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": 6.676145835751954,
+      "learning_rate": 3.2258064516129035e-07,
+      "loss": 1.0704,
+      "step": 1
+    },
+    {
+      "epoch": 0.008097165991902834,
+      "grad_norm": 6.651747207457028,
+      "learning_rate": 6.451612903225807e-07,
+      "loss": 1.0524,
+      "step": 2
+    },
+    {
+      "epoch": 0.012145748987854251,
+      "grad_norm": 6.640763269614994,
+      "learning_rate": 9.67741935483871e-07,
+      "loss": 1.0764,
+      "step": 3
+    },
+    {
+      "epoch": 0.016194331983805668,
+      "grad_norm": 6.364684524464274,
+      "learning_rate": 1.2903225806451614e-06,
+      "loss": 1.0412,
+      "step": 4
+    },
+    {
+      "epoch": 0.020242914979757085,
+      "grad_norm": 6.0785030057208385,
+      "learning_rate": 1.6129032258064516e-06,
+      "loss": 1.016,
+      "step": 5
+    },
+    {
+      "epoch": 0.024291497975708502,
+      "grad_norm": 6.071476363399392,
+      "learning_rate": 1.935483870967742e-06,
+      "loss": 1.0565,
+      "step": 6
+    },
+    {
+      "epoch": 0.02834008097165992,
+      "grad_norm": 4.849942102352849,
+      "learning_rate": 2.2580645161290324e-06,
+      "loss": 1.0066,
+      "step": 7
+    },
+    {
+      "epoch": 0.032388663967611336,
+      "grad_norm": 4.812594677152561,
+      "learning_rate": 2.580645161290323e-06,
+      "loss": 1.0316,
+      "step": 8
+    },
+    {
+      "epoch": 0.03643724696356275,
+      "grad_norm": 2.475818300394419,
+      "learning_rate": 2.903225806451613e-06,
+      "loss": 0.9264,
+      "step": 9
+    },
+    {
+      "epoch": 0.04048582995951417,
+      "grad_norm": 2.3392570021499015,
+      "learning_rate": 3.225806451612903e-06,
+      "loss": 0.9801,
+      "step": 10
+    },
+    {
+      "epoch": 0.044534412955465584,
+      "grad_norm": 2.230899943276188,
+      "learning_rate": 3.548387096774194e-06,
+      "loss": 0.9628,
+      "step": 11
+    },
+    {
+      "epoch": 0.048582995951417005,
+      "grad_norm": 4.428765714344261,
+      "learning_rate": 3.870967741935484e-06,
+      "loss": 0.9572,
+      "step": 12
+    },
+    {
+      "epoch": 0.05263157894736842,
+      "grad_norm": 4.732766236612769,
+      "learning_rate": 4.193548387096774e-06,
+      "loss": 0.9547,
+      "step": 13
+    },
+    {
+      "epoch": 0.05668016194331984,
+      "grad_norm": 4.252706598803878,
+      "learning_rate": 4.516129032258065e-06,
+      "loss": 0.8816,
+      "step": 14
+    },
+    {
+      "epoch": 0.06072874493927125,
+      "grad_norm": 4.087429123753276,
+      "learning_rate": 4.838709677419355e-06,
+      "loss": 0.9191,
+      "step": 15
+    },
+    {
+      "epoch": 0.06477732793522267,
+      "grad_norm": 3.301640567976228,
+      "learning_rate": 5.161290322580646e-06,
+      "loss": 0.8802,
+      "step": 16
+    },
+    {
+      "epoch": 0.06882591093117409,
+      "grad_norm": 2.945907185011285,
+      "learning_rate": 5.483870967741935e-06,
+      "loss": 0.8473,
+      "step": 17
+    },
+    {
+      "epoch": 0.0728744939271255,
+      "grad_norm": 2.8000468339784685,
+      "learning_rate": 5.806451612903226e-06,
+      "loss": 0.8777,
+      "step": 18
+    },
+    {
+      "epoch": 0.07692307692307693,
+      "grad_norm": 2.394931733556691,
+      "learning_rate": 6.129032258064517e-06,
+      "loss": 0.8289,
+      "step": 19
+    },
+    {
+      "epoch": 0.08097165991902834,
+      "grad_norm": 2.0216212639360642,
+      "learning_rate": 6.451612903225806e-06,
+      "loss": 0.8378,
+      "step": 20
+    },
+    {
+      "epoch": 0.08502024291497975,
+      "grad_norm": 1.5167177615708118,
+      "learning_rate": 6.774193548387097e-06,
+      "loss": 0.7935,
+      "step": 21
+    },
+    {
+      "epoch": 0.08906882591093117,
+      "grad_norm": 1.8827168272821462,
+      "learning_rate": 7.096774193548388e-06,
+      "loss": 0.8371,
+      "step": 22
+    },
+    {
+      "epoch": 0.0931174089068826,
+      "grad_norm": 1.8153318529170588,
+      "learning_rate": 7.4193548387096784e-06,
+      "loss": 0.7543,
+      "step": 23
+    },
+    {
+      "epoch": 0.09716599190283401,
+      "grad_norm": 1.8851433533012525,
+      "learning_rate": 7.741935483870968e-06,
+      "loss": 0.7897,
+      "step": 24
+    },
+    {
+      "epoch": 0.10121457489878542,
+      "grad_norm": 1.540084502338418,
+      "learning_rate": 8.064516129032258e-06,
+      "loss": 0.795,
+      "step": 25
+    },
+    {
+      "epoch": 0.10526315789473684,
+      "grad_norm": 1.2481397303900987,
+      "learning_rate": 8.387096774193549e-06,
+      "loss": 0.7862,
+      "step": 26
+    },
+    {
+      "epoch": 0.10931174089068826,
+      "grad_norm": 1.2423403546912808,
+      "learning_rate": 8.70967741935484e-06,
+      "loss": 0.7809,
+      "step": 27
+    },
+    {
+      "epoch": 0.11336032388663968,
+      "grad_norm": 1.4597097667902006,
+      "learning_rate": 9.03225806451613e-06,
+      "loss": 0.7578,
+      "step": 28
+    },
+    {
+      "epoch": 0.11740890688259109,
+      "grad_norm": 1.1760102176137328,
+      "learning_rate": 9.35483870967742e-06,
+      "loss": 0.7398,
+      "step": 29
+    },
+    {
+      "epoch": 0.1214574898785425,
+      "grad_norm": 0.9997036241703402,
+      "learning_rate": 9.67741935483871e-06,
+      "loss": 0.7547,
+      "step": 30
+    },
+    {
+      "epoch": 0.12550607287449392,
+      "grad_norm": 0.934136949452039,
+      "learning_rate": 1e-05,
+      "loss": 0.7211,
+      "step": 31
+    },
+    {
+      "epoch": 0.12955465587044535,
+      "grad_norm": 1.0055186934772933,
+      "learning_rate": 1.0322580645161291e-05,
+      "loss": 0.7444,
+      "step": 32
+    },
+    {
+      "epoch": 0.13360323886639677,
+      "grad_norm": 0.8652648882349199,
+      "learning_rate": 1.0645161290322582e-05,
+      "loss": 0.7594,
+      "step": 33
+    },
+    {
+      "epoch": 0.13765182186234817,
+      "grad_norm": 16.559698655485235,
+      "learning_rate": 1.096774193548387e-05,
+      "loss": 0.7595,
+      "step": 34
+    },
+    {
+      "epoch": 0.1417004048582996,
+      "grad_norm": 1.2905596390380618,
+      "learning_rate": 1.1290322580645164e-05,
+      "loss": 0.7274,
+      "step": 35
+    },
+    {
+      "epoch": 0.145748987854251,
+      "grad_norm": 1.0140552636015814,
+      "learning_rate": 1.1612903225806453e-05,
+      "loss": 0.7345,
+      "step": 36
+    },
+    {
+      "epoch": 0.14979757085020243,
+      "grad_norm": 0.7060379176173712,
+      "learning_rate": 1.1935483870967743e-05,
+      "loss": 0.7624,
+      "step": 37
+    },
+    {
+      "epoch": 0.15384615384615385,
+      "grad_norm": 0.7965408473924335,
+      "learning_rate": 1.2258064516129034e-05,
+      "loss": 0.7575,
+      "step": 38
+    },
+    {
+      "epoch": 0.15789473684210525,
+      "grad_norm": 0.9257518620172522,
+      "learning_rate": 1.2580645161290324e-05,
+      "loss": 0.7253,
+      "step": 39
+    },
+    {
+      "epoch": 0.16194331983805668,
+      "grad_norm": 0.8182119128938206,
+      "learning_rate": 1.2903225806451613e-05,
+      "loss": 0.7106,
+      "step": 40
+    },
+    {
+      "epoch": 0.1659919028340081,
+      "grad_norm": 0.7863783167388365,
+      "learning_rate": 1.3225806451612903e-05,
+      "loss": 0.7293,
+      "step": 41
+    },
+    {
+      "epoch": 0.1700404858299595,
+      "grad_norm": 0.8798857541610017,
+      "learning_rate": 1.3548387096774194e-05,
+      "loss": 0.7622,
+      "step": 42
+    },
+    {
+      "epoch": 0.17408906882591094,
+      "grad_norm": 0.7108387625742065,
+      "learning_rate": 1.3870967741935486e-05,
+      "loss": 0.7188,
+      "step": 43
+    },
+    {
+      "epoch": 0.17813765182186234,
+      "grad_norm": 0.7392131635415615,
+      "learning_rate": 1.4193548387096776e-05,
+      "loss": 0.7208,
+      "step": 44
+    },
+    {
+      "epoch": 0.18218623481781376,
+      "grad_norm": 0.7223455529614952,
+      "learning_rate": 1.4516129032258066e-05,
+      "loss": 0.7226,
+      "step": 45
+    },
+    {
+      "epoch": 0.1862348178137652,
+      "grad_norm": 0.7774432948429539,
+      "learning_rate": 1.4838709677419357e-05,
+      "loss": 0.6944,
+      "step": 46
+    },
+    {
+      "epoch": 0.1902834008097166,
+      "grad_norm": 0.6666733235407696,
+      "learning_rate": 1.5161290322580646e-05,
+      "loss": 0.7353,
+      "step": 47
+    },
+    {
+      "epoch": 0.19433198380566802,
+      "grad_norm": 0.7106718471903637,
+      "learning_rate": 1.5483870967741936e-05,
+      "loss": 0.7143,
+      "step": 48
+    },
+    {
+      "epoch": 0.19838056680161945,
+      "grad_norm": 0.6361648777890632,
+      "learning_rate": 1.5806451612903226e-05,
+      "loss": 0.7003,
+      "step": 49
+    },
+    {
+      "epoch": 0.20242914979757085,
+      "grad_norm": 0.6707410919119826,
+      "learning_rate": 1.6129032258064517e-05,
+      "loss": 0.6715,
+      "step": 50
+    },
+    {
+      "epoch": 0.20647773279352227,
+      "grad_norm": 0.6790494451909449,
+      "learning_rate": 1.6451612903225807e-05,
+      "loss": 0.7181,
+      "step": 51
+    },
+    {
+      "epoch": 0.21052631578947367,
+      "grad_norm": 0.7035975158912797,
+      "learning_rate": 1.6774193548387098e-05,
+      "loss": 0.7069,
+      "step": 52
+    },
+    {
+      "epoch": 0.2145748987854251,
+      "grad_norm": 0.6616161121629226,
+      "learning_rate": 1.7096774193548388e-05,
+      "loss": 0.6979,
+      "step": 53
+    },
+    {
+      "epoch": 0.21862348178137653,
+      "grad_norm": 0.6771539857140317,
+      "learning_rate": 1.741935483870968e-05,
+      "loss": 0.7035,
+      "step": 54
+    },
+    {
+      "epoch": 0.22267206477732793,
+      "grad_norm": 0.6735056414050704,
+      "learning_rate": 1.774193548387097e-05,
+      "loss": 0.6938,
+      "step": 55
+    },
+    {
+      "epoch": 0.22672064777327935,
+      "grad_norm": 0.736973242657467,
+      "learning_rate": 1.806451612903226e-05,
+      "loss": 0.6883,
+      "step": 56
+    },
+    {
+      "epoch": 0.23076923076923078,
+      "grad_norm": 0.8609213706953633,
+      "learning_rate": 1.838709677419355e-05,
+      "loss": 0.7049,
+      "step": 57
+    },
+    {
+      "epoch": 0.23481781376518218,
+      "grad_norm": 1.0631754590979579,
+      "learning_rate": 1.870967741935484e-05,
+      "loss": 0.674,
+      "step": 58
+    },
+    {
+      "epoch": 0.2388663967611336,
+      "grad_norm": 4.067089698476355,
+      "learning_rate": 1.903225806451613e-05,
+      "loss": 0.7024,
+      "step": 59
+    },
+    {
+      "epoch": 0.242914979757085,
+      "grad_norm": 1.224151023003773,
+      "learning_rate": 1.935483870967742e-05,
+      "loss": 0.675,
+      "step": 60
+    },
+    {
+      "epoch": 0.24696356275303644,
+      "grad_norm": 0.7089882103177129,
+      "learning_rate": 1.967741935483871e-05,
+      "loss": 0.6903,
+      "step": 61
+    },
+    {
+      "epoch": 0.25101214574898784,
+      "grad_norm": 7.430285613495958,
+      "learning_rate": 2e-05,
+      "loss": 0.6763,
+      "step": 62
+    },
+    {
+      "epoch": 0.2550607287449393,
+      "grad_norm": 1.614692402041524,
+      "learning_rate": 2.0322580645161292e-05,
+      "loss": 0.6937,
+      "step": 63
+    },
+    {
+      "epoch": 0.2591093117408907,
+      "grad_norm": 0.9738519244640278,
+      "learning_rate": 2.0645161290322582e-05,
+      "loss": 0.6801,
+      "step": 64
+    },
+    {
+      "epoch": 0.2631578947368421,
+      "grad_norm": 0.9050835318648012,
+      "learning_rate": 2.0967741935483873e-05,
+      "loss": 0.6772,
+      "step": 65
+    },
+    {
+      "epoch": 0.26720647773279355,
+      "grad_norm": 1.2578995848157943,
+      "learning_rate": 2.1290322580645163e-05,
+      "loss": 0.6698,
+      "step": 66
+    },
+    {
+      "epoch": 0.27125506072874495,
+      "grad_norm": 0.776745571621289,
+      "learning_rate": 2.161290322580645e-05,
+      "loss": 0.6857,
+      "step": 67
+    },
+    {
+      "epoch": 0.27530364372469635,
+      "grad_norm": 0.7972208128397074,
+      "learning_rate": 2.193548387096774e-05,
+      "loss": 0.6569,
+      "step": 68
+    },
+    {
+      "epoch": 0.2793522267206478,
+      "grad_norm": 1.2879033663387678,
+      "learning_rate": 2.225806451612903e-05,
+      "loss": 0.6885,
+      "step": 69
+    },
+    {
+      "epoch": 0.2834008097165992,
+      "grad_norm": 0.6990202105618228,
+      "learning_rate": 2.2580645161290328e-05,
+      "loss": 0.7075,
+      "step": 70
+    },
+    {
+      "epoch": 0.2874493927125506,
+      "grad_norm": 0.8590597647882539,
+      "learning_rate": 2.2903225806451615e-05,
+      "loss": 0.6863,
+      "step": 71
+    },
+    {
+      "epoch": 0.291497975708502,
+      "grad_norm": 0.6838594798352408,
+      "learning_rate": 2.3225806451612906e-05,
+      "loss": 0.6601,
+      "step": 72
+    },
+    {
+      "epoch": 0.29554655870445345,
+      "grad_norm": 0.8010564272771952,
+      "learning_rate": 2.3548387096774196e-05,
+      "loss": 0.6606,
+      "step": 73
+    },
+    {
+      "epoch": 0.29959514170040485,
+      "grad_norm": 0.7559394781056001,
+      "learning_rate": 2.3870967741935486e-05,
+      "loss": 0.6936,
+      "step": 74
+    },
+    {
+      "epoch": 0.30364372469635625,
+      "grad_norm": 0.713675965782304,
+      "learning_rate": 2.4193548387096777e-05,
+      "loss": 0.6853,
+      "step": 75
+    },
+    {
+      "epoch": 0.3076923076923077,
+      "grad_norm": 0.8615517027634002,
+      "learning_rate": 2.4516129032258067e-05,
+      "loss": 0.6935,
+      "step": 76
+    },
+    {
+      "epoch": 0.3117408906882591,
+      "grad_norm": 0.8492772767053012,
+      "learning_rate": 2.4838709677419358e-05,
+      "loss": 0.693,
+      "step": 77
+    },
+    {
+      "epoch": 0.3157894736842105,
+      "grad_norm": 0.9435154831964874,
+      "learning_rate": 2.5161290322580648e-05,
+      "loss": 0.7039,
+      "step": 78
+    },
+    {
+      "epoch": 0.31983805668016196,
+      "grad_norm": 0.6656371027378365,
+      "learning_rate": 2.548387096774194e-05,
+      "loss": 0.668,
+      "step": 79
+    },
+    {
+      "epoch": 0.32388663967611336,
+      "grad_norm": 0.8465135938014197,
+      "learning_rate": 2.5806451612903226e-05,
+      "loss": 0.6782,
+      "step": 80
+    },
+    {
+      "epoch": 0.32793522267206476,
+      "grad_norm": 0.902280580296946,
+      "learning_rate": 2.6129032258064516e-05,
+      "loss": 0.6694,
+      "step": 81
+    },
+    {
+      "epoch": 0.3319838056680162,
+      "grad_norm": 0.7486175177584709,
+      "learning_rate": 2.6451612903225806e-05,
+      "loss": 0.6658,
+      "step": 82
+    },
+    {
+      "epoch": 0.3360323886639676,
+      "grad_norm": 1.1701673617263375,
+      "learning_rate": 2.6774193548387097e-05,
+      "loss": 0.6802,
+      "step": 83
+    },
+    {
+      "epoch": 0.340080971659919,
+      "grad_norm": 0.7335179320209914,
+      "learning_rate": 2.7096774193548387e-05,
+      "loss": 0.6765,
+      "step": 84
+    },
+    {
+      "epoch": 0.3441295546558704,
+      "grad_norm": 0.8598699421930768,
+      "learning_rate": 2.7419354838709678e-05,
+      "loss": 0.6915,
+      "step": 85
+    },
+    {
+      "epoch": 0.3481781376518219,
+      "grad_norm": 0.9742006307370583,
+      "learning_rate": 2.774193548387097e-05,
+      "loss": 0.6856,
+      "step": 86
+    },
+    {
+      "epoch": 0.3522267206477733,
+      "grad_norm": 0.7318347937061167,
+      "learning_rate": 2.8064516129032262e-05,
+      "loss": 0.6625,
+      "step": 87
+    },
+    {
+      "epoch": 0.3562753036437247,
+      "grad_norm": 0.8616684422071693,
+      "learning_rate": 2.8387096774193552e-05,
+      "loss": 0.6791,
+      "step": 88
+    },
+    {
+      "epoch": 0.3603238866396761,
+      "grad_norm": 0.7756773070141108,
+      "learning_rate": 2.8709677419354843e-05,
+      "loss": 0.645,
+      "step": 89
+    },
+    {
+      "epoch": 0.3643724696356275,
+      "grad_norm": 0.9623597445585983,
+      "learning_rate": 2.9032258064516133e-05,
+      "loss": 0.6775,
+      "step": 90
+    },
+    {
+      "epoch": 0.3684210526315789,
+      "grad_norm": 1.162658215626486,
+      "learning_rate": 2.9354838709677423e-05,
+      "loss": 0.6568,
+      "step": 91
+    },
+    {
+      "epoch": 0.3724696356275304,
+      "grad_norm": 0.7771956471932099,
+      "learning_rate": 2.9677419354838714e-05,
+      "loss": 0.6663,
+      "step": 92
+    },
+    {
+      "epoch": 0.3765182186234818,
+      "grad_norm": 1.125913169231333,
+      "learning_rate": 3.0000000000000004e-05,
+      "loss": 0.6638,
+      "step": 93
+    },
+    {
+      "epoch": 0.3805668016194332,
+      "grad_norm": 0.8070681625512021,
+      "learning_rate": 3.032258064516129e-05,
+      "loss": 0.653,
+      "step": 94
+    },
+    {
+      "epoch": 0.38461538461538464,
+      "grad_norm": 1.066373637705334,
+      "learning_rate": 3.0645161290322585e-05,
+      "loss": 0.6559,
+      "step": 95
+    },
+    {
+      "epoch": 0.38866396761133604,
+      "grad_norm": 0.8813816891557215,
+      "learning_rate": 3.096774193548387e-05,
+      "loss": 0.6527,
+      "step": 96
+    },
+    {
+      "epoch": 0.39271255060728744,
+      "grad_norm": 1.0204131578845181,
+      "learning_rate": 3.1290322580645166e-05,
+      "loss": 0.6945,
+      "step": 97
+    },
+    {
+      "epoch": 0.3967611336032389,
+      "grad_norm": 0.8371480529684904,
+      "learning_rate": 3.161290322580645e-05,
+      "loss": 0.6462,
+      "step": 98
+    },
+    {
+      "epoch": 0.4008097165991903,
+      "grad_norm": 0.7150252989345846,
+      "learning_rate": 3.193548387096774e-05,
+      "loss": 0.6701,
+      "step": 99
+    },
+    {
+      "epoch": 0.4048582995951417,
+      "grad_norm": 0.7849315174794829,
+      "learning_rate": 3.2258064516129034e-05,
+      "loss": 0.6498,
+      "step": 100
+    },
+    {
+      "epoch": 0.4089068825910931,
+      "grad_norm": 0.7483228117124135,
+      "learning_rate": 3.258064516129033e-05,
+      "loss": 0.6758,
+      "step": 101
+    },
+    {
+      "epoch": 0.41295546558704455,
+      "grad_norm": 0.8348023656376803,
+      "learning_rate": 3.2903225806451614e-05,
+      "loss": 0.6731,
+      "step": 102
+    },
+    {
+      "epoch": 0.41700404858299595,
+      "grad_norm": 1.0152186141805926,
+      "learning_rate": 3.322580645161291e-05,
+      "loss": 0.6544,
+      "step": 103
+    },
+    {
+      "epoch": 0.42105263157894735,
+      "grad_norm": 0.788066310285257,
+      "learning_rate": 3.3548387096774195e-05,
+      "loss": 0.6576,
+      "step": 104
+    },
+    {
+      "epoch": 0.4251012145748988,
+      "grad_norm": 0.6893613938601308,
+      "learning_rate": 3.387096774193549e-05,
+      "loss": 0.6517,
+      "step": 105
+    },
+    {
+      "epoch": 0.4291497975708502,
+      "grad_norm": 0.9347530171022542,
+      "learning_rate": 3.4193548387096776e-05,
+      "loss": 0.6524,
+      "step": 106
+    },
+    {
+      "epoch": 0.4331983805668016,
+      "grad_norm": 0.7802472064706533,
+      "learning_rate": 3.451612903225807e-05,
+      "loss": 0.657,
+      "step": 107
+    },
+    {
+      "epoch": 0.43724696356275305,
+      "grad_norm": 0.7732297073075312,
+      "learning_rate": 3.483870967741936e-05,
+      "loss": 0.671,
+      "step": 108
+    },
+    {
+      "epoch": 0.44129554655870445,
+      "grad_norm": 0.7333333459747566,
+      "learning_rate": 3.516129032258065e-05,
+      "loss": 0.6561,
+      "step": 109
+    },
+    {
+      "epoch": 0.44534412955465585,
+      "grad_norm": 0.8884030311264579,
+      "learning_rate": 3.548387096774194e-05,
+      "loss": 0.6352,
+      "step": 110
+    },
+    {
+      "epoch": 0.4493927125506073,
+      "grad_norm": 0.923667507713691,
+      "learning_rate": 3.5806451612903225e-05,
+      "loss": 0.6811,
+      "step": 111
+    },
+    {
+      "epoch": 0.4534412955465587,
+      "grad_norm": 0.8991357671031381,
+      "learning_rate": 3.612903225806452e-05,
+      "loss": 0.6663,
+      "step": 112
+    },
+    {
+      "epoch": 0.4574898785425101,
+      "grad_norm": 0.9846341811619315,
+      "learning_rate": 3.6451612903225805e-05,
+      "loss": 0.6712,
+      "step": 113
+    },
+    {
+      "epoch": 0.46153846153846156,
+      "grad_norm": 0.8998910702482752,
+      "learning_rate": 3.67741935483871e-05,
+      "loss": 0.6476,
+      "step": 114
+    },
+    {
+      "epoch": 0.46558704453441296,
+      "grad_norm": 0.9316005543018158,
+      "learning_rate": 3.7096774193548386e-05,
+      "loss": 0.6676,
+      "step": 115
+    },
+    {
+      "epoch": 0.46963562753036436,
+      "grad_norm": 1.0338215545592702,
+      "learning_rate": 3.741935483870968e-05,
+      "loss": 0.6803,
+      "step": 116
+    },
+    {
+      "epoch": 0.47368421052631576,
+      "grad_norm": 0.8428780220690855,
+      "learning_rate": 3.7741935483870974e-05,
+      "loss": 0.6791,
+      "step": 117
+    },
+    {
+      "epoch": 0.4777327935222672,
+      "grad_norm": 1.032352536072718,
+      "learning_rate": 3.806451612903226e-05,
+      "loss": 0.6654,
+      "step": 118
+    },
+    {
+      "epoch": 0.4817813765182186,
+      "grad_norm": 0.9301545559038729,
+      "learning_rate": 3.8387096774193555e-05,
+      "loss": 0.6624,
+      "step": 119
+    },
+    {
+      "epoch": 0.48582995951417,
+      "grad_norm": 0.9229497508461302,
+      "learning_rate": 3.870967741935484e-05,
+      "loss": 0.6482,
+      "step": 120
+    },
+    {
+      "epoch": 0.4898785425101215,
+      "grad_norm": 0.7401521310226457,
+      "learning_rate": 3.9032258064516135e-05,
+      "loss": 0.6402,
+      "step": 121
+    },
+    {
+      "epoch": 0.4939271255060729,
+      "grad_norm": 0.7474827000951283,
+      "learning_rate": 3.935483870967742e-05,
+      "loss": 0.6429,
+      "step": 122
+    },
+    {
+      "epoch": 0.4979757085020243,
+      "grad_norm": 0.962105107972727,
+      "learning_rate": 3.9677419354838716e-05,
+      "loss": 0.6647,
+      "step": 123
+    },
+    {
+      "epoch": 0.5020242914979757,
+      "grad_norm": 1.067669949234214,
+      "learning_rate": 4e-05,
+      "loss": 0.6484,
+      "step": 124
+    },
+    {
+      "epoch": 0.5060728744939271,
+      "grad_norm": 0.7711273641186039,
+      "learning_rate": 3.9999920040266474e-05,
+      "loss": 0.6509,
+      "step": 125
+    },
+    {
+      "epoch": 0.5101214574898786,
+      "grad_norm": 0.8812284913996881,
+      "learning_rate": 3.999968016170525e-05,
+      "loss": 0.6473,
+      "step": 126
+    },
+    {
+      "epoch": 0.5141700404858299,
+      "grad_norm": 0.8052580938841782,
+      "learning_rate": 3.9999280366234394e-05,
+      "loss": 0.6484,
+      "step": 127
+    },
+    {
+      "epoch": 0.5182186234817814,
+      "grad_norm": 0.6373651640218577,
+      "learning_rate": 3.999872065705065e-05,
+      "loss": 0.639,
+      "step": 128
+    },
+    {
+      "epoch": 0.5222672064777328,
+      "grad_norm": 0.7552731237821471,
+      "learning_rate": 3.999800103862945e-05,
+      "loss": 0.6509,
+      "step": 129
+    },
+    {
+      "epoch": 0.5263157894736842,
+      "grad_norm": 0.7305499745805287,
+      "learning_rate": 3.9997121516724834e-05,
+      "loss": 0.6417,
+      "step": 130
+    },
+    {
+      "epoch": 0.5303643724696356,
+      "grad_norm": 0.6533056108339448,
+      "learning_rate": 3.999608209836944e-05,
+      "loss": 0.6617,
+      "step": 131
+    },
+    {
+      "epoch": 0.5344129554655871,
+      "grad_norm": 0.994536310660137,
+      "learning_rate": 3.999488279187443e-05,
+      "loss": 0.6339,
+      "step": 132
+    },
+    {
+      "epoch": 0.5384615384615384,
+      "grad_norm": 0.7729701139421329,
+      "learning_rate": 3.9993523606829425e-05,
+      "loss": 0.6532,
+      "step": 133
+    },
+    {
+      "epoch": 0.5425101214574899,
+      "grad_norm": 0.7911811470145342,
+      "learning_rate": 3.9992004554102435e-05,
+      "loss": 0.6414,
+      "step": 134
+    },
+    {
+      "epoch": 0.5465587044534413,
+      "grad_norm": 0.9351087163538077,
+      "learning_rate": 3.999032564583976e-05,
+      "loss": 0.6759,
+      "step": 135
+    },
+    {
+      "epoch": 0.5506072874493927,
+      "grad_norm": 0.7935214342903969,
+      "learning_rate": 3.998848689546591e-05,
+      "loss": 0.6309,
+      "step": 136
+    },
+    {
+      "epoch": 0.5546558704453441,
+      "grad_norm": 0.9834331595060852,
+      "learning_rate": 3.9986488317683496e-05,
+      "loss": 0.6064,
+      "step": 137
+    },
+    {
+      "epoch": 0.5587044534412956,
+      "grad_norm": 0.6870771306214882,
+      "learning_rate": 3.9984329928473067e-05,
+      "loss": 0.6445,
+      "step": 138
+    },
+    {
+      "epoch": 0.562753036437247,
+      "grad_norm": 0.7223469168481459,
+      "learning_rate": 3.998201174509306e-05,
+      "loss": 0.654,
+      "step": 139
+    },
+    {
+      "epoch": 0.5668016194331984,
+      "grad_norm": 0.686521484999555,
+      "learning_rate": 3.997953378607961e-05,
+      "loss": 0.6359,
+      "step": 140
+    },
+    {
+      "epoch": 0.5708502024291497,
+      "grad_norm": 0.6829251707737491,
+      "learning_rate": 3.9976896071246403e-05,
+      "loss": 0.6337,
+      "step": 141
+    },
+    {
+      "epoch": 0.5748987854251012,
+      "grad_norm": 0.6159386621199995,
+      "learning_rate": 3.997409862168454e-05,
+      "loss": 0.6636,
+      "step": 142
+    },
+    {
+      "epoch": 0.5789473684210527,
+      "grad_norm": 0.6757845242298995,
+      "learning_rate": 3.997114145976237e-05,
+      "loss": 0.6103,
+      "step": 143
+    },
+    {
+      "epoch": 0.582995951417004,
+      "grad_norm": 0.5887339193743253,
+      "learning_rate": 3.9968024609125245e-05,
+      "loss": 0.6364,
+      "step": 144
+    },
+    {
+      "epoch": 0.5870445344129555,
+      "grad_norm": 0.7750174217497999,
+      "learning_rate": 3.996474809469545e-05,
+      "loss": 0.6386,
+      "step": 145
+    },
+    {
+      "epoch": 0.5910931174089069,
+      "grad_norm": 0.7261766601526581,
+      "learning_rate": 3.996131194267189e-05,
+      "loss": 0.6343,
+      "step": 146
+    },
+    {
+      "epoch": 0.5951417004048583,
+      "grad_norm": 0.7959872031978891,
+      "learning_rate": 3.995771618052995e-05,
+      "loss": 0.6755,
+      "step": 147
+    },
+    {
+      "epoch": 0.5991902834008097,
+      "grad_norm": 0.7237182634340344,
+      "learning_rate": 3.9953960837021256e-05,
+      "loss": 0.6469,
+      "step": 148
+    },
+    {
+      "epoch": 0.6032388663967612,
+      "grad_norm": 0.8023710059241309,
+      "learning_rate": 3.995004594217343e-05,
+      "loss": 0.6525,
+      "step": 149
+    },
+    {
+      "epoch": 0.6072874493927125,
+      "grad_norm": 0.656947294797479,
+      "learning_rate": 3.994597152728987e-05,
+      "loss": 0.6564,
+      "step": 150
+    },
+    {
+      "epoch": 0.611336032388664,
+      "grad_norm": 0.7447878859977948,
+      "learning_rate": 3.994173762494947e-05,
+      "loss": 0.6267,
+      "step": 151
+    },
+    {
+      "epoch": 0.6153846153846154,
+      "grad_norm": 0.6150021707816908,
+      "learning_rate": 3.9937344269006424e-05,
+      "loss": 0.6354,
+      "step": 152
+    },
+    {
+      "epoch": 0.6194331983805668,
+      "grad_norm": 0.7633003570778907,
+      "learning_rate": 3.993279149458988e-05,
+      "loss": 0.6367,
+      "step": 153
+    },
+    {
+      "epoch": 0.6234817813765182,
+      "grad_norm": 0.6935539265317298,
+      "learning_rate": 3.9928079338103695e-05,
+      "loss": 0.6146,
+      "step": 154
+    },
+    {
+      "epoch": 0.6275303643724697,
+      "grad_norm": 0.7309323897714641,
+      "learning_rate": 3.992320783722616e-05,
+      "loss": 0.6202,
+      "step": 155
+    },
+    {
+      "epoch": 0.631578947368421,
+      "grad_norm": 0.6747337767738942,
+      "learning_rate": 3.991817703090965e-05,
+      "loss": 0.6422,
+      "step": 156
+    },
+    {
+      "epoch": 0.6356275303643725,
+      "grad_norm": 0.6069881485059311,
+      "learning_rate": 3.991298695938038e-05,
+      "loss": 0.6224,
+      "step": 157
+    },
+    {
+      "epoch": 0.6396761133603239,
+      "grad_norm": 0.6073651742371702,
+      "learning_rate": 3.9907637664138e-05,
+      "loss": 0.6392,
+      "step": 158
+    },
+    {
+      "epoch": 0.6437246963562753,
+      "grad_norm": 0.5962646189956513,
+      "learning_rate": 3.990212918795534e-05,
+      "loss": 0.6196,
+      "step": 159
+    },
+    {
+      "epoch": 0.6477732793522267,
+      "grad_norm": 0.5927546710110386,
+      "learning_rate": 3.9896461574878036e-05,
+      "loss": 0.6443,
+      "step": 160
+    },
+    {
+      "epoch": 0.6518218623481782,
+      "grad_norm": 0.638158213337266,
+      "learning_rate": 3.989063487022416e-05,
+      "loss": 0.6545,
+      "step": 161
+    },
+    {
+      "epoch": 0.6558704453441295,
+      "grad_norm": 0.5697670409377467,
+      "learning_rate": 3.9884649120583903e-05,
+      "loss": 0.6326,
+      "step": 162
+    },
+    {
+      "epoch": 0.659919028340081,
+      "grad_norm": 0.5677113211204636,
+      "learning_rate": 3.987850437381915e-05,
+      "loss": 0.647,
+      "step": 163
+    },
+    {
+      "epoch": 0.6639676113360324,
+      "grad_norm": 0.6198045492278332,
+      "learning_rate": 3.987220067906313e-05,
+      "loss": 0.6415,
+      "step": 164
+    },
+    {
+      "epoch": 0.6680161943319838,
+      "grad_norm": 0.7707345863999904,
+      "learning_rate": 3.986573808672002e-05,
+      "loss": 0.654,
+      "step": 165
+    },
+    {
+      "epoch": 0.6720647773279352,
+      "grad_norm": 0.5892667754636222,
+      "learning_rate": 3.9859116648464534e-05,
+      "loss": 0.6504,
+      "step": 166
+    },
+    {
+      "epoch": 0.6761133603238867,
+      "grad_norm": 0.6765276756500118,
+      "learning_rate": 3.985233641724152e-05,
+      "loss": 0.6381,
+      "step": 167
+    },
+    {
+      "epoch": 0.680161943319838,
+      "grad_norm": 0.5801283967566021,
+      "learning_rate": 3.984539744726554e-05,
+      "loss": 0.613,
+      "step": 168
+    },
+    {
+      "epoch": 0.6842105263157895,
+      "grad_norm": 0.6579438276222209,
+      "learning_rate": 3.983829979402038e-05,
+      "loss": 0.6207,
+      "step": 169
+    },
+    {
+      "epoch": 0.6882591093117408,
+      "grad_norm": 0.6932752795470885,
+      "learning_rate": 3.983104351425871e-05,
+      "loss": 0.6248,
+      "step": 170
+    },
+    {
+      "epoch": 0.6923076923076923,
+      "grad_norm": 0.6022867265851257,
+      "learning_rate": 3.982362866600154e-05,
+      "loss": 0.6373,
+      "step": 171
+    },
+    {
+      "epoch": 0.6963562753036437,
+      "grad_norm": 0.5984918882121035,
+      "learning_rate": 3.9816055308537806e-05,
+      "loss": 0.5986,
+      "step": 172
+    },
+    {
+      "epoch": 0.7004048582995951,
+      "grad_norm": 0.7226818111655777,
+      "learning_rate": 3.980832350242388e-05,
+      "loss": 0.6533,
+      "step": 173
+    },
+    {
+      "epoch": 0.7044534412955465,
+      "grad_norm": 0.6082950097846145,
+      "learning_rate": 3.980043330948306e-05,
+      "loss": 0.6212,
+      "step": 174
+    },
+    {
+      "epoch": 0.708502024291498,
+      "grad_norm": 0.7201052213714393,
+      "learning_rate": 3.9792384792805124e-05,
+      "loss": 0.6235,
+      "step": 175
+    },
+    {
+      "epoch": 0.7125506072874493,
+      "grad_norm": 0.6319600789825495,
+      "learning_rate": 3.978417801674579e-05,
+      "loss": 0.6215,
+      "step": 176
+    },
+    {
+      "epoch": 0.7165991902834008,
+      "grad_norm": 0.8057784503768127,
+      "learning_rate": 3.9775813046926244e-05,
+      "loss": 0.6499,
+      "step": 177
+    },
+    {
+      "epoch": 0.7206477732793523,
+      "grad_norm": 0.5827405776463184,
+      "learning_rate": 3.976728995023254e-05,
+      "loss": 0.6308,
+      "step": 178
+    },
+    {
+      "epoch": 0.7246963562753036,
+      "grad_norm": 0.7843677010311505,
+      "learning_rate": 3.975860879481514e-05,
+      "loss": 0.6383,
+      "step": 179
+    },
+    {
+      "epoch": 0.728744939271255,
+      "grad_norm": 0.6695973348481961,
+      "learning_rate": 3.974976965008833e-05,
+      "loss": 0.6307,
+      "step": 180
+    },
+    {
+      "epoch": 0.7327935222672065,
+      "grad_norm": 0.7252408974191159,
+      "learning_rate": 3.9740772586729674e-05,
+      "loss": 0.6304,
+      "step": 181
+    },
+    {
+      "epoch": 0.7368421052631579,
+      "grad_norm": 0.6347638694992496,
+      "learning_rate": 3.973161767667946e-05,
+      "loss": 0.6172,
+      "step": 182
+    },
+    {
+      "epoch": 0.7408906882591093,
+      "grad_norm": 0.5473806363394464,
+      "learning_rate": 3.972230499314009e-05,
+      "loss": 0.6507,
+      "step": 183
+    },
+    {
+      "epoch": 0.7449392712550608,
+      "grad_norm": 0.6735447680293731,
+      "learning_rate": 3.9712834610575545e-05,
+      "loss": 0.636,
+      "step": 184
+    },
+    {
+      "epoch": 0.7489878542510121,
+      "grad_norm": 0.5682038384054041,
+      "learning_rate": 3.9703206604710746e-05,
+      "loss": 0.6384,
+      "step": 185
+    },
+    {
+      "epoch": 0.7530364372469636,
+      "grad_norm": 0.5769661730650769,
+      "learning_rate": 3.969342105253098e-05,
+      "loss": 0.6402,
+      "step": 186
+    },
+    {
+      "epoch": 0.757085020242915,
+      "grad_norm": 0.5735978279952927,
+      "learning_rate": 3.9683478032281246e-05,
+      "loss": 0.6351,
+      "step": 187
+    },
+    {
+      "epoch": 0.7611336032388664,
+      "grad_norm": 0.6899209057419836,
+      "learning_rate": 3.967337762346568e-05,
+      "loss": 0.6368,
+      "step": 188
+    },
+    {
+      "epoch": 0.7651821862348178,
+      "grad_norm": 0.5826438364653346,
+      "learning_rate": 3.9663119906846885e-05,
+      "loss": 0.6436,
+      "step": 189
+    },
+    {
+      "epoch": 0.7692307692307693,
+      "grad_norm": 0.6458694516053831,
+      "learning_rate": 3.965270496444528e-05,
+      "loss": 0.6035,
+      "step": 190
+    },
+    {
+      "epoch": 0.7732793522267206,
+      "grad_norm": 0.65964625060409,
+      "learning_rate": 3.9642132879538465e-05,
+      "loss": 0.6226,
+      "step": 191
+    },
+    {
+      "epoch": 0.7773279352226721,
+      "grad_norm": 0.5461484607623821,
+      "learning_rate": 3.9631403736660566e-05,
+      "loss": 0.6069,
+      "step": 192
+    },
+    {
+      "epoch": 0.7813765182186235,
+      "grad_norm": 0.5864995445857661,
+      "learning_rate": 3.962051762160151e-05,
+      "loss": 0.6233,
+      "step": 193
+    },
+    {
+      "epoch": 0.7854251012145749,
+      "grad_norm": 0.565226954141749,
+      "learning_rate": 3.960947462140638e-05,
+      "loss": 0.6214,
+      "step": 194
+    },
+    {
+      "epoch": 0.7894736842105263,
+      "grad_norm": 0.6713961795792279,
+      "learning_rate": 3.959827482437472e-05,
+      "loss": 0.6492,
+      "step": 195
+    },
+    {
+      "epoch": 0.7935222672064778,
+      "grad_norm": 0.57817661558636,
+      "learning_rate": 3.958691832005981e-05,
+      "loss": 0.6373,
+      "step": 196
+    },
+    {
+      "epoch": 0.7975708502024291,
+      "grad_norm": 0.8296811384033103,
+      "learning_rate": 3.9575405199267944e-05,
+      "loss": 0.6304,
+      "step": 197
+    },
+    {
+      "epoch": 0.8016194331983806,
+      "grad_norm": 0.7228663736459721,
+      "learning_rate": 3.956373555405774e-05,
+      "loss": 0.5969,
+      "step": 198
+    },
+    {
+      "epoch": 0.805668016194332,
+      "grad_norm": 0.5705944851388396,
+      "learning_rate": 3.955190947773937e-05,
+      "loss": 0.6002,
+      "step": 199
+    },
+    {
+      "epoch": 0.8097165991902834,
+      "grad_norm": 0.6573022008852134,
+      "learning_rate": 3.9539927064873824e-05,
+      "loss": 0.642,
+      "step": 200
+    },
+    {
+      "epoch": 0.8137651821862348,
+      "grad_norm": 0.5003291883370451,
+      "learning_rate": 3.952778841127214e-05,
+      "loss": 0.6297,
+      "step": 201
+    },
+    {
+      "epoch": 0.8178137651821862,
+      "grad_norm": 0.6529576609316593,
+      "learning_rate": 3.95154936139947e-05,
+      "loss": 0.6254,
+      "step": 202
+    },
+    {
+      "epoch": 0.8218623481781376,
+      "grad_norm": 0.6022522868880364,
+      "learning_rate": 3.950304277135034e-05,
+      "loss": 0.6219,
+      "step": 203
+    },
+    {
+      "epoch": 0.8259109311740891,
+      "grad_norm": 0.5490555357375112,
+      "learning_rate": 3.94904359828957e-05,
+      "loss": 0.5874,
+      "step": 204
+    },
+    {
+      "epoch": 0.8299595141700404,
+      "grad_norm": 0.6695281152743667,
+      "learning_rate": 3.94776733494343e-05,
+      "loss": 0.5969,
+      "step": 205
+    },
+    {
+      "epoch": 0.8340080971659919,
+      "grad_norm": 0.5111005225179026,
+      "learning_rate": 3.9464754973015826e-05,
+      "loss": 0.6189,
+      "step": 206
+    },
+    {
+      "epoch": 0.8380566801619433,
+      "grad_norm": 0.5999801629772696,
+      "learning_rate": 3.9451680956935275e-05,
+      "loss": 0.6241,
+      "step": 207
+    },
+    {
+      "epoch": 0.8421052631578947,
+      "grad_norm": 0.6032337806972214,
+      "learning_rate": 3.943845140573213e-05,
+      "loss": 0.5935,
+      "step": 208
+    },
+    {
+      "epoch": 0.8461538461538461,
+      "grad_norm": 0.5148407149432419,
+      "learning_rate": 3.942506642518952e-05,
+      "loss": 0.6233,
+      "step": 209
+    },
+    {
+      "epoch": 0.8502024291497976,
+      "grad_norm": 0.6521962378957924,
+      "learning_rate": 3.9411526122333396e-05,
+      "loss": 0.6275,
+      "step": 210
+    },
+    {
+      "epoch": 0.854251012145749,
+      "grad_norm": 0.5154863329964349,
+      "learning_rate": 3.939783060543167e-05,
+      "loss": 0.612,
+      "step": 211
+    },
+    {
+      "epoch": 0.8582995951417004,
+      "grad_norm": 0.5358169786582904,
+      "learning_rate": 3.938397998399333e-05,
+      "loss": 0.6131,
+      "step": 212
+    },
+    {
+      "epoch": 0.8623481781376519,
+      "grad_norm": 0.5983017170289388,
+      "learning_rate": 3.936997436876756e-05,
+      "loss": 0.6086,
+      "step": 213
+    },
+    {
+      "epoch": 0.8663967611336032,
+      "grad_norm": 0.6597410216463754,
+      "learning_rate": 3.935581387174289e-05,
+      "loss": 0.6066,
+      "step": 214
+    },
+    {
+      "epoch": 0.8704453441295547,
+      "grad_norm": 0.6207711111751771,
+      "learning_rate": 3.9341498606146295e-05,
+      "loss": 0.6063,
+      "step": 215
+    },
+    {
+      "epoch": 0.8744939271255061,
+      "grad_norm": 0.6198368697669184,
+      "learning_rate": 3.932702868644224e-05,
+      "loss": 0.6031,
+      "step": 216
+    },
+    {
+      "epoch": 0.8785425101214575,
+      "grad_norm": 0.5440200057870043,
+      "learning_rate": 3.931240422833181e-05,
+      "loss": 0.6157,
+      "step": 217
+    },
+    {
+      "epoch": 0.8825910931174089,
+      "grad_norm": 0.5366175263681967,
+      "learning_rate": 3.9297625348751806e-05,
+      "loss": 0.6006,
+      "step": 218
+    },
+    {
+      "epoch": 0.8866396761133604,
+      "grad_norm": 0.4874724544075323,
+      "learning_rate": 3.9282692165873744e-05,
+      "loss": 0.5858,
+      "step": 219
+    },
+    {
+      "epoch": 0.8906882591093117,
+      "grad_norm": 0.5961362543623738,
+      "learning_rate": 3.9267604799102945e-05,
+      "loss": 0.5914,
+      "step": 220
+    },
+    {
+      "epoch": 0.8947368421052632,
+      "grad_norm": 0.5154347862579468,
+      "learning_rate": 3.9252363369077615e-05,
+      "loss": 0.6203,
+      "step": 221
+    },
+    {
+      "epoch": 0.8987854251012146,
+      "grad_norm": 0.6182867947680947,
+      "learning_rate": 3.9236967997667796e-05,
+      "loss": 0.61,
+      "step": 222
+    },
+    {
+      "epoch": 0.902834008097166,
+      "grad_norm": 0.5710292548279198,
+      "learning_rate": 3.9221418807974496e-05,
+      "loss": 0.6004,
+      "step": 223
+    },
+    {
+      "epoch": 0.9068825910931174,
+      "grad_norm": 0.5988323836046103,
+      "learning_rate": 3.92057159243286e-05,
+      "loss": 0.6024,
+      "step": 224
+    },
+    {
+      "epoch": 0.9109311740890689,
+      "grad_norm": 0.6496015582738439,
+      "learning_rate": 3.9189859472289956e-05,
+      "loss": 0.6174,
+      "step": 225
+    },
+    {
+      "epoch": 0.9149797570850202,
+      "grad_norm": 0.6411616303559133,
+      "learning_rate": 3.9173849578646324e-05,
+      "loss": 0.6115,
+      "step": 226
+    },
+    {
+      "epoch": 0.9190283400809717,
+      "grad_norm": 0.5557608368572162,
+      "learning_rate": 3.9157686371412395e-05,
+      "loss": 0.6099,
+      "step": 227
+    },
+    {
+      "epoch": 0.9230769230769231,
+      "grad_norm": 0.5870385284654848,
+      "learning_rate": 3.914136997982874e-05,
+      "loss": 0.6033,
+      "step": 228
+    },
+    {
+      "epoch": 0.9271255060728745,
+      "grad_norm": 0.4902175870469723,
+      "learning_rate": 3.912490053436079e-05,
+      "loss": 0.623,
+      "step": 229
+    },
+    {
+      "epoch": 0.9311740890688259,
+      "grad_norm": 0.5681804350536758,
+      "learning_rate": 3.9108278166697805e-05,
+      "loss": 0.6142,
+      "step": 230
+    },
+    {
+      "epoch": 0.9352226720647774,
+      "grad_norm": 0.5046255068672658,
+      "learning_rate": 3.909150300975178e-05,
+      "loss": 0.5937,
+      "step": 231
+    },
+    {
+      "epoch": 0.9392712550607287,
+      "grad_norm": 0.6180483940267972,
+      "learning_rate": 3.907457519765642e-05,
+      "loss": 0.5876,
+      "step": 232
+    },
+    {
+      "epoch": 0.9433198380566802,
+      "grad_norm": 0.5221832355060458,
+      "learning_rate": 3.905749486576607e-05,
+      "loss": 0.6076,
+      "step": 233
+    },
+    {
+      "epoch": 0.9473684210526315,
+      "grad_norm": 0.5294969022248267,
+      "learning_rate": 3.90402621506546e-05,
+      "loss": 0.6144,
+      "step": 234
+    },
+    {
+      "epoch": 0.951417004048583,
+      "grad_norm": 0.634153326266318,
+      "learning_rate": 3.902287719011435e-05,
+      "loss": 0.6163,
+      "step": 235
+    },
+    {
+      "epoch": 0.9554655870445344,
+      "grad_norm": 0.5395325205840868,
+      "learning_rate": 3.900534012315499e-05,
+      "loss": 0.5976,
+      "step": 236
+    },
+    {
+      "epoch": 0.9595141700404858,
+      "grad_norm": 0.553994176157336,
+      "learning_rate": 3.8987651090002454e-05,
+      "loss": 0.6147,
+      "step": 237
+    },
+    {
+      "epoch": 0.9635627530364372,
+      "grad_norm": 0.5327789333326687,
+      "learning_rate": 3.896981023209777e-05,
+      "loss": 0.6082,
+      "step": 238
+    },
+    {
+      "epoch": 0.9676113360323887,
+      "grad_norm": 0.46029322203905293,
+      "learning_rate": 3.8951817692095966e-05,
+      "loss": 0.5856,
+      "step": 239
+    },
+    {
+      "epoch": 0.97165991902834,
+      "grad_norm": 0.5330226181916579,
+      "learning_rate": 3.893367361386491e-05,
+      "loss": 0.6034,
+      "step": 240
+    },
+    {
+      "epoch": 0.9757085020242915,
+      "grad_norm": 0.519659054376207,
+      "learning_rate": 3.891537814248417e-05,
+      "loss": 0.6298,
+      "step": 241
+    },
+    {
+      "epoch": 0.979757085020243,
+      "grad_norm": 0.5803369158449125,
+      "learning_rate": 3.889693142424384e-05,
+      "loss": 0.6051,
+      "step": 242
+    },
+    {
+      "epoch": 0.9838056680161943,
+      "grad_norm": 0.5156063892815116,
+      "learning_rate": 3.8878333606643405e-05,
+      "loss": 0.605,
+      "step": 243
+    },
+    {
+      "epoch": 0.9878542510121457,
+      "grad_norm": 0.5428382301968697,
+      "learning_rate": 3.885958483839049e-05,
+      "loss": 0.6335,
+      "step": 244
+    },
+    {
+      "epoch": 0.9919028340080972,
+      "grad_norm": 0.5555406819217314,
+      "learning_rate": 3.8840685269399786e-05,
+      "loss": 0.6038,
+      "step": 245
+    },
+    {
+      "epoch": 0.9959514170040485,
+      "grad_norm": 0.5093466689115276,
+      "learning_rate": 3.882163505079171e-05,
+      "loss": 0.6055,
+      "step": 246
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.5239854798221179,
+      "learning_rate": 3.880243433489132e-05,
+      "loss": 0.6132,
+      "step": 247
+    },
+    {
+      "epoch": 1.0040485829959513,
+      "grad_norm": 0.7013728413894895,
+      "learning_rate": 3.878308327522702e-05,
+      "loss": 0.5192,
+      "step": 248
+    },
+    {
+      "epoch": 1.008097165991903,
+      "grad_norm": 0.5453148120271949,
+      "learning_rate": 3.876358202652936e-05,
+      "loss": 0.5109,
+      "step": 249
+    },
+    {
+      "epoch": 1.0121457489878543,
+      "grad_norm": 0.8651476410536194,
+      "learning_rate": 3.8743930744729835e-05,
+      "loss": 0.4954,
+      "step": 250
+    },
+    {
+      "epoch": 1.0161943319838056,
+      "grad_norm": 0.5437715293697754,
+      "learning_rate": 3.8724129586959535e-05,
+      "loss": 0.5008,
+      "step": 251
+    },
+    {
+      "epoch": 1.0202429149797572,
+      "grad_norm": 0.6983225848796252,
+      "learning_rate": 3.870417871154801e-05,
+      "loss": 0.4745,
+      "step": 252
+    },
+    {
+      "epoch": 1.0242914979757085,
+      "grad_norm": 0.5037561529617816,
+      "learning_rate": 3.868407827802192e-05,
+      "loss": 0.4663,
+      "step": 253
+    },
+    {
+      "epoch": 1.0283400809716599,
+      "grad_norm": 0.6295588174983189,
+      "learning_rate": 3.8663828447103804e-05,
+      "loss": 0.5042,
+      "step": 254
+    },
+    {
+      "epoch": 1.0323886639676114,
+      "grad_norm": 0.5584629127940681,
+      "learning_rate": 3.8643429380710765e-05,
+      "loss": 0.5215,
+      "step": 255
+    },
+    {
+      "epoch": 1.0364372469635628,
+      "grad_norm": 0.5425123485492565,
+      "learning_rate": 3.862288124195319e-05,
+      "loss": 0.5011,
+      "step": 256
+    },
+    {
+      "epoch": 1.040485829959514,
+      "grad_norm": 0.6024173703246088,
+      "learning_rate": 3.8602184195133464e-05,
+      "loss": 0.497,
+      "step": 257
+    },
+    {
+      "epoch": 1.0445344129554657,
+      "grad_norm": 0.554499662161646,
+      "learning_rate": 3.8581338405744604e-05,
+      "loss": 0.466,
+      "step": 258
+    },
+    {
+      "epoch": 1.048582995951417,
+      "grad_norm": 0.6359381906858471,
+      "learning_rate": 3.8560344040468996e-05,
+      "loss": 0.4898,
+      "step": 259
+    },
+    {
+      "epoch": 1.0526315789473684,
+      "grad_norm": 0.6193406831777888,
+      "learning_rate": 3.853920126717703e-05,
+      "loss": 0.492,
+      "step": 260
+    },
+    {
+      "epoch": 1.05668016194332,
+      "grad_norm": 0.5394567583834976,
+      "learning_rate": 3.851791025492574e-05,
+      "loss": 0.5006,
+      "step": 261
+    },
+    {
+      "epoch": 1.0607287449392713,
+      "grad_norm": 0.7584013978531776,
+      "learning_rate": 3.8496471173957515e-05,
+      "loss": 0.4968,
+      "step": 262
+    },
+    {
+      "epoch": 1.0647773279352226,
+      "grad_norm": 0.5840963917533393,
+      "learning_rate": 3.847488419569866e-05,
+      "loss": 0.4989,
+      "step": 263
+    },
+    {
+      "epoch": 1.0688259109311742,
+      "grad_norm": 0.6718956658324695,
+      "learning_rate": 3.8453149492758077e-05,
+      "loss": 0.4903,
+      "step": 264
+    },
+    {
+      "epoch": 1.0728744939271255,
+      "grad_norm": 0.6452480192432732,
+      "learning_rate": 3.8431267238925886e-05,
+      "loss": 0.4954,
+      "step": 265
+    },
+    {
+      "epoch": 1.0769230769230769,
+      "grad_norm": 0.5400584909490936,
+      "learning_rate": 3.840923760917199e-05,
+      "loss": 0.4946,
+      "step": 266
+    },
+    {
+      "epoch": 1.0809716599190284,
+      "grad_norm": 0.5121178745517256,
+      "learning_rate": 3.838706077964473e-05,
+      "loss": 0.4973,
+      "step": 267
+    },
+    {
+      "epoch": 1.0850202429149798,
+      "grad_norm": 0.5241540612484918,
+      "learning_rate": 3.836473692766944e-05,
+      "loss": 0.4775,
+      "step": 268
+    },
+    {
+      "epoch": 1.0890688259109311,
+      "grad_norm": 0.5159226111575658,
+      "learning_rate": 3.834226623174705e-05,
+      "loss": 0.4986,
+      "step": 269
+    },
+    {
+      "epoch": 1.0931174089068827,
+      "grad_norm": 0.5617842366377159,
+      "learning_rate": 3.831964887155264e-05,
+      "loss": 0.4995,
+      "step": 270
+    },
+    {
+      "epoch": 1.097165991902834,
+      "grad_norm": 0.5334587337722138,
+      "learning_rate": 3.829688502793403e-05,
+      "loss": 0.4718,
+      "step": 271
+    },
+    {
+      "epoch": 1.1012145748987854,
+      "grad_norm": 0.5475941142569396,
+      "learning_rate": 3.827397488291029e-05,
+      "loss": 0.4961,
+      "step": 272
+    },
+    {
+      "epoch": 1.1052631578947367,
+      "grad_norm": 0.529076988906975,
+      "learning_rate": 3.8250918619670343e-05,
+      "loss": 0.4821,
+      "step": 273
+    },
+    {
+      "epoch": 1.1093117408906883,
+      "grad_norm": 0.5224096685649228,
+      "learning_rate": 3.822771642257145e-05,
+      "loss": 0.4845,
+      "step": 274
+    },
+    {
+      "epoch": 1.1133603238866396,
+      "grad_norm": 0.5124445406236834,
+      "learning_rate": 3.820436847713776e-05,
+      "loss": 0.487,
+      "step": 275
+    },
+    {
+      "epoch": 1.117408906882591,
+      "grad_norm": 0.5413689602890345,
+      "learning_rate": 3.8180874970058826e-05,
+      "loss": 0.4755,
+      "step": 276
+    },
+    {
+      "epoch": 1.1214574898785425,
+      "grad_norm": 0.5837102578815405,
+      "learning_rate": 3.8157236089188103e-05,
+      "loss": 0.4983,
+      "step": 277
+    },
+    {
+      "epoch": 1.125506072874494,
+      "grad_norm": 0.6222543722334347,
+      "learning_rate": 3.813345202354145e-05,
+      "loss": 0.4847,
+      "step": 278
+    },
+    {
+      "epoch": 1.1295546558704452,
+      "grad_norm": 0.527222207543397,
+      "learning_rate": 3.810952296329563e-05,
+      "loss": 0.5064,
+      "step": 279
+    },
+    {
+      "epoch": 1.1336032388663968,
+      "grad_norm": 0.6278787914835064,
+      "learning_rate": 3.808544909978676e-05,
+      "loss": 0.509,
+      "step": 280
+    },
+    {
+      "epoch": 1.1376518218623481,
+      "grad_norm": 0.4849659582576152,
+      "learning_rate": 3.806123062550882e-05,
+      "loss": 0.4905,
+      "step": 281
+    },
+    {
+      "epoch": 1.1417004048582995,
+      "grad_norm": 0.5271833804377688,
+      "learning_rate": 3.803686773411208e-05,
+      "loss": 0.4841,
+      "step": 282
+    },
+    {
+      "epoch": 1.145748987854251,
+      "grad_norm": 0.5743402172666859,
+      "learning_rate": 3.801236062040158e-05,
+      "loss": 0.497,
+      "step": 283
+    },
+    {
+      "epoch": 1.1497975708502024,
+      "grad_norm": 0.486337962837047,
+      "learning_rate": 3.798770948033553e-05,
+      "loss": 0.5151,
+      "step": 284
+    },
+    {
+      "epoch": 1.1538461538461537,
+      "grad_norm": 0.5771632795662621,
+      "learning_rate": 3.796291451102381e-05,
+      "loss": 0.496,
+      "step": 285
+    },
+    {
+      "epoch": 1.1578947368421053,
+      "grad_norm": 0.4980112146188978,
+      "learning_rate": 3.793797591072631e-05,
+      "loss": 0.466,
+      "step": 286
+    },
+    {
+      "epoch": 1.1619433198380567,
+      "grad_norm": 0.546788493378395,
+      "learning_rate": 3.791289387885144e-05,
+      "loss": 0.5128,
+      "step": 287
+    },
+    {
+      "epoch": 1.165991902834008,
+      "grad_norm": 0.5583158531191775,
+      "learning_rate": 3.788766861595443e-05,
+      "loss": 0.491,
+      "step": 288
+    },
+    {
+      "epoch": 1.1700404858299596,
+      "grad_norm": 0.49057131131162346,
+      "learning_rate": 3.7862300323735835e-05,
+      "loss": 0.5031,
+      "step": 289
+    },
+    {
+      "epoch": 1.174089068825911,
+      "grad_norm": 0.5454763691761577,
+      "learning_rate": 3.783678920503983e-05,
+      "loss": 0.4905,
+      "step": 290
+    },
+    {
+      "epoch": 1.1781376518218623,
+      "grad_norm": 0.5605302804690921,
+      "learning_rate": 3.781113546385265e-05,
+      "loss": 0.5009,
+      "step": 291
+    },
+    {
+      "epoch": 1.1821862348178138,
+      "grad_norm": 0.6470229202710353,
+      "learning_rate": 3.7785339305300916e-05,
+      "loss": 0.4942,
+      "step": 292
+    },
+    {
+      "epoch": 1.1862348178137652,
+      "grad_norm": 0.5462767529196909,
+      "learning_rate": 3.7759400935650025e-05,
+      "loss": 0.4883,
+      "step": 293
+    },
+    {
+      "epoch": 1.1902834008097165,
+      "grad_norm": 0.5509273372030735,
+      "learning_rate": 3.77333205623025e-05,
+      "loss": 0.5028,
+      "step": 294
+    },
+    {
+      "epoch": 1.194331983805668,
+      "grad_norm": 0.6059325905728433,
+      "learning_rate": 3.7707098393796295e-05,
+      "loss": 0.4885,
+      "step": 295
+    },
+    {
+      "epoch": 1.1983805668016194,
+      "grad_norm": 0.4941761930470436,
+      "learning_rate": 3.7680734639803185e-05,
+      "loss": 0.4862,
+      "step": 296
+    },
+    {
+      "epoch": 1.2024291497975708,
+      "grad_norm": 0.7069300851849303,
+      "learning_rate": 3.765422951112704e-05,
+      "loss": 0.5124,
+      "step": 297
+    },
+    {
+      "epoch": 1.2064777327935223,
+      "grad_norm": 0.6226916271456361,
+      "learning_rate": 3.762758321970216e-05,
+      "loss": 0.4738,
+      "step": 298
+    },
+    {
+      "epoch": 1.2105263157894737,
+      "grad_norm": 0.6053744541931155,
+      "learning_rate": 3.7600795978591584e-05,
+      "loss": 0.4657,
+      "step": 299
+    },
+    {
+      "epoch": 1.214574898785425,
+      "grad_norm": 0.687063984360041,
+      "learning_rate": 3.757386800198538e-05,
+      "loss": 0.4993,
+      "step": 300
+    },
+    {
+      "epoch": 1.2186234817813766,
+      "grad_norm": 0.5389174858351748,
+      "learning_rate": 3.7546799505198925e-05,
+      "loss": 0.4845,
+      "step": 301
+    },
+    {
+      "epoch": 1.222672064777328,
+      "grad_norm": 0.6773870950784634,
+      "learning_rate": 3.7519590704671197e-05,
+      "loss": 0.493,
+      "step": 302
+    },
+    {
+      "epoch": 1.2267206477732793,
+      "grad_norm": 0.5036105818426734,
+      "learning_rate": 3.749224181796305e-05,
+      "loss": 0.4871,
+      "step": 303
+    },
+    {
+      "epoch": 1.2307692307692308,
+      "grad_norm": 0.6775326858840116,
+      "learning_rate": 3.746475306375545e-05,
+      "loss": 0.4921,
+      "step": 304
+    },
+    {
+      "epoch": 1.2348178137651822,
+      "grad_norm": 0.5374480234955004,
+      "learning_rate": 3.743712466184774e-05,
+      "loss": 0.4642,
+      "step": 305
+    },
+    {
+      "epoch": 1.2388663967611335,
+      "grad_norm": 0.5261783830801381,
+      "learning_rate": 3.7409356833155885e-05,
+      "loss": 0.5126,
+      "step": 306
+    },
+    {
+      "epoch": 1.242914979757085,
+      "grad_norm": 0.5159949646631686,
+      "learning_rate": 3.73814497997107e-05,
+      "loss": 0.4939,
+      "step": 307
+    },
+    {
+      "epoch": 1.2469635627530364,
+      "grad_norm": 0.4650253177390461,
+      "learning_rate": 3.7353403784656095e-05,
+      "loss": 0.4856,
+      "step": 308
+    },
+    {
+      "epoch": 1.2510121457489878,
+      "grad_norm": 0.5532038167176707,
+      "learning_rate": 3.732521901224724e-05,
+      "loss": 0.5023,
+      "step": 309
+    },
+    {
+      "epoch": 1.2550607287449393,
+      "grad_norm": 0.5202579794896558,
+      "learning_rate": 3.7296895707848844e-05,
+      "loss": 0.5009,
+      "step": 310
+    },
+    {
+      "epoch": 1.2591093117408907,
+      "grad_norm": 0.5659152182750654,
+      "learning_rate": 3.7268434097933275e-05,
+      "loss": 0.498,
+      "step": 311
+    },
+    {
+      "epoch": 1.263157894736842,
+      "grad_norm": 0.5374323647348807,
+      "learning_rate": 3.723983441007882e-05,
+      "loss": 0.4856,
+      "step": 312
+    },
+    {
+      "epoch": 1.2672064777327936,
+      "grad_norm": 0.5148961270874854,
+      "learning_rate": 3.721109687296781e-05,
+      "loss": 0.5053,
+      "step": 313
+    },
+    {
+      "epoch": 1.271255060728745,
+      "grad_norm": 0.5568042109763733,
+      "learning_rate": 3.718222171638484e-05,
+      "loss": 0.5004,
+      "step": 314
+    },
+    {
+      "epoch": 1.2753036437246963,
+      "grad_norm": 0.5249543475143171,
+      "learning_rate": 3.715320917121488e-05,
+      "loss": 0.4813,
+      "step": 315
+    },
+    {
+      "epoch": 1.2793522267206479,
+      "grad_norm": 0.5655355669134362,
+      "learning_rate": 3.712405946944149e-05,
+      "loss": 0.5021,
+      "step": 316
+    },
+    {
+      "epoch": 1.2834008097165992,
+      "grad_norm": 0.5077246543269255,
+      "learning_rate": 3.709477284414488e-05,
+      "loss": 0.4933,
+      "step": 317
+    },
+    {
+      "epoch": 1.2874493927125505,
+      "grad_norm": 0.5437905848908219,
+      "learning_rate": 3.7065349529500136e-05,
+      "loss": 0.5003,
+      "step": 318
+    },
+    {
+      "epoch": 1.291497975708502,
+      "grad_norm": 0.5803496076467756,
+      "learning_rate": 3.7035789760775305e-05,
+      "loss": 0.5022,
+      "step": 319
+    },
+    {
+      "epoch": 1.2955465587044535,
+      "grad_norm": 0.5009993699976063,
+      "learning_rate": 3.70060937743295e-05,
+      "loss": 0.4969,
+      "step": 320
+    },
+    {
+      "epoch": 1.2995951417004048,
+      "grad_norm": 0.6019472990112059,
+      "learning_rate": 3.6976261807611045e-05,
+      "loss": 0.4762,
+      "step": 321
+    },
+    {
+      "epoch": 1.3036437246963564,
+      "grad_norm": 0.5706529053472534,
+      "learning_rate": 3.694629409915555e-05,
+      "loss": 0.5001,
+      "step": 322
+    },
+    {
+      "epoch": 1.3076923076923077,
+      "grad_norm": 0.6659342952409797,
+      "learning_rate": 3.691619088858401e-05,
+      "loss": 0.4786,
+      "step": 323
+    },
+    {
+      "epoch": 1.311740890688259,
+      "grad_norm": 0.5519714400185408,
+      "learning_rate": 3.68859524166009e-05,
+      "loss": 0.4772,
+      "step": 324
+    },
+    {
+      "epoch": 1.3157894736842106,
+      "grad_norm": 0.47912048035833044,
+      "learning_rate": 3.6855578924992226e-05,
+      "loss": 0.4744,
+      "step": 325
+    },
+    {
+      "epoch": 1.319838056680162,
+      "grad_norm": 0.5055905029482664,
+      "learning_rate": 3.6825070656623626e-05,
+      "loss": 0.488,
+      "step": 326
+    },
+    {
+      "epoch": 1.3238866396761133,
+      "grad_norm": 0.5218048443780346,
+      "learning_rate": 3.67944278554384e-05,
+      "loss": 0.4825,
+      "step": 327
+    },
+    {
+      "epoch": 1.3279352226720649,
+      "grad_norm": 0.5120352458574533,
+      "learning_rate": 3.676365076645557e-05,
+      "loss": 0.4917,
+      "step": 328
+    },
+    {
+      "epoch": 1.3319838056680162,
+      "grad_norm": 0.5531512347096023,
+      "learning_rate": 3.673273963576791e-05,
+      "loss": 0.4765,
+      "step": 329
+    },
+    {
+      "epoch": 1.3360323886639676,
+      "grad_norm": 0.5686596261395994,
+      "learning_rate": 3.670169471054001e-05,
+      "loss": 0.4993,
+      "step": 330
+    },
+    {
+      "epoch": 1.3400809716599191,
+      "grad_norm": 0.5971129228516225,
+      "learning_rate": 3.6670516239006254e-05,
+      "loss": 0.484,
+      "step": 331
+    },
+    {
+      "epoch": 1.3441295546558705,
+      "grad_norm": 0.5594994983287205,
+      "learning_rate": 3.663920447046888e-05,
+      "loss": 0.4913,
+      "step": 332
+    },
+    {
+      "epoch": 1.3481781376518218,
+      "grad_norm": 0.5789855349431501,
+      "learning_rate": 3.6607759655295954e-05,
+      "loss": 0.496,
+      "step": 333
+    },
+    {
+      "epoch": 1.3522267206477734,
+      "grad_norm": 0.6115834346106374,
+      "learning_rate": 3.657618204491937e-05,
+      "loss": 0.5036,
+      "step": 334
+    },
+    {
+      "epoch": 1.3562753036437247,
+      "grad_norm": 0.46490494824323503,
+      "learning_rate": 3.6544471891832875e-05,
+      "loss": 0.495,
+      "step": 335
+    },
+    {
+      "epoch": 1.360323886639676,
+      "grad_norm": 0.5867231353837556,
+      "learning_rate": 3.6512629449589983e-05,
+      "loss": 0.491,
+      "step": 336
+    },
+    {
+      "epoch": 1.3643724696356276,
+      "grad_norm": 0.5194849863927012,
+      "learning_rate": 3.6480654972802045e-05,
+      "loss": 0.4945,
+      "step": 337
+    },
+    {
+      "epoch": 1.368421052631579,
+      "grad_norm": 0.5086759302546262,
+      "learning_rate": 3.644854871713611e-05,
+      "loss": 0.502,
+      "step": 338
+    },
+    {
+      "epoch": 1.3724696356275303,
+      "grad_norm": 0.4763292881825884,
+      "learning_rate": 3.6416310939312924e-05,
+      "loss": 0.4931,
+      "step": 339
+    },
+    {
+      "epoch": 1.376518218623482,
+      "grad_norm": 0.49765250308986503,
+      "learning_rate": 3.638394189710493e-05,
+      "loss": 0.4722,
+      "step": 340
+    },
+    {
+      "epoch": 1.3805668016194332,
+      "grad_norm": 0.4477577955620521,
+      "learning_rate": 3.635144184933412e-05,
+      "loss": 0.4857,
+      "step": 341
+    },
+    {
+      "epoch": 1.3846153846153846,
+      "grad_norm": 0.5027211601456778,
+      "learning_rate": 3.6318811055869994e-05,
+      "loss": 0.4972,
+      "step": 342
+    },
+    {
+      "epoch": 1.3886639676113361,
+      "grad_norm": 0.48614885916247297,
+      "learning_rate": 3.628604977762752e-05,
+      "loss": 0.4987,
+      "step": 343
+    },
+    {
+      "epoch": 1.3927125506072875,
+      "grad_norm": 0.5209957576852885,
+      "learning_rate": 3.6253158276565004e-05,
+      "loss": 0.4956,
+      "step": 344
+    },
+    {
+      "epoch": 1.3967611336032388,
+      "grad_norm": 0.4962806316799539,
+      "learning_rate": 3.6220136815682016e-05,
+      "loss": 0.4675,
+      "step": 345
+    },
+    {
+      "epoch": 1.4008097165991904,
+      "grad_norm": 0.4568986263371971,
+      "learning_rate": 3.618698565901727e-05,
+      "loss": 0.48,
+      "step": 346
+    },
+    {
+      "epoch": 1.4048582995951417,
+      "grad_norm": 0.4590805992315336,
+      "learning_rate": 3.615370507164654e-05,
+      "loss": 0.5002,
+      "step": 347
+    },
+    {
+      "epoch": 1.408906882591093,
+      "grad_norm": 0.486396447652078,
+      "learning_rate": 3.61202953196805e-05,
+      "loss": 0.4784,
+      "step": 348
+    },
+    {
+      "epoch": 1.4129554655870447,
+      "grad_norm": 0.49065285494712924,
+      "learning_rate": 3.6086756670262654e-05,
+      "loss": 0.4852,
+      "step": 349
+    },
+    {
+      "epoch": 1.417004048582996,
+      "grad_norm": 0.47189029125217546,
+      "learning_rate": 3.6053089391567135e-05,
+      "loss": 0.4991,
+      "step": 350
+    },
+    {
+      "epoch": 1.4210526315789473,
+      "grad_norm": 0.4695791774966969,
+      "learning_rate": 3.601929375279662e-05,
+      "loss": 0.503,
+      "step": 351
+    },
+    {
+      "epoch": 1.425101214574899,
+      "grad_norm": 0.46036840588668343,
+      "learning_rate": 3.598537002418012e-05,
+      "loss": 0.4807,
+      "step": 352
+    },
+    {
+      "epoch": 1.4291497975708503,
+      "grad_norm": 0.42275708871364703,
+      "learning_rate": 3.5951318476970885e-05,
+      "loss": 0.5044,
+      "step": 353
+    },
+    {
+      "epoch": 1.4331983805668016,
+      "grad_norm": 0.4711649196394444,
+      "learning_rate": 3.591713938344416e-05,
+      "loss": 0.4683,
+      "step": 354
+    },
+    {
+      "epoch": 1.4372469635627532,
+      "grad_norm": 0.4768502393877982,
+      "learning_rate": 3.588283301689507e-05,
+      "loss": 0.4876,
+      "step": 355
+    },
+    {
+      "epoch": 1.4412955465587045,
+      "grad_norm": 0.4612653903388778,
+      "learning_rate": 3.5848399651636424e-05,
+      "loss": 0.5025,
+      "step": 356
+    },
+    {
+      "epoch": 1.4453441295546559,
+      "grad_norm": 0.4564327763014901,
+      "learning_rate": 3.5813839562996473e-05,
+      "loss": 0.4855,
+      "step": 357
+    },
+    {
+      "epoch": 1.4493927125506074,
+      "grad_norm": 0.5379483258532513,
+      "learning_rate": 3.5779153027316775e-05,
+      "loss": 0.5114,
+      "step": 358
+    },
+    {
+      "epoch": 1.4534412955465588,
+      "grad_norm": 0.4249247350761855,
+      "learning_rate": 3.574434032194994e-05,
+      "loss": 0.4933,
+      "step": 359
+    },
+    {
+      "epoch": 1.45748987854251,
+      "grad_norm": 0.4455686526471224,
+      "learning_rate": 3.5709401725257436e-05,
+      "loss": 0.5126,
+      "step": 360
+    },
+    {
+      "epoch": 1.4615384615384617,
+      "grad_norm": 0.45293806545356763,
+      "learning_rate": 3.5674337516607346e-05,
+      "loss": 0.4943,
+      "step": 361
+    },
+    {
+      "epoch": 1.465587044534413,
+      "grad_norm": 0.41593596354340634,
+      "learning_rate": 3.563914797637216e-05,
+      "loss": 0.4713,
+      "step": 362
+    },
+    {
+      "epoch": 1.4696356275303644,
+      "grad_norm": 0.482709463499064,
+      "learning_rate": 3.560383338592649e-05,
+      "loss": 0.4879,
+      "step": 363
+    },
+    {
+      "epoch": 1.4736842105263157,
+      "grad_norm": 0.40594879457716454,
+      "learning_rate": 3.556839402764487e-05,
+      "loss": 0.4997,
+      "step": 364
+    },
+    {
+      "epoch": 1.4777327935222673,
+      "grad_norm": 0.46411147238122924,
+      "learning_rate": 3.553283018489946e-05,
+      "loss": 0.4986,
+      "step": 365
+    },
+    {
+      "epoch": 1.4817813765182186,
+      "grad_norm": 0.46063260965820657,
+      "learning_rate": 3.54971421420578e-05,
+      "loss": 0.4818,
+      "step": 366
+    },
+    {
+      "epoch": 1.48582995951417,
+      "grad_norm": 0.51898432843912,
+      "learning_rate": 3.5461330184480524e-05,
+      "loss": 0.4786,
+      "step": 367
+    },
+    {
+      "epoch": 1.4898785425101215,
+      "grad_norm": 0.42178526422992013,
+      "learning_rate": 3.54253945985191e-05,
+      "loss": 0.4771,
+      "step": 368
+    },
+    {
+      "epoch": 1.4939271255060729,
+      "grad_norm": 0.5436254689187641,
+      "learning_rate": 3.5389335671513524e-05,
+      "loss": 0.5031,
+      "step": 369
+    },
+    {
+      "epoch": 1.4979757085020242,
+      "grad_norm": 0.48499806969174836,
+      "learning_rate": 3.5353153691789986e-05,
+      "loss": 0.5056,
+      "step": 370
+    },
+    {
+      "epoch": 1.5020242914979756,
+      "grad_norm": 0.5245756831135092,
+      "learning_rate": 3.531684894865866e-05,
+      "loss": 0.5095,
+      "step": 371
+    },
+    {
+      "epoch": 1.5060728744939271,
+      "grad_norm": 0.45619571678649934,
+      "learning_rate": 3.528042173241128e-05,
+      "loss": 0.5041,
+      "step": 372
+    },
+    {
+      "epoch": 1.5101214574898787,
+      "grad_norm": 0.4654779169082846,
+      "learning_rate": 3.524387233431892e-05,
+      "loss": 0.4822,
+      "step": 373
+    },
+    {
+      "epoch": 1.5141700404858298,
+      "grad_norm": 0.5477462985341813,
+      "learning_rate": 3.520720104662958e-05,
+      "loss": 0.4876,
+      "step": 374
+    },
+    {
+      "epoch": 1.5182186234817814,
+      "grad_norm": 0.4083798538259133,
+      "learning_rate": 3.5170408162565904e-05,
+      "loss": 0.493,
+      "step": 375
+    },
+    {
+      "epoch": 1.522267206477733,
+      "grad_norm": 0.534831364868229,
+      "learning_rate": 3.513349397632281e-05,
+      "loss": 0.4855,
+      "step": 376
+    },
+    {
+      "epoch": 1.526315789473684,
+      "grad_norm": 0.48774230657188333,
+      "learning_rate": 3.5096458783065145e-05,
+      "loss": 0.483,
+      "step": 377
+    },
+    {
+      "epoch": 1.5303643724696356,
+      "grad_norm": 0.44499588794471956,
+      "learning_rate": 3.505930287892533e-05,
+      "loss": 0.4903,
+      "step": 378
+    },
+    {
+      "epoch": 1.5344129554655872,
+      "grad_norm": 0.497531186091603,
+      "learning_rate": 3.502202656100099e-05,
+      "loss": 0.4819,
+      "step": 379
+    },
+    {
+      "epoch": 1.5384615384615383,
+      "grad_norm": 0.5029292567437619,
+      "learning_rate": 3.498463012735256e-05,
+      "loss": 0.5026,
+      "step": 380
+    },
+    {
+      "epoch": 1.54251012145749,
+      "grad_norm": 0.5009226391762714,
+      "learning_rate": 3.494711387700094e-05,
+      "loss": 0.4892,
+      "step": 381
+    },
+    {
+      "epoch": 1.5465587044534415,
+      "grad_norm": 0.5074852671171413,
+      "learning_rate": 3.490947810992505e-05,
+      "loss": 0.4771,
+      "step": 382
+    },
+    {
+      "epoch": 1.5506072874493926,
+      "grad_norm": 0.47562307712939306,
+      "learning_rate": 3.48717231270595e-05,
+      "loss": 0.4999,
+      "step": 383
+    },
+    {
+      "epoch": 1.5546558704453441,
+      "grad_norm": 0.5892494216084615,
+      "learning_rate": 3.483384923029211e-05,
+      "loss": 0.4901,
+      "step": 384
+    },
+    {
+      "epoch": 1.5587044534412957,
+      "grad_norm": 0.48582178898583583,
+      "learning_rate": 3.479585672246156e-05,
+      "loss": 0.5003,
+      "step": 385
+    },
+    {
+      "epoch": 1.5627530364372468,
+      "grad_norm": 0.5439131509716082,
+      "learning_rate": 3.4757745907354924e-05,
+      "loss": 0.4897,
+      "step": 386
+    },
+    {
+      "epoch": 1.5668016194331984,
+      "grad_norm": 0.4783748445617113,
+      "learning_rate": 3.471951708970527e-05,
+      "loss": 0.4866,
+      "step": 387
+    },
+    {
+      "epoch": 1.5708502024291497,
+      "grad_norm": 0.5634841112014572,
+      "learning_rate": 3.468117057518921e-05,
+      "loss": 0.4781,
+      "step": 388
+    },
+    {
+      "epoch": 1.574898785425101,
+      "grad_norm": 0.4980372782713765,
+      "learning_rate": 3.4642706670424426e-05,
+      "loss": 0.4752,
+      "step": 389
+    },
+    {
+      "epoch": 1.5789473684210527,
+      "grad_norm": 0.5513554010421116,
+      "learning_rate": 3.460412568296731e-05,
+      "loss": 0.4807,
+      "step": 390
+    },
+    {
+      "epoch": 1.582995951417004,
+      "grad_norm": 0.4714406507916131,
+      "learning_rate": 3.456542792131039e-05,
+      "loss": 0.5066,
+      "step": 391
+    },
+    {
+      "epoch": 1.5870445344129553,
+      "grad_norm": 0.5212856851761791,
+      "learning_rate": 3.4526613694879936e-05,
+      "loss": 0.482,
+      "step": 392
+    },
+    {
+      "epoch": 1.591093117408907,
+      "grad_norm": 0.522635575577373,
+      "learning_rate": 3.448768331403348e-05,
+      "loss": 0.5096,
+      "step": 393
+    },
+    {
+      "epoch": 1.5951417004048583,
+      "grad_norm": 0.49552442726800133,
+      "learning_rate": 3.44486370900573e-05,
+      "loss": 0.5129,
+      "step": 394
+    },
+    {
+      "epoch": 1.5991902834008096,
+      "grad_norm": 0.6068577675380971,
+      "learning_rate": 3.440947533516396e-05,
+      "loss": 0.4769,
+      "step": 395
+    },
+    {
+      "epoch": 1.6032388663967612,
+      "grad_norm": 0.5224307854395852,
+      "learning_rate": 3.4370198362489816e-05,
+      "loss": 0.4944,
+      "step": 396
+    },
+    {
+      "epoch": 1.6072874493927125,
+      "grad_norm": 0.5076173469744203,
+      "learning_rate": 3.43308064860925e-05,
+      "loss": 0.4918,
+      "step": 397
+    },
+    {
+      "epoch": 1.6113360323886639,
+      "grad_norm": 0.49795845920790205,
+      "learning_rate": 3.4291300020948393e-05,
+      "loss": 0.4728,
+      "step": 398
+    },
+    {
+      "epoch": 1.6153846153846154,
+      "grad_norm": 0.48803603000532614,
+      "learning_rate": 3.425167928295015e-05,
+      "loss": 0.4874,
+      "step": 399
+    },
+    {
+      "epoch": 1.6194331983805668,
+      "grad_norm": 0.4902755670368885,
+      "learning_rate": 3.421194458890411e-05,
+      "loss": 0.4978,
+      "step": 400
+    },
+    {
+      "epoch": 1.623481781376518,
+      "grad_norm": 0.5342492273811823,
+      "learning_rate": 3.4172096256527865e-05,
+      "loss": 0.5015,
+      "step": 401
+    },
+    {
+      "epoch": 1.6275303643724697,
+      "grad_norm": 0.4974068610412381,
+      "learning_rate": 3.4132134604447595e-05,
+      "loss": 0.4766,
+      "step": 402
+    },
+    {
+      "epoch": 1.631578947368421,
+      "grad_norm": 0.500226929060587,
+      "learning_rate": 3.409205995219561e-05,
+      "loss": 0.4959,
+      "step": 403
+    },
+    {
+      "epoch": 1.6356275303643724,
+      "grad_norm": 0.46830813211939304,
+      "learning_rate": 3.4051872620207765e-05,
+      "loss": 0.4852,
+      "step": 404
+    },
+    {
+      "epoch": 1.639676113360324,
+      "grad_norm": 0.5119678390015756,
+      "learning_rate": 3.40115729298209e-05,
+      "loss": 0.4894,
+      "step": 405
+    },
+    {
+      "epoch": 1.6437246963562753,
+      "grad_norm": 0.4957235108195215,
+      "learning_rate": 3.397116120327025e-05,
+      "loss": 0.5095,
+      "step": 406
+    },
+    {
+      "epoch": 1.6477732793522266,
+      "grad_norm": 0.44435915854641356,
+      "learning_rate": 3.3930637763686916e-05,
+      "loss": 0.4974,
+      "step": 407
+    },
+    {
+      "epoch": 1.6518218623481782,
+      "grad_norm": 0.4820296626000625,
+      "learning_rate": 3.389000293509524e-05,
+      "loss": 0.4876,
+      "step": 408
+    },
+    {
+      "epoch": 1.6558704453441295,
+      "grad_norm": 0.42129964786955704,
+      "learning_rate": 3.3849257042410225e-05,
+      "loss": 0.4847,
+      "step": 409
+    },
+    {
+      "epoch": 1.6599190283400809,
+      "grad_norm": 0.4690595147236224,
+      "learning_rate": 3.380840041143494e-05,
+      "loss": 0.4817,
+      "step": 410
+    },
+    {
+      "epoch": 1.6639676113360324,
+      "grad_norm": 0.44142926172409885,
+      "learning_rate": 3.376743336885792e-05,
+      "loss": 0.497,
+      "step": 411
+    },
+    {
+      "epoch": 1.6680161943319838,
+      "grad_norm": 0.45861944279299915,
+      "learning_rate": 3.372635624225055e-05,
+      "loss": 0.4968,
+      "step": 412
+    },
+    {
+      "epoch": 1.6720647773279351,
+      "grad_norm": 0.4123852649716994,
+      "learning_rate": 3.3685169360064445e-05,
+      "loss": 0.4765,
+      "step": 413
+    },
+    {
+      "epoch": 1.6761133603238867,
+      "grad_norm": 0.4910430894084535,
+      "learning_rate": 3.3643873051628804e-05,
+      "loss": 0.5036,
+      "step": 414
+    },
+    {
+      "epoch": 1.680161943319838,
+      "grad_norm": 0.43383835270525994,
+      "learning_rate": 3.3602467647147814e-05,
+      "loss": 0.5058,
+      "step": 415
+    },
+    {
+      "epoch": 1.6842105263157894,
+      "grad_norm": 0.4708366824566064,
+      "learning_rate": 3.356095347769799e-05,
+      "loss": 0.477,
+      "step": 416
+    },
+    {
+      "epoch": 1.688259109311741,
+      "grad_norm": 0.4600963208429319,
+      "learning_rate": 3.351933087522552e-05,
+      "loss": 0.4827,
+      "step": 417
+    },
+    {
+      "epoch": 1.6923076923076923,
+      "grad_norm": 0.4127868137390034,
+      "learning_rate": 3.3477600172543626e-05,
+      "loss": 0.4813,
+      "step": 418
+    },
+    {
+      "epoch": 1.6963562753036436,
+      "grad_norm": 0.490798815467858,
+      "learning_rate": 3.3435761703329894e-05,
+      "loss": 0.5121,
+      "step": 419
+    },
+    {
+      "epoch": 1.7004048582995952,
+      "grad_norm": 0.5075446337646792,
+      "learning_rate": 3.3393815802123606e-05,
+      "loss": 0.4856,
+      "step": 420
+    },
+    {
+      "epoch": 1.7044534412955465,
+      "grad_norm": 0.4512769056361883,
+      "learning_rate": 3.335176280432308e-05,
+      "loss": 0.4609,
+      "step": 421
+    },
+    {
+      "epoch": 1.708502024291498,
+      "grad_norm": 0.4957380153733052,
+      "learning_rate": 3.330960304618295e-05,
+      "loss": 0.4802,
+      "step": 422
+    },
+    {
+      "epoch": 1.7125506072874495,
+      "grad_norm": 0.4778336657366039,
+      "learning_rate": 3.326733686481153e-05,
+      "loss": 0.4868,
+      "step": 423
+    },
+    {
+      "epoch": 1.7165991902834008,
+      "grad_norm": 0.5030096605442838,
+      "learning_rate": 3.3224964598168084e-05,
+      "loss": 0.4813,
+      "step": 424
+    },
+    {
+      "epoch": 1.7206477732793521,
+      "grad_norm": 0.4910190488127221,
+      "learning_rate": 3.318248658506012e-05,
+      "loss": 0.499,
+      "step": 425
+    },
+    {
+      "epoch": 1.7246963562753037,
+      "grad_norm": 0.4714712007657032,
+      "learning_rate": 3.313990316514069e-05,
+      "loss": 0.5137,
+      "step": 426
+    },
+    {
+      "epoch": 1.728744939271255,
+      "grad_norm": 0.5130363945077953,
+      "learning_rate": 3.309721467890571e-05,
+      "loss": 0.4872,
+      "step": 427
+    },
+    {
+      "epoch": 1.7327935222672064,
+      "grad_norm": 0.46411876686128695,
+      "learning_rate": 3.305442146769115e-05,
+      "loss": 0.4846,
+      "step": 428
+    },
+    {
+      "epoch": 1.736842105263158,
+      "grad_norm": 0.43726889687805,
+      "learning_rate": 3.30115238736704e-05,
+      "loss": 0.4895,
+      "step": 429
+    },
+    {
+      "epoch": 1.7408906882591093,
+      "grad_norm": 0.5151735629059927,
+      "learning_rate": 3.296852223985148e-05,
+      "loss": 0.4855,
+      "step": 430
+    },
+    {
+      "epoch": 1.7449392712550607,
+      "grad_norm": 0.4221336538946466,
+      "learning_rate": 3.2925416910074315e-05,
+      "loss": 0.4727,
+      "step": 431
+    },
+    {
+      "epoch": 1.7489878542510122,
+      "grad_norm": 0.4544905773768445,
+      "learning_rate": 3.288220822900796e-05,
+      "loss": 0.4893,
+      "step": 432
+    },
+    {
+      "epoch": 1.7530364372469636,
+      "grad_norm": 0.49910729069217,
+      "learning_rate": 3.283889654214788e-05,
+      "loss": 0.5088,
+      "step": 433
+    },
+    {
+      "epoch": 1.757085020242915,
+      "grad_norm": 0.4645196534023941,
+      "learning_rate": 3.279548219581317e-05,
+      "loss": 0.5104,
+      "step": 434
+    },
+    {
+      "epoch": 1.7611336032388665,
+      "grad_norm": 0.48861363650495326,
+      "learning_rate": 3.275196553714379e-05,
+      "loss": 0.5003,
+      "step": 435
+    },
+    {
+      "epoch": 1.7651821862348178,
+      "grad_norm": 0.44636851833804303,
+      "learning_rate": 3.270834691409778e-05,
+      "loss": 0.5031,
+      "step": 436
+    },
+    {
+      "epoch": 1.7692307692307692,
+      "grad_norm": 0.46476284977904686,
+      "learning_rate": 3.266462667544849e-05,
+      "loss": 0.4941,
+      "step": 437
+    },
+    {
+      "epoch": 1.7732793522267207,
+      "grad_norm": 0.4249765928520606,
+      "learning_rate": 3.262080517078178e-05,
+      "loss": 0.5022,
+      "step": 438
+    },
+    {
+      "epoch": 1.777327935222672,
+      "grad_norm": 0.5606044449957487,
+      "learning_rate": 3.257688275049323e-05,
+      "loss": 0.4929,
+      "step": 439
+    },
+    {
+      "epoch": 1.7813765182186234,
+      "grad_norm": 0.4175719697367263,
+      "learning_rate": 3.253285976578535e-05,
+      "loss": 0.4674,
+      "step": 440
+    },
+    {
+      "epoch": 1.785425101214575,
+      "grad_norm": 0.5485038409452522,
+      "learning_rate": 3.2488736568664756e-05,
+      "loss": 0.466,
+      "step": 441
+    },
+    {
+      "epoch": 1.7894736842105263,
+      "grad_norm": 0.44449844374937153,
+      "learning_rate": 3.244451351193935e-05,
+      "loss": 0.496,
+      "step": 442
+    },
+    {
+      "epoch": 1.7935222672064777,
+      "grad_norm": 0.5268221511179716,
+      "learning_rate": 3.240019094921551e-05,
+      "loss": 0.4897,
+      "step": 443
+    },
+    {
+      "epoch": 1.7975708502024292,
+      "grad_norm": 0.400674289682926,
+      "learning_rate": 3.235576923489527e-05,
+      "loss": 0.4803,
+      "step": 444
+    },
+    {
+      "epoch": 1.8016194331983806,
+      "grad_norm": 0.4696743698246016,
+      "learning_rate": 3.2311248724173476e-05,
+      "loss": 0.4779,
+      "step": 445
+    },
+    {
+      "epoch": 1.805668016194332,
+      "grad_norm": 0.46007673844835656,
+      "learning_rate": 3.226662977303494e-05,
+      "loss": 0.4897,
+      "step": 446
+    },
+    {
+      "epoch": 1.8097165991902835,
+      "grad_norm": 0.4335161200936929,
+      "learning_rate": 3.2221912738251626e-05,
+      "loss": 0.4738,
+      "step": 447
+    },
+    {
+      "epoch": 1.8137651821862348,
+      "grad_norm": 0.45797645063843107,
+      "learning_rate": 3.2177097977379736e-05,
+      "loss": 0.499,
+      "step": 448
+    },
+    {
+      "epoch": 1.8178137651821862,
+      "grad_norm": 0.4325369539481943,
+      "learning_rate": 3.21321858487569e-05,
+      "loss": 0.5008,
+      "step": 449
+    },
+    {
+      "epoch": 1.8218623481781377,
+      "grad_norm": 0.4403932539797545,
+      "learning_rate": 3.208717671149931e-05,
+      "loss": 0.4844,
+      "step": 450
+    },
+    {
+      "epoch": 1.825910931174089,
+      "grad_norm": 0.4551291894750698,
+      "learning_rate": 3.204207092549882e-05,
+      "loss": 0.4792,
+      "step": 451
+    },
+    {
+      "epoch": 1.8299595141700404,
+      "grad_norm": 0.48410131421371344,
+      "learning_rate": 3.19968688514201e-05,
+      "loss": 0.4917,
+      "step": 452
+    },
+    {
+      "epoch": 1.834008097165992,
+      "grad_norm": 0.43013333323321534,
+      "learning_rate": 3.195157085069774e-05,
+      "loss": 0.4732,
+      "step": 453
+    },
+    {
+      "epoch": 1.8380566801619433,
+      "grad_norm": 0.44992652672619116,
+      "learning_rate": 3.190617728553332e-05,
+      "loss": 0.5171,
+      "step": 454
+    },
+    {
+      "epoch": 1.8421052631578947,
+      "grad_norm": 0.4639023047402663,
+      "learning_rate": 3.1860688518892606e-05,
+      "loss": 0.4496,
+      "step": 455
+    },
+    {
+      "epoch": 1.8461538461538463,
+      "grad_norm": 0.3993090117703164,
+      "learning_rate": 3.181510491450254e-05,
+      "loss": 0.4977,
+      "step": 456
+    },
+    {
+      "epoch": 1.8502024291497976,
+      "grad_norm": 0.4726774719349067,
+      "learning_rate": 3.176942683684842e-05,
+      "loss": 0.4636,
+      "step": 457
+    },
+    {
+      "epoch": 1.854251012145749,
+      "grad_norm": 0.44283811725777583,
+      "learning_rate": 3.1723654651170934e-05,
+      "loss": 0.4784,
+      "step": 458
+    },
+    {
+      "epoch": 1.8582995951417005,
+      "grad_norm": 0.45118900350836855,
+      "learning_rate": 3.167778872346327e-05,
+      "loss": 0.5049,
+      "step": 459
+    },
+    {
+      "epoch": 1.8623481781376519,
+      "grad_norm": 0.45327003740955474,
+      "learning_rate": 3.163182942046815e-05,
+      "loss": 0.4963,
+      "step": 460
+    },
+    {
+      "epoch": 1.8663967611336032,
+      "grad_norm": 0.48672455059230896,
+      "learning_rate": 3.158577710967494e-05,
+      "loss": 0.485,
+      "step": 461
+    },
+    {
+      "epoch": 1.8704453441295548,
+      "grad_norm": 0.4523298920998827,
+      "learning_rate": 3.153963215931669e-05,
+      "loss": 0.4697,
+      "step": 462
+    },
+    {
+      "epoch": 1.874493927125506,
+      "grad_norm": 0.4769227958082447,
+      "learning_rate": 3.14933949383672e-05,
+      "loss": 0.4948,
+      "step": 463
+    },
+    {
+      "epoch": 1.8785425101214575,
+      "grad_norm": 0.4758191837455256,
+      "learning_rate": 3.144706581653805e-05,
+      "loss": 0.4931,
+      "step": 464
+    },
+    {
+      "epoch": 1.882591093117409,
+      "grad_norm": 0.4632489274026781,
+      "learning_rate": 3.140064516427566e-05,
+      "loss": 0.5007,
+      "step": 465
+    },
+    {
+      "epoch": 1.8866396761133604,
+      "grad_norm": 0.44392541406598734,
+      "learning_rate": 3.135413335275833e-05,
+      "loss": 0.4846,
+      "step": 466
+    },
+    {
+      "epoch": 1.8906882591093117,
+      "grad_norm": 0.5038188864938486,
+      "learning_rate": 3.130753075389327e-05,
+      "loss": 0.4776,
+      "step": 467
+    },
+    {
+      "epoch": 1.8947368421052633,
+      "grad_norm": 0.45660021235852327,
+      "learning_rate": 3.1260837740313625e-05,
+      "loss": 0.4759,
+      "step": 468
+    },
+    {
+      "epoch": 1.8987854251012146,
+      "grad_norm": 0.44307403780574883,
+      "learning_rate": 3.121405468537547e-05,
+      "loss": 0.483,
+      "step": 469
+    },
+    {
+      "epoch": 1.902834008097166,
+      "grad_norm": 0.46401608565854857,
+      "learning_rate": 3.116718196315488e-05,
+      "loss": 0.476,
+      "step": 470
+    },
+    {
+      "epoch": 1.9068825910931175,
+      "grad_norm": 0.4636585909367666,
+      "learning_rate": 3.1120219948444894e-05,
+      "loss": 0.4945,
+      "step": 471
+    },
+    {
+      "epoch": 1.9109311740890689,
+      "grad_norm": 0.4507072220102074,
+      "learning_rate": 3.107316901675252e-05,
+      "loss": 0.4841,
+      "step": 472
+    },
+    {
+      "epoch": 1.9149797570850202,
+      "grad_norm": 0.5269492954320466,
+      "learning_rate": 3.1026029544295745e-05,
+      "loss": 0.5032,
+      "step": 473
+    },
+    {
+      "epoch": 1.9190283400809718,
+      "grad_norm": 0.40772971810071384,
+      "learning_rate": 3.097880190800057e-05,
+      "loss": 0.4995,
+      "step": 474
+    },
+    {
+      "epoch": 1.9230769230769231,
+      "grad_norm": 0.5164410997961938,
+      "learning_rate": 3.093148648549788e-05,
+      "loss": 0.4925,
+      "step": 475
+    },
+    {
+      "epoch": 1.9271255060728745,
+      "grad_norm": 0.43921118342650717,
+      "learning_rate": 3.088408365512055e-05,
+      "loss": 0.4824,
+      "step": 476
+    },
+    {
+      "epoch": 1.931174089068826,
+      "grad_norm": 0.4954989261494303,
+      "learning_rate": 3.083659379590034e-05,
+      "loss": 0.5039,
+      "step": 477
+    },
+    {
+      "epoch": 1.9352226720647774,
+      "grad_norm": 0.4899993878836004,
+      "learning_rate": 3.0789017287564913e-05,
+      "loss": 0.5087,
+      "step": 478
+    },
+    {
+      "epoch": 1.9392712550607287,
+      "grad_norm": 0.4771398368733428,
+      "learning_rate": 3.074135451053475e-05,
+      "loss": 0.4774,
+      "step": 479
+    },
+    {
+      "epoch": 1.9433198380566803,
+      "grad_norm": 0.48560980583462204,
+      "learning_rate": 3.069360584592015e-05,
+      "loss": 0.4923,
+      "step": 480
+    },
+    {
+      "epoch": 1.9473684210526314,
+      "grad_norm": 0.46090282242556163,
+      "learning_rate": 3.064577167551816e-05,
+      "loss": 0.4967,
+      "step": 481
+    },
+    {
+      "epoch": 1.951417004048583,
+      "grad_norm": 0.4420101914271031,
+      "learning_rate": 3.059785238180953e-05,
+      "loss": 0.4816,
+      "step": 482
+    },
+    {
+      "epoch": 1.9554655870445345,
+      "grad_norm": 0.4736130476428284,
+      "learning_rate": 3.054984834795566e-05,
+      "loss": 0.5065,
+      "step": 483
+    },
+    {
+      "epoch": 1.9595141700404857,
+      "grad_norm": 0.44617126914045524,
+      "learning_rate": 3.0501759957795537e-05,
+      "loss": 0.4723,
+      "step": 484
+    },
+    {
+      "epoch": 1.9635627530364372,
+      "grad_norm": 0.4654097411394151,
+      "learning_rate": 3.0453587595842627e-05,
+      "loss": 0.469,
+      "step": 485
+    },
+    {
+      "epoch": 1.9676113360323888,
+      "grad_norm": 0.5225699961579328,
+      "learning_rate": 3.0405331647281856e-05,
+      "loss": 0.4861,
+      "step": 486
+    },
+    {
+      "epoch": 1.97165991902834,
+      "grad_norm": 0.44595239143999055,
+      "learning_rate": 3.0356992497966508e-05,
+      "loss": 0.4633,
+      "step": 487
+    },
+    {
+      "epoch": 1.9757085020242915,
+      "grad_norm": 0.5068756648541544,
+      "learning_rate": 3.0308570534415137e-05,
+      "loss": 0.4808,
+      "step": 488
+    },
+    {
+      "epoch": 1.979757085020243,
+      "grad_norm": 0.48262094502830755,
+      "learning_rate": 3.0260066143808467e-05,
+      "loss": 0.4687,
+      "step": 489
+    },
+    {
+      "epoch": 1.9838056680161942,
+      "grad_norm": 0.4740909700701981,
+      "learning_rate": 3.021147971398632e-05,
+      "loss": 0.4811,
+      "step": 490
+    },
+    {
+      "epoch": 1.9878542510121457,
+      "grad_norm": 0.4284026824000849,
+      "learning_rate": 3.0162811633444478e-05,
+      "loss": 0.4618,
+      "step": 491
+    },
+    {
+      "epoch": 1.9919028340080973,
+      "grad_norm": 0.4618143329425092,
+      "learning_rate": 3.0114062291331635e-05,
+      "loss": 0.4801,
+      "step": 492
+    },
+    {
+      "epoch": 1.9959514170040484,
+      "grad_norm": 0.4779190721423571,
+      "learning_rate": 3.0065232077446217e-05,
+      "loss": 0.4789,
+      "step": 493
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.48918195545902493,
+      "learning_rate": 3.001632138223332e-05,
+      "loss": 0.4697,
+      "step": 494
+    },
+    {
+      "epoch": 2.0040485829959516,
+      "grad_norm": 0.7008395485780805,
+      "learning_rate": 2.9967330596781565e-05,
+      "loss": 0.3728,
+      "step": 495
+    },
+    {
+      "epoch": 2.0080971659919027,
+      "grad_norm": 0.5456845902800004,
+      "learning_rate": 2.9918260112819952e-05,
+      "loss": 0.3334,
+      "step": 496
+    },
+    {
+      "epoch": 2.0121457489878543,
+      "grad_norm": 1.3542571473775147,
+      "learning_rate": 2.9869110322714778e-05,
+      "loss": 0.3721,
+      "step": 497
+    },
+    {
+      "epoch": 2.016194331983806,
+      "grad_norm": 0.6054218137606142,
+      "learning_rate": 2.9819881619466447e-05,
+      "loss": 0.3433,
+      "step": 498
+    },
+    {
+      "epoch": 2.020242914979757,
+      "grad_norm": 0.5645615826078025,
+      "learning_rate": 2.9770574396706364e-05,
+      "loss": 0.34,
+      "step": 499
+    },
+    {
+      "epoch": 2.0242914979757085,
+      "grad_norm": 0.5564419006970671,
+      "learning_rate": 2.972118904869376e-05,
+      "loss": 0.346,
+      "step": 500
+    },
+    {
+      "epoch": 2.02834008097166,
+      "grad_norm": 0.48353775333834265,
+      "learning_rate": 2.9671725970312563e-05,
+      "loss": 0.3473,
+      "step": 501
+    },
+    {
+      "epoch": 2.032388663967611,
+      "grad_norm": 0.5555211311360618,
+      "learning_rate": 2.9622185557068238e-05,
+      "loss": 0.3456,
+      "step": 502
+    },
+    {
+      "epoch": 2.0364372469635628,
+      "grad_norm": 0.4598262891071463,
+      "learning_rate": 2.95725682050846e-05,
+      "loss": 0.3458,
+      "step": 503
+    },
+    {
+      "epoch": 2.0404858299595143,
+      "grad_norm": 0.5666802954014268,
+      "learning_rate": 2.9522874311100684e-05,
+      "loss": 0.3539,
+      "step": 504
+    },
+    {
+      "epoch": 2.0445344129554655,
+      "grad_norm": 0.4557918691716244,
+      "learning_rate": 2.9473104272467526e-05,
+      "loss": 0.3303,
+      "step": 505
+    },
+    {
+      "epoch": 2.048582995951417,
+      "grad_norm": 0.5495635065938249,
+      "learning_rate": 2.942325848714505e-05,
+      "loss": 0.3259,
+      "step": 506
+    },
+    {
+      "epoch": 2.0526315789473686,
+      "grad_norm": 0.5118676603769023,
+      "learning_rate": 2.9373337353698804e-05,
+      "loss": 0.3469,
+      "step": 507
+    },
+    {
+      "epoch": 2.0566801619433197,
+      "grad_norm": 0.5005490210937674,
+      "learning_rate": 2.932334127129686e-05,
+      "loss": 0.3317,
+      "step": 508
+    },
+    {
+      "epoch": 2.0607287449392713,
+      "grad_norm": 0.530588600352991,
+      "learning_rate": 2.9273270639706547e-05,
+      "loss": 0.3449,
+      "step": 509
+    },
+    {
+      "epoch": 2.064777327935223,
+      "grad_norm": 0.4972106753084377,
+      "learning_rate": 2.922312585929131e-05,
+      "loss": 0.3424,
+      "step": 510
+    },
+    {
+      "epoch": 2.068825910931174,
+      "grad_norm": 0.4600370079727931,
+      "learning_rate": 2.9172907331007478e-05,
+      "loss": 0.3297,
+      "step": 511
+    },
+    {
+      "epoch": 2.0728744939271255,
+      "grad_norm": 0.4411508309961058,
+      "learning_rate": 2.9122615456401053e-05,
+      "loss": 0.3408,
+      "step": 512
+    },
+    {
+      "epoch": 2.076923076923077,
+      "grad_norm": 0.4145602259666457,
+      "learning_rate": 2.9072250637604535e-05,
+      "loss": 0.344,
+      "step": 513
+    },
+    {
+      "epoch": 2.080971659919028,
+      "grad_norm": 0.4832420071109493,
+      "learning_rate": 2.9021813277333667e-05,
+      "loss": 0.3375,
+      "step": 514
+    },
+    {
+      "epoch": 2.08502024291498,
+      "grad_norm": 0.42938347715636577,
+      "learning_rate": 2.8971303778884244e-05,
+      "loss": 0.3548,
+      "step": 515
+    },
+    {
+      "epoch": 2.0890688259109313,
+      "grad_norm": 0.4692165269239897,
+      "learning_rate": 2.8920722546128863e-05,
+      "loss": 0.3339,
+      "step": 516
+    },
+    {
+      "epoch": 2.0931174089068825,
+      "grad_norm": 0.43722791287336554,
+      "learning_rate": 2.887006998351372e-05,
+      "loss": 0.3548,
+      "step": 517
+    },
+    {
+      "epoch": 2.097165991902834,
+      "grad_norm": 0.4644355606037003,
+      "learning_rate": 2.881934649605535e-05,
+      "loss": 0.3446,
+      "step": 518
+    },
+    {
+      "epoch": 2.1012145748987856,
+      "grad_norm": 0.43032643922760594,
+      "learning_rate": 2.8768552489337408e-05,
+      "loss": 0.3486,
+      "step": 519
+    },
+    {
+      "epoch": 2.1052631578947367,
+      "grad_norm": 0.43098041953562244,
+      "learning_rate": 2.8717688369507425e-05,
+      "loss": 0.3553,
+      "step": 520
+    },
+    {
+      "epoch": 2.1093117408906883,
+      "grad_norm": 0.4806927439605183,
+      "learning_rate": 2.8666754543273535e-05,
+      "loss": 0.3415,
+      "step": 521
+    },
+    {
+      "epoch": 2.11336032388664,
+      "grad_norm": 0.4210088085688488,
+      "learning_rate": 2.8615751417901273e-05,
+      "loss": 0.3397,
+      "step": 522
+    },
+    {
+      "epoch": 2.117408906882591,
+      "grad_norm": 0.4288930482683762,
+      "learning_rate": 2.856467940121025e-05,
+      "loss": 0.3474,
+      "step": 523
+    },
+    {
+      "epoch": 2.1214574898785425,
+      "grad_norm": 0.4954159951421326,
+      "learning_rate": 2.8513538901570964e-05,
+      "loss": 0.33,
+      "step": 524
+    },
+    {
+      "epoch": 2.125506072874494,
+      "grad_norm": 0.4539619684119518,
+      "learning_rate": 2.846233032790149e-05,
+      "loss": 0.3362,
+      "step": 525
+    },
+    {
+      "epoch": 2.1295546558704452,
+      "grad_norm": 0.4887941566385534,
+      "learning_rate": 2.84110540896642e-05,
+      "loss": 0.3436,
+      "step": 526
+    },
+    {
+      "epoch": 2.133603238866397,
+      "grad_norm": 0.4895768638893197,
+      "learning_rate": 2.8359710596862555e-05,
+      "loss": 0.3137,
+      "step": 527
+    },
+    {
+      "epoch": 2.1376518218623484,
+      "grad_norm": 0.4026110369752579,
+      "learning_rate": 2.8308300260037734e-05,
+      "loss": 0.3356,
+      "step": 528
+    },
+    {
+      "epoch": 2.1417004048582995,
+      "grad_norm": 0.40669004886129867,
+      "learning_rate": 2.8256823490265424e-05,
+      "loss": 0.3309,
+      "step": 529
+    },
+    {
+      "epoch": 2.145748987854251,
+      "grad_norm": 0.4828532522489602,
+      "learning_rate": 2.8205280699152515e-05,
+      "loss": 0.3304,
+      "step": 530
+    },
+    {
+      "epoch": 2.1497975708502026,
+      "grad_norm": 0.41043677984248134,
+      "learning_rate": 2.8153672298833777e-05,
+      "loss": 0.3479,
+      "step": 531
+    },
+    {
+      "epoch": 2.1538461538461537,
+      "grad_norm": 0.4096313963317629,
+      "learning_rate": 2.8101998701968615e-05,
+      "loss": 0.3492,
+      "step": 532
+    },
+    {
+      "epoch": 2.1578947368421053,
+      "grad_norm": 0.4424089110629253,
+      "learning_rate": 2.805026032173773e-05,
+      "loss": 0.3414,
+      "step": 533
+    },
+    {
+      "epoch": 2.161943319838057,
+      "grad_norm": 0.4018193753359818,
+      "learning_rate": 2.7998457571839832e-05,
+      "loss": 0.332,
+      "step": 534
+    },
+    {
+      "epoch": 2.165991902834008,
+      "grad_norm": 0.4309375186021196,
+      "learning_rate": 2.7946590866488322e-05,
+      "loss": 0.3366,
+      "step": 535
+    },
+    {
+      "epoch": 2.1700404858299596,
+      "grad_norm": 0.39042762529884223,
+      "learning_rate": 2.7894660620408e-05,
+      "loss": 0.3277,
+      "step": 536
+    },
+    {
+      "epoch": 2.174089068825911,
+      "grad_norm": 0.4730092415753691,
+      "learning_rate": 2.784266724883173e-05,
+      "loss": 0.3418,
+      "step": 537
+    },
+    {
+      "epoch": 2.1781376518218623,
+      "grad_norm": 0.4076041464553606,
+      "learning_rate": 2.7790611167497128e-05,
+      "loss": 0.3328,
+      "step": 538
+    },
+    {
+      "epoch": 2.182186234817814,
+      "grad_norm": 0.44716003416967287,
+      "learning_rate": 2.7738492792643224e-05,
+      "loss": 0.3511,
+      "step": 539
+    },
+    {
+      "epoch": 2.1862348178137654,
+      "grad_norm": 0.41625508456595556,
+      "learning_rate": 2.768631254100716e-05,
+      "loss": 0.3281,
+      "step": 540
+    },
+    {
+      "epoch": 2.1902834008097165,
+      "grad_norm": 0.47606759286683453,
+      "learning_rate": 2.7634070829820836e-05,
+      "loss": 0.3346,
+      "step": 541
+    },
+    {
+      "epoch": 2.194331983805668,
+      "grad_norm": 0.4365680435487121,
+      "learning_rate": 2.7581768076807592e-05,
+      "loss": 0.3423,
+      "step": 542
+    },
+    {
+      "epoch": 2.1983805668016196,
+      "grad_norm": 0.4818507405517365,
+      "learning_rate": 2.7529404700178832e-05,
+      "loss": 0.3386,
+      "step": 543
+    },
+    {
+      "epoch": 2.2024291497975708,
+      "grad_norm": 0.41246469271538627,
+      "learning_rate": 2.7476981118630735e-05,
+      "loss": 0.3416,
+      "step": 544
+    },
+    {
+      "epoch": 2.2064777327935223,
+      "grad_norm": 0.45245534747297866,
+      "learning_rate": 2.742449775134085e-05,
+      "loss": 0.3331,
+      "step": 545
+    },
+    {
+      "epoch": 2.2105263157894735,
+      "grad_norm": 0.4466428250713661,
+      "learning_rate": 2.7371955017964788e-05,
+      "loss": 0.3287,
+      "step": 546
+    },
+    {
+      "epoch": 2.214574898785425,
+      "grad_norm": 0.43907239494381306,
+      "learning_rate": 2.7319353338632845e-05,
+      "loss": 0.3313,
+      "step": 547
+    },
+    {
+      "epoch": 2.2186234817813766,
+      "grad_norm": 0.410709599940642,
+      "learning_rate": 2.7266693133946645e-05,
+      "loss": 0.3513,
+      "step": 548
+    },
+    {
+      "epoch": 2.2226720647773277,
+      "grad_norm": 0.4149227286865324,
+      "learning_rate": 2.7213974824975785e-05,
+      "loss": 0.3361,
+      "step": 549
+    },
+    {
+      "epoch": 2.2267206477732793,
+      "grad_norm": 0.42807459928364344,
+      "learning_rate": 2.7161198833254464e-05,
+      "loss": 0.3518,
+      "step": 550
+    },
+    {
+      "epoch": 2.230769230769231,
+      "grad_norm": 0.4447152407937643,
+      "learning_rate": 2.7108365580778095e-05,
+      "loss": 0.3482,
+      "step": 551
+    },
+    {
+      "epoch": 2.234817813765182,
+      "grad_norm": 0.41567917147670286,
+      "learning_rate": 2.705547548999996e-05,
+      "loss": 0.3688,
+      "step": 552
+    },
+    {
+      "epoch": 2.2388663967611335,
+      "grad_norm": 0.457813784738024,
+      "learning_rate": 2.7002528983827817e-05,
+      "loss": 0.3496,
+      "step": 553
+    },
+    {
+      "epoch": 2.242914979757085,
+      "grad_norm": 0.41456407684902896,
+      "learning_rate": 2.6949526485620517e-05,
+      "loss": 0.3461,
+      "step": 554
+    },
+    {
+      "epoch": 2.246963562753036,
+      "grad_norm": 0.4560622217758233,
+      "learning_rate": 2.6896468419184623e-05,
+      "loss": 0.3456,
+      "step": 555
+    },
+    {
+      "epoch": 2.251012145748988,
+      "grad_norm": 0.41142037763416944,
+      "learning_rate": 2.6843355208771024e-05,
+      "loss": 0.3491,
+      "step": 556
+    },
+    {
+      "epoch": 2.2550607287449393,
+      "grad_norm": 0.4634165134172303,
+      "learning_rate": 2.6790187279071536e-05,
+      "loss": 0.3656,
+      "step": 557
+    },
+    {
+      "epoch": 2.2591093117408905,
+      "grad_norm": 0.3571633505565813,
+      "learning_rate": 2.67369650552155e-05,
+      "loss": 0.3437,
+      "step": 558
+    },
+    {
+      "epoch": 2.263157894736842,
+      "grad_norm": 0.45109577793523603,
+      "learning_rate": 2.6683688962766407e-05,
+      "loss": 0.375,
+      "step": 559
+    },
+    {
+      "epoch": 2.2672064777327936,
+      "grad_norm": 0.42296226389065467,
+      "learning_rate": 2.6630359427718468e-05,
+      "loss": 0.3492,
+      "step": 560
+    },
+    {
+      "epoch": 2.2712550607287447,
+      "grad_norm": 0.44608627933306305,
+      "learning_rate": 2.6576976876493225e-05,
+      "loss": 0.3427,
+      "step": 561
+    },
+    {
+      "epoch": 2.2753036437246963,
+      "grad_norm": 0.4150366397730266,
+      "learning_rate": 2.652354173593614e-05,
+      "loss": 0.3443,
+      "step": 562
+    },
+    {
+      "epoch": 2.279352226720648,
+      "grad_norm": 0.43357215282945716,
+      "learning_rate": 2.647005443331317e-05,
+      "loss": 0.3594,
+      "step": 563
+    },
+    {
+      "epoch": 2.283400809716599,
+      "grad_norm": 0.4014718641833213,
+      "learning_rate": 2.6416515396307354e-05,
+      "loss": 0.3409,
+      "step": 564
+    },
+    {
+      "epoch": 2.2874493927125505,
+      "grad_norm": 0.424593149144851,
+      "learning_rate": 2.636292505301542e-05,
+      "loss": 0.349,
+      "step": 565
+    },
+    {
+      "epoch": 2.291497975708502,
+      "grad_norm": 0.4124185641534694,
+      "learning_rate": 2.630928383194431e-05,
+      "loss": 0.3548,
+      "step": 566
+    },
+    {
+      "epoch": 2.2955465587044532,
+      "grad_norm": 0.46071104458703505,
+      "learning_rate": 2.625559216200781e-05,
+      "loss": 0.3623,
+      "step": 567
+    },
+    {
+      "epoch": 2.299595141700405,
+      "grad_norm": 0.43736765780013487,
+      "learning_rate": 2.620185047252308e-05,
+      "loss": 0.3214,
+      "step": 568
+    },
+    {
+      "epoch": 2.3036437246963564,
+      "grad_norm": 0.3972729259742856,
+      "learning_rate": 2.614805919320723e-05,
+      "loss": 0.3219,
+      "step": 569
+    },
+    {
+      "epoch": 2.3076923076923075,
+      "grad_norm": 0.3978018615859414,
+      "learning_rate": 2.6094218754173895e-05,
+      "loss": 0.3128,
+      "step": 570
+    },
+    {
+      "epoch": 2.311740890688259,
+      "grad_norm": 0.4225131337613495,
+      "learning_rate": 2.60403295859298e-05,
+      "loss": 0.34,
+      "step": 571
+    },
+    {
+      "epoch": 2.3157894736842106,
+      "grad_norm": 0.37057088727537996,
+      "learning_rate": 2.5986392119371297e-05,
+      "loss": 0.3648,
+      "step": 572
+    },
+    {
+      "epoch": 2.3198380566801617,
+      "grad_norm": 0.40258252967132524,
+      "learning_rate": 2.5932406785780928e-05,
+      "loss": 0.3324,
+      "step": 573
+    },
+    {
+      "epoch": 2.3238866396761133,
+      "grad_norm": 0.4131514306320514,
+      "learning_rate": 2.5878374016823987e-05,
+      "loss": 0.3385,
+      "step": 574
+    },
+    {
+      "epoch": 2.327935222672065,
+      "grad_norm": 0.42887069410554124,
+      "learning_rate": 2.5824294244545045e-05,
+      "loss": 0.3545,
+      "step": 575
+    },
+    {
+      "epoch": 2.331983805668016,
+      "grad_norm": 0.4254378960228597,
+      "learning_rate": 2.577016790136453e-05,
+      "loss": 0.3396,
+      "step": 576
+    },
+    {
+      "epoch": 2.3360323886639676,
+      "grad_norm": 0.4500607698756121,
+      "learning_rate": 2.5715995420075225e-05,
+      "loss": 0.3499,
+      "step": 577
+    },
+    {
+      "epoch": 2.340080971659919,
+      "grad_norm": 0.4149463670761266,
+      "learning_rate": 2.5661777233838864e-05,
+      "loss": 0.3491,
+      "step": 578
+    },
+    {
+      "epoch": 2.3441295546558703,
+      "grad_norm": 0.481555944088884,
+      "learning_rate": 2.5607513776182616e-05,
+      "loss": 0.3389,
+      "step": 579
+    },
+    {
+      "epoch": 2.348178137651822,
+      "grad_norm": 0.44106637913247987,
+      "learning_rate": 2.5553205480995633e-05,
+      "loss": 0.3528,
+      "step": 580
+    },
+    {
+      "epoch": 2.3522267206477734,
+      "grad_norm": 0.4403539251476022,
+      "learning_rate": 2.5498852782525604e-05,
+      "loss": 0.3399,
+      "step": 581
+    },
+    {
+      "epoch": 2.3562753036437245,
+      "grad_norm": 0.4542840046941966,
+      "learning_rate": 2.5444456115375252e-05,
+      "loss": 0.3489,
+      "step": 582
+    },
+    {
+      "epoch": 2.360323886639676,
+      "grad_norm": 0.4087376729545972,
+      "learning_rate": 2.539001591449889e-05,
+      "loss": 0.356,
+      "step": 583
+    },
+    {
+      "epoch": 2.3643724696356276,
+      "grad_norm": 0.42317076748469384,
+      "learning_rate": 2.53355326151989e-05,
+      "loss": 0.3389,
+      "step": 584
+    },
+    {
+      "epoch": 2.3684210526315788,
+      "grad_norm": 0.38329758535864505,
+      "learning_rate": 2.5281006653122288e-05,
+      "loss": 0.3488,
+      "step": 585
+    },
+    {
+      "epoch": 2.3724696356275303,
+      "grad_norm": 0.4027068388279372,
+      "learning_rate": 2.5226438464257217e-05,
+      "loss": 0.3449,
+      "step": 586
+    },
+    {
+      "epoch": 2.376518218623482,
+      "grad_norm": 0.37808801596446745,
+      "learning_rate": 2.5171828484929447e-05,
+      "loss": 0.3428,
+      "step": 587
+    },
+    {
+      "epoch": 2.380566801619433,
+      "grad_norm": 0.385117437562582,
+      "learning_rate": 2.5117177151798934e-05,
+      "loss": 0.3279,
+      "step": 588
+    },
+    {
+      "epoch": 2.3846153846153846,
+      "grad_norm": 0.39632249040757145,
+      "learning_rate": 2.506248490185627e-05,
+      "loss": 0.3545,
+      "step": 589
+    },
+    {
+      "epoch": 2.388663967611336,
+      "grad_norm": 0.38124745864841025,
+      "learning_rate": 2.500775217241923e-05,
+      "loss": 0.3263,
+      "step": 590
+    },
+    {
+      "epoch": 2.3927125506072873,
+      "grad_norm": 0.410938477091929,
+      "learning_rate": 2.4952979401129266e-05,
+      "loss": 0.3247,
+      "step": 591
+    },
+    {
+      "epoch": 2.396761133603239,
+      "grad_norm": 0.3733235703157113,
+      "learning_rate": 2.4898167025948e-05,
+      "loss": 0.3434,
+      "step": 592
+    },
+    {
+      "epoch": 2.4008097165991904,
+      "grad_norm": 0.379957463386622,
+      "learning_rate": 2.4843315485153703e-05,
+      "loss": 0.3738,
+      "step": 593
+    },
+    {
+      "epoch": 2.4048582995951415,
+      "grad_norm": 0.40234274124401437,
+      "learning_rate": 2.4788425217337856e-05,
+      "loss": 0.3481,
+      "step": 594
+    },
+    {
+      "epoch": 2.408906882591093,
+      "grad_norm": 0.3764667823584245,
+      "learning_rate": 2.4733496661401572e-05,
+      "loss": 0.3473,
+      "step": 595
+    },
+    {
+      "epoch": 2.4129554655870447,
+      "grad_norm": 0.38165525148813106,
+      "learning_rate": 2.4678530256552114e-05,
+      "loss": 0.3452,
+      "step": 596
+    },
+    {
+      "epoch": 2.417004048582996,
+      "grad_norm": 0.4123920386476137,
+      "learning_rate": 2.46235264422994e-05,
+      "loss": 0.3555,
+      "step": 597
+    },
+    {
+      "epoch": 2.4210526315789473,
+      "grad_norm": 0.3824475527463244,
+      "learning_rate": 2.4568485658452452e-05,
+      "loss": 0.359,
+      "step": 598
+    },
+    {
+      "epoch": 2.425101214574899,
+      "grad_norm": 0.39888803120739014,
+      "learning_rate": 2.451340834511592e-05,
+      "loss": 0.3526,
+      "step": 599
+    },
+    {
+      "epoch": 2.42914979757085,
+      "grad_norm": 0.44272441333209617,
+      "learning_rate": 2.4458294942686528e-05,
+      "loss": 0.3453,
+      "step": 600
+    },
+    {
+      "epoch": 2.4331983805668016,
+      "grad_norm": 0.38970102904091464,
+      "learning_rate": 2.440314589184958e-05,
+      "loss": 0.3385,
+      "step": 601
+    },
+    {
+      "epoch": 2.437246963562753,
+      "grad_norm": 0.44417448152245836,
+      "learning_rate": 2.434796163357541e-05,
+      "loss": 0.3696,
+      "step": 602
+    },
+    {
+      "epoch": 2.4412955465587043,
+      "grad_norm": 0.37478701287206345,
+      "learning_rate": 2.429274260911588e-05,
+      "loss": 0.3594,
+      "step": 603
+    },
+    {
+      "epoch": 2.445344129554656,
+      "grad_norm": 0.40565000299147147,
+      "learning_rate": 2.423748926000084e-05,
+      "loss": 0.3576,
+      "step": 604
+    },
+    {
+      "epoch": 2.4493927125506074,
+      "grad_norm": 0.42101360990669473,
+      "learning_rate": 2.418220202803459e-05,
+      "loss": 0.3726,
+      "step": 605
+    },
+    {
+      "epoch": 2.4534412955465585,
+      "grad_norm": 0.3939370791876475,
+      "learning_rate": 2.412688135529237e-05,
+      "loss": 0.3592,
+      "step": 606
+    },
+    {
+      "epoch": 2.45748987854251,
+      "grad_norm": 0.4683971849177693,
+      "learning_rate": 2.4071527684116807e-05,
+      "loss": 0.3217,
+      "step": 607
+    },
+    {
+      "epoch": 2.4615384615384617,
+      "grad_norm": 0.40219969352034385,
+      "learning_rate": 2.4016141457114372e-05,
+      "loss": 0.3517,
+      "step": 608
+    },
+    {
+      "epoch": 2.465587044534413,
+      "grad_norm": 0.38694321426686773,
+      "learning_rate": 2.396072311715187e-05,
+      "loss": 0.3719,
+      "step": 609
+    },
+    {
+      "epoch": 2.4696356275303644,
+      "grad_norm": 0.38689558643636857,
+      "learning_rate": 2.3905273107352863e-05,
+      "loss": 0.3269,
+      "step": 610
+    },
+    {
+      "epoch": 2.473684210526316,
+      "grad_norm": 0.41417729399936565,
+      "learning_rate": 2.3849791871094157e-05,
+      "loss": 0.3588,
+      "step": 611
+    },
+    {
+      "epoch": 2.477732793522267,
+      "grad_norm": 0.3786485376374925,
+      "learning_rate": 2.379427985200223e-05,
+      "loss": 0.3443,
+      "step": 612
+    },
+    {
+      "epoch": 2.4817813765182186,
+      "grad_norm": 0.4165428781226978,
+      "learning_rate": 2.373873749394972e-05,
+      "loss": 0.3342,
+      "step": 613
+    },
+    {
+      "epoch": 2.48582995951417,
+      "grad_norm": 0.4056505677154468,
+      "learning_rate": 2.3683165241051836e-05,
+      "loss": 0.348,
+      "step": 614
+    },
+    {
+      "epoch": 2.4898785425101213,
+      "grad_norm": 0.4329799605205043,
+      "learning_rate": 2.3627563537662823e-05,
+      "loss": 0.3518,
+      "step": 615
+    },
+    {
+      "epoch": 2.493927125506073,
+      "grad_norm": 0.4060157160397477,
+      "learning_rate": 2.3571932828372433e-05,
+      "loss": 0.3495,
+      "step": 616
+    },
+    {
+      "epoch": 2.4979757085020244,
+      "grad_norm": 0.42227072853106956,
+      "learning_rate": 2.351627355800233e-05,
+      "loss": 0.3355,
+      "step": 617
+    },
+    {
+      "epoch": 2.5020242914979756,
+      "grad_norm": 0.40804653260445806,
+      "learning_rate": 2.346058617160256e-05,
+      "loss": 0.3359,
+      "step": 618
+    },
+    {
+      "epoch": 2.506072874493927,
+      "grad_norm": 0.43812826629152285,
+      "learning_rate": 2.3404871114447976e-05,
+      "loss": 0.3739,
+      "step": 619
+    },
+    {
+      "epoch": 2.5101214574898787,
+      "grad_norm": 0.4072553295244795,
+      "learning_rate": 2.334912883203469e-05,
+      "loss": 0.341,
+      "step": 620
+    },
+    {
+      "epoch": 2.51417004048583,
+      "grad_norm": 0.41278085446791013,
+      "learning_rate": 2.3293359770076505e-05,
+      "loss": 0.3252,
+      "step": 621
+    },
+    {
+      "epoch": 2.5182186234817814,
+      "grad_norm": 0.4122610175804459,
+      "learning_rate": 2.323756437450136e-05,
+      "loss": 0.3757,
+      "step": 622
+    },
+    {
+      "epoch": 2.522267206477733,
+      "grad_norm": 0.40117913287539453,
+      "learning_rate": 2.3181743091447754e-05,
+      "loss": 0.3539,
+      "step": 623
+    },
+    {
+      "epoch": 2.526315789473684,
+      "grad_norm": 0.39733827513546605,
+      "learning_rate": 2.3125896367261164e-05,
+      "loss": 0.34,
+      "step": 624
+    },
+    {
+      "epoch": 2.5303643724696356,
+      "grad_norm": 0.3853013675397692,
+      "learning_rate": 2.307002464849052e-05,
+      "loss": 0.3171,
+      "step": 625
+    },
+    {
+      "epoch": 2.534412955465587,
+      "grad_norm": 0.4265768086935125,
+      "learning_rate": 2.30141283818846e-05,
+      "loss": 0.3509,
+      "step": 626
+    },
+    {
+      "epoch": 2.5384615384615383,
+      "grad_norm": 0.36605189478261435,
+      "learning_rate": 2.2958208014388462e-05,
+      "loss": 0.3403,
+      "step": 627
+    },
+    {
+      "epoch": 2.54251012145749,
+      "grad_norm": 0.41254895216193876,
+      "learning_rate": 2.2902263993139857e-05,
+      "loss": 0.3435,
+      "step": 628
+    },
+    {
+      "epoch": 2.5465587044534415,
+      "grad_norm": 0.38753685916653047,
+      "learning_rate": 2.2846296765465708e-05,
+      "loss": 0.3376,
+      "step": 629
+    },
+    {
+      "epoch": 2.5506072874493926,
+      "grad_norm": 0.3862627772189477,
+      "learning_rate": 2.2790306778878465e-05,
+      "loss": 0.3504,
+      "step": 630
+    },
+    {
+      "epoch": 2.554655870445344,
+      "grad_norm": 0.39508830356693764,
+      "learning_rate": 2.273429448107257e-05,
+      "loss": 0.3432,
+      "step": 631
+    },
+    {
+      "epoch": 2.5587044534412957,
+      "grad_norm": 0.39062387018022593,
+      "learning_rate": 2.2678260319920868e-05,
+      "loss": 0.3484,
+      "step": 632
+    },
+    {
+      "epoch": 2.562753036437247,
+      "grad_norm": 0.392767749444017,
+      "learning_rate": 2.262220474347102e-05,
+      "loss": 0.3518,
+      "step": 633
+    },
+    {
+      "epoch": 2.5668016194331984,
+      "grad_norm": 0.3688153777582563,
+      "learning_rate": 2.2566128199941918e-05,
+      "loss": 0.3451,
+      "step": 634
+    },
+    {
+      "epoch": 2.57085020242915,
+      "grad_norm": 0.3941267337694592,
+      "learning_rate": 2.2510031137720104e-05,
+      "loss": 0.3475,
+      "step": 635
+    },
+    {
+      "epoch": 2.574898785425101,
+      "grad_norm": 0.3940112299240392,
+      "learning_rate": 2.2453914005356206e-05,
+      "loss": 0.3496,
+      "step": 636
+    },
+    {
+      "epoch": 2.5789473684210527,
+      "grad_norm": 0.39851935060682225,
+      "learning_rate": 2.2397777251561302e-05,
+      "loss": 0.3733,
+      "step": 637
+    },
+    {
+      "epoch": 2.582995951417004,
+      "grad_norm": 0.4096182368238612,
+      "learning_rate": 2.2341621325203393e-05,
+      "loss": 0.3426,
+      "step": 638
+    },
+    {
+      "epoch": 2.5870445344129553,
+      "grad_norm": 0.38043016537701624,
+      "learning_rate": 2.2285446675303764e-05,
+      "loss": 0.3499,
+      "step": 639
+    },
+    {
+      "epoch": 2.591093117408907,
+      "grad_norm": 0.4334207494755431,
+      "learning_rate": 2.2229253751033423e-05,
+      "loss": 0.3487,
+      "step": 640
+    },
+    {
+      "epoch": 2.5951417004048585,
+      "grad_norm": 0.4405427003425078,
+      "learning_rate": 2.2173043001709492e-05,
+      "loss": 0.3362,
+      "step": 641
+    },
+    {
+      "epoch": 2.5991902834008096,
+      "grad_norm": 0.3901980376632828,
+      "learning_rate": 2.2116814876791627e-05,
+      "loss": 0.3632,
+      "step": 642
+    },
+    {
+      "epoch": 2.603238866396761,
+      "grad_norm": 0.45055879249886926,
+      "learning_rate": 2.2060569825878413e-05,
+      "loss": 0.342,
+      "step": 643
+    },
+    {
+      "epoch": 2.6072874493927127,
+      "grad_norm": 0.44307498189207845,
+      "learning_rate": 2.2004308298703777e-05,
+      "loss": 0.3244,
+      "step": 644
+    },
+    {
+      "epoch": 2.611336032388664,
+      "grad_norm": 0.4560081873181335,
+      "learning_rate": 2.1948030745133396e-05,
+      "loss": 0.3435,
+      "step": 645
+    },
+    {
+      "epoch": 2.6153846153846154,
+      "grad_norm": 0.4933425599250298,
+      "learning_rate": 2.1891737615161083e-05,
+      "loss": 0.3488,
+      "step": 646
+    },
+    {
+      "epoch": 2.619433198380567,
+      "grad_norm": 0.4003199826865547,
+      "learning_rate": 2.1835429358905218e-05,
+      "loss": 0.338,
+      "step": 647
+    },
+    {
+      "epoch": 2.623481781376518,
+      "grad_norm": 0.45361832417725956,
+      "learning_rate": 2.17791064266051e-05,
+      "loss": 0.3448,
+      "step": 648
+    },
+    {
+      "epoch": 2.6275303643724697,
+      "grad_norm": 0.3928054702830991,
+      "learning_rate": 2.172276926861741e-05,
+      "loss": 0.3367,
+      "step": 649
+    },
+    {
+      "epoch": 2.6315789473684212,
+      "grad_norm": 0.3950966136592473,
+      "learning_rate": 2.166641833541255e-05,
+      "loss": 0.3442,
+      "step": 650
+    },
+    {
+      "epoch": 2.6356275303643724,
+      "grad_norm": 0.42158877088084057,
+      "learning_rate": 2.161005407757109e-05,
+      "loss": 0.3338,
+      "step": 651
+    },
+    {
+      "epoch": 2.639676113360324,
+      "grad_norm": 0.411875181205174,
+      "learning_rate": 2.155367694578013e-05,
+      "loss": 0.3442,
+      "step": 652
+    },
+    {
+      "epoch": 2.6437246963562755,
+      "grad_norm": 0.4035690428466812,
+      "learning_rate": 2.1497287390829712e-05,
+      "loss": 0.3553,
+      "step": 653
+    },
+    {
+      "epoch": 2.6477732793522266,
+      "grad_norm": 0.41111611246731566,
+      "learning_rate": 2.1440885863609208e-05,
+      "loss": 0.3408,
+      "step": 654
+    },
+    {
+      "epoch": 2.651821862348178,
+      "grad_norm": 0.3818810672899174,
+      "learning_rate": 2.1384472815103737e-05,
+      "loss": 0.3555,
+      "step": 655
+    },
+    {
+      "epoch": 2.6558704453441297,
+      "grad_norm": 0.36744520611297565,
+      "learning_rate": 2.1328048696390534e-05,
+      "loss": 0.362,
+      "step": 656
+    },
+    {
+      "epoch": 2.659919028340081,
+      "grad_norm": 0.37241975031855257,
+      "learning_rate": 2.1271613958635342e-05,
+      "loss": 0.364,
+      "step": 657
+    },
+    {
+      "epoch": 2.6639676113360324,
+      "grad_norm": 0.37836733398775796,
+      "learning_rate": 2.1215169053088825e-05,
+      "loss": 0.3472,
+      "step": 658
+    },
+    {
+      "epoch": 2.668016194331984,
+      "grad_norm": 0.38090721959074,
+      "learning_rate": 2.1158714431082937e-05,
+      "loss": 0.3397,
+      "step": 659
+    },
+    {
+      "epoch": 2.672064777327935,
+      "grad_norm": 0.40618242763618134,
+      "learning_rate": 2.110225054402733e-05,
+      "loss": 0.3448,
+      "step": 660
+    },
+    {
+      "epoch": 2.6761133603238867,
+      "grad_norm": 0.41802669179143614,
+      "learning_rate": 2.1045777843405747e-05,
+      "loss": 0.3428,
+      "step": 661
+    },
+    {
+      "epoch": 2.6801619433198383,
+      "grad_norm": 0.3720445294724675,
+      "learning_rate": 2.09892967807724e-05,
+      "loss": 0.3342,
+      "step": 662
+    },
+    {
+      "epoch": 2.6842105263157894,
+      "grad_norm": 0.43030615448619214,
+      "learning_rate": 2.093280780774835e-05,
+      "loss": 0.3388,
+      "step": 663
+    },
+    {
+      "epoch": 2.688259109311741,
+      "grad_norm": 0.38210011735724797,
+      "learning_rate": 2.087631137601793e-05,
+      "loss": 0.3465,
+      "step": 664
+    },
+    {
+      "epoch": 2.6923076923076925,
+      "grad_norm": 0.4392220319407001,
+      "learning_rate": 2.0819807937325104e-05,
+      "loss": 0.3386,
+      "step": 665
+    },
+    {
+      "epoch": 2.6963562753036436,
+      "grad_norm": 0.4084426604051917,
+      "learning_rate": 2.0763297943469847e-05,
+      "loss": 0.353,
+      "step": 666
+    },
+    {
+      "epoch": 2.700404858299595,
+      "grad_norm": 0.3738265581629073,
+      "learning_rate": 2.070678184630458e-05,
+      "loss": 0.3643,
+      "step": 667
+    },
+    {
+      "epoch": 2.7044534412955468,
+      "grad_norm": 0.38853995359693755,
+      "learning_rate": 2.06502600977305e-05,
+      "loss": 0.332,
+      "step": 668
+    },
+    {
+      "epoch": 2.708502024291498,
+      "grad_norm": 0.4064623860410118,
+      "learning_rate": 2.0593733149694008e-05,
+      "loss": 0.3622,
+      "step": 669
+    },
+    {
+      "epoch": 2.7125506072874495,
+      "grad_norm": 0.4184804812694506,
+      "learning_rate": 2.0537201454183074e-05,
+      "loss": 0.3514,
+      "step": 670
+    },
+    {
+      "epoch": 2.716599190283401,
+      "grad_norm": 0.38512375110245994,
+      "learning_rate": 2.0480665463223626e-05,
+      "loss": 0.3549,
+      "step": 671
+    },
+    {
+      "epoch": 2.720647773279352,
+      "grad_norm": 0.42879659074585896,
+      "learning_rate": 2.042412562887594e-05,
+      "loss": 0.3433,
+      "step": 672
+    },
+    {
+      "epoch": 2.7246963562753037,
+      "grad_norm": 0.4174445737441004,
+      "learning_rate": 2.0367582403231033e-05,
+      "loss": 0.3535,
+      "step": 673
+    },
+    {
+      "epoch": 2.7287449392712553,
+      "grad_norm": 0.39745907839678996,
+      "learning_rate": 2.0311036238407023e-05,
+      "loss": 0.3499,
+      "step": 674
+    },
+    {
+      "epoch": 2.7327935222672064,
+      "grad_norm": 0.36482010163913414,
+      "learning_rate": 2.0254487586545534e-05,
+      "loss": 0.3559,
+      "step": 675
+    },
+    {
+      "epoch": 2.736842105263158,
+      "grad_norm": 0.3813095543973692,
+      "learning_rate": 2.0197936899808088e-05,
+      "loss": 0.3558,
+      "step": 676
+    },
+    {
+      "epoch": 2.7408906882591095,
+      "grad_norm": 0.3630937671305114,
+      "learning_rate": 2.0141384630372466e-05,
+      "loss": 0.3432,
+      "step": 677
+    },
+    {
+      "epoch": 2.7449392712550607,
+      "grad_norm": 0.38330087104795285,
+      "learning_rate": 2.0084831230429106e-05,
+      "loss": 0.3354,
+      "step": 678
+    },
+    {
+      "epoch": 2.748987854251012,
+      "grad_norm": 0.38134076574982123,
+      "learning_rate": 2.0028277152177492e-05,
+      "loss": 0.3176,
+      "step": 679
+    },
+    {
+      "epoch": 2.753036437246964,
+      "grad_norm": 0.37117780093656466,
+      "learning_rate": 1.9971722847822518e-05,
+      "loss": 0.3509,
+      "step": 680
+    },
+    {
+      "epoch": 2.757085020242915,
+      "grad_norm": 0.37961440160398685,
+      "learning_rate": 1.99151687695709e-05,
+      "loss": 0.3525,
+      "step": 681
+    },
+    {
+      "epoch": 2.7611336032388665,
+      "grad_norm": 0.35727525664826437,
+      "learning_rate": 1.9858615369627537e-05,
+      "loss": 0.3603,
+      "step": 682
+    },
+    {
+      "epoch": 2.765182186234818,
+      "grad_norm": 0.3741041414078424,
+      "learning_rate": 1.9802063100191916e-05,
+      "loss": 0.348,
+      "step": 683
+    },
+    {
+      "epoch": 2.769230769230769,
+      "grad_norm": 0.37262170674243233,
+      "learning_rate": 1.974551241345447e-05,
+      "loss": 0.325,
+      "step": 684
+    },
+    {
+      "epoch": 2.7732793522267207,
+      "grad_norm": 0.3494842652723946,
+      "learning_rate": 1.9688963761592984e-05,
+      "loss": 0.3434,
+      "step": 685
+    },
+    {
+      "epoch": 2.7773279352226723,
+      "grad_norm": 0.3844431653739888,
+      "learning_rate": 1.963241759676897e-05,
+      "loss": 0.3578,
+      "step": 686
+    },
+    {
+      "epoch": 2.7813765182186234,
+      "grad_norm": 0.3788136176708169,
+      "learning_rate": 1.9575874371124066e-05,
+      "loss": 0.3319,
+      "step": 687
+    },
+    {
+      "epoch": 2.785425101214575,
+      "grad_norm": 0.3950476376024907,
+      "learning_rate": 1.951933453677638e-05,
+      "loss": 0.3518,
+      "step": 688
+    },
+    {
+      "epoch": 2.7894736842105265,
+      "grad_norm": 0.39753285836515445,
+      "learning_rate": 1.9462798545816932e-05,
+      "loss": 0.371,
+      "step": 689
+    },
+    {
+      "epoch": 2.7935222672064777,
+      "grad_norm": 0.392422588842051,
+      "learning_rate": 1.9406266850305996e-05,
+      "loss": 0.3575,
+      "step": 690
+    },
+    {
+      "epoch": 2.7975708502024292,
+      "grad_norm": 0.38797611793261483,
+      "learning_rate": 1.9349739902269503e-05,
+      "loss": 0.3327,
+      "step": 691
+    },
+    {
+      "epoch": 2.801619433198381,
+      "grad_norm": 0.4181164244138108,
+      "learning_rate": 1.9293218153695425e-05,
+      "loss": 0.3299,
+      "step": 692
+    },
+    {
+      "epoch": 2.805668016194332,
+      "grad_norm": 0.37323392679719014,
+      "learning_rate": 1.9236702056530153e-05,
+      "loss": 0.3656,
+      "step": 693
+    },
+    {
+      "epoch": 2.8097165991902835,
+      "grad_norm": 0.38882776513078837,
+      "learning_rate": 1.9180192062674907e-05,
+      "loss": 0.3427,
+      "step": 694
+    },
+    {
+      "epoch": 2.813765182186235,
+      "grad_norm": 0.3991607995373239,
+      "learning_rate": 1.9123688623982076e-05,
+      "loss": 0.3533,
+      "step": 695
+    },
+    {
+      "epoch": 2.817813765182186,
+      "grad_norm": 0.3818352650651884,
+      "learning_rate": 1.906719219225165e-05,
+      "loss": 0.3514,
+      "step": 696
+    },
+    {
+      "epoch": 2.8218623481781377,
+      "grad_norm": 0.37743933646714334,
+      "learning_rate": 1.9010703219227603e-05,
+      "loss": 0.364,
+      "step": 697
+    },
+    {
+      "epoch": 2.8259109311740893,
+      "grad_norm": 0.40284699093732684,
+      "learning_rate": 1.8954222156594256e-05,
+      "loss": 0.3575,
+      "step": 698
+    },
+    {
+      "epoch": 2.8299595141700404,
+      "grad_norm": 0.3713888043503691,
+      "learning_rate": 1.8897749455972673e-05,
+      "loss": 0.3355,
+      "step": 699
+    },
+    {
+      "epoch": 2.834008097165992,
+      "grad_norm": 0.40928664827697114,
+      "learning_rate": 1.884128556891707e-05,
+      "loss": 0.338,
+      "step": 700
+    },
+    {
+      "epoch": 2.8380566801619436,
+      "grad_norm": 0.3714006027652269,
+      "learning_rate": 1.878483094691119e-05,
+      "loss": 0.36,
+      "step": 701
+    },
+    {
+      "epoch": 2.8421052631578947,
+      "grad_norm": 0.35674021810791257,
+      "learning_rate": 1.8728386041364664e-05,
+      "loss": 0.3264,
+      "step": 702
+    },
+    {
+      "epoch": 2.8461538461538463,
+      "grad_norm": 0.3800895227173709,
+      "learning_rate": 1.867195130360947e-05,
+      "loss": 0.3363,
+      "step": 703
+    },
+    {
+      "epoch": 2.850202429149798,
+      "grad_norm": 0.3802202727411522,
+      "learning_rate": 1.8615527184896263e-05,
+      "loss": 0.3256,
+      "step": 704
+    },
+    {
+      "epoch": 2.854251012145749,
+      "grad_norm": 0.3868973478428561,
+      "learning_rate": 1.8559114136390795e-05,
+      "loss": 0.3632,
+      "step": 705
+    },
+    {
+      "epoch": 2.8582995951417005,
+      "grad_norm": 0.353054334184619,
+      "learning_rate": 1.8502712609170298e-05,
+      "loss": 0.346,
+      "step": 706
+    },
+    {
+      "epoch": 2.862348178137652,
+      "grad_norm": 0.3808812261087705,
+      "learning_rate": 1.8446323054219876e-05,
+      "loss": 0.3592,
+      "step": 707
+    },
+    {
+      "epoch": 2.866396761133603,
+      "grad_norm": 0.3615101631172709,
+      "learning_rate": 1.838994592242891e-05,
+      "loss": 0.3558,
+      "step": 708
+    },
+    {
+      "epoch": 2.8704453441295548,
+      "grad_norm": 0.357309969810083,
+      "learning_rate": 1.8333581664587453e-05,
+      "loss": 0.3331,
+      "step": 709
+    },
+    {
+      "epoch": 2.8744939271255063,
+      "grad_norm": 0.367610406261884,
+      "learning_rate": 1.8277230731382593e-05,
+      "loss": 0.3607,
+      "step": 710
+    },
+    {
+      "epoch": 2.8785425101214575,
+      "grad_norm": 0.36605988577622417,
+      "learning_rate": 1.82208935733949e-05,
+      "loss": 0.3589,
+      "step": 711
+    },
+    {
+      "epoch": 2.882591093117409,
+      "grad_norm": 0.3893215180742917,
+      "learning_rate": 1.8164570641094793e-05,
+      "loss": 0.3299,
+      "step": 712
+    },
+    {
+      "epoch": 2.8866396761133606,
+      "grad_norm": 0.3943439463343553,
+      "learning_rate": 1.810826238483892e-05,
+      "loss": 0.3331,
+      "step": 713
+    },
+    {
+      "epoch": 2.8906882591093117,
+      "grad_norm": 0.38118177364969225,
+      "learning_rate": 1.8051969254866608e-05,
+      "loss": 0.3383,
+      "step": 714
+    },
+    {
+      "epoch": 2.8947368421052633,
+      "grad_norm": 0.3801481858643914,
+      "learning_rate": 1.7995691701296226e-05,
+      "loss": 0.3253,
+      "step": 715
+    },
+    {
+      "epoch": 2.898785425101215,
+      "grad_norm": 0.3779065872718539,
+      "learning_rate": 1.7939430174121593e-05,
+      "loss": 0.3533,
+      "step": 716
+    },
+    {
+      "epoch": 2.902834008097166,
+      "grad_norm": 0.36738781330959813,
+      "learning_rate": 1.788318512320838e-05,
+      "loss": 0.3317,
+      "step": 717
+    },
+    {
+      "epoch": 2.9068825910931175,
+      "grad_norm": 0.39233099110947983,
+      "learning_rate": 1.782695699829051e-05,
+      "loss": 0.3427,
+      "step": 718
+    },
+    {
+      "epoch": 2.910931174089069,
+      "grad_norm": 0.3682834224400836,
+      "learning_rate": 1.7770746248966587e-05,
+      "loss": 0.3509,
+      "step": 719
+    },
+    {
+      "epoch": 2.91497975708502,
+      "grad_norm": 0.3617848836619326,
+      "learning_rate": 1.7714553324696243e-05,
+      "loss": 0.3455,
+      "step": 720
+    },
+    {
+      "epoch": 2.919028340080972,
+      "grad_norm": 0.36350893840102205,
+      "learning_rate": 1.7658378674796614e-05,
+      "loss": 0.3522,
+      "step": 721
+    },
+    {
+      "epoch": 2.9230769230769234,
+      "grad_norm": 0.3511140991703137,
+      "learning_rate": 1.76022227484387e-05,
+      "loss": 0.3408,
+      "step": 722
+    },
+    {
+      "epoch": 2.9271255060728745,
+      "grad_norm": 0.3596761963859078,
+      "learning_rate": 1.7546085994643807e-05,
+      "loss": 0.3667,
+      "step": 723
+    },
+    {
+      "epoch": 2.931174089068826,
+      "grad_norm": 0.36989838420498794,
+      "learning_rate": 1.7489968862279902e-05,
+      "loss": 0.3408,
+      "step": 724
+    },
+    {
+      "epoch": 2.9352226720647776,
+      "grad_norm": 0.35441299637893864,
+      "learning_rate": 1.743387180005809e-05,
+      "loss": 0.3673,
+      "step": 725
+    },
+    {
+      "epoch": 2.9392712550607287,
+      "grad_norm": 0.3734396778867049,
+      "learning_rate": 1.737779525652899e-05,
+      "loss": 0.3182,
+      "step": 726
+    },
+    {
+      "epoch": 2.9433198380566803,
+      "grad_norm": 0.3994156249573723,
+      "learning_rate": 1.732173968007914e-05,
+      "loss": 0.3459,
+      "step": 727
+    },
+    {
+      "epoch": 2.9473684210526314,
+      "grad_norm": 0.3675677358817139,
+      "learning_rate": 1.7265705518927437e-05,
+      "loss": 0.3588,
+      "step": 728
+    },
+    {
+      "epoch": 2.951417004048583,
+      "grad_norm": 0.3721832498802279,
+      "learning_rate": 1.7209693221121542e-05,
+      "loss": 0.3565,
+      "step": 729
+    },
+    {
+      "epoch": 2.9554655870445345,
+      "grad_norm": 0.35618731279421223,
+      "learning_rate": 1.7153703234534302e-05,
+      "loss": 0.3354,
+      "step": 730
+    },
+    {
+      "epoch": 2.9595141700404857,
+      "grad_norm": 0.34097059782904565,
+      "learning_rate": 1.709773600686015e-05,
+      "loss": 0.3369,
+      "step": 731
+    },
+    {
+      "epoch": 2.9635627530364372,
+      "grad_norm": 0.36645912228750416,
+      "learning_rate": 1.7041791985611548e-05,
+      "loss": 0.3484,
+      "step": 732
+    },
+    {
+      "epoch": 2.967611336032389,
+      "grad_norm": 0.37180747039317946,
+      "learning_rate": 1.69858716181154e-05,
+      "loss": 0.3595,
+      "step": 733
+    },
+    {
+      "epoch": 2.97165991902834,
+      "grad_norm": 0.3658633444553923,
+      "learning_rate": 1.692997535150948e-05,
+      "loss": 0.3508,
+      "step": 734
+    },
+    {
+      "epoch": 2.9757085020242915,
+      "grad_norm": 0.3565476543940208,
+      "learning_rate": 1.687410363273884e-05,
+      "loss": 0.3445,
+      "step": 735
+    },
+    {
+      "epoch": 2.979757085020243,
+      "grad_norm": 0.368629568311787,
+      "learning_rate": 1.6818256908552257e-05,
+      "loss": 0.3458,
+      "step": 736
+    },
+    {
+      "epoch": 2.983805668016194,
+      "grad_norm": 0.39188950880693746,
+      "learning_rate": 1.6762435625498646e-05,
+      "loss": 0.3455,
+      "step": 737
+    },
+    {
+      "epoch": 2.9878542510121457,
+      "grad_norm": 0.3752921577476485,
+      "learning_rate": 1.67066402299235e-05,
+      "loss": 0.3458,
+      "step": 738
+    },
+    {
+      "epoch": 2.9919028340080973,
+      "grad_norm": 0.3687796149583698,
+      "learning_rate": 1.6650871167965313e-05,
+      "loss": 0.3529,
+      "step": 739
+    },
+    {
+      "epoch": 2.9959514170040484,
+      "grad_norm": 0.35874091719495055,
+      "learning_rate": 1.6595128885552028e-05,
+      "loss": 0.3457,
+      "step": 740
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 0.3719210493670856,
+      "learning_rate": 1.653941382839745e-05,
+      "loss": 0.333,
+      "step": 741
+    },
+    {
+      "epoch": 3.0040485829959516,
+      "grad_norm": 0.588732168499068,
+      "learning_rate": 1.6483726441997673e-05,
+      "loss": 0.2331,
+      "step": 742
+    },
+    {
+      "epoch": 3.0080971659919027,
+      "grad_norm": 0.3753309059726238,
+      "learning_rate": 1.642806717162757e-05,
+      "loss": 0.2217,
+      "step": 743
+    },
+    {
+      "epoch": 3.0121457489878543,
+      "grad_norm": 0.5181017164777393,
+      "learning_rate": 1.637243646233718e-05,
+      "loss": 0.2233,
+      "step": 744
+    },
+    {
+      "epoch": 3.016194331983806,
+      "grad_norm": 0.7233840667583458,
+      "learning_rate": 1.6316834758948174e-05,
+      "loss": 0.2542,
+      "step": 745
+    },
+    {
+      "epoch": 3.020242914979757,
+      "grad_norm": 0.4695932498418605,
+      "learning_rate": 1.6261262506050282e-05,
+      "loss": 0.2391,
+      "step": 746
+    },
+    {
+      "epoch": 3.0242914979757085,
+      "grad_norm": 0.5028999589470778,
+      "learning_rate": 1.620572014799777e-05,
+      "loss": 0.2655,
+      "step": 747
+    },
+    {
+      "epoch": 3.02834008097166,
+      "grad_norm": 0.5621397186608448,
+      "learning_rate": 1.6150208128905857e-05,
+      "loss": 0.2321,
+      "step": 748
+    },
+    {
+      "epoch": 3.032388663967611,
+      "grad_norm": 0.4197098472928208,
+      "learning_rate": 1.6094726892647147e-05,
+      "loss": 0.2487,
+      "step": 749
+    },
+    {
+      "epoch": 3.0364372469635628,
+      "grad_norm": 0.3909694672176846,
+      "learning_rate": 1.6039276882848135e-05,
+      "loss": 0.2312,
+      "step": 750
+    },
+    {
+      "epoch": 3.0404858299595143,
+      "grad_norm": 0.5116199980018481,
+      "learning_rate": 1.5983858542885635e-05,
+      "loss": 0.229,
+      "step": 751
+    },
+    {
+      "epoch": 3.0445344129554655,
+      "grad_norm": 0.44307189227766386,
+      "learning_rate": 1.5928472315883203e-05,
+      "loss": 0.2388,
+      "step": 752
+    },
+    {
+      "epoch": 3.048582995951417,
+      "grad_norm": 0.39336809744958195,
+      "learning_rate": 1.5873118644707633e-05,
+      "loss": 0.2407,
+      "step": 753
+    },
+    {
+      "epoch": 3.0526315789473686,
+      "grad_norm": 0.4276007858332031,
+      "learning_rate": 1.5817797971965413e-05,
+      "loss": 0.2352,
+      "step": 754
+    },
+    {
+      "epoch": 3.0566801619433197,
+      "grad_norm": 0.408440624304936,
+      "learning_rate": 1.576251073999917e-05,
+      "loss": 0.2269,
+      "step": 755
+    },
+    {
+      "epoch": 3.0607287449392713,
+      "grad_norm": 0.3811666741047265,
+      "learning_rate": 1.5707257390884126e-05,
+      "loss": 0.2099,
+      "step": 756
+    },
+    {
+      "epoch": 3.064777327935223,
+      "grad_norm": 0.38547485794301395,
+      "learning_rate": 1.5652038366424595e-05,
+      "loss": 0.2398,
+      "step": 757
+    },
+    {
+      "epoch": 3.068825910931174,
+      "grad_norm": 0.3800304686417564,
+      "learning_rate": 1.5596854108150424e-05,
+      "loss": 0.2421,
+      "step": 758
+    },
+    {
+      "epoch": 3.0728744939271255,
+      "grad_norm": 0.38622883283415843,
+      "learning_rate": 1.5541705057313476e-05,
+      "loss": 0.2233,
+      "step": 759
+    },
+    {
+      "epoch": 3.076923076923077,
+      "grad_norm": 0.40475675005093237,
+      "learning_rate": 1.5486591654884086e-05,
+      "loss": 0.2421,
+      "step": 760
+    },
+    {
+      "epoch": 3.080971659919028,
+      "grad_norm": 0.37849170242067864,
+      "learning_rate": 1.543151434154755e-05,
+      "loss": 0.231,
+      "step": 761
+    },
+    {
+      "epoch": 3.08502024291498,
+      "grad_norm": 0.36235182367621777,
+      "learning_rate": 1.5376473557700615e-05,
+      "loss": 0.2329,
+      "step": 762
+    },
+    {
+      "epoch": 3.0890688259109313,
+      "grad_norm": 0.3673582574077076,
+      "learning_rate": 1.5321469743447893e-05,
+      "loss": 0.2311,
+      "step": 763
+    },
+    {
+      "epoch": 3.0931174089068825,
+      "grad_norm": 0.35882993666315416,
+      "learning_rate": 1.5266503338598434e-05,
+      "loss": 0.2376,
+      "step": 764
+    },
+    {
+      "epoch": 3.097165991902834,
+      "grad_norm": 0.3347753120605697,
+      "learning_rate": 1.5211574782662149e-05,
+      "loss": 0.2185,
+      "step": 765
+    },
+    {
+      "epoch": 3.1012145748987856,
+      "grad_norm": 0.349446605898444,
+      "learning_rate": 1.5156684514846307e-05,
+      "loss": 0.2538,
+      "step": 766
+    },
+    {
+      "epoch": 3.1052631578947367,
+      "grad_norm": 0.3352992101612812,
+      "learning_rate": 1.5101832974052012e-05,
+      "loss": 0.2669,
+      "step": 767
+    },
+    {
+      "epoch": 3.1093117408906883,
+      "grad_norm": 0.3373100034899367,
+      "learning_rate": 1.5047020598870737e-05,
+      "loss": 0.2345,
+      "step": 768
+    },
+    {
+      "epoch": 3.11336032388664,
+      "grad_norm": 0.33566466785748006,
+      "learning_rate": 1.4992247827580778e-05,
+      "loss": 0.217,
+      "step": 769
+    },
+    {
+      "epoch": 3.117408906882591,
+      "grad_norm": 0.3348032729042543,
+      "learning_rate": 1.493751509814374e-05,
+      "loss": 0.2328,
+      "step": 770
+    },
+    {
+      "epoch": 3.1214574898785425,
+      "grad_norm": 0.32097272585133874,
+      "learning_rate": 1.4882822848201073e-05,
+      "loss": 0.2265,
+      "step": 771
+    },
+    {
+      "epoch": 3.125506072874494,
+      "grad_norm": 0.3512990722504121,
+      "learning_rate": 1.4828171515070553e-05,
+      "loss": 0.245,
+      "step": 772
+    },
+    {
+      "epoch": 3.1295546558704452,
+      "grad_norm": 0.32378986939214366,
+      "learning_rate": 1.4773561535742793e-05,
+      "loss": 0.2532,
+      "step": 773
+    },
+    {
+      "epoch": 3.133603238866397,
+      "grad_norm": 0.3425258748656735,
+      "learning_rate": 1.4718993346877715e-05,
+      "loss": 0.2205,
+      "step": 774
+    },
+    {
+      "epoch": 3.1376518218623484,
+      "grad_norm": 0.3262997794589077,
+      "learning_rate": 1.466446738480111e-05,
+      "loss": 0.2483,
+      "step": 775
+    },
+    {
+      "epoch": 3.1417004048582995,
+      "grad_norm": 0.32334076199522277,
+      "learning_rate": 1.460998408550112e-05,
+      "loss": 0.2234,
+      "step": 776
+    },
+    {
+      "epoch": 3.145748987854251,
+      "grad_norm": 0.33486562263158226,
+      "learning_rate": 1.4555543884624751e-05,
+      "loss": 0.2018,
+      "step": 777
+    },
+    {
+      "epoch": 3.1497975708502026,
+      "grad_norm": 0.3398587204493366,
+      "learning_rate": 1.4501147217474402e-05,
+      "loss": 0.2438,
+      "step": 778
+    },
+    {
+      "epoch": 3.1538461538461537,
+      "grad_norm": 0.3160046382147342,
+      "learning_rate": 1.444679451900437e-05,
+      "loss": 0.2175,
+      "step": 779
+    },
+    {
+      "epoch": 3.1578947368421053,
+      "grad_norm": 0.34380337082486967,
+      "learning_rate": 1.4392486223817397e-05,
+      "loss": 0.2433,
+      "step": 780
+    },
+    {
+      "epoch": 3.161943319838057,
+      "grad_norm": 0.34608170455640186,
+      "learning_rate": 1.4338222766161143e-05,
+      "loss": 0.2263,
+      "step": 781
+    },
+    {
+      "epoch": 3.165991902834008,
+      "grad_norm": 0.32472432197221,
+      "learning_rate": 1.4284004579924777e-05,
+      "loss": 0.2182,
+      "step": 782
+    },
+    {
+      "epoch": 3.1700404858299596,
+      "grad_norm": 0.35520713748578336,
+      "learning_rate": 1.4229832098635479e-05,
+      "loss": 0.2386,
+      "step": 783
+    },
+    {
+      "epoch": 3.174089068825911,
+      "grad_norm": 0.3572773924799936,
+      "learning_rate": 1.4175705755454963e-05,
+      "loss": 0.2299,
+      "step": 784
+    },
+    {
+      "epoch": 3.1781376518218623,
+      "grad_norm": 0.32987034631402995,
+      "learning_rate": 1.412162598317602e-05,
+      "loss": 0.2132,
+      "step": 785
+    },
+    {
+      "epoch": 3.182186234817814,
+      "grad_norm": 0.34783876126685365,
+      "learning_rate": 1.406759321421907e-05,
+      "loss": 0.2208,
+      "step": 786
+    },
+    {
+      "epoch": 3.1862348178137654,
+      "grad_norm": 0.3320381511144544,
+      "learning_rate": 1.401360788062871e-05,
+      "loss": 0.2299,
+      "step": 787
+    },
+    {
+      "epoch": 3.1902834008097165,
+      "grad_norm": 0.31991967545584704,
+      "learning_rate": 1.3959670414070204e-05,
+      "loss": 0.2299,
+      "step": 788
+    },
+    {
+      "epoch": 3.194331983805668,
+      "grad_norm": 0.32021383585855173,
+      "learning_rate": 1.3905781245826108e-05,
+      "loss": 0.2436,
+      "step": 789
+    },
+    {
+      "epoch": 3.1983805668016196,
+      "grad_norm": 0.3182469571914602,
+      "learning_rate": 1.3851940806792778e-05,
+      "loss": 0.2238,
+      "step": 790
+    },
+    {
+      "epoch": 3.2024291497975708,
+      "grad_norm": 0.3239652015898832,
+      "learning_rate": 1.379814952747693e-05,
+      "loss": 0.2356,
+      "step": 791
+    },
+    {
+      "epoch": 3.2064777327935223,
+      "grad_norm": 0.30961098271711507,
+      "learning_rate": 1.3744407837992193e-05,
+      "loss": 0.2161,
+      "step": 792
+    },
+    {
+      "epoch": 3.2105263157894735,
+      "grad_norm": 0.32557991629266747,
+      "learning_rate": 1.3690716168055692e-05,
+      "loss": 0.2313,
+      "step": 793
+    },
+    {
+      "epoch": 3.214574898785425,
+      "grad_norm": 0.32811885525332096,
+      "learning_rate": 1.363707494698459e-05,
+      "loss": 0.2364,
+      "step": 794
+    },
+    {
+      "epoch": 3.2186234817813766,
+      "grad_norm": 0.3144613212215647,
+      "learning_rate": 1.358348460369265e-05,
+      "loss": 0.2428,
+      "step": 795
+    },
+    {
+      "epoch": 3.2226720647773277,
+      "grad_norm": 0.31968194991876003,
+      "learning_rate": 1.352994556668684e-05,
+      "loss": 0.2441,
+      "step": 796
+    },
+    {
+      "epoch": 3.2267206477732793,
+      "grad_norm": 0.33758752939721054,
+      "learning_rate": 1.3476458264063865e-05,
+      "loss": 0.2237,
+      "step": 797
+    },
+    {
+      "epoch": 3.230769230769231,
+      "grad_norm": 0.32482666461958976,
+      "learning_rate": 1.342302312350678e-05,
+      "loss": 0.2464,
+      "step": 798
+    },
+    {
+      "epoch": 3.234817813765182,
+      "grad_norm": 0.32048915088558494,
+      "learning_rate": 1.3369640572281537e-05,
+      "loss": 0.2321,
+      "step": 799
+    },
+    {
+      "epoch": 3.2388663967611335,
+      "grad_norm": 0.32095741047409654,
+      "learning_rate": 1.3316311037233596e-05,
+      "loss": 0.2462,
+      "step": 800
+    },
+    {
+      "epoch": 3.242914979757085,
+      "grad_norm": 0.3228356263571317,
+      "learning_rate": 1.326303494478451e-05,
+      "loss": 0.2284,
+      "step": 801
+    },
+    {
+      "epoch": 3.246963562753036,
+      "grad_norm": 0.3120952976807808,
+      "learning_rate": 1.3209812720928472e-05,
+      "loss": 0.2319,
+      "step": 802
+    },
+    {
+      "epoch": 3.251012145748988,
+      "grad_norm": 0.3265348758918198,
+      "learning_rate": 1.315664479122898e-05,
+      "loss": 0.2324,
+      "step": 803
+    },
+    {
+      "epoch": 3.2550607287449393,
+      "grad_norm": 0.32460556359711873,
+      "learning_rate": 1.3103531580815378e-05,
+      "loss": 0.2276,
+      "step": 804
+    },
+    {
+      "epoch": 3.2591093117408905,
+      "grad_norm": 0.32787049253410977,
+      "learning_rate": 1.3050473514379488e-05,
+      "loss": 0.2447,
+      "step": 805
+    },
+    {
+      "epoch": 3.263157894736842,
+      "grad_norm": 0.32719273595501697,
+      "learning_rate": 1.2997471016172188e-05,
+      "loss": 0.2238,
+      "step": 806
+    },
+    {
+      "epoch": 3.2672064777327936,
+      "grad_norm": 0.6072149427017708,
+      "learning_rate": 1.2944524510000042e-05,
+      "loss": 0.2212,
+      "step": 807
+    },
+    {
+      "epoch": 3.2712550607287447,
+      "grad_norm": 0.3216771578066297,
+      "learning_rate": 1.2891634419221908e-05,
+      "loss": 0.2375,
+      "step": 808
+    },
+    {
+      "epoch": 3.2753036437246963,
+      "grad_norm": 0.3770772258059165,
+      "learning_rate": 1.2838801166745545e-05,
+      "loss": 0.2355,
+      "step": 809
+    },
+    {
+      "epoch": 3.279352226720648,
+      "grad_norm": 1.6955413691112553,
+      "learning_rate": 1.2786025175024216e-05,
+      "loss": 0.256,
+      "step": 810
+    },
+    {
+      "epoch": 3.283400809716599,
+      "grad_norm": 0.45070403981721396,
+      "learning_rate": 1.2733306866053357e-05,
+      "loss": 0.2436,
+      "step": 811
+    },
+    {
+      "epoch": 3.2874493927125505,
+      "grad_norm": 0.3250128947386442,
+      "learning_rate": 1.2680646661367163e-05,
+      "loss": 0.2407,
+      "step": 812
+    },
+    {
+      "epoch": 3.291497975708502,
+      "grad_norm": 0.3172284692127623,
+      "learning_rate": 1.2628044982035219e-05,
+      "loss": 0.246,
+      "step": 813
+    },
+    {
+      "epoch": 3.2955465587044532,
+      "grad_norm": 0.34427142495538837,
+      "learning_rate": 1.2575502248659155e-05,
+      "loss": 0.2292,
+      "step": 814
+    },
+    {
+      "epoch": 3.299595141700405,
+      "grad_norm": 0.3355695675585497,
+      "learning_rate": 1.2523018881369269e-05,
+      "loss": 0.2491,
+      "step": 815
+    },
+    {
+      "epoch": 3.3036437246963564,
+      "grad_norm": 0.3157645895886423,
+      "learning_rate": 1.2470595299821176e-05,
+      "loss": 0.2297,
+      "step": 816
+    },
+    {
+      "epoch": 3.3076923076923075,
+      "grad_norm": 0.3344706346908776,
+      "learning_rate": 1.2418231923192416e-05,
+      "loss": 0.2276,
+      "step": 817
+    },
+    {
+      "epoch": 3.311740890688259,
+      "grad_norm": 0.35549149755022225,
+      "learning_rate": 1.2365929170179165e-05,
+      "loss": 0.2435,
+      "step": 818
+    },
+    {
+      "epoch": 3.3157894736842106,
+      "grad_norm": 0.33968392667780395,
+      "learning_rate": 1.2313687458992848e-05,
+      "loss": 0.2213,
+      "step": 819
+    },
+    {
+      "epoch": 3.3198380566801617,
+      "grad_norm": 0.3346930126194641,
+      "learning_rate": 1.2261507207356785e-05,
+      "loss": 0.222,
+      "step": 820
+    },
+    {
+      "epoch": 3.3238866396761133,
+      "grad_norm": 0.48263973770663976,
+      "learning_rate": 1.220938883250288e-05,
+      "loss": 0.2332,
+      "step": 821
+    },
+    {
+      "epoch": 3.327935222672065,
+      "grad_norm": 0.32439648228106543,
+      "learning_rate": 1.215733275116827e-05,
+      "loss": 0.2332,
+      "step": 822
+    },
+    {
+      "epoch": 3.331983805668016,
+      "grad_norm": 0.3196249958216759,
+      "learning_rate": 1.2105339379592005e-05,
+      "loss": 0.226,
+      "step": 823
+    },
+    {
+      "epoch": 3.3360323886639676,
+      "grad_norm": 0.30961330518854085,
+      "learning_rate": 1.2053409133511681e-05,
+      "loss": 0.2431,
+      "step": 824
+    },
+    {
+      "epoch": 3.340080971659919,
+      "grad_norm": 0.33447980555174706,
+      "learning_rate": 1.2001542428160173e-05,
+      "loss": 0.247,
+      "step": 825
+    },
+    {
+      "epoch": 3.3441295546558703,
+      "grad_norm": 0.3390608993981762,
+      "learning_rate": 1.1949739678262278e-05,
+      "loss": 0.2269,
+      "step": 826
+    },
+    {
+      "epoch": 3.348178137651822,
+      "grad_norm": 0.34242489591876596,
+      "learning_rate": 1.189800129803139e-05,
+      "loss": 0.2555,
+      "step": 827
+    },
+    {
+      "epoch": 3.3522267206477734,
+      "grad_norm": 0.3214094726748209,
+      "learning_rate": 1.1846327701166228e-05,
+      "loss": 0.2589,
+      "step": 828
+    },
+    {
+      "epoch": 3.3562753036437245,
+      "grad_norm": 0.33021295389815236,
+      "learning_rate": 1.1794719300847493e-05,
+      "loss": 0.2306,
+      "step": 829
+    },
+    {
+      "epoch": 3.360323886639676,
+      "grad_norm": 0.3204404012966855,
+      "learning_rate": 1.1743176509734582e-05,
+      "loss": 0.213,
+      "step": 830
+    },
+    {
+      "epoch": 3.3643724696356276,
+      "grad_norm": 0.3116989899624952,
+      "learning_rate": 1.1691699739962275e-05,
+      "loss": 0.2228,
+      "step": 831
+    },
+    {
+      "epoch": 3.3684210526315788,
+      "grad_norm": 0.33326658660551167,
+      "learning_rate": 1.1640289403137456e-05,
+      "loss": 0.2183,
+      "step": 832
+    },
+    {
+      "epoch": 3.3724696356275303,
+      "grad_norm": 0.29857415327485276,
+      "learning_rate": 1.1588945910335802e-05,
+      "loss": 0.2555,
+      "step": 833
+    },
+    {
+      "epoch": 3.376518218623482,
+      "grad_norm": 0.3117369171971591,
+      "learning_rate": 1.1537669672098517e-05,
+      "loss": 0.2385,
+      "step": 834
+    },
+    {
+      "epoch": 3.380566801619433,
+      "grad_norm": 0.32262468055188914,
+      "learning_rate": 1.1486461098429036e-05,
+      "loss": 0.2322,
+      "step": 835
+    },
+    {
+      "epoch": 3.3846153846153846,
+      "grad_norm": 0.2976782207024069,
+      "learning_rate": 1.1435320598789747e-05,
+      "loss": 0.2514,
+      "step": 836
+    },
+    {
+      "epoch": 3.388663967611336,
+      "grad_norm": 0.3251165968488672,
+      "learning_rate": 1.1384248582098736e-05,
+      "loss": 0.239,
+      "step": 837
+    },
+    {
+      "epoch": 3.3927125506072873,
+      "grad_norm": 0.3197179650247191,
+      "learning_rate": 1.1333245456726467e-05,
+      "loss": 0.2315,
+      "step": 838
+    },
+    {
+      "epoch": 3.396761133603239,
+      "grad_norm": 0.3085088271138121,
+      "learning_rate": 1.128231163049258e-05,
+      "loss": 0.2346,
+      "step": 839
+    },
+    {
+      "epoch": 3.4008097165991904,
+      "grad_norm": 0.3348110895895118,
+      "learning_rate": 1.123144751066259e-05,
+      "loss": 0.203,
+      "step": 840
+    },
+    {
+      "epoch": 3.4048582995951415,
+      "grad_norm": 0.3047273321621733,
+      "learning_rate": 1.1180653503944657e-05,
+      "loss": 0.249,
+      "step": 841
+    },
+    {
+      "epoch": 3.408906882591093,
+      "grad_norm": 0.30545917690050883,
+      "learning_rate": 1.1129930016486287e-05,
+      "loss": 0.2355,
+      "step": 842
+    },
+    {
+      "epoch": 3.4129554655870447,
+      "grad_norm": 0.3168962854183849,
+      "learning_rate": 1.107927745387114e-05,
+      "loss": 0.2245,
+      "step": 843
+    },
+    {
+      "epoch": 3.417004048582996,
+      "grad_norm": 0.3162005205688716,
+      "learning_rate": 1.1028696221115766e-05,
+      "loss": 0.2517,
+      "step": 844
+    },
+    {
+      "epoch": 3.4210526315789473,
+      "grad_norm": 0.31005383821752813,
+      "learning_rate": 1.0978186722666341e-05,
+      "loss": 0.2071,
+      "step": 845
+    },
+    {
+      "epoch": 3.425101214574899,
+      "grad_norm": 0.31269033980234806,
+      "learning_rate": 1.0927749362395475e-05,
+      "loss": 0.2346,
+      "step": 846
+    },
+    {
+      "epoch": 3.42914979757085,
+      "grad_norm": 0.33492316307078956,
+      "learning_rate": 1.0877384543598954e-05,
+      "loss": 0.2224,
+      "step": 847
+    },
+    {
+      "epoch": 3.4331983805668016,
+      "grad_norm": 0.31042721029304216,
+      "learning_rate": 1.0827092668992529e-05,
+      "loss": 0.2409,
+      "step": 848
+    },
+    {
+      "epoch": 3.437246963562753,
+      "grad_norm": 0.3248688856650903,
+      "learning_rate": 1.077687414070869e-05,
+      "loss": 0.2329,
+      "step": 849
+    },
+    {
+      "epoch": 3.4412955465587043,
+      "grad_norm": 0.30073699421325334,
+      "learning_rate": 1.0726729360293451e-05,
+      "loss": 0.2438,
+      "step": 850
+    },
+    {
+      "epoch": 3.445344129554656,
+      "grad_norm": 0.32639289793347087,
+      "learning_rate": 1.0676658728703147e-05,
+      "loss": 0.2387,
+      "step": 851
+    },
+    {
+      "epoch": 3.4493927125506074,
+      "grad_norm": 0.3263411285692992,
+      "learning_rate": 1.0626662646301198e-05,
+      "loss": 0.2583,
+      "step": 852
+    },
+    {
+      "epoch": 3.4534412955465585,
+      "grad_norm": 0.31772689947235067,
+      "learning_rate": 1.0576741512854958e-05,
+      "loss": 0.2307,
+      "step": 853
+    },
+    {
+      "epoch": 3.45748987854251,
+      "grad_norm": 0.312577867441999,
+      "learning_rate": 1.052689572753247e-05,
+      "loss": 0.2214,
+      "step": 854
+    },
+    {
+      "epoch": 3.4615384615384617,
+      "grad_norm": 0.31744969359641595,
+      "learning_rate": 1.0477125688899328e-05,
+      "loss": 0.223,
+      "step": 855
+    },
+    {
+      "epoch": 3.465587044534413,
+      "grad_norm": 0.31638071841895066,
+      "learning_rate": 1.0427431794915406e-05,
+      "loss": 0.2174,
+      "step": 856
+    },
+    {
+      "epoch": 3.4696356275303644,
+      "grad_norm": 0.3367846947947248,
+      "learning_rate": 1.0377814442931769e-05,
+      "loss": 0.199,
+      "step": 857
+    },
+    {
+      "epoch": 3.473684210526316,
+      "grad_norm": 0.32553839499945536,
+      "learning_rate": 1.0328274029687439e-05,
+      "loss": 0.229,
+      "step": 858
+    },
+    {
+      "epoch": 3.477732793522267,
+      "grad_norm": 0.33258947787255194,
+      "learning_rate": 1.0278810951306253e-05,
+      "loss": 0.2312,
+      "step": 859
+    },
+    {
+      "epoch": 3.4817813765182186,
+      "grad_norm": 0.3314742981906798,
+      "learning_rate": 1.0229425603293647e-05,
+      "loss": 0.2376,
+      "step": 860
+    },
+    {
+      "epoch": 3.48582995951417,
+      "grad_norm": 0.31788742725596675,
+      "learning_rate": 1.0180118380533558e-05,
+      "loss": 0.2404,
+      "step": 861
+    },
+    {
+      "epoch": 3.4898785425101213,
+      "grad_norm": 0.32038925608694024,
+      "learning_rate": 1.0130889677285225e-05,
+      "loss": 0.2386,
+      "step": 862
+    },
+    {
+      "epoch": 3.493927125506073,
+      "grad_norm": 0.32319886957666444,
+      "learning_rate": 1.008173988718005e-05,
+      "loss": 0.2279,
+      "step": 863
+    },
+    {
+      "epoch": 3.4979757085020244,
+      "grad_norm": 0.32728888776442516,
+      "learning_rate": 1.0032669403218439e-05,
+      "loss": 0.206,
+      "step": 864
+    },
+    {
+      "epoch": 3.5020242914979756,
+      "grad_norm": 0.32393956568326276,
+      "learning_rate": 9.983678617766677e-06,
+      "loss": 0.2155,
+      "step": 865
+    },
+    {
+      "epoch": 3.506072874493927,
+      "grad_norm": 0.34136695802218503,
+      "learning_rate": 9.934767922553788e-06,
+      "loss": 0.2141,
+      "step": 866
+    },
+    {
+      "epoch": 3.5101214574898787,
+      "grad_norm": 0.31661002321047194,
+      "learning_rate": 9.885937708668371e-06,
+      "loss": 0.2256,
+      "step": 867
+    },
+    {
+      "epoch": 3.51417004048583,
+      "grad_norm": 0.33434411004206516,
+      "learning_rate": 9.837188366555524e-06,
+      "loss": 0.2163,
+      "step": 868
+    },
+    {
+      "epoch": 3.5182186234817814,
+      "grad_norm": 0.3596552294328643,
+      "learning_rate": 9.788520286013691e-06,
+      "loss": 0.243,
+      "step": 869
+    },
+    {
+      "epoch": 3.522267206477733,
+      "grad_norm": 0.34098069745566484,
+      "learning_rate": 9.739933856191537e-06,
+      "loss": 0.2374,
+      "step": 870
+    },
+    {
+      "epoch": 3.526315789473684,
+      "grad_norm": 0.3056190679606161,
+      "learning_rate": 9.691429465584866e-06,
+      "loss": 0.2425,
+      "step": 871
+    },
+    {
+      "epoch": 3.5303643724696356,
+      "grad_norm": 0.33842090510393247,
+      "learning_rate": 9.643007502033492e-06,
+      "loss": 0.2391,
+      "step": 872
+    },
+    {
+      "epoch": 3.534412955465587,
+      "grad_norm": 0.34111906790281016,
+      "learning_rate": 9.594668352718152e-06,
+      "loss": 0.2178,
+      "step": 873
+    },
+    {
+      "epoch": 3.5384615384615383,
+      "grad_norm": 0.32104114393038163,
+      "learning_rate": 9.546412404157383e-06,
+      "loss": 0.2376,
+      "step": 874
+    },
+    {
+      "epoch": 3.54251012145749,
+      "grad_norm": 0.31654400004528904,
+      "learning_rate": 9.498240042204471e-06,
+      "loss": 0.2151,
+      "step": 875
+    },
+    {
+      "epoch": 3.5465587044534415,
+      "grad_norm": 0.30821474830040774,
+      "learning_rate": 9.450151652044339e-06,
+      "loss": 0.2491,
+      "step": 876
+    },
+    {
+      "epoch": 3.5506072874493926,
+      "grad_norm": 0.31440709273592926,
+      "learning_rate": 9.402147618190472e-06,
+      "loss": 0.2246,
+      "step": 877
+    },
+    {
+      "epoch": 3.554655870445344,
+      "grad_norm": 0.30958436589914257,
+      "learning_rate": 9.354228324481841e-06,
+      "loss": 0.2456,
+      "step": 878
+    },
+    {
+      "epoch": 3.5587044534412957,
+      "grad_norm": 0.3198696563464883,
+      "learning_rate": 9.306394154079852e-06,
+      "loss": 0.207,
+      "step": 879
+    },
+    {
+      "epoch": 3.562753036437247,
+      "grad_norm": 0.3155238828526831,
+      "learning_rate": 9.258645489465254e-06,
+      "loss": 0.2048,
+      "step": 880
+    },
+    {
+      "epoch": 3.5668016194331984,
+      "grad_norm": 0.3158728684130103,
+      "learning_rate": 9.21098271243509e-06,
+      "loss": 0.2181,
+      "step": 881
+    },
+    {
+      "epoch": 3.57085020242915,
+      "grad_norm": 0.3234784038261893,
+      "learning_rate": 9.163406204099659e-06,
+      "loss": 0.2344,
+      "step": 882
+    },
+    {
+      "epoch": 3.574898785425101,
+      "grad_norm": 0.32445824908260484,
+      "learning_rate": 9.115916344879452e-06,
+      "loss": 0.22,
+      "step": 883
+    },
+    {
+      "epoch": 3.5789473684210527,
+      "grad_norm": 0.3175387079446192,
+      "learning_rate": 9.068513514502126e-06,
+      "loss": 0.2212,
+      "step": 884
+    },
+    {
+      "epoch": 3.582995951417004,
+      "grad_norm": 0.33424704550886614,
+      "learning_rate": 9.021198091999439e-06,
+      "loss": 0.2412,
+      "step": 885
+    },
+    {
+      "epoch": 3.5870445344129553,
+      "grad_norm": 0.31160495228468105,
+      "learning_rate": 8.973970455704251e-06,
+      "loss": 0.2342,
+      "step": 886
+    },
+    {
+      "epoch": 3.591093117408907,
+      "grad_norm": 0.31640142987593745,
+      "learning_rate": 8.926830983247495e-06,
+      "loss": 0.2311,
+      "step": 887
+    },
+    {
+      "epoch": 3.5951417004048585,
+      "grad_norm": 0.3094356144901928,
+      "learning_rate": 8.879780051555116e-06,
+      "loss": 0.2441,
+      "step": 888
+    },
+    {
+      "epoch": 3.5991902834008096,
+      "grad_norm": 0.31125472415255906,
+      "learning_rate": 8.83281803684512e-06,
+      "loss": 0.2322,
+      "step": 889
+    },
+    {
+      "epoch": 3.603238866396761,
+      "grad_norm": 0.3184238957849063,
+      "learning_rate": 8.785945314624532e-06,
+      "loss": 0.2255,
+      "step": 890
+    },
+    {
+      "epoch": 3.6072874493927127,
+      "grad_norm": 0.30909052386258695,
+      "learning_rate": 8.739162259686385e-06,
+      "loss": 0.2582,
+      "step": 891
+    },
+    {
+      "epoch": 3.611336032388664,
+      "grad_norm": 0.32750331312274705,
+      "learning_rate": 8.692469246106734e-06,
+      "loss": 0.2398,
+      "step": 892
+    },
+    {
+      "epoch": 3.6153846153846154,
+      "grad_norm": 0.3214288251774642,
+      "learning_rate": 8.645866647241677e-06,
+      "loss": 0.2311,
+      "step": 893
+    },
+    {
+      "epoch": 3.619433198380567,
+      "grad_norm": 0.31820005527645556,
+      "learning_rate": 8.599354835724348e-06,
+      "loss": 0.2244,
+      "step": 894
+    },
+    {
+      "epoch": 3.623481781376518,
+      "grad_norm": 0.30905579125858,
+      "learning_rate": 8.552934183461958e-06,
+      "loss": 0.245,
+      "step": 895
+    },
+    {
+      "epoch": 3.6275303643724697,
+      "grad_norm": 0.3089576603020441,
+      "learning_rate": 8.506605061632805e-06,
+      "loss": 0.2523,
+      "step": 896
+    },
+    {
+      "epoch": 3.6315789473684212,
+      "grad_norm": 0.31193196480587193,
+      "learning_rate": 8.460367840683308e-06,
+      "loss": 0.2346,
+      "step": 897
+    },
+    {
+      "epoch": 3.6356275303643724,
+      "grad_norm": 0.3400528443975443,
+      "learning_rate": 8.414222890325065e-06,
+      "loss": 0.2388,
+      "step": 898
+    },
+    {
+      "epoch": 3.639676113360324,
+      "grad_norm": 0.30840038422699834,
+      "learning_rate": 8.368170579531855e-06,
+      "loss": 0.2314,
+      "step": 899
+    },
+    {
+      "epoch": 3.6437246963562755,
+      "grad_norm": 0.31963848292710073,
+      "learning_rate": 8.322211276536733e-06,
+      "loss": 0.2237,
+      "step": 900
+    },
+    {
+      "epoch": 3.6477732793522266,
+      "grad_norm": 0.3096879744371264,
+      "learning_rate": 8.27634534882907e-06,
+      "loss": 0.2224,
+      "step": 901
+    },
+    {
+      "epoch": 3.651821862348178,
+      "grad_norm": 0.3132469097209949,
+      "learning_rate": 8.230573163151585e-06,
+      "loss": 0.2554,
+      "step": 902
+    },
+    {
+      "epoch": 3.6558704453441297,
+      "grad_norm": 0.32120786070229496,
+      "learning_rate": 8.184895085497466e-06,
+      "loss": 0.2107,
+      "step": 903
+    },
+    {
+      "epoch": 3.659919028340081,
+      "grad_norm": 0.3201309551053658,
+      "learning_rate": 8.139311481107401e-06,
+      "loss": 0.2227,
+      "step": 904
+    },
+    {
+      "epoch": 3.6639676113360324,
+      "grad_norm": 0.30666968753704116,
+      "learning_rate": 8.093822714466686e-06,
+      "loss": 0.2267,
+      "step": 905
+    },
+    {
+      "epoch": 3.668016194331984,
+      "grad_norm": 0.3161876314954758,
+      "learning_rate": 8.048429149302271e-06,
+      "loss": 0.2321,
+      "step": 906
+    },
+    {
+      "epoch": 3.672064777327935,
+      "grad_norm": 0.3195845624375299,
+      "learning_rate": 8.003131148579902e-06,
+      "loss": 0.23,
+      "step": 907
+    },
+    {
+      "epoch": 3.6761133603238867,
+      "grad_norm": 0.31547251704289464,
+      "learning_rate": 7.957929074501182e-06,
+      "loss": 0.2392,
+      "step": 908
+    },
+    {
+      "epoch": 3.6801619433198383,
+      "grad_norm": 0.3037208014217914,
+      "learning_rate": 7.912823288500695e-06,
+      "loss": 0.2319,
+      "step": 909
+    },
+    {
+      "epoch": 3.6842105263157894,
+      "grad_norm": 0.3159028165969539,
+      "learning_rate": 7.867814151243102e-06,
+      "loss": 0.2348,
+      "step": 910
+    },
+    {
+      "epoch": 3.688259109311741,
+      "grad_norm": 0.3160593996582869,
+      "learning_rate": 7.822902022620267e-06,
+      "loss": 0.2246,
+      "step": 911
+    },
+    {
+      "epoch": 3.6923076923076925,
+      "grad_norm": 0.3154425333597439,
+      "learning_rate": 7.778087261748377e-06,
+      "loss": 0.2179,
+      "step": 912
+    },
+    {
+      "epoch": 3.6963562753036436,
+      "grad_norm": 0.2991365820226933,
+      "learning_rate": 7.733370226965056e-06,
+      "loss": 0.2391,
+      "step": 913
+    },
+    {
+      "epoch": 3.700404858299595,
+      "grad_norm": 0.31079328086604135,
+      "learning_rate": 7.688751275826527e-06,
+      "loss": 0.2126,
+      "step": 914
+    },
+    {
+      "epoch": 3.7044534412955468,
+      "grad_norm": 0.3201341329439365,
+      "learning_rate": 7.644230765104735e-06,
+      "loss": 0.2549,
+      "step": 915
+    },
+    {
+      "epoch": 3.708502024291498,
+      "grad_norm": 0.32973759107322215,
+      "learning_rate": 7.599809050784502e-06,
+      "loss": 0.2248,
+      "step": 916
+    },
+    {
+      "epoch": 3.7125506072874495,
+      "grad_norm": 0.3242994253885812,
+      "learning_rate": 7.555486488060661e-06,
+      "loss": 0.2255,
+      "step": 917
+    },
+    {
+      "epoch": 3.716599190283401,
+      "grad_norm": 0.3024496473331279,
+      "learning_rate": 7.511263431335249e-06,
+      "loss": 0.2391,
+      "step": 918
+    },
+    {
+      "epoch": 3.720647773279352,
+      "grad_norm": 0.30281347174325146,
+      "learning_rate": 7.467140234214656e-06,
+      "loss": 0.2123,
+      "step": 919
+    },
+    {
+      "epoch": 3.7246963562753037,
+      "grad_norm": 0.3059135739162069,
+      "learning_rate": 7.423117249506777e-06,
+      "loss": 0.2289,
+      "step": 920
+    },
+    {
+      "epoch": 3.7287449392712553,
+      "grad_norm": 0.30913521102109387,
+      "learning_rate": 7.379194829218231e-06,
+      "loss": 0.2225,
+      "step": 921
+    },
+    {
+      "epoch": 3.7327935222672064,
+      "grad_norm": 0.29788703077699885,
+      "learning_rate": 7.335373324551518e-06,
+      "loss": 0.2451,
+      "step": 922
+    },
+    {
+      "epoch": 3.736842105263158,
+      "grad_norm": 0.29813534962295146,
+      "learning_rate": 7.291653085902224e-06,
+      "loss": 0.2714,
+      "step": 923
+    },
+    {
+      "epoch": 3.7408906882591095,
+      "grad_norm": 0.3176761811709042,
+      "learning_rate": 7.248034462856213e-06,
+      "loss": 0.2285,
+      "step": 924
+    },
+    {
+      "epoch": 3.7449392712550607,
+      "grad_norm": 0.3006943640881231,
+      "learning_rate": 7.20451780418683e-06,
+      "loss": 0.2251,
+      "step": 925
+    },
+    {
+      "epoch": 3.748987854251012,
+      "grad_norm": 0.3177311219483602,
+      "learning_rate": 7.161103457852128e-06,
+      "loss": 0.2345,
+      "step": 926
+    },
+    {
+      "epoch": 3.753036437246964,
+      "grad_norm": 0.3063830109539489,
+      "learning_rate": 7.117791770992046e-06,
+      "loss": 0.248,
+      "step": 927
+    },
+    {
+      "epoch": 3.757085020242915,
+      "grad_norm": 0.3139383318241536,
+      "learning_rate": 7.074583089925688e-06,
+      "loss": 0.2421,
+      "step": 928
+    },
+    {
+      "epoch": 3.7611336032388665,
+      "grad_norm": 0.306071868255141,
+      "learning_rate": 7.0314777601485155e-06,
+      "loss": 0.2444,
+      "step": 929
+    },
+    {
+      "epoch": 3.765182186234818,
+      "grad_norm": 0.31819721385594635,
+      "learning_rate": 6.988476126329604e-06,
+      "loss": 0.2253,
+      "step": 930
+    },
+    {
+      "epoch": 3.769230769230769,
+      "grad_norm": 0.30765793583368023,
+      "learning_rate": 6.945578532308857e-06,
+      "loss": 0.2121,
+      "step": 931
+    },
+    {
+      "epoch": 3.7732793522267207,
+      "grad_norm": 0.30296468886030137,
+      "learning_rate": 6.902785321094301e-06,
+      "loss": 0.2378,
+      "step": 932
+    },
+    {
+      "epoch": 3.7773279352226723,
+      "grad_norm": 0.30605457346482956,
+      "learning_rate": 6.8600968348593115e-06,
+      "loss": 0.2088,
+      "step": 933
+    },
+    {
+      "epoch": 3.7813765182186234,
+      "grad_norm": 0.3016200518598292,
+      "learning_rate": 6.817513414939892e-06,
+      "loss": 0.2308,
+      "step": 934
+    },
+    {
+      "epoch": 3.785425101214575,
+      "grad_norm": 0.31861679808530996,
+      "learning_rate": 6.775035401831924e-06,
+      "loss": 0.2341,
+      "step": 935
+    },
+    {
+      "epoch": 3.7894736842105265,
+      "grad_norm": 0.2969877695226738,
+      "learning_rate": 6.732663135188473e-06,
+      "loss": 0.2336,
+      "step": 936
+    },
+    {
+      "epoch": 3.7935222672064777,
+      "grad_norm": 0.31046116745248137,
+      "learning_rate": 6.690396953817055e-06,
+      "loss": 0.2198,
+      "step": 937
+    },
+    {
+      "epoch": 3.7975708502024292,
+      "grad_norm": 0.3059644137380178,
+      "learning_rate": 6.648237195676927e-06,
+      "loss": 0.2419,
+      "step": 938
+    },
+    {
+      "epoch": 3.801619433198381,
+      "grad_norm": 0.31612637550218625,
+      "learning_rate": 6.606184197876395e-06,
+      "loss": 0.2054,
+      "step": 939
+    },
+    {
+      "epoch": 3.805668016194332,
+      "grad_norm": 0.303067205147564,
+      "learning_rate": 6.564238296670109e-06,
+      "loss": 0.2242,
+      "step": 940
+    },
+    {
+      "epoch": 3.8097165991902835,
+      "grad_norm": 0.2924533968283581,
+      "learning_rate": 6.52239982745638e-06,
+      "loss": 0.2526,
+      "step": 941
+    },
+    {
+      "epoch": 3.813765182186235,
+      "grad_norm": 0.3086790825071644,
+      "learning_rate": 6.480669124774484e-06,
+      "loss": 0.2386,
+      "step": 942
+    },
+    {
+      "epoch": 3.817813765182186,
+      "grad_norm": 0.3027746732056239,
+      "learning_rate": 6.439046522302013e-06,
+      "loss": 0.2512,
+      "step": 943
+    },
+    {
+      "epoch": 3.8218623481781377,
+      "grad_norm": 0.30277902764750897,
+      "learning_rate": 6.397532352852191e-06,
+      "loss": 0.228,
+      "step": 944
+    },
+    {
+      "epoch": 3.8259109311740893,
+      "grad_norm": 0.3011780744587861,
+      "learning_rate": 6.356126948371202e-06,
+      "loss": 0.2524,
+      "step": 945
+    },
+    {
+      "epoch": 3.8299595141700404,
+      "grad_norm": 0.3147703932229423,
+      "learning_rate": 6.314830639935559e-06,
+      "loss": 0.2262,
+      "step": 946
+    },
+    {
+      "epoch": 3.834008097165992,
+      "grad_norm": 0.30596233858692323,
+      "learning_rate": 6.273643757749448e-06,
+      "loss": 0.2296,
+      "step": 947
+    },
+    {
+      "epoch": 3.8380566801619436,
+      "grad_norm": 0.31644416221303256,
+      "learning_rate": 6.2325666311420856e-06,
+      "loss": 0.2248,
+      "step": 948
+    },
+    {
+      "epoch": 3.8421052631578947,
+      "grad_norm": 0.29528713868860873,
+      "learning_rate": 6.1915995885650675e-06,
+      "loss": 0.2031,
+      "step": 949
+    },
+    {
+      "epoch": 3.8461538461538463,
+      "grad_norm": 0.2977385810863854,
+      "learning_rate": 6.150742957589784e-06,
+      "loss": 0.2239,
+      "step": 950
+    },
+    {
+      "epoch": 3.850202429149798,
+      "grad_norm": 0.31256442557475544,
+      "learning_rate": 6.109997064904774e-06,
+      "loss": 0.2292,
+      "step": 951
+    },
+    {
+      "epoch": 3.854251012145749,
+      "grad_norm": 0.3188557138874561,
+      "learning_rate": 6.069362236313092e-06,
+      "loss": 0.228,
+      "step": 952
+    },
+    {
+      "epoch": 3.8582995951417005,
+      "grad_norm": 0.30403856872031715,
+      "learning_rate": 6.0288387967297565e-06,
+      "loss": 0.2316,
+      "step": 953
+    },
+    {
+      "epoch": 3.862348178137652,
+      "grad_norm": 0.29651331205615905,
+      "learning_rate": 5.988427070179108e-06,
+      "loss": 0.2255,
+      "step": 954
+    },
+    {
+      "epoch": 3.866396761133603,
+      "grad_norm": 0.3065325600944174,
+      "learning_rate": 5.948127379792234e-06,
+      "loss": 0.2368,
+      "step": 955
+    },
+    {
+      "epoch": 3.8704453441295548,
+      "grad_norm": 0.29075240400345026,
+      "learning_rate": 5.90794004780439e-06,
+      "loss": 0.229,
+      "step": 956
+    },
+    {
+      "epoch": 3.8744939271255063,
+      "grad_norm": 0.3072420944159983,
+      "learning_rate": 5.867865395552405e-06,
+      "loss": 0.2332,
+      "step": 957
+    },
+    {
+      "epoch": 3.8785425101214575,
+      "grad_norm": 0.29865833937625796,
+      "learning_rate": 5.8279037434721364e-06,
+      "loss": 0.2314,
+      "step": 958
+    },
+    {
+      "epoch": 3.882591093117409,
+      "grad_norm": 0.30733299286498605,
+      "learning_rate": 5.788055411095892e-06,
+      "loss": 0.239,
+      "step": 959
+    },
+    {
+      "epoch": 3.8866396761133606,
+      "grad_norm": 0.2998440085372471,
+      "learning_rate": 5.7483207170498625e-06,
+      "loss": 0.2405,
+      "step": 960
+    },
+    {
+      "epoch": 3.8906882591093117,
+      "grad_norm": 0.2968859637133358,
+      "learning_rate": 5.708699979051613e-06,
+      "loss": 0.2138,
+      "step": 961
+    },
+    {
+      "epoch": 3.8947368421052633,
+      "grad_norm": 0.30012640387392603,
+      "learning_rate": 5.66919351390751e-06,
+      "loss": 0.2038,
+      "step": 962
+    },
+    {
+      "epoch": 3.898785425101215,
+      "grad_norm": 0.3084393321774455,
+      "learning_rate": 5.629801637510188e-06,
+      "loss": 0.2165,
+      "step": 963
+    },
+    {
+      "epoch": 3.902834008097166,
+      "grad_norm": 0.31245807885165156,
+      "learning_rate": 5.590524664836043e-06,
+      "loss": 0.2231,
+      "step": 964
+    },
+    {
+      "epoch": 3.9068825910931175,
+      "grad_norm": 0.2973127673615376,
+      "learning_rate": 5.551362909942708e-06,
+      "loss": 0.2458,
+      "step": 965
+    },
+    {
+      "epoch": 3.910931174089069,
+      "grad_norm": 0.29465598940623533,
+      "learning_rate": 5.51231668596653e-06,
+      "loss": 0.2382,
+      "step": 966
+    },
+    {
+      "epoch": 3.91497975708502,
+      "grad_norm": 0.3070495423006755,
+      "learning_rate": 5.47338630512007e-06,
+      "loss": 0.2266,
+      "step": 967
+    },
+    {
+      "epoch": 3.919028340080972,
+      "grad_norm": 0.29145368197636584,
+      "learning_rate": 5.434572078689617e-06,
+      "loss": 0.2453,
+      "step": 968
+    },
+    {
+      "epoch": 3.9230769230769234,
+      "grad_norm": 0.2945111193573569,
+      "learning_rate": 5.395874317032694e-06,
+      "loss": 0.22,
+      "step": 969
+    },
+    {
+      "epoch": 3.9271255060728745,
+      "grad_norm": 0.30149378348313277,
+      "learning_rate": 5.357293329575573e-06,
+      "loss": 0.2407,
+      "step": 970
+    },
+    {
+      "epoch": 3.931174089068826,
+      "grad_norm": 0.30111243119875614,
+      "learning_rate": 5.318829424810797e-06,
+      "loss": 0.2294,
+      "step": 971
+    },
+    {
+      "epoch": 3.9352226720647776,
+      "grad_norm": 0.30253554412570977,
+      "learning_rate": 5.280482910294728e-06,
+      "loss": 0.2277,
+      "step": 972
+    },
+    {
+      "epoch": 3.9392712550607287,
+      "grad_norm": 0.3126995155204342,
+      "learning_rate": 5.242254092645078e-06,
+      "loss": 0.2307,
+      "step": 973
+    },
+    {
+      "epoch": 3.9433198380566803,
+      "grad_norm": 0.3132273975529979,
+      "learning_rate": 5.2041432775384445e-06,
+      "loss": 0.2302,
+      "step": 974
+    },
+    {
+      "epoch": 3.9473684210526314,
+      "grad_norm": 0.2993408581032944,
+      "learning_rate": 5.166150769707892e-06,
+      "loss": 0.2128,
+      "step": 975
+    },
+    {
+      "epoch": 3.951417004048583,
+      "grad_norm": 0.294842582105119,
+      "learning_rate": 5.128276872940503e-06,
+      "loss": 0.2393,
+      "step": 976
+    },
+    {
+      "epoch": 3.9554655870445345,
+      "grad_norm": 0.29466004931828865,
+      "learning_rate": 5.090521890074953e-06,
+      "loss": 0.2187,
+      "step": 977
+    },
+    {
+      "epoch": 3.9595141700404857,
+      "grad_norm": 0.30309805889650543,
+      "learning_rate": 5.0528861229990675e-06,
+      "loss": 0.2145,
+      "step": 978
+    },
+    {
+      "epoch": 3.9635627530364372,
+      "grad_norm": 0.2943965095085482,
+      "learning_rate": 5.015369872647442e-06,
+      "loss": 0.2287,
+      "step": 979
+    },
+    {
+      "epoch": 3.967611336032389,
+      "grad_norm": 0.3016854809202041,
+      "learning_rate": 4.9779734389990195e-06,
+      "loss": 0.2173,
+      "step": 980
+    },
+    {
+      "epoch": 3.97165991902834,
+      "grad_norm": 0.3002343806718434,
+      "learning_rate": 4.940697121074675e-06,
+      "loss": 0.2325,
+      "step": 981
+    },
+    {
+      "epoch": 3.9757085020242915,
+      "grad_norm": 0.291573360229697,
+      "learning_rate": 4.903541216934864e-06,
+      "loss": 0.2379,
+      "step": 982
+    },
+    {
+      "epoch": 3.979757085020243,
+      "grad_norm": 0.3019309854120377,
+      "learning_rate": 4.8665060236772e-06,
+      "loss": 0.2273,
+      "step": 983
+    },
+    {
+      "epoch": 3.983805668016194,
+      "grad_norm": 0.3102919045385511,
+      "learning_rate": 4.829591837434102e-06,
+      "loss": 0.2432,
+      "step": 984
+    },
+    {
+      "epoch": 3.9878542510121457,
+      "grad_norm": 0.30912254975147463,
+      "learning_rate": 4.792798953370424e-06,
+      "loss": 0.2349,
+      "step": 985
+    },
+    {
+      "epoch": 3.9919028340080973,
+      "grad_norm": 0.3338722036721439,
+      "learning_rate": 4.756127665681083e-06,
+      "loss": 0.2099,
+      "step": 986
+    },
+    {
+      "epoch": 3.9959514170040484,
+      "grad_norm": 0.30654208511092973,
+      "learning_rate": 4.719578267588725e-06,
+      "loss": 0.2251,
+      "step": 987
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 0.3014897882993722,
+      "learning_rate": 4.6831510513413525e-06,
+      "loss": 0.2237,
+      "step": 988
+    },
+    {
+      "epoch": 4.004048582995951,
+      "grad_norm": 0.49396489368099256,
+      "learning_rate": 4.6468463082100176e-06,
+      "loss": 0.1625,
+      "step": 989
+    },
+    {
+      "epoch": 4.008097165991903,
+      "grad_norm": 0.4056058683586355,
+      "learning_rate": 4.610664328486485e-06,
+      "loss": 0.1856,
+      "step": 990
+    },
+    {
+      "epoch": 4.012145748987854,
+      "grad_norm": 0.27953827395783687,
+      "learning_rate": 4.5746054014809e-06,
+      "loss": 0.1728,
+      "step": 991
+    },
+    {
+      "epoch": 4.016194331983805,
+      "grad_norm": 0.27782485731196366,
+      "learning_rate": 4.5386698155194766e-06,
+      "loss": 0.157,
+      "step": 992
+    },
+    {
+      "epoch": 4.020242914979757,
+      "grad_norm": 0.3753358915898066,
+      "learning_rate": 4.502857857942204e-06,
+      "loss": 0.1782,
+      "step": 993
+    },
+    {
+      "epoch": 4.0242914979757085,
+      "grad_norm": 0.40438601282736847,
+      "learning_rate": 4.467169815100549e-06,
+      "loss": 0.1667,
+      "step": 994
+    },
+    {
+      "epoch": 4.02834008097166,
+      "grad_norm": 0.4065869735788397,
+      "learning_rate": 4.431605972355137e-06,
+      "loss": 0.1816,
+      "step": 995
+    },
+    {
+      "epoch": 4.032388663967612,
+      "grad_norm": 0.35597501262116993,
+      "learning_rate": 4.396166614073516e-06,
+      "loss": 0.1672,
+      "step": 996
+    },
+    {
+      "epoch": 4.036437246963563,
+      "grad_norm": 0.30242787612642086,
+      "learning_rate": 4.360852023627846e-06,
+      "loss": 0.172,
+      "step": 997
+    },
+    {
+      "epoch": 4.040485829959514,
+      "grad_norm": 0.34329035746166636,
+      "learning_rate": 4.325662483392656e-06,
+      "loss": 0.1717,
+      "step": 998
+    },
+    {
+      "epoch": 4.044534412955466,
+      "grad_norm": 0.33488787002715487,
+      "learning_rate": 4.29059827474257e-06,
+      "loss": 0.178,
+      "step": 999
+    },
+    {
+      "epoch": 4.048582995951417,
+      "grad_norm": 0.3326592558819102,
+      "learning_rate": 4.255659678050063e-06,
+      "loss": 0.1833,
+      "step": 1000
+    },
+    {
+      "epoch": 4.052631578947368,
+      "grad_norm": 0.2728358577775824,
+      "learning_rate": 4.22084697268323e-06,
+      "loss": 0.1521,
+      "step": 1001
+    },
+    {
+      "epoch": 4.05668016194332,
+      "grad_norm": 0.2798357050441634,
+      "learning_rate": 4.186160437003535e-06,
+      "loss": 0.178,
+      "step": 1002
+    },
+    {
+      "epoch": 4.060728744939271,
+      "grad_norm": 0.3045689664890807,
+      "learning_rate": 4.151600348363582e-06,
+      "loss": 0.1853,
+      "step": 1003
+    },
+    {
+      "epoch": 4.064777327935222,
+      "grad_norm": 0.3039614344775456,
+      "learning_rate": 4.11716698310493e-06,
+      "loss": 0.1646,
+      "step": 1004
+    },
+    {
+      "epoch": 4.068825910931174,
+      "grad_norm": 0.29301532332231695,
+      "learning_rate": 4.082860616555848e-06,
+      "loss": 0.1819,
+      "step": 1005
+    },
+    {
+      "epoch": 4.0728744939271255,
+      "grad_norm": 0.2880290490618023,
+      "learning_rate": 4.048681523029125e-06,
+      "loss": 0.1735,
+      "step": 1006
+    },
+    {
+      "epoch": 4.076923076923077,
+      "grad_norm": 0.2691245471926778,
+      "learning_rate": 4.014629975819882e-06,
+      "loss": 0.1665,
+      "step": 1007
+    },
+    {
+      "epoch": 4.080971659919029,
+      "grad_norm": 0.2661034889100644,
+      "learning_rate": 3.9807062472033855e-06,
+      "loss": 0.1696,
+      "step": 1008
+    },
+    {
+      "epoch": 4.08502024291498,
+      "grad_norm": 0.27481192504336094,
+      "learning_rate": 3.94691060843287e-06,
+      "loss": 0.1974,
+      "step": 1009
+    },
+    {
+      "epoch": 4.089068825910931,
+      "grad_norm": 0.2843252077356889,
+      "learning_rate": 3.913243329737353e-06,
+      "loss": 0.1875,
+      "step": 1010
+    },
+    {
+      "epoch": 4.093117408906883,
+      "grad_norm": 0.27863045730785035,
+      "learning_rate": 3.879704680319503e-06,
+      "loss": 0.1796,
+      "step": 1011
+    },
+    {
+      "epoch": 4.097165991902834,
+      "grad_norm": 0.27264811516034915,
+      "learning_rate": 3.846294928353471e-06,
+      "loss": 0.1506,
+      "step": 1012
+    },
+    {
+      "epoch": 4.101214574898785,
+      "grad_norm": 0.2578662633940662,
+      "learning_rate": 3.8130143409827345e-06,
+      "loss": 0.1666,
+      "step": 1013
+    },
+    {
+      "epoch": 4.105263157894737,
+      "grad_norm": 0.272796087881526,
+      "learning_rate": 3.7798631843179887e-06,
+      "loss": 0.1674,
+      "step": 1014
+    },
+    {
+      "epoch": 4.109311740890688,
+      "grad_norm": 0.25353066668966345,
+      "learning_rate": 3.7468417234349973e-06,
+      "loss": 0.1901,
+      "step": 1015
+    },
+    {
+      "epoch": 4.113360323886639,
+      "grad_norm": 0.25520403060834823,
+      "learning_rate": 3.7139502223724823e-06,
+      "loss": 0.15,
+      "step": 1016
+    },
+    {
+      "epoch": 4.117408906882591,
+      "grad_norm": 0.25923776325345826,
+      "learning_rate": 3.6811889441300095e-06,
+      "loss": 0.1666,
+      "step": 1017
+    },
+    {
+      "epoch": 4.1214574898785425,
+      "grad_norm": 0.2572892646123473,
+      "learning_rate": 3.648558150665886e-06,
+      "loss": 0.1671,
+      "step": 1018
+    },
+    {
+      "epoch": 4.125506072874494,
+      "grad_norm": 0.25927686863597244,
+      "learning_rate": 3.616058102895075e-06,
+      "loss": 0.168,
+      "step": 1019
+    },
+    {
+      "epoch": 4.129554655870446,
+      "grad_norm": 0.2560787657821041,
+      "learning_rate": 3.583689060687079e-06,
+      "loss": 0.1543,
+      "step": 1020
+    },
+    {
+      "epoch": 4.133603238866397,
+      "grad_norm": 0.2519071371317529,
+      "learning_rate": 3.5514512828639023e-06,
+      "loss": 0.1777,
+      "step": 1021
+    },
+    {
+      "epoch": 4.137651821862348,
+      "grad_norm": 0.25727502816965253,
+      "learning_rate": 3.51934502719796e-06,
+      "loss": 0.1704,
+      "step": 1022
+    },
+    {
+      "epoch": 4.1417004048583,
+      "grad_norm": 0.2624225115355182,
+      "learning_rate": 3.4873705504100185e-06,
+      "loss": 0.1798,
+      "step": 1023
+    },
+    {
+      "epoch": 4.145748987854251,
+      "grad_norm": 0.2602399795254667,
+      "learning_rate": 3.4555281081671345e-06,
+      "loss": 0.185,
+      "step": 1024
+    },
+    {
+      "epoch": 4.149797570850202,
+      "grad_norm": 0.24886950960320312,
+      "learning_rate": 3.4238179550806326e-06,
+      "loss": 0.1724,
+      "step": 1025
+    },
+    {
+      "epoch": 4.153846153846154,
+      "grad_norm": 0.2476300230326001,
+      "learning_rate": 3.3922403447040497e-06,
+      "loss": 0.1726,
+      "step": 1026
+    },
+    {
+      "epoch": 4.157894736842105,
+      "grad_norm": 0.2522943770680571,
+      "learning_rate": 3.3607955295311247e-06,
+      "loss": 0.1819,
+      "step": 1027
+    },
+    {
+      "epoch": 4.161943319838056,
+      "grad_norm": 0.25690965470579374,
+      "learning_rate": 3.32948376099375e-06,
+      "loss": 0.1725,
+      "step": 1028
+    },
+    {
+      "epoch": 4.165991902834008,
+      "grad_norm": 0.25856725905760375,
+      "learning_rate": 3.2983052894599976e-06,
+      "loss": 0.1562,
+      "step": 1029
+    },
+    {
+      "epoch": 4.17004048582996,
+      "grad_norm": 0.24695473188888264,
+      "learning_rate": 3.2672603642320945e-06,
+      "loss": 0.1581,
+      "step": 1030
+    },
+    {
+      "epoch": 4.174089068825911,
+      "grad_norm": 0.2548573270372512,
+      "learning_rate": 3.236349233544438e-06,
+      "loss": 0.1976,
+      "step": 1031
+    },
+    {
+      "epoch": 4.178137651821863,
+      "grad_norm": 0.24382274472227033,
+      "learning_rate": 3.205572144561604e-06,
+      "loss": 0.16,
+      "step": 1032
+    },
+    {
+      "epoch": 4.182186234817814,
+      "grad_norm": 0.25222886265557176,
+      "learning_rate": 3.174929343376374e-06,
+      "loss": 0.151,
+      "step": 1033
+    },
+    {
+      "epoch": 4.186234817813765,
+      "grad_norm": 0.24823024512445585,
+      "learning_rate": 3.1444210750077777e-06,
+      "loss": 0.1727,
+      "step": 1034
+    },
+    {
+      "epoch": 4.190283400809717,
+      "grad_norm": 0.2568167154281882,
+      "learning_rate": 3.1140475833991046e-06,
+      "loss": 0.1666,
+      "step": 1035
+    },
+    {
+      "epoch": 4.194331983805668,
+      "grad_norm": 0.256178383755902,
+      "learning_rate": 3.0838091114159895e-06,
+      "loss": 0.2022,
+      "step": 1036
+    },
+    {
+      "epoch": 4.198380566801619,
+      "grad_norm": 0.25321409768620257,
+      "learning_rate": 3.0537059008444524e-06,
+      "loss": 0.1725,
+      "step": 1037
+    },
+    {
+      "epoch": 4.202429149797571,
+      "grad_norm": 0.24860769800837207,
+      "learning_rate": 3.0237381923889563e-06,
+      "loss": 0.1708,
+      "step": 1038
+    },
+    {
+      "epoch": 4.206477732793522,
+      "grad_norm": 0.2598527354082951,
+      "learning_rate": 2.9939062256704997e-06,
+      "loss": 0.1736,
+      "step": 1039
+    },
+    {
+      "epoch": 4.2105263157894735,
+      "grad_norm": 0.26153211750584593,
+      "learning_rate": 2.964210239224696e-06,
+      "loss": 0.1594,
+      "step": 1040
+    },
+    {
+      "epoch": 4.2145748987854255,
+      "grad_norm": 0.26231956715348276,
+      "learning_rate": 2.934650470499869e-06,
+      "loss": 0.1604,
+      "step": 1041
+    },
+    {
+      "epoch": 4.218623481781377,
+      "grad_norm": 0.24770883324193382,
+      "learning_rate": 2.905227155855128e-06,
+      "loss": 0.166,
+      "step": 1042
+    },
+    {
+      "epoch": 4.222672064777328,
+      "grad_norm": 0.2556327594220709,
+      "learning_rate": 2.875940530558519e-06,
+      "loss": 0.1818,
+      "step": 1043
+    },
+    {
+      "epoch": 4.22672064777328,
+      "grad_norm": 0.2575151033370932,
+      "learning_rate": 2.8467908287851154e-06,
+      "loss": 0.1764,
+      "step": 1044
+    },
+    {
+      "epoch": 4.230769230769231,
+      "grad_norm": 0.260639567038207,
+      "learning_rate": 2.8177782836151603e-06,
+      "loss": 0.184,
+      "step": 1045
+    },
+    {
+      "epoch": 4.234817813765182,
+      "grad_norm": 0.2506828162270049,
+      "learning_rate": 2.7889031270321874e-06,
+      "loss": 0.1729,
+      "step": 1046
+    },
+    {
+      "epoch": 4.238866396761134,
+      "grad_norm": 0.254801254438143,
+      "learning_rate": 2.760165589921184e-06,
+      "loss": 0.1597,
+      "step": 1047
+    },
+    {
+      "epoch": 4.242914979757085,
+      "grad_norm": 0.24733829085673442,
+      "learning_rate": 2.7315659020667305e-06,
+      "loss": 0.1613,
+      "step": 1048
+    },
+    {
+      "epoch": 4.246963562753036,
+      "grad_norm": 0.2529785437191398,
+      "learning_rate": 2.703104292151164e-06,
+      "loss": 0.1831,
+      "step": 1049
+    },
+    {
+      "epoch": 4.251012145748988,
+      "grad_norm": 0.25605272074978175,
+      "learning_rate": 2.6747809877527607e-06,
+      "loss": 0.1601,
+      "step": 1050
+    },
+    {
+      "epoch": 4.255060728744939,
+      "grad_norm": 0.2500053045926688,
+      "learning_rate": 2.6465962153439082e-06,
+      "loss": 0.182,
+      "step": 1051
+    },
+    {
+      "epoch": 4.2591093117408905,
+      "grad_norm": 0.24766164111738012,
+      "learning_rate": 2.618550200289303e-06,
+      "loss": 0.1843,
+      "step": 1052
+    },
+    {
+      "epoch": 4.2631578947368425,
+      "grad_norm": 0.24666440104922702,
+      "learning_rate": 2.59064316684412e-06,
+      "loss": 0.1746,
+      "step": 1053
+    },
+    {
+      "epoch": 4.267206477732794,
+      "grad_norm": 0.24390261972847346,
+      "learning_rate": 2.562875338152264e-06,
+      "loss": 0.156,
+      "step": 1054
+    },
+    {
+      "epoch": 4.271255060728745,
+      "grad_norm": 0.24731557504116172,
+      "learning_rate": 2.5352469362445552e-06,
+      "loss": 0.1827,
+      "step": 1055
+    },
+    {
+      "epoch": 4.275303643724697,
+      "grad_norm": 0.2500336186431661,
+      "learning_rate": 2.5077581820369523e-06,
+      "loss": 0.1761,
+      "step": 1056
+    },
+    {
+      "epoch": 4.279352226720648,
+      "grad_norm": 0.24985799657464844,
+      "learning_rate": 2.480409295328803e-06,
+      "loss": 0.1813,
+      "step": 1057
+    },
+    {
+      "epoch": 4.283400809716599,
+      "grad_norm": 0.24557456736378738,
+      "learning_rate": 2.45320049480108e-06,
+      "loss": 0.171,
+      "step": 1058
+    },
+    {
+      "epoch": 4.287449392712551,
+      "grad_norm": 0.25159571069119796,
+      "learning_rate": 2.4261319980146293e-06,
+      "loss": 0.1867,
+      "step": 1059
+    },
+    {
+      "epoch": 4.291497975708502,
+      "grad_norm": 0.25724463806018455,
+      "learning_rate": 2.399204021408421e-06,
+      "loss": 0.1529,
+      "step": 1060
+    },
+    {
+      "epoch": 4.295546558704453,
+      "grad_norm": 0.2538488302804189,
+      "learning_rate": 2.3724167802978395e-06,
+      "loss": 0.1794,
+      "step": 1061
+    },
+    {
+      "epoch": 4.299595141700405,
+      "grad_norm": 0.2578596367045205,
+      "learning_rate": 2.3457704888729625e-06,
+      "loss": 0.1615,
+      "step": 1062
+    },
+    {
+      "epoch": 4.303643724696356,
+      "grad_norm": 0.2513684246319179,
+      "learning_rate": 2.3192653601968162e-06,
+      "loss": 0.1646,
+      "step": 1063
+    },
+    {
+      "epoch": 4.3076923076923075,
+      "grad_norm": 0.2553189976998563,
+      "learning_rate": 2.2929016062037057e-06,
+      "loss": 0.169,
+      "step": 1064
+    },
+    {
+      "epoch": 4.3117408906882595,
+      "grad_norm": 0.23903900510230583,
+      "learning_rate": 2.2666794376975055e-06,
+      "loss": 0.1736,
+      "step": 1065
+    },
+    {
+      "epoch": 4.315789473684211,
+      "grad_norm": 0.24991775047447365,
+      "learning_rate": 2.24059906434998e-06,
+      "loss": 0.192,
+      "step": 1066
+    },
+    {
+      "epoch": 4.319838056680162,
+      "grad_norm": 0.24837734110862064,
+      "learning_rate": 2.214660694699091e-06,
+      "loss": 0.1376,
+      "step": 1067
+    },
+    {
+      "epoch": 4.323886639676114,
+      "grad_norm": 0.2563611615622322,
+      "learning_rate": 2.188864536147355e-06,
+      "loss": 0.1579,
+      "step": 1068
+    },
+    {
+      "epoch": 4.327935222672065,
+      "grad_norm": 0.2597202708928397,
+      "learning_rate": 2.1632107949601757e-06,
+      "loss": 0.1626,
+      "step": 1069
+    },
+    {
+      "epoch": 4.331983805668016,
+      "grad_norm": 0.248075273739568,
+      "learning_rate": 2.137699676264171e-06,
+      "loss": 0.1773,
+      "step": 1070
+    },
+    {
+      "epoch": 4.336032388663968,
+      "grad_norm": 0.24708238104447408,
+      "learning_rate": 2.1123313840455738e-06,
+      "loss": 0.1861,
+      "step": 1071
+    },
+    {
+      "epoch": 4.340080971659919,
+      "grad_norm": 0.24375930765675918,
+      "learning_rate": 2.08710612114857e-06,
+      "loss": 0.16,
+      "step": 1072
+    },
+    {
+      "epoch": 4.34412955465587,
+      "grad_norm": 0.24752451220992744,
+      "learning_rate": 2.0620240892736953e-06,
+      "loss": 0.1746,
+      "step": 1073
+    },
+    {
+      "epoch": 4.348178137651822,
+      "grad_norm": 0.2531519601019274,
+      "learning_rate": 2.0370854889762e-06,
+      "loss": 0.1756,
+      "step": 1074
+    },
+    {
+      "epoch": 4.352226720647773,
+      "grad_norm": 0.2480315388726587,
+      "learning_rate": 2.0122905196644726e-06,
+      "loss": 0.1536,
+      "step": 1075
+    },
+    {
+      "epoch": 4.3562753036437245,
+      "grad_norm": 0.25330790337862163,
+      "learning_rate": 1.987639379598427e-06,
+      "loss": 0.1505,
+      "step": 1076
+    },
+    {
+      "epoch": 4.3603238866396765,
+      "grad_norm": 0.24603131365830844,
+      "learning_rate": 1.963132265887919e-06,
+      "loss": 0.1781,
+      "step": 1077
+    },
+    {
+      "epoch": 4.364372469635628,
+      "grad_norm": 0.2478677476051414,
+      "learning_rate": 1.9387693744911784e-06,
+      "loss": 0.1842,
+      "step": 1078
+    },
+    {
+      "epoch": 4.368421052631579,
+      "grad_norm": 0.263166161396794,
+      "learning_rate": 1.9145509002132368e-06,
+      "loss": 0.192,
+      "step": 1079
+    },
+    {
+      "epoch": 4.372469635627531,
+      "grad_norm": 0.24252378739917496,
+      "learning_rate": 1.8904770367043723e-06,
+      "loss": 0.1744,
+      "step": 1080
+    },
+    {
+      "epoch": 4.376518218623482,
+      "grad_norm": 0.2566390930600202,
+      "learning_rate": 1.8665479764585504e-06,
+      "loss": 0.1476,
+      "step": 1081
+    },
+    {
+      "epoch": 4.380566801619433,
+      "grad_norm": 0.2598200426819249,
+      "learning_rate": 1.8427639108118977e-06,
+      "loss": 0.1781,
+      "step": 1082
+    },
+    {
+      "epoch": 4.384615384615385,
+      "grad_norm": 0.26563376176376124,
+      "learning_rate": 1.8191250299411755e-06,
+      "loss": 0.1784,
+      "step": 1083
+    },
+    {
+      "epoch": 4.388663967611336,
+      "grad_norm": 0.25471424791299496,
+      "learning_rate": 1.7956315228622446e-06,
+      "loss": 0.166,
+      "step": 1084
+    },
+    {
+      "epoch": 4.392712550607287,
+      "grad_norm": 0.24596408793480246,
+      "learning_rate": 1.7722835774285553e-06,
+      "loss": 0.1852,
+      "step": 1085
+    },
+    {
+      "epoch": 4.396761133603239,
+      "grad_norm": 0.24005339315261004,
+      "learning_rate": 1.7490813803296624e-06,
+      "loss": 0.1435,
+      "step": 1086
+    },
+    {
+      "epoch": 4.40080971659919,
+      "grad_norm": 0.25545092566778277,
+      "learning_rate": 1.726025117089718e-06,
+      "loss": 0.1797,
+      "step": 1087
+    },
+    {
+      "epoch": 4.4048582995951415,
+      "grad_norm": 0.25066039352733027,
+      "learning_rate": 1.70311497206598e-06,
+      "loss": 0.1682,
+      "step": 1088
+    },
+    {
+      "epoch": 4.4089068825910935,
+      "grad_norm": 0.2566349591219324,
+      "learning_rate": 1.6803511284473638e-06,
+      "loss": 0.1496,
+      "step": 1089
+    },
+    {
+      "epoch": 4.412955465587045,
+      "grad_norm": 0.25688248089275467,
+      "learning_rate": 1.6577337682529538e-06,
+      "loss": 0.1559,
+      "step": 1090
+    },
+    {
+      "epoch": 4.417004048582996,
+      "grad_norm": 0.24917065804186586,
+      "learning_rate": 1.6352630723305617e-06,
+      "loss": 0.1489,
+      "step": 1091
+    },
+    {
+      "epoch": 4.421052631578947,
+      "grad_norm": 0.2580567276849716,
+      "learning_rate": 1.612939220355272e-06,
+      "loss": 0.1898,
+      "step": 1092
+    },
+    {
+      "epoch": 4.425101214574899,
+      "grad_norm": 0.2329310250089003,
+      "learning_rate": 1.5907623908280112e-06,
+      "loss": 0.1563,
+      "step": 1093
+    },
+    {
+      "epoch": 4.42914979757085,
+      "grad_norm": 0.24949079893760656,
+      "learning_rate": 1.568732761074121e-06,
+      "loss": 0.1698,
+      "step": 1094
+    },
+    {
+      "epoch": 4.433198380566802,
+      "grad_norm": 0.26562461739878984,
+      "learning_rate": 1.5468505072419237e-06,
+      "loss": 0.1722,
+      "step": 1095
+    },
+    {
+      "epoch": 4.437246963562753,
+      "grad_norm": 0.2442786718226581,
+      "learning_rate": 1.5251158043013448e-06,
+      "loss": 0.1775,
+      "step": 1096
+    },
+    {
+      "epoch": 4.441295546558704,
+      "grad_norm": 0.24697493515117527,
+      "learning_rate": 1.5035288260424885e-06,
+      "loss": 0.1702,
+      "step": 1097
+    },
+    {
+      "epoch": 4.445344129554655,
+      "grad_norm": 0.2480901007179765,
+      "learning_rate": 1.4820897450742623e-06,
+      "loss": 0.1512,
+      "step": 1098
+    },
+    {
+      "epoch": 4.449392712550607,
+      "grad_norm": 0.2507890922615234,
+      "learning_rate": 1.4607987328229767e-06,
+      "loss": 0.1963,
+      "step": 1099
+    },
+    {
+      "epoch": 4.4534412955465585,
+      "grad_norm": 0.25707423519159345,
+      "learning_rate": 1.4396559595310056e-06,
+      "loss": 0.1687,
+      "step": 1100
+    },
+    {
+      "epoch": 4.4574898785425106,
+      "grad_norm": 0.25604638690728987,
+      "learning_rate": 1.4186615942553973e-06,
+      "loss": 0.1645,
+      "step": 1101
+    },
+    {
+      "epoch": 4.461538461538462,
+      "grad_norm": 0.25335299739726364,
+      "learning_rate": 1.3978158048665424e-06,
+      "loss": 0.1641,
+      "step": 1102
+    },
+    {
+      "epoch": 4.465587044534413,
+      "grad_norm": 0.2415573603320239,
+      "learning_rate": 1.3771187580468115e-06,
+      "loss": 0.1542,
+      "step": 1103
+    },
+    {
+      "epoch": 4.469635627530364,
+      "grad_norm": 0.2585748470305878,
+      "learning_rate": 1.3565706192892392e-06,
+      "loss": 0.1698,
+      "step": 1104
+    },
+    {
+      "epoch": 4.473684210526316,
+      "grad_norm": 0.2402583435394757,
+      "learning_rate": 1.3361715528961993e-06,
+      "loss": 0.1589,
+      "step": 1105
+    },
+    {
+      "epoch": 4.477732793522267,
+      "grad_norm": 0.25556888458065147,
+      "learning_rate": 1.3159217219780817e-06,
+      "loss": 0.1607,
+      "step": 1106
+    },
+    {
+      "epoch": 4.481781376518219,
+      "grad_norm": 0.24929646907107691,
+      "learning_rate": 1.2958212884519949e-06,
+      "loss": 0.162,
+      "step": 1107
+    },
+    {
+      "epoch": 4.48582995951417,
+      "grad_norm": 0.25043086488777994,
+      "learning_rate": 1.2758704130404675e-06,
+      "loss": 0.1781,
+      "step": 1108
+    },
+    {
+      "epoch": 4.489878542510121,
+      "grad_norm": 0.25373466138977313,
+      "learning_rate": 1.2560692552701738e-06,
+      "loss": 0.1504,
+      "step": 1109
+    },
+    {
+      "epoch": 4.493927125506072,
+      "grad_norm": 0.2560088443042012,
+      "learning_rate": 1.2364179734706382e-06,
+      "loss": 0.1923,
+      "step": 1110
+    },
+    {
+      "epoch": 4.497975708502024,
+      "grad_norm": 0.24923988445156017,
+      "learning_rate": 1.2169167247729852e-06,
+      "loss": 0.1759,
+      "step": 1111
+    },
+    {
+      "epoch": 4.502024291497976,
+      "grad_norm": 0.24044493060508407,
+      "learning_rate": 1.1975656651086864e-06,
+      "loss": 0.1624,
+      "step": 1112
+    },
+    {
+      "epoch": 4.506072874493928,
+      "grad_norm": 0.2434762086745541,
+      "learning_rate": 1.1783649492082926e-06,
+      "loss": 0.1669,
+      "step": 1113
+    },
+    {
+      "epoch": 4.510121457489879,
+      "grad_norm": 0.24155307328060283,
+      "learning_rate": 1.1593147306002183e-06,
+      "loss": 0.1712,
+      "step": 1114
+    },
+    {
+      "epoch": 4.51417004048583,
+      "grad_norm": 0.25441211785782386,
+      "learning_rate": 1.140415161609505e-06,
+      "loss": 0.1773,
+      "step": 1115
+    },
+    {
+      "epoch": 4.518218623481781,
+      "grad_norm": 0.24650391844074135,
+      "learning_rate": 1.121666393356604e-06,
+      "loss": 0.1704,
+      "step": 1116
+    },
+    {
+      "epoch": 4.522267206477733,
+      "grad_norm": 0.2533164255225416,
+      "learning_rate": 1.1030685757561632e-06,
+      "loss": 0.1686,
+      "step": 1117
+    },
+    {
+      "epoch": 4.526315789473684,
+      "grad_norm": 0.2567408168967213,
+      "learning_rate": 1.0846218575158373e-06,
+      "loss": 0.1673,
+      "step": 1118
+    },
+    {
+      "epoch": 4.530364372469636,
+      "grad_norm": 0.23641532238753146,
+      "learning_rate": 1.066326386135097e-06,
+      "loss": 0.1468,
+      "step": 1119
+    },
+    {
+      "epoch": 4.534412955465587,
+      "grad_norm": 0.26085044165336374,
+      "learning_rate": 1.0481823079040398e-06,
+      "loss": 0.1735,
+      "step": 1120
+    },
+    {
+      "epoch": 4.538461538461538,
+      "grad_norm": 0.24827009641164954,
+      "learning_rate": 1.0301897679022344e-06,
+      "loss": 0.1798,
+      "step": 1121
+    },
+    {
+      "epoch": 4.5425101214574894,
+      "grad_norm": 0.2521454383471447,
+      "learning_rate": 1.0123489099975491e-06,
+      "loss": 0.1561,
+      "step": 1122
+    },
+    {
+      "epoch": 4.5465587044534415,
+      "grad_norm": 0.26219019886749806,
+      "learning_rate": 9.9465987684501e-07,
+      "loss": 0.1645,
+      "step": 1123
+    },
+    {
+      "epoch": 4.550607287449393,
+      "grad_norm": 0.24512783188093906,
+      "learning_rate": 9.771228098856534e-07,
+      "loss": 0.1697,
+      "step": 1124
+    },
+    {
+      "epoch": 4.554655870445345,
+      "grad_norm": 0.23972008327476343,
+      "learning_rate": 9.597378493454013e-07,
+      "loss": 0.1725,
+      "step": 1125
+    },
+    {
+      "epoch": 4.558704453441296,
+      "grad_norm": 0.2504393152256871,
+      "learning_rate": 9.425051342339353e-07,
+      "loss": 0.1748,
+      "step": 1126
+    },
+    {
+      "epoch": 4.562753036437247,
+      "grad_norm": 0.28801267258974866,
+      "learning_rate": 9.254248023435864e-07,
+      "loss": 0.1787,
+      "step": 1127
+    },
+    {
+      "epoch": 4.566801619433198,
+      "grad_norm": 0.24633999709608345,
+      "learning_rate": 9.084969902482288e-07,
+      "loss": 0.1609,
+      "step": 1128
+    },
+    {
+      "epoch": 4.57085020242915,
+      "grad_norm": 0.2455747913100564,
+      "learning_rate": 8.917218333021993e-07,
+      "loss": 0.1919,
+      "step": 1129
+    },
+    {
+      "epoch": 4.574898785425101,
+      "grad_norm": 0.2382438760702198,
+      "learning_rate": 8.750994656392109e-07,
+      "loss": 0.1704,
+      "step": 1130
+    },
+    {
+      "epoch": 4.578947368421053,
+      "grad_norm": 0.23993588280237213,
+      "learning_rate": 8.58630020171265e-07,
+      "loss": 0.1773,
+      "step": 1131
+    },
+    {
+      "epoch": 4.582995951417004,
+      "grad_norm": 0.2533712583483457,
+      "learning_rate": 8.423136285876099e-07,
+      "loss": 0.1746,
+      "step": 1132
+    },
+    {
+      "epoch": 4.587044534412955,
+      "grad_norm": 0.24471317080133106,
+      "learning_rate": 8.26150421353682e-07,
+      "loss": 0.1696,
+      "step": 1133
+    },
+    {
+      "epoch": 4.5910931174089065,
+      "grad_norm": 0.2504558780828328,
+      "learning_rate": 8.101405277100549e-07,
+      "loss": 0.1736,
+      "step": 1134
+    },
+    {
+      "epoch": 4.5951417004048585,
+      "grad_norm": 0.23847248566364548,
+      "learning_rate": 7.942840756714077e-07,
+      "loss": 0.1579,
+      "step": 1135
+    },
+    {
+      "epoch": 4.59919028340081,
+      "grad_norm": 0.2456733677299304,
+      "learning_rate": 7.785811920255093e-07,
+      "loss": 0.1769,
+      "step": 1136
+    },
+    {
+      "epoch": 4.603238866396762,
+      "grad_norm": 0.25049540162307404,
+      "learning_rate": 7.630320023322024e-07,
+      "loss": 0.1655,
+      "step": 1137
+    },
+    {
+      "epoch": 4.607287449392713,
+      "grad_norm": 0.2596492854366612,
+      "learning_rate": 7.476366309223903e-07,
+      "loss": 0.167,
+      "step": 1138
+    },
+    {
+      "epoch": 4.611336032388664,
+      "grad_norm": 0.25158052851062673,
+      "learning_rate": 7.323952008970537e-07,
+      "loss": 0.1712,
+      "step": 1139
+    },
+    {
+      "epoch": 4.615384615384615,
+      "grad_norm": 0.24744013934140313,
+      "learning_rate": 7.173078341262618e-07,
+      "loss": 0.14,
+      "step": 1140
+    },
+    {
+      "epoch": 4.619433198380567,
+      "grad_norm": 0.24583883933126802,
+      "learning_rate": 7.023746512481989e-07,
+      "loss": 0.1803,
+      "step": 1141
+    },
+    {
+      "epoch": 4.623481781376518,
+      "grad_norm": 0.2459401272722852,
+      "learning_rate": 6.875957716681902e-07,
+      "loss": 0.1874,
+      "step": 1142
+    },
+    {
+      "epoch": 4.62753036437247,
+      "grad_norm": 0.24531174670382513,
+      "learning_rate": 6.729713135577687e-07,
+      "loss": 0.1549,
+      "step": 1143
+    },
+    {
+      "epoch": 4.631578947368421,
+      "grad_norm": 0.2521625661367056,
+      "learning_rate": 6.585013938537143e-07,
+      "loss": 0.1734,
+      "step": 1144
+    },
+    {
+      "epoch": 4.635627530364372,
+      "grad_norm": 0.23893465769438188,
+      "learning_rate": 6.441861282571115e-07,
+      "loss": 0.1515,
+      "step": 1145
+    },
+    {
+      "epoch": 4.6396761133603235,
+      "grad_norm": 0.23748729527337437,
+      "learning_rate": 6.300256312324471e-07,
+      "loss": 0.1856,
+      "step": 1146
+    },
+    {
+      "epoch": 4.6437246963562755,
+      "grad_norm": 0.24903693904261298,
+      "learning_rate": 6.160200160066776e-07,
+      "loss": 0.1802,
+      "step": 1147
+    },
+    {
+      "epoch": 4.647773279352227,
+      "grad_norm": 0.2675978531612752,
+      "learning_rate": 6.021693945683327e-07,
+      "loss": 0.1925,
+      "step": 1148
+    },
+    {
+      "epoch": 4.651821862348179,
+      "grad_norm": 0.25298013723945734,
+      "learning_rate": 5.884738776666044e-07,
+      "loss": 0.159,
+      "step": 1149
+    },
+    {
+      "epoch": 4.65587044534413,
+      "grad_norm": 0.2556153522190733,
+      "learning_rate": 5.749335748104878e-07,
+      "loss": 0.1782,
+      "step": 1150
+    },
+    {
+      "epoch": 4.659919028340081,
+      "grad_norm": 0.24480976715582722,
+      "learning_rate": 5.615485942678778e-07,
+      "loss": 0.1735,
+      "step": 1151
+    },
+    {
+      "epoch": 4.663967611336032,
+      "grad_norm": 0.24831359588929278,
+      "learning_rate": 5.483190430647289e-07,
+      "loss": 0.1793,
+      "step": 1152
+    },
+    {
+      "epoch": 4.668016194331984,
+      "grad_norm": 0.24846109638005592,
+      "learning_rate": 5.352450269841747e-07,
+      "loss": 0.1699,
+      "step": 1153
+    },
+    {
+      "epoch": 4.672064777327935,
+      "grad_norm": 0.2501084265756062,
+      "learning_rate": 5.22326650565701e-07,
+      "loss": 0.1528,
+      "step": 1154
+    },
+    {
+      "epoch": 4.676113360323887,
+      "grad_norm": 0.24541337824543116,
+      "learning_rate": 5.095640171043048e-07,
+      "loss": 0.1813,
+      "step": 1155
+    },
+    {
+      "epoch": 4.680161943319838,
+      "grad_norm": 0.2419339422086418,
+      "learning_rate": 4.969572286496594e-07,
+      "loss": 0.1555,
+      "step": 1156
+    },
+    {
+      "epoch": 4.684210526315789,
+      "grad_norm": 0.23796476599002286,
+      "learning_rate": 4.845063860053079e-07,
+      "loss": 0.2083,
+      "step": 1157
+    },
+    {
+      "epoch": 4.6882591093117405,
+      "grad_norm": 0.25228514258173185,
+      "learning_rate": 4.7221158872785997e-07,
+      "loss": 0.1557,
+      "step": 1158
+    },
+    {
+      "epoch": 4.6923076923076925,
+      "grad_norm": 0.2500806468654822,
+      "learning_rate": 4.600729351261857e-07,
+      "loss": 0.2016,
+      "step": 1159
+    },
+    {
+      "epoch": 4.696356275303644,
+      "grad_norm": 0.24304838759010575,
+      "learning_rate": 4.4809052226063355e-07,
+      "loss": 0.1557,
+      "step": 1160
+    },
+    {
+      "epoch": 4.700404858299595,
+      "grad_norm": 0.2512947285092783,
+      "learning_rate": 4.3626444594226046e-07,
+      "loss": 0.1655,
+      "step": 1161
+    },
+    {
+      "epoch": 4.704453441295547,
+      "grad_norm": 0.24695962592780138,
+      "learning_rate": 4.245948007320588e-07,
+      "loss": 0.1809,
+      "step": 1162
+    },
+    {
+      "epoch": 4.708502024291498,
+      "grad_norm": 0.234465551026101,
+      "learning_rate": 4.130816799401971e-07,
+      "loss": 0.1978,
+      "step": 1163
+    },
+    {
+      "epoch": 4.712550607287449,
+      "grad_norm": 0.25429510860876253,
+      "learning_rate": 4.017251756252827e-07,
+      "loss": 0.1787,
+      "step": 1164
+    },
+    {
+      "epoch": 4.716599190283401,
+      "grad_norm": 0.2622920540457299,
+      "learning_rate": 3.9052537859362473e-07,
+      "loss": 0.1862,
+      "step": 1165
+    },
+    {
+      "epoch": 4.720647773279352,
+      "grad_norm": 0.2504575332130552,
+      "learning_rate": 3.7948237839849687e-07,
+      "loss": 0.182,
+      "step": 1166
+    },
+    {
+      "epoch": 4.724696356275303,
+      "grad_norm": 0.2519877370770394,
+      "learning_rate": 3.6859626333943797e-07,
+      "loss": 0.1618,
+      "step": 1167
+    },
+    {
+      "epoch": 4.728744939271255,
+      "grad_norm": 0.24162333717898604,
+      "learning_rate": 3.578671204615347e-07,
+      "loss": 0.1685,
+      "step": 1168
+    },
+    {
+      "epoch": 4.732793522267206,
+      "grad_norm": 0.2535835344303502,
+      "learning_rate": 3.472950355547267e-07,
+      "loss": 0.1642,
+      "step": 1169
+    },
+    {
+      "epoch": 4.7368421052631575,
+      "grad_norm": 0.2454742872922908,
+      "learning_rate": 3.3688009315312244e-07,
+      "loss": 0.1786,
+      "step": 1170
+    },
+    {
+      "epoch": 4.7408906882591095,
+      "grad_norm": 0.24985570454840672,
+      "learning_rate": 3.266223765343224e-07,
+      "loss": 0.1599,
+      "step": 1171
+    },
+    {
+      "epoch": 4.744939271255061,
+      "grad_norm": 0.2645202474013409,
+      "learning_rate": 3.16521967718757e-07,
+      "loss": 0.1809,
+      "step": 1172
+    },
+    {
+      "epoch": 4.748987854251012,
+      "grad_norm": 0.25015733257531036,
+      "learning_rate": 3.0657894746902726e-07,
+      "loss": 0.1752,
+      "step": 1173
+    },
+    {
+      "epoch": 4.753036437246964,
+      "grad_norm": 0.24617168051151114,
+      "learning_rate": 2.9679339528925654e-07,
+      "loss": 0.147,
+      "step": 1174
+    },
+    {
+      "epoch": 4.757085020242915,
+      "grad_norm": 0.24306120603124212,
+      "learning_rate": 2.871653894244597e-07,
+      "loss": 0.1674,
+      "step": 1175
+    },
+    {
+      "epoch": 4.761133603238866,
+      "grad_norm": 0.2539360426219087,
+      "learning_rate": 2.7769500685991266e-07,
+      "loss": 0.1601,
+      "step": 1176
+    },
+    {
+      "epoch": 4.765182186234818,
+      "grad_norm": 0.24956762392261314,
+      "learning_rate": 2.683823233205485e-07,
+      "loss": 0.1669,
+      "step": 1177
+    },
+    {
+      "epoch": 4.769230769230769,
+      "grad_norm": 0.2419935668417277,
+      "learning_rate": 2.592274132703287e-07,
+      "loss": 0.1569,
+      "step": 1178
+    },
+    {
+      "epoch": 4.77327935222672,
+      "grad_norm": 0.25311305853323435,
+      "learning_rate": 2.502303499116754e-07,
+      "loss": 0.177,
+      "step": 1179
+    },
+    {
+      "epoch": 4.777327935222672,
+      "grad_norm": 0.23998730854352884,
+      "learning_rate": 2.413912051848666e-07,
+      "loss": 0.1437,
+      "step": 1180
+    },
+    {
+      "epoch": 4.781376518218623,
+      "grad_norm": 0.25831339581395346,
+      "learning_rate": 2.3271004976746613e-07,
+      "loss": 0.1749,
+      "step": 1181
+    },
+    {
+      "epoch": 4.7854251012145745,
+      "grad_norm": 0.24713590869510593,
+      "learning_rate": 2.2418695307376393e-07,
+      "loss": 0.1431,
+      "step": 1182
+    },
+    {
+      "epoch": 4.7894736842105265,
+      "grad_norm": 0.2551679626161612,
+      "learning_rate": 2.1582198325420968e-07,
+      "loss": 0.159,
+      "step": 1183
+    },
+    {
+      "epoch": 4.793522267206478,
+      "grad_norm": 0.25827530939770504,
+      "learning_rate": 2.0761520719488447e-07,
+      "loss": 0.1669,
+      "step": 1184
+    },
+    {
+      "epoch": 4.797570850202429,
+      "grad_norm": 0.2529481419128299,
+      "learning_rate": 1.9956669051694798e-07,
+      "loss": 0.1614,
+      "step": 1185
+    },
+    {
+      "epoch": 4.801619433198381,
+      "grad_norm": 0.2537248742340826,
+      "learning_rate": 1.916764975761254e-07,
+      "loss": 0.1787,
+      "step": 1186
+    },
+    {
+      "epoch": 4.805668016194332,
+      "grad_norm": 0.255055046848462,
+      "learning_rate": 1.839446914621923e-07,
+      "loss": 0.1811,
+      "step": 1187
+    },
+    {
+      "epoch": 4.809716599190283,
+      "grad_norm": 0.24151351293929146,
+      "learning_rate": 1.7637133399845962e-07,
+      "loss": 0.1564,
+      "step": 1188
+    },
+    {
+      "epoch": 4.813765182186235,
+      "grad_norm": 0.23254673001004045,
+      "learning_rate": 1.6895648574129397e-07,
+      "loss": 0.1708,
+      "step": 1189
+    },
+    {
+      "epoch": 4.817813765182186,
+      "grad_norm": 0.2462271755777369,
+      "learning_rate": 1.6170020597962465e-07,
+      "loss": 0.159,
+      "step": 1190
+    },
+    {
+      "epoch": 4.821862348178137,
+      "grad_norm": 0.24792974575698146,
+      "learning_rate": 1.5460255273447078e-07,
+      "loss": 0.1614,
+      "step": 1191
+    },
+    {
+      "epoch": 4.825910931174089,
+      "grad_norm": 0.25510510030989453,
+      "learning_rate": 1.4766358275847715e-07,
+      "loss": 0.1716,
+      "step": 1192
+    },
+    {
+      "epoch": 4.82995951417004,
+      "grad_norm": 0.254606051492874,
+      "learning_rate": 1.4088335153546573e-07,
+      "loss": 0.1788,
+      "step": 1193
+    },
+    {
+      "epoch": 4.834008097165992,
+      "grad_norm": 0.2457615521267218,
+      "learning_rate": 1.3426191327998496e-07,
+      "loss": 0.1701,
+      "step": 1194
+    },
+    {
+      "epoch": 4.838056680161944,
+      "grad_norm": 0.23461277480676251,
+      "learning_rate": 1.277993209368744e-07,
+      "loss": 0.1833,
+      "step": 1195
+    },
+    {
+      "epoch": 4.842105263157895,
+      "grad_norm": 0.24274875483164615,
+      "learning_rate": 1.214956261808542e-07,
+      "loss": 0.1577,
+      "step": 1196
+    },
+    {
+      "epoch": 4.846153846153846,
+      "grad_norm": 0.26384328655455797,
+      "learning_rate": 1.1535087941609846e-07,
+      "loss": 0.1717,
+      "step": 1197
+    },
+    {
+      "epoch": 4.850202429149798,
+      "grad_norm": 0.24948763350743308,
+      "learning_rate": 1.0936512977583802e-07,
+      "loss": 0.1664,
+      "step": 1198
+    },
+    {
+      "epoch": 4.854251012145749,
+      "grad_norm": 0.25500085025527885,
+      "learning_rate": 1.0353842512196732e-07,
+      "loss": 0.1632,
+      "step": 1199
+    },
+    {
+      "epoch": 4.8582995951417,
+      "grad_norm": 0.242634273976212,
+      "learning_rate": 9.787081204466253e-08,
+      "loss": 0.1671,
+      "step": 1200
+    },
+    {
+      "epoch": 4.862348178137652,
+      "grad_norm": 0.2475137038133676,
+      "learning_rate": 9.236233586200627e-08,
+      "loss": 0.1655,
+      "step": 1201
+    },
+    {
+      "epoch": 4.866396761133603,
+      "grad_norm": 0.26115535736254347,
+      "learning_rate": 8.70130406196279e-08,
+      "loss": 0.1651,
+      "step": 1202
+    },
+    {
+      "epoch": 4.870445344129554,
+      "grad_norm": 0.25150358244093074,
+      "learning_rate": 8.182296909035047e-08,
+      "loss": 0.1522,
+      "step": 1203
+    },
+    {
+      "epoch": 4.874493927125506,
+      "grad_norm": 0.25264655804358593,
+      "learning_rate": 7.679216277384439e-08,
+      "loss": 0.1629,
+      "step": 1204
+    },
+    {
+      "epoch": 4.8785425101214575,
+      "grad_norm": 0.25518421398878166,
+      "learning_rate": 7.19206618963053e-08,
+      "loss": 0.169,
+      "step": 1205
+    },
+    {
+      "epoch": 4.882591093117409,
+      "grad_norm": 0.24706748409599627,
+      "learning_rate": 6.720850541012347e-08,
+      "loss": 0.1756,
+      "step": 1206
+    },
+    {
+      "epoch": 4.886639676113361,
+      "grad_norm": 0.24661991398785668,
+      "learning_rate": 6.265573099357714e-08,
+      "loss": 0.1648,
+      "step": 1207
+    },
+    {
+      "epoch": 4.890688259109312,
+      "grad_norm": 0.24401841305895533,
+      "learning_rate": 5.826237505053067e-08,
+      "loss": 0.1741,
+      "step": 1208
+    },
+    {
+      "epoch": 4.894736842105263,
+      "grad_norm": 0.2444505268872435,
+      "learning_rate": 5.4028472710137e-08,
+      "loss": 0.1697,
+      "step": 1209
+    },
+    {
+      "epoch": 4.898785425101215,
+      "grad_norm": 0.24186343118974954,
+      "learning_rate": 4.9954057826571146e-08,
+      "loss": 0.1691,
+      "step": 1210
+    },
+    {
+      "epoch": 4.902834008097166,
+      "grad_norm": 0.245111498488646,
+      "learning_rate": 4.6039162978743776e-08,
+      "loss": 0.1779,
+      "step": 1211
+    },
+    {
+      "epoch": 4.906882591093117,
+      "grad_norm": 0.25161241737577295,
+      "learning_rate": 4.228381947004812e-08,
+      "loss": 0.1387,
+      "step": 1212
+    },
+    {
+      "epoch": 4.910931174089069,
+      "grad_norm": 0.2584020834390518,
+      "learning_rate": 3.868805732811343e-08,
+      "loss": 0.1721,
+      "step": 1213
+    },
+    {
+      "epoch": 4.91497975708502,
+      "grad_norm": 0.24811434081921543,
+      "learning_rate": 3.525190530455636e-08,
+      "loss": 0.178,
+      "step": 1214
+    },
+    {
+      "epoch": 4.919028340080971,
+      "grad_norm": 0.25626398090374947,
+      "learning_rate": 3.197539087475887e-08,
+      "loss": 0.1986,
+      "step": 1215
+    },
+    {
+      "epoch": 4.923076923076923,
+      "grad_norm": 0.25154722740561647,
+      "learning_rate": 2.8858540237641786e-08,
+      "loss": 0.1861,
+      "step": 1216
+    },
+    {
+      "epoch": 4.9271255060728745,
+      "grad_norm": 0.24833042234760128,
+      "learning_rate": 2.5901378315458247e-08,
+      "loss": 0.1597,
+      "step": 1217
+    },
+    {
+      "epoch": 4.931174089068826,
+      "grad_norm": 0.24549947707501257,
+      "learning_rate": 2.3103928753598347e-08,
+      "loss": 0.1766,
+      "step": 1218
+    },
+    {
+      "epoch": 4.935222672064778,
+      "grad_norm": 0.24104160559960439,
+      "learning_rate": 2.0466213920393718e-08,
+      "loss": 0.1728,
+      "step": 1219
+    },
+    {
+      "epoch": 4.939271255060729,
+      "grad_norm": 0.239121013190386,
+      "learning_rate": 1.7988254906942115e-08,
+      "loss": 0.1723,
+      "step": 1220
+    },
+    {
+      "epoch": 4.94331983805668,
+      "grad_norm": 0.24886181290433967,
+      "learning_rate": 1.5670071526936447e-08,
+      "loss": 0.1732,
+      "step": 1221
+    },
+    {
+      "epoch": 4.947368421052632,
+      "grad_norm": 0.4214685355925688,
+      "learning_rate": 1.3511682316509344e-08,
+      "loss": 0.1712,
+      "step": 1222
+    },
+    {
+      "epoch": 4.951417004048583,
+      "grad_norm": 0.2507039095145625,
+      "learning_rate": 1.1513104534086605e-08,
+      "loss": 0.165,
+      "step": 1223
+    },
+    {
+      "epoch": 4.955465587044534,
+      "grad_norm": 0.2456973256536877,
+      "learning_rate": 9.67435416023843e-09,
+      "loss": 0.163,
+      "step": 1224
+    },
+    {
+      "epoch": 4.959514170040486,
+      "grad_norm": 0.24425323217640782,
+      "learning_rate": 7.995445897566178e-09,
+      "loss": 0.1909,
+      "step": 1225
+    },
+    {
+      "epoch": 4.963562753036437,
+      "grad_norm": 0.2502432858503313,
+      "learning_rate": 6.476393170578021e-09,
+      "loss": 0.1757,
+      "step": 1226
+    },
+    {
+      "epoch": 4.967611336032388,
+      "grad_norm": 0.25027165681101166,
+      "learning_rate": 5.11720812557348e-09,
+      "loss": 0.173,
+      "step": 1227
+    },
+    {
+      "epoch": 4.97165991902834,
+      "grad_norm": 0.23991937733042862,
+      "learning_rate": 3.917901630563492e-09,
+      "loss": 0.1621,
+      "step": 1228
+    },
+    {
+      "epoch": 4.9757085020242915,
+      "grad_norm": 0.2499677314714579,
+      "learning_rate": 2.878483275170485e-09,
+      "loss": 0.1688,
+      "step": 1229
+    },
+    {
+      "epoch": 4.979757085020243,
+      "grad_norm": 0.25069002273244795,
+      "learning_rate": 1.998961370555108e-09,
+      "loss": 0.1703,
+      "step": 1230
+    },
+    {
+      "epoch": 4.983805668016195,
+      "grad_norm": 0.24530945004557253,
+      "learning_rate": 1.2793429493518361e-09,
+      "loss": 0.1638,
+      "step": 1231
+    },
+    {
+      "epoch": 4.987854251012146,
+      "grad_norm": 0.25110680229124666,
+      "learning_rate": 7.196337656112384e-10,
+      "loss": 0.187,
+      "step": 1232
+    },
+    {
+      "epoch": 4.991902834008097,
+      "grad_norm": 0.24691891452780462,
+      "learning_rate": 3.198382947511292e-10,
+      "loss": 0.1801,
+      "step": 1233
+    },
+    {
+      "epoch": 4.995951417004049,
+      "grad_norm": 0.26254964081308574,
+      "learning_rate": 7.995973352770137e-11,
+      "loss": 0.1718,
+      "step": 1234
+    },
+    {
+      "epoch": 5.0,
+      "grad_norm": 0.2486934533194353,
+      "learning_rate": 0.0,
+      "loss": 0.1418,
+      "step": 1235
+    },
+    {
+      "epoch": 5.0,
+      "step": 1235,
+      "total_flos": 2.1356066527982387e+18,
+      "train_loss": 0.38383434477605316,
+      "train_runtime": 23418.5628,
+      "train_samples_per_second": 6.747,
+      "train_steps_per_second": 0.053
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 1235,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 5,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 2.1356066527982387e+18,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}