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": 6.631874662665014,
+      "learning_rate": 3.2258064516129035e-07,
+      "loss": 1.0385,
+      "step": 1
+    },
+    {
+      "epoch": 0.008097165991902834,
+      "grad_norm": 6.548507451360709,
+      "learning_rate": 6.451612903225807e-07,
+      "loss": 1.031,
+      "step": 2
+    },
+    {
+      "epoch": 0.012145748987854251,
+      "grad_norm": 6.490652920011975,
+      "learning_rate": 9.67741935483871e-07,
+      "loss": 1.0274,
+      "step": 3
+    },
+    {
+      "epoch": 0.016194331983805668,
+      "grad_norm": 6.36769162757928,
+      "learning_rate": 1.2903225806451614e-06,
+      "loss": 1.0245,
+      "step": 4
+    },
+    {
+      "epoch": 0.020242914979757085,
+      "grad_norm": 6.23528934646902,
+      "learning_rate": 1.6129032258064516e-06,
+      "loss": 1.0345,
+      "step": 5
+    },
+    {
+      "epoch": 0.024291497975708502,
+      "grad_norm": 5.887338065196814,
+      "learning_rate": 1.935483870967742e-06,
+      "loss": 1.0123,
+      "step": 6
+    },
+    {
+      "epoch": 0.02834008097165992,
+      "grad_norm": 4.640378011816809,
+      "learning_rate": 2.2580645161290324e-06,
+      "loss": 1.0003,
+      "step": 7
+    },
+    {
+      "epoch": 0.032388663967611336,
+      "grad_norm": 4.2314702218746625,
+      "learning_rate": 2.580645161290323e-06,
+      "loss": 0.9679,
+      "step": 8
+    },
+    {
+      "epoch": 0.03643724696356275,
+      "grad_norm": 2.786076697136498,
+      "learning_rate": 2.903225806451613e-06,
+      "loss": 0.9506,
+      "step": 9
+    },
+    {
+      "epoch": 0.04048582995951417,
+      "grad_norm": 2.6122170481341973,
+      "learning_rate": 3.225806451612903e-06,
+      "loss": 0.9559,
+      "step": 10
+    },
+    {
+      "epoch": 0.044534412955465584,
+      "grad_norm": 2.5909832415667826,
+      "learning_rate": 3.548387096774194e-06,
+      "loss": 0.9578,
+      "step": 11
+    },
+    {
+      "epoch": 0.048582995951417005,
+      "grad_norm": 3.991787968254686,
+      "learning_rate": 3.870967741935484e-06,
+      "loss": 0.921,
+      "step": 12
+    },
+    {
+      "epoch": 0.05263157894736842,
+      "grad_norm": 4.287169637096368,
+      "learning_rate": 4.193548387096774e-06,
+      "loss": 0.9434,
+      "step": 13
+    },
+    {
+      "epoch": 0.05668016194331984,
+      "grad_norm": 4.085561535862964,
+      "learning_rate": 4.516129032258065e-06,
+      "loss": 0.927,
+      "step": 14
+    },
+    {
+      "epoch": 0.06072874493927125,
+      "grad_norm": 3.608674084571379,
+      "learning_rate": 4.838709677419355e-06,
+      "loss": 0.9074,
+      "step": 15
+    },
+    {
+      "epoch": 0.06477732793522267,
+      "grad_norm": 2.788527818629598,
+      "learning_rate": 5.161290322580646e-06,
+      "loss": 0.8843,
+      "step": 16
+    },
+    {
+      "epoch": 0.06882591093117409,
+      "grad_norm": 2.6058068756353228,
+      "learning_rate": 5.483870967741935e-06,
+      "loss": 0.872,
+      "step": 17
+    },
+    {
+      "epoch": 0.0728744939271255,
+      "grad_norm": 2.154474479898299,
+      "learning_rate": 5.806451612903226e-06,
+      "loss": 0.8591,
+      "step": 18
+    },
+    {
+      "epoch": 0.07692307692307693,
+      "grad_norm": 1.5020146339668194,
+      "learning_rate": 6.129032258064517e-06,
+      "loss": 0.829,
+      "step": 19
+    },
+    {
+      "epoch": 0.08097165991902834,
+      "grad_norm": 1.0742321918843796,
+      "learning_rate": 6.451612903225806e-06,
+      "loss": 0.8281,
+      "step": 20
+    },
+    {
+      "epoch": 0.08502024291497975,
+      "grad_norm": 1.031856161098676,
+      "learning_rate": 6.774193548387097e-06,
+      "loss": 0.8231,
+      "step": 21
+    },
+    {
+      "epoch": 0.08906882591093117,
+      "grad_norm": 1.2044088408306228,
+      "learning_rate": 7.096774193548388e-06,
+      "loss": 0.8003,
+      "step": 22
+    },
+    {
+      "epoch": 0.0931174089068826,
+      "grad_norm": 1.0627097865039064,
+      "learning_rate": 7.4193548387096784e-06,
+      "loss": 0.7886,
+      "step": 23
+    },
+    {
+      "epoch": 0.09716599190283401,
+      "grad_norm": 0.9218379571288356,
+      "learning_rate": 7.741935483870968e-06,
+      "loss": 0.7891,
+      "step": 24
+    },
+    {
+      "epoch": 0.10121457489878542,
+      "grad_norm": 0.9023994707161116,
+      "learning_rate": 8.064516129032258e-06,
+      "loss": 0.7796,
+      "step": 25
+    },
+    {
+      "epoch": 0.10526315789473684,
+      "grad_norm": 0.9356721006100296,
+      "learning_rate": 8.387096774193549e-06,
+      "loss": 0.7659,
+      "step": 26
+    },
+    {
+      "epoch": 0.10931174089068826,
+      "grad_norm": 0.8795031500906507,
+      "learning_rate": 8.70967741935484e-06,
+      "loss": 0.757,
+      "step": 27
+    },
+    {
+      "epoch": 0.11336032388663968,
+      "grad_norm": 0.7710399208808013,
+      "learning_rate": 9.03225806451613e-06,
+      "loss": 0.7474,
+      "step": 28
+    },
+    {
+      "epoch": 0.11740890688259109,
+      "grad_norm": 0.8652771627665349,
+      "learning_rate": 9.35483870967742e-06,
+      "loss": 0.7344,
+      "step": 29
+    },
+    {
+      "epoch": 0.1214574898785425,
+      "grad_norm": 0.7944152226798468,
+      "learning_rate": 9.67741935483871e-06,
+      "loss": 0.7442,
+      "step": 30
+    },
+    {
+      "epoch": 0.12550607287449392,
+      "grad_norm": 0.7026662238856727,
+      "learning_rate": 1e-05,
+      "loss": 0.7422,
+      "step": 31
+    },
+    {
+      "epoch": 0.12955465587044535,
+      "grad_norm": 0.6718328165869512,
+      "learning_rate": 1.0322580645161291e-05,
+      "loss": 0.7378,
+      "step": 32
+    },
+    {
+      "epoch": 0.13360323886639677,
+      "grad_norm": 0.7074365082216153,
+      "learning_rate": 1.0645161290322582e-05,
+      "loss": 0.7364,
+      "step": 33
+    },
+    {
+      "epoch": 0.13765182186234817,
+      "grad_norm": 0.7821855361478934,
+      "learning_rate": 1.096774193548387e-05,
+      "loss": 0.7336,
+      "step": 34
+    },
+    {
+      "epoch": 0.1417004048582996,
+      "grad_norm": 0.6187419159372477,
+      "learning_rate": 1.1290322580645164e-05,
+      "loss": 0.7356,
+      "step": 35
+    },
+    {
+      "epoch": 0.145748987854251,
+      "grad_norm": 0.6381780371855188,
+      "learning_rate": 1.1612903225806453e-05,
+      "loss": 0.713,
+      "step": 36
+    },
+    {
+      "epoch": 0.14979757085020243,
+      "grad_norm": 0.6338857495262601,
+      "learning_rate": 1.1935483870967743e-05,
+      "loss": 0.7281,
+      "step": 37
+    },
+    {
+      "epoch": 0.15384615384615385,
+      "grad_norm": 0.5504629654896094,
+      "learning_rate": 1.2258064516129034e-05,
+      "loss": 0.7183,
+      "step": 38
+    },
+    {
+      "epoch": 0.15789473684210525,
+      "grad_norm": 0.5538160274480187,
+      "learning_rate": 1.2580645161290324e-05,
+      "loss": 0.7279,
+      "step": 39
+    },
+    {
+      "epoch": 0.16194331983805668,
+      "grad_norm": 0.6653781516639989,
+      "learning_rate": 1.2903225806451613e-05,
+      "loss": 0.702,
+      "step": 40
+    },
+    {
+      "epoch": 0.1659919028340081,
+      "grad_norm": 0.596645984853379,
+      "learning_rate": 1.3225806451612903e-05,
+      "loss": 0.7138,
+      "step": 41
+    },
+    {
+      "epoch": 0.1700404858299595,
+      "grad_norm": 0.5759995594123721,
+      "learning_rate": 1.3548387096774194e-05,
+      "loss": 0.7037,
+      "step": 42
+    },
+    {
+      "epoch": 0.17408906882591094,
+      "grad_norm": 0.6027342186967412,
+      "learning_rate": 1.3870967741935486e-05,
+      "loss": 0.7044,
+      "step": 43
+    },
+    {
+      "epoch": 0.17813765182186234,
+      "grad_norm": 0.4946628704201567,
+      "learning_rate": 1.4193548387096776e-05,
+      "loss": 0.7066,
+      "step": 44
+    },
+    {
+      "epoch": 0.18218623481781376,
+      "grad_norm": 0.6051051615718342,
+      "learning_rate": 1.4516129032258066e-05,
+      "loss": 0.7005,
+      "step": 45
+    },
+    {
+      "epoch": 0.1862348178137652,
+      "grad_norm": 0.5193206678096324,
+      "learning_rate": 1.4838709677419357e-05,
+      "loss": 0.7125,
+      "step": 46
+    },
+    {
+      "epoch": 0.1902834008097166,
+      "grad_norm": 0.5142375945344245,
+      "learning_rate": 1.5161290322580646e-05,
+      "loss": 0.7044,
+      "step": 47
+    },
+    {
+      "epoch": 0.19433198380566802,
+      "grad_norm": 0.6813531410165126,
+      "learning_rate": 1.5483870967741936e-05,
+      "loss": 0.6946,
+      "step": 48
+    },
+    {
+      "epoch": 0.19838056680161945,
+      "grad_norm": 0.5398925275114149,
+      "learning_rate": 1.5806451612903226e-05,
+      "loss": 0.6921,
+      "step": 49
+    },
+    {
+      "epoch": 0.20242914979757085,
+      "grad_norm": 0.5666659157632717,
+      "learning_rate": 1.6129032258064517e-05,
+      "loss": 0.6731,
+      "step": 50
+    },
+    {
+      "epoch": 0.20647773279352227,
+      "grad_norm": 0.4771307621048343,
+      "learning_rate": 1.6451612903225807e-05,
+      "loss": 0.6928,
+      "step": 51
+    },
+    {
+      "epoch": 0.21052631578947367,
+      "grad_norm": 0.6837680333965692,
+      "learning_rate": 1.6774193548387098e-05,
+      "loss": 0.6747,
+      "step": 52
+    },
+    {
+      "epoch": 0.2145748987854251,
+      "grad_norm": 0.6533283499023701,
+      "learning_rate": 1.7096774193548388e-05,
+      "loss": 0.6756,
+      "step": 53
+    },
+    {
+      "epoch": 0.21862348178137653,
+      "grad_norm": 0.6624354273074161,
+      "learning_rate": 1.741935483870968e-05,
+      "loss": 0.6833,
+      "step": 54
+    },
+    {
+      "epoch": 0.22267206477732793,
+      "grad_norm": 0.6640875341590745,
+      "learning_rate": 1.774193548387097e-05,
+      "loss": 0.6746,
+      "step": 55
+    },
+    {
+      "epoch": 0.22672064777327935,
+      "grad_norm": 0.5867252493701934,
+      "learning_rate": 1.806451612903226e-05,
+      "loss": 0.6741,
+      "step": 56
+    },
+    {
+      "epoch": 0.23076923076923078,
+      "grad_norm": 0.7510038065663002,
+      "learning_rate": 1.838709677419355e-05,
+      "loss": 0.6677,
+      "step": 57
+    },
+    {
+      "epoch": 0.23481781376518218,
+      "grad_norm": 0.9989003508230546,
+      "learning_rate": 1.870967741935484e-05,
+      "loss": 0.6727,
+      "step": 58
+    },
+    {
+      "epoch": 0.2388663967611336,
+      "grad_norm": 0.6730090855683594,
+      "learning_rate": 1.903225806451613e-05,
+      "loss": 0.6648,
+      "step": 59
+    },
+    {
+      "epoch": 0.242914979757085,
+      "grad_norm": 0.9135706773697393,
+      "learning_rate": 1.935483870967742e-05,
+      "loss": 0.6795,
+      "step": 60
+    },
+    {
+      "epoch": 0.24696356275303644,
+      "grad_norm": 0.7765667339779362,
+      "learning_rate": 1.967741935483871e-05,
+      "loss": 0.6759,
+      "step": 61
+    },
+    {
+      "epoch": 0.25101214574898784,
+      "grad_norm": 0.7688565350932767,
+      "learning_rate": 2e-05,
+      "loss": 0.6721,
+      "step": 62
+    },
+    {
+      "epoch": 0.2550607287449393,
+      "grad_norm": 0.8138643778624666,
+      "learning_rate": 2.0322580645161292e-05,
+      "loss": 0.6649,
+      "step": 63
+    },
+    {
+      "epoch": 0.2591093117408907,
+      "grad_norm": 0.6236667608440538,
+      "learning_rate": 2.0645161290322582e-05,
+      "loss": 0.6632,
+      "step": 64
+    },
+    {
+      "epoch": 0.2631578947368421,
+      "grad_norm": 0.9727642268398281,
+      "learning_rate": 2.0967741935483873e-05,
+      "loss": 0.6456,
+      "step": 65
+    },
+    {
+      "epoch": 0.26720647773279355,
+      "grad_norm": 1.2041830033558103,
+      "learning_rate": 2.1290322580645163e-05,
+      "loss": 0.6547,
+      "step": 66
+    },
+    {
+      "epoch": 0.27125506072874495,
+      "grad_norm": 1.1381360732270562,
+      "learning_rate": 2.161290322580645e-05,
+      "loss": 0.6645,
+      "step": 67
+    },
+    {
+      "epoch": 0.27530364372469635,
+      "grad_norm": 1.030597561890964,
+      "learning_rate": 2.193548387096774e-05,
+      "loss": 0.6628,
+      "step": 68
+    },
+    {
+      "epoch": 0.2793522267206478,
+      "grad_norm": 1.1513313626479782,
+      "learning_rate": 2.225806451612903e-05,
+      "loss": 0.6689,
+      "step": 69
+    },
+    {
+      "epoch": 0.2834008097165992,
+      "grad_norm": 0.9530917232053379,
+      "learning_rate": 2.2580645161290328e-05,
+      "loss": 0.6588,
+      "step": 70
+    },
+    {
+      "epoch": 0.2874493927125506,
+      "grad_norm": 0.8914189266852579,
+      "learning_rate": 2.2903225806451615e-05,
+      "loss": 0.6638,
+      "step": 71
+    },
+    {
+      "epoch": 0.291497975708502,
+      "grad_norm": 1.00860896411086,
+      "learning_rate": 2.3225806451612906e-05,
+      "loss": 0.6501,
+      "step": 72
+    },
+    {
+      "epoch": 0.29554655870445345,
+      "grad_norm": 1.547700341333535,
+      "learning_rate": 2.3548387096774196e-05,
+      "loss": 0.6493,
+      "step": 73
+    },
+    {
+      "epoch": 0.29959514170040485,
+      "grad_norm": 0.8663486408398609,
+      "learning_rate": 2.3870967741935486e-05,
+      "loss": 0.6494,
+      "step": 74
+    },
+    {
+      "epoch": 0.30364372469635625,
+      "grad_norm": 1.6117613823783286,
+      "learning_rate": 2.4193548387096777e-05,
+      "loss": 0.6543,
+      "step": 75
+    },
+    {
+      "epoch": 0.3076923076923077,
+      "grad_norm": 1.2159738891143026,
+      "learning_rate": 2.4516129032258067e-05,
+      "loss": 0.6433,
+      "step": 76
+    },
+    {
+      "epoch": 0.3117408906882591,
+      "grad_norm": 1.5885230963351813,
+      "learning_rate": 2.4838709677419358e-05,
+      "loss": 0.6538,
+      "step": 77
+    },
+    {
+      "epoch": 0.3157894736842105,
+      "grad_norm": 1.1505266215001204,
+      "learning_rate": 2.5161290322580648e-05,
+      "loss": 0.6847,
+      "step": 78
+    },
+    {
+      "epoch": 0.31983805668016196,
+      "grad_norm": 1.755520504594623,
+      "learning_rate": 2.548387096774194e-05,
+      "loss": 0.6446,
+      "step": 79
+    },
+    {
+      "epoch": 0.32388663967611336,
+      "grad_norm": 1.4613898922209339,
+      "learning_rate": 2.5806451612903226e-05,
+      "loss": 0.6543,
+      "step": 80
+    },
+    {
+      "epoch": 0.32793522267206476,
+      "grad_norm": 1.223909501561493,
+      "learning_rate": 2.6129032258064516e-05,
+      "loss": 0.6441,
+      "step": 81
+    },
+    {
+      "epoch": 0.3319838056680162,
+      "grad_norm": 1.6075933668268492,
+      "learning_rate": 2.6451612903225806e-05,
+      "loss": 0.6538,
+      "step": 82
+    },
+    {
+      "epoch": 0.3360323886639676,
+      "grad_norm": 1.1277694934542435,
+      "learning_rate": 2.6774193548387097e-05,
+      "loss": 0.6306,
+      "step": 83
+    },
+    {
+      "epoch": 0.340080971659919,
+      "grad_norm": 1.5782716237313572,
+      "learning_rate": 2.7096774193548387e-05,
+      "loss": 0.6492,
+      "step": 84
+    },
+    {
+      "epoch": 0.3441295546558704,
+      "grad_norm": 1.4259691117354685,
+      "learning_rate": 2.7419354838709678e-05,
+      "loss": 0.6497,
+      "step": 85
+    },
+    {
+      "epoch": 0.3481781376518219,
+      "grad_norm": 1.58662978331792,
+      "learning_rate": 2.774193548387097e-05,
+      "loss": 0.6502,
+      "step": 86
+    },
+    {
+      "epoch": 0.3522267206477733,
+      "grad_norm": 1.1279101264422828,
+      "learning_rate": 2.8064516129032262e-05,
+      "loss": 0.6495,
+      "step": 87
+    },
+    {
+      "epoch": 0.3562753036437247,
+      "grad_norm": 2.1308291344202672,
+      "learning_rate": 2.8387096774193552e-05,
+      "loss": 0.6565,
+      "step": 88
+    },
+    {
+      "epoch": 0.3603238866396761,
+      "grad_norm": 1.3556211827708233,
+      "learning_rate": 2.8709677419354843e-05,
+      "loss": 0.6562,
+      "step": 89
+    },
+    {
+      "epoch": 0.3643724696356275,
+      "grad_norm": 2.408198830784785,
+      "learning_rate": 2.9032258064516133e-05,
+      "loss": 0.6558,
+      "step": 90
+    },
+    {
+      "epoch": 0.3684210526315789,
+      "grad_norm": 1.9046306119077316,
+      "learning_rate": 2.9354838709677423e-05,
+      "loss": 0.6543,
+      "step": 91
+    },
+    {
+      "epoch": 0.3724696356275304,
+      "grad_norm": 1.8988748822691424,
+      "learning_rate": 2.9677419354838714e-05,
+      "loss": 0.6509,
+      "step": 92
+    },
+    {
+      "epoch": 0.3765182186234818,
+      "grad_norm": 1.7319791995486546,
+      "learning_rate": 3.0000000000000004e-05,
+      "loss": 0.6395,
+      "step": 93
+    },
+    {
+      "epoch": 0.3805668016194332,
+      "grad_norm": 1.9331772450767013,
+      "learning_rate": 3.032258064516129e-05,
+      "loss": 0.6348,
+      "step": 94
+    },
+    {
+      "epoch": 0.38461538461538464,
+      "grad_norm": 1.702343608525822,
+      "learning_rate": 3.0645161290322585e-05,
+      "loss": 0.6402,
+      "step": 95
+    },
+    {
+      "epoch": 0.38866396761133604,
+      "grad_norm": 1.7055882335235109,
+      "learning_rate": 3.096774193548387e-05,
+      "loss": 0.6376,
+      "step": 96
+    },
+    {
+      "epoch": 0.39271255060728744,
+      "grad_norm": 1.524770375094887,
+      "learning_rate": 3.1290322580645166e-05,
+      "loss": 0.6486,
+      "step": 97
+    },
+    {
+      "epoch": 0.3967611336032389,
+      "grad_norm": 1.4214958514970124,
+      "learning_rate": 3.161290322580645e-05,
+      "loss": 0.6472,
+      "step": 98
+    },
+    {
+      "epoch": 0.4008097165991903,
+      "grad_norm": 1.3560428966283342,
+      "learning_rate": 3.193548387096774e-05,
+      "loss": 0.6546,
+      "step": 99
+    },
+    {
+      "epoch": 0.4048582995951417,
+      "grad_norm": 1.5201622405371789,
+      "learning_rate": 3.2258064516129034e-05,
+      "loss": 0.6268,
+      "step": 100
+    },
+    {
+      "epoch": 0.4089068825910931,
+      "grad_norm": 1.0608452566440247,
+      "learning_rate": 3.258064516129033e-05,
+      "loss": 0.6485,
+      "step": 101
+    },
+    {
+      "epoch": 0.41295546558704455,
+      "grad_norm": 2.4006939871332333,
+      "learning_rate": 3.2903225806451614e-05,
+      "loss": 0.6305,
+      "step": 102
+    },
+    {
+      "epoch": 0.41700404858299595,
+      "grad_norm": 2.186661236022071,
+      "learning_rate": 3.322580645161291e-05,
+      "loss": 0.6427,
+      "step": 103
+    },
+    {
+      "epoch": 0.42105263157894735,
+      "grad_norm": 1.2799753617077716,
+      "learning_rate": 3.3548387096774195e-05,
+      "loss": 0.6353,
+      "step": 104
+    },
+    {
+      "epoch": 0.4251012145748988,
+      "grad_norm": 1.8351198642951358,
+      "learning_rate": 3.387096774193549e-05,
+      "loss": 0.6453,
+      "step": 105
+    },
+    {
+      "epoch": 0.4291497975708502,
+      "grad_norm": 1.136662753762272,
+      "learning_rate": 3.4193548387096776e-05,
+      "loss": 0.646,
+      "step": 106
+    },
+    {
+      "epoch": 0.4331983805668016,
+      "grad_norm": 1.835373623238951,
+      "learning_rate": 3.451612903225807e-05,
+      "loss": 0.6381,
+      "step": 107
+    },
+    {
+      "epoch": 0.43724696356275305,
+      "grad_norm": 1.4061102736835045,
+      "learning_rate": 3.483870967741936e-05,
+      "loss": 0.633,
+      "step": 108
+    },
+    {
+      "epoch": 0.44129554655870445,
+      "grad_norm": 2.178380303723755,
+      "learning_rate": 3.516129032258065e-05,
+      "loss": 0.6456,
+      "step": 109
+    },
+    {
+      "epoch": 0.44534412955465585,
+      "grad_norm": 1.4189140903127118,
+      "learning_rate": 3.548387096774194e-05,
+      "loss": 0.6321,
+      "step": 110
+    },
+    {
+      "epoch": 0.4493927125506073,
+      "grad_norm": 2.0741135438232634,
+      "learning_rate": 3.5806451612903225e-05,
+      "loss": 0.6305,
+      "step": 111
+    },
+    {
+      "epoch": 0.4534412955465587,
+      "grad_norm": 1.7556879816129585,
+      "learning_rate": 3.612903225806452e-05,
+      "loss": 0.6313,
+      "step": 112
+    },
+    {
+      "epoch": 0.4574898785425101,
+      "grad_norm": 1.6166509979986143,
+      "learning_rate": 3.6451612903225805e-05,
+      "loss": 0.6428,
+      "step": 113
+    },
+    {
+      "epoch": 0.46153846153846156,
+      "grad_norm": 1.854877261687921,
+      "learning_rate": 3.67741935483871e-05,
+      "loss": 0.612,
+      "step": 114
+    },
+    {
+      "epoch": 0.46558704453441296,
+      "grad_norm": 1.2606124871297024,
+      "learning_rate": 3.7096774193548386e-05,
+      "loss": 0.6404,
+      "step": 115
+    },
+    {
+      "epoch": 0.46963562753036436,
+      "grad_norm": 1.880061099181257,
+      "learning_rate": 3.741935483870968e-05,
+      "loss": 0.6485,
+      "step": 116
+    },
+    {
+      "epoch": 0.47368421052631576,
+      "grad_norm": 1.4236667071773774,
+      "learning_rate": 3.7741935483870974e-05,
+      "loss": 0.6299,
+      "step": 117
+    },
+    {
+      "epoch": 0.4777327935222672,
+      "grad_norm": 1.6582365227769678,
+      "learning_rate": 3.806451612903226e-05,
+      "loss": 0.6212,
+      "step": 118
+    },
+    {
+      "epoch": 0.4817813765182186,
+      "grad_norm": 1.304106433406067,
+      "learning_rate": 3.8387096774193555e-05,
+      "loss": 0.6353,
+      "step": 119
+    },
+    {
+      "epoch": 0.48582995951417,
+      "grad_norm": 1.2989246413780229,
+      "learning_rate": 3.870967741935484e-05,
+      "loss": 0.628,
+      "step": 120
+    },
+    {
+      "epoch": 0.4898785425101215,
+      "grad_norm": 1.2027951751667127,
+      "learning_rate": 3.9032258064516135e-05,
+      "loss": 0.641,
+      "step": 121
+    },
+    {
+      "epoch": 0.4939271255060729,
+      "grad_norm": 1.6699043597701675,
+      "learning_rate": 3.935483870967742e-05,
+      "loss": 0.6334,
+      "step": 122
+    },
+    {
+      "epoch": 0.4979757085020243,
+      "grad_norm": 1.0877312197103186,
+      "learning_rate": 3.9677419354838716e-05,
+      "loss": 0.6246,
+      "step": 123
+    },
+    {
+      "epoch": 0.5020242914979757,
+      "grad_norm": 1.3040839297933027,
+      "learning_rate": 4e-05,
+      "loss": 0.6343,
+      "step": 124
+    },
+    {
+      "epoch": 0.5060728744939271,
+      "grad_norm": 1.5643949607243766,
+      "learning_rate": 3.9999920040266474e-05,
+      "loss": 0.6388,
+      "step": 125
+    },
+    {
+      "epoch": 0.5101214574898786,
+      "grad_norm": 1.8455188218382552,
+      "learning_rate": 3.999968016170525e-05,
+      "loss": 0.6438,
+      "step": 126
+    },
+    {
+      "epoch": 0.5141700404858299,
+      "grad_norm": 1.1526709939346227,
+      "learning_rate": 3.9999280366234394e-05,
+      "loss": 0.6224,
+      "step": 127
+    },
+    {
+      "epoch": 0.5182186234817814,
+      "grad_norm": 1.322950424250958,
+      "learning_rate": 3.999872065705065e-05,
+      "loss": 0.6202,
+      "step": 128
+    },
+    {
+      "epoch": 0.5222672064777328,
+      "grad_norm": 2.012855191531723,
+      "learning_rate": 3.999800103862945e-05,
+      "loss": 0.6453,
+      "step": 129
+    },
+    {
+      "epoch": 0.5263157894736842,
+      "grad_norm": 1.233328511254604,
+      "learning_rate": 3.9997121516724834e-05,
+      "loss": 0.6173,
+      "step": 130
+    },
+    {
+      "epoch": 0.5303643724696356,
+      "grad_norm": 1.948425821389313,
+      "learning_rate": 3.999608209836944e-05,
+      "loss": 0.646,
+      "step": 131
+    },
+    {
+      "epoch": 0.5344129554655871,
+      "grad_norm": 1.656981921955592,
+      "learning_rate": 3.999488279187443e-05,
+      "loss": 0.6338,
+      "step": 132
+    },
+    {
+      "epoch": 0.5384615384615384,
+      "grad_norm": 1.4671877999337042,
+      "learning_rate": 3.9993523606829425e-05,
+      "loss": 0.6235,
+      "step": 133
+    },
+    {
+      "epoch": 0.5425101214574899,
+      "grad_norm": 1.5508762818711408,
+      "learning_rate": 3.9992004554102435e-05,
+      "loss": 0.6305,
+      "step": 134
+    },
+    {
+      "epoch": 0.5465587044534413,
+      "grad_norm": 1.7400816218333597,
+      "learning_rate": 3.999032564583976e-05,
+      "loss": 0.6255,
+      "step": 135
+    },
+    {
+      "epoch": 0.5506072874493927,
+      "grad_norm": 1.0392109158668208,
+      "learning_rate": 3.998848689546591e-05,
+      "loss": 0.6165,
+      "step": 136
+    },
+    {
+      "epoch": 0.5546558704453441,
+      "grad_norm": 2.51872375934213,
+      "learning_rate": 3.9986488317683496e-05,
+      "loss": 0.6469,
+      "step": 137
+    },
+    {
+      "epoch": 0.5587044534412956,
+      "grad_norm": 1.6621767946097803,
+      "learning_rate": 3.9984329928473067e-05,
+      "loss": 0.6329,
+      "step": 138
+    },
+    {
+      "epoch": 0.562753036437247,
+      "grad_norm": 2.4841437098952257,
+      "learning_rate": 3.998201174509306e-05,
+      "loss": 0.6336,
+      "step": 139
+    },
+    {
+      "epoch": 0.5668016194331984,
+      "grad_norm": 2.346949118797692,
+      "learning_rate": 3.997953378607961e-05,
+      "loss": 0.64,
+      "step": 140
+    },
+    {
+      "epoch": 0.5708502024291497,
+      "grad_norm": 1.819129856608556,
+      "learning_rate": 3.9976896071246403e-05,
+      "loss": 0.6397,
+      "step": 141
+    },
+    {
+      "epoch": 0.5748987854251012,
+      "grad_norm": 1.8407040212179933,
+      "learning_rate": 3.997409862168454e-05,
+      "loss": 0.6338,
+      "step": 142
+    },
+    {
+      "epoch": 0.5789473684210527,
+      "grad_norm": 1.8114090598004686,
+      "learning_rate": 3.997114145976237e-05,
+      "loss": 0.6299,
+      "step": 143
+    },
+    {
+      "epoch": 0.582995951417004,
+      "grad_norm": 1.5464878592835052,
+      "learning_rate": 3.9968024609125245e-05,
+      "loss": 0.6274,
+      "step": 144
+    },
+    {
+      "epoch": 0.5870445344129555,
+      "grad_norm": 1.860103807692065,
+      "learning_rate": 3.996474809469545e-05,
+      "loss": 0.628,
+      "step": 145
+    },
+    {
+      "epoch": 0.5910931174089069,
+      "grad_norm": 1.5544651142335488,
+      "learning_rate": 3.996131194267189e-05,
+      "loss": 0.6073,
+      "step": 146
+    },
+    {
+      "epoch": 0.5951417004048583,
+      "grad_norm": 1.8299549385002885,
+      "learning_rate": 3.995771618052995e-05,
+      "loss": 0.6088,
+      "step": 147
+    },
+    {
+      "epoch": 0.5991902834008097,
+      "grad_norm": 1.38225066899528,
+      "learning_rate": 3.9953960837021256e-05,
+      "loss": 0.6151,
+      "step": 148
+    },
+    {
+      "epoch": 0.6032388663967612,
+      "grad_norm": 1.5541263185799563,
+      "learning_rate": 3.995004594217343e-05,
+      "loss": 0.6167,
+      "step": 149
+    },
+    {
+      "epoch": 0.6072874493927125,
+      "grad_norm": 1.2926862566796276,
+      "learning_rate": 3.994597152728987e-05,
+      "loss": 0.6259,
+      "step": 150
+    },
+    {
+      "epoch": 0.611336032388664,
+      "grad_norm": 1.2005396368108905,
+      "learning_rate": 3.994173762494947e-05,
+      "loss": 0.6231,
+      "step": 151
+    },
+    {
+      "epoch": 0.6153846153846154,
+      "grad_norm": 1.3401578414324058,
+      "learning_rate": 3.9937344269006424e-05,
+      "loss": 0.6249,
+      "step": 152
+    },
+    {
+      "epoch": 0.6194331983805668,
+      "grad_norm": 1.0915084144428826,
+      "learning_rate": 3.993279149458988e-05,
+      "loss": 0.6262,
+      "step": 153
+    },
+    {
+      "epoch": 0.6234817813765182,
+      "grad_norm": 1.4443431804559288,
+      "learning_rate": 3.9928079338103695e-05,
+      "loss": 0.6102,
+      "step": 154
+    },
+    {
+      "epoch": 0.6275303643724697,
+      "grad_norm": 1.3986015912478134,
+      "learning_rate": 3.992320783722616e-05,
+      "loss": 0.6262,
+      "step": 155
+    },
+    {
+      "epoch": 0.631578947368421,
+      "grad_norm": 1.2480070624932145,
+      "learning_rate": 3.991817703090965e-05,
+      "loss": 0.62,
+      "step": 156
+    },
+    {
+      "epoch": 0.6356275303643725,
+      "grad_norm": 1.1647674734396063,
+      "learning_rate": 3.991298695938038e-05,
+      "loss": 0.6245,
+      "step": 157
+    },
+    {
+      "epoch": 0.6396761133603239,
+      "grad_norm": 1.6156460952970317,
+      "learning_rate": 3.9907637664138e-05,
+      "loss": 0.6288,
+      "step": 158
+    },
+    {
+      "epoch": 0.6437246963562753,
+      "grad_norm": 0.9537158685035974,
+      "learning_rate": 3.990212918795534e-05,
+      "loss": 0.6091,
+      "step": 159
+    },
+    {
+      "epoch": 0.6477732793522267,
+      "grad_norm": 1.9087347581337257,
+      "learning_rate": 3.9896461574878036e-05,
+      "loss": 0.6222,
+      "step": 160
+    },
+    {
+      "epoch": 0.6518218623481782,
+      "grad_norm": 1.4125991988753928,
+      "learning_rate": 3.989063487022416e-05,
+      "loss": 0.6083,
+      "step": 161
+    },
+    {
+      "epoch": 0.6558704453441295,
+      "grad_norm": 1.4000012558212978,
+      "learning_rate": 3.9884649120583903e-05,
+      "loss": 0.6159,
+      "step": 162
+    },
+    {
+      "epoch": 0.659919028340081,
+      "grad_norm": 1.3092013624177254,
+      "learning_rate": 3.987850437381915e-05,
+      "loss": 0.6116,
+      "step": 163
+    },
+    {
+      "epoch": 0.6639676113360324,
+      "grad_norm": 1.237963066793589,
+      "learning_rate": 3.987220067906313e-05,
+      "loss": 0.6199,
+      "step": 164
+    },
+    {
+      "epoch": 0.6680161943319838,
+      "grad_norm": 1.1042753540957966,
+      "learning_rate": 3.986573808672002e-05,
+      "loss": 0.6274,
+      "step": 165
+    },
+    {
+      "epoch": 0.6720647773279352,
+      "grad_norm": 1.2662226135211816,
+      "learning_rate": 3.9859116648464534e-05,
+      "loss": 0.6132,
+      "step": 166
+    },
+    {
+      "epoch": 0.6761133603238867,
+      "grad_norm": 0.9435583700571545,
+      "learning_rate": 3.985233641724152e-05,
+      "loss": 0.6185,
+      "step": 167
+    },
+    {
+      "epoch": 0.680161943319838,
+      "grad_norm": 1.325765884225414,
+      "learning_rate": 3.984539744726554e-05,
+      "loss": 0.6176,
+      "step": 168
+    },
+    {
+      "epoch": 0.6842105263157895,
+      "grad_norm": 0.8694927390681501,
+      "learning_rate": 3.983829979402038e-05,
+      "loss": 0.613,
+      "step": 169
+    },
+    {
+      "epoch": 0.6882591093117408,
+      "grad_norm": 1.2053798366874577,
+      "learning_rate": 3.983104351425871e-05,
+      "loss": 0.6209,
+      "step": 170
+    },
+    {
+      "epoch": 0.6923076923076923,
+      "grad_norm": 1.4016222061589507,
+      "learning_rate": 3.982362866600154e-05,
+      "loss": 0.6246,
+      "step": 171
+    },
+    {
+      "epoch": 0.6963562753036437,
+      "grad_norm": 1.391034260493299,
+      "learning_rate": 3.9816055308537806e-05,
+      "loss": 0.61,
+      "step": 172
+    },
+    {
+      "epoch": 0.7004048582995951,
+      "grad_norm": 0.9016642881848921,
+      "learning_rate": 3.980832350242388e-05,
+      "loss": 0.6046,
+      "step": 173
+    },
+    {
+      "epoch": 0.7044534412955465,
+      "grad_norm": 1.1648272886668665,
+      "learning_rate": 3.980043330948306e-05,
+      "loss": 0.6127,
+      "step": 174
+    },
+    {
+      "epoch": 0.708502024291498,
+      "grad_norm": 0.9253198907457892,
+      "learning_rate": 3.9792384792805124e-05,
+      "loss": 0.6185,
+      "step": 175
+    },
+    {
+      "epoch": 0.7125506072874493,
+      "grad_norm": 1.3400827799288066,
+      "learning_rate": 3.978417801674579e-05,
+      "loss": 0.6044,
+      "step": 176
+    },
+    {
+      "epoch": 0.7165991902834008,
+      "grad_norm": 1.2672575754380495,
+      "learning_rate": 3.9775813046926244e-05,
+      "loss": 0.6152,
+      "step": 177
+    },
+    {
+      "epoch": 0.7206477732793523,
+      "grad_norm": 1.1012844939569613,
+      "learning_rate": 3.976728995023254e-05,
+      "loss": 0.6094,
+      "step": 178
+    },
+    {
+      "epoch": 0.7246963562753036,
+      "grad_norm": 1.5385987251252273,
+      "learning_rate": 3.975860879481514e-05,
+      "loss": 0.6134,
+      "step": 179
+    },
+    {
+      "epoch": 0.728744939271255,
+      "grad_norm": 0.7880886978016494,
+      "learning_rate": 3.974976965008833e-05,
+      "loss": 0.6085,
+      "step": 180
+    },
+    {
+      "epoch": 0.7327935222672065,
+      "grad_norm": 1.4096778231590392,
+      "learning_rate": 3.9740772586729674e-05,
+      "loss": 0.6177,
+      "step": 181
+    },
+    {
+      "epoch": 0.7368421052631579,
+      "grad_norm": 1.3557757496983147,
+      "learning_rate": 3.973161767667946e-05,
+      "loss": 0.64,
+      "step": 182
+    },
+    {
+      "epoch": 0.7408906882591093,
+      "grad_norm": 0.9560985091377173,
+      "learning_rate": 3.972230499314009e-05,
+      "loss": 0.6168,
+      "step": 183
+    },
+    {
+      "epoch": 0.7449392712550608,
+      "grad_norm": 1.329864284859221,
+      "learning_rate": 3.9712834610575545e-05,
+      "loss": 0.6151,
+      "step": 184
+    },
+    {
+      "epoch": 0.7489878542510121,
+      "grad_norm": 1.2014052453465742,
+      "learning_rate": 3.9703206604710746e-05,
+      "loss": 0.6239,
+      "step": 185
+    },
+    {
+      "epoch": 0.7530364372469636,
+      "grad_norm": 0.995500177582187,
+      "learning_rate": 3.969342105253098e-05,
+      "loss": 0.6029,
+      "step": 186
+    },
+    {
+      "epoch": 0.757085020242915,
+      "grad_norm": 1.1357625459247451,
+      "learning_rate": 3.9683478032281246e-05,
+      "loss": 0.606,
+      "step": 187
+    },
+    {
+      "epoch": 0.7611336032388664,
+      "grad_norm": 1.2087377538210227,
+      "learning_rate": 3.967337762346568e-05,
+      "loss": 0.5985,
+      "step": 188
+    },
+    {
+      "epoch": 0.7651821862348178,
+      "grad_norm": 0.9483658089472433,
+      "learning_rate": 3.9663119906846885e-05,
+      "loss": 0.6187,
+      "step": 189
+    },
+    {
+      "epoch": 0.7692307692307693,
+      "grad_norm": 1.2720948015842775,
+      "learning_rate": 3.965270496444528e-05,
+      "loss": 0.6171,
+      "step": 190
+    },
+    {
+      "epoch": 0.7732793522267206,
+      "grad_norm": 0.7700138496208023,
+      "learning_rate": 3.9642132879538465e-05,
+      "loss": 0.607,
+      "step": 191
+    },
+    {
+      "epoch": 0.7773279352226721,
+      "grad_norm": 1.1555089639315497,
+      "learning_rate": 3.9631403736660566e-05,
+      "loss": 0.5979,
+      "step": 192
+    },
+    {
+      "epoch": 0.7813765182186235,
+      "grad_norm": 1.218209971930902,
+      "learning_rate": 3.962051762160151e-05,
+      "loss": 0.6052,
+      "step": 193
+    },
+    {
+      "epoch": 0.7854251012145749,
+      "grad_norm": 1.1724361357359896,
+      "learning_rate": 3.960947462140638e-05,
+      "loss": 0.5997,
+      "step": 194
+    },
+    {
+      "epoch": 0.7894736842105263,
+      "grad_norm": 1.1565701098240506,
+      "learning_rate": 3.959827482437472e-05,
+      "loss": 0.6049,
+      "step": 195
+    },
+    {
+      "epoch": 0.7935222672064778,
+      "grad_norm": 0.9496064113694452,
+      "learning_rate": 3.958691832005981e-05,
+      "loss": 0.6036,
+      "step": 196
+    },
+    {
+      "epoch": 0.7975708502024291,
+      "grad_norm": 0.9396592767838323,
+      "learning_rate": 3.9575405199267944e-05,
+      "loss": 0.602,
+      "step": 197
+    },
+    {
+      "epoch": 0.8016194331983806,
+      "grad_norm": 0.7870580617820654,
+      "learning_rate": 3.956373555405774e-05,
+      "loss": 0.5898,
+      "step": 198
+    },
+    {
+      "epoch": 0.805668016194332,
+      "grad_norm": 1.2618471587668636,
+      "learning_rate": 3.955190947773937e-05,
+      "loss": 0.6008,
+      "step": 199
+    },
+    {
+      "epoch": 0.8097165991902834,
+      "grad_norm": 1.28623941930131,
+      "learning_rate": 3.9539927064873824e-05,
+      "loss": 0.6104,
+      "step": 200
+    },
+    {
+      "epoch": 0.8137651821862348,
+      "grad_norm": 1.1090504302400086,
+      "learning_rate": 3.952778841127214e-05,
+      "loss": 0.5999,
+      "step": 201
+    },
+    {
+      "epoch": 0.8178137651821862,
+      "grad_norm": 1.0446261520562585,
+      "learning_rate": 3.95154936139947e-05,
+      "loss": 0.6153,
+      "step": 202
+    },
+    {
+      "epoch": 0.8218623481781376,
+      "grad_norm": 1.19197815479649,
+      "learning_rate": 3.950304277135034e-05,
+      "loss": 0.6042,
+      "step": 203
+    },
+    {
+      "epoch": 0.8259109311740891,
+      "grad_norm": 0.987581015474592,
+      "learning_rate": 3.94904359828957e-05,
+      "loss": 0.6149,
+      "step": 204
+    },
+    {
+      "epoch": 0.8299595141700404,
+      "grad_norm": 1.4373664419556806,
+      "learning_rate": 3.94776733494343e-05,
+      "loss": 0.6168,
+      "step": 205
+    },
+    {
+      "epoch": 0.8340080971659919,
+      "grad_norm": 0.9378289673451462,
+      "learning_rate": 3.9464754973015826e-05,
+      "loss": 0.5957,
+      "step": 206
+    },
+    {
+      "epoch": 0.8380566801619433,
+      "grad_norm": 0.9779168351778184,
+      "learning_rate": 3.9451680956935275e-05,
+      "loss": 0.6088,
+      "step": 207
+    },
+    {
+      "epoch": 0.8421052631578947,
+      "grad_norm": 1.29024528435315,
+      "learning_rate": 3.943845140573213e-05,
+      "loss": 0.6066,
+      "step": 208
+    },
+    {
+      "epoch": 0.8461538461538461,
+      "grad_norm": 0.9703994121465018,
+      "learning_rate": 3.942506642518952e-05,
+      "loss": 0.601,
+      "step": 209
+    },
+    {
+      "epoch": 0.8502024291497976,
+      "grad_norm": 0.9381433772169446,
+      "learning_rate": 3.9411526122333396e-05,
+      "loss": 0.6007,
+      "step": 210
+    },
+    {
+      "epoch": 0.854251012145749,
+      "grad_norm": 0.8851232969906474,
+      "learning_rate": 3.939783060543167e-05,
+      "loss": 0.6006,
+      "step": 211
+    },
+    {
+      "epoch": 0.8582995951417004,
+      "grad_norm": 0.919036646417073,
+      "learning_rate": 3.938397998399333e-05,
+      "loss": 0.6075,
+      "step": 212
+    },
+    {
+      "epoch": 0.8623481781376519,
+      "grad_norm": 1.1487181957155512,
+      "learning_rate": 3.936997436876756e-05,
+      "loss": 0.5998,
+      "step": 213
+    },
+    {
+      "epoch": 0.8663967611336032,
+      "grad_norm": 1.0449760476442898,
+      "learning_rate": 3.935581387174289e-05,
+      "loss": 0.602,
+      "step": 214
+    },
+    {
+      "epoch": 0.8704453441295547,
+      "grad_norm": 0.946023239109648,
+      "learning_rate": 3.9341498606146295e-05,
+      "loss": 0.6057,
+      "step": 215
+    },
+    {
+      "epoch": 0.8744939271255061,
+      "grad_norm": 1.1536556386649564,
+      "learning_rate": 3.932702868644224e-05,
+      "loss": 0.6114,
+      "step": 216
+    },
+    {
+      "epoch": 0.8785425101214575,
+      "grad_norm": 1.20595717996231,
+      "learning_rate": 3.931240422833181e-05,
+      "loss": 0.6159,
+      "step": 217
+    },
+    {
+      "epoch": 0.8825910931174089,
+      "grad_norm": 0.9623196068447484,
+      "learning_rate": 3.9297625348751806e-05,
+      "loss": 0.5975,
+      "step": 218
+    },
+    {
+      "epoch": 0.8866396761133604,
+      "grad_norm": 1.0346015346331674,
+      "learning_rate": 3.9282692165873744e-05,
+      "loss": 0.5968,
+      "step": 219
+    },
+    {
+      "epoch": 0.8906882591093117,
+      "grad_norm": 1.2231789239660333,
+      "learning_rate": 3.9267604799102945e-05,
+      "loss": 0.5995,
+      "step": 220
+    },
+    {
+      "epoch": 0.8947368421052632,
+      "grad_norm": 1.1232050456558562,
+      "learning_rate": 3.9252363369077615e-05,
+      "loss": 0.6086,
+      "step": 221
+    },
+    {
+      "epoch": 0.8987854251012146,
+      "grad_norm": 1.0864417714622983,
+      "learning_rate": 3.9236967997667796e-05,
+      "loss": 0.6093,
+      "step": 222
+    },
+    {
+      "epoch": 0.902834008097166,
+      "grad_norm": 0.7603461503324972,
+      "learning_rate": 3.9221418807974496e-05,
+      "loss": 0.6021,
+      "step": 223
+    },
+    {
+      "epoch": 0.9068825910931174,
+      "grad_norm": 0.8737041006719874,
+      "learning_rate": 3.92057159243286e-05,
+      "loss": 0.5951,
+      "step": 224
+    },
+    {
+      "epoch": 0.9109311740890689,
+      "grad_norm": 1.2332786714423813,
+      "learning_rate": 3.9189859472289956e-05,
+      "loss": 0.5988,
+      "step": 225
+    },
+    {
+      "epoch": 0.9149797570850202,
+      "grad_norm": 1.0489523286875717,
+      "learning_rate": 3.9173849578646324e-05,
+      "loss": 0.5924,
+      "step": 226
+    },
+    {
+      "epoch": 0.9190283400809717,
+      "grad_norm": 0.8710344308362433,
+      "learning_rate": 3.9157686371412395e-05,
+      "loss": 0.6032,
+      "step": 227
+    },
+    {
+      "epoch": 0.9230769230769231,
+      "grad_norm": 0.8446856691395843,
+      "learning_rate": 3.914136997982874e-05,
+      "loss": 0.6022,
+      "step": 228
+    },
+    {
+      "epoch": 0.9271255060728745,
+      "grad_norm": 0.9116698979814577,
+      "learning_rate": 3.912490053436079e-05,
+      "loss": 0.6167,
+      "step": 229
+    },
+    {
+      "epoch": 0.9311740890688259,
+      "grad_norm": 1.0186860742288248,
+      "learning_rate": 3.9108278166697805e-05,
+      "loss": 0.6182,
+      "step": 230
+    },
+    {
+      "epoch": 0.9352226720647774,
+      "grad_norm": 1.2305832065893032,
+      "learning_rate": 3.909150300975178e-05,
+      "loss": 0.5994,
+      "step": 231
+    },
+    {
+      "epoch": 0.9392712550607287,
+      "grad_norm": 0.8800976513204251,
+      "learning_rate": 3.907457519765642e-05,
+      "loss": 0.5929,
+      "step": 232
+    },
+    {
+      "epoch": 0.9433198380566802,
+      "grad_norm": 1.2469879635200654,
+      "learning_rate": 3.905749486576607e-05,
+      "loss": 0.6042,
+      "step": 233
+    },
+    {
+      "epoch": 0.9473684210526315,
+      "grad_norm": 0.8865400164219897,
+      "learning_rate": 3.90402621506546e-05,
+      "loss": 0.6093,
+      "step": 234
+    },
+    {
+      "epoch": 0.951417004048583,
+      "grad_norm": 0.8388666244728249,
+      "learning_rate": 3.902287719011435e-05,
+      "loss": 0.6003,
+      "step": 235
+    },
+    {
+      "epoch": 0.9554655870445344,
+      "grad_norm": 0.9537805095429923,
+      "learning_rate": 3.900534012315499e-05,
+      "loss": 0.5878,
+      "step": 236
+    },
+    {
+      "epoch": 0.9595141700404858,
+      "grad_norm": 1.150943368606977,
+      "learning_rate": 3.8987651090002454e-05,
+      "loss": 0.5984,
+      "step": 237
+    },
+    {
+      "epoch": 0.9635627530364372,
+      "grad_norm": 0.8995561163861182,
+      "learning_rate": 3.896981023209777e-05,
+      "loss": 0.6045,
+      "step": 238
+    },
+    {
+      "epoch": 0.9676113360323887,
+      "grad_norm": 0.8691783154056315,
+      "learning_rate": 3.8951817692095966e-05,
+      "loss": 0.5914,
+      "step": 239
+    },
+    {
+      "epoch": 0.97165991902834,
+      "grad_norm": 1.1053217075317228,
+      "learning_rate": 3.893367361386491e-05,
+      "loss": 0.5842,
+      "step": 240
+    },
+    {
+      "epoch": 0.9757085020242915,
+      "grad_norm": 0.7717711051305616,
+      "learning_rate": 3.891537814248417e-05,
+      "loss": 0.5928,
+      "step": 241
+    },
+    {
+      "epoch": 0.979757085020243,
+      "grad_norm": 0.8614063654134501,
+      "learning_rate": 3.889693142424384e-05,
+      "loss": 0.6011,
+      "step": 242
+    },
+    {
+      "epoch": 0.9838056680161943,
+      "grad_norm": 0.911875290901189,
+      "learning_rate": 3.8878333606643405e-05,
+      "loss": 0.5908,
+      "step": 243
+    },
+    {
+      "epoch": 0.9878542510121457,
+      "grad_norm": 0.82419641223558,
+      "learning_rate": 3.885958483839049e-05,
+      "loss": 0.5908,
+      "step": 244
+    },
+    {
+      "epoch": 0.9919028340080972,
+      "grad_norm": 0.8625146724121691,
+      "learning_rate": 3.8840685269399786e-05,
+      "loss": 0.594,
+      "step": 245
+    },
+    {
+      "epoch": 0.9959514170040485,
+      "grad_norm": 1.2488448803043803,
+      "learning_rate": 3.882163505079171e-05,
+      "loss": 0.5807,
+      "step": 246
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.6923439229749667,
+      "learning_rate": 3.880243433489132e-05,
+      "loss": 0.6022,
+      "step": 247
+    },
+    {
+      "epoch": 1.0040485829959513,
+      "grad_norm": 1.2875604117721682,
+      "learning_rate": 3.878308327522702e-05,
+      "loss": 0.549,
+      "step": 248
+    },
+    {
+      "epoch": 1.008097165991903,
+      "grad_norm": 0.608329125090819,
+      "learning_rate": 3.876358202652936e-05,
+      "loss": 0.5605,
+      "step": 249
+    },
+    {
+      "epoch": 1.0121457489878543,
+      "grad_norm": 0.8637774091928133,
+      "learning_rate": 3.8743930744729835e-05,
+      "loss": 0.5572,
+      "step": 250
+    },
+    {
+      "epoch": 1.0161943319838056,
+      "grad_norm": 1.1363955890162067,
+      "learning_rate": 3.8724129586959535e-05,
+      "loss": 0.5538,
+      "step": 251
+    },
+    {
+      "epoch": 1.0202429149797572,
+      "grad_norm": 1.1067787866844503,
+      "learning_rate": 3.870417871154801e-05,
+      "loss": 0.5581,
+      "step": 252
+    },
+    {
+      "epoch": 1.0242914979757085,
+      "grad_norm": 1.0374884392260322,
+      "learning_rate": 3.868407827802192e-05,
+      "loss": 0.5614,
+      "step": 253
+    },
+    {
+      "epoch": 1.0283400809716599,
+      "grad_norm": 0.9311994014578118,
+      "learning_rate": 3.8663828447103804e-05,
+      "loss": 0.5688,
+      "step": 254
+    },
+    {
+      "epoch": 1.0323886639676114,
+      "grad_norm": 0.9331655274464027,
+      "learning_rate": 3.8643429380710765e-05,
+      "loss": 0.5483,
+      "step": 255
+    },
+    {
+      "epoch": 1.0364372469635628,
+      "grad_norm": 0.9884395906558908,
+      "learning_rate": 3.862288124195319e-05,
+      "loss": 0.5486,
+      "step": 256
+    },
+    {
+      "epoch": 1.040485829959514,
+      "grad_norm": 1.0170252979651653,
+      "learning_rate": 3.8602184195133464e-05,
+      "loss": 0.5601,
+      "step": 257
+    },
+    {
+      "epoch": 1.0445344129554657,
+      "grad_norm": 1.1862052460322414,
+      "learning_rate": 3.8581338405744604e-05,
+      "loss": 0.5446,
+      "step": 258
+    },
+    {
+      "epoch": 1.048582995951417,
+      "grad_norm": 0.8931988740790985,
+      "learning_rate": 3.8560344040468996e-05,
+      "loss": 0.5751,
+      "step": 259
+    },
+    {
+      "epoch": 1.0526315789473684,
+      "grad_norm": 0.9457640598589242,
+      "learning_rate": 3.853920126717703e-05,
+      "loss": 0.5653,
+      "step": 260
+    },
+    {
+      "epoch": 1.05668016194332,
+      "grad_norm": 0.7317986592206058,
+      "learning_rate": 3.851791025492574e-05,
+      "loss": 0.5559,
+      "step": 261
+    },
+    {
+      "epoch": 1.0607287449392713,
+      "grad_norm": 1.0886822507307512,
+      "learning_rate": 3.8496471173957515e-05,
+      "loss": 0.5542,
+      "step": 262
+    },
+    {
+      "epoch": 1.0647773279352226,
+      "grad_norm": 1.1849575406346287,
+      "learning_rate": 3.847488419569866e-05,
+      "loss": 0.564,
+      "step": 263
+    },
+    {
+      "epoch": 1.0688259109311742,
+      "grad_norm": 0.7305974143472481,
+      "learning_rate": 3.8453149492758077e-05,
+      "loss": 0.5553,
+      "step": 264
+    },
+    {
+      "epoch": 1.0728744939271255,
+      "grad_norm": 1.048026984147303,
+      "learning_rate": 3.8431267238925886e-05,
+      "loss": 0.5522,
+      "step": 265
+    },
+    {
+      "epoch": 1.0769230769230769,
+      "grad_norm": 0.938707296339516,
+      "learning_rate": 3.840923760917199e-05,
+      "loss": 0.5692,
+      "step": 266
+    },
+    {
+      "epoch": 1.0809716599190284,
+      "grad_norm": 0.7054761916116961,
+      "learning_rate": 3.838706077964473e-05,
+      "loss": 0.5584,
+      "step": 267
+    },
+    {
+      "epoch": 1.0850202429149798,
+      "grad_norm": 0.5690428131253701,
+      "learning_rate": 3.836473692766944e-05,
+      "loss": 0.5561,
+      "step": 268
+    },
+    {
+      "epoch": 1.0890688259109311,
+      "grad_norm": 0.8224320785240753,
+      "learning_rate": 3.834226623174705e-05,
+      "loss": 0.5478,
+      "step": 269
+    },
+    {
+      "epoch": 1.0931174089068827,
+      "grad_norm": 1.2923887790811,
+      "learning_rate": 3.831964887155264e-05,
+      "loss": 0.573,
+      "step": 270
+    },
+    {
+      "epoch": 1.097165991902834,
+      "grad_norm": 0.7542103217690346,
+      "learning_rate": 3.829688502793403e-05,
+      "loss": 0.5452,
+      "step": 271
+    },
+    {
+      "epoch": 1.1012145748987854,
+      "grad_norm": 1.0264025433494681,
+      "learning_rate": 3.827397488291029e-05,
+      "loss": 0.5643,
+      "step": 272
+    },
+    {
+      "epoch": 1.1052631578947367,
+      "grad_norm": 0.9728112036252929,
+      "learning_rate": 3.8250918619670343e-05,
+      "loss": 0.5512,
+      "step": 273
+    },
+    {
+      "epoch": 1.1093117408906883,
+      "grad_norm": 1.2122701623009668,
+      "learning_rate": 3.822771642257145e-05,
+      "loss": 0.5717,
+      "step": 274
+    },
+    {
+      "epoch": 1.1133603238866396,
+      "grad_norm": 1.0114674887259005,
+      "learning_rate": 3.820436847713776e-05,
+      "loss": 0.5428,
+      "step": 275
+    },
+    {
+      "epoch": 1.117408906882591,
+      "grad_norm": 0.9288285307778745,
+      "learning_rate": 3.8180874970058826e-05,
+      "loss": 0.5394,
+      "step": 276
+    },
+    {
+      "epoch": 1.1214574898785425,
+      "grad_norm": 1.073315019600305,
+      "learning_rate": 3.8157236089188103e-05,
+      "loss": 0.5524,
+      "step": 277
+    },
+    {
+      "epoch": 1.125506072874494,
+      "grad_norm": 1.0055479951990578,
+      "learning_rate": 3.813345202354145e-05,
+      "loss": 0.547,
+      "step": 278
+    },
+    {
+      "epoch": 1.1295546558704452,
+      "grad_norm": 0.9758761202098775,
+      "learning_rate": 3.810952296329563e-05,
+      "loss": 0.5692,
+      "step": 279
+    },
+    {
+      "epoch": 1.1336032388663968,
+      "grad_norm": 0.9789813856055244,
+      "learning_rate": 3.808544909978676e-05,
+      "loss": 0.5492,
+      "step": 280
+    },
+    {
+      "epoch": 1.1376518218623481,
+      "grad_norm": 0.9107760060233573,
+      "learning_rate": 3.806123062550882e-05,
+      "loss": 0.5552,
+      "step": 281
+    },
+    {
+      "epoch": 1.1417004048582995,
+      "grad_norm": 1.0708200345839969,
+      "learning_rate": 3.803686773411208e-05,
+      "loss": 0.5423,
+      "step": 282
+    },
+    {
+      "epoch": 1.145748987854251,
+      "grad_norm": 0.9487852329144653,
+      "learning_rate": 3.801236062040158e-05,
+      "loss": 0.5541,
+      "step": 283
+    },
+    {
+      "epoch": 1.1497975708502024,
+      "grad_norm": 0.9538264334576009,
+      "learning_rate": 3.798770948033553e-05,
+      "loss": 0.5453,
+      "step": 284
+    },
+    {
+      "epoch": 1.1538461538461537,
+      "grad_norm": 0.9047651373978068,
+      "learning_rate": 3.796291451102381e-05,
+      "loss": 0.5525,
+      "step": 285
+    },
+    {
+      "epoch": 1.1578947368421053,
+      "grad_norm": 0.5881903896215751,
+      "learning_rate": 3.793797591072631e-05,
+      "loss": 0.5505,
+      "step": 286
+    },
+    {
+      "epoch": 1.1619433198380567,
+      "grad_norm": 0.7118484939531968,
+      "learning_rate": 3.791289387885144e-05,
+      "loss": 0.5536,
+      "step": 287
+    },
+    {
+      "epoch": 1.165991902834008,
+      "grad_norm": 1.0198547500552453,
+      "learning_rate": 3.788766861595443e-05,
+      "loss": 0.5595,
+      "step": 288
+    },
+    {
+      "epoch": 1.1700404858299596,
+      "grad_norm": 0.871312586313798,
+      "learning_rate": 3.7862300323735835e-05,
+      "loss": 0.5631,
+      "step": 289
+    },
+    {
+      "epoch": 1.174089068825911,
+      "grad_norm": 0.9462467088776504,
+      "learning_rate": 3.783678920503983e-05,
+      "loss": 0.5622,
+      "step": 290
+    },
+    {
+      "epoch": 1.1781376518218623,
+      "grad_norm": 0.8628342533261177,
+      "learning_rate": 3.781113546385265e-05,
+      "loss": 0.5415,
+      "step": 291
+    },
+    {
+      "epoch": 1.1821862348178138,
+      "grad_norm": 0.7583597624227196,
+      "learning_rate": 3.7785339305300916e-05,
+      "loss": 0.554,
+      "step": 292
+    },
+    {
+      "epoch": 1.1862348178137652,
+      "grad_norm": 0.8428947290283714,
+      "learning_rate": 3.7759400935650025e-05,
+      "loss": 0.5592,
+      "step": 293
+    },
+    {
+      "epoch": 1.1902834008097165,
+      "grad_norm": 0.6085974286944771,
+      "learning_rate": 3.77333205623025e-05,
+      "loss": 0.5545,
+      "step": 294
+    },
+    {
+      "epoch": 1.194331983805668,
+      "grad_norm": 0.7474668565320257,
+      "learning_rate": 3.7707098393796295e-05,
+      "loss": 0.5579,
+      "step": 295
+    },
+    {
+      "epoch": 1.1983805668016194,
+      "grad_norm": 0.710164185787316,
+      "learning_rate": 3.7680734639803185e-05,
+      "loss": 0.5516,
+      "step": 296
+    },
+    {
+      "epoch": 1.2024291497975708,
+      "grad_norm": 0.535409765421496,
+      "learning_rate": 3.765422951112704e-05,
+      "loss": 0.5627,
+      "step": 297
+    },
+    {
+      "epoch": 1.2064777327935223,
+      "grad_norm": 0.784988878233653,
+      "learning_rate": 3.762758321970216e-05,
+      "loss": 0.556,
+      "step": 298
+    },
+    {
+      "epoch": 1.2105263157894737,
+      "grad_norm": 0.9477286603943309,
+      "learning_rate": 3.7600795978591584e-05,
+      "loss": 0.5536,
+      "step": 299
+    },
+    {
+      "epoch": 1.214574898785425,
+      "grad_norm": 1.0129611491844364,
+      "learning_rate": 3.757386800198538e-05,
+      "loss": 0.5666,
+      "step": 300
+    },
+    {
+      "epoch": 1.2186234817813766,
+      "grad_norm": 1.0697980475718205,
+      "learning_rate": 3.7546799505198925e-05,
+      "loss": 0.56,
+      "step": 301
+    },
+    {
+      "epoch": 1.222672064777328,
+      "grad_norm": 0.7596163621962714,
+      "learning_rate": 3.7519590704671197e-05,
+      "loss": 0.5704,
+      "step": 302
+    },
+    {
+      "epoch": 1.2267206477732793,
+      "grad_norm": 0.6629350485895089,
+      "learning_rate": 3.749224181796305e-05,
+      "loss": 0.5677,
+      "step": 303
+    },
+    {
+      "epoch": 1.2307692307692308,
+      "grad_norm": 0.8534454809882229,
+      "learning_rate": 3.746475306375545e-05,
+      "loss": 0.5547,
+      "step": 304
+    },
+    {
+      "epoch": 1.2348178137651822,
+      "grad_norm": 0.9973445544047009,
+      "learning_rate": 3.743712466184774e-05,
+      "loss": 0.5686,
+      "step": 305
+    },
+    {
+      "epoch": 1.2388663967611335,
+      "grad_norm": 0.9167033671633634,
+      "learning_rate": 3.7409356833155885e-05,
+      "loss": 0.5487,
+      "step": 306
+    },
+    {
+      "epoch": 1.242914979757085,
+      "grad_norm": 0.7626957166478787,
+      "learning_rate": 3.73814497997107e-05,
+      "loss": 0.5507,
+      "step": 307
+    },
+    {
+      "epoch": 1.2469635627530364,
+      "grad_norm": 0.5952423218879487,
+      "learning_rate": 3.7353403784656095e-05,
+      "loss": 0.5549,
+      "step": 308
+    },
+    {
+      "epoch": 1.2510121457489878,
+      "grad_norm": 0.7903046852645126,
+      "learning_rate": 3.732521901224724e-05,
+      "loss": 0.5556,
+      "step": 309
+    },
+    {
+      "epoch": 1.2550607287449393,
+      "grad_norm": 0.7759225709815586,
+      "learning_rate": 3.7296895707848844e-05,
+      "loss": 0.5562,
+      "step": 310
+    },
+    {
+      "epoch": 1.2591093117408907,
+      "grad_norm": 0.9069024370903955,
+      "learning_rate": 3.7268434097933275e-05,
+      "loss": 0.5544,
+      "step": 311
+    },
+    {
+      "epoch": 1.263157894736842,
+      "grad_norm": 0.9654248087560245,
+      "learning_rate": 3.723983441007882e-05,
+      "loss": 0.5472,
+      "step": 312
+    },
+    {
+      "epoch": 1.2672064777327936,
+      "grad_norm": 0.8124999968691082,
+      "learning_rate": 3.721109687296781e-05,
+      "loss": 0.5615,
+      "step": 313
+    },
+    {
+      "epoch": 1.271255060728745,
+      "grad_norm": 0.6150435756942622,
+      "learning_rate": 3.718222171638484e-05,
+      "loss": 0.5518,
+      "step": 314
+    },
+    {
+      "epoch": 1.2753036437246963,
+      "grad_norm": 0.8919350775735388,
+      "learning_rate": 3.715320917121488e-05,
+      "loss": 0.5571,
+      "step": 315
+    },
+    {
+      "epoch": 1.2793522267206479,
+      "grad_norm": 0.810615299272276,
+      "learning_rate": 3.712405946944149e-05,
+      "loss": 0.5637,
+      "step": 316
+    },
+    {
+      "epoch": 1.2834008097165992,
+      "grad_norm": 0.8694366602081891,
+      "learning_rate": 3.709477284414488e-05,
+      "loss": 0.5535,
+      "step": 317
+    },
+    {
+      "epoch": 1.2874493927125505,
+      "grad_norm": 1.0307956393841948,
+      "learning_rate": 3.7065349529500136e-05,
+      "loss": 0.55,
+      "step": 318
+    },
+    {
+      "epoch": 1.291497975708502,
+      "grad_norm": 1.0367732013823783,
+      "learning_rate": 3.7035789760775305e-05,
+      "loss": 0.5508,
+      "step": 319
+    },
+    {
+      "epoch": 1.2955465587044535,
+      "grad_norm": 0.8442427064049621,
+      "learning_rate": 3.70060937743295e-05,
+      "loss": 0.5546,
+      "step": 320
+    },
+    {
+      "epoch": 1.2995951417004048,
+      "grad_norm": 0.8396231156316588,
+      "learning_rate": 3.6976261807611045e-05,
+      "loss": 0.5513,
+      "step": 321
+    },
+    {
+      "epoch": 1.3036437246963564,
+      "grad_norm": 0.7411085152980473,
+      "learning_rate": 3.694629409915555e-05,
+      "loss": 0.5383,
+      "step": 322
+    },
+    {
+      "epoch": 1.3076923076923077,
+      "grad_norm": 0.8330585620227693,
+      "learning_rate": 3.691619088858401e-05,
+      "loss": 0.5693,
+      "step": 323
+    },
+    {
+      "epoch": 1.311740890688259,
+      "grad_norm": 1.3292473904125965,
+      "learning_rate": 3.68859524166009e-05,
+      "loss": 0.5489,
+      "step": 324
+    },
+    {
+      "epoch": 1.3157894736842106,
+      "grad_norm": 0.5808901673723256,
+      "learning_rate": 3.6855578924992226e-05,
+      "loss": 0.542,
+      "step": 325
+    },
+    {
+      "epoch": 1.319838056680162,
+      "grad_norm": 1.3020998264352723,
+      "learning_rate": 3.6825070656623626e-05,
+      "loss": 0.5563,
+      "step": 326
+    },
+    {
+      "epoch": 1.3238866396761133,
+      "grad_norm": 0.7965343788900325,
+      "learning_rate": 3.67944278554384e-05,
+      "loss": 0.5534,
+      "step": 327
+    },
+    {
+      "epoch": 1.3279352226720649,
+      "grad_norm": 0.8485732152788362,
+      "learning_rate": 3.676365076645557e-05,
+      "loss": 0.5549,
+      "step": 328
+    },
+    {
+      "epoch": 1.3319838056680162,
+      "grad_norm": 0.8393752329368717,
+      "learning_rate": 3.673273963576791e-05,
+      "loss": 0.5483,
+      "step": 329
+    },
+    {
+      "epoch": 1.3360323886639676,
+      "grad_norm": 0.4899684927204911,
+      "learning_rate": 3.670169471054001e-05,
+      "loss": 0.5481,
+      "step": 330
+    },
+    {
+      "epoch": 1.3400809716599191,
+      "grad_norm": 0.5268590673732356,
+      "learning_rate": 3.6670516239006254e-05,
+      "loss": 0.5559,
+      "step": 331
+    },
+    {
+      "epoch": 1.3441295546558705,
+      "grad_norm": 0.6569519456095773,
+      "learning_rate": 3.663920447046888e-05,
+      "loss": 0.5557,
+      "step": 332
+    },
+    {
+      "epoch": 1.3481781376518218,
+      "grad_norm": 0.6651027098040514,
+      "learning_rate": 3.6607759655295954e-05,
+      "loss": 0.5662,
+      "step": 333
+    },
+    {
+      "epoch": 1.3522267206477734,
+      "grad_norm": 0.5721496322532571,
+      "learning_rate": 3.657618204491937e-05,
+      "loss": 0.5515,
+      "step": 334
+    },
+    {
+      "epoch": 1.3562753036437247,
+      "grad_norm": 0.694321188241962,
+      "learning_rate": 3.6544471891832875e-05,
+      "loss": 0.5575,
+      "step": 335
+    },
+    {
+      "epoch": 1.360323886639676,
+      "grad_norm": 0.785599377510318,
+      "learning_rate": 3.6512629449589983e-05,
+      "loss": 0.5435,
+      "step": 336
+    },
+    {
+      "epoch": 1.3643724696356276,
+      "grad_norm": 0.8219771398374719,
+      "learning_rate": 3.6480654972802045e-05,
+      "loss": 0.5558,
+      "step": 337
+    },
+    {
+      "epoch": 1.368421052631579,
+      "grad_norm": 0.6151357484379033,
+      "learning_rate": 3.644854871713611e-05,
+      "loss": 0.5384,
+      "step": 338
+    },
+    {
+      "epoch": 1.3724696356275303,
+      "grad_norm": 0.6269058086246903,
+      "learning_rate": 3.6416310939312924e-05,
+      "loss": 0.5494,
+      "step": 339
+    },
+    {
+      "epoch": 1.376518218623482,
+      "grad_norm": 0.6297707054487486,
+      "learning_rate": 3.638394189710493e-05,
+      "loss": 0.5575,
+      "step": 340
+    },
+    {
+      "epoch": 1.3805668016194332,
+      "grad_norm": 0.6507298977702192,
+      "learning_rate": 3.635144184933412e-05,
+      "loss": 0.5403,
+      "step": 341
+    },
+    {
+      "epoch": 1.3846153846153846,
+      "grad_norm": 0.9305628350037801,
+      "learning_rate": 3.6318811055869994e-05,
+      "loss": 0.5665,
+      "step": 342
+    },
+    {
+      "epoch": 1.3886639676113361,
+      "grad_norm": 0.8130256614482996,
+      "learning_rate": 3.628604977762752e-05,
+      "loss": 0.5497,
+      "step": 343
+    },
+    {
+      "epoch": 1.3927125506072875,
+      "grad_norm": 0.5966336738329922,
+      "learning_rate": 3.6253158276565004e-05,
+      "loss": 0.5497,
+      "step": 344
+    },
+    {
+      "epoch": 1.3967611336032388,
+      "grad_norm": 0.6615941139224183,
+      "learning_rate": 3.6220136815682016e-05,
+      "loss": 0.5496,
+      "step": 345
+    },
+    {
+      "epoch": 1.4008097165991904,
+      "grad_norm": 0.9606093364777856,
+      "learning_rate": 3.618698565901727e-05,
+      "loss": 0.5582,
+      "step": 346
+    },
+    {
+      "epoch": 1.4048582995951417,
+      "grad_norm": 0.4755487069411029,
+      "learning_rate": 3.615370507164654e-05,
+      "loss": 0.562,
+      "step": 347
+    },
+    {
+      "epoch": 1.408906882591093,
+      "grad_norm": 1.1088085224588071,
+      "learning_rate": 3.61202953196805e-05,
+      "loss": 0.5505,
+      "step": 348
+    },
+    {
+      "epoch": 1.4129554655870447,
+      "grad_norm": 1.103626556784009,
+      "learning_rate": 3.6086756670262654e-05,
+      "loss": 0.5618,
+      "step": 349
+    },
+    {
+      "epoch": 1.417004048582996,
+      "grad_norm": 0.5207579396963141,
+      "learning_rate": 3.6053089391567135e-05,
+      "loss": 0.5464,
+      "step": 350
+    },
+    {
+      "epoch": 1.4210526315789473,
+      "grad_norm": 1.2454825328874004,
+      "learning_rate": 3.601929375279662e-05,
+      "loss": 0.5592,
+      "step": 351
+    },
+    {
+      "epoch": 1.425101214574899,
+      "grad_norm": 0.936932813192785,
+      "learning_rate": 3.598537002418012e-05,
+      "loss": 0.5431,
+      "step": 352
+    },
+    {
+      "epoch": 1.4291497975708503,
+      "grad_norm": 0.8301892175877633,
+      "learning_rate": 3.5951318476970885e-05,
+      "loss": 0.5563,
+      "step": 353
+    },
+    {
+      "epoch": 1.4331983805668016,
+      "grad_norm": 0.8580637360454386,
+      "learning_rate": 3.591713938344416e-05,
+      "loss": 0.5451,
+      "step": 354
+    },
+    {
+      "epoch": 1.4372469635627532,
+      "grad_norm": 0.6628436929097958,
+      "learning_rate": 3.588283301689507e-05,
+      "loss": 0.5566,
+      "step": 355
+    },
+    {
+      "epoch": 1.4412955465587045,
+      "grad_norm": 0.8031297583772211,
+      "learning_rate": 3.5848399651636424e-05,
+      "loss": 0.5373,
+      "step": 356
+    },
+    {
+      "epoch": 1.4453441295546559,
+      "grad_norm": 0.8218495992635328,
+      "learning_rate": 3.5813839562996473e-05,
+      "loss": 0.5535,
+      "step": 357
+    },
+    {
+      "epoch": 1.4493927125506074,
+      "grad_norm": 0.9963450396735369,
+      "learning_rate": 3.5779153027316775e-05,
+      "loss": 0.5513,
+      "step": 358
+    },
+    {
+      "epoch": 1.4534412955465588,
+      "grad_norm": 0.9699443150693713,
+      "learning_rate": 3.574434032194994e-05,
+      "loss": 0.5496,
+      "step": 359
+    },
+    {
+      "epoch": 1.45748987854251,
+      "grad_norm": 0.9878076231476901,
+      "learning_rate": 3.5709401725257436e-05,
+      "loss": 0.5464,
+      "step": 360
+    },
+    {
+      "epoch": 1.4615384615384617,
+      "grad_norm": 0.819377974539712,
+      "learning_rate": 3.5674337516607346e-05,
+      "loss": 0.56,
+      "step": 361
+    },
+    {
+      "epoch": 1.465587044534413,
+      "grad_norm": 0.7192635321230456,
+      "learning_rate": 3.563914797637216e-05,
+      "loss": 0.535,
+      "step": 362
+    },
+    {
+      "epoch": 1.4696356275303644,
+      "grad_norm": 1.3140278223423238,
+      "learning_rate": 3.560383338592649e-05,
+      "loss": 0.5544,
+      "step": 363
+    },
+    {
+      "epoch": 1.4736842105263157,
+      "grad_norm": 0.6430500375364925,
+      "learning_rate": 3.556839402764487e-05,
+      "loss": 0.5587,
+      "step": 364
+    },
+    {
+      "epoch": 1.4777327935222673,
+      "grad_norm": 1.0368566152818102,
+      "learning_rate": 3.553283018489946e-05,
+      "loss": 0.5514,
+      "step": 365
+    },
+    {
+      "epoch": 1.4817813765182186,
+      "grad_norm": 1.3471988196120237,
+      "learning_rate": 3.54971421420578e-05,
+      "loss": 0.5487,
+      "step": 366
+    },
+    {
+      "epoch": 1.48582995951417,
+      "grad_norm": 0.711764914825473,
+      "learning_rate": 3.5461330184480524e-05,
+      "loss": 0.5507,
+      "step": 367
+    },
+    {
+      "epoch": 1.4898785425101215,
+      "grad_norm": 1.4782668517500157,
+      "learning_rate": 3.54253945985191e-05,
+      "loss": 0.5539,
+      "step": 368
+    },
+    {
+      "epoch": 1.4939271255060729,
+      "grad_norm": 0.7134269602300389,
+      "learning_rate": 3.5389335671513524e-05,
+      "loss": 0.5592,
+      "step": 369
+    },
+    {
+      "epoch": 1.4979757085020242,
+      "grad_norm": 1.5176633807553153,
+      "learning_rate": 3.5353153691789986e-05,
+      "loss": 0.5572,
+      "step": 370
+    },
+    {
+      "epoch": 1.5020242914979756,
+      "grad_norm": 0.7104816536182004,
+      "learning_rate": 3.531684894865866e-05,
+      "loss": 0.5372,
+      "step": 371
+    },
+    {
+      "epoch": 1.5060728744939271,
+      "grad_norm": 1.4754880574118892,
+      "learning_rate": 3.528042173241128e-05,
+      "loss": 0.56,
+      "step": 372
+    },
+    {
+      "epoch": 1.5101214574898787,
+      "grad_norm": 0.7713496780441383,
+      "learning_rate": 3.524387233431892e-05,
+      "loss": 0.5454,
+      "step": 373
+    },
+    {
+      "epoch": 1.5141700404858298,
+      "grad_norm": 1.4010600578550136,
+      "learning_rate": 3.520720104662958e-05,
+      "loss": 0.5546,
+      "step": 374
+    },
+    {
+      "epoch": 1.5182186234817814,
+      "grad_norm": 0.888829624134531,
+      "learning_rate": 3.5170408162565904e-05,
+      "loss": 0.5446,
+      "step": 375
+    },
+    {
+      "epoch": 1.522267206477733,
+      "grad_norm": 1.2095385465073498,
+      "learning_rate": 3.513349397632281e-05,
+      "loss": 0.5604,
+      "step": 376
+    },
+    {
+      "epoch": 1.526315789473684,
+      "grad_norm": 0.9885964017772495,
+      "learning_rate": 3.5096458783065145e-05,
+      "loss": 0.5445,
+      "step": 377
+    },
+    {
+      "epoch": 1.5303643724696356,
+      "grad_norm": 1.0949808890431958,
+      "learning_rate": 3.505930287892533e-05,
+      "loss": 0.5441,
+      "step": 378
+    },
+    {
+      "epoch": 1.5344129554655872,
+      "grad_norm": 0.8846645624830788,
+      "learning_rate": 3.502202656100099e-05,
+      "loss": 0.5397,
+      "step": 379
+    },
+    {
+      "epoch": 1.5384615384615383,
+      "grad_norm": 0.9651536930656075,
+      "learning_rate": 3.498463012735256e-05,
+      "loss": 0.5601,
+      "step": 380
+    },
+    {
+      "epoch": 1.54251012145749,
+      "grad_norm": 0.7436246789872134,
+      "learning_rate": 3.494711387700094e-05,
+      "loss": 0.5453,
+      "step": 381
+    },
+    {
+      "epoch": 1.5465587044534415,
+      "grad_norm": 0.9946780390974589,
+      "learning_rate": 3.490947810992505e-05,
+      "loss": 0.5522,
+      "step": 382
+    },
+    {
+      "epoch": 1.5506072874493926,
+      "grad_norm": 0.8818403529891639,
+      "learning_rate": 3.48717231270595e-05,
+      "loss": 0.5496,
+      "step": 383
+    },
+    {
+      "epoch": 1.5546558704453441,
+      "grad_norm": 0.9700687575276646,
+      "learning_rate": 3.483384923029211e-05,
+      "loss": 0.5499,
+      "step": 384
+    },
+    {
+      "epoch": 1.5587044534412957,
+      "grad_norm": 0.9518273680514633,
+      "learning_rate": 3.479585672246156e-05,
+      "loss": 0.5542,
+      "step": 385
+    },
+    {
+      "epoch": 1.5627530364372468,
+      "grad_norm": 0.8401946189650517,
+      "learning_rate": 3.4757745907354924e-05,
+      "loss": 0.5387,
+      "step": 386
+    },
+    {
+      "epoch": 1.5668016194331984,
+      "grad_norm": 1.0540452834648122,
+      "learning_rate": 3.471951708970527e-05,
+      "loss": 0.5479,
+      "step": 387
+    },
+    {
+      "epoch": 1.5708502024291497,
+      "grad_norm": 0.8462579588782351,
+      "learning_rate": 3.468117057518921e-05,
+      "loss": 0.5439,
+      "step": 388
+    },
+    {
+      "epoch": 1.574898785425101,
+      "grad_norm": 0.9803806648913636,
+      "learning_rate": 3.4642706670424426e-05,
+      "loss": 0.5542,
+      "step": 389
+    },
+    {
+      "epoch": 1.5789473684210527,
+      "grad_norm": 1.0888060139886786,
+      "learning_rate": 3.460412568296731e-05,
+      "loss": 0.5562,
+      "step": 390
+    },
+    {
+      "epoch": 1.582995951417004,
+      "grad_norm": 0.9038210760420289,
+      "learning_rate": 3.456542792131039e-05,
+      "loss": 0.5479,
+      "step": 391
+    },
+    {
+      "epoch": 1.5870445344129553,
+      "grad_norm": 0.9781672100371229,
+      "learning_rate": 3.4526613694879936e-05,
+      "loss": 0.5401,
+      "step": 392
+    },
+    {
+      "epoch": 1.591093117408907,
+      "grad_norm": 0.9513162031148149,
+      "learning_rate": 3.448768331403348e-05,
+      "loss": 0.5498,
+      "step": 393
+    },
+    {
+      "epoch": 1.5951417004048583,
+      "grad_norm": 0.9916178440436256,
+      "learning_rate": 3.44486370900573e-05,
+      "loss": 0.5639,
+      "step": 394
+    },
+    {
+      "epoch": 1.5991902834008096,
+      "grad_norm": 0.7953984981248674,
+      "learning_rate": 3.440947533516396e-05,
+      "loss": 0.5427,
+      "step": 395
+    },
+    {
+      "epoch": 1.6032388663967612,
+      "grad_norm": 0.7353961717781602,
+      "learning_rate": 3.4370198362489816e-05,
+      "loss": 0.5566,
+      "step": 396
+    },
+    {
+      "epoch": 1.6072874493927125,
+      "grad_norm": 0.8673833599227521,
+      "learning_rate": 3.43308064860925e-05,
+      "loss": 0.5409,
+      "step": 397
+    },
+    {
+      "epoch": 1.6113360323886639,
+      "grad_norm": 0.6906413837551205,
+      "learning_rate": 3.4291300020948393e-05,
+      "loss": 0.5439,
+      "step": 398
+    },
+    {
+      "epoch": 1.6153846153846154,
+      "grad_norm": 0.5958470564551748,
+      "learning_rate": 3.425167928295015e-05,
+      "loss": 0.5482,
+      "step": 399
+    },
+    {
+      "epoch": 1.6194331983805668,
+      "grad_norm": 0.7681715124806431,
+      "learning_rate": 3.421194458890411e-05,
+      "loss": 0.5558,
+      "step": 400
+    },
+    {
+      "epoch": 1.623481781376518,
+      "grad_norm": 0.6211045812583129,
+      "learning_rate": 3.4172096256527865e-05,
+      "loss": 0.5644,
+      "step": 401
+    },
+    {
+      "epoch": 1.6275303643724697,
+      "grad_norm": 0.4927453405881622,
+      "learning_rate": 3.4132134604447595e-05,
+      "loss": 0.5468,
+      "step": 402
+    },
+    {
+      "epoch": 1.631578947368421,
+      "grad_norm": 0.5179996979719879,
+      "learning_rate": 3.409205995219561e-05,
+      "loss": 0.5502,
+      "step": 403
+    },
+    {
+      "epoch": 1.6356275303643724,
+      "grad_norm": 0.577999985486223,
+      "learning_rate": 3.4051872620207765e-05,
+      "loss": 0.5509,
+      "step": 404
+    },
+    {
+      "epoch": 1.639676113360324,
+      "grad_norm": 0.6743646161005886,
+      "learning_rate": 3.40115729298209e-05,
+      "loss": 0.5518,
+      "step": 405
+    },
+    {
+      "epoch": 1.6437246963562753,
+      "grad_norm": 0.5264154995399457,
+      "learning_rate": 3.397116120327025e-05,
+      "loss": 0.5526,
+      "step": 406
+    },
+    {
+      "epoch": 1.6477732793522266,
+      "grad_norm": 0.69735820995284,
+      "learning_rate": 3.3930637763686916e-05,
+      "loss": 0.5442,
+      "step": 407
+    },
+    {
+      "epoch": 1.6518218623481782,
+      "grad_norm": 0.4954364824513076,
+      "learning_rate": 3.389000293509524e-05,
+      "loss": 0.5492,
+      "step": 408
+    },
+    {
+      "epoch": 1.6558704453441295,
+      "grad_norm": 0.6138114840665702,
+      "learning_rate": 3.3849257042410225e-05,
+      "loss": 0.5391,
+      "step": 409
+    },
+    {
+      "epoch": 1.6599190283400809,
+      "grad_norm": 0.5802346854157112,
+      "learning_rate": 3.380840041143494e-05,
+      "loss": 0.5509,
+      "step": 410
+    },
+    {
+      "epoch": 1.6639676113360324,
+      "grad_norm": 0.5602587768530689,
+      "learning_rate": 3.376743336885792e-05,
+      "loss": 0.5674,
+      "step": 411
+    },
+    {
+      "epoch": 1.6680161943319838,
+      "grad_norm": 0.6429942543376022,
+      "learning_rate": 3.372635624225055e-05,
+      "loss": 0.5457,
+      "step": 412
+    },
+    {
+      "epoch": 1.6720647773279351,
+      "grad_norm": 0.5483689682966513,
+      "learning_rate": 3.3685169360064445e-05,
+      "loss": 0.5493,
+      "step": 413
+    },
+    {
+      "epoch": 1.6761133603238867,
+      "grad_norm": 0.5699263580420085,
+      "learning_rate": 3.3643873051628804e-05,
+      "loss": 0.5507,
+      "step": 414
+    },
+    {
+      "epoch": 1.680161943319838,
+      "grad_norm": 0.5997342394340943,
+      "learning_rate": 3.3602467647147814e-05,
+      "loss": 0.5342,
+      "step": 415
+    },
+    {
+      "epoch": 1.6842105263157894,
+      "grad_norm": 0.6182111406170856,
+      "learning_rate": 3.356095347769799e-05,
+      "loss": 0.5626,
+      "step": 416
+    },
+    {
+      "epoch": 1.688259109311741,
+      "grad_norm": 0.4673536168865773,
+      "learning_rate": 3.351933087522552e-05,
+      "loss": 0.5313,
+      "step": 417
+    },
+    {
+      "epoch": 1.6923076923076923,
+      "grad_norm": 0.5941017576079565,
+      "learning_rate": 3.3477600172543626e-05,
+      "loss": 0.5493,
+      "step": 418
+    },
+    {
+      "epoch": 1.6963562753036436,
+      "grad_norm": 0.5140872633943973,
+      "learning_rate": 3.3435761703329894e-05,
+      "loss": 0.5466,
+      "step": 419
+    },
+    {
+      "epoch": 1.7004048582995952,
+      "grad_norm": 0.6227580793861384,
+      "learning_rate": 3.3393815802123606e-05,
+      "loss": 0.5523,
+      "step": 420
+    },
+    {
+      "epoch": 1.7044534412955465,
+      "grad_norm": 0.48935927185040884,
+      "learning_rate": 3.335176280432308e-05,
+      "loss": 0.5407,
+      "step": 421
+    },
+    {
+      "epoch": 1.708502024291498,
+      "grad_norm": 0.49074787091021743,
+      "learning_rate": 3.330960304618295e-05,
+      "loss": 0.5428,
+      "step": 422
+    },
+    {
+      "epoch": 1.7125506072874495,
+      "grad_norm": 0.4207039599208339,
+      "learning_rate": 3.326733686481153e-05,
+      "loss": 0.5467,
+      "step": 423
+    },
+    {
+      "epoch": 1.7165991902834008,
+      "grad_norm": 0.4851020387901387,
+      "learning_rate": 3.3224964598168084e-05,
+      "loss": 0.5412,
+      "step": 424
+    },
+    {
+      "epoch": 1.7206477732793521,
+      "grad_norm": 0.5084500368309117,
+      "learning_rate": 3.318248658506012e-05,
+      "loss": 0.5462,
+      "step": 425
+    },
+    {
+      "epoch": 1.7246963562753037,
+      "grad_norm": 0.5467731922962201,
+      "learning_rate": 3.313990316514069e-05,
+      "loss": 0.5521,
+      "step": 426
+    },
+    {
+      "epoch": 1.728744939271255,
+      "grad_norm": 0.6203315498622513,
+      "learning_rate": 3.309721467890571e-05,
+      "loss": 0.5481,
+      "step": 427
+    },
+    {
+      "epoch": 1.7327935222672064,
+      "grad_norm": 0.683328991682802,
+      "learning_rate": 3.305442146769115e-05,
+      "loss": 0.5404,
+      "step": 428
+    },
+    {
+      "epoch": 1.736842105263158,
+      "grad_norm": 0.5139376854876271,
+      "learning_rate": 3.30115238736704e-05,
+      "loss": 0.5366,
+      "step": 429
+    },
+    {
+      "epoch": 1.7408906882591093,
+      "grad_norm": 0.676226413244196,
+      "learning_rate": 3.296852223985148e-05,
+      "loss": 0.5442,
+      "step": 430
+    },
+    {
+      "epoch": 1.7449392712550607,
+      "grad_norm": 0.6203888556188033,
+      "learning_rate": 3.2925416910074315e-05,
+      "loss": 0.5484,
+      "step": 431
+    },
+    {
+      "epoch": 1.7489878542510122,
+      "grad_norm": 0.5219945489736889,
+      "learning_rate": 3.288220822900796e-05,
+      "loss": 0.5345,
+      "step": 432
+    },
+    {
+      "epoch": 1.7530364372469636,
+      "grad_norm": 0.686524629399908,
+      "learning_rate": 3.283889654214788e-05,
+      "loss": 0.5486,
+      "step": 433
+    },
+    {
+      "epoch": 1.757085020242915,
+      "grad_norm": 0.5107283766116995,
+      "learning_rate": 3.279548219581317e-05,
+      "loss": 0.544,
+      "step": 434
+    },
+    {
+      "epoch": 1.7611336032388665,
+      "grad_norm": 0.667148922936172,
+      "learning_rate": 3.275196553714379e-05,
+      "loss": 0.546,
+      "step": 435
+    },
+    {
+      "epoch": 1.7651821862348178,
+      "grad_norm": 0.5952709555267286,
+      "learning_rate": 3.270834691409778e-05,
+      "loss": 0.5374,
+      "step": 436
+    },
+    {
+      "epoch": 1.7692307692307692,
+      "grad_norm": 0.5976525187391994,
+      "learning_rate": 3.266462667544849e-05,
+      "loss": 0.5548,
+      "step": 437
+    },
+    {
+      "epoch": 1.7732793522267207,
+      "grad_norm": 0.7039673481316785,
+      "learning_rate": 3.262080517078178e-05,
+      "loss": 0.5476,
+      "step": 438
+    },
+    {
+      "epoch": 1.777327935222672,
+      "grad_norm": 0.4293157860988465,
+      "learning_rate": 3.257688275049323e-05,
+      "loss": 0.5367,
+      "step": 439
+    },
+    {
+      "epoch": 1.7813765182186234,
+      "grad_norm": 0.5698800217041271,
+      "learning_rate": 3.253285976578535e-05,
+      "loss": 0.5501,
+      "step": 440
+    },
+    {
+      "epoch": 1.785425101214575,
+      "grad_norm": 0.5581181285385733,
+      "learning_rate": 3.2488736568664756e-05,
+      "loss": 0.5492,
+      "step": 441
+    },
+    {
+      "epoch": 1.7894736842105263,
+      "grad_norm": 0.5912884482860133,
+      "learning_rate": 3.244451351193935e-05,
+      "loss": 0.5452,
+      "step": 442
+    },
+    {
+      "epoch": 1.7935222672064777,
+      "grad_norm": 0.5757472562807558,
+      "learning_rate": 3.240019094921551e-05,
+      "loss": 0.5513,
+      "step": 443
+    },
+    {
+      "epoch": 1.7975708502024292,
+      "grad_norm": 0.5260646112130739,
+      "learning_rate": 3.235576923489527e-05,
+      "loss": 0.545,
+      "step": 444
+    },
+    {
+      "epoch": 1.8016194331983806,
+      "grad_norm": 0.5271244417469538,
+      "learning_rate": 3.2311248724173476e-05,
+      "loss": 0.554,
+      "step": 445
+    },
+    {
+      "epoch": 1.805668016194332,
+      "grad_norm": 0.4780482428504692,
+      "learning_rate": 3.226662977303494e-05,
+      "loss": 0.5424,
+      "step": 446
+    },
+    {
+      "epoch": 1.8097165991902835,
+      "grad_norm": 0.5448503643148233,
+      "learning_rate": 3.2221912738251626e-05,
+      "loss": 0.5525,
+      "step": 447
+    },
+    {
+      "epoch": 1.8137651821862348,
+      "grad_norm": 0.6125339762685589,
+      "learning_rate": 3.2177097977379736e-05,
+      "loss": 0.5439,
+      "step": 448
+    },
+    {
+      "epoch": 1.8178137651821862,
+      "grad_norm": 0.6434789705799597,
+      "learning_rate": 3.21321858487569e-05,
+      "loss": 0.5436,
+      "step": 449
+    },
+    {
+      "epoch": 1.8218623481781377,
+      "grad_norm": 0.5496126193068852,
+      "learning_rate": 3.208717671149931e-05,
+      "loss": 0.537,
+      "step": 450
+    },
+    {
+      "epoch": 1.825910931174089,
+      "grad_norm": 0.4002515546707261,
+      "learning_rate": 3.204207092549882e-05,
+      "loss": 0.5311,
+      "step": 451
+    },
+    {
+      "epoch": 1.8299595141700404,
+      "grad_norm": 0.6040881202685932,
+      "learning_rate": 3.19968688514201e-05,
+      "loss": 0.5372,
+      "step": 452
+    },
+    {
+      "epoch": 1.834008097165992,
+      "grad_norm": 0.589220518445873,
+      "learning_rate": 3.195157085069774e-05,
+      "loss": 0.5478,
+      "step": 453
+    },
+    {
+      "epoch": 1.8380566801619433,
+      "grad_norm": 0.6478555060039672,
+      "learning_rate": 3.190617728553332e-05,
+      "loss": 0.5474,
+      "step": 454
+    },
+    {
+      "epoch": 1.8421052631578947,
+      "grad_norm": 0.4878053535438857,
+      "learning_rate": 3.1860688518892606e-05,
+      "loss": 0.5506,
+      "step": 455
+    },
+    {
+      "epoch": 1.8461538461538463,
+      "grad_norm": 0.440919196157813,
+      "learning_rate": 3.181510491450254e-05,
+      "loss": 0.5407,
+      "step": 456
+    },
+    {
+      "epoch": 1.8502024291497976,
+      "grad_norm": 0.4643669623097734,
+      "learning_rate": 3.176942683684842e-05,
+      "loss": 0.55,
+      "step": 457
+    },
+    {
+      "epoch": 1.854251012145749,
+      "grad_norm": 0.5082492258374482,
+      "learning_rate": 3.1723654651170934e-05,
+      "loss": 0.5475,
+      "step": 458
+    },
+    {
+      "epoch": 1.8582995951417005,
+      "grad_norm": 0.539929395171649,
+      "learning_rate": 3.167778872346327e-05,
+      "loss": 0.5472,
+      "step": 459
+    },
+    {
+      "epoch": 1.8623481781376519,
+      "grad_norm": 0.4542963233670436,
+      "learning_rate": 3.163182942046815e-05,
+      "loss": 0.5287,
+      "step": 460
+    },
+    {
+      "epoch": 1.8663967611336032,
+      "grad_norm": 0.44233519309179914,
+      "learning_rate": 3.158577710967494e-05,
+      "loss": 0.5367,
+      "step": 461
+    },
+    {
+      "epoch": 1.8704453441295548,
+      "grad_norm": 0.5681281628375966,
+      "learning_rate": 3.153963215931669e-05,
+      "loss": 0.5441,
+      "step": 462
+    },
+    {
+      "epoch": 1.874493927125506,
+      "grad_norm": 0.4522707410243331,
+      "learning_rate": 3.14933949383672e-05,
+      "loss": 0.5401,
+      "step": 463
+    },
+    {
+      "epoch": 1.8785425101214575,
+      "grad_norm": 0.493481619450058,
+      "learning_rate": 3.144706581653805e-05,
+      "loss": 0.5415,
+      "step": 464
+    },
+    {
+      "epoch": 1.882591093117409,
+      "grad_norm": 0.4795952077360407,
+      "learning_rate": 3.140064516427566e-05,
+      "loss": 0.5365,
+      "step": 465
+    },
+    {
+      "epoch": 1.8866396761133604,
+      "grad_norm": 0.41544204821228015,
+      "learning_rate": 3.135413335275833e-05,
+      "loss": 0.5473,
+      "step": 466
+    },
+    {
+      "epoch": 1.8906882591093117,
+      "grad_norm": 0.4909503366786034,
+      "learning_rate": 3.130753075389327e-05,
+      "loss": 0.5383,
+      "step": 467
+    },
+    {
+      "epoch": 1.8947368421052633,
+      "grad_norm": 0.38862670102149194,
+      "learning_rate": 3.1260837740313625e-05,
+      "loss": 0.5361,
+      "step": 468
+    },
+    {
+      "epoch": 1.8987854251012146,
+      "grad_norm": 0.46688286532629614,
+      "learning_rate": 3.121405468537547e-05,
+      "loss": 0.5393,
+      "step": 469
+    },
+    {
+      "epoch": 1.902834008097166,
+      "grad_norm": 0.3832571089324067,
+      "learning_rate": 3.116718196315488e-05,
+      "loss": 0.5363,
+      "step": 470
+    },
+    {
+      "epoch": 1.9068825910931175,
+      "grad_norm": 0.49437901143350155,
+      "learning_rate": 3.1120219948444894e-05,
+      "loss": 0.5422,
+      "step": 471
+    },
+    {
+      "epoch": 1.9109311740890689,
+      "grad_norm": 0.4788832345778935,
+      "learning_rate": 3.107316901675252e-05,
+      "loss": 0.5374,
+      "step": 472
+    },
+    {
+      "epoch": 1.9149797570850202,
+      "grad_norm": 0.4120469757435154,
+      "learning_rate": 3.1026029544295745e-05,
+      "loss": 0.5404,
+      "step": 473
+    },
+    {
+      "epoch": 1.9190283400809718,
+      "grad_norm": 0.5354308240168897,
+      "learning_rate": 3.097880190800057e-05,
+      "loss": 0.55,
+      "step": 474
+    },
+    {
+      "epoch": 1.9230769230769231,
+      "grad_norm": 0.5436233247893849,
+      "learning_rate": 3.093148648549788e-05,
+      "loss": 0.5431,
+      "step": 475
+    },
+    {
+      "epoch": 1.9271255060728745,
+      "grad_norm": 0.418374491922879,
+      "learning_rate": 3.088408365512055e-05,
+      "loss": 0.5389,
+      "step": 476
+    },
+    {
+      "epoch": 1.931174089068826,
+      "grad_norm": 0.560549293451037,
+      "learning_rate": 3.083659379590034e-05,
+      "loss": 0.5438,
+      "step": 477
+    },
+    {
+      "epoch": 1.9352226720647774,
+      "grad_norm": 0.5064028726025557,
+      "learning_rate": 3.0789017287564913e-05,
+      "loss": 0.5273,
+      "step": 478
+    },
+    {
+      "epoch": 1.9392712550607287,
+      "grad_norm": 0.49940525190049856,
+      "learning_rate": 3.074135451053475e-05,
+      "loss": 0.5348,
+      "step": 479
+    },
+    {
+      "epoch": 1.9433198380566803,
+      "grad_norm": 0.751505767414857,
+      "learning_rate": 3.069360584592015e-05,
+      "loss": 0.5488,
+      "step": 480
+    },
+    {
+      "epoch": 1.9473684210526314,
+      "grad_norm": 0.5742987715566087,
+      "learning_rate": 3.064577167551816e-05,
+      "loss": 0.5337,
+      "step": 481
+    },
+    {
+      "epoch": 1.951417004048583,
+      "grad_norm": 0.5917716133863006,
+      "learning_rate": 3.059785238180953e-05,
+      "loss": 0.5203,
+      "step": 482
+    },
+    {
+      "epoch": 1.9554655870445345,
+      "grad_norm": 0.7196658252448703,
+      "learning_rate": 3.054984834795566e-05,
+      "loss": 0.5306,
+      "step": 483
+    },
+    {
+      "epoch": 1.9595141700404857,
+      "grad_norm": 0.465175875544765,
+      "learning_rate": 3.0501759957795537e-05,
+      "loss": 0.5375,
+      "step": 484
+    },
+    {
+      "epoch": 1.9635627530364372,
+      "grad_norm": 0.8636398089610432,
+      "learning_rate": 3.0453587595842627e-05,
+      "loss": 0.5434,
+      "step": 485
+    },
+    {
+      "epoch": 1.9676113360323888,
+      "grad_norm": 0.7284265142462247,
+      "learning_rate": 3.0405331647281856e-05,
+      "loss": 0.5442,
+      "step": 486
+    },
+    {
+      "epoch": 1.97165991902834,
+      "grad_norm": 0.44847039316006654,
+      "learning_rate": 3.0356992497966508e-05,
+      "loss": 0.5224,
+      "step": 487
+    },
+    {
+      "epoch": 1.9757085020242915,
+      "grad_norm": 0.8830290737162946,
+      "learning_rate": 3.0308570534415137e-05,
+      "loss": 0.5444,
+      "step": 488
+    },
+    {
+      "epoch": 1.979757085020243,
+      "grad_norm": 0.717496367445215,
+      "learning_rate": 3.0260066143808467e-05,
+      "loss": 0.5379,
+      "step": 489
+    },
+    {
+      "epoch": 1.9838056680161942,
+      "grad_norm": 0.5977121012947413,
+      "learning_rate": 3.021147971398632e-05,
+      "loss": 0.5345,
+      "step": 490
+    },
+    {
+      "epoch": 1.9878542510121457,
+      "grad_norm": 0.7114902721719124,
+      "learning_rate": 3.0162811633444478e-05,
+      "loss": 0.5468,
+      "step": 491
+    },
+    {
+      "epoch": 1.9919028340080973,
+      "grad_norm": 0.6314810464072685,
+      "learning_rate": 3.0114062291331635e-05,
+      "loss": 0.5368,
+      "step": 492
+    },
+    {
+      "epoch": 1.9959514170040484,
+      "grad_norm": 0.615921229307917,
+      "learning_rate": 3.0065232077446217e-05,
+      "loss": 0.5429,
+      "step": 493
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.6055483970293614,
+      "learning_rate": 3.001632138223332e-05,
+      "loss": 0.5428,
+      "step": 494
+    },
+    {
+      "epoch": 2.0040485829959516,
+      "grad_norm": 0.5058086618840303,
+      "learning_rate": 2.9967330596781565e-05,
+      "loss": 0.4702,
+      "step": 495
+    },
+    {
+      "epoch": 2.0080971659919027,
+      "grad_norm": 0.5056601058909136,
+      "learning_rate": 2.9918260112819952e-05,
+      "loss": 0.4811,
+      "step": 496
+    },
+    {
+      "epoch": 2.0121457489878543,
+      "grad_norm": 0.565316410519056,
+      "learning_rate": 2.9869110322714778e-05,
+      "loss": 0.491,
+      "step": 497
+    },
+    {
+      "epoch": 2.016194331983806,
+      "grad_norm": 0.5735113428670706,
+      "learning_rate": 2.9819881619466447e-05,
+      "loss": 0.4809,
+      "step": 498
+    },
+    {
+      "epoch": 2.020242914979757,
+      "grad_norm": 0.5339879108496473,
+      "learning_rate": 2.9770574396706364e-05,
+      "loss": 0.4835,
+      "step": 499
+    },
+    {
+      "epoch": 2.0242914979757085,
+      "grad_norm": 0.6974691364950177,
+      "learning_rate": 2.972118904869376e-05,
+      "loss": 0.4844,
+      "step": 500
+    },
+    {
+      "epoch": 2.02834008097166,
+      "grad_norm": 0.6500832191193383,
+      "learning_rate": 2.9671725970312563e-05,
+      "loss": 0.4817,
+      "step": 501
+    },
+    {
+      "epoch": 2.032388663967611,
+      "grad_norm": 0.5889610107366744,
+      "learning_rate": 2.9622185557068238e-05,
+      "loss": 0.4725,
+      "step": 502
+    },
+    {
+      "epoch": 2.0364372469635628,
+      "grad_norm": 0.5733748207835166,
+      "learning_rate": 2.95725682050846e-05,
+      "loss": 0.4794,
+      "step": 503
+    },
+    {
+      "epoch": 2.0404858299595143,
+      "grad_norm": 0.5890481518177154,
+      "learning_rate": 2.9522874311100684e-05,
+      "loss": 0.4818,
+      "step": 504
+    },
+    {
+      "epoch": 2.0445344129554655,
+      "grad_norm": 0.5681092017156488,
+      "learning_rate": 2.9473104272467526e-05,
+      "loss": 0.4843,
+      "step": 505
+    },
+    {
+      "epoch": 2.048582995951417,
+      "grad_norm": 0.5956225942360169,
+      "learning_rate": 2.942325848714505e-05,
+      "loss": 0.4865,
+      "step": 506
+    },
+    {
+      "epoch": 2.0526315789473686,
+      "grad_norm": 0.557142814357193,
+      "learning_rate": 2.9373337353698804e-05,
+      "loss": 0.4859,
+      "step": 507
+    },
+    {
+      "epoch": 2.0566801619433197,
+      "grad_norm": 0.4987399810772276,
+      "learning_rate": 2.932334127129686e-05,
+      "loss": 0.4817,
+      "step": 508
+    },
+    {
+      "epoch": 2.0607287449392713,
+      "grad_norm": 0.6313289380093993,
+      "learning_rate": 2.9273270639706547e-05,
+      "loss": 0.4818,
+      "step": 509
+    },
+    {
+      "epoch": 2.064777327935223,
+      "grad_norm": 0.48830199752424264,
+      "learning_rate": 2.922312585929131e-05,
+      "loss": 0.4784,
+      "step": 510
+    },
+    {
+      "epoch": 2.068825910931174,
+      "grad_norm": 0.5340781903419197,
+      "learning_rate": 2.9172907331007478e-05,
+      "loss": 0.4765,
+      "step": 511
+    },
+    {
+      "epoch": 2.0728744939271255,
+      "grad_norm": 0.4987680174299772,
+      "learning_rate": 2.9122615456401053e-05,
+      "loss": 0.4696,
+      "step": 512
+    },
+    {
+      "epoch": 2.076923076923077,
+      "grad_norm": 0.6161140255451967,
+      "learning_rate": 2.9072250637604535e-05,
+      "loss": 0.4895,
+      "step": 513
+    },
+    {
+      "epoch": 2.080971659919028,
+      "grad_norm": 0.5714924218569997,
+      "learning_rate": 2.9021813277333667e-05,
+      "loss": 0.4733,
+      "step": 514
+    },
+    {
+      "epoch": 2.08502024291498,
+      "grad_norm": 0.525163793637916,
+      "learning_rate": 2.8971303778884244e-05,
+      "loss": 0.4735,
+      "step": 515
+    },
+    {
+      "epoch": 2.0890688259109313,
+      "grad_norm": 0.5638774516544787,
+      "learning_rate": 2.8920722546128863e-05,
+      "loss": 0.4839,
+      "step": 516
+    },
+    {
+      "epoch": 2.0931174089068825,
+      "grad_norm": 0.4440864620369476,
+      "learning_rate": 2.887006998351372e-05,
+      "loss": 0.4929,
+      "step": 517
+    },
+    {
+      "epoch": 2.097165991902834,
+      "grad_norm": 0.550984516075476,
+      "learning_rate": 2.881934649605535e-05,
+      "loss": 0.4724,
+      "step": 518
+    },
+    {
+      "epoch": 2.1012145748987856,
+      "grad_norm": 0.5464710162267568,
+      "learning_rate": 2.8768552489337408e-05,
+      "loss": 0.4873,
+      "step": 519
+    },
+    {
+      "epoch": 2.1052631578947367,
+      "grad_norm": 0.47209417577606766,
+      "learning_rate": 2.8717688369507425e-05,
+      "loss": 0.4775,
+      "step": 520
+    },
+    {
+      "epoch": 2.1093117408906883,
+      "grad_norm": 0.49945042682614915,
+      "learning_rate": 2.8666754543273535e-05,
+      "loss": 0.481,
+      "step": 521
+    },
+    {
+      "epoch": 2.11336032388664,
+      "grad_norm": 0.47919558565266873,
+      "learning_rate": 2.8615751417901273e-05,
+      "loss": 0.4773,
+      "step": 522
+    },
+    {
+      "epoch": 2.117408906882591,
+      "grad_norm": 0.40851479133871804,
+      "learning_rate": 2.856467940121025e-05,
+      "loss": 0.4916,
+      "step": 523
+    },
+    {
+      "epoch": 2.1214574898785425,
+      "grad_norm": 0.5727661710844698,
+      "learning_rate": 2.8513538901570964e-05,
+      "loss": 0.4768,
+      "step": 524
+    },
+    {
+      "epoch": 2.125506072874494,
+      "grad_norm": 0.5844723668086866,
+      "learning_rate": 2.846233032790149e-05,
+      "loss": 0.4961,
+      "step": 525
+    },
+    {
+      "epoch": 2.1295546558704452,
+      "grad_norm": 0.5467095546076047,
+      "learning_rate": 2.84110540896642e-05,
+      "loss": 0.4767,
+      "step": 526
+    },
+    {
+      "epoch": 2.133603238866397,
+      "grad_norm": 0.6289333104077229,
+      "learning_rate": 2.8359710596862555e-05,
+      "loss": 0.4878,
+      "step": 527
+    },
+    {
+      "epoch": 2.1376518218623484,
+      "grad_norm": 0.5482859886118306,
+      "learning_rate": 2.8308300260037734e-05,
+      "loss": 0.4823,
+      "step": 528
+    },
+    {
+      "epoch": 2.1417004048582995,
+      "grad_norm": 0.6543499534694505,
+      "learning_rate": 2.8256823490265424e-05,
+      "loss": 0.4918,
+      "step": 529
+    },
+    {
+      "epoch": 2.145748987854251,
+      "grad_norm": 0.5003815318499261,
+      "learning_rate": 2.8205280699152515e-05,
+      "loss": 0.4845,
+      "step": 530
+    },
+    {
+      "epoch": 2.1497975708502026,
+      "grad_norm": 0.5836151908352182,
+      "learning_rate": 2.8153672298833777e-05,
+      "loss": 0.4778,
+      "step": 531
+    },
+    {
+      "epoch": 2.1538461538461537,
+      "grad_norm": 0.5701729983774481,
+      "learning_rate": 2.8101998701968615e-05,
+      "loss": 0.4897,
+      "step": 532
+    },
+    {
+      "epoch": 2.1578947368421053,
+      "grad_norm": 0.43543451799866006,
+      "learning_rate": 2.805026032173773e-05,
+      "loss": 0.4825,
+      "step": 533
+    },
+    {
+      "epoch": 2.161943319838057,
+      "grad_norm": 0.6281710720217452,
+      "learning_rate": 2.7998457571839832e-05,
+      "loss": 0.4861,
+      "step": 534
+    },
+    {
+      "epoch": 2.165991902834008,
+      "grad_norm": 0.5001746610793399,
+      "learning_rate": 2.7946590866488322e-05,
+      "loss": 0.4869,
+      "step": 535
+    },
+    {
+      "epoch": 2.1700404858299596,
+      "grad_norm": 0.4441667354162335,
+      "learning_rate": 2.7894660620408e-05,
+      "loss": 0.4788,
+      "step": 536
+    },
+    {
+      "epoch": 2.174089068825911,
+      "grad_norm": 0.5596400077999788,
+      "learning_rate": 2.784266724883173e-05,
+      "loss": 0.4776,
+      "step": 537
+    },
+    {
+      "epoch": 2.1781376518218623,
+      "grad_norm": 0.3781981266401971,
+      "learning_rate": 2.7790611167497128e-05,
+      "loss": 0.4861,
+      "step": 538
+    },
+    {
+      "epoch": 2.182186234817814,
+      "grad_norm": 0.5504458526858863,
+      "learning_rate": 2.7738492792643224e-05,
+      "loss": 0.4884,
+      "step": 539
+    },
+    {
+      "epoch": 2.1862348178137654,
+      "grad_norm": 0.44056808123493885,
+      "learning_rate": 2.768631254100716e-05,
+      "loss": 0.485,
+      "step": 540
+    },
+    {
+      "epoch": 2.1902834008097165,
+      "grad_norm": 0.5309355546397907,
+      "learning_rate": 2.7634070829820836e-05,
+      "loss": 0.4732,
+      "step": 541
+    },
+    {
+      "epoch": 2.194331983805668,
+      "grad_norm": 0.4900861556489155,
+      "learning_rate": 2.7581768076807592e-05,
+      "loss": 0.4849,
+      "step": 542
+    },
+    {
+      "epoch": 2.1983805668016196,
+      "grad_norm": 0.455604054174655,
+      "learning_rate": 2.7529404700178832e-05,
+      "loss": 0.4865,
+      "step": 543
+    },
+    {
+      "epoch": 2.2024291497975708,
+      "grad_norm": 0.5714683330220061,
+      "learning_rate": 2.7476981118630735e-05,
+      "loss": 0.4884,
+      "step": 544
+    },
+    {
+      "epoch": 2.2064777327935223,
+      "grad_norm": 0.4892068991117692,
+      "learning_rate": 2.742449775134085e-05,
+      "loss": 0.491,
+      "step": 545
+    },
+    {
+      "epoch": 2.2105263157894735,
+      "grad_norm": 0.47339931417295955,
+      "learning_rate": 2.7371955017964788e-05,
+      "loss": 0.4853,
+      "step": 546
+    },
+    {
+      "epoch": 2.214574898785425,
+      "grad_norm": 0.5711822387675246,
+      "learning_rate": 2.7319353338632845e-05,
+      "loss": 0.4826,
+      "step": 547
+    },
+    {
+      "epoch": 2.2186234817813766,
+      "grad_norm": 0.42115002586782063,
+      "learning_rate": 2.7266693133946645e-05,
+      "loss": 0.4853,
+      "step": 548
+    },
+    {
+      "epoch": 2.2226720647773277,
+      "grad_norm": 0.45568476100747374,
+      "learning_rate": 2.7213974824975785e-05,
+      "loss": 0.4892,
+      "step": 549
+    },
+    {
+      "epoch": 2.2267206477732793,
+      "grad_norm": 0.4577316757942157,
+      "learning_rate": 2.7161198833254464e-05,
+      "loss": 0.4843,
+      "step": 550
+    },
+    {
+      "epoch": 2.230769230769231,
+      "grad_norm": 0.4990044513297792,
+      "learning_rate": 2.7108365580778095e-05,
+      "loss": 0.4797,
+      "step": 551
+    },
+    {
+      "epoch": 2.234817813765182,
+      "grad_norm": 0.48002238123699753,
+      "learning_rate": 2.705547548999996e-05,
+      "loss": 0.4897,
+      "step": 552
+    },
+    {
+      "epoch": 2.2388663967611335,
+      "grad_norm": 0.48363996113338587,
+      "learning_rate": 2.7002528983827817e-05,
+      "loss": 0.485,
+      "step": 553
+    },
+    {
+      "epoch": 2.242914979757085,
+      "grad_norm": 0.5537968031436836,
+      "learning_rate": 2.6949526485620517e-05,
+      "loss": 0.4836,
+      "step": 554
+    },
+    {
+      "epoch": 2.246963562753036,
+      "grad_norm": 0.3973725622962676,
+      "learning_rate": 2.6896468419184623e-05,
+      "loss": 0.4901,
+      "step": 555
+    },
+    {
+      "epoch": 2.251012145748988,
+      "grad_norm": 0.4698720034538705,
+      "learning_rate": 2.6843355208771024e-05,
+      "loss": 0.4802,
+      "step": 556
+    },
+    {
+      "epoch": 2.2550607287449393,
+      "grad_norm": 0.4316004455068638,
+      "learning_rate": 2.6790187279071536e-05,
+      "loss": 0.4766,
+      "step": 557
+    },
+    {
+      "epoch": 2.2591093117408905,
+      "grad_norm": 0.3793264526010281,
+      "learning_rate": 2.67369650552155e-05,
+      "loss": 0.4872,
+      "step": 558
+    },
+    {
+      "epoch": 2.263157894736842,
+      "grad_norm": 0.4182071073985035,
+      "learning_rate": 2.6683688962766407e-05,
+      "loss": 0.48,
+      "step": 559
+    },
+    {
+      "epoch": 2.2672064777327936,
+      "grad_norm": 0.40622684949596866,
+      "learning_rate": 2.6630359427718468e-05,
+      "loss": 0.482,
+      "step": 560
+    },
+    {
+      "epoch": 2.2712550607287447,
+      "grad_norm": 0.45467903662008446,
+      "learning_rate": 2.6576976876493225e-05,
+      "loss": 0.4984,
+      "step": 561
+    },
+    {
+      "epoch": 2.2753036437246963,
+      "grad_norm": 0.4517118288536089,
+      "learning_rate": 2.652354173593614e-05,
+      "loss": 0.475,
+      "step": 562
+    },
+    {
+      "epoch": 2.279352226720648,
+      "grad_norm": 0.36278770794523546,
+      "learning_rate": 2.647005443331317e-05,
+      "loss": 0.4908,
+      "step": 563
+    },
+    {
+      "epoch": 2.283400809716599,
+      "grad_norm": 0.5253646301714223,
+      "learning_rate": 2.6416515396307354e-05,
+      "loss": 0.4821,
+      "step": 564
+    },
+    {
+      "epoch": 2.2874493927125505,
+      "grad_norm": 0.38682969299767556,
+      "learning_rate": 2.636292505301542e-05,
+      "loss": 0.4871,
+      "step": 565
+    },
+    {
+      "epoch": 2.291497975708502,
+      "grad_norm": 0.4257834630960836,
+      "learning_rate": 2.630928383194431e-05,
+      "loss": 0.4729,
+      "step": 566
+    },
+    {
+      "epoch": 2.2955465587044532,
+      "grad_norm": 0.43881479360537934,
+      "learning_rate": 2.625559216200781e-05,
+      "loss": 0.4807,
+      "step": 567
+    },
+    {
+      "epoch": 2.299595141700405,
+      "grad_norm": 0.4608550967308715,
+      "learning_rate": 2.620185047252308e-05,
+      "loss": 0.4815,
+      "step": 568
+    },
+    {
+      "epoch": 2.3036437246963564,
+      "grad_norm": 0.42253841924801505,
+      "learning_rate": 2.614805919320723e-05,
+      "loss": 0.4931,
+      "step": 569
+    },
+    {
+      "epoch": 2.3076923076923075,
+      "grad_norm": 0.49006703175721084,
+      "learning_rate": 2.6094218754173895e-05,
+      "loss": 0.4894,
+      "step": 570
+    },
+    {
+      "epoch": 2.311740890688259,
+      "grad_norm": 0.440537011727239,
+      "learning_rate": 2.60403295859298e-05,
+      "loss": 0.4807,
+      "step": 571
+    },
+    {
+      "epoch": 2.3157894736842106,
+      "grad_norm": 0.4308091211994416,
+      "learning_rate": 2.5986392119371297e-05,
+      "loss": 0.4907,
+      "step": 572
+    },
+    {
+      "epoch": 2.3198380566801617,
+      "grad_norm": 0.5706552877494199,
+      "learning_rate": 2.5932406785780928e-05,
+      "loss": 0.489,
+      "step": 573
+    },
+    {
+      "epoch": 2.3238866396761133,
+      "grad_norm": 0.4053036340664704,
+      "learning_rate": 2.5878374016823987e-05,
+      "loss": 0.4849,
+      "step": 574
+    },
+    {
+      "epoch": 2.327935222672065,
+      "grad_norm": 0.46920338136936435,
+      "learning_rate": 2.5824294244545045e-05,
+      "loss": 0.4818,
+      "step": 575
+    },
+    {
+      "epoch": 2.331983805668016,
+      "grad_norm": 0.41295026875521396,
+      "learning_rate": 2.577016790136453e-05,
+      "loss": 0.4805,
+      "step": 576
+    },
+    {
+      "epoch": 2.3360323886639676,
+      "grad_norm": 0.4237290166075254,
+      "learning_rate": 2.5715995420075225e-05,
+      "loss": 0.4819,
+      "step": 577
+    },
+    {
+      "epoch": 2.340080971659919,
+      "grad_norm": 0.40495183567298526,
+      "learning_rate": 2.5661777233838864e-05,
+      "loss": 0.4804,
+      "step": 578
+    },
+    {
+      "epoch": 2.3441295546558703,
+      "grad_norm": 0.4392918778898318,
+      "learning_rate": 2.5607513776182616e-05,
+      "loss": 0.4902,
+      "step": 579
+    },
+    {
+      "epoch": 2.348178137651822,
+      "grad_norm": 0.520797389333115,
+      "learning_rate": 2.5553205480995633e-05,
+      "loss": 0.4833,
+      "step": 580
+    },
+    {
+      "epoch": 2.3522267206477734,
+      "grad_norm": 0.395815010281138,
+      "learning_rate": 2.5498852782525604e-05,
+      "loss": 0.4799,
+      "step": 581
+    },
+    {
+      "epoch": 2.3562753036437245,
+      "grad_norm": 0.4289596366007229,
+      "learning_rate": 2.5444456115375252e-05,
+      "loss": 0.4876,
+      "step": 582
+    },
+    {
+      "epoch": 2.360323886639676,
+      "grad_norm": 0.3761348357069343,
+      "learning_rate": 2.539001591449889e-05,
+      "loss": 0.4861,
+      "step": 583
+    },
+    {
+      "epoch": 2.3643724696356276,
+      "grad_norm": 0.3828796111714548,
+      "learning_rate": 2.53355326151989e-05,
+      "loss": 0.4894,
+      "step": 584
+    },
+    {
+      "epoch": 2.3684210526315788,
+      "grad_norm": 0.40316391850429417,
+      "learning_rate": 2.5281006653122288e-05,
+      "loss": 0.49,
+      "step": 585
+    },
+    {
+      "epoch": 2.3724696356275303,
+      "grad_norm": 0.4226510659587086,
+      "learning_rate": 2.5226438464257217e-05,
+      "loss": 0.4868,
+      "step": 586
+    },
+    {
+      "epoch": 2.376518218623482,
+      "grad_norm": 0.43458432728706947,
+      "learning_rate": 2.5171828484929447e-05,
+      "loss": 0.4873,
+      "step": 587
+    },
+    {
+      "epoch": 2.380566801619433,
+      "grad_norm": 0.4497062793545992,
+      "learning_rate": 2.5117177151798934e-05,
+      "loss": 0.4977,
+      "step": 588
+    },
+    {
+      "epoch": 2.3846153846153846,
+      "grad_norm": 0.41373701210482594,
+      "learning_rate": 2.506248490185627e-05,
+      "loss": 0.4828,
+      "step": 589
+    },
+    {
+      "epoch": 2.388663967611336,
+      "grad_norm": 0.39164657783205226,
+      "learning_rate": 2.500775217241923e-05,
+      "loss": 0.486,
+      "step": 590
+    },
+    {
+      "epoch": 2.3927125506072873,
+      "grad_norm": 0.3713131721088172,
+      "learning_rate": 2.4952979401129266e-05,
+      "loss": 0.4885,
+      "step": 591
+    },
+    {
+      "epoch": 2.396761133603239,
+      "grad_norm": 0.3933199356915679,
+      "learning_rate": 2.4898167025948e-05,
+      "loss": 0.4775,
+      "step": 592
+    },
+    {
+      "epoch": 2.4008097165991904,
+      "grad_norm": 0.40957742089666427,
+      "learning_rate": 2.4843315485153703e-05,
+      "loss": 0.4872,
+      "step": 593
+    },
+    {
+      "epoch": 2.4048582995951415,
+      "grad_norm": 0.4518740135976957,
+      "learning_rate": 2.4788425217337856e-05,
+      "loss": 0.4764,
+      "step": 594
+    },
+    {
+      "epoch": 2.408906882591093,
+      "grad_norm": 0.4287864779867401,
+      "learning_rate": 2.4733496661401572e-05,
+      "loss": 0.4878,
+      "step": 595
+    },
+    {
+      "epoch": 2.4129554655870447,
+      "grad_norm": 0.47869675037729714,
+      "learning_rate": 2.4678530256552114e-05,
+      "loss": 0.4889,
+      "step": 596
+    },
+    {
+      "epoch": 2.417004048582996,
+      "grad_norm": 0.5077874476810672,
+      "learning_rate": 2.46235264422994e-05,
+      "loss": 0.4826,
+      "step": 597
+    },
+    {
+      "epoch": 2.4210526315789473,
+      "grad_norm": 0.5979437353367683,
+      "learning_rate": 2.4568485658452452e-05,
+      "loss": 0.4856,
+      "step": 598
+    },
+    {
+      "epoch": 2.425101214574899,
+      "grad_norm": 0.41088498773992777,
+      "learning_rate": 2.451340834511592e-05,
+      "loss": 0.4752,
+      "step": 599
+    },
+    {
+      "epoch": 2.42914979757085,
+      "grad_norm": 0.6328038790574716,
+      "learning_rate": 2.4458294942686528e-05,
+      "loss": 0.4856,
+      "step": 600
+    },
+    {
+      "epoch": 2.4331983805668016,
+      "grad_norm": 0.4789077659077234,
+      "learning_rate": 2.440314589184958e-05,
+      "loss": 0.4834,
+      "step": 601
+    },
+    {
+      "epoch": 2.437246963562753,
+      "grad_norm": 0.4306254452378808,
+      "learning_rate": 2.434796163357541e-05,
+      "loss": 0.4905,
+      "step": 602
+    },
+    {
+      "epoch": 2.4412955465587043,
+      "grad_norm": 0.5676667077728513,
+      "learning_rate": 2.429274260911588e-05,
+      "loss": 0.488,
+      "step": 603
+    },
+    {
+      "epoch": 2.445344129554656,
+      "grad_norm": 0.4981836692169252,
+      "learning_rate": 2.423748926000084e-05,
+      "loss": 0.4941,
+      "step": 604
+    },
+    {
+      "epoch": 2.4493927125506074,
+      "grad_norm": 0.4208669630270274,
+      "learning_rate": 2.418220202803459e-05,
+      "loss": 0.4913,
+      "step": 605
+    },
+    {
+      "epoch": 2.4534412955465585,
+      "grad_norm": 0.4688820817829445,
+      "learning_rate": 2.412688135529237e-05,
+      "loss": 0.486,
+      "step": 606
+    },
+    {
+      "epoch": 2.45748987854251,
+      "grad_norm": 0.43486896317636126,
+      "learning_rate": 2.4071527684116807e-05,
+      "loss": 0.4817,
+      "step": 607
+    },
+    {
+      "epoch": 2.4615384615384617,
+      "grad_norm": 0.4301220538098788,
+      "learning_rate": 2.4016141457114372e-05,
+      "loss": 0.4761,
+      "step": 608
+    },
+    {
+      "epoch": 2.465587044534413,
+      "grad_norm": 0.5162682741865786,
+      "learning_rate": 2.396072311715187e-05,
+      "loss": 0.4883,
+      "step": 609
+    },
+    {
+      "epoch": 2.4696356275303644,
+      "grad_norm": 0.4010320947254095,
+      "learning_rate": 2.3905273107352863e-05,
+      "loss": 0.481,
+      "step": 610
+    },
+    {
+      "epoch": 2.473684210526316,
+      "grad_norm": 0.4345124114236493,
+      "learning_rate": 2.3849791871094157e-05,
+      "loss": 0.4827,
+      "step": 611
+    },
+    {
+      "epoch": 2.477732793522267,
+      "grad_norm": 0.4757812004178412,
+      "learning_rate": 2.379427985200223e-05,
+      "loss": 0.4876,
+      "step": 612
+    },
+    {
+      "epoch": 2.4817813765182186,
+      "grad_norm": 0.35360302556611894,
+      "learning_rate": 2.373873749394972e-05,
+      "loss": 0.4857,
+      "step": 613
+    },
+    {
+      "epoch": 2.48582995951417,
+      "grad_norm": 0.48779475484643625,
+      "learning_rate": 2.3683165241051836e-05,
+      "loss": 0.49,
+      "step": 614
+    },
+    {
+      "epoch": 2.4898785425101213,
+      "grad_norm": 0.45142835374721213,
+      "learning_rate": 2.3627563537662823e-05,
+      "loss": 0.4874,
+      "step": 615
+    },
+    {
+      "epoch": 2.493927125506073,
+      "grad_norm": 0.45016310478910543,
+      "learning_rate": 2.3571932828372433e-05,
+      "loss": 0.4766,
+      "step": 616
+    },
+    {
+      "epoch": 2.4979757085020244,
+      "grad_norm": 0.4282897699564636,
+      "learning_rate": 2.351627355800233e-05,
+      "loss": 0.491,
+      "step": 617
+    },
+    {
+      "epoch": 2.5020242914979756,
+      "grad_norm": 0.5696591529725176,
+      "learning_rate": 2.346058617160256e-05,
+      "loss": 0.4927,
+      "step": 618
+    },
+    {
+      "epoch": 2.506072874493927,
+      "grad_norm": 0.43043877691517834,
+      "learning_rate": 2.3404871114447976e-05,
+      "loss": 0.4807,
+      "step": 619
+    },
+    {
+      "epoch": 2.5101214574898787,
+      "grad_norm": 0.5772994127391283,
+      "learning_rate": 2.334912883203469e-05,
+      "loss": 0.4803,
+      "step": 620
+    },
+    {
+      "epoch": 2.51417004048583,
+      "grad_norm": 0.49753571327710056,
+      "learning_rate": 2.3293359770076505e-05,
+      "loss": 0.4989,
+      "step": 621
+    },
+    {
+      "epoch": 2.5182186234817814,
+      "grad_norm": 0.4591830679832783,
+      "learning_rate": 2.323756437450136e-05,
+      "loss": 0.4908,
+      "step": 622
+    },
+    {
+      "epoch": 2.522267206477733,
+      "grad_norm": 0.4760879066628539,
+      "learning_rate": 2.3181743091447754e-05,
+      "loss": 0.4847,
+      "step": 623
+    },
+    {
+      "epoch": 2.526315789473684,
+      "grad_norm": 0.3659608880645495,
+      "learning_rate": 2.3125896367261164e-05,
+      "loss": 0.4693,
+      "step": 624
+    },
+    {
+      "epoch": 2.5303643724696356,
+      "grad_norm": 0.4127098861305154,
+      "learning_rate": 2.307002464849052e-05,
+      "loss": 0.4847,
+      "step": 625
+    },
+    {
+      "epoch": 2.534412955465587,
+      "grad_norm": 0.3581984668322913,
+      "learning_rate": 2.30141283818846e-05,
+      "loss": 0.4806,
+      "step": 626
+    },
+    {
+      "epoch": 2.5384615384615383,
+      "grad_norm": 0.36762072003771795,
+      "learning_rate": 2.2958208014388462e-05,
+      "loss": 0.4919,
+      "step": 627
+    },
+    {
+      "epoch": 2.54251012145749,
+      "grad_norm": 0.471042748721673,
+      "learning_rate": 2.2902263993139857e-05,
+      "loss": 0.48,
+      "step": 628
+    },
+    {
+      "epoch": 2.5465587044534415,
+      "grad_norm": 0.3575818962352354,
+      "learning_rate": 2.2846296765465708e-05,
+      "loss": 0.4865,
+      "step": 629
+    },
+    {
+      "epoch": 2.5506072874493926,
+      "grad_norm": 0.5314641741036046,
+      "learning_rate": 2.2790306778878465e-05,
+      "loss": 0.4694,
+      "step": 630
+    },
+    {
+      "epoch": 2.554655870445344,
+      "grad_norm": 0.5171887239113744,
+      "learning_rate": 2.273429448107257e-05,
+      "loss": 0.4803,
+      "step": 631
+    },
+    {
+      "epoch": 2.5587044534412957,
+      "grad_norm": 0.3845312202089083,
+      "learning_rate": 2.2678260319920868e-05,
+      "loss": 0.4835,
+      "step": 632
+    },
+    {
+      "epoch": 2.562753036437247,
+      "grad_norm": 0.5609107808783654,
+      "learning_rate": 2.262220474347102e-05,
+      "loss": 0.4814,
+      "step": 633
+    },
+    {
+      "epoch": 2.5668016194331984,
+      "grad_norm": 0.39867336547318166,
+      "learning_rate": 2.2566128199941918e-05,
+      "loss": 0.4699,
+      "step": 634
+    },
+    {
+      "epoch": 2.57085020242915,
+      "grad_norm": 0.4336031275649315,
+      "learning_rate": 2.2510031137720104e-05,
+      "loss": 0.48,
+      "step": 635
+    },
+    {
+      "epoch": 2.574898785425101,
+      "grad_norm": 0.47194918246822004,
+      "learning_rate": 2.2453914005356206e-05,
+      "loss": 0.4872,
+      "step": 636
+    },
+    {
+      "epoch": 2.5789473684210527,
+      "grad_norm": 0.3623070850325174,
+      "learning_rate": 2.2397777251561302e-05,
+      "loss": 0.4858,
+      "step": 637
+    },
+    {
+      "epoch": 2.582995951417004,
+      "grad_norm": 0.49151051543346935,
+      "learning_rate": 2.2341621325203393e-05,
+      "loss": 0.474,
+      "step": 638
+    },
+    {
+      "epoch": 2.5870445344129553,
+      "grad_norm": 0.4760609908131527,
+      "learning_rate": 2.2285446675303764e-05,
+      "loss": 0.48,
+      "step": 639
+    },
+    {
+      "epoch": 2.591093117408907,
+      "grad_norm": 0.4248871327170189,
+      "learning_rate": 2.2229253751033423e-05,
+      "loss": 0.4849,
+      "step": 640
+    },
+    {
+      "epoch": 2.5951417004048585,
+      "grad_norm": 0.4531844251377083,
+      "learning_rate": 2.2173043001709492e-05,
+      "loss": 0.4817,
+      "step": 641
+    },
+    {
+      "epoch": 2.5991902834008096,
+      "grad_norm": 0.35257002764399015,
+      "learning_rate": 2.2116814876791627e-05,
+      "loss": 0.4869,
+      "step": 642
+    },
+    {
+      "epoch": 2.603238866396761,
+      "grad_norm": 0.40823189336338384,
+      "learning_rate": 2.2060569825878413e-05,
+      "loss": 0.4904,
+      "step": 643
+    },
+    {
+      "epoch": 2.6072874493927127,
+      "grad_norm": 0.3434790102811941,
+      "learning_rate": 2.2004308298703777e-05,
+      "loss": 0.4883,
+      "step": 644
+    },
+    {
+      "epoch": 2.611336032388664,
+      "grad_norm": 0.453687689789995,
+      "learning_rate": 2.1948030745133396e-05,
+      "loss": 0.4832,
+      "step": 645
+    },
+    {
+      "epoch": 2.6153846153846154,
+      "grad_norm": 0.4030869829892541,
+      "learning_rate": 2.1891737615161083e-05,
+      "loss": 0.4889,
+      "step": 646
+    },
+    {
+      "epoch": 2.619433198380567,
+      "grad_norm": 0.41961314087732643,
+      "learning_rate": 2.1835429358905218e-05,
+      "loss": 0.4839,
+      "step": 647
+    },
+    {
+      "epoch": 2.623481781376518,
+      "grad_norm": 0.39285356771617574,
+      "learning_rate": 2.17791064266051e-05,
+      "loss": 0.4803,
+      "step": 648
+    },
+    {
+      "epoch": 2.6275303643724697,
+      "grad_norm": 0.4196947312407116,
+      "learning_rate": 2.172276926861741e-05,
+      "loss": 0.4847,
+      "step": 649
+    },
+    {
+      "epoch": 2.6315789473684212,
+      "grad_norm": 0.4684037652118737,
+      "learning_rate": 2.166641833541255e-05,
+      "loss": 0.4821,
+      "step": 650
+    },
+    {
+      "epoch": 2.6356275303643724,
+      "grad_norm": 0.3078314090729351,
+      "learning_rate": 2.161005407757109e-05,
+      "loss": 0.4787,
+      "step": 651
+    },
+    {
+      "epoch": 2.639676113360324,
+      "grad_norm": 0.4424285683790446,
+      "learning_rate": 2.155367694578013e-05,
+      "loss": 0.4874,
+      "step": 652
+    },
+    {
+      "epoch": 2.6437246963562755,
+      "grad_norm": 0.34674935138309054,
+      "learning_rate": 2.1497287390829712e-05,
+      "loss": 0.485,
+      "step": 653
+    },
+    {
+      "epoch": 2.6477732793522266,
+      "grad_norm": 0.3994975488681126,
+      "learning_rate": 2.1440885863609208e-05,
+      "loss": 0.4793,
+      "step": 654
+    },
+    {
+      "epoch": 2.651821862348178,
+      "grad_norm": 0.3298399250025475,
+      "learning_rate": 2.1384472815103737e-05,
+      "loss": 0.4814,
+      "step": 655
+    },
+    {
+      "epoch": 2.6558704453441297,
+      "grad_norm": 0.36496246159051726,
+      "learning_rate": 2.1328048696390534e-05,
+      "loss": 0.4723,
+      "step": 656
+    },
+    {
+      "epoch": 2.659919028340081,
+      "grad_norm": 0.3726709239521405,
+      "learning_rate": 2.1271613958635342e-05,
+      "loss": 0.4858,
+      "step": 657
+    },
+    {
+      "epoch": 2.6639676113360324,
+      "grad_norm": 0.37646392950521956,
+      "learning_rate": 2.1215169053088825e-05,
+      "loss": 0.4921,
+      "step": 658
+    },
+    {
+      "epoch": 2.668016194331984,
+      "grad_norm": 0.4055037055661752,
+      "learning_rate": 2.1158714431082937e-05,
+      "loss": 0.4873,
+      "step": 659
+    },
+    {
+      "epoch": 2.672064777327935,
+      "grad_norm": 0.3477322510601515,
+      "learning_rate": 2.110225054402733e-05,
+      "loss": 0.4786,
+      "step": 660
+    },
+    {
+      "epoch": 2.6761133603238867,
+      "grad_norm": 0.43048104433771556,
+      "learning_rate": 2.1045777843405747e-05,
+      "loss": 0.4849,
+      "step": 661
+    },
+    {
+      "epoch": 2.6801619433198383,
+      "grad_norm": 0.39106338549742264,
+      "learning_rate": 2.09892967807724e-05,
+      "loss": 0.471,
+      "step": 662
+    },
+    {
+      "epoch": 2.6842105263157894,
+      "grad_norm": 0.421695118584839,
+      "learning_rate": 2.093280780774835e-05,
+      "loss": 0.4931,
+      "step": 663
+    },
+    {
+      "epoch": 2.688259109311741,
+      "grad_norm": 0.38262053898176535,
+      "learning_rate": 2.087631137601793e-05,
+      "loss": 0.4838,
+      "step": 664
+    },
+    {
+      "epoch": 2.6923076923076925,
+      "grad_norm": 0.4115642110094501,
+      "learning_rate": 2.0819807937325104e-05,
+      "loss": 0.4841,
+      "step": 665
+    },
+    {
+      "epoch": 2.6963562753036436,
+      "grad_norm": 0.3599240590578294,
+      "learning_rate": 2.0763297943469847e-05,
+      "loss": 0.4921,
+      "step": 666
+    },
+    {
+      "epoch": 2.700404858299595,
+      "grad_norm": 0.41007531524747276,
+      "learning_rate": 2.070678184630458e-05,
+      "loss": 0.486,
+      "step": 667
+    },
+    {
+      "epoch": 2.7044534412955468,
+      "grad_norm": 0.3562672454457947,
+      "learning_rate": 2.06502600977305e-05,
+      "loss": 0.4865,
+      "step": 668
+    },
+    {
+      "epoch": 2.708502024291498,
+      "grad_norm": 0.34727901160253527,
+      "learning_rate": 2.0593733149694008e-05,
+      "loss": 0.4728,
+      "step": 669
+    },
+    {
+      "epoch": 2.7125506072874495,
+      "grad_norm": 0.38332364966348653,
+      "learning_rate": 2.0537201454183074e-05,
+      "loss": 0.4873,
+      "step": 670
+    },
+    {
+      "epoch": 2.716599190283401,
+      "grad_norm": 0.350591504764532,
+      "learning_rate": 2.0480665463223626e-05,
+      "loss": 0.4899,
+      "step": 671
+    },
+    {
+      "epoch": 2.720647773279352,
+      "grad_norm": 0.45336153032511584,
+      "learning_rate": 2.042412562887594e-05,
+      "loss": 0.4759,
+      "step": 672
+    },
+    {
+      "epoch": 2.7246963562753037,
+      "grad_norm": 0.3772293788081394,
+      "learning_rate": 2.0367582403231033e-05,
+      "loss": 0.4807,
+      "step": 673
+    },
+    {
+      "epoch": 2.7287449392712553,
+      "grad_norm": 0.496537757218724,
+      "learning_rate": 2.0311036238407023e-05,
+      "loss": 0.4809,
+      "step": 674
+    },
+    {
+      "epoch": 2.7327935222672064,
+      "grad_norm": 0.3770728215684446,
+      "learning_rate": 2.0254487586545534e-05,
+      "loss": 0.4904,
+      "step": 675
+    },
+    {
+      "epoch": 2.736842105263158,
+      "grad_norm": 0.5985787770882349,
+      "learning_rate": 2.0197936899808088e-05,
+      "loss": 0.4827,
+      "step": 676
+    },
+    {
+      "epoch": 2.7408906882591095,
+      "grad_norm": 0.3416899624976584,
+      "learning_rate": 2.0141384630372466e-05,
+      "loss": 0.4897,
+      "step": 677
+    },
+    {
+      "epoch": 2.7449392712550607,
+      "grad_norm": 0.48114593130516814,
+      "learning_rate": 2.0084831230429106e-05,
+      "loss": 0.4888,
+      "step": 678
+    },
+    {
+      "epoch": 2.748987854251012,
+      "grad_norm": 0.36303554633428536,
+      "learning_rate": 2.0028277152177492e-05,
+      "loss": 0.4756,
+      "step": 679
+    },
+    {
+      "epoch": 2.753036437246964,
+      "grad_norm": 0.5087956230846838,
+      "learning_rate": 1.9971722847822518e-05,
+      "loss": 0.4833,
+      "step": 680
+    },
+    {
+      "epoch": 2.757085020242915,
+      "grad_norm": 0.33642456356569267,
+      "learning_rate": 1.99151687695709e-05,
+      "loss": 0.4803,
+      "step": 681
+    },
+    {
+      "epoch": 2.7611336032388665,
+      "grad_norm": 0.48088228316395554,
+      "learning_rate": 1.9858615369627537e-05,
+      "loss": 0.4728,
+      "step": 682
+    },
+    {
+      "epoch": 2.765182186234818,
+      "grad_norm": 0.35234656927496394,
+      "learning_rate": 1.9802063100191916e-05,
+      "loss": 0.4902,
+      "step": 683
+    },
+    {
+      "epoch": 2.769230769230769,
+      "grad_norm": 0.41966963017798997,
+      "learning_rate": 1.974551241345447e-05,
+      "loss": 0.4855,
+      "step": 684
+    },
+    {
+      "epoch": 2.7732793522267207,
+      "grad_norm": 0.3575608202943406,
+      "learning_rate": 1.9688963761592984e-05,
+      "loss": 0.478,
+      "step": 685
+    },
+    {
+      "epoch": 2.7773279352226723,
+      "grad_norm": 0.3990686341680936,
+      "learning_rate": 1.963241759676897e-05,
+      "loss": 0.4823,
+      "step": 686
+    },
+    {
+      "epoch": 2.7813765182186234,
+      "grad_norm": 0.3646643175156392,
+      "learning_rate": 1.9575874371124066e-05,
+      "loss": 0.4795,
+      "step": 687
+    },
+    {
+      "epoch": 2.785425101214575,
+      "grad_norm": 0.4168205032024432,
+      "learning_rate": 1.951933453677638e-05,
+      "loss": 0.4867,
+      "step": 688
+    },
+    {
+      "epoch": 2.7894736842105265,
+      "grad_norm": 0.3634995477487968,
+      "learning_rate": 1.9462798545816932e-05,
+      "loss": 0.4922,
+      "step": 689
+    },
+    {
+      "epoch": 2.7935222672064777,
+      "grad_norm": 0.3836021155554615,
+      "learning_rate": 1.9406266850305996e-05,
+      "loss": 0.4742,
+      "step": 690
+    },
+    {
+      "epoch": 2.7975708502024292,
+      "grad_norm": 0.3233725298178868,
+      "learning_rate": 1.9349739902269503e-05,
+      "loss": 0.4868,
+      "step": 691
+    },
+    {
+      "epoch": 2.801619433198381,
+      "grad_norm": 0.3781848475626321,
+      "learning_rate": 1.9293218153695425e-05,
+      "loss": 0.4809,
+      "step": 692
+    },
+    {
+      "epoch": 2.805668016194332,
+      "grad_norm": 0.35958374622053574,
+      "learning_rate": 1.9236702056530153e-05,
+      "loss": 0.4888,
+      "step": 693
+    },
+    {
+      "epoch": 2.8097165991902835,
+      "grad_norm": 0.3331794320133313,
+      "learning_rate": 1.9180192062674907e-05,
+      "loss": 0.4811,
+      "step": 694
+    },
+    {
+      "epoch": 2.813765182186235,
+      "grad_norm": 0.3498458294824225,
+      "learning_rate": 1.9123688623982076e-05,
+      "loss": 0.4887,
+      "step": 695
+    },
+    {
+      "epoch": 2.817813765182186,
+      "grad_norm": 0.366972603682273,
+      "learning_rate": 1.906719219225165e-05,
+      "loss": 0.4922,
+      "step": 696
+    },
+    {
+      "epoch": 2.8218623481781377,
+      "grad_norm": 0.27731452395574085,
+      "learning_rate": 1.9010703219227603e-05,
+      "loss": 0.4798,
+      "step": 697
+    },
+    {
+      "epoch": 2.8259109311740893,
+      "grad_norm": 0.3787148855047442,
+      "learning_rate": 1.8954222156594256e-05,
+      "loss": 0.4866,
+      "step": 698
+    },
+    {
+      "epoch": 2.8299595141700404,
+      "grad_norm": 0.28990062906703656,
+      "learning_rate": 1.8897749455972673e-05,
+      "loss": 0.4814,
+      "step": 699
+    },
+    {
+      "epoch": 2.834008097165992,
+      "grad_norm": 0.35805580554112115,
+      "learning_rate": 1.884128556891707e-05,
+      "loss": 0.4799,
+      "step": 700
+    },
+    {
+      "epoch": 2.8380566801619436,
+      "grad_norm": 0.3188394753158388,
+      "learning_rate": 1.878483094691119e-05,
+      "loss": 0.4955,
+      "step": 701
+    },
+    {
+      "epoch": 2.8421052631578947,
+      "grad_norm": 0.374953341478833,
+      "learning_rate": 1.8728386041364664e-05,
+      "loss": 0.4882,
+      "step": 702
+    },
+    {
+      "epoch": 2.8461538461538463,
+      "grad_norm": 0.3066447946476035,
+      "learning_rate": 1.867195130360947e-05,
+      "loss": 0.4767,
+      "step": 703
+    },
+    {
+      "epoch": 2.850202429149798,
+      "grad_norm": 0.359158482038112,
+      "learning_rate": 1.8615527184896263e-05,
+      "loss": 0.4839,
+      "step": 704
+    },
+    {
+      "epoch": 2.854251012145749,
+      "grad_norm": 0.4013226231502269,
+      "learning_rate": 1.8559114136390795e-05,
+      "loss": 0.485,
+      "step": 705
+    },
+    {
+      "epoch": 2.8582995951417005,
+      "grad_norm": 0.36991485450418715,
+      "learning_rate": 1.8502712609170298e-05,
+      "loss": 0.4845,
+      "step": 706
+    },
+    {
+      "epoch": 2.862348178137652,
+      "grad_norm": 0.3937040831142193,
+      "learning_rate": 1.8446323054219876e-05,
+      "loss": 0.4811,
+      "step": 707
+    },
+    {
+      "epoch": 2.866396761133603,
+      "grad_norm": 0.3194141362952962,
+      "learning_rate": 1.838994592242891e-05,
+      "loss": 0.4853,
+      "step": 708
+    },
+    {
+      "epoch": 2.8704453441295548,
+      "grad_norm": 0.38391999422995055,
+      "learning_rate": 1.8333581664587453e-05,
+      "loss": 0.489,
+      "step": 709
+    },
+    {
+      "epoch": 2.8744939271255063,
+      "grad_norm": 0.3690244184168881,
+      "learning_rate": 1.8277230731382593e-05,
+      "loss": 0.4936,
+      "step": 710
+    },
+    {
+      "epoch": 2.8785425101214575,
+      "grad_norm": 0.4136598709096259,
+      "learning_rate": 1.82208935733949e-05,
+      "loss": 0.4758,
+      "step": 711
+    },
+    {
+      "epoch": 2.882591093117409,
+      "grad_norm": 0.3609683346524138,
+      "learning_rate": 1.8164570641094793e-05,
+      "loss": 0.4822,
+      "step": 712
+    },
+    {
+      "epoch": 2.8866396761133606,
+      "grad_norm": 0.4125258983906132,
+      "learning_rate": 1.810826238483892e-05,
+      "loss": 0.4875,
+      "step": 713
+    },
+    {
+      "epoch": 2.8906882591093117,
+      "grad_norm": 0.3858028172917169,
+      "learning_rate": 1.8051969254866608e-05,
+      "loss": 0.4897,
+      "step": 714
+    },
+    {
+      "epoch": 2.8947368421052633,
+      "grad_norm": 0.3891901075293251,
+      "learning_rate": 1.7995691701296226e-05,
+      "loss": 0.4883,
+      "step": 715
+    },
+    {
+      "epoch": 2.898785425101215,
+      "grad_norm": 0.36704942902374055,
+      "learning_rate": 1.7939430174121593e-05,
+      "loss": 0.4937,
+      "step": 716
+    },
+    {
+      "epoch": 2.902834008097166,
+      "grad_norm": 0.3714817725141067,
+      "learning_rate": 1.788318512320838e-05,
+      "loss": 0.4768,
+      "step": 717
+    },
+    {
+      "epoch": 2.9068825910931175,
+      "grad_norm": 0.3313648994642028,
+      "learning_rate": 1.782695699829051e-05,
+      "loss": 0.4766,
+      "step": 718
+    },
+    {
+      "epoch": 2.910931174089069,
+      "grad_norm": 0.4157370651905882,
+      "learning_rate": 1.7770746248966587e-05,
+      "loss": 0.4813,
+      "step": 719
+    },
+    {
+      "epoch": 2.91497975708502,
+      "grad_norm": 0.39997075133243726,
+      "learning_rate": 1.7714553324696243e-05,
+      "loss": 0.4768,
+      "step": 720
+    },
+    {
+      "epoch": 2.919028340080972,
+      "grad_norm": 0.38776719993814557,
+      "learning_rate": 1.7658378674796614e-05,
+      "loss": 0.4782,
+      "step": 721
+    },
+    {
+      "epoch": 2.9230769230769234,
+      "grad_norm": 0.3369377060128966,
+      "learning_rate": 1.76022227484387e-05,
+      "loss": 0.4928,
+      "step": 722
+    },
+    {
+      "epoch": 2.9271255060728745,
+      "grad_norm": 0.338475423856054,
+      "learning_rate": 1.7546085994643807e-05,
+      "loss": 0.4769,
+      "step": 723
+    },
+    {
+      "epoch": 2.931174089068826,
+      "grad_norm": 0.3808888812058278,
+      "learning_rate": 1.7489968862279902e-05,
+      "loss": 0.4783,
+      "step": 724
+    },
+    {
+      "epoch": 2.9352226720647776,
+      "grad_norm": 0.32861673214209747,
+      "learning_rate": 1.743387180005809e-05,
+      "loss": 0.4769,
+      "step": 725
+    },
+    {
+      "epoch": 2.9392712550607287,
+      "grad_norm": 0.31842810505445324,
+      "learning_rate": 1.737779525652899e-05,
+      "loss": 0.4794,
+      "step": 726
+    },
+    {
+      "epoch": 2.9433198380566803,
+      "grad_norm": 0.3388858375007648,
+      "learning_rate": 1.732173968007914e-05,
+      "loss": 0.4797,
+      "step": 727
+    },
+    {
+      "epoch": 2.9473684210526314,
+      "grad_norm": 0.32104455872353854,
+      "learning_rate": 1.7265705518927437e-05,
+      "loss": 0.487,
+      "step": 728
+    },
+    {
+      "epoch": 2.951417004048583,
+      "grad_norm": 0.34940809394653766,
+      "learning_rate": 1.7209693221121542e-05,
+      "loss": 0.4767,
+      "step": 729
+    },
+    {
+      "epoch": 2.9554655870445345,
+      "grad_norm": 0.2907694348059836,
+      "learning_rate": 1.7153703234534302e-05,
+      "loss": 0.4864,
+      "step": 730
+    },
+    {
+      "epoch": 2.9595141700404857,
+      "grad_norm": 0.3242834484952137,
+      "learning_rate": 1.709773600686015e-05,
+      "loss": 0.4879,
+      "step": 731
+    },
+    {
+      "epoch": 2.9635627530364372,
+      "grad_norm": 0.32612253631306753,
+      "learning_rate": 1.7041791985611548e-05,
+      "loss": 0.4927,
+      "step": 732
+    },
+    {
+      "epoch": 2.967611336032389,
+      "grad_norm": 0.31022767015962616,
+      "learning_rate": 1.69858716181154e-05,
+      "loss": 0.4758,
+      "step": 733
+    },
+    {
+      "epoch": 2.97165991902834,
+      "grad_norm": 0.33840085316658636,
+      "learning_rate": 1.692997535150948e-05,
+      "loss": 0.4795,
+      "step": 734
+    },
+    {
+      "epoch": 2.9757085020242915,
+      "grad_norm": 0.269626731333484,
+      "learning_rate": 1.687410363273884e-05,
+      "loss": 0.4844,
+      "step": 735
+    },
+    {
+      "epoch": 2.979757085020243,
+      "grad_norm": 0.40198561226882984,
+      "learning_rate": 1.6818256908552257e-05,
+      "loss": 0.4931,
+      "step": 736
+    },
+    {
+      "epoch": 2.983805668016194,
+      "grad_norm": 0.3050057806349413,
+      "learning_rate": 1.6762435625498646e-05,
+      "loss": 0.4886,
+      "step": 737
+    },
+    {
+      "epoch": 2.9878542510121457,
+      "grad_norm": 0.39905770787501543,
+      "learning_rate": 1.67066402299235e-05,
+      "loss": 0.4843,
+      "step": 738
+    },
+    {
+      "epoch": 2.9919028340080973,
+      "grad_norm": 0.3173080029913261,
+      "learning_rate": 1.6650871167965313e-05,
+      "loss": 0.4742,
+      "step": 739
+    },
+    {
+      "epoch": 2.9959514170040484,
+      "grad_norm": 0.34236720446917546,
+      "learning_rate": 1.6595128885552028e-05,
+      "loss": 0.4858,
+      "step": 740
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 0.39157919657479173,
+      "learning_rate": 1.653941382839745e-05,
+      "loss": 0.4803,
+      "step": 741
+    },
+    {
+      "epoch": 3.0040485829959516,
+      "grad_norm": 0.4248392779088937,
+      "learning_rate": 1.6483726441997673e-05,
+      "loss": 0.4204,
+      "step": 742
+    },
+    {
+      "epoch": 3.0080971659919027,
+      "grad_norm": 0.34798200428399517,
+      "learning_rate": 1.642806717162757e-05,
+      "loss": 0.4254,
+      "step": 743
+    },
+    {
+      "epoch": 3.0121457489878543,
+      "grad_norm": 0.46464042448914733,
+      "learning_rate": 1.637243646233718e-05,
+      "loss": 0.4242,
+      "step": 744
+    },
+    {
+      "epoch": 3.016194331983806,
+      "grad_norm": 0.37847338734339553,
+      "learning_rate": 1.6316834758948174e-05,
+      "loss": 0.4183,
+      "step": 745
+    },
+    {
+      "epoch": 3.020242914979757,
+      "grad_norm": 0.37784501237362145,
+      "learning_rate": 1.6261262506050282e-05,
+      "loss": 0.4226,
+      "step": 746
+    },
+    {
+      "epoch": 3.0242914979757085,
+      "grad_norm": 0.4237146535195215,
+      "learning_rate": 1.620572014799777e-05,
+      "loss": 0.4238,
+      "step": 747
+    },
+    {
+      "epoch": 3.02834008097166,
+      "grad_norm": 0.4269569136151165,
+      "learning_rate": 1.6150208128905857e-05,
+      "loss": 0.426,
+      "step": 748
+    },
+    {
+      "epoch": 3.032388663967611,
+      "grad_norm": 0.3533448699959528,
+      "learning_rate": 1.6094726892647147e-05,
+      "loss": 0.4208,
+      "step": 749
+    },
+    {
+      "epoch": 3.0364372469635628,
+      "grad_norm": 0.41757970020242524,
+      "learning_rate": 1.6039276882848135e-05,
+      "loss": 0.4249,
+      "step": 750
+    },
+    {
+      "epoch": 3.0404858299595143,
+      "grad_norm": 0.36045913723650297,
+      "learning_rate": 1.5983858542885635e-05,
+      "loss": 0.4143,
+      "step": 751
+    },
+    {
+      "epoch": 3.0445344129554655,
+      "grad_norm": 0.38523465296054765,
+      "learning_rate": 1.5928472315883203e-05,
+      "loss": 0.4204,
+      "step": 752
+    },
+    {
+      "epoch": 3.048582995951417,
+      "grad_norm": 0.40118708562314176,
+      "learning_rate": 1.5873118644707633e-05,
+      "loss": 0.427,
+      "step": 753
+    },
+    {
+      "epoch": 3.0526315789473686,
+      "grad_norm": 0.3737385267775277,
+      "learning_rate": 1.5817797971965413e-05,
+      "loss": 0.4213,
+      "step": 754
+    },
+    {
+      "epoch": 3.0566801619433197,
+      "grad_norm": 0.3768369202521634,
+      "learning_rate": 1.576251073999917e-05,
+      "loss": 0.417,
+      "step": 755
+    },
+    {
+      "epoch": 3.0607287449392713,
+      "grad_norm": 0.34797211787568827,
+      "learning_rate": 1.5707257390884126e-05,
+      "loss": 0.4161,
+      "step": 756
+    },
+    {
+      "epoch": 3.064777327935223,
+      "grad_norm": 0.32612402362296455,
+      "learning_rate": 1.5652038366424595e-05,
+      "loss": 0.4114,
+      "step": 757
+    },
+    {
+      "epoch": 3.068825910931174,
+      "grad_norm": 0.3738799483440077,
+      "learning_rate": 1.5596854108150424e-05,
+      "loss": 0.4318,
+      "step": 758
+    },
+    {
+      "epoch": 3.0728744939271255,
+      "grad_norm": 0.3294867371146253,
+      "learning_rate": 1.5541705057313476e-05,
+      "loss": 0.4135,
+      "step": 759
+    },
+    {
+      "epoch": 3.076923076923077,
+      "grad_norm": 0.3126125987773846,
+      "learning_rate": 1.5486591654884086e-05,
+      "loss": 0.4239,
+      "step": 760
+    },
+    {
+      "epoch": 3.080971659919028,
+      "grad_norm": 0.3683586435531214,
+      "learning_rate": 1.543151434154755e-05,
+      "loss": 0.4258,
+      "step": 761
+    },
+    {
+      "epoch": 3.08502024291498,
+      "grad_norm": 0.30886367538789194,
+      "learning_rate": 1.5376473557700615e-05,
+      "loss": 0.4314,
+      "step": 762
+    },
+    {
+      "epoch": 3.0890688259109313,
+      "grad_norm": 0.3272302463764228,
+      "learning_rate": 1.5321469743447893e-05,
+      "loss": 0.4127,
+      "step": 763
+    },
+    {
+      "epoch": 3.0931174089068825,
+      "grad_norm": 0.33312870948406575,
+      "learning_rate": 1.5266503338598434e-05,
+      "loss": 0.4334,
+      "step": 764
+    },
+    {
+      "epoch": 3.097165991902834,
+      "grad_norm": 0.34887904356144855,
+      "learning_rate": 1.5211574782662149e-05,
+      "loss": 0.4282,
+      "step": 765
+    },
+    {
+      "epoch": 3.1012145748987856,
+      "grad_norm": 0.29723923909380207,
+      "learning_rate": 1.5156684514846307e-05,
+      "loss": 0.4254,
+      "step": 766
+    },
+    {
+      "epoch": 3.1052631578947367,
+      "grad_norm": 0.3399606774581855,
+      "learning_rate": 1.5101832974052012e-05,
+      "loss": 0.4131,
+      "step": 767
+    },
+    {
+      "epoch": 3.1093117408906883,
+      "grad_norm": 0.3531323518656231,
+      "learning_rate": 1.5047020598870737e-05,
+      "loss": 0.4328,
+      "step": 768
+    },
+    {
+      "epoch": 3.11336032388664,
+      "grad_norm": 0.32965820024411674,
+      "learning_rate": 1.4992247827580778e-05,
+      "loss": 0.424,
+      "step": 769
+    },
+    {
+      "epoch": 3.117408906882591,
+      "grad_norm": 0.313089271333722,
+      "learning_rate": 1.493751509814374e-05,
+      "loss": 0.4336,
+      "step": 770
+    },
+    {
+      "epoch": 3.1214574898785425,
+      "grad_norm": 0.31228386099963007,
+      "learning_rate": 1.4882822848201073e-05,
+      "loss": 0.4204,
+      "step": 771
+    },
+    {
+      "epoch": 3.125506072874494,
+      "grad_norm": 0.3321023184652476,
+      "learning_rate": 1.4828171515070553e-05,
+      "loss": 0.424,
+      "step": 772
+    },
+    {
+      "epoch": 3.1295546558704452,
+      "grad_norm": 0.3006792563259293,
+      "learning_rate": 1.4773561535742793e-05,
+      "loss": 0.4187,
+      "step": 773
+    },
+    {
+      "epoch": 3.133603238866397,
+      "grad_norm": 0.3354317955864067,
+      "learning_rate": 1.4718993346877715e-05,
+      "loss": 0.4158,
+      "step": 774
+    },
+    {
+      "epoch": 3.1376518218623484,
+      "grad_norm": 0.31595039750116655,
+      "learning_rate": 1.466446738480111e-05,
+      "loss": 0.4286,
+      "step": 775
+    },
+    {
+      "epoch": 3.1417004048582995,
+      "grad_norm": 0.37927149989461073,
+      "learning_rate": 1.460998408550112e-05,
+      "loss": 0.4239,
+      "step": 776
+    },
+    {
+      "epoch": 3.145748987854251,
+      "grad_norm": 0.3344001000938557,
+      "learning_rate": 1.4555543884624751e-05,
+      "loss": 0.4281,
+      "step": 777
+    },
+    {
+      "epoch": 3.1497975708502026,
+      "grad_norm": 0.337299469356144,
+      "learning_rate": 1.4501147217474402e-05,
+      "loss": 0.4221,
+      "step": 778
+    },
+    {
+      "epoch": 3.1538461538461537,
+      "grad_norm": 0.4039599255992912,
+      "learning_rate": 1.444679451900437e-05,
+      "loss": 0.4362,
+      "step": 779
+    },
+    {
+      "epoch": 3.1578947368421053,
+      "grad_norm": 0.3378786244634307,
+      "learning_rate": 1.4392486223817397e-05,
+      "loss": 0.4192,
+      "step": 780
+    },
+    {
+      "epoch": 3.161943319838057,
+      "grad_norm": 0.36751412405954326,
+      "learning_rate": 1.4338222766161143e-05,
+      "loss": 0.4168,
+      "step": 781
+    },
+    {
+      "epoch": 3.165991902834008,
+      "grad_norm": 0.3562935703314679,
+      "learning_rate": 1.4284004579924777e-05,
+      "loss": 0.4253,
+      "step": 782
+    },
+    {
+      "epoch": 3.1700404858299596,
+      "grad_norm": 0.3267830620463696,
+      "learning_rate": 1.4229832098635479e-05,
+      "loss": 0.4182,
+      "step": 783
+    },
+    {
+      "epoch": 3.174089068825911,
+      "grad_norm": 0.32711818965454675,
+      "learning_rate": 1.4175705755454963e-05,
+      "loss": 0.4252,
+      "step": 784
+    },
+    {
+      "epoch": 3.1781376518218623,
+      "grad_norm": 0.28445673273194816,
+      "learning_rate": 1.412162598317602e-05,
+      "loss": 0.4141,
+      "step": 785
+    },
+    {
+      "epoch": 3.182186234817814,
+      "grad_norm": 0.32787867068101867,
+      "learning_rate": 1.406759321421907e-05,
+      "loss": 0.4229,
+      "step": 786
+    },
+    {
+      "epoch": 3.1862348178137654,
+      "grad_norm": 0.30296223584136145,
+      "learning_rate": 1.401360788062871e-05,
+      "loss": 0.427,
+      "step": 787
+    },
+    {
+      "epoch": 3.1902834008097165,
+      "grad_norm": 0.29433855153814503,
+      "learning_rate": 1.3959670414070204e-05,
+      "loss": 0.4168,
+      "step": 788
+    },
+    {
+      "epoch": 3.194331983805668,
+      "grad_norm": 0.3074779002342039,
+      "learning_rate": 1.3905781245826108e-05,
+      "loss": 0.4285,
+      "step": 789
+    },
+    {
+      "epoch": 3.1983805668016196,
+      "grad_norm": 0.2881540222370516,
+      "learning_rate": 1.3851940806792778e-05,
+      "loss": 0.422,
+      "step": 790
+    },
+    {
+      "epoch": 3.2024291497975708,
+      "grad_norm": 0.3203902977170346,
+      "learning_rate": 1.379814952747693e-05,
+      "loss": 0.4234,
+      "step": 791
+    },
+    {
+      "epoch": 3.2064777327935223,
+      "grad_norm": 0.28569855860725085,
+      "learning_rate": 1.3744407837992193e-05,
+      "loss": 0.4136,
+      "step": 792
+    },
+    {
+      "epoch": 3.2105263157894735,
+      "grad_norm": 0.2824521359025065,
+      "learning_rate": 1.3690716168055692e-05,
+      "loss": 0.4077,
+      "step": 793
+    },
+    {
+      "epoch": 3.214574898785425,
+      "grad_norm": 0.31340340484151125,
+      "learning_rate": 1.363707494698459e-05,
+      "loss": 0.4242,
+      "step": 794
+    },
+    {
+      "epoch": 3.2186234817813766,
+      "grad_norm": 0.2983447607723525,
+      "learning_rate": 1.358348460369265e-05,
+      "loss": 0.4291,
+      "step": 795
+    },
+    {
+      "epoch": 3.2226720647773277,
+      "grad_norm": 0.3223706862166583,
+      "learning_rate": 1.352994556668684e-05,
+      "loss": 0.4259,
+      "step": 796
+    },
+    {
+      "epoch": 3.2267206477732793,
+      "grad_norm": 0.30561322893096676,
+      "learning_rate": 1.3476458264063865e-05,
+      "loss": 0.4193,
+      "step": 797
+    },
+    {
+      "epoch": 3.230769230769231,
+      "grad_norm": 0.34582946805346193,
+      "learning_rate": 1.342302312350678e-05,
+      "loss": 0.4214,
+      "step": 798
+    },
+    {
+      "epoch": 3.234817813765182,
+      "grad_norm": 0.2823508982613799,
+      "learning_rate": 1.3369640572281537e-05,
+      "loss": 0.4195,
+      "step": 799
+    },
+    {
+      "epoch": 3.2388663967611335,
+      "grad_norm": 0.3456003813396814,
+      "learning_rate": 1.3316311037233596e-05,
+      "loss": 0.4218,
+      "step": 800
+    },
+    {
+      "epoch": 3.242914979757085,
+      "grad_norm": 0.29492883024843847,
+      "learning_rate": 1.326303494478451e-05,
+      "loss": 0.4236,
+      "step": 801
+    },
+    {
+      "epoch": 3.246963562753036,
+      "grad_norm": 0.2823790513250997,
+      "learning_rate": 1.3209812720928472e-05,
+      "loss": 0.4262,
+      "step": 802
+    },
+    {
+      "epoch": 3.251012145748988,
+      "grad_norm": 0.2811229795269438,
+      "learning_rate": 1.315664479122898e-05,
+      "loss": 0.4148,
+      "step": 803
+    },
+    {
+      "epoch": 3.2550607287449393,
+      "grad_norm": 0.28414872485595827,
+      "learning_rate": 1.3103531580815378e-05,
+      "loss": 0.4194,
+      "step": 804
+    },
+    {
+      "epoch": 3.2591093117408905,
+      "grad_norm": 0.2787599210785627,
+      "learning_rate": 1.3050473514379488e-05,
+      "loss": 0.4249,
+      "step": 805
+    },
+    {
+      "epoch": 3.263157894736842,
+      "grad_norm": 0.30915174587869504,
+      "learning_rate": 1.2997471016172188e-05,
+      "loss": 0.4292,
+      "step": 806
+    },
+    {
+      "epoch": 3.2672064777327936,
+      "grad_norm": 0.2783136787183032,
+      "learning_rate": 1.2944524510000042e-05,
+      "loss": 0.4154,
+      "step": 807
+    },
+    {
+      "epoch": 3.2712550607287447,
+      "grad_norm": 0.26335676679549647,
+      "learning_rate": 1.2891634419221908e-05,
+      "loss": 0.4341,
+      "step": 808
+    },
+    {
+      "epoch": 3.2753036437246963,
+      "grad_norm": 0.28491783903067713,
+      "learning_rate": 1.2838801166745545e-05,
+      "loss": 0.4356,
+      "step": 809
+    },
+    {
+      "epoch": 3.279352226720648,
+      "grad_norm": 0.2901978626497605,
+      "learning_rate": 1.2786025175024216e-05,
+      "loss": 0.4187,
+      "step": 810
+    },
+    {
+      "epoch": 3.283400809716599,
+      "grad_norm": 0.25993380640732716,
+      "learning_rate": 1.2733306866053357e-05,
+      "loss": 0.4251,
+      "step": 811
+    },
+    {
+      "epoch": 3.2874493927125505,
+      "grad_norm": 0.3067340848812672,
+      "learning_rate": 1.2680646661367163e-05,
+      "loss": 0.4141,
+      "step": 812
+    },
+    {
+      "epoch": 3.291497975708502,
+      "grad_norm": 0.2598027928418247,
+      "learning_rate": 1.2628044982035219e-05,
+      "loss": 0.4196,
+      "step": 813
+    },
+    {
+      "epoch": 3.2955465587044532,
+      "grad_norm": 0.2952055852719839,
+      "learning_rate": 1.2575502248659155e-05,
+      "loss": 0.427,
+      "step": 814
+    },
+    {
+      "epoch": 3.299595141700405,
+      "grad_norm": 0.27197254387268816,
+      "learning_rate": 1.2523018881369269e-05,
+      "loss": 0.4175,
+      "step": 815
+    },
+    {
+      "epoch": 3.3036437246963564,
+      "grad_norm": 0.30740165941975933,
+      "learning_rate": 1.2470595299821176e-05,
+      "loss": 0.4283,
+      "step": 816
+    },
+    {
+      "epoch": 3.3076923076923075,
+      "grad_norm": 0.3177843484579447,
+      "learning_rate": 1.2418231923192416e-05,
+      "loss": 0.4251,
+      "step": 817
+    },
+    {
+      "epoch": 3.311740890688259,
+      "grad_norm": 0.29471560394646307,
+      "learning_rate": 1.2365929170179165e-05,
+      "loss": 0.4166,
+      "step": 818
+    },
+    {
+      "epoch": 3.3157894736842106,
+      "grad_norm": 0.30711425821310095,
+      "learning_rate": 1.2313687458992848e-05,
+      "loss": 0.4213,
+      "step": 819
+    },
+    {
+      "epoch": 3.3198380566801617,
+      "grad_norm": 0.2532134325001613,
+      "learning_rate": 1.2261507207356785e-05,
+      "loss": 0.418,
+      "step": 820
+    },
+    {
+      "epoch": 3.3238866396761133,
+      "grad_norm": 0.3115704146707845,
+      "learning_rate": 1.220938883250288e-05,
+      "loss": 0.4292,
+      "step": 821
+    },
+    {
+      "epoch": 3.327935222672065,
+      "grad_norm": 0.2703437972885868,
+      "learning_rate": 1.215733275116827e-05,
+      "loss": 0.4215,
+      "step": 822
+    },
+    {
+      "epoch": 3.331983805668016,
+      "grad_norm": 0.28973160457897035,
+      "learning_rate": 1.2105339379592005e-05,
+      "loss": 0.4164,
+      "step": 823
+    },
+    {
+      "epoch": 3.3360323886639676,
+      "grad_norm": 0.2822129990262938,
+      "learning_rate": 1.2053409133511681e-05,
+      "loss": 0.4265,
+      "step": 824
+    },
+    {
+      "epoch": 3.340080971659919,
+      "grad_norm": 0.28797454611199,
+      "learning_rate": 1.2001542428160173e-05,
+      "loss": 0.4233,
+      "step": 825
+    },
+    {
+      "epoch": 3.3441295546558703,
+      "grad_norm": 0.28627656775099936,
+      "learning_rate": 1.1949739678262278e-05,
+      "loss": 0.43,
+      "step": 826
+    },
+    {
+      "epoch": 3.348178137651822,
+      "grad_norm": 0.3391352451775233,
+      "learning_rate": 1.189800129803139e-05,
+      "loss": 0.4286,
+      "step": 827
+    },
+    {
+      "epoch": 3.3522267206477734,
+      "grad_norm": 0.2822842088790476,
+      "learning_rate": 1.1846327701166228e-05,
+      "loss": 0.4283,
+      "step": 828
+    },
+    {
+      "epoch": 3.3562753036437245,
+      "grad_norm": 0.31748828914893995,
+      "learning_rate": 1.1794719300847493e-05,
+      "loss": 0.4306,
+      "step": 829
+    },
+    {
+      "epoch": 3.360323886639676,
+      "grad_norm": 0.308915275464002,
+      "learning_rate": 1.1743176509734582e-05,
+      "loss": 0.4234,
+      "step": 830
+    },
+    {
+      "epoch": 3.3643724696356276,
+      "grad_norm": 0.302127104963124,
+      "learning_rate": 1.1691699739962275e-05,
+      "loss": 0.4303,
+      "step": 831
+    },
+    {
+      "epoch": 3.3684210526315788,
+      "grad_norm": 0.3076591121104755,
+      "learning_rate": 1.1640289403137456e-05,
+      "loss": 0.4306,
+      "step": 832
+    },
+    {
+      "epoch": 3.3724696356275303,
+      "grad_norm": 0.2939853718996732,
+      "learning_rate": 1.1588945910335802e-05,
+      "loss": 0.4218,
+      "step": 833
+    },
+    {
+      "epoch": 3.376518218623482,
+      "grad_norm": 0.3045445006878746,
+      "learning_rate": 1.1537669672098517e-05,
+      "loss": 0.4282,
+      "step": 834
+    },
+    {
+      "epoch": 3.380566801619433,
+      "grad_norm": 0.27793018978940376,
+      "learning_rate": 1.1486461098429036e-05,
+      "loss": 0.4168,
+      "step": 835
+    },
+    {
+      "epoch": 3.3846153846153846,
+      "grad_norm": 0.30031594955303437,
+      "learning_rate": 1.1435320598789747e-05,
+      "loss": 0.4268,
+      "step": 836
+    },
+    {
+      "epoch": 3.388663967611336,
+      "grad_norm": 0.3083746623436973,
+      "learning_rate": 1.1384248582098736e-05,
+      "loss": 0.4247,
+      "step": 837
+    },
+    {
+      "epoch": 3.3927125506072873,
+      "grad_norm": 0.31063449627996276,
+      "learning_rate": 1.1333245456726467e-05,
+      "loss": 0.4309,
+      "step": 838
+    },
+    {
+      "epoch": 3.396761133603239,
+      "grad_norm": 0.2775116483918811,
+      "learning_rate": 1.128231163049258e-05,
+      "loss": 0.4183,
+      "step": 839
+    },
+    {
+      "epoch": 3.4008097165991904,
+      "grad_norm": 0.2969841010860041,
+      "learning_rate": 1.123144751066259e-05,
+      "loss": 0.4221,
+      "step": 840
+    },
+    {
+      "epoch": 3.4048582995951415,
+      "grad_norm": 0.2577640435517256,
+      "learning_rate": 1.1180653503944657e-05,
+      "loss": 0.4218,
+      "step": 841
+    },
+    {
+      "epoch": 3.408906882591093,
+      "grad_norm": 0.3010686974086211,
+      "learning_rate": 1.1129930016486287e-05,
+      "loss": 0.4212,
+      "step": 842
+    },
+    {
+      "epoch": 3.4129554655870447,
+      "grad_norm": 0.28247573597435705,
+      "learning_rate": 1.107927745387114e-05,
+      "loss": 0.4283,
+      "step": 843
+    },
+    {
+      "epoch": 3.417004048582996,
+      "grad_norm": 0.27860019408006076,
+      "learning_rate": 1.1028696221115766e-05,
+      "loss": 0.4198,
+      "step": 844
+    },
+    {
+      "epoch": 3.4210526315789473,
+      "grad_norm": 0.29428727982211605,
+      "learning_rate": 1.0978186722666341e-05,
+      "loss": 0.4226,
+      "step": 845
+    },
+    {
+      "epoch": 3.425101214574899,
+      "grad_norm": 0.2829336213316925,
+      "learning_rate": 1.0927749362395475e-05,
+      "loss": 0.4225,
+      "step": 846
+    },
+    {
+      "epoch": 3.42914979757085,
+      "grad_norm": 0.2959117643787738,
+      "learning_rate": 1.0877384543598954e-05,
+      "loss": 0.4141,
+      "step": 847
+    },
+    {
+      "epoch": 3.4331983805668016,
+      "grad_norm": 0.32202952664200174,
+      "learning_rate": 1.0827092668992529e-05,
+      "loss": 0.4202,
+      "step": 848
+    },
+    {
+      "epoch": 3.437246963562753,
+      "grad_norm": 0.2850066893965643,
+      "learning_rate": 1.077687414070869e-05,
+      "loss": 0.4261,
+      "step": 849
+    },
+    {
+      "epoch": 3.4412955465587043,
+      "grad_norm": 0.29645786047718525,
+      "learning_rate": 1.0726729360293451e-05,
+      "loss": 0.4353,
+      "step": 850
+    },
+    {
+      "epoch": 3.445344129554656,
+      "grad_norm": 0.28185269725821266,
+      "learning_rate": 1.0676658728703147e-05,
+      "loss": 0.4181,
+      "step": 851
+    },
+    {
+      "epoch": 3.4493927125506074,
+      "grad_norm": 0.2548813517384349,
+      "learning_rate": 1.0626662646301198e-05,
+      "loss": 0.4287,
+      "step": 852
+    },
+    {
+      "epoch": 3.4534412955465585,
+      "grad_norm": 0.2954684551929161,
+      "learning_rate": 1.0576741512854958e-05,
+      "loss": 0.4197,
+      "step": 853
+    },
+    {
+      "epoch": 3.45748987854251,
+      "grad_norm": 0.2742897786945715,
+      "learning_rate": 1.052689572753247e-05,
+      "loss": 0.4268,
+      "step": 854
+    },
+    {
+      "epoch": 3.4615384615384617,
+      "grad_norm": 0.2648323396021302,
+      "learning_rate": 1.0477125688899328e-05,
+      "loss": 0.4256,
+      "step": 855
+    },
+    {
+      "epoch": 3.465587044534413,
+      "grad_norm": 0.26013443180675544,
+      "learning_rate": 1.0427431794915406e-05,
+      "loss": 0.4292,
+      "step": 856
+    },
+    {
+      "epoch": 3.4696356275303644,
+      "grad_norm": 0.2838106522726138,
+      "learning_rate": 1.0377814442931769e-05,
+      "loss": 0.4362,
+      "step": 857
+    },
+    {
+      "epoch": 3.473684210526316,
+      "grad_norm": 0.2686284762479322,
+      "learning_rate": 1.0328274029687439e-05,
+      "loss": 0.4248,
+      "step": 858
+    },
+    {
+      "epoch": 3.477732793522267,
+      "grad_norm": 0.2684865145399458,
+      "learning_rate": 1.0278810951306253e-05,
+      "loss": 0.4318,
+      "step": 859
+    },
+    {
+      "epoch": 3.4817813765182186,
+      "grad_norm": 0.29134688512762436,
+      "learning_rate": 1.0229425603293647e-05,
+      "loss": 0.4208,
+      "step": 860
+    },
+    {
+      "epoch": 3.48582995951417,
+      "grad_norm": 0.24532909498470964,
+      "learning_rate": 1.0180118380533558e-05,
+      "loss": 0.4253,
+      "step": 861
+    },
+    {
+      "epoch": 3.4898785425101213,
+      "grad_norm": 0.3556061898290794,
+      "learning_rate": 1.0130889677285225e-05,
+      "loss": 0.425,
+      "step": 862
+    },
+    {
+      "epoch": 3.493927125506073,
+      "grad_norm": 0.2679758133492998,
+      "learning_rate": 1.008173988718005e-05,
+      "loss": 0.4208,
+      "step": 863
+    },
+    {
+      "epoch": 3.4979757085020244,
+      "grad_norm": 0.2784492741699898,
+      "learning_rate": 1.0032669403218439e-05,
+      "loss": 0.4242,
+      "step": 864
+    },
+    {
+      "epoch": 3.5020242914979756,
+      "grad_norm": 0.29171985320318616,
+      "learning_rate": 9.983678617766677e-06,
+      "loss": 0.4171,
+      "step": 865
+    },
+    {
+      "epoch": 3.506072874493927,
+      "grad_norm": 0.2669650534581502,
+      "learning_rate": 9.934767922553788e-06,
+      "loss": 0.4248,
+      "step": 866
+    },
+    {
+      "epoch": 3.5101214574898787,
+      "grad_norm": 0.2652761099811369,
+      "learning_rate": 9.885937708668371e-06,
+      "loss": 0.4215,
+      "step": 867
+    },
+    {
+      "epoch": 3.51417004048583,
+      "grad_norm": 0.25276515273345895,
+      "learning_rate": 9.837188366555524e-06,
+      "loss": 0.4325,
+      "step": 868
+    },
+    {
+      "epoch": 3.5182186234817814,
+      "grad_norm": 0.26376605647594853,
+      "learning_rate": 9.788520286013691e-06,
+      "loss": 0.4224,
+      "step": 869
+    },
+    {
+      "epoch": 3.522267206477733,
+      "grad_norm": 0.2534118866392286,
+      "learning_rate": 9.739933856191537e-06,
+      "loss": 0.4286,
+      "step": 870
+    },
+    {
+      "epoch": 3.526315789473684,
+      "grad_norm": 0.2680777756022362,
+      "learning_rate": 9.691429465584866e-06,
+      "loss": 0.4175,
+      "step": 871
+    },
+    {
+      "epoch": 3.5303643724696356,
+      "grad_norm": 0.2331972725428116,
+      "learning_rate": 9.643007502033492e-06,
+      "loss": 0.429,
+      "step": 872
+    },
+    {
+      "epoch": 3.534412955465587,
+      "grad_norm": 0.2451571199770327,
+      "learning_rate": 9.594668352718152e-06,
+      "loss": 0.4216,
+      "step": 873
+    },
+    {
+      "epoch": 3.5384615384615383,
+      "grad_norm": 0.2582537621013891,
+      "learning_rate": 9.546412404157383e-06,
+      "loss": 0.4241,
+      "step": 874
+    },
+    {
+      "epoch": 3.54251012145749,
+      "grad_norm": 0.23663634105904996,
+      "learning_rate": 9.498240042204471e-06,
+      "loss": 0.4282,
+      "step": 875
+    },
+    {
+      "epoch": 3.5465587044534415,
+      "grad_norm": 0.24554488426028484,
+      "learning_rate": 9.450151652044339e-06,
+      "loss": 0.4287,
+      "step": 876
+    },
+    {
+      "epoch": 3.5506072874493926,
+      "grad_norm": 0.24497426816585285,
+      "learning_rate": 9.402147618190472e-06,
+      "loss": 0.4187,
+      "step": 877
+    },
+    {
+      "epoch": 3.554655870445344,
+      "grad_norm": 0.25994813852158266,
+      "learning_rate": 9.354228324481841e-06,
+      "loss": 0.4295,
+      "step": 878
+    },
+    {
+      "epoch": 3.5587044534412957,
+      "grad_norm": 0.2512337439881897,
+      "learning_rate": 9.306394154079852e-06,
+      "loss": 0.4325,
+      "step": 879
+    },
+    {
+      "epoch": 3.562753036437247,
+      "grad_norm": 0.24645742242889174,
+      "learning_rate": 9.258645489465254e-06,
+      "loss": 0.4168,
+      "step": 880
+    },
+    {
+      "epoch": 3.5668016194331984,
+      "grad_norm": 0.2650282574460377,
+      "learning_rate": 9.21098271243509e-06,
+      "loss": 0.4223,
+      "step": 881
+    },
+    {
+      "epoch": 3.57085020242915,
+      "grad_norm": 0.24849997260354237,
+      "learning_rate": 9.163406204099659e-06,
+      "loss": 0.4277,
+      "step": 882
+    },
+    {
+      "epoch": 3.574898785425101,
+      "grad_norm": 0.257034320034506,
+      "learning_rate": 9.115916344879452e-06,
+      "loss": 0.4309,
+      "step": 883
+    },
+    {
+      "epoch": 3.5789473684210527,
+      "grad_norm": 0.2592159857945639,
+      "learning_rate": 9.068513514502126e-06,
+      "loss": 0.4209,
+      "step": 884
+    },
+    {
+      "epoch": 3.582995951417004,
+      "grad_norm": 0.25898837393661983,
+      "learning_rate": 9.021198091999439e-06,
+      "loss": 0.4348,
+      "step": 885
+    },
+    {
+      "epoch": 3.5870445344129553,
+      "grad_norm": 0.28909351108882425,
+      "learning_rate": 8.973970455704251e-06,
+      "loss": 0.4252,
+      "step": 886
+    },
+    {
+      "epoch": 3.591093117408907,
+      "grad_norm": 0.24815348704757875,
+      "learning_rate": 8.926830983247495e-06,
+      "loss": 0.4224,
+      "step": 887
+    },
+    {
+      "epoch": 3.5951417004048585,
+      "grad_norm": 0.2652239628012717,
+      "learning_rate": 8.879780051555116e-06,
+      "loss": 0.4249,
+      "step": 888
+    },
+    {
+      "epoch": 3.5991902834008096,
+      "grad_norm": 0.257694835332905,
+      "learning_rate": 8.83281803684512e-06,
+      "loss": 0.4295,
+      "step": 889
+    },
+    {
+      "epoch": 3.603238866396761,
+      "grad_norm": 0.2788523171733285,
+      "learning_rate": 8.785945314624532e-06,
+      "loss": 0.4239,
+      "step": 890
+    },
+    {
+      "epoch": 3.6072874493927127,
+      "grad_norm": 0.2640082556636577,
+      "learning_rate": 8.739162259686385e-06,
+      "loss": 0.4297,
+      "step": 891
+    },
+    {
+      "epoch": 3.611336032388664,
+      "grad_norm": 0.2700593585594836,
+      "learning_rate": 8.692469246106734e-06,
+      "loss": 0.4222,
+      "step": 892
+    },
+    {
+      "epoch": 3.6153846153846154,
+      "grad_norm": 0.2732978745796162,
+      "learning_rate": 8.645866647241677e-06,
+      "loss": 0.4214,
+      "step": 893
+    },
+    {
+      "epoch": 3.619433198380567,
+      "grad_norm": 0.24987738586376784,
+      "learning_rate": 8.599354835724348e-06,
+      "loss": 0.4279,
+      "step": 894
+    },
+    {
+      "epoch": 3.623481781376518,
+      "grad_norm": 0.3222074718220882,
+      "learning_rate": 8.552934183461958e-06,
+      "loss": 0.4266,
+      "step": 895
+    },
+    {
+      "epoch": 3.6275303643724697,
+      "grad_norm": 0.24958561094377224,
+      "learning_rate": 8.506605061632805e-06,
+      "loss": 0.425,
+      "step": 896
+    },
+    {
+      "epoch": 3.6315789473684212,
+      "grad_norm": 0.26569073511906527,
+      "learning_rate": 8.460367840683308e-06,
+      "loss": 0.4297,
+      "step": 897
+    },
+    {
+      "epoch": 3.6356275303643724,
+      "grad_norm": 0.2677483522515821,
+      "learning_rate": 8.414222890325065e-06,
+      "loss": 0.424,
+      "step": 898
+    },
+    {
+      "epoch": 3.639676113360324,
+      "grad_norm": 0.2625010104345672,
+      "learning_rate": 8.368170579531855e-06,
+      "loss": 0.428,
+      "step": 899
+    },
+    {
+      "epoch": 3.6437246963562755,
+      "grad_norm": 0.2837607388161951,
+      "learning_rate": 8.322211276536733e-06,
+      "loss": 0.4215,
+      "step": 900
+    },
+    {
+      "epoch": 3.6477732793522266,
+      "grad_norm": 0.24171046285870748,
+      "learning_rate": 8.27634534882907e-06,
+      "loss": 0.4253,
+      "step": 901
+    },
+    {
+      "epoch": 3.651821862348178,
+      "grad_norm": 0.2763925816492832,
+      "learning_rate": 8.230573163151585e-06,
+      "loss": 0.41,
+      "step": 902
+    },
+    {
+      "epoch": 3.6558704453441297,
+      "grad_norm": 0.2683309209816837,
+      "learning_rate": 8.184895085497466e-06,
+      "loss": 0.4106,
+      "step": 903
+    },
+    {
+      "epoch": 3.659919028340081,
+      "grad_norm": 0.26790045216096026,
+      "learning_rate": 8.139311481107401e-06,
+      "loss": 0.4329,
+      "step": 904
+    },
+    {
+      "epoch": 3.6639676113360324,
+      "grad_norm": 0.2555669029332108,
+      "learning_rate": 8.093822714466686e-06,
+      "loss": 0.4202,
+      "step": 905
+    },
+    {
+      "epoch": 3.668016194331984,
+      "grad_norm": 0.2640436815405802,
+      "learning_rate": 8.048429149302271e-06,
+      "loss": 0.4274,
+      "step": 906
+    },
+    {
+      "epoch": 3.672064777327935,
+      "grad_norm": 0.24946912947898137,
+      "learning_rate": 8.003131148579902e-06,
+      "loss": 0.4204,
+      "step": 907
+    },
+    {
+      "epoch": 3.6761133603238867,
+      "grad_norm": 0.2563348969921724,
+      "learning_rate": 7.957929074501182e-06,
+      "loss": 0.4314,
+      "step": 908
+    },
+    {
+      "epoch": 3.6801619433198383,
+      "grad_norm": 0.25260908840781665,
+      "learning_rate": 7.912823288500695e-06,
+      "loss": 0.4197,
+      "step": 909
+    },
+    {
+      "epoch": 3.6842105263157894,
+      "grad_norm": 0.24241355009642734,
+      "learning_rate": 7.867814151243102e-06,
+      "loss": 0.4192,
+      "step": 910
+    },
+    {
+      "epoch": 3.688259109311741,
+      "grad_norm": 0.28180591499666713,
+      "learning_rate": 7.822902022620267e-06,
+      "loss": 0.4188,
+      "step": 911
+    },
+    {
+      "epoch": 3.6923076923076925,
+      "grad_norm": 0.2451459654038878,
+      "learning_rate": 7.778087261748377e-06,
+      "loss": 0.42,
+      "step": 912
+    },
+    {
+      "epoch": 3.6963562753036436,
+      "grad_norm": 0.24608176943011803,
+      "learning_rate": 7.733370226965056e-06,
+      "loss": 0.4254,
+      "step": 913
+    },
+    {
+      "epoch": 3.700404858299595,
+      "grad_norm": 0.24289636516857402,
+      "learning_rate": 7.688751275826527e-06,
+      "loss": 0.4246,
+      "step": 914
+    },
+    {
+      "epoch": 3.7044534412955468,
+      "grad_norm": 0.2293636872921345,
+      "learning_rate": 7.644230765104735e-06,
+      "loss": 0.4164,
+      "step": 915
+    },
+    {
+      "epoch": 3.708502024291498,
+      "grad_norm": 0.21888501367273863,
+      "learning_rate": 7.599809050784502e-06,
+      "loss": 0.4201,
+      "step": 916
+    },
+    {
+      "epoch": 3.7125506072874495,
+      "grad_norm": 0.27024087993208873,
+      "learning_rate": 7.555486488060661e-06,
+      "loss": 0.4134,
+      "step": 917
+    },
+    {
+      "epoch": 3.716599190283401,
+      "grad_norm": 0.24942867668930918,
+      "learning_rate": 7.511263431335249e-06,
+      "loss": 0.4222,
+      "step": 918
+    },
+    {
+      "epoch": 3.720647773279352,
+      "grad_norm": 0.21640214739360283,
+      "learning_rate": 7.467140234214656e-06,
+      "loss": 0.4292,
+      "step": 919
+    },
+    {
+      "epoch": 3.7246963562753037,
+      "grad_norm": 0.26284285561374154,
+      "learning_rate": 7.423117249506777e-06,
+      "loss": 0.4198,
+      "step": 920
+    },
+    {
+      "epoch": 3.7287449392712553,
+      "grad_norm": 0.24559228089952878,
+      "learning_rate": 7.379194829218231e-06,
+      "loss": 0.4319,
+      "step": 921
+    },
+    {
+      "epoch": 3.7327935222672064,
+      "grad_norm": 0.2447700444217082,
+      "learning_rate": 7.335373324551518e-06,
+      "loss": 0.4271,
+      "step": 922
+    },
+    {
+      "epoch": 3.736842105263158,
+      "grad_norm": 0.264100355532592,
+      "learning_rate": 7.291653085902224e-06,
+      "loss": 0.4331,
+      "step": 923
+    },
+    {
+      "epoch": 3.7408906882591095,
+      "grad_norm": 0.25330641212752986,
+      "learning_rate": 7.248034462856213e-06,
+      "loss": 0.4236,
+      "step": 924
+    },
+    {
+      "epoch": 3.7449392712550607,
+      "grad_norm": 0.23857995401744306,
+      "learning_rate": 7.20451780418683e-06,
+      "loss": 0.4096,
+      "step": 925
+    },
+    {
+      "epoch": 3.748987854251012,
+      "grad_norm": 0.2340296485837844,
+      "learning_rate": 7.161103457852128e-06,
+      "loss": 0.4191,
+      "step": 926
+    },
+    {
+      "epoch": 3.753036437246964,
+      "grad_norm": 0.23137873032572603,
+      "learning_rate": 7.117791770992046e-06,
+      "loss": 0.423,
+      "step": 927
+    },
+    {
+      "epoch": 3.757085020242915,
+      "grad_norm": 0.2285352602926768,
+      "learning_rate": 7.074583089925688e-06,
+      "loss": 0.4267,
+      "step": 928
+    },
+    {
+      "epoch": 3.7611336032388665,
+      "grad_norm": 0.2289710181623498,
+      "learning_rate": 7.0314777601485155e-06,
+      "loss": 0.4312,
+      "step": 929
+    },
+    {
+      "epoch": 3.765182186234818,
+      "grad_norm": 0.259280376035225,
+      "learning_rate": 6.988476126329604e-06,
+      "loss": 0.4247,
+      "step": 930
+    },
+    {
+      "epoch": 3.769230769230769,
+      "grad_norm": 0.25363961966900916,
+      "learning_rate": 6.945578532308857e-06,
+      "loss": 0.4205,
+      "step": 931
+    },
+    {
+      "epoch": 3.7732793522267207,
+      "grad_norm": 0.23721465020539576,
+      "learning_rate": 6.902785321094301e-06,
+      "loss": 0.4103,
+      "step": 932
+    },
+    {
+      "epoch": 3.7773279352226723,
+      "grad_norm": 0.2514806833474067,
+      "learning_rate": 6.8600968348593115e-06,
+      "loss": 0.4242,
+      "step": 933
+    },
+    {
+      "epoch": 3.7813765182186234,
+      "grad_norm": 0.23266160671272113,
+      "learning_rate": 6.817513414939892e-06,
+      "loss": 0.427,
+      "step": 934
+    },
+    {
+      "epoch": 3.785425101214575,
+      "grad_norm": 0.24919842923074753,
+      "learning_rate": 6.775035401831924e-06,
+      "loss": 0.4149,
+      "step": 935
+    },
+    {
+      "epoch": 3.7894736842105265,
+      "grad_norm": 0.24386672493512873,
+      "learning_rate": 6.732663135188473e-06,
+      "loss": 0.4218,
+      "step": 936
+    },
+    {
+      "epoch": 3.7935222672064777,
+      "grad_norm": 0.24028598943220253,
+      "learning_rate": 6.690396953817055e-06,
+      "loss": 0.4265,
+      "step": 937
+    },
+    {
+      "epoch": 3.7975708502024292,
+      "grad_norm": 0.2500344395623075,
+      "learning_rate": 6.648237195676927e-06,
+      "loss": 0.4116,
+      "step": 938
+    },
+    {
+      "epoch": 3.801619433198381,
+      "grad_norm": 0.23365246997413586,
+      "learning_rate": 6.606184197876395e-06,
+      "loss": 0.4232,
+      "step": 939
+    },
+    {
+      "epoch": 3.805668016194332,
+      "grad_norm": 0.2588513868404867,
+      "learning_rate": 6.564238296670109e-06,
+      "loss": 0.426,
+      "step": 940
+    },
+    {
+      "epoch": 3.8097165991902835,
+      "grad_norm": 0.2538956724344903,
+      "learning_rate": 6.52239982745638e-06,
+      "loss": 0.4152,
+      "step": 941
+    },
+    {
+      "epoch": 3.813765182186235,
+      "grad_norm": 0.253984662064166,
+      "learning_rate": 6.480669124774484e-06,
+      "loss": 0.4221,
+      "step": 942
+    },
+    {
+      "epoch": 3.817813765182186,
+      "grad_norm": 0.2372315973100364,
+      "learning_rate": 6.439046522302013e-06,
+      "loss": 0.4209,
+      "step": 943
+    },
+    {
+      "epoch": 3.8218623481781377,
+      "grad_norm": 0.23449897136083198,
+      "learning_rate": 6.397532352852191e-06,
+      "loss": 0.4279,
+      "step": 944
+    },
+    {
+      "epoch": 3.8259109311740893,
+      "grad_norm": 0.23415174556918272,
+      "learning_rate": 6.356126948371202e-06,
+      "loss": 0.4197,
+      "step": 945
+    },
+    {
+      "epoch": 3.8299595141700404,
+      "grad_norm": 0.2577289853500942,
+      "learning_rate": 6.314830639935559e-06,
+      "loss": 0.4294,
+      "step": 946
+    },
+    {
+      "epoch": 3.834008097165992,
+      "grad_norm": 0.22461855022071567,
+      "learning_rate": 6.273643757749448e-06,
+      "loss": 0.4267,
+      "step": 947
+    },
+    {
+      "epoch": 3.8380566801619436,
+      "grad_norm": 0.2496418163675073,
+      "learning_rate": 6.2325666311420856e-06,
+      "loss": 0.4205,
+      "step": 948
+    },
+    {
+      "epoch": 3.8421052631578947,
+      "grad_norm": 0.2351749567417615,
+      "learning_rate": 6.1915995885650675e-06,
+      "loss": 0.4212,
+      "step": 949
+    },
+    {
+      "epoch": 3.8461538461538463,
+      "grad_norm": 0.24295726171235468,
+      "learning_rate": 6.150742957589784e-06,
+      "loss": 0.4353,
+      "step": 950
+    },
+    {
+      "epoch": 3.850202429149798,
+      "grad_norm": 0.21916955654524206,
+      "learning_rate": 6.109997064904774e-06,
+      "loss": 0.4238,
+      "step": 951
+    },
+    {
+      "epoch": 3.854251012145749,
+      "grad_norm": 0.2518988140386885,
+      "learning_rate": 6.069362236313092e-06,
+      "loss": 0.4186,
+      "step": 952
+    },
+    {
+      "epoch": 3.8582995951417005,
+      "grad_norm": 0.23050554774410628,
+      "learning_rate": 6.0288387967297565e-06,
+      "loss": 0.4301,
+      "step": 953
+    },
+    {
+      "epoch": 3.862348178137652,
+      "grad_norm": 0.23086233051359734,
+      "learning_rate": 5.988427070179108e-06,
+      "loss": 0.427,
+      "step": 954
+    },
+    {
+      "epoch": 3.866396761133603,
+      "grad_norm": 0.23173648521696227,
+      "learning_rate": 5.948127379792234e-06,
+      "loss": 0.4228,
+      "step": 955
+    },
+    {
+      "epoch": 3.8704453441295548,
+      "grad_norm": 0.2543446217973367,
+      "learning_rate": 5.90794004780439e-06,
+      "loss": 0.4194,
+      "step": 956
+    },
+    {
+      "epoch": 3.8744939271255063,
+      "grad_norm": 0.22989343864315434,
+      "learning_rate": 5.867865395552405e-06,
+      "loss": 0.4246,
+      "step": 957
+    },
+    {
+      "epoch": 3.8785425101214575,
+      "grad_norm": 0.24540328468198278,
+      "learning_rate": 5.8279037434721364e-06,
+      "loss": 0.4231,
+      "step": 958
+    },
+    {
+      "epoch": 3.882591093117409,
+      "grad_norm": 0.24939256576874305,
+      "learning_rate": 5.788055411095892e-06,
+      "loss": 0.4155,
+      "step": 959
+    },
+    {
+      "epoch": 3.8866396761133606,
+      "grad_norm": 0.23778621999825,
+      "learning_rate": 5.7483207170498625e-06,
+      "loss": 0.4265,
+      "step": 960
+    },
+    {
+      "epoch": 3.8906882591093117,
+      "grad_norm": 0.23406855067645838,
+      "learning_rate": 5.708699979051613e-06,
+      "loss": 0.4236,
+      "step": 961
+    },
+    {
+      "epoch": 3.8947368421052633,
+      "grad_norm": 0.2198791241668043,
+      "learning_rate": 5.66919351390751e-06,
+      "loss": 0.4116,
+      "step": 962
+    },
+    {
+      "epoch": 3.898785425101215,
+      "grad_norm": 0.2282392237281033,
+      "learning_rate": 5.629801637510188e-06,
+      "loss": 0.4278,
+      "step": 963
+    },
+    {
+      "epoch": 3.902834008097166,
+      "grad_norm": 0.24584286948225065,
+      "learning_rate": 5.590524664836043e-06,
+      "loss": 0.4215,
+      "step": 964
+    },
+    {
+      "epoch": 3.9068825910931175,
+      "grad_norm": 0.22988385229367353,
+      "learning_rate": 5.551362909942708e-06,
+      "loss": 0.4161,
+      "step": 965
+    },
+    {
+      "epoch": 3.910931174089069,
+      "grad_norm": 0.2352264758710563,
+      "learning_rate": 5.51231668596653e-06,
+      "loss": 0.4281,
+      "step": 966
+    },
+    {
+      "epoch": 3.91497975708502,
+      "grad_norm": 0.22948108550753996,
+      "learning_rate": 5.47338630512007e-06,
+      "loss": 0.417,
+      "step": 967
+    },
+    {
+      "epoch": 3.919028340080972,
+      "grad_norm": 0.23788128170314501,
+      "learning_rate": 5.434572078689617e-06,
+      "loss": 0.4235,
+      "step": 968
+    },
+    {
+      "epoch": 3.9230769230769234,
+      "grad_norm": 0.2391727005237767,
+      "learning_rate": 5.395874317032694e-06,
+      "loss": 0.4257,
+      "step": 969
+    },
+    {
+      "epoch": 3.9271255060728745,
+      "grad_norm": 0.22407548545420855,
+      "learning_rate": 5.357293329575573e-06,
+      "loss": 0.4185,
+      "step": 970
+    },
+    {
+      "epoch": 3.931174089068826,
+      "grad_norm": 0.22226501557657494,
+      "learning_rate": 5.318829424810797e-06,
+      "loss": 0.4227,
+      "step": 971
+    },
+    {
+      "epoch": 3.9352226720647776,
+      "grad_norm": 0.23960767685154402,
+      "learning_rate": 5.280482910294728e-06,
+      "loss": 0.4216,
+      "step": 972
+    },
+    {
+      "epoch": 3.9392712550607287,
+      "grad_norm": 0.22093107030568077,
+      "learning_rate": 5.242254092645078e-06,
+      "loss": 0.4185,
+      "step": 973
+    },
+    {
+      "epoch": 3.9433198380566803,
+      "grad_norm": 0.2268778026354143,
+      "learning_rate": 5.2041432775384445e-06,
+      "loss": 0.4176,
+      "step": 974
+    },
+    {
+      "epoch": 3.9473684210526314,
+      "grad_norm": 0.23253224523086655,
+      "learning_rate": 5.166150769707892e-06,
+      "loss": 0.4298,
+      "step": 975
+    },
+    {
+      "epoch": 3.951417004048583,
+      "grad_norm": 0.22175980248246482,
+      "learning_rate": 5.128276872940503e-06,
+      "loss": 0.4191,
+      "step": 976
+    },
+    {
+      "epoch": 3.9554655870445345,
+      "grad_norm": 0.2389436638990379,
+      "learning_rate": 5.090521890074953e-06,
+      "loss": 0.4275,
+      "step": 977
+    },
+    {
+      "epoch": 3.9595141700404857,
+      "grad_norm": 0.22447163931106429,
+      "learning_rate": 5.0528861229990675e-06,
+      "loss": 0.423,
+      "step": 978
+    },
+    {
+      "epoch": 3.9635627530364372,
+      "grad_norm": 0.22137217752542263,
+      "learning_rate": 5.015369872647442e-06,
+      "loss": 0.4251,
+      "step": 979
+    },
+    {
+      "epoch": 3.967611336032389,
+      "grad_norm": 0.22111365814558073,
+      "learning_rate": 4.9779734389990195e-06,
+      "loss": 0.4287,
+      "step": 980
+    },
+    {
+      "epoch": 3.97165991902834,
+      "grad_norm": 0.23809894740002846,
+      "learning_rate": 4.940697121074675e-06,
+      "loss": 0.4254,
+      "step": 981
+    },
+    {
+      "epoch": 3.9757085020242915,
+      "grad_norm": 0.21916421575087208,
+      "learning_rate": 4.903541216934864e-06,
+      "loss": 0.4212,
+      "step": 982
+    },
+    {
+      "epoch": 3.979757085020243,
+      "grad_norm": 0.2284413241364092,
+      "learning_rate": 4.8665060236772e-06,
+      "loss": 0.424,
+      "step": 983
+    },
+    {
+      "epoch": 3.983805668016194,
+      "grad_norm": 0.21037880361083572,
+      "learning_rate": 4.829591837434102e-06,
+      "loss": 0.428,
+      "step": 984
+    },
+    {
+      "epoch": 3.9878542510121457,
+      "grad_norm": 0.233684060968751,
+      "learning_rate": 4.792798953370424e-06,
+      "loss": 0.4163,
+      "step": 985
+    },
+    {
+      "epoch": 3.9919028340080973,
+      "grad_norm": 0.22799931847804913,
+      "learning_rate": 4.756127665681083e-06,
+      "loss": 0.4245,
+      "step": 986
+    },
+    {
+      "epoch": 3.9959514170040484,
+      "grad_norm": 0.2215972970854936,
+      "learning_rate": 4.719578267588725e-06,
+      "loss": 0.4167,
+      "step": 987
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 0.2281625997980398,
+      "learning_rate": 4.6831510513413525e-06,
+      "loss": 0.4213,
+      "step": 988
+    },
+    {
+      "epoch": 4.004048582995951,
+      "grad_norm": 0.3832365557775448,
+      "learning_rate": 4.6468463082100176e-06,
+      "loss": 0.3823,
+      "step": 989
+    },
+    {
+      "epoch": 4.008097165991903,
+      "grad_norm": 0.29675667316155124,
+      "learning_rate": 4.610664328486485e-06,
+      "loss": 0.384,
+      "step": 990
+    },
+    {
+      "epoch": 4.012145748987854,
+      "grad_norm": 0.23800857289554533,
+      "learning_rate": 4.5746054014809e-06,
+      "loss": 0.3782,
+      "step": 991
+    },
+    {
+      "epoch": 4.016194331983805,
+      "grad_norm": 0.3145035550007243,
+      "learning_rate": 4.5386698155194766e-06,
+      "loss": 0.3868,
+      "step": 992
+    },
+    {
+      "epoch": 4.020242914979757,
+      "grad_norm": 0.36582898207095993,
+      "learning_rate": 4.502857857942204e-06,
+      "loss": 0.3786,
+      "step": 993
+    },
+    {
+      "epoch": 4.0242914979757085,
+      "grad_norm": 0.34899112698184315,
+      "learning_rate": 4.467169815100549e-06,
+      "loss": 0.3751,
+      "step": 994
+    },
+    {
+      "epoch": 4.02834008097166,
+      "grad_norm": 0.3151247173872968,
+      "learning_rate": 4.431605972355137e-06,
+      "loss": 0.3818,
+      "step": 995
+    },
+    {
+      "epoch": 4.032388663967612,
+      "grad_norm": 0.2985804797262812,
+      "learning_rate": 4.396166614073516e-06,
+      "loss": 0.3768,
+      "step": 996
+    },
+    {
+      "epoch": 4.036437246963563,
+      "grad_norm": 0.2964628595092906,
+      "learning_rate": 4.360852023627846e-06,
+      "loss": 0.384,
+      "step": 997
+    },
+    {
+      "epoch": 4.040485829959514,
+      "grad_norm": 0.29752644021464825,
+      "learning_rate": 4.325662483392656e-06,
+      "loss": 0.3819,
+      "step": 998
+    },
+    {
+      "epoch": 4.044534412955466,
+      "grad_norm": 0.3193254553107604,
+      "learning_rate": 4.29059827474257e-06,
+      "loss": 0.3751,
+      "step": 999
+    },
+    {
+      "epoch": 4.048582995951417,
+      "grad_norm": 0.2960667997073776,
+      "learning_rate": 4.255659678050063e-06,
+      "loss": 0.3751,
+      "step": 1000
+    },
+    {
+      "epoch": 4.052631578947368,
+      "grad_norm": 0.2822428238444285,
+      "learning_rate": 4.22084697268323e-06,
+      "loss": 0.3856,
+      "step": 1001
+    },
+    {
+      "epoch": 4.05668016194332,
+      "grad_norm": 0.2983923850516611,
+      "learning_rate": 4.186160437003535e-06,
+      "loss": 0.3808,
+      "step": 1002
+    },
+    {
+      "epoch": 4.060728744939271,
+      "grad_norm": 0.3232237552011415,
+      "learning_rate": 4.151600348363582e-06,
+      "loss": 0.38,
+      "step": 1003
+    },
+    {
+      "epoch": 4.064777327935222,
+      "grad_norm": 0.31943585288940457,
+      "learning_rate": 4.11716698310493e-06,
+      "loss": 0.381,
+      "step": 1004
+    },
+    {
+      "epoch": 4.068825910931174,
+      "grad_norm": 0.2691260290727423,
+      "learning_rate": 4.082860616555848e-06,
+      "loss": 0.3771,
+      "step": 1005
+    },
+    {
+      "epoch": 4.0728744939271255,
+      "grad_norm": 0.303702909162991,
+      "learning_rate": 4.048681523029125e-06,
+      "loss": 0.3787,
+      "step": 1006
+    },
+    {
+      "epoch": 4.076923076923077,
+      "grad_norm": 0.2755609908438066,
+      "learning_rate": 4.014629975819882e-06,
+      "loss": 0.3764,
+      "step": 1007
+    },
+    {
+      "epoch": 4.080971659919029,
+      "grad_norm": 0.29205683037620417,
+      "learning_rate": 3.9807062472033855e-06,
+      "loss": 0.3727,
+      "step": 1008
+    },
+    {
+      "epoch": 4.08502024291498,
+      "grad_norm": 0.2935750636550215,
+      "learning_rate": 3.94691060843287e-06,
+      "loss": 0.3805,
+      "step": 1009
+    },
+    {
+      "epoch": 4.089068825910931,
+      "grad_norm": 0.2792374988311523,
+      "learning_rate": 3.913243329737353e-06,
+      "loss": 0.3852,
+      "step": 1010
+    },
+    {
+      "epoch": 4.093117408906883,
+      "grad_norm": 0.24711787629034043,
+      "learning_rate": 3.879704680319503e-06,
+      "loss": 0.3826,
+      "step": 1011
+    },
+    {
+      "epoch": 4.097165991902834,
+      "grad_norm": 0.2723129492481472,
+      "learning_rate": 3.846294928353471e-06,
+      "loss": 0.3927,
+      "step": 1012
+    },
+    {
+      "epoch": 4.101214574898785,
+      "grad_norm": 0.28732125953787396,
+      "learning_rate": 3.8130143409827345e-06,
+      "loss": 0.3747,
+      "step": 1013
+    },
+    {
+      "epoch": 4.105263157894737,
+      "grad_norm": 0.2644968034101217,
+      "learning_rate": 3.7798631843179887e-06,
+      "loss": 0.3826,
+      "step": 1014
+    },
+    {
+      "epoch": 4.109311740890688,
+      "grad_norm": 0.23356584995295104,
+      "learning_rate": 3.7468417234349973e-06,
+      "loss": 0.3771,
+      "step": 1015
+    },
+    {
+      "epoch": 4.113360323886639,
+      "grad_norm": 0.24038933054398948,
+      "learning_rate": 3.7139502223724823e-06,
+      "loss": 0.3781,
+      "step": 1016
+    },
+    {
+      "epoch": 4.117408906882591,
+      "grad_norm": 0.24967233965872154,
+      "learning_rate": 3.6811889441300095e-06,
+      "loss": 0.3791,
+      "step": 1017
+    },
+    {
+      "epoch": 4.1214574898785425,
+      "grad_norm": 0.25863994611281543,
+      "learning_rate": 3.648558150665886e-06,
+      "loss": 0.3886,
+      "step": 1018
+    },
+    {
+      "epoch": 4.125506072874494,
+      "grad_norm": 0.23289380265193396,
+      "learning_rate": 3.616058102895075e-06,
+      "loss": 0.3789,
+      "step": 1019
+    },
+    {
+      "epoch": 4.129554655870446,
+      "grad_norm": 0.2388264821787797,
+      "learning_rate": 3.583689060687079e-06,
+      "loss": 0.3763,
+      "step": 1020
+    },
+    {
+      "epoch": 4.133603238866397,
+      "grad_norm": 0.23609011262229773,
+      "learning_rate": 3.5514512828639023e-06,
+      "loss": 0.3713,
+      "step": 1021
+    },
+    {
+      "epoch": 4.137651821862348,
+      "grad_norm": 0.22163909999824902,
+      "learning_rate": 3.51934502719796e-06,
+      "loss": 0.3761,
+      "step": 1022
+    },
+    {
+      "epoch": 4.1417004048583,
+      "grad_norm": 0.21321345703690173,
+      "learning_rate": 3.4873705504100185e-06,
+      "loss": 0.3767,
+      "step": 1023
+    },
+    {
+      "epoch": 4.145748987854251,
+      "grad_norm": 0.2367994660625239,
+      "learning_rate": 3.4555281081671345e-06,
+      "loss": 0.3757,
+      "step": 1024
+    },
+    {
+      "epoch": 4.149797570850202,
+      "grad_norm": 0.23820880973124783,
+      "learning_rate": 3.4238179550806326e-06,
+      "loss": 0.3822,
+      "step": 1025
+    },
+    {
+      "epoch": 4.153846153846154,
+      "grad_norm": 0.2312843553593752,
+      "learning_rate": 3.3922403447040497e-06,
+      "loss": 0.3821,
+      "step": 1026
+    },
+    {
+      "epoch": 4.157894736842105,
+      "grad_norm": 0.21243570968812964,
+      "learning_rate": 3.3607955295311247e-06,
+      "loss": 0.3805,
+      "step": 1027
+    },
+    {
+      "epoch": 4.161943319838056,
+      "grad_norm": 0.20673296003933803,
+      "learning_rate": 3.32948376099375e-06,
+      "loss": 0.3795,
+      "step": 1028
+    },
+    {
+      "epoch": 4.165991902834008,
+      "grad_norm": 0.2150397167534648,
+      "learning_rate": 3.2983052894599976e-06,
+      "loss": 0.3806,
+      "step": 1029
+    },
+    {
+      "epoch": 4.17004048582996,
+      "grad_norm": 0.2098152632490226,
+      "learning_rate": 3.2672603642320945e-06,
+      "loss": 0.3681,
+      "step": 1030
+    },
+    {
+      "epoch": 4.174089068825911,
+      "grad_norm": 0.2324992180538782,
+      "learning_rate": 3.236349233544438e-06,
+      "loss": 0.3827,
+      "step": 1031
+    },
+    {
+      "epoch": 4.178137651821863,
+      "grad_norm": 0.21325259030048538,
+      "learning_rate": 3.205572144561604e-06,
+      "loss": 0.3813,
+      "step": 1032
+    },
+    {
+      "epoch": 4.182186234817814,
+      "grad_norm": 0.21711094922573418,
+      "learning_rate": 3.174929343376374e-06,
+      "loss": 0.3813,
+      "step": 1033
+    },
+    {
+      "epoch": 4.186234817813765,
+      "grad_norm": 0.22964097696872218,
+      "learning_rate": 3.1444210750077777e-06,
+      "loss": 0.3898,
+      "step": 1034
+    },
+    {
+      "epoch": 4.190283400809717,
+      "grad_norm": 0.21046807463152742,
+      "learning_rate": 3.1140475833991046e-06,
+      "loss": 0.3715,
+      "step": 1035
+    },
+    {
+      "epoch": 4.194331983805668,
+      "grad_norm": 0.22166447521246746,
+      "learning_rate": 3.0838091114159895e-06,
+      "loss": 0.3801,
+      "step": 1036
+    },
+    {
+      "epoch": 4.198380566801619,
+      "grad_norm": 0.21376444774266223,
+      "learning_rate": 3.0537059008444524e-06,
+      "loss": 0.3801,
+      "step": 1037
+    },
+    {
+      "epoch": 4.202429149797571,
+      "grad_norm": 0.21408701770623725,
+      "learning_rate": 3.0237381923889563e-06,
+      "loss": 0.3821,
+      "step": 1038
+    },
+    {
+      "epoch": 4.206477732793522,
+      "grad_norm": 0.21098500253455396,
+      "learning_rate": 2.9939062256704997e-06,
+      "loss": 0.3785,
+      "step": 1039
+    },
+    {
+      "epoch": 4.2105263157894735,
+      "grad_norm": 0.21432444496631534,
+      "learning_rate": 2.964210239224696e-06,
+      "loss": 0.3698,
+      "step": 1040
+    },
+    {
+      "epoch": 4.2145748987854255,
+      "grad_norm": 0.21430827505451402,
+      "learning_rate": 2.934650470499869e-06,
+      "loss": 0.372,
+      "step": 1041
+    },
+    {
+      "epoch": 4.218623481781377,
+      "grad_norm": 0.21838372997643404,
+      "learning_rate": 2.905227155855128e-06,
+      "loss": 0.3811,
+      "step": 1042
+    },
+    {
+      "epoch": 4.222672064777328,
+      "grad_norm": 0.21072051387932947,
+      "learning_rate": 2.875940530558519e-06,
+      "loss": 0.3836,
+      "step": 1043
+    },
+    {
+      "epoch": 4.22672064777328,
+      "grad_norm": 0.22367327633947226,
+      "learning_rate": 2.8467908287851154e-06,
+      "loss": 0.3759,
+      "step": 1044
+    },
+    {
+      "epoch": 4.230769230769231,
+      "grad_norm": 0.22017166296365265,
+      "learning_rate": 2.8177782836151603e-06,
+      "loss": 0.3775,
+      "step": 1045
+    },
+    {
+      "epoch": 4.234817813765182,
+      "grad_norm": 0.21797667494097514,
+      "learning_rate": 2.7889031270321874e-06,
+      "loss": 0.3822,
+      "step": 1046
+    },
+    {
+      "epoch": 4.238866396761134,
+      "grad_norm": 0.20918177317863323,
+      "learning_rate": 2.760165589921184e-06,
+      "loss": 0.3849,
+      "step": 1047
+    },
+    {
+      "epoch": 4.242914979757085,
+      "grad_norm": 0.2173824247047903,
+      "learning_rate": 2.7315659020667305e-06,
+      "loss": 0.3725,
+      "step": 1048
+    },
+    {
+      "epoch": 4.246963562753036,
+      "grad_norm": 0.20473087567419054,
+      "learning_rate": 2.703104292151164e-06,
+      "loss": 0.3879,
+      "step": 1049
+    },
+    {
+      "epoch": 4.251012145748988,
+      "grad_norm": 0.20600797637409035,
+      "learning_rate": 2.6747809877527607e-06,
+      "loss": 0.3861,
+      "step": 1050
+    },
+    {
+      "epoch": 4.255060728744939,
+      "grad_norm": 0.22694600197923204,
+      "learning_rate": 2.6465962153439082e-06,
+      "loss": 0.3837,
+      "step": 1051
+    },
+    {
+      "epoch": 4.2591093117408905,
+      "grad_norm": 0.2162025099616817,
+      "learning_rate": 2.618550200289303e-06,
+      "loss": 0.3835,
+      "step": 1052
+    },
+    {
+      "epoch": 4.2631578947368425,
+      "grad_norm": 0.22236249217122328,
+      "learning_rate": 2.59064316684412e-06,
+      "loss": 0.3775,
+      "step": 1053
+    },
+    {
+      "epoch": 4.267206477732794,
+      "grad_norm": 0.2127690391267123,
+      "learning_rate": 2.562875338152264e-06,
+      "loss": 0.3858,
+      "step": 1054
+    },
+    {
+      "epoch": 4.271255060728745,
+      "grad_norm": 0.21753384028405137,
+      "learning_rate": 2.5352469362445552e-06,
+      "loss": 0.3772,
+      "step": 1055
+    },
+    {
+      "epoch": 4.275303643724697,
+      "grad_norm": 0.20879510707054993,
+      "learning_rate": 2.5077581820369523e-06,
+      "loss": 0.3806,
+      "step": 1056
+    },
+    {
+      "epoch": 4.279352226720648,
+      "grad_norm": 0.21889246901933707,
+      "learning_rate": 2.480409295328803e-06,
+      "loss": 0.3769,
+      "step": 1057
+    },
+    {
+      "epoch": 4.283400809716599,
+      "grad_norm": 0.20886239424427536,
+      "learning_rate": 2.45320049480108e-06,
+      "loss": 0.3775,
+      "step": 1058
+    },
+    {
+      "epoch": 4.287449392712551,
+      "grad_norm": 0.27408551162598654,
+      "learning_rate": 2.4261319980146293e-06,
+      "loss": 0.3776,
+      "step": 1059
+    },
+    {
+      "epoch": 4.291497975708502,
+      "grad_norm": 0.22624792563907253,
+      "learning_rate": 2.399204021408421e-06,
+      "loss": 0.3805,
+      "step": 1060
+    },
+    {
+      "epoch": 4.295546558704453,
+      "grad_norm": 0.20298382869247633,
+      "learning_rate": 2.3724167802978395e-06,
+      "loss": 0.375,
+      "step": 1061
+    },
+    {
+      "epoch": 4.299595141700405,
+      "grad_norm": 0.21257226708401938,
+      "learning_rate": 2.3457704888729625e-06,
+      "loss": 0.3832,
+      "step": 1062
+    },
+    {
+      "epoch": 4.303643724696356,
+      "grad_norm": 0.22482229056728828,
+      "learning_rate": 2.3192653601968162e-06,
+      "loss": 0.373,
+      "step": 1063
+    },
+    {
+      "epoch": 4.3076923076923075,
+      "grad_norm": 0.20809207313632816,
+      "learning_rate": 2.2929016062037057e-06,
+      "loss": 0.374,
+      "step": 1064
+    },
+    {
+      "epoch": 4.3117408906882595,
+      "grad_norm": 0.20766562869106417,
+      "learning_rate": 2.2666794376975055e-06,
+      "loss": 0.3777,
+      "step": 1065
+    },
+    {
+      "epoch": 4.315789473684211,
+      "grad_norm": 0.2148278970757649,
+      "learning_rate": 2.24059906434998e-06,
+      "loss": 0.3744,
+      "step": 1066
+    },
+    {
+      "epoch": 4.319838056680162,
+      "grad_norm": 0.2115196166537666,
+      "learning_rate": 2.214660694699091e-06,
+      "loss": 0.3818,
+      "step": 1067
+    },
+    {
+      "epoch": 4.323886639676114,
+      "grad_norm": 0.21020112688867396,
+      "learning_rate": 2.188864536147355e-06,
+      "loss": 0.3793,
+      "step": 1068
+    },
+    {
+      "epoch": 4.327935222672065,
+      "grad_norm": 0.21031102687159833,
+      "learning_rate": 2.1632107949601757e-06,
+      "loss": 0.3826,
+      "step": 1069
+    },
+    {
+      "epoch": 4.331983805668016,
+      "grad_norm": 0.22199064543549477,
+      "learning_rate": 2.137699676264171e-06,
+      "loss": 0.3803,
+      "step": 1070
+    },
+    {
+      "epoch": 4.336032388663968,
+      "grad_norm": 0.21676973056683682,
+      "learning_rate": 2.1123313840455738e-06,
+      "loss": 0.3767,
+      "step": 1071
+    },
+    {
+      "epoch": 4.340080971659919,
+      "grad_norm": 0.2183637155021383,
+      "learning_rate": 2.08710612114857e-06,
+      "loss": 0.3883,
+      "step": 1072
+    },
+    {
+      "epoch": 4.34412955465587,
+      "grad_norm": 0.21070433495833676,
+      "learning_rate": 2.0620240892736953e-06,
+      "loss": 0.3757,
+      "step": 1073
+    },
+    {
+      "epoch": 4.348178137651822,
+      "grad_norm": 0.21076856984886175,
+      "learning_rate": 2.0370854889762e-06,
+      "loss": 0.3808,
+      "step": 1074
+    },
+    {
+      "epoch": 4.352226720647773,
+      "grad_norm": 0.2113381498246684,
+      "learning_rate": 2.0122905196644726e-06,
+      "loss": 0.3666,
+      "step": 1075
+    },
+    {
+      "epoch": 4.3562753036437245,
+      "grad_norm": 0.22166466497270432,
+      "learning_rate": 1.987639379598427e-06,
+      "loss": 0.384,
+      "step": 1076
+    },
+    {
+      "epoch": 4.3603238866396765,
+      "grad_norm": 0.22261633009004897,
+      "learning_rate": 1.963132265887919e-06,
+      "loss": 0.3832,
+      "step": 1077
+    },
+    {
+      "epoch": 4.364372469635628,
+      "grad_norm": 0.20438896810649906,
+      "learning_rate": 1.9387693744911784e-06,
+      "loss": 0.3752,
+      "step": 1078
+    },
+    {
+      "epoch": 4.368421052631579,
+      "grad_norm": 0.21236779823337706,
+      "learning_rate": 1.9145509002132368e-06,
+      "loss": 0.3754,
+      "step": 1079
+    },
+    {
+      "epoch": 4.372469635627531,
+      "grad_norm": 0.20361913171604343,
+      "learning_rate": 1.8904770367043723e-06,
+      "loss": 0.3848,
+      "step": 1080
+    },
+    {
+      "epoch": 4.376518218623482,
+      "grad_norm": 0.21760007577401141,
+      "learning_rate": 1.8665479764585504e-06,
+      "loss": 0.3821,
+      "step": 1081
+    },
+    {
+      "epoch": 4.380566801619433,
+      "grad_norm": 0.22047970093043723,
+      "learning_rate": 1.8427639108118977e-06,
+      "loss": 0.3763,
+      "step": 1082
+    },
+    {
+      "epoch": 4.384615384615385,
+      "grad_norm": 0.2043773194017159,
+      "learning_rate": 1.8191250299411755e-06,
+      "loss": 0.38,
+      "step": 1083
+    },
+    {
+      "epoch": 4.388663967611336,
+      "grad_norm": 0.2083296698861204,
+      "learning_rate": 1.7956315228622446e-06,
+      "loss": 0.3969,
+      "step": 1084
+    },
+    {
+      "epoch": 4.392712550607287,
+      "grad_norm": 0.2208297441621872,
+      "learning_rate": 1.7722835774285553e-06,
+      "loss": 0.3908,
+      "step": 1085
+    },
+    {
+      "epoch": 4.396761133603239,
+      "grad_norm": 0.2121289898217183,
+      "learning_rate": 1.7490813803296624e-06,
+      "loss": 0.3766,
+      "step": 1086
+    },
+    {
+      "epoch": 4.40080971659919,
+      "grad_norm": 0.22389411093758368,
+      "learning_rate": 1.726025117089718e-06,
+      "loss": 0.3795,
+      "step": 1087
+    },
+    {
+      "epoch": 4.4048582995951415,
+      "grad_norm": 0.2107609477381942,
+      "learning_rate": 1.70311497206598e-06,
+      "loss": 0.3752,
+      "step": 1088
+    },
+    {
+      "epoch": 4.4089068825910935,
+      "grad_norm": 0.21050680251024023,
+      "learning_rate": 1.6803511284473638e-06,
+      "loss": 0.3772,
+      "step": 1089
+    },
+    {
+      "epoch": 4.412955465587045,
+      "grad_norm": 0.20572929406811205,
+      "learning_rate": 1.6577337682529538e-06,
+      "loss": 0.3897,
+      "step": 1090
+    },
+    {
+      "epoch": 4.417004048582996,
+      "grad_norm": 0.19742082216476006,
+      "learning_rate": 1.6352630723305617e-06,
+      "loss": 0.3808,
+      "step": 1091
+    },
+    {
+      "epoch": 4.421052631578947,
+      "grad_norm": 0.21353629735254426,
+      "learning_rate": 1.612939220355272e-06,
+      "loss": 0.3733,
+      "step": 1092
+    },
+    {
+      "epoch": 4.425101214574899,
+      "grad_norm": 0.20517700495707464,
+      "learning_rate": 1.5907623908280112e-06,
+      "loss": 0.3831,
+      "step": 1093
+    },
+    {
+      "epoch": 4.42914979757085,
+      "grad_norm": 0.2005827107997109,
+      "learning_rate": 1.568732761074121e-06,
+      "loss": 0.3759,
+      "step": 1094
+    },
+    {
+      "epoch": 4.433198380566802,
+      "grad_norm": 0.2261590858119055,
+      "learning_rate": 1.5468505072419237e-06,
+      "loss": 0.3837,
+      "step": 1095
+    },
+    {
+      "epoch": 4.437246963562753,
+      "grad_norm": 0.20950037464270194,
+      "learning_rate": 1.5251158043013448e-06,
+      "loss": 0.3714,
+      "step": 1096
+    },
+    {
+      "epoch": 4.441295546558704,
+      "grad_norm": 0.20611586477714755,
+      "learning_rate": 1.5035288260424885e-06,
+      "loss": 0.3875,
+      "step": 1097
+    },
+    {
+      "epoch": 4.445344129554655,
+      "grad_norm": 0.2017613481954189,
+      "learning_rate": 1.4820897450742623e-06,
+      "loss": 0.3791,
+      "step": 1098
+    },
+    {
+      "epoch": 4.449392712550607,
+      "grad_norm": 0.20243454425205193,
+      "learning_rate": 1.4607987328229767e-06,
+      "loss": 0.377,
+      "step": 1099
+    },
+    {
+      "epoch": 4.4534412955465585,
+      "grad_norm": 0.2063774514174568,
+      "learning_rate": 1.4396559595310056e-06,
+      "loss": 0.3831,
+      "step": 1100
+    },
+    {
+      "epoch": 4.4574898785425106,
+      "grad_norm": 0.2106181737761159,
+      "learning_rate": 1.4186615942553973e-06,
+      "loss": 0.3752,
+      "step": 1101
+    },
+    {
+      "epoch": 4.461538461538462,
+      "grad_norm": 0.22845369097379872,
+      "learning_rate": 1.3978158048665424e-06,
+      "loss": 0.3785,
+      "step": 1102
+    },
+    {
+      "epoch": 4.465587044534413,
+      "grad_norm": 0.1978760615128565,
+      "learning_rate": 1.3771187580468115e-06,
+      "loss": 0.3778,
+      "step": 1103
+    },
+    {
+      "epoch": 4.469635627530364,
+      "grad_norm": 0.20700857749452853,
+      "learning_rate": 1.3565706192892392e-06,
+      "loss": 0.3797,
+      "step": 1104
+    },
+    {
+      "epoch": 4.473684210526316,
+      "grad_norm": 0.21409943959340133,
+      "learning_rate": 1.3361715528961993e-06,
+      "loss": 0.3753,
+      "step": 1105
+    },
+    {
+      "epoch": 4.477732793522267,
+      "grad_norm": 0.20834759725899227,
+      "learning_rate": 1.3159217219780817e-06,
+      "loss": 0.3874,
+      "step": 1106
+    },
+    {
+      "epoch": 4.481781376518219,
+      "grad_norm": 0.20101483863736563,
+      "learning_rate": 1.2958212884519949e-06,
+      "loss": 0.3742,
+      "step": 1107
+    },
+    {
+      "epoch": 4.48582995951417,
+      "grad_norm": 0.20426855998889681,
+      "learning_rate": 1.2758704130404675e-06,
+      "loss": 0.3743,
+      "step": 1108
+    },
+    {
+      "epoch": 4.489878542510121,
+      "grad_norm": 0.20489959143927902,
+      "learning_rate": 1.2560692552701738e-06,
+      "loss": 0.378,
+      "step": 1109
+    },
+    {
+      "epoch": 4.493927125506072,
+      "grad_norm": 0.20790915685462974,
+      "learning_rate": 1.2364179734706382e-06,
+      "loss": 0.3714,
+      "step": 1110
+    },
+    {
+      "epoch": 4.497975708502024,
+      "grad_norm": 0.20547553540810673,
+      "learning_rate": 1.2169167247729852e-06,
+      "loss": 0.3782,
+      "step": 1111
+    },
+    {
+      "epoch": 4.502024291497976,
+      "grad_norm": 0.1999109133121712,
+      "learning_rate": 1.1975656651086864e-06,
+      "loss": 0.3799,
+      "step": 1112
+    },
+    {
+      "epoch": 4.506072874493928,
+      "grad_norm": 0.20060299328023012,
+      "learning_rate": 1.1783649492082926e-06,
+      "loss": 0.3723,
+      "step": 1113
+    },
+    {
+      "epoch": 4.510121457489879,
+      "grad_norm": 0.204091350596608,
+      "learning_rate": 1.1593147306002183e-06,
+      "loss": 0.3741,
+      "step": 1114
+    },
+    {
+      "epoch": 4.51417004048583,
+      "grad_norm": 0.20309747327339112,
+      "learning_rate": 1.140415161609505e-06,
+      "loss": 0.3734,
+      "step": 1115
+    },
+    {
+      "epoch": 4.518218623481781,
+      "grad_norm": 0.20328819857496652,
+      "learning_rate": 1.121666393356604e-06,
+      "loss": 0.3703,
+      "step": 1116
+    },
+    {
+      "epoch": 4.522267206477733,
+      "grad_norm": 0.20365621621703042,
+      "learning_rate": 1.1030685757561632e-06,
+      "loss": 0.3798,
+      "step": 1117
+    },
+    {
+      "epoch": 4.526315789473684,
+      "grad_norm": 0.20446397645036102,
+      "learning_rate": 1.0846218575158373e-06,
+      "loss": 0.3778,
+      "step": 1118
+    },
+    {
+      "epoch": 4.530364372469636,
+      "grad_norm": 0.21148144349727185,
+      "learning_rate": 1.066326386135097e-06,
+      "loss": 0.3788,
+      "step": 1119
+    },
+    {
+      "epoch": 4.534412955465587,
+      "grad_norm": 0.1939861358941975,
+      "learning_rate": 1.0481823079040398e-06,
+      "loss": 0.3753,
+      "step": 1120
+    },
+    {
+      "epoch": 4.538461538461538,
+      "grad_norm": 0.19875145723019536,
+      "learning_rate": 1.0301897679022344e-06,
+      "loss": 0.3808,
+      "step": 1121
+    },
+    {
+      "epoch": 4.5425101214574894,
+      "grad_norm": 0.1994390212003373,
+      "learning_rate": 1.0123489099975491e-06,
+      "loss": 0.381,
+      "step": 1122
+    },
+    {
+      "epoch": 4.5465587044534415,
+      "grad_norm": 0.20501991429392075,
+      "learning_rate": 9.9465987684501e-07,
+      "loss": 0.3656,
+      "step": 1123
+    },
+    {
+      "epoch": 4.550607287449393,
+      "grad_norm": 0.19948314972634065,
+      "learning_rate": 9.771228098856534e-07,
+      "loss": 0.3841,
+      "step": 1124
+    },
+    {
+      "epoch": 4.554655870445345,
+      "grad_norm": 0.2043821813409728,
+      "learning_rate": 9.597378493454013e-07,
+      "loss": 0.3716,
+      "step": 1125
+    },
+    {
+      "epoch": 4.558704453441296,
+      "grad_norm": 0.20331747832386476,
+      "learning_rate": 9.425051342339353e-07,
+      "loss": 0.375,
+      "step": 1126
+    },
+    {
+      "epoch": 4.562753036437247,
+      "grad_norm": 0.19700167472749597,
+      "learning_rate": 9.254248023435864e-07,
+      "loss": 0.381,
+      "step": 1127
+    },
+    {
+      "epoch": 4.566801619433198,
+      "grad_norm": 0.20073304668055664,
+      "learning_rate": 9.084969902482288e-07,
+      "loss": 0.3823,
+      "step": 1128
+    },
+    {
+      "epoch": 4.57085020242915,
+      "grad_norm": 0.20309889318754484,
+      "learning_rate": 8.917218333021993e-07,
+      "loss": 0.3848,
+      "step": 1129
+    },
+    {
+      "epoch": 4.574898785425101,
+      "grad_norm": 0.2026754387468518,
+      "learning_rate": 8.750994656392109e-07,
+      "loss": 0.3688,
+      "step": 1130
+    },
+    {
+      "epoch": 4.578947368421053,
+      "grad_norm": 0.20361576092566866,
+      "learning_rate": 8.58630020171265e-07,
+      "loss": 0.3752,
+      "step": 1131
+    },
+    {
+      "epoch": 4.582995951417004,
+      "grad_norm": 0.1969300192434805,
+      "learning_rate": 8.423136285876099e-07,
+      "loss": 0.3746,
+      "step": 1132
+    },
+    {
+      "epoch": 4.587044534412955,
+      "grad_norm": 0.21609669593925201,
+      "learning_rate": 8.26150421353682e-07,
+      "loss": 0.3811,
+      "step": 1133
+    },
+    {
+      "epoch": 4.5910931174089065,
+      "grad_norm": 0.20035664113721346,
+      "learning_rate": 8.101405277100549e-07,
+      "loss": 0.3785,
+      "step": 1134
+    },
+    {
+      "epoch": 4.5951417004048585,
+      "grad_norm": 0.19502390836765818,
+      "learning_rate": 7.942840756714077e-07,
+      "loss": 0.3806,
+      "step": 1135
+    },
+    {
+      "epoch": 4.59919028340081,
+      "grad_norm": 0.19895868940524059,
+      "learning_rate": 7.785811920255093e-07,
+      "loss": 0.3758,
+      "step": 1136
+    },
+    {
+      "epoch": 4.603238866396762,
+      "grad_norm": 0.20736249101464926,
+      "learning_rate": 7.630320023322024e-07,
+      "loss": 0.3756,
+      "step": 1137
+    },
+    {
+      "epoch": 4.607287449392713,
+      "grad_norm": 0.19560155575046284,
+      "learning_rate": 7.476366309223903e-07,
+      "loss": 0.3738,
+      "step": 1138
+    },
+    {
+      "epoch": 4.611336032388664,
+      "grad_norm": 0.19514288187148532,
+      "learning_rate": 7.323952008970537e-07,
+      "loss": 0.3859,
+      "step": 1139
+    },
+    {
+      "epoch": 4.615384615384615,
+      "grad_norm": 0.19980396866348923,
+      "learning_rate": 7.173078341262618e-07,
+      "loss": 0.3741,
+      "step": 1140
+    },
+    {
+      "epoch": 4.619433198380567,
+      "grad_norm": 0.20422179901721962,
+      "learning_rate": 7.023746512481989e-07,
+      "loss": 0.3826,
+      "step": 1141
+    },
+    {
+      "epoch": 4.623481781376518,
+      "grad_norm": 0.19814204425468057,
+      "learning_rate": 6.875957716681902e-07,
+      "loss": 0.3796,
+      "step": 1142
+    },
+    {
+      "epoch": 4.62753036437247,
+      "grad_norm": 0.19719087580708525,
+      "learning_rate": 6.729713135577687e-07,
+      "loss": 0.3823,
+      "step": 1143
+    },
+    {
+      "epoch": 4.631578947368421,
+      "grad_norm": 0.20232403393635562,
+      "learning_rate": 6.585013938537143e-07,
+      "loss": 0.3771,
+      "step": 1144
+    },
+    {
+      "epoch": 4.635627530364372,
+      "grad_norm": 0.20245241849957263,
+      "learning_rate": 6.441861282571115e-07,
+      "loss": 0.3872,
+      "step": 1145
+    },
+    {
+      "epoch": 4.6396761133603235,
+      "grad_norm": 0.20422439922349347,
+      "learning_rate": 6.300256312324471e-07,
+      "loss": 0.3718,
+      "step": 1146
+    },
+    {
+      "epoch": 4.6437246963562755,
+      "grad_norm": 0.19638583261105352,
+      "learning_rate": 6.160200160066776e-07,
+      "loss": 0.3814,
+      "step": 1147
+    },
+    {
+      "epoch": 4.647773279352227,
+      "grad_norm": 0.20040956309173688,
+      "learning_rate": 6.021693945683327e-07,
+      "loss": 0.3717,
+      "step": 1148
+    },
+    {
+      "epoch": 4.651821862348179,
+      "grad_norm": 0.20207591625095078,
+      "learning_rate": 5.884738776666044e-07,
+      "loss": 0.3841,
+      "step": 1149
+    },
+    {
+      "epoch": 4.65587044534413,
+      "grad_norm": 0.19832487920961847,
+      "learning_rate": 5.749335748104878e-07,
+      "loss": 0.3828,
+      "step": 1150
+    },
+    {
+      "epoch": 4.659919028340081,
+      "grad_norm": 0.20153003843392053,
+      "learning_rate": 5.615485942678778e-07,
+      "loss": 0.3826,
+      "step": 1151
+    },
+    {
+      "epoch": 4.663967611336032,
+      "grad_norm": 0.19908090054457342,
+      "learning_rate": 5.483190430647289e-07,
+      "loss": 0.3881,
+      "step": 1152
+    },
+    {
+      "epoch": 4.668016194331984,
+      "grad_norm": 0.19536843871049345,
+      "learning_rate": 5.352450269841747e-07,
+      "loss": 0.3767,
+      "step": 1153
+    },
+    {
+      "epoch": 4.672064777327935,
+      "grad_norm": 0.20202323262109637,
+      "learning_rate": 5.22326650565701e-07,
+      "loss": 0.3794,
+      "step": 1154
+    },
+    {
+      "epoch": 4.676113360323887,
+      "grad_norm": 0.20084499787316662,
+      "learning_rate": 5.095640171043048e-07,
+      "loss": 0.3859,
+      "step": 1155
+    },
+    {
+      "epoch": 4.680161943319838,
+      "grad_norm": 0.19426714265914108,
+      "learning_rate": 4.969572286496594e-07,
+      "loss": 0.3769,
+      "step": 1156
+    },
+    {
+      "epoch": 4.684210526315789,
+      "grad_norm": 0.20671614223906445,
+      "learning_rate": 4.845063860053079e-07,
+      "loss": 0.3876,
+      "step": 1157
+    },
+    {
+      "epoch": 4.6882591093117405,
+      "grad_norm": 0.19976182680814145,
+      "learning_rate": 4.7221158872785997e-07,
+      "loss": 0.3757,
+      "step": 1158
+    },
+    {
+      "epoch": 4.6923076923076925,
+      "grad_norm": 0.20121109481392807,
+      "learning_rate": 4.600729351261857e-07,
+      "loss": 0.3739,
+      "step": 1159
+    },
+    {
+      "epoch": 4.696356275303644,
+      "grad_norm": 0.20005351888518888,
+      "learning_rate": 4.4809052226063355e-07,
+      "loss": 0.3767,
+      "step": 1160
+    },
+    {
+      "epoch": 4.700404858299595,
+      "grad_norm": 0.20328336523485976,
+      "learning_rate": 4.3626444594226046e-07,
+      "loss": 0.3823,
+      "step": 1161
+    },
+    {
+      "epoch": 4.704453441295547,
+      "grad_norm": 0.19680638597620317,
+      "learning_rate": 4.245948007320588e-07,
+      "loss": 0.3756,
+      "step": 1162
+    },
+    {
+      "epoch": 4.708502024291498,
+      "grad_norm": 0.19741648818785884,
+      "learning_rate": 4.130816799401971e-07,
+      "loss": 0.3846,
+      "step": 1163
+    },
+    {
+      "epoch": 4.712550607287449,
+      "grad_norm": 0.1953096348900384,
+      "learning_rate": 4.017251756252827e-07,
+      "loss": 0.3746,
+      "step": 1164
+    },
+    {
+      "epoch": 4.716599190283401,
+      "grad_norm": 0.20260828644676163,
+      "learning_rate": 3.9052537859362473e-07,
+      "loss": 0.3863,
+      "step": 1165
+    },
+    {
+      "epoch": 4.720647773279352,
+      "grad_norm": 0.1978026590507277,
+      "learning_rate": 3.7948237839849687e-07,
+      "loss": 0.3794,
+      "step": 1166
+    },
+    {
+      "epoch": 4.724696356275303,
+      "grad_norm": 0.19407052717066137,
+      "learning_rate": 3.6859626333943797e-07,
+      "loss": 0.3738,
+      "step": 1167
+    },
+    {
+      "epoch": 4.728744939271255,
+      "grad_norm": 0.194838458179867,
+      "learning_rate": 3.578671204615347e-07,
+      "loss": 0.3793,
+      "step": 1168
+    },
+    {
+      "epoch": 4.732793522267206,
+      "grad_norm": 0.2080182364408555,
+      "learning_rate": 3.472950355547267e-07,
+      "loss": 0.3798,
+      "step": 1169
+    },
+    {
+      "epoch": 4.7368421052631575,
+      "grad_norm": 0.19582067749452292,
+      "learning_rate": 3.3688009315312244e-07,
+      "loss": 0.3775,
+      "step": 1170
+    },
+    {
+      "epoch": 4.7408906882591095,
+      "grad_norm": 0.19539244596854596,
+      "learning_rate": 3.266223765343224e-07,
+      "loss": 0.3731,
+      "step": 1171
+    },
+    {
+      "epoch": 4.744939271255061,
+      "grad_norm": 0.20256545562999173,
+      "learning_rate": 3.16521967718757e-07,
+      "loss": 0.3823,
+      "step": 1172
+    },
+    {
+      "epoch": 4.748987854251012,
+      "grad_norm": 0.19838078203553217,
+      "learning_rate": 3.0657894746902726e-07,
+      "loss": 0.3786,
+      "step": 1173
+    },
+    {
+      "epoch": 4.753036437246964,
+      "grad_norm": 0.20763805979780672,
+      "learning_rate": 2.9679339528925654e-07,
+      "loss": 0.3712,
+      "step": 1174
+    },
+    {
+      "epoch": 4.757085020242915,
+      "grad_norm": 0.19971202400414806,
+      "learning_rate": 2.871653894244597e-07,
+      "loss": 0.374,
+      "step": 1175
+    },
+    {
+      "epoch": 4.761133603238866,
+      "grad_norm": 0.19807891213461387,
+      "learning_rate": 2.7769500685991266e-07,
+      "loss": 0.3797,
+      "step": 1176
+    },
+    {
+      "epoch": 4.765182186234818,
+      "grad_norm": 0.19377557897416284,
+      "learning_rate": 2.683823233205485e-07,
+      "loss": 0.3759,
+      "step": 1177
+    },
+    {
+      "epoch": 4.769230769230769,
+      "grad_norm": 0.191168972434604,
+      "learning_rate": 2.592274132703287e-07,
+      "loss": 0.3665,
+      "step": 1178
+    },
+    {
+      "epoch": 4.77327935222672,
+      "grad_norm": 0.20758920488553578,
+      "learning_rate": 2.502303499116754e-07,
+      "loss": 0.3711,
+      "step": 1179
+    },
+    {
+      "epoch": 4.777327935222672,
+      "grad_norm": 0.20224080894830998,
+      "learning_rate": 2.413912051848666e-07,
+      "loss": 0.3843,
+      "step": 1180
+    },
+    {
+      "epoch": 4.781376518218623,
+      "grad_norm": 0.20537069208699343,
+      "learning_rate": 2.3271004976746613e-07,
+      "loss": 0.3788,
+      "step": 1181
+    },
+    {
+      "epoch": 4.7854251012145745,
+      "grad_norm": 0.19511346706067131,
+      "learning_rate": 2.2418695307376393e-07,
+      "loss": 0.3737,
+      "step": 1182
+    },
+    {
+      "epoch": 4.7894736842105265,
+      "grad_norm": 0.19872827099101556,
+      "learning_rate": 2.1582198325420968e-07,
+      "loss": 0.3838,
+      "step": 1183
+    },
+    {
+      "epoch": 4.793522267206478,
+      "grad_norm": 0.19927472621044962,
+      "learning_rate": 2.0761520719488447e-07,
+      "loss": 0.3798,
+      "step": 1184
+    },
+    {
+      "epoch": 4.797570850202429,
+      "grad_norm": 0.20010316635503095,
+      "learning_rate": 1.9956669051694798e-07,
+      "loss": 0.3835,
+      "step": 1185
+    },
+    {
+      "epoch": 4.801619433198381,
+      "grad_norm": 0.2007711698098253,
+      "learning_rate": 1.916764975761254e-07,
+      "loss": 0.3802,
+      "step": 1186
+    },
+    {
+      "epoch": 4.805668016194332,
+      "grad_norm": 0.19871470912006964,
+      "learning_rate": 1.839446914621923e-07,
+      "loss": 0.3888,
+      "step": 1187
+    },
+    {
+      "epoch": 4.809716599190283,
+      "grad_norm": 0.19454825549188806,
+      "learning_rate": 1.7637133399845962e-07,
+      "loss": 0.378,
+      "step": 1188
+    },
+    {
+      "epoch": 4.813765182186235,
+      "grad_norm": 0.2027584150135223,
+      "learning_rate": 1.6895648574129397e-07,
+      "loss": 0.3759,
+      "step": 1189
+    },
+    {
+      "epoch": 4.817813765182186,
+      "grad_norm": 0.20186070636771236,
+      "learning_rate": 1.6170020597962465e-07,
+      "loss": 0.3824,
+      "step": 1190
+    },
+    {
+      "epoch": 4.821862348178137,
+      "grad_norm": 0.20018915197766496,
+      "learning_rate": 1.5460255273447078e-07,
+      "loss": 0.3816,
+      "step": 1191
+    },
+    {
+      "epoch": 4.825910931174089,
+      "grad_norm": 0.20775327932673107,
+      "learning_rate": 1.4766358275847715e-07,
+      "loss": 0.3808,
+      "step": 1192
+    },
+    {
+      "epoch": 4.82995951417004,
+      "grad_norm": 0.20400145550766574,
+      "learning_rate": 1.4088335153546573e-07,
+      "loss": 0.3782,
+      "step": 1193
+    },
+    {
+      "epoch": 4.834008097165992,
+      "grad_norm": 0.2005630179545976,
+      "learning_rate": 1.3426191327998496e-07,
+      "loss": 0.3856,
+      "step": 1194
+    },
+    {
+      "epoch": 4.838056680161944,
+      "grad_norm": 0.197771797611627,
+      "learning_rate": 1.277993209368744e-07,
+      "loss": 0.3809,
+      "step": 1195
+    },
+    {
+      "epoch": 4.842105263157895,
+      "grad_norm": 0.1954464332274358,
+      "learning_rate": 1.214956261808542e-07,
+      "loss": 0.3815,
+      "step": 1196
+    },
+    {
+      "epoch": 4.846153846153846,
+      "grad_norm": 0.1984297934148147,
+      "learning_rate": 1.1535087941609846e-07,
+      "loss": 0.3767,
+      "step": 1197
+    },
+    {
+      "epoch": 4.850202429149798,
+      "grad_norm": 0.19028075946432652,
+      "learning_rate": 1.0936512977583802e-07,
+      "loss": 0.3762,
+      "step": 1198
+    },
+    {
+      "epoch": 4.854251012145749,
+      "grad_norm": 0.20637010572159606,
+      "learning_rate": 1.0353842512196732e-07,
+      "loss": 0.3754,
+      "step": 1199
+    },
+    {
+      "epoch": 4.8582995951417,
+      "grad_norm": 0.19830854512030083,
+      "learning_rate": 9.787081204466253e-08,
+      "loss": 0.3805,
+      "step": 1200
+    },
+    {
+      "epoch": 4.862348178137652,
+      "grad_norm": 0.19672838879752394,
+      "learning_rate": 9.236233586200627e-08,
+      "loss": 0.3789,
+      "step": 1201
+    },
+    {
+      "epoch": 4.866396761133603,
+      "grad_norm": 0.20197559209569377,
+      "learning_rate": 8.70130406196279e-08,
+      "loss": 0.3893,
+      "step": 1202
+    },
+    {
+      "epoch": 4.870445344129554,
+      "grad_norm": 0.1974529027786469,
+      "learning_rate": 8.182296909035047e-08,
+      "loss": 0.3801,
+      "step": 1203
+    },
+    {
+      "epoch": 4.874493927125506,
+      "grad_norm": 0.19554733831422533,
+      "learning_rate": 7.679216277384439e-08,
+      "loss": 0.3714,
+      "step": 1204
+    },
+    {
+      "epoch": 4.8785425101214575,
+      "grad_norm": 0.19525973278414716,
+      "learning_rate": 7.19206618963053e-08,
+      "loss": 0.3887,
+      "step": 1205
+    },
+    {
+      "epoch": 4.882591093117409,
+      "grad_norm": 0.19195706377631905,
+      "learning_rate": 6.720850541012347e-08,
+      "loss": 0.3833,
+      "step": 1206
+    },
+    {
+      "epoch": 4.886639676113361,
+      "grad_norm": 0.2005047946741055,
+      "learning_rate": 6.265573099357714e-08,
+      "loss": 0.3734,
+      "step": 1207
+    },
+    {
+      "epoch": 4.890688259109312,
+      "grad_norm": 0.19163325893173405,
+      "learning_rate": 5.826237505053067e-08,
+      "loss": 0.3728,
+      "step": 1208
+    },
+    {
+      "epoch": 4.894736842105263,
+      "grad_norm": 0.2070940242976453,
+      "learning_rate": 5.4028472710137e-08,
+      "loss": 0.3806,
+      "step": 1209
+    },
+    {
+      "epoch": 4.898785425101215,
+      "grad_norm": 0.19358020228493716,
+      "learning_rate": 4.9954057826571146e-08,
+      "loss": 0.3809,
+      "step": 1210
+    },
+    {
+      "epoch": 4.902834008097166,
+      "grad_norm": 0.19303239752211104,
+      "learning_rate": 4.6039162978743776e-08,
+      "loss": 0.376,
+      "step": 1211
+    },
+    {
+      "epoch": 4.906882591093117,
+      "grad_norm": 0.20446596602300712,
+      "learning_rate": 4.228381947004812e-08,
+      "loss": 0.3705,
+      "step": 1212
+    },
+    {
+      "epoch": 4.910931174089069,
+      "grad_norm": 0.19568850323476417,
+      "learning_rate": 3.868805732811343e-08,
+      "loss": 0.3765,
+      "step": 1213
+    },
+    {
+      "epoch": 4.91497975708502,
+      "grad_norm": 0.20159188656492336,
+      "learning_rate": 3.525190530455636e-08,
+      "loss": 0.3888,
+      "step": 1214
+    },
+    {
+      "epoch": 4.919028340080971,
+      "grad_norm": 0.20229038491146697,
+      "learning_rate": 3.197539087475887e-08,
+      "loss": 0.3826,
+      "step": 1215
+    },
+    {
+      "epoch": 4.923076923076923,
+      "grad_norm": 0.1960319482347613,
+      "learning_rate": 2.8858540237641786e-08,
+      "loss": 0.3777,
+      "step": 1216
+    },
+    {
+      "epoch": 4.9271255060728745,
+      "grad_norm": 0.20558501481799932,
+      "learning_rate": 2.5901378315458247e-08,
+      "loss": 0.374,
+      "step": 1217
+    },
+    {
+      "epoch": 4.931174089068826,
+      "grad_norm": 0.19364206401170356,
+      "learning_rate": 2.3103928753598347e-08,
+      "loss": 0.3757,
+      "step": 1218
+    },
+    {
+      "epoch": 4.935222672064778,
+      "grad_norm": 0.20162844280638123,
+      "learning_rate": 2.0466213920393718e-08,
+      "loss": 0.3849,
+      "step": 1219
+    },
+    {
+      "epoch": 4.939271255060729,
+      "grad_norm": 0.19431013967578784,
+      "learning_rate": 1.7988254906942115e-08,
+      "loss": 0.3894,
+      "step": 1220
+    },
+    {
+      "epoch": 4.94331983805668,
+      "grad_norm": 0.1945454834200414,
+      "learning_rate": 1.5670071526936447e-08,
+      "loss": 0.373,
+      "step": 1221
+    },
+    {
+      "epoch": 4.947368421052632,
+      "grad_norm": 0.19518937552090249,
+      "learning_rate": 1.3511682316509344e-08,
+      "loss": 0.3784,
+      "step": 1222
+    },
+    {
+      "epoch": 4.951417004048583,
+      "grad_norm": 0.20434370077359434,
+      "learning_rate": 1.1513104534086605e-08,
+      "loss": 0.3761,
+      "step": 1223
+    },
+    {
+      "epoch": 4.955465587044534,
+      "grad_norm": 0.1967372814439288,
+      "learning_rate": 9.67435416023843e-09,
+      "loss": 0.3766,
+      "step": 1224
+    },
+    {
+      "epoch": 4.959514170040486,
+      "grad_norm": 0.19500111550064078,
+      "learning_rate": 7.995445897566178e-09,
+      "loss": 0.3712,
+      "step": 1225
+    },
+    {
+      "epoch": 4.963562753036437,
+      "grad_norm": 0.19568948760871033,
+      "learning_rate": 6.476393170578021e-09,
+      "loss": 0.3778,
+      "step": 1226
+    },
+    {
+      "epoch": 4.967611336032388,
+      "grad_norm": 0.1981793227428959,
+      "learning_rate": 5.11720812557348e-09,
+      "loss": 0.3733,
+      "step": 1227
+    },
+    {
+      "epoch": 4.97165991902834,
+      "grad_norm": 0.19652756904555865,
+      "learning_rate": 3.917901630563492e-09,
+      "loss": 0.3842,
+      "step": 1228
+    },
+    {
+      "epoch": 4.9757085020242915,
+      "grad_norm": 0.20023648823410112,
+      "learning_rate": 2.878483275170485e-09,
+      "loss": 0.3797,
+      "step": 1229
+    },
+    {
+      "epoch": 4.979757085020243,
+      "grad_norm": 0.2027297942045997,
+      "learning_rate": 1.998961370555108e-09,
+      "loss": 0.3752,
+      "step": 1230
+    },
+    {
+      "epoch": 4.983805668016195,
+      "grad_norm": 0.19805636558717915,
+      "learning_rate": 1.2793429493518361e-09,
+      "loss": 0.3792,
+      "step": 1231
+    },
+    {
+      "epoch": 4.987854251012146,
+      "grad_norm": 0.19997107410353115,
+      "learning_rate": 7.196337656112384e-10,
+      "loss": 0.3737,
+      "step": 1232
+    },
+    {
+      "epoch": 4.991902834008097,
+      "grad_norm": 0.19017976741047934,
+      "learning_rate": 3.198382947511292e-10,
+      "loss": 0.3822,
+      "step": 1233
+    },
+    {
+      "epoch": 4.995951417004049,
+      "grad_norm": 0.19703579010175537,
+      "learning_rate": 7.995973352770137e-11,
+      "loss": 0.3772,
+      "step": 1234
+    },
+    {
+      "epoch": 5.0,
+      "grad_norm": 0.1947342814833797,
+      "learning_rate": 0.0,
+      "loss": 0.378,
+      "step": 1235
+    },
+    {
+      "epoch": 5.0,
+      "step": 1235,
+      "total_flos": 5.115840720351003e+18,
+      "train_loss": 0.5002805300328413,
+      "train_runtime": 59099.4794,
+      "train_samples_per_second": 2.673,
+      "train_steps_per_second": 0.021
+    }
+  ],
+  "logging_steps": 1.0,
+  "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": 5.115840720351003e+18,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}