diff --git "a/checkpoint-9990/trainer_state.json" "b/checkpoint-9990/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-9990/trainer_state.json"
@@ -0,0 +1,7122 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.7716073221595736,
+  "eval_steps": 1000,
+  "global_step": 9990,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0,
+      "eval_loss": 7.565340518951416,
+      "eval_runtime": 1.2901,
+      "eval_samples_per_second": 9.302,
+      "eval_steps_per_second": 1.55,
+      "step": 0
+    },
+    {
+      "epoch": 0.0007723797018614351,
+      "grad_norm": 2.703125,
+      "learning_rate": 1.2853470437017995e-06,
+      "loss": 7.3877,
+      "step": 10
+    },
+    {
+      "epoch": 0.0015447594037228702,
+      "grad_norm": 3.203125,
+      "learning_rate": 2.570694087403599e-06,
+      "loss": 7.4036,
+      "step": 20
+    },
+    {
+      "epoch": 0.0023171391055843054,
+      "grad_norm": 3.59375,
+      "learning_rate": 3.856041131105398e-06,
+      "loss": 7.3948,
+      "step": 30
+    },
+    {
+      "epoch": 0.0030895188074457403,
+      "grad_norm": 2.96875,
+      "learning_rate": 5.141388174807198e-06,
+      "loss": 7.3935,
+      "step": 40
+    },
+    {
+      "epoch": 0.0038618985093071753,
+      "grad_norm": 1.859375,
+      "learning_rate": 6.426735218508998e-06,
+      "loss": 7.3788,
+      "step": 50
+    },
+    {
+      "epoch": 0.004634278211168611,
+      "grad_norm": 1.3515625,
+      "learning_rate": 7.712082262210796e-06,
+      "loss": 7.3564,
+      "step": 60
+    },
+    {
+      "epoch": 0.005406657913030045,
+      "grad_norm": 1.296875,
+      "learning_rate": 8.997429305912597e-06,
+      "loss": 7.341,
+      "step": 70
+    },
+    {
+      "epoch": 0.006179037614891481,
+      "grad_norm": 1.21875,
+      "learning_rate": 1.0282776349614396e-05,
+      "loss": 7.3153,
+      "step": 80
+    },
+    {
+      "epoch": 0.006951417316752916,
+      "grad_norm": 1.2109375,
+      "learning_rate": 1.1568123393316196e-05,
+      "loss": 7.2943,
+      "step": 90
+    },
+    {
+      "epoch": 0.007723797018614351,
+      "grad_norm": 1.140625,
+      "learning_rate": 1.2853470437017995e-05,
+      "loss": 7.2795,
+      "step": 100
+    },
+    {
+      "epoch": 0.008496176720475786,
+      "grad_norm": 1.2421875,
+      "learning_rate": 1.4138817480719796e-05,
+      "loss": 7.2647,
+      "step": 110
+    },
+    {
+      "epoch": 0.009268556422337221,
+      "grad_norm": 1.1328125,
+      "learning_rate": 1.5424164524421593e-05,
+      "loss": 7.2653,
+      "step": 120
+    },
+    {
+      "epoch": 0.010040936124198657,
+      "grad_norm": 1.1953125,
+      "learning_rate": 1.6709511568123395e-05,
+      "loss": 7.2383,
+      "step": 130
+    },
+    {
+      "epoch": 0.01081331582606009,
+      "grad_norm": 1.125,
+      "learning_rate": 1.7994858611825194e-05,
+      "loss": 7.2339,
+      "step": 140
+    },
+    {
+      "epoch": 0.011585695527921526,
+      "grad_norm": 1.2734375,
+      "learning_rate": 1.9280205655526993e-05,
+      "loss": 7.2323,
+      "step": 150
+    },
+    {
+      "epoch": 0.012358075229782961,
+      "grad_norm": 1.0546875,
+      "learning_rate": 2.0565552699228792e-05,
+      "loss": 7.2395,
+      "step": 160
+    },
+    {
+      "epoch": 0.013130454931644397,
+      "grad_norm": 1.140625,
+      "learning_rate": 2.185089974293059e-05,
+      "loss": 7.2193,
+      "step": 170
+    },
+    {
+      "epoch": 0.013902834633505832,
+      "grad_norm": 1.0234375,
+      "learning_rate": 2.3136246786632393e-05,
+      "loss": 7.2236,
+      "step": 180
+    },
+    {
+      "epoch": 0.014675214335367266,
+      "grad_norm": 1.0703125,
+      "learning_rate": 2.4421593830334192e-05,
+      "loss": 7.2195,
+      "step": 190
+    },
+    {
+      "epoch": 0.015447594037228701,
+      "grad_norm": 1.1015625,
+      "learning_rate": 2.570694087403599e-05,
+      "loss": 7.2041,
+      "step": 200
+    },
+    {
+      "epoch": 0.01621997373909014,
+      "grad_norm": 1.0859375,
+      "learning_rate": 2.699228791773779e-05,
+      "loss": 7.1884,
+      "step": 210
+    },
+    {
+      "epoch": 0.016992353440951572,
+      "grad_norm": 1.0625,
+      "learning_rate": 2.827763496143959e-05,
+      "loss": 7.2075,
+      "step": 220
+    },
+    {
+      "epoch": 0.017764733142813006,
+      "grad_norm": 1.0625,
+      "learning_rate": 2.9562982005141387e-05,
+      "loss": 7.1882,
+      "step": 230
+    },
+    {
+      "epoch": 0.018537112844674443,
+      "grad_norm": 1.1640625,
+      "learning_rate": 3.0848329048843186e-05,
+      "loss": 7.1827,
+      "step": 240
+    },
+    {
+      "epoch": 0.019309492546535877,
+      "grad_norm": 1.09375,
+      "learning_rate": 3.213367609254499e-05,
+      "loss": 7.1755,
+      "step": 250
+    },
+    {
+      "epoch": 0.020081872248397314,
+      "grad_norm": 1.1015625,
+      "learning_rate": 3.341902313624679e-05,
+      "loss": 7.1796,
+      "step": 260
+    },
+    {
+      "epoch": 0.020854251950258747,
+      "grad_norm": 1.09375,
+      "learning_rate": 3.470437017994859e-05,
+      "loss": 7.1768,
+      "step": 270
+    },
+    {
+      "epoch": 0.02162663165212018,
+      "grad_norm": 1.078125,
+      "learning_rate": 3.598971722365039e-05,
+      "loss": 7.1824,
+      "step": 280
+    },
+    {
+      "epoch": 0.022399011353981618,
+      "grad_norm": 1.0546875,
+      "learning_rate": 3.727506426735219e-05,
+      "loss": 7.1571,
+      "step": 290
+    },
+    {
+      "epoch": 0.023171391055843052,
+      "grad_norm": 1.2890625,
+      "learning_rate": 3.8560411311053986e-05,
+      "loss": 7.1409,
+      "step": 300
+    },
+    {
+      "epoch": 0.02394377075770449,
+      "grad_norm": 1.171875,
+      "learning_rate": 3.9845758354755785e-05,
+      "loss": 7.1577,
+      "step": 310
+    },
+    {
+      "epoch": 0.024716150459565923,
+      "grad_norm": 1.0859375,
+      "learning_rate": 4.1131105398457584e-05,
+      "loss": 7.1455,
+      "step": 320
+    },
+    {
+      "epoch": 0.025488530161427356,
+      "grad_norm": 1.1015625,
+      "learning_rate": 4.241645244215938e-05,
+      "loss": 7.1469,
+      "step": 330
+    },
+    {
+      "epoch": 0.026260909863288794,
+      "grad_norm": 1.1015625,
+      "learning_rate": 4.370179948586118e-05,
+      "loss": 7.1404,
+      "step": 340
+    },
+    {
+      "epoch": 0.027033289565150227,
+      "grad_norm": 1.0703125,
+      "learning_rate": 4.498714652956299e-05,
+      "loss": 7.1291,
+      "step": 350
+    },
+    {
+      "epoch": 0.027805669267011664,
+      "grad_norm": 1.125,
+      "learning_rate": 4.6272493573264786e-05,
+      "loss": 7.1399,
+      "step": 360
+    },
+    {
+      "epoch": 0.028578048968873098,
+      "grad_norm": 1.09375,
+      "learning_rate": 4.7557840616966585e-05,
+      "loss": 7.1168,
+      "step": 370
+    },
+    {
+      "epoch": 0.029350428670734532,
+      "grad_norm": 1.1015625,
+      "learning_rate": 4.8843187660668383e-05,
+      "loss": 7.1164,
+      "step": 380
+    },
+    {
+      "epoch": 0.03012280837259597,
+      "grad_norm": 1.1171875,
+      "learning_rate": 4.999999921770817e-05,
+      "loss": 7.1345,
+      "step": 390
+    },
+    {
+      "epoch": 0.030895188074457403,
+      "grad_norm": 1.1015625,
+      "learning_rate": 4.999990534274693e-05,
+      "loss": 7.1199,
+      "step": 400
+    },
+    {
+      "epoch": 0.031667567776318836,
+      "grad_norm": 1.1171875,
+      "learning_rate": 4.999965501009142e-05,
+      "loss": 7.091,
+      "step": 410
+    },
+    {
+      "epoch": 0.03243994747818028,
+      "grad_norm": 1.0625,
+      "learning_rate": 4.999924822130828e-05,
+      "loss": 7.0851,
+      "step": 420
+    },
+    {
+      "epoch": 0.03321232718004171,
+      "grad_norm": 1.125,
+      "learning_rate": 4.9998684978943356e-05,
+      "loss": 7.0958,
+      "step": 430
+    },
+    {
+      "epoch": 0.033984706881903144,
+      "grad_norm": 1.109375,
+      "learning_rate": 4.9997965286521595e-05,
+      "loss": 7.0915,
+      "step": 440
+    },
+    {
+      "epoch": 0.03475708658376458,
+      "grad_norm": 1.125,
+      "learning_rate": 4.999708914854706e-05,
+      "loss": 7.0837,
+      "step": 450
+    },
+    {
+      "epoch": 0.03552946628562601,
+      "grad_norm": 1.078125,
+      "learning_rate": 4.999605657050293e-05,
+      "loss": 7.0776,
+      "step": 460
+    },
+    {
+      "epoch": 0.03630184598748745,
+      "grad_norm": 1.125,
+      "learning_rate": 4.999486755885141e-05,
+      "loss": 7.1053,
+      "step": 470
+    },
+    {
+      "epoch": 0.037074225689348886,
+      "grad_norm": 1.140625,
+      "learning_rate": 4.999352212103373e-05,
+      "loss": 7.0883,
+      "step": 480
+    },
+    {
+      "epoch": 0.03784660539121032,
+      "grad_norm": 1.0546875,
+      "learning_rate": 4.999202026547009e-05,
+      "loss": 7.0903,
+      "step": 490
+    },
+    {
+      "epoch": 0.03861898509307175,
+      "grad_norm": 1.0390625,
+      "learning_rate": 4.9990362001559595e-05,
+      "loss": 7.0613,
+      "step": 500
+    },
+    {
+      "epoch": 0.03939136479493319,
+      "grad_norm": 1.109375,
+      "learning_rate": 4.9988547339680215e-05,
+      "loss": 7.0579,
+      "step": 510
+    },
+    {
+      "epoch": 0.04016374449679463,
+      "grad_norm": 1.140625,
+      "learning_rate": 4.9986576291188705e-05,
+      "loss": 7.0879,
+      "step": 520
+    },
+    {
+      "epoch": 0.04093612419865606,
+      "grad_norm": 1.0859375,
+      "learning_rate": 4.998444886842053e-05,
+      "loss": 7.0837,
+      "step": 530
+    },
+    {
+      "epoch": 0.041708503900517495,
+      "grad_norm": 1.109375,
+      "learning_rate": 4.998216508468982e-05,
+      "loss": 7.0301,
+      "step": 540
+    },
+    {
+      "epoch": 0.04248088360237893,
+      "grad_norm": 1.203125,
+      "learning_rate": 4.9979724954289244e-05,
+      "loss": 7.0722,
+      "step": 550
+    },
+    {
+      "epoch": 0.04325326330424036,
+      "grad_norm": 1.0625,
+      "learning_rate": 4.997712849248994e-05,
+      "loss": 7.0718,
+      "step": 560
+    },
+    {
+      "epoch": 0.0440256430061018,
+      "grad_norm": 1.0859375,
+      "learning_rate": 4.9974375715541455e-05,
+      "loss": 7.0687,
+      "step": 570
+    },
+    {
+      "epoch": 0.044798022707963236,
+      "grad_norm": 1.09375,
+      "learning_rate": 4.9971466640671545e-05,
+      "loss": 7.0742,
+      "step": 580
+    },
+    {
+      "epoch": 0.04557040240982467,
+      "grad_norm": 1.09375,
+      "learning_rate": 4.996840128608618e-05,
+      "loss": 7.0538,
+      "step": 590
+    },
+    {
+      "epoch": 0.046342782111686104,
+      "grad_norm": 1.0625,
+      "learning_rate": 4.9965179670969366e-05,
+      "loss": 7.0351,
+      "step": 600
+    },
+    {
+      "epoch": 0.04711516181354754,
+      "grad_norm": 1.0703125,
+      "learning_rate": 4.996180181548303e-05,
+      "loss": 7.0381,
+      "step": 610
+    },
+    {
+      "epoch": 0.04788754151540898,
+      "grad_norm": 1.140625,
+      "learning_rate": 4.995826774076693e-05,
+      "loss": 7.0398,
+      "step": 620
+    },
+    {
+      "epoch": 0.04865992121727041,
+      "grad_norm": 1.140625,
+      "learning_rate": 4.9954577468938454e-05,
+      "loss": 7.0413,
+      "step": 630
+    },
+    {
+      "epoch": 0.049432300919131845,
+      "grad_norm": 1.1015625,
+      "learning_rate": 4.9950731023092565e-05,
+      "loss": 7.0511,
+      "step": 640
+    },
+    {
+      "epoch": 0.05020468062099328,
+      "grad_norm": 1.0703125,
+      "learning_rate": 4.994672842730159e-05,
+      "loss": 7.0258,
+      "step": 650
+    },
+    {
+      "epoch": 0.05097706032285471,
+      "grad_norm": 1.0546875,
+      "learning_rate": 4.994256970661509e-05,
+      "loss": 7.0258,
+      "step": 660
+    },
+    {
+      "epoch": 0.05174944002471615,
+      "grad_norm": 1.0703125,
+      "learning_rate": 4.993825488705973e-05,
+      "loss": 7.0358,
+      "step": 670
+    },
+    {
+      "epoch": 0.05252181972657759,
+      "grad_norm": 1.078125,
+      "learning_rate": 4.9933783995639084e-05,
+      "loss": 7.0288,
+      "step": 680
+    },
+    {
+      "epoch": 0.05329419942843902,
+      "grad_norm": 1.0546875,
+      "learning_rate": 4.992915706033345e-05,
+      "loss": 7.0159,
+      "step": 690
+    },
+    {
+      "epoch": 0.054066579130300454,
+      "grad_norm": 1.1015625,
+      "learning_rate": 4.992437411009975e-05,
+      "loss": 7.0396,
+      "step": 700
+    },
+    {
+      "epoch": 0.05483895883216189,
+      "grad_norm": 1.046875,
+      "learning_rate": 4.991943517487126e-05,
+      "loss": 7.041,
+      "step": 710
+    },
+    {
+      "epoch": 0.05561133853402333,
+      "grad_norm": 1.078125,
+      "learning_rate": 4.9914340285557467e-05,
+      "loss": 7.0264,
+      "step": 720
+    },
+    {
+      "epoch": 0.05638371823588476,
+      "grad_norm": 1.03125,
+      "learning_rate": 4.990908947404388e-05,
+      "loss": 7.0325,
+      "step": 730
+    },
+    {
+      "epoch": 0.057156097937746196,
+      "grad_norm": 1.046875,
+      "learning_rate": 4.990368277319183e-05,
+      "loss": 7.0152,
+      "step": 740
+    },
+    {
+      "epoch": 0.05792847763960763,
+      "grad_norm": 1.109375,
+      "learning_rate": 4.989812021683824e-05,
+      "loss": 7.0214,
+      "step": 750
+    },
+    {
+      "epoch": 0.058700857341469063,
+      "grad_norm": 1.1015625,
+      "learning_rate": 4.9892401839795416e-05,
+      "loss": 7.0358,
+      "step": 760
+    },
+    {
+      "epoch": 0.059473237043330504,
+      "grad_norm": 1.078125,
+      "learning_rate": 4.988652767785087e-05,
+      "loss": 7.0274,
+      "step": 770
+    },
+    {
+      "epoch": 0.06024561674519194,
+      "grad_norm": 1.0703125,
+      "learning_rate": 4.988049776776705e-05,
+      "loss": 7.01,
+      "step": 780
+    },
+    {
+      "epoch": 0.06101799644705337,
+      "grad_norm": 1.0703125,
+      "learning_rate": 4.9874312147281144e-05,
+      "loss": 6.9994,
+      "step": 790
+    },
+    {
+      "epoch": 0.061790376148914805,
+      "grad_norm": 1.0390625,
+      "learning_rate": 4.9867970855104794e-05,
+      "loss": 7.001,
+      "step": 800
+    },
+    {
+      "epoch": 0.06256275585077624,
+      "grad_norm": 1.03125,
+      "learning_rate": 4.986147393092393e-05,
+      "loss": 7.0192,
+      "step": 810
+    },
+    {
+      "epoch": 0.06333513555263767,
+      "grad_norm": 1.0234375,
+      "learning_rate": 4.985482141539844e-05,
+      "loss": 7.0005,
+      "step": 820
+    },
+    {
+      "epoch": 0.0641075152544991,
+      "grad_norm": 1.0625,
+      "learning_rate": 4.984801335016198e-05,
+      "loss": 6.9889,
+      "step": 830
+    },
+    {
+      "epoch": 0.06487989495636055,
+      "grad_norm": 1.0625,
+      "learning_rate": 4.984104977782168e-05,
+      "loss": 6.9976,
+      "step": 840
+    },
+    {
+      "epoch": 0.06565227465822199,
+      "grad_norm": 1.0625,
+      "learning_rate": 4.983393074195787e-05,
+      "loss": 7.0213,
+      "step": 850
+    },
+    {
+      "epoch": 0.06642465436008342,
+      "grad_norm": 1.0078125,
+      "learning_rate": 4.9826656287123855e-05,
+      "loss": 6.995,
+      "step": 860
+    },
+    {
+      "epoch": 0.06719703406194485,
+      "grad_norm": 1.015625,
+      "learning_rate": 4.981922645884558e-05,
+      "loss": 7.003,
+      "step": 870
+    },
+    {
+      "epoch": 0.06796941376380629,
+      "grad_norm": 1.078125,
+      "learning_rate": 4.981164130362136e-05,
+      "loss": 7.0236,
+      "step": 880
+    },
+    {
+      "epoch": 0.06874179346566772,
+      "grad_norm": 1.0234375,
+      "learning_rate": 4.9803900868921626e-05,
+      "loss": 6.9976,
+      "step": 890
+    },
+    {
+      "epoch": 0.06951417316752916,
+      "grad_norm": 1.015625,
+      "learning_rate": 4.979600520318858e-05,
+      "loss": 7.0057,
+      "step": 900
+    },
+    {
+      "epoch": 0.07028655286939059,
+      "grad_norm": 1.0625,
+      "learning_rate": 4.978795435583591e-05,
+      "loss": 6.9869,
+      "step": 910
+    },
+    {
+      "epoch": 0.07105893257125202,
+      "grad_norm": 1.2109375,
+      "learning_rate": 4.977974837724849e-05,
+      "loss": 7.0009,
+      "step": 920
+    },
+    {
+      "epoch": 0.07183131227311346,
+      "grad_norm": 1.0078125,
+      "learning_rate": 4.9771387318782056e-05,
+      "loss": 7.0129,
+      "step": 930
+    },
+    {
+      "epoch": 0.0726036919749749,
+      "grad_norm": 1.0390625,
+      "learning_rate": 4.9762871232762883e-05,
+      "loss": 6.9773,
+      "step": 940
+    },
+    {
+      "epoch": 0.07337607167683634,
+      "grad_norm": 1.046875,
+      "learning_rate": 4.9754200172487464e-05,
+      "loss": 6.9807,
+      "step": 950
+    },
+    {
+      "epoch": 0.07414845137869777,
+      "grad_norm": 1.0546875,
+      "learning_rate": 4.974537419222216e-05,
+      "loss": 6.9977,
+      "step": 960
+    },
+    {
+      "epoch": 0.0749208310805592,
+      "grad_norm": 1.0234375,
+      "learning_rate": 4.973639334720288e-05,
+      "loss": 6.9931,
+      "step": 970
+    },
+    {
+      "epoch": 0.07569321078242064,
+      "grad_norm": 1.078125,
+      "learning_rate": 4.972725769363474e-05,
+      "loss": 6.9979,
+      "step": 980
+    },
+    {
+      "epoch": 0.07646559048428207,
+      "grad_norm": 1.1015625,
+      "learning_rate": 4.971796728869168e-05,
+      "loss": 6.9738,
+      "step": 990
+    },
+    {
+      "epoch": 0.0772379701861435,
+      "grad_norm": 1.03125,
+      "learning_rate": 4.9708522190516135e-05,
+      "loss": 6.985,
+      "step": 1000
+    },
+    {
+      "epoch": 0.0772379701861435,
+      "eval_loss": 7.195706844329834,
+      "eval_runtime": 1.3795,
+      "eval_samples_per_second": 8.699,
+      "eval_steps_per_second": 1.45,
+      "step": 1000
+    },
+    {
+      "epoch": 0.07801034988800494,
+      "grad_norm": 1.09375,
+      "learning_rate": 4.969892245821867e-05,
+      "loss": 6.9962,
+      "step": 1010
+    },
+    {
+      "epoch": 0.07878272958986637,
+      "grad_norm": 1.078125,
+      "learning_rate": 4.9689168151877566e-05,
+      "loss": 6.9837,
+      "step": 1020
+    },
+    {
+      "epoch": 0.07955510929172781,
+      "grad_norm": 1.0234375,
+      "learning_rate": 4.967925933253853e-05,
+      "loss": 6.996,
+      "step": 1030
+    },
+    {
+      "epoch": 0.08032748899358925,
+      "grad_norm": 1.0625,
+      "learning_rate": 4.966919606221423e-05,
+      "loss": 6.9922,
+      "step": 1040
+    },
+    {
+      "epoch": 0.08109986869545069,
+      "grad_norm": 1.046875,
+      "learning_rate": 4.965897840388395e-05,
+      "loss": 6.9721,
+      "step": 1050
+    },
+    {
+      "epoch": 0.08187224839731212,
+      "grad_norm": 1.0625,
+      "learning_rate": 4.964860642149317e-05,
+      "loss": 6.9778,
+      "step": 1060
+    },
+    {
+      "epoch": 0.08264462809917356,
+      "grad_norm": 1.0703125,
+      "learning_rate": 4.963808017995321e-05,
+      "loss": 6.9568,
+      "step": 1070
+    },
+    {
+      "epoch": 0.08341700780103499,
+      "grad_norm": 1.0859375,
+      "learning_rate": 4.962739974514077e-05,
+      "loss": 6.989,
+      "step": 1080
+    },
+    {
+      "epoch": 0.08418938750289642,
+      "grad_norm": 1.0859375,
+      "learning_rate": 4.9616565183897554e-05,
+      "loss": 6.9685,
+      "step": 1090
+    },
+    {
+      "epoch": 0.08496176720475786,
+      "grad_norm": 1.03125,
+      "learning_rate": 4.960557656402983e-05,
+      "loss": 6.966,
+      "step": 1100
+    },
+    {
+      "epoch": 0.08573414690661929,
+      "grad_norm": 1.0390625,
+      "learning_rate": 4.959443395430804e-05,
+      "loss": 6.9458,
+      "step": 1110
+    },
+    {
+      "epoch": 0.08650652660848072,
+      "grad_norm": 1.015625,
+      "learning_rate": 4.958313742446632e-05,
+      "loss": 6.9388,
+      "step": 1120
+    },
+    {
+      "epoch": 0.08727890631034216,
+      "grad_norm": 1.0234375,
+      "learning_rate": 4.957168704520209e-05,
+      "loss": 6.9658,
+      "step": 1130
+    },
+    {
+      "epoch": 0.0880512860122036,
+      "grad_norm": 1.0390625,
+      "learning_rate": 4.956008288817563e-05,
+      "loss": 6.9727,
+      "step": 1140
+    },
+    {
+      "epoch": 0.08882366571406504,
+      "grad_norm": 1.03125,
+      "learning_rate": 4.95483250260096e-05,
+      "loss": 6.9777,
+      "step": 1150
+    },
+    {
+      "epoch": 0.08959604541592647,
+      "grad_norm": 1.078125,
+      "learning_rate": 4.9536413532288594e-05,
+      "loss": 6.9421,
+      "step": 1160
+    },
+    {
+      "epoch": 0.0903684251177879,
+      "grad_norm": 1.03125,
+      "learning_rate": 4.952434848155869e-05,
+      "loss": 6.9656,
+      "step": 1170
+    },
+    {
+      "epoch": 0.09114080481964934,
+      "grad_norm": 1.0078125,
+      "learning_rate": 4.951212994932697e-05,
+      "loss": 6.9749,
+      "step": 1180
+    },
+    {
+      "epoch": 0.09191318452151077,
+      "grad_norm": 1.109375,
+      "learning_rate": 4.949975801206107e-05,
+      "loss": 6.9558,
+      "step": 1190
+    },
+    {
+      "epoch": 0.09268556422337221,
+      "grad_norm": 1.03125,
+      "learning_rate": 4.948723274718867e-05,
+      "loss": 6.9572,
+      "step": 1200
+    },
+    {
+      "epoch": 0.09345794392523364,
+      "grad_norm": 1.0234375,
+      "learning_rate": 4.9474554233097026e-05,
+      "loss": 6.9558,
+      "step": 1210
+    },
+    {
+      "epoch": 0.09423032362709508,
+      "grad_norm": 1.015625,
+      "learning_rate": 4.946172254913248e-05,
+      "loss": 6.9313,
+      "step": 1220
+    },
+    {
+      "epoch": 0.09500270332895651,
+      "grad_norm": 1.078125,
+      "learning_rate": 4.944873777559996e-05,
+      "loss": 6.9627,
+      "step": 1230
+    },
+    {
+      "epoch": 0.09577508303081796,
+      "grad_norm": 1.0390625,
+      "learning_rate": 4.94355999937625e-05,
+      "loss": 6.9681,
+      "step": 1240
+    },
+    {
+      "epoch": 0.09654746273267939,
+      "grad_norm": 1.03125,
+      "learning_rate": 4.942230928584068e-05,
+      "loss": 6.9696,
+      "step": 1250
+    },
+    {
+      "epoch": 0.09731984243454082,
+      "grad_norm": 1.1015625,
+      "learning_rate": 4.9408865735012154e-05,
+      "loss": 6.9751,
+      "step": 1260
+    },
+    {
+      "epoch": 0.09809222213640226,
+      "grad_norm": 1.046875,
+      "learning_rate": 4.939526942541113e-05,
+      "loss": 6.9496,
+      "step": 1270
+    },
+    {
+      "epoch": 0.09886460183826369,
+      "grad_norm": 1.0,
+      "learning_rate": 4.93815204421278e-05,
+      "loss": 6.9535,
+      "step": 1280
+    },
+    {
+      "epoch": 0.09963698154012512,
+      "grad_norm": 1.0234375,
+      "learning_rate": 4.9367618871207885e-05,
+      "loss": 6.9631,
+      "step": 1290
+    },
+    {
+      "epoch": 0.10040936124198656,
+      "grad_norm": 1.0,
+      "learning_rate": 4.9353564799652014e-05,
+      "loss": 6.9414,
+      "step": 1300
+    },
+    {
+      "epoch": 0.10118174094384799,
+      "grad_norm": 1.0234375,
+      "learning_rate": 4.933935831541522e-05,
+      "loss": 6.9548,
+      "step": 1310
+    },
+    {
+      "epoch": 0.10195412064570943,
+      "grad_norm": 1.046875,
+      "learning_rate": 4.932499950740639e-05,
+      "loss": 6.9365,
+      "step": 1320
+    },
+    {
+      "epoch": 0.10272650034757086,
+      "grad_norm": 1.046875,
+      "learning_rate": 4.9310488465487704e-05,
+      "loss": 6.9616,
+      "step": 1330
+    },
+    {
+      "epoch": 0.1034988800494323,
+      "grad_norm": 1.03125,
+      "learning_rate": 4.9295825280474074e-05,
+      "loss": 6.9439,
+      "step": 1340
+    },
+    {
+      "epoch": 0.10427125975129374,
+      "grad_norm": 1.0390625,
+      "learning_rate": 4.9281010044132584e-05,
+      "loss": 6.9459,
+      "step": 1350
+    },
+    {
+      "epoch": 0.10504363945315517,
+      "grad_norm": 1.046875,
+      "learning_rate": 4.926604284918187e-05,
+      "loss": 6.9559,
+      "step": 1360
+    },
+    {
+      "epoch": 0.10581601915501661,
+      "grad_norm": 1.046875,
+      "learning_rate": 4.9250923789291615e-05,
+      "loss": 6.946,
+      "step": 1370
+    },
+    {
+      "epoch": 0.10658839885687804,
+      "grad_norm": 1.0078125,
+      "learning_rate": 4.923565295908191e-05,
+      "loss": 6.9574,
+      "step": 1380
+    },
+    {
+      "epoch": 0.10736077855873948,
+      "grad_norm": 1.0234375,
+      "learning_rate": 4.922023045412266e-05,
+      "loss": 6.9327,
+      "step": 1390
+    },
+    {
+      "epoch": 0.10813315826060091,
+      "grad_norm": 1.0078125,
+      "learning_rate": 4.920465637093302e-05,
+      "loss": 6.9403,
+      "step": 1400
+    },
+    {
+      "epoch": 0.10890553796246234,
+      "grad_norm": 1.0546875,
+      "learning_rate": 4.918893080698077e-05,
+      "loss": 6.953,
+      "step": 1410
+    },
+    {
+      "epoch": 0.10967791766432378,
+      "grad_norm": 0.98828125,
+      "learning_rate": 4.9173053860681694e-05,
+      "loss": 6.9444,
+      "step": 1420
+    },
+    {
+      "epoch": 0.11045029736618521,
+      "grad_norm": 1.0234375,
+      "learning_rate": 4.915702563139899e-05,
+      "loss": 6.939,
+      "step": 1430
+    },
+    {
+      "epoch": 0.11122267706804666,
+      "grad_norm": 0.98828125,
+      "learning_rate": 4.914084621944262e-05,
+      "loss": 6.953,
+      "step": 1440
+    },
+    {
+      "epoch": 0.11199505676990809,
+      "grad_norm": 1.0078125,
+      "learning_rate": 4.912451572606872e-05,
+      "loss": 6.9619,
+      "step": 1450
+    },
+    {
+      "epoch": 0.11276743647176952,
+      "grad_norm": 1.015625,
+      "learning_rate": 4.910803425347892e-05,
+      "loss": 6.931,
+      "step": 1460
+    },
+    {
+      "epoch": 0.11353981617363096,
+      "grad_norm": 1.03125,
+      "learning_rate": 4.909140190481974e-05,
+      "loss": 6.9716,
+      "step": 1470
+    },
+    {
+      "epoch": 0.11431219587549239,
+      "grad_norm": 1.0,
+      "learning_rate": 4.907461878418194e-05,
+      "loss": 6.9467,
+      "step": 1480
+    },
+    {
+      "epoch": 0.11508457557735383,
+      "grad_norm": 1.015625,
+      "learning_rate": 4.905768499659984e-05,
+      "loss": 6.9505,
+      "step": 1490
+    },
+    {
+      "epoch": 0.11585695527921526,
+      "grad_norm": 1.078125,
+      "learning_rate": 4.9040600648050704e-05,
+      "loss": 6.9373,
+      "step": 1500
+    },
+    {
+      "epoch": 0.1166293349810767,
+      "grad_norm": 0.984375,
+      "learning_rate": 4.9023365845454044e-05,
+      "loss": 6.9465,
+      "step": 1510
+    },
+    {
+      "epoch": 0.11740171468293813,
+      "grad_norm": 1.015625,
+      "learning_rate": 4.9005980696670975e-05,
+      "loss": 6.9179,
+      "step": 1520
+    },
+    {
+      "epoch": 0.11817409438479956,
+      "grad_norm": 0.98046875,
+      "learning_rate": 4.8988445310503505e-05,
+      "loss": 6.9397,
+      "step": 1530
+    },
+    {
+      "epoch": 0.11894647408666101,
+      "grad_norm": 1.015625,
+      "learning_rate": 4.897075979669391e-05,
+      "loss": 6.936,
+      "step": 1540
+    },
+    {
+      "epoch": 0.11971885378852244,
+      "grad_norm": 1.0390625,
+      "learning_rate": 4.8952924265923984e-05,
+      "loss": 6.9263,
+      "step": 1550
+    },
+    {
+      "epoch": 0.12049123349038388,
+      "grad_norm": 1.0,
+      "learning_rate": 4.89349388298144e-05,
+      "loss": 6.9298,
+      "step": 1560
+    },
+    {
+      "epoch": 0.12126361319224531,
+      "grad_norm": 0.9921875,
+      "learning_rate": 4.8916803600923974e-05,
+      "loss": 6.9218,
+      "step": 1570
+    },
+    {
+      "epoch": 0.12203599289410674,
+      "grad_norm": 1.0546875,
+      "learning_rate": 4.889851869274899e-05,
+      "loss": 6.9469,
+      "step": 1580
+    },
+    {
+      "epoch": 0.12280837259596818,
+      "grad_norm": 1.015625,
+      "learning_rate": 4.888008421972244e-05,
+      "loss": 6.9314,
+      "step": 1590
+    },
+    {
+      "epoch": 0.12358075229782961,
+      "grad_norm": 1.0078125,
+      "learning_rate": 4.886150029721339e-05,
+      "loss": 6.9246,
+      "step": 1600
+    },
+    {
+      "epoch": 0.12435313199969104,
+      "grad_norm": 1.0234375,
+      "learning_rate": 4.884276704152618e-05,
+      "loss": 6.9011,
+      "step": 1610
+    },
+    {
+      "epoch": 0.12512551170155248,
+      "grad_norm": 1.0078125,
+      "learning_rate": 4.8823884569899726e-05,
+      "loss": 6.9597,
+      "step": 1620
+    },
+    {
+      "epoch": 0.12589789140341393,
+      "grad_norm": 0.97265625,
+      "learning_rate": 4.8804853000506806e-05,
+      "loss": 6.9067,
+      "step": 1630
+    },
+    {
+      "epoch": 0.12667027110527534,
+      "grad_norm": 1.0703125,
+      "learning_rate": 4.878567245245329e-05,
+      "loss": 6.9405,
+      "step": 1640
+    },
+    {
+      "epoch": 0.1274426508071368,
+      "grad_norm": 1.0078125,
+      "learning_rate": 4.8766343045777386e-05,
+      "loss": 6.9444,
+      "step": 1650
+    },
+    {
+      "epoch": 0.1282150305089982,
+      "grad_norm": 0.984375,
+      "learning_rate": 4.874686490144895e-05,
+      "loss": 6.9338,
+      "step": 1660
+    },
+    {
+      "epoch": 0.12898741021085966,
+      "grad_norm": 0.99609375,
+      "learning_rate": 4.872723814136866e-05,
+      "loss": 6.9245,
+      "step": 1670
+    },
+    {
+      "epoch": 0.1297597899127211,
+      "grad_norm": 1.0703125,
+      "learning_rate": 4.870746288836727e-05,
+      "loss": 6.934,
+      "step": 1680
+    },
+    {
+      "epoch": 0.13053216961458253,
+      "grad_norm": 0.98046875,
+      "learning_rate": 4.8687539266204896e-05,
+      "loss": 6.9158,
+      "step": 1690
+    },
+    {
+      "epoch": 0.13130454931644397,
+      "grad_norm": 0.97265625,
+      "learning_rate": 4.866746739957015e-05,
+      "loss": 6.9198,
+      "step": 1700
+    },
+    {
+      "epoch": 0.1320769290183054,
+      "grad_norm": 1.0234375,
+      "learning_rate": 4.8647247414079435e-05,
+      "loss": 6.9156,
+      "step": 1710
+    },
+    {
+      "epoch": 0.13284930872016684,
+      "grad_norm": 1.078125,
+      "learning_rate": 4.862687943627612e-05,
+      "loss": 6.9489,
+      "step": 1720
+    },
+    {
+      "epoch": 0.13362168842202826,
+      "grad_norm": 1.0390625,
+      "learning_rate": 4.860636359362975e-05,
+      "loss": 6.918,
+      "step": 1730
+    },
+    {
+      "epoch": 0.1343940681238897,
+      "grad_norm": 1.0390625,
+      "learning_rate": 4.858570001453529e-05,
+      "loss": 6.9442,
+      "step": 1740
+    },
+    {
+      "epoch": 0.13516644782575113,
+      "grad_norm": 1.046875,
+      "learning_rate": 4.856488882831225e-05,
+      "loss": 6.9234,
+      "step": 1750
+    },
+    {
+      "epoch": 0.13593882752761258,
+      "grad_norm": 0.98828125,
+      "learning_rate": 4.854393016520394e-05,
+      "loss": 6.9181,
+      "step": 1760
+    },
+    {
+      "epoch": 0.136711207229474,
+      "grad_norm": 0.9921875,
+      "learning_rate": 4.852282415637662e-05,
+      "loss": 6.9348,
+      "step": 1770
+    },
+    {
+      "epoch": 0.13748358693133544,
+      "grad_norm": 1.03125,
+      "learning_rate": 4.850157093391869e-05,
+      "loss": 6.9299,
+      "step": 1780
+    },
+    {
+      "epoch": 0.1382559666331969,
+      "grad_norm": 1.03125,
+      "learning_rate": 4.848017063083986e-05,
+      "loss": 6.925,
+      "step": 1790
+    },
+    {
+      "epoch": 0.1390283463350583,
+      "grad_norm": 1.015625,
+      "learning_rate": 4.845862338107031e-05,
+      "loss": 6.9315,
+      "step": 1800
+    },
+    {
+      "epoch": 0.13980072603691976,
+      "grad_norm": 1.015625,
+      "learning_rate": 4.843692931945989e-05,
+      "loss": 6.9169,
+      "step": 1810
+    },
+    {
+      "epoch": 0.14057310573878118,
+      "grad_norm": 1.0625,
+      "learning_rate": 4.841508858177721e-05,
+      "loss": 6.9144,
+      "step": 1820
+    },
+    {
+      "epoch": 0.14134548544064263,
+      "grad_norm": 1.0234375,
+      "learning_rate": 4.8393101304708866e-05,
+      "loss": 6.9215,
+      "step": 1830
+    },
+    {
+      "epoch": 0.14211786514250405,
+      "grad_norm": 1.03125,
+      "learning_rate": 4.837096762585851e-05,
+      "loss": 6.9191,
+      "step": 1840
+    },
+    {
+      "epoch": 0.1428902448443655,
+      "grad_norm": 1.0,
+      "learning_rate": 4.834868768374604e-05,
+      "loss": 6.9206,
+      "step": 1850
+    },
+    {
+      "epoch": 0.1436626245462269,
+      "grad_norm": 1.0078125,
+      "learning_rate": 4.8326261617806735e-05,
+      "loss": 6.902,
+      "step": 1860
+    },
+    {
+      "epoch": 0.14443500424808836,
+      "grad_norm": 1.03125,
+      "learning_rate": 4.830368956839033e-05,
+      "loss": 6.9232,
+      "step": 1870
+    },
+    {
+      "epoch": 0.1452073839499498,
+      "grad_norm": 1.0078125,
+      "learning_rate": 4.8280971676760204e-05,
+      "loss": 6.9145,
+      "step": 1880
+    },
+    {
+      "epoch": 0.14597976365181123,
+      "grad_norm": 1.015625,
+      "learning_rate": 4.825810808509245e-05,
+      "loss": 6.9167,
+      "step": 1890
+    },
+    {
+      "epoch": 0.14675214335367268,
+      "grad_norm": 0.95703125,
+      "learning_rate": 4.8235098936475e-05,
+      "loss": 6.8914,
+      "step": 1900
+    },
+    {
+      "epoch": 0.1475245230555341,
+      "grad_norm": 0.96484375,
+      "learning_rate": 4.8211944374906735e-05,
+      "loss": 6.9187,
+      "step": 1910
+    },
+    {
+      "epoch": 0.14829690275739554,
+      "grad_norm": 1.0703125,
+      "learning_rate": 4.818864454529658e-05,
+      "loss": 6.8995,
+      "step": 1920
+    },
+    {
+      "epoch": 0.14906928245925696,
+      "grad_norm": 1.125,
+      "learning_rate": 4.81651995934626e-05,
+      "loss": 6.9161,
+      "step": 1930
+    },
+    {
+      "epoch": 0.1498416621611184,
+      "grad_norm": 1.015625,
+      "learning_rate": 4.814160966613104e-05,
+      "loss": 6.921,
+      "step": 1940
+    },
+    {
+      "epoch": 0.15061404186297983,
+      "grad_norm": 0.98046875,
+      "learning_rate": 4.811787491093552e-05,
+      "loss": 6.9216,
+      "step": 1950
+    },
+    {
+      "epoch": 0.15138642156484128,
+      "grad_norm": 1.0078125,
+      "learning_rate": 4.8093995476415996e-05,
+      "loss": 6.9262,
+      "step": 1960
+    },
+    {
+      "epoch": 0.1521588012667027,
+      "grad_norm": 1.015625,
+      "learning_rate": 4.8069971512017884e-05,
+      "loss": 6.9033,
+      "step": 1970
+    },
+    {
+      "epoch": 0.15293118096856415,
+      "grad_norm": 1.03125,
+      "learning_rate": 4.804580316809111e-05,
+      "loss": 6.8992,
+      "step": 1980
+    },
+    {
+      "epoch": 0.1537035606704256,
+      "grad_norm": 0.9921875,
+      "learning_rate": 4.802149059588918e-05,
+      "loss": 6.9153,
+      "step": 1990
+    },
+    {
+      "epoch": 0.154475940372287,
+      "grad_norm": 0.9921875,
+      "learning_rate": 4.799703394756824e-05,
+      "loss": 6.9115,
+      "step": 2000
+    },
+    {
+      "epoch": 0.154475940372287,
+      "eval_loss": 7.10474157333374,
+      "eval_runtime": 1.4731,
+      "eval_samples_per_second": 8.146,
+      "eval_steps_per_second": 1.358,
+      "step": 2000
+    },
+    {
+      "epoch": 0.15524832007414846,
+      "grad_norm": 1.046875,
+      "learning_rate": 4.79724333761861e-05,
+      "loss": 6.9161,
+      "step": 2010
+    },
+    {
+      "epoch": 0.15602069977600988,
+      "grad_norm": 0.99609375,
+      "learning_rate": 4.7947689035701285e-05,
+      "loss": 6.9294,
+      "step": 2020
+    },
+    {
+      "epoch": 0.15679307947787133,
+      "grad_norm": 0.98046875,
+      "learning_rate": 4.792280108097208e-05,
+      "loss": 6.9147,
+      "step": 2030
+    },
+    {
+      "epoch": 0.15756545917973275,
+      "grad_norm": 1.0078125,
+      "learning_rate": 4.789776966775556e-05,
+      "loss": 6.8972,
+      "step": 2040
+    },
+    {
+      "epoch": 0.1583378388815942,
+      "grad_norm": 1.03125,
+      "learning_rate": 4.78725949527066e-05,
+      "loss": 6.901,
+      "step": 2050
+    },
+    {
+      "epoch": 0.15911021858345561,
+      "grad_norm": 1.0078125,
+      "learning_rate": 4.784727709337692e-05,
+      "loss": 6.9306,
+      "step": 2060
+    },
+    {
+      "epoch": 0.15988259828531706,
+      "grad_norm": 1.015625,
+      "learning_rate": 4.782181624821406e-05,
+      "loss": 6.9092,
+      "step": 2070
+    },
+    {
+      "epoch": 0.1606549779871785,
+      "grad_norm": 1.015625,
+      "learning_rate": 4.779621257656045e-05,
+      "loss": 6.9171,
+      "step": 2080
+    },
+    {
+      "epoch": 0.16142735768903993,
+      "grad_norm": 0.97265625,
+      "learning_rate": 4.777046623865233e-05,
+      "loss": 6.922,
+      "step": 2090
+    },
+    {
+      "epoch": 0.16219973739090138,
+      "grad_norm": 1.0078125,
+      "learning_rate": 4.774457739561884e-05,
+      "loss": 6.9022,
+      "step": 2100
+    },
+    {
+      "epoch": 0.1629721170927628,
+      "grad_norm": 1.0078125,
+      "learning_rate": 4.771854620948092e-05,
+      "loss": 6.9137,
+      "step": 2110
+    },
+    {
+      "epoch": 0.16374449679462424,
+      "grad_norm": 1.078125,
+      "learning_rate": 4.769237284315038e-05,
+      "loss": 6.9188,
+      "step": 2120
+    },
+    {
+      "epoch": 0.16451687649648566,
+      "grad_norm": 0.98046875,
+      "learning_rate": 4.766605746042882e-05,
+      "loss": 6.9029,
+      "step": 2130
+    },
+    {
+      "epoch": 0.1652892561983471,
+      "grad_norm": 0.98828125,
+      "learning_rate": 4.76396002260066e-05,
+      "loss": 6.8852,
+      "step": 2140
+    },
+    {
+      "epoch": 0.16606163590020853,
+      "grad_norm": 1.0234375,
+      "learning_rate": 4.76130013054619e-05,
+      "loss": 6.9108,
+      "step": 2150
+    },
+    {
+      "epoch": 0.16683401560206998,
+      "grad_norm": 1.0546875,
+      "learning_rate": 4.758626086525956e-05,
+      "loss": 6.8932,
+      "step": 2160
+    },
+    {
+      "epoch": 0.1676063953039314,
+      "grad_norm": 0.98046875,
+      "learning_rate": 4.755937907275012e-05,
+      "loss": 6.893,
+      "step": 2170
+    },
+    {
+      "epoch": 0.16837877500579285,
+      "grad_norm": 0.99609375,
+      "learning_rate": 4.7532356096168765e-05,
+      "loss": 6.8868,
+      "step": 2180
+    },
+    {
+      "epoch": 0.1691511547076543,
+      "grad_norm": 0.9921875,
+      "learning_rate": 4.750519210463421e-05,
+      "loss": 6.8966,
+      "step": 2190
+    },
+    {
+      "epoch": 0.16992353440951571,
+      "grad_norm": 1.015625,
+      "learning_rate": 4.747788726814775e-05,
+      "loss": 6.9187,
+      "step": 2200
+    },
+    {
+      "epoch": 0.17069591411137716,
+      "grad_norm": 0.984375,
+      "learning_rate": 4.7450441757592076e-05,
+      "loss": 6.9059,
+      "step": 2210
+    },
+    {
+      "epoch": 0.17146829381323858,
+      "grad_norm": 1.0,
+      "learning_rate": 4.7422855744730313e-05,
+      "loss": 6.9143,
+      "step": 2220
+    },
+    {
+      "epoch": 0.17224067351510003,
+      "grad_norm": 1.0078125,
+      "learning_rate": 4.739512940220485e-05,
+      "loss": 6.8997,
+      "step": 2230
+    },
+    {
+      "epoch": 0.17301305321696145,
+      "grad_norm": 1.0,
+      "learning_rate": 4.736726290353634e-05,
+      "loss": 6.9122,
+      "step": 2240
+    },
+    {
+      "epoch": 0.1737854329188229,
+      "grad_norm": 1.0234375,
+      "learning_rate": 4.733925642312257e-05,
+      "loss": 6.911,
+      "step": 2250
+    },
+    {
+      "epoch": 0.17455781262068432,
+      "grad_norm": 0.98046875,
+      "learning_rate": 4.731111013623738e-05,
+      "loss": 6.8854,
+      "step": 2260
+    },
+    {
+      "epoch": 0.17533019232254576,
+      "grad_norm": 1.03125,
+      "learning_rate": 4.728282421902957e-05,
+      "loss": 6.8924,
+      "step": 2270
+    },
+    {
+      "epoch": 0.1761025720244072,
+      "grad_norm": 0.984375,
+      "learning_rate": 4.7254398848521754e-05,
+      "loss": 6.9027,
+      "step": 2280
+    },
+    {
+      "epoch": 0.17687495172626863,
+      "grad_norm": 1.0390625,
+      "learning_rate": 4.722583420260936e-05,
+      "loss": 6.8791,
+      "step": 2290
+    },
+    {
+      "epoch": 0.17764733142813008,
+      "grad_norm": 0.984375,
+      "learning_rate": 4.719713046005938e-05,
+      "loss": 6.9119,
+      "step": 2300
+    },
+    {
+      "epoch": 0.1784197111299915,
+      "grad_norm": 1.0078125,
+      "learning_rate": 4.7168287800509376e-05,
+      "loss": 6.906,
+      "step": 2310
+    },
+    {
+      "epoch": 0.17919209083185295,
+      "grad_norm": 1.0078125,
+      "learning_rate": 4.7139306404466256e-05,
+      "loss": 6.8932,
+      "step": 2320
+    },
+    {
+      "epoch": 0.17996447053371437,
+      "grad_norm": 1.0,
+      "learning_rate": 4.71101864533052e-05,
+      "loss": 6.9049,
+      "step": 2330
+    },
+    {
+      "epoch": 0.1807368502355758,
+      "grad_norm": 1.015625,
+      "learning_rate": 4.708092812926853e-05,
+      "loss": 6.8837,
+      "step": 2340
+    },
+    {
+      "epoch": 0.18150922993743723,
+      "grad_norm": 1.015625,
+      "learning_rate": 4.7051531615464514e-05,
+      "loss": 6.9055,
+      "step": 2350
+    },
+    {
+      "epoch": 0.18228160963929868,
+      "grad_norm": 1.0078125,
+      "learning_rate": 4.702199709586631e-05,
+      "loss": 6.895,
+      "step": 2360
+    },
+    {
+      "epoch": 0.1830539893411601,
+      "grad_norm": 0.9921875,
+      "learning_rate": 4.6992324755310703e-05,
+      "loss": 6.9278,
+      "step": 2370
+    },
+    {
+      "epoch": 0.18382636904302155,
+      "grad_norm": 1.0234375,
+      "learning_rate": 4.696251477949705e-05,
+      "loss": 6.9203,
+      "step": 2380
+    },
+    {
+      "epoch": 0.184598748744883,
+      "grad_norm": 1.015625,
+      "learning_rate": 4.693256735498605e-05,
+      "loss": 6.877,
+      "step": 2390
+    },
+    {
+      "epoch": 0.18537112844674442,
+      "grad_norm": 0.984375,
+      "learning_rate": 4.690248266919862e-05,
+      "loss": 6.9047,
+      "step": 2400
+    },
+    {
+      "epoch": 0.18614350814860586,
+      "grad_norm": 1.0234375,
+      "learning_rate": 4.68722609104147e-05,
+      "loss": 6.8985,
+      "step": 2410
+    },
+    {
+      "epoch": 0.18691588785046728,
+      "grad_norm": 0.97265625,
+      "learning_rate": 4.6841902267772067e-05,
+      "loss": 6.8853,
+      "step": 2420
+    },
+    {
+      "epoch": 0.18768826755232873,
+      "grad_norm": 0.99609375,
+      "learning_rate": 4.681140693126518e-05,
+      "loss": 6.9066,
+      "step": 2430
+    },
+    {
+      "epoch": 0.18846064725419015,
+      "grad_norm": 1.0234375,
+      "learning_rate": 4.678077509174394e-05,
+      "loss": 6.9068,
+      "step": 2440
+    },
+    {
+      "epoch": 0.1892330269560516,
+      "grad_norm": 1.0390625,
+      "learning_rate": 4.675000694091257e-05,
+      "loss": 6.8997,
+      "step": 2450
+    },
+    {
+      "epoch": 0.19000540665791302,
+      "grad_norm": 1.015625,
+      "learning_rate": 4.671910267132836e-05,
+      "loss": 6.8845,
+      "step": 2460
+    },
+    {
+      "epoch": 0.19077778635977446,
+      "grad_norm": 1.0234375,
+      "learning_rate": 4.6688062476400454e-05,
+      "loss": 6.9147,
+      "step": 2470
+    },
+    {
+      "epoch": 0.1915501660616359,
+      "grad_norm": 1.0,
+      "learning_rate": 4.66568865503887e-05,
+      "loss": 6.887,
+      "step": 2480
+    },
+    {
+      "epoch": 0.19232254576349733,
+      "grad_norm": 1.0546875,
+      "learning_rate": 4.662557508840237e-05,
+      "loss": 6.9085,
+      "step": 2490
+    },
+    {
+      "epoch": 0.19309492546535878,
+      "grad_norm": 1.015625,
+      "learning_rate": 4.659412828639897e-05,
+      "loss": 6.879,
+      "step": 2500
+    },
+    {
+      "epoch": 0.1938673051672202,
+      "grad_norm": 1.0234375,
+      "learning_rate": 4.6562546341183014e-05,
+      "loss": 6.8863,
+      "step": 2510
+    },
+    {
+      "epoch": 0.19463968486908165,
+      "grad_norm": 1.0078125,
+      "learning_rate": 4.653082945040478e-05,
+      "loss": 6.8804,
+      "step": 2520
+    },
+    {
+      "epoch": 0.19541206457094307,
+      "grad_norm": 1.0390625,
+      "learning_rate": 4.649897781255908e-05,
+      "loss": 6.8728,
+      "step": 2530
+    },
+    {
+      "epoch": 0.19618444427280451,
+      "grad_norm": 0.98828125,
+      "learning_rate": 4.646699162698402e-05,
+      "loss": 6.8812,
+      "step": 2540
+    },
+    {
+      "epoch": 0.19695682397466593,
+      "grad_norm": 1.015625,
+      "learning_rate": 4.643487109385978e-05,
+      "loss": 6.8859,
+      "step": 2550
+    },
+    {
+      "epoch": 0.19772920367652738,
+      "grad_norm": 0.98046875,
+      "learning_rate": 4.640261641420726e-05,
+      "loss": 6.8882,
+      "step": 2560
+    },
+    {
+      "epoch": 0.19850158337838883,
+      "grad_norm": 0.9921875,
+      "learning_rate": 4.637022778988697e-05,
+      "loss": 6.8911,
+      "step": 2570
+    },
+    {
+      "epoch": 0.19927396308025025,
+      "grad_norm": 1.0,
+      "learning_rate": 4.6337705423597635e-05,
+      "loss": 6.8904,
+      "step": 2580
+    },
+    {
+      "epoch": 0.2000463427821117,
+      "grad_norm": 0.9765625,
+      "learning_rate": 4.630504951887502e-05,
+      "loss": 6.902,
+      "step": 2590
+    },
+    {
+      "epoch": 0.20081872248397312,
+      "grad_norm": 1.0234375,
+      "learning_rate": 4.62722602800906e-05,
+      "loss": 6.8906,
+      "step": 2600
+    },
+    {
+      "epoch": 0.20159110218583456,
+      "grad_norm": 0.984375,
+      "learning_rate": 4.6239337912450275e-05,
+      "loss": 6.8955,
+      "step": 2610
+    },
+    {
+      "epoch": 0.20236348188769598,
+      "grad_norm": 1.03125,
+      "learning_rate": 4.620628262199317e-05,
+      "loss": 6.9105,
+      "step": 2620
+    },
+    {
+      "epoch": 0.20313586158955743,
+      "grad_norm": 0.97265625,
+      "learning_rate": 4.6173094615590224e-05,
+      "loss": 6.8825,
+      "step": 2630
+    },
+    {
+      "epoch": 0.20390824129141885,
+      "grad_norm": 0.96875,
+      "learning_rate": 4.613977410094299e-05,
+      "loss": 6.8738,
+      "step": 2640
+    },
+    {
+      "epoch": 0.2046806209932803,
+      "grad_norm": 1.0859375,
+      "learning_rate": 4.61063212865823e-05,
+      "loss": 6.8829,
+      "step": 2650
+    },
+    {
+      "epoch": 0.20545300069514172,
+      "grad_norm": 0.98828125,
+      "learning_rate": 4.6072736381866944e-05,
+      "loss": 6.8847,
+      "step": 2660
+    },
+    {
+      "epoch": 0.20622538039700317,
+      "grad_norm": 1.0234375,
+      "learning_rate": 4.6039019596982393e-05,
+      "loss": 6.8774,
+      "step": 2670
+    },
+    {
+      "epoch": 0.2069977600988646,
+      "grad_norm": 1.0078125,
+      "learning_rate": 4.6005171142939465e-05,
+      "loss": 6.8789,
+      "step": 2680
+    },
+    {
+      "epoch": 0.20777013980072603,
+      "grad_norm": 1.0390625,
+      "learning_rate": 4.5971191231572994e-05,
+      "loss": 6.8847,
+      "step": 2690
+    },
+    {
+      "epoch": 0.20854251950258748,
+      "grad_norm": 0.984375,
+      "learning_rate": 4.593708007554054e-05,
+      "loss": 6.8994,
+      "step": 2700
+    },
+    {
+      "epoch": 0.2093148992044489,
+      "grad_norm": 1.015625,
+      "learning_rate": 4.5902837888321024e-05,
+      "loss": 6.8881,
+      "step": 2710
+    },
+    {
+      "epoch": 0.21008727890631035,
+      "grad_norm": 0.984375,
+      "learning_rate": 4.58684648842134e-05,
+      "loss": 6.8651,
+      "step": 2720
+    },
+    {
+      "epoch": 0.21085965860817177,
+      "grad_norm": 0.95703125,
+      "learning_rate": 4.5833961278335327e-05,
+      "loss": 6.8649,
+      "step": 2730
+    },
+    {
+      "epoch": 0.21163203831003322,
+      "grad_norm": 0.98828125,
+      "learning_rate": 4.5799327286621795e-05,
+      "loss": 6.8934,
+      "step": 2740
+    },
+    {
+      "epoch": 0.21240441801189464,
+      "grad_norm": 1.0,
+      "learning_rate": 4.5764563125823815e-05,
+      "loss": 6.8793,
+      "step": 2750
+    },
+    {
+      "epoch": 0.21317679771375608,
+      "grad_norm": 0.97265625,
+      "learning_rate": 4.572966901350702e-05,
+      "loss": 6.8735,
+      "step": 2760
+    },
+    {
+      "epoch": 0.21394917741561753,
+      "grad_norm": 0.99609375,
+      "learning_rate": 4.5694645168050345e-05,
+      "loss": 6.8832,
+      "step": 2770
+    },
+    {
+      "epoch": 0.21472155711747895,
+      "grad_norm": 1.015625,
+      "learning_rate": 4.565949180864461e-05,
+      "loss": 6.8664,
+      "step": 2780
+    },
+    {
+      "epoch": 0.2154939368193404,
+      "grad_norm": 1.015625,
+      "learning_rate": 4.562420915529119e-05,
+      "loss": 6.9006,
+      "step": 2790
+    },
+    {
+      "epoch": 0.21626631652120182,
+      "grad_norm": 1.015625,
+      "learning_rate": 4.5588797428800636e-05,
+      "loss": 6.8599,
+      "step": 2800
+    },
+    {
+      "epoch": 0.21703869622306327,
+      "grad_norm": 0.984375,
+      "learning_rate": 4.555325685079127e-05,
+      "loss": 6.8699,
+      "step": 2810
+    },
+    {
+      "epoch": 0.21781107592492469,
+      "grad_norm": 0.96875,
+      "learning_rate": 4.55175876436878e-05,
+      "loss": 6.8722,
+      "step": 2820
+    },
+    {
+      "epoch": 0.21858345562678613,
+      "grad_norm": 1.0,
+      "learning_rate": 4.548179003071995e-05,
+      "loss": 6.8641,
+      "step": 2830
+    },
+    {
+      "epoch": 0.21935583532864755,
+      "grad_norm": 0.98046875,
+      "learning_rate": 4.5445864235921044e-05,
+      "loss": 6.8844,
+      "step": 2840
+    },
+    {
+      "epoch": 0.220128215030509,
+      "grad_norm": 0.9921875,
+      "learning_rate": 4.540981048412662e-05,
+      "loss": 6.9032,
+      "step": 2850
+    },
+    {
+      "epoch": 0.22090059473237042,
+      "grad_norm": 1.03125,
+      "learning_rate": 4.537362900097302e-05,
+      "loss": 6.8924,
+      "step": 2860
+    },
+    {
+      "epoch": 0.22167297443423187,
+      "grad_norm": 0.98828125,
+      "learning_rate": 4.533732001289593e-05,
+      "loss": 6.8881,
+      "step": 2870
+    },
+    {
+      "epoch": 0.22244535413609332,
+      "grad_norm": 1.0078125,
+      "learning_rate": 4.5300883747129055e-05,
+      "loss": 6.8808,
+      "step": 2880
+    },
+    {
+      "epoch": 0.22321773383795473,
+      "grad_norm": 0.99609375,
+      "learning_rate": 4.526432043170262e-05,
+      "loss": 6.882,
+      "step": 2890
+    },
+    {
+      "epoch": 0.22399011353981618,
+      "grad_norm": 1.0078125,
+      "learning_rate": 4.522763029544196e-05,
+      "loss": 6.8773,
+      "step": 2900
+    },
+    {
+      "epoch": 0.2247624932416776,
+      "grad_norm": 0.98046875,
+      "learning_rate": 4.519081356796613e-05,
+      "loss": 6.8684,
+      "step": 2910
+    },
+    {
+      "epoch": 0.22553487294353905,
+      "grad_norm": 0.9765625,
+      "learning_rate": 4.5153870479686385e-05,
+      "loss": 6.8759,
+      "step": 2920
+    },
+    {
+      "epoch": 0.22630725264540047,
+      "grad_norm": 0.98046875,
+      "learning_rate": 4.511680126180485e-05,
+      "loss": 6.8927,
+      "step": 2930
+    },
+    {
+      "epoch": 0.22707963234726192,
+      "grad_norm": 1.0234375,
+      "learning_rate": 4.507960614631295e-05,
+      "loss": 6.8713,
+      "step": 2940
+    },
+    {
+      "epoch": 0.22785201204912334,
+      "grad_norm": 0.98828125,
+      "learning_rate": 4.504228536599007e-05,
+      "loss": 6.891,
+      "step": 2950
+    },
+    {
+      "epoch": 0.22862439175098478,
+      "grad_norm": 0.9921875,
+      "learning_rate": 4.5004839154402004e-05,
+      "loss": 6.8738,
+      "step": 2960
+    },
+    {
+      "epoch": 0.22939677145284623,
+      "grad_norm": 1.0234375,
+      "learning_rate": 4.496726774589957e-05,
+      "loss": 6.8653,
+      "step": 2970
+    },
+    {
+      "epoch": 0.23016915115470765,
+      "grad_norm": 1.015625,
+      "learning_rate": 4.4929571375617084e-05,
+      "loss": 6.879,
+      "step": 2980
+    },
+    {
+      "epoch": 0.2309415308565691,
+      "grad_norm": 1.015625,
+      "learning_rate": 4.489175027947094e-05,
+      "loss": 6.8891,
+      "step": 2990
+    },
+    {
+      "epoch": 0.2314821966478721,
+      "eval_loss": 7.09494161605835,
+      "eval_runtime": 1.4843,
+      "eval_samples_per_second": 8.084,
+      "eval_steps_per_second": 1.347,
+      "step": 2997
+    },
+    {
+      "epoch": 0.23171391055843052,
+      "grad_norm": 1.0390625,
+      "learning_rate": 4.4853804694158074e-05,
+      "loss": 6.8979,
+      "step": 3000
+    },
+    {
+      "epoch": 0.23171391055843052,
+      "eval_loss": 7.093435287475586,
+      "eval_runtime": 0.6702,
+      "eval_samples_per_second": 17.905,
+      "eval_steps_per_second": 2.984,
+      "step": 3000
+    },
+    {
+      "epoch": 0.23248629026029197,
+      "grad_norm": 1.0,
+      "learning_rate": 4.481573485715456e-05,
+      "loss": 6.8842,
+      "step": 3010
+    },
+    {
+      "epoch": 0.2332586699621534,
+      "grad_norm": 1.015625,
+      "learning_rate": 4.477754100671405e-05,
+      "loss": 6.9093,
+      "step": 3020
+    },
+    {
+      "epoch": 0.23403104966401483,
+      "grad_norm": 1.0,
+      "learning_rate": 4.4739223381866315e-05,
+      "loss": 6.903,
+      "step": 3030
+    },
+    {
+      "epoch": 0.23480342936587625,
+      "grad_norm": 1.015625,
+      "learning_rate": 4.470078222241575e-05,
+      "loss": 6.8967,
+      "step": 3040
+    },
+    {
+      "epoch": 0.2355758090677377,
+      "grad_norm": 1.03125,
+      "learning_rate": 4.466221776893987e-05,
+      "loss": 6.8878,
+      "step": 3050
+    },
+    {
+      "epoch": 0.23634818876959912,
+      "grad_norm": 0.9921875,
+      "learning_rate": 4.4623530262787824e-05,
+      "loss": 6.8928,
+      "step": 3060
+    },
+    {
+      "epoch": 0.23712056847146057,
+      "grad_norm": 0.984375,
+      "learning_rate": 4.4584719946078846e-05,
+      "loss": 6.8557,
+      "step": 3070
+    },
+    {
+      "epoch": 0.23789294817332202,
+      "grad_norm": 0.98828125,
+      "learning_rate": 4.454578706170075e-05,
+      "loss": 6.8815,
+      "step": 3080
+    },
+    {
+      "epoch": 0.23866532787518344,
+      "grad_norm": 0.9921875,
+      "learning_rate": 4.4506731853308435e-05,
+      "loss": 6.8865,
+      "step": 3090
+    },
+    {
+      "epoch": 0.23943770757704488,
+      "grad_norm": 0.984375,
+      "learning_rate": 4.446755456532235e-05,
+      "loss": 6.8907,
+      "step": 3100
+    },
+    {
+      "epoch": 0.2402100872789063,
+      "grad_norm": 0.9765625,
+      "learning_rate": 4.442825544292693e-05,
+      "loss": 6.8811,
+      "step": 3110
+    },
+    {
+      "epoch": 0.24098246698076775,
+      "grad_norm": 1.0,
+      "learning_rate": 4.438883473206913e-05,
+      "loss": 6.8935,
+      "step": 3120
+    },
+    {
+      "epoch": 0.24175484668262917,
+      "grad_norm": 0.98828125,
+      "learning_rate": 4.4349292679456813e-05,
+      "loss": 6.8793,
+      "step": 3130
+    },
+    {
+      "epoch": 0.24252722638449062,
+      "grad_norm": 0.97265625,
+      "learning_rate": 4.4309629532557254e-05,
+      "loss": 6.9043,
+      "step": 3140
+    },
+    {
+      "epoch": 0.24329960608635204,
+      "grad_norm": 0.984375,
+      "learning_rate": 4.426984553959558e-05,
+      "loss": 6.8944,
+      "step": 3150
+    },
+    {
+      "epoch": 0.24407198578821349,
+      "grad_norm": 0.98828125,
+      "learning_rate": 4.422994094955319e-05,
+      "loss": 6.9129,
+      "step": 3160
+    },
+    {
+      "epoch": 0.24484436549007493,
+      "grad_norm": 0.98828125,
+      "learning_rate": 4.418991601216625e-05,
+      "loss": 6.8943,
+      "step": 3170
+    },
+    {
+      "epoch": 0.24561674519193635,
+      "grad_norm": 0.99609375,
+      "learning_rate": 4.4149770977924095e-05,
+      "loss": 6.8974,
+      "step": 3180
+    },
+    {
+      "epoch": 0.2463891248937978,
+      "grad_norm": 1.078125,
+      "learning_rate": 4.410950609806763e-05,
+      "loss": 6.8643,
+      "step": 3190
+    },
+    {
+      "epoch": 0.24716150459565922,
+      "grad_norm": 0.97265625,
+      "learning_rate": 4.406912162458783e-05,
+      "loss": 6.8718,
+      "step": 3200
+    },
+    {
+      "epoch": 0.24793388429752067,
+      "grad_norm": 0.984375,
+      "learning_rate": 4.402861781022412e-05,
+      "loss": 6.8917,
+      "step": 3210
+    },
+    {
+      "epoch": 0.2487062639993821,
+      "grad_norm": 1.0078125,
+      "learning_rate": 4.3987994908462795e-05,
+      "loss": 6.8998,
+      "step": 3220
+    },
+    {
+      "epoch": 0.24947864370124354,
+      "grad_norm": 1.0234375,
+      "learning_rate": 4.394725317353544e-05,
+      "loss": 6.8858,
+      "step": 3230
+    },
+    {
+      "epoch": 0.25025102340310496,
+      "grad_norm": 0.97265625,
+      "learning_rate": 4.390639286041733e-05,
+      "loss": 6.8749,
+      "step": 3240
+    },
+    {
+      "epoch": 0.2510234031049664,
+      "grad_norm": 0.9921875,
+      "learning_rate": 4.3865414224825855e-05,
+      "loss": 6.894,
+      "step": 3250
+    },
+    {
+      "epoch": 0.25179578280682785,
+      "grad_norm": 0.99609375,
+      "learning_rate": 4.382431752321889e-05,
+      "loss": 6.8933,
+      "step": 3260
+    },
+    {
+      "epoch": 0.2525681625086893,
+      "grad_norm": 1.0,
+      "learning_rate": 4.378310301279324e-05,
+      "loss": 6.8864,
+      "step": 3270
+    },
+    {
+      "epoch": 0.2533405422105507,
+      "grad_norm": 0.98828125,
+      "learning_rate": 4.3741770951482945e-05,
+      "loss": 6.8942,
+      "step": 3280
+    },
+    {
+      "epoch": 0.25411292191241214,
+      "grad_norm": 1.015625,
+      "learning_rate": 4.370032159795776e-05,
+      "loss": 6.8936,
+      "step": 3290
+    },
+    {
+      "epoch": 0.2548853016142736,
+      "grad_norm": 0.984375,
+      "learning_rate": 4.365875521162147e-05,
+      "loss": 6.8873,
+      "step": 3300
+    },
+    {
+      "epoch": 0.25565768131613503,
+      "grad_norm": 1.0,
+      "learning_rate": 4.361707205261031e-05,
+      "loss": 6.8824,
+      "step": 3310
+    },
+    {
+      "epoch": 0.2564300610179964,
+      "grad_norm": 1.03125,
+      "learning_rate": 4.3575272381791284e-05,
+      "loss": 6.8783,
+      "step": 3320
+    },
+    {
+      "epoch": 0.25720244071985787,
+      "grad_norm": 0.97265625,
+      "learning_rate": 4.353335646076062e-05,
+      "loss": 6.8892,
+      "step": 3330
+    },
+    {
+      "epoch": 0.2579748204217193,
+      "grad_norm": 0.984375,
+      "learning_rate": 4.349132455184202e-05,
+      "loss": 6.8845,
+      "step": 3340
+    },
+    {
+      "epoch": 0.25874720012358077,
+      "grad_norm": 1.0078125,
+      "learning_rate": 4.344917691808511e-05,
+      "loss": 6.9082,
+      "step": 3350
+    },
+    {
+      "epoch": 0.2595195798254422,
+      "grad_norm": 0.98046875,
+      "learning_rate": 4.340691382326375e-05,
+      "loss": 6.8759,
+      "step": 3360
+    },
+    {
+      "epoch": 0.2602919595273036,
+      "grad_norm": 1.0703125,
+      "learning_rate": 4.33645355318744e-05,
+      "loss": 6.8914,
+      "step": 3370
+    },
+    {
+      "epoch": 0.26106433922916505,
+      "grad_norm": 0.98828125,
+      "learning_rate": 4.332204230913445e-05,
+      "loss": 6.9077,
+      "step": 3380
+    },
+    {
+      "epoch": 0.2618367189310265,
+      "grad_norm": 1.0,
+      "learning_rate": 4.327943442098058e-05,
+      "loss": 6.8864,
+      "step": 3390
+    },
+    {
+      "epoch": 0.26260909863288795,
+      "grad_norm": 1.046875,
+      "learning_rate": 4.323671213406707e-05,
+      "loss": 6.879,
+      "step": 3400
+    },
+    {
+      "epoch": 0.26338147833474934,
+      "grad_norm": 1.0,
+      "learning_rate": 4.3193875715764163e-05,
+      "loss": 6.8829,
+      "step": 3410
+    },
+    {
+      "epoch": 0.2641538580366108,
+      "grad_norm": 0.98046875,
+      "learning_rate": 4.3150925434156355e-05,
+      "loss": 6.8839,
+      "step": 3420
+    },
+    {
+      "epoch": 0.26492623773847224,
+      "grad_norm": 0.98828125,
+      "learning_rate": 4.310786155804074e-05,
+      "loss": 6.8853,
+      "step": 3430
+    },
+    {
+      "epoch": 0.2656986174403337,
+      "grad_norm": 1.015625,
+      "learning_rate": 4.3064684356925336e-05,
+      "loss": 6.8669,
+      "step": 3440
+    },
+    {
+      "epoch": 0.2664709971421951,
+      "grad_norm": 1.0078125,
+      "learning_rate": 4.3021394101027386e-05,
+      "loss": 6.8854,
+      "step": 3450
+    },
+    {
+      "epoch": 0.2672433768440565,
+      "grad_norm": 0.9609375,
+      "learning_rate": 4.297799106127165e-05,
+      "loss": 6.8633,
+      "step": 3460
+    },
+    {
+      "epoch": 0.26801575654591797,
+      "grad_norm": 0.98828125,
+      "learning_rate": 4.2934475509288736e-05,
+      "loss": 6.8846,
+      "step": 3470
+    },
+    {
+      "epoch": 0.2687881362477794,
+      "grad_norm": 0.984375,
+      "learning_rate": 4.28908477174134e-05,
+      "loss": 6.8841,
+      "step": 3480
+    },
+    {
+      "epoch": 0.26956051594964087,
+      "grad_norm": 1.0234375,
+      "learning_rate": 4.284710795868282e-05,
+      "loss": 6.8811,
+      "step": 3490
+    },
+    {
+      "epoch": 0.27033289565150226,
+      "grad_norm": 1.03125,
+      "learning_rate": 4.280325650683491e-05,
+      "loss": 6.8686,
+      "step": 3500
+    },
+    {
+      "epoch": 0.2711052753533637,
+      "grad_norm": 1.0078125,
+      "learning_rate": 4.275929363630658e-05,
+      "loss": 6.8637,
+      "step": 3510
+    },
+    {
+      "epoch": 0.27187765505522515,
+      "grad_norm": 1.0234375,
+      "learning_rate": 4.271521962223206e-05,
+      "loss": 6.854,
+      "step": 3520
+    },
+    {
+      "epoch": 0.2726500347570866,
+      "grad_norm": 0.984375,
+      "learning_rate": 4.2671034740441134e-05,
+      "loss": 6.8826,
+      "step": 3530
+    },
+    {
+      "epoch": 0.273422414458948,
+      "grad_norm": 1.015625,
+      "learning_rate": 4.262673926745744e-05,
+      "loss": 6.8765,
+      "step": 3540
+    },
+    {
+      "epoch": 0.27419479416080944,
+      "grad_norm": 1.0078125,
+      "learning_rate": 4.2582333480496716e-05,
+      "loss": 6.8766,
+      "step": 3550
+    },
+    {
+      "epoch": 0.2749671738626709,
+      "grad_norm": 0.97265625,
+      "learning_rate": 4.253781765746511e-05,
+      "loss": 6.8937,
+      "step": 3560
+    },
+    {
+      "epoch": 0.27573955356453234,
+      "grad_norm": 1.0078125,
+      "learning_rate": 4.2493192076957395e-05,
+      "loss": 6.8719,
+      "step": 3570
+    },
+    {
+      "epoch": 0.2765119332663938,
+      "grad_norm": 1.0078125,
+      "learning_rate": 4.244845701825525e-05,
+      "loss": 6.8776,
+      "step": 3580
+    },
+    {
+      "epoch": 0.2772843129682552,
+      "grad_norm": 0.984375,
+      "learning_rate": 4.240361276132549e-05,
+      "loss": 6.8716,
+      "step": 3590
+    },
+    {
+      "epoch": 0.2780566926701166,
+      "grad_norm": 0.9765625,
+      "learning_rate": 4.235865958681835e-05,
+      "loss": 6.8821,
+      "step": 3600
+    },
+    {
+      "epoch": 0.27882907237197807,
+      "grad_norm": 1.0,
+      "learning_rate": 4.231359777606569e-05,
+      "loss": 6.8974,
+      "step": 3610
+    },
+    {
+      "epoch": 0.2796014520738395,
+      "grad_norm": 0.9921875,
+      "learning_rate": 4.226842761107926e-05,
+      "loss": 6.8608,
+      "step": 3620
+    },
+    {
+      "epoch": 0.2803738317757009,
+      "grad_norm": 1.03125,
+      "learning_rate": 4.222314937454892e-05,
+      "loss": 6.9033,
+      "step": 3630
+    },
+    {
+      "epoch": 0.28114621147756236,
+      "grad_norm": 1.0390625,
+      "learning_rate": 4.217776334984089e-05,
+      "loss": 6.8775,
+      "step": 3640
+    },
+    {
+      "epoch": 0.2819185911794238,
+      "grad_norm": 0.96875,
+      "learning_rate": 4.213226982099596e-05,
+      "loss": 6.8686,
+      "step": 3650
+    },
+    {
+      "epoch": 0.28269097088128525,
+      "grad_norm": 0.96875,
+      "learning_rate": 4.2086669072727706e-05,
+      "loss": 6.8794,
+      "step": 3660
+    },
+    {
+      "epoch": 0.2834633505831467,
+      "grad_norm": 1.0078125,
+      "learning_rate": 4.204096139042073e-05,
+      "loss": 6.8695,
+      "step": 3670
+    },
+    {
+      "epoch": 0.2842357302850081,
+      "grad_norm": 0.98046875,
+      "learning_rate": 4.1995147060128845e-05,
+      "loss": 6.8748,
+      "step": 3680
+    },
+    {
+      "epoch": 0.28500810998686954,
+      "grad_norm": 1.015625,
+      "learning_rate": 4.194922636857333e-05,
+      "loss": 6.8688,
+      "step": 3690
+    },
+    {
+      "epoch": 0.285780489688731,
+      "grad_norm": 0.9921875,
+      "learning_rate": 4.190319960314109e-05,
+      "loss": 6.8498,
+      "step": 3700
+    },
+    {
+      "epoch": 0.28655286939059244,
+      "grad_norm": 0.99609375,
+      "learning_rate": 4.185706705188289e-05,
+      "loss": 6.8917,
+      "step": 3710
+    },
+    {
+      "epoch": 0.2873252490924538,
+      "grad_norm": 1.0078125,
+      "learning_rate": 4.1810829003511496e-05,
+      "loss": 6.8745,
+      "step": 3720
+    },
+    {
+      "epoch": 0.2880976287943153,
+      "grad_norm": 0.98046875,
+      "learning_rate": 4.176448574739999e-05,
+      "loss": 6.8633,
+      "step": 3730
+    },
+    {
+      "epoch": 0.2888700084961767,
+      "grad_norm": 1.0234375,
+      "learning_rate": 4.171803757357977e-05,
+      "loss": 6.8865,
+      "step": 3740
+    },
+    {
+      "epoch": 0.28964238819803817,
+      "grad_norm": 1.296875,
+      "learning_rate": 4.167148477273896e-05,
+      "loss": 6.8663,
+      "step": 3750
+    },
+    {
+      "epoch": 0.2904147678998996,
+      "grad_norm": 0.99609375,
+      "learning_rate": 4.162482763622039e-05,
+      "loss": 6.8705,
+      "step": 3760
+    },
+    {
+      "epoch": 0.291187147601761,
+      "grad_norm": 1.0078125,
+      "learning_rate": 4.157806645601988e-05,
+      "loss": 6.8613,
+      "step": 3770
+    },
+    {
+      "epoch": 0.29195952730362246,
+      "grad_norm": 0.9921875,
+      "learning_rate": 4.1531201524784415e-05,
+      "loss": 6.8695,
+      "step": 3780
+    },
+    {
+      "epoch": 0.2927319070054839,
+      "grad_norm": 1.0078125,
+      "learning_rate": 4.1484233135810256e-05,
+      "loss": 6.8804,
+      "step": 3790
+    },
+    {
+      "epoch": 0.29350428670734535,
+      "grad_norm": 0.9609375,
+      "learning_rate": 4.1437161583041153e-05,
+      "loss": 6.8632,
+      "step": 3800
+    },
+    {
+      "epoch": 0.29427666640920674,
+      "grad_norm": 0.99609375,
+      "learning_rate": 4.1389987161066486e-05,
+      "loss": 6.8949,
+      "step": 3810
+    },
+    {
+      "epoch": 0.2950490461110682,
+      "grad_norm": 1.015625,
+      "learning_rate": 4.1342710165119424e-05,
+      "loss": 6.8764,
+      "step": 3820
+    },
+    {
+      "epoch": 0.29582142581292964,
+      "grad_norm": 0.96484375,
+      "learning_rate": 4.129533089107508e-05,
+      "loss": 6.8804,
+      "step": 3830
+    },
+    {
+      "epoch": 0.2965938055147911,
+      "grad_norm": 0.98828125,
+      "learning_rate": 4.1247849635448646e-05,
+      "loss": 6.8813,
+      "step": 3840
+    },
+    {
+      "epoch": 0.2973661852166525,
+      "grad_norm": 1.0,
+      "learning_rate": 4.120026669539357e-05,
+      "loss": 6.8804,
+      "step": 3850
+    },
+    {
+      "epoch": 0.2981385649185139,
+      "grad_norm": 0.98828125,
+      "learning_rate": 4.115258236869967e-05,
+      "loss": 6.8667,
+      "step": 3860
+    },
+    {
+      "epoch": 0.2989109446203754,
+      "grad_norm": 1.0,
+      "learning_rate": 4.110479695379126e-05,
+      "loss": 6.8738,
+      "step": 3870
+    },
+    {
+      "epoch": 0.2996833243222368,
+      "grad_norm": 1.0078125,
+      "learning_rate": 4.1056910749725284e-05,
+      "loss": 6.8655,
+      "step": 3880
+    },
+    {
+      "epoch": 0.30045570402409827,
+      "grad_norm": 0.96484375,
+      "learning_rate": 4.1008924056189515e-05,
+      "loss": 6.856,
+      "step": 3890
+    },
+    {
+      "epoch": 0.30122808372595966,
+      "grad_norm": 0.96484375,
+      "learning_rate": 4.096083717350056e-05,
+      "loss": 6.8701,
+      "step": 3900
+    },
+    {
+      "epoch": 0.3020004634278211,
+      "grad_norm": 1.0,
+      "learning_rate": 4.0912650402602094e-05,
+      "loss": 6.8704,
+      "step": 3910
+    },
+    {
+      "epoch": 0.30277284312968256,
+      "grad_norm": 0.984375,
+      "learning_rate": 4.086436404506288e-05,
+      "loss": 6.8535,
+      "step": 3920
+    },
+    {
+      "epoch": 0.303545222831544,
+      "grad_norm": 0.98046875,
+      "learning_rate": 4.081597840307496e-05,
+      "loss": 6.8587,
+      "step": 3930
+    },
+    {
+      "epoch": 0.3043176025334054,
+      "grad_norm": 1.0,
+      "learning_rate": 4.076749377945172e-05,
+      "loss": 6.862,
+      "step": 3940
+    },
+    {
+      "epoch": 0.30508998223526684,
+      "grad_norm": 0.9921875,
+      "learning_rate": 4.0718910477626004e-05,
+      "loss": 6.8891,
+      "step": 3950
+    },
+    {
+      "epoch": 0.3058623619371283,
+      "grad_norm": 0.984375,
+      "learning_rate": 4.0670228801648216e-05,
+      "loss": 6.8789,
+      "step": 3960
+    },
+    {
+      "epoch": 0.30663474163898974,
+      "grad_norm": 1.0078125,
+      "learning_rate": 4.0621449056184425e-05,
+      "loss": 6.8865,
+      "step": 3970
+    },
+    {
+      "epoch": 0.3074071213408512,
+      "grad_norm": 1.0078125,
+      "learning_rate": 4.057257154651444e-05,
+      "loss": 6.8527,
+      "step": 3980
+    },
+    {
+      "epoch": 0.3081795010427126,
+      "grad_norm": 0.99609375,
+      "learning_rate": 4.052359657852992e-05,
+      "loss": 6.8626,
+      "step": 3990
+    },
+    {
+      "epoch": 0.308951880744574,
+      "grad_norm": 0.97265625,
+      "learning_rate": 4.047452445873246e-05,
+      "loss": 6.8774,
+      "step": 4000
+    },
+    {
+      "epoch": 0.308951880744574,
+      "eval_loss": 7.063729763031006,
+      "eval_runtime": 0.6598,
+      "eval_samples_per_second": 18.187,
+      "eval_steps_per_second": 3.031,
+      "step": 4000
+    },
+    {
+      "epoch": 0.3097242604464355,
+      "grad_norm": 0.9765625,
+      "learning_rate": 4.0425355494231634e-05,
+      "loss": 6.8738,
+      "step": 4010
+    },
+    {
+      "epoch": 0.3104966401482969,
+      "grad_norm": 0.95703125,
+      "learning_rate": 4.037608999274314e-05,
+      "loss": 6.8691,
+      "step": 4020
+    },
+    {
+      "epoch": 0.3112690198501583,
+      "grad_norm": 0.99609375,
+      "learning_rate": 4.032672826258679e-05,
+      "loss": 6.8542,
+      "step": 4030
+    },
+    {
+      "epoch": 0.31204139955201976,
+      "grad_norm": 0.98046875,
+      "learning_rate": 4.027727061268467e-05,
+      "loss": 6.8479,
+      "step": 4040
+    },
+    {
+      "epoch": 0.3128137792538812,
+      "grad_norm": 0.98828125,
+      "learning_rate": 4.022771735255915e-05,
+      "loss": 6.8745,
+      "step": 4050
+    },
+    {
+      "epoch": 0.31358615895574266,
+      "grad_norm": 0.9921875,
+      "learning_rate": 4.017806879233095e-05,
+      "loss": 6.8743,
+      "step": 4060
+    },
+    {
+      "epoch": 0.3143585386576041,
+      "grad_norm": 0.96484375,
+      "learning_rate": 4.012832524271721e-05,
+      "loss": 6.8775,
+      "step": 4070
+    },
+    {
+      "epoch": 0.3151309183594655,
+      "grad_norm": 1.0234375,
+      "learning_rate": 4.0078487015029567e-05,
+      "loss": 6.8767,
+      "step": 4080
+    },
+    {
+      "epoch": 0.31590329806132694,
+      "grad_norm": 0.953125,
+      "learning_rate": 4.002855442117216e-05,
+      "loss": 6.8509,
+      "step": 4090
+    },
+    {
+      "epoch": 0.3166756777631884,
+      "grad_norm": 1.015625,
+      "learning_rate": 3.9978527773639715e-05,
+      "loss": 6.8595,
+      "step": 4100
+    },
+    {
+      "epoch": 0.31744805746504984,
+      "grad_norm": 0.96875,
+      "learning_rate": 3.9928407385515575e-05,
+      "loss": 6.8694,
+      "step": 4110
+    },
+    {
+      "epoch": 0.31822043716691123,
+      "grad_norm": 1.0234375,
+      "learning_rate": 3.9878193570469746e-05,
+      "loss": 6.8409,
+      "step": 4120
+    },
+    {
+      "epoch": 0.3189928168687727,
+      "grad_norm": 0.97265625,
+      "learning_rate": 3.982788664275692e-05,
+      "loss": 6.8401,
+      "step": 4130
+    },
+    {
+      "epoch": 0.3197651965706341,
+      "grad_norm": 1.0,
+      "learning_rate": 3.9777486917214524e-05,
+      "loss": 6.8685,
+      "step": 4140
+    },
+    {
+      "epoch": 0.3205375762724956,
+      "grad_norm": 0.95703125,
+      "learning_rate": 3.9726994709260746e-05,
+      "loss": 6.8817,
+      "step": 4150
+    },
+    {
+      "epoch": 0.321309955974357,
+      "grad_norm": 1.0,
+      "learning_rate": 3.9676410334892574e-05,
+      "loss": 6.8706,
+      "step": 4160
+    },
+    {
+      "epoch": 0.3220823356762184,
+      "grad_norm": 1.0234375,
+      "learning_rate": 3.962573411068377e-05,
+      "loss": 6.8734,
+      "step": 4170
+    },
+    {
+      "epoch": 0.32285471537807986,
+      "grad_norm": 1.0,
+      "learning_rate": 3.957496635378296e-05,
+      "loss": 6.8566,
+      "step": 4180
+    },
+    {
+      "epoch": 0.3236270950799413,
+      "grad_norm": 1.015625,
+      "learning_rate": 3.952410738191158e-05,
+      "loss": 6.8965,
+      "step": 4190
+    },
+    {
+      "epoch": 0.32439947478180275,
+      "grad_norm": 1.0,
+      "learning_rate": 3.9473157513361945e-05,
+      "loss": 6.8552,
+      "step": 4200
+    },
+    {
+      "epoch": 0.32517185448366415,
+      "grad_norm": 0.98828125,
+      "learning_rate": 3.9422117066995213e-05,
+      "loss": 6.8404,
+      "step": 4210
+    },
+    {
+      "epoch": 0.3259442341855256,
+      "grad_norm": 0.99609375,
+      "learning_rate": 3.937098636223943e-05,
+      "loss": 6.8643,
+      "step": 4220
+    },
+    {
+      "epoch": 0.32671661388738704,
+      "grad_norm": 0.99609375,
+      "learning_rate": 3.931976571908746e-05,
+      "loss": 6.8549,
+      "step": 4230
+    },
+    {
+      "epoch": 0.3274889935892485,
+      "grad_norm": 0.96875,
+      "learning_rate": 3.926845545809511e-05,
+      "loss": 6.8685,
+      "step": 4240
+    },
+    {
+      "epoch": 0.32826137329110994,
+      "grad_norm": 0.96484375,
+      "learning_rate": 3.921705590037897e-05,
+      "loss": 6.8504,
+      "step": 4250
+    },
+    {
+      "epoch": 0.32903375299297133,
+      "grad_norm": 0.95703125,
+      "learning_rate": 3.916556736761453e-05,
+      "loss": 6.8577,
+      "step": 4260
+    },
+    {
+      "epoch": 0.3298061326948328,
+      "grad_norm": 0.984375,
+      "learning_rate": 3.911399018203408e-05,
+      "loss": 6.86,
+      "step": 4270
+    },
+    {
+      "epoch": 0.3305785123966942,
+      "grad_norm": 0.984375,
+      "learning_rate": 3.906232466642475e-05,
+      "loss": 6.8517,
+      "step": 4280
+    },
+    {
+      "epoch": 0.33135089209855567,
+      "grad_norm": 1.015625,
+      "learning_rate": 3.901057114412646e-05,
+      "loss": 6.8543,
+      "step": 4290
+    },
+    {
+      "epoch": 0.33212327180041706,
+      "grad_norm": 1.3125,
+      "learning_rate": 3.89587299390299e-05,
+      "loss": 6.8725,
+      "step": 4300
+    },
+    {
+      "epoch": 0.3328956515022785,
+      "grad_norm": 0.99609375,
+      "learning_rate": 3.890680137557453e-05,
+      "loss": 6.8657,
+      "step": 4310
+    },
+    {
+      "epoch": 0.33366803120413996,
+      "grad_norm": 1.0390625,
+      "learning_rate": 3.8854785778746505e-05,
+      "loss": 6.8727,
+      "step": 4320
+    },
+    {
+      "epoch": 0.3344404109060014,
+      "grad_norm": 0.9609375,
+      "learning_rate": 3.880268347407667e-05,
+      "loss": 6.8567,
+      "step": 4330
+    },
+    {
+      "epoch": 0.3352127906078628,
+      "grad_norm": 0.984375,
+      "learning_rate": 3.8750494787638514e-05,
+      "loss": 6.8493,
+      "step": 4340
+    },
+    {
+      "epoch": 0.33598517030972425,
+      "grad_norm": 0.9765625,
+      "learning_rate": 3.869822004604615e-05,
+      "loss": 6.8699,
+      "step": 4350
+    },
+    {
+      "epoch": 0.3367575500115857,
+      "grad_norm": 1.0,
+      "learning_rate": 3.864585957645222e-05,
+      "loss": 6.8758,
+      "step": 4360
+    },
+    {
+      "epoch": 0.33752992971344714,
+      "grad_norm": 1.0,
+      "learning_rate": 3.859341370654591e-05,
+      "loss": 6.8695,
+      "step": 4370
+    },
+    {
+      "epoch": 0.3383023094153086,
+      "grad_norm": 1.015625,
+      "learning_rate": 3.8540882764550864e-05,
+      "loss": 6.8636,
+      "step": 4380
+    },
+    {
+      "epoch": 0.33907468911717,
+      "grad_norm": 1.015625,
+      "learning_rate": 3.848826707922311e-05,
+      "loss": 6.8448,
+      "step": 4390
+    },
+    {
+      "epoch": 0.33984706881903143,
+      "grad_norm": 0.99609375,
+      "learning_rate": 3.843556697984907e-05,
+      "loss": 6.8582,
+      "step": 4400
+    },
+    {
+      "epoch": 0.3406194485208929,
+      "grad_norm": 1.0078125,
+      "learning_rate": 3.8382782796243414e-05,
+      "loss": 6.8671,
+      "step": 4410
+    },
+    {
+      "epoch": 0.3413918282227543,
+      "grad_norm": 1.0,
+      "learning_rate": 3.832991485874706e-05,
+      "loss": 6.8367,
+      "step": 4420
+    },
+    {
+      "epoch": 0.3421642079246157,
+      "grad_norm": 0.953125,
+      "learning_rate": 3.82769634982251e-05,
+      "loss": 6.8499,
+      "step": 4430
+    },
+    {
+      "epoch": 0.34293658762647716,
+      "grad_norm": 0.9765625,
+      "learning_rate": 3.82239290460647e-05,
+      "loss": 6.8576,
+      "step": 4440
+    },
+    {
+      "epoch": 0.3437089673283386,
+      "grad_norm": 1.046875,
+      "learning_rate": 3.817081183417302e-05,
+      "loss": 6.8485,
+      "step": 4450
+    },
+    {
+      "epoch": 0.34448134703020006,
+      "grad_norm": 0.97265625,
+      "learning_rate": 3.81176121949752e-05,
+      "loss": 6.8648,
+      "step": 4460
+    },
+    {
+      "epoch": 0.3452537267320615,
+      "grad_norm": 0.97265625,
+      "learning_rate": 3.806433046141221e-05,
+      "loss": 6.8523,
+      "step": 4470
+    },
+    {
+      "epoch": 0.3460261064339229,
+      "grad_norm": 0.99609375,
+      "learning_rate": 3.801096696693879e-05,
+      "loss": 6.8566,
+      "step": 4480
+    },
+    {
+      "epoch": 0.34679848613578435,
+      "grad_norm": 0.96484375,
+      "learning_rate": 3.795752204552139e-05,
+      "loss": 6.8554,
+      "step": 4490
+    },
+    {
+      "epoch": 0.3475708658376458,
+      "grad_norm": 1.015625,
+      "learning_rate": 3.790399603163603e-05,
+      "loss": 6.8706,
+      "step": 4500
+    },
+    {
+      "epoch": 0.34834324553950724,
+      "grad_norm": 0.96875,
+      "learning_rate": 3.7850389260266265e-05,
+      "loss": 6.8654,
+      "step": 4510
+    },
+    {
+      "epoch": 0.34911562524136863,
+      "grad_norm": 1.1328125,
+      "learning_rate": 3.7796702066901016e-05,
+      "loss": 6.8633,
+      "step": 4520
+    },
+    {
+      "epoch": 0.3498880049432301,
+      "grad_norm": 1.0625,
+      "learning_rate": 3.774293478753255e-05,
+      "loss": 6.8582,
+      "step": 4530
+    },
+    {
+      "epoch": 0.3506603846450915,
+      "grad_norm": 0.94140625,
+      "learning_rate": 3.768908775865433e-05,
+      "loss": 6.8695,
+      "step": 4540
+    },
+    {
+      "epoch": 0.351432764346953,
+      "grad_norm": 0.9609375,
+      "learning_rate": 3.7635161317258896e-05,
+      "loss": 6.8649,
+      "step": 4550
+    },
+    {
+      "epoch": 0.3522051440488144,
+      "grad_norm": 0.97265625,
+      "learning_rate": 3.758115580083579e-05,
+      "loss": 6.8773,
+      "step": 4560
+    },
+    {
+      "epoch": 0.3529775237506758,
+      "grad_norm": 0.99609375,
+      "learning_rate": 3.752707154736945e-05,
+      "loss": 6.8594,
+      "step": 4570
+    },
+    {
+      "epoch": 0.35374990345253726,
+      "grad_norm": 0.96875,
+      "learning_rate": 3.747290889533707e-05,
+      "loss": 6.8624,
+      "step": 4580
+    },
+    {
+      "epoch": 0.3545222831543987,
+      "grad_norm": 0.99609375,
+      "learning_rate": 3.7418668183706444e-05,
+      "loss": 6.8572,
+      "step": 4590
+    },
+    {
+      "epoch": 0.35529466285626016,
+      "grad_norm": 0.96875,
+      "learning_rate": 3.7364349751933954e-05,
+      "loss": 6.8602,
+      "step": 4600
+    },
+    {
+      "epoch": 0.35606704255812155,
+      "grad_norm": 1.015625,
+      "learning_rate": 3.730995393996234e-05,
+      "loss": 6.8287,
+      "step": 4610
+    },
+    {
+      "epoch": 0.356839422259983,
+      "grad_norm": 1.0078125,
+      "learning_rate": 3.725548108821863e-05,
+      "loss": 6.8759,
+      "step": 4620
+    },
+    {
+      "epoch": 0.35761180196184444,
+      "grad_norm": 0.95703125,
+      "learning_rate": 3.720093153761197e-05,
+      "loss": 6.8598,
+      "step": 4630
+    },
+    {
+      "epoch": 0.3583841816637059,
+      "grad_norm": 1.0234375,
+      "learning_rate": 3.714630562953155e-05,
+      "loss": 6.845,
+      "step": 4640
+    },
+    {
+      "epoch": 0.35915656136556734,
+      "grad_norm": 1.0078125,
+      "learning_rate": 3.7091603705844404e-05,
+      "loss": 6.8466,
+      "step": 4650
+    },
+    {
+      "epoch": 0.35992894106742873,
+      "grad_norm": 0.97265625,
+      "learning_rate": 3.70368261088933e-05,
+      "loss": 6.8637,
+      "step": 4660
+    },
+    {
+      "epoch": 0.3607013207692902,
+      "grad_norm": 0.98046875,
+      "learning_rate": 3.6981973181494575e-05,
+      "loss": 6.8445,
+      "step": 4670
+    },
+    {
+      "epoch": 0.3614737004711516,
+      "grad_norm": 0.97265625,
+      "learning_rate": 3.692704526693607e-05,
+      "loss": 6.8595,
+      "step": 4680
+    },
+    {
+      "epoch": 0.3622460801730131,
+      "grad_norm": 1.03125,
+      "learning_rate": 3.6872042708974854e-05,
+      "loss": 6.8581,
+      "step": 4690
+    },
+    {
+      "epoch": 0.36301845987487447,
+      "grad_norm": 0.96875,
+      "learning_rate": 3.6816965851835185e-05,
+      "loss": 6.8532,
+      "step": 4700
+    },
+    {
+      "epoch": 0.3637908395767359,
+      "grad_norm": 0.97265625,
+      "learning_rate": 3.676181504020626e-05,
+      "loss": 6.8591,
+      "step": 4710
+    },
+    {
+      "epoch": 0.36456321927859736,
+      "grad_norm": 0.94140625,
+      "learning_rate": 3.670659061924018e-05,
+      "loss": 6.8465,
+      "step": 4720
+    },
+    {
+      "epoch": 0.3653355989804588,
+      "grad_norm": 0.98046875,
+      "learning_rate": 3.665129293454964e-05,
+      "loss": 6.857,
+      "step": 4730
+    },
+    {
+      "epoch": 0.3661079786823202,
+      "grad_norm": 0.99609375,
+      "learning_rate": 3.659592233220591e-05,
+      "loss": 6.8585,
+      "step": 4740
+    },
+    {
+      "epoch": 0.36688035838418165,
+      "grad_norm": 1.0234375,
+      "learning_rate": 3.6540479158736545e-05,
+      "loss": 6.8585,
+      "step": 4750
+    },
+    {
+      "epoch": 0.3676527380860431,
+      "grad_norm": 0.96484375,
+      "learning_rate": 3.648496376112333e-05,
+      "loss": 6.8336,
+      "step": 4760
+    },
+    {
+      "epoch": 0.36842511778790454,
+      "grad_norm": 0.9765625,
+      "learning_rate": 3.6429376486799994e-05,
+      "loss": 6.8759,
+      "step": 4770
+    },
+    {
+      "epoch": 0.369197497489766,
+      "grad_norm": 0.96875,
+      "learning_rate": 3.637371768365015e-05,
+      "loss": 6.8797,
+      "step": 4780
+    },
+    {
+      "epoch": 0.3699698771916274,
+      "grad_norm": 0.98046875,
+      "learning_rate": 3.6317987700005034e-05,
+      "loss": 6.8621,
+      "step": 4790
+    },
+    {
+      "epoch": 0.37074225689348883,
+      "grad_norm": 0.98828125,
+      "learning_rate": 3.626218688464135e-05,
+      "loss": 6.8494,
+      "step": 4800
+    },
+    {
+      "epoch": 0.3715146365953503,
+      "grad_norm": 0.9921875,
+      "learning_rate": 3.620631558677909e-05,
+      "loss": 6.8502,
+      "step": 4810
+    },
+    {
+      "epoch": 0.3722870162972117,
+      "grad_norm": 0.953125,
+      "learning_rate": 3.615037415607937e-05,
+      "loss": 6.8375,
+      "step": 4820
+    },
+    {
+      "epoch": 0.3730593959990731,
+      "grad_norm": 0.96875,
+      "learning_rate": 3.60943629426422e-05,
+      "loss": 6.858,
+      "step": 4830
+    },
+    {
+      "epoch": 0.37383177570093457,
+      "grad_norm": 0.96875,
+      "learning_rate": 3.6038282297004325e-05,
+      "loss": 6.8536,
+      "step": 4840
+    },
+    {
+      "epoch": 0.374604155402796,
+      "grad_norm": 0.94921875,
+      "learning_rate": 3.598213257013699e-05,
+      "loss": 6.8314,
+      "step": 4850
+    },
+    {
+      "epoch": 0.37537653510465746,
+      "grad_norm": 0.984375,
+      "learning_rate": 3.592591411344383e-05,
+      "loss": 6.8452,
+      "step": 4860
+    },
+    {
+      "epoch": 0.3761489148065189,
+      "grad_norm": 1.0078125,
+      "learning_rate": 3.5869627278758563e-05,
+      "loss": 6.8402,
+      "step": 4870
+    },
+    {
+      "epoch": 0.3769212945083803,
+      "grad_norm": 0.97265625,
+      "learning_rate": 3.581327241834287e-05,
+      "loss": 6.8416,
+      "step": 4880
+    },
+    {
+      "epoch": 0.37769367421024175,
+      "grad_norm": 0.96484375,
+      "learning_rate": 3.575684988488412e-05,
+      "loss": 6.8549,
+      "step": 4890
+    },
+    {
+      "epoch": 0.3784660539121032,
+      "grad_norm": 0.96484375,
+      "learning_rate": 3.570036003149327e-05,
+      "loss": 6.8531,
+      "step": 4900
+    },
+    {
+      "epoch": 0.37923843361396464,
+      "grad_norm": 0.9765625,
+      "learning_rate": 3.564380321170252e-05,
+      "loss": 6.8578,
+      "step": 4910
+    },
+    {
+      "epoch": 0.38001081331582603,
+      "grad_norm": 0.99609375,
+      "learning_rate": 3.55871797794632e-05,
+      "loss": 6.866,
+      "step": 4920
+    },
+    {
+      "epoch": 0.3807831930176875,
+      "grad_norm": 1.0,
+      "learning_rate": 3.553049008914352e-05,
+      "loss": 6.8638,
+      "step": 4930
+    },
+    {
+      "epoch": 0.38155557271954893,
+      "grad_norm": 0.9453125,
+      "learning_rate": 3.547373449552635e-05,
+      "loss": 6.8245,
+      "step": 4940
+    },
+    {
+      "epoch": 0.3823279524214104,
+      "grad_norm": 1.0234375,
+      "learning_rate": 3.5416913353807e-05,
+      "loss": 6.8403,
+      "step": 4950
+    },
+    {
+      "epoch": 0.3831003321232718,
+      "grad_norm": 0.95703125,
+      "learning_rate": 3.536002701959103e-05,
+      "loss": 6.8333,
+      "step": 4960
+    },
+    {
+      "epoch": 0.3838727118251332,
+      "grad_norm": 0.953125,
+      "learning_rate": 3.5303075848891963e-05,
+      "loss": 6.8469,
+      "step": 4970
+    },
+    {
+      "epoch": 0.38464509152699466,
+      "grad_norm": 0.97265625,
+      "learning_rate": 3.52460601981291e-05,
+      "loss": 6.8639,
+      "step": 4980
+    },
+    {
+      "epoch": 0.3854174712288561,
+      "grad_norm": 1.109375,
+      "learning_rate": 3.518898042412529e-05,
+      "loss": 6.8447,
+      "step": 4990
+    },
+    {
+      "epoch": 0.38618985093071756,
+      "grad_norm": 0.9765625,
+      "learning_rate": 3.5131836884104665e-05,
+      "loss": 6.8551,
+      "step": 5000
+    },
+    {
+      "epoch": 0.38618985093071756,
+      "eval_loss": 7.044839859008789,
+      "eval_runtime": 0.6994,
+      "eval_samples_per_second": 17.158,
+      "eval_steps_per_second": 2.86,
+      "step": 5000
+    },
+    {
+      "epoch": 0.38696223063257895,
+      "grad_norm": 0.98046875,
+      "learning_rate": 3.507462993569046e-05,
+      "loss": 6.8331,
+      "step": 5010
+    },
+    {
+      "epoch": 0.3877346103344404,
+      "grad_norm": 0.98828125,
+      "learning_rate": 3.5017359936902705e-05,
+      "loss": 6.8634,
+      "step": 5020
+    },
+    {
+      "epoch": 0.38850699003630185,
+      "grad_norm": 0.94921875,
+      "learning_rate": 3.496002724615604e-05,
+      "loss": 6.8505,
+      "step": 5030
+    },
+    {
+      "epoch": 0.3892793697381633,
+      "grad_norm": 0.95703125,
+      "learning_rate": 3.490263222225745e-05,
+      "loss": 6.8628,
+      "step": 5040
+    },
+    {
+      "epoch": 0.39005174944002474,
+      "grad_norm": 0.98828125,
+      "learning_rate": 3.484517522440401e-05,
+      "loss": 6.8415,
+      "step": 5050
+    },
+    {
+      "epoch": 0.39082412914188613,
+      "grad_norm": 1.0234375,
+      "learning_rate": 3.478765661218067e-05,
+      "loss": 6.854,
+      "step": 5060
+    },
+    {
+      "epoch": 0.3915965088437476,
+      "grad_norm": 0.9453125,
+      "learning_rate": 3.473007674555797e-05,
+      "loss": 6.8754,
+      "step": 5070
+    },
+    {
+      "epoch": 0.39236888854560903,
+      "grad_norm": 0.953125,
+      "learning_rate": 3.467243598488978e-05,
+      "loss": 6.8701,
+      "step": 5080
+    },
+    {
+      "epoch": 0.3931412682474705,
+      "grad_norm": 0.9765625,
+      "learning_rate": 3.461473469091112e-05,
+      "loss": 6.8428,
+      "step": 5090
+    },
+    {
+      "epoch": 0.39391364794933187,
+      "grad_norm": 0.953125,
+      "learning_rate": 3.455697322473579e-05,
+      "loss": 6.8462,
+      "step": 5100
+    },
+    {
+      "epoch": 0.3946860276511933,
+      "grad_norm": 0.98046875,
+      "learning_rate": 3.44991519478542e-05,
+      "loss": 6.8507,
+      "step": 5110
+    },
+    {
+      "epoch": 0.39545840735305476,
+      "grad_norm": 0.99609375,
+      "learning_rate": 3.4441271222131056e-05,
+      "loss": 6.8176,
+      "step": 5120
+    },
+    {
+      "epoch": 0.3962307870549162,
+      "grad_norm": 0.984375,
+      "learning_rate": 3.4383331409803144e-05,
+      "loss": 6.8567,
+      "step": 5130
+    },
+    {
+      "epoch": 0.39700316675677766,
+      "grad_norm": 0.953125,
+      "learning_rate": 3.432533287347699e-05,
+      "loss": 6.8461,
+      "step": 5140
+    },
+    {
+      "epoch": 0.39777554645863905,
+      "grad_norm": 0.984375,
+      "learning_rate": 3.426727597612669e-05,
+      "loss": 6.8656,
+      "step": 5150
+    },
+    {
+      "epoch": 0.3985479261605005,
+      "grad_norm": 0.96484375,
+      "learning_rate": 3.420916108109153e-05,
+      "loss": 6.8674,
+      "step": 5160
+    },
+    {
+      "epoch": 0.39932030586236195,
+      "grad_norm": 0.96484375,
+      "learning_rate": 3.415098855207379e-05,
+      "loss": 6.8727,
+      "step": 5170
+    },
+    {
+      "epoch": 0.4000926855642234,
+      "grad_norm": 0.9921875,
+      "learning_rate": 3.4092758753136436e-05,
+      "loss": 6.85,
+      "step": 5180
+    },
+    {
+      "epoch": 0.4008650652660848,
+      "grad_norm": 0.9765625,
+      "learning_rate": 3.403447204870086e-05,
+      "loss": 6.8666,
+      "step": 5190
+    },
+    {
+      "epoch": 0.40163744496794623,
+      "grad_norm": 1.0625,
+      "learning_rate": 3.3976128803544566e-05,
+      "loss": 6.8557,
+      "step": 5200
+    },
+    {
+      "epoch": 0.4024098246698077,
+      "grad_norm": 0.99609375,
+      "learning_rate": 3.391772938279894e-05,
+      "loss": 6.8489,
+      "step": 5210
+    },
+    {
+      "epoch": 0.40318220437166913,
+      "grad_norm": 1.0234375,
+      "learning_rate": 3.385927415194689e-05,
+      "loss": 6.8627,
+      "step": 5220
+    },
+    {
+      "epoch": 0.4039545840735305,
+      "grad_norm": 0.94921875,
+      "learning_rate": 3.380076347682065e-05,
+      "loss": 6.8479,
+      "step": 5230
+    },
+    {
+      "epoch": 0.40472696377539197,
+      "grad_norm": 0.98828125,
+      "learning_rate": 3.37421977235994e-05,
+      "loss": 6.8395,
+      "step": 5240
+    },
+    {
+      "epoch": 0.4054993434772534,
+      "grad_norm": 0.9765625,
+      "learning_rate": 3.3683577258807056e-05,
+      "loss": 6.8601,
+      "step": 5250
+    },
+    {
+      "epoch": 0.40627172317911486,
+      "grad_norm": 0.9765625,
+      "learning_rate": 3.36249024493099e-05,
+      "loss": 6.839,
+      "step": 5260
+    },
+    {
+      "epoch": 0.4070441028809763,
+      "grad_norm": 1.015625,
+      "learning_rate": 3.356617366231434e-05,
+      "loss": 6.8416,
+      "step": 5270
+    },
+    {
+      "epoch": 0.4078164825828377,
+      "grad_norm": 0.9765625,
+      "learning_rate": 3.35073912653646e-05,
+      "loss": 6.8659,
+      "step": 5280
+    },
+    {
+      "epoch": 0.40858886228469915,
+      "grad_norm": 0.97265625,
+      "learning_rate": 3.3448555626340404e-05,
+      "loss": 6.8331,
+      "step": 5290
+    },
+    {
+      "epoch": 0.4093612419865606,
+      "grad_norm": 0.98046875,
+      "learning_rate": 3.338966711345468e-05,
+      "loss": 6.8459,
+      "step": 5300
+    },
+    {
+      "epoch": 0.41013362168842205,
+      "grad_norm": 1.015625,
+      "learning_rate": 3.333072609525127e-05,
+      "loss": 6.886,
+      "step": 5310
+    },
+    {
+      "epoch": 0.41090600139028344,
+      "grad_norm": 0.9765625,
+      "learning_rate": 3.327173294060258e-05,
+      "loss": 6.839,
+      "step": 5320
+    },
+    {
+      "epoch": 0.4116783810921449,
+      "grad_norm": 0.98046875,
+      "learning_rate": 3.3212688018707355e-05,
+      "loss": 6.8351,
+      "step": 5330
+    },
+    {
+      "epoch": 0.41245076079400633,
+      "grad_norm": 0.95703125,
+      "learning_rate": 3.3153591699088264e-05,
+      "loss": 6.8359,
+      "step": 5340
+    },
+    {
+      "epoch": 0.4132231404958678,
+      "grad_norm": 0.97265625,
+      "learning_rate": 3.3094444351589675e-05,
+      "loss": 6.8497,
+      "step": 5350
+    },
+    {
+      "epoch": 0.4139955201977292,
+      "grad_norm": 0.9921875,
+      "learning_rate": 3.303524634637529e-05,
+      "loss": 6.8488,
+      "step": 5360
+    },
+    {
+      "epoch": 0.4147678998995906,
+      "grad_norm": 0.9609375,
+      "learning_rate": 3.2975998053925856e-05,
+      "loss": 6.8519,
+      "step": 5370
+    },
+    {
+      "epoch": 0.41554027960145207,
+      "grad_norm": 0.98046875,
+      "learning_rate": 3.2916699845036816e-05,
+      "loss": 6.8321,
+      "step": 5380
+    },
+    {
+      "epoch": 0.4163126593033135,
+      "grad_norm": 0.99609375,
+      "learning_rate": 3.285735209081602e-05,
+      "loss": 6.8448,
+      "step": 5390
+    },
+    {
+      "epoch": 0.41708503900517496,
+      "grad_norm": 0.99609375,
+      "learning_rate": 3.279795516268139e-05,
+      "loss": 6.8521,
+      "step": 5400
+    },
+    {
+      "epoch": 0.41785741870703635,
+      "grad_norm": 0.98046875,
+      "learning_rate": 3.2738509432358586e-05,
+      "loss": 6.8353,
+      "step": 5410
+    },
+    {
+      "epoch": 0.4186297984088978,
+      "grad_norm": 0.96875,
+      "learning_rate": 3.267901527187869e-05,
+      "loss": 6.8712,
+      "step": 5420
+    },
+    {
+      "epoch": 0.41940217811075925,
+      "grad_norm": 1.0,
+      "learning_rate": 3.261947305357588e-05,
+      "loss": 6.8533,
+      "step": 5430
+    },
+    {
+      "epoch": 0.4201745578126207,
+      "grad_norm": 1.0,
+      "learning_rate": 3.2559883150085104e-05,
+      "loss": 6.8552,
+      "step": 5440
+    },
+    {
+      "epoch": 0.42094693751448214,
+      "grad_norm": 0.9765625,
+      "learning_rate": 3.250024593433971e-05,
+      "loss": 6.8444,
+      "step": 5450
+    },
+    {
+      "epoch": 0.42171931721634354,
+      "grad_norm": 1.0390625,
+      "learning_rate": 3.244056177956919e-05,
+      "loss": 6.8675,
+      "step": 5460
+    },
+    {
+      "epoch": 0.422491696918205,
+      "grad_norm": 0.9765625,
+      "learning_rate": 3.238083105929674e-05,
+      "loss": 6.852,
+      "step": 5470
+    },
+    {
+      "epoch": 0.42326407662006643,
+      "grad_norm": 1.0078125,
+      "learning_rate": 3.232105414733703e-05,
+      "loss": 6.8369,
+      "step": 5480
+    },
+    {
+      "epoch": 0.4240364563219279,
+      "grad_norm": 1.03125,
+      "learning_rate": 3.226123141779376e-05,
+      "loss": 6.8513,
+      "step": 5490
+    },
+    {
+      "epoch": 0.42480883602378927,
+      "grad_norm": 0.98046875,
+      "learning_rate": 3.220136324505744e-05,
+      "loss": 6.8678,
+      "step": 5500
+    },
+    {
+      "epoch": 0.4255812157256507,
+      "grad_norm": 1.0078125,
+      "learning_rate": 3.21414500038029e-05,
+      "loss": 6.8557,
+      "step": 5510
+    },
+    {
+      "epoch": 0.42635359542751217,
+      "grad_norm": 0.9375,
+      "learning_rate": 3.20814920689871e-05,
+      "loss": 6.8482,
+      "step": 5520
+    },
+    {
+      "epoch": 0.4271259751293736,
+      "grad_norm": 0.9921875,
+      "learning_rate": 3.2021489815846625e-05,
+      "loss": 6.8521,
+      "step": 5530
+    },
+    {
+      "epoch": 0.42789835483123506,
+      "grad_norm": 0.9921875,
+      "learning_rate": 3.196144361989551e-05,
+      "loss": 6.867,
+      "step": 5540
+    },
+    {
+      "epoch": 0.42867073453309645,
+      "grad_norm": 0.96875,
+      "learning_rate": 3.1901353856922725e-05,
+      "loss": 6.856,
+      "step": 5550
+    },
+    {
+      "epoch": 0.4294431142349579,
+      "grad_norm": 0.92578125,
+      "learning_rate": 3.184122090298994e-05,
+      "loss": 6.8535,
+      "step": 5560
+    },
+    {
+      "epoch": 0.43021549393681935,
+      "grad_norm": 0.9765625,
+      "learning_rate": 3.178104513442909e-05,
+      "loss": 6.8453,
+      "step": 5570
+    },
+    {
+      "epoch": 0.4309878736386808,
+      "grad_norm": 0.9609375,
+      "learning_rate": 3.172082692784011e-05,
+      "loss": 6.8447,
+      "step": 5580
+    },
+    {
+      "epoch": 0.4317602533405422,
+      "grad_norm": 0.984375,
+      "learning_rate": 3.166056666008847e-05,
+      "loss": 6.8317,
+      "step": 5590
+    },
+    {
+      "epoch": 0.43253263304240364,
+      "grad_norm": 0.9453125,
+      "learning_rate": 3.160026470830292e-05,
+      "loss": 6.8224,
+      "step": 5600
+    },
+    {
+      "epoch": 0.4333050127442651,
+      "grad_norm": 1.015625,
+      "learning_rate": 3.153992144987306e-05,
+      "loss": 6.8339,
+      "step": 5610
+    },
+    {
+      "epoch": 0.43407739244612653,
+      "grad_norm": 0.98828125,
+      "learning_rate": 3.1479537262446996e-05,
+      "loss": 6.8412,
+      "step": 5620
+    },
+    {
+      "epoch": 0.434849772147988,
+      "grad_norm": 1.0078125,
+      "learning_rate": 3.141911252392898e-05,
+      "loss": 6.8494,
+      "step": 5630
+    },
+    {
+      "epoch": 0.43562215184984937,
+      "grad_norm": 0.93359375,
+      "learning_rate": 3.135864761247708e-05,
+      "loss": 6.86,
+      "step": 5640
+    },
+    {
+      "epoch": 0.4363945315517108,
+      "grad_norm": 0.98828125,
+      "learning_rate": 3.1298142906500724e-05,
+      "loss": 6.8555,
+      "step": 5650
+    },
+    {
+      "epoch": 0.43716691125357227,
+      "grad_norm": 0.9921875,
+      "learning_rate": 3.1237598784658446e-05,
+      "loss": 6.8476,
+      "step": 5660
+    },
+    {
+      "epoch": 0.4379392909554337,
+      "grad_norm": 0.984375,
+      "learning_rate": 3.11770156258554e-05,
+      "loss": 6.8556,
+      "step": 5670
+    },
+    {
+      "epoch": 0.4387116706572951,
+      "grad_norm": 0.97265625,
+      "learning_rate": 3.11163938092411e-05,
+      "loss": 6.8437,
+      "step": 5680
+    },
+    {
+      "epoch": 0.43948405035915655,
+      "grad_norm": 0.984375,
+      "learning_rate": 3.105573371420697e-05,
+      "loss": 6.8287,
+      "step": 5690
+    },
+    {
+      "epoch": 0.440256430061018,
+      "grad_norm": 0.984375,
+      "learning_rate": 3.099503572038396e-05,
+      "loss": 6.8604,
+      "step": 5700
+    },
+    {
+      "epoch": 0.44102880976287945,
+      "grad_norm": 0.9921875,
+      "learning_rate": 3.093430020764027e-05,
+      "loss": 6.8329,
+      "step": 5710
+    },
+    {
+      "epoch": 0.44180118946474084,
+      "grad_norm": 0.984375,
+      "learning_rate": 3.087352755607884e-05,
+      "loss": 6.8534,
+      "step": 5720
+    },
+    {
+      "epoch": 0.4425735691666023,
+      "grad_norm": 0.96875,
+      "learning_rate": 3.0812718146035094e-05,
+      "loss": 6.8614,
+      "step": 5730
+    },
+    {
+      "epoch": 0.44334594886846374,
+      "grad_norm": 1.015625,
+      "learning_rate": 3.075187235807446e-05,
+      "loss": 6.8376,
+      "step": 5740
+    },
+    {
+      "epoch": 0.4441183285703252,
+      "grad_norm": 0.96484375,
+      "learning_rate": 3.069099057299006e-05,
+      "loss": 6.8322,
+      "step": 5750
+    },
+    {
+      "epoch": 0.44489070827218663,
+      "grad_norm": 1.0078125,
+      "learning_rate": 3.0630073171800274e-05,
+      "loss": 6.8479,
+      "step": 5760
+    },
+    {
+      "epoch": 0.445663087974048,
+      "grad_norm": 0.98828125,
+      "learning_rate": 3.05691205357464e-05,
+      "loss": 6.8455,
+      "step": 5770
+    },
+    {
+      "epoch": 0.44643546767590947,
+      "grad_norm": 0.9765625,
+      "learning_rate": 3.0508133046290222e-05,
+      "loss": 6.8285,
+      "step": 5780
+    },
+    {
+      "epoch": 0.4472078473777709,
+      "grad_norm": 0.953125,
+      "learning_rate": 3.0447111085111685e-05,
+      "loss": 6.8536,
+      "step": 5790
+    },
+    {
+      "epoch": 0.44798022707963236,
+      "grad_norm": 1.0234375,
+      "learning_rate": 3.0386055034106436e-05,
+      "loss": 6.8497,
+      "step": 5800
+    },
+    {
+      "epoch": 0.44875260678149376,
+      "grad_norm": 0.9765625,
+      "learning_rate": 3.032496527538349e-05,
+      "loss": 6.8544,
+      "step": 5810
+    },
+    {
+      "epoch": 0.4495249864833552,
+      "grad_norm": 1.015625,
+      "learning_rate": 3.0263842191262785e-05,
+      "loss": 6.8447,
+      "step": 5820
+    },
+    {
+      "epoch": 0.45029736618521665,
+      "grad_norm": 1.03125,
+      "learning_rate": 3.0202686164272875e-05,
+      "loss": 6.848,
+      "step": 5830
+    },
+    {
+      "epoch": 0.4510697458870781,
+      "grad_norm": 0.9375,
+      "learning_rate": 3.0141497577148414e-05,
+      "loss": 6.846,
+      "step": 5840
+    },
+    {
+      "epoch": 0.45184212558893955,
+      "grad_norm": 0.97265625,
+      "learning_rate": 3.008027681282789e-05,
+      "loss": 6.8683,
+      "step": 5850
+    },
+    {
+      "epoch": 0.45261450529080094,
+      "grad_norm": 0.984375,
+      "learning_rate": 3.0019024254451117e-05,
+      "loss": 6.8317,
+      "step": 5860
+    },
+    {
+      "epoch": 0.4533868849926624,
+      "grad_norm": 0.9453125,
+      "learning_rate": 2.9957740285356932e-05,
+      "loss": 6.8394,
+      "step": 5870
+    },
+    {
+      "epoch": 0.45415926469452383,
+      "grad_norm": 0.9765625,
+      "learning_rate": 2.9896425289080706e-05,
+      "loss": 6.8416,
+      "step": 5880
+    },
+    {
+      "epoch": 0.4549316443963853,
+      "grad_norm": 0.98828125,
+      "learning_rate": 2.9835079649352016e-05,
+      "loss": 6.8535,
+      "step": 5890
+    },
+    {
+      "epoch": 0.4557040240982467,
+      "grad_norm": 0.9921875,
+      "learning_rate": 2.977370375009221e-05,
+      "loss": 6.8316,
+      "step": 5900
+    },
+    {
+      "epoch": 0.4564764038001081,
+      "grad_norm": 0.984375,
+      "learning_rate": 2.9712297975412014e-05,
+      "loss": 6.847,
+      "step": 5910
+    },
+    {
+      "epoch": 0.45724878350196957,
+      "grad_norm": 0.9609375,
+      "learning_rate": 2.9650862709609113e-05,
+      "loss": 6.8373,
+      "step": 5920
+    },
+    {
+      "epoch": 0.458021163203831,
+      "grad_norm": 1.0390625,
+      "learning_rate": 2.958939833716577e-05,
+      "loss": 6.8555,
+      "step": 5930
+    },
+    {
+      "epoch": 0.45879354290569246,
+      "grad_norm": 0.9453125,
+      "learning_rate": 2.9527905242746394e-05,
+      "loss": 6.8214,
+      "step": 5940
+    },
+    {
+      "epoch": 0.45956592260755386,
+      "grad_norm": 0.953125,
+      "learning_rate": 2.9466383811195163e-05,
+      "loss": 6.855,
+      "step": 5950
+    },
+    {
+      "epoch": 0.4603383023094153,
+      "grad_norm": 0.984375,
+      "learning_rate": 2.9404834427533578e-05,
+      "loss": 6.8496,
+      "step": 5960
+    },
+    {
+      "epoch": 0.46111068201127675,
+      "grad_norm": 0.99609375,
+      "learning_rate": 2.9343257476958093e-05,
+      "loss": 6.8341,
+      "step": 5970
+    },
+    {
+      "epoch": 0.4618830617131382,
+      "grad_norm": 0.96484375,
+      "learning_rate": 2.9281653344837666e-05,
+      "loss": 6.8292,
+      "step": 5980
+    },
+    {
+      "epoch": 0.4626554414149996,
+      "grad_norm": 0.9609375,
+      "learning_rate": 2.9220022416711385e-05,
+      "loss": 6.856,
+      "step": 5990
+    },
+    {
+      "epoch": 0.46342782111686104,
+      "grad_norm": 0.96484375,
+      "learning_rate": 2.9158365078286017e-05,
+      "loss": 6.8365,
+      "step": 6000
+    },
+    {
+      "epoch": 0.46342782111686104,
+      "eval_loss": 7.034400939941406,
+      "eval_runtime": 0.6635,
+      "eval_samples_per_second": 18.085,
+      "eval_steps_per_second": 3.014,
+      "step": 6000
+    },
+    {
+      "epoch": 0.4642002008187225,
+      "grad_norm": 0.984375,
+      "learning_rate": 2.9096681715433633e-05,
+      "loss": 6.838,
+      "step": 6010
+    },
+    {
+      "epoch": 0.46497258052058393,
+      "grad_norm": 1.015625,
+      "learning_rate": 2.903497271418915e-05,
+      "loss": 6.8395,
+      "step": 6020
+    },
+    {
+      "epoch": 0.4657449602224454,
+      "grad_norm": 1.0078125,
+      "learning_rate": 2.897323846074797e-05,
+      "loss": 6.8484,
+      "step": 6030
+    },
+    {
+      "epoch": 0.4665173399243068,
+      "grad_norm": 0.953125,
+      "learning_rate": 2.8911479341463503e-05,
+      "loss": 6.8481,
+      "step": 6040
+    },
+    {
+      "epoch": 0.4672897196261682,
+      "grad_norm": 0.99609375,
+      "learning_rate": 2.884969574284479e-05,
+      "loss": 6.8246,
+      "step": 6050
+    },
+    {
+      "epoch": 0.46806209932802967,
+      "grad_norm": 0.9921875,
+      "learning_rate": 2.8787888051554074e-05,
+      "loss": 6.8388,
+      "step": 6060
+    },
+    {
+      "epoch": 0.4688344790298911,
+      "grad_norm": 1.1015625,
+      "learning_rate": 2.872605665440436e-05,
+      "loss": 6.8552,
+      "step": 6070
+    },
+    {
+      "epoch": 0.4696068587317525,
+      "grad_norm": 0.94140625,
+      "learning_rate": 2.866420193835705e-05,
+      "loss": 6.8579,
+      "step": 6080
+    },
+    {
+      "epoch": 0.47037923843361396,
+      "grad_norm": 0.94921875,
+      "learning_rate": 2.8602324290519444e-05,
+      "loss": 6.8285,
+      "step": 6090
+    },
+    {
+      "epoch": 0.4711516181354754,
+      "grad_norm": 1.0234375,
+      "learning_rate": 2.8540424098142378e-05,
+      "loss": 6.8252,
+      "step": 6100
+    },
+    {
+      "epoch": 0.47192399783733685,
+      "grad_norm": 0.9609375,
+      "learning_rate": 2.847850174861776e-05,
+      "loss": 6.817,
+      "step": 6110
+    },
+    {
+      "epoch": 0.47269637753919824,
+      "grad_norm": 0.98828125,
+      "learning_rate": 2.8416557629476208e-05,
+      "loss": 6.8538,
+      "step": 6120
+    },
+    {
+      "epoch": 0.4734687572410597,
+      "grad_norm": 0.96875,
+      "learning_rate": 2.8354592128384522e-05,
+      "loss": 6.8479,
+      "step": 6130
+    },
+    {
+      "epoch": 0.47424113694292114,
+      "grad_norm": 0.95703125,
+      "learning_rate": 2.8292605633143367e-05,
+      "loss": 6.8364,
+      "step": 6140
+    },
+    {
+      "epoch": 0.4750135166447826,
+      "grad_norm": 0.96484375,
+      "learning_rate": 2.8230598531684765e-05,
+      "loss": 6.8607,
+      "step": 6150
+    },
+    {
+      "epoch": 0.47578589634664403,
+      "grad_norm": 0.96875,
+      "learning_rate": 2.8168571212069715e-05,
+      "loss": 6.8197,
+      "step": 6160
+    },
+    {
+      "epoch": 0.4765582760485054,
+      "grad_norm": 0.98828125,
+      "learning_rate": 2.810652406248574e-05,
+      "loss": 6.8443,
+      "step": 6170
+    },
+    {
+      "epoch": 0.4773306557503669,
+      "grad_norm": 0.96484375,
+      "learning_rate": 2.8044457471244478e-05,
+      "loss": 6.8261,
+      "step": 6180
+    },
+    {
+      "epoch": 0.4781030354522283,
+      "grad_norm": 0.98046875,
+      "learning_rate": 2.798237182677922e-05,
+      "loss": 6.851,
+      "step": 6190
+    },
+    {
+      "epoch": 0.47887541515408977,
+      "grad_norm": 1.0078125,
+      "learning_rate": 2.7920267517642513e-05,
+      "loss": 6.837,
+      "step": 6200
+    },
+    {
+      "epoch": 0.47964779485595116,
+      "grad_norm": 0.9921875,
+      "learning_rate": 2.7858144932503705e-05,
+      "loss": 6.8447,
+      "step": 6210
+    },
+    {
+      "epoch": 0.4804201745578126,
+      "grad_norm": 0.984375,
+      "learning_rate": 2.779600446014653e-05,
+      "loss": 6.8337,
+      "step": 6220
+    },
+    {
+      "epoch": 0.48119255425967405,
+      "grad_norm": 0.9609375,
+      "learning_rate": 2.7733846489466663e-05,
+      "loss": 6.8325,
+      "step": 6230
+    },
+    {
+      "epoch": 0.4819649339615355,
+      "grad_norm": 1.0078125,
+      "learning_rate": 2.7671671409469285e-05,
+      "loss": 6.8274,
+      "step": 6240
+    },
+    {
+      "epoch": 0.48273731366339695,
+      "grad_norm": 1.0078125,
+      "learning_rate": 2.760947960926665e-05,
+      "loss": 6.8467,
+      "step": 6250
+    },
+    {
+      "epoch": 0.48350969336525834,
+      "grad_norm": 0.98828125,
+      "learning_rate": 2.7547271478075675e-05,
+      "loss": 6.8435,
+      "step": 6260
+    },
+    {
+      "epoch": 0.4842820730671198,
+      "grad_norm": 0.94140625,
+      "learning_rate": 2.7485047405215443e-05,
+      "loss": 6.8219,
+      "step": 6270
+    },
+    {
+      "epoch": 0.48505445276898124,
+      "grad_norm": 0.9765625,
+      "learning_rate": 2.7422807780104842e-05,
+      "loss": 6.8419,
+      "step": 6280
+    },
+    {
+      "epoch": 0.4858268324708427,
+      "grad_norm": 0.97265625,
+      "learning_rate": 2.736055299226007e-05,
+      "loss": 6.8228,
+      "step": 6290
+    },
+    {
+      "epoch": 0.4865992121727041,
+      "grad_norm": 0.98046875,
+      "learning_rate": 2.7298283431292232e-05,
+      "loss": 6.8402,
+      "step": 6300
+    },
+    {
+      "epoch": 0.4873715918745655,
+      "grad_norm": 1.0,
+      "learning_rate": 2.7235999486904867e-05,
+      "loss": 6.8134,
+      "step": 6310
+    },
+    {
+      "epoch": 0.48814397157642697,
+      "grad_norm": 1.0,
+      "learning_rate": 2.7173701548891556e-05,
+      "loss": 6.8256,
+      "step": 6320
+    },
+    {
+      "epoch": 0.4889163512782884,
+      "grad_norm": 0.96484375,
+      "learning_rate": 2.7111390007133446e-05,
+      "loss": 6.843,
+      "step": 6330
+    },
+    {
+      "epoch": 0.48968873098014987,
+      "grad_norm": 0.9921875,
+      "learning_rate": 2.704906525159681e-05,
+      "loss": 6.8306,
+      "step": 6340
+    },
+    {
+      "epoch": 0.49046111068201126,
+      "grad_norm": 1.015625,
+      "learning_rate": 2.698672767233063e-05,
+      "loss": 6.8333,
+      "step": 6350
+    },
+    {
+      "epoch": 0.4912334903838727,
+      "grad_norm": 0.9765625,
+      "learning_rate": 2.692437765946415e-05,
+      "loss": 6.824,
+      "step": 6360
+    },
+    {
+      "epoch": 0.49200587008573415,
+      "grad_norm": 0.95703125,
+      "learning_rate": 2.6862015603204406e-05,
+      "loss": 6.8549,
+      "step": 6370
+    },
+    {
+      "epoch": 0.4927782497875956,
+      "grad_norm": 0.98046875,
+      "learning_rate": 2.679964189383382e-05,
+      "loss": 6.8584,
+      "step": 6380
+    },
+    {
+      "epoch": 0.493550629489457,
+      "grad_norm": 0.9609375,
+      "learning_rate": 2.6737256921707743e-05,
+      "loss": 6.8422,
+      "step": 6390
+    },
+    {
+      "epoch": 0.49432300919131844,
+      "grad_norm": 0.98046875,
+      "learning_rate": 2.6674861077252005e-05,
+      "loss": 6.8427,
+      "step": 6400
+    },
+    {
+      "epoch": 0.4950953888931799,
+      "grad_norm": 0.96484375,
+      "learning_rate": 2.6612454750960482e-05,
+      "loss": 6.852,
+      "step": 6410
+    },
+    {
+      "epoch": 0.49586776859504134,
+      "grad_norm": 0.9921875,
+      "learning_rate": 2.6550038333392645e-05,
+      "loss": 6.8469,
+      "step": 6420
+    },
+    {
+      "epoch": 0.4966401482969028,
+      "grad_norm": 0.9765625,
+      "learning_rate": 2.6487612215171142e-05,
+      "loss": 6.8168,
+      "step": 6430
+    },
+    {
+      "epoch": 0.4974125279987642,
+      "grad_norm": 0.9375,
+      "learning_rate": 2.64251767869793e-05,
+      "loss": 6.841,
+      "step": 6440
+    },
+    {
+      "epoch": 0.4981849077006256,
+      "grad_norm": 1.015625,
+      "learning_rate": 2.636273243955873e-05,
+      "loss": 6.8414,
+      "step": 6450
+    },
+    {
+      "epoch": 0.49895728740248707,
+      "grad_norm": 0.96875,
+      "learning_rate": 2.6300279563706842e-05,
+      "loss": 6.8561,
+      "step": 6460
+    },
+    {
+      "epoch": 0.4997296671043485,
+      "grad_norm": 1.0078125,
+      "learning_rate": 2.6237818550274447e-05,
+      "loss": 6.8294,
+      "step": 6470
+    },
+    {
+      "epoch": 0.5005020468062099,
+      "grad_norm": 0.9921875,
+      "learning_rate": 2.6175349790163274e-05,
+      "loss": 6.823,
+      "step": 6480
+    },
+    {
+      "epoch": 0.5012744265080714,
+      "grad_norm": 0.94921875,
+      "learning_rate": 2.6112873674323524e-05,
+      "loss": 6.847,
+      "step": 6490
+    },
+    {
+      "epoch": 0.5020468062099328,
+      "grad_norm": 0.99609375,
+      "learning_rate": 2.6050390593751427e-05,
+      "loss": 6.8129,
+      "step": 6500
+    },
+    {
+      "epoch": 0.5028191859117942,
+      "grad_norm": 0.98828125,
+      "learning_rate": 2.5987900939486843e-05,
+      "loss": 6.8278,
+      "step": 6510
+    },
+    {
+      "epoch": 0.5035915656136557,
+      "grad_norm": 0.96875,
+      "learning_rate": 2.5925405102610707e-05,
+      "loss": 6.8193,
+      "step": 6520
+    },
+    {
+      "epoch": 0.5043639453155171,
+      "grad_norm": 0.984375,
+      "learning_rate": 2.5862903474242706e-05,
+      "loss": 6.8456,
+      "step": 6530
+    },
+    {
+      "epoch": 0.5051363250173786,
+      "grad_norm": 0.96484375,
+      "learning_rate": 2.5800396445538734e-05,
+      "loss": 6.8333,
+      "step": 6540
+    },
+    {
+      "epoch": 0.50590870471924,
+      "grad_norm": 0.9609375,
+      "learning_rate": 2.573788440768851e-05,
+      "loss": 6.816,
+      "step": 6550
+    },
+    {
+      "epoch": 0.5066810844211014,
+      "grad_norm": 0.9765625,
+      "learning_rate": 2.567536775191307e-05,
+      "loss": 6.841,
+      "step": 6560
+    },
+    {
+      "epoch": 0.5074534641229629,
+      "grad_norm": 0.95703125,
+      "learning_rate": 2.5612846869462382e-05,
+      "loss": 6.8339,
+      "step": 6570
+    },
+    {
+      "epoch": 0.5082258438248243,
+      "grad_norm": 0.98046875,
+      "learning_rate": 2.5550322151612845e-05,
+      "loss": 6.8277,
+      "step": 6580
+    },
+    {
+      "epoch": 0.5089982235266857,
+      "grad_norm": 1.0,
+      "learning_rate": 2.5487793989664867e-05,
+      "loss": 6.8212,
+      "step": 6590
+    },
+    {
+      "epoch": 0.5097706032285472,
+      "grad_norm": 0.95703125,
+      "learning_rate": 2.5425262774940406e-05,
+      "loss": 6.8423,
+      "step": 6600
+    },
+    {
+      "epoch": 0.5105429829304086,
+      "grad_norm": 0.99609375,
+      "learning_rate": 2.5362728898780552e-05,
+      "loss": 6.8088,
+      "step": 6610
+    },
+    {
+      "epoch": 0.5113153626322701,
+      "grad_norm": 1.0234375,
+      "learning_rate": 2.530019275254301e-05,
+      "loss": 6.8632,
+      "step": 6620
+    },
+    {
+      "epoch": 0.5120877423341315,
+      "grad_norm": 0.953125,
+      "learning_rate": 2.5237654727599724e-05,
+      "loss": 6.8233,
+      "step": 6630
+    },
+    {
+      "epoch": 0.5128601220359928,
+      "grad_norm": 0.9765625,
+      "learning_rate": 2.5175115215334373e-05,
+      "loss": 6.82,
+      "step": 6640
+    },
+    {
+      "epoch": 0.5136325017378544,
+      "grad_norm": 0.9921875,
+      "learning_rate": 2.5112574607139962e-05,
+      "loss": 6.8315,
+      "step": 6650
+    },
+    {
+      "epoch": 0.5144048814397157,
+      "grad_norm": 1.03125,
+      "learning_rate": 2.5050033294416343e-05,
+      "loss": 6.8517,
+      "step": 6660
+    },
+    {
+      "epoch": 0.5151772611415772,
+      "grad_norm": 0.97265625,
+      "learning_rate": 2.4987491668567787e-05,
+      "loss": 6.863,
+      "step": 6670
+    },
+    {
+      "epoch": 0.5159496408434386,
+      "grad_norm": 0.95703125,
+      "learning_rate": 2.4924950121000512e-05,
+      "loss": 6.8277,
+      "step": 6680
+    },
+    {
+      "epoch": 0.5167220205453,
+      "grad_norm": 1.015625,
+      "learning_rate": 2.486240904312025e-05,
+      "loss": 6.8378,
+      "step": 6690
+    },
+    {
+      "epoch": 0.5174944002471615,
+      "grad_norm": 0.96875,
+      "learning_rate": 2.4799868826329814e-05,
+      "loss": 6.8373,
+      "step": 6700
+    },
+    {
+      "epoch": 0.5182667799490229,
+      "grad_norm": 0.97265625,
+      "learning_rate": 2.4737329862026596e-05,
+      "loss": 6.8284,
+      "step": 6710
+    },
+    {
+      "epoch": 0.5190391596508844,
+      "grad_norm": 1.03125,
+      "learning_rate": 2.467479254160017e-05,
+      "loss": 6.8338,
+      "step": 6720
+    },
+    {
+      "epoch": 0.5198115393527458,
+      "grad_norm": 0.98828125,
+      "learning_rate": 2.4612257256429806e-05,
+      "loss": 6.8446,
+      "step": 6730
+    },
+    {
+      "epoch": 0.5205839190546072,
+      "grad_norm": 1.015625,
+      "learning_rate": 2.4549724397882064e-05,
+      "loss": 6.7989,
+      "step": 6740
+    },
+    {
+      "epoch": 0.5213562987564687,
+      "grad_norm": 0.98828125,
+      "learning_rate": 2.4487194357308287e-05,
+      "loss": 6.8337,
+      "step": 6750
+    },
+    {
+      "epoch": 0.5221286784583301,
+      "grad_norm": 1.0,
+      "learning_rate": 2.4424667526042205e-05,
+      "loss": 6.8524,
+      "step": 6760
+    },
+    {
+      "epoch": 0.5229010581601915,
+      "grad_norm": 0.9765625,
+      "learning_rate": 2.436214429539745e-05,
+      "loss": 6.8342,
+      "step": 6770
+    },
+    {
+      "epoch": 0.523673437862053,
+      "grad_norm": 0.96875,
+      "learning_rate": 2.429962505666511e-05,
+      "loss": 6.8168,
+      "step": 6780
+    },
+    {
+      "epoch": 0.5244458175639144,
+      "grad_norm": 0.9765625,
+      "learning_rate": 2.4237110201111327e-05,
+      "loss": 6.8446,
+      "step": 6790
+    },
+    {
+      "epoch": 0.5252181972657759,
+      "grad_norm": 0.97265625,
+      "learning_rate": 2.417460011997478e-05,
+      "loss": 6.8268,
+      "step": 6800
+    },
+    {
+      "epoch": 0.5259905769676373,
+      "grad_norm": 0.98046875,
+      "learning_rate": 2.4112095204464276e-05,
+      "loss": 6.8415,
+      "step": 6810
+    },
+    {
+      "epoch": 0.5267629566694987,
+      "grad_norm": 0.96875,
+      "learning_rate": 2.4049595845756276e-05,
+      "loss": 6.8134,
+      "step": 6820
+    },
+    {
+      "epoch": 0.5275353363713602,
+      "grad_norm": 0.95703125,
+      "learning_rate": 2.398710243499252e-05,
+      "loss": 6.8049,
+      "step": 6830
+    },
+    {
+      "epoch": 0.5283077160732216,
+      "grad_norm": 0.97265625,
+      "learning_rate": 2.3924615363277465e-05,
+      "loss": 6.8186,
+      "step": 6840
+    },
+    {
+      "epoch": 0.5290800957750831,
+      "grad_norm": 0.95703125,
+      "learning_rate": 2.3862135021675918e-05,
+      "loss": 6.8366,
+      "step": 6850
+    },
+    {
+      "epoch": 0.5298524754769445,
+      "grad_norm": 0.9609375,
+      "learning_rate": 2.3799661801210564e-05,
+      "loss": 6.8343,
+      "step": 6860
+    },
+    {
+      "epoch": 0.5306248551788059,
+      "grad_norm": 0.99609375,
+      "learning_rate": 2.3737196092859537e-05,
+      "loss": 6.8227,
+      "step": 6870
+    },
+    {
+      "epoch": 0.5313972348806674,
+      "grad_norm": 0.984375,
+      "learning_rate": 2.367473828755394e-05,
+      "loss": 6.8358,
+      "step": 6880
+    },
+    {
+      "epoch": 0.5321696145825288,
+      "grad_norm": 0.95703125,
+      "learning_rate": 2.361228877617542e-05,
+      "loss": 6.8328,
+      "step": 6890
+    },
+    {
+      "epoch": 0.5329419942843902,
+      "grad_norm": 0.98828125,
+      "learning_rate": 2.3549847949553713e-05,
+      "loss": 6.8421,
+      "step": 6900
+    },
+    {
+      "epoch": 0.5337143739862517,
+      "grad_norm": 0.99609375,
+      "learning_rate": 2.3487416198464223e-05,
+      "loss": 6.8314,
+      "step": 6910
+    },
+    {
+      "epoch": 0.534486753688113,
+      "grad_norm": 0.96484375,
+      "learning_rate": 2.342499391362553e-05,
+      "loss": 6.8252,
+      "step": 6920
+    },
+    {
+      "epoch": 0.5352591333899746,
+      "grad_norm": 0.98828125,
+      "learning_rate": 2.336258148569699e-05,
+      "loss": 6.8508,
+      "step": 6930
+    },
+    {
+      "epoch": 0.5360315130918359,
+      "grad_norm": 1.0078125,
+      "learning_rate": 2.3300179305276258e-05,
+      "loss": 6.84,
+      "step": 6940
+    },
+    {
+      "epoch": 0.5368038927936973,
+      "grad_norm": 0.97265625,
+      "learning_rate": 2.3237787762896877e-05,
+      "loss": 6.8223,
+      "step": 6950
+    },
+    {
+      "epoch": 0.5375762724955588,
+      "grad_norm": 0.94140625,
+      "learning_rate": 2.3175407249025787e-05,
+      "loss": 6.8427,
+      "step": 6960
+    },
+    {
+      "epoch": 0.5383486521974202,
+      "grad_norm": 0.9609375,
+      "learning_rate": 2.3113038154060925e-05,
+      "loss": 6.8329,
+      "step": 6970
+    },
+    {
+      "epoch": 0.5391210318992817,
+      "grad_norm": 0.984375,
+      "learning_rate": 2.3050680868328756e-05,
+      "loss": 6.8199,
+      "step": 6980
+    },
+    {
+      "epoch": 0.5398934116011431,
+      "grad_norm": 0.9765625,
+      "learning_rate": 2.2988335782081855e-05,
+      "loss": 6.8295,
+      "step": 6990
+    },
+    {
+      "epoch": 0.5401251255117016,
+      "eval_loss": 7.0248188972473145,
+      "eval_runtime": 1.6753,
+      "eval_samples_per_second": 7.163,
+      "eval_steps_per_second": 1.194,
+      "step": 6993
+    },
+    {
+      "epoch": 0.5406657913030045,
+      "grad_norm": 1.0078125,
+      "learning_rate": 2.2926003285496428e-05,
+      "loss": 6.8556,
+      "step": 7000
+    },
+    {
+      "epoch": 0.5406657913030045,
+      "eval_loss": 7.024194717407227,
+      "eval_runtime": 0.6705,
+      "eval_samples_per_second": 17.897,
+      "eval_steps_per_second": 2.983,
+      "step": 7000
+    },
+    {
+      "epoch": 0.541438171004866,
+      "grad_norm": 0.96484375,
+      "learning_rate": 2.2863683768669904e-05,
+      "loss": 6.8122,
+      "step": 7010
+    },
+    {
+      "epoch": 0.5422105507067274,
+      "grad_norm": 0.99609375,
+      "learning_rate": 2.2801377621618464e-05,
+      "loss": 6.8228,
+      "step": 7020
+    },
+    {
+      "epoch": 0.5429829304085889,
+      "grad_norm": 0.9375,
+      "learning_rate": 2.273908523427465e-05,
+      "loss": 6.8193,
+      "step": 7030
+    },
+    {
+      "epoch": 0.5437553101104503,
+      "grad_norm": 0.95703125,
+      "learning_rate": 2.2676806996484856e-05,
+      "loss": 6.8411,
+      "step": 7040
+    },
+    {
+      "epoch": 0.5445276898123117,
+      "grad_norm": 1.0078125,
+      "learning_rate": 2.2614543298006946e-05,
+      "loss": 6.8139,
+      "step": 7050
+    },
+    {
+      "epoch": 0.5453000695141732,
+      "grad_norm": 0.984375,
+      "learning_rate": 2.2552294528507757e-05,
+      "loss": 6.8291,
+      "step": 7060
+    },
+    {
+      "epoch": 0.5460724492160346,
+      "grad_norm": 0.96875,
+      "learning_rate": 2.249006107756076e-05,
+      "loss": 6.8288,
+      "step": 7070
+    },
+    {
+      "epoch": 0.546844828917896,
+      "grad_norm": 1.0078125,
+      "learning_rate": 2.2427843334643503e-05,
+      "loss": 6.8558,
+      "step": 7080
+    },
+    {
+      "epoch": 0.5476172086197575,
+      "grad_norm": 0.98046875,
+      "learning_rate": 2.236564168913524e-05,
+      "loss": 6.8224,
+      "step": 7090
+    },
+    {
+      "epoch": 0.5483895883216189,
+      "grad_norm": 1.0,
+      "learning_rate": 2.2303456530314488e-05,
+      "loss": 6.8361,
+      "step": 7100
+    },
+    {
+      "epoch": 0.5491619680234804,
+      "grad_norm": 0.96875,
+      "learning_rate": 2.2241288247356572e-05,
+      "loss": 6.8415,
+      "step": 7110
+    },
+    {
+      "epoch": 0.5499343477253418,
+      "grad_norm": 0.98046875,
+      "learning_rate": 2.2179137229331243e-05,
+      "loss": 6.8309,
+      "step": 7120
+    },
+    {
+      "epoch": 0.5507067274272032,
+      "grad_norm": 1.0,
+      "learning_rate": 2.2117003865200154e-05,
+      "loss": 6.822,
+      "step": 7130
+    },
+    {
+      "epoch": 0.5514791071290647,
+      "grad_norm": 1.0,
+      "learning_rate": 2.2054888543814493e-05,
+      "loss": 6.8302,
+      "step": 7140
+    },
+    {
+      "epoch": 0.5522514868309261,
+      "grad_norm": 0.95703125,
+      "learning_rate": 2.1992791653912525e-05,
+      "loss": 6.859,
+      "step": 7150
+    },
+    {
+      "epoch": 0.5530238665327876,
+      "grad_norm": 0.984375,
+      "learning_rate": 2.1930713584117182e-05,
+      "loss": 6.8409,
+      "step": 7160
+    },
+    {
+      "epoch": 0.553796246234649,
+      "grad_norm": 0.98046875,
+      "learning_rate": 2.18686547229336e-05,
+      "loss": 6.8316,
+      "step": 7170
+    },
+    {
+      "epoch": 0.5545686259365104,
+      "grad_norm": 0.96875,
+      "learning_rate": 2.1806615458746698e-05,
+      "loss": 6.8137,
+      "step": 7180
+    },
+    {
+      "epoch": 0.5553410056383719,
+      "grad_norm": 0.99609375,
+      "learning_rate": 2.1744596179818753e-05,
+      "loss": 6.8187,
+      "step": 7190
+    },
+    {
+      "epoch": 0.5561133853402332,
+      "grad_norm": 0.9765625,
+      "learning_rate": 2.1682597274286978e-05,
+      "loss": 6.8303,
+      "step": 7200
+    },
+    {
+      "epoch": 0.5568857650420947,
+      "grad_norm": 0.94921875,
+      "learning_rate": 2.162061913016107e-05,
+      "loss": 6.831,
+      "step": 7210
+    },
+    {
+      "epoch": 0.5576581447439561,
+      "grad_norm": 0.94921875,
+      "learning_rate": 2.15586621353208e-05,
+      "loss": 6.8433,
+      "step": 7220
+    },
+    {
+      "epoch": 0.5584305244458175,
+      "grad_norm": 0.9609375,
+      "learning_rate": 2.1496726677513572e-05,
+      "loss": 6.8362,
+      "step": 7230
+    },
+    {
+      "epoch": 0.559202904147679,
+      "grad_norm": 0.97265625,
+      "learning_rate": 2.1434813144352018e-05,
+      "loss": 6.8411,
+      "step": 7240
+    },
+    {
+      "epoch": 0.5599752838495404,
+      "grad_norm": 0.9453125,
+      "learning_rate": 2.137292192331155e-05,
+      "loss": 6.8255,
+      "step": 7250
+    },
+    {
+      "epoch": 0.5607476635514018,
+      "grad_norm": 0.9765625,
+      "learning_rate": 2.1311053401727945e-05,
+      "loss": 6.844,
+      "step": 7260
+    },
+    {
+      "epoch": 0.5615200432532633,
+      "grad_norm": 0.96484375,
+      "learning_rate": 2.124920796679491e-05,
+      "loss": 6.847,
+      "step": 7270
+    },
+    {
+      "epoch": 0.5622924229551247,
+      "grad_norm": 0.97265625,
+      "learning_rate": 2.1187386005561688e-05,
+      "loss": 6.7946,
+      "step": 7280
+    },
+    {
+      "epoch": 0.5630648026569862,
+      "grad_norm": 0.96875,
+      "learning_rate": 2.1125587904930598e-05,
+      "loss": 6.8269,
+      "step": 7290
+    },
+    {
+      "epoch": 0.5638371823588476,
+      "grad_norm": 0.99609375,
+      "learning_rate": 2.1063814051654646e-05,
+      "loss": 6.8256,
+      "step": 7300
+    },
+    {
+      "epoch": 0.564609562060709,
+      "grad_norm": 1.015625,
+      "learning_rate": 2.1002064832335057e-05,
+      "loss": 6.8257,
+      "step": 7310
+    },
+    {
+      "epoch": 0.5653819417625705,
+      "grad_norm": 0.984375,
+      "learning_rate": 2.0940340633418945e-05,
+      "loss": 6.8182,
+      "step": 7320
+    },
+    {
+      "epoch": 0.5661543214644319,
+      "grad_norm": 0.94921875,
+      "learning_rate": 2.0878641841196788e-05,
+      "loss": 6.8362,
+      "step": 7330
+    },
+    {
+      "epoch": 0.5669267011662934,
+      "grad_norm": 0.95703125,
+      "learning_rate": 2.0816968841800093e-05,
+      "loss": 6.8235,
+      "step": 7340
+    },
+    {
+      "epoch": 0.5676990808681548,
+      "grad_norm": 0.96875,
+      "learning_rate": 2.07553220211989e-05,
+      "loss": 6.8445,
+      "step": 7350
+    },
+    {
+      "epoch": 0.5684714605700162,
+      "grad_norm": 0.9765625,
+      "learning_rate": 2.069370176519949e-05,
+      "loss": 6.821,
+      "step": 7360
+    },
+    {
+      "epoch": 0.5692438402718777,
+      "grad_norm": 0.96875,
+      "learning_rate": 2.0632108459441828e-05,
+      "loss": 6.8246,
+      "step": 7370
+    },
+    {
+      "epoch": 0.5700162199737391,
+      "grad_norm": 1.046875,
+      "learning_rate": 2.0570542489397233e-05,
+      "loss": 6.8322,
+      "step": 7380
+    },
+    {
+      "epoch": 0.5707885996756005,
+      "grad_norm": 0.98828125,
+      "learning_rate": 2.0509004240365952e-05,
+      "loss": 6.8288,
+      "step": 7390
+    },
+    {
+      "epoch": 0.571560979377462,
+      "grad_norm": 0.97265625,
+      "learning_rate": 2.0447494097474763e-05,
+      "loss": 6.8228,
+      "step": 7400
+    },
+    {
+      "epoch": 0.5723333590793234,
+      "grad_norm": 1.03125,
+      "learning_rate": 2.0386012445674517e-05,
+      "loss": 6.8267,
+      "step": 7410
+    },
+    {
+      "epoch": 0.5731057387811849,
+      "grad_norm": 0.98046875,
+      "learning_rate": 2.032455966973777e-05,
+      "loss": 6.8391,
+      "step": 7420
+    },
+    {
+      "epoch": 0.5738781184830463,
+      "grad_norm": 0.94140625,
+      "learning_rate": 2.0263136154256356e-05,
+      "loss": 6.8319,
+      "step": 7430
+    },
+    {
+      "epoch": 0.5746504981849077,
+      "grad_norm": 1.0078125,
+      "learning_rate": 2.020174228363901e-05,
+      "loss": 6.8387,
+      "step": 7440
+    },
+    {
+      "epoch": 0.5754228778867692,
+      "grad_norm": 0.98828125,
+      "learning_rate": 2.014037844210892e-05,
+      "loss": 6.8304,
+      "step": 7450
+    },
+    {
+      "epoch": 0.5761952575886305,
+      "grad_norm": 0.97265625,
+      "learning_rate": 2.0079045013701343e-05,
+      "loss": 6.8125,
+      "step": 7460
+    },
+    {
+      "epoch": 0.576967637290492,
+      "grad_norm": 1.0078125,
+      "learning_rate": 2.00177423822612e-05,
+      "loss": 6.8464,
+      "step": 7470
+    },
+    {
+      "epoch": 0.5777400169923534,
+      "grad_norm": 0.9375,
+      "learning_rate": 1.995647093144068e-05,
+      "loss": 6.857,
+      "step": 7480
+    },
+    {
+      "epoch": 0.5785123966942148,
+      "grad_norm": 0.9765625,
+      "learning_rate": 1.9895231044696845e-05,
+      "loss": 6.8169,
+      "step": 7490
+    },
+    {
+      "epoch": 0.5792847763960763,
+      "grad_norm": 0.97265625,
+      "learning_rate": 1.9834023105289187e-05,
+      "loss": 6.8129,
+      "step": 7500
+    },
+    {
+      "epoch": 0.5800571560979377,
+      "grad_norm": 0.99609375,
+      "learning_rate": 1.977284749627729e-05,
+      "loss": 6.8157,
+      "step": 7510
+    },
+    {
+      "epoch": 0.5808295357997992,
+      "grad_norm": 0.99609375,
+      "learning_rate": 1.9711704600518388e-05,
+      "loss": 6.8557,
+      "step": 7520
+    },
+    {
+      "epoch": 0.5816019155016606,
+      "grad_norm": 0.98828125,
+      "learning_rate": 1.9650594800664996e-05,
+      "loss": 6.8389,
+      "step": 7530
+    },
+    {
+      "epoch": 0.582374295203522,
+      "grad_norm": 0.953125,
+      "learning_rate": 1.958951847916249e-05,
+      "loss": 6.8395,
+      "step": 7540
+    },
+    {
+      "epoch": 0.5831466749053835,
+      "grad_norm": 0.94921875,
+      "learning_rate": 1.9528476018246742e-05,
+      "loss": 6.8159,
+      "step": 7550
+    },
+    {
+      "epoch": 0.5839190546072449,
+      "grad_norm": 0.99609375,
+      "learning_rate": 1.9467467799941692e-05,
+      "loss": 6.814,
+      "step": 7560
+    },
+    {
+      "epoch": 0.5846914343091063,
+      "grad_norm": 0.98046875,
+      "learning_rate": 1.9406494206057012e-05,
+      "loss": 6.8321,
+      "step": 7570
+    },
+    {
+      "epoch": 0.5854638140109678,
+      "grad_norm": 0.984375,
+      "learning_rate": 1.9345555618185648e-05,
+      "loss": 6.8442,
+      "step": 7580
+    },
+    {
+      "epoch": 0.5862361937128292,
+      "grad_norm": 0.9609375,
+      "learning_rate": 1.9284652417701498e-05,
+      "loss": 6.8397,
+      "step": 7590
+    },
+    {
+      "epoch": 0.5870085734146907,
+      "grad_norm": 0.9765625,
+      "learning_rate": 1.9223784985756947e-05,
+      "loss": 6.8461,
+      "step": 7600
+    },
+    {
+      "epoch": 0.5877809531165521,
+      "grad_norm": 0.9609375,
+      "learning_rate": 1.9162953703280595e-05,
+      "loss": 6.8199,
+      "step": 7610
+    },
+    {
+      "epoch": 0.5885533328184135,
+      "grad_norm": 0.9609375,
+      "learning_rate": 1.910215895097476e-05,
+      "loss": 6.8245,
+      "step": 7620
+    },
+    {
+      "epoch": 0.589325712520275,
+      "grad_norm": 0.9375,
+      "learning_rate": 1.9041401109313142e-05,
+      "loss": 6.8307,
+      "step": 7630
+    },
+    {
+      "epoch": 0.5900980922221364,
+      "grad_norm": 0.95703125,
+      "learning_rate": 1.8980680558538457e-05,
+      "loss": 6.8155,
+      "step": 7640
+    },
+    {
+      "epoch": 0.5908704719239979,
+      "grad_norm": 0.98046875,
+      "learning_rate": 1.891999767866005e-05,
+      "loss": 6.8385,
+      "step": 7650
+    },
+    {
+      "epoch": 0.5916428516258593,
+      "grad_norm": 0.99609375,
+      "learning_rate": 1.885935284945149e-05,
+      "loss": 6.8417,
+      "step": 7660
+    },
+    {
+      "epoch": 0.5924152313277207,
+      "grad_norm": 0.9453125,
+      "learning_rate": 1.879874645044822e-05,
+      "loss": 6.8094,
+      "step": 7670
+    },
+    {
+      "epoch": 0.5931876110295822,
+      "grad_norm": 0.98828125,
+      "learning_rate": 1.8738178860945165e-05,
+      "loss": 6.8371,
+      "step": 7680
+    },
+    {
+      "epoch": 0.5939599907314436,
+      "grad_norm": 0.94921875,
+      "learning_rate": 1.86776504599944e-05,
+      "loss": 6.8353,
+      "step": 7690
+    },
+    {
+      "epoch": 0.594732370433305,
+      "grad_norm": 0.95703125,
+      "learning_rate": 1.8617161626402703e-05,
+      "loss": 6.8206,
+      "step": 7700
+    },
+    {
+      "epoch": 0.5955047501351665,
+      "grad_norm": 1.0078125,
+      "learning_rate": 1.8556712738729245e-05,
+      "loss": 6.8238,
+      "step": 7710
+    },
+    {
+      "epoch": 0.5962771298370279,
+      "grad_norm": 0.99609375,
+      "learning_rate": 1.84963041752832e-05,
+      "loss": 6.8114,
+      "step": 7720
+    },
+    {
+      "epoch": 0.5970495095388894,
+      "grad_norm": 0.96875,
+      "learning_rate": 1.84359363141214e-05,
+      "loss": 6.8419,
+      "step": 7730
+    },
+    {
+      "epoch": 0.5978218892407507,
+      "grad_norm": 1.0078125,
+      "learning_rate": 1.837560953304591e-05,
+      "loss": 6.847,
+      "step": 7740
+    },
+    {
+      "epoch": 0.5985942689426121,
+      "grad_norm": 0.98046875,
+      "learning_rate": 1.831532420960173e-05,
+      "loss": 6.8259,
+      "step": 7750
+    },
+    {
+      "epoch": 0.5993666486444736,
+      "grad_norm": 0.9765625,
+      "learning_rate": 1.825508072107439e-05,
+      "loss": 6.8315,
+      "step": 7760
+    },
+    {
+      "epoch": 0.600139028346335,
+      "grad_norm": 0.97265625,
+      "learning_rate": 1.8194879444487627e-05,
+      "loss": 6.8244,
+      "step": 7770
+    },
+    {
+      "epoch": 0.6009114080481965,
+      "grad_norm": 0.95703125,
+      "learning_rate": 1.813472075660097e-05,
+      "loss": 6.814,
+      "step": 7780
+    },
+    {
+      "epoch": 0.6016837877500579,
+      "grad_norm": 1.015625,
+      "learning_rate": 1.8074605033907432e-05,
+      "loss": 6.8255,
+      "step": 7790
+    },
+    {
+      "epoch": 0.6024561674519193,
+      "grad_norm": 0.96484375,
+      "learning_rate": 1.8014532652631133e-05,
+      "loss": 6.8337,
+      "step": 7800
+    },
+    {
+      "epoch": 0.6032285471537808,
+      "grad_norm": 0.984375,
+      "learning_rate": 1.795450398872494e-05,
+      "loss": 6.8234,
+      "step": 7810
+    },
+    {
+      "epoch": 0.6040009268556422,
+      "grad_norm": 0.98046875,
+      "learning_rate": 1.7894519417868138e-05,
+      "loss": 6.8266,
+      "step": 7820
+    },
+    {
+      "epoch": 0.6047733065575037,
+      "grad_norm": 0.9921875,
+      "learning_rate": 1.7834579315464055e-05,
+      "loss": 6.8131,
+      "step": 7830
+    },
+    {
+      "epoch": 0.6055456862593651,
+      "grad_norm": 0.96484375,
+      "learning_rate": 1.7774684056637726e-05,
+      "loss": 6.7991,
+      "step": 7840
+    },
+    {
+      "epoch": 0.6063180659612265,
+      "grad_norm": 0.96484375,
+      "learning_rate": 1.77148340162335e-05,
+      "loss": 6.8137,
+      "step": 7850
+    },
+    {
+      "epoch": 0.607090445663088,
+      "grad_norm": 1.09375,
+      "learning_rate": 1.7655029568812808e-05,
+      "loss": 6.8336,
+      "step": 7860
+    },
+    {
+      "epoch": 0.6078628253649494,
+      "grad_norm": 1.015625,
+      "learning_rate": 1.7595271088651696e-05,
+      "loss": 6.8414,
+      "step": 7870
+    },
+    {
+      "epoch": 0.6086352050668108,
+      "grad_norm": 0.9765625,
+      "learning_rate": 1.753555894973853e-05,
+      "loss": 6.8144,
+      "step": 7880
+    },
+    {
+      "epoch": 0.6094075847686723,
+      "grad_norm": 0.98046875,
+      "learning_rate": 1.747589352577166e-05,
+      "loss": 6.8123,
+      "step": 7890
+    },
+    {
+      "epoch": 0.6101799644705337,
+      "grad_norm": 0.96875,
+      "learning_rate": 1.741627519015711e-05,
+      "loss": 6.8095,
+      "step": 7900
+    },
+    {
+      "epoch": 0.6109523441723952,
+      "grad_norm": 0.9921875,
+      "learning_rate": 1.7356704316006184e-05,
+      "loss": 6.8234,
+      "step": 7910
+    },
+    {
+      "epoch": 0.6117247238742566,
+      "grad_norm": 1.0078125,
+      "learning_rate": 1.729718127613314e-05,
+      "loss": 6.8459,
+      "step": 7920
+    },
+    {
+      "epoch": 0.612497103576118,
+      "grad_norm": 1.0390625,
+      "learning_rate": 1.723770644305289e-05,
+      "loss": 6.83,
+      "step": 7930
+    },
+    {
+      "epoch": 0.6132694832779795,
+      "grad_norm": 0.99609375,
+      "learning_rate": 1.717828018897868e-05,
+      "loss": 6.825,
+      "step": 7940
+    },
+    {
+      "epoch": 0.6140418629798409,
+      "grad_norm": 0.9609375,
+      "learning_rate": 1.711890288581969e-05,
+      "loss": 6.8261,
+      "step": 7950
+    },
+    {
+      "epoch": 0.6148142426817024,
+      "grad_norm": 0.953125,
+      "learning_rate": 1.7059574905178758e-05,
+      "loss": 6.839,
+      "step": 7960
+    },
+    {
+      "epoch": 0.6155866223835638,
+      "grad_norm": 0.9765625,
+      "learning_rate": 1.7000296618350053e-05,
+      "loss": 6.8461,
+      "step": 7970
+    },
+    {
+      "epoch": 0.6163590020854252,
+      "grad_norm": 0.9921875,
+      "learning_rate": 1.6941068396316757e-05,
+      "loss": 6.8297,
+      "step": 7980
+    },
+    {
+      "epoch": 0.6171313817872867,
+      "grad_norm": 0.96484375,
+      "learning_rate": 1.6881890609748706e-05,
+      "loss": 6.8128,
+      "step": 7990
+    },
+    {
+      "epoch": 0.617903761489148,
+      "grad_norm": 1.0234375,
+      "learning_rate": 1.6822763629000104e-05,
+      "loss": 6.8333,
+      "step": 8000
+    },
+    {
+      "epoch": 0.617903761489148,
+      "eval_loss": 7.016650676727295,
+      "eval_runtime": 0.801,
+      "eval_samples_per_second": 14.981,
+      "eval_steps_per_second": 2.497,
+      "step": 8000
+    },
+    {
+      "epoch": 0.6186761411910096,
+      "grad_norm": 0.99609375,
+      "learning_rate": 1.6763687824107195e-05,
+      "loss": 6.819,
+      "step": 8010
+    },
+    {
+      "epoch": 0.619448520892871,
+      "grad_norm": 0.98046875,
+      "learning_rate": 1.6704663564785948e-05,
+      "loss": 6.8198,
+      "step": 8020
+    },
+    {
+      "epoch": 0.6202209005947323,
+      "grad_norm": 1.0,
+      "learning_rate": 1.6645691220429748e-05,
+      "loss": 6.8266,
+      "step": 8030
+    },
+    {
+      "epoch": 0.6209932802965938,
+      "grad_norm": 1.0390625,
+      "learning_rate": 1.6586771160107068e-05,
+      "loss": 6.8368,
+      "step": 8040
+    },
+    {
+      "epoch": 0.6217656599984552,
+      "grad_norm": 0.9765625,
+      "learning_rate": 1.652790375255917e-05,
+      "loss": 6.8504,
+      "step": 8050
+    },
+    {
+      "epoch": 0.6225380397003166,
+      "grad_norm": 0.96875,
+      "learning_rate": 1.6469089366197816e-05,
+      "loss": 6.827,
+      "step": 8060
+    },
+    {
+      "epoch": 0.6233104194021781,
+      "grad_norm": 1.0078125,
+      "learning_rate": 1.6410328369102922e-05,
+      "loss": 6.8419,
+      "step": 8070
+    },
+    {
+      "epoch": 0.6240827991040395,
+      "grad_norm": 0.94921875,
+      "learning_rate": 1.635162112902029e-05,
+      "loss": 6.8355,
+      "step": 8080
+    },
+    {
+      "epoch": 0.624855178805901,
+      "grad_norm": 0.99609375,
+      "learning_rate": 1.629296801335928e-05,
+      "loss": 6.8151,
+      "step": 8090
+    },
+    {
+      "epoch": 0.6256275585077624,
+      "grad_norm": 1.015625,
+      "learning_rate": 1.6234369389190538e-05,
+      "loss": 6.8074,
+      "step": 8100
+    },
+    {
+      "epoch": 0.6263999382096238,
+      "grad_norm": 1.0,
+      "learning_rate": 1.617582562324368e-05,
+      "loss": 6.8179,
+      "step": 8110
+    },
+    {
+      "epoch": 0.6271723179114853,
+      "grad_norm": 0.9609375,
+      "learning_rate": 1.6117337081904993e-05,
+      "loss": 6.8419,
+      "step": 8120
+    },
+    {
+      "epoch": 0.6279446976133467,
+      "grad_norm": 0.921875,
+      "learning_rate": 1.6058904131215152e-05,
+      "loss": 6.7907,
+      "step": 8130
+    },
+    {
+      "epoch": 0.6287170773152082,
+      "grad_norm": 1.0,
+      "learning_rate": 1.6000527136866942e-05,
+      "loss": 6.7958,
+      "step": 8140
+    },
+    {
+      "epoch": 0.6294894570170696,
+      "grad_norm": 0.984375,
+      "learning_rate": 1.5942206464202937e-05,
+      "loss": 6.831,
+      "step": 8150
+    },
+    {
+      "epoch": 0.630261836718931,
+      "grad_norm": 1.0078125,
+      "learning_rate": 1.5883942478213247e-05,
+      "loss": 6.8351,
+      "step": 8160
+    },
+    {
+      "epoch": 0.6310342164207925,
+      "grad_norm": 0.9375,
+      "learning_rate": 1.5825735543533194e-05,
+      "loss": 6.805,
+      "step": 8170
+    },
+    {
+      "epoch": 0.6318065961226539,
+      "grad_norm": 1.0,
+      "learning_rate": 1.5767586024441068e-05,
+      "loss": 6.8327,
+      "step": 8180
+    },
+    {
+      "epoch": 0.6325789758245153,
+      "grad_norm": 1.0234375,
+      "learning_rate": 1.5709494284855855e-05,
+      "loss": 6.8061,
+      "step": 8190
+    },
+    {
+      "epoch": 0.6333513555263768,
+      "grad_norm": 0.953125,
+      "learning_rate": 1.5651460688334902e-05,
+      "loss": 6.8425,
+      "step": 8200
+    },
+    {
+      "epoch": 0.6341237352282382,
+      "grad_norm": 0.953125,
+      "learning_rate": 1.559348559807169e-05,
+      "loss": 6.8266,
+      "step": 8210
+    },
+    {
+      "epoch": 0.6348961149300997,
+      "grad_norm": 0.95703125,
+      "learning_rate": 1.5535569376893554e-05,
+      "loss": 6.8397,
+      "step": 8220
+    },
+    {
+      "epoch": 0.6356684946319611,
+      "grad_norm": 0.98828125,
+      "learning_rate": 1.5477712387259403e-05,
+      "loss": 6.8182,
+      "step": 8230
+    },
+    {
+      "epoch": 0.6364408743338225,
+      "grad_norm": 0.93359375,
+      "learning_rate": 1.5419914991257464e-05,
+      "loss": 6.8343,
+      "step": 8240
+    },
+    {
+      "epoch": 0.637213254035684,
+      "grad_norm": 0.93359375,
+      "learning_rate": 1.536217755060299e-05,
+      "loss": 6.8285,
+      "step": 8250
+    },
+    {
+      "epoch": 0.6379856337375454,
+      "grad_norm": 0.9375,
+      "learning_rate": 1.530450042663602e-05,
+      "loss": 6.8255,
+      "step": 8260
+    },
+    {
+      "epoch": 0.6387580134394069,
+      "grad_norm": 0.984375,
+      "learning_rate": 1.5246883980319121e-05,
+      "loss": 6.8244,
+      "step": 8270
+    },
+    {
+      "epoch": 0.6395303931412682,
+      "grad_norm": 0.98828125,
+      "learning_rate": 1.5189328572235115e-05,
+      "loss": 6.8182,
+      "step": 8280
+    },
+    {
+      "epoch": 0.6403027728431296,
+      "grad_norm": 1.0,
+      "learning_rate": 1.5131834562584816e-05,
+      "loss": 6.8362,
+      "step": 8290
+    },
+    {
+      "epoch": 0.6410751525449911,
+      "grad_norm": 0.9609375,
+      "learning_rate": 1.5074402311184788e-05,
+      "loss": 6.8347,
+      "step": 8300
+    },
+    {
+      "epoch": 0.6418475322468525,
+      "grad_norm": 0.97265625,
+      "learning_rate": 1.5017032177465115e-05,
+      "loss": 6.8046,
+      "step": 8310
+    },
+    {
+      "epoch": 0.642619911948714,
+      "grad_norm": 0.9921875,
+      "learning_rate": 1.49597245204671e-05,
+      "loss": 6.8417,
+      "step": 8320
+    },
+    {
+      "epoch": 0.6433922916505754,
+      "grad_norm": 0.94921875,
+      "learning_rate": 1.4902479698841054e-05,
+      "loss": 6.8511,
+      "step": 8330
+    },
+    {
+      "epoch": 0.6441646713524368,
+      "grad_norm": 0.9765625,
+      "learning_rate": 1.4845298070844043e-05,
+      "loss": 6.8339,
+      "step": 8340
+    },
+    {
+      "epoch": 0.6449370510542983,
+      "grad_norm": 0.97265625,
+      "learning_rate": 1.4788179994337658e-05,
+      "loss": 6.8313,
+      "step": 8350
+    },
+    {
+      "epoch": 0.6457094307561597,
+      "grad_norm": 0.953125,
+      "learning_rate": 1.4731125826785752e-05,
+      "loss": 6.832,
+      "step": 8360
+    },
+    {
+      "epoch": 0.6464818104580211,
+      "grad_norm": 0.953125,
+      "learning_rate": 1.4674135925252216e-05,
+      "loss": 6.8134,
+      "step": 8370
+    },
+    {
+      "epoch": 0.6472541901598826,
+      "grad_norm": 0.94921875,
+      "learning_rate": 1.4617210646398739e-05,
+      "loss": 6.8213,
+      "step": 8380
+    },
+    {
+      "epoch": 0.648026569861744,
+      "grad_norm": 0.98046875,
+      "learning_rate": 1.4560350346482598e-05,
+      "loss": 6.8196,
+      "step": 8390
+    },
+    {
+      "epoch": 0.6487989495636055,
+      "grad_norm": 0.98046875,
+      "learning_rate": 1.4503555381354378e-05,
+      "loss": 6.8283,
+      "step": 8400
+    },
+    {
+      "epoch": 0.6495713292654669,
+      "grad_norm": 0.99609375,
+      "learning_rate": 1.4446826106455819e-05,
+      "loss": 6.8338,
+      "step": 8410
+    },
+    {
+      "epoch": 0.6503437089673283,
+      "grad_norm": 0.9609375,
+      "learning_rate": 1.4390162876817503e-05,
+      "loss": 6.8591,
+      "step": 8420
+    },
+    {
+      "epoch": 0.6511160886691898,
+      "grad_norm": 1.015625,
+      "learning_rate": 1.4333566047056719e-05,
+      "loss": 6.8454,
+      "step": 8430
+    },
+    {
+      "epoch": 0.6518884683710512,
+      "grad_norm": 0.96875,
+      "learning_rate": 1.4277035971375192e-05,
+      "loss": 6.8344,
+      "step": 8440
+    },
+    {
+      "epoch": 0.6526608480729127,
+      "grad_norm": 0.96875,
+      "learning_rate": 1.4220573003556852e-05,
+      "loss": 6.8274,
+      "step": 8450
+    },
+    {
+      "epoch": 0.6534332277747741,
+      "grad_norm": 0.96484375,
+      "learning_rate": 1.4164177496965687e-05,
+      "loss": 6.8408,
+      "step": 8460
+    },
+    {
+      "epoch": 0.6542056074766355,
+      "grad_norm": 0.93359375,
+      "learning_rate": 1.4107849804543469e-05,
+      "loss": 6.8185,
+      "step": 8470
+    },
+    {
+      "epoch": 0.654977987178497,
+      "grad_norm": 0.9921875,
+      "learning_rate": 1.4051590278807556e-05,
+      "loss": 6.8323,
+      "step": 8480
+    },
+    {
+      "epoch": 0.6557503668803584,
+      "grad_norm": 0.96484375,
+      "learning_rate": 1.3995399271848725e-05,
+      "loss": 6.8133,
+      "step": 8490
+    },
+    {
+      "epoch": 0.6565227465822199,
+      "grad_norm": 0.95703125,
+      "learning_rate": 1.3939277135328901e-05,
+      "loss": 6.8146,
+      "step": 8500
+    },
+    {
+      "epoch": 0.6572951262840813,
+      "grad_norm": 0.95703125,
+      "learning_rate": 1.3883224220479029e-05,
+      "loss": 6.8294,
+      "step": 8510
+    },
+    {
+      "epoch": 0.6580675059859427,
+      "grad_norm": 0.9609375,
+      "learning_rate": 1.3827240878096845e-05,
+      "loss": 6.8135,
+      "step": 8520
+    },
+    {
+      "epoch": 0.6588398856878042,
+      "grad_norm": 0.9609375,
+      "learning_rate": 1.3771327458544635e-05,
+      "loss": 6.8181,
+      "step": 8530
+    },
+    {
+      "epoch": 0.6596122653896656,
+      "grad_norm": 0.9921875,
+      "learning_rate": 1.3715484311747134e-05,
+      "loss": 6.8189,
+      "step": 8540
+    },
+    {
+      "epoch": 0.660384645091527,
+      "grad_norm": 0.94921875,
+      "learning_rate": 1.3659711787189244e-05,
+      "loss": 6.8238,
+      "step": 8550
+    },
+    {
+      "epoch": 0.6611570247933884,
+      "grad_norm": 0.9609375,
+      "learning_rate": 1.3604010233913925e-05,
+      "loss": 6.8193,
+      "step": 8560
+    },
+    {
+      "epoch": 0.6619294044952498,
+      "grad_norm": 0.94140625,
+      "learning_rate": 1.3548380000519964e-05,
+      "loss": 6.8231,
+      "step": 8570
+    },
+    {
+      "epoch": 0.6627017841971113,
+      "grad_norm": 0.94921875,
+      "learning_rate": 1.3492821435159797e-05,
+      "loss": 6.8094,
+      "step": 8580
+    },
+    {
+      "epoch": 0.6634741638989727,
+      "grad_norm": 0.96484375,
+      "learning_rate": 1.3437334885537317e-05,
+      "loss": 6.8286,
+      "step": 8590
+    },
+    {
+      "epoch": 0.6642465436008341,
+      "grad_norm": 0.98046875,
+      "learning_rate": 1.3381920698905787e-05,
+      "loss": 6.8046,
+      "step": 8600
+    },
+    {
+      "epoch": 0.6650189233026956,
+      "grad_norm": 0.9921875,
+      "learning_rate": 1.3326579222065535e-05,
+      "loss": 6.8363,
+      "step": 8610
+    },
+    {
+      "epoch": 0.665791303004557,
+      "grad_norm": 0.95703125,
+      "learning_rate": 1.3271310801361852e-05,
+      "loss": 6.8251,
+      "step": 8620
+    },
+    {
+      "epoch": 0.6665636827064185,
+      "grad_norm": 0.94921875,
+      "learning_rate": 1.3216115782682853e-05,
+      "loss": 6.8133,
+      "step": 8630
+    },
+    {
+      "epoch": 0.6673360624082799,
+      "grad_norm": 0.97265625,
+      "learning_rate": 1.3160994511457264e-05,
+      "loss": 6.8271,
+      "step": 8640
+    },
+    {
+      "epoch": 0.6681084421101413,
+      "grad_norm": 0.96484375,
+      "learning_rate": 1.3105947332652252e-05,
+      "loss": 6.8514,
+      "step": 8650
+    },
+    {
+      "epoch": 0.6688808218120028,
+      "grad_norm": 0.984375,
+      "learning_rate": 1.3050974590771328e-05,
+      "loss": 6.8099,
+      "step": 8660
+    },
+    {
+      "epoch": 0.6696532015138642,
+      "grad_norm": 0.96484375,
+      "learning_rate": 1.2996076629852114e-05,
+      "loss": 6.8558,
+      "step": 8670
+    },
+    {
+      "epoch": 0.6704255812157256,
+      "grad_norm": 0.96875,
+      "learning_rate": 1.2941253793464255e-05,
+      "loss": 6.8245,
+      "step": 8680
+    },
+    {
+      "epoch": 0.6711979609175871,
+      "grad_norm": 0.98828125,
+      "learning_rate": 1.2886506424707245e-05,
+      "loss": 6.8361,
+      "step": 8690
+    },
+    {
+      "epoch": 0.6719703406194485,
+      "grad_norm": 1.0,
+      "learning_rate": 1.2831834866208243e-05,
+      "loss": 6.832,
+      "step": 8700
+    },
+    {
+      "epoch": 0.67274272032131,
+      "grad_norm": 0.9453125,
+      "learning_rate": 1.2777239460119995e-05,
+      "loss": 6.8522,
+      "step": 8710
+    },
+    {
+      "epoch": 0.6735151000231714,
+      "grad_norm": 0.98046875,
+      "learning_rate": 1.2722720548118655e-05,
+      "loss": 6.8379,
+      "step": 8720
+    },
+    {
+      "epoch": 0.6742874797250328,
+      "grad_norm": 0.95703125,
+      "learning_rate": 1.2668278471401628e-05,
+      "loss": 6.8216,
+      "step": 8730
+    },
+    {
+      "epoch": 0.6750598594268943,
+      "grad_norm": 0.96875,
+      "learning_rate": 1.2613913570685499e-05,
+      "loss": 6.8192,
+      "step": 8740
+    },
+    {
+      "epoch": 0.6758322391287557,
+      "grad_norm": 0.94921875,
+      "learning_rate": 1.2559626186203812e-05,
+      "loss": 6.8368,
+      "step": 8750
+    },
+    {
+      "epoch": 0.6766046188306172,
+      "grad_norm": 0.98046875,
+      "learning_rate": 1.2505416657705027e-05,
+      "loss": 6.8094,
+      "step": 8760
+    },
+    {
+      "epoch": 0.6773769985324786,
+      "grad_norm": 0.984375,
+      "learning_rate": 1.245128532445035e-05,
+      "loss": 6.8189,
+      "step": 8770
+    },
+    {
+      "epoch": 0.67814937823434,
+      "grad_norm": 0.96484375,
+      "learning_rate": 1.2397232525211591e-05,
+      "loss": 6.8342,
+      "step": 8780
+    },
+    {
+      "epoch": 0.6789217579362015,
+      "grad_norm": 0.92578125,
+      "learning_rate": 1.2343258598269092e-05,
+      "loss": 6.8394,
+      "step": 8790
+    },
+    {
+      "epoch": 0.6796941376380629,
+      "grad_norm": 0.94921875,
+      "learning_rate": 1.2289363881409585e-05,
+      "loss": 6.8193,
+      "step": 8800
+    },
+    {
+      "epoch": 0.6804665173399244,
+      "grad_norm": 0.96875,
+      "learning_rate": 1.2235548711924055e-05,
+      "loss": 6.8292,
+      "step": 8810
+    },
+    {
+      "epoch": 0.6812388970417858,
+      "grad_norm": 0.9453125,
+      "learning_rate": 1.218181342660569e-05,
+      "loss": 6.8358,
+      "step": 8820
+    },
+    {
+      "epoch": 0.6820112767436471,
+      "grad_norm": 0.9765625,
+      "learning_rate": 1.2128158361747702e-05,
+      "loss": 6.8498,
+      "step": 8830
+    },
+    {
+      "epoch": 0.6827836564455086,
+      "grad_norm": 0.97265625,
+      "learning_rate": 1.2074583853141274e-05,
+      "loss": 6.822,
+      "step": 8840
+    },
+    {
+      "epoch": 0.68355603614737,
+      "grad_norm": 0.9765625,
+      "learning_rate": 1.2021090236073454e-05,
+      "loss": 6.8273,
+      "step": 8850
+    },
+    {
+      "epoch": 0.6843284158492314,
+      "grad_norm": 0.98046875,
+      "learning_rate": 1.1967677845325018e-05,
+      "loss": 6.8194,
+      "step": 8860
+    },
+    {
+      "epoch": 0.6851007955510929,
+      "grad_norm": 0.9375,
+      "learning_rate": 1.19143470151684e-05,
+      "loss": 6.8348,
+      "step": 8870
+    },
+    {
+      "epoch": 0.6858731752529543,
+      "grad_norm": 0.984375,
+      "learning_rate": 1.1861098079365627e-05,
+      "loss": 6.8172,
+      "step": 8880
+    },
+    {
+      "epoch": 0.6866455549548158,
+      "grad_norm": 0.953125,
+      "learning_rate": 1.1807931371166193e-05,
+      "loss": 6.8189,
+      "step": 8890
+    },
+    {
+      "epoch": 0.6874179346566772,
+      "grad_norm": 1.0,
+      "learning_rate": 1.1754847223304966e-05,
+      "loss": 6.8213,
+      "step": 8900
+    },
+    {
+      "epoch": 0.6881903143585386,
+      "grad_norm": 0.98046875,
+      "learning_rate": 1.1701845968000155e-05,
+      "loss": 6.8348,
+      "step": 8910
+    },
+    {
+      "epoch": 0.6889626940604001,
+      "grad_norm": 0.9765625,
+      "learning_rate": 1.1648927936951168e-05,
+      "loss": 6.8143,
+      "step": 8920
+    },
+    {
+      "epoch": 0.6897350737622615,
+      "grad_norm": 0.96484375,
+      "learning_rate": 1.1596093461336593e-05,
+      "loss": 6.7887,
+      "step": 8930
+    },
+    {
+      "epoch": 0.690507453464123,
+      "grad_norm": 0.96875,
+      "learning_rate": 1.1543342871812098e-05,
+      "loss": 6.8407,
+      "step": 8940
+    },
+    {
+      "epoch": 0.6912798331659844,
+      "grad_norm": 0.99609375,
+      "learning_rate": 1.1490676498508343e-05,
+      "loss": 6.8382,
+      "step": 8950
+    },
+    {
+      "epoch": 0.6920522128678458,
+      "grad_norm": 0.9609375,
+      "learning_rate": 1.1438094671028954e-05,
+      "loss": 6.8115,
+      "step": 8960
+    },
+    {
+      "epoch": 0.6928245925697073,
+      "grad_norm": 0.9921875,
+      "learning_rate": 1.1385597718448448e-05,
+      "loss": 6.8152,
+      "step": 8970
+    },
+    {
+      "epoch": 0.6935969722715687,
+      "grad_norm": 0.94921875,
+      "learning_rate": 1.133318596931014e-05,
+      "loss": 6.816,
+      "step": 8980
+    },
+    {
+      "epoch": 0.6943693519734302,
+      "grad_norm": 0.9921875,
+      "learning_rate": 1.128085975162414e-05,
+      "loss": 6.8158,
+      "step": 8990
+    },
+    {
+      "epoch": 0.6951417316752916,
+      "grad_norm": 0.9609375,
+      "learning_rate": 1.1228619392865248e-05,
+      "loss": 6.8378,
+      "step": 9000
+    },
+    {
+      "epoch": 0.6951417316752916,
+      "eval_loss": 7.014639377593994,
+      "eval_runtime": 0.6909,
+      "eval_samples_per_second": 17.369,
+      "eval_steps_per_second": 2.895,
+      "step": 9000
+    },
+    {
+      "epoch": 0.695914111377153,
+      "grad_norm": 0.91796875,
+      "learning_rate": 1.1176465219970954e-05,
+      "loss": 6.8303,
+      "step": 9010
+    },
+    {
+      "epoch": 0.6966864910790145,
+      "grad_norm": 0.94140625,
+      "learning_rate": 1.1124397559339372e-05,
+      "loss": 6.8194,
+      "step": 9020
+    },
+    {
+      "epoch": 0.6974588707808759,
+      "grad_norm": 0.99609375,
+      "learning_rate": 1.1072416736827157e-05,
+      "loss": 6.8393,
+      "step": 9030
+    },
+    {
+      "epoch": 0.6982312504827373,
+      "grad_norm": 0.97265625,
+      "learning_rate": 1.1020523077747542e-05,
+      "loss": 6.8107,
+      "step": 9040
+    },
+    {
+      "epoch": 0.6990036301845988,
+      "grad_norm": 0.96484375,
+      "learning_rate": 1.0968716906868245e-05,
+      "loss": 6.8269,
+      "step": 9050
+    },
+    {
+      "epoch": 0.6997760098864602,
+      "grad_norm": 0.97265625,
+      "learning_rate": 1.0916998548409449e-05,
+      "loss": 6.8307,
+      "step": 9060
+    },
+    {
+      "epoch": 0.7005483895883217,
+      "grad_norm": 0.95703125,
+      "learning_rate": 1.0865368326041794e-05,
+      "loss": 6.8177,
+      "step": 9070
+    },
+    {
+      "epoch": 0.701320769290183,
+      "grad_norm": 0.97265625,
+      "learning_rate": 1.0813826562884313e-05,
+      "loss": 6.8308,
+      "step": 9080
+    },
+    {
+      "epoch": 0.7020931489920444,
+      "grad_norm": 0.984375,
+      "learning_rate": 1.076237358150245e-05,
+      "loss": 6.8162,
+      "step": 9090
+    },
+    {
+      "epoch": 0.702865528693906,
+      "grad_norm": 0.98046875,
+      "learning_rate": 1.0711009703906025e-05,
+      "loss": 6.8358,
+      "step": 9100
+    },
+    {
+      "epoch": 0.7036379083957673,
+      "grad_norm": 0.9765625,
+      "learning_rate": 1.0659735251547204e-05,
+      "loss": 6.8346,
+      "step": 9110
+    },
+    {
+      "epoch": 0.7044102880976288,
+      "grad_norm": 0.9765625,
+      "learning_rate": 1.0608550545318489e-05,
+      "loss": 6.8377,
+      "step": 9120
+    },
+    {
+      "epoch": 0.7051826677994902,
+      "grad_norm": 0.9609375,
+      "learning_rate": 1.0557455905550764e-05,
+      "loss": 6.833,
+      "step": 9130
+    },
+    {
+      "epoch": 0.7059550475013516,
+      "grad_norm": 0.9609375,
+      "learning_rate": 1.0506451652011215e-05,
+      "loss": 6.8228,
+      "step": 9140
+    },
+    {
+      "epoch": 0.7067274272032131,
+      "grad_norm": 0.97265625,
+      "learning_rate": 1.0455538103901351e-05,
+      "loss": 6.8116,
+      "step": 9150
+    },
+    {
+      "epoch": 0.7074998069050745,
+      "grad_norm": 0.97265625,
+      "learning_rate": 1.0404715579855042e-05,
+      "loss": 6.8423,
+      "step": 9160
+    },
+    {
+      "epoch": 0.7082721866069359,
+      "grad_norm": 0.97265625,
+      "learning_rate": 1.0353984397936501e-05,
+      "loss": 6.8284,
+      "step": 9170
+    },
+    {
+      "epoch": 0.7090445663087974,
+      "grad_norm": 0.9296875,
+      "learning_rate": 1.030334487563826e-05,
+      "loss": 6.8167,
+      "step": 9180
+    },
+    {
+      "epoch": 0.7098169460106588,
+      "grad_norm": 0.92578125,
+      "learning_rate": 1.0252797329879251e-05,
+      "loss": 6.832,
+      "step": 9190
+    },
+    {
+      "epoch": 0.7105893257125203,
+      "grad_norm": 0.95703125,
+      "learning_rate": 1.0202342077002758e-05,
+      "loss": 6.8038,
+      "step": 9200
+    },
+    {
+      "epoch": 0.7113617054143817,
+      "grad_norm": 0.9453125,
+      "learning_rate": 1.015197943277448e-05,
+      "loss": 6.8106,
+      "step": 9210
+    },
+    {
+      "epoch": 0.7121340851162431,
+      "grad_norm": 0.93359375,
+      "learning_rate": 1.0101709712380553e-05,
+      "loss": 6.8262,
+      "step": 9220
+    },
+    {
+      "epoch": 0.7129064648181046,
+      "grad_norm": 0.96484375,
+      "learning_rate": 1.0051533230425527e-05,
+      "loss": 6.8224,
+      "step": 9230
+    },
+    {
+      "epoch": 0.713678844519966,
+      "grad_norm": 0.98046875,
+      "learning_rate": 1.000145030093048e-05,
+      "loss": 6.8314,
+      "step": 9240
+    },
+    {
+      "epoch": 0.7144512242218275,
+      "grad_norm": 0.94140625,
+      "learning_rate": 9.951461237330964e-06,
+      "loss": 6.8123,
+      "step": 9250
+    },
+    {
+      "epoch": 0.7152236039236889,
+      "grad_norm": 1.0546875,
+      "learning_rate": 9.90156635247512e-06,
+      "loss": 6.844,
+      "step": 9260
+    },
+    {
+      "epoch": 0.7159959836255503,
+      "grad_norm": 0.984375,
+      "learning_rate": 9.851765958621686e-06,
+      "loss": 6.8326,
+      "step": 9270
+    },
+    {
+      "epoch": 0.7167683633274118,
+      "grad_norm": 0.95703125,
+      "learning_rate": 9.802060367438013e-06,
+      "loss": 6.8004,
+      "step": 9280
+    },
+    {
+      "epoch": 0.7175407430292732,
+      "grad_norm": 0.9453125,
+      "learning_rate": 9.752449889998178e-06,
+      "loss": 6.8414,
+      "step": 9290
+    },
+    {
+      "epoch": 0.7183131227311347,
+      "grad_norm": 0.97265625,
+      "learning_rate": 9.702934836781e-06,
+      "loss": 6.8128,
+      "step": 9300
+    },
+    {
+      "epoch": 0.7190855024329961,
+      "grad_norm": 0.9453125,
+      "learning_rate": 9.653515517668076e-06,
+      "loss": 6.8188,
+      "step": 9310
+    },
+    {
+      "epoch": 0.7198578821348575,
+      "grad_norm": 0.9765625,
+      "learning_rate": 9.604192241941909e-06,
+      "loss": 6.8317,
+      "step": 9320
+    },
+    {
+      "epoch": 0.720630261836719,
+      "grad_norm": 0.94140625,
+      "learning_rate": 9.554965318283884e-06,
+      "loss": 6.8294,
+      "step": 9330
+    },
+    {
+      "epoch": 0.7214026415385804,
+      "grad_norm": 0.94140625,
+      "learning_rate": 9.505835054772419e-06,
+      "loss": 6.8288,
+      "step": 9340
+    },
+    {
+      "epoch": 0.7221750212404418,
+      "grad_norm": 0.99609375,
+      "learning_rate": 9.456801758880995e-06,
+      "loss": 6.8147,
+      "step": 9350
+    },
+    {
+      "epoch": 0.7229474009423033,
+      "grad_norm": 0.94921875,
+      "learning_rate": 9.407865737476221e-06,
+      "loss": 6.8197,
+      "step": 9360
+    },
+    {
+      "epoch": 0.7237197806441646,
+      "grad_norm": 0.96875,
+      "learning_rate": 9.359027296815922e-06,
+      "loss": 6.8014,
+      "step": 9370
+    },
+    {
+      "epoch": 0.7244921603460261,
+      "grad_norm": 1.0078125,
+      "learning_rate": 9.310286742547286e-06,
+      "loss": 6.8195,
+      "step": 9380
+    },
+    {
+      "epoch": 0.7252645400478875,
+      "grad_norm": 0.984375,
+      "learning_rate": 9.261644379704836e-06,
+      "loss": 6.8393,
+      "step": 9390
+    },
+    {
+      "epoch": 0.7260369197497489,
+      "grad_norm": 0.9765625,
+      "learning_rate": 9.2131005127086e-06,
+      "loss": 6.8331,
+      "step": 9400
+    },
+    {
+      "epoch": 0.7268092994516104,
+      "grad_norm": 0.94921875,
+      "learning_rate": 9.16465544536219e-06,
+      "loss": 6.8498,
+      "step": 9410
+    },
+    {
+      "epoch": 0.7275816791534718,
+      "grad_norm": 0.96484375,
+      "learning_rate": 9.116309480850896e-06,
+      "loss": 6.8174,
+      "step": 9420
+    },
+    {
+      "epoch": 0.7283540588553333,
+      "grad_norm": 0.96484375,
+      "learning_rate": 9.068062921739803e-06,
+      "loss": 6.8289,
+      "step": 9430
+    },
+    {
+      "epoch": 0.7291264385571947,
+      "grad_norm": 1.0,
+      "learning_rate": 9.019916069971857e-06,
+      "loss": 6.8059,
+      "step": 9440
+    },
+    {
+      "epoch": 0.7298988182590561,
+      "grad_norm": 0.953125,
+      "learning_rate": 8.971869226865998e-06,
+      "loss": 6.8167,
+      "step": 9450
+    },
+    {
+      "epoch": 0.7306711979609176,
+      "grad_norm": 1.0234375,
+      "learning_rate": 8.923922693115332e-06,
+      "loss": 6.8478,
+      "step": 9460
+    },
+    {
+      "epoch": 0.731443577662779,
+      "grad_norm": 0.97265625,
+      "learning_rate": 8.876076768785144e-06,
+      "loss": 6.8241,
+      "step": 9470
+    },
+    {
+      "epoch": 0.7322159573646404,
+      "grad_norm": 0.953125,
+      "learning_rate": 8.828331753311076e-06,
+      "loss": 6.8402,
+      "step": 9480
+    },
+    {
+      "epoch": 0.7329883370665019,
+      "grad_norm": 1.03125,
+      "learning_rate": 8.780687945497271e-06,
+      "loss": 6.8236,
+      "step": 9490
+    },
+    {
+      "epoch": 0.7337607167683633,
+      "grad_norm": 0.96875,
+      "learning_rate": 8.733145643514478e-06,
+      "loss": 6.832,
+      "step": 9500
+    },
+    {
+      "epoch": 0.7345330964702248,
+      "grad_norm": 0.98828125,
+      "learning_rate": 8.685705144898168e-06,
+      "loss": 6.8297,
+      "step": 9510
+    },
+    {
+      "epoch": 0.7353054761720862,
+      "grad_norm": 1.03125,
+      "learning_rate": 8.638366746546722e-06,
+      "loss": 6.8337,
+      "step": 9520
+    },
+    {
+      "epoch": 0.7360778558739476,
+      "grad_norm": 0.9453125,
+      "learning_rate": 8.591130744719509e-06,
+      "loss": 6.8139,
+      "step": 9530
+    },
+    {
+      "epoch": 0.7368502355758091,
+      "grad_norm": 0.9609375,
+      "learning_rate": 8.543997435035095e-06,
+      "loss": 6.8182,
+      "step": 9540
+    },
+    {
+      "epoch": 0.7376226152776705,
+      "grad_norm": 0.98046875,
+      "learning_rate": 8.496967112469362e-06,
+      "loss": 6.8083,
+      "step": 9550
+    },
+    {
+      "epoch": 0.738394994979532,
+      "grad_norm": 0.9609375,
+      "learning_rate": 8.450040071353644e-06,
+      "loss": 6.8266,
+      "step": 9560
+    },
+    {
+      "epoch": 0.7391673746813934,
+      "grad_norm": 0.9453125,
+      "learning_rate": 8.403216605372934e-06,
+      "loss": 6.8182,
+      "step": 9570
+    },
+    {
+      "epoch": 0.7399397543832548,
+      "grad_norm": 0.9765625,
+      "learning_rate": 8.356497007563988e-06,
+      "loss": 6.84,
+      "step": 9580
+    },
+    {
+      "epoch": 0.7407121340851163,
+      "grad_norm": 0.94140625,
+      "learning_rate": 8.309881570313537e-06,
+      "loss": 6.8209,
+      "step": 9590
+    },
+    {
+      "epoch": 0.7414845137869777,
+      "grad_norm": 0.953125,
+      "learning_rate": 8.263370585356456e-06,
+      "loss": 6.8048,
+      "step": 9600
+    },
+    {
+      "epoch": 0.7422568934888392,
+      "grad_norm": 0.953125,
+      "learning_rate": 8.216964343773893e-06,
+      "loss": 6.8341,
+      "step": 9610
+    },
+    {
+      "epoch": 0.7430292731907006,
+      "grad_norm": 0.953125,
+      "learning_rate": 8.170663135991472e-06,
+      "loss": 6.8268,
+      "step": 9620
+    },
+    {
+      "epoch": 0.743801652892562,
+      "grad_norm": 0.95703125,
+      "learning_rate": 8.124467251777531e-06,
+      "loss": 6.8086,
+      "step": 9630
+    },
+    {
+      "epoch": 0.7445740325944235,
+      "grad_norm": 0.96875,
+      "learning_rate": 8.078376980241206e-06,
+      "loss": 6.8144,
+      "step": 9640
+    },
+    {
+      "epoch": 0.7453464122962848,
+      "grad_norm": 0.94921875,
+      "learning_rate": 8.032392609830707e-06,
+      "loss": 6.8222,
+      "step": 9650
+    },
+    {
+      "epoch": 0.7461187919981462,
+      "grad_norm": 0.97265625,
+      "learning_rate": 7.986514428331452e-06,
+      "loss": 6.8081,
+      "step": 9660
+    },
+    {
+      "epoch": 0.7468911717000077,
+      "grad_norm": 0.9453125,
+      "learning_rate": 7.94074272286431e-06,
+      "loss": 6.8137,
+      "step": 9670
+    },
+    {
+      "epoch": 0.7476635514018691,
+      "grad_norm": 0.94140625,
+      "learning_rate": 7.895077779883803e-06,
+      "loss": 6.8035,
+      "step": 9680
+    },
+    {
+      "epoch": 0.7484359311037306,
+      "grad_norm": 0.95703125,
+      "learning_rate": 7.849519885176266e-06,
+      "loss": 6.8088,
+      "step": 9690
+    },
+    {
+      "epoch": 0.749208310805592,
+      "grad_norm": 1.0703125,
+      "learning_rate": 7.804069323858091e-06,
+      "loss": 6.8103,
+      "step": 9700
+    },
+    {
+      "epoch": 0.7499806905074534,
+      "grad_norm": 1.0,
+      "learning_rate": 7.758726380373987e-06,
+      "loss": 6.8328,
+      "step": 9710
+    },
+    {
+      "epoch": 0.7507530702093149,
+      "grad_norm": 0.9453125,
+      "learning_rate": 7.71349133849511e-06,
+      "loss": 6.8084,
+      "step": 9720
+    },
+    {
+      "epoch": 0.7515254499111763,
+      "grad_norm": 0.96484375,
+      "learning_rate": 7.668364481317334e-06,
+      "loss": 6.8095,
+      "step": 9730
+    },
+    {
+      "epoch": 0.7522978296130378,
+      "grad_norm": 0.95703125,
+      "learning_rate": 7.623346091259504e-06,
+      "loss": 6.8118,
+      "step": 9740
+    },
+    {
+      "epoch": 0.7530702093148992,
+      "grad_norm": 0.9765625,
+      "learning_rate": 7.578436450061627e-06,
+      "loss": 6.808,
+      "step": 9750
+    },
+    {
+      "epoch": 0.7538425890167606,
+      "grad_norm": 0.9921875,
+      "learning_rate": 7.533635838783115e-06,
+      "loss": 6.8079,
+      "step": 9760
+    },
+    {
+      "epoch": 0.7546149687186221,
+      "grad_norm": 0.9453125,
+      "learning_rate": 7.488944537801054e-06,
+      "loss": 6.8049,
+      "step": 9770
+    },
+    {
+      "epoch": 0.7553873484204835,
+      "grad_norm": 0.9375,
+      "learning_rate": 7.444362826808407e-06,
+      "loss": 6.7937,
+      "step": 9780
+    },
+    {
+      "epoch": 0.756159728122345,
+      "grad_norm": 0.9609375,
+      "learning_rate": 7.399890984812305e-06,
+      "loss": 6.8273,
+      "step": 9790
+    },
+    {
+      "epoch": 0.7569321078242064,
+      "grad_norm": 0.9609375,
+      "learning_rate": 7.355529290132285e-06,
+      "loss": 6.8404,
+      "step": 9800
+    },
+    {
+      "epoch": 0.7577044875260678,
+      "grad_norm": 0.93359375,
+      "learning_rate": 7.311278020398518e-06,
+      "loss": 6.8323,
+      "step": 9810
+    },
+    {
+      "epoch": 0.7584768672279293,
+      "grad_norm": 0.99609375,
+      "learning_rate": 7.2671374525501294e-06,
+      "loss": 6.8189,
+      "step": 9820
+    },
+    {
+      "epoch": 0.7592492469297907,
+      "grad_norm": 0.96484375,
+      "learning_rate": 7.223107862833431e-06,
+      "loss": 6.8409,
+      "step": 9830
+    },
+    {
+      "epoch": 0.7600216266316521,
+      "grad_norm": 0.9765625,
+      "learning_rate": 7.179189526800173e-06,
+      "loss": 6.8037,
+      "step": 9840
+    },
+    {
+      "epoch": 0.7607940063335136,
+      "grad_norm": 0.9296875,
+      "learning_rate": 7.135382719305875e-06,
+      "loss": 6.8286,
+      "step": 9850
+    },
+    {
+      "epoch": 0.761566386035375,
+      "grad_norm": 0.9609375,
+      "learning_rate": 7.091687714508044e-06,
+      "loss": 6.8186,
+      "step": 9860
+    },
+    {
+      "epoch": 0.7623387657372365,
+      "grad_norm": 0.94921875,
+      "learning_rate": 7.048104785864509e-06,
+      "loss": 6.8407,
+      "step": 9870
+    },
+    {
+      "epoch": 0.7631111454390979,
+      "grad_norm": 1.0,
+      "learning_rate": 7.004634206131696e-06,
+      "loss": 6.8385,
+      "step": 9880
+    },
+    {
+      "epoch": 0.7638835251409593,
+      "grad_norm": 0.953125,
+      "learning_rate": 6.9612762473628855e-06,
+      "loss": 6.832,
+      "step": 9890
+    },
+    {
+      "epoch": 0.7646559048428208,
+      "grad_norm": 0.921875,
+      "learning_rate": 6.918031180906567e-06,
+      "loss": 6.8112,
+      "step": 9900
+    },
+    {
+      "epoch": 0.7654282845446821,
+      "grad_norm": 0.97265625,
+      "learning_rate": 6.874899277404689e-06,
+      "loss": 6.8291,
+      "step": 9910
+    },
+    {
+      "epoch": 0.7662006642465437,
+      "grad_norm": 0.9765625,
+      "learning_rate": 6.8318808067910074e-06,
+      "loss": 6.8402,
+      "step": 9920
+    },
+    {
+      "epoch": 0.766973043948405,
+      "grad_norm": 0.99609375,
+      "learning_rate": 6.788976038289374e-06,
+      "loss": 6.8414,
+      "step": 9930
+    },
+    {
+      "epoch": 0.7677454236502664,
+      "grad_norm": 0.9609375,
+      "learning_rate": 6.746185240412051e-06,
+      "loss": 6.8256,
+      "step": 9940
+    },
+    {
+      "epoch": 0.7685178033521279,
+      "grad_norm": 0.98046875,
+      "learning_rate": 6.703508680958015e-06,
+      "loss": 6.8026,
+      "step": 9950
+    },
+    {
+      "epoch": 0.7692901830539893,
+      "grad_norm": 0.98828125,
+      "learning_rate": 6.660946627011353e-06,
+      "loss": 6.7947,
+      "step": 9960
+    },
+    {
+      "epoch": 0.7700625627558507,
+      "grad_norm": 0.94921875,
+      "learning_rate": 6.618499344939491e-06,
+      "loss": 6.8282,
+      "step": 9970
+    },
+    {
+      "epoch": 0.7708349424577122,
+      "grad_norm": 0.98046875,
+      "learning_rate": 6.576167100391584e-06,
+      "loss": 6.8416,
+      "step": 9980
+    },
+    {
+      "epoch": 0.7716073221595736,
+      "grad_norm": 0.953125,
+      "learning_rate": 6.533950158296856e-06,
+      "loss": 6.8196,
+      "step": 9990
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 12947,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 999,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 2.29364368331988e+19,
+  "train_batch_size": 12,
+  "trial_name": null,
+  "trial_params": null
+}