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": 6.830692526846545,
+      "learning_rate": 3.2258064516129035e-07,
+      "loss": 1.0613,
+      "step": 1
+    },
+    {
+      "epoch": 0.008097165991902834,
+      "grad_norm": 6.536450268346033,
+      "learning_rate": 6.451612903225807e-07,
+      "loss": 1.0388,
+      "step": 2
+    },
+    {
+      "epoch": 0.012145748987854251,
+      "grad_norm": 6.549329536537879,
+      "learning_rate": 9.67741935483871e-07,
+      "loss": 1.0377,
+      "step": 3
+    },
+    {
+      "epoch": 0.016194331983805668,
+      "grad_norm": 6.542410327499724,
+      "learning_rate": 1.2903225806451614e-06,
+      "loss": 1.0352,
+      "step": 4
+    },
+    {
+      "epoch": 0.020242914979757085,
+      "grad_norm": 6.091955726348501,
+      "learning_rate": 1.6129032258064516e-06,
+      "loss": 1.025,
+      "step": 5
+    },
+    {
+      "epoch": 0.024291497975708502,
+      "grad_norm": 5.913478709538355,
+      "learning_rate": 1.935483870967742e-06,
+      "loss": 1.0016,
+      "step": 6
+    },
+    {
+      "epoch": 0.02834008097165992,
+      "grad_norm": 4.882635592110913,
+      "learning_rate": 2.2580645161290324e-06,
+      "loss": 0.9837,
+      "step": 7
+    },
+    {
+      "epoch": 0.032388663967611336,
+      "grad_norm": 4.510711567179104,
+      "learning_rate": 2.580645161290323e-06,
+      "loss": 0.9723,
+      "step": 8
+    },
+    {
+      "epoch": 0.03643724696356275,
+      "grad_norm": 2.9689525564024946,
+      "learning_rate": 2.903225806451613e-06,
+      "loss": 0.9438,
+      "step": 9
+    },
+    {
+      "epoch": 0.04048582995951417,
+      "grad_norm": 2.8308854540800152,
+      "learning_rate": 3.225806451612903e-06,
+      "loss": 0.9522,
+      "step": 10
+    },
+    {
+      "epoch": 0.044534412955465584,
+      "grad_norm": 2.6833703860590434,
+      "learning_rate": 3.548387096774194e-06,
+      "loss": 0.9358,
+      "step": 11
+    },
+    {
+      "epoch": 0.048582995951417005,
+      "grad_norm": 3.9709234477095103,
+      "learning_rate": 3.870967741935484e-06,
+      "loss": 0.9106,
+      "step": 12
+    },
+    {
+      "epoch": 0.05263157894736842,
+      "grad_norm": 4.213190782963625,
+      "learning_rate": 4.193548387096774e-06,
+      "loss": 0.9158,
+      "step": 13
+    },
+    {
+      "epoch": 0.05668016194331984,
+      "grad_norm": 3.9280476738047283,
+      "learning_rate": 4.516129032258065e-06,
+      "loss": 0.8939,
+      "step": 14
+    },
+    {
+      "epoch": 0.06072874493927125,
+      "grad_norm": 3.638424233684725,
+      "learning_rate": 4.838709677419355e-06,
+      "loss": 0.8983,
+      "step": 15
+    },
+    {
+      "epoch": 0.06477732793522267,
+      "grad_norm": 2.5523908422797623,
+      "learning_rate": 5.161290322580646e-06,
+      "loss": 0.8224,
+      "step": 16
+    },
+    {
+      "epoch": 0.06882591093117409,
+      "grad_norm": 2.4570009188207087,
+      "learning_rate": 5.483870967741935e-06,
+      "loss": 0.8261,
+      "step": 17
+    },
+    {
+      "epoch": 0.0728744939271255,
+      "grad_norm": 1.9950110684170208,
+      "learning_rate": 5.806451612903226e-06,
+      "loss": 0.8071,
+      "step": 18
+    },
+    {
+      "epoch": 0.07692307692307693,
+      "grad_norm": 1.5428421882302115,
+      "learning_rate": 6.129032258064517e-06,
+      "loss": 0.8185,
+      "step": 19
+    },
+    {
+      "epoch": 0.08097165991902834,
+      "grad_norm": 1.1812213300765628,
+      "learning_rate": 6.451612903225806e-06,
+      "loss": 0.783,
+      "step": 20
+    },
+    {
+      "epoch": 0.08502024291497975,
+      "grad_norm": 1.1642470493769075,
+      "learning_rate": 6.774193548387097e-06,
+      "loss": 0.7555,
+      "step": 21
+    },
+    {
+      "epoch": 0.08906882591093117,
+      "grad_norm": 1.3808175360621662,
+      "learning_rate": 7.096774193548388e-06,
+      "loss": 0.7628,
+      "step": 22
+    },
+    {
+      "epoch": 0.0931174089068826,
+      "grad_norm": 1.2116400663235827,
+      "learning_rate": 7.4193548387096784e-06,
+      "loss": 0.738,
+      "step": 23
+    },
+    {
+      "epoch": 0.09716599190283401,
+      "grad_norm": 0.9462862954197553,
+      "learning_rate": 7.741935483870968e-06,
+      "loss": 0.7422,
+      "step": 24
+    },
+    {
+      "epoch": 0.10121457489878542,
+      "grad_norm": 1.0094260323655277,
+      "learning_rate": 8.064516129032258e-06,
+      "loss": 0.7377,
+      "step": 25
+    },
+    {
+      "epoch": 0.10526315789473684,
+      "grad_norm": 1.012383816616436,
+      "learning_rate": 8.387096774193549e-06,
+      "loss": 0.7338,
+      "step": 26
+    },
+    {
+      "epoch": 0.10931174089068826,
+      "grad_norm": 1.0032672838029506,
+      "learning_rate": 8.70967741935484e-06,
+      "loss": 0.7225,
+      "step": 27
+    },
+    {
+      "epoch": 0.11336032388663968,
+      "grad_norm": 0.759394058590696,
+      "learning_rate": 9.03225806451613e-06,
+      "loss": 0.7142,
+      "step": 28
+    },
+    {
+      "epoch": 0.11740890688259109,
+      "grad_norm": 0.8694886839924276,
+      "learning_rate": 9.35483870967742e-06,
+      "loss": 0.7071,
+      "step": 29
+    },
+    {
+      "epoch": 0.1214574898785425,
+      "grad_norm": 0.9579853882638087,
+      "learning_rate": 9.67741935483871e-06,
+      "loss": 0.7059,
+      "step": 30
+    },
+    {
+      "epoch": 0.12550607287449392,
+      "grad_norm": 0.8942383307455263,
+      "learning_rate": 1e-05,
+      "loss": 0.7081,
+      "step": 31
+    },
+    {
+      "epoch": 0.12955465587044535,
+      "grad_norm": 0.6777952719513622,
+      "learning_rate": 1.0322580645161291e-05,
+      "loss": 0.6823,
+      "step": 32
+    },
+    {
+      "epoch": 0.13360323886639677,
+      "grad_norm": 0.6942507872748462,
+      "learning_rate": 1.0645161290322582e-05,
+      "loss": 0.7003,
+      "step": 33
+    },
+    {
+      "epoch": 0.13765182186234817,
+      "grad_norm": 0.861338561014134,
+      "learning_rate": 1.096774193548387e-05,
+      "loss": 0.6956,
+      "step": 34
+    },
+    {
+      "epoch": 0.1417004048582996,
+      "grad_norm": 0.7253144472612186,
+      "learning_rate": 1.1290322580645164e-05,
+      "loss": 0.6904,
+      "step": 35
+    },
+    {
+      "epoch": 0.145748987854251,
+      "grad_norm": 0.646350826269348,
+      "learning_rate": 1.1612903225806453e-05,
+      "loss": 0.6769,
+      "step": 36
+    },
+    {
+      "epoch": 0.14979757085020243,
+      "grad_norm": 0.6450453406626593,
+      "learning_rate": 1.1935483870967743e-05,
+      "loss": 0.6564,
+      "step": 37
+    },
+    {
+      "epoch": 0.15384615384615385,
+      "grad_norm": 0.6006364525154078,
+      "learning_rate": 1.2258064516129034e-05,
+      "loss": 0.6817,
+      "step": 38
+    },
+    {
+      "epoch": 0.15789473684210525,
+      "grad_norm": 0.592485229129751,
+      "learning_rate": 1.2580645161290324e-05,
+      "loss": 0.6698,
+      "step": 39
+    },
+    {
+      "epoch": 0.16194331983805668,
+      "grad_norm": 0.6352694561310026,
+      "learning_rate": 1.2903225806451613e-05,
+      "loss": 0.6763,
+      "step": 40
+    },
+    {
+      "epoch": 0.1659919028340081,
+      "grad_norm": 0.663832935616849,
+      "learning_rate": 1.3225806451612903e-05,
+      "loss": 0.6567,
+      "step": 41
+    },
+    {
+      "epoch": 0.1700404858299595,
+      "grad_norm": 0.6654730244385952,
+      "learning_rate": 1.3548387096774194e-05,
+      "loss": 0.6791,
+      "step": 42
+    },
+    {
+      "epoch": 0.17408906882591094,
+      "grad_norm": 0.5969633787419939,
+      "learning_rate": 1.3870967741935486e-05,
+      "loss": 0.6738,
+      "step": 43
+    },
+    {
+      "epoch": 0.17813765182186234,
+      "grad_norm": 0.8828606252734891,
+      "learning_rate": 1.4193548387096776e-05,
+      "loss": 0.6635,
+      "step": 44
+    },
+    {
+      "epoch": 0.18218623481781376,
+      "grad_norm": 0.6082500943010971,
+      "learning_rate": 1.4516129032258066e-05,
+      "loss": 0.654,
+      "step": 45
+    },
+    {
+      "epoch": 0.1862348178137652,
+      "grad_norm": 0.6361730885818654,
+      "learning_rate": 1.4838709677419357e-05,
+      "loss": 0.6582,
+      "step": 46
+    },
+    {
+      "epoch": 0.1902834008097166,
+      "grad_norm": 0.6847783975739485,
+      "learning_rate": 1.5161290322580646e-05,
+      "loss": 0.6382,
+      "step": 47
+    },
+    {
+      "epoch": 0.19433198380566802,
+      "grad_norm": 0.5146061337863245,
+      "learning_rate": 1.5483870967741936e-05,
+      "loss": 0.6496,
+      "step": 48
+    },
+    {
+      "epoch": 0.19838056680161945,
+      "grad_norm": 0.7330785134471545,
+      "learning_rate": 1.5806451612903226e-05,
+      "loss": 0.6441,
+      "step": 49
+    },
+    {
+      "epoch": 0.20242914979757085,
+      "grad_norm": 0.571870251859315,
+      "learning_rate": 1.6129032258064517e-05,
+      "loss": 0.6423,
+      "step": 50
+    },
+    {
+      "epoch": 0.20647773279352227,
+      "grad_norm": 0.5474982955174669,
+      "learning_rate": 1.6451612903225807e-05,
+      "loss": 0.653,
+      "step": 51
+    },
+    {
+      "epoch": 0.21052631578947367,
+      "grad_norm": 0.6693741602471058,
+      "learning_rate": 1.6774193548387098e-05,
+      "loss": 0.6426,
+      "step": 52
+    },
+    {
+      "epoch": 0.2145748987854251,
+      "grad_norm": 0.6493701451231005,
+      "learning_rate": 1.7096774193548388e-05,
+      "loss": 0.6427,
+      "step": 53
+    },
+    {
+      "epoch": 0.21862348178137653,
+      "grad_norm": 0.8024678480083329,
+      "learning_rate": 1.741935483870968e-05,
+      "loss": 0.6354,
+      "step": 54
+    },
+    {
+      "epoch": 0.22267206477732793,
+      "grad_norm": 0.5818576215407394,
+      "learning_rate": 1.774193548387097e-05,
+      "loss": 0.6271,
+      "step": 55
+    },
+    {
+      "epoch": 0.22672064777327935,
+      "grad_norm": 0.7436140238098647,
+      "learning_rate": 1.806451612903226e-05,
+      "loss": 0.6489,
+      "step": 56
+    },
+    {
+      "epoch": 0.23076923076923078,
+      "grad_norm": 0.6056081789156584,
+      "learning_rate": 1.838709677419355e-05,
+      "loss": 0.6325,
+      "step": 57
+    },
+    {
+      "epoch": 0.23481781376518218,
+      "grad_norm": 0.7250421581099505,
+      "learning_rate": 1.870967741935484e-05,
+      "loss": 0.6299,
+      "step": 58
+    },
+    {
+      "epoch": 0.2388663967611336,
+      "grad_norm": 1.0073933730699423,
+      "learning_rate": 1.903225806451613e-05,
+      "loss": 0.6324,
+      "step": 59
+    },
+    {
+      "epoch": 0.242914979757085,
+      "grad_norm": 0.6586423003991008,
+      "learning_rate": 1.935483870967742e-05,
+      "loss": 0.6329,
+      "step": 60
+    },
+    {
+      "epoch": 0.24696356275303644,
+      "grad_norm": 1.0397750510283632,
+      "learning_rate": 1.967741935483871e-05,
+      "loss": 0.6426,
+      "step": 61
+    },
+    {
+      "epoch": 0.25101214574898784,
+      "grad_norm": 1.1801787527201986,
+      "learning_rate": 2e-05,
+      "loss": 0.6332,
+      "step": 62
+    },
+    {
+      "epoch": 0.2550607287449393,
+      "grad_norm": 0.855981673948009,
+      "learning_rate": 2.0322580645161292e-05,
+      "loss": 0.6403,
+      "step": 63
+    },
+    {
+      "epoch": 0.2591093117408907,
+      "grad_norm": 1.2610747701336373,
+      "learning_rate": 2.0645161290322582e-05,
+      "loss": 0.633,
+      "step": 64
+    },
+    {
+      "epoch": 0.2631578947368421,
+      "grad_norm": 0.8111091476323337,
+      "learning_rate": 2.0967741935483873e-05,
+      "loss": 0.6331,
+      "step": 65
+    },
+    {
+      "epoch": 0.26720647773279355,
+      "grad_norm": 1.427914515060836,
+      "learning_rate": 2.1290322580645163e-05,
+      "loss": 0.6222,
+      "step": 66
+    },
+    {
+      "epoch": 0.27125506072874495,
+      "grad_norm": 0.7720803889396617,
+      "learning_rate": 2.161290322580645e-05,
+      "loss": 0.6257,
+      "step": 67
+    },
+    {
+      "epoch": 0.27530364372469635,
+      "grad_norm": 1.3394234051881018,
+      "learning_rate": 2.193548387096774e-05,
+      "loss": 0.6163,
+      "step": 68
+    },
+    {
+      "epoch": 0.2793522267206478,
+      "grad_norm": 0.7830707343706959,
+      "learning_rate": 2.225806451612903e-05,
+      "loss": 0.6234,
+      "step": 69
+    },
+    {
+      "epoch": 0.2834008097165992,
+      "grad_norm": 1.4841183497956283,
+      "learning_rate": 2.2580645161290328e-05,
+      "loss": 0.6277,
+      "step": 70
+    },
+    {
+      "epoch": 0.2874493927125506,
+      "grad_norm": 1.0814889389011118,
+      "learning_rate": 2.2903225806451615e-05,
+      "loss": 0.6201,
+      "step": 71
+    },
+    {
+      "epoch": 0.291497975708502,
+      "grad_norm": 1.6406418120989408,
+      "learning_rate": 2.3225806451612906e-05,
+      "loss": 0.6218,
+      "step": 72
+    },
+    {
+      "epoch": 0.29554655870445345,
+      "grad_norm": 1.5740218399751802,
+      "learning_rate": 2.3548387096774196e-05,
+      "loss": 0.617,
+      "step": 73
+    },
+    {
+      "epoch": 0.29959514170040485,
+      "grad_norm": 1.0226997015776658,
+      "learning_rate": 2.3870967741935486e-05,
+      "loss": 0.6248,
+      "step": 74
+    },
+    {
+      "epoch": 0.30364372469635625,
+      "grad_norm": 1.3554749808818545,
+      "learning_rate": 2.4193548387096777e-05,
+      "loss": 0.6244,
+      "step": 75
+    },
+    {
+      "epoch": 0.3076923076923077,
+      "grad_norm": 1.2100914461910603,
+      "learning_rate": 2.4516129032258067e-05,
+      "loss": 0.6312,
+      "step": 76
+    },
+    {
+      "epoch": 0.3117408906882591,
+      "grad_norm": 1.1388424434715814,
+      "learning_rate": 2.4838709677419358e-05,
+      "loss": 0.6226,
+      "step": 77
+    },
+    {
+      "epoch": 0.3157894736842105,
+      "grad_norm": 1.3196756209400482,
+      "learning_rate": 2.5161290322580648e-05,
+      "loss": 0.6281,
+      "step": 78
+    },
+    {
+      "epoch": 0.31983805668016196,
+      "grad_norm": 0.8371188957386274,
+      "learning_rate": 2.548387096774194e-05,
+      "loss": 0.6178,
+      "step": 79
+    },
+    {
+      "epoch": 0.32388663967611336,
+      "grad_norm": 1.4011635003224088,
+      "learning_rate": 2.5806451612903226e-05,
+      "loss": 0.6221,
+      "step": 80
+    },
+    {
+      "epoch": 0.32793522267206476,
+      "grad_norm": 0.8487766155600351,
+      "learning_rate": 2.6129032258064516e-05,
+      "loss": 0.6271,
+      "step": 81
+    },
+    {
+      "epoch": 0.3319838056680162,
+      "grad_norm": 1.2013061275113082,
+      "learning_rate": 2.6451612903225806e-05,
+      "loss": 0.6121,
+      "step": 82
+    },
+    {
+      "epoch": 0.3360323886639676,
+      "grad_norm": 0.8835050731002629,
+      "learning_rate": 2.6774193548387097e-05,
+      "loss": 0.6225,
+      "step": 83
+    },
+    {
+      "epoch": 0.340080971659919,
+      "grad_norm": 0.8736298795536251,
+      "learning_rate": 2.7096774193548387e-05,
+      "loss": 0.6054,
+      "step": 84
+    },
+    {
+      "epoch": 0.3441295546558704,
+      "grad_norm": 1.0563940673529495,
+      "learning_rate": 2.7419354838709678e-05,
+      "loss": 0.607,
+      "step": 85
+    },
+    {
+      "epoch": 0.3481781376518219,
+      "grad_norm": 0.7772831885801177,
+      "learning_rate": 2.774193548387097e-05,
+      "loss": 0.6141,
+      "step": 86
+    },
+    {
+      "epoch": 0.3522267206477733,
+      "grad_norm": 1.2482572457210457,
+      "learning_rate": 2.8064516129032262e-05,
+      "loss": 0.6163,
+      "step": 87
+    },
+    {
+      "epoch": 0.3562753036437247,
+      "grad_norm": 0.9048967059351996,
+      "learning_rate": 2.8387096774193552e-05,
+      "loss": 0.6144,
+      "step": 88
+    },
+    {
+      "epoch": 0.3603238866396761,
+      "grad_norm": 1.0701476589984942,
+      "learning_rate": 2.8709677419354843e-05,
+      "loss": 0.6121,
+      "step": 89
+    },
+    {
+      "epoch": 0.3643724696356275,
+      "grad_norm": 0.8791964845902998,
+      "learning_rate": 2.9032258064516133e-05,
+      "loss": 0.6142,
+      "step": 90
+    },
+    {
+      "epoch": 0.3684210526315789,
+      "grad_norm": 1.1468725468143601,
+      "learning_rate": 2.9354838709677423e-05,
+      "loss": 0.6047,
+      "step": 91
+    },
+    {
+      "epoch": 0.3724696356275304,
+      "grad_norm": 0.8187254197820539,
+      "learning_rate": 2.9677419354838714e-05,
+      "loss": 0.6087,
+      "step": 92
+    },
+    {
+      "epoch": 0.3765182186234818,
+      "grad_norm": 1.024453136521935,
+      "learning_rate": 3.0000000000000004e-05,
+      "loss": 0.6165,
+      "step": 93
+    },
+    {
+      "epoch": 0.3805668016194332,
+      "grad_norm": 1.1543510928767098,
+      "learning_rate": 3.032258064516129e-05,
+      "loss": 0.6027,
+      "step": 94
+    },
+    {
+      "epoch": 0.38461538461538464,
+      "grad_norm": 0.8341677494746017,
+      "learning_rate": 3.0645161290322585e-05,
+      "loss": 0.6145,
+      "step": 95
+    },
+    {
+      "epoch": 0.38866396761133604,
+      "grad_norm": 1.1471556644224667,
+      "learning_rate": 3.096774193548387e-05,
+      "loss": 0.617,
+      "step": 96
+    },
+    {
+      "epoch": 0.39271255060728744,
+      "grad_norm": 1.189402943642498,
+      "learning_rate": 3.1290322580645166e-05,
+      "loss": 0.6026,
+      "step": 97
+    },
+    {
+      "epoch": 0.3967611336032389,
+      "grad_norm": 1.2569841069298144,
+      "learning_rate": 3.161290322580645e-05,
+      "loss": 0.6112,
+      "step": 98
+    },
+    {
+      "epoch": 0.4008097165991903,
+      "grad_norm": 0.9236919976155931,
+      "learning_rate": 3.193548387096774e-05,
+      "loss": 0.6047,
+      "step": 99
+    },
+    {
+      "epoch": 0.4048582995951417,
+      "grad_norm": 0.898300266225928,
+      "learning_rate": 3.2258064516129034e-05,
+      "loss": 0.5926,
+      "step": 100
+    },
+    {
+      "epoch": 0.4089068825910931,
+      "grad_norm": 1.027290835491899,
+      "learning_rate": 3.258064516129033e-05,
+      "loss": 0.6071,
+      "step": 101
+    },
+    {
+      "epoch": 0.41295546558704455,
+      "grad_norm": 0.8372838718012654,
+      "learning_rate": 3.2903225806451614e-05,
+      "loss": 0.6052,
+      "step": 102
+    },
+    {
+      "epoch": 0.41700404858299595,
+      "grad_norm": 0.8302203560848547,
+      "learning_rate": 3.322580645161291e-05,
+      "loss": 0.6152,
+      "step": 103
+    },
+    {
+      "epoch": 0.42105263157894735,
+      "grad_norm": 0.9011483393267162,
+      "learning_rate": 3.3548387096774195e-05,
+      "loss": 0.6051,
+      "step": 104
+    },
+    {
+      "epoch": 0.4251012145748988,
+      "grad_norm": 0.866132380549316,
+      "learning_rate": 3.387096774193549e-05,
+      "loss": 0.6035,
+      "step": 105
+    },
+    {
+      "epoch": 0.4291497975708502,
+      "grad_norm": 1.3282796194242303,
+      "learning_rate": 3.4193548387096776e-05,
+      "loss": 0.6003,
+      "step": 106
+    },
+    {
+      "epoch": 0.4331983805668016,
+      "grad_norm": 0.9648248881144394,
+      "learning_rate": 3.451612903225807e-05,
+      "loss": 0.6093,
+      "step": 107
+    },
+    {
+      "epoch": 0.43724696356275305,
+      "grad_norm": 0.8697438032067571,
+      "learning_rate": 3.483870967741936e-05,
+      "loss": 0.6138,
+      "step": 108
+    },
+    {
+      "epoch": 0.44129554655870445,
+      "grad_norm": 0.946569645868162,
+      "learning_rate": 3.516129032258065e-05,
+      "loss": 0.6028,
+      "step": 109
+    },
+    {
+      "epoch": 0.44534412955465585,
+      "grad_norm": 1.086632496717751,
+      "learning_rate": 3.548387096774194e-05,
+      "loss": 0.611,
+      "step": 110
+    },
+    {
+      "epoch": 0.4493927125506073,
+      "grad_norm": 0.8832864087522655,
+      "learning_rate": 3.5806451612903225e-05,
+      "loss": 0.6022,
+      "step": 111
+    },
+    {
+      "epoch": 0.4534412955465587,
+      "grad_norm": 1.3127722110482944,
+      "learning_rate": 3.612903225806452e-05,
+      "loss": 0.6055,
+      "step": 112
+    },
+    {
+      "epoch": 0.4574898785425101,
+      "grad_norm": 1.1989976586336497,
+      "learning_rate": 3.6451612903225805e-05,
+      "loss": 0.6077,
+      "step": 113
+    },
+    {
+      "epoch": 0.46153846153846156,
+      "grad_norm": 0.9878920559099328,
+      "learning_rate": 3.67741935483871e-05,
+      "loss": 0.6137,
+      "step": 114
+    },
+    {
+      "epoch": 0.46558704453441296,
+      "grad_norm": 1.4912318524762471,
+      "learning_rate": 3.7096774193548386e-05,
+      "loss": 0.5923,
+      "step": 115
+    },
+    {
+      "epoch": 0.46963562753036436,
+      "grad_norm": 1.2125791051363142,
+      "learning_rate": 3.741935483870968e-05,
+      "loss": 0.6181,
+      "step": 116
+    },
+    {
+      "epoch": 0.47368421052631576,
+      "grad_norm": 1.4338269878747167,
+      "learning_rate": 3.7741935483870974e-05,
+      "loss": 0.6138,
+      "step": 117
+    },
+    {
+      "epoch": 0.4777327935222672,
+      "grad_norm": 1.601289083509039,
+      "learning_rate": 3.806451612903226e-05,
+      "loss": 0.6181,
+      "step": 118
+    },
+    {
+      "epoch": 0.4817813765182186,
+      "grad_norm": 1.230487296788433,
+      "learning_rate": 3.8387096774193555e-05,
+      "loss": 0.6094,
+      "step": 119
+    },
+    {
+      "epoch": 0.48582995951417,
+      "grad_norm": 1.9191067564291475,
+      "learning_rate": 3.870967741935484e-05,
+      "loss": 0.6094,
+      "step": 120
+    },
+    {
+      "epoch": 0.4898785425101215,
+      "grad_norm": 1.447872341009961,
+      "learning_rate": 3.9032258064516135e-05,
+      "loss": 0.6053,
+      "step": 121
+    },
+    {
+      "epoch": 0.4939271255060729,
+      "grad_norm": 1.9112191786565251,
+      "learning_rate": 3.935483870967742e-05,
+      "loss": 0.6064,
+      "step": 122
+    },
+    {
+      "epoch": 0.4979757085020243,
+      "grad_norm": 1.5869576003864163,
+      "learning_rate": 3.9677419354838716e-05,
+      "loss": 0.6136,
+      "step": 123
+    },
+    {
+      "epoch": 0.5020242914979757,
+      "grad_norm": 1.6954886129156599,
+      "learning_rate": 4e-05,
+      "loss": 0.5918,
+      "step": 124
+    },
+    {
+      "epoch": 0.5060728744939271,
+      "grad_norm": 1.2244846024802667,
+      "learning_rate": 3.9999920040266474e-05,
+      "loss": 0.6027,
+      "step": 125
+    },
+    {
+      "epoch": 0.5101214574898786,
+      "grad_norm": 1.654526416716931,
+      "learning_rate": 3.999968016170525e-05,
+      "loss": 0.6055,
+      "step": 126
+    },
+    {
+      "epoch": 0.5141700404858299,
+      "grad_norm": 1.3179806425543272,
+      "learning_rate": 3.9999280366234394e-05,
+      "loss": 0.5898,
+      "step": 127
+    },
+    {
+      "epoch": 0.5182186234817814,
+      "grad_norm": 1.7436793528317545,
+      "learning_rate": 3.999872065705065e-05,
+      "loss": 0.6051,
+      "step": 128
+    },
+    {
+      "epoch": 0.5222672064777328,
+      "grad_norm": 1.4360427765225672,
+      "learning_rate": 3.999800103862945e-05,
+      "loss": 0.5959,
+      "step": 129
+    },
+    {
+      "epoch": 0.5263157894736842,
+      "grad_norm": 1.3671442382084646,
+      "learning_rate": 3.9997121516724834e-05,
+      "loss": 0.5982,
+      "step": 130
+    },
+    {
+      "epoch": 0.5303643724696356,
+      "grad_norm": 1.2605129603145393,
+      "learning_rate": 3.999608209836944e-05,
+      "loss": 0.5843,
+      "step": 131
+    },
+    {
+      "epoch": 0.5344129554655871,
+      "grad_norm": 1.2101029947221198,
+      "learning_rate": 3.999488279187443e-05,
+      "loss": 0.6023,
+      "step": 132
+    },
+    {
+      "epoch": 0.5384615384615384,
+      "grad_norm": 1.1443208465754071,
+      "learning_rate": 3.9993523606829425e-05,
+      "loss": 0.5911,
+      "step": 133
+    },
+    {
+      "epoch": 0.5425101214574899,
+      "grad_norm": 1.16468674740541,
+      "learning_rate": 3.9992004554102435e-05,
+      "loss": 0.5856,
+      "step": 134
+    },
+    {
+      "epoch": 0.5465587044534413,
+      "grad_norm": 0.8708864168617115,
+      "learning_rate": 3.999032564583976e-05,
+      "loss": 0.5831,
+      "step": 135
+    },
+    {
+      "epoch": 0.5506072874493927,
+      "grad_norm": 1.118753224682545,
+      "learning_rate": 3.998848689546591e-05,
+      "loss": 0.5999,
+      "step": 136
+    },
+    {
+      "epoch": 0.5546558704453441,
+      "grad_norm": 0.6749641074649334,
+      "learning_rate": 3.9986488317683496e-05,
+      "loss": 0.5969,
+      "step": 137
+    },
+    {
+      "epoch": 0.5587044534412956,
+      "grad_norm": 0.8410398144090315,
+      "learning_rate": 3.9984329928473067e-05,
+      "loss": 0.601,
+      "step": 138
+    },
+    {
+      "epoch": 0.562753036437247,
+      "grad_norm": 0.7723239503236059,
+      "learning_rate": 3.998201174509306e-05,
+      "loss": 0.5977,
+      "step": 139
+    },
+    {
+      "epoch": 0.5668016194331984,
+      "grad_norm": 0.9579296869385828,
+      "learning_rate": 3.997953378607961e-05,
+      "loss": 0.5991,
+      "step": 140
+    },
+    {
+      "epoch": 0.5708502024291497,
+      "grad_norm": 0.5909580848055229,
+      "learning_rate": 3.9976896071246403e-05,
+      "loss": 0.6049,
+      "step": 141
+    },
+    {
+      "epoch": 0.5748987854251012,
+      "grad_norm": 0.8189334355885878,
+      "learning_rate": 3.997409862168454e-05,
+      "loss": 0.5994,
+      "step": 142
+    },
+    {
+      "epoch": 0.5789473684210527,
+      "grad_norm": 0.6443396498472426,
+      "learning_rate": 3.997114145976237e-05,
+      "loss": 0.589,
+      "step": 143
+    },
+    {
+      "epoch": 0.582995951417004,
+      "grad_norm": 0.8770787428282857,
+      "learning_rate": 3.9968024609125245e-05,
+      "loss": 0.5904,
+      "step": 144
+    },
+    {
+      "epoch": 0.5870445344129555,
+      "grad_norm": 0.8154968956261548,
+      "learning_rate": 3.996474809469545e-05,
+      "loss": 0.5939,
+      "step": 145
+    },
+    {
+      "epoch": 0.5910931174089069,
+      "grad_norm": 0.6298376618466626,
+      "learning_rate": 3.996131194267189e-05,
+      "loss": 0.5787,
+      "step": 146
+    },
+    {
+      "epoch": 0.5951417004048583,
+      "grad_norm": 0.8360834114274385,
+      "learning_rate": 3.995771618052995e-05,
+      "loss": 0.5861,
+      "step": 147
+    },
+    {
+      "epoch": 0.5991902834008097,
+      "grad_norm": 0.7018576497111711,
+      "learning_rate": 3.9953960837021256e-05,
+      "loss": 0.5891,
+      "step": 148
+    },
+    {
+      "epoch": 0.6032388663967612,
+      "grad_norm": 1.0506293415060293,
+      "learning_rate": 3.995004594217343e-05,
+      "loss": 0.5931,
+      "step": 149
+    },
+    {
+      "epoch": 0.6072874493927125,
+      "grad_norm": 0.9429431693011433,
+      "learning_rate": 3.994597152728987e-05,
+      "loss": 0.5871,
+      "step": 150
+    },
+    {
+      "epoch": 0.611336032388664,
+      "grad_norm": 0.7182836201584235,
+      "learning_rate": 3.994173762494947e-05,
+      "loss": 0.5856,
+      "step": 151
+    },
+    {
+      "epoch": 0.6153846153846154,
+      "grad_norm": 0.8574845866040939,
+      "learning_rate": 3.9937344269006424e-05,
+      "loss": 0.5932,
+      "step": 152
+    },
+    {
+      "epoch": 0.6194331983805668,
+      "grad_norm": 0.6453784261162039,
+      "learning_rate": 3.993279149458988e-05,
+      "loss": 0.5867,
+      "step": 153
+    },
+    {
+      "epoch": 0.6234817813765182,
+      "grad_norm": 0.748195857480529,
+      "learning_rate": 3.9928079338103695e-05,
+      "loss": 0.5882,
+      "step": 154
+    },
+    {
+      "epoch": 0.6275303643724697,
+      "grad_norm": 0.7862966782266547,
+      "learning_rate": 3.992320783722616e-05,
+      "loss": 0.5972,
+      "step": 155
+    },
+    {
+      "epoch": 0.631578947368421,
+      "grad_norm": 0.592175453607095,
+      "learning_rate": 3.991817703090965e-05,
+      "loss": 0.5953,
+      "step": 156
+    },
+    {
+      "epoch": 0.6356275303643725,
+      "grad_norm": 0.7408223487440692,
+      "learning_rate": 3.991298695938038e-05,
+      "loss": 0.5956,
+      "step": 157
+    },
+    {
+      "epoch": 0.6396761133603239,
+      "grad_norm": 0.7839923246313646,
+      "learning_rate": 3.9907637664138e-05,
+      "loss": 0.5893,
+      "step": 158
+    },
+    {
+      "epoch": 0.6437246963562753,
+      "grad_norm": 0.5749746763028375,
+      "learning_rate": 3.990212918795534e-05,
+      "loss": 0.5881,
+      "step": 159
+    },
+    {
+      "epoch": 0.6477732793522267,
+      "grad_norm": 1.1285157252701952,
+      "learning_rate": 3.9896461574878036e-05,
+      "loss": 0.5881,
+      "step": 160
+    },
+    {
+      "epoch": 0.6518218623481782,
+      "grad_norm": 0.77877107481799,
+      "learning_rate": 3.989063487022416e-05,
+      "loss": 0.5874,
+      "step": 161
+    },
+    {
+      "epoch": 0.6558704453441295,
+      "grad_norm": 0.7438136660167506,
+      "learning_rate": 3.9884649120583903e-05,
+      "loss": 0.5801,
+      "step": 162
+    },
+    {
+      "epoch": 0.659919028340081,
+      "grad_norm": 0.9128468335637063,
+      "learning_rate": 3.987850437381915e-05,
+      "loss": 0.587,
+      "step": 163
+    },
+    {
+      "epoch": 0.6639676113360324,
+      "grad_norm": 0.780729817727738,
+      "learning_rate": 3.987220067906313e-05,
+      "loss": 0.5837,
+      "step": 164
+    },
+    {
+      "epoch": 0.6680161943319838,
+      "grad_norm": 0.6368840450498557,
+      "learning_rate": 3.986573808672002e-05,
+      "loss": 0.5793,
+      "step": 165
+    },
+    {
+      "epoch": 0.6720647773279352,
+      "grad_norm": 0.6470694242097893,
+      "learning_rate": 3.9859116648464534e-05,
+      "loss": 0.5913,
+      "step": 166
+    },
+    {
+      "epoch": 0.6761133603238867,
+      "grad_norm": 0.674877851815755,
+      "learning_rate": 3.985233641724152e-05,
+      "loss": 0.5693,
+      "step": 167
+    },
+    {
+      "epoch": 0.680161943319838,
+      "grad_norm": 0.6858342131817494,
+      "learning_rate": 3.984539744726554e-05,
+      "loss": 0.5847,
+      "step": 168
+    },
+    {
+      "epoch": 0.6842105263157895,
+      "grad_norm": 0.9583333404685312,
+      "learning_rate": 3.983829979402038e-05,
+      "loss": 0.589,
+      "step": 169
+    },
+    {
+      "epoch": 0.6882591093117408,
+      "grad_norm": 0.8866877888256087,
+      "learning_rate": 3.983104351425871e-05,
+      "loss": 0.5863,
+      "step": 170
+    },
+    {
+      "epoch": 0.6923076923076923,
+      "grad_norm": 0.6319914784192151,
+      "learning_rate": 3.982362866600154e-05,
+      "loss": 0.5917,
+      "step": 171
+    },
+    {
+      "epoch": 0.6963562753036437,
+      "grad_norm": 0.6747835693960595,
+      "learning_rate": 3.9816055308537806e-05,
+      "loss": 0.5917,
+      "step": 172
+    },
+    {
+      "epoch": 0.7004048582995951,
+      "grad_norm": 0.475596187742653,
+      "learning_rate": 3.980832350242388e-05,
+      "loss": 0.5898,
+      "step": 173
+    },
+    {
+      "epoch": 0.7044534412955465,
+      "grad_norm": 0.7009170745311305,
+      "learning_rate": 3.980043330948306e-05,
+      "loss": 0.5792,
+      "step": 174
+    },
+    {
+      "epoch": 0.708502024291498,
+      "grad_norm": 0.6899334358937455,
+      "learning_rate": 3.9792384792805124e-05,
+      "loss": 0.5805,
+      "step": 175
+    },
+    {
+      "epoch": 0.7125506072874493,
+      "grad_norm": 0.9047599292938838,
+      "learning_rate": 3.978417801674579e-05,
+      "loss": 0.5933,
+      "step": 176
+    },
+    {
+      "epoch": 0.7165991902834008,
+      "grad_norm": 0.7628534411638808,
+      "learning_rate": 3.9775813046926244e-05,
+      "loss": 0.5789,
+      "step": 177
+    },
+    {
+      "epoch": 0.7206477732793523,
+      "grad_norm": 0.5547952610430908,
+      "learning_rate": 3.976728995023254e-05,
+      "loss": 0.5776,
+      "step": 178
+    },
+    {
+      "epoch": 0.7246963562753036,
+      "grad_norm": 0.8052364333626093,
+      "learning_rate": 3.975860879481514e-05,
+      "loss": 0.5785,
+      "step": 179
+    },
+    {
+      "epoch": 0.728744939271255,
+      "grad_norm": 1.0792876937994944,
+      "learning_rate": 3.974976965008833e-05,
+      "loss": 0.5798,
+      "step": 180
+    },
+    {
+      "epoch": 0.7327935222672065,
+      "grad_norm": 0.9049825083225307,
+      "learning_rate": 3.9740772586729674e-05,
+      "loss": 0.5837,
+      "step": 181
+    },
+    {
+      "epoch": 0.7368421052631579,
+      "grad_norm": 0.5890966484794977,
+      "learning_rate": 3.973161767667946e-05,
+      "loss": 0.593,
+      "step": 182
+    },
+    {
+      "epoch": 0.7408906882591093,
+      "grad_norm": 0.5523645048512419,
+      "learning_rate": 3.972230499314009e-05,
+      "loss": 0.59,
+      "step": 183
+    },
+    {
+      "epoch": 0.7449392712550608,
+      "grad_norm": 0.5807386844388263,
+      "learning_rate": 3.9712834610575545e-05,
+      "loss": 0.5831,
+      "step": 184
+    },
+    {
+      "epoch": 0.7489878542510121,
+      "grad_norm": 0.6194670284075097,
+      "learning_rate": 3.9703206604710746e-05,
+      "loss": 0.5851,
+      "step": 185
+    },
+    {
+      "epoch": 0.7530364372469636,
+      "grad_norm": 0.6760178616656414,
+      "learning_rate": 3.969342105253098e-05,
+      "loss": 0.5699,
+      "step": 186
+    },
+    {
+      "epoch": 0.757085020242915,
+      "grad_norm": 0.739015450112533,
+      "learning_rate": 3.9683478032281246e-05,
+      "loss": 0.5831,
+      "step": 187
+    },
+    {
+      "epoch": 0.7611336032388664,
+      "grad_norm": 0.605813728070305,
+      "learning_rate": 3.967337762346568e-05,
+      "loss": 0.5737,
+      "step": 188
+    },
+    {
+      "epoch": 0.7651821862348178,
+      "grad_norm": 0.6256080722266263,
+      "learning_rate": 3.9663119906846885e-05,
+      "loss": 0.5778,
+      "step": 189
+    },
+    {
+      "epoch": 0.7692307692307693,
+      "grad_norm": 0.8028157060605364,
+      "learning_rate": 3.965270496444528e-05,
+      "loss": 0.584,
+      "step": 190
+    },
+    {
+      "epoch": 0.7732793522267206,
+      "grad_norm": 0.5373136897036741,
+      "learning_rate": 3.9642132879538465e-05,
+      "loss": 0.5811,
+      "step": 191
+    },
+    {
+      "epoch": 0.7773279352226721,
+      "grad_norm": 0.5981964725636449,
+      "learning_rate": 3.9631403736660566e-05,
+      "loss": 0.5782,
+      "step": 192
+    },
+    {
+      "epoch": 0.7813765182186235,
+      "grad_norm": 0.5438911681809323,
+      "learning_rate": 3.962051762160151e-05,
+      "loss": 0.5831,
+      "step": 193
+    },
+    {
+      "epoch": 0.7854251012145749,
+      "grad_norm": 0.5344199586215209,
+      "learning_rate": 3.960947462140638e-05,
+      "loss": 0.5762,
+      "step": 194
+    },
+    {
+      "epoch": 0.7894736842105263,
+      "grad_norm": 0.5973671080335308,
+      "learning_rate": 3.959827482437472e-05,
+      "loss": 0.5768,
+      "step": 195
+    },
+    {
+      "epoch": 0.7935222672064778,
+      "grad_norm": 0.6720513143444918,
+      "learning_rate": 3.958691832005981e-05,
+      "loss": 0.5767,
+      "step": 196
+    },
+    {
+      "epoch": 0.7975708502024291,
+      "grad_norm": 0.6539502562926091,
+      "learning_rate": 3.9575405199267944e-05,
+      "loss": 0.5837,
+      "step": 197
+    },
+    {
+      "epoch": 0.8016194331983806,
+      "grad_norm": 0.5613192518611195,
+      "learning_rate": 3.956373555405774e-05,
+      "loss": 0.5786,
+      "step": 198
+    },
+    {
+      "epoch": 0.805668016194332,
+      "grad_norm": 0.4943716772594813,
+      "learning_rate": 3.955190947773937e-05,
+      "loss": 0.5726,
+      "step": 199
+    },
+    {
+      "epoch": 0.8097165991902834,
+      "grad_norm": 0.563803546138494,
+      "learning_rate": 3.9539927064873824e-05,
+      "loss": 0.5716,
+      "step": 200
+    },
+    {
+      "epoch": 0.8137651821862348,
+      "grad_norm": 0.5002334348842264,
+      "learning_rate": 3.952778841127214e-05,
+      "loss": 0.5822,
+      "step": 201
+    },
+    {
+      "epoch": 0.8178137651821862,
+      "grad_norm": 0.6052173982571983,
+      "learning_rate": 3.95154936139947e-05,
+      "loss": 0.576,
+      "step": 202
+    },
+    {
+      "epoch": 0.8218623481781376,
+      "grad_norm": 0.547209457143202,
+      "learning_rate": 3.950304277135034e-05,
+      "loss": 0.5834,
+      "step": 203
+    },
+    {
+      "epoch": 0.8259109311740891,
+      "grad_norm": 0.6321113023660295,
+      "learning_rate": 3.94904359828957e-05,
+      "loss": 0.58,
+      "step": 204
+    },
+    {
+      "epoch": 0.8299595141700404,
+      "grad_norm": 0.6012164448095793,
+      "learning_rate": 3.94776733494343e-05,
+      "loss": 0.585,
+      "step": 205
+    },
+    {
+      "epoch": 0.8340080971659919,
+      "grad_norm": 0.49841440921040164,
+      "learning_rate": 3.9464754973015826e-05,
+      "loss": 0.5685,
+      "step": 206
+    },
+    {
+      "epoch": 0.8380566801619433,
+      "grad_norm": 0.5354281974440851,
+      "learning_rate": 3.9451680956935275e-05,
+      "loss": 0.5686,
+      "step": 207
+    },
+    {
+      "epoch": 0.8421052631578947,
+      "grad_norm": 0.5243140598923297,
+      "learning_rate": 3.943845140573213e-05,
+      "loss": 0.5737,
+      "step": 208
+    },
+    {
+      "epoch": 0.8461538461538461,
+      "grad_norm": 0.5251161876463946,
+      "learning_rate": 3.942506642518952e-05,
+      "loss": 0.5719,
+      "step": 209
+    },
+    {
+      "epoch": 0.8502024291497976,
+      "grad_norm": 0.5329766481349006,
+      "learning_rate": 3.9411526122333396e-05,
+      "loss": 0.5717,
+      "step": 210
+    },
+    {
+      "epoch": 0.854251012145749,
+      "grad_norm": 0.5018200666221081,
+      "learning_rate": 3.939783060543167e-05,
+      "loss": 0.5687,
+      "step": 211
+    },
+    {
+      "epoch": 0.8582995951417004,
+      "grad_norm": 0.5103124984288306,
+      "learning_rate": 3.938397998399333e-05,
+      "loss": 0.5723,
+      "step": 212
+    },
+    {
+      "epoch": 0.8623481781376519,
+      "grad_norm": 0.5365802232267306,
+      "learning_rate": 3.936997436876756e-05,
+      "loss": 0.5846,
+      "step": 213
+    },
+    {
+      "epoch": 0.8663967611336032,
+      "grad_norm": 0.5044703350714909,
+      "learning_rate": 3.935581387174289e-05,
+      "loss": 0.5812,
+      "step": 214
+    },
+    {
+      "epoch": 0.8704453441295547,
+      "grad_norm": 0.629548489838727,
+      "learning_rate": 3.9341498606146295e-05,
+      "loss": 0.5718,
+      "step": 215
+    },
+    {
+      "epoch": 0.8744939271255061,
+      "grad_norm": 0.5680249465628956,
+      "learning_rate": 3.932702868644224e-05,
+      "loss": 0.5717,
+      "step": 216
+    },
+    {
+      "epoch": 0.8785425101214575,
+      "grad_norm": 0.45986144205314855,
+      "learning_rate": 3.931240422833181e-05,
+      "loss": 0.5707,
+      "step": 217
+    },
+    {
+      "epoch": 0.8825910931174089,
+      "grad_norm": 0.4910755317702273,
+      "learning_rate": 3.9297625348751806e-05,
+      "loss": 0.5695,
+      "step": 218
+    },
+    {
+      "epoch": 0.8866396761133604,
+      "grad_norm": 0.5804442458062041,
+      "learning_rate": 3.9282692165873744e-05,
+      "loss": 0.5711,
+      "step": 219
+    },
+    {
+      "epoch": 0.8906882591093117,
+      "grad_norm": 0.45873432938214237,
+      "learning_rate": 3.9267604799102945e-05,
+      "loss": 0.5769,
+      "step": 220
+    },
+    {
+      "epoch": 0.8947368421052632,
+      "grad_norm": 0.5336737346215097,
+      "learning_rate": 3.9252363369077615e-05,
+      "loss": 0.566,
+      "step": 221
+    },
+    {
+      "epoch": 0.8987854251012146,
+      "grad_norm": 0.5600370604999467,
+      "learning_rate": 3.9236967997667796e-05,
+      "loss": 0.573,
+      "step": 222
+    },
+    {
+      "epoch": 0.902834008097166,
+      "grad_norm": 0.532710437001981,
+      "learning_rate": 3.9221418807974496e-05,
+      "loss": 0.574,
+      "step": 223
+    },
+    {
+      "epoch": 0.9068825910931174,
+      "grad_norm": 0.5754736899776985,
+      "learning_rate": 3.92057159243286e-05,
+      "loss": 0.5743,
+      "step": 224
+    },
+    {
+      "epoch": 0.9109311740890689,
+      "grad_norm": 0.5068739058573446,
+      "learning_rate": 3.9189859472289956e-05,
+      "loss": 0.5727,
+      "step": 225
+    },
+    {
+      "epoch": 0.9149797570850202,
+      "grad_norm": 0.5301720329717613,
+      "learning_rate": 3.9173849578646324e-05,
+      "loss": 0.5791,
+      "step": 226
+    },
+    {
+      "epoch": 0.9190283400809717,
+      "grad_norm": 0.6327962777635012,
+      "learning_rate": 3.9157686371412395e-05,
+      "loss": 0.5731,
+      "step": 227
+    },
+    {
+      "epoch": 0.9230769230769231,
+      "grad_norm": 0.46113950500448714,
+      "learning_rate": 3.914136997982874e-05,
+      "loss": 0.5809,
+      "step": 228
+    },
+    {
+      "epoch": 0.9271255060728745,
+      "grad_norm": 0.5128593134387284,
+      "learning_rate": 3.912490053436079e-05,
+      "loss": 0.5708,
+      "step": 229
+    },
+    {
+      "epoch": 0.9311740890688259,
+      "grad_norm": 0.6510487635034299,
+      "learning_rate": 3.9108278166697805e-05,
+      "loss": 0.5787,
+      "step": 230
+    },
+    {
+      "epoch": 0.9352226720647774,
+      "grad_norm": 0.50041153430379,
+      "learning_rate": 3.909150300975178e-05,
+      "loss": 0.5701,
+      "step": 231
+    },
+    {
+      "epoch": 0.9392712550607287,
+      "grad_norm": 0.598360043890475,
+      "learning_rate": 3.907457519765642e-05,
+      "loss": 0.5682,
+      "step": 232
+    },
+    {
+      "epoch": 0.9433198380566802,
+      "grad_norm": 0.6422006485043619,
+      "learning_rate": 3.905749486576607e-05,
+      "loss": 0.5712,
+      "step": 233
+    },
+    {
+      "epoch": 0.9473684210526315,
+      "grad_norm": 0.48993146029760176,
+      "learning_rate": 3.90402621506546e-05,
+      "loss": 0.5751,
+      "step": 234
+    },
+    {
+      "epoch": 0.951417004048583,
+      "grad_norm": 0.5202019444091218,
+      "learning_rate": 3.902287719011435e-05,
+      "loss": 0.571,
+      "step": 235
+    },
+    {
+      "epoch": 0.9554655870445344,
+      "grad_norm": 0.5431611022006602,
+      "learning_rate": 3.900534012315499e-05,
+      "loss": 0.5737,
+      "step": 236
+    },
+    {
+      "epoch": 0.9595141700404858,
+      "grad_norm": 0.49940254758342234,
+      "learning_rate": 3.8987651090002454e-05,
+      "loss": 0.5743,
+      "step": 237
+    },
+    {
+      "epoch": 0.9635627530364372,
+      "grad_norm": 0.4666146656822946,
+      "learning_rate": 3.896981023209777e-05,
+      "loss": 0.5681,
+      "step": 238
+    },
+    {
+      "epoch": 0.9676113360323887,
+      "grad_norm": 0.5840341260361247,
+      "learning_rate": 3.8951817692095966e-05,
+      "loss": 0.5846,
+      "step": 239
+    },
+    {
+      "epoch": 0.97165991902834,
+      "grad_norm": 0.49990438119873204,
+      "learning_rate": 3.893367361386491e-05,
+      "loss": 0.5714,
+      "step": 240
+    },
+    {
+      "epoch": 0.9757085020242915,
+      "grad_norm": 0.5907336296119943,
+      "learning_rate": 3.891537814248417e-05,
+      "loss": 0.5786,
+      "step": 241
+    },
+    {
+      "epoch": 0.979757085020243,
+      "grad_norm": 0.6120034599914989,
+      "learning_rate": 3.889693142424384e-05,
+      "loss": 0.586,
+      "step": 242
+    },
+    {
+      "epoch": 0.9838056680161943,
+      "grad_norm": 0.512351307167834,
+      "learning_rate": 3.8878333606643405e-05,
+      "loss": 0.5761,
+      "step": 243
+    },
+    {
+      "epoch": 0.9878542510121457,
+      "grad_norm": 0.5576325628173413,
+      "learning_rate": 3.885958483839049e-05,
+      "loss": 0.5832,
+      "step": 244
+    },
+    {
+      "epoch": 0.9919028340080972,
+      "grad_norm": 0.5607896998316962,
+      "learning_rate": 3.8840685269399786e-05,
+      "loss": 0.5917,
+      "step": 245
+    },
+    {
+      "epoch": 0.9959514170040485,
+      "grad_norm": 0.5108249735004766,
+      "learning_rate": 3.882163505079171e-05,
+      "loss": 0.5704,
+      "step": 246
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.44274138190807594,
+      "learning_rate": 3.880243433489132e-05,
+      "loss": 0.5735,
+      "step": 247
+    },
+    {
+      "epoch": 1.0040485829959513,
+      "grad_norm": 0.5469055827191645,
+      "learning_rate": 3.878308327522702e-05,
+      "loss": 0.5248,
+      "step": 248
+    },
+    {
+      "epoch": 1.008097165991903,
+      "grad_norm": 0.7342609729108807,
+      "learning_rate": 3.876358202652936e-05,
+      "loss": 0.5156,
+      "step": 249
+    },
+    {
+      "epoch": 1.0121457489878543,
+      "grad_norm": 0.6539317148115079,
+      "learning_rate": 3.8743930744729835e-05,
+      "loss": 0.5098,
+      "step": 250
+    },
+    {
+      "epoch": 1.0161943319838056,
+      "grad_norm": 0.6773429952236245,
+      "learning_rate": 3.8724129586959535e-05,
+      "loss": 0.5137,
+      "step": 251
+    },
+    {
+      "epoch": 1.0202429149797572,
+      "grad_norm": 0.6254374084449372,
+      "learning_rate": 3.870417871154801e-05,
+      "loss": 0.5055,
+      "step": 252
+    },
+    {
+      "epoch": 1.0242914979757085,
+      "grad_norm": 0.4399066219326448,
+      "learning_rate": 3.868407827802192e-05,
+      "loss": 0.5104,
+      "step": 253
+    },
+    {
+      "epoch": 1.0283400809716599,
+      "grad_norm": 0.7785640928449381,
+      "learning_rate": 3.8663828447103804e-05,
+      "loss": 0.5139,
+      "step": 254
+    },
+    {
+      "epoch": 1.0323886639676114,
+      "grad_norm": 0.8700009126741948,
+      "learning_rate": 3.8643429380710765e-05,
+      "loss": 0.5138,
+      "step": 255
+    },
+    {
+      "epoch": 1.0364372469635628,
+      "grad_norm": 0.776925005400892,
+      "learning_rate": 3.862288124195319e-05,
+      "loss": 0.5186,
+      "step": 256
+    },
+    {
+      "epoch": 1.040485829959514,
+      "grad_norm": 0.7761006250019951,
+      "learning_rate": 3.8602184195133464e-05,
+      "loss": 0.5038,
+      "step": 257
+    },
+    {
+      "epoch": 1.0445344129554657,
+      "grad_norm": 0.6612069813321432,
+      "learning_rate": 3.8581338405744604e-05,
+      "loss": 0.5069,
+      "step": 258
+    },
+    {
+      "epoch": 1.048582995951417,
+      "grad_norm": 1.0712093890513747,
+      "learning_rate": 3.8560344040468996e-05,
+      "loss": 0.5235,
+      "step": 259
+    },
+    {
+      "epoch": 1.0526315789473684,
+      "grad_norm": 0.8266785869659954,
+      "learning_rate": 3.853920126717703e-05,
+      "loss": 0.5138,
+      "step": 260
+    },
+    {
+      "epoch": 1.05668016194332,
+      "grad_norm": 0.6098719735582414,
+      "learning_rate": 3.851791025492574e-05,
+      "loss": 0.5042,
+      "step": 261
+    },
+    {
+      "epoch": 1.0607287449392713,
+      "grad_norm": 0.725650101166445,
+      "learning_rate": 3.8496471173957515e-05,
+      "loss": 0.5102,
+      "step": 262
+    },
+    {
+      "epoch": 1.0647773279352226,
+      "grad_norm": 0.7364952122816476,
+      "learning_rate": 3.847488419569866e-05,
+      "loss": 0.5112,
+      "step": 263
+    },
+    {
+      "epoch": 1.0688259109311742,
+      "grad_norm": 0.6845389968771182,
+      "learning_rate": 3.8453149492758077e-05,
+      "loss": 0.5155,
+      "step": 264
+    },
+    {
+      "epoch": 1.0728744939271255,
+      "grad_norm": 0.6219737516978153,
+      "learning_rate": 3.8431267238925886e-05,
+      "loss": 0.5267,
+      "step": 265
+    },
+    {
+      "epoch": 1.0769230769230769,
+      "grad_norm": 0.6292372516184697,
+      "learning_rate": 3.840923760917199e-05,
+      "loss": 0.523,
+      "step": 266
+    },
+    {
+      "epoch": 1.0809716599190284,
+      "grad_norm": 0.5519585029974919,
+      "learning_rate": 3.838706077964473e-05,
+      "loss": 0.5053,
+      "step": 267
+    },
+    {
+      "epoch": 1.0850202429149798,
+      "grad_norm": 0.6335384302876944,
+      "learning_rate": 3.836473692766944e-05,
+      "loss": 0.5054,
+      "step": 268
+    },
+    {
+      "epoch": 1.0890688259109311,
+      "grad_norm": 0.6037351922254697,
+      "learning_rate": 3.834226623174705e-05,
+      "loss": 0.5083,
+      "step": 269
+    },
+    {
+      "epoch": 1.0931174089068827,
+      "grad_norm": 0.6474953416137176,
+      "learning_rate": 3.831964887155264e-05,
+      "loss": 0.5129,
+      "step": 270
+    },
+    {
+      "epoch": 1.097165991902834,
+      "grad_norm": 0.5404169466101993,
+      "learning_rate": 3.829688502793403e-05,
+      "loss": 0.5041,
+      "step": 271
+    },
+    {
+      "epoch": 1.1012145748987854,
+      "grad_norm": 0.6402335388962876,
+      "learning_rate": 3.827397488291029e-05,
+      "loss": 0.5096,
+      "step": 272
+    },
+    {
+      "epoch": 1.1052631578947367,
+      "grad_norm": 0.4482278014310702,
+      "learning_rate": 3.8250918619670343e-05,
+      "loss": 0.5159,
+      "step": 273
+    },
+    {
+      "epoch": 1.1093117408906883,
+      "grad_norm": 0.6089446713289474,
+      "learning_rate": 3.822771642257145e-05,
+      "loss": 0.5164,
+      "step": 274
+    },
+    {
+      "epoch": 1.1133603238866396,
+      "grad_norm": 0.6008828049824789,
+      "learning_rate": 3.820436847713776e-05,
+      "loss": 0.5115,
+      "step": 275
+    },
+    {
+      "epoch": 1.117408906882591,
+      "grad_norm": 0.6053686407662269,
+      "learning_rate": 3.8180874970058826e-05,
+      "loss": 0.5118,
+      "step": 276
+    },
+    {
+      "epoch": 1.1214574898785425,
+      "grad_norm": 0.48305040243117475,
+      "learning_rate": 3.8157236089188103e-05,
+      "loss": 0.5202,
+      "step": 277
+    },
+    {
+      "epoch": 1.125506072874494,
+      "grad_norm": 0.5649966093195584,
+      "learning_rate": 3.813345202354145e-05,
+      "loss": 0.5036,
+      "step": 278
+    },
+    {
+      "epoch": 1.1295546558704452,
+      "grad_norm": 0.653396795344135,
+      "learning_rate": 3.810952296329563e-05,
+      "loss": 0.5098,
+      "step": 279
+    },
+    {
+      "epoch": 1.1336032388663968,
+      "grad_norm": 0.5323962211577381,
+      "learning_rate": 3.808544909978676e-05,
+      "loss": 0.4985,
+      "step": 280
+    },
+    {
+      "epoch": 1.1376518218623481,
+      "grad_norm": 0.6333455496976804,
+      "learning_rate": 3.806123062550882e-05,
+      "loss": 0.5053,
+      "step": 281
+    },
+    {
+      "epoch": 1.1417004048582995,
+      "grad_norm": 0.5616367276555482,
+      "learning_rate": 3.803686773411208e-05,
+      "loss": 0.5098,
+      "step": 282
+    },
+    {
+      "epoch": 1.145748987854251,
+      "grad_norm": 0.5043600211165785,
+      "learning_rate": 3.801236062040158e-05,
+      "loss": 0.5025,
+      "step": 283
+    },
+    {
+      "epoch": 1.1497975708502024,
+      "grad_norm": 0.5512775546288408,
+      "learning_rate": 3.798770948033553e-05,
+      "loss": 0.5125,
+      "step": 284
+    },
+    {
+      "epoch": 1.1538461538461537,
+      "grad_norm": 0.5067541775533778,
+      "learning_rate": 3.796291451102381e-05,
+      "loss": 0.5121,
+      "step": 285
+    },
+    {
+      "epoch": 1.1578947368421053,
+      "grad_norm": 0.49519734817457844,
+      "learning_rate": 3.793797591072631e-05,
+      "loss": 0.4966,
+      "step": 286
+    },
+    {
+      "epoch": 1.1619433198380567,
+      "grad_norm": 0.4650917304396431,
+      "learning_rate": 3.791289387885144e-05,
+      "loss": 0.5047,
+      "step": 287
+    },
+    {
+      "epoch": 1.165991902834008,
+      "grad_norm": 0.458728939798487,
+      "learning_rate": 3.788766861595443e-05,
+      "loss": 0.5134,
+      "step": 288
+    },
+    {
+      "epoch": 1.1700404858299596,
+      "grad_norm": 0.4768159702840952,
+      "learning_rate": 3.7862300323735835e-05,
+      "loss": 0.5059,
+      "step": 289
+    },
+    {
+      "epoch": 1.174089068825911,
+      "grad_norm": 0.44793633297173846,
+      "learning_rate": 3.783678920503983e-05,
+      "loss": 0.518,
+      "step": 290
+    },
+    {
+      "epoch": 1.1781376518218623,
+      "grad_norm": 0.47079033268027465,
+      "learning_rate": 3.781113546385265e-05,
+      "loss": 0.5107,
+      "step": 291
+    },
+    {
+      "epoch": 1.1821862348178138,
+      "grad_norm": 0.4598221827205203,
+      "learning_rate": 3.7785339305300916e-05,
+      "loss": 0.5142,
+      "step": 292
+    },
+    {
+      "epoch": 1.1862348178137652,
+      "grad_norm": 0.44948049178219696,
+      "learning_rate": 3.7759400935650025e-05,
+      "loss": 0.5079,
+      "step": 293
+    },
+    {
+      "epoch": 1.1902834008097165,
+      "grad_norm": 0.47016802426698856,
+      "learning_rate": 3.77333205623025e-05,
+      "loss": 0.5159,
+      "step": 294
+    },
+    {
+      "epoch": 1.194331983805668,
+      "grad_norm": 0.4409473989170051,
+      "learning_rate": 3.7707098393796295e-05,
+      "loss": 0.5142,
+      "step": 295
+    },
+    {
+      "epoch": 1.1983805668016194,
+      "grad_norm": 0.4623067280665605,
+      "learning_rate": 3.7680734639803185e-05,
+      "loss": 0.5034,
+      "step": 296
+    },
+    {
+      "epoch": 1.2024291497975708,
+      "grad_norm": 0.45463658133361823,
+      "learning_rate": 3.765422951112704e-05,
+      "loss": 0.5222,
+      "step": 297
+    },
+    {
+      "epoch": 1.2064777327935223,
+      "grad_norm": 0.4774353552137865,
+      "learning_rate": 3.762758321970216e-05,
+      "loss": 0.5032,
+      "step": 298
+    },
+    {
+      "epoch": 1.2105263157894737,
+      "grad_norm": 0.40837158237921195,
+      "learning_rate": 3.7600795978591584e-05,
+      "loss": 0.5107,
+      "step": 299
+    },
+    {
+      "epoch": 1.214574898785425,
+      "grad_norm": 0.46441407108720173,
+      "learning_rate": 3.757386800198538e-05,
+      "loss": 0.516,
+      "step": 300
+    },
+    {
+      "epoch": 1.2186234817813766,
+      "grad_norm": 0.49009801733592673,
+      "learning_rate": 3.7546799505198925e-05,
+      "loss": 0.5036,
+      "step": 301
+    },
+    {
+      "epoch": 1.222672064777328,
+      "grad_norm": 0.4573767640526802,
+      "learning_rate": 3.7519590704671197e-05,
+      "loss": 0.5086,
+      "step": 302
+    },
+    {
+      "epoch": 1.2267206477732793,
+      "grad_norm": 0.4134817554137267,
+      "learning_rate": 3.749224181796305e-05,
+      "loss": 0.5134,
+      "step": 303
+    },
+    {
+      "epoch": 1.2307692307692308,
+      "grad_norm": 0.48425998763142214,
+      "learning_rate": 3.746475306375545e-05,
+      "loss": 0.5153,
+      "step": 304
+    },
+    {
+      "epoch": 1.2348178137651822,
+      "grad_norm": 0.4392108811365307,
+      "learning_rate": 3.743712466184774e-05,
+      "loss": 0.4959,
+      "step": 305
+    },
+    {
+      "epoch": 1.2388663967611335,
+      "grad_norm": 0.5255557234363541,
+      "learning_rate": 3.7409356833155885e-05,
+      "loss": 0.5035,
+      "step": 306
+    },
+    {
+      "epoch": 1.242914979757085,
+      "grad_norm": 0.4709945630710952,
+      "learning_rate": 3.73814497997107e-05,
+      "loss": 0.5137,
+      "step": 307
+    },
+    {
+      "epoch": 1.2469635627530364,
+      "grad_norm": 0.4521006904377329,
+      "learning_rate": 3.7353403784656095e-05,
+      "loss": 0.5001,
+      "step": 308
+    },
+    {
+      "epoch": 1.2510121457489878,
+      "grad_norm": 0.5402328880637371,
+      "learning_rate": 3.732521901224724e-05,
+      "loss": 0.5162,
+      "step": 309
+    },
+    {
+      "epoch": 1.2550607287449393,
+      "grad_norm": 0.569822678892258,
+      "learning_rate": 3.7296895707848844e-05,
+      "loss": 0.5065,
+      "step": 310
+    },
+    {
+      "epoch": 1.2591093117408907,
+      "grad_norm": 0.4152562186957632,
+      "learning_rate": 3.7268434097933275e-05,
+      "loss": 0.508,
+      "step": 311
+    },
+    {
+      "epoch": 1.263157894736842,
+      "grad_norm": 0.49881334726830767,
+      "learning_rate": 3.723983441007882e-05,
+      "loss": 0.4971,
+      "step": 312
+    },
+    {
+      "epoch": 1.2672064777327936,
+      "grad_norm": 0.5307764776268132,
+      "learning_rate": 3.721109687296781e-05,
+      "loss": 0.5169,
+      "step": 313
+    },
+    {
+      "epoch": 1.271255060728745,
+      "grad_norm": 0.46572774634655145,
+      "learning_rate": 3.718222171638484e-05,
+      "loss": 0.5063,
+      "step": 314
+    },
+    {
+      "epoch": 1.2753036437246963,
+      "grad_norm": 0.5174926792907492,
+      "learning_rate": 3.715320917121488e-05,
+      "loss": 0.5056,
+      "step": 315
+    },
+    {
+      "epoch": 1.2793522267206479,
+      "grad_norm": 0.557572241022666,
+      "learning_rate": 3.712405946944149e-05,
+      "loss": 0.514,
+      "step": 316
+    },
+    {
+      "epoch": 1.2834008097165992,
+      "grad_norm": 0.4812532882606843,
+      "learning_rate": 3.709477284414488e-05,
+      "loss": 0.5138,
+      "step": 317
+    },
+    {
+      "epoch": 1.2874493927125505,
+      "grad_norm": 0.500565521886519,
+      "learning_rate": 3.7065349529500136e-05,
+      "loss": 0.5192,
+      "step": 318
+    },
+    {
+      "epoch": 1.291497975708502,
+      "grad_norm": 0.5302243506022996,
+      "learning_rate": 3.7035789760775305e-05,
+      "loss": 0.5023,
+      "step": 319
+    },
+    {
+      "epoch": 1.2955465587044535,
+      "grad_norm": 0.44752765829902486,
+      "learning_rate": 3.70060937743295e-05,
+      "loss": 0.5138,
+      "step": 320
+    },
+    {
+      "epoch": 1.2995951417004048,
+      "grad_norm": 0.4302017115687141,
+      "learning_rate": 3.6976261807611045e-05,
+      "loss": 0.5142,
+      "step": 321
+    },
+    {
+      "epoch": 1.3036437246963564,
+      "grad_norm": 0.4178277695642101,
+      "learning_rate": 3.694629409915555e-05,
+      "loss": 0.5103,
+      "step": 322
+    },
+    {
+      "epoch": 1.3076923076923077,
+      "grad_norm": 0.42588133459263505,
+      "learning_rate": 3.691619088858401e-05,
+      "loss": 0.5026,
+      "step": 323
+    },
+    {
+      "epoch": 1.311740890688259,
+      "grad_norm": 0.4519838072220198,
+      "learning_rate": 3.68859524166009e-05,
+      "loss": 0.5007,
+      "step": 324
+    },
+    {
+      "epoch": 1.3157894736842106,
+      "grad_norm": 0.44606391795635064,
+      "learning_rate": 3.6855578924992226e-05,
+      "loss": 0.503,
+      "step": 325
+    },
+    {
+      "epoch": 1.319838056680162,
+      "grad_norm": 0.4686870538278423,
+      "learning_rate": 3.6825070656623626e-05,
+      "loss": 0.5027,
+      "step": 326
+    },
+    {
+      "epoch": 1.3238866396761133,
+      "grad_norm": 0.48421204695250775,
+      "learning_rate": 3.67944278554384e-05,
+      "loss": 0.5125,
+      "step": 327
+    },
+    {
+      "epoch": 1.3279352226720649,
+      "grad_norm": 0.45929121582590715,
+      "learning_rate": 3.676365076645557e-05,
+      "loss": 0.5204,
+      "step": 328
+    },
+    {
+      "epoch": 1.3319838056680162,
+      "grad_norm": 0.4747754357557623,
+      "learning_rate": 3.673273963576791e-05,
+      "loss": 0.4981,
+      "step": 329
+    },
+    {
+      "epoch": 1.3360323886639676,
+      "grad_norm": 0.5436280334523735,
+      "learning_rate": 3.670169471054001e-05,
+      "loss": 0.5176,
+      "step": 330
+    },
+    {
+      "epoch": 1.3400809716599191,
+      "grad_norm": 0.4739739787357102,
+      "learning_rate": 3.6670516239006254e-05,
+      "loss": 0.5026,
+      "step": 331
+    },
+    {
+      "epoch": 1.3441295546558705,
+      "grad_norm": 0.5416092184065678,
+      "learning_rate": 3.663920447046888e-05,
+      "loss": 0.5114,
+      "step": 332
+    },
+    {
+      "epoch": 1.3481781376518218,
+      "grad_norm": 0.4189325678903935,
+      "learning_rate": 3.6607759655295954e-05,
+      "loss": 0.5111,
+      "step": 333
+    },
+    {
+      "epoch": 1.3522267206477734,
+      "grad_norm": 0.4825727301643273,
+      "learning_rate": 3.657618204491937e-05,
+      "loss": 0.5132,
+      "step": 334
+    },
+    {
+      "epoch": 1.3562753036437247,
+      "grad_norm": 0.47142646193673216,
+      "learning_rate": 3.6544471891832875e-05,
+      "loss": 0.4927,
+      "step": 335
+    },
+    {
+      "epoch": 1.360323886639676,
+      "grad_norm": 0.42440283828422853,
+      "learning_rate": 3.6512629449589983e-05,
+      "loss": 0.5223,
+      "step": 336
+    },
+    {
+      "epoch": 1.3643724696356276,
+      "grad_norm": 0.47193857919534665,
+      "learning_rate": 3.6480654972802045e-05,
+      "loss": 0.5184,
+      "step": 337
+    },
+    {
+      "epoch": 1.368421052631579,
+      "grad_norm": 0.5307405315267112,
+      "learning_rate": 3.644854871713611e-05,
+      "loss": 0.5218,
+      "step": 338
+    },
+    {
+      "epoch": 1.3724696356275303,
+      "grad_norm": 0.49639883144049135,
+      "learning_rate": 3.6416310939312924e-05,
+      "loss": 0.5158,
+      "step": 339
+    },
+    {
+      "epoch": 1.376518218623482,
+      "grad_norm": 0.43797654810136744,
+      "learning_rate": 3.638394189710493e-05,
+      "loss": 0.5099,
+      "step": 340
+    },
+    {
+      "epoch": 1.3805668016194332,
+      "grad_norm": 0.5171951797993668,
+      "learning_rate": 3.635144184933412e-05,
+      "loss": 0.5038,
+      "step": 341
+    },
+    {
+      "epoch": 1.3846153846153846,
+      "grad_norm": 0.4276450521150851,
+      "learning_rate": 3.6318811055869994e-05,
+      "loss": 0.5181,
+      "step": 342
+    },
+    {
+      "epoch": 1.3886639676113361,
+      "grad_norm": 0.4917899149108472,
+      "learning_rate": 3.628604977762752e-05,
+      "loss": 0.5208,
+      "step": 343
+    },
+    {
+      "epoch": 1.3927125506072875,
+      "grad_norm": 0.43611742315875623,
+      "learning_rate": 3.6253158276565004e-05,
+      "loss": 0.5127,
+      "step": 344
+    },
+    {
+      "epoch": 1.3967611336032388,
+      "grad_norm": 0.49748532602779777,
+      "learning_rate": 3.6220136815682016e-05,
+      "loss": 0.5163,
+      "step": 345
+    },
+    {
+      "epoch": 1.4008097165991904,
+      "grad_norm": 0.43903535927012244,
+      "learning_rate": 3.618698565901727e-05,
+      "loss": 0.5151,
+      "step": 346
+    },
+    {
+      "epoch": 1.4048582995951417,
+      "grad_norm": 0.500827730845864,
+      "learning_rate": 3.615370507164654e-05,
+      "loss": 0.5119,
+      "step": 347
+    },
+    {
+      "epoch": 1.408906882591093,
+      "grad_norm": 0.4539942840005327,
+      "learning_rate": 3.61202953196805e-05,
+      "loss": 0.5026,
+      "step": 348
+    },
+    {
+      "epoch": 1.4129554655870447,
+      "grad_norm": 0.5164880959754536,
+      "learning_rate": 3.6086756670262654e-05,
+      "loss": 0.5069,
+      "step": 349
+    },
+    {
+      "epoch": 1.417004048582996,
+      "grad_norm": 0.5041412386254979,
+      "learning_rate": 3.6053089391567135e-05,
+      "loss": 0.4914,
+      "step": 350
+    },
+    {
+      "epoch": 1.4210526315789473,
+      "grad_norm": 0.5530276948513727,
+      "learning_rate": 3.601929375279662e-05,
+      "loss": 0.5118,
+      "step": 351
+    },
+    {
+      "epoch": 1.425101214574899,
+      "grad_norm": 0.46215458546375754,
+      "learning_rate": 3.598537002418012e-05,
+      "loss": 0.5061,
+      "step": 352
+    },
+    {
+      "epoch": 1.4291497975708503,
+      "grad_norm": 0.49854523440605847,
+      "learning_rate": 3.5951318476970885e-05,
+      "loss": 0.4902,
+      "step": 353
+    },
+    {
+      "epoch": 1.4331983805668016,
+      "grad_norm": 0.574972860828797,
+      "learning_rate": 3.591713938344416e-05,
+      "loss": 0.5126,
+      "step": 354
+    },
+    {
+      "epoch": 1.4372469635627532,
+      "grad_norm": 0.4922656944755679,
+      "learning_rate": 3.588283301689507e-05,
+      "loss": 0.5085,
+      "step": 355
+    },
+    {
+      "epoch": 1.4412955465587045,
+      "grad_norm": 0.628885646078091,
+      "learning_rate": 3.5848399651636424e-05,
+      "loss": 0.5046,
+      "step": 356
+    },
+    {
+      "epoch": 1.4453441295546559,
+      "grad_norm": 0.6023341533303207,
+      "learning_rate": 3.5813839562996473e-05,
+      "loss": 0.4985,
+      "step": 357
+    },
+    {
+      "epoch": 1.4493927125506074,
+      "grad_norm": 0.6493906079339662,
+      "learning_rate": 3.5779153027316775e-05,
+      "loss": 0.5083,
+      "step": 358
+    },
+    {
+      "epoch": 1.4534412955465588,
+      "grad_norm": 0.5300385569669067,
+      "learning_rate": 3.574434032194994e-05,
+      "loss": 0.5087,
+      "step": 359
+    },
+    {
+      "epoch": 1.45748987854251,
+      "grad_norm": 0.4575667411047454,
+      "learning_rate": 3.5709401725257436e-05,
+      "loss": 0.5156,
+      "step": 360
+    },
+    {
+      "epoch": 1.4615384615384617,
+      "grad_norm": 0.46046603886531023,
+      "learning_rate": 3.5674337516607346e-05,
+      "loss": 0.5119,
+      "step": 361
+    },
+    {
+      "epoch": 1.465587044534413,
+      "grad_norm": 0.5054428926301158,
+      "learning_rate": 3.563914797637216e-05,
+      "loss": 0.5094,
+      "step": 362
+    },
+    {
+      "epoch": 1.4696356275303644,
+      "grad_norm": 0.4293905203427863,
+      "learning_rate": 3.560383338592649e-05,
+      "loss": 0.515,
+      "step": 363
+    },
+    {
+      "epoch": 1.4736842105263157,
+      "grad_norm": 0.5698890006621762,
+      "learning_rate": 3.556839402764487e-05,
+      "loss": 0.5095,
+      "step": 364
+    },
+    {
+      "epoch": 1.4777327935222673,
+      "grad_norm": 0.4380459074460214,
+      "learning_rate": 3.553283018489946e-05,
+      "loss": 0.5015,
+      "step": 365
+    },
+    {
+      "epoch": 1.4817813765182186,
+      "grad_norm": 0.5395707148321868,
+      "learning_rate": 3.54971421420578e-05,
+      "loss": 0.5133,
+      "step": 366
+    },
+    {
+      "epoch": 1.48582995951417,
+      "grad_norm": 0.495048662820013,
+      "learning_rate": 3.5461330184480524e-05,
+      "loss": 0.5014,
+      "step": 367
+    },
+    {
+      "epoch": 1.4898785425101215,
+      "grad_norm": 0.5432970295889165,
+      "learning_rate": 3.54253945985191e-05,
+      "loss": 0.5153,
+      "step": 368
+    },
+    {
+      "epoch": 1.4939271255060729,
+      "grad_norm": 0.4850355072988799,
+      "learning_rate": 3.5389335671513524e-05,
+      "loss": 0.4988,
+      "step": 369
+    },
+    {
+      "epoch": 1.4979757085020242,
+      "grad_norm": 0.5250618015694943,
+      "learning_rate": 3.5353153691789986e-05,
+      "loss": 0.4994,
+      "step": 370
+    },
+    {
+      "epoch": 1.5020242914979756,
+      "grad_norm": 0.416000740413209,
+      "learning_rate": 3.531684894865866e-05,
+      "loss": 0.5042,
+      "step": 371
+    },
+    {
+      "epoch": 1.5060728744939271,
+      "grad_norm": 0.514398358894453,
+      "learning_rate": 3.528042173241128e-05,
+      "loss": 0.5286,
+      "step": 372
+    },
+    {
+      "epoch": 1.5101214574898787,
+      "grad_norm": 0.53756035022823,
+      "learning_rate": 3.524387233431892e-05,
+      "loss": 0.5046,
+      "step": 373
+    },
+    {
+      "epoch": 1.5141700404858298,
+      "grad_norm": 0.44427606070981,
+      "learning_rate": 3.520720104662958e-05,
+      "loss": 0.5047,
+      "step": 374
+    },
+    {
+      "epoch": 1.5182186234817814,
+      "grad_norm": 0.40937481341334186,
+      "learning_rate": 3.5170408162565904e-05,
+      "loss": 0.5158,
+      "step": 375
+    },
+    {
+      "epoch": 1.522267206477733,
+      "grad_norm": 0.5380671955038417,
+      "learning_rate": 3.513349397632281e-05,
+      "loss": 0.5013,
+      "step": 376
+    },
+    {
+      "epoch": 1.526315789473684,
+      "grad_norm": 0.4022079862767205,
+      "learning_rate": 3.5096458783065145e-05,
+      "loss": 0.5176,
+      "step": 377
+    },
+    {
+      "epoch": 1.5303643724696356,
+      "grad_norm": 0.44894271510180095,
+      "learning_rate": 3.505930287892533e-05,
+      "loss": 0.521,
+      "step": 378
+    },
+    {
+      "epoch": 1.5344129554655872,
+      "grad_norm": 0.4026566931289031,
+      "learning_rate": 3.502202656100099e-05,
+      "loss": 0.5035,
+      "step": 379
+    },
+    {
+      "epoch": 1.5384615384615383,
+      "grad_norm": 0.4433448900135455,
+      "learning_rate": 3.498463012735256e-05,
+      "loss": 0.5049,
+      "step": 380
+    },
+    {
+      "epoch": 1.54251012145749,
+      "grad_norm": 0.37469499198429956,
+      "learning_rate": 3.494711387700094e-05,
+      "loss": 0.5094,
+      "step": 381
+    },
+    {
+      "epoch": 1.5465587044534415,
+      "grad_norm": 0.4408500448154407,
+      "learning_rate": 3.490947810992505e-05,
+      "loss": 0.5176,
+      "step": 382
+    },
+    {
+      "epoch": 1.5506072874493926,
+      "grad_norm": 0.3958878481149186,
+      "learning_rate": 3.48717231270595e-05,
+      "loss": 0.5063,
+      "step": 383
+    },
+    {
+      "epoch": 1.5546558704453441,
+      "grad_norm": 0.48849648896282116,
+      "learning_rate": 3.483384923029211e-05,
+      "loss": 0.5322,
+      "step": 384
+    },
+    {
+      "epoch": 1.5587044534412957,
+      "grad_norm": 0.4379250293837225,
+      "learning_rate": 3.479585672246156e-05,
+      "loss": 0.5049,
+      "step": 385
+    },
+    {
+      "epoch": 1.5627530364372468,
+      "grad_norm": 0.43759567611198324,
+      "learning_rate": 3.4757745907354924e-05,
+      "loss": 0.498,
+      "step": 386
+    },
+    {
+      "epoch": 1.5668016194331984,
+      "grad_norm": 0.42677689991396284,
+      "learning_rate": 3.471951708970527e-05,
+      "loss": 0.5003,
+      "step": 387
+    },
+    {
+      "epoch": 1.5708502024291497,
+      "grad_norm": 0.4482777002517184,
+      "learning_rate": 3.468117057518921e-05,
+      "loss": 0.5077,
+      "step": 388
+    },
+    {
+      "epoch": 1.574898785425101,
+      "grad_norm": 0.46061323986297303,
+      "learning_rate": 3.4642706670424426e-05,
+      "loss": 0.5214,
+      "step": 389
+    },
+    {
+      "epoch": 1.5789473684210527,
+      "grad_norm": 0.3775167361922601,
+      "learning_rate": 3.460412568296731e-05,
+      "loss": 0.5147,
+      "step": 390
+    },
+    {
+      "epoch": 1.582995951417004,
+      "grad_norm": 0.41672510182060263,
+      "learning_rate": 3.456542792131039e-05,
+      "loss": 0.5038,
+      "step": 391
+    },
+    {
+      "epoch": 1.5870445344129553,
+      "grad_norm": 0.4955144298985715,
+      "learning_rate": 3.4526613694879936e-05,
+      "loss": 0.5143,
+      "step": 392
+    },
+    {
+      "epoch": 1.591093117408907,
+      "grad_norm": 0.4395849859025501,
+      "learning_rate": 3.448768331403348e-05,
+      "loss": 0.4964,
+      "step": 393
+    },
+    {
+      "epoch": 1.5951417004048583,
+      "grad_norm": 0.47444851340252103,
+      "learning_rate": 3.44486370900573e-05,
+      "loss": 0.5125,
+      "step": 394
+    },
+    {
+      "epoch": 1.5991902834008096,
+      "grad_norm": 0.4193083502239155,
+      "learning_rate": 3.440947533516396e-05,
+      "loss": 0.5007,
+      "step": 395
+    },
+    {
+      "epoch": 1.6032388663967612,
+      "grad_norm": 0.47725247319422065,
+      "learning_rate": 3.4370198362489816e-05,
+      "loss": 0.5204,
+      "step": 396
+    },
+    {
+      "epoch": 1.6072874493927125,
+      "grad_norm": 0.42692842704851125,
+      "learning_rate": 3.43308064860925e-05,
+      "loss": 0.5146,
+      "step": 397
+    },
+    {
+      "epoch": 1.6113360323886639,
+      "grad_norm": 0.453432670409512,
+      "learning_rate": 3.4291300020948393e-05,
+      "loss": 0.4923,
+      "step": 398
+    },
+    {
+      "epoch": 1.6153846153846154,
+      "grad_norm": 0.4345283733359831,
+      "learning_rate": 3.425167928295015e-05,
+      "loss": 0.5017,
+      "step": 399
+    },
+    {
+      "epoch": 1.6194331983805668,
+      "grad_norm": 0.4631633986558607,
+      "learning_rate": 3.421194458890411e-05,
+      "loss": 0.5004,
+      "step": 400
+    },
+    {
+      "epoch": 1.623481781376518,
+      "grad_norm": 0.42196799954610537,
+      "learning_rate": 3.4172096256527865e-05,
+      "loss": 0.5222,
+      "step": 401
+    },
+    {
+      "epoch": 1.6275303643724697,
+      "grad_norm": 0.41240654061677934,
+      "learning_rate": 3.4132134604447595e-05,
+      "loss": 0.513,
+      "step": 402
+    },
+    {
+      "epoch": 1.631578947368421,
+      "grad_norm": 0.43071461775026687,
+      "learning_rate": 3.409205995219561e-05,
+      "loss": 0.4981,
+      "step": 403
+    },
+    {
+      "epoch": 1.6356275303643724,
+      "grad_norm": 0.4708739792887388,
+      "learning_rate": 3.4051872620207765e-05,
+      "loss": 0.4954,
+      "step": 404
+    },
+    {
+      "epoch": 1.639676113360324,
+      "grad_norm": 0.6166866242170601,
+      "learning_rate": 3.40115729298209e-05,
+      "loss": 0.515,
+      "step": 405
+    },
+    {
+      "epoch": 1.6437246963562753,
+      "grad_norm": 0.42123430419655655,
+      "learning_rate": 3.397116120327025e-05,
+      "loss": 0.5219,
+      "step": 406
+    },
+    {
+      "epoch": 1.6477732793522266,
+      "grad_norm": 0.5093108322946034,
+      "learning_rate": 3.3930637763686916e-05,
+      "loss": 0.5085,
+      "step": 407
+    },
+    {
+      "epoch": 1.6518218623481782,
+      "grad_norm": 0.4238799374018401,
+      "learning_rate": 3.389000293509524e-05,
+      "loss": 0.5049,
+      "step": 408
+    },
+    {
+      "epoch": 1.6558704453441295,
+      "grad_norm": 0.4826982151049804,
+      "learning_rate": 3.3849257042410225e-05,
+      "loss": 0.5039,
+      "step": 409
+    },
+    {
+      "epoch": 1.6599190283400809,
+      "grad_norm": 0.3860938039561584,
+      "learning_rate": 3.380840041143494e-05,
+      "loss": 0.5033,
+      "step": 410
+    },
+    {
+      "epoch": 1.6639676113360324,
+      "grad_norm": 0.5646231445536974,
+      "learning_rate": 3.376743336885792e-05,
+      "loss": 0.5064,
+      "step": 411
+    },
+    {
+      "epoch": 1.6680161943319838,
+      "grad_norm": 0.48439848809833597,
+      "learning_rate": 3.372635624225055e-05,
+      "loss": 0.5129,
+      "step": 412
+    },
+    {
+      "epoch": 1.6720647773279351,
+      "grad_norm": 0.466294720817737,
+      "learning_rate": 3.3685169360064445e-05,
+      "loss": 0.514,
+      "step": 413
+    },
+    {
+      "epoch": 1.6761133603238867,
+      "grad_norm": 0.528120944375683,
+      "learning_rate": 3.3643873051628804e-05,
+      "loss": 0.4994,
+      "step": 414
+    },
+    {
+      "epoch": 1.680161943319838,
+      "grad_norm": 0.5816590309438979,
+      "learning_rate": 3.3602467647147814e-05,
+      "loss": 0.5137,
+      "step": 415
+    },
+    {
+      "epoch": 1.6842105263157894,
+      "grad_norm": 0.4218677552441572,
+      "learning_rate": 3.356095347769799e-05,
+      "loss": 0.4978,
+      "step": 416
+    },
+    {
+      "epoch": 1.688259109311741,
+      "grad_norm": 0.5380116686898821,
+      "learning_rate": 3.351933087522552e-05,
+      "loss": 0.5002,
+      "step": 417
+    },
+    {
+      "epoch": 1.6923076923076923,
+      "grad_norm": 0.6086775271669274,
+      "learning_rate": 3.3477600172543626e-05,
+      "loss": 0.5062,
+      "step": 418
+    },
+    {
+      "epoch": 1.6963562753036436,
+      "grad_norm": 0.4919987346910397,
+      "learning_rate": 3.3435761703329894e-05,
+      "loss": 0.5111,
+      "step": 419
+    },
+    {
+      "epoch": 1.7004048582995952,
+      "grad_norm": 0.44767754481782723,
+      "learning_rate": 3.3393815802123606e-05,
+      "loss": 0.5048,
+      "step": 420
+    },
+    {
+      "epoch": 1.7044534412955465,
+      "grad_norm": 0.44461165302419153,
+      "learning_rate": 3.335176280432308e-05,
+      "loss": 0.4992,
+      "step": 421
+    },
+    {
+      "epoch": 1.708502024291498,
+      "grad_norm": 0.41726874061591135,
+      "learning_rate": 3.330960304618295e-05,
+      "loss": 0.5061,
+      "step": 422
+    },
+    {
+      "epoch": 1.7125506072874495,
+      "grad_norm": 0.435371974003365,
+      "learning_rate": 3.326733686481153e-05,
+      "loss": 0.5025,
+      "step": 423
+    },
+    {
+      "epoch": 1.7165991902834008,
+      "grad_norm": 0.4126933311174565,
+      "learning_rate": 3.3224964598168084e-05,
+      "loss": 0.5063,
+      "step": 424
+    },
+    {
+      "epoch": 1.7206477732793521,
+      "grad_norm": 0.351187179886881,
+      "learning_rate": 3.318248658506012e-05,
+      "loss": 0.5098,
+      "step": 425
+    },
+    {
+      "epoch": 1.7246963562753037,
+      "grad_norm": 0.4181844278435788,
+      "learning_rate": 3.313990316514069e-05,
+      "loss": 0.5171,
+      "step": 426
+    },
+    {
+      "epoch": 1.728744939271255,
+      "grad_norm": 0.4070785516712774,
+      "learning_rate": 3.309721467890571e-05,
+      "loss": 0.5044,
+      "step": 427
+    },
+    {
+      "epoch": 1.7327935222672064,
+      "grad_norm": 0.34208117966468055,
+      "learning_rate": 3.305442146769115e-05,
+      "loss": 0.4985,
+      "step": 428
+    },
+    {
+      "epoch": 1.736842105263158,
+      "grad_norm": 0.4109423341443401,
+      "learning_rate": 3.30115238736704e-05,
+      "loss": 0.5124,
+      "step": 429
+    },
+    {
+      "epoch": 1.7408906882591093,
+      "grad_norm": 0.3618470740112121,
+      "learning_rate": 3.296852223985148e-05,
+      "loss": 0.4976,
+      "step": 430
+    },
+    {
+      "epoch": 1.7449392712550607,
+      "grad_norm": 0.42793602923232,
+      "learning_rate": 3.2925416910074315e-05,
+      "loss": 0.5122,
+      "step": 431
+    },
+    {
+      "epoch": 1.7489878542510122,
+      "grad_norm": 0.3919789787862844,
+      "learning_rate": 3.288220822900796e-05,
+      "loss": 0.4998,
+      "step": 432
+    },
+    {
+      "epoch": 1.7530364372469636,
+      "grad_norm": 0.4255709985627238,
+      "learning_rate": 3.283889654214788e-05,
+      "loss": 0.4839,
+      "step": 433
+    },
+    {
+      "epoch": 1.757085020242915,
+      "grad_norm": 0.38459764237458244,
+      "learning_rate": 3.279548219581317e-05,
+      "loss": 0.4947,
+      "step": 434
+    },
+    {
+      "epoch": 1.7611336032388665,
+      "grad_norm": 0.37809585270948337,
+      "learning_rate": 3.275196553714379e-05,
+      "loss": 0.4944,
+      "step": 435
+    },
+    {
+      "epoch": 1.7651821862348178,
+      "grad_norm": 0.4054209572705219,
+      "learning_rate": 3.270834691409778e-05,
+      "loss": 0.4939,
+      "step": 436
+    },
+    {
+      "epoch": 1.7692307692307692,
+      "grad_norm": 0.40525704577623306,
+      "learning_rate": 3.266462667544849e-05,
+      "loss": 0.4945,
+      "step": 437
+    },
+    {
+      "epoch": 1.7732793522267207,
+      "grad_norm": 0.41275898375101383,
+      "learning_rate": 3.262080517078178e-05,
+      "loss": 0.5046,
+      "step": 438
+    },
+    {
+      "epoch": 1.777327935222672,
+      "grad_norm": 0.40077194615461875,
+      "learning_rate": 3.257688275049323e-05,
+      "loss": 0.5192,
+      "step": 439
+    },
+    {
+      "epoch": 1.7813765182186234,
+      "grad_norm": 0.4161755018475676,
+      "learning_rate": 3.253285976578535e-05,
+      "loss": 0.5009,
+      "step": 440
+    },
+    {
+      "epoch": 1.785425101214575,
+      "grad_norm": 0.40086003638325424,
+      "learning_rate": 3.2488736568664756e-05,
+      "loss": 0.5051,
+      "step": 441
+    },
+    {
+      "epoch": 1.7894736842105263,
+      "grad_norm": 0.44017317361619035,
+      "learning_rate": 3.244451351193935e-05,
+      "loss": 0.501,
+      "step": 442
+    },
+    {
+      "epoch": 1.7935222672064777,
+      "grad_norm": 0.41035396328405727,
+      "learning_rate": 3.240019094921551e-05,
+      "loss": 0.4891,
+      "step": 443
+    },
+    {
+      "epoch": 1.7975708502024292,
+      "grad_norm": 0.3859102963823525,
+      "learning_rate": 3.235576923489527e-05,
+      "loss": 0.5048,
+      "step": 444
+    },
+    {
+      "epoch": 1.8016194331983806,
+      "grad_norm": 0.44770631429617685,
+      "learning_rate": 3.2311248724173476e-05,
+      "loss": 0.5021,
+      "step": 445
+    },
+    {
+      "epoch": 1.805668016194332,
+      "grad_norm": 0.4114938259171238,
+      "learning_rate": 3.226662977303494e-05,
+      "loss": 0.5038,
+      "step": 446
+    },
+    {
+      "epoch": 1.8097165991902835,
+      "grad_norm": 0.43424837233529723,
+      "learning_rate": 3.2221912738251626e-05,
+      "loss": 0.5152,
+      "step": 447
+    },
+    {
+      "epoch": 1.8137651821862348,
+      "grad_norm": 0.45907553789676464,
+      "learning_rate": 3.2177097977379736e-05,
+      "loss": 0.5058,
+      "step": 448
+    },
+    {
+      "epoch": 1.8178137651821862,
+      "grad_norm": 0.47924983743134264,
+      "learning_rate": 3.21321858487569e-05,
+      "loss": 0.5045,
+      "step": 449
+    },
+    {
+      "epoch": 1.8218623481781377,
+      "grad_norm": 0.48254049198575555,
+      "learning_rate": 3.208717671149931e-05,
+      "loss": 0.5057,
+      "step": 450
+    },
+    {
+      "epoch": 1.825910931174089,
+      "grad_norm": 0.4069114183780363,
+      "learning_rate": 3.204207092549882e-05,
+      "loss": 0.4886,
+      "step": 451
+    },
+    {
+      "epoch": 1.8299595141700404,
+      "grad_norm": 0.441115434731664,
+      "learning_rate": 3.19968688514201e-05,
+      "loss": 0.5006,
+      "step": 452
+    },
+    {
+      "epoch": 1.834008097165992,
+      "grad_norm": 0.3523180458750645,
+      "learning_rate": 3.195157085069774e-05,
+      "loss": 0.4964,
+      "step": 453
+    },
+    {
+      "epoch": 1.8380566801619433,
+      "grad_norm": 0.42431367499055617,
+      "learning_rate": 3.190617728553332e-05,
+      "loss": 0.5073,
+      "step": 454
+    },
+    {
+      "epoch": 1.8421052631578947,
+      "grad_norm": 0.3680837398809946,
+      "learning_rate": 3.1860688518892606e-05,
+      "loss": 0.5034,
+      "step": 455
+    },
+    {
+      "epoch": 1.8461538461538463,
+      "grad_norm": 0.43439687265387167,
+      "learning_rate": 3.181510491450254e-05,
+      "loss": 0.5055,
+      "step": 456
+    },
+    {
+      "epoch": 1.8502024291497976,
+      "grad_norm": 0.40528937779036217,
+      "learning_rate": 3.176942683684842e-05,
+      "loss": 0.5012,
+      "step": 457
+    },
+    {
+      "epoch": 1.854251012145749,
+      "grad_norm": 0.3909310234718009,
+      "learning_rate": 3.1723654651170934e-05,
+      "loss": 0.508,
+      "step": 458
+    },
+    {
+      "epoch": 1.8582995951417005,
+      "grad_norm": 0.43255111989832373,
+      "learning_rate": 3.167778872346327e-05,
+      "loss": 0.5058,
+      "step": 459
+    },
+    {
+      "epoch": 1.8623481781376519,
+      "grad_norm": 0.3896921140861326,
+      "learning_rate": 3.163182942046815e-05,
+      "loss": 0.5083,
+      "step": 460
+    },
+    {
+      "epoch": 1.8663967611336032,
+      "grad_norm": 0.4043727496655229,
+      "learning_rate": 3.158577710967494e-05,
+      "loss": 0.4964,
+      "step": 461
+    },
+    {
+      "epoch": 1.8704453441295548,
+      "grad_norm": 0.4094341969033801,
+      "learning_rate": 3.153963215931669e-05,
+      "loss": 0.4978,
+      "step": 462
+    },
+    {
+      "epoch": 1.874493927125506,
+      "grad_norm": 0.36898157235490586,
+      "learning_rate": 3.14933949383672e-05,
+      "loss": 0.5121,
+      "step": 463
+    },
+    {
+      "epoch": 1.8785425101214575,
+      "grad_norm": 0.3634384000428044,
+      "learning_rate": 3.144706581653805e-05,
+      "loss": 0.5022,
+      "step": 464
+    },
+    {
+      "epoch": 1.882591093117409,
+      "grad_norm": 0.45404963376809065,
+      "learning_rate": 3.140064516427566e-05,
+      "loss": 0.5091,
+      "step": 465
+    },
+    {
+      "epoch": 1.8866396761133604,
+      "grad_norm": 0.4082789452459126,
+      "learning_rate": 3.135413335275833e-05,
+      "loss": 0.504,
+      "step": 466
+    },
+    {
+      "epoch": 1.8906882591093117,
+      "grad_norm": 0.39114224987067897,
+      "learning_rate": 3.130753075389327e-05,
+      "loss": 0.5058,
+      "step": 467
+    },
+    {
+      "epoch": 1.8947368421052633,
+      "grad_norm": 0.3356867648123109,
+      "learning_rate": 3.1260837740313625e-05,
+      "loss": 0.5045,
+      "step": 468
+    },
+    {
+      "epoch": 1.8987854251012146,
+      "grad_norm": 0.4031993498266614,
+      "learning_rate": 3.121405468537547e-05,
+      "loss": 0.5074,
+      "step": 469
+    },
+    {
+      "epoch": 1.902834008097166,
+      "grad_norm": 0.4526037065843044,
+      "learning_rate": 3.116718196315488e-05,
+      "loss": 0.4977,
+      "step": 470
+    },
+    {
+      "epoch": 1.9068825910931175,
+      "grad_norm": 0.38600702016260635,
+      "learning_rate": 3.1120219948444894e-05,
+      "loss": 0.5,
+      "step": 471
+    },
+    {
+      "epoch": 1.9109311740890689,
+      "grad_norm": 0.484848810216074,
+      "learning_rate": 3.107316901675252e-05,
+      "loss": 0.5101,
+      "step": 472
+    },
+    {
+      "epoch": 1.9149797570850202,
+      "grad_norm": 0.3451508260022927,
+      "learning_rate": 3.1026029544295745e-05,
+      "loss": 0.4882,
+      "step": 473
+    },
+    {
+      "epoch": 1.9190283400809718,
+      "grad_norm": 0.4537293209321823,
+      "learning_rate": 3.097880190800057e-05,
+      "loss": 0.4957,
+      "step": 474
+    },
+    {
+      "epoch": 1.9230769230769231,
+      "grad_norm": 0.41740265470933335,
+      "learning_rate": 3.093148648549788e-05,
+      "loss": 0.508,
+      "step": 475
+    },
+    {
+      "epoch": 1.9271255060728745,
+      "grad_norm": 0.3798048946774691,
+      "learning_rate": 3.088408365512055e-05,
+      "loss": 0.511,
+      "step": 476
+    },
+    {
+      "epoch": 1.931174089068826,
+      "grad_norm": 0.4108912189737695,
+      "learning_rate": 3.083659379590034e-05,
+      "loss": 0.5132,
+      "step": 477
+    },
+    {
+      "epoch": 1.9352226720647774,
+      "grad_norm": 0.41953851691206373,
+      "learning_rate": 3.0789017287564913e-05,
+      "loss": 0.5048,
+      "step": 478
+    },
+    {
+      "epoch": 1.9392712550607287,
+      "grad_norm": 0.39985427425717696,
+      "learning_rate": 3.074135451053475e-05,
+      "loss": 0.4991,
+      "step": 479
+    },
+    {
+      "epoch": 1.9433198380566803,
+      "grad_norm": 0.5723671649472488,
+      "learning_rate": 3.069360584592015e-05,
+      "loss": 0.5048,
+      "step": 480
+    },
+    {
+      "epoch": 1.9473684210526314,
+      "grad_norm": 0.45136305438287383,
+      "learning_rate": 3.064577167551816e-05,
+      "loss": 0.5019,
+      "step": 481
+    },
+    {
+      "epoch": 1.951417004048583,
+      "grad_norm": 0.41468559578368674,
+      "learning_rate": 3.059785238180953e-05,
+      "loss": 0.4985,
+      "step": 482
+    },
+    {
+      "epoch": 1.9554655870445345,
+      "grad_norm": 0.4585999933005169,
+      "learning_rate": 3.054984834795566e-05,
+      "loss": 0.5157,
+      "step": 483
+    },
+    {
+      "epoch": 1.9595141700404857,
+      "grad_norm": 0.43504619941570805,
+      "learning_rate": 3.0501759957795537e-05,
+      "loss": 0.5101,
+      "step": 484
+    },
+    {
+      "epoch": 1.9635627530364372,
+      "grad_norm": 0.48512082184981964,
+      "learning_rate": 3.0453587595842627e-05,
+      "loss": 0.4973,
+      "step": 485
+    },
+    {
+      "epoch": 1.9676113360323888,
+      "grad_norm": 0.3831256590804324,
+      "learning_rate": 3.0405331647281856e-05,
+      "loss": 0.5077,
+      "step": 486
+    },
+    {
+      "epoch": 1.97165991902834,
+      "grad_norm": 0.5607665009707417,
+      "learning_rate": 3.0356992497966508e-05,
+      "loss": 0.4942,
+      "step": 487
+    },
+    {
+      "epoch": 1.9757085020242915,
+      "grad_norm": 0.45342194276144493,
+      "learning_rate": 3.0308570534415137e-05,
+      "loss": 0.4925,
+      "step": 488
+    },
+    {
+      "epoch": 1.979757085020243,
+      "grad_norm": 0.45504663293806485,
+      "learning_rate": 3.0260066143808467e-05,
+      "loss": 0.5083,
+      "step": 489
+    },
+    {
+      "epoch": 1.9838056680161942,
+      "grad_norm": 0.4056695991841582,
+      "learning_rate": 3.021147971398632e-05,
+      "loss": 0.488,
+      "step": 490
+    },
+    {
+      "epoch": 1.9878542510121457,
+      "grad_norm": 0.4740097055347121,
+      "learning_rate": 3.0162811633444478e-05,
+      "loss": 0.5095,
+      "step": 491
+    },
+    {
+      "epoch": 1.9919028340080973,
+      "grad_norm": 0.42291197328521124,
+      "learning_rate": 3.0114062291331635e-05,
+      "loss": 0.506,
+      "step": 492
+    },
+    {
+      "epoch": 1.9959514170040484,
+      "grad_norm": 0.4353319354979632,
+      "learning_rate": 3.0065232077446217e-05,
+      "loss": 0.5048,
+      "step": 493
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.44114154018227586,
+      "learning_rate": 3.001632138223332e-05,
+      "loss": 0.505,
+      "step": 494
+    },
+    {
+      "epoch": 2.0040485829959516,
+      "grad_norm": 0.5395569695238726,
+      "learning_rate": 2.9967330596781565e-05,
+      "loss": 0.436,
+      "step": 495
+    },
+    {
+      "epoch": 2.0080971659919027,
+      "grad_norm": 0.4480265120799233,
+      "learning_rate": 2.9918260112819952e-05,
+      "loss": 0.4241,
+      "step": 496
+    },
+    {
+      "epoch": 2.0121457489878543,
+      "grad_norm": 0.7178863937157588,
+      "learning_rate": 2.9869110322714778e-05,
+      "loss": 0.4171,
+      "step": 497
+    },
+    {
+      "epoch": 2.016194331983806,
+      "grad_norm": 0.523157783216145,
+      "learning_rate": 2.9819881619466447e-05,
+      "loss": 0.4078,
+      "step": 498
+    },
+    {
+      "epoch": 2.020242914979757,
+      "grad_norm": 0.4454160067438631,
+      "learning_rate": 2.9770574396706364e-05,
+      "loss": 0.4205,
+      "step": 499
+    },
+    {
+      "epoch": 2.0242914979757085,
+      "grad_norm": 0.460163701192616,
+      "learning_rate": 2.972118904869376e-05,
+      "loss": 0.4185,
+      "step": 500
+    },
+    {
+      "epoch": 2.02834008097166,
+      "grad_norm": 0.511257394601272,
+      "learning_rate": 2.9671725970312563e-05,
+      "loss": 0.4123,
+      "step": 501
+    },
+    {
+      "epoch": 2.032388663967611,
+      "grad_norm": 0.46366271529453446,
+      "learning_rate": 2.9622185557068238e-05,
+      "loss": 0.4153,
+      "step": 502
+    },
+    {
+      "epoch": 2.0364372469635628,
+      "grad_norm": 0.4979273584979808,
+      "learning_rate": 2.95725682050846e-05,
+      "loss": 0.4189,
+      "step": 503
+    },
+    {
+      "epoch": 2.0404858299595143,
+      "grad_norm": 0.4766110704131582,
+      "learning_rate": 2.9522874311100684e-05,
+      "loss": 0.4257,
+      "step": 504
+    },
+    {
+      "epoch": 2.0445344129554655,
+      "grad_norm": 0.48621866545547454,
+      "learning_rate": 2.9473104272467526e-05,
+      "loss": 0.4183,
+      "step": 505
+    },
+    {
+      "epoch": 2.048582995951417,
+      "grad_norm": 0.4813583507475465,
+      "learning_rate": 2.942325848714505e-05,
+      "loss": 0.4169,
+      "step": 506
+    },
+    {
+      "epoch": 2.0526315789473686,
+      "grad_norm": 0.5329206662354763,
+      "learning_rate": 2.9373337353698804e-05,
+      "loss": 0.4192,
+      "step": 507
+    },
+    {
+      "epoch": 2.0566801619433197,
+      "grad_norm": 0.5061240997935965,
+      "learning_rate": 2.932334127129686e-05,
+      "loss": 0.4118,
+      "step": 508
+    },
+    {
+      "epoch": 2.0607287449392713,
+      "grad_norm": 0.6049370165725839,
+      "learning_rate": 2.9273270639706547e-05,
+      "loss": 0.4223,
+      "step": 509
+    },
+    {
+      "epoch": 2.064777327935223,
+      "grad_norm": 0.4066810500230157,
+      "learning_rate": 2.922312585929131e-05,
+      "loss": 0.4233,
+      "step": 510
+    },
+    {
+      "epoch": 2.068825910931174,
+      "grad_norm": 0.4708387273927638,
+      "learning_rate": 2.9172907331007478e-05,
+      "loss": 0.4141,
+      "step": 511
+    },
+    {
+      "epoch": 2.0728744939271255,
+      "grad_norm": 0.40935498449716823,
+      "learning_rate": 2.9122615456401053e-05,
+      "loss": 0.4113,
+      "step": 512
+    },
+    {
+      "epoch": 2.076923076923077,
+      "grad_norm": 0.482306006797164,
+      "learning_rate": 2.9072250637604535e-05,
+      "loss": 0.4138,
+      "step": 513
+    },
+    {
+      "epoch": 2.080971659919028,
+      "grad_norm": 0.44153251257354553,
+      "learning_rate": 2.9021813277333667e-05,
+      "loss": 0.4127,
+      "step": 514
+    },
+    {
+      "epoch": 2.08502024291498,
+      "grad_norm": 0.4226278861280671,
+      "learning_rate": 2.8971303778884244e-05,
+      "loss": 0.4221,
+      "step": 515
+    },
+    {
+      "epoch": 2.0890688259109313,
+      "grad_norm": 0.4129149530976653,
+      "learning_rate": 2.8920722546128863e-05,
+      "loss": 0.4259,
+      "step": 516
+    },
+    {
+      "epoch": 2.0931174089068825,
+      "grad_norm": 0.45305710784921854,
+      "learning_rate": 2.887006998351372e-05,
+      "loss": 0.4258,
+      "step": 517
+    },
+    {
+      "epoch": 2.097165991902834,
+      "grad_norm": 0.39283436649160797,
+      "learning_rate": 2.881934649605535e-05,
+      "loss": 0.4198,
+      "step": 518
+    },
+    {
+      "epoch": 2.1012145748987856,
+      "grad_norm": 0.4377935601892014,
+      "learning_rate": 2.8768552489337408e-05,
+      "loss": 0.4111,
+      "step": 519
+    },
+    {
+      "epoch": 2.1052631578947367,
+      "grad_norm": 0.40442340837769836,
+      "learning_rate": 2.8717688369507425e-05,
+      "loss": 0.4224,
+      "step": 520
+    },
+    {
+      "epoch": 2.1093117408906883,
+      "grad_norm": 0.41396608980431326,
+      "learning_rate": 2.8666754543273535e-05,
+      "loss": 0.4235,
+      "step": 521
+    },
+    {
+      "epoch": 2.11336032388664,
+      "grad_norm": 0.4164634655241282,
+      "learning_rate": 2.8615751417901273e-05,
+      "loss": 0.4153,
+      "step": 522
+    },
+    {
+      "epoch": 2.117408906882591,
+      "grad_norm": 0.3743376448806877,
+      "learning_rate": 2.856467940121025e-05,
+      "loss": 0.4123,
+      "step": 523
+    },
+    {
+      "epoch": 2.1214574898785425,
+      "grad_norm": 0.4343837815497891,
+      "learning_rate": 2.8513538901570964e-05,
+      "loss": 0.4198,
+      "step": 524
+    },
+    {
+      "epoch": 2.125506072874494,
+      "grad_norm": 0.4309885388974649,
+      "learning_rate": 2.846233032790149e-05,
+      "loss": 0.4155,
+      "step": 525
+    },
+    {
+      "epoch": 2.1295546558704452,
+      "grad_norm": 0.43869254694819415,
+      "learning_rate": 2.84110540896642e-05,
+      "loss": 0.4249,
+      "step": 526
+    },
+    {
+      "epoch": 2.133603238866397,
+      "grad_norm": 0.4194464491652385,
+      "learning_rate": 2.8359710596862555e-05,
+      "loss": 0.4297,
+      "step": 527
+    },
+    {
+      "epoch": 2.1376518218623484,
+      "grad_norm": 0.43510056571739625,
+      "learning_rate": 2.8308300260037734e-05,
+      "loss": 0.4161,
+      "step": 528
+    },
+    {
+      "epoch": 2.1417004048582995,
+      "grad_norm": 0.3535443168727852,
+      "learning_rate": 2.8256823490265424e-05,
+      "loss": 0.4149,
+      "step": 529
+    },
+    {
+      "epoch": 2.145748987854251,
+      "grad_norm": 0.5223260125948621,
+      "learning_rate": 2.8205280699152515e-05,
+      "loss": 0.4185,
+      "step": 530
+    },
+    {
+      "epoch": 2.1497975708502026,
+      "grad_norm": 0.3716038143209856,
+      "learning_rate": 2.8153672298833777e-05,
+      "loss": 0.4171,
+      "step": 531
+    },
+    {
+      "epoch": 2.1538461538461537,
+      "grad_norm": 0.43494584000944686,
+      "learning_rate": 2.8101998701968615e-05,
+      "loss": 0.4152,
+      "step": 532
+    },
+    {
+      "epoch": 2.1578947368421053,
+      "grad_norm": 0.4411982860707066,
+      "learning_rate": 2.805026032173773e-05,
+      "loss": 0.4222,
+      "step": 533
+    },
+    {
+      "epoch": 2.161943319838057,
+      "grad_norm": 0.3550054622298736,
+      "learning_rate": 2.7998457571839832e-05,
+      "loss": 0.419,
+      "step": 534
+    },
+    {
+      "epoch": 2.165991902834008,
+      "grad_norm": 0.37537486853468727,
+      "learning_rate": 2.7946590866488322e-05,
+      "loss": 0.4234,
+      "step": 535
+    },
+    {
+      "epoch": 2.1700404858299596,
+      "grad_norm": 0.35190832179102416,
+      "learning_rate": 2.7894660620408e-05,
+      "loss": 0.4196,
+      "step": 536
+    },
+    {
+      "epoch": 2.174089068825911,
+      "grad_norm": 0.33472748264245294,
+      "learning_rate": 2.784266724883173e-05,
+      "loss": 0.4088,
+      "step": 537
+    },
+    {
+      "epoch": 2.1781376518218623,
+      "grad_norm": 0.3756869610844889,
+      "learning_rate": 2.7790611167497128e-05,
+      "loss": 0.4063,
+      "step": 538
+    },
+    {
+      "epoch": 2.182186234817814,
+      "grad_norm": 0.3386047200235205,
+      "learning_rate": 2.7738492792643224e-05,
+      "loss": 0.4087,
+      "step": 539
+    },
+    {
+      "epoch": 2.1862348178137654,
+      "grad_norm": 0.3813403191773099,
+      "learning_rate": 2.768631254100716e-05,
+      "loss": 0.4346,
+      "step": 540
+    },
+    {
+      "epoch": 2.1902834008097165,
+      "grad_norm": 0.3713992503642015,
+      "learning_rate": 2.7634070829820836e-05,
+      "loss": 0.4131,
+      "step": 541
+    },
+    {
+      "epoch": 2.194331983805668,
+      "grad_norm": 0.3946691134391921,
+      "learning_rate": 2.7581768076807592e-05,
+      "loss": 0.4196,
+      "step": 542
+    },
+    {
+      "epoch": 2.1983805668016196,
+      "grad_norm": 0.4397579870760827,
+      "learning_rate": 2.7529404700178832e-05,
+      "loss": 0.4197,
+      "step": 543
+    },
+    {
+      "epoch": 2.2024291497975708,
+      "grad_norm": 0.34258436876067955,
+      "learning_rate": 2.7476981118630735e-05,
+      "loss": 0.421,
+      "step": 544
+    },
+    {
+      "epoch": 2.2064777327935223,
+      "grad_norm": 0.46578742669402,
+      "learning_rate": 2.742449775134085e-05,
+      "loss": 0.4267,
+      "step": 545
+    },
+    {
+      "epoch": 2.2105263157894735,
+      "grad_norm": 0.46099910395316385,
+      "learning_rate": 2.7371955017964788e-05,
+      "loss": 0.4187,
+      "step": 546
+    },
+    {
+      "epoch": 2.214574898785425,
+      "grad_norm": 0.3331090826213787,
+      "learning_rate": 2.7319353338632845e-05,
+      "loss": 0.4203,
+      "step": 547
+    },
+    {
+      "epoch": 2.2186234817813766,
+      "grad_norm": 0.3715926828182525,
+      "learning_rate": 2.7266693133946645e-05,
+      "loss": 0.4324,
+      "step": 548
+    },
+    {
+      "epoch": 2.2226720647773277,
+      "grad_norm": 0.3552635306837622,
+      "learning_rate": 2.7213974824975785e-05,
+      "loss": 0.4177,
+      "step": 549
+    },
+    {
+      "epoch": 2.2267206477732793,
+      "grad_norm": 0.36416471383202625,
+      "learning_rate": 2.7161198833254464e-05,
+      "loss": 0.4205,
+      "step": 550
+    },
+    {
+      "epoch": 2.230769230769231,
+      "grad_norm": 0.4244284446677761,
+      "learning_rate": 2.7108365580778095e-05,
+      "loss": 0.411,
+      "step": 551
+    },
+    {
+      "epoch": 2.234817813765182,
+      "grad_norm": 0.4334464682523426,
+      "learning_rate": 2.705547548999996e-05,
+      "loss": 0.4203,
+      "step": 552
+    },
+    {
+      "epoch": 2.2388663967611335,
+      "grad_norm": 0.34878830261241345,
+      "learning_rate": 2.7002528983827817e-05,
+      "loss": 0.4239,
+      "step": 553
+    },
+    {
+      "epoch": 2.242914979757085,
+      "grad_norm": 0.3893953115120094,
+      "learning_rate": 2.6949526485620517e-05,
+      "loss": 0.415,
+      "step": 554
+    },
+    {
+      "epoch": 2.246963562753036,
+      "grad_norm": 0.34682093984253015,
+      "learning_rate": 2.6896468419184623e-05,
+      "loss": 0.4243,
+      "step": 555
+    },
+    {
+      "epoch": 2.251012145748988,
+      "grad_norm": 0.41150072782587577,
+      "learning_rate": 2.6843355208771024e-05,
+      "loss": 0.419,
+      "step": 556
+    },
+    {
+      "epoch": 2.2550607287449393,
+      "grad_norm": 0.36194337613679833,
+      "learning_rate": 2.6790187279071536e-05,
+      "loss": 0.4085,
+      "step": 557
+    },
+    {
+      "epoch": 2.2591093117408905,
+      "grad_norm": 0.36799326045889064,
+      "learning_rate": 2.67369650552155e-05,
+      "loss": 0.426,
+      "step": 558
+    },
+    {
+      "epoch": 2.263157894736842,
+      "grad_norm": 0.38760474968921194,
+      "learning_rate": 2.6683688962766407e-05,
+      "loss": 0.4298,
+      "step": 559
+    },
+    {
+      "epoch": 2.2672064777327936,
+      "grad_norm": 0.3515967886963262,
+      "learning_rate": 2.6630359427718468e-05,
+      "loss": 0.4119,
+      "step": 560
+    },
+    {
+      "epoch": 2.2712550607287447,
+      "grad_norm": 0.42439228572919,
+      "learning_rate": 2.6576976876493225e-05,
+      "loss": 0.4393,
+      "step": 561
+    },
+    {
+      "epoch": 2.2753036437246963,
+      "grad_norm": 0.3636347823670654,
+      "learning_rate": 2.652354173593614e-05,
+      "loss": 0.4322,
+      "step": 562
+    },
+    {
+      "epoch": 2.279352226720648,
+      "grad_norm": 0.3762542840998603,
+      "learning_rate": 2.647005443331317e-05,
+      "loss": 0.4398,
+      "step": 563
+    },
+    {
+      "epoch": 2.283400809716599,
+      "grad_norm": 0.39882218154111465,
+      "learning_rate": 2.6416515396307354e-05,
+      "loss": 0.4325,
+      "step": 564
+    },
+    {
+      "epoch": 2.2874493927125505,
+      "grad_norm": 0.37577699634200284,
+      "learning_rate": 2.636292505301542e-05,
+      "loss": 0.4205,
+      "step": 565
+    },
+    {
+      "epoch": 2.291497975708502,
+      "grad_norm": 0.35572007336077976,
+      "learning_rate": 2.630928383194431e-05,
+      "loss": 0.4282,
+      "step": 566
+    },
+    {
+      "epoch": 2.2955465587044532,
+      "grad_norm": 0.40489154175049274,
+      "learning_rate": 2.625559216200781e-05,
+      "loss": 0.4104,
+      "step": 567
+    },
+    {
+      "epoch": 2.299595141700405,
+      "grad_norm": 0.36667667799858766,
+      "learning_rate": 2.620185047252308e-05,
+      "loss": 0.4136,
+      "step": 568
+    },
+    {
+      "epoch": 2.3036437246963564,
+      "grad_norm": 0.3769055297250673,
+      "learning_rate": 2.614805919320723e-05,
+      "loss": 0.414,
+      "step": 569
+    },
+    {
+      "epoch": 2.3076923076923075,
+      "grad_norm": 0.35354437147537315,
+      "learning_rate": 2.6094218754173895e-05,
+      "loss": 0.4305,
+      "step": 570
+    },
+    {
+      "epoch": 2.311740890688259,
+      "grad_norm": 0.3631010238224682,
+      "learning_rate": 2.60403295859298e-05,
+      "loss": 0.4275,
+      "step": 571
+    },
+    {
+      "epoch": 2.3157894736842106,
+      "grad_norm": 0.3949889379979816,
+      "learning_rate": 2.5986392119371297e-05,
+      "loss": 0.4274,
+      "step": 572
+    },
+    {
+      "epoch": 2.3198380566801617,
+      "grad_norm": 0.360018191834738,
+      "learning_rate": 2.5932406785780928e-05,
+      "loss": 0.422,
+      "step": 573
+    },
+    {
+      "epoch": 2.3238866396761133,
+      "grad_norm": 0.41408835556620743,
+      "learning_rate": 2.5878374016823987e-05,
+      "loss": 0.425,
+      "step": 574
+    },
+    {
+      "epoch": 2.327935222672065,
+      "grad_norm": 0.3578572521829385,
+      "learning_rate": 2.5824294244545045e-05,
+      "loss": 0.416,
+      "step": 575
+    },
+    {
+      "epoch": 2.331983805668016,
+      "grad_norm": 0.3757712684154036,
+      "learning_rate": 2.577016790136453e-05,
+      "loss": 0.4176,
+      "step": 576
+    },
+    {
+      "epoch": 2.3360323886639676,
+      "grad_norm": 0.3781207146751753,
+      "learning_rate": 2.5715995420075225e-05,
+      "loss": 0.4231,
+      "step": 577
+    },
+    {
+      "epoch": 2.340080971659919,
+      "grad_norm": 0.40560344791214825,
+      "learning_rate": 2.5661777233838864e-05,
+      "loss": 0.4188,
+      "step": 578
+    },
+    {
+      "epoch": 2.3441295546558703,
+      "grad_norm": 0.39316471341695663,
+      "learning_rate": 2.5607513776182616e-05,
+      "loss": 0.4124,
+      "step": 579
+    },
+    {
+      "epoch": 2.348178137651822,
+      "grad_norm": 0.34353321698700545,
+      "learning_rate": 2.5553205480995633e-05,
+      "loss": 0.4292,
+      "step": 580
+    },
+    {
+      "epoch": 2.3522267206477734,
+      "grad_norm": 0.4125085330304987,
+      "learning_rate": 2.5498852782525604e-05,
+      "loss": 0.424,
+      "step": 581
+    },
+    {
+      "epoch": 2.3562753036437245,
+      "grad_norm": 0.39778149976943916,
+      "learning_rate": 2.5444456115375252e-05,
+      "loss": 0.4217,
+      "step": 582
+    },
+    {
+      "epoch": 2.360323886639676,
+      "grad_norm": 0.3850486016502432,
+      "learning_rate": 2.539001591449889e-05,
+      "loss": 0.4227,
+      "step": 583
+    },
+    {
+      "epoch": 2.3643724696356276,
+      "grad_norm": 0.4393234765968069,
+      "learning_rate": 2.53355326151989e-05,
+      "loss": 0.4138,
+      "step": 584
+    },
+    {
+      "epoch": 2.3684210526315788,
+      "grad_norm": 0.39040798779889757,
+      "learning_rate": 2.5281006653122288e-05,
+      "loss": 0.4375,
+      "step": 585
+    },
+    {
+      "epoch": 2.3724696356275303,
+      "grad_norm": 0.44457092168456713,
+      "learning_rate": 2.5226438464257217e-05,
+      "loss": 0.4237,
+      "step": 586
+    },
+    {
+      "epoch": 2.376518218623482,
+      "grad_norm": 0.34414383752633515,
+      "learning_rate": 2.5171828484929447e-05,
+      "loss": 0.42,
+      "step": 587
+    },
+    {
+      "epoch": 2.380566801619433,
+      "grad_norm": 0.36282059980807196,
+      "learning_rate": 2.5117177151798934e-05,
+      "loss": 0.4287,
+      "step": 588
+    },
+    {
+      "epoch": 2.3846153846153846,
+      "grad_norm": 0.35119446513191055,
+      "learning_rate": 2.506248490185627e-05,
+      "loss": 0.4371,
+      "step": 589
+    },
+    {
+      "epoch": 2.388663967611336,
+      "grad_norm": 0.3573207516306423,
+      "learning_rate": 2.500775217241923e-05,
+      "loss": 0.4159,
+      "step": 590
+    },
+    {
+      "epoch": 2.3927125506072873,
+      "grad_norm": 0.34055508306874904,
+      "learning_rate": 2.4952979401129266e-05,
+      "loss": 0.4198,
+      "step": 591
+    },
+    {
+      "epoch": 2.396761133603239,
+      "grad_norm": 0.37065452775569013,
+      "learning_rate": 2.4898167025948e-05,
+      "loss": 0.4259,
+      "step": 592
+    },
+    {
+      "epoch": 2.4008097165991904,
+      "grad_norm": 0.3821920175547532,
+      "learning_rate": 2.4843315485153703e-05,
+      "loss": 0.4223,
+      "step": 593
+    },
+    {
+      "epoch": 2.4048582995951415,
+      "grad_norm": 0.32842651057259215,
+      "learning_rate": 2.4788425217337856e-05,
+      "loss": 0.4251,
+      "step": 594
+    },
+    {
+      "epoch": 2.408906882591093,
+      "grad_norm": 0.3581897536378419,
+      "learning_rate": 2.4733496661401572e-05,
+      "loss": 0.4181,
+      "step": 595
+    },
+    {
+      "epoch": 2.4129554655870447,
+      "grad_norm": 0.35557301467724955,
+      "learning_rate": 2.4678530256552114e-05,
+      "loss": 0.4202,
+      "step": 596
+    },
+    {
+      "epoch": 2.417004048582996,
+      "grad_norm": 0.34258971201257704,
+      "learning_rate": 2.46235264422994e-05,
+      "loss": 0.439,
+      "step": 597
+    },
+    {
+      "epoch": 2.4210526315789473,
+      "grad_norm": 0.3815763605490339,
+      "learning_rate": 2.4568485658452452e-05,
+      "loss": 0.4138,
+      "step": 598
+    },
+    {
+      "epoch": 2.425101214574899,
+      "grad_norm": 0.33383681840166063,
+      "learning_rate": 2.451340834511592e-05,
+      "loss": 0.4286,
+      "step": 599
+    },
+    {
+      "epoch": 2.42914979757085,
+      "grad_norm": 0.3785531358827489,
+      "learning_rate": 2.4458294942686528e-05,
+      "loss": 0.4167,
+      "step": 600
+    },
+    {
+      "epoch": 2.4331983805668016,
+      "grad_norm": 0.36021123730401766,
+      "learning_rate": 2.440314589184958e-05,
+      "loss": 0.422,
+      "step": 601
+    },
+    {
+      "epoch": 2.437246963562753,
+      "grad_norm": 0.3615384922680494,
+      "learning_rate": 2.434796163357541e-05,
+      "loss": 0.4278,
+      "step": 602
+    },
+    {
+      "epoch": 2.4412955465587043,
+      "grad_norm": 0.4034454513594229,
+      "learning_rate": 2.429274260911588e-05,
+      "loss": 0.4226,
+      "step": 603
+    },
+    {
+      "epoch": 2.445344129554656,
+      "grad_norm": 0.3660016961780756,
+      "learning_rate": 2.423748926000084e-05,
+      "loss": 0.4183,
+      "step": 604
+    },
+    {
+      "epoch": 2.4493927125506074,
+      "grad_norm": 0.37321817332768087,
+      "learning_rate": 2.418220202803459e-05,
+      "loss": 0.4398,
+      "step": 605
+    },
+    {
+      "epoch": 2.4534412955465585,
+      "grad_norm": 0.3885363435275948,
+      "learning_rate": 2.412688135529237e-05,
+      "loss": 0.4101,
+      "step": 606
+    },
+    {
+      "epoch": 2.45748987854251,
+      "grad_norm": 0.39708496501181717,
+      "learning_rate": 2.4071527684116807e-05,
+      "loss": 0.4063,
+      "step": 607
+    },
+    {
+      "epoch": 2.4615384615384617,
+      "grad_norm": 0.36471625027219007,
+      "learning_rate": 2.4016141457114372e-05,
+      "loss": 0.414,
+      "step": 608
+    },
+    {
+      "epoch": 2.465587044534413,
+      "grad_norm": 0.42849613063847514,
+      "learning_rate": 2.396072311715187e-05,
+      "loss": 0.4176,
+      "step": 609
+    },
+    {
+      "epoch": 2.4696356275303644,
+      "grad_norm": 0.34330637949903936,
+      "learning_rate": 2.3905273107352863e-05,
+      "loss": 0.4158,
+      "step": 610
+    },
+    {
+      "epoch": 2.473684210526316,
+      "grad_norm": 0.37592920530354035,
+      "learning_rate": 2.3849791871094157e-05,
+      "loss": 0.4241,
+      "step": 611
+    },
+    {
+      "epoch": 2.477732793522267,
+      "grad_norm": 0.3511550029029552,
+      "learning_rate": 2.379427985200223e-05,
+      "loss": 0.4289,
+      "step": 612
+    },
+    {
+      "epoch": 2.4817813765182186,
+      "grad_norm": 0.3588799256666628,
+      "learning_rate": 2.373873749394972e-05,
+      "loss": 0.4206,
+      "step": 613
+    },
+    {
+      "epoch": 2.48582995951417,
+      "grad_norm": 0.33660980946147956,
+      "learning_rate": 2.3683165241051836e-05,
+      "loss": 0.4074,
+      "step": 614
+    },
+    {
+      "epoch": 2.4898785425101213,
+      "grad_norm": 0.40242776456756285,
+      "learning_rate": 2.3627563537662823e-05,
+      "loss": 0.4178,
+      "step": 615
+    },
+    {
+      "epoch": 2.493927125506073,
+      "grad_norm": 0.3254972670009655,
+      "learning_rate": 2.3571932828372433e-05,
+      "loss": 0.4161,
+      "step": 616
+    },
+    {
+      "epoch": 2.4979757085020244,
+      "grad_norm": 0.3693117038908051,
+      "learning_rate": 2.351627355800233e-05,
+      "loss": 0.4245,
+      "step": 617
+    },
+    {
+      "epoch": 2.5020242914979756,
+      "grad_norm": 0.34358661692750114,
+      "learning_rate": 2.346058617160256e-05,
+      "loss": 0.4179,
+      "step": 618
+    },
+    {
+      "epoch": 2.506072874493927,
+      "grad_norm": 0.3855885183019052,
+      "learning_rate": 2.3404871114447976e-05,
+      "loss": 0.4215,
+      "step": 619
+    },
+    {
+      "epoch": 2.5101214574898787,
+      "grad_norm": 0.4363345262361219,
+      "learning_rate": 2.334912883203469e-05,
+      "loss": 0.4209,
+      "step": 620
+    },
+    {
+      "epoch": 2.51417004048583,
+      "grad_norm": 0.3906724338444202,
+      "learning_rate": 2.3293359770076505e-05,
+      "loss": 0.4195,
+      "step": 621
+    },
+    {
+      "epoch": 2.5182186234817814,
+      "grad_norm": 0.4037772583822802,
+      "learning_rate": 2.323756437450136e-05,
+      "loss": 0.4225,
+      "step": 622
+    },
+    {
+      "epoch": 2.522267206477733,
+      "grad_norm": 0.38091771883749836,
+      "learning_rate": 2.3181743091447754e-05,
+      "loss": 0.4156,
+      "step": 623
+    },
+    {
+      "epoch": 2.526315789473684,
+      "grad_norm": 0.3463517408546957,
+      "learning_rate": 2.3125896367261164e-05,
+      "loss": 0.4147,
+      "step": 624
+    },
+    {
+      "epoch": 2.5303643724696356,
+      "grad_norm": 0.423398809991408,
+      "learning_rate": 2.307002464849052e-05,
+      "loss": 0.4114,
+      "step": 625
+    },
+    {
+      "epoch": 2.534412955465587,
+      "grad_norm": 0.3530358887329171,
+      "learning_rate": 2.30141283818846e-05,
+      "loss": 0.4314,
+      "step": 626
+    },
+    {
+      "epoch": 2.5384615384615383,
+      "grad_norm": 0.3714671525649094,
+      "learning_rate": 2.2958208014388462e-05,
+      "loss": 0.4202,
+      "step": 627
+    },
+    {
+      "epoch": 2.54251012145749,
+      "grad_norm": 0.4565895489981728,
+      "learning_rate": 2.2902263993139857e-05,
+      "loss": 0.4328,
+      "step": 628
+    },
+    {
+      "epoch": 2.5465587044534415,
+      "grad_norm": 0.3452182879453706,
+      "learning_rate": 2.2846296765465708e-05,
+      "loss": 0.4182,
+      "step": 629
+    },
+    {
+      "epoch": 2.5506072874493926,
+      "grad_norm": 0.44182383406555376,
+      "learning_rate": 2.2790306778878465e-05,
+      "loss": 0.4163,
+      "step": 630
+    },
+    {
+      "epoch": 2.554655870445344,
+      "grad_norm": 0.37095120870396797,
+      "learning_rate": 2.273429448107257e-05,
+      "loss": 0.4257,
+      "step": 631
+    },
+    {
+      "epoch": 2.5587044534412957,
+      "grad_norm": 0.4134866330870738,
+      "learning_rate": 2.2678260319920868e-05,
+      "loss": 0.417,
+      "step": 632
+    },
+    {
+      "epoch": 2.562753036437247,
+      "grad_norm": 0.3906674198707889,
+      "learning_rate": 2.262220474347102e-05,
+      "loss": 0.4133,
+      "step": 633
+    },
+    {
+      "epoch": 2.5668016194331984,
+      "grad_norm": 0.38216273616644114,
+      "learning_rate": 2.2566128199941918e-05,
+      "loss": 0.4301,
+      "step": 634
+    },
+    {
+      "epoch": 2.57085020242915,
+      "grad_norm": 0.39280919999051567,
+      "learning_rate": 2.2510031137720104e-05,
+      "loss": 0.4292,
+      "step": 635
+    },
+    {
+      "epoch": 2.574898785425101,
+      "grad_norm": 0.38889377987481183,
+      "learning_rate": 2.2453914005356206e-05,
+      "loss": 0.4224,
+      "step": 636
+    },
+    {
+      "epoch": 2.5789473684210527,
+      "grad_norm": 0.4275096376435638,
+      "learning_rate": 2.2397777251561302e-05,
+      "loss": 0.4376,
+      "step": 637
+    },
+    {
+      "epoch": 2.582995951417004,
+      "grad_norm": 0.3785690144992792,
+      "learning_rate": 2.2341621325203393e-05,
+      "loss": 0.4267,
+      "step": 638
+    },
+    {
+      "epoch": 2.5870445344129553,
+      "grad_norm": 0.4472308191570152,
+      "learning_rate": 2.2285446675303764e-05,
+      "loss": 0.4132,
+      "step": 639
+    },
+    {
+      "epoch": 2.591093117408907,
+      "grad_norm": 0.37516045143419824,
+      "learning_rate": 2.2229253751033423e-05,
+      "loss": 0.4306,
+      "step": 640
+    },
+    {
+      "epoch": 2.5951417004048585,
+      "grad_norm": 0.4316328222267528,
+      "learning_rate": 2.2173043001709492e-05,
+      "loss": 0.4165,
+      "step": 641
+    },
+    {
+      "epoch": 2.5991902834008096,
+      "grad_norm": 0.35729930076735666,
+      "learning_rate": 2.2116814876791627e-05,
+      "loss": 0.4091,
+      "step": 642
+    },
+    {
+      "epoch": 2.603238866396761,
+      "grad_norm": 0.44309428896817765,
+      "learning_rate": 2.2060569825878413e-05,
+      "loss": 0.4167,
+      "step": 643
+    },
+    {
+      "epoch": 2.6072874493927127,
+      "grad_norm": 0.3662392973630227,
+      "learning_rate": 2.2004308298703777e-05,
+      "loss": 0.4104,
+      "step": 644
+    },
+    {
+      "epoch": 2.611336032388664,
+      "grad_norm": 0.3828899729855955,
+      "learning_rate": 2.1948030745133396e-05,
+      "loss": 0.4194,
+      "step": 645
+    },
+    {
+      "epoch": 2.6153846153846154,
+      "grad_norm": 0.3856439658052398,
+      "learning_rate": 2.1891737615161083e-05,
+      "loss": 0.4256,
+      "step": 646
+    },
+    {
+      "epoch": 2.619433198380567,
+      "grad_norm": 0.34841691762707944,
+      "learning_rate": 2.1835429358905218e-05,
+      "loss": 0.4138,
+      "step": 647
+    },
+    {
+      "epoch": 2.623481781376518,
+      "grad_norm": 0.36678434644175534,
+      "learning_rate": 2.17791064266051e-05,
+      "loss": 0.4089,
+      "step": 648
+    },
+    {
+      "epoch": 2.6275303643724697,
+      "grad_norm": 0.33304211432675246,
+      "learning_rate": 2.172276926861741e-05,
+      "loss": 0.4363,
+      "step": 649
+    },
+    {
+      "epoch": 2.6315789473684212,
+      "grad_norm": 0.3908056633055117,
+      "learning_rate": 2.166641833541255e-05,
+      "loss": 0.4246,
+      "step": 650
+    },
+    {
+      "epoch": 2.6356275303643724,
+      "grad_norm": 0.3339366620606824,
+      "learning_rate": 2.161005407757109e-05,
+      "loss": 0.4263,
+      "step": 651
+    },
+    {
+      "epoch": 2.639676113360324,
+      "grad_norm": 0.3542295100172441,
+      "learning_rate": 2.155367694578013e-05,
+      "loss": 0.43,
+      "step": 652
+    },
+    {
+      "epoch": 2.6437246963562755,
+      "grad_norm": 0.37019225025258246,
+      "learning_rate": 2.1497287390829712e-05,
+      "loss": 0.4356,
+      "step": 653
+    },
+    {
+      "epoch": 2.6477732793522266,
+      "grad_norm": 0.4229832101939057,
+      "learning_rate": 2.1440885863609208e-05,
+      "loss": 0.4158,
+      "step": 654
+    },
+    {
+      "epoch": 2.651821862348178,
+      "grad_norm": 0.36470104956106375,
+      "learning_rate": 2.1384472815103737e-05,
+      "loss": 0.4186,
+      "step": 655
+    },
+    {
+      "epoch": 2.6558704453441297,
+      "grad_norm": 0.46093000178248755,
+      "learning_rate": 2.1328048696390534e-05,
+      "loss": 0.4246,
+      "step": 656
+    },
+    {
+      "epoch": 2.659919028340081,
+      "grad_norm": 0.3928374437501149,
+      "learning_rate": 2.1271613958635342e-05,
+      "loss": 0.4208,
+      "step": 657
+    },
+    {
+      "epoch": 2.6639676113360324,
+      "grad_norm": 0.5042423290844023,
+      "learning_rate": 2.1215169053088825e-05,
+      "loss": 0.4264,
+      "step": 658
+    },
+    {
+      "epoch": 2.668016194331984,
+      "grad_norm": 0.3917139521465958,
+      "learning_rate": 2.1158714431082937e-05,
+      "loss": 0.4259,
+      "step": 659
+    },
+    {
+      "epoch": 2.672064777327935,
+      "grad_norm": 0.46788527741159597,
+      "learning_rate": 2.110225054402733e-05,
+      "loss": 0.4268,
+      "step": 660
+    },
+    {
+      "epoch": 2.6761133603238867,
+      "grad_norm": 0.36633232202229704,
+      "learning_rate": 2.1045777843405747e-05,
+      "loss": 0.4277,
+      "step": 661
+    },
+    {
+      "epoch": 2.6801619433198383,
+      "grad_norm": 0.4151595534640669,
+      "learning_rate": 2.09892967807724e-05,
+      "loss": 0.4259,
+      "step": 662
+    },
+    {
+      "epoch": 2.6842105263157894,
+      "grad_norm": 0.3565008255244946,
+      "learning_rate": 2.093280780774835e-05,
+      "loss": 0.4338,
+      "step": 663
+    },
+    {
+      "epoch": 2.688259109311741,
+      "grad_norm": 0.3890943645922177,
+      "learning_rate": 2.087631137601793e-05,
+      "loss": 0.4154,
+      "step": 664
+    },
+    {
+      "epoch": 2.6923076923076925,
+      "grad_norm": 0.37502393691955654,
+      "learning_rate": 2.0819807937325104e-05,
+      "loss": 0.4237,
+      "step": 665
+    },
+    {
+      "epoch": 2.6963562753036436,
+      "grad_norm": 0.4025767018791826,
+      "learning_rate": 2.0763297943469847e-05,
+      "loss": 0.4246,
+      "step": 666
+    },
+    {
+      "epoch": 2.700404858299595,
+      "grad_norm": 0.38664686763503164,
+      "learning_rate": 2.070678184630458e-05,
+      "loss": 0.4102,
+      "step": 667
+    },
+    {
+      "epoch": 2.7044534412955468,
+      "grad_norm": 0.34597945690115045,
+      "learning_rate": 2.06502600977305e-05,
+      "loss": 0.4215,
+      "step": 668
+    },
+    {
+      "epoch": 2.708502024291498,
+      "grad_norm": 0.36652219507195916,
+      "learning_rate": 2.0593733149694008e-05,
+      "loss": 0.4201,
+      "step": 669
+    },
+    {
+      "epoch": 2.7125506072874495,
+      "grad_norm": 0.35632040110540064,
+      "learning_rate": 2.0537201454183074e-05,
+      "loss": 0.4124,
+      "step": 670
+    },
+    {
+      "epoch": 2.716599190283401,
+      "grad_norm": 0.36419284809428104,
+      "learning_rate": 2.0480665463223626e-05,
+      "loss": 0.4243,
+      "step": 671
+    },
+    {
+      "epoch": 2.720647773279352,
+      "grad_norm": 0.3508896799103833,
+      "learning_rate": 2.042412562887594e-05,
+      "loss": 0.4266,
+      "step": 672
+    },
+    {
+      "epoch": 2.7246963562753037,
+      "grad_norm": 0.329784010919007,
+      "learning_rate": 2.0367582403231033e-05,
+      "loss": 0.4079,
+      "step": 673
+    },
+    {
+      "epoch": 2.7287449392712553,
+      "grad_norm": 0.3367768846149369,
+      "learning_rate": 2.0311036238407023e-05,
+      "loss": 0.4161,
+      "step": 674
+    },
+    {
+      "epoch": 2.7327935222672064,
+      "grad_norm": 0.3797101248134474,
+      "learning_rate": 2.0254487586545534e-05,
+      "loss": 0.4194,
+      "step": 675
+    },
+    {
+      "epoch": 2.736842105263158,
+      "grad_norm": 0.39013334667483474,
+      "learning_rate": 2.0197936899808088e-05,
+      "loss": 0.4207,
+      "step": 676
+    },
+    {
+      "epoch": 2.7408906882591095,
+      "grad_norm": 0.33003374575736305,
+      "learning_rate": 2.0141384630372466e-05,
+      "loss": 0.4219,
+      "step": 677
+    },
+    {
+      "epoch": 2.7449392712550607,
+      "grad_norm": 0.3513927133010393,
+      "learning_rate": 2.0084831230429106e-05,
+      "loss": 0.4231,
+      "step": 678
+    },
+    {
+      "epoch": 2.748987854251012,
+      "grad_norm": 0.322739382048087,
+      "learning_rate": 2.0028277152177492e-05,
+      "loss": 0.4208,
+      "step": 679
+    },
+    {
+      "epoch": 2.753036437246964,
+      "grad_norm": 0.32606774839910047,
+      "learning_rate": 1.9971722847822518e-05,
+      "loss": 0.4316,
+      "step": 680
+    },
+    {
+      "epoch": 2.757085020242915,
+      "grad_norm": 0.40733049990998055,
+      "learning_rate": 1.99151687695709e-05,
+      "loss": 0.4216,
+      "step": 681
+    },
+    {
+      "epoch": 2.7611336032388665,
+      "grad_norm": 0.31071851574678633,
+      "learning_rate": 1.9858615369627537e-05,
+      "loss": 0.4362,
+      "step": 682
+    },
+    {
+      "epoch": 2.765182186234818,
+      "grad_norm": 0.31814568962168344,
+      "learning_rate": 1.9802063100191916e-05,
+      "loss": 0.4259,
+      "step": 683
+    },
+    {
+      "epoch": 2.769230769230769,
+      "grad_norm": 0.30228125005895623,
+      "learning_rate": 1.974551241345447e-05,
+      "loss": 0.4211,
+      "step": 684
+    },
+    {
+      "epoch": 2.7732793522267207,
+      "grad_norm": 0.33792029624969716,
+      "learning_rate": 1.9688963761592984e-05,
+      "loss": 0.4169,
+      "step": 685
+    },
+    {
+      "epoch": 2.7773279352226723,
+      "grad_norm": 0.31262825777957765,
+      "learning_rate": 1.963241759676897e-05,
+      "loss": 0.4221,
+      "step": 686
+    },
+    {
+      "epoch": 2.7813765182186234,
+      "grad_norm": 0.3183513459794375,
+      "learning_rate": 1.9575874371124066e-05,
+      "loss": 0.4237,
+      "step": 687
+    },
+    {
+      "epoch": 2.785425101214575,
+      "grad_norm": 0.3303695221238289,
+      "learning_rate": 1.951933453677638e-05,
+      "loss": 0.4312,
+      "step": 688
+    },
+    {
+      "epoch": 2.7894736842105265,
+      "grad_norm": 0.294978144630511,
+      "learning_rate": 1.9462798545816932e-05,
+      "loss": 0.4085,
+      "step": 689
+    },
+    {
+      "epoch": 2.7935222672064777,
+      "grad_norm": 0.3099603965335112,
+      "learning_rate": 1.9406266850305996e-05,
+      "loss": 0.4193,
+      "step": 690
+    },
+    {
+      "epoch": 2.7975708502024292,
+      "grad_norm": 0.33903194597427766,
+      "learning_rate": 1.9349739902269503e-05,
+      "loss": 0.4319,
+      "step": 691
+    },
+    {
+      "epoch": 2.801619433198381,
+      "grad_norm": 0.34008745626736375,
+      "learning_rate": 1.9293218153695425e-05,
+      "loss": 0.4209,
+      "step": 692
+    },
+    {
+      "epoch": 2.805668016194332,
+      "grad_norm": 0.32450229242748707,
+      "learning_rate": 1.9236702056530153e-05,
+      "loss": 0.4213,
+      "step": 693
+    },
+    {
+      "epoch": 2.8097165991902835,
+      "grad_norm": 0.3398432634086294,
+      "learning_rate": 1.9180192062674907e-05,
+      "loss": 0.4215,
+      "step": 694
+    },
+    {
+      "epoch": 2.813765182186235,
+      "grad_norm": 0.3252230385933103,
+      "learning_rate": 1.9123688623982076e-05,
+      "loss": 0.4224,
+      "step": 695
+    },
+    {
+      "epoch": 2.817813765182186,
+      "grad_norm": 0.3055511852720172,
+      "learning_rate": 1.906719219225165e-05,
+      "loss": 0.4217,
+      "step": 696
+    },
+    {
+      "epoch": 2.8218623481781377,
+      "grad_norm": 0.34337007709799316,
+      "learning_rate": 1.9010703219227603e-05,
+      "loss": 0.4294,
+      "step": 697
+    },
+    {
+      "epoch": 2.8259109311740893,
+      "grad_norm": 0.31669284682777565,
+      "learning_rate": 1.8954222156594256e-05,
+      "loss": 0.4268,
+      "step": 698
+    },
+    {
+      "epoch": 2.8299595141700404,
+      "grad_norm": 0.3427893703085702,
+      "learning_rate": 1.8897749455972673e-05,
+      "loss": 0.428,
+      "step": 699
+    },
+    {
+      "epoch": 2.834008097165992,
+      "grad_norm": 0.30153695890586757,
+      "learning_rate": 1.884128556891707e-05,
+      "loss": 0.4212,
+      "step": 700
+    },
+    {
+      "epoch": 2.8380566801619436,
+      "grad_norm": 0.356532535745721,
+      "learning_rate": 1.878483094691119e-05,
+      "loss": 0.4204,
+      "step": 701
+    },
+    {
+      "epoch": 2.8421052631578947,
+      "grad_norm": 0.33178888719705946,
+      "learning_rate": 1.8728386041364664e-05,
+      "loss": 0.4241,
+      "step": 702
+    },
+    {
+      "epoch": 2.8461538461538463,
+      "grad_norm": 0.35696840446918376,
+      "learning_rate": 1.867195130360947e-05,
+      "loss": 0.4256,
+      "step": 703
+    },
+    {
+      "epoch": 2.850202429149798,
+      "grad_norm": 0.37666020196636224,
+      "learning_rate": 1.8615527184896263e-05,
+      "loss": 0.4343,
+      "step": 704
+    },
+    {
+      "epoch": 2.854251012145749,
+      "grad_norm": 0.2982291611533658,
+      "learning_rate": 1.8559114136390795e-05,
+      "loss": 0.4251,
+      "step": 705
+    },
+    {
+      "epoch": 2.8582995951417005,
+      "grad_norm": 0.3817387624842967,
+      "learning_rate": 1.8502712609170298e-05,
+      "loss": 0.4173,
+      "step": 706
+    },
+    {
+      "epoch": 2.862348178137652,
+      "grad_norm": 0.3096234988007026,
+      "learning_rate": 1.8446323054219876e-05,
+      "loss": 0.4239,
+      "step": 707
+    },
+    {
+      "epoch": 2.866396761133603,
+      "grad_norm": 0.3338887251660209,
+      "learning_rate": 1.838994592242891e-05,
+      "loss": 0.4209,
+      "step": 708
+    },
+    {
+      "epoch": 2.8704453441295548,
+      "grad_norm": 0.368857723464722,
+      "learning_rate": 1.8333581664587453e-05,
+      "loss": 0.4234,
+      "step": 709
+    },
+    {
+      "epoch": 2.8744939271255063,
+      "grad_norm": 0.33361698135287593,
+      "learning_rate": 1.8277230731382593e-05,
+      "loss": 0.4257,
+      "step": 710
+    },
+    {
+      "epoch": 2.8785425101214575,
+      "grad_norm": 0.3389029989545042,
+      "learning_rate": 1.82208935733949e-05,
+      "loss": 0.422,
+      "step": 711
+    },
+    {
+      "epoch": 2.882591093117409,
+      "grad_norm": 0.31728873345722614,
+      "learning_rate": 1.8164570641094793e-05,
+      "loss": 0.428,
+      "step": 712
+    },
+    {
+      "epoch": 2.8866396761133606,
+      "grad_norm": 0.36308849530037207,
+      "learning_rate": 1.810826238483892e-05,
+      "loss": 0.4046,
+      "step": 713
+    },
+    {
+      "epoch": 2.8906882591093117,
+      "grad_norm": 0.3116940677197923,
+      "learning_rate": 1.8051969254866608e-05,
+      "loss": 0.4118,
+      "step": 714
+    },
+    {
+      "epoch": 2.8947368421052633,
+      "grad_norm": 0.3251586114610725,
+      "learning_rate": 1.7995691701296226e-05,
+      "loss": 0.4225,
+      "step": 715
+    },
+    {
+      "epoch": 2.898785425101215,
+      "grad_norm": 0.3561211080128548,
+      "learning_rate": 1.7939430174121593e-05,
+      "loss": 0.4256,
+      "step": 716
+    },
+    {
+      "epoch": 2.902834008097166,
+      "grad_norm": 0.3119575762680143,
+      "learning_rate": 1.788318512320838e-05,
+      "loss": 0.4278,
+      "step": 717
+    },
+    {
+      "epoch": 2.9068825910931175,
+      "grad_norm": 0.34282321825931666,
+      "learning_rate": 1.782695699829051e-05,
+      "loss": 0.3998,
+      "step": 718
+    },
+    {
+      "epoch": 2.910931174089069,
+      "grad_norm": 0.32955818613015714,
+      "learning_rate": 1.7770746248966587e-05,
+      "loss": 0.4175,
+      "step": 719
+    },
+    {
+      "epoch": 2.91497975708502,
+      "grad_norm": 0.32459239180619953,
+      "learning_rate": 1.7714553324696243e-05,
+      "loss": 0.4201,
+      "step": 720
+    },
+    {
+      "epoch": 2.919028340080972,
+      "grad_norm": 0.38702194359199327,
+      "learning_rate": 1.7658378674796614e-05,
+      "loss": 0.42,
+      "step": 721
+    },
+    {
+      "epoch": 2.9230769230769234,
+      "grad_norm": 0.3272957562750429,
+      "learning_rate": 1.76022227484387e-05,
+      "loss": 0.4241,
+      "step": 722
+    },
+    {
+      "epoch": 2.9271255060728745,
+      "grad_norm": 0.3668051825092378,
+      "learning_rate": 1.7546085994643807e-05,
+      "loss": 0.4188,
+      "step": 723
+    },
+    {
+      "epoch": 2.931174089068826,
+      "grad_norm": 0.32071480198331725,
+      "learning_rate": 1.7489968862279902e-05,
+      "loss": 0.4263,
+      "step": 724
+    },
+    {
+      "epoch": 2.9352226720647776,
+      "grad_norm": 0.36417056720484264,
+      "learning_rate": 1.743387180005809e-05,
+      "loss": 0.4111,
+      "step": 725
+    },
+    {
+      "epoch": 2.9392712550607287,
+      "grad_norm": 0.3496128626002752,
+      "learning_rate": 1.737779525652899e-05,
+      "loss": 0.426,
+      "step": 726
+    },
+    {
+      "epoch": 2.9433198380566803,
+      "grad_norm": 0.33528505046146495,
+      "learning_rate": 1.732173968007914e-05,
+      "loss": 0.4247,
+      "step": 727
+    },
+    {
+      "epoch": 2.9473684210526314,
+      "grad_norm": 0.3284008892750232,
+      "learning_rate": 1.7265705518927437e-05,
+      "loss": 0.4283,
+      "step": 728
+    },
+    {
+      "epoch": 2.951417004048583,
+      "grad_norm": 0.3098727860313402,
+      "learning_rate": 1.7209693221121542e-05,
+      "loss": 0.4185,
+      "step": 729
+    },
+    {
+      "epoch": 2.9554655870445345,
+      "grad_norm": 0.30011532709114946,
+      "learning_rate": 1.7153703234534302e-05,
+      "loss": 0.4206,
+      "step": 730
+    },
+    {
+      "epoch": 2.9595141700404857,
+      "grad_norm": 0.30631169380594747,
+      "learning_rate": 1.709773600686015e-05,
+      "loss": 0.4223,
+      "step": 731
+    },
+    {
+      "epoch": 2.9635627530364372,
+      "grad_norm": 0.3239424707684689,
+      "learning_rate": 1.7041791985611548e-05,
+      "loss": 0.4351,
+      "step": 732
+    },
+    {
+      "epoch": 2.967611336032389,
+      "grad_norm": 0.30869608265694654,
+      "learning_rate": 1.69858716181154e-05,
+      "loss": 0.424,
+      "step": 733
+    },
+    {
+      "epoch": 2.97165991902834,
+      "grad_norm": 0.3267496894600305,
+      "learning_rate": 1.692997535150948e-05,
+      "loss": 0.4053,
+      "step": 734
+    },
+    {
+      "epoch": 2.9757085020242915,
+      "grad_norm": 0.3045523857135247,
+      "learning_rate": 1.687410363273884e-05,
+      "loss": 0.4114,
+      "step": 735
+    },
+    {
+      "epoch": 2.979757085020243,
+      "grad_norm": 0.3252090974144984,
+      "learning_rate": 1.6818256908552257e-05,
+      "loss": 0.424,
+      "step": 736
+    },
+    {
+      "epoch": 2.983805668016194,
+      "grad_norm": 0.36514494566627587,
+      "learning_rate": 1.6762435625498646e-05,
+      "loss": 0.4224,
+      "step": 737
+    },
+    {
+      "epoch": 2.9878542510121457,
+      "grad_norm": 0.33310298406491473,
+      "learning_rate": 1.67066402299235e-05,
+      "loss": 0.41,
+      "step": 738
+    },
+    {
+      "epoch": 2.9919028340080973,
+      "grad_norm": 0.3203535110938392,
+      "learning_rate": 1.6650871167965313e-05,
+      "loss": 0.4394,
+      "step": 739
+    },
+    {
+      "epoch": 2.9959514170040484,
+      "grad_norm": 0.3964362431291779,
+      "learning_rate": 1.6595128885552028e-05,
+      "loss": 0.4083,
+      "step": 740
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 0.32330397481223355,
+      "learning_rate": 1.653941382839745e-05,
+      "loss": 0.4085,
+      "step": 741
+    },
+    {
+      "epoch": 3.0040485829959516,
+      "grad_norm": 0.5862140247926768,
+      "learning_rate": 1.6483726441997673e-05,
+      "loss": 0.3352,
+      "step": 742
+    },
+    {
+      "epoch": 3.0080971659919027,
+      "grad_norm": 0.3975936991755836,
+      "learning_rate": 1.642806717162757e-05,
+      "loss": 0.3353,
+      "step": 743
+    },
+    {
+      "epoch": 3.0121457489878543,
+      "grad_norm": 0.7028391117415878,
+      "learning_rate": 1.637243646233718e-05,
+      "loss": 0.339,
+      "step": 744
+    },
+    {
+      "epoch": 3.016194331983806,
+      "grad_norm": 0.4640248111225813,
+      "learning_rate": 1.6316834758948174e-05,
+      "loss": 0.3366,
+      "step": 745
+    },
+    {
+      "epoch": 3.020242914979757,
+      "grad_norm": 0.5180603871700086,
+      "learning_rate": 1.6261262506050282e-05,
+      "loss": 0.3444,
+      "step": 746
+    },
+    {
+      "epoch": 3.0242914979757085,
+      "grad_norm": 0.4506843641114074,
+      "learning_rate": 1.620572014799777e-05,
+      "loss": 0.3223,
+      "step": 747
+    },
+    {
+      "epoch": 3.02834008097166,
+      "grad_norm": 0.4515020170872718,
+      "learning_rate": 1.6150208128905857e-05,
+      "loss": 0.3214,
+      "step": 748
+    },
+    {
+      "epoch": 3.032388663967611,
+      "grad_norm": 0.4183147332942801,
+      "learning_rate": 1.6094726892647147e-05,
+      "loss": 0.3165,
+      "step": 749
+    },
+    {
+      "epoch": 3.0364372469635628,
+      "grad_norm": 0.46121224409497624,
+      "learning_rate": 1.6039276882848135e-05,
+      "loss": 0.3264,
+      "step": 750
+    },
+    {
+      "epoch": 3.0404858299595143,
+      "grad_norm": 0.5030665706476984,
+      "learning_rate": 1.5983858542885635e-05,
+      "loss": 0.3294,
+      "step": 751
+    },
+    {
+      "epoch": 3.0445344129554655,
+      "grad_norm": 0.3936162330275694,
+      "learning_rate": 1.5928472315883203e-05,
+      "loss": 0.3289,
+      "step": 752
+    },
+    {
+      "epoch": 3.048582995951417,
+      "grad_norm": 0.5168815566536247,
+      "learning_rate": 1.5873118644707633e-05,
+      "loss": 0.3143,
+      "step": 753
+    },
+    {
+      "epoch": 3.0526315789473686,
+      "grad_norm": 0.4178908433148446,
+      "learning_rate": 1.5817797971965413e-05,
+      "loss": 0.3314,
+      "step": 754
+    },
+    {
+      "epoch": 3.0566801619433197,
+      "grad_norm": 0.3706905806494907,
+      "learning_rate": 1.576251073999917e-05,
+      "loss": 0.3312,
+      "step": 755
+    },
+    {
+      "epoch": 3.0607287449392713,
+      "grad_norm": 0.39834912338529377,
+      "learning_rate": 1.5707257390884126e-05,
+      "loss": 0.3272,
+      "step": 756
+    },
+    {
+      "epoch": 3.064777327935223,
+      "grad_norm": 0.4194909897139284,
+      "learning_rate": 1.5652038366424595e-05,
+      "loss": 0.3272,
+      "step": 757
+    },
+    {
+      "epoch": 3.068825910931174,
+      "grad_norm": 0.3857812796577152,
+      "learning_rate": 1.5596854108150424e-05,
+      "loss": 0.3304,
+      "step": 758
+    },
+    {
+      "epoch": 3.0728744939271255,
+      "grad_norm": 0.38847617453183697,
+      "learning_rate": 1.5541705057313476e-05,
+      "loss": 0.3383,
+      "step": 759
+    },
+    {
+      "epoch": 3.076923076923077,
+      "grad_norm": 0.3997642998147756,
+      "learning_rate": 1.5486591654884086e-05,
+      "loss": 0.339,
+      "step": 760
+    },
+    {
+      "epoch": 3.080971659919028,
+      "grad_norm": 0.39473897488237364,
+      "learning_rate": 1.543151434154755e-05,
+      "loss": 0.3253,
+      "step": 761
+    },
+    {
+      "epoch": 3.08502024291498,
+      "grad_norm": 0.42282223724837786,
+      "learning_rate": 1.5376473557700615e-05,
+      "loss": 0.3149,
+      "step": 762
+    },
+    {
+      "epoch": 3.0890688259109313,
+      "grad_norm": 0.37942816313786376,
+      "learning_rate": 1.5321469743447893e-05,
+      "loss": 0.3494,
+      "step": 763
+    },
+    {
+      "epoch": 3.0931174089068825,
+      "grad_norm": 0.40522309490483527,
+      "learning_rate": 1.5266503338598434e-05,
+      "loss": 0.3377,
+      "step": 764
+    },
+    {
+      "epoch": 3.097165991902834,
+      "grad_norm": 0.3855880160338621,
+      "learning_rate": 1.5211574782662149e-05,
+      "loss": 0.3353,
+      "step": 765
+    },
+    {
+      "epoch": 3.1012145748987856,
+      "grad_norm": 0.3306407862957468,
+      "learning_rate": 1.5156684514846307e-05,
+      "loss": 0.334,
+      "step": 766
+    },
+    {
+      "epoch": 3.1052631578947367,
+      "grad_norm": 0.41888402157985083,
+      "learning_rate": 1.5101832974052012e-05,
+      "loss": 0.3271,
+      "step": 767
+    },
+    {
+      "epoch": 3.1093117408906883,
+      "grad_norm": 0.3733247448172879,
+      "learning_rate": 1.5047020598870737e-05,
+      "loss": 0.333,
+      "step": 768
+    },
+    {
+      "epoch": 3.11336032388664,
+      "grad_norm": 0.34127378426927435,
+      "learning_rate": 1.4992247827580778e-05,
+      "loss": 0.341,
+      "step": 769
+    },
+    {
+      "epoch": 3.117408906882591,
+      "grad_norm": 0.3670545420403164,
+      "learning_rate": 1.493751509814374e-05,
+      "loss": 0.3262,
+      "step": 770
+    },
+    {
+      "epoch": 3.1214574898785425,
+      "grad_norm": 0.37435072174489875,
+      "learning_rate": 1.4882822848201073e-05,
+      "loss": 0.3342,
+      "step": 771
+    },
+    {
+      "epoch": 3.125506072874494,
+      "grad_norm": 0.34552005262199753,
+      "learning_rate": 1.4828171515070553e-05,
+      "loss": 0.333,
+      "step": 772
+    },
+    {
+      "epoch": 3.1295546558704452,
+      "grad_norm": 0.3516640555098388,
+      "learning_rate": 1.4773561535742793e-05,
+      "loss": 0.3126,
+      "step": 773
+    },
+    {
+      "epoch": 3.133603238866397,
+      "grad_norm": 0.34793981930530693,
+      "learning_rate": 1.4718993346877715e-05,
+      "loss": 0.3279,
+      "step": 774
+    },
+    {
+      "epoch": 3.1376518218623484,
+      "grad_norm": 0.35171486334329854,
+      "learning_rate": 1.466446738480111e-05,
+      "loss": 0.3364,
+      "step": 775
+    },
+    {
+      "epoch": 3.1417004048582995,
+      "grad_norm": 0.3246101708069989,
+      "learning_rate": 1.460998408550112e-05,
+      "loss": 0.3238,
+      "step": 776
+    },
+    {
+      "epoch": 3.145748987854251,
+      "grad_norm": 0.3618088041230471,
+      "learning_rate": 1.4555543884624751e-05,
+      "loss": 0.3357,
+      "step": 777
+    },
+    {
+      "epoch": 3.1497975708502026,
+      "grad_norm": 0.35053747790000755,
+      "learning_rate": 1.4501147217474402e-05,
+      "loss": 0.3163,
+      "step": 778
+    },
+    {
+      "epoch": 3.1538461538461537,
+      "grad_norm": 0.3223808876593083,
+      "learning_rate": 1.444679451900437e-05,
+      "loss": 0.3372,
+      "step": 779
+    },
+    {
+      "epoch": 3.1578947368421053,
+      "grad_norm": 0.34891550595159093,
+      "learning_rate": 1.4392486223817397e-05,
+      "loss": 0.3298,
+      "step": 780
+    },
+    {
+      "epoch": 3.161943319838057,
+      "grad_norm": 0.3450471525027687,
+      "learning_rate": 1.4338222766161143e-05,
+      "loss": 0.3338,
+      "step": 781
+    },
+    {
+      "epoch": 3.165991902834008,
+      "grad_norm": 0.32690063629384275,
+      "learning_rate": 1.4284004579924777e-05,
+      "loss": 0.3259,
+      "step": 782
+    },
+    {
+      "epoch": 3.1700404858299596,
+      "grad_norm": 0.34420194952264915,
+      "learning_rate": 1.4229832098635479e-05,
+      "loss": 0.3505,
+      "step": 783
+    },
+    {
+      "epoch": 3.174089068825911,
+      "grad_norm": 0.3345433869963605,
+      "learning_rate": 1.4175705755454963e-05,
+      "loss": 0.3337,
+      "step": 784
+    },
+    {
+      "epoch": 3.1781376518218623,
+      "grad_norm": 0.35408068008001964,
+      "learning_rate": 1.412162598317602e-05,
+      "loss": 0.3255,
+      "step": 785
+    },
+    {
+      "epoch": 3.182186234817814,
+      "grad_norm": 0.32709180704154633,
+      "learning_rate": 1.406759321421907e-05,
+      "loss": 0.3371,
+      "step": 786
+    },
+    {
+      "epoch": 3.1862348178137654,
+      "grad_norm": 0.32675455971291684,
+      "learning_rate": 1.401360788062871e-05,
+      "loss": 0.3316,
+      "step": 787
+    },
+    {
+      "epoch": 3.1902834008097165,
+      "grad_norm": 0.3237570153191568,
+      "learning_rate": 1.3959670414070204e-05,
+      "loss": 0.3198,
+      "step": 788
+    },
+    {
+      "epoch": 3.194331983805668,
+      "grad_norm": 0.3476109372248779,
+      "learning_rate": 1.3905781245826108e-05,
+      "loss": 0.3267,
+      "step": 789
+    },
+    {
+      "epoch": 3.1983805668016196,
+      "grad_norm": 0.31098977728079874,
+      "learning_rate": 1.3851940806792778e-05,
+      "loss": 0.3412,
+      "step": 790
+    },
+    {
+      "epoch": 3.2024291497975708,
+      "grad_norm": 0.379843201776324,
+      "learning_rate": 1.379814952747693e-05,
+      "loss": 0.3409,
+      "step": 791
+    },
+    {
+      "epoch": 3.2064777327935223,
+      "grad_norm": 0.31975566144481965,
+      "learning_rate": 1.3744407837992193e-05,
+      "loss": 0.3286,
+      "step": 792
+    },
+    {
+      "epoch": 3.2105263157894735,
+      "grad_norm": 0.30901169100893844,
+      "learning_rate": 1.3690716168055692e-05,
+      "loss": 0.326,
+      "step": 793
+    },
+    {
+      "epoch": 3.214574898785425,
+      "grad_norm": 0.37294227508827243,
+      "learning_rate": 1.363707494698459e-05,
+      "loss": 0.3337,
+      "step": 794
+    },
+    {
+      "epoch": 3.2186234817813766,
+      "grad_norm": 0.32015449482435165,
+      "learning_rate": 1.358348460369265e-05,
+      "loss": 0.3289,
+      "step": 795
+    },
+    {
+      "epoch": 3.2226720647773277,
+      "grad_norm": 0.3277037316761255,
+      "learning_rate": 1.352994556668684e-05,
+      "loss": 0.3304,
+      "step": 796
+    },
+    {
+      "epoch": 3.2267206477732793,
+      "grad_norm": 0.3326103875066167,
+      "learning_rate": 1.3476458264063865e-05,
+      "loss": 0.3305,
+      "step": 797
+    },
+    {
+      "epoch": 3.230769230769231,
+      "grad_norm": 0.3307200249011326,
+      "learning_rate": 1.342302312350678e-05,
+      "loss": 0.3265,
+      "step": 798
+    },
+    {
+      "epoch": 3.234817813765182,
+      "grad_norm": 0.31564516860650377,
+      "learning_rate": 1.3369640572281537e-05,
+      "loss": 0.334,
+      "step": 799
+    },
+    {
+      "epoch": 3.2388663967611335,
+      "grad_norm": 0.3130830633932455,
+      "learning_rate": 1.3316311037233596e-05,
+      "loss": 0.3373,
+      "step": 800
+    },
+    {
+      "epoch": 3.242914979757085,
+      "grad_norm": 0.31721460818888925,
+      "learning_rate": 1.326303494478451e-05,
+      "loss": 0.332,
+      "step": 801
+    },
+    {
+      "epoch": 3.246963562753036,
+      "grad_norm": 0.31838305487397656,
+      "learning_rate": 1.3209812720928472e-05,
+      "loss": 0.3414,
+      "step": 802
+    },
+    {
+      "epoch": 3.251012145748988,
+      "grad_norm": 0.30664152724415833,
+      "learning_rate": 1.315664479122898e-05,
+      "loss": 0.328,
+      "step": 803
+    },
+    {
+      "epoch": 3.2550607287449393,
+      "grad_norm": 0.3237295263059715,
+      "learning_rate": 1.3103531580815378e-05,
+      "loss": 0.3347,
+      "step": 804
+    },
+    {
+      "epoch": 3.2591093117408905,
+      "grad_norm": 0.32701478297912856,
+      "learning_rate": 1.3050473514379488e-05,
+      "loss": 0.3272,
+      "step": 805
+    },
+    {
+      "epoch": 3.263157894736842,
+      "grad_norm": 0.3278433045687872,
+      "learning_rate": 1.2997471016172188e-05,
+      "loss": 0.3298,
+      "step": 806
+    },
+    {
+      "epoch": 3.2672064777327936,
+      "grad_norm": 0.3276027250468151,
+      "learning_rate": 1.2944524510000042e-05,
+      "loss": 0.3265,
+      "step": 807
+    },
+    {
+      "epoch": 3.2712550607287447,
+      "grad_norm": 0.3174224705755455,
+      "learning_rate": 1.2891634419221908e-05,
+      "loss": 0.3436,
+      "step": 808
+    },
+    {
+      "epoch": 3.2753036437246963,
+      "grad_norm": 0.295062934688034,
+      "learning_rate": 1.2838801166745545e-05,
+      "loss": 0.3401,
+      "step": 809
+    },
+    {
+      "epoch": 3.279352226720648,
+      "grad_norm": 0.35331800965849947,
+      "learning_rate": 1.2786025175024216e-05,
+      "loss": 0.3241,
+      "step": 810
+    },
+    {
+      "epoch": 3.283400809716599,
+      "grad_norm": 0.29602124483204195,
+      "learning_rate": 1.2733306866053357e-05,
+      "loss": 0.3325,
+      "step": 811
+    },
+    {
+      "epoch": 3.2874493927125505,
+      "grad_norm": 0.32115742242083256,
+      "learning_rate": 1.2680646661367163e-05,
+      "loss": 0.3297,
+      "step": 812
+    },
+    {
+      "epoch": 3.291497975708502,
+      "grad_norm": 0.3242334768836147,
+      "learning_rate": 1.2628044982035219e-05,
+      "loss": 0.3409,
+      "step": 813
+    },
+    {
+      "epoch": 3.2955465587044532,
+      "grad_norm": 0.3152629735461682,
+      "learning_rate": 1.2575502248659155e-05,
+      "loss": 0.3371,
+      "step": 814
+    },
+    {
+      "epoch": 3.299595141700405,
+      "grad_norm": 0.31144555608640234,
+      "learning_rate": 1.2523018881369269e-05,
+      "loss": 0.3277,
+      "step": 815
+    },
+    {
+      "epoch": 3.3036437246963564,
+      "grad_norm": 0.3012582045823565,
+      "learning_rate": 1.2470595299821176e-05,
+      "loss": 0.3289,
+      "step": 816
+    },
+    {
+      "epoch": 3.3076923076923075,
+      "grad_norm": 0.3287893071431169,
+      "learning_rate": 1.2418231923192416e-05,
+      "loss": 0.3268,
+      "step": 817
+    },
+    {
+      "epoch": 3.311740890688259,
+      "grad_norm": 0.33618609921155107,
+      "learning_rate": 1.2365929170179165e-05,
+      "loss": 0.3369,
+      "step": 818
+    },
+    {
+      "epoch": 3.3157894736842106,
+      "grad_norm": 0.28717724015162976,
+      "learning_rate": 1.2313687458992848e-05,
+      "loss": 0.3259,
+      "step": 819
+    },
+    {
+      "epoch": 3.3198380566801617,
+      "grad_norm": 0.32620887996960823,
+      "learning_rate": 1.2261507207356785e-05,
+      "loss": 0.3298,
+      "step": 820
+    },
+    {
+      "epoch": 3.3238866396761133,
+      "grad_norm": 0.3278095442275866,
+      "learning_rate": 1.220938883250288e-05,
+      "loss": 0.3351,
+      "step": 821
+    },
+    {
+      "epoch": 3.327935222672065,
+      "grad_norm": 0.2916394612465855,
+      "learning_rate": 1.215733275116827e-05,
+      "loss": 0.3241,
+      "step": 822
+    },
+    {
+      "epoch": 3.331983805668016,
+      "grad_norm": 0.354845302407426,
+      "learning_rate": 1.2105339379592005e-05,
+      "loss": 0.3286,
+      "step": 823
+    },
+    {
+      "epoch": 3.3360323886639676,
+      "grad_norm": 0.32429416098954467,
+      "learning_rate": 1.2053409133511681e-05,
+      "loss": 0.3222,
+      "step": 824
+    },
+    {
+      "epoch": 3.340080971659919,
+      "grad_norm": 0.3118071613503536,
+      "learning_rate": 1.2001542428160173e-05,
+      "loss": 0.3274,
+      "step": 825
+    },
+    {
+      "epoch": 3.3441295546558703,
+      "grad_norm": 0.3300641194085066,
+      "learning_rate": 1.1949739678262278e-05,
+      "loss": 0.3294,
+      "step": 826
+    },
+    {
+      "epoch": 3.348178137651822,
+      "grad_norm": 0.3526317895904749,
+      "learning_rate": 1.189800129803139e-05,
+      "loss": 0.3243,
+      "step": 827
+    },
+    {
+      "epoch": 3.3522267206477734,
+      "grad_norm": 0.32299440896407194,
+      "learning_rate": 1.1846327701166228e-05,
+      "loss": 0.3412,
+      "step": 828
+    },
+    {
+      "epoch": 3.3562753036437245,
+      "grad_norm": 0.33087461131560686,
+      "learning_rate": 1.1794719300847493e-05,
+      "loss": 0.3324,
+      "step": 829
+    },
+    {
+      "epoch": 3.360323886639676,
+      "grad_norm": 0.3095764024637083,
+      "learning_rate": 1.1743176509734582e-05,
+      "loss": 0.3322,
+      "step": 830
+    },
+    {
+      "epoch": 3.3643724696356276,
+      "grad_norm": 0.33679473916211283,
+      "learning_rate": 1.1691699739962275e-05,
+      "loss": 0.332,
+      "step": 831
+    },
+    {
+      "epoch": 3.3684210526315788,
+      "grad_norm": 0.31329732108353964,
+      "learning_rate": 1.1640289403137456e-05,
+      "loss": 0.3274,
+      "step": 832
+    },
+    {
+      "epoch": 3.3724696356275303,
+      "grad_norm": 0.33938244382255206,
+      "learning_rate": 1.1588945910335802e-05,
+      "loss": 0.3246,
+      "step": 833
+    },
+    {
+      "epoch": 3.376518218623482,
+      "grad_norm": 0.29893383575602395,
+      "learning_rate": 1.1537669672098517e-05,
+      "loss": 0.3332,
+      "step": 834
+    },
+    {
+      "epoch": 3.380566801619433,
+      "grad_norm": 0.3174688342645778,
+      "learning_rate": 1.1486461098429036e-05,
+      "loss": 0.3217,
+      "step": 835
+    },
+    {
+      "epoch": 3.3846153846153846,
+      "grad_norm": 0.3230088721338807,
+      "learning_rate": 1.1435320598789747e-05,
+      "loss": 0.3308,
+      "step": 836
+    },
+    {
+      "epoch": 3.388663967611336,
+      "grad_norm": 0.3134805554682831,
+      "learning_rate": 1.1384248582098736e-05,
+      "loss": 0.3264,
+      "step": 837
+    },
+    {
+      "epoch": 3.3927125506072873,
+      "grad_norm": 0.31866455364852075,
+      "learning_rate": 1.1333245456726467e-05,
+      "loss": 0.3267,
+      "step": 838
+    },
+    {
+      "epoch": 3.396761133603239,
+      "grad_norm": 0.3205805633504907,
+      "learning_rate": 1.128231163049258e-05,
+      "loss": 0.3242,
+      "step": 839
+    },
+    {
+      "epoch": 3.4008097165991904,
+      "grad_norm": 0.3139603052925247,
+      "learning_rate": 1.123144751066259e-05,
+      "loss": 0.3353,
+      "step": 840
+    },
+    {
+      "epoch": 3.4048582995951415,
+      "grad_norm": 0.3302453249214024,
+      "learning_rate": 1.1180653503944657e-05,
+      "loss": 0.3345,
+      "step": 841
+    },
+    {
+      "epoch": 3.408906882591093,
+      "grad_norm": 0.3162291712506773,
+      "learning_rate": 1.1129930016486287e-05,
+      "loss": 0.3316,
+      "step": 842
+    },
+    {
+      "epoch": 3.4129554655870447,
+      "grad_norm": 0.3191138263962919,
+      "learning_rate": 1.107927745387114e-05,
+      "loss": 0.3448,
+      "step": 843
+    },
+    {
+      "epoch": 3.417004048582996,
+      "grad_norm": 0.32614939349902516,
+      "learning_rate": 1.1028696221115766e-05,
+      "loss": 0.328,
+      "step": 844
+    },
+    {
+      "epoch": 3.4210526315789473,
+      "grad_norm": 0.31471179795425414,
+      "learning_rate": 1.0978186722666341e-05,
+      "loss": 0.3279,
+      "step": 845
+    },
+    {
+      "epoch": 3.425101214574899,
+      "grad_norm": 0.3458271681521562,
+      "learning_rate": 1.0927749362395475e-05,
+      "loss": 0.3135,
+      "step": 846
+    },
+    {
+      "epoch": 3.42914979757085,
+      "grad_norm": 0.33016711415905703,
+      "learning_rate": 1.0877384543598954e-05,
+      "loss": 0.3272,
+      "step": 847
+    },
+    {
+      "epoch": 3.4331983805668016,
+      "grad_norm": 0.34843406026476764,
+      "learning_rate": 1.0827092668992529e-05,
+      "loss": 0.3284,
+      "step": 848
+    },
+    {
+      "epoch": 3.437246963562753,
+      "grad_norm": 0.31423406161849593,
+      "learning_rate": 1.077687414070869e-05,
+      "loss": 0.3408,
+      "step": 849
+    },
+    {
+      "epoch": 3.4412955465587043,
+      "grad_norm": 0.3389775814987399,
+      "learning_rate": 1.0726729360293451e-05,
+      "loss": 0.3365,
+      "step": 850
+    },
+    {
+      "epoch": 3.445344129554656,
+      "grad_norm": 0.34174820047373017,
+      "learning_rate": 1.0676658728703147e-05,
+      "loss": 0.3422,
+      "step": 851
+    },
+    {
+      "epoch": 3.4493927125506074,
+      "grad_norm": 0.2949160641500599,
+      "learning_rate": 1.0626662646301198e-05,
+      "loss": 0.3398,
+      "step": 852
+    },
+    {
+      "epoch": 3.4534412955465585,
+      "grad_norm": 0.3537243656300022,
+      "learning_rate": 1.0576741512854958e-05,
+      "loss": 0.3247,
+      "step": 853
+    },
+    {
+      "epoch": 3.45748987854251,
+      "grad_norm": 0.3012151452079631,
+      "learning_rate": 1.052689572753247e-05,
+      "loss": 0.3441,
+      "step": 854
+    },
+    {
+      "epoch": 3.4615384615384617,
+      "grad_norm": 0.3081776592530041,
+      "learning_rate": 1.0477125688899328e-05,
+      "loss": 0.3334,
+      "step": 855
+    },
+    {
+      "epoch": 3.465587044534413,
+      "grad_norm": 0.3073221068375501,
+      "learning_rate": 1.0427431794915406e-05,
+      "loss": 0.3365,
+      "step": 856
+    },
+    {
+      "epoch": 3.4696356275303644,
+      "grad_norm": 0.3192736756468624,
+      "learning_rate": 1.0377814442931769e-05,
+      "loss": 0.3246,
+      "step": 857
+    },
+    {
+      "epoch": 3.473684210526316,
+      "grad_norm": 0.33513906849060876,
+      "learning_rate": 1.0328274029687439e-05,
+      "loss": 0.3316,
+      "step": 858
+    },
+    {
+      "epoch": 3.477732793522267,
+      "grad_norm": 0.28163598245930727,
+      "learning_rate": 1.0278810951306253e-05,
+      "loss": 0.3365,
+      "step": 859
+    },
+    {
+      "epoch": 3.4817813765182186,
+      "grad_norm": 0.3824939634522167,
+      "learning_rate": 1.0229425603293647e-05,
+      "loss": 0.3229,
+      "step": 860
+    },
+    {
+      "epoch": 3.48582995951417,
+      "grad_norm": 0.3115166522017471,
+      "learning_rate": 1.0180118380533558e-05,
+      "loss": 0.3469,
+      "step": 861
+    },
+    {
+      "epoch": 3.4898785425101213,
+      "grad_norm": 0.30281356005800436,
+      "learning_rate": 1.0130889677285225e-05,
+      "loss": 0.3192,
+      "step": 862
+    },
+    {
+      "epoch": 3.493927125506073,
+      "grad_norm": 0.30252397100125633,
+      "learning_rate": 1.008173988718005e-05,
+      "loss": 0.3302,
+      "step": 863
+    },
+    {
+      "epoch": 3.4979757085020244,
+      "grad_norm": 0.29432599341156035,
+      "learning_rate": 1.0032669403218439e-05,
+      "loss": 0.3247,
+      "step": 864
+    },
+    {
+      "epoch": 3.5020242914979756,
+      "grad_norm": 0.29924576129547265,
+      "learning_rate": 9.983678617766677e-06,
+      "loss": 0.3272,
+      "step": 865
+    },
+    {
+      "epoch": 3.506072874493927,
+      "grad_norm": 0.3064899849747424,
+      "learning_rate": 9.934767922553788e-06,
+      "loss": 0.3298,
+      "step": 866
+    },
+    {
+      "epoch": 3.5101214574898787,
+      "grad_norm": 0.28242764422670363,
+      "learning_rate": 9.885937708668371e-06,
+      "loss": 0.335,
+      "step": 867
+    },
+    {
+      "epoch": 3.51417004048583,
+      "grad_norm": 0.3029189018802603,
+      "learning_rate": 9.837188366555524e-06,
+      "loss": 0.3415,
+      "step": 868
+    },
+    {
+      "epoch": 3.5182186234817814,
+      "grad_norm": 0.30934273594639106,
+      "learning_rate": 9.788520286013691e-06,
+      "loss": 0.3408,
+      "step": 869
+    },
+    {
+      "epoch": 3.522267206477733,
+      "grad_norm": 0.30732960951715843,
+      "learning_rate": 9.739933856191537e-06,
+      "loss": 0.3297,
+      "step": 870
+    },
+    {
+      "epoch": 3.526315789473684,
+      "grad_norm": 0.29638849049594435,
+      "learning_rate": 9.691429465584866e-06,
+      "loss": 0.3238,
+      "step": 871
+    },
+    {
+      "epoch": 3.5303643724696356,
+      "grad_norm": 0.2972488268754308,
+      "learning_rate": 9.643007502033492e-06,
+      "loss": 0.3243,
+      "step": 872
+    },
+    {
+      "epoch": 3.534412955465587,
+      "grad_norm": 0.30374844776436294,
+      "learning_rate": 9.594668352718152e-06,
+      "loss": 0.3375,
+      "step": 873
+    },
+    {
+      "epoch": 3.5384615384615383,
+      "grad_norm": 0.30739346935093814,
+      "learning_rate": 9.546412404157383e-06,
+      "loss": 0.3279,
+      "step": 874
+    },
+    {
+      "epoch": 3.54251012145749,
+      "grad_norm": 0.288958699681425,
+      "learning_rate": 9.498240042204471e-06,
+      "loss": 0.3253,
+      "step": 875
+    },
+    {
+      "epoch": 3.5465587044534415,
+      "grad_norm": 0.2971189062475765,
+      "learning_rate": 9.450151652044339e-06,
+      "loss": 0.3249,
+      "step": 876
+    },
+    {
+      "epoch": 3.5506072874493926,
+      "grad_norm": 0.2934696844243091,
+      "learning_rate": 9.402147618190472e-06,
+      "loss": 0.3319,
+      "step": 877
+    },
+    {
+      "epoch": 3.554655870445344,
+      "grad_norm": 0.30319283411104503,
+      "learning_rate": 9.354228324481841e-06,
+      "loss": 0.3333,
+      "step": 878
+    },
+    {
+      "epoch": 3.5587044534412957,
+      "grad_norm": 0.2984801244096838,
+      "learning_rate": 9.306394154079852e-06,
+      "loss": 0.3411,
+      "step": 879
+    },
+    {
+      "epoch": 3.562753036437247,
+      "grad_norm": 0.29931085440872396,
+      "learning_rate": 9.258645489465254e-06,
+      "loss": 0.3352,
+      "step": 880
+    },
+    {
+      "epoch": 3.5668016194331984,
+      "grad_norm": 0.32047606057780387,
+      "learning_rate": 9.21098271243509e-06,
+      "loss": 0.3266,
+      "step": 881
+    },
+    {
+      "epoch": 3.57085020242915,
+      "grad_norm": 0.3177765829317743,
+      "learning_rate": 9.163406204099659e-06,
+      "loss": 0.3315,
+      "step": 882
+    },
+    {
+      "epoch": 3.574898785425101,
+      "grad_norm": 0.3092967524814139,
+      "learning_rate": 9.115916344879452e-06,
+      "loss": 0.3368,
+      "step": 883
+    },
+    {
+      "epoch": 3.5789473684210527,
+      "grad_norm": 0.3169874395785488,
+      "learning_rate": 9.068513514502126e-06,
+      "loss": 0.3413,
+      "step": 884
+    },
+    {
+      "epoch": 3.582995951417004,
+      "grad_norm": 0.2981316520643636,
+      "learning_rate": 9.021198091999439e-06,
+      "loss": 0.3414,
+      "step": 885
+    },
+    {
+      "epoch": 3.5870445344129553,
+      "grad_norm": 0.3124533121112578,
+      "learning_rate": 8.973970455704251e-06,
+      "loss": 0.3308,
+      "step": 886
+    },
+    {
+      "epoch": 3.591093117408907,
+      "grad_norm": 0.32967653206058295,
+      "learning_rate": 8.926830983247495e-06,
+      "loss": 0.3329,
+      "step": 887
+    },
+    {
+      "epoch": 3.5951417004048585,
+      "grad_norm": 0.2924601878750102,
+      "learning_rate": 8.879780051555116e-06,
+      "loss": 0.3361,
+      "step": 888
+    },
+    {
+      "epoch": 3.5991902834008096,
+      "grad_norm": 0.3190786058555971,
+      "learning_rate": 8.83281803684512e-06,
+      "loss": 0.3379,
+      "step": 889
+    },
+    {
+      "epoch": 3.603238866396761,
+      "grad_norm": 0.33582645816526935,
+      "learning_rate": 8.785945314624532e-06,
+      "loss": 0.3269,
+      "step": 890
+    },
+    {
+      "epoch": 3.6072874493927127,
+      "grad_norm": 0.2954917814115415,
+      "learning_rate": 8.739162259686385e-06,
+      "loss": 0.3368,
+      "step": 891
+    },
+    {
+      "epoch": 3.611336032388664,
+      "grad_norm": 0.322345033733098,
+      "learning_rate": 8.692469246106734e-06,
+      "loss": 0.3302,
+      "step": 892
+    },
+    {
+      "epoch": 3.6153846153846154,
+      "grad_norm": 0.3319163879244466,
+      "learning_rate": 8.645866647241677e-06,
+      "loss": 0.3287,
+      "step": 893
+    },
+    {
+      "epoch": 3.619433198380567,
+      "grad_norm": 0.28378487884926523,
+      "learning_rate": 8.599354835724348e-06,
+      "loss": 0.3367,
+      "step": 894
+    },
+    {
+      "epoch": 3.623481781376518,
+      "grad_norm": 0.3144070765253364,
+      "learning_rate": 8.552934183461958e-06,
+      "loss": 0.334,
+      "step": 895
+    },
+    {
+      "epoch": 3.6275303643724697,
+      "grad_norm": 0.30532745670403705,
+      "learning_rate": 8.506605061632805e-06,
+      "loss": 0.3234,
+      "step": 896
+    },
+    {
+      "epoch": 3.6315789473684212,
+      "grad_norm": 0.28686164323633334,
+      "learning_rate": 8.460367840683308e-06,
+      "loss": 0.3248,
+      "step": 897
+    },
+    {
+      "epoch": 3.6356275303643724,
+      "grad_norm": 0.3080036218088754,
+      "learning_rate": 8.414222890325065e-06,
+      "loss": 0.3302,
+      "step": 898
+    },
+    {
+      "epoch": 3.639676113360324,
+      "grad_norm": 0.2995214966569133,
+      "learning_rate": 8.368170579531855e-06,
+      "loss": 0.3415,
+      "step": 899
+    },
+    {
+      "epoch": 3.6437246963562755,
+      "grad_norm": 0.3123995779348358,
+      "learning_rate": 8.322211276536733e-06,
+      "loss": 0.3391,
+      "step": 900
+    },
+    {
+      "epoch": 3.6477732793522266,
+      "grad_norm": 0.3033867433757769,
+      "learning_rate": 8.27634534882907e-06,
+      "loss": 0.3312,
+      "step": 901
+    },
+    {
+      "epoch": 3.651821862348178,
+      "grad_norm": 0.2975816482312663,
+      "learning_rate": 8.230573163151585e-06,
+      "loss": 0.3175,
+      "step": 902
+    },
+    {
+      "epoch": 3.6558704453441297,
+      "grad_norm": 0.30234120679631893,
+      "learning_rate": 8.184895085497466e-06,
+      "loss": 0.3396,
+      "step": 903
+    },
+    {
+      "epoch": 3.659919028340081,
+      "grad_norm": 0.33384387563561374,
+      "learning_rate": 8.139311481107401e-06,
+      "loss": 0.3387,
+      "step": 904
+    },
+    {
+      "epoch": 3.6639676113360324,
+      "grad_norm": 0.3181701788784848,
+      "learning_rate": 8.093822714466686e-06,
+      "loss": 0.3248,
+      "step": 905
+    },
+    {
+      "epoch": 3.668016194331984,
+      "grad_norm": 0.2865857754858248,
+      "learning_rate": 8.048429149302271e-06,
+      "loss": 0.3428,
+      "step": 906
+    },
+    {
+      "epoch": 3.672064777327935,
+      "grad_norm": 0.3308173863522947,
+      "learning_rate": 8.003131148579902e-06,
+      "loss": 0.3348,
+      "step": 907
+    },
+    {
+      "epoch": 3.6761133603238867,
+      "grad_norm": 0.3359557749006318,
+      "learning_rate": 7.957929074501182e-06,
+      "loss": 0.3318,
+      "step": 908
+    },
+    {
+      "epoch": 3.6801619433198383,
+      "grad_norm": 0.2869993916746231,
+      "learning_rate": 7.912823288500695e-06,
+      "loss": 0.3336,
+      "step": 909
+    },
+    {
+      "epoch": 3.6842105263157894,
+      "grad_norm": 0.3075978305358129,
+      "learning_rate": 7.867814151243102e-06,
+      "loss": 0.3334,
+      "step": 910
+    },
+    {
+      "epoch": 3.688259109311741,
+      "grad_norm": 0.3106457178033204,
+      "learning_rate": 7.822902022620267e-06,
+      "loss": 0.3277,
+      "step": 911
+    },
+    {
+      "epoch": 3.6923076923076925,
+      "grad_norm": 0.314246258634998,
+      "learning_rate": 7.778087261748377e-06,
+      "loss": 0.3297,
+      "step": 912
+    },
+    {
+      "epoch": 3.6963562753036436,
+      "grad_norm": 0.30772841621226554,
+      "learning_rate": 7.733370226965056e-06,
+      "loss": 0.3396,
+      "step": 913
+    },
+    {
+      "epoch": 3.700404858299595,
+      "grad_norm": 0.31894157029288045,
+      "learning_rate": 7.688751275826527e-06,
+      "loss": 0.3465,
+      "step": 914
+    },
+    {
+      "epoch": 3.7044534412955468,
+      "grad_norm": 0.30830407330345105,
+      "learning_rate": 7.644230765104735e-06,
+      "loss": 0.3181,
+      "step": 915
+    },
+    {
+      "epoch": 3.708502024291498,
+      "grad_norm": 0.31843649950462605,
+      "learning_rate": 7.599809050784502e-06,
+      "loss": 0.337,
+      "step": 916
+    },
+    {
+      "epoch": 3.7125506072874495,
+      "grad_norm": 0.2941368397901161,
+      "learning_rate": 7.555486488060661e-06,
+      "loss": 0.3327,
+      "step": 917
+    },
+    {
+      "epoch": 3.716599190283401,
+      "grad_norm": 0.3145484691244138,
+      "learning_rate": 7.511263431335249e-06,
+      "loss": 0.3275,
+      "step": 918
+    },
+    {
+      "epoch": 3.720647773279352,
+      "grad_norm": 0.3134855183175215,
+      "learning_rate": 7.467140234214656e-06,
+      "loss": 0.3444,
+      "step": 919
+    },
+    {
+      "epoch": 3.7246963562753037,
+      "grad_norm": 0.3148535001045878,
+      "learning_rate": 7.423117249506777e-06,
+      "loss": 0.3212,
+      "step": 920
+    },
+    {
+      "epoch": 3.7287449392712553,
+      "grad_norm": 0.2949547660418575,
+      "learning_rate": 7.379194829218231e-06,
+      "loss": 0.3328,
+      "step": 921
+    },
+    {
+      "epoch": 3.7327935222672064,
+      "grad_norm": 0.3086328241925601,
+      "learning_rate": 7.335373324551518e-06,
+      "loss": 0.3431,
+      "step": 922
+    },
+    {
+      "epoch": 3.736842105263158,
+      "grad_norm": 0.2890836714268375,
+      "learning_rate": 7.291653085902224e-06,
+      "loss": 0.327,
+      "step": 923
+    },
+    {
+      "epoch": 3.7408906882591095,
+      "grad_norm": 0.28721070040710234,
+      "learning_rate": 7.248034462856213e-06,
+      "loss": 0.3432,
+      "step": 924
+    },
+    {
+      "epoch": 3.7449392712550607,
+      "grad_norm": 0.31021331185858425,
+      "learning_rate": 7.20451780418683e-06,
+      "loss": 0.3304,
+      "step": 925
+    },
+    {
+      "epoch": 3.748987854251012,
+      "grad_norm": 0.3032719066976004,
+      "learning_rate": 7.161103457852128e-06,
+      "loss": 0.3209,
+      "step": 926
+    },
+    {
+      "epoch": 3.753036437246964,
+      "grad_norm": 0.29582530583042044,
+      "learning_rate": 7.117791770992046e-06,
+      "loss": 0.3304,
+      "step": 927
+    },
+    {
+      "epoch": 3.757085020242915,
+      "grad_norm": 0.2849470534034771,
+      "learning_rate": 7.074583089925688e-06,
+      "loss": 0.3326,
+      "step": 928
+    },
+    {
+      "epoch": 3.7611336032388665,
+      "grad_norm": 0.285844636018954,
+      "learning_rate": 7.0314777601485155e-06,
+      "loss": 0.328,
+      "step": 929
+    },
+    {
+      "epoch": 3.765182186234818,
+      "grad_norm": 0.2952078297157979,
+      "learning_rate": 6.988476126329604e-06,
+      "loss": 0.3371,
+      "step": 930
+    },
+    {
+      "epoch": 3.769230769230769,
+      "grad_norm": 0.28484535149732315,
+      "learning_rate": 6.945578532308857e-06,
+      "loss": 0.3275,
+      "step": 931
+    },
+    {
+      "epoch": 3.7732793522267207,
+      "grad_norm": 0.2893292908751114,
+      "learning_rate": 6.902785321094301e-06,
+      "loss": 0.3286,
+      "step": 932
+    },
+    {
+      "epoch": 3.7773279352226723,
+      "grad_norm": 0.29296152688589394,
+      "learning_rate": 6.8600968348593115e-06,
+      "loss": 0.3162,
+      "step": 933
+    },
+    {
+      "epoch": 3.7813765182186234,
+      "grad_norm": 0.2847837885863179,
+      "learning_rate": 6.817513414939892e-06,
+      "loss": 0.3453,
+      "step": 934
+    },
+    {
+      "epoch": 3.785425101214575,
+      "grad_norm": 0.2889734015042347,
+      "learning_rate": 6.775035401831924e-06,
+      "loss": 0.3378,
+      "step": 935
+    },
+    {
+      "epoch": 3.7894736842105265,
+      "grad_norm": 0.28779182168172524,
+      "learning_rate": 6.732663135188473e-06,
+      "loss": 0.334,
+      "step": 936
+    },
+    {
+      "epoch": 3.7935222672064777,
+      "grad_norm": 0.2829215144341817,
+      "learning_rate": 6.690396953817055e-06,
+      "loss": 0.332,
+      "step": 937
+    },
+    {
+      "epoch": 3.7975708502024292,
+      "grad_norm": 0.2907895584239673,
+      "learning_rate": 6.648237195676927e-06,
+      "loss": 0.326,
+      "step": 938
+    },
+    {
+      "epoch": 3.801619433198381,
+      "grad_norm": 0.28882503220267125,
+      "learning_rate": 6.606184197876395e-06,
+      "loss": 0.3378,
+      "step": 939
+    },
+    {
+      "epoch": 3.805668016194332,
+      "grad_norm": 0.28204536979760186,
+      "learning_rate": 6.564238296670109e-06,
+      "loss": 0.3312,
+      "step": 940
+    },
+    {
+      "epoch": 3.8097165991902835,
+      "grad_norm": 0.28511723249604176,
+      "learning_rate": 6.52239982745638e-06,
+      "loss": 0.3427,
+      "step": 941
+    },
+    {
+      "epoch": 3.813765182186235,
+      "grad_norm": 0.3022119568146914,
+      "learning_rate": 6.480669124774484e-06,
+      "loss": 0.343,
+      "step": 942
+    },
+    {
+      "epoch": 3.817813765182186,
+      "grad_norm": 0.31678568785184086,
+      "learning_rate": 6.439046522302013e-06,
+      "loss": 0.3238,
+      "step": 943
+    },
+    {
+      "epoch": 3.8218623481781377,
+      "grad_norm": 0.2852051416880907,
+      "learning_rate": 6.397532352852191e-06,
+      "loss": 0.3248,
+      "step": 944
+    },
+    {
+      "epoch": 3.8259109311740893,
+      "grad_norm": 0.30011455872609544,
+      "learning_rate": 6.356126948371202e-06,
+      "loss": 0.3287,
+      "step": 945
+    },
+    {
+      "epoch": 3.8299595141700404,
+      "grad_norm": 0.3221112060069891,
+      "learning_rate": 6.314830639935559e-06,
+      "loss": 0.3275,
+      "step": 946
+    },
+    {
+      "epoch": 3.834008097165992,
+      "grad_norm": 0.29551163830857624,
+      "learning_rate": 6.273643757749448e-06,
+      "loss": 0.3187,
+      "step": 947
+    },
+    {
+      "epoch": 3.8380566801619436,
+      "grad_norm": 0.294340418678824,
+      "learning_rate": 6.2325666311420856e-06,
+      "loss": 0.3219,
+      "step": 948
+    },
+    {
+      "epoch": 3.8421052631578947,
+      "grad_norm": 0.30044044766104316,
+      "learning_rate": 6.1915995885650675e-06,
+      "loss": 0.3315,
+      "step": 949
+    },
+    {
+      "epoch": 3.8461538461538463,
+      "grad_norm": 0.29251699038762596,
+      "learning_rate": 6.150742957589784e-06,
+      "loss": 0.3189,
+      "step": 950
+    },
+    {
+      "epoch": 3.850202429149798,
+      "grad_norm": 0.2842268992226119,
+      "learning_rate": 6.109997064904774e-06,
+      "loss": 0.3373,
+      "step": 951
+    },
+    {
+      "epoch": 3.854251012145749,
+      "grad_norm": 0.2974664779912128,
+      "learning_rate": 6.069362236313092e-06,
+      "loss": 0.3218,
+      "step": 952
+    },
+    {
+      "epoch": 3.8582995951417005,
+      "grad_norm": 0.2934139443216481,
+      "learning_rate": 6.0288387967297565e-06,
+      "loss": 0.3207,
+      "step": 953
+    },
+    {
+      "epoch": 3.862348178137652,
+      "grad_norm": 0.30302392561914854,
+      "learning_rate": 5.988427070179108e-06,
+      "loss": 0.3268,
+      "step": 954
+    },
+    {
+      "epoch": 3.866396761133603,
+      "grad_norm": 0.284501493853288,
+      "learning_rate": 5.948127379792234e-06,
+      "loss": 0.3238,
+      "step": 955
+    },
+    {
+      "epoch": 3.8704453441295548,
+      "grad_norm": 0.30559041088669875,
+      "learning_rate": 5.90794004780439e-06,
+      "loss": 0.3367,
+      "step": 956
+    },
+    {
+      "epoch": 3.8744939271255063,
+      "grad_norm": 0.28851168026987284,
+      "learning_rate": 5.867865395552405e-06,
+      "loss": 0.3347,
+      "step": 957
+    },
+    {
+      "epoch": 3.8785425101214575,
+      "grad_norm": 0.27367182459790923,
+      "learning_rate": 5.8279037434721364e-06,
+      "loss": 0.3301,
+      "step": 958
+    },
+    {
+      "epoch": 3.882591093117409,
+      "grad_norm": 0.2846530633562061,
+      "learning_rate": 5.788055411095892e-06,
+      "loss": 0.3237,
+      "step": 959
+    },
+    {
+      "epoch": 3.8866396761133606,
+      "grad_norm": 0.3050603713141295,
+      "learning_rate": 5.7483207170498625e-06,
+      "loss": 0.3239,
+      "step": 960
+    },
+    {
+      "epoch": 3.8906882591093117,
+      "grad_norm": 0.2904717895958458,
+      "learning_rate": 5.708699979051613e-06,
+      "loss": 0.3232,
+      "step": 961
+    },
+    {
+      "epoch": 3.8947368421052633,
+      "grad_norm": 0.28111950556186105,
+      "learning_rate": 5.66919351390751e-06,
+      "loss": 0.3351,
+      "step": 962
+    },
+    {
+      "epoch": 3.898785425101215,
+      "grad_norm": 0.3165009857596391,
+      "learning_rate": 5.629801637510188e-06,
+      "loss": 0.3311,
+      "step": 963
+    },
+    {
+      "epoch": 3.902834008097166,
+      "grad_norm": 0.2910024536649084,
+      "learning_rate": 5.590524664836043e-06,
+      "loss": 0.319,
+      "step": 964
+    },
+    {
+      "epoch": 3.9068825910931175,
+      "grad_norm": 0.2927224740907455,
+      "learning_rate": 5.551362909942708e-06,
+      "loss": 0.3303,
+      "step": 965
+    },
+    {
+      "epoch": 3.910931174089069,
+      "grad_norm": 0.2945200115881548,
+      "learning_rate": 5.51231668596653e-06,
+      "loss": 0.329,
+      "step": 966
+    },
+    {
+      "epoch": 3.91497975708502,
+      "grad_norm": 0.3043554436192971,
+      "learning_rate": 5.47338630512007e-06,
+      "loss": 0.3311,
+      "step": 967
+    },
+    {
+      "epoch": 3.919028340080972,
+      "grad_norm": 0.2875397042066636,
+      "learning_rate": 5.434572078689617e-06,
+      "loss": 0.3317,
+      "step": 968
+    },
+    {
+      "epoch": 3.9230769230769234,
+      "grad_norm": 0.2874127602951936,
+      "learning_rate": 5.395874317032694e-06,
+      "loss": 0.3323,
+      "step": 969
+    },
+    {
+      "epoch": 3.9271255060728745,
+      "grad_norm": 0.2870475482107977,
+      "learning_rate": 5.357293329575573e-06,
+      "loss": 0.3298,
+      "step": 970
+    },
+    {
+      "epoch": 3.931174089068826,
+      "grad_norm": 0.30323145088667325,
+      "learning_rate": 5.318829424810797e-06,
+      "loss": 0.3325,
+      "step": 971
+    },
+    {
+      "epoch": 3.9352226720647776,
+      "grad_norm": 0.2802699636172721,
+      "learning_rate": 5.280482910294728e-06,
+      "loss": 0.3284,
+      "step": 972
+    },
+    {
+      "epoch": 3.9392712550607287,
+      "grad_norm": 0.2824954676973631,
+      "learning_rate": 5.242254092645078e-06,
+      "loss": 0.3225,
+      "step": 973
+    },
+    {
+      "epoch": 3.9433198380566803,
+      "grad_norm": 0.27859528611112416,
+      "learning_rate": 5.2041432775384445e-06,
+      "loss": 0.3335,
+      "step": 974
+    },
+    {
+      "epoch": 3.9473684210526314,
+      "grad_norm": 0.28826730684251967,
+      "learning_rate": 5.166150769707892e-06,
+      "loss": 0.3146,
+      "step": 975
+    },
+    {
+      "epoch": 3.951417004048583,
+      "grad_norm": 0.3029593134550885,
+      "learning_rate": 5.128276872940503e-06,
+      "loss": 0.3332,
+      "step": 976
+    },
+    {
+      "epoch": 3.9554655870445345,
+      "grad_norm": 0.28279875474175287,
+      "learning_rate": 5.090521890074953e-06,
+      "loss": 0.3175,
+      "step": 977
+    },
+    {
+      "epoch": 3.9595141700404857,
+      "grad_norm": 0.2985132750363249,
+      "learning_rate": 5.0528861229990675e-06,
+      "loss": 0.3271,
+      "step": 978
+    },
+    {
+      "epoch": 3.9635627530364372,
+      "grad_norm": 0.28593338585740175,
+      "learning_rate": 5.015369872647442e-06,
+      "loss": 0.3293,
+      "step": 979
+    },
+    {
+      "epoch": 3.967611336032389,
+      "grad_norm": 0.2778081614643638,
+      "learning_rate": 4.9779734389990195e-06,
+      "loss": 0.3275,
+      "step": 980
+    },
+    {
+      "epoch": 3.97165991902834,
+      "grad_norm": 0.2880437730492521,
+      "learning_rate": 4.940697121074675e-06,
+      "loss": 0.325,
+      "step": 981
+    },
+    {
+      "epoch": 3.9757085020242915,
+      "grad_norm": 0.2818577098269561,
+      "learning_rate": 4.903541216934864e-06,
+      "loss": 0.3217,
+      "step": 982
+    },
+    {
+      "epoch": 3.979757085020243,
+      "grad_norm": 0.27611040371418427,
+      "learning_rate": 4.8665060236772e-06,
+      "loss": 0.3294,
+      "step": 983
+    },
+    {
+      "epoch": 3.983805668016194,
+      "grad_norm": 0.2801932157326473,
+      "learning_rate": 4.829591837434102e-06,
+      "loss": 0.328,
+      "step": 984
+    },
+    {
+      "epoch": 3.9878542510121457,
+      "grad_norm": 0.2833247346475826,
+      "learning_rate": 4.792798953370424e-06,
+      "loss": 0.3326,
+      "step": 985
+    },
+    {
+      "epoch": 3.9919028340080973,
+      "grad_norm": 0.2752033949905642,
+      "learning_rate": 4.756127665681083e-06,
+      "loss": 0.3353,
+      "step": 986
+    },
+    {
+      "epoch": 3.9959514170040484,
+      "grad_norm": 0.2802812112551736,
+      "learning_rate": 4.719578267588725e-06,
+      "loss": 0.327,
+      "step": 987
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 0.2836390675604979,
+      "learning_rate": 4.6831510513413525e-06,
+      "loss": 0.32,
+      "step": 988
+    },
+    {
+      "epoch": 4.004048582995951,
+      "grad_norm": 0.5028362928299511,
+      "learning_rate": 4.6468463082100176e-06,
+      "loss": 0.2803,
+      "step": 989
+    },
+    {
+      "epoch": 4.008097165991903,
+      "grad_norm": 0.444058472540308,
+      "learning_rate": 4.610664328486485e-06,
+      "loss": 0.2553,
+      "step": 990
+    },
+    {
+      "epoch": 4.012145748987854,
+      "grad_norm": 0.3393272803803114,
+      "learning_rate": 4.5746054014809e-06,
+      "loss": 0.2605,
+      "step": 991
+    },
+    {
+      "epoch": 4.016194331983805,
+      "grad_norm": 0.42630542371763264,
+      "learning_rate": 4.5386698155194766e-06,
+      "loss": 0.2578,
+      "step": 992
+    },
+    {
+      "epoch": 4.020242914979757,
+      "grad_norm": 0.5402980678411002,
+      "learning_rate": 4.502857857942204e-06,
+      "loss": 0.2684,
+      "step": 993
+    },
+    {
+      "epoch": 4.0242914979757085,
+      "grad_norm": 0.5465097428726662,
+      "learning_rate": 4.467169815100549e-06,
+      "loss": 0.2719,
+      "step": 994
+    },
+    {
+      "epoch": 4.02834008097166,
+      "grad_norm": 0.3821457116712654,
+      "learning_rate": 4.431605972355137e-06,
+      "loss": 0.2797,
+      "step": 995
+    },
+    {
+      "epoch": 4.032388663967612,
+      "grad_norm": 0.45887333075919357,
+      "learning_rate": 4.396166614073516e-06,
+      "loss": 0.2587,
+      "step": 996
+    },
+    {
+      "epoch": 4.036437246963563,
+      "grad_norm": 0.44965189822889823,
+      "learning_rate": 4.360852023627846e-06,
+      "loss": 0.2616,
+      "step": 997
+    },
+    {
+      "epoch": 4.040485829959514,
+      "grad_norm": 0.4282018100914342,
+      "learning_rate": 4.325662483392656e-06,
+      "loss": 0.2582,
+      "step": 998
+    },
+    {
+      "epoch": 4.044534412955466,
+      "grad_norm": 0.3484384362676654,
+      "learning_rate": 4.29059827474257e-06,
+      "loss": 0.248,
+      "step": 999
+    },
+    {
+      "epoch": 4.048582995951417,
+      "grad_norm": 0.35036749724178406,
+      "learning_rate": 4.255659678050063e-06,
+      "loss": 0.273,
+      "step": 1000
+    },
+    {
+      "epoch": 4.052631578947368,
+      "grad_norm": 0.3876197650638699,
+      "learning_rate": 4.22084697268323e-06,
+      "loss": 0.2592,
+      "step": 1001
+    },
+    {
+      "epoch": 4.05668016194332,
+      "grad_norm": 0.3974738571870152,
+      "learning_rate": 4.186160437003535e-06,
+      "loss": 0.2661,
+      "step": 1002
+    },
+    {
+      "epoch": 4.060728744939271,
+      "grad_norm": 0.35201290763591186,
+      "learning_rate": 4.151600348363582e-06,
+      "loss": 0.2477,
+      "step": 1003
+    },
+    {
+      "epoch": 4.064777327935222,
+      "grad_norm": 0.331707078493155,
+      "learning_rate": 4.11716698310493e-06,
+      "loss": 0.2572,
+      "step": 1004
+    },
+    {
+      "epoch": 4.068825910931174,
+      "grad_norm": 0.32715499717831387,
+      "learning_rate": 4.082860616555848e-06,
+      "loss": 0.2584,
+      "step": 1005
+    },
+    {
+      "epoch": 4.0728744939271255,
+      "grad_norm": 0.33897035209833937,
+      "learning_rate": 4.048681523029125e-06,
+      "loss": 0.2651,
+      "step": 1006
+    },
+    {
+      "epoch": 4.076923076923077,
+      "grad_norm": 0.35783445632652583,
+      "learning_rate": 4.014629975819882e-06,
+      "loss": 0.2622,
+      "step": 1007
+    },
+    {
+      "epoch": 4.080971659919029,
+      "grad_norm": 0.3346304640837659,
+      "learning_rate": 3.9807062472033855e-06,
+      "loss": 0.2596,
+      "step": 1008
+    },
+    {
+      "epoch": 4.08502024291498,
+      "grad_norm": 0.33446777657555143,
+      "learning_rate": 3.94691060843287e-06,
+      "loss": 0.2598,
+      "step": 1009
+    },
+    {
+      "epoch": 4.089068825910931,
+      "grad_norm": 0.3069316373444891,
+      "learning_rate": 3.913243329737353e-06,
+      "loss": 0.2605,
+      "step": 1010
+    },
+    {
+      "epoch": 4.093117408906883,
+      "grad_norm": 0.33967818361823193,
+      "learning_rate": 3.879704680319503e-06,
+      "loss": 0.2653,
+      "step": 1011
+    },
+    {
+      "epoch": 4.097165991902834,
+      "grad_norm": 0.3356114676647828,
+      "learning_rate": 3.846294928353471e-06,
+      "loss": 0.2493,
+      "step": 1012
+    },
+    {
+      "epoch": 4.101214574898785,
+      "grad_norm": 0.3276247533808605,
+      "learning_rate": 3.8130143409827345e-06,
+      "loss": 0.2651,
+      "step": 1013
+    },
+    {
+      "epoch": 4.105263157894737,
+      "grad_norm": 0.31281305191162934,
+      "learning_rate": 3.7798631843179887e-06,
+      "loss": 0.2725,
+      "step": 1014
+    },
+    {
+      "epoch": 4.109311740890688,
+      "grad_norm": 0.31019693477638366,
+      "learning_rate": 3.7468417234349973e-06,
+      "loss": 0.266,
+      "step": 1015
+    },
+    {
+      "epoch": 4.113360323886639,
+      "grad_norm": 0.30199767193173715,
+      "learning_rate": 3.7139502223724823e-06,
+      "loss": 0.2679,
+      "step": 1016
+    },
+    {
+      "epoch": 4.117408906882591,
+      "grad_norm": 0.3034243754212267,
+      "learning_rate": 3.6811889441300095e-06,
+      "loss": 0.2619,
+      "step": 1017
+    },
+    {
+      "epoch": 4.1214574898785425,
+      "grad_norm": 0.32303544076299623,
+      "learning_rate": 3.648558150665886e-06,
+      "loss": 0.2589,
+      "step": 1018
+    },
+    {
+      "epoch": 4.125506072874494,
+      "grad_norm": 0.29670438253860926,
+      "learning_rate": 3.616058102895075e-06,
+      "loss": 0.2722,
+      "step": 1019
+    },
+    {
+      "epoch": 4.129554655870446,
+      "grad_norm": 0.30436688916269034,
+      "learning_rate": 3.583689060687079e-06,
+      "loss": 0.25,
+      "step": 1020
+    },
+    {
+      "epoch": 4.133603238866397,
+      "grad_norm": 0.2938303846078055,
+      "learning_rate": 3.5514512828639023e-06,
+      "loss": 0.2569,
+      "step": 1021
+    },
+    {
+      "epoch": 4.137651821862348,
+      "grad_norm": 0.2951759836929609,
+      "learning_rate": 3.51934502719796e-06,
+      "loss": 0.269,
+      "step": 1022
+    },
+    {
+      "epoch": 4.1417004048583,
+      "grad_norm": 0.32627565451644297,
+      "learning_rate": 3.4873705504100185e-06,
+      "loss": 0.2631,
+      "step": 1023
+    },
+    {
+      "epoch": 4.145748987854251,
+      "grad_norm": 0.2970882480543895,
+      "learning_rate": 3.4555281081671345e-06,
+      "loss": 0.2639,
+      "step": 1024
+    },
+    {
+      "epoch": 4.149797570850202,
+      "grad_norm": 0.29383145274448436,
+      "learning_rate": 3.4238179550806326e-06,
+      "loss": 0.262,
+      "step": 1025
+    },
+    {
+      "epoch": 4.153846153846154,
+      "grad_norm": 0.3017645535217455,
+      "learning_rate": 3.3922403447040497e-06,
+      "loss": 0.2694,
+      "step": 1026
+    },
+    {
+      "epoch": 4.157894736842105,
+      "grad_norm": 0.29940689866908987,
+      "learning_rate": 3.3607955295311247e-06,
+      "loss": 0.2571,
+      "step": 1027
+    },
+    {
+      "epoch": 4.161943319838056,
+      "grad_norm": 0.29309200920441403,
+      "learning_rate": 3.32948376099375e-06,
+      "loss": 0.2643,
+      "step": 1028
+    },
+    {
+      "epoch": 4.165991902834008,
+      "grad_norm": 0.29620496172771504,
+      "learning_rate": 3.2983052894599976e-06,
+      "loss": 0.2602,
+      "step": 1029
+    },
+    {
+      "epoch": 4.17004048582996,
+      "grad_norm": 0.29526876200048524,
+      "learning_rate": 3.2672603642320945e-06,
+      "loss": 0.2498,
+      "step": 1030
+    },
+    {
+      "epoch": 4.174089068825911,
+      "grad_norm": 0.303107504791561,
+      "learning_rate": 3.236349233544438e-06,
+      "loss": 0.2677,
+      "step": 1031
+    },
+    {
+      "epoch": 4.178137651821863,
+      "grad_norm": 0.30524560522763494,
+      "learning_rate": 3.205572144561604e-06,
+      "loss": 0.2616,
+      "step": 1032
+    },
+    {
+      "epoch": 4.182186234817814,
+      "grad_norm": 0.294693551454714,
+      "learning_rate": 3.174929343376374e-06,
+      "loss": 0.2483,
+      "step": 1033
+    },
+    {
+      "epoch": 4.186234817813765,
+      "grad_norm": 0.2944754089136599,
+      "learning_rate": 3.1444210750077777e-06,
+      "loss": 0.2742,
+      "step": 1034
+    },
+    {
+      "epoch": 4.190283400809717,
+      "grad_norm": 0.2827447131136314,
+      "learning_rate": 3.1140475833991046e-06,
+      "loss": 0.2808,
+      "step": 1035
+    },
+    {
+      "epoch": 4.194331983805668,
+      "grad_norm": 0.30284875124560884,
+      "learning_rate": 3.0838091114159895e-06,
+      "loss": 0.2504,
+      "step": 1036
+    },
+    {
+      "epoch": 4.198380566801619,
+      "grad_norm": 0.2933704190417536,
+      "learning_rate": 3.0537059008444524e-06,
+      "loss": 0.2669,
+      "step": 1037
+    },
+    {
+      "epoch": 4.202429149797571,
+      "grad_norm": 0.2961454489760588,
+      "learning_rate": 3.0237381923889563e-06,
+      "loss": 0.2743,
+      "step": 1038
+    },
+    {
+      "epoch": 4.206477732793522,
+      "grad_norm": 0.2857800405705651,
+      "learning_rate": 2.9939062256704997e-06,
+      "loss": 0.2677,
+      "step": 1039
+    },
+    {
+      "epoch": 4.2105263157894735,
+      "grad_norm": 0.28706810442879316,
+      "learning_rate": 2.964210239224696e-06,
+      "loss": 0.2496,
+      "step": 1040
+    },
+    {
+      "epoch": 4.2145748987854255,
+      "grad_norm": 0.28136294218780683,
+      "learning_rate": 2.934650470499869e-06,
+      "loss": 0.2582,
+      "step": 1041
+    },
+    {
+      "epoch": 4.218623481781377,
+      "grad_norm": 0.28843971844348143,
+      "learning_rate": 2.905227155855128e-06,
+      "loss": 0.2554,
+      "step": 1042
+    },
+    {
+      "epoch": 4.222672064777328,
+      "grad_norm": 0.2921086704378538,
+      "learning_rate": 2.875940530558519e-06,
+      "loss": 0.2726,
+      "step": 1043
+    },
+    {
+      "epoch": 4.22672064777328,
+      "grad_norm": 0.30691999517540286,
+      "learning_rate": 2.8467908287851154e-06,
+      "loss": 0.2659,
+      "step": 1044
+    },
+    {
+      "epoch": 4.230769230769231,
+      "grad_norm": 0.28488214189043176,
+      "learning_rate": 2.8177782836151603e-06,
+      "loss": 0.2515,
+      "step": 1045
+    },
+    {
+      "epoch": 4.234817813765182,
+      "grad_norm": 0.2881857686599114,
+      "learning_rate": 2.7889031270321874e-06,
+      "loss": 0.2634,
+      "step": 1046
+    },
+    {
+      "epoch": 4.238866396761134,
+      "grad_norm": 0.3009794472448827,
+      "learning_rate": 2.760165589921184e-06,
+      "loss": 0.261,
+      "step": 1047
+    },
+    {
+      "epoch": 4.242914979757085,
+      "grad_norm": 0.3012695039437794,
+      "learning_rate": 2.7315659020667305e-06,
+      "loss": 0.2412,
+      "step": 1048
+    },
+    {
+      "epoch": 4.246963562753036,
+      "grad_norm": 0.29758461262344643,
+      "learning_rate": 2.703104292151164e-06,
+      "loss": 0.2541,
+      "step": 1049
+    },
+    {
+      "epoch": 4.251012145748988,
+      "grad_norm": 0.28522003475014296,
+      "learning_rate": 2.6747809877527607e-06,
+      "loss": 0.2656,
+      "step": 1050
+    },
+    {
+      "epoch": 4.255060728744939,
+      "grad_norm": 0.2796732927964386,
+      "learning_rate": 2.6465962153439082e-06,
+      "loss": 0.2722,
+      "step": 1051
+    },
+    {
+      "epoch": 4.2591093117408905,
+      "grad_norm": 0.29061789298461316,
+      "learning_rate": 2.618550200289303e-06,
+      "loss": 0.2706,
+      "step": 1052
+    },
+    {
+      "epoch": 4.2631578947368425,
+      "grad_norm": 0.29182649366684604,
+      "learning_rate": 2.59064316684412e-06,
+      "loss": 0.2643,
+      "step": 1053
+    },
+    {
+      "epoch": 4.267206477732794,
+      "grad_norm": 0.30432468995317224,
+      "learning_rate": 2.562875338152264e-06,
+      "loss": 0.2521,
+      "step": 1054
+    },
+    {
+      "epoch": 4.271255060728745,
+      "grad_norm": 0.2980595345555326,
+      "learning_rate": 2.5352469362445552e-06,
+      "loss": 0.2455,
+      "step": 1055
+    },
+    {
+      "epoch": 4.275303643724697,
+      "grad_norm": 0.30162284390525157,
+      "learning_rate": 2.5077581820369523e-06,
+      "loss": 0.2661,
+      "step": 1056
+    },
+    {
+      "epoch": 4.279352226720648,
+      "grad_norm": 0.30095780552442897,
+      "learning_rate": 2.480409295328803e-06,
+      "loss": 0.2503,
+      "step": 1057
+    },
+    {
+      "epoch": 4.283400809716599,
+      "grad_norm": 0.2906010785478982,
+      "learning_rate": 2.45320049480108e-06,
+      "loss": 0.263,
+      "step": 1058
+    },
+    {
+      "epoch": 4.287449392712551,
+      "grad_norm": 0.29378618608940626,
+      "learning_rate": 2.4261319980146293e-06,
+      "loss": 0.2687,
+      "step": 1059
+    },
+    {
+      "epoch": 4.291497975708502,
+      "grad_norm": 0.28756196312570526,
+      "learning_rate": 2.399204021408421e-06,
+      "loss": 0.2681,
+      "step": 1060
+    },
+    {
+      "epoch": 4.295546558704453,
+      "grad_norm": 0.34200309994191197,
+      "learning_rate": 2.3724167802978395e-06,
+      "loss": 0.252,
+      "step": 1061
+    },
+    {
+      "epoch": 4.299595141700405,
+      "grad_norm": 0.277058244246811,
+      "learning_rate": 2.3457704888729625e-06,
+      "loss": 0.2697,
+      "step": 1062
+    },
+    {
+      "epoch": 4.303643724696356,
+      "grad_norm": 0.2933283072650462,
+      "learning_rate": 2.3192653601968162e-06,
+      "loss": 0.2592,
+      "step": 1063
+    },
+    {
+      "epoch": 4.3076923076923075,
+      "grad_norm": 0.2988564937076045,
+      "learning_rate": 2.2929016062037057e-06,
+      "loss": 0.2566,
+      "step": 1064
+    },
+    {
+      "epoch": 4.3117408906882595,
+      "grad_norm": 0.2883650194642767,
+      "learning_rate": 2.2666794376975055e-06,
+      "loss": 0.2601,
+      "step": 1065
+    },
+    {
+      "epoch": 4.315789473684211,
+      "grad_norm": 0.28937480777134594,
+      "learning_rate": 2.24059906434998e-06,
+      "loss": 0.2574,
+      "step": 1066
+    },
+    {
+      "epoch": 4.319838056680162,
+      "grad_norm": 0.299442669232713,
+      "learning_rate": 2.214660694699091e-06,
+      "loss": 0.2589,
+      "step": 1067
+    },
+    {
+      "epoch": 4.323886639676114,
+      "grad_norm": 0.2843670023183634,
+      "learning_rate": 2.188864536147355e-06,
+      "loss": 0.2639,
+      "step": 1068
+    },
+    {
+      "epoch": 4.327935222672065,
+      "grad_norm": 0.2802005861305596,
+      "learning_rate": 2.1632107949601757e-06,
+      "loss": 0.2698,
+      "step": 1069
+    },
+    {
+      "epoch": 4.331983805668016,
+      "grad_norm": 0.30456593240908925,
+      "learning_rate": 2.137699676264171e-06,
+      "loss": 0.2642,
+      "step": 1070
+    },
+    {
+      "epoch": 4.336032388663968,
+      "grad_norm": 0.28597444147186024,
+      "learning_rate": 2.1123313840455738e-06,
+      "loss": 0.257,
+      "step": 1071
+    },
+    {
+      "epoch": 4.340080971659919,
+      "grad_norm": 0.2758235144378006,
+      "learning_rate": 2.08710612114857e-06,
+      "loss": 0.2551,
+      "step": 1072
+    },
+    {
+      "epoch": 4.34412955465587,
+      "grad_norm": 0.2880705855360486,
+      "learning_rate": 2.0620240892736953e-06,
+      "loss": 0.2479,
+      "step": 1073
+    },
+    {
+      "epoch": 4.348178137651822,
+      "grad_norm": 0.276330421847386,
+      "learning_rate": 2.0370854889762e-06,
+      "loss": 0.2686,
+      "step": 1074
+    },
+    {
+      "epoch": 4.352226720647773,
+      "grad_norm": 0.2799368804504245,
+      "learning_rate": 2.0122905196644726e-06,
+      "loss": 0.2501,
+      "step": 1075
+    },
+    {
+      "epoch": 4.3562753036437245,
+      "grad_norm": 0.29985425493801965,
+      "learning_rate": 1.987639379598427e-06,
+      "loss": 0.2529,
+      "step": 1076
+    },
+    {
+      "epoch": 4.3603238866396765,
+      "grad_norm": 0.28769921911970436,
+      "learning_rate": 1.963132265887919e-06,
+      "loss": 0.2533,
+      "step": 1077
+    },
+    {
+      "epoch": 4.364372469635628,
+      "grad_norm": 0.2752317883613011,
+      "learning_rate": 1.9387693744911784e-06,
+      "loss": 0.2584,
+      "step": 1078
+    },
+    {
+      "epoch": 4.368421052631579,
+      "grad_norm": 0.28532490818549555,
+      "learning_rate": 1.9145509002132368e-06,
+      "loss": 0.2631,
+      "step": 1079
+    },
+    {
+      "epoch": 4.372469635627531,
+      "grad_norm": 0.28486492489545434,
+      "learning_rate": 1.8904770367043723e-06,
+      "loss": 0.2512,
+      "step": 1080
+    },
+    {
+      "epoch": 4.376518218623482,
+      "grad_norm": 0.27980583420419836,
+      "learning_rate": 1.8665479764585504e-06,
+      "loss": 0.2653,
+      "step": 1081
+    },
+    {
+      "epoch": 4.380566801619433,
+      "grad_norm": 0.2856860824416601,
+      "learning_rate": 1.8427639108118977e-06,
+      "loss": 0.2568,
+      "step": 1082
+    },
+    {
+      "epoch": 4.384615384615385,
+      "grad_norm": 0.2841745646865941,
+      "learning_rate": 1.8191250299411755e-06,
+      "loss": 0.2613,
+      "step": 1083
+    },
+    {
+      "epoch": 4.388663967611336,
+      "grad_norm": 0.2888632463192244,
+      "learning_rate": 1.7956315228622446e-06,
+      "loss": 0.2728,
+      "step": 1084
+    },
+    {
+      "epoch": 4.392712550607287,
+      "grad_norm": 0.29005322972861086,
+      "learning_rate": 1.7722835774285553e-06,
+      "loss": 0.2597,
+      "step": 1085
+    },
+    {
+      "epoch": 4.396761133603239,
+      "grad_norm": 0.278413957513732,
+      "learning_rate": 1.7490813803296624e-06,
+      "loss": 0.2569,
+      "step": 1086
+    },
+    {
+      "epoch": 4.40080971659919,
+      "grad_norm": 0.2857495038214909,
+      "learning_rate": 1.726025117089718e-06,
+      "loss": 0.2653,
+      "step": 1087
+    },
+    {
+      "epoch": 4.4048582995951415,
+      "grad_norm": 0.28385834790138437,
+      "learning_rate": 1.70311497206598e-06,
+      "loss": 0.2676,
+      "step": 1088
+    },
+    {
+      "epoch": 4.4089068825910935,
+      "grad_norm": 0.28376532897565215,
+      "learning_rate": 1.6803511284473638e-06,
+      "loss": 0.2553,
+      "step": 1089
+    },
+    {
+      "epoch": 4.412955465587045,
+      "grad_norm": 0.29456436945011955,
+      "learning_rate": 1.6577337682529538e-06,
+      "loss": 0.2475,
+      "step": 1090
+    },
+    {
+      "epoch": 4.417004048582996,
+      "grad_norm": 0.2830497102149869,
+      "learning_rate": 1.6352630723305617e-06,
+      "loss": 0.2627,
+      "step": 1091
+    },
+    {
+      "epoch": 4.421052631578947,
+      "grad_norm": 0.3021592680854813,
+      "learning_rate": 1.612939220355272e-06,
+      "loss": 0.2512,
+      "step": 1092
+    },
+    {
+      "epoch": 4.425101214574899,
+      "grad_norm": 0.27581533387724677,
+      "learning_rate": 1.5907623908280112e-06,
+      "loss": 0.2684,
+      "step": 1093
+    },
+    {
+      "epoch": 4.42914979757085,
+      "grad_norm": 0.2727719429300977,
+      "learning_rate": 1.568732761074121e-06,
+      "loss": 0.2617,
+      "step": 1094
+    },
+    {
+      "epoch": 4.433198380566802,
+      "grad_norm": 0.30298416636667364,
+      "learning_rate": 1.5468505072419237e-06,
+      "loss": 0.2399,
+      "step": 1095
+    },
+    {
+      "epoch": 4.437246963562753,
+      "grad_norm": 0.2691177917613172,
+      "learning_rate": 1.5251158043013448e-06,
+      "loss": 0.2628,
+      "step": 1096
+    },
+    {
+      "epoch": 4.441295546558704,
+      "grad_norm": 0.28361008902752677,
+      "learning_rate": 1.5035288260424885e-06,
+      "loss": 0.2555,
+      "step": 1097
+    },
+    {
+      "epoch": 4.445344129554655,
+      "grad_norm": 0.28790025861081114,
+      "learning_rate": 1.4820897450742623e-06,
+      "loss": 0.2623,
+      "step": 1098
+    },
+    {
+      "epoch": 4.449392712550607,
+      "grad_norm": 0.2834650913605142,
+      "learning_rate": 1.4607987328229767e-06,
+      "loss": 0.2661,
+      "step": 1099
+    },
+    {
+      "epoch": 4.4534412955465585,
+      "grad_norm": 0.30333452033247404,
+      "learning_rate": 1.4396559595310056e-06,
+      "loss": 0.232,
+      "step": 1100
+    },
+    {
+      "epoch": 4.4574898785425106,
+      "grad_norm": 0.2926001356041047,
+      "learning_rate": 1.4186615942553973e-06,
+      "loss": 0.2476,
+      "step": 1101
+    },
+    {
+      "epoch": 4.461538461538462,
+      "grad_norm": 0.27963038958736,
+      "learning_rate": 1.3978158048665424e-06,
+      "loss": 0.264,
+      "step": 1102
+    },
+    {
+      "epoch": 4.465587044534413,
+      "grad_norm": 0.3009687078059311,
+      "learning_rate": 1.3771187580468115e-06,
+      "loss": 0.242,
+      "step": 1103
+    },
+    {
+      "epoch": 4.469635627530364,
+      "grad_norm": 0.2860744595462306,
+      "learning_rate": 1.3565706192892392e-06,
+      "loss": 0.2592,
+      "step": 1104
+    },
+    {
+      "epoch": 4.473684210526316,
+      "grad_norm": 0.2962659783667669,
+      "learning_rate": 1.3361715528961993e-06,
+      "loss": 0.2539,
+      "step": 1105
+    },
+    {
+      "epoch": 4.477732793522267,
+      "grad_norm": 0.2866429188296955,
+      "learning_rate": 1.3159217219780817e-06,
+      "loss": 0.2533,
+      "step": 1106
+    },
+    {
+      "epoch": 4.481781376518219,
+      "grad_norm": 0.2709857198264384,
+      "learning_rate": 1.2958212884519949e-06,
+      "loss": 0.2588,
+      "step": 1107
+    },
+    {
+      "epoch": 4.48582995951417,
+      "grad_norm": 0.2848899995497738,
+      "learning_rate": 1.2758704130404675e-06,
+      "loss": 0.2592,
+      "step": 1108
+    },
+    {
+      "epoch": 4.489878542510121,
+      "grad_norm": 0.28298702669036574,
+      "learning_rate": 1.2560692552701738e-06,
+      "loss": 0.2618,
+      "step": 1109
+    },
+    {
+      "epoch": 4.493927125506072,
+      "grad_norm": 0.2834581079271533,
+      "learning_rate": 1.2364179734706382e-06,
+      "loss": 0.2658,
+      "step": 1110
+    },
+    {
+      "epoch": 4.497975708502024,
+      "grad_norm": 0.28753252468038404,
+      "learning_rate": 1.2169167247729852e-06,
+      "loss": 0.2686,
+      "step": 1111
+    },
+    {
+      "epoch": 4.502024291497976,
+      "grad_norm": 0.27488950321867517,
+      "learning_rate": 1.1975656651086864e-06,
+      "loss": 0.2787,
+      "step": 1112
+    },
+    {
+      "epoch": 4.506072874493928,
+      "grad_norm": 0.28134495296656,
+      "learning_rate": 1.1783649492082926e-06,
+      "loss": 0.2535,
+      "step": 1113
+    },
+    {
+      "epoch": 4.510121457489879,
+      "grad_norm": 0.2727274915844921,
+      "learning_rate": 1.1593147306002183e-06,
+      "loss": 0.2672,
+      "step": 1114
+    },
+    {
+      "epoch": 4.51417004048583,
+      "grad_norm": 0.2755878346276235,
+      "learning_rate": 1.140415161609505e-06,
+      "loss": 0.272,
+      "step": 1115
+    },
+    {
+      "epoch": 4.518218623481781,
+      "grad_norm": 0.2621179099883972,
+      "learning_rate": 1.121666393356604e-06,
+      "loss": 0.2668,
+      "step": 1116
+    },
+    {
+      "epoch": 4.522267206477733,
+      "grad_norm": 0.28353116176937304,
+      "learning_rate": 1.1030685757561632e-06,
+      "loss": 0.2458,
+      "step": 1117
+    },
+    {
+      "epoch": 4.526315789473684,
+      "grad_norm": 0.2860275497103349,
+      "learning_rate": 1.0846218575158373e-06,
+      "loss": 0.2539,
+      "step": 1118
+    },
+    {
+      "epoch": 4.530364372469636,
+      "grad_norm": 0.279229673080097,
+      "learning_rate": 1.066326386135097e-06,
+      "loss": 0.2659,
+      "step": 1119
+    },
+    {
+      "epoch": 4.534412955465587,
+      "grad_norm": 0.262862870744325,
+      "learning_rate": 1.0481823079040398e-06,
+      "loss": 0.2642,
+      "step": 1120
+    },
+    {
+      "epoch": 4.538461538461538,
+      "grad_norm": 0.286269105046779,
+      "learning_rate": 1.0301897679022344e-06,
+      "loss": 0.2546,
+      "step": 1121
+    },
+    {
+      "epoch": 4.5425101214574894,
+      "grad_norm": 0.27321970291089964,
+      "learning_rate": 1.0123489099975491e-06,
+      "loss": 0.2717,
+      "step": 1122
+    },
+    {
+      "epoch": 4.5465587044534415,
+      "grad_norm": 0.2819317399056701,
+      "learning_rate": 9.9465987684501e-07,
+      "loss": 0.2691,
+      "step": 1123
+    },
+    {
+      "epoch": 4.550607287449393,
+      "grad_norm": 0.27635222506144425,
+      "learning_rate": 9.771228098856534e-07,
+      "loss": 0.2783,
+      "step": 1124
+    },
+    {
+      "epoch": 4.554655870445345,
+      "grad_norm": 0.2717148399128772,
+      "learning_rate": 9.597378493454013e-07,
+      "loss": 0.2541,
+      "step": 1125
+    },
+    {
+      "epoch": 4.558704453441296,
+      "grad_norm": 0.28589946207746464,
+      "learning_rate": 9.425051342339353e-07,
+      "loss": 0.258,
+      "step": 1126
+    },
+    {
+      "epoch": 4.562753036437247,
+      "grad_norm": 0.28736973194789095,
+      "learning_rate": 9.254248023435864e-07,
+      "loss": 0.2614,
+      "step": 1127
+    },
+    {
+      "epoch": 4.566801619433198,
+      "grad_norm": 0.2783156761902675,
+      "learning_rate": 9.084969902482288e-07,
+      "loss": 0.2541,
+      "step": 1128
+    },
+    {
+      "epoch": 4.57085020242915,
+      "grad_norm": 0.2853971983068151,
+      "learning_rate": 8.917218333021993e-07,
+      "loss": 0.2512,
+      "step": 1129
+    },
+    {
+      "epoch": 4.574898785425101,
+      "grad_norm": 0.28023454203977455,
+      "learning_rate": 8.750994656392109e-07,
+      "loss": 0.2417,
+      "step": 1130
+    },
+    {
+      "epoch": 4.578947368421053,
+      "grad_norm": 0.28781669707411467,
+      "learning_rate": 8.58630020171265e-07,
+      "loss": 0.262,
+      "step": 1131
+    },
+    {
+      "epoch": 4.582995951417004,
+      "grad_norm": 0.2844633966276769,
+      "learning_rate": 8.423136285876099e-07,
+      "loss": 0.2523,
+      "step": 1132
+    },
+    {
+      "epoch": 4.587044534412955,
+      "grad_norm": 0.2842511922343382,
+      "learning_rate": 8.26150421353682e-07,
+      "loss": 0.2649,
+      "step": 1133
+    },
+    {
+      "epoch": 4.5910931174089065,
+      "grad_norm": 0.2824989393563515,
+      "learning_rate": 8.101405277100549e-07,
+      "loss": 0.2558,
+      "step": 1134
+    },
+    {
+      "epoch": 4.5951417004048585,
+      "grad_norm": 0.2764562718972228,
+      "learning_rate": 7.942840756714077e-07,
+      "loss": 0.2697,
+      "step": 1135
+    },
+    {
+      "epoch": 4.59919028340081,
+      "grad_norm": 0.26153342095073,
+      "learning_rate": 7.785811920255093e-07,
+      "loss": 0.2743,
+      "step": 1136
+    },
+    {
+      "epoch": 4.603238866396762,
+      "grad_norm": 0.2781473609428542,
+      "learning_rate": 7.630320023322024e-07,
+      "loss": 0.2655,
+      "step": 1137
+    },
+    {
+      "epoch": 4.607287449392713,
+      "grad_norm": 0.2747423400811156,
+      "learning_rate": 7.476366309223903e-07,
+      "loss": 0.2687,
+      "step": 1138
+    },
+    {
+      "epoch": 4.611336032388664,
+      "grad_norm": 0.2990045370458084,
+      "learning_rate": 7.323952008970537e-07,
+      "loss": 0.2573,
+      "step": 1139
+    },
+    {
+      "epoch": 4.615384615384615,
+      "grad_norm": 0.28734427618173936,
+      "learning_rate": 7.173078341262618e-07,
+      "loss": 0.2447,
+      "step": 1140
+    },
+    {
+      "epoch": 4.619433198380567,
+      "grad_norm": 0.2928657821916152,
+      "learning_rate": 7.023746512481989e-07,
+      "loss": 0.2536,
+      "step": 1141
+    },
+    {
+      "epoch": 4.623481781376518,
+      "grad_norm": 0.28580376359874143,
+      "learning_rate": 6.875957716681902e-07,
+      "loss": 0.2487,
+      "step": 1142
+    },
+    {
+      "epoch": 4.62753036437247,
+      "grad_norm": 0.28655686989907114,
+      "learning_rate": 6.729713135577687e-07,
+      "loss": 0.2634,
+      "step": 1143
+    },
+    {
+      "epoch": 4.631578947368421,
+      "grad_norm": 0.2797077329446172,
+      "learning_rate": 6.585013938537143e-07,
+      "loss": 0.2771,
+      "step": 1144
+    },
+    {
+      "epoch": 4.635627530364372,
+      "grad_norm": 0.2701144555023815,
+      "learning_rate": 6.441861282571115e-07,
+      "loss": 0.2682,
+      "step": 1145
+    },
+    {
+      "epoch": 4.6396761133603235,
+      "grad_norm": 0.2749601889023287,
+      "learning_rate": 6.300256312324471e-07,
+      "loss": 0.2616,
+      "step": 1146
+    },
+    {
+      "epoch": 4.6437246963562755,
+      "grad_norm": 0.27896941884318444,
+      "learning_rate": 6.160200160066776e-07,
+      "loss": 0.2699,
+      "step": 1147
+    },
+    {
+      "epoch": 4.647773279352227,
+      "grad_norm": 0.27794625602023393,
+      "learning_rate": 6.021693945683327e-07,
+      "loss": 0.2695,
+      "step": 1148
+    },
+    {
+      "epoch": 4.651821862348179,
+      "grad_norm": 0.2701206039437078,
+      "learning_rate": 5.884738776666044e-07,
+      "loss": 0.2557,
+      "step": 1149
+    },
+    {
+      "epoch": 4.65587044534413,
+      "grad_norm": 0.2826816724460482,
+      "learning_rate": 5.749335748104878e-07,
+      "loss": 0.2651,
+      "step": 1150
+    },
+    {
+      "epoch": 4.659919028340081,
+      "grad_norm": 0.28234367377791575,
+      "learning_rate": 5.615485942678778e-07,
+      "loss": 0.2652,
+      "step": 1151
+    },
+    {
+      "epoch": 4.663967611336032,
+      "grad_norm": 0.27370741526766307,
+      "learning_rate": 5.483190430647289e-07,
+      "loss": 0.2693,
+      "step": 1152
+    },
+    {
+      "epoch": 4.668016194331984,
+      "grad_norm": 0.2774305979979346,
+      "learning_rate": 5.352450269841747e-07,
+      "loss": 0.2535,
+      "step": 1153
+    },
+    {
+      "epoch": 4.672064777327935,
+      "grad_norm": 0.27128434870305695,
+      "learning_rate": 5.22326650565701e-07,
+      "loss": 0.2719,
+      "step": 1154
+    },
+    {
+      "epoch": 4.676113360323887,
+      "grad_norm": 0.27355210923712264,
+      "learning_rate": 5.095640171043048e-07,
+      "loss": 0.2824,
+      "step": 1155
+    },
+    {
+      "epoch": 4.680161943319838,
+      "grad_norm": 0.29033019099957114,
+      "learning_rate": 4.969572286496594e-07,
+      "loss": 0.2402,
+      "step": 1156
+    },
+    {
+      "epoch": 4.684210526315789,
+      "grad_norm": 0.28291288348945154,
+      "learning_rate": 4.845063860053079e-07,
+      "loss": 0.2563,
+      "step": 1157
+    },
+    {
+      "epoch": 4.6882591093117405,
+      "grad_norm": 0.276557500779358,
+      "learning_rate": 4.7221158872785997e-07,
+      "loss": 0.2517,
+      "step": 1158
+    },
+    {
+      "epoch": 4.6923076923076925,
+      "grad_norm": 0.2774099080579215,
+      "learning_rate": 4.600729351261857e-07,
+      "loss": 0.2688,
+      "step": 1159
+    },
+    {
+      "epoch": 4.696356275303644,
+      "grad_norm": 0.2699201303452072,
+      "learning_rate": 4.4809052226063355e-07,
+      "loss": 0.2722,
+      "step": 1160
+    },
+    {
+      "epoch": 4.700404858299595,
+      "grad_norm": 0.26896907506885176,
+      "learning_rate": 4.3626444594226046e-07,
+      "loss": 0.2773,
+      "step": 1161
+    },
+    {
+      "epoch": 4.704453441295547,
+      "grad_norm": 0.2736400915015439,
+      "learning_rate": 4.245948007320588e-07,
+      "loss": 0.2646,
+      "step": 1162
+    },
+    {
+      "epoch": 4.708502024291498,
+      "grad_norm": 0.26992270100390586,
+      "learning_rate": 4.130816799401971e-07,
+      "loss": 0.2534,
+      "step": 1163
+    },
+    {
+      "epoch": 4.712550607287449,
+      "grad_norm": 0.26912032005972425,
+      "learning_rate": 4.017251756252827e-07,
+      "loss": 0.2643,
+      "step": 1164
+    },
+    {
+      "epoch": 4.716599190283401,
+      "grad_norm": 0.2980166532306395,
+      "learning_rate": 3.9052537859362473e-07,
+      "loss": 0.258,
+      "step": 1165
+    },
+    {
+      "epoch": 4.720647773279352,
+      "grad_norm": 0.27770289526034575,
+      "learning_rate": 3.7948237839849687e-07,
+      "loss": 0.252,
+      "step": 1166
+    },
+    {
+      "epoch": 4.724696356275303,
+      "grad_norm": 0.2802608758655415,
+      "learning_rate": 3.6859626333943797e-07,
+      "loss": 0.2572,
+      "step": 1167
+    },
+    {
+      "epoch": 4.728744939271255,
+      "grad_norm": 0.2723910855014645,
+      "learning_rate": 3.578671204615347e-07,
+      "loss": 0.2545,
+      "step": 1168
+    },
+    {
+      "epoch": 4.732793522267206,
+      "grad_norm": 0.2863276119417557,
+      "learning_rate": 3.472950355547267e-07,
+      "loss": 0.2505,
+      "step": 1169
+    },
+    {
+      "epoch": 4.7368421052631575,
+      "grad_norm": 0.2894243949738839,
+      "learning_rate": 3.3688009315312244e-07,
+      "loss": 0.2525,
+      "step": 1170
+    },
+    {
+      "epoch": 4.7408906882591095,
+      "grad_norm": 0.2853453785197192,
+      "learning_rate": 3.266223765343224e-07,
+      "loss": 0.2446,
+      "step": 1171
+    },
+    {
+      "epoch": 4.744939271255061,
+      "grad_norm": 0.27160242009398883,
+      "learning_rate": 3.16521967718757e-07,
+      "loss": 0.2597,
+      "step": 1172
+    },
+    {
+      "epoch": 4.748987854251012,
+      "grad_norm": 0.2886551473651144,
+      "learning_rate": 3.0657894746902726e-07,
+      "loss": 0.2514,
+      "step": 1173
+    },
+    {
+      "epoch": 4.753036437246964,
+      "grad_norm": 0.2819591152091504,
+      "learning_rate": 2.9679339528925654e-07,
+      "loss": 0.2582,
+      "step": 1174
+    },
+    {
+      "epoch": 4.757085020242915,
+      "grad_norm": 0.2752485696008143,
+      "learning_rate": 2.871653894244597e-07,
+      "loss": 0.2567,
+      "step": 1175
+    },
+    {
+      "epoch": 4.761133603238866,
+      "grad_norm": 0.27764649207765935,
+      "learning_rate": 2.7769500685991266e-07,
+      "loss": 0.2622,
+      "step": 1176
+    },
+    {
+      "epoch": 4.765182186234818,
+      "grad_norm": 0.28484087512641104,
+      "learning_rate": 2.683823233205485e-07,
+      "loss": 0.2571,
+      "step": 1177
+    },
+    {
+      "epoch": 4.769230769230769,
+      "grad_norm": 0.2738176735265105,
+      "learning_rate": 2.592274132703287e-07,
+      "loss": 0.2675,
+      "step": 1178
+    },
+    {
+      "epoch": 4.77327935222672,
+      "grad_norm": 0.2725765755576917,
+      "learning_rate": 2.502303499116754e-07,
+      "loss": 0.2643,
+      "step": 1179
+    },
+    {
+      "epoch": 4.777327935222672,
+      "grad_norm": 0.2792053561897689,
+      "learning_rate": 2.413912051848666e-07,
+      "loss": 0.2597,
+      "step": 1180
+    },
+    {
+      "epoch": 4.781376518218623,
+      "grad_norm": 0.27368115275347993,
+      "learning_rate": 2.3271004976746613e-07,
+      "loss": 0.2607,
+      "step": 1181
+    },
+    {
+      "epoch": 4.7854251012145745,
+      "grad_norm": 0.2763573587886316,
+      "learning_rate": 2.2418695307376393e-07,
+      "loss": 0.2616,
+      "step": 1182
+    },
+    {
+      "epoch": 4.7894736842105265,
+      "grad_norm": 0.2645702958528518,
+      "learning_rate": 2.1582198325420968e-07,
+      "loss": 0.2644,
+      "step": 1183
+    },
+    {
+      "epoch": 4.793522267206478,
+      "grad_norm": 0.271631321698289,
+      "learning_rate": 2.0761520719488447e-07,
+      "loss": 0.265,
+      "step": 1184
+    },
+    {
+      "epoch": 4.797570850202429,
+      "grad_norm": 0.271397041040225,
+      "learning_rate": 1.9956669051694798e-07,
+      "loss": 0.262,
+      "step": 1185
+    },
+    {
+      "epoch": 4.801619433198381,
+      "grad_norm": 0.2810651980416332,
+      "learning_rate": 1.916764975761254e-07,
+      "loss": 0.2609,
+      "step": 1186
+    },
+    {
+      "epoch": 4.805668016194332,
+      "grad_norm": 0.2855918502903814,
+      "learning_rate": 1.839446914621923e-07,
+      "loss": 0.2551,
+      "step": 1187
+    },
+    {
+      "epoch": 4.809716599190283,
+      "grad_norm": 0.28213721542454906,
+      "learning_rate": 1.7637133399845962e-07,
+      "loss": 0.2612,
+      "step": 1188
+    },
+    {
+      "epoch": 4.813765182186235,
+      "grad_norm": 0.27644619145286414,
+      "learning_rate": 1.6895648574129397e-07,
+      "loss": 0.264,
+      "step": 1189
+    },
+    {
+      "epoch": 4.817813765182186,
+      "grad_norm": 0.2842087052460375,
+      "learning_rate": 1.6170020597962465e-07,
+      "loss": 0.2753,
+      "step": 1190
+    },
+    {
+      "epoch": 4.821862348178137,
+      "grad_norm": 0.2822942750877723,
+      "learning_rate": 1.5460255273447078e-07,
+      "loss": 0.2623,
+      "step": 1191
+    },
+    {
+      "epoch": 4.825910931174089,
+      "grad_norm": 0.28606612555693567,
+      "learning_rate": 1.4766358275847715e-07,
+      "loss": 0.2679,
+      "step": 1192
+    },
+    {
+      "epoch": 4.82995951417004,
+      "grad_norm": 0.2809521177436041,
+      "learning_rate": 1.4088335153546573e-07,
+      "loss": 0.2544,
+      "step": 1193
+    },
+    {
+      "epoch": 4.834008097165992,
+      "grad_norm": 0.2900160608945642,
+      "learning_rate": 1.3426191327998496e-07,
+      "loss": 0.2499,
+      "step": 1194
+    },
+    {
+      "epoch": 4.838056680161944,
+      "grad_norm": 0.27306622630773364,
+      "learning_rate": 1.277993209368744e-07,
+      "loss": 0.278,
+      "step": 1195
+    },
+    {
+      "epoch": 4.842105263157895,
+      "grad_norm": 0.3008547735900923,
+      "learning_rate": 1.214956261808542e-07,
+      "loss": 0.253,
+      "step": 1196
+    },
+    {
+      "epoch": 4.846153846153846,
+      "grad_norm": 0.2627502777932794,
+      "learning_rate": 1.1535087941609846e-07,
+      "loss": 0.2681,
+      "step": 1197
+    },
+    {
+      "epoch": 4.850202429149798,
+      "grad_norm": 0.2796632582165713,
+      "learning_rate": 1.0936512977583802e-07,
+      "loss": 0.2695,
+      "step": 1198
+    },
+    {
+      "epoch": 4.854251012145749,
+      "grad_norm": 0.26821493468957125,
+      "learning_rate": 1.0353842512196732e-07,
+      "loss": 0.2648,
+      "step": 1199
+    },
+    {
+      "epoch": 4.8582995951417,
+      "grad_norm": 0.2754139623019979,
+      "learning_rate": 9.787081204466253e-08,
+      "loss": 0.2494,
+      "step": 1200
+    },
+    {
+      "epoch": 4.862348178137652,
+      "grad_norm": 0.27545342218758334,
+      "learning_rate": 9.236233586200627e-08,
+      "loss": 0.2619,
+      "step": 1201
+    },
+    {
+      "epoch": 4.866396761133603,
+      "grad_norm": 0.28094360785859407,
+      "learning_rate": 8.70130406196279e-08,
+      "loss": 0.2677,
+      "step": 1202
+    },
+    {
+      "epoch": 4.870445344129554,
+      "grad_norm": 0.26068538542082687,
+      "learning_rate": 8.182296909035047e-08,
+      "loss": 0.2792,
+      "step": 1203
+    },
+    {
+      "epoch": 4.874493927125506,
+      "grad_norm": 0.2694676458814123,
+      "learning_rate": 7.679216277384439e-08,
+      "loss": 0.2699,
+      "step": 1204
+    },
+    {
+      "epoch": 4.8785425101214575,
+      "grad_norm": 0.26232310948961385,
+      "learning_rate": 7.19206618963053e-08,
+      "loss": 0.2688,
+      "step": 1205
+    },
+    {
+      "epoch": 4.882591093117409,
+      "grad_norm": 0.27764777121778395,
+      "learning_rate": 6.720850541012347e-08,
+      "loss": 0.2678,
+      "step": 1206
+    },
+    {
+      "epoch": 4.886639676113361,
+      "grad_norm": 0.2866645584016821,
+      "learning_rate": 6.265573099357714e-08,
+      "loss": 0.267,
+      "step": 1207
+    },
+    {
+      "epoch": 4.890688259109312,
+      "grad_norm": 0.2748588317327751,
+      "learning_rate": 5.826237505053067e-08,
+      "loss": 0.2606,
+      "step": 1208
+    },
+    {
+      "epoch": 4.894736842105263,
+      "grad_norm": 0.2813428168202467,
+      "learning_rate": 5.4028472710137e-08,
+      "loss": 0.2618,
+      "step": 1209
+    },
+    {
+      "epoch": 4.898785425101215,
+      "grad_norm": 0.2719853081615888,
+      "learning_rate": 4.9954057826571146e-08,
+      "loss": 0.2604,
+      "step": 1210
+    },
+    {
+      "epoch": 4.902834008097166,
+      "grad_norm": 0.2690033060646696,
+      "learning_rate": 4.6039162978743776e-08,
+      "loss": 0.2614,
+      "step": 1211
+    },
+    {
+      "epoch": 4.906882591093117,
+      "grad_norm": 0.2609241078784742,
+      "learning_rate": 4.228381947004812e-08,
+      "loss": 0.2629,
+      "step": 1212
+    },
+    {
+      "epoch": 4.910931174089069,
+      "grad_norm": 0.28280551464157616,
+      "learning_rate": 3.868805732811343e-08,
+      "loss": 0.2514,
+      "step": 1213
+    },
+    {
+      "epoch": 4.91497975708502,
+      "grad_norm": 0.26964254554549627,
+      "learning_rate": 3.525190530455636e-08,
+      "loss": 0.2677,
+      "step": 1214
+    },
+    {
+      "epoch": 4.919028340080971,
+      "grad_norm": 0.29272546790340975,
+      "learning_rate": 3.197539087475887e-08,
+      "loss": 0.2546,
+      "step": 1215
+    },
+    {
+      "epoch": 4.923076923076923,
+      "grad_norm": 0.2755735528010214,
+      "learning_rate": 2.8858540237641786e-08,
+      "loss": 0.2572,
+      "step": 1216
+    },
+    {
+      "epoch": 4.9271255060728745,
+      "grad_norm": 0.27178291930623627,
+      "learning_rate": 2.5901378315458247e-08,
+      "loss": 0.2589,
+      "step": 1217
+    },
+    {
+      "epoch": 4.931174089068826,
+      "grad_norm": 0.28460084824112,
+      "learning_rate": 2.3103928753598347e-08,
+      "loss": 0.2551,
+      "step": 1218
+    },
+    {
+      "epoch": 4.935222672064778,
+      "grad_norm": 0.27359481456785667,
+      "learning_rate": 2.0466213920393718e-08,
+      "loss": 0.249,
+      "step": 1219
+    },
+    {
+      "epoch": 4.939271255060729,
+      "grad_norm": 0.2840456451869775,
+      "learning_rate": 1.7988254906942115e-08,
+      "loss": 0.2661,
+      "step": 1220
+    },
+    {
+      "epoch": 4.94331983805668,
+      "grad_norm": 0.28553723342633675,
+      "learning_rate": 1.5670071526936447e-08,
+      "loss": 0.2507,
+      "step": 1221
+    },
+    {
+      "epoch": 4.947368421052632,
+      "grad_norm": 0.2730335160364624,
+      "learning_rate": 1.3511682316509344e-08,
+      "loss": 0.2647,
+      "step": 1222
+    },
+    {
+      "epoch": 4.951417004048583,
+      "grad_norm": 0.28992281897144445,
+      "learning_rate": 1.1513104534086605e-08,
+      "loss": 0.2593,
+      "step": 1223
+    },
+    {
+      "epoch": 4.955465587044534,
+      "grad_norm": 0.26512584273950307,
+      "learning_rate": 9.67435416023843e-09,
+      "loss": 0.2505,
+      "step": 1224
+    },
+    {
+      "epoch": 4.959514170040486,
+      "grad_norm": 0.27456236465568334,
+      "learning_rate": 7.995445897566178e-09,
+      "loss": 0.2582,
+      "step": 1225
+    },
+    {
+      "epoch": 4.963562753036437,
+      "grad_norm": 0.28123386491401337,
+      "learning_rate": 6.476393170578021e-09,
+      "loss": 0.2582,
+      "step": 1226
+    },
+    {
+      "epoch": 4.967611336032388,
+      "grad_norm": 0.2769024740773072,
+      "learning_rate": 5.11720812557348e-09,
+      "loss": 0.2624,
+      "step": 1227
+    },
+    {
+      "epoch": 4.97165991902834,
+      "grad_norm": 0.27082846515665815,
+      "learning_rate": 3.917901630563492e-09,
+      "loss": 0.2623,
+      "step": 1228
+    },
+    {
+      "epoch": 4.9757085020242915,
+      "grad_norm": 0.2779711412698715,
+      "learning_rate": 2.878483275170485e-09,
+      "loss": 0.2536,
+      "step": 1229
+    },
+    {
+      "epoch": 4.979757085020243,
+      "grad_norm": 0.28334700585203515,
+      "learning_rate": 1.998961370555108e-09,
+      "loss": 0.2598,
+      "step": 1230
+    },
+    {
+      "epoch": 4.983805668016195,
+      "grad_norm": 0.2703159924343393,
+      "learning_rate": 1.2793429493518361e-09,
+      "loss": 0.2649,
+      "step": 1231
+    },
+    {
+      "epoch": 4.987854251012146,
+      "grad_norm": 0.2751311517716602,
+      "learning_rate": 7.196337656112384e-10,
+      "loss": 0.2573,
+      "step": 1232
+    },
+    {
+      "epoch": 4.991902834008097,
+      "grad_norm": 0.2795343950642818,
+      "learning_rate": 3.198382947511292e-10,
+      "loss": 0.2533,
+      "step": 1233
+    },
+    {
+      "epoch": 4.995951417004049,
+      "grad_norm": 0.27442587879261976,
+      "learning_rate": 7.995973352770137e-11,
+      "loss": 0.2629,
+      "step": 1234
+    },
+    {
+      "epoch": 5.0,
+      "grad_norm": 0.2684733005610841,
+      "learning_rate": 0.0,
+      "loss": 0.258,
+      "step": 1235
+    },
+    {
+      "epoch": 5.0,
+      "step": 1235,
+      "total_flos": 1975292565913600.0,
+      "train_loss": 0.43101271609304403,
+      "train_runtime": 24851.894,
+      "train_samples_per_second": 6.358,
+      "train_steps_per_second": 0.05
+    }
+  ],
+  "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": 1975292565913600.0,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}