diff --git "a/results/checkpoint-8922/trainer_state.json" "b/results/checkpoint-8922/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/results/checkpoint-8922/trainer_state.json"
@@ -0,0 +1,6413 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 6.0,
+  "eval_steps": 500,
+  "global_step": 8922,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.006724949562878279,
+      "grad_norm": 7.708007335662842,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 5.3528,
+      "step": 10
+    },
+    {
+      "epoch": 0.013449899125756557,
+      "grad_norm": 4.7876434326171875,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 5.3265,
+      "step": 20
+    },
+    {
+      "epoch": 0.020174848688634835,
+      "grad_norm": 4.998968124389648,
+      "learning_rate": 3e-06,
+      "loss": 5.2177,
+      "step": 30
+    },
+    {
+      "epoch": 0.026899798251513115,
+      "grad_norm": 4.760256290435791,
+      "learning_rate": 3.9e-06,
+      "loss": 5.2683,
+      "step": 40
+    },
+    {
+      "epoch": 0.03362474781439139,
+      "grad_norm": 6.136503219604492,
+      "learning_rate": 4.9000000000000005e-06,
+      "loss": 5.0256,
+      "step": 50
+    },
+    {
+      "epoch": 0.04034969737726967,
+      "grad_norm": 4.262901782989502,
+      "learning_rate": 5.9e-06,
+      "loss": 4.9378,
+      "step": 60
+    },
+    {
+      "epoch": 0.04707464694014795,
+      "grad_norm": 3.3165853023529053,
+      "learning_rate": 6.900000000000001e-06,
+      "loss": 4.6847,
+      "step": 70
+    },
+    {
+      "epoch": 0.05379959650302623,
+      "grad_norm": 3.6227917671203613,
+      "learning_rate": 7.9e-06,
+      "loss": 4.573,
+      "step": 80
+    },
+    {
+      "epoch": 0.0605245460659045,
+      "grad_norm": 3.023857593536377,
+      "learning_rate": 8.8e-06,
+      "loss": 4.2629,
+      "step": 90
+    },
+    {
+      "epoch": 0.06724949562878278,
+      "grad_norm": 2.9306392669677734,
+      "learning_rate": 9.800000000000001e-06,
+      "loss": 3.9944,
+      "step": 100
+    },
+    {
+      "epoch": 0.07397444519166106,
+      "grad_norm": 2.7784805297851562,
+      "learning_rate": 1.0700000000000001e-05,
+      "loss": 3.5371,
+      "step": 110
+    },
+    {
+      "epoch": 0.08069939475453934,
+      "grad_norm": 2.2752227783203125,
+      "learning_rate": 1.1700000000000001e-05,
+      "loss": 3.5498,
+      "step": 120
+    },
+    {
+      "epoch": 0.08742434431741762,
+      "grad_norm": 2.1082851886749268,
+      "learning_rate": 1.27e-05,
+      "loss": 3.3127,
+      "step": 130
+    },
+    {
+      "epoch": 0.0941492938802959,
+      "grad_norm": 2.923049211502075,
+      "learning_rate": 1.3700000000000001e-05,
+      "loss": 3.102,
+      "step": 140
+    },
+    {
+      "epoch": 0.10087424344317418,
+      "grad_norm": 2.354473352432251,
+      "learning_rate": 1.47e-05,
+      "loss": 2.9127,
+      "step": 150
+    },
+    {
+      "epoch": 0.10759919300605246,
+      "grad_norm": 2.0444324016571045,
+      "learning_rate": 1.5700000000000002e-05,
+      "loss": 2.8566,
+      "step": 160
+    },
+    {
+      "epoch": 0.11432414256893074,
+      "grad_norm": 3.295562267303467,
+      "learning_rate": 1.6700000000000003e-05,
+      "loss": 2.6766,
+      "step": 170
+    },
+    {
+      "epoch": 0.121049092131809,
+      "grad_norm": 21.337621688842773,
+      "learning_rate": 1.77e-05,
+      "loss": 2.4635,
+      "step": 180
+    },
+    {
+      "epoch": 0.12777404169468728,
+      "grad_norm": 2.203986167907715,
+      "learning_rate": 1.87e-05,
+      "loss": 2.4118,
+      "step": 190
+    },
+    {
+      "epoch": 0.13449899125756556,
+      "grad_norm": 2.7684690952301025,
+      "learning_rate": 1.97e-05,
+      "loss": 2.2789,
+      "step": 200
+    },
+    {
+      "epoch": 0.14122394082044384,
+      "grad_norm": 2.3202223777770996,
+      "learning_rate": 2.07e-05,
+      "loss": 2.1204,
+      "step": 210
+    },
+    {
+      "epoch": 0.14794889038332212,
+      "grad_norm": 2.229384660720825,
+      "learning_rate": 2.1700000000000002e-05,
+      "loss": 1.9699,
+      "step": 220
+    },
+    {
+      "epoch": 0.1546738399462004,
+      "grad_norm": 2.0208075046539307,
+      "learning_rate": 2.2700000000000003e-05,
+      "loss": 1.935,
+      "step": 230
+    },
+    {
+      "epoch": 0.16139878950907868,
+      "grad_norm": 2.1106619834899902,
+      "learning_rate": 2.37e-05,
+      "loss": 1.8648,
+      "step": 240
+    },
+    {
+      "epoch": 0.16812373907195696,
+      "grad_norm": 2.151031017303467,
+      "learning_rate": 2.47e-05,
+      "loss": 1.7949,
+      "step": 250
+    },
+    {
+      "epoch": 0.17484868863483524,
+      "grad_norm": 2.3149192333221436,
+      "learning_rate": 2.57e-05,
+      "loss": 1.6146,
+      "step": 260
+    },
+    {
+      "epoch": 0.18157363819771352,
+      "grad_norm": 2.266303539276123,
+      "learning_rate": 2.6700000000000002e-05,
+      "loss": 1.6113,
+      "step": 270
+    },
+    {
+      "epoch": 0.1882985877605918,
+      "grad_norm": NaN,
+      "learning_rate": 2.7500000000000004e-05,
+      "loss": 1.5469,
+      "step": 280
+    },
+    {
+      "epoch": 0.19502353732347008,
+      "grad_norm": 2.277637243270874,
+      "learning_rate": 2.8499999999999998e-05,
+      "loss": 1.5391,
+      "step": 290
+    },
+    {
+      "epoch": 0.20174848688634836,
+      "grad_norm": 2.165311574935913,
+      "learning_rate": 2.95e-05,
+      "loss": 1.407,
+      "step": 300
+    },
+    {
+      "epoch": 0.20847343644922664,
+      "grad_norm": 2.2575840950012207,
+      "learning_rate": 3.05e-05,
+      "loss": 1.2985,
+      "step": 310
+    },
+    {
+      "epoch": 0.21519838601210492,
+      "grad_norm": 2.3073315620422363,
+      "learning_rate": 3.15e-05,
+      "loss": 1.3224,
+      "step": 320
+    },
+    {
+      "epoch": 0.2219233355749832,
+      "grad_norm": 2.1623892784118652,
+      "learning_rate": 3.2500000000000004e-05,
+      "loss": 1.314,
+      "step": 330
+    },
+    {
+      "epoch": 0.22864828513786148,
+      "grad_norm": 1.951863408088684,
+      "learning_rate": 3.3400000000000005e-05,
+      "loss": 1.257,
+      "step": 340
+    },
+    {
+      "epoch": 0.23537323470073973,
+      "grad_norm": 2.1350021362304688,
+      "learning_rate": 3.4399999999999996e-05,
+      "loss": 1.1251,
+      "step": 350
+    },
+    {
+      "epoch": 0.242098184263618,
+      "grad_norm": 2.1534841060638428,
+      "learning_rate": 3.53e-05,
+      "loss": 1.1653,
+      "step": 360
+    },
+    {
+      "epoch": 0.2488231338264963,
+      "grad_norm": 1.8336793184280396,
+      "learning_rate": 3.62e-05,
+      "loss": 1.1436,
+      "step": 370
+    },
+    {
+      "epoch": 0.25554808338937457,
+      "grad_norm": 1.8173967599868774,
+      "learning_rate": 3.72e-05,
+      "loss": 1.065,
+      "step": 380
+    },
+    {
+      "epoch": 0.26227303295225285,
+      "grad_norm": 1.9319089651107788,
+      "learning_rate": 3.82e-05,
+      "loss": 1.0498,
+      "step": 390
+    },
+    {
+      "epoch": 0.26899798251513113,
+      "grad_norm": 5.745934009552002,
+      "learning_rate": 3.9200000000000004e-05,
+      "loss": 1.0344,
+      "step": 400
+    },
+    {
+      "epoch": 0.2757229320780094,
+      "grad_norm": 1.7822935581207275,
+      "learning_rate": 4.02e-05,
+      "loss": 1.0393,
+      "step": 410
+    },
+    {
+      "epoch": 0.2824478816408877,
+      "grad_norm": 1.6039519309997559,
+      "learning_rate": 4.12e-05,
+      "loss": 0.9708,
+      "step": 420
+    },
+    {
+      "epoch": 0.28917283120376597,
+      "grad_norm": 2.8318278789520264,
+      "learning_rate": 4.22e-05,
+      "loss": 0.9958,
+      "step": 430
+    },
+    {
+      "epoch": 0.29589778076664425,
+      "grad_norm": 2.0125811100006104,
+      "learning_rate": 4.32e-05,
+      "loss": 0.9762,
+      "step": 440
+    },
+    {
+      "epoch": 0.3026227303295225,
+      "grad_norm": 1.8111716508865356,
+      "learning_rate": 4.4200000000000004e-05,
+      "loss": 0.8681,
+      "step": 450
+    },
+    {
+      "epoch": 0.3093476798924008,
+      "grad_norm": 1.900375247001648,
+      "learning_rate": 4.52e-05,
+      "loss": 0.8475,
+      "step": 460
+    },
+    {
+      "epoch": 0.3160726294552791,
+      "grad_norm": 2.1886966228485107,
+      "learning_rate": 4.61e-05,
+      "loss": 0.9559,
+      "step": 470
+    },
+    {
+      "epoch": 0.32279757901815737,
+      "grad_norm": 1.6458282470703125,
+      "learning_rate": 4.71e-05,
+      "loss": 0.83,
+      "step": 480
+    },
+    {
+      "epoch": 0.32952252858103565,
+      "grad_norm": 1.9217121601104736,
+      "learning_rate": 4.8100000000000004e-05,
+      "loss": 0.8023,
+      "step": 490
+    },
+    {
+      "epoch": 0.3362474781439139,
+      "grad_norm": 1.5879414081573486,
+      "learning_rate": 4.91e-05,
+      "loss": 0.7311,
+      "step": 500
+    },
+    {
+      "epoch": 0.3362474781439139,
+      "eval_loss": 0.6057256460189819,
+      "eval_runtime": 14.7732,
+      "eval_samples_per_second": 178.973,
+      "eval_steps_per_second": 22.405,
+      "step": 500
+    },
+    {
+      "epoch": 0.3429724277067922,
+      "grad_norm": 1.7010548114776611,
+      "learning_rate": 4.9994063167893614e-05,
+      "loss": 0.8544,
+      "step": 510
+    },
+    {
+      "epoch": 0.3496973772696705,
+      "grad_norm": 1.644062876701355,
+      "learning_rate": 4.993469484682973e-05,
+      "loss": 0.7183,
+      "step": 520
+    },
+    {
+      "epoch": 0.35642232683254876,
+      "grad_norm": 1.8534233570098877,
+      "learning_rate": 4.987532652576585e-05,
+      "loss": 0.7635,
+      "step": 530
+    },
+    {
+      "epoch": 0.36314727639542704,
+      "grad_norm": 1.9692680835723877,
+      "learning_rate": 4.9815958204701975e-05,
+      "loss": 0.7794,
+      "step": 540
+    },
+    {
+      "epoch": 0.3698722259583053,
+      "grad_norm": 2.2176966667175293,
+      "learning_rate": 4.975658988363809e-05,
+      "loss": 0.6438,
+      "step": 550
+    },
+    {
+      "epoch": 0.3765971755211836,
+      "grad_norm": 1.6217116117477417,
+      "learning_rate": 4.969722156257421e-05,
+      "loss": 0.7305,
+      "step": 560
+    },
+    {
+      "epoch": 0.3833221250840619,
+      "grad_norm": 1.4970368146896362,
+      "learning_rate": 4.9637853241510336e-05,
+      "loss": 0.6838,
+      "step": 570
+    },
+    {
+      "epoch": 0.39004707464694016,
+      "grad_norm": 1.6664597988128662,
+      "learning_rate": 4.957848492044645e-05,
+      "loss": 0.6605,
+      "step": 580
+    },
+    {
+      "epoch": 0.39677202420981844,
+      "grad_norm": 1.796850323677063,
+      "learning_rate": 4.951911659938257e-05,
+      "loss": 0.5982,
+      "step": 590
+    },
+    {
+      "epoch": 0.4034969737726967,
+      "grad_norm": 1.7564682960510254,
+      "learning_rate": 4.945974827831869e-05,
+      "loss": 0.6208,
+      "step": 600
+    },
+    {
+      "epoch": 0.410221923335575,
+      "grad_norm": 1.770803689956665,
+      "learning_rate": 4.940037995725481e-05,
+      "loss": 0.6699,
+      "step": 610
+    },
+    {
+      "epoch": 0.4169468728984533,
+      "grad_norm": 1.7247834205627441,
+      "learning_rate": 4.934101163619093e-05,
+      "loss": 0.6711,
+      "step": 620
+    },
+    {
+      "epoch": 0.42367182246133156,
+      "grad_norm": 1.7561867237091064,
+      "learning_rate": 4.928164331512705e-05,
+      "loss": 0.6286,
+      "step": 630
+    },
+    {
+      "epoch": 0.43039677202420984,
+      "grad_norm": 1.5144267082214355,
+      "learning_rate": 4.922227499406317e-05,
+      "loss": 0.6011,
+      "step": 640
+    },
+    {
+      "epoch": 0.4371217215870881,
+      "grad_norm": 1.731772780418396,
+      "learning_rate": 4.9162906672999294e-05,
+      "loss": 0.5068,
+      "step": 650
+    },
+    {
+      "epoch": 0.4438466711499664,
+      "grad_norm": 1.797025203704834,
+      "learning_rate": 4.9103538351935406e-05,
+      "loss": 0.5727,
+      "step": 660
+    },
+    {
+      "epoch": 0.4505716207128447,
+      "grad_norm": 1.615449070930481,
+      "learning_rate": 4.904417003087153e-05,
+      "loss": 0.6116,
+      "step": 670
+    },
+    {
+      "epoch": 0.45729657027572296,
+      "grad_norm": 1.2763652801513672,
+      "learning_rate": 4.898480170980765e-05,
+      "loss": 0.5898,
+      "step": 680
+    },
+    {
+      "epoch": 0.46402151983860124,
+      "grad_norm": 1.553597331047058,
+      "learning_rate": 4.8925433388743767e-05,
+      "loss": 0.5822,
+      "step": 690
+    },
+    {
+      "epoch": 0.47074646940147946,
+      "grad_norm": 1.478072166442871,
+      "learning_rate": 4.886606506767989e-05,
+      "loss": 0.565,
+      "step": 700
+    },
+    {
+      "epoch": 0.47747141896435774,
+      "grad_norm": 1.3752771615982056,
+      "learning_rate": 4.880669674661601e-05,
+      "loss": 0.5575,
+      "step": 710
+    },
+    {
+      "epoch": 0.484196368527236,
+      "grad_norm": 1.5757756233215332,
+      "learning_rate": 4.874732842555213e-05,
+      "loss": 0.5506,
+      "step": 720
+    },
+    {
+      "epoch": 0.4909213180901143,
+      "grad_norm": 1.8003489971160889,
+      "learning_rate": 4.8687960104488246e-05,
+      "loss": 0.5185,
+      "step": 730
+    },
+    {
+      "epoch": 0.4976462676529926,
+      "grad_norm": 1.8582769632339478,
+      "learning_rate": 4.8628591783424364e-05,
+      "loss": 0.4633,
+      "step": 740
+    },
+    {
+      "epoch": 0.5043712172158709,
+      "grad_norm": 1.8770896196365356,
+      "learning_rate": 4.856922346236049e-05,
+      "loss": 0.5392,
+      "step": 750
+    },
+    {
+      "epoch": 0.5110961667787491,
+      "grad_norm": 1.782762885093689,
+      "learning_rate": 4.8509855141296607e-05,
+      "loss": 0.5292,
+      "step": 760
+    },
+    {
+      "epoch": 0.5178211163416274,
+      "grad_norm": 1.5999263525009155,
+      "learning_rate": 4.8450486820232725e-05,
+      "loss": 0.5244,
+      "step": 770
+    },
+    {
+      "epoch": 0.5245460659045057,
+      "grad_norm": 1.4285874366760254,
+      "learning_rate": 4.839111849916885e-05,
+      "loss": 0.5169,
+      "step": 780
+    },
+    {
+      "epoch": 0.531271015467384,
+      "grad_norm": 1.814850091934204,
+      "learning_rate": 4.833175017810497e-05,
+      "loss": 0.4731,
+      "step": 790
+    },
+    {
+      "epoch": 0.5379959650302623,
+      "grad_norm": 1.5589072704315186,
+      "learning_rate": 4.8272381857041086e-05,
+      "loss": 0.5805,
+      "step": 800
+    },
+    {
+      "epoch": 0.5447209145931405,
+      "grad_norm": 1.4412167072296143,
+      "learning_rate": 4.8213013535977204e-05,
+      "loss": 0.5044,
+      "step": 810
+    },
+    {
+      "epoch": 0.5514458641560188,
+      "grad_norm": 1.56436288356781,
+      "learning_rate": 4.815364521491332e-05,
+      "loss": 0.4496,
+      "step": 820
+    },
+    {
+      "epoch": 0.5581708137188971,
+      "grad_norm": 1.2886687517166138,
+      "learning_rate": 4.8094276893849447e-05,
+      "loss": 0.4655,
+      "step": 830
+    },
+    {
+      "epoch": 0.5648957632817754,
+      "grad_norm": 1.4733515977859497,
+      "learning_rate": 4.8034908572785565e-05,
+      "loss": 0.4916,
+      "step": 840
+    },
+    {
+      "epoch": 0.5716207128446537,
+      "grad_norm": 1.6645103693008423,
+      "learning_rate": 4.797554025172168e-05,
+      "loss": 0.5046,
+      "step": 850
+    },
+    {
+      "epoch": 0.5783456624075319,
+      "grad_norm": 1.0857338905334473,
+      "learning_rate": 4.791617193065781e-05,
+      "loss": 0.4996,
+      "step": 860
+    },
+    {
+      "epoch": 0.5850706119704102,
+      "grad_norm": 1.4320952892303467,
+      "learning_rate": 4.7856803609593926e-05,
+      "loss": 0.4788,
+      "step": 870
+    },
+    {
+      "epoch": 0.5917955615332885,
+      "grad_norm": 1.6821132898330688,
+      "learning_rate": 4.7797435288530044e-05,
+      "loss": 0.5284,
+      "step": 880
+    },
+    {
+      "epoch": 0.5985205110961668,
+      "grad_norm": 1.1405266523361206,
+      "learning_rate": 4.773806696746616e-05,
+      "loss": 0.5179,
+      "step": 890
+    },
+    {
+      "epoch": 0.605245460659045,
+      "grad_norm": 1.2930107116699219,
+      "learning_rate": 4.767869864640228e-05,
+      "loss": 0.4225,
+      "step": 900
+    },
+    {
+      "epoch": 0.6119704102219233,
+      "grad_norm": 1.2401055097579956,
+      "learning_rate": 4.7619330325338405e-05,
+      "loss": 0.4995,
+      "step": 910
+    },
+    {
+      "epoch": 0.6186953597848016,
+      "grad_norm": 1.5603152513504028,
+      "learning_rate": 4.755996200427452e-05,
+      "loss": 0.4895,
+      "step": 920
+    },
+    {
+      "epoch": 0.6254203093476799,
+      "grad_norm": 1.407322883605957,
+      "learning_rate": 4.750059368321064e-05,
+      "loss": 0.3682,
+      "step": 930
+    },
+    {
+      "epoch": 0.6321452589105582,
+      "grad_norm": 1.2464901208877563,
+      "learning_rate": 4.7441225362146766e-05,
+      "loss": 0.4341,
+      "step": 940
+    },
+    {
+      "epoch": 0.6388702084734365,
+      "grad_norm": 1.2959908246994019,
+      "learning_rate": 4.738185704108288e-05,
+      "loss": 0.4965,
+      "step": 950
+    },
+    {
+      "epoch": 0.6455951580363147,
+      "grad_norm": 1.4426651000976562,
+      "learning_rate": 4.7322488720018995e-05,
+      "loss": 0.4894,
+      "step": 960
+    },
+    {
+      "epoch": 0.652320107599193,
+      "grad_norm": 1.3427236080169678,
+      "learning_rate": 4.726312039895512e-05,
+      "loss": 0.4193,
+      "step": 970
+    },
+    {
+      "epoch": 0.6590450571620713,
+      "grad_norm": 1.3153373003005981,
+      "learning_rate": 4.720375207789124e-05,
+      "loss": 0.4719,
+      "step": 980
+    },
+    {
+      "epoch": 0.6657700067249496,
+      "grad_norm": 1.2846280336380005,
+      "learning_rate": 4.714438375682736e-05,
+      "loss": 0.4027,
+      "step": 990
+    },
+    {
+      "epoch": 0.6724949562878278,
+      "grad_norm": 1.559920072555542,
+      "learning_rate": 4.708501543576348e-05,
+      "loss": 0.4952,
+      "step": 1000
+    },
+    {
+      "epoch": 0.6724949562878278,
+      "eval_loss": 0.3311997950077057,
+      "eval_runtime": 14.4715,
+      "eval_samples_per_second": 182.704,
+      "eval_steps_per_second": 22.873,
+      "step": 1000
+    },
+    {
+      "epoch": 0.6792199058507061,
+      "grad_norm": 1.4075298309326172,
+      "learning_rate": 4.70256471146996e-05,
+      "loss": 0.392,
+      "step": 1010
+    },
+    {
+      "epoch": 0.6859448554135844,
+      "grad_norm": 1.2165625095367432,
+      "learning_rate": 4.6966278793635724e-05,
+      "loss": 0.3714,
+      "step": 1020
+    },
+    {
+      "epoch": 0.6926698049764627,
+      "grad_norm": 1.2659556865692139,
+      "learning_rate": 4.6906910472571835e-05,
+      "loss": 0.4422,
+      "step": 1030
+    },
+    {
+      "epoch": 0.699394754539341,
+      "grad_norm": 1.6019095182418823,
+      "learning_rate": 4.684754215150795e-05,
+      "loss": 0.4335,
+      "step": 1040
+    },
+    {
+      "epoch": 0.7061197041022192,
+      "grad_norm": 1.350982904434204,
+      "learning_rate": 4.678817383044408e-05,
+      "loss": 0.4364,
+      "step": 1050
+    },
+    {
+      "epoch": 0.7128446536650975,
+      "grad_norm": 1.059822678565979,
+      "learning_rate": 4.6728805509380196e-05,
+      "loss": 0.3972,
+      "step": 1060
+    },
+    {
+      "epoch": 0.7195696032279758,
+      "grad_norm": 1.1738590002059937,
+      "learning_rate": 4.6669437188316314e-05,
+      "loss": 0.4753,
+      "step": 1070
+    },
+    {
+      "epoch": 0.7262945527908541,
+      "grad_norm": 1.3485381603240967,
+      "learning_rate": 4.661006886725244e-05,
+      "loss": 0.4033,
+      "step": 1080
+    },
+    {
+      "epoch": 0.7330195023537324,
+      "grad_norm": 1.2352713346481323,
+      "learning_rate": 4.655070054618856e-05,
+      "loss": 0.3844,
+      "step": 1090
+    },
+    {
+      "epoch": 0.7397444519166106,
+      "grad_norm": 0.9524800181388855,
+      "learning_rate": 4.6491332225124675e-05,
+      "loss": 0.3292,
+      "step": 1100
+    },
+    {
+      "epoch": 0.7464694014794889,
+      "grad_norm": 1.5953115224838257,
+      "learning_rate": 4.643196390406079e-05,
+      "loss": 0.4262,
+      "step": 1110
+    },
+    {
+      "epoch": 0.7531943510423672,
+      "grad_norm": 1.0904059410095215,
+      "learning_rate": 4.637259558299691e-05,
+      "loss": 0.3794,
+      "step": 1120
+    },
+    {
+      "epoch": 0.7599193006052455,
+      "grad_norm": 1.1243659257888794,
+      "learning_rate": 4.6313227261933036e-05,
+      "loss": 0.3931,
+      "step": 1130
+    },
+    {
+      "epoch": 0.7666442501681238,
+      "grad_norm": 1.7865381240844727,
+      "learning_rate": 4.6253858940869154e-05,
+      "loss": 0.3758,
+      "step": 1140
+    },
+    {
+      "epoch": 0.773369199731002,
+      "grad_norm": 1.322513222694397,
+      "learning_rate": 4.619449061980527e-05,
+      "loss": 0.3805,
+      "step": 1150
+    },
+    {
+      "epoch": 0.7800941492938803,
+      "grad_norm": 1.1556380987167358,
+      "learning_rate": 4.61351222987414e-05,
+      "loss": 0.4068,
+      "step": 1160
+    },
+    {
+      "epoch": 0.7868190988567586,
+      "grad_norm": 4.545334339141846,
+      "learning_rate": 4.6075753977677515e-05,
+      "loss": 0.3823,
+      "step": 1170
+    },
+    {
+      "epoch": 0.7935440484196369,
+      "grad_norm": 1.322127103805542,
+      "learning_rate": 4.601638565661363e-05,
+      "loss": 0.4151,
+      "step": 1180
+    },
+    {
+      "epoch": 0.8002689979825152,
+      "grad_norm": 0.9689624905586243,
+      "learning_rate": 4.595701733554975e-05,
+      "loss": 0.3771,
+      "step": 1190
+    },
+    {
+      "epoch": 0.8069939475453934,
+      "grad_norm": 1.1350356340408325,
+      "learning_rate": 4.589764901448587e-05,
+      "loss": 0.3708,
+      "step": 1200
+    },
+    {
+      "epoch": 0.8137188971082717,
+      "grad_norm": 1.15723717212677,
+      "learning_rate": 4.5838280693421994e-05,
+      "loss": 0.4063,
+      "step": 1210
+    },
+    {
+      "epoch": 0.82044384667115,
+      "grad_norm": 1.017109990119934,
+      "learning_rate": 4.577891237235811e-05,
+      "loss": 0.4116,
+      "step": 1220
+    },
+    {
+      "epoch": 0.8271687962340283,
+      "grad_norm": 1.16199791431427,
+      "learning_rate": 4.571954405129423e-05,
+      "loss": 0.4085,
+      "step": 1230
+    },
+    {
+      "epoch": 0.8338937457969066,
+      "grad_norm": 1.7634154558181763,
+      "learning_rate": 4.5660175730230355e-05,
+      "loss": 0.369,
+      "step": 1240
+    },
+    {
+      "epoch": 0.8406186953597848,
+      "grad_norm": 1.144355297088623,
+      "learning_rate": 4.5600807409166466e-05,
+      "loss": 0.3807,
+      "step": 1250
+    },
+    {
+      "epoch": 0.8473436449226631,
+      "grad_norm": 1.0942720174789429,
+      "learning_rate": 4.554143908810259e-05,
+      "loss": 0.3907,
+      "step": 1260
+    },
+    {
+      "epoch": 0.8540685944855414,
+      "grad_norm": 1.1029889583587646,
+      "learning_rate": 4.548207076703871e-05,
+      "loss": 0.3803,
+      "step": 1270
+    },
+    {
+      "epoch": 0.8607935440484197,
+      "grad_norm": 0.9530211687088013,
+      "learning_rate": 4.542270244597483e-05,
+      "loss": 0.3334,
+      "step": 1280
+    },
+    {
+      "epoch": 0.867518493611298,
+      "grad_norm": 1.1777485609054565,
+      "learning_rate": 4.536333412491095e-05,
+      "loss": 0.3641,
+      "step": 1290
+    },
+    {
+      "epoch": 0.8742434431741762,
+      "grad_norm": 1.2794089317321777,
+      "learning_rate": 4.530396580384707e-05,
+      "loss": 0.3405,
+      "step": 1300
+    },
+    {
+      "epoch": 0.8809683927370545,
+      "grad_norm": 1.0702439546585083,
+      "learning_rate": 4.524459748278319e-05,
+      "loss": 0.396,
+      "step": 1310
+    },
+    {
+      "epoch": 0.8876933422999328,
+      "grad_norm": 1.5103455781936646,
+      "learning_rate": 4.518522916171931e-05,
+      "loss": 0.4143,
+      "step": 1320
+    },
+    {
+      "epoch": 0.8944182918628111,
+      "grad_norm": 1.0924547910690308,
+      "learning_rate": 4.5125860840655424e-05,
+      "loss": 0.3117,
+      "step": 1330
+    },
+    {
+      "epoch": 0.9011432414256894,
+      "grad_norm": 1.2703629732131958,
+      "learning_rate": 4.506649251959155e-05,
+      "loss": 0.3284,
+      "step": 1340
+    },
+    {
+      "epoch": 0.9078681909885676,
+      "grad_norm": 1.138601541519165,
+      "learning_rate": 4.500712419852767e-05,
+      "loss": 0.325,
+      "step": 1350
+    },
+    {
+      "epoch": 0.9145931405514459,
+      "grad_norm": 1.1391545534133911,
+      "learning_rate": 4.4947755877463785e-05,
+      "loss": 0.3468,
+      "step": 1360
+    },
+    {
+      "epoch": 0.9213180901143242,
+      "grad_norm": 1.068265676498413,
+      "learning_rate": 4.488838755639991e-05,
+      "loss": 0.4056,
+      "step": 1370
+    },
+    {
+      "epoch": 0.9280430396772025,
+      "grad_norm": 1.1206265687942505,
+      "learning_rate": 4.482901923533603e-05,
+      "loss": 0.3516,
+      "step": 1380
+    },
+    {
+      "epoch": 0.9347679892400808,
+      "grad_norm": 1.178755283355713,
+      "learning_rate": 4.4769650914272146e-05,
+      "loss": 0.3368,
+      "step": 1390
+    },
+    {
+      "epoch": 0.9414929388029589,
+      "grad_norm": 1.0818071365356445,
+      "learning_rate": 4.4710282593208264e-05,
+      "loss": 0.3453,
+      "step": 1400
+    },
+    {
+      "epoch": 0.9482178883658372,
+      "grad_norm": 1.2789335250854492,
+      "learning_rate": 4.465091427214438e-05,
+      "loss": 0.3647,
+      "step": 1410
+    },
+    {
+      "epoch": 0.9549428379287155,
+      "grad_norm": 1.2054712772369385,
+      "learning_rate": 4.459154595108051e-05,
+      "loss": 0.361,
+      "step": 1420
+    },
+    {
+      "epoch": 0.9616677874915938,
+      "grad_norm": 1.1972471475601196,
+      "learning_rate": 4.4532177630016625e-05,
+      "loss": 0.3843,
+      "step": 1430
+    },
+    {
+      "epoch": 0.968392737054472,
+      "grad_norm": 1.224511981010437,
+      "learning_rate": 4.4472809308952743e-05,
+      "loss": 0.3281,
+      "step": 1440
+    },
+    {
+      "epoch": 0.9751176866173503,
+      "grad_norm": 1.6475602388381958,
+      "learning_rate": 4.441344098788887e-05,
+      "loss": 0.3414,
+      "step": 1450
+    },
+    {
+      "epoch": 0.9818426361802286,
+      "grad_norm": 1.2841564416885376,
+      "learning_rate": 4.4354072666824986e-05,
+      "loss": 0.326,
+      "step": 1460
+    },
+    {
+      "epoch": 0.9885675857431069,
+      "grad_norm": 1.0296547412872314,
+      "learning_rate": 4.4294704345761104e-05,
+      "loss": 0.3432,
+      "step": 1470
+    },
+    {
+      "epoch": 0.9952925353059852,
+      "grad_norm": 1.5025538206100464,
+      "learning_rate": 4.423533602469722e-05,
+      "loss": 0.3704,
+      "step": 1480
+    },
+    {
+      "epoch": 1.0020174848688634,
+      "grad_norm": 0.8826779723167419,
+      "learning_rate": 4.417596770363334e-05,
+      "loss": 0.3386,
+      "step": 1490
+    },
+    {
+      "epoch": 1.0087424344317417,
+      "grad_norm": 1.1697226762771606,
+      "learning_rate": 4.4116599382569465e-05,
+      "loss": 0.3141,
+      "step": 1500
+    },
+    {
+      "epoch": 1.0087424344317417,
+      "eval_loss": 0.26416271924972534,
+      "eval_runtime": 14.3375,
+      "eval_samples_per_second": 184.411,
+      "eval_steps_per_second": 23.086,
+      "step": 1500
+    },
+    {
+      "epoch": 1.01546738399462,
+      "grad_norm": 1.0717393159866333,
+      "learning_rate": 4.4057231061505583e-05,
+      "loss": 0.3196,
+      "step": 1510
+    },
+    {
+      "epoch": 1.0221923335574983,
+      "grad_norm": 1.0421812534332275,
+      "learning_rate": 4.39978627404417e-05,
+      "loss": 0.3434,
+      "step": 1520
+    },
+    {
+      "epoch": 1.0289172831203766,
+      "grad_norm": 1.326445460319519,
+      "learning_rate": 4.3938494419377826e-05,
+      "loss": 0.3639,
+      "step": 1530
+    },
+    {
+      "epoch": 1.0356422326832548,
+      "grad_norm": 0.7816503047943115,
+      "learning_rate": 4.3879126098313944e-05,
+      "loss": 0.3278,
+      "step": 1540
+    },
+    {
+      "epoch": 1.0423671822461331,
+      "grad_norm": 1.133551836013794,
+      "learning_rate": 4.3819757777250056e-05,
+      "loss": 0.3527,
+      "step": 1550
+    },
+    {
+      "epoch": 1.0490921318090114,
+      "grad_norm": 1.0650932788848877,
+      "learning_rate": 4.376038945618618e-05,
+      "loss": 0.3401,
+      "step": 1560
+    },
+    {
+      "epoch": 1.0558170813718897,
+      "grad_norm": 1.3639334440231323,
+      "learning_rate": 4.37010211351223e-05,
+      "loss": 0.372,
+      "step": 1570
+    },
+    {
+      "epoch": 1.062542030934768,
+      "grad_norm": 1.162127137184143,
+      "learning_rate": 4.3641652814058423e-05,
+      "loss": 0.3862,
+      "step": 1580
+    },
+    {
+      "epoch": 1.0692669804976462,
+      "grad_norm": 1.0656379461288452,
+      "learning_rate": 4.358228449299454e-05,
+      "loss": 0.302,
+      "step": 1590
+    },
+    {
+      "epoch": 1.0759919300605245,
+      "grad_norm": 1.2670329809188843,
+      "learning_rate": 4.352291617193066e-05,
+      "loss": 0.3013,
+      "step": 1600
+    },
+    {
+      "epoch": 1.0827168796234028,
+      "grad_norm": 1.1603080034255981,
+      "learning_rate": 4.3463547850866784e-05,
+      "loss": 0.2655,
+      "step": 1610
+    },
+    {
+      "epoch": 1.089441829186281,
+      "grad_norm": 1.0313270092010498,
+      "learning_rate": 4.34041795298029e-05,
+      "loss": 0.3272,
+      "step": 1620
+    },
+    {
+      "epoch": 1.0961667787491594,
+      "grad_norm": 0.8739041090011597,
+      "learning_rate": 4.3344811208739014e-05,
+      "loss": 0.3031,
+      "step": 1630
+    },
+    {
+      "epoch": 1.1028917283120376,
+      "grad_norm": 1.0449531078338623,
+      "learning_rate": 4.328544288767514e-05,
+      "loss": 0.2969,
+      "step": 1640
+    },
+    {
+      "epoch": 1.109616677874916,
+      "grad_norm": 1.065719485282898,
+      "learning_rate": 4.322607456661126e-05,
+      "loss": 0.3422,
+      "step": 1650
+    },
+    {
+      "epoch": 1.1163416274377942,
+      "grad_norm": 1.2147116661071777,
+      "learning_rate": 4.3166706245547375e-05,
+      "loss": 0.309,
+      "step": 1660
+    },
+    {
+      "epoch": 1.1230665770006725,
+      "grad_norm": 0.8757501840591431,
+      "learning_rate": 4.31073379244835e-05,
+      "loss": 0.2781,
+      "step": 1670
+    },
+    {
+      "epoch": 1.1297915265635508,
+      "grad_norm": 1.0971019268035889,
+      "learning_rate": 4.304796960341962e-05,
+      "loss": 0.2931,
+      "step": 1680
+    },
+    {
+      "epoch": 1.136516476126429,
+      "grad_norm": 1.1437770128250122,
+      "learning_rate": 4.298860128235574e-05,
+      "loss": 0.2595,
+      "step": 1690
+    },
+    {
+      "epoch": 1.1432414256893073,
+      "grad_norm": 1.025486707687378,
+      "learning_rate": 4.2929232961291854e-05,
+      "loss": 0.2598,
+      "step": 1700
+    },
+    {
+      "epoch": 1.1499663752521856,
+      "grad_norm": 1.2916457653045654,
+      "learning_rate": 4.286986464022797e-05,
+      "loss": 0.2544,
+      "step": 1710
+    },
+    {
+      "epoch": 1.1566913248150639,
+      "grad_norm": 1.05109441280365,
+      "learning_rate": 4.28104963191641e-05,
+      "loss": 0.3443,
+      "step": 1720
+    },
+    {
+      "epoch": 1.1634162743779422,
+      "grad_norm": 0.974449634552002,
+      "learning_rate": 4.2751127998100215e-05,
+      "loss": 0.2725,
+      "step": 1730
+    },
+    {
+      "epoch": 1.1701412239408204,
+      "grad_norm": 0.9677799940109253,
+      "learning_rate": 4.269175967703633e-05,
+      "loss": 0.3194,
+      "step": 1740
+    },
+    {
+      "epoch": 1.1768661735036987,
+      "grad_norm": 1.2493693828582764,
+      "learning_rate": 4.263239135597246e-05,
+      "loss": 0.3004,
+      "step": 1750
+    },
+    {
+      "epoch": 1.183591123066577,
+      "grad_norm": 1.0732406377792358,
+      "learning_rate": 4.2573023034908576e-05,
+      "loss": 0.2952,
+      "step": 1760
+    },
+    {
+      "epoch": 1.1903160726294553,
+      "grad_norm": 1.2802927494049072,
+      "learning_rate": 4.2513654713844694e-05,
+      "loss": 0.2894,
+      "step": 1770
+    },
+    {
+      "epoch": 1.1970410221923335,
+      "grad_norm": 0.8654187917709351,
+      "learning_rate": 4.245428639278081e-05,
+      "loss": 0.2714,
+      "step": 1780
+    },
+    {
+      "epoch": 1.2037659717552118,
+      "grad_norm": 1.3110740184783936,
+      "learning_rate": 4.239491807171693e-05,
+      "loss": 0.2993,
+      "step": 1790
+    },
+    {
+      "epoch": 1.21049092131809,
+      "grad_norm": 1.2888758182525635,
+      "learning_rate": 4.2335549750653055e-05,
+      "loss": 0.3015,
+      "step": 1800
+    },
+    {
+      "epoch": 1.2172158708809684,
+      "grad_norm": 1.1518973112106323,
+      "learning_rate": 4.227618142958917e-05,
+      "loss": 0.2795,
+      "step": 1810
+    },
+    {
+      "epoch": 1.2239408204438467,
+      "grad_norm": 1.0696250200271606,
+      "learning_rate": 4.221681310852529e-05,
+      "loss": 0.2756,
+      "step": 1820
+    },
+    {
+      "epoch": 1.230665770006725,
+      "grad_norm": 0.879612147808075,
+      "learning_rate": 4.2157444787461416e-05,
+      "loss": 0.3179,
+      "step": 1830
+    },
+    {
+      "epoch": 1.2373907195696032,
+      "grad_norm": 0.8837944269180298,
+      "learning_rate": 4.2098076466397534e-05,
+      "loss": 0.2439,
+      "step": 1840
+    },
+    {
+      "epoch": 1.2441156691324815,
+      "grad_norm": 1.0999325513839722,
+      "learning_rate": 4.203870814533365e-05,
+      "loss": 0.3235,
+      "step": 1850
+    },
+    {
+      "epoch": 1.2508406186953598,
+      "grad_norm": 1.1159007549285889,
+      "learning_rate": 4.197933982426977e-05,
+      "loss": 0.2855,
+      "step": 1860
+    },
+    {
+      "epoch": 1.257565568258238,
+      "grad_norm": 0.933502733707428,
+      "learning_rate": 4.191997150320589e-05,
+      "loss": 0.277,
+      "step": 1870
+    },
+    {
+      "epoch": 1.2642905178211163,
+      "grad_norm": 1.038533091545105,
+      "learning_rate": 4.186060318214201e-05,
+      "loss": 0.3594,
+      "step": 1880
+    },
+    {
+      "epoch": 1.2710154673839946,
+      "grad_norm": 1.0289478302001953,
+      "learning_rate": 4.180123486107813e-05,
+      "loss": 0.271,
+      "step": 1890
+    },
+    {
+      "epoch": 1.277740416946873,
+      "grad_norm": 0.97165846824646,
+      "learning_rate": 4.174186654001425e-05,
+      "loss": 0.2784,
+      "step": 1900
+    },
+    {
+      "epoch": 1.2844653665097512,
+      "grad_norm": 1.3086190223693848,
+      "learning_rate": 4.1682498218950374e-05,
+      "loss": 0.2883,
+      "step": 1910
+    },
+    {
+      "epoch": 1.2911903160726295,
+      "grad_norm": 0.9382519721984863,
+      "learning_rate": 4.162312989788649e-05,
+      "loss": 0.2936,
+      "step": 1920
+    },
+    {
+      "epoch": 1.2979152656355077,
+      "grad_norm": 0.9957454800605774,
+      "learning_rate": 4.156376157682261e-05,
+      "loss": 0.3137,
+      "step": 1930
+    },
+    {
+      "epoch": 1.304640215198386,
+      "grad_norm": 0.8603612184524536,
+      "learning_rate": 4.150439325575873e-05,
+      "loss": 0.2735,
+      "step": 1940
+    },
+    {
+      "epoch": 1.3113651647612643,
+      "grad_norm": 0.9475612044334412,
+      "learning_rate": 4.1445024934694846e-05,
+      "loss": 0.2763,
+      "step": 1950
+    },
+    {
+      "epoch": 1.3180901143241426,
+      "grad_norm": 1.2513048648834229,
+      "learning_rate": 4.138565661363097e-05,
+      "loss": 0.254,
+      "step": 1960
+    },
+    {
+      "epoch": 1.3248150638870209,
+      "grad_norm": 1.067569375038147,
+      "learning_rate": 4.132628829256709e-05,
+      "loss": 0.235,
+      "step": 1970
+    },
+    {
+      "epoch": 1.3315400134498991,
+      "grad_norm": 1.2715383768081665,
+      "learning_rate": 4.126691997150321e-05,
+      "loss": 0.2932,
+      "step": 1980
+    },
+    {
+      "epoch": 1.3382649630127774,
+      "grad_norm": 0.8481950163841248,
+      "learning_rate": 4.120755165043933e-05,
+      "loss": 0.2962,
+      "step": 1990
+    },
+    {
+      "epoch": 1.3449899125756557,
+      "grad_norm": 0.9146913290023804,
+      "learning_rate": 4.114818332937544e-05,
+      "loss": 0.2988,
+      "step": 2000
+    },
+    {
+      "epoch": 1.3449899125756557,
+      "eval_loss": 0.22625228762626648,
+      "eval_runtime": 14.3844,
+      "eval_samples_per_second": 183.81,
+      "eval_steps_per_second": 23.011,
+      "step": 2000
+    },
+    {
+      "epoch": 1.351714862138534,
+      "grad_norm": 1.3533391952514648,
+      "learning_rate": 4.108881500831157e-05,
+      "loss": 0.2837,
+      "step": 2010
+    },
+    {
+      "epoch": 1.3584398117014123,
+      "grad_norm": 1.1099708080291748,
+      "learning_rate": 4.1029446687247686e-05,
+      "loss": 0.2536,
+      "step": 2020
+    },
+    {
+      "epoch": 1.3651647612642905,
+      "grad_norm": 0.7431579828262329,
+      "learning_rate": 4.0970078366183804e-05,
+      "loss": 0.2192,
+      "step": 2030
+    },
+    {
+      "epoch": 1.3718897108271688,
+      "grad_norm": 0.8818496465682983,
+      "learning_rate": 4.091071004511993e-05,
+      "loss": 0.2197,
+      "step": 2040
+    },
+    {
+      "epoch": 1.378614660390047,
+      "grad_norm": 1.0587776899337769,
+      "learning_rate": 4.085134172405605e-05,
+      "loss": 0.2696,
+      "step": 2050
+    },
+    {
+      "epoch": 1.3853396099529254,
+      "grad_norm": 1.0173654556274414,
+      "learning_rate": 4.0791973402992165e-05,
+      "loss": 0.2839,
+      "step": 2060
+    },
+    {
+      "epoch": 1.3920645595158037,
+      "grad_norm": 1.1157002449035645,
+      "learning_rate": 4.073260508192828e-05,
+      "loss": 0.2893,
+      "step": 2070
+    },
+    {
+      "epoch": 1.398789509078682,
+      "grad_norm": 0.9677891135215759,
+      "learning_rate": 4.06732367608644e-05,
+      "loss": 0.2471,
+      "step": 2080
+    },
+    {
+      "epoch": 1.4055144586415602,
+      "grad_norm": 0.6478651762008667,
+      "learning_rate": 4.0613868439800526e-05,
+      "loss": 0.222,
+      "step": 2090
+    },
+    {
+      "epoch": 1.4122394082044385,
+      "grad_norm": 0.9797048568725586,
+      "learning_rate": 4.0554500118736644e-05,
+      "loss": 0.2671,
+      "step": 2100
+    },
+    {
+      "epoch": 1.4189643577673168,
+      "grad_norm": 0.9667022824287415,
+      "learning_rate": 4.049513179767276e-05,
+      "loss": 0.2672,
+      "step": 2110
+    },
+    {
+      "epoch": 1.425689307330195,
+      "grad_norm": 0.7505651712417603,
+      "learning_rate": 4.043576347660889e-05,
+      "loss": 0.2854,
+      "step": 2120
+    },
+    {
+      "epoch": 1.4324142568930733,
+      "grad_norm": 0.9649038910865784,
+      "learning_rate": 4.0376395155545005e-05,
+      "loss": 0.2819,
+      "step": 2130
+    },
+    {
+      "epoch": 1.4391392064559516,
+      "grad_norm": 1.242560863494873,
+      "learning_rate": 4.031702683448112e-05,
+      "loss": 0.2844,
+      "step": 2140
+    },
+    {
+      "epoch": 1.44586415601883,
+      "grad_norm": 1.0524150133132935,
+      "learning_rate": 4.025765851341724e-05,
+      "loss": 0.2925,
+      "step": 2150
+    },
+    {
+      "epoch": 1.4525891055817082,
+      "grad_norm": 1.135675072669983,
+      "learning_rate": 4.019829019235336e-05,
+      "loss": 0.2961,
+      "step": 2160
+    },
+    {
+      "epoch": 1.4593140551445865,
+      "grad_norm": 1.071516752243042,
+      "learning_rate": 4.0138921871289484e-05,
+      "loss": 0.2502,
+      "step": 2170
+    },
+    {
+      "epoch": 1.4660390047074647,
+      "grad_norm": 0.862398087978363,
+      "learning_rate": 4.00795535502256e-05,
+      "loss": 0.2646,
+      "step": 2180
+    },
+    {
+      "epoch": 1.472763954270343,
+      "grad_norm": 1.5514510869979858,
+      "learning_rate": 4.002018522916172e-05,
+      "loss": 0.2775,
+      "step": 2190
+    },
+    {
+      "epoch": 1.4794889038332213,
+      "grad_norm": 1.1586170196533203,
+      "learning_rate": 3.9960816908097845e-05,
+      "loss": 0.2749,
+      "step": 2200
+    },
+    {
+      "epoch": 1.4862138533960996,
+      "grad_norm": 0.7154861688613892,
+      "learning_rate": 3.990144858703396e-05,
+      "loss": 0.2357,
+      "step": 2210
+    },
+    {
+      "epoch": 1.4929388029589779,
+      "grad_norm": 0.9317541122436523,
+      "learning_rate": 3.9842080265970075e-05,
+      "loss": 0.2601,
+      "step": 2220
+    },
+    {
+      "epoch": 1.4996637525218561,
+      "grad_norm": 1.2377835512161255,
+      "learning_rate": 3.97827119449062e-05,
+      "loss": 0.2765,
+      "step": 2230
+    },
+    {
+      "epoch": 1.5063887020847344,
+      "grad_norm": 0.704967200756073,
+      "learning_rate": 3.972334362384232e-05,
+      "loss": 0.2587,
+      "step": 2240
+    },
+    {
+      "epoch": 1.5131136516476127,
+      "grad_norm": 1.0047144889831543,
+      "learning_rate": 3.9663975302778436e-05,
+      "loss": 0.2535,
+      "step": 2250
+    },
+    {
+      "epoch": 1.519838601210491,
+      "grad_norm": 1.10822331905365,
+      "learning_rate": 3.960460698171456e-05,
+      "loss": 0.2447,
+      "step": 2260
+    },
+    {
+      "epoch": 1.5265635507733692,
+      "grad_norm": 1.041707992553711,
+      "learning_rate": 3.954523866065068e-05,
+      "loss": 0.2761,
+      "step": 2270
+    },
+    {
+      "epoch": 1.5332885003362475,
+      "grad_norm": 0.8031961917877197,
+      "learning_rate": 3.94858703395868e-05,
+      "loss": 0.2171,
+      "step": 2280
+    },
+    {
+      "epoch": 1.5400134498991258,
+      "grad_norm": 0.9082476496696472,
+      "learning_rate": 3.942650201852292e-05,
+      "loss": 0.2653,
+      "step": 2290
+    },
+    {
+      "epoch": 1.546738399462004,
+      "grad_norm": 0.9983156323432922,
+      "learning_rate": 3.936713369745903e-05,
+      "loss": 0.2718,
+      "step": 2300
+    },
+    {
+      "epoch": 1.5534633490248824,
+      "grad_norm": 0.7857641577720642,
+      "learning_rate": 3.930776537639516e-05,
+      "loss": 0.2479,
+      "step": 2310
+    },
+    {
+      "epoch": 1.5601882985877606,
+      "grad_norm": 0.7820498943328857,
+      "learning_rate": 3.9248397055331276e-05,
+      "loss": 0.2826,
+      "step": 2320
+    },
+    {
+      "epoch": 1.5669132481506387,
+      "grad_norm": 0.9067456722259521,
+      "learning_rate": 3.9189028734267394e-05,
+      "loss": 0.3029,
+      "step": 2330
+    },
+    {
+      "epoch": 1.573638197713517,
+      "grad_norm": 0.7091119289398193,
+      "learning_rate": 3.912966041320352e-05,
+      "loss": 0.222,
+      "step": 2340
+    },
+    {
+      "epoch": 1.5803631472763953,
+      "grad_norm": 1.0687780380249023,
+      "learning_rate": 3.9070292092139637e-05,
+      "loss": 0.2741,
+      "step": 2350
+    },
+    {
+      "epoch": 1.5870880968392735,
+      "grad_norm": 0.7274528741836548,
+      "learning_rate": 3.9010923771075755e-05,
+      "loss": 0.275,
+      "step": 2360
+    },
+    {
+      "epoch": 1.5938130464021518,
+      "grad_norm": 0.7796964645385742,
+      "learning_rate": 3.895155545001187e-05,
+      "loss": 0.2754,
+      "step": 2370
+    },
+    {
+      "epoch": 1.60053799596503,
+      "grad_norm": 0.9450284838676453,
+      "learning_rate": 3.889218712894799e-05,
+      "loss": 0.2318,
+      "step": 2380
+    },
+    {
+      "epoch": 1.6072629455279084,
+      "grad_norm": 0.9865807890892029,
+      "learning_rate": 3.8832818807884116e-05,
+      "loss": 0.2851,
+      "step": 2390
+    },
+    {
+      "epoch": 1.6139878950907867,
+      "grad_norm": 0.9114990830421448,
+      "learning_rate": 3.8773450486820234e-05,
+      "loss": 0.2232,
+      "step": 2400
+    },
+    {
+      "epoch": 1.620712844653665,
+      "grad_norm": 1.1912641525268555,
+      "learning_rate": 3.871408216575635e-05,
+      "loss": 0.2125,
+      "step": 2410
+    },
+    {
+      "epoch": 1.6274377942165432,
+      "grad_norm": 0.9785919189453125,
+      "learning_rate": 3.8654713844692477e-05,
+      "loss": 0.2779,
+      "step": 2420
+    },
+    {
+      "epoch": 1.6341627437794215,
+      "grad_norm": 1.6103061437606812,
+      "learning_rate": 3.8595345523628595e-05,
+      "loss": 0.242,
+      "step": 2430
+    },
+    {
+      "epoch": 1.6408876933422998,
+      "grad_norm": 1.161123275756836,
+      "learning_rate": 3.853597720256471e-05,
+      "loss": 0.2297,
+      "step": 2440
+    },
+    {
+      "epoch": 1.647612642905178,
+      "grad_norm": 0.8608954548835754,
+      "learning_rate": 3.847660888150083e-05,
+      "loss": 0.2907,
+      "step": 2450
+    },
+    {
+      "epoch": 1.6543375924680563,
+      "grad_norm": 0.7341004610061646,
+      "learning_rate": 3.841724056043695e-05,
+      "loss": 0.2325,
+      "step": 2460
+    },
+    {
+      "epoch": 1.6610625420309346,
+      "grad_norm": 0.976824939250946,
+      "learning_rate": 3.8357872239373074e-05,
+      "loss": 0.2631,
+      "step": 2470
+    },
+    {
+      "epoch": 1.667787491593813,
+      "grad_norm": 0.9651874303817749,
+      "learning_rate": 3.829850391830919e-05,
+      "loss": 0.2731,
+      "step": 2480
+    },
+    {
+      "epoch": 1.6745124411566912,
+      "grad_norm": 0.856626570224762,
+      "learning_rate": 3.823913559724531e-05,
+      "loss": 0.221,
+      "step": 2490
+    },
+    {
+      "epoch": 1.6812373907195695,
+      "grad_norm": 0.990326464176178,
+      "learning_rate": 3.8179767276181435e-05,
+      "loss": 0.3172,
+      "step": 2500
+    },
+    {
+      "epoch": 1.6812373907195695,
+      "eval_loss": 0.2056342214345932,
+      "eval_runtime": 14.2733,
+      "eval_samples_per_second": 185.241,
+      "eval_steps_per_second": 23.19,
+      "step": 2500
+    },
+    {
+      "epoch": 1.6879623402824477,
+      "grad_norm": 0.8716689348220825,
+      "learning_rate": 3.812039895511755e-05,
+      "loss": 0.242,
+      "step": 2510
+    },
+    {
+      "epoch": 1.694687289845326,
+      "grad_norm": 1.0509111881256104,
+      "learning_rate": 3.806103063405367e-05,
+      "loss": 0.2674,
+      "step": 2520
+    },
+    {
+      "epoch": 1.7014122394082043,
+      "grad_norm": 0.7472453713417053,
+      "learning_rate": 3.800166231298979e-05,
+      "loss": 0.2856,
+      "step": 2530
+    },
+    {
+      "epoch": 1.7081371889710826,
+      "grad_norm": 0.9000864624977112,
+      "learning_rate": 3.794229399192591e-05,
+      "loss": 0.3039,
+      "step": 2540
+    },
+    {
+      "epoch": 1.7148621385339609,
+      "grad_norm": 0.8913015127182007,
+      "learning_rate": 3.788292567086203e-05,
+      "loss": 0.3029,
+      "step": 2550
+    },
+    {
+      "epoch": 1.7215870880968391,
+      "grad_norm": 0.8225172758102417,
+      "learning_rate": 3.782355734979815e-05,
+      "loss": 0.2137,
+      "step": 2560
+    },
+    {
+      "epoch": 1.7283120376597174,
+      "grad_norm": 0.9405450224876404,
+      "learning_rate": 3.776418902873427e-05,
+      "loss": 0.2064,
+      "step": 2570
+    },
+    {
+      "epoch": 1.7350369872225957,
+      "grad_norm": 0.9194093942642212,
+      "learning_rate": 3.770482070767039e-05,
+      "loss": 0.2301,
+      "step": 2580
+    },
+    {
+      "epoch": 1.741761936785474,
+      "grad_norm": 0.8586764931678772,
+      "learning_rate": 3.764545238660651e-05,
+      "loss": 0.2553,
+      "step": 2590
+    },
+    {
+      "epoch": 1.7484868863483523,
+      "grad_norm": 1.3633416891098022,
+      "learning_rate": 3.758608406554263e-05,
+      "loss": 0.2307,
+      "step": 2600
+    },
+    {
+      "epoch": 1.7552118359112305,
+      "grad_norm": 0.7213522791862488,
+      "learning_rate": 3.752671574447875e-05,
+      "loss": 0.251,
+      "step": 2610
+    },
+    {
+      "epoch": 1.7619367854741088,
+      "grad_norm": 1.1806386709213257,
+      "learning_rate": 3.7467347423414865e-05,
+      "loss": 0.2753,
+      "step": 2620
+    },
+    {
+      "epoch": 1.768661735036987,
+      "grad_norm": 1.203027367591858,
+      "learning_rate": 3.740797910235099e-05,
+      "loss": 0.2878,
+      "step": 2630
+    },
+    {
+      "epoch": 1.7753866845998654,
+      "grad_norm": 0.9682397842407227,
+      "learning_rate": 3.734861078128711e-05,
+      "loss": 0.2722,
+      "step": 2640
+    },
+    {
+      "epoch": 1.7821116341627437,
+      "grad_norm": 1.2024784088134766,
+      "learning_rate": 3.7289242460223226e-05,
+      "loss": 0.1792,
+      "step": 2650
+    },
+    {
+      "epoch": 1.788836583725622,
+      "grad_norm": 0.9232836365699768,
+      "learning_rate": 3.722987413915935e-05,
+      "loss": 0.2255,
+      "step": 2660
+    },
+    {
+      "epoch": 1.7955615332885002,
+      "grad_norm": 0.8811613917350769,
+      "learning_rate": 3.717050581809546e-05,
+      "loss": 0.2608,
+      "step": 2670
+    },
+    {
+      "epoch": 1.8022864828513785,
+      "grad_norm": 0.9594545364379883,
+      "learning_rate": 3.711113749703159e-05,
+      "loss": 0.2906,
+      "step": 2680
+    },
+    {
+      "epoch": 1.8090114324142568,
+      "grad_norm": 0.8096931576728821,
+      "learning_rate": 3.7051769175967705e-05,
+      "loss": 0.2148,
+      "step": 2690
+    },
+    {
+      "epoch": 1.815736381977135,
+      "grad_norm": 1.0215779542922974,
+      "learning_rate": 3.699240085490382e-05,
+      "loss": 0.2833,
+      "step": 2700
+    },
+    {
+      "epoch": 1.8224613315400133,
+      "grad_norm": 0.8041989207267761,
+      "learning_rate": 3.693303253383995e-05,
+      "loss": 0.1985,
+      "step": 2710
+    },
+    {
+      "epoch": 1.8291862811028916,
+      "grad_norm": 1.1806946992874146,
+      "learning_rate": 3.6873664212776066e-05,
+      "loss": 0.2714,
+      "step": 2720
+    },
+    {
+      "epoch": 1.8359112306657699,
+      "grad_norm": 0.9091224670410156,
+      "learning_rate": 3.6814295891712184e-05,
+      "loss": 0.2034,
+      "step": 2730
+    },
+    {
+      "epoch": 1.8426361802286482,
+      "grad_norm": 0.7493769526481628,
+      "learning_rate": 3.675492757064831e-05,
+      "loss": 0.2219,
+      "step": 2740
+    },
+    {
+      "epoch": 1.8493611297915264,
+      "grad_norm": 0.8476701974868774,
+      "learning_rate": 3.669555924958442e-05,
+      "loss": 0.2407,
+      "step": 2750
+    },
+    {
+      "epoch": 1.8560860793544047,
+      "grad_norm": 0.8047271370887756,
+      "learning_rate": 3.6636190928520545e-05,
+      "loss": 0.2388,
+      "step": 2760
+    },
+    {
+      "epoch": 1.862811028917283,
+      "grad_norm": 0.8351836800575256,
+      "learning_rate": 3.657682260745666e-05,
+      "loss": 0.2393,
+      "step": 2770
+    },
+    {
+      "epoch": 1.8695359784801613,
+      "grad_norm": 0.8370159268379211,
+      "learning_rate": 3.651745428639278e-05,
+      "loss": 0.2252,
+      "step": 2780
+    },
+    {
+      "epoch": 1.8762609280430396,
+      "grad_norm": 1.0017287731170654,
+      "learning_rate": 3.6458085965328906e-05,
+      "loss": 0.2861,
+      "step": 2790
+    },
+    {
+      "epoch": 1.8829858776059178,
+      "grad_norm": 0.6298861503601074,
+      "learning_rate": 3.6398717644265024e-05,
+      "loss": 0.2139,
+      "step": 2800
+    },
+    {
+      "epoch": 1.8897108271687961,
+      "grad_norm": 0.7649684548377991,
+      "learning_rate": 3.633934932320114e-05,
+      "loss": 0.2377,
+      "step": 2810
+    },
+    {
+      "epoch": 1.8964357767316744,
+      "grad_norm": 0.9333203434944153,
+      "learning_rate": 3.627998100213726e-05,
+      "loss": 0.2546,
+      "step": 2820
+    },
+    {
+      "epoch": 1.9031607262945527,
+      "grad_norm": 0.8119333386421204,
+      "learning_rate": 3.622061268107338e-05,
+      "loss": 0.2792,
+      "step": 2830
+    },
+    {
+      "epoch": 1.909885675857431,
+      "grad_norm": 0.8402873873710632,
+      "learning_rate": 3.6161244360009496e-05,
+      "loss": 0.2776,
+      "step": 2840
+    },
+    {
+      "epoch": 1.9166106254203092,
+      "grad_norm": 0.6565370559692383,
+      "learning_rate": 3.610187603894562e-05,
+      "loss": 0.2081,
+      "step": 2850
+    },
+    {
+      "epoch": 1.9233355749831875,
+      "grad_norm": 0.818225085735321,
+      "learning_rate": 3.604250771788174e-05,
+      "loss": 0.2271,
+      "step": 2860
+    },
+    {
+      "epoch": 1.9300605245460658,
+      "grad_norm": 0.6470263004302979,
+      "learning_rate": 3.5983139396817864e-05,
+      "loss": 0.2026,
+      "step": 2870
+    },
+    {
+      "epoch": 1.936785474108944,
+      "grad_norm": 0.8524275422096252,
+      "learning_rate": 3.592377107575398e-05,
+      "loss": 0.2628,
+      "step": 2880
+    },
+    {
+      "epoch": 1.9435104236718224,
+      "grad_norm": 0.7875599265098572,
+      "learning_rate": 3.58644027546901e-05,
+      "loss": 0.281,
+      "step": 2890
+    },
+    {
+      "epoch": 1.9502353732347006,
+      "grad_norm": 0.9621008038520813,
+      "learning_rate": 3.580503443362622e-05,
+      "loss": 0.2349,
+      "step": 2900
+    },
+    {
+      "epoch": 1.956960322797579,
+      "grad_norm": 0.8459897041320801,
+      "learning_rate": 3.5745666112562336e-05,
+      "loss": 0.2372,
+      "step": 2910
+    },
+    {
+      "epoch": 1.9636852723604572,
+      "grad_norm": 0.9341444969177246,
+      "learning_rate": 3.5686297791498454e-05,
+      "loss": 0.2533,
+      "step": 2920
+    },
+    {
+      "epoch": 1.9704102219233355,
+      "grad_norm": 0.6931284666061401,
+      "learning_rate": 3.562692947043458e-05,
+      "loss": 0.1903,
+      "step": 2930
+    },
+    {
+      "epoch": 1.9771351714862138,
+      "grad_norm": 0.8395922780036926,
+      "learning_rate": 3.55675611493707e-05,
+      "loss": 0.2997,
+      "step": 2940
+    },
+    {
+      "epoch": 1.983860121049092,
+      "grad_norm": 1.1813199520111084,
+      "learning_rate": 3.5508192828306815e-05,
+      "loss": 0.2437,
+      "step": 2950
+    },
+    {
+      "epoch": 1.9905850706119703,
+      "grad_norm": 1.0604130029678345,
+      "learning_rate": 3.544882450724294e-05,
+      "loss": 0.2402,
+      "step": 2960
+    },
+    {
+      "epoch": 1.9973100201748486,
+      "grad_norm": 0.7427442073822021,
+      "learning_rate": 3.538945618617905e-05,
+      "loss": 0.2111,
+      "step": 2970
+    },
+    {
+      "epoch": 2.004034969737727,
+      "grad_norm": 0.7738786339759827,
+      "learning_rate": 3.5330087865115176e-05,
+      "loss": 0.2503,
+      "step": 2980
+    },
+    {
+      "epoch": 2.010759919300605,
+      "grad_norm": 1.0489563941955566,
+      "learning_rate": 3.5270719544051294e-05,
+      "loss": 0.2051,
+      "step": 2990
+    },
+    {
+      "epoch": 2.0174848688634834,
+      "grad_norm": 1.0015239715576172,
+      "learning_rate": 3.521135122298741e-05,
+      "loss": 0.2148,
+      "step": 3000
+    },
+    {
+      "epoch": 2.0174848688634834,
+      "eval_loss": 0.19253146648406982,
+      "eval_runtime": 14.4364,
+      "eval_samples_per_second": 183.148,
+      "eval_steps_per_second": 22.928,
+      "step": 3000
+    },
+    {
+      "epoch": 2.0242098184263617,
+      "grad_norm": 0.7455950975418091,
+      "learning_rate": 3.515198290192354e-05,
+      "loss": 0.2406,
+      "step": 3010
+    },
+    {
+      "epoch": 2.03093476798924,
+      "grad_norm": 1.0238157510757446,
+      "learning_rate": 3.5092614580859655e-05,
+      "loss": 0.2556,
+      "step": 3020
+    },
+    {
+      "epoch": 2.0376597175521183,
+      "grad_norm": 0.7209094166755676,
+      "learning_rate": 3.5033246259795774e-05,
+      "loss": 0.2255,
+      "step": 3030
+    },
+    {
+      "epoch": 2.0443846671149966,
+      "grad_norm": 0.7461678981781006,
+      "learning_rate": 3.49738779387319e-05,
+      "loss": 0.1926,
+      "step": 3040
+    },
+    {
+      "epoch": 2.051109616677875,
+      "grad_norm": 0.7917340993881226,
+      "learning_rate": 3.491450961766801e-05,
+      "loss": 0.28,
+      "step": 3050
+    },
+    {
+      "epoch": 2.057834566240753,
+      "grad_norm": 0.7438961863517761,
+      "learning_rate": 3.4855141296604135e-05,
+      "loss": 0.1955,
+      "step": 3060
+    },
+    {
+      "epoch": 2.0645595158036314,
+      "grad_norm": 0.8378620147705078,
+      "learning_rate": 3.479577297554025e-05,
+      "loss": 0.1863,
+      "step": 3070
+    },
+    {
+      "epoch": 2.0712844653665097,
+      "grad_norm": 0.8809778690338135,
+      "learning_rate": 3.473640465447637e-05,
+      "loss": 0.2263,
+      "step": 3080
+    },
+    {
+      "epoch": 2.078009414929388,
+      "grad_norm": 0.6806073784828186,
+      "learning_rate": 3.4677036333412495e-05,
+      "loss": 0.1948,
+      "step": 3090
+    },
+    {
+      "epoch": 2.0847343644922662,
+      "grad_norm": 1.061790108680725,
+      "learning_rate": 3.4617668012348614e-05,
+      "loss": 0.2436,
+      "step": 3100
+    },
+    {
+      "epoch": 2.0914593140551445,
+      "grad_norm": 0.9499590992927551,
+      "learning_rate": 3.455829969128473e-05,
+      "loss": 0.2388,
+      "step": 3110
+    },
+    {
+      "epoch": 2.098184263618023,
+      "grad_norm": 0.8609486222267151,
+      "learning_rate": 3.449893137022085e-05,
+      "loss": 0.2675,
+      "step": 3120
+    },
+    {
+      "epoch": 2.104909213180901,
+      "grad_norm": 0.7025455236434937,
+      "learning_rate": 3.443956304915697e-05,
+      "loss": 0.2014,
+      "step": 3130
+    },
+    {
+      "epoch": 2.1116341627437794,
+      "grad_norm": 0.783028244972229,
+      "learning_rate": 3.438019472809309e-05,
+      "loss": 0.1954,
+      "step": 3140
+    },
+    {
+      "epoch": 2.1183591123066576,
+      "grad_norm": 0.767192542552948,
+      "learning_rate": 3.432082640702921e-05,
+      "loss": 0.2294,
+      "step": 3150
+    },
+    {
+      "epoch": 2.125084061869536,
+      "grad_norm": 1.0050126314163208,
+      "learning_rate": 3.426145808596533e-05,
+      "loss": 0.2217,
+      "step": 3160
+    },
+    {
+      "epoch": 2.131809011432414,
+      "grad_norm": 0.9885833263397217,
+      "learning_rate": 3.4202089764901454e-05,
+      "loss": 0.2299,
+      "step": 3170
+    },
+    {
+      "epoch": 2.1385339609952925,
+      "grad_norm": 0.959437906742096,
+      "learning_rate": 3.414272144383757e-05,
+      "loss": 0.1762,
+      "step": 3180
+    },
+    {
+      "epoch": 2.1452589105581708,
+      "grad_norm": 0.8083593249320984,
+      "learning_rate": 3.408335312277369e-05,
+      "loss": 0.1829,
+      "step": 3190
+    },
+    {
+      "epoch": 2.151983860121049,
+      "grad_norm": 1.081933856010437,
+      "learning_rate": 3.402398480170981e-05,
+      "loss": 0.2263,
+      "step": 3200
+    },
+    {
+      "epoch": 2.1587088096839273,
+      "grad_norm": 1.1069368124008179,
+      "learning_rate": 3.3964616480645926e-05,
+      "loss": 0.2492,
+      "step": 3210
+    },
+    {
+      "epoch": 2.1654337592468056,
+      "grad_norm": 0.8485414385795593,
+      "learning_rate": 3.390524815958205e-05,
+      "loss": 0.2201,
+      "step": 3220
+    },
+    {
+      "epoch": 2.172158708809684,
+      "grad_norm": 1.1156219244003296,
+      "learning_rate": 3.384587983851817e-05,
+      "loss": 0.2494,
+      "step": 3230
+    },
+    {
+      "epoch": 2.178883658372562,
+      "grad_norm": 0.8588371276855469,
+      "learning_rate": 3.378651151745429e-05,
+      "loss": 0.2784,
+      "step": 3240
+    },
+    {
+      "epoch": 2.1856086079354404,
+      "grad_norm": 0.9314185380935669,
+      "learning_rate": 3.372714319639041e-05,
+      "loss": 0.2127,
+      "step": 3250
+    },
+    {
+      "epoch": 2.1923335574983187,
+      "grad_norm": 1.0111339092254639,
+      "learning_rate": 3.366777487532653e-05,
+      "loss": 0.2082,
+      "step": 3260
+    },
+    {
+      "epoch": 2.199058507061197,
+      "grad_norm": 0.8263574242591858,
+      "learning_rate": 3.360840655426265e-05,
+      "loss": 0.2052,
+      "step": 3270
+    },
+    {
+      "epoch": 2.2057834566240753,
+      "grad_norm": 0.9010327458381653,
+      "learning_rate": 3.3549038233198766e-05,
+      "loss": 0.246,
+      "step": 3280
+    },
+    {
+      "epoch": 2.2125084061869535,
+      "grad_norm": 1.0510506629943848,
+      "learning_rate": 3.3489669912134884e-05,
+      "loss": 0.2138,
+      "step": 3290
+    },
+    {
+      "epoch": 2.219233355749832,
+      "grad_norm": 1.0803182125091553,
+      "learning_rate": 3.343030159107101e-05,
+      "loss": 0.2478,
+      "step": 3300
+    },
+    {
+      "epoch": 2.22595830531271,
+      "grad_norm": 0.7982410788536072,
+      "learning_rate": 3.337093327000713e-05,
+      "loss": 0.2249,
+      "step": 3310
+    },
+    {
+      "epoch": 2.2326832548755884,
+      "grad_norm": 0.7922587990760803,
+      "learning_rate": 3.3311564948943245e-05,
+      "loss": 0.2598,
+      "step": 3320
+    },
+    {
+      "epoch": 2.2394082044384667,
+      "grad_norm": 0.752295732498169,
+      "learning_rate": 3.325219662787937e-05,
+      "loss": 0.2036,
+      "step": 3330
+    },
+    {
+      "epoch": 2.246133154001345,
+      "grad_norm": 0.9120996594429016,
+      "learning_rate": 3.319282830681548e-05,
+      "loss": 0.2507,
+      "step": 3340
+    },
+    {
+      "epoch": 2.2528581035642232,
+      "grad_norm": 1.0284003019332886,
+      "learning_rate": 3.3133459985751606e-05,
+      "loss": 0.2191,
+      "step": 3350
+    },
+    {
+      "epoch": 2.2595830531271015,
+      "grad_norm": 0.6306409239768982,
+      "learning_rate": 3.3074091664687724e-05,
+      "loss": 0.2093,
+      "step": 3360
+    },
+    {
+      "epoch": 2.26630800268998,
+      "grad_norm": 0.8642512559890747,
+      "learning_rate": 3.301472334362384e-05,
+      "loss": 0.2246,
+      "step": 3370
+    },
+    {
+      "epoch": 2.273032952252858,
+      "grad_norm": 0.7505747079849243,
+      "learning_rate": 3.295535502255997e-05,
+      "loss": 0.1806,
+      "step": 3380
+    },
+    {
+      "epoch": 2.2797579018157363,
+      "grad_norm": 0.7727899551391602,
+      "learning_rate": 3.2901923533602466e-05,
+      "loss": 0.204,
+      "step": 3390
+    },
+    {
+      "epoch": 2.2864828513786146,
+      "grad_norm": 0.7602193355560303,
+      "learning_rate": 3.284255521253859e-05,
+      "loss": 0.243,
+      "step": 3400
+    },
+    {
+      "epoch": 2.293207800941493,
+      "grad_norm": 0.6114295125007629,
+      "learning_rate": 3.278318689147471e-05,
+      "loss": 0.2212,
+      "step": 3410
+    },
+    {
+      "epoch": 2.299932750504371,
+      "grad_norm": 0.8132253289222717,
+      "learning_rate": 3.272381857041083e-05,
+      "loss": 0.2095,
+      "step": 3420
+    },
+    {
+      "epoch": 2.3066577000672495,
+      "grad_norm": 1.0195751190185547,
+      "learning_rate": 3.266445024934695e-05,
+      "loss": 0.2549,
+      "step": 3430
+    },
+    {
+      "epoch": 2.3133826496301277,
+      "grad_norm": 0.8007863163948059,
+      "learning_rate": 3.260508192828307e-05,
+      "loss": 0.2355,
+      "step": 3440
+    },
+    {
+      "epoch": 2.320107599193006,
+      "grad_norm": 0.736564040184021,
+      "learning_rate": 3.254571360721919e-05,
+      "loss": 0.2084,
+      "step": 3450
+    },
+    {
+      "epoch": 2.3268325487558843,
+      "grad_norm": 0.7039320468902588,
+      "learning_rate": 3.2492282118261695e-05,
+      "loss": 0.187,
+      "step": 3460
+    },
+    {
+      "epoch": 2.3335574983187626,
+      "grad_norm": 0.7210522890090942,
+      "learning_rate": 3.243291379719782e-05,
+      "loss": 0.2183,
+      "step": 3470
+    },
+    {
+      "epoch": 2.340282447881641,
+      "grad_norm": 1.0226560831069946,
+      "learning_rate": 3.237354547613394e-05,
+      "loss": 0.1965,
+      "step": 3480
+    },
+    {
+      "epoch": 2.347007397444519,
+      "grad_norm": 0.9947569966316223,
+      "learning_rate": 3.2314177155070056e-05,
+      "loss": 0.283,
+      "step": 3490
+    },
+    {
+      "epoch": 2.3537323470073974,
+      "grad_norm": 0.68961101770401,
+      "learning_rate": 3.225480883400618e-05,
+      "loss": 0.1731,
+      "step": 3500
+    },
+    {
+      "epoch": 2.3537323470073974,
+      "eval_loss": 0.18062594532966614,
+      "eval_runtime": 14.5269,
+      "eval_samples_per_second": 182.007,
+      "eval_steps_per_second": 22.785,
+      "step": 3500
+    },
+    {
+      "epoch": 2.3604572965702757,
+      "grad_norm": 1.0577141046524048,
+      "learning_rate": 3.219544051294229e-05,
+      "loss": 0.2143,
+      "step": 3510
+    },
+    {
+      "epoch": 2.367182246133154,
+      "grad_norm": 0.7405422925949097,
+      "learning_rate": 3.2136072191878416e-05,
+      "loss": 0.1852,
+      "step": 3520
+    },
+    {
+      "epoch": 2.3739071956960323,
+      "grad_norm": 0.7497440576553345,
+      "learning_rate": 3.2076703870814535e-05,
+      "loss": 0.2103,
+      "step": 3530
+    },
+    {
+      "epoch": 2.3806321452589105,
+      "grad_norm": 1.0656136274337769,
+      "learning_rate": 3.201733554975065e-05,
+      "loss": 0.2264,
+      "step": 3540
+    },
+    {
+      "epoch": 2.387357094821789,
+      "grad_norm": 0.7102001905441284,
+      "learning_rate": 3.195796722868678e-05,
+      "loss": 0.212,
+      "step": 3550
+    },
+    {
+      "epoch": 2.394082044384667,
+      "grad_norm": 0.681854248046875,
+      "learning_rate": 3.1898598907622896e-05,
+      "loss": 0.2084,
+      "step": 3560
+    },
+    {
+      "epoch": 2.4008069939475454,
+      "grad_norm": 0.8357053995132446,
+      "learning_rate": 3.1839230586559014e-05,
+      "loss": 0.2552,
+      "step": 3570
+    },
+    {
+      "epoch": 2.4075319435104237,
+      "grad_norm": 0.9674760699272156,
+      "learning_rate": 3.177986226549513e-05,
+      "loss": 0.1648,
+      "step": 3580
+    },
+    {
+      "epoch": 2.414256893073302,
+      "grad_norm": 0.9936577081680298,
+      "learning_rate": 3.172049394443125e-05,
+      "loss": 0.281,
+      "step": 3590
+    },
+    {
+      "epoch": 2.42098184263618,
+      "grad_norm": 0.8406476974487305,
+      "learning_rate": 3.1661125623367375e-05,
+      "loss": 0.2074,
+      "step": 3600
+    },
+    {
+      "epoch": 2.4277067921990585,
+      "grad_norm": 0.6320979595184326,
+      "learning_rate": 3.160175730230349e-05,
+      "loss": 0.2265,
+      "step": 3610
+    },
+    {
+      "epoch": 2.4344317417619368,
+      "grad_norm": 1.0120488405227661,
+      "learning_rate": 3.154238898123961e-05,
+      "loss": 0.2178,
+      "step": 3620
+    },
+    {
+      "epoch": 2.441156691324815,
+      "grad_norm": 0.6850121021270752,
+      "learning_rate": 3.1483020660175736e-05,
+      "loss": 0.1898,
+      "step": 3630
+    },
+    {
+      "epoch": 2.4478816408876933,
+      "grad_norm": 0.7465779185295105,
+      "learning_rate": 3.1423652339111854e-05,
+      "loss": 0.2227,
+      "step": 3640
+    },
+    {
+      "epoch": 2.4546065904505716,
+      "grad_norm": 0.769567608833313,
+      "learning_rate": 3.136428401804797e-05,
+      "loss": 0.2123,
+      "step": 3650
+    },
+    {
+      "epoch": 2.46133154001345,
+      "grad_norm": 0.7620592713356018,
+      "learning_rate": 3.130491569698409e-05,
+      "loss": 0.2213,
+      "step": 3660
+    },
+    {
+      "epoch": 2.468056489576328,
+      "grad_norm": 0.6534302234649658,
+      "learning_rate": 3.124554737592021e-05,
+      "loss": 0.182,
+      "step": 3670
+    },
+    {
+      "epoch": 2.4747814391392065,
+      "grad_norm": 1.0276516675949097,
+      "learning_rate": 3.118617905485633e-05,
+      "loss": 0.2901,
+      "step": 3680
+    },
+    {
+      "epoch": 2.4815063887020847,
+      "grad_norm": 0.6156563758850098,
+      "learning_rate": 3.112681073379245e-05,
+      "loss": 0.195,
+      "step": 3690
+    },
+    {
+      "epoch": 2.488231338264963,
+      "grad_norm": 0.9594860672950745,
+      "learning_rate": 3.106744241272857e-05,
+      "loss": 0.1914,
+      "step": 3700
+    },
+    {
+      "epoch": 2.4949562878278413,
+      "grad_norm": 0.935619592666626,
+      "learning_rate": 3.1008074091664694e-05,
+      "loss": 0.1992,
+      "step": 3710
+    },
+    {
+      "epoch": 2.5016812373907196,
+      "grad_norm": 1.0625261068344116,
+      "learning_rate": 3.094870577060081e-05,
+      "loss": 0.1935,
+      "step": 3720
+    },
+    {
+      "epoch": 2.508406186953598,
+      "grad_norm": 0.7582225799560547,
+      "learning_rate": 3.088933744953692e-05,
+      "loss": 0.2272,
+      "step": 3730
+    },
+    {
+      "epoch": 2.515131136516476,
+      "grad_norm": 0.5857738852500916,
+      "learning_rate": 3.082996912847305e-05,
+      "loss": 0.1747,
+      "step": 3740
+    },
+    {
+      "epoch": 2.5218560860793544,
+      "grad_norm": 0.735443651676178,
+      "learning_rate": 3.0770600807409166e-05,
+      "loss": 0.2151,
+      "step": 3750
+    },
+    {
+      "epoch": 2.5285810356422327,
+      "grad_norm": 0.7247041463851929,
+      "learning_rate": 3.0711232486345284e-05,
+      "loss": 0.1893,
+      "step": 3760
+    },
+    {
+      "epoch": 2.535305985205111,
+      "grad_norm": 0.6171685457229614,
+      "learning_rate": 3.065186416528141e-05,
+      "loss": 0.2087,
+      "step": 3770
+    },
+    {
+      "epoch": 2.5420309347679892,
+      "grad_norm": 0.8305767178535461,
+      "learning_rate": 3.059249584421753e-05,
+      "loss": 0.2012,
+      "step": 3780
+    },
+    {
+      "epoch": 2.5487558843308675,
+      "grad_norm": 0.7687706351280212,
+      "learning_rate": 3.053312752315365e-05,
+      "loss": 0.1859,
+      "step": 3790
+    },
+    {
+      "epoch": 2.555480833893746,
+      "grad_norm": 0.7182943224906921,
+      "learning_rate": 3.047375920208977e-05,
+      "loss": 0.2004,
+      "step": 3800
+    },
+    {
+      "epoch": 2.562205783456624,
+      "grad_norm": 0.8720729947090149,
+      "learning_rate": 3.0414390881025884e-05,
+      "loss": 0.1964,
+      "step": 3810
+    },
+    {
+      "epoch": 2.5689307330195024,
+      "grad_norm": 0.7959709167480469,
+      "learning_rate": 3.0355022559962003e-05,
+      "loss": 0.21,
+      "step": 3820
+    },
+    {
+      "epoch": 2.5756556825823806,
+      "grad_norm": 0.7620719075202942,
+      "learning_rate": 3.0295654238898124e-05,
+      "loss": 0.2158,
+      "step": 3830
+    },
+    {
+      "epoch": 2.582380632145259,
+      "grad_norm": 0.6374610662460327,
+      "learning_rate": 3.0236285917834245e-05,
+      "loss": 0.1785,
+      "step": 3840
+    },
+    {
+      "epoch": 2.589105581708137,
+      "grad_norm": 0.9034950137138367,
+      "learning_rate": 3.0176917596770367e-05,
+      "loss": 0.2177,
+      "step": 3850
+    },
+    {
+      "epoch": 2.5958305312710155,
+      "grad_norm": 0.7911028861999512,
+      "learning_rate": 3.0117549275706485e-05,
+      "loss": 0.1848,
+      "step": 3860
+    },
+    {
+      "epoch": 2.6025554808338938,
+      "grad_norm": 0.7024137377738953,
+      "learning_rate": 3.0058180954642606e-05,
+      "loss": 0.1929,
+      "step": 3870
+    },
+    {
+      "epoch": 2.609280430396772,
+      "grad_norm": 0.7845777273178101,
+      "learning_rate": 2.999881263357872e-05,
+      "loss": 0.2002,
+      "step": 3880
+    },
+    {
+      "epoch": 2.6160053799596503,
+      "grad_norm": 0.8540692329406738,
+      "learning_rate": 2.9939444312514843e-05,
+      "loss": 0.2275,
+      "step": 3890
+    },
+    {
+      "epoch": 2.6227303295225286,
+      "grad_norm": 0.707000732421875,
+      "learning_rate": 2.988007599145096e-05,
+      "loss": 0.2423,
+      "step": 3900
+    },
+    {
+      "epoch": 2.629455279085407,
+      "grad_norm": 0.8943628668785095,
+      "learning_rate": 2.9820707670387082e-05,
+      "loss": 0.1889,
+      "step": 3910
+    },
+    {
+      "epoch": 2.636180228648285,
+      "grad_norm": 0.7365798354148865,
+      "learning_rate": 2.9761339349323204e-05,
+      "loss": 0.1885,
+      "step": 3920
+    },
+    {
+      "epoch": 2.6429051782111634,
+      "grad_norm": 0.8606659173965454,
+      "learning_rate": 2.9701971028259325e-05,
+      "loss": 0.2033,
+      "step": 3930
+    },
+    {
+      "epoch": 2.6496301277740417,
+      "grad_norm": 1.0965065956115723,
+      "learning_rate": 2.9642602707195443e-05,
+      "loss": 0.2234,
+      "step": 3940
+    },
+    {
+      "epoch": 2.65635507733692,
+      "grad_norm": 0.8194990754127502,
+      "learning_rate": 2.9583234386131564e-05,
+      "loss": 0.1973,
+      "step": 3950
+    },
+    {
+      "epoch": 2.6630800268997983,
+      "grad_norm": 0.9065597653388977,
+      "learning_rate": 2.952386606506768e-05,
+      "loss": 0.2189,
+      "step": 3960
+    },
+    {
+      "epoch": 2.6698049764626766,
+      "grad_norm": 0.7854435443878174,
+      "learning_rate": 2.94644977440038e-05,
+      "loss": 0.2179,
+      "step": 3970
+    },
+    {
+      "epoch": 2.676529926025555,
+      "grad_norm": 0.741438627243042,
+      "learning_rate": 2.940512942293992e-05,
+      "loss": 0.2321,
+      "step": 3980
+    },
+    {
+      "epoch": 2.683254875588433,
+      "grad_norm": 0.6715734601020813,
+      "learning_rate": 2.934576110187604e-05,
+      "loss": 0.2344,
+      "step": 3990
+    },
+    {
+      "epoch": 2.6899798251513114,
+      "grad_norm": 0.94840407371521,
+      "learning_rate": 2.928639278081216e-05,
+      "loss": 0.2516,
+      "step": 4000
+    },
+    {
+      "epoch": 2.6899798251513114,
+      "eval_loss": 0.17142708599567413,
+      "eval_runtime": 14.4655,
+      "eval_samples_per_second": 182.78,
+      "eval_steps_per_second": 22.882,
+      "step": 4000
+    },
+    {
+      "epoch": 2.6967047747141897,
+      "grad_norm": 1.161207914352417,
+      "learning_rate": 2.922702445974828e-05,
+      "loss": 0.1899,
+      "step": 4010
+    },
+    {
+      "epoch": 2.703429724277068,
+      "grad_norm": 1.0559875965118408,
+      "learning_rate": 2.91676561386844e-05,
+      "loss": 0.2161,
+      "step": 4020
+    },
+    {
+      "epoch": 2.7101546738399462,
+      "grad_norm": 1.0273319482803345,
+      "learning_rate": 2.9108287817620516e-05,
+      "loss": 0.2025,
+      "step": 4030
+    },
+    {
+      "epoch": 2.7168796234028245,
+      "grad_norm": 0.8442862033843994,
+      "learning_rate": 2.9048919496556637e-05,
+      "loss": 0.1655,
+      "step": 4040
+    },
+    {
+      "epoch": 2.723604572965703,
+      "grad_norm": 0.8517313003540039,
+      "learning_rate": 2.898955117549276e-05,
+      "loss": 0.2349,
+      "step": 4050
+    },
+    {
+      "epoch": 2.730329522528581,
+      "grad_norm": 0.7692966461181641,
+      "learning_rate": 2.8930182854428877e-05,
+      "loss": 0.2297,
+      "step": 4060
+    },
+    {
+      "epoch": 2.7370544720914594,
+      "grad_norm": 0.6031805276870728,
+      "learning_rate": 2.8870814533364998e-05,
+      "loss": 0.1809,
+      "step": 4070
+    },
+    {
+      "epoch": 2.7437794216543376,
+      "grad_norm": 0.9225636720657349,
+      "learning_rate": 2.881144621230112e-05,
+      "loss": 0.1832,
+      "step": 4080
+    },
+    {
+      "epoch": 2.750504371217216,
+      "grad_norm": 0.8155885338783264,
+      "learning_rate": 2.8752077891237238e-05,
+      "loss": 0.1758,
+      "step": 4090
+    },
+    {
+      "epoch": 2.757229320780094,
+      "grad_norm": 0.7248073220252991,
+      "learning_rate": 2.869270957017336e-05,
+      "loss": 0.1832,
+      "step": 4100
+    },
+    {
+      "epoch": 2.7639542703429725,
+      "grad_norm": 0.6624840497970581,
+      "learning_rate": 2.8633341249109474e-05,
+      "loss": 0.1891,
+      "step": 4110
+    },
+    {
+      "epoch": 2.7706792199058508,
+      "grad_norm": 0.4667346179485321,
+      "learning_rate": 2.8573972928045595e-05,
+      "loss": 0.21,
+      "step": 4120
+    },
+    {
+      "epoch": 2.777404169468729,
+      "grad_norm": 0.8075745105743408,
+      "learning_rate": 2.8514604606981717e-05,
+      "loss": 0.205,
+      "step": 4130
+    },
+    {
+      "epoch": 2.7841291190316073,
+      "grad_norm": 0.8244682550430298,
+      "learning_rate": 2.8455236285917835e-05,
+      "loss": 0.1878,
+      "step": 4140
+    },
+    {
+      "epoch": 2.7908540685944856,
+      "grad_norm": 0.8454658389091492,
+      "learning_rate": 2.8395867964853956e-05,
+      "loss": 0.1849,
+      "step": 4150
+    },
+    {
+      "epoch": 2.797579018157364,
+      "grad_norm": 0.8751473426818848,
+      "learning_rate": 2.8336499643790078e-05,
+      "loss": 0.2139,
+      "step": 4160
+    },
+    {
+      "epoch": 2.804303967720242,
+      "grad_norm": 0.8629645705223083,
+      "learning_rate": 2.8277131322726196e-05,
+      "loss": 0.1862,
+      "step": 4170
+    },
+    {
+      "epoch": 2.8110289172831204,
+      "grad_norm": 0.6888182759284973,
+      "learning_rate": 2.821776300166231e-05,
+      "loss": 0.2068,
+      "step": 4180
+    },
+    {
+      "epoch": 2.8177538668459987,
+      "grad_norm": 0.7422506809234619,
+      "learning_rate": 2.8158394680598432e-05,
+      "loss": 0.1572,
+      "step": 4190
+    },
+    {
+      "epoch": 2.824478816408877,
+      "grad_norm": 0.6318463683128357,
+      "learning_rate": 2.8099026359534553e-05,
+      "loss": 0.2644,
+      "step": 4200
+    },
+    {
+      "epoch": 2.8312037659717553,
+      "grad_norm": 0.81563800573349,
+      "learning_rate": 2.803965803847067e-05,
+      "loss": 0.2106,
+      "step": 4210
+    },
+    {
+      "epoch": 2.8379287155346336,
+      "grad_norm": 0.8201348781585693,
+      "learning_rate": 2.7980289717406793e-05,
+      "loss": 0.2318,
+      "step": 4220
+    },
+    {
+      "epoch": 2.844653665097512,
+      "grad_norm": 0.8415096402168274,
+      "learning_rate": 2.7920921396342914e-05,
+      "loss": 0.1859,
+      "step": 4230
+    },
+    {
+      "epoch": 2.85137861466039,
+      "grad_norm": 0.7930416464805603,
+      "learning_rate": 2.7861553075279036e-05,
+      "loss": 0.2404,
+      "step": 4240
+    },
+    {
+      "epoch": 2.8581035642232684,
+      "grad_norm": 0.7852760553359985,
+      "learning_rate": 2.7802184754215154e-05,
+      "loss": 0.1959,
+      "step": 4250
+    },
+    {
+      "epoch": 2.8648285137861467,
+      "grad_norm": 0.6727051138877869,
+      "learning_rate": 2.774281643315127e-05,
+      "loss": 0.1696,
+      "step": 4260
+    },
+    {
+      "epoch": 2.871553463349025,
+      "grad_norm": 0.7168065905570984,
+      "learning_rate": 2.768344811208739e-05,
+      "loss": 0.1896,
+      "step": 4270
+    },
+    {
+      "epoch": 2.8782784129119032,
+      "grad_norm": 0.6939849853515625,
+      "learning_rate": 2.762407979102351e-05,
+      "loss": 0.2072,
+      "step": 4280
+    },
+    {
+      "epoch": 2.8850033624747815,
+      "grad_norm": 1.2583644390106201,
+      "learning_rate": 2.756471146995963e-05,
+      "loss": 0.2281,
+      "step": 4290
+    },
+    {
+      "epoch": 2.89172831203766,
+      "grad_norm": 0.8161798119544983,
+      "learning_rate": 2.750534314889575e-05,
+      "loss": 0.1871,
+      "step": 4300
+    },
+    {
+      "epoch": 2.898453261600538,
+      "grad_norm": 0.7994545102119446,
+      "learning_rate": 2.7445974827831872e-05,
+      "loss": 0.1638,
+      "step": 4310
+    },
+    {
+      "epoch": 2.9051782111634163,
+      "grad_norm": 0.6566575169563293,
+      "learning_rate": 2.738660650676799e-05,
+      "loss": 0.1552,
+      "step": 4320
+    },
+    {
+      "epoch": 2.9119031607262946,
+      "grad_norm": 1.0358943939208984,
+      "learning_rate": 2.732723818570411e-05,
+      "loss": 0.1851,
+      "step": 4330
+    },
+    {
+      "epoch": 2.918628110289173,
+      "grad_norm": 0.6944819092750549,
+      "learning_rate": 2.7267869864640227e-05,
+      "loss": 0.1477,
+      "step": 4340
+    },
+    {
+      "epoch": 2.925353059852051,
+      "grad_norm": 0.88713139295578,
+      "learning_rate": 2.7208501543576348e-05,
+      "loss": 0.1927,
+      "step": 4350
+    },
+    {
+      "epoch": 2.9320780094149295,
+      "grad_norm": 0.8105150461196899,
+      "learning_rate": 2.714913322251247e-05,
+      "loss": 0.1958,
+      "step": 4360
+    },
+    {
+      "epoch": 2.9388029589778077,
+      "grad_norm": 0.6686846613883972,
+      "learning_rate": 2.7089764901448588e-05,
+      "loss": 0.1935,
+      "step": 4370
+    },
+    {
+      "epoch": 2.945527908540686,
+      "grad_norm": 0.9116830229759216,
+      "learning_rate": 2.703039658038471e-05,
+      "loss": 0.1814,
+      "step": 4380
+    },
+    {
+      "epoch": 2.9522528581035643,
+      "grad_norm": 0.5856004357337952,
+      "learning_rate": 2.697102825932083e-05,
+      "loss": 0.1746,
+      "step": 4390
+    },
+    {
+      "epoch": 2.9589778076664426,
+      "grad_norm": 0.8741239309310913,
+      "learning_rate": 2.691165993825695e-05,
+      "loss": 0.1816,
+      "step": 4400
+    },
+    {
+      "epoch": 2.965702757229321,
+      "grad_norm": 0.7971295714378357,
+      "learning_rate": 2.6852291617193063e-05,
+      "loss": 0.2379,
+      "step": 4410
+    },
+    {
+      "epoch": 2.972427706792199,
+      "grad_norm": 0.657518744468689,
+      "learning_rate": 2.6792923296129185e-05,
+      "loss": 0.2032,
+      "step": 4420
+    },
+    {
+      "epoch": 2.9791526563550774,
+      "grad_norm": 0.6854186058044434,
+      "learning_rate": 2.6733554975065306e-05,
+      "loss": 0.2078,
+      "step": 4430
+    },
+    {
+      "epoch": 2.9858776059179557,
+      "grad_norm": 0.9783342480659485,
+      "learning_rate": 2.6674186654001428e-05,
+      "loss": 0.1903,
+      "step": 4440
+    },
+    {
+      "epoch": 2.992602555480834,
+      "grad_norm": 0.8605186939239502,
+      "learning_rate": 2.6614818332937546e-05,
+      "loss": 0.2001,
+      "step": 4450
+    },
+    {
+      "epoch": 2.9993275050437123,
+      "grad_norm": 0.4880738854408264,
+      "learning_rate": 2.6555450011873667e-05,
+      "loss": 0.2122,
+      "step": 4460
+    },
+    {
+      "epoch": 3.0060524546065905,
+      "grad_norm": 0.7420046329498291,
+      "learning_rate": 2.649608169080979e-05,
+      "loss": 0.2074,
+      "step": 4470
+    },
+    {
+      "epoch": 3.012777404169469,
+      "grad_norm": 0.8112709522247314,
+      "learning_rate": 2.6436713369745903e-05,
+      "loss": 0.1728,
+      "step": 4480
+    },
+    {
+      "epoch": 3.019502353732347,
+      "grad_norm": 0.5927948355674744,
+      "learning_rate": 2.637734504868202e-05,
+      "loss": 0.1838,
+      "step": 4490
+    },
+    {
+      "epoch": 3.0262273032952254,
+      "grad_norm": 0.7261831164360046,
+      "learning_rate": 2.6317976727618143e-05,
+      "loss": 0.1431,
+      "step": 4500
+    },
+    {
+      "epoch": 3.0262273032952254,
+      "eval_loss": 0.16571946442127228,
+      "eval_runtime": 14.5789,
+      "eval_samples_per_second": 181.358,
+      "eval_steps_per_second": 22.704,
+      "step": 4500
+    },
+    {
+      "epoch": 3.0329522528581037,
+      "grad_norm": 0.6484703421592712,
+      "learning_rate": 2.6258608406554264e-05,
+      "loss": 0.1987,
+      "step": 4510
+    },
+    {
+      "epoch": 3.039677202420982,
+      "grad_norm": 0.7991098165512085,
+      "learning_rate": 2.6199240085490386e-05,
+      "loss": 0.2086,
+      "step": 4520
+    },
+    {
+      "epoch": 3.04640215198386,
+      "grad_norm": 0.7324103713035583,
+      "learning_rate": 2.6139871764426504e-05,
+      "loss": 0.1974,
+      "step": 4530
+    },
+    {
+      "epoch": 3.0531271015467385,
+      "grad_norm": 0.6129199266433716,
+      "learning_rate": 2.6080503443362625e-05,
+      "loss": 0.2471,
+      "step": 4540
+    },
+    {
+      "epoch": 3.0598520511096168,
+      "grad_norm": 0.7959029674530029,
+      "learning_rate": 2.6021135122298747e-05,
+      "loss": 0.1996,
+      "step": 4550
+    },
+    {
+      "epoch": 3.066577000672495,
+      "grad_norm": 0.8219399452209473,
+      "learning_rate": 2.596176680123486e-05,
+      "loss": 0.1978,
+      "step": 4560
+    },
+    {
+      "epoch": 3.0733019502353733,
+      "grad_norm": 0.7939678430557251,
+      "learning_rate": 2.590239848017098e-05,
+      "loss": 0.2257,
+      "step": 4570
+    },
+    {
+      "epoch": 3.0800268997982516,
+      "grad_norm": 0.5718218088150024,
+      "learning_rate": 2.58430301591071e-05,
+      "loss": 0.1844,
+      "step": 4580
+    },
+    {
+      "epoch": 3.08675184936113,
+      "grad_norm": 0.8061422109603882,
+      "learning_rate": 2.5783661838043222e-05,
+      "loss": 0.2007,
+      "step": 4590
+    },
+    {
+      "epoch": 3.093476798924008,
+      "grad_norm": 0.6342537999153137,
+      "learning_rate": 2.572429351697934e-05,
+      "loss": 0.1582,
+      "step": 4600
+    },
+    {
+      "epoch": 3.1002017484868865,
+      "grad_norm": 0.8753138780593872,
+      "learning_rate": 2.5664925195915462e-05,
+      "loss": 0.1767,
+      "step": 4610
+    },
+    {
+      "epoch": 3.1069266980497647,
+      "grad_norm": 0.5665940642356873,
+      "learning_rate": 2.5605556874851583e-05,
+      "loss": 0.2006,
+      "step": 4620
+    },
+    {
+      "epoch": 3.113651647612643,
+      "grad_norm": 0.7274763584136963,
+      "learning_rate": 2.5546188553787698e-05,
+      "loss": 0.197,
+      "step": 4630
+    },
+    {
+      "epoch": 3.1203765971755213,
+      "grad_norm": 0.9588426351547241,
+      "learning_rate": 2.548682023272382e-05,
+      "loss": 0.1797,
+      "step": 4640
+    },
+    {
+      "epoch": 3.1271015467383996,
+      "grad_norm": 0.7508196830749512,
+      "learning_rate": 2.5427451911659938e-05,
+      "loss": 0.1681,
+      "step": 4650
+    },
+    {
+      "epoch": 3.133826496301278,
+      "grad_norm": 0.9038055539131165,
+      "learning_rate": 2.536808359059606e-05,
+      "loss": 0.1689,
+      "step": 4660
+    },
+    {
+      "epoch": 3.140551445864156,
+      "grad_norm": 0.8027380108833313,
+      "learning_rate": 2.530871526953218e-05,
+      "loss": 0.2027,
+      "step": 4670
+    },
+    {
+      "epoch": 3.1472763954270344,
+      "grad_norm": 0.8716439008712769,
+      "learning_rate": 2.52493469484683e-05,
+      "loss": 0.2026,
+      "step": 4680
+    },
+    {
+      "epoch": 3.1540013449899127,
+      "grad_norm": 0.6943103075027466,
+      "learning_rate": 2.518997862740442e-05,
+      "loss": 0.1811,
+      "step": 4690
+    },
+    {
+      "epoch": 3.160726294552791,
+      "grad_norm": 0.6706501245498657,
+      "learning_rate": 2.5130610306340535e-05,
+      "loss": 0.1751,
+      "step": 4700
+    },
+    {
+      "epoch": 3.1674512441156693,
+      "grad_norm": 0.7474800944328308,
+      "learning_rate": 2.5071241985276656e-05,
+      "loss": 0.1824,
+      "step": 4710
+    },
+    {
+      "epoch": 3.1741761936785475,
+      "grad_norm": 0.7841882109642029,
+      "learning_rate": 2.5011873664212778e-05,
+      "loss": 0.1923,
+      "step": 4720
+    },
+    {
+      "epoch": 3.180901143241426,
+      "grad_norm": 0.7595344185829163,
+      "learning_rate": 2.4952505343148896e-05,
+      "loss": 0.1989,
+      "step": 4730
+    },
+    {
+      "epoch": 3.187626092804304,
+      "grad_norm": 0.7175992727279663,
+      "learning_rate": 2.4893137022085017e-05,
+      "loss": 0.2077,
+      "step": 4740
+    },
+    {
+      "epoch": 3.1943510423671824,
+      "grad_norm": 0.8540729880332947,
+      "learning_rate": 2.483376870102114e-05,
+      "loss": 0.2082,
+      "step": 4750
+    },
+    {
+      "epoch": 3.2010759919300606,
+      "grad_norm": 0.8834501504898071,
+      "learning_rate": 2.4774400379957257e-05,
+      "loss": 0.1883,
+      "step": 4760
+    },
+    {
+      "epoch": 3.207800941492939,
+      "grad_norm": 1.0018720626831055,
+      "learning_rate": 2.4715032058893375e-05,
+      "loss": 0.1939,
+      "step": 4770
+    },
+    {
+      "epoch": 3.214525891055817,
+      "grad_norm": 0.8706895709037781,
+      "learning_rate": 2.4655663737829496e-05,
+      "loss": 0.171,
+      "step": 4780
+    },
+    {
+      "epoch": 3.2212508406186955,
+      "grad_norm": 0.9072573781013489,
+      "learning_rate": 2.4596295416765618e-05,
+      "loss": 0.1868,
+      "step": 4790
+    },
+    {
+      "epoch": 3.2279757901815738,
+      "grad_norm": 0.7303891777992249,
+      "learning_rate": 2.4536927095701732e-05,
+      "loss": 0.1806,
+      "step": 4800
+    },
+    {
+      "epoch": 3.234700739744452,
+      "grad_norm": 0.7761268615722656,
+      "learning_rate": 2.4477558774637854e-05,
+      "loss": 0.1857,
+      "step": 4810
+    },
+    {
+      "epoch": 3.2414256893073303,
+      "grad_norm": 0.5720176100730896,
+      "learning_rate": 2.4418190453573975e-05,
+      "loss": 0.1915,
+      "step": 4820
+    },
+    {
+      "epoch": 3.2481506388702086,
+      "grad_norm": 0.7509124875068665,
+      "learning_rate": 2.4358822132510097e-05,
+      "loss": 0.1643,
+      "step": 4830
+    },
+    {
+      "epoch": 3.254875588433087,
+      "grad_norm": 0.963586688041687,
+      "learning_rate": 2.429945381144621e-05,
+      "loss": 0.1942,
+      "step": 4840
+    },
+    {
+      "epoch": 3.261600537995965,
+      "grad_norm": 0.688130259513855,
+      "learning_rate": 2.4240085490382333e-05,
+      "loss": 0.1746,
+      "step": 4850
+    },
+    {
+      "epoch": 3.2683254875588434,
+      "grad_norm": 0.6827595233917236,
+      "learning_rate": 2.4180717169318454e-05,
+      "loss": 0.1877,
+      "step": 4860
+    },
+    {
+      "epoch": 3.2750504371217217,
+      "grad_norm": 0.8125810027122498,
+      "learning_rate": 2.4121348848254572e-05,
+      "loss": 0.1902,
+      "step": 4870
+    },
+    {
+      "epoch": 3.2817753866846,
+      "grad_norm": 0.6463695764541626,
+      "learning_rate": 2.406198052719069e-05,
+      "loss": 0.2067,
+      "step": 4880
+    },
+    {
+      "epoch": 3.2885003362474783,
+      "grad_norm": 0.3192298412322998,
+      "learning_rate": 2.4002612206126812e-05,
+      "loss": 0.1757,
+      "step": 4890
+    },
+    {
+      "epoch": 3.2952252858103566,
+      "grad_norm": 0.9306578040122986,
+      "learning_rate": 2.3943243885062933e-05,
+      "loss": 0.1811,
+      "step": 4900
+    },
+    {
+      "epoch": 3.301950235373235,
+      "grad_norm": 1.080837607383728,
+      "learning_rate": 2.388387556399905e-05,
+      "loss": 0.2022,
+      "step": 4910
+    },
+    {
+      "epoch": 3.308675184936113,
+      "grad_norm": 0.6997596621513367,
+      "learning_rate": 2.382450724293517e-05,
+      "loss": 0.1729,
+      "step": 4920
+    },
+    {
+      "epoch": 3.3154001344989914,
+      "grad_norm": 0.6311511993408203,
+      "learning_rate": 2.376513892187129e-05,
+      "loss": 0.1493,
+      "step": 4930
+    },
+    {
+      "epoch": 3.3221250840618697,
+      "grad_norm": 0.7515596747398376,
+      "learning_rate": 2.3705770600807412e-05,
+      "loss": 0.1829,
+      "step": 4940
+    },
+    {
+      "epoch": 3.328850033624748,
+      "grad_norm": 0.6540161371231079,
+      "learning_rate": 2.364640227974353e-05,
+      "loss": 0.2014,
+      "step": 4950
+    },
+    {
+      "epoch": 3.3355749831876262,
+      "grad_norm": 0.6269936561584473,
+      "learning_rate": 2.358703395867965e-05,
+      "loss": 0.1896,
+      "step": 4960
+    },
+    {
+      "epoch": 3.3422999327505045,
+      "grad_norm": 0.871743381023407,
+      "learning_rate": 2.352766563761577e-05,
+      "loss": 0.1899,
+      "step": 4970
+    },
+    {
+      "epoch": 3.349024882313383,
+      "grad_norm": 0.6379810571670532,
+      "learning_rate": 2.3468297316551888e-05,
+      "loss": 0.1796,
+      "step": 4980
+    },
+    {
+      "epoch": 3.355749831876261,
+      "grad_norm": 0.6702184677124023,
+      "learning_rate": 2.340892899548801e-05,
+      "loss": 0.239,
+      "step": 4990
+    },
+    {
+      "epoch": 3.3624747814391394,
+      "grad_norm": 0.649553120136261,
+      "learning_rate": 2.3349560674424127e-05,
+      "loss": 0.1607,
+      "step": 5000
+    },
+    {
+      "epoch": 3.3624747814391394,
+      "eval_loss": 0.16000543534755707,
+      "eval_runtime": 14.3892,
+      "eval_samples_per_second": 183.75,
+      "eval_steps_per_second": 23.003,
+      "step": 5000
+    },
+    {
+      "epoch": 3.3691997310020176,
+      "grad_norm": 0.7338672280311584,
+      "learning_rate": 2.329019235336025e-05,
+      "loss": 0.1769,
+      "step": 5010
+    },
+    {
+      "epoch": 3.375924680564896,
+      "grad_norm": 0.9036253690719604,
+      "learning_rate": 2.3230824032296367e-05,
+      "loss": 0.1823,
+      "step": 5020
+    },
+    {
+      "epoch": 3.382649630127774,
+      "grad_norm": 0.8734234571456909,
+      "learning_rate": 2.317145571123249e-05,
+      "loss": 0.1667,
+      "step": 5030
+    },
+    {
+      "epoch": 3.3893745796906525,
+      "grad_norm": 0.9217347502708435,
+      "learning_rate": 2.3112087390168607e-05,
+      "loss": 0.1559,
+      "step": 5040
+    },
+    {
+      "epoch": 3.3960995292535308,
+      "grad_norm": 0.7761852741241455,
+      "learning_rate": 2.3052719069104728e-05,
+      "loss": 0.1663,
+      "step": 5050
+    },
+    {
+      "epoch": 3.402824478816409,
+      "grad_norm": 0.8161054849624634,
+      "learning_rate": 2.2993350748040846e-05,
+      "loss": 0.1958,
+      "step": 5060
+    },
+    {
+      "epoch": 3.4095494283792873,
+      "grad_norm": 0.8758025765419006,
+      "learning_rate": 2.2933982426976968e-05,
+      "loss": 0.1977,
+      "step": 5070
+    },
+    {
+      "epoch": 3.4162743779421656,
+      "grad_norm": 0.9359439611434937,
+      "learning_rate": 2.2874614105913086e-05,
+      "loss": 0.1823,
+      "step": 5080
+    },
+    {
+      "epoch": 3.422999327505044,
+      "grad_norm": 0.5680366158485413,
+      "learning_rate": 2.2815245784849207e-05,
+      "loss": 0.1989,
+      "step": 5090
+    },
+    {
+      "epoch": 3.429724277067922,
+      "grad_norm": 0.8269837498664856,
+      "learning_rate": 2.2755877463785325e-05,
+      "loss": 0.1878,
+      "step": 5100
+    },
+    {
+      "epoch": 3.4364492266308004,
+      "grad_norm": 0.6142411828041077,
+      "learning_rate": 2.2696509142721447e-05,
+      "loss": 0.2008,
+      "step": 5110
+    },
+    {
+      "epoch": 3.4431741761936787,
+      "grad_norm": 0.730137050151825,
+      "learning_rate": 2.2637140821657565e-05,
+      "loss": 0.184,
+      "step": 5120
+    },
+    {
+      "epoch": 3.449899125756557,
+      "grad_norm": 0.6891915798187256,
+      "learning_rate": 2.2577772500593683e-05,
+      "loss": 0.1677,
+      "step": 5130
+    },
+    {
+      "epoch": 3.4566240753194353,
+      "grad_norm": 0.7893393039703369,
+      "learning_rate": 2.2518404179529804e-05,
+      "loss": 0.2016,
+      "step": 5140
+    },
+    {
+      "epoch": 3.4633490248823136,
+      "grad_norm": 0.9286131262779236,
+      "learning_rate": 2.246497269057231e-05,
+      "loss": 0.2178,
+      "step": 5150
+    },
+    {
+      "epoch": 3.470073974445192,
+      "grad_norm": 0.9323815703392029,
+      "learning_rate": 2.2405604369508432e-05,
+      "loss": 0.16,
+      "step": 5160
+    },
+    {
+      "epoch": 3.47679892400807,
+      "grad_norm": 0.6662344336509705,
+      "learning_rate": 2.2346236048444553e-05,
+      "loss": 0.1703,
+      "step": 5170
+    },
+    {
+      "epoch": 3.4835238735709484,
+      "grad_norm": 0.6510077118873596,
+      "learning_rate": 2.228686772738067e-05,
+      "loss": 0.1631,
+      "step": 5180
+    },
+    {
+      "epoch": 3.4902488231338267,
+      "grad_norm": 0.7270941734313965,
+      "learning_rate": 2.222749940631679e-05,
+      "loss": 0.1555,
+      "step": 5190
+    },
+    {
+      "epoch": 3.496973772696705,
+      "grad_norm": 0.7204037308692932,
+      "learning_rate": 2.216813108525291e-05,
+      "loss": 0.2068,
+      "step": 5200
+    },
+    {
+      "epoch": 3.503698722259583,
+      "grad_norm": 0.4908240735530853,
+      "learning_rate": 2.2108762764189032e-05,
+      "loss": 0.1468,
+      "step": 5210
+    },
+    {
+      "epoch": 3.510423671822461,
+      "grad_norm": 0.9486116766929626,
+      "learning_rate": 2.2049394443125147e-05,
+      "loss": 0.156,
+      "step": 5220
+    },
+    {
+      "epoch": 3.5171486213853393,
+      "grad_norm": 0.8911501169204712,
+      "learning_rate": 2.199002612206127e-05,
+      "loss": 0.1853,
+      "step": 5230
+    },
+    {
+      "epoch": 3.5238735709482176,
+      "grad_norm": 0.7860977649688721,
+      "learning_rate": 2.193065780099739e-05,
+      "loss": 0.1714,
+      "step": 5240
+    },
+    {
+      "epoch": 3.530598520511096,
+      "grad_norm": 0.9978109002113342,
+      "learning_rate": 2.1871289479933508e-05,
+      "loss": 0.1569,
+      "step": 5250
+    },
+    {
+      "epoch": 3.537323470073974,
+      "grad_norm": 0.8278188109397888,
+      "learning_rate": 2.1811921158869626e-05,
+      "loss": 0.1977,
+      "step": 5260
+    },
+    {
+      "epoch": 3.5440484196368525,
+      "grad_norm": 0.75886470079422,
+      "learning_rate": 2.1752552837805748e-05,
+      "loss": 0.2145,
+      "step": 5270
+    },
+    {
+      "epoch": 3.5507733691997307,
+      "grad_norm": 0.608791172504425,
+      "learning_rate": 2.169318451674187e-05,
+      "loss": 0.1586,
+      "step": 5280
+    },
+    {
+      "epoch": 3.557498318762609,
+      "grad_norm": 0.8277237415313721,
+      "learning_rate": 2.1633816195677987e-05,
+      "loss": 0.213,
+      "step": 5290
+    },
+    {
+      "epoch": 3.5642232683254873,
+      "grad_norm": 0.6545449495315552,
+      "learning_rate": 2.1574447874614105e-05,
+      "loss": 0.1797,
+      "step": 5300
+    },
+    {
+      "epoch": 3.5709482178883656,
+      "grad_norm": 0.8395527005195618,
+      "learning_rate": 2.1515079553550227e-05,
+      "loss": 0.1776,
+      "step": 5310
+    },
+    {
+      "epoch": 3.577673167451244,
+      "grad_norm": 0.6380318403244019,
+      "learning_rate": 2.1455711232486348e-05,
+      "loss": 0.1342,
+      "step": 5320
+    },
+    {
+      "epoch": 3.584398117014122,
+      "grad_norm": 0.751994252204895,
+      "learning_rate": 2.1396342911422466e-05,
+      "loss": 0.1525,
+      "step": 5330
+    },
+    {
+      "epoch": 3.5911230665770004,
+      "grad_norm": 0.6051616668701172,
+      "learning_rate": 2.1336974590358584e-05,
+      "loss": 0.1682,
+      "step": 5340
+    },
+    {
+      "epoch": 3.5978480161398787,
+      "grad_norm": 0.8227441906929016,
+      "learning_rate": 2.1277606269294706e-05,
+      "loss": 0.2026,
+      "step": 5350
+    },
+    {
+      "epoch": 3.604572965702757,
+      "grad_norm": 0.6999370455741882,
+      "learning_rate": 2.1218237948230827e-05,
+      "loss": 0.1954,
+      "step": 5360
+    },
+    {
+      "epoch": 3.6112979152656353,
+      "grad_norm": 0.6701903343200684,
+      "learning_rate": 2.1158869627166945e-05,
+      "loss": 0.1559,
+      "step": 5370
+    },
+    {
+      "epoch": 3.6180228648285135,
+      "grad_norm": 1.0471676588058472,
+      "learning_rate": 2.1099501306103063e-05,
+      "loss": 0.1504,
+      "step": 5380
+    },
+    {
+      "epoch": 3.624747814391392,
+      "grad_norm": 0.8120239973068237,
+      "learning_rate": 2.1040132985039185e-05,
+      "loss": 0.1757,
+      "step": 5390
+    },
+    {
+      "epoch": 3.63147276395427,
+      "grad_norm": 0.7743443846702576,
+      "learning_rate": 2.0980764663975303e-05,
+      "loss": 0.2083,
+      "step": 5400
+    },
+    {
+      "epoch": 3.6381977135171484,
+      "grad_norm": 0.630879282951355,
+      "learning_rate": 2.0921396342911424e-05,
+      "loss": 0.1435,
+      "step": 5410
+    },
+    {
+      "epoch": 3.6449226630800267,
+      "grad_norm": 1.7099244594573975,
+      "learning_rate": 2.0862028021847542e-05,
+      "loss": 0.2272,
+      "step": 5420
+    },
+    {
+      "epoch": 3.651647612642905,
+      "grad_norm": 0.7191898226737976,
+      "learning_rate": 2.0802659700783664e-05,
+      "loss": 0.2123,
+      "step": 5430
+    },
+    {
+      "epoch": 3.658372562205783,
+      "grad_norm": 0.5158668160438538,
+      "learning_rate": 2.0743291379719782e-05,
+      "loss": 0.1714,
+      "step": 5440
+    },
+    {
+      "epoch": 3.6650975117686615,
+      "grad_norm": 0.591200590133667,
+      "learning_rate": 2.0683923058655903e-05,
+      "loss": 0.1435,
+      "step": 5450
+    },
+    {
+      "epoch": 3.6718224613315398,
+      "grad_norm": 0.6342896819114685,
+      "learning_rate": 2.062455473759202e-05,
+      "loss": 0.1615,
+      "step": 5460
+    },
+    {
+      "epoch": 3.678547410894418,
+      "grad_norm": 0.6707589030265808,
+      "learning_rate": 2.0565186416528143e-05,
+      "loss": 0.1646,
+      "step": 5470
+    },
+    {
+      "epoch": 3.6852723604572963,
+      "grad_norm": 0.9689118266105652,
+      "learning_rate": 2.050581809546426e-05,
+      "loss": 0.2129,
+      "step": 5480
+    },
+    {
+      "epoch": 3.6919973100201746,
+      "grad_norm": 0.7816817760467529,
+      "learning_rate": 2.0446449774400382e-05,
+      "loss": 0.1631,
+      "step": 5490
+    },
+    {
+      "epoch": 3.698722259583053,
+      "grad_norm": 0.7102280855178833,
+      "learning_rate": 2.03870814533365e-05,
+      "loss": 0.1895,
+      "step": 5500
+    },
+    {
+      "epoch": 3.698722259583053,
+      "eval_loss": 0.1554509997367859,
+      "eval_runtime": 14.4049,
+      "eval_samples_per_second": 183.549,
+      "eval_steps_per_second": 22.978,
+      "step": 5500
+    },
+    {
+      "epoch": 3.705447209145931,
+      "grad_norm": 0.8417555689811707,
+      "learning_rate": 2.0327713132272622e-05,
+      "loss": 0.2083,
+      "step": 5510
+    },
+    {
+      "epoch": 3.7121721587088095,
+      "grad_norm": 0.6126737594604492,
+      "learning_rate": 2.026834481120874e-05,
+      "loss": 0.1675,
+      "step": 5520
+    },
+    {
+      "epoch": 3.7188971082716877,
+      "grad_norm": 0.8237993717193604,
+      "learning_rate": 2.020897649014486e-05,
+      "loss": 0.1737,
+      "step": 5530
+    },
+    {
+      "epoch": 3.725622057834566,
+      "grad_norm": 0.531822144985199,
+      "learning_rate": 2.014960816908098e-05,
+      "loss": 0.1557,
+      "step": 5540
+    },
+    {
+      "epoch": 3.7323470073974443,
+      "grad_norm": 0.8391734957695007,
+      "learning_rate": 2.0090239848017097e-05,
+      "loss": 0.1905,
+      "step": 5550
+    },
+    {
+      "epoch": 3.7390719569603226,
+      "grad_norm": 0.7321840524673462,
+      "learning_rate": 2.003087152695322e-05,
+      "loss": 0.2007,
+      "step": 5560
+    },
+    {
+      "epoch": 3.745796906523201,
+      "grad_norm": 1.057217001914978,
+      "learning_rate": 1.997150320588934e-05,
+      "loss": 0.2366,
+      "step": 5570
+    },
+    {
+      "epoch": 3.752521856086079,
+      "grad_norm": 0.48040035367012024,
+      "learning_rate": 1.991213488482546e-05,
+      "loss": 0.1783,
+      "step": 5580
+    },
+    {
+      "epoch": 3.7592468056489574,
+      "grad_norm": 0.9027122855186462,
+      "learning_rate": 1.9852766563761576e-05,
+      "loss": 0.1978,
+      "step": 5590
+    },
+    {
+      "epoch": 3.7659717552118357,
+      "grad_norm": 0.8584364652633667,
+      "learning_rate": 1.9793398242697698e-05,
+      "loss": 0.1962,
+      "step": 5600
+    },
+    {
+      "epoch": 3.772696704774714,
+      "grad_norm": 0.8134893178939819,
+      "learning_rate": 1.9734029921633816e-05,
+      "loss": 0.1544,
+      "step": 5610
+    },
+    {
+      "epoch": 3.7794216543375923,
+      "grad_norm": 0.7096035480499268,
+      "learning_rate": 1.9674661600569937e-05,
+      "loss": 0.1601,
+      "step": 5620
+    },
+    {
+      "epoch": 3.7861466039004705,
+      "grad_norm": 0.7845041751861572,
+      "learning_rate": 1.9615293279506056e-05,
+      "loss": 0.2081,
+      "step": 5630
+    },
+    {
+      "epoch": 3.792871553463349,
+      "grad_norm": 0.5299544334411621,
+      "learning_rate": 1.9555924958442177e-05,
+      "loss": 0.1587,
+      "step": 5640
+    },
+    {
+      "epoch": 3.799596503026227,
+      "grad_norm": 0.5104948878288269,
+      "learning_rate": 1.9496556637378295e-05,
+      "loss": 0.2026,
+      "step": 5650
+    },
+    {
+      "epoch": 3.8063214525891054,
+      "grad_norm": 0.7379271984100342,
+      "learning_rate": 1.9437188316314416e-05,
+      "loss": 0.1811,
+      "step": 5660
+    },
+    {
+      "epoch": 3.8130464021519836,
+      "grad_norm": 0.686158299446106,
+      "learning_rate": 1.9377819995250535e-05,
+      "loss": 0.1743,
+      "step": 5670
+    },
+    {
+      "epoch": 3.819771351714862,
+      "grad_norm": 0.7175692915916443,
+      "learning_rate": 1.9318451674186656e-05,
+      "loss": 0.1751,
+      "step": 5680
+    },
+    {
+      "epoch": 3.82649630127774,
+      "grad_norm": 0.6327086687088013,
+      "learning_rate": 1.9259083353122774e-05,
+      "loss": 0.1471,
+      "step": 5690
+    },
+    {
+      "epoch": 3.8332212508406185,
+      "grad_norm": 0.8163675665855408,
+      "learning_rate": 1.9199715032058892e-05,
+      "loss": 0.1989,
+      "step": 5700
+    },
+    {
+      "epoch": 3.8399462004034968,
+      "grad_norm": 0.6851422190666199,
+      "learning_rate": 1.9140346710995014e-05,
+      "loss": 0.182,
+      "step": 5710
+    },
+    {
+      "epoch": 3.846671149966375,
+      "grad_norm": 0.9112792611122131,
+      "learning_rate": 1.9080978389931135e-05,
+      "loss": 0.2003,
+      "step": 5720
+    },
+    {
+      "epoch": 3.8533960995292533,
+      "grad_norm": 0.7452694773674011,
+      "learning_rate": 1.9021610068867253e-05,
+      "loss": 0.1823,
+      "step": 5730
+    },
+    {
+      "epoch": 3.8601210490921316,
+      "grad_norm": 0.8061573505401611,
+      "learning_rate": 1.896224174780337e-05,
+      "loss": 0.2005,
+      "step": 5740
+    },
+    {
+      "epoch": 3.86684599865501,
+      "grad_norm": 0.8620453476905823,
+      "learning_rate": 1.8902873426739493e-05,
+      "loss": 0.1951,
+      "step": 5750
+    },
+    {
+      "epoch": 3.873570948217888,
+      "grad_norm": 0.7951462864875793,
+      "learning_rate": 1.8843505105675614e-05,
+      "loss": 0.1879,
+      "step": 5760
+    },
+    {
+      "epoch": 3.8802958977807664,
+      "grad_norm": 0.787812352180481,
+      "learning_rate": 1.8784136784611732e-05,
+      "loss": 0.1486,
+      "step": 5770
+    },
+    {
+      "epoch": 3.8870208473436447,
+      "grad_norm": 0.8405473232269287,
+      "learning_rate": 1.872476846354785e-05,
+      "loss": 0.1864,
+      "step": 5780
+    },
+    {
+      "epoch": 3.893745796906523,
+      "grad_norm": 0.7713287472724915,
+      "learning_rate": 1.866540014248397e-05,
+      "loss": 0.1632,
+      "step": 5790
+    },
+    {
+      "epoch": 3.9004707464694013,
+      "grad_norm": 1.1021727323532104,
+      "learning_rate": 1.8606031821420093e-05,
+      "loss": 0.1763,
+      "step": 5800
+    },
+    {
+      "epoch": 3.9071956960322796,
+      "grad_norm": 0.9041099548339844,
+      "learning_rate": 1.8546663500356208e-05,
+      "loss": 0.177,
+      "step": 5810
+    },
+    {
+      "epoch": 3.913920645595158,
+      "grad_norm": 0.607937216758728,
+      "learning_rate": 1.848729517929233e-05,
+      "loss": 0.1498,
+      "step": 5820
+    },
+    {
+      "epoch": 3.920645595158036,
+      "grad_norm": 0.8985883593559265,
+      "learning_rate": 1.842792685822845e-05,
+      "loss": 0.1786,
+      "step": 5830
+    },
+    {
+      "epoch": 3.9273705447209144,
+      "grad_norm": 0.7832037210464478,
+      "learning_rate": 1.8368558537164572e-05,
+      "loss": 0.2394,
+      "step": 5840
+    },
+    {
+      "epoch": 3.9340954942837927,
+      "grad_norm": 0.8359716534614563,
+      "learning_rate": 1.8309190216100687e-05,
+      "loss": 0.1774,
+      "step": 5850
+    },
+    {
+      "epoch": 3.940820443846671,
+      "grad_norm": 0.7045271992683411,
+      "learning_rate": 1.824982189503681e-05,
+      "loss": 0.1667,
+      "step": 5860
+    },
+    {
+      "epoch": 3.9475453934095492,
+      "grad_norm": 0.48406705260276794,
+      "learning_rate": 1.819045357397293e-05,
+      "loss": 0.1927,
+      "step": 5870
+    },
+    {
+      "epoch": 3.9542703429724275,
+      "grad_norm": 0.6424108743667603,
+      "learning_rate": 1.813108525290905e-05,
+      "loss": 0.1612,
+      "step": 5880
+    },
+    {
+      "epoch": 3.960995292535306,
+      "grad_norm": 0.9854601621627808,
+      "learning_rate": 1.8071716931845166e-05,
+      "loss": 0.1902,
+      "step": 5890
+    },
+    {
+      "epoch": 3.967720242098184,
+      "grad_norm": 0.596906304359436,
+      "learning_rate": 1.8012348610781287e-05,
+      "loss": 0.1659,
+      "step": 5900
+    },
+    {
+      "epoch": 3.9744451916610624,
+      "grad_norm": 0.5105352401733398,
+      "learning_rate": 1.795298028971741e-05,
+      "loss": 0.1498,
+      "step": 5910
+    },
+    {
+      "epoch": 3.9811701412239406,
+      "grad_norm": 0.7021915316581726,
+      "learning_rate": 1.789361196865353e-05,
+      "loss": 0.1908,
+      "step": 5920
+    },
+    {
+      "epoch": 3.987895090786819,
+      "grad_norm": 0.6894943714141846,
+      "learning_rate": 1.7834243647589645e-05,
+      "loss": 0.1869,
+      "step": 5930
+    },
+    {
+      "epoch": 3.994620040349697,
+      "grad_norm": 0.6956019997596741,
+      "learning_rate": 1.7774875326525766e-05,
+      "loss": 0.1761,
+      "step": 5940
+    },
+    {
+      "epoch": 4.0013449899125755,
+      "grad_norm": 0.7093386650085449,
+      "learning_rate": 1.7715507005461888e-05,
+      "loss": 0.1433,
+      "step": 5950
+    },
+    {
+      "epoch": 4.008069939475454,
+      "grad_norm": 0.8302282094955444,
+      "learning_rate": 1.7656138684398006e-05,
+      "loss": 0.1663,
+      "step": 5960
+    },
+    {
+      "epoch": 4.014794889038332,
+      "grad_norm": 0.7187122702598572,
+      "learning_rate": 1.7596770363334124e-05,
+      "loss": 0.1556,
+      "step": 5970
+    },
+    {
+      "epoch": 4.02151983860121,
+      "grad_norm": 0.7609972357749939,
+      "learning_rate": 1.7537402042270245e-05,
+      "loss": 0.2126,
+      "step": 5980
+    },
+    {
+      "epoch": 4.028244788164089,
+      "grad_norm": 0.7078454494476318,
+      "learning_rate": 1.7478033721206367e-05,
+      "loss": 0.178,
+      "step": 5990
+    },
+    {
+      "epoch": 4.034969737726967,
+      "grad_norm": 0.7371071577072144,
+      "learning_rate": 1.7418665400142485e-05,
+      "loss": 0.1715,
+      "step": 6000
+    },
+    {
+      "epoch": 4.034969737726967,
+      "eval_loss": 0.15210556983947754,
+      "eval_runtime": 14.4339,
+      "eval_samples_per_second": 183.18,
+      "eval_steps_per_second": 22.932,
+      "step": 6000
+    },
+    {
+      "epoch": 4.041694687289845,
+      "grad_norm": 0.9265919327735901,
+      "learning_rate": 1.7359297079078603e-05,
+      "loss": 0.1895,
+      "step": 6010
+    },
+    {
+      "epoch": 4.048419636852723,
+      "grad_norm": 0.5793859362602234,
+      "learning_rate": 1.7299928758014724e-05,
+      "loss": 0.1478,
+      "step": 6020
+    },
+    {
+      "epoch": 4.055144586415602,
+      "grad_norm": 0.622623085975647,
+      "learning_rate": 1.7240560436950846e-05,
+      "loss": 0.1858,
+      "step": 6030
+    },
+    {
+      "epoch": 4.06186953597848,
+      "grad_norm": 0.9503956437110901,
+      "learning_rate": 1.7181192115886964e-05,
+      "loss": 0.1935,
+      "step": 6040
+    },
+    {
+      "epoch": 4.068594485541358,
+      "grad_norm": 0.6633917689323425,
+      "learning_rate": 1.7121823794823082e-05,
+      "loss": 0.1561,
+      "step": 6050
+    },
+    {
+      "epoch": 4.0753194351042366,
+      "grad_norm": 0.7264837622642517,
+      "learning_rate": 1.7062455473759204e-05,
+      "loss": 0.1657,
+      "step": 6060
+    },
+    {
+      "epoch": 4.082044384667115,
+      "grad_norm": 0.739024817943573,
+      "learning_rate": 1.7003087152695325e-05,
+      "loss": 0.1573,
+      "step": 6070
+    },
+    {
+      "epoch": 4.088769334229993,
+      "grad_norm": 0.7400847673416138,
+      "learning_rate": 1.6943718831631443e-05,
+      "loss": 0.1804,
+      "step": 6080
+    },
+    {
+      "epoch": 4.095494283792871,
+      "grad_norm": 0.6965304017066956,
+      "learning_rate": 1.688435051056756e-05,
+      "loss": 0.1962,
+      "step": 6090
+    },
+    {
+      "epoch": 4.10221923335575,
+      "grad_norm": 0.6275157928466797,
+      "learning_rate": 1.6824982189503683e-05,
+      "loss": 0.1547,
+      "step": 6100
+    },
+    {
+      "epoch": 4.108944182918628,
+      "grad_norm": 0.677739679813385,
+      "learning_rate": 1.67656138684398e-05,
+      "loss": 0.1491,
+      "step": 6110
+    },
+    {
+      "epoch": 4.115669132481506,
+      "grad_norm": 0.8116381764411926,
+      "learning_rate": 1.6706245547375922e-05,
+      "loss": 0.1611,
+      "step": 6120
+    },
+    {
+      "epoch": 4.1223940820443845,
+      "grad_norm": 0.809904158115387,
+      "learning_rate": 1.664687722631204e-05,
+      "loss": 0.2239,
+      "step": 6130
+    },
+    {
+      "epoch": 4.129119031607263,
+      "grad_norm": 0.7069700956344604,
+      "learning_rate": 1.658750890524816e-05,
+      "loss": 0.1372,
+      "step": 6140
+    },
+    {
+      "epoch": 4.135843981170141,
+      "grad_norm": 0.8203169703483582,
+      "learning_rate": 1.652814058418428e-05,
+      "loss": 0.1438,
+      "step": 6150
+    },
+    {
+      "epoch": 4.142568930733019,
+      "grad_norm": 0.8196160197257996,
+      "learning_rate": 1.64687722631204e-05,
+      "loss": 0.1926,
+      "step": 6160
+    },
+    {
+      "epoch": 4.149293880295898,
+      "grad_norm": 0.8232800364494324,
+      "learning_rate": 1.640940394205652e-05,
+      "loss": 0.1669,
+      "step": 6170
+    },
+    {
+      "epoch": 4.156018829858776,
+      "grad_norm": 0.5236471891403198,
+      "learning_rate": 1.635003562099264e-05,
+      "loss": 0.1396,
+      "step": 6180
+    },
+    {
+      "epoch": 4.162743779421654,
+      "grad_norm": 0.8701181411743164,
+      "learning_rate": 1.629066729992876e-05,
+      "loss": 0.1826,
+      "step": 6190
+    },
+    {
+      "epoch": 4.1694687289845325,
+      "grad_norm": 0.7005910277366638,
+      "learning_rate": 1.6231298978864877e-05,
+      "loss": 0.1656,
+      "step": 6200
+    },
+    {
+      "epoch": 4.176193678547411,
+      "grad_norm": 0.8140985369682312,
+      "learning_rate": 1.6171930657800998e-05,
+      "loss": 0.1595,
+      "step": 6210
+    },
+    {
+      "epoch": 4.182918628110289,
+      "grad_norm": 0.7422236204147339,
+      "learning_rate": 1.6112562336737116e-05,
+      "loss": 0.158,
+      "step": 6220
+    },
+    {
+      "epoch": 4.189643577673167,
+      "grad_norm": 0.7477454543113708,
+      "learning_rate": 1.6053194015673238e-05,
+      "loss": 0.1812,
+      "step": 6230
+    },
+    {
+      "epoch": 4.196368527236046,
+      "grad_norm": 0.897112250328064,
+      "learning_rate": 1.5993825694609356e-05,
+      "loss": 0.1797,
+      "step": 6240
+    },
+    {
+      "epoch": 4.203093476798924,
+      "grad_norm": 0.5914967656135559,
+      "learning_rate": 1.5934457373545477e-05,
+      "loss": 0.1567,
+      "step": 6250
+    },
+    {
+      "epoch": 4.209818426361802,
+      "grad_norm": 0.8895167112350464,
+      "learning_rate": 1.5875089052481595e-05,
+      "loss": 0.1777,
+      "step": 6260
+    },
+    {
+      "epoch": 4.21654337592468,
+      "grad_norm": 0.9183871150016785,
+      "learning_rate": 1.5815720731417717e-05,
+      "loss": 0.1546,
+      "step": 6270
+    },
+    {
+      "epoch": 4.223268325487559,
+      "grad_norm": 0.7540081739425659,
+      "learning_rate": 1.5756352410353835e-05,
+      "loss": 0.1457,
+      "step": 6280
+    },
+    {
+      "epoch": 4.229993275050437,
+      "grad_norm": 0.7206098437309265,
+      "learning_rate": 1.5696984089289956e-05,
+      "loss": 0.1653,
+      "step": 6290
+    },
+    {
+      "epoch": 4.236718224613315,
+      "grad_norm": 0.7160133719444275,
+      "learning_rate": 1.5637615768226074e-05,
+      "loss": 0.1619,
+      "step": 6300
+    },
+    {
+      "epoch": 4.2434431741761935,
+      "grad_norm": 0.7305388450622559,
+      "learning_rate": 1.5578247447162196e-05,
+      "loss": 0.1402,
+      "step": 6310
+    },
+    {
+      "epoch": 4.250168123739072,
+      "grad_norm": 0.8916037082672119,
+      "learning_rate": 1.5518879126098314e-05,
+      "loss": 0.1731,
+      "step": 6320
+    },
+    {
+      "epoch": 4.25689307330195,
+      "grad_norm": 0.9328885078430176,
+      "learning_rate": 1.5459510805034435e-05,
+      "loss": 0.2115,
+      "step": 6330
+    },
+    {
+      "epoch": 4.263618022864828,
+      "grad_norm": 0.667841374874115,
+      "learning_rate": 1.5400142483970553e-05,
+      "loss": 0.1815,
+      "step": 6340
+    },
+    {
+      "epoch": 4.270342972427707,
+      "grad_norm": 0.7957150340080261,
+      "learning_rate": 1.5340774162906675e-05,
+      "loss": 0.1469,
+      "step": 6350
+    },
+    {
+      "epoch": 4.277067921990585,
+      "grad_norm": 0.4576648771762848,
+      "learning_rate": 1.5281405841842793e-05,
+      "loss": 0.1527,
+      "step": 6360
+    },
+    {
+      "epoch": 4.283792871553463,
+      "grad_norm": 0.6559432148933411,
+      "learning_rate": 1.5222037520778911e-05,
+      "loss": 0.1891,
+      "step": 6370
+    },
+    {
+      "epoch": 4.2905178211163415,
+      "grad_norm": 0.8562697768211365,
+      "learning_rate": 1.5162669199715032e-05,
+      "loss": 0.1845,
+      "step": 6380
+    },
+    {
+      "epoch": 4.29724277067922,
+      "grad_norm": 0.6783193945884705,
+      "learning_rate": 1.5103300878651152e-05,
+      "loss": 0.1537,
+      "step": 6390
+    },
+    {
+      "epoch": 4.303967720242098,
+      "grad_norm": 0.5136592388153076,
+      "learning_rate": 1.5043932557587274e-05,
+      "loss": 0.1328,
+      "step": 6400
+    },
+    {
+      "epoch": 4.310692669804976,
+      "grad_norm": 0.5365772843360901,
+      "learning_rate": 1.498456423652339e-05,
+      "loss": 0.1574,
+      "step": 6410
+    },
+    {
+      "epoch": 4.317417619367855,
+      "grad_norm": 0.7236724495887756,
+      "learning_rate": 1.4925195915459512e-05,
+      "loss": 0.1378,
+      "step": 6420
+    },
+    {
+      "epoch": 4.324142568930733,
+      "grad_norm": 0.6155509948730469,
+      "learning_rate": 1.4865827594395631e-05,
+      "loss": 0.1645,
+      "step": 6430
+    },
+    {
+      "epoch": 4.330867518493611,
+      "grad_norm": 1.0150562524795532,
+      "learning_rate": 1.4806459273331753e-05,
+      "loss": 0.1757,
+      "step": 6440
+    },
+    {
+      "epoch": 4.3375924680564895,
+      "grad_norm": 0.5872318148612976,
+      "learning_rate": 1.4747090952267869e-05,
+      "loss": 0.185,
+      "step": 6450
+    },
+    {
+      "epoch": 4.344317417619368,
+      "grad_norm": 0.6680309772491455,
+      "learning_rate": 1.468772263120399e-05,
+      "loss": 0.1732,
+      "step": 6460
+    },
+    {
+      "epoch": 4.351042367182246,
+      "grad_norm": 0.6384024024009705,
+      "learning_rate": 1.462835431014011e-05,
+      "loss": 0.1821,
+      "step": 6470
+    },
+    {
+      "epoch": 4.357767316745124,
+      "grad_norm": 0.8072052001953125,
+      "learning_rate": 1.456898598907623e-05,
+      "loss": 0.1928,
+      "step": 6480
+    },
+    {
+      "epoch": 4.364492266308003,
+      "grad_norm": 0.5495839715003967,
+      "learning_rate": 1.4509617668012348e-05,
+      "loss": 0.1674,
+      "step": 6490
+    },
+    {
+      "epoch": 4.371217215870881,
+      "grad_norm": 0.9522832632064819,
+      "learning_rate": 1.445024934694847e-05,
+      "loss": 0.1604,
+      "step": 6500
+    },
+    {
+      "epoch": 4.371217215870881,
+      "eval_loss": 0.1505712866783142,
+      "eval_runtime": 14.3577,
+      "eval_samples_per_second": 184.152,
+      "eval_steps_per_second": 23.054,
+      "step": 6500
+    },
+    {
+      "epoch": 4.377942165433759,
+      "grad_norm": 0.8312244415283203,
+      "learning_rate": 1.439088102588459e-05,
+      "loss": 0.1754,
+      "step": 6510
+    },
+    {
+      "epoch": 4.384667114996637,
+      "grad_norm": 0.6350352764129639,
+      "learning_rate": 1.4331512704820707e-05,
+      "loss": 0.1438,
+      "step": 6520
+    },
+    {
+      "epoch": 4.391392064559516,
+      "grad_norm": 0.8806383609771729,
+      "learning_rate": 1.4272144383756827e-05,
+      "loss": 0.1789,
+      "step": 6530
+    },
+    {
+      "epoch": 4.398117014122394,
+      "grad_norm": 0.6274639964103699,
+      "learning_rate": 1.4212776062692949e-05,
+      "loss": 0.1364,
+      "step": 6540
+    },
+    {
+      "epoch": 4.404841963685272,
+      "grad_norm": 0.6913734674453735,
+      "learning_rate": 1.4153407741629068e-05,
+      "loss": 0.1934,
+      "step": 6550
+    },
+    {
+      "epoch": 4.4115669132481505,
+      "grad_norm": 0.6295379400253296,
+      "learning_rate": 1.4094039420565186e-05,
+      "loss": 0.1629,
+      "step": 6560
+    },
+    {
+      "epoch": 4.418291862811029,
+      "grad_norm": 0.5869171023368835,
+      "learning_rate": 1.4040607931607694e-05,
+      "loss": 0.1667,
+      "step": 6570
+    },
+    {
+      "epoch": 4.425016812373907,
+      "grad_norm": 0.9569641351699829,
+      "learning_rate": 1.3981239610543814e-05,
+      "loss": 0.1467,
+      "step": 6580
+    },
+    {
+      "epoch": 4.431741761936785,
+      "grad_norm": 1.2260146141052246,
+      "learning_rate": 1.3921871289479934e-05,
+      "loss": 0.1812,
+      "step": 6590
+    },
+    {
+      "epoch": 4.438466711499664,
+      "grad_norm": 0.7992416620254517,
+      "learning_rate": 1.3862502968416055e-05,
+      "loss": 0.1325,
+      "step": 6600
+    },
+    {
+      "epoch": 4.445191661062542,
+      "grad_norm": 0.5538522601127625,
+      "learning_rate": 1.3803134647352173e-05,
+      "loss": 0.2007,
+      "step": 6610
+    },
+    {
+      "epoch": 4.45191661062542,
+      "grad_norm": 0.6790997982025146,
+      "learning_rate": 1.3743766326288293e-05,
+      "loss": 0.1648,
+      "step": 6620
+    },
+    {
+      "epoch": 4.4586415601882985,
+      "grad_norm": 0.7720995545387268,
+      "learning_rate": 1.3684398005224413e-05,
+      "loss": 0.1498,
+      "step": 6630
+    },
+    {
+      "epoch": 4.465366509751177,
+      "grad_norm": 0.7530688643455505,
+      "learning_rate": 1.3625029684160531e-05,
+      "loss": 0.1279,
+      "step": 6640
+    },
+    {
+      "epoch": 4.472091459314055,
+      "grad_norm": 0.8694820404052734,
+      "learning_rate": 1.3565661363096653e-05,
+      "loss": 0.1605,
+      "step": 6650
+    },
+    {
+      "epoch": 4.478816408876933,
+      "grad_norm": 0.6193166971206665,
+      "learning_rate": 1.3506293042032772e-05,
+      "loss": 0.195,
+      "step": 6660
+    },
+    {
+      "epoch": 4.485541358439812,
+      "grad_norm": 0.7043260931968689,
+      "learning_rate": 1.3446924720968892e-05,
+      "loss": 0.1727,
+      "step": 6670
+    },
+    {
+      "epoch": 4.49226630800269,
+      "grad_norm": 0.9062890410423279,
+      "learning_rate": 1.338755639990501e-05,
+      "loss": 0.1667,
+      "step": 6680
+    },
+    {
+      "epoch": 4.498991257565568,
+      "grad_norm": 0.7411687970161438,
+      "learning_rate": 1.332818807884113e-05,
+      "loss": 0.1906,
+      "step": 6690
+    },
+    {
+      "epoch": 4.5057162071284464,
+      "grad_norm": 0.8305112719535828,
+      "learning_rate": 1.3268819757777251e-05,
+      "loss": 0.1809,
+      "step": 6700
+    },
+    {
+      "epoch": 4.512441156691325,
+      "grad_norm": 0.6523670554161072,
+      "learning_rate": 1.3209451436713371e-05,
+      "loss": 0.2148,
+      "step": 6710
+    },
+    {
+      "epoch": 4.519166106254203,
+      "grad_norm": 0.8279176354408264,
+      "learning_rate": 1.3150083115649489e-05,
+      "loss": 0.1497,
+      "step": 6720
+    },
+    {
+      "epoch": 4.525891055817081,
+      "grad_norm": 0.4953828454017639,
+      "learning_rate": 1.3090714794585609e-05,
+      "loss": 0.1579,
+      "step": 6730
+    },
+    {
+      "epoch": 4.53261600537996,
+      "grad_norm": 0.9850824475288391,
+      "learning_rate": 1.303134647352173e-05,
+      "loss": 0.1906,
+      "step": 6740
+    },
+    {
+      "epoch": 4.539340954942838,
+      "grad_norm": 0.872776985168457,
+      "learning_rate": 1.297197815245785e-05,
+      "loss": 0.1642,
+      "step": 6750
+    },
+    {
+      "epoch": 4.546065904505716,
+      "grad_norm": 0.548650860786438,
+      "learning_rate": 1.2912609831393968e-05,
+      "loss": 0.145,
+      "step": 6760
+    },
+    {
+      "epoch": 4.552790854068594,
+      "grad_norm": 0.5439342856407166,
+      "learning_rate": 1.2853241510330088e-05,
+      "loss": 0.1569,
+      "step": 6770
+    },
+    {
+      "epoch": 4.559515803631473,
+      "grad_norm": 0.543531596660614,
+      "learning_rate": 1.279387318926621e-05,
+      "loss": 0.1542,
+      "step": 6780
+    },
+    {
+      "epoch": 4.566240753194351,
+      "grad_norm": 0.5698922276496887,
+      "learning_rate": 1.2734504868202326e-05,
+      "loss": 0.1577,
+      "step": 6790
+    },
+    {
+      "epoch": 4.572965702757229,
+      "grad_norm": 1.2096151113510132,
+      "learning_rate": 1.2675136547138447e-05,
+      "loss": 0.1717,
+      "step": 6800
+    },
+    {
+      "epoch": 4.5796906523201075,
+      "grad_norm": 0.664527416229248,
+      "learning_rate": 1.2615768226074567e-05,
+      "loss": 0.1918,
+      "step": 6810
+    },
+    {
+      "epoch": 4.586415601882986,
+      "grad_norm": 1.1124138832092285,
+      "learning_rate": 1.2556399905010688e-05,
+      "loss": 0.1684,
+      "step": 6820
+    },
+    {
+      "epoch": 4.593140551445864,
+      "grad_norm": 0.7068644165992737,
+      "learning_rate": 1.2497031583946807e-05,
+      "loss": 0.1421,
+      "step": 6830
+    },
+    {
+      "epoch": 4.599865501008742,
+      "grad_norm": 0.8015205264091492,
+      "learning_rate": 1.2437663262882926e-05,
+      "loss": 0.1561,
+      "step": 6840
+    },
+    {
+      "epoch": 4.606590450571621,
+      "grad_norm": 0.7207475304603577,
+      "learning_rate": 1.2378294941819046e-05,
+      "loss": 0.156,
+      "step": 6850
+    },
+    {
+      "epoch": 4.613315400134499,
+      "grad_norm": 0.8778402209281921,
+      "learning_rate": 1.2318926620755166e-05,
+      "loss": 0.17,
+      "step": 6860
+    },
+    {
+      "epoch": 4.620040349697377,
+      "grad_norm": 0.8167802095413208,
+      "learning_rate": 1.2259558299691286e-05,
+      "loss": 0.188,
+      "step": 6870
+    },
+    {
+      "epoch": 4.6267652992602555,
+      "grad_norm": 0.6325931549072266,
+      "learning_rate": 1.2200189978627405e-05,
+      "loss": 0.1491,
+      "step": 6880
+    },
+    {
+      "epoch": 4.633490248823134,
+      "grad_norm": 0.8713521957397461,
+      "learning_rate": 1.2140821657563525e-05,
+      "loss": 0.1405,
+      "step": 6890
+    },
+    {
+      "epoch": 4.640215198386012,
+      "grad_norm": 0.617329478263855,
+      "learning_rate": 1.2081453336499645e-05,
+      "loss": 0.1627,
+      "step": 6900
+    },
+    {
+      "epoch": 4.64694014794889,
+      "grad_norm": 0.683966338634491,
+      "learning_rate": 1.2022085015435763e-05,
+      "loss": 0.1395,
+      "step": 6910
+    },
+    {
+      "epoch": 4.653665097511769,
+      "grad_norm": 0.7148402333259583,
+      "learning_rate": 1.1962716694371884e-05,
+      "loss": 0.1865,
+      "step": 6920
+    },
+    {
+      "epoch": 4.660390047074647,
+      "grad_norm": 0.616771399974823,
+      "learning_rate": 1.1903348373308002e-05,
+      "loss": 0.1566,
+      "step": 6930
+    },
+    {
+      "epoch": 4.667114996637525,
+      "grad_norm": 0.6236411333084106,
+      "learning_rate": 1.1843980052244124e-05,
+      "loss": 0.1986,
+      "step": 6940
+    },
+    {
+      "epoch": 4.673839946200403,
+      "grad_norm": 0.8944858312606812,
+      "learning_rate": 1.1784611731180242e-05,
+      "loss": 0.1612,
+      "step": 6950
+    },
+    {
+      "epoch": 4.680564895763282,
+      "grad_norm": 0.7702111601829529,
+      "learning_rate": 1.1725243410116363e-05,
+      "loss": 0.1892,
+      "step": 6960
+    },
+    {
+      "epoch": 4.68728984532616,
+      "grad_norm": 0.6784073710441589,
+      "learning_rate": 1.1665875089052481e-05,
+      "loss": 0.2,
+      "step": 6970
+    },
+    {
+      "epoch": 4.694014794889038,
+      "grad_norm": 1.0141595602035522,
+      "learning_rate": 1.1606506767988603e-05,
+      "loss": 0.1485,
+      "step": 6980
+    },
+    {
+      "epoch": 4.700739744451917,
+      "grad_norm": 0.7536572813987732,
+      "learning_rate": 1.1547138446924721e-05,
+      "loss": 0.1602,
+      "step": 6990
+    },
+    {
+      "epoch": 4.707464694014795,
+      "grad_norm": 0.7322705984115601,
+      "learning_rate": 1.1487770125860842e-05,
+      "loss": 0.1422,
+      "step": 7000
+    },
+    {
+      "epoch": 4.707464694014795,
+      "eval_loss": 0.1484505832195282,
+      "eval_runtime": 14.4562,
+      "eval_samples_per_second": 182.897,
+      "eval_steps_per_second": 22.897,
+      "step": 7000
+    },
+    {
+      "epoch": 4.714189643577673,
+      "grad_norm": 0.7850281000137329,
+      "learning_rate": 1.142840180479696e-05,
+      "loss": 0.1565,
+      "step": 7010
+    },
+    {
+      "epoch": 4.720914593140551,
+      "grad_norm": 0.656444787979126,
+      "learning_rate": 1.1369033483733082e-05,
+      "loss": 0.1388,
+      "step": 7020
+    },
+    {
+      "epoch": 4.72763954270343,
+      "grad_norm": 0.8533453345298767,
+      "learning_rate": 1.13096651626692e-05,
+      "loss": 0.2173,
+      "step": 7030
+    },
+    {
+      "epoch": 4.734364492266308,
+      "grad_norm": 0.5954863429069519,
+      "learning_rate": 1.125029684160532e-05,
+      "loss": 0.1494,
+      "step": 7040
+    },
+    {
+      "epoch": 4.741089441829186,
+      "grad_norm": 0.7292872667312622,
+      "learning_rate": 1.119092852054144e-05,
+      "loss": 0.1808,
+      "step": 7050
+    },
+    {
+      "epoch": 4.7478143913920645,
+      "grad_norm": 0.5989605784416199,
+      "learning_rate": 1.113156019947756e-05,
+      "loss": 0.1474,
+      "step": 7060
+    },
+    {
+      "epoch": 4.754539340954943,
+      "grad_norm": 0.870201826095581,
+      "learning_rate": 1.1072191878413679e-05,
+      "loss": 0.1767,
+      "step": 7070
+    },
+    {
+      "epoch": 4.761264290517821,
+      "grad_norm": 0.7545794248580933,
+      "learning_rate": 1.1012823557349799e-05,
+      "loss": 0.1726,
+      "step": 7080
+    },
+    {
+      "epoch": 4.767989240080699,
+      "grad_norm": 0.5619848370552063,
+      "learning_rate": 1.0953455236285919e-05,
+      "loss": 0.154,
+      "step": 7090
+    },
+    {
+      "epoch": 4.774714189643578,
+      "grad_norm": 0.6141572594642639,
+      "learning_rate": 1.0894086915222038e-05,
+      "loss": 0.1506,
+      "step": 7100
+    },
+    {
+      "epoch": 4.781439139206456,
+      "grad_norm": 0.5468729734420776,
+      "learning_rate": 1.0834718594158158e-05,
+      "loss": 0.1809,
+      "step": 7110
+    },
+    {
+      "epoch": 4.788164088769334,
+      "grad_norm": 0.7549372315406799,
+      "learning_rate": 1.0775350273094278e-05,
+      "loss": 0.1434,
+      "step": 7120
+    },
+    {
+      "epoch": 4.7948890383322125,
+      "grad_norm": 0.5954113602638245,
+      "learning_rate": 1.0715981952030398e-05,
+      "loss": 0.1773,
+      "step": 7130
+    },
+    {
+      "epoch": 4.801613987895091,
+      "grad_norm": 0.9040542840957642,
+      "learning_rate": 1.0656613630966517e-05,
+      "loss": 0.1802,
+      "step": 7140
+    },
+    {
+      "epoch": 4.808338937457969,
+      "grad_norm": 0.8996601700782776,
+      "learning_rate": 1.0597245309902637e-05,
+      "loss": 0.1771,
+      "step": 7150
+    },
+    {
+      "epoch": 4.815063887020847,
+      "grad_norm": 0.5393683910369873,
+      "learning_rate": 1.0537876988838755e-05,
+      "loss": 0.1463,
+      "step": 7160
+    },
+    {
+      "epoch": 4.821788836583726,
+      "grad_norm": 0.5825424194335938,
+      "learning_rate": 1.0478508667774877e-05,
+      "loss": 0.17,
+      "step": 7170
+    },
+    {
+      "epoch": 4.828513786146604,
+      "grad_norm": 0.7498705983161926,
+      "learning_rate": 1.0419140346710995e-05,
+      "loss": 0.1753,
+      "step": 7180
+    },
+    {
+      "epoch": 4.835238735709482,
+      "grad_norm": 0.6035088300704956,
+      "learning_rate": 1.0359772025647114e-05,
+      "loss": 0.1459,
+      "step": 7190
+    },
+    {
+      "epoch": 4.84196368527236,
+      "grad_norm": 0.7723816633224487,
+      "learning_rate": 1.0300403704583234e-05,
+      "loss": 0.1585,
+      "step": 7200
+    },
+    {
+      "epoch": 4.848688634835239,
+      "grad_norm": 0.6356052160263062,
+      "learning_rate": 1.0241035383519354e-05,
+      "loss": 0.1658,
+      "step": 7210
+    },
+    {
+      "epoch": 4.855413584398117,
+      "grad_norm": 0.9170238971710205,
+      "learning_rate": 1.0181667062455474e-05,
+      "loss": 0.1738,
+      "step": 7220
+    },
+    {
+      "epoch": 4.862138533960995,
+      "grad_norm": 0.5520994067192078,
+      "learning_rate": 1.0122298741391594e-05,
+      "loss": 0.1839,
+      "step": 7230
+    },
+    {
+      "epoch": 4.8688634835238735,
+      "grad_norm": 0.6071659326553345,
+      "learning_rate": 1.0062930420327713e-05,
+      "loss": 0.1802,
+      "step": 7240
+    },
+    {
+      "epoch": 4.875588433086752,
+      "grad_norm": 0.7233278751373291,
+      "learning_rate": 1.0003562099263833e-05,
+      "loss": 0.1741,
+      "step": 7250
+    },
+    {
+      "epoch": 4.88231338264963,
+      "grad_norm": 0.7937821745872498,
+      "learning_rate": 9.944193778199953e-06,
+      "loss": 0.1784,
+      "step": 7260
+    },
+    {
+      "epoch": 4.889038332212508,
+      "grad_norm": 0.5812748670578003,
+      "learning_rate": 9.884825457136073e-06,
+      "loss": 0.1916,
+      "step": 7270
+    },
+    {
+      "epoch": 4.895763281775387,
+      "grad_norm": 0.5226556062698364,
+      "learning_rate": 9.825457136072192e-06,
+      "loss": 0.1482,
+      "step": 7280
+    },
+    {
+      "epoch": 4.902488231338265,
+      "grad_norm": 0.7629415392875671,
+      "learning_rate": 9.766088815008312e-06,
+      "loss": 0.2192,
+      "step": 7290
+    },
+    {
+      "epoch": 4.909213180901143,
+      "grad_norm": 0.7357354760169983,
+      "learning_rate": 9.706720493944432e-06,
+      "loss": 0.1695,
+      "step": 7300
+    },
+    {
+      "epoch": 4.9159381304640215,
+      "grad_norm": 0.6955385804176331,
+      "learning_rate": 9.647352172880552e-06,
+      "loss": 0.1841,
+      "step": 7310
+    },
+    {
+      "epoch": 4.9226630800269,
+      "grad_norm": 0.803431510925293,
+      "learning_rate": 9.587983851816671e-06,
+      "loss": 0.1883,
+      "step": 7320
+    },
+    {
+      "epoch": 4.929388029589778,
+      "grad_norm": 0.5103302001953125,
+      "learning_rate": 9.528615530752791e-06,
+      "loss": 0.1581,
+      "step": 7330
+    },
+    {
+      "epoch": 4.936112979152656,
+      "grad_norm": 0.7879159450531006,
+      "learning_rate": 9.46924720968891e-06,
+      "loss": 0.1355,
+      "step": 7340
+    },
+    {
+      "epoch": 4.942837928715535,
+      "grad_norm": 0.6451155543327332,
+      "learning_rate": 9.40987888862503e-06,
+      "loss": 0.1633,
+      "step": 7350
+    },
+    {
+      "epoch": 4.949562878278413,
+      "grad_norm": 0.8154376745223999,
+      "learning_rate": 9.350510567561149e-06,
+      "loss": 0.1468,
+      "step": 7360
+    },
+    {
+      "epoch": 4.956287827841291,
+      "grad_norm": 0.6419748663902283,
+      "learning_rate": 9.29114224649727e-06,
+      "loss": 0.178,
+      "step": 7370
+    },
+    {
+      "epoch": 4.9630127774041695,
+      "grad_norm": 0.8760988116264343,
+      "learning_rate": 9.231773925433388e-06,
+      "loss": 0.1849,
+      "step": 7380
+    },
+    {
+      "epoch": 4.969737726967048,
+      "grad_norm": 0.8139458894729614,
+      "learning_rate": 9.17240560436951e-06,
+      "loss": 0.149,
+      "step": 7390
+    },
+    {
+      "epoch": 4.976462676529926,
+      "grad_norm": 0.565104603767395,
+      "learning_rate": 9.113037283305628e-06,
+      "loss": 0.1271,
+      "step": 7400
+    },
+    {
+      "epoch": 4.983187626092804,
+      "grad_norm": 0.8110491037368774,
+      "learning_rate": 9.05366896224175e-06,
+      "loss": 0.1803,
+      "step": 7410
+    },
+    {
+      "epoch": 4.989912575655683,
+      "grad_norm": 1.0213757753372192,
+      "learning_rate": 8.994300641177867e-06,
+      "loss": 0.1923,
+      "step": 7420
+    },
+    {
+      "epoch": 4.996637525218561,
+      "grad_norm": 0.7190690636634827,
+      "learning_rate": 8.934932320113989e-06,
+      "loss": 0.1763,
+      "step": 7430
+    },
+    {
+      "epoch": 5.003362474781439,
+      "grad_norm": 0.8246272206306458,
+      "learning_rate": 8.875563999050107e-06,
+      "loss": 0.1486,
+      "step": 7440
+    },
+    {
+      "epoch": 5.010087424344317,
+      "grad_norm": 0.5950284600257874,
+      "learning_rate": 8.816195677986228e-06,
+      "loss": 0.2139,
+      "step": 7450
+    },
+    {
+      "epoch": 5.016812373907196,
+      "grad_norm": 0.7800911664962769,
+      "learning_rate": 8.756827356922346e-06,
+      "loss": 0.1364,
+      "step": 7460
+    },
+    {
+      "epoch": 5.023537323470074,
+      "grad_norm": 0.8360269069671631,
+      "learning_rate": 8.697459035858466e-06,
+      "loss": 0.1636,
+      "step": 7470
+    },
+    {
+      "epoch": 5.030262273032952,
+      "grad_norm": 0.5287578105926514,
+      "learning_rate": 8.638090714794586e-06,
+      "loss": 0.1711,
+      "step": 7480
+    },
+    {
+      "epoch": 5.0369872225958305,
+      "grad_norm": 0.907789945602417,
+      "learning_rate": 8.578722393730706e-06,
+      "loss": 0.176,
+      "step": 7490
+    },
+    {
+      "epoch": 5.043712172158709,
+      "grad_norm": 0.6482927799224854,
+      "learning_rate": 8.519354072666825e-06,
+      "loss": 0.1526,
+      "step": 7500
+    },
+    {
+      "epoch": 5.043712172158709,
+      "eval_loss": 0.14650234580039978,
+      "eval_runtime": 14.2858,
+      "eval_samples_per_second": 185.079,
+      "eval_steps_per_second": 23.17,
+      "step": 7500
+    },
+    {
+      "epoch": 5.050437121721587,
+      "grad_norm": 0.7564139366149902,
+      "learning_rate": 8.459985751602945e-06,
+      "loss": 0.178,
+      "step": 7510
+    },
+    {
+      "epoch": 5.057162071284465,
+      "grad_norm": 0.7940858602523804,
+      "learning_rate": 8.400617430539065e-06,
+      "loss": 0.1412,
+      "step": 7520
+    },
+    {
+      "epoch": 5.063887020847344,
+      "grad_norm": 0.9502273201942444,
+      "learning_rate": 8.341249109475185e-06,
+      "loss": 0.1552,
+      "step": 7530
+    },
+    {
+      "epoch": 5.070611970410222,
+      "grad_norm": 0.530156135559082,
+      "learning_rate": 8.281880788411304e-06,
+      "loss": 0.1297,
+      "step": 7540
+    },
+    {
+      "epoch": 5.0773369199731,
+      "grad_norm": 0.632544755935669,
+      "learning_rate": 8.222512467347424e-06,
+      "loss": 0.1596,
+      "step": 7550
+    },
+    {
+      "epoch": 5.0840618695359785,
+      "grad_norm": 0.6239969730377197,
+      "learning_rate": 8.163144146283544e-06,
+      "loss": 0.1603,
+      "step": 7560
+    },
+    {
+      "epoch": 5.090786819098857,
+      "grad_norm": 0.6600353121757507,
+      "learning_rate": 8.103775825219664e-06,
+      "loss": 0.1512,
+      "step": 7570
+    },
+    {
+      "epoch": 5.097511768661735,
+      "grad_norm": 0.6660837531089783,
+      "learning_rate": 8.044407504155783e-06,
+      "loss": 0.1666,
+      "step": 7580
+    },
+    {
+      "epoch": 5.104236718224613,
+      "grad_norm": 0.5615241527557373,
+      "learning_rate": 7.985039183091903e-06,
+      "loss": 0.1626,
+      "step": 7590
+    },
+    {
+      "epoch": 5.110961667787492,
+      "grad_norm": 0.8264844417572021,
+      "learning_rate": 7.925670862028023e-06,
+      "loss": 0.1772,
+      "step": 7600
+    },
+    {
+      "epoch": 5.11768661735037,
+      "grad_norm": 0.8056089282035828,
+      "learning_rate": 7.866302540964143e-06,
+      "loss": 0.2064,
+      "step": 7610
+    },
+    {
+      "epoch": 5.124411566913248,
+      "grad_norm": 0.6176988482475281,
+      "learning_rate": 7.80693421990026e-06,
+      "loss": 0.162,
+      "step": 7620
+    },
+    {
+      "epoch": 5.1311365164761265,
+      "grad_norm": 0.6516183018684387,
+      "learning_rate": 7.74756589883638e-06,
+      "loss": 0.1632,
+      "step": 7630
+    },
+    {
+      "epoch": 5.137861466039005,
+      "grad_norm": 0.7382626533508301,
+      "learning_rate": 7.6881975777725e-06,
+      "loss": 0.1617,
+      "step": 7640
+    },
+    {
+      "epoch": 5.144586415601883,
+      "grad_norm": 0.7340312600135803,
+      "learning_rate": 7.628829256708621e-06,
+      "loss": 0.1519,
+      "step": 7650
+    },
+    {
+      "epoch": 5.151311365164761,
+      "grad_norm": 0.700930655002594,
+      "learning_rate": 7.56946093564474e-06,
+      "loss": 0.1409,
+      "step": 7660
+    },
+    {
+      "epoch": 5.15803631472764,
+      "grad_norm": 0.8751327395439148,
+      "learning_rate": 7.5100926145808605e-06,
+      "loss": 0.1915,
+      "step": 7670
+    },
+    {
+      "epoch": 5.164761264290518,
+      "grad_norm": 0.5034623146057129,
+      "learning_rate": 7.450724293516979e-06,
+      "loss": 0.1334,
+      "step": 7680
+    },
+    {
+      "epoch": 5.171486213853396,
+      "grad_norm": 0.7203763127326965,
+      "learning_rate": 7.3913559724531e-06,
+      "loss": 0.1682,
+      "step": 7690
+    },
+    {
+      "epoch": 5.178211163416274,
+      "grad_norm": 0.6495651006698608,
+      "learning_rate": 7.331987651389219e-06,
+      "loss": 0.1692,
+      "step": 7700
+    },
+    {
+      "epoch": 5.184936112979153,
+      "grad_norm": 0.6066912412643433,
+      "learning_rate": 7.2726193303253395e-06,
+      "loss": 0.1388,
+      "step": 7710
+    },
+    {
+      "epoch": 5.191661062542031,
+      "grad_norm": 0.6121117472648621,
+      "learning_rate": 7.213251009261458e-06,
+      "loss": 0.2234,
+      "step": 7720
+    },
+    {
+      "epoch": 5.198386012104909,
+      "grad_norm": 0.5731104612350464,
+      "learning_rate": 7.153882688197579e-06,
+      "loss": 0.1626,
+      "step": 7730
+    },
+    {
+      "epoch": 5.2051109616677875,
+      "grad_norm": 0.7278406023979187,
+      "learning_rate": 7.094514367133698e-06,
+      "loss": 0.1439,
+      "step": 7740
+    },
+    {
+      "epoch": 5.211835911230666,
+      "grad_norm": 0.7343350648880005,
+      "learning_rate": 7.035146046069817e-06,
+      "loss": 0.1619,
+      "step": 7750
+    },
+    {
+      "epoch": 5.218560860793544,
+      "grad_norm": 0.7979176640510559,
+      "learning_rate": 6.9757777250059375e-06,
+      "loss": 0.1873,
+      "step": 7760
+    },
+    {
+      "epoch": 5.225285810356422,
+      "grad_norm": 0.7361353039741516,
+      "learning_rate": 6.916409403942056e-06,
+      "loss": 0.1744,
+      "step": 7770
+    },
+    {
+      "epoch": 5.232010759919301,
+      "grad_norm": 1.0325379371643066,
+      "learning_rate": 6.857041082878177e-06,
+      "loss": 0.1671,
+      "step": 7780
+    },
+    {
+      "epoch": 5.238735709482179,
+      "grad_norm": 0.5549724102020264,
+      "learning_rate": 6.797672761814296e-06,
+      "loss": 0.1389,
+      "step": 7790
+    },
+    {
+      "epoch": 5.245460659045057,
+      "grad_norm": 0.6693004965782166,
+      "learning_rate": 6.7383044407504165e-06,
+      "loss": 0.1227,
+      "step": 7800
+    },
+    {
+      "epoch": 5.2521856086079355,
+      "grad_norm": 0.6957047581672668,
+      "learning_rate": 6.678936119686535e-06,
+      "loss": 0.142,
+      "step": 7810
+    },
+    {
+      "epoch": 5.258910558170814,
+      "grad_norm": 0.5779728889465332,
+      "learning_rate": 6.619567798622656e-06,
+      "loss": 0.1951,
+      "step": 7820
+    },
+    {
+      "epoch": 5.265635507733692,
+      "grad_norm": 0.5068587064743042,
+      "learning_rate": 6.560199477558775e-06,
+      "loss": 0.1603,
+      "step": 7830
+    },
+    {
+      "epoch": 5.27236045729657,
+      "grad_norm": 0.8560962080955505,
+      "learning_rate": 6.500831156494895e-06,
+      "loss": 0.1657,
+      "step": 7840
+    },
+    {
+      "epoch": 5.279085406859449,
+      "grad_norm": 0.9463856816291809,
+      "learning_rate": 6.4414628354310145e-06,
+      "loss": 0.1372,
+      "step": 7850
+    },
+    {
+      "epoch": 5.285810356422327,
+      "grad_norm": 0.5416259765625,
+      "learning_rate": 6.382094514367134e-06,
+      "loss": 0.1368,
+      "step": 7860
+    },
+    {
+      "epoch": 5.292535305985205,
+      "grad_norm": 0.35940733551979065,
+      "learning_rate": 6.322726193303254e-06,
+      "loss": 0.1313,
+      "step": 7870
+    },
+    {
+      "epoch": 5.299260255548083,
+      "grad_norm": 0.5176042914390564,
+      "learning_rate": 6.263357872239374e-06,
+      "loss": 0.1549,
+      "step": 7880
+    },
+    {
+      "epoch": 5.305985205110962,
+      "grad_norm": 0.8194878101348877,
+      "learning_rate": 6.2039895511754935e-06,
+      "loss": 0.1332,
+      "step": 7890
+    },
+    {
+      "epoch": 5.31271015467384,
+      "grad_norm": 0.718500018119812,
+      "learning_rate": 6.144621230111612e-06,
+      "loss": 0.1632,
+      "step": 7900
+    },
+    {
+      "epoch": 5.319435104236718,
+      "grad_norm": 0.6662378907203674,
+      "learning_rate": 6.085252909047732e-06,
+      "loss": 0.1731,
+      "step": 7910
+    },
+    {
+      "epoch": 5.326160053799597,
+      "grad_norm": 0.5502232313156128,
+      "learning_rate": 6.025884587983852e-06,
+      "loss": 0.179,
+      "step": 7920
+    },
+    {
+      "epoch": 5.332885003362475,
+      "grad_norm": 0.7982697486877441,
+      "learning_rate": 5.966516266919972e-06,
+      "loss": 0.1705,
+      "step": 7930
+    },
+    {
+      "epoch": 5.339609952925353,
+      "grad_norm": 0.6752830147743225,
+      "learning_rate": 5.9071479458560915e-06,
+      "loss": 0.1539,
+      "step": 7940
+    },
+    {
+      "epoch": 5.346334902488231,
+      "grad_norm": 0.7998142242431641,
+      "learning_rate": 5.847779624792211e-06,
+      "loss": 0.1536,
+      "step": 7950
+    },
+    {
+      "epoch": 5.35305985205111,
+      "grad_norm": 0.8491772413253784,
+      "learning_rate": 5.78841130372833e-06,
+      "loss": 0.1715,
+      "step": 7960
+    },
+    {
+      "epoch": 5.359784801613988,
+      "grad_norm": 0.6874935626983643,
+      "learning_rate": 5.72904298266445e-06,
+      "loss": 0.1375,
+      "step": 7970
+    },
+    {
+      "epoch": 5.366509751176866,
+      "grad_norm": 0.5897092223167419,
+      "learning_rate": 5.66967466160057e-06,
+      "loss": 0.1329,
+      "step": 7980
+    },
+    {
+      "epoch": 5.3732347007397445,
+      "grad_norm": 0.6634560227394104,
+      "learning_rate": 5.610306340536689e-06,
+      "loss": 0.1531,
+      "step": 7990
+    },
+    {
+      "epoch": 5.379959650302623,
+      "grad_norm": 0.7213748097419739,
+      "learning_rate": 5.550938019472809e-06,
+      "loss": 0.1606,
+      "step": 8000
+    },
+    {
+      "epoch": 5.379959650302623,
+      "eval_loss": 0.1459866315126419,
+      "eval_runtime": 14.4136,
+      "eval_samples_per_second": 183.437,
+      "eval_steps_per_second": 22.964,
+      "step": 8000
+    },
+    {
+      "epoch": 5.386684599865501,
+      "grad_norm": 0.7976545095443726,
+      "learning_rate": 5.491569698408929e-06,
+      "loss": 0.1638,
+      "step": 8010
+    },
+    {
+      "epoch": 5.393409549428379,
+      "grad_norm": 0.6053023338317871,
+      "learning_rate": 5.432201377345049e-06,
+      "loss": 0.1468,
+      "step": 8020
+    },
+    {
+      "epoch": 5.400134498991258,
+      "grad_norm": 0.6205041408538818,
+      "learning_rate": 5.3728330562811684e-06,
+      "loss": 0.1819,
+      "step": 8030
+    },
+    {
+      "epoch": 5.406859448554136,
+      "grad_norm": 0.8174419403076172,
+      "learning_rate": 5.313464735217288e-06,
+      "loss": 0.1687,
+      "step": 8040
+    },
+    {
+      "epoch": 5.413584398117014,
+      "grad_norm": 0.7457543611526489,
+      "learning_rate": 5.254096414153408e-06,
+      "loss": 0.1437,
+      "step": 8050
+    },
+    {
+      "epoch": 5.4203093476798925,
+      "grad_norm": 0.4972638487815857,
+      "learning_rate": 5.194728093089528e-06,
+      "loss": 0.2305,
+      "step": 8060
+    },
+    {
+      "epoch": 5.427034297242771,
+      "grad_norm": 0.49321138858795166,
+      "learning_rate": 5.1353597720256475e-06,
+      "loss": 0.1395,
+      "step": 8070
+    },
+    {
+      "epoch": 5.433759246805649,
+      "grad_norm": 0.6880173683166504,
+      "learning_rate": 5.075991450961767e-06,
+      "loss": 0.1116,
+      "step": 8080
+    },
+    {
+      "epoch": 5.440484196368527,
+      "grad_norm": 0.6798646450042725,
+      "learning_rate": 5.016623129897887e-06,
+      "loss": 0.1767,
+      "step": 8090
+    },
+    {
+      "epoch": 5.447209145931406,
+      "grad_norm": 0.5538681745529175,
+      "learning_rate": 4.957254808834006e-06,
+      "loss": 0.1596,
+      "step": 8100
+    },
+    {
+      "epoch": 5.453934095494284,
+      "grad_norm": 0.5008218884468079,
+      "learning_rate": 4.897886487770126e-06,
+      "loss": 0.1554,
+      "step": 8110
+    },
+    {
+      "epoch": 5.460659045057162,
+      "grad_norm": 0.6006589531898499,
+      "learning_rate": 4.8385181667062454e-06,
+      "loss": 0.1496,
+      "step": 8120
+    },
+    {
+      "epoch": 5.46738399462004,
+      "grad_norm": 0.7322587966918945,
+      "learning_rate": 4.779149845642365e-06,
+      "loss": 0.1592,
+      "step": 8130
+    },
+    {
+      "epoch": 5.474108944182919,
+      "grad_norm": 0.7133284211158752,
+      "learning_rate": 4.719781524578485e-06,
+      "loss": 0.128,
+      "step": 8140
+    },
+    {
+      "epoch": 5.480833893745797,
+      "grad_norm": 0.5079299807548523,
+      "learning_rate": 4.660413203514605e-06,
+      "loss": 0.1412,
+      "step": 8150
+    },
+    {
+      "epoch": 5.487558843308675,
+      "grad_norm": 0.4279893636703491,
+      "learning_rate": 4.6010448824507245e-06,
+      "loss": 0.1681,
+      "step": 8160
+    },
+    {
+      "epoch": 5.4942837928715536,
+      "grad_norm": 0.6305129528045654,
+      "learning_rate": 4.541676561386844e-06,
+      "loss": 0.1574,
+      "step": 8170
+    },
+    {
+      "epoch": 5.501008742434432,
+      "grad_norm": 0.6287931203842163,
+      "learning_rate": 4.482308240322964e-06,
+      "loss": 0.1691,
+      "step": 8180
+    },
+    {
+      "epoch": 5.50773369199731,
+      "grad_norm": 0.6833222508430481,
+      "learning_rate": 4.422939919259084e-06,
+      "loss": 0.1763,
+      "step": 8190
+    },
+    {
+      "epoch": 5.514458641560188,
+      "grad_norm": 0.621030867099762,
+      "learning_rate": 4.3635715981952035e-06,
+      "loss": 0.1394,
+      "step": 8200
+    },
+    {
+      "epoch": 5.521183591123067,
+      "grad_norm": 0.9984930753707886,
+      "learning_rate": 4.304203277131323e-06,
+      "loss": 0.1822,
+      "step": 8210
+    },
+    {
+      "epoch": 5.527908540685945,
+      "grad_norm": 0.7831975221633911,
+      "learning_rate": 4.244834956067443e-06,
+      "loss": 0.1812,
+      "step": 8220
+    },
+    {
+      "epoch": 5.534633490248823,
+      "grad_norm": 0.6752682328224182,
+      "learning_rate": 4.185466635003563e-06,
+      "loss": 0.1622,
+      "step": 8230
+    },
+    {
+      "epoch": 5.5413584398117015,
+      "grad_norm": 0.6233469843864441,
+      "learning_rate": 4.126098313939682e-06,
+      "loss": 0.1404,
+      "step": 8240
+    },
+    {
+      "epoch": 5.54808338937458,
+      "grad_norm": 0.825937032699585,
+      "learning_rate": 4.0667299928758015e-06,
+      "loss": 0.1266,
+      "step": 8250
+    },
+    {
+      "epoch": 5.554808338937458,
+      "grad_norm": 0.587255597114563,
+      "learning_rate": 4.007361671811921e-06,
+      "loss": 0.1724,
+      "step": 8260
+    },
+    {
+      "epoch": 5.561533288500336,
+      "grad_norm": 0.7307100892066956,
+      "learning_rate": 3.947993350748041e-06,
+      "loss": 0.1809,
+      "step": 8270
+    },
+    {
+      "epoch": 5.568258238063215,
+      "grad_norm": 0.6363770365715027,
+      "learning_rate": 3.888625029684161e-06,
+      "loss": 0.1444,
+      "step": 8280
+    },
+    {
+      "epoch": 5.574983187626093,
+      "grad_norm": 0.7173252701759338,
+      "learning_rate": 3.8292567086202805e-06,
+      "loss": 0.159,
+      "step": 8290
+    },
+    {
+      "epoch": 5.581708137188971,
+      "grad_norm": 0.6081172227859497,
+      "learning_rate": 3.7698883875564e-06,
+      "loss": 0.1433,
+      "step": 8300
+    },
+    {
+      "epoch": 5.5884330867518495,
+      "grad_norm": 1.3353235721588135,
+      "learning_rate": 3.7105200664925196e-06,
+      "loss": 0.1883,
+      "step": 8310
+    },
+    {
+      "epoch": 5.595158036314728,
+      "grad_norm": 0.4842171370983124,
+      "learning_rate": 3.6511517454286394e-06,
+      "loss": 0.1457,
+      "step": 8320
+    },
+    {
+      "epoch": 5.601882985877606,
+      "grad_norm": 0.8106813430786133,
+      "learning_rate": 3.591783424364759e-06,
+      "loss": 0.1681,
+      "step": 8330
+    },
+    {
+      "epoch": 5.608607935440484,
+      "grad_norm": 0.838200032711029,
+      "learning_rate": 3.532415103300879e-06,
+      "loss": 0.1572,
+      "step": 8340
+    },
+    {
+      "epoch": 5.615332885003363,
+      "grad_norm": 0.42966827750205994,
+      "learning_rate": 3.4730467822369987e-06,
+      "loss": 0.1395,
+      "step": 8350
+    },
+    {
+      "epoch": 5.622057834566241,
+      "grad_norm": 0.806328296661377,
+      "learning_rate": 3.4136784611731184e-06,
+      "loss": 0.2106,
+      "step": 8360
+    },
+    {
+      "epoch": 5.628782784129119,
+      "grad_norm": 0.6712527275085449,
+      "learning_rate": 3.354310140109238e-06,
+      "loss": 0.1519,
+      "step": 8370
+    },
+    {
+      "epoch": 5.635507733691997,
+      "grad_norm": 0.708582878112793,
+      "learning_rate": 3.294941819045357e-06,
+      "loss": 0.1394,
+      "step": 8380
+    },
+    {
+      "epoch": 5.642232683254876,
+      "grad_norm": 0.6497735977172852,
+      "learning_rate": 3.235573497981477e-06,
+      "loss": 0.1647,
+      "step": 8390
+    },
+    {
+      "epoch": 5.648957632817754,
+      "grad_norm": 0.6724647283554077,
+      "learning_rate": 3.1762051769175966e-06,
+      "loss": 0.1555,
+      "step": 8400
+    },
+    {
+      "epoch": 5.655682582380632,
+      "grad_norm": 0.7497691512107849,
+      "learning_rate": 3.116836855853717e-06,
+      "loss": 0.1468,
+      "step": 8410
+    },
+    {
+      "epoch": 5.6624075319435105,
+      "grad_norm": 0.6995181441307068,
+      "learning_rate": 3.057468534789836e-06,
+      "loss": 0.1491,
+      "step": 8420
+    },
+    {
+      "epoch": 5.669132481506389,
+      "grad_norm": 0.6918231248855591,
+      "learning_rate": 2.998100213725956e-06,
+      "loss": 0.1567,
+      "step": 8430
+    },
+    {
+      "epoch": 5.675857431069267,
+      "grad_norm": 0.700425386428833,
+      "learning_rate": 2.9387318926620757e-06,
+      "loss": 0.1493,
+      "step": 8440
+    },
+    {
+      "epoch": 5.682582380632145,
+      "grad_norm": 0.5970633029937744,
+      "learning_rate": 2.8793635715981954e-06,
+      "loss": 0.1779,
+      "step": 8450
+    },
+    {
+      "epoch": 5.689307330195024,
+      "grad_norm": 0.703995943069458,
+      "learning_rate": 2.819995250534315e-06,
+      "loss": 0.1482,
+      "step": 8460
+    },
+    {
+      "epoch": 5.696032279757902,
+      "grad_norm": 0.624579906463623,
+      "learning_rate": 2.760626929470435e-06,
+      "loss": 0.1697,
+      "step": 8470
+    },
+    {
+      "epoch": 5.70275722932078,
+      "grad_norm": 0.6102532744407654,
+      "learning_rate": 2.7012586084065547e-06,
+      "loss": 0.1518,
+      "step": 8480
+    },
+    {
+      "epoch": 5.7094821788836585,
+      "grad_norm": 0.7269166707992554,
+      "learning_rate": 2.641890287342674e-06,
+      "loss": 0.1841,
+      "step": 8490
+    },
+    {
+      "epoch": 5.716207128446537,
+      "grad_norm": 0.5745391249656677,
+      "learning_rate": 2.582521966278794e-06,
+      "loss": 0.1512,
+      "step": 8500
+    },
+    {
+      "epoch": 5.716207128446537,
+      "eval_loss": 0.144563689827919,
+      "eval_runtime": 14.4525,
+      "eval_samples_per_second": 182.944,
+      "eval_steps_per_second": 22.903,
+      "step": 8500
+    },
+    {
+      "epoch": 5.722932078009415,
+      "grad_norm": 0.89340740442276,
+      "learning_rate": 2.5231536452149136e-06,
+      "loss": 0.1528,
+      "step": 8510
+    },
+    {
+      "epoch": 5.729657027572293,
+      "grad_norm": 0.8557310700416565,
+      "learning_rate": 2.4637853241510333e-06,
+      "loss": 0.1803,
+      "step": 8520
+    },
+    {
+      "epoch": 5.736381977135172,
+      "grad_norm": 0.7893068790435791,
+      "learning_rate": 2.404417003087153e-06,
+      "loss": 0.145,
+      "step": 8530
+    },
+    {
+      "epoch": 5.74310692669805,
+      "grad_norm": 0.548139214515686,
+      "learning_rate": 2.3450486820232724e-06,
+      "loss": 0.1753,
+      "step": 8540
+    },
+    {
+      "epoch": 5.749831876260928,
+      "grad_norm": 0.7231383323669434,
+      "learning_rate": 2.285680360959392e-06,
+      "loss": 0.1417,
+      "step": 8550
+    },
+    {
+      "epoch": 5.7565568258238065,
+      "grad_norm": 0.7758724689483643,
+      "learning_rate": 2.226312039895512e-06,
+      "loss": 0.1606,
+      "step": 8560
+    },
+    {
+      "epoch": 5.763281775386685,
+      "grad_norm": 0.692232072353363,
+      "learning_rate": 2.1669437188316313e-06,
+      "loss": 0.1343,
+      "step": 8570
+    },
+    {
+      "epoch": 5.770006724949563,
+      "grad_norm": 0.6432099938392639,
+      "learning_rate": 2.107575397767751e-06,
+      "loss": 0.187,
+      "step": 8580
+    },
+    {
+      "epoch": 5.776731674512441,
+      "grad_norm": 0.581702709197998,
+      "learning_rate": 2.048207076703871e-06,
+      "loss": 0.1344,
+      "step": 8590
+    },
+    {
+      "epoch": 5.78345662407532,
+      "grad_norm": 0.7130106091499329,
+      "learning_rate": 1.9888387556399906e-06,
+      "loss": 0.1501,
+      "step": 8600
+    },
+    {
+      "epoch": 5.790181573638198,
+      "grad_norm": 0.7084644436836243,
+      "learning_rate": 1.9294704345761103e-06,
+      "loss": 0.1725,
+      "step": 8610
+    },
+    {
+      "epoch": 5.796906523201076,
+      "grad_norm": 0.6888975501060486,
+      "learning_rate": 1.87010211351223e-06,
+      "loss": 0.184,
+      "step": 8620
+    },
+    {
+      "epoch": 5.803631472763954,
+      "grad_norm": 0.636211633682251,
+      "learning_rate": 1.8107337924483494e-06,
+      "loss": 0.17,
+      "step": 8630
+    },
+    {
+      "epoch": 5.810356422326833,
+      "grad_norm": 0.4802153408527374,
+      "learning_rate": 1.7513654713844692e-06,
+      "loss": 0.1428,
+      "step": 8640
+    },
+    {
+      "epoch": 5.817081371889711,
+      "grad_norm": 0.5505760908126831,
+      "learning_rate": 1.691997150320589e-06,
+      "loss": 0.177,
+      "step": 8650
+    },
+    {
+      "epoch": 5.823806321452589,
+      "grad_norm": 0.7836665511131287,
+      "learning_rate": 1.6326288292567087e-06,
+      "loss": 0.1725,
+      "step": 8660
+    },
+    {
+      "epoch": 5.8305312710154675,
+      "grad_norm": 0.5304895043373108,
+      "learning_rate": 1.5732605081928285e-06,
+      "loss": 0.1339,
+      "step": 8670
+    },
+    {
+      "epoch": 5.837256220578346,
+      "grad_norm": 0.7424430251121521,
+      "learning_rate": 1.513892187128948e-06,
+      "loss": 0.1441,
+      "step": 8680
+    },
+    {
+      "epoch": 5.843981170141224,
+      "grad_norm": 0.9562227129936218,
+      "learning_rate": 1.4545238660650678e-06,
+      "loss": 0.1688,
+      "step": 8690
+    },
+    {
+      "epoch": 5.850706119704102,
+      "grad_norm": 0.6445603966712952,
+      "learning_rate": 1.3951555450011875e-06,
+      "loss": 0.148,
+      "step": 8700
+    },
+    {
+      "epoch": 5.857431069266981,
+      "grad_norm": 1.0529707670211792,
+      "learning_rate": 1.335787223937307e-06,
+      "loss": 0.1908,
+      "step": 8710
+    },
+    {
+      "epoch": 5.864156018829859,
+      "grad_norm": 1.1476075649261475,
+      "learning_rate": 1.2764189028734268e-06,
+      "loss": 0.1497,
+      "step": 8720
+    },
+    {
+      "epoch": 5.870880968392737,
+      "grad_norm": 0.548938512802124,
+      "learning_rate": 1.2170505818095466e-06,
+      "loss": 0.1498,
+      "step": 8730
+    },
+    {
+      "epoch": 5.8776059179556155,
+      "grad_norm": 0.5004345178604126,
+      "learning_rate": 1.1576822607456662e-06,
+      "loss": 0.1912,
+      "step": 8740
+    },
+    {
+      "epoch": 5.884330867518494,
+      "grad_norm": 0.8999865055084229,
+      "learning_rate": 1.0983139396817857e-06,
+      "loss": 0.1627,
+      "step": 8750
+    },
+    {
+      "epoch": 5.891055817081372,
+      "grad_norm": 0.8366420269012451,
+      "learning_rate": 1.0389456186179055e-06,
+      "loss": 0.1417,
+      "step": 8760
+    },
+    {
+      "epoch": 5.89778076664425,
+      "grad_norm": 0.7696868181228638,
+      "learning_rate": 9.795772975540252e-07,
+      "loss": 0.1651,
+      "step": 8770
+    },
+    {
+      "epoch": 5.904505716207129,
+      "grad_norm": 0.8378667831420898,
+      "learning_rate": 9.202089764901449e-07,
+      "loss": 0.1524,
+      "step": 8780
+    },
+    {
+      "epoch": 5.911230665770007,
+      "grad_norm": 0.5676979422569275,
+      "learning_rate": 8.608406554262645e-07,
+      "loss": 0.1355,
+      "step": 8790
+    },
+    {
+      "epoch": 5.917955615332885,
+      "grad_norm": 0.8161872625350952,
+      "learning_rate": 8.014723343623843e-07,
+      "loss": 0.1469,
+      "step": 8800
+    },
+    {
+      "epoch": 5.9246805648957634,
+      "grad_norm": 0.6396122574806213,
+      "learning_rate": 7.421040132985039e-07,
+      "loss": 0.1343,
+      "step": 8810
+    },
+    {
+      "epoch": 5.931405514458642,
+      "grad_norm": 0.49894294142723083,
+      "learning_rate": 6.827356922346237e-07,
+      "loss": 0.1436,
+      "step": 8820
+    },
+    {
+      "epoch": 5.93813046402152,
+      "grad_norm": 1.815820336341858,
+      "learning_rate": 6.233673711707434e-07,
+      "loss": 0.1587,
+      "step": 8830
+    },
+    {
+      "epoch": 5.944855413584398,
+      "grad_norm": 0.7993642091751099,
+      "learning_rate": 5.63999050106863e-07,
+      "loss": 0.1967,
+      "step": 8840
+    },
+    {
+      "epoch": 5.951580363147277,
+      "grad_norm": 0.9096683859825134,
+      "learning_rate": 5.046307290429827e-07,
+      "loss": 0.1467,
+      "step": 8850
+    },
+    {
+      "epoch": 5.958305312710155,
+      "grad_norm": 0.8135556578636169,
+      "learning_rate": 4.452624079791023e-07,
+      "loss": 0.1708,
+      "step": 8860
+    },
+    {
+      "epoch": 5.965030262273033,
+      "grad_norm": 0.699492335319519,
+      "learning_rate": 3.8589408691522203e-07,
+      "loss": 0.144,
+      "step": 8870
+    },
+    {
+      "epoch": 5.971755211835911,
+      "grad_norm": 0.707664430141449,
+      "learning_rate": 3.2652576585134174e-07,
+      "loss": 0.1314,
+      "step": 8880
+    },
+    {
+      "epoch": 5.97848016139879,
+      "grad_norm": 0.7655701637268066,
+      "learning_rate": 2.671574447874614e-07,
+      "loss": 0.1449,
+      "step": 8890
+    },
+    {
+      "epoch": 5.985205110961668,
+      "grad_norm": 0.7005723714828491,
+      "learning_rate": 2.077891237235811e-07,
+      "loss": 0.1502,
+      "step": 8900
+    },
+    {
+      "epoch": 5.991930060524546,
+      "grad_norm": 0.5278902649879456,
+      "learning_rate": 1.4842080265970078e-07,
+      "loss": 0.1591,
+      "step": 8910
+    },
+    {
+      "epoch": 5.9986550100874245,
+      "grad_norm": 0.9299124479293823,
+      "learning_rate": 8.905248159582048e-08,
+      "loss": 0.1779,
+      "step": 8920
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 8922,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 6,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.506527567364096e+16,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}