
    {
      "epoch": 31.00187902187902,
      "grad_norm": 0.0005613996763713658,
      "learning_rate": 4.201344201344202e-06,
      "loss": 0.693,
      "step": 24160
    },
    {
      "epoch": 31.002136422136424,
      "grad_norm": 0.013050409033894539,
      "learning_rate": 4.198484198484199e-06,
      "loss": 0.0005,
      "step": 24170
    },
    {
      "epoch": 31.002393822393824,
      "grad_norm": 0.00033386447466909885,
      "learning_rate": 4.195624195624196e-06,
      "loss": 0.2389,
      "step": 24180
    },
    {
      "epoch": 31.002651222651224,
      "grad_norm": 0.0003567083622328937,
      "learning_rate": 4.192764192764194e-06,
      "loss": 0.2853,
      "step": 24190
    },
    {
      "epoch": 31.002908622908624,
      "grad_norm": 0.0013058069162070751,
      "learning_rate": 4.18990418990419e-06,
      "loss": 0.002,
      "step": 24200
    },
    {
      "epoch": 31.003166023166024,
      "grad_norm": 0.0475904643535614,
      "learning_rate": 4.187044187044187e-06,
      "loss": 0.0008,
      "step": 24210
    },
    {
      "epoch": 31.003423423423424,
      "grad_norm": 0.00036910572089254856,
      "learning_rate": 4.184184184184185e-06,
      "loss": 0.0004,
      "step": 24220
    },
    {
      "epoch": 31.003680823680824,
      "grad_norm": 0.3366129398345947,
      "learning_rate": 4.181324181324182e-06,
      "loss": 0.3606,
      "step": 24230
    },
    {
      "epoch": 31.003938223938224,
      "grad_norm": 0.0003529265522956848,
      "learning_rate": 4.1784641784641785e-06,
      "loss": 0.4474,
      "step": 24240
    },
    {
      "epoch": 31.004195624195624,
      "grad_norm": 0.2059638500213623,
      "learning_rate": 4.175604175604176e-06,
      "loss": 0.0879,
      "step": 24250
    },
    {
      "epoch": 31.004453024453024,
      "grad_norm": 0.00036747619742527604,
      "learning_rate": 4.172744172744173e-06,
      "loss": 0.0002,
      "step": 24260
    },
    {
      "epoch": 31.004710424710424,
      "grad_norm": 217.43408203125,
      "learning_rate": 4.16988416988417e-06,
      "loss": 0.5037,
      "step": 24270
    },
    {
      "epoch": 31.004967824967824,
      "grad_norm": 0.061313144862651825,
      "learning_rate": 4.167024167024168e-06,
      "loss": 0.0009,
      "step": 24280
    },
    {
      "epoch": 31.005225225225225,
      "grad_norm": 0.010008785873651505,
      "learning_rate": 4.164164164164164e-06,
      "loss": 0.0001,
      "step": 24290
    },
    {
      "epoch": 31.005482625482625,
      "grad_norm": 0.0011877252254635096,
      "learning_rate": 4.1613041613041615e-06,
      "loss": 2.6035,
      "step": 24300
    },
    {
      "epoch": 31.005740025740025,
      "grad_norm": 0.049051184207201004,
      "learning_rate": 4.158444158444159e-06,
      "loss": 0.4011,
      "step": 24310
    },
    {
      "epoch": 31.005997425997425,
      "grad_norm": 0.004785649012774229,
      "learning_rate": 4.155584155584156e-06,
      "loss": 0.6697,
      "step": 24320
    },
    {
      "epoch": 31.006254826254825,
      "grad_norm": 0.00534382788464427,
      "learning_rate": 4.152724152724153e-06,
      "loss": 0.4767,
      "step": 24330
    },
    {
      "epoch": 31.006512226512225,
      "grad_norm": 742.736083984375,
      "learning_rate": 4.14986414986415e-06,
      "loss": 0.5331,
      "step": 24340
    },
    {
      "epoch": 31.00676962676963,
      "grad_norm": 0.00026568840257823467,
      "learning_rate": 4.147004147004147e-06,
      "loss": 0.2242,
      "step": 24350
    },
    {
      "epoch": 31.00702702702703,
      "grad_norm": 0.0005520674167200923,
      "learning_rate": 4.1441441441441446e-06,
      "loss": 0.6095,
      "step": 24360
    },
    {
      "epoch": 31.00728442728443,
      "grad_norm": 0.019691677764058113,
      "learning_rate": 4.141284141284142e-06,
      "loss": 0.0005,
      "step": 24370
    },
    {
      "epoch": 31.00754182754183,
      "grad_norm": 0.022847900167107582,
      "learning_rate": 4.138424138424138e-06,
      "loss": 0.0009,
      "step": 24380
    },
    {
      "epoch": 31.00779922779923,
      "grad_norm": 0.00487444456666708,
      "learning_rate": 4.135564135564136e-06,
      "loss": 0.9965,
      "step": 24390
    },
    {
      "epoch": 31.00805662805663,
      "grad_norm": 1136.6373291015625,
      "learning_rate": 4.132704132704133e-06,
      "loss": 1.1896,
      "step": 24400
    },
    {
      "epoch": 31.00831402831403,
      "grad_norm": 0.0009329996537417173,
      "learning_rate": 4.12984412984413e-06,
      "loss": 0.4841,
      "step": 24410
    },
    {
      "epoch": 31.00857142857143,
      "grad_norm": 0.0002942239516414702,
      "learning_rate": 4.126984126984127e-06,
      "loss": 0.4838,
      "step": 24420
    },
    {
      "epoch": 31.00882882882883,
      "grad_norm": 0.000892312207724899,
      "learning_rate": 4.124124124124124e-06,
      "loss": 0.0002,
      "step": 24430
    },
    {
      "epoch": 31.00908622908623,
      "grad_norm": 84.1487808227539,
      "learning_rate": 4.121264121264121e-06,
      "loss": 0.5793,
      "step": 24440
    },
    {
      "epoch": 31.00934362934363,
      "grad_norm": 0.0861109271645546,
      "learning_rate": 4.118404118404119e-06,
      "loss": 0.7574,
      "step": 24450
    },
    {
      "epoch": 31.00960102960103,
      "grad_norm": 0.053891342133283615,
      "learning_rate": 4.115544115544116e-06,
      "loss": 0.4293,
      "step": 24460
    },
    {
      "epoch": 31.00985842985843,
      "grad_norm": 0.022826682776212692,
      "learning_rate": 4.112684112684113e-06,
      "loss": 0.1316,
      "step": 24470
    },
    {
      "epoch": 31.01011583011583,
      "grad_norm": 0.012865704484283924,
      "learning_rate": 4.109824109824111e-06,
      "loss": 0.5083,
      "step": 24480
    },
    {
      "epoch": 31.01037323037323,
      "grad_norm": 0.0002833681064657867,
      "learning_rate": 4.106964106964107e-06,
      "loss": 0.9789,
      "step": 24490
    },
    {
      "epoch": 31.01063063063063,
      "grad_norm": 0.1123170480132103,
      "learning_rate": 4.1041041041041045e-06,
      "loss": 0.3515,
      "step": 24500
    },
    {
      "epoch": 31.01088803088803,
      "grad_norm": 0.026177385821938515,
      "learning_rate": 4.101244101244102e-06,
      "loss": 0.0005,
      "step": 24510
    },
    {
      "epoch": 31.01114543114543,
      "grad_norm": 0.00027682280051521957,
      "learning_rate": 4.098384098384099e-06,
      "loss": 0.6438,
      "step": 24520
    },
    {
      "epoch": 31.011402831402833,
      "grad_norm": 0.029603002592921257,
      "learning_rate": 4.095524095524096e-06,
      "loss": 0.6032,
      "step": 24530
    },
    {
      "epoch": 31.011660231660233,
      "grad_norm": 0.0008603780879639089,
      "learning_rate": 4.092664092664093e-06,
      "loss": 0.6519,
      "step": 24540
    },
    {
      "epoch": 31.011917631917633,
      "grad_norm": 0.012208040803670883,
      "learning_rate": 4.08980408980409e-06,
      "loss": 0.7892,
      "step": 24550
    },
    {
      "epoch": 31.012175032175033,
      "grad_norm": 0.00032583274878561497,
      "learning_rate": 4.0869440869440875e-06,
      "loss": 0.002,
      "step": 24560
    },
    {
      "epoch": 31.012432432432433,
      "grad_norm": 0.00026643762248568237,
      "learning_rate": 4.084084084084085e-06,
      "loss": 0.0021,
      "step": 24570
    },
    {
      "epoch": 31.012689832689833,
      "grad_norm": 0.002847940195351839,
      "learning_rate": 4.081224081224081e-06,
      "loss": 0.9015,
      "step": 24580
    },
    {
      "epoch": 31.012947232947234,
      "grad_norm": 0.016298290342092514,
      "learning_rate": 4.078364078364079e-06,
      "loss": 0.0007,
      "step": 24590
    },
    {
      "epoch": 31.013204633204634,
      "grad_norm": 0.04232637211680412,
      "learning_rate": 4.075504075504076e-06,
      "loss": 0.3838,
      "step": 24600
    },
    {
      "epoch": 31.013462033462034,
      "grad_norm": 0.0029244639445096254,
      "learning_rate": 4.072644072644073e-06,
      "loss": 0.6542,
      "step": 24610
    },
    {
      "epoch": 31.013719433719434,
      "grad_norm": 0.07914727926254272,
      "learning_rate": 4.0697840697840706e-06,
      "loss": 0.2355,
      "step": 24620
    },
    {
      "epoch": 31.013976833976834,
      "grad_norm": 0.07038512825965881,
      "learning_rate": 4.066924066924067e-06,
      "loss": 0.5107,
      "step": 24630
    },
    {
      "epoch": 31.014234234234234,
      "grad_norm": 0.00048367600538767874,
      "learning_rate": 4.064064064064064e-06,
      "loss": 0.6973,
      "step": 24640
    },
    {
      "epoch": 31.014491634491634,
      "grad_norm": 0.00034090675762854517,
      "learning_rate": 4.061204061204062e-06,
      "loss": 0.1097,
      "step": 24650
    },
    {
      "epoch": 31.014749034749034,
      "grad_norm": 421.32574462890625,
      "learning_rate": 4.058344058344059e-06,
      "loss": 1.0028,
      "step": 24660
    },
    {
      "epoch": 31.015006435006434,
      "grad_norm": 0.11087636649608612,
      "learning_rate": 4.0554840554840554e-06,
      "loss": 0.0003,
      "step": 24670
    },
    {
      "epoch": 31.015263835263834,
      "grad_norm": 0.05585281923413277,
      "learning_rate": 4.052624052624053e-06,
      "loss": 0.6875,
      "step": 24680
    },
    {
      "epoch": 31.015521235521234,
      "grad_norm": 0.04771759733557701,
      "learning_rate": 4.04976404976405e-06,
      "loss": 0.0004,
      "step": 24690
    },
    {
      "epoch": 31.015778635778634,
      "grad_norm": 0.010960754007101059,
      "learning_rate": 4.046904046904047e-06,
      "loss": 1.7557,
      "step": 24700
    },
    {
      "epoch": 31.016036036036034,
      "grad_norm": 0.17178845405578613,
      "learning_rate": 4.044044044044044e-06,
      "loss": 0.5631,
      "step": 24710
    },
    {
      "epoch": 31.016293436293438,
      "grad_norm": 0.07653289288282394,
      "learning_rate": 4.041184041184041e-06,
      "loss": 0.324,
      "step": 24720
    },
    {
      "epoch": 31.016550836550838,
      "grad_norm": 0.00036575511330738664,
      "learning_rate": 4.0383240383240385e-06,
      "loss": 1.0434,
      "step": 24730
    },
    {
      "epoch": 31.016808236808238,
      "grad_norm": 0.047172967344522476,
      "learning_rate": 4.035464035464036e-06,
      "loss": 0.4783,
      "step": 24740
    },
    {
      "epoch": 31.017065637065638,
      "grad_norm": 8.044768333435059,
      "learning_rate": 4.032604032604033e-06,
      "loss": 0.0029,
      "step": 24750
    },
    {
      "epoch": 31.017323037323038,
      "grad_norm": 0.007932187058031559,
      "learning_rate": 4.02974402974403e-06,
      "loss": 0.0013,
      "step": 24760
    },
    {
      "epoch": 31.01758043758044,
      "grad_norm": 0.02719135954976082,
      "learning_rate": 4.026884026884027e-06,
      "loss": 0.6169,
      "step": 24770
    },
    {
      "epoch": 31.01783783783784,
      "grad_norm": 0.00028610503068193793,
      "learning_rate": 4.024024024024024e-06,
      "loss": 0.0003,
      "step": 24780
    },
    {
      "epoch": 31.01809523809524,
      "grad_norm": 0.021465850993990898,
      "learning_rate": 4.0211640211640215e-06,
      "loss": 0.495,
      "step": 24790
    },
    {
      "epoch": 31.01835263835264,
      "grad_norm": 3.276745319366455,
      "learning_rate": 4.018304018304018e-06,
      "loss": 0.0036,
      "step": 24800
    },
    {
      "epoch": 31.01861003861004,
      "grad_norm": 0.0002894099452532828,
      "learning_rate": 4.015444015444015e-06,
      "loss": 0.0085,
      "step": 24810
    },
    {
      "epoch": 31.01886743886744,
      "grad_norm": 0.00047004703083075583,
      "learning_rate": 4.012584012584013e-06,
      "loss": 0.6427,
      "step": 24820
    },
    {
      "epoch": 31.01912483912484,
      "grad_norm": 0.005355236120522022,
      "learning_rate": 4.00972400972401e-06,
      "loss": 0.0008,
      "step": 24830
    },
    {
      "epoch": 31.01938223938224,
      "grad_norm": 9.087779998779297,
      "learning_rate": 4.006864006864007e-06,
      "loss": 0.4508,
      "step": 24840
    },
    {
      "epoch": 31.01963963963964,
      "grad_norm": 0.08665086328983307,
      "learning_rate": 4.004004004004005e-06,
      "loss": 1.1483,
      "step": 24850
    },
    {
      "epoch": 31.01989703989704,
      "grad_norm": 0.0019484072690829635,
      "learning_rate": 4.001144001144002e-06,
      "loss": 1.7687,
      "step": 24860
    },
    {
      "epoch": 31.02,
      "eval_accuracy": 0.8695652173913043,
      "eval_loss": 0.6157187819480896,
      "eval_runtime": 13.4004,
      "eval_samples_per_second": 3.433,
      "eval_steps_per_second": 3.433,
      "step": 24864
    },
    {
      "epoch": 32.00015444015444,
      "grad_norm": 0.020324936136603355,
      "learning_rate": 3.998283998283999e-06,
      "loss": 0.7948,
      "step": 24870
    },
    {
      "epoch": 32.00041184041184,
      "grad_norm": 0.0005573392263613641,
      "learning_rate": 3.995423995423996e-06,
      "loss": 0.0005,
      "step": 24880
    },
    {
      "epoch": 32.00066924066924,
      "grad_norm": 0.022215578705072403,
      "learning_rate": 3.992563992563993e-06,
      "loss": 0.0014,
      "step": 24890
    },
    {
      "epoch": 32.00092664092664,
      "grad_norm": 0.2555524408817291,
      "learning_rate": 3.98970398970399e-06,
      "loss": 0.4709,
      "step": 24900
    },
    {
      "epoch": 32.00118404118404,
      "grad_norm": 0.0002522377180866897,
      "learning_rate": 3.986843986843988e-06,
      "loss": 0.5217,
      "step": 24910
    },
    {
      "epoch": 32.00144144144144,
      "grad_norm": 0.0034697242081165314,
      "learning_rate": 3.983983983983984e-06,
      "loss": 1.5856,
      "step": 24920
    },
    {
      "epoch": 32.00169884169884,
      "grad_norm": 0.0024689610581845045,
      "learning_rate": 3.9811239811239814e-06,
      "loss": 0.0005,
      "step": 24930
    },
    {
      "epoch": 32.00195624195624,
      "grad_norm": 0.00033844856079667807,
      "learning_rate": 3.978263978263979e-06,
      "loss": 0.0468,
      "step": 24940
    },
    {
      "epoch": 32.00221364221364,
      "grad_norm": 1.361294150352478,
      "learning_rate": 3.975403975403976e-06,
      "loss": 0.3848,
      "step": 24950
    },
    {
      "epoch": 32.00247104247104,
      "grad_norm": 0.0002234355779364705,
      "learning_rate": 3.9725439725439725e-06,
      "loss": 0.001,
      "step": 24960
    },
    {
      "epoch": 32.00272844272844,
      "grad_norm": 105.61128997802734,
      "learning_rate": 3.96968396968397e-06,
      "loss": 0.5425,
      "step": 24970
    },
    {
      "epoch": 32.002985842985844,
      "grad_norm": 0.0008264643256552517,
      "learning_rate": 3.966823966823967e-06,
      "loss": 0.0005,
      "step": 24980
    },
    {
      "epoch": 32.00324324324324,
      "grad_norm": 0.0004085392283741385,
      "learning_rate": 3.9639639639639645e-06,
      "loss": 0.8064,
      "step": 24990
    },
    {
      "epoch": 32.003500643500644,
      "grad_norm": 0.009747138246893883,
      "learning_rate": 3.961103961103962e-06,
      "loss": 0.2604,
      "step": 25000
    },
    {
      "epoch": 32.00375804375804,
      "grad_norm": 0.002869596006348729,
      "learning_rate": 3.958243958243958e-06,
      "loss": 0.716,
      "step": 25010
    },
    {
      "epoch": 32.004015444015444,
      "grad_norm": 0.008356648497283459,
      "learning_rate": 3.955383955383956e-06,
      "loss": 0.6306,
      "step": 25020
    },
    {
      "epoch": 32.00427284427285,
      "grad_norm": 0.0006081182509660721,
      "learning_rate": 3.952523952523953e-06,
      "loss": 0.0037,
      "step": 25030
    },
    {
      "epoch": 32.004530244530244,
      "grad_norm": 0.0005686648073606193,
      "learning_rate": 3.94966394966395e-06,
      "loss": 0.0006,
      "step": 25040
    },
    {
      "epoch": 32.00478764478765,
      "grad_norm": 0.0034270593896508217,
      "learning_rate": 3.946803946803947e-06,
      "loss": 0.0019,
      "step": 25050
    },
    {
      "epoch": 32.005045045045044,
      "grad_norm": 0.023503242060542107,
      "learning_rate": 3.943943943943944e-06,
      "loss": 0.0014,
      "step": 25060
    },
    {
      "epoch": 32.00530244530245,
      "grad_norm": 59.9887580871582,
      "learning_rate": 3.941083941083941e-06,
      "loss": 1.1883,
      "step": 25070
    },
    {
      "epoch": 32.005559845559844,
      "grad_norm": 0.11845483630895615,
      "learning_rate": 3.938223938223939e-06,
      "loss": 0.001,
      "step": 25080
    },
    {
      "epoch": 32.00581724581725,
      "grad_norm": 0.020497610792517662,
      "learning_rate": 3.935363935363936e-06,
      "loss": 0.3378,
      "step": 25090
    },
    {
      "epoch": 32.006074646074644,
      "grad_norm": 0.007013235706835985,
      "learning_rate": 3.932503932503932e-06,
      "loss": 1.4059,
      "step": 25100
    },
    {
      "epoch": 32.00633204633205,
      "grad_norm": 0.02268589660525322,
      "learning_rate": 3.92964392964393e-06,
      "loss": 0.7253,
      "step": 25110
    },
    {
      "epoch": 32.006589446589444,
      "grad_norm": 0.00040192139567807317,
      "learning_rate": 3.926783926783927e-06,
      "loss": 1.0919,
      "step": 25120
    },
    {
      "epoch": 32.00684684684685,
      "grad_norm": 0.00022598865325562656,
      "learning_rate": 3.923923923923924e-06,
      "loss": 0.0539,
      "step": 25130
    },
    {
      "epoch": 32.007104247104245,
      "grad_norm": 0.029369860887527466,
      "learning_rate": 3.921063921063921e-06,
      "loss": 0.0015,
      "step": 25140
    },
    {
      "epoch": 32.00736164736165,
      "grad_norm": 0.0013951669679954648,
      "learning_rate": 3.918203918203918e-06,
      "loss": 0.003,
      "step": 25150
    },
    {
      "epoch": 32.007619047619045,
      "grad_norm": 253.09658813476562,
      "learning_rate": 3.9153439153439155e-06,
      "loss": 0.536,
      "step": 25160
    },
    {
      "epoch": 32.00787644787645,
      "grad_norm": 155.09249877929688,
      "learning_rate": 3.912483912483913e-06,
      "loss": 0.9932,
      "step": 25170
    },
    {
      "epoch": 32.008133848133845,
      "grad_norm": 0.008935080841183662,
      "learning_rate": 3.90962390962391e-06,
      "loss": 0.5176,
      "step": 25180
    },
    {
      "epoch": 32.00839124839125,
      "grad_norm": 0.010084914974868298,
      "learning_rate": 3.9067639067639066e-06,
      "loss": 0.5228,
      "step": 25190
    },
    {
      "epoch": 32.00864864864865,
      "grad_norm": 107.02832794189453,
      "learning_rate": 3.903903903903904e-06,
      "loss": 1.1572,
      "step": 25200
    },
    {
      "epoch": 32.00890604890605,
      "grad_norm": 0.013532307930290699,
      "learning_rate": 3.901043901043902e-06,
      "loss": 0.5614,
      "step": 25210
    },
    {
      "epoch": 32.00916344916345,
      "grad_norm": 739.5260620117188,
      "learning_rate": 3.8981838981838985e-06,
      "loss": 0.4832,
      "step": 25220
    },
    {
      "epoch": 32.00942084942085,
      "grad_norm": 113.25529479980469,
      "learning_rate": 3.895323895323896e-06,
      "loss": 1.1142,
      "step": 25230
    },
    {
      "epoch": 32.00967824967825,
      "grad_norm": 0.4678283929824829,
      "learning_rate": 3.892463892463893e-06,
      "loss": 0.5052,
      "step": 25240
    },
    {
      "epoch": 32.00993564993565,
      "grad_norm": 229.15992736816406,
      "learning_rate": 3.8896038896038905e-06,
      "loss": 0.5591,
      "step": 25250
    },
    {
      "epoch": 32.01019305019305,
      "grad_norm": 135.36276245117188,
      "learning_rate": 3.886743886743887e-06,
      "loss": 0.8544,
      "step": 25260
    },
    {
      "epoch": 32.01045045045045,
      "grad_norm": 98.59178161621094,
      "learning_rate": 3.883883883883884e-06,
      "loss": 1.0966,
      "step": 25270
    },
    {
      "epoch": 32.01070785070785,
      "grad_norm": 56.150657653808594,
      "learning_rate": 3.8810238810238816e-06,
      "loss": 0.2389,
      "step": 25280
    },
    {
      "epoch": 32.01096525096525,
      "grad_norm": 0.33948537707328796,
      "learning_rate": 3.878163878163879e-06,
      "loss": 0.2457,
      "step": 25290
    },
    {
      "epoch": 32.01122265122265,
      "grad_norm": 0.014132946729660034,
      "learning_rate": 3.875303875303875e-06,
      "loss": 0.001,
      "step": 25300
    },
    {
      "epoch": 32.01148005148005,
      "grad_norm": 0.06854890286922455,
      "learning_rate": 3.872443872443873e-06,
      "loss": 0.0011,
      "step": 25310
    },
    {
      "epoch": 32.01173745173745,
      "grad_norm": 0.43976348638534546,
      "learning_rate": 3.86958386958387e-06,
      "loss": 0.648,
      "step": 25320
    },
    {
      "epoch": 32.01199485199485,
      "grad_norm": 0.00041257721022702754,
      "learning_rate": 3.866723866723867e-06,
      "loss": 0.001,
      "step": 25330
    },
    {
      "epoch": 32.01225225225225,
      "grad_norm": 0.00022966216783970594,
      "learning_rate": 3.863863863863865e-06,
      "loss": 0.4409,
      "step": 25340
    },
    {
      "epoch": 32.01250965250965,
      "grad_norm": 0.000402516481699422,
      "learning_rate": 3.861003861003861e-06,
      "loss": 0.0005,
      "step": 25350
    },
    {
      "epoch": 32.01276705276705,
      "grad_norm": 0.010474207811057568,
      "learning_rate": 3.858143858143858e-06,
      "loss": 0.579,
      "step": 25360
    },
    {
      "epoch": 32.01302445302445,
      "grad_norm": 0.21792848408222198,
      "learning_rate": 3.855283855283856e-06,
      "loss": 0.8084,
      "step": 25370
    },
    {
      "epoch": 32.01328185328185,
      "grad_norm": 0.009377431124448776,
      "learning_rate": 3.852423852423853e-06,
      "loss": 0.4596,
      "step": 25380
    },
    {
      "epoch": 32.01353925353926,
      "grad_norm": 0.01633077673614025,
      "learning_rate": 3.8495638495638495e-06,
      "loss": 0.0007,
      "step": 25390
    },
    {
      "epoch": 32.01379665379665,
      "grad_norm": 0.10646045953035355,
      "learning_rate": 3.846703846703847e-06,
      "loss": 0.3559,
      "step": 25400
    },
    {
      "epoch": 32.01405405405406,
      "grad_norm": 0.03003956936299801,
      "learning_rate": 3.843843843843844e-06,
      "loss": 0.0277,
      "step": 25410
    },
    {
      "epoch": 32.01431145431145,
      "grad_norm": 0.0002730604028329253,
      "learning_rate": 3.8409838409838415e-06,
      "loss": 0.1572,
      "step": 25420
    },
    {
      "epoch": 32.01456885456886,
      "grad_norm": 0.0002827415300998837,
      "learning_rate": 3.838123838123839e-06,
      "loss": 0.006,
      "step": 25430
    },
    {
      "epoch": 32.014826254826254,
      "grad_norm": 0.01532386802136898,
      "learning_rate": 3.835263835263835e-06,
      "loss": 0.0006,
      "step": 25440
    },
    {
      "epoch": 32.01508365508366,
      "grad_norm": 0.0004180877876933664,
      "learning_rate": 3.8324038324038326e-06,
      "loss": 0.0003,
      "step": 25450
    },
    {
      "epoch": 32.015341055341054,
      "grad_norm": 0.00066907680593431,
      "learning_rate": 3.82954382954383e-06,
      "loss": 0.4843,
      "step": 25460
    },
    {
      "epoch": 32.01559845559846,
      "grad_norm": 0.00021530227968469262,
      "learning_rate": 3.826683826683827e-06,
      "loss": 2.1196,
      "step": 25470
    },
    {
      "epoch": 32.015855855855854,
      "grad_norm": 0.05011071264743805,
      "learning_rate": 3.823823823823824e-06,
      "loss": 0.0141,
      "step": 25480
    },
    {
      "epoch": 32.01611325611326,
      "grad_norm": 0.005272657610476017,
      "learning_rate": 3.820963820963821e-06,
      "loss": 0.0011,
      "step": 25490
    },
    {
      "epoch": 32.016370656370654,
      "grad_norm": 0.08962148427963257,
      "learning_rate": 3.818103818103818e-06,
      "loss": 0.0014,
      "step": 25500
    },
    {
      "epoch": 32.01662805662806,
      "grad_norm": 0.09578029066324234,
      "learning_rate": 3.815243815243816e-06,
      "loss": 0.675,
      "step": 25510
    },
    {
      "epoch": 32.016885456885454,
      "grad_norm": 0.04941503703594208,
      "learning_rate": 3.8123838123838125e-06,
      "loss": 0.0019,
      "step": 25520
    },
    {
      "epoch": 32.01714285714286,
      "grad_norm": 0.3917921185493469,
      "learning_rate": 3.80952380952381e-06,
      "loss": 0.0006,
      "step": 25530
    },
    {
      "epoch": 32.017400257400254,
      "grad_norm": 0.053205542266368866,
      "learning_rate": 3.8066638066638067e-06,
      "loss": 0.0003,
      "step": 25540
    },
    {
      "epoch": 32.01765765765766,
      "grad_norm": 0.0002905249421019107,
      "learning_rate": 3.803803803803804e-06,
      "loss": 0.8974,
      "step": 25550
    },
    {
      "epoch": 32.017915057915054,
      "grad_norm": 105.43701934814453,
      "learning_rate": 3.800943800943801e-06,
      "loss": 1.1082,
      "step": 25560
    },
    {
      "epoch": 32.01817245817246,
      "grad_norm": 0.18681477010250092,
      "learning_rate": 3.7980837980837982e-06,
      "loss": 0.0003,
      "step": 25570
    },
    {
      "epoch": 32.01842985842986,
      "grad_norm": 0.0005921496194787323,
      "learning_rate": 3.795223795223796e-06,
      "loss": 0.0322,
      "step": 25580
    },
    {
      "epoch": 32.01868725868726,
      "grad_norm": 0.0002967441687360406,
      "learning_rate": 3.792363792363793e-06,
      "loss": 0.0008,
      "step": 25590
    },
    {
      "epoch": 32.01894465894466,
      "grad_norm": 0.03639024868607521,
      "learning_rate": 3.78950378950379e-06,
      "loss": 0.1673,
      "step": 25600
    },
    {
      "epoch": 32.01920205920206,
      "grad_norm": 0.00020979787223041058,
      "learning_rate": 3.786643786643787e-06,
      "loss": 0.6608,
      "step": 25610
    },
    {
      "epoch": 32.01945945945946,
      "grad_norm": 0.0004491186118684709,
      "learning_rate": 3.7837837837837844e-06,
      "loss": 0.0006,
      "step": 25620
    },
    {
      "epoch": 32.01971685971686,
      "grad_norm": 0.14153051376342773,
      "learning_rate": 3.7809237809237813e-06,
      "loss": 0.4424,
      "step": 25630
    },
    {
      "epoch": 32.01997425997426,
      "grad_norm": 0.1150650754570961,
      "learning_rate": 3.7780637780637786e-06,
      "loss": 0.0036,
      "step": 25640
    },
    {
      "epoch": 32.02,
      "eval_accuracy": 0.8478260869565217,
      "eval_loss": 1.0988526344299316,
      "eval_runtime": 13.5282,
      "eval_samples_per_second": 3.4,
      "eval_steps_per_second": 3.4,
      "step": 25641
    },
    {
      "epoch": 33.00023166023166,
      "grad_norm": 0.0002022027620114386,
      "learning_rate": 3.7752037752037755e-06,
      "loss": 0.0001,
      "step": 25650
    },
    {
      "epoch": 33.00048906048906,
      "grad_norm": 0.6226126551628113,
      "learning_rate": 3.772343772343773e-06,
      "loss": 0.6352,
      "step": 25660
    },
    {
      "epoch": 33.00074646074646,
      "grad_norm": 0.00021000955894123763,
      "learning_rate": 3.76948376948377e-06,
      "loss": 0.0012,
      "step": 25670
    },
    {
      "epoch": 33.00100386100386,
      "grad_norm": 0.0005035571521148086,
      "learning_rate": 3.766623766623767e-06,
      "loss": 0.0003,
      "step": 25680
    },
    {
      "epoch": 33.00126126126126,
      "grad_norm": 0.0008679556776769459,
      "learning_rate": 3.7637637637637643e-06,
      "loss": 0.0001,
      "step": 25690
    },
    {
      "epoch": 33.00151866151866,
      "grad_norm": 0.000201143222511746,
      "learning_rate": 3.7609037609037612e-06,
      "loss": 0.0049,
      "step": 25700
    },
    {
      "epoch": 33.00177606177606,
      "grad_norm": 0.00031288861646316946,
      "learning_rate": 3.7580437580437585e-06,
      "loss": 0.0017,
      "step": 25710
    },
    {
      "epoch": 33.00203346203346,
      "grad_norm": 0.0003889285144396126,
      "learning_rate": 3.7551837551837554e-06,
      "loss": 0.0616,
      "step": 25720
    },
    {
      "epoch": 33.00229086229086,
      "grad_norm": 0.034657277166843414,
      "learning_rate": 3.7523237523237528e-06,
      "loss": 0.7658,
      "step": 25730
    },
    {
      "epoch": 33.00254826254826,
      "grad_norm": 0.0001886966492747888,
      "learning_rate": 3.7494637494637496e-06,
      "loss": 0.0004,
      "step": 25740
    },
    {
      "epoch": 33.00280566280566,
      "grad_norm": 0.007791679818183184,
      "learning_rate": 3.746603746603747e-06,
      "loss": 0.0003,
      "step": 25750
    },
    {
      "epoch": 33.00306306306306,
      "grad_norm": 0.013618108816444874,
      "learning_rate": 3.743743743743744e-06,
      "loss": 0.2169,
      "step": 25760
    },
    {
      "epoch": 33.00332046332046,
      "grad_norm": 0.00634233420714736,
      "learning_rate": 3.740883740883741e-06,
      "loss": 0.0002,
      "step": 25770
    },
    {
      "epoch": 33.00357786357787,
      "grad_norm": 0.0011721225455403328,
      "learning_rate": 3.7380237380237385e-06,
      "loss": 0.4377,
      "step": 25780
    },
    {
      "epoch": 33.00383526383526,
      "grad_norm": 0.0002262504567625001,
      "learning_rate": 3.7351637351637354e-06,
      "loss": 0.0301,
      "step": 25790
    },
    {
      "epoch": 33.00409266409267,
      "grad_norm": 0.00018854282097890973,
      "learning_rate": 3.7323037323037327e-06,
      "loss": 0.8319,
      "step": 25800
    },
    {
      "epoch": 33.004350064350064,
      "grad_norm": 0.13399730622768402,
      "learning_rate": 3.7294437294437296e-06,
      "loss": 0.092,
      "step": 25810
    },
    {
      "epoch": 33.00460746460747,
      "grad_norm": 0.02621803991496563,
      "learning_rate": 3.726583726583727e-06,
      "loss": 0.4453,
      "step": 25820
    },
    {
      "epoch": 33.004864864864864,
      "grad_norm": 2.0014894008636475,
      "learning_rate": 3.723723723723724e-06,
      "loss": 0.1793,
      "step": 25830
    },
    {
      "epoch": 33.00512226512227,
      "grad_norm": 214.54891967773438,
      "learning_rate": 3.720863720863721e-06,
      "loss": 0.0317,
      "step": 25840
    },
    {
      "epoch": 33.005379665379664,
      "grad_norm": 0.005550840869545937,
      "learning_rate": 3.718003718003718e-06,
      "loss": 0.7145,
      "step": 25850
    },
    {
      "epoch": 33.00563706563707,
      "grad_norm": 0.007037818897515535,
      "learning_rate": 3.7151437151437153e-06,
      "loss": 0.7924,
      "step": 25860
    },
    {
      "epoch": 33.005894465894464,
      "grad_norm": 0.009896584786474705,
      "learning_rate": 3.7122837122837122e-06,
      "loss": 0.4399,
      "step": 25870
    },
    {
      "epoch": 33.00615186615187,
      "grad_norm": 0.17795123159885406,
      "learning_rate": 3.7094237094237095e-06,
      "loss": 1.1506,
      "step": 25880
    },
    {
      "epoch": 33.006409266409264,
      "grad_norm": 0.0003295901115052402,
      "learning_rate": 3.706563706563707e-06,
      "loss": 0.7586,
      "step": 25890
    },
    {
      "epoch": 33.00666666666667,
      "grad_norm": 0.006462703924626112,
      "learning_rate": 3.7037037037037037e-06,
      "loss": 0.0015,
      "step": 25900
    },
    {
      "epoch": 33.006924066924064,
      "grad_norm": 0.0007809298695065081,
      "learning_rate": 3.700843700843701e-06,
      "loss": 0.0002,
      "step": 25910
    },
    {
      "epoch": 33.00718146718147,
      "grad_norm": 0.10808756947517395,
      "learning_rate": 3.697983697983698e-06,
      "loss": 1.0891,
      "step": 25920
    },
    {
      "epoch": 33.007438867438864,
      "grad_norm": 0.0005664303316734731,
      "learning_rate": 3.6951236951236953e-06,
      "loss": 1.0756,
      "step": 25930
    },
    {
      "epoch": 33.00769626769627,
      "grad_norm": 0.39256948232650757,
      "learning_rate": 3.692263692263692e-06,
      "loss": 0.6975,
      "step": 25940
    },
    {
      "epoch": 33.007953667953664,
      "grad_norm": 0.05600004643201828,
      "learning_rate": 3.68940368940369e-06,
      "loss": 0.0004,
      "step": 25950
    },
    {
      "epoch": 33.00821106821107,
      "grad_norm": 0.0007221155683510005,
      "learning_rate": 3.6865436865436872e-06,
      "loss": 1.3454,
      "step": 25960
    },
    {
      "epoch": 33.00846846846847,
      "grad_norm": 0.0249467883259058,
      "learning_rate": 3.683683683683684e-06,
      "loss": 0.0002,
      "step": 25970
    },
    {
      "epoch": 33.00872586872587,
      "grad_norm": 0.01132212020456791,
      "learning_rate": 3.6808236808236814e-06,
      "loss": 0.8555,
      "step": 25980
    },
    {
      "epoch": 33.00898326898327,
      "grad_norm": 0.20580242574214935,
      "learning_rate": 3.6779636779636783e-06,
      "loss": 0.0009,
      "step": 25990
    },
    {
      "epoch": 33.00924066924067,
      "grad_norm": 0.004588236566632986,
      "learning_rate": 3.6751036751036756e-06,
      "loss": 0.0003,
      "step": 26000
    },
    {
      "epoch": 33.00949806949807,
      "grad_norm": 51.92314529418945,
      "learning_rate": 3.6722436722436725e-06,
      "loss": 0.768,
      "step": 26010
    },
    {
      "epoch": 33.00975546975547,
      "grad_norm": 0.023511553183197975,
      "learning_rate": 3.66938366938367e-06,
      "loss": 0.1939,
      "step": 26020
    },
    {
      "epoch": 33.01001287001287,
      "grad_norm": 0.00020286561630200595,
      "learning_rate": 3.666523666523667e-06,
      "loss": 1.453,
      "step": 26030
    },
    {
      "epoch": 33.01027027027027,
      "grad_norm": 0.03150779381394386,
      "learning_rate": 3.663663663663664e-06,
      "loss": 0.5429,
      "step": 26040
    },
    {
      "epoch": 33.01052767052767,
      "grad_norm": 1292.7117919921875,
      "learning_rate": 3.6608036608036614e-06,
      "loss": 0.4714,
      "step": 26050
    },
    {
      "epoch": 33.01078507078507,
      "grad_norm": 0.0039928508922457695,
      "learning_rate": 3.6579436579436583e-06,
      "loss": 0.0025,
      "step": 26060
    },
    {
      "epoch": 33.01104247104247,
      "grad_norm": 0.004026371985673904,
      "learning_rate": 3.6550836550836556e-06,
      "loss": 0.2678,
      "step": 26070
    },
    {
      "epoch": 33.01129987129987,
      "grad_norm": 0.001805619103834033,
      "learning_rate": 3.6522236522236525e-06,
      "loss": 0.0078,
      "step": 26080
    },
    {
      "epoch": 33.01155727155727,
      "grad_norm": 0.0019118794007226825,
      "learning_rate": 3.6493636493636498e-06,
      "loss": 0.5989,
      "step": 26090
    },
    {
      "epoch": 33.01181467181467,
      "grad_norm": 0.003230776870623231,
      "learning_rate": 3.6465036465036467e-06,
      "loss": 0.0005,
      "step": 26100
    },
    {
      "epoch": 33.01207207207207,
      "grad_norm": 0.00032252943492494524,
      "learning_rate": 3.643643643643644e-06,
      "loss": 0.0008,
      "step": 26110
    },
    {
      "epoch": 33.01232947232947,
      "grad_norm": 0.0008710320107638836,
      "learning_rate": 3.6407836407836413e-06,
      "loss": 0.4259,
      "step": 26120
    },
    {
      "epoch": 33.01258687258687,
      "grad_norm": 0.014103787951171398,
      "learning_rate": 3.637923637923638e-06,
      "loss": 0.001,
      "step": 26130
    },
    {
      "epoch": 33.012844272844276,
      "grad_norm": 0.06376767158508301,
      "learning_rate": 3.6350636350636355e-06,
      "loss": 0.0009,
      "step": 26140
    },
    {
      "epoch": 33.01310167310167,
      "grad_norm": 0.0013607190921902657,
      "learning_rate": 3.6322036322036324e-06,
      "loss": 0.066,
      "step": 26150
    },
    {
      "epoch": 33.01335907335908,
      "grad_norm": 0.00027021009009331465,
      "learning_rate": 3.6293436293436297e-06,
      "loss": 0.0029,
      "step": 26160
    },
    {
      "epoch": 33.01361647361647,
      "grad_norm": 0.01180395856499672,
      "learning_rate": 3.6264836264836266e-06,
      "loss": 0.0004,
      "step": 26170
    },
    {
      "epoch": 33.01387387387388,
      "grad_norm": 0.04038818180561066,
      "learning_rate": 3.623623623623624e-06,
      "loss": 0.0004,
      "step": 26180
    },
    {
      "epoch": 33.01413127413127,
      "grad_norm": 0.016394222155213356,
      "learning_rate": 3.620763620763621e-06,
      "loss": 0.7459,
      "step": 26190
    },
    {
      "epoch": 33.01438867438868,
      "grad_norm": 577.7449340820312,
      "learning_rate": 3.617903617903618e-06,
      "loss": 0.2242,
      "step": 26200
    },
    {
      "epoch": 33.01464607464607,
      "grad_norm": 0.015523134730756283,
      "learning_rate": 3.615043615043615e-06,
      "loss": 0.0001,
      "step": 26210
    },
    {
      "epoch": 33.01490347490348,
      "grad_norm": 0.02036185935139656,
      "learning_rate": 3.6121836121836124e-06,
      "loss": 1.3635,
      "step": 26220
    },
    {
      "epoch": 33.01516087516087,
      "grad_norm": 45.18238830566406,
      "learning_rate": 3.6093236093236097e-06,
      "loss": 2.0981,
      "step": 26230
    },
    {
      "epoch": 33.01541827541828,
      "grad_norm": 0.09267007559537888,
      "learning_rate": 3.6064636064636066e-06,
      "loss": 0.639,
      "step": 26240
    },
    {
      "epoch": 33.01567567567567,
      "grad_norm": 1.0800601243972778,
      "learning_rate": 3.603603603603604e-06,
      "loss": 0.0284,
      "step": 26250
    },
    {
      "epoch": 33.01593307593308,
      "grad_norm": 0.0005480324034579098,
      "learning_rate": 3.6007436007436008e-06,
      "loss": 0.9278,
      "step": 26260
    },
    {
      "epoch": 33.016190476190474,
      "grad_norm": 0.27580541372299194,
      "learning_rate": 3.597883597883598e-06,
      "loss": 0.6449,
      "step": 26270
    },
    {
      "epoch": 33.01644787644788,
      "grad_norm": 0.0772259458899498,
      "learning_rate": 3.595023595023595e-06,
      "loss": 0.0012,
      "step": 26280
    },
    {
      "epoch": 33.016705276705274,
      "grad_norm": 0.036952629685401917,
      "learning_rate": 3.5921635921635923e-06,
      "loss": 0.0406,
      "step": 26290
    },
    {
      "epoch": 33.01696267696268,
      "grad_norm": 0.0002245551295345649,
      "learning_rate": 3.589303589303589e-06,
      "loss": 0.001,
      "step": 26300
    },
    {
      "epoch": 33.017220077220074,
      "grad_norm": 3.1717610359191895,
      "learning_rate": 3.5864435864435865e-06,
      "loss": 0.001,
      "step": 26310
    },
    {
      "epoch": 33.01747747747748,
      "grad_norm": 0.0382111519575119,
      "learning_rate": 3.5835835835835834e-06,
      "loss": 0.4692,
      "step": 26320
    },
    {
      "epoch": 33.01773487773488,
      "grad_norm": 0.0640905350446701,
      "learning_rate": 3.580723580723581e-06,
      "loss": 0.5739,
      "step": 26330
    },
    {
      "epoch": 33.01799227799228,
      "grad_norm": 0.00023219177091959864,
      "learning_rate": 3.5778635778635785e-06,
      "loss": 0.9233,
      "step": 26340
    },
    {
      "epoch": 33.01824967824968,
      "grad_norm": 0.040953151881694794,
      "learning_rate": 3.5750035750035753e-06,
      "loss": 0.0011,
      "step": 26350
    },
    {
      "epoch": 33.01850707850708,
      "grad_norm": 16.874282836914062,
      "learning_rate": 3.5721435721435727e-06,
      "loss": 0.6596,
      "step": 26360
    },
    {
      "epoch": 33.01876447876448,
      "grad_norm": 0.0005724704824388027,
      "learning_rate": 3.56928356928357e-06,
      "loss": 0.0016,
      "step": 26370
    },
    {
      "epoch": 33.01902187902188,
      "grad_norm": 401.7337341308594,
      "learning_rate": 3.566423566423567e-06,
      "loss": 0.4095,
      "step": 26380
    },
    {
      "epoch": 33.01927927927928,
      "grad_norm": 0.0027081759180873632,
      "learning_rate": 3.563563563563564e-06,
      "loss": 1.1057,
      "step": 26390
    },
    {
      "epoch": 33.01953667953668,
      "grad_norm": 0.11730697751045227,
      "learning_rate": 3.560703560703561e-06,
      "loss": 1.2431,
      "step": 26400
    },
    {
      "epoch": 33.01979407979408,
      "grad_norm": 0.0015030332142487168,
      "learning_rate": 3.5578435578435584e-06,
      "loss": 0.6257,
      "step": 26410
    },
    {
      "epoch": 33.02,
      "eval_accuracy": 0.8260869565217391,
      "eval_loss": 0.9700731039047241,
      "eval_runtime": 13.4331,
      "eval_samples_per_second": 3.424,
      "eval_steps_per_second": 3.424,
      "step": 26418
    },
    {
      "epoch": 34.00005148005148,
      "grad_norm": 0.21256451308727264,
      "learning_rate": 3.5549835549835553e-06,
      "loss": 1.1551,
      "step": 26420
    },
    {
      "epoch": 34.00030888030888,
      "grad_norm": 0.005612839478999376,
      "learning_rate": 3.5521235521235526e-06,
      "loss": 0.5618,
      "step": 26430
    },
    {
      "epoch": 34.00056628056628,
      "grad_norm": 0.0003517004370223731,
      "learning_rate": 3.5492635492635495e-06,
      "loss": 0.0014,
      "step": 26440
    },
    {
      "epoch": 34.00082368082368,
      "grad_norm": 0.01810491643846035,
      "learning_rate": 3.546403546403547e-06,
      "loss": 0.6831,
      "step": 26450
    },
    {
      "epoch": 34.00108108108108,
      "grad_norm": 0.00031709542963653803,
      "learning_rate": 3.5435435435435437e-06,
      "loss": 0.0011,
      "step": 26460
    },
    {
      "epoch": 34.00133848133848,
      "grad_norm": 0.0002846124698407948,
      "learning_rate": 3.540683540683541e-06,
      "loss": 0.4127,
      "step": 26470
    },
    {
      "epoch": 34.00159588159588,
      "grad_norm": 0.00024031595967244357,
      "learning_rate": 3.5378235378235383e-06,
      "loss": 0.0572,
      "step": 26480
    },
    {
      "epoch": 34.00185328185328,
      "grad_norm": 42.76933288574219,
      "learning_rate": 3.5349635349635352e-06,
      "loss": 0.0024,
      "step": 26490
    },
    {
      "epoch": 34.00211068211068,
      "grad_norm": 0.061154335737228394,
      "learning_rate": 3.5321035321035326e-06,
      "loss": 0.0007,
      "step": 26500
    },
    {
      "epoch": 34.00236808236808,
      "grad_norm": 0.0024257488548755646,
      "learning_rate": 3.5292435292435294e-06,
      "loss": 0.0002,
      "step": 26510
    },
    {
      "epoch": 34.00262548262548,
      "grad_norm": 0.0003811272617895156,
      "learning_rate": 3.5263835263835268e-06,
      "loss": 0.0004,
      "step": 26520
    },
    {
      "epoch": 34.002882882882886,
      "grad_norm": 0.0008752444991841912,
      "learning_rate": 3.5235235235235237e-06,
      "loss": 0.4638,
      "step": 26530
    },
    {
      "epoch": 34.00314028314028,
      "grad_norm": 0.001001368393190205,
      "learning_rate": 3.520663520663521e-06,
      "loss": 0.6171,
      "step": 26540
    },
    {
      "epoch": 34.00339768339769,
      "grad_norm": 0.13201755285263062,
      "learning_rate": 3.517803517803518e-06,
      "loss": 1.412,
      "step": 26550
    },
    {
      "epoch": 34.00365508365508,
      "grad_norm": 0.00029380101477727294,
      "learning_rate": 3.514943514943515e-06,
      "loss": 0.0001,
      "step": 26560
    },
    {
      "epoch": 34.00391248391249,
      "grad_norm": 0.07130458950996399,
      "learning_rate": 3.512083512083512e-06,
      "loss": 0.4946,
      "step": 26570
    },
    {
      "epoch": 34.00416988416988,
      "grad_norm": 0.025920139625668526,
      "learning_rate": 3.5092235092235094e-06,
      "loss": 0.8401,
      "step": 26580
    },
    {
      "epoch": 34.00442728442729,
      "grad_norm": 97.6776351928711,
      "learning_rate": 3.5063635063635067e-06,
      "loss": 0.6201,
      "step": 26590
    },
    {
      "epoch": 34.00468468468468,
      "grad_norm": 0.0005312726716510952,
      "learning_rate": 3.5035035035035036e-06,
      "loss": 0.0005,
      "step": 26600
    },
    {
      "epoch": 34.00494208494209,
      "grad_norm": 0.012904675677418709,
      "learning_rate": 3.500643500643501e-06,
      "loss": 0.0123,
      "step": 26610
    },
    {
      "epoch": 34.00519948519948,
      "grad_norm": 0.006391404662281275,
      "learning_rate": 3.497783497783498e-06,
      "loss": 0.0015,
      "step": 26620
    },
    {
      "epoch": 34.00545688545689,
      "grad_norm": 0.009654326364398003,
      "learning_rate": 3.494923494923495e-06,
      "loss": 0.0014,
      "step": 26630
    },
    {
      "epoch": 34.005714285714284,
      "grad_norm": 0.002458939328789711,
      "learning_rate": 3.492063492063492e-06,
      "loss": 0.5645,
      "step": 26640
    },
    {
      "epoch": 34.00597168597169,
      "grad_norm": 0.05094308778643608,
      "learning_rate": 3.4892034892034893e-06,
      "loss": 0.0003,
      "step": 26650
    },
    {
      "epoch": 34.006229086229084,
      "grad_norm": 0.01629532128572464,
      "learning_rate": 3.4863434863434862e-06,
      "loss": 0.0007,
      "step": 26660
    },
    {
      "epoch": 34.00648648648649,
      "grad_norm": 0.001868433435447514,
      "learning_rate": 3.4834834834834835e-06,
      "loss": 0.001,
      "step": 26670
    },
    {
      "epoch": 34.006743886743884,
      "grad_norm": 0.024310791864991188,
      "learning_rate": 3.480623480623481e-06,
      "loss": 0.0005,
      "step": 26680
    },
    {
      "epoch": 34.00700128700129,
      "grad_norm": 0.0002190342784160748,
      "learning_rate": 3.4777634777634777e-06,
      "loss": 0.7477,
      "step": 26690
    },
    {
      "epoch": 34.007258687258684,
      "grad_norm": 0.00019727973267436028,
      "learning_rate": 3.4749034749034755e-06,
      "loss": 0.7275,
      "step": 26700
    },
    {
      "epoch": 34.00751608751609,
      "grad_norm": 90.7525634765625,
      "learning_rate": 3.4720434720434724e-06,
      "loss": 0.8832,
      "step": 26710
    },
    {
      "epoch": 34.00777348777349,
      "grad_norm": 476.1946716308594,
      "learning_rate": 3.4691834691834697e-06,
      "loss": 0.8877,
      "step": 26720
    },
    {
      "epoch": 34.00803088803089,
      "grad_norm": 0.012554511427879333,
      "learning_rate": 3.466323466323467e-06,
      "loss": 0.0015,
      "step": 26730
    },
    {
      "epoch": 34.00828828828829,
      "grad_norm": 0.17711642384529114,
      "learning_rate": 3.463463463463464e-06,
      "loss": 0.6558,
      "step": 26740
    },
    {
      "epoch": 34.00854568854569,
      "grad_norm": 0.00020986588788218796,
      "learning_rate": 3.4606034606034612e-06,
      "loss": 0.0043,
      "step": 26750
    },
    {
      "epoch": 34.00880308880309,
      "grad_norm": 0.11533048748970032,
      "learning_rate": 3.457743457743458e-06,
      "loss": 0.0004,
      "step": 26760
    },
    {
      "epoch": 34.00906048906049,
      "grad_norm": 0.0015373900532722473,
      "learning_rate": 3.4548834548834554e-06,
      "loss": 0.6428,
      "step": 26770
    },
    {
      "epoch": 34.00931788931789,
      "grad_norm": 0.0005574403912760317,
      "learning_rate": 3.4520234520234523e-06,
      "loss": 0.0005,
      "step": 26780
    },
    {
      "epoch": 34.00957528957529,
      "grad_norm": 0.006771887186914682,
      "learning_rate": 3.4491634491634496e-06,
      "loss": 0.0007,
      "step": 26790
    },
    {
      "epoch": 34.00983268983269,
      "grad_norm": 0.018115438520908356,
      "learning_rate": 3.4463034463034465e-06,
      "loss": 0.4471,
      "step": 26800
    },
    {
      "epoch": 34.01009009009009,
      "grad_norm": 0.0823565125465393,
      "learning_rate": 3.443443443443444e-06,
      "loss": 0.5876,
      "step": 26810
    },
    {
      "epoch": 34.01034749034749,
      "grad_norm": 0.01303765270859003,
      "learning_rate": 3.440583440583441e-06,
      "loss": 1.1356,
      "step": 26820
    },
    {
      "epoch": 34.01060489060489,
      "grad_norm": 0.0010142725659534335,
      "learning_rate": 3.437723437723438e-06,
      "loss": 0.0068,
      "step": 26830
    },
    {
      "epoch": 34.01086229086229,
      "grad_norm": 0.09757779538631439,
      "learning_rate": 3.4348634348634354e-06,
      "loss": 1.2388,
      "step": 26840
    },
    {
      "epoch": 34.01111969111969,
      "grad_norm": 0.053004637360572815,
      "learning_rate": 3.4320034320034323e-06,
      "loss": 0.0015,
      "step": 26850
    },
    {
      "epoch": 34.01137709137709,
      "grad_norm": 0.11356706917285919,
      "learning_rate": 3.4291434291434296e-06,
      "loss": 0.004,
      "step": 26860
    },
    {
      "epoch": 34.01163449163449,
      "grad_norm": 0.10561950504779816,
      "learning_rate": 3.4262834262834265e-06,
      "loss": 0.0004,
      "step": 26870
    },
    {
      "epoch": 34.01189189189189,
      "grad_norm": 0.0003318323288112879,
      "learning_rate": 3.423423423423424e-06,
      "loss": 0.0005,
      "step": 26880
    },
    {
      "epoch": 34.01214929214929,
      "grad_norm": 0.0002169922081520781,
      "learning_rate": 3.4205634205634207e-06,
      "loss": 0.0003,
      "step": 26890
    },
    {
      "epoch": 34.01240669240669,
      "grad_norm": 0.07491030544042587,
      "learning_rate": 3.417703417703418e-06,
      "loss": 0.0004,
      "step": 26900
    },
    {
      "epoch": 34.012664092664096,
      "grad_norm": 0.6812867522239685,
      "learning_rate": 3.414843414843415e-06,
      "loss": 0.0472,
      "step": 26910
    },
    {
      "epoch": 34.01292149292149,
      "grad_norm": 0.08314365893602371,
      "learning_rate": 3.411983411983412e-06,
      "loss": 0.0003,
      "step": 26920
    },
    {
      "epoch": 34.013178893178896,
      "grad_norm": 0.05190544202923775,
      "learning_rate": 3.4091234091234095e-06,
      "loss": 0.0892,
      "step": 26930
    },
    {
      "epoch": 34.01343629343629,
      "grad_norm": 0.03272266685962677,
      "learning_rate": 3.4062634062634064e-06,
      "loss": 0.0001,
      "step": 26940
    },
    {
      "epoch": 34.013693693693696,
      "grad_norm": 0.039729151874780655,
      "learning_rate": 3.4034034034034037e-06,
      "loss": 0.6702,
      "step": 26950
    },
    {
      "epoch": 34.01395109395109,
      "grad_norm": 0.00022988698037806898,
      "learning_rate": 3.4005434005434006e-06,
      "loss": 1.3983,
      "step": 26960
    },
    {
      "epoch": 34.014208494208496,
      "grad_norm": 0.005083046387881041,
      "learning_rate": 3.397683397683398e-06,
      "loss": 0.0006,
      "step": 26970
    },
    {
      "epoch": 34.01446589446589,
      "grad_norm": 271.5005187988281,
      "learning_rate": 3.394823394823395e-06,
      "loss": 1.3513,
      "step": 26980
    },
    {
      "epoch": 34.0147232947233,
      "grad_norm": 109.3269271850586,
      "learning_rate": 3.391963391963392e-06,
      "loss": 0.5311,
      "step": 26990
    },
    {
      "epoch": 34.01498069498069,
      "grad_norm": 0.0002787598641589284,
      "learning_rate": 3.389103389103389e-06,
      "loss": 0.021,
      "step": 27000
    },
    {
      "epoch": 34.0152380952381,
      "grad_norm": 0.00025051116244867444,
      "learning_rate": 3.3862433862433864e-06,
      "loss": 0.0021,
      "step": 27010
    },
    {
      "epoch": 34.01549549549549,
      "grad_norm": 0.0006493125110864639,
      "learning_rate": 3.3833833833833833e-06,
      "loss": 0.4824,
      "step": 27020
    },
    {
      "epoch": 34.0157528957529,
      "grad_norm": 0.0007675174274481833,
      "learning_rate": 3.3805233805233806e-06,
      "loss": 0.1143,
      "step": 27030
    },
    {
      "epoch": 34.01601029601029,
      "grad_norm": 0.1199033185839653,
      "learning_rate": 3.377663377663378e-06,
      "loss": 0.0004,
      "step": 27040
    },
    {
      "epoch": 34.0162676962677,
      "grad_norm": 0.0002853220503311604,
      "learning_rate": 3.3748033748033748e-06,
      "loss": 0.0003,
      "step": 27050
    },
    {
      "epoch": 34.01652509652509,
      "grad_norm": 0.08917862176895142,
      "learning_rate": 3.371943371943372e-06,
      "loss": 0.0004,
      "step": 27060
    },
    {
      "epoch": 34.0167824967825,
      "grad_norm": 0.0002305887028342113,
      "learning_rate": 3.369083369083369e-06,
      "loss": 0.6585,
      "step": 27070
    },
    {
      "epoch": 34.01703989703989,
      "grad_norm": 0.04280995950102806,
      "learning_rate": 3.3662233662233667e-06,
      "loss": 0.2125,
      "step": 27080
    },
    {
      "epoch": 34.0172972972973,
      "grad_norm": 0.005567069165408611,
      "learning_rate": 3.363363363363364e-06,
      "loss": 0.5481,
      "step": 27090
    },
    {
      "epoch": 34.0175546975547,
      "grad_norm": 0.017813654616475105,
      "learning_rate": 3.360503360503361e-06,
      "loss": 0.0007,
      "step": 27100
    },
    {
      "epoch": 34.0178120978121,
      "grad_norm": 92.60106658935547,
      "learning_rate": 3.3576433576433583e-06,
      "loss": 0.7287,
      "step": 27110
    },
    {
      "epoch": 34.0180694980695,
      "grad_norm": 0.10312855243682861,
      "learning_rate": 3.354783354783355e-06,
      "loss": 0.001,
      "step": 27120
    },
    {
      "epoch": 34.0183268983269,
      "grad_norm": 0.05467154458165169,
      "learning_rate": 3.3519233519233525e-06,
      "loss": 0.001,
      "step": 27130
    },
    {
      "epoch": 34.0185842985843,
      "grad_norm": 0.10436591506004333,
      "learning_rate": 3.3490633490633494e-06,
      "loss": 1.2953,
      "step": 27140
    },
    {
      "epoch": 34.0188416988417,
      "grad_norm": 0.06830277293920517,
      "learning_rate": 3.3462033462033467e-06,
      "loss": 0.0005,
      "step": 27150
    },
    {
      "epoch": 34.0190990990991,
      "grad_norm": 0.029437022283673286,
      "learning_rate": 3.3433433433433436e-06,
      "loss": 0.0002,
      "step": 27160
    },
    {
      "epoch": 34.0193564993565,
      "grad_norm": 2062.808837890625,
      "learning_rate": 3.340483340483341e-06,
      "loss": 0.4116,
      "step": 27170
    },
    {
      "epoch": 34.0196138996139,
      "grad_norm": 0.00020808493718504906,
      "learning_rate": 3.337623337623338e-06,
      "loss": 0.081,
      "step": 27180
    },
    {
      "epoch": 34.0198712998713,
      "grad_norm": 0.00027362650143913925,
      "learning_rate": 3.334763334763335e-06,
      "loss": 0.8758,
      "step": 27190
    },
    {
      "epoch": 34.02,
      "eval_accuracy": 0.8695652173913043,
      "eval_loss": 0.7715871930122375,
      "eval_runtime": 13.3796,
      "eval_samples_per_second": 3.438,
      "eval_steps_per_second": 3.438,
      "step": 27195
    },
    {
      "epoch": 35.0001287001287,
      "grad_norm": 0.00017348758410662413,
      "learning_rate": 3.3319033319033324e-06,
      "loss": 1.0786,
      "step": 27200
    },
    {
      "epoch": 35.0003861003861,
      "grad_norm": 0.0002206278295489028,
      "learning_rate": 3.3290433290433293e-06,
      "loss": 0.6193,
      "step": 27210
    },
    {
      "epoch": 35.0006435006435,
      "grad_norm": 0.0069132959470152855,
      "learning_rate": 3.3261833261833266e-06,
      "loss": 0.5806,
      "step": 27220
    },
    {
      "epoch": 35.0009009009009,
      "grad_norm": 0.3105623424053192,
      "learning_rate": 3.3233233233233235e-06,
      "loss": 0.4066,
      "step": 27230
    },
    {
      "epoch": 35.0011583011583,
      "grad_norm": 0.00029745243955403566,
      "learning_rate": 3.320463320463321e-06,
      "loss": 0.0074,
      "step": 27240
    },
    {
      "epoch": 35.0014157014157,
      "grad_norm": 0.005316603928804398,
      "learning_rate": 3.3176033176033177e-06,
      "loss": 0.1438,
      "step": 27250
    },
    {
      "epoch": 35.0016731016731,
      "grad_norm": 0.00018269273277837783,
      "learning_rate": 3.314743314743315e-06,
      "loss": 0.0002,
      "step": 27260
    },
    {
      "epoch": 35.0019305019305,
      "grad_norm": 0.0015236226608976722,
      "learning_rate": 3.311883311883312e-06,
      "loss": 0.5008,
      "step": 27270
    },
    {
      "epoch": 35.0021879021879,
      "grad_norm": 0.021271970123052597,
      "learning_rate": 3.3090233090233092e-06,
      "loss": 0.0003,
      "step": 27280
    },
    {
      "epoch": 35.0024453024453,
      "grad_norm": 0.0007113626343198121,
      "learning_rate": 3.3061633061633066e-06,
      "loss": 1.2656,
      "step": 27290
    },
    {
      "epoch": 35.002702702702706,
      "grad_norm": 0.3437110185623169,
      "learning_rate": 3.3033033033033035e-06,
      "loss": 0.0008,
      "step": 27300
    },
    {
      "epoch": 35.0029601029601,
      "grad_norm": 0.042603738605976105,
      "learning_rate": 3.3004433004433008e-06,
      "loss": 0.7157,
      "step": 27310
    },
    {
      "epoch": 35.003217503217506,
      "grad_norm": 0.0068983961828053,
      "learning_rate": 3.2975832975832977e-06,
      "loss": 0.5931,
      "step": 27320
    },
    {
      "epoch": 35.0034749034749,
      "grad_norm": 0.023130234330892563,
      "learning_rate": 3.294723294723295e-06,
      "loss": 0.0004,
      "step": 27330
    },
    {
      "epoch": 35.003732303732306,
      "grad_norm": 0.0014327826211228967,
      "learning_rate": 3.291863291863292e-06,
      "loss": 0.4843,
      "step": 27340
    },
    {
      "epoch": 35.0039897039897,
      "grad_norm": 0.0004222426505293697,
      "learning_rate": 3.289003289003289e-06,
      "loss": 0.5337,
      "step": 27350
    },
    {
      "epoch": 35.004247104247106,
      "grad_norm": 0.0003112400881946087,
      "learning_rate": 3.286143286143286e-06,
      "loss": 0.6144,
      "step": 27360
    },
    {
      "epoch": 35.0045045045045,
      "grad_norm": 0.00023315486032515764,
      "learning_rate": 3.2832832832832834e-06,
      "loss": 0.3117,
      "step": 27370
    },
    {
      "epoch": 35.00476190476191,
      "grad_norm": 0.01298606488853693,
      "learning_rate": 3.2804232804232807e-06,
      "loss": 0.2838,
      "step": 27380
    },
    {
      "epoch": 35.0050193050193,
      "grad_norm": 161.7285919189453,
      "learning_rate": 3.2775632775632776e-06,
      "loss": 0.5173,
      "step": 27390
    },
    {
      "epoch": 35.00527670527671,
      "grad_norm": 0.0003556629817467183,
      "learning_rate": 3.274703274703275e-06,
      "loss": 0.6745,
      "step": 27400
    },
    {
      "epoch": 35.0055341055341,
      "grad_norm": 0.003990973811596632,
      "learning_rate": 3.271843271843272e-06,
      "loss": 0.0003,
      "step": 27410
    },
    {
      "epoch": 35.00579150579151,
      "grad_norm": 0.017456263303756714,
      "learning_rate": 3.268983268983269e-06,
      "loss": 0.0003,
      "step": 27420
    },
    {
      "epoch": 35.0060489060489,
      "grad_norm": 377.7240905761719,
      "learning_rate": 3.266123266123266e-06,
      "loss": 0.0187,
      "step": 27430
    },
    {
      "epoch": 35.00630630630631,
      "grad_norm": 0.051734283566474915,
      "learning_rate": 3.2632632632632633e-06,
      "loss": 0.0003,
      "step": 27440
    },
    {
      "epoch": 35.0065637065637,
      "grad_norm": 0.053064629435539246,
      "learning_rate": 3.260403260403261e-06,
      "loss": 0.0024,
      "step": 27450
    },
    {
      "epoch": 35.00682110682111,
      "grad_norm": 0.16284862160682678,
      "learning_rate": 3.257543257543258e-06,
      "loss": 0.9852,
      "step": 27460
    },
    {
      "epoch": 35.0070785070785,
      "grad_norm": 0.0042984443716704845,
      "learning_rate": 3.2546832546832553e-06,
      "loss": 0.3026,
      "step": 27470
    },
    {
      "epoch": 35.00733590733591,
      "grad_norm": 0.00021199326147325337,
      "learning_rate": 3.251823251823252e-06,
      "loss": 0.3458,
      "step": 27480
    },
    {
      "epoch": 35.00759330759331,
      "grad_norm": 1034.568115234375,
      "learning_rate": 3.2489632489632495e-06,
      "loss": 0.381,
      "step": 27490
    },
    {
      "epoch": 35.00785070785071,
      "grad_norm": 0.12477569282054901,
      "learning_rate": 3.2461032461032464e-06,
      "loss": 0.0004,
      "step": 27500
    },
    {
      "epoch": 35.00810810810811,
      "grad_norm": 0.042156290262937546,
      "learning_rate": 3.2432432432432437e-06,
      "loss": 0.8273,
      "step": 27510
    },
    {
      "epoch": 35.00836550836551,
      "grad_norm": 0.003120782785117626,
      "learning_rate": 3.240383240383241e-06,
      "loss": 0.0007,
      "step": 27520
    },
    {
      "epoch": 35.00862290862291,
      "grad_norm": 131.19398498535156,
      "learning_rate": 3.237523237523238e-06,
      "loss": 0.5689,
      "step": 27530
    },
    {
      "epoch": 35.00888030888031,
      "grad_norm": 0.00033146803616546094,
      "learning_rate": 3.2346632346632352e-06,
      "loss": 0.0002,
      "step": 27540
    },
    {
      "epoch": 35.00913770913771,
      "grad_norm": 152.272216796875,
      "learning_rate": 3.231803231803232e-06,
      "loss": 0.6175,
      "step": 27550
    },
    {
      "epoch": 35.00939510939511,
      "grad_norm": 0.016422213986516,
      "learning_rate": 3.2289432289432294e-06,
      "loss": 0.6991,
      "step": 27560
    },
    {
      "epoch": 35.00965250965251,
      "grad_norm": 0.02891531027853489,
      "learning_rate": 3.2260832260832263e-06,
      "loss": 0.0145,
      "step": 27570
    },
    {
      "epoch": 35.00990990990991,
      "grad_norm": 0.034770023077726364,
      "learning_rate": 3.2232232232232236e-06,
      "loss": 1.3164,
      "step": 27580
    },
    {
      "epoch": 35.01016731016731,
      "grad_norm": 0.2806760370731354,
      "learning_rate": 3.2203632203632205e-06,
      "loss": 0.1784,
      "step": 27590
    },
    {
      "epoch": 35.01042471042471,
      "grad_norm": 58.18006896972656,
      "learning_rate": 3.217503217503218e-06,
      "loss": 1.1186,
      "step": 27600
    },
    {
      "epoch": 35.01068211068211,
      "grad_norm": 0.025718385353684425,
      "learning_rate": 3.2146432146432147e-06,
      "loss": 0.0003,
      "step": 27610
    },
    {
      "epoch": 35.01093951093951,
      "grad_norm": 0.0030768548604100943,
      "learning_rate": 3.211783211783212e-06,
      "loss": 0.0002,
      "step": 27620
    },
    {
      "epoch": 35.01119691119691,
      "grad_norm": 0.010804054327309132,
      "learning_rate": 3.2089232089232094e-06,
      "loss": 0.7885,
      "step": 27630
    },
    {
      "epoch": 35.01145431145431,
      "grad_norm": 250.21238708496094,
      "learning_rate": 3.2060632060632063e-06,
      "loss": 0.4832,
      "step": 27640
    },
    {
      "epoch": 35.01171171171171,
      "grad_norm": 57.603599548339844,
      "learning_rate": 3.2032032032032036e-06,
      "loss": 1.1185,
      "step": 27650
    },
    {
      "epoch": 35.011969111969115,
      "grad_norm": 0.0002042093692580238,
      "learning_rate": 3.2003432003432005e-06,
      "loss": 0.6971,
      "step": 27660
    },
    {
      "epoch": 35.01222651222651,
      "grad_norm": 584.781982421875,
      "learning_rate": 3.197483197483198e-06,
      "loss": 0.0292,
      "step": 27670
    },
    {
      "epoch": 35.012483912483916,
      "grad_norm": 0.11098294705152512,
      "learning_rate": 3.1946231946231947e-06,
      "loss": 0.0022,
      "step": 27680
    },
    {
      "epoch": 35.01274131274131,
      "grad_norm": 0.0507424995303154,
      "learning_rate": 3.191763191763192e-06,
      "loss": 0.0008,
      "step": 27690
    },
    {
      "epoch": 35.012998712998716,
      "grad_norm": 0.00076050974894315,
      "learning_rate": 3.188903188903189e-06,
      "loss": 0.5183,
      "step": 27700
    },
    {
      "epoch": 35.01325611325611,
      "grad_norm": 0.00023733789566904306,
      "learning_rate": 3.1860431860431862e-06,
      "loss": 0.0005,
      "step": 27710
    },
    {
      "epoch": 35.013513513513516,
      "grad_norm": 0.0004732540692202747,
      "learning_rate": 3.183183183183183e-06,
      "loss": 0.0005,
      "step": 27720
    },
    {
      "epoch": 35.01377091377091,
      "grad_norm": 0.0001804508501663804,
      "learning_rate": 3.1803231803231804e-06,
      "loss": 0.0007,
      "step": 27730
    },
    {
      "epoch": 35.014028314028316,
      "grad_norm": 0.1139734536409378,
      "learning_rate": 3.1774631774631777e-06,
      "loss": 0.0011,
      "step": 27740
    },
    {
      "epoch": 35.01428571428571,
      "grad_norm": 0.15720994770526886,
      "learning_rate": 3.1746031746031746e-06,
      "loss": 0.0006,
      "step": 27750
    },
    {
      "epoch": 35.014543114543116,
      "grad_norm": 0.005728594027459621,
      "learning_rate": 3.171743171743172e-06,
      "loss": 0.2534,
      "step": 27760
    },
    {
      "epoch": 35.01480051480051,
      "grad_norm": 0.00016303425945807248,
      "learning_rate": 3.168883168883169e-06,
      "loss": 0.4451,
      "step": 27770
    },
    {
      "epoch": 35.015057915057916,
      "grad_norm": 0.07668960839509964,
      "learning_rate": 3.166023166023166e-06,
      "loss": 0.0004,
      "step": 27780
    },
    {
      "epoch": 35.01531531531531,
      "grad_norm": 0.0008509852923452854,
      "learning_rate": 3.163163163163163e-06,
      "loss": 1.3581,
      "step": 27790
    },
    {
      "epoch": 35.015572715572716,
      "grad_norm": 0.3970625400543213,
      "learning_rate": 3.1603031603031604e-06,
      "loss": 0.0006,
      "step": 27800
    },
    {
      "epoch": 35.01583011583011,
      "grad_norm": 0.49919551610946655,
      "learning_rate": 3.1574431574431573e-06,
      "loss": 0.0007,
      "step": 27810
    },
    {
      "epoch": 35.01608751608752,
      "grad_norm": 0.4793255627155304,
      "learning_rate": 3.154583154583155e-06,
      "loss": 0.0008,
      "step": 27820
    },
    {
      "epoch": 35.01634491634491,
      "grad_norm": 0.12254327535629272,
      "learning_rate": 3.1517231517231523e-06,
      "loss": 0.4944,
      "step": 27830
    },
    {
      "epoch": 35.01660231660232,
      "grad_norm": 0.051711589097976685,
      "learning_rate": 3.148863148863149e-06,
      "loss": 0.0002,
      "step": 27840
    },
    {
      "epoch": 35.01685971685972,
      "grad_norm": 0.00018205131345894188,
      "learning_rate": 3.1460031460031465e-06,
      "loss": 0.0002,
      "step": 27850
    },
    {
      "epoch": 35.01711711711712,
      "grad_norm": 0.00019165266712661833,
      "learning_rate": 3.1431431431431434e-06,
      "loss": 0.8208,
      "step": 27860
    },
    {
      "epoch": 35.01737451737452,
      "grad_norm": 0.0013323579914867878,
      "learning_rate": 3.1402831402831407e-06,
      "loss": 0.0122,
      "step": 27870
    },
    {
      "epoch": 35.01763191763192,
      "grad_norm": 0.00027685629902407527,
      "learning_rate": 3.137423137423138e-06,
      "loss": 0.0122,
      "step": 27880
    },
    {
      "epoch": 35.01788931788932,
      "grad_norm": 0.01839580573141575,
      "learning_rate": 3.134563134563135e-06,
      "loss": 0.6481,
      "step": 27890
    },
    {
      "epoch": 35.01814671814672,
      "grad_norm": 0.0037593143060803413,
      "learning_rate": 3.1317031317031323e-06,
      "loss": 0.0001,
      "step": 27900
    },
    {
      "epoch": 35.01840411840412,
      "grad_norm": 1.110424518585205,
      "learning_rate": 3.128843128843129e-06,
      "loss": 0.5803,
      "step": 27910
    },
    {
      "epoch": 35.01866151866152,
      "grad_norm": 0.0064668068662285805,
      "learning_rate": 3.1259831259831265e-06,
      "loss": 0.0005,
      "step": 27920
    },
    {
      "epoch": 35.01891891891892,
      "grad_norm": 0.00647960789501667,
      "learning_rate": 3.1231231231231234e-06,
      "loss": 0.0038,
      "step": 27930
    },
    {
      "epoch": 35.01917631917632,
      "grad_norm": 0.03522304818034172,
      "learning_rate": 3.1202631202631207e-06,
      "loss": 0.0729,
      "step": 27940
    },
    {
      "epoch": 35.01943371943372,
      "grad_norm": 1.5073069334030151,
      "learning_rate": 3.1174031174031176e-06,
      "loss": 0.6523,
      "step": 27950
    },
    {
      "epoch": 35.01969111969112,
      "grad_norm": 114.54082489013672,
      "learning_rate": 3.114543114543115e-06,
      "loss": 1.3528,
      "step": 27960
    },
    {
      "epoch": 35.01994851994852,
      "grad_norm": 0.0004169368767179549,
      "learning_rate": 3.1116831116831118e-06,
      "loss": 0.5316,
      "step": 27970
    },
    {
      "epoch": 35.02,
      "eval_accuracy": 0.8478260869565217,
      "eval_loss": 0.917574405670166,
      "eval_runtime": 14.3651,
      "eval_samples_per_second": 3.202,
      "eval_steps_per_second": 3.202,
      "step": 27972
    },
    {
      "epoch": 36.00020592020592,
      "grad_norm": 0.00026314210845157504,
      "learning_rate": 3.108823108823109e-06,
      "loss": 0.7594,
      "step": 27980
    },
    {
      "epoch": 36.00046332046332,
      "grad_norm": 0.06673434376716614,
      "learning_rate": 3.1059631059631064e-06,
      "loss": 0.0005,
      "step": 27990
    },
    {
      "epoch": 36.00072072072072,
      "grad_norm": 0.0002622131141833961,
      "learning_rate": 3.1031031031031033e-06,
      "loss": 0.5207,
      "step": 28000
    },
    {
      "epoch": 36.00097812097812,
      "grad_norm": 0.007425097282975912,
      "learning_rate": 3.1002431002431006e-06,
      "loss": 0.1283,
      "step": 28010
    },
    {
      "epoch": 36.00123552123552,
      "grad_norm": 678.264404296875,
      "learning_rate": 3.0973830973830975e-06,
      "loss": 0.2884,
      "step": 28020
    },
    {
      "epoch": 36.00149292149292,
      "grad_norm": 0.00021443287550937384,
      "learning_rate": 3.094523094523095e-06,
      "loss": 0.0003,
      "step": 28030
    },
    {
      "epoch": 36.00175032175032,
      "grad_norm": 0.02657965198159218,
      "learning_rate": 3.0916630916630917e-06,
      "loss": 0.0007,
      "step": 28040
    },
    {
      "epoch": 36.002007722007725,
      "grad_norm": 0.08193910866975784,
      "learning_rate": 3.088803088803089e-06,
      "loss": 0.0443,
      "step": 28050
    },
    {
      "epoch": 36.00226512226512,
      "grad_norm": 0.13849781453609467,
      "learning_rate": 3.085943085943086e-06,
      "loss": 0.0004,
      "step": 28060
    },
    {
      "epoch": 36.002522522522526,
      "grad_norm": 0.002490939339622855,
      "learning_rate": 3.0830830830830832e-06,
      "loss": 0.001,
      "step": 28070
    },
    {
      "epoch": 36.00277992277992,
      "grad_norm": 0.020132167264819145,
      "learning_rate": 3.0802230802230806e-06,
      "loss": 0.0011,
      "step": 28080
    },
    {
      "epoch": 36.003037323037326,
      "grad_norm": 0.0001936969201778993,
      "learning_rate": 3.0773630773630775e-06,
      "loss": 0.6241,
      "step": 28090
    },
    {
      "epoch": 36.00329472329472,
      "grad_norm": 0.00044216442620381713,
      "learning_rate": 3.0745030745030748e-06,
      "loss": 0.4918,
      "step": 28100
    },
    {
      "epoch": 36.003552123552126,
      "grad_norm": 0.0017083561979234219,
      "learning_rate": 3.0716430716430717e-06,
      "loss": 0.1778,
      "step": 28110
    },
    {
      "epoch": 36.00380952380952,
      "grad_norm": 0.01577197201550007,
      "learning_rate": 3.068783068783069e-06,
      "loss": 0.0003,
      "step": 28120
    },
    {
      "epoch": 36.004066924066926,
      "grad_norm": 0.03330492228269577,
      "learning_rate": 3.065923065923066e-06,
      "loss": 0.0003,
      "step": 28130
    },
    {
      "epoch": 36.00432432432432,
      "grad_norm": 0.009137704968452454,
      "learning_rate": 3.063063063063063e-06,
      "loss": 0.5943,
      "step": 28140
    },
    {
      "epoch": 36.004581724581726,
      "grad_norm": 1.6320407390594482,
      "learning_rate": 3.06020306020306e-06,
      "loss": 1.329,
      "step": 28150
    },
    {
      "epoch": 36.00483912483912,
      "grad_norm": 0.001672191545367241,
      "learning_rate": 3.0573430573430574e-06,
      "loss": 0.8598,
      "step": 28160
    },
    {
      "epoch": 36.005096525096526,
      "grad_norm": 0.0024032602086663246,
      "learning_rate": 3.0544830544830543e-06,
      "loss": 0.001,
      "step": 28170
    },
    {
      "epoch": 36.00535392535392,
      "grad_norm": 720.0140380859375,
      "learning_rate": 3.0516230516230516e-06,
      "loss": 0.0433,
      "step": 28180
    },
    {
      "epoch": 36.005611325611326,
      "grad_norm": 0.01528639905154705,
      "learning_rate": 3.048763048763049e-06,
      "loss": 0.1401,
      "step": 28190
    },
    {
      "epoch": 36.00586872586872,
      "grad_norm": 0.12506289780139923,
      "learning_rate": 3.0459030459030462e-06,
      "loss": 0.0004,
      "step": 28200
    },
    {
      "epoch": 36.00612612612613,
      "grad_norm": 0.012107270769774914,
      "learning_rate": 3.0430430430430436e-06,
      "loss": 0.5241,
      "step": 28210
    },
    {
      "epoch": 36.00638352638352,
      "grad_norm": 0.003177333390340209,
      "learning_rate": 3.040183040183041e-06,
      "loss": 0.7474,
      "step": 28220
    },
    {
      "epoch": 36.00664092664093,
      "grad_norm": 0.00022184972476679832,
      "learning_rate": 3.0373230373230378e-06,
      "loss": 0.143,
      "step": 28230
    },
    {
      "epoch": 36.00689832689833,
      "grad_norm": 0.022349726408720016,
      "learning_rate": 3.034463034463035e-06,
      "loss": 0.6655,
      "step": 28240
    },
    {
      "epoch": 36.00715572715573,
      "grad_norm": 0.0002947472967207432,
      "learning_rate": 3.031603031603032e-06,
      "loss": 0.0025,
      "step": 28250
    },
    {
      "epoch": 36.00741312741313,
      "grad_norm": 0.0007760194130241871,
      "learning_rate": 3.0287430287430293e-06,
      "loss": 0.0004,
      "step": 28260
    },
    {
      "epoch": 36.00767052767053,
      "grad_norm": 116.30671691894531,
      "learning_rate": 3.025883025883026e-06,
      "loss": 0.7956,
      "step": 28270
    },
    {
      "epoch": 36.00792792792793,
      "grad_norm": 0.00033408516901545227,
      "learning_rate": 3.0230230230230235e-06,
      "loss": 0.0,
      "step": 28280
    },
    {
      "epoch": 36.00818532818533,
      "grad_norm": 0.0354166217148304,
      "learning_rate": 3.0201630201630204e-06,
      "loss": 0.0003,
      "step": 28290
    },
    {
      "epoch": 36.00844272844273,
      "grad_norm": 0.0003061169118154794,
      "learning_rate": 3.0173030173030177e-06,
      "loss": 0.0001,
      "step": 28300
    },
    {
      "epoch": 36.00870012870013,
      "grad_norm": 0.003866319777444005,
      "learning_rate": 3.0144430144430146e-06,
      "loss": 1.8578,
      "step": 28310
    },
    {
      "epoch": 36.00895752895753,
      "grad_norm": 0.0002617123245727271,
      "learning_rate": 3.011583011583012e-06,
      "loss": 0.6685,
      "step": 28320
    },
    {
      "epoch": 36.00921492921493,
      "grad_norm": 0.00019824574701488018,
      "learning_rate": 3.0087230087230092e-06,
      "loss": 0.0003,
      "step": 28330
    },
    {
      "epoch": 36.00947232947233,
      "grad_norm": 0.00023168852203525603,
      "learning_rate": 3.005863005863006e-06,
      "loss": 1.0579,
      "step": 28340
    },
    {
      "epoch": 36.00972972972973,
      "grad_norm": 0.11857693642377853,
      "learning_rate": 3.0030030030030034e-06,
      "loss": 0.0007,
      "step": 28350
    },
    {
      "epoch": 36.00998712998713,
      "grad_norm": 0.039131905883550644,
      "learning_rate": 3.0001430001430003e-06,
      "loss": 0.0009,
      "step": 28360
    },
    {
      "epoch": 36.01024453024453,
      "grad_norm": 0.009820118546485901,
      "learning_rate": 2.9972829972829977e-06,
      "loss": 0.8717,
      "step": 28370
    },
    {
      "epoch": 36.01050193050193,
      "grad_norm": 0.00021176527661737055,
      "learning_rate": 2.9944229944229945e-06,
      "loss": 0.0004,
      "step": 28380
    },
    {
      "epoch": 36.01075933075933,
      "grad_norm": 0.18563033640384674,
      "learning_rate": 2.991562991562992e-06,
      "loss": 1.3191,
      "step": 28390
    },
    {
      "epoch": 36.01101673101673,
      "grad_norm": 0.00015481845184694976,
      "learning_rate": 2.9887029887029888e-06,
      "loss": 1.5741,
      "step": 28400
    },
    {
      "epoch": 36.01127413127413,
      "grad_norm": 0.0009244285756722093,
      "learning_rate": 2.985842985842986e-06,
      "loss": 0.0011,
      "step": 28410
    },
    {
      "epoch": 36.01153153153153,
      "grad_norm": 0.08380212634801865,
      "learning_rate": 2.982982982982983e-06,
      "loss": 0.0014,
      "step": 28420
    },
    {
      "epoch": 36.011788931788935,
      "grad_norm": 0.01747632957994938,
      "learning_rate": 2.9801229801229803e-06,
      "loss": 0.036,
      "step": 28430
    },
    {
      "epoch": 36.01204633204633,
      "grad_norm": 143.46397399902344,
      "learning_rate": 2.9772629772629776e-06,
      "loss": 0.7545,
      "step": 28440
    },
    {
      "epoch": 36.012303732303735,
      "grad_norm": 0.00015246524708345532,
      "learning_rate": 2.9744029744029745e-06,
      "loss": 0.0012,
      "step": 28450
    },
    {
      "epoch": 36.01256113256113,
      "grad_norm": 0.2742994725704193,
      "learning_rate": 2.971542971542972e-06,
      "loss": 0.0007,
      "step": 28460
    },
    {
      "epoch": 36.012818532818535,
      "grad_norm": 92.01673126220703,
      "learning_rate": 2.9686829686829687e-06,
      "loss": 1.2388,
      "step": 28470
    },
    {
      "epoch": 36.01307593307593,
      "grad_norm": 0.010131720453500748,
      "learning_rate": 2.965822965822966e-06,
      "loss": 0.0005,
      "step": 28480
    },
    {
      "epoch": 36.013333333333335,
      "grad_norm": 156.0325164794922,
      "learning_rate": 2.962962962962963e-06,
      "loss": 1.373,
      "step": 28490
    },
    {
      "epoch": 36.01359073359073,
      "grad_norm": 0.016434205695986748,
      "learning_rate": 2.9601029601029602e-06,
      "loss": 0.0017,
      "step": 28500
    },
    {
      "epoch": 36.013848133848136,
      "grad_norm": 0.0648985505104065,
      "learning_rate": 2.957242957242957e-06,
      "loss": 0.6322,
      "step": 28510
    },
    {
      "epoch": 36.01410553410553,
      "grad_norm": 0.18576735258102417,
      "learning_rate": 2.9543829543829544e-06,
      "loss": 0.0007,
      "step": 28520
    },
    {
      "epoch": 36.014362934362936,
      "grad_norm": 0.014449382200837135,
      "learning_rate": 2.9515229515229517e-06,
      "loss": 1.0034,
      "step": 28530
    },
    {
      "epoch": 36.01462033462033,
      "grad_norm": 0.0007821593899279833,
      "learning_rate": 2.9486629486629486e-06,
      "loss": 0.5917,
      "step": 28540
    },
    {
      "epoch": 36.014877734877736,
      "grad_norm": 0.00015635178715456277,
      "learning_rate": 2.945802945802946e-06,
      "loss": 1.1249,
      "step": 28550
    },
    {
      "epoch": 36.01513513513513,
      "grad_norm": 0.10193686932325363,
      "learning_rate": 2.942942942942943e-06,
      "loss": 0.9982,
      "step": 28560
    },
    {
      "epoch": 36.015392535392536,
      "grad_norm": 0.07530353218317032,
      "learning_rate": 2.9400829400829406e-06,
      "loss": 0.0005,
      "step": 28570
    },
    {
      "epoch": 36.01564993564993,
      "grad_norm": 0.026423171162605286,
      "learning_rate": 2.937222937222938e-06,
      "loss": 0.0012,
      "step": 28580
    },
    {
      "epoch": 36.015907335907336,
      "grad_norm": 0.0017025470733642578,
      "learning_rate": 2.934362934362935e-06,
      "loss": 0.0004,
      "step": 28590
    },
    {
      "epoch": 36.01616473616474,
      "grad_norm": 0.00017218876746483147,
      "learning_rate": 2.931502931502932e-06,
      "loss": 0.0006,
      "step": 28600
    },
    {
      "epoch": 36.016422136422136,
      "grad_norm": 0.00020879317889921367,
      "learning_rate": 2.928642928642929e-06,
      "loss": 0.0002,
      "step": 28610
    },
    {
      "epoch": 36.01667953667954,
      "grad_norm": 0.00023135027731768787,
      "learning_rate": 2.9257829257829263e-06,
      "loss": 0.5356,
      "step": 28620
    },
    {
      "epoch": 36.016936936936936,
      "grad_norm": 0.0003467734786681831,
      "learning_rate": 2.9229229229229232e-06,
      "loss": 0.0003,
      "step": 28630
    },
    {
      "epoch": 36.01719433719434,
      "grad_norm": 0.0012130647664889693,
      "learning_rate": 2.9200629200629205e-06,
      "loss": 0.6772,
      "step": 28640
    },
    {
      "epoch": 36.01745173745174,
      "grad_norm": 0.00030612791306339204,
      "learning_rate": 2.9172029172029174e-06,
      "loss": 0.0006,
      "step": 28650
    },
    {
      "epoch": 36.01770913770914,
      "grad_norm": 0.023205522447824478,
      "learning_rate": 2.9143429143429147e-06,
      "loss": 0.0006,
      "step": 28660
    },
    {
      "epoch": 36.01796653796654,
      "grad_norm": 0.017860963940620422,
      "learning_rate": 2.9114829114829116e-06,
      "loss": 0.303,
      "step": 28670
    },
    {
      "epoch": 36.01822393822394,
      "grad_norm": 0.037889543920755386,
      "learning_rate": 2.908622908622909e-06,
      "loss": 0.9838,
      "step": 28680
    },
    {
      "epoch": 36.01848133848134,
      "grad_norm": 0.061780329793691635,
      "learning_rate": 2.9057629057629063e-06,
      "loss": 0.0001,
      "step": 28690
    },
    {
      "epoch": 36.01873873873874,
      "grad_norm": 0.0001773426338331774,
      "learning_rate": 2.902902902902903e-06,
      "loss": 0.0002,
      "step": 28700
    },
    {
      "epoch": 36.01899613899614,
      "grad_norm": 0.09331512451171875,
      "learning_rate": 2.9000429000429005e-06,
      "loss": 0.0009,
      "step": 28710
    },
    {
      "epoch": 36.01925353925354,
      "grad_norm": 0.00026550382608547807,
      "learning_rate": 2.8971828971828974e-06,
      "loss": 0.0003,
      "step": 28720
    },
    {
      "epoch": 36.01951093951094,
      "grad_norm": 0.01748647913336754,
      "learning_rate": 2.8943228943228947e-06,
      "loss": 0.0004,
      "step": 28730
    },
    {
      "epoch": 36.01976833976834,
      "grad_norm": 0.0793762281537056,
      "learning_rate": 2.8914628914628916e-06,
      "loss": 0.0003,
      "step": 28740
    },
    {
      "epoch": 36.02,
      "eval_accuracy": 0.8478260869565217,
      "eval_loss": 1.074432373046875,
      "eval_runtime": 13.3897,
      "eval_samples_per_second": 3.435,
      "eval_steps_per_second": 3.435,
      "step": 28749
    },
    {
      "epoch": 37.00002574002574,
      "grad_norm": 0.017261644825339317,
      "learning_rate": 2.888602888602889e-06,
      "loss": 0.3016,
      "step": 28750
    },
    {
      "epoch": 37.00028314028314,
      "grad_norm": 0.00018201515194959939,
      "learning_rate": 2.8857428857428858e-06,
      "loss": 0.0005,
      "step": 28760
    },
    {
      "epoch": 37.00054054054054,
      "grad_norm": 0.09265059977769852,
      "learning_rate": 2.882882882882883e-06,
      "loss": 0.0004,
      "step": 28770
    },
    {
      "epoch": 37.00079794079794,
      "grad_norm": 0.0002512499631848186,
      "learning_rate": 2.8800228800228804e-06,
      "loss": 0.6045,
      "step": 28780
    },
    {
      "epoch": 37.00105534105534,
      "grad_norm": 0.021960824728012085,
      "learning_rate": 2.8771628771628773e-06,
      "loss": 0.4089,
      "step": 28790
    },
    {
      "epoch": 37.00131274131274,
      "grad_norm": 0.00020900735398754478,
      "learning_rate": 2.8743028743028746e-06,
      "loss": 0.5131,
      "step": 28800
    },
    {
      "epoch": 37.00157014157014,
      "grad_norm": 0.00446329265832901,
      "learning_rate": 2.8714428714428715e-06,
      "loss": 0.0002,
      "step": 28810
    },
    {
      "epoch": 37.001827541827545,
      "grad_norm": 0.009537389501929283,
      "learning_rate": 2.868582868582869e-06,
      "loss": 0.0019,
      "step": 28820
    },
    {
      "epoch": 37.00208494208494,
      "grad_norm": 0.05497797206044197,
      "learning_rate": 2.8657228657228657e-06,
      "loss": 0.005,
      "step": 28830
    },
    {
      "epoch": 37.002342342342345,
      "grad_norm": 0.00030364299891516566,
      "learning_rate": 2.862862862862863e-06,
      "loss": 0.6848,
      "step": 28840
    },
    {
      "epoch": 37.00259974259974,
      "grad_norm": 0.004607612267136574,
      "learning_rate": 2.86000286000286e-06,
      "loss": 0.8919,
      "step": 28850
    },
    {
      "epoch": 37.002857142857145,
      "grad_norm": 0.002171023515984416,
      "learning_rate": 2.8571428571428573e-06,
      "loss": 0.5662,
      "step": 28860
    },
    {
      "epoch": 37.00311454311454,
      "grad_norm": 0.00719066895544529,
      "learning_rate": 2.854282854282854e-06,
      "loss": 0.0003,
      "step": 28870
    },
    {
      "epoch": 37.003371943371945,
      "grad_norm": 0.08742506057024002,
      "learning_rate": 2.8514228514228515e-06,
      "loss": 0.7729,
      "step": 28880
    },
    {
      "epoch": 37.00362934362934,
      "grad_norm": 0.004545649979263544,
      "learning_rate": 2.8485628485628488e-06,
      "loss": 0.0001,
      "step": 28890
    },
    {
      "epoch": 37.003886743886746,
      "grad_norm": 0.00015914670075289905,
      "learning_rate": 2.8457028457028457e-06,
      "loss": 0.7993,
      "step": 28900
    },
    {
      "epoch": 37.00414414414414,
      "grad_norm": 0.052335429936647415,
      "learning_rate": 2.842842842842843e-06,
      "loss": 1.0677,
      "step": 28910
    },
    {
      "epoch": 37.004401544401546,
      "grad_norm": 0.00018764731066767126,
      "learning_rate": 2.83998283998284e-06,
      "loss": 0.0002,
      "step": 28920
    },
    {
      "epoch": 37.00465894465894,
      "grad_norm": 0.00018724115216173232,
      "learning_rate": 2.837122837122837e-06,
      "loss": 0.2418,
      "step": 28930
    },
    {
      "epoch": 37.004916344916346,
      "grad_norm": 0.004428797867149115,
      "learning_rate": 2.834262834262834e-06,
      "loss": 0.0005,
      "step": 28940
    },
    {
      "epoch": 37.00517374517374,
      "grad_norm": 0.004312903620302677,
      "learning_rate": 2.831402831402832e-06,
      "loss": 0.7043,
      "step": 28950
    },
    {
      "epoch": 37.005431145431146,
      "grad_norm": 465.94049072265625,
      "learning_rate": 2.828542828542829e-06,
      "loss": 0.0731,
      "step": 28960
    },
    {
      "epoch": 37.00568854568854,
      "grad_norm": 0.021510232239961624,
      "learning_rate": 2.825682825682826e-06,
      "loss": 0.0002,
      "step": 28970
    },
    {
      "epoch": 37.005945945945946,
      "grad_norm": 0.00018029431521426886,
      "learning_rate": 2.8228228228228234e-06,
      "loss": 0.0026,
      "step": 28980
    },
    {
      "epoch": 37.00620334620335,
      "grad_norm": 0.0017462585819885135,
      "learning_rate": 2.8199628199628202e-06,
      "loss": 0.2045,
      "step": 28990
    },
    {
      "epoch": 37.006460746460746,
      "grad_norm": 0.00019625762070063502,
      "learning_rate": 2.8171028171028176e-06,
      "loss": 0.0001,
      "step": 29000
    },
    {
      "epoch": 37.00671814671815,
      "grad_norm": 0.024823561310768127,
      "learning_rate": 2.8142428142428145e-06,
      "loss": 0.0451,
      "step": 29010
    },
    {
      "epoch": 37.006975546975546,
      "grad_norm": 0.00015227263793349266,
      "learning_rate": 2.8113828113828118e-06,
      "loss": 0.106,
      "step": 29020
    },
    {
      "epoch": 37.00723294723295,
      "grad_norm": 0.00022190021991264075,
      "learning_rate": 2.808522808522809e-06,
      "loss": 0.5192,
      "step": 29030
    },
    {
      "epoch": 37.00749034749035,
      "grad_norm": 0.00021235458552837372,
      "learning_rate": 2.805662805662806e-06,
      "loss": 0.8077,
      "step": 29040
    },
    {
      "epoch": 37.00774774774775,
      "grad_norm": 0.0744086354970932,
      "learning_rate": 2.8028028028028033e-06,
      "loss": 0.7212,
      "step": 29050
    },
    {
      "epoch": 37.00800514800515,
      "grad_norm": 0.028785334900021553,
      "learning_rate": 2.7999427999428e-06,
      "loss": 0.022,
      "step": 29060
    },
    {
      "epoch": 37.00826254826255,
      "grad_norm": 0.0001730113581288606,
      "learning_rate": 2.7970827970827975e-06,
      "loss": 0.007,
      "step": 29070
    },
    {
      "epoch": 37.00851994851995,
      "grad_norm": 0.00028456852305680513,
      "learning_rate": 2.7942227942227944e-06,
      "loss": 1.4105,
      "step": 29080
    },
    {
      "epoch": 37.00877734877735,
      "grad_norm": 0.2067694216966629,
      "learning_rate": 2.7913627913627917e-06,
      "loss": 0.0902,
      "step": 29090
    },
    {
      "epoch": 37.00903474903475,
      "grad_norm": 4.730393409729004,
      "learning_rate": 2.7885027885027886e-06,
      "loss": 0.0014,
      "step": 29100
    },
    {
      "epoch": 37.00929214929215,
      "grad_norm": 280.6280212402344,
      "learning_rate": 2.785642785642786e-06,
      "loss": 0.7313,
      "step": 29110
    },
    {
      "epoch": 37.00954954954955,
      "grad_norm": 0.0012725815176963806,
      "learning_rate": 2.782782782782783e-06,
      "loss": 0.0001,
      "step": 29120
    },
    {
      "epoch": 37.00980694980695,
      "grad_norm": 0.10952453315258026,
      "learning_rate": 2.77992277992278e-06,
      "loss": 0.0004,
      "step": 29130
    },
    {
      "epoch": 37.01006435006435,
      "grad_norm": 0.674946665763855,
      "learning_rate": 2.7770627770627775e-06,
      "loss": 0.3738,
      "step": 29140
    },
    {
      "epoch": 37.01032175032175,
      "grad_norm": 0.006087814457714558,
      "learning_rate": 2.7742027742027743e-06,
      "loss": 0.0005,
      "step": 29150
    },
    {
      "epoch": 37.01057915057915,
      "grad_norm": 0.008630234748125076,
      "learning_rate": 2.7713427713427717e-06,
      "loss": 0.0002,
      "step": 29160
    },
    {
      "epoch": 37.01083655083655,
      "grad_norm": 0.0001840332115534693,
      "learning_rate": 2.7684827684827686e-06,
      "loss": 0.0002,
      "step": 29170
    },
    {
      "epoch": 37.011093951093955,
      "grad_norm": 0.0036154910922050476,
      "learning_rate": 2.765622765622766e-06,
      "loss": 0.0005,
      "step": 29180
    },
    {
      "epoch": 37.01135135135135,
      "grad_norm": 0.02959679253399372,
      "learning_rate": 2.7627627627627628e-06,
      "loss": 0.6275,
      "step": 29190
    },
    {
      "epoch": 37.011608751608755,
      "grad_norm": 0.01243556383997202,
      "learning_rate": 2.75990275990276e-06,
      "loss": 0.0002,
      "step": 29200
    },
    {
      "epoch": 37.01186615186615,
      "grad_norm": 0.07791236788034439,
      "learning_rate": 2.757042757042757e-06,
      "loss": 0.8283,
      "step": 29210
    },
    {
      "epoch": 37.012123552123555,
      "grad_norm": 0.012300877831876278,
      "learning_rate": 2.7541827541827543e-06,
      "loss": 0.0005,
      "step": 29220
    },
    {
      "epoch": 37.01238095238095,
      "grad_norm": 0.0001772117830114439,
      "learning_rate": 2.7513227513227516e-06,
      "loss": 0.2855,
      "step": 29230
    },
    {
      "epoch": 37.012638352638355,
      "grad_norm": 0.010969296097755432,
      "learning_rate": 2.7484627484627485e-06,
      "loss": 0.0034,
      "step": 29240
    },
    {
      "epoch": 37.01289575289575,
      "grad_norm": 0.00019608347793109715,
      "learning_rate": 2.745602745602746e-06,
      "loss": 0.7902,
      "step": 29250
    },
    {
      "epoch": 37.013153153153155,
      "grad_norm": 0.020125016570091248,
      "learning_rate": 2.7427427427427427e-06,
      "loss": 0.0003,
      "step": 29260
    },
    {
      "epoch": 37.01341055341055,
      "grad_norm": 0.0045450483448803425,
      "learning_rate": 2.73988273988274e-06,
      "loss": 0.0001,
      "step": 29270
    },
    {
      "epoch": 37.013667953667955,
      "grad_norm": 0.0002109996130457148,
      "learning_rate": 2.737022737022737e-06,
      "loss": 0.0001,
      "step": 29280
    },
    {
      "epoch": 37.01392535392535,
      "grad_norm": 0.00018298572103958577,
      "learning_rate": 2.7341627341627342e-06,
      "loss": 1.0783,
      "step": 29290
    },
    {
      "epoch": 37.014182754182755,
      "grad_norm": 0.08364742249250412,
      "learning_rate": 2.731302731302731e-06,
      "loss": 0.3974,
      "step": 29300
    },
    {
      "epoch": 37.01444015444015,
      "grad_norm": 0.032810915261507034,
      "learning_rate": 2.7284427284427284e-06,
      "loss": 0.2355,
      "step": 29310
    },
    {
      "epoch": 37.014697554697555,
      "grad_norm": 0.30951929092407227,
      "learning_rate": 2.725582725582726e-06,
      "loss": 0.001,
      "step": 29320
    },
    {
      "epoch": 37.01495495495495,
      "grad_norm": 0.0006139618926681578,
      "learning_rate": 2.722722722722723e-06,
      "loss": 0.0841,
      "step": 29330
    },
    {
      "epoch": 37.015212355212356,
      "grad_norm": 0.034589651972055435,
      "learning_rate": 2.7198627198627204e-06,
      "loss": 0.0009,
      "step": 29340
    },
    {
      "epoch": 37.01546975546975,
      "grad_norm": 0.0001762336032697931,
      "learning_rate": 2.7170027170027173e-06,
      "loss": 0.001,
      "step": 29350
    },
    {
      "epoch": 37.015727155727156,
      "grad_norm": 0.0002292945428052917,
      "learning_rate": 2.7141427141427146e-06,
      "loss": 0.1161,
      "step": 29360
    },
    {
      "epoch": 37.01598455598456,
      "grad_norm": 0.0370081327855587,
      "learning_rate": 2.7112827112827115e-06,
      "loss": 0.0004,
      "step": 29370
    },
    {
      "epoch": 37.016241956241956,
      "grad_norm": 0.04158321022987366,
      "learning_rate": 2.708422708422709e-06,
      "loss": 0.0004,
      "step": 29380
    },
    {
      "epoch": 37.01649935649936,
      "grad_norm": 0.004836615175008774,
      "learning_rate": 2.705562705562706e-06,
      "loss": 0.7461,
      "step": 29390
    },
    {
      "epoch": 37.016756756756756,
      "grad_norm": 0.025562850758433342,
      "learning_rate": 2.702702702702703e-06,
      "loss": 0.5474,
      "step": 29400
    },
    {
      "epoch": 37.01701415701416,
      "grad_norm": 0.00021090406517032534,
      "learning_rate": 2.6998426998427003e-06,
      "loss": 0.0001,
      "step": 29410
    },
    {
      "epoch": 37.017271557271556,
      "grad_norm": 0.00023956765653565526,
      "learning_rate": 2.6969826969826972e-06,
      "loss": 0.1954,
      "step": 29420
    },
    {
      "epoch": 37.01752895752896,
      "grad_norm": 0.0024973272811621428,
      "learning_rate": 2.6941226941226945e-06,
      "loss": 0.0784,
      "step": 29430
    },
    {
      "epoch": 37.017786357786356,
      "grad_norm": 0.03849514201283455,
      "learning_rate": 2.6912626912626914e-06,
      "loss": 0.6281,
      "step": 29440
    },
    {
      "epoch": 37.01804375804376,
      "grad_norm": 0.0002072292409138754,
      "learning_rate": 2.6884026884026887e-06,
      "loss": 1.933,
      "step": 29450
    },
    {
      "epoch": 37.018301158301156,
      "grad_norm": 0.0017715301364660263,
      "learning_rate": 2.6855426855426856e-06,
      "loss": 0.0,
      "step": 29460
    },
    {
      "epoch": 37.01855855855856,
      "grad_norm": 0.00020758021855726838,
      "learning_rate": 2.682682682682683e-06,
      "loss": 0.8772,
      "step": 29470
    },
    {
      "epoch": 37.018815958815956,
      "grad_norm": 0.036694519221782684,
      "learning_rate": 2.6798226798226803e-06,
      "loss": 0.0022,
      "step": 29480
    },
    {
      "epoch": 37.01907335907336,
      "grad_norm": 0.5028038024902344,
      "learning_rate": 2.676962676962677e-06,
      "loss": 0.0005,
      "step": 29490
    },
    {
      "epoch": 37.01933075933076,
      "grad_norm": 0.08280939608812332,
      "learning_rate": 2.6741026741026745e-06,
      "loss": 0.7947,
      "step": 29500
    },
    {
      "epoch": 37.01958815958816,
      "grad_norm": 0.031048625707626343,
      "learning_rate": 2.6712426712426714e-06,
      "loss": 0.0004,
      "step": 29510
    },
    {
      "epoch": 37.01984555984556,
      "grad_norm": 0.08677981793880463,
      "learning_rate": 2.6683826683826687e-06,
      "loss": 0.0005,
      "step": 29520
    },
    {
      "epoch": 37.02,
      "eval_accuracy": 0.8695652173913043,
      "eval_loss": 0.9154806733131409,
      "eval_runtime": 13.414,
      "eval_samples_per_second": 3.429,
      "eval_steps_per_second": 3.429,
      "step": 29526
    },
    {
      "epoch": 38.00010296010296,
      "grad_norm": 57.652462005615234,
      "learning_rate": 2.6655226655226656e-06,
      "loss": 2.1915,
      "step": 29530
    },
    {
      "epoch": 38.00036036036036,
      "grad_norm": 3.105980396270752,
      "learning_rate": 2.662662662662663e-06,
      "loss": 0.0005,
      "step": 29540
    },
    {
      "epoch": 38.00061776061776,
      "grad_norm": 0.3790825307369232,
      "learning_rate": 2.65980265980266e-06,
      "loss": 0.0005,
      "step": 29550
    },
    {
      "epoch": 38.00087516087516,
      "grad_norm": 0.0002181073505198583,
      "learning_rate": 2.656942656942657e-06,
      "loss": 0.3792,
      "step": 29560
    },
    {
      "epoch": 38.001132561132565,
      "grad_norm": 0.0001607909653102979,
      "learning_rate": 2.654082654082654e-06,
      "loss": 0.0001,
      "step": 29570
    },
    {
      "epoch": 38.00138996138996,
      "grad_norm": 0.10881076753139496,
      "learning_rate": 2.6512226512226513e-06,
      "loss": 0.7041,
      "step": 29580
    },
    {
      "epoch": 38.001647361647365,
      "grad_norm": 615.3811645507812,
      "learning_rate": 2.6483626483626486e-06,
      "loss": 0.0372,
      "step": 29590
    },
    {
      "epoch": 38.00190476190476,
      "grad_norm": 0.00022765577887184918,
      "learning_rate": 2.6455026455026455e-06,
      "loss": 0.0003,
      "step": 29600
    },
    {
      "epoch": 38.002162162162165,
      "grad_norm": 0.0011044471757486463,
      "learning_rate": 2.642642642642643e-06,
      "loss": 0.0001,
      "step": 29610
    },
    {
      "epoch": 38.00241956241956,
      "grad_norm": 0.43733513355255127,
      "learning_rate": 2.6397826397826397e-06,
      "loss": 0.0011,
      "step": 29620
    },
    {
      "epoch": 38.002676962676965,
      "grad_norm": 0.00021193055727053434,
      "learning_rate": 2.636922636922637e-06,
      "loss": 0.8158,
      "step": 29630
    },
    {
      "epoch": 38.00293436293436,
      "grad_norm": 0.00447743758559227,
      "learning_rate": 2.634062634062634e-06,
      "loss": 0.5663,
      "step": 29640
    },
    {
      "epoch": 38.003191763191765,
      "grad_norm": 576.5663452148438,
      "learning_rate": 2.6312026312026313e-06,
      "loss": 0.4441,
      "step": 29650
    },
    {
      "epoch": 38.00344916344916,
      "grad_norm": 0.006028820760548115,
      "learning_rate": 2.628342628342628e-06,
      "loss": 0.0002,
      "step": 29660
    },
    {
      "epoch": 38.003706563706565,
      "grad_norm": 0.01456479262560606,
      "learning_rate": 2.6254826254826255e-06,
      "loss": 0.3368,
      "step": 29670
    },
    {
      "epoch": 38.00396396396396,
      "grad_norm": 0.006685736123472452,
      "learning_rate": 2.6226226226226224e-06,
      "loss": 0.0004,
      "step": 29680
    },
    {
      "epoch": 38.004221364221365,
      "grad_norm": 0.016805648803710938,
      "learning_rate": 2.61976261976262e-06,
      "loss": 0.1144,
      "step": 29690
    },
    {
      "epoch": 38.00447876447876,
      "grad_norm": 0.00028289741021580994,
      "learning_rate": 2.6169026169026174e-06,
      "loss": 0.0001,
      "step": 29700
    },
    {
      "epoch": 38.004736164736165,
      "grad_norm": 0.00025284860748797655,
      "learning_rate": 2.6140426140426143e-06,
      "loss": 0.4801,
      "step": 29710
    },
    {
      "epoch": 38.00499356499356,
      "grad_norm": 0.0012302598915994167,
      "learning_rate": 2.6111826111826116e-06,
      "loss": 0.6057,
      "step": 29720
    },
    {
      "epoch": 38.005250965250966,
      "grad_norm": 0.04306511580944061,
      "learning_rate": 2.608322608322609e-06,
      "loss": 0.4271,
      "step": 29730
    },
    {
      "epoch": 38.00550836550836,
      "grad_norm": 0.1301436573266983,
      "learning_rate": 2.605462605462606e-06,
      "loss": 0.0005,
      "step": 29740
    },
    {
      "epoch": 38.005765765765766,
      "grad_norm": 0.14988690614700317,
      "learning_rate": 2.602602602602603e-06,
      "loss": 0.0001,
      "step": 29750
    },
    {
      "epoch": 38.00602316602317,
      "grad_norm": 0.002918446669355035,
      "learning_rate": 2.5997425997426e-06,
      "loss": 0.0005,
      "step": 29760
    },
    {
      "epoch": 38.006280566280566,
      "grad_norm": 0.005147375166416168,
      "learning_rate": 2.5968825968825974e-06,
      "loss": 0.7509,
      "step": 29770
    },
    {
      "epoch": 38.00653796653797,
      "grad_norm": 0.00018102384638041258,
      "learning_rate": 2.5940225940225943e-06,
      "loss": 0.6546,
      "step": 29780
    },
    {
      "epoch": 38.006795366795366,
      "grad_norm": 0.03137935698032379,
      "learning_rate": 2.5911625911625916e-06,
      "loss": 0.5553,
      "step": 29790
    },
    {
      "epoch": 38.00705276705277,
      "grad_norm": 0.005157016683369875,
      "learning_rate": 2.5883025883025885e-06,
      "loss": 0.0019,
      "step": 29800
    },
    {
      "epoch": 38.007310167310166,
      "grad_norm": 0.00020223679894115776,
      "learning_rate": 2.5854425854425858e-06,
      "loss": 0.0002,
      "step": 29810
    },
    {
      "epoch": 38.00756756756757,
      "grad_norm": 0.10321886092424393,
      "learning_rate": 2.5825825825825827e-06,
      "loss": 0.0002,
      "step": 29820
    },
    {
      "epoch": 38.007824967824966,
      "grad_norm": 0.0003532360424287617,
      "learning_rate": 2.57972257972258e-06,
      "loss": 0.0004,
      "step": 29830
    },
    {
      "epoch": 38.00808236808237,
      "grad_norm": 0.0043226853013038635,
      "learning_rate": 2.5768625768625773e-06,
      "loss": 0.4117,
      "step": 29840
    },
    {
      "epoch": 38.008339768339766,
      "grad_norm": 0.017380457371473312,
      "learning_rate": 2.574002574002574e-06,
      "loss": 0.0001,
      "step": 29850
    },
    {
      "epoch": 38.00859716859717,
      "grad_norm": 0.0010595513740554452,
      "learning_rate": 2.5711425711425715e-06,
      "loss": 0.0008,
      "step": 29860
    },
    {
      "epoch": 38.00885456885457,
      "grad_norm": 0.00034468816011212766,
      "learning_rate": 2.5682825682825684e-06,
      "loss": 0.5743,
      "step": 29870
    },
    {
      "epoch": 38.00911196911197,
      "grad_norm": 0.0005547988112084568,
      "learning_rate": 2.5654225654225657e-06,
      "loss": 0.0002,
      "step": 29880
    },
    {
      "epoch": 38.00936936936937,
      "grad_norm": 0.00622868025675416,
      "learning_rate": 2.5625625625625626e-06,
      "loss": 0.0002,
      "step": 29890
    },
    {
      "epoch": 38.00962676962677,
      "grad_norm": 0.01443453598767519,
      "learning_rate": 2.55970255970256e-06,
      "loss": 0.0008,
      "step": 29900
    },
    {
      "epoch": 38.00988416988417,
      "grad_norm": 0.00020728138042613864,
      "learning_rate": 2.556842556842557e-06,
      "loss": 0.0002,
      "step": 29910
    },
    {
      "epoch": 38.01014157014157,
      "grad_norm": 504.9588928222656,
      "learning_rate": 2.553982553982554e-06,
      "loss": 0.4423,
      "step": 29920
    },
    {
      "epoch": 38.01039897039897,
      "grad_norm": 0.012988753616809845,
      "learning_rate": 2.5511225511225515e-06,
      "loss": 0.0001,
      "step": 29930
    },
    {
      "epoch": 38.01065637065637,
      "grad_norm": 0.0057409703731536865,
      "learning_rate": 2.5482625482625484e-06,
      "loss": 0.0052,
      "step": 29940
    },
    {
      "epoch": 38.010913770913774,
      "grad_norm": 0.00012208484986331314,
      "learning_rate": 2.5454025454025457e-06,
      "loss": 0.001,
      "step": 29950
    },
    {
      "epoch": 38.01117117117117,
      "grad_norm": 0.02826821058988571,
      "learning_rate": 2.5425425425425426e-06,
      "loss": 0.0002,
      "step": 29960
    },
    {
      "epoch": 38.011428571428574,
      "grad_norm": 352.06964111328125,
      "learning_rate": 2.53968253968254e-06,
      "loss": 1.1301,
      "step": 29970
    },
    {
      "epoch": 38.01168597168597,
      "grad_norm": 0.0001997912477236241,
      "learning_rate": 2.5368225368225368e-06,
      "loss": 0.0001,
      "step": 29980
    },
    {
      "epoch": 38.011943371943374,
      "grad_norm": 0.00014716804435011,
      "learning_rate": 2.533962533962534e-06,
      "loss": 1.1997,
      "step": 29990
    },
    {
      "epoch": 38.01220077220077,
      "grad_norm": 2.6033856868743896,
      "learning_rate": 2.531102531102531e-06,
      "loss": 0.0005,
      "step": 30000
    },
    {
      "epoch": 38.012458172458174,
      "grad_norm": 0.00014160611317493021,
      "learning_rate": 2.5282425282425283e-06,
      "loss": 0.1244,
      "step": 30010
    },
    {
      "epoch": 38.01271557271557,
      "grad_norm": 0.0004008954856544733,
      "learning_rate": 2.525382525382525e-06,
      "loss": 0.0001,
      "step": 30020
    },
    {
      "epoch": 38.012972972972975,
      "grad_norm": 0.0023208346683532,
      "learning_rate": 2.5225225225225225e-06,
      "loss": 0.0044,
      "step": 30030
    },
    {
      "epoch": 38.01323037323037,
      "grad_norm": 0.003921369090676308,
      "learning_rate": 2.51966251966252e-06,
      "loss": 0.0006,
      "step": 30040
    },
    {
      "epoch": 38.013487773487775,
      "grad_norm": 0.005466674454510212,
      "learning_rate": 2.5168025168025167e-06,
      "loss": 0.0001,
      "step": 30050
    },
    {
      "epoch": 38.01374517374517,
      "grad_norm": 0.00016683118883520365,
      "learning_rate": 2.513942513942514e-06,
      "loss": 0.0001,
      "step": 30060
    },
    {
      "epoch": 38.014002574002575,
      "grad_norm": 0.00019211266771890223,
      "learning_rate": 2.5110825110825113e-06,
      "loss": 0.0003,
      "step": 30070
    },
    {
      "epoch": 38.01425997425997,
      "grad_norm": 0.00022057870228309184,
      "learning_rate": 2.5082225082225087e-06,
      "loss": 0.0004,
      "step": 30080
    },
    {
      "epoch": 38.014517374517375,
      "grad_norm": 0.00017979362746700644,
      "learning_rate": 2.505362505362506e-06,
      "loss": 0.0006,
      "step": 30090
    },
    {
      "epoch": 38.01477477477477,
      "grad_norm": 0.00035356191801838577,
      "learning_rate": 2.502502502502503e-06,
      "loss": 0.0001,
      "step": 30100
    },
    {
      "epoch": 38.015032175032175,
      "grad_norm": 0.010907981544733047,
      "learning_rate": 2.4996424996424998e-06,
      "loss": 0.1697,
      "step": 30110
    },
    {
      "epoch": 38.01528957528958,
      "grad_norm": 0.03007809817790985,
      "learning_rate": 2.496782496782497e-06,
      "loss": 0.0002,
      "step": 30120
    },
    {
      "epoch": 38.015546975546975,
      "grad_norm": 0.001186450943350792,
      "learning_rate": 2.493922493922494e-06,
      "loss": 0.0283,
      "step": 30130
    },
    {
      "epoch": 38.01580437580438,
      "grad_norm": 0.021413983777165413,
      "learning_rate": 2.4910624910624913e-06,
      "loss": 0.0001,
      "step": 30140
    },
    {
      "epoch": 38.016061776061775,
      "grad_norm": 0.00018151052063331008,
      "learning_rate": 2.488202488202488e-06,
      "loss": 0.768,
      "step": 30150
    },
    {
      "epoch": 38.01631917631918,
      "grad_norm": 0.00312127941288054,
      "learning_rate": 2.4853424853424855e-06,
      "loss": 0.0005,
      "step": 30160
    },
    {
      "epoch": 38.016576576576576,
      "grad_norm": 0.0001102391179301776,
      "learning_rate": 2.482482482482483e-06,
      "loss": 0.0001,
      "step": 30170
    },
    {
      "epoch": 38.01683397683398,
      "grad_norm": 0.0002471956831868738,
      "learning_rate": 2.47962247962248e-06,
      "loss": 0.0007,
      "step": 30180
    },
    {
      "epoch": 38.017091377091376,
      "grad_norm": 0.0004663833533413708,
      "learning_rate": 2.476762476762477e-06,
      "loss": 0.0129,
      "step": 30190
    },
    {
      "epoch": 38.01734877734878,
      "grad_norm": 0.00016829062951728702,
      "learning_rate": 2.4739024739024743e-06,
      "loss": 0.3392,
      "step": 30200
    },
    {
      "epoch": 38.017606177606176,
      "grad_norm": 0.003199836239218712,
      "learning_rate": 2.4710424710424712e-06,
      "loss": 0.7917,
      "step": 30210
    },
    {
      "epoch": 38.01786357786358,
      "grad_norm": 0.0024653258733451366,
      "learning_rate": 2.4681824681824685e-06,
      "loss": 0.4412,
      "step": 30220
    },
    {
      "epoch": 38.018120978120976,
      "grad_norm": 0.01302412524819374,
      "learning_rate": 2.4653224653224654e-06,
      "loss": 0.0155,
      "step": 30230
    },
    {
      "epoch": 38.01837837837838,
      "grad_norm": 0.00023153946676757187,
      "learning_rate": 2.4624624624624628e-06,
      "loss": 0.0001,
      "step": 30240
    },
    {
      "epoch": 38.018635778635776,
      "grad_norm": 0.010636151768267155,
      "learning_rate": 2.4596024596024596e-06,
      "loss": 0.0001,
      "step": 30250
    },
    {
      "epoch": 38.01889317889318,
      "grad_norm": 0.019402464851737022,
      "learning_rate": 2.456742456742457e-06,
      "loss": 0.8413,
      "step": 30260
    },
    {
      "epoch": 38.019150579150576,
      "grad_norm": 0.00025949705741368234,
      "learning_rate": 2.453882453882454e-06,
      "loss": 0.0014,
      "step": 30270
    },
    {
      "epoch": 38.01940797940798,
      "grad_norm": 0.014147725887596607,
      "learning_rate": 2.451022451022451e-06,
      "loss": 0.0003,
      "step": 30280
    },
    {
      "epoch": 38.019665379665376,
      "grad_norm": 0.00033936629188247025,
      "learning_rate": 2.4481624481624485e-06,
      "loss": 0.0024,
      "step": 30290
    },
    {
      "epoch": 38.01992277992278,
      "grad_norm": 0.00021008138719480485,
      "learning_rate": 2.4453024453024454e-06,
      "loss": 0.2137,
      "step": 30300
    },
    {
      "epoch": 38.02,
      "eval_accuracy": 0.9130434782608695,
      "eval_loss": 0.5413678884506226,
      "eval_runtime": 13.3642,
      "eval_samples_per_second": 3.442,
      "eval_steps_per_second": 3.442,
      "step": 30303
    },
    {
      "epoch": 39.00018018018018,
      "grad_norm": 0.047418005764484406,
      "learning_rate": 2.4424424424424427e-06,
      "loss": 0.9208,
      "step": 30310
    },
    {
      "epoch": 39.00043758043758,
      "grad_norm": 0.010265672579407692,
      "learning_rate": 2.4395824395824396e-06,
      "loss": 0.0001,
      "step": 30320
    },
    {
      "epoch": 39.00069498069498,
      "grad_norm": 0.0024453650694340467,
      "learning_rate": 2.436722436722437e-06,
      "loss": 0.0001,
      "step": 30330
    },
    {
      "epoch": 39.000952380952384,
      "grad_norm": 0.007709293160587549,
      "learning_rate": 2.433862433862434e-06,
      "loss": 0.0023,
      "step": 30340
    },
    {
      "epoch": 39.00120978120978,
      "grad_norm": 0.03284775838255882,
      "learning_rate": 2.431002431002431e-06,
      "loss": 0.0002,
      "step": 30350
    },
    {
      "epoch": 39.001467181467184,
      "grad_norm": 0.00015525065828114748,
      "learning_rate": 2.4281424281424284e-06,
      "loss": 0.0152,
      "step": 30360
    },
    {
      "epoch": 39.00172458172458,
      "grad_norm": 0.0003922375326510519,
      "learning_rate": 2.4252824252824258e-06,
      "loss": 1.1854,
      "step": 30370
    },
    {
      "epoch": 39.001981981981984,
      "grad_norm": 0.0003774458891712129,
      "learning_rate": 2.4224224224224226e-06,
      "loss": 0.0861,
      "step": 30380
    },
    {
      "epoch": 39.00223938223938,
      "grad_norm": 0.00016582466196268797,
      "learning_rate": 2.41956241956242e-06,
      "loss": 0.0001,
      "step": 30390
    },
    {
      "epoch": 39.002496782496785,
      "grad_norm": 0.00012327576405368745,
      "learning_rate": 2.416702416702417e-06,
      "loss": 0.0001,
      "step": 30400
    },
    {
      "epoch": 39.00275418275418,
      "grad_norm": 0.06711563467979431,
      "learning_rate": 2.413842413842414e-06,
      "loss": 0.0002,
      "step": 30410
    },
    {
      "epoch": 39.003011583011585,
      "grad_norm": 0.06605685502290726,
      "learning_rate": 2.410982410982411e-06,
      "loss": 0.0006,
      "step": 30420
    },
    {
      "epoch": 39.00326898326898,
      "grad_norm": 0.00011760670895455405,
      "learning_rate": 2.4081224081224084e-06,
      "loss": 1.4446,
      "step": 30430
    },
    {
      "epoch": 39.003526383526385,
      "grad_norm": 0.015367030166089535,
      "learning_rate": 2.4052624052624053e-06,
      "loss": 0.731,
      "step": 30440
    },
    {
      "epoch": 39.00378378378378,
      "grad_norm": 612.0079345703125,
      "learning_rate": 2.4024024024024026e-06,
      "loss": 0.4344,
      "step": 30450
    },
    {
      "epoch": 39.004041184041185,
      "grad_norm": 0.16250044107437134,
      "learning_rate": 2.3995423995424e-06,
      "loss": 0.0002,
      "step": 30460
    },
    {
      "epoch": 39.00429858429858,
      "grad_norm": 0.00013442897761706263,
      "learning_rate": 2.396682396682397e-06,
      "loss": 0.0035,
      "step": 30470
    },
    {
      "epoch": 39.004555984555985,
      "grad_norm": 0.03660891577601433,
      "learning_rate": 2.393822393822394e-06,
      "loss": 0.0001,
      "step": 30480
    },
    {
      "epoch": 39.00481338481338,
      "grad_norm": 0.01504859421402216,
      "learning_rate": 2.390962390962391e-06,
      "loss": 0.0011,
      "step": 30490
    },
    {
      "epoch": 39.005070785070785,
      "grad_norm": 0.6826884150505066,
      "learning_rate": 2.3881023881023883e-06,
      "loss": 0.0003,
      "step": 30500
    },
    {
      "epoch": 39.00532818532819,
      "grad_norm": 0.0002525031450204551,
      "learning_rate": 2.3852423852423852e-06,
      "loss": 0.0001,
      "step": 30510
    },
    {
      "epoch": 39.005585585585585,
      "grad_norm": 0.00015134667046368122,
      "learning_rate": 2.3823823823823825e-06,
      "loss": 0.0003,
      "step": 30520
    },
    {
      "epoch": 39.00584298584299,
      "grad_norm": 0.045858077704906464,
      "learning_rate": 2.37952237952238e-06,
      "loss": 0.0002,
      "step": 30530
    },
    {
      "epoch": 39.006100386100385,
      "grad_norm": 0.000713690766133368,
      "learning_rate": 2.376662376662377e-06,
      "loss": 0.0001,
      "step": 30540
    },
    {
      "epoch": 39.00635778635779,
      "grad_norm": 0.00014348112745210528,
      "learning_rate": 2.373802373802374e-06,
      "loss": 0.0206,
      "step": 30550
    },
    {
      "epoch": 39.006615186615186,
      "grad_norm": 0.017485586926341057,
      "learning_rate": 2.3709423709423714e-06,
      "loss": 0.5571,
      "step": 30560
    },
    {
      "epoch": 39.00687258687259,
      "grad_norm": 0.0076943435706198215,
      "learning_rate": 2.3680823680823683e-06,
      "loss": 0.0036,
      "step": 30570
    },
    {
      "epoch": 39.007129987129986,
      "grad_norm": 0.006811995059251785,
      "learning_rate": 2.3652223652223656e-06,
      "loss": 0.0001,
      "step": 30580
    },
    {
      "epoch": 39.00738738738739,
      "grad_norm": 0.00021085226035211235,
      "learning_rate": 2.3623623623623625e-06,
      "loss": 0.0002,
      "step": 30590
    },
    {
      "epoch": 39.007644787644786,
      "grad_norm": 0.00011505138536449522,
      "learning_rate": 2.35950235950236e-06,
      "loss": 0.0006,
      "step": 30600
    },
    {
      "epoch": 39.00790218790219,
      "grad_norm": 0.0032912762835621834,
      "learning_rate": 2.3566423566423567e-06,
      "loss": 0.0001,
      "step": 30610
    },
    {
      "epoch": 39.008159588159586,
      "grad_norm": 0.00032590367482043803,
      "learning_rate": 2.353782353782354e-06,
      "loss": 0.0005,
      "step": 30620
    },
    {
      "epoch": 39.00841698841699,
      "grad_norm": 0.10024252533912659,
      "learning_rate": 2.3509223509223513e-06,
      "loss": 0.0002,
      "step": 30630
    },
    {
      "epoch": 39.008674388674386,
      "grad_norm": 0.0001177530357381329,
      "learning_rate": 2.348062348062348e-06,
      "loss": 0.8669,
      "step": 30640
    },
    {
      "epoch": 39.00893178893179,
      "grad_norm": 0.009781216271221638,
      "learning_rate": 2.3452023452023455e-06,
      "loss": 0.0001,
      "step": 30650
    },
    {
      "epoch": 39.009189189189186,
      "grad_norm": 0.0025387846399098635,
      "learning_rate": 2.3423423423423424e-06,
      "loss": 0.0011,
      "step": 30660
    },
    {
      "epoch": 39.00944658944659,
      "grad_norm": 0.00022111479484010488,
      "learning_rate": 2.3394823394823397e-06,
      "loss": 0.0857,
      "step": 30670
    },
    {
      "epoch": 39.009703989703986,
      "grad_norm": 0.004516247659921646,
      "learning_rate": 2.3366223366223366e-06,
      "loss": 0.0012,
      "step": 30680
    },
    {
      "epoch": 39.00996138996139,
      "grad_norm": 0.00013777939602732658,
      "learning_rate": 2.333762333762334e-06,
      "loss": 0.6783,
      "step": 30690
    },
    {
      "epoch": 39.010218790218794,
      "grad_norm": 0.00011778510815929621,
      "learning_rate": 2.330902330902331e-06,
      "loss": 0.9963,
      "step": 30700
    },
    {
      "epoch": 39.01047619047619,
      "grad_norm": 0.010461336001753807,
      "learning_rate": 2.328042328042328e-06,
      "loss": 0.7411,
      "step": 30710
    },
    {
      "epoch": 39.010733590733594,
      "grad_norm": 0.016033122316002846,
      "learning_rate": 2.3251823251823255e-06,
      "loss": 0.6801,
      "step": 30720
    },
    {
      "epoch": 39.01099099099099,
      "grad_norm": 0.003915534354746342,
      "learning_rate": 2.3223223223223228e-06,
      "loss": 0.603,
      "step": 30730
    },
    {
      "epoch": 39.011248391248394,
      "grad_norm": 0.02247695066034794,
      "learning_rate": 2.3194623194623197e-06,
      "loss": 0.033,
      "step": 30740
    },
    {
      "epoch": 39.01150579150579,
      "grad_norm": 0.008587141521275043,
      "learning_rate": 2.316602316602317e-06,
      "loss": 0.6416,
      "step": 30750
    },
    {
      "epoch": 39.011763191763194,
      "grad_norm": 0.0012964047491550446,
      "learning_rate": 2.313742313742314e-06,
      "loss": 0.0001,
      "step": 30760
    },
    {
      "epoch": 39.01202059202059,
      "grad_norm": 0.007788754999637604,
      "learning_rate": 2.310882310882311e-06,
      "loss": 0.0002,
      "step": 30770
    },
    {
      "epoch": 39.012277992277994,
      "grad_norm": 0.00040538437315262854,
      "learning_rate": 2.308022308022308e-06,
      "loss": 0.7998,
      "step": 30780
    },
    {
      "epoch": 39.01253539253539,
      "grad_norm": 0.00013059898628853261,
      "learning_rate": 2.3051623051623054e-06,
      "loss": 0.6301,
      "step": 30790
    },
    {
      "epoch": 39.012792792792794,
      "grad_norm": 0.010551397688686848,
      "learning_rate": 2.3023023023023023e-06,
      "loss": 0.0051,
      "step": 30800
    },
    {
      "epoch": 39.01305019305019,
      "grad_norm": 0.00023208599304780364,
      "learning_rate": 2.2994422994422996e-06,
      "loss": 0.6185,
      "step": 30810
    },
    {
      "epoch": 39.013307593307594,
      "grad_norm": 0.003290843451395631,
      "learning_rate": 2.296582296582297e-06,
      "loss": 0.0001,
      "step": 30820
    },
    {
      "epoch": 39.01356499356499,
      "grad_norm": 0.005354827735573053,
      "learning_rate": 2.293722293722294e-06,
      "loss": 0.0002,
      "step": 30830
    },
    {
      "epoch": 39.013822393822394,
      "grad_norm": 0.001084979739971459,
      "learning_rate": 2.290862290862291e-06,
      "loss": 0.0002,
      "step": 30840
    },
    {
      "epoch": 39.01407979407979,
      "grad_norm": 0.07646601647138596,
      "learning_rate": 2.288002288002288e-06,
      "loss": 0.0003,
      "step": 30850
    },
    {
      "epoch": 39.014337194337195,
      "grad_norm": 0.004698839038610458,
      "learning_rate": 2.2851422851422854e-06,
      "loss": 0.214,
      "step": 30860
    },
    {
      "epoch": 39.01459459459459,
      "grad_norm": 0.0005614709225483239,
      "learning_rate": 2.2822822822822822e-06,
      "loss": 0.0001,
      "step": 30870
    },
    {
      "epoch": 39.014851994851995,
      "grad_norm": 0.00012098588194930926,
      "learning_rate": 2.2794222794222796e-06,
      "loss": 0.0002,
      "step": 30880
    },
    {
      "epoch": 39.0151093951094,
      "grad_norm": 0.13583892583847046,
      "learning_rate": 2.2765622765622765e-06,
      "loss": 0.0007,
      "step": 30890
    },
    {
      "epoch": 39.015366795366795,
      "grad_norm": 0.00043689829180948436,
      "learning_rate": 2.2737022737022738e-06,
      "loss": 0.0003,
      "step": 30900
    },
    {
      "epoch": 39.0156241956242,
      "grad_norm": 0.002140444004908204,
      "learning_rate": 2.270842270842271e-06,
      "loss": 1.9341,
      "step": 30910
    },
    {
      "epoch": 39.015881595881595,
      "grad_norm": 0.04902840033173561,
      "learning_rate": 2.2679822679822684e-06,
      "loss": 0.0004,
      "step": 30920
    },
    {
      "epoch": 39.016138996139,
      "grad_norm": 0.00013759516878053546,
      "learning_rate": 2.2651222651222653e-06,
      "loss": 0.4478,
      "step": 30930
    },
    {
      "epoch": 39.016396396396395,
      "grad_norm": 158.75863647460938,
      "learning_rate": 2.2622622622622626e-06,
      "loss": 1.3205,
      "step": 30940
    },
    {
      "epoch": 39.0166537966538,
      "grad_norm": 0.49175313115119934,
      "learning_rate": 2.2594022594022595e-06,
      "loss": 0.0002,
      "step": 30950
    },
    {
      "epoch": 39.016911196911195,
      "grad_norm": 0.036261558532714844,
      "learning_rate": 2.256542256542257e-06,
      "loss": 0.0001,
      "step": 30960
    },
    {
      "epoch": 39.0171685971686,
      "grad_norm": 0.00017356501484755427,
      "learning_rate": 2.2536822536822537e-06,
      "loss": 0.5837,
      "step": 30970
    },
    {
      "epoch": 39.017425997425995,
      "grad_norm": 0.000755143875721842,
      "learning_rate": 2.250822250822251e-06,
      "loss": 0.0003,
      "step": 30980
    },
    {
      "epoch": 39.0176833976834,
      "grad_norm": 0.0005234266864135861,
      "learning_rate": 2.2479622479622483e-06,
      "loss": 0.0002,
      "step": 30990
    },
    {
      "epoch": 39.017940797940796,
      "grad_norm": 0.0005713241989724338,
      "learning_rate": 2.2451022451022452e-06,
      "loss": 0.0001,
      "step": 31000
    },
    {
      "epoch": 39.0181981981982,
      "grad_norm": 0.005339688155800104,
      "learning_rate": 2.2422422422422426e-06,
      "loss": 0.5243,
      "step": 31010
    },
    {
      "epoch": 39.018455598455596,
      "grad_norm": 0.00022389739751815796,
      "learning_rate": 2.2393822393822394e-06,
      "loss": 0.0005,
      "step": 31020
    },
    {
      "epoch": 39.018712998713,
      "grad_norm": 0.00026733646518550813,
      "learning_rate": 2.2365222365222368e-06,
      "loss": 0.0002,
      "step": 31030
    },
    {
      "epoch": 39.018970398970396,
      "grad_norm": 0.009177249856293201,
      "learning_rate": 2.2336622336622337e-06,
      "loss": 0.2119,
      "step": 31040
    },
    {
      "epoch": 39.0192277992278,
      "grad_norm": 0.12196987867355347,
      "learning_rate": 2.230802230802231e-06,
      "loss": 0.0001,
      "step": 31050
    },
    {
      "epoch": 39.0194851994852,
      "grad_norm": 0.021783724427223206,
      "learning_rate": 2.227942227942228e-06,
      "loss": 0.691,
      "step": 31060
    },
    {
      "epoch": 39.0197425997426,
      "grad_norm": 0.007660671602934599,
      "learning_rate": 2.225082225082225e-06,
      "loss": 0.0002,
      "step": 31070
    },
    {
      "epoch": 39.02,
      "grad_norm": 0.0010431857081130147,
      "learning_rate": 2.222222222222222e-06,
      "loss": 0.0001,
      "step": 31080
    },
    {
      "epoch": 39.02,
      "eval_accuracy": 0.8260869565217391,
      "eval_loss": 0.9474976658821106,
      "eval_runtime": 13.4713,
      "eval_samples_per_second": 3.415,
      "eval_steps_per_second": 3.415,
      "step": 31080
    },
    {
      "epoch": 40.000257400257404,
      "grad_norm": 1.181172251701355,
      "learning_rate": 2.2193622193622194e-06,
      "loss": 0.0002,
      "step": 31090
    },
    {
      "epoch": 40.0005148005148,
      "grad_norm": 318.863037109375,
      "learning_rate": 2.2165022165022167e-06,
      "loss": 0.7819,
      "step": 31100
    },
    {
      "epoch": 40.000772200772204,
      "grad_norm": 0.00014746197848580778,
      "learning_rate": 2.213642213642214e-06,
      "loss": 0.0002,
      "step": 31110
    },
    {
      "epoch": 40.0010296010296,
      "grad_norm": 144.7231903076172,
      "learning_rate": 2.210782210782211e-06,
      "loss": 0.7112,
      "step": 31120
    },
    {
      "epoch": 40.001287001287004,
      "grad_norm": 0.00015773468476254493,
      "learning_rate": 2.2079222079222082e-06,
      "loss": 0.2482,
      "step": 31130
    },
    {
      "epoch": 40.0015444015444,
      "grad_norm": 0.0001816485309973359,
      "learning_rate": 2.205062205062205e-06,
      "loss": 1.637,
      "step": 31140
    },
    {
      "epoch": 40.001801801801804,
      "grad_norm": 0.034062862396240234,
      "learning_rate": 2.2022022022022024e-06,
      "loss": 0.5285,
      "step": 31150
    },
    {
      "epoch": 40.0020592020592,
      "grad_norm": 0.0027663600631058216,
      "learning_rate": 2.1993421993421998e-06,
      "loss": 0.6391,
      "step": 31160
    },
    {
      "epoch": 40.002316602316604,
      "grad_norm": 0.010885432362556458,
      "learning_rate": 2.1964821964821967e-06,
      "loss": 0.0001,
      "step": 31170
    },
    {
      "epoch": 40.002574002574,
      "grad_norm": 0.04000730440020561,
      "learning_rate": 2.193622193622194e-06,
      "loss": 0.0004,
      "step": 31180
    },
    {
      "epoch": 40.002831402831404,
      "grad_norm": 0.004812656436115503,
      "learning_rate": 2.190762190762191e-06,
      "loss": 0.0017,
      "step": 31190
    },
    {
      "epoch": 40.0030888030888,
      "grad_norm": 0.06480229645967484,
      "learning_rate": 2.187902187902188e-06,
      "loss": 0.3643,
      "step": 31200
    },
    {
      "epoch": 40.003346203346204,
      "grad_norm": 0.003973096143454313,
      "learning_rate": 2.185042185042185e-06,
      "loss": 0.0004,
      "step": 31210
    },
    {
      "epoch": 40.0036036036036,
      "grad_norm": 0.0010743311140686274,
      "learning_rate": 2.1821821821821824e-06,
      "loss": 0.0001,
      "step": 31220
    },
    {
      "epoch": 40.003861003861005,
      "grad_norm": 0.01999141089618206,
      "learning_rate": 2.1793221793221793e-06,
      "loss": 0.6939,
      "step": 31230
    },
    {
      "epoch": 40.0041184041184,
      "grad_norm": 0.0030548074282705784,
      "learning_rate": 2.1764621764621766e-06,
      "loss": 0.8081,
      "step": 31240
    },
    {
      "epoch": 40.004375804375805,
      "grad_norm": 0.004580581560730934,
      "learning_rate": 2.1736021736021735e-06,
      "loss": 0.0003,
      "step": 31250
    },
    {
      "epoch": 40.0046332046332,
      "grad_norm": 0.6698066592216492,
      "learning_rate": 2.170742170742171e-06,
      "loss": 0.0003,
      "step": 31260
    },
    {
      "epoch": 40.004890604890605,
      "grad_norm": 0.00941177923232317,
      "learning_rate": 2.167882167882168e-06,
      "loss": 0.0002,
      "step": 31270
    },
    {
      "epoch": 40.00514800514801,
      "grad_norm": 0.0010183549020439386,
      "learning_rate": 2.1650221650221654e-06,
      "loss": 0.0001,
      "step": 31280
    },
    {
      "epoch": 40.005405405405405,
      "grad_norm": 8.488154411315918,
      "learning_rate": 2.1621621621621623e-06,
      "loss": 0.0007,
      "step": 31290
    },
    {
      "epoch": 40.00566280566281,
      "grad_norm": 0.0022410727106034756,
      "learning_rate": 2.1593021593021596e-06,
      "loss": 0.0209,
      "step": 31300
    },
    {
      "epoch": 40.005920205920205,
      "grad_norm": 0.034624092280864716,
      "learning_rate": 2.1564421564421565e-06,
      "loss": 0.0003,
      "step": 31310
    },
    {
      "epoch": 40.00617760617761,
      "grad_norm": 0.01160801388323307,
      "learning_rate": 2.153582153582154e-06,
      "loss": 0.5756,
      "step": 31320
    },
    {
      "epoch": 40.006435006435005,
      "grad_norm": 0.00036862597335129976,
      "learning_rate": 2.150722150722151e-06,
      "loss": 0.0001,
      "step": 31330
    },
    {
      "epoch": 40.00669240669241,
      "grad_norm": 0.00013573635078500956,
      "learning_rate": 2.147862147862148e-06,
      "loss": 0.9687,
      "step": 31340
    },
    {
      "epoch": 40.006949806949805,
      "grad_norm": 0.0024660457856953144,
      "learning_rate": 2.1450021450021454e-06,
      "loss": 0.0002,
      "step": 31350
    },
    {
      "epoch": 40.00720720720721,
      "grad_norm": 0.0038319462910294533,
      "learning_rate": 2.1421421421421423e-06,
      "loss": 0.0001,
      "step": 31360
    },
    {
      "epoch": 40.007464607464605,
      "grad_norm": 0.005298302508890629,
      "learning_rate": 2.1392821392821396e-06,
      "loss": 0.0001,
      "step": 31370
    },
    {
      "epoch": 40.00772200772201,
      "grad_norm": 0.00020501506514847279,
      "learning_rate": 2.1364221364221365e-06,
      "loss": 0.0001,
      "step": 31380
    },
    {
      "epoch": 40.007979407979406,
      "grad_norm": 0.003177186707034707,
      "learning_rate": 2.133562133562134e-06,
      "loss": 0.0004,
      "step": 31390
    },
    {
      "epoch": 40.00823680823681,
      "grad_norm": 0.0029437614139169455,
      "learning_rate": 2.1307021307021307e-06,
      "loss": 0.0002,
      "step": 31400
    },
    {
      "epoch": 40.008494208494206,
      "grad_norm": 0.00016003473137971014,
      "learning_rate": 2.127842127842128e-06,
      "loss": 0.0005,
      "step": 31410
    },
    {
      "epoch": 40.00875160875161,
      "grad_norm": 3.366640090942383,
      "learning_rate": 2.124982124982125e-06,
      "loss": 0.0017,
      "step": 31420
    },
    {
      "epoch": 40.009009009009006,
      "grad_norm": 0.00017837573250290006,
      "learning_rate": 2.1221221221221222e-06,
      "loss": 0.0,
      "step": 31430
    },
    {
      "epoch": 40.00926640926641,
      "grad_norm": 0.00028988305712118745,
      "learning_rate": 2.1192621192621195e-06,
      "loss": 0.0001,
      "step": 31440
    },
    {
      "epoch": 40.00952380952381,
      "grad_norm": 0.06588441133499146,
      "learning_rate": 2.1164021164021164e-06,
      "loss": 0.755,
      "step": 31450
    },
    {
      "epoch": 40.00978120978121,
      "grad_norm": 0.00015987700317054987,
      "learning_rate": 2.1135421135421137e-06,
      "loss": 0.2524,
      "step": 31460
    },
    {
      "epoch": 40.01003861003861,
      "grad_norm": 0.002687298459932208,
      "learning_rate": 2.110682110682111e-06,
      "loss": 0.001,
      "step": 31470
    },
    {
      "epoch": 40.01029601029601,
      "grad_norm": 0.016389530152082443,
      "learning_rate": 2.107822107822108e-06,
      "loss": 0.6573,
      "step": 31480
    },
    {
      "epoch": 40.01055341055341,
      "grad_norm": 0.0076734754256904125,
      "learning_rate": 2.1049621049621053e-06,
      "loss": 0.7998,
      "step": 31490
    },
    {
      "epoch": 40.01081081081081,
      "grad_norm": 0.3663455843925476,
      "learning_rate": 2.102102102102102e-06,
      "loss": 0.0005,
      "step": 31500
    },
    {
      "epoch": 40.01106821106821,
      "grad_norm": 0.0001318360009463504,
      "learning_rate": 2.0992420992420995e-06,
      "loss": 0.0001,
      "step": 31510
    },
    {
      "epoch": 40.01132561132561,
      "grad_norm": 0.5110870003700256,
      "learning_rate": 2.096382096382097e-06,
      "loss": 0.0766,
      "step": 31520
    },
    {
      "epoch": 40.011583011583014,
      "grad_norm": 0.00019474855798762292,
      "learning_rate": 2.0935220935220937e-06,
      "loss": 0.0004,
      "step": 31530
    },
    {
      "epoch": 40.01184041184041,
      "grad_norm": 0.04191095754504204,
      "learning_rate": 2.090662090662091e-06,
      "loss": 0.2068,
      "step": 31540
    },
    {
      "epoch": 40.012097812097814,
      "grad_norm": 0.08732067048549652,
      "learning_rate": 2.087802087802088e-06,
      "loss": 0.0001,
      "step": 31550
    },
    {
      "epoch": 40.01235521235521,
      "grad_norm": 0.00027026841416954994,
      "learning_rate": 2.084942084942085e-06,
      "loss": 0.0035,
      "step": 31560
    },
    {
      "epoch": 40.012612612612614,
      "grad_norm": 0.0018748992588371038,
      "learning_rate": 2.082082082082082e-06,
      "loss": 0.4388,
      "step": 31570
    },
    {
      "epoch": 40.01287001287001,
      "grad_norm": 0.00020655177650041878,
      "learning_rate": 2.0792220792220794e-06,
      "loss": 0.0002,
      "step": 31580
    },
    {
      "epoch": 40.013127413127414,
      "grad_norm": 0.0026093849446624517,
      "learning_rate": 2.0763620763620763e-06,
      "loss": 0.6074,
      "step": 31590
    },
    {
      "epoch": 40.01338481338481,
      "grad_norm": 0.0001526171836303547,
      "learning_rate": 2.0735020735020736e-06,
      "loss": 0.6795,
      "step": 31600
    },
    {
      "epoch": 40.013642213642214,
      "grad_norm": 0.0050857714377343655,
      "learning_rate": 2.070642070642071e-06,
      "loss": 0.0106,
      "step": 31610
    },
    {
      "epoch": 40.01389961389961,
      "grad_norm": 0.17710165679454803,
      "learning_rate": 2.067782067782068e-06,
      "loss": 0.0004,
      "step": 31620
    },
    {
      "epoch": 40.014157014157014,
      "grad_norm": 0.23658306896686554,
      "learning_rate": 2.064922064922065e-06,
      "loss": 0.0094,
      "step": 31630
    },
    {
      "epoch": 40.01441441441442,
      "grad_norm": 0.00034343061270192266,
      "learning_rate": 2.062062062062062e-06,
      "loss": 0.0001,
      "step": 31640
    },
    {
      "epoch": 40.014671814671814,
      "grad_norm": 0.00010646691225701943,
      "learning_rate": 2.0592020592020594e-06,
      "loss": 0.0001,
      "step": 31650
    },
    {
      "epoch": 40.01492921492922,
      "grad_norm": 0.0020296790171414614,
      "learning_rate": 2.0563420563420567e-06,
      "loss": 1.1483,
      "step": 31660
    },
    {
      "epoch": 40.015186615186614,
      "grad_norm": 0.00015698590141255409,
      "learning_rate": 2.0534820534820536e-06,
      "loss": 0.0,
      "step": 31670
    },
    {
      "epoch": 40.01544401544402,
      "grad_norm": 0.00016668056196067482,
      "learning_rate": 2.050622050622051e-06,
      "loss": 0.0001,
      "step": 31680
    },
    {
      "epoch": 40.015701415701415,
      "grad_norm": 0.0005182051681913435,
      "learning_rate": 2.047762047762048e-06,
      "loss": 0.0001,
      "step": 31690
    },
    {
      "epoch": 40.01595881595882,
      "grad_norm": 0.2584684491157532,
      "learning_rate": 2.044902044902045e-06,
      "loss": 0.0003,
      "step": 31700
    },
    {
      "epoch": 40.016216216216215,
      "grad_norm": 0.18375691771507263,
      "learning_rate": 2.0420420420420424e-06,
      "loss": 0.0001,
      "step": 31710
    },
    {
      "epoch": 40.01647361647362,
      "grad_norm": 0.00018776406068354845,
      "learning_rate": 2.0391820391820393e-06,
      "loss": 0.0,
      "step": 31720
    },
    {
      "epoch": 40.016731016731015,
      "grad_norm": 0.017002640292048454,
      "learning_rate": 2.0363220363220366e-06,
      "loss": 0.0001,
      "step": 31730
    },
    {
      "epoch": 40.01698841698842,
      "grad_norm": 0.0015852090436965227,
      "learning_rate": 2.0334620334620335e-06,
      "loss": 0.0001,
      "step": 31740
    },
    {
      "epoch": 40.017245817245815,
      "grad_norm": 0.0003951498947571963,
      "learning_rate": 2.030602030602031e-06,
      "loss": 0.0003,
      "step": 31750
    },
    {
      "epoch": 40.01750321750322,
      "grad_norm": 0.00012422552390489727,
      "learning_rate": 2.0277420277420277e-06,
      "loss": 1.2037,
      "step": 31760
    },
    {
      "epoch": 40.017760617760615,
      "grad_norm": 0.019571000710129738,
      "learning_rate": 2.024882024882025e-06,
      "loss": 0.0001,
      "step": 31770
    },
    {
      "epoch": 40.01801801801802,
      "grad_norm": 0.00016542627417948097,
      "learning_rate": 2.022022022022022e-06,
      "loss": 0.0102,
      "step": 31780
    },
    {
      "epoch": 40.018275418275415,
      "grad_norm": 0.029966507107019424,
      "learning_rate": 2.0191620191620192e-06,
      "loss": 0.6281,
      "step": 31790
    },
    {
      "epoch": 40.01853281853282,
      "grad_norm": 0.00015886305482126772,
      "learning_rate": 2.0163020163020166e-06,
      "loss": 0.0002,
      "step": 31800
    },
    {
      "epoch": 40.018790218790215,
      "grad_norm": 0.005060768220573664,
      "learning_rate": 2.0134420134420135e-06,
      "loss": 0.0001,
      "step": 31810
    },
    {
      "epoch": 40.01904761904762,
      "grad_norm": 0.0025004101917147636,
      "learning_rate": 2.0105820105820108e-06,
      "loss": 0.2131,
      "step": 31820
    },
    {
      "epoch": 40.01930501930502,
      "grad_norm": 0.000141930824611336,
      "learning_rate": 2.0077220077220077e-06,
      "loss": 0.5586,
      "step": 31830
    },
    {
      "epoch": 40.01956241956242,
      "grad_norm": 0.1792692095041275,
      "learning_rate": 2.004862004862005e-06,
      "loss": 0.0003,
      "step": 31840
    },
    {
      "epoch": 40.01981981981982,
      "grad_norm": 0.002483024261891842,
      "learning_rate": 2.0020020020020023e-06,
      "loss": 0.0022,
      "step": 31850
    },
    {
      "epoch": 40.02,
      "eval_accuracy": 0.8478260869565217,
      "eval_loss": 1.057834267616272,
      "eval_runtime": 13.3792,
      "eval_samples_per_second": 3.438,
      "eval_steps_per_second": 3.438,
      "step": 31857
    },
    {
      "epoch": 41.00007722007722,
      "grad_norm": 0.0013759952271357179,
      "learning_rate": 1.9991419991419996e-06,
      "loss": 0.7711,
      "step": 31860
    },
    {
      "epoch": 41.00033462033462,
      "grad_norm": 0.0012656886829063296,
      "learning_rate": 1.9962819962819965e-06,
      "loss": 0.0,
      "step": 31870
    },
    {
      "epoch": 41.00059202059202,
      "grad_norm": 0.012028630822896957,
      "learning_rate": 1.993421993421994e-06,
      "loss": 0.8055,
      "step": 31880
    },
    {
      "epoch": 41.00084942084942,
      "grad_norm": 0.0012869435595348477,
      "learning_rate": 1.9905619905619907e-06,
      "loss": 0.0002,
      "step": 31890
    },
    {
      "epoch": 41.00110682110682,
      "grad_norm": 0.0013365877093747258,
      "learning_rate": 1.987701987701988e-06,
      "loss": 0.0001,
      "step": 31900
    },
    {
      "epoch": 41.00136422136422,
      "grad_norm": 0.05966697260737419,
      "learning_rate": 1.984841984841985e-06,
      "loss": 0.0009,
      "step": 31910
    },
    {
      "epoch": 41.001621621621624,
      "grad_norm": 548.8873901367188,
      "learning_rate": 1.9819819819819822e-06,
      "loss": 0.6607,
      "step": 31920
    },
    {
      "epoch": 41.00187902187902,
      "grad_norm": 0.00041702756425365806,
      "learning_rate": 1.979121979121979e-06,
      "loss": 0.5822,
      "step": 31930
    },
    {
      "epoch": 41.002136422136424,
      "grad_norm": 0.0010571606690064073,
      "learning_rate": 1.9762619762619764e-06,
      "loss": 0.0001,
      "step": 31940
    },
    {
      "epoch": 41.00239382239382,
      "grad_norm": 0.0010254288790747523,
      "learning_rate": 1.9734019734019733e-06,
      "loss": 0.0024,
      "step": 31950
    },
    {
      "epoch": 41.002651222651224,
      "grad_norm": 0.0024843502324074507,
      "learning_rate": 1.9705419705419707e-06,
      "loss": 0.0002,
      "step": 31960
    },
    {
      "epoch": 41.00290862290862,
      "grad_norm": 0.0002254961582366377,
      "learning_rate": 1.967681967681968e-06,
      "loss": 1.3888,
      "step": 31970
    },
    {
      "epoch": 41.003166023166024,
      "grad_norm": 0.0001984317641472444,
      "learning_rate": 1.964821964821965e-06,
      "loss": 0.0087,
      "step": 31980
    },
    {
      "epoch": 41.00342342342342,
      "grad_norm": 0.00011000720405718312,
      "learning_rate": 1.961961961961962e-06,
      "loss": 0.0045,
      "step": 31990
    },
    {
      "epoch": 41.003680823680824,
      "grad_norm": 0.015376497060060501,
      "learning_rate": 1.959101959101959e-06,
      "loss": 0.0006,
      "step": 32000
    },
    {
      "epoch": 41.00393822393822,
      "grad_norm": 0.0003337947418913245,
      "learning_rate": 1.9562419562419564e-06,
      "loss": 0.0001,
      "step": 32010
    },
    {
      "epoch": 41.004195624195624,
      "grad_norm": 0.00041247939225286245,
      "learning_rate": 1.9533819533819533e-06,
      "loss": 0.0001,
      "step": 32020
    },
    {
      "epoch": 41.00445302445303,
      "grad_norm": 0.001562226563692093,
      "learning_rate": 1.950521950521951e-06,
      "loss": 0.0003,
      "step": 32030
    },
    {
      "epoch": 41.004710424710424,
      "grad_norm": 0.00014134070079308003,
      "learning_rate": 1.947661947661948e-06,
      "loss": 0.0,
      "step": 32040
    },
    {
      "epoch": 41.00496782496783,
      "grad_norm": 0.00018032017396762967,
      "learning_rate": 1.9448019448019452e-06,
      "loss": 0.0001,
      "step": 32050
    },
    {
      "epoch": 41.005225225225225,
      "grad_norm": 0.009866469539701939,
      "learning_rate": 1.941941941941942e-06,
      "loss": 0.0002,
      "step": 32060
    },
    {
      "epoch": 41.00548262548263,
      "grad_norm": 0.0001611222542123869,
      "learning_rate": 1.9390819390819394e-06,
      "loss": 0.0004,
      "step": 32070
    },
    {
      "epoch": 41.005740025740025,
      "grad_norm": 0.0015964763006195426,
      "learning_rate": 1.9362219362219363e-06,
      "loss": 0.0001,
      "step": 32080
    },
    {
      "epoch": 41.00599742599743,
      "grad_norm": 0.0021054623648524284,
      "learning_rate": 1.9333619333619337e-06,
      "loss": 0.7093,
      "step": 32090
    },
    {
      "epoch": 41.006254826254825,
      "grad_norm": 0.00017562608991283923,
      "learning_rate": 1.9305019305019305e-06,
      "loss": 0.0001,
      "step": 32100
    },
    {
      "epoch": 41.00651222651223,
      "grad_norm": 0.00044666044414043427,
      "learning_rate": 1.927641927641928e-06,
      "loss": 0.0,
      "step": 32110
    },
    {
      "epoch": 41.006769626769625,
      "grad_norm": 0.0006799431866966188,
      "learning_rate": 1.9247819247819248e-06,
      "loss": 0.0001,
      "step": 32120
    },
    {
      "epoch": 41.00702702702703,
      "grad_norm": 0.0002899241226259619,
      "learning_rate": 1.921921921921922e-06,
      "loss": 0.0001,
      "step": 32130
    },
    {
      "epoch": 41.007284427284425,
      "grad_norm": 0.0003254143812227994,
      "learning_rate": 1.9190619190619194e-06,
      "loss": 0.0001,
      "step": 32140
    },
    {
      "epoch": 41.00754182754183,
      "grad_norm": 9.479661093791947e-05,
      "learning_rate": 1.9162019162019163e-06,
      "loss": 0.0,
      "step": 32150
    },
    {
      "epoch": 41.007799227799225,
      "grad_norm": 0.0019586803391575813,
      "learning_rate": 1.9133419133419136e-06,
      "loss": 0.0001,
      "step": 32160
    },
    {
      "epoch": 41.00805662805663,
      "grad_norm": 0.0001963915565283969,
      "learning_rate": 1.9104819104819105e-06,
      "loss": 0.0001,
      "step": 32170
    },
    {
      "epoch": 41.008314028314025,
      "grad_norm": 0.00012030902871629223,
      "learning_rate": 1.907621907621908e-06,
      "loss": 0.0217,
      "step": 32180
    },
    {
      "epoch": 41.00857142857143,
      "grad_norm": 0.00992583204060793,
      "learning_rate": 1.904761904761905e-06,
      "loss": 0.0048,
      "step": 32190
    },
    {
      "epoch": 41.008828828828825,
      "grad_norm": 0.0002358455676585436,
      "learning_rate": 1.901901901901902e-06,
      "loss": 0.8052,
      "step": 32200
    },
    {
      "epoch": 41.00908622908623,
      "grad_norm": 0.0581081286072731,
      "learning_rate": 1.8990418990418991e-06,
      "loss": 0.0001,
      "step": 32210
    },
    {
      "epoch": 41.00934362934363,
      "grad_norm": 2753.812255859375,
      "learning_rate": 1.8961818961818964e-06,
      "loss": 0.975,
      "step": 32220
    },
    {
      "epoch": 41.00960102960103,
      "grad_norm": 0.001163011766038835,
      "learning_rate": 1.8933218933218935e-06,
      "loss": 0.0001,
      "step": 32230
    },
    {
      "epoch": 41.00985842985843,
      "grad_norm": 0.04853722080588341,
      "learning_rate": 1.8904618904618906e-06,
      "loss": 0.5376,
      "step": 32240
    },
    {
      "epoch": 41.01011583011583,
      "grad_norm": 0.00030269287526607513,
      "learning_rate": 1.8876018876018877e-06,
      "loss": 0.0001,
      "step": 32250
    },
    {
      "epoch": 41.01037323037323,
      "grad_norm": 0.00014615173859056085,
      "learning_rate": 1.884741884741885e-06,
      "loss": 0.0001,
      "step": 32260
    },
    {
      "epoch": 41.01063063063063,
      "grad_norm": 0.00015686519327573478,
      "learning_rate": 1.8818818818818822e-06,
      "loss": 0.0001,
      "step": 32270
    },
    {
      "epoch": 41.01088803088803,
      "grad_norm": 0.00044415026786737144,
      "learning_rate": 1.8790218790218793e-06,
      "loss": 0.0001,
      "step": 32280
    },
    {
      "epoch": 41.01114543114543,
      "grad_norm": 0.00014384774840436876,
      "learning_rate": 1.8761618761618764e-06,
      "loss": 0.0001,
      "step": 32290
    },
    {
      "epoch": 41.01140283140283,
      "grad_norm": 0.00019766957848332822,
      "learning_rate": 1.8733018733018735e-06,
      "loss": 0.0008,
      "step": 32300
    },
    {
      "epoch": 41.01166023166023,
      "grad_norm": 0.00011445001291576773,
      "learning_rate": 1.8704418704418706e-06,
      "loss": 0.0002,
      "step": 32310
    },
    {
      "epoch": 41.01191763191763,
      "grad_norm": 0.003054665867239237,
      "learning_rate": 1.8675818675818677e-06,
      "loss": 0.0004,
      "step": 32320
    },
    {
      "epoch": 41.01217503217503,
      "grad_norm": 0.00013076022150926292,
      "learning_rate": 1.8647218647218648e-06,
      "loss": 0.0669,
      "step": 32330
    },
    {
      "epoch": 41.01243243243243,
      "grad_norm": 0.0016582345124334097,
      "learning_rate": 1.861861861861862e-06,
      "loss": 0.9401,
      "step": 32340
    },
    {
      "epoch": 41.01268983268983,
      "grad_norm": 0.001524147461168468,
      "learning_rate": 1.859001859001859e-06,
      "loss": 0.0002,
      "step": 32350
    },
    {
      "epoch": 41.01294723294723,
      "grad_norm": 0.007222165819257498,
      "learning_rate": 1.8561418561418561e-06,
      "loss": 0.0002,
      "step": 32360
    },
    {
      "epoch": 41.01320463320463,
      "grad_norm": 0.00016654320643283427,
      "learning_rate": 1.8532818532818534e-06,
      "loss": 0.0,
      "step": 32370
    },
    {
      "epoch": 41.013462033462034,
      "grad_norm": 0.0013375479029491544,
      "learning_rate": 1.8504218504218505e-06,
      "loss": 0.2057,
      "step": 32380
    },
    {
      "epoch": 41.01371943371943,
      "grad_norm": 0.010129381902515888,
      "learning_rate": 1.8475618475618476e-06,
      "loss": 0.856,
      "step": 32390
    },
    {
      "epoch": 41.013976833976834,
      "grad_norm": 0.00015381610137410462,
      "learning_rate": 1.844701844701845e-06,
      "loss": 0.0002,
      "step": 32400
    },
    {
      "epoch": 41.01423423423424,
      "grad_norm": 0.00014513394853565842,
      "learning_rate": 1.841841841841842e-06,
      "loss": 0.0001,
      "step": 32410
    },
    {
      "epoch": 41.014491634491634,
      "grad_norm": 0.00014010151789989322,
      "learning_rate": 1.8389818389818392e-06,
      "loss": 0.6838,
      "step": 32420
    },
    {
      "epoch": 41.01474903474904,
      "grad_norm": 0.0005243502091616392,
      "learning_rate": 1.8361218361218363e-06,
      "loss": 0.8736,
      "step": 32430
    },
    {
      "epoch": 41.015006435006434,
      "grad_norm": 0.004362246487289667,
      "learning_rate": 1.8332618332618336e-06,
      "loss": 0.144,
      "step": 32440
    },
    {
      "epoch": 41.01526383526384,
      "grad_norm": 0.003542808350175619,
      "learning_rate": 1.8304018304018307e-06,
      "loss": 0.2341,
      "step": 32450
    },
    {
      "epoch": 41.015521235521234,
      "grad_norm": 0.00015553591947536916,
      "learning_rate": 1.8275418275418278e-06,
      "loss": 0.0001,
      "step": 32460
    },
    {
      "epoch": 41.01577863577864,
      "grad_norm": 0.00012182680075056851,
      "learning_rate": 1.8246818246818249e-06,
      "loss": 0.519,
      "step": 32470
    },
    {
      "epoch": 41.016036036036034,
      "grad_norm": 0.03936436027288437,
      "learning_rate": 1.821821821821822e-06,
      "loss": 0.0025,
      "step": 32480
    },
    {
      "epoch": 41.01629343629344,
      "grad_norm": 0.0007767226779833436,
      "learning_rate": 1.818961818961819e-06,
      "loss": 0.0,
      "step": 32490
    },
    {
      "epoch": 41.016550836550834,
      "grad_norm": 0.0001796115393517539,
      "learning_rate": 1.8161018161018162e-06,
      "loss": 0.0001,
      "step": 32500
    },
    {
      "epoch": 41.01680823680824,
      "grad_norm": 88.20077514648438,
      "learning_rate": 1.8132418132418133e-06,
      "loss": 1.2879,
      "step": 32510
    },
    {
      "epoch": 41.017065637065635,
      "grad_norm": 0.006687271874397993,
      "learning_rate": 1.8103818103818104e-06,
      "loss": 0.0003,
      "step": 32520
    },
    {
      "epoch": 41.01732303732304,
      "grad_norm": 0.00014735499280504882,
      "learning_rate": 1.8075218075218075e-06,
      "loss": 0.5289,
      "step": 32530
    },
    {
      "epoch": 41.017580437580435,
      "grad_norm": 0.00012671224249061197,
      "learning_rate": 1.8046618046618048e-06,
      "loss": 0.6025,
      "step": 32540
    },
    {
      "epoch": 41.01783783783784,
      "grad_norm": 0.00012978336599189788,
      "learning_rate": 1.801801801801802e-06,
      "loss": 0.0004,
      "step": 32550
    },
    {
      "epoch": 41.018095238095235,
      "grad_norm": 0.0001393378188367933,
      "learning_rate": 1.798941798941799e-06,
      "loss": 0.7477,
      "step": 32560
    },
    {
      "epoch": 41.01835263835264,
      "grad_norm": 0.00018257521151099354,
      "learning_rate": 1.7960817960817961e-06,
      "loss": 0.7311,
      "step": 32570
    },
    {
      "epoch": 41.01861003861004,
      "grad_norm": 0.00029136036755517125,
      "learning_rate": 1.7932217932217933e-06,
      "loss": 0.0002,
      "step": 32580
    },
    {
      "epoch": 41.01886743886744,
      "grad_norm": 1.4105013608932495,
      "learning_rate": 1.7903617903617906e-06,
      "loss": 0.0005,
      "step": 32590
    },
    {
      "epoch": 41.01912483912484,
      "grad_norm": 0.00852140597999096,
      "learning_rate": 1.7875017875017877e-06,
      "loss": 0.2117,
      "step": 32600
    },
    {
      "epoch": 41.01938223938224,
      "grad_norm": 0.004909228533506393,
      "learning_rate": 1.784641784641785e-06,
      "loss": 0.0001,
      "step": 32610
    },
    {
      "epoch": 41.01963963963964,
      "grad_norm": 0.00011149414058309048,
      "learning_rate": 1.781781781781782e-06,
      "loss": 0.0001,
      "step": 32620
    },
    {
      "epoch": 41.01989703989704,
      "grad_norm": 0.00012031041114823893,
      "learning_rate": 1.7789217789217792e-06,
      "loss": 0.0002,
      "step": 32630
    },
    {
      "epoch": 41.02,
      "eval_accuracy": 0.8695652173913043,
      "eval_loss": 0.9567879438400269,
      "eval_runtime": 13.3891,
      "eval_samples_per_second": 3.436,
      "eval_steps_per_second": 3.436,
      "step": 32634
    },
    {
      "epoch": 42.00015444015444,
      "grad_norm": 0.00020218401914462447,
      "learning_rate": 1.7760617760617763e-06,
      "loss": 0.0728,
      "step": 32640
    },
    {
      "epoch": 42.00041184041184,
      "grad_norm": 0.00011965364683419466,
      "learning_rate": 1.7732017732017734e-06,
      "loss": 0.0005,
      "step": 32650
    },
    {
      "epoch": 42.00066924066924,
      "grad_norm": 0.00013755726104136556,
      "learning_rate": 1.7703417703417705e-06,
      "loss": 0.0001,
      "step": 32660
    },
    {
      "epoch": 42.00092664092664,
      "grad_norm": 0.003897932590916753,
      "learning_rate": 1.7674817674817676e-06,
      "loss": 0.5112,
      "step": 32670
    },
    {
      "epoch": 42.00118404118404,
      "grad_norm": 0.048069994896650314,
      "learning_rate": 1.7646217646217647e-06,
      "loss": 0.0002,
      "step": 32680
    },
    {
      "epoch": 42.00144144144144,
      "grad_norm": 0.00015302524843718857,
      "learning_rate": 1.7617617617617618e-06,
      "loss": 0.0002,
      "step": 32690
    },
    {
      "epoch": 42.00169884169884,
      "grad_norm": 0.0002129245549440384,
      "learning_rate": 1.758901758901759e-06,
      "loss": 0.0001,
      "step": 32700
    },
    {
      "epoch": 42.00195624195624,
      "grad_norm": 0.004387426655739546,
      "learning_rate": 1.756041756041756e-06,
      "loss": 0.0001,
      "step": 32710
    },
    {
      "epoch": 42.00221364221364,
      "grad_norm": 0.01488508377224207,
      "learning_rate": 1.7531817531817534e-06,
      "loss": 0.0001,
      "step": 32720
    },
    {
      "epoch": 42.00247104247104,
      "grad_norm": 0.00032001244835555553,
      "learning_rate": 1.7503217503217505e-06,
      "loss": 0.005,
      "step": 32730
    },
    {
      "epoch": 42.00272844272844,
      "grad_norm": 0.002208408433943987,
      "learning_rate": 1.7474617474617476e-06,
      "loss": 0.5717,
      "step": 32740
    },
    {
      "epoch": 42.002985842985844,
      "grad_norm": 0.00012605212396010756,
      "learning_rate": 1.7446017446017447e-06,
      "loss": 0.0,
      "step": 32750
    },
    {
      "epoch": 42.00324324324324,
      "grad_norm": 0.0014979569241404533,
      "learning_rate": 1.7417417417417418e-06,
      "loss": 0.0,
      "step": 32760
    },
    {
      "epoch": 42.003500643500644,
      "grad_norm": 0.003774981014430523,
      "learning_rate": 1.7388817388817389e-06,
      "loss": 0.0001,
      "step": 32770
    },
    {
      "epoch": 42.00375804375804,
      "grad_norm": 0.0028361384756863117,
      "learning_rate": 1.7360217360217362e-06,
      "loss": 0.7856,
      "step": 32780
    },
    {
      "epoch": 42.004015444015444,
      "grad_norm": 0.0033012754283845425,
      "learning_rate": 1.7331617331617335e-06,
      "loss": 0.0003,
      "step": 32790
    },
    {
      "epoch": 42.00427284427285,
      "grad_norm": 0.009487603791058064,
      "learning_rate": 1.7303017303017306e-06,
      "loss": 0.0003,
      "step": 32800
    },
    {
      "epoch": 42.004530244530244,
      "grad_norm": 0.00019623312982730567,
      "learning_rate": 1.7274417274417277e-06,
      "loss": 0.0001,
      "step": 32810
    },
    {
      "epoch": 42.00478764478765,
      "grad_norm": 0.00011614287359407172,
      "learning_rate": 1.7245817245817248e-06,
      "loss": 0.6143,
      "step": 32820
    },
    {
      "epoch": 42.005045045045044,
      "grad_norm": 0.0002354454481974244,
      "learning_rate": 1.721721721721722e-06,
      "loss": 0.6405,
      "step": 32830
    },
    {
      "epoch": 42.00530244530245,
      "grad_norm": 0.03378787264227867,
      "learning_rate": 1.718861718861719e-06,
      "loss": 0.0001,
      "step": 32840
    },
    {
      "epoch": 42.005559845559844,
      "grad_norm": 0.00012451726070139557,
      "learning_rate": 1.7160017160017161e-06,
      "loss": 0.0002,
      "step": 32850
    },
    {
      "epoch": 42.00581724581725,
      "grad_norm": 0.0002813052851706743,
      "learning_rate": 1.7131417131417132e-06,
      "loss": 0.0002,
      "step": 32860
    },
    {
      "epoch": 42.006074646074644,
      "grad_norm": 0.0054261996410787106,
      "learning_rate": 1.7102817102817103e-06,
      "loss": 0.0001,
      "step": 32870
    },
    {
      "epoch": 42.00633204633205,
      "grad_norm": 0.00012816753587685525,
      "learning_rate": 1.7074217074217074e-06,
      "loss": 0.0002,
      "step": 32880
    },
    {
      "epoch": 42.006589446589444,
      "grad_norm": 0.0003143631329294294,
      "learning_rate": 1.7045617045617048e-06,
      "loss": 0.0042,
      "step": 32890
    },
    {
      "epoch": 42.00684684684685,
      "grad_norm": 0.00046908739022910595,
      "learning_rate": 1.7017017017017019e-06,
      "loss": 0.781,
      "step": 32900
    },
    {
      "epoch": 42.007104247104245,
      "grad_norm": 0.03740377351641655,
      "learning_rate": 1.698841698841699e-06,
      "loss": 0.0001,
      "step": 32910
    },
    {
      "epoch": 42.00736164736165,
      "grad_norm": 0.0009596379823051393,
      "learning_rate": 1.695981695981696e-06,
      "loss": 0.0001,
      "step": 32920
    },
    {
      "epoch": 42.007619047619045,
      "grad_norm": 0.012743496336042881,
      "learning_rate": 1.6931216931216932e-06,
      "loss": 0.0001,
      "step": 32930
    },
    {
      "epoch": 42.00787644787645,
      "grad_norm": 0.002485014731064439,
      "learning_rate": 1.6902616902616903e-06,
      "loss": 0.6629,
      "step": 32940
    },
    {
      "epoch": 42.008133848133845,
      "grad_norm": 0.0100320503115654,
      "learning_rate": 1.6874016874016874e-06,
      "loss": 0.0003,
      "step": 32950
    },
    {
      "epoch": 42.00839124839125,
      "grad_norm": 0.010429958812892437,
      "learning_rate": 1.6845416845416845e-06,
      "loss": 0.0003,
      "step": 32960
    },
    {
      "epoch": 42.00864864864865,
      "grad_norm": 0.01115471962839365,
      "learning_rate": 1.681681681681682e-06,
      "loss": 0.0003,
      "step": 32970
    },
    {
      "epoch": 42.00890604890605,
      "grad_norm": 144.6180877685547,
      "learning_rate": 1.6788216788216791e-06,
      "loss": 0.7403,
      "step": 32980
    },
    {
      "epoch": 42.00916344916345,
      "grad_norm": 1396.6766357421875,
      "learning_rate": 1.6759616759616762e-06,
      "loss": 0.6216,
      "step": 32990
    },
    {
      "epoch": 42.00942084942085,
      "grad_norm": 0.017528045922517776,
      "learning_rate": 1.6731016731016733e-06,
      "loss": 0.0001,
      "step": 33000
    },
    {
      "epoch": 42.00967824967825,
      "grad_norm": 0.0125763900578022,
      "learning_rate": 1.6702416702416704e-06,
      "loss": 0.0001,
      "step": 33010
    },
    {
      "epoch": 42.00993564993565,
      "grad_norm": 0.0016611794708296657,
      "learning_rate": 1.6673816673816675e-06,
      "loss": 0.0,
      "step": 33020
    },
    {
      "epoch": 42.01019305019305,
      "grad_norm": 0.0032114211935549974,
      "learning_rate": 1.6645216645216646e-06,
      "loss": 0.808,
      "step": 33030
    },
    {
      "epoch": 42.01045045045045,
      "grad_norm": 0.04180116951465607,
      "learning_rate": 1.6616616616616618e-06,
      "loss": 0.0006,
      "step": 33040
    },
    {
      "epoch": 42.01070785070785,
      "grad_norm": 0.32980695366859436,
      "learning_rate": 1.6588016588016589e-06,
      "loss": 0.0002,
      "step": 33050
    },
    {
      "epoch": 42.01096525096525,
      "grad_norm": 0.008240937255322933,
      "learning_rate": 1.655941655941656e-06,
      "loss": 0.6066,
      "step": 33060
    },
    {
      "epoch": 42.01122265122265,
      "grad_norm": 0.00017696853319648653,
      "learning_rate": 1.6530816530816533e-06,
      "loss": 0.0,
      "step": 33070
    },
    {
      "epoch": 42.01148005148005,
      "grad_norm": 0.00016973470337688923,
      "learning_rate": 1.6502216502216504e-06,
      "loss": 0.0001,
      "step": 33080
    },
    {
      "epoch": 42.01173745173745,
      "grad_norm": 0.0002044615539489314,
      "learning_rate": 1.6473616473616475e-06,
      "loss": 0.0001,
      "step": 33090
    },
    {
      "epoch": 42.01199485199485,
      "grad_norm": 0.008063897490501404,
      "learning_rate": 1.6445016445016446e-06,
      "loss": 0.0001,
      "step": 33100
    },
    {
      "epoch": 42.01225225225225,
      "grad_norm": 0.00011532429925864562,
      "learning_rate": 1.6416416416416417e-06,
      "loss": 0.0001,
      "step": 33110
    },
    {
      "epoch": 42.01250965250965,
      "grad_norm": 0.054353538900613785,
      "learning_rate": 1.6387816387816388e-06,
      "loss": 0.0002,
      "step": 33120
    },
    {
      "epoch": 42.01276705276705,
      "grad_norm": 0.00019760456052608788,
      "learning_rate": 1.635921635921636e-06,
      "loss": 0.0001,
      "step": 33130
    },
    {
      "epoch": 42.01302445302445,
      "grad_norm": 0.052227336913347244,
      "learning_rate": 1.633061633061633e-06,
      "loss": 0.0002,
      "step": 33140
    },
    {
      "epoch": 42.01328185328185,
      "grad_norm": 0.08783034235239029,
      "learning_rate": 1.6302016302016305e-06,
      "loss": 0.4019,
      "step": 33150
    },
    {
      "epoch": 42.01353925353926,
      "grad_norm": 0.007176580838859081,
      "learning_rate": 1.6273416273416276e-06,
      "loss": 0.0001,
      "step": 33160
    },
    {
      "epoch": 42.01379665379665,
      "grad_norm": 0.14134147763252258,
      "learning_rate": 1.6244816244816247e-06,
      "loss": 0.0002,
      "step": 33170
    },
    {
      "epoch": 42.01405405405406,
      "grad_norm": 0.004731915425509214,
      "learning_rate": 1.6216216216216219e-06,
      "loss": 0.0001,
      "step": 33180
    },
    {
      "epoch": 42.01431145431145,
      "grad_norm": 0.0035558564122766256,
      "learning_rate": 1.618761618761619e-06,
      "loss": 0.7162,
      "step": 33190
    },
    {
      "epoch": 42.01456885456886,
      "grad_norm": 0.015174226835370064,
      "learning_rate": 1.615901615901616e-06,
      "loss": 0.0001,
      "step": 33200
    },
    {
      "epoch": 42.014826254826254,
      "grad_norm": 0.010268859565258026,
      "learning_rate": 1.6130416130416132e-06,
      "loss": 0.0002,
      "step": 33210
    },
    {
      "epoch": 42.01508365508366,
      "grad_norm": 0.00010535326146055013,
      "learning_rate": 1.6101816101816103e-06,
      "loss": 0.0005,
      "step": 33220
    },
    {
      "epoch": 42.015341055341054,
      "grad_norm": 0.00044961340609006584,
      "learning_rate": 1.6073216073216074e-06,
      "loss": 0.0001,
      "step": 33230
    },
    {
      "epoch": 42.01559845559846,
      "grad_norm": 0.00024218950420618057,
      "learning_rate": 1.6044616044616047e-06,
      "loss": 0.0001,
      "step": 33240
    },
    {
      "epoch": 42.015855855855854,
      "grad_norm": 0.013476267457008362,
      "learning_rate": 1.6016016016016018e-06,
      "loss": 0.0022,
      "step": 33250
    },
    {
      "epoch": 42.01611325611326,
      "grad_norm": 0.01417740061879158,
      "learning_rate": 1.598741598741599e-06,
      "loss": 0.0005,
      "step": 33260
    },
    {
      "epoch": 42.016370656370654,
      "grad_norm": 0.00013865696382708848,
      "learning_rate": 1.595881595881596e-06,
      "loss": 0.0001,
      "step": 33270
    },
    {
      "epoch": 42.01662805662806,
      "grad_norm": 0.009123004972934723,
      "learning_rate": 1.5930215930215931e-06,
      "loss": 0.0001,
      "step": 33280
    },
    {
      "epoch": 42.016885456885454,
      "grad_norm": 0.0426633283495903,
      "learning_rate": 1.5901615901615902e-06,
      "loss": 0.0001,
      "step": 33290
    },
    {
      "epoch": 42.01714285714286,
      "grad_norm": 0.00010858638415811583,
      "learning_rate": 1.5873015873015873e-06,
      "loss": 0.0001,
      "step": 33300
    },
    {
      "epoch": 42.017400257400254,
      "grad_norm": 0.01740678958594799,
      "learning_rate": 1.5844415844415844e-06,
      "loss": 0.6447,
      "step": 33310
    },
    {
      "epoch": 42.01765765765766,
      "grad_norm": 0.0025633275508880615,
      "learning_rate": 1.5815815815815815e-06,
      "loss": 0.738,
      "step": 33320
    },
    {
      "epoch": 42.017915057915054,
      "grad_norm": 0.06856297701597214,
      "learning_rate": 1.5787215787215786e-06,
      "loss": 0.0003,
      "step": 33330
    },
    {
      "epoch": 42.01817245817246,
      "grad_norm": 0.006150628440082073,
      "learning_rate": 1.5758615758615762e-06,
      "loss": 0.0001,
      "step": 33340
    },
    {
      "epoch": 42.01842985842986,
      "grad_norm": 0.0021094612311571836,
      "learning_rate": 1.5730015730015733e-06,
      "loss": 0.3257,
      "step": 33350
    },
    {
      "epoch": 42.01868725868726,
      "grad_norm": 0.003045695135369897,
      "learning_rate": 1.5701415701415704e-06,
      "loss": 0.0001,
      "step": 33360
    },
    {
      "epoch": 42.01894465894466,
      "grad_norm": 0.002286852104589343,
      "learning_rate": 1.5672815672815675e-06,
      "loss": 0.6679,
      "step": 33370
    },
    {
      "epoch": 42.01920205920206,
      "grad_norm": 0.002063462510704994,
      "learning_rate": 1.5644215644215646e-06,
      "loss": 0.0001,
      "step": 33380
    },
    {
      "epoch": 42.01945945945946,
      "grad_norm": 0.00013045952073298395,
      "learning_rate": 1.5615615615615617e-06,
      "loss": 0.6961,
      "step": 33390
    },
    {
      "epoch": 42.01971685971686,
      "grad_norm": 0.0018174780998378992,
      "learning_rate": 1.5587015587015588e-06,
      "loss": 0.1587,
      "step": 33400
    },
    {
      "epoch": 42.01997425997426,
      "grad_norm": 0.0012939365115016699,
      "learning_rate": 1.5558415558415559e-06,
      "loss": 0.0004,
      "step": 33410
    },
    {
      "epoch": 42.02,
      "eval_accuracy": 0.8695652173913043,
      "eval_loss": 0.9224266409873962,
      "eval_runtime": 13.4044,
      "eval_samples_per_second": 3.432,
      "eval_steps_per_second": 3.432,
      "step": 33411
    },
    {
      "epoch": 43.00023166023166,
      "grad_norm": 0.01454122643917799,
      "learning_rate": 1.5529815529815532e-06,
      "loss": 0.0002,
      "step": 33420
    },
    {
      "epoch": 43.00048906048906,
      "grad_norm": 0.011734708212316036,
      "learning_rate": 1.5501215501215503e-06,
      "loss": 0.0004,
      "step": 33430
    },
    {
      "epoch": 43.00074646074646,
      "grad_norm": 0.0002132106601493433,
      "learning_rate": 1.5472615472615474e-06,
      "loss": 0.6844,
      "step": 33440
    },
    {
      "epoch": 43.00100386100386,
      "grad_norm": 0.0011723071802407503,
      "learning_rate": 1.5444015444015445e-06,
      "loss": 0.0023,
      "step": 33450
    },
    {
      "epoch": 43.00126126126126,
      "grad_norm": 0.006746041122823954,
      "learning_rate": 1.5415415415415416e-06,
      "loss": 0.0006,
      "step": 33460
    },
    {
      "epoch": 43.00151866151866,
      "grad_norm": 0.0007373033440671861,
      "learning_rate": 1.5386815386815387e-06,
      "loss": 1.6536,
      "step": 33470
    },
    {
      "epoch": 43.00177606177606,
      "grad_norm": 0.00580536900088191,
      "learning_rate": 1.5358215358215358e-06,
      "loss": 0.0003,
      "step": 33480
    },
    {
      "epoch": 43.00203346203346,
      "grad_norm": 0.00019677856471389532,
      "learning_rate": 1.532961532961533e-06,
      "loss": 0.0002,
      "step": 33490
    },
    {
      "epoch": 43.00229086229086,
      "grad_norm": 0.006432530004531145,
      "learning_rate": 1.53010153010153e-06,
      "loss": 0.7492,
      "step": 33500
    },
    {
      "epoch": 43.00254826254826,
      "grad_norm": 0.003972894046455622,
      "learning_rate": 1.5272415272415271e-06,
      "loss": 0.0002,
      "step": 33510
    },
    {
      "epoch": 43.00280566280566,
      "grad_norm": 0.004245332442224026,
      "learning_rate": 1.5243815243815245e-06,
      "loss": 0.0001,
      "step": 33520
    },
    {
      "epoch": 43.00306306306306,
      "grad_norm": 0.06526504456996918,
      "learning_rate": 1.5215215215215218e-06,
      "loss": 0.0001,
      "step": 33530
    },
    {
      "epoch": 43.00332046332046,
      "grad_norm": 0.004649861715734005,
      "learning_rate": 1.5186615186615189e-06,
      "loss": 0.0002,
      "step": 33540
    },
    {
      "epoch": 43.00357786357787,
      "grad_norm": 9.96380767901428e-05,
      "learning_rate": 1.515801515801516e-06,
      "loss": 0.0005,
      "step": 33550
    },
    {
      "epoch": 43.00383526383526,
      "grad_norm": 0.005374431610107422,
      "learning_rate": 1.512941512941513e-06,
      "loss": 0.0002,
      "step": 33560
    },
    {
      "epoch": 43.00409266409267,
      "grad_norm": 0.00018989373347721994,
      "learning_rate": 1.5100815100815102e-06,
      "loss": 0.0,
      "step": 33570
    },
    {
      "epoch": 43.004350064350064,
      "grad_norm": 0.008197961375117302,
      "learning_rate": 1.5072215072215073e-06,
      "loss": 0.0001,
      "step": 33580
    },
    {
      "epoch": 43.00460746460747,
      "grad_norm": 0.012032076716423035,
      "learning_rate": 1.5043615043615046e-06,
      "loss": 0.0001,
      "step": 33590
    },
    {
      "epoch": 43.004864864864864,
      "grad_norm": 0.00011511553020682186,
      "learning_rate": 1.5015015015015017e-06,
      "loss": 0.0005,
      "step": 33600
    },
    {
      "epoch": 43.00512226512227,
      "grad_norm": 0.00012108176451874897,
      "learning_rate": 1.4986414986414988e-06,
      "loss": 0.0001,
      "step": 33610
    },
    {
      "epoch": 43.005379665379664,
      "grad_norm": 0.00024196848971769214,
      "learning_rate": 1.495781495781496e-06,
      "loss": 0.1988,
      "step": 33620
    },
    {
      "epoch": 43.00563706563707,
      "grad_norm": 9.511480311630294e-05,
      "learning_rate": 1.492921492921493e-06,
      "loss": 0.0001,
      "step": 33630
    },
    {
      "epoch": 43.005894465894464,
      "grad_norm": 0.0001018472685245797,
      "learning_rate": 1.4900614900614901e-06,
      "loss": 0.34,
      "step": 33640
    },
    {
      "epoch": 43.00615186615187,
      "grad_norm": 217.52088928222656,
      "learning_rate": 1.4872014872014872e-06,
      "loss": 0.5665,
      "step": 33650
    },
    {
      "epoch": 43.006409266409264,
      "grad_norm": 0.02129535749554634,
      "learning_rate": 1.4843414843414843e-06,
      "loss": 0.0062,
      "step": 33660
    },
    {
      "epoch": 43.00666666666667,
      "grad_norm": 0.049556683748960495,
      "learning_rate": 1.4814814814814815e-06,
      "loss": 0.0007,
      "step": 33670
    },
    {
      "epoch": 43.006924066924064,
      "grad_norm": 0.0052298568189144135,
      "learning_rate": 1.4786214786214786e-06,
      "loss": 0.0003,
      "step": 33680
    },
    {
      "epoch": 43.00718146718147,
      "grad_norm": 0.041593704372644424,
      "learning_rate": 1.4757614757614759e-06,
      "loss": 0.0003,
      "step": 33690
    },
    {
      "epoch": 43.007438867438864,
      "grad_norm": 0.00040065948269329965,
      "learning_rate": 1.472901472901473e-06,
      "loss": 0.3137,
      "step": 33700
    },
    {
      "epoch": 43.00769626769627,
      "grad_norm": 0.030675746500492096,
      "learning_rate": 1.4700414700414703e-06,
      "loss": 0.0003,
      "step": 33710
    },
    {
      "epoch": 43.007953667953664,
      "grad_norm": 0.0022851689718663692,
      "learning_rate": 1.4671814671814674e-06,
      "loss": 0.0,
      "step": 33720
    },
    {
      "epoch": 43.00821106821107,
      "grad_norm": 101.7066650390625,
      "learning_rate": 1.4643214643214645e-06,
      "loss": 1.5302,
      "step": 33730
    },
    {
      "epoch": 43.00846846846847,
      "grad_norm": 0.00015403941506519914,
      "learning_rate": 1.4614614614614616e-06,
      "loss": 0.0001,
      "step": 33740
    },
    {
      "epoch": 43.00872586872587,
      "grad_norm": 0.0009902218589559197,
      "learning_rate": 1.4586014586014587e-06,
      "loss": 0.0001,
      "step": 33750
    },
    {
      "epoch": 43.00898326898327,
      "grad_norm": 0.0025801286101341248,
      "learning_rate": 1.4557414557414558e-06,
      "loss": 0.0002,
      "step": 33760
    },
    {
      "epoch": 43.00924066924067,
      "grad_norm": 0.00010533163003856316,
      "learning_rate": 1.4528814528814531e-06,
      "loss": 0.0003,
      "step": 33770
    },
    {
      "epoch": 43.00949806949807,
      "grad_norm": 39.131736755371094,
      "learning_rate": 1.4500214500214502e-06,
      "loss": 0.0072,
      "step": 33780
    },
    {
      "epoch": 43.00975546975547,
      "grad_norm": 0.00011560120765352622,
      "learning_rate": 1.4471614471614473e-06,
      "loss": 0.0001,
      "step": 33790
    },
    {
      "epoch": 43.01001287001287,
      "grad_norm": 0.014448067173361778,
      "learning_rate": 1.4443014443014444e-06,
      "loss": 0.0001,
      "step": 33800
    },
    {
      "epoch": 43.01027027027027,
      "grad_norm": 0.05728612467646599,
      "learning_rate": 1.4414414414414416e-06,
      "loss": 0.0005,
      "step": 33810
    },
    {
      "epoch": 43.01052767052767,
      "grad_norm": 0.0005739001790061593,
      "learning_rate": 1.4385814385814387e-06,
      "loss": 0.0001,
      "step": 33820
    },
    {
      "epoch": 43.01078507078507,
      "grad_norm": 0.00013156767818145454,
      "learning_rate": 1.4357214357214358e-06,
      "loss": 0.0001,
      "step": 33830
    },
    {
      "epoch": 43.01104247104247,
      "grad_norm": 0.0010646074078977108,
      "learning_rate": 1.4328614328614329e-06,
      "loss": 0.0001,
      "step": 33840
    },
    {
      "epoch": 43.01129987129987,
      "grad_norm": 0.00016523963131476194,
      "learning_rate": 1.43000143000143e-06,
      "loss": 0.0001,
      "step": 33850
    },
    {
      "epoch": 43.01155727155727,
      "grad_norm": 0.01101844571530819,
      "learning_rate": 1.427141427141427e-06,
      "loss": 1.569,
      "step": 33860
    },
    {
      "epoch": 43.01181467181467,
      "grad_norm": 9.992035484174266e-05,
      "learning_rate": 1.4242814242814244e-06,
      "loss": 0.0001,
      "step": 33870
    },
    {
      "epoch": 43.01207207207207,
      "grad_norm": 0.0001401646004524082,
      "learning_rate": 1.4214214214214215e-06,
      "loss": 0.0004,
      "step": 33880
    },
    {
      "epoch": 43.01232947232947,
      "grad_norm": 0.027710143476724625,
      "learning_rate": 1.4185614185614186e-06,
      "loss": 0.0002,
      "step": 33890
    },
    {
      "epoch": 43.01258687258687,
      "grad_norm": 0.002572421682998538,
      "learning_rate": 1.415701415701416e-06,
      "loss": 0.002,
      "step": 33900
    },
    {
      "epoch": 43.012844272844276,
      "grad_norm": 0.002672908827662468,
      "learning_rate": 1.412841412841413e-06,
      "loss": 0.4745,
      "step": 33910
    },
    {
      "epoch": 43.01310167310167,
      "grad_norm": 0.00011232582619413733,
      "learning_rate": 1.4099814099814101e-06,
      "loss": 0.4356,
      "step": 33920
    },
    {
      "epoch": 43.01335907335908,
      "grad_norm": 0.010794050060212612,
      "learning_rate": 1.4071214071214072e-06,
      "loss": 0.0004,
      "step": 33930
    },
    {
      "epoch": 43.01361647361647,
      "grad_norm": 0.021403295919299126,
      "learning_rate": 1.4042614042614045e-06,
      "loss": 0.0001,
      "step": 33940
    },
    {
      "epoch": 43.01387387387388,
      "grad_norm": 0.003955175634473562,
      "learning_rate": 1.4014014014014016e-06,
      "loss": 0.0003,
      "step": 33950
    },
    {
      "epoch": 43.01413127413127,
      "grad_norm": 0.0001854293659562245,
      "learning_rate": 1.3985413985413988e-06,
      "loss": 0.0001,
      "step": 33960
    },
    {
      "epoch": 43.01438867438868,
      "grad_norm": 0.004186289384961128,
      "learning_rate": 1.3956813956813959e-06,
      "loss": 0.0,
      "step": 33970
    },
    {
      "epoch": 43.01464607464607,
      "grad_norm": 0.003534914692863822,
      "learning_rate": 1.392821392821393e-06,
      "loss": 0.0002,
      "step": 33980
    },
    {
      "epoch": 43.01490347490348,
      "grad_norm": 0.004942872561514378,
      "learning_rate": 1.38996138996139e-06,
      "loss": 0.0001,
      "step": 33990
    },
    {
      "epoch": 43.01516087516087,
      "grad_norm": 0.020380662754178047,
      "learning_rate": 1.3871013871013872e-06,
      "loss": 0.6943,
      "step": 34000
    },
    {
      "epoch": 43.01541827541828,
      "grad_norm": 0.013380438089370728,
      "learning_rate": 1.3842413842413843e-06,
      "loss": 0.0002,
      "step": 34010
    },
    {
      "epoch": 43.01567567567567,
      "grad_norm": 0.00011852156603708863,
      "learning_rate": 1.3813813813813814e-06,
      "loss": 0.0001,
      "step": 34020
    },
    {
      "epoch": 43.01593307593308,
      "grad_norm": 0.02771720103919506,
      "learning_rate": 1.3785213785213785e-06,
      "loss": 0.0558,
      "step": 34030
    },
    {
      "epoch": 43.016190476190474,
      "grad_norm": 0.004810509271919727,
      "learning_rate": 1.3756613756613758e-06,
      "loss": 0.0002,
      "step": 34040
    },
    {
      "epoch": 43.01644787644788,
      "grad_norm": 0.001095824409276247,
      "learning_rate": 1.372801372801373e-06,
      "loss": 0.0982,
      "step": 34050
    },
    {
      "epoch": 43.016705276705274,
      "grad_norm": 0.028301896527409554,
      "learning_rate": 1.36994136994137e-06,
      "loss": 0.0057,
      "step": 34060
    },
    {
      "epoch": 43.01696267696268,
      "grad_norm": 8.951701602200046e-05,
      "learning_rate": 1.3670813670813671e-06,
      "loss": 0.0,
      "step": 34070
    },
    {
      "epoch": 43.017220077220074,
      "grad_norm": 0.00012243095261510462,
      "learning_rate": 1.3642213642213642e-06,
      "loss": 0.0019,
      "step": 34080
    },
    {
      "epoch": 43.01747747747748,
      "grad_norm": 0.08369873464107513,
      "learning_rate": 1.3613613613613615e-06,
      "loss": 0.0001,
      "step": 34090
    },
    {
      "epoch": 43.01773487773488,
      "grad_norm": 0.12311464548110962,
      "learning_rate": 1.3585013585013586e-06,
      "loss": 0.0035,
      "step": 34100
    },
    {
      "epoch": 43.01799227799228,
      "grad_norm": 0.0011633503017947078,
      "learning_rate": 1.3556413556413557e-06,
      "loss": 0.0001,
      "step": 34110
    },
    {
      "epoch": 43.01824967824968,
      "grad_norm": 0.2869292199611664,
      "learning_rate": 1.352781352781353e-06,
      "loss": 0.6535,
      "step": 34120
    },
    {
      "epoch": 43.01850707850708,
      "grad_norm": 0.0009136784356087446,
      "learning_rate": 1.3499213499213502e-06,
      "loss": 0.0001,
      "step": 34130
    },
    {
      "epoch": 43.01876447876448,
      "grad_norm": 0.017915774136781693,
      "learning_rate": 1.3470613470613473e-06,
      "loss": 0.7692,
      "step": 34140
    },
    {
      "epoch": 43.01902187902188,
      "grad_norm": 0.013292906805872917,
      "learning_rate": 1.3442013442013444e-06,
      "loss": 0.1451,
      "step": 34150
    },
    {
      "epoch": 43.01927927927928,
      "grad_norm": 0.49672406911849976,
      "learning_rate": 1.3413413413413415e-06,
      "loss": 0.0002,
      "step": 34160
    },
    {
      "epoch": 43.01953667953668,
      "grad_norm": 0.018338019028306007,
      "learning_rate": 1.3384813384813386e-06,
      "loss": 0.3841,
      "step": 34170
    },
    {
      "epoch": 43.01979407979408,
      "grad_norm": 0.0016882263589650393,
      "learning_rate": 1.3356213356213357e-06,
      "loss": 0.0,
      "step": 34180
    },
    {
      "epoch": 43.02,
      "eval_accuracy": 0.8913043478260869,
      "eval_loss": 0.8144087791442871,
      "eval_runtime": 13.4412,
      "eval_samples_per_second": 3.422,
      "eval_steps_per_second": 3.422,
      "step": 34188
    },
    {
      "epoch": 44.00005148005148,
      "grad_norm": 0.00010920155182247981,
      "learning_rate": 1.3327613327613328e-06,
      "loss": 0.0001,
      "step": 34190
    },
    {
      "epoch": 44.00030888030888,
      "grad_norm": 0.00023170172062236816,
      "learning_rate": 1.32990132990133e-06,
      "loss": 0.0001,
      "step": 34200
    },
    {
      "epoch": 44.00056628056628,
      "grad_norm": 0.00012434182281140238,
      "learning_rate": 1.327041327041327e-06,
      "loss": 0.0,
      "step": 34210
    },
    {
      "epoch": 44.00082368082368,
      "grad_norm": 0.0024254657328128815,
      "learning_rate": 1.3241813241813243e-06,
      "loss": 0.0001,
      "step": 34220
    },
    {
      "epoch": 44.00108108108108,
      "grad_norm": 0.001370239071547985,
      "learning_rate": 1.3213213213213214e-06,
      "loss": 0.0002,
      "step": 34230
    },
    {
      "epoch": 44.00133848133848,
      "grad_norm": 0.0040776995010674,
      "learning_rate": 1.3184613184613185e-06,
      "loss": 0.0001,
      "step": 34240
    },
    {
      "epoch": 44.00159588159588,
      "grad_norm": 0.0010753278620541096,
      "learning_rate": 1.3156013156013156e-06,
      "loss": 0.0021,
      "step": 34250
    },
    {
      "epoch": 44.00185328185328,
      "grad_norm": 0.00011391542648198083,
      "learning_rate": 1.3127413127413127e-06,
      "loss": 0.0004,
      "step": 34260
    },
    {
      "epoch": 44.00211068211068,
      "grad_norm": 0.00021202664356678724,
      "learning_rate": 1.30988130988131e-06,
      "loss": 0.3768,
      "step": 34270
    },
    {
      "epoch": 44.00236808236808,
      "grad_norm": 0.0013385345228016376,
      "learning_rate": 1.3070213070213072e-06,
      "loss": 0.0,
      "step": 34280
    },
    {
      "epoch": 44.00262548262548,
      "grad_norm": 9.8246535344515e-05,
      "learning_rate": 1.3041613041613045e-06,
      "loss": 0.0001,
      "step": 34290
    },
    {
      "epoch": 44.002882882882886,
      "grad_norm": 0.00012314043124206364,
      "learning_rate": 1.3013013013013016e-06,
      "loss": 0.0,
      "step": 34300
    },
    {
      "epoch": 44.00314028314028,
      "grad_norm": 0.00023305923969019204,
      "learning_rate": 1.2984412984412987e-06,
      "loss": 0.0001,
      "step": 34310
    },
    {
      "epoch": 44.00339768339769,
      "grad_norm": 0.006541033275425434,
      "learning_rate": 1.2955812955812958e-06,
      "loss": 0.0002,
      "step": 34320
    },
    {
      "epoch": 44.00365508365508,
      "grad_norm": 0.008542906492948532,
      "learning_rate": 1.2927212927212929e-06,
      "loss": 0.0002,
      "step": 34330
    },
    {
      "epoch": 44.00391248391249,
      "grad_norm": 0.00016316522669512779,
      "learning_rate": 1.28986128986129e-06,
      "loss": 0.5992,
      "step": 34340
    },
    {
      "epoch": 44.00416988416988,
      "grad_norm": 1539.5146484375,
      "learning_rate": 1.287001287001287e-06,
      "loss": 0.1944,
      "step": 34350
    },
    {
      "epoch": 44.00442728442729,
      "grad_norm": 0.0014140707207843661,
      "learning_rate": 1.2841412841412842e-06,
      "loss": 0.0,
      "step": 34360
    },
    {
      "epoch": 44.00468468468468,
      "grad_norm": 0.00015174942382145673,
      "learning_rate": 1.2812812812812813e-06,
      "loss": 0.0,
      "step": 34370
    },
    {
      "epoch": 44.00494208494209,
      "grad_norm": 0.026701737195253372,
      "learning_rate": 1.2784212784212784e-06,
      "loss": 0.0,
      "step": 34380
    },
    {
      "epoch": 44.00519948519948,
      "grad_norm": 0.00013119878713041544,
      "learning_rate": 1.2755612755612757e-06,
      "loss": 0.0001,
      "step": 34390
    },
    {
      "epoch": 44.00545688545689,
      "grad_norm": 0.00010385591303929687,
      "learning_rate": 1.2727012727012728e-06,
      "loss": 0.0,
      "step": 34400
    },
    {
      "epoch": 44.005714285714284,
      "grad_norm": 0.0004606852598953992,
      "learning_rate": 1.26984126984127e-06,
      "loss": 0.0,
      "step": 34410
    },
    {
      "epoch": 44.00597168597169,
      "grad_norm": 0.002633896190673113,
      "learning_rate": 1.266981266981267e-06,
      "loss": 0.0001,
      "step": 34420
    },
    {
      "epoch": 44.006229086229084,
      "grad_norm": 0.001557589159347117,
      "learning_rate": 1.2641212641212641e-06,
      "loss": 0.0,
      "step": 34430
    },
    {
      "epoch": 44.00648648648649,
      "grad_norm": 0.03303082287311554,
      "learning_rate": 1.2612612612612613e-06,
      "loss": 0.0023,
      "step": 34440
    },
    {
      "epoch": 44.006743886743884,
      "grad_norm": 0.00011313861614326015,
      "learning_rate": 1.2584012584012584e-06,
      "loss": 0.0001,
      "step": 34450
    },
    {
      "epoch": 44.00700128700129,
      "grad_norm": 0.02742510288953781,
      "learning_rate": 1.2555412555412557e-06,
      "loss": 0.0001,
      "step": 34460
    },
    {
      "epoch": 44.007258687258684,
      "grad_norm": 0.0010636849328875542,
      "learning_rate": 1.252681252681253e-06,
      "loss": 0.0,
      "step": 34470
    },
    {
      "epoch": 44.00751608751609,
      "grad_norm": 9.581966878613457e-05,
      "learning_rate": 1.2498212498212499e-06,
      "loss": 0.0,
      "step": 34480
    },
    {
      "epoch": 44.00777348777349,
      "grad_norm": 1947.2777099609375,
      "learning_rate": 1.246961246961247e-06,
      "loss": 0.3187,
      "step": 34490
    },
    {
      "epoch": 44.00803088803089,
      "grad_norm": 0.00014478244702331722,
      "learning_rate": 1.244101244101244e-06,
      "loss": 0.6529,
      "step": 34500
    },
    {
      "epoch": 44.00828828828829,
      "grad_norm": 0.0012537414440885186,
      "learning_rate": 1.2412412412412414e-06,
      "loss": 0.0097,
      "step": 34510
    },
    {
      "epoch": 44.00854568854569,
      "grad_norm": 0.00011815100879175588,
      "learning_rate": 1.2383812383812385e-06,
      "loss": 0.4375,
      "step": 34520
    },
    {
      "epoch": 44.00880308880309,
      "grad_norm": 0.00031987254624255,
      "learning_rate": 1.2355212355212356e-06,
      "loss": 0.0001,
      "step": 34530
    },
    {
      "epoch": 44.00906048906049,
      "grad_norm": 0.0006125512300059199,
      "learning_rate": 1.2326612326612327e-06,
      "loss": 0.0002,
      "step": 34540
    },
    {
      "epoch": 44.00931788931789,
      "grad_norm": 8.829426224110648e-05,
      "learning_rate": 1.2298012298012298e-06,
      "loss": 0.6567,
      "step": 34550
    },
    {
      "epoch": 44.00957528957529,
      "grad_norm": 0.0020937060471624136,
      "learning_rate": 1.226941226941227e-06,
      "loss": 0.0001,
      "step": 34560
    },
    {
      "epoch": 44.00983268983269,
      "grad_norm": 0.0015685156686231494,
      "learning_rate": 1.2240812240812242e-06,
      "loss": 0.0,
      "step": 34570
    },
    {
      "epoch": 44.01009009009009,
      "grad_norm": 0.0005102535942569375,
      "learning_rate": 1.2212212212212213e-06,
      "loss": 1.1215,
      "step": 34580
    },
    {
      "epoch": 44.01034749034749,
      "grad_norm": 0.0015307324938476086,
      "learning_rate": 1.2183612183612185e-06,
      "loss": 0.0,
      "step": 34590
    },
    {
      "epoch": 44.01060489060489,
      "grad_norm": 0.00014208511856850237,
      "learning_rate": 1.2155012155012156e-06,
      "loss": 0.0038,
      "step": 34600
    },
    {
      "epoch": 44.01086229086229,
      "grad_norm": 0.04187508672475815,
      "learning_rate": 1.2126412126412129e-06,
      "loss": 0.6378,
      "step": 34610
    },
    {
      "epoch": 44.01111969111969,
      "grad_norm": 0.0010099312057718635,
      "learning_rate": 1.20978120978121e-06,
      "loss": 0.0,
      "step": 34620
    },
    {
      "epoch": 44.01137709137709,
      "grad_norm": 0.00010222556011285633,
      "learning_rate": 1.206921206921207e-06,
      "loss": 0.0,
      "step": 34630
    },
    {
      "epoch": 44.01163449163449,
      "grad_norm": 0.01367103774100542,
      "learning_rate": 1.2040612040612042e-06,
      "loss": 0.0,
      "step": 34640
    },
    {
      "epoch": 44.01189189189189,
      "grad_norm": 0.0009156797314062715,
      "learning_rate": 1.2012012012012013e-06,
      "loss": 0.0001,
      "step": 34650
    },
    {
      "epoch": 44.01214929214929,
      "grad_norm": 0.020239707082509995,
      "learning_rate": 1.1983411983411984e-06,
      "loss": 0.0,
      "step": 34660
    },
    {
      "epoch": 44.01240669240669,
      "grad_norm": 0.00010953854507533833,
      "learning_rate": 1.1954811954811955e-06,
      "loss": 0.0006,
      "step": 34670
    },
    {
      "epoch": 44.012664092664096,
      "grad_norm": 0.00010805286729009822,
      "learning_rate": 1.1926211926211926e-06,
      "loss": 0.6203,
      "step": 34680
    },
    {
      "epoch": 44.01292149292149,
      "grad_norm": 3.057513952255249,
      "learning_rate": 1.18976118976119e-06,
      "loss": 0.0004,
      "step": 34690
    },
    {
      "epoch": 44.013178893178896,
      "grad_norm": 40.827030181884766,
      "learning_rate": 1.186901186901187e-06,
      "loss": 0.0015,
      "step": 34700
    },
    {
      "epoch": 44.01343629343629,
      "grad_norm": 0.03450491279363632,
      "learning_rate": 1.1840411840411841e-06,
      "loss": 0.0001,
      "step": 34710
    },
    {
      "epoch": 44.013693693693696,
      "grad_norm": 0.00013904368097428232,
      "learning_rate": 1.1811811811811812e-06,
      "loss": 0.0,
      "step": 34720
    },
    {
      "epoch": 44.01395109395109,
      "grad_norm": 7.998216460691765e-05,
      "learning_rate": 1.1783211783211783e-06,
      "loss": 0.1652,
      "step": 34730
    },
    {
      "epoch": 44.014208494208496,
      "grad_norm": 0.00040033727418631315,
      "learning_rate": 1.1754611754611757e-06,
      "loss": 0.0001,
      "step": 34740
    },
    {
      "epoch": 44.01446589446589,
      "grad_norm": 0.00010231852502329275,
      "learning_rate": 1.1726011726011728e-06,
      "loss": 0.0005,
      "step": 34750
    },
    {
      "epoch": 44.0147232947233,
      "grad_norm": 0.0009837534744292498,
      "learning_rate": 1.1697411697411699e-06,
      "loss": 0.7573,
      "step": 34760
    },
    {
      "epoch": 44.01498069498069,
      "grad_norm": 0.003812544047832489,
      "learning_rate": 1.166881166881167e-06,
      "loss": 0.0025,
      "step": 34770
    },
    {
      "epoch": 44.0152380952381,
      "grad_norm": 0.0010106575209647417,
      "learning_rate": 1.164021164021164e-06,
      "loss": 0.0003,
      "step": 34780
    },
    {
      "epoch": 44.01549549549549,
      "grad_norm": 8.687885565450415e-05,
      "learning_rate": 1.1611611611611614e-06,
      "loss": 0.0001,
      "step": 34790
    },
    {
      "epoch": 44.0157528957529,
      "grad_norm": 0.0001750994415488094,
      "learning_rate": 1.1583011583011585e-06,
      "loss": 0.0001,
      "step": 34800
    },
    {
      "epoch": 44.01601029601029,
      "grad_norm": 0.0008143062004819512,
      "learning_rate": 1.1554411554411556e-06,
      "loss": 0.0001,
      "step": 34810
    },
    {
      "epoch": 44.0162676962677,
      "grad_norm": 0.0009398268884979188,
      "learning_rate": 1.1525811525811527e-06,
      "loss": 0.0001,
      "step": 34820
    },
    {
      "epoch": 44.01652509652509,
      "grad_norm": 0.0001232112990692258,
      "learning_rate": 1.1497211497211498e-06,
      "loss": 0.0001,
      "step": 34830
    },
    {
      "epoch": 44.0167824967825,
      "grad_norm": 0.006615647114813328,
      "learning_rate": 1.146861146861147e-06,
      "loss": 0.0567,
      "step": 34840
    },
    {
      "epoch": 44.01703989703989,
      "grad_norm": 0.0021121406462043524,
      "learning_rate": 1.144001144001144e-06,
      "loss": 0.0003,
      "step": 34850
    },
    {
      "epoch": 44.0172972972973,
      "grad_norm": 9.626035898691043e-05,
      "learning_rate": 1.1411411411411411e-06,
      "loss": 0.0001,
      "step": 34860
    },
    {
      "epoch": 44.0175546975547,
      "grad_norm": 0.00023816150496713817,
      "learning_rate": 1.1382811382811382e-06,
      "loss": 0.0001,
      "step": 34870
    },
    {
      "epoch": 44.0178120978121,
      "grad_norm": 0.00011487925803521648,
      "learning_rate": 1.1354211354211355e-06,
      "loss": 0.0002,
      "step": 34880
    },
    {
      "epoch": 44.0180694980695,
      "grad_norm": 0.009908772073686123,
      "learning_rate": 1.1325611325611326e-06,
      "loss": 1.6454,
      "step": 34890
    },
    {
      "epoch": 44.0183268983269,
      "grad_norm": 0.004898641724139452,
      "learning_rate": 1.1297011297011298e-06,
      "loss": 0.6973,
      "step": 34900
    },
    {
      "epoch": 44.0185842985843,
      "grad_norm": 9.289468289352953e-05,
      "learning_rate": 1.1268411268411269e-06,
      "loss": 0.0001,
      "step": 34910
    },
    {
      "epoch": 44.0188416988417,
      "grad_norm": 0.006554802879691124,
      "learning_rate": 1.1239811239811242e-06,
      "loss": 0.7225,
      "step": 34920
    },
    {
      "epoch": 44.0190990990991,
      "grad_norm": 0.002455679466947913,
      "learning_rate": 1.1211211211211213e-06,
      "loss": 0.0004,
      "step": 34930
    },
    {
      "epoch": 44.0193564993565,
      "grad_norm": 0.07329325377941132,
      "learning_rate": 1.1182611182611184e-06,
      "loss": 0.0002,
      "step": 34940
    },
    {
      "epoch": 44.0196138996139,
      "grad_norm": 0.00023317259910982102,
      "learning_rate": 1.1154011154011155e-06,
      "loss": 0.0001,
      "step": 34950
    },
    {
      "epoch": 44.0198712998713,
      "grad_norm": 0.04382844269275665,
      "learning_rate": 1.1125411125411126e-06,
      "loss": 0.0001,
      "step": 34960
    },
    {
      "epoch": 44.02,
      "eval_accuracy": 0.8913043478260869,
      "eval_loss": 0.9010905623435974,
      "eval_runtime": 13.3897,
      "eval_samples_per_second": 3.435,
      "eval_steps_per_second": 3.435,
      "step": 34965
    },
    {
      "epoch": 45.0001287001287,
      "grad_norm": 0.005866614170372486,
      "learning_rate": 1.1096811096811097e-06,
      "loss": 0.0001,
      "step": 34970
    },
    {
      "epoch": 45.0003861003861,
      "grad_norm": 0.00011965092562604696,
      "learning_rate": 1.106821106821107e-06,
      "loss": 0.1422,
      "step": 34980
    },
    {
      "epoch": 45.0006435006435,
      "grad_norm": 0.0015373146161437035,
      "learning_rate": 1.1039611039611041e-06,
      "loss": 0.0014,
      "step": 34990
    },
    {
      "epoch": 45.0009009009009,
      "grad_norm": 0.02903032675385475,
      "learning_rate": 1.1011011011011012e-06,
      "loss": 0.2738,
      "step": 35000
    },
    {
      "epoch": 45.0011583011583,
      "grad_norm": 0.00014115635713096708,
      "learning_rate": 1.0982410982410983e-06,
      "loss": 0.0451,
      "step": 35010
    },
    {
      "epoch": 45.0014157014157,
      "grad_norm": 0.00010962053056573495,
      "learning_rate": 1.0953810953810954e-06,
      "loss": 0.0,
      "step": 35020
    },
    {
      "epoch": 45.0016731016731,
      "grad_norm": 0.0038448001723736525,
      "learning_rate": 1.0925210925210925e-06,
      "loss": 0.0001,
      "step": 35030
    },
    {
      "epoch": 45.0019305019305,
      "grad_norm": 0.00015873707889113575,
      "learning_rate": 1.0896610896610896e-06,
      "loss": 0.0001,
      "step": 35040
    },
    {
      "epoch": 45.0021879021879,
      "grad_norm": 0.003100020345300436,
      "learning_rate": 1.0868010868010867e-06,
      "loss": 0.0,
      "step": 35050
    },
    {
      "epoch": 45.0024453024453,
      "grad_norm": 0.00020110583864152431,
      "learning_rate": 1.083941083941084e-06,
      "loss": 0.0,
      "step": 35060
    },
    {
      "epoch": 45.002702702702706,
      "grad_norm": 0.00012975893332622945,
      "learning_rate": 1.0810810810810812e-06,
      "loss": 0.0001,
      "step": 35070
    },
    {
      "epoch": 45.0029601029601,
      "grad_norm": 9.7593292593956e-05,
      "learning_rate": 1.0782210782210783e-06,
      "loss": 0.0006,
      "step": 35080
    },
    {
      "epoch": 45.003217503217506,
      "grad_norm": 0.02327113039791584,
      "learning_rate": 1.0753610753610756e-06,
      "loss": 0.728,
      "step": 35090
    },
    {
      "epoch": 45.0034749034749,
      "grad_norm": 0.03272142633795738,
      "learning_rate": 1.0725010725010727e-06,
      "loss": 0.0001,
      "step": 35100
    },
    {
      "epoch": 45.003732303732306,
      "grad_norm": 0.026449372991919518,
      "learning_rate": 1.0696410696410698e-06,
      "loss": 0.0001,
      "step": 35110
    },
    {
      "epoch": 45.0039897039897,
      "grad_norm": 0.0006191537249833345,
      "learning_rate": 1.066781066781067e-06,
      "loss": 0.0829,
      "step": 35120
    },
    {
      "epoch": 45.004247104247106,
      "grad_norm": 9.959912858903408e-05,
      "learning_rate": 1.063921063921064e-06,
      "loss": 0.0001,
      "step": 35130
    },
    {
      "epoch": 45.0045045045045,
      "grad_norm": 0.18045932054519653,
      "learning_rate": 1.0610610610610611e-06,
      "loss": 0.0001,
      "step": 35140
    },
    {
      "epoch": 45.00476190476191,
      "grad_norm": 0.0007838630117475986,
      "learning_rate": 1.0582010582010582e-06,
      "loss": 0.0,
      "step": 35150
    },
    {
      "epoch": 45.0050193050193,
      "grad_norm": 0.06128101423382759,
      "learning_rate": 1.0553410553410555e-06,
      "loss": 0.0001,
      "step": 35160
    },
    {
      "epoch": 45.00527670527671,
      "grad_norm": 0.018773818388581276,
      "learning_rate": 1.0524810524810526e-06,
      "loss": 0.0002,
      "step": 35170
    },
    {
      "epoch": 45.0055341055341,
      "grad_norm": 0.01248914934694767,
      "learning_rate": 1.0496210496210497e-06,
      "loss": 0.0005,
      "step": 35180
    },
    {
      "epoch": 45.00579150579151,
      "grad_norm": 0.0021200559567660093,
      "learning_rate": 1.0467610467610468e-06,
      "loss": 0.0,
      "step": 35190
    },
    {
      "epoch": 45.0060489060489,
      "grad_norm": 0.0001151402248069644,
      "learning_rate": 1.043901043901044e-06,
      "loss": 0.0002,
      "step": 35200
    },
    {
      "epoch": 45.00630630630631,
      "grad_norm": 0.00010334582475479692,
      "learning_rate": 1.041041041041041e-06,
      "loss": 0.0001,
      "step": 35210
    },
    {
      "epoch": 45.0065637065637,
      "grad_norm": 0.00017368808039464056,
      "learning_rate": 1.0381810381810382e-06,
      "loss": 0.7997,
      "step": 35220
    },
    {
      "epoch": 45.00682110682111,
      "grad_norm": 0.0025292395148426294,
      "learning_rate": 1.0353210353210355e-06,
      "loss": 1.6189,
      "step": 35230
    },
    {
      "epoch": 45.0070785070785,
      "grad_norm": 0.0001754663826432079,
      "learning_rate": 1.0324610324610326e-06,
      "loss": 0.0001,
      "step": 35240
    },
    {
      "epoch": 45.00733590733591,
      "grad_norm": 0.14910262823104858,
      "learning_rate": 1.0296010296010297e-06,
      "loss": 0.0002,
      "step": 35250
    },
    {
      "epoch": 45.00759330759331,
      "grad_norm": 0.00010118578211404383,
      "learning_rate": 1.0267410267410268e-06,
      "loss": 0.0594,
      "step": 35260
    },
    {
      "epoch": 45.00785070785071,
      "grad_norm": 0.001283423276618123,
      "learning_rate": 1.023881023881024e-06,
      "loss": 0.0001,
      "step": 35270
    },
    {
      "epoch": 45.00810810810811,
      "grad_norm": 0.0028013254050165415,
      "learning_rate": 1.0210210210210212e-06,
      "loss": 0.0,
      "step": 35280
    },
    {
      "epoch": 45.00836550836551,
      "grad_norm": 0.0008561990689486265,
      "learning_rate": 1.0181610181610183e-06,
      "loss": 0.0001,
      "step": 35290
    },
    {
      "epoch": 45.00862290862291,
      "grad_norm": 0.00010187633597524837,
      "learning_rate": 1.0153010153010154e-06,
      "loss": 0.0,
      "step": 35300
    },
    {
      "epoch": 45.00888030888031,
      "grad_norm": 0.008484124206006527,
      "learning_rate": 1.0124410124410125e-06,
      "loss": 0.8,
      "step": 35310
    },
    {
      "epoch": 45.00913770913771,
      "grad_norm": 0.0002765395911410451,
      "learning_rate": 1.0095810095810096e-06,
      "loss": 0.0,
      "step": 35320
    },
    {
      "epoch": 45.00939510939511,
      "grad_norm": 3.5977704524993896,
      "learning_rate": 1.0067210067210067e-06,
      "loss": 0.0006,
      "step": 35330
    },
    {
      "epoch": 45.00965250965251,
      "grad_norm": 0.0013253793586045504,
      "learning_rate": 1.0038610038610038e-06,
      "loss": 0.0525,
      "step": 35340
    },
    {
      "epoch": 45.00990990990991,
      "grad_norm": 0.002857430372387171,
      "learning_rate": 1.0010010010010011e-06,
      "loss": 0.0001,
      "step": 35350
    },
    {
      "epoch": 45.01016731016731,
      "grad_norm": 0.006012418307363987,
      "learning_rate": 9.981409981409983e-07,
      "loss": 0.0,
      "step": 35360
    },
    {
      "epoch": 45.01042471042471,
      "grad_norm": 0.046247564256191254,
      "learning_rate": 9.952809952809954e-07,
      "loss": 0.0004,
      "step": 35370
    },
    {
      "epoch": 45.01068211068211,
      "grad_norm": 0.0015133422566577792,
      "learning_rate": 9.924209924209925e-07,
      "loss": 0.0,
      "step": 35380
    },
    {
      "epoch": 45.01093951093951,
      "grad_norm": 0.0030911520589143038,
      "learning_rate": 9.895609895609896e-07,
      "loss": 0.0002,
      "step": 35390
    },
    {
      "epoch": 45.01119691119691,
      "grad_norm": 0.00036034543882124126,
      "learning_rate": 9.867009867009867e-07,
      "loss": 0.7626,
      "step": 35400
    },
    {
      "epoch": 45.01145431145431,
      "grad_norm": 0.0031866899225860834,
      "learning_rate": 9.83840983840984e-07,
      "loss": 0.0,
      "step": 35410
    },
    {
      "epoch": 45.01171171171171,
      "grad_norm": 0.009315420873463154,
      "learning_rate": 9.80980980980981e-07,
      "loss": 0.0001,
      "step": 35420
    },
    {
      "epoch": 45.011969111969115,
      "grad_norm": 0.00187424395699054,
      "learning_rate": 9.781209781209782e-07,
      "loss": 0.8233,
      "step": 35430
    },
    {
      "epoch": 45.01222651222651,
      "grad_norm": 0.007069937419146299,
      "learning_rate": 9.752609752609755e-07,
      "loss": 0.7884,
      "step": 35440
    },
    {
      "epoch": 45.012483912483916,
      "grad_norm": 0.00011888705921592191,
      "learning_rate": 9.724009724009726e-07,
      "loss": 0.0006,
      "step": 35450
    },
    {
      "epoch": 45.01274131274131,
      "grad_norm": 0.00011121502029709518,
      "learning_rate": 9.695409695409697e-07,
      "loss": 0.0017,
      "step": 35460
    },
    {
      "epoch": 45.012998712998716,
      "grad_norm": 0.00048588740173727274,
      "learning_rate": 9.666809666809668e-07,
      "loss": 0.0017,
      "step": 35470
    },
    {
      "epoch": 45.01325611325611,
      "grad_norm": 0.0035627170000225306,
      "learning_rate": 9.63820963820964e-07,
      "loss": 0.0001,
      "step": 35480
    },
    {
      "epoch": 45.013513513513516,
      "grad_norm": 0.0016204885905608535,
      "learning_rate": 9.60960960960961e-07,
      "loss": 0.0,
      "step": 35490
    },
    {
      "epoch": 45.01377091377091,
      "grad_norm": 1.0253617763519287,
      "learning_rate": 9.581009581009581e-07,
      "loss": 0.0004,
      "step": 35500
    },
    {
      "epoch": 45.014028314028316,
      "grad_norm": 0.7940694093704224,
      "learning_rate": 9.552409552409552e-07,
      "loss": 0.0002,
      "step": 35510
    },
    {
      "epoch": 45.01428571428571,
      "grad_norm": 8.757525210967287e-05,
      "learning_rate": 9.523809523809525e-07,
      "loss": 0.0,
      "step": 35520
    },
    {
      "epoch": 45.014543114543116,
      "grad_norm": 0.014474915340542793,
      "learning_rate": 9.495209495209496e-07,
      "loss": 0.0159,
      "step": 35530
    },
    {
      "epoch": 45.01480051480051,
      "grad_norm": 0.01842646673321724,
      "learning_rate": 9.466609466609468e-07,
      "loss": 0.0001,
      "step": 35540
    },
    {
      "epoch": 45.015057915057916,
      "grad_norm": 0.004527392331510782,
      "learning_rate": 9.438009438009439e-07,
      "loss": 0.0001,
      "step": 35550
    },
    {
      "epoch": 45.01531531531531,
      "grad_norm": 0.00041223526932299137,
      "learning_rate": 9.409409409409411e-07,
      "loss": 0.0,
      "step": 35560
    },
    {
      "epoch": 45.015572715572716,
      "grad_norm": 0.000709443585947156,
      "learning_rate": 9.380809380809382e-07,
      "loss": 0.0001,
      "step": 35570
    },
    {
      "epoch": 45.01583011583011,
      "grad_norm": 0.00010205066064372659,
      "learning_rate": 9.352209352209353e-07,
      "loss": 1.0062,
      "step": 35580
    },
    {
      "epoch": 45.01608751608752,
      "grad_norm": 0.0038576810620725155,
      "learning_rate": 9.323609323609324e-07,
      "loss": 0.0,
      "step": 35590
    },
    {
      "epoch": 45.01634491634491,
      "grad_norm": 0.0001053248270181939,
      "learning_rate": 9.295009295009295e-07,
      "loss": 0.3508,
      "step": 35600
    },
    {
      "epoch": 45.01660231660232,
      "grad_norm": 0.006075258832424879,
      "learning_rate": 9.266409266409267e-07,
      "loss": 0.0001,
      "step": 35610
    },
    {
      "epoch": 45.01685971685972,
      "grad_norm": 0.00010978049976984039,
      "learning_rate": 9.237809237809238e-07,
      "loss": 0.0001,
      "step": 35620
    },
    {
      "epoch": 45.01711711711712,
      "grad_norm": 0.0025920916814357042,
      "learning_rate": 9.20920920920921e-07,
      "loss": 0.0002,
      "step": 35630
    },
    {
      "epoch": 45.01737451737452,
      "grad_norm": 0.0029896306805312634,
      "learning_rate": 9.180609180609181e-07,
      "loss": 0.0,
      "step": 35640
    },
    {
      "epoch": 45.01763191763192,
      "grad_norm": 0.004719553980976343,
      "learning_rate": 9.152009152009153e-07,
      "loss": 0.0,
      "step": 35650
    },
    {
      "epoch": 45.01788931788932,
      "grad_norm": 0.01105885673314333,
      "learning_rate": 9.123409123409124e-07,
      "loss": 0.013,
      "step": 35660
    },
    {
      "epoch": 45.01814671814672,
      "grad_norm": 344.09503173828125,
      "learning_rate": 9.094809094809096e-07,
      "loss": 0.5786,
      "step": 35670
    },
    {
      "epoch": 45.01840411840412,
      "grad_norm": 0.00010508803825359792,
      "learning_rate": 9.066209066209067e-07,
      "loss": 0.0,
      "step": 35680
    },
    {
      "epoch": 45.01866151866152,
      "grad_norm": 0.0030339634977281094,
      "learning_rate": 9.037609037609038e-07,
      "loss": 0.0001,
      "step": 35690
    },
    {
      "epoch": 45.01891891891892,
      "grad_norm": 0.08380759507417679,
      "learning_rate": 9.00900900900901e-07,
      "loss": 0.0001,
      "step": 35700
    },
    {
      "epoch": 45.01917631917632,
      "grad_norm": 50.496097564697266,
      "learning_rate": 8.980408980408981e-07,
      "loss": 0.0095,
      "step": 35710
    },
    {
      "epoch": 45.01943371943372,
      "grad_norm": 0.00014803580415900797,
      "learning_rate": 8.951808951808953e-07,
      "loss": 0.0001,
      "step": 35720
    },
    {
      "epoch": 45.01969111969112,
      "grad_norm": 9.634919842937961e-05,
      "learning_rate": 8.923208923208925e-07,
      "loss": 0.0001,
      "step": 35730
    },
    {
      "epoch": 45.01994851994852,
      "grad_norm": 909.0189208984375,
      "learning_rate": 8.894608894608896e-07,
      "loss": 0.6661,
      "step": 35740
    },
    {
      "epoch": 45.02,
      "eval_accuracy": 0.8478260869565217,
      "eval_loss": 1.0514214038848877,
      "eval_runtime": 13.4683,
      "eval_samples_per_second": 3.415,
      "eval_steps_per_second": 3.415,
      "step": 35742
    },
    {
      "epoch": 46.00020592020592,
      "grad_norm": 0.007054031360894442,
      "learning_rate": 8.866008866008867e-07,
      "loss": 0.0001,
      "step": 35750
    },
    {
      "epoch": 46.00046332046332,
      "grad_norm": 0.004835424479097128,
      "learning_rate": 8.837408837408838e-07,
      "loss": 0.0002,
      "step": 35760
    },
    {
      "epoch": 46.00072072072072,
      "grad_norm": 0.00011531000927789137,
      "learning_rate": 8.808808808808809e-07,
      "loss": 0.0,
      "step": 35770
    },
    {
      "epoch": 46.00097812097812,
      "grad_norm": 0.001735526486299932,
      "learning_rate": 8.78020878020878e-07,
      "loss": 0.0002,
      "step": 35780
    },
    {
      "epoch": 46.00123552123552,
      "grad_norm": 0.0001596819784026593,
      "learning_rate": 8.751608751608752e-07,
      "loss": 0.0,
      "step": 35790
    },
    {
      "epoch": 46.00149292149292,
      "grad_norm": 0.0007862728089094162,
      "learning_rate": 8.723008723008723e-07,
      "loss": 0.0001,
      "step": 35800
    },
    {
      "epoch": 46.00175032175032,
      "grad_norm": 0.00195333338342607,
      "learning_rate": 8.694408694408694e-07,
      "loss": 0.4595,
      "step": 35810
    },
    {
      "epoch": 46.002007722007725,
      "grad_norm": 0.03994870185852051,
      "learning_rate": 8.665808665808668e-07,
      "loss": 0.0002,
      "step": 35820
    },
    {
      "epoch": 46.00226512226512,
      "grad_norm": 0.012519776821136475,
      "learning_rate": 8.637208637208639e-07,
      "loss": 0.0,
      "step": 35830
    },
    {
      "epoch": 46.002522522522526,
      "grad_norm": 0.004235388711094856,
      "learning_rate": 8.60860860860861e-07,
      "loss": 0.0,
      "step": 35840
    },
    {
      "epoch": 46.00277992277992,
      "grad_norm": 0.00017162068979814649,
      "learning_rate": 8.580008580008581e-07,
      "loss": 0.4289,
      "step": 35850
    },
    {
      "epoch": 46.003037323037326,
      "grad_norm": 0.00011935268412344158,
      "learning_rate": 8.551408551408552e-07,
      "loss": 0.0002,
      "step": 35860
    },
    {
      "epoch": 46.00329472329472,
      "grad_norm": 0.00021309407020453364,
      "learning_rate": 8.522808522808524e-07,
      "loss": 0.0001,
      "step": 35870
    },
    {
      "epoch": 46.003552123552126,
      "grad_norm": 0.03302145376801491,
      "learning_rate": 8.494208494208495e-07,
      "loss": 0.0002,
      "step": 35880
    },
    {
      "epoch": 46.00380952380952,
      "grad_norm": 0.00010466259845998138,
      "learning_rate": 8.465608465608466e-07,
      "loss": 0.0001,
      "step": 35890
    },
    {
      "epoch": 46.004066924066926,
      "grad_norm": 0.0013877922901883721,
      "learning_rate": 8.437008437008437e-07,
      "loss": 0.0001,
      "step": 35900
    },
    {
      "epoch": 46.00432432432432,
      "grad_norm": 0.006681135855615139,
      "learning_rate": 8.40840840840841e-07,
      "loss": 0.0001,
      "step": 35910
    },
    {
      "epoch": 46.004581724581726,
      "grad_norm": 0.00010747831402113661,
      "learning_rate": 8.379808379808381e-07,
      "loss": 0.0,
      "step": 35920
    },
    {
      "epoch": 46.00483912483912,
      "grad_norm": 0.002587482100352645,
      "learning_rate": 8.351208351208352e-07,
      "loss": 0.0,
      "step": 35930
    },
    {
      "epoch": 46.005096525096526,
      "grad_norm": 0.00039881718112155795,
      "learning_rate": 8.322608322608323e-07,
      "loss": 0.495,
      "step": 35940
    },
    {
      "epoch": 46.00535392535392,
      "grad_norm": 8.408023131778464e-05,
      "learning_rate": 8.294008294008294e-07,
      "loss": 0.0,
      "step": 35950
    },
    {
      "epoch": 46.005611325611326,
      "grad_norm": 0.007310986518859863,
      "learning_rate": 8.265408265408266e-07,
      "loss": 0.0003,
      "step": 35960
    },
    {
      "epoch": 46.00586872586872,
      "grad_norm": 0.0002170134976040572,
      "learning_rate": 8.236808236808237e-07,
      "loss": 0.8696,
      "step": 35970
    },
    {
      "epoch": 46.00612612612613,
      "grad_norm": 0.007625481579452753,
      "learning_rate": 8.208208208208208e-07,
      "loss": 0.0001,
      "step": 35980
    },
    {
      "epoch": 46.00638352638352,
      "grad_norm": 8.718504977878183e-05,
      "learning_rate": 8.17960817960818e-07,
      "loss": 0.0001,
      "step": 35990
    },
    {
      "epoch": 46.00664092664093,
      "grad_norm": 0.00394862936809659,
      "learning_rate": 8.151008151008153e-07,
      "loss": 0.0002,
      "step": 36000
    },
    {
      "epoch": 46.00689832689833,
      "grad_norm": 9.449994831811637e-05,
      "learning_rate": 8.122408122408124e-07,
      "loss": 0.2956,
      "step": 36010
    },
    {
      "epoch": 46.00715572715573,
      "grad_norm": 0.004683338571339846,
      "learning_rate": 8.093808093808095e-07,
      "loss": 0.0001,
      "step": 36020
    },
    {
      "epoch": 46.00741312741313,
      "grad_norm": 0.00012272816093172878,
      "learning_rate": 8.065208065208066e-07,
      "loss": 0.0003,
      "step": 36030
    },
    {
      "epoch": 46.00767052767053,
      "grad_norm": 0.061585329473018646,
      "learning_rate": 8.036608036608037e-07,
      "loss": 0.0,
      "step": 36040
    },
    {
      "epoch": 46.00792792792793,
      "grad_norm": 6465.0048828125,
      "learning_rate": 8.008008008008009e-07,
      "loss": 0.3987,
      "step": 36050
    },
    {
      "epoch": 46.00818532818533,
      "grad_norm": 0.0032067273277789354,
      "learning_rate": 7.97940797940798e-07,
      "loss": 0.0,
      "step": 36060
    },
    {
      "epoch": 46.00844272844273,
      "grad_norm": 0.004086290951818228,
      "learning_rate": 7.950807950807951e-07,
      "loss": 0.9153,
      "step": 36070
    },
    {
      "epoch": 46.00870012870013,
      "grad_norm": 0.24981355667114258,
      "learning_rate": 7.922207922207922e-07,
      "loss": 0.0001,
      "step": 36080
    },
    {
      "epoch": 46.00895752895753,
      "grad_norm": 0.027362490072846413,
      "learning_rate": 7.893607893607893e-07,
      "loss": 0.0001,
      "step": 36090
    },
    {
      "epoch": 46.00921492921493,
      "grad_norm": 0.008485764265060425,
      "learning_rate": 7.865007865007866e-07,
      "loss": 0.187,
      "step": 36100
    },
    {
      "epoch": 46.00947232947233,
      "grad_norm": 0.00039889125037007034,
      "learning_rate": 7.836407836407837e-07,
      "loss": 0.0001,
      "step": 36110
    },
    {
      "epoch": 46.00972972972973,
      "grad_norm": 0.10861273854970932,
      "learning_rate": 7.807807807807808e-07,
      "loss": 0.0002,
      "step": 36120
    },
    {
      "epoch": 46.00998712998713,
      "grad_norm": 0.003896931419149041,
      "learning_rate": 7.779207779207779e-07,
      "loss": 0.0,
      "step": 36130
    },
    {
      "epoch": 46.01024453024453,
      "grad_norm": 0.001219980069436133,
      "learning_rate": 7.750607750607752e-07,
      "loss": 0.0001,
      "step": 36140
    },
    {
      "epoch": 46.01050193050193,
      "grad_norm": 0.00012802051787730306,
      "learning_rate": 7.722007722007723e-07,
      "loss": 0.0001,
      "step": 36150
    },
    {
      "epoch": 46.01075933075933,
      "grad_norm": 0.001733460114337504,
      "learning_rate": 7.693407693407694e-07,
      "loss": 0.0,
      "step": 36160
    },
    {
      "epoch": 46.01101673101673,
      "grad_norm": 0.01749817654490471,
      "learning_rate": 7.664807664807665e-07,
      "loss": 0.0002,
      "step": 36170
    },
    {
      "epoch": 46.01127413127413,
      "grad_norm": 0.02637387067079544,
      "learning_rate": 7.636207636207636e-07,
      "loss": 0.0001,
      "step": 36180
    },
    {
      "epoch": 46.01153153153153,
      "grad_norm": 0.0005767493858002126,
      "learning_rate": 7.607607607607609e-07,
      "loss": 0.0001,
      "step": 36190
    },
    {
      "epoch": 46.011788931788935,
      "grad_norm": 0.00018126668874174356,
      "learning_rate": 7.57900757900758e-07,
      "loss": 0.003,
      "step": 36200
    },
    {
      "epoch": 46.01204633204633,
      "grad_norm": 0.00014345711679197848,
      "learning_rate": 7.550407550407551e-07,
      "loss": 0.0001,
      "step": 36210
    },
    {
      "epoch": 46.012303732303735,
      "grad_norm": 0.009981050156056881,
      "learning_rate": 7.521807521807523e-07,
      "loss": 0.0001,
      "step": 36220
    },
    {
      "epoch": 46.01256113256113,
      "grad_norm": 0.00011738213652279228,
      "learning_rate": 7.493207493207494e-07,
      "loss": 0.0,
      "step": 36230
    },
    {
      "epoch": 46.012818532818535,
      "grad_norm": 1049.6097412109375,
      "learning_rate": 7.464607464607465e-07,
      "loss": 1.1861,
      "step": 36240
    },
    {
      "epoch": 46.01307593307593,
      "grad_norm": 187.1683349609375,
      "learning_rate": 7.436007436007436e-07,
      "loss": 0.787,
      "step": 36250
    },
    {
      "epoch": 46.013333333333335,
      "grad_norm": 0.0012502912431955338,
      "learning_rate": 7.407407407407407e-07,
      "loss": 0.0389,
      "step": 36260
    },
    {
      "epoch": 46.01359073359073,
      "grad_norm": 0.00012157708260929212,
      "learning_rate": 7.378807378807379e-07,
      "loss": 0.0001,
      "step": 36270
    },
    {
      "epoch": 46.013848133848136,
      "grad_norm": 0.002692732261493802,
      "learning_rate": 7.350207350207351e-07,
      "loss": 0.0001,
      "step": 36280
    },
    {
      "epoch": 46.01410553410553,
      "grad_norm": 0.0001023231161525473,
      "learning_rate": 7.321607321607323e-07,
      "loss": 0.9306,
      "step": 36290
    },
    {
      "epoch": 46.014362934362936,
      "grad_norm": 0.10785011947154999,
      "learning_rate": 7.293007293007294e-07,
      "loss": 0.0001,
      "step": 36300
    },
    {
      "epoch": 46.01462033462033,
      "grad_norm": 0.004624065011739731,
      "learning_rate": 7.264407264407266e-07,
      "loss": 0.0001,
      "step": 36310
    },
    {
      "epoch": 46.014877734877736,
      "grad_norm": 0.0320514477789402,
      "learning_rate": 7.235807235807237e-07,
      "loss": 0.6644,
      "step": 36320
    },
    {
      "epoch": 46.01513513513513,
      "grad_norm": 0.07943771779537201,
      "learning_rate": 7.207207207207208e-07,
      "loss": 0.0001,
      "step": 36330
    },
    {
      "epoch": 46.015392535392536,
      "grad_norm": 0.0013853806303814054,
      "learning_rate": 7.178607178607179e-07,
      "loss": 0.1372,
      "step": 36340
    },
    {
      "epoch": 46.01564993564993,
      "grad_norm": 0.00339730572886765,
      "learning_rate": 7.15000715000715e-07,
      "loss": 0.0,
      "step": 36350
    },
    {
      "epoch": 46.015907335907336,
      "grad_norm": 0.00013834710989613086,
      "learning_rate": 7.121407121407122e-07,
      "loss": 0.0002,
      "step": 36360
    },
    {
      "epoch": 46.01616473616474,
      "grad_norm": 0.01571325585246086,
      "learning_rate": 7.092807092807093e-07,
      "loss": 0.0001,
      "step": 36370
    },
    {
      "epoch": 46.016422136422136,
      "grad_norm": 0.00020861165830865502,
      "learning_rate": 7.064207064207065e-07,
      "loss": 0.0002,
      "step": 36380
    },
    {
      "epoch": 46.01667953667954,
      "grad_norm": 30.054359436035156,
      "learning_rate": 7.035607035607036e-07,
      "loss": 0.7005,
      "step": 36390
    },
    {
      "epoch": 46.016936936936936,
      "grad_norm": 9.627743565943092e-05,
      "learning_rate": 7.007007007007008e-07,
      "loss": 0.0007,
      "step": 36400
    },
    {
      "epoch": 46.01719433719434,
      "grad_norm": 0.005623129662126303,
      "learning_rate": 6.978406978406979e-07,
      "loss": 0.0,
      "step": 36410
    },
    {
      "epoch": 46.01745173745174,
      "grad_norm": 0.04780065640807152,
      "learning_rate": 6.94980694980695e-07,
      "loss": 0.0001,
      "step": 36420
    },
    {
      "epoch": 46.01770913770914,
      "grad_norm": 0.00010322517482563853,
      "learning_rate": 6.921206921206921e-07,
      "loss": 0.0002,
      "step": 36430
    },
    {
      "epoch": 46.01796653796654,
      "grad_norm": 0.0031095435842871666,
      "learning_rate": 6.892606892606892e-07,
      "loss": 0.7759,
      "step": 36440
    },
    {
      "epoch": 46.01822393822394,
      "grad_norm": 0.00010584430856397375,
      "learning_rate": 6.864006864006865e-07,
      "loss": 0.0,
      "step": 36450
    },
    {
      "epoch": 46.01848133848134,
      "grad_norm": 0.00017411337466910481,
      "learning_rate": 6.835406835406836e-07,
      "loss": 0.0,
      "step": 36460
    },
    {
      "epoch": 46.01873873873874,
      "grad_norm": 0.008927185088396072,
      "learning_rate": 6.806806806806808e-07,
      "loss": 0.0001,
      "step": 36470
    },
    {
      "epoch": 46.01899613899614,
      "grad_norm": 0.0015496767591685057,
      "learning_rate": 6.778206778206779e-07,
      "loss": 0.0001,
      "step": 36480
    },
    {
      "epoch": 46.01925353925354,
      "grad_norm": 0.0566856823861599,
      "learning_rate": 6.749606749606751e-07,
      "loss": 0.2111,
      "step": 36490
    },
    {
      "epoch": 46.01951093951094,
      "grad_norm": 0.00018851790810003877,
      "learning_rate": 6.721006721006722e-07,
      "loss": 0.0005,
      "step": 36500
    },
    {
      "epoch": 46.01976833976834,
      "grad_norm": 0.005153292324393988,
      "learning_rate": 6.692406692406693e-07,
      "loss": 0.0001,
      "step": 36510
    },
    {
      "epoch": 46.02,
      "eval_accuracy": 0.8913043478260869,
      "eval_loss": 0.8273398280143738,
      "eval_runtime": 13.4638,
      "eval_samples_per_second": 3.417,
      "eval_steps_per_second": 3.417,
      "step": 36519
    },
    {
      "epoch": 47.00002574002574,
      "grad_norm": 0.00952563900500536,
      "learning_rate": 6.663806663806664e-07,
      "loss": 0.1634,
      "step": 36520
    },
    {
      "epoch": 47.00028314028314,
      "grad_norm": 0.05393436178565025,
      "learning_rate": 6.635206635206635e-07,
      "loss": 0.0001,
      "step": 36530
    },
    {
      "epoch": 47.00054054054054,
      "grad_norm": 0.07490912079811096,
      "learning_rate": 6.606606606606607e-07,
      "loss": 0.0001,
      "step": 36540
    },
    {
      "epoch": 47.00079794079794,
      "grad_norm": 8.754934242460877e-05,
      "learning_rate": 6.578006578006578e-07,
      "loss": 0.0,
      "step": 36550
    },
    {
      "epoch": 47.00105534105534,
      "grad_norm": 0.003464779816567898,
      "learning_rate": 6.54940654940655e-07,
      "loss": 0.0001,
      "step": 36560
    },
    {
      "epoch": 47.00131274131274,
      "grad_norm": 0.00010122011735802516,
      "learning_rate": 6.520806520806522e-07,
      "loss": 0.0002,
      "step": 36570
    },
    {
      "epoch": 47.00157014157014,
      "grad_norm": 199.99314880371094,
      "learning_rate": 6.492206492206493e-07,
      "loss": 0.7026,
      "step": 36580
    },
    {
      "epoch": 47.001827541827545,
      "grad_norm": 0.0013138225767761469,
      "learning_rate": 6.463606463606464e-07,
      "loss": 0.0,
      "step": 36590
    },
    {
      "epoch": 47.00208494208494,
      "grad_norm": 0.00027888649492524564,
      "learning_rate": 6.435006435006435e-07,
      "loss": 0.0002,
      "step": 36600
    },
    {
      "epoch": 47.002342342342345,
      "grad_norm": 380.2559814453125,
      "learning_rate": 6.406406406406407e-07,
      "loss": 0.7617,
      "step": 36610
    },
    {
      "epoch": 47.00259974259974,
      "grad_norm": 0.0002311421267222613,
      "learning_rate": 6.377806377806379e-07,
      "loss": 0.0,
      "step": 36620
    },
    {
      "epoch": 47.002857142857145,
      "grad_norm": 0.002484462922438979,
      "learning_rate": 6.34920634920635e-07,
      "loss": 0.0001,
      "step": 36630
    },
    {
      "epoch": 47.00311454311454,
      "grad_norm": 0.3317424952983856,
      "learning_rate": 6.320606320606321e-07,
      "loss": 0.0001,
      "step": 36640
    },
    {
      "epoch": 47.003371943371945,
      "grad_norm": 0.0013046682579442859,
      "learning_rate": 6.292006292006292e-07,
      "loss": 0.0,
      "step": 36650
    },
    {
      "epoch": 47.00362934362934,
      "grad_norm": 0.0008944895234890282,
      "learning_rate": 6.263406263406265e-07,
      "loss": 0.4445,
      "step": 36660
    },
    {
      "epoch": 47.003886743886746,
      "grad_norm": 0.01090352050960064,
      "learning_rate": 6.234806234806235e-07,
      "loss": 0.0001,
      "step": 36670
    },
    {
      "epoch": 47.00414414414414,
      "grad_norm": 0.00010999095684383065,
      "learning_rate": 6.206206206206207e-07,
      "loss": 0.0001,
      "step": 36680
    },
    {
      "epoch": 47.004401544401546,
      "grad_norm": 0.0012661719229072332,
      "learning_rate": 6.177606177606178e-07,
      "loss": 0.0002,
      "step": 36690
    },
    {
      "epoch": 47.00465894465894,
      "grad_norm": 0.003985162358731031,
      "learning_rate": 6.149006149006149e-07,
      "loss": 0.0,
      "step": 36700
    },
    {
      "epoch": 47.004916344916346,
      "grad_norm": 0.00013246751041151583,
      "learning_rate": 6.120406120406121e-07,
      "loss": 0.0035,
      "step": 36710
    },
    {
      "epoch": 47.00517374517374,
      "grad_norm": 0.027062317356467247,
      "learning_rate": 6.091806091806092e-07,
      "loss": 0.5117,
      "step": 36720
    },
    {
      "epoch": 47.005431145431146,
      "grad_norm": 0.004527709446847439,
      "learning_rate": 6.063206063206064e-07,
      "loss": 0.001,
      "step": 36730
    },
    {
      "epoch": 47.00568854568854,
      "grad_norm": 0.00012451779912225902,
      "learning_rate": 6.034606034606035e-07,
      "loss": 0.0001,
      "step": 36740
    },
    {
      "epoch": 47.005945945945946,
      "grad_norm": 0.08242174237966537,
      "learning_rate": 6.006006006006006e-07,
      "loss": 0.0001,
      "step": 36750
    },
    {
      "epoch": 47.00620334620335,
      "grad_norm": 0.0005578892887569964,
      "learning_rate": 5.977405977405978e-07,
      "loss": 0.0001,
      "step": 36760
    },
    {
      "epoch": 47.006460746460746,
      "grad_norm": 0.0012626610696315765,
      "learning_rate": 5.94880594880595e-07,
      "loss": 0.298,
      "step": 36770
    },
    {
      "epoch": 47.00671814671815,
      "grad_norm": 0.00010453144204802811,
      "learning_rate": 5.920205920205921e-07,
      "loss": 0.0001,
      "step": 36780
    },
    {
      "epoch": 47.006975546975546,
      "grad_norm": 0.000996722374111414,
      "learning_rate": 5.891605891605892e-07,
      "loss": 0.0,
      "step": 36790
    },
    {
      "epoch": 47.00723294723295,
      "grad_norm": 9.595556184649467e-05,
      "learning_rate": 5.863005863005864e-07,
      "loss": 0.0027,
      "step": 36800
    },
    {
      "epoch": 47.00749034749035,
      "grad_norm": 1.0915254354476929,
      "learning_rate": 5.834405834405835e-07,
      "loss": 0.0003,
      "step": 36810
    },
    {
      "epoch": 47.00774774774775,
      "grad_norm": 0.00010088778799399734,
      "learning_rate": 5.805805805805807e-07,
      "loss": 0.3989,
      "step": 36820
    },
    {
      "epoch": 47.00800514800515,
      "grad_norm": 0.00810331478714943,
      "learning_rate": 5.777205777205778e-07,
      "loss": 0.0001,
      "step": 36830
    },
    {
      "epoch": 47.00826254826255,
      "grad_norm": 7.67952369642444e-05,
      "learning_rate": 5.748605748605749e-07,
      "loss": 0.0001,
      "step": 36840
    },
    {
      "epoch": 47.00851994851995,
      "grad_norm": 0.0010103165404871106,
      "learning_rate": 5.72000572000572e-07,
      "loss": 0.0,
      "step": 36850
    },
    {
      "epoch": 47.00877734877735,
      "grad_norm": 0.007856501266360283,
      "learning_rate": 5.691405691405691e-07,
      "loss": 0.0,
      "step": 36860
    },
    {
      "epoch": 47.00903474903475,
      "grad_norm": 0.0008229284430854023,
      "learning_rate": 5.662805662805663e-07,
      "loss": 0.0001,
      "step": 36870
    },
    {
      "epoch": 47.00929214929215,
      "grad_norm": 0.0014737971359863877,
      "learning_rate": 5.634205634205634e-07,
      "loss": 0.6789,
      "step": 36880
    },
    {
      "epoch": 47.00954954954955,
      "grad_norm": 0.10162188857793808,
      "learning_rate": 5.605605605605606e-07,
      "loss": 0.0001,
      "step": 36890
    },
    {
      "epoch": 47.00980694980695,
      "grad_norm": 0.00013945061073172837,
      "learning_rate": 5.577005577005577e-07,
      "loss": 0.5182,
      "step": 36900
    },
    {
      "epoch": 47.01006435006435,
      "grad_norm": 0.0001227227330673486,
      "learning_rate": 5.548405548405548e-07,
      "loss": 0.0,
      "step": 36910
    },
    {
      "epoch": 47.01032175032175,
      "grad_norm": 9.417844557901844e-05,
      "learning_rate": 5.519805519805521e-07,
      "loss": 0.0151,
      "step": 36920
    },
    {
      "epoch": 47.01057915057915,
      "grad_norm": 0.020008182153105736,
      "learning_rate": 5.491205491205492e-07,
      "loss": 0.0005,
      "step": 36930
    },
    {
      "epoch": 47.01083655083655,
      "grad_norm": 0.00020941771799698472,
      "learning_rate": 5.462605462605463e-07,
      "loss": 0.0013,
      "step": 36940
    },
    {
      "epoch": 47.011093951093955,
      "grad_norm": 0.05506380647420883,
      "learning_rate": 5.434005434005434e-07,
      "loss": 0.0001,
      "step": 36950
    },
    {
      "epoch": 47.01135135135135,
      "grad_norm": 0.0007317409035749733,
      "learning_rate": 5.405405405405406e-07,
      "loss": 0.0003,
      "step": 36960
    },
    {
      "epoch": 47.011608751608755,
      "grad_norm": 0.005294485948979855,
      "learning_rate": 5.376805376805378e-07,
      "loss": 0.116,
      "step": 36970
    },
    {
      "epoch": 47.01186615186615,
      "grad_norm": 9.708470315672457e-05,
      "learning_rate": 5.348205348205349e-07,
      "loss": 0.0001,
      "step": 36980
    },
    {
      "epoch": 47.012123552123555,
      "grad_norm": 0.0071139405481517315,
      "learning_rate": 5.31960531960532e-07,
      "loss": 0.0,
      "step": 36990
    },
    {
      "epoch": 47.01238095238095,
      "grad_norm": 0.002544434741139412,
      "learning_rate": 5.291005291005291e-07,
      "loss": 0.0,
      "step": 37000
    },
    {
      "epoch": 47.012638352638355,
      "grad_norm": 0.06576728820800781,
      "learning_rate": 5.262405262405263e-07,
      "loss": 0.0001,
      "step": 37010
    },
    {
      "epoch": 47.01289575289575,
      "grad_norm": 7.955427281558514e-05,
      "learning_rate": 5.233805233805234e-07,
      "loss": 0.0001,
      "step": 37020
    },
    {
      "epoch": 47.013153153153155,
      "grad_norm": 9.721639071358368e-05,
      "learning_rate": 5.205205205205205e-07,
      "loss": 0.0,
      "step": 37030
    },
    {
      "epoch": 47.01341055341055,
      "grad_norm": 0.049452438950538635,
      "learning_rate": 5.176605176605177e-07,
      "loss": 0.0,
      "step": 37040
    },
    {
      "epoch": 47.013667953667955,
      "grad_norm": 0.00011968828039243817,
      "learning_rate": 5.148005148005148e-07,
      "loss": 0.0001,
      "step": 37050
    },
    {
      "epoch": 47.01392535392535,
      "grad_norm": 0.0015481059672310948,
      "learning_rate": 5.11940511940512e-07,
      "loss": 0.0,
      "step": 37060
    },
    {
      "epoch": 47.014182754182755,
      "grad_norm": 0.0013427004450932145,
      "learning_rate": 5.090805090805092e-07,
      "loss": 0.9195,
      "step": 37070
    },
    {
      "epoch": 47.01444015444015,
      "grad_norm": 0.001583964447490871,
      "learning_rate": 5.062205062205063e-07,
      "loss": 0.0001,
      "step": 37080
    },
    {
      "epoch": 47.014697554697555,
      "grad_norm": 0.014870808459818363,
      "learning_rate": 5.033605033605034e-07,
      "loss": 0.735,
      "step": 37090
    },
    {
      "epoch": 47.01495495495495,
      "grad_norm": 9.150934783974662e-05,
      "learning_rate": 5.005005005005006e-07,
      "loss": 0.0001,
      "step": 37100
    },
    {
      "epoch": 47.015212355212356,
      "grad_norm": 0.009138006716966629,
      "learning_rate": 4.976404976404977e-07,
      "loss": 0.0,
      "step": 37110
    },
    {
      "epoch": 47.01546975546975,
      "grad_norm": 0.00016449633403681219,
      "learning_rate": 4.947804947804948e-07,
      "loss": 0.0,
      "step": 37120
    },
    {
      "epoch": 47.015727155727156,
      "grad_norm": 0.001054841559380293,
      "learning_rate": 4.91920491920492e-07,
      "loss": 0.0405,
      "step": 37130
    },
    {
      "epoch": 47.01598455598456,
      "grad_norm": 0.00013857490557711571,
      "learning_rate": 4.890604890604891e-07,
      "loss": 0.0001,
      "step": 37140
    },
    {
      "epoch": 47.016241956241956,
      "grad_norm": 0.0011062290286645293,
      "learning_rate": 4.862004862004863e-07,
      "loss": 0.0001,
      "step": 37150
    },
    {
      "epoch": 47.01649935649936,
      "grad_norm": 0.0005065145669505,
      "learning_rate": 4.833404833404834e-07,
      "loss": 1.0933,
      "step": 37160
    },
    {
      "epoch": 47.016756756756756,
      "grad_norm": 0.029881233349442482,
      "learning_rate": 4.804804804804805e-07,
      "loss": 0.0,
      "step": 37170
    },
    {
      "epoch": 47.01701415701416,
      "grad_norm": 0.009099447168409824,
      "learning_rate": 4.776204776204776e-07,
      "loss": 0.0002,
      "step": 37180
    },
    {
      "epoch": 47.017271557271556,
      "grad_norm": 0.003578891744837165,
      "learning_rate": 4.747604747604748e-07,
      "loss": 0.7147,
      "step": 37190
    },
    {
      "epoch": 47.01752895752896,
      "grad_norm": 0.0010232250206172466,
      "learning_rate": 4.7190047190047194e-07,
      "loss": 0.0001,
      "step": 37200
    },
    {
      "epoch": 47.017786357786356,
      "grad_norm": 0.00027414108626544476,
      "learning_rate": 4.690404690404691e-07,
      "loss": 0.0002,
      "step": 37210
    },
    {
      "epoch": 47.01804375804376,
      "grad_norm": 0.07519324123859406,
      "learning_rate": 4.661804661804662e-07,
      "loss": 0.0,
      "step": 37220
    },
    {
      "epoch": 47.018301158301156,
      "grad_norm": 0.00012404580775182694,
      "learning_rate": 4.6332046332046336e-07,
      "loss": 0.0001,
      "step": 37230
    },
    {
      "epoch": 47.01855855855856,
      "grad_norm": 0.0008826501434668899,
      "learning_rate": 4.604604604604605e-07,
      "loss": 0.0,
      "step": 37240
    },
    {
      "epoch": 47.018815958815956,
      "grad_norm": 0.0012050550431013107,
      "learning_rate": 4.5760045760045767e-07,
      "loss": 0.0001,
      "step": 37250
    },
    {
      "epoch": 47.01907335907336,
      "grad_norm": 6.927676440682262e-05,
      "learning_rate": 4.547404547404548e-07,
      "loss": 0.7524,
      "step": 37260
    },
    {
      "epoch": 47.01933075933076,
      "grad_norm": 0.00022927786631044,
      "learning_rate": 4.518804518804519e-07,
      "loss": 0.2072,
      "step": 37270
    },
    {
      "epoch": 47.01958815958816,
      "grad_norm": 0.0018792460905387998,
      "learning_rate": 4.4902044902044904e-07,
      "loss": 0.0001,
      "step": 37280
    },
    {
      "epoch": 47.01984555984556,
      "grad_norm": 0.003901133546605706,
      "learning_rate": 4.4616044616044625e-07,
      "loss": 0.0,
      "step": 37290
    },
    {
      "epoch": 47.02,
      "eval_accuracy": 0.8913043478260869,
      "eval_loss": 0.7096506357192993,
      "eval_runtime": 13.4018,
      "eval_samples_per_second": 3.432,
      "eval_steps_per_second": 3.432,
      "step": 37296
    },
    {
      "epoch": 48.00010296010296,
      "grad_norm": 0.0013056639581918716,
      "learning_rate": 4.4330044330044335e-07,
      "loss": 0.5536,
      "step": 37300
    },
    {
      "epoch": 48.00036036036036,
      "grad_norm": 8.839034853735939e-05,
      "learning_rate": 4.4044044044044046e-07,
      "loss": 0.0001,
      "step": 37310
    },
    {
      "epoch": 48.00061776061776,
      "grad_norm": 0.02358631044626236,
      "learning_rate": 4.375804375804376e-07,
      "loss": 0.0001,
      "step": 37320
    },
    {
      "epoch": 48.00087516087516,
      "grad_norm": 0.00010084093810291961,
      "learning_rate": 4.347204347204347e-07,
      "loss": 0.7232,
      "step": 37330
    },
    {
      "epoch": 48.001132561132565,
      "grad_norm": 0.0037173698656260967,
      "learning_rate": 4.3186043186043193e-07,
      "loss": 0.0001,
      "step": 37340
    },
    {
      "epoch": 48.00138996138996,
      "grad_norm": 0.0001251946232514456,
      "learning_rate": 4.2900042900042903e-07,
      "loss": 0.0001,
      "step": 37350
    },
    {
      "epoch": 48.001647361647365,
      "grad_norm": 0.01967187225818634,
      "learning_rate": 4.261404261404262e-07,
      "loss": 0.0001,
      "step": 37360
    },
    {
      "epoch": 48.00190476190476,
      "grad_norm": 0.010023032315075397,
      "learning_rate": 4.232804232804233e-07,
      "loss": 0.0,
      "step": 37370
    },
    {
      "epoch": 48.002162162162165,
      "grad_norm": 0.06681618839502335,
      "learning_rate": 4.204204204204205e-07,
      "loss": 0.0008,
      "step": 37380
    },
    {
      "epoch": 48.00241956241956,
      "grad_norm": 0.021821511909365654,
      "learning_rate": 4.175604175604176e-07,
      "loss": 0.0002,
      "step": 37390
    },
    {
      "epoch": 48.002676962676965,
      "grad_norm": 0.0001355089625576511,
      "learning_rate": 4.147004147004147e-07,
      "loss": 0.5061,
      "step": 37400
    },
    {
      "epoch": 48.00293436293436,
      "grad_norm": 0.038224369287490845,
      "learning_rate": 4.1184041184041187e-07,
      "loss": 0.668,
      "step": 37410
    },
    {
      "epoch": 48.003191763191765,
      "grad_norm": 0.00025807946803979576,
      "learning_rate": 4.08980408980409e-07,
      "loss": 0.0,
      "step": 37420
    },
    {
      "epoch": 48.00344916344916,
      "grad_norm": 0.00012352765770629048,
      "learning_rate": 4.061204061204062e-07,
      "loss": 0.0001,
      "step": 37430
    },
    {
      "epoch": 48.003706563706565,
      "grad_norm": 0.021487237885594368,
      "learning_rate": 4.032604032604033e-07,
      "loss": 0.0001,
      "step": 37440
    },
    {
      "epoch": 48.00396396396396,
      "grad_norm": 1613.104736328125,
      "learning_rate": 4.0040040040040045e-07,
      "loss": 0.2208,
      "step": 37450
    },
    {
      "epoch": 48.004221364221365,
      "grad_norm": 0.05000975728034973,
      "learning_rate": 3.9754039754039755e-07,
      "loss": 0.0001,
      "step": 37460
    },
    {
      "epoch": 48.00447876447876,
      "grad_norm": 0.0008557880646549165,
      "learning_rate": 3.9468039468039466e-07,
      "loss": 0.0001,
      "step": 37470
    },
    {
      "epoch": 48.004736164736165,
      "grad_norm": 0.0027655167505145073,
      "learning_rate": 3.9182039182039187e-07,
      "loss": 0.0001,
      "step": 37480
    },
    {
      "epoch": 48.00499356499356,
      "grad_norm": 0.11204931885004044,
      "learning_rate": 3.8896038896038897e-07,
      "loss": 0.0002,
      "step": 37490
    },
    {
      "epoch": 48.005250965250966,
      "grad_norm": 0.08630715310573578,
      "learning_rate": 3.8610038610038613e-07,
      "loss": 0.0004,
      "step": 37500
    },
    {
      "epoch": 48.00550836550836,
      "grad_norm": 0.0018811404006555676,
      "learning_rate": 3.8324038324038323e-07,
      "loss": 0.0,
      "step": 37510
    },
    {
      "epoch": 48.005765765765766,
      "grad_norm": 0.0012397312093526125,
      "learning_rate": 3.8038038038038044e-07,
      "loss": 0.0,
      "step": 37520
    },
    {
      "epoch": 48.00602316602317,
      "grad_norm": 0.002020134124904871,
      "learning_rate": 3.7752037752037755e-07,
      "loss": 0.0001,
      "step": 37530
    },
    {
      "epoch": 48.006280566280566,
      "grad_norm": 0.029110634699463844,
      "learning_rate": 3.746603746603747e-07,
      "loss": 0.0001,
      "step": 37540
    },
    {
      "epoch": 48.00653796653797,
      "grad_norm": 0.0019776870030909777,
      "learning_rate": 3.718003718003718e-07,
      "loss": 0.0003,
      "step": 37550
    },
    {
      "epoch": 48.006795366795366,
      "grad_norm": 0.07283729314804077,
      "learning_rate": 3.6894036894036897e-07,
      "loss": 0.0003,
      "step": 37560
    },
    {
      "epoch": 48.00705276705277,
      "grad_norm": 0.00014232056855689734,
      "learning_rate": 3.660803660803661e-07,
      "loss": 0.0193,
      "step": 37570
    },
    {
      "epoch": 48.007310167310166,
      "grad_norm": 0.17029236257076263,
      "learning_rate": 3.632203632203633e-07,
      "loss": 0.0001,
      "step": 37580
    },
    {
      "epoch": 48.00756756756757,
      "grad_norm": 0.02278299257159233,
      "learning_rate": 3.603603603603604e-07,
      "loss": 0.0001,
      "step": 37590
    },
    {
      "epoch": 48.007824967824966,
      "grad_norm": 0.00011213924881303683,
      "learning_rate": 3.575003575003575e-07,
      "loss": 0.0001,
      "step": 37600
    },
    {
      "epoch": 48.00808236808237,
      "grad_norm": 0.0008133391384035349,
      "learning_rate": 3.5464035464035465e-07,
      "loss": 0.0301,
      "step": 37610
    },
    {
      "epoch": 48.008339768339766,
      "grad_norm": 0.0009331071632914245,
      "learning_rate": 3.517803517803518e-07,
      "loss": 0.0,
      "step": 37620
    },
    {
      "epoch": 48.00859716859717,
      "grad_norm": 8.731058915145695e-05,
      "learning_rate": 3.4892034892034896e-07,
      "loss": 0.0002,
      "step": 37630
    },
    {
      "epoch": 48.00885456885457,
      "grad_norm": 0.011118177324533463,
      "learning_rate": 3.4606034606034607e-07,
      "loss": 0.0,
      "step": 37640
    },
    {
      "epoch": 48.00911196911197,
      "grad_norm": 0.00018614377768244594,
      "learning_rate": 3.4320034320034323e-07,
      "loss": 0.0,
      "step": 37650
    },
    {
      "epoch": 48.00936936936937,
      "grad_norm": 8.32139776321128e-05,
      "learning_rate": 3.403403403403404e-07,
      "loss": 0.0,
      "step": 37660
    },
    {
      "epoch": 48.00962676962677,
      "grad_norm": 0.021617701277136803,
      "learning_rate": 3.3748033748033754e-07,
      "loss": 0.0001,
      "step": 37670
    },
    {
      "epoch": 48.00988416988417,
      "grad_norm": 0.029266489669680595,
      "learning_rate": 3.3462033462033465e-07,
      "loss": 0.0002,
      "step": 37680
    },
    {
      "epoch": 48.01014157014157,
      "grad_norm": 0.0012129413662478328,
      "learning_rate": 3.3176033176033175e-07,
      "loss": 0.0006,
      "step": 37690
    },
    {
      "epoch": 48.01039897039897,
      "grad_norm": 0.005568423308432102,
      "learning_rate": 3.289003289003289e-07,
      "loss": 0.0,
      "step": 37700
    },
    {
      "epoch": 48.01065637065637,
      "grad_norm": 0.0016812310786917806,
      "learning_rate": 3.260403260403261e-07,
      "loss": 0.1846,
      "step": 37710
    },
    {
      "epoch": 48.010913770913774,
      "grad_norm": 0.0011652051471173763,
      "learning_rate": 3.231803231803232e-07,
      "loss": 0.0002,
      "step": 37720
    },
    {
      "epoch": 48.01117117117117,
      "grad_norm": 8.414344483753666e-05,
      "learning_rate": 3.2032032032032033e-07,
      "loss": 0.0001,
      "step": 37730
    },
    {
      "epoch": 48.011428571428574,
      "grad_norm": 0.0011205601040273905,
      "learning_rate": 3.174603174603175e-07,
      "loss": 0.0,
      "step": 37740
    },
    {
      "epoch": 48.01168597168597,
      "grad_norm": 0.0004462198994588107,
      "learning_rate": 3.146003146003146e-07,
      "loss": 0.0001,
      "step": 37750
    },
    {
      "epoch": 48.011943371943374,
      "grad_norm": 0.0015702313976362348,
      "learning_rate": 3.1174031174031175e-07,
      "loss": 0.0,
      "step": 37760
    },
    {
      "epoch": 48.01220077220077,
      "grad_norm": 0.00011221476597711444,
      "learning_rate": 3.088803088803089e-07,
      "loss": 0.8953,
      "step": 37770
    },
    {
      "epoch": 48.012458172458174,
      "grad_norm": 0.0006436722469516098,
      "learning_rate": 3.0602030602030606e-07,
      "loss": 0.0009,
      "step": 37780
    },
    {
      "epoch": 48.01271557271557,
      "grad_norm": 3.6860032081604004,
      "learning_rate": 3.031603031603032e-07,
      "loss": 0.4928,
      "step": 37790
    },
    {
      "epoch": 48.012972972972975,
      "grad_norm": 0.0011715868022292852,
      "learning_rate": 3.003003003003003e-07,
      "loss": 0.0,
      "step": 37800
    },
    {
      "epoch": 48.01323037323037,
      "grad_norm": 339.0137634277344,
      "learning_rate": 2.974402974402975e-07,
      "loss": 0.7127,
      "step": 37810
    },
    {
      "epoch": 48.013487773487775,
      "grad_norm": 0.00025066794478334486,
      "learning_rate": 2.945802945802946e-07,
      "loss": 0.0,
      "step": 37820
    },
    {
      "epoch": 48.01374517374517,
      "grad_norm": 0.0003706521529238671,
      "learning_rate": 2.9172029172029174e-07,
      "loss": 0.0,
      "step": 37830
    },
    {
      "epoch": 48.014002574002575,
      "grad_norm": 0.00023929473536554724,
      "learning_rate": 2.888602888602889e-07,
      "loss": 0.0,
      "step": 37840
    },
    {
      "epoch": 48.01425997425997,
      "grad_norm": 0.0029671115335077047,
      "learning_rate": 2.86000286000286e-07,
      "loss": 0.0001,
      "step": 37850
    },
    {
      "epoch": 48.014517374517375,
      "grad_norm": 0.03844938799738884,
      "learning_rate": 2.8314028314028316e-07,
      "loss": 0.0001,
      "step": 37860
    },
    {
      "epoch": 48.01477477477477,
      "grad_norm": 3.0009055137634277,
      "learning_rate": 2.802802802802803e-07,
      "loss": 0.0005,
      "step": 37870
    },
    {
      "epoch": 48.015032175032175,
      "grad_norm": 0.028874320909380913,
      "learning_rate": 2.774202774202774e-07,
      "loss": 0.0001,
      "step": 37880
    },
    {
      "epoch": 48.01528957528958,
      "grad_norm": 0.011996056884527206,
      "learning_rate": 2.745602745602746e-07,
      "loss": 0.0,
      "step": 37890
    },
    {
      "epoch": 48.015546975546975,
      "grad_norm": 0.0015611989656463265,
      "learning_rate": 2.717002717002717e-07,
      "loss": 0.0096,
      "step": 37900
    },
    {
      "epoch": 48.01580437580438,
      "grad_norm": 0.0020811811555176973,
      "learning_rate": 2.688402688402689e-07,
      "loss": 0.9104,
      "step": 37910
    },
    {
      "epoch": 48.016061776061775,
      "grad_norm": 0.007127484772354364,
      "learning_rate": 2.65980265980266e-07,
      "loss": 0.0001,
      "step": 37920
    },
    {
      "epoch": 48.01631917631918,
      "grad_norm": 0.0019268447067588568,
      "learning_rate": 2.6312026312026316e-07,
      "loss": 0.0002,
      "step": 37930
    },
    {
      "epoch": 48.016576576576576,
      "grad_norm": 9.653266897657886e-05,
      "learning_rate": 2.6026026026026026e-07,
      "loss": 0.0,
      "step": 37940
    },
    {
      "epoch": 48.01683397683398,
      "grad_norm": 0.00013772105739917606,
      "learning_rate": 2.574002574002574e-07,
      "loss": 0.0001,
      "step": 37950
    },
    {
      "epoch": 48.017091377091376,
      "grad_norm": 0.0012763019185513258,
      "learning_rate": 2.545402545402546e-07,
      "loss": 0.0,
      "step": 37960
    },
    {
      "epoch": 48.01734877734878,
      "grad_norm": 0.0019373123068362474,
      "learning_rate": 2.516802516802517e-07,
      "loss": 0.0,
      "step": 37970
    },
    {
      "epoch": 48.017606177606176,
      "grad_norm": 0.0032342688646167517,
      "learning_rate": 2.4882024882024884e-07,
      "loss": 0.0,
      "step": 37980
    },
    {
      "epoch": 48.01786357786358,
      "grad_norm": 0.008433103561401367,
      "learning_rate": 2.45960245960246e-07,
      "loss": 0.0001,
      "step": 37990
    },
    {
      "epoch": 48.018120978120976,
      "grad_norm": 8.630425145383924e-05,
      "learning_rate": 2.4310024310024315e-07,
      "loss": 0.0,
      "step": 38000
    },
    {
      "epoch": 48.01837837837838,
      "grad_norm": 0.00015035143587738276,
      "learning_rate": 2.4024024024024026e-07,
      "loss": 0.966,
      "step": 38010
    },
    {
      "epoch": 48.018635778635776,
      "grad_norm": 0.0001188807946164161,
      "learning_rate": 2.373802373802374e-07,
      "loss": 0.0001,
      "step": 38020
    },
    {
      "epoch": 48.01889317889318,
      "grad_norm": 8.999750571092591e-05,
      "learning_rate": 2.3452023452023455e-07,
      "loss": 0.0,
      "step": 38030
    },
    {
      "epoch": 48.019150579150576,
      "grad_norm": 0.008205074816942215,
      "learning_rate": 2.3166023166023168e-07,
      "loss": 0.8337,
      "step": 38040
    },
    {
      "epoch": 48.01940797940798,
      "grad_norm": 3213.668701171875,
      "learning_rate": 2.2880022880022884e-07,
      "loss": 0.1913,
      "step": 38050
    },
    {
      "epoch": 48.019665379665376,
      "grad_norm": 0.00011710776743711904,
      "learning_rate": 2.2594022594022594e-07,
      "loss": 0.0001,
      "step": 38060
    },
    {
      "epoch": 48.01992277992278,
      "grad_norm": 8.933376375352964e-05,
      "learning_rate": 2.2308022308022312e-07,
      "loss": 0.6658,
      "step": 38070
    },
    {
      "epoch": 48.02,
      "eval_accuracy": 0.8913043478260869,
      "eval_loss": 0.831529438495636,
      "eval_runtime": 13.4096,
      "eval_samples_per_second": 3.43,
      "eval_steps_per_second": 3.43,
      "step": 38073
    },
    {
      "epoch": 49.00018018018018,
      "grad_norm": 0.07369162142276764,
      "learning_rate": 2.2022022022022023e-07,
      "loss": 0.0,
      "step": 38080
    },
    {
      "epoch": 49.00043758043758,
      "grad_norm": 0.5429767370223999,
      "learning_rate": 2.1736021736021736e-07,
      "loss": 0.6264,
      "step": 38090
    },
    {
      "epoch": 49.00069498069498,
      "grad_norm": 0.00045454761129803956,
      "learning_rate": 2.1450021450021452e-07,
      "loss": 0.0005,
      "step": 38100
    },
    {
      "epoch": 49.000952380952384,
      "grad_norm": 0.0005570737412199378,
      "learning_rate": 2.1164021164021165e-07,
      "loss": 0.0002,
      "step": 38110
    },
    {
      "epoch": 49.00120978120978,
      "grad_norm": 9.545540524413809e-05,
      "learning_rate": 2.087802087802088e-07,
      "loss": 0.0001,
      "step": 38120
    },
    {
      "epoch": 49.001467181467184,
      "grad_norm": 0.0019436365691944957,
      "learning_rate": 2.0592020592020594e-07,
      "loss": 0.0,
      "step": 38130
    },
    {
      "epoch": 49.00172458172458,
      "grad_norm": 0.020056238397955894,
      "learning_rate": 2.030602030602031e-07,
      "loss": 0.0001,
      "step": 38140
    },
    {
      "epoch": 49.001981981981984,
      "grad_norm": 0.020614050328731537,
      "learning_rate": 2.0020020020020022e-07,
      "loss": 0.8467,
      "step": 38150
    },
    {
      "epoch": 49.00223938223938,
      "grad_norm": 0.0006698184879496694,
      "learning_rate": 1.9734019734019733e-07,
      "loss": 0.0,
      "step": 38160
    },
    {
      "epoch": 49.002496782496785,
      "grad_norm": 0.0013321598526090384,
      "learning_rate": 1.9448019448019449e-07,
      "loss": 0.4252,
      "step": 38170
    },
    {
      "epoch": 49.00275418275418,
      "grad_norm": 0.03110680542886257,
      "learning_rate": 1.9162019162019162e-07,
      "loss": 0.0,
      "step": 38180
    },
    {
      "epoch": 49.003011583011585,
      "grad_norm": 0.0001595134090166539,
      "learning_rate": 1.8876018876018877e-07,
      "loss": 0.0001,
      "step": 38190
    },
    {
      "epoch": 49.00326898326898,
      "grad_norm": 0.002772995037958026,
      "learning_rate": 1.859001859001859e-07,
      "loss": 0.0,
      "step": 38200
    },
    {
      "epoch": 49.003526383526385,
      "grad_norm": 0.00313870538957417,
      "learning_rate": 1.8304018304018306e-07,
      "loss": 0.0001,
      "step": 38210
    },
    {
      "epoch": 49.00378378378378,
      "grad_norm": 0.00930901151150465,
      "learning_rate": 1.801801801801802e-07,
      "loss": 0.0002,
      "step": 38220
    },
    {
      "epoch": 49.004041184041185,
      "grad_norm": 0.00011457037180662155,
      "learning_rate": 1.7732017732017732e-07,
      "loss": 0.0001,
      "step": 38230
    },
    {
      "epoch": 49.00429858429858,
      "grad_norm": 0.012397563084959984,
      "learning_rate": 1.7446017446017448e-07,
      "loss": 0.0001,
      "step": 38240
    },
    {
      "epoch": 49.004555984555985,
      "grad_norm": 0.011724770069122314,
      "learning_rate": 1.7160017160017161e-07,
      "loss": 0.0001,
      "step": 38250
    },
    {
      "epoch": 49.00481338481338,
      "grad_norm": 7.894120790297166e-05,
      "learning_rate": 1.6874016874016877e-07,
      "loss": 0.0001,
      "step": 38260
    },
    {
      "epoch": 49.005070785070785,
      "grad_norm": 0.004039763938635588,
      "learning_rate": 1.6588016588016588e-07,
      "loss": 0.0,
      "step": 38270
    },
    {
      "epoch": 49.00532818532819,
      "grad_norm": 0.0477052703499794,
      "learning_rate": 1.6302016302016306e-07,
      "loss": 0.0001,
      "step": 38280
    },
    {
      "epoch": 49.005585585585585,
      "grad_norm": 0.010268225334584713,
      "learning_rate": 1.6016016016016016e-07,
      "loss": 0.0001,
      "step": 38290
    },
    {
      "epoch": 49.00584298584299,
      "grad_norm": 0.03354515880346298,
      "learning_rate": 1.573001573001573e-07,
      "loss": 0.0001,
      "step": 38300
    },
    {
      "epoch": 49.006100386100385,
      "grad_norm": 0.00013072919682599604,
      "learning_rate": 1.5444015444015445e-07,
      "loss": 0.0,
      "step": 38310
    },
    {
      "epoch": 49.00635778635779,
      "grad_norm": 0.00982525385916233,
      "learning_rate": 1.515801515801516e-07,
      "loss": 1.4921,
      "step": 38320
    },
    {
      "epoch": 49.006615186615186,
      "grad_norm": 0.01039874367415905,
      "learning_rate": 1.4872014872014874e-07,
      "loss": 0.0,
      "step": 38330
    },
    {
      "epoch": 49.00687258687259,
      "grad_norm": 0.0014984187437221408,
      "learning_rate": 1.4586014586014587e-07,
      "loss": 0.0,
      "step": 38340
    },
    {
      "epoch": 49.007129987129986,
      "grad_norm": 0.00010380911407992244,
      "learning_rate": 1.43000143000143e-07,
      "loss": 0.0,
      "step": 38350
    },
    {
      "epoch": 49.00738738738739,
      "grad_norm": 0.00010410812683403492,
      "learning_rate": 1.4014014014014016e-07,
      "loss": 0.0,
      "step": 38360
    },
    {
      "epoch": 49.007644787644786,
      "grad_norm": 9.377163951285183e-05,
      "learning_rate": 1.372801372801373e-07,
      "loss": 0.0,
      "step": 38370
    },
    {
      "epoch": 49.00790218790219,
      "grad_norm": 0.0019170086598023772,
      "learning_rate": 1.3442013442013445e-07,
      "loss": 0.0001,
      "step": 38380
    },
    {
      "epoch": 49.008159588159586,
      "grad_norm": 0.00012893872917629778,
      "learning_rate": 1.3156013156013158e-07,
      "loss": 1.0042,
      "step": 38390
    },
    {
      "epoch": 49.00841698841699,
      "grad_norm": 7.869133696658537e-05,
      "learning_rate": 1.287001287001287e-07,
      "loss": 0.0008,
      "step": 38400
    },
    {
      "epoch": 49.008674388674386,
      "grad_norm": 0.00012004271411569789,
      "learning_rate": 1.2584012584012584e-07,
      "loss": 0.0,
      "step": 38410
    },
    {
      "epoch": 49.00893178893179,
      "grad_norm": 0.0014769094996154308,
      "learning_rate": 1.22980122980123e-07,
      "loss": 0.0009,
      "step": 38420
    },
    {
      "epoch": 49.009189189189186,
      "grad_norm": 0.003862373996526003,
      "learning_rate": 1.2012012012012013e-07,
      "loss": 0.0001,
      "step": 38430
    },
    {
      "epoch": 49.00944658944659,
      "grad_norm": 0.0006913309334777296,
      "learning_rate": 1.1726011726011727e-07,
      "loss": 0.0003,
      "step": 38440
    },
    {
      "epoch": 49.009703989703986,
      "grad_norm": 0.0021493281237781048,
      "learning_rate": 1.1440011440011442e-07,
      "loss": 0.0004,
      "step": 38450
    },
    {
      "epoch": 49.00996138996139,
      "grad_norm": 0.00039128007483668625,
      "learning_rate": 1.1154011154011156e-07,
      "loss": 0.0001,
      "step": 38460
    },
    {
      "epoch": 49.010218790218794,
      "grad_norm": 0.003142331726849079,
      "learning_rate": 1.0868010868010868e-07,
      "loss": 0.0002,
      "step": 38470
    },
    {
      "epoch": 49.01047619047619,
      "grad_norm": 0.0017112774075940251,
      "learning_rate": 1.0582010582010582e-07,
      "loss": 0.0,
      "step": 38480
    },
    {
      "epoch": 49.010733590733594,
      "grad_norm": 0.0016064579831436276,
      "learning_rate": 1.0296010296010297e-07,
      "loss": 0.0,
      "step": 38490
    },
    {
      "epoch": 49.01099099099099,
      "grad_norm": 8.536849054507911e-05,
      "learning_rate": 1.0010010010010011e-07,
      "loss": 0.851,
      "step": 38500
    },
    {
      "epoch": 49.011248391248394,
      "grad_norm": 0.000131205641082488,
      "learning_rate": 9.724009724009724e-08,
      "loss": 0.5012,
      "step": 38510
    },
    {
      "epoch": 49.01150579150579,
      "grad_norm": 0.00011044665006920695,
      "learning_rate": 9.438009438009439e-08,
      "loss": 0.0,
      "step": 38520
    },
    {
      "epoch": 49.011763191763194,
      "grad_norm": 0.0001505005348008126,
      "learning_rate": 9.152009152009153e-08,
      "loss": 0.0,
      "step": 38530
    },
    {
      "epoch": 49.01202059202059,
      "grad_norm": 0.0006344806170091033,
      "learning_rate": 8.866008866008866e-08,
      "loss": 0.0001,
      "step": 38540
    },
    {
      "epoch": 49.012277992277994,
      "grad_norm": 0.01434845756739378,
      "learning_rate": 8.580008580008581e-08,
      "loss": 0.6324,
      "step": 38550
    },
    {
      "epoch": 49.01253539253539,
      "grad_norm": 0.0017600515857338905,
      "learning_rate": 8.294008294008294e-08,
      "loss": 0.0009,
      "step": 38560
    },
    {
      "epoch": 49.012792792792794,
      "grad_norm": 0.0006000173161737621,
      "learning_rate": 8.008008008008008e-08,
      "loss": 0.0,
      "step": 38570
    },
    {
      "epoch": 49.01305019305019,
      "grad_norm": 0.0001577949442435056,
      "learning_rate": 7.722007722007723e-08,
      "loss": 0.0001,
      "step": 38580
    },
    {
      "epoch": 49.013307593307594,
      "grad_norm": 0.0021368926391005516,
      "learning_rate": 7.436007436007437e-08,
      "loss": 0.0002,
      "step": 38590
    },
    {
      "epoch": 49.01356499356499,
      "grad_norm": 0.02230561152100563,
      "learning_rate": 7.15000715000715e-08,
      "loss": 0.0001,
      "step": 38600
    },
    {
      "epoch": 49.013822393822394,
      "grad_norm": 0.00016711508214939386,
      "learning_rate": 6.864006864006865e-08,
      "loss": 0.8394,
      "step": 38610
    },
    {
      "epoch": 49.01407979407979,
      "grad_norm": 0.004358669742941856,
      "learning_rate": 6.578006578006579e-08,
      "loss": 0.0003,
      "step": 38620
    },
    {
      "epoch": 49.014337194337195,
      "grad_norm": 0.0006673623574897647,
      "learning_rate": 6.292006292006292e-08,
      "loss": 0.0,
      "step": 38630
    },
    {
      "epoch": 49.01459459459459,
      "grad_norm": 605.3671264648438,
      "learning_rate": 6.006006006006006e-08,
      "loss": 0.089,
      "step": 38640
    },
    {
      "epoch": 49.014851994851995,
      "grad_norm": 0.0001498550409451127,
      "learning_rate": 5.720005720005721e-08,
      "loss": 0.65,
      "step": 38650
    },
    {
      "epoch": 49.0151093951094,
      "grad_norm": 0.0008860075031407177,
      "learning_rate": 5.434005434005434e-08,
      "loss": 0.0001,
      "step": 38660
    },
    {
      "epoch": 49.015366795366795,
      "grad_norm": 0.0030720012728124857,
      "learning_rate": 5.1480051480051484e-08,
      "loss": 0.0,
      "step": 38670
    },
    {
      "epoch": 49.0156241956242,
      "grad_norm": 0.0020494323689490557,
      "learning_rate": 4.862004862004862e-08,
      "loss": 0.0001,
      "step": 38680
    },
    {
      "epoch": 49.015881595881595,
      "grad_norm": 0.0007796235731802881,
      "learning_rate": 4.5760045760045766e-08,
      "loss": 0.0001,
      "step": 38690
    },
    {
      "epoch": 49.016138996139,
      "grad_norm": 0.000528481206856668,
      "learning_rate": 4.2900042900042903e-08,
      "loss": 0.0002,
      "step": 38700
    },
    {
      "epoch": 49.016396396396395,
      "grad_norm": 0.0076874056831002235,
      "learning_rate": 4.004004004004004e-08,
      "loss": 0.0001,
      "step": 38710
    },
    {
      "epoch": 49.0166537966538,
      "grad_norm": 0.00017405615653842688,
      "learning_rate": 3.7180037180037185e-08,
      "loss": 0.7798,
      "step": 38720
    },
    {
      "epoch": 49.016911196911195,
      "grad_norm": 0.002475382061675191,
      "learning_rate": 3.432003432003432e-08,
      "loss": 0.0,
      "step": 38730
    },
    {
      "epoch": 49.0171685971686,
      "grad_norm": 0.0001674262312008068,
      "learning_rate": 3.146003146003146e-08,
      "loss": 0.7042,
      "step": 38740
    },
    {
      "epoch": 49.017425997425995,
      "grad_norm": 0.0012000445276498795,
      "learning_rate": 2.8600028600028604e-08,
      "loss": 0.0001,
      "step": 38750
    },
    {
      "epoch": 49.0176833976834,
      "grad_norm": 0.00012928430805914104,
      "learning_rate": 2.5740025740025742e-08,
      "loss": 0.0,
      "step": 38760
    },
    {
      "epoch": 49.017940797940796,
      "grad_norm": 0.0012572959531098604,
      "learning_rate": 2.2880022880022883e-08,
      "loss": 0.0001,
      "step": 38770
    },
    {
      "epoch": 49.0181981981982,
      "grad_norm": 0.00019746992620639503,
      "learning_rate": 2.002002002002002e-08,
      "loss": 0.0018,
      "step": 38780
    },
    {
      "epoch": 49.018455598455596,
      "grad_norm": 0.020095746964216232,
      "learning_rate": 1.716001716001716e-08,
      "loss": 0.0001,
      "step": 38790
    },
    {
      "epoch": 49.018712998713,
      "grad_norm": 9.711348684504628e-05,
      "learning_rate": 1.4300014300014302e-08,
      "loss": 0.0002,
      "step": 38800
    },
    {
      "epoch": 49.018970398970396,
      "grad_norm": 0.009967109188437462,
      "learning_rate": 1.1440011440011441e-08,
      "loss": 0.0007,
      "step": 38810
    },
    {
      "epoch": 49.0192277992278,
      "grad_norm": 0.0005083610885776579,
      "learning_rate": 8.58000858000858e-09,
      "loss": 0.0001,
      "step": 38820
    },
    {
      "epoch": 49.0194851994852,
      "grad_norm": 0.00010245986777590588,
      "learning_rate": 5.720005720005721e-09,
      "loss": 0.0001,
      "step": 38830
    },
    {
      "epoch": 49.0197425997426,
      "grad_norm": 0.014355776831507683,
      "learning_rate": 2.8600028600028604e-09,
      "loss": 0.0001,
      "step": 38840
    },
    {
      "epoch": 49.02,
      "grad_norm": 0.00027233664877712727,
      "learning_rate": 0.0,
      "loss": 0.0,
      "step": 38850
    },
    {
      "epoch": 49.02,
      "eval_accuracy": 0.8913043478260869,
      "eval_loss": 0.857639491558075,
      "eval_runtime": 15.9664,
      "eval_samples_per_second": 2.881,
      "eval_steps_per_second": 2.881,
      "step": 38850
    },
    {
      "epoch": 49.02,
      "step": 38850,
      "total_flos": 1.7059137234927944e+20,
      "train_loss": 0.625811186066474,
      "train_runtime": 28909.5385,
      "train_samples_per_second": 1.344,
      "train_steps_per_second": 1.344
    },
    {
      "epoch": 49.02,
      "eval_accuracy": 0.9130434782608695,
      "eval_loss": 0.42057543992996216,
      "eval_runtime": 13.3636,
      "eval_samples_per_second": 3.442,
      "eval_steps_per_second": 3.442,
      "step": 38850
    },
    {
      "epoch": 49.02,
      "eval_accuracy": 0.9130434782608695,
      "eval_loss": 0.42057543992996216,
      "eval_runtime": 13.3586,
      "eval_samples_per_second": 3.443,
      "eval_steps_per_second": 3.443,
      "step": 38850
    }
  ],
  "logging_steps": 10,
  "max_steps": 38850,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 9223372036854775807,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 1.7059137234927944e+20,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}