diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,8687 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 5.0,
+  "eval_steps": 500,
+  "global_step": 1235,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.004048582995951417,
+      "grad_norm": 6.93506761290288,
+      "learning_rate": 3.2258064516129035e-07,
+      "loss": 0.8242,
+      "step": 1
+    },
+    {
+      "epoch": 0.008097165991902834,
+      "grad_norm": 6.728224827085898,
+      "learning_rate": 6.451612903225807e-07,
+      "loss": 0.8128,
+      "step": 2
+    },
+    {
+      "epoch": 0.012145748987854251,
+      "grad_norm": 7.035054840392815,
+      "learning_rate": 9.67741935483871e-07,
+      "loss": 0.8348,
+      "step": 3
+    },
+    {
+      "epoch": 0.016194331983805668,
+      "grad_norm": 6.7402549266709055,
+      "learning_rate": 1.2903225806451614e-06,
+      "loss": 0.8257,
+      "step": 4
+    },
+    {
+      "epoch": 0.020242914979757085,
+      "grad_norm": 6.489691641936819,
+      "learning_rate": 1.6129032258064516e-06,
+      "loss": 0.8118,
+      "step": 5
+    },
+    {
+      "epoch": 0.024291497975708502,
+      "grad_norm": 6.338195009976488,
+      "learning_rate": 1.935483870967742e-06,
+      "loss": 0.7996,
+      "step": 6
+    },
+    {
+      "epoch": 0.02834008097165992,
+      "grad_norm": 5.152704069270391,
+      "learning_rate": 2.2580645161290324e-06,
+      "loss": 0.7672,
+      "step": 7
+    },
+    {
+      "epoch": 0.032388663967611336,
+      "grad_norm": 4.798612580067465,
+      "learning_rate": 2.580645161290323e-06,
+      "loss": 0.7511,
+      "step": 8
+    },
+    {
+      "epoch": 0.03643724696356275,
+      "grad_norm": 2.8007333356229913,
+      "learning_rate": 2.903225806451613e-06,
+      "loss": 0.7011,
+      "step": 9
+    },
+    {
+      "epoch": 0.04048582995951417,
+      "grad_norm": 2.605095448704562,
+      "learning_rate": 3.225806451612903e-06,
+      "loss": 0.6772,
+      "step": 10
+    },
+    {
+      "epoch": 0.044534412955465584,
+      "grad_norm": 2.4624734316229975,
+      "learning_rate": 3.548387096774194e-06,
+      "loss": 0.6823,
+      "step": 11
+    },
+    {
+      "epoch": 0.048582995951417005,
+      "grad_norm": 3.7450297024026455,
+      "learning_rate": 3.870967741935484e-06,
+      "loss": 0.663,
+      "step": 12
+    },
+    {
+      "epoch": 0.05263157894736842,
+      "grad_norm": 3.93553787691552,
+      "learning_rate": 4.193548387096774e-06,
+      "loss": 0.6547,
+      "step": 13
+    },
+    {
+      "epoch": 0.05668016194331984,
+      "grad_norm": 3.662340019258872,
+      "learning_rate": 4.516129032258065e-06,
+      "loss": 0.6132,
+      "step": 14
+    },
+    {
+      "epoch": 0.06072874493927125,
+      "grad_norm": 3.5159875294057974,
+      "learning_rate": 4.838709677419355e-06,
+      "loss": 0.6179,
+      "step": 15
+    },
+    {
+      "epoch": 0.06477732793522267,
+      "grad_norm": 3.1190240225639934,
+      "learning_rate": 5.161290322580646e-06,
+      "loss": 0.604,
+      "step": 16
+    },
+    {
+      "epoch": 0.06882591093117409,
+      "grad_norm": 2.836877626848926,
+      "learning_rate": 5.483870967741935e-06,
+      "loss": 0.5651,
+      "step": 17
+    },
+    {
+      "epoch": 0.0728744939271255,
+      "grad_norm": 2.474145127817769,
+      "learning_rate": 5.806451612903226e-06,
+      "loss": 0.5684,
+      "step": 18
+    },
+    {
+      "epoch": 0.07692307692307693,
+      "grad_norm": 1.9198777259414883,
+      "learning_rate": 6.129032258064517e-06,
+      "loss": 0.5633,
+      "step": 19
+    },
+    {
+      "epoch": 0.08097165991902834,
+      "grad_norm": 1.3285025022283075,
+      "learning_rate": 6.451612903225806e-06,
+      "loss": 0.5614,
+      "step": 20
+    },
+    {
+      "epoch": 0.08502024291497975,
+      "grad_norm": 1.4035030189821487,
+      "learning_rate": 6.774193548387097e-06,
+      "loss": 0.5269,
+      "step": 21
+    },
+    {
+      "epoch": 0.08906882591093117,
+      "grad_norm": 1.5976695560738816,
+      "learning_rate": 7.096774193548388e-06,
+      "loss": 0.5329,
+      "step": 22
+    },
+    {
+      "epoch": 0.0931174089068826,
+      "grad_norm": 1.6406676288237352,
+      "learning_rate": 7.4193548387096784e-06,
+      "loss": 0.5318,
+      "step": 23
+    },
+    {
+      "epoch": 0.09716599190283401,
+      "grad_norm": 1.0783580355237576,
+      "learning_rate": 7.741935483870968e-06,
+      "loss": 0.5127,
+      "step": 24
+    },
+    {
+      "epoch": 0.10121457489878542,
+      "grad_norm": 0.7951049969611887,
+      "learning_rate": 8.064516129032258e-06,
+      "loss": 0.497,
+      "step": 25
+    },
+    {
+      "epoch": 0.10526315789473684,
+      "grad_norm": 0.9560824026186652,
+      "learning_rate": 8.387096774193549e-06,
+      "loss": 0.4936,
+      "step": 26
+    },
+    {
+      "epoch": 0.10931174089068826,
+      "grad_norm": 1.0709715681963936,
+      "learning_rate": 8.70967741935484e-06,
+      "loss": 0.4772,
+      "step": 27
+    },
+    {
+      "epoch": 0.11336032388663968,
+      "grad_norm": 0.9991818998969109,
+      "learning_rate": 9.03225806451613e-06,
+      "loss": 0.4849,
+      "step": 28
+    },
+    {
+      "epoch": 0.11740890688259109,
+      "grad_norm": 0.726147060039181,
+      "learning_rate": 9.35483870967742e-06,
+      "loss": 0.4859,
+      "step": 29
+    },
+    {
+      "epoch": 0.1214574898785425,
+      "grad_norm": 0.7734854963942096,
+      "learning_rate": 9.67741935483871e-06,
+      "loss": 0.4735,
+      "step": 30
+    },
+    {
+      "epoch": 0.12550607287449392,
+      "grad_norm": 0.943000177568488,
+      "learning_rate": 1e-05,
+      "loss": 0.4882,
+      "step": 31
+    },
+    {
+      "epoch": 0.12955465587044535,
+      "grad_norm": 0.8279430026570178,
+      "learning_rate": 1.0322580645161291e-05,
+      "loss": 0.4742,
+      "step": 32
+    },
+    {
+      "epoch": 0.13360323886639677,
+      "grad_norm": 0.6323454709029194,
+      "learning_rate": 1.0645161290322582e-05,
+      "loss": 0.4603,
+      "step": 33
+    },
+    {
+      "epoch": 0.13765182186234817,
+      "grad_norm": 0.699825695799157,
+      "learning_rate": 1.096774193548387e-05,
+      "loss": 0.4592,
+      "step": 34
+    },
+    {
+      "epoch": 0.1417004048582996,
+      "grad_norm": 0.5713477165035835,
+      "learning_rate": 1.1290322580645164e-05,
+      "loss": 0.4542,
+      "step": 35
+    },
+    {
+      "epoch": 0.145748987854251,
+      "grad_norm": 0.5821676115557833,
+      "learning_rate": 1.1612903225806453e-05,
+      "loss": 0.4561,
+      "step": 36
+    },
+    {
+      "epoch": 0.14979757085020243,
+      "grad_norm": 0.6686992879987348,
+      "learning_rate": 1.1935483870967743e-05,
+      "loss": 0.4718,
+      "step": 37
+    },
+    {
+      "epoch": 0.15384615384615385,
+      "grad_norm": 0.5687294153405139,
+      "learning_rate": 1.2258064516129034e-05,
+      "loss": 0.4432,
+      "step": 38
+    },
+    {
+      "epoch": 0.15789473684210525,
+      "grad_norm": 0.6128323530177185,
+      "learning_rate": 1.2580645161290324e-05,
+      "loss": 0.4519,
+      "step": 39
+    },
+    {
+      "epoch": 0.16194331983805668,
+      "grad_norm": 0.5479341958360333,
+      "learning_rate": 1.2903225806451613e-05,
+      "loss": 0.4423,
+      "step": 40
+    },
+    {
+      "epoch": 0.1659919028340081,
+      "grad_norm": 0.5973983075030315,
+      "learning_rate": 1.3225806451612903e-05,
+      "loss": 0.4338,
+      "step": 41
+    },
+    {
+      "epoch": 0.1700404858299595,
+      "grad_norm": 0.617820570478686,
+      "learning_rate": 1.3548387096774194e-05,
+      "loss": 0.4487,
+      "step": 42
+    },
+    {
+      "epoch": 0.17408906882591094,
+      "grad_norm": 0.5785046132805214,
+      "learning_rate": 1.3870967741935486e-05,
+      "loss": 0.4519,
+      "step": 43
+    },
+    {
+      "epoch": 0.17813765182186234,
+      "grad_norm": 0.5373361538081993,
+      "learning_rate": 1.4193548387096776e-05,
+      "loss": 0.4532,
+      "step": 44
+    },
+    {
+      "epoch": 0.18218623481781376,
+      "grad_norm": 0.5771857419268738,
+      "learning_rate": 1.4516129032258066e-05,
+      "loss": 0.4424,
+      "step": 45
+    },
+    {
+      "epoch": 0.1862348178137652,
+      "grad_norm": 0.5378198560112833,
+      "learning_rate": 1.4838709677419357e-05,
+      "loss": 0.4572,
+      "step": 46
+    },
+    {
+      "epoch": 0.1902834008097166,
+      "grad_norm": 0.47970232695607457,
+      "learning_rate": 1.5161290322580646e-05,
+      "loss": 0.4276,
+      "step": 47
+    },
+    {
+      "epoch": 0.19433198380566802,
+      "grad_norm": 0.6663165748577663,
+      "learning_rate": 1.5483870967741936e-05,
+      "loss": 0.4596,
+      "step": 48
+    },
+    {
+      "epoch": 0.19838056680161945,
+      "grad_norm": 0.5149183481622055,
+      "learning_rate": 1.5806451612903226e-05,
+      "loss": 0.42,
+      "step": 49
+    },
+    {
+      "epoch": 0.20242914979757085,
+      "grad_norm": 0.4721809849135527,
+      "learning_rate": 1.6129032258064517e-05,
+      "loss": 0.448,
+      "step": 50
+    },
+    {
+      "epoch": 0.20647773279352227,
+      "grad_norm": 0.5956314654053572,
+      "learning_rate": 1.6451612903225807e-05,
+      "loss": 0.4307,
+      "step": 51
+    },
+    {
+      "epoch": 0.21052631578947367,
+      "grad_norm": 0.5848382884051311,
+      "learning_rate": 1.6774193548387098e-05,
+      "loss": 0.4318,
+      "step": 52
+    },
+    {
+      "epoch": 0.2145748987854251,
+      "grad_norm": 0.493053257643866,
+      "learning_rate": 1.7096774193548388e-05,
+      "loss": 0.4363,
+      "step": 53
+    },
+    {
+      "epoch": 0.21862348178137653,
+      "grad_norm": 0.586284904163215,
+      "learning_rate": 1.741935483870968e-05,
+      "loss": 0.4347,
+      "step": 54
+    },
+    {
+      "epoch": 0.22267206477732793,
+      "grad_norm": 0.5206787949419898,
+      "learning_rate": 1.774193548387097e-05,
+      "loss": 0.4193,
+      "step": 55
+    },
+    {
+      "epoch": 0.22672064777327935,
+      "grad_norm": 0.5129895599752715,
+      "learning_rate": 1.806451612903226e-05,
+      "loss": 0.4128,
+      "step": 56
+    },
+    {
+      "epoch": 0.23076923076923078,
+      "grad_norm": 0.579563724742604,
+      "learning_rate": 1.838709677419355e-05,
+      "loss": 0.433,
+      "step": 57
+    },
+    {
+      "epoch": 0.23481781376518218,
+      "grad_norm": 0.5349696769781325,
+      "learning_rate": 1.870967741935484e-05,
+      "loss": 0.4102,
+      "step": 58
+    },
+    {
+      "epoch": 0.2388663967611336,
+      "grad_norm": 0.5243840497976267,
+      "learning_rate": 1.903225806451613e-05,
+      "loss": 0.4316,
+      "step": 59
+    },
+    {
+      "epoch": 0.242914979757085,
+      "grad_norm": 0.5851495419970012,
+      "learning_rate": 1.935483870967742e-05,
+      "loss": 0.4194,
+      "step": 60
+    },
+    {
+      "epoch": 0.24696356275303644,
+      "grad_norm": 0.5806027888993117,
+      "learning_rate": 1.967741935483871e-05,
+      "loss": 0.4087,
+      "step": 61
+    },
+    {
+      "epoch": 0.25101214574898784,
+      "grad_norm": 0.6040158729023761,
+      "learning_rate": 2e-05,
+      "loss": 0.43,
+      "step": 62
+    },
+    {
+      "epoch": 0.2550607287449393,
+      "grad_norm": 0.47880449632793637,
+      "learning_rate": 2.0322580645161292e-05,
+      "loss": 0.4221,
+      "step": 63
+    },
+    {
+      "epoch": 0.2591093117408907,
+      "grad_norm": 0.6697008222563356,
+      "learning_rate": 2.0645161290322582e-05,
+      "loss": 0.4149,
+      "step": 64
+    },
+    {
+      "epoch": 0.2631578947368421,
+      "grad_norm": 0.5450357803180627,
+      "learning_rate": 2.0967741935483873e-05,
+      "loss": 0.4173,
+      "step": 65
+    },
+    {
+      "epoch": 0.26720647773279355,
+      "grad_norm": 0.6740382351043922,
+      "learning_rate": 2.1290322580645163e-05,
+      "loss": 0.4167,
+      "step": 66
+    },
+    {
+      "epoch": 0.27125506072874495,
+      "grad_norm": 0.5490409016613031,
+      "learning_rate": 2.161290322580645e-05,
+      "loss": 0.4218,
+      "step": 67
+    },
+    {
+      "epoch": 0.27530364372469635,
+      "grad_norm": 0.5325953455218867,
+      "learning_rate": 2.193548387096774e-05,
+      "loss": 0.4108,
+      "step": 68
+    },
+    {
+      "epoch": 0.2793522267206478,
+      "grad_norm": 0.6363524923613719,
+      "learning_rate": 2.225806451612903e-05,
+      "loss": 0.4207,
+      "step": 69
+    },
+    {
+      "epoch": 0.2834008097165992,
+      "grad_norm": 0.5494518588772667,
+      "learning_rate": 2.2580645161290328e-05,
+      "loss": 0.403,
+      "step": 70
+    },
+    {
+      "epoch": 0.2874493927125506,
+      "grad_norm": 0.6685541367365478,
+      "learning_rate": 2.2903225806451615e-05,
+      "loss": 0.4087,
+      "step": 71
+    },
+    {
+      "epoch": 0.291497975708502,
+      "grad_norm": 0.7310629860142822,
+      "learning_rate": 2.3225806451612906e-05,
+      "loss": 0.4128,
+      "step": 72
+    },
+    {
+      "epoch": 0.29554655870445345,
+      "grad_norm": 0.7311685465414492,
+      "learning_rate": 2.3548387096774196e-05,
+      "loss": 0.4217,
+      "step": 73
+    },
+    {
+      "epoch": 0.29959514170040485,
+      "grad_norm": 0.9339152260004198,
+      "learning_rate": 2.3870967741935486e-05,
+      "loss": 0.421,
+      "step": 74
+    },
+    {
+      "epoch": 0.30364372469635625,
+      "grad_norm": 0.6169540993208952,
+      "learning_rate": 2.4193548387096777e-05,
+      "loss": 0.3834,
+      "step": 75
+    },
+    {
+      "epoch": 0.3076923076923077,
+      "grad_norm": 0.8010241138940468,
+      "learning_rate": 2.4516129032258067e-05,
+      "loss": 0.4213,
+      "step": 76
+    },
+    {
+      "epoch": 0.3117408906882591,
+      "grad_norm": 0.6219741835518974,
+      "learning_rate": 2.4838709677419358e-05,
+      "loss": 0.405,
+      "step": 77
+    },
+    {
+      "epoch": 0.3157894736842105,
+      "grad_norm": 0.7285668575632289,
+      "learning_rate": 2.5161290322580648e-05,
+      "loss": 0.4246,
+      "step": 78
+    },
+    {
+      "epoch": 0.31983805668016196,
+      "grad_norm": 0.6425883435572683,
+      "learning_rate": 2.548387096774194e-05,
+      "loss": 0.3904,
+      "step": 79
+    },
+    {
+      "epoch": 0.32388663967611336,
+      "grad_norm": 0.705108039315312,
+      "learning_rate": 2.5806451612903226e-05,
+      "loss": 0.4107,
+      "step": 80
+    },
+    {
+      "epoch": 0.32793522267206476,
+      "grad_norm": 0.7819843971441914,
+      "learning_rate": 2.6129032258064516e-05,
+      "loss": 0.3995,
+      "step": 81
+    },
+    {
+      "epoch": 0.3319838056680162,
+      "grad_norm": 0.8219519120269072,
+      "learning_rate": 2.6451612903225806e-05,
+      "loss": 0.4313,
+      "step": 82
+    },
+    {
+      "epoch": 0.3360323886639676,
+      "grad_norm": 0.7928863428544981,
+      "learning_rate": 2.6774193548387097e-05,
+      "loss": 0.4047,
+      "step": 83
+    },
+    {
+      "epoch": 0.340080971659919,
+      "grad_norm": 0.7859160069504129,
+      "learning_rate": 2.7096774193548387e-05,
+      "loss": 0.3973,
+      "step": 84
+    },
+    {
+      "epoch": 0.3441295546558704,
+      "grad_norm": 0.8282560022580644,
+      "learning_rate": 2.7419354838709678e-05,
+      "loss": 0.405,
+      "step": 85
+    },
+    {
+      "epoch": 0.3481781376518219,
+      "grad_norm": 0.7109674136872353,
+      "learning_rate": 2.774193548387097e-05,
+      "loss": 0.3955,
+      "step": 86
+    },
+    {
+      "epoch": 0.3522267206477733,
+      "grad_norm": 0.8661680361693244,
+      "learning_rate": 2.8064516129032262e-05,
+      "loss": 0.4004,
+      "step": 87
+    },
+    {
+      "epoch": 0.3562753036437247,
+      "grad_norm": 0.85560487726791,
+      "learning_rate": 2.8387096774193552e-05,
+      "loss": 0.3933,
+      "step": 88
+    },
+    {
+      "epoch": 0.3603238866396761,
+      "grad_norm": 0.8674758800664015,
+      "learning_rate": 2.8709677419354843e-05,
+      "loss": 0.4065,
+      "step": 89
+    },
+    {
+      "epoch": 0.3643724696356275,
+      "grad_norm": 0.8174457388357976,
+      "learning_rate": 2.9032258064516133e-05,
+      "loss": 0.431,
+      "step": 90
+    },
+    {
+      "epoch": 0.3684210526315789,
+      "grad_norm": 0.5712446612256497,
+      "learning_rate": 2.9354838709677423e-05,
+      "loss": 0.4126,
+      "step": 91
+    },
+    {
+      "epoch": 0.3724696356275304,
+      "grad_norm": 0.7878518140700018,
+      "learning_rate": 2.9677419354838714e-05,
+      "loss": 0.4098,
+      "step": 92
+    },
+    {
+      "epoch": 0.3765182186234818,
+      "grad_norm": 0.6223816992740082,
+      "learning_rate": 3.0000000000000004e-05,
+      "loss": 0.4065,
+      "step": 93
+    },
+    {
+      "epoch": 0.3805668016194332,
+      "grad_norm": 0.8075859571665865,
+      "learning_rate": 3.032258064516129e-05,
+      "loss": 0.3897,
+      "step": 94
+    },
+    {
+      "epoch": 0.38461538461538464,
+      "grad_norm": 0.6231672487431316,
+      "learning_rate": 3.0645161290322585e-05,
+      "loss": 0.4,
+      "step": 95
+    },
+    {
+      "epoch": 0.38866396761133604,
+      "grad_norm": 0.8206513586378699,
+      "learning_rate": 3.096774193548387e-05,
+      "loss": 0.3947,
+      "step": 96
+    },
+    {
+      "epoch": 0.39271255060728744,
+      "grad_norm": 0.7134315425126568,
+      "learning_rate": 3.1290322580645166e-05,
+      "loss": 0.4019,
+      "step": 97
+    },
+    {
+      "epoch": 0.3967611336032389,
+      "grad_norm": 0.5843975412240461,
+      "learning_rate": 3.161290322580645e-05,
+      "loss": 0.399,
+      "step": 98
+    },
+    {
+      "epoch": 0.4008097165991903,
+      "grad_norm": 0.6692318794704084,
+      "learning_rate": 3.193548387096774e-05,
+      "loss": 0.4009,
+      "step": 99
+    },
+    {
+      "epoch": 0.4048582995951417,
+      "grad_norm": 0.5842871436991638,
+      "learning_rate": 3.2258064516129034e-05,
+      "loss": 0.3935,
+      "step": 100
+    },
+    {
+      "epoch": 0.4089068825910931,
+      "grad_norm": 0.7046129827075374,
+      "learning_rate": 3.258064516129033e-05,
+      "loss": 0.4005,
+      "step": 101
+    },
+    {
+      "epoch": 0.41295546558704455,
+      "grad_norm": 0.6841320367744452,
+      "learning_rate": 3.2903225806451614e-05,
+      "loss": 0.3823,
+      "step": 102
+    },
+    {
+      "epoch": 0.41700404858299595,
+      "grad_norm": 0.6097245991721043,
+      "learning_rate": 3.322580645161291e-05,
+      "loss": 0.4115,
+      "step": 103
+    },
+    {
+      "epoch": 0.42105263157894735,
+      "grad_norm": 0.7501113502776863,
+      "learning_rate": 3.3548387096774195e-05,
+      "loss": 0.401,
+      "step": 104
+    },
+    {
+      "epoch": 0.4251012145748988,
+      "grad_norm": 0.6481938396071341,
+      "learning_rate": 3.387096774193549e-05,
+      "loss": 0.4002,
+      "step": 105
+    },
+    {
+      "epoch": 0.4291497975708502,
+      "grad_norm": 0.7030525966292959,
+      "learning_rate": 3.4193548387096776e-05,
+      "loss": 0.3992,
+      "step": 106
+    },
+    {
+      "epoch": 0.4331983805668016,
+      "grad_norm": 0.6555379492678538,
+      "learning_rate": 3.451612903225807e-05,
+      "loss": 0.3869,
+      "step": 107
+    },
+    {
+      "epoch": 0.43724696356275305,
+      "grad_norm": 0.7064589543416501,
+      "learning_rate": 3.483870967741936e-05,
+      "loss": 0.4109,
+      "step": 108
+    },
+    {
+      "epoch": 0.44129554655870445,
+      "grad_norm": 0.5897435638383827,
+      "learning_rate": 3.516129032258065e-05,
+      "loss": 0.4105,
+      "step": 109
+    },
+    {
+      "epoch": 0.44534412955465585,
+      "grad_norm": 0.6600929416782032,
+      "learning_rate": 3.548387096774194e-05,
+      "loss": 0.4026,
+      "step": 110
+    },
+    {
+      "epoch": 0.4493927125506073,
+      "grad_norm": 0.5335397922300148,
+      "learning_rate": 3.5806451612903225e-05,
+      "loss": 0.4161,
+      "step": 111
+    },
+    {
+      "epoch": 0.4534412955465587,
+      "grad_norm": 0.689375995043661,
+      "learning_rate": 3.612903225806452e-05,
+      "loss": 0.3946,
+      "step": 112
+    },
+    {
+      "epoch": 0.4574898785425101,
+      "grad_norm": 1.1044171628560384,
+      "learning_rate": 3.6451612903225805e-05,
+      "loss": 0.4147,
+      "step": 113
+    },
+    {
+      "epoch": 0.46153846153846156,
+      "grad_norm": 0.753399018265872,
+      "learning_rate": 3.67741935483871e-05,
+      "loss": 0.4181,
+      "step": 114
+    },
+    {
+      "epoch": 0.46558704453441296,
+      "grad_norm": 0.6495187863751273,
+      "learning_rate": 3.7096774193548386e-05,
+      "loss": 0.4045,
+      "step": 115
+    },
+    {
+      "epoch": 0.46963562753036436,
+      "grad_norm": 0.712497838362611,
+      "learning_rate": 3.741935483870968e-05,
+      "loss": 0.4103,
+      "step": 116
+    },
+    {
+      "epoch": 0.47368421052631576,
+      "grad_norm": 0.46814033484466794,
+      "learning_rate": 3.7741935483870974e-05,
+      "loss": 0.394,
+      "step": 117
+    },
+    {
+      "epoch": 0.4777327935222672,
+      "grad_norm": 0.7536294157443522,
+      "learning_rate": 3.806451612903226e-05,
+      "loss": 0.4028,
+      "step": 118
+    },
+    {
+      "epoch": 0.4817813765182186,
+      "grad_norm": 0.7533427972400936,
+      "learning_rate": 3.8387096774193555e-05,
+      "loss": 0.3931,
+      "step": 119
+    },
+    {
+      "epoch": 0.48582995951417,
+      "grad_norm": 0.49982819639129805,
+      "learning_rate": 3.870967741935484e-05,
+      "loss": 0.4021,
+      "step": 120
+    },
+    {
+      "epoch": 0.4898785425101215,
+      "grad_norm": 0.6115188815938445,
+      "learning_rate": 3.9032258064516135e-05,
+      "loss": 0.4071,
+      "step": 121
+    },
+    {
+      "epoch": 0.4939271255060729,
+      "grad_norm": 0.6239774409174472,
+      "learning_rate": 3.935483870967742e-05,
+      "loss": 0.3918,
+      "step": 122
+    },
+    {
+      "epoch": 0.4979757085020243,
+      "grad_norm": 0.6328812066665706,
+      "learning_rate": 3.9677419354838716e-05,
+      "loss": 0.4146,
+      "step": 123
+    },
+    {
+      "epoch": 0.5020242914979757,
+      "grad_norm": 0.8512716735359367,
+      "learning_rate": 4e-05,
+      "loss": 0.39,
+      "step": 124
+    },
+    {
+      "epoch": 0.5060728744939271,
+      "grad_norm": 0.7888328054339072,
+      "learning_rate": 3.9999920040266474e-05,
+      "loss": 0.3906,
+      "step": 125
+    },
+    {
+      "epoch": 0.5101214574898786,
+      "grad_norm": 0.6587051880834979,
+      "learning_rate": 3.999968016170525e-05,
+      "loss": 0.4125,
+      "step": 126
+    },
+    {
+      "epoch": 0.5141700404858299,
+      "grad_norm": 0.5434524269787779,
+      "learning_rate": 3.9999280366234394e-05,
+      "loss": 0.4056,
+      "step": 127
+    },
+    {
+      "epoch": 0.5182186234817814,
+      "grad_norm": 0.6310317446854713,
+      "learning_rate": 3.999872065705065e-05,
+      "loss": 0.3806,
+      "step": 128
+    },
+    {
+      "epoch": 0.5222672064777328,
+      "grad_norm": 0.5173898752150703,
+      "learning_rate": 3.999800103862945e-05,
+      "loss": 0.4011,
+      "step": 129
+    },
+    {
+      "epoch": 0.5263157894736842,
+      "grad_norm": 0.5234762709205704,
+      "learning_rate": 3.9997121516724834e-05,
+      "loss": 0.403,
+      "step": 130
+    },
+    {
+      "epoch": 0.5303643724696356,
+      "grad_norm": 0.5458408810919101,
+      "learning_rate": 3.999608209836944e-05,
+      "loss": 0.4033,
+      "step": 131
+    },
+    {
+      "epoch": 0.5344129554655871,
+      "grad_norm": 0.6191459234932507,
+      "learning_rate": 3.999488279187443e-05,
+      "loss": 0.3832,
+      "step": 132
+    },
+    {
+      "epoch": 0.5384615384615384,
+      "grad_norm": 0.4598670087742607,
+      "learning_rate": 3.9993523606829425e-05,
+      "loss": 0.4204,
+      "step": 133
+    },
+    {
+      "epoch": 0.5425101214574899,
+      "grad_norm": 0.6334057836511542,
+      "learning_rate": 3.9992004554102435e-05,
+      "loss": 0.407,
+      "step": 134
+    },
+    {
+      "epoch": 0.5465587044534413,
+      "grad_norm": 0.6124768566854322,
+      "learning_rate": 3.999032564583976e-05,
+      "loss": 0.4086,
+      "step": 135
+    },
+    {
+      "epoch": 0.5506072874493927,
+      "grad_norm": 0.5767155527030136,
+      "learning_rate": 3.998848689546591e-05,
+      "loss": 0.3996,
+      "step": 136
+    },
+    {
+      "epoch": 0.5546558704453441,
+      "grad_norm": 0.5631422488809599,
+      "learning_rate": 3.9986488317683496e-05,
+      "loss": 0.3914,
+      "step": 137
+    },
+    {
+      "epoch": 0.5587044534412956,
+      "grad_norm": 0.6270590130270602,
+      "learning_rate": 3.9984329928473067e-05,
+      "loss": 0.4107,
+      "step": 138
+    },
+    {
+      "epoch": 0.562753036437247,
+      "grad_norm": 0.46813475796961507,
+      "learning_rate": 3.998201174509306e-05,
+      "loss": 0.4014,
+      "step": 139
+    },
+    {
+      "epoch": 0.5668016194331984,
+      "grad_norm": 0.5317413186008069,
+      "learning_rate": 3.997953378607961e-05,
+      "loss": 0.3862,
+      "step": 140
+    },
+    {
+      "epoch": 0.5708502024291497,
+      "grad_norm": 0.5190601298233718,
+      "learning_rate": 3.9976896071246403e-05,
+      "loss": 0.3918,
+      "step": 141
+    },
+    {
+      "epoch": 0.5748987854251012,
+      "grad_norm": 0.534245548357502,
+      "learning_rate": 3.997409862168454e-05,
+      "loss": 0.3994,
+      "step": 142
+    },
+    {
+      "epoch": 0.5789473684210527,
+      "grad_norm": 0.5447643745974791,
+      "learning_rate": 3.997114145976237e-05,
+      "loss": 0.3918,
+      "step": 143
+    },
+    {
+      "epoch": 0.582995951417004,
+      "grad_norm": 0.6332773838253363,
+      "learning_rate": 3.9968024609125245e-05,
+      "loss": 0.404,
+      "step": 144
+    },
+    {
+      "epoch": 0.5870445344129555,
+      "grad_norm": 0.5446950155878345,
+      "learning_rate": 3.996474809469545e-05,
+      "loss": 0.3996,
+      "step": 145
+    },
+    {
+      "epoch": 0.5910931174089069,
+      "grad_norm": 0.5807345519270956,
+      "learning_rate": 3.996131194267189e-05,
+      "loss": 0.4139,
+      "step": 146
+    },
+    {
+      "epoch": 0.5951417004048583,
+      "grad_norm": 0.5966916586131096,
+      "learning_rate": 3.995771618052995e-05,
+      "loss": 0.389,
+      "step": 147
+    },
+    {
+      "epoch": 0.5991902834008097,
+      "grad_norm": 0.42703686890700354,
+      "learning_rate": 3.9953960837021256e-05,
+      "loss": 0.3915,
+      "step": 148
+    },
+    {
+      "epoch": 0.6032388663967612,
+      "grad_norm": 0.5532906129707901,
+      "learning_rate": 3.995004594217343e-05,
+      "loss": 0.4087,
+      "step": 149
+    },
+    {
+      "epoch": 0.6072874493927125,
+      "grad_norm": 0.48809650951300415,
+      "learning_rate": 3.994597152728987e-05,
+      "loss": 0.3876,
+      "step": 150
+    },
+    {
+      "epoch": 0.611336032388664,
+      "grad_norm": 0.6558661778920176,
+      "learning_rate": 3.994173762494947e-05,
+      "loss": 0.383,
+      "step": 151
+    },
+    {
+      "epoch": 0.6153846153846154,
+      "grad_norm": 0.6766180996396752,
+      "learning_rate": 3.9937344269006424e-05,
+      "loss": 0.3968,
+      "step": 152
+    },
+    {
+      "epoch": 0.6194331983805668,
+      "grad_norm": 0.5761134212575872,
+      "learning_rate": 3.993279149458988e-05,
+      "loss": 0.4052,
+      "step": 153
+    },
+    {
+      "epoch": 0.6234817813765182,
+      "grad_norm": 0.5481155328015522,
+      "learning_rate": 3.9928079338103695e-05,
+      "loss": 0.4084,
+      "step": 154
+    },
+    {
+      "epoch": 0.6275303643724697,
+      "grad_norm": 0.5913378757386736,
+      "learning_rate": 3.992320783722616e-05,
+      "loss": 0.3954,
+      "step": 155
+    },
+    {
+      "epoch": 0.631578947368421,
+      "grad_norm": 0.5879794756566398,
+      "learning_rate": 3.991817703090965e-05,
+      "loss": 0.3863,
+      "step": 156
+    },
+    {
+      "epoch": 0.6356275303643725,
+      "grad_norm": 0.699233897943011,
+      "learning_rate": 3.991298695938038e-05,
+      "loss": 0.3963,
+      "step": 157
+    },
+    {
+      "epoch": 0.6396761133603239,
+      "grad_norm": 0.5234543838619287,
+      "learning_rate": 3.9907637664138e-05,
+      "loss": 0.3911,
+      "step": 158
+    },
+    {
+      "epoch": 0.6437246963562753,
+      "grad_norm": 0.7207905236756805,
+      "learning_rate": 3.990212918795534e-05,
+      "loss": 0.3992,
+      "step": 159
+    },
+    {
+      "epoch": 0.6477732793522267,
+      "grad_norm": 0.6124515323101511,
+      "learning_rate": 3.9896461574878036e-05,
+      "loss": 0.3916,
+      "step": 160
+    },
+    {
+      "epoch": 0.6518218623481782,
+      "grad_norm": 0.5836686183287692,
+      "learning_rate": 3.989063487022416e-05,
+      "loss": 0.3696,
+      "step": 161
+    },
+    {
+      "epoch": 0.6558704453441295,
+      "grad_norm": 0.5960418034097117,
+      "learning_rate": 3.9884649120583903e-05,
+      "loss": 0.3819,
+      "step": 162
+    },
+    {
+      "epoch": 0.659919028340081,
+      "grad_norm": 0.6167871837949725,
+      "learning_rate": 3.987850437381915e-05,
+      "loss": 0.3979,
+      "step": 163
+    },
+    {
+      "epoch": 0.6639676113360324,
+      "grad_norm": 0.6642480598392148,
+      "learning_rate": 3.987220067906313e-05,
+      "loss": 0.3821,
+      "step": 164
+    },
+    {
+      "epoch": 0.6680161943319838,
+      "grad_norm": 0.5772082618876039,
+      "learning_rate": 3.986573808672002e-05,
+      "loss": 0.3977,
+      "step": 165
+    },
+    {
+      "epoch": 0.6720647773279352,
+      "grad_norm": 0.5951108619251065,
+      "learning_rate": 3.9859116648464534e-05,
+      "loss": 0.4013,
+      "step": 166
+    },
+    {
+      "epoch": 0.6761133603238867,
+      "grad_norm": 0.7835363214539045,
+      "learning_rate": 3.985233641724152e-05,
+      "loss": 0.3848,
+      "step": 167
+    },
+    {
+      "epoch": 0.680161943319838,
+      "grad_norm": 0.5955841053192503,
+      "learning_rate": 3.984539744726554e-05,
+      "loss": 0.3791,
+      "step": 168
+    },
+    {
+      "epoch": 0.6842105263157895,
+      "grad_norm": 0.5992724323481706,
+      "learning_rate": 3.983829979402038e-05,
+      "loss": 0.392,
+      "step": 169
+    },
+    {
+      "epoch": 0.6882591093117408,
+      "grad_norm": 0.5774278133478388,
+      "learning_rate": 3.983104351425871e-05,
+      "loss": 0.3884,
+      "step": 170
+    },
+    {
+      "epoch": 0.6923076923076923,
+      "grad_norm": 0.5884545127799761,
+      "learning_rate": 3.982362866600154e-05,
+      "loss": 0.3837,
+      "step": 171
+    },
+    {
+      "epoch": 0.6963562753036437,
+      "grad_norm": 0.4905885670665835,
+      "learning_rate": 3.9816055308537806e-05,
+      "loss": 0.4119,
+      "step": 172
+    },
+    {
+      "epoch": 0.7004048582995951,
+      "grad_norm": 0.569254572802065,
+      "learning_rate": 3.980832350242388e-05,
+      "loss": 0.391,
+      "step": 173
+    },
+    {
+      "epoch": 0.7044534412955465,
+      "grad_norm": 0.5184664769595553,
+      "learning_rate": 3.980043330948306e-05,
+      "loss": 0.3739,
+      "step": 174
+    },
+    {
+      "epoch": 0.708502024291498,
+      "grad_norm": 0.5067783270507124,
+      "learning_rate": 3.9792384792805124e-05,
+      "loss": 0.3921,
+      "step": 175
+    },
+    {
+      "epoch": 0.7125506072874493,
+      "grad_norm": 0.4466296648838256,
+      "learning_rate": 3.978417801674579e-05,
+      "loss": 0.3838,
+      "step": 176
+    },
+    {
+      "epoch": 0.7165991902834008,
+      "grad_norm": 0.6164819122210317,
+      "learning_rate": 3.9775813046926244e-05,
+      "loss": 0.3927,
+      "step": 177
+    },
+    {
+      "epoch": 0.7206477732793523,
+      "grad_norm": 0.4137223952147918,
+      "learning_rate": 3.976728995023254e-05,
+      "loss": 0.3884,
+      "step": 178
+    },
+    {
+      "epoch": 0.7246963562753036,
+      "grad_norm": 0.5934213706548929,
+      "learning_rate": 3.975860879481514e-05,
+      "loss": 0.387,
+      "step": 179
+    },
+    {
+      "epoch": 0.728744939271255,
+      "grad_norm": 0.5199015883412167,
+      "learning_rate": 3.974976965008833e-05,
+      "loss": 0.3926,
+      "step": 180
+    },
+    {
+      "epoch": 0.7327935222672065,
+      "grad_norm": 0.5215626116122729,
+      "learning_rate": 3.9740772586729674e-05,
+      "loss": 0.3813,
+      "step": 181
+    },
+    {
+      "epoch": 0.7368421052631579,
+      "grad_norm": 0.4873285470271915,
+      "learning_rate": 3.973161767667946e-05,
+      "loss": 0.3888,
+      "step": 182
+    },
+    {
+      "epoch": 0.7408906882591093,
+      "grad_norm": 0.5143600194726559,
+      "learning_rate": 3.972230499314009e-05,
+      "loss": 0.3977,
+      "step": 183
+    },
+    {
+      "epoch": 0.7449392712550608,
+      "grad_norm": 0.4456745092887104,
+      "learning_rate": 3.9712834610575545e-05,
+      "loss": 0.3763,
+      "step": 184
+    },
+    {
+      "epoch": 0.7489878542510121,
+      "grad_norm": 0.43428410667566114,
+      "learning_rate": 3.9703206604710746e-05,
+      "loss": 0.3903,
+      "step": 185
+    },
+    {
+      "epoch": 0.7530364372469636,
+      "grad_norm": 0.47100981675160913,
+      "learning_rate": 3.969342105253098e-05,
+      "loss": 0.3833,
+      "step": 186
+    },
+    {
+      "epoch": 0.757085020242915,
+      "grad_norm": 0.3961510024581055,
+      "learning_rate": 3.9683478032281246e-05,
+      "loss": 0.3874,
+      "step": 187
+    },
+    {
+      "epoch": 0.7611336032388664,
+      "grad_norm": 0.4878205588742952,
+      "learning_rate": 3.967337762346568e-05,
+      "loss": 0.3732,
+      "step": 188
+    },
+    {
+      "epoch": 0.7651821862348178,
+      "grad_norm": 0.38351143681352556,
+      "learning_rate": 3.9663119906846885e-05,
+      "loss": 0.3741,
+      "step": 189
+    },
+    {
+      "epoch": 0.7692307692307693,
+      "grad_norm": 0.3809945829102143,
+      "learning_rate": 3.965270496444528e-05,
+      "loss": 0.3555,
+      "step": 190
+    },
+    {
+      "epoch": 0.7732793522267206,
+      "grad_norm": 0.4545258505191857,
+      "learning_rate": 3.9642132879538465e-05,
+      "loss": 0.3825,
+      "step": 191
+    },
+    {
+      "epoch": 0.7773279352226721,
+      "grad_norm": 0.4127862365668851,
+      "learning_rate": 3.9631403736660566e-05,
+      "loss": 0.3906,
+      "step": 192
+    },
+    {
+      "epoch": 0.7813765182186235,
+      "grad_norm": 0.44245687487974644,
+      "learning_rate": 3.962051762160151e-05,
+      "loss": 0.3882,
+      "step": 193
+    },
+    {
+      "epoch": 0.7854251012145749,
+      "grad_norm": 0.4302155784844054,
+      "learning_rate": 3.960947462140638e-05,
+      "loss": 0.3999,
+      "step": 194
+    },
+    {
+      "epoch": 0.7894736842105263,
+      "grad_norm": 0.5111409514039196,
+      "learning_rate": 3.959827482437472e-05,
+      "loss": 0.4124,
+      "step": 195
+    },
+    {
+      "epoch": 0.7935222672064778,
+      "grad_norm": 0.45780012221405236,
+      "learning_rate": 3.958691832005981e-05,
+      "loss": 0.3831,
+      "step": 196
+    },
+    {
+      "epoch": 0.7975708502024291,
+      "grad_norm": 0.4482786709169187,
+      "learning_rate": 3.9575405199267944e-05,
+      "loss": 0.3753,
+      "step": 197
+    },
+    {
+      "epoch": 0.8016194331983806,
+      "grad_norm": 0.5257515107380353,
+      "learning_rate": 3.956373555405774e-05,
+      "loss": 0.3868,
+      "step": 198
+    },
+    {
+      "epoch": 0.805668016194332,
+      "grad_norm": 0.4610489989535344,
+      "learning_rate": 3.955190947773937e-05,
+      "loss": 0.3844,
+      "step": 199
+    },
+    {
+      "epoch": 0.8097165991902834,
+      "grad_norm": 0.4840698327670045,
+      "learning_rate": 3.9539927064873824e-05,
+      "loss": 0.3813,
+      "step": 200
+    },
+    {
+      "epoch": 0.8137651821862348,
+      "grad_norm": 0.49004035371747273,
+      "learning_rate": 3.952778841127214e-05,
+      "loss": 0.394,
+      "step": 201
+    },
+    {
+      "epoch": 0.8178137651821862,
+      "grad_norm": 0.45114913935600626,
+      "learning_rate": 3.95154936139947e-05,
+      "loss": 0.3877,
+      "step": 202
+    },
+    {
+      "epoch": 0.8218623481781376,
+      "grad_norm": 0.5234751282751309,
+      "learning_rate": 3.950304277135034e-05,
+      "loss": 0.3695,
+      "step": 203
+    },
+    {
+      "epoch": 0.8259109311740891,
+      "grad_norm": 0.46028373977422005,
+      "learning_rate": 3.94904359828957e-05,
+      "loss": 0.3677,
+      "step": 204
+    },
+    {
+      "epoch": 0.8299595141700404,
+      "grad_norm": 0.47312773498704647,
+      "learning_rate": 3.94776733494343e-05,
+      "loss": 0.3918,
+      "step": 205
+    },
+    {
+      "epoch": 0.8340080971659919,
+      "grad_norm": 0.40243465255732586,
+      "learning_rate": 3.9464754973015826e-05,
+      "loss": 0.3759,
+      "step": 206
+    },
+    {
+      "epoch": 0.8380566801619433,
+      "grad_norm": 0.46426484089175446,
+      "learning_rate": 3.9451680956935275e-05,
+      "loss": 0.3688,
+      "step": 207
+    },
+    {
+      "epoch": 0.8421052631578947,
+      "grad_norm": 0.4843157186248196,
+      "learning_rate": 3.943845140573213e-05,
+      "loss": 0.3657,
+      "step": 208
+    },
+    {
+      "epoch": 0.8461538461538461,
+      "grad_norm": 0.4933624063466679,
+      "learning_rate": 3.942506642518952e-05,
+      "loss": 0.3781,
+      "step": 209
+    },
+    {
+      "epoch": 0.8502024291497976,
+      "grad_norm": 0.553727849261982,
+      "learning_rate": 3.9411526122333396e-05,
+      "loss": 0.3932,
+      "step": 210
+    },
+    {
+      "epoch": 0.854251012145749,
+      "grad_norm": 0.4924549968122828,
+      "learning_rate": 3.939783060543167e-05,
+      "loss": 0.3732,
+      "step": 211
+    },
+    {
+      "epoch": 0.8582995951417004,
+      "grad_norm": 0.5375696994594666,
+      "learning_rate": 3.938397998399333e-05,
+      "loss": 0.3997,
+      "step": 212
+    },
+    {
+      "epoch": 0.8623481781376519,
+      "grad_norm": 0.4844112194526399,
+      "learning_rate": 3.936997436876756e-05,
+      "loss": 0.3761,
+      "step": 213
+    },
+    {
+      "epoch": 0.8663967611336032,
+      "grad_norm": 0.435236586649903,
+      "learning_rate": 3.935581387174289e-05,
+      "loss": 0.3775,
+      "step": 214
+    },
+    {
+      "epoch": 0.8704453441295547,
+      "grad_norm": 0.5737411764604883,
+      "learning_rate": 3.9341498606146295e-05,
+      "loss": 0.3854,
+      "step": 215
+    },
+    {
+      "epoch": 0.8744939271255061,
+      "grad_norm": 0.44261126869304446,
+      "learning_rate": 3.932702868644224e-05,
+      "loss": 0.3799,
+      "step": 216
+    },
+    {
+      "epoch": 0.8785425101214575,
+      "grad_norm": 0.5455289063343999,
+      "learning_rate": 3.931240422833181e-05,
+      "loss": 0.3687,
+      "step": 217
+    },
+    {
+      "epoch": 0.8825910931174089,
+      "grad_norm": 0.45681208005560725,
+      "learning_rate": 3.9297625348751806e-05,
+      "loss": 0.3777,
+      "step": 218
+    },
+    {
+      "epoch": 0.8866396761133604,
+      "grad_norm": 0.53126494124608,
+      "learning_rate": 3.9282692165873744e-05,
+      "loss": 0.3933,
+      "step": 219
+    },
+    {
+      "epoch": 0.8906882591093117,
+      "grad_norm": 0.5143209676885587,
+      "learning_rate": 3.9267604799102945e-05,
+      "loss": 0.3972,
+      "step": 220
+    },
+    {
+      "epoch": 0.8947368421052632,
+      "grad_norm": 0.609388114264725,
+      "learning_rate": 3.9252363369077615e-05,
+      "loss": 0.373,
+      "step": 221
+    },
+    {
+      "epoch": 0.8987854251012146,
+      "grad_norm": 0.47035111925504147,
+      "learning_rate": 3.9236967997667796e-05,
+      "loss": 0.3832,
+      "step": 222
+    },
+    {
+      "epoch": 0.902834008097166,
+      "grad_norm": 0.614664003226842,
+      "learning_rate": 3.9221418807974496e-05,
+      "loss": 0.3885,
+      "step": 223
+    },
+    {
+      "epoch": 0.9068825910931174,
+      "grad_norm": 0.3759060244427104,
+      "learning_rate": 3.92057159243286e-05,
+      "loss": 0.391,
+      "step": 224
+    },
+    {
+      "epoch": 0.9109311740890689,
+      "grad_norm": 0.4804216262526576,
+      "learning_rate": 3.9189859472289956e-05,
+      "loss": 0.3913,
+      "step": 225
+    },
+    {
+      "epoch": 0.9149797570850202,
+      "grad_norm": 0.4442047102038536,
+      "learning_rate": 3.9173849578646324e-05,
+      "loss": 0.4015,
+      "step": 226
+    },
+    {
+      "epoch": 0.9190283400809717,
+      "grad_norm": 0.4529637098914157,
+      "learning_rate": 3.9157686371412395e-05,
+      "loss": 0.387,
+      "step": 227
+    },
+    {
+      "epoch": 0.9230769230769231,
+      "grad_norm": 0.53900982706863,
+      "learning_rate": 3.914136997982874e-05,
+      "loss": 0.3725,
+      "step": 228
+    },
+    {
+      "epoch": 0.9271255060728745,
+      "grad_norm": 0.5300123015005833,
+      "learning_rate": 3.912490053436079e-05,
+      "loss": 0.3725,
+      "step": 229
+    },
+    {
+      "epoch": 0.9311740890688259,
+      "grad_norm": 0.5012710473007886,
+      "learning_rate": 3.9108278166697805e-05,
+      "loss": 0.3982,
+      "step": 230
+    },
+    {
+      "epoch": 0.9352226720647774,
+      "grad_norm": 0.6153883391595075,
+      "learning_rate": 3.909150300975178e-05,
+      "loss": 0.3733,
+      "step": 231
+    },
+    {
+      "epoch": 0.9392712550607287,
+      "grad_norm": 0.4260667524916343,
+      "learning_rate": 3.907457519765642e-05,
+      "loss": 0.3872,
+      "step": 232
+    },
+    {
+      "epoch": 0.9433198380566802,
+      "grad_norm": 0.4773811964588878,
+      "learning_rate": 3.905749486576607e-05,
+      "loss": 0.3822,
+      "step": 233
+    },
+    {
+      "epoch": 0.9473684210526315,
+      "grad_norm": 0.4684802295583382,
+      "learning_rate": 3.90402621506546e-05,
+      "loss": 0.3709,
+      "step": 234
+    },
+    {
+      "epoch": 0.951417004048583,
+      "grad_norm": 0.4618438137747802,
+      "learning_rate": 3.902287719011435e-05,
+      "loss": 0.3815,
+      "step": 235
+    },
+    {
+      "epoch": 0.9554655870445344,
+      "grad_norm": 0.5021742192925732,
+      "learning_rate": 3.900534012315499e-05,
+      "loss": 0.3738,
+      "step": 236
+    },
+    {
+      "epoch": 0.9595141700404858,
+      "grad_norm": 0.5281679432931484,
+      "learning_rate": 3.8987651090002454e-05,
+      "loss": 0.3734,
+      "step": 237
+    },
+    {
+      "epoch": 0.9635627530364372,
+      "grad_norm": 0.44307996132829164,
+      "learning_rate": 3.896981023209777e-05,
+      "loss": 0.3826,
+      "step": 238
+    },
+    {
+      "epoch": 0.9676113360323887,
+      "grad_norm": 0.5144153436659205,
+      "learning_rate": 3.8951817692095966e-05,
+      "loss": 0.3773,
+      "step": 239
+    },
+    {
+      "epoch": 0.97165991902834,
+      "grad_norm": 0.4807520942071718,
+      "learning_rate": 3.893367361386491e-05,
+      "loss": 0.365,
+      "step": 240
+    },
+    {
+      "epoch": 0.9757085020242915,
+      "grad_norm": 0.4755812964409327,
+      "learning_rate": 3.891537814248417e-05,
+      "loss": 0.3843,
+      "step": 241
+    },
+    {
+      "epoch": 0.979757085020243,
+      "grad_norm": 0.5019184328774097,
+      "learning_rate": 3.889693142424384e-05,
+      "loss": 0.3666,
+      "step": 242
+    },
+    {
+      "epoch": 0.9838056680161943,
+      "grad_norm": 0.4706497881376087,
+      "learning_rate": 3.8878333606643405e-05,
+      "loss": 0.3723,
+      "step": 243
+    },
+    {
+      "epoch": 0.9878542510121457,
+      "grad_norm": 0.4811876955349954,
+      "learning_rate": 3.885958483839049e-05,
+      "loss": 0.3631,
+      "step": 244
+    },
+    {
+      "epoch": 0.9919028340080972,
+      "grad_norm": 0.5815668079953942,
+      "learning_rate": 3.8840685269399786e-05,
+      "loss": 0.3735,
+      "step": 245
+    },
+    {
+      "epoch": 0.9959514170040485,
+      "grad_norm": 0.48091163974739365,
+      "learning_rate": 3.882163505079171e-05,
+      "loss": 0.3731,
+      "step": 246
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.4509356840032675,
+      "learning_rate": 3.880243433489132e-05,
+      "loss": 0.3888,
+      "step": 247
+    },
+    {
+      "epoch": 1.0040485829959513,
+      "grad_norm": 0.5572581377615234,
+      "learning_rate": 3.878308327522702e-05,
+      "loss": 0.3349,
+      "step": 248
+    },
+    {
+      "epoch": 1.008097165991903,
+      "grad_norm": 0.41065127974265225,
+      "learning_rate": 3.876358202652936e-05,
+      "loss": 0.3149,
+      "step": 249
+    },
+    {
+      "epoch": 1.0121457489878543,
+      "grad_norm": 0.6009767571892967,
+      "learning_rate": 3.8743930744729835e-05,
+      "loss": 0.3441,
+      "step": 250
+    },
+    {
+      "epoch": 1.0161943319838056,
+      "grad_norm": 0.46982603663011246,
+      "learning_rate": 3.8724129586959535e-05,
+      "loss": 0.329,
+      "step": 251
+    },
+    {
+      "epoch": 1.0202429149797572,
+      "grad_norm": 0.4855707465034801,
+      "learning_rate": 3.870417871154801e-05,
+      "loss": 0.309,
+      "step": 252
+    },
+    {
+      "epoch": 1.0242914979757085,
+      "grad_norm": 0.5432588791509528,
+      "learning_rate": 3.868407827802192e-05,
+      "loss": 0.3228,
+      "step": 253
+    },
+    {
+      "epoch": 1.0283400809716599,
+      "grad_norm": 0.4546597621219417,
+      "learning_rate": 3.8663828447103804e-05,
+      "loss": 0.3269,
+      "step": 254
+    },
+    {
+      "epoch": 1.0323886639676114,
+      "grad_norm": 0.48825808984888264,
+      "learning_rate": 3.8643429380710765e-05,
+      "loss": 0.3179,
+      "step": 255
+    },
+    {
+      "epoch": 1.0364372469635628,
+      "grad_norm": 0.44126150180405355,
+      "learning_rate": 3.862288124195319e-05,
+      "loss": 0.3203,
+      "step": 256
+    },
+    {
+      "epoch": 1.040485829959514,
+      "grad_norm": 0.4497015105145206,
+      "learning_rate": 3.8602184195133464e-05,
+      "loss": 0.3182,
+      "step": 257
+    },
+    {
+      "epoch": 1.0445344129554657,
+      "grad_norm": 0.467931869617115,
+      "learning_rate": 3.8581338405744604e-05,
+      "loss": 0.3217,
+      "step": 258
+    },
+    {
+      "epoch": 1.048582995951417,
+      "grad_norm": 0.434583895011882,
+      "learning_rate": 3.8560344040468996e-05,
+      "loss": 0.3235,
+      "step": 259
+    },
+    {
+      "epoch": 1.0526315789473684,
+      "grad_norm": 0.4701709878382886,
+      "learning_rate": 3.853920126717703e-05,
+      "loss": 0.3206,
+      "step": 260
+    },
+    {
+      "epoch": 1.05668016194332,
+      "grad_norm": 0.4252701004547774,
+      "learning_rate": 3.851791025492574e-05,
+      "loss": 0.3216,
+      "step": 261
+    },
+    {
+      "epoch": 1.0607287449392713,
+      "grad_norm": 0.45391910340942726,
+      "learning_rate": 3.8496471173957515e-05,
+      "loss": 0.3138,
+      "step": 262
+    },
+    {
+      "epoch": 1.0647773279352226,
+      "grad_norm": 0.4695293553183844,
+      "learning_rate": 3.847488419569866e-05,
+      "loss": 0.3161,
+      "step": 263
+    },
+    {
+      "epoch": 1.0688259109311742,
+      "grad_norm": 0.4693343014283496,
+      "learning_rate": 3.8453149492758077e-05,
+      "loss": 0.3073,
+      "step": 264
+    },
+    {
+      "epoch": 1.0728744939271255,
+      "grad_norm": 0.422723532632553,
+      "learning_rate": 3.8431267238925886e-05,
+      "loss": 0.318,
+      "step": 265
+    },
+    {
+      "epoch": 1.0769230769230769,
+      "grad_norm": 0.4188025309845467,
+      "learning_rate": 3.840923760917199e-05,
+      "loss": 0.3124,
+      "step": 266
+    },
+    {
+      "epoch": 1.0809716599190284,
+      "grad_norm": 0.4293427778823667,
+      "learning_rate": 3.838706077964473e-05,
+      "loss": 0.3172,
+      "step": 267
+    },
+    {
+      "epoch": 1.0850202429149798,
+      "grad_norm": 0.5024615112472096,
+      "learning_rate": 3.836473692766944e-05,
+      "loss": 0.3303,
+      "step": 268
+    },
+    {
+      "epoch": 1.0890688259109311,
+      "grad_norm": 0.3582524241644417,
+      "learning_rate": 3.834226623174705e-05,
+      "loss": 0.3248,
+      "step": 269
+    },
+    {
+      "epoch": 1.0931174089068827,
+      "grad_norm": 0.5193363339961823,
+      "learning_rate": 3.831964887155264e-05,
+      "loss": 0.312,
+      "step": 270
+    },
+    {
+      "epoch": 1.097165991902834,
+      "grad_norm": 0.41012677364109523,
+      "learning_rate": 3.829688502793403e-05,
+      "loss": 0.3177,
+      "step": 271
+    },
+    {
+      "epoch": 1.1012145748987854,
+      "grad_norm": 0.5339362954204108,
+      "learning_rate": 3.827397488291029e-05,
+      "loss": 0.3194,
+      "step": 272
+    },
+    {
+      "epoch": 1.1052631578947367,
+      "grad_norm": 0.48942659083201084,
+      "learning_rate": 3.8250918619670343e-05,
+      "loss": 0.3194,
+      "step": 273
+    },
+    {
+      "epoch": 1.1093117408906883,
+      "grad_norm": 0.34999812115895607,
+      "learning_rate": 3.822771642257145e-05,
+      "loss": 0.3175,
+      "step": 274
+    },
+    {
+      "epoch": 1.1133603238866396,
+      "grad_norm": 0.43327067808644804,
+      "learning_rate": 3.820436847713776e-05,
+      "loss": 0.3047,
+      "step": 275
+    },
+    {
+      "epoch": 1.117408906882591,
+      "grad_norm": 0.4063452220230677,
+      "learning_rate": 3.8180874970058826e-05,
+      "loss": 0.3215,
+      "step": 276
+    },
+    {
+      "epoch": 1.1214574898785425,
+      "grad_norm": 0.3754042933478335,
+      "learning_rate": 3.8157236089188103e-05,
+      "loss": 0.3148,
+      "step": 277
+    },
+    {
+      "epoch": 1.125506072874494,
+      "grad_norm": 0.45842473803773337,
+      "learning_rate": 3.813345202354145e-05,
+      "loss": 0.3232,
+      "step": 278
+    },
+    {
+      "epoch": 1.1295546558704452,
+      "grad_norm": 0.4188035673484333,
+      "learning_rate": 3.810952296329563e-05,
+      "loss": 0.3204,
+      "step": 279
+    },
+    {
+      "epoch": 1.1336032388663968,
+      "grad_norm": 0.3726397486596435,
+      "learning_rate": 3.808544909978676e-05,
+      "loss": 0.327,
+      "step": 280
+    },
+    {
+      "epoch": 1.1376518218623481,
+      "grad_norm": 0.4621911404019513,
+      "learning_rate": 3.806123062550882e-05,
+      "loss": 0.3248,
+      "step": 281
+    },
+    {
+      "epoch": 1.1417004048582995,
+      "grad_norm": 0.4109615393116153,
+      "learning_rate": 3.803686773411208e-05,
+      "loss": 0.3433,
+      "step": 282
+    },
+    {
+      "epoch": 1.145748987854251,
+      "grad_norm": 0.43789713033998123,
+      "learning_rate": 3.801236062040158e-05,
+      "loss": 0.3196,
+      "step": 283
+    },
+    {
+      "epoch": 1.1497975708502024,
+      "grad_norm": 0.4657741296850761,
+      "learning_rate": 3.798770948033553e-05,
+      "loss": 0.3411,
+      "step": 284
+    },
+    {
+      "epoch": 1.1538461538461537,
+      "grad_norm": 0.40367449292390845,
+      "learning_rate": 3.796291451102381e-05,
+      "loss": 0.3042,
+      "step": 285
+    },
+    {
+      "epoch": 1.1578947368421053,
+      "grad_norm": 0.4926490177912692,
+      "learning_rate": 3.793797591072631e-05,
+      "loss": 0.3224,
+      "step": 286
+    },
+    {
+      "epoch": 1.1619433198380567,
+      "grad_norm": 0.42774064308681875,
+      "learning_rate": 3.791289387885144e-05,
+      "loss": 0.3387,
+      "step": 287
+    },
+    {
+      "epoch": 1.165991902834008,
+      "grad_norm": 0.4255132745987894,
+      "learning_rate": 3.788766861595443e-05,
+      "loss": 0.3249,
+      "step": 288
+    },
+    {
+      "epoch": 1.1700404858299596,
+      "grad_norm": 0.44305850645557215,
+      "learning_rate": 3.7862300323735835e-05,
+      "loss": 0.3269,
+      "step": 289
+    },
+    {
+      "epoch": 1.174089068825911,
+      "grad_norm": 0.3493992201440272,
+      "learning_rate": 3.783678920503983e-05,
+      "loss": 0.3187,
+      "step": 290
+    },
+    {
+      "epoch": 1.1781376518218623,
+      "grad_norm": 0.4181824663331357,
+      "learning_rate": 3.781113546385265e-05,
+      "loss": 0.3311,
+      "step": 291
+    },
+    {
+      "epoch": 1.1821862348178138,
+      "grad_norm": 0.3904168967679695,
+      "learning_rate": 3.7785339305300916e-05,
+      "loss": 0.3166,
+      "step": 292
+    },
+    {
+      "epoch": 1.1862348178137652,
+      "grad_norm": 0.40670106559031116,
+      "learning_rate": 3.7759400935650025e-05,
+      "loss": 0.3286,
+      "step": 293
+    },
+    {
+      "epoch": 1.1902834008097165,
+      "grad_norm": 0.457355248296084,
+      "learning_rate": 3.77333205623025e-05,
+      "loss": 0.3062,
+      "step": 294
+    },
+    {
+      "epoch": 1.194331983805668,
+      "grad_norm": 0.41134318193408037,
+      "learning_rate": 3.7707098393796295e-05,
+      "loss": 0.3285,
+      "step": 295
+    },
+    {
+      "epoch": 1.1983805668016194,
+      "grad_norm": 0.48239864236675767,
+      "learning_rate": 3.7680734639803185e-05,
+      "loss": 0.3274,
+      "step": 296
+    },
+    {
+      "epoch": 1.2024291497975708,
+      "grad_norm": 0.41386146405011925,
+      "learning_rate": 3.765422951112704e-05,
+      "loss": 0.3246,
+      "step": 297
+    },
+    {
+      "epoch": 1.2064777327935223,
+      "grad_norm": 0.4275059079918467,
+      "learning_rate": 3.762758321970216e-05,
+      "loss": 0.3368,
+      "step": 298
+    },
+    {
+      "epoch": 1.2105263157894737,
+      "grad_norm": 0.4083797268200254,
+      "learning_rate": 3.7600795978591584e-05,
+      "loss": 0.3199,
+      "step": 299
+    },
+    {
+      "epoch": 1.214574898785425,
+      "grad_norm": 0.4206150734603961,
+      "learning_rate": 3.757386800198538e-05,
+      "loss": 0.3178,
+      "step": 300
+    },
+    {
+      "epoch": 1.2186234817813766,
+      "grad_norm": 0.39709287031232504,
+      "learning_rate": 3.7546799505198925e-05,
+      "loss": 0.3099,
+      "step": 301
+    },
+    {
+      "epoch": 1.222672064777328,
+      "grad_norm": 0.41126758866528396,
+      "learning_rate": 3.7519590704671197e-05,
+      "loss": 0.3102,
+      "step": 302
+    },
+    {
+      "epoch": 1.2267206477732793,
+      "grad_norm": 0.4147977651058878,
+      "learning_rate": 3.749224181796305e-05,
+      "loss": 0.3381,
+      "step": 303
+    },
+    {
+      "epoch": 1.2307692307692308,
+      "grad_norm": 0.3819212898538071,
+      "learning_rate": 3.746475306375545e-05,
+      "loss": 0.3325,
+      "step": 304
+    },
+    {
+      "epoch": 1.2348178137651822,
+      "grad_norm": 0.4254602826336854,
+      "learning_rate": 3.743712466184774e-05,
+      "loss": 0.3201,
+      "step": 305
+    },
+    {
+      "epoch": 1.2388663967611335,
+      "grad_norm": 0.5152127780036394,
+      "learning_rate": 3.7409356833155885e-05,
+      "loss": 0.337,
+      "step": 306
+    },
+    {
+      "epoch": 1.242914979757085,
+      "grad_norm": 0.42966155326455174,
+      "learning_rate": 3.73814497997107e-05,
+      "loss": 0.3335,
+      "step": 307
+    },
+    {
+      "epoch": 1.2469635627530364,
+      "grad_norm": 0.5333909509132816,
+      "learning_rate": 3.7353403784656095e-05,
+      "loss": 0.3299,
+      "step": 308
+    },
+    {
+      "epoch": 1.2510121457489878,
+      "grad_norm": 0.39028824283995883,
+      "learning_rate": 3.732521901224724e-05,
+      "loss": 0.3377,
+      "step": 309
+    },
+    {
+      "epoch": 1.2550607287449393,
+      "grad_norm": 0.4767587775757094,
+      "learning_rate": 3.7296895707848844e-05,
+      "loss": 0.3113,
+      "step": 310
+    },
+    {
+      "epoch": 1.2591093117408907,
+      "grad_norm": 0.3939631012576733,
+      "learning_rate": 3.7268434097933275e-05,
+      "loss": 0.3293,
+      "step": 311
+    },
+    {
+      "epoch": 1.263157894736842,
+      "grad_norm": 0.41609649609807714,
+      "learning_rate": 3.723983441007882e-05,
+      "loss": 0.316,
+      "step": 312
+    },
+    {
+      "epoch": 1.2672064777327936,
+      "grad_norm": 0.4208973620060245,
+      "learning_rate": 3.721109687296781e-05,
+      "loss": 0.3274,
+      "step": 313
+    },
+    {
+      "epoch": 1.271255060728745,
+      "grad_norm": 0.390466469362456,
+      "learning_rate": 3.718222171638484e-05,
+      "loss": 0.3224,
+      "step": 314
+    },
+    {
+      "epoch": 1.2753036437246963,
+      "grad_norm": 0.3473704400620932,
+      "learning_rate": 3.715320917121488e-05,
+      "loss": 0.3225,
+      "step": 315
+    },
+    {
+      "epoch": 1.2793522267206479,
+      "grad_norm": 0.40948573090244206,
+      "learning_rate": 3.712405946944149e-05,
+      "loss": 0.3478,
+      "step": 316
+    },
+    {
+      "epoch": 1.2834008097165992,
+      "grad_norm": 0.36239037710218797,
+      "learning_rate": 3.709477284414488e-05,
+      "loss": 0.306,
+      "step": 317
+    },
+    {
+      "epoch": 1.2874493927125505,
+      "grad_norm": 0.4865837113609504,
+      "learning_rate": 3.7065349529500136e-05,
+      "loss": 0.3235,
+      "step": 318
+    },
+    {
+      "epoch": 1.291497975708502,
+      "grad_norm": 0.4296744521210969,
+      "learning_rate": 3.7035789760775305e-05,
+      "loss": 0.3142,
+      "step": 319
+    },
+    {
+      "epoch": 1.2955465587044535,
+      "grad_norm": 0.5260541600486818,
+      "learning_rate": 3.70060937743295e-05,
+      "loss": 0.3238,
+      "step": 320
+    },
+    {
+      "epoch": 1.2995951417004048,
+      "grad_norm": 0.46019920188248814,
+      "learning_rate": 3.6976261807611045e-05,
+      "loss": 0.3209,
+      "step": 321
+    },
+    {
+      "epoch": 1.3036437246963564,
+      "grad_norm": 0.45440615212350854,
+      "learning_rate": 3.694629409915555e-05,
+      "loss": 0.3361,
+      "step": 322
+    },
+    {
+      "epoch": 1.3076923076923077,
+      "grad_norm": 0.41149369081588866,
+      "learning_rate": 3.691619088858401e-05,
+      "loss": 0.3159,
+      "step": 323
+    },
+    {
+      "epoch": 1.311740890688259,
+      "grad_norm": 0.3998869544460914,
+      "learning_rate": 3.68859524166009e-05,
+      "loss": 0.3182,
+      "step": 324
+    },
+    {
+      "epoch": 1.3157894736842106,
+      "grad_norm": 0.3624608398612328,
+      "learning_rate": 3.6855578924992226e-05,
+      "loss": 0.3256,
+      "step": 325
+    },
+    {
+      "epoch": 1.319838056680162,
+      "grad_norm": 0.4032281575808946,
+      "learning_rate": 3.6825070656623626e-05,
+      "loss": 0.3236,
+      "step": 326
+    },
+    {
+      "epoch": 1.3238866396761133,
+      "grad_norm": 0.36326082733564324,
+      "learning_rate": 3.67944278554384e-05,
+      "loss": 0.3108,
+      "step": 327
+    },
+    {
+      "epoch": 1.3279352226720649,
+      "grad_norm": 0.4039389061871749,
+      "learning_rate": 3.676365076645557e-05,
+      "loss": 0.3132,
+      "step": 328
+    },
+    {
+      "epoch": 1.3319838056680162,
+      "grad_norm": 0.38645107708969006,
+      "learning_rate": 3.673273963576791e-05,
+      "loss": 0.3248,
+      "step": 329
+    },
+    {
+      "epoch": 1.3360323886639676,
+      "grad_norm": 0.3906114225596417,
+      "learning_rate": 3.670169471054001e-05,
+      "loss": 0.315,
+      "step": 330
+    },
+    {
+      "epoch": 1.3400809716599191,
+      "grad_norm": 0.4556071133073667,
+      "learning_rate": 3.6670516239006254e-05,
+      "loss": 0.3355,
+      "step": 331
+    },
+    {
+      "epoch": 1.3441295546558705,
+      "grad_norm": 0.3848988626228359,
+      "learning_rate": 3.663920447046888e-05,
+      "loss": 0.3105,
+      "step": 332
+    },
+    {
+      "epoch": 1.3481781376518218,
+      "grad_norm": 0.3906571819613406,
+      "learning_rate": 3.6607759655295954e-05,
+      "loss": 0.331,
+      "step": 333
+    },
+    {
+      "epoch": 1.3522267206477734,
+      "grad_norm": 0.3850925574600048,
+      "learning_rate": 3.657618204491937e-05,
+      "loss": 0.3049,
+      "step": 334
+    },
+    {
+      "epoch": 1.3562753036437247,
+      "grad_norm": 0.34467492311310594,
+      "learning_rate": 3.6544471891832875e-05,
+      "loss": 0.3136,
+      "step": 335
+    },
+    {
+      "epoch": 1.360323886639676,
+      "grad_norm": 0.38525939625248157,
+      "learning_rate": 3.6512629449589983e-05,
+      "loss": 0.3154,
+      "step": 336
+    },
+    {
+      "epoch": 1.3643724696356276,
+      "grad_norm": 0.32636252135397664,
+      "learning_rate": 3.6480654972802045e-05,
+      "loss": 0.3373,
+      "step": 337
+    },
+    {
+      "epoch": 1.368421052631579,
+      "grad_norm": 0.43498756918820614,
+      "learning_rate": 3.644854871713611e-05,
+      "loss": 0.3317,
+      "step": 338
+    },
+    {
+      "epoch": 1.3724696356275303,
+      "grad_norm": 0.3527032240010863,
+      "learning_rate": 3.6416310939312924e-05,
+      "loss": 0.3141,
+      "step": 339
+    },
+    {
+      "epoch": 1.376518218623482,
+      "grad_norm": 0.41253981326904815,
+      "learning_rate": 3.638394189710493e-05,
+      "loss": 0.3095,
+      "step": 340
+    },
+    {
+      "epoch": 1.3805668016194332,
+      "grad_norm": 0.42222751058665803,
+      "learning_rate": 3.635144184933412e-05,
+      "loss": 0.3178,
+      "step": 341
+    },
+    {
+      "epoch": 1.3846153846153846,
+      "grad_norm": 0.3787982878149811,
+      "learning_rate": 3.6318811055869994e-05,
+      "loss": 0.3286,
+      "step": 342
+    },
+    {
+      "epoch": 1.3886639676113361,
+      "grad_norm": 0.40254276701920544,
+      "learning_rate": 3.628604977762752e-05,
+      "loss": 0.3307,
+      "step": 343
+    },
+    {
+      "epoch": 1.3927125506072875,
+      "grad_norm": 0.34520536715521427,
+      "learning_rate": 3.6253158276565004e-05,
+      "loss": 0.3239,
+      "step": 344
+    },
+    {
+      "epoch": 1.3967611336032388,
+      "grad_norm": 0.45496635698307597,
+      "learning_rate": 3.6220136815682016e-05,
+      "loss": 0.3301,
+      "step": 345
+    },
+    {
+      "epoch": 1.4008097165991904,
+      "grad_norm": 0.36788839311297855,
+      "learning_rate": 3.618698565901727e-05,
+      "loss": 0.3227,
+      "step": 346
+    },
+    {
+      "epoch": 1.4048582995951417,
+      "grad_norm": 0.3903143606967964,
+      "learning_rate": 3.615370507164654e-05,
+      "loss": 0.3157,
+      "step": 347
+    },
+    {
+      "epoch": 1.408906882591093,
+      "grad_norm": 0.39549045614765066,
+      "learning_rate": 3.61202953196805e-05,
+      "loss": 0.3204,
+      "step": 348
+    },
+    {
+      "epoch": 1.4129554655870447,
+      "grad_norm": 0.33888129727454713,
+      "learning_rate": 3.6086756670262654e-05,
+      "loss": 0.3216,
+      "step": 349
+    },
+    {
+      "epoch": 1.417004048582996,
+      "grad_norm": 0.35660233734474206,
+      "learning_rate": 3.6053089391567135e-05,
+      "loss": 0.3235,
+      "step": 350
+    },
+    {
+      "epoch": 1.4210526315789473,
+      "grad_norm": 0.38519926891536616,
+      "learning_rate": 3.601929375279662e-05,
+      "loss": 0.3242,
+      "step": 351
+    },
+    {
+      "epoch": 1.425101214574899,
+      "grad_norm": 0.4419434302249723,
+      "learning_rate": 3.598537002418012e-05,
+      "loss": 0.3091,
+      "step": 352
+    },
+    {
+      "epoch": 1.4291497975708503,
+      "grad_norm": 0.3791331215539678,
+      "learning_rate": 3.5951318476970885e-05,
+      "loss": 0.3321,
+      "step": 353
+    },
+    {
+      "epoch": 1.4331983805668016,
+      "grad_norm": 0.4434242503779165,
+      "learning_rate": 3.591713938344416e-05,
+      "loss": 0.3447,
+      "step": 354
+    },
+    {
+      "epoch": 1.4372469635627532,
+      "grad_norm": 0.5051207527299485,
+      "learning_rate": 3.588283301689507e-05,
+      "loss": 0.3241,
+      "step": 355
+    },
+    {
+      "epoch": 1.4412955465587045,
+      "grad_norm": 0.43013920628618085,
+      "learning_rate": 3.5848399651636424e-05,
+      "loss": 0.3436,
+      "step": 356
+    },
+    {
+      "epoch": 1.4453441295546559,
+      "grad_norm": 0.4930612017785623,
+      "learning_rate": 3.5813839562996473e-05,
+      "loss": 0.3176,
+      "step": 357
+    },
+    {
+      "epoch": 1.4493927125506074,
+      "grad_norm": 0.5116760604834883,
+      "learning_rate": 3.5779153027316775e-05,
+      "loss": 0.3233,
+      "step": 358
+    },
+    {
+      "epoch": 1.4534412955465588,
+      "grad_norm": 0.4849192212346232,
+      "learning_rate": 3.574434032194994e-05,
+      "loss": 0.3344,
+      "step": 359
+    },
+    {
+      "epoch": 1.45748987854251,
+      "grad_norm": 0.47256155186379895,
+      "learning_rate": 3.5709401725257436e-05,
+      "loss": 0.3301,
+      "step": 360
+    },
+    {
+      "epoch": 1.4615384615384617,
+      "grad_norm": 0.4598082515641089,
+      "learning_rate": 3.5674337516607346e-05,
+      "loss": 0.3099,
+      "step": 361
+    },
+    {
+      "epoch": 1.465587044534413,
+      "grad_norm": 0.4221136409288697,
+      "learning_rate": 3.563914797637216e-05,
+      "loss": 0.3291,
+      "step": 362
+    },
+    {
+      "epoch": 1.4696356275303644,
+      "grad_norm": 0.45207601607703785,
+      "learning_rate": 3.560383338592649e-05,
+      "loss": 0.3204,
+      "step": 363
+    },
+    {
+      "epoch": 1.4736842105263157,
+      "grad_norm": 0.472213500054122,
+      "learning_rate": 3.556839402764487e-05,
+      "loss": 0.3326,
+      "step": 364
+    },
+    {
+      "epoch": 1.4777327935222673,
+      "grad_norm": 0.38729461085102423,
+      "learning_rate": 3.553283018489946e-05,
+      "loss": 0.3288,
+      "step": 365
+    },
+    {
+      "epoch": 1.4817813765182186,
+      "grad_norm": 0.444595334918768,
+      "learning_rate": 3.54971421420578e-05,
+      "loss": 0.3279,
+      "step": 366
+    },
+    {
+      "epoch": 1.48582995951417,
+      "grad_norm": 0.42171467811009067,
+      "learning_rate": 3.5461330184480524e-05,
+      "loss": 0.3411,
+      "step": 367
+    },
+    {
+      "epoch": 1.4898785425101215,
+      "grad_norm": 0.38575592096528233,
+      "learning_rate": 3.54253945985191e-05,
+      "loss": 0.3204,
+      "step": 368
+    },
+    {
+      "epoch": 1.4939271255060729,
+      "grad_norm": 0.43141559772545246,
+      "learning_rate": 3.5389335671513524e-05,
+      "loss": 0.3275,
+      "step": 369
+    },
+    {
+      "epoch": 1.4979757085020242,
+      "grad_norm": 0.4025597595743156,
+      "learning_rate": 3.5353153691789986e-05,
+      "loss": 0.3105,
+      "step": 370
+    },
+    {
+      "epoch": 1.5020242914979756,
+      "grad_norm": 0.42033291132798334,
+      "learning_rate": 3.531684894865866e-05,
+      "loss": 0.3152,
+      "step": 371
+    },
+    {
+      "epoch": 1.5060728744939271,
+      "grad_norm": 0.42853182634093523,
+      "learning_rate": 3.528042173241128e-05,
+      "loss": 0.3307,
+      "step": 372
+    },
+    {
+      "epoch": 1.5101214574898787,
+      "grad_norm": 0.4170201296098199,
+      "learning_rate": 3.524387233431892e-05,
+      "loss": 0.3156,
+      "step": 373
+    },
+    {
+      "epoch": 1.5141700404858298,
+      "grad_norm": 0.36577555129968875,
+      "learning_rate": 3.520720104662958e-05,
+      "loss": 0.3382,
+      "step": 374
+    },
+    {
+      "epoch": 1.5182186234817814,
+      "grad_norm": 0.3921960111459719,
+      "learning_rate": 3.5170408162565904e-05,
+      "loss": 0.3063,
+      "step": 375
+    },
+    {
+      "epoch": 1.522267206477733,
+      "grad_norm": 0.3887702794050287,
+      "learning_rate": 3.513349397632281e-05,
+      "loss": 0.3133,
+      "step": 376
+    },
+    {
+      "epoch": 1.526315789473684,
+      "grad_norm": 0.4404945777877588,
+      "learning_rate": 3.5096458783065145e-05,
+      "loss": 0.3257,
+      "step": 377
+    },
+    {
+      "epoch": 1.5303643724696356,
+      "grad_norm": 0.3832814619152055,
+      "learning_rate": 3.505930287892533e-05,
+      "loss": 0.3361,
+      "step": 378
+    },
+    {
+      "epoch": 1.5344129554655872,
+      "grad_norm": 0.4773073948935644,
+      "learning_rate": 3.502202656100099e-05,
+      "loss": 0.343,
+      "step": 379
+    },
+    {
+      "epoch": 1.5384615384615383,
+      "grad_norm": 0.41582206335345245,
+      "learning_rate": 3.498463012735256e-05,
+      "loss": 0.3259,
+      "step": 380
+    },
+    {
+      "epoch": 1.54251012145749,
+      "grad_norm": 0.4442067911306152,
+      "learning_rate": 3.494711387700094e-05,
+      "loss": 0.32,
+      "step": 381
+    },
+    {
+      "epoch": 1.5465587044534415,
+      "grad_norm": 0.39235099485285707,
+      "learning_rate": 3.490947810992505e-05,
+      "loss": 0.3234,
+      "step": 382
+    },
+    {
+      "epoch": 1.5506072874493926,
+      "grad_norm": 0.3777093759789124,
+      "learning_rate": 3.48717231270595e-05,
+      "loss": 0.3165,
+      "step": 383
+    },
+    {
+      "epoch": 1.5546558704453441,
+      "grad_norm": 0.39453876030772544,
+      "learning_rate": 3.483384923029211e-05,
+      "loss": 0.3321,
+      "step": 384
+    },
+    {
+      "epoch": 1.5587044534412957,
+      "grad_norm": 0.3789580656635722,
+      "learning_rate": 3.479585672246156e-05,
+      "loss": 0.306,
+      "step": 385
+    },
+    {
+      "epoch": 1.5627530364372468,
+      "grad_norm": 0.35651193990710806,
+      "learning_rate": 3.4757745907354924e-05,
+      "loss": 0.3107,
+      "step": 386
+    },
+    {
+      "epoch": 1.5668016194331984,
+      "grad_norm": 0.5000266073244674,
+      "learning_rate": 3.471951708970527e-05,
+      "loss": 0.3412,
+      "step": 387
+    },
+    {
+      "epoch": 1.5708502024291497,
+      "grad_norm": 0.36120999044589075,
+      "learning_rate": 3.468117057518921e-05,
+      "loss": 0.3257,
+      "step": 388
+    },
+    {
+      "epoch": 1.574898785425101,
+      "grad_norm": 0.4100501000292806,
+      "learning_rate": 3.4642706670424426e-05,
+      "loss": 0.3332,
+      "step": 389
+    },
+    {
+      "epoch": 1.5789473684210527,
+      "grad_norm": 0.44762896140367725,
+      "learning_rate": 3.460412568296731e-05,
+      "loss": 0.3379,
+      "step": 390
+    },
+    {
+      "epoch": 1.582995951417004,
+      "grad_norm": 0.3688657660896109,
+      "learning_rate": 3.456542792131039e-05,
+      "loss": 0.3405,
+      "step": 391
+    },
+    {
+      "epoch": 1.5870445344129553,
+      "grad_norm": 0.4235840718444496,
+      "learning_rate": 3.4526613694879936e-05,
+      "loss": 0.3077,
+      "step": 392
+    },
+    {
+      "epoch": 1.591093117408907,
+      "grad_norm": 0.3639024927517413,
+      "learning_rate": 3.448768331403348e-05,
+      "loss": 0.3291,
+      "step": 393
+    },
+    {
+      "epoch": 1.5951417004048583,
+      "grad_norm": 0.41921320378989196,
+      "learning_rate": 3.44486370900573e-05,
+      "loss": 0.3229,
+      "step": 394
+    },
+    {
+      "epoch": 1.5991902834008096,
+      "grad_norm": 0.37080739979921495,
+      "learning_rate": 3.440947533516396e-05,
+      "loss": 0.3291,
+      "step": 395
+    },
+    {
+      "epoch": 1.6032388663967612,
+      "grad_norm": 0.4296905922061334,
+      "learning_rate": 3.4370198362489816e-05,
+      "loss": 0.3259,
+      "step": 396
+    },
+    {
+      "epoch": 1.6072874493927125,
+      "grad_norm": 0.3272752599427859,
+      "learning_rate": 3.43308064860925e-05,
+      "loss": 0.3394,
+      "step": 397
+    },
+    {
+      "epoch": 1.6113360323886639,
+      "grad_norm": 0.4391690079988431,
+      "learning_rate": 3.4291300020948393e-05,
+      "loss": 0.3384,
+      "step": 398
+    },
+    {
+      "epoch": 1.6153846153846154,
+      "grad_norm": 0.36148334875813115,
+      "learning_rate": 3.425167928295015e-05,
+      "loss": 0.3066,
+      "step": 399
+    },
+    {
+      "epoch": 1.6194331983805668,
+      "grad_norm": 0.4146354913414668,
+      "learning_rate": 3.421194458890411e-05,
+      "loss": 0.3139,
+      "step": 400
+    },
+    {
+      "epoch": 1.623481781376518,
+      "grad_norm": 0.37666738764667373,
+      "learning_rate": 3.4172096256527865e-05,
+      "loss": 0.3199,
+      "step": 401
+    },
+    {
+      "epoch": 1.6275303643724697,
+      "grad_norm": 0.33278272669425274,
+      "learning_rate": 3.4132134604447595e-05,
+      "loss": 0.3077,
+      "step": 402
+    },
+    {
+      "epoch": 1.631578947368421,
+      "grad_norm": 0.3950448904819945,
+      "learning_rate": 3.409205995219561e-05,
+      "loss": 0.334,
+      "step": 403
+    },
+    {
+      "epoch": 1.6356275303643724,
+      "grad_norm": 0.35680792188548127,
+      "learning_rate": 3.4051872620207765e-05,
+      "loss": 0.3176,
+      "step": 404
+    },
+    {
+      "epoch": 1.639676113360324,
+      "grad_norm": 0.3338575072403953,
+      "learning_rate": 3.40115729298209e-05,
+      "loss": 0.3254,
+      "step": 405
+    },
+    {
+      "epoch": 1.6437246963562753,
+      "grad_norm": 0.4040105095492605,
+      "learning_rate": 3.397116120327025e-05,
+      "loss": 0.3095,
+      "step": 406
+    },
+    {
+      "epoch": 1.6477732793522266,
+      "grad_norm": 0.33546694695567714,
+      "learning_rate": 3.3930637763686916e-05,
+      "loss": 0.3218,
+      "step": 407
+    },
+    {
+      "epoch": 1.6518218623481782,
+      "grad_norm": 0.3831973914743605,
+      "learning_rate": 3.389000293509524e-05,
+      "loss": 0.3296,
+      "step": 408
+    },
+    {
+      "epoch": 1.6558704453441295,
+      "grad_norm": 0.41754091048815456,
+      "learning_rate": 3.3849257042410225e-05,
+      "loss": 0.3191,
+      "step": 409
+    },
+    {
+      "epoch": 1.6599190283400809,
+      "grad_norm": 0.3834294770330887,
+      "learning_rate": 3.380840041143494e-05,
+      "loss": 0.3479,
+      "step": 410
+    },
+    {
+      "epoch": 1.6639676113360324,
+      "grad_norm": 0.39906828580951514,
+      "learning_rate": 3.376743336885792e-05,
+      "loss": 0.3207,
+      "step": 411
+    },
+    {
+      "epoch": 1.6680161943319838,
+      "grad_norm": 0.4231103602024218,
+      "learning_rate": 3.372635624225055e-05,
+      "loss": 0.326,
+      "step": 412
+    },
+    {
+      "epoch": 1.6720647773279351,
+      "grad_norm": 0.38122525039532834,
+      "learning_rate": 3.3685169360064445e-05,
+      "loss": 0.3241,
+      "step": 413
+    },
+    {
+      "epoch": 1.6761133603238867,
+      "grad_norm": 0.35627566195368127,
+      "learning_rate": 3.3643873051628804e-05,
+      "loss": 0.3165,
+      "step": 414
+    },
+    {
+      "epoch": 1.680161943319838,
+      "grad_norm": 0.3620124786431755,
+      "learning_rate": 3.3602467647147814e-05,
+      "loss": 0.328,
+      "step": 415
+    },
+    {
+      "epoch": 1.6842105263157894,
+      "grad_norm": 0.4175393230465348,
+      "learning_rate": 3.356095347769799e-05,
+      "loss": 0.3342,
+      "step": 416
+    },
+    {
+      "epoch": 1.688259109311741,
+      "grad_norm": 0.355891215863639,
+      "learning_rate": 3.351933087522552e-05,
+      "loss": 0.3229,
+      "step": 417
+    },
+    {
+      "epoch": 1.6923076923076923,
+      "grad_norm": 0.36983568284399787,
+      "learning_rate": 3.3477600172543626e-05,
+      "loss": 0.3074,
+      "step": 418
+    },
+    {
+      "epoch": 1.6963562753036436,
+      "grad_norm": 0.36963191191286904,
+      "learning_rate": 3.3435761703329894e-05,
+      "loss": 0.3282,
+      "step": 419
+    },
+    {
+      "epoch": 1.7004048582995952,
+      "grad_norm": 0.3963905384383013,
+      "learning_rate": 3.3393815802123606e-05,
+      "loss": 0.3249,
+      "step": 420
+    },
+    {
+      "epoch": 1.7044534412955465,
+      "grad_norm": 0.34255123087154354,
+      "learning_rate": 3.335176280432308e-05,
+      "loss": 0.3225,
+      "step": 421
+    },
+    {
+      "epoch": 1.708502024291498,
+      "grad_norm": 0.4086537778245258,
+      "learning_rate": 3.330960304618295e-05,
+      "loss": 0.3284,
+      "step": 422
+    },
+    {
+      "epoch": 1.7125506072874495,
+      "grad_norm": 0.37633447114373525,
+      "learning_rate": 3.326733686481153e-05,
+      "loss": 0.3264,
+      "step": 423
+    },
+    {
+      "epoch": 1.7165991902834008,
+      "grad_norm": 0.4466116735864922,
+      "learning_rate": 3.3224964598168084e-05,
+      "loss": 0.3243,
+      "step": 424
+    },
+    {
+      "epoch": 1.7206477732793521,
+      "grad_norm": 0.3873421673794995,
+      "learning_rate": 3.318248658506012e-05,
+      "loss": 0.319,
+      "step": 425
+    },
+    {
+      "epoch": 1.7246963562753037,
+      "grad_norm": 0.41843131803598854,
+      "learning_rate": 3.313990316514069e-05,
+      "loss": 0.3101,
+      "step": 426
+    },
+    {
+      "epoch": 1.728744939271255,
+      "grad_norm": 0.40430137738885813,
+      "learning_rate": 3.309721467890571e-05,
+      "loss": 0.3222,
+      "step": 427
+    },
+    {
+      "epoch": 1.7327935222672064,
+      "grad_norm": 0.3997277751146518,
+      "learning_rate": 3.305442146769115e-05,
+      "loss": 0.3195,
+      "step": 428
+    },
+    {
+      "epoch": 1.736842105263158,
+      "grad_norm": 0.3656737633697745,
+      "learning_rate": 3.30115238736704e-05,
+      "loss": 0.3175,
+      "step": 429
+    },
+    {
+      "epoch": 1.7408906882591093,
+      "grad_norm": 0.4093980760876903,
+      "learning_rate": 3.296852223985148e-05,
+      "loss": 0.325,
+      "step": 430
+    },
+    {
+      "epoch": 1.7449392712550607,
+      "grad_norm": 0.3530362170070343,
+      "learning_rate": 3.2925416910074315e-05,
+      "loss": 0.3313,
+      "step": 431
+    },
+    {
+      "epoch": 1.7489878542510122,
+      "grad_norm": 0.3810126053669929,
+      "learning_rate": 3.288220822900796e-05,
+      "loss": 0.3316,
+      "step": 432
+    },
+    {
+      "epoch": 1.7530364372469636,
+      "grad_norm": 0.3522277541957332,
+      "learning_rate": 3.283889654214788e-05,
+      "loss": 0.2966,
+      "step": 433
+    },
+    {
+      "epoch": 1.757085020242915,
+      "grad_norm": 0.38209512436641135,
+      "learning_rate": 3.279548219581317e-05,
+      "loss": 0.3151,
+      "step": 434
+    },
+    {
+      "epoch": 1.7611336032388665,
+      "grad_norm": 0.3679807783762031,
+      "learning_rate": 3.275196553714379e-05,
+      "loss": 0.3,
+      "step": 435
+    },
+    {
+      "epoch": 1.7651821862348178,
+      "grad_norm": 0.4178192658564299,
+      "learning_rate": 3.270834691409778e-05,
+      "loss": 0.3197,
+      "step": 436
+    },
+    {
+      "epoch": 1.7692307692307692,
+      "grad_norm": 0.33617261055917336,
+      "learning_rate": 3.266462667544849e-05,
+      "loss": 0.3168,
+      "step": 437
+    },
+    {
+      "epoch": 1.7732793522267207,
+      "grad_norm": 0.38723595354827317,
+      "learning_rate": 3.262080517078178e-05,
+      "loss": 0.3016,
+      "step": 438
+    },
+    {
+      "epoch": 1.777327935222672,
+      "grad_norm": 0.3335158948112509,
+      "learning_rate": 3.257688275049323e-05,
+      "loss": 0.3144,
+      "step": 439
+    },
+    {
+      "epoch": 1.7813765182186234,
+      "grad_norm": 0.38629505113547796,
+      "learning_rate": 3.253285976578535e-05,
+      "loss": 0.3267,
+      "step": 440
+    },
+    {
+      "epoch": 1.785425101214575,
+      "grad_norm": 0.3702301404109381,
+      "learning_rate": 3.2488736568664756e-05,
+      "loss": 0.3173,
+      "step": 441
+    },
+    {
+      "epoch": 1.7894736842105263,
+      "grad_norm": 0.34583451779129654,
+      "learning_rate": 3.244451351193935e-05,
+      "loss": 0.3206,
+      "step": 442
+    },
+    {
+      "epoch": 1.7935222672064777,
+      "grad_norm": 0.3925057813942527,
+      "learning_rate": 3.240019094921551e-05,
+      "loss": 0.3085,
+      "step": 443
+    },
+    {
+      "epoch": 1.7975708502024292,
+      "grad_norm": 0.3976517386157222,
+      "learning_rate": 3.235576923489527e-05,
+      "loss": 0.3194,
+      "step": 444
+    },
+    {
+      "epoch": 1.8016194331983806,
+      "grad_norm": 0.4290178399939706,
+      "learning_rate": 3.2311248724173476e-05,
+      "loss": 0.311,
+      "step": 445
+    },
+    {
+      "epoch": 1.805668016194332,
+      "grad_norm": 0.3573301717042051,
+      "learning_rate": 3.226662977303494e-05,
+      "loss": 0.3216,
+      "step": 446
+    },
+    {
+      "epoch": 1.8097165991902835,
+      "grad_norm": 0.3948301288484063,
+      "learning_rate": 3.2221912738251626e-05,
+      "loss": 0.3204,
+      "step": 447
+    },
+    {
+      "epoch": 1.8137651821862348,
+      "grad_norm": 0.3651655031382838,
+      "learning_rate": 3.2177097977379736e-05,
+      "loss": 0.3246,
+      "step": 448
+    },
+    {
+      "epoch": 1.8178137651821862,
+      "grad_norm": 0.39613879110273736,
+      "learning_rate": 3.21321858487569e-05,
+      "loss": 0.3203,
+      "step": 449
+    },
+    {
+      "epoch": 1.8218623481781377,
+      "grad_norm": 0.35887325446176044,
+      "learning_rate": 3.208717671149931e-05,
+      "loss": 0.3133,
+      "step": 450
+    },
+    {
+      "epoch": 1.825910931174089,
+      "grad_norm": 0.3533173235168531,
+      "learning_rate": 3.204207092549882e-05,
+      "loss": 0.3304,
+      "step": 451
+    },
+    {
+      "epoch": 1.8299595141700404,
+      "grad_norm": 0.39904440398284596,
+      "learning_rate": 3.19968688514201e-05,
+      "loss": 0.3212,
+      "step": 452
+    },
+    {
+      "epoch": 1.834008097165992,
+      "grad_norm": 0.3231148695967858,
+      "learning_rate": 3.195157085069774e-05,
+      "loss": 0.3275,
+      "step": 453
+    },
+    {
+      "epoch": 1.8380566801619433,
+      "grad_norm": 0.35747649421373745,
+      "learning_rate": 3.190617728553332e-05,
+      "loss": 0.3257,
+      "step": 454
+    },
+    {
+      "epoch": 1.8421052631578947,
+      "grad_norm": 0.352225262740757,
+      "learning_rate": 3.1860688518892606e-05,
+      "loss": 0.2996,
+      "step": 455
+    },
+    {
+      "epoch": 1.8461538461538463,
+      "grad_norm": 0.40086462077162616,
+      "learning_rate": 3.181510491450254e-05,
+      "loss": 0.3285,
+      "step": 456
+    },
+    {
+      "epoch": 1.8502024291497976,
+      "grad_norm": 0.326614585445406,
+      "learning_rate": 3.176942683684842e-05,
+      "loss": 0.3339,
+      "step": 457
+    },
+    {
+      "epoch": 1.854251012145749,
+      "grad_norm": 0.3969302533583997,
+      "learning_rate": 3.1723654651170934e-05,
+      "loss": 0.3118,
+      "step": 458
+    },
+    {
+      "epoch": 1.8582995951417005,
+      "grad_norm": 0.30684494868341394,
+      "learning_rate": 3.167778872346327e-05,
+      "loss": 0.315,
+      "step": 459
+    },
+    {
+      "epoch": 1.8623481781376519,
+      "grad_norm": 0.36772166606737844,
+      "learning_rate": 3.163182942046815e-05,
+      "loss": 0.3255,
+      "step": 460
+    },
+    {
+      "epoch": 1.8663967611336032,
+      "grad_norm": 0.36209055851995414,
+      "learning_rate": 3.158577710967494e-05,
+      "loss": 0.3205,
+      "step": 461
+    },
+    {
+      "epoch": 1.8704453441295548,
+      "grad_norm": 0.3788095260766203,
+      "learning_rate": 3.153963215931669e-05,
+      "loss": 0.2975,
+      "step": 462
+    },
+    {
+      "epoch": 1.874493927125506,
+      "grad_norm": 0.364654650921325,
+      "learning_rate": 3.14933949383672e-05,
+      "loss": 0.3294,
+      "step": 463
+    },
+    {
+      "epoch": 1.8785425101214575,
+      "grad_norm": 0.4284651069080952,
+      "learning_rate": 3.144706581653805e-05,
+      "loss": 0.3126,
+      "step": 464
+    },
+    {
+      "epoch": 1.882591093117409,
+      "grad_norm": 0.3672897408181576,
+      "learning_rate": 3.140064516427566e-05,
+      "loss": 0.3088,
+      "step": 465
+    },
+    {
+      "epoch": 1.8866396761133604,
+      "grad_norm": 0.3704717735791531,
+      "learning_rate": 3.135413335275833e-05,
+      "loss": 0.336,
+      "step": 466
+    },
+    {
+      "epoch": 1.8906882591093117,
+      "grad_norm": 0.46116653709519373,
+      "learning_rate": 3.130753075389327e-05,
+      "loss": 0.3243,
+      "step": 467
+    },
+    {
+      "epoch": 1.8947368421052633,
+      "grad_norm": 0.3947875978929145,
+      "learning_rate": 3.1260837740313625e-05,
+      "loss": 0.3168,
+      "step": 468
+    },
+    {
+      "epoch": 1.8987854251012146,
+      "grad_norm": 0.405807578911147,
+      "learning_rate": 3.121405468537547e-05,
+      "loss": 0.3042,
+      "step": 469
+    },
+    {
+      "epoch": 1.902834008097166,
+      "grad_norm": 0.36832163009890684,
+      "learning_rate": 3.116718196315488e-05,
+      "loss": 0.3143,
+      "step": 470
+    },
+    {
+      "epoch": 1.9068825910931175,
+      "grad_norm": 0.37315402996140035,
+      "learning_rate": 3.1120219948444894e-05,
+      "loss": 0.3264,
+      "step": 471
+    },
+    {
+      "epoch": 1.9109311740890689,
+      "grad_norm": 0.3640197151290267,
+      "learning_rate": 3.107316901675252e-05,
+      "loss": 0.305,
+      "step": 472
+    },
+    {
+      "epoch": 1.9149797570850202,
+      "grad_norm": 0.3580597128815145,
+      "learning_rate": 3.1026029544295745e-05,
+      "loss": 0.3228,
+      "step": 473
+    },
+    {
+      "epoch": 1.9190283400809718,
+      "grad_norm": 0.37643435765074384,
+      "learning_rate": 3.097880190800057e-05,
+      "loss": 0.3142,
+      "step": 474
+    },
+    {
+      "epoch": 1.9230769230769231,
+      "grad_norm": 0.37776496135684334,
+      "learning_rate": 3.093148648549788e-05,
+      "loss": 0.3237,
+      "step": 475
+    },
+    {
+      "epoch": 1.9271255060728745,
+      "grad_norm": 0.3570845535021339,
+      "learning_rate": 3.088408365512055e-05,
+      "loss": 0.3215,
+      "step": 476
+    },
+    {
+      "epoch": 1.931174089068826,
+      "grad_norm": 0.393313414316562,
+      "learning_rate": 3.083659379590034e-05,
+      "loss": 0.3176,
+      "step": 477
+    },
+    {
+      "epoch": 1.9352226720647774,
+      "grad_norm": 0.3567756107272182,
+      "learning_rate": 3.0789017287564913e-05,
+      "loss": 0.3212,
+      "step": 478
+    },
+    {
+      "epoch": 1.9392712550607287,
+      "grad_norm": 0.31877296313779435,
+      "learning_rate": 3.074135451053475e-05,
+      "loss": 0.3239,
+      "step": 479
+    },
+    {
+      "epoch": 1.9433198380566803,
+      "grad_norm": 0.37931311020033653,
+      "learning_rate": 3.069360584592015e-05,
+      "loss": 0.3177,
+      "step": 480
+    },
+    {
+      "epoch": 1.9473684210526314,
+      "grad_norm": 0.37599797235485777,
+      "learning_rate": 3.064577167551816e-05,
+      "loss": 0.319,
+      "step": 481
+    },
+    {
+      "epoch": 1.951417004048583,
+      "grad_norm": 0.33666659459533577,
+      "learning_rate": 3.059785238180953e-05,
+      "loss": 0.3267,
+      "step": 482
+    },
+    {
+      "epoch": 1.9554655870445345,
+      "grad_norm": 0.37146738454468514,
+      "learning_rate": 3.054984834795566e-05,
+      "loss": 0.3172,
+      "step": 483
+    },
+    {
+      "epoch": 1.9595141700404857,
+      "grad_norm": 0.36819771080086294,
+      "learning_rate": 3.0501759957795537e-05,
+      "loss": 0.3322,
+      "step": 484
+    },
+    {
+      "epoch": 1.9635627530364372,
+      "grad_norm": 0.3691353523665943,
+      "learning_rate": 3.0453587595842627e-05,
+      "loss": 0.315,
+      "step": 485
+    },
+    {
+      "epoch": 1.9676113360323888,
+      "grad_norm": 0.37084723892585214,
+      "learning_rate": 3.0405331647281856e-05,
+      "loss": 0.319,
+      "step": 486
+    },
+    {
+      "epoch": 1.97165991902834,
+      "grad_norm": 0.3805131079773682,
+      "learning_rate": 3.0356992497966508e-05,
+      "loss": 0.3096,
+      "step": 487
+    },
+    {
+      "epoch": 1.9757085020242915,
+      "grad_norm": 0.3834579259366537,
+      "learning_rate": 3.0308570534415137e-05,
+      "loss": 0.3194,
+      "step": 488
+    },
+    {
+      "epoch": 1.979757085020243,
+      "grad_norm": 0.3835705265856652,
+      "learning_rate": 3.0260066143808467e-05,
+      "loss": 0.3194,
+      "step": 489
+    },
+    {
+      "epoch": 1.9838056680161942,
+      "grad_norm": 0.3202447289135067,
+      "learning_rate": 3.021147971398632e-05,
+      "loss": 0.333,
+      "step": 490
+    },
+    {
+      "epoch": 1.9878542510121457,
+      "grad_norm": 0.4303776600028691,
+      "learning_rate": 3.0162811633444478e-05,
+      "loss": 0.3219,
+      "step": 491
+    },
+    {
+      "epoch": 1.9919028340080973,
+      "grad_norm": 0.32813224783849326,
+      "learning_rate": 3.0114062291331635e-05,
+      "loss": 0.2979,
+      "step": 492
+    },
+    {
+      "epoch": 1.9959514170040484,
+      "grad_norm": 0.3533569651740726,
+      "learning_rate": 3.0065232077446217e-05,
+      "loss": 0.3192,
+      "step": 493
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.3186166669335711,
+      "learning_rate": 3.001632138223332e-05,
+      "loss": 0.316,
+      "step": 494
+    },
+    {
+      "epoch": 2.0040485829959516,
+      "grad_norm": 0.4560487276987198,
+      "learning_rate": 2.9967330596781565e-05,
+      "loss": 0.2683,
+      "step": 495
+    },
+    {
+      "epoch": 2.0080971659919027,
+      "grad_norm": 0.3554238473200862,
+      "learning_rate": 2.9918260112819952e-05,
+      "loss": 0.2424,
+      "step": 496
+    },
+    {
+      "epoch": 2.0121457489878543,
+      "grad_norm": 0.6121066264878348,
+      "learning_rate": 2.9869110322714778e-05,
+      "loss": 0.2423,
+      "step": 497
+    },
+    {
+      "epoch": 2.016194331983806,
+      "grad_norm": 0.436610325579155,
+      "learning_rate": 2.9819881619466447e-05,
+      "loss": 0.2581,
+      "step": 498
+    },
+    {
+      "epoch": 2.020242914979757,
+      "grad_norm": 0.37169681724465564,
+      "learning_rate": 2.9770574396706364e-05,
+      "loss": 0.2569,
+      "step": 499
+    },
+    {
+      "epoch": 2.0242914979757085,
+      "grad_norm": 0.39898364423349814,
+      "learning_rate": 2.972118904869376e-05,
+      "loss": 0.2557,
+      "step": 500
+    },
+    {
+      "epoch": 2.02834008097166,
+      "grad_norm": 0.33915799296305293,
+      "learning_rate": 2.9671725970312563e-05,
+      "loss": 0.2598,
+      "step": 501
+    },
+    {
+      "epoch": 2.032388663967611,
+      "grad_norm": 0.3706054219804499,
+      "learning_rate": 2.9622185557068238e-05,
+      "loss": 0.2447,
+      "step": 502
+    },
+    {
+      "epoch": 2.0364372469635628,
+      "grad_norm": 0.4010863998780629,
+      "learning_rate": 2.95725682050846e-05,
+      "loss": 0.2495,
+      "step": 503
+    },
+    {
+      "epoch": 2.0404858299595143,
+      "grad_norm": 0.3758139978165267,
+      "learning_rate": 2.9522874311100684e-05,
+      "loss": 0.238,
+      "step": 504
+    },
+    {
+      "epoch": 2.0445344129554655,
+      "grad_norm": 0.38758398472304106,
+      "learning_rate": 2.9473104272467526e-05,
+      "loss": 0.2428,
+      "step": 505
+    },
+    {
+      "epoch": 2.048582995951417,
+      "grad_norm": 0.40953888235428804,
+      "learning_rate": 2.942325848714505e-05,
+      "loss": 0.2391,
+      "step": 506
+    },
+    {
+      "epoch": 2.0526315789473686,
+      "grad_norm": 0.3790298773736337,
+      "learning_rate": 2.9373337353698804e-05,
+      "loss": 0.2343,
+      "step": 507
+    },
+    {
+      "epoch": 2.0566801619433197,
+      "grad_norm": 0.41802372758856954,
+      "learning_rate": 2.932334127129686e-05,
+      "loss": 0.2614,
+      "step": 508
+    },
+    {
+      "epoch": 2.0607287449392713,
+      "grad_norm": 0.4139610154470518,
+      "learning_rate": 2.9273270639706547e-05,
+      "loss": 0.2544,
+      "step": 509
+    },
+    {
+      "epoch": 2.064777327935223,
+      "grad_norm": 0.3446473643807053,
+      "learning_rate": 2.922312585929131e-05,
+      "loss": 0.245,
+      "step": 510
+    },
+    {
+      "epoch": 2.068825910931174,
+      "grad_norm": 0.3821792533043968,
+      "learning_rate": 2.9172907331007478e-05,
+      "loss": 0.2336,
+      "step": 511
+    },
+    {
+      "epoch": 2.0728744939271255,
+      "grad_norm": 0.349146234034953,
+      "learning_rate": 2.9122615456401053e-05,
+      "loss": 0.252,
+      "step": 512
+    },
+    {
+      "epoch": 2.076923076923077,
+      "grad_norm": 0.4152655923026081,
+      "learning_rate": 2.9072250637604535e-05,
+      "loss": 0.248,
+      "step": 513
+    },
+    {
+      "epoch": 2.080971659919028,
+      "grad_norm": 0.3316176884215142,
+      "learning_rate": 2.9021813277333667e-05,
+      "loss": 0.2263,
+      "step": 514
+    },
+    {
+      "epoch": 2.08502024291498,
+      "grad_norm": 0.3821048382357758,
+      "learning_rate": 2.8971303778884244e-05,
+      "loss": 0.2344,
+      "step": 515
+    },
+    {
+      "epoch": 2.0890688259109313,
+      "grad_norm": 0.34508221553875673,
+      "learning_rate": 2.8920722546128863e-05,
+      "loss": 0.2678,
+      "step": 516
+    },
+    {
+      "epoch": 2.0931174089068825,
+      "grad_norm": 0.32113056003060875,
+      "learning_rate": 2.887006998351372e-05,
+      "loss": 0.2311,
+      "step": 517
+    },
+    {
+      "epoch": 2.097165991902834,
+      "grad_norm": 0.3297963442324316,
+      "learning_rate": 2.881934649605535e-05,
+      "loss": 0.2648,
+      "step": 518
+    },
+    {
+      "epoch": 2.1012145748987856,
+      "grad_norm": 0.3429233375917042,
+      "learning_rate": 2.8768552489337408e-05,
+      "loss": 0.2282,
+      "step": 519
+    },
+    {
+      "epoch": 2.1052631578947367,
+      "grad_norm": 0.34635197324560446,
+      "learning_rate": 2.8717688369507425e-05,
+      "loss": 0.2508,
+      "step": 520
+    },
+    {
+      "epoch": 2.1093117408906883,
+      "grad_norm": 0.3542479518204499,
+      "learning_rate": 2.8666754543273535e-05,
+      "loss": 0.2233,
+      "step": 521
+    },
+    {
+      "epoch": 2.11336032388664,
+      "grad_norm": 0.3539483376063812,
+      "learning_rate": 2.8615751417901273e-05,
+      "loss": 0.247,
+      "step": 522
+    },
+    {
+      "epoch": 2.117408906882591,
+      "grad_norm": 0.3490741561404649,
+      "learning_rate": 2.856467940121025e-05,
+      "loss": 0.2198,
+      "step": 523
+    },
+    {
+      "epoch": 2.1214574898785425,
+      "grad_norm": 0.39672484691334015,
+      "learning_rate": 2.8513538901570964e-05,
+      "loss": 0.2448,
+      "step": 524
+    },
+    {
+      "epoch": 2.125506072874494,
+      "grad_norm": 0.3604009132104185,
+      "learning_rate": 2.846233032790149e-05,
+      "loss": 0.2511,
+      "step": 525
+    },
+    {
+      "epoch": 2.1295546558704452,
+      "grad_norm": 0.34458732829360694,
+      "learning_rate": 2.84110540896642e-05,
+      "loss": 0.2365,
+      "step": 526
+    },
+    {
+      "epoch": 2.133603238866397,
+      "grad_norm": 0.3687986206275351,
+      "learning_rate": 2.8359710596862555e-05,
+      "loss": 0.2482,
+      "step": 527
+    },
+    {
+      "epoch": 2.1376518218623484,
+      "grad_norm": 0.3250732116238457,
+      "learning_rate": 2.8308300260037734e-05,
+      "loss": 0.2416,
+      "step": 528
+    },
+    {
+      "epoch": 2.1417004048582995,
+      "grad_norm": 0.3751692778551227,
+      "learning_rate": 2.8256823490265424e-05,
+      "loss": 0.2371,
+      "step": 529
+    },
+    {
+      "epoch": 2.145748987854251,
+      "grad_norm": 0.34442993999401766,
+      "learning_rate": 2.8205280699152515e-05,
+      "loss": 0.2524,
+      "step": 530
+    },
+    {
+      "epoch": 2.1497975708502026,
+      "grad_norm": 0.33024867635986266,
+      "learning_rate": 2.8153672298833777e-05,
+      "loss": 0.2495,
+      "step": 531
+    },
+    {
+      "epoch": 2.1538461538461537,
+      "grad_norm": 0.3085232890739815,
+      "learning_rate": 2.8101998701968615e-05,
+      "loss": 0.2378,
+      "step": 532
+    },
+    {
+      "epoch": 2.1578947368421053,
+      "grad_norm": 0.3593821692799787,
+      "learning_rate": 2.805026032173773e-05,
+      "loss": 0.2607,
+      "step": 533
+    },
+    {
+      "epoch": 2.161943319838057,
+      "grad_norm": 0.32585387850459285,
+      "learning_rate": 2.7998457571839832e-05,
+      "loss": 0.2342,
+      "step": 534
+    },
+    {
+      "epoch": 2.165991902834008,
+      "grad_norm": 0.3808265563970907,
+      "learning_rate": 2.7946590866488322e-05,
+      "loss": 0.2549,
+      "step": 535
+    },
+    {
+      "epoch": 2.1700404858299596,
+      "grad_norm": 0.3459287945780527,
+      "learning_rate": 2.7894660620408e-05,
+      "loss": 0.2482,
+      "step": 536
+    },
+    {
+      "epoch": 2.174089068825911,
+      "grad_norm": 0.3598757319143665,
+      "learning_rate": 2.784266724883173e-05,
+      "loss": 0.2508,
+      "step": 537
+    },
+    {
+      "epoch": 2.1781376518218623,
+      "grad_norm": 0.39099067252267733,
+      "learning_rate": 2.7790611167497128e-05,
+      "loss": 0.2289,
+      "step": 538
+    },
+    {
+      "epoch": 2.182186234817814,
+      "grad_norm": 0.35094795996312633,
+      "learning_rate": 2.7738492792643224e-05,
+      "loss": 0.2368,
+      "step": 539
+    },
+    {
+      "epoch": 2.1862348178137654,
+      "grad_norm": 0.3498796589133733,
+      "learning_rate": 2.768631254100716e-05,
+      "loss": 0.2308,
+      "step": 540
+    },
+    {
+      "epoch": 2.1902834008097165,
+      "grad_norm": 0.39584964566909264,
+      "learning_rate": 2.7634070829820836e-05,
+      "loss": 0.2394,
+      "step": 541
+    },
+    {
+      "epoch": 2.194331983805668,
+      "grad_norm": 0.35577829346066514,
+      "learning_rate": 2.7581768076807592e-05,
+      "loss": 0.2409,
+      "step": 542
+    },
+    {
+      "epoch": 2.1983805668016196,
+      "grad_norm": 0.32653864537318983,
+      "learning_rate": 2.7529404700178832e-05,
+      "loss": 0.2508,
+      "step": 543
+    },
+    {
+      "epoch": 2.2024291497975708,
+      "grad_norm": 0.3651388657715699,
+      "learning_rate": 2.7476981118630735e-05,
+      "loss": 0.2455,
+      "step": 544
+    },
+    {
+      "epoch": 2.2064777327935223,
+      "grad_norm": 0.3169612254707923,
+      "learning_rate": 2.742449775134085e-05,
+      "loss": 0.229,
+      "step": 545
+    },
+    {
+      "epoch": 2.2105263157894735,
+      "grad_norm": 0.3470505588985843,
+      "learning_rate": 2.7371955017964788e-05,
+      "loss": 0.2671,
+      "step": 546
+    },
+    {
+      "epoch": 2.214574898785425,
+      "grad_norm": 0.3360185696746806,
+      "learning_rate": 2.7319353338632845e-05,
+      "loss": 0.2299,
+      "step": 547
+    },
+    {
+      "epoch": 2.2186234817813766,
+      "grad_norm": 0.31341904823366695,
+      "learning_rate": 2.7266693133946645e-05,
+      "loss": 0.2536,
+      "step": 548
+    },
+    {
+      "epoch": 2.2226720647773277,
+      "grad_norm": 0.3002729962334511,
+      "learning_rate": 2.7213974824975785e-05,
+      "loss": 0.2426,
+      "step": 549
+    },
+    {
+      "epoch": 2.2267206477732793,
+      "grad_norm": 0.3323111215111652,
+      "learning_rate": 2.7161198833254464e-05,
+      "loss": 0.2401,
+      "step": 550
+    },
+    {
+      "epoch": 2.230769230769231,
+      "grad_norm": 0.33973773702580534,
+      "learning_rate": 2.7108365580778095e-05,
+      "loss": 0.2307,
+      "step": 551
+    },
+    {
+      "epoch": 2.234817813765182,
+      "grad_norm": 0.305915621273617,
+      "learning_rate": 2.705547548999996e-05,
+      "loss": 0.2564,
+      "step": 552
+    },
+    {
+      "epoch": 2.2388663967611335,
+      "grad_norm": 0.3256666081839865,
+      "learning_rate": 2.7002528983827817e-05,
+      "loss": 0.243,
+      "step": 553
+    },
+    {
+      "epoch": 2.242914979757085,
+      "grad_norm": 0.32884823854512124,
+      "learning_rate": 2.6949526485620517e-05,
+      "loss": 0.2479,
+      "step": 554
+    },
+    {
+      "epoch": 2.246963562753036,
+      "grad_norm": 0.3238758666788886,
+      "learning_rate": 2.6896468419184623e-05,
+      "loss": 0.257,
+      "step": 555
+    },
+    {
+      "epoch": 2.251012145748988,
+      "grad_norm": 0.3174306496395586,
+      "learning_rate": 2.6843355208771024e-05,
+      "loss": 0.2454,
+      "step": 556
+    },
+    {
+      "epoch": 2.2550607287449393,
+      "grad_norm": 0.3196426127537085,
+      "learning_rate": 2.6790187279071536e-05,
+      "loss": 0.244,
+      "step": 557
+    },
+    {
+      "epoch": 2.2591093117408905,
+      "grad_norm": 0.32138580918876464,
+      "learning_rate": 2.67369650552155e-05,
+      "loss": 0.2353,
+      "step": 558
+    },
+    {
+      "epoch": 2.263157894736842,
+      "grad_norm": 0.3684328726682199,
+      "learning_rate": 2.6683688962766407e-05,
+      "loss": 0.2645,
+      "step": 559
+    },
+    {
+      "epoch": 2.2672064777327936,
+      "grad_norm": 0.3141850948071082,
+      "learning_rate": 2.6630359427718468e-05,
+      "loss": 0.2611,
+      "step": 560
+    },
+    {
+      "epoch": 2.2712550607287447,
+      "grad_norm": 0.31066229836919096,
+      "learning_rate": 2.6576976876493225e-05,
+      "loss": 0.2187,
+      "step": 561
+    },
+    {
+      "epoch": 2.2753036437246963,
+      "grad_norm": 0.3419772812213054,
+      "learning_rate": 2.652354173593614e-05,
+      "loss": 0.2505,
+      "step": 562
+    },
+    {
+      "epoch": 2.279352226720648,
+      "grad_norm": 0.2941899718051732,
+      "learning_rate": 2.647005443331317e-05,
+      "loss": 0.2397,
+      "step": 563
+    },
+    {
+      "epoch": 2.283400809716599,
+      "grad_norm": 0.36007283623142367,
+      "learning_rate": 2.6416515396307354e-05,
+      "loss": 0.2658,
+      "step": 564
+    },
+    {
+      "epoch": 2.2874493927125505,
+      "grad_norm": 0.30851698003993533,
+      "learning_rate": 2.636292505301542e-05,
+      "loss": 0.2477,
+      "step": 565
+    },
+    {
+      "epoch": 2.291497975708502,
+      "grad_norm": 0.35951204008514837,
+      "learning_rate": 2.630928383194431e-05,
+      "loss": 0.2282,
+      "step": 566
+    },
+    {
+      "epoch": 2.2955465587044532,
+      "grad_norm": 0.29885333135009107,
+      "learning_rate": 2.625559216200781e-05,
+      "loss": 0.233,
+      "step": 567
+    },
+    {
+      "epoch": 2.299595141700405,
+      "grad_norm": 0.3371076524286233,
+      "learning_rate": 2.620185047252308e-05,
+      "loss": 0.2198,
+      "step": 568
+    },
+    {
+      "epoch": 2.3036437246963564,
+      "grad_norm": 0.3470722966696769,
+      "learning_rate": 2.614805919320723e-05,
+      "loss": 0.24,
+      "step": 569
+    },
+    {
+      "epoch": 2.3076923076923075,
+      "grad_norm": 0.31669640822704187,
+      "learning_rate": 2.6094218754173895e-05,
+      "loss": 0.2325,
+      "step": 570
+    },
+    {
+      "epoch": 2.311740890688259,
+      "grad_norm": 0.3308651763150751,
+      "learning_rate": 2.60403295859298e-05,
+      "loss": 0.2516,
+      "step": 571
+    },
+    {
+      "epoch": 2.3157894736842106,
+      "grad_norm": 0.33503359863694354,
+      "learning_rate": 2.5986392119371297e-05,
+      "loss": 0.2533,
+      "step": 572
+    },
+    {
+      "epoch": 2.3198380566801617,
+      "grad_norm": 0.31824984583681154,
+      "learning_rate": 2.5932406785780928e-05,
+      "loss": 0.2467,
+      "step": 573
+    },
+    {
+      "epoch": 2.3238866396761133,
+      "grad_norm": 0.33353116912510605,
+      "learning_rate": 2.5878374016823987e-05,
+      "loss": 0.2369,
+      "step": 574
+    },
+    {
+      "epoch": 2.327935222672065,
+      "grad_norm": 0.303035512016234,
+      "learning_rate": 2.5824294244545045e-05,
+      "loss": 0.2507,
+      "step": 575
+    },
+    {
+      "epoch": 2.331983805668016,
+      "grad_norm": 0.34990950608748417,
+      "learning_rate": 2.577016790136453e-05,
+      "loss": 0.2451,
+      "step": 576
+    },
+    {
+      "epoch": 2.3360323886639676,
+      "grad_norm": 0.3187899712353704,
+      "learning_rate": 2.5715995420075225e-05,
+      "loss": 0.2557,
+      "step": 577
+    },
+    {
+      "epoch": 2.340080971659919,
+      "grad_norm": 0.3889010911367558,
+      "learning_rate": 2.5661777233838864e-05,
+      "loss": 0.2623,
+      "step": 578
+    },
+    {
+      "epoch": 2.3441295546558703,
+      "grad_norm": 0.3035172647458326,
+      "learning_rate": 2.5607513776182616e-05,
+      "loss": 0.2695,
+      "step": 579
+    },
+    {
+      "epoch": 2.348178137651822,
+      "grad_norm": 0.33061911229131474,
+      "learning_rate": 2.5553205480995633e-05,
+      "loss": 0.248,
+      "step": 580
+    },
+    {
+      "epoch": 2.3522267206477734,
+      "grad_norm": 0.3372258413597226,
+      "learning_rate": 2.5498852782525604e-05,
+      "loss": 0.2603,
+      "step": 581
+    },
+    {
+      "epoch": 2.3562753036437245,
+      "grad_norm": 0.30918003999781146,
+      "learning_rate": 2.5444456115375252e-05,
+      "loss": 0.2361,
+      "step": 582
+    },
+    {
+      "epoch": 2.360323886639676,
+      "grad_norm": 0.3012471799249609,
+      "learning_rate": 2.539001591449889e-05,
+      "loss": 0.2476,
+      "step": 583
+    },
+    {
+      "epoch": 2.3643724696356276,
+      "grad_norm": 0.2969032172211022,
+      "learning_rate": 2.53355326151989e-05,
+      "loss": 0.2539,
+      "step": 584
+    },
+    {
+      "epoch": 2.3684210526315788,
+      "grad_norm": 0.33182481102199723,
+      "learning_rate": 2.5281006653122288e-05,
+      "loss": 0.23,
+      "step": 585
+    },
+    {
+      "epoch": 2.3724696356275303,
+      "grad_norm": 0.3157440127005082,
+      "learning_rate": 2.5226438464257217e-05,
+      "loss": 0.2486,
+      "step": 586
+    },
+    {
+      "epoch": 2.376518218623482,
+      "grad_norm": 0.3067389422894455,
+      "learning_rate": 2.5171828484929447e-05,
+      "loss": 0.2624,
+      "step": 587
+    },
+    {
+      "epoch": 2.380566801619433,
+      "grad_norm": 0.35008576449072465,
+      "learning_rate": 2.5117177151798934e-05,
+      "loss": 0.2527,
+      "step": 588
+    },
+    {
+      "epoch": 2.3846153846153846,
+      "grad_norm": 0.3415130537658597,
+      "learning_rate": 2.506248490185627e-05,
+      "loss": 0.2721,
+      "step": 589
+    },
+    {
+      "epoch": 2.388663967611336,
+      "grad_norm": 0.334679204756773,
+      "learning_rate": 2.500775217241923e-05,
+      "loss": 0.2389,
+      "step": 590
+    },
+    {
+      "epoch": 2.3927125506072873,
+      "grad_norm": 0.3152579599900844,
+      "learning_rate": 2.4952979401129266e-05,
+      "loss": 0.2284,
+      "step": 591
+    },
+    {
+      "epoch": 2.396761133603239,
+      "grad_norm": 0.3158862297390935,
+      "learning_rate": 2.4898167025948e-05,
+      "loss": 0.2448,
+      "step": 592
+    },
+    {
+      "epoch": 2.4008097165991904,
+      "grad_norm": 0.32815175630113214,
+      "learning_rate": 2.4843315485153703e-05,
+      "loss": 0.2348,
+      "step": 593
+    },
+    {
+      "epoch": 2.4048582995951415,
+      "grad_norm": 0.33194083540882824,
+      "learning_rate": 2.4788425217337856e-05,
+      "loss": 0.2421,
+      "step": 594
+    },
+    {
+      "epoch": 2.408906882591093,
+      "grad_norm": 0.32554990084936336,
+      "learning_rate": 2.4733496661401572e-05,
+      "loss": 0.2488,
+      "step": 595
+    },
+    {
+      "epoch": 2.4129554655870447,
+      "grad_norm": 0.3267524297342017,
+      "learning_rate": 2.4678530256552114e-05,
+      "loss": 0.2478,
+      "step": 596
+    },
+    {
+      "epoch": 2.417004048582996,
+      "grad_norm": 0.33036163337840885,
+      "learning_rate": 2.46235264422994e-05,
+      "loss": 0.236,
+      "step": 597
+    },
+    {
+      "epoch": 2.4210526315789473,
+      "grad_norm": 0.3065684272654906,
+      "learning_rate": 2.4568485658452452e-05,
+      "loss": 0.2391,
+      "step": 598
+    },
+    {
+      "epoch": 2.425101214574899,
+      "grad_norm": 0.3541741867580982,
+      "learning_rate": 2.451340834511592e-05,
+      "loss": 0.2382,
+      "step": 599
+    },
+    {
+      "epoch": 2.42914979757085,
+      "grad_norm": 0.3109994511217404,
+      "learning_rate": 2.4458294942686528e-05,
+      "loss": 0.251,
+      "step": 600
+    },
+    {
+      "epoch": 2.4331983805668016,
+      "grad_norm": 0.3103307370062228,
+      "learning_rate": 2.440314589184958e-05,
+      "loss": 0.2337,
+      "step": 601
+    },
+    {
+      "epoch": 2.437246963562753,
+      "grad_norm": 0.33174182406243713,
+      "learning_rate": 2.434796163357541e-05,
+      "loss": 0.2366,
+      "step": 602
+    },
+    {
+      "epoch": 2.4412955465587043,
+      "grad_norm": 0.3155907073863106,
+      "learning_rate": 2.429274260911588e-05,
+      "loss": 0.2107,
+      "step": 603
+    },
+    {
+      "epoch": 2.445344129554656,
+      "grad_norm": 0.3200373950668486,
+      "learning_rate": 2.423748926000084e-05,
+      "loss": 0.2338,
+      "step": 604
+    },
+    {
+      "epoch": 2.4493927125506074,
+      "grad_norm": 0.3321344456442205,
+      "learning_rate": 2.418220202803459e-05,
+      "loss": 0.249,
+      "step": 605
+    },
+    {
+      "epoch": 2.4534412955465585,
+      "grad_norm": 0.30214102077881544,
+      "learning_rate": 2.412688135529237e-05,
+      "loss": 0.2433,
+      "step": 606
+    },
+    {
+      "epoch": 2.45748987854251,
+      "grad_norm": 0.32817198799217984,
+      "learning_rate": 2.4071527684116807e-05,
+      "loss": 0.2465,
+      "step": 607
+    },
+    {
+      "epoch": 2.4615384615384617,
+      "grad_norm": 0.3173796251578628,
+      "learning_rate": 2.4016141457114372e-05,
+      "loss": 0.2442,
+      "step": 608
+    },
+    {
+      "epoch": 2.465587044534413,
+      "grad_norm": 0.3049814018055747,
+      "learning_rate": 2.396072311715187e-05,
+      "loss": 0.2497,
+      "step": 609
+    },
+    {
+      "epoch": 2.4696356275303644,
+      "grad_norm": 0.35925299487125006,
+      "learning_rate": 2.3905273107352863e-05,
+      "loss": 0.271,
+      "step": 610
+    },
+    {
+      "epoch": 2.473684210526316,
+      "grad_norm": 0.3117346532780196,
+      "learning_rate": 2.3849791871094157e-05,
+      "loss": 0.2538,
+      "step": 611
+    },
+    {
+      "epoch": 2.477732793522267,
+      "grad_norm": 0.29419136351101804,
+      "learning_rate": 2.379427985200223e-05,
+      "loss": 0.2414,
+      "step": 612
+    },
+    {
+      "epoch": 2.4817813765182186,
+      "grad_norm": 0.35743508551440817,
+      "learning_rate": 2.373873749394972e-05,
+      "loss": 0.2633,
+      "step": 613
+    },
+    {
+      "epoch": 2.48582995951417,
+      "grad_norm": 0.27432309560970863,
+      "learning_rate": 2.3683165241051836e-05,
+      "loss": 0.2841,
+      "step": 614
+    },
+    {
+      "epoch": 2.4898785425101213,
+      "grad_norm": 0.3105468064529887,
+      "learning_rate": 2.3627563537662823e-05,
+      "loss": 0.2451,
+      "step": 615
+    },
+    {
+      "epoch": 2.493927125506073,
+      "grad_norm": 0.28830446627312906,
+      "learning_rate": 2.3571932828372433e-05,
+      "loss": 0.2465,
+      "step": 616
+    },
+    {
+      "epoch": 2.4979757085020244,
+      "grad_norm": 0.2906113590088792,
+      "learning_rate": 2.351627355800233e-05,
+      "loss": 0.245,
+      "step": 617
+    },
+    {
+      "epoch": 2.5020242914979756,
+      "grad_norm": 0.2979056543999439,
+      "learning_rate": 2.346058617160256e-05,
+      "loss": 0.2408,
+      "step": 618
+    },
+    {
+      "epoch": 2.506072874493927,
+      "grad_norm": 0.3010636013292708,
+      "learning_rate": 2.3404871114447976e-05,
+      "loss": 0.2363,
+      "step": 619
+    },
+    {
+      "epoch": 2.5101214574898787,
+      "grad_norm": 0.30103589810567,
+      "learning_rate": 2.334912883203469e-05,
+      "loss": 0.2445,
+      "step": 620
+    },
+    {
+      "epoch": 2.51417004048583,
+      "grad_norm": 0.3115877783147523,
+      "learning_rate": 2.3293359770076505e-05,
+      "loss": 0.2575,
+      "step": 621
+    },
+    {
+      "epoch": 2.5182186234817814,
+      "grad_norm": 0.3025998329816071,
+      "learning_rate": 2.323756437450136e-05,
+      "loss": 0.2382,
+      "step": 622
+    },
+    {
+      "epoch": 2.522267206477733,
+      "grad_norm": 0.3272657507757977,
+      "learning_rate": 2.3181743091447754e-05,
+      "loss": 0.2391,
+      "step": 623
+    },
+    {
+      "epoch": 2.526315789473684,
+      "grad_norm": 0.3074949013908285,
+      "learning_rate": 2.3125896367261164e-05,
+      "loss": 0.2352,
+      "step": 624
+    },
+    {
+      "epoch": 2.5303643724696356,
+      "grad_norm": 0.29806295692146156,
+      "learning_rate": 2.307002464849052e-05,
+      "loss": 0.245,
+      "step": 625
+    },
+    {
+      "epoch": 2.534412955465587,
+      "grad_norm": 0.31775939563894634,
+      "learning_rate": 2.30141283818846e-05,
+      "loss": 0.2477,
+      "step": 626
+    },
+    {
+      "epoch": 2.5384615384615383,
+      "grad_norm": 0.28657791363256163,
+      "learning_rate": 2.2958208014388462e-05,
+      "loss": 0.2312,
+      "step": 627
+    },
+    {
+      "epoch": 2.54251012145749,
+      "grad_norm": 0.28859285836161014,
+      "learning_rate": 2.2902263993139857e-05,
+      "loss": 0.2349,
+      "step": 628
+    },
+    {
+      "epoch": 2.5465587044534415,
+      "grad_norm": 0.30745608682791153,
+      "learning_rate": 2.2846296765465708e-05,
+      "loss": 0.2518,
+      "step": 629
+    },
+    {
+      "epoch": 2.5506072874493926,
+      "grad_norm": 0.315243974074004,
+      "learning_rate": 2.2790306778878465e-05,
+      "loss": 0.2304,
+      "step": 630
+    },
+    {
+      "epoch": 2.554655870445344,
+      "grad_norm": 0.28627960006075165,
+      "learning_rate": 2.273429448107257e-05,
+      "loss": 0.233,
+      "step": 631
+    },
+    {
+      "epoch": 2.5587044534412957,
+      "grad_norm": 0.3098260111748755,
+      "learning_rate": 2.2678260319920868e-05,
+      "loss": 0.2573,
+      "step": 632
+    },
+    {
+      "epoch": 2.562753036437247,
+      "grad_norm": 0.312894702696894,
+      "learning_rate": 2.262220474347102e-05,
+      "loss": 0.2518,
+      "step": 633
+    },
+    {
+      "epoch": 2.5668016194331984,
+      "grad_norm": 0.30717850436933175,
+      "learning_rate": 2.2566128199941918e-05,
+      "loss": 0.263,
+      "step": 634
+    },
+    {
+      "epoch": 2.57085020242915,
+      "grad_norm": 0.3305992535114179,
+      "learning_rate": 2.2510031137720104e-05,
+      "loss": 0.2427,
+      "step": 635
+    },
+    {
+      "epoch": 2.574898785425101,
+      "grad_norm": 0.29834646710070006,
+      "learning_rate": 2.2453914005356206e-05,
+      "loss": 0.2333,
+      "step": 636
+    },
+    {
+      "epoch": 2.5789473684210527,
+      "grad_norm": 0.3151347247113424,
+      "learning_rate": 2.2397777251561302e-05,
+      "loss": 0.2557,
+      "step": 637
+    },
+    {
+      "epoch": 2.582995951417004,
+      "grad_norm": 0.3108257787772622,
+      "learning_rate": 2.2341621325203393e-05,
+      "loss": 0.2295,
+      "step": 638
+    },
+    {
+      "epoch": 2.5870445344129553,
+      "grad_norm": 0.2836314463815351,
+      "learning_rate": 2.2285446675303764e-05,
+      "loss": 0.2326,
+      "step": 639
+    },
+    {
+      "epoch": 2.591093117408907,
+      "grad_norm": 0.32368193647676935,
+      "learning_rate": 2.2229253751033423e-05,
+      "loss": 0.2487,
+      "step": 640
+    },
+    {
+      "epoch": 2.5951417004048585,
+      "grad_norm": 0.3249030778378911,
+      "learning_rate": 2.2173043001709492e-05,
+      "loss": 0.259,
+      "step": 641
+    },
+    {
+      "epoch": 2.5991902834008096,
+      "grad_norm": 0.3242209618774341,
+      "learning_rate": 2.2116814876791627e-05,
+      "loss": 0.2675,
+      "step": 642
+    },
+    {
+      "epoch": 2.603238866396761,
+      "grad_norm": 0.3290481195553767,
+      "learning_rate": 2.2060569825878413e-05,
+      "loss": 0.2445,
+      "step": 643
+    },
+    {
+      "epoch": 2.6072874493927127,
+      "grad_norm": 0.2984331597910281,
+      "learning_rate": 2.2004308298703777e-05,
+      "loss": 0.2375,
+      "step": 644
+    },
+    {
+      "epoch": 2.611336032388664,
+      "grad_norm": 0.3223696809583322,
+      "learning_rate": 2.1948030745133396e-05,
+      "loss": 0.2521,
+      "step": 645
+    },
+    {
+      "epoch": 2.6153846153846154,
+      "grad_norm": 0.29739342886050146,
+      "learning_rate": 2.1891737615161083e-05,
+      "loss": 0.238,
+      "step": 646
+    },
+    {
+      "epoch": 2.619433198380567,
+      "grad_norm": 0.34920243207267776,
+      "learning_rate": 2.1835429358905218e-05,
+      "loss": 0.2459,
+      "step": 647
+    },
+    {
+      "epoch": 2.623481781376518,
+      "grad_norm": 0.2969560642435508,
+      "learning_rate": 2.17791064266051e-05,
+      "loss": 0.2432,
+      "step": 648
+    },
+    {
+      "epoch": 2.6275303643724697,
+      "grad_norm": 0.3294588517064404,
+      "learning_rate": 2.172276926861741e-05,
+      "loss": 0.2446,
+      "step": 649
+    },
+    {
+      "epoch": 2.6315789473684212,
+      "grad_norm": 0.3224992208187403,
+      "learning_rate": 2.166641833541255e-05,
+      "loss": 0.2355,
+      "step": 650
+    },
+    {
+      "epoch": 2.6356275303643724,
+      "grad_norm": 0.32860443098409153,
+      "learning_rate": 2.161005407757109e-05,
+      "loss": 0.2546,
+      "step": 651
+    },
+    {
+      "epoch": 2.639676113360324,
+      "grad_norm": 0.3357953806249753,
+      "learning_rate": 2.155367694578013e-05,
+      "loss": 0.2438,
+      "step": 652
+    },
+    {
+      "epoch": 2.6437246963562755,
+      "grad_norm": 0.3250629892491294,
+      "learning_rate": 2.1497287390829712e-05,
+      "loss": 0.2539,
+      "step": 653
+    },
+    {
+      "epoch": 2.6477732793522266,
+      "grad_norm": 0.3412642663979935,
+      "learning_rate": 2.1440885863609208e-05,
+      "loss": 0.2424,
+      "step": 654
+    },
+    {
+      "epoch": 2.651821862348178,
+      "grad_norm": 0.2909592311412711,
+      "learning_rate": 2.1384472815103737e-05,
+      "loss": 0.24,
+      "step": 655
+    },
+    {
+      "epoch": 2.6558704453441297,
+      "grad_norm": 0.3127314938713807,
+      "learning_rate": 2.1328048696390534e-05,
+      "loss": 0.2458,
+      "step": 656
+    },
+    {
+      "epoch": 2.659919028340081,
+      "grad_norm": 0.31171395498581694,
+      "learning_rate": 2.1271613958635342e-05,
+      "loss": 0.2442,
+      "step": 657
+    },
+    {
+      "epoch": 2.6639676113360324,
+      "grad_norm": 0.30434682346335795,
+      "learning_rate": 2.1215169053088825e-05,
+      "loss": 0.2423,
+      "step": 658
+    },
+    {
+      "epoch": 2.668016194331984,
+      "grad_norm": 0.33392493411790336,
+      "learning_rate": 2.1158714431082937e-05,
+      "loss": 0.2349,
+      "step": 659
+    },
+    {
+      "epoch": 2.672064777327935,
+      "grad_norm": 0.2830656628500496,
+      "learning_rate": 2.110225054402733e-05,
+      "loss": 0.2559,
+      "step": 660
+    },
+    {
+      "epoch": 2.6761133603238867,
+      "grad_norm": 0.3140906532591819,
+      "learning_rate": 2.1045777843405747e-05,
+      "loss": 0.2432,
+      "step": 661
+    },
+    {
+      "epoch": 2.6801619433198383,
+      "grad_norm": 0.28276812509879656,
+      "learning_rate": 2.09892967807724e-05,
+      "loss": 0.2535,
+      "step": 662
+    },
+    {
+      "epoch": 2.6842105263157894,
+      "grad_norm": 0.32132653319604615,
+      "learning_rate": 2.093280780774835e-05,
+      "loss": 0.2409,
+      "step": 663
+    },
+    {
+      "epoch": 2.688259109311741,
+      "grad_norm": 0.284520285685773,
+      "learning_rate": 2.087631137601793e-05,
+      "loss": 0.2452,
+      "step": 664
+    },
+    {
+      "epoch": 2.6923076923076925,
+      "grad_norm": 0.28846306692154616,
+      "learning_rate": 2.0819807937325104e-05,
+      "loss": 0.2394,
+      "step": 665
+    },
+    {
+      "epoch": 2.6963562753036436,
+      "grad_norm": 0.28971206647107023,
+      "learning_rate": 2.0763297943469847e-05,
+      "loss": 0.2386,
+      "step": 666
+    },
+    {
+      "epoch": 2.700404858299595,
+      "grad_norm": 0.28959889721879484,
+      "learning_rate": 2.070678184630458e-05,
+      "loss": 0.2423,
+      "step": 667
+    },
+    {
+      "epoch": 2.7044534412955468,
+      "grad_norm": 0.29467233319654607,
+      "learning_rate": 2.06502600977305e-05,
+      "loss": 0.2555,
+      "step": 668
+    },
+    {
+      "epoch": 2.708502024291498,
+      "grad_norm": 0.3315536499555472,
+      "learning_rate": 2.0593733149694008e-05,
+      "loss": 0.2456,
+      "step": 669
+    },
+    {
+      "epoch": 2.7125506072874495,
+      "grad_norm": 0.3198318982614178,
+      "learning_rate": 2.0537201454183074e-05,
+      "loss": 0.2425,
+      "step": 670
+    },
+    {
+      "epoch": 2.716599190283401,
+      "grad_norm": 0.31603710656544165,
+      "learning_rate": 2.0480665463223626e-05,
+      "loss": 0.2593,
+      "step": 671
+    },
+    {
+      "epoch": 2.720647773279352,
+      "grad_norm": 0.3231420161363104,
+      "learning_rate": 2.042412562887594e-05,
+      "loss": 0.2572,
+      "step": 672
+    },
+    {
+      "epoch": 2.7246963562753037,
+      "grad_norm": 0.2830022045135199,
+      "learning_rate": 2.0367582403231033e-05,
+      "loss": 0.2563,
+      "step": 673
+    },
+    {
+      "epoch": 2.7287449392712553,
+      "grad_norm": 0.324889670764517,
+      "learning_rate": 2.0311036238407023e-05,
+      "loss": 0.2537,
+      "step": 674
+    },
+    {
+      "epoch": 2.7327935222672064,
+      "grad_norm": 0.3031830987789163,
+      "learning_rate": 2.0254487586545534e-05,
+      "loss": 0.2358,
+      "step": 675
+    },
+    {
+      "epoch": 2.736842105263158,
+      "grad_norm": 0.3156424618604592,
+      "learning_rate": 2.0197936899808088e-05,
+      "loss": 0.2509,
+      "step": 676
+    },
+    {
+      "epoch": 2.7408906882591095,
+      "grad_norm": 0.30636944589717013,
+      "learning_rate": 2.0141384630372466e-05,
+      "loss": 0.2496,
+      "step": 677
+    },
+    {
+      "epoch": 2.7449392712550607,
+      "grad_norm": 0.29198359795472417,
+      "learning_rate": 2.0084831230429106e-05,
+      "loss": 0.2423,
+      "step": 678
+    },
+    {
+      "epoch": 2.748987854251012,
+      "grad_norm": 0.34351361181237366,
+      "learning_rate": 2.0028277152177492e-05,
+      "loss": 0.2549,
+      "step": 679
+    },
+    {
+      "epoch": 2.753036437246964,
+      "grad_norm": 0.3009232657175083,
+      "learning_rate": 1.9971722847822518e-05,
+      "loss": 0.2378,
+      "step": 680
+    },
+    {
+      "epoch": 2.757085020242915,
+      "grad_norm": 0.33721311519047215,
+      "learning_rate": 1.99151687695709e-05,
+      "loss": 0.2615,
+      "step": 681
+    },
+    {
+      "epoch": 2.7611336032388665,
+      "grad_norm": 0.2818266683078424,
+      "learning_rate": 1.9858615369627537e-05,
+      "loss": 0.2538,
+      "step": 682
+    },
+    {
+      "epoch": 2.765182186234818,
+      "grad_norm": 0.2966632015269009,
+      "learning_rate": 1.9802063100191916e-05,
+      "loss": 0.2584,
+      "step": 683
+    },
+    {
+      "epoch": 2.769230769230769,
+      "grad_norm": 0.29148071468943165,
+      "learning_rate": 1.974551241345447e-05,
+      "loss": 0.2516,
+      "step": 684
+    },
+    {
+      "epoch": 2.7732793522267207,
+      "grad_norm": 0.30012301767003047,
+      "learning_rate": 1.9688963761592984e-05,
+      "loss": 0.2414,
+      "step": 685
+    },
+    {
+      "epoch": 2.7773279352226723,
+      "grad_norm": 0.2965207156869325,
+      "learning_rate": 1.963241759676897e-05,
+      "loss": 0.2439,
+      "step": 686
+    },
+    {
+      "epoch": 2.7813765182186234,
+      "grad_norm": 0.28923041760337465,
+      "learning_rate": 1.9575874371124066e-05,
+      "loss": 0.2333,
+      "step": 687
+    },
+    {
+      "epoch": 2.785425101214575,
+      "grad_norm": 0.3081010939188119,
+      "learning_rate": 1.951933453677638e-05,
+      "loss": 0.2557,
+      "step": 688
+    },
+    {
+      "epoch": 2.7894736842105265,
+      "grad_norm": 0.2972032894823422,
+      "learning_rate": 1.9462798545816932e-05,
+      "loss": 0.2501,
+      "step": 689
+    },
+    {
+      "epoch": 2.7935222672064777,
+      "grad_norm": 0.2976987832792086,
+      "learning_rate": 1.9406266850305996e-05,
+      "loss": 0.2495,
+      "step": 690
+    },
+    {
+      "epoch": 2.7975708502024292,
+      "grad_norm": 0.2861740450974801,
+      "learning_rate": 1.9349739902269503e-05,
+      "loss": 0.2649,
+      "step": 691
+    },
+    {
+      "epoch": 2.801619433198381,
+      "grad_norm": 0.28727577512005575,
+      "learning_rate": 1.9293218153695425e-05,
+      "loss": 0.2351,
+      "step": 692
+    },
+    {
+      "epoch": 2.805668016194332,
+      "grad_norm": 0.31416779882313717,
+      "learning_rate": 1.9236702056530153e-05,
+      "loss": 0.2509,
+      "step": 693
+    },
+    {
+      "epoch": 2.8097165991902835,
+      "grad_norm": 0.2951243649277555,
+      "learning_rate": 1.9180192062674907e-05,
+      "loss": 0.239,
+      "step": 694
+    },
+    {
+      "epoch": 2.813765182186235,
+      "grad_norm": 0.29642720657934596,
+      "learning_rate": 1.9123688623982076e-05,
+      "loss": 0.2427,
+      "step": 695
+    },
+    {
+      "epoch": 2.817813765182186,
+      "grad_norm": 0.29756607305133304,
+      "learning_rate": 1.906719219225165e-05,
+      "loss": 0.2405,
+      "step": 696
+    },
+    {
+      "epoch": 2.8218623481781377,
+      "grad_norm": 0.2854501121220625,
+      "learning_rate": 1.9010703219227603e-05,
+      "loss": 0.2336,
+      "step": 697
+    },
+    {
+      "epoch": 2.8259109311740893,
+      "grad_norm": 0.3420021905988869,
+      "learning_rate": 1.8954222156594256e-05,
+      "loss": 0.2487,
+      "step": 698
+    },
+    {
+      "epoch": 2.8299595141700404,
+      "grad_norm": 0.31042146525017716,
+      "learning_rate": 1.8897749455972673e-05,
+      "loss": 0.2453,
+      "step": 699
+    },
+    {
+      "epoch": 2.834008097165992,
+      "grad_norm": 0.3229885430060254,
+      "learning_rate": 1.884128556891707e-05,
+      "loss": 0.2435,
+      "step": 700
+    },
+    {
+      "epoch": 2.8380566801619436,
+      "grad_norm": 0.31378598886250486,
+      "learning_rate": 1.878483094691119e-05,
+      "loss": 0.2426,
+      "step": 701
+    },
+    {
+      "epoch": 2.8421052631578947,
+      "grad_norm": 0.34033531384495064,
+      "learning_rate": 1.8728386041364664e-05,
+      "loss": 0.241,
+      "step": 702
+    },
+    {
+      "epoch": 2.8461538461538463,
+      "grad_norm": 0.2967319970529675,
+      "learning_rate": 1.867195130360947e-05,
+      "loss": 0.2458,
+      "step": 703
+    },
+    {
+      "epoch": 2.850202429149798,
+      "grad_norm": 0.34240898048922774,
+      "learning_rate": 1.8615527184896263e-05,
+      "loss": 0.2557,
+      "step": 704
+    },
+    {
+      "epoch": 2.854251012145749,
+      "grad_norm": 0.3012526072601195,
+      "learning_rate": 1.8559114136390795e-05,
+      "loss": 0.2193,
+      "step": 705
+    },
+    {
+      "epoch": 2.8582995951417005,
+      "grad_norm": 0.2933172534809712,
+      "learning_rate": 1.8502712609170298e-05,
+      "loss": 0.2387,
+      "step": 706
+    },
+    {
+      "epoch": 2.862348178137652,
+      "grad_norm": 0.336090774994267,
+      "learning_rate": 1.8446323054219876e-05,
+      "loss": 0.2544,
+      "step": 707
+    },
+    {
+      "epoch": 2.866396761133603,
+      "grad_norm": 0.30720849324739213,
+      "learning_rate": 1.838994592242891e-05,
+      "loss": 0.2392,
+      "step": 708
+    },
+    {
+      "epoch": 2.8704453441295548,
+      "grad_norm": 0.3060125487450167,
+      "learning_rate": 1.8333581664587453e-05,
+      "loss": 0.2454,
+      "step": 709
+    },
+    {
+      "epoch": 2.8744939271255063,
+      "grad_norm": 0.32895391325832535,
+      "learning_rate": 1.8277230731382593e-05,
+      "loss": 0.2346,
+      "step": 710
+    },
+    {
+      "epoch": 2.8785425101214575,
+      "grad_norm": 0.2884526651169379,
+      "learning_rate": 1.82208935733949e-05,
+      "loss": 0.2346,
+      "step": 711
+    },
+    {
+      "epoch": 2.882591093117409,
+      "grad_norm": 0.35161023521941526,
+      "learning_rate": 1.8164570641094793e-05,
+      "loss": 0.2505,
+      "step": 712
+    },
+    {
+      "epoch": 2.8866396761133606,
+      "grad_norm": 0.2972409328467491,
+      "learning_rate": 1.810826238483892e-05,
+      "loss": 0.2244,
+      "step": 713
+    },
+    {
+      "epoch": 2.8906882591093117,
+      "grad_norm": 0.3025334165417715,
+      "learning_rate": 1.8051969254866608e-05,
+      "loss": 0.2652,
+      "step": 714
+    },
+    {
+      "epoch": 2.8947368421052633,
+      "grad_norm": 0.3063983840083053,
+      "learning_rate": 1.7995691701296226e-05,
+      "loss": 0.2347,
+      "step": 715
+    },
+    {
+      "epoch": 2.898785425101215,
+      "grad_norm": 0.3274958404235151,
+      "learning_rate": 1.7939430174121593e-05,
+      "loss": 0.2488,
+      "step": 716
+    },
+    {
+      "epoch": 2.902834008097166,
+      "grad_norm": 0.3201473653274444,
+      "learning_rate": 1.788318512320838e-05,
+      "loss": 0.255,
+      "step": 717
+    },
+    {
+      "epoch": 2.9068825910931175,
+      "grad_norm": 0.3338012047723535,
+      "learning_rate": 1.782695699829051e-05,
+      "loss": 0.2398,
+      "step": 718
+    },
+    {
+      "epoch": 2.910931174089069,
+      "grad_norm": 0.31824451378182333,
+      "learning_rate": 1.7770746248966587e-05,
+      "loss": 0.2618,
+      "step": 719
+    },
+    {
+      "epoch": 2.91497975708502,
+      "grad_norm": 0.3194752409510773,
+      "learning_rate": 1.7714553324696243e-05,
+      "loss": 0.2194,
+      "step": 720
+    },
+    {
+      "epoch": 2.919028340080972,
+      "grad_norm": 0.3397832228526603,
+      "learning_rate": 1.7658378674796614e-05,
+      "loss": 0.2382,
+      "step": 721
+    },
+    {
+      "epoch": 2.9230769230769234,
+      "grad_norm": 0.3337663062992993,
+      "learning_rate": 1.76022227484387e-05,
+      "loss": 0.2472,
+      "step": 722
+    },
+    {
+      "epoch": 2.9271255060728745,
+      "grad_norm": 0.3046521012195735,
+      "learning_rate": 1.7546085994643807e-05,
+      "loss": 0.2554,
+      "step": 723
+    },
+    {
+      "epoch": 2.931174089068826,
+      "grad_norm": 0.31998720647848433,
+      "learning_rate": 1.7489968862279902e-05,
+      "loss": 0.2508,
+      "step": 724
+    },
+    {
+      "epoch": 2.9352226720647776,
+      "grad_norm": 0.3442874595590751,
+      "learning_rate": 1.743387180005809e-05,
+      "loss": 0.2483,
+      "step": 725
+    },
+    {
+      "epoch": 2.9392712550607287,
+      "grad_norm": 0.3220838163097024,
+      "learning_rate": 1.737779525652899e-05,
+      "loss": 0.2413,
+      "step": 726
+    },
+    {
+      "epoch": 2.9433198380566803,
+      "grad_norm": 0.3099538144405799,
+      "learning_rate": 1.732173968007914e-05,
+      "loss": 0.2456,
+      "step": 727
+    },
+    {
+      "epoch": 2.9473684210526314,
+      "grad_norm": 0.31375036028898534,
+      "learning_rate": 1.7265705518927437e-05,
+      "loss": 0.2307,
+      "step": 728
+    },
+    {
+      "epoch": 2.951417004048583,
+      "grad_norm": 0.29948474262468877,
+      "learning_rate": 1.7209693221121542e-05,
+      "loss": 0.24,
+      "step": 729
+    },
+    {
+      "epoch": 2.9554655870445345,
+      "grad_norm": 0.2731308766279527,
+      "learning_rate": 1.7153703234534302e-05,
+      "loss": 0.2519,
+      "step": 730
+    },
+    {
+      "epoch": 2.9595141700404857,
+      "grad_norm": 0.32404339912480606,
+      "learning_rate": 1.709773600686015e-05,
+      "loss": 0.2528,
+      "step": 731
+    },
+    {
+      "epoch": 2.9635627530364372,
+      "grad_norm": 0.3007688469360073,
+      "learning_rate": 1.7041791985611548e-05,
+      "loss": 0.2425,
+      "step": 732
+    },
+    {
+      "epoch": 2.967611336032389,
+      "grad_norm": 0.27020685614536005,
+      "learning_rate": 1.69858716181154e-05,
+      "loss": 0.2365,
+      "step": 733
+    },
+    {
+      "epoch": 2.97165991902834,
+      "grad_norm": 0.319956875766256,
+      "learning_rate": 1.692997535150948e-05,
+      "loss": 0.2519,
+      "step": 734
+    },
+    {
+      "epoch": 2.9757085020242915,
+      "grad_norm": 0.3182727678214156,
+      "learning_rate": 1.687410363273884e-05,
+      "loss": 0.2474,
+      "step": 735
+    },
+    {
+      "epoch": 2.979757085020243,
+      "grad_norm": 0.26666418052869306,
+      "learning_rate": 1.6818256908552257e-05,
+      "loss": 0.2476,
+      "step": 736
+    },
+    {
+      "epoch": 2.983805668016194,
+      "grad_norm": 0.29210374183686116,
+      "learning_rate": 1.6762435625498646e-05,
+      "loss": 0.2484,
+      "step": 737
+    },
+    {
+      "epoch": 2.9878542510121457,
+      "grad_norm": 0.2834219105840551,
+      "learning_rate": 1.67066402299235e-05,
+      "loss": 0.2273,
+      "step": 738
+    },
+    {
+      "epoch": 2.9919028340080973,
+      "grad_norm": 0.27938713341284566,
+      "learning_rate": 1.6650871167965313e-05,
+      "loss": 0.2311,
+      "step": 739
+    },
+    {
+      "epoch": 2.9959514170040484,
+      "grad_norm": 0.2884697655908823,
+      "learning_rate": 1.6595128885552028e-05,
+      "loss": 0.2359,
+      "step": 740
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 0.2639811516597644,
+      "learning_rate": 1.653941382839745e-05,
+      "loss": 0.2403,
+      "step": 741
+    },
+    {
+      "epoch": 3.0040485829959516,
+      "grad_norm": 0.3886013815926394,
+      "learning_rate": 1.6483726441997673e-05,
+      "loss": 0.1871,
+      "step": 742
+    },
+    {
+      "epoch": 3.0080971659919027,
+      "grad_norm": 0.2971072656123463,
+      "learning_rate": 1.642806717162757e-05,
+      "loss": 0.1755,
+      "step": 743
+    },
+    {
+      "epoch": 3.0121457489878543,
+      "grad_norm": 0.5244628098615847,
+      "learning_rate": 1.637243646233718e-05,
+      "loss": 0.1888,
+      "step": 744
+    },
+    {
+      "epoch": 3.016194331983806,
+      "grad_norm": 0.3994573965638901,
+      "learning_rate": 1.6316834758948174e-05,
+      "loss": 0.1695,
+      "step": 745
+    },
+    {
+      "epoch": 3.020242914979757,
+      "grad_norm": 0.42788358623824835,
+      "learning_rate": 1.6261262506050282e-05,
+      "loss": 0.1754,
+      "step": 746
+    },
+    {
+      "epoch": 3.0242914979757085,
+      "grad_norm": 0.4252459993419149,
+      "learning_rate": 1.620572014799777e-05,
+      "loss": 0.1757,
+      "step": 747
+    },
+    {
+      "epoch": 3.02834008097166,
+      "grad_norm": 0.3258964030138234,
+      "learning_rate": 1.6150208128905857e-05,
+      "loss": 0.1984,
+      "step": 748
+    },
+    {
+      "epoch": 3.032388663967611,
+      "grad_norm": 0.36125882784996954,
+      "learning_rate": 1.6094726892647147e-05,
+      "loss": 0.1805,
+      "step": 749
+    },
+    {
+      "epoch": 3.0364372469635628,
+      "grad_norm": 0.3533157585232186,
+      "learning_rate": 1.6039276882848135e-05,
+      "loss": 0.1776,
+      "step": 750
+    },
+    {
+      "epoch": 3.0404858299595143,
+      "grad_norm": 0.3415595893981855,
+      "learning_rate": 1.5983858542885635e-05,
+      "loss": 0.1801,
+      "step": 751
+    },
+    {
+      "epoch": 3.0445344129554655,
+      "grad_norm": 0.3313682502203542,
+      "learning_rate": 1.5928472315883203e-05,
+      "loss": 0.1796,
+      "step": 752
+    },
+    {
+      "epoch": 3.048582995951417,
+      "grad_norm": 0.3056656734006266,
+      "learning_rate": 1.5873118644707633e-05,
+      "loss": 0.1707,
+      "step": 753
+    },
+    {
+      "epoch": 3.0526315789473686,
+      "grad_norm": 0.3246520167856373,
+      "learning_rate": 1.5817797971965413e-05,
+      "loss": 0.2053,
+      "step": 754
+    },
+    {
+      "epoch": 3.0566801619433197,
+      "grad_norm": 0.3284848941588058,
+      "learning_rate": 1.576251073999917e-05,
+      "loss": 0.1729,
+      "step": 755
+    },
+    {
+      "epoch": 3.0607287449392713,
+      "grad_norm": 0.29316577245669406,
+      "learning_rate": 1.5707257390884126e-05,
+      "loss": 0.168,
+      "step": 756
+    },
+    {
+      "epoch": 3.064777327935223,
+      "grad_norm": 0.34646554182294137,
+      "learning_rate": 1.5652038366424595e-05,
+      "loss": 0.178,
+      "step": 757
+    },
+    {
+      "epoch": 3.068825910931174,
+      "grad_norm": 0.32287848034429906,
+      "learning_rate": 1.5596854108150424e-05,
+      "loss": 0.1784,
+      "step": 758
+    },
+    {
+      "epoch": 3.0728744939271255,
+      "grad_norm": 0.30465302758066304,
+      "learning_rate": 1.5541705057313476e-05,
+      "loss": 0.1641,
+      "step": 759
+    },
+    {
+      "epoch": 3.076923076923077,
+      "grad_norm": 0.30645529002063454,
+      "learning_rate": 1.5486591654884086e-05,
+      "loss": 0.18,
+      "step": 760
+    },
+    {
+      "epoch": 3.080971659919028,
+      "grad_norm": 0.30144326483530937,
+      "learning_rate": 1.543151434154755e-05,
+      "loss": 0.1957,
+      "step": 761
+    },
+    {
+      "epoch": 3.08502024291498,
+      "grad_norm": 0.3012928039877883,
+      "learning_rate": 1.5376473557700615e-05,
+      "loss": 0.1611,
+      "step": 762
+    },
+    {
+      "epoch": 3.0890688259109313,
+      "grad_norm": 0.2827203338272862,
+      "learning_rate": 1.5321469743447893e-05,
+      "loss": 0.1822,
+      "step": 763
+    },
+    {
+      "epoch": 3.0931174089068825,
+      "grad_norm": 0.28736959682809277,
+      "learning_rate": 1.5266503338598434e-05,
+      "loss": 0.18,
+      "step": 764
+    },
+    {
+      "epoch": 3.097165991902834,
+      "grad_norm": 0.30590466534857874,
+      "learning_rate": 1.5211574782662149e-05,
+      "loss": 0.172,
+      "step": 765
+    },
+    {
+      "epoch": 3.1012145748987856,
+      "grad_norm": 0.29410914767672214,
+      "learning_rate": 1.5156684514846307e-05,
+      "loss": 0.1875,
+      "step": 766
+    },
+    {
+      "epoch": 3.1052631578947367,
+      "grad_norm": 0.28117112134821853,
+      "learning_rate": 1.5101832974052012e-05,
+      "loss": 0.1707,
+      "step": 767
+    },
+    {
+      "epoch": 3.1093117408906883,
+      "grad_norm": 0.28545822124179093,
+      "learning_rate": 1.5047020598870737e-05,
+      "loss": 0.1758,
+      "step": 768
+    },
+    {
+      "epoch": 3.11336032388664,
+      "grad_norm": 0.2764121189670712,
+      "learning_rate": 1.4992247827580778e-05,
+      "loss": 0.1816,
+      "step": 769
+    },
+    {
+      "epoch": 3.117408906882591,
+      "grad_norm": 0.2812085549208583,
+      "learning_rate": 1.493751509814374e-05,
+      "loss": 0.1779,
+      "step": 770
+    },
+    {
+      "epoch": 3.1214574898785425,
+      "grad_norm": 0.3058176460524668,
+      "learning_rate": 1.4882822848201073e-05,
+      "loss": 0.1738,
+      "step": 771
+    },
+    {
+      "epoch": 3.125506072874494,
+      "grad_norm": 0.29104115083253246,
+      "learning_rate": 1.4828171515070553e-05,
+      "loss": 0.1713,
+      "step": 772
+    },
+    {
+      "epoch": 3.1295546558704452,
+      "grad_norm": 0.28745169350539074,
+      "learning_rate": 1.4773561535742793e-05,
+      "loss": 0.1772,
+      "step": 773
+    },
+    {
+      "epoch": 3.133603238866397,
+      "grad_norm": 0.2820214901766566,
+      "learning_rate": 1.4718993346877715e-05,
+      "loss": 0.1795,
+      "step": 774
+    },
+    {
+      "epoch": 3.1376518218623484,
+      "grad_norm": 0.29599626726890715,
+      "learning_rate": 1.466446738480111e-05,
+      "loss": 0.1852,
+      "step": 775
+    },
+    {
+      "epoch": 3.1417004048582995,
+      "grad_norm": 0.27806222133511316,
+      "learning_rate": 1.460998408550112e-05,
+      "loss": 0.1802,
+      "step": 776
+    },
+    {
+      "epoch": 3.145748987854251,
+      "grad_norm": 0.27897213985619773,
+      "learning_rate": 1.4555543884624751e-05,
+      "loss": 0.1543,
+      "step": 777
+    },
+    {
+      "epoch": 3.1497975708502026,
+      "grad_norm": 0.2722380749474777,
+      "learning_rate": 1.4501147217474402e-05,
+      "loss": 0.1982,
+      "step": 778
+    },
+    {
+      "epoch": 3.1538461538461537,
+      "grad_norm": 0.27414016529811713,
+      "learning_rate": 1.444679451900437e-05,
+      "loss": 0.1738,
+      "step": 779
+    },
+    {
+      "epoch": 3.1578947368421053,
+      "grad_norm": 0.29816079262886447,
+      "learning_rate": 1.4392486223817397e-05,
+      "loss": 0.1503,
+      "step": 780
+    },
+    {
+      "epoch": 3.161943319838057,
+      "grad_norm": 0.2756307863297978,
+      "learning_rate": 1.4338222766161143e-05,
+      "loss": 0.188,
+      "step": 781
+    },
+    {
+      "epoch": 3.165991902834008,
+      "grad_norm": 0.30327995074901354,
+      "learning_rate": 1.4284004579924777e-05,
+      "loss": 0.1624,
+      "step": 782
+    },
+    {
+      "epoch": 3.1700404858299596,
+      "grad_norm": 0.28669244322268317,
+      "learning_rate": 1.4229832098635479e-05,
+      "loss": 0.1709,
+      "step": 783
+    },
+    {
+      "epoch": 3.174089068825911,
+      "grad_norm": 0.2666713803567914,
+      "learning_rate": 1.4175705755454963e-05,
+      "loss": 0.1874,
+      "step": 784
+    },
+    {
+      "epoch": 3.1781376518218623,
+      "grad_norm": 0.2902908578428875,
+      "learning_rate": 1.412162598317602e-05,
+      "loss": 0.1653,
+      "step": 785
+    },
+    {
+      "epoch": 3.182186234817814,
+      "grad_norm": 0.29965217368922575,
+      "learning_rate": 1.406759321421907e-05,
+      "loss": 0.18,
+      "step": 786
+    },
+    {
+      "epoch": 3.1862348178137654,
+      "grad_norm": 0.29352437673937215,
+      "learning_rate": 1.401360788062871e-05,
+      "loss": 0.1921,
+      "step": 787
+    },
+    {
+      "epoch": 3.1902834008097165,
+      "grad_norm": 0.3023137881318398,
+      "learning_rate": 1.3959670414070204e-05,
+      "loss": 0.1786,
+      "step": 788
+    },
+    {
+      "epoch": 3.194331983805668,
+      "grad_norm": 0.29641460260567687,
+      "learning_rate": 1.3905781245826108e-05,
+      "loss": 0.1667,
+      "step": 789
+    },
+    {
+      "epoch": 3.1983805668016196,
+      "grad_norm": 0.28008638179957585,
+      "learning_rate": 1.3851940806792778e-05,
+      "loss": 0.1785,
+      "step": 790
+    },
+    {
+      "epoch": 3.2024291497975708,
+      "grad_norm": 0.28845068293456866,
+      "learning_rate": 1.379814952747693e-05,
+      "loss": 0.1825,
+      "step": 791
+    },
+    {
+      "epoch": 3.2064777327935223,
+      "grad_norm": 0.28672905956442424,
+      "learning_rate": 1.3744407837992193e-05,
+      "loss": 0.1691,
+      "step": 792
+    },
+    {
+      "epoch": 3.2105263157894735,
+      "grad_norm": 0.2754148972088073,
+      "learning_rate": 1.3690716168055692e-05,
+      "loss": 0.1714,
+      "step": 793
+    },
+    {
+      "epoch": 3.214574898785425,
+      "grad_norm": 0.29192095385956734,
+      "learning_rate": 1.363707494698459e-05,
+      "loss": 0.2016,
+      "step": 794
+    },
+    {
+      "epoch": 3.2186234817813766,
+      "grad_norm": 0.27560978787760787,
+      "learning_rate": 1.358348460369265e-05,
+      "loss": 0.1781,
+      "step": 795
+    },
+    {
+      "epoch": 3.2226720647773277,
+      "grad_norm": 0.27324156970887886,
+      "learning_rate": 1.352994556668684e-05,
+      "loss": 0.1824,
+      "step": 796
+    },
+    {
+      "epoch": 3.2267206477732793,
+      "grad_norm": 0.2955289414338238,
+      "learning_rate": 1.3476458264063865e-05,
+      "loss": 0.1806,
+      "step": 797
+    },
+    {
+      "epoch": 3.230769230769231,
+      "grad_norm": 0.28381611416381125,
+      "learning_rate": 1.342302312350678e-05,
+      "loss": 0.1696,
+      "step": 798
+    },
+    {
+      "epoch": 3.234817813765182,
+      "grad_norm": 0.26618561589604617,
+      "learning_rate": 1.3369640572281537e-05,
+      "loss": 0.1824,
+      "step": 799
+    },
+    {
+      "epoch": 3.2388663967611335,
+      "grad_norm": 0.2814198058998902,
+      "learning_rate": 1.3316311037233596e-05,
+      "loss": 0.1784,
+      "step": 800
+    },
+    {
+      "epoch": 3.242914979757085,
+      "grad_norm": 0.2740188226895593,
+      "learning_rate": 1.326303494478451e-05,
+      "loss": 0.1715,
+      "step": 801
+    },
+    {
+      "epoch": 3.246963562753036,
+      "grad_norm": 0.2804678392774074,
+      "learning_rate": 1.3209812720928472e-05,
+      "loss": 0.1763,
+      "step": 802
+    },
+    {
+      "epoch": 3.251012145748988,
+      "grad_norm": 0.2658528888278077,
+      "learning_rate": 1.315664479122898e-05,
+      "loss": 0.1747,
+      "step": 803
+    },
+    {
+      "epoch": 3.2550607287449393,
+      "grad_norm": 0.2681773535385801,
+      "learning_rate": 1.3103531580815378e-05,
+      "loss": 0.184,
+      "step": 804
+    },
+    {
+      "epoch": 3.2591093117408905,
+      "grad_norm": 0.27876872428708677,
+      "learning_rate": 1.3050473514379488e-05,
+      "loss": 0.1776,
+      "step": 805
+    },
+    {
+      "epoch": 3.263157894736842,
+      "grad_norm": 0.28958844509685483,
+      "learning_rate": 1.2997471016172188e-05,
+      "loss": 0.1951,
+      "step": 806
+    },
+    {
+      "epoch": 3.2672064777327936,
+      "grad_norm": 0.2655563669290789,
+      "learning_rate": 1.2944524510000042e-05,
+      "loss": 0.1576,
+      "step": 807
+    },
+    {
+      "epoch": 3.2712550607287447,
+      "grad_norm": 0.2705041249293009,
+      "learning_rate": 1.2891634419221908e-05,
+      "loss": 0.1575,
+      "step": 808
+    },
+    {
+      "epoch": 3.2753036437246963,
+      "grad_norm": 0.2865603234359412,
+      "learning_rate": 1.2838801166745545e-05,
+      "loss": 0.1752,
+      "step": 809
+    },
+    {
+      "epoch": 3.279352226720648,
+      "grad_norm": 0.2647146098201965,
+      "learning_rate": 1.2786025175024216e-05,
+      "loss": 0.1849,
+      "step": 810
+    },
+    {
+      "epoch": 3.283400809716599,
+      "grad_norm": 0.25239172599232,
+      "learning_rate": 1.2733306866053357e-05,
+      "loss": 0.1914,
+      "step": 811
+    },
+    {
+      "epoch": 3.2874493927125505,
+      "grad_norm": 0.2619058656882606,
+      "learning_rate": 1.2680646661367163e-05,
+      "loss": 0.1588,
+      "step": 812
+    },
+    {
+      "epoch": 3.291497975708502,
+      "grad_norm": 0.26915308729517395,
+      "learning_rate": 1.2628044982035219e-05,
+      "loss": 0.1961,
+      "step": 813
+    },
+    {
+      "epoch": 3.2955465587044532,
+      "grad_norm": 0.26543936921441713,
+      "learning_rate": 1.2575502248659155e-05,
+      "loss": 0.1775,
+      "step": 814
+    },
+    {
+      "epoch": 3.299595141700405,
+      "grad_norm": 0.26804088996332026,
+      "learning_rate": 1.2523018881369269e-05,
+      "loss": 0.1909,
+      "step": 815
+    },
+    {
+      "epoch": 3.3036437246963564,
+      "grad_norm": 0.25223201926955896,
+      "learning_rate": 1.2470595299821176e-05,
+      "loss": 0.1859,
+      "step": 816
+    },
+    {
+      "epoch": 3.3076923076923075,
+      "grad_norm": 0.2591335514641335,
+      "learning_rate": 1.2418231923192416e-05,
+      "loss": 0.1842,
+      "step": 817
+    },
+    {
+      "epoch": 3.311740890688259,
+      "grad_norm": 0.2699577886376258,
+      "learning_rate": 1.2365929170179165e-05,
+      "loss": 0.1861,
+      "step": 818
+    },
+    {
+      "epoch": 3.3157894736842106,
+      "grad_norm": 0.27627979176281997,
+      "learning_rate": 1.2313687458992848e-05,
+      "loss": 0.1647,
+      "step": 819
+    },
+    {
+      "epoch": 3.3198380566801617,
+      "grad_norm": 0.2643215876309311,
+      "learning_rate": 1.2261507207356785e-05,
+      "loss": 0.1579,
+      "step": 820
+    },
+    {
+      "epoch": 3.3238866396761133,
+      "grad_norm": 0.268795946296114,
+      "learning_rate": 1.220938883250288e-05,
+      "loss": 0.1852,
+      "step": 821
+    },
+    {
+      "epoch": 3.327935222672065,
+      "grad_norm": 0.2567743917564207,
+      "learning_rate": 1.215733275116827e-05,
+      "loss": 0.1667,
+      "step": 822
+    },
+    {
+      "epoch": 3.331983805668016,
+      "grad_norm": 0.2641732107199845,
+      "learning_rate": 1.2105339379592005e-05,
+      "loss": 0.1608,
+      "step": 823
+    },
+    {
+      "epoch": 3.3360323886639676,
+      "grad_norm": 0.25430868827634107,
+      "learning_rate": 1.2053409133511681e-05,
+      "loss": 0.1725,
+      "step": 824
+    },
+    {
+      "epoch": 3.340080971659919,
+      "grad_norm": 0.26847384522552104,
+      "learning_rate": 1.2001542428160173e-05,
+      "loss": 0.1719,
+      "step": 825
+    },
+    {
+      "epoch": 3.3441295546558703,
+      "grad_norm": 0.277785240969414,
+      "learning_rate": 1.1949739678262278e-05,
+      "loss": 0.1746,
+      "step": 826
+    },
+    {
+      "epoch": 3.348178137651822,
+      "grad_norm": 0.2598615762042909,
+      "learning_rate": 1.189800129803139e-05,
+      "loss": 0.1835,
+      "step": 827
+    },
+    {
+      "epoch": 3.3522267206477734,
+      "grad_norm": 0.27442052410187096,
+      "learning_rate": 1.1846327701166228e-05,
+      "loss": 0.1654,
+      "step": 828
+    },
+    {
+      "epoch": 3.3562753036437245,
+      "grad_norm": 0.27316030584696177,
+      "learning_rate": 1.1794719300847493e-05,
+      "loss": 0.1762,
+      "step": 829
+    },
+    {
+      "epoch": 3.360323886639676,
+      "grad_norm": 0.29473402118378644,
+      "learning_rate": 1.1743176509734582e-05,
+      "loss": 0.1582,
+      "step": 830
+    },
+    {
+      "epoch": 3.3643724696356276,
+      "grad_norm": 0.2665449630636611,
+      "learning_rate": 1.1691699739962275e-05,
+      "loss": 0.187,
+      "step": 831
+    },
+    {
+      "epoch": 3.3684210526315788,
+      "grad_norm": 0.2569561094233251,
+      "learning_rate": 1.1640289403137456e-05,
+      "loss": 0.1775,
+      "step": 832
+    },
+    {
+      "epoch": 3.3724696356275303,
+      "grad_norm": 0.274192330781389,
+      "learning_rate": 1.1588945910335802e-05,
+      "loss": 0.19,
+      "step": 833
+    },
+    {
+      "epoch": 3.376518218623482,
+      "grad_norm": 0.2601898167536849,
+      "learning_rate": 1.1537669672098517e-05,
+      "loss": 0.1767,
+      "step": 834
+    },
+    {
+      "epoch": 3.380566801619433,
+      "grad_norm": 0.2613685882414638,
+      "learning_rate": 1.1486461098429036e-05,
+      "loss": 0.1741,
+      "step": 835
+    },
+    {
+      "epoch": 3.3846153846153846,
+      "grad_norm": 0.2640867776799683,
+      "learning_rate": 1.1435320598789747e-05,
+      "loss": 0.1816,
+      "step": 836
+    },
+    {
+      "epoch": 3.388663967611336,
+      "grad_norm": 0.25529303038449647,
+      "learning_rate": 1.1384248582098736e-05,
+      "loss": 0.1787,
+      "step": 837
+    },
+    {
+      "epoch": 3.3927125506072873,
+      "grad_norm": 0.25802868757997516,
+      "learning_rate": 1.1333245456726467e-05,
+      "loss": 0.186,
+      "step": 838
+    },
+    {
+      "epoch": 3.396761133603239,
+      "grad_norm": 0.27100687807388946,
+      "learning_rate": 1.128231163049258e-05,
+      "loss": 0.1507,
+      "step": 839
+    },
+    {
+      "epoch": 3.4008097165991904,
+      "grad_norm": 0.26252197179660197,
+      "learning_rate": 1.123144751066259e-05,
+      "loss": 0.1503,
+      "step": 840
+    },
+    {
+      "epoch": 3.4048582995951415,
+      "grad_norm": 0.26840078676193885,
+      "learning_rate": 1.1180653503944657e-05,
+      "loss": 0.1703,
+      "step": 841
+    },
+    {
+      "epoch": 3.408906882591093,
+      "grad_norm": 0.2815923672121167,
+      "learning_rate": 1.1129930016486287e-05,
+      "loss": 0.1795,
+      "step": 842
+    },
+    {
+      "epoch": 3.4129554655870447,
+      "grad_norm": 0.27210622797098205,
+      "learning_rate": 1.107927745387114e-05,
+      "loss": 0.1768,
+      "step": 843
+    },
+    {
+      "epoch": 3.417004048582996,
+      "grad_norm": 0.2611840286526943,
+      "learning_rate": 1.1028696221115766e-05,
+      "loss": 0.1721,
+      "step": 844
+    },
+    {
+      "epoch": 3.4210526315789473,
+      "grad_norm": 0.2529894506901741,
+      "learning_rate": 1.0978186722666341e-05,
+      "loss": 0.1827,
+      "step": 845
+    },
+    {
+      "epoch": 3.425101214574899,
+      "grad_norm": 0.25783395993603975,
+      "learning_rate": 1.0927749362395475e-05,
+      "loss": 0.1889,
+      "step": 846
+    },
+    {
+      "epoch": 3.42914979757085,
+      "grad_norm": 0.2553836673823914,
+      "learning_rate": 1.0877384543598954e-05,
+      "loss": 0.1791,
+      "step": 847
+    },
+    {
+      "epoch": 3.4331983805668016,
+      "grad_norm": 0.2585740684545916,
+      "learning_rate": 1.0827092668992529e-05,
+      "loss": 0.1875,
+      "step": 848
+    },
+    {
+      "epoch": 3.437246963562753,
+      "grad_norm": 0.26207858867404865,
+      "learning_rate": 1.077687414070869e-05,
+      "loss": 0.1645,
+      "step": 849
+    },
+    {
+      "epoch": 3.4412955465587043,
+      "grad_norm": 0.26594271048195395,
+      "learning_rate": 1.0726729360293451e-05,
+      "loss": 0.177,
+      "step": 850
+    },
+    {
+      "epoch": 3.445344129554656,
+      "grad_norm": 0.26898069967518723,
+      "learning_rate": 1.0676658728703147e-05,
+      "loss": 0.1856,
+      "step": 851
+    },
+    {
+      "epoch": 3.4493927125506074,
+      "grad_norm": 0.27224301859103595,
+      "learning_rate": 1.0626662646301198e-05,
+      "loss": 0.1618,
+      "step": 852
+    },
+    {
+      "epoch": 3.4534412955465585,
+      "grad_norm": 0.2884673902382327,
+      "learning_rate": 1.0576741512854958e-05,
+      "loss": 0.1835,
+      "step": 853
+    },
+    {
+      "epoch": 3.45748987854251,
+      "grad_norm": 0.2581842629492105,
+      "learning_rate": 1.052689572753247e-05,
+      "loss": 0.1706,
+      "step": 854
+    },
+    {
+      "epoch": 3.4615384615384617,
+      "grad_norm": 0.26632055690428413,
+      "learning_rate": 1.0477125688899328e-05,
+      "loss": 0.1711,
+      "step": 855
+    },
+    {
+      "epoch": 3.465587044534413,
+      "grad_norm": 0.2662078857474226,
+      "learning_rate": 1.0427431794915406e-05,
+      "loss": 0.1767,
+      "step": 856
+    },
+    {
+      "epoch": 3.4696356275303644,
+      "grad_norm": 0.26903932119175317,
+      "learning_rate": 1.0377814442931769e-05,
+      "loss": 0.1755,
+      "step": 857
+    },
+    {
+      "epoch": 3.473684210526316,
+      "grad_norm": 0.2578382352545654,
+      "learning_rate": 1.0328274029687439e-05,
+      "loss": 0.1642,
+      "step": 858
+    },
+    {
+      "epoch": 3.477732793522267,
+      "grad_norm": 0.2743691786787268,
+      "learning_rate": 1.0278810951306253e-05,
+      "loss": 0.1736,
+      "step": 859
+    },
+    {
+      "epoch": 3.4817813765182186,
+      "grad_norm": 0.2578289122235886,
+      "learning_rate": 1.0229425603293647e-05,
+      "loss": 0.1921,
+      "step": 860
+    },
+    {
+      "epoch": 3.48582995951417,
+      "grad_norm": 0.2730238710678897,
+      "learning_rate": 1.0180118380533558e-05,
+      "loss": 0.1712,
+      "step": 861
+    },
+    {
+      "epoch": 3.4898785425101213,
+      "grad_norm": 0.25919424007125674,
+      "learning_rate": 1.0130889677285225e-05,
+      "loss": 0.174,
+      "step": 862
+    },
+    {
+      "epoch": 3.493927125506073,
+      "grad_norm": 0.26897856430803885,
+      "learning_rate": 1.008173988718005e-05,
+      "loss": 0.1753,
+      "step": 863
+    },
+    {
+      "epoch": 3.4979757085020244,
+      "grad_norm": 0.25142755816447054,
+      "learning_rate": 1.0032669403218439e-05,
+      "loss": 0.175,
+      "step": 864
+    },
+    {
+      "epoch": 3.5020242914979756,
+      "grad_norm": 0.2647618110329252,
+      "learning_rate": 9.983678617766677e-06,
+      "loss": 0.1834,
+      "step": 865
+    },
+    {
+      "epoch": 3.506072874493927,
+      "grad_norm": 0.2625509805957499,
+      "learning_rate": 9.934767922553788e-06,
+      "loss": 0.1678,
+      "step": 866
+    },
+    {
+      "epoch": 3.5101214574898787,
+      "grad_norm": 0.2592637525918784,
+      "learning_rate": 9.885937708668371e-06,
+      "loss": 0.19,
+      "step": 867
+    },
+    {
+      "epoch": 3.51417004048583,
+      "grad_norm": 0.2576204767800957,
+      "learning_rate": 9.837188366555524e-06,
+      "loss": 0.1694,
+      "step": 868
+    },
+    {
+      "epoch": 3.5182186234817814,
+      "grad_norm": 0.26677501769182754,
+      "learning_rate": 9.788520286013691e-06,
+      "loss": 0.1925,
+      "step": 869
+    },
+    {
+      "epoch": 3.522267206477733,
+      "grad_norm": 0.26096521074264334,
+      "learning_rate": 9.739933856191537e-06,
+      "loss": 0.172,
+      "step": 870
+    },
+    {
+      "epoch": 3.526315789473684,
+      "grad_norm": 0.26236414500829486,
+      "learning_rate": 9.691429465584866e-06,
+      "loss": 0.18,
+      "step": 871
+    },
+    {
+      "epoch": 3.5303643724696356,
+      "grad_norm": 0.2585945139723032,
+      "learning_rate": 9.643007502033492e-06,
+      "loss": 0.1838,
+      "step": 872
+    },
+    {
+      "epoch": 3.534412955465587,
+      "grad_norm": 0.25985027640878106,
+      "learning_rate": 9.594668352718152e-06,
+      "loss": 0.1623,
+      "step": 873
+    },
+    {
+      "epoch": 3.5384615384615383,
+      "grad_norm": 0.2538477319356308,
+      "learning_rate": 9.546412404157383e-06,
+      "loss": 0.1522,
+      "step": 874
+    },
+    {
+      "epoch": 3.54251012145749,
+      "grad_norm": 0.2608244132049447,
+      "learning_rate": 9.498240042204471e-06,
+      "loss": 0.1813,
+      "step": 875
+    },
+    {
+      "epoch": 3.5465587044534415,
+      "grad_norm": 0.24746651832198052,
+      "learning_rate": 9.450151652044339e-06,
+      "loss": 0.1757,
+      "step": 876
+    },
+    {
+      "epoch": 3.5506072874493926,
+      "grad_norm": 0.2643742546163019,
+      "learning_rate": 9.402147618190472e-06,
+      "loss": 0.1708,
+      "step": 877
+    },
+    {
+      "epoch": 3.554655870445344,
+      "grad_norm": 0.2573936150769756,
+      "learning_rate": 9.354228324481841e-06,
+      "loss": 0.187,
+      "step": 878
+    },
+    {
+      "epoch": 3.5587044534412957,
+      "grad_norm": 0.26318799100002926,
+      "learning_rate": 9.306394154079852e-06,
+      "loss": 0.1551,
+      "step": 879
+    },
+    {
+      "epoch": 3.562753036437247,
+      "grad_norm": 0.250361904002301,
+      "learning_rate": 9.258645489465254e-06,
+      "loss": 0.177,
+      "step": 880
+    },
+    {
+      "epoch": 3.5668016194331984,
+      "grad_norm": 0.2609750691886062,
+      "learning_rate": 9.21098271243509e-06,
+      "loss": 0.1729,
+      "step": 881
+    },
+    {
+      "epoch": 3.57085020242915,
+      "grad_norm": 0.25899166475515606,
+      "learning_rate": 9.163406204099659e-06,
+      "loss": 0.1641,
+      "step": 882
+    },
+    {
+      "epoch": 3.574898785425101,
+      "grad_norm": 0.2515099369753396,
+      "learning_rate": 9.115916344879452e-06,
+      "loss": 0.1628,
+      "step": 883
+    },
+    {
+      "epoch": 3.5789473684210527,
+      "grad_norm": 0.2594611363772046,
+      "learning_rate": 9.068513514502126e-06,
+      "loss": 0.1575,
+      "step": 884
+    },
+    {
+      "epoch": 3.582995951417004,
+      "grad_norm": 0.2620622746829401,
+      "learning_rate": 9.021198091999439e-06,
+      "loss": 0.1796,
+      "step": 885
+    },
+    {
+      "epoch": 3.5870445344129553,
+      "grad_norm": 0.26152852392852516,
+      "learning_rate": 8.973970455704251e-06,
+      "loss": 0.1682,
+      "step": 886
+    },
+    {
+      "epoch": 3.591093117408907,
+      "grad_norm": 0.2522455102693992,
+      "learning_rate": 8.926830983247495e-06,
+      "loss": 0.1661,
+      "step": 887
+    },
+    {
+      "epoch": 3.5951417004048585,
+      "grad_norm": 0.2515498593179143,
+      "learning_rate": 8.879780051555116e-06,
+      "loss": 0.1609,
+      "step": 888
+    },
+    {
+      "epoch": 3.5991902834008096,
+      "grad_norm": 0.2800414127926491,
+      "learning_rate": 8.83281803684512e-06,
+      "loss": 0.1604,
+      "step": 889
+    },
+    {
+      "epoch": 3.603238866396761,
+      "grad_norm": 0.2915618687525542,
+      "learning_rate": 8.785945314624532e-06,
+      "loss": 0.1757,
+      "step": 890
+    },
+    {
+      "epoch": 3.6072874493927127,
+      "grad_norm": 0.2576028612150738,
+      "learning_rate": 8.739162259686385e-06,
+      "loss": 0.177,
+      "step": 891
+    },
+    {
+      "epoch": 3.611336032388664,
+      "grad_norm": 0.25803661237017034,
+      "learning_rate": 8.692469246106734e-06,
+      "loss": 0.1787,
+      "step": 892
+    },
+    {
+      "epoch": 3.6153846153846154,
+      "grad_norm": 0.2611716800248797,
+      "learning_rate": 8.645866647241677e-06,
+      "loss": 0.1774,
+      "step": 893
+    },
+    {
+      "epoch": 3.619433198380567,
+      "grad_norm": 0.2604285443715423,
+      "learning_rate": 8.599354835724348e-06,
+      "loss": 0.1745,
+      "step": 894
+    },
+    {
+      "epoch": 3.623481781376518,
+      "grad_norm": 0.2545414803739149,
+      "learning_rate": 8.552934183461958e-06,
+      "loss": 0.1976,
+      "step": 895
+    },
+    {
+      "epoch": 3.6275303643724697,
+      "grad_norm": 0.26254106546382666,
+      "learning_rate": 8.506605061632805e-06,
+      "loss": 0.1714,
+      "step": 896
+    },
+    {
+      "epoch": 3.6315789473684212,
+      "grad_norm": 0.26944530044549,
+      "learning_rate": 8.460367840683308e-06,
+      "loss": 0.1795,
+      "step": 897
+    },
+    {
+      "epoch": 3.6356275303643724,
+      "grad_norm": 0.24688042904007684,
+      "learning_rate": 8.414222890325065e-06,
+      "loss": 0.187,
+      "step": 898
+    },
+    {
+      "epoch": 3.639676113360324,
+      "grad_norm": 0.26746093369743734,
+      "learning_rate": 8.368170579531855e-06,
+      "loss": 0.1484,
+      "step": 899
+    },
+    {
+      "epoch": 3.6437246963562755,
+      "grad_norm": 0.2621125824602388,
+      "learning_rate": 8.322211276536733e-06,
+      "loss": 0.1567,
+      "step": 900
+    },
+    {
+      "epoch": 3.6477732793522266,
+      "grad_norm": 0.2631100549499284,
+      "learning_rate": 8.27634534882907e-06,
+      "loss": 0.1694,
+      "step": 901
+    },
+    {
+      "epoch": 3.651821862348178,
+      "grad_norm": 0.251891979503433,
+      "learning_rate": 8.230573163151585e-06,
+      "loss": 0.1565,
+      "step": 902
+    },
+    {
+      "epoch": 3.6558704453441297,
+      "grad_norm": 0.2597821661679219,
+      "learning_rate": 8.184895085497466e-06,
+      "loss": 0.1793,
+      "step": 903
+    },
+    {
+      "epoch": 3.659919028340081,
+      "grad_norm": 0.251753805963078,
+      "learning_rate": 8.139311481107401e-06,
+      "loss": 0.1743,
+      "step": 904
+    },
+    {
+      "epoch": 3.6639676113360324,
+      "grad_norm": 0.2729795353448808,
+      "learning_rate": 8.093822714466686e-06,
+      "loss": 0.1646,
+      "step": 905
+    },
+    {
+      "epoch": 3.668016194331984,
+      "grad_norm": 0.2524516752519574,
+      "learning_rate": 8.048429149302271e-06,
+      "loss": 0.1726,
+      "step": 906
+    },
+    {
+      "epoch": 3.672064777327935,
+      "grad_norm": 0.2448404283931466,
+      "learning_rate": 8.003131148579902e-06,
+      "loss": 0.1796,
+      "step": 907
+    },
+    {
+      "epoch": 3.6761133603238867,
+      "grad_norm": 0.24754425146426248,
+      "learning_rate": 7.957929074501182e-06,
+      "loss": 0.1829,
+      "step": 908
+    },
+    {
+      "epoch": 3.6801619433198383,
+      "grad_norm": 0.24342333299561106,
+      "learning_rate": 7.912823288500695e-06,
+      "loss": 0.1759,
+      "step": 909
+    },
+    {
+      "epoch": 3.6842105263157894,
+      "grad_norm": 0.24372561284397123,
+      "learning_rate": 7.867814151243102e-06,
+      "loss": 0.1815,
+      "step": 910
+    },
+    {
+      "epoch": 3.688259109311741,
+      "grad_norm": 0.2595132543130526,
+      "learning_rate": 7.822902022620267e-06,
+      "loss": 0.1629,
+      "step": 911
+    },
+    {
+      "epoch": 3.6923076923076925,
+      "grad_norm": 0.2708256100633346,
+      "learning_rate": 7.778087261748377e-06,
+      "loss": 0.177,
+      "step": 912
+    },
+    {
+      "epoch": 3.6963562753036436,
+      "grad_norm": 0.2680508010873069,
+      "learning_rate": 7.733370226965056e-06,
+      "loss": 0.1812,
+      "step": 913
+    },
+    {
+      "epoch": 3.700404858299595,
+      "grad_norm": 0.24992803547897352,
+      "learning_rate": 7.688751275826527e-06,
+      "loss": 0.1832,
+      "step": 914
+    },
+    {
+      "epoch": 3.7044534412955468,
+      "grad_norm": 0.277383858422052,
+      "learning_rate": 7.644230765104735e-06,
+      "loss": 0.1651,
+      "step": 915
+    },
+    {
+      "epoch": 3.708502024291498,
+      "grad_norm": 0.26934149052801015,
+      "learning_rate": 7.599809050784502e-06,
+      "loss": 0.1766,
+      "step": 916
+    },
+    {
+      "epoch": 3.7125506072874495,
+      "grad_norm": 0.25483589738697554,
+      "learning_rate": 7.555486488060661e-06,
+      "loss": 0.171,
+      "step": 917
+    },
+    {
+      "epoch": 3.716599190283401,
+      "grad_norm": 0.26682581594210614,
+      "learning_rate": 7.511263431335249e-06,
+      "loss": 0.1696,
+      "step": 918
+    },
+    {
+      "epoch": 3.720647773279352,
+      "grad_norm": 0.2612364059016844,
+      "learning_rate": 7.467140234214656e-06,
+      "loss": 0.1654,
+      "step": 919
+    },
+    {
+      "epoch": 3.7246963562753037,
+      "grad_norm": 0.262721678381597,
+      "learning_rate": 7.423117249506777e-06,
+      "loss": 0.1801,
+      "step": 920
+    },
+    {
+      "epoch": 3.7287449392712553,
+      "grad_norm": 0.25921454588506704,
+      "learning_rate": 7.379194829218231e-06,
+      "loss": 0.1796,
+      "step": 921
+    },
+    {
+      "epoch": 3.7327935222672064,
+      "grad_norm": 0.2538316724143131,
+      "learning_rate": 7.335373324551518e-06,
+      "loss": 0.1897,
+      "step": 922
+    },
+    {
+      "epoch": 3.736842105263158,
+      "grad_norm": 0.28454267121969906,
+      "learning_rate": 7.291653085902224e-06,
+      "loss": 0.1764,
+      "step": 923
+    },
+    {
+      "epoch": 3.7408906882591095,
+      "grad_norm": 0.26165435314435875,
+      "learning_rate": 7.248034462856213e-06,
+      "loss": 0.1699,
+      "step": 924
+    },
+    {
+      "epoch": 3.7449392712550607,
+      "grad_norm": 0.2516277902695165,
+      "learning_rate": 7.20451780418683e-06,
+      "loss": 0.1752,
+      "step": 925
+    },
+    {
+      "epoch": 3.748987854251012,
+      "grad_norm": 0.2622653264440862,
+      "learning_rate": 7.161103457852128e-06,
+      "loss": 0.1907,
+      "step": 926
+    },
+    {
+      "epoch": 3.753036437246964,
+      "grad_norm": 0.26829957264575516,
+      "learning_rate": 7.117791770992046e-06,
+      "loss": 0.1935,
+      "step": 927
+    },
+    {
+      "epoch": 3.757085020242915,
+      "grad_norm": 0.26986135068422384,
+      "learning_rate": 7.074583089925688e-06,
+      "loss": 0.1839,
+      "step": 928
+    },
+    {
+      "epoch": 3.7611336032388665,
+      "grad_norm": 0.256856071999415,
+      "learning_rate": 7.0314777601485155e-06,
+      "loss": 0.1844,
+      "step": 929
+    },
+    {
+      "epoch": 3.765182186234818,
+      "grad_norm": 0.2503394793351316,
+      "learning_rate": 6.988476126329604e-06,
+      "loss": 0.1676,
+      "step": 930
+    },
+    {
+      "epoch": 3.769230769230769,
+      "grad_norm": 0.271719261774843,
+      "learning_rate": 6.945578532308857e-06,
+      "loss": 0.1652,
+      "step": 931
+    },
+    {
+      "epoch": 3.7732793522267207,
+      "grad_norm": 0.26277028695066745,
+      "learning_rate": 6.902785321094301e-06,
+      "loss": 0.1837,
+      "step": 932
+    },
+    {
+      "epoch": 3.7773279352226723,
+      "grad_norm": 0.2401457237285394,
+      "learning_rate": 6.8600968348593115e-06,
+      "loss": 0.1784,
+      "step": 933
+    },
+    {
+      "epoch": 3.7813765182186234,
+      "grad_norm": 0.2568452658632874,
+      "learning_rate": 6.817513414939892e-06,
+      "loss": 0.1792,
+      "step": 934
+    },
+    {
+      "epoch": 3.785425101214575,
+      "grad_norm": 0.26717716553063375,
+      "learning_rate": 6.775035401831924e-06,
+      "loss": 0.1821,
+      "step": 935
+    },
+    {
+      "epoch": 3.7894736842105265,
+      "grad_norm": 0.24292407989332876,
+      "learning_rate": 6.732663135188473e-06,
+      "loss": 0.1809,
+      "step": 936
+    },
+    {
+      "epoch": 3.7935222672064777,
+      "grad_norm": 0.24988930340368062,
+      "learning_rate": 6.690396953817055e-06,
+      "loss": 0.1897,
+      "step": 937
+    },
+    {
+      "epoch": 3.7975708502024292,
+      "grad_norm": 0.2521072996569165,
+      "learning_rate": 6.648237195676927e-06,
+      "loss": 0.1686,
+      "step": 938
+    },
+    {
+      "epoch": 3.801619433198381,
+      "grad_norm": 0.24665314226353946,
+      "learning_rate": 6.606184197876395e-06,
+      "loss": 0.178,
+      "step": 939
+    },
+    {
+      "epoch": 3.805668016194332,
+      "grad_norm": 0.25611205102829615,
+      "learning_rate": 6.564238296670109e-06,
+      "loss": 0.1707,
+      "step": 940
+    },
+    {
+      "epoch": 3.8097165991902835,
+      "grad_norm": 0.2529699416981883,
+      "learning_rate": 6.52239982745638e-06,
+      "loss": 0.1732,
+      "step": 941
+    },
+    {
+      "epoch": 3.813765182186235,
+      "grad_norm": 0.24234753597019795,
+      "learning_rate": 6.480669124774484e-06,
+      "loss": 0.1838,
+      "step": 942
+    },
+    {
+      "epoch": 3.817813765182186,
+      "grad_norm": 0.2506482875290628,
+      "learning_rate": 6.439046522302013e-06,
+      "loss": 0.1695,
+      "step": 943
+    },
+    {
+      "epoch": 3.8218623481781377,
+      "grad_norm": 0.2485413688517496,
+      "learning_rate": 6.397532352852191e-06,
+      "loss": 0.1879,
+      "step": 944
+    },
+    {
+      "epoch": 3.8259109311740893,
+      "grad_norm": 0.24626711379691865,
+      "learning_rate": 6.356126948371202e-06,
+      "loss": 0.1696,
+      "step": 945
+    },
+    {
+      "epoch": 3.8299595141700404,
+      "grad_norm": 0.2587365870790454,
+      "learning_rate": 6.314830639935559e-06,
+      "loss": 0.1775,
+      "step": 946
+    },
+    {
+      "epoch": 3.834008097165992,
+      "grad_norm": 0.24446530708492506,
+      "learning_rate": 6.273643757749448e-06,
+      "loss": 0.1824,
+      "step": 947
+    },
+    {
+      "epoch": 3.8380566801619436,
+      "grad_norm": 0.2461675961039812,
+      "learning_rate": 6.2325666311420856e-06,
+      "loss": 0.1791,
+      "step": 948
+    },
+    {
+      "epoch": 3.8421052631578947,
+      "grad_norm": 0.24953194296591408,
+      "learning_rate": 6.1915995885650675e-06,
+      "loss": 0.1747,
+      "step": 949
+    },
+    {
+      "epoch": 3.8461538461538463,
+      "grad_norm": 0.25272637357833966,
+      "learning_rate": 6.150742957589784e-06,
+      "loss": 0.1603,
+      "step": 950
+    },
+    {
+      "epoch": 3.850202429149798,
+      "grad_norm": 0.23976741153410136,
+      "learning_rate": 6.109997064904774e-06,
+      "loss": 0.1823,
+      "step": 951
+    },
+    {
+      "epoch": 3.854251012145749,
+      "grad_norm": 0.24693492889288737,
+      "learning_rate": 6.069362236313092e-06,
+      "loss": 0.1643,
+      "step": 952
+    },
+    {
+      "epoch": 3.8582995951417005,
+      "grad_norm": 0.24680064511321456,
+      "learning_rate": 6.0288387967297565e-06,
+      "loss": 0.171,
+      "step": 953
+    },
+    {
+      "epoch": 3.862348178137652,
+      "grad_norm": 0.24854843374243712,
+      "learning_rate": 5.988427070179108e-06,
+      "loss": 0.1726,
+      "step": 954
+    },
+    {
+      "epoch": 3.866396761133603,
+      "grad_norm": 0.2628097088099847,
+      "learning_rate": 5.948127379792234e-06,
+      "loss": 0.1842,
+      "step": 955
+    },
+    {
+      "epoch": 3.8704453441295548,
+      "grad_norm": 0.25909809443195053,
+      "learning_rate": 5.90794004780439e-06,
+      "loss": 0.1739,
+      "step": 956
+    },
+    {
+      "epoch": 3.8744939271255063,
+      "grad_norm": 0.25585166863461056,
+      "learning_rate": 5.867865395552405e-06,
+      "loss": 0.1777,
+      "step": 957
+    },
+    {
+      "epoch": 3.8785425101214575,
+      "grad_norm": 0.25142636546134106,
+      "learning_rate": 5.8279037434721364e-06,
+      "loss": 0.1857,
+      "step": 958
+    },
+    {
+      "epoch": 3.882591093117409,
+      "grad_norm": 0.25273440016904336,
+      "learning_rate": 5.788055411095892e-06,
+      "loss": 0.1648,
+      "step": 959
+    },
+    {
+      "epoch": 3.8866396761133606,
+      "grad_norm": 0.24729712691923836,
+      "learning_rate": 5.7483207170498625e-06,
+      "loss": 0.1729,
+      "step": 960
+    },
+    {
+      "epoch": 3.8906882591093117,
+      "grad_norm": 0.26532306587655236,
+      "learning_rate": 5.708699979051613e-06,
+      "loss": 0.1787,
+      "step": 961
+    },
+    {
+      "epoch": 3.8947368421052633,
+      "grad_norm": 0.24241189175968927,
+      "learning_rate": 5.66919351390751e-06,
+      "loss": 0.184,
+      "step": 962
+    },
+    {
+      "epoch": 3.898785425101215,
+      "grad_norm": 0.24409101490560906,
+      "learning_rate": 5.629801637510188e-06,
+      "loss": 0.1814,
+      "step": 963
+    },
+    {
+      "epoch": 3.902834008097166,
+      "grad_norm": 0.2562914111779602,
+      "learning_rate": 5.590524664836043e-06,
+      "loss": 0.1553,
+      "step": 964
+    },
+    {
+      "epoch": 3.9068825910931175,
+      "grad_norm": 0.26430948276942734,
+      "learning_rate": 5.551362909942708e-06,
+      "loss": 0.1691,
+      "step": 965
+    },
+    {
+      "epoch": 3.910931174089069,
+      "grad_norm": 0.24796684346559528,
+      "learning_rate": 5.51231668596653e-06,
+      "loss": 0.1885,
+      "step": 966
+    },
+    {
+      "epoch": 3.91497975708502,
+      "grad_norm": 0.2502659530156121,
+      "learning_rate": 5.47338630512007e-06,
+      "loss": 0.1599,
+      "step": 967
+    },
+    {
+      "epoch": 3.919028340080972,
+      "grad_norm": 0.24447011290830825,
+      "learning_rate": 5.434572078689617e-06,
+      "loss": 0.1878,
+      "step": 968
+    },
+    {
+      "epoch": 3.9230769230769234,
+      "grad_norm": 0.23632164048104023,
+      "learning_rate": 5.395874317032694e-06,
+      "loss": 0.1917,
+      "step": 969
+    },
+    {
+      "epoch": 3.9271255060728745,
+      "grad_norm": 0.2535149550456872,
+      "learning_rate": 5.357293329575573e-06,
+      "loss": 0.1778,
+      "step": 970
+    },
+    {
+      "epoch": 3.931174089068826,
+      "grad_norm": 0.2584263921996791,
+      "learning_rate": 5.318829424810797e-06,
+      "loss": 0.1504,
+      "step": 971
+    },
+    {
+      "epoch": 3.9352226720647776,
+      "grad_norm": 0.26137515149581975,
+      "learning_rate": 5.280482910294728e-06,
+      "loss": 0.1786,
+      "step": 972
+    },
+    {
+      "epoch": 3.9392712550607287,
+      "grad_norm": 0.24978563762890965,
+      "learning_rate": 5.242254092645078e-06,
+      "loss": 0.1715,
+      "step": 973
+    },
+    {
+      "epoch": 3.9433198380566803,
+      "grad_norm": 0.2546898831617932,
+      "learning_rate": 5.2041432775384445e-06,
+      "loss": 0.1554,
+      "step": 974
+    },
+    {
+      "epoch": 3.9473684210526314,
+      "grad_norm": 0.24797252251889357,
+      "learning_rate": 5.166150769707892e-06,
+      "loss": 0.1746,
+      "step": 975
+    },
+    {
+      "epoch": 3.951417004048583,
+      "grad_norm": 0.24612548905484172,
+      "learning_rate": 5.128276872940503e-06,
+      "loss": 0.1607,
+      "step": 976
+    },
+    {
+      "epoch": 3.9554655870445345,
+      "grad_norm": 0.25094499869576686,
+      "learning_rate": 5.090521890074953e-06,
+      "loss": 0.1741,
+      "step": 977
+    },
+    {
+      "epoch": 3.9595141700404857,
+      "grad_norm": 0.24852188462661892,
+      "learning_rate": 5.0528861229990675e-06,
+      "loss": 0.1765,
+      "step": 978
+    },
+    {
+      "epoch": 3.9635627530364372,
+      "grad_norm": 0.24295036812406617,
+      "learning_rate": 5.015369872647442e-06,
+      "loss": 0.1882,
+      "step": 979
+    },
+    {
+      "epoch": 3.967611336032389,
+      "grad_norm": 0.25662600238743116,
+      "learning_rate": 4.9779734389990195e-06,
+      "loss": 0.1815,
+      "step": 980
+    },
+    {
+      "epoch": 3.97165991902834,
+      "grad_norm": 0.263347050816017,
+      "learning_rate": 4.940697121074675e-06,
+      "loss": 0.1769,
+      "step": 981
+    },
+    {
+      "epoch": 3.9757085020242915,
+      "grad_norm": 0.25086943442096354,
+      "learning_rate": 4.903541216934864e-06,
+      "loss": 0.1768,
+      "step": 982
+    },
+    {
+      "epoch": 3.979757085020243,
+      "grad_norm": 0.2340244639635422,
+      "learning_rate": 4.8665060236772e-06,
+      "loss": 0.1773,
+      "step": 983
+    },
+    {
+      "epoch": 3.983805668016194,
+      "grad_norm": 0.23839027099990628,
+      "learning_rate": 4.829591837434102e-06,
+      "loss": 0.1788,
+      "step": 984
+    },
+    {
+      "epoch": 3.9878542510121457,
+      "grad_norm": 0.23689970870425883,
+      "learning_rate": 4.792798953370424e-06,
+      "loss": 0.1725,
+      "step": 985
+    },
+    {
+      "epoch": 3.9919028340080973,
+      "grad_norm": 0.25146806532261967,
+      "learning_rate": 4.756127665681083e-06,
+      "loss": 0.1499,
+      "step": 986
+    },
+    {
+      "epoch": 3.9959514170040484,
+      "grad_norm": 0.23552711059190606,
+      "learning_rate": 4.719578267588725e-06,
+      "loss": 0.1852,
+      "step": 987
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 0.23804332836820086,
+      "learning_rate": 4.6831510513413525e-06,
+      "loss": 0.1734,
+      "step": 988
+    },
+    {
+      "epoch": 4.004048582995951,
+      "grad_norm": 0.3339008318078217,
+      "learning_rate": 4.6468463082100176e-06,
+      "loss": 0.1434,
+      "step": 989
+    },
+    {
+      "epoch": 4.008097165991903,
+      "grad_norm": 0.28964943245361874,
+      "learning_rate": 4.610664328486485e-06,
+      "loss": 0.1312,
+      "step": 990
+    },
+    {
+      "epoch": 4.012145748987854,
+      "grad_norm": 0.21940029137101089,
+      "learning_rate": 4.5746054014809e-06,
+      "loss": 0.1079,
+      "step": 991
+    },
+    {
+      "epoch": 4.016194331983805,
+      "grad_norm": 0.2617718525075635,
+      "learning_rate": 4.5386698155194766e-06,
+      "loss": 0.1223,
+      "step": 992
+    },
+    {
+      "epoch": 4.020242914979757,
+      "grad_norm": 0.36442767902836787,
+      "learning_rate": 4.502857857942204e-06,
+      "loss": 0.1473,
+      "step": 993
+    },
+    {
+      "epoch": 4.0242914979757085,
+      "grad_norm": 0.3602736353181695,
+      "learning_rate": 4.467169815100549e-06,
+      "loss": 0.1492,
+      "step": 994
+    },
+    {
+      "epoch": 4.02834008097166,
+      "grad_norm": 0.3039873112723803,
+      "learning_rate": 4.431605972355137e-06,
+      "loss": 0.1342,
+      "step": 995
+    },
+    {
+      "epoch": 4.032388663967612,
+      "grad_norm": 0.22994535892480614,
+      "learning_rate": 4.396166614073516e-06,
+      "loss": 0.1334,
+      "step": 996
+    },
+    {
+      "epoch": 4.036437246963563,
+      "grad_norm": 0.28207699703330225,
+      "learning_rate": 4.360852023627846e-06,
+      "loss": 0.1236,
+      "step": 997
+    },
+    {
+      "epoch": 4.040485829959514,
+      "grad_norm": 0.3133666018722317,
+      "learning_rate": 4.325662483392656e-06,
+      "loss": 0.1397,
+      "step": 998
+    },
+    {
+      "epoch": 4.044534412955466,
+      "grad_norm": 0.29479857177414853,
+      "learning_rate": 4.29059827474257e-06,
+      "loss": 0.1293,
+      "step": 999
+    },
+    {
+      "epoch": 4.048582995951417,
+      "grad_norm": 0.2438543809637198,
+      "learning_rate": 4.255659678050063e-06,
+      "loss": 0.1255,
+      "step": 1000
+    },
+    {
+      "epoch": 4.052631578947368,
+      "grad_norm": 0.23056189460216275,
+      "learning_rate": 4.22084697268323e-06,
+      "loss": 0.1394,
+      "step": 1001
+    },
+    {
+      "epoch": 4.05668016194332,
+      "grad_norm": 0.22852533462925081,
+      "learning_rate": 4.186160437003535e-06,
+      "loss": 0.1355,
+      "step": 1002
+    },
+    {
+      "epoch": 4.060728744939271,
+      "grad_norm": 0.24532169068202803,
+      "learning_rate": 4.151600348363582e-06,
+      "loss": 0.1438,
+      "step": 1003
+    },
+    {
+      "epoch": 4.064777327935222,
+      "grad_norm": 0.2679875397766209,
+      "learning_rate": 4.11716698310493e-06,
+      "loss": 0.1383,
+      "step": 1004
+    },
+    {
+      "epoch": 4.068825910931174,
+      "grad_norm": 0.26111547069095675,
+      "learning_rate": 4.082860616555848e-06,
+      "loss": 0.1385,
+      "step": 1005
+    },
+    {
+      "epoch": 4.0728744939271255,
+      "grad_norm": 0.22042134091631396,
+      "learning_rate": 4.048681523029125e-06,
+      "loss": 0.1233,
+      "step": 1006
+    },
+    {
+      "epoch": 4.076923076923077,
+      "grad_norm": 0.22162782454233076,
+      "learning_rate": 4.014629975819882e-06,
+      "loss": 0.1333,
+      "step": 1007
+    },
+    {
+      "epoch": 4.080971659919029,
+      "grad_norm": 0.22585968812104484,
+      "learning_rate": 3.9807062472033855e-06,
+      "loss": 0.1302,
+      "step": 1008
+    },
+    {
+      "epoch": 4.08502024291498,
+      "grad_norm": 0.2201041296047578,
+      "learning_rate": 3.94691060843287e-06,
+      "loss": 0.1249,
+      "step": 1009
+    },
+    {
+      "epoch": 4.089068825910931,
+      "grad_norm": 0.21195882832097263,
+      "learning_rate": 3.913243329737353e-06,
+      "loss": 0.1404,
+      "step": 1010
+    },
+    {
+      "epoch": 4.093117408906883,
+      "grad_norm": 0.22364343817471066,
+      "learning_rate": 3.879704680319503e-06,
+      "loss": 0.1278,
+      "step": 1011
+    },
+    {
+      "epoch": 4.097165991902834,
+      "grad_norm": 0.2243312127912922,
+      "learning_rate": 3.846294928353471e-06,
+      "loss": 0.1311,
+      "step": 1012
+    },
+    {
+      "epoch": 4.101214574898785,
+      "grad_norm": 0.22570257242078134,
+      "learning_rate": 3.8130143409827345e-06,
+      "loss": 0.1391,
+      "step": 1013
+    },
+    {
+      "epoch": 4.105263157894737,
+      "grad_norm": 0.22354246470644362,
+      "learning_rate": 3.7798631843179887e-06,
+      "loss": 0.1442,
+      "step": 1014
+    },
+    {
+      "epoch": 4.109311740890688,
+      "grad_norm": 0.22610269586302795,
+      "learning_rate": 3.7468417234349973e-06,
+      "loss": 0.1326,
+      "step": 1015
+    },
+    {
+      "epoch": 4.113360323886639,
+      "grad_norm": 0.21769113626051587,
+      "learning_rate": 3.7139502223724823e-06,
+      "loss": 0.1349,
+      "step": 1016
+    },
+    {
+      "epoch": 4.117408906882591,
+      "grad_norm": 0.21860421986348444,
+      "learning_rate": 3.6811889441300095e-06,
+      "loss": 0.1344,
+      "step": 1017
+    },
+    {
+      "epoch": 4.1214574898785425,
+      "grad_norm": 0.22237320919756215,
+      "learning_rate": 3.648558150665886e-06,
+      "loss": 0.1403,
+      "step": 1018
+    },
+    {
+      "epoch": 4.125506072874494,
+      "grad_norm": 0.21660078587875936,
+      "learning_rate": 3.616058102895075e-06,
+      "loss": 0.1262,
+      "step": 1019
+    },
+    {
+      "epoch": 4.129554655870446,
+      "grad_norm": 0.21683756260947518,
+      "learning_rate": 3.583689060687079e-06,
+      "loss": 0.1296,
+      "step": 1020
+    },
+    {
+      "epoch": 4.133603238866397,
+      "grad_norm": 0.22862954089505375,
+      "learning_rate": 3.5514512828639023e-06,
+      "loss": 0.1196,
+      "step": 1021
+    },
+    {
+      "epoch": 4.137651821862348,
+      "grad_norm": 0.21466277975387513,
+      "learning_rate": 3.51934502719796e-06,
+      "loss": 0.1153,
+      "step": 1022
+    },
+    {
+      "epoch": 4.1417004048583,
+      "grad_norm": 0.21263662079629536,
+      "learning_rate": 3.4873705504100185e-06,
+      "loss": 0.1568,
+      "step": 1023
+    },
+    {
+      "epoch": 4.145748987854251,
+      "grad_norm": 0.20950307473065186,
+      "learning_rate": 3.4555281081671345e-06,
+      "loss": 0.1465,
+      "step": 1024
+    },
+    {
+      "epoch": 4.149797570850202,
+      "grad_norm": 0.21230578951172438,
+      "learning_rate": 3.4238179550806326e-06,
+      "loss": 0.111,
+      "step": 1025
+    },
+    {
+      "epoch": 4.153846153846154,
+      "grad_norm": 0.2116454523519856,
+      "learning_rate": 3.3922403447040497e-06,
+      "loss": 0.1408,
+      "step": 1026
+    },
+    {
+      "epoch": 4.157894736842105,
+      "grad_norm": 0.20857972241523584,
+      "learning_rate": 3.3607955295311247e-06,
+      "loss": 0.137,
+      "step": 1027
+    },
+    {
+      "epoch": 4.161943319838056,
+      "grad_norm": 0.21094539701695098,
+      "learning_rate": 3.32948376099375e-06,
+      "loss": 0.1356,
+      "step": 1028
+    },
+    {
+      "epoch": 4.165991902834008,
+      "grad_norm": 0.21780551195642683,
+      "learning_rate": 3.2983052894599976e-06,
+      "loss": 0.1162,
+      "step": 1029
+    },
+    {
+      "epoch": 4.17004048582996,
+      "grad_norm": 0.22435707899178484,
+      "learning_rate": 3.2672603642320945e-06,
+      "loss": 0.1347,
+      "step": 1030
+    },
+    {
+      "epoch": 4.174089068825911,
+      "grad_norm": 0.226635899684846,
+      "learning_rate": 3.236349233544438e-06,
+      "loss": 0.1479,
+      "step": 1031
+    },
+    {
+      "epoch": 4.178137651821863,
+      "grad_norm": 0.21702416050389728,
+      "learning_rate": 3.205572144561604e-06,
+      "loss": 0.127,
+      "step": 1032
+    },
+    {
+      "epoch": 4.182186234817814,
+      "grad_norm": 0.21461449420806158,
+      "learning_rate": 3.174929343376374e-06,
+      "loss": 0.1076,
+      "step": 1033
+    },
+    {
+      "epoch": 4.186234817813765,
+      "grad_norm": 0.21633585199543698,
+      "learning_rate": 3.1444210750077777e-06,
+      "loss": 0.1218,
+      "step": 1034
+    },
+    {
+      "epoch": 4.190283400809717,
+      "grad_norm": 0.2066708480015723,
+      "learning_rate": 3.1140475833991046e-06,
+      "loss": 0.1182,
+      "step": 1035
+    },
+    {
+      "epoch": 4.194331983805668,
+      "grad_norm": 0.21735116027961213,
+      "learning_rate": 3.0838091114159895e-06,
+      "loss": 0.1149,
+      "step": 1036
+    },
+    {
+      "epoch": 4.198380566801619,
+      "grad_norm": 0.21680036547466372,
+      "learning_rate": 3.0537059008444524e-06,
+      "loss": 0.1218,
+      "step": 1037
+    },
+    {
+      "epoch": 4.202429149797571,
+      "grad_norm": 0.21780925170646043,
+      "learning_rate": 3.0237381923889563e-06,
+      "loss": 0.1419,
+      "step": 1038
+    },
+    {
+      "epoch": 4.206477732793522,
+      "grad_norm": 0.2257694997717373,
+      "learning_rate": 2.9939062256704997e-06,
+      "loss": 0.1293,
+      "step": 1039
+    },
+    {
+      "epoch": 4.2105263157894735,
+      "grad_norm": 0.211410055293199,
+      "learning_rate": 2.964210239224696e-06,
+      "loss": 0.1286,
+      "step": 1040
+    },
+    {
+      "epoch": 4.2145748987854255,
+      "grad_norm": 0.210980896069176,
+      "learning_rate": 2.934650470499869e-06,
+      "loss": 0.1359,
+      "step": 1041
+    },
+    {
+      "epoch": 4.218623481781377,
+      "grad_norm": 0.209080833995568,
+      "learning_rate": 2.905227155855128e-06,
+      "loss": 0.1341,
+      "step": 1042
+    },
+    {
+      "epoch": 4.222672064777328,
+      "grad_norm": 0.2179851731397519,
+      "learning_rate": 2.875940530558519e-06,
+      "loss": 0.1366,
+      "step": 1043
+    },
+    {
+      "epoch": 4.22672064777328,
+      "grad_norm": 0.21607680112533698,
+      "learning_rate": 2.8467908287851154e-06,
+      "loss": 0.1232,
+      "step": 1044
+    },
+    {
+      "epoch": 4.230769230769231,
+      "grad_norm": 0.21902848474732228,
+      "learning_rate": 2.8177782836151603e-06,
+      "loss": 0.1431,
+      "step": 1045
+    },
+    {
+      "epoch": 4.234817813765182,
+      "grad_norm": 0.21197688104432177,
+      "learning_rate": 2.7889031270321874e-06,
+      "loss": 0.1233,
+      "step": 1046
+    },
+    {
+      "epoch": 4.238866396761134,
+      "grad_norm": 0.21223587631184485,
+      "learning_rate": 2.760165589921184e-06,
+      "loss": 0.1285,
+      "step": 1047
+    },
+    {
+      "epoch": 4.242914979757085,
+      "grad_norm": 0.21746310118232884,
+      "learning_rate": 2.7315659020667305e-06,
+      "loss": 0.1347,
+      "step": 1048
+    },
+    {
+      "epoch": 4.246963562753036,
+      "grad_norm": 0.21413278770250715,
+      "learning_rate": 2.703104292151164e-06,
+      "loss": 0.1159,
+      "step": 1049
+    },
+    {
+      "epoch": 4.251012145748988,
+      "grad_norm": 0.21636351672466098,
+      "learning_rate": 2.6747809877527607e-06,
+      "loss": 0.1279,
+      "step": 1050
+    },
+    {
+      "epoch": 4.255060728744939,
+      "grad_norm": 0.22822176430500332,
+      "learning_rate": 2.6465962153439082e-06,
+      "loss": 0.1372,
+      "step": 1051
+    },
+    {
+      "epoch": 4.2591093117408905,
+      "grad_norm": 0.23764759464779442,
+      "learning_rate": 2.618550200289303e-06,
+      "loss": 0.1488,
+      "step": 1052
+    },
+    {
+      "epoch": 4.2631578947368425,
+      "grad_norm": 0.21915574745473126,
+      "learning_rate": 2.59064316684412e-06,
+      "loss": 0.1447,
+      "step": 1053
+    },
+    {
+      "epoch": 4.267206477732794,
+      "grad_norm": 0.2144372254920631,
+      "learning_rate": 2.562875338152264e-06,
+      "loss": 0.1328,
+      "step": 1054
+    },
+    {
+      "epoch": 4.271255060728745,
+      "grad_norm": 0.20531897174455874,
+      "learning_rate": 2.5352469362445552e-06,
+      "loss": 0.1422,
+      "step": 1055
+    },
+    {
+      "epoch": 4.275303643724697,
+      "grad_norm": 0.20623352890086377,
+      "learning_rate": 2.5077581820369523e-06,
+      "loss": 0.1494,
+      "step": 1056
+    },
+    {
+      "epoch": 4.279352226720648,
+      "grad_norm": 0.20347801114393138,
+      "learning_rate": 2.480409295328803e-06,
+      "loss": 0.136,
+      "step": 1057
+    },
+    {
+      "epoch": 4.283400809716599,
+      "grad_norm": 0.20456495581175885,
+      "learning_rate": 2.45320049480108e-06,
+      "loss": 0.1494,
+      "step": 1058
+    },
+    {
+      "epoch": 4.287449392712551,
+      "grad_norm": 0.20414983960310243,
+      "learning_rate": 2.4261319980146293e-06,
+      "loss": 0.1379,
+      "step": 1059
+    },
+    {
+      "epoch": 4.291497975708502,
+      "grad_norm": 0.22231872713291664,
+      "learning_rate": 2.399204021408421e-06,
+      "loss": 0.1322,
+      "step": 1060
+    },
+    {
+      "epoch": 4.295546558704453,
+      "grad_norm": 0.20864690596643096,
+      "learning_rate": 2.3724167802978395e-06,
+      "loss": 0.1151,
+      "step": 1061
+    },
+    {
+      "epoch": 4.299595141700405,
+      "grad_norm": 0.2158010751630423,
+      "learning_rate": 2.3457704888729625e-06,
+      "loss": 0.1414,
+      "step": 1062
+    },
+    {
+      "epoch": 4.303643724696356,
+      "grad_norm": 0.21860521182304482,
+      "learning_rate": 2.3192653601968162e-06,
+      "loss": 0.1277,
+      "step": 1063
+    },
+    {
+      "epoch": 4.3076923076923075,
+      "grad_norm": 0.2193088673027459,
+      "learning_rate": 2.2929016062037057e-06,
+      "loss": 0.1104,
+      "step": 1064
+    },
+    {
+      "epoch": 4.3117408906882595,
+      "grad_norm": 0.22071697486922578,
+      "learning_rate": 2.2666794376975055e-06,
+      "loss": 0.1408,
+      "step": 1065
+    },
+    {
+      "epoch": 4.315789473684211,
+      "grad_norm": 0.20758115284985845,
+      "learning_rate": 2.24059906434998e-06,
+      "loss": 0.1433,
+      "step": 1066
+    },
+    {
+      "epoch": 4.319838056680162,
+      "grad_norm": 0.21780776969047014,
+      "learning_rate": 2.214660694699091e-06,
+      "loss": 0.1322,
+      "step": 1067
+    },
+    {
+      "epoch": 4.323886639676114,
+      "grad_norm": 0.20663407569726247,
+      "learning_rate": 2.188864536147355e-06,
+      "loss": 0.1285,
+      "step": 1068
+    },
+    {
+      "epoch": 4.327935222672065,
+      "grad_norm": 0.21488646894100424,
+      "learning_rate": 2.1632107949601757e-06,
+      "loss": 0.1094,
+      "step": 1069
+    },
+    {
+      "epoch": 4.331983805668016,
+      "grad_norm": 0.21212707994719002,
+      "learning_rate": 2.137699676264171e-06,
+      "loss": 0.1264,
+      "step": 1070
+    },
+    {
+      "epoch": 4.336032388663968,
+      "grad_norm": 0.21070234565541585,
+      "learning_rate": 2.1123313840455738e-06,
+      "loss": 0.1222,
+      "step": 1071
+    },
+    {
+      "epoch": 4.340080971659919,
+      "grad_norm": 0.2064034489154814,
+      "learning_rate": 2.08710612114857e-06,
+      "loss": 0.119,
+      "step": 1072
+    },
+    {
+      "epoch": 4.34412955465587,
+      "grad_norm": 0.2080555701835878,
+      "learning_rate": 2.0620240892736953e-06,
+      "loss": 0.1268,
+      "step": 1073
+    },
+    {
+      "epoch": 4.348178137651822,
+      "grad_norm": 0.2112015059745621,
+      "learning_rate": 2.0370854889762e-06,
+      "loss": 0.1488,
+      "step": 1074
+    },
+    {
+      "epoch": 4.352226720647773,
+      "grad_norm": 0.20529006454510543,
+      "learning_rate": 2.0122905196644726e-06,
+      "loss": 0.1385,
+      "step": 1075
+    },
+    {
+      "epoch": 4.3562753036437245,
+      "grad_norm": 0.20384118068367812,
+      "learning_rate": 1.987639379598427e-06,
+      "loss": 0.1133,
+      "step": 1076
+    },
+    {
+      "epoch": 4.3603238866396765,
+      "grad_norm": 0.202036716714986,
+      "learning_rate": 1.963132265887919e-06,
+      "loss": 0.1318,
+      "step": 1077
+    },
+    {
+      "epoch": 4.364372469635628,
+      "grad_norm": 0.2234119098621468,
+      "learning_rate": 1.9387693744911784e-06,
+      "loss": 0.1468,
+      "step": 1078
+    },
+    {
+      "epoch": 4.368421052631579,
+      "grad_norm": 0.20992659455988374,
+      "learning_rate": 1.9145509002132368e-06,
+      "loss": 0.1213,
+      "step": 1079
+    },
+    {
+      "epoch": 4.372469635627531,
+      "grad_norm": 0.21010215679052544,
+      "learning_rate": 1.8904770367043723e-06,
+      "loss": 0.1384,
+      "step": 1080
+    },
+    {
+      "epoch": 4.376518218623482,
+      "grad_norm": 0.23362051898167863,
+      "learning_rate": 1.8665479764585504e-06,
+      "loss": 0.1225,
+      "step": 1081
+    },
+    {
+      "epoch": 4.380566801619433,
+      "grad_norm": 0.2231263025886564,
+      "learning_rate": 1.8427639108118977e-06,
+      "loss": 0.1534,
+      "step": 1082
+    },
+    {
+      "epoch": 4.384615384615385,
+      "grad_norm": 0.21043598421232865,
+      "learning_rate": 1.8191250299411755e-06,
+      "loss": 0.1364,
+      "step": 1083
+    },
+    {
+      "epoch": 4.388663967611336,
+      "grad_norm": 0.2015769256202967,
+      "learning_rate": 1.7956315228622446e-06,
+      "loss": 0.1379,
+      "step": 1084
+    },
+    {
+      "epoch": 4.392712550607287,
+      "grad_norm": 0.20954924177570644,
+      "learning_rate": 1.7722835774285553e-06,
+      "loss": 0.1276,
+      "step": 1085
+    },
+    {
+      "epoch": 4.396761133603239,
+      "grad_norm": 0.20514486429837978,
+      "learning_rate": 1.7490813803296624e-06,
+      "loss": 0.1418,
+      "step": 1086
+    },
+    {
+      "epoch": 4.40080971659919,
+      "grad_norm": 0.21510782117325444,
+      "learning_rate": 1.726025117089718e-06,
+      "loss": 0.1204,
+      "step": 1087
+    },
+    {
+      "epoch": 4.4048582995951415,
+      "grad_norm": 0.2089657588771132,
+      "learning_rate": 1.70311497206598e-06,
+      "loss": 0.1085,
+      "step": 1088
+    },
+    {
+      "epoch": 4.4089068825910935,
+      "grad_norm": 0.2249639847799371,
+      "learning_rate": 1.6803511284473638e-06,
+      "loss": 0.1553,
+      "step": 1089
+    },
+    {
+      "epoch": 4.412955465587045,
+      "grad_norm": 0.2120254888677494,
+      "learning_rate": 1.6577337682529538e-06,
+      "loss": 0.1288,
+      "step": 1090
+    },
+    {
+      "epoch": 4.417004048582996,
+      "grad_norm": 0.20265291837332483,
+      "learning_rate": 1.6352630723305617e-06,
+      "loss": 0.1347,
+      "step": 1091
+    },
+    {
+      "epoch": 4.421052631578947,
+      "grad_norm": 0.2078881888560106,
+      "learning_rate": 1.612939220355272e-06,
+      "loss": 0.1133,
+      "step": 1092
+    },
+    {
+      "epoch": 4.425101214574899,
+      "grad_norm": 0.21391882575314541,
+      "learning_rate": 1.5907623908280112e-06,
+      "loss": 0.111,
+      "step": 1093
+    },
+    {
+      "epoch": 4.42914979757085,
+      "grad_norm": 0.2120018428807029,
+      "learning_rate": 1.568732761074121e-06,
+      "loss": 0.1491,
+      "step": 1094
+    },
+    {
+      "epoch": 4.433198380566802,
+      "grad_norm": 0.21323982167659078,
+      "learning_rate": 1.5468505072419237e-06,
+      "loss": 0.1259,
+      "step": 1095
+    },
+    {
+      "epoch": 4.437246963562753,
+      "grad_norm": 0.21419416838371216,
+      "learning_rate": 1.5251158043013448e-06,
+      "loss": 0.1257,
+      "step": 1096
+    },
+    {
+      "epoch": 4.441295546558704,
+      "grad_norm": 0.22041711077045845,
+      "learning_rate": 1.5035288260424885e-06,
+      "loss": 0.127,
+      "step": 1097
+    },
+    {
+      "epoch": 4.445344129554655,
+      "grad_norm": 0.20985964507013768,
+      "learning_rate": 1.4820897450742623e-06,
+      "loss": 0.1401,
+      "step": 1098
+    },
+    {
+      "epoch": 4.449392712550607,
+      "grad_norm": 0.21135170174551432,
+      "learning_rate": 1.4607987328229767e-06,
+      "loss": 0.1158,
+      "step": 1099
+    },
+    {
+      "epoch": 4.4534412955465585,
+      "grad_norm": 0.21204588823406892,
+      "learning_rate": 1.4396559595310056e-06,
+      "loss": 0.1375,
+      "step": 1100
+    },
+    {
+      "epoch": 4.4574898785425106,
+      "grad_norm": 0.2110381644554301,
+      "learning_rate": 1.4186615942553973e-06,
+      "loss": 0.1398,
+      "step": 1101
+    },
+    {
+      "epoch": 4.461538461538462,
+      "grad_norm": 0.2088775098171412,
+      "learning_rate": 1.3978158048665424e-06,
+      "loss": 0.1226,
+      "step": 1102
+    },
+    {
+      "epoch": 4.465587044534413,
+      "grad_norm": 0.20886442704100364,
+      "learning_rate": 1.3771187580468115e-06,
+      "loss": 0.1408,
+      "step": 1103
+    },
+    {
+      "epoch": 4.469635627530364,
+      "grad_norm": 0.20625999414160162,
+      "learning_rate": 1.3565706192892392e-06,
+      "loss": 0.1246,
+      "step": 1104
+    },
+    {
+      "epoch": 4.473684210526316,
+      "grad_norm": 0.2167329292389523,
+      "learning_rate": 1.3361715528961993e-06,
+      "loss": 0.139,
+      "step": 1105
+    },
+    {
+      "epoch": 4.477732793522267,
+      "grad_norm": 0.20103429312474616,
+      "learning_rate": 1.3159217219780817e-06,
+      "loss": 0.134,
+      "step": 1106
+    },
+    {
+      "epoch": 4.481781376518219,
+      "grad_norm": 0.21266081028807576,
+      "learning_rate": 1.2958212884519949e-06,
+      "loss": 0.1393,
+      "step": 1107
+    },
+    {
+      "epoch": 4.48582995951417,
+      "grad_norm": 0.21033097876363907,
+      "learning_rate": 1.2758704130404675e-06,
+      "loss": 0.1179,
+      "step": 1108
+    },
+    {
+      "epoch": 4.489878542510121,
+      "grad_norm": 0.21386898763707304,
+      "learning_rate": 1.2560692552701738e-06,
+      "loss": 0.1238,
+      "step": 1109
+    },
+    {
+      "epoch": 4.493927125506072,
+      "grad_norm": 0.216279144416779,
+      "learning_rate": 1.2364179734706382e-06,
+      "loss": 0.1339,
+      "step": 1110
+    },
+    {
+      "epoch": 4.497975708502024,
+      "grad_norm": 0.21413660988941627,
+      "learning_rate": 1.2169167247729852e-06,
+      "loss": 0.1335,
+      "step": 1111
+    },
+    {
+      "epoch": 4.502024291497976,
+      "grad_norm": 0.2078037515660601,
+      "learning_rate": 1.1975656651086864e-06,
+      "loss": 0.1268,
+      "step": 1112
+    },
+    {
+      "epoch": 4.506072874493928,
+      "grad_norm": 0.21212891372726717,
+      "learning_rate": 1.1783649492082926e-06,
+      "loss": 0.1414,
+      "step": 1113
+    },
+    {
+      "epoch": 4.510121457489879,
+      "grad_norm": 0.1994182023112096,
+      "learning_rate": 1.1593147306002183e-06,
+      "loss": 0.1483,
+      "step": 1114
+    },
+    {
+      "epoch": 4.51417004048583,
+      "grad_norm": 0.21625964314730856,
+      "learning_rate": 1.140415161609505e-06,
+      "loss": 0.1457,
+      "step": 1115
+    },
+    {
+      "epoch": 4.518218623481781,
+      "grad_norm": 0.20923150541262578,
+      "learning_rate": 1.121666393356604e-06,
+      "loss": 0.1391,
+      "step": 1116
+    },
+    {
+      "epoch": 4.522267206477733,
+      "grad_norm": 0.2126859596690903,
+      "learning_rate": 1.1030685757561632e-06,
+      "loss": 0.1463,
+      "step": 1117
+    },
+    {
+      "epoch": 4.526315789473684,
+      "grad_norm": 0.2063691920972948,
+      "learning_rate": 1.0846218575158373e-06,
+      "loss": 0.1403,
+      "step": 1118
+    },
+    {
+      "epoch": 4.530364372469636,
+      "grad_norm": 0.21146980494536796,
+      "learning_rate": 1.066326386135097e-06,
+      "loss": 0.137,
+      "step": 1119
+    },
+    {
+      "epoch": 4.534412955465587,
+      "grad_norm": 0.21715881273879653,
+      "learning_rate": 1.0481823079040398e-06,
+      "loss": 0.1342,
+      "step": 1120
+    },
+    {
+      "epoch": 4.538461538461538,
+      "grad_norm": 0.21248186737318994,
+      "learning_rate": 1.0301897679022344e-06,
+      "loss": 0.128,
+      "step": 1121
+    },
+    {
+      "epoch": 4.5425101214574894,
+      "grad_norm": 0.20799865884645974,
+      "learning_rate": 1.0123489099975491e-06,
+      "loss": 0.1321,
+      "step": 1122
+    },
+    {
+      "epoch": 4.5465587044534415,
+      "grad_norm": 0.20645310399754974,
+      "learning_rate": 9.9465987684501e-07,
+      "loss": 0.1405,
+      "step": 1123
+    },
+    {
+      "epoch": 4.550607287449393,
+      "grad_norm": 0.21036557377115628,
+      "learning_rate": 9.771228098856534e-07,
+      "loss": 0.1398,
+      "step": 1124
+    },
+    {
+      "epoch": 4.554655870445345,
+      "grad_norm": 0.20937384500709283,
+      "learning_rate": 9.597378493454013e-07,
+      "loss": 0.1501,
+      "step": 1125
+    },
+    {
+      "epoch": 4.558704453441296,
+      "grad_norm": 0.20935273333021634,
+      "learning_rate": 9.425051342339353e-07,
+      "loss": 0.1429,
+      "step": 1126
+    },
+    {
+      "epoch": 4.562753036437247,
+      "grad_norm": 0.1959518734979928,
+      "learning_rate": 9.254248023435864e-07,
+      "loss": 0.1343,
+      "step": 1127
+    },
+    {
+      "epoch": 4.566801619433198,
+      "grad_norm": 0.20524543111967986,
+      "learning_rate": 9.084969902482288e-07,
+      "loss": 0.1319,
+      "step": 1128
+    },
+    {
+      "epoch": 4.57085020242915,
+      "grad_norm": 0.2025128899227959,
+      "learning_rate": 8.917218333021993e-07,
+      "loss": 0.1396,
+      "step": 1129
+    },
+    {
+      "epoch": 4.574898785425101,
+      "grad_norm": 0.21693070133625916,
+      "learning_rate": 8.750994656392109e-07,
+      "loss": 0.1308,
+      "step": 1130
+    },
+    {
+      "epoch": 4.578947368421053,
+      "grad_norm": 0.2114720756756507,
+      "learning_rate": 8.58630020171265e-07,
+      "loss": 0.1293,
+      "step": 1131
+    },
+    {
+      "epoch": 4.582995951417004,
+      "grad_norm": 0.21306978375820915,
+      "learning_rate": 8.423136285876099e-07,
+      "loss": 0.1385,
+      "step": 1132
+    },
+    {
+      "epoch": 4.587044534412955,
+      "grad_norm": 0.21796278920188467,
+      "learning_rate": 8.26150421353682e-07,
+      "loss": 0.1116,
+      "step": 1133
+    },
+    {
+      "epoch": 4.5910931174089065,
+      "grad_norm": 0.20666409241106176,
+      "learning_rate": 8.101405277100549e-07,
+      "loss": 0.1344,
+      "step": 1134
+    },
+    {
+      "epoch": 4.5951417004048585,
+      "grad_norm": 0.2148847324931492,
+      "learning_rate": 7.942840756714077e-07,
+      "loss": 0.1351,
+      "step": 1135
+    },
+    {
+      "epoch": 4.59919028340081,
+      "grad_norm": 0.2116348192260234,
+      "learning_rate": 7.785811920255093e-07,
+      "loss": 0.1153,
+      "step": 1136
+    },
+    {
+      "epoch": 4.603238866396762,
+      "grad_norm": 0.21231840664100013,
+      "learning_rate": 7.630320023322024e-07,
+      "loss": 0.136,
+      "step": 1137
+    },
+    {
+      "epoch": 4.607287449392713,
+      "grad_norm": 0.21032983452562912,
+      "learning_rate": 7.476366309223903e-07,
+      "loss": 0.124,
+      "step": 1138
+    },
+    {
+      "epoch": 4.611336032388664,
+      "grad_norm": 0.21197857986569255,
+      "learning_rate": 7.323952008970537e-07,
+      "loss": 0.1354,
+      "step": 1139
+    },
+    {
+      "epoch": 4.615384615384615,
+      "grad_norm": 0.2297243443661598,
+      "learning_rate": 7.173078341262618e-07,
+      "loss": 0.1191,
+      "step": 1140
+    },
+    {
+      "epoch": 4.619433198380567,
+      "grad_norm": 0.2047338174280469,
+      "learning_rate": 7.023746512481989e-07,
+      "loss": 0.1269,
+      "step": 1141
+    },
+    {
+      "epoch": 4.623481781376518,
+      "grad_norm": 0.20462364900196134,
+      "learning_rate": 6.875957716681902e-07,
+      "loss": 0.1555,
+      "step": 1142
+    },
+    {
+      "epoch": 4.62753036437247,
+      "grad_norm": 0.21222119909800843,
+      "learning_rate": 6.729713135577687e-07,
+      "loss": 0.1391,
+      "step": 1143
+    },
+    {
+      "epoch": 4.631578947368421,
+      "grad_norm": 0.21122480617953987,
+      "learning_rate": 6.585013938537143e-07,
+      "loss": 0.1237,
+      "step": 1144
+    },
+    {
+      "epoch": 4.635627530364372,
+      "grad_norm": 0.21003095418705386,
+      "learning_rate": 6.441861282571115e-07,
+      "loss": 0.143,
+      "step": 1145
+    },
+    {
+      "epoch": 4.6396761133603235,
+      "grad_norm": 0.21736101941956215,
+      "learning_rate": 6.300256312324471e-07,
+      "loss": 0.1362,
+      "step": 1146
+    },
+    {
+      "epoch": 4.6437246963562755,
+      "grad_norm": 0.21782561836626604,
+      "learning_rate": 6.160200160066776e-07,
+      "loss": 0.1315,
+      "step": 1147
+    },
+    {
+      "epoch": 4.647773279352227,
+      "grad_norm": 0.19543437051825474,
+      "learning_rate": 6.021693945683327e-07,
+      "loss": 0.1352,
+      "step": 1148
+    },
+    {
+      "epoch": 4.651821862348179,
+      "grad_norm": 0.21359919955358853,
+      "learning_rate": 5.884738776666044e-07,
+      "loss": 0.1498,
+      "step": 1149
+    },
+    {
+      "epoch": 4.65587044534413,
+      "grad_norm": 0.20384763112178264,
+      "learning_rate": 5.749335748104878e-07,
+      "loss": 0.1446,
+      "step": 1150
+    },
+    {
+      "epoch": 4.659919028340081,
+      "grad_norm": 0.20904552075586394,
+      "learning_rate": 5.615485942678778e-07,
+      "loss": 0.134,
+      "step": 1151
+    },
+    {
+      "epoch": 4.663967611336032,
+      "grad_norm": 0.20522899378328852,
+      "learning_rate": 5.483190430647289e-07,
+      "loss": 0.1472,
+      "step": 1152
+    },
+    {
+      "epoch": 4.668016194331984,
+      "grad_norm": 0.207665200591191,
+      "learning_rate": 5.352450269841747e-07,
+      "loss": 0.1527,
+      "step": 1153
+    },
+    {
+      "epoch": 4.672064777327935,
+      "grad_norm": 0.2064685879071364,
+      "learning_rate": 5.22326650565701e-07,
+      "loss": 0.1389,
+      "step": 1154
+    },
+    {
+      "epoch": 4.676113360323887,
+      "grad_norm": 0.2072814003650562,
+      "learning_rate": 5.095640171043048e-07,
+      "loss": 0.1401,
+      "step": 1155
+    },
+    {
+      "epoch": 4.680161943319838,
+      "grad_norm": 0.21515192994061916,
+      "learning_rate": 4.969572286496594e-07,
+      "loss": 0.1314,
+      "step": 1156
+    },
+    {
+      "epoch": 4.684210526315789,
+      "grad_norm": 0.20671410856220865,
+      "learning_rate": 4.845063860053079e-07,
+      "loss": 0.1278,
+      "step": 1157
+    },
+    {
+      "epoch": 4.6882591093117405,
+      "grad_norm": 0.2078991135134485,
+      "learning_rate": 4.7221158872785997e-07,
+      "loss": 0.1382,
+      "step": 1158
+    },
+    {
+      "epoch": 4.6923076923076925,
+      "grad_norm": 0.22002983244713123,
+      "learning_rate": 4.600729351261857e-07,
+      "loss": 0.133,
+      "step": 1159
+    },
+    {
+      "epoch": 4.696356275303644,
+      "grad_norm": 0.21129534461488791,
+      "learning_rate": 4.4809052226063355e-07,
+      "loss": 0.1279,
+      "step": 1160
+    },
+    {
+      "epoch": 4.700404858299595,
+      "grad_norm": 0.20775209280532372,
+      "learning_rate": 4.3626444594226046e-07,
+      "loss": 0.1117,
+      "step": 1161
+    },
+    {
+      "epoch": 4.704453441295547,
+      "grad_norm": 0.2126351897289371,
+      "learning_rate": 4.245948007320588e-07,
+      "loss": 0.1344,
+      "step": 1162
+    },
+    {
+      "epoch": 4.708502024291498,
+      "grad_norm": 0.20998789263784903,
+      "learning_rate": 4.130816799401971e-07,
+      "loss": 0.1221,
+      "step": 1163
+    },
+    {
+      "epoch": 4.712550607287449,
+      "grad_norm": 0.20890041886351562,
+      "learning_rate": 4.017251756252827e-07,
+      "loss": 0.1316,
+      "step": 1164
+    },
+    {
+      "epoch": 4.716599190283401,
+      "grad_norm": 0.20342894391654126,
+      "learning_rate": 3.9052537859362473e-07,
+      "loss": 0.1277,
+      "step": 1165
+    },
+    {
+      "epoch": 4.720647773279352,
+      "grad_norm": 0.20476530853956593,
+      "learning_rate": 3.7948237839849687e-07,
+      "loss": 0.1367,
+      "step": 1166
+    },
+    {
+      "epoch": 4.724696356275303,
+      "grad_norm": 0.2043069981282894,
+      "learning_rate": 3.6859626333943797e-07,
+      "loss": 0.142,
+      "step": 1167
+    },
+    {
+      "epoch": 4.728744939271255,
+      "grad_norm": 0.20426832004312873,
+      "learning_rate": 3.578671204615347e-07,
+      "loss": 0.1251,
+      "step": 1168
+    },
+    {
+      "epoch": 4.732793522267206,
+      "grad_norm": 0.22893397095651788,
+      "learning_rate": 3.472950355547267e-07,
+      "loss": 0.114,
+      "step": 1169
+    },
+    {
+      "epoch": 4.7368421052631575,
+      "grad_norm": 0.219878391232362,
+      "learning_rate": 3.3688009315312244e-07,
+      "loss": 0.13,
+      "step": 1170
+    },
+    {
+      "epoch": 4.7408906882591095,
+      "grad_norm": 0.21278916222036665,
+      "learning_rate": 3.266223765343224e-07,
+      "loss": 0.1234,
+      "step": 1171
+    },
+    {
+      "epoch": 4.744939271255061,
+      "grad_norm": 0.2036994274463091,
+      "learning_rate": 3.16521967718757e-07,
+      "loss": 0.1405,
+      "step": 1172
+    },
+    {
+      "epoch": 4.748987854251012,
+      "grad_norm": 0.2139494291175298,
+      "learning_rate": 3.0657894746902726e-07,
+      "loss": 0.1183,
+      "step": 1173
+    },
+    {
+      "epoch": 4.753036437246964,
+      "grad_norm": 0.20838619422712262,
+      "learning_rate": 2.9679339528925654e-07,
+      "loss": 0.1285,
+      "step": 1174
+    },
+    {
+      "epoch": 4.757085020242915,
+      "grad_norm": 0.20016956258522542,
+      "learning_rate": 2.871653894244597e-07,
+      "loss": 0.1489,
+      "step": 1175
+    },
+    {
+      "epoch": 4.761133603238866,
+      "grad_norm": 0.2146900433570998,
+      "learning_rate": 2.7769500685991266e-07,
+      "loss": 0.1227,
+      "step": 1176
+    },
+    {
+      "epoch": 4.765182186234818,
+      "grad_norm": 0.19784504899304092,
+      "learning_rate": 2.683823233205485e-07,
+      "loss": 0.1324,
+      "step": 1177
+    },
+    {
+      "epoch": 4.769230769230769,
+      "grad_norm": 0.21238311572526775,
+      "learning_rate": 2.592274132703287e-07,
+      "loss": 0.116,
+      "step": 1178
+    },
+    {
+      "epoch": 4.77327935222672,
+      "grad_norm": 0.2098890292083296,
+      "learning_rate": 2.502303499116754e-07,
+      "loss": 0.1358,
+      "step": 1179
+    },
+    {
+      "epoch": 4.777327935222672,
+      "grad_norm": 0.21008165069041132,
+      "learning_rate": 2.413912051848666e-07,
+      "loss": 0.1316,
+      "step": 1180
+    },
+    {
+      "epoch": 4.781376518218623,
+      "grad_norm": 0.20897815422241178,
+      "learning_rate": 2.3271004976746613e-07,
+      "loss": 0.1346,
+      "step": 1181
+    },
+    {
+      "epoch": 4.7854251012145745,
+      "grad_norm": 0.21305881903975965,
+      "learning_rate": 2.2418695307376393e-07,
+      "loss": 0.1128,
+      "step": 1182
+    },
+    {
+      "epoch": 4.7894736842105265,
+      "grad_norm": 0.2129608211228651,
+      "learning_rate": 2.1582198325420968e-07,
+      "loss": 0.1365,
+      "step": 1183
+    },
+    {
+      "epoch": 4.793522267206478,
+      "grad_norm": 0.2053332527248821,
+      "learning_rate": 2.0761520719488447e-07,
+      "loss": 0.1204,
+      "step": 1184
+    },
+    {
+      "epoch": 4.797570850202429,
+      "grad_norm": 0.20534205648055648,
+      "learning_rate": 1.9956669051694798e-07,
+      "loss": 0.1391,
+      "step": 1185
+    },
+    {
+      "epoch": 4.801619433198381,
+      "grad_norm": 0.20662278423511543,
+      "learning_rate": 1.916764975761254e-07,
+      "loss": 0.1343,
+      "step": 1186
+    },
+    {
+      "epoch": 4.805668016194332,
+      "grad_norm": 0.20921693799221855,
+      "learning_rate": 1.839446914621923e-07,
+      "loss": 0.1376,
+      "step": 1187
+    },
+    {
+      "epoch": 4.809716599190283,
+      "grad_norm": 0.20840803592716256,
+      "learning_rate": 1.7637133399845962e-07,
+      "loss": 0.1482,
+      "step": 1188
+    },
+    {
+      "epoch": 4.813765182186235,
+      "grad_norm": 0.20740036666229184,
+      "learning_rate": 1.6895648574129397e-07,
+      "loss": 0.1343,
+      "step": 1189
+    },
+    {
+      "epoch": 4.817813765182186,
+      "grad_norm": 0.20575303371944845,
+      "learning_rate": 1.6170020597962465e-07,
+      "loss": 0.1334,
+      "step": 1190
+    },
+    {
+      "epoch": 4.821862348178137,
+      "grad_norm": 0.20249700671348173,
+      "learning_rate": 1.5460255273447078e-07,
+      "loss": 0.1459,
+      "step": 1191
+    },
+    {
+      "epoch": 4.825910931174089,
+      "grad_norm": 0.20766572960248242,
+      "learning_rate": 1.4766358275847715e-07,
+      "loss": 0.1319,
+      "step": 1192
+    },
+    {
+      "epoch": 4.82995951417004,
+      "grad_norm": 0.2129765294329246,
+      "learning_rate": 1.4088335153546573e-07,
+      "loss": 0.1223,
+      "step": 1193
+    },
+    {
+      "epoch": 4.834008097165992,
+      "grad_norm": 0.20726044610278632,
+      "learning_rate": 1.3426191327998496e-07,
+      "loss": 0.1429,
+      "step": 1194
+    },
+    {
+      "epoch": 4.838056680161944,
+      "grad_norm": 0.20087638057557472,
+      "learning_rate": 1.277993209368744e-07,
+      "loss": 0.1295,
+      "step": 1195
+    },
+    {
+      "epoch": 4.842105263157895,
+      "grad_norm": 0.2053364062821857,
+      "learning_rate": 1.214956261808542e-07,
+      "loss": 0.1417,
+      "step": 1196
+    },
+    {
+      "epoch": 4.846153846153846,
+      "grad_norm": 0.20669512653476094,
+      "learning_rate": 1.1535087941609846e-07,
+      "loss": 0.1076,
+      "step": 1197
+    },
+    {
+      "epoch": 4.850202429149798,
+      "grad_norm": 0.2000462147719165,
+      "learning_rate": 1.0936512977583802e-07,
+      "loss": 0.145,
+      "step": 1198
+    },
+    {
+      "epoch": 4.854251012145749,
+      "grad_norm": 0.20733552677840925,
+      "learning_rate": 1.0353842512196732e-07,
+      "loss": 0.1169,
+      "step": 1199
+    },
+    {
+      "epoch": 4.8582995951417,
+      "grad_norm": 0.20924282835585767,
+      "learning_rate": 9.787081204466253e-08,
+      "loss": 0.1117,
+      "step": 1200
+    },
+    {
+      "epoch": 4.862348178137652,
+      "grad_norm": 0.20597993429398134,
+      "learning_rate": 9.236233586200627e-08,
+      "loss": 0.1356,
+      "step": 1201
+    },
+    {
+      "epoch": 4.866396761133603,
+      "grad_norm": 0.20083834973249262,
+      "learning_rate": 8.70130406196279e-08,
+      "loss": 0.1412,
+      "step": 1202
+    },
+    {
+      "epoch": 4.870445344129554,
+      "grad_norm": 0.20760335250816522,
+      "learning_rate": 8.182296909035047e-08,
+      "loss": 0.13,
+      "step": 1203
+    },
+    {
+      "epoch": 4.874493927125506,
+      "grad_norm": 0.2082953385766791,
+      "learning_rate": 7.679216277384439e-08,
+      "loss": 0.1379,
+      "step": 1204
+    },
+    {
+      "epoch": 4.8785425101214575,
+      "grad_norm": 0.2071728201936558,
+      "learning_rate": 7.19206618963053e-08,
+      "loss": 0.1234,
+      "step": 1205
+    },
+    {
+      "epoch": 4.882591093117409,
+      "grad_norm": 0.20522223355983968,
+      "learning_rate": 6.720850541012347e-08,
+      "loss": 0.1279,
+      "step": 1206
+    },
+    {
+      "epoch": 4.886639676113361,
+      "grad_norm": 0.21094262626937457,
+      "learning_rate": 6.265573099357714e-08,
+      "loss": 0.1406,
+      "step": 1207
+    },
+    {
+      "epoch": 4.890688259109312,
+      "grad_norm": 0.20072809351036994,
+      "learning_rate": 5.826237505053067e-08,
+      "loss": 0.141,
+      "step": 1208
+    },
+    {
+      "epoch": 4.894736842105263,
+      "grad_norm": 0.20057366296103865,
+      "learning_rate": 5.4028472710137e-08,
+      "loss": 0.1441,
+      "step": 1209
+    },
+    {
+      "epoch": 4.898785425101215,
+      "grad_norm": 0.20024474580876273,
+      "learning_rate": 4.9954057826571146e-08,
+      "loss": 0.1275,
+      "step": 1210
+    },
+    {
+      "epoch": 4.902834008097166,
+      "grad_norm": 0.20981461421043124,
+      "learning_rate": 4.6039162978743776e-08,
+      "loss": 0.1332,
+      "step": 1211
+    },
+    {
+      "epoch": 4.906882591093117,
+      "grad_norm": 0.20651990800591916,
+      "learning_rate": 4.228381947004812e-08,
+      "loss": 0.1305,
+      "step": 1212
+    },
+    {
+      "epoch": 4.910931174089069,
+      "grad_norm": 0.21338502625637626,
+      "learning_rate": 3.868805732811343e-08,
+      "loss": 0.1245,
+      "step": 1213
+    },
+    {
+      "epoch": 4.91497975708502,
+      "grad_norm": 0.19992338324364373,
+      "learning_rate": 3.525190530455636e-08,
+      "loss": 0.1287,
+      "step": 1214
+    },
+    {
+      "epoch": 4.919028340080971,
+      "grad_norm": 0.21092967065393262,
+      "learning_rate": 3.197539087475887e-08,
+      "loss": 0.1311,
+      "step": 1215
+    },
+    {
+      "epoch": 4.923076923076923,
+      "grad_norm": 0.2114997592276131,
+      "learning_rate": 2.8858540237641786e-08,
+      "loss": 0.1248,
+      "step": 1216
+    },
+    {
+      "epoch": 4.9271255060728745,
+      "grad_norm": 0.21437969165612109,
+      "learning_rate": 2.5901378315458247e-08,
+      "loss": 0.1213,
+      "step": 1217
+    },
+    {
+      "epoch": 4.931174089068826,
+      "grad_norm": 0.20721243276044354,
+      "learning_rate": 2.3103928753598347e-08,
+      "loss": 0.1556,
+      "step": 1218
+    },
+    {
+      "epoch": 4.935222672064778,
+      "grad_norm": 0.2042215084339711,
+      "learning_rate": 2.0466213920393718e-08,
+      "loss": 0.1303,
+      "step": 1219
+    },
+    {
+      "epoch": 4.939271255060729,
+      "grad_norm": 0.20454332253195753,
+      "learning_rate": 1.7988254906942115e-08,
+      "loss": 0.1179,
+      "step": 1220
+    },
+    {
+      "epoch": 4.94331983805668,
+      "grad_norm": 0.20967993764009474,
+      "learning_rate": 1.5670071526936447e-08,
+      "loss": 0.1376,
+      "step": 1221
+    },
+    {
+      "epoch": 4.947368421052632,
+      "grad_norm": 0.20629188496545584,
+      "learning_rate": 1.3511682316509344e-08,
+      "loss": 0.1389,
+      "step": 1222
+    },
+    {
+      "epoch": 4.951417004048583,
+      "grad_norm": 0.20686823534954574,
+      "learning_rate": 1.1513104534086605e-08,
+      "loss": 0.1344,
+      "step": 1223
+    },
+    {
+      "epoch": 4.955465587044534,
+      "grad_norm": 0.20376233307112013,
+      "learning_rate": 9.67435416023843e-09,
+      "loss": 0.1144,
+      "step": 1224
+    },
+    {
+      "epoch": 4.959514170040486,
+      "grad_norm": 0.20590506587591445,
+      "learning_rate": 7.995445897566178e-09,
+      "loss": 0.1247,
+      "step": 1225
+    },
+    {
+      "epoch": 4.963562753036437,
+      "grad_norm": 0.20810776605196601,
+      "learning_rate": 6.476393170578021e-09,
+      "loss": 0.1343,
+      "step": 1226
+    },
+    {
+      "epoch": 4.967611336032388,
+      "grad_norm": 0.19959506920729506,
+      "learning_rate": 5.11720812557348e-09,
+      "loss": 0.1324,
+      "step": 1227
+    },
+    {
+      "epoch": 4.97165991902834,
+      "grad_norm": 0.20351932711279971,
+      "learning_rate": 3.917901630563492e-09,
+      "loss": 0.1269,
+      "step": 1228
+    },
+    {
+      "epoch": 4.9757085020242915,
+      "grad_norm": 0.1971425023510774,
+      "learning_rate": 2.878483275170485e-09,
+      "loss": 0.1161,
+      "step": 1229
+    },
+    {
+      "epoch": 4.979757085020243,
+      "grad_norm": 0.20120923882875416,
+      "learning_rate": 1.998961370555108e-09,
+      "loss": 0.1316,
+      "step": 1230
+    },
+    {
+      "epoch": 4.983805668016195,
+      "grad_norm": 0.19861919395805938,
+      "learning_rate": 1.2793429493518361e-09,
+      "loss": 0.1256,
+      "step": 1231
+    },
+    {
+      "epoch": 4.987854251012146,
+      "grad_norm": 0.21413358463736268,
+      "learning_rate": 7.196337656112384e-10,
+      "loss": 0.1271,
+      "step": 1232
+    },
+    {
+      "epoch": 4.991902834008097,
+      "grad_norm": 0.203356285700396,
+      "learning_rate": 3.198382947511292e-10,
+      "loss": 0.1299,
+      "step": 1233
+    },
+    {
+      "epoch": 4.995951417004049,
+      "grad_norm": 0.2037275878915362,
+      "learning_rate": 7.995973352770137e-11,
+      "loss": 0.14,
+      "step": 1234
+    },
+    {
+      "epoch": 5.0,
+      "grad_norm": 0.2079503349524466,
+      "learning_rate": 0.0,
+      "loss": 0.1287,
+      "step": 1235
+    },
+    {
+      "epoch": 5.0,
+      "step": 1235,
+      "total_flos": 1.9749926564213555e+18,
+      "train_loss": 0.2607566904623499,
+      "train_runtime": 24011.273,
+      "train_samples_per_second": 6.58,
+      "train_steps_per_second": 0.051
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 1235,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 5,
+  "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": 1.9749926564213555e+18,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}