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": 6.91439982454023,
+      "learning_rate": 3.2258064516129035e-07,
+      "loss": 1.0196,
+      "step": 1
+    },
+    {
+      "epoch": 0.008097165991902834,
+      "grad_norm": 6.611037188476071,
+      "learning_rate": 6.451612903225807e-07,
+      "loss": 0.9882,
+      "step": 2
+    },
+    {
+      "epoch": 0.012145748987854251,
+      "grad_norm": 6.278886740193047,
+      "learning_rate": 9.67741935483871e-07,
+      "loss": 0.9669,
+      "step": 3
+    },
+    {
+      "epoch": 0.016194331983805668,
+      "grad_norm": 6.441840681271121,
+      "learning_rate": 1.2903225806451614e-06,
+      "loss": 0.988,
+      "step": 4
+    },
+    {
+      "epoch": 0.020242914979757085,
+      "grad_norm": 6.157029002828562,
+      "learning_rate": 1.6129032258064516e-06,
+      "loss": 0.972,
+      "step": 5
+    },
+    {
+      "epoch": 0.024291497975708502,
+      "grad_norm": 6.158500091034868,
+      "learning_rate": 1.935483870967742e-06,
+      "loss": 0.9827,
+      "step": 6
+    },
+    {
+      "epoch": 0.02834008097165992,
+      "grad_norm": 4.718627939258586,
+      "learning_rate": 2.2580645161290324e-06,
+      "loss": 0.914,
+      "step": 7
+    },
+    {
+      "epoch": 0.032388663967611336,
+      "grad_norm": 4.497877260787213,
+      "learning_rate": 2.580645161290323e-06,
+      "loss": 0.9168,
+      "step": 8
+    },
+    {
+      "epoch": 0.03643724696356275,
+      "grad_norm": 2.3962827721593016,
+      "learning_rate": 2.903225806451613e-06,
+      "loss": 0.8814,
+      "step": 9
+    },
+    {
+      "epoch": 0.04048582995951417,
+      "grad_norm": 2.080266256377777,
+      "learning_rate": 3.225806451612903e-06,
+      "loss": 0.8809,
+      "step": 10
+    },
+    {
+      "epoch": 0.044534412955465584,
+      "grad_norm": 1.9959611802819488,
+      "learning_rate": 3.548387096774194e-06,
+      "loss": 0.8664,
+      "step": 11
+    },
+    {
+      "epoch": 0.048582995951417005,
+      "grad_norm": 4.205777891988513,
+      "learning_rate": 3.870967741935484e-06,
+      "loss": 0.8835,
+      "step": 12
+    },
+    {
+      "epoch": 0.05263157894736842,
+      "grad_norm": 4.522952429341937,
+      "learning_rate": 4.193548387096774e-06,
+      "loss": 0.8782,
+      "step": 13
+    },
+    {
+      "epoch": 0.05668016194331984,
+      "grad_norm": 4.516865648123622,
+      "learning_rate": 4.516129032258065e-06,
+      "loss": 0.8659,
+      "step": 14
+    },
+    {
+      "epoch": 0.06072874493927125,
+      "grad_norm": 4.03569261682077,
+      "learning_rate": 4.838709677419355e-06,
+      "loss": 0.8271,
+      "step": 15
+    },
+    {
+      "epoch": 0.06477732793522267,
+      "grad_norm": 3.0461304566981857,
+      "learning_rate": 5.161290322580646e-06,
+      "loss": 0.8025,
+      "step": 16
+    },
+    {
+      "epoch": 0.06882591093117409,
+      "grad_norm": 3.0370759485579453,
+      "learning_rate": 5.483870967741935e-06,
+      "loss": 0.8174,
+      "step": 17
+    },
+    {
+      "epoch": 0.0728744939271255,
+      "grad_norm": 2.710176753723551,
+      "learning_rate": 5.806451612903226e-06,
+      "loss": 0.8106,
+      "step": 18
+    },
+    {
+      "epoch": 0.07692307692307693,
+      "grad_norm": 2.1858104247843007,
+      "learning_rate": 6.129032258064517e-06,
+      "loss": 0.7864,
+      "step": 19
+    },
+    {
+      "epoch": 0.08097165991902834,
+      "grad_norm": 1.6838246540294233,
+      "learning_rate": 6.451612903225806e-06,
+      "loss": 0.7652,
+      "step": 20
+    },
+    {
+      "epoch": 0.08502024291497975,
+      "grad_norm": 1.5246797344499399,
+      "learning_rate": 6.774193548387097e-06,
+      "loss": 0.7571,
+      "step": 21
+    },
+    {
+      "epoch": 0.08906882591093117,
+      "grad_norm": 1.8068334515148556,
+      "learning_rate": 7.096774193548388e-06,
+      "loss": 0.7447,
+      "step": 22
+    },
+    {
+      "epoch": 0.0931174089068826,
+      "grad_norm": 1.6828411518282573,
+      "learning_rate": 7.4193548387096784e-06,
+      "loss": 0.7092,
+      "step": 23
+    },
+    {
+      "epoch": 0.09716599190283401,
+      "grad_norm": 1.524374939316922,
+      "learning_rate": 7.741935483870968e-06,
+      "loss": 0.7359,
+      "step": 24
+    },
+    {
+      "epoch": 0.10121457489878542,
+      "grad_norm": 1.2526428734672972,
+      "learning_rate": 8.064516129032258e-06,
+      "loss": 0.7328,
+      "step": 25
+    },
+    {
+      "epoch": 0.10526315789473684,
+      "grad_norm": 1.2628308585628552,
+      "learning_rate": 8.387096774193549e-06,
+      "loss": 0.719,
+      "step": 26
+    },
+    {
+      "epoch": 0.10931174089068826,
+      "grad_norm": 1.2628649787441562,
+      "learning_rate": 8.70967741935484e-06,
+      "loss": 0.7053,
+      "step": 27
+    },
+    {
+      "epoch": 0.11336032388663968,
+      "grad_norm": 1.0293571426158756,
+      "learning_rate": 9.03225806451613e-06,
+      "loss": 0.7019,
+      "step": 28
+    },
+    {
+      "epoch": 0.11740890688259109,
+      "grad_norm": 0.9222650708287791,
+      "learning_rate": 9.35483870967742e-06,
+      "loss": 0.7128,
+      "step": 29
+    },
+    {
+      "epoch": 0.1214574898785425,
+      "grad_norm": 0.9818879538054957,
+      "learning_rate": 9.67741935483871e-06,
+      "loss": 0.6904,
+      "step": 30
+    },
+    {
+      "epoch": 0.12550607287449392,
+      "grad_norm": 0.9891340835931939,
+      "learning_rate": 1e-05,
+      "loss": 0.6999,
+      "step": 31
+    },
+    {
+      "epoch": 0.12955465587044535,
+      "grad_norm": 0.7328011459288261,
+      "learning_rate": 1.0322580645161291e-05,
+      "loss": 0.6686,
+      "step": 32
+    },
+    {
+      "epoch": 0.13360323886639677,
+      "grad_norm": 0.8630089626287336,
+      "learning_rate": 1.0645161290322582e-05,
+      "loss": 0.6949,
+      "step": 33
+    },
+    {
+      "epoch": 0.13765182186234817,
+      "grad_norm": 0.7468528581985123,
+      "learning_rate": 1.096774193548387e-05,
+      "loss": 0.6746,
+      "step": 34
+    },
+    {
+      "epoch": 0.1417004048582996,
+      "grad_norm": 0.821207525407642,
+      "learning_rate": 1.1290322580645164e-05,
+      "loss": 0.697,
+      "step": 35
+    },
+    {
+      "epoch": 0.145748987854251,
+      "grad_norm": 0.7064014798913736,
+      "learning_rate": 1.1612903225806453e-05,
+      "loss": 0.6737,
+      "step": 36
+    },
+    {
+      "epoch": 0.14979757085020243,
+      "grad_norm": 0.6875662645693122,
+      "learning_rate": 1.1935483870967743e-05,
+      "loss": 0.6649,
+      "step": 37
+    },
+    {
+      "epoch": 0.15384615384615385,
+      "grad_norm": 0.7748158298436761,
+      "learning_rate": 1.2258064516129034e-05,
+      "loss": 0.681,
+      "step": 38
+    },
+    {
+      "epoch": 0.15789473684210525,
+      "grad_norm": 0.5716478382725939,
+      "learning_rate": 1.2580645161290324e-05,
+      "loss": 0.6553,
+      "step": 39
+    },
+    {
+      "epoch": 0.16194331983805668,
+      "grad_norm": 0.6491806252956833,
+      "learning_rate": 1.2903225806451613e-05,
+      "loss": 0.6461,
+      "step": 40
+    },
+    {
+      "epoch": 0.1659919028340081,
+      "grad_norm": 0.6362627509359275,
+      "learning_rate": 1.3225806451612903e-05,
+      "loss": 0.65,
+      "step": 41
+    },
+    {
+      "epoch": 0.1700404858299595,
+      "grad_norm": 0.6283291958194658,
+      "learning_rate": 1.3548387096774194e-05,
+      "loss": 0.647,
+      "step": 42
+    },
+    {
+      "epoch": 0.17408906882591094,
+      "grad_norm": 0.5855834292724604,
+      "learning_rate": 1.3870967741935486e-05,
+      "loss": 0.6481,
+      "step": 43
+    },
+    {
+      "epoch": 0.17813765182186234,
+      "grad_norm": 0.6345158027632454,
+      "learning_rate": 1.4193548387096776e-05,
+      "loss": 0.6684,
+      "step": 44
+    },
+    {
+      "epoch": 0.18218623481781376,
+      "grad_norm": 0.6228254192866781,
+      "learning_rate": 1.4516129032258066e-05,
+      "loss": 0.6474,
+      "step": 45
+    },
+    {
+      "epoch": 0.1862348178137652,
+      "grad_norm": 0.6299501327478599,
+      "learning_rate": 1.4838709677419357e-05,
+      "loss": 0.6374,
+      "step": 46
+    },
+    {
+      "epoch": 0.1902834008097166,
+      "grad_norm": 0.6817385613299207,
+      "learning_rate": 1.5161290322580646e-05,
+      "loss": 0.6442,
+      "step": 47
+    },
+    {
+      "epoch": 0.19433198380566802,
+      "grad_norm": 0.5127662341993566,
+      "learning_rate": 1.5483870967741936e-05,
+      "loss": 0.6207,
+      "step": 48
+    },
+    {
+      "epoch": 0.19838056680161945,
+      "grad_norm": 0.6116836035014853,
+      "learning_rate": 1.5806451612903226e-05,
+      "loss": 0.6431,
+      "step": 49
+    },
+    {
+      "epoch": 0.20242914979757085,
+      "grad_norm": 0.5441745715157339,
+      "learning_rate": 1.6129032258064517e-05,
+      "loss": 0.6247,
+      "step": 50
+    },
+    {
+      "epoch": 0.20647773279352227,
+      "grad_norm": 0.6226876052525784,
+      "learning_rate": 1.6451612903225807e-05,
+      "loss": 0.6419,
+      "step": 51
+    },
+    {
+      "epoch": 0.21052631578947367,
+      "grad_norm": 0.570402908164821,
+      "learning_rate": 1.6774193548387098e-05,
+      "loss": 0.6402,
+      "step": 52
+    },
+    {
+      "epoch": 0.2145748987854251,
+      "grad_norm": 0.6220646227235918,
+      "learning_rate": 1.7096774193548388e-05,
+      "loss": 0.6173,
+      "step": 53
+    },
+    {
+      "epoch": 0.21862348178137653,
+      "grad_norm": 0.7037580104482473,
+      "learning_rate": 1.741935483870968e-05,
+      "loss": 0.636,
+      "step": 54
+    },
+    {
+      "epoch": 0.22267206477732793,
+      "grad_norm": 0.6011381597146701,
+      "learning_rate": 1.774193548387097e-05,
+      "loss": 0.6222,
+      "step": 55
+    },
+    {
+      "epoch": 0.22672064777327935,
+      "grad_norm": 0.6515291696355174,
+      "learning_rate": 1.806451612903226e-05,
+      "loss": 0.6331,
+      "step": 56
+    },
+    {
+      "epoch": 0.23076923076923078,
+      "grad_norm": 0.6755844180489725,
+      "learning_rate": 1.838709677419355e-05,
+      "loss": 0.6593,
+      "step": 57
+    },
+    {
+      "epoch": 0.23481781376518218,
+      "grad_norm": 0.5510743927801369,
+      "learning_rate": 1.870967741935484e-05,
+      "loss": 0.6473,
+      "step": 58
+    },
+    {
+      "epoch": 0.2388663967611336,
+      "grad_norm": 0.6770299532355615,
+      "learning_rate": 1.903225806451613e-05,
+      "loss": 0.6309,
+      "step": 59
+    },
+    {
+      "epoch": 0.242914979757085,
+      "grad_norm": 0.6423597211942355,
+      "learning_rate": 1.935483870967742e-05,
+      "loss": 0.6252,
+      "step": 60
+    },
+    {
+      "epoch": 0.24696356275303644,
+      "grad_norm": 0.660732938037667,
+      "learning_rate": 1.967741935483871e-05,
+      "loss": 0.617,
+      "step": 61
+    },
+    {
+      "epoch": 0.25101214574898784,
+      "grad_norm": 0.6854456133836372,
+      "learning_rate": 2e-05,
+      "loss": 0.5987,
+      "step": 62
+    },
+    {
+      "epoch": 0.2550607287449393,
+      "grad_norm": 0.7166384654399082,
+      "learning_rate": 2.0322580645161292e-05,
+      "loss": 0.6203,
+      "step": 63
+    },
+    {
+      "epoch": 0.2591093117408907,
+      "grad_norm": 0.7253063880375059,
+      "learning_rate": 2.0645161290322582e-05,
+      "loss": 0.6104,
+      "step": 64
+    },
+    {
+      "epoch": 0.2631578947368421,
+      "grad_norm": 0.6992533150725144,
+      "learning_rate": 2.0967741935483873e-05,
+      "loss": 0.6057,
+      "step": 65
+    },
+    {
+      "epoch": 0.26720647773279355,
+      "grad_norm": 0.7164914246623904,
+      "learning_rate": 2.1290322580645163e-05,
+      "loss": 0.61,
+      "step": 66
+    },
+    {
+      "epoch": 0.27125506072874495,
+      "grad_norm": 0.6626575127213616,
+      "learning_rate": 2.161290322580645e-05,
+      "loss": 0.6235,
+      "step": 67
+    },
+    {
+      "epoch": 0.27530364372469635,
+      "grad_norm": 0.7516237957375272,
+      "learning_rate": 2.193548387096774e-05,
+      "loss": 0.6116,
+      "step": 68
+    },
+    {
+      "epoch": 0.2793522267206478,
+      "grad_norm": 0.6994556876257593,
+      "learning_rate": 2.225806451612903e-05,
+      "loss": 0.595,
+      "step": 69
+    },
+    {
+      "epoch": 0.2834008097165992,
+      "grad_norm": 0.9160108788355905,
+      "learning_rate": 2.2580645161290328e-05,
+      "loss": 0.633,
+      "step": 70
+    },
+    {
+      "epoch": 0.2874493927125506,
+      "grad_norm": 1.1683948884804234,
+      "learning_rate": 2.2903225806451615e-05,
+      "loss": 0.6113,
+      "step": 71
+    },
+    {
+      "epoch": 0.291497975708502,
+      "grad_norm": 0.7256967826053178,
+      "learning_rate": 2.3225806451612906e-05,
+      "loss": 0.6121,
+      "step": 72
+    },
+    {
+      "epoch": 0.29554655870445345,
+      "grad_norm": 1.3617378163598515,
+      "learning_rate": 2.3548387096774196e-05,
+      "loss": 0.6053,
+      "step": 73
+    },
+    {
+      "epoch": 0.29959514170040485,
+      "grad_norm": 0.8743160469080165,
+      "learning_rate": 2.3870967741935486e-05,
+      "loss": 0.6072,
+      "step": 74
+    },
+    {
+      "epoch": 0.30364372469635625,
+      "grad_norm": 1.0282732161394155,
+      "learning_rate": 2.4193548387096777e-05,
+      "loss": 0.5826,
+      "step": 75
+    },
+    {
+      "epoch": 0.3076923076923077,
+      "grad_norm": 1.0728736512459587,
+      "learning_rate": 2.4516129032258067e-05,
+      "loss": 0.6075,
+      "step": 76
+    },
+    {
+      "epoch": 0.3117408906882591,
+      "grad_norm": 0.963875339815632,
+      "learning_rate": 2.4838709677419358e-05,
+      "loss": 0.6082,
+      "step": 77
+    },
+    {
+      "epoch": 0.3157894736842105,
+      "grad_norm": 0.9941159297332919,
+      "learning_rate": 2.5161290322580648e-05,
+      "loss": 0.6003,
+      "step": 78
+    },
+    {
+      "epoch": 0.31983805668016196,
+      "grad_norm": 0.7321927081101347,
+      "learning_rate": 2.548387096774194e-05,
+      "loss": 0.5643,
+      "step": 79
+    },
+    {
+      "epoch": 0.32388663967611336,
+      "grad_norm": 0.9121658408572525,
+      "learning_rate": 2.5806451612903226e-05,
+      "loss": 0.5924,
+      "step": 80
+    },
+    {
+      "epoch": 0.32793522267206476,
+      "grad_norm": 0.8502872049331047,
+      "learning_rate": 2.6129032258064516e-05,
+      "loss": 0.6002,
+      "step": 81
+    },
+    {
+      "epoch": 0.3319838056680162,
+      "grad_norm": 0.7567987954161355,
+      "learning_rate": 2.6451612903225806e-05,
+      "loss": 0.607,
+      "step": 82
+    },
+    {
+      "epoch": 0.3360323886639676,
+      "grad_norm": 0.9086390894254913,
+      "learning_rate": 2.6774193548387097e-05,
+      "loss": 0.5757,
+      "step": 83
+    },
+    {
+      "epoch": 0.340080971659919,
+      "grad_norm": 0.7476645541143958,
+      "learning_rate": 2.7096774193548387e-05,
+      "loss": 0.5821,
+      "step": 84
+    },
+    {
+      "epoch": 0.3441295546558704,
+      "grad_norm": 1.254807107757769,
+      "learning_rate": 2.7419354838709678e-05,
+      "loss": 0.6125,
+      "step": 85
+    },
+    {
+      "epoch": 0.3481781376518219,
+      "grad_norm": 1.023721801689678,
+      "learning_rate": 2.774193548387097e-05,
+      "loss": 0.5838,
+      "step": 86
+    },
+    {
+      "epoch": 0.3522267206477733,
+      "grad_norm": 0.8484450416231974,
+      "learning_rate": 2.8064516129032262e-05,
+      "loss": 0.5935,
+      "step": 87
+    },
+    {
+      "epoch": 0.3562753036437247,
+      "grad_norm": 0.9438231708881947,
+      "learning_rate": 2.8387096774193552e-05,
+      "loss": 0.5875,
+      "step": 88
+    },
+    {
+      "epoch": 0.3603238866396761,
+      "grad_norm": 0.8749904810648811,
+      "learning_rate": 2.8709677419354843e-05,
+      "loss": 0.589,
+      "step": 89
+    },
+    {
+      "epoch": 0.3643724696356275,
+      "grad_norm": 0.9245182636951439,
+      "learning_rate": 2.9032258064516133e-05,
+      "loss": 0.5945,
+      "step": 90
+    },
+    {
+      "epoch": 0.3684210526315789,
+      "grad_norm": 1.083312055677822,
+      "learning_rate": 2.9354838709677423e-05,
+      "loss": 0.596,
+      "step": 91
+    },
+    {
+      "epoch": 0.3724696356275304,
+      "grad_norm": 0.7127477677643788,
+      "learning_rate": 2.9677419354838714e-05,
+      "loss": 0.6,
+      "step": 92
+    },
+    {
+      "epoch": 0.3765182186234818,
+      "grad_norm": 1.007724055136697,
+      "learning_rate": 3.0000000000000004e-05,
+      "loss": 0.5992,
+      "step": 93
+    },
+    {
+      "epoch": 0.3805668016194332,
+      "grad_norm": 0.7586587438442745,
+      "learning_rate": 3.032258064516129e-05,
+      "loss": 0.589,
+      "step": 94
+    },
+    {
+      "epoch": 0.38461538461538464,
+      "grad_norm": 0.6635143318534671,
+      "learning_rate": 3.0645161290322585e-05,
+      "loss": 0.5661,
+      "step": 95
+    },
+    {
+      "epoch": 0.38866396761133604,
+      "grad_norm": 0.810666316426532,
+      "learning_rate": 3.096774193548387e-05,
+      "loss": 0.5948,
+      "step": 96
+    },
+    {
+      "epoch": 0.39271255060728744,
+      "grad_norm": 0.6900246200571897,
+      "learning_rate": 3.1290322580645166e-05,
+      "loss": 0.6037,
+      "step": 97
+    },
+    {
+      "epoch": 0.3967611336032389,
+      "grad_norm": 0.7226387784353655,
+      "learning_rate": 3.161290322580645e-05,
+      "loss": 0.5701,
+      "step": 98
+    },
+    {
+      "epoch": 0.4008097165991903,
+      "grad_norm": 0.9619920740492134,
+      "learning_rate": 3.193548387096774e-05,
+      "loss": 0.5829,
+      "step": 99
+    },
+    {
+      "epoch": 0.4048582995951417,
+      "grad_norm": 1.0774685720482906,
+      "learning_rate": 3.2258064516129034e-05,
+      "loss": 0.596,
+      "step": 100
+    },
+    {
+      "epoch": 0.4089068825910931,
+      "grad_norm": 0.583347649416544,
+      "learning_rate": 3.258064516129033e-05,
+      "loss": 0.5876,
+      "step": 101
+    },
+    {
+      "epoch": 0.41295546558704455,
+      "grad_norm": 0.8104843665616732,
+      "learning_rate": 3.2903225806451614e-05,
+      "loss": 0.578,
+      "step": 102
+    },
+    {
+      "epoch": 0.41700404858299595,
+      "grad_norm": 0.654322439107127,
+      "learning_rate": 3.322580645161291e-05,
+      "loss": 0.5729,
+      "step": 103
+    },
+    {
+      "epoch": 0.42105263157894735,
+      "grad_norm": 0.7319442733715082,
+      "learning_rate": 3.3548387096774195e-05,
+      "loss": 0.5853,
+      "step": 104
+    },
+    {
+      "epoch": 0.4251012145748988,
+      "grad_norm": 0.7651350429293301,
+      "learning_rate": 3.387096774193549e-05,
+      "loss": 0.5921,
+      "step": 105
+    },
+    {
+      "epoch": 0.4291497975708502,
+      "grad_norm": 0.7567213374828199,
+      "learning_rate": 3.4193548387096776e-05,
+      "loss": 0.5677,
+      "step": 106
+    },
+    {
+      "epoch": 0.4331983805668016,
+      "grad_norm": 0.7266492677748358,
+      "learning_rate": 3.451612903225807e-05,
+      "loss": 0.5874,
+      "step": 107
+    },
+    {
+      "epoch": 0.43724696356275305,
+      "grad_norm": 0.7691152767158361,
+      "learning_rate": 3.483870967741936e-05,
+      "loss": 0.5889,
+      "step": 108
+    },
+    {
+      "epoch": 0.44129554655870445,
+      "grad_norm": 1.0185218664241045,
+      "learning_rate": 3.516129032258065e-05,
+      "loss": 0.5658,
+      "step": 109
+    },
+    {
+      "epoch": 0.44534412955465585,
+      "grad_norm": 1.282866908781599,
+      "learning_rate": 3.548387096774194e-05,
+      "loss": 0.5838,
+      "step": 110
+    },
+    {
+      "epoch": 0.4493927125506073,
+      "grad_norm": 0.7482459752205654,
+      "learning_rate": 3.5806451612903225e-05,
+      "loss": 0.5961,
+      "step": 111
+    },
+    {
+      "epoch": 0.4534412955465587,
+      "grad_norm": 0.7238753196957406,
+      "learning_rate": 3.612903225806452e-05,
+      "loss": 0.591,
+      "step": 112
+    },
+    {
+      "epoch": 0.4574898785425101,
+      "grad_norm": 1.0954889848190619,
+      "learning_rate": 3.6451612903225805e-05,
+      "loss": 0.573,
+      "step": 113
+    },
+    {
+      "epoch": 0.46153846153846156,
+      "grad_norm": 1.2727689764461676,
+      "learning_rate": 3.67741935483871e-05,
+      "loss": 0.5782,
+      "step": 114
+    },
+    {
+      "epoch": 0.46558704453441296,
+      "grad_norm": 0.7528211073474118,
+      "learning_rate": 3.7096774193548386e-05,
+      "loss": 0.5904,
+      "step": 115
+    },
+    {
+      "epoch": 0.46963562753036436,
+      "grad_norm": 1.164341549295574,
+      "learning_rate": 3.741935483870968e-05,
+      "loss": 0.5932,
+      "step": 116
+    },
+    {
+      "epoch": 0.47368421052631576,
+      "grad_norm": 0.8718118075534099,
+      "learning_rate": 3.7741935483870974e-05,
+      "loss": 0.5829,
+      "step": 117
+    },
+    {
+      "epoch": 0.4777327935222672,
+      "grad_norm": 0.7550281585152258,
+      "learning_rate": 3.806451612903226e-05,
+      "loss": 0.5835,
+      "step": 118
+    },
+    {
+      "epoch": 0.4817813765182186,
+      "grad_norm": 0.6946069581298773,
+      "learning_rate": 3.8387096774193555e-05,
+      "loss": 0.5742,
+      "step": 119
+    },
+    {
+      "epoch": 0.48582995951417,
+      "grad_norm": 0.8372358842911266,
+      "learning_rate": 3.870967741935484e-05,
+      "loss": 0.5879,
+      "step": 120
+    },
+    {
+      "epoch": 0.4898785425101215,
+      "grad_norm": 0.9326235668038219,
+      "learning_rate": 3.9032258064516135e-05,
+      "loss": 0.5828,
+      "step": 121
+    },
+    {
+      "epoch": 0.4939271255060729,
+      "grad_norm": 0.789401540489468,
+      "learning_rate": 3.935483870967742e-05,
+      "loss": 0.5702,
+      "step": 122
+    },
+    {
+      "epoch": 0.4979757085020243,
+      "grad_norm": 0.8407320058765433,
+      "learning_rate": 3.9677419354838716e-05,
+      "loss": 0.5816,
+      "step": 123
+    },
+    {
+      "epoch": 0.5020242914979757,
+      "grad_norm": 0.7532861504929652,
+      "learning_rate": 4e-05,
+      "loss": 0.5628,
+      "step": 124
+    },
+    {
+      "epoch": 0.5060728744939271,
+      "grad_norm": 0.8004118596273322,
+      "learning_rate": 3.9999920040266474e-05,
+      "loss": 0.6148,
+      "step": 125
+    },
+    {
+      "epoch": 0.5101214574898786,
+      "grad_norm": 0.8107427007596643,
+      "learning_rate": 3.999968016170525e-05,
+      "loss": 0.5737,
+      "step": 126
+    },
+    {
+      "epoch": 0.5141700404858299,
+      "grad_norm": 0.7782769442810809,
+      "learning_rate": 3.9999280366234394e-05,
+      "loss": 0.5737,
+      "step": 127
+    },
+    {
+      "epoch": 0.5182186234817814,
+      "grad_norm": 0.8048302306313392,
+      "learning_rate": 3.999872065705065e-05,
+      "loss": 0.5781,
+      "step": 128
+    },
+    {
+      "epoch": 0.5222672064777328,
+      "grad_norm": 0.7653125065176388,
+      "learning_rate": 3.999800103862945e-05,
+      "loss": 0.565,
+      "step": 129
+    },
+    {
+      "epoch": 0.5263157894736842,
+      "grad_norm": 0.6842006874927496,
+      "learning_rate": 3.9997121516724834e-05,
+      "loss": 0.5854,
+      "step": 130
+    },
+    {
+      "epoch": 0.5303643724696356,
+      "grad_norm": 0.6763997689896543,
+      "learning_rate": 3.999608209836944e-05,
+      "loss": 0.5724,
+      "step": 131
+    },
+    {
+      "epoch": 0.5344129554655871,
+      "grad_norm": 0.7504476630825971,
+      "learning_rate": 3.999488279187443e-05,
+      "loss": 0.5841,
+      "step": 132
+    },
+    {
+      "epoch": 0.5384615384615384,
+      "grad_norm": 0.7574808423011655,
+      "learning_rate": 3.9993523606829425e-05,
+      "loss": 0.5927,
+      "step": 133
+    },
+    {
+      "epoch": 0.5425101214574899,
+      "grad_norm": 0.7923611426367804,
+      "learning_rate": 3.9992004554102435e-05,
+      "loss": 0.5624,
+      "step": 134
+    },
+    {
+      "epoch": 0.5465587044534413,
+      "grad_norm": 0.9132648103456406,
+      "learning_rate": 3.999032564583976e-05,
+      "loss": 0.577,
+      "step": 135
+    },
+    {
+      "epoch": 0.5506072874493927,
+      "grad_norm": 0.8761191244341112,
+      "learning_rate": 3.998848689546591e-05,
+      "loss": 0.5616,
+      "step": 136
+    },
+    {
+      "epoch": 0.5546558704453441,
+      "grad_norm": 0.7022307724776943,
+      "learning_rate": 3.9986488317683496e-05,
+      "loss": 0.5653,
+      "step": 137
+    },
+    {
+      "epoch": 0.5587044534412956,
+      "grad_norm": 0.7695028519356771,
+      "learning_rate": 3.9984329928473067e-05,
+      "loss": 0.5733,
+      "step": 138
+    },
+    {
+      "epoch": 0.562753036437247,
+      "grad_norm": 0.9622412173497,
+      "learning_rate": 3.998201174509306e-05,
+      "loss": 0.5617,
+      "step": 139
+    },
+    {
+      "epoch": 0.5668016194331984,
+      "grad_norm": 0.7997308057857767,
+      "learning_rate": 3.997953378607961e-05,
+      "loss": 0.555,
+      "step": 140
+    },
+    {
+      "epoch": 0.5708502024291497,
+      "grad_norm": 0.8481310893370686,
+      "learning_rate": 3.9976896071246403e-05,
+      "loss": 0.5809,
+      "step": 141
+    },
+    {
+      "epoch": 0.5748987854251012,
+      "grad_norm": 0.8003275544802565,
+      "learning_rate": 3.997409862168454e-05,
+      "loss": 0.5602,
+      "step": 142
+    },
+    {
+      "epoch": 0.5789473684210527,
+      "grad_norm": 0.837786611435764,
+      "learning_rate": 3.997114145976237e-05,
+      "loss": 0.5501,
+      "step": 143
+    },
+    {
+      "epoch": 0.582995951417004,
+      "grad_norm": 0.8094920341564575,
+      "learning_rate": 3.9968024609125245e-05,
+      "loss": 0.5796,
+      "step": 144
+    },
+    {
+      "epoch": 0.5870445344129555,
+      "grad_norm": 0.6485036818731799,
+      "learning_rate": 3.996474809469545e-05,
+      "loss": 0.5494,
+      "step": 145
+    },
+    {
+      "epoch": 0.5910931174089069,
+      "grad_norm": 1.0429103062939176,
+      "learning_rate": 3.996131194267189e-05,
+      "loss": 0.5757,
+      "step": 146
+    },
+    {
+      "epoch": 0.5951417004048583,
+      "grad_norm": 0.7688494201748384,
+      "learning_rate": 3.995771618052995e-05,
+      "loss": 0.5757,
+      "step": 147
+    },
+    {
+      "epoch": 0.5991902834008097,
+      "grad_norm": 0.6566952471012729,
+      "learning_rate": 3.9953960837021256e-05,
+      "loss": 0.5766,
+      "step": 148
+    },
+    {
+      "epoch": 0.6032388663967612,
+      "grad_norm": 0.6959331977853267,
+      "learning_rate": 3.995004594217343e-05,
+      "loss": 0.5602,
+      "step": 149
+    },
+    {
+      "epoch": 0.6072874493927125,
+      "grad_norm": 0.5823629833578193,
+      "learning_rate": 3.994597152728987e-05,
+      "loss": 0.5782,
+      "step": 150
+    },
+    {
+      "epoch": 0.611336032388664,
+      "grad_norm": 0.5274472530930019,
+      "learning_rate": 3.994173762494947e-05,
+      "loss": 0.5548,
+      "step": 151
+    },
+    {
+      "epoch": 0.6153846153846154,
+      "grad_norm": 0.5613112391057612,
+      "learning_rate": 3.9937344269006424e-05,
+      "loss": 0.5507,
+      "step": 152
+    },
+    {
+      "epoch": 0.6194331983805668,
+      "grad_norm": 0.689022410267624,
+      "learning_rate": 3.993279149458988e-05,
+      "loss": 0.5688,
+      "step": 153
+    },
+    {
+      "epoch": 0.6234817813765182,
+      "grad_norm": 0.7688929918466414,
+      "learning_rate": 3.9928079338103695e-05,
+      "loss": 0.5737,
+      "step": 154
+    },
+    {
+      "epoch": 0.6275303643724697,
+      "grad_norm": 0.6856895544738326,
+      "learning_rate": 3.992320783722616e-05,
+      "loss": 0.5653,
+      "step": 155
+    },
+    {
+      "epoch": 0.631578947368421,
+      "grad_norm": 0.6212998256635267,
+      "learning_rate": 3.991817703090965e-05,
+      "loss": 0.5438,
+      "step": 156
+    },
+    {
+      "epoch": 0.6356275303643725,
+      "grad_norm": 0.5524879697209308,
+      "learning_rate": 3.991298695938038e-05,
+      "loss": 0.5721,
+      "step": 157
+    },
+    {
+      "epoch": 0.6396761133603239,
+      "grad_norm": 0.6747716547496253,
+      "learning_rate": 3.9907637664138e-05,
+      "loss": 0.5782,
+      "step": 158
+    },
+    {
+      "epoch": 0.6437246963562753,
+      "grad_norm": 0.5859661457318545,
+      "learning_rate": 3.990212918795534e-05,
+      "loss": 0.5584,
+      "step": 159
+    },
+    {
+      "epoch": 0.6477732793522267,
+      "grad_norm": 0.6829391063606716,
+      "learning_rate": 3.9896461574878036e-05,
+      "loss": 0.5611,
+      "step": 160
+    },
+    {
+      "epoch": 0.6518218623481782,
+      "grad_norm": 0.5990797840068647,
+      "learning_rate": 3.989063487022416e-05,
+      "loss": 0.5426,
+      "step": 161
+    },
+    {
+      "epoch": 0.6558704453441295,
+      "grad_norm": 0.60923749447732,
+      "learning_rate": 3.9884649120583903e-05,
+      "loss": 0.5595,
+      "step": 162
+    },
+    {
+      "epoch": 0.659919028340081,
+      "grad_norm": 0.5422937660670134,
+      "learning_rate": 3.987850437381915e-05,
+      "loss": 0.5509,
+      "step": 163
+    },
+    {
+      "epoch": 0.6639676113360324,
+      "grad_norm": 0.7162973864083567,
+      "learning_rate": 3.987220067906313e-05,
+      "loss": 0.5552,
+      "step": 164
+    },
+    {
+      "epoch": 0.6680161943319838,
+      "grad_norm": 0.6472901327658984,
+      "learning_rate": 3.986573808672002e-05,
+      "loss": 0.562,
+      "step": 165
+    },
+    {
+      "epoch": 0.6720647773279352,
+      "grad_norm": 0.5741317155225052,
+      "learning_rate": 3.9859116648464534e-05,
+      "loss": 0.5707,
+      "step": 166
+    },
+    {
+      "epoch": 0.6761133603238867,
+      "grad_norm": 0.7266697019451276,
+      "learning_rate": 3.985233641724152e-05,
+      "loss": 0.5541,
+      "step": 167
+    },
+    {
+      "epoch": 0.680161943319838,
+      "grad_norm": 0.6401844667198147,
+      "learning_rate": 3.984539744726554e-05,
+      "loss": 0.5482,
+      "step": 168
+    },
+    {
+      "epoch": 0.6842105263157895,
+      "grad_norm": 0.6799681867933521,
+      "learning_rate": 3.983829979402038e-05,
+      "loss": 0.5498,
+      "step": 169
+    },
+    {
+      "epoch": 0.6882591093117408,
+      "grad_norm": 0.7339850338428117,
+      "learning_rate": 3.983104351425871e-05,
+      "loss": 0.5234,
+      "step": 170
+    },
+    {
+      "epoch": 0.6923076923076923,
+      "grad_norm": 0.6525153932760162,
+      "learning_rate": 3.982362866600154e-05,
+      "loss": 0.548,
+      "step": 171
+    },
+    {
+      "epoch": 0.6963562753036437,
+      "grad_norm": 0.7305153777107973,
+      "learning_rate": 3.9816055308537806e-05,
+      "loss": 0.5526,
+      "step": 172
+    },
+    {
+      "epoch": 0.7004048582995951,
+      "grad_norm": 0.6094044413694832,
+      "learning_rate": 3.980832350242388e-05,
+      "loss": 0.5447,
+      "step": 173
+    },
+    {
+      "epoch": 0.7044534412955465,
+      "grad_norm": 0.7587254524342729,
+      "learning_rate": 3.980043330948306e-05,
+      "loss": 0.5651,
+      "step": 174
+    },
+    {
+      "epoch": 0.708502024291498,
+      "grad_norm": 0.5899423798334114,
+      "learning_rate": 3.9792384792805124e-05,
+      "loss": 0.5538,
+      "step": 175
+    },
+    {
+      "epoch": 0.7125506072874493,
+      "grad_norm": 0.6144651752662363,
+      "learning_rate": 3.978417801674579e-05,
+      "loss": 0.5668,
+      "step": 176
+    },
+    {
+      "epoch": 0.7165991902834008,
+      "grad_norm": 0.6903000525165026,
+      "learning_rate": 3.9775813046926244e-05,
+      "loss": 0.5683,
+      "step": 177
+    },
+    {
+      "epoch": 0.7206477732793523,
+      "grad_norm": 0.48255263318264857,
+      "learning_rate": 3.976728995023254e-05,
+      "loss": 0.5389,
+      "step": 178
+    },
+    {
+      "epoch": 0.7246963562753036,
+      "grad_norm": 0.6506746273045,
+      "learning_rate": 3.975860879481514e-05,
+      "loss": 0.5505,
+      "step": 179
+    },
+    {
+      "epoch": 0.728744939271255,
+      "grad_norm": 0.6291452493274834,
+      "learning_rate": 3.974976965008833e-05,
+      "loss": 0.5631,
+      "step": 180
+    },
+    {
+      "epoch": 0.7327935222672065,
+      "grad_norm": 0.5995085886645466,
+      "learning_rate": 3.9740772586729674e-05,
+      "loss": 0.52,
+      "step": 181
+    },
+    {
+      "epoch": 0.7368421052631579,
+      "grad_norm": 0.5717052760953413,
+      "learning_rate": 3.973161767667946e-05,
+      "loss": 0.5546,
+      "step": 182
+    },
+    {
+      "epoch": 0.7408906882591093,
+      "grad_norm": 0.5859438392594744,
+      "learning_rate": 3.972230499314009e-05,
+      "loss": 0.5719,
+      "step": 183
+    },
+    {
+      "epoch": 0.7449392712550608,
+      "grad_norm": 0.5188152804864878,
+      "learning_rate": 3.9712834610575545e-05,
+      "loss": 0.5303,
+      "step": 184
+    },
+    {
+      "epoch": 0.7489878542510121,
+      "grad_norm": 0.5304209178468712,
+      "learning_rate": 3.9703206604710746e-05,
+      "loss": 0.5566,
+      "step": 185
+    },
+    {
+      "epoch": 0.7530364372469636,
+      "grad_norm": 0.6272524427419848,
+      "learning_rate": 3.969342105253098e-05,
+      "loss": 0.5692,
+      "step": 186
+    },
+    {
+      "epoch": 0.757085020242915,
+      "grad_norm": 0.5375902222077092,
+      "learning_rate": 3.9683478032281246e-05,
+      "loss": 0.5391,
+      "step": 187
+    },
+    {
+      "epoch": 0.7611336032388664,
+      "grad_norm": 0.6031890162646101,
+      "learning_rate": 3.967337762346568e-05,
+      "loss": 0.5368,
+      "step": 188
+    },
+    {
+      "epoch": 0.7651821862348178,
+      "grad_norm": 0.6203984063268676,
+      "learning_rate": 3.9663119906846885e-05,
+      "loss": 0.5522,
+      "step": 189
+    },
+    {
+      "epoch": 0.7692307692307693,
+      "grad_norm": 0.8003622370413697,
+      "learning_rate": 3.965270496444528e-05,
+      "loss": 0.5687,
+      "step": 190
+    },
+    {
+      "epoch": 0.7732793522267206,
+      "grad_norm": 0.5953046598976753,
+      "learning_rate": 3.9642132879538465e-05,
+      "loss": 0.5461,
+      "step": 191
+    },
+    {
+      "epoch": 0.7773279352226721,
+      "grad_norm": 0.7038127589971238,
+      "learning_rate": 3.9631403736660566e-05,
+      "loss": 0.564,
+      "step": 192
+    },
+    {
+      "epoch": 0.7813765182186235,
+      "grad_norm": 0.653066831830131,
+      "learning_rate": 3.962051762160151e-05,
+      "loss": 0.5416,
+      "step": 193
+    },
+    {
+      "epoch": 0.7854251012145749,
+      "grad_norm": 0.5476263538266813,
+      "learning_rate": 3.960947462140638e-05,
+      "loss": 0.5348,
+      "step": 194
+    },
+    {
+      "epoch": 0.7894736842105263,
+      "grad_norm": 0.6637033259598588,
+      "learning_rate": 3.959827482437472e-05,
+      "loss": 0.5615,
+      "step": 195
+    },
+    {
+      "epoch": 0.7935222672064778,
+      "grad_norm": 0.5909093905779422,
+      "learning_rate": 3.958691832005981e-05,
+      "loss": 0.5323,
+      "step": 196
+    },
+    {
+      "epoch": 0.7975708502024291,
+      "grad_norm": 0.6266426984187328,
+      "learning_rate": 3.9575405199267944e-05,
+      "loss": 0.5402,
+      "step": 197
+    },
+    {
+      "epoch": 0.8016194331983806,
+      "grad_norm": 0.5986574836994142,
+      "learning_rate": 3.956373555405774e-05,
+      "loss": 0.5516,
+      "step": 198
+    },
+    {
+      "epoch": 0.805668016194332,
+      "grad_norm": 0.630393512432598,
+      "learning_rate": 3.955190947773937e-05,
+      "loss": 0.5574,
+      "step": 199
+    },
+    {
+      "epoch": 0.8097165991902834,
+      "grad_norm": 0.5177894328511737,
+      "learning_rate": 3.9539927064873824e-05,
+      "loss": 0.546,
+      "step": 200
+    },
+    {
+      "epoch": 0.8137651821862348,
+      "grad_norm": 0.5077701781248308,
+      "learning_rate": 3.952778841127214e-05,
+      "loss": 0.5466,
+      "step": 201
+    },
+    {
+      "epoch": 0.8178137651821862,
+      "grad_norm": 0.5822367150718275,
+      "learning_rate": 3.95154936139947e-05,
+      "loss": 0.5459,
+      "step": 202
+    },
+    {
+      "epoch": 0.8218623481781376,
+      "grad_norm": 0.4847045783814016,
+      "learning_rate": 3.950304277135034e-05,
+      "loss": 0.552,
+      "step": 203
+    },
+    {
+      "epoch": 0.8259109311740891,
+      "grad_norm": 0.5443227512204581,
+      "learning_rate": 3.94904359828957e-05,
+      "loss": 0.5521,
+      "step": 204
+    },
+    {
+      "epoch": 0.8299595141700404,
+      "grad_norm": 0.6128292478362033,
+      "learning_rate": 3.94776733494343e-05,
+      "loss": 0.5458,
+      "step": 205
+    },
+    {
+      "epoch": 0.8340080971659919,
+      "grad_norm": 0.6331972607895184,
+      "learning_rate": 3.9464754973015826e-05,
+      "loss": 0.5531,
+      "step": 206
+    },
+    {
+      "epoch": 0.8380566801619433,
+      "grad_norm": 0.5688610586067223,
+      "learning_rate": 3.9451680956935275e-05,
+      "loss": 0.5477,
+      "step": 207
+    },
+    {
+      "epoch": 0.8421052631578947,
+      "grad_norm": 0.7168708401867817,
+      "learning_rate": 3.943845140573213e-05,
+      "loss": 0.529,
+      "step": 208
+    },
+    {
+      "epoch": 0.8461538461538461,
+      "grad_norm": 0.49070687097280036,
+      "learning_rate": 3.942506642518952e-05,
+      "loss": 0.5206,
+      "step": 209
+    },
+    {
+      "epoch": 0.8502024291497976,
+      "grad_norm": 0.6061519947469395,
+      "learning_rate": 3.9411526122333396e-05,
+      "loss": 0.5359,
+      "step": 210
+    },
+    {
+      "epoch": 0.854251012145749,
+      "grad_norm": 0.6360953131601655,
+      "learning_rate": 3.939783060543167e-05,
+      "loss": 0.5572,
+      "step": 211
+    },
+    {
+      "epoch": 0.8582995951417004,
+      "grad_norm": 0.5864100611131109,
+      "learning_rate": 3.938397998399333e-05,
+      "loss": 0.54,
+      "step": 212
+    },
+    {
+      "epoch": 0.8623481781376519,
+      "grad_norm": 0.5368866714963305,
+      "learning_rate": 3.936997436876756e-05,
+      "loss": 0.5279,
+      "step": 213
+    },
+    {
+      "epoch": 0.8663967611336032,
+      "grad_norm": 0.6064693960477965,
+      "learning_rate": 3.935581387174289e-05,
+      "loss": 0.5207,
+      "step": 214
+    },
+    {
+      "epoch": 0.8704453441295547,
+      "grad_norm": 0.5670236317254922,
+      "learning_rate": 3.9341498606146295e-05,
+      "loss": 0.553,
+      "step": 215
+    },
+    {
+      "epoch": 0.8744939271255061,
+      "grad_norm": 0.5518077488405978,
+      "learning_rate": 3.932702868644224e-05,
+      "loss": 0.5578,
+      "step": 216
+    },
+    {
+      "epoch": 0.8785425101214575,
+      "grad_norm": 0.657372653458077,
+      "learning_rate": 3.931240422833181e-05,
+      "loss": 0.5445,
+      "step": 217
+    },
+    {
+      "epoch": 0.8825910931174089,
+      "grad_norm": 0.47198071835649563,
+      "learning_rate": 3.9297625348751806e-05,
+      "loss": 0.5303,
+      "step": 218
+    },
+    {
+      "epoch": 0.8866396761133604,
+      "grad_norm": 0.5646492966152917,
+      "learning_rate": 3.9282692165873744e-05,
+      "loss": 0.5384,
+      "step": 219
+    },
+    {
+      "epoch": 0.8906882591093117,
+      "grad_norm": 0.4374198949141667,
+      "learning_rate": 3.9267604799102945e-05,
+      "loss": 0.5179,
+      "step": 220
+    },
+    {
+      "epoch": 0.8947368421052632,
+      "grad_norm": 0.5047493480972215,
+      "learning_rate": 3.9252363369077615e-05,
+      "loss": 0.5175,
+      "step": 221
+    },
+    {
+      "epoch": 0.8987854251012146,
+      "grad_norm": 0.5248414654736869,
+      "learning_rate": 3.9236967997667796e-05,
+      "loss": 0.5402,
+      "step": 222
+    },
+    {
+      "epoch": 0.902834008097166,
+      "grad_norm": 0.49415845854979135,
+      "learning_rate": 3.9221418807974496e-05,
+      "loss": 0.5294,
+      "step": 223
+    },
+    {
+      "epoch": 0.9068825910931174,
+      "grad_norm": 0.45352363596772516,
+      "learning_rate": 3.92057159243286e-05,
+      "loss": 0.5152,
+      "step": 224
+    },
+    {
+      "epoch": 0.9109311740890689,
+      "grad_norm": 0.5237720934744394,
+      "learning_rate": 3.9189859472289956e-05,
+      "loss": 0.5501,
+      "step": 225
+    },
+    {
+      "epoch": 0.9149797570850202,
+      "grad_norm": 0.5042526229191596,
+      "learning_rate": 3.9173849578646324e-05,
+      "loss": 0.5375,
+      "step": 226
+    },
+    {
+      "epoch": 0.9190283400809717,
+      "grad_norm": 0.5616676133218438,
+      "learning_rate": 3.9157686371412395e-05,
+      "loss": 0.5407,
+      "step": 227
+    },
+    {
+      "epoch": 0.9230769230769231,
+      "grad_norm": 0.5603116560216412,
+      "learning_rate": 3.914136997982874e-05,
+      "loss": 0.5464,
+      "step": 228
+    },
+    {
+      "epoch": 0.9271255060728745,
+      "grad_norm": 0.5454160753243187,
+      "learning_rate": 3.912490053436079e-05,
+      "loss": 0.5557,
+      "step": 229
+    },
+    {
+      "epoch": 0.9311740890688259,
+      "grad_norm": 0.6224274023606,
+      "learning_rate": 3.9108278166697805e-05,
+      "loss": 0.5239,
+      "step": 230
+    },
+    {
+      "epoch": 0.9352226720647774,
+      "grad_norm": 0.5559748886968797,
+      "learning_rate": 3.909150300975178e-05,
+      "loss": 0.5421,
+      "step": 231
+    },
+    {
+      "epoch": 0.9392712550607287,
+      "grad_norm": 0.5763654800233462,
+      "learning_rate": 3.907457519765642e-05,
+      "loss": 0.5392,
+      "step": 232
+    },
+    {
+      "epoch": 0.9433198380566802,
+      "grad_norm": 0.5320675395701142,
+      "learning_rate": 3.905749486576607e-05,
+      "loss": 0.5449,
+      "step": 233
+    },
+    {
+      "epoch": 0.9473684210526315,
+      "grad_norm": 0.5103900503739126,
+      "learning_rate": 3.90402621506546e-05,
+      "loss": 0.5552,
+      "step": 234
+    },
+    {
+      "epoch": 0.951417004048583,
+      "grad_norm": 0.539334671061716,
+      "learning_rate": 3.902287719011435e-05,
+      "loss": 0.5476,
+      "step": 235
+    },
+    {
+      "epoch": 0.9554655870445344,
+      "grad_norm": 0.4447987187272607,
+      "learning_rate": 3.900534012315499e-05,
+      "loss": 0.5261,
+      "step": 236
+    },
+    {
+      "epoch": 0.9595141700404858,
+      "grad_norm": 0.545623894384561,
+      "learning_rate": 3.8987651090002454e-05,
+      "loss": 0.5275,
+      "step": 237
+    },
+    {
+      "epoch": 0.9635627530364372,
+      "grad_norm": 0.43847787595474413,
+      "learning_rate": 3.896981023209777e-05,
+      "loss": 0.5283,
+      "step": 238
+    },
+    {
+      "epoch": 0.9676113360323887,
+      "grad_norm": 0.4539807861620778,
+      "learning_rate": 3.8951817692095966e-05,
+      "loss": 0.5369,
+      "step": 239
+    },
+    {
+      "epoch": 0.97165991902834,
+      "grad_norm": 0.47517900519793377,
+      "learning_rate": 3.893367361386491e-05,
+      "loss": 0.5223,
+      "step": 240
+    },
+    {
+      "epoch": 0.9757085020242915,
+      "grad_norm": 0.48029008482514574,
+      "learning_rate": 3.891537814248417e-05,
+      "loss": 0.5417,
+      "step": 241
+    },
+    {
+      "epoch": 0.979757085020243,
+      "grad_norm": 0.47346952252518976,
+      "learning_rate": 3.889693142424384e-05,
+      "loss": 0.5568,
+      "step": 242
+    },
+    {
+      "epoch": 0.9838056680161943,
+      "grad_norm": 0.5086960876927074,
+      "learning_rate": 3.8878333606643405e-05,
+      "loss": 0.5324,
+      "step": 243
+    },
+    {
+      "epoch": 0.9878542510121457,
+      "grad_norm": 0.5472311207766616,
+      "learning_rate": 3.885958483839049e-05,
+      "loss": 0.55,
+      "step": 244
+    },
+    {
+      "epoch": 0.9919028340080972,
+      "grad_norm": 0.6031836444293271,
+      "learning_rate": 3.8840685269399786e-05,
+      "loss": 0.504,
+      "step": 245
+    },
+    {
+      "epoch": 0.9959514170040485,
+      "grad_norm": 0.4790625450709841,
+      "learning_rate": 3.882163505079171e-05,
+      "loss": 0.5116,
+      "step": 246
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.7121918386594018,
+      "learning_rate": 3.880243433489132e-05,
+      "loss": 0.544,
+      "step": 247
+    },
+    {
+      "epoch": 1.0040485829959513,
+      "grad_norm": 0.6553054042179141,
+      "learning_rate": 3.878308327522702e-05,
+      "loss": 0.4562,
+      "step": 248
+    },
+    {
+      "epoch": 1.008097165991903,
+      "grad_norm": 0.6571079506796456,
+      "learning_rate": 3.876358202652936e-05,
+      "loss": 0.4604,
+      "step": 249
+    },
+    {
+      "epoch": 1.0121457489878543,
+      "grad_norm": 0.7799892953195962,
+      "learning_rate": 3.8743930744729835e-05,
+      "loss": 0.4815,
+      "step": 250
+    },
+    {
+      "epoch": 1.0161943319838056,
+      "grad_norm": 0.5926158078098009,
+      "learning_rate": 3.8724129586959535e-05,
+      "loss": 0.4483,
+      "step": 251
+    },
+    {
+      "epoch": 1.0202429149797572,
+      "grad_norm": 0.6194190825674073,
+      "learning_rate": 3.870417871154801e-05,
+      "loss": 0.4545,
+      "step": 252
+    },
+    {
+      "epoch": 1.0242914979757085,
+      "grad_norm": 0.6287953682138896,
+      "learning_rate": 3.868407827802192e-05,
+      "loss": 0.4595,
+      "step": 253
+    },
+    {
+      "epoch": 1.0283400809716599,
+      "grad_norm": 0.5125463903869024,
+      "learning_rate": 3.8663828447103804e-05,
+      "loss": 0.451,
+      "step": 254
+    },
+    {
+      "epoch": 1.0323886639676114,
+      "grad_norm": 0.7381156220561103,
+      "learning_rate": 3.8643429380710765e-05,
+      "loss": 0.4551,
+      "step": 255
+    },
+    {
+      "epoch": 1.0364372469635628,
+      "grad_norm": 0.5007144078387045,
+      "learning_rate": 3.862288124195319e-05,
+      "loss": 0.454,
+      "step": 256
+    },
+    {
+      "epoch": 1.040485829959514,
+      "grad_norm": 0.5940471232118957,
+      "learning_rate": 3.8602184195133464e-05,
+      "loss": 0.4641,
+      "step": 257
+    },
+    {
+      "epoch": 1.0445344129554657,
+      "grad_norm": 0.5830176346543179,
+      "learning_rate": 3.8581338405744604e-05,
+      "loss": 0.447,
+      "step": 258
+    },
+    {
+      "epoch": 1.048582995951417,
+      "grad_norm": 0.49074472288006676,
+      "learning_rate": 3.8560344040468996e-05,
+      "loss": 0.4702,
+      "step": 259
+    },
+    {
+      "epoch": 1.0526315789473684,
+      "grad_norm": 0.6096815764842323,
+      "learning_rate": 3.853920126717703e-05,
+      "loss": 0.4568,
+      "step": 260
+    },
+    {
+      "epoch": 1.05668016194332,
+      "grad_norm": 0.7570542230678967,
+      "learning_rate": 3.851791025492574e-05,
+      "loss": 0.4765,
+      "step": 261
+    },
+    {
+      "epoch": 1.0607287449392713,
+      "grad_norm": 0.49749881130006296,
+      "learning_rate": 3.8496471173957515e-05,
+      "loss": 0.4562,
+      "step": 262
+    },
+    {
+      "epoch": 1.0647773279352226,
+      "grad_norm": 0.7922827228634954,
+      "learning_rate": 3.847488419569866e-05,
+      "loss": 0.4696,
+      "step": 263
+    },
+    {
+      "epoch": 1.0688259109311742,
+      "grad_norm": 0.6220162989396968,
+      "learning_rate": 3.8453149492758077e-05,
+      "loss": 0.455,
+      "step": 264
+    },
+    {
+      "epoch": 1.0728744939271255,
+      "grad_norm": 0.59458043740469,
+      "learning_rate": 3.8431267238925886e-05,
+      "loss": 0.4589,
+      "step": 265
+    },
+    {
+      "epoch": 1.0769230769230769,
+      "grad_norm": 0.5365203865497199,
+      "learning_rate": 3.840923760917199e-05,
+      "loss": 0.4648,
+      "step": 266
+    },
+    {
+      "epoch": 1.0809716599190284,
+      "grad_norm": 0.4923737039176017,
+      "learning_rate": 3.838706077964473e-05,
+      "loss": 0.469,
+      "step": 267
+    },
+    {
+      "epoch": 1.0850202429149798,
+      "grad_norm": 0.4992818749928396,
+      "learning_rate": 3.836473692766944e-05,
+      "loss": 0.4506,
+      "step": 268
+    },
+    {
+      "epoch": 1.0890688259109311,
+      "grad_norm": 0.465433091862857,
+      "learning_rate": 3.834226623174705e-05,
+      "loss": 0.4502,
+      "step": 269
+    },
+    {
+      "epoch": 1.0931174089068827,
+      "grad_norm": 0.529003922126649,
+      "learning_rate": 3.831964887155264e-05,
+      "loss": 0.4527,
+      "step": 270
+    },
+    {
+      "epoch": 1.097165991902834,
+      "grad_norm": 0.6049675736645987,
+      "learning_rate": 3.829688502793403e-05,
+      "loss": 0.4501,
+      "step": 271
+    },
+    {
+      "epoch": 1.1012145748987854,
+      "grad_norm": 0.6790913032288739,
+      "learning_rate": 3.827397488291029e-05,
+      "loss": 0.4511,
+      "step": 272
+    },
+    {
+      "epoch": 1.1052631578947367,
+      "grad_norm": 0.5125162216042488,
+      "learning_rate": 3.8250918619670343e-05,
+      "loss": 0.4577,
+      "step": 273
+    },
+    {
+      "epoch": 1.1093117408906883,
+      "grad_norm": 0.7690032997741356,
+      "learning_rate": 3.822771642257145e-05,
+      "loss": 0.4561,
+      "step": 274
+    },
+    {
+      "epoch": 1.1133603238866396,
+      "grad_norm": 0.5063914131538898,
+      "learning_rate": 3.820436847713776e-05,
+      "loss": 0.4419,
+      "step": 275
+    },
+    {
+      "epoch": 1.117408906882591,
+      "grad_norm": 0.6016761335545834,
+      "learning_rate": 3.8180874970058826e-05,
+      "loss": 0.4537,
+      "step": 276
+    },
+    {
+      "epoch": 1.1214574898785425,
+      "grad_norm": 0.673358289197658,
+      "learning_rate": 3.8157236089188103e-05,
+      "loss": 0.4477,
+      "step": 277
+    },
+    {
+      "epoch": 1.125506072874494,
+      "grad_norm": 0.4844366943377918,
+      "learning_rate": 3.813345202354145e-05,
+      "loss": 0.4531,
+      "step": 278
+    },
+    {
+      "epoch": 1.1295546558704452,
+      "grad_norm": 0.6528336405126467,
+      "learning_rate": 3.810952296329563e-05,
+      "loss": 0.4626,
+      "step": 279
+    },
+    {
+      "epoch": 1.1336032388663968,
+      "grad_norm": 0.5287026612962245,
+      "learning_rate": 3.808544909978676e-05,
+      "loss": 0.4421,
+      "step": 280
+    },
+    {
+      "epoch": 1.1376518218623481,
+      "grad_norm": 0.5282744988624211,
+      "learning_rate": 3.806123062550882e-05,
+      "loss": 0.4465,
+      "step": 281
+    },
+    {
+      "epoch": 1.1417004048582995,
+      "grad_norm": 0.47519547501537857,
+      "learning_rate": 3.803686773411208e-05,
+      "loss": 0.4563,
+      "step": 282
+    },
+    {
+      "epoch": 1.145748987854251,
+      "grad_norm": 0.5762244771189334,
+      "learning_rate": 3.801236062040158e-05,
+      "loss": 0.453,
+      "step": 283
+    },
+    {
+      "epoch": 1.1497975708502024,
+      "grad_norm": 0.4939088303199037,
+      "learning_rate": 3.798770948033553e-05,
+      "loss": 0.4627,
+      "step": 284
+    },
+    {
+      "epoch": 1.1538461538461537,
+      "grad_norm": 0.5022660607947965,
+      "learning_rate": 3.796291451102381e-05,
+      "loss": 0.4603,
+      "step": 285
+    },
+    {
+      "epoch": 1.1578947368421053,
+      "grad_norm": 0.46351431616284977,
+      "learning_rate": 3.793797591072631e-05,
+      "loss": 0.4478,
+      "step": 286
+    },
+    {
+      "epoch": 1.1619433198380567,
+      "grad_norm": 0.4224068656831366,
+      "learning_rate": 3.791289387885144e-05,
+      "loss": 0.4568,
+      "step": 287
+    },
+    {
+      "epoch": 1.165991902834008,
+      "grad_norm": 0.4628791338899818,
+      "learning_rate": 3.788766861595443e-05,
+      "loss": 0.4646,
+      "step": 288
+    },
+    {
+      "epoch": 1.1700404858299596,
+      "grad_norm": 0.4469136716055259,
+      "learning_rate": 3.7862300323735835e-05,
+      "loss": 0.4724,
+      "step": 289
+    },
+    {
+      "epoch": 1.174089068825911,
+      "grad_norm": 0.5146739880638586,
+      "learning_rate": 3.783678920503983e-05,
+      "loss": 0.4571,
+      "step": 290
+    },
+    {
+      "epoch": 1.1781376518218623,
+      "grad_norm": 0.41372791074240706,
+      "learning_rate": 3.781113546385265e-05,
+      "loss": 0.4401,
+      "step": 291
+    },
+    {
+      "epoch": 1.1821862348178138,
+      "grad_norm": 0.48895899900514483,
+      "learning_rate": 3.7785339305300916e-05,
+      "loss": 0.4658,
+      "step": 292
+    },
+    {
+      "epoch": 1.1862348178137652,
+      "grad_norm": 0.4038924484499077,
+      "learning_rate": 3.7759400935650025e-05,
+      "loss": 0.4765,
+      "step": 293
+    },
+    {
+      "epoch": 1.1902834008097165,
+      "grad_norm": 0.49375022667356216,
+      "learning_rate": 3.77333205623025e-05,
+      "loss": 0.4564,
+      "step": 294
+    },
+    {
+      "epoch": 1.194331983805668,
+      "grad_norm": 0.41403718755150476,
+      "learning_rate": 3.7707098393796295e-05,
+      "loss": 0.4475,
+      "step": 295
+    },
+    {
+      "epoch": 1.1983805668016194,
+      "grad_norm": 0.5350580304637399,
+      "learning_rate": 3.7680734639803185e-05,
+      "loss": 0.4619,
+      "step": 296
+    },
+    {
+      "epoch": 1.2024291497975708,
+      "grad_norm": 0.508844983113203,
+      "learning_rate": 3.765422951112704e-05,
+      "loss": 0.4653,
+      "step": 297
+    },
+    {
+      "epoch": 1.2064777327935223,
+      "grad_norm": 0.5183595509082722,
+      "learning_rate": 3.762758321970216e-05,
+      "loss": 0.4504,
+      "step": 298
+    },
+    {
+      "epoch": 1.2105263157894737,
+      "grad_norm": 0.5085065341352923,
+      "learning_rate": 3.7600795978591584e-05,
+      "loss": 0.4628,
+      "step": 299
+    },
+    {
+      "epoch": 1.214574898785425,
+      "grad_norm": 0.46075828778773936,
+      "learning_rate": 3.757386800198538e-05,
+      "loss": 0.4727,
+      "step": 300
+    },
+    {
+      "epoch": 1.2186234817813766,
+      "grad_norm": 0.492741697468255,
+      "learning_rate": 3.7546799505198925e-05,
+      "loss": 0.4557,
+      "step": 301
+    },
+    {
+      "epoch": 1.222672064777328,
+      "grad_norm": 0.48099059638011854,
+      "learning_rate": 3.7519590704671197e-05,
+      "loss": 0.456,
+      "step": 302
+    },
+    {
+      "epoch": 1.2267206477732793,
+      "grad_norm": 0.4463513028182158,
+      "learning_rate": 3.749224181796305e-05,
+      "loss": 0.4523,
+      "step": 303
+    },
+    {
+      "epoch": 1.2307692307692308,
+      "grad_norm": 0.5674990086791625,
+      "learning_rate": 3.746475306375545e-05,
+      "loss": 0.466,
+      "step": 304
+    },
+    {
+      "epoch": 1.2348178137651822,
+      "grad_norm": 0.46528601233453704,
+      "learning_rate": 3.743712466184774e-05,
+      "loss": 0.4609,
+      "step": 305
+    },
+    {
+      "epoch": 1.2388663967611335,
+      "grad_norm": 0.4609216822599589,
+      "learning_rate": 3.7409356833155885e-05,
+      "loss": 0.4478,
+      "step": 306
+    },
+    {
+      "epoch": 1.242914979757085,
+      "grad_norm": 0.481529742354291,
+      "learning_rate": 3.73814497997107e-05,
+      "loss": 0.4508,
+      "step": 307
+    },
+    {
+      "epoch": 1.2469635627530364,
+      "grad_norm": 0.4723765427263988,
+      "learning_rate": 3.7353403784656095e-05,
+      "loss": 0.4609,
+      "step": 308
+    },
+    {
+      "epoch": 1.2510121457489878,
+      "grad_norm": 0.4659689303965229,
+      "learning_rate": 3.732521901224724e-05,
+      "loss": 0.4482,
+      "step": 309
+    },
+    {
+      "epoch": 1.2550607287449393,
+      "grad_norm": 0.4907940302227885,
+      "learning_rate": 3.7296895707848844e-05,
+      "loss": 0.4528,
+      "step": 310
+    },
+    {
+      "epoch": 1.2591093117408907,
+      "grad_norm": 0.39307438823083635,
+      "learning_rate": 3.7268434097933275e-05,
+      "loss": 0.4518,
+      "step": 311
+    },
+    {
+      "epoch": 1.263157894736842,
+      "grad_norm": 0.41378770570751233,
+      "learning_rate": 3.723983441007882e-05,
+      "loss": 0.4401,
+      "step": 312
+    },
+    {
+      "epoch": 1.2672064777327936,
+      "grad_norm": 0.41917614301680384,
+      "learning_rate": 3.721109687296781e-05,
+      "loss": 0.4655,
+      "step": 313
+    },
+    {
+      "epoch": 1.271255060728745,
+      "grad_norm": 0.43715583600906444,
+      "learning_rate": 3.718222171638484e-05,
+      "loss": 0.4692,
+      "step": 314
+    },
+    {
+      "epoch": 1.2753036437246963,
+      "grad_norm": 0.47507758370230246,
+      "learning_rate": 3.715320917121488e-05,
+      "loss": 0.4788,
+      "step": 315
+    },
+    {
+      "epoch": 1.2793522267206479,
+      "grad_norm": 0.4306839516095726,
+      "learning_rate": 3.712405946944149e-05,
+      "loss": 0.4628,
+      "step": 316
+    },
+    {
+      "epoch": 1.2834008097165992,
+      "grad_norm": 0.5491297236095899,
+      "learning_rate": 3.709477284414488e-05,
+      "loss": 0.4677,
+      "step": 317
+    },
+    {
+      "epoch": 1.2874493927125505,
+      "grad_norm": 0.42812475565820957,
+      "learning_rate": 3.7065349529500136e-05,
+      "loss": 0.4456,
+      "step": 318
+    },
+    {
+      "epoch": 1.291497975708502,
+      "grad_norm": 0.525328771378448,
+      "learning_rate": 3.7035789760775305e-05,
+      "loss": 0.4411,
+      "step": 319
+    },
+    {
+      "epoch": 1.2955465587044535,
+      "grad_norm": 0.4357801751466913,
+      "learning_rate": 3.70060937743295e-05,
+      "loss": 0.4616,
+      "step": 320
+    },
+    {
+      "epoch": 1.2995951417004048,
+      "grad_norm": 0.47557078633387345,
+      "learning_rate": 3.6976261807611045e-05,
+      "loss": 0.4574,
+      "step": 321
+    },
+    {
+      "epoch": 1.3036437246963564,
+      "grad_norm": 0.5025291275724787,
+      "learning_rate": 3.694629409915555e-05,
+      "loss": 0.4426,
+      "step": 322
+    },
+    {
+      "epoch": 1.3076923076923077,
+      "grad_norm": 0.5130563821734926,
+      "learning_rate": 3.691619088858401e-05,
+      "loss": 0.456,
+      "step": 323
+    },
+    {
+      "epoch": 1.311740890688259,
+      "grad_norm": 0.5140664449895996,
+      "learning_rate": 3.68859524166009e-05,
+      "loss": 0.4644,
+      "step": 324
+    },
+    {
+      "epoch": 1.3157894736842106,
+      "grad_norm": 0.5812611015802189,
+      "learning_rate": 3.6855578924992226e-05,
+      "loss": 0.4682,
+      "step": 325
+    },
+    {
+      "epoch": 1.319838056680162,
+      "grad_norm": 0.45905855648478644,
+      "learning_rate": 3.6825070656623626e-05,
+      "loss": 0.4555,
+      "step": 326
+    },
+    {
+      "epoch": 1.3238866396761133,
+      "grad_norm": 0.4709055380419974,
+      "learning_rate": 3.67944278554384e-05,
+      "loss": 0.4531,
+      "step": 327
+    },
+    {
+      "epoch": 1.3279352226720649,
+      "grad_norm": 0.433740378564086,
+      "learning_rate": 3.676365076645557e-05,
+      "loss": 0.4487,
+      "step": 328
+    },
+    {
+      "epoch": 1.3319838056680162,
+      "grad_norm": 0.44011545391998835,
+      "learning_rate": 3.673273963576791e-05,
+      "loss": 0.4554,
+      "step": 329
+    },
+    {
+      "epoch": 1.3360323886639676,
+      "grad_norm": 0.4140634135390608,
+      "learning_rate": 3.670169471054001e-05,
+      "loss": 0.4567,
+      "step": 330
+    },
+    {
+      "epoch": 1.3400809716599191,
+      "grad_norm": 0.4854407617441044,
+      "learning_rate": 3.6670516239006254e-05,
+      "loss": 0.4506,
+      "step": 331
+    },
+    {
+      "epoch": 1.3441295546558705,
+      "grad_norm": 0.4215393761867604,
+      "learning_rate": 3.663920447046888e-05,
+      "loss": 0.4549,
+      "step": 332
+    },
+    {
+      "epoch": 1.3481781376518218,
+      "grad_norm": 0.47680585801970166,
+      "learning_rate": 3.6607759655295954e-05,
+      "loss": 0.456,
+      "step": 333
+    },
+    {
+      "epoch": 1.3522267206477734,
+      "grad_norm": 0.5545356647012596,
+      "learning_rate": 3.657618204491937e-05,
+      "loss": 0.4494,
+      "step": 334
+    },
+    {
+      "epoch": 1.3562753036437247,
+      "grad_norm": 0.5387553513533647,
+      "learning_rate": 3.6544471891832875e-05,
+      "loss": 0.4499,
+      "step": 335
+    },
+    {
+      "epoch": 1.360323886639676,
+      "grad_norm": 0.4922643748519295,
+      "learning_rate": 3.6512629449589983e-05,
+      "loss": 0.4586,
+      "step": 336
+    },
+    {
+      "epoch": 1.3643724696356276,
+      "grad_norm": 0.6438009631968139,
+      "learning_rate": 3.6480654972802045e-05,
+      "loss": 0.4506,
+      "step": 337
+    },
+    {
+      "epoch": 1.368421052631579,
+      "grad_norm": 0.44522082504936566,
+      "learning_rate": 3.644854871713611e-05,
+      "loss": 0.4634,
+      "step": 338
+    },
+    {
+      "epoch": 1.3724696356275303,
+      "grad_norm": 0.5606824557228997,
+      "learning_rate": 3.6416310939312924e-05,
+      "loss": 0.4493,
+      "step": 339
+    },
+    {
+      "epoch": 1.376518218623482,
+      "grad_norm": 0.4655710895382937,
+      "learning_rate": 3.638394189710493e-05,
+      "loss": 0.4524,
+      "step": 340
+    },
+    {
+      "epoch": 1.3805668016194332,
+      "grad_norm": 0.5082616038866344,
+      "learning_rate": 3.635144184933412e-05,
+      "loss": 0.4451,
+      "step": 341
+    },
+    {
+      "epoch": 1.3846153846153846,
+      "grad_norm": 0.45839659592331755,
+      "learning_rate": 3.6318811055869994e-05,
+      "loss": 0.476,
+      "step": 342
+    },
+    {
+      "epoch": 1.3886639676113361,
+      "grad_norm": 0.5167527307346722,
+      "learning_rate": 3.628604977762752e-05,
+      "loss": 0.4532,
+      "step": 343
+    },
+    {
+      "epoch": 1.3927125506072875,
+      "grad_norm": 0.4695158141354274,
+      "learning_rate": 3.6253158276565004e-05,
+      "loss": 0.4586,
+      "step": 344
+    },
+    {
+      "epoch": 1.3967611336032388,
+      "grad_norm": 0.4844521052837252,
+      "learning_rate": 3.6220136815682016e-05,
+      "loss": 0.4591,
+      "step": 345
+    },
+    {
+      "epoch": 1.4008097165991904,
+      "grad_norm": 0.49681254692948734,
+      "learning_rate": 3.618698565901727e-05,
+      "loss": 0.4375,
+      "step": 346
+    },
+    {
+      "epoch": 1.4048582995951417,
+      "grad_norm": 0.48317929924849795,
+      "learning_rate": 3.615370507164654e-05,
+      "loss": 0.4753,
+      "step": 347
+    },
+    {
+      "epoch": 1.408906882591093,
+      "grad_norm": 0.5332784959507774,
+      "learning_rate": 3.61202953196805e-05,
+      "loss": 0.4627,
+      "step": 348
+    },
+    {
+      "epoch": 1.4129554655870447,
+      "grad_norm": 0.5427369833275021,
+      "learning_rate": 3.6086756670262654e-05,
+      "loss": 0.4445,
+      "step": 349
+    },
+    {
+      "epoch": 1.417004048582996,
+      "grad_norm": 0.4096317609883185,
+      "learning_rate": 3.6053089391567135e-05,
+      "loss": 0.4529,
+      "step": 350
+    },
+    {
+      "epoch": 1.4210526315789473,
+      "grad_norm": 0.4889462776438449,
+      "learning_rate": 3.601929375279662e-05,
+      "loss": 0.4552,
+      "step": 351
+    },
+    {
+      "epoch": 1.425101214574899,
+      "grad_norm": 0.40937923366593765,
+      "learning_rate": 3.598537002418012e-05,
+      "loss": 0.453,
+      "step": 352
+    },
+    {
+      "epoch": 1.4291497975708503,
+      "grad_norm": 0.5225759289445184,
+      "learning_rate": 3.5951318476970885e-05,
+      "loss": 0.4511,
+      "step": 353
+    },
+    {
+      "epoch": 1.4331983805668016,
+      "grad_norm": 0.4723629280376106,
+      "learning_rate": 3.591713938344416e-05,
+      "loss": 0.4471,
+      "step": 354
+    },
+    {
+      "epoch": 1.4372469635627532,
+      "grad_norm": 0.5256042113138615,
+      "learning_rate": 3.588283301689507e-05,
+      "loss": 0.455,
+      "step": 355
+    },
+    {
+      "epoch": 1.4412955465587045,
+      "grad_norm": 0.4287549899629411,
+      "learning_rate": 3.5848399651636424e-05,
+      "loss": 0.4601,
+      "step": 356
+    },
+    {
+      "epoch": 1.4453441295546559,
+      "grad_norm": 0.4099135113305582,
+      "learning_rate": 3.5813839562996473e-05,
+      "loss": 0.4254,
+      "step": 357
+    },
+    {
+      "epoch": 1.4493927125506074,
+      "grad_norm": 0.42348989139876947,
+      "learning_rate": 3.5779153027316775e-05,
+      "loss": 0.4456,
+      "step": 358
+    },
+    {
+      "epoch": 1.4534412955465588,
+      "grad_norm": 0.431082958717516,
+      "learning_rate": 3.574434032194994e-05,
+      "loss": 0.4462,
+      "step": 359
+    },
+    {
+      "epoch": 1.45748987854251,
+      "grad_norm": 0.4553833506122703,
+      "learning_rate": 3.5709401725257436e-05,
+      "loss": 0.459,
+      "step": 360
+    },
+    {
+      "epoch": 1.4615384615384617,
+      "grad_norm": 0.49211234410259597,
+      "learning_rate": 3.5674337516607346e-05,
+      "loss": 0.4532,
+      "step": 361
+    },
+    {
+      "epoch": 1.465587044534413,
+      "grad_norm": 0.37268907761339004,
+      "learning_rate": 3.563914797637216e-05,
+      "loss": 0.4496,
+      "step": 362
+    },
+    {
+      "epoch": 1.4696356275303644,
+      "grad_norm": 0.4583923786310263,
+      "learning_rate": 3.560383338592649e-05,
+      "loss": 0.4482,
+      "step": 363
+    },
+    {
+      "epoch": 1.4736842105263157,
+      "grad_norm": 0.42124207077659914,
+      "learning_rate": 3.556839402764487e-05,
+      "loss": 0.4696,
+      "step": 364
+    },
+    {
+      "epoch": 1.4777327935222673,
+      "grad_norm": 0.43547395196206545,
+      "learning_rate": 3.553283018489946e-05,
+      "loss": 0.4503,
+      "step": 365
+    },
+    {
+      "epoch": 1.4817813765182186,
+      "grad_norm": 0.4822686066033594,
+      "learning_rate": 3.54971421420578e-05,
+      "loss": 0.4595,
+      "step": 366
+    },
+    {
+      "epoch": 1.48582995951417,
+      "grad_norm": 0.39530868557037485,
+      "learning_rate": 3.5461330184480524e-05,
+      "loss": 0.4523,
+      "step": 367
+    },
+    {
+      "epoch": 1.4898785425101215,
+      "grad_norm": 0.4145276793641515,
+      "learning_rate": 3.54253945985191e-05,
+      "loss": 0.441,
+      "step": 368
+    },
+    {
+      "epoch": 1.4939271255060729,
+      "grad_norm": 0.49540272085879217,
+      "learning_rate": 3.5389335671513524e-05,
+      "loss": 0.4311,
+      "step": 369
+    },
+    {
+      "epoch": 1.4979757085020242,
+      "grad_norm": 0.49998708727690977,
+      "learning_rate": 3.5353153691789986e-05,
+      "loss": 0.4531,
+      "step": 370
+    },
+    {
+      "epoch": 1.5020242914979756,
+      "grad_norm": 0.4785533040127285,
+      "learning_rate": 3.531684894865866e-05,
+      "loss": 0.4439,
+      "step": 371
+    },
+    {
+      "epoch": 1.5060728744939271,
+      "grad_norm": 0.56915121573665,
+      "learning_rate": 3.528042173241128e-05,
+      "loss": 0.4553,
+      "step": 372
+    },
+    {
+      "epoch": 1.5101214574898787,
+      "grad_norm": 0.42058120812263305,
+      "learning_rate": 3.524387233431892e-05,
+      "loss": 0.4508,
+      "step": 373
+    },
+    {
+      "epoch": 1.5141700404858298,
+      "grad_norm": 0.7201989951222124,
+      "learning_rate": 3.520720104662958e-05,
+      "loss": 0.4686,
+      "step": 374
+    },
+    {
+      "epoch": 1.5182186234817814,
+      "grad_norm": 0.45124997541107287,
+      "learning_rate": 3.5170408162565904e-05,
+      "loss": 0.4515,
+      "step": 375
+    },
+    {
+      "epoch": 1.522267206477733,
+      "grad_norm": 0.551946018555688,
+      "learning_rate": 3.513349397632281e-05,
+      "loss": 0.4562,
+      "step": 376
+    },
+    {
+      "epoch": 1.526315789473684,
+      "grad_norm": 0.5425515623384918,
+      "learning_rate": 3.5096458783065145e-05,
+      "loss": 0.4677,
+      "step": 377
+    },
+    {
+      "epoch": 1.5303643724696356,
+      "grad_norm": 0.4565938391414513,
+      "learning_rate": 3.505930287892533e-05,
+      "loss": 0.4444,
+      "step": 378
+    },
+    {
+      "epoch": 1.5344129554655872,
+      "grad_norm": 0.5677202867697491,
+      "learning_rate": 3.502202656100099e-05,
+      "loss": 0.4487,
+      "step": 379
+    },
+    {
+      "epoch": 1.5384615384615383,
+      "grad_norm": 0.46372917466480174,
+      "learning_rate": 3.498463012735256e-05,
+      "loss": 0.4612,
+      "step": 380
+    },
+    {
+      "epoch": 1.54251012145749,
+      "grad_norm": 0.44555393183602965,
+      "learning_rate": 3.494711387700094e-05,
+      "loss": 0.4353,
+      "step": 381
+    },
+    {
+      "epoch": 1.5465587044534415,
+      "grad_norm": 0.5264552572088259,
+      "learning_rate": 3.490947810992505e-05,
+      "loss": 0.4495,
+      "step": 382
+    },
+    {
+      "epoch": 1.5506072874493926,
+      "grad_norm": 0.465182515608989,
+      "learning_rate": 3.48717231270595e-05,
+      "loss": 0.4652,
+      "step": 383
+    },
+    {
+      "epoch": 1.5546558704453441,
+      "grad_norm": 0.4890218280358757,
+      "learning_rate": 3.483384923029211e-05,
+      "loss": 0.433,
+      "step": 384
+    },
+    {
+      "epoch": 1.5587044534412957,
+      "grad_norm": 0.4567865246141412,
+      "learning_rate": 3.479585672246156e-05,
+      "loss": 0.4355,
+      "step": 385
+    },
+    {
+      "epoch": 1.5627530364372468,
+      "grad_norm": 0.4245604780289654,
+      "learning_rate": 3.4757745907354924e-05,
+      "loss": 0.4557,
+      "step": 386
+    },
+    {
+      "epoch": 1.5668016194331984,
+      "grad_norm": 0.4303734360060921,
+      "learning_rate": 3.471951708970527e-05,
+      "loss": 0.4494,
+      "step": 387
+    },
+    {
+      "epoch": 1.5708502024291497,
+      "grad_norm": 0.39655768511787487,
+      "learning_rate": 3.468117057518921e-05,
+      "loss": 0.4522,
+      "step": 388
+    },
+    {
+      "epoch": 1.574898785425101,
+      "grad_norm": 0.4194012254242455,
+      "learning_rate": 3.4642706670424426e-05,
+      "loss": 0.4462,
+      "step": 389
+    },
+    {
+      "epoch": 1.5789473684210527,
+      "grad_norm": 0.39127899427292223,
+      "learning_rate": 3.460412568296731e-05,
+      "loss": 0.452,
+      "step": 390
+    },
+    {
+      "epoch": 1.582995951417004,
+      "grad_norm": 0.4819771477648677,
+      "learning_rate": 3.456542792131039e-05,
+      "loss": 0.4399,
+      "step": 391
+    },
+    {
+      "epoch": 1.5870445344129553,
+      "grad_norm": 0.37387353253779637,
+      "learning_rate": 3.4526613694879936e-05,
+      "loss": 0.4368,
+      "step": 392
+    },
+    {
+      "epoch": 1.591093117408907,
+      "grad_norm": 0.45073753842619463,
+      "learning_rate": 3.448768331403348e-05,
+      "loss": 0.4316,
+      "step": 393
+    },
+    {
+      "epoch": 1.5951417004048583,
+      "grad_norm": 0.37851678762843194,
+      "learning_rate": 3.44486370900573e-05,
+      "loss": 0.4471,
+      "step": 394
+    },
+    {
+      "epoch": 1.5991902834008096,
+      "grad_norm": 0.4161457598189429,
+      "learning_rate": 3.440947533516396e-05,
+      "loss": 0.4567,
+      "step": 395
+    },
+    {
+      "epoch": 1.6032388663967612,
+      "grad_norm": 0.3709473595576599,
+      "learning_rate": 3.4370198362489816e-05,
+      "loss": 0.4659,
+      "step": 396
+    },
+    {
+      "epoch": 1.6072874493927125,
+      "grad_norm": 0.4714348056763181,
+      "learning_rate": 3.43308064860925e-05,
+      "loss": 0.4474,
+      "step": 397
+    },
+    {
+      "epoch": 1.6113360323886639,
+      "grad_norm": 0.41177482536098003,
+      "learning_rate": 3.4291300020948393e-05,
+      "loss": 0.4421,
+      "step": 398
+    },
+    {
+      "epoch": 1.6153846153846154,
+      "grad_norm": 0.46671555797237907,
+      "learning_rate": 3.425167928295015e-05,
+      "loss": 0.4557,
+      "step": 399
+    },
+    {
+      "epoch": 1.6194331983805668,
+      "grad_norm": 0.4951146079399618,
+      "learning_rate": 3.421194458890411e-05,
+      "loss": 0.46,
+      "step": 400
+    },
+    {
+      "epoch": 1.623481781376518,
+      "grad_norm": 0.47461047917912264,
+      "learning_rate": 3.4172096256527865e-05,
+      "loss": 0.4369,
+      "step": 401
+    },
+    {
+      "epoch": 1.6275303643724697,
+      "grad_norm": 0.3828570938496667,
+      "learning_rate": 3.4132134604447595e-05,
+      "loss": 0.448,
+      "step": 402
+    },
+    {
+      "epoch": 1.631578947368421,
+      "grad_norm": 0.43937533391558825,
+      "learning_rate": 3.409205995219561e-05,
+      "loss": 0.4511,
+      "step": 403
+    },
+    {
+      "epoch": 1.6356275303643724,
+      "grad_norm": 0.4142648318808585,
+      "learning_rate": 3.4051872620207765e-05,
+      "loss": 0.4476,
+      "step": 404
+    },
+    {
+      "epoch": 1.639676113360324,
+      "grad_norm": 0.3724903754548483,
+      "learning_rate": 3.40115729298209e-05,
+      "loss": 0.447,
+      "step": 405
+    },
+    {
+      "epoch": 1.6437246963562753,
+      "grad_norm": 0.42907737408952035,
+      "learning_rate": 3.397116120327025e-05,
+      "loss": 0.4498,
+      "step": 406
+    },
+    {
+      "epoch": 1.6477732793522266,
+      "grad_norm": 0.434526486559879,
+      "learning_rate": 3.3930637763686916e-05,
+      "loss": 0.4367,
+      "step": 407
+    },
+    {
+      "epoch": 1.6518218623481782,
+      "grad_norm": 0.41991616566774004,
+      "learning_rate": 3.389000293509524e-05,
+      "loss": 0.4432,
+      "step": 408
+    },
+    {
+      "epoch": 1.6558704453441295,
+      "grad_norm": 0.5687081198998807,
+      "learning_rate": 3.3849257042410225e-05,
+      "loss": 0.4579,
+      "step": 409
+    },
+    {
+      "epoch": 1.6599190283400809,
+      "grad_norm": 0.46330723977300126,
+      "learning_rate": 3.380840041143494e-05,
+      "loss": 0.4621,
+      "step": 410
+    },
+    {
+      "epoch": 1.6639676113360324,
+      "grad_norm": 0.41272029624944684,
+      "learning_rate": 3.376743336885792e-05,
+      "loss": 0.4615,
+      "step": 411
+    },
+    {
+      "epoch": 1.6680161943319838,
+      "grad_norm": 0.46588938111311384,
+      "learning_rate": 3.372635624225055e-05,
+      "loss": 0.4457,
+      "step": 412
+    },
+    {
+      "epoch": 1.6720647773279351,
+      "grad_norm": 0.46901569380813096,
+      "learning_rate": 3.3685169360064445e-05,
+      "loss": 0.4341,
+      "step": 413
+    },
+    {
+      "epoch": 1.6761133603238867,
+      "grad_norm": 0.39099002115388537,
+      "learning_rate": 3.3643873051628804e-05,
+      "loss": 0.4644,
+      "step": 414
+    },
+    {
+      "epoch": 1.680161943319838,
+      "grad_norm": 0.43087035873684376,
+      "learning_rate": 3.3602467647147814e-05,
+      "loss": 0.4589,
+      "step": 415
+    },
+    {
+      "epoch": 1.6842105263157894,
+      "grad_norm": 0.5032238813626763,
+      "learning_rate": 3.356095347769799e-05,
+      "loss": 0.4396,
+      "step": 416
+    },
+    {
+      "epoch": 1.688259109311741,
+      "grad_norm": 0.4576147740557068,
+      "learning_rate": 3.351933087522552e-05,
+      "loss": 0.4595,
+      "step": 417
+    },
+    {
+      "epoch": 1.6923076923076923,
+      "grad_norm": 0.43745267189936266,
+      "learning_rate": 3.3477600172543626e-05,
+      "loss": 0.4426,
+      "step": 418
+    },
+    {
+      "epoch": 1.6963562753036436,
+      "grad_norm": 0.3991173234011665,
+      "learning_rate": 3.3435761703329894e-05,
+      "loss": 0.4634,
+      "step": 419
+    },
+    {
+      "epoch": 1.7004048582995952,
+      "grad_norm": 0.42863573926924486,
+      "learning_rate": 3.3393815802123606e-05,
+      "loss": 0.4578,
+      "step": 420
+    },
+    {
+      "epoch": 1.7044534412955465,
+      "grad_norm": 0.3482082292483338,
+      "learning_rate": 3.335176280432308e-05,
+      "loss": 0.4349,
+      "step": 421
+    },
+    {
+      "epoch": 1.708502024291498,
+      "grad_norm": 0.4635337399741752,
+      "learning_rate": 3.330960304618295e-05,
+      "loss": 0.4537,
+      "step": 422
+    },
+    {
+      "epoch": 1.7125506072874495,
+      "grad_norm": 0.3473173102619737,
+      "learning_rate": 3.326733686481153e-05,
+      "loss": 0.4541,
+      "step": 423
+    },
+    {
+      "epoch": 1.7165991902834008,
+      "grad_norm": 0.4075587027665116,
+      "learning_rate": 3.3224964598168084e-05,
+      "loss": 0.4312,
+      "step": 424
+    },
+    {
+      "epoch": 1.7206477732793521,
+      "grad_norm": 0.37077595228585924,
+      "learning_rate": 3.318248658506012e-05,
+      "loss": 0.458,
+      "step": 425
+    },
+    {
+      "epoch": 1.7246963562753037,
+      "grad_norm": 0.39243366312897177,
+      "learning_rate": 3.313990316514069e-05,
+      "loss": 0.4335,
+      "step": 426
+    },
+    {
+      "epoch": 1.728744939271255,
+      "grad_norm": 0.4123232528092884,
+      "learning_rate": 3.309721467890571e-05,
+      "loss": 0.4557,
+      "step": 427
+    },
+    {
+      "epoch": 1.7327935222672064,
+      "grad_norm": 0.4239598839493016,
+      "learning_rate": 3.305442146769115e-05,
+      "loss": 0.4474,
+      "step": 428
+    },
+    {
+      "epoch": 1.736842105263158,
+      "grad_norm": 0.3461111559411,
+      "learning_rate": 3.30115238736704e-05,
+      "loss": 0.442,
+      "step": 429
+    },
+    {
+      "epoch": 1.7408906882591093,
+      "grad_norm": 0.5198280910715624,
+      "learning_rate": 3.296852223985148e-05,
+      "loss": 0.45,
+      "step": 430
+    },
+    {
+      "epoch": 1.7449392712550607,
+      "grad_norm": 0.40804012762722286,
+      "learning_rate": 3.2925416910074315e-05,
+      "loss": 0.4489,
+      "step": 431
+    },
+    {
+      "epoch": 1.7489878542510122,
+      "grad_norm": 0.46543686551625635,
+      "learning_rate": 3.288220822900796e-05,
+      "loss": 0.454,
+      "step": 432
+    },
+    {
+      "epoch": 1.7530364372469636,
+      "grad_norm": 0.42512853450907856,
+      "learning_rate": 3.283889654214788e-05,
+      "loss": 0.4412,
+      "step": 433
+    },
+    {
+      "epoch": 1.757085020242915,
+      "grad_norm": 0.3495548808342774,
+      "learning_rate": 3.279548219581317e-05,
+      "loss": 0.4433,
+      "step": 434
+    },
+    {
+      "epoch": 1.7611336032388665,
+      "grad_norm": 0.42056901355071724,
+      "learning_rate": 3.275196553714379e-05,
+      "loss": 0.4469,
+      "step": 435
+    },
+    {
+      "epoch": 1.7651821862348178,
+      "grad_norm": 0.3774903055247291,
+      "learning_rate": 3.270834691409778e-05,
+      "loss": 0.4436,
+      "step": 436
+    },
+    {
+      "epoch": 1.7692307692307692,
+      "grad_norm": 0.34970601191885214,
+      "learning_rate": 3.266462667544849e-05,
+      "loss": 0.4586,
+      "step": 437
+    },
+    {
+      "epoch": 1.7732793522267207,
+      "grad_norm": 0.44005096198981497,
+      "learning_rate": 3.262080517078178e-05,
+      "loss": 0.455,
+      "step": 438
+    },
+    {
+      "epoch": 1.777327935222672,
+      "grad_norm": 0.34653988675735636,
+      "learning_rate": 3.257688275049323e-05,
+      "loss": 0.4529,
+      "step": 439
+    },
+    {
+      "epoch": 1.7813765182186234,
+      "grad_norm": 0.37048600461114817,
+      "learning_rate": 3.253285976578535e-05,
+      "loss": 0.4374,
+      "step": 440
+    },
+    {
+      "epoch": 1.785425101214575,
+      "grad_norm": 0.31723815868641075,
+      "learning_rate": 3.2488736568664756e-05,
+      "loss": 0.461,
+      "step": 441
+    },
+    {
+      "epoch": 1.7894736842105263,
+      "grad_norm": 0.36936574729489574,
+      "learning_rate": 3.244451351193935e-05,
+      "loss": 0.4345,
+      "step": 442
+    },
+    {
+      "epoch": 1.7935222672064777,
+      "grad_norm": 0.39139430113185264,
+      "learning_rate": 3.240019094921551e-05,
+      "loss": 0.4567,
+      "step": 443
+    },
+    {
+      "epoch": 1.7975708502024292,
+      "grad_norm": 0.3846881655549945,
+      "learning_rate": 3.235576923489527e-05,
+      "loss": 0.4588,
+      "step": 444
+    },
+    {
+      "epoch": 1.8016194331983806,
+      "grad_norm": 0.3604423565090035,
+      "learning_rate": 3.2311248724173476e-05,
+      "loss": 0.4427,
+      "step": 445
+    },
+    {
+      "epoch": 1.805668016194332,
+      "grad_norm": 0.34379514353125873,
+      "learning_rate": 3.226662977303494e-05,
+      "loss": 0.4456,
+      "step": 446
+    },
+    {
+      "epoch": 1.8097165991902835,
+      "grad_norm": 0.38412205049689624,
+      "learning_rate": 3.2221912738251626e-05,
+      "loss": 0.4444,
+      "step": 447
+    },
+    {
+      "epoch": 1.8137651821862348,
+      "grad_norm": 0.38616394363291334,
+      "learning_rate": 3.2177097977379736e-05,
+      "loss": 0.4509,
+      "step": 448
+    },
+    {
+      "epoch": 1.8178137651821862,
+      "grad_norm": 0.38349582850916764,
+      "learning_rate": 3.21321858487569e-05,
+      "loss": 0.443,
+      "step": 449
+    },
+    {
+      "epoch": 1.8218623481781377,
+      "grad_norm": 0.403247491709115,
+      "learning_rate": 3.208717671149931e-05,
+      "loss": 0.4363,
+      "step": 450
+    },
+    {
+      "epoch": 1.825910931174089,
+      "grad_norm": 0.4048898522561361,
+      "learning_rate": 3.204207092549882e-05,
+      "loss": 0.4571,
+      "step": 451
+    },
+    {
+      "epoch": 1.8299595141700404,
+      "grad_norm": 0.40285225627122395,
+      "learning_rate": 3.19968688514201e-05,
+      "loss": 0.4351,
+      "step": 452
+    },
+    {
+      "epoch": 1.834008097165992,
+      "grad_norm": 0.4620505059473818,
+      "learning_rate": 3.195157085069774e-05,
+      "loss": 0.4386,
+      "step": 453
+    },
+    {
+      "epoch": 1.8380566801619433,
+      "grad_norm": 0.3554313778046748,
+      "learning_rate": 3.190617728553332e-05,
+      "loss": 0.4488,
+      "step": 454
+    },
+    {
+      "epoch": 1.8421052631578947,
+      "grad_norm": 0.4686033681174479,
+      "learning_rate": 3.1860688518892606e-05,
+      "loss": 0.4609,
+      "step": 455
+    },
+    {
+      "epoch": 1.8461538461538463,
+      "grad_norm": 0.41525000403055606,
+      "learning_rate": 3.181510491450254e-05,
+      "loss": 0.4483,
+      "step": 456
+    },
+    {
+      "epoch": 1.8502024291497976,
+      "grad_norm": 0.4162993289071702,
+      "learning_rate": 3.176942683684842e-05,
+      "loss": 0.4379,
+      "step": 457
+    },
+    {
+      "epoch": 1.854251012145749,
+      "grad_norm": 0.4934698195822326,
+      "learning_rate": 3.1723654651170934e-05,
+      "loss": 0.4391,
+      "step": 458
+    },
+    {
+      "epoch": 1.8582995951417005,
+      "grad_norm": 0.38583284427476466,
+      "learning_rate": 3.167778872346327e-05,
+      "loss": 0.4194,
+      "step": 459
+    },
+    {
+      "epoch": 1.8623481781376519,
+      "grad_norm": 0.43454467715306866,
+      "learning_rate": 3.163182942046815e-05,
+      "loss": 0.4506,
+      "step": 460
+    },
+    {
+      "epoch": 1.8663967611336032,
+      "grad_norm": 0.44231868593988105,
+      "learning_rate": 3.158577710967494e-05,
+      "loss": 0.4465,
+      "step": 461
+    },
+    {
+      "epoch": 1.8704453441295548,
+      "grad_norm": 0.4122322965123194,
+      "learning_rate": 3.153963215931669e-05,
+      "loss": 0.4443,
+      "step": 462
+    },
+    {
+      "epoch": 1.874493927125506,
+      "grad_norm": 0.4701790216354679,
+      "learning_rate": 3.14933949383672e-05,
+      "loss": 0.4621,
+      "step": 463
+    },
+    {
+      "epoch": 1.8785425101214575,
+      "grad_norm": 0.37431910901107235,
+      "learning_rate": 3.144706581653805e-05,
+      "loss": 0.4538,
+      "step": 464
+    },
+    {
+      "epoch": 1.882591093117409,
+      "grad_norm": 0.48394811616632033,
+      "learning_rate": 3.140064516427566e-05,
+      "loss": 0.4365,
+      "step": 465
+    },
+    {
+      "epoch": 1.8866396761133604,
+      "grad_norm": 0.39819266805106307,
+      "learning_rate": 3.135413335275833e-05,
+      "loss": 0.4451,
+      "step": 466
+    },
+    {
+      "epoch": 1.8906882591093117,
+      "grad_norm": 0.39672228471363574,
+      "learning_rate": 3.130753075389327e-05,
+      "loss": 0.4377,
+      "step": 467
+    },
+    {
+      "epoch": 1.8947368421052633,
+      "grad_norm": 0.4380670426294559,
+      "learning_rate": 3.1260837740313625e-05,
+      "loss": 0.4361,
+      "step": 468
+    },
+    {
+      "epoch": 1.8987854251012146,
+      "grad_norm": 0.4802355465465301,
+      "learning_rate": 3.121405468537547e-05,
+      "loss": 0.466,
+      "step": 469
+    },
+    {
+      "epoch": 1.902834008097166,
+      "grad_norm": 0.4370553428123631,
+      "learning_rate": 3.116718196315488e-05,
+      "loss": 0.4602,
+      "step": 470
+    },
+    {
+      "epoch": 1.9068825910931175,
+      "grad_norm": 0.4182027897094699,
+      "learning_rate": 3.1120219948444894e-05,
+      "loss": 0.4365,
+      "step": 471
+    },
+    {
+      "epoch": 1.9109311740890689,
+      "grad_norm": 0.4509036231108313,
+      "learning_rate": 3.107316901675252e-05,
+      "loss": 0.4364,
+      "step": 472
+    },
+    {
+      "epoch": 1.9149797570850202,
+      "grad_norm": 0.3944205679712307,
+      "learning_rate": 3.1026029544295745e-05,
+      "loss": 0.4232,
+      "step": 473
+    },
+    {
+      "epoch": 1.9190283400809718,
+      "grad_norm": 0.5513478303789943,
+      "learning_rate": 3.097880190800057e-05,
+      "loss": 0.4559,
+      "step": 474
+    },
+    {
+      "epoch": 1.9230769230769231,
+      "grad_norm": 0.4102329762136008,
+      "learning_rate": 3.093148648549788e-05,
+      "loss": 0.4559,
+      "step": 475
+    },
+    {
+      "epoch": 1.9271255060728745,
+      "grad_norm": 0.49997030941246917,
+      "learning_rate": 3.088408365512055e-05,
+      "loss": 0.4452,
+      "step": 476
+    },
+    {
+      "epoch": 1.931174089068826,
+      "grad_norm": 0.43281054576663136,
+      "learning_rate": 3.083659379590034e-05,
+      "loss": 0.4604,
+      "step": 477
+    },
+    {
+      "epoch": 1.9352226720647774,
+      "grad_norm": 0.478693409610471,
+      "learning_rate": 3.0789017287564913e-05,
+      "loss": 0.4275,
+      "step": 478
+    },
+    {
+      "epoch": 1.9392712550607287,
+      "grad_norm": 0.42073269532387264,
+      "learning_rate": 3.074135451053475e-05,
+      "loss": 0.4282,
+      "step": 479
+    },
+    {
+      "epoch": 1.9433198380566803,
+      "grad_norm": 0.43967086483277956,
+      "learning_rate": 3.069360584592015e-05,
+      "loss": 0.4403,
+      "step": 480
+    },
+    {
+      "epoch": 1.9473684210526314,
+      "grad_norm": 0.4283023732321674,
+      "learning_rate": 3.064577167551816e-05,
+      "loss": 0.4483,
+      "step": 481
+    },
+    {
+      "epoch": 1.951417004048583,
+      "grad_norm": 0.4327970436998257,
+      "learning_rate": 3.059785238180953e-05,
+      "loss": 0.4501,
+      "step": 482
+    },
+    {
+      "epoch": 1.9554655870445345,
+      "grad_norm": 0.3550541828308529,
+      "learning_rate": 3.054984834795566e-05,
+      "loss": 0.4374,
+      "step": 483
+    },
+    {
+      "epoch": 1.9595141700404857,
+      "grad_norm": 0.445659806281629,
+      "learning_rate": 3.0501759957795537e-05,
+      "loss": 0.4536,
+      "step": 484
+    },
+    {
+      "epoch": 1.9635627530364372,
+      "grad_norm": 0.4685105169762957,
+      "learning_rate": 3.0453587595842627e-05,
+      "loss": 0.4544,
+      "step": 485
+    },
+    {
+      "epoch": 1.9676113360323888,
+      "grad_norm": 0.4602483095876176,
+      "learning_rate": 3.0405331647281856e-05,
+      "loss": 0.4568,
+      "step": 486
+    },
+    {
+      "epoch": 1.97165991902834,
+      "grad_norm": 0.5339556644403494,
+      "learning_rate": 3.0356992497966508e-05,
+      "loss": 0.4558,
+      "step": 487
+    },
+    {
+      "epoch": 1.9757085020242915,
+      "grad_norm": 0.39048814759801526,
+      "learning_rate": 3.0308570534415137e-05,
+      "loss": 0.4465,
+      "step": 488
+    },
+    {
+      "epoch": 1.979757085020243,
+      "grad_norm": 0.5093589774116066,
+      "learning_rate": 3.0260066143808467e-05,
+      "loss": 0.4434,
+      "step": 489
+    },
+    {
+      "epoch": 1.9838056680161942,
+      "grad_norm": 0.4613116925286783,
+      "learning_rate": 3.021147971398632e-05,
+      "loss": 0.435,
+      "step": 490
+    },
+    {
+      "epoch": 1.9878542510121457,
+      "grad_norm": 0.47258482981342753,
+      "learning_rate": 3.0162811633444478e-05,
+      "loss": 0.4371,
+      "step": 491
+    },
+    {
+      "epoch": 1.9919028340080973,
+      "grad_norm": 0.4848296847876648,
+      "learning_rate": 3.0114062291331635e-05,
+      "loss": 0.4498,
+      "step": 492
+    },
+    {
+      "epoch": 1.9959514170040484,
+      "grad_norm": 0.3884129521247856,
+      "learning_rate": 3.0065232077446217e-05,
+      "loss": 0.461,
+      "step": 493
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.4900388768015001,
+      "learning_rate": 3.001632138223332e-05,
+      "loss": 0.4308,
+      "step": 494
+    },
+    {
+      "epoch": 2.0040485829959516,
+      "grad_norm": 0.5507101857069697,
+      "learning_rate": 2.9967330596781565e-05,
+      "loss": 0.3424,
+      "step": 495
+    },
+    {
+      "epoch": 2.0080971659919027,
+      "grad_norm": 0.3879718818253577,
+      "learning_rate": 2.9918260112819952e-05,
+      "loss": 0.332,
+      "step": 496
+    },
+    {
+      "epoch": 2.0121457489878543,
+      "grad_norm": 0.8744803440722018,
+      "learning_rate": 2.9869110322714778e-05,
+      "loss": 0.3557,
+      "step": 497
+    },
+    {
+      "epoch": 2.016194331983806,
+      "grad_norm": 0.4893006266969581,
+      "learning_rate": 2.9819881619466447e-05,
+      "loss": 0.3581,
+      "step": 498
+    },
+    {
+      "epoch": 2.020242914979757,
+      "grad_norm": 0.5261932413619123,
+      "learning_rate": 2.9770574396706364e-05,
+      "loss": 0.3555,
+      "step": 499
+    },
+    {
+      "epoch": 2.0242914979757085,
+      "grad_norm": 0.43769378386216784,
+      "learning_rate": 2.972118904869376e-05,
+      "loss": 0.3642,
+      "step": 500
+    },
+    {
+      "epoch": 2.02834008097166,
+      "grad_norm": 0.43404283158922685,
+      "learning_rate": 2.9671725970312563e-05,
+      "loss": 0.3462,
+      "step": 501
+    },
+    {
+      "epoch": 2.032388663967611,
+      "grad_norm": 0.4388651409762131,
+      "learning_rate": 2.9622185557068238e-05,
+      "loss": 0.3462,
+      "step": 502
+    },
+    {
+      "epoch": 2.0364372469635628,
+      "grad_norm": 0.47239787141299805,
+      "learning_rate": 2.95725682050846e-05,
+      "loss": 0.3478,
+      "step": 503
+    },
+    {
+      "epoch": 2.0404858299595143,
+      "grad_norm": 0.40756640146424883,
+      "learning_rate": 2.9522874311100684e-05,
+      "loss": 0.3396,
+      "step": 504
+    },
+    {
+      "epoch": 2.0445344129554655,
+      "grad_norm": 0.512100493557915,
+      "learning_rate": 2.9473104272467526e-05,
+      "loss": 0.3517,
+      "step": 505
+    },
+    {
+      "epoch": 2.048582995951417,
+      "grad_norm": 0.38828696577421135,
+      "learning_rate": 2.942325848714505e-05,
+      "loss": 0.3543,
+      "step": 506
+    },
+    {
+      "epoch": 2.0526315789473686,
+      "grad_norm": 0.40614918042756054,
+      "learning_rate": 2.9373337353698804e-05,
+      "loss": 0.3423,
+      "step": 507
+    },
+    {
+      "epoch": 2.0566801619433197,
+      "grad_norm": 0.41389617606543433,
+      "learning_rate": 2.932334127129686e-05,
+      "loss": 0.35,
+      "step": 508
+    },
+    {
+      "epoch": 2.0607287449392713,
+      "grad_norm": 0.44043895146787854,
+      "learning_rate": 2.9273270639706547e-05,
+      "loss": 0.344,
+      "step": 509
+    },
+    {
+      "epoch": 2.064777327935223,
+      "grad_norm": 0.44062570900983594,
+      "learning_rate": 2.922312585929131e-05,
+      "loss": 0.3288,
+      "step": 510
+    },
+    {
+      "epoch": 2.068825910931174,
+      "grad_norm": 0.4253996005003732,
+      "learning_rate": 2.9172907331007478e-05,
+      "loss": 0.34,
+      "step": 511
+    },
+    {
+      "epoch": 2.0728744939271255,
+      "grad_norm": 0.470048749786572,
+      "learning_rate": 2.9122615456401053e-05,
+      "loss": 0.351,
+      "step": 512
+    },
+    {
+      "epoch": 2.076923076923077,
+      "grad_norm": 0.4373821651538964,
+      "learning_rate": 2.9072250637604535e-05,
+      "loss": 0.3234,
+      "step": 513
+    },
+    {
+      "epoch": 2.080971659919028,
+      "grad_norm": 0.43747180076340053,
+      "learning_rate": 2.9021813277333667e-05,
+      "loss": 0.3385,
+      "step": 514
+    },
+    {
+      "epoch": 2.08502024291498,
+      "grad_norm": 0.48435953961795264,
+      "learning_rate": 2.8971303778884244e-05,
+      "loss": 0.3419,
+      "step": 515
+    },
+    {
+      "epoch": 2.0890688259109313,
+      "grad_norm": 0.45204189770601577,
+      "learning_rate": 2.8920722546128863e-05,
+      "loss": 0.3425,
+      "step": 516
+    },
+    {
+      "epoch": 2.0931174089068825,
+      "grad_norm": 0.4453817403721837,
+      "learning_rate": 2.887006998351372e-05,
+      "loss": 0.3611,
+      "step": 517
+    },
+    {
+      "epoch": 2.097165991902834,
+      "grad_norm": 0.36931681705132163,
+      "learning_rate": 2.881934649605535e-05,
+      "loss": 0.3386,
+      "step": 518
+    },
+    {
+      "epoch": 2.1012145748987856,
+      "grad_norm": 0.47030084078302226,
+      "learning_rate": 2.8768552489337408e-05,
+      "loss": 0.3512,
+      "step": 519
+    },
+    {
+      "epoch": 2.1052631578947367,
+      "grad_norm": 0.4078601508057191,
+      "learning_rate": 2.8717688369507425e-05,
+      "loss": 0.346,
+      "step": 520
+    },
+    {
+      "epoch": 2.1093117408906883,
+      "grad_norm": 0.43345242440588244,
+      "learning_rate": 2.8666754543273535e-05,
+      "loss": 0.3441,
+      "step": 521
+    },
+    {
+      "epoch": 2.11336032388664,
+      "grad_norm": 0.41420305060860874,
+      "learning_rate": 2.8615751417901273e-05,
+      "loss": 0.3511,
+      "step": 522
+    },
+    {
+      "epoch": 2.117408906882591,
+      "grad_norm": 0.4154000622315714,
+      "learning_rate": 2.856467940121025e-05,
+      "loss": 0.3438,
+      "step": 523
+    },
+    {
+      "epoch": 2.1214574898785425,
+      "grad_norm": 0.3913586473761936,
+      "learning_rate": 2.8513538901570964e-05,
+      "loss": 0.3503,
+      "step": 524
+    },
+    {
+      "epoch": 2.125506072874494,
+      "grad_norm": 0.37088193482370824,
+      "learning_rate": 2.846233032790149e-05,
+      "loss": 0.3343,
+      "step": 525
+    },
+    {
+      "epoch": 2.1295546558704452,
+      "grad_norm": 0.39694573610261846,
+      "learning_rate": 2.84110540896642e-05,
+      "loss": 0.34,
+      "step": 526
+    },
+    {
+      "epoch": 2.133603238866397,
+      "grad_norm": 0.4649195582832834,
+      "learning_rate": 2.8359710596862555e-05,
+      "loss": 0.3415,
+      "step": 527
+    },
+    {
+      "epoch": 2.1376518218623484,
+      "grad_norm": 0.3845156496940679,
+      "learning_rate": 2.8308300260037734e-05,
+      "loss": 0.3412,
+      "step": 528
+    },
+    {
+      "epoch": 2.1417004048582995,
+      "grad_norm": 0.43384515712777005,
+      "learning_rate": 2.8256823490265424e-05,
+      "loss": 0.3251,
+      "step": 529
+    },
+    {
+      "epoch": 2.145748987854251,
+      "grad_norm": 0.4101446040034745,
+      "learning_rate": 2.8205280699152515e-05,
+      "loss": 0.3571,
+      "step": 530
+    },
+    {
+      "epoch": 2.1497975708502026,
+      "grad_norm": 0.3809669861431564,
+      "learning_rate": 2.8153672298833777e-05,
+      "loss": 0.3454,
+      "step": 531
+    },
+    {
+      "epoch": 2.1538461538461537,
+      "grad_norm": 0.435053463033773,
+      "learning_rate": 2.8101998701968615e-05,
+      "loss": 0.346,
+      "step": 532
+    },
+    {
+      "epoch": 2.1578947368421053,
+      "grad_norm": 0.3752140548797898,
+      "learning_rate": 2.805026032173773e-05,
+      "loss": 0.3574,
+      "step": 533
+    },
+    {
+      "epoch": 2.161943319838057,
+      "grad_norm": 0.45291622505722595,
+      "learning_rate": 2.7998457571839832e-05,
+      "loss": 0.3436,
+      "step": 534
+    },
+    {
+      "epoch": 2.165991902834008,
+      "grad_norm": 0.39222393876239126,
+      "learning_rate": 2.7946590866488322e-05,
+      "loss": 0.3385,
+      "step": 535
+    },
+    {
+      "epoch": 2.1700404858299596,
+      "grad_norm": 0.3950543106007906,
+      "learning_rate": 2.7894660620408e-05,
+      "loss": 0.342,
+      "step": 536
+    },
+    {
+      "epoch": 2.174089068825911,
+      "grad_norm": 0.4007001359526838,
+      "learning_rate": 2.784266724883173e-05,
+      "loss": 0.347,
+      "step": 537
+    },
+    {
+      "epoch": 2.1781376518218623,
+      "grad_norm": 0.38976930716766794,
+      "learning_rate": 2.7790611167497128e-05,
+      "loss": 0.3298,
+      "step": 538
+    },
+    {
+      "epoch": 2.182186234817814,
+      "grad_norm": 0.4184460931227305,
+      "learning_rate": 2.7738492792643224e-05,
+      "loss": 0.3492,
+      "step": 539
+    },
+    {
+      "epoch": 2.1862348178137654,
+      "grad_norm": 0.40861884209412114,
+      "learning_rate": 2.768631254100716e-05,
+      "loss": 0.3589,
+      "step": 540
+    },
+    {
+      "epoch": 2.1902834008097165,
+      "grad_norm": 0.38874575234451086,
+      "learning_rate": 2.7634070829820836e-05,
+      "loss": 0.3556,
+      "step": 541
+    },
+    {
+      "epoch": 2.194331983805668,
+      "grad_norm": 0.42903284381961587,
+      "learning_rate": 2.7581768076807592e-05,
+      "loss": 0.3519,
+      "step": 542
+    },
+    {
+      "epoch": 2.1983805668016196,
+      "grad_norm": 0.34489904308249736,
+      "learning_rate": 2.7529404700178832e-05,
+      "loss": 0.326,
+      "step": 543
+    },
+    {
+      "epoch": 2.2024291497975708,
+      "grad_norm": 0.44409508108506585,
+      "learning_rate": 2.7476981118630735e-05,
+      "loss": 0.3544,
+      "step": 544
+    },
+    {
+      "epoch": 2.2064777327935223,
+      "grad_norm": 0.3451311033556776,
+      "learning_rate": 2.742449775134085e-05,
+      "loss": 0.3361,
+      "step": 545
+    },
+    {
+      "epoch": 2.2105263157894735,
+      "grad_norm": 0.45418125858293296,
+      "learning_rate": 2.7371955017964788e-05,
+      "loss": 0.3488,
+      "step": 546
+    },
+    {
+      "epoch": 2.214574898785425,
+      "grad_norm": 0.3430827837974501,
+      "learning_rate": 2.7319353338632845e-05,
+      "loss": 0.3438,
+      "step": 547
+    },
+    {
+      "epoch": 2.2186234817813766,
+      "grad_norm": 0.3949013508645307,
+      "learning_rate": 2.7266693133946645e-05,
+      "loss": 0.3397,
+      "step": 548
+    },
+    {
+      "epoch": 2.2226720647773277,
+      "grad_norm": 0.47462024446519874,
+      "learning_rate": 2.7213974824975785e-05,
+      "loss": 0.3579,
+      "step": 549
+    },
+    {
+      "epoch": 2.2267206477732793,
+      "grad_norm": 0.4143167819867855,
+      "learning_rate": 2.7161198833254464e-05,
+      "loss": 0.3441,
+      "step": 550
+    },
+    {
+      "epoch": 2.230769230769231,
+      "grad_norm": 0.46236487106575963,
+      "learning_rate": 2.7108365580778095e-05,
+      "loss": 0.359,
+      "step": 551
+    },
+    {
+      "epoch": 2.234817813765182,
+      "grad_norm": 0.3469917008191411,
+      "learning_rate": 2.705547548999996e-05,
+      "loss": 0.3469,
+      "step": 552
+    },
+    {
+      "epoch": 2.2388663967611335,
+      "grad_norm": 0.46583689713782095,
+      "learning_rate": 2.7002528983827817e-05,
+      "loss": 0.3535,
+      "step": 553
+    },
+    {
+      "epoch": 2.242914979757085,
+      "grad_norm": 0.4025818811594365,
+      "learning_rate": 2.6949526485620517e-05,
+      "loss": 0.3516,
+      "step": 554
+    },
+    {
+      "epoch": 2.246963562753036,
+      "grad_norm": 0.4100335565662676,
+      "learning_rate": 2.6896468419184623e-05,
+      "loss": 0.3529,
+      "step": 555
+    },
+    {
+      "epoch": 2.251012145748988,
+      "grad_norm": 0.41541853428933734,
+      "learning_rate": 2.6843355208771024e-05,
+      "loss": 0.3488,
+      "step": 556
+    },
+    {
+      "epoch": 2.2550607287449393,
+      "grad_norm": 0.36257428220155596,
+      "learning_rate": 2.6790187279071536e-05,
+      "loss": 0.347,
+      "step": 557
+    },
+    {
+      "epoch": 2.2591093117408905,
+      "grad_norm": 0.43814369648146095,
+      "learning_rate": 2.67369650552155e-05,
+      "loss": 0.3631,
+      "step": 558
+    },
+    {
+      "epoch": 2.263157894736842,
+      "grad_norm": 0.36766774573350963,
+      "learning_rate": 2.6683688962766407e-05,
+      "loss": 0.3541,
+      "step": 559
+    },
+    {
+      "epoch": 2.2672064777327936,
+      "grad_norm": 0.3529844721810865,
+      "learning_rate": 2.6630359427718468e-05,
+      "loss": 0.3612,
+      "step": 560
+    },
+    {
+      "epoch": 2.2712550607287447,
+      "grad_norm": 0.3329630652327371,
+      "learning_rate": 2.6576976876493225e-05,
+      "loss": 0.3504,
+      "step": 561
+    },
+    {
+      "epoch": 2.2753036437246963,
+      "grad_norm": 0.3477964857023041,
+      "learning_rate": 2.652354173593614e-05,
+      "loss": 0.333,
+      "step": 562
+    },
+    {
+      "epoch": 2.279352226720648,
+      "grad_norm": 0.3663826784185901,
+      "learning_rate": 2.647005443331317e-05,
+      "loss": 0.3261,
+      "step": 563
+    },
+    {
+      "epoch": 2.283400809716599,
+      "grad_norm": 0.34866907880373327,
+      "learning_rate": 2.6416515396307354e-05,
+      "loss": 0.3645,
+      "step": 564
+    },
+    {
+      "epoch": 2.2874493927125505,
+      "grad_norm": 0.3968600684586525,
+      "learning_rate": 2.636292505301542e-05,
+      "loss": 0.352,
+      "step": 565
+    },
+    {
+      "epoch": 2.291497975708502,
+      "grad_norm": 0.36750816253736346,
+      "learning_rate": 2.630928383194431e-05,
+      "loss": 0.3507,
+      "step": 566
+    },
+    {
+      "epoch": 2.2955465587044532,
+      "grad_norm": 0.36602067334841415,
+      "learning_rate": 2.625559216200781e-05,
+      "loss": 0.3539,
+      "step": 567
+    },
+    {
+      "epoch": 2.299595141700405,
+      "grad_norm": 0.37785772928061906,
+      "learning_rate": 2.620185047252308e-05,
+      "loss": 0.375,
+      "step": 568
+    },
+    {
+      "epoch": 2.3036437246963564,
+      "grad_norm": 0.37190397317503926,
+      "learning_rate": 2.614805919320723e-05,
+      "loss": 0.3662,
+      "step": 569
+    },
+    {
+      "epoch": 2.3076923076923075,
+      "grad_norm": 0.3498655172963449,
+      "learning_rate": 2.6094218754173895e-05,
+      "loss": 0.3461,
+      "step": 570
+    },
+    {
+      "epoch": 2.311740890688259,
+      "grad_norm": 0.3464906706483118,
+      "learning_rate": 2.60403295859298e-05,
+      "loss": 0.3458,
+      "step": 571
+    },
+    {
+      "epoch": 2.3157894736842106,
+      "grad_norm": 0.38740242992397556,
+      "learning_rate": 2.5986392119371297e-05,
+      "loss": 0.3462,
+      "step": 572
+    },
+    {
+      "epoch": 2.3198380566801617,
+      "grad_norm": 0.3535659178280249,
+      "learning_rate": 2.5932406785780928e-05,
+      "loss": 0.3523,
+      "step": 573
+    },
+    {
+      "epoch": 2.3238866396761133,
+      "grad_norm": 0.33051725074960336,
+      "learning_rate": 2.5878374016823987e-05,
+      "loss": 0.352,
+      "step": 574
+    },
+    {
+      "epoch": 2.327935222672065,
+      "grad_norm": 0.3821375769031829,
+      "learning_rate": 2.5824294244545045e-05,
+      "loss": 0.3531,
+      "step": 575
+    },
+    {
+      "epoch": 2.331983805668016,
+      "grad_norm": 0.33891514437393677,
+      "learning_rate": 2.577016790136453e-05,
+      "loss": 0.35,
+      "step": 576
+    },
+    {
+      "epoch": 2.3360323886639676,
+      "grad_norm": 0.37118393047132003,
+      "learning_rate": 2.5715995420075225e-05,
+      "loss": 0.3491,
+      "step": 577
+    },
+    {
+      "epoch": 2.340080971659919,
+      "grad_norm": 0.3591476881335863,
+      "learning_rate": 2.5661777233838864e-05,
+      "loss": 0.3493,
+      "step": 578
+    },
+    {
+      "epoch": 2.3441295546558703,
+      "grad_norm": 0.35998231876013875,
+      "learning_rate": 2.5607513776182616e-05,
+      "loss": 0.3338,
+      "step": 579
+    },
+    {
+      "epoch": 2.348178137651822,
+      "grad_norm": 0.4115655758056906,
+      "learning_rate": 2.5553205480995633e-05,
+      "loss": 0.3541,
+      "step": 580
+    },
+    {
+      "epoch": 2.3522267206477734,
+      "grad_norm": 0.35650051184545484,
+      "learning_rate": 2.5498852782525604e-05,
+      "loss": 0.3384,
+      "step": 581
+    },
+    {
+      "epoch": 2.3562753036437245,
+      "grad_norm": 0.36306804255030556,
+      "learning_rate": 2.5444456115375252e-05,
+      "loss": 0.3435,
+      "step": 582
+    },
+    {
+      "epoch": 2.360323886639676,
+      "grad_norm": 0.37637934112324856,
+      "learning_rate": 2.539001591449889e-05,
+      "loss": 0.3406,
+      "step": 583
+    },
+    {
+      "epoch": 2.3643724696356276,
+      "grad_norm": 0.3861181381712958,
+      "learning_rate": 2.53355326151989e-05,
+      "loss": 0.3408,
+      "step": 584
+    },
+    {
+      "epoch": 2.3684210526315788,
+      "grad_norm": 0.36486236953343265,
+      "learning_rate": 2.5281006653122288e-05,
+      "loss": 0.3489,
+      "step": 585
+    },
+    {
+      "epoch": 2.3724696356275303,
+      "grad_norm": 0.3609048563224692,
+      "learning_rate": 2.5226438464257217e-05,
+      "loss": 0.3354,
+      "step": 586
+    },
+    {
+      "epoch": 2.376518218623482,
+      "grad_norm": 0.33511225917968473,
+      "learning_rate": 2.5171828484929447e-05,
+      "loss": 0.3466,
+      "step": 587
+    },
+    {
+      "epoch": 2.380566801619433,
+      "grad_norm": 0.34185451857551497,
+      "learning_rate": 2.5117177151798934e-05,
+      "loss": 0.3672,
+      "step": 588
+    },
+    {
+      "epoch": 2.3846153846153846,
+      "grad_norm": 0.36483088920155415,
+      "learning_rate": 2.506248490185627e-05,
+      "loss": 0.3538,
+      "step": 589
+    },
+    {
+      "epoch": 2.388663967611336,
+      "grad_norm": 0.3243193955482822,
+      "learning_rate": 2.500775217241923e-05,
+      "loss": 0.3608,
+      "step": 590
+    },
+    {
+      "epoch": 2.3927125506072873,
+      "grad_norm": 0.3462447910825581,
+      "learning_rate": 2.4952979401129266e-05,
+      "loss": 0.3379,
+      "step": 591
+    },
+    {
+      "epoch": 2.396761133603239,
+      "grad_norm": 0.3668402626870958,
+      "learning_rate": 2.4898167025948e-05,
+      "loss": 0.3447,
+      "step": 592
+    },
+    {
+      "epoch": 2.4008097165991904,
+      "grad_norm": 0.3674940506574202,
+      "learning_rate": 2.4843315485153703e-05,
+      "loss": 0.3511,
+      "step": 593
+    },
+    {
+      "epoch": 2.4048582995951415,
+      "grad_norm": 0.33731951933472587,
+      "learning_rate": 2.4788425217337856e-05,
+      "loss": 0.3428,
+      "step": 594
+    },
+    {
+      "epoch": 2.408906882591093,
+      "grad_norm": 0.33635001320736585,
+      "learning_rate": 2.4733496661401572e-05,
+      "loss": 0.3499,
+      "step": 595
+    },
+    {
+      "epoch": 2.4129554655870447,
+      "grad_norm": 0.34702616883019916,
+      "learning_rate": 2.4678530256552114e-05,
+      "loss": 0.3503,
+      "step": 596
+    },
+    {
+      "epoch": 2.417004048582996,
+      "grad_norm": 0.3368799318269111,
+      "learning_rate": 2.46235264422994e-05,
+      "loss": 0.3393,
+      "step": 597
+    },
+    {
+      "epoch": 2.4210526315789473,
+      "grad_norm": 0.37837705044988557,
+      "learning_rate": 2.4568485658452452e-05,
+      "loss": 0.3335,
+      "step": 598
+    },
+    {
+      "epoch": 2.425101214574899,
+      "grad_norm": 0.3784155341390421,
+      "learning_rate": 2.451340834511592e-05,
+      "loss": 0.3555,
+      "step": 599
+    },
+    {
+      "epoch": 2.42914979757085,
+      "grad_norm": 0.34623063434006285,
+      "learning_rate": 2.4458294942686528e-05,
+      "loss": 0.3534,
+      "step": 600
+    },
+    {
+      "epoch": 2.4331983805668016,
+      "grad_norm": 0.39804642564098114,
+      "learning_rate": 2.440314589184958e-05,
+      "loss": 0.3483,
+      "step": 601
+    },
+    {
+      "epoch": 2.437246963562753,
+      "grad_norm": 0.33898516781308935,
+      "learning_rate": 2.434796163357541e-05,
+      "loss": 0.3622,
+      "step": 602
+    },
+    {
+      "epoch": 2.4412955465587043,
+      "grad_norm": 0.3911583030395931,
+      "learning_rate": 2.429274260911588e-05,
+      "loss": 0.3566,
+      "step": 603
+    },
+    {
+      "epoch": 2.445344129554656,
+      "grad_norm": 0.3461297082333668,
+      "learning_rate": 2.423748926000084e-05,
+      "loss": 0.3522,
+      "step": 604
+    },
+    {
+      "epoch": 2.4493927125506074,
+      "grad_norm": 0.37118156174815786,
+      "learning_rate": 2.418220202803459e-05,
+      "loss": 0.353,
+      "step": 605
+    },
+    {
+      "epoch": 2.4534412955465585,
+      "grad_norm": 0.30264198663106845,
+      "learning_rate": 2.412688135529237e-05,
+      "loss": 0.3567,
+      "step": 606
+    },
+    {
+      "epoch": 2.45748987854251,
+      "grad_norm": 0.3768285223697634,
+      "learning_rate": 2.4071527684116807e-05,
+      "loss": 0.3484,
+      "step": 607
+    },
+    {
+      "epoch": 2.4615384615384617,
+      "grad_norm": 0.32717809277931514,
+      "learning_rate": 2.4016141457114372e-05,
+      "loss": 0.3455,
+      "step": 608
+    },
+    {
+      "epoch": 2.465587044534413,
+      "grad_norm": 0.34904207429999606,
+      "learning_rate": 2.396072311715187e-05,
+      "loss": 0.3471,
+      "step": 609
+    },
+    {
+      "epoch": 2.4696356275303644,
+      "grad_norm": 0.3390790674701622,
+      "learning_rate": 2.3905273107352863e-05,
+      "loss": 0.3458,
+      "step": 610
+    },
+    {
+      "epoch": 2.473684210526316,
+      "grad_norm": 0.3379252861157204,
+      "learning_rate": 2.3849791871094157e-05,
+      "loss": 0.3497,
+      "step": 611
+    },
+    {
+      "epoch": 2.477732793522267,
+      "grad_norm": 0.33014989453014293,
+      "learning_rate": 2.379427985200223e-05,
+      "loss": 0.3352,
+      "step": 612
+    },
+    {
+      "epoch": 2.4817813765182186,
+      "grad_norm": 0.35021468861560795,
+      "learning_rate": 2.373873749394972e-05,
+      "loss": 0.3519,
+      "step": 613
+    },
+    {
+      "epoch": 2.48582995951417,
+      "grad_norm": 0.327651312675687,
+      "learning_rate": 2.3683165241051836e-05,
+      "loss": 0.342,
+      "step": 614
+    },
+    {
+      "epoch": 2.4898785425101213,
+      "grad_norm": 0.37999643201958205,
+      "learning_rate": 2.3627563537662823e-05,
+      "loss": 0.3665,
+      "step": 615
+    },
+    {
+      "epoch": 2.493927125506073,
+      "grad_norm": 0.37105945094993986,
+      "learning_rate": 2.3571932828372433e-05,
+      "loss": 0.3582,
+      "step": 616
+    },
+    {
+      "epoch": 2.4979757085020244,
+      "grad_norm": 0.33969507587939707,
+      "learning_rate": 2.351627355800233e-05,
+      "loss": 0.3469,
+      "step": 617
+    },
+    {
+      "epoch": 2.5020242914979756,
+      "grad_norm": 0.3335833769101476,
+      "learning_rate": 2.346058617160256e-05,
+      "loss": 0.3518,
+      "step": 618
+    },
+    {
+      "epoch": 2.506072874493927,
+      "grad_norm": 0.3612622508207503,
+      "learning_rate": 2.3404871114447976e-05,
+      "loss": 0.3612,
+      "step": 619
+    },
+    {
+      "epoch": 2.5101214574898787,
+      "grad_norm": 0.33065558169980497,
+      "learning_rate": 2.334912883203469e-05,
+      "loss": 0.3639,
+      "step": 620
+    },
+    {
+      "epoch": 2.51417004048583,
+      "grad_norm": 0.38613105738450004,
+      "learning_rate": 2.3293359770076505e-05,
+      "loss": 0.34,
+      "step": 621
+    },
+    {
+      "epoch": 2.5182186234817814,
+      "grad_norm": 0.3462382326277488,
+      "learning_rate": 2.323756437450136e-05,
+      "loss": 0.3527,
+      "step": 622
+    },
+    {
+      "epoch": 2.522267206477733,
+      "grad_norm": 0.352991768627804,
+      "learning_rate": 2.3181743091447754e-05,
+      "loss": 0.3351,
+      "step": 623
+    },
+    {
+      "epoch": 2.526315789473684,
+      "grad_norm": 0.348229678878419,
+      "learning_rate": 2.3125896367261164e-05,
+      "loss": 0.3619,
+      "step": 624
+    },
+    {
+      "epoch": 2.5303643724696356,
+      "grad_norm": 0.33975773690610234,
+      "learning_rate": 2.307002464849052e-05,
+      "loss": 0.3499,
+      "step": 625
+    },
+    {
+      "epoch": 2.534412955465587,
+      "grad_norm": 0.3445903725825857,
+      "learning_rate": 2.30141283818846e-05,
+      "loss": 0.353,
+      "step": 626
+    },
+    {
+      "epoch": 2.5384615384615383,
+      "grad_norm": 0.32834198744379195,
+      "learning_rate": 2.2958208014388462e-05,
+      "loss": 0.3504,
+      "step": 627
+    },
+    {
+      "epoch": 2.54251012145749,
+      "grad_norm": 0.34548427264667936,
+      "learning_rate": 2.2902263993139857e-05,
+      "loss": 0.3488,
+      "step": 628
+    },
+    {
+      "epoch": 2.5465587044534415,
+      "grad_norm": 0.3765161490467458,
+      "learning_rate": 2.2846296765465708e-05,
+      "loss": 0.3575,
+      "step": 629
+    },
+    {
+      "epoch": 2.5506072874493926,
+      "grad_norm": 0.324342557810993,
+      "learning_rate": 2.2790306778878465e-05,
+      "loss": 0.3538,
+      "step": 630
+    },
+    {
+      "epoch": 2.554655870445344,
+      "grad_norm": 0.32774526472201865,
+      "learning_rate": 2.273429448107257e-05,
+      "loss": 0.3403,
+      "step": 631
+    },
+    {
+      "epoch": 2.5587044534412957,
+      "grad_norm": 0.3448554047303887,
+      "learning_rate": 2.2678260319920868e-05,
+      "loss": 0.3289,
+      "step": 632
+    },
+    {
+      "epoch": 2.562753036437247,
+      "grad_norm": 0.33880028759027897,
+      "learning_rate": 2.262220474347102e-05,
+      "loss": 0.3605,
+      "step": 633
+    },
+    {
+      "epoch": 2.5668016194331984,
+      "grad_norm": 0.3680577804348161,
+      "learning_rate": 2.2566128199941918e-05,
+      "loss": 0.36,
+      "step": 634
+    },
+    {
+      "epoch": 2.57085020242915,
+      "grad_norm": 0.34080925394181816,
+      "learning_rate": 2.2510031137720104e-05,
+      "loss": 0.356,
+      "step": 635
+    },
+    {
+      "epoch": 2.574898785425101,
+      "grad_norm": 0.3385339194514327,
+      "learning_rate": 2.2453914005356206e-05,
+      "loss": 0.3402,
+      "step": 636
+    },
+    {
+      "epoch": 2.5789473684210527,
+      "grad_norm": 0.3707580631241485,
+      "learning_rate": 2.2397777251561302e-05,
+      "loss": 0.3554,
+      "step": 637
+    },
+    {
+      "epoch": 2.582995951417004,
+      "grad_norm": 0.3612421424566549,
+      "learning_rate": 2.2341621325203393e-05,
+      "loss": 0.3478,
+      "step": 638
+    },
+    {
+      "epoch": 2.5870445344129553,
+      "grad_norm": 0.3710477876766041,
+      "learning_rate": 2.2285446675303764e-05,
+      "loss": 0.3332,
+      "step": 639
+    },
+    {
+      "epoch": 2.591093117408907,
+      "grad_norm": 0.3830800049319667,
+      "learning_rate": 2.2229253751033423e-05,
+      "loss": 0.3527,
+      "step": 640
+    },
+    {
+      "epoch": 2.5951417004048585,
+      "grad_norm": 0.34180260873606927,
+      "learning_rate": 2.2173043001709492e-05,
+      "loss": 0.3404,
+      "step": 641
+    },
+    {
+      "epoch": 2.5991902834008096,
+      "grad_norm": 0.4270486927072495,
+      "learning_rate": 2.2116814876791627e-05,
+      "loss": 0.3377,
+      "step": 642
+    },
+    {
+      "epoch": 2.603238866396761,
+      "grad_norm": 0.38884893360842726,
+      "learning_rate": 2.2060569825878413e-05,
+      "loss": 0.3658,
+      "step": 643
+    },
+    {
+      "epoch": 2.6072874493927127,
+      "grad_norm": 0.35655248677573165,
+      "learning_rate": 2.2004308298703777e-05,
+      "loss": 0.3478,
+      "step": 644
+    },
+    {
+      "epoch": 2.611336032388664,
+      "grad_norm": 0.4232371968814269,
+      "learning_rate": 2.1948030745133396e-05,
+      "loss": 0.3463,
+      "step": 645
+    },
+    {
+      "epoch": 2.6153846153846154,
+      "grad_norm": 0.42742417681899103,
+      "learning_rate": 2.1891737615161083e-05,
+      "loss": 0.3661,
+      "step": 646
+    },
+    {
+      "epoch": 2.619433198380567,
+      "grad_norm": 0.36046853797961165,
+      "learning_rate": 2.1835429358905218e-05,
+      "loss": 0.3402,
+      "step": 647
+    },
+    {
+      "epoch": 2.623481781376518,
+      "grad_norm": 0.38890742400037814,
+      "learning_rate": 2.17791064266051e-05,
+      "loss": 0.3428,
+      "step": 648
+    },
+    {
+      "epoch": 2.6275303643724697,
+      "grad_norm": 0.3301807457107487,
+      "learning_rate": 2.172276926861741e-05,
+      "loss": 0.3479,
+      "step": 649
+    },
+    {
+      "epoch": 2.6315789473684212,
+      "grad_norm": 0.36025700371370184,
+      "learning_rate": 2.166641833541255e-05,
+      "loss": 0.3343,
+      "step": 650
+    },
+    {
+      "epoch": 2.6356275303643724,
+      "grad_norm": 0.3506237990292698,
+      "learning_rate": 2.161005407757109e-05,
+      "loss": 0.3551,
+      "step": 651
+    },
+    {
+      "epoch": 2.639676113360324,
+      "grad_norm": 0.34085381510158885,
+      "learning_rate": 2.155367694578013e-05,
+      "loss": 0.3487,
+      "step": 652
+    },
+    {
+      "epoch": 2.6437246963562755,
+      "grad_norm": 0.3347873222063248,
+      "learning_rate": 2.1497287390829712e-05,
+      "loss": 0.3588,
+      "step": 653
+    },
+    {
+      "epoch": 2.6477732793522266,
+      "grad_norm": 0.36676990744553933,
+      "learning_rate": 2.1440885863609208e-05,
+      "loss": 0.342,
+      "step": 654
+    },
+    {
+      "epoch": 2.651821862348178,
+      "grad_norm": 0.3581881253438231,
+      "learning_rate": 2.1384472815103737e-05,
+      "loss": 0.3561,
+      "step": 655
+    },
+    {
+      "epoch": 2.6558704453441297,
+      "grad_norm": 0.35398240347027193,
+      "learning_rate": 2.1328048696390534e-05,
+      "loss": 0.3438,
+      "step": 656
+    },
+    {
+      "epoch": 2.659919028340081,
+      "grad_norm": 0.35150064902823147,
+      "learning_rate": 2.1271613958635342e-05,
+      "loss": 0.3512,
+      "step": 657
+    },
+    {
+      "epoch": 2.6639676113360324,
+      "grad_norm": 0.3586639998651549,
+      "learning_rate": 2.1215169053088825e-05,
+      "loss": 0.3389,
+      "step": 658
+    },
+    {
+      "epoch": 2.668016194331984,
+      "grad_norm": 0.36512153683349013,
+      "learning_rate": 2.1158714431082937e-05,
+      "loss": 0.3447,
+      "step": 659
+    },
+    {
+      "epoch": 2.672064777327935,
+      "grad_norm": 0.3174003915750069,
+      "learning_rate": 2.110225054402733e-05,
+      "loss": 0.3465,
+      "step": 660
+    },
+    {
+      "epoch": 2.6761133603238867,
+      "grad_norm": 0.38620023976403883,
+      "learning_rate": 2.1045777843405747e-05,
+      "loss": 0.3473,
+      "step": 661
+    },
+    {
+      "epoch": 2.6801619433198383,
+      "grad_norm": 0.33300516084582443,
+      "learning_rate": 2.09892967807724e-05,
+      "loss": 0.3477,
+      "step": 662
+    },
+    {
+      "epoch": 2.6842105263157894,
+      "grad_norm": 0.386136945817044,
+      "learning_rate": 2.093280780774835e-05,
+      "loss": 0.3584,
+      "step": 663
+    },
+    {
+      "epoch": 2.688259109311741,
+      "grad_norm": 0.36692852022554034,
+      "learning_rate": 2.087631137601793e-05,
+      "loss": 0.3507,
+      "step": 664
+    },
+    {
+      "epoch": 2.6923076923076925,
+      "grad_norm": 0.33409429205746743,
+      "learning_rate": 2.0819807937325104e-05,
+      "loss": 0.3434,
+      "step": 665
+    },
+    {
+      "epoch": 2.6963562753036436,
+      "grad_norm": 0.46003704750873914,
+      "learning_rate": 2.0763297943469847e-05,
+      "loss": 0.3494,
+      "step": 666
+    },
+    {
+      "epoch": 2.700404858299595,
+      "grad_norm": 0.33933931084900926,
+      "learning_rate": 2.070678184630458e-05,
+      "loss": 0.3517,
+      "step": 667
+    },
+    {
+      "epoch": 2.7044534412955468,
+      "grad_norm": 0.4117323284271442,
+      "learning_rate": 2.06502600977305e-05,
+      "loss": 0.3279,
+      "step": 668
+    },
+    {
+      "epoch": 2.708502024291498,
+      "grad_norm": 0.3433373249217794,
+      "learning_rate": 2.0593733149694008e-05,
+      "loss": 0.3561,
+      "step": 669
+    },
+    {
+      "epoch": 2.7125506072874495,
+      "grad_norm": 0.36451177640402527,
+      "learning_rate": 2.0537201454183074e-05,
+      "loss": 0.3177,
+      "step": 670
+    },
+    {
+      "epoch": 2.716599190283401,
+      "grad_norm": 0.37558123734035914,
+      "learning_rate": 2.0480665463223626e-05,
+      "loss": 0.36,
+      "step": 671
+    },
+    {
+      "epoch": 2.720647773279352,
+      "grad_norm": 0.3278921450693575,
+      "learning_rate": 2.042412562887594e-05,
+      "loss": 0.3397,
+      "step": 672
+    },
+    {
+      "epoch": 2.7246963562753037,
+      "grad_norm": 0.37086516233620276,
+      "learning_rate": 2.0367582403231033e-05,
+      "loss": 0.3554,
+      "step": 673
+    },
+    {
+      "epoch": 2.7287449392712553,
+      "grad_norm": 0.35029259046055045,
+      "learning_rate": 2.0311036238407023e-05,
+      "loss": 0.3538,
+      "step": 674
+    },
+    {
+      "epoch": 2.7327935222672064,
+      "grad_norm": 0.36378184199178976,
+      "learning_rate": 2.0254487586545534e-05,
+      "loss": 0.3448,
+      "step": 675
+    },
+    {
+      "epoch": 2.736842105263158,
+      "grad_norm": 0.34837020062514773,
+      "learning_rate": 2.0197936899808088e-05,
+      "loss": 0.3521,
+      "step": 676
+    },
+    {
+      "epoch": 2.7408906882591095,
+      "grad_norm": 0.35854137155263516,
+      "learning_rate": 2.0141384630372466e-05,
+      "loss": 0.3479,
+      "step": 677
+    },
+    {
+      "epoch": 2.7449392712550607,
+      "grad_norm": 0.3524813287017685,
+      "learning_rate": 2.0084831230429106e-05,
+      "loss": 0.351,
+      "step": 678
+    },
+    {
+      "epoch": 2.748987854251012,
+      "grad_norm": 0.3642358286744724,
+      "learning_rate": 2.0028277152177492e-05,
+      "loss": 0.3437,
+      "step": 679
+    },
+    {
+      "epoch": 2.753036437246964,
+      "grad_norm": 0.3402072970595609,
+      "learning_rate": 1.9971722847822518e-05,
+      "loss": 0.3504,
+      "step": 680
+    },
+    {
+      "epoch": 2.757085020242915,
+      "grad_norm": 0.33894130086552077,
+      "learning_rate": 1.99151687695709e-05,
+      "loss": 0.3474,
+      "step": 681
+    },
+    {
+      "epoch": 2.7611336032388665,
+      "grad_norm": 0.38056972633211433,
+      "learning_rate": 1.9858615369627537e-05,
+      "loss": 0.3569,
+      "step": 682
+    },
+    {
+      "epoch": 2.765182186234818,
+      "grad_norm": 0.3188592828111863,
+      "learning_rate": 1.9802063100191916e-05,
+      "loss": 0.3478,
+      "step": 683
+    },
+    {
+      "epoch": 2.769230769230769,
+      "grad_norm": 0.3704916967161989,
+      "learning_rate": 1.974551241345447e-05,
+      "loss": 0.3375,
+      "step": 684
+    },
+    {
+      "epoch": 2.7732793522267207,
+      "grad_norm": 0.3475222243818235,
+      "learning_rate": 1.9688963761592984e-05,
+      "loss": 0.3317,
+      "step": 685
+    },
+    {
+      "epoch": 2.7773279352226723,
+      "grad_norm": 0.2977079095166456,
+      "learning_rate": 1.963241759676897e-05,
+      "loss": 0.364,
+      "step": 686
+    },
+    {
+      "epoch": 2.7813765182186234,
+      "grad_norm": 0.35090380959774037,
+      "learning_rate": 1.9575874371124066e-05,
+      "loss": 0.3459,
+      "step": 687
+    },
+    {
+      "epoch": 2.785425101214575,
+      "grad_norm": 0.322254346599763,
+      "learning_rate": 1.951933453677638e-05,
+      "loss": 0.3563,
+      "step": 688
+    },
+    {
+      "epoch": 2.7894736842105265,
+      "grad_norm": 0.3235149909737108,
+      "learning_rate": 1.9462798545816932e-05,
+      "loss": 0.3282,
+      "step": 689
+    },
+    {
+      "epoch": 2.7935222672064777,
+      "grad_norm": 0.352249707219789,
+      "learning_rate": 1.9406266850305996e-05,
+      "loss": 0.3408,
+      "step": 690
+    },
+    {
+      "epoch": 2.7975708502024292,
+      "grad_norm": 0.3208296316951392,
+      "learning_rate": 1.9349739902269503e-05,
+      "loss": 0.3237,
+      "step": 691
+    },
+    {
+      "epoch": 2.801619433198381,
+      "grad_norm": 0.34777547098687356,
+      "learning_rate": 1.9293218153695425e-05,
+      "loss": 0.3426,
+      "step": 692
+    },
+    {
+      "epoch": 2.805668016194332,
+      "grad_norm": 0.3511744144612769,
+      "learning_rate": 1.9236702056530153e-05,
+      "loss": 0.3426,
+      "step": 693
+    },
+    {
+      "epoch": 2.8097165991902835,
+      "grad_norm": 0.3507035152504313,
+      "learning_rate": 1.9180192062674907e-05,
+      "loss": 0.331,
+      "step": 694
+    },
+    {
+      "epoch": 2.813765182186235,
+      "grad_norm": 0.3580709902006066,
+      "learning_rate": 1.9123688623982076e-05,
+      "loss": 0.3433,
+      "step": 695
+    },
+    {
+      "epoch": 2.817813765182186,
+      "grad_norm": 0.3095425324928812,
+      "learning_rate": 1.906719219225165e-05,
+      "loss": 0.3335,
+      "step": 696
+    },
+    {
+      "epoch": 2.8218623481781377,
+      "grad_norm": 0.38578262047581335,
+      "learning_rate": 1.9010703219227603e-05,
+      "loss": 0.3318,
+      "step": 697
+    },
+    {
+      "epoch": 2.8259109311740893,
+      "grad_norm": 0.32582506876189876,
+      "learning_rate": 1.8954222156594256e-05,
+      "loss": 0.3391,
+      "step": 698
+    },
+    {
+      "epoch": 2.8299595141700404,
+      "grad_norm": 0.35371666856839956,
+      "learning_rate": 1.8897749455972673e-05,
+      "loss": 0.3537,
+      "step": 699
+    },
+    {
+      "epoch": 2.834008097165992,
+      "grad_norm": 0.3596895875484218,
+      "learning_rate": 1.884128556891707e-05,
+      "loss": 0.3509,
+      "step": 700
+    },
+    {
+      "epoch": 2.8380566801619436,
+      "grad_norm": 0.348741905848361,
+      "learning_rate": 1.878483094691119e-05,
+      "loss": 0.3412,
+      "step": 701
+    },
+    {
+      "epoch": 2.8421052631578947,
+      "grad_norm": 0.3260550959644138,
+      "learning_rate": 1.8728386041364664e-05,
+      "loss": 0.3573,
+      "step": 702
+    },
+    {
+      "epoch": 2.8461538461538463,
+      "grad_norm": 0.3501807896576499,
+      "learning_rate": 1.867195130360947e-05,
+      "loss": 0.3474,
+      "step": 703
+    },
+    {
+      "epoch": 2.850202429149798,
+      "grad_norm": 0.32222577207197745,
+      "learning_rate": 1.8615527184896263e-05,
+      "loss": 0.3498,
+      "step": 704
+    },
+    {
+      "epoch": 2.854251012145749,
+      "grad_norm": 0.33264826005499964,
+      "learning_rate": 1.8559114136390795e-05,
+      "loss": 0.3479,
+      "step": 705
+    },
+    {
+      "epoch": 2.8582995951417005,
+      "grad_norm": 0.32083024431996215,
+      "learning_rate": 1.8502712609170298e-05,
+      "loss": 0.3428,
+      "step": 706
+    },
+    {
+      "epoch": 2.862348178137652,
+      "grad_norm": 0.32563799916560077,
+      "learning_rate": 1.8446323054219876e-05,
+      "loss": 0.3479,
+      "step": 707
+    },
+    {
+      "epoch": 2.866396761133603,
+      "grad_norm": 0.34497786459635726,
+      "learning_rate": 1.838994592242891e-05,
+      "loss": 0.3475,
+      "step": 708
+    },
+    {
+      "epoch": 2.8704453441295548,
+      "grad_norm": 0.3375667175695045,
+      "learning_rate": 1.8333581664587453e-05,
+      "loss": 0.3464,
+      "step": 709
+    },
+    {
+      "epoch": 2.8744939271255063,
+      "grad_norm": 0.3315463897448291,
+      "learning_rate": 1.8277230731382593e-05,
+      "loss": 0.3448,
+      "step": 710
+    },
+    {
+      "epoch": 2.8785425101214575,
+      "grad_norm": 0.3051796795034602,
+      "learning_rate": 1.82208935733949e-05,
+      "loss": 0.355,
+      "step": 711
+    },
+    {
+      "epoch": 2.882591093117409,
+      "grad_norm": 0.38379333020488976,
+      "learning_rate": 1.8164570641094793e-05,
+      "loss": 0.3386,
+      "step": 712
+    },
+    {
+      "epoch": 2.8866396761133606,
+      "grad_norm": 0.3080787185429063,
+      "learning_rate": 1.810826238483892e-05,
+      "loss": 0.3447,
+      "step": 713
+    },
+    {
+      "epoch": 2.8906882591093117,
+      "grad_norm": 0.3610147509211246,
+      "learning_rate": 1.8051969254866608e-05,
+      "loss": 0.3367,
+      "step": 714
+    },
+    {
+      "epoch": 2.8947368421052633,
+      "grad_norm": 0.3094924096539449,
+      "learning_rate": 1.7995691701296226e-05,
+      "loss": 0.3408,
+      "step": 715
+    },
+    {
+      "epoch": 2.898785425101215,
+      "grad_norm": 0.3207352054109797,
+      "learning_rate": 1.7939430174121593e-05,
+      "loss": 0.3664,
+      "step": 716
+    },
+    {
+      "epoch": 2.902834008097166,
+      "grad_norm": 0.33153276785994495,
+      "learning_rate": 1.788318512320838e-05,
+      "loss": 0.3384,
+      "step": 717
+    },
+    {
+      "epoch": 2.9068825910931175,
+      "grad_norm": 0.33169738190052245,
+      "learning_rate": 1.782695699829051e-05,
+      "loss": 0.3397,
+      "step": 718
+    },
+    {
+      "epoch": 2.910931174089069,
+      "grad_norm": 0.34949496811576575,
+      "learning_rate": 1.7770746248966587e-05,
+      "loss": 0.3454,
+      "step": 719
+    },
+    {
+      "epoch": 2.91497975708502,
+      "grad_norm": 0.3343979811519819,
+      "learning_rate": 1.7714553324696243e-05,
+      "loss": 0.3436,
+      "step": 720
+    },
+    {
+      "epoch": 2.919028340080972,
+      "grad_norm": 0.37257663589119666,
+      "learning_rate": 1.7658378674796614e-05,
+      "loss": 0.3677,
+      "step": 721
+    },
+    {
+      "epoch": 2.9230769230769234,
+      "grad_norm": 0.3067289619487916,
+      "learning_rate": 1.76022227484387e-05,
+      "loss": 0.3407,
+      "step": 722
+    },
+    {
+      "epoch": 2.9271255060728745,
+      "grad_norm": 0.36177756996636706,
+      "learning_rate": 1.7546085994643807e-05,
+      "loss": 0.3393,
+      "step": 723
+    },
+    {
+      "epoch": 2.931174089068826,
+      "grad_norm": 0.32883052044643424,
+      "learning_rate": 1.7489968862279902e-05,
+      "loss": 0.344,
+      "step": 724
+    },
+    {
+      "epoch": 2.9352226720647776,
+      "grad_norm": 0.312129388143621,
+      "learning_rate": 1.743387180005809e-05,
+      "loss": 0.3274,
+      "step": 725
+    },
+    {
+      "epoch": 2.9392712550607287,
+      "grad_norm": 0.3423011357734857,
+      "learning_rate": 1.737779525652899e-05,
+      "loss": 0.3463,
+      "step": 726
+    },
+    {
+      "epoch": 2.9433198380566803,
+      "grad_norm": 0.30327954478788494,
+      "learning_rate": 1.732173968007914e-05,
+      "loss": 0.3513,
+      "step": 727
+    },
+    {
+      "epoch": 2.9473684210526314,
+      "grad_norm": 0.3155028078078164,
+      "learning_rate": 1.7265705518927437e-05,
+      "loss": 0.3414,
+      "step": 728
+    },
+    {
+      "epoch": 2.951417004048583,
+      "grad_norm": 0.3387517022825865,
+      "learning_rate": 1.7209693221121542e-05,
+      "loss": 0.348,
+      "step": 729
+    },
+    {
+      "epoch": 2.9554655870445345,
+      "grad_norm": 0.3181000355094372,
+      "learning_rate": 1.7153703234534302e-05,
+      "loss": 0.3667,
+      "step": 730
+    },
+    {
+      "epoch": 2.9595141700404857,
+      "grad_norm": 0.3193290865090587,
+      "learning_rate": 1.709773600686015e-05,
+      "loss": 0.3479,
+      "step": 731
+    },
+    {
+      "epoch": 2.9635627530364372,
+      "grad_norm": 0.3371404047200397,
+      "learning_rate": 1.7041791985611548e-05,
+      "loss": 0.3395,
+      "step": 732
+    },
+    {
+      "epoch": 2.967611336032389,
+      "grad_norm": 0.34144916257825725,
+      "learning_rate": 1.69858716181154e-05,
+      "loss": 0.3585,
+      "step": 733
+    },
+    {
+      "epoch": 2.97165991902834,
+      "grad_norm": 0.29311956394406596,
+      "learning_rate": 1.692997535150948e-05,
+      "loss": 0.3423,
+      "step": 734
+    },
+    {
+      "epoch": 2.9757085020242915,
+      "grad_norm": 0.3513962008784661,
+      "learning_rate": 1.687410363273884e-05,
+      "loss": 0.3377,
+      "step": 735
+    },
+    {
+      "epoch": 2.979757085020243,
+      "grad_norm": 0.3194519180534689,
+      "learning_rate": 1.6818256908552257e-05,
+      "loss": 0.3385,
+      "step": 736
+    },
+    {
+      "epoch": 2.983805668016194,
+      "grad_norm": 0.3290373043912981,
+      "learning_rate": 1.6762435625498646e-05,
+      "loss": 0.3451,
+      "step": 737
+    },
+    {
+      "epoch": 2.9878542510121457,
+      "grad_norm": 0.34834538538245213,
+      "learning_rate": 1.67066402299235e-05,
+      "loss": 0.3552,
+      "step": 738
+    },
+    {
+      "epoch": 2.9919028340080973,
+      "grad_norm": 0.32483116807805246,
+      "learning_rate": 1.6650871167965313e-05,
+      "loss": 0.3563,
+      "step": 739
+    },
+    {
+      "epoch": 2.9959514170040484,
+      "grad_norm": 0.3305416374021896,
+      "learning_rate": 1.6595128885552028e-05,
+      "loss": 0.3414,
+      "step": 740
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 0.3358045009977234,
+      "learning_rate": 1.653941382839745e-05,
+      "loss": 0.3355,
+      "step": 741
+    },
+    {
+      "epoch": 3.0040485829959516,
+      "grad_norm": 0.5458372130725518,
+      "learning_rate": 1.6483726441997673e-05,
+      "loss": 0.2545,
+      "step": 742
+    },
+    {
+      "epoch": 3.0080971659919027,
+      "grad_norm": 0.33452067290519627,
+      "learning_rate": 1.642806717162757e-05,
+      "loss": 0.2648,
+      "step": 743
+    },
+    {
+      "epoch": 3.0121457489878543,
+      "grad_norm": 0.7122445827758006,
+      "learning_rate": 1.637243646233718e-05,
+      "loss": 0.2475,
+      "step": 744
+    },
+    {
+      "epoch": 3.016194331983806,
+      "grad_norm": 0.5274123081184598,
+      "learning_rate": 1.6316834758948174e-05,
+      "loss": 0.2453,
+      "step": 745
+    },
+    {
+      "epoch": 3.020242914979757,
+      "grad_norm": 0.44088626371668166,
+      "learning_rate": 1.6261262506050282e-05,
+      "loss": 0.2585,
+      "step": 746
+    },
+    {
+      "epoch": 3.0242914979757085,
+      "grad_norm": 0.5002804264231067,
+      "learning_rate": 1.620572014799777e-05,
+      "loss": 0.2625,
+      "step": 747
+    },
+    {
+      "epoch": 3.02834008097166,
+      "grad_norm": 0.42714491108712827,
+      "learning_rate": 1.6150208128905857e-05,
+      "loss": 0.2561,
+      "step": 748
+    },
+    {
+      "epoch": 3.032388663967611,
+      "grad_norm": 0.39500050099469114,
+      "learning_rate": 1.6094726892647147e-05,
+      "loss": 0.2649,
+      "step": 749
+    },
+    {
+      "epoch": 3.0364372469635628,
+      "grad_norm": 0.40470319920169184,
+      "learning_rate": 1.6039276882848135e-05,
+      "loss": 0.2592,
+      "step": 750
+    },
+    {
+      "epoch": 3.0404858299595143,
+      "grad_norm": 0.4137884026409026,
+      "learning_rate": 1.5983858542885635e-05,
+      "loss": 0.2383,
+      "step": 751
+    },
+    {
+      "epoch": 3.0445344129554655,
+      "grad_norm": 0.368270044465049,
+      "learning_rate": 1.5928472315883203e-05,
+      "loss": 0.2429,
+      "step": 752
+    },
+    {
+      "epoch": 3.048582995951417,
+      "grad_norm": 0.3325167515921634,
+      "learning_rate": 1.5873118644707633e-05,
+      "loss": 0.2596,
+      "step": 753
+    },
+    {
+      "epoch": 3.0526315789473686,
+      "grad_norm": 0.401250044905098,
+      "learning_rate": 1.5817797971965413e-05,
+      "loss": 0.2495,
+      "step": 754
+    },
+    {
+      "epoch": 3.0566801619433197,
+      "grad_norm": 0.3733813699178917,
+      "learning_rate": 1.576251073999917e-05,
+      "loss": 0.2562,
+      "step": 755
+    },
+    {
+      "epoch": 3.0607287449392713,
+      "grad_norm": 0.3499149897875523,
+      "learning_rate": 1.5707257390884126e-05,
+      "loss": 0.2532,
+      "step": 756
+    },
+    {
+      "epoch": 3.064777327935223,
+      "grad_norm": 0.3506776916607793,
+      "learning_rate": 1.5652038366424595e-05,
+      "loss": 0.2548,
+      "step": 757
+    },
+    {
+      "epoch": 3.068825910931174,
+      "grad_norm": 0.4093940690293295,
+      "learning_rate": 1.5596854108150424e-05,
+      "loss": 0.265,
+      "step": 758
+    },
+    {
+      "epoch": 3.0728744939271255,
+      "grad_norm": 0.36454955708391457,
+      "learning_rate": 1.5541705057313476e-05,
+      "loss": 0.2492,
+      "step": 759
+    },
+    {
+      "epoch": 3.076923076923077,
+      "grad_norm": 0.34272985795705485,
+      "learning_rate": 1.5486591654884086e-05,
+      "loss": 0.2536,
+      "step": 760
+    },
+    {
+      "epoch": 3.080971659919028,
+      "grad_norm": 0.36250380093547274,
+      "learning_rate": 1.543151434154755e-05,
+      "loss": 0.2337,
+      "step": 761
+    },
+    {
+      "epoch": 3.08502024291498,
+      "grad_norm": 0.36748384461646566,
+      "learning_rate": 1.5376473557700615e-05,
+      "loss": 0.2597,
+      "step": 762
+    },
+    {
+      "epoch": 3.0890688259109313,
+      "grad_norm": 0.30561431771959596,
+      "learning_rate": 1.5321469743447893e-05,
+      "loss": 0.2597,
+      "step": 763
+    },
+    {
+      "epoch": 3.0931174089068825,
+      "grad_norm": 0.34552528550058204,
+      "learning_rate": 1.5266503338598434e-05,
+      "loss": 0.2484,
+      "step": 764
+    },
+    {
+      "epoch": 3.097165991902834,
+      "grad_norm": 0.33053385128677826,
+      "learning_rate": 1.5211574782662149e-05,
+      "loss": 0.2655,
+      "step": 765
+    },
+    {
+      "epoch": 3.1012145748987856,
+      "grad_norm": 0.32072886415106017,
+      "learning_rate": 1.5156684514846307e-05,
+      "loss": 0.2623,
+      "step": 766
+    },
+    {
+      "epoch": 3.1052631578947367,
+      "grad_norm": 0.3199276590083028,
+      "learning_rate": 1.5101832974052012e-05,
+      "loss": 0.2617,
+      "step": 767
+    },
+    {
+      "epoch": 3.1093117408906883,
+      "grad_norm": 0.36949224124354973,
+      "learning_rate": 1.5047020598870737e-05,
+      "loss": 0.2596,
+      "step": 768
+    },
+    {
+      "epoch": 3.11336032388664,
+      "grad_norm": 0.3228380310454622,
+      "learning_rate": 1.4992247827580778e-05,
+      "loss": 0.2529,
+      "step": 769
+    },
+    {
+      "epoch": 3.117408906882591,
+      "grad_norm": 0.33139525986164087,
+      "learning_rate": 1.493751509814374e-05,
+      "loss": 0.2559,
+      "step": 770
+    },
+    {
+      "epoch": 3.1214574898785425,
+      "grad_norm": 0.33121289129799886,
+      "learning_rate": 1.4882822848201073e-05,
+      "loss": 0.2381,
+      "step": 771
+    },
+    {
+      "epoch": 3.125506072874494,
+      "grad_norm": 0.3329300084918852,
+      "learning_rate": 1.4828171515070553e-05,
+      "loss": 0.2528,
+      "step": 772
+    },
+    {
+      "epoch": 3.1295546558704452,
+      "grad_norm": 0.3309569266814212,
+      "learning_rate": 1.4773561535742793e-05,
+      "loss": 0.2515,
+      "step": 773
+    },
+    {
+      "epoch": 3.133603238866397,
+      "grad_norm": 0.32465174609651853,
+      "learning_rate": 1.4718993346877715e-05,
+      "loss": 0.2596,
+      "step": 774
+    },
+    {
+      "epoch": 3.1376518218623484,
+      "grad_norm": 0.31912663151346277,
+      "learning_rate": 1.466446738480111e-05,
+      "loss": 0.2536,
+      "step": 775
+    },
+    {
+      "epoch": 3.1417004048582995,
+      "grad_norm": 0.3449328267868318,
+      "learning_rate": 1.460998408550112e-05,
+      "loss": 0.271,
+      "step": 776
+    },
+    {
+      "epoch": 3.145748987854251,
+      "grad_norm": 0.3104443841647944,
+      "learning_rate": 1.4555543884624751e-05,
+      "loss": 0.257,
+      "step": 777
+    },
+    {
+      "epoch": 3.1497975708502026,
+      "grad_norm": 0.3361816538839216,
+      "learning_rate": 1.4501147217474402e-05,
+      "loss": 0.2653,
+      "step": 778
+    },
+    {
+      "epoch": 3.1538461538461537,
+      "grad_norm": 0.3214399995611792,
+      "learning_rate": 1.444679451900437e-05,
+      "loss": 0.2667,
+      "step": 779
+    },
+    {
+      "epoch": 3.1578947368421053,
+      "grad_norm": 0.32805584097794643,
+      "learning_rate": 1.4392486223817397e-05,
+      "loss": 0.2688,
+      "step": 780
+    },
+    {
+      "epoch": 3.161943319838057,
+      "grad_norm": 0.32059817963213566,
+      "learning_rate": 1.4338222766161143e-05,
+      "loss": 0.2675,
+      "step": 781
+    },
+    {
+      "epoch": 3.165991902834008,
+      "grad_norm": 0.31770176019481977,
+      "learning_rate": 1.4284004579924777e-05,
+      "loss": 0.2532,
+      "step": 782
+    },
+    {
+      "epoch": 3.1700404858299596,
+      "grad_norm": 0.34034866137653685,
+      "learning_rate": 1.4229832098635479e-05,
+      "loss": 0.2544,
+      "step": 783
+    },
+    {
+      "epoch": 3.174089068825911,
+      "grad_norm": 0.31158177087287425,
+      "learning_rate": 1.4175705755454963e-05,
+      "loss": 0.2591,
+      "step": 784
+    },
+    {
+      "epoch": 3.1781376518218623,
+      "grad_norm": 0.3497188432294514,
+      "learning_rate": 1.412162598317602e-05,
+      "loss": 0.2567,
+      "step": 785
+    },
+    {
+      "epoch": 3.182186234817814,
+      "grad_norm": 0.3062154828947262,
+      "learning_rate": 1.406759321421907e-05,
+      "loss": 0.2618,
+      "step": 786
+    },
+    {
+      "epoch": 3.1862348178137654,
+      "grad_norm": 0.31315895854750625,
+      "learning_rate": 1.401360788062871e-05,
+      "loss": 0.2563,
+      "step": 787
+    },
+    {
+      "epoch": 3.1902834008097165,
+      "grad_norm": 0.2854172614945123,
+      "learning_rate": 1.3959670414070204e-05,
+      "loss": 0.2559,
+      "step": 788
+    },
+    {
+      "epoch": 3.194331983805668,
+      "grad_norm": 0.29117581781342117,
+      "learning_rate": 1.3905781245826108e-05,
+      "loss": 0.2409,
+      "step": 789
+    },
+    {
+      "epoch": 3.1983805668016196,
+      "grad_norm": 0.29177565540686506,
+      "learning_rate": 1.3851940806792778e-05,
+      "loss": 0.2295,
+      "step": 790
+    },
+    {
+      "epoch": 3.2024291497975708,
+      "grad_norm": 0.2932474868888365,
+      "learning_rate": 1.379814952747693e-05,
+      "loss": 0.2517,
+      "step": 791
+    },
+    {
+      "epoch": 3.2064777327935223,
+      "grad_norm": 0.3029207539614302,
+      "learning_rate": 1.3744407837992193e-05,
+      "loss": 0.2583,
+      "step": 792
+    },
+    {
+      "epoch": 3.2105263157894735,
+      "grad_norm": 0.2969835690950678,
+      "learning_rate": 1.3690716168055692e-05,
+      "loss": 0.2604,
+      "step": 793
+    },
+    {
+      "epoch": 3.214574898785425,
+      "grad_norm": 0.31265569204409555,
+      "learning_rate": 1.363707494698459e-05,
+      "loss": 0.2477,
+      "step": 794
+    },
+    {
+      "epoch": 3.2186234817813766,
+      "grad_norm": 0.29980465268650647,
+      "learning_rate": 1.358348460369265e-05,
+      "loss": 0.2578,
+      "step": 795
+    },
+    {
+      "epoch": 3.2226720647773277,
+      "grad_norm": 0.3066746890851102,
+      "learning_rate": 1.352994556668684e-05,
+      "loss": 0.2475,
+      "step": 796
+    },
+    {
+      "epoch": 3.2267206477732793,
+      "grad_norm": 0.2984162867518012,
+      "learning_rate": 1.3476458264063865e-05,
+      "loss": 0.2479,
+      "step": 797
+    },
+    {
+      "epoch": 3.230769230769231,
+      "grad_norm": 0.2855305565948594,
+      "learning_rate": 1.342302312350678e-05,
+      "loss": 0.2673,
+      "step": 798
+    },
+    {
+      "epoch": 3.234817813765182,
+      "grad_norm": 0.30054756319079146,
+      "learning_rate": 1.3369640572281537e-05,
+      "loss": 0.2549,
+      "step": 799
+    },
+    {
+      "epoch": 3.2388663967611335,
+      "grad_norm": 0.28214936732084056,
+      "learning_rate": 1.3316311037233596e-05,
+      "loss": 0.2688,
+      "step": 800
+    },
+    {
+      "epoch": 3.242914979757085,
+      "grad_norm": 0.27559382901923046,
+      "learning_rate": 1.326303494478451e-05,
+      "loss": 0.2615,
+      "step": 801
+    },
+    {
+      "epoch": 3.246963562753036,
+      "grad_norm": 0.27384705600208903,
+      "learning_rate": 1.3209812720928472e-05,
+      "loss": 0.2736,
+      "step": 802
+    },
+    {
+      "epoch": 3.251012145748988,
+      "grad_norm": 0.30396868888016587,
+      "learning_rate": 1.315664479122898e-05,
+      "loss": 0.2486,
+      "step": 803
+    },
+    {
+      "epoch": 3.2550607287449393,
+      "grad_norm": 0.283239959186242,
+      "learning_rate": 1.3103531580815378e-05,
+      "loss": 0.2539,
+      "step": 804
+    },
+    {
+      "epoch": 3.2591093117408905,
+      "grad_norm": 0.2971086076468887,
+      "learning_rate": 1.3050473514379488e-05,
+      "loss": 0.2505,
+      "step": 805
+    },
+    {
+      "epoch": 3.263157894736842,
+      "grad_norm": 0.31442485321575053,
+      "learning_rate": 1.2997471016172188e-05,
+      "loss": 0.248,
+      "step": 806
+    },
+    {
+      "epoch": 3.2672064777327936,
+      "grad_norm": 0.28479504611652406,
+      "learning_rate": 1.2944524510000042e-05,
+      "loss": 0.2531,
+      "step": 807
+    },
+    {
+      "epoch": 3.2712550607287447,
+      "grad_norm": 0.2816725733962436,
+      "learning_rate": 1.2891634419221908e-05,
+      "loss": 0.2547,
+      "step": 808
+    },
+    {
+      "epoch": 3.2753036437246963,
+      "grad_norm": 0.27641133494540704,
+      "learning_rate": 1.2838801166745545e-05,
+      "loss": 0.2506,
+      "step": 809
+    },
+    {
+      "epoch": 3.279352226720648,
+      "grad_norm": 0.2994096223196522,
+      "learning_rate": 1.2786025175024216e-05,
+      "loss": 0.2454,
+      "step": 810
+    },
+    {
+      "epoch": 3.283400809716599,
+      "grad_norm": 0.29400967580417725,
+      "learning_rate": 1.2733306866053357e-05,
+      "loss": 0.2626,
+      "step": 811
+    },
+    {
+      "epoch": 3.2874493927125505,
+      "grad_norm": 0.2788423854325679,
+      "learning_rate": 1.2680646661367163e-05,
+      "loss": 0.2669,
+      "step": 812
+    },
+    {
+      "epoch": 3.291497975708502,
+      "grad_norm": 0.2939539879314061,
+      "learning_rate": 1.2628044982035219e-05,
+      "loss": 0.2482,
+      "step": 813
+    },
+    {
+      "epoch": 3.2955465587044532,
+      "grad_norm": 0.28745563828058707,
+      "learning_rate": 1.2575502248659155e-05,
+      "loss": 0.2546,
+      "step": 814
+    },
+    {
+      "epoch": 3.299595141700405,
+      "grad_norm": 0.2981321632272293,
+      "learning_rate": 1.2523018881369269e-05,
+      "loss": 0.2459,
+      "step": 815
+    },
+    {
+      "epoch": 3.3036437246963564,
+      "grad_norm": 0.3059839346370724,
+      "learning_rate": 1.2470595299821176e-05,
+      "loss": 0.263,
+      "step": 816
+    },
+    {
+      "epoch": 3.3076923076923075,
+      "grad_norm": 0.3046802554719862,
+      "learning_rate": 1.2418231923192416e-05,
+      "loss": 0.261,
+      "step": 817
+    },
+    {
+      "epoch": 3.311740890688259,
+      "grad_norm": 0.30506879964118055,
+      "learning_rate": 1.2365929170179165e-05,
+      "loss": 0.2711,
+      "step": 818
+    },
+    {
+      "epoch": 3.3157894736842106,
+      "grad_norm": 0.2898660147238817,
+      "learning_rate": 1.2313687458992848e-05,
+      "loss": 0.2515,
+      "step": 819
+    },
+    {
+      "epoch": 3.3198380566801617,
+      "grad_norm": 0.32644298198842014,
+      "learning_rate": 1.2261507207356785e-05,
+      "loss": 0.2532,
+      "step": 820
+    },
+    {
+      "epoch": 3.3238866396761133,
+      "grad_norm": 0.303912223224435,
+      "learning_rate": 1.220938883250288e-05,
+      "loss": 0.2611,
+      "step": 821
+    },
+    {
+      "epoch": 3.327935222672065,
+      "grad_norm": 0.2939969575986051,
+      "learning_rate": 1.215733275116827e-05,
+      "loss": 0.2483,
+      "step": 822
+    },
+    {
+      "epoch": 3.331983805668016,
+      "grad_norm": 0.28690668044805373,
+      "learning_rate": 1.2105339379592005e-05,
+      "loss": 0.2368,
+      "step": 823
+    },
+    {
+      "epoch": 3.3360323886639676,
+      "grad_norm": 0.29752764891923594,
+      "learning_rate": 1.2053409133511681e-05,
+      "loss": 0.2481,
+      "step": 824
+    },
+    {
+      "epoch": 3.340080971659919,
+      "grad_norm": 0.30380809091514405,
+      "learning_rate": 1.2001542428160173e-05,
+      "loss": 0.2449,
+      "step": 825
+    },
+    {
+      "epoch": 3.3441295546558703,
+      "grad_norm": 0.2892852854901141,
+      "learning_rate": 1.1949739678262278e-05,
+      "loss": 0.239,
+      "step": 826
+    },
+    {
+      "epoch": 3.348178137651822,
+      "grad_norm": 0.29140802122888354,
+      "learning_rate": 1.189800129803139e-05,
+      "loss": 0.2541,
+      "step": 827
+    },
+    {
+      "epoch": 3.3522267206477734,
+      "grad_norm": 0.2824708604550116,
+      "learning_rate": 1.1846327701166228e-05,
+      "loss": 0.2624,
+      "step": 828
+    },
+    {
+      "epoch": 3.3562753036437245,
+      "grad_norm": 0.28774617617000076,
+      "learning_rate": 1.1794719300847493e-05,
+      "loss": 0.2613,
+      "step": 829
+    },
+    {
+      "epoch": 3.360323886639676,
+      "grad_norm": 0.29558273520062733,
+      "learning_rate": 1.1743176509734582e-05,
+      "loss": 0.2667,
+      "step": 830
+    },
+    {
+      "epoch": 3.3643724696356276,
+      "grad_norm": 0.30328771735691945,
+      "learning_rate": 1.1691699739962275e-05,
+      "loss": 0.2514,
+      "step": 831
+    },
+    {
+      "epoch": 3.3684210526315788,
+      "grad_norm": 0.2932656059086371,
+      "learning_rate": 1.1640289403137456e-05,
+      "loss": 0.2676,
+      "step": 832
+    },
+    {
+      "epoch": 3.3724696356275303,
+      "grad_norm": 0.29612873777618265,
+      "learning_rate": 1.1588945910335802e-05,
+      "loss": 0.2533,
+      "step": 833
+    },
+    {
+      "epoch": 3.376518218623482,
+      "grad_norm": 0.30287503695667906,
+      "learning_rate": 1.1537669672098517e-05,
+      "loss": 0.252,
+      "step": 834
+    },
+    {
+      "epoch": 3.380566801619433,
+      "grad_norm": 0.29338525839458185,
+      "learning_rate": 1.1486461098429036e-05,
+      "loss": 0.2441,
+      "step": 835
+    },
+    {
+      "epoch": 3.3846153846153846,
+      "grad_norm": 0.28734578142829936,
+      "learning_rate": 1.1435320598789747e-05,
+      "loss": 0.246,
+      "step": 836
+    },
+    {
+      "epoch": 3.388663967611336,
+      "grad_norm": 0.2868066146594245,
+      "learning_rate": 1.1384248582098736e-05,
+      "loss": 0.244,
+      "step": 837
+    },
+    {
+      "epoch": 3.3927125506072873,
+      "grad_norm": 0.2852534248469639,
+      "learning_rate": 1.1333245456726467e-05,
+      "loss": 0.2378,
+      "step": 838
+    },
+    {
+      "epoch": 3.396761133603239,
+      "grad_norm": 0.28454732867329363,
+      "learning_rate": 1.128231163049258e-05,
+      "loss": 0.2519,
+      "step": 839
+    },
+    {
+      "epoch": 3.4008097165991904,
+      "grad_norm": 0.30550751362052914,
+      "learning_rate": 1.123144751066259e-05,
+      "loss": 0.2474,
+      "step": 840
+    },
+    {
+      "epoch": 3.4048582995951415,
+      "grad_norm": 0.27993891077021127,
+      "learning_rate": 1.1180653503944657e-05,
+      "loss": 0.2574,
+      "step": 841
+    },
+    {
+      "epoch": 3.408906882591093,
+      "grad_norm": 0.2895110566416651,
+      "learning_rate": 1.1129930016486287e-05,
+      "loss": 0.2541,
+      "step": 842
+    },
+    {
+      "epoch": 3.4129554655870447,
+      "grad_norm": 0.29552180466596656,
+      "learning_rate": 1.107927745387114e-05,
+      "loss": 0.2623,
+      "step": 843
+    },
+    {
+      "epoch": 3.417004048582996,
+      "grad_norm": 0.2983043846747731,
+      "learning_rate": 1.1028696221115766e-05,
+      "loss": 0.2459,
+      "step": 844
+    },
+    {
+      "epoch": 3.4210526315789473,
+      "grad_norm": 0.2866268640818632,
+      "learning_rate": 1.0978186722666341e-05,
+      "loss": 0.2514,
+      "step": 845
+    },
+    {
+      "epoch": 3.425101214574899,
+      "grad_norm": 0.29135774372307316,
+      "learning_rate": 1.0927749362395475e-05,
+      "loss": 0.2545,
+      "step": 846
+    },
+    {
+      "epoch": 3.42914979757085,
+      "grad_norm": 0.291892031062897,
+      "learning_rate": 1.0877384543598954e-05,
+      "loss": 0.2456,
+      "step": 847
+    },
+    {
+      "epoch": 3.4331983805668016,
+      "grad_norm": 0.2874059616332184,
+      "learning_rate": 1.0827092668992529e-05,
+      "loss": 0.2673,
+      "step": 848
+    },
+    {
+      "epoch": 3.437246963562753,
+      "grad_norm": 0.2911350247987412,
+      "learning_rate": 1.077687414070869e-05,
+      "loss": 0.234,
+      "step": 849
+    },
+    {
+      "epoch": 3.4412955465587043,
+      "grad_norm": 0.2979230563964137,
+      "learning_rate": 1.0726729360293451e-05,
+      "loss": 0.2651,
+      "step": 850
+    },
+    {
+      "epoch": 3.445344129554656,
+      "grad_norm": 0.2869442005450121,
+      "learning_rate": 1.0676658728703147e-05,
+      "loss": 0.2553,
+      "step": 851
+    },
+    {
+      "epoch": 3.4493927125506074,
+      "grad_norm": 0.29206441080757356,
+      "learning_rate": 1.0626662646301198e-05,
+      "loss": 0.2526,
+      "step": 852
+    },
+    {
+      "epoch": 3.4534412955465585,
+      "grad_norm": 0.28153489372106333,
+      "learning_rate": 1.0576741512854958e-05,
+      "loss": 0.2527,
+      "step": 853
+    },
+    {
+      "epoch": 3.45748987854251,
+      "grad_norm": 0.2743720018526786,
+      "learning_rate": 1.052689572753247e-05,
+      "loss": 0.2578,
+      "step": 854
+    },
+    {
+      "epoch": 3.4615384615384617,
+      "grad_norm": 0.29274206208772147,
+      "learning_rate": 1.0477125688899328e-05,
+      "loss": 0.2272,
+      "step": 855
+    },
+    {
+      "epoch": 3.465587044534413,
+      "grad_norm": 0.29020297650521387,
+      "learning_rate": 1.0427431794915406e-05,
+      "loss": 0.2737,
+      "step": 856
+    },
+    {
+      "epoch": 3.4696356275303644,
+      "grad_norm": 0.30095258826246085,
+      "learning_rate": 1.0377814442931769e-05,
+      "loss": 0.2536,
+      "step": 857
+    },
+    {
+      "epoch": 3.473684210526316,
+      "grad_norm": 0.2944553710779163,
+      "learning_rate": 1.0328274029687439e-05,
+      "loss": 0.2569,
+      "step": 858
+    },
+    {
+      "epoch": 3.477732793522267,
+      "grad_norm": 0.29902677799557653,
+      "learning_rate": 1.0278810951306253e-05,
+      "loss": 0.2562,
+      "step": 859
+    },
+    {
+      "epoch": 3.4817813765182186,
+      "grad_norm": 0.29112412565356,
+      "learning_rate": 1.0229425603293647e-05,
+      "loss": 0.2591,
+      "step": 860
+    },
+    {
+      "epoch": 3.48582995951417,
+      "grad_norm": 0.28661963594997464,
+      "learning_rate": 1.0180118380533558e-05,
+      "loss": 0.265,
+      "step": 861
+    },
+    {
+      "epoch": 3.4898785425101213,
+      "grad_norm": 0.29747233486767144,
+      "learning_rate": 1.0130889677285225e-05,
+      "loss": 0.2482,
+      "step": 862
+    },
+    {
+      "epoch": 3.493927125506073,
+      "grad_norm": 0.3023928799114594,
+      "learning_rate": 1.008173988718005e-05,
+      "loss": 0.2583,
+      "step": 863
+    },
+    {
+      "epoch": 3.4979757085020244,
+      "grad_norm": 0.27627591272670465,
+      "learning_rate": 1.0032669403218439e-05,
+      "loss": 0.2476,
+      "step": 864
+    },
+    {
+      "epoch": 3.5020242914979756,
+      "grad_norm": 0.2938985273183619,
+      "learning_rate": 9.983678617766677e-06,
+      "loss": 0.2489,
+      "step": 865
+    },
+    {
+      "epoch": 3.506072874493927,
+      "grad_norm": 0.3014912778221674,
+      "learning_rate": 9.934767922553788e-06,
+      "loss": 0.2538,
+      "step": 866
+    },
+    {
+      "epoch": 3.5101214574898787,
+      "grad_norm": 0.26870580200125815,
+      "learning_rate": 9.885937708668371e-06,
+      "loss": 0.2712,
+      "step": 867
+    },
+    {
+      "epoch": 3.51417004048583,
+      "grad_norm": 0.2861244202875831,
+      "learning_rate": 9.837188366555524e-06,
+      "loss": 0.2727,
+      "step": 868
+    },
+    {
+      "epoch": 3.5182186234817814,
+      "grad_norm": 0.30095058123769525,
+      "learning_rate": 9.788520286013691e-06,
+      "loss": 0.2613,
+      "step": 869
+    },
+    {
+      "epoch": 3.522267206477733,
+      "grad_norm": 0.27510259648830176,
+      "learning_rate": 9.739933856191537e-06,
+      "loss": 0.2475,
+      "step": 870
+    },
+    {
+      "epoch": 3.526315789473684,
+      "grad_norm": 0.2894124701883834,
+      "learning_rate": 9.691429465584866e-06,
+      "loss": 0.2518,
+      "step": 871
+    },
+    {
+      "epoch": 3.5303643724696356,
+      "grad_norm": 0.29272064504702183,
+      "learning_rate": 9.643007502033492e-06,
+      "loss": 0.2534,
+      "step": 872
+    },
+    {
+      "epoch": 3.534412955465587,
+      "grad_norm": 0.28570392352674884,
+      "learning_rate": 9.594668352718152e-06,
+      "loss": 0.2641,
+      "step": 873
+    },
+    {
+      "epoch": 3.5384615384615383,
+      "grad_norm": 0.2802382658223346,
+      "learning_rate": 9.546412404157383e-06,
+      "loss": 0.2715,
+      "step": 874
+    },
+    {
+      "epoch": 3.54251012145749,
+      "grad_norm": 0.2788825661578156,
+      "learning_rate": 9.498240042204471e-06,
+      "loss": 0.2764,
+      "step": 875
+    },
+    {
+      "epoch": 3.5465587044534415,
+      "grad_norm": 0.27663540559266525,
+      "learning_rate": 9.450151652044339e-06,
+      "loss": 0.2667,
+      "step": 876
+    },
+    {
+      "epoch": 3.5506072874493926,
+      "grad_norm": 0.2886900377301999,
+      "learning_rate": 9.402147618190472e-06,
+      "loss": 0.2692,
+      "step": 877
+    },
+    {
+      "epoch": 3.554655870445344,
+      "grad_norm": 0.279858426098991,
+      "learning_rate": 9.354228324481841e-06,
+      "loss": 0.2443,
+      "step": 878
+    },
+    {
+      "epoch": 3.5587044534412957,
+      "grad_norm": 0.28672960611983356,
+      "learning_rate": 9.306394154079852e-06,
+      "loss": 0.2563,
+      "step": 879
+    },
+    {
+      "epoch": 3.562753036437247,
+      "grad_norm": 0.2913084751128345,
+      "learning_rate": 9.258645489465254e-06,
+      "loss": 0.2462,
+      "step": 880
+    },
+    {
+      "epoch": 3.5668016194331984,
+      "grad_norm": 0.2917343639227354,
+      "learning_rate": 9.21098271243509e-06,
+      "loss": 0.2616,
+      "step": 881
+    },
+    {
+      "epoch": 3.57085020242915,
+      "grad_norm": 0.2863137481156259,
+      "learning_rate": 9.163406204099659e-06,
+      "loss": 0.2609,
+      "step": 882
+    },
+    {
+      "epoch": 3.574898785425101,
+      "grad_norm": 0.2937686390316401,
+      "learning_rate": 9.115916344879452e-06,
+      "loss": 0.2642,
+      "step": 883
+    },
+    {
+      "epoch": 3.5789473684210527,
+      "grad_norm": 0.27774808240040405,
+      "learning_rate": 9.068513514502126e-06,
+      "loss": 0.2449,
+      "step": 884
+    },
+    {
+      "epoch": 3.582995951417004,
+      "grad_norm": 0.2837742741594504,
+      "learning_rate": 9.021198091999439e-06,
+      "loss": 0.2644,
+      "step": 885
+    },
+    {
+      "epoch": 3.5870445344129553,
+      "grad_norm": 0.2788648104282438,
+      "learning_rate": 8.973970455704251e-06,
+      "loss": 0.2447,
+      "step": 886
+    },
+    {
+      "epoch": 3.591093117408907,
+      "grad_norm": 0.2850176565785955,
+      "learning_rate": 8.926830983247495e-06,
+      "loss": 0.2606,
+      "step": 887
+    },
+    {
+      "epoch": 3.5951417004048585,
+      "grad_norm": 0.29066892103574893,
+      "learning_rate": 8.879780051555116e-06,
+      "loss": 0.2522,
+      "step": 888
+    },
+    {
+      "epoch": 3.5991902834008096,
+      "grad_norm": 0.28662345110237275,
+      "learning_rate": 8.83281803684512e-06,
+      "loss": 0.2505,
+      "step": 889
+    },
+    {
+      "epoch": 3.603238866396761,
+      "grad_norm": 0.27808025666681946,
+      "learning_rate": 8.785945314624532e-06,
+      "loss": 0.2548,
+      "step": 890
+    },
+    {
+      "epoch": 3.6072874493927127,
+      "grad_norm": 0.2975282065651329,
+      "learning_rate": 8.739162259686385e-06,
+      "loss": 0.2646,
+      "step": 891
+    },
+    {
+      "epoch": 3.611336032388664,
+      "grad_norm": 0.30397589985816936,
+      "learning_rate": 8.692469246106734e-06,
+      "loss": 0.2388,
+      "step": 892
+    },
+    {
+      "epoch": 3.6153846153846154,
+      "grad_norm": 0.28323433271783577,
+      "learning_rate": 8.645866647241677e-06,
+      "loss": 0.2562,
+      "step": 893
+    },
+    {
+      "epoch": 3.619433198380567,
+      "grad_norm": 0.29337341052630794,
+      "learning_rate": 8.599354835724348e-06,
+      "loss": 0.2545,
+      "step": 894
+    },
+    {
+      "epoch": 3.623481781376518,
+      "grad_norm": 0.304754352743764,
+      "learning_rate": 8.552934183461958e-06,
+      "loss": 0.2323,
+      "step": 895
+    },
+    {
+      "epoch": 3.6275303643724697,
+      "grad_norm": 0.2826457679396477,
+      "learning_rate": 8.506605061632805e-06,
+      "loss": 0.2737,
+      "step": 896
+    },
+    {
+      "epoch": 3.6315789473684212,
+      "grad_norm": 0.2931297542315252,
+      "learning_rate": 8.460367840683308e-06,
+      "loss": 0.237,
+      "step": 897
+    },
+    {
+      "epoch": 3.6356275303643724,
+      "grad_norm": 0.2920070644638484,
+      "learning_rate": 8.414222890325065e-06,
+      "loss": 0.2759,
+      "step": 898
+    },
+    {
+      "epoch": 3.639676113360324,
+      "grad_norm": 0.3085891399447288,
+      "learning_rate": 8.368170579531855e-06,
+      "loss": 0.2695,
+      "step": 899
+    },
+    {
+      "epoch": 3.6437246963562755,
+      "grad_norm": 0.27236880176569506,
+      "learning_rate": 8.322211276536733e-06,
+      "loss": 0.2565,
+      "step": 900
+    },
+    {
+      "epoch": 3.6477732793522266,
+      "grad_norm": 0.28626448640884,
+      "learning_rate": 8.27634534882907e-06,
+      "loss": 0.2524,
+      "step": 901
+    },
+    {
+      "epoch": 3.651821862348178,
+      "grad_norm": 0.29934442408500206,
+      "learning_rate": 8.230573163151585e-06,
+      "loss": 0.2557,
+      "step": 902
+    },
+    {
+      "epoch": 3.6558704453441297,
+      "grad_norm": 0.28756902772644477,
+      "learning_rate": 8.184895085497466e-06,
+      "loss": 0.2487,
+      "step": 903
+    },
+    {
+      "epoch": 3.659919028340081,
+      "grad_norm": 0.29265719549116803,
+      "learning_rate": 8.139311481107401e-06,
+      "loss": 0.2459,
+      "step": 904
+    },
+    {
+      "epoch": 3.6639676113360324,
+      "grad_norm": 0.291579197610245,
+      "learning_rate": 8.093822714466686e-06,
+      "loss": 0.24,
+      "step": 905
+    },
+    {
+      "epoch": 3.668016194331984,
+      "grad_norm": 0.29974939472345713,
+      "learning_rate": 8.048429149302271e-06,
+      "loss": 0.2476,
+      "step": 906
+    },
+    {
+      "epoch": 3.672064777327935,
+      "grad_norm": 0.28764248458538505,
+      "learning_rate": 8.003131148579902e-06,
+      "loss": 0.2518,
+      "step": 907
+    },
+    {
+      "epoch": 3.6761133603238867,
+      "grad_norm": 0.2816022638577354,
+      "learning_rate": 7.957929074501182e-06,
+      "loss": 0.2502,
+      "step": 908
+    },
+    {
+      "epoch": 3.6801619433198383,
+      "grad_norm": 0.2835818684612923,
+      "learning_rate": 7.912823288500695e-06,
+      "loss": 0.2483,
+      "step": 909
+    },
+    {
+      "epoch": 3.6842105263157894,
+      "grad_norm": 0.28359698691191504,
+      "learning_rate": 7.867814151243102e-06,
+      "loss": 0.2612,
+      "step": 910
+    },
+    {
+      "epoch": 3.688259109311741,
+      "grad_norm": 0.27511532411902595,
+      "learning_rate": 7.822902022620267e-06,
+      "loss": 0.2472,
+      "step": 911
+    },
+    {
+      "epoch": 3.6923076923076925,
+      "grad_norm": 0.2779552155845146,
+      "learning_rate": 7.778087261748377e-06,
+      "loss": 0.2696,
+      "step": 912
+    },
+    {
+      "epoch": 3.6963562753036436,
+      "grad_norm": 0.27905799286635163,
+      "learning_rate": 7.733370226965056e-06,
+      "loss": 0.2646,
+      "step": 913
+    },
+    {
+      "epoch": 3.700404858299595,
+      "grad_norm": 0.2799854066250846,
+      "learning_rate": 7.688751275826527e-06,
+      "loss": 0.2424,
+      "step": 914
+    },
+    {
+      "epoch": 3.7044534412955468,
+      "grad_norm": 0.2774389887749684,
+      "learning_rate": 7.644230765104735e-06,
+      "loss": 0.2597,
+      "step": 915
+    },
+    {
+      "epoch": 3.708502024291498,
+      "grad_norm": 0.2760566115522948,
+      "learning_rate": 7.599809050784502e-06,
+      "loss": 0.2398,
+      "step": 916
+    },
+    {
+      "epoch": 3.7125506072874495,
+      "grad_norm": 0.3031427031939067,
+      "learning_rate": 7.555486488060661e-06,
+      "loss": 0.2334,
+      "step": 917
+    },
+    {
+      "epoch": 3.716599190283401,
+      "grad_norm": 0.28007630678765044,
+      "learning_rate": 7.511263431335249e-06,
+      "loss": 0.26,
+      "step": 918
+    },
+    {
+      "epoch": 3.720647773279352,
+      "grad_norm": 0.29225705887342507,
+      "learning_rate": 7.467140234214656e-06,
+      "loss": 0.2529,
+      "step": 919
+    },
+    {
+      "epoch": 3.7246963562753037,
+      "grad_norm": 0.2898341816868658,
+      "learning_rate": 7.423117249506777e-06,
+      "loss": 0.2291,
+      "step": 920
+    },
+    {
+      "epoch": 3.7287449392712553,
+      "grad_norm": 0.2781055256640129,
+      "learning_rate": 7.379194829218231e-06,
+      "loss": 0.2494,
+      "step": 921
+    },
+    {
+      "epoch": 3.7327935222672064,
+      "grad_norm": 0.2817901840869381,
+      "learning_rate": 7.335373324551518e-06,
+      "loss": 0.2495,
+      "step": 922
+    },
+    {
+      "epoch": 3.736842105263158,
+      "grad_norm": 0.25866345169804983,
+      "learning_rate": 7.291653085902224e-06,
+      "loss": 0.2498,
+      "step": 923
+    },
+    {
+      "epoch": 3.7408906882591095,
+      "grad_norm": 0.2716674409501263,
+      "learning_rate": 7.248034462856213e-06,
+      "loss": 0.2565,
+      "step": 924
+    },
+    {
+      "epoch": 3.7449392712550607,
+      "grad_norm": 0.2769447139336569,
+      "learning_rate": 7.20451780418683e-06,
+      "loss": 0.261,
+      "step": 925
+    },
+    {
+      "epoch": 3.748987854251012,
+      "grad_norm": 0.2775316519482268,
+      "learning_rate": 7.161103457852128e-06,
+      "loss": 0.2494,
+      "step": 926
+    },
+    {
+      "epoch": 3.753036437246964,
+      "grad_norm": 0.27252131224184367,
+      "learning_rate": 7.117791770992046e-06,
+      "loss": 0.2586,
+      "step": 927
+    },
+    {
+      "epoch": 3.757085020242915,
+      "grad_norm": 0.27207478359564624,
+      "learning_rate": 7.074583089925688e-06,
+      "loss": 0.2525,
+      "step": 928
+    },
+    {
+      "epoch": 3.7611336032388665,
+      "grad_norm": 0.28924205695581884,
+      "learning_rate": 7.0314777601485155e-06,
+      "loss": 0.2433,
+      "step": 929
+    },
+    {
+      "epoch": 3.765182186234818,
+      "grad_norm": 0.26946979557333284,
+      "learning_rate": 6.988476126329604e-06,
+      "loss": 0.2567,
+      "step": 930
+    },
+    {
+      "epoch": 3.769230769230769,
+      "grad_norm": 0.27782281749657534,
+      "learning_rate": 6.945578532308857e-06,
+      "loss": 0.2609,
+      "step": 931
+    },
+    {
+      "epoch": 3.7732793522267207,
+      "grad_norm": 0.28387680994606085,
+      "learning_rate": 6.902785321094301e-06,
+      "loss": 0.2603,
+      "step": 932
+    },
+    {
+      "epoch": 3.7773279352226723,
+      "grad_norm": 0.26310756946820796,
+      "learning_rate": 6.8600968348593115e-06,
+      "loss": 0.251,
+      "step": 933
+    },
+    {
+      "epoch": 3.7813765182186234,
+      "grad_norm": 0.27077371365910097,
+      "learning_rate": 6.817513414939892e-06,
+      "loss": 0.2439,
+      "step": 934
+    },
+    {
+      "epoch": 3.785425101214575,
+      "grad_norm": 0.29808162888669354,
+      "learning_rate": 6.775035401831924e-06,
+      "loss": 0.2272,
+      "step": 935
+    },
+    {
+      "epoch": 3.7894736842105265,
+      "grad_norm": 0.27408405441179223,
+      "learning_rate": 6.732663135188473e-06,
+      "loss": 0.238,
+      "step": 936
+    },
+    {
+      "epoch": 3.7935222672064777,
+      "grad_norm": 0.28062952642080813,
+      "learning_rate": 6.690396953817055e-06,
+      "loss": 0.2485,
+      "step": 937
+    },
+    {
+      "epoch": 3.7975708502024292,
+      "grad_norm": 0.3091915959678639,
+      "learning_rate": 6.648237195676927e-06,
+      "loss": 0.2506,
+      "step": 938
+    },
+    {
+      "epoch": 3.801619433198381,
+      "grad_norm": 0.28022644463759494,
+      "learning_rate": 6.606184197876395e-06,
+      "loss": 0.2565,
+      "step": 939
+    },
+    {
+      "epoch": 3.805668016194332,
+      "grad_norm": 0.2687778365811384,
+      "learning_rate": 6.564238296670109e-06,
+      "loss": 0.2462,
+      "step": 940
+    },
+    {
+      "epoch": 3.8097165991902835,
+      "grad_norm": 0.28272777466874993,
+      "learning_rate": 6.52239982745638e-06,
+      "loss": 0.2512,
+      "step": 941
+    },
+    {
+      "epoch": 3.813765182186235,
+      "grad_norm": 0.2755131600338027,
+      "learning_rate": 6.480669124774484e-06,
+      "loss": 0.2768,
+      "step": 942
+    },
+    {
+      "epoch": 3.817813765182186,
+      "grad_norm": 0.29505847615098446,
+      "learning_rate": 6.439046522302013e-06,
+      "loss": 0.2445,
+      "step": 943
+    },
+    {
+      "epoch": 3.8218623481781377,
+      "grad_norm": 0.28360385424929313,
+      "learning_rate": 6.397532352852191e-06,
+      "loss": 0.2514,
+      "step": 944
+    },
+    {
+      "epoch": 3.8259109311740893,
+      "grad_norm": 0.2655030256565956,
+      "learning_rate": 6.356126948371202e-06,
+      "loss": 0.243,
+      "step": 945
+    },
+    {
+      "epoch": 3.8299595141700404,
+      "grad_norm": 0.27285087466358016,
+      "learning_rate": 6.314830639935559e-06,
+      "loss": 0.2592,
+      "step": 946
+    },
+    {
+      "epoch": 3.834008097165992,
+      "grad_norm": 0.27554707944474865,
+      "learning_rate": 6.273643757749448e-06,
+      "loss": 0.2441,
+      "step": 947
+    },
+    {
+      "epoch": 3.8380566801619436,
+      "grad_norm": 0.28594353705942543,
+      "learning_rate": 6.2325666311420856e-06,
+      "loss": 0.2432,
+      "step": 948
+    },
+    {
+      "epoch": 3.8421052631578947,
+      "grad_norm": 0.2780347447154959,
+      "learning_rate": 6.1915995885650675e-06,
+      "loss": 0.2456,
+      "step": 949
+    },
+    {
+      "epoch": 3.8461538461538463,
+      "grad_norm": 0.27003589161208386,
+      "learning_rate": 6.150742957589784e-06,
+      "loss": 0.2301,
+      "step": 950
+    },
+    {
+      "epoch": 3.850202429149798,
+      "grad_norm": 0.2785557548705016,
+      "learning_rate": 6.109997064904774e-06,
+      "loss": 0.267,
+      "step": 951
+    },
+    {
+      "epoch": 3.854251012145749,
+      "grad_norm": 0.2759354827136701,
+      "learning_rate": 6.069362236313092e-06,
+      "loss": 0.2629,
+      "step": 952
+    },
+    {
+      "epoch": 3.8582995951417005,
+      "grad_norm": 0.27802499366725253,
+      "learning_rate": 6.0288387967297565e-06,
+      "loss": 0.2507,
+      "step": 953
+    },
+    {
+      "epoch": 3.862348178137652,
+      "grad_norm": 0.26553840095066106,
+      "learning_rate": 5.988427070179108e-06,
+      "loss": 0.2476,
+      "step": 954
+    },
+    {
+      "epoch": 3.866396761133603,
+      "grad_norm": 0.27017986709608066,
+      "learning_rate": 5.948127379792234e-06,
+      "loss": 0.2443,
+      "step": 955
+    },
+    {
+      "epoch": 3.8704453441295548,
+      "grad_norm": 0.2744668667217903,
+      "learning_rate": 5.90794004780439e-06,
+      "loss": 0.261,
+      "step": 956
+    },
+    {
+      "epoch": 3.8744939271255063,
+      "grad_norm": 0.26783378432248556,
+      "learning_rate": 5.867865395552405e-06,
+      "loss": 0.2547,
+      "step": 957
+    },
+    {
+      "epoch": 3.8785425101214575,
+      "grad_norm": 0.27127167398121366,
+      "learning_rate": 5.8279037434721364e-06,
+      "loss": 0.2475,
+      "step": 958
+    },
+    {
+      "epoch": 3.882591093117409,
+      "grad_norm": 0.26796693970002844,
+      "learning_rate": 5.788055411095892e-06,
+      "loss": 0.2585,
+      "step": 959
+    },
+    {
+      "epoch": 3.8866396761133606,
+      "grad_norm": 0.263909788031495,
+      "learning_rate": 5.7483207170498625e-06,
+      "loss": 0.2498,
+      "step": 960
+    },
+    {
+      "epoch": 3.8906882591093117,
+      "grad_norm": 0.2568981226705071,
+      "learning_rate": 5.708699979051613e-06,
+      "loss": 0.2497,
+      "step": 961
+    },
+    {
+      "epoch": 3.8947368421052633,
+      "grad_norm": 0.2537809266675906,
+      "learning_rate": 5.66919351390751e-06,
+      "loss": 0.2486,
+      "step": 962
+    },
+    {
+      "epoch": 3.898785425101215,
+      "grad_norm": 0.2731340810945399,
+      "learning_rate": 5.629801637510188e-06,
+      "loss": 0.2411,
+      "step": 963
+    },
+    {
+      "epoch": 3.902834008097166,
+      "grad_norm": 0.26786534690866937,
+      "learning_rate": 5.590524664836043e-06,
+      "loss": 0.2408,
+      "step": 964
+    },
+    {
+      "epoch": 3.9068825910931175,
+      "grad_norm": 0.2758912944752292,
+      "learning_rate": 5.551362909942708e-06,
+      "loss": 0.2693,
+      "step": 965
+    },
+    {
+      "epoch": 3.910931174089069,
+      "grad_norm": 0.2694467282199529,
+      "learning_rate": 5.51231668596653e-06,
+      "loss": 0.2546,
+      "step": 966
+    },
+    {
+      "epoch": 3.91497975708502,
+      "grad_norm": 0.2708275809897375,
+      "learning_rate": 5.47338630512007e-06,
+      "loss": 0.2618,
+      "step": 967
+    },
+    {
+      "epoch": 3.919028340080972,
+      "grad_norm": 0.26548829000516244,
+      "learning_rate": 5.434572078689617e-06,
+      "loss": 0.2628,
+      "step": 968
+    },
+    {
+      "epoch": 3.9230769230769234,
+      "grad_norm": 0.27711127352311127,
+      "learning_rate": 5.395874317032694e-06,
+      "loss": 0.2465,
+      "step": 969
+    },
+    {
+      "epoch": 3.9271255060728745,
+      "grad_norm": 0.27022416751226674,
+      "learning_rate": 5.357293329575573e-06,
+      "loss": 0.2442,
+      "step": 970
+    },
+    {
+      "epoch": 3.931174089068826,
+      "grad_norm": 0.27594131850390013,
+      "learning_rate": 5.318829424810797e-06,
+      "loss": 0.2457,
+      "step": 971
+    },
+    {
+      "epoch": 3.9352226720647776,
+      "grad_norm": 0.278332783523553,
+      "learning_rate": 5.280482910294728e-06,
+      "loss": 0.2673,
+      "step": 972
+    },
+    {
+      "epoch": 3.9392712550607287,
+      "grad_norm": 0.2619672027625368,
+      "learning_rate": 5.242254092645078e-06,
+      "loss": 0.2548,
+      "step": 973
+    },
+    {
+      "epoch": 3.9433198380566803,
+      "grad_norm": 0.2597608477989495,
+      "learning_rate": 5.2041432775384445e-06,
+      "loss": 0.2691,
+      "step": 974
+    },
+    {
+      "epoch": 3.9473684210526314,
+      "grad_norm": 0.2704464342479962,
+      "learning_rate": 5.166150769707892e-06,
+      "loss": 0.2505,
+      "step": 975
+    },
+    {
+      "epoch": 3.951417004048583,
+      "grad_norm": 0.2730422681749816,
+      "learning_rate": 5.128276872940503e-06,
+      "loss": 0.2588,
+      "step": 976
+    },
+    {
+      "epoch": 3.9554655870445345,
+      "grad_norm": 0.2729852542768067,
+      "learning_rate": 5.090521890074953e-06,
+      "loss": 0.2593,
+      "step": 977
+    },
+    {
+      "epoch": 3.9595141700404857,
+      "grad_norm": 0.267444100192479,
+      "learning_rate": 5.0528861229990675e-06,
+      "loss": 0.2384,
+      "step": 978
+    },
+    {
+      "epoch": 3.9635627530364372,
+      "grad_norm": 0.2684351544956101,
+      "learning_rate": 5.015369872647442e-06,
+      "loss": 0.2647,
+      "step": 979
+    },
+    {
+      "epoch": 3.967611336032389,
+      "grad_norm": 0.28388560159531234,
+      "learning_rate": 4.9779734389990195e-06,
+      "loss": 0.2541,
+      "step": 980
+    },
+    {
+      "epoch": 3.97165991902834,
+      "grad_norm": 0.27252045297474226,
+      "learning_rate": 4.940697121074675e-06,
+      "loss": 0.2391,
+      "step": 981
+    },
+    {
+      "epoch": 3.9757085020242915,
+      "grad_norm": 0.2614734280828838,
+      "learning_rate": 4.903541216934864e-06,
+      "loss": 0.2713,
+      "step": 982
+    },
+    {
+      "epoch": 3.979757085020243,
+      "grad_norm": 0.2602662575160767,
+      "learning_rate": 4.8665060236772e-06,
+      "loss": 0.2505,
+      "step": 983
+    },
+    {
+      "epoch": 3.983805668016194,
+      "grad_norm": 0.2749280446228705,
+      "learning_rate": 4.829591837434102e-06,
+      "loss": 0.2491,
+      "step": 984
+    },
+    {
+      "epoch": 3.9878542510121457,
+      "grad_norm": 0.2709674593383219,
+      "learning_rate": 4.792798953370424e-06,
+      "loss": 0.2392,
+      "step": 985
+    },
+    {
+      "epoch": 3.9919028340080973,
+      "grad_norm": 0.2606720069523475,
+      "learning_rate": 4.756127665681083e-06,
+      "loss": 0.2622,
+      "step": 986
+    },
+    {
+      "epoch": 3.9959514170040484,
+      "grad_norm": 0.27313646275761916,
+      "learning_rate": 4.719578267588725e-06,
+      "loss": 0.2584,
+      "step": 987
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 0.26979482242937736,
+      "learning_rate": 4.6831510513413525e-06,
+      "loss": 0.2555,
+      "step": 988
+    },
+    {
+      "epoch": 4.004048582995951,
+      "grad_norm": 0.4935647574678857,
+      "learning_rate": 4.6468463082100176e-06,
+      "loss": 0.1892,
+      "step": 989
+    },
+    {
+      "epoch": 4.008097165991903,
+      "grad_norm": 0.3443459124893285,
+      "learning_rate": 4.610664328486485e-06,
+      "loss": 0.1995,
+      "step": 990
+    },
+    {
+      "epoch": 4.012145748987854,
+      "grad_norm": 0.256535004572781,
+      "learning_rate": 4.5746054014809e-06,
+      "loss": 0.1747,
+      "step": 991
+    },
+    {
+      "epoch": 4.016194331983805,
+      "grad_norm": 0.3778672395193411,
+      "learning_rate": 4.5386698155194766e-06,
+      "loss": 0.1948,
+      "step": 992
+    },
+    {
+      "epoch": 4.020242914979757,
+      "grad_norm": 0.5252821724922182,
+      "learning_rate": 4.502857857942204e-06,
+      "loss": 0.2028,
+      "step": 993
+    },
+    {
+      "epoch": 4.0242914979757085,
+      "grad_norm": 0.410690872258837,
+      "learning_rate": 4.467169815100549e-06,
+      "loss": 0.1927,
+      "step": 994
+    },
+    {
+      "epoch": 4.02834008097166,
+      "grad_norm": 0.30176104907063955,
+      "learning_rate": 4.431605972355137e-06,
+      "loss": 0.2008,
+      "step": 995
+    },
+    {
+      "epoch": 4.032388663967612,
+      "grad_norm": 0.3537374540392164,
+      "learning_rate": 4.396166614073516e-06,
+      "loss": 0.1939,
+      "step": 996
+    },
+    {
+      "epoch": 4.036437246963563,
+      "grad_norm": 0.38835460149339923,
+      "learning_rate": 4.360852023627846e-06,
+      "loss": 0.2006,
+      "step": 997
+    },
+    {
+      "epoch": 4.040485829959514,
+      "grad_norm": 0.3943386209832084,
+      "learning_rate": 4.325662483392656e-06,
+      "loss": 0.1858,
+      "step": 998
+    },
+    {
+      "epoch": 4.044534412955466,
+      "grad_norm": 0.3013714090851541,
+      "learning_rate": 4.29059827474257e-06,
+      "loss": 0.1905,
+      "step": 999
+    },
+    {
+      "epoch": 4.048582995951417,
+      "grad_norm": 0.27592959210063916,
+      "learning_rate": 4.255659678050063e-06,
+      "loss": 0.1732,
+      "step": 1000
+    },
+    {
+      "epoch": 4.052631578947368,
+      "grad_norm": 0.3605117999055238,
+      "learning_rate": 4.22084697268323e-06,
+      "loss": 0.1948,
+      "step": 1001
+    },
+    {
+      "epoch": 4.05668016194332,
+      "grad_norm": 0.4081850690320335,
+      "learning_rate": 4.186160437003535e-06,
+      "loss": 0.195,
+      "step": 1002
+    },
+    {
+      "epoch": 4.060728744939271,
+      "grad_norm": 0.3393557454100399,
+      "learning_rate": 4.151600348363582e-06,
+      "loss": 0.1993,
+      "step": 1003
+    },
+    {
+      "epoch": 4.064777327935222,
+      "grad_norm": 0.27265951397472377,
+      "learning_rate": 4.11716698310493e-06,
+      "loss": 0.2,
+      "step": 1004
+    },
+    {
+      "epoch": 4.068825910931174,
+      "grad_norm": 0.2907281506658532,
+      "learning_rate": 4.082860616555848e-06,
+      "loss": 0.2067,
+      "step": 1005
+    },
+    {
+      "epoch": 4.0728744939271255,
+      "grad_norm": 0.35223174592200746,
+      "learning_rate": 4.048681523029125e-06,
+      "loss": 0.1905,
+      "step": 1006
+    },
+    {
+      "epoch": 4.076923076923077,
+      "grad_norm": 0.3256235031913515,
+      "learning_rate": 4.014629975819882e-06,
+      "loss": 0.1954,
+      "step": 1007
+    },
+    {
+      "epoch": 4.080971659919029,
+      "grad_norm": 0.2947065231870294,
+      "learning_rate": 3.9807062472033855e-06,
+      "loss": 0.2068,
+      "step": 1008
+    },
+    {
+      "epoch": 4.08502024291498,
+      "grad_norm": 0.2786609335886333,
+      "learning_rate": 3.94691060843287e-06,
+      "loss": 0.199,
+      "step": 1009
+    },
+    {
+      "epoch": 4.089068825910931,
+      "grad_norm": 0.29817711314134443,
+      "learning_rate": 3.913243329737353e-06,
+      "loss": 0.2071,
+      "step": 1010
+    },
+    {
+      "epoch": 4.093117408906883,
+      "grad_norm": 0.2951366724173837,
+      "learning_rate": 3.879704680319503e-06,
+      "loss": 0.1906,
+      "step": 1011
+    },
+    {
+      "epoch": 4.097165991902834,
+      "grad_norm": 0.2926385700461069,
+      "learning_rate": 3.846294928353471e-06,
+      "loss": 0.2103,
+      "step": 1012
+    },
+    {
+      "epoch": 4.101214574898785,
+      "grad_norm": 0.27622457639908343,
+      "learning_rate": 3.8130143409827345e-06,
+      "loss": 0.1813,
+      "step": 1013
+    },
+    {
+      "epoch": 4.105263157894737,
+      "grad_norm": 0.2817170233243934,
+      "learning_rate": 3.7798631843179887e-06,
+      "loss": 0.1848,
+      "step": 1014
+    },
+    {
+      "epoch": 4.109311740890688,
+      "grad_norm": 0.2771665953839175,
+      "learning_rate": 3.7468417234349973e-06,
+      "loss": 0.2025,
+      "step": 1015
+    },
+    {
+      "epoch": 4.113360323886639,
+      "grad_norm": 0.2576244760574326,
+      "learning_rate": 3.7139502223724823e-06,
+      "loss": 0.2009,
+      "step": 1016
+    },
+    {
+      "epoch": 4.117408906882591,
+      "grad_norm": 0.26672890992048,
+      "learning_rate": 3.6811889441300095e-06,
+      "loss": 0.1907,
+      "step": 1017
+    },
+    {
+      "epoch": 4.1214574898785425,
+      "grad_norm": 0.2614979990287259,
+      "learning_rate": 3.648558150665886e-06,
+      "loss": 0.1973,
+      "step": 1018
+    },
+    {
+      "epoch": 4.125506072874494,
+      "grad_norm": 0.248588311025278,
+      "learning_rate": 3.616058102895075e-06,
+      "loss": 0.1952,
+      "step": 1019
+    },
+    {
+      "epoch": 4.129554655870446,
+      "grad_norm": 0.2472384331934195,
+      "learning_rate": 3.583689060687079e-06,
+      "loss": 0.1857,
+      "step": 1020
+    },
+    {
+      "epoch": 4.133603238866397,
+      "grad_norm": 0.2485710135895015,
+      "learning_rate": 3.5514512828639023e-06,
+      "loss": 0.1876,
+      "step": 1021
+    },
+    {
+      "epoch": 4.137651821862348,
+      "grad_norm": 0.2584463052223274,
+      "learning_rate": 3.51934502719796e-06,
+      "loss": 0.1995,
+      "step": 1022
+    },
+    {
+      "epoch": 4.1417004048583,
+      "grad_norm": 0.26562706463646724,
+      "learning_rate": 3.4873705504100185e-06,
+      "loss": 0.1883,
+      "step": 1023
+    },
+    {
+      "epoch": 4.145748987854251,
+      "grad_norm": 0.261968915679784,
+      "learning_rate": 3.4555281081671345e-06,
+      "loss": 0.1817,
+      "step": 1024
+    },
+    {
+      "epoch": 4.149797570850202,
+      "grad_norm": 0.23614185281140015,
+      "learning_rate": 3.4238179550806326e-06,
+      "loss": 0.1788,
+      "step": 1025
+    },
+    {
+      "epoch": 4.153846153846154,
+      "grad_norm": 0.24557638244745958,
+      "learning_rate": 3.3922403447040497e-06,
+      "loss": 0.1927,
+      "step": 1026
+    },
+    {
+      "epoch": 4.157894736842105,
+      "grad_norm": 0.2529798466676429,
+      "learning_rate": 3.3607955295311247e-06,
+      "loss": 0.1814,
+      "step": 1027
+    },
+    {
+      "epoch": 4.161943319838056,
+      "grad_norm": 0.26174070190750287,
+      "learning_rate": 3.32948376099375e-06,
+      "loss": 0.1896,
+      "step": 1028
+    },
+    {
+      "epoch": 4.165991902834008,
+      "grad_norm": 0.24030437360333315,
+      "learning_rate": 3.2983052894599976e-06,
+      "loss": 0.2016,
+      "step": 1029
+    },
+    {
+      "epoch": 4.17004048582996,
+      "grad_norm": 0.23923766951124495,
+      "learning_rate": 3.2672603642320945e-06,
+      "loss": 0.2012,
+      "step": 1030
+    },
+    {
+      "epoch": 4.174089068825911,
+      "grad_norm": 0.25142041943634036,
+      "learning_rate": 3.236349233544438e-06,
+      "loss": 0.1978,
+      "step": 1031
+    },
+    {
+      "epoch": 4.178137651821863,
+      "grad_norm": 0.25834759172759264,
+      "learning_rate": 3.205572144561604e-06,
+      "loss": 0.1846,
+      "step": 1032
+    },
+    {
+      "epoch": 4.182186234817814,
+      "grad_norm": 0.2451762987951842,
+      "learning_rate": 3.174929343376374e-06,
+      "loss": 0.1935,
+      "step": 1033
+    },
+    {
+      "epoch": 4.186234817813765,
+      "grad_norm": 0.25330931602696144,
+      "learning_rate": 3.1444210750077777e-06,
+      "loss": 0.1814,
+      "step": 1034
+    },
+    {
+      "epoch": 4.190283400809717,
+      "grad_norm": 0.26137523617074376,
+      "learning_rate": 3.1140475833991046e-06,
+      "loss": 0.1781,
+      "step": 1035
+    },
+    {
+      "epoch": 4.194331983805668,
+      "grad_norm": 0.2531087732790562,
+      "learning_rate": 3.0838091114159895e-06,
+      "loss": 0.1964,
+      "step": 1036
+    },
+    {
+      "epoch": 4.198380566801619,
+      "grad_norm": 0.2601925938427493,
+      "learning_rate": 3.0537059008444524e-06,
+      "loss": 0.1839,
+      "step": 1037
+    },
+    {
+      "epoch": 4.202429149797571,
+      "grad_norm": 0.2453818176572395,
+      "learning_rate": 3.0237381923889563e-06,
+      "loss": 0.1818,
+      "step": 1038
+    },
+    {
+      "epoch": 4.206477732793522,
+      "grad_norm": 0.2431137899771681,
+      "learning_rate": 2.9939062256704997e-06,
+      "loss": 0.1881,
+      "step": 1039
+    },
+    {
+      "epoch": 4.2105263157894735,
+      "grad_norm": 0.25575922090534586,
+      "learning_rate": 2.964210239224696e-06,
+      "loss": 0.1806,
+      "step": 1040
+    },
+    {
+      "epoch": 4.2145748987854255,
+      "grad_norm": 0.2461434223936982,
+      "learning_rate": 2.934650470499869e-06,
+      "loss": 0.1967,
+      "step": 1041
+    },
+    {
+      "epoch": 4.218623481781377,
+      "grad_norm": 0.24008323684276747,
+      "learning_rate": 2.905227155855128e-06,
+      "loss": 0.1964,
+      "step": 1042
+    },
+    {
+      "epoch": 4.222672064777328,
+      "grad_norm": 0.24448690784349453,
+      "learning_rate": 2.875940530558519e-06,
+      "loss": 0.1969,
+      "step": 1043
+    },
+    {
+      "epoch": 4.22672064777328,
+      "grad_norm": 0.236517667384313,
+      "learning_rate": 2.8467908287851154e-06,
+      "loss": 0.2003,
+      "step": 1044
+    },
+    {
+      "epoch": 4.230769230769231,
+      "grad_norm": 0.2490580354223857,
+      "learning_rate": 2.8177782836151603e-06,
+      "loss": 0.1865,
+      "step": 1045
+    },
+    {
+      "epoch": 4.234817813765182,
+      "grad_norm": 0.2471302908018022,
+      "learning_rate": 2.7889031270321874e-06,
+      "loss": 0.1857,
+      "step": 1046
+    },
+    {
+      "epoch": 4.238866396761134,
+      "grad_norm": 0.2552026919338716,
+      "learning_rate": 2.760165589921184e-06,
+      "loss": 0.1901,
+      "step": 1047
+    },
+    {
+      "epoch": 4.242914979757085,
+      "grad_norm": 0.24421187035984113,
+      "learning_rate": 2.7315659020667305e-06,
+      "loss": 0.1911,
+      "step": 1048
+    },
+    {
+      "epoch": 4.246963562753036,
+      "grad_norm": 0.24617816751724433,
+      "learning_rate": 2.703104292151164e-06,
+      "loss": 0.1755,
+      "step": 1049
+    },
+    {
+      "epoch": 4.251012145748988,
+      "grad_norm": 0.23844352776938524,
+      "learning_rate": 2.6747809877527607e-06,
+      "loss": 0.1793,
+      "step": 1050
+    },
+    {
+      "epoch": 4.255060728744939,
+      "grad_norm": 0.24605592369729687,
+      "learning_rate": 2.6465962153439082e-06,
+      "loss": 0.2008,
+      "step": 1051
+    },
+    {
+      "epoch": 4.2591093117408905,
+      "grad_norm": 0.244529004356487,
+      "learning_rate": 2.618550200289303e-06,
+      "loss": 0.1922,
+      "step": 1052
+    },
+    {
+      "epoch": 4.2631578947368425,
+      "grad_norm": 0.24169118119595723,
+      "learning_rate": 2.59064316684412e-06,
+      "loss": 0.1847,
+      "step": 1053
+    },
+    {
+      "epoch": 4.267206477732794,
+      "grad_norm": 0.24227457649663092,
+      "learning_rate": 2.562875338152264e-06,
+      "loss": 0.1954,
+      "step": 1054
+    },
+    {
+      "epoch": 4.271255060728745,
+      "grad_norm": 0.24982433918777802,
+      "learning_rate": 2.5352469362445552e-06,
+      "loss": 0.213,
+      "step": 1055
+    },
+    {
+      "epoch": 4.275303643724697,
+      "grad_norm": 0.23286092289975927,
+      "learning_rate": 2.5077581820369523e-06,
+      "loss": 0.1824,
+      "step": 1056
+    },
+    {
+      "epoch": 4.279352226720648,
+      "grad_norm": 0.24352916070278605,
+      "learning_rate": 2.480409295328803e-06,
+      "loss": 0.2037,
+      "step": 1057
+    },
+    {
+      "epoch": 4.283400809716599,
+      "grad_norm": 0.24094886239378338,
+      "learning_rate": 2.45320049480108e-06,
+      "loss": 0.2088,
+      "step": 1058
+    },
+    {
+      "epoch": 4.287449392712551,
+      "grad_norm": 0.25546612274256775,
+      "learning_rate": 2.4261319980146293e-06,
+      "loss": 0.1944,
+      "step": 1059
+    },
+    {
+      "epoch": 4.291497975708502,
+      "grad_norm": 0.24112972791659443,
+      "learning_rate": 2.399204021408421e-06,
+      "loss": 0.208,
+      "step": 1060
+    },
+    {
+      "epoch": 4.295546558704453,
+      "grad_norm": 0.2513914859782865,
+      "learning_rate": 2.3724167802978395e-06,
+      "loss": 0.1904,
+      "step": 1061
+    },
+    {
+      "epoch": 4.299595141700405,
+      "grad_norm": 0.2415276992879767,
+      "learning_rate": 2.3457704888729625e-06,
+      "loss": 0.2011,
+      "step": 1062
+    },
+    {
+      "epoch": 4.303643724696356,
+      "grad_norm": 0.24637368366858228,
+      "learning_rate": 2.3192653601968162e-06,
+      "loss": 0.1983,
+      "step": 1063
+    },
+    {
+      "epoch": 4.3076923076923075,
+      "grad_norm": 0.24906663794292266,
+      "learning_rate": 2.2929016062037057e-06,
+      "loss": 0.1978,
+      "step": 1064
+    },
+    {
+      "epoch": 4.3117408906882595,
+      "grad_norm": 0.24839160185694195,
+      "learning_rate": 2.2666794376975055e-06,
+      "loss": 0.201,
+      "step": 1065
+    },
+    {
+      "epoch": 4.315789473684211,
+      "grad_norm": 0.23685054337568634,
+      "learning_rate": 2.24059906434998e-06,
+      "loss": 0.1855,
+      "step": 1066
+    },
+    {
+      "epoch": 4.319838056680162,
+      "grad_norm": 0.2430243992793424,
+      "learning_rate": 2.214660694699091e-06,
+      "loss": 0.1979,
+      "step": 1067
+    },
+    {
+      "epoch": 4.323886639676114,
+      "grad_norm": 0.24175053527093407,
+      "learning_rate": 2.188864536147355e-06,
+      "loss": 0.1883,
+      "step": 1068
+    },
+    {
+      "epoch": 4.327935222672065,
+      "grad_norm": 0.24309524766020105,
+      "learning_rate": 2.1632107949601757e-06,
+      "loss": 0.2064,
+      "step": 1069
+    },
+    {
+      "epoch": 4.331983805668016,
+      "grad_norm": 0.2446466749903697,
+      "learning_rate": 2.137699676264171e-06,
+      "loss": 0.1993,
+      "step": 1070
+    },
+    {
+      "epoch": 4.336032388663968,
+      "grad_norm": 0.24261738293919724,
+      "learning_rate": 2.1123313840455738e-06,
+      "loss": 0.1831,
+      "step": 1071
+    },
+    {
+      "epoch": 4.340080971659919,
+      "grad_norm": 0.234684337005504,
+      "learning_rate": 2.08710612114857e-06,
+      "loss": 0.2025,
+      "step": 1072
+    },
+    {
+      "epoch": 4.34412955465587,
+      "grad_norm": 0.2416401712953408,
+      "learning_rate": 2.0620240892736953e-06,
+      "loss": 0.2061,
+      "step": 1073
+    },
+    {
+      "epoch": 4.348178137651822,
+      "grad_norm": 0.24796600600138194,
+      "learning_rate": 2.0370854889762e-06,
+      "loss": 0.2058,
+      "step": 1074
+    },
+    {
+      "epoch": 4.352226720647773,
+      "grad_norm": 0.23834625515082974,
+      "learning_rate": 2.0122905196644726e-06,
+      "loss": 0.1917,
+      "step": 1075
+    },
+    {
+      "epoch": 4.3562753036437245,
+      "grad_norm": 0.2468130785255841,
+      "learning_rate": 1.987639379598427e-06,
+      "loss": 0.192,
+      "step": 1076
+    },
+    {
+      "epoch": 4.3603238866396765,
+      "grad_norm": 0.23990577614127315,
+      "learning_rate": 1.963132265887919e-06,
+      "loss": 0.2131,
+      "step": 1077
+    },
+    {
+      "epoch": 4.364372469635628,
+      "grad_norm": 0.2471311844071165,
+      "learning_rate": 1.9387693744911784e-06,
+      "loss": 0.2053,
+      "step": 1078
+    },
+    {
+      "epoch": 4.368421052631579,
+      "grad_norm": 0.2466938556914585,
+      "learning_rate": 1.9145509002132368e-06,
+      "loss": 0.2013,
+      "step": 1079
+    },
+    {
+      "epoch": 4.372469635627531,
+      "grad_norm": 0.2356346955313169,
+      "learning_rate": 1.8904770367043723e-06,
+      "loss": 0.1894,
+      "step": 1080
+    },
+    {
+      "epoch": 4.376518218623482,
+      "grad_norm": 0.24229471950563136,
+      "learning_rate": 1.8665479764585504e-06,
+      "loss": 0.1852,
+      "step": 1081
+    },
+    {
+      "epoch": 4.380566801619433,
+      "grad_norm": 0.24068844178762783,
+      "learning_rate": 1.8427639108118977e-06,
+      "loss": 0.1876,
+      "step": 1082
+    },
+    {
+      "epoch": 4.384615384615385,
+      "grad_norm": 0.24158730148277363,
+      "learning_rate": 1.8191250299411755e-06,
+      "loss": 0.1871,
+      "step": 1083
+    },
+    {
+      "epoch": 4.388663967611336,
+      "grad_norm": 0.2428671113864186,
+      "learning_rate": 1.7956315228622446e-06,
+      "loss": 0.1947,
+      "step": 1084
+    },
+    {
+      "epoch": 4.392712550607287,
+      "grad_norm": 0.23552650061021232,
+      "learning_rate": 1.7722835774285553e-06,
+      "loss": 0.1964,
+      "step": 1085
+    },
+    {
+      "epoch": 4.396761133603239,
+      "grad_norm": 0.23301342030034616,
+      "learning_rate": 1.7490813803296624e-06,
+      "loss": 0.1883,
+      "step": 1086
+    },
+    {
+      "epoch": 4.40080971659919,
+      "grad_norm": 0.23338941634423585,
+      "learning_rate": 1.726025117089718e-06,
+      "loss": 0.2002,
+      "step": 1087
+    },
+    {
+      "epoch": 4.4048582995951415,
+      "grad_norm": 0.23045763323644639,
+      "learning_rate": 1.70311497206598e-06,
+      "loss": 0.1707,
+      "step": 1088
+    },
+    {
+      "epoch": 4.4089068825910935,
+      "grad_norm": 0.24288496124232925,
+      "learning_rate": 1.6803511284473638e-06,
+      "loss": 0.1969,
+      "step": 1089
+    },
+    {
+      "epoch": 4.412955465587045,
+      "grad_norm": 0.23570254481055308,
+      "learning_rate": 1.6577337682529538e-06,
+      "loss": 0.1915,
+      "step": 1090
+    },
+    {
+      "epoch": 4.417004048582996,
+      "grad_norm": 0.23668513832885174,
+      "learning_rate": 1.6352630723305617e-06,
+      "loss": 0.1985,
+      "step": 1091
+    },
+    {
+      "epoch": 4.421052631578947,
+      "grad_norm": 0.242603644877871,
+      "learning_rate": 1.612939220355272e-06,
+      "loss": 0.1947,
+      "step": 1092
+    },
+    {
+      "epoch": 4.425101214574899,
+      "grad_norm": 0.24565676340229292,
+      "learning_rate": 1.5907623908280112e-06,
+      "loss": 0.2073,
+      "step": 1093
+    },
+    {
+      "epoch": 4.42914979757085,
+      "grad_norm": 0.24723463079214028,
+      "learning_rate": 1.568732761074121e-06,
+      "loss": 0.1794,
+      "step": 1094
+    },
+    {
+      "epoch": 4.433198380566802,
+      "grad_norm": 0.2537096418767233,
+      "learning_rate": 1.5468505072419237e-06,
+      "loss": 0.2142,
+      "step": 1095
+    },
+    {
+      "epoch": 4.437246963562753,
+      "grad_norm": 0.23525290173479504,
+      "learning_rate": 1.5251158043013448e-06,
+      "loss": 0.194,
+      "step": 1096
+    },
+    {
+      "epoch": 4.441295546558704,
+      "grad_norm": 0.2382218130987789,
+      "learning_rate": 1.5035288260424885e-06,
+      "loss": 0.1905,
+      "step": 1097
+    },
+    {
+      "epoch": 4.445344129554655,
+      "grad_norm": 0.2415269513533817,
+      "learning_rate": 1.4820897450742623e-06,
+      "loss": 0.1916,
+      "step": 1098
+    },
+    {
+      "epoch": 4.449392712550607,
+      "grad_norm": 0.2399542378201055,
+      "learning_rate": 1.4607987328229767e-06,
+      "loss": 0.1962,
+      "step": 1099
+    },
+    {
+      "epoch": 4.4534412955465585,
+      "grad_norm": 0.2293634693986604,
+      "learning_rate": 1.4396559595310056e-06,
+      "loss": 0.2041,
+      "step": 1100
+    },
+    {
+      "epoch": 4.4574898785425106,
+      "grad_norm": 0.2378255134601964,
+      "learning_rate": 1.4186615942553973e-06,
+      "loss": 0.2047,
+      "step": 1101
+    },
+    {
+      "epoch": 4.461538461538462,
+      "grad_norm": 0.23730687976152826,
+      "learning_rate": 1.3978158048665424e-06,
+      "loss": 0.1933,
+      "step": 1102
+    },
+    {
+      "epoch": 4.465587044534413,
+      "grad_norm": 0.23780726323659296,
+      "learning_rate": 1.3771187580468115e-06,
+      "loss": 0.1934,
+      "step": 1103
+    },
+    {
+      "epoch": 4.469635627530364,
+      "grad_norm": 0.24139170518841607,
+      "learning_rate": 1.3565706192892392e-06,
+      "loss": 0.1966,
+      "step": 1104
+    },
+    {
+      "epoch": 4.473684210526316,
+      "grad_norm": 0.23436080866516998,
+      "learning_rate": 1.3361715528961993e-06,
+      "loss": 0.183,
+      "step": 1105
+    },
+    {
+      "epoch": 4.477732793522267,
+      "grad_norm": 0.24477977689648922,
+      "learning_rate": 1.3159217219780817e-06,
+      "loss": 0.1948,
+      "step": 1106
+    },
+    {
+      "epoch": 4.481781376518219,
+      "grad_norm": 0.2438553700360888,
+      "learning_rate": 1.2958212884519949e-06,
+      "loss": 0.1889,
+      "step": 1107
+    },
+    {
+      "epoch": 4.48582995951417,
+      "grad_norm": 0.24443096403431305,
+      "learning_rate": 1.2758704130404675e-06,
+      "loss": 0.1973,
+      "step": 1108
+    },
+    {
+      "epoch": 4.489878542510121,
+      "grad_norm": 0.2418227043885346,
+      "learning_rate": 1.2560692552701738e-06,
+      "loss": 0.2003,
+      "step": 1109
+    },
+    {
+      "epoch": 4.493927125506072,
+      "grad_norm": 0.24284800450568603,
+      "learning_rate": 1.2364179734706382e-06,
+      "loss": 0.1947,
+      "step": 1110
+    },
+    {
+      "epoch": 4.497975708502024,
+      "grad_norm": 0.23551116586926432,
+      "learning_rate": 1.2169167247729852e-06,
+      "loss": 0.2033,
+      "step": 1111
+    },
+    {
+      "epoch": 4.502024291497976,
+      "grad_norm": 0.23610563881039692,
+      "learning_rate": 1.1975656651086864e-06,
+      "loss": 0.1871,
+      "step": 1112
+    },
+    {
+      "epoch": 4.506072874493928,
+      "grad_norm": 0.24101485302920117,
+      "learning_rate": 1.1783649492082926e-06,
+      "loss": 0.1945,
+      "step": 1113
+    },
+    {
+      "epoch": 4.510121457489879,
+      "grad_norm": 0.2383851852174509,
+      "learning_rate": 1.1593147306002183e-06,
+      "loss": 0.1918,
+      "step": 1114
+    },
+    {
+      "epoch": 4.51417004048583,
+      "grad_norm": 0.2334034912974562,
+      "learning_rate": 1.140415161609505e-06,
+      "loss": 0.1865,
+      "step": 1115
+    },
+    {
+      "epoch": 4.518218623481781,
+      "grad_norm": 0.2482796824730159,
+      "learning_rate": 1.121666393356604e-06,
+      "loss": 0.2006,
+      "step": 1116
+    },
+    {
+      "epoch": 4.522267206477733,
+      "grad_norm": 0.23574457940389645,
+      "learning_rate": 1.1030685757561632e-06,
+      "loss": 0.1887,
+      "step": 1117
+    },
+    {
+      "epoch": 4.526315789473684,
+      "grad_norm": 0.2522172651249002,
+      "learning_rate": 1.0846218575158373e-06,
+      "loss": 0.2018,
+      "step": 1118
+    },
+    {
+      "epoch": 4.530364372469636,
+      "grad_norm": 0.24419152520381965,
+      "learning_rate": 1.066326386135097e-06,
+      "loss": 0.2031,
+      "step": 1119
+    },
+    {
+      "epoch": 4.534412955465587,
+      "grad_norm": 0.23428058080672773,
+      "learning_rate": 1.0481823079040398e-06,
+      "loss": 0.2034,
+      "step": 1120
+    },
+    {
+      "epoch": 4.538461538461538,
+      "grad_norm": 0.23512102136954652,
+      "learning_rate": 1.0301897679022344e-06,
+      "loss": 0.1817,
+      "step": 1121
+    },
+    {
+      "epoch": 4.5425101214574894,
+      "grad_norm": 0.2429060454210732,
+      "learning_rate": 1.0123489099975491e-06,
+      "loss": 0.2041,
+      "step": 1122
+    },
+    {
+      "epoch": 4.5465587044534415,
+      "grad_norm": 0.23706143852861034,
+      "learning_rate": 9.9465987684501e-07,
+      "loss": 0.1935,
+      "step": 1123
+    },
+    {
+      "epoch": 4.550607287449393,
+      "grad_norm": 0.23259619928658143,
+      "learning_rate": 9.771228098856534e-07,
+      "loss": 0.2138,
+      "step": 1124
+    },
+    {
+      "epoch": 4.554655870445345,
+      "grad_norm": 0.24676814130179894,
+      "learning_rate": 9.597378493454013e-07,
+      "loss": 0.1838,
+      "step": 1125
+    },
+    {
+      "epoch": 4.558704453441296,
+      "grad_norm": 0.2355324847119727,
+      "learning_rate": 9.425051342339353e-07,
+      "loss": 0.2099,
+      "step": 1126
+    },
+    {
+      "epoch": 4.562753036437247,
+      "grad_norm": 0.23960498740319677,
+      "learning_rate": 9.254248023435864e-07,
+      "loss": 0.2178,
+      "step": 1127
+    },
+    {
+      "epoch": 4.566801619433198,
+      "grad_norm": 0.23474366337066746,
+      "learning_rate": 9.084969902482288e-07,
+      "loss": 0.2017,
+      "step": 1128
+    },
+    {
+      "epoch": 4.57085020242915,
+      "grad_norm": 0.2384409329729265,
+      "learning_rate": 8.917218333021993e-07,
+      "loss": 0.1876,
+      "step": 1129
+    },
+    {
+      "epoch": 4.574898785425101,
+      "grad_norm": 0.2360410469059819,
+      "learning_rate": 8.750994656392109e-07,
+      "loss": 0.18,
+      "step": 1130
+    },
+    {
+      "epoch": 4.578947368421053,
+      "grad_norm": 0.23889281296823006,
+      "learning_rate": 8.58630020171265e-07,
+      "loss": 0.1981,
+      "step": 1131
+    },
+    {
+      "epoch": 4.582995951417004,
+      "grad_norm": 0.22931180861631728,
+      "learning_rate": 8.423136285876099e-07,
+      "loss": 0.2115,
+      "step": 1132
+    },
+    {
+      "epoch": 4.587044534412955,
+      "grad_norm": 0.23866959097572069,
+      "learning_rate": 8.26150421353682e-07,
+      "loss": 0.2113,
+      "step": 1133
+    },
+    {
+      "epoch": 4.5910931174089065,
+      "grad_norm": 0.23306921278373494,
+      "learning_rate": 8.101405277100549e-07,
+      "loss": 0.198,
+      "step": 1134
+    },
+    {
+      "epoch": 4.5951417004048585,
+      "grad_norm": 0.23176037693255347,
+      "learning_rate": 7.942840756714077e-07,
+      "loss": 0.1963,
+      "step": 1135
+    },
+    {
+      "epoch": 4.59919028340081,
+      "grad_norm": 0.2345597871132039,
+      "learning_rate": 7.785811920255093e-07,
+      "loss": 0.1972,
+      "step": 1136
+    },
+    {
+      "epoch": 4.603238866396762,
+      "grad_norm": 0.24507659315401956,
+      "learning_rate": 7.630320023322024e-07,
+      "loss": 0.1977,
+      "step": 1137
+    },
+    {
+      "epoch": 4.607287449392713,
+      "grad_norm": 0.23267276342724605,
+      "learning_rate": 7.476366309223903e-07,
+      "loss": 0.2101,
+      "step": 1138
+    },
+    {
+      "epoch": 4.611336032388664,
+      "grad_norm": 0.24453517057729873,
+      "learning_rate": 7.323952008970537e-07,
+      "loss": 0.1945,
+      "step": 1139
+    },
+    {
+      "epoch": 4.615384615384615,
+      "grad_norm": 0.2349157233004698,
+      "learning_rate": 7.173078341262618e-07,
+      "loss": 0.1892,
+      "step": 1140
+    },
+    {
+      "epoch": 4.619433198380567,
+      "grad_norm": 0.23502003999944335,
+      "learning_rate": 7.023746512481989e-07,
+      "loss": 0.1992,
+      "step": 1141
+    },
+    {
+      "epoch": 4.623481781376518,
+      "grad_norm": 0.23095882969876586,
+      "learning_rate": 6.875957716681902e-07,
+      "loss": 0.1815,
+      "step": 1142
+    },
+    {
+      "epoch": 4.62753036437247,
+      "grad_norm": 0.23798669776702192,
+      "learning_rate": 6.729713135577687e-07,
+      "loss": 0.181,
+      "step": 1143
+    },
+    {
+      "epoch": 4.631578947368421,
+      "grad_norm": 0.23515420900241807,
+      "learning_rate": 6.585013938537143e-07,
+      "loss": 0.2059,
+      "step": 1144
+    },
+    {
+      "epoch": 4.635627530364372,
+      "grad_norm": 0.23030677711026926,
+      "learning_rate": 6.441861282571115e-07,
+      "loss": 0.19,
+      "step": 1145
+    },
+    {
+      "epoch": 4.6396761133603235,
+      "grad_norm": 0.23489800825809704,
+      "learning_rate": 6.300256312324471e-07,
+      "loss": 0.193,
+      "step": 1146
+    },
+    {
+      "epoch": 4.6437246963562755,
+      "grad_norm": 0.2340301110065373,
+      "learning_rate": 6.160200160066776e-07,
+      "loss": 0.1891,
+      "step": 1147
+    },
+    {
+      "epoch": 4.647773279352227,
+      "grad_norm": 0.2470025436076655,
+      "learning_rate": 6.021693945683327e-07,
+      "loss": 0.1882,
+      "step": 1148
+    },
+    {
+      "epoch": 4.651821862348179,
+      "grad_norm": 0.23943212910792133,
+      "learning_rate": 5.884738776666044e-07,
+      "loss": 0.1982,
+      "step": 1149
+    },
+    {
+      "epoch": 4.65587044534413,
+      "grad_norm": 0.23822001533456258,
+      "learning_rate": 5.749335748104878e-07,
+      "loss": 0.1866,
+      "step": 1150
+    },
+    {
+      "epoch": 4.659919028340081,
+      "grad_norm": 0.24073650156919707,
+      "learning_rate": 5.615485942678778e-07,
+      "loss": 0.2002,
+      "step": 1151
+    },
+    {
+      "epoch": 4.663967611336032,
+      "grad_norm": 0.24001034727948428,
+      "learning_rate": 5.483190430647289e-07,
+      "loss": 0.2016,
+      "step": 1152
+    },
+    {
+      "epoch": 4.668016194331984,
+      "grad_norm": 0.23365624794520887,
+      "learning_rate": 5.352450269841747e-07,
+      "loss": 0.2023,
+      "step": 1153
+    },
+    {
+      "epoch": 4.672064777327935,
+      "grad_norm": 0.23843604323959014,
+      "learning_rate": 5.22326650565701e-07,
+      "loss": 0.1917,
+      "step": 1154
+    },
+    {
+      "epoch": 4.676113360323887,
+      "grad_norm": 0.2355389887105177,
+      "learning_rate": 5.095640171043048e-07,
+      "loss": 0.2102,
+      "step": 1155
+    },
+    {
+      "epoch": 4.680161943319838,
+      "grad_norm": 0.23989988911732749,
+      "learning_rate": 4.969572286496594e-07,
+      "loss": 0.1849,
+      "step": 1156
+    },
+    {
+      "epoch": 4.684210526315789,
+      "grad_norm": 0.2327617783325008,
+      "learning_rate": 4.845063860053079e-07,
+      "loss": 0.2062,
+      "step": 1157
+    },
+    {
+      "epoch": 4.6882591093117405,
+      "grad_norm": 0.2317732696390293,
+      "learning_rate": 4.7221158872785997e-07,
+      "loss": 0.1941,
+      "step": 1158
+    },
+    {
+      "epoch": 4.6923076923076925,
+      "grad_norm": 0.2400355475033211,
+      "learning_rate": 4.600729351261857e-07,
+      "loss": 0.1744,
+      "step": 1159
+    },
+    {
+      "epoch": 4.696356275303644,
+      "grad_norm": 0.24424548601689222,
+      "learning_rate": 4.4809052226063355e-07,
+      "loss": 0.1917,
+      "step": 1160
+    },
+    {
+      "epoch": 4.700404858299595,
+      "grad_norm": 0.2529837469000429,
+      "learning_rate": 4.3626444594226046e-07,
+      "loss": 0.185,
+      "step": 1161
+    },
+    {
+      "epoch": 4.704453441295547,
+      "grad_norm": 0.2339542343118446,
+      "learning_rate": 4.245948007320588e-07,
+      "loss": 0.1923,
+      "step": 1162
+    },
+    {
+      "epoch": 4.708502024291498,
+      "grad_norm": 0.23977833865721118,
+      "learning_rate": 4.130816799401971e-07,
+      "loss": 0.1752,
+      "step": 1163
+    },
+    {
+      "epoch": 4.712550607287449,
+      "grad_norm": 0.24566962274277024,
+      "learning_rate": 4.017251756252827e-07,
+      "loss": 0.18,
+      "step": 1164
+    },
+    {
+      "epoch": 4.716599190283401,
+      "grad_norm": 0.2386843476296561,
+      "learning_rate": 3.9052537859362473e-07,
+      "loss": 0.201,
+      "step": 1165
+    },
+    {
+      "epoch": 4.720647773279352,
+      "grad_norm": 0.23954521879954765,
+      "learning_rate": 3.7948237839849687e-07,
+      "loss": 0.1935,
+      "step": 1166
+    },
+    {
+      "epoch": 4.724696356275303,
+      "grad_norm": 0.23780785014303082,
+      "learning_rate": 3.6859626333943797e-07,
+      "loss": 0.2205,
+      "step": 1167
+    },
+    {
+      "epoch": 4.728744939271255,
+      "grad_norm": 0.23445984703754938,
+      "learning_rate": 3.578671204615347e-07,
+      "loss": 0.1815,
+      "step": 1168
+    },
+    {
+      "epoch": 4.732793522267206,
+      "grad_norm": 0.23802277073216463,
+      "learning_rate": 3.472950355547267e-07,
+      "loss": 0.1941,
+      "step": 1169
+    },
+    {
+      "epoch": 4.7368421052631575,
+      "grad_norm": 0.24562977477297276,
+      "learning_rate": 3.3688009315312244e-07,
+      "loss": 0.171,
+      "step": 1170
+    },
+    {
+      "epoch": 4.7408906882591095,
+      "grad_norm": 0.24594685311973935,
+      "learning_rate": 3.266223765343224e-07,
+      "loss": 0.1844,
+      "step": 1171
+    },
+    {
+      "epoch": 4.744939271255061,
+      "grad_norm": 0.24154732427551692,
+      "learning_rate": 3.16521967718757e-07,
+      "loss": 0.1839,
+      "step": 1172
+    },
+    {
+      "epoch": 4.748987854251012,
+      "grad_norm": 0.2439425116491534,
+      "learning_rate": 3.0657894746902726e-07,
+      "loss": 0.2142,
+      "step": 1173
+    },
+    {
+      "epoch": 4.753036437246964,
+      "grad_norm": 0.22971352419542995,
+      "learning_rate": 2.9679339528925654e-07,
+      "loss": 0.1973,
+      "step": 1174
+    },
+    {
+      "epoch": 4.757085020242915,
+      "grad_norm": 0.23734033558760068,
+      "learning_rate": 2.871653894244597e-07,
+      "loss": 0.2043,
+      "step": 1175
+    },
+    {
+      "epoch": 4.761133603238866,
+      "grad_norm": 0.2445339766306351,
+      "learning_rate": 2.7769500685991266e-07,
+      "loss": 0.1803,
+      "step": 1176
+    },
+    {
+      "epoch": 4.765182186234818,
+      "grad_norm": 0.22825589371230942,
+      "learning_rate": 2.683823233205485e-07,
+      "loss": 0.1886,
+      "step": 1177
+    },
+    {
+      "epoch": 4.769230769230769,
+      "grad_norm": 0.23258379192829648,
+      "learning_rate": 2.592274132703287e-07,
+      "loss": 0.1882,
+      "step": 1178
+    },
+    {
+      "epoch": 4.77327935222672,
+      "grad_norm": 0.24434342372447093,
+      "learning_rate": 2.502303499116754e-07,
+      "loss": 0.1897,
+      "step": 1179
+    },
+    {
+      "epoch": 4.777327935222672,
+      "grad_norm": 0.2431609263739724,
+      "learning_rate": 2.413912051848666e-07,
+      "loss": 0.1843,
+      "step": 1180
+    },
+    {
+      "epoch": 4.781376518218623,
+      "grad_norm": 0.23075455326319744,
+      "learning_rate": 2.3271004976746613e-07,
+      "loss": 0.2032,
+      "step": 1181
+    },
+    {
+      "epoch": 4.7854251012145745,
+      "grad_norm": 0.2339016274921721,
+      "learning_rate": 2.2418695307376393e-07,
+      "loss": 0.2101,
+      "step": 1182
+    },
+    {
+      "epoch": 4.7894736842105265,
+      "grad_norm": 0.2380861929471764,
+      "learning_rate": 2.1582198325420968e-07,
+      "loss": 0.2001,
+      "step": 1183
+    },
+    {
+      "epoch": 4.793522267206478,
+      "grad_norm": 0.2413892813777306,
+      "learning_rate": 2.0761520719488447e-07,
+      "loss": 0.1916,
+      "step": 1184
+    },
+    {
+      "epoch": 4.797570850202429,
+      "grad_norm": 0.24212725326105963,
+      "learning_rate": 1.9956669051694798e-07,
+      "loss": 0.211,
+      "step": 1185
+    },
+    {
+      "epoch": 4.801619433198381,
+      "grad_norm": 0.23373268501438738,
+      "learning_rate": 1.916764975761254e-07,
+      "loss": 0.2175,
+      "step": 1186
+    },
+    {
+      "epoch": 4.805668016194332,
+      "grad_norm": 0.24256240789200972,
+      "learning_rate": 1.839446914621923e-07,
+      "loss": 0.2003,
+      "step": 1187
+    },
+    {
+      "epoch": 4.809716599190283,
+      "grad_norm": 0.23760816156685916,
+      "learning_rate": 1.7637133399845962e-07,
+      "loss": 0.1859,
+      "step": 1188
+    },
+    {
+      "epoch": 4.813765182186235,
+      "grad_norm": 0.22812430737322542,
+      "learning_rate": 1.6895648574129397e-07,
+      "loss": 0.2173,
+      "step": 1189
+    },
+    {
+      "epoch": 4.817813765182186,
+      "grad_norm": 0.23585902567235847,
+      "learning_rate": 1.6170020597962465e-07,
+      "loss": 0.2279,
+      "step": 1190
+    },
+    {
+      "epoch": 4.821862348178137,
+      "grad_norm": 0.24749443556418266,
+      "learning_rate": 1.5460255273447078e-07,
+      "loss": 0.1985,
+      "step": 1191
+    },
+    {
+      "epoch": 4.825910931174089,
+      "grad_norm": 0.23024250744995206,
+      "learning_rate": 1.4766358275847715e-07,
+      "loss": 0.1876,
+      "step": 1192
+    },
+    {
+      "epoch": 4.82995951417004,
+      "grad_norm": 0.24108919208438648,
+      "learning_rate": 1.4088335153546573e-07,
+      "loss": 0.1993,
+      "step": 1193
+    },
+    {
+      "epoch": 4.834008097165992,
+      "grad_norm": 0.23877193395157584,
+      "learning_rate": 1.3426191327998496e-07,
+      "loss": 0.185,
+      "step": 1194
+    },
+    {
+      "epoch": 4.838056680161944,
+      "grad_norm": 0.23769586196452697,
+      "learning_rate": 1.277993209368744e-07,
+      "loss": 0.1951,
+      "step": 1195
+    },
+    {
+      "epoch": 4.842105263157895,
+      "grad_norm": 0.2422648088024461,
+      "learning_rate": 1.214956261808542e-07,
+      "loss": 0.2059,
+      "step": 1196
+    },
+    {
+      "epoch": 4.846153846153846,
+      "grad_norm": 0.22857546045739716,
+      "learning_rate": 1.1535087941609846e-07,
+      "loss": 0.2068,
+      "step": 1197
+    },
+    {
+      "epoch": 4.850202429149798,
+      "grad_norm": 0.23573177823650449,
+      "learning_rate": 1.0936512977583802e-07,
+      "loss": 0.1852,
+      "step": 1198
+    },
+    {
+      "epoch": 4.854251012145749,
+      "grad_norm": 0.23738754500456075,
+      "learning_rate": 1.0353842512196732e-07,
+      "loss": 0.185,
+      "step": 1199
+    },
+    {
+      "epoch": 4.8582995951417,
+      "grad_norm": 0.2312247888934274,
+      "learning_rate": 9.787081204466253e-08,
+      "loss": 0.1973,
+      "step": 1200
+    },
+    {
+      "epoch": 4.862348178137652,
+      "grad_norm": 0.23621776917922513,
+      "learning_rate": 9.236233586200627e-08,
+      "loss": 0.195,
+      "step": 1201
+    },
+    {
+      "epoch": 4.866396761133603,
+      "grad_norm": 0.22971855978189543,
+      "learning_rate": 8.70130406196279e-08,
+      "loss": 0.2045,
+      "step": 1202
+    },
+    {
+      "epoch": 4.870445344129554,
+      "grad_norm": 0.23029624511964433,
+      "learning_rate": 8.182296909035047e-08,
+      "loss": 0.21,
+      "step": 1203
+    },
+    {
+      "epoch": 4.874493927125506,
+      "grad_norm": 0.24339899127991382,
+      "learning_rate": 7.679216277384439e-08,
+      "loss": 0.197,
+      "step": 1204
+    },
+    {
+      "epoch": 4.8785425101214575,
+      "grad_norm": 0.2338673757246494,
+      "learning_rate": 7.19206618963053e-08,
+      "loss": 0.187,
+      "step": 1205
+    },
+    {
+      "epoch": 4.882591093117409,
+      "grad_norm": 0.2367769072214976,
+      "learning_rate": 6.720850541012347e-08,
+      "loss": 0.1896,
+      "step": 1206
+    },
+    {
+      "epoch": 4.886639676113361,
+      "grad_norm": 0.24488826939798403,
+      "learning_rate": 6.265573099357714e-08,
+      "loss": 0.1932,
+      "step": 1207
+    },
+    {
+      "epoch": 4.890688259109312,
+      "grad_norm": 0.2363706776052818,
+      "learning_rate": 5.826237505053067e-08,
+      "loss": 0.1972,
+      "step": 1208
+    },
+    {
+      "epoch": 4.894736842105263,
+      "grad_norm": 0.24020213550624214,
+      "learning_rate": 5.4028472710137e-08,
+      "loss": 0.1853,
+      "step": 1209
+    },
+    {
+      "epoch": 4.898785425101215,
+      "grad_norm": 0.24267226202515355,
+      "learning_rate": 4.9954057826571146e-08,
+      "loss": 0.2002,
+      "step": 1210
+    },
+    {
+      "epoch": 4.902834008097166,
+      "grad_norm": 0.23838144278126647,
+      "learning_rate": 4.6039162978743776e-08,
+      "loss": 0.2052,
+      "step": 1211
+    },
+    {
+      "epoch": 4.906882591093117,
+      "grad_norm": 0.2350391581585882,
+      "learning_rate": 4.228381947004812e-08,
+      "loss": 0.1822,
+      "step": 1212
+    },
+    {
+      "epoch": 4.910931174089069,
+      "grad_norm": 0.23216186801774458,
+      "learning_rate": 3.868805732811343e-08,
+      "loss": 0.2094,
+      "step": 1213
+    },
+    {
+      "epoch": 4.91497975708502,
+      "grad_norm": 0.22834352409135683,
+      "learning_rate": 3.525190530455636e-08,
+      "loss": 0.2004,
+      "step": 1214
+    },
+    {
+      "epoch": 4.919028340080971,
+      "grad_norm": 0.22868330764099346,
+      "learning_rate": 3.197539087475887e-08,
+      "loss": 0.209,
+      "step": 1215
+    },
+    {
+      "epoch": 4.923076923076923,
+      "grad_norm": 0.2378079547163325,
+      "learning_rate": 2.8858540237641786e-08,
+      "loss": 0.1853,
+      "step": 1216
+    },
+    {
+      "epoch": 4.9271255060728745,
+      "grad_norm": 0.23566850374778392,
+      "learning_rate": 2.5901378315458247e-08,
+      "loss": 0.204,
+      "step": 1217
+    },
+    {
+      "epoch": 4.931174089068826,
+      "grad_norm": 0.23419594224736853,
+      "learning_rate": 2.3103928753598347e-08,
+      "loss": 0.2105,
+      "step": 1218
+    },
+    {
+      "epoch": 4.935222672064778,
+      "grad_norm": 0.2482634803192749,
+      "learning_rate": 2.0466213920393718e-08,
+      "loss": 0.1979,
+      "step": 1219
+    },
+    {
+      "epoch": 4.939271255060729,
+      "grad_norm": 0.23409249528111814,
+      "learning_rate": 1.7988254906942115e-08,
+      "loss": 0.1796,
+      "step": 1220
+    },
+    {
+      "epoch": 4.94331983805668,
+      "grad_norm": 0.22889145853119658,
+      "learning_rate": 1.5670071526936447e-08,
+      "loss": 0.1898,
+      "step": 1221
+    },
+    {
+      "epoch": 4.947368421052632,
+      "grad_norm": 0.23213559622161084,
+      "learning_rate": 1.3511682316509344e-08,
+      "loss": 0.201,
+      "step": 1222
+    },
+    {
+      "epoch": 4.951417004048583,
+      "grad_norm": 0.23407696235440295,
+      "learning_rate": 1.1513104534086605e-08,
+      "loss": 0.1808,
+      "step": 1223
+    },
+    {
+      "epoch": 4.955465587044534,
+      "grad_norm": 0.2378931852315451,
+      "learning_rate": 9.67435416023843e-09,
+      "loss": 0.1814,
+      "step": 1224
+    },
+    {
+      "epoch": 4.959514170040486,
+      "grad_norm": 0.2346122779840439,
+      "learning_rate": 7.995445897566178e-09,
+      "loss": 0.1969,
+      "step": 1225
+    },
+    {
+      "epoch": 4.963562753036437,
+      "grad_norm": 0.23308233587412155,
+      "learning_rate": 6.476393170578021e-09,
+      "loss": 0.2193,
+      "step": 1226
+    },
+    {
+      "epoch": 4.967611336032388,
+      "grad_norm": 0.23869925447736656,
+      "learning_rate": 5.11720812557348e-09,
+      "loss": 0.183,
+      "step": 1227
+    },
+    {
+      "epoch": 4.97165991902834,
+      "grad_norm": 0.23069990964440032,
+      "learning_rate": 3.917901630563492e-09,
+      "loss": 0.2018,
+      "step": 1228
+    },
+    {
+      "epoch": 4.9757085020242915,
+      "grad_norm": 0.23533474009059926,
+      "learning_rate": 2.878483275170485e-09,
+      "loss": 0.1792,
+      "step": 1229
+    },
+    {
+      "epoch": 4.979757085020243,
+      "grad_norm": 0.2390097200265693,
+      "learning_rate": 1.998961370555108e-09,
+      "loss": 0.1975,
+      "step": 1230
+    },
+    {
+      "epoch": 4.983805668016195,
+      "grad_norm": 0.23435713739884773,
+      "learning_rate": 1.2793429493518361e-09,
+      "loss": 0.2058,
+      "step": 1231
+    },
+    {
+      "epoch": 4.987854251012146,
+      "grad_norm": 0.23153277001829967,
+      "learning_rate": 7.196337656112384e-10,
+      "loss": 0.2002,
+      "step": 1232
+    },
+    {
+      "epoch": 4.991902834008097,
+      "grad_norm": 0.23176717417319245,
+      "learning_rate": 3.198382947511292e-10,
+      "loss": 0.207,
+      "step": 1233
+    },
+    {
+      "epoch": 4.995951417004049,
+      "grad_norm": 0.23767084377547923,
+      "learning_rate": 7.995973352770137e-11,
+      "loss": 0.1972,
+      "step": 1234
+    },
+    {
+      "epoch": 5.0,
+      "grad_norm": 0.23677320912006,
+      "learning_rate": 0.0,
+      "loss": 0.1925,
+      "step": 1235
+    },
+    {
+      "epoch": 5.0,
+      "step": 1235,
+      "total_flos": 2.6827061386668933e+18,
+      "train_loss": 0.37129853870945906,
+      "train_runtime": 25199.3425,
+      "train_samples_per_second": 6.27,
+      "train_steps_per_second": 0.049
+    }
+  ],
+  "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": 2.6827061386668933e+18,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}