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": 6.59632565694887,
+      "learning_rate": 3.2258064516129035e-07,
+      "loss": 0.8483,
+      "step": 1
+    },
+    {
+      "epoch": 0.008097165991902834,
+      "grad_norm": 6.482285035307976,
+      "learning_rate": 6.451612903225807e-07,
+      "loss": 0.8284,
+      "step": 2
+    },
+    {
+      "epoch": 0.012145748987854251,
+      "grad_norm": 6.635223310570105,
+      "learning_rate": 9.67741935483871e-07,
+      "loss": 0.8493,
+      "step": 3
+    },
+    {
+      "epoch": 0.016194331983805668,
+      "grad_norm": 6.50667767532137,
+      "learning_rate": 1.2903225806451614e-06,
+      "loss": 0.846,
+      "step": 4
+    },
+    {
+      "epoch": 0.020242914979757085,
+      "grad_norm": 6.286944173716275,
+      "learning_rate": 1.6129032258064516e-06,
+      "loss": 0.8223,
+      "step": 5
+    },
+    {
+      "epoch": 0.024291497975708502,
+      "grad_norm": 5.982085856376136,
+      "learning_rate": 1.935483870967742e-06,
+      "loss": 0.8097,
+      "step": 6
+    },
+    {
+      "epoch": 0.02834008097165992,
+      "grad_norm": 4.710802633872772,
+      "learning_rate": 2.2580645161290324e-06,
+      "loss": 0.7437,
+      "step": 7
+    },
+    {
+      "epoch": 0.032388663967611336,
+      "grad_norm": 4.346605276136481,
+      "learning_rate": 2.580645161290323e-06,
+      "loss": 0.7444,
+      "step": 8
+    },
+    {
+      "epoch": 0.03643724696356275,
+      "grad_norm": 2.5053498132848353,
+      "learning_rate": 2.903225806451613e-06,
+      "loss": 0.7179,
+      "step": 9
+    },
+    {
+      "epoch": 0.04048582995951417,
+      "grad_norm": 2.2542601088214904,
+      "learning_rate": 3.225806451612903e-06,
+      "loss": 0.6985,
+      "step": 10
+    },
+    {
+      "epoch": 0.044534412955465584,
+      "grad_norm": 2.2032571765792985,
+      "learning_rate": 3.548387096774194e-06,
+      "loss": 0.7179,
+      "step": 11
+    },
+    {
+      "epoch": 0.048582995951417005,
+      "grad_norm": 3.647945495628868,
+      "learning_rate": 3.870967741935484e-06,
+      "loss": 0.6832,
+      "step": 12
+    },
+    {
+      "epoch": 0.05263157894736842,
+      "grad_norm": 3.95683048115789,
+      "learning_rate": 4.193548387096774e-06,
+      "loss": 0.6861,
+      "step": 13
+    },
+    {
+      "epoch": 0.05668016194331984,
+      "grad_norm": 3.8250476402419076,
+      "learning_rate": 4.516129032258065e-06,
+      "loss": 0.6654,
+      "step": 14
+    },
+    {
+      "epoch": 0.06072874493927125,
+      "grad_norm": 3.813623308512108,
+      "learning_rate": 4.838709677419355e-06,
+      "loss": 0.675,
+      "step": 15
+    },
+    {
+      "epoch": 0.06477732793522267,
+      "grad_norm": 3.342551759067519,
+      "learning_rate": 5.161290322580646e-06,
+      "loss": 0.658,
+      "step": 16
+    },
+    {
+      "epoch": 0.06882591093117409,
+      "grad_norm": 3.0546880024091987,
+      "learning_rate": 5.483870967741935e-06,
+      "loss": 0.6156,
+      "step": 17
+    },
+    {
+      "epoch": 0.0728744939271255,
+      "grad_norm": 2.4438730008951737,
+      "learning_rate": 5.806451612903226e-06,
+      "loss": 0.6101,
+      "step": 18
+    },
+    {
+      "epoch": 0.07692307692307693,
+      "grad_norm": 1.8709324873543496,
+      "learning_rate": 6.129032258064517e-06,
+      "loss": 0.5976,
+      "step": 19
+    },
+    {
+      "epoch": 0.08097165991902834,
+      "grad_norm": 1.4384005363744663,
+      "learning_rate": 6.451612903225806e-06,
+      "loss": 0.5987,
+      "step": 20
+    },
+    {
+      "epoch": 0.08502024291497975,
+      "grad_norm": 1.5182481532788914,
+      "learning_rate": 6.774193548387097e-06,
+      "loss": 0.5877,
+      "step": 21
+    },
+    {
+      "epoch": 0.08906882591093117,
+      "grad_norm": 1.749470623957126,
+      "learning_rate": 7.096774193548388e-06,
+      "loss": 0.5671,
+      "step": 22
+    },
+    {
+      "epoch": 0.0931174089068826,
+      "grad_norm": 1.5376806727286971,
+      "learning_rate": 7.4193548387096784e-06,
+      "loss": 0.5716,
+      "step": 23
+    },
+    {
+      "epoch": 0.09716599190283401,
+      "grad_norm": 1.0451594917535298,
+      "learning_rate": 7.741935483870968e-06,
+      "loss": 0.5487,
+      "step": 24
+    },
+    {
+      "epoch": 0.10121457489878542,
+      "grad_norm": 0.7973668507410387,
+      "learning_rate": 8.064516129032258e-06,
+      "loss": 0.5603,
+      "step": 25
+    },
+    {
+      "epoch": 0.10526315789473684,
+      "grad_norm": 1.0989073039605004,
+      "learning_rate": 8.387096774193549e-06,
+      "loss": 0.5404,
+      "step": 26
+    },
+    {
+      "epoch": 0.10931174089068826,
+      "grad_norm": 1.046449446461705,
+      "learning_rate": 8.70967741935484e-06,
+      "loss": 0.5269,
+      "step": 27
+    },
+    {
+      "epoch": 0.11336032388663968,
+      "grad_norm": 0.7455215591846296,
+      "learning_rate": 9.03225806451613e-06,
+      "loss": 0.529,
+      "step": 28
+    },
+    {
+      "epoch": 0.11740890688259109,
+      "grad_norm": 0.7684061098839466,
+      "learning_rate": 9.35483870967742e-06,
+      "loss": 0.5079,
+      "step": 29
+    },
+    {
+      "epoch": 0.1214574898785425,
+      "grad_norm": 0.96638040087196,
+      "learning_rate": 9.67741935483871e-06,
+      "loss": 0.5589,
+      "step": 30
+    },
+    {
+      "epoch": 0.12550607287449392,
+      "grad_norm": 0.7709136873134572,
+      "learning_rate": 1e-05,
+      "loss": 0.502,
+      "step": 31
+    },
+    {
+      "epoch": 0.12955465587044535,
+      "grad_norm": 0.6981605474634434,
+      "learning_rate": 1.0322580645161291e-05,
+      "loss": 0.5315,
+      "step": 32
+    },
+    {
+      "epoch": 0.13360323886639677,
+      "grad_norm": 0.7968959302830291,
+      "learning_rate": 1.0645161290322582e-05,
+      "loss": 0.5508,
+      "step": 33
+    },
+    {
+      "epoch": 0.13765182186234817,
+      "grad_norm": 0.6647669046209145,
+      "learning_rate": 1.096774193548387e-05,
+      "loss": 0.5097,
+      "step": 34
+    },
+    {
+      "epoch": 0.1417004048582996,
+      "grad_norm": 0.5201591544512119,
+      "learning_rate": 1.1290322580645164e-05,
+      "loss": 0.4961,
+      "step": 35
+    },
+    {
+      "epoch": 0.145748987854251,
+      "grad_norm": 0.5593960116198085,
+      "learning_rate": 1.1612903225806453e-05,
+      "loss": 0.4952,
+      "step": 36
+    },
+    {
+      "epoch": 0.14979757085020243,
+      "grad_norm": 0.6942288258736914,
+      "learning_rate": 1.1935483870967743e-05,
+      "loss": 0.5223,
+      "step": 37
+    },
+    {
+      "epoch": 0.15384615384615385,
+      "grad_norm": 0.5786641053572443,
+      "learning_rate": 1.2258064516129034e-05,
+      "loss": 0.5158,
+      "step": 38
+    },
+    {
+      "epoch": 0.15789473684210525,
+      "grad_norm": 0.6435609001759982,
+      "learning_rate": 1.2580645161290324e-05,
+      "loss": 0.488,
+      "step": 39
+    },
+    {
+      "epoch": 0.16194331983805668,
+      "grad_norm": 0.5739288256977902,
+      "learning_rate": 1.2903225806451613e-05,
+      "loss": 0.4921,
+      "step": 40
+    },
+    {
+      "epoch": 0.1659919028340081,
+      "grad_norm": 0.5871542080738715,
+      "learning_rate": 1.3225806451612903e-05,
+      "loss": 0.4992,
+      "step": 41
+    },
+    {
+      "epoch": 0.1700404858299595,
+      "grad_norm": 0.5992075767212771,
+      "learning_rate": 1.3548387096774194e-05,
+      "loss": 0.4955,
+      "step": 42
+    },
+    {
+      "epoch": 0.17408906882591094,
+      "grad_norm": 0.604287588668187,
+      "learning_rate": 1.3870967741935486e-05,
+      "loss": 0.4901,
+      "step": 43
+    },
+    {
+      "epoch": 0.17813765182186234,
+      "grad_norm": 0.7058143226748904,
+      "learning_rate": 1.4193548387096776e-05,
+      "loss": 0.5004,
+      "step": 44
+    },
+    {
+      "epoch": 0.18218623481781376,
+      "grad_norm": 0.7217350252908545,
+      "learning_rate": 1.4516129032258066e-05,
+      "loss": 0.4685,
+      "step": 45
+    },
+    {
+      "epoch": 0.1862348178137652,
+      "grad_norm": 0.7176598910364846,
+      "learning_rate": 1.4838709677419357e-05,
+      "loss": 0.468,
+      "step": 46
+    },
+    {
+      "epoch": 0.1902834008097166,
+      "grad_norm": 0.6404297801661366,
+      "learning_rate": 1.5161290322580646e-05,
+      "loss": 0.4877,
+      "step": 47
+    },
+    {
+      "epoch": 0.19433198380566802,
+      "grad_norm": 0.6480231058628698,
+      "learning_rate": 1.5483870967741936e-05,
+      "loss": 0.4819,
+      "step": 48
+    },
+    {
+      "epoch": 0.19838056680161945,
+      "grad_norm": 0.6539909126030891,
+      "learning_rate": 1.5806451612903226e-05,
+      "loss": 0.4996,
+      "step": 49
+    },
+    {
+      "epoch": 0.20242914979757085,
+      "grad_norm": 0.535393170340933,
+      "learning_rate": 1.6129032258064517e-05,
+      "loss": 0.471,
+      "step": 50
+    },
+    {
+      "epoch": 0.20647773279352227,
+      "grad_norm": 0.7017141116079592,
+      "learning_rate": 1.6451612903225807e-05,
+      "loss": 0.4689,
+      "step": 51
+    },
+    {
+      "epoch": 0.21052631578947367,
+      "grad_norm": 0.5201697440445429,
+      "learning_rate": 1.6774193548387098e-05,
+      "loss": 0.4809,
+      "step": 52
+    },
+    {
+      "epoch": 0.2145748987854251,
+      "grad_norm": 0.7197134549012842,
+      "learning_rate": 1.7096774193548388e-05,
+      "loss": 0.4903,
+      "step": 53
+    },
+    {
+      "epoch": 0.21862348178137653,
+      "grad_norm": 0.5187256379416644,
+      "learning_rate": 1.741935483870968e-05,
+      "loss": 0.4664,
+      "step": 54
+    },
+    {
+      "epoch": 0.22267206477732793,
+      "grad_norm": 0.6588080498232608,
+      "learning_rate": 1.774193548387097e-05,
+      "loss": 0.4635,
+      "step": 55
+    },
+    {
+      "epoch": 0.22672064777327935,
+      "grad_norm": 0.6641750476098391,
+      "learning_rate": 1.806451612903226e-05,
+      "loss": 0.4947,
+      "step": 56
+    },
+    {
+      "epoch": 0.23076923076923078,
+      "grad_norm": 0.6506069871396042,
+      "learning_rate": 1.838709677419355e-05,
+      "loss": 0.4661,
+      "step": 57
+    },
+    {
+      "epoch": 0.23481781376518218,
+      "grad_norm": 0.7491716021573606,
+      "learning_rate": 1.870967741935484e-05,
+      "loss": 0.4637,
+      "step": 58
+    },
+    {
+      "epoch": 0.2388663967611336,
+      "grad_norm": 0.6679788826562139,
+      "learning_rate": 1.903225806451613e-05,
+      "loss": 0.4753,
+      "step": 59
+    },
+    {
+      "epoch": 0.242914979757085,
+      "grad_norm": 0.7775070194477786,
+      "learning_rate": 1.935483870967742e-05,
+      "loss": 0.4511,
+      "step": 60
+    },
+    {
+      "epoch": 0.24696356275303644,
+      "grad_norm": 0.708368434391324,
+      "learning_rate": 1.967741935483871e-05,
+      "loss": 0.4432,
+      "step": 61
+    },
+    {
+      "epoch": 0.25101214574898784,
+      "grad_norm": 0.6635962134288701,
+      "learning_rate": 2e-05,
+      "loss": 0.4362,
+      "step": 62
+    },
+    {
+      "epoch": 0.2550607287449393,
+      "grad_norm": 0.7108051205544056,
+      "learning_rate": 2.0322580645161292e-05,
+      "loss": 0.462,
+      "step": 63
+    },
+    {
+      "epoch": 0.2591093117408907,
+      "grad_norm": 0.5348871302034774,
+      "learning_rate": 2.0645161290322582e-05,
+      "loss": 0.4864,
+      "step": 64
+    },
+    {
+      "epoch": 0.2631578947368421,
+      "grad_norm": 0.6566511380889785,
+      "learning_rate": 2.0967741935483873e-05,
+      "loss": 0.4811,
+      "step": 65
+    },
+    {
+      "epoch": 0.26720647773279355,
+      "grad_norm": 0.504206968653019,
+      "learning_rate": 2.1290322580645163e-05,
+      "loss": 0.4684,
+      "step": 66
+    },
+    {
+      "epoch": 0.27125506072874495,
+      "grad_norm": 0.6108983850986108,
+      "learning_rate": 2.161290322580645e-05,
+      "loss": 0.467,
+      "step": 67
+    },
+    {
+      "epoch": 0.27530364372469635,
+      "grad_norm": 0.5163713527270782,
+      "learning_rate": 2.193548387096774e-05,
+      "loss": 0.4594,
+      "step": 68
+    },
+    {
+      "epoch": 0.2793522267206478,
+      "grad_norm": 0.7510596240419984,
+      "learning_rate": 2.225806451612903e-05,
+      "loss": 0.4702,
+      "step": 69
+    },
+    {
+      "epoch": 0.2834008097165992,
+      "grad_norm": 0.5679551087010409,
+      "learning_rate": 2.2580645161290328e-05,
+      "loss": 0.4584,
+      "step": 70
+    },
+    {
+      "epoch": 0.2874493927125506,
+      "grad_norm": 0.5496850694840804,
+      "learning_rate": 2.2903225806451615e-05,
+      "loss": 0.4721,
+      "step": 71
+    },
+    {
+      "epoch": 0.291497975708502,
+      "grad_norm": 0.6231017885111215,
+      "learning_rate": 2.3225806451612906e-05,
+      "loss": 0.4805,
+      "step": 72
+    },
+    {
+      "epoch": 0.29554655870445345,
+      "grad_norm": 0.4875436561172546,
+      "learning_rate": 2.3548387096774196e-05,
+      "loss": 0.4589,
+      "step": 73
+    },
+    {
+      "epoch": 0.29959514170040485,
+      "grad_norm": 0.5219559973265657,
+      "learning_rate": 2.3870967741935486e-05,
+      "loss": 0.4644,
+      "step": 74
+    },
+    {
+      "epoch": 0.30364372469635625,
+      "grad_norm": 0.5513526651831108,
+      "learning_rate": 2.4193548387096777e-05,
+      "loss": 0.4766,
+      "step": 75
+    },
+    {
+      "epoch": 0.3076923076923077,
+      "grad_norm": 0.4977478347703895,
+      "learning_rate": 2.4516129032258067e-05,
+      "loss": 0.4436,
+      "step": 76
+    },
+    {
+      "epoch": 0.3117408906882591,
+      "grad_norm": 0.7101631374791924,
+      "learning_rate": 2.4838709677419358e-05,
+      "loss": 0.4729,
+      "step": 77
+    },
+    {
+      "epoch": 0.3157894736842105,
+      "grad_norm": 0.6181618034899764,
+      "learning_rate": 2.5161290322580648e-05,
+      "loss": 0.4704,
+      "step": 78
+    },
+    {
+      "epoch": 0.31983805668016196,
+      "grad_norm": 0.5288330702809824,
+      "learning_rate": 2.548387096774194e-05,
+      "loss": 0.4356,
+      "step": 79
+    },
+    {
+      "epoch": 0.32388663967611336,
+      "grad_norm": 0.705870592440069,
+      "learning_rate": 2.5806451612903226e-05,
+      "loss": 0.4785,
+      "step": 80
+    },
+    {
+      "epoch": 0.32793522267206476,
+      "grad_norm": 0.6501425840457334,
+      "learning_rate": 2.6129032258064516e-05,
+      "loss": 0.459,
+      "step": 81
+    },
+    {
+      "epoch": 0.3319838056680162,
+      "grad_norm": 0.7809524783574461,
+      "learning_rate": 2.6451612903225806e-05,
+      "loss": 0.462,
+      "step": 82
+    },
+    {
+      "epoch": 0.3360323886639676,
+      "grad_norm": 0.9236787452242621,
+      "learning_rate": 2.6774193548387097e-05,
+      "loss": 0.4648,
+      "step": 83
+    },
+    {
+      "epoch": 0.340080971659919,
+      "grad_norm": 0.5910898766311075,
+      "learning_rate": 2.7096774193548387e-05,
+      "loss": 0.4518,
+      "step": 84
+    },
+    {
+      "epoch": 0.3441295546558704,
+      "grad_norm": 0.7073885036104086,
+      "learning_rate": 2.7419354838709678e-05,
+      "loss": 0.4353,
+      "step": 85
+    },
+    {
+      "epoch": 0.3481781376518219,
+      "grad_norm": 0.7578645165498111,
+      "learning_rate": 2.774193548387097e-05,
+      "loss": 0.4264,
+      "step": 86
+    },
+    {
+      "epoch": 0.3522267206477733,
+      "grad_norm": 0.5399050409191374,
+      "learning_rate": 2.8064516129032262e-05,
+      "loss": 0.4379,
+      "step": 87
+    },
+    {
+      "epoch": 0.3562753036437247,
+      "grad_norm": 0.7577611831561102,
+      "learning_rate": 2.8387096774193552e-05,
+      "loss": 0.4645,
+      "step": 88
+    },
+    {
+      "epoch": 0.3603238866396761,
+      "grad_norm": 0.7586315639643703,
+      "learning_rate": 2.8709677419354843e-05,
+      "loss": 0.4484,
+      "step": 89
+    },
+    {
+      "epoch": 0.3643724696356275,
+      "grad_norm": 0.7210708331646452,
+      "learning_rate": 2.9032258064516133e-05,
+      "loss": 0.4589,
+      "step": 90
+    },
+    {
+      "epoch": 0.3684210526315789,
+      "grad_norm": 0.6611816160306119,
+      "learning_rate": 2.9354838709677423e-05,
+      "loss": 0.447,
+      "step": 91
+    },
+    {
+      "epoch": 0.3724696356275304,
+      "grad_norm": 0.8224092515529183,
+      "learning_rate": 2.9677419354838714e-05,
+      "loss": 0.4522,
+      "step": 92
+    },
+    {
+      "epoch": 0.3765182186234818,
+      "grad_norm": 0.6819768048155815,
+      "learning_rate": 3.0000000000000004e-05,
+      "loss": 0.4474,
+      "step": 93
+    },
+    {
+      "epoch": 0.3805668016194332,
+      "grad_norm": 1.0198996121966875,
+      "learning_rate": 3.032258064516129e-05,
+      "loss": 0.4683,
+      "step": 94
+    },
+    {
+      "epoch": 0.38461538461538464,
+      "grad_norm": 0.8926709771842097,
+      "learning_rate": 3.0645161290322585e-05,
+      "loss": 0.4555,
+      "step": 95
+    },
+    {
+      "epoch": 0.38866396761133604,
+      "grad_norm": 0.6206357962283122,
+      "learning_rate": 3.096774193548387e-05,
+      "loss": 0.4352,
+      "step": 96
+    },
+    {
+      "epoch": 0.39271255060728744,
+      "grad_norm": 0.6069826981963206,
+      "learning_rate": 3.1290322580645166e-05,
+      "loss": 0.4562,
+      "step": 97
+    },
+    {
+      "epoch": 0.3967611336032389,
+      "grad_norm": 0.6967314698715243,
+      "learning_rate": 3.161290322580645e-05,
+      "loss": 0.4506,
+      "step": 98
+    },
+    {
+      "epoch": 0.4008097165991903,
+      "grad_norm": 0.7844956322113878,
+      "learning_rate": 3.193548387096774e-05,
+      "loss": 0.4769,
+      "step": 99
+    },
+    {
+      "epoch": 0.4048582995951417,
+      "grad_norm": 0.7000681669129033,
+      "learning_rate": 3.2258064516129034e-05,
+      "loss": 0.4413,
+      "step": 100
+    },
+    {
+      "epoch": 0.4089068825910931,
+      "grad_norm": 0.6608389761492879,
+      "learning_rate": 3.258064516129033e-05,
+      "loss": 0.4644,
+      "step": 101
+    },
+    {
+      "epoch": 0.41295546558704455,
+      "grad_norm": 0.9677925392641267,
+      "learning_rate": 3.2903225806451614e-05,
+      "loss": 0.4685,
+      "step": 102
+    },
+    {
+      "epoch": 0.41700404858299595,
+      "grad_norm": 0.8928376160526392,
+      "learning_rate": 3.322580645161291e-05,
+      "loss": 0.4459,
+      "step": 103
+    },
+    {
+      "epoch": 0.42105263157894735,
+      "grad_norm": 0.7533760691562938,
+      "learning_rate": 3.3548387096774195e-05,
+      "loss": 0.4392,
+      "step": 104
+    },
+    {
+      "epoch": 0.4251012145748988,
+      "grad_norm": 0.9357446214772136,
+      "learning_rate": 3.387096774193549e-05,
+      "loss": 0.4681,
+      "step": 105
+    },
+    {
+      "epoch": 0.4291497975708502,
+      "grad_norm": 0.8582290058575414,
+      "learning_rate": 3.4193548387096776e-05,
+      "loss": 0.4516,
+      "step": 106
+    },
+    {
+      "epoch": 0.4331983805668016,
+      "grad_norm": 0.9381939119523846,
+      "learning_rate": 3.451612903225807e-05,
+      "loss": 0.4681,
+      "step": 107
+    },
+    {
+      "epoch": 0.43724696356275305,
+      "grad_norm": 0.7353805086600403,
+      "learning_rate": 3.483870967741936e-05,
+      "loss": 0.4343,
+      "step": 108
+    },
+    {
+      "epoch": 0.44129554655870445,
+      "grad_norm": 0.8438540152744822,
+      "learning_rate": 3.516129032258065e-05,
+      "loss": 0.4524,
+      "step": 109
+    },
+    {
+      "epoch": 0.44534412955465585,
+      "grad_norm": 0.8719238279652175,
+      "learning_rate": 3.548387096774194e-05,
+      "loss": 0.4516,
+      "step": 110
+    },
+    {
+      "epoch": 0.4493927125506073,
+      "grad_norm": 0.7574537473514303,
+      "learning_rate": 3.5806451612903225e-05,
+      "loss": 0.4406,
+      "step": 111
+    },
+    {
+      "epoch": 0.4534412955465587,
+      "grad_norm": 0.8247802917617257,
+      "learning_rate": 3.612903225806452e-05,
+      "loss": 0.4623,
+      "step": 112
+    },
+    {
+      "epoch": 0.4574898785425101,
+      "grad_norm": 0.8609385436001783,
+      "learning_rate": 3.6451612903225805e-05,
+      "loss": 0.4589,
+      "step": 113
+    },
+    {
+      "epoch": 0.46153846153846156,
+      "grad_norm": 0.9558518954415965,
+      "learning_rate": 3.67741935483871e-05,
+      "loss": 0.4667,
+      "step": 114
+    },
+    {
+      "epoch": 0.46558704453441296,
+      "grad_norm": 0.883596686420253,
+      "learning_rate": 3.7096774193548386e-05,
+      "loss": 0.4499,
+      "step": 115
+    },
+    {
+      "epoch": 0.46963562753036436,
+      "grad_norm": 0.9302743287027387,
+      "learning_rate": 3.741935483870968e-05,
+      "loss": 0.4575,
+      "step": 116
+    },
+    {
+      "epoch": 0.47368421052631576,
+      "grad_norm": 0.893775936189633,
+      "learning_rate": 3.7741935483870974e-05,
+      "loss": 0.4368,
+      "step": 117
+    },
+    {
+      "epoch": 0.4777327935222672,
+      "grad_norm": 1.0762933925575693,
+      "learning_rate": 3.806451612903226e-05,
+      "loss": 0.4672,
+      "step": 118
+    },
+    {
+      "epoch": 0.4817813765182186,
+      "grad_norm": 0.8760446284434412,
+      "learning_rate": 3.8387096774193555e-05,
+      "loss": 0.4369,
+      "step": 119
+    },
+    {
+      "epoch": 0.48582995951417,
+      "grad_norm": 0.7365084436013108,
+      "learning_rate": 3.870967741935484e-05,
+      "loss": 0.4474,
+      "step": 120
+    },
+    {
+      "epoch": 0.4898785425101215,
+      "grad_norm": 0.7331940631503605,
+      "learning_rate": 3.9032258064516135e-05,
+      "loss": 0.4348,
+      "step": 121
+    },
+    {
+      "epoch": 0.4939271255060729,
+      "grad_norm": 0.6534088193382599,
+      "learning_rate": 3.935483870967742e-05,
+      "loss": 0.4434,
+      "step": 122
+    },
+    {
+      "epoch": 0.4979757085020243,
+      "grad_norm": 0.8643140950548386,
+      "learning_rate": 3.9677419354838716e-05,
+      "loss": 0.4427,
+      "step": 123
+    },
+    {
+      "epoch": 0.5020242914979757,
+      "grad_norm": 0.5565409171263412,
+      "learning_rate": 4e-05,
+      "loss": 0.4649,
+      "step": 124
+    },
+    {
+      "epoch": 0.5060728744939271,
+      "grad_norm": 0.9914408388083505,
+      "learning_rate": 3.9999920040266474e-05,
+      "loss": 0.4564,
+      "step": 125
+    },
+    {
+      "epoch": 0.5101214574898786,
+      "grad_norm": 0.8191708900433339,
+      "learning_rate": 3.999968016170525e-05,
+      "loss": 0.4526,
+      "step": 126
+    },
+    {
+      "epoch": 0.5141700404858299,
+      "grad_norm": 0.7231458176946067,
+      "learning_rate": 3.9999280366234394e-05,
+      "loss": 0.4351,
+      "step": 127
+    },
+    {
+      "epoch": 0.5182186234817814,
+      "grad_norm": 0.8654594575436407,
+      "learning_rate": 3.999872065705065e-05,
+      "loss": 0.4575,
+      "step": 128
+    },
+    {
+      "epoch": 0.5222672064777328,
+      "grad_norm": 0.6591349442864614,
+      "learning_rate": 3.999800103862945e-05,
+      "loss": 0.4409,
+      "step": 129
+    },
+    {
+      "epoch": 0.5263157894736842,
+      "grad_norm": 0.7601914161420072,
+      "learning_rate": 3.9997121516724834e-05,
+      "loss": 0.4301,
+      "step": 130
+    },
+    {
+      "epoch": 0.5303643724696356,
+      "grad_norm": 0.7458095525018785,
+      "learning_rate": 3.999608209836944e-05,
+      "loss": 0.4287,
+      "step": 131
+    },
+    {
+      "epoch": 0.5344129554655871,
+      "grad_norm": 0.5606494945052176,
+      "learning_rate": 3.999488279187443e-05,
+      "loss": 0.446,
+      "step": 132
+    },
+    {
+      "epoch": 0.5384615384615384,
+      "grad_norm": 0.5787593583035353,
+      "learning_rate": 3.9993523606829425e-05,
+      "loss": 0.4211,
+      "step": 133
+    },
+    {
+      "epoch": 0.5425101214574899,
+      "grad_norm": 0.6510097405454264,
+      "learning_rate": 3.9992004554102435e-05,
+      "loss": 0.433,
+      "step": 134
+    },
+    {
+      "epoch": 0.5465587044534413,
+      "grad_norm": 0.5815225304417536,
+      "learning_rate": 3.999032564583976e-05,
+      "loss": 0.4501,
+      "step": 135
+    },
+    {
+      "epoch": 0.5506072874493927,
+      "grad_norm": 0.5906274158015927,
+      "learning_rate": 3.998848689546591e-05,
+      "loss": 0.4417,
+      "step": 136
+    },
+    {
+      "epoch": 0.5546558704453441,
+      "grad_norm": 0.48931519220650227,
+      "learning_rate": 3.9986488317683496e-05,
+      "loss": 0.4442,
+      "step": 137
+    },
+    {
+      "epoch": 0.5587044534412956,
+      "grad_norm": 0.5192492666048689,
+      "learning_rate": 3.9984329928473067e-05,
+      "loss": 0.4282,
+      "step": 138
+    },
+    {
+      "epoch": 0.562753036437247,
+      "grad_norm": 0.4897803822454282,
+      "learning_rate": 3.998201174509306e-05,
+      "loss": 0.4372,
+      "step": 139
+    },
+    {
+      "epoch": 0.5668016194331984,
+      "grad_norm": 0.5831156656620063,
+      "learning_rate": 3.997953378607961e-05,
+      "loss": 0.4811,
+      "step": 140
+    },
+    {
+      "epoch": 0.5708502024291497,
+      "grad_norm": 0.5451650830579229,
+      "learning_rate": 3.9976896071246403e-05,
+      "loss": 0.4592,
+      "step": 141
+    },
+    {
+      "epoch": 0.5748987854251012,
+      "grad_norm": 0.44914071984099796,
+      "learning_rate": 3.997409862168454e-05,
+      "loss": 0.4205,
+      "step": 142
+    },
+    {
+      "epoch": 0.5789473684210527,
+      "grad_norm": 0.48398926859701746,
+      "learning_rate": 3.997114145976237e-05,
+      "loss": 0.4591,
+      "step": 143
+    },
+    {
+      "epoch": 0.582995951417004,
+      "grad_norm": 0.5296489528832491,
+      "learning_rate": 3.9968024609125245e-05,
+      "loss": 0.4414,
+      "step": 144
+    },
+    {
+      "epoch": 0.5870445344129555,
+      "grad_norm": 0.48869503036941275,
+      "learning_rate": 3.996474809469545e-05,
+      "loss": 0.4578,
+      "step": 145
+    },
+    {
+      "epoch": 0.5910931174089069,
+      "grad_norm": 0.4893763205635697,
+      "learning_rate": 3.996131194267189e-05,
+      "loss": 0.4569,
+      "step": 146
+    },
+    {
+      "epoch": 0.5951417004048583,
+      "grad_norm": 0.5165915999690879,
+      "learning_rate": 3.995771618052995e-05,
+      "loss": 0.4538,
+      "step": 147
+    },
+    {
+      "epoch": 0.5991902834008097,
+      "grad_norm": 0.6218695227446923,
+      "learning_rate": 3.9953960837021256e-05,
+      "loss": 0.4357,
+      "step": 148
+    },
+    {
+      "epoch": 0.6032388663967612,
+      "grad_norm": 0.8642353550479025,
+      "learning_rate": 3.995004594217343e-05,
+      "loss": 0.4531,
+      "step": 149
+    },
+    {
+      "epoch": 0.6072874493927125,
+      "grad_norm": 0.5050096160071239,
+      "learning_rate": 3.994597152728987e-05,
+      "loss": 0.4915,
+      "step": 150
+    },
+    {
+      "epoch": 0.611336032388664,
+      "grad_norm": 0.4971290103582981,
+      "learning_rate": 3.994173762494947e-05,
+      "loss": 0.426,
+      "step": 151
+    },
+    {
+      "epoch": 0.6153846153846154,
+      "grad_norm": 0.6057906846812047,
+      "learning_rate": 3.9937344269006424e-05,
+      "loss": 0.45,
+      "step": 152
+    },
+    {
+      "epoch": 0.6194331983805668,
+      "grad_norm": 0.4612526852277003,
+      "learning_rate": 3.993279149458988e-05,
+      "loss": 0.4475,
+      "step": 153
+    },
+    {
+      "epoch": 0.6234817813765182,
+      "grad_norm": 0.5248040984788032,
+      "learning_rate": 3.9928079338103695e-05,
+      "loss": 0.426,
+      "step": 154
+    },
+    {
+      "epoch": 0.6275303643724697,
+      "grad_norm": 0.4695215974608953,
+      "learning_rate": 3.992320783722616e-05,
+      "loss": 0.4662,
+      "step": 155
+    },
+    {
+      "epoch": 0.631578947368421,
+      "grad_norm": 0.7133329645806846,
+      "learning_rate": 3.991817703090965e-05,
+      "loss": 0.4462,
+      "step": 156
+    },
+    {
+      "epoch": 0.6356275303643725,
+      "grad_norm": 0.5702660345203943,
+      "learning_rate": 3.991298695938038e-05,
+      "loss": 0.4418,
+      "step": 157
+    },
+    {
+      "epoch": 0.6396761133603239,
+      "grad_norm": 0.5069983340509612,
+      "learning_rate": 3.9907637664138e-05,
+      "loss": 0.4331,
+      "step": 158
+    },
+    {
+      "epoch": 0.6437246963562753,
+      "grad_norm": 0.514320366348836,
+      "learning_rate": 3.990212918795534e-05,
+      "loss": 0.4301,
+      "step": 159
+    },
+    {
+      "epoch": 0.6477732793522267,
+      "grad_norm": 0.47387431847249917,
+      "learning_rate": 3.9896461574878036e-05,
+      "loss": 0.4438,
+      "step": 160
+    },
+    {
+      "epoch": 0.6518218623481782,
+      "grad_norm": 0.5327661712319929,
+      "learning_rate": 3.989063487022416e-05,
+      "loss": 0.444,
+      "step": 161
+    },
+    {
+      "epoch": 0.6558704453441295,
+      "grad_norm": 0.4409495182552322,
+      "learning_rate": 3.9884649120583903e-05,
+      "loss": 0.4025,
+      "step": 162
+    },
+    {
+      "epoch": 0.659919028340081,
+      "grad_norm": 0.491789664254398,
+      "learning_rate": 3.987850437381915e-05,
+      "loss": 0.4529,
+      "step": 163
+    },
+    {
+      "epoch": 0.6639676113360324,
+      "grad_norm": 0.4460045899918977,
+      "learning_rate": 3.987220067906313e-05,
+      "loss": 0.4415,
+      "step": 164
+    },
+    {
+      "epoch": 0.6680161943319838,
+      "grad_norm": 0.4591010275623563,
+      "learning_rate": 3.986573808672002e-05,
+      "loss": 0.4462,
+      "step": 165
+    },
+    {
+      "epoch": 0.6720647773279352,
+      "grad_norm": 0.48857392126918303,
+      "learning_rate": 3.9859116648464534e-05,
+      "loss": 0.4458,
+      "step": 166
+    },
+    {
+      "epoch": 0.6761133603238867,
+      "grad_norm": 0.47988280681516676,
+      "learning_rate": 3.985233641724152e-05,
+      "loss": 0.4092,
+      "step": 167
+    },
+    {
+      "epoch": 0.680161943319838,
+      "grad_norm": 0.5331464801051145,
+      "learning_rate": 3.984539744726554e-05,
+      "loss": 0.4296,
+      "step": 168
+    },
+    {
+      "epoch": 0.6842105263157895,
+      "grad_norm": 0.5416977195316846,
+      "learning_rate": 3.983829979402038e-05,
+      "loss": 0.4393,
+      "step": 169
+    },
+    {
+      "epoch": 0.6882591093117408,
+      "grad_norm": 0.5576744339814727,
+      "learning_rate": 3.983104351425871e-05,
+      "loss": 0.4268,
+      "step": 170
+    },
+    {
+      "epoch": 0.6923076923076923,
+      "grad_norm": 0.6292964504338814,
+      "learning_rate": 3.982362866600154e-05,
+      "loss": 0.4606,
+      "step": 171
+    },
+    {
+      "epoch": 0.6963562753036437,
+      "grad_norm": 0.6419362202246865,
+      "learning_rate": 3.9816055308537806e-05,
+      "loss": 0.437,
+      "step": 172
+    },
+    {
+      "epoch": 0.7004048582995951,
+      "grad_norm": 0.5315089267893636,
+      "learning_rate": 3.980832350242388e-05,
+      "loss": 0.4385,
+      "step": 173
+    },
+    {
+      "epoch": 0.7044534412955465,
+      "grad_norm": 0.6223428039165441,
+      "learning_rate": 3.980043330948306e-05,
+      "loss": 0.4515,
+      "step": 174
+    },
+    {
+      "epoch": 0.708502024291498,
+      "grad_norm": 0.5462774723152037,
+      "learning_rate": 3.9792384792805124e-05,
+      "loss": 0.4425,
+      "step": 175
+    },
+    {
+      "epoch": 0.7125506072874493,
+      "grad_norm": 0.5117549480306245,
+      "learning_rate": 3.978417801674579e-05,
+      "loss": 0.425,
+      "step": 176
+    },
+    {
+      "epoch": 0.7165991902834008,
+      "grad_norm": 0.6110250701775746,
+      "learning_rate": 3.9775813046926244e-05,
+      "loss": 0.4393,
+      "step": 177
+    },
+    {
+      "epoch": 0.7206477732793523,
+      "grad_norm": 0.5684402310414942,
+      "learning_rate": 3.976728995023254e-05,
+      "loss": 0.4451,
+      "step": 178
+    },
+    {
+      "epoch": 0.7246963562753036,
+      "grad_norm": 0.47371357843648293,
+      "learning_rate": 3.975860879481514e-05,
+      "loss": 0.4564,
+      "step": 179
+    },
+    {
+      "epoch": 0.728744939271255,
+      "grad_norm": 0.5340442260876178,
+      "learning_rate": 3.974976965008833e-05,
+      "loss": 0.4485,
+      "step": 180
+    },
+    {
+      "epoch": 0.7327935222672065,
+      "grad_norm": 0.5386538274972921,
+      "learning_rate": 3.9740772586729674e-05,
+      "loss": 0.4398,
+      "step": 181
+    },
+    {
+      "epoch": 0.7368421052631579,
+      "grad_norm": 0.5896583451001711,
+      "learning_rate": 3.973161767667946e-05,
+      "loss": 0.4453,
+      "step": 182
+    },
+    {
+      "epoch": 0.7408906882591093,
+      "grad_norm": 0.47277244962222714,
+      "learning_rate": 3.972230499314009e-05,
+      "loss": 0.4127,
+      "step": 183
+    },
+    {
+      "epoch": 0.7449392712550608,
+      "grad_norm": 0.5307340962129721,
+      "learning_rate": 3.9712834610575545e-05,
+      "loss": 0.4504,
+      "step": 184
+    },
+    {
+      "epoch": 0.7489878542510121,
+      "grad_norm": 0.5165472174611158,
+      "learning_rate": 3.9703206604710746e-05,
+      "loss": 0.427,
+      "step": 185
+    },
+    {
+      "epoch": 0.7530364372469636,
+      "grad_norm": 0.5236236240040879,
+      "learning_rate": 3.969342105253098e-05,
+      "loss": 0.4506,
+      "step": 186
+    },
+    {
+      "epoch": 0.757085020242915,
+      "grad_norm": 0.5423830662668447,
+      "learning_rate": 3.9683478032281246e-05,
+      "loss": 0.414,
+      "step": 187
+    },
+    {
+      "epoch": 0.7611336032388664,
+      "grad_norm": 0.5294625977399926,
+      "learning_rate": 3.967337762346568e-05,
+      "loss": 0.4419,
+      "step": 188
+    },
+    {
+      "epoch": 0.7651821862348178,
+      "grad_norm": 0.4504806778773605,
+      "learning_rate": 3.9663119906846885e-05,
+      "loss": 0.4152,
+      "step": 189
+    },
+    {
+      "epoch": 0.7692307692307693,
+      "grad_norm": 0.5861508386326344,
+      "learning_rate": 3.965270496444528e-05,
+      "loss": 0.4579,
+      "step": 190
+    },
+    {
+      "epoch": 0.7732793522267206,
+      "grad_norm": 0.5211792903492162,
+      "learning_rate": 3.9642132879538465e-05,
+      "loss": 0.4444,
+      "step": 191
+    },
+    {
+      "epoch": 0.7773279352226721,
+      "grad_norm": 0.5628770023497878,
+      "learning_rate": 3.9631403736660566e-05,
+      "loss": 0.4309,
+      "step": 192
+    },
+    {
+      "epoch": 0.7813765182186235,
+      "grad_norm": 0.5444284104736354,
+      "learning_rate": 3.962051762160151e-05,
+      "loss": 0.4483,
+      "step": 193
+    },
+    {
+      "epoch": 0.7854251012145749,
+      "grad_norm": 0.5478206046086028,
+      "learning_rate": 3.960947462140638e-05,
+      "loss": 0.438,
+      "step": 194
+    },
+    {
+      "epoch": 0.7894736842105263,
+      "grad_norm": 0.5430077649232334,
+      "learning_rate": 3.959827482437472e-05,
+      "loss": 0.4282,
+      "step": 195
+    },
+    {
+      "epoch": 0.7935222672064778,
+      "grad_norm": 0.6390486493679963,
+      "learning_rate": 3.958691832005981e-05,
+      "loss": 0.4357,
+      "step": 196
+    },
+    {
+      "epoch": 0.7975708502024291,
+      "grad_norm": 0.6445162315631702,
+      "learning_rate": 3.9575405199267944e-05,
+      "loss": 0.435,
+      "step": 197
+    },
+    {
+      "epoch": 0.8016194331983806,
+      "grad_norm": 0.6700333927016985,
+      "learning_rate": 3.956373555405774e-05,
+      "loss": 0.4193,
+      "step": 198
+    },
+    {
+      "epoch": 0.805668016194332,
+      "grad_norm": 0.4464722599812626,
+      "learning_rate": 3.955190947773937e-05,
+      "loss": 0.4255,
+      "step": 199
+    },
+    {
+      "epoch": 0.8097165991902834,
+      "grad_norm": 0.7351153965161137,
+      "learning_rate": 3.9539927064873824e-05,
+      "loss": 0.4396,
+      "step": 200
+    },
+    {
+      "epoch": 0.8137651821862348,
+      "grad_norm": 0.4373565867596197,
+      "learning_rate": 3.952778841127214e-05,
+      "loss": 0.4472,
+      "step": 201
+    },
+    {
+      "epoch": 0.8178137651821862,
+      "grad_norm": 0.654705250518949,
+      "learning_rate": 3.95154936139947e-05,
+      "loss": 0.4364,
+      "step": 202
+    },
+    {
+      "epoch": 0.8218623481781376,
+      "grad_norm": 0.45839774034554553,
+      "learning_rate": 3.950304277135034e-05,
+      "loss": 0.4593,
+      "step": 203
+    },
+    {
+      "epoch": 0.8259109311740891,
+      "grad_norm": 0.5447605292662818,
+      "learning_rate": 3.94904359828957e-05,
+      "loss": 0.4382,
+      "step": 204
+    },
+    {
+      "epoch": 0.8299595141700404,
+      "grad_norm": 0.44545060531180214,
+      "learning_rate": 3.94776733494343e-05,
+      "loss": 0.416,
+      "step": 205
+    },
+    {
+      "epoch": 0.8340080971659919,
+      "grad_norm": 0.5438949702634067,
+      "learning_rate": 3.9464754973015826e-05,
+      "loss": 0.4565,
+      "step": 206
+    },
+    {
+      "epoch": 0.8380566801619433,
+      "grad_norm": 0.44431341333160634,
+      "learning_rate": 3.9451680956935275e-05,
+      "loss": 0.4398,
+      "step": 207
+    },
+    {
+      "epoch": 0.8421052631578947,
+      "grad_norm": 0.5797343880317395,
+      "learning_rate": 3.943845140573213e-05,
+      "loss": 0.431,
+      "step": 208
+    },
+    {
+      "epoch": 0.8461538461538461,
+      "grad_norm": 0.49318128133359773,
+      "learning_rate": 3.942506642518952e-05,
+      "loss": 0.4391,
+      "step": 209
+    },
+    {
+      "epoch": 0.8502024291497976,
+      "grad_norm": 0.6620418391249285,
+      "learning_rate": 3.9411526122333396e-05,
+      "loss": 0.4457,
+      "step": 210
+    },
+    {
+      "epoch": 0.854251012145749,
+      "grad_norm": 0.45823803643026034,
+      "learning_rate": 3.939783060543167e-05,
+      "loss": 0.4323,
+      "step": 211
+    },
+    {
+      "epoch": 0.8582995951417004,
+      "grad_norm": 0.6198413003090952,
+      "learning_rate": 3.938397998399333e-05,
+      "loss": 0.4418,
+      "step": 212
+    },
+    {
+      "epoch": 0.8623481781376519,
+      "grad_norm": 0.5205986185518108,
+      "learning_rate": 3.936997436876756e-05,
+      "loss": 0.4285,
+      "step": 213
+    },
+    {
+      "epoch": 0.8663967611336032,
+      "grad_norm": 0.42876204346045876,
+      "learning_rate": 3.935581387174289e-05,
+      "loss": 0.4457,
+      "step": 214
+    },
+    {
+      "epoch": 0.8704453441295547,
+      "grad_norm": 0.4785970176236987,
+      "learning_rate": 3.9341498606146295e-05,
+      "loss": 0.4236,
+      "step": 215
+    },
+    {
+      "epoch": 0.8744939271255061,
+      "grad_norm": 0.42253668409374107,
+      "learning_rate": 3.932702868644224e-05,
+      "loss": 0.4347,
+      "step": 216
+    },
+    {
+      "epoch": 0.8785425101214575,
+      "grad_norm": 0.4972143787577108,
+      "learning_rate": 3.931240422833181e-05,
+      "loss": 0.4547,
+      "step": 217
+    },
+    {
+      "epoch": 0.8825910931174089,
+      "grad_norm": 0.543002739749849,
+      "learning_rate": 3.9297625348751806e-05,
+      "loss": 0.4326,
+      "step": 218
+    },
+    {
+      "epoch": 0.8866396761133604,
+      "grad_norm": 0.46336239875388285,
+      "learning_rate": 3.9282692165873744e-05,
+      "loss": 0.4453,
+      "step": 219
+    },
+    {
+      "epoch": 0.8906882591093117,
+      "grad_norm": 0.44310139757577555,
+      "learning_rate": 3.9267604799102945e-05,
+      "loss": 0.4371,
+      "step": 220
+    },
+    {
+      "epoch": 0.8947368421052632,
+      "grad_norm": 0.4994855010907901,
+      "learning_rate": 3.9252363369077615e-05,
+      "loss": 0.4606,
+      "step": 221
+    },
+    {
+      "epoch": 0.8987854251012146,
+      "grad_norm": 0.4948889694460968,
+      "learning_rate": 3.9236967997667796e-05,
+      "loss": 0.4411,
+      "step": 222
+    },
+    {
+      "epoch": 0.902834008097166,
+      "grad_norm": 0.4494092023301079,
+      "learning_rate": 3.9221418807974496e-05,
+      "loss": 0.4332,
+      "step": 223
+    },
+    {
+      "epoch": 0.9068825910931174,
+      "grad_norm": 0.4114792162419875,
+      "learning_rate": 3.92057159243286e-05,
+      "loss": 0.4319,
+      "step": 224
+    },
+    {
+      "epoch": 0.9109311740890689,
+      "grad_norm": 0.488957794119599,
+      "learning_rate": 3.9189859472289956e-05,
+      "loss": 0.4452,
+      "step": 225
+    },
+    {
+      "epoch": 0.9149797570850202,
+      "grad_norm": 0.43594556112182525,
+      "learning_rate": 3.9173849578646324e-05,
+      "loss": 0.4352,
+      "step": 226
+    },
+    {
+      "epoch": 0.9190283400809717,
+      "grad_norm": 0.5105429654499162,
+      "learning_rate": 3.9157686371412395e-05,
+      "loss": 0.4295,
+      "step": 227
+    },
+    {
+      "epoch": 0.9230769230769231,
+      "grad_norm": 0.4531439149763565,
+      "learning_rate": 3.914136997982874e-05,
+      "loss": 0.4363,
+      "step": 228
+    },
+    {
+      "epoch": 0.9271255060728745,
+      "grad_norm": 0.5760719369844288,
+      "learning_rate": 3.912490053436079e-05,
+      "loss": 0.422,
+      "step": 229
+    },
+    {
+      "epoch": 0.9311740890688259,
+      "grad_norm": 0.5343710752827381,
+      "learning_rate": 3.9108278166697805e-05,
+      "loss": 0.4131,
+      "step": 230
+    },
+    {
+      "epoch": 0.9352226720647774,
+      "grad_norm": 0.4342570288750887,
+      "learning_rate": 3.909150300975178e-05,
+      "loss": 0.417,
+      "step": 231
+    },
+    {
+      "epoch": 0.9392712550607287,
+      "grad_norm": 0.4943033870078154,
+      "learning_rate": 3.907457519765642e-05,
+      "loss": 0.4302,
+      "step": 232
+    },
+    {
+      "epoch": 0.9433198380566802,
+      "grad_norm": 0.4736500590937538,
+      "learning_rate": 3.905749486576607e-05,
+      "loss": 0.4197,
+      "step": 233
+    },
+    {
+      "epoch": 0.9473684210526315,
+      "grad_norm": 0.5300415181782997,
+      "learning_rate": 3.90402621506546e-05,
+      "loss": 0.4407,
+      "step": 234
+    },
+    {
+      "epoch": 0.951417004048583,
+      "grad_norm": 0.4706423727766026,
+      "learning_rate": 3.902287719011435e-05,
+      "loss": 0.4346,
+      "step": 235
+    },
+    {
+      "epoch": 0.9554655870445344,
+      "grad_norm": 0.462950786800888,
+      "learning_rate": 3.900534012315499e-05,
+      "loss": 0.4225,
+      "step": 236
+    },
+    {
+      "epoch": 0.9595141700404858,
+      "grad_norm": 0.48206865885725714,
+      "learning_rate": 3.8987651090002454e-05,
+      "loss": 0.4325,
+      "step": 237
+    },
+    {
+      "epoch": 0.9635627530364372,
+      "grad_norm": 0.5509844913162009,
+      "learning_rate": 3.896981023209777e-05,
+      "loss": 0.4128,
+      "step": 238
+    },
+    {
+      "epoch": 0.9676113360323887,
+      "grad_norm": 0.4366044289293754,
+      "learning_rate": 3.8951817692095966e-05,
+      "loss": 0.4312,
+      "step": 239
+    },
+    {
+      "epoch": 0.97165991902834,
+      "grad_norm": 0.6220811036277273,
+      "learning_rate": 3.893367361386491e-05,
+      "loss": 0.4251,
+      "step": 240
+    },
+    {
+      "epoch": 0.9757085020242915,
+      "grad_norm": 0.4130760028728943,
+      "learning_rate": 3.891537814248417e-05,
+      "loss": 0.4122,
+      "step": 241
+    },
+    {
+      "epoch": 0.979757085020243,
+      "grad_norm": 0.6042233197407489,
+      "learning_rate": 3.889693142424384e-05,
+      "loss": 0.4476,
+      "step": 242
+    },
+    {
+      "epoch": 0.9838056680161943,
+      "grad_norm": 0.46290240010210365,
+      "learning_rate": 3.8878333606643405e-05,
+      "loss": 0.4256,
+      "step": 243
+    },
+    {
+      "epoch": 0.9878542510121457,
+      "grad_norm": 0.4258685621240959,
+      "learning_rate": 3.885958483839049e-05,
+      "loss": 0.4297,
+      "step": 244
+    },
+    {
+      "epoch": 0.9919028340080972,
+      "grad_norm": 0.4606540098627949,
+      "learning_rate": 3.8840685269399786e-05,
+      "loss": 0.4138,
+      "step": 245
+    },
+    {
+      "epoch": 0.9959514170040485,
+      "grad_norm": 0.3881936586524609,
+      "learning_rate": 3.882163505079171e-05,
+      "loss": 0.4454,
+      "step": 246
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.5320606172807268,
+      "learning_rate": 3.880243433489132e-05,
+      "loss": 0.4383,
+      "step": 247
+    },
+    {
+      "epoch": 1.0040485829959513,
+      "grad_norm": 0.4317467035074675,
+      "learning_rate": 3.878308327522702e-05,
+      "loss": 0.3604,
+      "step": 248
+    },
+    {
+      "epoch": 1.008097165991903,
+      "grad_norm": 0.47056533606004813,
+      "learning_rate": 3.876358202652936e-05,
+      "loss": 0.361,
+      "step": 249
+    },
+    {
+      "epoch": 1.0121457489878543,
+      "grad_norm": 0.4560752311054184,
+      "learning_rate": 3.8743930744729835e-05,
+      "loss": 0.3549,
+      "step": 250
+    },
+    {
+      "epoch": 1.0161943319838056,
+      "grad_norm": 0.41723200903288543,
+      "learning_rate": 3.8724129586959535e-05,
+      "loss": 0.3528,
+      "step": 251
+    },
+    {
+      "epoch": 1.0202429149797572,
+      "grad_norm": 0.4842848486186935,
+      "learning_rate": 3.870417871154801e-05,
+      "loss": 0.3799,
+      "step": 252
+    },
+    {
+      "epoch": 1.0242914979757085,
+      "grad_norm": 0.48656259611942065,
+      "learning_rate": 3.868407827802192e-05,
+      "loss": 0.365,
+      "step": 253
+    },
+    {
+      "epoch": 1.0283400809716599,
+      "grad_norm": 0.4383282158178253,
+      "learning_rate": 3.8663828447103804e-05,
+      "loss": 0.3508,
+      "step": 254
+    },
+    {
+      "epoch": 1.0323886639676114,
+      "grad_norm": 0.47335640647619615,
+      "learning_rate": 3.8643429380710765e-05,
+      "loss": 0.3715,
+      "step": 255
+    },
+    {
+      "epoch": 1.0364372469635628,
+      "grad_norm": 0.42065312911333197,
+      "learning_rate": 3.862288124195319e-05,
+      "loss": 0.3723,
+      "step": 256
+    },
+    {
+      "epoch": 1.040485829959514,
+      "grad_norm": 0.4623667493191412,
+      "learning_rate": 3.8602184195133464e-05,
+      "loss": 0.3836,
+      "step": 257
+    },
+    {
+      "epoch": 1.0445344129554657,
+      "grad_norm": 0.42793700878272517,
+      "learning_rate": 3.8581338405744604e-05,
+      "loss": 0.3603,
+      "step": 258
+    },
+    {
+      "epoch": 1.048582995951417,
+      "grad_norm": 0.43353240548004507,
+      "learning_rate": 3.8560344040468996e-05,
+      "loss": 0.3658,
+      "step": 259
+    },
+    {
+      "epoch": 1.0526315789473684,
+      "grad_norm": 0.46340208050772325,
+      "learning_rate": 3.853920126717703e-05,
+      "loss": 0.3608,
+      "step": 260
+    },
+    {
+      "epoch": 1.05668016194332,
+      "grad_norm": 0.49261996547334314,
+      "learning_rate": 3.851791025492574e-05,
+      "loss": 0.3671,
+      "step": 261
+    },
+    {
+      "epoch": 1.0607287449392713,
+      "grad_norm": 0.5379506066918786,
+      "learning_rate": 3.8496471173957515e-05,
+      "loss": 0.3532,
+      "step": 262
+    },
+    {
+      "epoch": 1.0647773279352226,
+      "grad_norm": 0.5010632671658889,
+      "learning_rate": 3.847488419569866e-05,
+      "loss": 0.3813,
+      "step": 263
+    },
+    {
+      "epoch": 1.0688259109311742,
+      "grad_norm": 0.5088713691956003,
+      "learning_rate": 3.8453149492758077e-05,
+      "loss": 0.3559,
+      "step": 264
+    },
+    {
+      "epoch": 1.0728744939271255,
+      "grad_norm": 0.45377479902395707,
+      "learning_rate": 3.8431267238925886e-05,
+      "loss": 0.3797,
+      "step": 265
+    },
+    {
+      "epoch": 1.0769230769230769,
+      "grad_norm": 0.4777444389196963,
+      "learning_rate": 3.840923760917199e-05,
+      "loss": 0.3877,
+      "step": 266
+    },
+    {
+      "epoch": 1.0809716599190284,
+      "grad_norm": 0.47082797859856107,
+      "learning_rate": 3.838706077964473e-05,
+      "loss": 0.3782,
+      "step": 267
+    },
+    {
+      "epoch": 1.0850202429149798,
+      "grad_norm": 0.3993835389579811,
+      "learning_rate": 3.836473692766944e-05,
+      "loss": 0.3665,
+      "step": 268
+    },
+    {
+      "epoch": 1.0890688259109311,
+      "grad_norm": 0.5307289558865997,
+      "learning_rate": 3.834226623174705e-05,
+      "loss": 0.3673,
+      "step": 269
+    },
+    {
+      "epoch": 1.0931174089068827,
+      "grad_norm": 0.43997436243131877,
+      "learning_rate": 3.831964887155264e-05,
+      "loss": 0.3525,
+      "step": 270
+    },
+    {
+      "epoch": 1.097165991902834,
+      "grad_norm": 0.46726268935330784,
+      "learning_rate": 3.829688502793403e-05,
+      "loss": 0.3548,
+      "step": 271
+    },
+    {
+      "epoch": 1.1012145748987854,
+      "grad_norm": 0.44364788973837227,
+      "learning_rate": 3.827397488291029e-05,
+      "loss": 0.3571,
+      "step": 272
+    },
+    {
+      "epoch": 1.1052631578947367,
+      "grad_norm": 0.4171138899334268,
+      "learning_rate": 3.8250918619670343e-05,
+      "loss": 0.3619,
+      "step": 273
+    },
+    {
+      "epoch": 1.1093117408906883,
+      "grad_norm": 0.4988230849584687,
+      "learning_rate": 3.822771642257145e-05,
+      "loss": 0.3861,
+      "step": 274
+    },
+    {
+      "epoch": 1.1133603238866396,
+      "grad_norm": 0.38095917372457866,
+      "learning_rate": 3.820436847713776e-05,
+      "loss": 0.3707,
+      "step": 275
+    },
+    {
+      "epoch": 1.117408906882591,
+      "grad_norm": 0.500645248766726,
+      "learning_rate": 3.8180874970058826e-05,
+      "loss": 0.3695,
+      "step": 276
+    },
+    {
+      "epoch": 1.1214574898785425,
+      "grad_norm": 0.42880779724839624,
+      "learning_rate": 3.8157236089188103e-05,
+      "loss": 0.3579,
+      "step": 277
+    },
+    {
+      "epoch": 1.125506072874494,
+      "grad_norm": 0.4620332837928879,
+      "learning_rate": 3.813345202354145e-05,
+      "loss": 0.3582,
+      "step": 278
+    },
+    {
+      "epoch": 1.1295546558704452,
+      "grad_norm": 0.43531374865274763,
+      "learning_rate": 3.810952296329563e-05,
+      "loss": 0.3806,
+      "step": 279
+    },
+    {
+      "epoch": 1.1336032388663968,
+      "grad_norm": 0.38989396300062973,
+      "learning_rate": 3.808544909978676e-05,
+      "loss": 0.3559,
+      "step": 280
+    },
+    {
+      "epoch": 1.1376518218623481,
+      "grad_norm": 0.47107692305378857,
+      "learning_rate": 3.806123062550882e-05,
+      "loss": 0.3714,
+      "step": 281
+    },
+    {
+      "epoch": 1.1417004048582995,
+      "grad_norm": 0.48872376016104113,
+      "learning_rate": 3.803686773411208e-05,
+      "loss": 0.3449,
+      "step": 282
+    },
+    {
+      "epoch": 1.145748987854251,
+      "grad_norm": 0.4326517564107585,
+      "learning_rate": 3.801236062040158e-05,
+      "loss": 0.3692,
+      "step": 283
+    },
+    {
+      "epoch": 1.1497975708502024,
+      "grad_norm": 0.44362891158914636,
+      "learning_rate": 3.798770948033553e-05,
+      "loss": 0.3544,
+      "step": 284
+    },
+    {
+      "epoch": 1.1538461538461537,
+      "grad_norm": 0.44037020186539383,
+      "learning_rate": 3.796291451102381e-05,
+      "loss": 0.3805,
+      "step": 285
+    },
+    {
+      "epoch": 1.1578947368421053,
+      "grad_norm": 0.41812381655045605,
+      "learning_rate": 3.793797591072631e-05,
+      "loss": 0.3592,
+      "step": 286
+    },
+    {
+      "epoch": 1.1619433198380567,
+      "grad_norm": 0.4370071925368987,
+      "learning_rate": 3.791289387885144e-05,
+      "loss": 0.3741,
+      "step": 287
+    },
+    {
+      "epoch": 1.165991902834008,
+      "grad_norm": 0.4252020122204045,
+      "learning_rate": 3.788766861595443e-05,
+      "loss": 0.3556,
+      "step": 288
+    },
+    {
+      "epoch": 1.1700404858299596,
+      "grad_norm": 0.46991702280641834,
+      "learning_rate": 3.7862300323735835e-05,
+      "loss": 0.3844,
+      "step": 289
+    },
+    {
+      "epoch": 1.174089068825911,
+      "grad_norm": 0.40614665919111437,
+      "learning_rate": 3.783678920503983e-05,
+      "loss": 0.3734,
+      "step": 290
+    },
+    {
+      "epoch": 1.1781376518218623,
+      "grad_norm": 0.38505914246085077,
+      "learning_rate": 3.781113546385265e-05,
+      "loss": 0.3676,
+      "step": 291
+    },
+    {
+      "epoch": 1.1821862348178138,
+      "grad_norm": 0.4183185874002469,
+      "learning_rate": 3.7785339305300916e-05,
+      "loss": 0.3772,
+      "step": 292
+    },
+    {
+      "epoch": 1.1862348178137652,
+      "grad_norm": 0.4220142251060403,
+      "learning_rate": 3.7759400935650025e-05,
+      "loss": 0.3694,
+      "step": 293
+    },
+    {
+      "epoch": 1.1902834008097165,
+      "grad_norm": 0.38095931836175373,
+      "learning_rate": 3.77333205623025e-05,
+      "loss": 0.3744,
+      "step": 294
+    },
+    {
+      "epoch": 1.194331983805668,
+      "grad_norm": 0.36782638315697674,
+      "learning_rate": 3.7707098393796295e-05,
+      "loss": 0.3663,
+      "step": 295
+    },
+    {
+      "epoch": 1.1983805668016194,
+      "grad_norm": 0.4710115223772437,
+      "learning_rate": 3.7680734639803185e-05,
+      "loss": 0.3804,
+      "step": 296
+    },
+    {
+      "epoch": 1.2024291497975708,
+      "grad_norm": 0.4292865033067806,
+      "learning_rate": 3.765422951112704e-05,
+      "loss": 0.3726,
+      "step": 297
+    },
+    {
+      "epoch": 1.2064777327935223,
+      "grad_norm": 0.4433679356256906,
+      "learning_rate": 3.762758321970216e-05,
+      "loss": 0.368,
+      "step": 298
+    },
+    {
+      "epoch": 1.2105263157894737,
+      "grad_norm": 0.4135937344168482,
+      "learning_rate": 3.7600795978591584e-05,
+      "loss": 0.3558,
+      "step": 299
+    },
+    {
+      "epoch": 1.214574898785425,
+      "grad_norm": 0.4025688962123073,
+      "learning_rate": 3.757386800198538e-05,
+      "loss": 0.3898,
+      "step": 300
+    },
+    {
+      "epoch": 1.2186234817813766,
+      "grad_norm": 0.3737379306410504,
+      "learning_rate": 3.7546799505198925e-05,
+      "loss": 0.3555,
+      "step": 301
+    },
+    {
+      "epoch": 1.222672064777328,
+      "grad_norm": 0.3892654431634836,
+      "learning_rate": 3.7519590704671197e-05,
+      "loss": 0.3618,
+      "step": 302
+    },
+    {
+      "epoch": 1.2267206477732793,
+      "grad_norm": 0.37388086101090795,
+      "learning_rate": 3.749224181796305e-05,
+      "loss": 0.3725,
+      "step": 303
+    },
+    {
+      "epoch": 1.2307692307692308,
+      "grad_norm": 0.3583119440103304,
+      "learning_rate": 3.746475306375545e-05,
+      "loss": 0.3729,
+      "step": 304
+    },
+    {
+      "epoch": 1.2348178137651822,
+      "grad_norm": 0.42378728486161815,
+      "learning_rate": 3.743712466184774e-05,
+      "loss": 0.3756,
+      "step": 305
+    },
+    {
+      "epoch": 1.2388663967611335,
+      "grad_norm": 0.40431471030366345,
+      "learning_rate": 3.7409356833155885e-05,
+      "loss": 0.372,
+      "step": 306
+    },
+    {
+      "epoch": 1.242914979757085,
+      "grad_norm": 0.44890455536578605,
+      "learning_rate": 3.73814497997107e-05,
+      "loss": 0.3792,
+      "step": 307
+    },
+    {
+      "epoch": 1.2469635627530364,
+      "grad_norm": 0.3783121740775092,
+      "learning_rate": 3.7353403784656095e-05,
+      "loss": 0.3598,
+      "step": 308
+    },
+    {
+      "epoch": 1.2510121457489878,
+      "grad_norm": 0.44334424874757544,
+      "learning_rate": 3.732521901224724e-05,
+      "loss": 0.3711,
+      "step": 309
+    },
+    {
+      "epoch": 1.2550607287449393,
+      "grad_norm": 0.3756029238353637,
+      "learning_rate": 3.7296895707848844e-05,
+      "loss": 0.3449,
+      "step": 310
+    },
+    {
+      "epoch": 1.2591093117408907,
+      "grad_norm": 0.46224605713985484,
+      "learning_rate": 3.7268434097933275e-05,
+      "loss": 0.3726,
+      "step": 311
+    },
+    {
+      "epoch": 1.263157894736842,
+      "grad_norm": 0.4014978784891467,
+      "learning_rate": 3.723983441007882e-05,
+      "loss": 0.3609,
+      "step": 312
+    },
+    {
+      "epoch": 1.2672064777327936,
+      "grad_norm": 0.42428163626725374,
+      "learning_rate": 3.721109687296781e-05,
+      "loss": 0.3465,
+      "step": 313
+    },
+    {
+      "epoch": 1.271255060728745,
+      "grad_norm": 0.42115579982012485,
+      "learning_rate": 3.718222171638484e-05,
+      "loss": 0.3746,
+      "step": 314
+    },
+    {
+      "epoch": 1.2753036437246963,
+      "grad_norm": 0.42781928834838395,
+      "learning_rate": 3.715320917121488e-05,
+      "loss": 0.3592,
+      "step": 315
+    },
+    {
+      "epoch": 1.2793522267206479,
+      "grad_norm": 0.423575360163584,
+      "learning_rate": 3.712405946944149e-05,
+      "loss": 0.3708,
+      "step": 316
+    },
+    {
+      "epoch": 1.2834008097165992,
+      "grad_norm": 0.48824305365281373,
+      "learning_rate": 3.709477284414488e-05,
+      "loss": 0.3666,
+      "step": 317
+    },
+    {
+      "epoch": 1.2874493927125505,
+      "grad_norm": 0.4092042464413155,
+      "learning_rate": 3.7065349529500136e-05,
+      "loss": 0.3558,
+      "step": 318
+    },
+    {
+      "epoch": 1.291497975708502,
+      "grad_norm": 0.4449685298556577,
+      "learning_rate": 3.7035789760775305e-05,
+      "loss": 0.367,
+      "step": 319
+    },
+    {
+      "epoch": 1.2955465587044535,
+      "grad_norm": 0.4715194880533309,
+      "learning_rate": 3.70060937743295e-05,
+      "loss": 0.3554,
+      "step": 320
+    },
+    {
+      "epoch": 1.2995951417004048,
+      "grad_norm": 0.46215856774956,
+      "learning_rate": 3.6976261807611045e-05,
+      "loss": 0.3672,
+      "step": 321
+    },
+    {
+      "epoch": 1.3036437246963564,
+      "grad_norm": 0.3937927904885563,
+      "learning_rate": 3.694629409915555e-05,
+      "loss": 0.3597,
+      "step": 322
+    },
+    {
+      "epoch": 1.3076923076923077,
+      "grad_norm": 0.532891132275317,
+      "learning_rate": 3.691619088858401e-05,
+      "loss": 0.3781,
+      "step": 323
+    },
+    {
+      "epoch": 1.311740890688259,
+      "grad_norm": 0.38079678167435826,
+      "learning_rate": 3.68859524166009e-05,
+      "loss": 0.3863,
+      "step": 324
+    },
+    {
+      "epoch": 1.3157894736842106,
+      "grad_norm": 0.5688426008201034,
+      "learning_rate": 3.6855578924992226e-05,
+      "loss": 0.3866,
+      "step": 325
+    },
+    {
+      "epoch": 1.319838056680162,
+      "grad_norm": 0.39540843909913725,
+      "learning_rate": 3.6825070656623626e-05,
+      "loss": 0.3615,
+      "step": 326
+    },
+    {
+      "epoch": 1.3238866396761133,
+      "grad_norm": 0.5208803890378622,
+      "learning_rate": 3.67944278554384e-05,
+      "loss": 0.3688,
+      "step": 327
+    },
+    {
+      "epoch": 1.3279352226720649,
+      "grad_norm": 0.4546054901001856,
+      "learning_rate": 3.676365076645557e-05,
+      "loss": 0.3709,
+      "step": 328
+    },
+    {
+      "epoch": 1.3319838056680162,
+      "grad_norm": 0.46521611686303527,
+      "learning_rate": 3.673273963576791e-05,
+      "loss": 0.3377,
+      "step": 329
+    },
+    {
+      "epoch": 1.3360323886639676,
+      "grad_norm": 0.48968747920658723,
+      "learning_rate": 3.670169471054001e-05,
+      "loss": 0.3572,
+      "step": 330
+    },
+    {
+      "epoch": 1.3400809716599191,
+      "grad_norm": 0.38776434143690636,
+      "learning_rate": 3.6670516239006254e-05,
+      "loss": 0.3819,
+      "step": 331
+    },
+    {
+      "epoch": 1.3441295546558705,
+      "grad_norm": 0.45270638866906415,
+      "learning_rate": 3.663920447046888e-05,
+      "loss": 0.3952,
+      "step": 332
+    },
+    {
+      "epoch": 1.3481781376518218,
+      "grad_norm": 0.4025063900559035,
+      "learning_rate": 3.6607759655295954e-05,
+      "loss": 0.3488,
+      "step": 333
+    },
+    {
+      "epoch": 1.3522267206477734,
+      "grad_norm": 0.4459319977413574,
+      "learning_rate": 3.657618204491937e-05,
+      "loss": 0.3798,
+      "step": 334
+    },
+    {
+      "epoch": 1.3562753036437247,
+      "grad_norm": 0.5243668972028495,
+      "learning_rate": 3.6544471891832875e-05,
+      "loss": 0.357,
+      "step": 335
+    },
+    {
+      "epoch": 1.360323886639676,
+      "grad_norm": 0.39276339171247865,
+      "learning_rate": 3.6512629449589983e-05,
+      "loss": 0.3668,
+      "step": 336
+    },
+    {
+      "epoch": 1.3643724696356276,
+      "grad_norm": 0.4775479269157366,
+      "learning_rate": 3.6480654972802045e-05,
+      "loss": 0.3704,
+      "step": 337
+    },
+    {
+      "epoch": 1.368421052631579,
+      "grad_norm": 0.42160650298866686,
+      "learning_rate": 3.644854871713611e-05,
+      "loss": 0.362,
+      "step": 338
+    },
+    {
+      "epoch": 1.3724696356275303,
+      "grad_norm": 0.4063779636208966,
+      "learning_rate": 3.6416310939312924e-05,
+      "loss": 0.3607,
+      "step": 339
+    },
+    {
+      "epoch": 1.376518218623482,
+      "grad_norm": 0.4505741299483498,
+      "learning_rate": 3.638394189710493e-05,
+      "loss": 0.3827,
+      "step": 340
+    },
+    {
+      "epoch": 1.3805668016194332,
+      "grad_norm": 0.5069475271069037,
+      "learning_rate": 3.635144184933412e-05,
+      "loss": 0.3909,
+      "step": 341
+    },
+    {
+      "epoch": 1.3846153846153846,
+      "grad_norm": 0.4000310493158753,
+      "learning_rate": 3.6318811055869994e-05,
+      "loss": 0.3647,
+      "step": 342
+    },
+    {
+      "epoch": 1.3886639676113361,
+      "grad_norm": 0.43187336783458435,
+      "learning_rate": 3.628604977762752e-05,
+      "loss": 0.3716,
+      "step": 343
+    },
+    {
+      "epoch": 1.3927125506072875,
+      "grad_norm": 0.44118110757545786,
+      "learning_rate": 3.6253158276565004e-05,
+      "loss": 0.3836,
+      "step": 344
+    },
+    {
+      "epoch": 1.3967611336032388,
+      "grad_norm": 0.34406214290785275,
+      "learning_rate": 3.6220136815682016e-05,
+      "loss": 0.3521,
+      "step": 345
+    },
+    {
+      "epoch": 1.4008097165991904,
+      "grad_norm": 0.3998542989739769,
+      "learning_rate": 3.618698565901727e-05,
+      "loss": 0.3595,
+      "step": 346
+    },
+    {
+      "epoch": 1.4048582995951417,
+      "grad_norm": 0.37938565717817246,
+      "learning_rate": 3.615370507164654e-05,
+      "loss": 0.3864,
+      "step": 347
+    },
+    {
+      "epoch": 1.408906882591093,
+      "grad_norm": 0.4248928228089392,
+      "learning_rate": 3.61202953196805e-05,
+      "loss": 0.3478,
+      "step": 348
+    },
+    {
+      "epoch": 1.4129554655870447,
+      "grad_norm": 0.4080545378213579,
+      "learning_rate": 3.6086756670262654e-05,
+      "loss": 0.3575,
+      "step": 349
+    },
+    {
+      "epoch": 1.417004048582996,
+      "grad_norm": 0.40179890266043483,
+      "learning_rate": 3.6053089391567135e-05,
+      "loss": 0.3768,
+      "step": 350
+    },
+    {
+      "epoch": 1.4210526315789473,
+      "grad_norm": 0.3687416390452529,
+      "learning_rate": 3.601929375279662e-05,
+      "loss": 0.3445,
+      "step": 351
+    },
+    {
+      "epoch": 1.425101214574899,
+      "grad_norm": 0.405895759366919,
+      "learning_rate": 3.598537002418012e-05,
+      "loss": 0.3325,
+      "step": 352
+    },
+    {
+      "epoch": 1.4291497975708503,
+      "grad_norm": 0.4151686442688146,
+      "learning_rate": 3.5951318476970885e-05,
+      "loss": 0.3755,
+      "step": 353
+    },
+    {
+      "epoch": 1.4331983805668016,
+      "grad_norm": 0.42275190350984554,
+      "learning_rate": 3.591713938344416e-05,
+      "loss": 0.3591,
+      "step": 354
+    },
+    {
+      "epoch": 1.4372469635627532,
+      "grad_norm": 0.45579810717874886,
+      "learning_rate": 3.588283301689507e-05,
+      "loss": 0.3637,
+      "step": 355
+    },
+    {
+      "epoch": 1.4412955465587045,
+      "grad_norm": 0.40638925588636476,
+      "learning_rate": 3.5848399651636424e-05,
+      "loss": 0.3831,
+      "step": 356
+    },
+    {
+      "epoch": 1.4453441295546559,
+      "grad_norm": 0.4579956606470859,
+      "learning_rate": 3.5813839562996473e-05,
+      "loss": 0.3485,
+      "step": 357
+    },
+    {
+      "epoch": 1.4493927125506074,
+      "grad_norm": 0.4057724149806602,
+      "learning_rate": 3.5779153027316775e-05,
+      "loss": 0.3642,
+      "step": 358
+    },
+    {
+      "epoch": 1.4534412955465588,
+      "grad_norm": 0.4542913306587732,
+      "learning_rate": 3.574434032194994e-05,
+      "loss": 0.3826,
+      "step": 359
+    },
+    {
+      "epoch": 1.45748987854251,
+      "grad_norm": 0.48054551804384704,
+      "learning_rate": 3.5709401725257436e-05,
+      "loss": 0.3572,
+      "step": 360
+    },
+    {
+      "epoch": 1.4615384615384617,
+      "grad_norm": 0.46155432186647144,
+      "learning_rate": 3.5674337516607346e-05,
+      "loss": 0.3822,
+      "step": 361
+    },
+    {
+      "epoch": 1.465587044534413,
+      "grad_norm": 0.459087827496654,
+      "learning_rate": 3.563914797637216e-05,
+      "loss": 0.3834,
+      "step": 362
+    },
+    {
+      "epoch": 1.4696356275303644,
+      "grad_norm": 0.3953808969398796,
+      "learning_rate": 3.560383338592649e-05,
+      "loss": 0.3622,
+      "step": 363
+    },
+    {
+      "epoch": 1.4736842105263157,
+      "grad_norm": 0.41387046469296307,
+      "learning_rate": 3.556839402764487e-05,
+      "loss": 0.3863,
+      "step": 364
+    },
+    {
+      "epoch": 1.4777327935222673,
+      "grad_norm": 0.37303506115525487,
+      "learning_rate": 3.553283018489946e-05,
+      "loss": 0.3791,
+      "step": 365
+    },
+    {
+      "epoch": 1.4817813765182186,
+      "grad_norm": 0.40279312471577583,
+      "learning_rate": 3.54971421420578e-05,
+      "loss": 0.3798,
+      "step": 366
+    },
+    {
+      "epoch": 1.48582995951417,
+      "grad_norm": 0.35695913612606506,
+      "learning_rate": 3.5461330184480524e-05,
+      "loss": 0.3804,
+      "step": 367
+    },
+    {
+      "epoch": 1.4898785425101215,
+      "grad_norm": 0.4467096755732761,
+      "learning_rate": 3.54253945985191e-05,
+      "loss": 0.3806,
+      "step": 368
+    },
+    {
+      "epoch": 1.4939271255060729,
+      "grad_norm": 0.42203535722912433,
+      "learning_rate": 3.5389335671513524e-05,
+      "loss": 0.3636,
+      "step": 369
+    },
+    {
+      "epoch": 1.4979757085020242,
+      "grad_norm": 0.4549894595118823,
+      "learning_rate": 3.5353153691789986e-05,
+      "loss": 0.3602,
+      "step": 370
+    },
+    {
+      "epoch": 1.5020242914979756,
+      "grad_norm": 0.4919352976367366,
+      "learning_rate": 3.531684894865866e-05,
+      "loss": 0.3681,
+      "step": 371
+    },
+    {
+      "epoch": 1.5060728744939271,
+      "grad_norm": 0.35602875384649957,
+      "learning_rate": 3.528042173241128e-05,
+      "loss": 0.3768,
+      "step": 372
+    },
+    {
+      "epoch": 1.5101214574898787,
+      "grad_norm": 0.5287619763845682,
+      "learning_rate": 3.524387233431892e-05,
+      "loss": 0.3445,
+      "step": 373
+    },
+    {
+      "epoch": 1.5141700404858298,
+      "grad_norm": 0.3479493766401925,
+      "learning_rate": 3.520720104662958e-05,
+      "loss": 0.3819,
+      "step": 374
+    },
+    {
+      "epoch": 1.5182186234817814,
+      "grad_norm": 0.5077375373624382,
+      "learning_rate": 3.5170408162565904e-05,
+      "loss": 0.3574,
+      "step": 375
+    },
+    {
+      "epoch": 1.522267206477733,
+      "grad_norm": 0.44341475274753184,
+      "learning_rate": 3.513349397632281e-05,
+      "loss": 0.3884,
+      "step": 376
+    },
+    {
+      "epoch": 1.526315789473684,
+      "grad_norm": 0.4211823326273393,
+      "learning_rate": 3.5096458783065145e-05,
+      "loss": 0.3914,
+      "step": 377
+    },
+    {
+      "epoch": 1.5303643724696356,
+      "grad_norm": 0.47489063394452224,
+      "learning_rate": 3.505930287892533e-05,
+      "loss": 0.3836,
+      "step": 378
+    },
+    {
+      "epoch": 1.5344129554655872,
+      "grad_norm": 0.33782364832604367,
+      "learning_rate": 3.502202656100099e-05,
+      "loss": 0.369,
+      "step": 379
+    },
+    {
+      "epoch": 1.5384615384615383,
+      "grad_norm": 0.4603922421414736,
+      "learning_rate": 3.498463012735256e-05,
+      "loss": 0.3853,
+      "step": 380
+    },
+    {
+      "epoch": 1.54251012145749,
+      "grad_norm": 0.3771365848217366,
+      "learning_rate": 3.494711387700094e-05,
+      "loss": 0.3741,
+      "step": 381
+    },
+    {
+      "epoch": 1.5465587044534415,
+      "grad_norm": 0.41687597356134254,
+      "learning_rate": 3.490947810992505e-05,
+      "loss": 0.3689,
+      "step": 382
+    },
+    {
+      "epoch": 1.5506072874493926,
+      "grad_norm": 0.38923344071630706,
+      "learning_rate": 3.48717231270595e-05,
+      "loss": 0.3623,
+      "step": 383
+    },
+    {
+      "epoch": 1.5546558704453441,
+      "grad_norm": 0.3990350261559954,
+      "learning_rate": 3.483384923029211e-05,
+      "loss": 0.3695,
+      "step": 384
+    },
+    {
+      "epoch": 1.5587044534412957,
+      "grad_norm": 0.4080496467546424,
+      "learning_rate": 3.479585672246156e-05,
+      "loss": 0.3658,
+      "step": 385
+    },
+    {
+      "epoch": 1.5627530364372468,
+      "grad_norm": 0.37272958083853913,
+      "learning_rate": 3.4757745907354924e-05,
+      "loss": 0.3559,
+      "step": 386
+    },
+    {
+      "epoch": 1.5668016194331984,
+      "grad_norm": 0.39642520107073614,
+      "learning_rate": 3.471951708970527e-05,
+      "loss": 0.3667,
+      "step": 387
+    },
+    {
+      "epoch": 1.5708502024291497,
+      "grad_norm": 0.3737650969194724,
+      "learning_rate": 3.468117057518921e-05,
+      "loss": 0.378,
+      "step": 388
+    },
+    {
+      "epoch": 1.574898785425101,
+      "grad_norm": 0.3973164411688128,
+      "learning_rate": 3.4642706670424426e-05,
+      "loss": 0.364,
+      "step": 389
+    },
+    {
+      "epoch": 1.5789473684210527,
+      "grad_norm": 0.40706285425600347,
+      "learning_rate": 3.460412568296731e-05,
+      "loss": 0.3734,
+      "step": 390
+    },
+    {
+      "epoch": 1.582995951417004,
+      "grad_norm": 0.40225657090445155,
+      "learning_rate": 3.456542792131039e-05,
+      "loss": 0.381,
+      "step": 391
+    },
+    {
+      "epoch": 1.5870445344129553,
+      "grad_norm": 0.4116436876256783,
+      "learning_rate": 3.4526613694879936e-05,
+      "loss": 0.3681,
+      "step": 392
+    },
+    {
+      "epoch": 1.591093117408907,
+      "grad_norm": 0.44730789975476354,
+      "learning_rate": 3.448768331403348e-05,
+      "loss": 0.3647,
+      "step": 393
+    },
+    {
+      "epoch": 1.5951417004048583,
+      "grad_norm": 0.41482432225800603,
+      "learning_rate": 3.44486370900573e-05,
+      "loss": 0.3872,
+      "step": 394
+    },
+    {
+      "epoch": 1.5991902834008096,
+      "grad_norm": 0.4224321134627349,
+      "learning_rate": 3.440947533516396e-05,
+      "loss": 0.3814,
+      "step": 395
+    },
+    {
+      "epoch": 1.6032388663967612,
+      "grad_norm": 0.3842650033450959,
+      "learning_rate": 3.4370198362489816e-05,
+      "loss": 0.366,
+      "step": 396
+    },
+    {
+      "epoch": 1.6072874493927125,
+      "grad_norm": 0.36060724821901935,
+      "learning_rate": 3.43308064860925e-05,
+      "loss": 0.3531,
+      "step": 397
+    },
+    {
+      "epoch": 1.6113360323886639,
+      "grad_norm": 0.42739520995952685,
+      "learning_rate": 3.4291300020948393e-05,
+      "loss": 0.377,
+      "step": 398
+    },
+    {
+      "epoch": 1.6153846153846154,
+      "grad_norm": 0.3468902186288037,
+      "learning_rate": 3.425167928295015e-05,
+      "loss": 0.3737,
+      "step": 399
+    },
+    {
+      "epoch": 1.6194331983805668,
+      "grad_norm": 0.4567412075228189,
+      "learning_rate": 3.421194458890411e-05,
+      "loss": 0.3546,
+      "step": 400
+    },
+    {
+      "epoch": 1.623481781376518,
+      "grad_norm": 0.39791192143968934,
+      "learning_rate": 3.4172096256527865e-05,
+      "loss": 0.3905,
+      "step": 401
+    },
+    {
+      "epoch": 1.6275303643724697,
+      "grad_norm": 0.3638895335968851,
+      "learning_rate": 3.4132134604447595e-05,
+      "loss": 0.3687,
+      "step": 402
+    },
+    {
+      "epoch": 1.631578947368421,
+      "grad_norm": 0.3762040889947907,
+      "learning_rate": 3.409205995219561e-05,
+      "loss": 0.3574,
+      "step": 403
+    },
+    {
+      "epoch": 1.6356275303643724,
+      "grad_norm": 0.36350552397661384,
+      "learning_rate": 3.4051872620207765e-05,
+      "loss": 0.3539,
+      "step": 404
+    },
+    {
+      "epoch": 1.639676113360324,
+      "grad_norm": 0.3183831741845232,
+      "learning_rate": 3.40115729298209e-05,
+      "loss": 0.3523,
+      "step": 405
+    },
+    {
+      "epoch": 1.6437246963562753,
+      "grad_norm": 0.337619990380765,
+      "learning_rate": 3.397116120327025e-05,
+      "loss": 0.3716,
+      "step": 406
+    },
+    {
+      "epoch": 1.6477732793522266,
+      "grad_norm": 0.365619986782008,
+      "learning_rate": 3.3930637763686916e-05,
+      "loss": 0.3705,
+      "step": 407
+    },
+    {
+      "epoch": 1.6518218623481782,
+      "grad_norm": 0.34916873763266354,
+      "learning_rate": 3.389000293509524e-05,
+      "loss": 0.3677,
+      "step": 408
+    },
+    {
+      "epoch": 1.6558704453441295,
+      "grad_norm": 0.36778856104340335,
+      "learning_rate": 3.3849257042410225e-05,
+      "loss": 0.3476,
+      "step": 409
+    },
+    {
+      "epoch": 1.6599190283400809,
+      "grad_norm": 0.38337457303570543,
+      "learning_rate": 3.380840041143494e-05,
+      "loss": 0.3697,
+      "step": 410
+    },
+    {
+      "epoch": 1.6639676113360324,
+      "grad_norm": 0.3699757678694376,
+      "learning_rate": 3.376743336885792e-05,
+      "loss": 0.3615,
+      "step": 411
+    },
+    {
+      "epoch": 1.6680161943319838,
+      "grad_norm": 0.3658774063903156,
+      "learning_rate": 3.372635624225055e-05,
+      "loss": 0.3717,
+      "step": 412
+    },
+    {
+      "epoch": 1.6720647773279351,
+      "grad_norm": 0.3538847919225455,
+      "learning_rate": 3.3685169360064445e-05,
+      "loss": 0.3695,
+      "step": 413
+    },
+    {
+      "epoch": 1.6761133603238867,
+      "grad_norm": 0.3638544739278751,
+      "learning_rate": 3.3643873051628804e-05,
+      "loss": 0.362,
+      "step": 414
+    },
+    {
+      "epoch": 1.680161943319838,
+      "grad_norm": 0.3124825648915713,
+      "learning_rate": 3.3602467647147814e-05,
+      "loss": 0.3751,
+      "step": 415
+    },
+    {
+      "epoch": 1.6842105263157894,
+      "grad_norm": 0.33438834760805236,
+      "learning_rate": 3.356095347769799e-05,
+      "loss": 0.3736,
+      "step": 416
+    },
+    {
+      "epoch": 1.688259109311741,
+      "grad_norm": 0.36463263097741816,
+      "learning_rate": 3.351933087522552e-05,
+      "loss": 0.3704,
+      "step": 417
+    },
+    {
+      "epoch": 1.6923076923076923,
+      "grad_norm": 0.3724323689942102,
+      "learning_rate": 3.3477600172543626e-05,
+      "loss": 0.3773,
+      "step": 418
+    },
+    {
+      "epoch": 1.6963562753036436,
+      "grad_norm": 0.36795242334486217,
+      "learning_rate": 3.3435761703329894e-05,
+      "loss": 0.3875,
+      "step": 419
+    },
+    {
+      "epoch": 1.7004048582995952,
+      "grad_norm": 0.3903190366042933,
+      "learning_rate": 3.3393815802123606e-05,
+      "loss": 0.3884,
+      "step": 420
+    },
+    {
+      "epoch": 1.7044534412955465,
+      "grad_norm": 0.3834690906461387,
+      "learning_rate": 3.335176280432308e-05,
+      "loss": 0.3706,
+      "step": 421
+    },
+    {
+      "epoch": 1.708502024291498,
+      "grad_norm": 0.36382219517025893,
+      "learning_rate": 3.330960304618295e-05,
+      "loss": 0.3925,
+      "step": 422
+    },
+    {
+      "epoch": 1.7125506072874495,
+      "grad_norm": 0.3831345203424723,
+      "learning_rate": 3.326733686481153e-05,
+      "loss": 0.3849,
+      "step": 423
+    },
+    {
+      "epoch": 1.7165991902834008,
+      "grad_norm": 0.37823625692272006,
+      "learning_rate": 3.3224964598168084e-05,
+      "loss": 0.358,
+      "step": 424
+    },
+    {
+      "epoch": 1.7206477732793521,
+      "grad_norm": 0.42088971539100434,
+      "learning_rate": 3.318248658506012e-05,
+      "loss": 0.3625,
+      "step": 425
+    },
+    {
+      "epoch": 1.7246963562753037,
+      "grad_norm": 0.4021970461737732,
+      "learning_rate": 3.313990316514069e-05,
+      "loss": 0.3773,
+      "step": 426
+    },
+    {
+      "epoch": 1.728744939271255,
+      "grad_norm": 0.42976290888537794,
+      "learning_rate": 3.309721467890571e-05,
+      "loss": 0.3911,
+      "step": 427
+    },
+    {
+      "epoch": 1.7327935222672064,
+      "grad_norm": 0.40573442614534216,
+      "learning_rate": 3.305442146769115e-05,
+      "loss": 0.3731,
+      "step": 428
+    },
+    {
+      "epoch": 1.736842105263158,
+      "grad_norm": 0.4087385612475439,
+      "learning_rate": 3.30115238736704e-05,
+      "loss": 0.3556,
+      "step": 429
+    },
+    {
+      "epoch": 1.7408906882591093,
+      "grad_norm": 0.33911119281220664,
+      "learning_rate": 3.296852223985148e-05,
+      "loss": 0.3601,
+      "step": 430
+    },
+    {
+      "epoch": 1.7449392712550607,
+      "grad_norm": 0.3771330439846614,
+      "learning_rate": 3.2925416910074315e-05,
+      "loss": 0.3703,
+      "step": 431
+    },
+    {
+      "epoch": 1.7489878542510122,
+      "grad_norm": 0.33409322567964883,
+      "learning_rate": 3.288220822900796e-05,
+      "loss": 0.3488,
+      "step": 432
+    },
+    {
+      "epoch": 1.7530364372469636,
+      "grad_norm": 0.35791809890487847,
+      "learning_rate": 3.283889654214788e-05,
+      "loss": 0.3529,
+      "step": 433
+    },
+    {
+      "epoch": 1.757085020242915,
+      "grad_norm": 0.3773614721839675,
+      "learning_rate": 3.279548219581317e-05,
+      "loss": 0.3635,
+      "step": 434
+    },
+    {
+      "epoch": 1.7611336032388665,
+      "grad_norm": 0.38152470925864174,
+      "learning_rate": 3.275196553714379e-05,
+      "loss": 0.3827,
+      "step": 435
+    },
+    {
+      "epoch": 1.7651821862348178,
+      "grad_norm": 0.41101483282991214,
+      "learning_rate": 3.270834691409778e-05,
+      "loss": 0.3675,
+      "step": 436
+    },
+    {
+      "epoch": 1.7692307692307692,
+      "grad_norm": 0.3792417783931052,
+      "learning_rate": 3.266462667544849e-05,
+      "loss": 0.3757,
+      "step": 437
+    },
+    {
+      "epoch": 1.7732793522267207,
+      "grad_norm": 0.3742092869863632,
+      "learning_rate": 3.262080517078178e-05,
+      "loss": 0.3873,
+      "step": 438
+    },
+    {
+      "epoch": 1.777327935222672,
+      "grad_norm": 0.380350952713812,
+      "learning_rate": 3.257688275049323e-05,
+      "loss": 0.351,
+      "step": 439
+    },
+    {
+      "epoch": 1.7813765182186234,
+      "grad_norm": 0.37043341093728377,
+      "learning_rate": 3.253285976578535e-05,
+      "loss": 0.3595,
+      "step": 440
+    },
+    {
+      "epoch": 1.785425101214575,
+      "grad_norm": 0.3654133362240301,
+      "learning_rate": 3.2488736568664756e-05,
+      "loss": 0.3612,
+      "step": 441
+    },
+    {
+      "epoch": 1.7894736842105263,
+      "grad_norm": 0.35845680845794753,
+      "learning_rate": 3.244451351193935e-05,
+      "loss": 0.3806,
+      "step": 442
+    },
+    {
+      "epoch": 1.7935222672064777,
+      "grad_norm": 0.3885342990601441,
+      "learning_rate": 3.240019094921551e-05,
+      "loss": 0.3712,
+      "step": 443
+    },
+    {
+      "epoch": 1.7975708502024292,
+      "grad_norm": 0.3891445815748651,
+      "learning_rate": 3.235576923489527e-05,
+      "loss": 0.384,
+      "step": 444
+    },
+    {
+      "epoch": 1.8016194331983806,
+      "grad_norm": 0.37235110282911055,
+      "learning_rate": 3.2311248724173476e-05,
+      "loss": 0.3598,
+      "step": 445
+    },
+    {
+      "epoch": 1.805668016194332,
+      "grad_norm": 0.4218588633550091,
+      "learning_rate": 3.226662977303494e-05,
+      "loss": 0.37,
+      "step": 446
+    },
+    {
+      "epoch": 1.8097165991902835,
+      "grad_norm": 0.4212906978236678,
+      "learning_rate": 3.2221912738251626e-05,
+      "loss": 0.3739,
+      "step": 447
+    },
+    {
+      "epoch": 1.8137651821862348,
+      "grad_norm": 0.4763800503092454,
+      "learning_rate": 3.2177097977379736e-05,
+      "loss": 0.3689,
+      "step": 448
+    },
+    {
+      "epoch": 1.8178137651821862,
+      "grad_norm": 0.510878863142854,
+      "learning_rate": 3.21321858487569e-05,
+      "loss": 0.3977,
+      "step": 449
+    },
+    {
+      "epoch": 1.8218623481781377,
+      "grad_norm": 0.4760436687982531,
+      "learning_rate": 3.208717671149931e-05,
+      "loss": 0.3771,
+      "step": 450
+    },
+    {
+      "epoch": 1.825910931174089,
+      "grad_norm": 0.43400337213203355,
+      "learning_rate": 3.204207092549882e-05,
+      "loss": 0.3661,
+      "step": 451
+    },
+    {
+      "epoch": 1.8299595141700404,
+      "grad_norm": 0.4944477178985351,
+      "learning_rate": 3.19968688514201e-05,
+      "loss": 0.361,
+      "step": 452
+    },
+    {
+      "epoch": 1.834008097165992,
+      "grad_norm": 0.37665123620001456,
+      "learning_rate": 3.195157085069774e-05,
+      "loss": 0.3584,
+      "step": 453
+    },
+    {
+      "epoch": 1.8380566801619433,
+      "grad_norm": 0.42334197045382577,
+      "learning_rate": 3.190617728553332e-05,
+      "loss": 0.3711,
+      "step": 454
+    },
+    {
+      "epoch": 1.8421052631578947,
+      "grad_norm": 0.3375871765205956,
+      "learning_rate": 3.1860688518892606e-05,
+      "loss": 0.3523,
+      "step": 455
+    },
+    {
+      "epoch": 1.8461538461538463,
+      "grad_norm": 0.46099885761247894,
+      "learning_rate": 3.181510491450254e-05,
+      "loss": 0.3873,
+      "step": 456
+    },
+    {
+      "epoch": 1.8502024291497976,
+      "grad_norm": 0.34046152170897603,
+      "learning_rate": 3.176942683684842e-05,
+      "loss": 0.3463,
+      "step": 457
+    },
+    {
+      "epoch": 1.854251012145749,
+      "grad_norm": 0.4311345526534373,
+      "learning_rate": 3.1723654651170934e-05,
+      "loss": 0.37,
+      "step": 458
+    },
+    {
+      "epoch": 1.8582995951417005,
+      "grad_norm": 0.34879026964779625,
+      "learning_rate": 3.167778872346327e-05,
+      "loss": 0.3434,
+      "step": 459
+    },
+    {
+      "epoch": 1.8623481781376519,
+      "grad_norm": 0.4041363566718014,
+      "learning_rate": 3.163182942046815e-05,
+      "loss": 0.358,
+      "step": 460
+    },
+    {
+      "epoch": 1.8663967611336032,
+      "grad_norm": 0.3752973106174396,
+      "learning_rate": 3.158577710967494e-05,
+      "loss": 0.3557,
+      "step": 461
+    },
+    {
+      "epoch": 1.8704453441295548,
+      "grad_norm": 0.34677397337545457,
+      "learning_rate": 3.153963215931669e-05,
+      "loss": 0.3703,
+      "step": 462
+    },
+    {
+      "epoch": 1.874493927125506,
+      "grad_norm": 0.38415091982874133,
+      "learning_rate": 3.14933949383672e-05,
+      "loss": 0.3665,
+      "step": 463
+    },
+    {
+      "epoch": 1.8785425101214575,
+      "grad_norm": 0.3175694716985712,
+      "learning_rate": 3.144706581653805e-05,
+      "loss": 0.3755,
+      "step": 464
+    },
+    {
+      "epoch": 1.882591093117409,
+      "grad_norm": 0.3482774224126065,
+      "learning_rate": 3.140064516427566e-05,
+      "loss": 0.3787,
+      "step": 465
+    },
+    {
+      "epoch": 1.8866396761133604,
+      "grad_norm": 0.3403662684034545,
+      "learning_rate": 3.135413335275833e-05,
+      "loss": 0.3654,
+      "step": 466
+    },
+    {
+      "epoch": 1.8906882591093117,
+      "grad_norm": 0.3240125018587614,
+      "learning_rate": 3.130753075389327e-05,
+      "loss": 0.3708,
+      "step": 467
+    },
+    {
+      "epoch": 1.8947368421052633,
+      "grad_norm": 0.3677975257976491,
+      "learning_rate": 3.1260837740313625e-05,
+      "loss": 0.3595,
+      "step": 468
+    },
+    {
+      "epoch": 1.8987854251012146,
+      "grad_norm": 0.3182515692559823,
+      "learning_rate": 3.121405468537547e-05,
+      "loss": 0.3521,
+      "step": 469
+    },
+    {
+      "epoch": 1.902834008097166,
+      "grad_norm": 0.3834043827894139,
+      "learning_rate": 3.116718196315488e-05,
+      "loss": 0.3544,
+      "step": 470
+    },
+    {
+      "epoch": 1.9068825910931175,
+      "grad_norm": 0.34077481550219035,
+      "learning_rate": 3.1120219948444894e-05,
+      "loss": 0.3784,
+      "step": 471
+    },
+    {
+      "epoch": 1.9109311740890689,
+      "grad_norm": 0.35899682456263976,
+      "learning_rate": 3.107316901675252e-05,
+      "loss": 0.3792,
+      "step": 472
+    },
+    {
+      "epoch": 1.9149797570850202,
+      "grad_norm": 0.3293814863068676,
+      "learning_rate": 3.1026029544295745e-05,
+      "loss": 0.3718,
+      "step": 473
+    },
+    {
+      "epoch": 1.9190283400809718,
+      "grad_norm": 0.398279709488126,
+      "learning_rate": 3.097880190800057e-05,
+      "loss": 0.3613,
+      "step": 474
+    },
+    {
+      "epoch": 1.9230769230769231,
+      "grad_norm": 0.3303075579225027,
+      "learning_rate": 3.093148648549788e-05,
+      "loss": 0.3529,
+      "step": 475
+    },
+    {
+      "epoch": 1.9271255060728745,
+      "grad_norm": 0.39595695619392146,
+      "learning_rate": 3.088408365512055e-05,
+      "loss": 0.3672,
+      "step": 476
+    },
+    {
+      "epoch": 1.931174089068826,
+      "grad_norm": 0.3747937731107122,
+      "learning_rate": 3.083659379590034e-05,
+      "loss": 0.3904,
+      "step": 477
+    },
+    {
+      "epoch": 1.9352226720647774,
+      "grad_norm": 0.3533258553890501,
+      "learning_rate": 3.0789017287564913e-05,
+      "loss": 0.3591,
+      "step": 478
+    },
+    {
+      "epoch": 1.9392712550607287,
+      "grad_norm": 0.37748019141775524,
+      "learning_rate": 3.074135451053475e-05,
+      "loss": 0.3979,
+      "step": 479
+    },
+    {
+      "epoch": 1.9433198380566803,
+      "grad_norm": 0.3538604688222132,
+      "learning_rate": 3.069360584592015e-05,
+      "loss": 0.3581,
+      "step": 480
+    },
+    {
+      "epoch": 1.9473684210526314,
+      "grad_norm": 0.37787361463008456,
+      "learning_rate": 3.064577167551816e-05,
+      "loss": 0.351,
+      "step": 481
+    },
+    {
+      "epoch": 1.951417004048583,
+      "grad_norm": 0.3646444104893767,
+      "learning_rate": 3.059785238180953e-05,
+      "loss": 0.3759,
+      "step": 482
+    },
+    {
+      "epoch": 1.9554655870445345,
+      "grad_norm": 0.3741801251347151,
+      "learning_rate": 3.054984834795566e-05,
+      "loss": 0.3618,
+      "step": 483
+    },
+    {
+      "epoch": 1.9595141700404857,
+      "grad_norm": 0.4065627324471616,
+      "learning_rate": 3.0501759957795537e-05,
+      "loss": 0.3465,
+      "step": 484
+    },
+    {
+      "epoch": 1.9635627530364372,
+      "grad_norm": 0.36726436573305865,
+      "learning_rate": 3.0453587595842627e-05,
+      "loss": 0.3598,
+      "step": 485
+    },
+    {
+      "epoch": 1.9676113360323888,
+      "grad_norm": 0.35541638538055786,
+      "learning_rate": 3.0405331647281856e-05,
+      "loss": 0.3507,
+      "step": 486
+    },
+    {
+      "epoch": 1.97165991902834,
+      "grad_norm": 0.33340136157076966,
+      "learning_rate": 3.0356992497966508e-05,
+      "loss": 0.3654,
+      "step": 487
+    },
+    {
+      "epoch": 1.9757085020242915,
+      "grad_norm": 0.35199964291913893,
+      "learning_rate": 3.0308570534415137e-05,
+      "loss": 0.3719,
+      "step": 488
+    },
+    {
+      "epoch": 1.979757085020243,
+      "grad_norm": 0.33716853878524844,
+      "learning_rate": 3.0260066143808467e-05,
+      "loss": 0.3632,
+      "step": 489
+    },
+    {
+      "epoch": 1.9838056680161942,
+      "grad_norm": 0.37196867968164393,
+      "learning_rate": 3.021147971398632e-05,
+      "loss": 0.3678,
+      "step": 490
+    },
+    {
+      "epoch": 1.9878542510121457,
+      "grad_norm": 0.34979986540384805,
+      "learning_rate": 3.0162811633444478e-05,
+      "loss": 0.373,
+      "step": 491
+    },
+    {
+      "epoch": 1.9919028340080973,
+      "grad_norm": 0.35278128698807615,
+      "learning_rate": 3.0114062291331635e-05,
+      "loss": 0.3597,
+      "step": 492
+    },
+    {
+      "epoch": 1.9959514170040484,
+      "grad_norm": 0.33746197461428196,
+      "learning_rate": 3.0065232077446217e-05,
+      "loss": 0.3609,
+      "step": 493
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.3252634930157202,
+      "learning_rate": 3.001632138223332e-05,
+      "loss": 0.3632,
+      "step": 494
+    },
+    {
+      "epoch": 2.0040485829959516,
+      "grad_norm": 0.5050374144309877,
+      "learning_rate": 2.9967330596781565e-05,
+      "loss": 0.2872,
+      "step": 495
+    },
+    {
+      "epoch": 2.0080971659919027,
+      "grad_norm": 0.6161418665716476,
+      "learning_rate": 2.9918260112819952e-05,
+      "loss": 0.2874,
+      "step": 496
+    },
+    {
+      "epoch": 2.0121457489878543,
+      "grad_norm": 0.39427763753679795,
+      "learning_rate": 2.9869110322714778e-05,
+      "loss": 0.2797,
+      "step": 497
+    },
+    {
+      "epoch": 2.016194331983806,
+      "grad_norm": 0.44345210698179327,
+      "learning_rate": 2.9819881619466447e-05,
+      "loss": 0.2842,
+      "step": 498
+    },
+    {
+      "epoch": 2.020242914979757,
+      "grad_norm": 0.3861260943883524,
+      "learning_rate": 2.9770574396706364e-05,
+      "loss": 0.2714,
+      "step": 499
+    },
+    {
+      "epoch": 2.0242914979757085,
+      "grad_norm": 0.44304533975870797,
+      "learning_rate": 2.972118904869376e-05,
+      "loss": 0.2576,
+      "step": 500
+    },
+    {
+      "epoch": 2.02834008097166,
+      "grad_norm": 0.3891924829362501,
+      "learning_rate": 2.9671725970312563e-05,
+      "loss": 0.2677,
+      "step": 501
+    },
+    {
+      "epoch": 2.032388663967611,
+      "grad_norm": 0.484415319229292,
+      "learning_rate": 2.9622185557068238e-05,
+      "loss": 0.2804,
+      "step": 502
+    },
+    {
+      "epoch": 2.0364372469635628,
+      "grad_norm": 0.3611129872457288,
+      "learning_rate": 2.95725682050846e-05,
+      "loss": 0.2727,
+      "step": 503
+    },
+    {
+      "epoch": 2.0404858299595143,
+      "grad_norm": 0.4674049005434611,
+      "learning_rate": 2.9522874311100684e-05,
+      "loss": 0.2653,
+      "step": 504
+    },
+    {
+      "epoch": 2.0445344129554655,
+      "grad_norm": 0.40996126215550377,
+      "learning_rate": 2.9473104272467526e-05,
+      "loss": 0.2977,
+      "step": 505
+    },
+    {
+      "epoch": 2.048582995951417,
+      "grad_norm": 0.3758267725613359,
+      "learning_rate": 2.942325848714505e-05,
+      "loss": 0.2819,
+      "step": 506
+    },
+    {
+      "epoch": 2.0526315789473686,
+      "grad_norm": 0.3904681273943319,
+      "learning_rate": 2.9373337353698804e-05,
+      "loss": 0.2923,
+      "step": 507
+    },
+    {
+      "epoch": 2.0566801619433197,
+      "grad_norm": 0.39178331912837905,
+      "learning_rate": 2.932334127129686e-05,
+      "loss": 0.2771,
+      "step": 508
+    },
+    {
+      "epoch": 2.0607287449392713,
+      "grad_norm": 0.36080397140206283,
+      "learning_rate": 2.9273270639706547e-05,
+      "loss": 0.2838,
+      "step": 509
+    },
+    {
+      "epoch": 2.064777327935223,
+      "grad_norm": 0.398272069762543,
+      "learning_rate": 2.922312585929131e-05,
+      "loss": 0.288,
+      "step": 510
+    },
+    {
+      "epoch": 2.068825910931174,
+      "grad_norm": 0.38183455633937513,
+      "learning_rate": 2.9172907331007478e-05,
+      "loss": 0.2813,
+      "step": 511
+    },
+    {
+      "epoch": 2.0728744939271255,
+      "grad_norm": 0.42731214893892705,
+      "learning_rate": 2.9122615456401053e-05,
+      "loss": 0.3003,
+      "step": 512
+    },
+    {
+      "epoch": 2.076923076923077,
+      "grad_norm": 0.3756498091049423,
+      "learning_rate": 2.9072250637604535e-05,
+      "loss": 0.2842,
+      "step": 513
+    },
+    {
+      "epoch": 2.080971659919028,
+      "grad_norm": 0.3982921836360366,
+      "learning_rate": 2.9021813277333667e-05,
+      "loss": 0.2754,
+      "step": 514
+    },
+    {
+      "epoch": 2.08502024291498,
+      "grad_norm": 0.39087061347621344,
+      "learning_rate": 2.8971303778884244e-05,
+      "loss": 0.2894,
+      "step": 515
+    },
+    {
+      "epoch": 2.0890688259109313,
+      "grad_norm": 0.36160789552185213,
+      "learning_rate": 2.8920722546128863e-05,
+      "loss": 0.2671,
+      "step": 516
+    },
+    {
+      "epoch": 2.0931174089068825,
+      "grad_norm": 0.45823932275162865,
+      "learning_rate": 2.887006998351372e-05,
+      "loss": 0.2791,
+      "step": 517
+    },
+    {
+      "epoch": 2.097165991902834,
+      "grad_norm": 0.34746545600614653,
+      "learning_rate": 2.881934649605535e-05,
+      "loss": 0.2798,
+      "step": 518
+    },
+    {
+      "epoch": 2.1012145748987856,
+      "grad_norm": 0.4521660857633365,
+      "learning_rate": 2.8768552489337408e-05,
+      "loss": 0.2975,
+      "step": 519
+    },
+    {
+      "epoch": 2.1052631578947367,
+      "grad_norm": 0.36510608199176153,
+      "learning_rate": 2.8717688369507425e-05,
+      "loss": 0.2692,
+      "step": 520
+    },
+    {
+      "epoch": 2.1093117408906883,
+      "grad_norm": 0.4274611596708236,
+      "learning_rate": 2.8666754543273535e-05,
+      "loss": 0.287,
+      "step": 521
+    },
+    {
+      "epoch": 2.11336032388664,
+      "grad_norm": 0.37415607182461474,
+      "learning_rate": 2.8615751417901273e-05,
+      "loss": 0.2654,
+      "step": 522
+    },
+    {
+      "epoch": 2.117408906882591,
+      "grad_norm": 0.37705567757644215,
+      "learning_rate": 2.856467940121025e-05,
+      "loss": 0.3002,
+      "step": 523
+    },
+    {
+      "epoch": 2.1214574898785425,
+      "grad_norm": 0.383442298389726,
+      "learning_rate": 2.8513538901570964e-05,
+      "loss": 0.2707,
+      "step": 524
+    },
+    {
+      "epoch": 2.125506072874494,
+      "grad_norm": 0.3430855150063715,
+      "learning_rate": 2.846233032790149e-05,
+      "loss": 0.2769,
+      "step": 525
+    },
+    {
+      "epoch": 2.1295546558704452,
+      "grad_norm": 0.3696511211365157,
+      "learning_rate": 2.84110540896642e-05,
+      "loss": 0.2773,
+      "step": 526
+    },
+    {
+      "epoch": 2.133603238866397,
+      "grad_norm": 0.34199336151051,
+      "learning_rate": 2.8359710596862555e-05,
+      "loss": 0.2627,
+      "step": 527
+    },
+    {
+      "epoch": 2.1376518218623484,
+      "grad_norm": 0.4042220901811012,
+      "learning_rate": 2.8308300260037734e-05,
+      "loss": 0.2942,
+      "step": 528
+    },
+    {
+      "epoch": 2.1417004048582995,
+      "grad_norm": 0.319260289070332,
+      "learning_rate": 2.8256823490265424e-05,
+      "loss": 0.2725,
+      "step": 529
+    },
+    {
+      "epoch": 2.145748987854251,
+      "grad_norm": 0.3594889320604298,
+      "learning_rate": 2.8205280699152515e-05,
+      "loss": 0.2601,
+      "step": 530
+    },
+    {
+      "epoch": 2.1497975708502026,
+      "grad_norm": 0.34653541794510745,
+      "learning_rate": 2.8153672298833777e-05,
+      "loss": 0.2841,
+      "step": 531
+    },
+    {
+      "epoch": 2.1538461538461537,
+      "grad_norm": 0.36506111257146084,
+      "learning_rate": 2.8101998701968615e-05,
+      "loss": 0.2702,
+      "step": 532
+    },
+    {
+      "epoch": 2.1578947368421053,
+      "grad_norm": 0.3773069480804255,
+      "learning_rate": 2.805026032173773e-05,
+      "loss": 0.2695,
+      "step": 533
+    },
+    {
+      "epoch": 2.161943319838057,
+      "grad_norm": 0.37036228853179576,
+      "learning_rate": 2.7998457571839832e-05,
+      "loss": 0.2765,
+      "step": 534
+    },
+    {
+      "epoch": 2.165991902834008,
+      "grad_norm": 0.36254792758556914,
+      "learning_rate": 2.7946590866488322e-05,
+      "loss": 0.2801,
+      "step": 535
+    },
+    {
+      "epoch": 2.1700404858299596,
+      "grad_norm": 0.40452392534953663,
+      "learning_rate": 2.7894660620408e-05,
+      "loss": 0.2886,
+      "step": 536
+    },
+    {
+      "epoch": 2.174089068825911,
+      "grad_norm": 0.35874920967738294,
+      "learning_rate": 2.784266724883173e-05,
+      "loss": 0.2535,
+      "step": 537
+    },
+    {
+      "epoch": 2.1781376518218623,
+      "grad_norm": 0.4161148655401944,
+      "learning_rate": 2.7790611167497128e-05,
+      "loss": 0.2876,
+      "step": 538
+    },
+    {
+      "epoch": 2.182186234817814,
+      "grad_norm": 0.3598455285660406,
+      "learning_rate": 2.7738492792643224e-05,
+      "loss": 0.2777,
+      "step": 539
+    },
+    {
+      "epoch": 2.1862348178137654,
+      "grad_norm": 0.34724135566623315,
+      "learning_rate": 2.768631254100716e-05,
+      "loss": 0.2796,
+      "step": 540
+    },
+    {
+      "epoch": 2.1902834008097165,
+      "grad_norm": 0.40721199232437255,
+      "learning_rate": 2.7634070829820836e-05,
+      "loss": 0.2902,
+      "step": 541
+    },
+    {
+      "epoch": 2.194331983805668,
+      "grad_norm": 0.3614853925136745,
+      "learning_rate": 2.7581768076807592e-05,
+      "loss": 0.3103,
+      "step": 542
+    },
+    {
+      "epoch": 2.1983805668016196,
+      "grad_norm": 0.4119416628662494,
+      "learning_rate": 2.7529404700178832e-05,
+      "loss": 0.2937,
+      "step": 543
+    },
+    {
+      "epoch": 2.2024291497975708,
+      "grad_norm": 0.32116299150030747,
+      "learning_rate": 2.7476981118630735e-05,
+      "loss": 0.2903,
+      "step": 544
+    },
+    {
+      "epoch": 2.2064777327935223,
+      "grad_norm": 0.3840992865010118,
+      "learning_rate": 2.742449775134085e-05,
+      "loss": 0.2695,
+      "step": 545
+    },
+    {
+      "epoch": 2.2105263157894735,
+      "grad_norm": 0.3447461088823134,
+      "learning_rate": 2.7371955017964788e-05,
+      "loss": 0.2657,
+      "step": 546
+    },
+    {
+      "epoch": 2.214574898785425,
+      "grad_norm": 0.3696122301691591,
+      "learning_rate": 2.7319353338632845e-05,
+      "loss": 0.2629,
+      "step": 547
+    },
+    {
+      "epoch": 2.2186234817813766,
+      "grad_norm": 0.38036563073232393,
+      "learning_rate": 2.7266693133946645e-05,
+      "loss": 0.3006,
+      "step": 548
+    },
+    {
+      "epoch": 2.2226720647773277,
+      "grad_norm": 0.3294169855192188,
+      "learning_rate": 2.7213974824975785e-05,
+      "loss": 0.2735,
+      "step": 549
+    },
+    {
+      "epoch": 2.2267206477732793,
+      "grad_norm": 0.40208376032021076,
+      "learning_rate": 2.7161198833254464e-05,
+      "loss": 0.2934,
+      "step": 550
+    },
+    {
+      "epoch": 2.230769230769231,
+      "grad_norm": 0.35090571254941155,
+      "learning_rate": 2.7108365580778095e-05,
+      "loss": 0.2919,
+      "step": 551
+    },
+    {
+      "epoch": 2.234817813765182,
+      "grad_norm": 0.33208024729581,
+      "learning_rate": 2.705547548999996e-05,
+      "loss": 0.2714,
+      "step": 552
+    },
+    {
+      "epoch": 2.2388663967611335,
+      "grad_norm": 0.374266663842396,
+      "learning_rate": 2.7002528983827817e-05,
+      "loss": 0.2736,
+      "step": 553
+    },
+    {
+      "epoch": 2.242914979757085,
+      "grad_norm": 0.32836527766874485,
+      "learning_rate": 2.6949526485620517e-05,
+      "loss": 0.2743,
+      "step": 554
+    },
+    {
+      "epoch": 2.246963562753036,
+      "grad_norm": 0.40384789128043086,
+      "learning_rate": 2.6896468419184623e-05,
+      "loss": 0.3034,
+      "step": 555
+    },
+    {
+      "epoch": 2.251012145748988,
+      "grad_norm": 0.3131031003632895,
+      "learning_rate": 2.6843355208771024e-05,
+      "loss": 0.2916,
+      "step": 556
+    },
+    {
+      "epoch": 2.2550607287449393,
+      "grad_norm": 0.4054848551846526,
+      "learning_rate": 2.6790187279071536e-05,
+      "loss": 0.3009,
+      "step": 557
+    },
+    {
+      "epoch": 2.2591093117408905,
+      "grad_norm": 0.36204512943051503,
+      "learning_rate": 2.67369650552155e-05,
+      "loss": 0.2918,
+      "step": 558
+    },
+    {
+      "epoch": 2.263157894736842,
+      "grad_norm": 0.33573096395748514,
+      "learning_rate": 2.6683688962766407e-05,
+      "loss": 0.2884,
+      "step": 559
+    },
+    {
+      "epoch": 2.2672064777327936,
+      "grad_norm": 0.3350140359260613,
+      "learning_rate": 2.6630359427718468e-05,
+      "loss": 0.265,
+      "step": 560
+    },
+    {
+      "epoch": 2.2712550607287447,
+      "grad_norm": 0.3773179749899134,
+      "learning_rate": 2.6576976876493225e-05,
+      "loss": 0.2767,
+      "step": 561
+    },
+    {
+      "epoch": 2.2753036437246963,
+      "grad_norm": 0.31037945633746655,
+      "learning_rate": 2.652354173593614e-05,
+      "loss": 0.2801,
+      "step": 562
+    },
+    {
+      "epoch": 2.279352226720648,
+      "grad_norm": 0.3704010320357341,
+      "learning_rate": 2.647005443331317e-05,
+      "loss": 0.2982,
+      "step": 563
+    },
+    {
+      "epoch": 2.283400809716599,
+      "grad_norm": 0.3439120452679456,
+      "learning_rate": 2.6416515396307354e-05,
+      "loss": 0.2809,
+      "step": 564
+    },
+    {
+      "epoch": 2.2874493927125505,
+      "grad_norm": 0.3286465706415457,
+      "learning_rate": 2.636292505301542e-05,
+      "loss": 0.285,
+      "step": 565
+    },
+    {
+      "epoch": 2.291497975708502,
+      "grad_norm": 0.3269534014865223,
+      "learning_rate": 2.630928383194431e-05,
+      "loss": 0.2857,
+      "step": 566
+    },
+    {
+      "epoch": 2.2955465587044532,
+      "grad_norm": 0.32310779903587755,
+      "learning_rate": 2.625559216200781e-05,
+      "loss": 0.2835,
+      "step": 567
+    },
+    {
+      "epoch": 2.299595141700405,
+      "grad_norm": 0.3341348870583215,
+      "learning_rate": 2.620185047252308e-05,
+      "loss": 0.273,
+      "step": 568
+    },
+    {
+      "epoch": 2.3036437246963564,
+      "grad_norm": 0.3455674751291652,
+      "learning_rate": 2.614805919320723e-05,
+      "loss": 0.3069,
+      "step": 569
+    },
+    {
+      "epoch": 2.3076923076923075,
+      "grad_norm": 0.32954528800371236,
+      "learning_rate": 2.6094218754173895e-05,
+      "loss": 0.2744,
+      "step": 570
+    },
+    {
+      "epoch": 2.311740890688259,
+      "grad_norm": 0.33938571492467173,
+      "learning_rate": 2.60403295859298e-05,
+      "loss": 0.2804,
+      "step": 571
+    },
+    {
+      "epoch": 2.3157894736842106,
+      "grad_norm": 0.3461023328066336,
+      "learning_rate": 2.5986392119371297e-05,
+      "loss": 0.2812,
+      "step": 572
+    },
+    {
+      "epoch": 2.3198380566801617,
+      "grad_norm": 0.3651169496606674,
+      "learning_rate": 2.5932406785780928e-05,
+      "loss": 0.2776,
+      "step": 573
+    },
+    {
+      "epoch": 2.3238866396761133,
+      "grad_norm": 0.3177358490049366,
+      "learning_rate": 2.5878374016823987e-05,
+      "loss": 0.2921,
+      "step": 574
+    },
+    {
+      "epoch": 2.327935222672065,
+      "grad_norm": 0.3736917770073717,
+      "learning_rate": 2.5824294244545045e-05,
+      "loss": 0.2907,
+      "step": 575
+    },
+    {
+      "epoch": 2.331983805668016,
+      "grad_norm": 0.32432882915485045,
+      "learning_rate": 2.577016790136453e-05,
+      "loss": 0.2823,
+      "step": 576
+    },
+    {
+      "epoch": 2.3360323886639676,
+      "grad_norm": 0.36812511123761293,
+      "learning_rate": 2.5715995420075225e-05,
+      "loss": 0.285,
+      "step": 577
+    },
+    {
+      "epoch": 2.340080971659919,
+      "grad_norm": 0.3417552901160492,
+      "learning_rate": 2.5661777233838864e-05,
+      "loss": 0.2814,
+      "step": 578
+    },
+    {
+      "epoch": 2.3441295546558703,
+      "grad_norm": 0.32747012008555865,
+      "learning_rate": 2.5607513776182616e-05,
+      "loss": 0.2772,
+      "step": 579
+    },
+    {
+      "epoch": 2.348178137651822,
+      "grad_norm": 0.31699640427527187,
+      "learning_rate": 2.5553205480995633e-05,
+      "loss": 0.2893,
+      "step": 580
+    },
+    {
+      "epoch": 2.3522267206477734,
+      "grad_norm": 0.327984041216824,
+      "learning_rate": 2.5498852782525604e-05,
+      "loss": 0.2827,
+      "step": 581
+    },
+    {
+      "epoch": 2.3562753036437245,
+      "grad_norm": 0.3377979583208305,
+      "learning_rate": 2.5444456115375252e-05,
+      "loss": 0.2793,
+      "step": 582
+    },
+    {
+      "epoch": 2.360323886639676,
+      "grad_norm": 0.3106898613618173,
+      "learning_rate": 2.539001591449889e-05,
+      "loss": 0.2817,
+      "step": 583
+    },
+    {
+      "epoch": 2.3643724696356276,
+      "grad_norm": 0.33735077057587887,
+      "learning_rate": 2.53355326151989e-05,
+      "loss": 0.2877,
+      "step": 584
+    },
+    {
+      "epoch": 2.3684210526315788,
+      "grad_norm": 0.3278405047929037,
+      "learning_rate": 2.5281006653122288e-05,
+      "loss": 0.2853,
+      "step": 585
+    },
+    {
+      "epoch": 2.3724696356275303,
+      "grad_norm": 0.3421973207980533,
+      "learning_rate": 2.5226438464257217e-05,
+      "loss": 0.2879,
+      "step": 586
+    },
+    {
+      "epoch": 2.376518218623482,
+      "grad_norm": 0.3458885155959433,
+      "learning_rate": 2.5171828484929447e-05,
+      "loss": 0.2646,
+      "step": 587
+    },
+    {
+      "epoch": 2.380566801619433,
+      "grad_norm": 0.33976591575058274,
+      "learning_rate": 2.5117177151798934e-05,
+      "loss": 0.286,
+      "step": 588
+    },
+    {
+      "epoch": 2.3846153846153846,
+      "grad_norm": 0.3605062531656976,
+      "learning_rate": 2.506248490185627e-05,
+      "loss": 0.2711,
+      "step": 589
+    },
+    {
+      "epoch": 2.388663967611336,
+      "grad_norm": 0.3523369401127506,
+      "learning_rate": 2.500775217241923e-05,
+      "loss": 0.3068,
+      "step": 590
+    },
+    {
+      "epoch": 2.3927125506072873,
+      "grad_norm": 0.3445280527975442,
+      "learning_rate": 2.4952979401129266e-05,
+      "loss": 0.2895,
+      "step": 591
+    },
+    {
+      "epoch": 2.396761133603239,
+      "grad_norm": 0.3300830110755918,
+      "learning_rate": 2.4898167025948e-05,
+      "loss": 0.3009,
+      "step": 592
+    },
+    {
+      "epoch": 2.4008097165991904,
+      "grad_norm": 0.3471052284476298,
+      "learning_rate": 2.4843315485153703e-05,
+      "loss": 0.2843,
+      "step": 593
+    },
+    {
+      "epoch": 2.4048582995951415,
+      "grad_norm": 0.3071071328165256,
+      "learning_rate": 2.4788425217337856e-05,
+      "loss": 0.2838,
+      "step": 594
+    },
+    {
+      "epoch": 2.408906882591093,
+      "grad_norm": 0.34161775449055776,
+      "learning_rate": 2.4733496661401572e-05,
+      "loss": 0.2967,
+      "step": 595
+    },
+    {
+      "epoch": 2.4129554655870447,
+      "grad_norm": 0.3337406618570239,
+      "learning_rate": 2.4678530256552114e-05,
+      "loss": 0.2884,
+      "step": 596
+    },
+    {
+      "epoch": 2.417004048582996,
+      "grad_norm": 0.3569276652327532,
+      "learning_rate": 2.46235264422994e-05,
+      "loss": 0.297,
+      "step": 597
+    },
+    {
+      "epoch": 2.4210526315789473,
+      "grad_norm": 0.3523387694226063,
+      "learning_rate": 2.4568485658452452e-05,
+      "loss": 0.2902,
+      "step": 598
+    },
+    {
+      "epoch": 2.425101214574899,
+      "grad_norm": 0.35310161077124463,
+      "learning_rate": 2.451340834511592e-05,
+      "loss": 0.2809,
+      "step": 599
+    },
+    {
+      "epoch": 2.42914979757085,
+      "grad_norm": 0.3935907167298501,
+      "learning_rate": 2.4458294942686528e-05,
+      "loss": 0.2788,
+      "step": 600
+    },
+    {
+      "epoch": 2.4331983805668016,
+      "grad_norm": 0.3191192362843582,
+      "learning_rate": 2.440314589184958e-05,
+      "loss": 0.2935,
+      "step": 601
+    },
+    {
+      "epoch": 2.437246963562753,
+      "grad_norm": 0.3524127375358563,
+      "learning_rate": 2.434796163357541e-05,
+      "loss": 0.2953,
+      "step": 602
+    },
+    {
+      "epoch": 2.4412955465587043,
+      "grad_norm": 0.31339454545769624,
+      "learning_rate": 2.429274260911588e-05,
+      "loss": 0.2932,
+      "step": 603
+    },
+    {
+      "epoch": 2.445344129554656,
+      "grad_norm": 0.33324535964460195,
+      "learning_rate": 2.423748926000084e-05,
+      "loss": 0.304,
+      "step": 604
+    },
+    {
+      "epoch": 2.4493927125506074,
+      "grad_norm": 0.3162599060888613,
+      "learning_rate": 2.418220202803459e-05,
+      "loss": 0.2688,
+      "step": 605
+    },
+    {
+      "epoch": 2.4534412955465585,
+      "grad_norm": 0.32646610161794204,
+      "learning_rate": 2.412688135529237e-05,
+      "loss": 0.2957,
+      "step": 606
+    },
+    {
+      "epoch": 2.45748987854251,
+      "grad_norm": 0.3251187008853546,
+      "learning_rate": 2.4071527684116807e-05,
+      "loss": 0.2633,
+      "step": 607
+    },
+    {
+      "epoch": 2.4615384615384617,
+      "grad_norm": 0.30999166707088677,
+      "learning_rate": 2.4016141457114372e-05,
+      "loss": 0.3064,
+      "step": 608
+    },
+    {
+      "epoch": 2.465587044534413,
+      "grad_norm": 0.3031269758000732,
+      "learning_rate": 2.396072311715187e-05,
+      "loss": 0.2745,
+      "step": 609
+    },
+    {
+      "epoch": 2.4696356275303644,
+      "grad_norm": 0.35458924874921977,
+      "learning_rate": 2.3905273107352863e-05,
+      "loss": 0.295,
+      "step": 610
+    },
+    {
+      "epoch": 2.473684210526316,
+      "grad_norm": 0.30302903348734755,
+      "learning_rate": 2.3849791871094157e-05,
+      "loss": 0.2694,
+      "step": 611
+    },
+    {
+      "epoch": 2.477732793522267,
+      "grad_norm": 0.31326371742074727,
+      "learning_rate": 2.379427985200223e-05,
+      "loss": 0.29,
+      "step": 612
+    },
+    {
+      "epoch": 2.4817813765182186,
+      "grad_norm": 0.3244896931349663,
+      "learning_rate": 2.373873749394972e-05,
+      "loss": 0.2942,
+      "step": 613
+    },
+    {
+      "epoch": 2.48582995951417,
+      "grad_norm": 0.3052954905739353,
+      "learning_rate": 2.3683165241051836e-05,
+      "loss": 0.2979,
+      "step": 614
+    },
+    {
+      "epoch": 2.4898785425101213,
+      "grad_norm": 0.33168810133326293,
+      "learning_rate": 2.3627563537662823e-05,
+      "loss": 0.2975,
+      "step": 615
+    },
+    {
+      "epoch": 2.493927125506073,
+      "grad_norm": 0.30835344437105033,
+      "learning_rate": 2.3571932828372433e-05,
+      "loss": 0.2876,
+      "step": 616
+    },
+    {
+      "epoch": 2.4979757085020244,
+      "grad_norm": 0.34465983048846227,
+      "learning_rate": 2.351627355800233e-05,
+      "loss": 0.2639,
+      "step": 617
+    },
+    {
+      "epoch": 2.5020242914979756,
+      "grad_norm": 0.33522320642558373,
+      "learning_rate": 2.346058617160256e-05,
+      "loss": 0.2889,
+      "step": 618
+    },
+    {
+      "epoch": 2.506072874493927,
+      "grad_norm": 0.3507051987631235,
+      "learning_rate": 2.3404871114447976e-05,
+      "loss": 0.3123,
+      "step": 619
+    },
+    {
+      "epoch": 2.5101214574898787,
+      "grad_norm": 0.310232211380619,
+      "learning_rate": 2.334912883203469e-05,
+      "loss": 0.2895,
+      "step": 620
+    },
+    {
+      "epoch": 2.51417004048583,
+      "grad_norm": 0.3418850826180368,
+      "learning_rate": 2.3293359770076505e-05,
+      "loss": 0.2942,
+      "step": 621
+    },
+    {
+      "epoch": 2.5182186234817814,
+      "grad_norm": 0.2976945148935071,
+      "learning_rate": 2.323756437450136e-05,
+      "loss": 0.2916,
+      "step": 622
+    },
+    {
+      "epoch": 2.522267206477733,
+      "grad_norm": 0.3148436140887398,
+      "learning_rate": 2.3181743091447754e-05,
+      "loss": 0.2915,
+      "step": 623
+    },
+    {
+      "epoch": 2.526315789473684,
+      "grad_norm": 0.34377564472074446,
+      "learning_rate": 2.3125896367261164e-05,
+      "loss": 0.3076,
+      "step": 624
+    },
+    {
+      "epoch": 2.5303643724696356,
+      "grad_norm": 0.3360642446889338,
+      "learning_rate": 2.307002464849052e-05,
+      "loss": 0.2859,
+      "step": 625
+    },
+    {
+      "epoch": 2.534412955465587,
+      "grad_norm": 0.3060252229752874,
+      "learning_rate": 2.30141283818846e-05,
+      "loss": 0.2951,
+      "step": 626
+    },
+    {
+      "epoch": 2.5384615384615383,
+      "grad_norm": 0.34787488223379087,
+      "learning_rate": 2.2958208014388462e-05,
+      "loss": 0.2918,
+      "step": 627
+    },
+    {
+      "epoch": 2.54251012145749,
+      "grad_norm": 0.28838494442047147,
+      "learning_rate": 2.2902263993139857e-05,
+      "loss": 0.2889,
+      "step": 628
+    },
+    {
+      "epoch": 2.5465587044534415,
+      "grad_norm": 0.30128693269720785,
+      "learning_rate": 2.2846296765465708e-05,
+      "loss": 0.2834,
+      "step": 629
+    },
+    {
+      "epoch": 2.5506072874493926,
+      "grad_norm": 0.32986925080932655,
+      "learning_rate": 2.2790306778878465e-05,
+      "loss": 0.2656,
+      "step": 630
+    },
+    {
+      "epoch": 2.554655870445344,
+      "grad_norm": 0.31049934398061235,
+      "learning_rate": 2.273429448107257e-05,
+      "loss": 0.2948,
+      "step": 631
+    },
+    {
+      "epoch": 2.5587044534412957,
+      "grad_norm": 0.3142136619906757,
+      "learning_rate": 2.2678260319920868e-05,
+      "loss": 0.2907,
+      "step": 632
+    },
+    {
+      "epoch": 2.562753036437247,
+      "grad_norm": 0.3255856532084294,
+      "learning_rate": 2.262220474347102e-05,
+      "loss": 0.2895,
+      "step": 633
+    },
+    {
+      "epoch": 2.5668016194331984,
+      "grad_norm": 0.32901269097639635,
+      "learning_rate": 2.2566128199941918e-05,
+      "loss": 0.3021,
+      "step": 634
+    },
+    {
+      "epoch": 2.57085020242915,
+      "grad_norm": 0.3136374892545854,
+      "learning_rate": 2.2510031137720104e-05,
+      "loss": 0.287,
+      "step": 635
+    },
+    {
+      "epoch": 2.574898785425101,
+      "grad_norm": 0.340068525849097,
+      "learning_rate": 2.2453914005356206e-05,
+      "loss": 0.3087,
+      "step": 636
+    },
+    {
+      "epoch": 2.5789473684210527,
+      "grad_norm": 0.3011554201185437,
+      "learning_rate": 2.2397777251561302e-05,
+      "loss": 0.2973,
+      "step": 637
+    },
+    {
+      "epoch": 2.582995951417004,
+      "grad_norm": 0.2904783586123931,
+      "learning_rate": 2.2341621325203393e-05,
+      "loss": 0.2841,
+      "step": 638
+    },
+    {
+      "epoch": 2.5870445344129553,
+      "grad_norm": 0.2992102910700875,
+      "learning_rate": 2.2285446675303764e-05,
+      "loss": 0.2715,
+      "step": 639
+    },
+    {
+      "epoch": 2.591093117408907,
+      "grad_norm": 0.33612963426149073,
+      "learning_rate": 2.2229253751033423e-05,
+      "loss": 0.313,
+      "step": 640
+    },
+    {
+      "epoch": 2.5951417004048585,
+      "grad_norm": 0.3367365100916747,
+      "learning_rate": 2.2173043001709492e-05,
+      "loss": 0.298,
+      "step": 641
+    },
+    {
+      "epoch": 2.5991902834008096,
+      "grad_norm": 0.30162027392363694,
+      "learning_rate": 2.2116814876791627e-05,
+      "loss": 0.289,
+      "step": 642
+    },
+    {
+      "epoch": 2.603238866396761,
+      "grad_norm": 0.34693931756158125,
+      "learning_rate": 2.2060569825878413e-05,
+      "loss": 0.281,
+      "step": 643
+    },
+    {
+      "epoch": 2.6072874493927127,
+      "grad_norm": 0.32003983870987796,
+      "learning_rate": 2.2004308298703777e-05,
+      "loss": 0.3031,
+      "step": 644
+    },
+    {
+      "epoch": 2.611336032388664,
+      "grad_norm": 0.30698660272909783,
+      "learning_rate": 2.1948030745133396e-05,
+      "loss": 0.2827,
+      "step": 645
+    },
+    {
+      "epoch": 2.6153846153846154,
+      "grad_norm": 0.29369392215178114,
+      "learning_rate": 2.1891737615161083e-05,
+      "loss": 0.2889,
+      "step": 646
+    },
+    {
+      "epoch": 2.619433198380567,
+      "grad_norm": 0.3099192583606761,
+      "learning_rate": 2.1835429358905218e-05,
+      "loss": 0.3085,
+      "step": 647
+    },
+    {
+      "epoch": 2.623481781376518,
+      "grad_norm": 0.31814068783796795,
+      "learning_rate": 2.17791064266051e-05,
+      "loss": 0.2916,
+      "step": 648
+    },
+    {
+      "epoch": 2.6275303643724697,
+      "grad_norm": 0.3016282505738383,
+      "learning_rate": 2.172276926861741e-05,
+      "loss": 0.2966,
+      "step": 649
+    },
+    {
+      "epoch": 2.6315789473684212,
+      "grad_norm": 0.31064124779274194,
+      "learning_rate": 2.166641833541255e-05,
+      "loss": 0.2824,
+      "step": 650
+    },
+    {
+      "epoch": 2.6356275303643724,
+      "grad_norm": 0.30946954627330076,
+      "learning_rate": 2.161005407757109e-05,
+      "loss": 0.273,
+      "step": 651
+    },
+    {
+      "epoch": 2.639676113360324,
+      "grad_norm": 0.32340097018505826,
+      "learning_rate": 2.155367694578013e-05,
+      "loss": 0.2773,
+      "step": 652
+    },
+    {
+      "epoch": 2.6437246963562755,
+      "grad_norm": 0.3477477441308052,
+      "learning_rate": 2.1497287390829712e-05,
+      "loss": 0.3076,
+      "step": 653
+    },
+    {
+      "epoch": 2.6477732793522266,
+      "grad_norm": 0.35768480272872344,
+      "learning_rate": 2.1440885863609208e-05,
+      "loss": 0.2725,
+      "step": 654
+    },
+    {
+      "epoch": 2.651821862348178,
+      "grad_norm": 0.3330276496710868,
+      "learning_rate": 2.1384472815103737e-05,
+      "loss": 0.2821,
+      "step": 655
+    },
+    {
+      "epoch": 2.6558704453441297,
+      "grad_norm": 0.3388599497492239,
+      "learning_rate": 2.1328048696390534e-05,
+      "loss": 0.2896,
+      "step": 656
+    },
+    {
+      "epoch": 2.659919028340081,
+      "grad_norm": 0.3386253869713379,
+      "learning_rate": 2.1271613958635342e-05,
+      "loss": 0.2768,
+      "step": 657
+    },
+    {
+      "epoch": 2.6639676113360324,
+      "grad_norm": 0.3468683138768928,
+      "learning_rate": 2.1215169053088825e-05,
+      "loss": 0.292,
+      "step": 658
+    },
+    {
+      "epoch": 2.668016194331984,
+      "grad_norm": 0.3071352239927404,
+      "learning_rate": 2.1158714431082937e-05,
+      "loss": 0.2623,
+      "step": 659
+    },
+    {
+      "epoch": 2.672064777327935,
+      "grad_norm": 0.3195589477949123,
+      "learning_rate": 2.110225054402733e-05,
+      "loss": 0.2718,
+      "step": 660
+    },
+    {
+      "epoch": 2.6761133603238867,
+      "grad_norm": 0.3231634560848976,
+      "learning_rate": 2.1045777843405747e-05,
+      "loss": 0.2959,
+      "step": 661
+    },
+    {
+      "epoch": 2.6801619433198383,
+      "grad_norm": 0.30252828525282327,
+      "learning_rate": 2.09892967807724e-05,
+      "loss": 0.2819,
+      "step": 662
+    },
+    {
+      "epoch": 2.6842105263157894,
+      "grad_norm": 0.30870327136829145,
+      "learning_rate": 2.093280780774835e-05,
+      "loss": 0.2764,
+      "step": 663
+    },
+    {
+      "epoch": 2.688259109311741,
+      "grad_norm": 0.31594227869268854,
+      "learning_rate": 2.087631137601793e-05,
+      "loss": 0.2731,
+      "step": 664
+    },
+    {
+      "epoch": 2.6923076923076925,
+      "grad_norm": 0.3086844966652639,
+      "learning_rate": 2.0819807937325104e-05,
+      "loss": 0.3008,
+      "step": 665
+    },
+    {
+      "epoch": 2.6963562753036436,
+      "grad_norm": 0.31888267236510204,
+      "learning_rate": 2.0763297943469847e-05,
+      "loss": 0.2757,
+      "step": 666
+    },
+    {
+      "epoch": 2.700404858299595,
+      "grad_norm": 0.3232690554382307,
+      "learning_rate": 2.070678184630458e-05,
+      "loss": 0.271,
+      "step": 667
+    },
+    {
+      "epoch": 2.7044534412955468,
+      "grad_norm": 0.3151708472462376,
+      "learning_rate": 2.06502600977305e-05,
+      "loss": 0.2854,
+      "step": 668
+    },
+    {
+      "epoch": 2.708502024291498,
+      "grad_norm": 0.32018264342398656,
+      "learning_rate": 2.0593733149694008e-05,
+      "loss": 0.283,
+      "step": 669
+    },
+    {
+      "epoch": 2.7125506072874495,
+      "grad_norm": 0.3220465915833933,
+      "learning_rate": 2.0537201454183074e-05,
+      "loss": 0.2936,
+      "step": 670
+    },
+    {
+      "epoch": 2.716599190283401,
+      "grad_norm": 0.3172132017701568,
+      "learning_rate": 2.0480665463223626e-05,
+      "loss": 0.2945,
+      "step": 671
+    },
+    {
+      "epoch": 2.720647773279352,
+      "grad_norm": 0.3499915922541871,
+      "learning_rate": 2.042412562887594e-05,
+      "loss": 0.2924,
+      "step": 672
+    },
+    {
+      "epoch": 2.7246963562753037,
+      "grad_norm": 0.3159386584575246,
+      "learning_rate": 2.0367582403231033e-05,
+      "loss": 0.2918,
+      "step": 673
+    },
+    {
+      "epoch": 2.7287449392712553,
+      "grad_norm": 0.33106671395406245,
+      "learning_rate": 2.0311036238407023e-05,
+      "loss": 0.2668,
+      "step": 674
+    },
+    {
+      "epoch": 2.7327935222672064,
+      "grad_norm": 0.34036934891589393,
+      "learning_rate": 2.0254487586545534e-05,
+      "loss": 0.2802,
+      "step": 675
+    },
+    {
+      "epoch": 2.736842105263158,
+      "grad_norm": 0.346226006355185,
+      "learning_rate": 2.0197936899808088e-05,
+      "loss": 0.2784,
+      "step": 676
+    },
+    {
+      "epoch": 2.7408906882591095,
+      "grad_norm": 0.3236868650397469,
+      "learning_rate": 2.0141384630372466e-05,
+      "loss": 0.2933,
+      "step": 677
+    },
+    {
+      "epoch": 2.7449392712550607,
+      "grad_norm": 0.347880343042189,
+      "learning_rate": 2.0084831230429106e-05,
+      "loss": 0.2863,
+      "step": 678
+    },
+    {
+      "epoch": 2.748987854251012,
+      "grad_norm": 0.3200228469580614,
+      "learning_rate": 2.0028277152177492e-05,
+      "loss": 0.2773,
+      "step": 679
+    },
+    {
+      "epoch": 2.753036437246964,
+      "grad_norm": 0.35762260706271887,
+      "learning_rate": 1.9971722847822518e-05,
+      "loss": 0.3066,
+      "step": 680
+    },
+    {
+      "epoch": 2.757085020242915,
+      "grad_norm": 0.3092093213592242,
+      "learning_rate": 1.99151687695709e-05,
+      "loss": 0.2911,
+      "step": 681
+    },
+    {
+      "epoch": 2.7611336032388665,
+      "grad_norm": 0.31714776270034084,
+      "learning_rate": 1.9858615369627537e-05,
+      "loss": 0.2956,
+      "step": 682
+    },
+    {
+      "epoch": 2.765182186234818,
+      "grad_norm": 0.3092010938434201,
+      "learning_rate": 1.9802063100191916e-05,
+      "loss": 0.2941,
+      "step": 683
+    },
+    {
+      "epoch": 2.769230769230769,
+      "grad_norm": 0.3227824521192695,
+      "learning_rate": 1.974551241345447e-05,
+      "loss": 0.306,
+      "step": 684
+    },
+    {
+      "epoch": 2.7732793522267207,
+      "grad_norm": 0.30697722888592655,
+      "learning_rate": 1.9688963761592984e-05,
+      "loss": 0.2813,
+      "step": 685
+    },
+    {
+      "epoch": 2.7773279352226723,
+      "grad_norm": 0.28985818170905975,
+      "learning_rate": 1.963241759676897e-05,
+      "loss": 0.2725,
+      "step": 686
+    },
+    {
+      "epoch": 2.7813765182186234,
+      "grad_norm": 0.3123034107192393,
+      "learning_rate": 1.9575874371124066e-05,
+      "loss": 0.279,
+      "step": 687
+    },
+    {
+      "epoch": 2.785425101214575,
+      "grad_norm": 0.3225780199343664,
+      "learning_rate": 1.951933453677638e-05,
+      "loss": 0.2745,
+      "step": 688
+    },
+    {
+      "epoch": 2.7894736842105265,
+      "grad_norm": 0.30868132572063706,
+      "learning_rate": 1.9462798545816932e-05,
+      "loss": 0.2638,
+      "step": 689
+    },
+    {
+      "epoch": 2.7935222672064777,
+      "grad_norm": 0.33706285383278284,
+      "learning_rate": 1.9406266850305996e-05,
+      "loss": 0.2977,
+      "step": 690
+    },
+    {
+      "epoch": 2.7975708502024292,
+      "grad_norm": 0.3328281673347713,
+      "learning_rate": 1.9349739902269503e-05,
+      "loss": 0.2993,
+      "step": 691
+    },
+    {
+      "epoch": 2.801619433198381,
+      "grad_norm": 0.3385918715678388,
+      "learning_rate": 1.9293218153695425e-05,
+      "loss": 0.2949,
+      "step": 692
+    },
+    {
+      "epoch": 2.805668016194332,
+      "grad_norm": 0.3678084997733253,
+      "learning_rate": 1.9236702056530153e-05,
+      "loss": 0.2857,
+      "step": 693
+    },
+    {
+      "epoch": 2.8097165991902835,
+      "grad_norm": 0.3067132501848614,
+      "learning_rate": 1.9180192062674907e-05,
+      "loss": 0.2649,
+      "step": 694
+    },
+    {
+      "epoch": 2.813765182186235,
+      "grad_norm": 0.30518252010146557,
+      "learning_rate": 1.9123688623982076e-05,
+      "loss": 0.2903,
+      "step": 695
+    },
+    {
+      "epoch": 2.817813765182186,
+      "grad_norm": 0.35703599407724607,
+      "learning_rate": 1.906719219225165e-05,
+      "loss": 0.2868,
+      "step": 696
+    },
+    {
+      "epoch": 2.8218623481781377,
+      "grad_norm": 0.3262658000413752,
+      "learning_rate": 1.9010703219227603e-05,
+      "loss": 0.2956,
+      "step": 697
+    },
+    {
+      "epoch": 2.8259109311740893,
+      "grad_norm": 0.29827246813359404,
+      "learning_rate": 1.8954222156594256e-05,
+      "loss": 0.2801,
+      "step": 698
+    },
+    {
+      "epoch": 2.8299595141700404,
+      "grad_norm": 0.34436462571072485,
+      "learning_rate": 1.8897749455972673e-05,
+      "loss": 0.2787,
+      "step": 699
+    },
+    {
+      "epoch": 2.834008097165992,
+      "grad_norm": 0.3631960731204254,
+      "learning_rate": 1.884128556891707e-05,
+      "loss": 0.289,
+      "step": 700
+    },
+    {
+      "epoch": 2.8380566801619436,
+      "grad_norm": 0.31358539505682415,
+      "learning_rate": 1.878483094691119e-05,
+      "loss": 0.2806,
+      "step": 701
+    },
+    {
+      "epoch": 2.8421052631578947,
+      "grad_norm": 0.3933209081068681,
+      "learning_rate": 1.8728386041364664e-05,
+      "loss": 0.2824,
+      "step": 702
+    },
+    {
+      "epoch": 2.8461538461538463,
+      "grad_norm": 0.2963412135338255,
+      "learning_rate": 1.867195130360947e-05,
+      "loss": 0.2831,
+      "step": 703
+    },
+    {
+      "epoch": 2.850202429149798,
+      "grad_norm": 0.3164229923387106,
+      "learning_rate": 1.8615527184896263e-05,
+      "loss": 0.2823,
+      "step": 704
+    },
+    {
+      "epoch": 2.854251012145749,
+      "grad_norm": 0.301366321958825,
+      "learning_rate": 1.8559114136390795e-05,
+      "loss": 0.2703,
+      "step": 705
+    },
+    {
+      "epoch": 2.8582995951417005,
+      "grad_norm": 0.3302607206422732,
+      "learning_rate": 1.8502712609170298e-05,
+      "loss": 0.2867,
+      "step": 706
+    },
+    {
+      "epoch": 2.862348178137652,
+      "grad_norm": 0.3279993830760607,
+      "learning_rate": 1.8446323054219876e-05,
+      "loss": 0.2751,
+      "step": 707
+    },
+    {
+      "epoch": 2.866396761133603,
+      "grad_norm": 0.35120057704675617,
+      "learning_rate": 1.838994592242891e-05,
+      "loss": 0.2736,
+      "step": 708
+    },
+    {
+      "epoch": 2.8704453441295548,
+      "grad_norm": 0.31262471819363696,
+      "learning_rate": 1.8333581664587453e-05,
+      "loss": 0.2859,
+      "step": 709
+    },
+    {
+      "epoch": 2.8744939271255063,
+      "grad_norm": 0.30068915740875374,
+      "learning_rate": 1.8277230731382593e-05,
+      "loss": 0.2761,
+      "step": 710
+    },
+    {
+      "epoch": 2.8785425101214575,
+      "grad_norm": 0.3316279987523849,
+      "learning_rate": 1.82208935733949e-05,
+      "loss": 0.2898,
+      "step": 711
+    },
+    {
+      "epoch": 2.882591093117409,
+      "grad_norm": 0.3050321809052951,
+      "learning_rate": 1.8164570641094793e-05,
+      "loss": 0.2811,
+      "step": 712
+    },
+    {
+      "epoch": 2.8866396761133606,
+      "grad_norm": 0.3482784437719155,
+      "learning_rate": 1.810826238483892e-05,
+      "loss": 0.2883,
+      "step": 713
+    },
+    {
+      "epoch": 2.8906882591093117,
+      "grad_norm": 0.2984838776837278,
+      "learning_rate": 1.8051969254866608e-05,
+      "loss": 0.2974,
+      "step": 714
+    },
+    {
+      "epoch": 2.8947368421052633,
+      "grad_norm": 0.3559100100146726,
+      "learning_rate": 1.7995691701296226e-05,
+      "loss": 0.2727,
+      "step": 715
+    },
+    {
+      "epoch": 2.898785425101215,
+      "grad_norm": 0.3179923658023191,
+      "learning_rate": 1.7939430174121593e-05,
+      "loss": 0.2861,
+      "step": 716
+    },
+    {
+      "epoch": 2.902834008097166,
+      "grad_norm": 0.3170512297762442,
+      "learning_rate": 1.788318512320838e-05,
+      "loss": 0.2814,
+      "step": 717
+    },
+    {
+      "epoch": 2.9068825910931175,
+      "grad_norm": 0.30470054974944777,
+      "learning_rate": 1.782695699829051e-05,
+      "loss": 0.2878,
+      "step": 718
+    },
+    {
+      "epoch": 2.910931174089069,
+      "grad_norm": 0.3282642804900981,
+      "learning_rate": 1.7770746248966587e-05,
+      "loss": 0.2799,
+      "step": 719
+    },
+    {
+      "epoch": 2.91497975708502,
+      "grad_norm": 0.3066003031770754,
+      "learning_rate": 1.7714553324696243e-05,
+      "loss": 0.2752,
+      "step": 720
+    },
+    {
+      "epoch": 2.919028340080972,
+      "grad_norm": 0.33960450401219633,
+      "learning_rate": 1.7658378674796614e-05,
+      "loss": 0.273,
+      "step": 721
+    },
+    {
+      "epoch": 2.9230769230769234,
+      "grad_norm": 0.30652868195287136,
+      "learning_rate": 1.76022227484387e-05,
+      "loss": 0.2794,
+      "step": 722
+    },
+    {
+      "epoch": 2.9271255060728745,
+      "grad_norm": 0.35767756527194283,
+      "learning_rate": 1.7546085994643807e-05,
+      "loss": 0.2789,
+      "step": 723
+    },
+    {
+      "epoch": 2.931174089068826,
+      "grad_norm": 0.33243588335500923,
+      "learning_rate": 1.7489968862279902e-05,
+      "loss": 0.2959,
+      "step": 724
+    },
+    {
+      "epoch": 2.9352226720647776,
+      "grad_norm": 0.30264380243625366,
+      "learning_rate": 1.743387180005809e-05,
+      "loss": 0.287,
+      "step": 725
+    },
+    {
+      "epoch": 2.9392712550607287,
+      "grad_norm": 0.31083232057860777,
+      "learning_rate": 1.737779525652899e-05,
+      "loss": 0.286,
+      "step": 726
+    },
+    {
+      "epoch": 2.9433198380566803,
+      "grad_norm": 0.31513372204097984,
+      "learning_rate": 1.732173968007914e-05,
+      "loss": 0.2793,
+      "step": 727
+    },
+    {
+      "epoch": 2.9473684210526314,
+      "grad_norm": 0.3310613315685983,
+      "learning_rate": 1.7265705518927437e-05,
+      "loss": 0.285,
+      "step": 728
+    },
+    {
+      "epoch": 2.951417004048583,
+      "grad_norm": 0.3168223682410759,
+      "learning_rate": 1.7209693221121542e-05,
+      "loss": 0.2938,
+      "step": 729
+    },
+    {
+      "epoch": 2.9554655870445345,
+      "grad_norm": 0.3338117291763535,
+      "learning_rate": 1.7153703234534302e-05,
+      "loss": 0.292,
+      "step": 730
+    },
+    {
+      "epoch": 2.9595141700404857,
+      "grad_norm": 0.3194412696584913,
+      "learning_rate": 1.709773600686015e-05,
+      "loss": 0.2783,
+      "step": 731
+    },
+    {
+      "epoch": 2.9635627530364372,
+      "grad_norm": 0.3380942606196546,
+      "learning_rate": 1.7041791985611548e-05,
+      "loss": 0.2892,
+      "step": 732
+    },
+    {
+      "epoch": 2.967611336032389,
+      "grad_norm": 0.3525884086489862,
+      "learning_rate": 1.69858716181154e-05,
+      "loss": 0.2649,
+      "step": 733
+    },
+    {
+      "epoch": 2.97165991902834,
+      "grad_norm": 0.29780315142320424,
+      "learning_rate": 1.692997535150948e-05,
+      "loss": 0.2785,
+      "step": 734
+    },
+    {
+      "epoch": 2.9757085020242915,
+      "grad_norm": 0.37156804926885073,
+      "learning_rate": 1.687410363273884e-05,
+      "loss": 0.287,
+      "step": 735
+    },
+    {
+      "epoch": 2.979757085020243,
+      "grad_norm": 0.366545147899131,
+      "learning_rate": 1.6818256908552257e-05,
+      "loss": 0.2894,
+      "step": 736
+    },
+    {
+      "epoch": 2.983805668016194,
+      "grad_norm": 0.28598951446097853,
+      "learning_rate": 1.6762435625498646e-05,
+      "loss": 0.289,
+      "step": 737
+    },
+    {
+      "epoch": 2.9878542510121457,
+      "grad_norm": 0.35830991851602095,
+      "learning_rate": 1.67066402299235e-05,
+      "loss": 0.2701,
+      "step": 738
+    },
+    {
+      "epoch": 2.9919028340080973,
+      "grad_norm": 0.3257029583275646,
+      "learning_rate": 1.6650871167965313e-05,
+      "loss": 0.2945,
+      "step": 739
+    },
+    {
+      "epoch": 2.9959514170040484,
+      "grad_norm": 0.3007401232065952,
+      "learning_rate": 1.6595128885552028e-05,
+      "loss": 0.2806,
+      "step": 740
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 0.3260125568808646,
+      "learning_rate": 1.653941382839745e-05,
+      "loss": 0.2831,
+      "step": 741
+    },
+    {
+      "epoch": 3.0040485829959516,
+      "grad_norm": 0.48390710162908424,
+      "learning_rate": 1.6483726441997673e-05,
+      "loss": 0.2056,
+      "step": 742
+    },
+    {
+      "epoch": 3.0080971659919027,
+      "grad_norm": 0.3169053891446322,
+      "learning_rate": 1.642806717162757e-05,
+      "loss": 0.1959,
+      "step": 743
+    },
+    {
+      "epoch": 3.0121457489878543,
+      "grad_norm": 0.5496226301969515,
+      "learning_rate": 1.637243646233718e-05,
+      "loss": 0.207,
+      "step": 744
+    },
+    {
+      "epoch": 3.016194331983806,
+      "grad_norm": 0.3694289920726758,
+      "learning_rate": 1.6316834758948174e-05,
+      "loss": 0.1944,
+      "step": 745
+    },
+    {
+      "epoch": 3.020242914979757,
+      "grad_norm": 0.423222877085196,
+      "learning_rate": 1.6261262506050282e-05,
+      "loss": 0.2035,
+      "step": 746
+    },
+    {
+      "epoch": 3.0242914979757085,
+      "grad_norm": 0.40888632559736166,
+      "learning_rate": 1.620572014799777e-05,
+      "loss": 0.2042,
+      "step": 747
+    },
+    {
+      "epoch": 3.02834008097166,
+      "grad_norm": 0.3532807229148997,
+      "learning_rate": 1.6150208128905857e-05,
+      "loss": 0.1978,
+      "step": 748
+    },
+    {
+      "epoch": 3.032388663967611,
+      "grad_norm": 0.3514822762111404,
+      "learning_rate": 1.6094726892647147e-05,
+      "loss": 0.2076,
+      "step": 749
+    },
+    {
+      "epoch": 3.0364372469635628,
+      "grad_norm": 0.36912858955685063,
+      "learning_rate": 1.6039276882848135e-05,
+      "loss": 0.2083,
+      "step": 750
+    },
+    {
+      "epoch": 3.0404858299595143,
+      "grad_norm": 0.35050308371253175,
+      "learning_rate": 1.5983858542885635e-05,
+      "loss": 0.2083,
+      "step": 751
+    },
+    {
+      "epoch": 3.0445344129554655,
+      "grad_norm": 0.3215930545793862,
+      "learning_rate": 1.5928472315883203e-05,
+      "loss": 0.2105,
+      "step": 752
+    },
+    {
+      "epoch": 3.048582995951417,
+      "grad_norm": 0.3357223235546437,
+      "learning_rate": 1.5873118644707633e-05,
+      "loss": 0.1894,
+      "step": 753
+    },
+    {
+      "epoch": 3.0526315789473686,
+      "grad_norm": 0.33719649463924567,
+      "learning_rate": 1.5817797971965413e-05,
+      "loss": 0.2158,
+      "step": 754
+    },
+    {
+      "epoch": 3.0566801619433197,
+      "grad_norm": 0.29628121318860734,
+      "learning_rate": 1.576251073999917e-05,
+      "loss": 0.2178,
+      "step": 755
+    },
+    {
+      "epoch": 3.0607287449392713,
+      "grad_norm": 0.35790480431531785,
+      "learning_rate": 1.5707257390884126e-05,
+      "loss": 0.2072,
+      "step": 756
+    },
+    {
+      "epoch": 3.064777327935223,
+      "grad_norm": 0.33796510656382045,
+      "learning_rate": 1.5652038366424595e-05,
+      "loss": 0.2012,
+      "step": 757
+    },
+    {
+      "epoch": 3.068825910931174,
+      "grad_norm": 0.298274515354508,
+      "learning_rate": 1.5596854108150424e-05,
+      "loss": 0.1995,
+      "step": 758
+    },
+    {
+      "epoch": 3.0728744939271255,
+      "grad_norm": 0.333298146027429,
+      "learning_rate": 1.5541705057313476e-05,
+      "loss": 0.2188,
+      "step": 759
+    },
+    {
+      "epoch": 3.076923076923077,
+      "grad_norm": 0.31690170265995715,
+      "learning_rate": 1.5486591654884086e-05,
+      "loss": 0.1854,
+      "step": 760
+    },
+    {
+      "epoch": 3.080971659919028,
+      "grad_norm": 0.33707391111683593,
+      "learning_rate": 1.543151434154755e-05,
+      "loss": 0.2018,
+      "step": 761
+    },
+    {
+      "epoch": 3.08502024291498,
+      "grad_norm": 0.3179985627843678,
+      "learning_rate": 1.5376473557700615e-05,
+      "loss": 0.2014,
+      "step": 762
+    },
+    {
+      "epoch": 3.0890688259109313,
+      "grad_norm": 0.3294320700487532,
+      "learning_rate": 1.5321469743447893e-05,
+      "loss": 0.1906,
+      "step": 763
+    },
+    {
+      "epoch": 3.0931174089068825,
+      "grad_norm": 0.3026120700796653,
+      "learning_rate": 1.5266503338598434e-05,
+      "loss": 0.1992,
+      "step": 764
+    },
+    {
+      "epoch": 3.097165991902834,
+      "grad_norm": 0.2920876911654093,
+      "learning_rate": 1.5211574782662149e-05,
+      "loss": 0.1887,
+      "step": 765
+    },
+    {
+      "epoch": 3.1012145748987856,
+      "grad_norm": 0.29750441540291606,
+      "learning_rate": 1.5156684514846307e-05,
+      "loss": 0.2167,
+      "step": 766
+    },
+    {
+      "epoch": 3.1052631578947367,
+      "grad_norm": 0.3058814757971374,
+      "learning_rate": 1.5101832974052012e-05,
+      "loss": 0.2079,
+      "step": 767
+    },
+    {
+      "epoch": 3.1093117408906883,
+      "grad_norm": 0.2948290067967974,
+      "learning_rate": 1.5047020598870737e-05,
+      "loss": 0.2058,
+      "step": 768
+    },
+    {
+      "epoch": 3.11336032388664,
+      "grad_norm": 0.30858201060731694,
+      "learning_rate": 1.4992247827580778e-05,
+      "loss": 0.2027,
+      "step": 769
+    },
+    {
+      "epoch": 3.117408906882591,
+      "grad_norm": 0.3074534377907505,
+      "learning_rate": 1.493751509814374e-05,
+      "loss": 0.1929,
+      "step": 770
+    },
+    {
+      "epoch": 3.1214574898785425,
+      "grad_norm": 0.2881606864496859,
+      "learning_rate": 1.4882822848201073e-05,
+      "loss": 0.2071,
+      "step": 771
+    },
+    {
+      "epoch": 3.125506072874494,
+      "grad_norm": 0.3104726164329907,
+      "learning_rate": 1.4828171515070553e-05,
+      "loss": 0.1831,
+      "step": 772
+    },
+    {
+      "epoch": 3.1295546558704452,
+      "grad_norm": 0.30548151013212843,
+      "learning_rate": 1.4773561535742793e-05,
+      "loss": 0.1997,
+      "step": 773
+    },
+    {
+      "epoch": 3.133603238866397,
+      "grad_norm": 0.2914341839762302,
+      "learning_rate": 1.4718993346877715e-05,
+      "loss": 0.1905,
+      "step": 774
+    },
+    {
+      "epoch": 3.1376518218623484,
+      "grad_norm": 0.33819706433432667,
+      "learning_rate": 1.466446738480111e-05,
+      "loss": 0.2216,
+      "step": 775
+    },
+    {
+      "epoch": 3.1417004048582995,
+      "grad_norm": 0.3169924341420627,
+      "learning_rate": 1.460998408550112e-05,
+      "loss": 0.2041,
+      "step": 776
+    },
+    {
+      "epoch": 3.145748987854251,
+      "grad_norm": 0.31182787962660696,
+      "learning_rate": 1.4555543884624751e-05,
+      "loss": 0.1839,
+      "step": 777
+    },
+    {
+      "epoch": 3.1497975708502026,
+      "grad_norm": 0.32900327908077937,
+      "learning_rate": 1.4501147217474402e-05,
+      "loss": 0.1879,
+      "step": 778
+    },
+    {
+      "epoch": 3.1538461538461537,
+      "grad_norm": 0.3334197775838478,
+      "learning_rate": 1.444679451900437e-05,
+      "loss": 0.2018,
+      "step": 779
+    },
+    {
+      "epoch": 3.1578947368421053,
+      "grad_norm": 0.29410494239717516,
+      "learning_rate": 1.4392486223817397e-05,
+      "loss": 0.2138,
+      "step": 780
+    },
+    {
+      "epoch": 3.161943319838057,
+      "grad_norm": 0.3172735364835795,
+      "learning_rate": 1.4338222766161143e-05,
+      "loss": 0.1972,
+      "step": 781
+    },
+    {
+      "epoch": 3.165991902834008,
+      "grad_norm": 0.3420425451818212,
+      "learning_rate": 1.4284004579924777e-05,
+      "loss": 0.1864,
+      "step": 782
+    },
+    {
+      "epoch": 3.1700404858299596,
+      "grad_norm": 0.2847480590009165,
+      "learning_rate": 1.4229832098635479e-05,
+      "loss": 0.1996,
+      "step": 783
+    },
+    {
+      "epoch": 3.174089068825911,
+      "grad_norm": 0.3104319297971531,
+      "learning_rate": 1.4175705755454963e-05,
+      "loss": 0.2205,
+      "step": 784
+    },
+    {
+      "epoch": 3.1781376518218623,
+      "grad_norm": 0.32932172959987793,
+      "learning_rate": 1.412162598317602e-05,
+      "loss": 0.1961,
+      "step": 785
+    },
+    {
+      "epoch": 3.182186234817814,
+      "grad_norm": 0.3105305818668368,
+      "learning_rate": 1.406759321421907e-05,
+      "loss": 0.2069,
+      "step": 786
+    },
+    {
+      "epoch": 3.1862348178137654,
+      "grad_norm": 0.34418410765474555,
+      "learning_rate": 1.401360788062871e-05,
+      "loss": 0.2095,
+      "step": 787
+    },
+    {
+      "epoch": 3.1902834008097165,
+      "grad_norm": 0.30176049131804755,
+      "learning_rate": 1.3959670414070204e-05,
+      "loss": 0.2017,
+      "step": 788
+    },
+    {
+      "epoch": 3.194331983805668,
+      "grad_norm": 0.30870584484769925,
+      "learning_rate": 1.3905781245826108e-05,
+      "loss": 0.2017,
+      "step": 789
+    },
+    {
+      "epoch": 3.1983805668016196,
+      "grad_norm": 0.3121165753222951,
+      "learning_rate": 1.3851940806792778e-05,
+      "loss": 0.2154,
+      "step": 790
+    },
+    {
+      "epoch": 3.2024291497975708,
+      "grad_norm": 0.29256071181358406,
+      "learning_rate": 1.379814952747693e-05,
+      "loss": 0.1891,
+      "step": 791
+    },
+    {
+      "epoch": 3.2064777327935223,
+      "grad_norm": 0.2993332459637832,
+      "learning_rate": 1.3744407837992193e-05,
+      "loss": 0.1834,
+      "step": 792
+    },
+    {
+      "epoch": 3.2105263157894735,
+      "grad_norm": 0.28546500776972045,
+      "learning_rate": 1.3690716168055692e-05,
+      "loss": 0.1946,
+      "step": 793
+    },
+    {
+      "epoch": 3.214574898785425,
+      "grad_norm": 0.3195278072524679,
+      "learning_rate": 1.363707494698459e-05,
+      "loss": 0.2058,
+      "step": 794
+    },
+    {
+      "epoch": 3.2186234817813766,
+      "grad_norm": 0.3160416563715032,
+      "learning_rate": 1.358348460369265e-05,
+      "loss": 0.2034,
+      "step": 795
+    },
+    {
+      "epoch": 3.2226720647773277,
+      "grad_norm": 0.2969403627733425,
+      "learning_rate": 1.352994556668684e-05,
+      "loss": 0.2051,
+      "step": 796
+    },
+    {
+      "epoch": 3.2267206477732793,
+      "grad_norm": 0.29988929406942283,
+      "learning_rate": 1.3476458264063865e-05,
+      "loss": 0.1933,
+      "step": 797
+    },
+    {
+      "epoch": 3.230769230769231,
+      "grad_norm": 0.3294588171954754,
+      "learning_rate": 1.342302312350678e-05,
+      "loss": 0.1927,
+      "step": 798
+    },
+    {
+      "epoch": 3.234817813765182,
+      "grad_norm": 0.2757988680619944,
+      "learning_rate": 1.3369640572281537e-05,
+      "loss": 0.2012,
+      "step": 799
+    },
+    {
+      "epoch": 3.2388663967611335,
+      "grad_norm": 0.33994395718641135,
+      "learning_rate": 1.3316311037233596e-05,
+      "loss": 0.2122,
+      "step": 800
+    },
+    {
+      "epoch": 3.242914979757085,
+      "grad_norm": 0.30623221423399766,
+      "learning_rate": 1.326303494478451e-05,
+      "loss": 0.2033,
+      "step": 801
+    },
+    {
+      "epoch": 3.246963562753036,
+      "grad_norm": 0.2902883678228579,
+      "learning_rate": 1.3209812720928472e-05,
+      "loss": 0.2075,
+      "step": 802
+    },
+    {
+      "epoch": 3.251012145748988,
+      "grad_norm": 0.3053799040228012,
+      "learning_rate": 1.315664479122898e-05,
+      "loss": 0.1998,
+      "step": 803
+    },
+    {
+      "epoch": 3.2550607287449393,
+      "grad_norm": 0.30622321119918444,
+      "learning_rate": 1.3103531580815378e-05,
+      "loss": 0.176,
+      "step": 804
+    },
+    {
+      "epoch": 3.2591093117408905,
+      "grad_norm": 0.3298714267344877,
+      "learning_rate": 1.3050473514379488e-05,
+      "loss": 0.2179,
+      "step": 805
+    },
+    {
+      "epoch": 3.263157894736842,
+      "grad_norm": 0.2969461490408847,
+      "learning_rate": 1.2997471016172188e-05,
+      "loss": 0.2146,
+      "step": 806
+    },
+    {
+      "epoch": 3.2672064777327936,
+      "grad_norm": 0.3162687093024044,
+      "learning_rate": 1.2944524510000042e-05,
+      "loss": 0.185,
+      "step": 807
+    },
+    {
+      "epoch": 3.2712550607287447,
+      "grad_norm": 0.30440346892296405,
+      "learning_rate": 1.2891634419221908e-05,
+      "loss": 0.1957,
+      "step": 808
+    },
+    {
+      "epoch": 3.2753036437246963,
+      "grad_norm": 0.3084327333540835,
+      "learning_rate": 1.2838801166745545e-05,
+      "loss": 0.203,
+      "step": 809
+    },
+    {
+      "epoch": 3.279352226720648,
+      "grad_norm": 0.31531054597235086,
+      "learning_rate": 1.2786025175024216e-05,
+      "loss": 0.1934,
+      "step": 810
+    },
+    {
+      "epoch": 3.283400809716599,
+      "grad_norm": 0.3183079096081402,
+      "learning_rate": 1.2733306866053357e-05,
+      "loss": 0.1899,
+      "step": 811
+    },
+    {
+      "epoch": 3.2874493927125505,
+      "grad_norm": 0.3190005483690834,
+      "learning_rate": 1.2680646661367163e-05,
+      "loss": 0.2027,
+      "step": 812
+    },
+    {
+      "epoch": 3.291497975708502,
+      "grad_norm": 0.311895372448144,
+      "learning_rate": 1.2628044982035219e-05,
+      "loss": 0.1834,
+      "step": 813
+    },
+    {
+      "epoch": 3.2955465587044532,
+      "grad_norm": 0.33670379813548623,
+      "learning_rate": 1.2575502248659155e-05,
+      "loss": 0.2053,
+      "step": 814
+    },
+    {
+      "epoch": 3.299595141700405,
+      "grad_norm": 0.30005263733077486,
+      "learning_rate": 1.2523018881369269e-05,
+      "loss": 0.2074,
+      "step": 815
+    },
+    {
+      "epoch": 3.3036437246963564,
+      "grad_norm": 0.30634505887949365,
+      "learning_rate": 1.2470595299821176e-05,
+      "loss": 0.2021,
+      "step": 816
+    },
+    {
+      "epoch": 3.3076923076923075,
+      "grad_norm": 0.30189560415441513,
+      "learning_rate": 1.2418231923192416e-05,
+      "loss": 0.1965,
+      "step": 817
+    },
+    {
+      "epoch": 3.311740890688259,
+      "grad_norm": 0.316115867074452,
+      "learning_rate": 1.2365929170179165e-05,
+      "loss": 0.2121,
+      "step": 818
+    },
+    {
+      "epoch": 3.3157894736842106,
+      "grad_norm": 0.30538191014553917,
+      "learning_rate": 1.2313687458992848e-05,
+      "loss": 0.217,
+      "step": 819
+    },
+    {
+      "epoch": 3.3198380566801617,
+      "grad_norm": 0.32473225068579625,
+      "learning_rate": 1.2261507207356785e-05,
+      "loss": 0.2013,
+      "step": 820
+    },
+    {
+      "epoch": 3.3238866396761133,
+      "grad_norm": 0.3085463006898807,
+      "learning_rate": 1.220938883250288e-05,
+      "loss": 0.2148,
+      "step": 821
+    },
+    {
+      "epoch": 3.327935222672065,
+      "grad_norm": 0.2939163361845062,
+      "learning_rate": 1.215733275116827e-05,
+      "loss": 0.2098,
+      "step": 822
+    },
+    {
+      "epoch": 3.331983805668016,
+      "grad_norm": 0.3387400730958776,
+      "learning_rate": 1.2105339379592005e-05,
+      "loss": 0.1982,
+      "step": 823
+    },
+    {
+      "epoch": 3.3360323886639676,
+      "grad_norm": 0.31275257287024716,
+      "learning_rate": 1.2053409133511681e-05,
+      "loss": 0.1906,
+      "step": 824
+    },
+    {
+      "epoch": 3.340080971659919,
+      "grad_norm": 0.3060975457469417,
+      "learning_rate": 1.2001542428160173e-05,
+      "loss": 0.1876,
+      "step": 825
+    },
+    {
+      "epoch": 3.3441295546558703,
+      "grad_norm": 0.3336210668753272,
+      "learning_rate": 1.1949739678262278e-05,
+      "loss": 0.2001,
+      "step": 826
+    },
+    {
+      "epoch": 3.348178137651822,
+      "grad_norm": 0.312575224785405,
+      "learning_rate": 1.189800129803139e-05,
+      "loss": 0.1943,
+      "step": 827
+    },
+    {
+      "epoch": 3.3522267206477734,
+      "grad_norm": 0.2998058199945868,
+      "learning_rate": 1.1846327701166228e-05,
+      "loss": 0.1976,
+      "step": 828
+    },
+    {
+      "epoch": 3.3562753036437245,
+      "grad_norm": 0.33581700913397133,
+      "learning_rate": 1.1794719300847493e-05,
+      "loss": 0.2118,
+      "step": 829
+    },
+    {
+      "epoch": 3.360323886639676,
+      "grad_norm": 0.291180866156165,
+      "learning_rate": 1.1743176509734582e-05,
+      "loss": 0.1909,
+      "step": 830
+    },
+    {
+      "epoch": 3.3643724696356276,
+      "grad_norm": 0.33359303278009494,
+      "learning_rate": 1.1691699739962275e-05,
+      "loss": 0.1938,
+      "step": 831
+    },
+    {
+      "epoch": 3.3684210526315788,
+      "grad_norm": 0.3097555549761639,
+      "learning_rate": 1.1640289403137456e-05,
+      "loss": 0.1841,
+      "step": 832
+    },
+    {
+      "epoch": 3.3724696356275303,
+      "grad_norm": 0.2904372592593639,
+      "learning_rate": 1.1588945910335802e-05,
+      "loss": 0.2098,
+      "step": 833
+    },
+    {
+      "epoch": 3.376518218623482,
+      "grad_norm": 0.3017739049017231,
+      "learning_rate": 1.1537669672098517e-05,
+      "loss": 0.2009,
+      "step": 834
+    },
+    {
+      "epoch": 3.380566801619433,
+      "grad_norm": 0.3138344411054483,
+      "learning_rate": 1.1486461098429036e-05,
+      "loss": 0.2165,
+      "step": 835
+    },
+    {
+      "epoch": 3.3846153846153846,
+      "grad_norm": 0.2916026260127747,
+      "learning_rate": 1.1435320598789747e-05,
+      "loss": 0.2005,
+      "step": 836
+    },
+    {
+      "epoch": 3.388663967611336,
+      "grad_norm": 0.30747658659859434,
+      "learning_rate": 1.1384248582098736e-05,
+      "loss": 0.2149,
+      "step": 837
+    },
+    {
+      "epoch": 3.3927125506072873,
+      "grad_norm": 0.3191246169542393,
+      "learning_rate": 1.1333245456726467e-05,
+      "loss": 0.1981,
+      "step": 838
+    },
+    {
+      "epoch": 3.396761133603239,
+      "grad_norm": 0.30638957002541206,
+      "learning_rate": 1.128231163049258e-05,
+      "loss": 0.1964,
+      "step": 839
+    },
+    {
+      "epoch": 3.4008097165991904,
+      "grad_norm": 0.31672833422019253,
+      "learning_rate": 1.123144751066259e-05,
+      "loss": 0.1951,
+      "step": 840
+    },
+    {
+      "epoch": 3.4048582995951415,
+      "grad_norm": 0.2976509777973181,
+      "learning_rate": 1.1180653503944657e-05,
+      "loss": 0.1948,
+      "step": 841
+    },
+    {
+      "epoch": 3.408906882591093,
+      "grad_norm": 0.26530264218551053,
+      "learning_rate": 1.1129930016486287e-05,
+      "loss": 0.1998,
+      "step": 842
+    },
+    {
+      "epoch": 3.4129554655870447,
+      "grad_norm": 0.2875960584265089,
+      "learning_rate": 1.107927745387114e-05,
+      "loss": 0.1862,
+      "step": 843
+    },
+    {
+      "epoch": 3.417004048582996,
+      "grad_norm": 0.3167529120556994,
+      "learning_rate": 1.1028696221115766e-05,
+      "loss": 0.1682,
+      "step": 844
+    },
+    {
+      "epoch": 3.4210526315789473,
+      "grad_norm": 0.30706824581839864,
+      "learning_rate": 1.0978186722666341e-05,
+      "loss": 0.1889,
+      "step": 845
+    },
+    {
+      "epoch": 3.425101214574899,
+      "grad_norm": 0.3099753820074774,
+      "learning_rate": 1.0927749362395475e-05,
+      "loss": 0.203,
+      "step": 846
+    },
+    {
+      "epoch": 3.42914979757085,
+      "grad_norm": 0.29752873505862226,
+      "learning_rate": 1.0877384543598954e-05,
+      "loss": 0.1865,
+      "step": 847
+    },
+    {
+      "epoch": 3.4331983805668016,
+      "grad_norm": 0.2955108568357823,
+      "learning_rate": 1.0827092668992529e-05,
+      "loss": 0.1974,
+      "step": 848
+    },
+    {
+      "epoch": 3.437246963562753,
+      "grad_norm": 0.2968626839105756,
+      "learning_rate": 1.077687414070869e-05,
+      "loss": 0.1895,
+      "step": 849
+    },
+    {
+      "epoch": 3.4412955465587043,
+      "grad_norm": 0.30862946961492194,
+      "learning_rate": 1.0726729360293451e-05,
+      "loss": 0.1921,
+      "step": 850
+    },
+    {
+      "epoch": 3.445344129554656,
+      "grad_norm": 0.3204379776816058,
+      "learning_rate": 1.0676658728703147e-05,
+      "loss": 0.2044,
+      "step": 851
+    },
+    {
+      "epoch": 3.4493927125506074,
+      "grad_norm": 0.29529729894902784,
+      "learning_rate": 1.0626662646301198e-05,
+      "loss": 0.201,
+      "step": 852
+    },
+    {
+      "epoch": 3.4534412955465585,
+      "grad_norm": 0.28940040361410413,
+      "learning_rate": 1.0576741512854958e-05,
+      "loss": 0.1918,
+      "step": 853
+    },
+    {
+      "epoch": 3.45748987854251,
+      "grad_norm": 0.31441918312750994,
+      "learning_rate": 1.052689572753247e-05,
+      "loss": 0.2012,
+      "step": 854
+    },
+    {
+      "epoch": 3.4615384615384617,
+      "grad_norm": 0.29241861432054833,
+      "learning_rate": 1.0477125688899328e-05,
+      "loss": 0.1932,
+      "step": 855
+    },
+    {
+      "epoch": 3.465587044534413,
+      "grad_norm": 0.3019214404281462,
+      "learning_rate": 1.0427431794915406e-05,
+      "loss": 0.1924,
+      "step": 856
+    },
+    {
+      "epoch": 3.4696356275303644,
+      "grad_norm": 0.29054485307291683,
+      "learning_rate": 1.0377814442931769e-05,
+      "loss": 0.2086,
+      "step": 857
+    },
+    {
+      "epoch": 3.473684210526316,
+      "grad_norm": 0.2687964609903366,
+      "learning_rate": 1.0328274029687439e-05,
+      "loss": 0.2081,
+      "step": 858
+    },
+    {
+      "epoch": 3.477732793522267,
+      "grad_norm": 0.2989045579746449,
+      "learning_rate": 1.0278810951306253e-05,
+      "loss": 0.202,
+      "step": 859
+    },
+    {
+      "epoch": 3.4817813765182186,
+      "grad_norm": 0.29600389462073884,
+      "learning_rate": 1.0229425603293647e-05,
+      "loss": 0.1949,
+      "step": 860
+    },
+    {
+      "epoch": 3.48582995951417,
+      "grad_norm": 0.2981382680576097,
+      "learning_rate": 1.0180118380533558e-05,
+      "loss": 0.1578,
+      "step": 861
+    },
+    {
+      "epoch": 3.4898785425101213,
+      "grad_norm": 0.2701856565087567,
+      "learning_rate": 1.0130889677285225e-05,
+      "loss": 0.2135,
+      "step": 862
+    },
+    {
+      "epoch": 3.493927125506073,
+      "grad_norm": 0.3117025247197708,
+      "learning_rate": 1.008173988718005e-05,
+      "loss": 0.1854,
+      "step": 863
+    },
+    {
+      "epoch": 3.4979757085020244,
+      "grad_norm": 0.28699361782559013,
+      "learning_rate": 1.0032669403218439e-05,
+      "loss": 0.2143,
+      "step": 864
+    },
+    {
+      "epoch": 3.5020242914979756,
+      "grad_norm": 0.28038990291832444,
+      "learning_rate": 9.983678617766677e-06,
+      "loss": 0.2117,
+      "step": 865
+    },
+    {
+      "epoch": 3.506072874493927,
+      "grad_norm": 0.3029319629631064,
+      "learning_rate": 9.934767922553788e-06,
+      "loss": 0.1991,
+      "step": 866
+    },
+    {
+      "epoch": 3.5101214574898787,
+      "grad_norm": 0.2864477269870901,
+      "learning_rate": 9.885937708668371e-06,
+      "loss": 0.2111,
+      "step": 867
+    },
+    {
+      "epoch": 3.51417004048583,
+      "grad_norm": 0.2778865086170692,
+      "learning_rate": 9.837188366555524e-06,
+      "loss": 0.2013,
+      "step": 868
+    },
+    {
+      "epoch": 3.5182186234817814,
+      "grad_norm": 0.2895835932529354,
+      "learning_rate": 9.788520286013691e-06,
+      "loss": 0.1999,
+      "step": 869
+    },
+    {
+      "epoch": 3.522267206477733,
+      "grad_norm": 0.2940136491809492,
+      "learning_rate": 9.739933856191537e-06,
+      "loss": 0.198,
+      "step": 870
+    },
+    {
+      "epoch": 3.526315789473684,
+      "grad_norm": 0.27380411690271356,
+      "learning_rate": 9.691429465584866e-06,
+      "loss": 0.2005,
+      "step": 871
+    },
+    {
+      "epoch": 3.5303643724696356,
+      "grad_norm": 0.2831710848148934,
+      "learning_rate": 9.643007502033492e-06,
+      "loss": 0.2127,
+      "step": 872
+    },
+    {
+      "epoch": 3.534412955465587,
+      "grad_norm": 0.28141613506501184,
+      "learning_rate": 9.594668352718152e-06,
+      "loss": 0.1893,
+      "step": 873
+    },
+    {
+      "epoch": 3.5384615384615383,
+      "grad_norm": 0.29213978228504695,
+      "learning_rate": 9.546412404157383e-06,
+      "loss": 0.209,
+      "step": 874
+    },
+    {
+      "epoch": 3.54251012145749,
+      "grad_norm": 0.282089372372099,
+      "learning_rate": 9.498240042204471e-06,
+      "loss": 0.2003,
+      "step": 875
+    },
+    {
+      "epoch": 3.5465587044534415,
+      "grad_norm": 0.3076322505788656,
+      "learning_rate": 9.450151652044339e-06,
+      "loss": 0.1819,
+      "step": 876
+    },
+    {
+      "epoch": 3.5506072874493926,
+      "grad_norm": 0.2761947380297417,
+      "learning_rate": 9.402147618190472e-06,
+      "loss": 0.2106,
+      "step": 877
+    },
+    {
+      "epoch": 3.554655870445344,
+      "grad_norm": 0.2731562534787993,
+      "learning_rate": 9.354228324481841e-06,
+      "loss": 0.2089,
+      "step": 878
+    },
+    {
+      "epoch": 3.5587044534412957,
+      "grad_norm": 0.28148325029877713,
+      "learning_rate": 9.306394154079852e-06,
+      "loss": 0.2033,
+      "step": 879
+    },
+    {
+      "epoch": 3.562753036437247,
+      "grad_norm": 0.29341174551913923,
+      "learning_rate": 9.258645489465254e-06,
+      "loss": 0.2008,
+      "step": 880
+    },
+    {
+      "epoch": 3.5668016194331984,
+      "grad_norm": 0.3006118752409928,
+      "learning_rate": 9.21098271243509e-06,
+      "loss": 0.2008,
+      "step": 881
+    },
+    {
+      "epoch": 3.57085020242915,
+      "grad_norm": 0.28872857575888483,
+      "learning_rate": 9.163406204099659e-06,
+      "loss": 0.2112,
+      "step": 882
+    },
+    {
+      "epoch": 3.574898785425101,
+      "grad_norm": 0.27342463199453526,
+      "learning_rate": 9.115916344879452e-06,
+      "loss": 0.196,
+      "step": 883
+    },
+    {
+      "epoch": 3.5789473684210527,
+      "grad_norm": 0.3004322366387542,
+      "learning_rate": 9.068513514502126e-06,
+      "loss": 0.1848,
+      "step": 884
+    },
+    {
+      "epoch": 3.582995951417004,
+      "grad_norm": 0.2780074255566303,
+      "learning_rate": 9.021198091999439e-06,
+      "loss": 0.1963,
+      "step": 885
+    },
+    {
+      "epoch": 3.5870445344129553,
+      "grad_norm": 0.288851907586006,
+      "learning_rate": 8.973970455704251e-06,
+      "loss": 0.1999,
+      "step": 886
+    },
+    {
+      "epoch": 3.591093117408907,
+      "grad_norm": 0.27973637726474954,
+      "learning_rate": 8.926830983247495e-06,
+      "loss": 0.2072,
+      "step": 887
+    },
+    {
+      "epoch": 3.5951417004048585,
+      "grad_norm": 0.2932202581297176,
+      "learning_rate": 8.879780051555116e-06,
+      "loss": 0.1966,
+      "step": 888
+    },
+    {
+      "epoch": 3.5991902834008096,
+      "grad_norm": 0.2944128460242157,
+      "learning_rate": 8.83281803684512e-06,
+      "loss": 0.2136,
+      "step": 889
+    },
+    {
+      "epoch": 3.603238866396761,
+      "grad_norm": 0.26848438495769633,
+      "learning_rate": 8.785945314624532e-06,
+      "loss": 0.207,
+      "step": 890
+    },
+    {
+      "epoch": 3.6072874493927127,
+      "grad_norm": 0.2809573728438526,
+      "learning_rate": 8.739162259686385e-06,
+      "loss": 0.2106,
+      "step": 891
+    },
+    {
+      "epoch": 3.611336032388664,
+      "grad_norm": 0.27409047738220144,
+      "learning_rate": 8.692469246106734e-06,
+      "loss": 0.204,
+      "step": 892
+    },
+    {
+      "epoch": 3.6153846153846154,
+      "grad_norm": 0.26499781089451413,
+      "learning_rate": 8.645866647241677e-06,
+      "loss": 0.1986,
+      "step": 893
+    },
+    {
+      "epoch": 3.619433198380567,
+      "grad_norm": 0.2865378810117944,
+      "learning_rate": 8.599354835724348e-06,
+      "loss": 0.2015,
+      "step": 894
+    },
+    {
+      "epoch": 3.623481781376518,
+      "grad_norm": 0.30326938312243606,
+      "learning_rate": 8.552934183461958e-06,
+      "loss": 0.1963,
+      "step": 895
+    },
+    {
+      "epoch": 3.6275303643724697,
+      "grad_norm": 0.2849631573485576,
+      "learning_rate": 8.506605061632805e-06,
+      "loss": 0.2017,
+      "step": 896
+    },
+    {
+      "epoch": 3.6315789473684212,
+      "grad_norm": 0.27646110361872434,
+      "learning_rate": 8.460367840683308e-06,
+      "loss": 0.1979,
+      "step": 897
+    },
+    {
+      "epoch": 3.6356275303643724,
+      "grad_norm": 0.2836154771244939,
+      "learning_rate": 8.414222890325065e-06,
+      "loss": 0.2143,
+      "step": 898
+    },
+    {
+      "epoch": 3.639676113360324,
+      "grad_norm": 0.29833010777546465,
+      "learning_rate": 8.368170579531855e-06,
+      "loss": 0.2111,
+      "step": 899
+    },
+    {
+      "epoch": 3.6437246963562755,
+      "grad_norm": 0.2603052113301202,
+      "learning_rate": 8.322211276536733e-06,
+      "loss": 0.2049,
+      "step": 900
+    },
+    {
+      "epoch": 3.6477732793522266,
+      "grad_norm": 0.2854502636450143,
+      "learning_rate": 8.27634534882907e-06,
+      "loss": 0.2074,
+      "step": 901
+    },
+    {
+      "epoch": 3.651821862348178,
+      "grad_norm": 0.29103575829699463,
+      "learning_rate": 8.230573163151585e-06,
+      "loss": 0.1951,
+      "step": 902
+    },
+    {
+      "epoch": 3.6558704453441297,
+      "grad_norm": 0.2974198994385354,
+      "learning_rate": 8.184895085497466e-06,
+      "loss": 0.1843,
+      "step": 903
+    },
+    {
+      "epoch": 3.659919028340081,
+      "grad_norm": 0.27861674607051423,
+      "learning_rate": 8.139311481107401e-06,
+      "loss": 0.1953,
+      "step": 904
+    },
+    {
+      "epoch": 3.6639676113360324,
+      "grad_norm": 0.29875818316349895,
+      "learning_rate": 8.093822714466686e-06,
+      "loss": 0.1989,
+      "step": 905
+    },
+    {
+      "epoch": 3.668016194331984,
+      "grad_norm": 0.30468929762473973,
+      "learning_rate": 8.048429149302271e-06,
+      "loss": 0.2173,
+      "step": 906
+    },
+    {
+      "epoch": 3.672064777327935,
+      "grad_norm": 0.2742581578227668,
+      "learning_rate": 8.003131148579902e-06,
+      "loss": 0.1983,
+      "step": 907
+    },
+    {
+      "epoch": 3.6761133603238867,
+      "grad_norm": 0.3210674869503059,
+      "learning_rate": 7.957929074501182e-06,
+      "loss": 0.2119,
+      "step": 908
+    },
+    {
+      "epoch": 3.6801619433198383,
+      "grad_norm": 0.27777020722669926,
+      "learning_rate": 7.912823288500695e-06,
+      "loss": 0.201,
+      "step": 909
+    },
+    {
+      "epoch": 3.6842105263157894,
+      "grad_norm": 0.287674236045348,
+      "learning_rate": 7.867814151243102e-06,
+      "loss": 0.1931,
+      "step": 910
+    },
+    {
+      "epoch": 3.688259109311741,
+      "grad_norm": 0.2712456399154733,
+      "learning_rate": 7.822902022620267e-06,
+      "loss": 0.1974,
+      "step": 911
+    },
+    {
+      "epoch": 3.6923076923076925,
+      "grad_norm": 0.28355512521702314,
+      "learning_rate": 7.778087261748377e-06,
+      "loss": 0.211,
+      "step": 912
+    },
+    {
+      "epoch": 3.6963562753036436,
+      "grad_norm": 0.27569761310558827,
+      "learning_rate": 7.733370226965056e-06,
+      "loss": 0.1822,
+      "step": 913
+    },
+    {
+      "epoch": 3.700404858299595,
+      "grad_norm": 0.2711457530100778,
+      "learning_rate": 7.688751275826527e-06,
+      "loss": 0.2107,
+      "step": 914
+    },
+    {
+      "epoch": 3.7044534412955468,
+      "grad_norm": 0.27697208544415575,
+      "learning_rate": 7.644230765104735e-06,
+      "loss": 0.1916,
+      "step": 915
+    },
+    {
+      "epoch": 3.708502024291498,
+      "grad_norm": 0.2694376586528273,
+      "learning_rate": 7.599809050784502e-06,
+      "loss": 0.1895,
+      "step": 916
+    },
+    {
+      "epoch": 3.7125506072874495,
+      "grad_norm": 0.27273578501547946,
+      "learning_rate": 7.555486488060661e-06,
+      "loss": 0.2025,
+      "step": 917
+    },
+    {
+      "epoch": 3.716599190283401,
+      "grad_norm": 0.281806923836313,
+      "learning_rate": 7.511263431335249e-06,
+      "loss": 0.2053,
+      "step": 918
+    },
+    {
+      "epoch": 3.720647773279352,
+      "grad_norm": 0.29396271750603786,
+      "learning_rate": 7.467140234214656e-06,
+      "loss": 0.198,
+      "step": 919
+    },
+    {
+      "epoch": 3.7246963562753037,
+      "grad_norm": 0.2824154921130078,
+      "learning_rate": 7.423117249506777e-06,
+      "loss": 0.1897,
+      "step": 920
+    },
+    {
+      "epoch": 3.7287449392712553,
+      "grad_norm": 0.27519786912568633,
+      "learning_rate": 7.379194829218231e-06,
+      "loss": 0.1867,
+      "step": 921
+    },
+    {
+      "epoch": 3.7327935222672064,
+      "grad_norm": 0.28393990037148115,
+      "learning_rate": 7.335373324551518e-06,
+      "loss": 0.2078,
+      "step": 922
+    },
+    {
+      "epoch": 3.736842105263158,
+      "grad_norm": 0.2746207361198436,
+      "learning_rate": 7.291653085902224e-06,
+      "loss": 0.196,
+      "step": 923
+    },
+    {
+      "epoch": 3.7408906882591095,
+      "grad_norm": 0.28668571420934236,
+      "learning_rate": 7.248034462856213e-06,
+      "loss": 0.1916,
+      "step": 924
+    },
+    {
+      "epoch": 3.7449392712550607,
+      "grad_norm": 0.27869601434965496,
+      "learning_rate": 7.20451780418683e-06,
+      "loss": 0.1741,
+      "step": 925
+    },
+    {
+      "epoch": 3.748987854251012,
+      "grad_norm": 0.2798887918305361,
+      "learning_rate": 7.161103457852128e-06,
+      "loss": 0.2167,
+      "step": 926
+    },
+    {
+      "epoch": 3.753036437246964,
+      "grad_norm": 0.2747496116936131,
+      "learning_rate": 7.117791770992046e-06,
+      "loss": 0.1953,
+      "step": 927
+    },
+    {
+      "epoch": 3.757085020242915,
+      "grad_norm": 0.27934709571159855,
+      "learning_rate": 7.074583089925688e-06,
+      "loss": 0.1868,
+      "step": 928
+    },
+    {
+      "epoch": 3.7611336032388665,
+      "grad_norm": 0.27994463705860717,
+      "learning_rate": 7.0314777601485155e-06,
+      "loss": 0.1879,
+      "step": 929
+    },
+    {
+      "epoch": 3.765182186234818,
+      "grad_norm": 0.30072353784599054,
+      "learning_rate": 6.988476126329604e-06,
+      "loss": 0.2004,
+      "step": 930
+    },
+    {
+      "epoch": 3.769230769230769,
+      "grad_norm": 0.27829467051088785,
+      "learning_rate": 6.945578532308857e-06,
+      "loss": 0.2022,
+      "step": 931
+    },
+    {
+      "epoch": 3.7732793522267207,
+      "grad_norm": 0.2748344413786139,
+      "learning_rate": 6.902785321094301e-06,
+      "loss": 0.1932,
+      "step": 932
+    },
+    {
+      "epoch": 3.7773279352226723,
+      "grad_norm": 0.3024954759509495,
+      "learning_rate": 6.8600968348593115e-06,
+      "loss": 0.1876,
+      "step": 933
+    },
+    {
+      "epoch": 3.7813765182186234,
+      "grad_norm": 0.28543773111140947,
+      "learning_rate": 6.817513414939892e-06,
+      "loss": 0.1983,
+      "step": 934
+    },
+    {
+      "epoch": 3.785425101214575,
+      "grad_norm": 0.26428824051100547,
+      "learning_rate": 6.775035401831924e-06,
+      "loss": 0.2041,
+      "step": 935
+    },
+    {
+      "epoch": 3.7894736842105265,
+      "grad_norm": 0.27692085251791543,
+      "learning_rate": 6.732663135188473e-06,
+      "loss": 0.2005,
+      "step": 936
+    },
+    {
+      "epoch": 3.7935222672064777,
+      "grad_norm": 0.28528202975791167,
+      "learning_rate": 6.690396953817055e-06,
+      "loss": 0.1879,
+      "step": 937
+    },
+    {
+      "epoch": 3.7975708502024292,
+      "grad_norm": 0.257713893985085,
+      "learning_rate": 6.648237195676927e-06,
+      "loss": 0.1939,
+      "step": 938
+    },
+    {
+      "epoch": 3.801619433198381,
+      "grad_norm": 0.2803846921119148,
+      "learning_rate": 6.606184197876395e-06,
+      "loss": 0.2071,
+      "step": 939
+    },
+    {
+      "epoch": 3.805668016194332,
+      "grad_norm": 0.2749900730775925,
+      "learning_rate": 6.564238296670109e-06,
+      "loss": 0.1944,
+      "step": 940
+    },
+    {
+      "epoch": 3.8097165991902835,
+      "grad_norm": 0.2803604977202947,
+      "learning_rate": 6.52239982745638e-06,
+      "loss": 0.211,
+      "step": 941
+    },
+    {
+      "epoch": 3.813765182186235,
+      "grad_norm": 0.26873976462823956,
+      "learning_rate": 6.480669124774484e-06,
+      "loss": 0.2065,
+      "step": 942
+    },
+    {
+      "epoch": 3.817813765182186,
+      "grad_norm": 0.2880002077278946,
+      "learning_rate": 6.439046522302013e-06,
+      "loss": 0.1864,
+      "step": 943
+    },
+    {
+      "epoch": 3.8218623481781377,
+      "grad_norm": 0.26357125153260474,
+      "learning_rate": 6.397532352852191e-06,
+      "loss": 0.2086,
+      "step": 944
+    },
+    {
+      "epoch": 3.8259109311740893,
+      "grad_norm": 0.27084184692189106,
+      "learning_rate": 6.356126948371202e-06,
+      "loss": 0.1892,
+      "step": 945
+    },
+    {
+      "epoch": 3.8299595141700404,
+      "grad_norm": 0.2820979951374823,
+      "learning_rate": 6.314830639935559e-06,
+      "loss": 0.1861,
+      "step": 946
+    },
+    {
+      "epoch": 3.834008097165992,
+      "grad_norm": 0.2805226350040575,
+      "learning_rate": 6.273643757749448e-06,
+      "loss": 0.2037,
+      "step": 947
+    },
+    {
+      "epoch": 3.8380566801619436,
+      "grad_norm": 0.2905137765817934,
+      "learning_rate": 6.2325666311420856e-06,
+      "loss": 0.1886,
+      "step": 948
+    },
+    {
+      "epoch": 3.8421052631578947,
+      "grad_norm": 0.2954721804174607,
+      "learning_rate": 6.1915995885650675e-06,
+      "loss": 0.2063,
+      "step": 949
+    },
+    {
+      "epoch": 3.8461538461538463,
+      "grad_norm": 0.28419119284639244,
+      "learning_rate": 6.150742957589784e-06,
+      "loss": 0.2002,
+      "step": 950
+    },
+    {
+      "epoch": 3.850202429149798,
+      "grad_norm": 0.26617129578501797,
+      "learning_rate": 6.109997064904774e-06,
+      "loss": 0.2034,
+      "step": 951
+    },
+    {
+      "epoch": 3.854251012145749,
+      "grad_norm": 0.2612920547129711,
+      "learning_rate": 6.069362236313092e-06,
+      "loss": 0.1898,
+      "step": 952
+    },
+    {
+      "epoch": 3.8582995951417005,
+      "grad_norm": 0.2832244385700315,
+      "learning_rate": 6.0288387967297565e-06,
+      "loss": 0.2093,
+      "step": 953
+    },
+    {
+      "epoch": 3.862348178137652,
+      "grad_norm": 0.2710024397380795,
+      "learning_rate": 5.988427070179108e-06,
+      "loss": 0.1915,
+      "step": 954
+    },
+    {
+      "epoch": 3.866396761133603,
+      "grad_norm": 0.2753206384501534,
+      "learning_rate": 5.948127379792234e-06,
+      "loss": 0.2063,
+      "step": 955
+    },
+    {
+      "epoch": 3.8704453441295548,
+      "grad_norm": 0.26397782813633025,
+      "learning_rate": 5.90794004780439e-06,
+      "loss": 0.2069,
+      "step": 956
+    },
+    {
+      "epoch": 3.8744939271255063,
+      "grad_norm": 0.27357815167313954,
+      "learning_rate": 5.867865395552405e-06,
+      "loss": 0.2053,
+      "step": 957
+    },
+    {
+      "epoch": 3.8785425101214575,
+      "grad_norm": 0.27377151569451785,
+      "learning_rate": 5.8279037434721364e-06,
+      "loss": 0.1907,
+      "step": 958
+    },
+    {
+      "epoch": 3.882591093117409,
+      "grad_norm": 0.2845900678328638,
+      "learning_rate": 5.788055411095892e-06,
+      "loss": 0.1817,
+      "step": 959
+    },
+    {
+      "epoch": 3.8866396761133606,
+      "grad_norm": 0.29352462242090666,
+      "learning_rate": 5.7483207170498625e-06,
+      "loss": 0.206,
+      "step": 960
+    },
+    {
+      "epoch": 3.8906882591093117,
+      "grad_norm": 0.2696463974105453,
+      "learning_rate": 5.708699979051613e-06,
+      "loss": 0.202,
+      "step": 961
+    },
+    {
+      "epoch": 3.8947368421052633,
+      "grad_norm": 0.26836151365053984,
+      "learning_rate": 5.66919351390751e-06,
+      "loss": 0.2055,
+      "step": 962
+    },
+    {
+      "epoch": 3.898785425101215,
+      "grad_norm": 0.2787339693654075,
+      "learning_rate": 5.629801637510188e-06,
+      "loss": 0.1938,
+      "step": 963
+    },
+    {
+      "epoch": 3.902834008097166,
+      "grad_norm": 0.27768288105084105,
+      "learning_rate": 5.590524664836043e-06,
+      "loss": 0.1968,
+      "step": 964
+    },
+    {
+      "epoch": 3.9068825910931175,
+      "grad_norm": 0.26746007927501125,
+      "learning_rate": 5.551362909942708e-06,
+      "loss": 0.2097,
+      "step": 965
+    },
+    {
+      "epoch": 3.910931174089069,
+      "grad_norm": 0.2792009774574196,
+      "learning_rate": 5.51231668596653e-06,
+      "loss": 0.1952,
+      "step": 966
+    },
+    {
+      "epoch": 3.91497975708502,
+      "grad_norm": 0.28036744222908533,
+      "learning_rate": 5.47338630512007e-06,
+      "loss": 0.1789,
+      "step": 967
+    },
+    {
+      "epoch": 3.919028340080972,
+      "grad_norm": 0.2855717073581853,
+      "learning_rate": 5.434572078689617e-06,
+      "loss": 0.1879,
+      "step": 968
+    },
+    {
+      "epoch": 3.9230769230769234,
+      "grad_norm": 0.27511397575929836,
+      "learning_rate": 5.395874317032694e-06,
+      "loss": 0.1856,
+      "step": 969
+    },
+    {
+      "epoch": 3.9271255060728745,
+      "grad_norm": 0.281832300592089,
+      "learning_rate": 5.357293329575573e-06,
+      "loss": 0.2103,
+      "step": 970
+    },
+    {
+      "epoch": 3.931174089068826,
+      "grad_norm": 0.28073798009315115,
+      "learning_rate": 5.318829424810797e-06,
+      "loss": 0.1927,
+      "step": 971
+    },
+    {
+      "epoch": 3.9352226720647776,
+      "grad_norm": 0.28700436712319904,
+      "learning_rate": 5.280482910294728e-06,
+      "loss": 0.1832,
+      "step": 972
+    },
+    {
+      "epoch": 3.9392712550607287,
+      "grad_norm": 0.25518404200848877,
+      "learning_rate": 5.242254092645078e-06,
+      "loss": 0.1964,
+      "step": 973
+    },
+    {
+      "epoch": 3.9433198380566803,
+      "grad_norm": 0.2611927348611384,
+      "learning_rate": 5.2041432775384445e-06,
+      "loss": 0.2054,
+      "step": 974
+    },
+    {
+      "epoch": 3.9473684210526314,
+      "grad_norm": 0.27628223901303783,
+      "learning_rate": 5.166150769707892e-06,
+      "loss": 0.1915,
+      "step": 975
+    },
+    {
+      "epoch": 3.951417004048583,
+      "grad_norm": 0.2844781851304111,
+      "learning_rate": 5.128276872940503e-06,
+      "loss": 0.2026,
+      "step": 976
+    },
+    {
+      "epoch": 3.9554655870445345,
+      "grad_norm": 0.27058945062246503,
+      "learning_rate": 5.090521890074953e-06,
+      "loss": 0.2009,
+      "step": 977
+    },
+    {
+      "epoch": 3.9595141700404857,
+      "grad_norm": 0.2695108974372428,
+      "learning_rate": 5.0528861229990675e-06,
+      "loss": 0.1962,
+      "step": 978
+    },
+    {
+      "epoch": 3.9635627530364372,
+      "grad_norm": 0.2775507777175757,
+      "learning_rate": 5.015369872647442e-06,
+      "loss": 0.2004,
+      "step": 979
+    },
+    {
+      "epoch": 3.967611336032389,
+      "grad_norm": 0.2943813366272714,
+      "learning_rate": 4.9779734389990195e-06,
+      "loss": 0.1792,
+      "step": 980
+    },
+    {
+      "epoch": 3.97165991902834,
+      "grad_norm": 0.28359252110871846,
+      "learning_rate": 4.940697121074675e-06,
+      "loss": 0.2043,
+      "step": 981
+    },
+    {
+      "epoch": 3.9757085020242915,
+      "grad_norm": 0.283059174411216,
+      "learning_rate": 4.903541216934864e-06,
+      "loss": 0.2109,
+      "step": 982
+    },
+    {
+      "epoch": 3.979757085020243,
+      "grad_norm": 0.2734151231132047,
+      "learning_rate": 4.8665060236772e-06,
+      "loss": 0.1778,
+      "step": 983
+    },
+    {
+      "epoch": 3.983805668016194,
+      "grad_norm": 0.2832928065916919,
+      "learning_rate": 4.829591837434102e-06,
+      "loss": 0.1973,
+      "step": 984
+    },
+    {
+      "epoch": 3.9878542510121457,
+      "grad_norm": 0.28128685985849045,
+      "learning_rate": 4.792798953370424e-06,
+      "loss": 0.2162,
+      "step": 985
+    },
+    {
+      "epoch": 3.9919028340080973,
+      "grad_norm": 0.2798899436131163,
+      "learning_rate": 4.756127665681083e-06,
+      "loss": 0.1891,
+      "step": 986
+    },
+    {
+      "epoch": 3.9959514170040484,
+      "grad_norm": 0.2716367401868908,
+      "learning_rate": 4.719578267588725e-06,
+      "loss": 0.2014,
+      "step": 987
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 0.2787794689124958,
+      "learning_rate": 4.6831510513413525e-06,
+      "loss": 0.1803,
+      "step": 988
+    },
+    {
+      "epoch": 4.004048582995951,
+      "grad_norm": 0.4644246528837289,
+      "learning_rate": 4.6468463082100176e-06,
+      "loss": 0.1481,
+      "step": 989
+    },
+    {
+      "epoch": 4.008097165991903,
+      "grad_norm": 0.37537465961527033,
+      "learning_rate": 4.610664328486485e-06,
+      "loss": 0.1415,
+      "step": 990
+    },
+    {
+      "epoch": 4.012145748987854,
+      "grad_norm": 0.28797902030436007,
+      "learning_rate": 4.5746054014809e-06,
+      "loss": 0.1563,
+      "step": 991
+    },
+    {
+      "epoch": 4.016194331983805,
+      "grad_norm": 0.3224045791743065,
+      "learning_rate": 4.5386698155194766e-06,
+      "loss": 0.1347,
+      "step": 992
+    },
+    {
+      "epoch": 4.020242914979757,
+      "grad_norm": 0.4025680380841487,
+      "learning_rate": 4.502857857942204e-06,
+      "loss": 0.1377,
+      "step": 993
+    },
+    {
+      "epoch": 4.0242914979757085,
+      "grad_norm": 0.47540203079815374,
+      "learning_rate": 4.467169815100549e-06,
+      "loss": 0.1534,
+      "step": 994
+    },
+    {
+      "epoch": 4.02834008097166,
+      "grad_norm": 0.36859927512421947,
+      "learning_rate": 4.431605972355137e-06,
+      "loss": 0.1412,
+      "step": 995
+    },
+    {
+      "epoch": 4.032388663967612,
+      "grad_norm": 0.2945818803908006,
+      "learning_rate": 4.396166614073516e-06,
+      "loss": 0.1536,
+      "step": 996
+    },
+    {
+      "epoch": 4.036437246963563,
+      "grad_norm": 0.35299674033199374,
+      "learning_rate": 4.360852023627846e-06,
+      "loss": 0.1462,
+      "step": 997
+    },
+    {
+      "epoch": 4.040485829959514,
+      "grad_norm": 0.37193437192031376,
+      "learning_rate": 4.325662483392656e-06,
+      "loss": 0.1326,
+      "step": 998
+    },
+    {
+      "epoch": 4.044534412955466,
+      "grad_norm": 0.3728402201560996,
+      "learning_rate": 4.29059827474257e-06,
+      "loss": 0.1329,
+      "step": 999
+    },
+    {
+      "epoch": 4.048582995951417,
+      "grad_norm": 0.3023897183785313,
+      "learning_rate": 4.255659678050063e-06,
+      "loss": 0.1516,
+      "step": 1000
+    },
+    {
+      "epoch": 4.052631578947368,
+      "grad_norm": 0.36598348692763033,
+      "learning_rate": 4.22084697268323e-06,
+      "loss": 0.1473,
+      "step": 1001
+    },
+    {
+      "epoch": 4.05668016194332,
+      "grad_norm": 0.30828489315794505,
+      "learning_rate": 4.186160437003535e-06,
+      "loss": 0.1494,
+      "step": 1002
+    },
+    {
+      "epoch": 4.060728744939271,
+      "grad_norm": 0.3546847213677791,
+      "learning_rate": 4.151600348363582e-06,
+      "loss": 0.1515,
+      "step": 1003
+    },
+    {
+      "epoch": 4.064777327935222,
+      "grad_norm": 0.3092351379167999,
+      "learning_rate": 4.11716698310493e-06,
+      "loss": 0.1558,
+      "step": 1004
+    },
+    {
+      "epoch": 4.068825910931174,
+      "grad_norm": 0.3034713479163597,
+      "learning_rate": 4.082860616555848e-06,
+      "loss": 0.1524,
+      "step": 1005
+    },
+    {
+      "epoch": 4.0728744939271255,
+      "grad_norm": 0.2643750055449553,
+      "learning_rate": 4.048681523029125e-06,
+      "loss": 0.1484,
+      "step": 1006
+    },
+    {
+      "epoch": 4.076923076923077,
+      "grad_norm": 0.2749196104247523,
+      "learning_rate": 4.014629975819882e-06,
+      "loss": 0.138,
+      "step": 1007
+    },
+    {
+      "epoch": 4.080971659919029,
+      "grad_norm": 0.2684689490337929,
+      "learning_rate": 3.9807062472033855e-06,
+      "loss": 0.1348,
+      "step": 1008
+    },
+    {
+      "epoch": 4.08502024291498,
+      "grad_norm": 0.27887255932911664,
+      "learning_rate": 3.94691060843287e-06,
+      "loss": 0.1414,
+      "step": 1009
+    },
+    {
+      "epoch": 4.089068825910931,
+      "grad_norm": 0.27789829736986843,
+      "learning_rate": 3.913243329737353e-06,
+      "loss": 0.1534,
+      "step": 1010
+    },
+    {
+      "epoch": 4.093117408906883,
+      "grad_norm": 0.263544072270295,
+      "learning_rate": 3.879704680319503e-06,
+      "loss": 0.1531,
+      "step": 1011
+    },
+    {
+      "epoch": 4.097165991902834,
+      "grad_norm": 0.26842353633268445,
+      "learning_rate": 3.846294928353471e-06,
+      "loss": 0.1529,
+      "step": 1012
+    },
+    {
+      "epoch": 4.101214574898785,
+      "grad_norm": 0.2891563112928309,
+      "learning_rate": 3.8130143409827345e-06,
+      "loss": 0.1243,
+      "step": 1013
+    },
+    {
+      "epoch": 4.105263157894737,
+      "grad_norm": 0.26204840811108693,
+      "learning_rate": 3.7798631843179887e-06,
+      "loss": 0.1325,
+      "step": 1014
+    },
+    {
+      "epoch": 4.109311740890688,
+      "grad_norm": 0.2647169094329937,
+      "learning_rate": 3.7468417234349973e-06,
+      "loss": 0.1333,
+      "step": 1015
+    },
+    {
+      "epoch": 4.113360323886639,
+      "grad_norm": 0.2771123462073878,
+      "learning_rate": 3.7139502223724823e-06,
+      "loss": 0.141,
+      "step": 1016
+    },
+    {
+      "epoch": 4.117408906882591,
+      "grad_norm": 0.26018487494199866,
+      "learning_rate": 3.6811889441300095e-06,
+      "loss": 0.1565,
+      "step": 1017
+    },
+    {
+      "epoch": 4.1214574898785425,
+      "grad_norm": 0.25877894186381,
+      "learning_rate": 3.648558150665886e-06,
+      "loss": 0.1417,
+      "step": 1018
+    },
+    {
+      "epoch": 4.125506072874494,
+      "grad_norm": 0.24947178515509208,
+      "learning_rate": 3.616058102895075e-06,
+      "loss": 0.1464,
+      "step": 1019
+    },
+    {
+      "epoch": 4.129554655870446,
+      "grad_norm": 0.2474144995698884,
+      "learning_rate": 3.583689060687079e-06,
+      "loss": 0.1353,
+      "step": 1020
+    },
+    {
+      "epoch": 4.133603238866397,
+      "grad_norm": 0.25248736232414076,
+      "learning_rate": 3.5514512828639023e-06,
+      "loss": 0.1367,
+      "step": 1021
+    },
+    {
+      "epoch": 4.137651821862348,
+      "grad_norm": 0.2627602979367308,
+      "learning_rate": 3.51934502719796e-06,
+      "loss": 0.1477,
+      "step": 1022
+    },
+    {
+      "epoch": 4.1417004048583,
+      "grad_norm": 0.2607706359613178,
+      "learning_rate": 3.4873705504100185e-06,
+      "loss": 0.1408,
+      "step": 1023
+    },
+    {
+      "epoch": 4.145748987854251,
+      "grad_norm": 0.2650694979240986,
+      "learning_rate": 3.4555281081671345e-06,
+      "loss": 0.151,
+      "step": 1024
+    },
+    {
+      "epoch": 4.149797570850202,
+      "grad_norm": 0.25298884112211295,
+      "learning_rate": 3.4238179550806326e-06,
+      "loss": 0.1458,
+      "step": 1025
+    },
+    {
+      "epoch": 4.153846153846154,
+      "grad_norm": 0.25386071328852566,
+      "learning_rate": 3.3922403447040497e-06,
+      "loss": 0.1401,
+      "step": 1026
+    },
+    {
+      "epoch": 4.157894736842105,
+      "grad_norm": 0.27304866020875074,
+      "learning_rate": 3.3607955295311247e-06,
+      "loss": 0.141,
+      "step": 1027
+    },
+    {
+      "epoch": 4.161943319838056,
+      "grad_norm": 0.2534873774897347,
+      "learning_rate": 3.32948376099375e-06,
+      "loss": 0.1351,
+      "step": 1028
+    },
+    {
+      "epoch": 4.165991902834008,
+      "grad_norm": 0.26538177087847864,
+      "learning_rate": 3.2983052894599976e-06,
+      "loss": 0.1128,
+      "step": 1029
+    },
+    {
+      "epoch": 4.17004048582996,
+      "grad_norm": 0.25775589889137185,
+      "learning_rate": 3.2672603642320945e-06,
+      "loss": 0.1426,
+      "step": 1030
+    },
+    {
+      "epoch": 4.174089068825911,
+      "grad_norm": 0.26735108487928666,
+      "learning_rate": 3.236349233544438e-06,
+      "loss": 0.1299,
+      "step": 1031
+    },
+    {
+      "epoch": 4.178137651821863,
+      "grad_norm": 0.24813269054708476,
+      "learning_rate": 3.205572144561604e-06,
+      "loss": 0.1633,
+      "step": 1032
+    },
+    {
+      "epoch": 4.182186234817814,
+      "grad_norm": 0.25263682503819335,
+      "learning_rate": 3.174929343376374e-06,
+      "loss": 0.1359,
+      "step": 1033
+    },
+    {
+      "epoch": 4.186234817813765,
+      "grad_norm": 0.2602820996695367,
+      "learning_rate": 3.1444210750077777e-06,
+      "loss": 0.1406,
+      "step": 1034
+    },
+    {
+      "epoch": 4.190283400809717,
+      "grad_norm": 0.25584206988988495,
+      "learning_rate": 3.1140475833991046e-06,
+      "loss": 0.1484,
+      "step": 1035
+    },
+    {
+      "epoch": 4.194331983805668,
+      "grad_norm": 0.26773209353197186,
+      "learning_rate": 3.0838091114159895e-06,
+      "loss": 0.127,
+      "step": 1036
+    },
+    {
+      "epoch": 4.198380566801619,
+      "grad_norm": 0.250940232296162,
+      "learning_rate": 3.0537059008444524e-06,
+      "loss": 0.1313,
+      "step": 1037
+    },
+    {
+      "epoch": 4.202429149797571,
+      "grad_norm": 0.2565374249948775,
+      "learning_rate": 3.0237381923889563e-06,
+      "loss": 0.1405,
+      "step": 1038
+    },
+    {
+      "epoch": 4.206477732793522,
+      "grad_norm": 0.25451636699635366,
+      "learning_rate": 2.9939062256704997e-06,
+      "loss": 0.1482,
+      "step": 1039
+    },
+    {
+      "epoch": 4.2105263157894735,
+      "grad_norm": 0.2619737393768085,
+      "learning_rate": 2.964210239224696e-06,
+      "loss": 0.1431,
+      "step": 1040
+    },
+    {
+      "epoch": 4.2145748987854255,
+      "grad_norm": 0.2572430761256009,
+      "learning_rate": 2.934650470499869e-06,
+      "loss": 0.1215,
+      "step": 1041
+    },
+    {
+      "epoch": 4.218623481781377,
+      "grad_norm": 0.25941463058303615,
+      "learning_rate": 2.905227155855128e-06,
+      "loss": 0.1302,
+      "step": 1042
+    },
+    {
+      "epoch": 4.222672064777328,
+      "grad_norm": 0.24078550435655782,
+      "learning_rate": 2.875940530558519e-06,
+      "loss": 0.1436,
+      "step": 1043
+    },
+    {
+      "epoch": 4.22672064777328,
+      "grad_norm": 0.2491831574202586,
+      "learning_rate": 2.8467908287851154e-06,
+      "loss": 0.1565,
+      "step": 1044
+    },
+    {
+      "epoch": 4.230769230769231,
+      "grad_norm": 0.2554528452528475,
+      "learning_rate": 2.8177782836151603e-06,
+      "loss": 0.1499,
+      "step": 1045
+    },
+    {
+      "epoch": 4.234817813765182,
+      "grad_norm": 0.24171603904393274,
+      "learning_rate": 2.7889031270321874e-06,
+      "loss": 0.1499,
+      "step": 1046
+    },
+    {
+      "epoch": 4.238866396761134,
+      "grad_norm": 0.2548585766659158,
+      "learning_rate": 2.760165589921184e-06,
+      "loss": 0.1339,
+      "step": 1047
+    },
+    {
+      "epoch": 4.242914979757085,
+      "grad_norm": 0.25901674713407596,
+      "learning_rate": 2.7315659020667305e-06,
+      "loss": 0.1374,
+      "step": 1048
+    },
+    {
+      "epoch": 4.246963562753036,
+      "grad_norm": 0.23946871291982744,
+      "learning_rate": 2.703104292151164e-06,
+      "loss": 0.1472,
+      "step": 1049
+    },
+    {
+      "epoch": 4.251012145748988,
+      "grad_norm": 0.27207050068548233,
+      "learning_rate": 2.6747809877527607e-06,
+      "loss": 0.1464,
+      "step": 1050
+    },
+    {
+      "epoch": 4.255060728744939,
+      "grad_norm": 0.2559919973897953,
+      "learning_rate": 2.6465962153439082e-06,
+      "loss": 0.1245,
+      "step": 1051
+    },
+    {
+      "epoch": 4.2591093117408905,
+      "grad_norm": 0.26618749846380624,
+      "learning_rate": 2.618550200289303e-06,
+      "loss": 0.143,
+      "step": 1052
+    },
+    {
+      "epoch": 4.2631578947368425,
+      "grad_norm": 0.26043292416677954,
+      "learning_rate": 2.59064316684412e-06,
+      "loss": 0.1474,
+      "step": 1053
+    },
+    {
+      "epoch": 4.267206477732794,
+      "grad_norm": 0.26403906827576445,
+      "learning_rate": 2.562875338152264e-06,
+      "loss": 0.1333,
+      "step": 1054
+    },
+    {
+      "epoch": 4.271255060728745,
+      "grad_norm": 0.2855096329847352,
+      "learning_rate": 2.5352469362445552e-06,
+      "loss": 0.1301,
+      "step": 1055
+    },
+    {
+      "epoch": 4.275303643724697,
+      "grad_norm": 0.2627061566600244,
+      "learning_rate": 2.5077581820369523e-06,
+      "loss": 0.1389,
+      "step": 1056
+    },
+    {
+      "epoch": 4.279352226720648,
+      "grad_norm": 0.2560402589139545,
+      "learning_rate": 2.480409295328803e-06,
+      "loss": 0.147,
+      "step": 1057
+    },
+    {
+      "epoch": 4.283400809716599,
+      "grad_norm": 0.2725267587234124,
+      "learning_rate": 2.45320049480108e-06,
+      "loss": 0.1332,
+      "step": 1058
+    },
+    {
+      "epoch": 4.287449392712551,
+      "grad_norm": 0.27129796433663994,
+      "learning_rate": 2.4261319980146293e-06,
+      "loss": 0.1559,
+      "step": 1059
+    },
+    {
+      "epoch": 4.291497975708502,
+      "grad_norm": 0.25803469435480064,
+      "learning_rate": 2.399204021408421e-06,
+      "loss": 0.1518,
+      "step": 1060
+    },
+    {
+      "epoch": 4.295546558704453,
+      "grad_norm": 0.26195180759635256,
+      "learning_rate": 2.3724167802978395e-06,
+      "loss": 0.1446,
+      "step": 1061
+    },
+    {
+      "epoch": 4.299595141700405,
+      "grad_norm": 0.2516421356711692,
+      "learning_rate": 2.3457704888729625e-06,
+      "loss": 0.1429,
+      "step": 1062
+    },
+    {
+      "epoch": 4.303643724696356,
+      "grad_norm": 0.25641210383626367,
+      "learning_rate": 2.3192653601968162e-06,
+      "loss": 0.1301,
+      "step": 1063
+    },
+    {
+      "epoch": 4.3076923076923075,
+      "grad_norm": 0.2583092394867026,
+      "learning_rate": 2.2929016062037057e-06,
+      "loss": 0.1452,
+      "step": 1064
+    },
+    {
+      "epoch": 4.3117408906882595,
+      "grad_norm": 0.2557876703020485,
+      "learning_rate": 2.2666794376975055e-06,
+      "loss": 0.1351,
+      "step": 1065
+    },
+    {
+      "epoch": 4.315789473684211,
+      "grad_norm": 0.26470877078129657,
+      "learning_rate": 2.24059906434998e-06,
+      "loss": 0.1283,
+      "step": 1066
+    },
+    {
+      "epoch": 4.319838056680162,
+      "grad_norm": 0.25320040915394215,
+      "learning_rate": 2.214660694699091e-06,
+      "loss": 0.1486,
+      "step": 1067
+    },
+    {
+      "epoch": 4.323886639676114,
+      "grad_norm": 0.24570113678541947,
+      "learning_rate": 2.188864536147355e-06,
+      "loss": 0.1393,
+      "step": 1068
+    },
+    {
+      "epoch": 4.327935222672065,
+      "grad_norm": 0.24631282069587168,
+      "learning_rate": 2.1632107949601757e-06,
+      "loss": 0.1403,
+      "step": 1069
+    },
+    {
+      "epoch": 4.331983805668016,
+      "grad_norm": 0.257059953180816,
+      "learning_rate": 2.137699676264171e-06,
+      "loss": 0.1446,
+      "step": 1070
+    },
+    {
+      "epoch": 4.336032388663968,
+      "grad_norm": 0.26200313017085697,
+      "learning_rate": 2.1123313840455738e-06,
+      "loss": 0.1493,
+      "step": 1071
+    },
+    {
+      "epoch": 4.340080971659919,
+      "grad_norm": 0.2583411370169722,
+      "learning_rate": 2.08710612114857e-06,
+      "loss": 0.1268,
+      "step": 1072
+    },
+    {
+      "epoch": 4.34412955465587,
+      "grad_norm": 0.26383688959424934,
+      "learning_rate": 2.0620240892736953e-06,
+      "loss": 0.1337,
+      "step": 1073
+    },
+    {
+      "epoch": 4.348178137651822,
+      "grad_norm": 0.27177330866583654,
+      "learning_rate": 2.0370854889762e-06,
+      "loss": 0.1434,
+      "step": 1074
+    },
+    {
+      "epoch": 4.352226720647773,
+      "grad_norm": 0.2639420186559343,
+      "learning_rate": 2.0122905196644726e-06,
+      "loss": 0.1294,
+      "step": 1075
+    },
+    {
+      "epoch": 4.3562753036437245,
+      "grad_norm": 0.2547989810017792,
+      "learning_rate": 1.987639379598427e-06,
+      "loss": 0.1451,
+      "step": 1076
+    },
+    {
+      "epoch": 4.3603238866396765,
+      "grad_norm": 0.25960895236787657,
+      "learning_rate": 1.963132265887919e-06,
+      "loss": 0.1307,
+      "step": 1077
+    },
+    {
+      "epoch": 4.364372469635628,
+      "grad_norm": 0.25810979375833343,
+      "learning_rate": 1.9387693744911784e-06,
+      "loss": 0.142,
+      "step": 1078
+    },
+    {
+      "epoch": 4.368421052631579,
+      "grad_norm": 0.2603846403873938,
+      "learning_rate": 1.9145509002132368e-06,
+      "loss": 0.1416,
+      "step": 1079
+    },
+    {
+      "epoch": 4.372469635627531,
+      "grad_norm": 0.25091321914793047,
+      "learning_rate": 1.8904770367043723e-06,
+      "loss": 0.1378,
+      "step": 1080
+    },
+    {
+      "epoch": 4.376518218623482,
+      "grad_norm": 0.26355719758439655,
+      "learning_rate": 1.8665479764585504e-06,
+      "loss": 0.1391,
+      "step": 1081
+    },
+    {
+      "epoch": 4.380566801619433,
+      "grad_norm": 0.2608437488287452,
+      "learning_rate": 1.8427639108118977e-06,
+      "loss": 0.1332,
+      "step": 1082
+    },
+    {
+      "epoch": 4.384615384615385,
+      "grad_norm": 0.23680332525823117,
+      "learning_rate": 1.8191250299411755e-06,
+      "loss": 0.1544,
+      "step": 1083
+    },
+    {
+      "epoch": 4.388663967611336,
+      "grad_norm": 0.2741237586615027,
+      "learning_rate": 1.7956315228622446e-06,
+      "loss": 0.1352,
+      "step": 1084
+    },
+    {
+      "epoch": 4.392712550607287,
+      "grad_norm": 0.25039938506045517,
+      "learning_rate": 1.7722835774285553e-06,
+      "loss": 0.1494,
+      "step": 1085
+    },
+    {
+      "epoch": 4.396761133603239,
+      "grad_norm": 0.26368949824574667,
+      "learning_rate": 1.7490813803296624e-06,
+      "loss": 0.1327,
+      "step": 1086
+    },
+    {
+      "epoch": 4.40080971659919,
+      "grad_norm": 0.26363811832430956,
+      "learning_rate": 1.726025117089718e-06,
+      "loss": 0.136,
+      "step": 1087
+    },
+    {
+      "epoch": 4.4048582995951415,
+      "grad_norm": 0.25586054642847267,
+      "learning_rate": 1.70311497206598e-06,
+      "loss": 0.1477,
+      "step": 1088
+    },
+    {
+      "epoch": 4.4089068825910935,
+      "grad_norm": 0.2505581404548255,
+      "learning_rate": 1.6803511284473638e-06,
+      "loss": 0.1393,
+      "step": 1089
+    },
+    {
+      "epoch": 4.412955465587045,
+      "grad_norm": 0.2714087019476765,
+      "learning_rate": 1.6577337682529538e-06,
+      "loss": 0.1411,
+      "step": 1090
+    },
+    {
+      "epoch": 4.417004048582996,
+      "grad_norm": 0.2541986483731322,
+      "learning_rate": 1.6352630723305617e-06,
+      "loss": 0.121,
+      "step": 1091
+    },
+    {
+      "epoch": 4.421052631578947,
+      "grad_norm": 0.25454578985109577,
+      "learning_rate": 1.612939220355272e-06,
+      "loss": 0.1477,
+      "step": 1092
+    },
+    {
+      "epoch": 4.425101214574899,
+      "grad_norm": 0.2601528579843554,
+      "learning_rate": 1.5907623908280112e-06,
+      "loss": 0.1341,
+      "step": 1093
+    },
+    {
+      "epoch": 4.42914979757085,
+      "grad_norm": 0.24845121761204048,
+      "learning_rate": 1.568732761074121e-06,
+      "loss": 0.1351,
+      "step": 1094
+    },
+    {
+      "epoch": 4.433198380566802,
+      "grad_norm": 0.26406023938888873,
+      "learning_rate": 1.5468505072419237e-06,
+      "loss": 0.1471,
+      "step": 1095
+    },
+    {
+      "epoch": 4.437246963562753,
+      "grad_norm": 0.25567899012829426,
+      "learning_rate": 1.5251158043013448e-06,
+      "loss": 0.1403,
+      "step": 1096
+    },
+    {
+      "epoch": 4.441295546558704,
+      "grad_norm": 0.25096085097594617,
+      "learning_rate": 1.5035288260424885e-06,
+      "loss": 0.1446,
+      "step": 1097
+    },
+    {
+      "epoch": 4.445344129554655,
+      "grad_norm": 0.2489882705644177,
+      "learning_rate": 1.4820897450742623e-06,
+      "loss": 0.1398,
+      "step": 1098
+    },
+    {
+      "epoch": 4.449392712550607,
+      "grad_norm": 0.25073586702078754,
+      "learning_rate": 1.4607987328229767e-06,
+      "loss": 0.1352,
+      "step": 1099
+    },
+    {
+      "epoch": 4.4534412955465585,
+      "grad_norm": 0.2582454078378835,
+      "learning_rate": 1.4396559595310056e-06,
+      "loss": 0.1422,
+      "step": 1100
+    },
+    {
+      "epoch": 4.4574898785425106,
+      "grad_norm": 0.25361671619598986,
+      "learning_rate": 1.4186615942553973e-06,
+      "loss": 0.1303,
+      "step": 1101
+    },
+    {
+      "epoch": 4.461538461538462,
+      "grad_norm": 0.24752152723569198,
+      "learning_rate": 1.3978158048665424e-06,
+      "loss": 0.1388,
+      "step": 1102
+    },
+    {
+      "epoch": 4.465587044534413,
+      "grad_norm": 0.26057055407386076,
+      "learning_rate": 1.3771187580468115e-06,
+      "loss": 0.1583,
+      "step": 1103
+    },
+    {
+      "epoch": 4.469635627530364,
+      "grad_norm": 0.2595000419577279,
+      "learning_rate": 1.3565706192892392e-06,
+      "loss": 0.1545,
+      "step": 1104
+    },
+    {
+      "epoch": 4.473684210526316,
+      "grad_norm": 0.24889526198667605,
+      "learning_rate": 1.3361715528961993e-06,
+      "loss": 0.1512,
+      "step": 1105
+    },
+    {
+      "epoch": 4.477732793522267,
+      "grad_norm": 0.25979688842394943,
+      "learning_rate": 1.3159217219780817e-06,
+      "loss": 0.1442,
+      "step": 1106
+    },
+    {
+      "epoch": 4.481781376518219,
+      "grad_norm": 0.2553658611473135,
+      "learning_rate": 1.2958212884519949e-06,
+      "loss": 0.1411,
+      "step": 1107
+    },
+    {
+      "epoch": 4.48582995951417,
+      "grad_norm": 0.24950075838778407,
+      "learning_rate": 1.2758704130404675e-06,
+      "loss": 0.1354,
+      "step": 1108
+    },
+    {
+      "epoch": 4.489878542510121,
+      "grad_norm": 0.26860093494169507,
+      "learning_rate": 1.2560692552701738e-06,
+      "loss": 0.1374,
+      "step": 1109
+    },
+    {
+      "epoch": 4.493927125506072,
+      "grad_norm": 0.2658862916454609,
+      "learning_rate": 1.2364179734706382e-06,
+      "loss": 0.1377,
+      "step": 1110
+    },
+    {
+      "epoch": 4.497975708502024,
+      "grad_norm": 0.26484532961937957,
+      "learning_rate": 1.2169167247729852e-06,
+      "loss": 0.1324,
+      "step": 1111
+    },
+    {
+      "epoch": 4.502024291497976,
+      "grad_norm": 0.2448009028331799,
+      "learning_rate": 1.1975656651086864e-06,
+      "loss": 0.1478,
+      "step": 1112
+    },
+    {
+      "epoch": 4.506072874493928,
+      "grad_norm": 0.2456948685133228,
+      "learning_rate": 1.1783649492082926e-06,
+      "loss": 0.1474,
+      "step": 1113
+    },
+    {
+      "epoch": 4.510121457489879,
+      "grad_norm": 0.2671337612375431,
+      "learning_rate": 1.1593147306002183e-06,
+      "loss": 0.1245,
+      "step": 1114
+    },
+    {
+      "epoch": 4.51417004048583,
+      "grad_norm": 0.24661955257012708,
+      "learning_rate": 1.140415161609505e-06,
+      "loss": 0.1236,
+      "step": 1115
+    },
+    {
+      "epoch": 4.518218623481781,
+      "grad_norm": 0.25541739036475225,
+      "learning_rate": 1.121666393356604e-06,
+      "loss": 0.1296,
+      "step": 1116
+    },
+    {
+      "epoch": 4.522267206477733,
+      "grad_norm": 0.2701969005856409,
+      "learning_rate": 1.1030685757561632e-06,
+      "loss": 0.1343,
+      "step": 1117
+    },
+    {
+      "epoch": 4.526315789473684,
+      "grad_norm": 0.26509448405264957,
+      "learning_rate": 1.0846218575158373e-06,
+      "loss": 0.1416,
+      "step": 1118
+    },
+    {
+      "epoch": 4.530364372469636,
+      "grad_norm": 0.25008773502692994,
+      "learning_rate": 1.066326386135097e-06,
+      "loss": 0.1315,
+      "step": 1119
+    },
+    {
+      "epoch": 4.534412955465587,
+      "grad_norm": 0.2699351198506807,
+      "learning_rate": 1.0481823079040398e-06,
+      "loss": 0.134,
+      "step": 1120
+    },
+    {
+      "epoch": 4.538461538461538,
+      "grad_norm": 0.2626802933456165,
+      "learning_rate": 1.0301897679022344e-06,
+      "loss": 0.1351,
+      "step": 1121
+    },
+    {
+      "epoch": 4.5425101214574894,
+      "grad_norm": 0.2491722958124618,
+      "learning_rate": 1.0123489099975491e-06,
+      "loss": 0.135,
+      "step": 1122
+    },
+    {
+      "epoch": 4.5465587044534415,
+      "grad_norm": 0.2652265948501109,
+      "learning_rate": 9.9465987684501e-07,
+      "loss": 0.1332,
+      "step": 1123
+    },
+    {
+      "epoch": 4.550607287449393,
+      "grad_norm": 0.24923169718406835,
+      "learning_rate": 9.771228098856534e-07,
+      "loss": 0.141,
+      "step": 1124
+    },
+    {
+      "epoch": 4.554655870445345,
+      "grad_norm": 0.2484165020397499,
+      "learning_rate": 9.597378493454013e-07,
+      "loss": 0.1487,
+      "step": 1125
+    },
+    {
+      "epoch": 4.558704453441296,
+      "grad_norm": 0.2494476068213255,
+      "learning_rate": 9.425051342339353e-07,
+      "loss": 0.1368,
+      "step": 1126
+    },
+    {
+      "epoch": 4.562753036437247,
+      "grad_norm": 0.2561981918380986,
+      "learning_rate": 9.254248023435864e-07,
+      "loss": 0.1357,
+      "step": 1127
+    },
+    {
+      "epoch": 4.566801619433198,
+      "grad_norm": 0.25364136948973043,
+      "learning_rate": 9.084969902482288e-07,
+      "loss": 0.1321,
+      "step": 1128
+    },
+    {
+      "epoch": 4.57085020242915,
+      "grad_norm": 0.24369022495217754,
+      "learning_rate": 8.917218333021993e-07,
+      "loss": 0.1437,
+      "step": 1129
+    },
+    {
+      "epoch": 4.574898785425101,
+      "grad_norm": 0.2536231932013965,
+      "learning_rate": 8.750994656392109e-07,
+      "loss": 0.1339,
+      "step": 1130
+    },
+    {
+      "epoch": 4.578947368421053,
+      "grad_norm": 0.2470658348522141,
+      "learning_rate": 8.58630020171265e-07,
+      "loss": 0.1488,
+      "step": 1131
+    },
+    {
+      "epoch": 4.582995951417004,
+      "grad_norm": 0.2595941276479585,
+      "learning_rate": 8.423136285876099e-07,
+      "loss": 0.1437,
+      "step": 1132
+    },
+    {
+      "epoch": 4.587044534412955,
+      "grad_norm": 0.26615799185155015,
+      "learning_rate": 8.26150421353682e-07,
+      "loss": 0.1256,
+      "step": 1133
+    },
+    {
+      "epoch": 4.5910931174089065,
+      "grad_norm": 0.2625765734646271,
+      "learning_rate": 8.101405277100549e-07,
+      "loss": 0.1575,
+      "step": 1134
+    },
+    {
+      "epoch": 4.5951417004048585,
+      "grad_norm": 0.28247869991273467,
+      "learning_rate": 7.942840756714077e-07,
+      "loss": 0.1422,
+      "step": 1135
+    },
+    {
+      "epoch": 4.59919028340081,
+      "grad_norm": 0.24361843056912033,
+      "learning_rate": 7.785811920255093e-07,
+      "loss": 0.1363,
+      "step": 1136
+    },
+    {
+      "epoch": 4.603238866396762,
+      "grad_norm": 0.26082483333494894,
+      "learning_rate": 7.630320023322024e-07,
+      "loss": 0.142,
+      "step": 1137
+    },
+    {
+      "epoch": 4.607287449392713,
+      "grad_norm": 0.25862950582989147,
+      "learning_rate": 7.476366309223903e-07,
+      "loss": 0.1572,
+      "step": 1138
+    },
+    {
+      "epoch": 4.611336032388664,
+      "grad_norm": 0.2562145166980071,
+      "learning_rate": 7.323952008970537e-07,
+      "loss": 0.1485,
+      "step": 1139
+    },
+    {
+      "epoch": 4.615384615384615,
+      "grad_norm": 0.25010637816312914,
+      "learning_rate": 7.173078341262618e-07,
+      "loss": 0.1448,
+      "step": 1140
+    },
+    {
+      "epoch": 4.619433198380567,
+      "grad_norm": 0.2494078057135322,
+      "learning_rate": 7.023746512481989e-07,
+      "loss": 0.1536,
+      "step": 1141
+    },
+    {
+      "epoch": 4.623481781376518,
+      "grad_norm": 0.2659983974609215,
+      "learning_rate": 6.875957716681902e-07,
+      "loss": 0.1278,
+      "step": 1142
+    },
+    {
+      "epoch": 4.62753036437247,
+      "grad_norm": 0.24408733535350732,
+      "learning_rate": 6.729713135577687e-07,
+      "loss": 0.1548,
+      "step": 1143
+    },
+    {
+      "epoch": 4.631578947368421,
+      "grad_norm": 0.25691296084193965,
+      "learning_rate": 6.585013938537143e-07,
+      "loss": 0.1528,
+      "step": 1144
+    },
+    {
+      "epoch": 4.635627530364372,
+      "grad_norm": 0.24486223584642433,
+      "learning_rate": 6.441861282571115e-07,
+      "loss": 0.1401,
+      "step": 1145
+    },
+    {
+      "epoch": 4.6396761133603235,
+      "grad_norm": 0.2454771800832961,
+      "learning_rate": 6.300256312324471e-07,
+      "loss": 0.1547,
+      "step": 1146
+    },
+    {
+      "epoch": 4.6437246963562755,
+      "grad_norm": 0.2592157727594957,
+      "learning_rate": 6.160200160066776e-07,
+      "loss": 0.1422,
+      "step": 1147
+    },
+    {
+      "epoch": 4.647773279352227,
+      "grad_norm": 0.2518188291640924,
+      "learning_rate": 6.021693945683327e-07,
+      "loss": 0.1471,
+      "step": 1148
+    },
+    {
+      "epoch": 4.651821862348179,
+      "grad_norm": 0.23901791894522237,
+      "learning_rate": 5.884738776666044e-07,
+      "loss": 0.1454,
+      "step": 1149
+    },
+    {
+      "epoch": 4.65587044534413,
+      "grad_norm": 0.24642010025744376,
+      "learning_rate": 5.749335748104878e-07,
+      "loss": 0.129,
+      "step": 1150
+    },
+    {
+      "epoch": 4.659919028340081,
+      "grad_norm": 0.24608605984816534,
+      "learning_rate": 5.615485942678778e-07,
+      "loss": 0.1465,
+      "step": 1151
+    },
+    {
+      "epoch": 4.663967611336032,
+      "grad_norm": 0.2543936361054934,
+      "learning_rate": 5.483190430647289e-07,
+      "loss": 0.1518,
+      "step": 1152
+    },
+    {
+      "epoch": 4.668016194331984,
+      "grad_norm": 0.2607329083479994,
+      "learning_rate": 5.352450269841747e-07,
+      "loss": 0.1244,
+      "step": 1153
+    },
+    {
+      "epoch": 4.672064777327935,
+      "grad_norm": 0.25465054039071405,
+      "learning_rate": 5.22326650565701e-07,
+      "loss": 0.1394,
+      "step": 1154
+    },
+    {
+      "epoch": 4.676113360323887,
+      "grad_norm": 0.26903817138917474,
+      "learning_rate": 5.095640171043048e-07,
+      "loss": 0.1308,
+      "step": 1155
+    },
+    {
+      "epoch": 4.680161943319838,
+      "grad_norm": 0.24942576194271515,
+      "learning_rate": 4.969572286496594e-07,
+      "loss": 0.1362,
+      "step": 1156
+    },
+    {
+      "epoch": 4.684210526315789,
+      "grad_norm": 0.2553840731365522,
+      "learning_rate": 4.845063860053079e-07,
+      "loss": 0.1254,
+      "step": 1157
+    },
+    {
+      "epoch": 4.6882591093117405,
+      "grad_norm": 0.2531470094686325,
+      "learning_rate": 4.7221158872785997e-07,
+      "loss": 0.1302,
+      "step": 1158
+    },
+    {
+      "epoch": 4.6923076923076925,
+      "grad_norm": 0.25173057276385746,
+      "learning_rate": 4.600729351261857e-07,
+      "loss": 0.1302,
+      "step": 1159
+    },
+    {
+      "epoch": 4.696356275303644,
+      "grad_norm": 0.2614576032667149,
+      "learning_rate": 4.4809052226063355e-07,
+      "loss": 0.1229,
+      "step": 1160
+    },
+    {
+      "epoch": 4.700404858299595,
+      "grad_norm": 0.2612477304594784,
+      "learning_rate": 4.3626444594226046e-07,
+      "loss": 0.1368,
+      "step": 1161
+    },
+    {
+      "epoch": 4.704453441295547,
+      "grad_norm": 0.26064338721802754,
+      "learning_rate": 4.245948007320588e-07,
+      "loss": 0.1538,
+      "step": 1162
+    },
+    {
+      "epoch": 4.708502024291498,
+      "grad_norm": 0.2608040248357144,
+      "learning_rate": 4.130816799401971e-07,
+      "loss": 0.1499,
+      "step": 1163
+    },
+    {
+      "epoch": 4.712550607287449,
+      "grad_norm": 0.24300954132968058,
+      "learning_rate": 4.017251756252827e-07,
+      "loss": 0.1388,
+      "step": 1164
+    },
+    {
+      "epoch": 4.716599190283401,
+      "grad_norm": 0.2565529787071644,
+      "learning_rate": 3.9052537859362473e-07,
+      "loss": 0.1278,
+      "step": 1165
+    },
+    {
+      "epoch": 4.720647773279352,
+      "grad_norm": 0.24290923299320835,
+      "learning_rate": 3.7948237839849687e-07,
+      "loss": 0.1267,
+      "step": 1166
+    },
+    {
+      "epoch": 4.724696356275303,
+      "grad_norm": 0.24291379924283807,
+      "learning_rate": 3.6859626333943797e-07,
+      "loss": 0.1503,
+      "step": 1167
+    },
+    {
+      "epoch": 4.728744939271255,
+      "grad_norm": 0.2471536265874195,
+      "learning_rate": 3.578671204615347e-07,
+      "loss": 0.1425,
+      "step": 1168
+    },
+    {
+      "epoch": 4.732793522267206,
+      "grad_norm": 0.2512054909615427,
+      "learning_rate": 3.472950355547267e-07,
+      "loss": 0.156,
+      "step": 1169
+    },
+    {
+      "epoch": 4.7368421052631575,
+      "grad_norm": 0.24572411695716825,
+      "learning_rate": 3.3688009315312244e-07,
+      "loss": 0.1529,
+      "step": 1170
+    },
+    {
+      "epoch": 4.7408906882591095,
+      "grad_norm": 0.24869847816300855,
+      "learning_rate": 3.266223765343224e-07,
+      "loss": 0.133,
+      "step": 1171
+    },
+    {
+      "epoch": 4.744939271255061,
+      "grad_norm": 0.25450850287256993,
+      "learning_rate": 3.16521967718757e-07,
+      "loss": 0.1357,
+      "step": 1172
+    },
+    {
+      "epoch": 4.748987854251012,
+      "grad_norm": 0.24902486326034107,
+      "learning_rate": 3.0657894746902726e-07,
+      "loss": 0.1315,
+      "step": 1173
+    },
+    {
+      "epoch": 4.753036437246964,
+      "grad_norm": 0.2548557003975382,
+      "learning_rate": 2.9679339528925654e-07,
+      "loss": 0.1392,
+      "step": 1174
+    },
+    {
+      "epoch": 4.757085020242915,
+      "grad_norm": 0.23921812697825937,
+      "learning_rate": 2.871653894244597e-07,
+      "loss": 0.1434,
+      "step": 1175
+    },
+    {
+      "epoch": 4.761133603238866,
+      "grad_norm": 0.2550271490577432,
+      "learning_rate": 2.7769500685991266e-07,
+      "loss": 0.1353,
+      "step": 1176
+    },
+    {
+      "epoch": 4.765182186234818,
+      "grad_norm": 0.2546346942663952,
+      "learning_rate": 2.683823233205485e-07,
+      "loss": 0.1424,
+      "step": 1177
+    },
+    {
+      "epoch": 4.769230769230769,
+      "grad_norm": 0.2514774398133411,
+      "learning_rate": 2.592274132703287e-07,
+      "loss": 0.1368,
+      "step": 1178
+    },
+    {
+      "epoch": 4.77327935222672,
+      "grad_norm": 0.24782682550568202,
+      "learning_rate": 2.502303499116754e-07,
+      "loss": 0.1339,
+      "step": 1179
+    },
+    {
+      "epoch": 4.777327935222672,
+      "grad_norm": 0.2606293652736473,
+      "learning_rate": 2.413912051848666e-07,
+      "loss": 0.1461,
+      "step": 1180
+    },
+    {
+      "epoch": 4.781376518218623,
+      "grad_norm": 0.250251493855873,
+      "learning_rate": 2.3271004976746613e-07,
+      "loss": 0.128,
+      "step": 1181
+    },
+    {
+      "epoch": 4.7854251012145745,
+      "grad_norm": 0.24391271705677392,
+      "learning_rate": 2.2418695307376393e-07,
+      "loss": 0.1468,
+      "step": 1182
+    },
+    {
+      "epoch": 4.7894736842105265,
+      "grad_norm": 0.2423612189545789,
+      "learning_rate": 2.1582198325420968e-07,
+      "loss": 0.1389,
+      "step": 1183
+    },
+    {
+      "epoch": 4.793522267206478,
+      "grad_norm": 0.2479709831049085,
+      "learning_rate": 2.0761520719488447e-07,
+      "loss": 0.1166,
+      "step": 1184
+    },
+    {
+      "epoch": 4.797570850202429,
+      "grad_norm": 0.23422943191117515,
+      "learning_rate": 1.9956669051694798e-07,
+      "loss": 0.1285,
+      "step": 1185
+    },
+    {
+      "epoch": 4.801619433198381,
+      "grad_norm": 0.23767747492323837,
+      "learning_rate": 1.916764975761254e-07,
+      "loss": 0.1585,
+      "step": 1186
+    },
+    {
+      "epoch": 4.805668016194332,
+      "grad_norm": 0.24811006499300867,
+      "learning_rate": 1.839446914621923e-07,
+      "loss": 0.1439,
+      "step": 1187
+    },
+    {
+      "epoch": 4.809716599190283,
+      "grad_norm": 0.2559273715301149,
+      "learning_rate": 1.7637133399845962e-07,
+      "loss": 0.1515,
+      "step": 1188
+    },
+    {
+      "epoch": 4.813765182186235,
+      "grad_norm": 0.24447574692298774,
+      "learning_rate": 1.6895648574129397e-07,
+      "loss": 0.1461,
+      "step": 1189
+    },
+    {
+      "epoch": 4.817813765182186,
+      "grad_norm": 0.25675438703871967,
+      "learning_rate": 1.6170020597962465e-07,
+      "loss": 0.1471,
+      "step": 1190
+    },
+    {
+      "epoch": 4.821862348178137,
+      "grad_norm": 0.250712714624097,
+      "learning_rate": 1.5460255273447078e-07,
+      "loss": 0.1359,
+      "step": 1191
+    },
+    {
+      "epoch": 4.825910931174089,
+      "grad_norm": 0.24452835115968025,
+      "learning_rate": 1.4766358275847715e-07,
+      "loss": 0.1359,
+      "step": 1192
+    },
+    {
+      "epoch": 4.82995951417004,
+      "grad_norm": 0.2538095245695539,
+      "learning_rate": 1.4088335153546573e-07,
+      "loss": 0.1414,
+      "step": 1193
+    },
+    {
+      "epoch": 4.834008097165992,
+      "grad_norm": 0.2551022528891488,
+      "learning_rate": 1.3426191327998496e-07,
+      "loss": 0.1347,
+      "step": 1194
+    },
+    {
+      "epoch": 4.838056680161944,
+      "grad_norm": 0.24351798094218502,
+      "learning_rate": 1.277993209368744e-07,
+      "loss": 0.1553,
+      "step": 1195
+    },
+    {
+      "epoch": 4.842105263157895,
+      "grad_norm": 0.2576262283674437,
+      "learning_rate": 1.214956261808542e-07,
+      "loss": 0.1584,
+      "step": 1196
+    },
+    {
+      "epoch": 4.846153846153846,
+      "grad_norm": 0.2677393778581336,
+      "learning_rate": 1.1535087941609846e-07,
+      "loss": 0.1471,
+      "step": 1197
+    },
+    {
+      "epoch": 4.850202429149798,
+      "grad_norm": 0.2440996098920378,
+      "learning_rate": 1.0936512977583802e-07,
+      "loss": 0.135,
+      "step": 1198
+    },
+    {
+      "epoch": 4.854251012145749,
+      "grad_norm": 0.25014770841592593,
+      "learning_rate": 1.0353842512196732e-07,
+      "loss": 0.1457,
+      "step": 1199
+    },
+    {
+      "epoch": 4.8582995951417,
+      "grad_norm": 0.2504941847188282,
+      "learning_rate": 9.787081204466253e-08,
+      "loss": 0.1433,
+      "step": 1200
+    },
+    {
+      "epoch": 4.862348178137652,
+      "grad_norm": 0.2582717356782261,
+      "learning_rate": 9.236233586200627e-08,
+      "loss": 0.1354,
+      "step": 1201
+    },
+    {
+      "epoch": 4.866396761133603,
+      "grad_norm": 0.24415902552821606,
+      "learning_rate": 8.70130406196279e-08,
+      "loss": 0.1431,
+      "step": 1202
+    },
+    {
+      "epoch": 4.870445344129554,
+      "grad_norm": 0.25228477657523196,
+      "learning_rate": 8.182296909035047e-08,
+      "loss": 0.1324,
+      "step": 1203
+    },
+    {
+      "epoch": 4.874493927125506,
+      "grad_norm": 0.26379503736387466,
+      "learning_rate": 7.679216277384439e-08,
+      "loss": 0.1271,
+      "step": 1204
+    },
+    {
+      "epoch": 4.8785425101214575,
+      "grad_norm": 0.252618327828462,
+      "learning_rate": 7.19206618963053e-08,
+      "loss": 0.1455,
+      "step": 1205
+    },
+    {
+      "epoch": 4.882591093117409,
+      "grad_norm": 0.2546835444154793,
+      "learning_rate": 6.720850541012347e-08,
+      "loss": 0.1368,
+      "step": 1206
+    },
+    {
+      "epoch": 4.886639676113361,
+      "grad_norm": 0.23975786937148613,
+      "learning_rate": 6.265573099357714e-08,
+      "loss": 0.1321,
+      "step": 1207
+    },
+    {
+      "epoch": 4.890688259109312,
+      "grad_norm": 0.24123826516960076,
+      "learning_rate": 5.826237505053067e-08,
+      "loss": 0.1347,
+      "step": 1208
+    },
+    {
+      "epoch": 4.894736842105263,
+      "grad_norm": 0.2541847731327919,
+      "learning_rate": 5.4028472710137e-08,
+      "loss": 0.1352,
+      "step": 1209
+    },
+    {
+      "epoch": 4.898785425101215,
+      "grad_norm": 0.24949521339296135,
+      "learning_rate": 4.9954057826571146e-08,
+      "loss": 0.1377,
+      "step": 1210
+    },
+    {
+      "epoch": 4.902834008097166,
+      "grad_norm": 0.24598524779301784,
+      "learning_rate": 4.6039162978743776e-08,
+      "loss": 0.1333,
+      "step": 1211
+    },
+    {
+      "epoch": 4.906882591093117,
+      "grad_norm": 0.2602126485243393,
+      "learning_rate": 4.228381947004812e-08,
+      "loss": 0.1422,
+      "step": 1212
+    },
+    {
+      "epoch": 4.910931174089069,
+      "grad_norm": 0.24700340547266855,
+      "learning_rate": 3.868805732811343e-08,
+      "loss": 0.1336,
+      "step": 1213
+    },
+    {
+      "epoch": 4.91497975708502,
+      "grad_norm": 0.25561139966458585,
+      "learning_rate": 3.525190530455636e-08,
+      "loss": 0.1362,
+      "step": 1214
+    },
+    {
+      "epoch": 4.919028340080971,
+      "grad_norm": 0.25432268972317223,
+      "learning_rate": 3.197539087475887e-08,
+      "loss": 0.1566,
+      "step": 1215
+    },
+    {
+      "epoch": 4.923076923076923,
+      "grad_norm": 0.2520153147404582,
+      "learning_rate": 2.8858540237641786e-08,
+      "loss": 0.1242,
+      "step": 1216
+    },
+    {
+      "epoch": 4.9271255060728745,
+      "grad_norm": 0.2509634172278358,
+      "learning_rate": 2.5901378315458247e-08,
+      "loss": 0.1373,
+      "step": 1217
+    },
+    {
+      "epoch": 4.931174089068826,
+      "grad_norm": 0.24804095306048957,
+      "learning_rate": 2.3103928753598347e-08,
+      "loss": 0.1455,
+      "step": 1218
+    },
+    {
+      "epoch": 4.935222672064778,
+      "grad_norm": 0.2458796958047569,
+      "learning_rate": 2.0466213920393718e-08,
+      "loss": 0.1313,
+      "step": 1219
+    },
+    {
+      "epoch": 4.939271255060729,
+      "grad_norm": 0.25139949013728796,
+      "learning_rate": 1.7988254906942115e-08,
+      "loss": 0.1319,
+      "step": 1220
+    },
+    {
+      "epoch": 4.94331983805668,
+      "grad_norm": 0.257351988547275,
+      "learning_rate": 1.5670071526936447e-08,
+      "loss": 0.1432,
+      "step": 1221
+    },
+    {
+      "epoch": 4.947368421052632,
+      "grad_norm": 0.25040931288135504,
+      "learning_rate": 1.3511682316509344e-08,
+      "loss": 0.1372,
+      "step": 1222
+    },
+    {
+      "epoch": 4.951417004048583,
+      "grad_norm": 0.25168759217945796,
+      "learning_rate": 1.1513104534086605e-08,
+      "loss": 0.1426,
+      "step": 1223
+    },
+    {
+      "epoch": 4.955465587044534,
+      "grad_norm": 0.2469950347371207,
+      "learning_rate": 9.67435416023843e-09,
+      "loss": 0.1545,
+      "step": 1224
+    },
+    {
+      "epoch": 4.959514170040486,
+      "grad_norm": 0.2572986078984373,
+      "learning_rate": 7.995445897566178e-09,
+      "loss": 0.143,
+      "step": 1225
+    },
+    {
+      "epoch": 4.963562753036437,
+      "grad_norm": 0.25740773925566535,
+      "learning_rate": 6.476393170578021e-09,
+      "loss": 0.1443,
+      "step": 1226
+    },
+    {
+      "epoch": 4.967611336032388,
+      "grad_norm": 0.23966814255770746,
+      "learning_rate": 5.11720812557348e-09,
+      "loss": 0.1349,
+      "step": 1227
+    },
+    {
+      "epoch": 4.97165991902834,
+      "grad_norm": 0.2547660687308002,
+      "learning_rate": 3.917901630563492e-09,
+      "loss": 0.1381,
+      "step": 1228
+    },
+    {
+      "epoch": 4.9757085020242915,
+      "grad_norm": 0.2562394569920061,
+      "learning_rate": 2.878483275170485e-09,
+      "loss": 0.1289,
+      "step": 1229
+    },
+    {
+      "epoch": 4.979757085020243,
+      "grad_norm": 0.2564884404248968,
+      "learning_rate": 1.998961370555108e-09,
+      "loss": 0.1353,
+      "step": 1230
+    },
+    {
+      "epoch": 4.983805668016195,
+      "grad_norm": 0.26477425177895325,
+      "learning_rate": 1.2793429493518361e-09,
+      "loss": 0.1292,
+      "step": 1231
+    },
+    {
+      "epoch": 4.987854251012146,
+      "grad_norm": 0.2595517712890884,
+      "learning_rate": 7.196337656112384e-10,
+      "loss": 0.1199,
+      "step": 1232
+    },
+    {
+      "epoch": 4.991902834008097,
+      "grad_norm": 0.2610874356207304,
+      "learning_rate": 3.198382947511292e-10,
+      "loss": 0.1366,
+      "step": 1233
+    },
+    {
+      "epoch": 4.995951417004049,
+      "grad_norm": 0.2530918266054629,
+      "learning_rate": 7.995973352770137e-11,
+      "loss": 0.1439,
+      "step": 1234
+    },
+    {
+      "epoch": 5.0,
+      "grad_norm": 0.24871891746519426,
+      "learning_rate": 0.0,
+      "loss": 0.1299,
+      "step": 1235
+    },
+    {
+      "epoch": 5.0,
+      "step": 1235,
+      "total_flos": 1515501703856128.0,
+      "train_loss": 0.2937546003202678,
+      "train_runtime": 24376.4965,
+      "train_samples_per_second": 6.482,
+      "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": 1515501703856128.0,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}