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": 6.658193228444498,
+      "learning_rate": 3.2258064516129035e-07,
+      "loss": 1.0821,
+      "step": 1
+    },
+    {
+      "epoch": 0.008097165991902834,
+      "grad_norm": 6.352718267289265,
+      "learning_rate": 6.451612903225807e-07,
+      "loss": 1.0742,
+      "step": 2
+    },
+    {
+      "epoch": 0.012145748987854251,
+      "grad_norm": 6.323666997782271,
+      "learning_rate": 9.67741935483871e-07,
+      "loss": 1.061,
+      "step": 3
+    },
+    {
+      "epoch": 0.016194331983805668,
+      "grad_norm": 6.523066216367048,
+      "learning_rate": 1.2903225806451614e-06,
+      "loss": 1.068,
+      "step": 4
+    },
+    {
+      "epoch": 0.020242914979757085,
+      "grad_norm": 6.301405407361487,
+      "learning_rate": 1.6129032258064516e-06,
+      "loss": 1.0823,
+      "step": 5
+    },
+    {
+      "epoch": 0.024291497975708502,
+      "grad_norm": 5.875443557964236,
+      "learning_rate": 1.935483870967742e-06,
+      "loss": 1.0568,
+      "step": 6
+    },
+    {
+      "epoch": 0.02834008097165992,
+      "grad_norm": 4.4352788247317845,
+      "learning_rate": 2.2580645161290324e-06,
+      "loss": 0.9962,
+      "step": 7
+    },
+    {
+      "epoch": 0.032388663967611336,
+      "grad_norm": 4.200772821630771,
+      "learning_rate": 2.580645161290323e-06,
+      "loss": 1.0167,
+      "step": 8
+    },
+    {
+      "epoch": 0.03643724696356275,
+      "grad_norm": 2.524821088243099,
+      "learning_rate": 2.903225806451613e-06,
+      "loss": 0.9916,
+      "step": 9
+    },
+    {
+      "epoch": 0.04048582995951417,
+      "grad_norm": 2.387740123202893,
+      "learning_rate": 3.225806451612903e-06,
+      "loss": 0.9601,
+      "step": 10
+    },
+    {
+      "epoch": 0.044534412955465584,
+      "grad_norm": 2.3172984278387183,
+      "learning_rate": 3.548387096774194e-06,
+      "loss": 0.9671,
+      "step": 11
+    },
+    {
+      "epoch": 0.048582995951417005,
+      "grad_norm": 3.311387383836343,
+      "learning_rate": 3.870967741935484e-06,
+      "loss": 0.9196,
+      "step": 12
+    },
+    {
+      "epoch": 0.05263157894736842,
+      "grad_norm": 3.4732884201832426,
+      "learning_rate": 4.193548387096774e-06,
+      "loss": 0.9425,
+      "step": 13
+    },
+    {
+      "epoch": 0.05668016194331984,
+      "grad_norm": 3.0815422807782333,
+      "learning_rate": 4.516129032258065e-06,
+      "loss": 0.9033,
+      "step": 14
+    },
+    {
+      "epoch": 0.06072874493927125,
+      "grad_norm": 2.6881381710828354,
+      "learning_rate": 4.838709677419355e-06,
+      "loss": 0.9099,
+      "step": 15
+    },
+    {
+      "epoch": 0.06477732793522267,
+      "grad_norm": 1.9706885243052445,
+      "learning_rate": 5.161290322580646e-06,
+      "loss": 0.89,
+      "step": 16
+    },
+    {
+      "epoch": 0.06882591093117409,
+      "grad_norm": 2.171162519651841,
+      "learning_rate": 5.483870967741935e-06,
+      "loss": 0.8733,
+      "step": 17
+    },
+    {
+      "epoch": 0.0728744939271255,
+      "grad_norm": 2.151279935119755,
+      "learning_rate": 5.806451612903226e-06,
+      "loss": 0.8909,
+      "step": 18
+    },
+    {
+      "epoch": 0.07692307692307693,
+      "grad_norm": 1.550426462525218,
+      "learning_rate": 6.129032258064517e-06,
+      "loss": 0.8504,
+      "step": 19
+    },
+    {
+      "epoch": 0.08097165991902834,
+      "grad_norm": 1.1629895958060532,
+      "learning_rate": 6.451612903225806e-06,
+      "loss": 0.8136,
+      "step": 20
+    },
+    {
+      "epoch": 0.08502024291497975,
+      "grad_norm": 1.2379719512395777,
+      "learning_rate": 6.774193548387097e-06,
+      "loss": 0.8206,
+      "step": 21
+    },
+    {
+      "epoch": 0.08906882591093117,
+      "grad_norm": 1.3492860428262017,
+      "learning_rate": 7.096774193548388e-06,
+      "loss": 0.8339,
+      "step": 22
+    },
+    {
+      "epoch": 0.0931174089068826,
+      "grad_norm": 1.327795669890712,
+      "learning_rate": 7.4193548387096784e-06,
+      "loss": 0.7928,
+      "step": 23
+    },
+    {
+      "epoch": 0.09716599190283401,
+      "grad_norm": 0.9728575184652763,
+      "learning_rate": 7.741935483870968e-06,
+      "loss": 0.8024,
+      "step": 24
+    },
+    {
+      "epoch": 0.10121457489878542,
+      "grad_norm": 0.9521898838856783,
+      "learning_rate": 8.064516129032258e-06,
+      "loss": 0.7876,
+      "step": 25
+    },
+    {
+      "epoch": 0.10526315789473684,
+      "grad_norm": 0.9396228882406651,
+      "learning_rate": 8.387096774193549e-06,
+      "loss": 0.8028,
+      "step": 26
+    },
+    {
+      "epoch": 0.10931174089068826,
+      "grad_norm": 0.9186047980986983,
+      "learning_rate": 8.70967741935484e-06,
+      "loss": 0.7945,
+      "step": 27
+    },
+    {
+      "epoch": 0.11336032388663968,
+      "grad_norm": 0.748412888810471,
+      "learning_rate": 9.03225806451613e-06,
+      "loss": 0.7669,
+      "step": 28
+    },
+    {
+      "epoch": 0.11740890688259109,
+      "grad_norm": 0.7611302870799146,
+      "learning_rate": 9.35483870967742e-06,
+      "loss": 0.7832,
+      "step": 29
+    },
+    {
+      "epoch": 0.1214574898785425,
+      "grad_norm": 0.7476835515774477,
+      "learning_rate": 9.67741935483871e-06,
+      "loss": 0.7375,
+      "step": 30
+    },
+    {
+      "epoch": 0.12550607287449392,
+      "grad_norm": 0.7278137957099209,
+      "learning_rate": 1e-05,
+      "loss": 0.7586,
+      "step": 31
+    },
+    {
+      "epoch": 0.12955465587044535,
+      "grad_norm": 0.6792860302932225,
+      "learning_rate": 1.0322580645161291e-05,
+      "loss": 0.7559,
+      "step": 32
+    },
+    {
+      "epoch": 0.13360323886639677,
+      "grad_norm": 0.6630006280538379,
+      "learning_rate": 1.0645161290322582e-05,
+      "loss": 0.7607,
+      "step": 33
+    },
+    {
+      "epoch": 0.13765182186234817,
+      "grad_norm": 0.7488528869115579,
+      "learning_rate": 1.096774193548387e-05,
+      "loss": 0.7617,
+      "step": 34
+    },
+    {
+      "epoch": 0.1417004048582996,
+      "grad_norm": 0.7575182779993863,
+      "learning_rate": 1.1290322580645164e-05,
+      "loss": 0.7592,
+      "step": 35
+    },
+    {
+      "epoch": 0.145748987854251,
+      "grad_norm": 0.6408759836977601,
+      "learning_rate": 1.1612903225806453e-05,
+      "loss": 0.7555,
+      "step": 36
+    },
+    {
+      "epoch": 0.14979757085020243,
+      "grad_norm": 0.7974919639947601,
+      "learning_rate": 1.1935483870967743e-05,
+      "loss": 0.7385,
+      "step": 37
+    },
+    {
+      "epoch": 0.15384615384615385,
+      "grad_norm": 0.777771112093958,
+      "learning_rate": 1.2258064516129034e-05,
+      "loss": 0.7568,
+      "step": 38
+    },
+    {
+      "epoch": 0.15789473684210525,
+      "grad_norm": 0.6313223740483368,
+      "learning_rate": 1.2580645161290324e-05,
+      "loss": 0.7395,
+      "step": 39
+    },
+    {
+      "epoch": 0.16194331983805668,
+      "grad_norm": 0.777807502292222,
+      "learning_rate": 1.2903225806451613e-05,
+      "loss": 0.7433,
+      "step": 40
+    },
+    {
+      "epoch": 0.1659919028340081,
+      "grad_norm": 0.641575916250809,
+      "learning_rate": 1.3225806451612903e-05,
+      "loss": 0.719,
+      "step": 41
+    },
+    {
+      "epoch": 0.1700404858299595,
+      "grad_norm": 0.6903281846114853,
+      "learning_rate": 1.3548387096774194e-05,
+      "loss": 0.7417,
+      "step": 42
+    },
+    {
+      "epoch": 0.17408906882591094,
+      "grad_norm": 0.8926293513106064,
+      "learning_rate": 1.3870967741935486e-05,
+      "loss": 0.7324,
+      "step": 43
+    },
+    {
+      "epoch": 0.17813765182186234,
+      "grad_norm": 0.5561835660074063,
+      "learning_rate": 1.4193548387096776e-05,
+      "loss": 0.7438,
+      "step": 44
+    },
+    {
+      "epoch": 0.18218623481781376,
+      "grad_norm": 1.016351081654112,
+      "learning_rate": 1.4516129032258066e-05,
+      "loss": 0.7097,
+      "step": 45
+    },
+    {
+      "epoch": 0.1862348178137652,
+      "grad_norm": 0.566570461098952,
+      "learning_rate": 1.4838709677419357e-05,
+      "loss": 0.718,
+      "step": 46
+    },
+    {
+      "epoch": 0.1902834008097166,
+      "grad_norm": 0.7198680625602359,
+      "learning_rate": 1.5161290322580646e-05,
+      "loss": 0.7308,
+      "step": 47
+    },
+    {
+      "epoch": 0.19433198380566802,
+      "grad_norm": 0.6009861802736409,
+      "learning_rate": 1.5483870967741936e-05,
+      "loss": 0.7374,
+      "step": 48
+    },
+    {
+      "epoch": 0.19838056680161945,
+      "grad_norm": 0.6416068130929297,
+      "learning_rate": 1.5806451612903226e-05,
+      "loss": 0.73,
+      "step": 49
+    },
+    {
+      "epoch": 0.20242914979757085,
+      "grad_norm": 0.6430174071233516,
+      "learning_rate": 1.6129032258064517e-05,
+      "loss": 0.7398,
+      "step": 50
+    },
+    {
+      "epoch": 0.20647773279352227,
+      "grad_norm": 0.7542930219931179,
+      "learning_rate": 1.6451612903225807e-05,
+      "loss": 0.729,
+      "step": 51
+    },
+    {
+      "epoch": 0.21052631578947367,
+      "grad_norm": 0.6502007798694655,
+      "learning_rate": 1.6774193548387098e-05,
+      "loss": 0.7221,
+      "step": 52
+    },
+    {
+      "epoch": 0.2145748987854251,
+      "grad_norm": 0.9397932261088281,
+      "learning_rate": 1.7096774193548388e-05,
+      "loss": 0.7271,
+      "step": 53
+    },
+    {
+      "epoch": 0.21862348178137653,
+      "grad_norm": 0.5965836177984346,
+      "learning_rate": 1.741935483870968e-05,
+      "loss": 0.7235,
+      "step": 54
+    },
+    {
+      "epoch": 0.22267206477732793,
+      "grad_norm": 0.7687092466723786,
+      "learning_rate": 1.774193548387097e-05,
+      "loss": 0.6864,
+      "step": 55
+    },
+    {
+      "epoch": 0.22672064777327935,
+      "grad_norm": 0.6621344295002758,
+      "learning_rate": 1.806451612903226e-05,
+      "loss": 0.722,
+      "step": 56
+    },
+    {
+      "epoch": 0.23076923076923078,
+      "grad_norm": 0.6090875880707193,
+      "learning_rate": 1.838709677419355e-05,
+      "loss": 0.7164,
+      "step": 57
+    },
+    {
+      "epoch": 0.23481781376518218,
+      "grad_norm": 0.5577484777648477,
+      "learning_rate": 1.870967741935484e-05,
+      "loss": 0.7171,
+      "step": 58
+    },
+    {
+      "epoch": 0.2388663967611336,
+      "grad_norm": 0.7099103867698489,
+      "learning_rate": 1.903225806451613e-05,
+      "loss": 0.7073,
+      "step": 59
+    },
+    {
+      "epoch": 0.242914979757085,
+      "grad_norm": 0.567656399530447,
+      "learning_rate": 1.935483870967742e-05,
+      "loss": 0.6965,
+      "step": 60
+    },
+    {
+      "epoch": 0.24696356275303644,
+      "grad_norm": 0.786520919288324,
+      "learning_rate": 1.967741935483871e-05,
+      "loss": 0.6922,
+      "step": 61
+    },
+    {
+      "epoch": 0.25101214574898784,
+      "grad_norm": 0.6875520101771185,
+      "learning_rate": 2e-05,
+      "loss": 0.7101,
+      "step": 62
+    },
+    {
+      "epoch": 0.2550607287449393,
+      "grad_norm": 0.8030036571478365,
+      "learning_rate": 2.0322580645161292e-05,
+      "loss": 0.6993,
+      "step": 63
+    },
+    {
+      "epoch": 0.2591093117408907,
+      "grad_norm": 0.8417743057946424,
+      "learning_rate": 2.0645161290322582e-05,
+      "loss": 0.7061,
+      "step": 64
+    },
+    {
+      "epoch": 0.2631578947368421,
+      "grad_norm": 0.7520867030201985,
+      "learning_rate": 2.0967741935483873e-05,
+      "loss": 0.704,
+      "step": 65
+    },
+    {
+      "epoch": 0.26720647773279355,
+      "grad_norm": 0.9315576803559299,
+      "learning_rate": 2.1290322580645163e-05,
+      "loss": 0.7124,
+      "step": 66
+    },
+    {
+      "epoch": 0.27125506072874495,
+      "grad_norm": 0.9379708974905925,
+      "learning_rate": 2.161290322580645e-05,
+      "loss": 0.7149,
+      "step": 67
+    },
+    {
+      "epoch": 0.27530364372469635,
+      "grad_norm": 0.8074012272474558,
+      "learning_rate": 2.193548387096774e-05,
+      "loss": 0.695,
+      "step": 68
+    },
+    {
+      "epoch": 0.2793522267206478,
+      "grad_norm": 0.5872782498899629,
+      "learning_rate": 2.225806451612903e-05,
+      "loss": 0.6959,
+      "step": 69
+    },
+    {
+      "epoch": 0.2834008097165992,
+      "grad_norm": 0.7664306722890953,
+      "learning_rate": 2.2580645161290328e-05,
+      "loss": 0.7123,
+      "step": 70
+    },
+    {
+      "epoch": 0.2874493927125506,
+      "grad_norm": 0.7912520132247243,
+      "learning_rate": 2.2903225806451615e-05,
+      "loss": 0.704,
+      "step": 71
+    },
+    {
+      "epoch": 0.291497975708502,
+      "grad_norm": 0.7861567542357676,
+      "learning_rate": 2.3225806451612906e-05,
+      "loss": 0.701,
+      "step": 72
+    },
+    {
+      "epoch": 0.29554655870445345,
+      "grad_norm": 0.8085100161210469,
+      "learning_rate": 2.3548387096774196e-05,
+      "loss": 0.7005,
+      "step": 73
+    },
+    {
+      "epoch": 0.29959514170040485,
+      "grad_norm": 0.9084310085741203,
+      "learning_rate": 2.3870967741935486e-05,
+      "loss": 0.6765,
+      "step": 74
+    },
+    {
+      "epoch": 0.30364372469635625,
+      "grad_norm": 0.9352820101961716,
+      "learning_rate": 2.4193548387096777e-05,
+      "loss": 0.718,
+      "step": 75
+    },
+    {
+      "epoch": 0.3076923076923077,
+      "grad_norm": 0.8401709298013247,
+      "learning_rate": 2.4516129032258067e-05,
+      "loss": 0.7133,
+      "step": 76
+    },
+    {
+      "epoch": 0.3117408906882591,
+      "grad_norm": 0.6218718469479543,
+      "learning_rate": 2.4838709677419358e-05,
+      "loss": 0.7082,
+      "step": 77
+    },
+    {
+      "epoch": 0.3157894736842105,
+      "grad_norm": 0.8371160532882184,
+      "learning_rate": 2.5161290322580648e-05,
+      "loss": 0.7053,
+      "step": 78
+    },
+    {
+      "epoch": 0.31983805668016196,
+      "grad_norm": 0.7028928954366277,
+      "learning_rate": 2.548387096774194e-05,
+      "loss": 0.6878,
+      "step": 79
+    },
+    {
+      "epoch": 0.32388663967611336,
+      "grad_norm": 0.6929756671279174,
+      "learning_rate": 2.5806451612903226e-05,
+      "loss": 0.6959,
+      "step": 80
+    },
+    {
+      "epoch": 0.32793522267206476,
+      "grad_norm": 0.8405608591574003,
+      "learning_rate": 2.6129032258064516e-05,
+      "loss": 0.6969,
+      "step": 81
+    },
+    {
+      "epoch": 0.3319838056680162,
+      "grad_norm": 0.7577922815053332,
+      "learning_rate": 2.6451612903225806e-05,
+      "loss": 0.6961,
+      "step": 82
+    },
+    {
+      "epoch": 0.3360323886639676,
+      "grad_norm": 1.1708241964139425,
+      "learning_rate": 2.6774193548387097e-05,
+      "loss": 0.6997,
+      "step": 83
+    },
+    {
+      "epoch": 0.340080971659919,
+      "grad_norm": 0.7961454891100223,
+      "learning_rate": 2.7096774193548387e-05,
+      "loss": 0.6939,
+      "step": 84
+    },
+    {
+      "epoch": 0.3441295546558704,
+      "grad_norm": 0.8501954252840936,
+      "learning_rate": 2.7419354838709678e-05,
+      "loss": 0.7188,
+      "step": 85
+    },
+    {
+      "epoch": 0.3481781376518219,
+      "grad_norm": 0.7272653999849925,
+      "learning_rate": 2.774193548387097e-05,
+      "loss": 0.6903,
+      "step": 86
+    },
+    {
+      "epoch": 0.3522267206477733,
+      "grad_norm": 0.926775578221197,
+      "learning_rate": 2.8064516129032262e-05,
+      "loss": 0.6864,
+      "step": 87
+    },
+    {
+      "epoch": 0.3562753036437247,
+      "grad_norm": 1.4145317657358436,
+      "learning_rate": 2.8387096774193552e-05,
+      "loss": 0.719,
+      "step": 88
+    },
+    {
+      "epoch": 0.3603238866396761,
+      "grad_norm": 0.7499779998433067,
+      "learning_rate": 2.8709677419354843e-05,
+      "loss": 0.6944,
+      "step": 89
+    },
+    {
+      "epoch": 0.3643724696356275,
+      "grad_norm": 1.4678063180652459,
+      "learning_rate": 2.9032258064516133e-05,
+      "loss": 0.688,
+      "step": 90
+    },
+    {
+      "epoch": 0.3684210526315789,
+      "grad_norm": 0.800773971107059,
+      "learning_rate": 2.9354838709677423e-05,
+      "loss": 0.702,
+      "step": 91
+    },
+    {
+      "epoch": 0.3724696356275304,
+      "grad_norm": 0.7707338751097913,
+      "learning_rate": 2.9677419354838714e-05,
+      "loss": 0.6975,
+      "step": 92
+    },
+    {
+      "epoch": 0.3765182186234818,
+      "grad_norm": 0.9899124172536848,
+      "learning_rate": 3.0000000000000004e-05,
+      "loss": 0.7125,
+      "step": 93
+    },
+    {
+      "epoch": 0.3805668016194332,
+      "grad_norm": 1.1048803490324055,
+      "learning_rate": 3.032258064516129e-05,
+      "loss": 0.6813,
+      "step": 94
+    },
+    {
+      "epoch": 0.38461538461538464,
+      "grad_norm": 0.7805948272612039,
+      "learning_rate": 3.0645161290322585e-05,
+      "loss": 0.6733,
+      "step": 95
+    },
+    {
+      "epoch": 0.38866396761133604,
+      "grad_norm": 0.8986961485658556,
+      "learning_rate": 3.096774193548387e-05,
+      "loss": 0.6898,
+      "step": 96
+    },
+    {
+      "epoch": 0.39271255060728744,
+      "grad_norm": 1.4303696994901112,
+      "learning_rate": 3.1290322580645166e-05,
+      "loss": 0.6943,
+      "step": 97
+    },
+    {
+      "epoch": 0.3967611336032389,
+      "grad_norm": 0.7163411135649276,
+      "learning_rate": 3.161290322580645e-05,
+      "loss": 0.6855,
+      "step": 98
+    },
+    {
+      "epoch": 0.4008097165991903,
+      "grad_norm": 1.2701832087355207,
+      "learning_rate": 3.193548387096774e-05,
+      "loss": 0.6868,
+      "step": 99
+    },
+    {
+      "epoch": 0.4048582995951417,
+      "grad_norm": 0.8726934541406115,
+      "learning_rate": 3.2258064516129034e-05,
+      "loss": 0.6722,
+      "step": 100
+    },
+    {
+      "epoch": 0.4089068825910931,
+      "grad_norm": 0.7686531754926523,
+      "learning_rate": 3.258064516129033e-05,
+      "loss": 0.7115,
+      "step": 101
+    },
+    {
+      "epoch": 0.41295546558704455,
+      "grad_norm": 1.066497558327885,
+      "learning_rate": 3.2903225806451614e-05,
+      "loss": 0.6963,
+      "step": 102
+    },
+    {
+      "epoch": 0.41700404858299595,
+      "grad_norm": 1.028012541511116,
+      "learning_rate": 3.322580645161291e-05,
+      "loss": 0.6946,
+      "step": 103
+    },
+    {
+      "epoch": 0.42105263157894735,
+      "grad_norm": 0.8614820271674268,
+      "learning_rate": 3.3548387096774195e-05,
+      "loss": 0.6806,
+      "step": 104
+    },
+    {
+      "epoch": 0.4251012145748988,
+      "grad_norm": 0.7442226880460873,
+      "learning_rate": 3.387096774193549e-05,
+      "loss": 0.7038,
+      "step": 105
+    },
+    {
+      "epoch": 0.4291497975708502,
+      "grad_norm": 0.8737033478409245,
+      "learning_rate": 3.4193548387096776e-05,
+      "loss": 0.6973,
+      "step": 106
+    },
+    {
+      "epoch": 0.4331983805668016,
+      "grad_norm": 0.786879117731717,
+      "learning_rate": 3.451612903225807e-05,
+      "loss": 0.6919,
+      "step": 107
+    },
+    {
+      "epoch": 0.43724696356275305,
+      "grad_norm": 0.7759678422503173,
+      "learning_rate": 3.483870967741936e-05,
+      "loss": 0.6823,
+      "step": 108
+    },
+    {
+      "epoch": 0.44129554655870445,
+      "grad_norm": 0.9294119248888878,
+      "learning_rate": 3.516129032258065e-05,
+      "loss": 0.7058,
+      "step": 109
+    },
+    {
+      "epoch": 0.44534412955465585,
+      "grad_norm": 1.0547396829031759,
+      "learning_rate": 3.548387096774194e-05,
+      "loss": 0.6791,
+      "step": 110
+    },
+    {
+      "epoch": 0.4493927125506073,
+      "grad_norm": 1.2590058841902494,
+      "learning_rate": 3.5806451612903225e-05,
+      "loss": 0.7164,
+      "step": 111
+    },
+    {
+      "epoch": 0.4534412955465587,
+      "grad_norm": 0.7285616675584562,
+      "learning_rate": 3.612903225806452e-05,
+      "loss": 0.7035,
+      "step": 112
+    },
+    {
+      "epoch": 0.4574898785425101,
+      "grad_norm": 1.0651229394447368,
+      "learning_rate": 3.6451612903225805e-05,
+      "loss": 0.6849,
+      "step": 113
+    },
+    {
+      "epoch": 0.46153846153846156,
+      "grad_norm": 1.1037474727969976,
+      "learning_rate": 3.67741935483871e-05,
+      "loss": 0.7034,
+      "step": 114
+    },
+    {
+      "epoch": 0.46558704453441296,
+      "grad_norm": 0.9633233174022997,
+      "learning_rate": 3.7096774193548386e-05,
+      "loss": 0.709,
+      "step": 115
+    },
+    {
+      "epoch": 0.46963562753036436,
+      "grad_norm": 0.8906177029787045,
+      "learning_rate": 3.741935483870968e-05,
+      "loss": 0.7058,
+      "step": 116
+    },
+    {
+      "epoch": 0.47368421052631576,
+      "grad_norm": 0.9281638229119158,
+      "learning_rate": 3.7741935483870974e-05,
+      "loss": 0.687,
+      "step": 117
+    },
+    {
+      "epoch": 0.4777327935222672,
+      "grad_norm": 0.9904472120759535,
+      "learning_rate": 3.806451612903226e-05,
+      "loss": 0.6855,
+      "step": 118
+    },
+    {
+      "epoch": 0.4817813765182186,
+      "grad_norm": 1.0357333413964982,
+      "learning_rate": 3.8387096774193555e-05,
+      "loss": 0.6815,
+      "step": 119
+    },
+    {
+      "epoch": 0.48582995951417,
+      "grad_norm": 1.0223809901311933,
+      "learning_rate": 3.870967741935484e-05,
+      "loss": 0.687,
+      "step": 120
+    },
+    {
+      "epoch": 0.4898785425101215,
+      "grad_norm": 1.140604858410277,
+      "learning_rate": 3.9032258064516135e-05,
+      "loss": 0.6833,
+      "step": 121
+    },
+    {
+      "epoch": 0.4939271255060729,
+      "grad_norm": 0.6855821130836728,
+      "learning_rate": 3.935483870967742e-05,
+      "loss": 0.6814,
+      "step": 122
+    },
+    {
+      "epoch": 0.4979757085020243,
+      "grad_norm": 1.0681562747443636,
+      "learning_rate": 3.9677419354838716e-05,
+      "loss": 0.6793,
+      "step": 123
+    },
+    {
+      "epoch": 0.5020242914979757,
+      "grad_norm": 0.9272233493489369,
+      "learning_rate": 4e-05,
+      "loss": 0.6867,
+      "step": 124
+    },
+    {
+      "epoch": 0.5060728744939271,
+      "grad_norm": 1.0352733994401442,
+      "learning_rate": 3.9999920040266474e-05,
+      "loss": 0.6889,
+      "step": 125
+    },
+    {
+      "epoch": 0.5101214574898786,
+      "grad_norm": 0.715719291266791,
+      "learning_rate": 3.999968016170525e-05,
+      "loss": 0.6635,
+      "step": 126
+    },
+    {
+      "epoch": 0.5141700404858299,
+      "grad_norm": 0.8018151908810274,
+      "learning_rate": 3.9999280366234394e-05,
+      "loss": 0.6746,
+      "step": 127
+    },
+    {
+      "epoch": 0.5182186234817814,
+      "grad_norm": 0.9055791220406902,
+      "learning_rate": 3.999872065705065e-05,
+      "loss": 0.6897,
+      "step": 128
+    },
+    {
+      "epoch": 0.5222672064777328,
+      "grad_norm": 1.0593951752301194,
+      "learning_rate": 3.999800103862945e-05,
+      "loss": 0.7045,
+      "step": 129
+    },
+    {
+      "epoch": 0.5263157894736842,
+      "grad_norm": 0.7141766950807921,
+      "learning_rate": 3.9997121516724834e-05,
+      "loss": 0.6712,
+      "step": 130
+    },
+    {
+      "epoch": 0.5303643724696356,
+      "grad_norm": 0.936722954165768,
+      "learning_rate": 3.999608209836944e-05,
+      "loss": 0.6966,
+      "step": 131
+    },
+    {
+      "epoch": 0.5344129554655871,
+      "grad_norm": 0.6815097394984646,
+      "learning_rate": 3.999488279187443e-05,
+      "loss": 0.686,
+      "step": 132
+    },
+    {
+      "epoch": 0.5384615384615384,
+      "grad_norm": 0.71878544805492,
+      "learning_rate": 3.9993523606829425e-05,
+      "loss": 0.6859,
+      "step": 133
+    },
+    {
+      "epoch": 0.5425101214574899,
+      "grad_norm": 0.739567965731752,
+      "learning_rate": 3.9992004554102435e-05,
+      "loss": 0.6743,
+      "step": 134
+    },
+    {
+      "epoch": 0.5465587044534413,
+      "grad_norm": 0.7733399781741298,
+      "learning_rate": 3.999032564583976e-05,
+      "loss": 0.6851,
+      "step": 135
+    },
+    {
+      "epoch": 0.5506072874493927,
+      "grad_norm": 0.8512149863329531,
+      "learning_rate": 3.998848689546591e-05,
+      "loss": 0.6886,
+      "step": 136
+    },
+    {
+      "epoch": 0.5546558704453441,
+      "grad_norm": 0.7103075568757141,
+      "learning_rate": 3.9986488317683496e-05,
+      "loss": 0.7052,
+      "step": 137
+    },
+    {
+      "epoch": 0.5587044534412956,
+      "grad_norm": 0.8125622345075315,
+      "learning_rate": 3.9984329928473067e-05,
+      "loss": 0.6916,
+      "step": 138
+    },
+    {
+      "epoch": 0.562753036437247,
+      "grad_norm": 0.856865076306539,
+      "learning_rate": 3.998201174509306e-05,
+      "loss": 0.7007,
+      "step": 139
+    },
+    {
+      "epoch": 0.5668016194331984,
+      "grad_norm": 0.7770493968416377,
+      "learning_rate": 3.997953378607961e-05,
+      "loss": 0.6917,
+      "step": 140
+    },
+    {
+      "epoch": 0.5708502024291497,
+      "grad_norm": 0.6601699616759417,
+      "learning_rate": 3.9976896071246403e-05,
+      "loss": 0.6762,
+      "step": 141
+    },
+    {
+      "epoch": 0.5748987854251012,
+      "grad_norm": 0.5951929103665123,
+      "learning_rate": 3.997409862168454e-05,
+      "loss": 0.7289,
+      "step": 142
+    },
+    {
+      "epoch": 0.5789473684210527,
+      "grad_norm": 0.6345124923179682,
+      "learning_rate": 3.997114145976237e-05,
+      "loss": 0.6933,
+      "step": 143
+    },
+    {
+      "epoch": 0.582995951417004,
+      "grad_norm": 0.7041980989058875,
+      "learning_rate": 3.9968024609125245e-05,
+      "loss": 0.6983,
+      "step": 144
+    },
+    {
+      "epoch": 0.5870445344129555,
+      "grad_norm": 0.7848436018845351,
+      "learning_rate": 3.996474809469545e-05,
+      "loss": 0.7084,
+      "step": 145
+    },
+    {
+      "epoch": 0.5910931174089069,
+      "grad_norm": 0.6564232554834082,
+      "learning_rate": 3.996131194267189e-05,
+      "loss": 0.6964,
+      "step": 146
+    },
+    {
+      "epoch": 0.5951417004048583,
+      "grad_norm": 0.7615467510725065,
+      "learning_rate": 3.995771618052995e-05,
+      "loss": 0.6916,
+      "step": 147
+    },
+    {
+      "epoch": 0.5991902834008097,
+      "grad_norm": 0.7363719997848097,
+      "learning_rate": 3.9953960837021256e-05,
+      "loss": 0.6927,
+      "step": 148
+    },
+    {
+      "epoch": 0.6032388663967612,
+      "grad_norm": 0.6924934698503076,
+      "learning_rate": 3.995004594217343e-05,
+      "loss": 0.6839,
+      "step": 149
+    },
+    {
+      "epoch": 0.6072874493927125,
+      "grad_norm": 0.7003630119967684,
+      "learning_rate": 3.994597152728987e-05,
+      "loss": 0.6946,
+      "step": 150
+    },
+    {
+      "epoch": 0.611336032388664,
+      "grad_norm": 0.6474169821162569,
+      "learning_rate": 3.994173762494947e-05,
+      "loss": 0.6937,
+      "step": 151
+    },
+    {
+      "epoch": 0.6153846153846154,
+      "grad_norm": 0.6550440693425719,
+      "learning_rate": 3.9937344269006424e-05,
+      "loss": 0.6748,
+      "step": 152
+    },
+    {
+      "epoch": 0.6194331983805668,
+      "grad_norm": 0.5848728833812986,
+      "learning_rate": 3.993279149458988e-05,
+      "loss": 0.6686,
+      "step": 153
+    },
+    {
+      "epoch": 0.6234817813765182,
+      "grad_norm": 0.7086217827944161,
+      "learning_rate": 3.9928079338103695e-05,
+      "loss": 0.6711,
+      "step": 154
+    },
+    {
+      "epoch": 0.6275303643724697,
+      "grad_norm": 0.5547558345028344,
+      "learning_rate": 3.992320783722616e-05,
+      "loss": 0.6836,
+      "step": 155
+    },
+    {
+      "epoch": 0.631578947368421,
+      "grad_norm": 0.5855961012494776,
+      "learning_rate": 3.991817703090965e-05,
+      "loss": 0.7077,
+      "step": 156
+    },
+    {
+      "epoch": 0.6356275303643725,
+      "grad_norm": 0.6377749728478641,
+      "learning_rate": 3.991298695938038e-05,
+      "loss": 0.6932,
+      "step": 157
+    },
+    {
+      "epoch": 0.6396761133603239,
+      "grad_norm": 0.6846919843984827,
+      "learning_rate": 3.9907637664138e-05,
+      "loss": 0.6777,
+      "step": 158
+    },
+    {
+      "epoch": 0.6437246963562753,
+      "grad_norm": 0.6645907951435576,
+      "learning_rate": 3.990212918795534e-05,
+      "loss": 0.6908,
+      "step": 159
+    },
+    {
+      "epoch": 0.6477732793522267,
+      "grad_norm": 0.5686979007041219,
+      "learning_rate": 3.9896461574878036e-05,
+      "loss": 0.6701,
+      "step": 160
+    },
+    {
+      "epoch": 0.6518218623481782,
+      "grad_norm": 0.7564626145420931,
+      "learning_rate": 3.989063487022416e-05,
+      "loss": 0.6564,
+      "step": 161
+    },
+    {
+      "epoch": 0.6558704453441295,
+      "grad_norm": 0.6330057436342083,
+      "learning_rate": 3.9884649120583903e-05,
+      "loss": 0.6811,
+      "step": 162
+    },
+    {
+      "epoch": 0.659919028340081,
+      "grad_norm": 0.5704192170650445,
+      "learning_rate": 3.987850437381915e-05,
+      "loss": 0.6817,
+      "step": 163
+    },
+    {
+      "epoch": 0.6639676113360324,
+      "grad_norm": 0.6362569999134506,
+      "learning_rate": 3.987220067906313e-05,
+      "loss": 0.6867,
+      "step": 164
+    },
+    {
+      "epoch": 0.6680161943319838,
+      "grad_norm": 0.5640337044655594,
+      "learning_rate": 3.986573808672002e-05,
+      "loss": 0.6894,
+      "step": 165
+    },
+    {
+      "epoch": 0.6720647773279352,
+      "grad_norm": 0.6403772338618361,
+      "learning_rate": 3.9859116648464534e-05,
+      "loss": 0.6821,
+      "step": 166
+    },
+    {
+      "epoch": 0.6761133603238867,
+      "grad_norm": 0.5448428512252852,
+      "learning_rate": 3.985233641724152e-05,
+      "loss": 0.6677,
+      "step": 167
+    },
+    {
+      "epoch": 0.680161943319838,
+      "grad_norm": 0.6584550852688358,
+      "learning_rate": 3.984539744726554e-05,
+      "loss": 0.6705,
+      "step": 168
+    },
+    {
+      "epoch": 0.6842105263157895,
+      "grad_norm": 0.7187892627588385,
+      "learning_rate": 3.983829979402038e-05,
+      "loss": 0.6812,
+      "step": 169
+    },
+    {
+      "epoch": 0.6882591093117408,
+      "grad_norm": 0.5812803846740467,
+      "learning_rate": 3.983104351425871e-05,
+      "loss": 0.6857,
+      "step": 170
+    },
+    {
+      "epoch": 0.6923076923076923,
+      "grad_norm": 0.5365184799377494,
+      "learning_rate": 3.982362866600154e-05,
+      "loss": 0.6865,
+      "step": 171
+    },
+    {
+      "epoch": 0.6963562753036437,
+      "grad_norm": 0.7315470832350932,
+      "learning_rate": 3.9816055308537806e-05,
+      "loss": 0.6971,
+      "step": 172
+    },
+    {
+      "epoch": 0.7004048582995951,
+      "grad_norm": 0.8064531326627215,
+      "learning_rate": 3.980832350242388e-05,
+      "loss": 0.6742,
+      "step": 173
+    },
+    {
+      "epoch": 0.7044534412955465,
+      "grad_norm": 0.5850617191398704,
+      "learning_rate": 3.980043330948306e-05,
+      "loss": 0.6805,
+      "step": 174
+    },
+    {
+      "epoch": 0.708502024291498,
+      "grad_norm": 0.5330119558171106,
+      "learning_rate": 3.9792384792805124e-05,
+      "loss": 0.6737,
+      "step": 175
+    },
+    {
+      "epoch": 0.7125506072874493,
+      "grad_norm": 0.6181454250803062,
+      "learning_rate": 3.978417801674579e-05,
+      "loss": 0.6739,
+      "step": 176
+    },
+    {
+      "epoch": 0.7165991902834008,
+      "grad_norm": 0.4876069392027213,
+      "learning_rate": 3.9775813046926244e-05,
+      "loss": 0.694,
+      "step": 177
+    },
+    {
+      "epoch": 0.7206477732793523,
+      "grad_norm": 0.5917179465002591,
+      "learning_rate": 3.976728995023254e-05,
+      "loss": 0.6959,
+      "step": 178
+    },
+    {
+      "epoch": 0.7246963562753036,
+      "grad_norm": 0.6446840418494931,
+      "learning_rate": 3.975860879481514e-05,
+      "loss": 0.6843,
+      "step": 179
+    },
+    {
+      "epoch": 0.728744939271255,
+      "grad_norm": 0.6725158942709312,
+      "learning_rate": 3.974976965008833e-05,
+      "loss": 0.6938,
+      "step": 180
+    },
+    {
+      "epoch": 0.7327935222672065,
+      "grad_norm": 0.7050351940444379,
+      "learning_rate": 3.9740772586729674e-05,
+      "loss": 0.6776,
+      "step": 181
+    },
+    {
+      "epoch": 0.7368421052631579,
+      "grad_norm": 1.1900303792119429,
+      "learning_rate": 3.973161767667946e-05,
+      "loss": 0.6672,
+      "step": 182
+    },
+    {
+      "epoch": 0.7408906882591093,
+      "grad_norm": 0.5807011572012102,
+      "learning_rate": 3.972230499314009e-05,
+      "loss": 0.6738,
+      "step": 183
+    },
+    {
+      "epoch": 0.7449392712550608,
+      "grad_norm": 1.0980917950610483,
+      "learning_rate": 3.9712834610575545e-05,
+      "loss": 0.6897,
+      "step": 184
+    },
+    {
+      "epoch": 0.7489878542510121,
+      "grad_norm": 0.6399464383014477,
+      "learning_rate": 3.9703206604710746e-05,
+      "loss": 0.6954,
+      "step": 185
+    },
+    {
+      "epoch": 0.7530364372469636,
+      "grad_norm": 0.7258069510224404,
+      "learning_rate": 3.969342105253098e-05,
+      "loss": 0.6905,
+      "step": 186
+    },
+    {
+      "epoch": 0.757085020242915,
+      "grad_norm": 0.6866495256537453,
+      "learning_rate": 3.9683478032281246e-05,
+      "loss": 0.675,
+      "step": 187
+    },
+    {
+      "epoch": 0.7611336032388664,
+      "grad_norm": 0.6447766310513656,
+      "learning_rate": 3.967337762346568e-05,
+      "loss": 0.6955,
+      "step": 188
+    },
+    {
+      "epoch": 0.7651821862348178,
+      "grad_norm": 0.5920558290130435,
+      "learning_rate": 3.9663119906846885e-05,
+      "loss": 0.7084,
+      "step": 189
+    },
+    {
+      "epoch": 0.7692307692307693,
+      "grad_norm": 0.7873469098232636,
+      "learning_rate": 3.965270496444528e-05,
+      "loss": 0.6816,
+      "step": 190
+    },
+    {
+      "epoch": 0.7732793522267206,
+      "grad_norm": 0.561880499229307,
+      "learning_rate": 3.9642132879538465e-05,
+      "loss": 0.6932,
+      "step": 191
+    },
+    {
+      "epoch": 0.7773279352226721,
+      "grad_norm": 0.7169990988327277,
+      "learning_rate": 3.9631403736660566e-05,
+      "loss": 0.6653,
+      "step": 192
+    },
+    {
+      "epoch": 0.7813765182186235,
+      "grad_norm": 0.6279546627575615,
+      "learning_rate": 3.962051762160151e-05,
+      "loss": 0.6668,
+      "step": 193
+    },
+    {
+      "epoch": 0.7854251012145749,
+      "grad_norm": 0.6440307483980716,
+      "learning_rate": 3.960947462140638e-05,
+      "loss": 0.6889,
+      "step": 194
+    },
+    {
+      "epoch": 0.7894736842105263,
+      "grad_norm": 0.6821603889212943,
+      "learning_rate": 3.959827482437472e-05,
+      "loss": 0.6812,
+      "step": 195
+    },
+    {
+      "epoch": 0.7935222672064778,
+      "grad_norm": 0.7623287650895649,
+      "learning_rate": 3.958691832005981e-05,
+      "loss": 0.6736,
+      "step": 196
+    },
+    {
+      "epoch": 0.7975708502024291,
+      "grad_norm": 0.6954146438883124,
+      "learning_rate": 3.9575405199267944e-05,
+      "loss": 0.6697,
+      "step": 197
+    },
+    {
+      "epoch": 0.8016194331983806,
+      "grad_norm": 0.6139919177617866,
+      "learning_rate": 3.956373555405774e-05,
+      "loss": 0.678,
+      "step": 198
+    },
+    {
+      "epoch": 0.805668016194332,
+      "grad_norm": 0.6893441216094816,
+      "learning_rate": 3.955190947773937e-05,
+      "loss": 0.6861,
+      "step": 199
+    },
+    {
+      "epoch": 0.8097165991902834,
+      "grad_norm": 0.5731894312263041,
+      "learning_rate": 3.9539927064873824e-05,
+      "loss": 0.6756,
+      "step": 200
+    },
+    {
+      "epoch": 0.8137651821862348,
+      "grad_norm": 0.7774627650582961,
+      "learning_rate": 3.952778841127214e-05,
+      "loss": 0.6784,
+      "step": 201
+    },
+    {
+      "epoch": 0.8178137651821862,
+      "grad_norm": 0.6119955800822422,
+      "learning_rate": 3.95154936139947e-05,
+      "loss": 0.6697,
+      "step": 202
+    },
+    {
+      "epoch": 0.8218623481781376,
+      "grad_norm": 0.6379974473827404,
+      "learning_rate": 3.950304277135034e-05,
+      "loss": 0.6901,
+      "step": 203
+    },
+    {
+      "epoch": 0.8259109311740891,
+      "grad_norm": 0.5522722303307719,
+      "learning_rate": 3.94904359828957e-05,
+      "loss": 0.6757,
+      "step": 204
+    },
+    {
+      "epoch": 0.8299595141700404,
+      "grad_norm": 0.6279934405936006,
+      "learning_rate": 3.94776733494343e-05,
+      "loss": 0.6631,
+      "step": 205
+    },
+    {
+      "epoch": 0.8340080971659919,
+      "grad_norm": 0.6359210023062812,
+      "learning_rate": 3.9464754973015826e-05,
+      "loss": 0.6906,
+      "step": 206
+    },
+    {
+      "epoch": 0.8380566801619433,
+      "grad_norm": 0.7359317619059215,
+      "learning_rate": 3.9451680956935275e-05,
+      "loss": 0.6743,
+      "step": 207
+    },
+    {
+      "epoch": 0.8421052631578947,
+      "grad_norm": 0.6227769183652431,
+      "learning_rate": 3.943845140573213e-05,
+      "loss": 0.6847,
+      "step": 208
+    },
+    {
+      "epoch": 0.8461538461538461,
+      "grad_norm": 0.5854935358323026,
+      "learning_rate": 3.942506642518952e-05,
+      "loss": 0.6843,
+      "step": 209
+    },
+    {
+      "epoch": 0.8502024291497976,
+      "grad_norm": 0.7365093810318033,
+      "learning_rate": 3.9411526122333396e-05,
+      "loss": 0.6861,
+      "step": 210
+    },
+    {
+      "epoch": 0.854251012145749,
+      "grad_norm": 0.561373092073052,
+      "learning_rate": 3.939783060543167e-05,
+      "loss": 0.6606,
+      "step": 211
+    },
+    {
+      "epoch": 0.8582995951417004,
+      "grad_norm": 0.5909229445676151,
+      "learning_rate": 3.938397998399333e-05,
+      "loss": 0.6734,
+      "step": 212
+    },
+    {
+      "epoch": 0.8623481781376519,
+      "grad_norm": 0.566302984231421,
+      "learning_rate": 3.936997436876756e-05,
+      "loss": 0.6747,
+      "step": 213
+    },
+    {
+      "epoch": 0.8663967611336032,
+      "grad_norm": 0.5957776082391579,
+      "learning_rate": 3.935581387174289e-05,
+      "loss": 0.674,
+      "step": 214
+    },
+    {
+      "epoch": 0.8704453441295547,
+      "grad_norm": 0.573440850864575,
+      "learning_rate": 3.9341498606146295e-05,
+      "loss": 0.6716,
+      "step": 215
+    },
+    {
+      "epoch": 0.8744939271255061,
+      "grad_norm": 0.626748953064233,
+      "learning_rate": 3.932702868644224e-05,
+      "loss": 0.6904,
+      "step": 216
+    },
+    {
+      "epoch": 0.8785425101214575,
+      "grad_norm": 0.5238919947294322,
+      "learning_rate": 3.931240422833181e-05,
+      "loss": 0.6823,
+      "step": 217
+    },
+    {
+      "epoch": 0.8825910931174089,
+      "grad_norm": 0.5144360282026947,
+      "learning_rate": 3.9297625348751806e-05,
+      "loss": 0.6751,
+      "step": 218
+    },
+    {
+      "epoch": 0.8866396761133604,
+      "grad_norm": 0.5143401472586091,
+      "learning_rate": 3.9282692165873744e-05,
+      "loss": 0.667,
+      "step": 219
+    },
+    {
+      "epoch": 0.8906882591093117,
+      "grad_norm": 0.5210633912333873,
+      "learning_rate": 3.9267604799102945e-05,
+      "loss": 0.673,
+      "step": 220
+    },
+    {
+      "epoch": 0.8947368421052632,
+      "grad_norm": 0.5130831681008017,
+      "learning_rate": 3.9252363369077615e-05,
+      "loss": 0.6897,
+      "step": 221
+    },
+    {
+      "epoch": 0.8987854251012146,
+      "grad_norm": 0.612952395635294,
+      "learning_rate": 3.9236967997667796e-05,
+      "loss": 0.6809,
+      "step": 222
+    },
+    {
+      "epoch": 0.902834008097166,
+      "grad_norm": 0.5818354956411964,
+      "learning_rate": 3.9221418807974496e-05,
+      "loss": 0.6801,
+      "step": 223
+    },
+    {
+      "epoch": 0.9068825910931174,
+      "grad_norm": 0.5860616783994034,
+      "learning_rate": 3.92057159243286e-05,
+      "loss": 0.6791,
+      "step": 224
+    },
+    {
+      "epoch": 0.9109311740890689,
+      "grad_norm": 0.4418313832093877,
+      "learning_rate": 3.9189859472289956e-05,
+      "loss": 0.6873,
+      "step": 225
+    },
+    {
+      "epoch": 0.9149797570850202,
+      "grad_norm": 0.5462537992705272,
+      "learning_rate": 3.9173849578646324e-05,
+      "loss": 0.6758,
+      "step": 226
+    },
+    {
+      "epoch": 0.9190283400809717,
+      "grad_norm": 0.5005030988750496,
+      "learning_rate": 3.9157686371412395e-05,
+      "loss": 0.6765,
+      "step": 227
+    },
+    {
+      "epoch": 0.9230769230769231,
+      "grad_norm": 0.548968808204632,
+      "learning_rate": 3.914136997982874e-05,
+      "loss": 0.6718,
+      "step": 228
+    },
+    {
+      "epoch": 0.9271255060728745,
+      "grad_norm": 0.5231947263759159,
+      "learning_rate": 3.912490053436079e-05,
+      "loss": 0.6631,
+      "step": 229
+    },
+    {
+      "epoch": 0.9311740890688259,
+      "grad_norm": 0.5116525872254679,
+      "learning_rate": 3.9108278166697805e-05,
+      "loss": 0.6786,
+      "step": 230
+    },
+    {
+      "epoch": 0.9352226720647774,
+      "grad_norm": 0.6223552528542323,
+      "learning_rate": 3.909150300975178e-05,
+      "loss": 0.6847,
+      "step": 231
+    },
+    {
+      "epoch": 0.9392712550607287,
+      "grad_norm": 0.5458617730119272,
+      "learning_rate": 3.907457519765642e-05,
+      "loss": 0.6657,
+      "step": 232
+    },
+    {
+      "epoch": 0.9433198380566802,
+      "grad_norm": 0.6516581679187323,
+      "learning_rate": 3.905749486576607e-05,
+      "loss": 0.6848,
+      "step": 233
+    },
+    {
+      "epoch": 0.9473684210526315,
+      "grad_norm": 0.585598298050118,
+      "learning_rate": 3.90402621506546e-05,
+      "loss": 0.6788,
+      "step": 234
+    },
+    {
+      "epoch": 0.951417004048583,
+      "grad_norm": 0.5768584325860954,
+      "learning_rate": 3.902287719011435e-05,
+      "loss": 0.6542,
+      "step": 235
+    },
+    {
+      "epoch": 0.9554655870445344,
+      "grad_norm": 0.6250769741671467,
+      "learning_rate": 3.900534012315499e-05,
+      "loss": 0.6706,
+      "step": 236
+    },
+    {
+      "epoch": 0.9595141700404858,
+      "grad_norm": 0.5085245738688717,
+      "learning_rate": 3.8987651090002454e-05,
+      "loss": 0.6521,
+      "step": 237
+    },
+    {
+      "epoch": 0.9635627530364372,
+      "grad_norm": 0.6946082017078093,
+      "learning_rate": 3.896981023209777e-05,
+      "loss": 0.6499,
+      "step": 238
+    },
+    {
+      "epoch": 0.9676113360323887,
+      "grad_norm": 0.5724913348523432,
+      "learning_rate": 3.8951817692095966e-05,
+      "loss": 0.659,
+      "step": 239
+    },
+    {
+      "epoch": 0.97165991902834,
+      "grad_norm": 0.6934604542457313,
+      "learning_rate": 3.893367361386491e-05,
+      "loss": 0.6673,
+      "step": 240
+    },
+    {
+      "epoch": 0.9757085020242915,
+      "grad_norm": 0.6253213659960687,
+      "learning_rate": 3.891537814248417e-05,
+      "loss": 0.6598,
+      "step": 241
+    },
+    {
+      "epoch": 0.979757085020243,
+      "grad_norm": 0.6203464110417592,
+      "learning_rate": 3.889693142424384e-05,
+      "loss": 0.6723,
+      "step": 242
+    },
+    {
+      "epoch": 0.9838056680161943,
+      "grad_norm": 0.5009564747877816,
+      "learning_rate": 3.8878333606643405e-05,
+      "loss": 0.6729,
+      "step": 243
+    },
+    {
+      "epoch": 0.9878542510121457,
+      "grad_norm": 0.6229003889695709,
+      "learning_rate": 3.885958483839049e-05,
+      "loss": 0.6708,
+      "step": 244
+    },
+    {
+      "epoch": 0.9919028340080972,
+      "grad_norm": 0.6362077222506495,
+      "learning_rate": 3.8840685269399786e-05,
+      "loss": 0.6706,
+      "step": 245
+    },
+    {
+      "epoch": 0.9959514170040485,
+      "grad_norm": 0.48569689078193795,
+      "learning_rate": 3.882163505079171e-05,
+      "loss": 0.6779,
+      "step": 246
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.6471688495613968,
+      "learning_rate": 3.880243433489132e-05,
+      "loss": 0.6484,
+      "step": 247
+    },
+    {
+      "epoch": 1.0040485829959513,
+      "grad_norm": 0.6321285223720292,
+      "learning_rate": 3.878308327522702e-05,
+      "loss": 0.5761,
+      "step": 248
+    },
+    {
+      "epoch": 1.008097165991903,
+      "grad_norm": 0.7350188410950859,
+      "learning_rate": 3.876358202652936e-05,
+      "loss": 0.5664,
+      "step": 249
+    },
+    {
+      "epoch": 1.0121457489878543,
+      "grad_norm": 0.7520079611322513,
+      "learning_rate": 3.8743930744729835e-05,
+      "loss": 0.5681,
+      "step": 250
+    },
+    {
+      "epoch": 1.0161943319838056,
+      "grad_norm": 0.562600027128556,
+      "learning_rate": 3.8724129586959535e-05,
+      "loss": 0.5563,
+      "step": 251
+    },
+    {
+      "epoch": 1.0202429149797572,
+      "grad_norm": 0.684760191702537,
+      "learning_rate": 3.870417871154801e-05,
+      "loss": 0.5491,
+      "step": 252
+    },
+    {
+      "epoch": 1.0242914979757085,
+      "grad_norm": 0.7142125843136301,
+      "learning_rate": 3.868407827802192e-05,
+      "loss": 0.5436,
+      "step": 253
+    },
+    {
+      "epoch": 1.0283400809716599,
+      "grad_norm": 0.5692097170258332,
+      "learning_rate": 3.8663828447103804e-05,
+      "loss": 0.5442,
+      "step": 254
+    },
+    {
+      "epoch": 1.0323886639676114,
+      "grad_norm": 0.6043303102780874,
+      "learning_rate": 3.8643429380710765e-05,
+      "loss": 0.5618,
+      "step": 255
+    },
+    {
+      "epoch": 1.0364372469635628,
+      "grad_norm": 0.7689001289560873,
+      "learning_rate": 3.862288124195319e-05,
+      "loss": 0.5714,
+      "step": 256
+    },
+    {
+      "epoch": 1.040485829959514,
+      "grad_norm": 0.5418531454956804,
+      "learning_rate": 3.8602184195133464e-05,
+      "loss": 0.5444,
+      "step": 257
+    },
+    {
+      "epoch": 1.0445344129554657,
+      "grad_norm": 0.7205281149179622,
+      "learning_rate": 3.8581338405744604e-05,
+      "loss": 0.5486,
+      "step": 258
+    },
+    {
+      "epoch": 1.048582995951417,
+      "grad_norm": 0.5893886019072904,
+      "learning_rate": 3.8560344040468996e-05,
+      "loss": 0.5543,
+      "step": 259
+    },
+    {
+      "epoch": 1.0526315789473684,
+      "grad_norm": 0.6418224846865742,
+      "learning_rate": 3.853920126717703e-05,
+      "loss": 0.562,
+      "step": 260
+    },
+    {
+      "epoch": 1.05668016194332,
+      "grad_norm": 0.6515578931292385,
+      "learning_rate": 3.851791025492574e-05,
+      "loss": 0.566,
+      "step": 261
+    },
+    {
+      "epoch": 1.0607287449392713,
+      "grad_norm": 0.6060223859946785,
+      "learning_rate": 3.8496471173957515e-05,
+      "loss": 0.5613,
+      "step": 262
+    },
+    {
+      "epoch": 1.0647773279352226,
+      "grad_norm": 0.6809890468793064,
+      "learning_rate": 3.847488419569866e-05,
+      "loss": 0.5518,
+      "step": 263
+    },
+    {
+      "epoch": 1.0688259109311742,
+      "grad_norm": 0.6852694352286175,
+      "learning_rate": 3.8453149492758077e-05,
+      "loss": 0.5596,
+      "step": 264
+    },
+    {
+      "epoch": 1.0728744939271255,
+      "grad_norm": 0.690200278316907,
+      "learning_rate": 3.8431267238925886e-05,
+      "loss": 0.5612,
+      "step": 265
+    },
+    {
+      "epoch": 1.0769230769230769,
+      "grad_norm": 0.583526547396153,
+      "learning_rate": 3.840923760917199e-05,
+      "loss": 0.5582,
+      "step": 266
+    },
+    {
+      "epoch": 1.0809716599190284,
+      "grad_norm": 0.6605252519077704,
+      "learning_rate": 3.838706077964473e-05,
+      "loss": 0.5667,
+      "step": 267
+    },
+    {
+      "epoch": 1.0850202429149798,
+      "grad_norm": 0.6233322821532359,
+      "learning_rate": 3.836473692766944e-05,
+      "loss": 0.5688,
+      "step": 268
+    },
+    {
+      "epoch": 1.0890688259109311,
+      "grad_norm": 0.5949218557177777,
+      "learning_rate": 3.834226623174705e-05,
+      "loss": 0.54,
+      "step": 269
+    },
+    {
+      "epoch": 1.0931174089068827,
+      "grad_norm": 0.6317627543838211,
+      "learning_rate": 3.831964887155264e-05,
+      "loss": 0.5716,
+      "step": 270
+    },
+    {
+      "epoch": 1.097165991902834,
+      "grad_norm": 0.7616836590922457,
+      "learning_rate": 3.829688502793403e-05,
+      "loss": 0.57,
+      "step": 271
+    },
+    {
+      "epoch": 1.1012145748987854,
+      "grad_norm": 0.5347677720244529,
+      "learning_rate": 3.827397488291029e-05,
+      "loss": 0.5578,
+      "step": 272
+    },
+    {
+      "epoch": 1.1052631578947367,
+      "grad_norm": 0.7266467163409042,
+      "learning_rate": 3.8250918619670343e-05,
+      "loss": 0.5427,
+      "step": 273
+    },
+    {
+      "epoch": 1.1093117408906883,
+      "grad_norm": 0.5442101912145858,
+      "learning_rate": 3.822771642257145e-05,
+      "loss": 0.5752,
+      "step": 274
+    },
+    {
+      "epoch": 1.1133603238866396,
+      "grad_norm": 0.6483344455603085,
+      "learning_rate": 3.820436847713776e-05,
+      "loss": 0.5441,
+      "step": 275
+    },
+    {
+      "epoch": 1.117408906882591,
+      "grad_norm": 0.6416749758081938,
+      "learning_rate": 3.8180874970058826e-05,
+      "loss": 0.5451,
+      "step": 276
+    },
+    {
+      "epoch": 1.1214574898785425,
+      "grad_norm": 0.6198946910783929,
+      "learning_rate": 3.8157236089188103e-05,
+      "loss": 0.5596,
+      "step": 277
+    },
+    {
+      "epoch": 1.125506072874494,
+      "grad_norm": 0.665177127225933,
+      "learning_rate": 3.813345202354145e-05,
+      "loss": 0.5501,
+      "step": 278
+    },
+    {
+      "epoch": 1.1295546558704452,
+      "grad_norm": 0.5135613768570633,
+      "learning_rate": 3.810952296329563e-05,
+      "loss": 0.5454,
+      "step": 279
+    },
+    {
+      "epoch": 1.1336032388663968,
+      "grad_norm": 0.6882857435933701,
+      "learning_rate": 3.808544909978676e-05,
+      "loss": 0.555,
+      "step": 280
+    },
+    {
+      "epoch": 1.1376518218623481,
+      "grad_norm": 0.5014304837648347,
+      "learning_rate": 3.806123062550882e-05,
+      "loss": 0.5714,
+      "step": 281
+    },
+    {
+      "epoch": 1.1417004048582995,
+      "grad_norm": 0.6655201782381905,
+      "learning_rate": 3.803686773411208e-05,
+      "loss": 0.5592,
+      "step": 282
+    },
+    {
+      "epoch": 1.145748987854251,
+      "grad_norm": 0.5544561834359856,
+      "learning_rate": 3.801236062040158e-05,
+      "loss": 0.5473,
+      "step": 283
+    },
+    {
+      "epoch": 1.1497975708502024,
+      "grad_norm": 0.5151567198313756,
+      "learning_rate": 3.798770948033553e-05,
+      "loss": 0.561,
+      "step": 284
+    },
+    {
+      "epoch": 1.1538461538461537,
+      "grad_norm": 0.6512798731911204,
+      "learning_rate": 3.796291451102381e-05,
+      "loss": 0.5439,
+      "step": 285
+    },
+    {
+      "epoch": 1.1578947368421053,
+      "grad_norm": 0.5271788488417714,
+      "learning_rate": 3.793797591072631e-05,
+      "loss": 0.5345,
+      "step": 286
+    },
+    {
+      "epoch": 1.1619433198380567,
+      "grad_norm": 0.704724941440883,
+      "learning_rate": 3.791289387885144e-05,
+      "loss": 0.5834,
+      "step": 287
+    },
+    {
+      "epoch": 1.165991902834008,
+      "grad_norm": 0.9346575152713579,
+      "learning_rate": 3.788766861595443e-05,
+      "loss": 0.5598,
+      "step": 288
+    },
+    {
+      "epoch": 1.1700404858299596,
+      "grad_norm": 0.677415878323858,
+      "learning_rate": 3.7862300323735835e-05,
+      "loss": 0.5568,
+      "step": 289
+    },
+    {
+      "epoch": 1.174089068825911,
+      "grad_norm": 0.5787191744112385,
+      "learning_rate": 3.783678920503983e-05,
+      "loss": 0.5538,
+      "step": 290
+    },
+    {
+      "epoch": 1.1781376518218623,
+      "grad_norm": 0.5472918454114685,
+      "learning_rate": 3.781113546385265e-05,
+      "loss": 0.5457,
+      "step": 291
+    },
+    {
+      "epoch": 1.1821862348178138,
+      "grad_norm": 0.5451977067173039,
+      "learning_rate": 3.7785339305300916e-05,
+      "loss": 0.5546,
+      "step": 292
+    },
+    {
+      "epoch": 1.1862348178137652,
+      "grad_norm": 0.5700810426580409,
+      "learning_rate": 3.7759400935650025e-05,
+      "loss": 0.561,
+      "step": 293
+    },
+    {
+      "epoch": 1.1902834008097165,
+      "grad_norm": 0.5849908142762105,
+      "learning_rate": 3.77333205623025e-05,
+      "loss": 0.5554,
+      "step": 294
+    },
+    {
+      "epoch": 1.194331983805668,
+      "grad_norm": 0.5236495932350917,
+      "learning_rate": 3.7707098393796295e-05,
+      "loss": 0.5481,
+      "step": 295
+    },
+    {
+      "epoch": 1.1983805668016194,
+      "grad_norm": 0.5797940557773889,
+      "learning_rate": 3.7680734639803185e-05,
+      "loss": 0.556,
+      "step": 296
+    },
+    {
+      "epoch": 1.2024291497975708,
+      "grad_norm": 0.628616010436192,
+      "learning_rate": 3.765422951112704e-05,
+      "loss": 0.5658,
+      "step": 297
+    },
+    {
+      "epoch": 1.2064777327935223,
+      "grad_norm": 0.527183639374829,
+      "learning_rate": 3.762758321970216e-05,
+      "loss": 0.538,
+      "step": 298
+    },
+    {
+      "epoch": 1.2105263157894737,
+      "grad_norm": 0.593286004439089,
+      "learning_rate": 3.7600795978591584e-05,
+      "loss": 0.5408,
+      "step": 299
+    },
+    {
+      "epoch": 1.214574898785425,
+      "grad_norm": 0.5210669249230252,
+      "learning_rate": 3.757386800198538e-05,
+      "loss": 0.5794,
+      "step": 300
+    },
+    {
+      "epoch": 1.2186234817813766,
+      "grad_norm": 0.5261058383289915,
+      "learning_rate": 3.7546799505198925e-05,
+      "loss": 0.5596,
+      "step": 301
+    },
+    {
+      "epoch": 1.222672064777328,
+      "grad_norm": 0.4997364671316612,
+      "learning_rate": 3.7519590704671197e-05,
+      "loss": 0.5687,
+      "step": 302
+    },
+    {
+      "epoch": 1.2267206477732793,
+      "grad_norm": 0.49249946195631084,
+      "learning_rate": 3.749224181796305e-05,
+      "loss": 0.5718,
+      "step": 303
+    },
+    {
+      "epoch": 1.2307692307692308,
+      "grad_norm": 0.5358939012404814,
+      "learning_rate": 3.746475306375545e-05,
+      "loss": 0.5429,
+      "step": 304
+    },
+    {
+      "epoch": 1.2348178137651822,
+      "grad_norm": 0.47645522837330945,
+      "learning_rate": 3.743712466184774e-05,
+      "loss": 0.5595,
+      "step": 305
+    },
+    {
+      "epoch": 1.2388663967611335,
+      "grad_norm": 0.5068986457682251,
+      "learning_rate": 3.7409356833155885e-05,
+      "loss": 0.5635,
+      "step": 306
+    },
+    {
+      "epoch": 1.242914979757085,
+      "grad_norm": 0.47799694237549034,
+      "learning_rate": 3.73814497997107e-05,
+      "loss": 0.5537,
+      "step": 307
+    },
+    {
+      "epoch": 1.2469635627530364,
+      "grad_norm": 0.6265149991949317,
+      "learning_rate": 3.7353403784656095e-05,
+      "loss": 0.5493,
+      "step": 308
+    },
+    {
+      "epoch": 1.2510121457489878,
+      "grad_norm": 0.6707373771787015,
+      "learning_rate": 3.732521901224724e-05,
+      "loss": 0.554,
+      "step": 309
+    },
+    {
+      "epoch": 1.2550607287449393,
+      "grad_norm": 0.4965896370700255,
+      "learning_rate": 3.7296895707848844e-05,
+      "loss": 0.5641,
+      "step": 310
+    },
+    {
+      "epoch": 1.2591093117408907,
+      "grad_norm": 0.7052911533562827,
+      "learning_rate": 3.7268434097933275e-05,
+      "loss": 0.5745,
+      "step": 311
+    },
+    {
+      "epoch": 1.263157894736842,
+      "grad_norm": 0.6224183057882141,
+      "learning_rate": 3.723983441007882e-05,
+      "loss": 0.5563,
+      "step": 312
+    },
+    {
+      "epoch": 1.2672064777327936,
+      "grad_norm": 0.4973646369619287,
+      "learning_rate": 3.721109687296781e-05,
+      "loss": 0.57,
+      "step": 313
+    },
+    {
+      "epoch": 1.271255060728745,
+      "grad_norm": 0.6016648918726423,
+      "learning_rate": 3.718222171638484e-05,
+      "loss": 0.5645,
+      "step": 314
+    },
+    {
+      "epoch": 1.2753036437246963,
+      "grad_norm": 0.5994510871761279,
+      "learning_rate": 3.715320917121488e-05,
+      "loss": 0.5546,
+      "step": 315
+    },
+    {
+      "epoch": 1.2793522267206479,
+      "grad_norm": 0.5245040353872992,
+      "learning_rate": 3.712405946944149e-05,
+      "loss": 0.5743,
+      "step": 316
+    },
+    {
+      "epoch": 1.2834008097165992,
+      "grad_norm": 0.5983754071261111,
+      "learning_rate": 3.709477284414488e-05,
+      "loss": 0.5841,
+      "step": 317
+    },
+    {
+      "epoch": 1.2874493927125505,
+      "grad_norm": 0.6381958594862511,
+      "learning_rate": 3.7065349529500136e-05,
+      "loss": 0.5689,
+      "step": 318
+    },
+    {
+      "epoch": 1.291497975708502,
+      "grad_norm": 0.5043852527973852,
+      "learning_rate": 3.7035789760775305e-05,
+      "loss": 0.5555,
+      "step": 319
+    },
+    {
+      "epoch": 1.2955465587044535,
+      "grad_norm": 0.8004627345208959,
+      "learning_rate": 3.70060937743295e-05,
+      "loss": 0.5642,
+      "step": 320
+    },
+    {
+      "epoch": 1.2995951417004048,
+      "grad_norm": 0.46890280085686337,
+      "learning_rate": 3.6976261807611045e-05,
+      "loss": 0.5504,
+      "step": 321
+    },
+    {
+      "epoch": 1.3036437246963564,
+      "grad_norm": 0.7764587241123002,
+      "learning_rate": 3.694629409915555e-05,
+      "loss": 0.5589,
+      "step": 322
+    },
+    {
+      "epoch": 1.3076923076923077,
+      "grad_norm": 0.5471959856572183,
+      "learning_rate": 3.691619088858401e-05,
+      "loss": 0.543,
+      "step": 323
+    },
+    {
+      "epoch": 1.311740890688259,
+      "grad_norm": 0.5791922709653066,
+      "learning_rate": 3.68859524166009e-05,
+      "loss": 0.559,
+      "step": 324
+    },
+    {
+      "epoch": 1.3157894736842106,
+      "grad_norm": 0.5719971226310504,
+      "learning_rate": 3.6855578924992226e-05,
+      "loss": 0.5551,
+      "step": 325
+    },
+    {
+      "epoch": 1.319838056680162,
+      "grad_norm": 0.5127700993108518,
+      "learning_rate": 3.6825070656623626e-05,
+      "loss": 0.5662,
+      "step": 326
+    },
+    {
+      "epoch": 1.3238866396761133,
+      "grad_norm": 0.5077416499689852,
+      "learning_rate": 3.67944278554384e-05,
+      "loss": 0.5582,
+      "step": 327
+    },
+    {
+      "epoch": 1.3279352226720649,
+      "grad_norm": 0.5803305850332873,
+      "learning_rate": 3.676365076645557e-05,
+      "loss": 0.5522,
+      "step": 328
+    },
+    {
+      "epoch": 1.3319838056680162,
+      "grad_norm": 0.618085609903998,
+      "learning_rate": 3.673273963576791e-05,
+      "loss": 0.5464,
+      "step": 329
+    },
+    {
+      "epoch": 1.3360323886639676,
+      "grad_norm": 0.5651449558851476,
+      "learning_rate": 3.670169471054001e-05,
+      "loss": 0.5419,
+      "step": 330
+    },
+    {
+      "epoch": 1.3400809716599191,
+      "grad_norm": 0.6101561983196607,
+      "learning_rate": 3.6670516239006254e-05,
+      "loss": 0.5667,
+      "step": 331
+    },
+    {
+      "epoch": 1.3441295546558705,
+      "grad_norm": 0.5632584006028271,
+      "learning_rate": 3.663920447046888e-05,
+      "loss": 0.5457,
+      "step": 332
+    },
+    {
+      "epoch": 1.3481781376518218,
+      "grad_norm": 0.7030442280117203,
+      "learning_rate": 3.6607759655295954e-05,
+      "loss": 0.5633,
+      "step": 333
+    },
+    {
+      "epoch": 1.3522267206477734,
+      "grad_norm": 0.6322463933566796,
+      "learning_rate": 3.657618204491937e-05,
+      "loss": 0.5621,
+      "step": 334
+    },
+    {
+      "epoch": 1.3562753036437247,
+      "grad_norm": 0.6577627568985589,
+      "learning_rate": 3.6544471891832875e-05,
+      "loss": 0.5528,
+      "step": 335
+    },
+    {
+      "epoch": 1.360323886639676,
+      "grad_norm": 0.6131766420276452,
+      "learning_rate": 3.6512629449589983e-05,
+      "loss": 0.5702,
+      "step": 336
+    },
+    {
+      "epoch": 1.3643724696356276,
+      "grad_norm": 0.6427954716669986,
+      "learning_rate": 3.6480654972802045e-05,
+      "loss": 0.5831,
+      "step": 337
+    },
+    {
+      "epoch": 1.368421052631579,
+      "grad_norm": 0.5553100265313777,
+      "learning_rate": 3.644854871713611e-05,
+      "loss": 0.5586,
+      "step": 338
+    },
+    {
+      "epoch": 1.3724696356275303,
+      "grad_norm": 0.6913196832765729,
+      "learning_rate": 3.6416310939312924e-05,
+      "loss": 0.5604,
+      "step": 339
+    },
+    {
+      "epoch": 1.376518218623482,
+      "grad_norm": 0.5207685624593228,
+      "learning_rate": 3.638394189710493e-05,
+      "loss": 0.5619,
+      "step": 340
+    },
+    {
+      "epoch": 1.3805668016194332,
+      "grad_norm": 0.6143330619949505,
+      "learning_rate": 3.635144184933412e-05,
+      "loss": 0.5733,
+      "step": 341
+    },
+    {
+      "epoch": 1.3846153846153846,
+      "grad_norm": 0.641812434299892,
+      "learning_rate": 3.6318811055869994e-05,
+      "loss": 0.5746,
+      "step": 342
+    },
+    {
+      "epoch": 1.3886639676113361,
+      "grad_norm": 0.619399895159573,
+      "learning_rate": 3.628604977762752e-05,
+      "loss": 0.5677,
+      "step": 343
+    },
+    {
+      "epoch": 1.3927125506072875,
+      "grad_norm": 0.6096290431401343,
+      "learning_rate": 3.6253158276565004e-05,
+      "loss": 0.5617,
+      "step": 344
+    },
+    {
+      "epoch": 1.3967611336032388,
+      "grad_norm": 0.6906327496768488,
+      "learning_rate": 3.6220136815682016e-05,
+      "loss": 0.5629,
+      "step": 345
+    },
+    {
+      "epoch": 1.4008097165991904,
+      "grad_norm": 0.5232606579797187,
+      "learning_rate": 3.618698565901727e-05,
+      "loss": 0.5754,
+      "step": 346
+    },
+    {
+      "epoch": 1.4048582995951417,
+      "grad_norm": 0.6937489660050271,
+      "learning_rate": 3.615370507164654e-05,
+      "loss": 0.5712,
+      "step": 347
+    },
+    {
+      "epoch": 1.408906882591093,
+      "grad_norm": 0.5953378007476958,
+      "learning_rate": 3.61202953196805e-05,
+      "loss": 0.5632,
+      "step": 348
+    },
+    {
+      "epoch": 1.4129554655870447,
+      "grad_norm": 0.5225756955810039,
+      "learning_rate": 3.6086756670262654e-05,
+      "loss": 0.5722,
+      "step": 349
+    },
+    {
+      "epoch": 1.417004048582996,
+      "grad_norm": 0.5718608045826205,
+      "learning_rate": 3.6053089391567135e-05,
+      "loss": 0.5608,
+      "step": 350
+    },
+    {
+      "epoch": 1.4210526315789473,
+      "grad_norm": 0.46840099975553934,
+      "learning_rate": 3.601929375279662e-05,
+      "loss": 0.5628,
+      "step": 351
+    },
+    {
+      "epoch": 1.425101214574899,
+      "grad_norm": 0.5946629225693222,
+      "learning_rate": 3.598537002418012e-05,
+      "loss": 0.5518,
+      "step": 352
+    },
+    {
+      "epoch": 1.4291497975708503,
+      "grad_norm": 0.4995178001521157,
+      "learning_rate": 3.5951318476970885e-05,
+      "loss": 0.5713,
+      "step": 353
+    },
+    {
+      "epoch": 1.4331983805668016,
+      "grad_norm": 0.5166337019728015,
+      "learning_rate": 3.591713938344416e-05,
+      "loss": 0.545,
+      "step": 354
+    },
+    {
+      "epoch": 1.4372469635627532,
+      "grad_norm": 0.6182807584629387,
+      "learning_rate": 3.588283301689507e-05,
+      "loss": 0.5536,
+      "step": 355
+    },
+    {
+      "epoch": 1.4412955465587045,
+      "grad_norm": 0.4518075769317617,
+      "learning_rate": 3.5848399651636424e-05,
+      "loss": 0.5635,
+      "step": 356
+    },
+    {
+      "epoch": 1.4453441295546559,
+      "grad_norm": 0.5582588691872968,
+      "learning_rate": 3.5813839562996473e-05,
+      "loss": 0.5658,
+      "step": 357
+    },
+    {
+      "epoch": 1.4493927125506074,
+      "grad_norm": 0.5253932893179281,
+      "learning_rate": 3.5779153027316775e-05,
+      "loss": 0.5746,
+      "step": 358
+    },
+    {
+      "epoch": 1.4534412955465588,
+      "grad_norm": 0.458207231001632,
+      "learning_rate": 3.574434032194994e-05,
+      "loss": 0.545,
+      "step": 359
+    },
+    {
+      "epoch": 1.45748987854251,
+      "grad_norm": 0.5412313342992535,
+      "learning_rate": 3.5709401725257436e-05,
+      "loss": 0.5681,
+      "step": 360
+    },
+    {
+      "epoch": 1.4615384615384617,
+      "grad_norm": 0.49335402834134756,
+      "learning_rate": 3.5674337516607346e-05,
+      "loss": 0.5697,
+      "step": 361
+    },
+    {
+      "epoch": 1.465587044534413,
+      "grad_norm": 0.4966591222613841,
+      "learning_rate": 3.563914797637216e-05,
+      "loss": 0.5603,
+      "step": 362
+    },
+    {
+      "epoch": 1.4696356275303644,
+      "grad_norm": 0.6029090367382667,
+      "learning_rate": 3.560383338592649e-05,
+      "loss": 0.556,
+      "step": 363
+    },
+    {
+      "epoch": 1.4736842105263157,
+      "grad_norm": 0.4885502496406368,
+      "learning_rate": 3.556839402764487e-05,
+      "loss": 0.5621,
+      "step": 364
+    },
+    {
+      "epoch": 1.4777327935222673,
+      "grad_norm": 0.495076872773626,
+      "learning_rate": 3.553283018489946e-05,
+      "loss": 0.5427,
+      "step": 365
+    },
+    {
+      "epoch": 1.4817813765182186,
+      "grad_norm": 0.5738379589593847,
+      "learning_rate": 3.54971421420578e-05,
+      "loss": 0.5709,
+      "step": 366
+    },
+    {
+      "epoch": 1.48582995951417,
+      "grad_norm": 0.5070870252979219,
+      "learning_rate": 3.5461330184480524e-05,
+      "loss": 0.5553,
+      "step": 367
+    },
+    {
+      "epoch": 1.4898785425101215,
+      "grad_norm": 0.5884067409481301,
+      "learning_rate": 3.54253945985191e-05,
+      "loss": 0.5596,
+      "step": 368
+    },
+    {
+      "epoch": 1.4939271255060729,
+      "grad_norm": 0.5038173358497086,
+      "learning_rate": 3.5389335671513524e-05,
+      "loss": 0.5734,
+      "step": 369
+    },
+    {
+      "epoch": 1.4979757085020242,
+      "grad_norm": 0.45356245765532305,
+      "learning_rate": 3.5353153691789986e-05,
+      "loss": 0.5639,
+      "step": 370
+    },
+    {
+      "epoch": 1.5020242914979756,
+      "grad_norm": 0.5131292043305852,
+      "learning_rate": 3.531684894865866e-05,
+      "loss": 0.5665,
+      "step": 371
+    },
+    {
+      "epoch": 1.5060728744939271,
+      "grad_norm": 0.4718542926663491,
+      "learning_rate": 3.528042173241128e-05,
+      "loss": 0.5691,
+      "step": 372
+    },
+    {
+      "epoch": 1.5101214574898787,
+      "grad_norm": 0.4642393442544735,
+      "learning_rate": 3.524387233431892e-05,
+      "loss": 0.5593,
+      "step": 373
+    },
+    {
+      "epoch": 1.5141700404858298,
+      "grad_norm": 0.4825852286950471,
+      "learning_rate": 3.520720104662958e-05,
+      "loss": 0.5747,
+      "step": 374
+    },
+    {
+      "epoch": 1.5182186234817814,
+      "grad_norm": 0.45691902423271663,
+      "learning_rate": 3.5170408162565904e-05,
+      "loss": 0.5637,
+      "step": 375
+    },
+    {
+      "epoch": 1.522267206477733,
+      "grad_norm": 0.4988027783212776,
+      "learning_rate": 3.513349397632281e-05,
+      "loss": 0.5562,
+      "step": 376
+    },
+    {
+      "epoch": 1.526315789473684,
+      "grad_norm": 0.49915524196775585,
+      "learning_rate": 3.5096458783065145e-05,
+      "loss": 0.5669,
+      "step": 377
+    },
+    {
+      "epoch": 1.5303643724696356,
+      "grad_norm": 0.4828993766268075,
+      "learning_rate": 3.505930287892533e-05,
+      "loss": 0.5677,
+      "step": 378
+    },
+    {
+      "epoch": 1.5344129554655872,
+      "grad_norm": 0.5554856770442007,
+      "learning_rate": 3.502202656100099e-05,
+      "loss": 0.5711,
+      "step": 379
+    },
+    {
+      "epoch": 1.5384615384615383,
+      "grad_norm": 0.49959722720034355,
+      "learning_rate": 3.498463012735256e-05,
+      "loss": 0.5712,
+      "step": 380
+    },
+    {
+      "epoch": 1.54251012145749,
+      "grad_norm": 0.5059684653386233,
+      "learning_rate": 3.494711387700094e-05,
+      "loss": 0.5603,
+      "step": 381
+    },
+    {
+      "epoch": 1.5465587044534415,
+      "grad_norm": 0.5118893174716055,
+      "learning_rate": 3.490947810992505e-05,
+      "loss": 0.5692,
+      "step": 382
+    },
+    {
+      "epoch": 1.5506072874493926,
+      "grad_norm": 0.5193570422649706,
+      "learning_rate": 3.48717231270595e-05,
+      "loss": 0.5568,
+      "step": 383
+    },
+    {
+      "epoch": 1.5546558704453441,
+      "grad_norm": 0.4716390753584926,
+      "learning_rate": 3.483384923029211e-05,
+      "loss": 0.5816,
+      "step": 384
+    },
+    {
+      "epoch": 1.5587044534412957,
+      "grad_norm": 0.45939754483790024,
+      "learning_rate": 3.479585672246156e-05,
+      "loss": 0.574,
+      "step": 385
+    },
+    {
+      "epoch": 1.5627530364372468,
+      "grad_norm": 0.4916759919268338,
+      "learning_rate": 3.4757745907354924e-05,
+      "loss": 0.561,
+      "step": 386
+    },
+    {
+      "epoch": 1.5668016194331984,
+      "grad_norm": 0.47145375340020734,
+      "learning_rate": 3.471951708970527e-05,
+      "loss": 0.5529,
+      "step": 387
+    },
+    {
+      "epoch": 1.5708502024291497,
+      "grad_norm": 0.45788164644693946,
+      "learning_rate": 3.468117057518921e-05,
+      "loss": 0.5533,
+      "step": 388
+    },
+    {
+      "epoch": 1.574898785425101,
+      "grad_norm": 0.5576559793292503,
+      "learning_rate": 3.4642706670424426e-05,
+      "loss": 0.5861,
+      "step": 389
+    },
+    {
+      "epoch": 1.5789473684210527,
+      "grad_norm": 0.4331968688562831,
+      "learning_rate": 3.460412568296731e-05,
+      "loss": 0.559,
+      "step": 390
+    },
+    {
+      "epoch": 1.582995951417004,
+      "grad_norm": 0.5301188337415963,
+      "learning_rate": 3.456542792131039e-05,
+      "loss": 0.5597,
+      "step": 391
+    },
+    {
+      "epoch": 1.5870445344129553,
+      "grad_norm": 0.5188409543632893,
+      "learning_rate": 3.4526613694879936e-05,
+      "loss": 0.5639,
+      "step": 392
+    },
+    {
+      "epoch": 1.591093117408907,
+      "grad_norm": 0.5622449452717668,
+      "learning_rate": 3.448768331403348e-05,
+      "loss": 0.5662,
+      "step": 393
+    },
+    {
+      "epoch": 1.5951417004048583,
+      "grad_norm": 0.49324509403248196,
+      "learning_rate": 3.44486370900573e-05,
+      "loss": 0.5611,
+      "step": 394
+    },
+    {
+      "epoch": 1.5991902834008096,
+      "grad_norm": 0.5808507562474285,
+      "learning_rate": 3.440947533516396e-05,
+      "loss": 0.572,
+      "step": 395
+    },
+    {
+      "epoch": 1.6032388663967612,
+      "grad_norm": 0.4570644722508165,
+      "learning_rate": 3.4370198362489816e-05,
+      "loss": 0.5653,
+      "step": 396
+    },
+    {
+      "epoch": 1.6072874493927125,
+      "grad_norm": 0.506494178173887,
+      "learning_rate": 3.43308064860925e-05,
+      "loss": 0.5542,
+      "step": 397
+    },
+    {
+      "epoch": 1.6113360323886639,
+      "grad_norm": 0.5749175284596194,
+      "learning_rate": 3.4291300020948393e-05,
+      "loss": 0.5597,
+      "step": 398
+    },
+    {
+      "epoch": 1.6153846153846154,
+      "grad_norm": 0.4991905117366417,
+      "learning_rate": 3.425167928295015e-05,
+      "loss": 0.5481,
+      "step": 399
+    },
+    {
+      "epoch": 1.6194331983805668,
+      "grad_norm": 0.5839442241952063,
+      "learning_rate": 3.421194458890411e-05,
+      "loss": 0.5608,
+      "step": 400
+    },
+    {
+      "epoch": 1.623481781376518,
+      "grad_norm": 0.5334217711148604,
+      "learning_rate": 3.4172096256527865e-05,
+      "loss": 0.5779,
+      "step": 401
+    },
+    {
+      "epoch": 1.6275303643724697,
+      "grad_norm": 0.47220992624685,
+      "learning_rate": 3.4132134604447595e-05,
+      "loss": 0.55,
+      "step": 402
+    },
+    {
+      "epoch": 1.631578947368421,
+      "grad_norm": 0.5471105557361146,
+      "learning_rate": 3.409205995219561e-05,
+      "loss": 0.5637,
+      "step": 403
+    },
+    {
+      "epoch": 1.6356275303643724,
+      "grad_norm": 0.5587864185135106,
+      "learning_rate": 3.4051872620207765e-05,
+      "loss": 0.5688,
+      "step": 404
+    },
+    {
+      "epoch": 1.639676113360324,
+      "grad_norm": 0.5572734636073006,
+      "learning_rate": 3.40115729298209e-05,
+      "loss": 0.5634,
+      "step": 405
+    },
+    {
+      "epoch": 1.6437246963562753,
+      "grad_norm": 0.6602340217458353,
+      "learning_rate": 3.397116120327025e-05,
+      "loss": 0.575,
+      "step": 406
+    },
+    {
+      "epoch": 1.6477732793522266,
+      "grad_norm": 0.5086096826062502,
+      "learning_rate": 3.3930637763686916e-05,
+      "loss": 0.5563,
+      "step": 407
+    },
+    {
+      "epoch": 1.6518218623481782,
+      "grad_norm": 0.5798683696141924,
+      "learning_rate": 3.389000293509524e-05,
+      "loss": 0.5644,
+      "step": 408
+    },
+    {
+      "epoch": 1.6558704453441295,
+      "grad_norm": 0.5174979984030831,
+      "learning_rate": 3.3849257042410225e-05,
+      "loss": 0.5729,
+      "step": 409
+    },
+    {
+      "epoch": 1.6599190283400809,
+      "grad_norm": 0.48309599594435754,
+      "learning_rate": 3.380840041143494e-05,
+      "loss": 0.5446,
+      "step": 410
+    },
+    {
+      "epoch": 1.6639676113360324,
+      "grad_norm": 0.5845961646659847,
+      "learning_rate": 3.376743336885792e-05,
+      "loss": 0.5741,
+      "step": 411
+    },
+    {
+      "epoch": 1.6680161943319838,
+      "grad_norm": 0.5819994654178141,
+      "learning_rate": 3.372635624225055e-05,
+      "loss": 0.5819,
+      "step": 412
+    },
+    {
+      "epoch": 1.6720647773279351,
+      "grad_norm": 0.5586045820956032,
+      "learning_rate": 3.3685169360064445e-05,
+      "loss": 0.5542,
+      "step": 413
+    },
+    {
+      "epoch": 1.6761133603238867,
+      "grad_norm": 0.6459011530939155,
+      "learning_rate": 3.3643873051628804e-05,
+      "loss": 0.5659,
+      "step": 414
+    },
+    {
+      "epoch": 1.680161943319838,
+      "grad_norm": 0.5493640895962423,
+      "learning_rate": 3.3602467647147814e-05,
+      "loss": 0.5698,
+      "step": 415
+    },
+    {
+      "epoch": 1.6842105263157894,
+      "grad_norm": 0.6391071199856935,
+      "learning_rate": 3.356095347769799e-05,
+      "loss": 0.5598,
+      "step": 416
+    },
+    {
+      "epoch": 1.688259109311741,
+      "grad_norm": 0.5550182459439932,
+      "learning_rate": 3.351933087522552e-05,
+      "loss": 0.5643,
+      "step": 417
+    },
+    {
+      "epoch": 1.6923076923076923,
+      "grad_norm": 0.5331427413278176,
+      "learning_rate": 3.3477600172543626e-05,
+      "loss": 0.5534,
+      "step": 418
+    },
+    {
+      "epoch": 1.6963562753036436,
+      "grad_norm": 0.514137415274564,
+      "learning_rate": 3.3435761703329894e-05,
+      "loss": 0.5553,
+      "step": 419
+    },
+    {
+      "epoch": 1.7004048582995952,
+      "grad_norm": 0.5166077635161821,
+      "learning_rate": 3.3393815802123606e-05,
+      "loss": 0.5567,
+      "step": 420
+    },
+    {
+      "epoch": 1.7044534412955465,
+      "grad_norm": 0.49170628368586694,
+      "learning_rate": 3.335176280432308e-05,
+      "loss": 0.5434,
+      "step": 421
+    },
+    {
+      "epoch": 1.708502024291498,
+      "grad_norm": 0.5045128618138327,
+      "learning_rate": 3.330960304618295e-05,
+      "loss": 0.5587,
+      "step": 422
+    },
+    {
+      "epoch": 1.7125506072874495,
+      "grad_norm": 0.4856087232474049,
+      "learning_rate": 3.326733686481153e-05,
+      "loss": 0.5657,
+      "step": 423
+    },
+    {
+      "epoch": 1.7165991902834008,
+      "grad_norm": 0.48075210394066653,
+      "learning_rate": 3.3224964598168084e-05,
+      "loss": 0.5644,
+      "step": 424
+    },
+    {
+      "epoch": 1.7206477732793521,
+      "grad_norm": 0.47411002399206076,
+      "learning_rate": 3.318248658506012e-05,
+      "loss": 0.5519,
+      "step": 425
+    },
+    {
+      "epoch": 1.7246963562753037,
+      "grad_norm": 0.5083887085180088,
+      "learning_rate": 3.313990316514069e-05,
+      "loss": 0.5646,
+      "step": 426
+    },
+    {
+      "epoch": 1.728744939271255,
+      "grad_norm": 0.4859558322312172,
+      "learning_rate": 3.309721467890571e-05,
+      "loss": 0.5599,
+      "step": 427
+    },
+    {
+      "epoch": 1.7327935222672064,
+      "grad_norm": 0.516200574949808,
+      "learning_rate": 3.305442146769115e-05,
+      "loss": 0.5395,
+      "step": 428
+    },
+    {
+      "epoch": 1.736842105263158,
+      "grad_norm": 0.4948914761983246,
+      "learning_rate": 3.30115238736704e-05,
+      "loss": 0.5657,
+      "step": 429
+    },
+    {
+      "epoch": 1.7408906882591093,
+      "grad_norm": 0.4500366534222148,
+      "learning_rate": 3.296852223985148e-05,
+      "loss": 0.5466,
+      "step": 430
+    },
+    {
+      "epoch": 1.7449392712550607,
+      "grad_norm": 0.4671027812254386,
+      "learning_rate": 3.2925416910074315e-05,
+      "loss": 0.5604,
+      "step": 431
+    },
+    {
+      "epoch": 1.7489878542510122,
+      "grad_norm": 0.4735953795989333,
+      "learning_rate": 3.288220822900796e-05,
+      "loss": 0.5728,
+      "step": 432
+    },
+    {
+      "epoch": 1.7530364372469636,
+      "grad_norm": 0.46945777687143375,
+      "learning_rate": 3.283889654214788e-05,
+      "loss": 0.5524,
+      "step": 433
+    },
+    {
+      "epoch": 1.757085020242915,
+      "grad_norm": 0.473204857911194,
+      "learning_rate": 3.279548219581317e-05,
+      "loss": 0.5724,
+      "step": 434
+    },
+    {
+      "epoch": 1.7611336032388665,
+      "grad_norm": 0.4998257456139568,
+      "learning_rate": 3.275196553714379e-05,
+      "loss": 0.5652,
+      "step": 435
+    },
+    {
+      "epoch": 1.7651821862348178,
+      "grad_norm": 0.49644844400066107,
+      "learning_rate": 3.270834691409778e-05,
+      "loss": 0.567,
+      "step": 436
+    },
+    {
+      "epoch": 1.7692307692307692,
+      "grad_norm": 0.4912329159986039,
+      "learning_rate": 3.266462667544849e-05,
+      "loss": 0.5708,
+      "step": 437
+    },
+    {
+      "epoch": 1.7732793522267207,
+      "grad_norm": 0.5221383226253078,
+      "learning_rate": 3.262080517078178e-05,
+      "loss": 0.5628,
+      "step": 438
+    },
+    {
+      "epoch": 1.777327935222672,
+      "grad_norm": 0.526638268759069,
+      "learning_rate": 3.257688275049323e-05,
+      "loss": 0.5678,
+      "step": 439
+    },
+    {
+      "epoch": 1.7813765182186234,
+      "grad_norm": 0.4609541790658492,
+      "learning_rate": 3.253285976578535e-05,
+      "loss": 0.5698,
+      "step": 440
+    },
+    {
+      "epoch": 1.785425101214575,
+      "grad_norm": 0.4852821432734768,
+      "learning_rate": 3.2488736568664756e-05,
+      "loss": 0.5504,
+      "step": 441
+    },
+    {
+      "epoch": 1.7894736842105263,
+      "grad_norm": 0.44610032244418074,
+      "learning_rate": 3.244451351193935e-05,
+      "loss": 0.5696,
+      "step": 442
+    },
+    {
+      "epoch": 1.7935222672064777,
+      "grad_norm": 0.6016919162596763,
+      "learning_rate": 3.240019094921551e-05,
+      "loss": 0.5529,
+      "step": 443
+    },
+    {
+      "epoch": 1.7975708502024292,
+      "grad_norm": 0.4685233940376834,
+      "learning_rate": 3.235576923489527e-05,
+      "loss": 0.5619,
+      "step": 444
+    },
+    {
+      "epoch": 1.8016194331983806,
+      "grad_norm": 0.5879394038218164,
+      "learning_rate": 3.2311248724173476e-05,
+      "loss": 0.563,
+      "step": 445
+    },
+    {
+      "epoch": 1.805668016194332,
+      "grad_norm": 0.47252491552776243,
+      "learning_rate": 3.226662977303494e-05,
+      "loss": 0.5634,
+      "step": 446
+    },
+    {
+      "epoch": 1.8097165991902835,
+      "grad_norm": 0.5485511146670748,
+      "learning_rate": 3.2221912738251626e-05,
+      "loss": 0.5612,
+      "step": 447
+    },
+    {
+      "epoch": 1.8137651821862348,
+      "grad_norm": 0.5545106109656849,
+      "learning_rate": 3.2177097977379736e-05,
+      "loss": 0.5588,
+      "step": 448
+    },
+    {
+      "epoch": 1.8178137651821862,
+      "grad_norm": 0.43753346980295654,
+      "learning_rate": 3.21321858487569e-05,
+      "loss": 0.548,
+      "step": 449
+    },
+    {
+      "epoch": 1.8218623481781377,
+      "grad_norm": 0.5560601740750067,
+      "learning_rate": 3.208717671149931e-05,
+      "loss": 0.5494,
+      "step": 450
+    },
+    {
+      "epoch": 1.825910931174089,
+      "grad_norm": 0.45576603036469393,
+      "learning_rate": 3.204207092549882e-05,
+      "loss": 0.5729,
+      "step": 451
+    },
+    {
+      "epoch": 1.8299595141700404,
+      "grad_norm": 0.4712453972015793,
+      "learning_rate": 3.19968688514201e-05,
+      "loss": 0.5733,
+      "step": 452
+    },
+    {
+      "epoch": 1.834008097165992,
+      "grad_norm": 0.4897015061788262,
+      "learning_rate": 3.195157085069774e-05,
+      "loss": 0.5473,
+      "step": 453
+    },
+    {
+      "epoch": 1.8380566801619433,
+      "grad_norm": 0.4697294508599066,
+      "learning_rate": 3.190617728553332e-05,
+      "loss": 0.5648,
+      "step": 454
+    },
+    {
+      "epoch": 1.8421052631578947,
+      "grad_norm": 0.4526872368548364,
+      "learning_rate": 3.1860688518892606e-05,
+      "loss": 0.5619,
+      "step": 455
+    },
+    {
+      "epoch": 1.8461538461538463,
+      "grad_norm": 0.5144055038084834,
+      "learning_rate": 3.181510491450254e-05,
+      "loss": 0.5773,
+      "step": 456
+    },
+    {
+      "epoch": 1.8502024291497976,
+      "grad_norm": 0.45801978233336854,
+      "learning_rate": 3.176942683684842e-05,
+      "loss": 0.5574,
+      "step": 457
+    },
+    {
+      "epoch": 1.854251012145749,
+      "grad_norm": 0.4462627878006374,
+      "learning_rate": 3.1723654651170934e-05,
+      "loss": 0.5575,
+      "step": 458
+    },
+    {
+      "epoch": 1.8582995951417005,
+      "grad_norm": 0.5175835190361675,
+      "learning_rate": 3.167778872346327e-05,
+      "loss": 0.5594,
+      "step": 459
+    },
+    {
+      "epoch": 1.8623481781376519,
+      "grad_norm": 0.4347874987620905,
+      "learning_rate": 3.163182942046815e-05,
+      "loss": 0.5538,
+      "step": 460
+    },
+    {
+      "epoch": 1.8663967611336032,
+      "grad_norm": 0.4153743621518593,
+      "learning_rate": 3.158577710967494e-05,
+      "loss": 0.5438,
+      "step": 461
+    },
+    {
+      "epoch": 1.8704453441295548,
+      "grad_norm": 0.47425444684751245,
+      "learning_rate": 3.153963215931669e-05,
+      "loss": 0.5622,
+      "step": 462
+    },
+    {
+      "epoch": 1.874493927125506,
+      "grad_norm": 0.4348614984898632,
+      "learning_rate": 3.14933949383672e-05,
+      "loss": 0.5603,
+      "step": 463
+    },
+    {
+      "epoch": 1.8785425101214575,
+      "grad_norm": 0.45338584575155766,
+      "learning_rate": 3.144706581653805e-05,
+      "loss": 0.5589,
+      "step": 464
+    },
+    {
+      "epoch": 1.882591093117409,
+      "grad_norm": 0.429344798789128,
+      "learning_rate": 3.140064516427566e-05,
+      "loss": 0.5608,
+      "step": 465
+    },
+    {
+      "epoch": 1.8866396761133604,
+      "grad_norm": 0.4953153191681917,
+      "learning_rate": 3.135413335275833e-05,
+      "loss": 0.5867,
+      "step": 466
+    },
+    {
+      "epoch": 1.8906882591093117,
+      "grad_norm": 0.467573996084903,
+      "learning_rate": 3.130753075389327e-05,
+      "loss": 0.5516,
+      "step": 467
+    },
+    {
+      "epoch": 1.8947368421052633,
+      "grad_norm": 0.4612721179014494,
+      "learning_rate": 3.1260837740313625e-05,
+      "loss": 0.5606,
+      "step": 468
+    },
+    {
+      "epoch": 1.8987854251012146,
+      "grad_norm": 0.4831944851376742,
+      "learning_rate": 3.121405468537547e-05,
+      "loss": 0.5529,
+      "step": 469
+    },
+    {
+      "epoch": 1.902834008097166,
+      "grad_norm": 0.4324128509663495,
+      "learning_rate": 3.116718196315488e-05,
+      "loss": 0.5534,
+      "step": 470
+    },
+    {
+      "epoch": 1.9068825910931175,
+      "grad_norm": 0.4645206443486525,
+      "learning_rate": 3.1120219948444894e-05,
+      "loss": 0.5744,
+      "step": 471
+    },
+    {
+      "epoch": 1.9109311740890689,
+      "grad_norm": 0.4680798459089111,
+      "learning_rate": 3.107316901675252e-05,
+      "loss": 0.576,
+      "step": 472
+    },
+    {
+      "epoch": 1.9149797570850202,
+      "grad_norm": 0.4204254832983892,
+      "learning_rate": 3.1026029544295745e-05,
+      "loss": 0.5545,
+      "step": 473
+    },
+    {
+      "epoch": 1.9190283400809718,
+      "grad_norm": 0.4822594503904122,
+      "learning_rate": 3.097880190800057e-05,
+      "loss": 0.5727,
+      "step": 474
+    },
+    {
+      "epoch": 1.9230769230769231,
+      "grad_norm": 0.4431510972746328,
+      "learning_rate": 3.093148648549788e-05,
+      "loss": 0.5582,
+      "step": 475
+    },
+    {
+      "epoch": 1.9271255060728745,
+      "grad_norm": 0.41044748947144416,
+      "learning_rate": 3.088408365512055e-05,
+      "loss": 0.5673,
+      "step": 476
+    },
+    {
+      "epoch": 1.931174089068826,
+      "grad_norm": 0.4180128801294434,
+      "learning_rate": 3.083659379590034e-05,
+      "loss": 0.5663,
+      "step": 477
+    },
+    {
+      "epoch": 1.9352226720647774,
+      "grad_norm": 0.4234882069462685,
+      "learning_rate": 3.0789017287564913e-05,
+      "loss": 0.5496,
+      "step": 478
+    },
+    {
+      "epoch": 1.9392712550607287,
+      "grad_norm": 0.43940627548274,
+      "learning_rate": 3.074135451053475e-05,
+      "loss": 0.5573,
+      "step": 479
+    },
+    {
+      "epoch": 1.9433198380566803,
+      "grad_norm": 0.41828453841924457,
+      "learning_rate": 3.069360584592015e-05,
+      "loss": 0.5841,
+      "step": 480
+    },
+    {
+      "epoch": 1.9473684210526314,
+      "grad_norm": 0.46176244335134803,
+      "learning_rate": 3.064577167551816e-05,
+      "loss": 0.5629,
+      "step": 481
+    },
+    {
+      "epoch": 1.951417004048583,
+      "grad_norm": 0.44595903595197006,
+      "learning_rate": 3.059785238180953e-05,
+      "loss": 0.567,
+      "step": 482
+    },
+    {
+      "epoch": 1.9554655870445345,
+      "grad_norm": 0.44446632974241157,
+      "learning_rate": 3.054984834795566e-05,
+      "loss": 0.5586,
+      "step": 483
+    },
+    {
+      "epoch": 1.9595141700404857,
+      "grad_norm": 0.40889631795057674,
+      "learning_rate": 3.0501759957795537e-05,
+      "loss": 0.5527,
+      "step": 484
+    },
+    {
+      "epoch": 1.9635627530364372,
+      "grad_norm": 0.5190929130044892,
+      "learning_rate": 3.0453587595842627e-05,
+      "loss": 0.555,
+      "step": 485
+    },
+    {
+      "epoch": 1.9676113360323888,
+      "grad_norm": 0.4384791415127682,
+      "learning_rate": 3.0405331647281856e-05,
+      "loss": 0.566,
+      "step": 486
+    },
+    {
+      "epoch": 1.97165991902834,
+      "grad_norm": 0.4859034167813243,
+      "learning_rate": 3.0356992497966508e-05,
+      "loss": 0.5637,
+      "step": 487
+    },
+    {
+      "epoch": 1.9757085020242915,
+      "grad_norm": 0.47966638862658934,
+      "learning_rate": 3.0308570534415137e-05,
+      "loss": 0.5459,
+      "step": 488
+    },
+    {
+      "epoch": 1.979757085020243,
+      "grad_norm": 0.5097985669983889,
+      "learning_rate": 3.0260066143808467e-05,
+      "loss": 0.5644,
+      "step": 489
+    },
+    {
+      "epoch": 1.9838056680161942,
+      "grad_norm": 0.5111600622755891,
+      "learning_rate": 3.021147971398632e-05,
+      "loss": 0.5649,
+      "step": 490
+    },
+    {
+      "epoch": 1.9878542510121457,
+      "grad_norm": 0.5920960993731526,
+      "learning_rate": 3.0162811633444478e-05,
+      "loss": 0.5757,
+      "step": 491
+    },
+    {
+      "epoch": 1.9919028340080973,
+      "grad_norm": 0.41427275672874453,
+      "learning_rate": 3.0114062291331635e-05,
+      "loss": 0.561,
+      "step": 492
+    },
+    {
+      "epoch": 1.9959514170040484,
+      "grad_norm": 0.6081232349812942,
+      "learning_rate": 3.0065232077446217e-05,
+      "loss": 0.556,
+      "step": 493
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.4115196790822288,
+      "learning_rate": 3.001632138223332e-05,
+      "loss": 0.545,
+      "step": 494
+    },
+    {
+      "epoch": 2.0040485829959516,
+      "grad_norm": 0.7697623088681668,
+      "learning_rate": 2.9967330596781565e-05,
+      "loss": 0.4124,
+      "step": 495
+    },
+    {
+      "epoch": 2.0080971659919027,
+      "grad_norm": 1.0198063485329862,
+      "learning_rate": 2.9918260112819952e-05,
+      "loss": 0.4076,
+      "step": 496
+    },
+    {
+      "epoch": 2.0121457489878543,
+      "grad_norm": 0.6346101182023476,
+      "learning_rate": 2.9869110322714778e-05,
+      "loss": 0.4084,
+      "step": 497
+    },
+    {
+      "epoch": 2.016194331983806,
+      "grad_norm": 0.5046103858120733,
+      "learning_rate": 2.9819881619466447e-05,
+      "loss": 0.4066,
+      "step": 498
+    },
+    {
+      "epoch": 2.020242914979757,
+      "grad_norm": 0.5344295462243199,
+      "learning_rate": 2.9770574396706364e-05,
+      "loss": 0.4163,
+      "step": 499
+    },
+    {
+      "epoch": 2.0242914979757085,
+      "grad_norm": 0.5965443948782839,
+      "learning_rate": 2.972118904869376e-05,
+      "loss": 0.4186,
+      "step": 500
+    },
+    {
+      "epoch": 2.02834008097166,
+      "grad_norm": 0.5516470811168162,
+      "learning_rate": 2.9671725970312563e-05,
+      "loss": 0.4109,
+      "step": 501
+    },
+    {
+      "epoch": 2.032388663967611,
+      "grad_norm": 0.4968286658741776,
+      "learning_rate": 2.9622185557068238e-05,
+      "loss": 0.4054,
+      "step": 502
+    },
+    {
+      "epoch": 2.0364372469635628,
+      "grad_norm": 0.5397398421979092,
+      "learning_rate": 2.95725682050846e-05,
+      "loss": 0.4034,
+      "step": 503
+    },
+    {
+      "epoch": 2.0404858299595143,
+      "grad_norm": 0.5101796385424205,
+      "learning_rate": 2.9522874311100684e-05,
+      "loss": 0.4118,
+      "step": 504
+    },
+    {
+      "epoch": 2.0445344129554655,
+      "grad_norm": 0.5241222825374852,
+      "learning_rate": 2.9473104272467526e-05,
+      "loss": 0.3956,
+      "step": 505
+    },
+    {
+      "epoch": 2.048582995951417,
+      "grad_norm": 0.5594150196178385,
+      "learning_rate": 2.942325848714505e-05,
+      "loss": 0.4113,
+      "step": 506
+    },
+    {
+      "epoch": 2.0526315789473686,
+      "grad_norm": 0.6127109269898036,
+      "learning_rate": 2.9373337353698804e-05,
+      "loss": 0.4153,
+      "step": 507
+    },
+    {
+      "epoch": 2.0566801619433197,
+      "grad_norm": 0.5075374823781003,
+      "learning_rate": 2.932334127129686e-05,
+      "loss": 0.3973,
+      "step": 508
+    },
+    {
+      "epoch": 2.0607287449392713,
+      "grad_norm": 0.5349047025455763,
+      "learning_rate": 2.9273270639706547e-05,
+      "loss": 0.407,
+      "step": 509
+    },
+    {
+      "epoch": 2.064777327935223,
+      "grad_norm": 0.6041536879014583,
+      "learning_rate": 2.922312585929131e-05,
+      "loss": 0.4071,
+      "step": 510
+    },
+    {
+      "epoch": 2.068825910931174,
+      "grad_norm": 0.536191152245298,
+      "learning_rate": 2.9172907331007478e-05,
+      "loss": 0.423,
+      "step": 511
+    },
+    {
+      "epoch": 2.0728744939271255,
+      "grad_norm": 0.3908075790437079,
+      "learning_rate": 2.9122615456401053e-05,
+      "loss": 0.3889,
+      "step": 512
+    },
+    {
+      "epoch": 2.076923076923077,
+      "grad_norm": 0.5230043023192886,
+      "learning_rate": 2.9072250637604535e-05,
+      "loss": 0.4257,
+      "step": 513
+    },
+    {
+      "epoch": 2.080971659919028,
+      "grad_norm": 0.4322269531955865,
+      "learning_rate": 2.9021813277333667e-05,
+      "loss": 0.4151,
+      "step": 514
+    },
+    {
+      "epoch": 2.08502024291498,
+      "grad_norm": 0.45926959239103016,
+      "learning_rate": 2.8971303778884244e-05,
+      "loss": 0.4107,
+      "step": 515
+    },
+    {
+      "epoch": 2.0890688259109313,
+      "grad_norm": 0.60854853014569,
+      "learning_rate": 2.8920722546128863e-05,
+      "loss": 0.3834,
+      "step": 516
+    },
+    {
+      "epoch": 2.0931174089068825,
+      "grad_norm": 0.4458872123233851,
+      "learning_rate": 2.887006998351372e-05,
+      "loss": 0.4255,
+      "step": 517
+    },
+    {
+      "epoch": 2.097165991902834,
+      "grad_norm": 0.4941581284473403,
+      "learning_rate": 2.881934649605535e-05,
+      "loss": 0.4158,
+      "step": 518
+    },
+    {
+      "epoch": 2.1012145748987856,
+      "grad_norm": 0.5069210287537788,
+      "learning_rate": 2.8768552489337408e-05,
+      "loss": 0.405,
+      "step": 519
+    },
+    {
+      "epoch": 2.1052631578947367,
+      "grad_norm": 0.42411383433476785,
+      "learning_rate": 2.8717688369507425e-05,
+      "loss": 0.4173,
+      "step": 520
+    },
+    {
+      "epoch": 2.1093117408906883,
+      "grad_norm": 0.4791610741120199,
+      "learning_rate": 2.8666754543273535e-05,
+      "loss": 0.3986,
+      "step": 521
+    },
+    {
+      "epoch": 2.11336032388664,
+      "grad_norm": 0.5272597678983434,
+      "learning_rate": 2.8615751417901273e-05,
+      "loss": 0.4096,
+      "step": 522
+    },
+    {
+      "epoch": 2.117408906882591,
+      "grad_norm": 0.5362121923316208,
+      "learning_rate": 2.856467940121025e-05,
+      "loss": 0.4181,
+      "step": 523
+    },
+    {
+      "epoch": 2.1214574898785425,
+      "grad_norm": 0.48977760428291156,
+      "learning_rate": 2.8513538901570964e-05,
+      "loss": 0.4241,
+      "step": 524
+    },
+    {
+      "epoch": 2.125506072874494,
+      "grad_norm": 0.5954809537445128,
+      "learning_rate": 2.846233032790149e-05,
+      "loss": 0.3923,
+      "step": 525
+    },
+    {
+      "epoch": 2.1295546558704452,
+      "grad_norm": 0.5165395682184515,
+      "learning_rate": 2.84110540896642e-05,
+      "loss": 0.4059,
+      "step": 526
+    },
+    {
+      "epoch": 2.133603238866397,
+      "grad_norm": 0.5445776754673729,
+      "learning_rate": 2.8359710596862555e-05,
+      "loss": 0.3965,
+      "step": 527
+    },
+    {
+      "epoch": 2.1376518218623484,
+      "grad_norm": 0.5106036783526501,
+      "learning_rate": 2.8308300260037734e-05,
+      "loss": 0.4157,
+      "step": 528
+    },
+    {
+      "epoch": 2.1417004048582995,
+      "grad_norm": 0.4575742065248807,
+      "learning_rate": 2.8256823490265424e-05,
+      "loss": 0.4069,
+      "step": 529
+    },
+    {
+      "epoch": 2.145748987854251,
+      "grad_norm": 0.4621882942026104,
+      "learning_rate": 2.8205280699152515e-05,
+      "loss": 0.3962,
+      "step": 530
+    },
+    {
+      "epoch": 2.1497975708502026,
+      "grad_norm": 0.4387741120292675,
+      "learning_rate": 2.8153672298833777e-05,
+      "loss": 0.4107,
+      "step": 531
+    },
+    {
+      "epoch": 2.1538461538461537,
+      "grad_norm": 0.5204969768861246,
+      "learning_rate": 2.8101998701968615e-05,
+      "loss": 0.4105,
+      "step": 532
+    },
+    {
+      "epoch": 2.1578947368421053,
+      "grad_norm": 0.5345534303453878,
+      "learning_rate": 2.805026032173773e-05,
+      "loss": 0.4079,
+      "step": 533
+    },
+    {
+      "epoch": 2.161943319838057,
+      "grad_norm": 0.4957126557531803,
+      "learning_rate": 2.7998457571839832e-05,
+      "loss": 0.4135,
+      "step": 534
+    },
+    {
+      "epoch": 2.165991902834008,
+      "grad_norm": 0.459799091896473,
+      "learning_rate": 2.7946590866488322e-05,
+      "loss": 0.4185,
+      "step": 535
+    },
+    {
+      "epoch": 2.1700404858299596,
+      "grad_norm": 0.4317990598954756,
+      "learning_rate": 2.7894660620408e-05,
+      "loss": 0.4147,
+      "step": 536
+    },
+    {
+      "epoch": 2.174089068825911,
+      "grad_norm": 0.4443683267614298,
+      "learning_rate": 2.784266724883173e-05,
+      "loss": 0.4064,
+      "step": 537
+    },
+    {
+      "epoch": 2.1781376518218623,
+      "grad_norm": 0.4633239524433981,
+      "learning_rate": 2.7790611167497128e-05,
+      "loss": 0.4055,
+      "step": 538
+    },
+    {
+      "epoch": 2.182186234817814,
+      "grad_norm": 0.4140818918988567,
+      "learning_rate": 2.7738492792643224e-05,
+      "loss": 0.3944,
+      "step": 539
+    },
+    {
+      "epoch": 2.1862348178137654,
+      "grad_norm": 0.461384213602541,
+      "learning_rate": 2.768631254100716e-05,
+      "loss": 0.3923,
+      "step": 540
+    },
+    {
+      "epoch": 2.1902834008097165,
+      "grad_norm": 0.4424949864429909,
+      "learning_rate": 2.7634070829820836e-05,
+      "loss": 0.4124,
+      "step": 541
+    },
+    {
+      "epoch": 2.194331983805668,
+      "grad_norm": 0.45573184554447654,
+      "learning_rate": 2.7581768076807592e-05,
+      "loss": 0.4028,
+      "step": 542
+    },
+    {
+      "epoch": 2.1983805668016196,
+      "grad_norm": 0.41553082941234176,
+      "learning_rate": 2.7529404700178832e-05,
+      "loss": 0.3817,
+      "step": 543
+    },
+    {
+      "epoch": 2.2024291497975708,
+      "grad_norm": 0.5195410099431053,
+      "learning_rate": 2.7476981118630735e-05,
+      "loss": 0.4018,
+      "step": 544
+    },
+    {
+      "epoch": 2.2064777327935223,
+      "grad_norm": 0.4285719593475236,
+      "learning_rate": 2.742449775134085e-05,
+      "loss": 0.4076,
+      "step": 545
+    },
+    {
+      "epoch": 2.2105263157894735,
+      "grad_norm": 0.49756738966120523,
+      "learning_rate": 2.7371955017964788e-05,
+      "loss": 0.4094,
+      "step": 546
+    },
+    {
+      "epoch": 2.214574898785425,
+      "grad_norm": 0.40615587097518935,
+      "learning_rate": 2.7319353338632845e-05,
+      "loss": 0.4168,
+      "step": 547
+    },
+    {
+      "epoch": 2.2186234817813766,
+      "grad_norm": 0.44056656708983066,
+      "learning_rate": 2.7266693133946645e-05,
+      "loss": 0.4233,
+      "step": 548
+    },
+    {
+      "epoch": 2.2226720647773277,
+      "grad_norm": 0.4796574356587187,
+      "learning_rate": 2.7213974824975785e-05,
+      "loss": 0.4162,
+      "step": 549
+    },
+    {
+      "epoch": 2.2267206477732793,
+      "grad_norm": 0.4155351813146497,
+      "learning_rate": 2.7161198833254464e-05,
+      "loss": 0.442,
+      "step": 550
+    },
+    {
+      "epoch": 2.230769230769231,
+      "grad_norm": 0.4448959851154466,
+      "learning_rate": 2.7108365580778095e-05,
+      "loss": 0.4159,
+      "step": 551
+    },
+    {
+      "epoch": 2.234817813765182,
+      "grad_norm": 0.46074016679994756,
+      "learning_rate": 2.705547548999996e-05,
+      "loss": 0.4084,
+      "step": 552
+    },
+    {
+      "epoch": 2.2388663967611335,
+      "grad_norm": 0.49630455370489895,
+      "learning_rate": 2.7002528983827817e-05,
+      "loss": 0.4062,
+      "step": 553
+    },
+    {
+      "epoch": 2.242914979757085,
+      "grad_norm": 0.4545707919076836,
+      "learning_rate": 2.6949526485620517e-05,
+      "loss": 0.4032,
+      "step": 554
+    },
+    {
+      "epoch": 2.246963562753036,
+      "grad_norm": 0.5079625644356227,
+      "learning_rate": 2.6896468419184623e-05,
+      "loss": 0.4178,
+      "step": 555
+    },
+    {
+      "epoch": 2.251012145748988,
+      "grad_norm": 0.42713430431205146,
+      "learning_rate": 2.6843355208771024e-05,
+      "loss": 0.3716,
+      "step": 556
+    },
+    {
+      "epoch": 2.2550607287449393,
+      "grad_norm": 0.4407437793539872,
+      "learning_rate": 2.6790187279071536e-05,
+      "loss": 0.4237,
+      "step": 557
+    },
+    {
+      "epoch": 2.2591093117408905,
+      "grad_norm": 1.0065222551193822,
+      "learning_rate": 2.67369650552155e-05,
+      "loss": 0.4096,
+      "step": 558
+    },
+    {
+      "epoch": 2.263157894736842,
+      "grad_norm": 0.4929845031525788,
+      "learning_rate": 2.6683688962766407e-05,
+      "loss": 0.4126,
+      "step": 559
+    },
+    {
+      "epoch": 2.2672064777327936,
+      "grad_norm": 0.42029763146665333,
+      "learning_rate": 2.6630359427718468e-05,
+      "loss": 0.4122,
+      "step": 560
+    },
+    {
+      "epoch": 2.2712550607287447,
+      "grad_norm": 0.4498946366745028,
+      "learning_rate": 2.6576976876493225e-05,
+      "loss": 0.4258,
+      "step": 561
+    },
+    {
+      "epoch": 2.2753036437246963,
+      "grad_norm": 0.5156467225745298,
+      "learning_rate": 2.652354173593614e-05,
+      "loss": 0.4196,
+      "step": 562
+    },
+    {
+      "epoch": 2.279352226720648,
+      "grad_norm": 0.4388455807844743,
+      "learning_rate": 2.647005443331317e-05,
+      "loss": 0.4048,
+      "step": 563
+    },
+    {
+      "epoch": 2.283400809716599,
+      "grad_norm": 0.5033988520708915,
+      "learning_rate": 2.6416515396307354e-05,
+      "loss": 0.4316,
+      "step": 564
+    },
+    {
+      "epoch": 2.2874493927125505,
+      "grad_norm": 0.5054297527392886,
+      "learning_rate": 2.636292505301542e-05,
+      "loss": 0.4095,
+      "step": 565
+    },
+    {
+      "epoch": 2.291497975708502,
+      "grad_norm": 0.5394019686601541,
+      "learning_rate": 2.630928383194431e-05,
+      "loss": 0.4188,
+      "step": 566
+    },
+    {
+      "epoch": 2.2955465587044532,
+      "grad_norm": 0.49642160355833487,
+      "learning_rate": 2.625559216200781e-05,
+      "loss": 0.4301,
+      "step": 567
+    },
+    {
+      "epoch": 2.299595141700405,
+      "grad_norm": 0.4835745318358811,
+      "learning_rate": 2.620185047252308e-05,
+      "loss": 0.4248,
+      "step": 568
+    },
+    {
+      "epoch": 2.3036437246963564,
+      "grad_norm": 0.4102022287526124,
+      "learning_rate": 2.614805919320723e-05,
+      "loss": 0.4143,
+      "step": 569
+    },
+    {
+      "epoch": 2.3076923076923075,
+      "grad_norm": 0.4661606193018389,
+      "learning_rate": 2.6094218754173895e-05,
+      "loss": 0.4244,
+      "step": 570
+    },
+    {
+      "epoch": 2.311740890688259,
+      "grad_norm": 0.4272980274554298,
+      "learning_rate": 2.60403295859298e-05,
+      "loss": 0.4138,
+      "step": 571
+    },
+    {
+      "epoch": 2.3157894736842106,
+      "grad_norm": 0.5183464218396063,
+      "learning_rate": 2.5986392119371297e-05,
+      "loss": 0.423,
+      "step": 572
+    },
+    {
+      "epoch": 2.3198380566801617,
+      "grad_norm": 0.3998997029160212,
+      "learning_rate": 2.5932406785780928e-05,
+      "loss": 0.4128,
+      "step": 573
+    },
+    {
+      "epoch": 2.3238866396761133,
+      "grad_norm": 0.49937190519324837,
+      "learning_rate": 2.5878374016823987e-05,
+      "loss": 0.4138,
+      "step": 574
+    },
+    {
+      "epoch": 2.327935222672065,
+      "grad_norm": 0.43881991050392205,
+      "learning_rate": 2.5824294244545045e-05,
+      "loss": 0.4237,
+      "step": 575
+    },
+    {
+      "epoch": 2.331983805668016,
+      "grad_norm": 0.4878982070728604,
+      "learning_rate": 2.577016790136453e-05,
+      "loss": 0.4223,
+      "step": 576
+    },
+    {
+      "epoch": 2.3360323886639676,
+      "grad_norm": 0.44251419257347574,
+      "learning_rate": 2.5715995420075225e-05,
+      "loss": 0.4142,
+      "step": 577
+    },
+    {
+      "epoch": 2.340080971659919,
+      "grad_norm": 0.47936063775901877,
+      "learning_rate": 2.5661777233838864e-05,
+      "loss": 0.4209,
+      "step": 578
+    },
+    {
+      "epoch": 2.3441295546558703,
+      "grad_norm": 0.4316561163002032,
+      "learning_rate": 2.5607513776182616e-05,
+      "loss": 0.4114,
+      "step": 579
+    },
+    {
+      "epoch": 2.348178137651822,
+      "grad_norm": 0.4445882074918383,
+      "learning_rate": 2.5553205480995633e-05,
+      "loss": 0.398,
+      "step": 580
+    },
+    {
+      "epoch": 2.3522267206477734,
+      "grad_norm": 0.5008458348425079,
+      "learning_rate": 2.5498852782525604e-05,
+      "loss": 0.4182,
+      "step": 581
+    },
+    {
+      "epoch": 2.3562753036437245,
+      "grad_norm": 0.45444964154406003,
+      "learning_rate": 2.5444456115375252e-05,
+      "loss": 0.4189,
+      "step": 582
+    },
+    {
+      "epoch": 2.360323886639676,
+      "grad_norm": 0.45418625187407297,
+      "learning_rate": 2.539001591449889e-05,
+      "loss": 0.4213,
+      "step": 583
+    },
+    {
+      "epoch": 2.3643724696356276,
+      "grad_norm": 0.42539220224486257,
+      "learning_rate": 2.53355326151989e-05,
+      "loss": 0.4106,
+      "step": 584
+    },
+    {
+      "epoch": 2.3684210526315788,
+      "grad_norm": 0.4743093083356267,
+      "learning_rate": 2.5281006653122288e-05,
+      "loss": 0.4287,
+      "step": 585
+    },
+    {
+      "epoch": 2.3724696356275303,
+      "grad_norm": 0.423220939570028,
+      "learning_rate": 2.5226438464257217e-05,
+      "loss": 0.4062,
+      "step": 586
+    },
+    {
+      "epoch": 2.376518218623482,
+      "grad_norm": 0.6018873685240169,
+      "learning_rate": 2.5171828484929447e-05,
+      "loss": 0.4325,
+      "step": 587
+    },
+    {
+      "epoch": 2.380566801619433,
+      "grad_norm": 0.44823742535496736,
+      "learning_rate": 2.5117177151798934e-05,
+      "loss": 0.415,
+      "step": 588
+    },
+    {
+      "epoch": 2.3846153846153846,
+      "grad_norm": 0.42527148686736055,
+      "learning_rate": 2.506248490185627e-05,
+      "loss": 0.4248,
+      "step": 589
+    },
+    {
+      "epoch": 2.388663967611336,
+      "grad_norm": 0.43247406559454327,
+      "learning_rate": 2.500775217241923e-05,
+      "loss": 0.4056,
+      "step": 590
+    },
+    {
+      "epoch": 2.3927125506072873,
+      "grad_norm": 0.49354112218666224,
+      "learning_rate": 2.4952979401129266e-05,
+      "loss": 0.4069,
+      "step": 591
+    },
+    {
+      "epoch": 2.396761133603239,
+      "grad_norm": 0.45903184419087906,
+      "learning_rate": 2.4898167025948e-05,
+      "loss": 0.426,
+      "step": 592
+    },
+    {
+      "epoch": 2.4008097165991904,
+      "grad_norm": 0.45459099036656264,
+      "learning_rate": 2.4843315485153703e-05,
+      "loss": 0.4336,
+      "step": 593
+    },
+    {
+      "epoch": 2.4048582995951415,
+      "grad_norm": 0.4759132448587322,
+      "learning_rate": 2.4788425217337856e-05,
+      "loss": 0.401,
+      "step": 594
+    },
+    {
+      "epoch": 2.408906882591093,
+      "grad_norm": 0.6678916392541466,
+      "learning_rate": 2.4733496661401572e-05,
+      "loss": 0.4089,
+      "step": 595
+    },
+    {
+      "epoch": 2.4129554655870447,
+      "grad_norm": 0.4665642328120312,
+      "learning_rate": 2.4678530256552114e-05,
+      "loss": 0.4141,
+      "step": 596
+    },
+    {
+      "epoch": 2.417004048582996,
+      "grad_norm": 0.5216781576759644,
+      "learning_rate": 2.46235264422994e-05,
+      "loss": 0.4016,
+      "step": 597
+    },
+    {
+      "epoch": 2.4210526315789473,
+      "grad_norm": 0.463144042219119,
+      "learning_rate": 2.4568485658452452e-05,
+      "loss": 0.396,
+      "step": 598
+    },
+    {
+      "epoch": 2.425101214574899,
+      "grad_norm": 0.48272458987479405,
+      "learning_rate": 2.451340834511592e-05,
+      "loss": 0.3998,
+      "step": 599
+    },
+    {
+      "epoch": 2.42914979757085,
+      "grad_norm": 0.42580216894596934,
+      "learning_rate": 2.4458294942686528e-05,
+      "loss": 0.4084,
+      "step": 600
+    },
+    {
+      "epoch": 2.4331983805668016,
+      "grad_norm": 0.526810145094761,
+      "learning_rate": 2.440314589184958e-05,
+      "loss": 0.4228,
+      "step": 601
+    },
+    {
+      "epoch": 2.437246963562753,
+      "grad_norm": 0.42370676526608425,
+      "learning_rate": 2.434796163357541e-05,
+      "loss": 0.3985,
+      "step": 602
+    },
+    {
+      "epoch": 2.4412955465587043,
+      "grad_norm": 0.48838232676915994,
+      "learning_rate": 2.429274260911588e-05,
+      "loss": 0.4193,
+      "step": 603
+    },
+    {
+      "epoch": 2.445344129554656,
+      "grad_norm": 0.47828846658352786,
+      "learning_rate": 2.423748926000084e-05,
+      "loss": 0.4098,
+      "step": 604
+    },
+    {
+      "epoch": 2.4493927125506074,
+      "grad_norm": 0.4598448821615105,
+      "learning_rate": 2.418220202803459e-05,
+      "loss": 0.4175,
+      "step": 605
+    },
+    {
+      "epoch": 2.4534412955465585,
+      "grad_norm": 0.46125946834221515,
+      "learning_rate": 2.412688135529237e-05,
+      "loss": 0.4063,
+      "step": 606
+    },
+    {
+      "epoch": 2.45748987854251,
+      "grad_norm": 0.4374068389341486,
+      "learning_rate": 2.4071527684116807e-05,
+      "loss": 0.4227,
+      "step": 607
+    },
+    {
+      "epoch": 2.4615384615384617,
+      "grad_norm": 0.4809270745312397,
+      "learning_rate": 2.4016141457114372e-05,
+      "loss": 0.4194,
+      "step": 608
+    },
+    {
+      "epoch": 2.465587044534413,
+      "grad_norm": 0.42892611294968186,
+      "learning_rate": 2.396072311715187e-05,
+      "loss": 0.414,
+      "step": 609
+    },
+    {
+      "epoch": 2.4696356275303644,
+      "grad_norm": 0.5633557221380807,
+      "learning_rate": 2.3905273107352863e-05,
+      "loss": 0.4029,
+      "step": 610
+    },
+    {
+      "epoch": 2.473684210526316,
+      "grad_norm": 0.43887202866917674,
+      "learning_rate": 2.3849791871094157e-05,
+      "loss": 0.4191,
+      "step": 611
+    },
+    {
+      "epoch": 2.477732793522267,
+      "grad_norm": 0.5609149516586832,
+      "learning_rate": 2.379427985200223e-05,
+      "loss": 0.4306,
+      "step": 612
+    },
+    {
+      "epoch": 2.4817813765182186,
+      "grad_norm": 0.4432853180385889,
+      "learning_rate": 2.373873749394972e-05,
+      "loss": 0.3994,
+      "step": 613
+    },
+    {
+      "epoch": 2.48582995951417,
+      "grad_norm": 0.44637338655810493,
+      "learning_rate": 2.3683165241051836e-05,
+      "loss": 0.4295,
+      "step": 614
+    },
+    {
+      "epoch": 2.4898785425101213,
+      "grad_norm": 0.4311240253807926,
+      "learning_rate": 2.3627563537662823e-05,
+      "loss": 0.4113,
+      "step": 615
+    },
+    {
+      "epoch": 2.493927125506073,
+      "grad_norm": 0.4422549583019162,
+      "learning_rate": 2.3571932828372433e-05,
+      "loss": 0.4232,
+      "step": 616
+    },
+    {
+      "epoch": 2.4979757085020244,
+      "grad_norm": 0.42925214634900477,
+      "learning_rate": 2.351627355800233e-05,
+      "loss": 0.402,
+      "step": 617
+    },
+    {
+      "epoch": 2.5020242914979756,
+      "grad_norm": 0.4573193876591057,
+      "learning_rate": 2.346058617160256e-05,
+      "loss": 0.4055,
+      "step": 618
+    },
+    {
+      "epoch": 2.506072874493927,
+      "grad_norm": 0.43975100865593214,
+      "learning_rate": 2.3404871114447976e-05,
+      "loss": 0.4052,
+      "step": 619
+    },
+    {
+      "epoch": 2.5101214574898787,
+      "grad_norm": 0.4380995768771331,
+      "learning_rate": 2.334912883203469e-05,
+      "loss": 0.4175,
+      "step": 620
+    },
+    {
+      "epoch": 2.51417004048583,
+      "grad_norm": 0.4488253864930049,
+      "learning_rate": 2.3293359770076505e-05,
+      "loss": 0.432,
+      "step": 621
+    },
+    {
+      "epoch": 2.5182186234817814,
+      "grad_norm": 0.46995940604925085,
+      "learning_rate": 2.323756437450136e-05,
+      "loss": 0.4256,
+      "step": 622
+    },
+    {
+      "epoch": 2.522267206477733,
+      "grad_norm": 0.47028216031639103,
+      "learning_rate": 2.3181743091447754e-05,
+      "loss": 0.41,
+      "step": 623
+    },
+    {
+      "epoch": 2.526315789473684,
+      "grad_norm": 0.41942171592245153,
+      "learning_rate": 2.3125896367261164e-05,
+      "loss": 0.4206,
+      "step": 624
+    },
+    {
+      "epoch": 2.5303643724696356,
+      "grad_norm": 0.4185907865405384,
+      "learning_rate": 2.307002464849052e-05,
+      "loss": 0.4224,
+      "step": 625
+    },
+    {
+      "epoch": 2.534412955465587,
+      "grad_norm": 0.4995015704245525,
+      "learning_rate": 2.30141283818846e-05,
+      "loss": 0.4198,
+      "step": 626
+    },
+    {
+      "epoch": 2.5384615384615383,
+      "grad_norm": 0.43408235231155556,
+      "learning_rate": 2.2958208014388462e-05,
+      "loss": 0.4027,
+      "step": 627
+    },
+    {
+      "epoch": 2.54251012145749,
+      "grad_norm": 0.5075172861014853,
+      "learning_rate": 2.2902263993139857e-05,
+      "loss": 0.4264,
+      "step": 628
+    },
+    {
+      "epoch": 2.5465587044534415,
+      "grad_norm": 0.44005738182570703,
+      "learning_rate": 2.2846296765465708e-05,
+      "loss": 0.4154,
+      "step": 629
+    },
+    {
+      "epoch": 2.5506072874493926,
+      "grad_norm": 0.4268202951662054,
+      "learning_rate": 2.2790306778878465e-05,
+      "loss": 0.4012,
+      "step": 630
+    },
+    {
+      "epoch": 2.554655870445344,
+      "grad_norm": 0.4871067404713711,
+      "learning_rate": 2.273429448107257e-05,
+      "loss": 0.3959,
+      "step": 631
+    },
+    {
+      "epoch": 2.5587044534412957,
+      "grad_norm": 0.41024547952587764,
+      "learning_rate": 2.2678260319920868e-05,
+      "loss": 0.3963,
+      "step": 632
+    },
+    {
+      "epoch": 2.562753036437247,
+      "grad_norm": 0.4447921065120892,
+      "learning_rate": 2.262220474347102e-05,
+      "loss": 0.4256,
+      "step": 633
+    },
+    {
+      "epoch": 2.5668016194331984,
+      "grad_norm": 0.42857223136033035,
+      "learning_rate": 2.2566128199941918e-05,
+      "loss": 0.4067,
+      "step": 634
+    },
+    {
+      "epoch": 2.57085020242915,
+      "grad_norm": 0.42053518599016226,
+      "learning_rate": 2.2510031137720104e-05,
+      "loss": 0.4186,
+      "step": 635
+    },
+    {
+      "epoch": 2.574898785425101,
+      "grad_norm": 0.462394432673237,
+      "learning_rate": 2.2453914005356206e-05,
+      "loss": 0.414,
+      "step": 636
+    },
+    {
+      "epoch": 2.5789473684210527,
+      "grad_norm": 0.4283972970276092,
+      "learning_rate": 2.2397777251561302e-05,
+      "loss": 0.4079,
+      "step": 637
+    },
+    {
+      "epoch": 2.582995951417004,
+      "grad_norm": 0.43078737035825804,
+      "learning_rate": 2.2341621325203393e-05,
+      "loss": 0.4148,
+      "step": 638
+    },
+    {
+      "epoch": 2.5870445344129553,
+      "grad_norm": 0.4019635982595694,
+      "learning_rate": 2.2285446675303764e-05,
+      "loss": 0.4235,
+      "step": 639
+    },
+    {
+      "epoch": 2.591093117408907,
+      "grad_norm": 0.4220347845370312,
+      "learning_rate": 2.2229253751033423e-05,
+      "loss": 0.4267,
+      "step": 640
+    },
+    {
+      "epoch": 2.5951417004048585,
+      "grad_norm": 0.40284954366034664,
+      "learning_rate": 2.2173043001709492e-05,
+      "loss": 0.4265,
+      "step": 641
+    },
+    {
+      "epoch": 2.5991902834008096,
+      "grad_norm": 0.4276873318175535,
+      "learning_rate": 2.2116814876791627e-05,
+      "loss": 0.3945,
+      "step": 642
+    },
+    {
+      "epoch": 2.603238866396761,
+      "grad_norm": 0.41493334683932875,
+      "learning_rate": 2.2060569825878413e-05,
+      "loss": 0.4211,
+      "step": 643
+    },
+    {
+      "epoch": 2.6072874493927127,
+      "grad_norm": 0.417198274634009,
+      "learning_rate": 2.2004308298703777e-05,
+      "loss": 0.4048,
+      "step": 644
+    },
+    {
+      "epoch": 2.611336032388664,
+      "grad_norm": 0.4737799111056316,
+      "learning_rate": 2.1948030745133396e-05,
+      "loss": 0.4136,
+      "step": 645
+    },
+    {
+      "epoch": 2.6153846153846154,
+      "grad_norm": 0.43330319645673276,
+      "learning_rate": 2.1891737615161083e-05,
+      "loss": 0.4223,
+      "step": 646
+    },
+    {
+      "epoch": 2.619433198380567,
+      "grad_norm": 0.40649823102139226,
+      "learning_rate": 2.1835429358905218e-05,
+      "loss": 0.4069,
+      "step": 647
+    },
+    {
+      "epoch": 2.623481781376518,
+      "grad_norm": 0.38581836737364406,
+      "learning_rate": 2.17791064266051e-05,
+      "loss": 0.4263,
+      "step": 648
+    },
+    {
+      "epoch": 2.6275303643724697,
+      "grad_norm": 0.46183898055851474,
+      "learning_rate": 2.172276926861741e-05,
+      "loss": 0.4217,
+      "step": 649
+    },
+    {
+      "epoch": 2.6315789473684212,
+      "grad_norm": 0.41730502167781375,
+      "learning_rate": 2.166641833541255e-05,
+      "loss": 0.4202,
+      "step": 650
+    },
+    {
+      "epoch": 2.6356275303643724,
+      "grad_norm": 0.41708209869273644,
+      "learning_rate": 2.161005407757109e-05,
+      "loss": 0.4173,
+      "step": 651
+    },
+    {
+      "epoch": 2.639676113360324,
+      "grad_norm": 0.4249668595084292,
+      "learning_rate": 2.155367694578013e-05,
+      "loss": 0.4262,
+      "step": 652
+    },
+    {
+      "epoch": 2.6437246963562755,
+      "grad_norm": 5.129207277020445,
+      "learning_rate": 2.1497287390829712e-05,
+      "loss": 0.408,
+      "step": 653
+    },
+    {
+      "epoch": 2.6477732793522266,
+      "grad_norm": 0.9178116865441093,
+      "learning_rate": 2.1440885863609208e-05,
+      "loss": 0.4233,
+      "step": 654
+    },
+    {
+      "epoch": 2.651821862348178,
+      "grad_norm": 14.354976186197778,
+      "learning_rate": 2.1384472815103737e-05,
+      "loss": 0.4746,
+      "step": 655
+    },
+    {
+      "epoch": 2.6558704453441297,
+      "grad_norm": 0.5559634939914085,
+      "learning_rate": 2.1328048696390534e-05,
+      "loss": 0.4263,
+      "step": 656
+    },
+    {
+      "epoch": 2.659919028340081,
+      "grad_norm": 0.42217725415344953,
+      "learning_rate": 2.1271613958635342e-05,
+      "loss": 0.4125,
+      "step": 657
+    },
+    {
+      "epoch": 2.6639676113360324,
+      "grad_norm": 0.47337226498709634,
+      "learning_rate": 2.1215169053088825e-05,
+      "loss": 0.411,
+      "step": 658
+    },
+    {
+      "epoch": 2.668016194331984,
+      "grad_norm": 0.42234147344077094,
+      "learning_rate": 2.1158714431082937e-05,
+      "loss": 0.4037,
+      "step": 659
+    },
+    {
+      "epoch": 2.672064777327935,
+      "grad_norm": 0.5267559070566546,
+      "learning_rate": 2.110225054402733e-05,
+      "loss": 0.4255,
+      "step": 660
+    },
+    {
+      "epoch": 2.6761133603238867,
+      "grad_norm": 0.4306562774017678,
+      "learning_rate": 2.1045777843405747e-05,
+      "loss": 0.4159,
+      "step": 661
+    },
+    {
+      "epoch": 2.6801619433198383,
+      "grad_norm": 0.43500004535030357,
+      "learning_rate": 2.09892967807724e-05,
+      "loss": 0.4092,
+      "step": 662
+    },
+    {
+      "epoch": 2.6842105263157894,
+      "grad_norm": 0.46326799030443694,
+      "learning_rate": 2.093280780774835e-05,
+      "loss": 0.4222,
+      "step": 663
+    },
+    {
+      "epoch": 2.688259109311741,
+      "grad_norm": 0.41605318474554387,
+      "learning_rate": 2.087631137601793e-05,
+      "loss": 0.3945,
+      "step": 664
+    },
+    {
+      "epoch": 2.6923076923076925,
+      "grad_norm": 0.4647327945324304,
+      "learning_rate": 2.0819807937325104e-05,
+      "loss": 0.4268,
+      "step": 665
+    },
+    {
+      "epoch": 2.6963562753036436,
+      "grad_norm": 0.4563491125942353,
+      "learning_rate": 2.0763297943469847e-05,
+      "loss": 0.4133,
+      "step": 666
+    },
+    {
+      "epoch": 2.700404858299595,
+      "grad_norm": 0.43855920080386607,
+      "learning_rate": 2.070678184630458e-05,
+      "loss": 0.4062,
+      "step": 667
+    },
+    {
+      "epoch": 2.7044534412955468,
+      "grad_norm": 0.44694332043563184,
+      "learning_rate": 2.06502600977305e-05,
+      "loss": 0.4122,
+      "step": 668
+    },
+    {
+      "epoch": 2.708502024291498,
+      "grad_norm": 0.42163133839414624,
+      "learning_rate": 2.0593733149694008e-05,
+      "loss": 0.4193,
+      "step": 669
+    },
+    {
+      "epoch": 2.7125506072874495,
+      "grad_norm": 0.4610589034038727,
+      "learning_rate": 2.0537201454183074e-05,
+      "loss": 0.4121,
+      "step": 670
+    },
+    {
+      "epoch": 2.716599190283401,
+      "grad_norm": 0.38267551209906336,
+      "learning_rate": 2.0480665463223626e-05,
+      "loss": 0.4193,
+      "step": 671
+    },
+    {
+      "epoch": 2.720647773279352,
+      "grad_norm": 0.4242850383948593,
+      "learning_rate": 2.042412562887594e-05,
+      "loss": 0.4167,
+      "step": 672
+    },
+    {
+      "epoch": 2.7246963562753037,
+      "grad_norm": 0.3708087140513297,
+      "learning_rate": 2.0367582403231033e-05,
+      "loss": 0.4306,
+      "step": 673
+    },
+    {
+      "epoch": 2.7287449392712553,
+      "grad_norm": 0.4192675726609841,
+      "learning_rate": 2.0311036238407023e-05,
+      "loss": 0.4063,
+      "step": 674
+    },
+    {
+      "epoch": 2.7327935222672064,
+      "grad_norm": 0.41301344048654015,
+      "learning_rate": 2.0254487586545534e-05,
+      "loss": 0.4027,
+      "step": 675
+    },
+    {
+      "epoch": 2.736842105263158,
+      "grad_norm": 0.4018847185171801,
+      "learning_rate": 2.0197936899808088e-05,
+      "loss": 0.411,
+      "step": 676
+    },
+    {
+      "epoch": 2.7408906882591095,
+      "grad_norm": 0.39663086577951695,
+      "learning_rate": 2.0141384630372466e-05,
+      "loss": 0.4136,
+      "step": 677
+    },
+    {
+      "epoch": 2.7449392712550607,
+      "grad_norm": 0.3888289373967148,
+      "learning_rate": 2.0084831230429106e-05,
+      "loss": 0.4221,
+      "step": 678
+    },
+    {
+      "epoch": 2.748987854251012,
+      "grad_norm": 0.3866373645646477,
+      "learning_rate": 2.0028277152177492e-05,
+      "loss": 0.4065,
+      "step": 679
+    },
+    {
+      "epoch": 2.753036437246964,
+      "grad_norm": 0.38669793000242386,
+      "learning_rate": 1.9971722847822518e-05,
+      "loss": 0.4247,
+      "step": 680
+    },
+    {
+      "epoch": 2.757085020242915,
+      "grad_norm": 0.39569311342281444,
+      "learning_rate": 1.99151687695709e-05,
+      "loss": 0.4013,
+      "step": 681
+    },
+    {
+      "epoch": 2.7611336032388665,
+      "grad_norm": 0.3749629757471426,
+      "learning_rate": 1.9858615369627537e-05,
+      "loss": 0.4138,
+      "step": 682
+    },
+    {
+      "epoch": 2.765182186234818,
+      "grad_norm": 0.3967884479645164,
+      "learning_rate": 1.9802063100191916e-05,
+      "loss": 0.4235,
+      "step": 683
+    },
+    {
+      "epoch": 2.769230769230769,
+      "grad_norm": 0.39006976434576945,
+      "learning_rate": 1.974551241345447e-05,
+      "loss": 0.4248,
+      "step": 684
+    },
+    {
+      "epoch": 2.7732793522267207,
+      "grad_norm": 0.3930737669346127,
+      "learning_rate": 1.9688963761592984e-05,
+      "loss": 0.3993,
+      "step": 685
+    },
+    {
+      "epoch": 2.7773279352226723,
+      "grad_norm": 0.3906586526657853,
+      "learning_rate": 1.963241759676897e-05,
+      "loss": 0.436,
+      "step": 686
+    },
+    {
+      "epoch": 2.7813765182186234,
+      "grad_norm": 0.4718078578660584,
+      "learning_rate": 1.9575874371124066e-05,
+      "loss": 0.4055,
+      "step": 687
+    },
+    {
+      "epoch": 2.785425101214575,
+      "grad_norm": 0.40318940284230775,
+      "learning_rate": 1.951933453677638e-05,
+      "loss": 0.4258,
+      "step": 688
+    },
+    {
+      "epoch": 2.7894736842105265,
+      "grad_norm": 0.4115953396679594,
+      "learning_rate": 1.9462798545816932e-05,
+      "loss": 0.4133,
+      "step": 689
+    },
+    {
+      "epoch": 2.7935222672064777,
+      "grad_norm": 0.4327538753687804,
+      "learning_rate": 1.9406266850305996e-05,
+      "loss": 0.4197,
+      "step": 690
+    },
+    {
+      "epoch": 2.7975708502024292,
+      "grad_norm": 0.40074089782226635,
+      "learning_rate": 1.9349739902269503e-05,
+      "loss": 0.4044,
+      "step": 691
+    },
+    {
+      "epoch": 2.801619433198381,
+      "grad_norm": 0.38909774319956325,
+      "learning_rate": 1.9293218153695425e-05,
+      "loss": 0.4108,
+      "step": 692
+    },
+    {
+      "epoch": 2.805668016194332,
+      "grad_norm": 0.4316013386751565,
+      "learning_rate": 1.9236702056530153e-05,
+      "loss": 0.4121,
+      "step": 693
+    },
+    {
+      "epoch": 2.8097165991902835,
+      "grad_norm": 0.3901147397916779,
+      "learning_rate": 1.9180192062674907e-05,
+      "loss": 0.4086,
+      "step": 694
+    },
+    {
+      "epoch": 2.813765182186235,
+      "grad_norm": 0.4163650396641431,
+      "learning_rate": 1.9123688623982076e-05,
+      "loss": 0.415,
+      "step": 695
+    },
+    {
+      "epoch": 2.817813765182186,
+      "grad_norm": 0.4043054670863953,
+      "learning_rate": 1.906719219225165e-05,
+      "loss": 0.4053,
+      "step": 696
+    },
+    {
+      "epoch": 2.8218623481781377,
+      "grad_norm": 0.4101973270414887,
+      "learning_rate": 1.9010703219227603e-05,
+      "loss": 0.4267,
+      "step": 697
+    },
+    {
+      "epoch": 2.8259109311740893,
+      "grad_norm": 0.5365782338279556,
+      "learning_rate": 1.8954222156594256e-05,
+      "loss": 0.4353,
+      "step": 698
+    },
+    {
+      "epoch": 2.8299595141700404,
+      "grad_norm": 0.3976581919307605,
+      "learning_rate": 1.8897749455972673e-05,
+      "loss": 0.4195,
+      "step": 699
+    },
+    {
+      "epoch": 2.834008097165992,
+      "grad_norm": 0.4722525888754367,
+      "learning_rate": 1.884128556891707e-05,
+      "loss": 0.4028,
+      "step": 700
+    },
+    {
+      "epoch": 2.8380566801619436,
+      "grad_norm": 0.43120691677644074,
+      "learning_rate": 1.878483094691119e-05,
+      "loss": 0.3891,
+      "step": 701
+    },
+    {
+      "epoch": 2.8421052631578947,
+      "grad_norm": 0.3808765792125265,
+      "learning_rate": 1.8728386041364664e-05,
+      "loss": 0.4031,
+      "step": 702
+    },
+    {
+      "epoch": 2.8461538461538463,
+      "grad_norm": 0.49828760724719107,
+      "learning_rate": 1.867195130360947e-05,
+      "loss": 0.4261,
+      "step": 703
+    },
+    {
+      "epoch": 2.850202429149798,
+      "grad_norm": 0.38707702711548697,
+      "learning_rate": 1.8615527184896263e-05,
+      "loss": 0.42,
+      "step": 704
+    },
+    {
+      "epoch": 2.854251012145749,
+      "grad_norm": 0.470377118543371,
+      "learning_rate": 1.8559114136390795e-05,
+      "loss": 0.4121,
+      "step": 705
+    },
+    {
+      "epoch": 2.8582995951417005,
+      "grad_norm": 0.44339235859691856,
+      "learning_rate": 1.8502712609170298e-05,
+      "loss": 0.4188,
+      "step": 706
+    },
+    {
+      "epoch": 2.862348178137652,
+      "grad_norm": 0.4189024456881989,
+      "learning_rate": 1.8446323054219876e-05,
+      "loss": 0.4119,
+      "step": 707
+    },
+    {
+      "epoch": 2.866396761133603,
+      "grad_norm": 0.40467484886710203,
+      "learning_rate": 1.838994592242891e-05,
+      "loss": 0.412,
+      "step": 708
+    },
+    {
+      "epoch": 2.8704453441295548,
+      "grad_norm": 0.4320436401018974,
+      "learning_rate": 1.8333581664587453e-05,
+      "loss": 0.4001,
+      "step": 709
+    },
+    {
+      "epoch": 2.8744939271255063,
+      "grad_norm": 0.47210243822596704,
+      "learning_rate": 1.8277230731382593e-05,
+      "loss": 0.4018,
+      "step": 710
+    },
+    {
+      "epoch": 2.8785425101214575,
+      "grad_norm": 0.4247314460376957,
+      "learning_rate": 1.82208935733949e-05,
+      "loss": 0.4273,
+      "step": 711
+    },
+    {
+      "epoch": 2.882591093117409,
+      "grad_norm": 0.3859257644572033,
+      "learning_rate": 1.8164570641094793e-05,
+      "loss": 0.4149,
+      "step": 712
+    },
+    {
+      "epoch": 2.8866396761133606,
+      "grad_norm": 0.42614212305696814,
+      "learning_rate": 1.810826238483892e-05,
+      "loss": 0.4187,
+      "step": 713
+    },
+    {
+      "epoch": 2.8906882591093117,
+      "grad_norm": 0.41477413536864527,
+      "learning_rate": 1.8051969254866608e-05,
+      "loss": 0.4225,
+      "step": 714
+    },
+    {
+      "epoch": 2.8947368421052633,
+      "grad_norm": 0.4145041720994417,
+      "learning_rate": 1.7995691701296226e-05,
+      "loss": 0.3964,
+      "step": 715
+    },
+    {
+      "epoch": 2.898785425101215,
+      "grad_norm": 0.4002884708278908,
+      "learning_rate": 1.7939430174121593e-05,
+      "loss": 0.4186,
+      "step": 716
+    },
+    {
+      "epoch": 2.902834008097166,
+      "grad_norm": 0.4396884751002073,
+      "learning_rate": 1.788318512320838e-05,
+      "loss": 0.4156,
+      "step": 717
+    },
+    {
+      "epoch": 2.9068825910931175,
+      "grad_norm": 0.41974600367262954,
+      "learning_rate": 1.782695699829051e-05,
+      "loss": 0.4096,
+      "step": 718
+    },
+    {
+      "epoch": 2.910931174089069,
+      "grad_norm": 0.37589477207706345,
+      "learning_rate": 1.7770746248966587e-05,
+      "loss": 0.399,
+      "step": 719
+    },
+    {
+      "epoch": 2.91497975708502,
+      "grad_norm": 0.4149034488989687,
+      "learning_rate": 1.7714553324696243e-05,
+      "loss": 0.4047,
+      "step": 720
+    },
+    {
+      "epoch": 2.919028340080972,
+      "grad_norm": 0.45973683589769676,
+      "learning_rate": 1.7658378674796614e-05,
+      "loss": 0.4155,
+      "step": 721
+    },
+    {
+      "epoch": 2.9230769230769234,
+      "grad_norm": 0.40299576432080675,
+      "learning_rate": 1.76022227484387e-05,
+      "loss": 0.4124,
+      "step": 722
+    },
+    {
+      "epoch": 2.9271255060728745,
+      "grad_norm": 0.42766663615254813,
+      "learning_rate": 1.7546085994643807e-05,
+      "loss": 0.4168,
+      "step": 723
+    },
+    {
+      "epoch": 2.931174089068826,
+      "grad_norm": 0.43397108767144504,
+      "learning_rate": 1.7489968862279902e-05,
+      "loss": 0.4115,
+      "step": 724
+    },
+    {
+      "epoch": 2.9352226720647776,
+      "grad_norm": 0.4276233971481483,
+      "learning_rate": 1.743387180005809e-05,
+      "loss": 0.4085,
+      "step": 725
+    },
+    {
+      "epoch": 2.9392712550607287,
+      "grad_norm": 0.44085716384040063,
+      "learning_rate": 1.737779525652899e-05,
+      "loss": 0.4256,
+      "step": 726
+    },
+    {
+      "epoch": 2.9433198380566803,
+      "grad_norm": 0.43151444889211976,
+      "learning_rate": 1.732173968007914e-05,
+      "loss": 0.4118,
+      "step": 727
+    },
+    {
+      "epoch": 2.9473684210526314,
+      "grad_norm": 0.41901294211574763,
+      "learning_rate": 1.7265705518927437e-05,
+      "loss": 0.4162,
+      "step": 728
+    },
+    {
+      "epoch": 2.951417004048583,
+      "grad_norm": 0.3729581513397867,
+      "learning_rate": 1.7209693221121542e-05,
+      "loss": 0.4284,
+      "step": 729
+    },
+    {
+      "epoch": 2.9554655870445345,
+      "grad_norm": 0.4270355585776899,
+      "learning_rate": 1.7153703234534302e-05,
+      "loss": 0.403,
+      "step": 730
+    },
+    {
+      "epoch": 2.9595141700404857,
+      "grad_norm": 0.3939808326040104,
+      "learning_rate": 1.709773600686015e-05,
+      "loss": 0.4001,
+      "step": 731
+    },
+    {
+      "epoch": 2.9635627530364372,
+      "grad_norm": 0.3769728490123247,
+      "learning_rate": 1.7041791985611548e-05,
+      "loss": 0.4023,
+      "step": 732
+    },
+    {
+      "epoch": 2.967611336032389,
+      "grad_norm": 0.4571149089456342,
+      "learning_rate": 1.69858716181154e-05,
+      "loss": 0.422,
+      "step": 733
+    },
+    {
+      "epoch": 2.97165991902834,
+      "grad_norm": 0.3883583400498467,
+      "learning_rate": 1.692997535150948e-05,
+      "loss": 0.4101,
+      "step": 734
+    },
+    {
+      "epoch": 2.9757085020242915,
+      "grad_norm": 0.4158989887539294,
+      "learning_rate": 1.687410363273884e-05,
+      "loss": 0.4097,
+      "step": 735
+    },
+    {
+      "epoch": 2.979757085020243,
+      "grad_norm": 0.43357234199844147,
+      "learning_rate": 1.6818256908552257e-05,
+      "loss": 0.415,
+      "step": 736
+    },
+    {
+      "epoch": 2.983805668016194,
+      "grad_norm": 0.4036689900551293,
+      "learning_rate": 1.6762435625498646e-05,
+      "loss": 0.4264,
+      "step": 737
+    },
+    {
+      "epoch": 2.9878542510121457,
+      "grad_norm": 0.4749232903804516,
+      "learning_rate": 1.67066402299235e-05,
+      "loss": 0.4114,
+      "step": 738
+    },
+    {
+      "epoch": 2.9919028340080973,
+      "grad_norm": 0.46129696421007105,
+      "learning_rate": 1.6650871167965313e-05,
+      "loss": 0.4239,
+      "step": 739
+    },
+    {
+      "epoch": 2.9959514170040484,
+      "grad_norm": 0.4249955924781897,
+      "learning_rate": 1.6595128885552028e-05,
+      "loss": 0.4026,
+      "step": 740
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 0.4291315987581198,
+      "learning_rate": 1.653941382839745e-05,
+      "loss": 0.3984,
+      "step": 741
+    },
+    {
+      "epoch": 3.0040485829959516,
+      "grad_norm": 0.7599082557525318,
+      "learning_rate": 1.6483726441997673e-05,
+      "loss": 0.2975,
+      "step": 742
+    },
+    {
+      "epoch": 3.0080971659919027,
+      "grad_norm": 0.44865146408489864,
+      "learning_rate": 1.642806717162757e-05,
+      "loss": 0.2788,
+      "step": 743
+    },
+    {
+      "epoch": 3.0121457489878543,
+      "grad_norm": 0.8907645712179667,
+      "learning_rate": 1.637243646233718e-05,
+      "loss": 0.2821,
+      "step": 744
+    },
+    {
+      "epoch": 3.016194331983806,
+      "grad_norm": 0.6578510800206772,
+      "learning_rate": 1.6316834758948174e-05,
+      "loss": 0.2996,
+      "step": 745
+    },
+    {
+      "epoch": 3.020242914979757,
+      "grad_norm": 0.5956427565325288,
+      "learning_rate": 1.6261262506050282e-05,
+      "loss": 0.2897,
+      "step": 746
+    },
+    {
+      "epoch": 3.0242914979757085,
+      "grad_norm": 0.667071495427492,
+      "learning_rate": 1.620572014799777e-05,
+      "loss": 0.2789,
+      "step": 747
+    },
+    {
+      "epoch": 3.02834008097166,
+      "grad_norm": 0.5006461595818111,
+      "learning_rate": 1.6150208128905857e-05,
+      "loss": 0.2959,
+      "step": 748
+    },
+    {
+      "epoch": 3.032388663967611,
+      "grad_norm": 0.46628471774162605,
+      "learning_rate": 1.6094726892647147e-05,
+      "loss": 0.276,
+      "step": 749
+    },
+    {
+      "epoch": 3.0364372469635628,
+      "grad_norm": 0.48943106960421545,
+      "learning_rate": 1.6039276882848135e-05,
+      "loss": 0.2731,
+      "step": 750
+    },
+    {
+      "epoch": 3.0404858299595143,
+      "grad_norm": 0.4768622280319648,
+      "learning_rate": 1.5983858542885635e-05,
+      "loss": 0.2873,
+      "step": 751
+    },
+    {
+      "epoch": 3.0445344129554655,
+      "grad_norm": 0.47165651424303506,
+      "learning_rate": 1.5928472315883203e-05,
+      "loss": 0.2756,
+      "step": 752
+    },
+    {
+      "epoch": 3.048582995951417,
+      "grad_norm": 0.43115333388824695,
+      "learning_rate": 1.5873118644707633e-05,
+      "loss": 0.278,
+      "step": 753
+    },
+    {
+      "epoch": 3.0526315789473686,
+      "grad_norm": 0.43667141427401,
+      "learning_rate": 1.5817797971965413e-05,
+      "loss": 0.2833,
+      "step": 754
+    },
+    {
+      "epoch": 3.0566801619433197,
+      "grad_norm": 0.4756902563946376,
+      "learning_rate": 1.576251073999917e-05,
+      "loss": 0.2608,
+      "step": 755
+    },
+    {
+      "epoch": 3.0607287449392713,
+      "grad_norm": 0.42251443650859133,
+      "learning_rate": 1.5707257390884126e-05,
+      "loss": 0.2788,
+      "step": 756
+    },
+    {
+      "epoch": 3.064777327935223,
+      "grad_norm": 0.42888254879881255,
+      "learning_rate": 1.5652038366424595e-05,
+      "loss": 0.2823,
+      "step": 757
+    },
+    {
+      "epoch": 3.068825910931174,
+      "grad_norm": 0.469909041180784,
+      "learning_rate": 1.5596854108150424e-05,
+      "loss": 0.2863,
+      "step": 758
+    },
+    {
+      "epoch": 3.0728744939271255,
+      "grad_norm": 0.41455937260630005,
+      "learning_rate": 1.5541705057313476e-05,
+      "loss": 0.2831,
+      "step": 759
+    },
+    {
+      "epoch": 3.076923076923077,
+      "grad_norm": 0.4297343356147595,
+      "learning_rate": 1.5486591654884086e-05,
+      "loss": 0.2744,
+      "step": 760
+    },
+    {
+      "epoch": 3.080971659919028,
+      "grad_norm": 0.4221206298103517,
+      "learning_rate": 1.543151434154755e-05,
+      "loss": 0.2685,
+      "step": 761
+    },
+    {
+      "epoch": 3.08502024291498,
+      "grad_norm": 0.36357329941888156,
+      "learning_rate": 1.5376473557700615e-05,
+      "loss": 0.2945,
+      "step": 762
+    },
+    {
+      "epoch": 3.0890688259109313,
+      "grad_norm": 0.44140593840629255,
+      "learning_rate": 1.5321469743447893e-05,
+      "loss": 0.2723,
+      "step": 763
+    },
+    {
+      "epoch": 3.0931174089068825,
+      "grad_norm": 0.41009098473388705,
+      "learning_rate": 1.5266503338598434e-05,
+      "loss": 0.2799,
+      "step": 764
+    },
+    {
+      "epoch": 3.097165991902834,
+      "grad_norm": 0.4017464448358975,
+      "learning_rate": 1.5211574782662149e-05,
+      "loss": 0.2708,
+      "step": 765
+    },
+    {
+      "epoch": 3.1012145748987856,
+      "grad_norm": 0.4295742554553543,
+      "learning_rate": 1.5156684514846307e-05,
+      "loss": 0.2826,
+      "step": 766
+    },
+    {
+      "epoch": 3.1052631578947367,
+      "grad_norm": 0.39665384143433163,
+      "learning_rate": 1.5101832974052012e-05,
+      "loss": 0.2675,
+      "step": 767
+    },
+    {
+      "epoch": 3.1093117408906883,
+      "grad_norm": 0.4108021696566653,
+      "learning_rate": 1.5047020598870737e-05,
+      "loss": 0.2823,
+      "step": 768
+    },
+    {
+      "epoch": 3.11336032388664,
+      "grad_norm": 0.38924238051652493,
+      "learning_rate": 1.4992247827580778e-05,
+      "loss": 0.2817,
+      "step": 769
+    },
+    {
+      "epoch": 3.117408906882591,
+      "grad_norm": 0.4207843904959568,
+      "learning_rate": 1.493751509814374e-05,
+      "loss": 0.2761,
+      "step": 770
+    },
+    {
+      "epoch": 3.1214574898785425,
+      "grad_norm": 0.3795119944704963,
+      "learning_rate": 1.4882822848201073e-05,
+      "loss": 0.2739,
+      "step": 771
+    },
+    {
+      "epoch": 3.125506072874494,
+      "grad_norm": 0.41361146522916203,
+      "learning_rate": 1.4828171515070553e-05,
+      "loss": 0.2868,
+      "step": 772
+    },
+    {
+      "epoch": 3.1295546558704452,
+      "grad_norm": 0.4018580963399373,
+      "learning_rate": 1.4773561535742793e-05,
+      "loss": 0.2753,
+      "step": 773
+    },
+    {
+      "epoch": 3.133603238866397,
+      "grad_norm": 0.3822180586385162,
+      "learning_rate": 1.4718993346877715e-05,
+      "loss": 0.275,
+      "step": 774
+    },
+    {
+      "epoch": 3.1376518218623484,
+      "grad_norm": 0.4176650738520671,
+      "learning_rate": 1.466446738480111e-05,
+      "loss": 0.27,
+      "step": 775
+    },
+    {
+      "epoch": 3.1417004048582995,
+      "grad_norm": 0.3834973148631733,
+      "learning_rate": 1.460998408550112e-05,
+      "loss": 0.285,
+      "step": 776
+    },
+    {
+      "epoch": 3.145748987854251,
+      "grad_norm": 0.41692467170504305,
+      "learning_rate": 1.4555543884624751e-05,
+      "loss": 0.2762,
+      "step": 777
+    },
+    {
+      "epoch": 3.1497975708502026,
+      "grad_norm": 0.38698145744980517,
+      "learning_rate": 1.4501147217474402e-05,
+      "loss": 0.2819,
+      "step": 778
+    },
+    {
+      "epoch": 3.1538461538461537,
+      "grad_norm": 0.38960726432592285,
+      "learning_rate": 1.444679451900437e-05,
+      "loss": 0.2872,
+      "step": 779
+    },
+    {
+      "epoch": 3.1578947368421053,
+      "grad_norm": 0.3740216752637036,
+      "learning_rate": 1.4392486223817397e-05,
+      "loss": 0.2827,
+      "step": 780
+    },
+    {
+      "epoch": 3.161943319838057,
+      "grad_norm": 0.4000609628729081,
+      "learning_rate": 1.4338222766161143e-05,
+      "loss": 0.2618,
+      "step": 781
+    },
+    {
+      "epoch": 3.165991902834008,
+      "grad_norm": 0.39579911921933675,
+      "learning_rate": 1.4284004579924777e-05,
+      "loss": 0.2823,
+      "step": 782
+    },
+    {
+      "epoch": 3.1700404858299596,
+      "grad_norm": 0.38373280500884405,
+      "learning_rate": 1.4229832098635479e-05,
+      "loss": 0.2853,
+      "step": 783
+    },
+    {
+      "epoch": 3.174089068825911,
+      "grad_norm": 0.3841627315670691,
+      "learning_rate": 1.4175705755454963e-05,
+      "loss": 0.2956,
+      "step": 784
+    },
+    {
+      "epoch": 3.1781376518218623,
+      "grad_norm": 0.3751361073899476,
+      "learning_rate": 1.412162598317602e-05,
+      "loss": 0.266,
+      "step": 785
+    },
+    {
+      "epoch": 3.182186234817814,
+      "grad_norm": 0.35883844409893656,
+      "learning_rate": 1.406759321421907e-05,
+      "loss": 0.276,
+      "step": 786
+    },
+    {
+      "epoch": 3.1862348178137654,
+      "grad_norm": 0.4057307647702272,
+      "learning_rate": 1.401360788062871e-05,
+      "loss": 0.284,
+      "step": 787
+    },
+    {
+      "epoch": 3.1902834008097165,
+      "grad_norm": 0.390305792755178,
+      "learning_rate": 1.3959670414070204e-05,
+      "loss": 0.275,
+      "step": 788
+    },
+    {
+      "epoch": 3.194331983805668,
+      "grad_norm": 0.4154865945131834,
+      "learning_rate": 1.3905781245826108e-05,
+      "loss": 0.2645,
+      "step": 789
+    },
+    {
+      "epoch": 3.1983805668016196,
+      "grad_norm": 0.39920677092287454,
+      "learning_rate": 1.3851940806792778e-05,
+      "loss": 0.2712,
+      "step": 790
+    },
+    {
+      "epoch": 3.2024291497975708,
+      "grad_norm": 0.3674121950603693,
+      "learning_rate": 1.379814952747693e-05,
+      "loss": 0.2913,
+      "step": 791
+    },
+    {
+      "epoch": 3.2064777327935223,
+      "grad_norm": 0.39454400233785275,
+      "learning_rate": 1.3744407837992193e-05,
+      "loss": 0.2622,
+      "step": 792
+    },
+    {
+      "epoch": 3.2105263157894735,
+      "grad_norm": 0.39073107537839863,
+      "learning_rate": 1.3690716168055692e-05,
+      "loss": 0.2706,
+      "step": 793
+    },
+    {
+      "epoch": 3.214574898785425,
+      "grad_norm": 0.3877792332688356,
+      "learning_rate": 1.363707494698459e-05,
+      "loss": 0.2798,
+      "step": 794
+    },
+    {
+      "epoch": 3.2186234817813766,
+      "grad_norm": 0.3999007038753181,
+      "learning_rate": 1.358348460369265e-05,
+      "loss": 0.284,
+      "step": 795
+    },
+    {
+      "epoch": 3.2226720647773277,
+      "grad_norm": 0.36007435995428855,
+      "learning_rate": 1.352994556668684e-05,
+      "loss": 0.28,
+      "step": 796
+    },
+    {
+      "epoch": 3.2267206477732793,
+      "grad_norm": 0.38339261902961536,
+      "learning_rate": 1.3476458264063865e-05,
+      "loss": 0.2836,
+      "step": 797
+    },
+    {
+      "epoch": 3.230769230769231,
+      "grad_norm": 0.3714216235522727,
+      "learning_rate": 1.342302312350678e-05,
+      "loss": 0.2798,
+      "step": 798
+    },
+    {
+      "epoch": 3.234817813765182,
+      "grad_norm": 0.37243921266756197,
+      "learning_rate": 1.3369640572281537e-05,
+      "loss": 0.2976,
+      "step": 799
+    },
+    {
+      "epoch": 3.2388663967611335,
+      "grad_norm": 0.36771220092177653,
+      "learning_rate": 1.3316311037233596e-05,
+      "loss": 0.2743,
+      "step": 800
+    },
+    {
+      "epoch": 3.242914979757085,
+      "grad_norm": 0.4051499136003373,
+      "learning_rate": 1.326303494478451e-05,
+      "loss": 0.2795,
+      "step": 801
+    },
+    {
+      "epoch": 3.246963562753036,
+      "grad_norm": 0.3596511764096029,
+      "learning_rate": 1.3209812720928472e-05,
+      "loss": 0.281,
+      "step": 802
+    },
+    {
+      "epoch": 3.251012145748988,
+      "grad_norm": 0.37951765014375033,
+      "learning_rate": 1.315664479122898e-05,
+      "loss": 0.2685,
+      "step": 803
+    },
+    {
+      "epoch": 3.2550607287449393,
+      "grad_norm": 0.4215905307127568,
+      "learning_rate": 1.3103531580815378e-05,
+      "loss": 0.2719,
+      "step": 804
+    },
+    {
+      "epoch": 3.2591093117408905,
+      "grad_norm": 0.38433501552319393,
+      "learning_rate": 1.3050473514379488e-05,
+      "loss": 0.284,
+      "step": 805
+    },
+    {
+      "epoch": 3.263157894736842,
+      "grad_norm": 0.38488693310063193,
+      "learning_rate": 1.2997471016172188e-05,
+      "loss": 0.2815,
+      "step": 806
+    },
+    {
+      "epoch": 3.2672064777327936,
+      "grad_norm": 0.3757497420925305,
+      "learning_rate": 1.2944524510000042e-05,
+      "loss": 0.2673,
+      "step": 807
+    },
+    {
+      "epoch": 3.2712550607287447,
+      "grad_norm": 0.3707840461951492,
+      "learning_rate": 1.2891634419221908e-05,
+      "loss": 0.2943,
+      "step": 808
+    },
+    {
+      "epoch": 3.2753036437246963,
+      "grad_norm": 0.37591297649902444,
+      "learning_rate": 1.2838801166745545e-05,
+      "loss": 0.2744,
+      "step": 809
+    },
+    {
+      "epoch": 3.279352226720648,
+      "grad_norm": 0.38894547495605436,
+      "learning_rate": 1.2786025175024216e-05,
+      "loss": 0.2712,
+      "step": 810
+    },
+    {
+      "epoch": 3.283400809716599,
+      "grad_norm": 0.38150876448525983,
+      "learning_rate": 1.2733306866053357e-05,
+      "loss": 0.2766,
+      "step": 811
+    },
+    {
+      "epoch": 3.2874493927125505,
+      "grad_norm": 0.37105040322424193,
+      "learning_rate": 1.2680646661367163e-05,
+      "loss": 0.2644,
+      "step": 812
+    },
+    {
+      "epoch": 3.291497975708502,
+      "grad_norm": 0.37931124520813964,
+      "learning_rate": 1.2628044982035219e-05,
+      "loss": 0.2671,
+      "step": 813
+    },
+    {
+      "epoch": 3.2955465587044532,
+      "grad_norm": 0.3959615523346858,
+      "learning_rate": 1.2575502248659155e-05,
+      "loss": 0.2752,
+      "step": 814
+    },
+    {
+      "epoch": 3.299595141700405,
+      "grad_norm": 0.37172480104481576,
+      "learning_rate": 1.2523018881369269e-05,
+      "loss": 0.2817,
+      "step": 815
+    },
+    {
+      "epoch": 3.3036437246963564,
+      "grad_norm": 0.400650562920213,
+      "learning_rate": 1.2470595299821176e-05,
+      "loss": 0.2643,
+      "step": 816
+    },
+    {
+      "epoch": 3.3076923076923075,
+      "grad_norm": 0.3706701445591225,
+      "learning_rate": 1.2418231923192416e-05,
+      "loss": 0.2755,
+      "step": 817
+    },
+    {
+      "epoch": 3.311740890688259,
+      "grad_norm": 0.35659126774985145,
+      "learning_rate": 1.2365929170179165e-05,
+      "loss": 0.2953,
+      "step": 818
+    },
+    {
+      "epoch": 3.3157894736842106,
+      "grad_norm": 0.39890607259601807,
+      "learning_rate": 1.2313687458992848e-05,
+      "loss": 0.2745,
+      "step": 819
+    },
+    {
+      "epoch": 3.3198380566801617,
+      "grad_norm": 0.3815824703036372,
+      "learning_rate": 1.2261507207356785e-05,
+      "loss": 0.2782,
+      "step": 820
+    },
+    {
+      "epoch": 3.3238866396761133,
+      "grad_norm": 0.3571240007226501,
+      "learning_rate": 1.220938883250288e-05,
+      "loss": 0.2736,
+      "step": 821
+    },
+    {
+      "epoch": 3.327935222672065,
+      "grad_norm": 0.4026991338646853,
+      "learning_rate": 1.215733275116827e-05,
+      "loss": 0.2946,
+      "step": 822
+    },
+    {
+      "epoch": 3.331983805668016,
+      "grad_norm": 0.3538953295466929,
+      "learning_rate": 1.2105339379592005e-05,
+      "loss": 0.2837,
+      "step": 823
+    },
+    {
+      "epoch": 3.3360323886639676,
+      "grad_norm": 0.3494380621624193,
+      "learning_rate": 1.2053409133511681e-05,
+      "loss": 0.2777,
+      "step": 824
+    },
+    {
+      "epoch": 3.340080971659919,
+      "grad_norm": 0.39524032144870497,
+      "learning_rate": 1.2001542428160173e-05,
+      "loss": 0.2645,
+      "step": 825
+    },
+    {
+      "epoch": 3.3441295546558703,
+      "grad_norm": 0.41310521626204877,
+      "learning_rate": 1.1949739678262278e-05,
+      "loss": 0.2825,
+      "step": 826
+    },
+    {
+      "epoch": 3.348178137651822,
+      "grad_norm": 0.3687757832621318,
+      "learning_rate": 1.189800129803139e-05,
+      "loss": 0.2627,
+      "step": 827
+    },
+    {
+      "epoch": 3.3522267206477734,
+      "grad_norm": 0.3764560466321167,
+      "learning_rate": 1.1846327701166228e-05,
+      "loss": 0.2652,
+      "step": 828
+    },
+    {
+      "epoch": 3.3562753036437245,
+      "grad_norm": 0.39500670267058374,
+      "learning_rate": 1.1794719300847493e-05,
+      "loss": 0.2926,
+      "step": 829
+    },
+    {
+      "epoch": 3.360323886639676,
+      "grad_norm": 0.3938528257996328,
+      "learning_rate": 1.1743176509734582e-05,
+      "loss": 0.2793,
+      "step": 830
+    },
+    {
+      "epoch": 3.3643724696356276,
+      "grad_norm": 0.4019965641557246,
+      "learning_rate": 1.1691699739962275e-05,
+      "loss": 0.2711,
+      "step": 831
+    },
+    {
+      "epoch": 3.3684210526315788,
+      "grad_norm": 0.38018985032505165,
+      "learning_rate": 1.1640289403137456e-05,
+      "loss": 0.2716,
+      "step": 832
+    },
+    {
+      "epoch": 3.3724696356275303,
+      "grad_norm": 0.41365461480986015,
+      "learning_rate": 1.1588945910335802e-05,
+      "loss": 0.2892,
+      "step": 833
+    },
+    {
+      "epoch": 3.376518218623482,
+      "grad_norm": 0.39262608285413,
+      "learning_rate": 1.1537669672098517e-05,
+      "loss": 0.2727,
+      "step": 834
+    },
+    {
+      "epoch": 3.380566801619433,
+      "grad_norm": 0.41167853541116267,
+      "learning_rate": 1.1486461098429036e-05,
+      "loss": 0.281,
+      "step": 835
+    },
+    {
+      "epoch": 3.3846153846153846,
+      "grad_norm": 0.48795488631282424,
+      "learning_rate": 1.1435320598789747e-05,
+      "loss": 0.2815,
+      "step": 836
+    },
+    {
+      "epoch": 3.388663967611336,
+      "grad_norm": 0.3937871960431845,
+      "learning_rate": 1.1384248582098736e-05,
+      "loss": 0.2691,
+      "step": 837
+    },
+    {
+      "epoch": 3.3927125506072873,
+      "grad_norm": 0.4138454988274364,
+      "learning_rate": 1.1333245456726467e-05,
+      "loss": 0.2724,
+      "step": 838
+    },
+    {
+      "epoch": 3.396761133603239,
+      "grad_norm": 0.3691019907657625,
+      "learning_rate": 1.128231163049258e-05,
+      "loss": 0.3074,
+      "step": 839
+    },
+    {
+      "epoch": 3.4008097165991904,
+      "grad_norm": 0.3672336922385863,
+      "learning_rate": 1.123144751066259e-05,
+      "loss": 0.2775,
+      "step": 840
+    },
+    {
+      "epoch": 3.4048582995951415,
+      "grad_norm": 0.4010371878031593,
+      "learning_rate": 1.1180653503944657e-05,
+      "loss": 0.2754,
+      "step": 841
+    },
+    {
+      "epoch": 3.408906882591093,
+      "grad_norm": 0.35891037125883235,
+      "learning_rate": 1.1129930016486287e-05,
+      "loss": 0.2931,
+      "step": 842
+    },
+    {
+      "epoch": 3.4129554655870447,
+      "grad_norm": 0.39082553545420207,
+      "learning_rate": 1.107927745387114e-05,
+      "loss": 0.2788,
+      "step": 843
+    },
+    {
+      "epoch": 3.417004048582996,
+      "grad_norm": 0.374792884419042,
+      "learning_rate": 1.1028696221115766e-05,
+      "loss": 0.2848,
+      "step": 844
+    },
+    {
+      "epoch": 3.4210526315789473,
+      "grad_norm": 0.3483593350499838,
+      "learning_rate": 1.0978186722666341e-05,
+      "loss": 0.2754,
+      "step": 845
+    },
+    {
+      "epoch": 3.425101214574899,
+      "grad_norm": 0.3655236372894745,
+      "learning_rate": 1.0927749362395475e-05,
+      "loss": 0.2788,
+      "step": 846
+    },
+    {
+      "epoch": 3.42914979757085,
+      "grad_norm": 0.36829927847813754,
+      "learning_rate": 1.0877384543598954e-05,
+      "loss": 0.284,
+      "step": 847
+    },
+    {
+      "epoch": 3.4331983805668016,
+      "grad_norm": 0.3684217144160631,
+      "learning_rate": 1.0827092668992529e-05,
+      "loss": 0.2737,
+      "step": 848
+    },
+    {
+      "epoch": 3.437246963562753,
+      "grad_norm": 0.38858509192657487,
+      "learning_rate": 1.077687414070869e-05,
+      "loss": 0.2735,
+      "step": 849
+    },
+    {
+      "epoch": 3.4412955465587043,
+      "grad_norm": 0.37457439240568835,
+      "learning_rate": 1.0726729360293451e-05,
+      "loss": 0.2534,
+      "step": 850
+    },
+    {
+      "epoch": 3.445344129554656,
+      "grad_norm": 0.4045892632217714,
+      "learning_rate": 1.0676658728703147e-05,
+      "loss": 0.2892,
+      "step": 851
+    },
+    {
+      "epoch": 3.4493927125506074,
+      "grad_norm": 0.3781884398145671,
+      "learning_rate": 1.0626662646301198e-05,
+      "loss": 0.2724,
+      "step": 852
+    },
+    {
+      "epoch": 3.4534412955465585,
+      "grad_norm": 0.37243897860384595,
+      "learning_rate": 1.0576741512854958e-05,
+      "loss": 0.276,
+      "step": 853
+    },
+    {
+      "epoch": 3.45748987854251,
+      "grad_norm": 0.41267100800275053,
+      "learning_rate": 1.052689572753247e-05,
+      "loss": 0.2947,
+      "step": 854
+    },
+    {
+      "epoch": 3.4615384615384617,
+      "grad_norm": 0.3848434842169259,
+      "learning_rate": 1.0477125688899328e-05,
+      "loss": 0.2826,
+      "step": 855
+    },
+    {
+      "epoch": 3.465587044534413,
+      "grad_norm": 0.38383377585505335,
+      "learning_rate": 1.0427431794915406e-05,
+      "loss": 0.2999,
+      "step": 856
+    },
+    {
+      "epoch": 3.4696356275303644,
+      "grad_norm": 0.4212140580463511,
+      "learning_rate": 1.0377814442931769e-05,
+      "loss": 0.2806,
+      "step": 857
+    },
+    {
+      "epoch": 3.473684210526316,
+      "grad_norm": 0.3732117717910858,
+      "learning_rate": 1.0328274029687439e-05,
+      "loss": 0.2822,
+      "step": 858
+    },
+    {
+      "epoch": 3.477732793522267,
+      "grad_norm": 0.3678359802356167,
+      "learning_rate": 1.0278810951306253e-05,
+      "loss": 0.2845,
+      "step": 859
+    },
+    {
+      "epoch": 3.4817813765182186,
+      "grad_norm": 0.4075455973201401,
+      "learning_rate": 1.0229425603293647e-05,
+      "loss": 0.2794,
+      "step": 860
+    },
+    {
+      "epoch": 3.48582995951417,
+      "grad_norm": 0.3710607055874896,
+      "learning_rate": 1.0180118380533558e-05,
+      "loss": 0.2694,
+      "step": 861
+    },
+    {
+      "epoch": 3.4898785425101213,
+      "grad_norm": 0.36049087199319785,
+      "learning_rate": 1.0130889677285225e-05,
+      "loss": 0.2825,
+      "step": 862
+    },
+    {
+      "epoch": 3.493927125506073,
+      "grad_norm": 0.3831440731390451,
+      "learning_rate": 1.008173988718005e-05,
+      "loss": 0.2648,
+      "step": 863
+    },
+    {
+      "epoch": 3.4979757085020244,
+      "grad_norm": 0.38503508927029595,
+      "learning_rate": 1.0032669403218439e-05,
+      "loss": 0.2639,
+      "step": 864
+    },
+    {
+      "epoch": 3.5020242914979756,
+      "grad_norm": 0.36375390259517354,
+      "learning_rate": 9.983678617766677e-06,
+      "loss": 0.2788,
+      "step": 865
+    },
+    {
+      "epoch": 3.506072874493927,
+      "grad_norm": 0.3835140792487109,
+      "learning_rate": 9.934767922553788e-06,
+      "loss": 0.2732,
+      "step": 866
+    },
+    {
+      "epoch": 3.5101214574898787,
+      "grad_norm": 0.36002744005307397,
+      "learning_rate": 9.885937708668371e-06,
+      "loss": 0.2699,
+      "step": 867
+    },
+    {
+      "epoch": 3.51417004048583,
+      "grad_norm": 0.3670352541817853,
+      "learning_rate": 9.837188366555524e-06,
+      "loss": 0.2902,
+      "step": 868
+    },
+    {
+      "epoch": 3.5182186234817814,
+      "grad_norm": 0.365662589157744,
+      "learning_rate": 9.788520286013691e-06,
+      "loss": 0.2742,
+      "step": 869
+    },
+    {
+      "epoch": 3.522267206477733,
+      "grad_norm": 0.3715159897612988,
+      "learning_rate": 9.739933856191537e-06,
+      "loss": 0.2774,
+      "step": 870
+    },
+    {
+      "epoch": 3.526315789473684,
+      "grad_norm": 0.36953059105092745,
+      "learning_rate": 9.691429465584866e-06,
+      "loss": 0.2519,
+      "step": 871
+    },
+    {
+      "epoch": 3.5303643724696356,
+      "grad_norm": 0.36530945244768737,
+      "learning_rate": 9.643007502033492e-06,
+      "loss": 0.2944,
+      "step": 872
+    },
+    {
+      "epoch": 3.534412955465587,
+      "grad_norm": 0.37561008697637666,
+      "learning_rate": 9.594668352718152e-06,
+      "loss": 0.2729,
+      "step": 873
+    },
+    {
+      "epoch": 3.5384615384615383,
+      "grad_norm": 0.38131418137322487,
+      "learning_rate": 9.546412404157383e-06,
+      "loss": 0.2785,
+      "step": 874
+    },
+    {
+      "epoch": 3.54251012145749,
+      "grad_norm": 0.3799041572446109,
+      "learning_rate": 9.498240042204471e-06,
+      "loss": 0.2988,
+      "step": 875
+    },
+    {
+      "epoch": 3.5465587044534415,
+      "grad_norm": 0.35778972838982903,
+      "learning_rate": 9.450151652044339e-06,
+      "loss": 0.2674,
+      "step": 876
+    },
+    {
+      "epoch": 3.5506072874493926,
+      "grad_norm": 0.45161757390302626,
+      "learning_rate": 9.402147618190472e-06,
+      "loss": 0.2605,
+      "step": 877
+    },
+    {
+      "epoch": 3.554655870445344,
+      "grad_norm": 0.4163480667352032,
+      "learning_rate": 9.354228324481841e-06,
+      "loss": 0.2898,
+      "step": 878
+    },
+    {
+      "epoch": 3.5587044534412957,
+      "grad_norm": 0.38475675918299035,
+      "learning_rate": 9.306394154079852e-06,
+      "loss": 0.2811,
+      "step": 879
+    },
+    {
+      "epoch": 3.562753036437247,
+      "grad_norm": 0.36508013369146075,
+      "learning_rate": 9.258645489465254e-06,
+      "loss": 0.2702,
+      "step": 880
+    },
+    {
+      "epoch": 3.5668016194331984,
+      "grad_norm": 0.3642502186427338,
+      "learning_rate": 9.21098271243509e-06,
+      "loss": 0.2728,
+      "step": 881
+    },
+    {
+      "epoch": 3.57085020242915,
+      "grad_norm": 0.3612121681435541,
+      "learning_rate": 9.163406204099659e-06,
+      "loss": 0.2646,
+      "step": 882
+    },
+    {
+      "epoch": 3.574898785425101,
+      "grad_norm": 0.3586372471905153,
+      "learning_rate": 9.115916344879452e-06,
+      "loss": 0.2618,
+      "step": 883
+    },
+    {
+      "epoch": 3.5789473684210527,
+      "grad_norm": 0.37765536311497433,
+      "learning_rate": 9.068513514502126e-06,
+      "loss": 0.2568,
+      "step": 884
+    },
+    {
+      "epoch": 3.582995951417004,
+      "grad_norm": 0.37806398492936794,
+      "learning_rate": 9.021198091999439e-06,
+      "loss": 0.2828,
+      "step": 885
+    },
+    {
+      "epoch": 3.5870445344129553,
+      "grad_norm": 0.3729387043208613,
+      "learning_rate": 8.973970455704251e-06,
+      "loss": 0.2728,
+      "step": 886
+    },
+    {
+      "epoch": 3.591093117408907,
+      "grad_norm": 0.3762780040640659,
+      "learning_rate": 8.926830983247495e-06,
+      "loss": 0.2757,
+      "step": 887
+    },
+    {
+      "epoch": 3.5951417004048585,
+      "grad_norm": 0.36710025359624043,
+      "learning_rate": 8.879780051555116e-06,
+      "loss": 0.2883,
+      "step": 888
+    },
+    {
+      "epoch": 3.5991902834008096,
+      "grad_norm": 0.38748134229617426,
+      "learning_rate": 8.83281803684512e-06,
+      "loss": 0.2666,
+      "step": 889
+    },
+    {
+      "epoch": 3.603238866396761,
+      "grad_norm": 0.3999553372861661,
+      "learning_rate": 8.785945314624532e-06,
+      "loss": 0.2797,
+      "step": 890
+    },
+    {
+      "epoch": 3.6072874493927127,
+      "grad_norm": 0.3696051622703846,
+      "learning_rate": 8.739162259686385e-06,
+      "loss": 0.2875,
+      "step": 891
+    },
+    {
+      "epoch": 3.611336032388664,
+      "grad_norm": 0.38570416422671316,
+      "learning_rate": 8.692469246106734e-06,
+      "loss": 0.2948,
+      "step": 892
+    },
+    {
+      "epoch": 3.6153846153846154,
+      "grad_norm": 0.3726879789369337,
+      "learning_rate": 8.645866647241677e-06,
+      "loss": 0.2661,
+      "step": 893
+    },
+    {
+      "epoch": 3.619433198380567,
+      "grad_norm": 0.3696511984037539,
+      "learning_rate": 8.599354835724348e-06,
+      "loss": 0.2712,
+      "step": 894
+    },
+    {
+      "epoch": 3.623481781376518,
+      "grad_norm": 0.36357395640387763,
+      "learning_rate": 8.552934183461958e-06,
+      "loss": 0.2771,
+      "step": 895
+    },
+    {
+      "epoch": 3.6275303643724697,
+      "grad_norm": 0.40226393098121593,
+      "learning_rate": 8.506605061632805e-06,
+      "loss": 0.286,
+      "step": 896
+    },
+    {
+      "epoch": 3.6315789473684212,
+      "grad_norm": 0.38051839663783976,
+      "learning_rate": 8.460367840683308e-06,
+      "loss": 0.2795,
+      "step": 897
+    },
+    {
+      "epoch": 3.6356275303643724,
+      "grad_norm": 0.37700854897742564,
+      "learning_rate": 8.414222890325065e-06,
+      "loss": 0.2643,
+      "step": 898
+    },
+    {
+      "epoch": 3.639676113360324,
+      "grad_norm": 0.39153730115923524,
+      "learning_rate": 8.368170579531855e-06,
+      "loss": 0.2764,
+      "step": 899
+    },
+    {
+      "epoch": 3.6437246963562755,
+      "grad_norm": 0.3835900351362234,
+      "learning_rate": 8.322211276536733e-06,
+      "loss": 0.2938,
+      "step": 900
+    },
+    {
+      "epoch": 3.6477732793522266,
+      "grad_norm": 0.3629030814154145,
+      "learning_rate": 8.27634534882907e-06,
+      "loss": 0.2748,
+      "step": 901
+    },
+    {
+      "epoch": 3.651821862348178,
+      "grad_norm": 0.3585946663692,
+      "learning_rate": 8.230573163151585e-06,
+      "loss": 0.2785,
+      "step": 902
+    },
+    {
+      "epoch": 3.6558704453441297,
+      "grad_norm": 0.35160321296504515,
+      "learning_rate": 8.184895085497466e-06,
+      "loss": 0.2877,
+      "step": 903
+    },
+    {
+      "epoch": 3.659919028340081,
+      "grad_norm": 0.36221545229621815,
+      "learning_rate": 8.139311481107401e-06,
+      "loss": 0.2727,
+      "step": 904
+    },
+    {
+      "epoch": 3.6639676113360324,
+      "grad_norm": 0.37314553815905277,
+      "learning_rate": 8.093822714466686e-06,
+      "loss": 0.2809,
+      "step": 905
+    },
+    {
+      "epoch": 3.668016194331984,
+      "grad_norm": 0.3701087632667543,
+      "learning_rate": 8.048429149302271e-06,
+      "loss": 0.2707,
+      "step": 906
+    },
+    {
+      "epoch": 3.672064777327935,
+      "grad_norm": 0.3718585132163012,
+      "learning_rate": 8.003131148579902e-06,
+      "loss": 0.2679,
+      "step": 907
+    },
+    {
+      "epoch": 3.6761133603238867,
+      "grad_norm": 0.3707324864677274,
+      "learning_rate": 7.957929074501182e-06,
+      "loss": 0.2947,
+      "step": 908
+    },
+    {
+      "epoch": 3.6801619433198383,
+      "grad_norm": 0.36185433330604483,
+      "learning_rate": 7.912823288500695e-06,
+      "loss": 0.2627,
+      "step": 909
+    },
+    {
+      "epoch": 3.6842105263157894,
+      "grad_norm": 0.36045985815144416,
+      "learning_rate": 7.867814151243102e-06,
+      "loss": 0.2756,
+      "step": 910
+    },
+    {
+      "epoch": 3.688259109311741,
+      "grad_norm": 0.3573391268647463,
+      "learning_rate": 7.822902022620267e-06,
+      "loss": 0.2761,
+      "step": 911
+    },
+    {
+      "epoch": 3.6923076923076925,
+      "grad_norm": 0.375679954011263,
+      "learning_rate": 7.778087261748377e-06,
+      "loss": 0.2932,
+      "step": 912
+    },
+    {
+      "epoch": 3.6963562753036436,
+      "grad_norm": 0.36706858218462174,
+      "learning_rate": 7.733370226965056e-06,
+      "loss": 0.2885,
+      "step": 913
+    },
+    {
+      "epoch": 3.700404858299595,
+      "grad_norm": 0.37993703570434856,
+      "learning_rate": 7.688751275826527e-06,
+      "loss": 0.2665,
+      "step": 914
+    },
+    {
+      "epoch": 3.7044534412955468,
+      "grad_norm": 0.37438993826743416,
+      "learning_rate": 7.644230765104735e-06,
+      "loss": 0.2831,
+      "step": 915
+    },
+    {
+      "epoch": 3.708502024291498,
+      "grad_norm": 0.3739670651395066,
+      "learning_rate": 7.599809050784502e-06,
+      "loss": 0.2491,
+      "step": 916
+    },
+    {
+      "epoch": 3.7125506072874495,
+      "grad_norm": 0.3668734262504267,
+      "learning_rate": 7.555486488060661e-06,
+      "loss": 0.2796,
+      "step": 917
+    },
+    {
+      "epoch": 3.716599190283401,
+      "grad_norm": 0.3580127089782473,
+      "learning_rate": 7.511263431335249e-06,
+      "loss": 0.2928,
+      "step": 918
+    },
+    {
+      "epoch": 3.720647773279352,
+      "grad_norm": 0.3868325419960579,
+      "learning_rate": 7.467140234214656e-06,
+      "loss": 0.2684,
+      "step": 919
+    },
+    {
+      "epoch": 3.7246963562753037,
+      "grad_norm": 0.3589005996471209,
+      "learning_rate": 7.423117249506777e-06,
+      "loss": 0.2776,
+      "step": 920
+    },
+    {
+      "epoch": 3.7287449392712553,
+      "grad_norm": 0.35746080026326266,
+      "learning_rate": 7.379194829218231e-06,
+      "loss": 0.288,
+      "step": 921
+    },
+    {
+      "epoch": 3.7327935222672064,
+      "grad_norm": 0.3734397413255103,
+      "learning_rate": 7.335373324551518e-06,
+      "loss": 0.292,
+      "step": 922
+    },
+    {
+      "epoch": 3.736842105263158,
+      "grad_norm": 0.36837702902793606,
+      "learning_rate": 7.291653085902224e-06,
+      "loss": 0.269,
+      "step": 923
+    },
+    {
+      "epoch": 3.7408906882591095,
+      "grad_norm": 0.3537809394451439,
+      "learning_rate": 7.248034462856213e-06,
+      "loss": 0.2933,
+      "step": 924
+    },
+    {
+      "epoch": 3.7449392712550607,
+      "grad_norm": 0.3792234991806172,
+      "learning_rate": 7.20451780418683e-06,
+      "loss": 0.2657,
+      "step": 925
+    },
+    {
+      "epoch": 3.748987854251012,
+      "grad_norm": 0.37832291452477357,
+      "learning_rate": 7.161103457852128e-06,
+      "loss": 0.2905,
+      "step": 926
+    },
+    {
+      "epoch": 3.753036437246964,
+      "grad_norm": 0.36622249223167347,
+      "learning_rate": 7.117791770992046e-06,
+      "loss": 0.2743,
+      "step": 927
+    },
+    {
+      "epoch": 3.757085020242915,
+      "grad_norm": 0.36420376948135047,
+      "learning_rate": 7.074583089925688e-06,
+      "loss": 0.2557,
+      "step": 928
+    },
+    {
+      "epoch": 3.7611336032388665,
+      "grad_norm": 0.36702305957323267,
+      "learning_rate": 7.0314777601485155e-06,
+      "loss": 0.2736,
+      "step": 929
+    },
+    {
+      "epoch": 3.765182186234818,
+      "grad_norm": 0.38444429212848014,
+      "learning_rate": 6.988476126329604e-06,
+      "loss": 0.2641,
+      "step": 930
+    },
+    {
+      "epoch": 3.769230769230769,
+      "grad_norm": 0.3577670345252928,
+      "learning_rate": 6.945578532308857e-06,
+      "loss": 0.2653,
+      "step": 931
+    },
+    {
+      "epoch": 3.7732793522267207,
+      "grad_norm": 0.37619228542031635,
+      "learning_rate": 6.902785321094301e-06,
+      "loss": 0.2744,
+      "step": 932
+    },
+    {
+      "epoch": 3.7773279352226723,
+      "grad_norm": 0.34735095437242436,
+      "learning_rate": 6.8600968348593115e-06,
+      "loss": 0.273,
+      "step": 933
+    },
+    {
+      "epoch": 3.7813765182186234,
+      "grad_norm": 0.3665503455420503,
+      "learning_rate": 6.817513414939892e-06,
+      "loss": 0.268,
+      "step": 934
+    },
+    {
+      "epoch": 3.785425101214575,
+      "grad_norm": 0.37074378651640955,
+      "learning_rate": 6.775035401831924e-06,
+      "loss": 0.2712,
+      "step": 935
+    },
+    {
+      "epoch": 3.7894736842105265,
+      "grad_norm": 0.3773611453609929,
+      "learning_rate": 6.732663135188473e-06,
+      "loss": 0.286,
+      "step": 936
+    },
+    {
+      "epoch": 3.7935222672064777,
+      "grad_norm": 0.3586537471567054,
+      "learning_rate": 6.690396953817055e-06,
+      "loss": 0.2795,
+      "step": 937
+    },
+    {
+      "epoch": 3.7975708502024292,
+      "grad_norm": 0.38295065368423886,
+      "learning_rate": 6.648237195676927e-06,
+      "loss": 0.2561,
+      "step": 938
+    },
+    {
+      "epoch": 3.801619433198381,
+      "grad_norm": 0.3940591899631688,
+      "learning_rate": 6.606184197876395e-06,
+      "loss": 0.2638,
+      "step": 939
+    },
+    {
+      "epoch": 3.805668016194332,
+      "grad_norm": 0.38001558083903697,
+      "learning_rate": 6.564238296670109e-06,
+      "loss": 0.2787,
+      "step": 940
+    },
+    {
+      "epoch": 3.8097165991902835,
+      "grad_norm": 0.37582148964391643,
+      "learning_rate": 6.52239982745638e-06,
+      "loss": 0.2431,
+      "step": 941
+    },
+    {
+      "epoch": 3.813765182186235,
+      "grad_norm": 0.3607884921981072,
+      "learning_rate": 6.480669124774484e-06,
+      "loss": 0.284,
+      "step": 942
+    },
+    {
+      "epoch": 3.817813765182186,
+      "grad_norm": 0.3486527264694192,
+      "learning_rate": 6.439046522302013e-06,
+      "loss": 0.2915,
+      "step": 943
+    },
+    {
+      "epoch": 3.8218623481781377,
+      "grad_norm": 0.3773970406434678,
+      "learning_rate": 6.397532352852191e-06,
+      "loss": 0.2632,
+      "step": 944
+    },
+    {
+      "epoch": 3.8259109311740893,
+      "grad_norm": 0.366291853755756,
+      "learning_rate": 6.356126948371202e-06,
+      "loss": 0.2824,
+      "step": 945
+    },
+    {
+      "epoch": 3.8299595141700404,
+      "grad_norm": 0.36137364231300206,
+      "learning_rate": 6.314830639935559e-06,
+      "loss": 0.2713,
+      "step": 946
+    },
+    {
+      "epoch": 3.834008097165992,
+      "grad_norm": 0.35743170762042414,
+      "learning_rate": 6.273643757749448e-06,
+      "loss": 0.2788,
+      "step": 947
+    },
+    {
+      "epoch": 3.8380566801619436,
+      "grad_norm": 0.3614638941082431,
+      "learning_rate": 6.2325666311420856e-06,
+      "loss": 0.2792,
+      "step": 948
+    },
+    {
+      "epoch": 3.8421052631578947,
+      "grad_norm": 0.35682694163549633,
+      "learning_rate": 6.1915995885650675e-06,
+      "loss": 0.2762,
+      "step": 949
+    },
+    {
+      "epoch": 3.8461538461538463,
+      "grad_norm": 0.3629794734985952,
+      "learning_rate": 6.150742957589784e-06,
+      "loss": 0.2967,
+      "step": 950
+    },
+    {
+      "epoch": 3.850202429149798,
+      "grad_norm": 0.360470587989926,
+      "learning_rate": 6.109997064904774e-06,
+      "loss": 0.2705,
+      "step": 951
+    },
+    {
+      "epoch": 3.854251012145749,
+      "grad_norm": 0.3773790206121749,
+      "learning_rate": 6.069362236313092e-06,
+      "loss": 0.2771,
+      "step": 952
+    },
+    {
+      "epoch": 3.8582995951417005,
+      "grad_norm": 0.3933747431734778,
+      "learning_rate": 6.0288387967297565e-06,
+      "loss": 0.269,
+      "step": 953
+    },
+    {
+      "epoch": 3.862348178137652,
+      "grad_norm": 0.34492997683986226,
+      "learning_rate": 5.988427070179108e-06,
+      "loss": 0.2891,
+      "step": 954
+    },
+    {
+      "epoch": 3.866396761133603,
+      "grad_norm": 0.34391029561515846,
+      "learning_rate": 5.948127379792234e-06,
+      "loss": 0.2723,
+      "step": 955
+    },
+    {
+      "epoch": 3.8704453441295548,
+      "grad_norm": 0.35207024011823285,
+      "learning_rate": 5.90794004780439e-06,
+      "loss": 0.2685,
+      "step": 956
+    },
+    {
+      "epoch": 3.8744939271255063,
+      "grad_norm": 0.361022033733948,
+      "learning_rate": 5.867865395552405e-06,
+      "loss": 0.274,
+      "step": 957
+    },
+    {
+      "epoch": 3.8785425101214575,
+      "grad_norm": 0.36959867807304286,
+      "learning_rate": 5.8279037434721364e-06,
+      "loss": 0.2695,
+      "step": 958
+    },
+    {
+      "epoch": 3.882591093117409,
+      "grad_norm": 0.35581666803591006,
+      "learning_rate": 5.788055411095892e-06,
+      "loss": 0.2646,
+      "step": 959
+    },
+    {
+      "epoch": 3.8866396761133606,
+      "grad_norm": 0.3571339401089168,
+      "learning_rate": 5.7483207170498625e-06,
+      "loss": 0.2827,
+      "step": 960
+    },
+    {
+      "epoch": 3.8906882591093117,
+      "grad_norm": 0.39418735281356976,
+      "learning_rate": 5.708699979051613e-06,
+      "loss": 0.2809,
+      "step": 961
+    },
+    {
+      "epoch": 3.8947368421052633,
+      "grad_norm": 0.353825507709019,
+      "learning_rate": 5.66919351390751e-06,
+      "loss": 0.2737,
+      "step": 962
+    },
+    {
+      "epoch": 3.898785425101215,
+      "grad_norm": 0.35985385488997473,
+      "learning_rate": 5.629801637510188e-06,
+      "loss": 0.2718,
+      "step": 963
+    },
+    {
+      "epoch": 3.902834008097166,
+      "grad_norm": 0.363517015426541,
+      "learning_rate": 5.590524664836043e-06,
+      "loss": 0.2594,
+      "step": 964
+    },
+    {
+      "epoch": 3.9068825910931175,
+      "grad_norm": 0.3714450303436371,
+      "learning_rate": 5.551362909942708e-06,
+      "loss": 0.2741,
+      "step": 965
+    },
+    {
+      "epoch": 3.910931174089069,
+      "grad_norm": 0.3575602173611453,
+      "learning_rate": 5.51231668596653e-06,
+      "loss": 0.259,
+      "step": 966
+    },
+    {
+      "epoch": 3.91497975708502,
+      "grad_norm": 0.3491051435208793,
+      "learning_rate": 5.47338630512007e-06,
+      "loss": 0.2711,
+      "step": 967
+    },
+    {
+      "epoch": 3.919028340080972,
+      "grad_norm": 0.3676478735330877,
+      "learning_rate": 5.434572078689617e-06,
+      "loss": 0.2707,
+      "step": 968
+    },
+    {
+      "epoch": 3.9230769230769234,
+      "grad_norm": 0.3650525543042498,
+      "learning_rate": 5.395874317032694e-06,
+      "loss": 0.2692,
+      "step": 969
+    },
+    {
+      "epoch": 3.9271255060728745,
+      "grad_norm": 0.3523927307698294,
+      "learning_rate": 5.357293329575573e-06,
+      "loss": 0.2615,
+      "step": 970
+    },
+    {
+      "epoch": 3.931174089068826,
+      "grad_norm": 0.3240787655334934,
+      "learning_rate": 5.318829424810797e-06,
+      "loss": 0.2928,
+      "step": 971
+    },
+    {
+      "epoch": 3.9352226720647776,
+      "grad_norm": 0.3615626769368519,
+      "learning_rate": 5.280482910294728e-06,
+      "loss": 0.2502,
+      "step": 972
+    },
+    {
+      "epoch": 3.9392712550607287,
+      "grad_norm": 0.3595524393575295,
+      "learning_rate": 5.242254092645078e-06,
+      "loss": 0.2653,
+      "step": 973
+    },
+    {
+      "epoch": 3.9433198380566803,
+      "grad_norm": 0.36296674648506005,
+      "learning_rate": 5.2041432775384445e-06,
+      "loss": 0.2683,
+      "step": 974
+    },
+    {
+      "epoch": 3.9473684210526314,
+      "grad_norm": 0.3788898692977803,
+      "learning_rate": 5.166150769707892e-06,
+      "loss": 0.2734,
+      "step": 975
+    },
+    {
+      "epoch": 3.951417004048583,
+      "grad_norm": 0.3384850472901298,
+      "learning_rate": 5.128276872940503e-06,
+      "loss": 0.2773,
+      "step": 976
+    },
+    {
+      "epoch": 3.9554655870445345,
+      "grad_norm": 0.35990213269151394,
+      "learning_rate": 5.090521890074953e-06,
+      "loss": 0.2739,
+      "step": 977
+    },
+    {
+      "epoch": 3.9595141700404857,
+      "grad_norm": 0.3466063749591334,
+      "learning_rate": 5.0528861229990675e-06,
+      "loss": 0.276,
+      "step": 978
+    },
+    {
+      "epoch": 3.9635627530364372,
+      "grad_norm": 0.36904293323284715,
+      "learning_rate": 5.015369872647442e-06,
+      "loss": 0.2637,
+      "step": 979
+    },
+    {
+      "epoch": 3.967611336032389,
+      "grad_norm": 0.3518533143954396,
+      "learning_rate": 4.9779734389990195e-06,
+      "loss": 0.2849,
+      "step": 980
+    },
+    {
+      "epoch": 3.97165991902834,
+      "grad_norm": 0.3556627751148852,
+      "learning_rate": 4.940697121074675e-06,
+      "loss": 0.2724,
+      "step": 981
+    },
+    {
+      "epoch": 3.9757085020242915,
+      "grad_norm": 0.34711117115023027,
+      "learning_rate": 4.903541216934864e-06,
+      "loss": 0.2781,
+      "step": 982
+    },
+    {
+      "epoch": 3.979757085020243,
+      "grad_norm": 0.3632355932931168,
+      "learning_rate": 4.8665060236772e-06,
+      "loss": 0.273,
+      "step": 983
+    },
+    {
+      "epoch": 3.983805668016194,
+      "grad_norm": 0.36473205047441853,
+      "learning_rate": 4.829591837434102e-06,
+      "loss": 0.2719,
+      "step": 984
+    },
+    {
+      "epoch": 3.9878542510121457,
+      "grad_norm": 0.37106282945515384,
+      "learning_rate": 4.792798953370424e-06,
+      "loss": 0.2723,
+      "step": 985
+    },
+    {
+      "epoch": 3.9919028340080973,
+      "grad_norm": 0.3652521291947608,
+      "learning_rate": 4.756127665681083e-06,
+      "loss": 0.2607,
+      "step": 986
+    },
+    {
+      "epoch": 3.9959514170040484,
+      "grad_norm": 0.3444158363917552,
+      "learning_rate": 4.719578267588725e-06,
+      "loss": 0.2711,
+      "step": 987
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 0.3522648138932278,
+      "learning_rate": 4.6831510513413525e-06,
+      "loss": 0.2674,
+      "step": 988
+    },
+    {
+      "epoch": 4.004048582995951,
+      "grad_norm": 0.6267371525116809,
+      "learning_rate": 4.6468463082100176e-06,
+      "loss": 0.1981,
+      "step": 989
+    },
+    {
+      "epoch": 4.008097165991903,
+      "grad_norm": 0.4733790144284881,
+      "learning_rate": 4.610664328486485e-06,
+      "loss": 0.2004,
+      "step": 990
+    },
+    {
+      "epoch": 4.012145748987854,
+      "grad_norm": 0.38307086215483366,
+      "learning_rate": 4.5746054014809e-06,
+      "loss": 0.1721,
+      "step": 991
+    },
+    {
+      "epoch": 4.016194331983805,
+      "grad_norm": 0.4223486529987876,
+      "learning_rate": 4.5386698155194766e-06,
+      "loss": 0.1963,
+      "step": 992
+    },
+    {
+      "epoch": 4.020242914979757,
+      "grad_norm": 0.5138593272385482,
+      "learning_rate": 4.502857857942204e-06,
+      "loss": 0.1787,
+      "step": 993
+    },
+    {
+      "epoch": 4.0242914979757085,
+      "grad_norm": 0.6064469261171828,
+      "learning_rate": 4.467169815100549e-06,
+      "loss": 0.2015,
+      "step": 994
+    },
+    {
+      "epoch": 4.02834008097166,
+      "grad_norm": 0.5171334651039372,
+      "learning_rate": 4.431605972355137e-06,
+      "loss": 0.1831,
+      "step": 995
+    },
+    {
+      "epoch": 4.032388663967612,
+      "grad_norm": 0.4002862800078213,
+      "learning_rate": 4.396166614073516e-06,
+      "loss": 0.1799,
+      "step": 996
+    },
+    {
+      "epoch": 4.036437246963563,
+      "grad_norm": 0.4058191730143565,
+      "learning_rate": 4.360852023627846e-06,
+      "loss": 0.1985,
+      "step": 997
+    },
+    {
+      "epoch": 4.040485829959514,
+      "grad_norm": 0.47100668892934044,
+      "learning_rate": 4.325662483392656e-06,
+      "loss": 0.2168,
+      "step": 998
+    },
+    {
+      "epoch": 4.044534412955466,
+      "grad_norm": 0.4859099981082729,
+      "learning_rate": 4.29059827474257e-06,
+      "loss": 0.1889,
+      "step": 999
+    },
+    {
+      "epoch": 4.048582995951417,
+      "grad_norm": 0.42012200773908825,
+      "learning_rate": 4.255659678050063e-06,
+      "loss": 0.1757,
+      "step": 1000
+    },
+    {
+      "epoch": 4.052631578947368,
+      "grad_norm": 0.34564340622533524,
+      "learning_rate": 4.22084697268323e-06,
+      "loss": 0.1882,
+      "step": 1001
+    },
+    {
+      "epoch": 4.05668016194332,
+      "grad_norm": 0.3615063792551274,
+      "learning_rate": 4.186160437003535e-06,
+      "loss": 0.1753,
+      "step": 1002
+    },
+    {
+      "epoch": 4.060728744939271,
+      "grad_norm": 0.3905972822461221,
+      "learning_rate": 4.151600348363582e-06,
+      "loss": 0.1933,
+      "step": 1003
+    },
+    {
+      "epoch": 4.064777327935222,
+      "grad_norm": 0.4555032839109062,
+      "learning_rate": 4.11716698310493e-06,
+      "loss": 0.1841,
+      "step": 1004
+    },
+    {
+      "epoch": 4.068825910931174,
+      "grad_norm": 0.4126472795784923,
+      "learning_rate": 4.082860616555848e-06,
+      "loss": 0.1995,
+      "step": 1005
+    },
+    {
+      "epoch": 4.0728744939271255,
+      "grad_norm": 0.38085350964281234,
+      "learning_rate": 4.048681523029125e-06,
+      "loss": 0.2029,
+      "step": 1006
+    },
+    {
+      "epoch": 4.076923076923077,
+      "grad_norm": 0.3750242070953552,
+      "learning_rate": 4.014629975819882e-06,
+      "loss": 0.1872,
+      "step": 1007
+    },
+    {
+      "epoch": 4.080971659919029,
+      "grad_norm": 0.35294207327318694,
+      "learning_rate": 3.9807062472033855e-06,
+      "loss": 0.2007,
+      "step": 1008
+    },
+    {
+      "epoch": 4.08502024291498,
+      "grad_norm": 0.3634518972453983,
+      "learning_rate": 3.94691060843287e-06,
+      "loss": 0.1995,
+      "step": 1009
+    },
+    {
+      "epoch": 4.089068825910931,
+      "grad_norm": 0.35752185992970653,
+      "learning_rate": 3.913243329737353e-06,
+      "loss": 0.198,
+      "step": 1010
+    },
+    {
+      "epoch": 4.093117408906883,
+      "grad_norm": 0.3609569776853851,
+      "learning_rate": 3.879704680319503e-06,
+      "loss": 0.1746,
+      "step": 1011
+    },
+    {
+      "epoch": 4.097165991902834,
+      "grad_norm": 0.35319047254444424,
+      "learning_rate": 3.846294928353471e-06,
+      "loss": 0.1905,
+      "step": 1012
+    },
+    {
+      "epoch": 4.101214574898785,
+      "grad_norm": 0.3470905057727203,
+      "learning_rate": 3.8130143409827345e-06,
+      "loss": 0.174,
+      "step": 1013
+    },
+    {
+      "epoch": 4.105263157894737,
+      "grad_norm": 0.3680929137296742,
+      "learning_rate": 3.7798631843179887e-06,
+      "loss": 0.1863,
+      "step": 1014
+    },
+    {
+      "epoch": 4.109311740890688,
+      "grad_norm": 0.35071330326127936,
+      "learning_rate": 3.7468417234349973e-06,
+      "loss": 0.2137,
+      "step": 1015
+    },
+    {
+      "epoch": 4.113360323886639,
+      "grad_norm": 0.3323765040008887,
+      "learning_rate": 3.7139502223724823e-06,
+      "loss": 0.2115,
+      "step": 1016
+    },
+    {
+      "epoch": 4.117408906882591,
+      "grad_norm": 0.3261028439882454,
+      "learning_rate": 3.6811889441300095e-06,
+      "loss": 0.1896,
+      "step": 1017
+    },
+    {
+      "epoch": 4.1214574898785425,
+      "grad_norm": 0.3365566414359312,
+      "learning_rate": 3.648558150665886e-06,
+      "loss": 0.1994,
+      "step": 1018
+    },
+    {
+      "epoch": 4.125506072874494,
+      "grad_norm": 0.3458863131753691,
+      "learning_rate": 3.616058102895075e-06,
+      "loss": 0.2028,
+      "step": 1019
+    },
+    {
+      "epoch": 4.129554655870446,
+      "grad_norm": 0.3364862802134336,
+      "learning_rate": 3.583689060687079e-06,
+      "loss": 0.1977,
+      "step": 1020
+    },
+    {
+      "epoch": 4.133603238866397,
+      "grad_norm": 0.33599010565839604,
+      "learning_rate": 3.5514512828639023e-06,
+      "loss": 0.1852,
+      "step": 1021
+    },
+    {
+      "epoch": 4.137651821862348,
+      "grad_norm": 0.3364802909687814,
+      "learning_rate": 3.51934502719796e-06,
+      "loss": 0.1806,
+      "step": 1022
+    },
+    {
+      "epoch": 4.1417004048583,
+      "grad_norm": 0.3240453278767888,
+      "learning_rate": 3.4873705504100185e-06,
+      "loss": 0.2109,
+      "step": 1023
+    },
+    {
+      "epoch": 4.145748987854251,
+      "grad_norm": 0.32908097895961125,
+      "learning_rate": 3.4555281081671345e-06,
+      "loss": 0.1965,
+      "step": 1024
+    },
+    {
+      "epoch": 4.149797570850202,
+      "grad_norm": 0.33164663189013666,
+      "learning_rate": 3.4238179550806326e-06,
+      "loss": 0.2058,
+      "step": 1025
+    },
+    {
+      "epoch": 4.153846153846154,
+      "grad_norm": 0.32865585862217983,
+      "learning_rate": 3.3922403447040497e-06,
+      "loss": 0.1842,
+      "step": 1026
+    },
+    {
+      "epoch": 4.157894736842105,
+      "grad_norm": 0.3227338106448975,
+      "learning_rate": 3.3607955295311247e-06,
+      "loss": 0.1791,
+      "step": 1027
+    },
+    {
+      "epoch": 4.161943319838056,
+      "grad_norm": 0.322295878483904,
+      "learning_rate": 3.32948376099375e-06,
+      "loss": 0.1915,
+      "step": 1028
+    },
+    {
+      "epoch": 4.165991902834008,
+      "grad_norm": 0.3174765094854709,
+      "learning_rate": 3.2983052894599976e-06,
+      "loss": 0.1934,
+      "step": 1029
+    },
+    {
+      "epoch": 4.17004048582996,
+      "grad_norm": 0.324124194027386,
+      "learning_rate": 3.2672603642320945e-06,
+      "loss": 0.1994,
+      "step": 1030
+    },
+    {
+      "epoch": 4.174089068825911,
+      "grad_norm": 0.32679079145336815,
+      "learning_rate": 3.236349233544438e-06,
+      "loss": 0.1778,
+      "step": 1031
+    },
+    {
+      "epoch": 4.178137651821863,
+      "grad_norm": 0.33014879084892623,
+      "learning_rate": 3.205572144561604e-06,
+      "loss": 0.1938,
+      "step": 1032
+    },
+    {
+      "epoch": 4.182186234817814,
+      "grad_norm": 0.3358262659180513,
+      "learning_rate": 3.174929343376374e-06,
+      "loss": 0.1907,
+      "step": 1033
+    },
+    {
+      "epoch": 4.186234817813765,
+      "grad_norm": 0.31242394510788807,
+      "learning_rate": 3.1444210750077777e-06,
+      "loss": 0.1803,
+      "step": 1034
+    },
+    {
+      "epoch": 4.190283400809717,
+      "grad_norm": 0.32573678941514894,
+      "learning_rate": 3.1140475833991046e-06,
+      "loss": 0.192,
+      "step": 1035
+    },
+    {
+      "epoch": 4.194331983805668,
+      "grad_norm": 0.3135638823940378,
+      "learning_rate": 3.0838091114159895e-06,
+      "loss": 0.1732,
+      "step": 1036
+    },
+    {
+      "epoch": 4.198380566801619,
+      "grad_norm": 0.33565829141259984,
+      "learning_rate": 3.0537059008444524e-06,
+      "loss": 0.1994,
+      "step": 1037
+    },
+    {
+      "epoch": 4.202429149797571,
+      "grad_norm": 0.34649214604495754,
+      "learning_rate": 3.0237381923889563e-06,
+      "loss": 0.1963,
+      "step": 1038
+    },
+    {
+      "epoch": 4.206477732793522,
+      "grad_norm": 0.3480560009768896,
+      "learning_rate": 2.9939062256704997e-06,
+      "loss": 0.1923,
+      "step": 1039
+    },
+    {
+      "epoch": 4.2105263157894735,
+      "grad_norm": 0.32717603286167135,
+      "learning_rate": 2.964210239224696e-06,
+      "loss": 0.1895,
+      "step": 1040
+    },
+    {
+      "epoch": 4.2145748987854255,
+      "grad_norm": 0.30938797350351993,
+      "learning_rate": 2.934650470499869e-06,
+      "loss": 0.2019,
+      "step": 1041
+    },
+    {
+      "epoch": 4.218623481781377,
+      "grad_norm": 0.31488216901167154,
+      "learning_rate": 2.905227155855128e-06,
+      "loss": 0.1824,
+      "step": 1042
+    },
+    {
+      "epoch": 4.222672064777328,
+      "grad_norm": 0.3224222517927278,
+      "learning_rate": 2.875940530558519e-06,
+      "loss": 0.1868,
+      "step": 1043
+    },
+    {
+      "epoch": 4.22672064777328,
+      "grad_norm": 0.3231762195934621,
+      "learning_rate": 2.8467908287851154e-06,
+      "loss": 0.1828,
+      "step": 1044
+    },
+    {
+      "epoch": 4.230769230769231,
+      "grad_norm": 0.3232372234559517,
+      "learning_rate": 2.8177782836151603e-06,
+      "loss": 0.1863,
+      "step": 1045
+    },
+    {
+      "epoch": 4.234817813765182,
+      "grad_norm": 0.31534246145536293,
+      "learning_rate": 2.7889031270321874e-06,
+      "loss": 0.1899,
+      "step": 1046
+    },
+    {
+      "epoch": 4.238866396761134,
+      "grad_norm": 0.31799496188627424,
+      "learning_rate": 2.760165589921184e-06,
+      "loss": 0.1953,
+      "step": 1047
+    },
+    {
+      "epoch": 4.242914979757085,
+      "grad_norm": 0.3326348222854781,
+      "learning_rate": 2.7315659020667305e-06,
+      "loss": 0.1852,
+      "step": 1048
+    },
+    {
+      "epoch": 4.246963562753036,
+      "grad_norm": 0.34207178303204144,
+      "learning_rate": 2.703104292151164e-06,
+      "loss": 0.183,
+      "step": 1049
+    },
+    {
+      "epoch": 4.251012145748988,
+      "grad_norm": 0.3367597833254254,
+      "learning_rate": 2.6747809877527607e-06,
+      "loss": 0.1696,
+      "step": 1050
+    },
+    {
+      "epoch": 4.255060728744939,
+      "grad_norm": 0.3200632236276798,
+      "learning_rate": 2.6465962153439082e-06,
+      "loss": 0.1966,
+      "step": 1051
+    },
+    {
+      "epoch": 4.2591093117408905,
+      "grad_norm": 0.31679487654329214,
+      "learning_rate": 2.618550200289303e-06,
+      "loss": 0.1997,
+      "step": 1052
+    },
+    {
+      "epoch": 4.2631578947368425,
+      "grad_norm": 0.3238318421048846,
+      "learning_rate": 2.59064316684412e-06,
+      "loss": 0.2,
+      "step": 1053
+    },
+    {
+      "epoch": 4.267206477732794,
+      "grad_norm": 0.3299469331354682,
+      "learning_rate": 2.562875338152264e-06,
+      "loss": 0.2023,
+      "step": 1054
+    },
+    {
+      "epoch": 4.271255060728745,
+      "grad_norm": 0.31130666665976814,
+      "learning_rate": 2.5352469362445552e-06,
+      "loss": 0.2079,
+      "step": 1055
+    },
+    {
+      "epoch": 4.275303643724697,
+      "grad_norm": 0.3255378902727481,
+      "learning_rate": 2.5077581820369523e-06,
+      "loss": 0.1917,
+      "step": 1056
+    },
+    {
+      "epoch": 4.279352226720648,
+      "grad_norm": 0.33159058842289685,
+      "learning_rate": 2.480409295328803e-06,
+      "loss": 0.1895,
+      "step": 1057
+    },
+    {
+      "epoch": 4.283400809716599,
+      "grad_norm": 0.3110789655909086,
+      "learning_rate": 2.45320049480108e-06,
+      "loss": 0.201,
+      "step": 1058
+    },
+    {
+      "epoch": 4.287449392712551,
+      "grad_norm": 0.34118270562621267,
+      "learning_rate": 2.4261319980146293e-06,
+      "loss": 0.1893,
+      "step": 1059
+    },
+    {
+      "epoch": 4.291497975708502,
+      "grad_norm": 0.32889656851438764,
+      "learning_rate": 2.399204021408421e-06,
+      "loss": 0.2048,
+      "step": 1060
+    },
+    {
+      "epoch": 4.295546558704453,
+      "grad_norm": 0.30165509633939347,
+      "learning_rate": 2.3724167802978395e-06,
+      "loss": 0.1901,
+      "step": 1061
+    },
+    {
+      "epoch": 4.299595141700405,
+      "grad_norm": 0.3304974434448672,
+      "learning_rate": 2.3457704888729625e-06,
+      "loss": 0.1889,
+      "step": 1062
+    },
+    {
+      "epoch": 4.303643724696356,
+      "grad_norm": 0.3339156270561477,
+      "learning_rate": 2.3192653601968162e-06,
+      "loss": 0.1801,
+      "step": 1063
+    },
+    {
+      "epoch": 4.3076923076923075,
+      "grad_norm": 0.31782102152217884,
+      "learning_rate": 2.2929016062037057e-06,
+      "loss": 0.1846,
+      "step": 1064
+    },
+    {
+      "epoch": 4.3117408906882595,
+      "grad_norm": 0.32534349356892667,
+      "learning_rate": 2.2666794376975055e-06,
+      "loss": 0.1993,
+      "step": 1065
+    },
+    {
+      "epoch": 4.315789473684211,
+      "grad_norm": 0.35282959701922706,
+      "learning_rate": 2.24059906434998e-06,
+      "loss": 0.1847,
+      "step": 1066
+    },
+    {
+      "epoch": 4.319838056680162,
+      "grad_norm": 0.3451710799308233,
+      "learning_rate": 2.214660694699091e-06,
+      "loss": 0.1949,
+      "step": 1067
+    },
+    {
+      "epoch": 4.323886639676114,
+      "grad_norm": 0.32167227662690695,
+      "learning_rate": 2.188864536147355e-06,
+      "loss": 0.1993,
+      "step": 1068
+    },
+    {
+      "epoch": 4.327935222672065,
+      "grad_norm": 0.32474066015579867,
+      "learning_rate": 2.1632107949601757e-06,
+      "loss": 0.2067,
+      "step": 1069
+    },
+    {
+      "epoch": 4.331983805668016,
+      "grad_norm": 0.3382010689836065,
+      "learning_rate": 2.137699676264171e-06,
+      "loss": 0.1746,
+      "step": 1070
+    },
+    {
+      "epoch": 4.336032388663968,
+      "grad_norm": 0.32291481325626054,
+      "learning_rate": 2.1123313840455738e-06,
+      "loss": 0.189,
+      "step": 1071
+    },
+    {
+      "epoch": 4.340080971659919,
+      "grad_norm": 0.31651971292689035,
+      "learning_rate": 2.08710612114857e-06,
+      "loss": 0.2,
+      "step": 1072
+    },
+    {
+      "epoch": 4.34412955465587,
+      "grad_norm": 0.34426859423045947,
+      "learning_rate": 2.0620240892736953e-06,
+      "loss": 0.1932,
+      "step": 1073
+    },
+    {
+      "epoch": 4.348178137651822,
+      "grad_norm": 0.3137111195849621,
+      "learning_rate": 2.0370854889762e-06,
+      "loss": 0.1846,
+      "step": 1074
+    },
+    {
+      "epoch": 4.352226720647773,
+      "grad_norm": 0.330225274254349,
+      "learning_rate": 2.0122905196644726e-06,
+      "loss": 0.1789,
+      "step": 1075
+    },
+    {
+      "epoch": 4.3562753036437245,
+      "grad_norm": 0.29942354520583697,
+      "learning_rate": 1.987639379598427e-06,
+      "loss": 0.1979,
+      "step": 1076
+    },
+    {
+      "epoch": 4.3603238866396765,
+      "grad_norm": 0.3431767637802596,
+      "learning_rate": 1.963132265887919e-06,
+      "loss": 0.1815,
+      "step": 1077
+    },
+    {
+      "epoch": 4.364372469635628,
+      "grad_norm": 0.3199302834914742,
+      "learning_rate": 1.9387693744911784e-06,
+      "loss": 0.1948,
+      "step": 1078
+    },
+    {
+      "epoch": 4.368421052631579,
+      "grad_norm": 0.33462494451557884,
+      "learning_rate": 1.9145509002132368e-06,
+      "loss": 0.182,
+      "step": 1079
+    },
+    {
+      "epoch": 4.372469635627531,
+      "grad_norm": 0.3216429020811058,
+      "learning_rate": 1.8904770367043723e-06,
+      "loss": 0.1946,
+      "step": 1080
+    },
+    {
+      "epoch": 4.376518218623482,
+      "grad_norm": 0.32865755113811546,
+      "learning_rate": 1.8665479764585504e-06,
+      "loss": 0.196,
+      "step": 1081
+    },
+    {
+      "epoch": 4.380566801619433,
+      "grad_norm": 0.3196403175059523,
+      "learning_rate": 1.8427639108118977e-06,
+      "loss": 0.1932,
+      "step": 1082
+    },
+    {
+      "epoch": 4.384615384615385,
+      "grad_norm": 0.3163098364513389,
+      "learning_rate": 1.8191250299411755e-06,
+      "loss": 0.1893,
+      "step": 1083
+    },
+    {
+      "epoch": 4.388663967611336,
+      "grad_norm": 0.3173882989643301,
+      "learning_rate": 1.7956315228622446e-06,
+      "loss": 0.2055,
+      "step": 1084
+    },
+    {
+      "epoch": 4.392712550607287,
+      "grad_norm": 0.3211404136298008,
+      "learning_rate": 1.7722835774285553e-06,
+      "loss": 0.2032,
+      "step": 1085
+    },
+    {
+      "epoch": 4.396761133603239,
+      "grad_norm": 0.3331910979386506,
+      "learning_rate": 1.7490813803296624e-06,
+      "loss": 0.188,
+      "step": 1086
+    },
+    {
+      "epoch": 4.40080971659919,
+      "grad_norm": 0.3245486931895574,
+      "learning_rate": 1.726025117089718e-06,
+      "loss": 0.1849,
+      "step": 1087
+    },
+    {
+      "epoch": 4.4048582995951415,
+      "grad_norm": 0.31290464801272216,
+      "learning_rate": 1.70311497206598e-06,
+      "loss": 0.1834,
+      "step": 1088
+    },
+    {
+      "epoch": 4.4089068825910935,
+      "grad_norm": 0.3387620775364345,
+      "learning_rate": 1.6803511284473638e-06,
+      "loss": 0.1888,
+      "step": 1089
+    },
+    {
+      "epoch": 4.412955465587045,
+      "grad_norm": 0.33089021000719054,
+      "learning_rate": 1.6577337682529538e-06,
+      "loss": 0.197,
+      "step": 1090
+    },
+    {
+      "epoch": 4.417004048582996,
+      "grad_norm": 0.3186055099484369,
+      "learning_rate": 1.6352630723305617e-06,
+      "loss": 0.192,
+      "step": 1091
+    },
+    {
+      "epoch": 4.421052631578947,
+      "grad_norm": 0.321683908558758,
+      "learning_rate": 1.612939220355272e-06,
+      "loss": 0.2003,
+      "step": 1092
+    },
+    {
+      "epoch": 4.425101214574899,
+      "grad_norm": 0.3208635051278085,
+      "learning_rate": 1.5907623908280112e-06,
+      "loss": 0.1909,
+      "step": 1093
+    },
+    {
+      "epoch": 4.42914979757085,
+      "grad_norm": 0.33002578541590194,
+      "learning_rate": 1.568732761074121e-06,
+      "loss": 0.1877,
+      "step": 1094
+    },
+    {
+      "epoch": 4.433198380566802,
+      "grad_norm": 0.3169926185155109,
+      "learning_rate": 1.5468505072419237e-06,
+      "loss": 0.1803,
+      "step": 1095
+    },
+    {
+      "epoch": 4.437246963562753,
+      "grad_norm": 0.3155520569331371,
+      "learning_rate": 1.5251158043013448e-06,
+      "loss": 0.2006,
+      "step": 1096
+    },
+    {
+      "epoch": 4.441295546558704,
+      "grad_norm": 0.3400570125282252,
+      "learning_rate": 1.5035288260424885e-06,
+      "loss": 0.2043,
+      "step": 1097
+    },
+    {
+      "epoch": 4.445344129554655,
+      "grad_norm": 0.30725946328972875,
+      "learning_rate": 1.4820897450742623e-06,
+      "loss": 0.1903,
+      "step": 1098
+    },
+    {
+      "epoch": 4.449392712550607,
+      "grad_norm": 0.3199655609390277,
+      "learning_rate": 1.4607987328229767e-06,
+      "loss": 0.1767,
+      "step": 1099
+    },
+    {
+      "epoch": 4.4534412955465585,
+      "grad_norm": 0.3161782751451782,
+      "learning_rate": 1.4396559595310056e-06,
+      "loss": 0.2066,
+      "step": 1100
+    },
+    {
+      "epoch": 4.4574898785425106,
+      "grad_norm": 0.3125534617629706,
+      "learning_rate": 1.4186615942553973e-06,
+      "loss": 0.1934,
+      "step": 1101
+    },
+    {
+      "epoch": 4.461538461538462,
+      "grad_norm": 0.3323632464703027,
+      "learning_rate": 1.3978158048665424e-06,
+      "loss": 0.1805,
+      "step": 1102
+    },
+    {
+      "epoch": 4.465587044534413,
+      "grad_norm": 0.3292728224569826,
+      "learning_rate": 1.3771187580468115e-06,
+      "loss": 0.1859,
+      "step": 1103
+    },
+    {
+      "epoch": 4.469635627530364,
+      "grad_norm": 0.3088665009761456,
+      "learning_rate": 1.3565706192892392e-06,
+      "loss": 0.1944,
+      "step": 1104
+    },
+    {
+      "epoch": 4.473684210526316,
+      "grad_norm": 0.3205394736937808,
+      "learning_rate": 1.3361715528961993e-06,
+      "loss": 0.184,
+      "step": 1105
+    },
+    {
+      "epoch": 4.477732793522267,
+      "grad_norm": 0.3060378509807788,
+      "learning_rate": 1.3159217219780817e-06,
+      "loss": 0.1932,
+      "step": 1106
+    },
+    {
+      "epoch": 4.481781376518219,
+      "grad_norm": 0.31923811279569825,
+      "learning_rate": 1.2958212884519949e-06,
+      "loss": 0.1935,
+      "step": 1107
+    },
+    {
+      "epoch": 4.48582995951417,
+      "grad_norm": 0.329267855429911,
+      "learning_rate": 1.2758704130404675e-06,
+      "loss": 0.1849,
+      "step": 1108
+    },
+    {
+      "epoch": 4.489878542510121,
+      "grad_norm": 0.329642511819745,
+      "learning_rate": 1.2560692552701738e-06,
+      "loss": 0.1831,
+      "step": 1109
+    },
+    {
+      "epoch": 4.493927125506072,
+      "grad_norm": 0.33678227459923415,
+      "learning_rate": 1.2364179734706382e-06,
+      "loss": 0.1849,
+      "step": 1110
+    },
+    {
+      "epoch": 4.497975708502024,
+      "grad_norm": 0.32013058461679467,
+      "learning_rate": 1.2169167247729852e-06,
+      "loss": 0.188,
+      "step": 1111
+    },
+    {
+      "epoch": 4.502024291497976,
+      "grad_norm": 0.3195130553541573,
+      "learning_rate": 1.1975656651086864e-06,
+      "loss": 0.1891,
+      "step": 1112
+    },
+    {
+      "epoch": 4.506072874493928,
+      "grad_norm": 0.32878990983928186,
+      "learning_rate": 1.1783649492082926e-06,
+      "loss": 0.1754,
+      "step": 1113
+    },
+    {
+      "epoch": 4.510121457489879,
+      "grad_norm": 0.32340016521090165,
+      "learning_rate": 1.1593147306002183e-06,
+      "loss": 0.1836,
+      "step": 1114
+    },
+    {
+      "epoch": 4.51417004048583,
+      "grad_norm": 0.34008970521365417,
+      "learning_rate": 1.140415161609505e-06,
+      "loss": 0.198,
+      "step": 1115
+    },
+    {
+      "epoch": 4.518218623481781,
+      "grad_norm": 0.31678690262346954,
+      "learning_rate": 1.121666393356604e-06,
+      "loss": 0.1917,
+      "step": 1116
+    },
+    {
+      "epoch": 4.522267206477733,
+      "grad_norm": 0.3257702816127247,
+      "learning_rate": 1.1030685757561632e-06,
+      "loss": 0.1779,
+      "step": 1117
+    },
+    {
+      "epoch": 4.526315789473684,
+      "grad_norm": 0.31742398404366085,
+      "learning_rate": 1.0846218575158373e-06,
+      "loss": 0.2147,
+      "step": 1118
+    },
+    {
+      "epoch": 4.530364372469636,
+      "grad_norm": 0.3212237159641466,
+      "learning_rate": 1.066326386135097e-06,
+      "loss": 0.1893,
+      "step": 1119
+    },
+    {
+      "epoch": 4.534412955465587,
+      "grad_norm": 0.32597840109145665,
+      "learning_rate": 1.0481823079040398e-06,
+      "loss": 0.1664,
+      "step": 1120
+    },
+    {
+      "epoch": 4.538461538461538,
+      "grad_norm": 0.3300817363073614,
+      "learning_rate": 1.0301897679022344e-06,
+      "loss": 0.1846,
+      "step": 1121
+    },
+    {
+      "epoch": 4.5425101214574894,
+      "grad_norm": 0.3266886776394248,
+      "learning_rate": 1.0123489099975491e-06,
+      "loss": 0.1806,
+      "step": 1122
+    },
+    {
+      "epoch": 4.5465587044534415,
+      "grad_norm": 0.3359505247796292,
+      "learning_rate": 9.9465987684501e-07,
+      "loss": 0.1608,
+      "step": 1123
+    },
+    {
+      "epoch": 4.550607287449393,
+      "grad_norm": 0.3212657288445211,
+      "learning_rate": 9.771228098856534e-07,
+      "loss": 0.2062,
+      "step": 1124
+    },
+    {
+      "epoch": 4.554655870445345,
+      "grad_norm": 0.3063305784566777,
+      "learning_rate": 9.597378493454013e-07,
+      "loss": 0.1803,
+      "step": 1125
+    },
+    {
+      "epoch": 4.558704453441296,
+      "grad_norm": 0.3292299578487427,
+      "learning_rate": 9.425051342339353e-07,
+      "loss": 0.1768,
+      "step": 1126
+    },
+    {
+      "epoch": 4.562753036437247,
+      "grad_norm": 0.3314358880496163,
+      "learning_rate": 9.254248023435864e-07,
+      "loss": 0.1757,
+      "step": 1127
+    },
+    {
+      "epoch": 4.566801619433198,
+      "grad_norm": 0.32548702708923805,
+      "learning_rate": 9.084969902482288e-07,
+      "loss": 0.1768,
+      "step": 1128
+    },
+    {
+      "epoch": 4.57085020242915,
+      "grad_norm": 0.3237540624328683,
+      "learning_rate": 8.917218333021993e-07,
+      "loss": 0.1849,
+      "step": 1129
+    },
+    {
+      "epoch": 4.574898785425101,
+      "grad_norm": 0.3301400005763725,
+      "learning_rate": 8.750994656392109e-07,
+      "loss": 0.1792,
+      "step": 1130
+    },
+    {
+      "epoch": 4.578947368421053,
+      "grad_norm": 0.32520706747204015,
+      "learning_rate": 8.58630020171265e-07,
+      "loss": 0.182,
+      "step": 1131
+    },
+    {
+      "epoch": 4.582995951417004,
+      "grad_norm": 0.31999344958774967,
+      "learning_rate": 8.423136285876099e-07,
+      "loss": 0.1994,
+      "step": 1132
+    },
+    {
+      "epoch": 4.587044534412955,
+      "grad_norm": 0.31749727829650326,
+      "learning_rate": 8.26150421353682e-07,
+      "loss": 0.2052,
+      "step": 1133
+    },
+    {
+      "epoch": 4.5910931174089065,
+      "grad_norm": 0.33729820072097094,
+      "learning_rate": 8.101405277100549e-07,
+      "loss": 0.1614,
+      "step": 1134
+    },
+    {
+      "epoch": 4.5951417004048585,
+      "grad_norm": 0.3108498439128492,
+      "learning_rate": 7.942840756714077e-07,
+      "loss": 0.1865,
+      "step": 1135
+    },
+    {
+      "epoch": 4.59919028340081,
+      "grad_norm": 0.3281964526190255,
+      "learning_rate": 7.785811920255093e-07,
+      "loss": 0.1904,
+      "step": 1136
+    },
+    {
+      "epoch": 4.603238866396762,
+      "grad_norm": 0.3258796856481552,
+      "learning_rate": 7.630320023322024e-07,
+      "loss": 0.1932,
+      "step": 1137
+    },
+    {
+      "epoch": 4.607287449392713,
+      "grad_norm": 0.32283710237373564,
+      "learning_rate": 7.476366309223903e-07,
+      "loss": 0.1951,
+      "step": 1138
+    },
+    {
+      "epoch": 4.611336032388664,
+      "grad_norm": 0.33177182233096814,
+      "learning_rate": 7.323952008970537e-07,
+      "loss": 0.199,
+      "step": 1139
+    },
+    {
+      "epoch": 4.615384615384615,
+      "grad_norm": 0.32214042871287635,
+      "learning_rate": 7.173078341262618e-07,
+      "loss": 0.1886,
+      "step": 1140
+    },
+    {
+      "epoch": 4.619433198380567,
+      "grad_norm": 0.32677649592799973,
+      "learning_rate": 7.023746512481989e-07,
+      "loss": 0.1774,
+      "step": 1141
+    },
+    {
+      "epoch": 4.623481781376518,
+      "grad_norm": 0.3124301941408824,
+      "learning_rate": 6.875957716681902e-07,
+      "loss": 0.186,
+      "step": 1142
+    },
+    {
+      "epoch": 4.62753036437247,
+      "grad_norm": 0.3372729257331374,
+      "learning_rate": 6.729713135577687e-07,
+      "loss": 0.1892,
+      "step": 1143
+    },
+    {
+      "epoch": 4.631578947368421,
+      "grad_norm": 0.3233989581254828,
+      "learning_rate": 6.585013938537143e-07,
+      "loss": 0.1956,
+      "step": 1144
+    },
+    {
+      "epoch": 4.635627530364372,
+      "grad_norm": 0.3323282955483524,
+      "learning_rate": 6.441861282571115e-07,
+      "loss": 0.1822,
+      "step": 1145
+    },
+    {
+      "epoch": 4.6396761133603235,
+      "grad_norm": 0.3252765872903235,
+      "learning_rate": 6.300256312324471e-07,
+      "loss": 0.1714,
+      "step": 1146
+    },
+    {
+      "epoch": 4.6437246963562755,
+      "grad_norm": 0.33412996326724226,
+      "learning_rate": 6.160200160066776e-07,
+      "loss": 0.1874,
+      "step": 1147
+    },
+    {
+      "epoch": 4.647773279352227,
+      "grad_norm": 0.32705147984172905,
+      "learning_rate": 6.021693945683327e-07,
+      "loss": 0.182,
+      "step": 1148
+    },
+    {
+      "epoch": 4.651821862348179,
+      "grad_norm": 0.310385198263215,
+      "learning_rate": 5.884738776666044e-07,
+      "loss": 0.2067,
+      "step": 1149
+    },
+    {
+      "epoch": 4.65587044534413,
+      "grad_norm": 0.312362453519246,
+      "learning_rate": 5.749335748104878e-07,
+      "loss": 0.195,
+      "step": 1150
+    },
+    {
+      "epoch": 4.659919028340081,
+      "grad_norm": 0.3124797970815278,
+      "learning_rate": 5.615485942678778e-07,
+      "loss": 0.207,
+      "step": 1151
+    },
+    {
+      "epoch": 4.663967611336032,
+      "grad_norm": 0.31494653003570905,
+      "learning_rate": 5.483190430647289e-07,
+      "loss": 0.1731,
+      "step": 1152
+    },
+    {
+      "epoch": 4.668016194331984,
+      "grad_norm": 0.3217045219981912,
+      "learning_rate": 5.352450269841747e-07,
+      "loss": 0.1703,
+      "step": 1153
+    },
+    {
+      "epoch": 4.672064777327935,
+      "grad_norm": 0.3119223731942894,
+      "learning_rate": 5.22326650565701e-07,
+      "loss": 0.1919,
+      "step": 1154
+    },
+    {
+      "epoch": 4.676113360323887,
+      "grad_norm": 0.3115889032488499,
+      "learning_rate": 5.095640171043048e-07,
+      "loss": 0.1907,
+      "step": 1155
+    },
+    {
+      "epoch": 4.680161943319838,
+      "grad_norm": 0.3123962747704769,
+      "learning_rate": 4.969572286496594e-07,
+      "loss": 0.2115,
+      "step": 1156
+    },
+    {
+      "epoch": 4.684210526315789,
+      "grad_norm": 0.31281455740216485,
+      "learning_rate": 4.845063860053079e-07,
+      "loss": 0.2018,
+      "step": 1157
+    },
+    {
+      "epoch": 4.6882591093117405,
+      "grad_norm": 0.31187433815844046,
+      "learning_rate": 4.7221158872785997e-07,
+      "loss": 0.1877,
+      "step": 1158
+    },
+    {
+      "epoch": 4.6923076923076925,
+      "grad_norm": 0.32314185191915273,
+      "learning_rate": 4.600729351261857e-07,
+      "loss": 0.177,
+      "step": 1159
+    },
+    {
+      "epoch": 4.696356275303644,
+      "grad_norm": 0.32361167698083193,
+      "learning_rate": 4.4809052226063355e-07,
+      "loss": 0.2012,
+      "step": 1160
+    },
+    {
+      "epoch": 4.700404858299595,
+      "grad_norm": 0.3255871123635018,
+      "learning_rate": 4.3626444594226046e-07,
+      "loss": 0.1889,
+      "step": 1161
+    },
+    {
+      "epoch": 4.704453441295547,
+      "grad_norm": 0.33045390947311887,
+      "learning_rate": 4.245948007320588e-07,
+      "loss": 0.1871,
+      "step": 1162
+    },
+    {
+      "epoch": 4.708502024291498,
+      "grad_norm": 0.3151687949475516,
+      "learning_rate": 4.130816799401971e-07,
+      "loss": 0.1758,
+      "step": 1163
+    },
+    {
+      "epoch": 4.712550607287449,
+      "grad_norm": 0.31828731112718806,
+      "learning_rate": 4.017251756252827e-07,
+      "loss": 0.1903,
+      "step": 1164
+    },
+    {
+      "epoch": 4.716599190283401,
+      "grad_norm": 0.3326616304546707,
+      "learning_rate": 3.9052537859362473e-07,
+      "loss": 0.1952,
+      "step": 1165
+    },
+    {
+      "epoch": 4.720647773279352,
+      "grad_norm": 0.3126000682439883,
+      "learning_rate": 3.7948237839849687e-07,
+      "loss": 0.2151,
+      "step": 1166
+    },
+    {
+      "epoch": 4.724696356275303,
+      "grad_norm": 0.3093927815138861,
+      "learning_rate": 3.6859626333943797e-07,
+      "loss": 0.1908,
+      "step": 1167
+    },
+    {
+      "epoch": 4.728744939271255,
+      "grad_norm": 0.3222286313868019,
+      "learning_rate": 3.578671204615347e-07,
+      "loss": 0.1764,
+      "step": 1168
+    },
+    {
+      "epoch": 4.732793522267206,
+      "grad_norm": 0.3197329036683679,
+      "learning_rate": 3.472950355547267e-07,
+      "loss": 0.1915,
+      "step": 1169
+    },
+    {
+      "epoch": 4.7368421052631575,
+      "grad_norm": 0.3099018845375203,
+      "learning_rate": 3.3688009315312244e-07,
+      "loss": 0.1888,
+      "step": 1170
+    },
+    {
+      "epoch": 4.7408906882591095,
+      "grad_norm": 0.33252502991898286,
+      "learning_rate": 3.266223765343224e-07,
+      "loss": 0.1929,
+      "step": 1171
+    },
+    {
+      "epoch": 4.744939271255061,
+      "grad_norm": 0.3276562713111983,
+      "learning_rate": 3.16521967718757e-07,
+      "loss": 0.1734,
+      "step": 1172
+    },
+    {
+      "epoch": 4.748987854251012,
+      "grad_norm": 0.32144536038460036,
+      "learning_rate": 3.0657894746902726e-07,
+      "loss": 0.1866,
+      "step": 1173
+    },
+    {
+      "epoch": 4.753036437246964,
+      "grad_norm": 0.30820190208022463,
+      "learning_rate": 2.9679339528925654e-07,
+      "loss": 0.2024,
+      "step": 1174
+    },
+    {
+      "epoch": 4.757085020242915,
+      "grad_norm": 0.3076434269080316,
+      "learning_rate": 2.871653894244597e-07,
+      "loss": 0.206,
+      "step": 1175
+    },
+    {
+      "epoch": 4.761133603238866,
+      "grad_norm": 0.31791571745026875,
+      "learning_rate": 2.7769500685991266e-07,
+      "loss": 0.1753,
+      "step": 1176
+    },
+    {
+      "epoch": 4.765182186234818,
+      "grad_norm": 0.3270727572861424,
+      "learning_rate": 2.683823233205485e-07,
+      "loss": 0.1993,
+      "step": 1177
+    },
+    {
+      "epoch": 4.769230769230769,
+      "grad_norm": 0.3269860443579605,
+      "learning_rate": 2.592274132703287e-07,
+      "loss": 0.1906,
+      "step": 1178
+    },
+    {
+      "epoch": 4.77327935222672,
+      "grad_norm": 0.320383262264032,
+      "learning_rate": 2.502303499116754e-07,
+      "loss": 0.1995,
+      "step": 1179
+    },
+    {
+      "epoch": 4.777327935222672,
+      "grad_norm": 0.32313068398676525,
+      "learning_rate": 2.413912051848666e-07,
+      "loss": 0.1915,
+      "step": 1180
+    },
+    {
+      "epoch": 4.781376518218623,
+      "grad_norm": 0.3270645339329153,
+      "learning_rate": 2.3271004976746613e-07,
+      "loss": 0.1865,
+      "step": 1181
+    },
+    {
+      "epoch": 4.7854251012145745,
+      "grad_norm": 0.32554139819082506,
+      "learning_rate": 2.2418695307376393e-07,
+      "loss": 0.205,
+      "step": 1182
+    },
+    {
+      "epoch": 4.7894736842105265,
+      "grad_norm": 0.332594616939356,
+      "learning_rate": 2.1582198325420968e-07,
+      "loss": 0.1798,
+      "step": 1183
+    },
+    {
+      "epoch": 4.793522267206478,
+      "grad_norm": 0.32330177024822987,
+      "learning_rate": 2.0761520719488447e-07,
+      "loss": 0.1724,
+      "step": 1184
+    },
+    {
+      "epoch": 4.797570850202429,
+      "grad_norm": 0.32514563571901195,
+      "learning_rate": 1.9956669051694798e-07,
+      "loss": 0.1792,
+      "step": 1185
+    },
+    {
+      "epoch": 4.801619433198381,
+      "grad_norm": 0.33743296972479553,
+      "learning_rate": 1.916764975761254e-07,
+      "loss": 0.1796,
+      "step": 1186
+    },
+    {
+      "epoch": 4.805668016194332,
+      "grad_norm": 0.3033578453911393,
+      "learning_rate": 1.839446914621923e-07,
+      "loss": 0.1928,
+      "step": 1187
+    },
+    {
+      "epoch": 4.809716599190283,
+      "grad_norm": 0.33270952860767466,
+      "learning_rate": 1.7637133399845962e-07,
+      "loss": 0.1877,
+      "step": 1188
+    },
+    {
+      "epoch": 4.813765182186235,
+      "grad_norm": 0.31222676182688264,
+      "learning_rate": 1.6895648574129397e-07,
+      "loss": 0.2003,
+      "step": 1189
+    },
+    {
+      "epoch": 4.817813765182186,
+      "grad_norm": 0.31701617056089526,
+      "learning_rate": 1.6170020597962465e-07,
+      "loss": 0.2054,
+      "step": 1190
+    },
+    {
+      "epoch": 4.821862348178137,
+      "grad_norm": 0.3264901040975524,
+      "learning_rate": 1.5460255273447078e-07,
+      "loss": 0.1827,
+      "step": 1191
+    },
+    {
+      "epoch": 4.825910931174089,
+      "grad_norm": 0.32713036809084245,
+      "learning_rate": 1.4766358275847715e-07,
+      "loss": 0.1795,
+      "step": 1192
+    },
+    {
+      "epoch": 4.82995951417004,
+      "grad_norm": 0.3220763057185667,
+      "learning_rate": 1.4088335153546573e-07,
+      "loss": 0.1671,
+      "step": 1193
+    },
+    {
+      "epoch": 4.834008097165992,
+      "grad_norm": 0.32437248885683695,
+      "learning_rate": 1.3426191327998496e-07,
+      "loss": 0.1914,
+      "step": 1194
+    },
+    {
+      "epoch": 4.838056680161944,
+      "grad_norm": 0.3110435983335065,
+      "learning_rate": 1.277993209368744e-07,
+      "loss": 0.1998,
+      "step": 1195
+    },
+    {
+      "epoch": 4.842105263157895,
+      "grad_norm": 0.30132773646267835,
+      "learning_rate": 1.214956261808542e-07,
+      "loss": 0.1993,
+      "step": 1196
+    },
+    {
+      "epoch": 4.846153846153846,
+      "grad_norm": 0.3185842512688148,
+      "learning_rate": 1.1535087941609846e-07,
+      "loss": 0.1871,
+      "step": 1197
+    },
+    {
+      "epoch": 4.850202429149798,
+      "grad_norm": 0.3308563082756751,
+      "learning_rate": 1.0936512977583802e-07,
+      "loss": 0.1866,
+      "step": 1198
+    },
+    {
+      "epoch": 4.854251012145749,
+      "grad_norm": 0.3220562178844999,
+      "learning_rate": 1.0353842512196732e-07,
+      "loss": 0.1607,
+      "step": 1199
+    },
+    {
+      "epoch": 4.8582995951417,
+      "grad_norm": 0.32628896284772047,
+      "learning_rate": 9.787081204466253e-08,
+      "loss": 0.195,
+      "step": 1200
+    },
+    {
+      "epoch": 4.862348178137652,
+      "grad_norm": 0.3215454300839537,
+      "learning_rate": 9.236233586200627e-08,
+      "loss": 0.1912,
+      "step": 1201
+    },
+    {
+      "epoch": 4.866396761133603,
+      "grad_norm": 0.3095511490035609,
+      "learning_rate": 8.70130406196279e-08,
+      "loss": 0.2046,
+      "step": 1202
+    },
+    {
+      "epoch": 4.870445344129554,
+      "grad_norm": 0.33696992414694626,
+      "learning_rate": 8.182296909035047e-08,
+      "loss": 0.2008,
+      "step": 1203
+    },
+    {
+      "epoch": 4.874493927125506,
+      "grad_norm": 0.30139945384772665,
+      "learning_rate": 7.679216277384439e-08,
+      "loss": 0.1865,
+      "step": 1204
+    },
+    {
+      "epoch": 4.8785425101214575,
+      "grad_norm": 0.3274679199284958,
+      "learning_rate": 7.19206618963053e-08,
+      "loss": 0.1704,
+      "step": 1205
+    },
+    {
+      "epoch": 4.882591093117409,
+      "grad_norm": 0.3277430747435027,
+      "learning_rate": 6.720850541012347e-08,
+      "loss": 0.1878,
+      "step": 1206
+    },
+    {
+      "epoch": 4.886639676113361,
+      "grad_norm": 0.3139760230212527,
+      "learning_rate": 6.265573099357714e-08,
+      "loss": 0.1989,
+      "step": 1207
+    },
+    {
+      "epoch": 4.890688259109312,
+      "grad_norm": 0.32819980642539404,
+      "learning_rate": 5.826237505053067e-08,
+      "loss": 0.1783,
+      "step": 1208
+    },
+    {
+      "epoch": 4.894736842105263,
+      "grad_norm": 0.3243415934602552,
+      "learning_rate": 5.4028472710137e-08,
+      "loss": 0.2067,
+      "step": 1209
+    },
+    {
+      "epoch": 4.898785425101215,
+      "grad_norm": 0.3184277332603675,
+      "learning_rate": 4.9954057826571146e-08,
+      "loss": 0.1887,
+      "step": 1210
+    },
+    {
+      "epoch": 4.902834008097166,
+      "grad_norm": 0.32599665652542414,
+      "learning_rate": 4.6039162978743776e-08,
+      "loss": 0.1765,
+      "step": 1211
+    },
+    {
+      "epoch": 4.906882591093117,
+      "grad_norm": 0.3170670704111622,
+      "learning_rate": 4.228381947004812e-08,
+      "loss": 0.2037,
+      "step": 1212
+    },
+    {
+      "epoch": 4.910931174089069,
+      "grad_norm": 0.3210384987694565,
+      "learning_rate": 3.868805732811343e-08,
+      "loss": 0.1911,
+      "step": 1213
+    },
+    {
+      "epoch": 4.91497975708502,
+      "grad_norm": 0.3141679462438823,
+      "learning_rate": 3.525190530455636e-08,
+      "loss": 0.2014,
+      "step": 1214
+    },
+    {
+      "epoch": 4.919028340080971,
+      "grad_norm": 0.3237474393338069,
+      "learning_rate": 3.197539087475887e-08,
+      "loss": 0.1796,
+      "step": 1215
+    },
+    {
+      "epoch": 4.923076923076923,
+      "grad_norm": 0.32244106599390004,
+      "learning_rate": 2.8858540237641786e-08,
+      "loss": 0.1894,
+      "step": 1216
+    },
+    {
+      "epoch": 4.9271255060728745,
+      "grad_norm": 0.3236517601156332,
+      "learning_rate": 2.5901378315458247e-08,
+      "loss": 0.184,
+      "step": 1217
+    },
+    {
+      "epoch": 4.931174089068826,
+      "grad_norm": 0.3300186986784507,
+      "learning_rate": 2.3103928753598347e-08,
+      "loss": 0.1994,
+      "step": 1218
+    },
+    {
+      "epoch": 4.935222672064778,
+      "grad_norm": 0.3166244127907655,
+      "learning_rate": 2.0466213920393718e-08,
+      "loss": 0.1964,
+      "step": 1219
+    },
+    {
+      "epoch": 4.939271255060729,
+      "grad_norm": 0.3116946030039317,
+      "learning_rate": 1.7988254906942115e-08,
+      "loss": 0.1813,
+      "step": 1220
+    },
+    {
+      "epoch": 4.94331983805668,
+      "grad_norm": 0.31999600243105947,
+      "learning_rate": 1.5670071526936447e-08,
+      "loss": 0.186,
+      "step": 1221
+    },
+    {
+      "epoch": 4.947368421052632,
+      "grad_norm": 0.30401472520121636,
+      "learning_rate": 1.3511682316509344e-08,
+      "loss": 0.1981,
+      "step": 1222
+    },
+    {
+      "epoch": 4.951417004048583,
+      "grad_norm": 0.3088318429239499,
+      "learning_rate": 1.1513104534086605e-08,
+      "loss": 0.168,
+      "step": 1223
+    },
+    {
+      "epoch": 4.955465587044534,
+      "grad_norm": 0.30200360837352336,
+      "learning_rate": 9.67435416023843e-09,
+      "loss": 0.2008,
+      "step": 1224
+    },
+    {
+      "epoch": 4.959514170040486,
+      "grad_norm": 0.332946206467641,
+      "learning_rate": 7.995445897566178e-09,
+      "loss": 0.1951,
+      "step": 1225
+    },
+    {
+      "epoch": 4.963562753036437,
+      "grad_norm": 0.32748577197801176,
+      "learning_rate": 6.476393170578021e-09,
+      "loss": 0.2047,
+      "step": 1226
+    },
+    {
+      "epoch": 4.967611336032388,
+      "grad_norm": 0.3096219035694745,
+      "learning_rate": 5.11720812557348e-09,
+      "loss": 0.1888,
+      "step": 1227
+    },
+    {
+      "epoch": 4.97165991902834,
+      "grad_norm": 0.32632863588139505,
+      "learning_rate": 3.917901630563492e-09,
+      "loss": 0.174,
+      "step": 1228
+    },
+    {
+      "epoch": 4.9757085020242915,
+      "grad_norm": 0.3350546782031217,
+      "learning_rate": 2.878483275170485e-09,
+      "loss": 0.1799,
+      "step": 1229
+    },
+    {
+      "epoch": 4.979757085020243,
+      "grad_norm": 0.33371464341106516,
+      "learning_rate": 1.998961370555108e-09,
+      "loss": 0.1936,
+      "step": 1230
+    },
+    {
+      "epoch": 4.983805668016195,
+      "grad_norm": 0.3183107199165395,
+      "learning_rate": 1.2793429493518361e-09,
+      "loss": 0.1747,
+      "step": 1231
+    },
+    {
+      "epoch": 4.987854251012146,
+      "grad_norm": 0.32976450492614073,
+      "learning_rate": 7.196337656112384e-10,
+      "loss": 0.1987,
+      "step": 1232
+    },
+    {
+      "epoch": 4.991902834008097,
+      "grad_norm": 0.3225216133190626,
+      "learning_rate": 3.198382947511292e-10,
+      "loss": 0.1888,
+      "step": 1233
+    },
+    {
+      "epoch": 4.995951417004049,
+      "grad_norm": 0.32132392411948235,
+      "learning_rate": 7.995973352770137e-11,
+      "loss": 0.171,
+      "step": 1234
+    },
+    {
+      "epoch": 5.0,
+      "grad_norm": 0.3116456803465757,
+      "learning_rate": 0.0,
+      "loss": 0.1954,
+      "step": 1235
+    },
+    {
+      "epoch": 5.0,
+      "step": 1235,
+      "total_flos": 1359537785995264.0,
+      "train_loss": 0.4319087717335234,
+      "train_runtime": 19656.7744,
+      "train_samples_per_second": 8.038,
+      "train_steps_per_second": 0.063
+    }
+  ],
+  "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": 1359537785995264.0,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}